1 2013-08-14 Xinliang David Li <davidxl@google.com>
2 * config/i386/i386.c (ix86_option_override_internal):
3 Fix potential unitialized variable error.
5 2013-08-14 Xinliang David Li <davidxl@google.com>
7 * config/i386/i386.opt: Define two new options.
8 * config/i386/x86-tune.def: Add arch selector field in macros.
9 * config/i386/i386.h: Adjust macro definition.
10 * config/i386/i386.c (ix86_option_override_internal):
12 (parse_mtune_ctrl_str): New function.
13 (set_ix86_tune_features): New function.
14 (ix86_function_specific_restore): Call the new helper function.
16 2013-08-14 Andrey Belevantsev <abel@ispras.ru>
18 PR rtl-optimization/57662
19 * sel-sched.c (code_motion_process_successors): When the current insn
20 is removed after the recursive traversal, break from the loop.
21 Add comments and debug printouts.
23 2013-08-14 Jakub Jelinek <jakub@redhat.com>
24 Alexandre Oliva <aoliva@redhat.com>
27 * config/i386/i386.c (ix86_delegitimize_address): For CM_MEDIUM_PIC
28 and CM_LARGE_PIC ix86_cmodel fall thru into the -m32 code, handle
29 there also UNSPEC_PLTOFF.
31 2013-08-14 Marek Polacek <polacek@redhat.com>
33 * ipa-inline-analysis.c (add_clause): Avoid shifting integer
34 NUM_CONDITIONS bit positions.
36 2013-08-13 Cary Coutant <ccoutant@google.com>
38 * dwarf2out.c (CHECKSUM_BLOCK): New macro.
39 (attr_checksum): Hash vector contents instead of pointer.
40 (attr_checksum_ordered): Likewise.
42 2013-08-13 Uros Bizjak <ubizjak@gmail.com>
44 * config/i386/sse.md (*sse2_maskmovdqu): Emit addr32 prefix
45 when Pmode != word_mode. Add length_address attribute.
46 (sse3_monitor_<mode>): Merge from sse3_monitor and
47 sse3_monitor64_<mode> insn patterns. Emit addr32 prefix when
48 Pmode != word_mode. Update insn length attribute.
49 * config/i386/i386.c (ix86_option_override_internal): Update
50 ix86_gen_monitor selection for merged sse3_monitor insn.
52 2013-08-13 Julian Brown <julian@codesourcery.com>
54 * config/rs6000/rs6000.c (rs6000_legitimize_reload_address): Don't
55 perform invalid legitimization on greater-than-word-size modes for
58 2013-08-13 Vladimir Makarov <vmakarov@redhat.com>
60 * ira.c (setup_class_translate_array): Use aclass instead of cl
61 for classes not fully covered by allocno classes.
63 2013-08-13 Jakub Jelinek <jakub@redhat.com>
65 PR tree-optimization/57661
66 * tree-inline.h (struct copy_body_data): Add blocks_to_copy field.
67 * tree-inline.c (tree_function_versioning): Initialize it.
68 (remap_gimple_stmt): Return GIMPLE_NOP for MEM_REF lhs clobber stmts
69 if id->blocks_to_copy and MEM_REF's SSA_NAME is defined in a block
70 that is not being copied.
73 * asan.c (instrument_strlen_call): Fix typo in comment.
74 Use char * type even for the lhs of POINTER_PLUS_EXPR.
76 2013-08-13 Steve Ellcey <sellcey@mips.com>
78 * config/mips/mips.md (prefetch): Use lw instead of ld on
79 loongson in 32bit mode.
81 2013-08-13 Nick Clifton <nickc@redhat.com>
83 * config.gcc: (avr-linux): Allow for tmake_file not being empty.
85 2013-08-13 Jan Hubicka <jh@suse.cz>
87 * cgraph.c (cgraph_turn_edge_to_speculative): Return newly
88 introduced edge; fix typo in sanity check.
89 (cgraph_resolve_speculation): Export; improve diagnostic.
90 (cgraph_redirect_edge_call_stmt_to_callee): Better diagnostic; cancel
91 speculation at type mismatch.
92 * cgraph.h (cgraph_turn_edge_to_speculative): Update.
93 (cgraph_resolve_speculation): Declare.
94 (symtab_can_be_discarded): New function.
95 * value-prof.c (gimple_ic_transform): Remove actual transform code.
96 * ipa-inline-transform.c (speculation_removed): New global var.
97 (clone_inlined_nodes): See if speculation can be removed.
98 (inline_call): If speculations was removed, we growths may not match.
99 * ipa-inline.c (can_inline_edge_p): Add DISREGARD_LIMITS parameter.
100 (speculation_useful_p): New function.
101 (resolve_noninline_speculation): New function.
102 (inline_small_functions): Resolve useless speculations.
103 * ipa-inline.h (speculation_useful_p): Declare
104 * ipa.c (can_replace_by_local_alias): Simplify.
105 (ipa_profile): Produce speculative calls in non-lto, too;
106 add simple cost model; produce local aliases.
108 2013-08-13 David Malcolm <dmalcolm@redhat.com>
110 * config/i386/t-i386 (i386.o): Rename stray PIPELINE_H to
113 2013-08-12 Paolo Carlini <paolo.carlini@oracle.com>
115 * config/i386/i386.c (ix86_function_versions): Use error + inform.
117 2013-08-12 Uros Bizjak <ubizjak@gmail.com>
119 * config/i386/i386.md (floatunssi<mode>2 expand): Use MODEF mode
120 iterator instead of X87MODEF.
122 2013-08-12 Perez Read <netfirewall@gmail.com>
125 * config/i386/i386.md (*movabs<mode>_1): Add <ptrsize> PTR before
126 operand 0 for intel asm alternative.
127 (*movabs<mode>_2): Ditto for operand 1.
129 2013-08-12 James Greenhalgh <james.greenhalgh@arm.com>
131 * config/aarch64/arm_none.h
132 (vdup<bhsd>_lane_<su><8,16,32,64>): Fix macro call.
134 2013-08-12 Nick Clifton <nickc@redhat.com>
136 * config.gcc (m32r-linux): Allow for tmake_file not being empty.
138 2013-08-12 Yuri Rumyantsev <ysrumyan@gmail.com>
140 * config/i386/i386.md (floatunssi<mode>2 expand): Add new
141 expand for QI/HImode operand to produce more effictive code for
142 unsigned char(short) --> float(double) conversion.
144 2013-08-12 Alexander Monakov <amonakov@ispras.ru>
146 * doc/invoke.texi: Mention that -ftls-model does not force the final
149 2013-08-12 Marek Polacek <polacek@redhat.com>
150 Marc Glisse <marc.glisse@inria.fr>
152 PR tree-optimization/57980
153 * tree-tailcall.c (process_assignment): Call build_minus_one_cst
154 when creating -1 constant.
156 2013-08-10 Jan Hubicka <jh@suse.cz>
158 Workaround binutils PR14342.
159 * tree-profile.c (init_ic_make_global_vars): Add LTO path.
160 (gimple_init_edge_profiler): Likewise.
161 (gimple_gen_ic_func_profiler): Likewise.
163 2013-08-09 Jan Hubicka <jh@suse.cz>
165 * cgraph.c (cgraph_create_edge_1): Clear speculative flag.
167 2013-08-09 Xinliang David Li <davidxl@google.com>
169 * config/i386/stringop.def: New file.
170 * config/i386/stringop.opt: New file.
171 * config/i386/i386-opts.h: Include stringopt.def.
172 * config/i386/i386.opt: Include stringopt.opt.
173 * config/i386/i386.c (ix86_option_override_internal):
174 Override default size based stringop inline strategies with options.
175 * config/i386/i386.c (ix86_parse_stringop_strategy_string):
178 2013-08-09 Jan Hubicka <jh@suse.cz>
180 * ipa-ref.c (ipa_clear_stmts_in_references): Clear lto_stmt_uid, too.
182 2013-08-09 Jan Hubicka <jh@suse.cz>
184 * cgraph.c (cgraph_resolve_speculation): Cut frequency to
186 (dump_cgraph_node): Dump profile-id.
187 * cgraph.h (cgraph_indirect_call_info): Add common_target_id
188 and common_target_probability.
189 * lto-cgraph.c (lto_output_edge): Stream common targets.
190 (lto_output_node): Stream profile ids.
191 (input_node): Stream profile ids.
192 (input_edge): Stream common targets.
193 * lto-streamer-in.c (fixup_call_stmt_edges_1): Fix formatting.
194 * ipa.c: Include value-prof.h
195 (ipa_profile_generate_summary): Turn indirect call statement histograms
197 (ipa_profile): Turn common targets into speculative edges.
199 2013-08-09 Jan Hubicka <jh@suse.cz>
201 * cgraph.h (cgraph_node): Add profile_id.
202 * value-prof.c (cgraph_node_map): Turn into pointer_map.
203 (init_node_map): Rewrite to handle hashes increas of incremental IDs.
204 (del_node_map): Update.
205 (find_func_by_funcdef_no): Replace by ...
206 (find_func_by_profile_id): ... this one.
207 (gimple_ic_transform): Do not remove useful histograms when
208 speculation is not done; dump info when indirect call removal
210 * value-prof.h (find_func_by_profile_id, gimple_ic): Declare.
211 * gcov-io.h (__gcov_indirect_call_profiler): Replace by ...
212 (__gcov_indirect_call_profiler_v2): .. this one.
213 * profile.h (init_node_map): Update.
214 * coverage.c (coverage_compute_profile_id): New function.
215 * coverage.h (coverage_compute_profile_id): Declare.
216 * tree-profile.c (init_ic_make_global_vars): Make
217 __gcov_indirect_call_callee and __gcov_indirect_call_counters global.
218 (gimple_init_edge_profiler): Update prototype of
219 __gcov_indirect_call_profiler.
220 (gimple_gen_ic_func_profiler): Simplify.
221 (tree_profiling): Use init_node_map
223 2013-08-09 Jan Hubicka <jh@suse.cz>
225 * cgraphbuild.c (cgraph_rebuild_references): Rebuild only
226 non-speculative refs.
227 * cgraph.c (cgraph_update_edge_in_call_site_hash): New function.
228 (cgraph_add_edge_to_call_site_hash): Deal with speculative calls.
229 (cgraph_set_call_stmt): Likewise.
230 (cgraph_create_edge_1): Fix release checking compilatoin;
232 (cgraph_free_edge): Free indirect info.
233 (cgraph_turn_edge_to_speculative): New function.
234 (cgraph_speculative_call_info): New function.
235 (cgraph_make_edge_direct): Return direct edge; handle speculation.
236 (cgraph_redirect_edge_call_stmt_to_callee): Expand speculative edges.
237 (dump_cgraph_node): Dump speculation.
238 (verify_edge_count_and_frequency): Accept speculative edges.
239 (verify_edge_corresponds_to_fndecl): Handle partitioned cgraph.
240 (verify_cgraph_node): Handle speculation.
241 * cgraph.h (cgraph_edge): Add SPECULATIVE flag.
242 (cgraph_set_call_stmt): Update prototype.
243 (cgraph_make_edge_direct): Update prototype.
244 (cgraph_speculative_call_info): Declare.
245 * ipa-cp.c (ipcp_discover_new_direct_edges): Be ready for edge
246 to change; update call of ipa_find_references.
247 * ipa-ref.c (ipa_record_reference): Fix return value; clear
248 lto_stmt_uid and speculative flags.
249 (ipa_dump_references): Dump speculation.
250 (ipa_clone_references): Clone speculative flag.
251 (ipa_clone_referring): Likewise.
252 (ipa_clone_ref): New function.
253 (ipa_find_reference): Look into lto_stmt_uids
254 (ipa_clear_stmts_in_references): Do not clear speculative calls.
255 * ipa-ref.h (ipa_ref): Add lto_stmt_uid and speculative flags.
256 (ipa_find_reference): Update declaration.
257 (ipa_clone_ref): Declare.
258 * lto-cgraph.c (lto_output_edge): Make lto_stmt_uids start from 0;
259 stream speculative flag.
260 (lto_output_ref): Stream statements uids and speculation.
261 (input_ref): Likewise.
262 (input_edge): Stream speuclation.
263 * cgraphclones.c (cgraph_clone_edge): Clone speculation.
264 (cgraph_set_call_stmt_including_clones): Handle speculation.
265 * ipa-inline.c (heap_edge_removal_hook): New function.
266 (inline_small_functions): Register it.
267 * lto-streamer-in.c (fixup_call_stmt_edges_1): Bounds checking;
268 also initialize refs.
269 * ipa-prop.c (ipa_make_edge_direct_to_target): Be ready for
271 (try_make_edge_direct_simple_call): Likewise.
272 (try_make_edge_direct_simple_call): Likewise.
273 (update_indirect_edges_after_inlining): Likewise.
274 (remove_described_reference): Look proper lto_stmt_uid.
275 (propagate_controlled_uses): Likewise.
276 (propagate_controlled_uses): Liekwise.
277 * tree-inline.c (copy_bb): Copy speculative edges.
278 (redirect_all_calls): New function.
279 (copy_cfg_body): Do redirection after loop info is updated.
280 (delete_unreachable_blocks_update_callgraph): Updadte speculation.
282 2013-08-09 Jan Hubicka <jh@suse.cz>
284 * lto-streamer-out.c (output_function): Renumber PHIs.
285 * lto-streamer-in.c (input_function): Likewise.
287 2013-08-09 James Greenhalgh <james.greenhalgh@arm.com>
289 * config/aarch64/aarch64-simd-builtins.def (get_lane_signed): Remove.
290 (get_lane_unsigned): Likewise.
291 (dup_lane_scalar): Likewise.
292 (get_lane): enable for VALL.
293 * config/aarch64/aarch64-simd.md
294 (aarch64_dup_lane_scalar<mode>): Remove.
295 (aarch64_get_lane_signed<mode>): Likewise.
296 (aarch64_get_lane_unsigned<mode>): Likewise.
297 (aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): New.
298 (aarch64_get_lane_zero_extendsi<mode>): Likewise.
299 (aarch64_get_lane<mode>): Enable for all vector modes.
300 (aarch64_get_lanedi): Remove misleading constraints.
301 * config/aarch64/arm_neon.h
302 (__aarch64_vget_lane_any): Define.
303 (__aarch64_vget<q>_lane_<fpsu><8,16,32,64>): Likewise.
304 (vget<q>_lane_<fpsu><8,16,32,64>): Use __aarch64_vget_lane macros.
305 (vdup<bhsd>_lane_<su><8,16,32,64>): Likewise.
306 * config/aarch64/iterators.md (VDQQH): New.
310 2013-08-09 Eric Botcazou <ebotcazou@adacore.com>
312 * configure.ac: Add GAS check for LEON instructions on SPARC.
313 * configure: Regenerate.
314 * config.in: Likewise.
315 * config.gcc (with_cpu): Remove sparc-leon*-* and deal with LEON in the
317 * config/sparc/sparc.opt (LEON, LEON3): New masks.
318 * config/sparc/sparc.h (ASM_CPU32_DEFAULT_SPEC): Set to AS_LEON_FLAG
320 (ASM_CPU_SPEC): Pass AS_LEON_FLAG if -mcpu=leon or -mcpu=leon3.
321 (AS_LEON_FLAG): New macro.
322 * config/sparc/sparc.c (sparc_option_override): Set MASK_LEON for leon
323 and MASK_LEON3 for leon3 and unset them if HAVE_AS_LEON is not defined.
324 Deal with LEON and LEON3 for the memory model.
325 * config/sparc/sync.md (atomic_compare_and_swap<mode>): Enable if LEON3
326 (atomic_compare_and_swap<mode>_1): Likewise.
327 (*atomic_compare_and_swap<mode>_1): Likewise.
329 2013-08-09 Zhenqiang Chen <zhenqiang.chen@linaro.org>
331 * config/arm/neon.md (vcond): Fix floating-point vector
332 comparisons against 0.
334 2013-08-08 Vladimir Makarov <vmakarov@redhat.com>
336 * lra-constraints.c (emit_spill_move): Remove assert.
337 (process_alt_operands): Add more debugging
338 output. Increase reject for spilling into memory. Decrease
339 reject for reloading scratch.
340 (split_reg): Use HARD_REGNO_CALLER_SAVE_MODE.
342 2013-08-08 Steve Ellcey <sellcey@mips.com>
344 * config/mips/mti-linux.h (SYSROOT_SUFFIX_SPEC): Add nan2008.
345 * config/mips/t-mti-elf (MULTILIB_OPTIONS): Make mips16 and
346 micromips incompatible. Add nan2008.
347 (MULTILIB_DIRNAMES): Add nan2008.
348 (MULTILIB_EXCEPTIONS): Remove mips16/micromips entry.
349 * config/mips/t-mti-linux (MULTILIB_OPTIONS): Make mips16
350 and micromips incompatible. Add nan2008.
351 (MULTILIB_DIRNAMES): Add nan2008.
352 (MULTILIB_EXCEPTIONS): Remove mips16/micromips entry.
354 2013-08-08 Richard Sandiford <rdsandiford@googlemail.com>
356 PR rtl-optimization/58079
357 * combine.c (combine_simplify_rtx): Avoid using SUBST if
358 simplify_comparison has widened a comparison with an integer.
360 2013-08-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
362 * config/arm/neon.md (movmisalign<mode>): Disable when we
363 don't allow unaligned accesses.
364 (*movmisalign<mode>_neon_store): Likewise.
365 (*movmisalign<mode>_neon_load): Likewise.
366 (*movmisalign<mode>_neon_store): Likewise.
367 (*movmisalign<mode>_neon_load): Likewise.
369 2013-08-08 Jan Hubicka <jh@suse.cz>
371 * cgraphbuild.c (build_cgraph_edges): Do not walk into debugs.
372 (make_pass_rebuild_cgraph_edges): Also clear references.
373 * cgraph.c (verify_cgraph_node): Add basic ipa-ref verifier.
374 * ipa-inline-transform.c (inline_transform): Remove all references
376 * cgraphunit.c (expand_function): Remove all references after
378 * ipa-ref.c (ipa_ref_has_aliases_p): Fix formatting.
379 (ipa_find_reference): Rewrite to iterator.
380 (remove_stmt_references): Likewise.
381 (ipa_clear_stmts_in_references): New function.
382 * ipa-ref.h (ipa_clear_stmts_in_references): Declare.
383 * cgraphclones.c (cgraph_materialize_all_clones): Remove or
385 * ipa-split.c (split_function): Remove references in split function.
387 2013-08-08 Richard Earnshaw <rearnsha@arm.com>
390 * config/arm/arm/neon.md (neon_vld1_dupdi): New expand pattern.
391 (neon_vld1_dup<mode> VD iterator): Iterate over VD not VDX.
393 2013-08-08 Richard Earnshaw <rearnsha@arm.com>
396 * config/arm/arm.c (aapcs_vfp_allocate): Decompose the argument if the
397 suggested mode for the assignment isn't compatible with the
400 2013-08-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
403 * config/arm/arm.h (MALLOC_ABI_ALIGNMENT): Define.
405 2013-08-07 Xinliang David Li <davidxl@google.com>
407 * config/i386/i386.opt: New option -mtune-ctrl=.
408 * config/i386/x86-tune.def: New file.
409 * config/i386/i386.h: include x86-tune.def.
410 * config/i386/i386.c (ix86_option_override_internal):
411 Parsing -mtune-ctrl= option and set tune features.
413 2013-08-07 Oleg Endo <olegendo@gcc.gnu.org>
416 * config/rs6000/rs6000.c (gen_2arg_fn_t): Remove typedef.
417 (rs6000_emit_swdiv, rs6000_emit_swrsqrt): Don't cast result of GEN_FCN
420 2013-08-07 Eric Botcazou <ebotcazou@adacore.com>
422 * rtl.h (update_alignments): Declare.
423 * final.c (grow_label_align): New function extracted from...
424 (shorten_branches): ...here. Call it.
425 (update_alignments): New function.
426 * reorg.c (sibling_labels): New variable.
427 (get_label_before): Add SIBLING parameter. If it is non-zero, push
428 the new label along with it onto the sibling_labels vector.
429 (fill_simple_delay_slots): Adjust call to get_label_before.
430 (fill_slots_from_thread): Likewise.
431 (relax_delay_slots): Likewise.
432 (make_return_insns): Likewise.
433 (dbr_schedule): Invoke update_alignment on the sibling_labels vector.
435 2013-08-07 Eric Botcazou <ebotcazou@adacore.com>
437 * diagnostic.c (diagnostic_classify_diagnostic): Accept zero index and
438 document its semantics.
439 (diagnostic_report_diagnostic): Adjust accordingly.
441 2013-08-07 David Malcolm <dmalcolm@redhat.com>
443 * config/sparc/sparc.c (insert_pass_work_around_errata): Move into...
444 (sparc_option_override): ...and port to new C++ pass API.
445 * config/sparc/t-sparc (sparc.o): Add dep on CONTEXT_H
447 2013-08-07 Peter Bergner <bergner@vnet.ibm.com>
449 * config/rs6000/rs6000.c (htm_expand_builtin) <case 0>: Remove.
451 2013-08-06 Caroline Tice <cmtice@google.com>
453 * gcc.c (VTABLE_VERIFICATION_SPEC): New definition.
454 (LINK_COMMAND_SPEC): Add VTABLE_VERIFICATION_SPEC.
455 * tree-pass.h: Add pass_vtable_verify.
456 * varasm.c (assemble_variable): Add code to properly set the comdat
457 section and name for the .vtable_map_vars section.
458 (assemble_vtyv_preinit_initializer): New function.
459 (default_sectin_type_flags): Make sure .vtable_map_vars section has
461 * output.h: Add function decl for assemble_vtv_preinit_initializer.
462 * vtable-verify.c: New file.
463 * vtable-verify.h: New file.
464 * flag-types.h (enum vtv_priority): Defintions for flag_vtable_verify
465 initialiation levels.
466 * timevar.def (TV_VTABLE_VERIFICATION): New definition.
467 * passes.def: Insert pass_vtable_verify.
468 * aclocal.m4: Reorder includes.
469 * doc/invoke.texi: Document the -fvtable-verify=, -fvtv-debug, and
470 -fvtv-counts options.
471 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add vtv_start*.o,
472 as appropriate, if -fvtable-verify=... is used.
473 (GNU_USER_TARGET_ENDFILE_SPEC): Add vtv_end*.o as appropriate, if
474 -fvtable-verify=... is used.
475 * Makefile.in (OBJS): Add vtable-verify.o to list.
476 (vtable-verify.o): Add new build rule.
477 (GTFILES): Add vtable-verify.c to list.
478 * common.opt (fvtable-verify=): New flag.
479 (vtv_priority): Values for fvtable-verify= flag.
480 (fvtv-counts): New flag.
481 (fvtv-debug): New flag.
482 * tree.h (save_vtable_map_decl): New extern function decl.
484 2013-08-07 David Malcolm <dmalcolm@redhat.com>
486 * config/rl78/rl78.c (rl78_devirt_pass): Convert from a struct to...
487 (pass_rl78_devirt): ...new subclass of rtl_opt_pass along with...
488 (pass_data_rl78_devirt): ...new pass_data instance and...
489 (make_pass_rl78_devirt): ...new function.
490 (rl78_asm_file_start): Port pass registration to new C++ API.
492 2013-08-07 David Malcolm <dmalcolm@redhat.com>
494 * coretypes.h (rtl_opt_pass): Add.
496 * config/epiphany/epiphany.c (pass_mode_switch_use): New.
497 (epiphany_init): Port to new C++ pass API.
498 (epiphany_optimize_mode_switching): Likewise.
499 * pass_manager.h (pass_manager::get_pass_split_all_insns): New.
500 (pass_manager::get_pass_mode_switching): New.
501 (pass_manager::get_pass_peephole2): New.
502 * mode-switching.c (pass_mode_switching): Add clone method.
503 * recog.c (pass_peephole2): Add clone method.
504 (pass_split_all_insns): Add clone method.
506 2013-08-06 David Malcolm <dmalcolm@redhat.com>
508 * config/mips/mips.c (insert_pass_mips_machine_reorg2): Move into...
509 (mips_option_override): ...here, porting to new C++ API for passes.
511 2013-08-06 Jan Hubicka <jh@suse.cz>
513 * cgraph.c (cgraph_get_body): New function based on lto.c
515 * cgraph.h (cgraph_get_body): Declare.
516 * cgraphclones.c (cgraph_create_virtual_clone): Commonize WPA and
518 * cgraphunit.c (expand_function): Get body prior expanding.
519 * ipa.c (function_and_variable_visibility): Use gimple_has_body_p test.
520 * lto-cgraph.c (lto_output_node): Do not stream bodies we don't
522 * passes.c (do_per_function_toporder): Get body.
523 * tree-inline.c (expand_call_inline): Get body prior inlining it.
524 * tree-ssa-structalias.c (ipa_pta_execute): Get body; skip clones.
526 2013-08-06 Martin Jambor <mjambor@suse.cz>
529 * cgraphunit.c (cgraph_finalize_function): Assert that nested function
530 is not re-finalized. Rename second parameter to no_collect.
532 2013-08-06 Martin Jambor <mjambor@suse.cz>
535 * gimple-ssa-strength-reduction.c (replace_ref): Make sure built
536 MEM_REF has proper alignment information.
538 2013-08-05 Oleg Endo <olegendo@gcc.gnu.org>
541 * recog.h (rtx (*insn_gen_fn) (rtx, ...)): Replace typedef with new
543 * expr.c (move_by_pieces_1, store_by_pieces_2): Replace argument
544 rtx (*) (rtx, ...) with insn_gen_fn.
545 * genoutput.c (output_insn_data): Cast gen_? function pointers to
546 insn_gen_fn::stored_funcptr. Add initializer braces.
548 2013-08-05 David Malcolm <dmalcolm@redhat.com>
550 Rewrite how instances of passes are cloned to remove assumptions
551 about their sizes (thus allowing pass subclasses to have
552 additional data fields, albeit non-GC-managed ones at this point).
554 * passes.c (make_pass_instance): Now that passes have clone
555 methods, rewrite this function to eliminate XNEW and memcpy
556 calls that used hardcoded sizes. Since this function no longer
557 creates pass instances, rename it to...
558 (add_pass_instance): ...this. Document the old way that passes were
559 numbered and flagged, and rework this function to continue using it.
560 (next_pass_1): Add an initial_pass argument for use by
562 (position_pass): When adding multiple instances of a pass, use
563 the pass's clone method, rather than relying on the XNEW/memcpy
564 within the former make_pass_instance (now add_pass_instance).
565 (pass_manager::pass_manager): When invoking next_pass_1, also supply
566 the initial instance of the current pass within the pass manager.
568 2013-08-05 David Malcolm <dmalcolm@redhat.com>
570 This is the automated part of the conversion of passes from C
571 structs to C++ classes.
573 Patch autogenerated by refactor_passes.py from
574 https://github.com/davidmalcolm/gcc-refactoring-scripts
575 revision 03fe39476a4c4ea450b49e087cfa817b5f92021e
577 * asan.c (pass_asan): Convert from a global struct to a subclass of
578 gimple_opt_pass along with...
579 (pass_data_asan): ...new pass_data instance and...
580 (make_pass_asan): ...new function.
581 (pass_asan_O0): Convert from a global struct to a subclass of
582 gimple_opt_pass along with...
583 (pass_data_asan_O0): ...new pass_data instance and...
584 (make_pass_asan_O0): ...new function.
585 * auto-inc-dec.c (pass_inc_dec): Convert from a global struct to a
586 subclass of rtl_opt_pass along with...
587 (pass_data_inc_dec): ...new pass_data instance and...
588 (make_pass_inc_dec): ...new function.
589 * bb-reorder.c (pass_reorder_blocks): Convert from a global struct to
590 a subclass of rtl_opt_pass along with...
591 (pass_data_reorder_blocks): ...new pass_data instance and...
592 (make_pass_reorder_blocks): ...new function.
593 (pass_duplicate_computed_gotos): Convert from a global struct to a
594 subclass of rtl_opt_pass along with...
595 (pass_data_duplicate_computed_gotos): ...new pass_data instance and...
596 (make_pass_duplicate_computed_gotos): ...new function.
597 (pass_partition_blocks): Convert from a global struct to a subclass of
598 rtl_opt_pass along with...
599 (pass_data_partition_blocks): ...new pass_data instance and...
600 (make_pass_partition_blocks): ...new function.
601 * bt-load.c (pass_branch_target_load_optimize1): Convert from a global
602 struct to a subclass of rtl_opt_pass along with...
603 (pass_data_branch_target_load_optimize1): ...new pass_data instance
605 (make_pass_branch_target_load_optimize1): ...new function.
606 (pass_branch_target_load_optimize2): Convert from a global struct to a
607 subclass of rtl_opt_pass along with...
608 (pass_data_branch_target_load_optimize2): ...new pass_data instance
610 (make_pass_branch_target_load_optimize2): ...new function.
611 * cfgcleanup.c (pass_jump): Convert from a global struct to a subclass
612 of rtl_opt_pass along with...
613 (pass_data_jump): ...new pass_data instance and...
614 (make_pass_jump): ...new function.
615 (pass_jump2): Convert from a global struct to a subclass of
616 rtl_opt_pass along with...
617 (pass_data_jump2): ...new pass_data instance and...
618 (make_pass_jump2): ...new function.
619 * cfgexpand.c (pass_expand): Convert from a global struct to a
620 subclass of rtl_opt_pass along with...
621 (pass_data_expand): ...new pass_data instance and...
622 (make_pass_expand): ...new function.
623 * cfgrtl.c (pass_free_cfg): Convert from a global struct to a subclass
624 of rtl_opt_pass along with...
625 (pass_data_free_cfg): ...new pass_data instance and...
626 (make_pass_free_cfg): ...new function.
627 (pass_into_cfg_layout_mode): Convert from a global struct to a
628 subclass of rtl_opt_pass along with...
629 (pass_data_into_cfg_layout_mode): ...new pass_data instance and...
630 (make_pass_into_cfg_layout_mode): ...new function.
631 (pass_outof_cfg_layout_mode): Convert from a global struct to a
632 subclass of rtl_opt_pass along with...
633 (pass_data_outof_cfg_layout_mode): ...new pass_data instance and...
634 (make_pass_outof_cfg_layout_mode): ...new function.
635 * cgraphbuild.c (pass_build_cgraph_edges): Convert from a global
636 struct to a subclass of gimple_opt_pass along with...
637 (pass_data_build_cgraph_edges): ...new pass_data instance and...
638 (make_pass_build_cgraph_edges): ...new function.
639 (pass_rebuild_cgraph_edges): Convert from a global struct to a
640 subclass of gimple_opt_pass along with...
641 (pass_data_rebuild_cgraph_edges): ...new pass_data instance and...
642 (make_pass_rebuild_cgraph_edges): ...new function.
643 (pass_remove_cgraph_callee_edges): Convert from a global struct to a
644 subclass of gimple_opt_pass along with...
645 (pass_data_remove_cgraph_callee_edges): ...new pass_data instance
647 (make_pass_remove_cgraph_callee_edges): ...new function.
648 * combine-stack-adj.c (pass_stack_adjustments): Convert from a global
649 struct to a subclass of rtl_opt_pass along with...
650 (pass_data_stack_adjustments): ...new pass_data instance and...
651 (make_pass_stack_adjustments): ...new function.
652 * combine.c (pass_combine): Convert from a global struct to a subclass
653 of rtl_opt_pass along with...
654 (pass_data_combine): ...new pass_data instance and...
655 (make_pass_combine): ...new function.
656 * compare-elim.c (pass_compare_elim_after_reload): Convert from a
657 global struct to a subclass of rtl_opt_pass along with...
658 (pass_data_compare_elim_after_reload): ...new pass_data instance
660 (make_pass_compare_elim_after_reload): ...new function.
661 * cprop.c (pass_rtl_cprop): Convert from a global struct to a subclass
662 of rtl_opt_pass along with...
663 (pass_data_rtl_cprop): ...new pass_data instance and...
664 (make_pass_rtl_cprop): ...new function.
665 * cse.c (pass_cse): Convert from a global struct to a subclass of
666 rtl_opt_pass along with...
667 (pass_data_cse): ...new pass_data instance and...
668 (make_pass_cse): ...new function.
669 (pass_cse2): Convert from a global struct to a subclass of
670 rtl_opt_pass along with...
671 (pass_data_cse2): ...new pass_data instance and...
672 (make_pass_cse2): ...new function.
673 (pass_cse_after_global_opts): Convert from a global struct to a
674 subclass of rtl_opt_pass along with...
675 (pass_data_cse_after_global_opts): ...new pass_data instance and...
676 (make_pass_cse_after_global_opts): ...new function.
677 * dce.c (pass_ud_rtl_dce): Convert from a global struct to a subclass
678 of rtl_opt_pass along with...
679 (pass_data_ud_rtl_dce): ...new pass_data instance and...
680 (make_pass_ud_rtl_dce): ...new function.
681 (pass_fast_rtl_dce): Convert from a global struct to a subclass of
682 rtl_opt_pass along with...
683 (pass_data_fast_rtl_dce): ...new pass_data instance and...
684 (make_pass_fast_rtl_dce): ...new function.
685 * df-core.c (pass_df_initialize_opt): Convert from a global struct to
686 a subclass of rtl_opt_pass along with...
687 (pass_data_df_initialize_opt): ...new pass_data instance and...
688 (make_pass_df_initialize_opt): ...new function.
689 (pass_df_initialize_no_opt): Convert from a global struct to a
690 subclass of rtl_opt_pass along with...
691 (pass_data_df_initialize_no_opt): ...new pass_data instance and...
692 (make_pass_df_initialize_no_opt): ...new function.
693 (pass_df_finish): Convert from a global struct to a subclass of
694 rtl_opt_pass along with...
695 (pass_data_df_finish): ...new pass_data instance and...
696 (make_pass_df_finish): ...new function.
697 * dse.c (pass_rtl_dse1): Convert from a global struct to a subclass of
698 rtl_opt_pass along with...
699 (pass_data_rtl_dse1): ...new pass_data instance and...
700 (make_pass_rtl_dse1): ...new function.
701 (pass_rtl_dse2): Convert from a global struct to a subclass of
702 rtl_opt_pass along with...
703 (pass_data_rtl_dse2): ...new pass_data instance and...
704 (make_pass_rtl_dse2): ...new function.
705 * dwarf2cfi.c (pass_dwarf2_frame): Convert from a global struct to a
706 subclass of rtl_opt_pass along with...
707 (pass_data_dwarf2_frame): ...new pass_data instance and...
708 (make_pass_dwarf2_frame): ...new function.
709 * except.c (pass_set_nothrow_function_flags): Convert from a global
710 struct to a subclass of rtl_opt_pass along with...
711 (pass_data_set_nothrow_function_flags): ...new pass_data instance
713 (make_pass_set_nothrow_function_flags): ...new function.
714 (pass_convert_to_eh_region_ranges): Convert from a global struct to a
715 subclass of rtl_opt_pass along with...
716 (pass_data_convert_to_eh_region_ranges): ...new pass_data instance
718 (make_pass_convert_to_eh_region_ranges): ...new function.
719 * final.c (pass_compute_alignments): Convert from a global struct to a
720 subclass of rtl_opt_pass along with...
721 (pass_data_compute_alignments): ...new pass_data instance and...
722 (make_pass_compute_alignments): ...new function.
723 (pass_final): Convert from a global struct to a subclass of
724 rtl_opt_pass along with...
725 (pass_data_final): ...new pass_data instance and...
726 (make_pass_final): ...new function.
727 (pass_shorten_branches): Convert from a global struct to a subclass of
728 rtl_opt_pass along with...
729 (pass_data_shorten_branches): ...new pass_data instance and...
730 (make_pass_shorten_branches): ...new function.
731 (pass_clean_state): Convert from a global struct to a subclass of
732 rtl_opt_pass along with...
733 (pass_data_clean_state): ...new pass_data instance and...
734 (make_pass_clean_state): ...new function.
735 * function.c (pass_instantiate_virtual_regs): Convert from a global
736 struct to a subclass of rtl_opt_pass along with...
737 (pass_data_instantiate_virtual_regs): ...new pass_data instance and...
738 (make_pass_instantiate_virtual_regs): ...new function.
739 (pass_leaf_regs): Convert from a global struct to a subclass of
740 rtl_opt_pass along with...
741 (pass_data_leaf_regs): ...new pass_data instance and...
742 (make_pass_leaf_regs): ...new function.
743 (pass_thread_prologue_and_epilogue): Convert from a global struct to a
744 subclass of rtl_opt_pass along with...
745 (pass_data_thread_prologue_and_epilogue): ...new pass_data instance
747 (make_pass_thread_prologue_and_epilogue): ...new function.
748 (pass_match_asm_constraints): Convert from a global struct to a
749 subclass of rtl_opt_pass along with...
750 (pass_data_match_asm_constraints): ...new pass_data instance and...
751 (make_pass_match_asm_constraints): ...new function.
752 * fwprop.c (pass_rtl_fwprop): Convert from a global struct to a
753 subclass of rtl_opt_pass along with...
754 (pass_data_rtl_fwprop): ...new pass_data instance and...
755 (make_pass_rtl_fwprop): ...new function.
756 (pass_rtl_fwprop_addr): Convert from a global struct to a subclass of
757 rtl_opt_pass along with...
758 (pass_data_rtl_fwprop_addr): ...new pass_data instance and...
759 (make_pass_rtl_fwprop_addr): ...new function.
760 * gcse.c (pass_rtl_pre): Convert from a global struct to a subclass of
761 rtl_opt_pass along with...
762 (pass_data_rtl_pre): ...new pass_data instance and...
763 (make_pass_rtl_pre): ...new function.
764 (pass_rtl_hoist): Convert from a global struct to a subclass of
765 rtl_opt_pass along with...
766 (pass_data_rtl_hoist): ...new pass_data instance and...
767 (make_pass_rtl_hoist): ...new function.
768 * gimple-low.c (pass_lower_cf): Convert from a global struct to a
769 subclass of gimple_opt_pass along with...
770 (pass_data_lower_cf): ...new pass_data instance and...
771 (make_pass_lower_cf): ...new function.
772 * gimple-ssa-strength-reduction.c (pass_strength_reduction): Convert
773 from a global struct to a subclass of gimple_opt_pass along with...
774 (pass_data_strength_reduction): ...new pass_data instance and...
775 (make_pass_strength_reduction): ...new function.
776 * ifcvt.c (pass_rtl_ifcvt): Convert from a global struct to a subclass
777 of rtl_opt_pass along with...
778 (pass_data_rtl_ifcvt): ...new pass_data instance and...
779 (make_pass_rtl_ifcvt): ...new function.
780 (pass_if_after_combine): Convert from a global struct to a subclass of
781 rtl_opt_pass along with...
782 (pass_data_if_after_combine): ...new pass_data instance and...
783 (make_pass_if_after_combine): ...new function.
784 (pass_if_after_reload): Convert from a global struct to a subclass of
785 rtl_opt_pass along with...
786 (pass_data_if_after_reload): ...new pass_data instance and...
787 (make_pass_if_after_reload): ...new function.
788 * init-regs.c (pass_initialize_regs): Convert from a global struct to
789 a subclass of rtl_opt_pass along with...
790 (pass_data_initialize_regs): ...new pass_data instance and...
791 (make_pass_initialize_regs): ...new function.
792 * ipa-cp.c (pass_ipa_cp): Convert from a global struct to a subclass
793 of ipa_opt_pass_d along with...
794 (pass_data_ipa_cp): ...new pass_data instance and...
795 (make_pass_ipa_cp): ...new function.
796 * ipa-inline-analysis.c (pass_inline_parameters): Convert from a
797 global struct to a subclass of gimple_opt_pass along with...
798 (pass_data_inline_parameters): ...new pass_data instance and...
799 (make_pass_inline_parameters): ...new function.
800 * ipa-inline.c (pass_early_inline): Convert from a global struct to a
801 subclass of gimple_opt_pass along with...
802 (pass_data_early_inline): ...new pass_data instance and...
803 (make_pass_early_inline): ...new function.
804 (pass_ipa_inline): Convert from a global struct to a subclass of
805 ipa_opt_pass_d along with...
806 (pass_data_ipa_inline): ...new pass_data instance and...
807 (make_pass_ipa_inline): ...new function.
808 * ipa-pure-const.c (pass_local_pure_const): Convert from a global
809 struct to a subclass of gimple_opt_pass along with...
810 (pass_data_local_pure_const): ...new pass_data instance and...
811 (make_pass_local_pure_const): ...new function.
812 (pass_ipa_pure_const): Convert from a global struct to a subclass of
813 ipa_opt_pass_d along with...
814 (pass_data_ipa_pure_const): ...new pass_data instance and...
815 (make_pass_ipa_pure_const): ...new function.
816 * ipa-reference.c (pass_ipa_reference): Convert from a global struct
817 to a subclass of ipa_opt_pass_d along with...
818 (pass_data_ipa_reference): ...new pass_data instance and...
819 (make_pass_ipa_reference): ...new function.
820 * ipa-split.c (pass_split_functions): Convert from a global struct to
821 a subclass of gimple_opt_pass along with...
822 (pass_data_split_functions): ...new pass_data instance and...
823 (make_pass_split_functions): ...new function.
824 (pass_feedback_split_functions): Convert from a global struct to a
825 subclass of gimple_opt_pass along with...
826 (pass_data_feedback_split_functions): ...new pass_data instance and...
827 (make_pass_feedback_split_functions): ...new function.
828 * ipa.c (pass_ipa_function_and_variable_visibility): Convert from a
829 global struct to a subclass of simple_ipa_opt_pass along with...
830 (pass_data_ipa_function_and_variable_visibility): ...new pass_data
832 (make_pass_ipa_function_and_variable_visibility): ...new function.
833 (pass_ipa_free_inline_summary): Convert from a global struct to a
834 subclass of simple_ipa_opt_pass along with...
835 (pass_data_ipa_free_inline_summary): ...new pass_data instance and...
836 (make_pass_ipa_free_inline_summary): ...new function.
837 (pass_ipa_whole_program_visibility): Convert from a global struct to a
838 subclass of ipa_opt_pass_d along with...
839 (pass_data_ipa_whole_program_visibility): ...new pass_data instance
841 (make_pass_ipa_whole_program_visibility): ...new function.
842 (pass_ipa_profile): Convert from a global struct to a subclass of
843 ipa_opt_pass_d along with...
844 (pass_data_ipa_profile): ...new pass_data instance and...
845 (make_pass_ipa_profile): ...new function.
846 (pass_ipa_cdtor_merge): Convert from a global struct to a subclass of
847 ipa_opt_pass_d along with...
848 (pass_data_ipa_cdtor_merge): ...new pass_data instance and...
849 (make_pass_ipa_cdtor_merge): ...new function.
850 * ira.c (pass_ira): Convert from a global struct to a subclass of
851 rtl_opt_pass along with...
852 (pass_data_ira): ...new pass_data instance and...
853 (make_pass_ira): ...new function.
854 (pass_reload): Convert from a global struct to a subclass of
855 rtl_opt_pass along with...
856 (pass_data_reload): ...new pass_data instance and...
857 (make_pass_reload): ...new function.
858 * jump.c (pass_cleanup_barriers): Convert from a global struct to a
859 subclass of rtl_opt_pass along with...
860 (pass_data_cleanup_barriers): ...new pass_data instance and...
861 (make_pass_cleanup_barriers): ...new function.
862 * loop-init.c (pass_loop2): Convert from a global struct to a subclass
863 of rtl_opt_pass along with...
864 (pass_data_loop2): ...new pass_data instance and...
865 (make_pass_loop2): ...new function.
866 (pass_rtl_loop_init): Convert from a global struct to a subclass of
867 rtl_opt_pass along with...
868 (pass_data_rtl_loop_init): ...new pass_data instance and...
869 (make_pass_rtl_loop_init): ...new function.
870 (pass_rtl_loop_done): Convert from a global struct to a subclass of
871 rtl_opt_pass along with...
872 (pass_data_rtl_loop_done): ...new pass_data instance and...
873 (make_pass_rtl_loop_done): ...new function.
874 (pass_rtl_move_loop_invariants): Convert from a global struct to a
875 subclass of rtl_opt_pass along with...
876 (pass_data_rtl_move_loop_invariants): ...new pass_data instance and...
877 (make_pass_rtl_move_loop_invariants): ...new function.
878 (pass_rtl_unswitch): Convert from a global struct to a subclass of
879 rtl_opt_pass along with...
880 (pass_data_rtl_unswitch): ...new pass_data instance and...
881 (make_pass_rtl_unswitch): ...new function.
882 (pass_rtl_unroll_and_peel_loops): Convert from a global struct to a
883 subclass of rtl_opt_pass along with...
884 (pass_data_rtl_unroll_and_peel_loops): ...new pass_data instance
886 (make_pass_rtl_unroll_and_peel_loops): ...new function.
887 (pass_rtl_doloop): Convert from a global struct to a subclass of
888 rtl_opt_pass along with...
889 (pass_data_rtl_doloop): ...new pass_data instance and...
890 (make_pass_rtl_doloop): ...new function.
891 * lower-subreg.c (pass_lower_subreg): Convert from a global struct to
892 a subclass of rtl_opt_pass along with...
893 (pass_data_lower_subreg): ...new pass_data instance and...
894 (make_pass_lower_subreg): ...new function.
895 (pass_lower_subreg2): Convert from a global struct to a subclass of
896 rtl_opt_pass along with...
897 (pass_data_lower_subreg2): ...new pass_data instance and...
898 (make_pass_lower_subreg2): ...new function.
899 * lto-streamer-out.c (pass_ipa_lto_gimple_out): Convert from a global
900 struct to a subclass of ipa_opt_pass_d along with...
901 (pass_data_ipa_lto_gimple_out): ...new pass_data instance and...
902 (make_pass_ipa_lto_gimple_out): ...new function.
903 (pass_ipa_lto_finish_out): Convert from a global struct to a subclass
904 of ipa_opt_pass_d along with...
905 (pass_data_ipa_lto_finish_out): ...new pass_data instance and...
906 (make_pass_ipa_lto_finish_out): ...new function.
907 * mode-switching.c (pass_mode_switching): Convert from a global struct
908 to a subclass of rtl_opt_pass along with...
909 (pass_data_mode_switching): ...new pass_data instance and...
910 (make_pass_mode_switching): ...new function.
911 * modulo-sched.c (pass_sms): Convert from a global struct to a
912 subclass of rtl_opt_pass along with...
913 (pass_data_sms): ...new pass_data instance and...
914 (make_pass_sms): ...new function.
915 * omp-low.c (pass_expand_omp): Convert from a global struct to a
916 subclass of gimple_opt_pass along with...
917 (pass_data_expand_omp): ...new pass_data instance and...
918 (make_pass_expand_omp): ...new function.
919 (pass_lower_omp): Convert from a global struct to a subclass of
920 gimple_opt_pass along with...
921 (pass_data_lower_omp): ...new pass_data instance and...
922 (make_pass_lower_omp): ...new function.
923 (pass_diagnose_omp_blocks): Convert from a global struct to a subclass
924 of gimple_opt_pass along with...
925 (pass_data_diagnose_omp_blocks): ...new pass_data instance and...
926 (make_pass_diagnose_omp_blocks): ...new function.
927 * passes.c (pass_early_local_passes): Convert from a global struct to
928 a subclass of simple_ipa_opt_pass along with...
929 (pass_data_early_local_passes): ...new pass_data instance and...
930 (make_pass_early_local_passes): ...new function.
931 (pass_all_early_optimizations): Convert from a global struct to a
932 subclass of gimple_opt_pass along with...
933 (pass_data_all_early_optimizations): ...new pass_data instance and...
934 (make_pass_all_early_optimizations): ...new function.
935 (pass_all_optimizations): Convert from a global struct to a subclass
936 of gimple_opt_pass along with...
937 (pass_data_all_optimizations): ...new pass_data instance and...
938 (make_pass_all_optimizations): ...new function.
939 (pass_all_optimizations_g): Convert from a global struct to a subclass
940 of gimple_opt_pass along with...
941 (pass_data_all_optimizations_g): ...new pass_data instance and...
942 (make_pass_all_optimizations_g): ...new function.
943 (pass_rest_of_compilation): Convert from a global struct to a subclass
944 of rtl_opt_pass along with...
945 (pass_data_rest_of_compilation): ...new pass_data instance and...
946 (make_pass_rest_of_compilation): ...new function.
947 (pass_postreload): Convert from a global struct to a subclass of
948 rtl_opt_pass along with...
949 (pass_data_postreload): ...new pass_data instance and...
950 (make_pass_postreload): ...new function.
951 * postreload-gcse.c (pass_gcse2): Convert from a global struct to a
952 subclass of rtl_opt_pass along with...
953 (pass_data_gcse2): ...new pass_data instance and...
954 (make_pass_gcse2): ...new function.
955 * postreload.c (pass_postreload_cse): Convert from a global struct to
956 a subclass of rtl_opt_pass along with...
957 (pass_data_postreload_cse): ...new pass_data instance and...
958 (make_pass_postreload_cse): ...new function.
959 * predict.c (pass_profile): Convert from a global struct to a subclass
960 of gimple_opt_pass along with...
961 (pass_data_profile): ...new pass_data instance and...
962 (make_pass_profile): ...new function.
963 (pass_strip_predict_hints): Convert from a global struct to a subclass
964 of gimple_opt_pass along with...
965 (pass_data_strip_predict_hints): ...new pass_data instance and...
966 (make_pass_strip_predict_hints): ...new function.
967 * recog.c (pass_peephole2): Convert from a global struct to a subclass
968 of rtl_opt_pass along with...
969 (pass_data_peephole2): ...new pass_data instance and...
970 (make_pass_peephole2): ...new function.
971 (pass_split_all_insns): Convert from a global struct to a subclass of
972 rtl_opt_pass along with...
973 (pass_data_split_all_insns): ...new pass_data instance and...
974 (make_pass_split_all_insns): ...new function.
975 (pass_split_after_reload): Convert from a global struct to a subclass
976 of rtl_opt_pass along with...
977 (pass_data_split_after_reload): ...new pass_data instance and...
978 (make_pass_split_after_reload): ...new function.
979 (pass_split_before_regstack): Convert from a global struct to a
980 subclass of rtl_opt_pass along with...
981 (pass_data_split_before_regstack): ...new pass_data instance and...
982 (make_pass_split_before_regstack): ...new function.
983 (pass_split_before_sched2): Convert from a global struct to a subclass
984 of rtl_opt_pass along with...
985 (pass_data_split_before_sched2): ...new pass_data instance and...
986 (make_pass_split_before_sched2): ...new function.
987 (pass_split_for_shorten_branches): Convert from a global struct to a
988 subclass of rtl_opt_pass along with...
989 (pass_data_split_for_shorten_branches): ...new pass_data instance
991 (make_pass_split_for_shorten_branches): ...new function.
992 * ree.c (pass_ree): Convert from a global struct to a subclass of
993 rtl_opt_pass along with...
994 (pass_data_ree): ...new pass_data instance and...
995 (make_pass_ree): ...new function.
996 * reg-stack.c (pass_stack_regs): Convert from a global struct to a
997 subclass of rtl_opt_pass along with...
998 (pass_data_stack_regs): ...new pass_data instance and...
999 (make_pass_stack_regs): ...new function.
1000 (pass_stack_regs_run): Convert from a global struct to a subclass of
1001 rtl_opt_pass along with...
1002 (pass_data_stack_regs_run): ...new pass_data instance and...
1003 (make_pass_stack_regs_run): ...new function.
1004 * regcprop.c (pass_cprop_hardreg): Convert from a global struct to a
1005 subclass of rtl_opt_pass along with...
1006 (pass_data_cprop_hardreg): ...new pass_data instance and...
1007 (make_pass_cprop_hardreg): ...new function.
1008 * reginfo.c (pass_reginfo_init): Convert from a global struct to a
1009 subclass of rtl_opt_pass along with...
1010 (pass_data_reginfo_init): ...new pass_data instance and...
1011 (make_pass_reginfo_init): ...new function.
1012 * regmove.c (pass_regmove): Convert from a global struct to a subclass
1013 of rtl_opt_pass along with...
1014 (pass_data_regmove): ...new pass_data instance and...
1015 (make_pass_regmove): ...new function.
1016 * regrename.c (pass_regrename): Convert from a global struct to a
1017 subclass of rtl_opt_pass along with...
1018 (pass_data_regrename): ...new pass_data instance and...
1019 (make_pass_regrename): ...new function.
1020 * reorg.c (pass_delay_slots): Convert from a global struct to a
1021 subclass of rtl_opt_pass along with...
1022 (pass_data_delay_slots): ...new pass_data instance and...
1023 (make_pass_delay_slots): ...new function.
1024 (pass_machine_reorg): Convert from a global struct to a subclass of
1025 rtl_opt_pass along with...
1026 (pass_data_machine_reorg): ...new pass_data instance and...
1027 (make_pass_machine_reorg): ...new function.
1028 * sched-rgn.c (pass_sched): Convert from a global struct to a subclass
1029 of rtl_opt_pass along with...
1030 (pass_data_sched): ...new pass_data instance and...
1031 (make_pass_sched): ...new function.
1032 (pass_sched2): Convert from a global struct to a subclass of
1033 rtl_opt_pass along with...
1034 (pass_data_sched2): ...new pass_data instance and...
1035 (make_pass_sched2): ...new function.
1036 * stack-ptr-mod.c (pass_stack_ptr_mod): Convert from a global struct
1037 to a subclass of rtl_opt_pass along with...
1038 (pass_data_stack_ptr_mod): ...new pass_data instance and...
1039 (make_pass_stack_ptr_mod): ...new function.
1040 * store-motion.c (pass_rtl_store_motion): Convert from a global struct
1041 to a subclass of rtl_opt_pass along with...
1042 (pass_data_rtl_store_motion): ...new pass_data instance and...
1043 (make_pass_rtl_store_motion): ...new function.
1044 * tracer.c (pass_tracer): Convert from a global struct to a subclass
1045 of gimple_opt_pass along with...
1046 (pass_data_tracer): ...new pass_data instance and...
1047 (make_pass_tracer): ...new function.
1048 * trans-mem.c (pass_diagnose_tm_blocks): Convert from a global struct
1049 to a subclass of gimple_opt_pass along with...
1050 (pass_data_diagnose_tm_blocks): ...new pass_data instance and...
1051 (make_pass_diagnose_tm_blocks): ...new function.
1052 (pass_lower_tm): Convert from a global struct to a subclass of
1053 gimple_opt_pass along with...
1054 (pass_data_lower_tm): ...new pass_data instance and...
1055 (make_pass_lower_tm): ...new function.
1056 (pass_tm_init): Convert from a global struct to a subclass of
1057 gimple_opt_pass along with...
1058 (pass_data_tm_init): ...new pass_data instance and...
1059 (make_pass_tm_init): ...new function.
1060 (pass_tm_mark): Convert from a global struct to a subclass of
1061 gimple_opt_pass along with...
1062 (pass_data_tm_mark): ...new pass_data instance and...
1063 (make_pass_tm_mark): ...new function.
1064 (pass_tm_edges): Convert from a global struct to a subclass of
1065 gimple_opt_pass along with...
1066 (pass_data_tm_edges): ...new pass_data instance and...
1067 (make_pass_tm_edges): ...new function.
1068 (pass_tm_memopt): Convert from a global struct to a subclass of
1069 gimple_opt_pass along with...
1070 (pass_data_tm_memopt): ...new pass_data instance and...
1071 (make_pass_tm_memopt): ...new function.
1072 (pass_ipa_tm): Convert from a global struct to a subclass of
1073 simple_ipa_opt_pass along with...
1074 (pass_data_ipa_tm): ...new pass_data instance and...
1075 (make_pass_ipa_tm): ...new function.
1076 * tree-call-cdce.c (pass_call_cdce): Convert from a global struct to a
1077 subclass of gimple_opt_pass along with...
1078 (pass_data_call_cdce): ...new pass_data instance and...
1079 (make_pass_call_cdce): ...new function.
1080 * tree-cfg.c (pass_build_cfg): Convert from a global struct to a
1081 subclass of gimple_opt_pass along with...
1082 (pass_data_build_cfg): ...new pass_data instance and...
1083 (make_pass_build_cfg): ...new function.
1084 (pass_split_crit_edges): Convert from a global struct to a subclass of
1085 gimple_opt_pass along with...
1086 (pass_data_split_crit_edges): ...new pass_data instance and...
1087 (make_pass_split_crit_edges): ...new function.
1088 (pass_warn_function_return): Convert from a global struct to a
1089 subclass of gimple_opt_pass along with...
1090 (pass_data_warn_function_return): ...new pass_data instance and...
1091 (make_pass_warn_function_return): ...new function.
1092 (pass_warn_function_noreturn): Convert from a global struct to a
1093 subclass of gimple_opt_pass along with...
1094 (pass_data_warn_function_noreturn): ...new pass_data instance and...
1095 (make_pass_warn_function_noreturn): ...new function.
1096 (pass_warn_unused_result): Convert from a global struct to a subclass
1097 of gimple_opt_pass along with...
1098 (pass_data_warn_unused_result): ...new pass_data instance and...
1099 (make_pass_warn_unused_result): ...new function.
1100 * tree-cfgcleanup.c (pass_merge_phi): Convert from a global struct to
1101 a subclass of gimple_opt_pass along with...
1102 (pass_data_merge_phi): ...new pass_data instance and...
1103 (make_pass_merge_phi): ...new function.
1104 * tree-complex.c (pass_lower_complex): Convert from a global struct to
1105 a subclass of gimple_opt_pass along with...
1106 (pass_data_lower_complex): ...new pass_data instance and...
1107 (make_pass_lower_complex): ...new function.
1108 (pass_lower_complex_O0): Convert from a global struct to a subclass of
1109 gimple_opt_pass along with...
1110 (pass_data_lower_complex_O0): ...new pass_data instance and...
1111 (make_pass_lower_complex_O0): ...new function.
1112 * tree-eh.c (pass_lower_eh): Convert from a global struct to a
1113 subclass of gimple_opt_pass along with...
1114 (pass_data_lower_eh): ...new pass_data instance and...
1115 (make_pass_lower_eh): ...new function.
1116 (pass_refactor_eh): Convert from a global struct to a subclass of
1117 gimple_opt_pass along with...
1118 (pass_data_refactor_eh): ...new pass_data instance and...
1119 (make_pass_refactor_eh): ...new function.
1120 (pass_lower_resx): Convert from a global struct to a subclass of
1121 gimple_opt_pass along with...
1122 (pass_data_lower_resx): ...new pass_data instance and...
1123 (make_pass_lower_resx): ...new function.
1124 (pass_lower_eh_dispatch): Convert from a global struct to a subclass
1125 of gimple_opt_pass along with...
1126 (pass_data_lower_eh_dispatch): ...new pass_data instance and...
1127 (make_pass_lower_eh_dispatch): ...new function.
1128 (pass_cleanup_eh): Convert from a global struct to a subclass of
1129 gimple_opt_pass along with...
1130 (pass_data_cleanup_eh): ...new pass_data instance and...
1131 (make_pass_cleanup_eh): ...new function.
1132 * tree-emutls.c (pass_ipa_lower_emutls): Convert from a global struct
1133 to a subclass of simple_ipa_opt_pass along with...
1134 (pass_data_ipa_lower_emutls): ...new pass_data instance and...
1135 (make_pass_ipa_lower_emutls): ...new function.
1136 * tree-if-conv.c (pass_if_conversion): Convert from a global struct to
1137 a subclass of gimple_opt_pass along with...
1138 (pass_data_if_conversion): ...new pass_data instance and...
1139 (make_pass_if_conversion): ...new function.
1140 * tree-into-ssa.c (pass_build_ssa): Convert from a global struct to a
1141 subclass of gimple_opt_pass along with...
1142 (pass_data_build_ssa): ...new pass_data instance and...
1143 (make_pass_build_ssa): ...new function.
1144 * tree-loop-distribution.c (pass_loop_distribution): Convert from a
1145 global struct to a subclass of gimple_opt_pass along with...
1146 (pass_data_loop_distribution): ...new pass_data instance and...
1147 (make_pass_loop_distribution): ...new function.
1148 * tree-mudflap.c (pass_mudflap_1): Convert from a global struct to a
1149 subclass of gimple_opt_pass along with...
1150 (pass_data_mudflap_1): ...new pass_data instance and...
1151 (make_pass_mudflap_1): ...new function.
1152 (pass_mudflap_2): Convert from a global struct to a subclass of
1153 gimple_opt_pass along with...
1154 (pass_data_mudflap_2): ...new pass_data instance and...
1155 (make_pass_mudflap_2): ...new function.
1156 * tree-nomudflap.c (pass_mudflap_1): Convert from a global struct to a
1157 subclass of gimple_opt_pass along with...
1158 (pass_data_mudflap_1): ...new pass_data instance and...
1159 (make_pass_mudflap_1): ...new function.
1160 (pass_mudflap_2): Convert from a global struct to a subclass of
1161 gimple_opt_pass along with...
1162 (pass_data_mudflap_2): ...new pass_data instance and...
1163 (make_pass_mudflap_2): ...new function.
1164 * tree-nrv.c (pass_nrv): Convert from a global struct to a subclass of
1165 gimple_opt_pass along with...
1166 (pass_data_nrv): ...new pass_data instance and...
1167 (make_pass_nrv): ...new function.
1168 (pass_return_slot): Convert from a global struct to a subclass of
1169 gimple_opt_pass along with...
1170 (pass_data_return_slot): ...new pass_data instance and...
1171 (make_pass_return_slot): ...new function.
1172 * tree-object-size.c (pass_object_sizes): Convert from a global struct
1173 to a subclass of gimple_opt_pass along with...
1174 (pass_data_object_sizes): ...new pass_data instance and...
1175 (make_pass_object_sizes): ...new function.
1176 * tree-optimize.c (pass_cleanup_cfg_post_optimizing): Convert from a
1177 global struct to a subclass of gimple_opt_pass along with...
1178 (pass_data_cleanup_cfg_post_optimizing): ...new pass_data instance
1180 (make_pass_cleanup_cfg_post_optimizing): ...new function.
1181 (pass_fixup_cfg): Convert from a global struct to a subclass of
1182 gimple_opt_pass along with...
1183 (pass_data_fixup_cfg): ...new pass_data instance and...
1184 (make_pass_fixup_cfg): ...new function.
1185 * tree-pass.h (pass_mudflap_1): Replace declaration with that of...
1186 (make_pass_mudflap_1): ...new function.
1187 (pass_mudflap_2): Replace declaration with that of...
1188 (make_pass_mudflap_2): ...new function.
1189 (pass_asan): Replace declaration with that of...
1190 (make_pass_asan): ...new function.
1191 (pass_asan_O0): Replace declaration with that of...
1192 (make_pass_asan_O0): ...new function.
1193 (pass_tsan): Replace declaration with that of...
1194 (make_pass_tsan): ...new function.
1195 (pass_tsan_O0): Replace declaration with that of...
1196 (make_pass_tsan_O0): ...new function.
1197 (pass_lower_cf): Replace declaration with that of...
1198 (make_pass_lower_cf): ...new function.
1199 (pass_refactor_eh): Replace declaration with that of...
1200 (make_pass_refactor_eh): ...new function.
1201 (pass_lower_eh): Replace declaration with that of...
1202 (make_pass_lower_eh): ...new function.
1203 (pass_lower_eh_dispatch): Replace declaration with that of...
1204 (make_pass_lower_eh_dispatch): ...new function.
1205 (pass_lower_resx): Replace declaration with that of...
1206 (make_pass_lower_resx): ...new function.
1207 (pass_build_cfg): Replace declaration with that of...
1208 (make_pass_build_cfg): ...new function.
1209 (pass_early_tree_profile): Replace declaration with that of...
1210 (make_pass_early_tree_profile): ...new function.
1211 (pass_cleanup_eh): Replace declaration with that of...
1212 (make_pass_cleanup_eh): ...new function.
1213 (pass_sra): Replace declaration with that of...
1214 (make_pass_sra): ...new function.
1215 (pass_sra_early): Replace declaration with that of...
1216 (make_pass_sra_early): ...new function.
1217 (pass_early_ipa_sra): Replace declaration with that of...
1218 (make_pass_early_ipa_sra): ...new function.
1219 (pass_tail_recursion): Replace declaration with that of...
1220 (make_pass_tail_recursion): ...new function.
1221 (pass_tail_calls): Replace declaration with that of...
1222 (make_pass_tail_calls): ...new function.
1223 (pass_tree_loop): Replace declaration with that of...
1224 (make_pass_tree_loop): ...new function.
1225 (pass_tree_loop_init): Replace declaration with that of...
1226 (make_pass_tree_loop_init): ...new function.
1227 (pass_lim): Replace declaration with that of...
1228 (make_pass_lim): ...new function.
1229 (pass_tree_unswitch): Replace declaration with that of...
1230 (make_pass_tree_unswitch): ...new function.
1231 (pass_predcom): Replace declaration with that of...
1232 (make_pass_predcom): ...new function.
1233 (pass_iv_canon): Replace declaration with that of...
1234 (make_pass_iv_canon): ...new function.
1235 (pass_scev_cprop): Replace declaration with that of...
1236 (make_pass_scev_cprop): ...new function.
1237 (pass_empty_loop): Replace declaration with that of...
1238 (make_pass_empty_loop): ...new function.
1239 (pass_record_bounds): Replace declaration with that of...
1240 (make_pass_record_bounds): ...new function.
1241 (pass_graphite): Replace declaration with that of...
1242 (make_pass_graphite): ...new function.
1243 (pass_graphite_transforms): Replace declaration with that of...
1244 (make_pass_graphite_transforms): ...new function.
1245 (pass_if_conversion): Replace declaration with that of...
1246 (make_pass_if_conversion): ...new function.
1247 (pass_loop_distribution): Replace declaration with that of...
1248 (make_pass_loop_distribution): ...new function.
1249 (pass_vectorize): Replace declaration with that of...
1250 (make_pass_vectorize): ...new function.
1251 (pass_slp_vectorize): Replace declaration with that of...
1252 (make_pass_slp_vectorize): ...new function.
1253 (pass_complete_unroll): Replace declaration with that of...
1254 (make_pass_complete_unroll): ...new function.
1255 (pass_complete_unrolli): Replace declaration with that of...
1256 (make_pass_complete_unrolli): ...new function.
1257 (pass_parallelize_loops): Replace declaration with that of...
1258 (make_pass_parallelize_loops): ...new function.
1259 (pass_loop_prefetch): Replace declaration with that of...
1260 (make_pass_loop_prefetch): ...new function.
1261 (pass_iv_optimize): Replace declaration with that of...
1262 (make_pass_iv_optimize): ...new function.
1263 (pass_tree_loop_done): Replace declaration with that of...
1264 (make_pass_tree_loop_done): ...new function.
1265 (pass_ch): Replace declaration with that of...
1266 (make_pass_ch): ...new function.
1267 (pass_ccp): Replace declaration with that of...
1268 (make_pass_ccp): ...new function.
1269 (pass_phi_only_cprop): Replace declaration with that of...
1270 (make_pass_phi_only_cprop): ...new function.
1271 (pass_build_ssa): Replace declaration with that of...
1272 (make_pass_build_ssa): ...new function.
1273 (pass_build_alias): Replace declaration with that of...
1274 (make_pass_build_alias): ...new function.
1275 (pass_build_ealias): Replace declaration with that of...
1276 (make_pass_build_ealias): ...new function.
1277 (pass_dominator): Replace declaration with that of...
1278 (make_pass_dominator): ...new function.
1279 (pass_dce): Replace declaration with that of...
1280 (make_pass_dce): ...new function.
1281 (pass_dce_loop): Replace declaration with that of...
1282 (make_pass_dce_loop): ...new function.
1283 (pass_cd_dce): Replace declaration with that of...
1284 (make_pass_cd_dce): ...new function.
1285 (pass_call_cdce): Replace declaration with that of...
1286 (make_pass_call_cdce): ...new function.
1287 (pass_merge_phi): Replace declaration with that of...
1288 (make_pass_merge_phi): ...new function.
1289 (pass_split_crit_edges): Replace declaration with that of...
1290 (make_pass_split_crit_edges): ...new function.
1291 (pass_pre): Replace declaration with that of...
1292 (make_pass_pre): ...new function.
1293 (pass_profile): Replace declaration with that of...
1294 (make_pass_profile): ...new function.
1295 (pass_strip_predict_hints): Replace declaration with that of...
1296 (make_pass_strip_predict_hints): ...new function.
1297 (pass_lower_complex_O0): Replace declaration with that of...
1298 (make_pass_lower_complex_O0): ...new function.
1299 (pass_lower_complex): Replace declaration with that of...
1300 (make_pass_lower_complex): ...new function.
1301 (pass_lower_vector): Replace declaration with that of...
1302 (make_pass_lower_vector): ...new function.
1303 (pass_lower_vector_ssa): Replace declaration with that of...
1304 (make_pass_lower_vector_ssa): ...new function.
1305 (pass_lower_omp): Replace declaration with that of...
1306 (make_pass_lower_omp): ...new function.
1307 (pass_diagnose_omp_blocks): Replace declaration with that of...
1308 (make_pass_diagnose_omp_blocks): ...new function.
1309 (pass_expand_omp): Replace declaration with that of...
1310 (make_pass_expand_omp): ...new function.
1311 (pass_expand_omp_ssa): Replace declaration with that of...
1312 (make_pass_expand_omp_ssa): ...new function.
1313 (pass_object_sizes): Replace declaration with that of...
1314 (make_pass_object_sizes): ...new function.
1315 (pass_strlen): Replace declaration with that of...
1316 (make_pass_strlen): ...new function.
1317 (pass_fold_builtins): Replace declaration with that of...
1318 (make_pass_fold_builtins): ...new function.
1319 (pass_stdarg): Replace declaration with that of...
1320 (make_pass_stdarg): ...new function.
1321 (pass_early_warn_uninitialized): Replace declaration with that of...
1322 (make_pass_early_warn_uninitialized): ...new function.
1323 (pass_late_warn_uninitialized): Replace declaration with that of...
1324 (make_pass_late_warn_uninitialized): ...new function.
1325 (pass_cse_reciprocals): Replace declaration with that of...
1326 (make_pass_cse_reciprocals): ...new function.
1327 (pass_cse_sincos): Replace declaration with that of...
1328 (make_pass_cse_sincos): ...new function.
1329 (pass_optimize_bswap): Replace declaration with that of...
1330 (make_pass_optimize_bswap): ...new function.
1331 (pass_optimize_widening_mul): Replace declaration with that of...
1332 (make_pass_optimize_widening_mul): ...new function.
1333 (pass_warn_function_return): Replace declaration with that of...
1334 (make_pass_warn_function_return): ...new function.
1335 (pass_warn_function_noreturn): Replace declaration with that of...
1336 (make_pass_warn_function_noreturn): ...new function.
1337 (pass_cselim): Replace declaration with that of...
1338 (make_pass_cselim): ...new function.
1339 (pass_phiopt): Replace declaration with that of...
1340 (make_pass_phiopt): ...new function.
1341 (pass_forwprop): Replace declaration with that of...
1342 (make_pass_forwprop): ...new function.
1343 (pass_phiprop): Replace declaration with that of...
1344 (make_pass_phiprop): ...new function.
1345 (pass_tree_ifcombine): Replace declaration with that of...
1346 (make_pass_tree_ifcombine): ...new function.
1347 (pass_dse): Replace declaration with that of...
1348 (make_pass_dse): ...new function.
1349 (pass_nrv): Replace declaration with that of...
1350 (make_pass_nrv): ...new function.
1351 (pass_rename_ssa_copies): Replace declaration with that of...
1352 (make_pass_rename_ssa_copies): ...new function.
1353 (pass_sink_code): Replace declaration with that of...
1354 (make_pass_sink_code): ...new function.
1355 (pass_fre): Replace declaration with that of...
1356 (make_pass_fre): ...new function.
1357 (pass_check_data_deps): Replace declaration with that of...
1358 (make_pass_check_data_deps): ...new function.
1359 (pass_copy_prop): Replace declaration with that of...
1360 (make_pass_copy_prop): ...new function.
1361 (pass_vrp): Replace declaration with that of...
1362 (make_pass_vrp): ...new function.
1363 (pass_uncprop): Replace declaration with that of...
1364 (make_pass_uncprop): ...new function.
1365 (pass_return_slot): Replace declaration with that of...
1366 (make_pass_return_slot): ...new function.
1367 (pass_reassoc): Replace declaration with that of...
1368 (make_pass_reassoc): ...new function.
1369 (pass_rebuild_cgraph_edges): Replace declaration with that of...
1370 (make_pass_rebuild_cgraph_edges): ...new function.
1371 (pass_remove_cgraph_callee_edges): Replace declaration with that of...
1372 (make_pass_remove_cgraph_callee_edges): ...new function.
1373 (pass_build_cgraph_edges): Replace declaration with that of...
1374 (make_pass_build_cgraph_edges): ...new function.
1375 (pass_local_pure_const): Replace declaration with that of...
1376 (make_pass_local_pure_const): ...new function.
1377 (pass_tracer): Replace declaration with that of...
1378 (make_pass_tracer): ...new function.
1379 (pass_warn_unused_result): Replace declaration with that of...
1380 (make_pass_warn_unused_result): ...new function.
1381 (pass_diagnose_tm_blocks): Replace declaration with that of...
1382 (make_pass_diagnose_tm_blocks): ...new function.
1383 (pass_lower_tm): Replace declaration with that of...
1384 (make_pass_lower_tm): ...new function.
1385 (pass_tm_init): Replace declaration with that of...
1386 (make_pass_tm_init): ...new function.
1387 (pass_tm_mark): Replace declaration with that of...
1388 (make_pass_tm_mark): ...new function.
1389 (pass_tm_memopt): Replace declaration with that of...
1390 (make_pass_tm_memopt): ...new function.
1391 (pass_tm_edges): Replace declaration with that of...
1392 (make_pass_tm_edges): ...new function.
1393 (pass_split_functions): Replace declaration with that of...
1394 (make_pass_split_functions): ...new function.
1395 (pass_feedback_split_functions): Replace declaration with that of...
1396 (make_pass_feedback_split_functions): ...new function.
1397 (pass_strength_reduction): Replace declaration with that of...
1398 (make_pass_strength_reduction): ...new function.
1399 (pass_ipa_lower_emutls): Replace declaration with that of...
1400 (make_pass_ipa_lower_emutls): ...new function.
1401 (pass_ipa_function_and_variable_visibility): Replace declaration with
1403 (make_pass_ipa_function_and_variable_visibility): ...new function.
1404 (pass_ipa_tree_profile): Replace declaration with that of...
1405 (make_pass_ipa_tree_profile): ...new function.
1406 (pass_early_local_passes): Replace declaration with that of...
1407 (make_pass_early_local_passes): ...new function.
1408 (pass_ipa_whole_program_visibility): Replace declaration with that
1410 (make_pass_ipa_whole_program_visibility): ...new function.
1411 (pass_ipa_lto_gimple_out): Replace declaration with that of...
1412 (make_pass_ipa_lto_gimple_out): ...new function.
1413 (pass_ipa_increase_alignment): Replace declaration with that of...
1414 (make_pass_ipa_increase_alignment): ...new function.
1415 (pass_ipa_inline): Replace declaration with that of...
1416 (make_pass_ipa_inline): ...new function.
1417 (pass_ipa_free_lang_data): Replace declaration with that of...
1418 (make_pass_ipa_free_lang_data): ...new function.
1419 (pass_ipa_free_inline_summary): Replace declaration with that of...
1420 (make_pass_ipa_free_inline_summary): ...new function.
1421 (pass_ipa_cp): Replace declaration with that of...
1422 (make_pass_ipa_cp): ...new function.
1423 (pass_ipa_reference): Replace declaration with that of...
1424 (make_pass_ipa_reference): ...new function.
1425 (pass_ipa_pure_const): Replace declaration with that of...
1426 (make_pass_ipa_pure_const): ...new function.
1427 (pass_ipa_pta): Replace declaration with that of...
1428 (make_pass_ipa_pta): ...new function.
1429 (pass_ipa_lto_finish_out): Replace declaration with that of...
1430 (make_pass_ipa_lto_finish_out): ...new function.
1431 (pass_ipa_tm): Replace declaration with that of...
1432 (make_pass_ipa_tm): ...new function.
1433 (pass_ipa_profile): Replace declaration with that of...
1434 (make_pass_ipa_profile): ...new function.
1435 (pass_ipa_cdtor_merge): Replace declaration with that of...
1436 (make_pass_ipa_cdtor_merge): ...new function.
1437 (pass_cleanup_cfg_post_optimizing): Replace declaration with that
1439 (make_pass_cleanup_cfg_post_optimizing): ...new function.
1440 (pass_init_datastructures): Replace declaration with that of...
1441 (make_pass_init_datastructures): ...new function.
1442 (pass_fixup_cfg): Replace declaration with that of...
1443 (make_pass_fixup_cfg): ...new function.
1444 (pass_expand): Replace declaration with that of...
1445 (make_pass_expand): ...new function.
1446 (pass_instantiate_virtual_regs): Replace declaration with that of...
1447 (make_pass_instantiate_virtual_regs): ...new function.
1448 (pass_rtl_fwprop): Replace declaration with that of...
1449 (make_pass_rtl_fwprop): ...new function.
1450 (pass_rtl_fwprop_addr): Replace declaration with that of...
1451 (make_pass_rtl_fwprop_addr): ...new function.
1452 (pass_jump): Replace declaration with that of...
1453 (make_pass_jump): ...new function.
1454 (pass_jump2): Replace declaration with that of...
1455 (make_pass_jump2): ...new function.
1456 (pass_lower_subreg): Replace declaration with that of...
1457 (make_pass_lower_subreg): ...new function.
1458 (pass_cse): Replace declaration with that of...
1459 (make_pass_cse): ...new function.
1460 (pass_fast_rtl_dce): Replace declaration with that of...
1461 (make_pass_fast_rtl_dce): ...new function.
1462 (pass_ud_rtl_dce): Replace declaration with that of...
1463 (make_pass_ud_rtl_dce): ...new function.
1464 (pass_rtl_dce): Replace declaration with that of...
1465 (make_pass_rtl_dce): ...new function.
1466 (pass_rtl_dse1): Replace declaration with that of...
1467 (make_pass_rtl_dse1): ...new function.
1468 (pass_rtl_dse2): Replace declaration with that of...
1469 (make_pass_rtl_dse2): ...new function.
1470 (pass_rtl_dse3): Replace declaration with that of...
1471 (make_pass_rtl_dse3): ...new function.
1472 (pass_rtl_cprop): Replace declaration with that of...
1473 (make_pass_rtl_cprop): ...new function.
1474 (pass_rtl_pre): Replace declaration with that of...
1475 (make_pass_rtl_pre): ...new function.
1476 (pass_rtl_hoist): Replace declaration with that of...
1477 (make_pass_rtl_hoist): ...new function.
1478 (pass_rtl_store_motion): Replace declaration with that of...
1479 (make_pass_rtl_store_motion): ...new function.
1480 (pass_cse_after_global_opts): Replace declaration with that of...
1481 (make_pass_cse_after_global_opts): ...new function.
1482 (pass_rtl_ifcvt): Replace declaration with that of...
1483 (make_pass_rtl_ifcvt): ...new function.
1484 (pass_into_cfg_layout_mode): Replace declaration with that of...
1485 (make_pass_into_cfg_layout_mode): ...new function.
1486 (pass_outof_cfg_layout_mode): Replace declaration with that of...
1487 (make_pass_outof_cfg_layout_mode): ...new function.
1488 (pass_loop2): Replace declaration with that of...
1489 (make_pass_loop2): ...new function.
1490 (pass_rtl_loop_init): Replace declaration with that of...
1491 (make_pass_rtl_loop_init): ...new function.
1492 (pass_rtl_move_loop_invariants): Replace declaration with that of...
1493 (make_pass_rtl_move_loop_invariants): ...new function.
1494 (pass_rtl_unswitch): Replace declaration with that of...
1495 (make_pass_rtl_unswitch): ...new function.
1496 (pass_rtl_unroll_and_peel_loops): Replace declaration with that of...
1497 (make_pass_rtl_unroll_and_peel_loops): ...new function.
1498 (pass_rtl_doloop): Replace declaration with that of...
1499 (make_pass_rtl_doloop): ...new function.
1500 (pass_rtl_loop_done): Replace declaration with that of...
1501 (make_pass_rtl_loop_done): ...new function.
1502 (pass_web): Replace declaration with that of...
1503 (make_pass_web): ...new function.
1504 (pass_cse2): Replace declaration with that of...
1505 (make_pass_cse2): ...new function.
1506 (pass_df_initialize_opt): Replace declaration with that of...
1507 (make_pass_df_initialize_opt): ...new function.
1508 (pass_df_initialize_no_opt): Replace declaration with that of...
1509 (make_pass_df_initialize_no_opt): ...new function.
1510 (pass_reginfo_init): Replace declaration with that of...
1511 (make_pass_reginfo_init): ...new function.
1512 (pass_inc_dec): Replace declaration with that of...
1513 (make_pass_inc_dec): ...new function.
1514 (pass_stack_ptr_mod): Replace declaration with that of...
1515 (make_pass_stack_ptr_mod): ...new function.
1516 (pass_initialize_regs): Replace declaration with that of...
1517 (make_pass_initialize_regs): ...new function.
1518 (pass_combine): Replace declaration with that of...
1519 (make_pass_combine): ...new function.
1520 (pass_if_after_combine): Replace declaration with that of...
1521 (make_pass_if_after_combine): ...new function.
1522 (pass_ree): Replace declaration with that of...
1523 (make_pass_ree): ...new function.
1524 (pass_partition_blocks): Replace declaration with that of...
1525 (make_pass_partition_blocks): ...new function.
1526 (pass_match_asm_constraints): Replace declaration with that of...
1527 (make_pass_match_asm_constraints): ...new function.
1528 (pass_regmove): Replace declaration with that of...
1529 (make_pass_regmove): ...new function.
1530 (pass_split_all_insns): Replace declaration with that of...
1531 (make_pass_split_all_insns): ...new function.
1532 (pass_fast_rtl_byte_dce): Replace declaration with that of...
1533 (make_pass_fast_rtl_byte_dce): ...new function.
1534 (pass_lower_subreg2): Replace declaration with that of...
1535 (make_pass_lower_subreg2): ...new function.
1536 (pass_mode_switching): Replace declaration with that of...
1537 (make_pass_mode_switching): ...new function.
1538 (pass_sms): Replace declaration with that of...
1539 (make_pass_sms): ...new function.
1540 (pass_sched): Replace declaration with that of...
1541 (make_pass_sched): ...new function.
1542 (pass_ira): Replace declaration with that of...
1543 (make_pass_ira): ...new function.
1544 (pass_reload): Replace declaration with that of...
1545 (make_pass_reload): ...new function.
1546 (pass_clean_state): Replace declaration with that of...
1547 (make_pass_clean_state): ...new function.
1548 (pass_branch_prob): Replace declaration with that of...
1549 (make_pass_branch_prob): ...new function.
1550 (pass_value_profile_transformations): Replace declaration with that
1552 (make_pass_value_profile_transformations): ...new function.
1553 (pass_postreload_cse): Replace declaration with that of...
1554 (make_pass_postreload_cse): ...new function.
1555 (pass_gcse2): Replace declaration with that of...
1556 (make_pass_gcse2): ...new function.
1557 (pass_split_after_reload): Replace declaration with that of...
1558 (make_pass_split_after_reload): ...new function.
1559 (pass_branch_target_load_optimize1): Replace declaration with that
1561 (make_pass_branch_target_load_optimize1): ...new function.
1562 (pass_thread_prologue_and_epilogue): Replace declaration with that
1564 (make_pass_thread_prologue_and_epilogue): ...new function.
1565 (pass_stack_adjustments): Replace declaration with that of...
1566 (make_pass_stack_adjustments): ...new function.
1567 (pass_peephole2): Replace declaration with that of...
1568 (make_pass_peephole2): ...new function.
1569 (pass_if_after_reload): Replace declaration with that of...
1570 (make_pass_if_after_reload): ...new function.
1571 (pass_regrename): Replace declaration with that of...
1572 (make_pass_regrename): ...new function.
1573 (pass_cprop_hardreg): Replace declaration with that of...
1574 (make_pass_cprop_hardreg): ...new function.
1575 (pass_reorder_blocks): Replace declaration with that of...
1576 (make_pass_reorder_blocks): ...new function.
1577 (pass_branch_target_load_optimize2): Replace declaration with that
1579 (make_pass_branch_target_load_optimize2): ...new function.
1580 (pass_leaf_regs): Replace declaration with that of...
1581 (make_pass_leaf_regs): ...new function.
1582 (pass_split_before_sched2): Replace declaration with that of...
1583 (make_pass_split_before_sched2): ...new function.
1584 (pass_compare_elim_after_reload): Replace declaration with that of...
1585 (make_pass_compare_elim_after_reload): ...new function.
1586 (pass_sched2): Replace declaration with that of...
1587 (make_pass_sched2): ...new function.
1588 (pass_stack_regs): Replace declaration with that of...
1589 (make_pass_stack_regs): ...new function.
1590 (pass_stack_regs_run): Replace declaration with that of...
1591 (make_pass_stack_regs_run): ...new function.
1592 (pass_df_finish): Replace declaration with that of...
1593 (make_pass_df_finish): ...new function.
1594 (pass_compute_alignments): Replace declaration with that of...
1595 (make_pass_compute_alignments): ...new function.
1596 (pass_duplicate_computed_gotos): Replace declaration with that of...
1597 (make_pass_duplicate_computed_gotos): ...new function.
1598 (pass_variable_tracking): Replace declaration with that of...
1599 (make_pass_variable_tracking): ...new function.
1600 (pass_free_cfg): Replace declaration with that of...
1601 (make_pass_free_cfg): ...new function.
1602 (pass_machine_reorg): Replace declaration with that of...
1603 (make_pass_machine_reorg): ...new function.
1604 (pass_cleanup_barriers): Replace declaration with that of...
1605 (make_pass_cleanup_barriers): ...new function.
1606 (pass_delay_slots): Replace declaration with that of...
1607 (make_pass_delay_slots): ...new function.
1608 (pass_split_for_shorten_branches): Replace declaration with that of...
1609 (make_pass_split_for_shorten_branches): ...new function.
1610 (pass_split_before_regstack): Replace declaration with that of...
1611 (make_pass_split_before_regstack): ...new function.
1612 (pass_convert_to_eh_region_ranges): Replace declaration with that
1614 (make_pass_convert_to_eh_region_ranges): ...new function.
1615 (pass_shorten_branches): Replace declaration with that of...
1616 (make_pass_shorten_branches): ...new function.
1617 (pass_set_nothrow_function_flags): Replace declaration with that of...
1618 (make_pass_set_nothrow_function_flags): ...new function.
1619 (pass_dwarf2_frame): Replace declaration with that of...
1620 (make_pass_dwarf2_frame): ...new function.
1621 (pass_final): Replace declaration with that of...
1622 (make_pass_final): ...new function.
1623 (pass_rtl_seqabstr): Replace declaration with that of...
1624 (make_pass_rtl_seqabstr): ...new function.
1625 (pass_release_ssa_names): Replace declaration with that of...
1626 (make_pass_release_ssa_names): ...new function.
1627 (pass_early_inline): Replace declaration with that of...
1628 (make_pass_early_inline): ...new function.
1629 (pass_inline_parameters): Replace declaration with that of...
1630 (make_pass_inline_parameters): ...new function.
1631 (pass_update_address_taken): Replace declaration with that of...
1632 (make_pass_update_address_taken): ...new function.
1633 (pass_convert_switch): Replace declaration with that of...
1634 (make_pass_convert_switch): ...new function.
1635 * tree-profile.c (pass_ipa_tree_profile): Convert from a global struct
1636 to a subclass of simple_ipa_opt_pass along with...
1637 (pass_data_ipa_tree_profile): ...new pass_data instance and...
1638 (make_pass_ipa_tree_profile): ...new function.
1639 * tree-sra.c (pass_sra_early): Convert from a global struct to a
1640 subclass of gimple_opt_pass along with...
1641 (pass_data_sra_early): ...new pass_data instance and...
1642 (make_pass_sra_early): ...new function.
1643 (pass_sra): Convert from a global struct to a subclass of
1644 gimple_opt_pass along with...
1645 (pass_data_sra): ...new pass_data instance and...
1646 (make_pass_sra): ...new function.
1647 (pass_early_ipa_sra): Convert from a global struct to a subclass of
1648 gimple_opt_pass along with...
1649 (pass_data_early_ipa_sra): ...new pass_data instance and...
1650 (make_pass_early_ipa_sra): ...new function.
1651 * tree-ssa-ccp.c (pass_ccp): Convert from a global struct to a
1652 subclass of gimple_opt_pass along with...
1653 (pass_data_ccp): ...new pass_data instance and...
1654 (make_pass_ccp): ...new function.
1655 (pass_fold_builtins): Convert from a global struct to a subclass of
1656 gimple_opt_pass along with...
1657 (pass_data_fold_builtins): ...new pass_data instance and...
1658 (make_pass_fold_builtins): ...new function.
1659 * tree-ssa-copy.c (pass_copy_prop): Convert from a global struct to a
1660 subclass of gimple_opt_pass along with...
1661 (pass_data_copy_prop): ...new pass_data instance and...
1662 (make_pass_copy_prop): ...new function.
1663 * tree-ssa-copyrename.c (pass_rename_ssa_copies): Convert from a
1664 global struct to a subclass of gimple_opt_pass along with...
1665 (pass_data_rename_ssa_copies): ...new pass_data instance and...
1666 (make_pass_rename_ssa_copies): ...new function.
1667 * tree-ssa-dce.c (pass_dce): Convert from a global struct to a
1668 subclass of gimple_opt_pass along with...
1669 (pass_data_dce): ...new pass_data instance and...
1670 (make_pass_dce): ...new function.
1671 (pass_dce_loop): Convert from a global struct to a subclass of
1672 gimple_opt_pass along with...
1673 (pass_data_dce_loop): ...new pass_data instance and...
1674 (make_pass_dce_loop): ...new function.
1675 (pass_cd_dce): Convert from a global struct to a subclass of
1676 gimple_opt_pass along with...
1677 (pass_data_cd_dce): ...new pass_data instance and...
1678 (make_pass_cd_dce): ...new function.
1679 * tree-ssa-dom.c (pass_dominator): Convert from a global struct to a
1680 subclass of gimple_opt_pass along with...
1681 (pass_data_dominator): ...new pass_data instance and...
1682 (make_pass_dominator): ...new function.
1683 (pass_phi_only_cprop): Convert from a global struct to a subclass of
1684 gimple_opt_pass along with...
1685 (pass_data_phi_only_cprop): ...new pass_data instance and...
1686 (make_pass_phi_only_cprop): ...new function.
1687 * tree-ssa-dse.c (pass_dse): Convert from a global struct to a
1688 subclass of gimple_opt_pass along with...
1689 (pass_data_dse): ...new pass_data instance and...
1690 (make_pass_dse): ...new function.
1691 * tree-ssa-forwprop.c (pass_forwprop): Convert from a global struct to
1692 a subclass of gimple_opt_pass along with...
1693 (pass_data_forwprop): ...new pass_data instance and...
1694 (make_pass_forwprop): ...new function.
1695 * tree-ssa-ifcombine.c (pass_tree_ifcombine): Convert from a global
1696 struct to a subclass of gimple_opt_pass along with...
1697 (pass_data_tree_ifcombine): ...new pass_data instance and...
1698 (make_pass_tree_ifcombine): ...new function.
1699 * tree-ssa-loop-ch.c (pass_ch): Convert from a global struct to a
1700 subclass of gimple_opt_pass along with...
1701 (pass_data_ch): ...new pass_data instance and...
1702 (make_pass_ch): ...new function.
1703 * tree-ssa-loop.c (pass_tree_loop): Convert from a global struct to a
1704 subclass of gimple_opt_pass along with...
1705 (pass_data_tree_loop): ...new pass_data instance and...
1706 (make_pass_tree_loop): ...new function.
1707 (pass_tree_loop_init): Convert from a global struct to a subclass of
1708 gimple_opt_pass along with...
1709 (pass_data_tree_loop_init): ...new pass_data instance and...
1710 (make_pass_tree_loop_init): ...new function.
1711 (pass_lim): Convert from a global struct to a subclass of
1712 gimple_opt_pass along with...
1713 (pass_data_lim): ...new pass_data instance and...
1714 (make_pass_lim): ...new function.
1715 (pass_tree_unswitch): Convert from a global struct to a subclass of
1716 gimple_opt_pass along with...
1717 (pass_data_tree_unswitch): ...new pass_data instance and...
1718 (make_pass_tree_unswitch): ...new function.
1719 (pass_predcom): Convert from a global struct to a subclass of
1720 gimple_opt_pass along with...
1721 (pass_data_predcom): ...new pass_data instance and...
1722 (make_pass_predcom): ...new function.
1723 (pass_vectorize): Convert from a global struct to a subclass of
1724 gimple_opt_pass along with...
1725 (pass_data_vectorize): ...new pass_data instance and...
1726 (make_pass_vectorize): ...new function.
1727 (pass_graphite): Convert from a global struct to a subclass of
1728 gimple_opt_pass along with...
1729 (pass_data_graphite): ...new pass_data instance and...
1730 (make_pass_graphite): ...new function.
1731 (pass_graphite_transforms): Convert from a global struct to a subclass
1732 of gimple_opt_pass along with...
1733 (pass_data_graphite_transforms): ...new pass_data instance and...
1734 (make_pass_graphite_transforms): ...new function.
1735 (pass_check_data_deps): Convert from a global struct to a subclass of
1736 gimple_opt_pass along with...
1737 (pass_data_check_data_deps): ...new pass_data instance and...
1738 (make_pass_check_data_deps): ...new function.
1739 (pass_iv_canon): Convert from a global struct to a subclass of
1740 gimple_opt_pass along with...
1741 (pass_data_iv_canon): ...new pass_data instance and...
1742 (make_pass_iv_canon): ...new function.
1743 (pass_scev_cprop): Convert from a global struct to a subclass of
1744 gimple_opt_pass along with...
1745 (pass_data_scev_cprop): ...new pass_data instance and...
1746 (make_pass_scev_cprop): ...new function.
1747 (pass_record_bounds): Convert from a global struct to a subclass of
1748 gimple_opt_pass along with...
1749 (pass_data_record_bounds): ...new pass_data instance and...
1750 (make_pass_record_bounds): ...new function.
1751 (pass_complete_unroll): Convert from a global struct to a subclass of
1752 gimple_opt_pass along with...
1753 (pass_data_complete_unroll): ...new pass_data instance and...
1754 (make_pass_complete_unroll): ...new function.
1755 (pass_complete_unrolli): Convert from a global struct to a subclass of
1756 gimple_opt_pass along with...
1757 (pass_data_complete_unrolli): ...new pass_data instance and...
1758 (make_pass_complete_unrolli): ...new function.
1759 (pass_parallelize_loops): Convert from a global struct to a subclass
1760 of gimple_opt_pass along with...
1761 (pass_data_parallelize_loops): ...new pass_data instance and...
1762 (make_pass_parallelize_loops): ...new function.
1763 (pass_loop_prefetch): Convert from a global struct to a subclass of
1764 gimple_opt_pass along with...
1765 (pass_data_loop_prefetch): ...new pass_data instance and...
1766 (make_pass_loop_prefetch): ...new function.
1767 (pass_iv_optimize): Convert from a global struct to a subclass of
1768 gimple_opt_pass along with...
1769 (pass_data_iv_optimize): ...new pass_data instance and...
1770 (make_pass_iv_optimize): ...new function.
1771 (pass_tree_loop_done): Convert from a global struct to a subclass of
1772 gimple_opt_pass along with...
1773 (pass_data_tree_loop_done): ...new pass_data instance and...
1774 (make_pass_tree_loop_done): ...new function.
1775 * tree-ssa-math-opts.c (pass_cse_reciprocals): Convert from a global
1776 struct to a subclass of gimple_opt_pass along with...
1777 (pass_data_cse_reciprocals): ...new pass_data instance and...
1778 (make_pass_cse_reciprocals): ...new function.
1779 (pass_cse_sincos): Convert from a global struct to a subclass of
1780 gimple_opt_pass along with...
1781 (pass_data_cse_sincos): ...new pass_data instance and...
1782 (make_pass_cse_sincos): ...new function.
1783 (pass_optimize_bswap): Convert from a global struct to a subclass of
1784 gimple_opt_pass along with...
1785 (pass_data_optimize_bswap): ...new pass_data instance and...
1786 (make_pass_optimize_bswap): ...new function.
1787 (pass_optimize_widening_mul): Convert from a global struct to a
1788 subclass of gimple_opt_pass along with...
1789 (pass_data_optimize_widening_mul): ...new pass_data instance and...
1790 (make_pass_optimize_widening_mul): ...new function.
1791 * tree-ssa-phiopt.c (pass_phiopt): Convert from a global struct to a
1792 subclass of gimple_opt_pass along with...
1793 (pass_data_phiopt): ...new pass_data instance and...
1794 (make_pass_phiopt): ...new function.
1795 (pass_cselim): Convert from a global struct to a subclass of
1796 gimple_opt_pass along with...
1797 (pass_data_cselim): ...new pass_data instance and...
1798 (make_pass_cselim): ...new function.
1799 * tree-ssa-phiprop.c (pass_phiprop): Convert from a global struct to a
1800 subclass of gimple_opt_pass along with...
1801 (pass_data_phiprop): ...new pass_data instance and...
1802 (make_pass_phiprop): ...new function.
1803 * tree-ssa-pre.c (pass_pre): Convert from a global struct to a
1804 subclass of gimple_opt_pass along with...
1805 (pass_data_pre): ...new pass_data instance and...
1806 (make_pass_pre): ...new function.
1807 (pass_fre): Convert from a global struct to a subclass of
1808 gimple_opt_pass along with...
1809 (pass_data_fre): ...new pass_data instance and...
1810 (make_pass_fre): ...new function.
1811 * tree-ssa-reassoc.c (pass_reassoc): Convert from a global struct to a
1812 subclass of gimple_opt_pass along with...
1813 (pass_data_reassoc): ...new pass_data instance and...
1814 (make_pass_reassoc): ...new function.
1815 * tree-ssa-sink.c (pass_sink_code): Convert from a global struct to a
1816 subclass of gimple_opt_pass along with...
1817 (pass_data_sink_code): ...new pass_data instance and...
1818 (make_pass_sink_code): ...new function.
1819 * tree-ssa-strlen.c (pass_strlen): Convert from a global struct to a
1820 subclass of gimple_opt_pass along with...
1821 (pass_data_strlen): ...new pass_data instance and...
1822 (make_pass_strlen): ...new function.
1823 * tree-ssa-structalias.c (pass_build_alias): Convert from a global
1824 struct to a subclass of gimple_opt_pass along with...
1825 (pass_data_build_alias): ...new pass_data instance and...
1826 (make_pass_build_alias): ...new function.
1827 (pass_build_ealias): Convert from a global struct to a subclass of
1828 gimple_opt_pass along with...
1829 (pass_data_build_ealias): ...new pass_data instance and...
1830 (make_pass_build_ealias): ...new function.
1831 (pass_ipa_pta): Convert from a global struct to a subclass of
1832 simple_ipa_opt_pass along with...
1833 (pass_data_ipa_pta): ...new pass_data instance and...
1834 (make_pass_ipa_pta): ...new function.
1835 * tree-ssa-uncprop.c (pass_uncprop): Convert from a global struct to a
1836 subclass of gimple_opt_pass along with...
1837 (pass_data_uncprop): ...new pass_data instance and...
1838 (make_pass_uncprop): ...new function.
1839 * tree-ssa-uninit.c (pass_late_warn_uninitialized): Convert from a
1840 global struct to a subclass of gimple_opt_pass along with...
1841 (pass_data_late_warn_uninitialized): ...new pass_data instance and...
1842 (make_pass_late_warn_uninitialized): ...new function.
1843 * tree-ssa.c (pass_init_datastructures): Convert from a global struct
1844 to a subclass of gimple_opt_pass along with...
1845 (pass_data_init_datastructures): ...new pass_data instance and...
1846 (make_pass_init_datastructures): ...new function.
1847 (pass_early_warn_uninitialized): Convert from a global struct to a
1848 subclass of gimple_opt_pass along with...
1849 (pass_data_early_warn_uninitialized): ...new pass_data instance and...
1850 (make_pass_early_warn_uninitialized): ...new function.
1851 (pass_update_address_taken): Convert from a global struct to a
1852 subclass of gimple_opt_pass along with...
1853 (pass_data_update_address_taken): ...new pass_data instance and...
1854 (make_pass_update_address_taken): ...new function.
1855 * tree-ssanames.c (pass_release_ssa_names): Convert from a global
1856 struct to a subclass of gimple_opt_pass along with...
1857 (pass_data_release_ssa_names): ...new pass_data instance and...
1858 (make_pass_release_ssa_names): ...new function.
1859 * tree-stdarg.c (pass_stdarg): Convert from a global struct to a
1860 subclass of gimple_opt_pass along with...
1861 (pass_data_stdarg): ...new pass_data instance and...
1862 (make_pass_stdarg): ...new function.
1863 * tree-switch-conversion.c (pass_convert_switch): Convert from a
1864 global struct to a subclass of gimple_opt_pass along with...
1865 (pass_data_convert_switch): ...new pass_data instance and...
1866 (make_pass_convert_switch): ...new function.
1867 * tree-tailcall.c (pass_tail_recursion): Convert from a global struct
1868 to a subclass of gimple_opt_pass along with...
1869 (pass_data_tail_recursion): ...new pass_data instance and...
1870 (make_pass_tail_recursion): ...new function.
1871 (pass_tail_calls): Convert from a global struct to a subclass of
1872 gimple_opt_pass along with...
1873 (pass_data_tail_calls): ...new pass_data instance and...
1874 (make_pass_tail_calls): ...new function.
1875 * tree-vect-generic.c (pass_lower_vector): Convert from a global
1876 struct to a subclass of gimple_opt_pass along with...
1877 (pass_data_lower_vector): ...new pass_data instance and...
1878 (make_pass_lower_vector): ...new function.
1879 (pass_lower_vector_ssa): Convert from a global struct to a subclass of
1880 gimple_opt_pass along with...
1881 (pass_data_lower_vector_ssa): ...new pass_data instance and...
1882 (make_pass_lower_vector_ssa): ...new function.
1883 * tree-vectorizer.c (pass_slp_vectorize): Convert from a global struct
1884 to a subclass of gimple_opt_pass along with...
1885 (pass_data_slp_vectorize): ...new pass_data instance and...
1886 (make_pass_slp_vectorize): ...new function.
1887 (pass_ipa_increase_alignment): Convert from a global struct to a
1888 subclass of simple_ipa_opt_pass along with...
1889 (pass_data_ipa_increase_alignment): ...new pass_data instance and...
1890 (make_pass_ipa_increase_alignment): ...new function.
1891 * tree-vrp.c (pass_vrp): Convert from a global struct to a subclass of
1892 gimple_opt_pass along with...
1893 (pass_data_vrp): ...new pass_data instance and...
1894 (make_pass_vrp): ...new function.
1895 * tree.c (pass_ipa_free_lang_data): Convert from a global struct to a
1896 subclass of simple_ipa_opt_pass along with...
1897 (pass_data_ipa_free_lang_data): ...new pass_data instance and...
1898 (make_pass_ipa_free_lang_data): ...new function.
1899 * tsan.c (pass_tsan): Convert from a global struct to a subclass of
1900 gimple_opt_pass along with...
1901 (pass_data_tsan): ...new pass_data instance and...
1902 (make_pass_tsan): ...new function.
1903 (pass_tsan_O0): Convert from a global struct to a subclass of
1904 gimple_opt_pass along with...
1905 (pass_data_tsan_O0): ...new pass_data instance and...
1906 (make_pass_tsan_O0): ...new function.
1907 * var-tracking.c (pass_variable_tracking): Convert from a global
1908 struct to a subclass of rtl_opt_pass along with...
1909 (pass_data_variable_tracking): ...new pass_data instance and...
1910 (make_pass_variable_tracking): ...new function.
1911 * web.c (pass_web): Convert from a global struct to a subclass of
1912 rtl_opt_pass along with...
1913 (pass_data_web): ...new pass_data instance and...
1914 (make_pass_web): ...new function.
1915 * config/epiphany/epiphany.h (pass_mode_switch_use): Replace
1916 declaration with that of...
1917 (make_pass_mode_switch_use): ...new function.
1918 (pass_resolve_sw_modes): Replace declaration with that of...
1919 (make_pass_resolve_sw_modes): ...new function.
1920 * config/epiphany/mode-switch-use.c (pass_mode_switch_use): Convert
1921 from a global struct to a subclass of rtl_opt_pass along with...
1922 (pass_data_mode_switch_use): ...new pass_data instance and...
1923 (make_pass_mode_switch_use): ...new function.
1924 * config/epiphany/resolve-sw-modes.c (pass_resolve_sw_modes): Convert
1925 from a global struct to a subclass of rtl_opt_pass along with...
1926 (pass_data_resolve_sw_modes): ...new pass_data instance and...
1927 (make_pass_resolve_sw_modes): ...new function.
1928 * config/i386/i386.c (pass_insert_vzeroupper): Convert from a global
1929 struct to a subclass of rtl_opt_pass along with...
1930 (pass_data_insert_vzeroupper): ...new pass_data instance and...
1931 (make_pass_insert_vzeroupper): ...new function.
1932 * config/sparc/sparc.c (pass_work_around_errata): Convert from a
1933 global struct to a subclass of rtl_opt_pass along with...
1934 (pass_data_work_around_errata): ...new pass_data instance and...
1935 (make_pass_work_around_errata): ...new function.
1936 * config/mips/mips.c (pass_mips_machine_reorg2): Convert from a global
1937 struct to a subclass of rtl_opt_pass along with...
1938 (pass_data_mips_machine_reorg2): ...new pass_data instance and...
1939 (make_pass_mips_machine_reorg2): ...new function.
1941 2013-08-05 David Malcolm <dmalcolm@redhat.com>
1943 * passes.c (pass_manager::operator new): New.
1945 2013-08-05 David Malcolm <dmalcolm@redhat.com>
1947 Handwritten part of conversion of passes to C++ classes.
1949 * Makefile.in (PASS_MANAGER_H): Add dep on pass-instances.def.
1950 (toplev.o): Add dep on PASS_MANAGER_H.
1951 * cgraphunit.c (cgraph_process_new_functions): Rework invocation
1952 of early local pases to reflect this moving from a global to a
1953 member of gcc::pass_manager.
1954 (cgraph_add_new_function): Likewise.
1955 * lto-cgraph.c (lto_output_node): Update for conversion of
1956 struct ipa_opt_pass_d to a C++ subclass of opt_pass.
1957 * passes.c (opt_pass::clone): New.
1958 (opt_pass::gate): New.
1959 (opt_pass::execute): New.
1960 (opt_pass::opt_pass): New.
1961 (pass_manager::execute_early_local_passes): New.
1962 (pass_manager::execute_pass_mode_switching): new.
1963 (finish_optimization_passes): Convert to...
1964 (pass_manager::finish_optimization_passes): ...this.
1965 (finish_optimization_passes): Update for conversion of passes to
1967 (register_dump_files_1): Use has_gate since we cannot portably
1968 check a vtable entry against NULL.
1969 (dump_one_pass): Likewise.
1970 (ipa_write_summaries_2): Likewise.
1971 (ipa_write_optimization_summaries_1): Likewise.
1972 (ipa_read_summaries_1): Likewise.
1973 (ipa_read_optimization_summaries_1): Likewise.
1974 (execute_ipa_stmt_fixups): Likewise.
1975 (pass_manager::pass_manager): Rewrite pass-creation, invoking
1976 pass-creation functions rather than wiring up globals, and
1977 storing the results in fields of pass_manager generated using
1979 (pass_manager::dump_profile_report): Update for conversion of
1980 passes to C++ classes.
1981 (pass_manager::execute_ipa_summary_passes): Likewise.
1982 (execute_one_ipa_transform_pass): Likewise.
1983 (execute_one_pass): Use has_gate and has_execute since we cannot
1984 portably check a vtable entry against NULL.
1985 * pass_manager.h (pass_manager::finish_optimization_passes): New.
1986 (pass_manager): Use pass-instances.def to add fields for the
1987 various pass instances.
1988 * toplev.c (finalize): Update for move of
1989 finish_optimization_passes to a method of gcc::pass_manager.
1990 * toplev.h (finish_optimization_passes): Move to method of class
1992 * tree-pass.h (struct pass_data): New.
1993 (opt_pass): Convert to C++ class, make it a subclass of pass_data.
1994 (opt_pass::gate): Convert to virtual function.
1995 (opt_pass::~opt_pass): New.
1996 (opt_pass::clone): New.
1997 (opt_pass::execute): Convert to virtual function.
1998 (opt_pass::opt_pass): New.
1999 (opt_pass::ctxt_): new.
2000 (gimple_opt_pass): Convert to subclass of opt_pass.
2001 (gimple_opt_pass::gimple_opt_pass): New.
2002 (rtl_opt_pass): Convert to subclass of opt_pass.
2003 (rtl_opt_pass::rtl_opt_pass): New.
2004 (ipa_opt_pass_d): Convert to subclass of opt_pass.
2005 (ipa_opt_pass_d::ipa_opt_pass_d): New.
2006 (simple_ipa_opt_pass): Convert to subclass of opt_pass.
2007 (simple_ipa_opt_pass::simple_ipa_opt_pass): New.
2008 * config/i386/i386.c (rest_of_handle_insert_vzeroupper): Rework
2009 invocation of pass_mode_switching to reflect this moving from a
2010 global to a member of gcc::pass_manager.
2011 (ix86_option_override): Rework how pass_insert_vzeroupper is
2012 added to the pass_manager to reflect autogenerated changes.
2013 * config/i386/t-i386 (i386.o) Add deps on CONTEXT_H and PASS_MANAGER_H.
2015 2013-08-05 Richard Earnshaw <rearnsha@arm.com>
2017 PR rtl-optimization/57708
2018 * recog.c (peep2_find_free_register): Validate all regs in a
2021 2013-08-05 Jan Hubicka <jh@suse.cz>
2024 * cgraph.c (verify_cgraph_node): Accept local flags from other
2026 * ipa.c (symtab_remove_unreachable_nodes): Do not clear local flag.
2027 (function_and_variable_visibility): Likewise.
2028 * trans-mem.c (ipa_tm_create_version): TM versions are not local.
2030 2013-08-05 Gabriel Dos Reis <gdr@integrable-solutions.net>
2032 * graph.c (init_graph_slim_pretty_print): Remove.
2033 (print_graph_cfg): Do not call it. Use local pretty printer.
2034 (start_graph_dump): Likewise.
2036 2013-08-05 Gabriel Dos Reis <gdr@integrable-solutions.net>
2038 * gimple-pretty-print.c (buffer): Remove.
2039 (initialized): Likewise.
2040 (maybe_init_pretty_print): Likewise.
2041 (print_gimple_stmt): Do not call it. Use non-static local
2042 pretty_printer variable.
2043 (print_gimple_expr): Likewise.
2044 (print_gimple_seq): Likewise.
2045 (gimple_dump_bb): Likewise.
2047 2013-08-05 Gabriel Dos Reis <gdr@integrable-solutions.net>
2049 * asan.c (asan_pp): Remove.
2050 (asan_pp_initialized): Likewise.
2051 (asan_pp_initialize): Likewise.
2052 (asan_pp_string): Take a pretty_printer parameter. Adjust callers.
2053 (asan_emit_stack_protection): Tidy. Use local pretty printer.
2054 (asan_add_global): Likewise.
2056 2013-08-04 Gabriel Dos Reis <gdr@integrable-solutions.net>
2058 * pretty-print.h (pp_base): Remove. Adjust dependent macros.
2059 * diagnostic.h (diagnostic_flush_buffer): Adjust.
2060 * pretty-print.c (pp_formatted_text_data): Likewise.
2061 (pp_indent): Rename from pp_base_indent.
2062 (pp_format): Rename from pp_base_format.
2063 (pp_output_formatted_text): Rename from pp_base_output_formatted_text.
2064 (pp_format_verbatim): Rename from pp_base_format_verbatim.
2065 (pp_flush): Rename from pp_base_flush.
2066 (pp_set_line_maximum_length): Rename from
2067 pp_base_set_line_maximum_length.
2068 (pp_clear_output_area): Rename from pp_base_clear_output_area.
2069 (pp_set_prefix): Rename from pp_base_set_prefix.
2070 (pp_destroy_prefix): Rename from pp_base_destroy_prefix.
2071 (pp_emit_prefix): Rename from pp_base_emit_prefix.
2072 (pp_append_text): Rename from pp_base_append_text.
2073 (pp_formatted_text): Rename from pp_base_formatted_text.
2074 (pp_last_position_in_text): Rename from pp_base_last_position_in_text.
2075 (pp_remaining_character_count_for_line): Rename from
2076 pp_base_remaining_character_count_for_line.
2077 (pp_newline): Rename from pp_base_newline.
2078 (pp_character): Rename from pp_base_character.
2079 (pp_string): Rename from pp_base_string.
2080 (pp_maybe_space): Rename from pp_base_maybe_space.
2081 * asan.c (asan_pp_string): Adjust.
2082 (asan_emit_stack_protection): Likewise.
2083 (asan_add_global): Likewise.
2084 * sched-vis.c (str_pattern_slim): Adjust pretty printer function call.
2085 * tree-mudflap.c (mf_varname_tree): Likewise.
2086 * tree-pretty-print.c (pp_tree_identifier): Rename from
2087 pp_base_tree_identifier.
2088 * tree-pretty-print.h (pp_tree_identifier): Remove macro definition.
2089 Declare as function.
2091 2013-08-03 Gabriel Dos Reis <gdr@integrable-solutions.net>
2093 * pretty-print.h (pp_bar_bar): New.
2094 (pp_ampersand_ampersand): Likewise.
2095 (pp_less_equal): Likewise.
2096 (pp_greater_equal): Likewise.
2097 * gimple-pretty-print.c (dump_ternary_rhs): Use specialized pretty
2098 printer functions instead of pp_string or operators and punctuators.
2099 (dump_gimple_call): Likewise.
2100 (dump_gimple_omp_for): Likewise.
2101 (dump_gimple_transaction): Likewise.
2102 (dump_gimple_phi): Likewise.
2103 (pp_gimple_stmt_1): Likewise.
2104 * sched-vis.c (print_insn): Likewise.
2105 * tree-mudflap.c (mf_varname_tree): Likewise.
2106 * tree-pretty-print.c (dump_block_node): Likewise.
2107 (dump_generic_node): Likewise.
2109 2013-08-02 Jan Hubicka <jh@suse.cz>
2111 * lto-cgraph.c (compute_ltrans_boundary): Add abstract origins into
2113 * lto-streamer-out.c (tree_is_indexable): Results decls and
2114 parm decls are not indexable.
2115 (DFS_write_tree_body): Do not follow args and results.
2116 (hash_tree): Likewise.
2117 (output_functions): Rearrange so struct function is needed
2118 only when real body is output; be able to also ouptut abstract
2119 functions; output DECL_ARGUMENTS and DECL_RESULT.
2120 (lto_output): When not in WPA, ale store abstract functions.
2121 (write_symbol): Do not care about RESULT_DECL.
2122 (output_symbol_p): Handle correctly sbtract decls.
2123 * lto-streamer-in.c (input_function): Rearrange so struct
2124 function can be NULL at entry; allow streaming of
2125 functions w/o body; store DECL_ARGUMENTS and DECL_RESULT.
2126 * ipa.c (symtab_remove_unreachable_nodes): Silence confused
2127 sanity check during LTO.
2128 * tree-streamer-out.c (write_ts_decl_non_common_tree_pointers): Skip
2129 RESULT_DECl and DECL_ARGUMENTS.
2130 * tree-streamer-in.c (lto_input_ts_decl_non_common_tree_pointers):
2133 2013-08-03 Gabriel Dos Reis <gdr@integrable-solutions.net>
2135 * pretty-print.h (pp_underscore): New.
2137 * gimple-pretty-print.c (dump_unary_rhs): Use specialized pretty
2138 printer functions instead of pp_character.
2139 (dump_binary_rhs): Likewise.
2140 (dump_ternary_rhs): Likewise.
2141 (dump_gimple_call_args): Likewise.
2142 (pp_points_to_solution): Likewise.
2143 (dump_gimple_call): Likewise.
2144 (dump_gimple_switch): Likewise.
2145 (dump_gimple_cond): Likewise.
2146 (dump_gimple_bind): Likewise.
2147 (dump_gimple_try): Likewise.
2148 (dump_gimple_omp_for): Likewise.
2149 (dump_gimple_omp_continue): Likewise.
2150 (dump_gimple_omp_single): Likewise.
2151 (dump_gimple_omp_sections): Likewise.
2152 (dump_gimple_omp_block): Likewise.
2153 (dump_gimple_omp_critical): Likewise.
2154 (dump_gimple_transaction): Likewise.
2155 (dump_gimple_asm): Likewise.
2156 (dump_gimple_phi): Likewise.
2157 (dump_gimple_omp_parallel): Likewise.
2158 (dump_gimple_omp_task): Likewise.
2159 (dump_gimple_omp_atomic_load): Likewise.
2160 (dump_gimple_omp_atomic_store): Likewise.
2161 (dump_gimple_mem_ops): Likewise.
2162 (pp_gimple_stmt_1): Likewise.
2163 (pp_cfg_jump): Likewise.
2164 (dump_implicit_edges): Likewise.
2165 (gimple_dump_bb_for_graph): Likewise.
2166 * graph.c (draw_cfg_node): Likewise.
2167 * langhooks.c (lhd_print_error_function): Likewise.
2168 * sched-vis.c (print_exp): Likewise.
2169 (print_value): Likewise.
2170 (print_pattern): Likewise.
2171 (print_insn): Likewise.
2172 (rtl_dump_bb_for_graph): Likewise.
2173 * tree-pretty-print.c (dump_function_declaration): Likewise.
2174 (dump_array_domain): Likewise.
2175 (dump_omp_clause): Likewise.
2176 (dump_location): Likewise.
2177 (dump_generic_node): Likewise.
2178 (print_struct_decl): Likewise.
2179 * diagnostic.c (diagnostic_show_locus): Use pp_space.
2181 2013-08-03 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
2183 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Update
2184 candidate table when replacing a candidate statement.
2185 (replace_rhs_if_not_dup): Likewise.
2186 (replace_one_candidate): Likewise.
2188 2013-08-02 Jan Hubicka <jh@suse.cz>
2189 Martin Liska <marxin.liska@gmail.com>
2191 * cgraphunit.c (add_new_function): Fix logic when adding from
2193 (assemble_thunk): Rename to ...
2194 (expand_thunk); .. this one; export; get it working with
2195 general functions; make produced gimple valid.
2196 * cgraph.h (expand_thunk): Declare.
2198 2013-08-02 Jan Hubicka <jh@suse.cz>
2200 * ipa-cp.c (gather_context_independent_values): Use
2201 ipa_get_param_move_cost.
2202 (get_replacement_map): Remove PARAM; move parameter folding
2204 (create_specialized_node): Update.
2205 * ipa-prop.c (ipa_populate_param_decls): Do not look for origins;
2206 assert that we have gimple body; update move_cost.
2207 (count_formal_params): Assert that we have gimple body.
2208 (ipa_dump_param): New function.
2209 (ipa_alloc_node_params): Break out from ...
2210 (ipa_initialize_node_params): ... here.
2211 (ipa_get_vector_of_formal_parms): ICE when used in WPA.
2212 (ipa_write_node_info): Stream move costs.
2213 (ipa_read_node_info): Read move costs.
2214 (ipa_update_after_lto_read): Do not recompute node params.
2215 * ipa-prop.h (ipa_param_descriptor): Add move_cost.
2216 (ipa_get_param): Check we are not in WPA.
2217 (ipa_get_param_move_cost): New.
2218 * tree-inline.c (tree_function_versioning): Fold replacement as needed.
2219 * ipa-inline-analysis.c (inline_node_duplication_hook): Expect only
2220 parm numbers to be present.
2222 2013-08-02 Vladimir Makarov <vmakarov@redhat.com>
2224 PR rtl-optimization/58048
2225 * lra-constraints.c (process_alt_operands): Don't check asm
2226 operand on register.
2228 2013-08-02 Eric Botcazou <ebotcazou@adacore.com>
2230 * config/sparc/sparc.c (sparc_emit_membar_for_model) <SMM_TSO>: Add
2231 the implied StoreLoad barrier for atomic operations if before.
2233 2013-08-02 Jan Hubicka <jh@suse.cz>
2234 Martin Liska <marxin.liska@gmail.com>
2236 * cgraph.c (cgraph_function_body_availability): Do not check
2238 * cgraph.h (symtab_for_node_and_aliases, symtab_nonoverwritable_alias,
2239 symtab_node_availability): Declare.
2240 * ipa.c (can_replace_by_local_alias): New.
2241 (function_and_variable_visibility): Use it.
2242 * symtab.c (symtab_for_node_and_aliases,
2243 symtab_nonoverwritable_alias_1, symtab_nonoverwritable_alias): New.
2245 2013-08-02 Vladimir Makarov <vmakarov@redhat.com>
2247 PR rtl-optimization/57963
2248 * lra-constraints.c (reverse_equiv_p, contains_reloaded_insn_p): New.
2249 (lra_constraints): Use them.
2251 2013-08-02 Sofiane Naci <sofiane.naci@arm.com>
2253 * config/arm/types.md (define_attr "type"): Add "load_acq"
2255 * config/arm/cortex-a53.md (cortex_a53_load1): Update for attribute
2257 (cortex_a53_store1): Likewise.
2259 2013-08-01 Jan Hubicka <jh@suse.cz>
2261 * ipa.c (symtab_remove_unreachable_nodes): Nodes in other
2262 partitions are not needed.
2264 2013-08-01 Uros Bizjak <ubizjak@gmail.com>
2266 * config/i386/i386.h (MAYBE_NON_Q_CLASS_P): New.
2267 * config/i386/i386.c (ix86_secondary_reload): Use INTEGER_CLASS_P and
2268 MAYBE_NON_Q_CLASS_P where appropriate.
2270 2013-08-01 Jan Hubicka <jh@suse.cz>
2272 * cgraph.h (release_function_body): Declare.
2273 * tree.c (free_lang_data_in_decl): Free, parameters and return values
2274 of unused delcarations.
2276 2013-08-01 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2278 * config/arm/arm.md (minmax_arithsi_non_canon): Emit canonical
2279 RTL form when subtracting a constant.
2281 2013-08-01 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2283 * config/arm/arm.md (peepholes for eq (reg1) (reg2/imm)):
2284 Generate canonical plus rtx with negated immediate instead of minus
2286 * config/arm/arm.c (thumb2_reorg): Handle ADCS <Rd>, <Rn> case.
2288 2013-08-01 Jan Hubicka <jh@suse.cz>
2290 * cgraph.c (cgraph_release_function_body): Use used_as_abstract_origin.
2291 (cgraph_release_function_body): Likewise.
2292 (cgraph_can_remove_if_no_direct_calls_p): Likewise.
2293 * cgraph.h (cgrpah_node): Rename abstract_and_needed
2294 to used_as_abstract_origin.
2295 * tree-inline-transfrom.c (can_remove_node_now_p_1): Do not remove
2296 symbols used as abstract origins.
2297 * cgraphunit.c (analyze_functions): Update.
2298 * ipa.c (symtab_remove_unreachable_nodes): Recompute
2299 used_as_abstract_origin.
2300 * tree-inline.c (tree_function_versioning): Update
2301 used_as_abstract_origin; be ready for DECL_RESULT and
2302 DECL_ARGUMENTS to be NULL.
2304 * lto-symtab.c (lto_symtab_merge_symbols): Merge duplicated nodes
2305 for abstract functions.
2306 * cgraph.h (symtab_real_symbol_p): Abstract declarations are not
2309 2013-08-01 Jan Hubicka <jh@suse.cz>
2311 * profile.c (compute_value_histograms): Fix thinko.
2313 2013-08-01 Sofiane Naci <sofiane.naci@arm.com>
2315 * config.gcc (aarch64*-*-*): Add aarch-common.o to extra_objs. Add
2316 aarch-common-protos.h to extra_headers.
2317 (aarch64*-*-*): Add arm/aarch-common-protos.h to tm_p_file.
2318 * config/aarch64/aarch64.md: Include "../arm/cortex-a53.md".
2319 * config/aarch64/t-aarch64 (aarch-common.o): Define.
2321 2013-08-01 Sofiane Naci <sofiane.naci@arm.com>
2323 * config/aarch64/aarch64.md (define_attr "type"): Delete.
2324 Include "../arm/types.md". Define "type" attribute for all patterns.
2325 * config/aarch64/aarch64-simd.md (move_lo_quad_<mode>): Update for
2328 2013-07-31 Michael Meissner <meissner@linux.vnet.ibm.com>
2330 * config/rs6000/predicates.md (fusion_gpr_addis): New predicates
2331 to support power8 load fusion.
2332 (fusion_gpr_mem_load): Likewise.
2334 * config/rs6000/rs6000-modes.def (PTImode): Update a comment.
2336 * config/rs6000/rs6000-protos.h (fusion_gpr_load_p): New
2337 declarations for power8 load fusion.
2338 (emit_fusion_gpr_load): Likewise.
2340 * config/rs6000/rs6000.c (rs6000_option_override_internal): If
2341 tuning for power8, turn on fusion mode by default. Turn on sign
2342 extending fusion mode if normal fusion mode is on, and we are at
2344 (fusion_gpr_load_p): New function, return true if we can fuse an
2345 addis instruction with a dependent load to a GPR.
2346 (emit_fusion_gpr_load): Emit the instructions for power8 load
2349 * config/rs6000/vsx.md (VSX_M2): New iterator for fusion peepholes.
2350 (VSX load fusion peepholes): New peepholes to fuse together an
2351 addi instruction with a VSX load instruction.
2353 * config/rs6000/rs6000.md (GPR load fusion peepholes): New
2354 peepholes to fuse an addis instruction with a load to a GPR base
2355 register. If we are supporting sign extending fusions, convert
2356 sign extending loads to zero extending loads and add an explicit
2359 2013-07-31 Sofiane Naci <sofiane.naci@arm.com>
2361 * config.gcc (arm*-*-*): Add aarch-common.o to extra_objs. Add
2362 aarch-common-protos.h to extra_headers.
2363 (arm*-*-*): Add arm/aarch-common-protos.h to tm_p_file.
2364 * config/arm/arm.c (arm_early_load_addr_dep): Move from here to ...
2365 (arm_early_store_addr_dep): Likewise.
2366 (arm_no_early_alu_shift_dep): Likewise.
2367 (arm_no_early_alu_shift_value_dep): Likewise.
2368 (arm_no_early_mul_dep): Likewise.
2369 (arm_no_early_store_addr_dep): Likewise.
2370 (arm_mac_accumulator_is_mul_result): Likewise.
2371 (arm_mac_accumulator_is_result): Likewise.
2372 * config/arm/aarch-common.c: ... here. New file.
2373 * config/arm/arm-protos.h (arm_early_load_addr_dep): Move from
2375 (arm_early_store_addr_dep): Likewise.
2376 (arm_no_early_alu_shift_dep): Likewise.
2377 (arm_no_early_alu_shift_value_dep): Likewise.
2378 (arm_no_early_mul_dep): Likewise.
2379 (arm_no_early_store_addr_dep): Likewise.
2380 (arm_mac_accumulator_is_mul_result): Likewise.
2381 (arm_mac_accumulator_is_result): Likewise.
2382 * config/arm/aarch-common-protos.h: ... here. New file.
2383 * config/arm/t-arm (aarch-common.o): Define.
2385 2013-07-31 Sofiane Naci <sofiane.naci@arm.com>
2387 * config/arm/arm.md: Include new file "types.md".
2388 (define_attr "type"): Move from here to ...
2389 (define_attr "mul32"): Likewise.
2390 (define_attr "mul64"): Likewise.
2391 * config/arm/types.md: ... here. New file.
2393 2013-07-31 Sebastian Huber <sebastian.huber@embedded-brains.de>
2395 * config.gcc (*-*-rtems*): Use __cxa_atexit by default.
2396 * config/rs6000/rtems.h (TARGET_LIBGCC_SDATA_SECTION): Define.
2398 2013-07-31 Jan-Benedict Glaw <jbglaw@lug-owl.de>
2400 * gen-pass-instances.awk: Fix offset of substr().
2402 2013-07-31 David Malcolm <dmalcolm@redhat.com>
2404 * Makefile.in (pass-instances.def): New.
2405 (passes.o): Replace dependency on passes.def with one on
2408 * gen-pass-instances.awk: New.
2410 * passes.c (pass_manager::pass_manager): Use pass-instances.def
2411 rather than passes.def, updating local definition of NEXT_PASS
2412 macro to add an extra NUM parameter (currently unused).
2414 2013-07-30 David Malcolm <dmalcolm@redhat.com>
2416 * Makefile.in (PASS_MANAGER_H): New.
2417 (lto-cgraph.o): Depend on CONTEXT_H and PASS_MANAGER_H.
2418 (passes.o): Likewise.
2419 (statistics.o): Likewise.
2420 (cgraphunit.o): Likewise.
2421 (context.o): Depend on PASS_MANAGER_H.
2423 * pass_manager.h: New.
2425 * cgraphunit.c (cgraph_add_new_function): Update for moves
2426 of globals to fields of pass_manager.
2427 (analyze_function): Likewise.
2428 (expand_function): Likewise.
2429 (ipa_passes): Likewise.
2430 (compile): Likewise.
2432 * context.c (context::context): New.
2433 * context.h (context::context): New.
2434 (context::get_passes): New.
2435 (context::passes_): New.
2437 * lto-cgraph.c (input_node): Update for moves of globals to
2438 fields of pass_manager.
2440 * passes.c (all_passes): Remove, in favor of a field of the
2441 same name within the new class pass_manager.
2442 (all_small_ipa_passes): Likewise.
2443 (all_lowering_passes): Likewise.
2444 (all_regular_ipa_passes): Likewise.
2445 (all_late_ipa_passes): Likewise.
2446 (all_lto_gen_passes): Likewise.
2447 (passes_by_id): Likewise.
2448 (passes_by_id_size): Likewise.
2449 (gcc_pass_lists): Remove, in favor of "pass_lists" field within
2450 the new class pass_manager.
2451 (set_pass_for_id): Convert to...
2452 (pass_manager::set_pass_for_id): ...method.
2453 (get_pass_for_id): Convert to...
2454 (pass_manager::get_pass_for_id): ...method.
2455 (register_one_dump_file): Move body of implementation into...
2456 (pass_manager::register_one_dump_file): ...here.
2457 (register_dump_files_1): Convert to...
2458 (pass_manager::register_dump_files_1): ...method.
2459 (register_dump_files): Convert to...
2460 (pass_manager::register_dump_files): ...method.
2461 (create_pass_tab): Update for moves of globals to fields of
2463 (dump_passes): Move body of implementation into...
2464 (pass_manager::dump_passes): ...here.
2465 (register_pass): Move body of implementation into...
2466 (pass_manager::register_pass): ...here.
2467 (init_optimization_passes): Convert into...
2468 (pass_manager::pass_manager): ...constructor for new
2469 pass_manager class, and initialize the pass_lists array.
2470 (check_profile_consistency): Update for moves of globals to
2471 fields of pass_manager.
2472 (dump_profile_report): Move body of implementation into...
2473 (pass_manager::dump_profile_report): ...here.
2474 (ipa_write_summaries_1): Update for moves of pass lists from
2475 being globals to fields of pass_manager.
2476 (ipa_write_optimization_summaries): Likewise.
2477 (ipa_read_summaries): Likewise.
2478 (ipa_read_optimization_summaries): Likewise.
2479 (execute_all_ipa_stmt_fixups): Likewise.
2481 * statistics.c (statistics_fini): Update for moves of globals to
2482 fields of pass_manager.
2484 * toplev.c (general_init): Replace call to
2485 init_optimization_passes with construction of the pass_manager
2488 * tree-pass.h (all_passes): Remove, in favor of a field of the
2489 same name within the new class pass_manager.
2490 (all_small_ipa_passes): Likewise.
2491 (all_lowering_passes): Likewise.
2492 (all_regular_ipa_passes): Likewise.
2493 (all_lto_gen_passes): Likewise.
2494 (all_late_ipa_passes): Likewise.
2495 (passes_by_id): Likewise.
2496 (passes_by_id_size): Likewise.
2497 (gcc_pass_lists): Remove, in favor of "pass_lists" field within
2498 the new class pass_manager.
2499 (get_pass_for_id): Remove.
2501 2013-07-30 Richard Earnshaw <rearnsha@arm.com>
2503 * config.gcc (arm): Require 64-bit host-wide-int for all ARM target
2506 2013-07-30 Richard Earnshaw <rearnsha@arm.com>
2508 * arm.md (mulhi3): New expand pattern.
2510 2013-07-30 Jan Hubicka <jh@suse.cz>
2511 Martin Liska <marxin.liska@gmail.com>
2513 * profile.c (compute_value_histograms): Do not ICE when
2514 there is mismatch only on some counters.
2516 2013-07-30 Zhenqiang Chen <zhenqiang.chen@linaro.org>
2518 PR rtl-optimization/57637
2519 * function.c (move_insn_for_shrink_wrap): Also check the
2520 GEN set of the LIVE problem for the liveness analysis
2521 if it exists, otherwise give up.
2523 2013-07-29 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
2525 PR tree-optimization/57993
2526 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Record
2527 replaced statement in the candidate table.
2528 (phi_add_costs): Return infinite cost when the hidden basis does
2529 not dominate all phis on which the candidate is dependent.
2530 (replace_one_candidate): Record replaced statement in the
2533 2013-07-29 Joern Rennecke <joern.rennecke@embecosm.com>
2535 * config/epiphany/epiphany.md (*isub_i+2): New peephole.
2536 (ashlv2si3): New expander.
2537 (*ashlv2si3_i): New define_insn_and_split.
2538 * predicates.md (float_operation): Allow patterns with three
2541 PR rtl-optimization/58021
2542 * mode-switching.c (create_pre_exit): Always split off preceding
2543 insns if we are not at the basic block head.
2545 2013-07-29 Maciej W. Rozycki <macro@codesourcery.com>
2547 * config/mips/linux.h (GLIBC_DYNAMIC_LINKER): Handle `-mnan=2008'.
2548 (UCLIBC_DYNAMIC_LINKER): New macro.
2549 * config/mips/linux64.h (GLIBC_DYNAMIC_LINKER32): Handle
2551 (GLIBC_DYNAMIC_LINKER64, GLIBC_DYNAMIC_LINKERN32): Likewise.
2552 (UCLIBC_DYNAMIC_LINKER32): Undefine macro first. Handle
2554 (UCLIBC_DYNAMIC_LINKER64): Redefine macro.
2555 (UCLIBC_DYNAMIC_LINKERN32): Likewise.
2556 * config/mips/mips-modes.def: Remove RESET_FLOAT_FORMAT calls
2557 for SF and DF modes. Use ieee_quad_format for TF mode.
2558 * config/mips/mips-opts.h (mips_ieee_754_setting): New enum.
2559 * config/mips/mips.c (mips_file_start): Output a `.nan' directive.
2560 (mips_option_override): Handle `-mnan=legacy'.
2561 * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Handle
2562 `-mabs=2008' and `-mnan=2008'.
2563 (OPTION_DEFAULT_SPECS): Add "nan" default.
2564 (ASM_SPEC): Handle `-mnan='.
2565 [!HAVE_AS_NAN] (HAVE_AS_NAN): New macro.
2566 * config/mips/mips.md (abs<mode>2): Handle `-mabs=2008', update
2567 comment accordingly.
2568 (neg<mode>2): Likewise.
2569 * config/mips/mips.opt (mabs, mnan): New options.
2570 * doc/install.texi (Configuration): Document `--with-nan=' option.
2571 * doc/invoke.texi (Option Summary): List MIPS `-mabs=' and
2573 (MIPS Options): Document them.
2574 * config.gcc <mips*-*-*>: Handle `--with-nan='.
2575 * configure.ac <mips*-*-*>: Check for GAS `-mnan=2008' support.
2576 * configure: Regenerate.
2577 * config.in: Regenerate.
2579 2013-07-29 Uros Bizjak <ubizjak@gmail.com>
2581 * config/i386/i386.md (float post-reload splitters): Do not check
2582 for subregs of SSE registers.
2584 2013-07-29 Uros Bizjak <ubizjak@gmail.com>
2585 H.J. Lu <hongjiu.lu@intel.com>
2589 * config/i386/i386.md (post-reload splitter
2590 to avoid partial SSE reg dependency stalls): New pattern.
2592 2013-07-29 Dominik Vogt <vogt@linux.vnet.ibm.com>
2594 * config/s390/s390.md ("movcc"): Swap load and store instructions.
2596 2013-07-27 Joern Rennecke <joern.rennecke@embecosm.com>
2598 * config/epiphany/epiphany.c (epiphany_compute_frame_size):
2599 Also reserve space for saving UNKNOWN_REGNUM for leaf functions.
2601 2013-07-26 Cary Coutant <ccoutant@google.com>
2603 * dwarf2out.c (die_checksum_ordered): Don't include template
2604 instantiations in signature.
2605 (is_template_parameter): New function.
2606 (is_template_instantiation): New function.
2607 (generate_skeleton_bottom_up): Don't include template instantiations
2609 (generate_skeleton): Likewise.
2610 (break_out_comdat_types): Move recursive call to break out nested
2612 (prune_unused_types_mark_generic_parms_dies): Call
2613 is_template_parameter.
2615 2013-07-26 Ian Bolton <ian.bolton@arm.com>
2617 * config/aarch64/aarch64.md (neg<mode>2): Offer alternative that
2618 uses vector registers.
2619 * config/aarch64/iterators.md: Add attributes rtn and vas.
2621 2013-07-26 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2622 Richard Earnshaw <richard.earnshaw@arm.com>
2624 * combine.c (simplify_comparison): Re-canonicalize operands
2626 * config/arm/arm.md (movcond_addsi): New splitter.
2628 2013-07-25 Sterling Augustine <saugustine@google.com>
2630 * dwarf2out.c (size_of_pubnames): Move code to...
2631 (include_pubname_in_output): ...here. New.
2632 (want_pubnames): Rearrange.
2633 (output_pubnames): Call include_pubname_in_output. Move assertion.
2635 2013-07-25 Cameron McInally <cameron.mcinally@nyu.edu>
2637 * doc/extend.texi: Fix return types for __builtin_ia32_cmp*s builtins.
2639 2013-07-25 Cameron McInally <cameron.mcinally@nyu.edu>
2642 * doc/extend.texi: Remove obsolete builtins. Fix
2643 typo for __builtin_ia32_loadss and __builtin_ia32_cmpnltss.
2645 2013-07-25 Jan Hubicka <jh@suse.cz>
2647 * cgraph.c (release_function_body): Break out from ...
2648 (cgraph_release_function_body): ... this one; also release DECL_RESULT
2650 * ipa-cp.c (get_replacement_map): Add parm_num argument; do not set
2651 old_tree in the map.
2652 (create_specialized_node): Update.
2653 * lto-cgraph.c (output_node_opt_summary): Do not translate old_tree
2655 * cgraphclones.c (cgraph_create_virtual_clone): Do not copy
2656 DECL_ARGUMENTS, DECL_INITIAL and DECL_RESULT.
2657 * ipa-prop.c (ipa_populate_param_decls): Look for origin of clones.
2658 * tree-inline.c (initialize_cfun): Initialize DECL_ARGUMENTS and
2661 2013-07-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2663 * config/arm/arm.md (arm_addsi3, addsi3_carryin_<optab>,
2664 addsi3_carryin_alt2_<optab>): Correct output template.
2666 2013-07-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2668 * config/arm/arm-fixed.md (ssmulsa3, usmulusa3):
2669 Adjust for arm_restrict_it.
2670 Remove trailing whitespace.
2672 2013-07-25  Mark Kettenis  <kettenis@openbsd.org>
2674 * config/pa/pa.c (pa_trampoline_init): Emit __enable_execute_stack
2675 libcall if HAVE_ENABLE_EXECUTE_STACK is defined.
2677 * config.gcc (hppa-*-openbsd*): Don't set tmake_file.
2679 2013-07-25 Vladimir Makarov <vmakarov@redhat.com>
2681 PR rtl-optimization/57960
2682 * lra-constraints.c (process_alt_operands): Use the right mode
2683 when checking strict_low.
2685 2013-07-25 Jan Hubicka <jh@suse.cz>
2687 * lto-symtab.c (lto_cgraph_replace_node): Release function body.
2688 * cgraph.c (cgraph_remove_node): Do not release function body
2689 when in cgraph streaming.
2690 * ipa.c (process_references, symtab_remove_unreachable_nodes): Objects
2691 in other partitions are not considered reachable; fix handling of
2694 2013-07-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
2696 * config/arm/arm.md (*sibcall_insn): Remove unnecessary space.
2698 2013-07-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
2703 * config/arm/arm.md ("*sibcall_insn): Replace use of
2704 Ss with US. Adjust output for v5 and v4t.
2705 (*sibcall_value_insn): Likewise and loosen predicate on operand0.
2707 * config/arm/constraints.md ("Ss"): Rename to US.
2709 2013-07-25 Terry Guo <terry.guo@arm.com>
2711 * config/arm/arm.c (thumb1_size_rtx_costs): Assign proper cost for
2712 shift_add/shift_sub0/shift_sub1 RTXs.
2714 2013-07-24 Bill Schmidt <wschmidt@linux.ibm.com>
2715 Anton Blanchard <anton@au1.ibm.com>
2717 * config/rs6000/altivec.md (altivec_vpkpx): Handle little endian.
2718 (altivec_vpks<VI_char>ss): Likewise.
2719 (altivec_vpks<VI_char>us): Likewise.
2720 (altivec_vpku<VI_char>us): Likewise.
2721 (altivec_vpku<VI_char>um): Likewise.
2723 2013-07-24 David Malcolm <dmalcolm@redhat.com>
2725 Introduce context class.
2727 * Makefile.in (CONTEXT_H): New.
2728 (OBJS): Add context.o.
2729 (toplev.o): Add CONTEXT_H to dependencies.
2732 * toplev.c (general_init): Create the singleton gcc::context instance.
2738 2013-07-24 Joern Rennecke <joern.rennecke@embecosm.com>
2740 PR rtl-optimization/57968
2741 * mode-switching.c (create_pre_exit): Allow instructions that
2742 don't set a return register to need a non-exit mode.
2744 2013-07-24 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
2745 Anton Blanchard <anton@au1.ibm.com>
2747 * config/rs6000/vector.md (vec_realign_load_<mode>): Reorder input
2748 operands to vperm for little endian.
2749 * config/rs6000/rs6000.c (rs6000_expand_builtin): Use lvsr instead
2750 of lvsl to create the control mask for a vperm for little endian.
2752 2013-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2753 Anton Blanchard <anton@au1.ibm.com>
2755 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Reverse
2756 two operands for little-endian.
2758 2013-07-23 Steve Ellcey <sellcey@mips.com>
2760 * config/mips/mips.c (mips_case_values_threshold): New.
2761 (TARGET_CASE_VALUES_THRESHOLD): Define.
2763 2013-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2764 Anton Blanchard <anton@au1.ibm.com>
2766 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Correct
2767 selection of field for vector splat in little endian mode.
2769 2013-07-23 Michael Meissner <meissner@linux.vnet.ibm.com>
2771 * config/rs6000/vector.md (xor<mode>3): Move 128-bit boolean
2772 expanders to rs6000.md.
2773 (ior<mode>3): Likewise.
2774 (and<mode>3): Likewise.
2775 (one_cmpl<mode>2): Likewise.
2776 (nor<mode>3): Likewise.
2777 (andc<mode>3): Likewise.
2778 (eqv<mode>3): Likewise.
2779 (nand<mode>3): Likewise.
2780 (orc<mode>3): Likewise.
2782 * config/rs6000/rs6000-protos.h (rs6000_split_logical): New
2785 * config/rs6000/rs6000.c (rs6000_split_logical_inner): Add support
2786 to split multi-word logical operations.
2787 (rs6000_split_logical_di): Likewise.
2788 (rs6000_split_logical): Likewise.
2790 * config/rs6000/vsx.md (VSX_L2): Delete, no longer used.
2791 (vsx_and<mode>3_32bit): Move 128-bit logical insns to rs6000.md,
2792 and allow TImode operations in 32-bit.
2793 (vsx_and<mode>3_64bit): Likewise.
2794 (vsx_ior<mode>3_32bit): Likewise.
2795 (vsx_ior<mode>3_64bit): Likewise.
2796 (vsx_xor<mode>3_32bit): Likewise.
2797 (vsx_xor<mode>3_64bit): Likewise.
2798 (vsx_one_cmpl<mode>2_32bit): Likewise.
2799 (vsx_one_cmpl<mode>2_64bit): Likewise.
2800 (vsx_nor<mode>3_32bit): Likewise.
2801 (vsx_nor<mode>3_64bit): Likewise.
2802 (vsx_andc<mode>3_32bit): Likewise.
2803 (vsx_andc<mode>3_64bit): Likewise.
2804 (vsx_eqv<mode>3_32bit): Likewise.
2805 (vsx_eqv<mode>3_64bit): Likewise.
2806 (vsx_nand<mode>3_32bit): Likewise.
2807 (vsx_nand<mode>3_64bit): Likewise.
2808 (vsx_orc<mode>3_32bit): Likewise.
2809 (vsx_orc<mode>3_64bit): Likewise.
2811 * config/rs6000/rs6000.h (VLOGICAL_REGNO_P): Always allow vector
2812 logical types in GPRs.
2814 * config/rs6000/altivec.md (altivec_and<mode>3): Move 128-bit
2815 logical insns to rs6000.md, and allow TImode operations in
2817 (altivec_ior<mode>3): Likewise.
2818 (altivec_xor<mode>3): Likewise.
2819 (altivec_one_cmpl<mode>2): Likewise.
2820 (altivec_nor<mode>3): Likewise.
2821 (altivec_andc<mode>3): Likewise.
2823 * config/rs6000/rs6000.md (BOOL_128): New mode iterators and mode
2824 attributes for moving the 128-bit logical operations into
2826 (BOOL_REGS_OUTPUT): Likewise.
2827 (BOOL_REGS_OP1): Likewise.
2828 (BOOL_REGS_OP2): Likewise.
2829 (BOOL_REGS_UNARY): Likewise.
2830 (BOOL_REGS_AND_CR0): Likewise.
2831 (one_cmpl<mode>2): Add support for DI logical operations on
2832 32-bit, splitting the operations to 32-bit.
2836 (and<mode>3, 128-bit types): Rewrite 2013-06-06 logical operator
2837 changes to combine the 32/64-bit code, allow logical operations on
2838 TI mode in 32-bit, and to use similar match_operator patterns like
2839 scalar mode uses. Combine the Altivec and VSX code for logical
2840 operations, and move it here.
2841 (ior<mode>3, 128-bit types): Likewise.
2842 (xor<mode>3, 128-bit types): Likewise.
2843 (one_cmpl<mode>3, 128-bit types): Likewise.
2844 (nor<mode>3, 128-bit types): Likewise.
2845 (andc<mode>3, 128-bit types): Likewise.
2846 (eqv<mode>3, 128-bit types): Likewise.
2847 (nand<mode>3, 128-bit types): Likewise.
2848 (orc<mode>3, 128-bit types): Likewise.
2849 (and<mode>3_internal): Likewise.
2850 (bool<mode>3_internal): Likewise.
2851 (boolc<mode>3_internal1): Likewise.
2852 (boolc<mode>3_internal2): Likewise.
2853 (boolcc<mode>3_internal1): Likewise.
2854 (boolcc<mode>3_internal2): Likewise.
2855 (eqv<mode>3_internal1): Likewise.
2856 (eqv<mode>3_internal2): Likewise.
2857 (one_cmpl1<mode>3_internal): Likewise.
2859 2013-07-23 David Holsgrove <david.holsgrove@xilinx.com>
2861 * config/microblaze/microblaze.c (microblaze_expand_prologue):
2862 Rename flag_stack_usage to flag_stack_usage_info.
2864 2013-07-23 David Holsgrove <david.holsgrove@xilinx.com>
2866 * config/microblaze/sync.md: New file.
2867 * config/microblaze/microblaze.md: Include sync.md
2868 * config/microblaze/microblaze.c: Add print_operand 'y'.
2869 * config/microblaze/constraints.md: Add memory_contraint
2870 'Q' which is a single register.
2872 2013-07-23 Eric Botcazou <ebotcazou@adacore.com>
2874 * doc/invoke.texi (SPARC Options): Document new leon3 processor value.
2876 2013-07-22 Po-Chun Chang <pchang9@cs.wisc.edu>
2878 * reload.c (find_reloads): Exit loop once we find this operand
2879 cannot be reloaded somehow for this alternative.
2881 * reload.c (find_reloads): Exit loop once we find a hard register.
2883 * rtlanal.c (computed_jump_p): Exit loop once we find label
2886 * i386.c (ix86_pad_returns): Exit loop after setting replace.
2888 * cfgloopmanip.c (remove_path): Exit loop after setting
2891 * gensupport.c (subst_dup): Avoid loop if code is not
2892 MATCH_DUP nor MATCH_OP_DUP.
2894 2013-07-23 Nicklas Bo Jensen <nbjensen@gmail.com>
2896 * doc/md.texi (Machine-Specific Peephole Optimizers): Fix a typo.
2898 2013-07-23 Yufeng Zhang <yufeng.zhang@arm.com>
2900 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Also return
2901 true for SP_REGNUM if mode == ptr_mode.
2902 * config/aarch64/aarch64.h (ADDITIONAL_REGISTER_NAMES): Add "wsp"
2903 with value R0_REGNUM + 31.
2905 2013-07-23 Yufeng Zhang <yufeng.zhang@arm.com>
2907 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): In big-endian,
2908 pad pointer-typed argument downward.
2910 2013-07-23 Yufeng Zhang <yufeng.zhang@arm.com>
2912 * config/aarch64/aarch64.h (TARGET_CPU_CPP_BUILTINS): Define _ILP32
2913 and __ILP32__ when the ILP32 model is in use.
2915 2013-07-23 Yufeng Zhang <yufeng.zhang@arm.com>
2917 * config/aarch64/aarch64.c (POINTER_BYTES): New define.
2918 (aarch64_load_symref_appropriately): In the case of
2919 SYMBOL_SMALL_ABSOLUTE, use the mode of 'dest' instead of Pmode
2920 to generate new rtx; likewise to the case of SYMBOL_SMALL_GOT.
2921 (aarch64_expand_mov_immediate): In the case of SYMBOL_FORCE_TO_MEM,
2922 change to pass 'ptr_mode' to force_const_mem and zero-extend 'mem'
2923 if 'mode' doesn't equal to 'ptr_mode'.
2924 (aarch64_output_mi_thunk): Add an assertion on the alignment of
2925 'vcall_offset'; change to call aarch64_emit_move differently depending
2926 on whether 'Pmode' equals to 'ptr_mode' or not; use 'POINTER_BYTES'
2927 to calculate the upper bound of 'vcall_offset'.
2928 (aarch64_cannot_force_const_mem): Change to also return true if
2930 (aarch64_legitimize_reload_address): In the case of large
2931 displacements, add new local variable 'xmode' and an assertion
2932 based on it; change to use 'xmode' to generate the new rtx and
2934 (aarch64_asm_trampoline_template): Change to generate the template
2935 differently depending on TARGET_ILP32 or not; change to use
2936 'POINTER_BYTES' in the argument passed to assemble_aligned_integer.
2937 (aarch64_trampoline_size): Removed.
2938 (aarch64_trampoline_init): Add new local constant 'tramp_code_sz'
2939 and replace immediate literals with it. Change to use 'ptr_mode'
2940 instead of 'DImode' and call convert_memory_address if the mode
2941 of 'fnaddr' doesn't equal to 'ptr_mode'.
2942 (aarch64_elf_asm_constructor): Change to use assemble_aligned_integer
2944 (aarch64_elf_asm_destructor): Likewise.
2945 * config/aarch64/aarch64.h (TRAMPOLINE_SIZE): Change to be dependent
2946 on TARGET_ILP32 instead of aarch64_trampoline_size.
2947 * config/aarch64/aarch64.md (movsi_aarch64): Add new alternatives
2948 of 'mov' between WSP and W registers as well as 'adr' and 'adrp'.
2949 (loadwb_pair<GPI:mode>_<PTR:mode>): Rename to ...
2950 (loadwb_pair<GPI:mode>_<P:mode>): ... this. Replace PTR with P.
2951 (storewb_pair<GPI:mode>_<PTR:mode>): Likewise; rename to ...
2952 (storewb_pair<GPI:mode>_<P:mode>): ... this.
2953 (add_losym): Change to 'define_expand' and call gen_add_losym_<mode>
2954 depending on the value of 'mode'.
2955 (add_losym_<mode>): New.
2956 (ldr_got_small_<mode>): New, based on ldr_got_small.
2957 (ldr_got_small): Remove.
2958 (ldr_got_small_sidi): New.
2959 * config/aarch64/iterators.md (P): New.
2960 (PTR): Change to 'ptr_mode' in the condition.
2962 2013-07-23 Yufeng Zhang <yufeng.zhang@arm.com>
2964 * config.gcc (aarch64*-*-*): Support --with-abi.
2965 (aarch64*-*-elf): Support --with-multilib-list.
2966 (aarch64*-*-linux*): Likewise.
2967 (supported_defaults): Add abi to aarch64*-*-*.
2968 * configure.ac: Mention AArch64 for --with-multilib-list.
2969 * configure: Re-generated.
2970 * config/aarch64/biarchilp32.h: New file.
2971 * config/aarch64/biarchlp64.h: New file.
2972 * config/aarch64/aarch64-elf.h (ENDIAN_SPEC): New define.
2974 (MULTILIB_DEFAULTS): Ditto.
2975 (DRIVER_SELF_SPECS): Ditto.
2976 (ASM_SPEC): Update to also substitute -mabi.
2977 * config/aarch64/aarch64-elf-raw.h (LINK_SPEC): Add linker script
2978 file whose name depends on -mabi= and -mbig-endian.
2979 * config/aarch64/aarch64.h (LONG_TYPE_SIZE): Change to depend on
2981 (POINTER_SIZE): New define.
2982 (POINTERS_EXTEND_UNSIGNED): Ditto.
2983 (enum aarch64_abi_type): New enumeration tag.
2984 (AARCH64_ABI_LP64, AARCH64_ABI_ILP32): New enumerators.
2985 (AARCH64_ABI_DEFAULT): Define to AARCH64_ABI_LP64 if undefined.
2986 (TARGET_ILP32): New define.
2987 * config/aarch64/aarch64.opt (mabi): New.
2989 (ilp32, lp64): New values for -mabi.
2990 * config/aarch64/t-aarch64 (comma): New define.
2991 (MULTILIB_OPTIONS): Ditto.
2992 (MULTILIB_DIRNAMES): Ditto.
2993 * config/aarch64/t-aarch64-linux (MULTIARCH_DIRNAME): New define.
2994 * doc/invoke.texi: Document -mabi for AArch64.
2996 2013-07-23 Georg-Johann Lay <avr@gjlay.de>
2998 * config/avr/avr.md: Explain asm print modifier 'r' for REG.
3000 2013-07-22 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
3001 Anton Blanchard <anton@au1.ibm.com>
3003 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Fix
3004 endianness when selecting field to splat.
3006 2013-07-22 Eric Christopher <echristo@gmail.com>
3008 * dwarf2out.c (die_odr_checksum): New function to use
3009 CHECKSUM_ macros and ULEB128 for DIE tag.
3010 (generate_type_signature): Use.
3012 2013-07-22 Eric Botcazou <ebotcazou@adacore.com>
3014 * config.gcc (sparc*-*-*): Accept leon3 processor.
3015 (sparc-leon*-*): Merge with sparc*-*-* and add leon3 support.
3016 * doc/invoke.texi (SPARC Options): Adjust -mfix-ut699 entry.
3017 * config/sparc/sparc-opts.h (enum processor_type): Add PROCESSOR_LEON3.
3018 * config/sparc/sparc.opt (enum processor_type): Add leon3.
3019 (mfix-ut699): Adjust comment.
3020 * config/sparc/sparc.h (TARGET_CPU_leon3): New define.
3021 (CPP_CPU32_DEFAULT_SPEC): Add leon3 support.
3022 (CPP_CPU_SPEC): Likewise.
3023 (ASM_CPU_SPEC): Likewise.
3024 * config/sparc/sparc.c (leon3_cost): New constant.
3025 (sparc_option_override): Add leon3 support.
3026 (mem_ref): New function.
3027 (sparc_gate_work_around_errata): Return true if -mfix-ut699 is enabled.
3028 (sparc_do_work_around_errata): Look into the instruction in the delay
3029 slot and adjust accordingly. Add fix for the data cache nullify issues
3030 of the UT699. Change insertion position for the NOP.
3031 * config/sparc/leon.md (leon_fpalu, leon_fpmds, write_buf): Delete.
3032 (leon3_load): New reservation.
3033 (leon_store): Bump latency to 2.
3034 (grfpu): New automaton.
3035 (grfpu_alu): New unit.
3036 (grfpu_ds): Likewise.
3037 (leon_fp_alu): Adjust.
3038 (leon_fp_mult): Delete.
3039 (leon_fp_div): Split into leon_fp_divs and leon_fp_divd.
3040 (leon_fp_sqrt): Split into leon_fp_sqrts and leon_fp_sqrtd.
3041 * config/sparc/sparc.md (cpu): Add leon3.
3042 * config/sparc/sync.md (atomic_exchangesi): Disable if -mfix-ut699.
3044 (atomic_test_and_set): Likewise.
3047 2013-07-22 Jürgen Urban <JuergenUrban@gmx.de>
3049 * config.gcc (mips*-*-*): Add --with-fpu support. Make single the
3050 default for R5900 targets.
3051 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Handle --with-fpu.
3052 (ISA_HAS_LDC1_SDC1): Set to false for TARGET_MIPS5900.
3053 * config/mips/mips.c (mips_option_override): Report an error for
3054 -march=r5900 -mhard-float -mdouble-float. Use spu_single_format
3055 for -march=r5900 -mhard-float.
3057 2013-07-22 Po-Chun Chang <pchang9@cs.wisc.edu>
3059 * df-problems.c (can_move_insns_across): Exit loop once we
3060 find a non-fixed, non-global register.
3062 * ipa-pure-const.c (propagate_nothrow): Exit loop after
3065 * omega.c (omega_eliminate_red): Break after setting red_found.
3066 (omega_problem_has_red_equations): Similarly after setting found.
3067 (omega_query_variable): Similarly after setting coupled.
3069 2013-07-22 Marek Polacek <polacek@redhat.com>
3071 * gimplify.c: Don't include gimple.h twice.
3073 2013-07-22 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3075 * config/arm/constraints.md (Pd): Allow TARGET_THUMB
3076 instead of TARGET_THUMB1.
3077 (Pz): New constraint.
3078 * config/arm/arm.md (arm_addsi3): Add alternatives for 16-bit
3080 (compare_negsi_si): Likewise.
3081 (compare_addsi2_op0): Likewise.
3082 (compare_addsi2_op1): Likewise.
3083 (addsi3_carryin_<optab>): Likewise.
3084 (addsi3_carryin_alt2_<optab>): Likewise.
3085 (addsi3_carryin_shift_<optab>): Disable cond_exec variant
3086 for arm_restrict_it.
3087 (subsi3_carryin): Likewise.
3088 (arm_subsi3_insn): Add alternatives for 16-bit encoding.
3089 (minmax_arithsi): Disable for arm_restrict_it.
3090 (minmax_arithsi_non_canon): Adjust for arm_restrict_it.
3091 (satsi_<SAT:code>): Disable cond_exec variant for arm_restrict_it.
3092 (satsi_<SAT:code>_shift): Likewise.
3093 (arm_shiftsi3): Add alternative for 16-bit encoding.
3094 (arm32_movhf): Disable for arm_restrict_it.
3095 (arm_cmpdi_unsigned): Add alternatives for 16-bit encoding.
3096 (arm_movtas_ze): Disable cond_exec variant for arm_restrict_it.
3098 2013-07-22 Sofiane Naci <sofiane.naci@arm.com>
3100 * config/arm/arm.md (attribute "insn"): Delete.
3101 (attribute "type"): Add "mov_imm", "mov_reg", "mov_shift",
3102 "mov_shift_reg", "mvn_imm", "mvn_reg", "mvn_shift" and "mvn_shift_reg".
3103 (not_shiftsi): Update for attribute change.
3104 (not_shiftsi_compare0): Likewise.
3105 (not_shiftsi_compare0_scratch): Likewise.
3106 (arm_one_cmplsi2): Likewise.
3107 (thumb1_one_cmplsi2): Likewise.
3108 (notsi_compare0): Likewise.
3109 (notsi_compare0_scratch): Likewise.
3110 (thumb1_movdi_insn): Likewise.
3111 (arm_movsi_insn): Likewise.
3112 (movhi_insn_arch4): Likewise.
3113 (movhi_bytes): Likewise.
3114 (arm_movqi_insn): Likewise.
3115 (thumb1_movqi_insn): Likewise.
3116 (arm32_movhf): Likewise.
3117 (thumb1_movhf): Likewise.
3118 (arm_movsf_soft_insn): Likewise.
3119 (thumb1_movsf_insn): Likewise.
3120 (thumb_movdf_insn): Likewise.
3121 (movsicc_insn): Likewise.
3122 (movsfcc_soft_insn): Likewise.
3123 (and_scc): Likewise.
3124 (cond_move): Likewise.
3125 (if_move_not): Likewise.
3126 (if_not_move): Likewise.
3127 (if_shift_move): Likewise.
3128 (if_move_shift): Likewise.
3129 (if_shift_shift): Likewise.
3130 (if_not_arith): Likewise.
3131 (if_arith_not): Likewise.
3132 (cond_move_not): Likewise.
3133 * config/arm/neon.md (neon_mov<mode>): Update for attribute change.
3134 (neon_mov<mode>): Likewise.
3135 * config/arm/vfp.md (arm_movsi_vfp): Update for attribute change.
3136 (thumb2_movsi_vfp): Likewise.
3137 (movsf_vfp): Likewise.
3138 (thumb2_movsf_vfp): Likewise.
3139 * config/arm/arm.c (xscale_sched_adjust_cost): Update for attribute
3141 (cortexa7_older_only): Likewise.
3142 (cortexa7_younger): Likewise.
3143 * config/arm/arm1020e.md (1020alu_op): Update for attribute change.
3144 (1020alu_shift_op): Likewise.
3145 (1020alu_shift_reg_op): Likewise.
3146 * config/arm/arm1026ejs.md (alu_op): Update for attribute change.
3147 (alu_shift_op): Likewise.
3148 (alu_shift_reg_op): Likewise.
3149 * config/arm/arm1136jfs.md (11_alu_op): Update for attribute change.
3150 (11_alu_shift_op): Likewise.
3151 (11_alu_shift_reg_op): Likewise.
3152 * config/arm/arm926ejs.md (9_alu_op): Update for attribute change.
3153 (9_alu_shift_reg_op): Likewise.
3154 * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
3156 (cortex_a15_alu_shift): Likewise.
3157 (cortex_a15_alu_shift_reg): Likewise.
3158 * config/arm/cortex-a5.md (cortex_a5_alu): Update for attribute change.
3159 (cortex_a5_alu_shift): Likewise.
3160 * config/arm/cortex-a53.md (cortex_a53_alu): Update for attribute
3162 (cortex_a53_alu_shift): Likewise.
3163 * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for attribute
3165 (cortex_a7_alu_reg): Likewise.
3166 (cortex_a7_alu_shift): Likewise.
3167 * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
3168 (cortex_a8_alu_shift): Likewise.
3169 (cortex_a8_alu_shift_reg): Likewise.
3170 (cortex_a8_mov): Likewise.
3171 * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute change.
3172 (cortex_a9_dp_shift): Likewise.
3173 * config/arm/cortex-m4.md (cortex_m4_alu): Update for attribute change.
3174 * config/arm/cortex-r4.md (cortex_r4_alu): Update for attribute change.
3175 (cortex_r4_mov): Likewise.
3176 (cortex_r4_alu_shift): Likewise.
3177 (cortex_r4_alu_shift_reg): Likewise.
3178 * config/arm/fa526.md (526_alu_op): Update for attribute change.
3179 (526_alu_shift_op): Likewise.
3180 * config/arm/fa606te.md (606te_alu_op): Update for attribute change.
3181 * config/arm/fa626te.md (626te_alu_op): Update for attribute change.
3182 (626te_alu_shift_op): Likewise.
3183 * config/arm/fa726te.md (726te_shift_op): Update for attribute change.
3184 (726te_alu_op): Likewise.
3185 (726te_alu_shift_op): Likewise.
3186 (726te_alu_shift_reg_op): Likewise.
3187 * config/arm/fmp626.md (mp626_alu_op): Update for attribute change.
3188 (mp626_alu_shift_op): Likewise.
3189 * config/arm/marvell-pj4.md (pj4_alu_e1): Update for attribute change.
3190 (pj4_alu_e1_conds): Likewise.
3191 (pj4_alu): Likewise.
3192 (pj4_alu_conds): Likewise.
3193 (pj4_shift): Likewise.
3194 (pj4_shift_conds): Likewise.
3195 (pj4_alu_shift): Likewise.
3196 (pj4_alu_shift_conds): Likewise.
3198 2013-07-22 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3200 * config/arm/predicates.md (shiftable_operator_strict_it):
3202 * config/arm/thumb2.md (thumb_andsi_not_shiftsi_si):
3203 Disable cond_exec version for arm_restrict_it.
3204 (thumb2_smaxsi3): Convert to generate cond_exec.
3205 (thumb2_sminsi3): Likewise.
3206 (thumb32_umaxsi3): Likewise.
3207 (thumb2_uminsi3): Likewise.
3208 (thumb2_abssi2): Adjust constraints for arm_restrict_it.
3209 (thumb2_neg_abssi2): Likewise.
3210 (thumb2_mov_scc): Add alternative for 16-bit encoding.
3211 (thumb2_movsicc_insn): Adjust alternatives.
3212 (thumb2_mov_negscc): Disable for arm_restrict_it.
3213 (thumb2_mov_negscc_strict_it): New pattern.
3214 (thumb2_mov_notscc_strict_it): New pattern.
3215 (thumb2_mov_notscc): Disable for arm_restrict_it.
3216 (thumb2_ior_scc): Likewise.
3217 (thumb2_ior_scc_strict_it): New pattern.
3218 (thumb2_cond_move): Adjust for arm_restrict_it.
3219 (thumb2_cond_arith): Disable for arm_restrict_it.
3220 (thumb2_cond_arith_strict_it): New pattern.
3221 (thumb2_cond_sub): Adjust for arm_restrict_it.
3222 (thumb2_movcond): Likewise.
3223 (thumb2_extendqisi_v6): Disable cond_exec variant for arm_restrict_it.
3224 (thumb2_zero_extendhisi2_v6): Likewise.
3225 (thumb2_zero_extendqisi2_v6): Likewise.
3226 (orsi_notsi_si): Likewise.
3227 (orsi_not_shiftsi_si): Likewise.
3229 2013-07-22 Georg-Johann Lay <avr@gjlay.de>
3231 * config/avr/avr.c (avr_out_xload): No SBIS around LPM so that
3232 instruction sequence is 1 byte shorter.
3234 2013-07-22 Uros Bizjak <ubizjak@gmail.com>
3236 * config/i386/i386.md (nonlocal_goto_receiver): Delete insn if
3237 it is not needed after split.
3239 2013-07-20 Iain Sandoe <iain@codesourcery.com>
3242 * config/i386/i386.c (output_set_got) [TARGET_MACHO]: Adjust to emit a
3243 second label for nonlocal goto receivers. Don't output pic base labels
3244 unless we're producing PIC; mark that action unreachable().
3245 (ix86_save_reg): If the function contains a nonlocal label, save the
3247 * config/darwin-protos.h (machopic_should_output_picbase_label): New.
3248 * gcc/config/darwin.c (emitted_pic_label_num): New GTY.
3249 (update_pic_label_number_if_needed): New.
3250 (machopic_output_function_base_name): Adjust for nonlocal receiver
3252 (machopic_should_output_picbase_label): New.
3253 * config/i386/i386.md (enum unspecv): UNSPECV_NLGR: New.
3254 (nonlocal_goto_receiver): New insn and split.
3256 2013-07-20 James Greenhalgh <james.greenhalgh@arm.com>
3258 * config/aarch64/aarch64-builtins.c
3259 (aarch64_fold_builtin): Fold abs in all modes.
3260 * config/aarch64/aarch64-simd-builtins.def
3261 (abs): Enable for all modes.
3262 * config/aarch64/arm_neon.h
3263 (vabs<q>_s<8,16,32,64): Rewrite using builtins.
3264 (vabs_f64): Add missing intrinsic.
3266 2013-07-19 Ian Bolton <ian.bolton@arm.com>
3268 * config/aarch64/arm_neon.h (vabs_s64): New function
3270 2013-07-19 Georg-Johann Lay <avr@gjlay.de>
3273 * config/avr/avr-fixed.md (round<mode>3_const): Turn expander to insn.
3274 * config/avr/avr.md (adjust_len): Add `round'.
3275 * config/avr/avr-protos.h (avr_out_round): New prototype.
3276 (avr_out_plus): Add `out_label' argument.
3277 * config/avr/avr.c (avr_out_plus_1): Add `out_label' argument.
3278 (avr_out_plus): Pass down `out_label' to avr_out_plus_1.
3279 Handle the case where `insn' is just a pattern.
3280 (avr_out_bitop): Handle the case where `insn' is just a pattern.
3281 (avr_out_round): New function.
3282 (avr_adjust_insn_length): Handle ADJUST_LEN_ROUND.
3284 2013-07-18 David Holsgrove <david.holsgrove@xilinx.com>
3286 * config/microblaze/microblaze.c (microblaze_expand_prologue):
3287 Add check for flag_stack_usage to handle -fstack-usage support
3289 2013-07-18 Pat Haugen <pthaugen@us.ibm.com>
3291 * config/rs6000/rs6000.c (rs6000_option_override_internal): Adjust flag
3292 interaction for new Power8 flags and VSX.
3294 2013-07-18 Sriraman Tallam <tmsriram@google.com>
3297 * tree-inline.c (expand_call_inline): Emit errors during
3298 early_inlining only if optimization is not turned on.
3300 2013-07-18 David Malcolm <dmalcolm@redhat.com>
3304 * passes.c (init_optimization_passes): Move the construction of
3305 the pass hierarchy into a new passes.def file.
3307 * Makefile.in (passes.o): Add dependency on passes.def.
3309 2013-07-18 David Malcolm <dmalcolm@redhat.com>
3311 * passes.c (init_optimization_passes): Introduce macros for
3312 constructing the tree of passes (INSERT_PASSES_AFTER,
3313 PUSH_INSERT_PASSES_WITHIN, POP_INSERT_PASSES,
3314 TERMINATE_PASS_LIST).
3316 2013-07-18 Vladimir Makarov <vmakarov@redhat.com>
3317 Wei Mi <wmi@google.com>
3319 PR rtl-optimization/57878
3320 * lra-assigns.c (assign_by_spills): Move non_reload_pseudos to the
3322 (reload_pseudo_compare_func): Check nregs first for reload
3325 2013-07-18 David Malcolm <dmalcolm@redhat.com>
3327 * tree-pass.h (pass_ipa_lto_wpa_fixup): Remove redundant decl.
3329 2013-07-18 Po-Chun Chang <pchang9@cs.wisc.edu>
3331 * read-rtl.c (validate_const_int): Once an invalid character is
3332 seen, quit the loop.
3334 * gengtype.c (write_roots): Similarly once we find the "deletable"
3335 or "if_marked" option.
3337 2013-07-18 Sofiane Naci <sofiane.naci@arm.com>
3339 * config/arm/arm.md (attribute "insn"): Delete values "mrs", "msr",
3340 "xtab" and "sat". Move value "clz" from here to ...
3341 (attriubte "type"): ... here.
3342 (satsi_<SAT:code>): Delete "insn" attribute.
3343 (satsi_<SAT:code>_shift): Likewise.
3344 (arm_zero_extendqisi2addsi): Likewise.
3345 (arm_extendqisi2addsi): Likewise.
3346 (clzsi2): Update for attribute changes.
3347 (rbitsi2): Likewise.
3348 * config/arm/arm-fixed.md (arm_ssatsihi_shift): Delete "insn"
3350 (arm_usatsihi): Likewise.
3351 * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
3353 2013-07-18 Sofiane Naci <sofiane.naci@arm.com>
3355 * config/arm/arm.md (attribute "type"): Rename "simple_alu_imm" to
3356 "arlo_imm". Rename "alu_reg" to "arlo_reg". Rename "simple_alu_shift"
3357 to "extend". Split "alu_shift" into "shift" and "arlo_shift". Split
3358 "alu_shift_reg" into "shift_reg" and "arlo_shift_reg". List types
3359 in alphabetical order.
3360 (attribute "core_cycles"): Update for attribute changes.
3361 (arm_addsi3): Likewise.
3362 (addsi3_compare0): Likewise.
3363 (addsi3_compare0_scratch): Likewise.
3364 (addsi3_compare_op1): Likewise.
3365 (addsi3_compare_op2): Likewise.
3366 (compare_addsi2_op0): Likewise.
3367 (compare_addsi2_op1): Likewise.
3368 (addsi3_carryin_shift_<optab>): Likewise.
3369 (subsi3_carryin_shift): Likewise.
3370 (rsbsi3_carryin_shift): Likewise.
3371 (arm_subsi3_insn): Likewise.
3372 (subsi3_compare0): Likewise.
3373 (subsi3_compare): Likewise.
3374 (arm_andsi3_insn): Likewise.
3375 (thumb1_andsi3_insn): Likewise.
3376 (andsi3_compare0): Likewise.
3377 (andsi3_compare0_scratch): Likewise.
3378 (zeroextractsi_compare0_scratch
3379 (andsi_not_shiftsi_si): Likewise.
3380 (iorsi3_insn): Likewise.
3381 (iorsi3_compare0): Likewise.
3382 (iorsi3_compare0_scratch): Likewise.
3383 (arm_xorsi3): Likewise.
3384 (thumb1_xorsi3_insn): Likewise.
3385 (xorsi3_compare0): Likewise.
3386 (xorsi3_compare0_scratch): Likewise.
3387 (satsi_<SAT:code>_shift): Likewise.
3389 (arm_shiftsi3): Likewise.
3390 (shiftsi3_compare0): Likewise.
3391 (not_shiftsi): Likewise.
3392 (not_shiftsi_compare0): Likewise.
3393 (not_shiftsi_compare0_scratch): Likewise.
3394 (arm_one_cmplsi2): Likewise.
3395 (thumb_one_complsi2): Likewise.
3396 (notsi_compare0): Likewise.
3397 (notsi_compare0_scratch): Likewise.
3398 (thumb1_zero_extendhisi2): Likewise.
3399 (arm_zero_extendhisi2): Likewise.
3400 (arm_zero_extendhisi2_v6): Likewise.
3401 (arm_zero_extendhisi2addsi): Likewise.
3402 (thumb1_zero_extendqisi2): Likewise.
3403 (thumb1_zero_extendqisi2_v6): Likewise.
3404 (arm_zero_extendqisi2): Likewise.
3405 (arm_zero_extendqisi2_v6): Likewise.
3406 (arm_zero_extendqisi2addsi): Likewise.
3407 (thumb1_extendhisi2): Likewise.
3408 (arm_extendhisi2): Likewise.
3409 (arm_extendhisi2_v6): Likewise.
3410 (arm_extendqisi): Likewise.
3411 (arm_extendqisi_v6): Likewise.
3412 (arm_extendqisi2addsi): Likewise.
3413 (thumb1_extendqisi2): Likewise.
3414 (thumb1_movdi_insn): Likewise.
3415 (arm_movsi_insn): Likewise.
3416 (movsi_compare0): Likewise.
3417 (movhi_insn_arch4): Likewise.
3418 (movhi_bytes): Likewise.
3419 (arm_movqi_insn): Likewise.
3420 (thumb1_movqi_insn): Likewise.
3421 (arm32_movhf): Likewise.
3422 (thumb1_movhf): Likewise.
3423 (arm_movsf_soft_insn): Likewise.
3424 (thumb1_movsf_insn): Likewise.
3425 (movdf_soft_insn): Likewise.
3426 (thumb_movdf_insn): Likewise.
3427 (arm_cmpsi_insn): Likewise.
3428 (cmpsi_shiftsi): Likewise.
3429 (cmpsi_shiftsi_swp): Likewise.
3430 (arm_cmpsi_negshiftsi_si): Likewise.
3431 (movsicc_insn): Likewise.
3432 (movsfcc_soft_insn): Likewise.
3433 (arith_shiftsi): Likewise.
3434 (arith_shiftsi_compare0
3435 (arith_shiftsi_compare0_scratch
3436 (sub_shiftsi): Likewise.
3437 (sub_shiftsi_compare0
3438 (sub_shiftsi_compare0_scratch
3439 (and_scc): Likewise.
3440 (cond_move): Likewise.
3441 (if_plus_move): Likewise.
3442 (if_move_plus): Likewise.
3443 (if_move_not): Likewise.
3444 (if_not_move): Likewise.
3445 (if_shift_move): Likewise.
3446 (if_move_shift): Likewise.
3447 (if_shift_shift): Likewise.
3448 (if_not_arith): Likewise.
3449 (if_arith_not): Likewise.
3450 (cond_move_not): Likewise.
3451 (thumb1_ashlsi3): Set type attribute.
3452 (thumb1_ashrsi3): Likewise.
3453 (thumb1_lshrsi3): Likewise.
3454 (thumb1_rotrsi3): Likewise.
3455 (shiftsi3_compare0_scratch): Likewise.
3456 * config/arm/neon.md (neon_mov<mode>): Update for attribute changes.
3457 (neon_mov<mode>): Likewise.
3458 * config/arm/thumb2.md (thumb_andsi_not_shiftsi_si): Update for
3460 (thumb2_movsi_insn): Likewise.
3461 (thumb2_cmpsi_neg_shiftsi): Likewise.
3462 (thumb2_extendqisi_v6): Likewise.
3463 (thumb2_zero_extendhisi2_v6): Likewise.
3464 (thumb2_zero_extendqisi2_v6): Likewise.
3465 (thumb2_shiftsi3_short): Likewise.
3466 (thumb2_addsi3_compare0_scratch): Likewise.
3467 (orsi_not_shiftsi_si): Likewise.
3468 * config/arm/vfp.md (arm_movsi_vfp): Update for attribute changes.
3469 * config/arm/arm-fixed.md (arm_ssatsihi_shift): Update for attribute
3471 * config/arm/arm1020e.md (1020alu_op): Update for attribute changes.
3472 (1020alu_shift_op): Likewise.
3473 (1020alu_shift_reg_op): Likewise.
3474 * config/arm/arm1026ejs.md (alu_op): Update for attribute changes.
3475 (alu_shift_op): Likewise.
3476 (alu_shift_reg_op): Likewise.
3477 * config/arm/arm1136jfs.md (11_alu_op): Update for attribute changes.
3478 (11_alu_shift_op): Likewise.
3479 (11_alu_shift_reg_op): Likewise.
3480 * config/arm/arm926ejs.md (9_alu_op): Update for attribute changes.
3481 (9_alu_shift_reg_op): Likewise.
3482 * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
3484 (cortex_a15_alu_shift): Likewise.
3485 (cortex_a15_alu_shift_reg): Likewise.
3486 * config/arm/cortex-a5.md (cortex_a5_alu): Update for attribute
3488 (cortex_a5_alu_shift): Likewise.
3489 * config/arm/cortex-a53.md (cortex_a53_alu) : Update for attribute
3491 (cortex_a53_alu_shift): Likewise.
3492 * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for attribute
3494 (cortex_a7_alu_reg): Likewise.
3495 (cortex_a7_alu_shift): Likewise.
3496 * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute
3498 (cortex_a8_alu_shift): Likewise.
3499 (cortex_a8_alu_shift_reg): Likewise.
3500 (cortex_a8_mov): Likewise.
3501 * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute changes.
3502 (cortex_a9_dp_shift): Likewise.
3503 * config/arm/cortex-m4.md (cortex_m4_alu): Update for attribute
3505 * config/arm/cortex-r4.md (cortex_r4_alu): Update for attribute
3507 (cortex_r4_mov): Likewise.
3508 (cortex_r4_alu_shift): Likewise.
3509 (cortex_r4_alu_shift_reg): Likewise.
3510 * config/arm/fa526.md (526_alu_op): Update for attribute changes.
3511 (526_alu_shift_op): Likewise.
3512 * config/arm/fa606te.md (606te_alu_op): Update for attribute changes.
3513 * config/arm/fa626te.md (626te_alu_op): Update for attribute changes.
3514 (626te_alu_shift_op): Likewise.
3515 * config/arm/fa726te.md (726te_shift_op): Update for attribute changes.
3516 (726te_alu_op): Likewise.
3517 (726te_alu_shift_op): Likewise.
3518 (726te_alu_shift_reg_op): Likewise.
3519 * config/arm/fmp626.md (mp626_alu_op): Update for attribute changes.
3520 (mp626_alu_shift_op): Likewise.
3521 * config/arm/marvell-pj4.md (pj4_alu_e1): Update for attribute changes.
3522 (pj4_alu_e1_conds): Likewise.
3523 (pj4_alu): Likewise.
3524 (pj4_alu_conds): Likewise.
3525 (pj4_shift): Likewise.
3526 (pj4_shift_conds): Likewise.
3527 (pj4_alu_shift): Likewise.
3528 (pj4_alu_shift_conds): Likewise.
3529 * config/arm/arm.c (xscale_sched_adjust_cost): Update for attribute
3531 (cortexa7_older_only): Likewise.
3532 (cortexa7_younger): Likewise.
3534 2013-07-18 David Malcolm <dmalcolm@redhat.com>
3536 * ipa-pure-const.c (generate_summary): Rename to...
3537 (pure_const_generate_summary): ... this.
3539 2013-07-17 Iain Sandoe <iain@codesourcery.com>
3541 * config/rs6000/darwin.h (REGISTER_NAMES): Add HTM registers.
3543 2013-07-17 Yvan Roux <yvan.roux@linaro.org>
3546 * config/arm/arm.c (gen_movmem_ldrd_strd): Fix unaligned load/store
3549 2013-07-17 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
3551 * config/s390/s390.c: (s390_expand_builtin): Allow -mhtm to be
3552 enabled without -march=zEC12.
3553 * config/s390/s390.h (TARGET_HTM): Do not require EC12 machine
3556 2013-07-16 Maciej W. Rozycki <macro@codesourcery.com>
3558 * config/mips/mips.h (ISA_HAS_FP4): Correct formatting.
3559 (ISA_HAS_FP_MADD4_MSUB4): Also enable for ISA_MIPS32R2.
3560 (ISA_HAS_NMADD4_NMSUB4): Remove the MODE argument; rewrite in
3561 terms of ISA_HAS_FP4, and also enable for ISA_MIPS32R2.
3562 (ISA_HAS_NMADD3_NMSUB3): Remove the MODE argument.
3563 * config/mips/mips.c (mips_rtx_costs) <PLUS>: Check for
3564 ISA_HAS_FP_MADD4_MSUB4 || ISA_HAS_FP_MADD3_MSUB3 rather than
3566 <MINUS, NEG>: Update according to changes to ISA_HAS_NMADD4_NMSUB4
3567 and ISA_HAS_NMADD3_NMSUB3.
3568 * config/mips/mips.md (nmadd4<mode>, nmadd3<mode>): Likewise.
3569 (nmadd4<mode>_fastmath, nmadd3<mode>_fastmath): Likewise.
3570 (nmsub4<mode>, nmsub3<mode>): Likewise.
3571 (nmsub4<mode>_fastmath, nmsub3<mode>_fastmath): Likewise.
3573 2013-07-16 Maciej W. Rozycki <macro@codesourcery.com>
3575 * config/mips/mips.h (ISA_HAS_NMADD4_NMSUB4): Remove
3576 TARGET_MIPS5400 checking.
3578 2013-07-16 Jakub Jelinek <jakub@redhat.com>
3579 Peter Bergner <bergner@vnet.ibm.com>
3581 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTERS): Mention HTM
3582 registers in the comment.
3583 (DWARF_FRAME_REGISTERS): Subtract also the 3 HTM registers.
3584 (DWARF_REG_TO_UNWIND_COLUMN): Use DWARF_FRAME_REGISTERS
3585 rather than FIRST_PSEUDO_REGISTERS.
3587 2013-07-16 Peter Bergner <bergner@vnet.ibm.com>
3589 * config/rs6000/rs6000.c (rs6000_option_override_internal): Do not
3590 enable extra ISA flags with TARGET_HTM.
3592 2013-07-16 Maciej W. Rozycki <macro@codesourcery.com>
3594 * config/mips/mips.h (ISA_HAS_MULS, ISA_HAS_MSAC, ISA_HAS_MACC):
3597 2013-07-15 Cong Hou <congh@google.com>
3599 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use hash function
3600 in compare function for sorting.
3602 2013-07-15 Peter Bergner <bergner@vnet.ibm.com>
3604 * config.gcc (powerpc*-*-*): Install htmintrin.h and htmxlintrin.h.
3605 * config/rs6000/t-rs6000 (MD_INCLUDES): Add htm.md.
3606 * config/rs6000/rs6000.opt: Add -mhtm option.
3607 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Add OPTION_MASK_HTM.
3608 (ISA_2_7_MASKS_SERVER): Add OPTION_MASK_HTM.
3609 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Define
3610 __HTM__ if the HTM instructions are available.
3611 * config/rs6000/predicates.md (u3bit_cint_operand, u10bit_cint_operand,
3612 htm_spr_reg_operand): New define_predicates.
3613 * config/rs6000/rs6000.md (define_attr "type"): Add htm.
3614 (TFHAR_REGNO, TFIAR_REGNO, TEXASR_REGNO): New define_constants.
3616 * config/rs6000/rs6000-builtin.def (BU_HTM_0, BU_HTM_1, BU_HTM_2,
3617 BU_HTM_3, BU_HTM_SPR0, BU_HTM_SPR1): Add support macros for defining
3618 HTM builtin functions.
3619 * config/rs6000/rs6000.c (RS6000_BUILTIN_H): New macro.
3620 (rs6000_reg_names, alt_reg_names): Add HTM SPR register names.
3621 (rs6000_init_hard_regno_mode_ok): Add support for HTM instructions.
3622 (rs6000_builtin_mask_calculate): Likewise.
3623 (rs6000_option_override_internal): Likewise.
3624 (bdesc_htm): Add new HTM builtin support.
3625 (htm_spr_num): New function.
3626 (htm_spr_regno): Likewise.
3627 (rs6000_htm_spr_icode): Likewise.
3628 (htm_expand_builtin): Likewise.
3629 (htm_init_builtins): Likewise.
3630 (rs6000_expand_builtin): Add support for HTM builtin functions.
3631 (rs6000_init_builtins): Likewise.
3632 (rs6000_invalid_builtin, rs6000_opt_mask): Add support for -mhtm
3634 * config/rs6000/rs6000.h (ASM_CPU_SPEC): Add support for -mhtm.
3635 (TARGET_HTM, MASK_HTM): Define macros.
3636 (FIRST_PSEUDO_REGISTER): Adjust for new HTM SPR registers.
3637 (FIXED_REGISTERS): Likewise.
3638 (CALL_USED_REGISTERS): Likewise.
3639 (CALL_REALLY_USED_REGISTERS): Likewise.
3640 (REG_ALLOC_ORDER): Likewise.
3641 (enum reg_class): Likewise.
3642 (REG_CLASS_NAMES): Likewise.
3643 (REG_CLASS_CONTENTS): Likewise.
3644 (REGISTER_NAMES): Likewise.
3645 (ADDITIONAL_REGISTER_NAMES): Likewise.
3646 (RS6000_BTC_SPR, RS6000_BTC_VOID, RS6000_BTC_32BIT, RS6000_BTC_64BIT,
3647 RS6000_BTC_MISC_MASK, RS6000_BTM_HTM): New macros.
3648 (RS6000_BTM_COMMON): Add RS6000_BTM_HTM.
3649 * config/rs6000/htm.md: New file.
3650 * config/rs6000/htmintrin.h: New file.
3651 * config/rs6000/htmxlintrin.h: New file.
3653 2013-07-15 Marcus Shawcroft <marcus.shawcroft@arm.com>
3655 * config/aarch64/aarch64-protos.h (aarch64_symbol_type):
3656 Define SYMBOL_TINY_GOT, update comment.
3657 * config/aarch64/aarch64.c
3658 (aarch64_load_symref_appropriately): Handle SYMBOL_TINY_GOT.
3659 (aarch64_expand_mov_immediate): Likewise.
3660 (aarch64_print_operand): Likewise.
3661 (aarch64_classify_symbol): Likewise.
3662 * config/aarch64/aarch64.md (UNSPEC_GOTTINYPIC): Define.
3663 (ldr_got_tiny): Define.
3665 2013-07-13 Tobias Grosser <tobias@grosser.es>
3667 PR tree-optimization/54094
3668 * graphite-clast-to-gimple.c (translate_clast_for_loop): Derive the
3669 scheduling dimension for the parallelism check from the polyhedral
3670 information in the AST.
3671 * graphite-dependences.c (carries_deps): Do not assume the schedule is
3674 2013-07-13 Jason Merrill <jason@redhat.com>
3676 * print-tree.c (debug_vec_tree): Use debug_raw.
3677 (debug_raw (vec<tree, va_gc> &)): New.
3678 (debug_raw (vec<tree, va_gc> *)): New.
3679 * tree.h: Declare them.
3681 2013-07-13 Bin Cheng <bin.cheng@arm.com>
3683 * ifcvt.c (ifcvt_after_combine): New static variable.
3684 (cheap_bb_rtx_cost_p): Set scale to REG_BR_PROB_BASE when optimizing
3686 (if_convert): New parameter after_combine. Set ifcvt_after_combine.
3687 (rest_of_handle_if_conversion, rest_of_handle_if_after_combine,
3688 rest_of_handle_if_after_reload): Pass new argument for if_convert.
3690 2013-07-12 Maciej W. Rozycki <macro@codesourcery.com>
3692 * config/mips/mips.c (mips_expand_call): Remove empty statement.
3694 2013-07-12 Michael Matz <matz@suse.de>
3697 * convert.c (convert_to_real): Reject non-float inner types.
3699 2013-07-12 Tejas Belagod <tejas.belagod@arm.com>
3701 * config/aarch64/aarch64-protos.h
3702 (aarch64_simd_immediate_valid_for_move): Remove.
3703 * config/aarch64/aarch64.c (simd_immediate_info): New member.
3704 (aarch64_simd_valid_immediate): Recognize idioms for shifting ones
3706 (aarch64_output_simd_mov_immediate): Print the correct shift specifier.
3708 2013-07-11 Steve Ellcey <sellcey@mips.com>
3710 * config/mips/mips.c (mips_conditional_register_usage): Do not
3711 use t[0-7] registers in MIPS16 mode when optimizing for size.
3713 2013-07-11 Sriraman Tallam <tmsriram@google.com>
3715 * config/i386/i386.c (dispatch_function_versions): Fix array
3716 indexing of function_version_info to match actual_versions.
3718 2013-07-11 Teresa Johnson <tejohnson@google.com>
3720 * vec.h (struct va_gc): Move release out-of-line.
3721 (va_gc::release): Call ggc_free on released vec.
3723 2013-07-11 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
3725 * config/rs6000/rs6000.md (""*tls_gd_low<TLSmode:tls_abi_suffix>"):
3726 Require GOT register as additional operand in UNSPEC.
3727 ("*tls_ld_low<TLSmode:tls_abi_suffix>"): Likewise.
3728 ("*tls_got_dtprel_low<TLSmode:tls_abi_suffix>"): Likewise.
3729 ("*tls_got_tprel_low<TLSmode:tls_abi_suffix>"): Likewise.
3730 ("*tls_gd<TLSmode:tls_abi_suffix>"): Update splitter.
3731 ("*tls_ld<TLSmode:tls_abi_suffix>"): Likewise.
3732 ("tls_got_dtprel_<TLSmode:tls_abi_suffix>"): Likewise.
3733 ("tls_got_tprel_<TLSmode:tls_abi_suffix>"): Likewise.
3735 2013-07-11 Georg-Johann Lay <avr@gjlay.de>
3738 * config/avr/avr.c (avr_set_current_function): Sanity-check signal
3739 name seen by assembler/linker rather if available.
3741 2013-07-11 Andreas Schwab <schwab@suse.de>
3743 * config/aarch64/aarch64-linux.h (CPP_SPEC): Define.
3745 2013-07-10 Vladimir Makarov <vmakarov@redhat.com>
3747 * lra-constraints.c (curr_insn_transform): Switch off optional reloads.
3749 2013-07-10 Joseph Myers <joseph@codesourcery.com>
3751 * doc/tm.texi.in: Move hook documentation to ....
3752 * target.def: ... here.
3754 * doc/tm.texi.in (TARGET_CANONICALIZE_COMPARISON): Remove stray
3756 * doc/tm.texi: Regenerate.
3758 2013-07-10 Paolo Carlini <paolo.carlini@oracle.com>
3761 * doc/invoke.texi: Document -Wconditionally-supported.
3763 2013-07-10 Georg-Johann Lay <avr@gjlay.de>
3766 * config/avr/avr.c (avr_prologue_setup_frame): Trunk -size to mode
3769 2013-07-10 Georg-Johann Lay <avr@gjlay.de>
3772 * config/avr/avr-mcus.def (atmega16hva, atmega16hva2, atmega16hvb)
3773 (atmega16m1, atmega16u4, atmega32a, atmega32c1, atmega32hvb)
3774 (atmega32m1, atmega32u4, atmega32u6, atmega64c1, atmega64m1):
3775 Remove duplicate devices.
3776 * config/avr/gen-avr-mmcu-texi.c (print_mcus): Fail on duplicate MCUs.
3777 * config/avr/t-multilib: Regenerate.
3778 * config/avr/avr-tables.opt: Regenerate.
3779 * doc/avr-mmcu.texi: Regenerate.
3781 2013-07-10 Georg-Johann Lay <avr@gjlay.de>
3784 * config/avr/avr.opt (Waddr-space-convert): Fix typo.
3786 2013-07-10 Graham Stott <graham.stott@btinternet.com>
3788 * config/mips/mips.c (mips_rtx_costs): Very slightly increase
3789 the cost of MULT when optimizing for size.
3791 2013-07-10 Jan-Benedict Glaw <jbglaw@lug-owl.de>
3793 * config/cr16/cr16-protos.h: Don't include target.h.
3795 2013-07-09 Joseph Myers <joseph@codesourcery.com>
3797 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Only
3798 adjust register size for TDmode and TFmode for VSX registers.
3800 2013-07-08 Kai Tietz <ktietz@redhat.com>
3803 * config/i386/i386.c (TARGET_FUNCTION_ATTRIBUTE_INLINABLE_P): Define as
3804 hook_bool_const_tree_true.
3806 2013-07-08 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
3808 * config/s390/s390.c: Replace F*_REGNUM with FPR*_REGNUM.
3809 * config/s390/s390.h: Remove F*_REGNUM macro definitions.
3810 * config/s390/s390.md: Define FPR*_REGNUM constants.
3811 Fix FPR2_REGNUM constant (18 -> 17).
3812 ("*trunc<BFP:mode><DFP_ALL:mode>2")
3813 ("*trunc<DFP_ALL:mode><BFP:mode>2")
3814 ("trunc<BFP:mode><DFP_ALL:mode>2")
3815 ("trunc<DFP_ALL:mode><BFP:mode>2")
3816 ("*extend<BFP:mode><DFP_ALL:mode>2")
3817 ("*extend<DFP_ALL:mode><BFP:mode>2")
3818 ("extend<BFP:mode><DFP_ALL:mode>2")
3819 ("extend<DFP_ALL:mode><BFP:mode>2"): Replace FPR2_REGNUM with
3822 2013-07-08 Graham Stott <graham.stott@btinternet.com>
3824 * Makefile.in: (c-family-warn): Define to $(STRICT_WARN)
3826 2013-07-08 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
3828 * config/s390/s390.c: Rename cfun_set_fpr_bit to cfun_set_fpr_save
3829 and cfun_fpr_bit_p to cfun_fpr_save_p.
3830 (s390_frame_area, s390_register_info, s390_frame_info)
3831 (s390_emit_prologue, s390_emit_epilogue)
3832 (s390_conditional_register_usage): Use the *_REGNUM macros for FPR
3834 * config/s390/s390.h: Define *_REGNUM macros for floating point
3837 2013-07-08 Eric Botcazou <ebotcazou@adacore.com>
3839 * Makefile.in (tree-ssa-reassoc.o): Add dependency on $(PARAMS_H).
3841 2013-07-08 Po-Chun Chang <pchang9@cs.wisc.edu>
3843 PR rtl-optimization/57786
3844 * combine.c (distribute_notes) <case REG_DEAD>: Change all_used to bool
3845 and break out of the loop when it is set to false.
3847 2013-07-08 Jakub Jelinek <jakub@redhat.com>
3850 * simplify-rtx.c (simplify_unary_operation_1) <case ZERO_EXTEND>:
3851 Simplify (zero_extend:SI (subreg:QI (and:SI (reg:SI)
3852 (const_int 63)) 0)).
3853 * combine.c (make_extraction): Create ZERO_EXTEND or SIGN_EXTEND
3854 using simplify_gen_unary instead of gen_rtx_*_EXTEND.
3855 * config/i386/i386.md (*jcc_bt<mode>_1): New define_insn_and_split.
3857 PR rtl-optimization/57829
3858 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Ensure that
3859 mask bits outside of mode are just sign-extension from mode to HWI.
3861 2013-07-08 Michael Zolotukhin <michael.v.zolotukhin@gmail.com>
3863 * config/i386/i386-opts.h (enum stringop_alg): Add vector_loop.
3864 * config/i386/i386.c (expand_set_or_movmem_via_loop): Use
3865 adjust_address instead of change_address to keep info about alignment.
3866 (emit_strmov): Remove.
3867 (emit_memmov): New function.
3868 (expand_movmem_epilogue): Refactor to properly handle bigger sizes.
3869 (expand_movmem_epilogue): Likewise and return updated rtx for
3871 (expand_constant_movmem_prologue): Likewise and return updated rtx for
3872 destination and source.
3873 (decide_alignment): Refactor, handle vector_loop.
3874 (ix86_expand_movmem): Likewise.
3875 (ix86_expand_setmem): Likewise.
3876 * config/i386/i386.opt (Enum): Add vector_loop to option stringop_alg.
3878 2013-07-07 Uros Bizjak <ubizjak@gmail.com>
3880 * config/i386/driver-i386.c (host_detect_local_cpu): Do not check
3881 signature_TM2_ebx, it interferes with signature_INTEL_ebx.
3883 2013-07-06 Uros Bizjak <ubizjak@gmail.com>
3885 * config/i386/sse.md (sse_movlhps): Change alternative 3
3886 of operand 2 to "m".
3888 2013-07-06 Uros Bizjak <ubizjak@gmail.com>
3891 * config/i386/sse.md (iptr): New mode attribute.
3892 (sse2_movq128): Add pointer size overrides for Intel asm dialect.
3893 (<sse>_vm<plusminus_insn><mode>3): Ditto.
3894 (<sse>_vmmul<mode>3): Ditto.
3895 (<sse>_vmdiv<mode>3): Ditto.
3896 (sse_vmrcpv4sf2): Ditto.
3897 (<sse>_vmsqrt<mode>2): Ditto.
3898 (sse_vmrsqrtv4sf2): Ditto.
3899 (<sse>_vm<code><mode>3): Ditto.
3900 (avx_vmcmp<mode>3): Ditto.
3901 (<sse>_vmmaskcmp<mode>3): Ditto.
3902 (<sse>_comi): Ditto.
3903 (<sse>_ucomi): Ditto.
3904 (*xop_vmfrcz_<mode>): Ditto.
3905 (*fmai_fmadd_<mode>): Ditto.
3906 (*fmai_fmsub_<mode>): Ditto.
3907 (*fmai_fnmadd_<mode>): Ditto.
3908 (*fmai_fnmsub_<mode>): Ditto.
3909 (*fma4i_vmfmadd_<mode>): Ditto.
3910 (*fma4i_vmfmsub_<mode>): Ditto.
3911 (*fma4i_vmfnmadd_<mode>): Ditto.
3912 (*fma4i_vmfnmsub_<mode>): Ditto.
3913 (*xop_vmfrcz_<mode>): Ditto.
3914 (sse_cvtps2pi): Ditto.
3915 (sse_cvttps2pi): Ditto.
3916 (sse_cvtss2si): Ditto.
3917 (sse_cvtss2si_2): Ditto.
3918 (sse_cvtss2siq_2): Ditto.
3919 (sse_cvttss2si): Ditto.
3920 (sse_cvttss2siq): Ditto.
3921 (sse_cvtsd2si): Ditto.
3922 (sse_cvtsd2si_2): Ditto.
3923 (sse_cvtsd2siq_2): Ditto.
3924 (sse_cvttsd2si): Ditto.
3925 (sse_cvttsd2siq): Ditto.
3926 (sse_cvtsd2ss): Ditto.
3927 (sse_cvtss2sd): Ditto.
3928 (avx2_pbroadcast<mode>): Ditto.
3929 (avx2_pbroadcast<mode>_1): Ditto.
3930 (*avx_vperm_broadcast_v4sf): Ditto.
3932 (sse_movhlps): Ditto for movlp[sd]/movhp[sd] alternatives.
3933 (sse_movlhps): Ditto.
3934 (sse_storehps): Ditto.
3935 (sse_loadhps): Ditto.
3936 (sse_storelps): Ditto.
3937 (sse_loadlps): Ditto.
3938 (*vec_concatv4sf): Ditto.
3939 (*vec_interleave_highv2df): Ditto.
3940 (*vec_interleave_lowv2df): Ditto.
3941 (*vec_extractv2df_1_sse): Ditto.
3942 (*vec_extractv2df_0_sse): Ditto.
3943 (sse2_storelpd): Ditto.
3944 (sse2_loadlpd): Ditto.
3945 (sse2_movsd): Ditto.
3946 (*vec_concatv4si): Ditto.
3947 (vec_concatv2di): Ditto.
3949 * config/i386/mmx.md (mmx_punpcklbw): Add pointer size overrides
3950 for Intel asm dialect.
3951 (mmx_punpcklwd): Ditto.
3952 (mmx_punpckldq): Ditto.
3954 * config/i386/i386.c (ix86_print_operand) ['H']: Output 'qword ptr'
3955 for intel assembler dialect.
3957 2013-07-06 Jakub Jelinek <jakub@redhat.com>
3960 * fold-const.c (tree_call_nonnegative_warnv_p): Return true
3961 for BUILT_IN_C{LZ,LRSB}*.
3962 * tree.h (CASE_INT_FN): Add FN##IMAX case.
3963 * tree-vrp.c (extract_range_basic): Handle
3964 BUILT_IN_{FFS,PARITY,POPCOUNT,C{LZ,TZ,LRSB}}*. For
3965 BUILT_IN_CONSTANT_P if argument isn't (D) of PARM_DECL,
3966 fall thru to code calling set_value*.
3967 * builtins.c (expand_builtin): Remove *IMAX cases.
3968 (fold_builtin_bitop): For BUILT_IN_CLRSB* return NULL_TREE
3969 if width is bigger than 2*HWI.
3971 2013-07-05 Vladimir Makarov <vmakarov@redhat.com>
3973 PR rtl-optimization/55342
3974 * lra-int.h (lra_subreg_reload_pseudos): New.
3975 * lra.c: Add undoing optional reloads to the block diagram.
3976 (lra_subreg_reload_pseudos): New.
3977 (lra_optional_reload_pseudos): Change comments.
3978 (lra): Init and clear lra_subreg_reload_pseudos. Clear
3979 lra_optional_reload_pseudos after undo transformations.
3980 * lra-assigns.c (pseudo_prefix_title): New.
3981 (lra_setup_reg_renumber): Use it.
3982 (spill_for): Ditto. Check subreg reload pseudos too.
3983 (assign_by_spills): Consider subreg reload pseudos too.
3984 * lra-constraints.c (simplify_operand_subreg): Use
3985 lra_subreg_reload_pseudos instead of lra_optional_reload_pseudos.
3986 (curr_insn_transform): Recognize and do optional reloads.
3987 (undo_optional_reloads): New.
3988 (lra_undo_inheritance): Call undo_optional_reloads.
3990 2013-07-05 Thomas Quinot <quinot@adacore.com>
3992 * tree-complex.c (expand_complex_operations_1): Fix typo.
3994 2013-07-04 Tejas Belagod <tejas.belagod@arm.com>
3996 * config/aarch64/aarch64-protos.h (cpu_vector_cost): New.
3997 (tune_params): New member 'const vec_costs'.
3998 * config/aarch64/aarch64.c (generic_vector_cost): New.
3999 (generic_tunings): New member 'generic_vector_cost'.
4000 (aarch64_builtin_vectorization_cost): New.
4001 (aarch64_add_stmt_cost): New.
4002 (TARGET_VECTORIZE_ADD_STMT_COST): New.
4003 (TARGET_VECTORIZE_BUILTIN_VECTORIZATION_COST): New.
4005 2013-07-03 Jakub Jelinek <jakub@redhat.com>
4008 * config/i386/predicates.md (vsib_address_operand): Disallow
4009 SYMBOL_REF or LABEL_REF in parts.disp if TARGET_64BIT && flag_pic.
4011 2013-07-03 Hans-Peter Nilsson <hp@bitrange.com>
4014 * stmt.c (expand_nl_goto_receiver): Remove almost-copy of
4015 expand_builtin_setjmp_receiver.
4016 (expand_label): Adjust, call expand_builtin_setjmp_receiver
4017 with NULL for the label parameter.
4018 * builtins.c (expand_builtin_setjmp_receiver): Don't clobber
4019 the frame-pointer. Adjust comments.
4020 [HAVE_builtin_setjmp_receiver]: Emit builtin_setjmp_receiver
4021 only if LABEL is non-NULL.
4023 2013-07-03 Yufeng Zhang <yufeng.zhang@arm.com>
4025 * config/aarch64/aarch64.h (enum arm_abi_type): Remove.
4026 (ARM_ABI_AAPCS64): Ditto.
4028 (ARM_DEFAULT_ABI): Ditto.
4030 2013-07-03 James Greenhalgh <james.greenhalgh@arm.com>
4032 * config/aarch64/aarch64-builtins.c
4033 (aarch64_simd_expand_builtin): Handle AARCH64_SIMD_STORE1.
4034 * config/aarch64/aarch64-simd-builtins.def (ld1): New.
4036 * config/aarch64/aarch64-simd.md
4037 (aarch64_ld1<VALL:mode>): New.
4038 (aarch64_st1<VALL:mode>): Likewise.
4039 * config/aarch64/arm_neon.h
4040 (vld1<q>_<fpsu><8, 16, 32, 64>): Convert to RTL builtins.
4042 2013-07-02 Sriraman Tallam <tmsriram@google.com>
4044 * config/i386/i386.c (gate_insert_vzeroupper): Check if
4046 (ix86_option_override_internal):Turn on all -mavx target flags by
4047 default as they are dependent on AVX anyway.
4049 2013-07-02 Cary Coutant <ccoutant@google.com>
4051 * dwarf2out.c (loc_checksum): Call hash_loc_operands for a
4053 (loc_checksum_ordered): Likewise.
4054 (hash_loc_operands): Remove inline keyword.
4056 2013-07-02 Jakub Jelinek <jakub@redhat.com>
4058 PR tree-optimization/57741
4059 * tree-vect-loop.c (vect_is_simple_iv_evolution): Disallow
4060 non-INTEGRAL_TYPE_P non-SCALAR_FLOAT_TYPE_P SSA_NAME step_exprs,
4061 or SCALAR_FLOAT_TYPE_P SSA_NAMEs if !flag_associative_math.
4062 Allow REAL_CST step_exprs if flag_associative_math.
4063 (get_initial_def_for_induction): Handle SCALAR_FLOAT_TYPE_P step_expr.
4065 2013-07-02 Ian Bolton <ian.bolton@arm.com>
4067 * config/aarch64/aarch64-simd.md (absdi2): Support abs for DI mode.
4069 2013-07-02 Ian Bolton <ian.bolton@arm.com>
4071 * config/aarch64/aarch64.md (*extr_insv_reg<mode>): New pattern.
4073 2013-07-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4075 * config/arm/arm.md (arm_andsi3_insn): Add alternatives for 16-bit
4077 (iorsi3_insn): Likewise.
4078 (arm_xorsi3): Likewise.
4080 2013-07-01 Sofiane Naci <sofiane.naci@arm.com>
4082 * arm.md (attribute "wtype"): Delete. Move attribute values from here
4084 (attribute "type"): ... here, and prefix with "wmmx_".
4085 (attribute "core_cycles"): Update for attribute changes.
4086 * iwmmxt.md (tbcstv8qi): Update for attribute changes.
4087 (tbcstv4hi): Likewise.
4088 (tbcstv2si): Likewise.
4089 (iwmmxt_iordi3): Likewise.
4090 (iwmmxt_xordi3): Likewise.
4091 (iwmmxt_anddi3): Likewise.
4092 (iwmmxt_nanddi3): Likewise.
4093 (iwmmxt_arm_movdi): Likewise.
4094 (iwmmxt_movsi_insn): Likewise.
4095 (mov<mode>_internal): Likewise.
4096 (and<mode>3_iwmmxt): Likewise.
4097 (ior<mode>3_iwmmxt): Likewise.
4098 (xor<mode>3_iwmmxt): Likewise.
4099 (add<mode>3_iwmmxt): Likewise.
4100 (ssaddv8qi3): Likewise.
4101 (ssaddv4hi3): Likewise.
4102 (ssaddv2si3): Likewise.
4103 (usaddv8qi3): Likewise.
4104 (usaddv4hi3): Likewise.
4105 (usaddv2si3): Likewise.
4106 (sub<mode>3_iwmmxt): Likewise.
4107 (sssubv8qi3): Likewise.
4108 (sssubv4hi3): Likewise.
4109 (sssubv2si3): Likewise.
4110 (ussubv8qi3): Likewise.
4111 (ussubv4hi3): Likewise.
4112 (ussubv2si3): Likewise.
4113 (mulv4hi3_iwmmxt): Likewise.
4114 (smulv4hi3_highpart): Likewise.
4115 (umulv4hi3_highpart): Likewise.
4116 (iwmmxt_wmacs): Likewise.
4117 (iwmmxt_wmacsz): Likewise.
4118 (iwmmxt_wmacu): Likewise.
4119 (iwmmxt_wmacuz): Likewise.
4120 (iwmmxt_clrdi): Likewise.
4121 (iwmmxt_clrv8qi): Likewise.
4122 (iwmmxt_clr4hi): Likewise.
4123 (iwmmxt_clr2si): Likewise.
4124 (iwmmxt_uavgrndv8qi3): Likewise.
4125 (iwmmxt_uavgrndv4hi3): Likewise.
4126 (iwmmxt_uavgv8qi3): Likewise.
4127 (iwmmxt_uavgv4hi3): Likewise.
4128 (iwmmxt_tinsrb): Likewise.
4129 (iwmmxt_tinsrh): Likewise.
4130 (iwmmxt_tinsrw): Likewise.
4131 (iwmmxt_textrmub): Likewise.
4132 (iwmmxt_textrmsb): Likewise.
4133 (iwmmxt_textrmuh): Likewise.
4134 (iwmmxt_textrmsh): Likewise.
4135 (iwmmxt_textrmw): Likewise.
4136 (iwmxxt_wshufh): Likewise.
4137 (eqv8qi3): Likewise.
4138 (eqv4hi3): Likewise.
4139 (eqv2si3): Likewise.
4140 (gtuv8qi3): Likewise.
4141 (gtuv4hi3): Likewise.
4142 (gtuv2si3): Likewise.
4143 (gtv8qi3): Likewise.
4144 (gtv4hi3): Likewise.
4145 (gtv2si3): Likewise.
4146 (smax<mode>3_iwmmxt): Likewise.
4147 (umax<mode>3_iwmmxt): Likewise.
4148 (smin<mode>3_iwmmxt): Likewise.
4149 (umin<mode>3_iwmmxt): Likewise.
4150 (iwmmxt_wpackhss): Likewise.
4151 (iwmmxt_wpackwss): Likewise.
4152 (iwmmxt_wpackdss): Likewise.
4153 (iwmmxt_wpackhus): Likewise.
4154 (iwmmxt_wpackwus): Likewise.
4155 (iwmmxt_wpackdus): Likewise.
4156 (iwmmxt_wunpckihb): Likewise.
4157 (iwmmxt_wunpckihh): Likewise.
4158 (iwmmxt_wunpckihw): Likewise.
4159 (iwmmxt_wunpckilb): Likewise.
4160 (iwmmxt_wunpckilh): Likewise.
4161 (iwmmxt_wunpckilw): Likewise.
4162 (iwmmxt_wunpckehub): Likewise.
4163 (iwmmxt_wunpckehuh): Likewise.
4164 (iwmmxt_wunpckehuw): Likewise.
4165 (iwmmxt_wunpckehsb): Likewise.
4166 (iwmmxt_wunpckehsh): Likewise.
4167 (iwmmxt_wunpckehsw): Likewise.
4168 (iwmmxt_wunpckelub): Likewise.
4169 (iwmmxt_wunpckeluh): Likewise.
4170 (iwmmxt_wunpckeluw): Likewise.
4171 (iwmmxt_wunpckelsb): Likewise.
4172 (iwmmxt_wunpckelsh): Likewise.
4173 (iwmmxt_wunpckelsw): Likewise.
4174 (ror<mode>3): Likewise.
4175 (ashr<mode>3_iwmmxt): Likewise.
4176 (lshr<mode>3_iwmmxt): Likewise.
4177 (ashl<mode>3_iwmmxt): Likewise.
4178 (ror<mode>3_di): Likewise.
4179 (ashr<mode>3_di): Likewise.
4180 (lshr<mode>3_di): Likewise.
4181 (ashl<mode>3_di): Likewise.
4182 (iwmmxt_wmadds): Likewise.
4183 (iwmmxt_wmaddu): Likewise.
4184 (iwmmxt_tmia): Likewise.
4185 (iwmmxt_tmiaph): Likewise.
4186 (iwmmxt_tmiabb): Likewise.
4187 (iwmmxt_tmiatb): Likewise.
4188 (iwmmxt_tmiabt): Likewise.
4189 (iwmmxt_tmiatt): Likewise.
4190 (iwmmxt_tmovmskb): Likewise.
4191 (iwmmxt_tmovmskh): Likewise.
4192 (iwmmxt_tmovmskw): Likewise.
4193 (iwmmxt_waccb): Likewise.
4194 (iwmmxt_wacch): Likewise.
4195 (iwmmxt_waccw): Likewise.
4196 (iwmmxt_waligni): Likewise.
4197 (iwmmxt_walignr): Likewise.
4198 (iwmmxt_walignr0): Likewise.
4199 (iwmmxt_walignr1): Likewise.
4200 (iwmmxt_walignr2): Likewise.
4201 (iwmmxt_walignr3): Likewise.
4202 (iwmmxt_wsadb): Likewise.
4203 (iwmmxt_wsadh): Likewise.
4204 (iwmmxt_wsadbz): Likewise.
4205 (iwmmxt_wsadhz): Likewise.
4206 * iwmmxt2.md (iwmmxt_wabs<mode>3): Update for attribute changes.
4207 (iwmmxt_wabsdiffb): Likewise.
4208 (iwmmxt_wabsdiffh): Likewise.
4209 (iwmmxt_wabsdiffw): Likewise.
4210 (iwmmxt_waddsubhx): Likewise
4211 (iwmmxt_wsubaddhx): Likewise.
4212 (addc<mode>3): Likewise.
4213 (iwmmxt_avg4): Likewise.
4214 (iwmmxt_avg4r): Likewise.
4215 (iwmmxt_wmaddsx): Likewise.
4216 (iwmmxt_wmaddux): Likewise.
4217 (iwmmxt_wmaddsn): Likewise.
4218 (iwmmxt_wmaddun): Likewise.
4219 (iwmmxt_wmulwsm): Likewise.
4220 (iwmmxt_wmulwum): Likewise.
4221 (iwmmxt_wmulsmr): Likewise.
4222 (iwmmxt_wmulumr): Likewise.
4223 (iwmmxt_wmulwsmr): Likewise.
4224 (iwmmxt_wmulwumr): Likewise.
4225 (iwmmxt_wmulwl): Likewise.
4226 (iwmmxt_wqmulm): Likewise.
4227 (iwmmxt_wqmulwm): Likewise.
4228 (iwmmxt_wqmulmr): Likewise.
4229 (iwmmxt_wqmulwmr): Likewise.
4230 (iwmmxt_waddbhusm): Likewise.
4231 (iwmmxt_waddbhusl): Likewise.
4232 (iwmmxt_wqmiabb): Likewise.
4233 (iwmmxt_wqmiabt): Likewise.
4234 (iwmmxt_wqmiatb): Likewise.
4235 (iwmmxt_wqmiatt): Likewise.
4236 (iwmmxt_wqmiabbn): Likewise.
4237 (iwmmxt_wqmiabtn): Likewise.
4238 (iwmmxt_wqmiatbn): Likewise.
4239 (iwmmxt_wqmiattn): Likewise.
4240 (iwmmxt_wmiabb): Likewise.
4241 (iwmmxt_wmiabt): Likewise.
4242 (iwmmxt_wmiatb): Likewise.
4243 (iwmmxt_wmiatt): Likewise.
4244 (iwmmxt_wmiabbn): Likewise.
4245 (iwmmxt_wmiabtn): Likewise.
4246 (iwmmxt_wmiatbn): Likewise.
4247 (iwmmxt_wmiattn): Likewise.
4248 (iwmmxt_wmiawbb): Likewise.
4249 (iwmmxt_wmiawbt): Likewise.
4250 (iwmmxt_wmiawtb): Likewise.
4251 (iwmmxt_wmiawtt): Likewise.
4252 (iwmmxt_wmiawbbn): Likewise.
4253 (iwmmxt_wmiawbtn): Likewise.
4254 (iwmmxt_wmiawtbn): Likewise.
4255 (iwmmxt_wmiawttn): Likewise.
4256 (iwmmxt_wmerge): Likewise.
4257 (iwmmxt_tandc<mode>3): Likewise.
4258 (iwmmxt_torc<mode>3): Likewise.
4259 (iwmmxt_torvsc<mode>3): Likewise.
4260 (iwmmxt_textrc<mode>3): Likewise.
4261 * marvell-f-iwmmxt.md (wmmxt_shift): Update for attribute changes.
4262 (wmmxt_pack): Likewise.
4263 (wmmxt_mult_c1): Likewise.
4264 (wmmxt_mult_c2): Likewise.
4265 (wmmxt_alu_c1): Likewise.
4266 (wmmxt_alu_c2): Likewise.
4267 (wmmxt_alu_c3): Likewise.
4268 (wmmxt_transfer_c1): Likewise.
4269 (wmmxt_transfer_c2): Likewise.
4270 (wmmxt_transfer_c3): Likewise.
4271 (marvell_f_iwmmxt_wstr): Likewise.
4272 (marvell_f_iwmmxt_wldr): Likewise.
4274 2013-06-29 Yufeng Zhang <yufeng.zhang@arm.com>
4276 * config/aarch64/aarch64.c: Remove junk from the beginning of the file.
4278 2013-06-28 Vladimir Makarov <vmakarov@redhat.com>
4281 2013-06-28 Vladimir Makarov <vmakarov@redhat.com>
4282 * lra-constraints.c (need_for_split_p): Check call used hard regs
4283 living through calls.
4285 * lra-constraints.c (inherit_in_ebb): Reset live_hard_regs for
4286 call used regs for call insn.
4288 2013-06-28 Jakub Jelinek <jakub@redhat.com>
4291 * config/i386/i386.c (ix86_expand_builtin): If target == NULL and
4292 mode is VOIDmode, don't create a VOIDmode pseudo to copy result into.
4294 2013-06-28 Balaji V. Iyer <balaji.v.iyer@intel.com>
4296 * builtins.def: Fixed the function type of CILKPLUS_BUILTIN.
4298 2013-06-28 Vladimir Makarov <vmakarov@redhat.com>
4300 * lra-constraints.c (need_for_split_p): Check call used hard regs
4301 living through calls.
4303 2013-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
4306 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Do not allow PTImode
4307 to tie with any other modes. Eliminate Altivec vector mode tests,
4308 since these are a subset of ALTIVEC or VSX vector modes. Simplify
4309 code, to return 0 if testing MODE2 for a condition, if we've
4310 already tested MODE1 for the same condition.
4312 2013-06-28 Marcus Shawcroft <marcus.shawcroft@arm.com>
4314 * config/aarch64/aarch64.c (aarch64_cannot_force_const_mem): Adjust
4317 2013-06-28 Marcus Shawcroft <marcus.shawcroft@arm.com>
4319 * config/aarch64/aarch64-protos.h (aarch64_symbol_type):
4320 Update comment w.r.t SYMBOL_TINY_ABSOLUTE.
4322 2013-06-28 Marcus Shawcroft <marcus.shawcroft@arm.com>
4324 * config/aarch64/aarch64-protos.h (aarch64_classify_symbol_expression):
4326 (aarch64_symbolic_constant_p): Remove.
4327 * config/aarch64/aarch64.c (aarch64_classify_symbol_expression): Remove
4328 static. Fix line length and white space.
4329 (aarch64_symbolic_constant_p): Remove.
4330 * config/aarch64/predicates.md (aarch64_valid_symref):
4331 Use aarch64_classify_symbol_expression.
4333 2013-06-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4335 * config/arm/constraints.md (Ts): New constraint.
4336 * config/arm/arm.md (arm_movqi_insn): Add alternatives for
4338 (compare_scc): Use "Ts" constraint for operand 0.
4339 (ior_scc_scc): Likewise.
4340 (and_scc_scc): Likewise.
4341 (and_scc_scc_nodom): Likewise.
4342 (ior_scc_scc_cmp): Likewise for operand 7.
4343 (and_scc_scc_cmp): Likewise.
4344 * config/arm/thumb2.md (thumb2_movsi_insn):
4345 Add alternatives for 16-bit encodings.
4346 (thumb2_movhi_insn): Likewise.
4347 (thumb2_movsicc_insn): Likewise.
4348 (thumb2_and_scc): Take 'and' outside cond_exec. Use "Ts" constraint.
4349 (thumb2_negscc): Use "Ts" constraint.
4350 Move mvn instruction outside cond_exec block.
4351 * config/arm/vfp.md (thumb2_movsi_vfp): Add alternatives
4352 for 16-bit encodings.
4354 2013-06-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4356 * config/arm/arm.md (arm_mulsi3_v6): Add alternative for 16-bit
4358 (mulsi3addsi_v6): Disable predicable variant for arm_restrict_it.
4359 (mulsi3subsi): Likewise.
4360 (mulsidi3adddi): Likewise.
4361 (mulsidi3_v6): Likewise.
4362 (umulsidi3_v6): Likewise.
4363 (umulsidi3adddi_v6): Likewise.
4364 (smulsi3_highpart_v6): Likewise.
4365 (umulsi3_highpart_v6): Likewise.
4366 (mulhisi3tb): Likewise.
4367 (mulhisi3bt): Likewise.
4368 (mulhisi3tt): Likewise.
4369 (maddhisi4): Likewise.
4370 (maddhisi4tb): Likewise.
4371 (maddhisi4tt): Likewise.
4372 (maddhidi4): Likewise.
4373 (maddhidi4tb): Likewise.
4374 (maddhidi4tt): Likewise.
4375 (zeroextractsi_compare0_scratch): Likewise.
4376 (insv_zero): Likewise.
4377 (insv_t2): Likewise.
4378 (anddi_notzesidi_di): Likewise.
4379 (anddi_notsesidi_di): Likewise.
4380 (andsi_notsi_si): Likewise.
4381 (iordi_zesidi_di): Likewise.
4382 (xordi_zesidi_di): Likewise.
4383 (andsi_iorsi3_notsi): Likewise.
4385 (smax_m1): Likewise.
4387 (not_shiftsi): Likewise.
4388 (unaligned_loadsi): Likewise.
4389 (unaligned_loadhis): Likewise.
4390 (unaligned_loadhiu): Likewise.
4391 (unaligned_storesi): Likewise.
4392 (unaligned_storehi): Likewise.
4393 (extv_reg): Likewise.
4394 (extzv_t2): Likewise.
4396 (udivsi3): Likewise.
4397 (arm_zero_extendhisi2addsi): Likewise.
4398 (arm_zero_extendqisi2addsi): Likewise.
4399 (compareqi_eq0): Likewise.
4400 (arm_extendhisi2_v6): Likewise.
4401 (arm_extendqisi2addsi): Likewise.
4402 (arm_movt): Likewise.
4403 (thumb2_ldrd): Likewise.
4404 (thumb2_ldrd_base): Likewise.
4405 (thumb2_ldrd_base_neg): Likewise.
4406 (thumb2_strd): Likewise.
4407 (thumb2_strd_base): Likewise.
4408 (thumb2_strd_base_neg): Likewise.
4409 (arm_negsi2): Add alternative for 16-bit encoding.
4410 (arm_one_cmplsi2): Likewise.
4412 2013-06-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4414 * config/arm/predicates.md (arm_cond_move_operator): New predicate.
4415 * config/arm/arm.md (movsfcc): Use arm_cond_move_operator predicate.
4416 (movdfcc): Likewise.
4417 * config/arm/vfp.md (*thumb2_movsf_vfp):
4418 Disable predication for arm_restrict_it.
4419 (*thumb2_movsfcc_vfp): Disable for arm_restrict_it.
4420 (*thumb2_movdfcc_vfp): Likewise.
4421 (*abssf2_vfp, *absdf2_vfp, *negsf2_vfp, *negdf2_vfp,*addsf3_vfp,
4422 *adddf3_vfp, *subsf3_vfp, *subdf3_vfpc, *divsf3_vfp,*divdf3_vfp,
4423 *mulsf3_vfp, *muldf3_vfp, *mulsf3negsf_vfp, *muldf3negdf_vfp,
4424 *mulsf3addsf_vfp, *muldf3adddf_vfp, *mulsf3subsf_vfp,
4425 *muldf3subdf_vfp, *mulsf3negsfaddsf_vfp, *fmuldf3negdfadddf_vfp,
4426 *mulsf3negsfsubsf_vfp, *muldf3negdfsubdf_vfp, *fma<SDF:mode>4,
4427 *fmsub<SDF:mode>4, *fnmsub<SDF:mode>4, *fnmadd<SDF:mode>4,
4428 *extendsfdf2_vfp, *truncdfsf2_vfp, *extendhfsf2, *truncsfhf2,
4429 *truncsisf2_vfp, *truncsidf2_vfp, fixuns_truncsfsi2, fixuns_truncdfsi2,
4430 *floatsisf2_vfp, *floatsidf2_vfp, floatunssisf2, floatunssidf2,
4431 *sqrtsf2_vfp, *sqrtdf2_vfp, *cmpsf_vfp, *cmpsf_trap_vfp, *cmpdf_vfp,
4432 *cmpdf_trap_vfp, <vrint_pattern><SDF:mode>2):
4433 Disable predication for arm_restrict_it.
4435 2013-06-28 Kirill Yukhin <kirill.yukhin@intel.com>
4437 * config/i386/bmiintrin.h (_bextr_u32): New.
4438 (_bextr_u64): Ditto.
4440 2013-06-27 Richard Sandiford <rdsandiford@googlemail.com>
4442 * config.gcc (mips*-mti-elf*, mips*-sde-elf*, mips64r5900-*-elf*)
4443 (mips64r5900el-*-elf*): Include mips/n32-elf.h.
4444 * config/mips/sde.h (LOCAL_LABEL_PREFIX, NO_DOLLAR_IN_LABEL)
4445 (LONG_DOUBLE_TYPE_SIZE, LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Move to...
4446 * config/mips/n32-elf.h: ...this new file.
4448 2013-06-27 Marc Glisse <marc.glisse@inria.fr>
4451 * config/i386/i386.c (enum ix86_builtins, bdesc_args): Remove
4452 IX86_BUILTIN_CMPNGTSS and IX86_BUILTIN_CMPNGESS.
4454 2013-06-27 Catherine Moore <clm@codesourcery.com>
4456 * config/mips/mips-tables.opt: Regenerate.
4457 * config/mips/mips-cpus.def: Add m14ke and m14kec.
4458 * config/mips/mips.h (BASE_DRIVER_SELF_SPECS): m14ke* implies -mdspr2.
4459 * doc/invoke.texi: Add -m14kc.
4461 2013-06-27 Jakub Jelinek <jakub@redhat.com>
4464 * config/i386/i386.md (bmi_bextr_<mode>): Swap predicates and
4465 constraints of operand 1 and 2.
4468 * config/i386/i386.md (bmi2_bzhi_<mode>3): Swap AND arguments
4469 to match RTL canonicalization. Swap predicates and
4470 constraints of operand 1 and 2.
4472 2013-06-27 Vladimir Makarov <vmakarov@redhat.com>
4474 * lra-constraints.c (inherit_in_ebb): Process static hard regs too.
4475 Process OP_INOUT regs for splitting too.
4477 2013-06-27 Jakub Jelinek <jakub@redhat.com>
4479 * tree-vect-stmts.c (vectorizable_store): Move ptr_incr var
4480 decl before the loop, initialize to NULL.
4481 (vectorizable_load): Initialize ptr_incr to NULL.
4483 2013-06-27 Martin Jambor <mjambor@suse.cz>
4486 * ipa-ref.h (ipa_maybe_record_reference): Declare.
4487 * ipa-ref.c (ipa_maybe_record_reference): New function.
4488 * cgraphclones.c (cgraph_create_virtual_clone): Use it.
4489 * ipa-cp.c (create_specialized_node): Record potential references from
4491 * Makefile.in (ipa-ref.o): Add IPA_REF_H to dependencies.
4493 2013-06-27 Yufeng Zhang <yufeng.zhang@arm.com>
4495 * config/aarch64/aarch64.c (aarch64_force_temporary): Add an extra
4496 parameter 'mode' of type 'enum machine_mode mode'; change to pass
4497 'mode' to force_reg.
4498 (aarch64_add_offset): Update calls to aarch64_force_temporary.
4499 (aarch64_expand_mov_immediate): Likewise.
4501 2013-06-27 Yufeng Zhang <yufeng.zhang@arm.com>
4503 * config/aarch64/aarch64.c (aarch64_add_offset): Change to pass
4504 'mode' to aarch64_plus_immediate and gen_rtx_PLUS.
4506 2013-06-27 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
4508 * config/s390/s390.c: Rename UNSPEC_CCU_TO_INT to
4509 UNSPEC_STRCMPCC_TO_INT and UNSPEC_CCZ_TO_INT to UNSPEC_CC_TO_INT.
4510 (struct machine_function): Add tbegin_p.
4511 (s390_canonicalize_comparison): Fold CC mode compares to
4512 conditional jump if possible.
4513 (s390_emit_jump): Return the emitted jump.
4514 (s390_branch_condition_mask, s390_branch_condition_mnemonic):
4515 Handle CCRAWmode compares.
4516 (s390_option_override): Default to -mhtm if available.
4517 (s390_reg_clobbered_rtx): Handle floating point regs as well.
4518 (s390_regs_ever_clobbered): Use s390_regs_ever_clobbered also for
4519 FPRs instead of df_regs_ever_live_p.
4520 (s390_optimize_nonescaping_tx): New function.
4521 (s390_init_frame_layout): Extend clobbered_regs array to cover
4523 (s390_emit_prologue): Call s390_optimize_nonescaping_tx.
4524 (s390_expand_tbegin): New function.
4525 (enum s390_builtin): New enum definition.
4526 (code_for_builtin): New array definition.
4527 (s390_init_builtins): New function.
4528 (s390_expand_builtin): New function.
4529 (TARGET_INIT_BUILTINS): Define.
4530 (TARGET_EXPAND_BUILTIN): Define.
4531 * common/config/s390/s390-common.c (processor_flags_table): Add PF_TX.
4532 * config/s390/predicates.md (s390_comparison): Handle CCRAWmode.
4533 (s390_alc_comparison): Likewise.
4534 * config/s390/s390-modes.def: Add CCRAWmode.
4535 * config/s390/s390.h (processor_flags): Add PF_TX.
4536 (TARGET_CPU_HTM): Define macro.
4537 (TARGET_HTM): Define macro.
4538 (TARGET_CPU_CPP_BUILTINS): Define __HTM__ for htm.
4539 * config/s390/s390.md: Rename UNSPEC_CCU_TO_INT to
4540 UNSPEC_STRCMPCC_TO_INT and UNSPEC_CCZ_TO_INT to UNSPEC_CC_TO_INT.
4541 (UNSPECV_TBEGIN, UNSPECV_TBEGINC, UNSPECV_TEND, UNSPECV_TABORT)
4542 (UNSPECV_ETND, UNSPECV_NTSTG, UNSPECV_PPA): New unspecv enum values.
4543 (TBEGIN_MASK, TBEGINC_MASK): New constants.
4544 ("*cc_to_int"): Move up.
4545 ("*mov<mode>cc", "*cjump_64", "*cjump_31"): Accept integer
4546 constants other than 0.
4547 ("*ccraw_to_int"): New insn and splitter definition.
4548 ("tbegin", "tbegin_nofloat", "tbegin_retry")
4549 ("tbegin_retry_nofloat", "tbeginc", "tend", "tabort")
4550 ("tx_assist"): New expander.
4551 ("tbegin_1", "tbegin_nofloat_1", "*tbeginc_1", "*tend_1")
4552 ("*tabort_1", "etnd", "ntstg", "*ppa"): New insn definition.
4553 * config/s390/s390.opt: Add -mhtm option.
4554 * config/s390/s390-protos.h (s390_emit_jump): Add return type.
4555 * config/s390/htmxlintrin.h: New file.
4556 * config/s390/htmintrin.h: New file.
4557 * config/s390/s390intrin.h: New file.
4558 * doc/extend.texi: Document htm builtins.
4559 * config.gcc: Add the new header files to extra_headers.
4561 2013-06-26 Thomas Schwinge <thomas@codesourcery.com>
4563 * config/i386/gnu.h [TARGET_LIBC_PROVIDES_SSP]
4564 (TARGET_CAN_SPLIT_STACK, TARGET_THREAD_SPLIT_STACK_OFFSET): Undefine.
4566 2013-06-26 Michael Meissner <meissner@linux.vnet.ibm.com>
4567 Pat Haugen <pthaugen@us.ibm.com>
4568 Peter Bergner <bergner@vnet.ibm.com>
4570 * config/rs6000/power8.md: New.
4571 * config/rs6000/rs6000-cpus.def (RS6000_CPU table): Adjust processor
4572 setting for power8 entry.
4573 * config/rs6000/t-rs6000 (MD_INCLUDES): Add power8.md.
4574 * config/rs6000/rs6000.c (is_microcoded_insn, is_cracked_insn): Adjust
4575 test for Power4/Power5 only.
4576 (insn_must_be_first_in_group, insn_must_be_last_in_group): Add Power8
4578 (force_new_group): Adjust comment.
4579 * config/rs6000/rs6000.md: Include power8.md.
4581 2013-06-26 Greta Yorsh <Greta.Yorsh@arm.com>
4583 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Define macro.
4584 * config/arm/arm-protos.h (arm_max_conditional_execute): New
4586 (tune_params): Update comment.
4587 * config/arm/arm.c (arm_cortex_a15_tune): Set max_cond_insns to 2.
4588 (arm_max_conditional_execute): New function.
4589 (thumb2_final_prescan_insn): Use max_insn_skipped and
4590 MAX_INSN_PER_IT_BLOCK to compute maximum instructions in a block.
4592 2013-06-25 Jakub Jelinek <jakub@redhat.com>
4594 PR tree-optimization/57705
4595 * tree-vect-loop.c (vect_is_simple_iv_evolution): Allow
4596 SSA_NAME step, provided that it is not defined inside the loop.
4597 (vect_analyze_scalar_cycles_1): Disallow SSA_NAME step in nested loop.
4598 (get_initial_def_for_induction): Handle SSA_NAME IV step.
4600 2013-06-25 Martin Jambor <mjambor@suse.cz>
4603 * cgraph.h (cgraph_indirect_call_info): New flag member_ptr.
4604 * ipa-prop.c (ipa_print_node_jump_functions): Mark member pointer
4606 (ipa_note_param_call): Initialize member_ptr flag.
4607 (ipa_analyze_indirect_call_uses): Set member_ptr flag.
4608 (ipa_make_edge_direct_to_target): Bail out if member_ptr is set.
4609 (ipa_write_indirect_edge_info): Stream member_ptr flag.
4610 (ipa_read_indirect_edge_info): Likewise.
4612 2013-06-25 Richard Biener <rguenther@suse.de>
4615 * passes.c (init_optimization_passes): Move pass_fold_builtins
4616 and pass_dce earlier with -Og.
4618 2013-06-25 Eric Botcazou <ebotcazou@adacore.com>
4620 * expr.c (expand_expr_real_1) <ARRAY_REF>: Fix formatting glitches.
4621 <BIT_FIELD_REF>: Remove trailing TAB.
4622 * varasm.c (output_constructor_bitfield): Fix formatting glitch and
4625 2013-06-24 Martin Jambor <mjambor@suse.cz>
4627 PR tree-optimization/57358
4628 * ipa-prop.c (ipa_func_spec_opts_forbid_analysis_p): New function.
4629 (ipa_compute_jump_functions_for_edge): Bail out if it returns true.
4630 (ipa_analyze_params_uses): Generate pessimistic info when true.
4632 2013-06-24 Martin Jambor <mjambor@suse.cz>
4634 PR tree-optimization/57539
4635 * cgraphclones.c (cgraph_clone_node): Add parameter new_inlined_to, set
4636 global.inlined_to of the new node to it. All callers changed.
4637 * ipa-inline-transform.c (clone_inlined_nodes): New variable
4638 inlining_into, pass it to cgraph_clone_node.
4639 * ipa-prop.c (ipa_propagate_indirect_call_infos): Do not call
4640 ipa_free_edge_args_substructures.
4641 (ipa_edge_duplication_hook): Only add edges from inlined nodes to
4642 rdesc linked list. Do not assert rdesc edges have inlined caller.
4643 Assert we have found an rdesc in the rdesc list.
4645 2013-06-24 Richard Biener <rguenther@suse.de>
4647 * pointer-set.h (struct pointer_set_t): Move here from pointer-set.c.
4648 (pointer_set_lookup): Declare.
4649 (class pointer_map): New template class implementing a
4650 generic pointer to T map.
4651 (pointer_map<T>::pointer_map, pointer_map<T>::~pointer_map,
4652 pointer_map<T>::contains, pointer_map<T>::insert,
4653 pointer_map<T>::traverse): New functions.
4654 * pointer-set.c (struct pointer_set_t): Moved to pointer-set.h.
4655 (pointer_set_lookup): New function.
4656 (pointer_set_contains): Use pointer_set_lookup.
4657 (pointer_set_insert): Likewise.
4658 (insert_aux): Remove.
4659 (struct pointer_map_t): Embed a pointer_set_t.
4660 (pointer_map_create): Adjust.
4661 (pointer_map_destroy): Likewise.
4662 (pointer_map_contains): Likewise.
4663 (pointer_map_insert): Likewise.
4664 (pointer_map_traverse): Likewise.
4665 * tree-streamer.h (struct streamer_tree_cache_d): Use a
4666 pointer_map<unsigned> instead of a pointer_map_t.
4667 * tree-streamer.c (streamer_tree_cache_insert_1): Adjust.
4668 (streamer_tree_cache_lookup): Likewise.
4669 (streamer_tree_cache_create): Likewise.
4670 (streamer_tree_cache_delete): Likewise.
4671 * lto-streamer.h (struct lto_tree_ref_encoder): Use a
4672 pointer_map<unsigned> instead of a pointer_map_t.
4673 (lto_init_tree_ref_encoder): Adjust.
4674 (lto_destroy_tree_ref_encoder): Likewise.
4675 * lto-section-out.c (lto_output_decl_index): Likewise.
4676 (lto_record_function_out_decl_state): Likewise.
4677 * dominance.c (iterate_fix_dominators): Use pointer_map<int>.
4679 2013-06-24 Richard Biener <rguenther@suse.de>
4681 PR tree-optimization/57488
4682 * tree-ssa-pre.c (insert): Clear NEW sets before each iteration.
4684 2013-06-24 Alan Modra <amodra@gmail.com>
4686 * config/rs6000/rs6000.c (vspltis_constant): Correct for little-endian.
4687 (gen_easy_altivec_constant): Likewise.
4688 * config/rs6000/predicates.md (easy_vector_constant_add_self,
4689 easy_vector_constant_msb): Likewise.
4691 2013-06-23 Jakub Jelinek <jakub@redhat.com>
4694 * common/config/i386/i386-common.c (ix86_handle_option): For OPT_mlzcnt
4695 add missing return true.
4697 2013-06-23 Oleg Endo <olegendo@gcc.gnu.org>
4700 * config/sh/predicates.md (general_extend_operand): Invoke
4701 general_movsrc_operand for memory operands.
4702 (general_movsrc_operand): Allow reg+reg addressing, do not use
4703 general_operand for memory operands.
4705 2013-06-23 Sriraman Tallam <tmsriram@google.com>
4707 * config/i386/i386.c (ix86_pragma_target_parse): Restore target
4708 when current target options does not apply.
4709 * config/i386/i386-protos.h (ix86_reset_previous_fndecl): New function.
4710 * config/i386/i386.c (ix86_reset_previous_fndecl): Ditto.
4711 * config/i386/bmiintrin.h: Pass appropriate target
4712 attributes to header.
4713 * config/i386/mmintrin.h: Ditto.
4714 * config/i386/nmmintrin.h: Ditto.
4715 * config/i386/avx2intrin.h: Ditto.
4716 * config/i386/fxsrintrin.h: Ditto.
4717 * config/i386/tbmintrin.h: Ditto.
4718 * config/i386/xsaveintrin.h: Ditto.
4719 * config/i386/f16cintrin.h: Ditto.
4720 * config/i386/xtestintrin.h: Ditto.
4721 * config/i386/xsaveoptintrin.h: Ditto.
4722 * config/i386/bmi2intrin.h: Ditto.
4723 * config/i386/lzcntintrin.h: Ditto.
4724 * config/i386/smmintrin.h: Ditto.
4725 * config/i386/wmmintrin.h: Ditto.
4726 * config/i386/x86intrin.h: Remove all header include guards.
4727 * config/i386/prfchwintrin.h: Ditto.
4728 * config/i386/pmmintrin.h: Ditto.
4729 * config/i386/tmmintrin.h: Ditto.
4730 * config/i386/xmmintrin.h: Ditto.
4731 * config/i386/popcntintrin.h: Ditto.
4732 * config/i386/rdseedintrin.h: Ditto.
4733 * config/i386/ammintrin.h: Ditto.
4734 * config/i386/emmintrin.h: Ditto.
4735 * config/i386/immintrin.h: Remove all header include guards.
4736 * config/i386/fma4intrin.h: Ditto.
4737 * config/i386/lwpintrin.h: Ditto.
4738 * config/i386/xopintrin.h: Ditto.
4739 * config/i386/ia32intrin.h: Ditto.
4740 * config/i386/avxintrin.h: Ditto.
4741 * config/i386/rtmintrin.h: Ditto.
4742 * config/i386/fmaintrin.h: Ditto.
4743 * config/i386/mm3dnow.h: Ditto.
4745 2013-06-22 Sriraman Tallam <tmsriram@google.com>
4747 * common/config/i386/i386-common.c: Handle LZCNT.
4749 2013-06-22 Andi Kleen <ak@linux.intel.com>
4751 * doc/extend.texi: Use __atomic_store_n instead of
4752 __atomic_store in HLE example.
4754 2013-06-22 Oleg Endo <olegendo@gcc.gnu.org>
4756 * config/sh/sh.c: Remove <cstdlib> workaround.
4758 2013-06-21 Andi Kleen <ak@linux.intel.com>
4760 * doc/extend.texi: Dont use __atomic_clear in HLE example. Fix typo.
4762 2013-06-21 Andi Kleen <ak@linux.intel.com>
4764 * doc/extend.texi: Document that __atomic_clear and
4765 __atomic_test_and_set should only be used with bool.
4767 2013-06-20 Jan Hubicka <jh@suse.cz>
4769 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Use
4771 * tree.c (decls_same_for_odr): New function.
4772 (same_for_edr): New function.
4773 (types_same_for_odr): New function.
4774 (get_binfo_at_offset): Use it.
4775 * tree.h (types_same_for_odr): Declare.
4777 2013-06-20 Oleg Endo <olegendo@gcc.gnu.org>
4778 Jason Merrill <jason@redhat.com>
4780 * system.h: Include <cstdlib> as well as <stdlib.h>.
4782 2013-06-20 Uros Bizjak <ubizjak@gmail.com>
4785 * config/i386/i386.c (construct_container): Report error if
4786 long double is used with disabled x87 float returns.
4788 2013-06-20 Jan Hubicka <jh@suse.cz>
4790 * lto-cgraph.c (input_symtab): Do not set cgraph state.
4792 2013-06-20 Joern Rennecke <joern.rennecke@embecosm.com>
4794 PR rtl-optimization/57425
4795 PR rtl-optimization/57569
4796 * alias.c (write_dependence_p): Remove parameters mem_mode and
4797 canon_mem_addr. Add parameters x_mode, x_addr and x_canonicalized.
4798 Changed all callers.
4799 (canon_anti_dependence): Get comments and semantics in sync.
4800 Add parameter mem_canonicalized. Changed all callers.
4801 * rtl.h (canon_anti_dependence): Update prototype.
4803 2013-06-20 Richard Biener <rguenther@suse.de>
4805 * data-streamer-in.c (streamer_read_uhwi): Optimize single
4806 byte case, inline streamer_read_uchar and defer section
4809 2013-06-20 Richard Biener <rguenther@suse.de>
4811 PR tree-optimization/57584
4812 * tree-ssa-loop-niter.c (expand_simple_operations): Avoid including
4813 SSA names into the expanded expression that take part in
4814 abnormal coalescing.
4816 2013-06-19 Sharad Singhai <singhai@google.com>
4818 * gcov.c (print_usage): Handle new option.
4819 (process_args): Ditto.
4820 (get_gcov_intermediate_filename): New function.
4821 (output_intermediate_file): New function.
4822 (output_gcov_file): New function
4823 (generate_results): Handle new option.
4824 (release_function): Relase demangled name.
4825 (read_graph_file): Handle demangled name.
4826 (output_lines): Ditto.
4827 * doc/gcov.texi: Document gcov intermediate format.
4829 2013-06-19 Vladimir Makarov <vmakarov@redhat.com>
4832 * lra.c (emit_add3_insn, emit_add2_insn): New functions.
4833 (lra_emit_add): Use the functions. Add comment about Y as an
4836 2013-06-19 David Edelsohn <dje.gcc@gmail.com>
4839 * collect2.c (collect_atexit): New.
4840 (collect_exit): Delete.
4841 (main): Register collect_atexit with atexit.
4842 (collect_wait): Change collect_exit to exit.
4844 * collect2.h (collect_exit): Delete.
4845 * tlink.c (do_tlink): Rename exit to ret. Change collect_exit to exit.
4847 2013-06-19 Wei Mi <wmi@google.com>
4849 PR rtl-optimization/57518
4850 * ira.c (set_paradoxical_subreg): Set pdx_subregs[regno]
4851 if regno is used in paradoxical subreg.
4852 (update_equiv_regs): Check pdx_subregs[regno] before
4853 set a reg to be equivalent with a mem.
4855 2013-06-19 Matthias Klose <doko@ubuntu.com>
4858 * file-find.h (find_a_file): Add a mode parameter.
4859 * file-find.c (find_a_file): Likewise.
4860 * gcc-ar.c (main): Call find_a_file with R_OK for the plugin,
4861 with X_OK for the executables.
4862 * collect2.c (main): Call find_a_file with X_OK.
4864 2013-06-19 Steve Ellcey <sellcey@mips.com>
4867 * config/mips/mips.md (casesi_internal_mips16_<mode>):
4868 Use NEXT_INSN instead of next_real_insn.
4870 2013-06-19 Jan Hubicka <jh@suse.cz>
4872 * cgraph.h (const_value_known_p): Replace by ...
4873 (ctor_for_folding): .. this one.
4874 * cgraphunit.c (process_function_and_variable_attributes): Use it.
4875 * lto-cgraph.c (compute_ltrans_boundary): Use ctor_for_folding.
4876 * expr.c (expand_expr_real_1): Likewise.
4877 (string_constant): Likewise.
4878 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Likewise.
4879 * ipa.c (process_references): Likewise.
4880 (symtab_remove_unreachable_nodes): Likewise.
4881 * ipa-inline-analysis.c (param_change_prob): Likewise.
4882 * gimple-fold.c (canonicalize_constructor_val): Likewise.
4883 (get_base_constructor): Likwise.
4884 * varpool.c (varpool_remove_node): Likewise.
4885 (varpool_remove_initializer): LIkewise.
4886 (dump_varpool_node): LIkwise.
4887 (const_value_known_p): Rewrite to ...
4888 (ctor_for_folding): ... this one.
4890 2013-06-19 Jakub Jelinek <jakub@redhat.com>
4893 * gcc-ar.c (main): If not CROSS_DIRECTORY_STRUCTURE, look for
4894 PERSONALITY in $PATH derived prefixes.
4896 2013-06-19 Jeff Law <law@redhat.com>
4898 * tree-ssa-forwprop.c (simplify_bitwise_binary_boolean): Fix typo
4901 * tree-ssa-forwprop.c (simplify_bitwise_binary_boolean): New function.
4902 (simplify_bitwise_binary): Use it to simpify certain binary ops on
4905 2013-06-19 Sofiane Naci <sofiane.naci@arm.com>
4907 * config/arm/vfp.md: Move VFP instruction classification documentation
4909 * config/arm/arm.md: ... here. Update instruction classification
4912 2013-06-19 Richard Earnshaw <rearnsha@arm.com>
4914 arm.md (split for eq(reg, 0)): Add variants for ARMv5 and Thumb2.
4915 (peepholes for eq(reg, not-0)): Ensure condition register is dead after
4916 pattern. Use more efficient sequences on ARMv5 and Thumb2.
4918 2013-06-19 Steven Bosscher <steven@gcc.gnu.org>
4921 * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
4922 with NEXT_INSN. Use tablejump_p to check for jump table data
4925 2013-06-19 Paolo Carlini <paolo.carlini@oracle.com>
4928 * doc/cpp.texi [Standard Predefined Macros, __cplusplus]: Document
4929 that now in C++ the value is correct per the C++ standards.
4931 2013-06-19 Richard Biener <rguenther@suse.de>
4933 * expr.c (expand_expr_real_1): Use SCOPE_FILE_SCOPE_P to check
4936 2013-06-19 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
4939 2013-06-18 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
4942 * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
4943 with next_active_insn.
4945 2013-06-18 Sriraman Tallam <tmsriram@google.com>
4947 * ipa-inline.c (inline_always_inline_functions): Pretend always_inline
4948 functions are inlined during failures to flag an error.
4949 * tree-inline.c (expand_call_inline): Allow the error to be flagged
4950 in early inline pass.
4952 2013-06-18 H.J. Lu <hongjiu.lu@intel.com>
4954 * config/i386/i386.c (initial_ix86_tune_features): Fix a typo
4957 2013-06-18 Julian Brown <julian@codesourcery.com>
4959 * config/arm/arm.c (neon_vector_mem_operand): Add strict argument.
4960 Permit virtual register pre-reload if !strict.
4961 (coproc_secondary_reload_class): Adjust for neon_vector_mem_operand
4963 * config/arm/arm-protos.h (neon_vector_mem_operand): Adjust
4965 * config/arm/neon.md (movmisalign<mode>): Use
4966 neon_perm_struct_or_reg_operand instead of
4967 neon_struct_or_register_operand.
4968 (*movmisalign<mode>_neon_load, *movmisalign<mode>_neon_store): Use
4969 neon_permissive_struct_operand instead of neon_struct_operand.
4970 * config/arm/constraints.md (Un, Um, Us): Adjust calls to
4971 neon_vector_mem_operand.
4972 * config/arm/predicates.md (neon_struct_operand): Adjust call to
4973 neon_vector_mem_operand.
4974 (neon_permissive_struct_operand): New.
4975 (neon_struct_or_register_operand): Rename to...
4976 (neon_perm_struct_or_reg_operand): This. Adjust call to
4977 neon_vector_mem_operand.
4979 2013-06-18 Richard Biener <rguenther@suse.de>
4981 * Makefile.in (LTO_STREAMER_H): Add pointer-set.h dependency.
4982 * lto-streamer.h: Include pointer-set.h.
4983 (struct lto_decl_slot): Remove.
4984 (struct lto_tree_ref_encoder): Make tree_hash_table a pointer-map.
4985 Remove next_index entry.
4986 (lto_hash_decl_slot_node, lto_eq_decl_slot_node,
4987 lto_hash_type_slot_node, lto_eq_type_slot_node): Remove.
4988 (lto_init_tree_ref_encoder): Adjust.
4989 (lto_destroy_tree_ref_encoder): Likewise.
4990 * lto-section-out.c (lto_hash_decl_slot_node, lto_eq_decl_slot_node,
4991 lto_hash_type_slot_node, lto_eq_type_slot_node): Remove.
4992 (lto_output_decl_index): Adjust.
4993 (lto_new_out_decl_state): Likewise.
4994 (lto_record_function_out_decl_state): Likewise.
4995 * lto-streamer-out.c (copy_function): Likewise.
4997 2013-06-18 Richard Biener <rguenther@suse.de>
4999 * Makefile.in (cgraphunit.o): Add $(CFGLOOP_H) dependency.
5000 * cgraphunit.c: Include cfgloop.h.
5001 (init_lowered_empty_function): Initialize the loop tree.
5002 (assemble_thunk): Insert new BBs into loops.
5004 2013-06-18 Richard Biener <rguenther@suse.de>
5006 * tree-streamer.h (streamer_tree_cache_create): Adjust prototype.
5007 * tree-streamer.c (streamer_tree_cache_create): Make maintaining
5008 the map from cache entry to cache index optional.
5009 (streamer_tree_cache_replace_tree): Adjust accordingly.
5010 (streamer_tree_cache_append): Likewise.
5011 (streamer_tree_cache_delete): Likewise.
5012 * lto-streamer-in.c (lto_data_in_create): Do not maintain the
5013 streamer cache map from cache entry to cache index.
5014 * lto-streamer-out.c (create_output_block): Adjust.
5016 2013-06-18 Sofiane Naci <sofiane.naci@arm.com>
5018 * config/arm/arm.md (attribute "insn"): Move multiplication and
5019 division attributes to...
5020 (attribute "type"): ... here. Remove mult.
5021 (attribute "mul32"): New attribute.
5022 (attribute "mul64"): Add umaal.
5023 (*arm_mulsi3): Update attributes.
5024 (*arm_mulsi3_v6): Likewise.
5025 (*thumb_mulsi3): Likewise.
5026 (*thumb_mulsi3_v6): Likewise.
5027 (*mulsi3_compare0): Likewise.
5028 (*mulsi3_compare0_v6): Likewise.
5029 (*mulsi_compare0_scratch): Likewise.
5030 (*mulsi_compare0_scratch_v6): Likewise.
5031 (*mulsi3addsi): Likewise.
5032 (*mulsi3addsi_v6): Likewise.
5033 (*mulsi3addsi_compare0): Likewise.
5034 (*mulsi3addsi_compare0_v6): Likewise.
5035 (*mulsi3addsi_compare0_scratch): Likewise.
5036 (*mulsi3addsi_compare0_scratch_v6): Likewise.
5037 (*mulsi3subsi): Likewise.
5038 (*mulsidi3adddi): Likewise.
5039 (*mulsi3addsi_v6): Likewise.
5040 (*mulsidi3adddi_v6): Likewise.
5041 (*mulsidi3_nov6): Likewise.
5042 (*mulsidi3_v6): Likewise.
5043 (*umulsidi3_nov6): Likewise.
5044 (*umulsidi3_v6): Likewise.
5045 (*umulsidi3adddi): Likewise.
5046 (*umulsidi3adddi_v6): Likewise.
5047 (*smulsi3_highpart_nov6): Likewise.
5048 (*smulsi3_highpart_v6): Likewise.
5049 (*umulsi3_highpart_nov6): Likewise.
5050 (*umulsi3_highpart_v6): Likewise.
5051 (mulhisi3): Likewise.
5052 (*mulhisi3tb): Likewise.
5053 (*mulhisi3bt): Likewise.
5054 (*mulhisi3tt): Likewise.
5055 (maddhisi4): Likewise.
5056 (*maddhisi4tb): Likewise.
5057 (*maddhisi4tt): Likewise.
5058 (maddhidi4): Likewise.
5059 (*maddhidi4tb): Likewise.
5060 (*maddhidi4tt): Likewise.
5062 (udivsi3): Likewise.
5063 * config/arm/thumb2.md (thumb2_mulsi_short): Update attributes.
5064 (thumb2_mulsi_short_compare0): Likewise.
5065 (thumb2_mulsi_short_compare0_scratch): Likewise.
5066 * config/arm/arm1020e.md (1020mult1): Update attribute change.
5067 (1020mult2): Likewise.
5068 (1020mult3): Likewise.
5069 (1020mult4): Likewise.
5070 (1020mult5): Likewise.
5071 (1020mult6): Likewise.
5072 * config/arm/cortex-a15.md (cortex_a15_mult32): Update attribute
5074 (cortex_a15_mult64): Likewise.
5075 (cortex_a15_sdiv): Likewise.
5076 (cortex_a15_udiv): Likewise.
5077 * config/arm/arm1026ejs.md (mult1): Update attribute change.
5083 * config/arm/marvell-pj4.md (pj4_ir_mul): Update attribute change.
5084 (pj4_ir_div): Likewise.
5085 * config/arm/arm1136jfs.md (11_mult1): Update attribute change.
5086 (11_mult2): Likewise.
5087 (11_mult3): Likewise.
5088 (11_mult4): Likewise.
5089 (11_mult5): Likewise.
5090 (11_mult6): Likewise.
5091 (11_mult7): Likewise.
5092 * config/arm/cortex-a8.md (cortex_a8_mul): Update attribute change.
5093 (cortex_a8_mla): Likewise.
5094 (cortex_a8_mull): Likewise.
5095 (cortex_a8_smulwy): Likewise.
5096 (cortex_a8_smlald): Likewise.
5097 * config/arm/cortex-m4.md (cortex_m4_alu): Update attribute change.
5098 * config/arm/cortex-r4.md (cortex_r4_mul_4): Update attribute change.
5099 (cortex_r4_mul_3): Likewise.
5100 (cortex_r4_mla_4): Likewise.
5101 (cortex_r4_mla_3): Likewise.
5102 (cortex_r4_smlald): Likewise.
5103 (cortex_r4_mull): Likewise.
5104 (cortex_r4_sdiv): Likewise.
5105 (cortex_r4_udiv): Likewise.
5106 * config/arm/cortex-a7.md (cortex_a7_mul): Update attribute change.
5107 (cortex_a7_idiv): Likewise.
5108 * config/arm/arm926ejs.md (9_mult1): Update attribute change.
5109 (9_mult2): Likewise.
5110 (9_mult3): Likewise.
5111 (9_mult4): Likewise.
5112 (9_mult5): Likewise.
5113 (9_mult6): Likewise.
5114 * config/arm/cortex-a53.md (cortex_a53_mul): Update attribute change.
5115 (cortex_a53_sdiv): Likewise.
5116 (cortex_a53_udiv): Likewise.
5117 * config/arm/fa726te.md (726te_mult_op): Update attribute change.
5118 * config/arm/fmp626.md (mp626_mult1): Update attribute change.
5119 (mp626_mult2): Likewise.
5120 (mp626_mult3): Likewise.
5121 (mp626_mult4): Likewise.
5122 * config/arm/fa526.md (526_mult1): Update attribute change.
5123 (526_mult2): Likewise.
5124 * config/arm/arm-generic.md (mult): Update attribute change.
5125 (mult_ldsched_strongarm): Likewise.
5126 (mult_ldsched): Likewise.
5127 (multi_cycle): Likewise.
5128 * config/arm/cortex-a5.md (cortex_a5_mul): Update attribute change.
5129 * config/arm/fa606te.md (606te_mult1): Update attribute change.
5130 (606te_mult2): Likewise.
5131 (606te_mult3): Likewise.
5132 (606te_mult4): Likewise.
5133 * config/arm/cortex-a9.md (cortex_a9_mult16): Update attribute change.
5134 (cortex_a9_mac16): Likewise.
5135 (cortex_a9_multiply): Likewise.
5136 (cortex_a9_mac): Likewise.
5137 (cortex_a9_multiply_long): Likewise.
5138 * config/arm/fa626te.md (626te_mult1): Update attribute change.
5139 (626te_mult2): Likewise.
5140 (626te_mult3): Likewise.
5141 (626te_mult4): Likewise.
5143 2013-06-18 Richard Biener <rguenther@suse.de>
5146 * lto-symtab.c (lto_symtab_merge_decls): Process nodes properly.
5148 2013-06-18 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
5151 * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
5152 with next_active_insn.
5154 2013-06-18 Alan Modra <amodra@gmail.com>
5156 * config/rs6000/rs6000.h (enum data_align): New.
5157 (LOCAL_ALIGNMENT, DATA_ALIGNMENT): Use rs6000_data_alignment.
5158 (DATA_ABI_ALIGNMENT): Define.
5159 (CONSTANT_ALIGNMENT): Correct comment.
5160 * config/rs6000/rs6000-protos.h (rs6000_data_alignment): Declare.
5161 * config/rs6000/rs6000.c (rs6000_data_alignment): New function.
5163 2013-06-17 David Malcolm <dmalcolm@redhat.com>
5165 * ggc-page.c (ggc_pch_write_object) <d>: Remove erroneous
5166 ATTRIBUTE_UNUSED marking.
5168 2013-06-17 Sofiane Naci <sofiane.naci@arm.com>
5170 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>): Add r<-w
5171 alternative and update.
5172 (aarch64_dup_lanedi): Delete.
5173 * config/aarch64/arm_neon.h (vdup<bhsd>_lane_*): Update.
5174 * config/aarch64/aarch64-simd-builtins.def: Update.
5176 2013-06-17 Richard Biener <rguenther@suse.de>
5178 * lto-streamer.h (enum LTO_tags): Add LTO_tree_scc.
5179 (lto_input_scc): Declare.
5180 (lto_input_tree_1): Likewise.
5181 (struct lto_stats_d): Add num_tree_bodies_output and
5182 num_pickle_refs_output.
5183 * lto-streamer-in.c (lto_read_body): Use streamer_tree_cache_get_tree.
5184 (lto_read_tree_1): Split out from ...
5185 (lto_read_tree): ... this.
5186 (lto_input_scc): New function.
5187 (lto_input_tree_1): Split out from ...
5188 (lto_input_tree): ... this. Handle LTO_tree_scc.
5189 (lto_data_in_create): Create the streamer cache without hashes.
5190 * lto-streamer-out.c (create_output_block): Create the streamer
5191 cache with hashes when not doing WPA.
5192 (lto_write_tree_1): Split out from ...
5193 (lto_write_tree): ... this.
5194 (get_symbol_initial_value): New function.
5195 (lto_output_tree_1): Split out from ...
5196 (lto_output_tree): ... this. Write trees as series of SCCs
5197 using a DFS walk via DFS_write_tree.
5198 (struct sccs, struct scc_entry): New types.
5199 (next_dfs_num, sccstack, sccstate, sccstate_obstack): New globals.
5200 (DFS_write_tree_body): New function.
5201 (DFS_write_tree): Likewise.
5202 (hash_tree): Likewise.
5203 (scc_entry_compare): Likewise.
5204 (hash_scc): Likewise.
5205 (tree_is_indexable): DEBUG_EXPR_DECLs are local entities.
5206 * tree-streamer-in.c (lto_input_ts_list_tree_pointers): Stream
5207 TREE_CHAIN as regular reference.
5208 (streamer_read_integer_cst): Remove.
5209 (streamer_get_pickled_tree): Adjust.
5210 * tree-streamer-out.c (streamer_write_chain): Disable streaming
5211 of DECL_EXTERNALs in BLOCK_VARS for now.
5212 (write_ts_list_tree_pointers): Stream TREE_CHAIN as regular
5214 * tree-streamer.c (streamer_tree_cache_add_to_node_array):
5215 Add hash value argument and record that if hashes are recorded
5217 (streamer_tree_cache_insert_1): Adjust.
5218 (streamer_tree_cache_insert): Likewise.
5219 (streamer_tree_cache_insert_at): Rename to ...
5220 (streamer_tree_cache_replace_tree): ... this and adjust.
5221 (streamer_tree_cache_append): Adjust.
5222 (record_common_node): Likewise.
5223 (streamer_tree_cache_create): Add argument whether to
5224 record hash values together with trees.
5225 (streamer_tree_cache_delete): Adjust.
5226 * tree-streamer.h (struct streamer_tree_cache_d): Add
5228 (streamer_read_integer_cst): Remove.
5229 (streamer_tree_cache_insert): Adjust.
5230 (streamer_tree_cache_append): Likewise.
5231 (streamer_tree_cache_insert_at): Rename to ...
5232 (streamer_tree_cache_replace_tree): ... this and adjust.
5233 (streamer_tree_cache_create): Add argument whether to record hashes.
5234 (streamer_tree_cache_get): Rename to ...
5235 (streamer_tree_cache_get_tree): ... this.
5236 (streamer_tree_cache_get_hash): New function.
5237 * tree.c (cache_integer_cst): New function.
5238 * tree.h (cache_integer_cst): Declare.
5239 (ANON_AGGRNAME_FORMAT, ANON_AGGRNAME_P): Move here from cp/cp-tree.h.
5240 * lto-symtab.c (lto_varpool_replace_node): Only release
5241 DECL_INITIAL of non-prevailing decls.
5242 * varpool.c (varpool_remove_initializer): Do not release
5243 DECL_INITIAL when we are still in CGRAPH_LTO_STREAMING.
5245 2013-06-16 Jürgen Urban <JuergenUrban@gmx.de>
5247 * config/mips/mips.h (ISA_HAS_MUL3): Include TARGET_MIPS5900.
5248 (ISA_HAS_MULT, ISA_HAS_DMULT, ISA_HAS_DIV, ISA_HAS_DDIV): New macros.
5249 * config/mips/mips.md (mul<mode>3, mul<mode>3_internal)
5250 (mul<mode>3_r4000): Require ISA_HAS_<D>MULT.
5251 (mul<mode>3_mul3): Handle TARGET_MIPS5900.
5252 (mulsidi3_64bit_dmul): Remove redundant TARGET_64BIT test.
5253 (<su>muldi3_highpart, <su>muldi3_highpart_internal, <u>mulditi3)
5254 (<u>mulditi3_internal, <u>mulditi3_r4000): Require ISA_HAS_DMULT
5255 instead of TARGET_64BIT.
5256 (divmod<mode>4, udivmod<mode>4, <u>divmod<GPR:mode>4_hilo_<HILO:mode>):
5257 Require ISA_HAS_<D>DIV.
5259 2013-06-16 Richard Sandiford <rdsandiford@googlemail.com>
5261 * config.gcc (mips*-mti-linux*, mips64*-*-linux*, mipsisa64*-*-linux*)
5262 (mips*-*-linux*): Move default with_llsc setting to where other
5264 (mips*-*-vxworks*): Move with_arch default from with_cpu block to
5266 (mips64r5900-*-*, mips64r5900el-*-*, mipsr5900-*-*, mipsr5900el-*-*):
5267 Likewise. Remove default with_tune setting. Move default float
5268 setting to its own block. Handle with_llsc in the same block as above.
5270 2013-06-16 Joern Rennecke <joern.rennecke@embecosm.com>
5272 PR rtl-optimization/57425
5273 PR rtl-optimization/57569
5274 * alias.c (write_dependence_p): Add new parameters mem_mode,
5275 canon_mem_addr and mem_canonicalized. Change type of writep to bool.
5276 Changed all callers.
5277 (canon_anti_dependence): New function.
5278 * cse.c (check_dependence): Use canon_anti_dependence.
5279 * cselib.c (cselib_invalidate_mem): Likewise.
5280 * rtl.h (canon_anti_dependence): Declare.
5282 2013-06-16 Jürgen Urban <JuergenUrban@gmx.de>
5284 * config/mips/mips.h (ISA_HAS_LL_SC): Exclude TARGET_MIPS5900.
5285 * config/mips/mips.c (mips_start_ll_sc_sync_block): Output
5286 ".set mips3" for 64-bit targets.
5288 2013-06-15 Dehao Chen <dehao@google.com>
5290 * tree-flow.h (gimple_check_call_matching_types): Add new argument.
5291 * gimple-low.c (gimple_check_call_matching_types): Likewise.
5292 (gimple_check_call_args): Likewise.
5293 * value-prof.c (check_ic_target): Likewise.
5294 * ipa-inline.c (early_inliner): Likewise.
5295 * ipa-prop.c (update_indirect_edges_after_inlining): Likewise.
5296 * cgraph.c (cgraph_create_edge_1): Likewise.
5297 (cgraph_make_edge_direct): Likewise.
5299 2013-06-14 Michael Meissner <meissner@linux.vnet.ibm.com>
5302 * config/rs6000/rs6000.md (mov<mode>_ppc64): Call
5303 rs6000_output_move_128bit to handle emitting quad memory
5304 operations. Set attribute length to 8 bytes.
5306 2013-06-14 Vidya Praveen <vidyapraveen@arm.com>
5308 * config/aarch64/aarch64-simd.md (aarch64_<su>mlal_lo<mode>):
5310 (aarch64_<su>mlal_hi<mode>, aarch64_<su>mlsl_lo<mode>): Likewise.
5311 (aarch64_<su>mlsl_hi<mode>, aarch64_<su>mlal<mode>): Likewise.
5312 (aarch64_<su>mlsl<mode>): Likewise.
5314 2013-06-14 Mike Stump <mikestump@comcast.net>
5316 * Makefile.in (TARGET_H): Add insn-codes.h.
5318 2013-06-14 Alan Modra <amodra@gmail.com>
5322 * expr.c (expand_expr_real_1 <normal_inner_ref>): Pass
5323 EXPAND_MEMORY and EXPAND_WRITE to recursive call. Don't use
5324 bitfield expansion when EXPAND_MEMORY.
5325 (expand_expr_real_1 <VIEW_CONVERT_EXPR>): Pass modifier likewise.
5327 2013-06-13 Michael Meissner <meissner@linux.vnet.ibm.com>
5329 * config/rs6000/rs6000.c (rs6000_option_override_internal): Move
5330 test for clearing quad memory on 32-bit later.
5332 2013-06-13 Marc Glisse <marc.glisse@inria.fr>
5334 * fold-const.c (negate_expr_p): Handle VECTOR_CST.
5335 (fold_negate_expr): Likewise.
5336 (fold_real_zero_addition_p): Handle vectors.
5337 (fold_binary_loc) <PLUS_EXPR, MINUS_EXPR>: Likewise.
5339 2013-06-14 Alan Modra <amodra@gmail.com>
5341 * varasm.c (force_const_mem): Revert 2013-06-07 change.
5343 2013-06-13 Jan Hubicka <jh@suse.cz>
5345 * ipa.c (cgraph_externally_visible_p, varpool_externally_visible_p):
5346 Local comdats are not externally visible.
5347 * symtab.c (dump_symtab_base): Dump externally visible.
5348 (verify_symtab_base): Verify back links in the symtab hash.
5350 2013-06-13 Bin Cheng <bin.cheng@arm.com>
5352 * fold-const.c (operand_equal_p): Consider NOP_EXPR and
5353 CONVERT_EXPR as equal nodes.
5355 2013-06-13 Bin Cheng <bin.cheng@arm.com>
5357 * rtlanal.c (noop_move_p): Check the code to be executed for COND_EXEC.
5359 2013-06-13 Marc Glisse <marc.glisse@inria.fr>
5361 * tree-ssa-forwprop.c (simplify_bitwise_binary, associate_plusminus):
5362 Generalize to complex and vector.
5363 * tree.c (build_all_ones_cst): New function.
5364 * tree.h (build_all_ones_cst): Declare it.
5366 2013-06-13 Alan Modra <amodra@gmail.com>
5368 * config/rs6000/rs6000.h (LONG_DOUBLE_LARGE_FIRST): Define.
5369 * config/rs6000/rs6000.md (signbittf2): New insn.
5370 (extenddftf2_internal): Use LONG_DOUBLE_LARGE_FIRST.
5371 (abstf2_internal, cmptf_internal2): Likewise.
5372 * config/rs6000/spe.md (spe_abstf2_cmp, spe_abstf2_tst): Likewise.
5374 2013-06-12 Michael Meissner <meissner@linux.vnet.ibm.com>
5375 Pat Haugen <pthaugen@us.ibm.com>
5376 Peter Bergner <bergner@vnet.ibm.com>
5378 * config/rs6000/rs6000.c (emit_load_locked): Add support for
5379 power8 byte, half-word, and quad-word atomic instructions.
5380 (emit_store_conditional): Likewise.
5381 (rs6000_expand_atomic_compare_and_swap): Likewise.
5382 (rs6000_expand_atomic_op): Likewise.
5384 * config/rs6000/sync.md (larx): Add new modes for power8.
5386 (AINT): New mode iterator to include TImode as well as normal
5387 integer modes on power8.
5388 (fetchop_pred): Use int_reg_operand instead of gpc_reg_operand so
5389 that VSX registers are not considered. Use AINT mode iterator
5390 instead of INT1 to allow inclusion of quad word atomic operations
5392 (load_locked<mode>): Likewise.
5393 (store_conditional<mode>): Likewise.
5394 (atomic_compare_and_swap<mode>): Likewise.
5395 (atomic_exchange<mode>): Likewise.
5396 (atomic_nand<mode>): Likewise.
5397 (atomic_fetch_<fetchop_name><mode>): Likewise.
5398 (atomic_nand_fetch<mode>): Likewise.
5399 (mem_thread_fence): Use gen_loadsync_<mode> instead of enumerating
5401 (ATOMIC): On power8, add QImode, HImode modes.
5402 (load_locked<QHI:mode>_si): Varients of load_locked for QI/HI
5403 modes that promote to SImode.
5404 (load_lockedti): Convert TImode arguments to PTImode, so that we
5405 get a guaranteed even/odd register pair.
5406 (load_lockedpti): Likewise.
5407 (store_conditionalti): Likewise.
5408 (store_conditionalpti): Likewise.
5410 * config/rs6000/rs6000.md (QHI): New mode iterator for power8
5411 atomic load/store instructions.
5414 2013-06-12 Richard Sandiford <rdsandiford@googlemail.com>
5416 * config/mips/mips.md (extended_mips16): Include GOT and constant-pool
5418 (insn_count): New attribute, with most cases extracted from...
5419 (length): ...here. Redefine most cases in terms of insn_count.
5420 (single_insn): Delete.
5421 (can_delay): Use insn_count to check for single instructions.
5422 (*mul<mode>3_r4300, mul<mode>3_r4000, *mul_acc_si, *mul_acc_si_r3900)
5423 (*msac_using_macc, *mul_sub_si, <u>mulsidi3_32bit_r4000)
5424 (<u>mulsidi3_64bit_r4000, <su>muldi3_highpart_internal)
5425 (<su>mulsi3_highpart_split, <su>muldi3_highpart_internal)
5426 (<u>mulditi3_r4000, *div<mode>3, *recip<mode>3, divmod<mode>4)
5427 (udivmod<mode>4, sqrt<mode>2, *rsqrt<mode>a, *rsqrt<mode>b)
5428 (fix_truncdfsi2_macro, fix_truncsfsi2_macro, *lea_high64)
5429 (*lea64, cprestore_<mode>, clear_hazard_<mode>, <unnamed insn>)
5430 (casesi_internal_mips16_<mode>, *tls_get_tp_<mode>_split)
5431 (tls_get_tp_mips16, *tls_get_tp_mips16_call_<mode>): Use "insn_count"
5432 rather than "length".
5433 (tls_get_tp_<mode>): Likewise. Remove redundant "no_delay" attribute.
5434 * config/mips/mips-ps-3d.md (mips_c_cond_4s, mips_cabs_cond_4s):
5435 Use "insn_count" rather than "length".
5436 * config/mips/mips-dsp.md
5437 (mips_l<SHORT:size><u>x_ext<GPR:mode>_<P:mode>)
5438 (mips_l<GPR:size>x_<P:mode>, *mips_lw<u>x_<P:mode>_ext): Remove
5441 2013-06-12 Marc Glisse <marc.glisse@inria.fr>
5443 PR tree-optimization/57361
5444 * tree-ssa-dse.c (dse_possible_dead_store_p): Handle self-assignment.
5446 2013-06-12 Sofiane Naci <sofiane.naci@arm.com>
5448 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Convert
5450 (aarch64_simd_combine<mode>): New instruction expansion.
5451 * config/aarch64/aarch64-protos.h (aarch64_split_simd_combine): New
5453 * config/aarch64/aarch64.c (aarch64_split_combine): New function.
5454 * config/aarch64/iterators.md (Vdbl): Add entry for DF.
5456 2013-06-12 Jan Hubicka <jh@suse.cz>
5458 * cgraph.c (verify_edge_corresponds_to_fndecl): Be lax about
5459 decl has when in streaming stage.
5460 * lto-symtab.c (lto_symtab_merge_symbols): Likewise.
5461 * cgraph.h (cgraph_state): Add CGRAPH_LTO_STREAMING.
5463 2013-06-12 Roland Stigge <stigge@antcom.de>
5466 * config/rs6000/t-linux (MULTIARCH_DIRNAME): Fix SPE version detection.
5468 2013-06-12 Jakub Jelinek <jakub@redhat.com>
5470 PR tree-optimization/57537
5471 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): If
5472 vect_handle_widen_op_by_const, convert oprnd1 to half_type1.
5474 2013-06-12 Richard Biener <rguenther@suse.de>
5476 * data-streamer.h (streamer_write_char_stream): CSE
5477 obs->current_pointer.
5478 * data-streamer-out.c (streamer_write_uhwi_stream): Inline
5479 streamer_write_char_stream manually and optimize the resulting loop.
5480 (streamer_write_hwi_stream): Likewise.
5482 2013-06-12 Jan Hubicka <jh@suse.cz>
5484 * lto-symtab.c (lto_symtab_merge_symbols): Populate symtab hashtable.
5485 * cgraph.h (varpool_create_empty_node): Declare.
5486 * lto-cgraph.c (input_node, input_varpool_node): Forcingly create
5488 * symtab.c (symtab_unregister_node): Be lax about missin entries
5490 (symtab_get_node): Update comment.
5491 * varpool.c (varpool_create_empty_node): Break out from ...
5492 (varpool_node_for_decl): ... here.
5493 * lto-streamer.h (lto_file_decl_data): Add RESOLUTION_MAP.
5495 2013-06-12 Eric Botcazou <ebotcazou@adacore.com>
5497 * expr.c (expand_expr_real_1) <TARGET_MEM_REF>: Use straight-line flow.
5498 <MEM_REF>: Use 'type' instead of TREE_TYPE (exp) and tidy up the first
5499 part. Use straight-line flow at the end.
5500 <COMPONENT_REF>: Remove superfluous else.
5501 <VIEW_CONVERT_EXPR>: Use 'type' instead of TREE_TYPE (exp).
5503 2013-06-12 Jakub Jelinek <jakub@redhat.com>
5506 * varasm.c (decl_binds_to_current_def_p): Call binds_local_p
5507 target hook even for !TREE_PUBLIC decls. If no resolution info
5508 is available, return false for common and external decls.
5510 2013-06-12 Kaushik Phatak <kaushik.phatak@kpitcummins.com>
5512 * config/rl78/constraints.md (U): New constraint.
5513 * config/rl78/rl78.md (*mulqi3_rl78,*mulhi3_rl78,*mulhi3_g13): Add
5516 2013-06-11 Michael Meissner <meissner@linux.vnet.ibm.com>
5519 * config/rs6000/driver-rs6000.c (elf_platform): Make buffer static
5520 to allow returning address to AT_PLATFORM name.
5522 2013-06-11 Jan Hubicka <jh@suse.cz>
5524 * cgraph.c (cgraph_create_function_alias): Set weakref flag.
5525 * cgraph.h (symtab_node_base): Add weakref flag.
5526 * cgraphunit.c (cgraph_reset_node): Clear weakref flag.
5527 (handle_alias_pairs): Set weakref flag, do not set DECL_EXTERNAL.
5528 (output_weakrefs): Use weakref flag.
5529 * fold-const.c (simple_operand_p): Handle WEAK.
5530 * gimple-fold.c (can_refer_decl_in_current_unit_p): Drop weakref.
5531 * ipa.c (varpool_externally_visible_p): Drop weakref.
5532 (function_and_variable_visibility): Update comment; fix weakref
5533 sanity checks; do not clear DECL_WEAK on them.
5534 * lto-cgraph.c (lto_output_node): update.
5535 (lto_output_varpool_node): Update.
5536 (input_overwrite_node): Update.
5537 (input_node): Update.
5538 (input_varpool_node): Update.
5539 * lto-symtab.c (lto_symtab_symbol_p): Do not special case weakrefs.
5540 (lto_symtab_merge_symbols): Add sanity check.
5541 (lto_symtab_prevailing_decl): Do not special case weakrefs.
5542 * passes.c (rest_of_decl_compilation): Set static flag, too.
5543 * symtab.c (dump_symtab_base): Dump weakref.
5544 (verify_symtab_base): Sanity check weakrefs.
5545 (symtab_make_decl_local): Remove duplicated code.
5546 (symtab_alias_ultimate_target): Simplify.
5547 * varpool.c (varpool_create_variable_alias): Set weakref flag.
5549 2013-06-11 DJ Delorie <dj@redhat.com>
5551 * config/rl78/rl78.c (TARGET_UNWIND_WORD_MODE): Define.
5552 (rl78_unwind_word_mode): New.
5554 2013-06-11 David Malcolm <dmalcolm@redhat.com>
5556 * final.c (debug_prefix_maps): Make static.
5558 2013-06-11 David Malcolm <dmalcolm@redhat.com>
5560 * function.c (initial_trampoline): Remove stray copy.
5562 2013-06-11 Sofiane Naci <sofiane.naci@arm.com>
5564 * config/aarch64/aarch64-simd.md (move_lo_quad_<mode>): Update.
5566 2013-06-11 Martin Jambor <mjambor@suse.cz>
5568 * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that param_index is
5569 within bounds at the beginning of the function.
5571 2013-06-11 Alan Modra <amodra@gmail.com>
5573 * varasm.c (get_section): Don't die on !DECL_P decl. Tidy error
5575 (get_named_section): Don't NULL !DECL_P decl.
5577 2013-06-11 Igor Zamyatin <igor.zamyatin@intel.com>
5579 * doc/invoke.texi (core-avx2): Document.
5581 (atom): Updated with MOVBE.
5583 2013-06-11 Richard Biener <rguenther@suse.de>
5585 * collect2.c (main): Do not redirect ld stdout/stderr when debugging.
5587 2013-06-11 Anton Blanchard <anton@samba.org>
5589 * config/rs6000/rs6000.c (rs6000_adjust_atomic_subword): Calculate
5590 correct shift value in little-endian mode.
5592 2013-06-11 Jakub Jelinek <jakub@redhat.com>
5595 * varasm.c (get_variable_align): Move #endif to the right place.
5597 2013-06-10 Cary Coutant <ccoutant@google.com>
5599 * dwarf2out.c (hash_external_ref): Use die_symbol or signature
5600 for hash so that hash table traversal order is deterministic.
5602 2013-06-10 Michael Meissner <meissner@linux.vnet.ibm.com>
5603 Pat Haugen <pthaugen@us.ibm.com>
5604 Peter Bergner <bergner@vnet.ibm.com>
5606 * config/rs6000/vector.md (GPR move splitter): Do not split moves
5607 of vectors in GPRS if they are direct moves or quad word load or
5610 * config/rs6000/rs6000-protos.h (rs6000_output_move_128bit): Add
5612 (direct_move_p): Likewise.
5613 (quad_load_store_p): Likewise.
5615 * config/rs6000/rs6000.c (enum rs6000_reg_type): Simplify register
5616 classes into bins based on the physical register type.
5617 (reg_class_to_reg_type): Likewise.
5618 (IS_STD_REG_TYPE): Likewise.
5619 (IS_FP_VECT_REG_TYPE): Likewise.
5620 (reload_fpr_gpr): Arrays to determine what insn to use if we can
5621 use direct move instructions.
5622 (reload_gpr_vsx): Likewise.
5623 (reload_vsx_gpr): Likewise.
5624 (rs6000_init_hard_regno_mode_ok): Precalculate the register type
5625 information that is a simplification of register classes. Also
5626 precalculate direct move reload helpers.
5627 (direct_move_p): New function to return true if the operation can
5628 be done as a direct move instruciton.
5629 (quad_load_store_p): New function to return true if the operation
5630 is a quad memory operation.
5631 (rs6000_legitimize_address): If quad memory, only allow register
5632 indirect for TImode addresses.
5633 (rs6000_legitimate_address_p): Likewise.
5634 (enum reload_reg_type): Delete, replace with rs6000_reg_type.
5635 (rs6000_reload_register_type): Likewise.
5636 (register_to_reg_type): Return register type.
5637 (rs6000_secondary_reload_simple_move): New helper function for
5638 secondary reload and secondary memory needed to identify anything
5639 that is a simple move, and does not need reloading.
5640 (rs6000_secondary_reload_direct_move): New helper function for
5641 secondary reload to identify cases that can be done with several
5642 instructions via the direct move instructions.
5643 (rs6000_secondary_reload_move): New helper function for secondary
5644 reload to identify moves between register types that can be done.
5645 (rs6000_secondary_reload): Add support for quad memory operations
5646 and for direct move.
5647 (rs6000_secondary_memory_needed): Likewise.
5648 (rs6000_debug_secondary_memory_needed): Change argument names.
5649 (rs6000_output_move_128bit): New function to return the move to
5650 use for 128-bit moves, including knowing about the various
5651 limitations of quad memory operations.
5653 * config/rs6000/vsx.md (vsx_mov<mode>): Add support for quad
5654 memory operations. call rs6000_output_move_128bit for the actual
5655 instruciton(s) to generate.
5656 (vsx_movti_64bit): Likewise.
5658 * config/rs6000/rs6000.md (UNSPEC_P8V_FMRGOW): New unspec values.
5659 (UNSPEC_P8V_MTVSRWZ): Likewise.
5660 (UNSPEC_P8V_RELOAD_FROM_GPR): Likewise.
5661 (UNSPEC_P8V_MTVSRD): Likewise.
5662 (UNSPEC_P8V_XXPERMDI): Likewise.
5663 (UNSPEC_P8V_RELOAD_FROM_VSX): Likewise.
5664 (UNSPEC_FUSION_GPR): Likewise.
5665 (FMOVE128_GPR): New iterator for direct move.
5666 (f32_lv): New mode attribute for load/store of SFmode/SDmode values.
5669 (zero_extend<mode>di2_internal1): Add support for power8 32-bit
5670 loads and direct move instructions.
5671 (zero_extendsidi2_lfiwzx): Likewise.
5672 (extendsidi2_lfiwax): Likewise.
5673 (extendsidi2_nocell): Likewise.
5674 (floatsi<mode>2_lfiwax): Likewise.
5676 (floatunssi<mode>2_lfiwzx): Likewise.
5678 (fix_trunc<mode>_stfiwx): Likewise.
5679 (fixuns_trunc<mode>_stfiwx): Likewise.
5680 (mov<mode>_hardfloat, 32-bit floating point): Likewise.
5681 (mov<move>_hardfloat64, 64-bit floating point): Likewise.
5682 (parity<mode>2_cmpb): Set length/type attr.
5683 (unnamed shift right patterns, mov<mode>_internal2): Change type attr
5684 for 'mr.' to fast_compare.
5685 (bpermd_<mode>): Change type attr to popcnt.
5686 (p8_fmrgow_<mode>): New insns for power8 direct move support.
5687 (p8_mtvsrwz_1): Likewise.
5688 (p8_mtvsrwz_2): Likewise.
5689 (reload_fpr_from_gpr<mode>): Likewise.
5690 (p8_mtvsrd_1): Likewise.
5691 (p8_mtvsrd_2): Likewise.
5692 (p8_xxpermdi_<mode>): Likewise.
5693 (reload_vsx_from_gpr<mode>): Likewise.
5694 (reload_vsx_from_gprsf): Likewise.
5695 (p8_mfvsrd_3_<mode>): LIkewise.
5696 (reload_gpr_from_vsx<mode>): Likewise.
5697 (reload_gpr_from_vsxsf): Likewise.
5698 (p8_mfvsrd_4_disf): Likewise.
5699 (multi-word GPR splits): Do not split direct moves or quad memory
5702 2013-06-10 David Malcolm <dmalcolm@redhat.com>
5704 * tree-into-ssa.c (interesting_blocks): Make static.
5706 2013-06-10 Jakub Jelinek <jakub@redhat.com>
5709 * varasm.c (align_variable): Don't use DATA_ALIGNMENT or
5710 CONSTANT_ALIGNMENT if !decl_binds_to_current_def_p (decl).
5711 Use DATA_ABI_ALIGNMENT for that case instead if defined.
5712 (get_variable_align): New function.
5713 (get_variable_section, emit_bss, emit_common,
5714 assemble_variable_contents, place_block_symbol): Use
5715 get_variable_align instead of DECL_ALIGN.
5716 (assemble_noswitch_variable): Add align argument, use it
5717 instead of DECL_ALIGN.
5718 (assemble_variable): Adjust caller. Use get_variable_align
5719 instead of DECL_ALIGN.
5720 * config/i386/i386.h (DATA_ALIGNMENT): Adjust x86_data_alignment
5722 (DATA_ABI_ALIGNMENT): Define.
5723 * config/i386/i386-protos.h (x86_data_alignment): Adjust prototype.
5724 * config/i386/i386.c (x86_data_alignment): Add opt argument. If
5725 opt is false, only return the psABI mandated alignment increase.
5726 * config/c6x/c6x.h (DATA_ALIGNMENT): Renamed to...
5727 (DATA_ABI_ALIGNMENT): ... this.
5728 * config/mmix/mmix.h (DATA_ALIGNMENT): Renamed to...
5729 (DATA_ABI_ALIGNMENT): ... this.
5730 * config/mmix/mmix.c (mmix_data_alignment): Adjust function comment.
5731 * config/s390/s390.h (DATA_ALIGNMENT): Renamed to...
5732 (DATA_ABI_ALIGNMENT): ... this.
5733 * doc/tm.texi.in (DATA_ABI_ALIGNMENT): Document.
5734 * doc/tm.texi: Regenerated.
5736 2013-06-10 Uros Bizjak <ubizjak@gmail.com>
5738 * config/alpha/alpha.c (alpha_emit_xfloating_compare): Also use
5739 cmp_code to construct REG_EQUAL note.
5741 2013-06-09 Jakub Jelinek <jakub@redhat.com>
5744 * config/i386/i386.md (TARGET_READ_MODIFY_WRITE peepholes): Ensure
5745 that operands[2] doesn't overlap with operands[0].
5747 2013-06-09 David Edelsohn <dje.gcc@gmail.com>
5748 Jan Hubicka <jh@suse.cz>
5750 * config/rs6000/rs6000.c (print_operand, 'z'): Remove historical
5751 hack to mark symbols as used.
5753 2013-06-08 Vladimir Makarov <vmakarov@redhat.com>
5755 PR rtl-optimization/57559
5756 * lra-constraints.c (process_alt_operands): Don't discourage
5757 memory with known offset for offsetable memory constraint.
5758 * lra.c (lra_emit_add): Exchange y and z for 2-op add insn.
5760 2013-06-08 Eric Botcazou <ebotcazou@adacore.com>
5762 * varasm.c (struct oc_local_state): Reorder fields.
5763 (output_constructor_bitfield): Replace OUTER parameter with BIT_OFFSET
5764 and adjust accordingly.
5765 (output_constructor): Reorder initialization code and adjust call to
5766 output_constructor_bitfield.
5768 2013-06-07 Jan Hubicka <jh@suse.cz>
5770 * symtab.c (symtab_resolve_alias): Do not remove alias attribute.
5772 2013-06-07 David Malcolm <dmalcolm@redhat.com>
5774 * tree-object-size.c (unknown): Make const.
5776 2013-06-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
5778 * config/s390/s390.md (cpu_facility): Add cpu_zarch.
5779 ("*movmem_short", "*clrmem_short", "*cmpmem_short): Use cpu_zarch
5780 for last alternative in the cpu_facility attribute.
5782 2013-06-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5785 * config/arm/arm.md (*xordi3_insn): Change to insn_and_split.
5786 (xordi3): Change operand 2 constraint to arm_xordi_operand.
5787 * config/arm/arm.c (const_ok_for_dimode_op): Handle XOR.
5788 * config/arm/constraints.md (Dg): New constraint.
5789 * config/arm/neon.md (xordi3_neon): Remove.
5790 (neon_veor<mode>): Generate xordi3 instead of xordi3_neon.
5791 * config/arm/predicates.md (arm_xordi_operand): New predicate.
5793 2013-06-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5795 * config/arm/arm.md (anddi3_insn): Remove duplicate alternatives.
5796 Clean up alternatives.
5798 2013-06-07 Alan Modra <amodra@gmail.com>
5800 * config/rs6000/rs6000.c (setup_incoming_varargs): Round up
5803 2013-06-07 Alan Modra <amodra@gmail.com>
5805 * varasm.c (force_const_mem): Assert mode is not VOID or BLK.
5807 2013-06-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5809 * config/arm/constraints.md (Df): New constraint.
5810 * config/arm/arm.md (iordi3_insn): Use Df constraint instead of De.
5811 Correct length attribute for last two alternatives.
5813 2013-06-07 Alan Modra <amodra@gmail.com>
5815 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
5816 override user -mfp-in-toc.
5817 (offsettable_ok_by_alignment): Consider just the current access
5818 rather than the whole object, unless BLKmode. Handle
5819 CONSTANT_POOL_ADDRESS_P constants that lack a decl too.
5820 (use_toc_relative_ref): Allow CONSTANT_POOL_ADDRESS_P constants
5821 for -mcmodel=medium.
5822 * config/rs6000/linux64.h (SUBSUBTARGET_OVERRIDE_OPTIONS): Don't
5823 override user -mfp-in-toc or -msum-in-toc. Default to
5824 -mno-fp-in-toc for -mcmodel=medium.
5826 2013-06-06 DJ Delorie <dj@redhat.com>
5828 * config/rl78/rl78.c (rl78_valid_pointer_mode): New, implements
5829 TARGET_VALID_POINTER_MODE.
5831 2013-06-06 Michael Meissner <meissner@linux.vnet.ibm.com>
5832 Pat Haugen <pthaugen@us.ibm.com>
5833 Peter Bergner <bergner@vnet.ibm.com>
5835 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
5836 Document new power8 builtins.
5838 * config/rs6000/vector.md (and<mode>3): Add a clobber/scratch of a
5839 condition code register, to allow 128-bit logical operations to be
5840 done in the VSX or GPR registers.
5841 (nor<mode>3): Use the canonical form for nor.
5842 (eqv<mode>3): Add expanders for power8 xxleqv, xxlnand, xxlorc,
5843 vclz*, and vpopcnt* vector instructions.
5844 (nand<mode>3): Likewise.
5845 (orc<mode>3): Likewise.
5846 (clz<mode>2): LIkewise.
5847 (popcount<mode>2): Likewise.
5849 * config/rs6000/predicates.md (int_reg_operand): Rework tests so
5850 that only the GPRs are recognized.
5852 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
5853 support for new power8 builtins.
5855 * config/rs6000/rs6000-builtin.def (xscvspdpn): Add new power8
5857 (xscvdpspn): Likewise.
5863 (vpopcnt): Likewise.
5864 (vpopcntb): Likewise.
5865 (vpopcnth): Likewise.
5866 (vpopcntw): Likewise.
5867 (vpopcntd): Likewise.
5872 (eqv_v16qi3): Likewise.
5873 (eqv_v8hi3): Likewise.
5874 (eqv_v4si3): Likewise.
5875 (eqv_v2di3): Likewise.
5876 (eqv_v4sf3): Likewise.
5877 (eqv_v2df3): Likewise.
5879 (nand_v16qi3): Likewise.
5880 (nand_v8hi3): Likewise.
5881 (nand_v4si3): Likewise.
5882 (nand_v2di3): Likewise.
5883 (nand_v4sf3): Likewise.
5884 (nand_v2df3): Likewise.
5886 (orc_v16qi3): Likewise.
5887 (orc_v8hi3): Likewise.
5888 (orc_v4si3): Likewise.
5889 (orc_v2di3): Likewise.
5890 (orc_v4sf3): Likewise.
5891 (orc_v2df3): Likewise.
5893 * config/rs6000/rs6000.c (rs6000_option_override_internal): Only
5894 allow power8 quad mode in 64-bit.
5895 (rs6000_builtin_vectorized_function): Add support to vectorize
5896 ISA 2.07 count leading zeros, population count builtins.
5897 (rs6000_expand_vector_init): On ISA 2.07 use xscvdpspn to form
5898 V4SF vectors instead of xscvdpsp to avoid IEEE related traps.
5899 (builtin_function_type): Add vgbbd builtin function which takes an
5901 (altivec_expand_vec_perm_const): Add support for new power8 merge
5904 * config/rs6000/vsx.md (VSX_L2): New iterator for 128-bit types,
5905 that does not include TImdoe for use with 32-bit.
5906 (UNSPEC_VSX_CVSPDPN): Support for power8 xscvdpspn and xscvspdpn
5908 (UNSPEC_VSX_CVDPSPN): Likewise.
5909 (vsx_xscvdpspn): Likewise.
5910 (vsx_xscvspdpn): Likewise.
5911 (vsx_xscvdpspn_scalar): Likewise.
5912 (vsx_xscvspdpn_directmove): Likewise.
5913 (vsx_and<mode>3): Split logical operations into 32-bit and
5914 64-bit. Add support to do logical operations on TImode as well as
5915 VSX vector types. Allow logical operations to be done in either
5916 VSX registers or in general purpose registers in 64-bit mode. Add
5917 splitters if GPRs were used. For AND, add clobber of CCmode to
5918 allow use of ANDI on GPRs. Rewrite nor to use the canonical RTL
5920 (vsx_and<mode>3_32bit): Likewise.
5921 (vsx_and<mode>3_64bit): Likewise.
5922 (vsx_ior<mode>3): Likewise.
5923 (vsx_ior<mode>3_32bit): Likewise.
5924 (vsx_ior<mode>3_64bit): Likewise.
5925 (vsx_xor<mode>3): Likewise.
5926 (vsx_xor<mode>3_32bit): Likewise.
5927 (vsx_xor<mode>3_64bit): Likewise.
5928 (vsx_one_cmpl<mode>2): Likewise.
5929 (vsx_one_cmpl<mode>2_32bit): Likewise.
5930 (vsx_one_cmpl<mode>2_64bit): Likewise.
5931 (vsx_nor<mode>3): Likewise.
5932 (vsx_nor<mode>3_32bit): Likewise.
5933 (vsx_nor<mode>3_64bit): Likewise.
5934 (vsx_andc<mode>3): Likewise.
5935 (vsx_andc<mode>3_32bit): Likewise.
5936 (vsx_andc<mode>3_64bit): Likewise.
5937 (vsx_eqv<mode>3_32bit): Add support for power8 xxleqv, xxlnand,
5938 and xxlorc instructions.
5939 (vsx_eqv<mode>3_64bit): Likewise.
5940 (vsx_nand<mode>3_32bit): Likewise.
5941 (vsx_nand<mode>3_64bit): Likewise.
5942 (vsx_orc<mode>3_32bit): Likewise.
5943 (vsx_orc<mode>3_64bit): Likewise.
5945 * config/rs6000/rs6000.h (VLOGICAL_REGNO_P): Update comment.
5947 * config/rs6000/altivec.md (UNSPEC_VGBBD): Add power8 vgbbd
5949 (p8_vmrgew): Add power8 vmrgew and vmrgow instructions.
5950 (p8_vmrgow): Likewise.
5951 (altivec_and<mode>3): Add clobber of CCmode to allow AND using
5952 GPRs to be split under VSX.
5953 (p8v_clz<mode>2): Add power8 count leading zero support.
5954 (p8v_popcount<mode>2): Add power8 population count support.
5955 (p8v_vgbbd): Add power8 gather bits by bytes by doubleword
5958 * config/rs6000/rs6000.md (eqv<mode>3): Add support for powerp eqv
5961 * config/rs6000/altivec.h (vec_eqv): Add defines to export power8
5963 (vec_nand): Likewise.
5964 (vec_vclz): Likewise.
5965 (vec_vclzb): Likewise.
5966 (vec_vclzd): Likewise.
5967 (vec_vclzh): Likewise.
5968 (vec_vclzw): Likewise.
5969 (vec_vgbbd): Likewise.
5970 (vec_vmrgew): Likewise.
5971 (vec_vmrgow): Likewise.
5972 (vec_vpopcnt): Likewise.
5973 (vec_vpopcntb): Likewise.
5974 (vec_vpopcntd): Likewise.
5975 (vec_vpopcnth): Likewise.
5976 (vec_vpopcntw): Likewise.
5978 2013-06-06 Vladimir Makarov <vmakarov@redhat.com>
5980 PR rtl-optimization/57468
5981 * config/i386/i386.c (inline_secondary_memory_needed): Ignore
5984 2013-06-06 Vladimir Makarov <vmakarov@redhat.com>
5986 PR rtl-optimization/57459
5987 * lra-constraints.c (update_ebb_live_info): Fix typo for operand
5988 type when setting live regs.
5990 2013-06-06 Vladimir Makarov <vmakarov@redhat.com>
5992 * config/s390/s390.opt (mlra): New option.
5993 * config/s390/s390.c (s390_decompose_address): Check displacement
5994 for all registers for LRA.
5995 (s390_secondary_reload): Don't used secondary reloads for LRA.
5996 (s390_lra_p): New function.
5997 (TARGET_LRA_P): Define.
5998 * config/s390/s390.md (*movmem_short, *clrmem_short): Change value
5999 of attribute cpu_facility to zarch for the last alternative.
6000 (*cmpmem_short): Ditto.
6002 2013-06-06 Eric Botcazou <ebotcazou@adacore.com>
6004 * config/arm/arm.c (arm_r3_live_at_start_p): New predicate.
6005 (arm_compute_static_chain_stack_bytes): Use it. Tidy up.
6006 (arm_expand_prologue): Likewise.
6008 2013-06-06 Teresa Johnson <tejohnson@google.com>
6011 * ifcvt.c (find_if_case_1): Replace BB_COPY_PARTITION with assert
6012 as this is now done by redirect_edge_and_branch_force.
6013 * function.c (thread_prologue_and_epilogue_insns): Insert new bb after
6014 barriers, and fix interaction with splitting.
6015 * emit-rtl.c (try_split): Copy REG_CROSSING_JUMP notes.
6016 * cfgcleanup.c (try_forward_edges): Fix early return value to properly
6017 reflect changes made in the routine.
6018 * bb-reorder.c (emit_barrier_after_bb): Move to cfgrtl.c.
6019 (fix_up_fall_thru_edges): Remove incorrect check for bb layout order
6020 since this is called in cfglayout mode, and replace partition fixup
6021 with assert as that is now done by force_nonfallthru_and_redirect.
6022 (add_reg_crossing_jump_notes): Handle the fact that some jumps may
6023 already be marked with region crossing note.
6024 (insert_section_boundary_note): Make non-static, gate on flag
6025 has_bb_partition, rewrite to also check for multiple partitions.
6026 (rest_of_handle_reorder_blocks): Remove call to
6027 insert_section_boundary_note, now done later during free_cfg.
6028 (duplicate_computed_gotos): Don't duplicate partition crossing edge.
6029 * bb-reorder.h (insert_section_boundary_note): Declare.
6030 * Makefile.in (cfgrtl.o): Depend on bb-reorder.h
6031 * cfgrtl.c (rest_of_pass_free_cfg): If partitions exist
6032 invoke insert_section_boundary_note.
6033 (try_redirect_by_replacing_jump): Remove unnecessary
6034 check for region crossing note.
6035 (fixup_partition_crossing): New function.
6036 (rtl_redirect_edge_and_branch): Fixup partition boundaries.
6037 (emit_barrier_after_bb): Move here from bb-reorder.c, handle insertion
6038 in non-cfglayout mode.
6039 (force_nonfallthru_and_redirect): Fixup partition boundaries,
6040 remove old code that tried to do this. Emit barrier correctly
6041 when we are in cfglayout mode.
6042 (last_bb_in_partition): New function.
6043 (rtl_split_edge): Correctly fixup partition boundaries.
6044 (commit_one_edge_insertion): Remove old code that tried to
6045 fixup region crossing edge since this is now handled in
6046 split_block, and set up insertion point correctly since
6047 block may now end in a jump.
6048 (verify_hot_cold_block_grouping): Guard against checking when not in
6049 linearized RTL mode.
6050 (rtl_verify_edges): Add checks for incorrect/missing REG_CROSSING_JUMP
6052 (rtl_verify_flow_info_1): Move verify_hot_cold_block_grouping to
6053 rtl_verify_flow_info, so not called in cfglayout mode.
6054 (rtl_verify_flow_info): Move verify_hot_cold_block_grouping here.
6055 (fixup_reorder_chain): Remove old code that attempted to fixup region
6056 crossing note as this is now handled in force_nonfallthru_and_redirect.
6057 (duplicate_insn_chain): Don't duplicate switch section notes.
6058 (rtl_can_remove_branch_p): Remove unnecessary check for region crossing
6060 * basic-block.h (emit_barrier_after_bb): Declare.
6062 2013-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6064 * config/arm/arm-fixed.md (add<mode>3,usadd<mode>3,ssadd<mode>3,
6065 sub<mode>3, ussub<mode>3, sssub<mode>3, arm_ssatsihi_shift,
6066 arm_usatsihi): Adjust alternatives for arm_restrict_it.
6068 2013-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6070 * config/arm/arm-ldmstm.ml: Set "predicable_short_it" to "no"
6072 * config/arm/ldmstm.md: Regenerate.
6074 2013-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6076 * config/arm/sync.md (atomic_loaddi_1):
6077 Disable predication for arm_restrict_it.
6078 (arm_load_exclusive<mode>): Likewise.
6079 (arm_load_exclusivesi): Likewise.
6080 (arm_load_exclusivedi): Likewise.
6081 (arm_load_acquire_exclusive<mode>): Likewise.
6082 (arm_load_acquire_exclusivesi): Likewise.
6083 (arm_load_acquire_exclusivedi): Likewise.
6084 (arm_store_exclusive<mode>): Likewise.
6085 (arm_store_exclusive<mode>): Likewise.
6086 (arm_store_release_exclusivedi): Likewise.
6087 (arm_store_release_exclusive<mode>): Likewise.
6089 2013-06-06 Richard Biener <rguenther@suse.de>
6091 * lto-streamer.h (enum LTO_tags): Move LTO_tree_pickle_reference
6093 (lto_tag_is_tree_code_p): Adjust.
6094 (lto_tag_is_gimple_code_p): Likewise.
6095 (lto_gimple_code_to_tag): Likewise.
6096 (lto_tag_to_gimple_code): Likewise.
6097 (lto_tree_code_to_tag): Likewise.
6098 (lto_tag_to_tree_code): Likewise.
6099 * data-streamer.h (streamer_write_hwi_in_range): Use
6100 uhwi streaming to stream the normalized range.
6101 (streamer_read_hwi_in_range): Likewise.
6103 2013-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6105 * config/arm/arm.md (enabled_for_depr_it): New attribute.
6106 (predicable_short_it): Likewise.
6107 (predicated): Likewise.
6108 (enabled): Handle above.
6109 (define_cond_exec): Set predicated attribute to yes.
6111 2013-06-05 Mike Stump <mikestump@comcast.net>
6113 * gdbinit.in (__FUNCTION__): Add.
6115 2013-06-05 Uros Bizjak <ubizjak@gmail.com>
6117 * config/alpha/alpha.c (alpha_emit_conditional_move): Swap all
6118 GE, GT, GEU and GTU compares, modulo DImode compares with zero.
6120 2013-06-05 Jan Hubicka <jh@suse.cz>
6122 * varasm.c (mark_decl_referenced): Revert the removal until targets
6125 2013-06-05 David Edelsohn <dje.gcc@gmail.com>
6127 * config/rs6000/rs6000.c (print_operand, 'z'): Use DECL_PRESERVE_P
6128 instead of mark_decl_referenced.
6130 2013-06-05 Jan Hubicka <jh@suse.cz>
6132 * cgraph.c (cgraph_remove_node): Clear forced_by_abi.
6133 (cgraph_node_cannot_be_local_p_1): Honnor symbol.forced_by_abi
6134 and symtab_used_from_object_file_p.
6135 (cgraph_make_node_local_1): Clear forced_by_abi.
6136 (cgraph_can_remove_if_no_direct_calls_and): Use forced_by_abi
6137 * cgraph.h (symtab_node_base): Add forced_by_abi.
6138 (decide_is_variable_needed): Remove.
6139 (varpool_can_remove_if_no_refs): Honnor symbol.forced_by_abi.
6140 * cgraphunit.c (cgraph_decide_is_function_needed): Rename to ..
6141 (decide_is_symbol_needed): ... this one; handle symbols in general;
6142 always analyze virtuals; honnor forced_by_abi.
6143 (cgraph_finalize_function): Update.
6144 (varpool_finalize_decl): Update.
6145 (symbol_defined_and_needed): Remove.
6146 (analyze_functions): Update.
6147 * lto-cgraph.c (lto_output_node, lto_output_varpool_node,
6148 output_refs, input_overwrite_node): Handle forced_by_abi.
6149 * ipa.c (cgraph_address_taken_from_non_vtable_p): Rename to ...
6150 (address_taken_from_non_vtable_p): ... this one.
6151 (comdat_can_be_unshared_p_1): New function.
6152 (cgraph_comdat_can_be_unshared_p): Rename to ...
6153 (comdat_can_be_unshared_p): ... this one; handle symbols in general.
6154 (varpool_externally_visible_p): Use comdat_can_be_unshared_p.
6155 (function_and_variable_visibility): Clear forced_by_abi as needed.
6156 * trans-mem.c (ipa_tm_mark_forced_by_abi_node): New functoin.
6157 (ipa_tm_create_version_alias, ipa_tm_create_version): Update.
6158 * symtab.c (dump_symtab_base): Dump forced_by_abi.
6159 * varpool.c (decide_is_variable_needed): Remove.
6161 2013-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6163 * config/arm/arm.c (MAX_INSN_PER_IT_BLOCK): New macro.
6164 (arm_option_override): Override arm_restrict_it where appropriate.
6165 (thumb2_final_prescan_insn): Use MAX_INSN_PER_IT_BLOCK.
6166 * config/arm/arm.opt (mrestrict-it): New command-line option.
6167 * doc/invoke.texi: Document -mrestrict-it.
6169 2013-06-05 David Malcolm <dmalcolm@redhat.com>
6171 * tsan.c (tsan_atomic_table): Make const.
6173 2013-06-05 Richard Biener <rguenther@suse.de>
6175 * tree-streamer.c (streamer_tree_cache_insert_1): Update the
6176 index associated with the tree we are supposed to replace.
6177 * tree-streamer-out.c (pack_ts_base_value_fields): Output
6178 TREE_ASM_WRITTEN as zero for everything but SSA names.
6180 2013-06-05 David Malcolm <dmalcolm@redhat.com>
6182 * tree-ssa-structalias.c (call_stmt_vars): Make static.
6184 2013-06-04 Jan Hubicka <jh@suse.cz>
6186 * lto-cgraph.c (get_alias_symbol): Remove weakref sanity check.
6187 (input_node, input_varpool_node): Handle correctly external same
6189 * ipa.c (symtab_remove_unreachable_nodes): Do not remove external
6190 nodes at ltrans stage.
6192 2013-06-04 Jan Hubicka <jh@suse.cz>
6194 * ipa-inline.c (update_caller_keys): Fix availability test.
6195 (update_callee_keys): Likewise.
6196 * symtab.c (symtab_alias_ultimate_target): Make availaiblity logic
6197 to follow ELF standard.
6199 2013-06-04 Jürgen Urban <JuergenUrban@gmx.de>
6201 * config.gcc (mipsr5900-*-elf*, mipsr5900el-*-elf*, mips64r5900-*-elf*)
6202 (mips64r5900el-*-elf*): New configurations.
6203 * config/mips/mips-cpus.def (r5900): New processor.
6204 * config/mips/mips-tables.opt: Regenerate.
6205 * config/mips/mips.c (mips_rtx_cost_data): Add an R5900 entry.
6206 (mips_issue_rate): Handle PROCESSOR_R5900.
6207 (mips_reorg_process_insns): Force reorder mode for the R5900.
6208 * config/mips/mips.h (TARGET_MIPS5900): Define.
6209 (ISA_HAS_CONDMOVE, ISA_HAS_PREFETCH, ISA_HAS_HILO_INTERLOCKS): Include
6211 (ISA_HAS_LOAD_DELAY, ISA_HAS_XFER_DELAY, ISA_HAS_FCMP_DELAY): Exclude
6213 * config/mips/mips.md (processor): Add r5900.
6214 (MOVECC): Disallow CCmode conditions for TARGET_MIPS5900.
6216 2013-06-04 Ian Bolton <ian.bolton@arm.com>
6218 * config/aarch64/aarch64.md (*mov<mode>_aarch64): Call
6219 into function to generate MOVI instruction.
6220 * config/aarch64/aarch64.c (aarch64_simd_container_mode): New function.
6221 (aarch64_preferred_simd_mode): Turn into wrapper.
6222 (aarch64_output_scalar_simd_mov_immediate): New function.
6223 * config/aarch64/aarch64-protos.h: Add prototype for above.
6225 2013-06-04 Ian Bolton <ian.bolton@arm.com>
6227 * config/aarch64/aarch64.c (simd_immediate_info): Remove
6228 element_char member.
6229 (sizetochar): Return signed char.
6230 (aarch64_simd_valid_immediate): Remove elchar and other
6231 unnecessary variables.
6232 (aarch64_output_simd_mov_immediate): Take rtx instead of &rtx.
6233 Calculate element_char as required.
6234 * config/aarch64/aarch64-protos.h: Update and move prototype
6235 for aarch64_output_simd_mov_immediate.
6236 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
6239 2013-06-04 Ian Bolton <ian.bolton@arm.com>
6241 * config/aarch64/aarch64.c (simd_immediate_info): Struct to hold
6242 information completed by aarch64_simd_valid_immediate.
6243 (aarch64_legitimate_constant_p): Update arguments.
6244 (aarch64_simd_valid_immediate): Work with struct rather than many
6246 (aarch64_simd_scalar_immediate_valid_for_move): Update arguments.
6247 (aarch64_simd_make_constant): Update arguments.
6248 (aarch64_output_simd_mov_immediate): Work with struct rather than
6249 many pointers. Output immediate directly rather than as operand.
6250 * config/aarch64/aarch64-protos.h (aarch64_simd_valid_immediate):
6252 * config/aarch64/constraints.md (Dn): Update arguments.
6254 2013-06-04 Ian Bolton <ian.bolton@arm.com>
6256 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): No
6258 (aarch64_simd_immediate_valid_for_move): Remove.
6259 (aarch64_simd_scalar_immediate_valid_for_move): Update call.
6260 (aarch64_simd_make_constant): Update call.
6261 (aarch64_output_simd_mov_immediate): Update call.
6262 * config/aarch64/aarch64-protos.h (aarch64_simd_valid_immediate):
6264 * config/aarch64/constraints.md (Dn): Update call.
6266 2013-06-04 Ian Bolton <ian.bolton@arm.com>
6268 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Change
6269 return type to bool for prototype.
6270 (aarch64_legitimate_constant_p): Check for true instead of not -1.
6271 (aarch64_simd_valid_immediate): Fix up each return to return a bool.
6272 (aarch64_simd_immediate_valid_for_move): Update retval for bool.
6274 2013-06-04 Catherine Moore <clm@codesourcery.com>
6276 * config/mips/mips.opt (meva): New.
6277 * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Define __mips_eva.
6278 (ASM_SPEC): Handle -meva.
6279 * doc/invoke.texi (meva): Document.
6281 2013-06-04 Alan Modra <amodra@gmail.com>
6283 * config/rs6000/rs6000.c (output_toc): Correct little-endian float
6286 2013-06-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6288 * rtl.def: Add extra fourth optional field to define_cond_exec.
6289 * gensupport.c (process_one_cond_exec): Process attributes from
6291 * doc/md.texi: Document fourth field in define_cond_exec.
6293 2013-06-04 Eric Botcazou <ebotcazou@adacore.com>
6295 * expmed.c (extract_bit_field_1): In the larger-than-a-word case, factor
6296 out the processing order as in store_bit_field_1.
6298 2013-06-04 Jan Hubicka <jh@suse.cz>
6301 * cgraphunit.c (cgraph_process_same_body_aliases): Create
6302 non-VAR_DECL node if it does not exist yet.
6304 2013-06-03 Richard Sandiford <rdsandiford@googlemail.com>
6306 * config.gcc (mipsisa64sr71k-*-elf*, mipsisa64sb1-*-elf*)
6307 (mipsisa64sb1el-*-elf*, mips64-*-elf*, mips64el-*-elf*)
6308 (mips64orion-*-elf*, mips64orionel-*-elf*): Remove
6309 target_cpu_default setting.
6311 2013-06-03 Teresa Johnson <tejohnson@google.com>
6313 * dumpfile.c (opt_info_switch_p): Change -fopt-info
6314 default to -fopt-info=optimized instead of all.
6315 * doc/invoke.texi: Ditto.
6316 * tree-vectorizer.c (vectorize_loops): Emit loop vectorization
6317 success under MSG_OPTIMIZED_LOCATIONS, and use dump_printf_loc.
6318 (execute_vect_slp): Emit BB vectorization success under
6319 MSG_OPTIMIZED_LOCATIONS.
6320 * tree-vect-slp.c (vect_slp_transform_bb): Change
6321 MSG_OPTIMIZED_LOCATIONS to MSG_NOTE.
6322 * tree-vect-loop.c (vect_transform_loop): Ditto.
6324 2013-06-03 Jason Merrill <jason@redhat.com>
6327 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
6328 Use TARGET_EXPR for C++.
6330 2013-06-03 Jakub Jelinek <jakub@redhat.com>
6332 PR rtl-optimization/57268
6333 * sched-deps.c (sched_analyze_2): Don't flush_pending_lists
6334 if DEBUG_INSN_P (insn).
6337 2013-05-31 Dinar Temirbulatov <dinar@kugelworks.com>
6339 PR rtl-optimization/57268
6340 * sched-deps.c (sched_analyze_2): Flush dependence lists if
6341 the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
6343 2013-06-03 Yuri Rumyantsev <yuri.s.rumyantsev@intel.com>
6345 * config/i386/i386.c (ix86_lea_outperforms): Fix formatting.
6346 (ix86_avoid_lea_for_addr): Likewise.
6347 (exact_dependency_1): Likewise.
6348 (ix86_adjust_cost): Likewise.
6349 (swap_top_of_ready_list): Fix formatting and !reload_completed check
6351 (do_reorder_for_imul): Fix typo, formatting and
6352 !reload_completed check removed.
6353 (ix86_sched_reorder): Fix typo and formatting.
6354 (fold_builtin_cpu): Move M_INTEL_SLM at the end of processor types
6357 2013-06-03 Sofiane Naci <sofiane.naci@arm.com>
6359 * config/aarch64/aarch64.md (*movdi_aarch64): Define "simd" attribute.
6361 2013-06-03 Eric Botcazou <ebotcazou@adacore.com>
6363 * varasm.c (output_constant) <CONSTRUCTOR>: Minor formatting tweak.
6364 <STRING_CST>: Likewise.
6365 <VECTOR_CST>: Likewise.
6367 2013-06-01 Janus Weil <janus@gcc.gnu.org>
6368 Mikael Morin <mikael@gcc.gnu.org>
6370 * configure.ac: Add AC_HEADER_TIOCGWINSZ macro.
6371 * config.in: Regenerated.
6372 * configure: Regenerated.
6374 2013-06-01 Jan Hubicka <jh@suse.cz>
6377 * cgraphunit.c (compile): When weakref is not supported,
6378 set up transparent aliases before final output pass.
6379 * varasm.c (assemble_alias): Do not try to do it here.
6381 2013-06-01 Jan Hubicka <jh@suse.cz>
6384 * passes.c (for_per_function): Skip unanalyzed functions.
6386 2013-06-01 Jan Hubicka <jh@suse.cz>
6388 * lto-symtab.c (lto_symtab_merge_cgraph_nodes_1): Rename to ...
6389 (lto_symtab_merge_symbols_1): ... this one.
6390 (lto_symtab_merge_cgraph_nodes): Rename to ...
6391 (lto_symtab_merge_symbols): ... this one; simplify.
6392 * cgraph.c (same_body_aliases_done): Rename to ...
6393 (cpp_implicit_aliases_done): ... this one.
6394 (cgraph_create_function_alias): Update.
6395 (cgraph_same_body_alias): Update.
6396 (dump_cgraph_node): Remove alias dumping; simplify thunk dumping.
6397 (verify_edge_corresponds_to_fndecl): Simplify.
6398 * cgraph.h (symtab_node_base): Add cpp_implicit_alias, alias_target.
6399 (cgraph_node): Remove same_body_alias.
6400 (varpool_node): Remove alias_of and extra_name_alias.
6401 (same_body_aliases_done): Rename to ..
6402 (cpp_implicit_aliases_done): ... this one.
6403 (symtab_alias_ultimate_target): Add default parameter.
6404 (symtab_resolve_alias): New function.
6405 (fixup_same_cpp_alias_visibility): Declare.
6406 (cgraph_function_node): Add default parameter.
6407 (cgraph_node_asm_name): Likewise.
6408 (cgraph_function_or_thunk_node): Add default parameter; do
6409 not ICE when it is NULL.
6410 (varpool_variable_node): Likewise.
6411 * tree-emutls.c (create_emultls_var): Update.
6412 (ipa_lower_emutls): Update.
6413 * cgraphunit.c (cgraph_decide_is_function_needed): Update.
6414 (cgraph_reset_node): Reset alias info.
6415 (cgraph_finalize_function): Update.
6416 (fixup_same_cpp_alias_visibility): Move to symtab.c.
6417 (analyze_function): Simplify.
6418 (cgraph_process_same_body_aliases): Simplify.
6419 (analyze_functions): Fixup same body aliases.
6420 (handle_alias_pairs): Simplify.
6421 (assemble_thunk): Update.
6422 (assemble_thunks_and_aliases): Update.
6423 (output_weakrefs): Rewrite.
6424 * lto-cgraph.c (lto_output_node): Rewrite alias handling.
6425 (lto_output_varpool_node): Likewise.
6426 (compute_ltrans_boundary): Remve assert.
6427 (get_alias_symbol): New functoin.
6428 (input_node): Rewrite alias handling.
6429 (input_varpool_node): Likewise.
6430 * ipa-pure-const.c (propagate_pure_const): Fix formating.
6431 * ipa.c (process_references): Handle weakrefs correctly.
6432 (symtab_remove_unreachable_nodes): Likewise.
6433 * trans-mem.c (get_cg_data): Update.
6434 (ipa_tm_create_version_alias): Update.
6435 (ipa_tm_execute): Update.
6436 * symtab.c (dump_symtab_base): Dump aliases.
6437 (verify_symtab_base): Verify aliases.
6438 (symtab_node_availability): New function.
6439 (symtab_alias_ultimate_target): Simplify.
6440 (fixup_same_cpp_alias_visibility): Move here from cgraphunit.c;
6441 handle all the fixup cases.
6442 (symtab_resolve_alias): New function.
6443 * passes.c (ipa_write_summaries): Handle weakrefs.
6444 * varpool.c (varpool_analyze_node): Simplify.
6445 (assemble_aliases): Update.
6446 (varpool_create_variable_alias): Simplify.
6447 (varpool_extra_name_alias): Simplify.
6448 * lto-streamer.h (lto_symtab_merge_cgraph_nodes): Rename to...
6449 (lto_symtab_merge_symbols): ... this one.
6451 2013-06-01 Dinar Temirbulatov <dinar@kugelworks.com>
6454 PR rtl-optimization/57268
6455 * sched-deps.c (sched_analyze_2): Flush dependence lists if
6456 the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
6458 2013-06-01 Tobias Burnus <burnus@net-b.de>
6461 2013-05-31 Tobias Burnus <burnus@net-b.de>
6464 * tree-ssa-math-opts.c (execute_cse_sincos): Move check
6467 2013-05-31 Dinar Temirbulatov <dinar@kugelworks.com>
6469 PR rtl-optimization/57268
6470 * sched-deps.c (sched_analyze_2): Flush dependence lists if
6471 the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
6473 2013-05-31 Eric Botcazou <ebotcazou@adacore.com>
6475 * config/rs6000/predicates.md (rs6000_cbranch_operator): Accept some
6476 unordered comparison operators when -fno-trapping-math is in effect
6478 * config/rs6000/rs6000.c (rs6000_generate_compare): Remove dead code
6479 and implement unordered comparison operators properly on the e500.
6481 2013-05-31 Eric Botcazou <ebotcazou@adacore.com>
6483 * simplify-rtx.c (simplify_byte_swapping_operation): Use proper macro
6484 for constant scalar integers.
6485 (simplify_relational_operation_1): Likewise.
6487 2013-05-31 Segher Boessenkool <segher@kernel.crashing.org>
6489 * config/rs6000/rs6000-opts.h (enum processor_type): Reorder.
6490 * config/rs6000/rs6000.md (cpu): Reorder. Split long line.
6493 2013-05-31 Yuri Rumyantsev <yuri.s.rumyantsev@intel.com>
6494 Igor Zamyatin <igor.zamyatin@intel.com>
6496 Silvermont (SLM) architecture performance tuning.
6497 * config/i386/i386.h (enum ix86_tune_indices): Add
6498 X86_TUNE_SPLIT_MEM_OPND_FOR_FP_CONVERTS.
6499 (TARGET_SPLIT_MEM_OPND_FOR_FP_CONVERTS): New define.
6501 * config/i386/i386.c (initial_ix86_tune_features)
6502 <X86_TUNE_SPLIT_MEM_OPND_FOR_FP_CONVERTS>: Initialize.
6503 (ix86_lea_outperforms): Handle Silvermont tuning.
6504 (ix86_avoid_lea_for_add): Add new argument to ix86_lea_outperforms
6506 (ix86_use_lea_for_mov): Likewise.
6507 (ix86_avoid_lea_for_addr): Likewise.
6508 (ix86_lea_for_add_ok): Likewise.
6509 (exact_dependency_1): New function.
6510 (exact_store_load_dependency): Likewise.
6511 (ix86_adjust_cost): Handle Silvermont tuning.
6512 (do_reoder_for_imul): Likewise.
6513 (swap_top_of_ready_list): New function.
6514 (ix86_sched_reorder): Changed to handle Silvermont tuning.
6516 * config/i386/i386.md (peepholes that split memory operand in fp
6519 2013-05-31 Marcus Shawcroft <marcus.shawcroft@arm.com>
6521 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
6522 Remove un-necessary braces.
6524 2013-05-31 Marcus Shawcroft <marcus.shawcroft@arm.com>
6526 * config/aarch64/aarch64.c (aarch64_classify_symbol):
6527 Use SYMBOL_TINY_ABSOLUTE for AARCH64_CMODEL_TINY_PIC.
6529 2013-05-31 Tobias Burnus <burnus@net-b.de>
6532 * tree-ssa-math-opts.c (execute_cse_sincos): Move check further up.
6534 2013-05-31 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6537 * config/arm/arm.c (const_ok_for_dimode_op): Handle IOR.
6538 * config/arm/arm.md (*iordi3_insn): Change to insn_and_split.
6539 * config/arm/neon.md (iordi3_neon): Remove.
6540 (neon_vorr<mode>): Generate iordi3 instead of iordi3_neon.
6541 * config/arm/predicates.md (imm_for_neon_logic_operand):
6542 Move to earlier in the file.
6543 (neon_logic_op2): Likewise.
6544 (arm_iordi_operand_neon): New predicate.
6546 2013-05-31 Richard Biener <rguenther@suse.de>
6548 PR tree-optimization/57478
6549 PR tree-optimization/57453
6550 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Uses in PHI nodes
6553 2013-05-31 Kaushik Phatak <kaushik.phatak@kpitcummins.com>
6555 * config/rl78/rl78.md (mulqi3,mulhi3): New define_expands.
6556 (*mulqi3_rl78,*mulhi3_rl78,*mulhi3_g13): New define_insns.
6558 2013-05-30 Tobias Burnus <burnus@net-b.de>
6559 Thomas Koenig <tkoenig@gcc.gnu.org>
6562 * tree-ssa-math-opts.c (execute_cse_sincos): Optimize
6563 powi (-1.0, k) to (k & 1) ? -1.0 : 1.0.
6565 2013-05-30 Steven Bosscher <steven@gcc.gnu.org>
6567 * rtlanal.c (tablejump_p): Expect table and label to be adjacent.
6569 2013-05-30 Vladimir Makarov <vmakarov@redhat.com>
6571 * target.def (register_usage_leveling_p): New hook.
6572 * targhooks.c (default_register_usage_leveling_p): New.
6573 * targhooks.h (default_register_usage_leveling_p): New prototype.
6574 * lra-assigns.c (register_usage_leveling_p): Use the hook.
6575 * doc/tm.texi.in (TARGET_REGISTER_USAGE_LEVELING_P): New hook.
6576 * doc/tm.texi: Update.
6577 * config/i386/i386.c (TARGET_REGISTER_USAGE_LEVELING_P): Define.
6579 2013-05-30 Ian Bolton <ian.bolton@arm.com>
6581 * config/aarch64/aarch64.md (insv<mode>): New define_expand.
6582 (*insv_reg<mode>): New define_insn.
6584 2013-05-30 Joern Rennecke <joern.rennecke@embecosm.com>
6586 PR rtl-optimization/57439
6587 * postreload.c (move2add_valid_value_p): Check that we have
6588 a zero subreg_regno_offset when accessing the register in
6591 2013-05-30 Yuri Rumyantsev <yuri.s.rumyantsev@intel.com>
6592 Igor Zamyatin <igor.zamyatin@intel.com>
6594 Silvermont (SLM) architecture pipeline model, tuning and
6596 * config.gcc: Add slm config options and target.
6598 * config/i386/slm.md: New.
6600 * config/i386/driver-i386.c (host_detect_local_cpu): Check movbe.
6602 * gcc/config/i386/i386-c.c (ix86_target_macros_internal): New case
6604 (ix86_target_macros_internal): Likewise.
6606 * gcc/config/i386/i386.c (slm_cost): New cost.
6607 (m_SLM): New macro flag.
6608 (initial_ix86_tune_features): Set m_SLM.
6609 (x86_accumulate_outgoing_args): Likewise.
6610 (x86_arch_always_fancy_math_387): Likewise.
6611 (processor_target_table): Add slm cost.
6612 (cpu_names): Add slm cpu name.
6613 (x86_option_override_internal): Set SLM ISA.
6614 (ix86_issue_rate): New case PROCESSOR_SLM.
6615 (ia32_multipass_dfa_lookahead): Likewise.
6616 (fold_builtin_cpu): Add slm.
6618 * config/i386/i386.h (TARGET_SLM): New target macro.
6619 (target_cpu_default): Add TARGET_CPU_DEFAULT_slm.
6620 (processor_type): Add PROCESSOR_SLM.
6622 * config/i386/i386.md (cpu): Add new value "slm".
6623 (slm.md): Include slm.md.
6625 2013-05-30 Bernd Schmidt <bernds@codesourcery.com>
6626 Zhenqiang Chen <zhenqiang.chen@linaro.org>
6628 * config/arm/arm-protos.h: Add and update function protos.
6629 * config/arm/arm.c (use_simple_return_p): New added.
6630 (thumb2_expand_return): Check simple_return flag.
6631 * config/arm/arm.md: Add simple_return and conditional simple_return.
6632 * config/arm/iterators.md: Add iterator for return and simple_return.
6634 2013-05-30 Zhenqiang Chen <zhenqiang.chen@linaro.org>
6636 * config/arm/arm.c (arm_add_cfa_adjust_cfa_note): New added.
6637 (arm_emit_multi_reg_pop): Add REG_CFA_ADJUST_CFA notes.
6638 (arm_emit_vfp_multi_reg_pop): Likewise.
6639 (thumb2_emit_ldrd_pop): Likewise.
6640 (arm_expand_epilogue): Add misc REG_CFA notes.
6641 (arm_unwind_emit): Skip REG_CFA_ADJUST_CFA and REG_CFA_RESTORE.
6643 2013-05-29 Lawrence Crowl <crowl@google.com>
6645 * config/arm/t-arm: Update for below.
6647 * config/arm/arm.c (arm_libcall_uses_aapcs_base::libcall_htab):
6648 Change type to hash_table. Update dependent calls and types.
6650 * config/i386/t-cygming: Update for below.
6652 * config/i386/t-interix: Update for below.
6654 * config/i386/winnt.c (i386_pe_section_type_flags::htab):
6655 Change type to hash_table. Update dependent calls and types.
6656 (i386_find_on_wrapper_list::wrappers): Likewise.
6658 * config/ia64/t-ia64: Update for below.
6660 * config/ia64/ia64.c (bundle_state_table):
6661 Change type to hash_table. Update dependent calls and types.
6663 * config/mips/mips.c (mips_reorg_process_insns::htab):
6664 Change type to hash_table. Update dependent calls and types.
6666 * config/sol2.c (solaris_comdat_htab):
6667 Change type to hash_table. Update dependent calls and types.
6669 * config/t-sol2: Update for above.
6671 2013-05-29 Teresa Johnson <tejohnson@google.com>
6673 * passes.c (dump_passes): Use FOR_EACH_FUNCTION since
6674 functions are not yet marked as defined.
6676 2013-05-29 Michael Meissner <meissner@linux.vnet.ibm.com>
6677 Pat Haugen <pthaugen@us.ibm.com>
6678 Peter Bergner <bergner@vnet.ibm.com>
6680 * config/rs6000/vector.md (VEC_I): Add support for new power8 V2DI
6684 (vrotl<mode>3): Likewise.
6685 (vashl<mode>3): Likewise.
6686 (vlshr<mode>3): Likewise.
6687 (vashr<mode>3): Likewise.
6689 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
6690 support for power8 V2DI builtins.
6692 * config/rs6000/rs6000-builtin.def (abs_v2di): Add support for
6693 power8 V2DI builtins.
6694 (vupkhsw): Likewise.
6695 (vupklsw): Likewise.
6696 (vaddudm): Likewise.
6701 (vpkudum): Likewise.
6702 (vpksdss): Likewise.
6703 (vpkudus): Likewise.
6704 (vpksdus): Likewise.
6709 (vsubudm): Likewise.
6710 (vcmpequd): Likewise.
6711 (vcmpgtsd): Likewise.
6712 (vcmpgtud): Likewise.
6713 (vcmpequd_p): Likewise.
6714 (vcmpgtsd_p): Likewise.
6715 (vcmpgtud_p): Likewise.
6716 (vupkhsw): Likewise.
6717 (vupklsw): Likewise.
6718 (vaddudm): Likewise.
6723 (vpksdss): Likewise.
6724 (vpksdus): Likewise.
6725 (vpkudum): Likewise.
6726 (vpkudus): Likewise.
6731 (vsubudm): Likewise.
6733 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Add
6734 support for power8 V2DI instructions.
6736 * config/rs6000/altivec.md (UNSPEC_VPKUHUM): Add support for
6737 power8 V2DI instructions. Combine pack and unpack insns to use an
6738 iterator for each mode. Check whether a particular mode supports
6739 Altivec instructions instead of just checking TARGET_ALTIVEC.
6740 (UNSPEC_VPKUWUM): Likewise.
6741 (UNSPEC_VPKSHSS): Likewise.
6742 (UNSPEC_VPKSWSS): Likewise.
6743 (UNSPEC_VPKUHUS): Likewise.
6744 (UNSPEC_VPKSHUS): Likewise.
6745 (UNSPEC_VPKUWUS): Likewise.
6746 (UNSPEC_VPKSWUS): Likewise.
6747 (UNSPEC_VPACK_SIGN_SIGN_SAT): Likewise.
6748 (UNSPEC_VPACK_SIGN_UNS_SAT): Likewise.
6749 (UNSPEC_VPACK_UNS_UNS_SAT): Likewise.
6750 (UNSPEC_VPACK_UNS_UNS_MOD): Likewise.
6751 (UNSPEC_VUPKHSB): Likewise.
6752 (UNSPEC_VUNPACK_HI_SIGN): Likewise.
6753 (UNSPEC_VUNPACK_LO_SIGN): Likewise.
6754 (UNSPEC_VUPKHSH): Likewise.
6755 (UNSPEC_VUPKLSB): Likewise.
6756 (UNSPEC_VUPKLSH): Likewise.
6758 (VI_char): Likewise.
6759 (VI_scalar): Likewise.
6760 (VI_unit): Likewise.
6762 (VP_small): Likewise.
6763 (VP_small_lc): Likewise.
6764 (VU_char): Likewise.
6765 (add<mode>3): Likewise.
6766 (altivec_vaddcuw): Likewise.
6767 (altivec_vaddu<VI_char>s): Likewise.
6768 (altivec_vadds<VI_char>s): Likewise.
6769 (sub<mode>3): Likewise.
6770 (altivec_vsubcuw): Likewise.
6771 (altivec_vsubu<VI_char>s): Likewise.
6772 (altivec_vsubs<VI_char>s): Likewise.
6773 (altivec_vavgs<VI_char>): Likewise.
6774 (altivec_vcmpbfp): Likewise.
6775 (altivec_eq<mode>): Likewise.
6776 (altivec_gt<mode>): Likewise.
6777 (altivec_gtu<mode>): Likewise.
6778 (umax<mode>3): Likewise.
6779 (smax<mode>3): Likewise.
6780 (umin<mode>3): Likewise.
6781 (smin<mode>3): Likewise.
6782 (altivec_vpkuhum): Likewise.
6783 (altivec_vpkuwum): Likewise.
6784 (altivec_vpkshss): Likewise.
6785 (altivec_vpkswss): Likewise.
6786 (altivec_vpkuhus): Likewise.
6787 (altivec_vpkshus): Likewise.
6788 (altivec_vpkuwus): Likewise.
6789 (altivec_vpkswus): Likewise.
6790 (altivec_vpks<VI_char>ss): Likewise.
6791 (altivec_vpks<VI_char>us): Likewise.
6792 (altivec_vpku<VI_char>us): Likewise.
6793 (altivec_vpku<VI_char>um): Likewise.
6794 (altivec_vrl<VI_char>): Likewise.
6795 (altivec_vsl<VI_char>): Likewise.
6796 (altivec_vsr<VI_char>): Likewise.
6797 (altivec_vsra<VI_char>): Likewise.
6798 (altivec_vsldoi_<mode>): Likewise.
6799 (altivec_vupkhsb): Likewise.
6800 (altivec_vupkhs<VU_char>): Likewise.
6801 (altivec_vupkls<VU_char>): Likewise.
6802 (altivec_vupkhsh): Likewise.
6803 (altivec_vupklsb): Likewise.
6804 (altivec_vupklsh): Likewise.
6805 (altivec_vcmpequ<VI_char>_p): Likewise.
6806 (altivec_vcmpgts<VI_char>_p): Likewise.
6807 (altivec_vcmpgtu<VI_char>_p): Likewise.
6808 (abs<mode>2): Likewise.
6809 (vec_unpacks_hi_v16qi): Likewise.
6810 (vec_unpacks_hi_v8hi): Likewise.
6811 (vec_unpacks_lo_v16qi): Likewise.
6812 (vec_unpacks_hi_<VP_small_lc>): Likewise.
6813 (vec_unpacks_lo_v8hi): Likewise.
6814 (vec_unpacks_lo_<VP_small_lc>): Likewise.
6815 (vec_pack_trunc_v8h): Likewise.
6816 (vec_pack_trunc_v4si): Likewise.
6817 (vec_pack_trunc_<mode>): Likewise.
6819 * config/rs6000/altivec.h (vec_vaddudm): Add defines for power8
6821 (vec_vmaxsd): Likewise.
6822 (vec_vmaxud): Likewise.
6823 (vec_vminsd): Likewise.
6824 (vec_vminud): Likewise.
6825 (vec_vpksdss): Likewise.
6826 (vec_vpksdus): Likewise.
6827 (vec_vpkudum): Likewise.
6828 (vec_vpkudus): Likewise.
6829 (vec_vrld): Likewise.
6830 (vec_vsld): Likewise.
6831 (vec_vsrad): Likewise.
6832 (vec_vsrd): Likewise.
6833 (vec_vsubudm): Likewise.
6834 (vec_vupkhsw): Likewise.
6835 (vec_vupklsw): Likewise.
6837 2013-05-29 Jan Hubicka <jh@suse.cz>
6839 * cgraph.h (symtab_node_base): Add definition, alias and analyzed
6840 flags; reorder rest of fields in more consistent way.
6841 (varpool_node): Remove analyzed, finalized and alias.
6842 (cgraph_ndoe): Likewise.
6843 (symtab_alias_ultimate_target): New function.
6844 (cgraph_function_node): Move offline.
6845 (cgraph_reset_node): Declare.
6846 (cgraph_comdat_can_be_unshared_p): Remove.
6847 (varpool_remove_initializer): Declare.
6848 (varpool_first_defined_variable, varpool_next_defined_variable
6849 cgraph_first_defined_function, cgraph_next_defined_function): Update.
6850 (cgraph_function_with_gimple_body_p): Update.
6851 (varpool_all_refs_explicit_p): Update.
6852 (symtab_alias_target): New function.
6853 (cgraph_alias_aliased_node, varpool_alias_aliased_node): Rename to ...
6854 (cgraph_alias_target, varpool_alias_target): .. this one; simplify.
6855 (cgraph_function_or_thunk_node): Simplify using
6856 symtab_alias_ultimate_target.
6857 (varpool_variable_node): Likewise.
6858 * cgraph.c (cgraph_create_function_alias): Update.
6859 (cgraph_add_thunk): Update.
6860 (cgraph_remove_node): Update.
6861 (dump_cgraph_node): Do not dump removed flags.
6862 (cgraph_function_body_availability): Update.
6863 (cgraph_propagate_frequency): Update.
6864 (verify_cgraph_node): Check sanity of local flag.
6865 (cgraph_function_node): Move here from cgraph.h; revamp for
6866 cgraph_function_or_thunk_node.
6867 * lto-symtab.c (lto_varpool_replace_node): Update.
6868 (lto_symtab_resolve_can_prevail_p): Update.
6869 (lto_symtab_merge_cgraph_nodes): Update.
6870 * ipa-cp.c (determine_versionability, initialize_node_lattices,
6871 propagate_constants_accross_call, devirtualization_time_bonus,
6872 ipcp_propagate_stage): Update.
6873 * tree-emutls.c (create_emultls_var, ipa_lower_emutls): Update.
6874 * ipa-inline-transform.c (clone_inlined_nodes,
6875 preserve_function_body_p): Update.
6876 * ipa-reference.c (propagate): Update.
6877 (write_node_summary_p): Update.
6878 * toplev.c (wrapup_global_declaration_2): Update.
6879 * cgraphunit.c (cgraph_analyze_function): Rename to ...
6880 (analyze_function) ... this one.
6881 (cgraph_process_new_functions): Update.
6882 (cgraph_reset_node): Export.
6883 (cgraph_finalize_function): Update.
6884 (cgraph_add_new_function): Update.
6885 (process_function_and_variable_attributes): Update.
6886 (varpool_finalize_decl): Update.
6887 (symbol_finalized): Remove.
6888 (symbol_finalized_and_needed): Rename to ...
6889 (symbol_defined_and_needed): ... update.
6890 (cgraph_analyze_functions): Update.
6891 (handle_alias_pairs): Update.
6892 (mark_functions_to_output): Update.
6893 (assemble_thunk): Update.
6894 (output_in_order): Update.
6895 (output_weakrefs): Update.
6896 (finalize_compilation_unit): Update.
6897 * lto-cgraph.c (reachable_from_other_partition_p, lto_output_node,
6898 lto_output_varpool_node, compute_ltrans_boundary, input_overwrite_node,
6899 input_node, input_varpool_node): Update.
6900 * dbxout.c (dbxout_expand_expr): Update.
6901 * cgraphclones.c (cgraph_clone_node): Update.
6902 (cgraph_copy_node_for_versioning): Update.
6903 (cgraph_materialize_clone): Update.
6904 (cgraph_materialize_all_clones): Update.
6905 * ipa-pure-const.c (analyze_function, pure_const_write_summary,
6906 propagate_pure_const, propagate_nothrow): Update.
6907 * lto-streamer-out.c (lto_output, write_symbol): Update.
6908 * ipa-utils.c (ipa_reverse_postorder): Update.
6909 * ipa-inline.c (can_inline_edge_p): Update.
6910 (update_caller_keys, ipa_inline): Update.
6911 * dwarf2out.c (reference_to_unused,
6912 premark_types_used_by_global_vars_helper): Update.
6913 * tree-eh.c (tree_could_trap_p): Update.
6914 * ipa-split.c (consider_split, execute_split_functions): Update.
6915 * ipa.c (cgraph_non_local_node_p_1, cgraph_local_node_p,
6916 has_addr_references_p): Update; move ahead in file for better
6918 (process_references): Simplify.
6919 (symtab_remove_unreachable_nodes): Update; cleanup way function/var
6921 (cgraph_comdat_can_be_unshared_p): Make static.
6922 (cgraph_externally_visible_p): Update.
6923 (varpool_externally_visible_p): Update.
6924 (function_and_variable_visibility): Update.
6925 * trans-mem.c (get_cg_data, ipa_tm_mayenterirr_function,
6926 ipa_tm_mark_force_output_node): Update.
6927 * ipa-inline-analysis.c (dump_inline_summary, initialize_inline_failed,
6928 estimate_edge_devirt_benefit, inline_generate_summary,
6929 inline_write_summary): Update.
6930 * gimple-fold.c (can_refer_decl_in_current_unit_p): Update.
6931 * ipa-prop.c (ipa_compute_jump_functions): Update.
6932 (ipa_print_node_params, ipa_prop_read_section,
6933 ipa_update_after_lto_read, read_replacements_section): Update.
6934 * varasm.c (mark_decl_referenced): Update.
6935 (assemble_alias, dump_tm_clone_pairs): Update.
6936 * tree-inline.c (copy_bb): Update.
6937 (estimate_num_insns, optimize_inline_calls, tree_function_versioning):
6939 * symtab.c (dump_symtab_base): Print new flags.
6940 (verify_symtab_base): Verify new flags.
6941 (symtab_alias_ultimate_target): New function.
6942 * tree-ssa-structalias.c (get_constraint_for_ssa_var,
6943 create_variable_info_for, associate_varinfo_to_alias, ipa_pta_execute):
6945 * passes.c (ipa_write_summaries, ipa_write_optimization_summaries):
6947 * i386.c (ix86_get_function_versions_dispatcher,
6948 ix86_generate_version_dispatcher_body): Update.
6949 (fold_builtin_cpu): Use varpool_add_new_variable.
6950 * varpool.c (varpool_remove_initializer): Break out from ...
6951 (varpool_remove_node): ... this one.
6952 (dump_varpool_node, varpool_node_for_asm,
6953 cgraph_variable_initializer_availability, varpool_analyze_node,
6954 varpool_assemble_decl, varpool_remove_unreferenced_decls,
6955 varpool_finalize_named_section_flags, varpool_create_variable_alias):
6958 2013-05-29 Jan Hubicka <jh@suse.cz>
6960 * passes.c (init_optimization_passes): Move OMP expansion into lowering.
6962 2013-05-29 Easwaran Raman <eraman@google.com>
6964 PR tree-optimization/57442
6965 * tree-ssa-reassoc.c (appears_later_in_bb): Return correct value
6966 when control exits the main loop.
6968 2013-05-29 Sandeep Kumar Singh <Sandeep.Singh2@kpitcummins.com>
6970 * rx/rx.h (TARGET_CPU_CPP_BUILTINS): Add macros for RX100, RX200,
6972 * rx/rx.opt: Add macro for rx100 with string rx100 and value RX100.
6973 * rx/rx-opts.h (rx_cpu_types): Add new cpu type rx100.
6974 * rx/t-rx: Add rx100 under multi library matches option for nofpu
6977 2013-05-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6979 PR tree-optimization/57441
6980 * gimple-ssa-strength-reduction.c (analyze_candidates_and_replace):
6981 Don't limit size of incr_vec to number of candidates.
6983 2013-05-29 Steve Ellcey <sellcey@imgtec.com>
6985 * config/mips/mti-linux.h (SYSROOT_SUFFIX_SPEC): Add micromips
6986 and mips16 directories.
6987 * config/mips/t-mti-linux (MULTILIB_OPTIONS): Add micromips and mips16.
6988 (MULTILIB_DIRNAMES): Ditto.
6989 (MULTILIB_EXCEPTIONS): Add new exceptions.
6990 * config/mips/t-mti-elf (MULTILIB_OPTIONS): Add micromips.
6991 (MULTILIB_DIRNAMES): Ditto.
6992 (MULTILIB_EXCEPTIONS): Add new exceptions.
6994 2012-05-29 Chris Schlumberger-Socha <chris.schlumberger-socha@arm.com>
6995 Marcus Shawcroft <marcus.shawcroft@arm.com>
6997 * config/aarch64/aarch64-protos.h (aarch64_symbol_type): Define
6998 SYMBOL_TINY_ABSOLUTE.
6999 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately): Handle
7000 SYMBOL_TINY_ABSOLUTE.
7001 (aarch64_expand_mov_immediate): Likewise.
7002 (aarch64_classify_symbol): Likewise.
7003 (aarch64_mov_operand_p): Remove ATTRIBUTE_UNUSED.
7004 Permit SYMBOL_TINY_ABSOLUTE.
7005 * config/aarch64/predicates.md (aarch64_mov_operand): Permit CONST.
7007 2013-05-29 Chris Schlumberger-Socha <chris.schlumberger-socha@arm.com>
7008 Marcus Shawcroft <marcus.shawcroft@arm.com>
7010 * config/aarch64/aarch64.c (aarch64_classify_symbol): Remove comment.
7011 Refactor if/switch. Replace gcc_assert with if.
7013 2013-05-29 Ganesh Gopalasubramanian <Ganesh.Gopalasubramanian@amd.com>
7015 * config/i386/i386.c (initial_ix86_tune_features): Enable
7016 FP Reassociation for AMD bdver1 and bdver2.
7018 2013-05-29 Martin Jambor <mjambor@suse.cz>
7020 * tree-cfg.c (verify_expr): Verify that BIT_FIELD_REF, REALPART_EXPR
7021 and IMAGPART_EXPR do not occur within other handled_components.
7023 2013-05-29 Richard Biener <rguenther@suse.de>
7025 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Guard vinfo
7026 access on whether the use is in the BB we currently try to
7028 (vect_bb_vectorization_profitable_p): Pass the BB we currently
7029 vectorize to vect_bb_slp_scalar_cost.
7031 2013-05-29 Richard Biener <rguenther@suse.de>
7033 * tree-vect-slp.c (vect_bb_slp_scalar_cost): New function
7034 computing scalar cost offsetted by stmts that are kept live
7036 (vect_bb_vectorization_profitable_p): Use vect_bb_slp_scalar_cost
7037 for computation of scalar cost.
7039 2013-05-28 Steve Ellcey <sellcey@mips.com>
7041 * config/mips/mips-cpus.def (mips32r2): Change processor type.
7043 2013-05-28 Balaji V. Iyer <balaji.v.iyer@intel.com>
7045 * doc/extend.texi (C Extensions): Added documentation about Cilk Plus
7046 array notation built-in reduction functions.
7047 * doc/passes.texi (Passes): Added documentation about changes done
7049 * doc/invoke.texi (C Dialect Options): Added documentation about
7050 the -fcilkplus flag.
7051 * Makefile.in (C_COMMON_OBJS): Added c-family/array-notation-common.o.
7052 (BUILTINS_DEF): Depend on cilkplus.def.
7053 * builtins.def: Include cilkplus.def. Define DEF_CILKPLUS_BUILTIN.
7054 * builtin-types.def: Define BT_FN_INT_PTR_PTR_PTR.
7055 * cilkplus.def: New file.
7057 2013-05-28 Joern Rennecke <joern.rennecke@embecosm.com>
7059 PR rtl-optimization/57439
7060 * postreload.c (move2add_use_add2_insn): Use gen_lowpart_common.
7062 2013-05-28 Easwaran Raman <eraman@google.com>
7064 PR tree-optimization/57337
7065 * tree-ssa-reassoc.c (appears_later_in_bb): New function.
7066 (find_insert_point): Correctly identify the insertion point
7067 when two statements with the same UID is compared.
7069 2013-05-28 Richard Biener <rguenther@suse.de>
7071 PR tree-optimization/56787
7072 * tree-vect-data-refs.c (vect_analyze_data_refs): Drop clobbers
7073 from the list of data references.
7074 * tree-vect-loop.c (vect_determine_vectorization_factor): Skip
7076 (vect_analyze_loop_operations): Likewise.
7077 (vect_transform_loop): Remove clobbers.
7079 2013-05-28 Martin Jambor <mjambor@suse.cz>
7081 * tree-cfg.c (verify_expr): Verify that BIT_FIELD_REFs, IMAGPART_EXPRs
7082 and REALPART_EXPRs have scalar type.
7084 2013-05-28 Richard Biener <rguenther@suse.de>
7086 PR tree-optimization/57411
7087 * tree-ssa-copy.c (may_propagate_copy): Cannot propagate
7089 * tree-ssa-dom.c (eliminate_const_or_copy): Special-case
7090 virtual operand propagation.
7092 2013-05-28 Eric Botcazou <ebotcazou@adacore.com>
7094 * config/sparc/sparc.c (sparc_expand_vec_perm_bmask): Use %g0 as
7095 destination register for bmasksi_vis.
7096 (vector_init_bshuffle): Likewise.
7097 * config/sparc/sparc.md (vec_perm_constv8qi): Likewise.
7099 2013-05-28 Eric Botcazou <ebotcazou@adacore.com>
7101 * doc/invoke.texi (SPARC Options): Document -mfix-ut699.
7102 * builtins.c (expand_builtin_mathfn) <BUILT_IN_SQRT>: Try to widen the
7103 mode if the instruction isn't available in the original mode.
7104 * config/sparc/sparc.opt (mfix-ut699): New option.
7105 * config/sparc/sparc.md (muldf3_extend): Disable if -mfix-ut699.
7106 (divdf3): Turn into expander.
7107 (divdf3_nofix): New insn.
7108 (divdf3_fix): Likewise.
7109 (divsf3): Disable if -mfix-ut699.
7110 (sqrtdf2): Turn into expander.
7111 (sqrtdf2_nofix): New insn.
7112 (sqrtdf2_fix): Likewise.
7113 (sqrtsf2): Disable if -mfix-ut699.
7115 2013-05-27 Richard Biener <rguenther@suse.de>
7118 * omp-low.c (expand_omp_atomic_pipeline): Use the correct latch
7119 block for the new loop.
7121 2013-05-27 Richard Biener <rguenther@suse.de>
7123 PR tree-optimization/57343
7124 * tree-ssa-loop-niter.c (number_of_iterations_ne_max): Do not
7125 use multiple_of_p if not TYPE_OVERFLOW_UNDEFINED.
7126 (number_of_iterations_cond): Do not build the folded tree.
7128 2013-05-27 Richard Biener <rguenther@suse.de>
7132 * fold-const.c (operand_equal_p): Compare FIELD_DECLs with
7133 OEP_CONSTANT_ADDRESS_OF retained.
7135 PR tree-optimization/57417
7136 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Fix test
7138 (set_ssa_val_to): Compare addresses using
7139 get_addr_base_and_unit_offset.
7141 2013-05-27 Joern Rennecke <joern.rennecke@embecosm.com>
7143 PR rtl-optimization/56833
7144 * postreload.c (move2add_record_mode): New function.
7145 (move2add_record_sym_value, move2add_valid_value_p): Likewise.
7146 (move2add_use_add2_insn): Use move2add_record_sym_value.
7147 (move2add_use_add3_insn): Likewise.
7148 (reload_cse_move2add): Use move2add_valid_value_p and
7149 move2add_record_mode. Invalidate call-clobbered and REG_INC
7150 affected regs by setting reg_mode to VOIDmode.
7151 (move2add_note_store): Don't pretend the inside of a SUBREG is
7152 the actual destination. Invalidate single/leading registers by
7153 setting reg_mode to VOIDmode.
7154 Use move2add_record_sym_value, move2add_valid_value_p and
7155 move2add_record_mode.
7157 2013-05-27 Richard Biener <rguenther@suse.de>
7159 PR tree-optimization/57396
7160 * tree-affine.c (double_int_constant_multiple_p): Properly
7161 return false for val == 0 and div != 0.
7163 2013-05-25 Richard Sandiford <rdsandiford@googlemail.com>
7165 * config/mips/mips.h: Use #elif in preprocessor conditions.
7167 2013-05-25 Richard Sandiford <rdsandiford@googlemail.com>
7170 * config/mips/constraints.md (kl): New constraint.
7171 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Delete.
7172 (divmod<mode>4_internal): Rename to divmod<mode>4. Use "kl" as the
7173 constraint for operand 0. Split after CSE for MIPS16. Emit a move
7175 (udivmod<mode>4_internal): Likewise udivmod<mode>4.
7177 2013-05-25 Richard Sandiford <rdsandiford@googlemail.com>
7180 * config/mips/mips.c (mips_can_inline_p): New function.
7181 (TARGET_CAN_INLINE_P): Define.
7183 2013-05-25 Steven Bosscher <steven@gcc.gnu.org>
7185 * sched-int.h (ds_t, dw_t): Make unsigned int.
7186 Fix documentation that describes how all the ds_t bits are used.
7187 Reserve the last bit for delayed-branch scheduling.
7188 (BITS_PER_DEP_STATUS): Move to ds_t typedef.
7189 (BITS_PER_DEP_WEAK): Fix definition and documentation.
7190 (gen_dep_weak_1): Remove prototype.
7191 * sched-deps.c (get_dep_weak_1): Make static.
7192 * target.def (speculate_insn, needs_block_p, gen_spec_check,
7193 get_insn_spec_ds, get_insn_checked_ds): Adjust hook prototypes.
7194 * doc/tm.texi: Regenerate.
7195 * config/ia64/ia64.c (ia64_needs_block_p): Update prototype.
7197 2013-05-24 Steven Bosscher <steven@gcc.gnu.org>
7200 * haifa-sched.c (sched_extend_bb): Ignore DEBUG_INSNs.
7202 2013-05-24 Nathan Sidwell <nathan@codesourcery.com>
7203 Sandra Loosemore <sandra@codesourcery.com>
7205 * config.gcc (powerpc-*): Allow native for with-cpu.
7207 2013-05-24 Jeff Law <law@redhat.com>
7209 PR tree-optimization/57124
7210 * tree-vrp.c (simplify_cond_using_ranges): Only simplify a
7211 conversion feeding a condition if the range has an overflow
7212 if -fstrict-overflow. Add warnings for when we do make the
7215 2013-05-24 Dehao Chen <dehao@google.com>
7217 * tree-cfg.c (locus_discrim_map): Fix the typo.
7218 (locus_discrim_hasher): Likewise.
7219 (locus_discrim_hasher::hash): Likewise.
7220 (locus_discrim_hasher::equal): Likewise.
7222 2013-05-24 Martin Jambor <mjambor@suse.cz>
7224 PR tree-optimization/57294
7225 * cgraph.h (ipa_record_stmt_references): Declare.
7226 * cgraphbuild.c (ipa_record_stmt_references): New function.
7227 (build_cgraph_edges): Use ipa_record_stmt_references.
7228 (rebuild_cgraph_edges): Likewise.
7229 (cgraph_rebuild_references): Likewise.
7230 * ipa-prop.c (ipa_modify_call_arguments): Discard references
7231 associated with the old statement and build references from the
7232 newly built statements.
7233 * ipa-ref.c (ipa_remove_stmt_references): New function.
7234 * ipa-ref.h (ipa_remove_stmt_references): Declare.
7236 2013-05-24 Vladimir Makarov <vmakarov@redhat.com>
7238 * lra-constraints.c (emit_spill_move): Use smaller mode for
7240 (check_and_process_move): Consider mem-reg moves for secondary
7242 (curr_insn_transform): Don't lose insns emitted before for
7243 secondary memory moves.
7244 (inherit_in_ebb): Mark defined reg. Add usage only if it is not a
7245 reg set up in the current insn.
7247 2013-05-24 Dehao Chen <dehao@google.com>
7249 * tree-cfg.c (locus_descrim_hasher::hash): Change discriminator
7251 (locus_descrim_hasher::equal): Likewise.
7252 (build_gimple_cfg): New discriminator assignment algorithm.
7253 (make_edges): Likewise.
7254 (next_discriminator_for_locus): Likewise.
7255 (same_line_p): Likewise.
7256 (assign_discriminators): Likewise.
7257 (make_cond_expr_edges): Likewise.
7258 (make_gimple_switch_edges): Likewise.
7259 (make_goto_expr_edges): Likewise.
7260 (make_gimple_asm_edges): Likewise.
7262 2013-05-24 Ian Bolton <ian.bolton@arm.com>
7264 * config/aarch64/aarch64.c (aarch64_print_operand): Change the
7265 X format specifier to only display bottom 16 bits.
7266 * config/aarch64/aarch64.md (insv_imm<mode>): Allow any size of
7267 immediate to match for operand 2, since it will be masked.
7269 2013-05-24 Richard Biener <rguenther@suse.de>
7271 PR tree-optimization/57287
7272 * tree-ssa-uninit.c (compute_uninit_opnds_pos): Disregard
7273 all SSA names that occur in abnormal PHIs.
7275 2013-05-24 Alexander Ivchenko <alexander.ivchenko@intel.com>
7278 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Check
7279 that index is not negative.
7281 2013-05-24 Eric Botcazou <ebotcazou@adacore.com>
7283 PR rtl-optimization/55177
7284 * simplify-rtx.c (simplify_unary_operation_1) <NOT>: Deal with BSWAP.
7285 (simplify_byte_swapping_operation): New.
7286 (simplify_binary_operation_1): Call it for AND, IOR and XOR.
7287 (simplify_relational_operation_1): Deal with BSWAP.
7289 2013-05-23 Richard Henderson <rth@redhat.com>
7292 * config/i386/i386.c (ix86_seh_fixup_eh_fallthru): New.
7293 (ix86_reorg): Call it.
7295 2013-05-23 Uros Bizjak <ubizjak@gmail.com>
7298 * config/alpha/alpha.md (unspec): Add UNSPEC_XFLT_COMPARE.
7299 * config/alpha/alpha.c (alpha_emit_xfloating_compare): Construct
7300 REG_EQUAL note as UNSPEC_XFLT_COMPARE unspec.
7302 2013-05-23 Christian Bruel <christian.bruel@st.com>
7305 * config/arm/arm.c (arm_dwarf_register_span): Do not use dbx number.
7307 2013-05-23 Chris Schlumberger-Socha <chris.schlumberger-socha@arm.com>
7308 Marcus Shawcroft <marcus.shawcroft@arm.com>
7310 * config/aarch64/aarch64.md (*movdi_aarch64): Replace Usa with S.
7311 * config/aarch64/constraints.md (Usa): Remove.
7312 * doc/md.texi (AArch64 Usa): Remove.
7314 2013-05-23 Chris Schlumberger-Socha <chris.schlumberger-socha@arm.com>
7315 Marcus Shawcroft <marcus.shawcroft@arm.com>
7317 * config/aarch64/aarch64-protos.h (aarch64_mov_operand_p): Define.
7318 * config/aarch64/aarch64.c (aarch64_mov_operand_p): Define.
7319 * config/aarch64/predicates.md (aarch64_const_address): Remove.
7320 (aarch64_mov_operand): Use aarch64_mov_operand_p.
7322 2013-05-23 Vidya Praveen <vidyapraveen@arm.com>
7324 * config/aarch64/aarch64-simd.md (clzv4si2): Support for CLZ
7325 instruction (AdvSIMD).
7326 * config/aarch64/aarch64-builtins.c
7327 (aarch64_builtin_vectorized_function): Handler for BUILT_IN_CLZ.
7328 * config/aarch64/aarch-simd-builtins.def: Entry for CLZ.
7330 2013-05-23 Martin Jambor <mjambor@suse.cz>
7333 * tree.h (contains_bitfld_component_ref_p): Declare.
7334 * tree-sra.c (contains_bitfld_comp_ref_p): Move...
7335 * tree.c (contains_bitfld_component_ref_p): ...here. Adjust its
7337 * ipa-prop.c (determine_known_aggregate_parts): Check that LHS does
7338 not access a bit-field. Assert all final offsets are byte-aligned.
7340 2013-05-23 Richard Biener <rguenther@suse.de>
7342 PR tree-optimization/57380
7343 * tree-ssa-phiprop.c (propagate_with_phi): Do not require at
7344 least one invariant or re-used load.
7345 * passes.c (init_optimization_passes): Move pass_phiprop before
7348 2013-05-23 James Greenhalgh <james.greenhalgh@arm.com>
7350 * config/aarch64/aarch64-simd.md
7351 (aarch64_cm<optab>di): Add clobber of CC_REGNUM to unsplit pattern.
7353 2013-05-23 Richard Biener <rguenther@suse.de>
7356 * fold-const.c (operand_equal_p): Compare FIELD_DECLs with
7357 OEP_CONSTANT_ADDRESS_OF retained.
7359 2013-05-23 Jakub Jelinek <jakub@redhat.com>
7362 * expmed.c (store_split_bit_field): If op0 is a REG or SUBREG of a REG,
7363 don't lower unit. Handle unit not being always BITS_PER_WORD.
7365 2013-05-23 Richard Biener <rguenther@suse.de>
7367 PR rtl-optimization/57341
7368 * ira.c (validate_equiv_mem_from_store): Use anti_dependence
7369 instead of true_dependence.
7371 2013-05-22 David Malcolm <dmalcolm@redhat.com>
7373 * bb-reorder.c (branch_threshold): Make const.
7374 (exec_threshold): Ditto.
7376 2013-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
7377 Pat Haugen <pthaugen@us.ibm.com>
7378 Peter Bergner <bergner@vnet.ibm.com>
7380 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions): Add
7381 documentation for the power8 crypto builtins.
7383 * config/rs6000/t-rs6000 (MD_INCLUDES): Add crypto.md.
7385 * config/rs6000/rs6000-builtin.def (BU_P8V_AV_1): Add support
7386 macros for defining power8 builtin functions.
7387 (BU_P8V_AV_2): Likewise.
7388 (BU_P8V_AV_P): Likewise.
7389 (BU_P8V_VSX_1): Likewise.
7390 (BU_P8V_OVERLOAD_1): Likewise.
7391 (BU_P8V_OVERLOAD_2): Likewise.
7392 (BU_CRYPTO_1): Likewise.
7393 (BU_CRYPTO_2): Likewise.
7394 (BU_CRYPTO_3): Likewise.
7395 (BU_CRYPTO_OVERLOAD_1): Likewise.
7396 (BU_CRYPTO_OVERLOAD_2): Likewise.
7397 (XSCVSPDP): Fix typo, point to the correct instruction.
7398 (VCIPHER): Add power8 crypto builtins.
7399 (VCIPHERLAST): Likewise.
7400 (VNCIPHER): Likewise.
7401 (VNCIPHERLAST): Likewise.
7402 (VPMSUMB): Likewise.
7403 (VPMSUMH): Likewise.
7404 (VPMSUMW): Likewise.
7405 (VPERMXOR_V2DI): Likewise.
7406 (VPERMXOR_V4SI: Likewise.
7407 (VPERMXOR_V8HI: Likewise.
7408 (VPERMXOR_V16QI: Likewise.
7409 (VSHASIGMAW): Likewise.
7410 (VSHASIGMAD): Likewise.
7412 (VPERMXOR): Likewise.
7413 (VSHASIGMA): Likewise.
7415 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Define
7416 __CRYPTO__ if the crypto instructions are available.
7417 (altivec_overloaded_builtins): Add support for overloaded power8
7420 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
7421 support for power8 crypto builtins.
7422 (builtin_function_type): Likewise.
7423 (altivec_init_builtins): Add support for builtins that take vector
7424 long long (V2DI) arguments.
7426 * config/rs6000/crypto.md: New file, define power8 crypto
7429 2013-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
7430 Pat Haugen <pthaugen@us.ibm.com>
7431 Peter Bergner <bergner@vnet.ibm.com>
7433 * doc/invoke.texi (Option Summary): Add power8 options.
7434 (RS/6000 and PowerPC Options): Likewise.
7436 * doc/md.texi (PowerPC and IBM RS6000 constraints): Update to use
7437 constraints.md instead of rs6000.h. Reorder w* constraints. Add
7438 wm, wn, wr documentation.
7440 * gcc/config/rs6000/constraints.md (wm): New constraint for VSX
7441 registers if direct move instructions are enabled.
7442 (wn): New constraint for no registers.
7443 (wq): New constraint for quad word even GPR registers.
7444 (wr): New constraint if 64-bit instructions are enabled.
7445 (wv): New constraint if power8 vector instructions are enabled.
7446 (wQ): New constraint for quad word memory locations.
7448 * gcc/config/rs6000/predicates.md (const_0_to_15_operand): New
7449 constraint for 0..15 for crypto instructions.
7450 (gpc_reg_operand): If VSX allow registers in VSX registers as well
7451 as GPR and floating point registers.
7452 (int_reg_operand): New predicate to match only GPR registers.
7453 (base_reg_operand): New predicate to match base registers.
7454 (quad_int_reg_operand): New predicate to match even GPR registers
7455 for quad memory operations.
7456 (vsx_reg_or_cint_operand): New predicate to allow vector logical
7457 operations in both GPR and VSX registers.
7458 (quad_memory_operand): New predicate for quad memory operations.
7459 (reg_or_indexed_operand): New predicate for direct move support.
7461 * gcc/config/rs6000/rs6000-cpus.def (ISA_2_5_MASKS_EMBEDDED):
7462 Inherit from ISA_2_4_MASKS, not ISA_2_2_MASKS.
7463 (ISA_2_7_MASKS_SERVER): New mask for ISA 2.07 (i.e. power8).
7464 (POWERPC_MASKS): Add power8 options.
7465 (power8 cpu): Use ISA_2_7_MASKS_SERVER instead of specifying the
7468 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
7469 Define _ARCH_PWR8 and __POWER8_VECTOR__ for power8.
7471 * gcc/config/rs6000/rs6000.opt (-mvsx-timode): Add documentation.
7472 (-mpower8-fusion): New power8 options.
7473 (-mpower8-fusion-sign): Likewise.
7474 (-mpower8-vector): Likewise.
7475 (-mcrypto): Likewise.
7476 (-mdirect-move): Likewise.
7477 (-mquad-memory): Likewise.
7479 * gcc/config/rs6000/rs6000.c (power8_cost): Initial definition for
7481 (rs6000_hard_regno_mode_ok): Make PTImode only match even GPR
7483 (rs6000_debug_reg_print): Print the base register class if -mdebug=reg.
7484 (rs6000_debug_vector_unit): Add p8_vector.
7485 (rs6000_debug_reg_global): If -mdebug=reg, print power8 constraint
7486 definitions. Also print fusion state.
7487 (rs6000_init_hard_regno_mode_ok): Set up power8 constraints.
7488 (rs6000_builtin_mask_calculate): Add power8 builtin support.
7489 (rs6000_option_override_internal): Add support for power8.
7490 (rs6000_common_init_builtins): Add debugging for skipped builtins
7492 (rs6000_adjust_cost): Add power8 support.
7493 (rs6000_issue_rate): Likewise.
7494 (insn_must_be_first_in_group): Likewise.
7495 (insn_must_be_last_in_group): Likewise.
7496 (force_new_group): Likewise.
7497 (rs6000_register_move_cost): Likewise.
7498 (rs6000_opt_masks): Likewise.
7500 * config/rs6000/rs6000.h (ASM_CPU_POWER8_SPEC): If we don't have a
7501 power8 capable assembler, default to power7 options.
7502 (TARGET_DIRECT_MOVE): Likewise.
7503 (TARGET_CRYPTO): Likewise.
7504 (TARGET_P8_VECTOR): Likewise.
7505 (VECTOR_UNIT_P8_VECTOR_P): Define power8 vector support.
7506 (VECTOR_UNIT_VSX_OR_P8_VECTOR_P): Likewise.
7507 (VECTOR_MEM_P8_VECTOR_P): Likewise.
7508 (VECTOR_MEM_VSX_OR_P8_VECTOR_P): Likewise.
7509 (VECTOR_MEM_ALTIVEC_OR_VSX_P): Likewise.
7510 (TARGET_XSCVDPSPN): Likewise.
7511 (TARGET_XSCVSPDPN): Likewsie.
7512 (TARGET_SYNC_HI_QI): Likewise.
7513 (TARGET_SYNC_TI): Likewise.
7514 (MASK_CRYPTO): Likewise.
7515 (MASK_DIRECT_MOVE): Likewise.
7516 (MASK_P8_FUSION): Likewise.
7517 (MASK_P8_VECTOR): Likewise.
7518 (REG_ALLOC_ORDER): Move fr13 to be lower in priority so that the TFmode
7519 temporary used by some of the direct move instructions to get two FP
7520 temporary registers does not force creation of a stack frame.
7521 (VLOGICAL_REGNO_P): Allow vector logical operations in GPRs.
7522 (MODES_TIEABLE_P): Move the VSX tests above the Altivec tests so
7523 that any VSX registers are tieable, even if they are also an
7524 Altivec vector mode.
7525 (r6000_reg_class_enum): Add wm, wr, wv constraints.
7526 (RS6000_BTM_P8_VECTOR): Power8 builtin support.
7527 (RS6000_BTM_CRYPTO): Likewise.
7528 (RS6000_BTM_COMMON): Likewise.
7530 * config/rs6000/rs6000.md (cpu attribute): Add power8.
7531 * config/rs6000/rs6000-opts.h (PROCESSOR_POWER8): Likewise.
7532 (enum rs6000_vector): Add power8 vector support.
7534 2013-05-22 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
7538 * config/arm/arm.c (any_sibcall_uses_r3): Rename to ..
7539 (any_sibcall_could_use_r3): this and handle indirect calls.
7540 (arm_get_frame_offsets): Rename use of any_sibcall_uses_r3.
7542 2013-05-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7544 * config/rs6000/rs6000.h (MALLOC_ABI_ALIGNMENT): New #define.
7546 2013-05-22 Richard Biener <rguenther@suse.de>
7549 * profile.c (branch_prob): Do not split blocks that are
7550 abnormally receiving from ECF_RETURNS_TWICE functions.
7552 2013-05-22 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
7554 * recog.c (offsettable_address_addr_space_p): Fix calculation of
7555 address mode. Move pointer mode initialization to the same place.
7557 2013-05-22 Michael Zolotukhin <michael.v.zolotukhin@intel.com>
7559 * read-rtl.c (copy_rtx_for_iterators): Continue applying iterators
7560 while it has any effect.
7562 2013-05-21 Easwaran Raman <eraman@google.com>
7564 PR tree-optimization/57322
7565 * tree-ssa-reassoc.c (build_and_add_sum): If a BB is empty, set the
7566 UID of the statement added to the BB to be 1.
7568 2013-05-21 Jakub Jelinek <jakub@redhat.com>
7570 PR tree-optimization/57331
7571 * tree-vrp.c (simplify_cond_using_ranges): Don't optimize comparison
7572 of conversion from pointer type to integral type with integer.
7574 2013-05-21 Martin Jambor <mjambor@suse.cz>
7577 * ipa-prop.c (ipa_read_node_info): Process param_used and
7578 controlled_uses in the same order as when writing.
7580 2013-05-21 Magnus Granberg <baldrick@free.fr>
7583 * Makefile.in (PLUGIN_HEADERS): Add $(TARGET_H).
7585 2013-05-21 Richard Biener <rguenther@suse.de>
7587 PR tree-optimization/57318
7588 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Do not
7589 estimate stmts with side-effects as likely eliminated.
7591 2013-05-21 Richard Biener <rguenther@suse.de>
7593 PR tree-optimization/57330
7594 * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Properly
7595 preserve the call stmts fntype.
7597 2013-05-21 Richard Biener <rguenther@suse.de>
7599 PR tree-optimization/57303
7600 * tree-ssa-sink.c (statement_sink_location): Improve killing
7601 stmt detection and properly handle self-assignments.
7603 2013-05-21 Christian Bruel <christian.bruel@st.com>
7605 * dwarf2out.c (multiple_reg_loc_descriptor): Use dbx_reg_number for
7606 spanning registers. LEAF_REG_REMAP is supported only for contiguous
7607 registers. Set register size out of the PARALLEL loop.
7609 2013-05-20 Oleg Endo <olegendo@gcc.gnu.org>
7612 * config/sh/sh.md (fmasf4): Remove empty constraints strings.
7613 (*fmasf4, *fmasf4_media): New insns.
7615 2013-05-19 Richard Sandiford <rdsandiford@googlemail.com>
7617 * config/mips/mips.h (BASE_INSN_LENGTH, NOP_INSN_LENGTH): New macros.
7618 * config/mips/mips.c (mips_symbol_insns, mips_address_insns)
7619 (mips_const_insns, mips_split_const_insns, mips_load_store_insns)
7620 (mips_idiv_insns): Update the comments to say that the returned
7621 instruction counts are in units of BASE_INSN_LENGTH.
7622 (mips_adjust_insn_length): Multiply the mips_load_label_num_insns
7623 by BASE_INSN_LENGTH rather than 4. Add the jump separately,
7624 using 2 rather than 4 as the length of indirect MIPS16 and
7625 microMIPS jumps. Use NOP_INSN_LENGTH rather than 4 as the
7626 length of a NOP. Don't divide MIPS16 lengths by 2.
7627 (mips16_split_long_branches): Assume a branch is long if the
7628 length is greater than 4 rather than 8.
7629 * config/mips/mips.md (length): Give MIPS16 lengths directly,
7630 rather than multiplying them by 2. Multiply instruction counts
7631 by BASE_INSN_LENGTH rather than 4.
7632 (*jump_mips16, tls_get_tp_mips16_<mode>)
7633 (*tls_get_tp_mips16_call_<mode>): Divide lengths by 2.
7635 2013-05-19 Richard Sandiford <rdsandiford@googlemail.com>
7637 * config/mips/mips.md (extended_mips16): Remove branch case.
7638 (length): Remove duplicated extended_mips16 test.
7640 2013-05-19 Richard Sandiford <rdsandiford@googlemail.com>
7642 * config/mips/t-sde: Don't build 64-bit microMIPS multilibs.
7644 2013-05-18 Richard Sandiford <rdsandiford@googlemail.com>
7646 * recog.h (Recog_data): Rename to...
7647 (recog_data_d): ...this.
7648 (recog_data): Update accordingly.
7649 * recog.c (recog_data): Likewise.
7650 * reload.c (save_recog_data): Likewise.
7651 * config/picochip/picochip.c (picochip_saved_recog_data): Likewise.
7652 (picochip_save_recog_data, picochip_restore_recog_data): Likewise.
7654 2013-05-17 Julian Brown <julian@codesourcery.com>
7656 * gcse.c (compute_ld_motion_mems): If a non-simple MEM is
7657 found in a REG_EQUAL note, invalidate it.
7659 2013-05-17 Easwaran Raman <eraman@google.com>
7661 * tree-ssa-reassoc.c (find_insert_point): New function.
7662 (insert_stmt_after): Likewise.
7663 (get_def_stmt): Likewise.
7664 (ensure_ops_are_available): Likewise.
7665 (not_dominated_by): Likewise.
7666 (rewrite_expr_tree): Do not move statements beyond what is
7667 necessary. Remove call to swap_ops_for_binary_stmt...
7668 (reassociate_bb): ... and move it here.
7669 (build_and_add_sum): Assign UIDs for new statements.
7670 (linearize_expr): Likewise.
7671 (do_reassoc): Renumber gimple statement UIDs.
7673 2013-05-17 Jan Hubicka <jh@suse.cz>
7675 * lto-symtab.c (lto_symtab_merge_cgraph_nodes): Resolve cross module
7677 * cgraph.c (dump_cgraph_node): Do not ice on unresolved alias.
7678 * cgraphunit.c (handle_alias_pairs): Store target of unresolved
7680 (output_weakrefs): Update.
7682 2013-05-17 Po-Chun Chang <pchang9@cs.wisc.edu>
7683 Martin Jambor <mjambor@suse.cz>
7686 * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Break when a
7687 value that corresponds to the given aggval is found in values vector.
7689 2013-05-17 Uros Bizjak <ubizjak@gmail.com>
7691 * config/i386/driver-i386.c (host_detect_local_cpu): Pass mmx, 3dnow,
7692 sse, sse2, sse3, ssse3 and sse4a flags to options.
7694 2013-05-17 David Malcolm <dmalcolm@redhat.com>
7696 * gengtype-state.c: (s_expr_writer): New class, to handle
7697 prettifying of output layout of s-expressions.
7698 (state_writer): New class, to write out gtype.state.
7699 (state_written_type_count): Move this variable into member data of
7701 (s_expr_writer::s_expr_writer): New code: constructor for new class
7702 (state_writer::state_writer(): ditto
7703 (s_expr_writer::write_new_line): New function
7704 (s_expr_writer::write_any_indent): ditto
7705 (s_expr_writer::begin_s_expr): ditto
7706 (s_expr_writer::end_s_expr): ditto
7707 (write_state_fileloc): convert to method of state_writer...
7708 (state_writer:: write_state_fileloc): ...and use methods of
7709 s_expr_writer to write indentation into the gtype.state output file
7710 to visually represent the hierarchical structure of the list
7712 (write_state_fields): ditto, renaming to...
7713 (state_writer::write_state_fields)
7714 (write_state_a_string): ditto, renaming to...
7715 (state_writer::write_state_a_string)
7716 (write_state_string_option): ditto, renaming to...
7717 (state_writer::write_state_string_option)
7718 (write_state_type_option): ditto, renaming to...
7719 (state_writer::write_state_type_option)
7720 (write_state_nested_option): ditto, renaming to...
7721 (state_writer::write_state_nested_option)
7722 (write_state_option): ditto, renaming to...
7723 (state_writer::write_state_option)
7724 (write_state_options): ditto, renaming to...
7725 (state_writer::write_state_options)
7726 (write_state_lang_bitmap): ditto, renaming to...
7727 (state_writer::write_state_lang_bitmap)
7728 (write_state_version): ditto, renaming to...
7729 (state_writer::write_state_version)
7730 (write_state_scalar_type): ditto, renaming to...
7731 (state_writer::write_state_scalar_type)
7732 (write_state_string_type): ditto, renaming to...
7733 (state_writer::write_state_string_type)
7734 (write_state_undefined_type): ditto, renaming to...
7735 (state_writer::write_state_undefined_type)
7736 (write_state_struct_union_type): ditto, renaming to...
7737 (state_writer::write_state_struct_union_type)
7738 (write_state_struct_type): ditto, renaming to...
7739 (state_writer::write_state_struct_type)
7740 (write_state_user_struct_type): ditto, renaming to...
7741 (state_writer::write_state_user_struct_type)
7742 (write_state_lang_struct_type): ditto, renaming to...
7743 (state_writer::write_state_lang_struct_type)
7744 (write_state_param_struct_type): ditto, renaming to...
7745 (state_writer::write_state_param_struct_type)
7746 (write_state_pointer_type): ditto, renaming to...
7747 (state_writer::write_state_pointer_type)
7748 (write_state_array_type): ditto, renaming to...
7749 (state_writer::write_state_array_type)
7750 (write_state_gc_used): ditto, renaming to...
7751 (state_writer::write_state_gc_used)
7752 (write_state_common_type_content): ditto, renaming to...
7753 (state_writer::write_state_common_type_content)
7754 (write_state_type): ditto, renaming to...
7755 (state_writer::write_state_type)
7756 (write_state_pair_list): ditto, renaming to...
7757 (state_writer::write_state_pair_list)
7758 (write_state_pair): ditto, renaming to...
7759 (state_writer::write_state_pair)
7760 (write_state_typedefs): ditto, renaming to...
7761 (state_writer::write_state_typedefs)
7762 (write_state_structures): ditto, renaming to...
7763 (state_writer::write_state_structures)
7764 (write_state_param_structs): ditto, renaming to...
7765 (state_writer::write_state_param_structs)
7766 (write_state_variables): ditto, renaming to...
7767 (state_writer::write_state_variables)
7768 (write_state_srcdir): ditto, renaming to...
7769 (state_writer::write_state_srcdir)
7770 (write_state_files_list): ditto, renaming to...
7771 (state_writer::write_state_files_list)
7772 (write_state_languages): ditto, renaming to...
7773 (state_writer::write_state_languages)
7774 (write_state): create a state_writer instance and use it when
7775 writing out the state file
7777 2013-05-17 Mike Stump <mikestump@comcast.net>
7779 PR rtl-optimization/57304
7780 * web.c (union_match_dups): Ensure that DF_REF_LOC exists before
7781 accessing DF_REF_REAL_LOC.
7783 2013-05-17 Jakub Jelinek <jakub@redhat.com>
7785 PR rtl-optimization/57281
7786 PR rtl-optimization/57300
7787 * config/i386/i386.md (extendsidi2_1 dead reg splitter): Remove.
7788 (extendsidi2_1 peephole2s): Add instead 2 new peephole2s, that undo
7789 what the other splitter did if the registers are dead.
7791 2013-05-17 Richard Biener <rguenther@suse.de>
7793 * tree-ssa-alias.c (stmt_kills_ref_p_1): Properly compare
7796 2013-05-17 Jakub Jelinek <jakub@redhat.com>
7798 * gcc.c (SANITIZER_SPEC): Reject -fsanitize=address -fsanitize=thread
7801 2013-05-17 Marek Polacek <polacek@redhat.com>
7803 * tree-ssa-strlen.c (handle_char_store): Don't invalidate cached
7804 length when doing non-zero store of storing '\0' to '\0'.
7806 2013-05-17 Jakub Jelinek <jakub@redhat.com>
7808 * tree-vect-patterns.c (vect_recog_rotate_pattern): For
7809 vect_external_def oprnd1 with loop_vinfo, try to emit
7810 optional cast, negation and and stmts on the loop preheader
7811 edge instead of into the pattern def seq.
7813 PR tree-optimization/57051
7814 * fold-const.c (const_binop) <case VEC_LSHIFT_EXPR,
7815 case VEC_RSHIFT_EXPR>: Fix BYTES_BIG_ENDIAN handling.
7817 2013-05-16 Nick Clifton <nickc@redhat.com>
7819 * config/rl78/rl78.c (rl78_attribute_table): Add naked.
7820 (rl78_is_naked_func): New function.
7821 (rl78_expand_prologue): Skip prologue generation for naked functions.
7822 (rl78_expand_epilogue): Skip epilogue generation for naked functions.
7823 * doc/extend.texi (naked): Add RL78 to the list of processors
7824 that supports this attribute.
7826 2013-05-16 Jeff Law <law@redhat.com>
7828 * Makefile.in (tree-switch-conversion.o): Depend on $(OPTABS_H).
7830 2013-05-16 Uros Bizjak <ubizjak@gmail.com>
7832 * config/i386/driver-i386.c (host_detect_local_cpu): Determine
7833 cache parameters using detect_caches_amd also for CYRIX,
7834 NSC and TM2 signatures.
7836 2013-05-16 Uros Bizjak <ubizjak@gmail.com>
7837 Dzianis Kahanovich <mahatma@eu.by>
7841 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
7842 VIA/Centaur processors and determine their cache parameters
7843 using detect_caches_amd.
7845 2013-05-16 Teresa Johnson <tejohnson@google.com>
7847 * cfgrtl.c (verify_hot_cold_block_grouping): Return err.
7848 (rtl_verify_edges): New function.
7849 (rtl_verify_bb_insns): Ditto.
7850 (rtl_verify_bb_pointers): Ditto.
7851 (rtl_verify_bb_insn_chain): Ditto.
7852 (rtl_verify_fallthru): Ditto.
7853 (rtl_verify_bb_layout): Ditto.
7854 (rtl_verify_flow_info_1): Outline checks into new functions.
7855 (rtl_verify_flow_info): Ditto.
7857 2013-05-16 Steve Ellcey <sellcey@imgtec.com>
7859 * cfghooks.c (copy_bbs): Add update_dominance argument.
7860 * cfghooks.h (copy_bbs): Update prototype.
7861 * tree-cfg.c (gimple_duplicate_sese_region):
7862 Add update_dominance argument.
7863 * tree-flow.h (gimple_duplicate_sese_region): Update prototype.
7864 * tree-ssa-loop-ch.c (copy_loop_headers): Update
7865 gimple_duplicate_sese_region call.
7866 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg):
7867 Update copy_bbs call.
7868 * cfgloopmanip.c (duplicate_loop_to_header_edge): Ditto.
7869 * trans-mem.c (ipa_uninstrument_transaction): Ditto.
7871 2013-05-16 Jakub Jelinek <jakub@redhat.com>
7873 * tree-vectorizer.h (NUM_PATTERNS): Increment.
7874 * tree-vect-patterns.c (vect_vect_recog_func_ptrs): Add
7875 vect_recog_rotate_pattern.
7876 (vect_recog_rotate_pattern): New function.
7878 2013-05-16 Jason Merrill <jason@redhat.com>
7880 * Makefile.in (LLINKER): New variable.
7881 (mostlyclean): Remove link mutex.
7882 * configure.ac: Handle --enable-link-mutex.
7883 * lock-and-run.sh: New script.
7885 2013-05-16 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
7888 * config/arm/arm.c (arm_function_ok_for_sibcall): Add check
7891 2013-05-16 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
7893 * reorg.c (link_cc0_insns): Wrap in #ifdef HAVE_cc0.
7895 2013-05-16 Greta Yorsh <Greta.Yorsh@arm.com>
7897 * config/arm/arm-protos.h (gen_movmem_ldrd_strd): New declaration.
7898 * config/arm/arm.c (next_consecutive_mem): New function.
7899 (gen_movmem_ldrd_strd): Likewise.
7900 * config/arm/arm.md (movmemqi): Update condition and code.
7901 (unaligned_loaddi, unaligned_storedi): New patterns.
7903 2013-05-16 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
7905 * config.gcc: Obsolete *-*-solaris2.9*.
7906 * doc/install.texi (Specific, *-*-solaris2*): Document it.
7908 2013-05-16 Richard Biener <rguenther@suse.de>
7910 * passes.c (init_optimization_passes): Move pass_parallelize_loops
7911 earlier, after GRAPHITE transforms and IV canonicalization.
7913 2013-05-16 Jakub Jelinek <jakub@redhat.com>
7915 * omp-low.c (extract_omp_for_data): For collapsed loops,
7916 if at least one of the loops is known at compile time to
7917 iterate zero times, set count to 0.
7918 (expand_omp_regimplify_p): New function.
7919 (expand_omp_for_generic): For collapsed loops, if at least
7920 one of the loops isn't known to iterate at least once,
7921 add runtime check with setting count to 0.
7922 (expand_omp_for_static_nochunk, expand_omp_for_static_chunk):
7923 For unsigned types if it isn't known at compile time that
7924 the loop will iterate at least once, add runtime check to bypass
7925 the whole loop if initial condition isn't true.
7927 2013-05-16 Nathan Sidwell <nathan@codesourcery.com>
7929 * varasm.c (default_use_anchors_for_symbol_p): Use decl_replaceable_p.
7931 2013-05-16 Marc Glisse <marc.glisse@inria.fr>
7934 * fold-const.c (fold_ternary_loc) <VEC_COND_EXPR>: Disable some
7935 transformations to avoid an infinite loop.
7937 2013-05-16 Marek Polacek <polacek@redhat.com>
7939 * tree-scalar-evolution.c (scev_const_prop): Add more dumps.
7941 2013-05-15 Leif Ekblad <leif@rdos.net>
7943 * config/i386/i386.c (ix86_decompose_address): Use
7944 DEFAULT_TLS_SEG_REG to access TLS segment register.
7945 * config/i386/i386.h (DEFAULT_TLS_SEG_REG): New define.
7946 * config/i386/rdos.h (DEFAULT_TLS_SEG_REG): Ditto.
7947 (TARGET_TLS_DIRECT_SEG_REFS_DEFAULT): Ditto.
7949 2013-05-15 Richard Sandiford <rdsandiford@googlemail.com>
7952 * config/mips/mips.c (mips_function_ok_for_sibcall): Don't allow
7953 sibling calls to functions that would normally be lazily bound,
7954 unless $gp is call-clobbered.
7956 2013-05-15 Uros Bizjak <ubizjak@gmail.com>
7958 * config/i386/i386.c (ix86_option_override_internal): Update
7959 processor_alias_table for missing PTA_PRFCHW and PTA_FXSR flags. Add
7960 PTA_POPCNT to corei7 entry. Do not enable SSE prefetch on
7961 non-SSE 3dNow! targets. Enable TARGET_PRFCHW for TARGET_3DNOW targets.
7962 * config/i386/i386.md (prefetch): Enable for TARGET_PRFCHW instead
7964 (*prefetch_3dnow): Enable for TARGET_PRFCHW only.
7966 2013-05-15 Andreas Schwab <schwab@suse.de>
7968 * config/m68k/m68k.md (*rotlhi3_lowpart, *rotlqi3_lowpart): Name
7969 for rotlhi3+1 and rotlqi3+1, resp. Fix reference to non-existing
7972 2013-05-15 Teresa Johnson <tejohnson@google.com>
7974 * loop-unroll.c (report_unroll_peel): Check decision before
7975 emitting unroll/peel message.
7977 2013-05-15 Teresa Johnson <tejohnson@google.com>
7979 * function.h (has_bb_partition): New rtl_data flag.
7980 (bb_reorder_complete): Ditto.
7981 * cfgcleanup.c (try_crossjump_to_edge): Check for has_bb_partition
7982 instead of flag_reorder_blocks_and_partition.
7983 * cfgrtl.c (verify_hot_cold_block_grouping): Moved from bb-reorder.c,
7984 with some enhancements.
7985 (rtl_verify_flow_info_1): Call verify_hot_cold_block_grouping.
7986 * bb-reorder.c (connect_traces): Check for has_bb_partition
7987 instead of flag_reorder_blocks_and_partition.
7988 (verify_hot_cold_block_grouping): Moved to cfgrtl.c.
7989 (reorder_basic_blocks): Set bb_reorder_complete flag, remove call to
7990 verify_hot_cold_block_grouping.
7991 (partition_hot_cold_basic_blocks): Set has_bb_partition.
7993 2013-05-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
7996 * config/arm/predicates.md (call_insn_operand): New predicate.
7997 * config/arm/constraints.md ("Cs", "Ss"): New constraints.
7998 * config/arm/arm.md (*call_insn, *call_value_insn): Match only
7999 if insn is not a tail call.
8000 (*sibcall_insn, *sibcall_value_insn): Adjust for tailcalling through
8002 * config/arm/arm.h (enum reg_class): New caller save register class.
8003 (REG_CLASS_NAMES): Likewise.
8004 (REG_CLASS_CONTENTS): Likewise.
8005 * config/arm/arm.c (arm_function_ok_for_sibcall): Allow tailcalling
8008 2013-05-15 Richard Biener <rguenther@suse.de>
8010 * tree-vect-loop.c (vect_transform_loop): Use MSG_NOTE instead
8011 of MSG_OPTIMIZED_LOCATIONS.
8012 * tree-vect-slp.c (vect_make_slp_decision): Likewise.
8013 (vect_slp_transform_bb): Indicate location in MSG_OPTIMIZED_LOCATIONS
8015 * tree-vectorizer.c (vectorize_loops): Use MSG_NOTE instead
8016 of MSG_OPTIMIZED_LOCATIONS.
8017 (execute_vect_slp): Likewise.
8018 * tree-vect-loop-manip.c (vect_do_peeling_for_loop_bound): Likewise.
8019 (vect_create_cond_for_alias_checks): Likewise.
8020 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Likewise.
8021 (vect_recog_widen_mult_pattern): Likewise.
8022 (vect_recog_widen_sum_pattern): Likewise.
8023 (vect_recog_over_widening_pattern): Likewise.
8024 (vect_recog_widen_shift_pattern): Likewise.
8025 (vect_recog_vector_vector_shift_pattern): Likewise.
8026 (vect_recog_divmod_pattern): Likewise.
8027 (vect_recog_mixed_size_cond_pattern): Likewise.
8028 (vect_recog_bool_pattern): Likewise.
8029 (vect_pattern_recog_1): Likewise.
8031 2013-05-15 Martin Jambor <mjambor@suse.cz>
8033 * ipa-prop.c (ipa_make_edge_direct_to_target): Redirect calls to
8034 non-functions to builtin_unreachable.
8035 * ipa-inline-transform.c (inline_call): Do not assert estimates were
8036 correct when new direct edges were discovered.
8038 2013-05-15 Martin Jambor <mjambor@suse.cz>
8040 * ipa-prop.c (ipa_print_node_jump_functions): Print symbol order in
8041 header, print symbol order instead of node uid, print more information
8042 about indirect edge targets.
8043 (ipa_make_edge_direct_to_target): Print symbol order instead of node
8045 (ipa_make_edge_direct_to_target): Likewise.
8046 (remove_described_reference): Likewise.
8047 (propagate_controlled_uses): Likewise.
8048 (ipa_print_node_params): Also print symbol order.
8049 (ipcp_transform_function): Print symbol order instead of node uids.
8050 * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Likewise.
8051 (cgraph_get_create_real_symbol_node): Likewise.
8052 * ipa-cp.c (print_lattice): Likewise.
8053 (print_all_lattices): Likewise.
8054 (determine_versionability): Likewise.
8055 (initialize_node_lattices): Likewise.
8056 (estimate_local_effects): Likewise.
8057 (update_profiling_info): Likewise.
8058 (create_specialized_node): Likewise.
8059 (perhaps_add_new_callers): Likewise.
8060 (decide_about_value): Likewise.
8061 (decide_whether_version_node): Likewise.
8062 (identify_dead_nodes): Likewise.
8063 * ipa-inline-analysis.c (dump_inline_edge_summary): Likewise.
8064 (dump_inline_summary): Likewise.
8065 (estimate_node_size_and_time): Likewise.
8066 (inline_analyze_function): Likewise.
8067 * ipa-inline.c (report_inline_failed_reason): Likewise.
8068 (want_early_inline_function_p): Likewise.
8069 (edge_badness): Likewise.
8070 (update_edge_key): Likewise.
8071 (inline_small_functions): Likewise. Add dumping of order to two other
8073 * ipa-pure-const.c (pure_const_read_summary): Print symbol order
8074 instead of node uids.
8075 (propagate_pure_const): Likewise.
8076 (propagate_pure_const): Likewise.
8077 * ipa-utils.c (dump_cgraph_node_set): Likewise.
8078 * lto-cgraph.c (input_node): Explicitly specify we dump uid.
8079 * lto-symtab.c (lto_cgraph_replace_node): Print symbol order instead
8081 * tree-pretty-print.c (dump_function_header): Likewise.
8082 * tree-sra.c (convert_callers_for_node): Dump in traditional format.
8083 Print symbol order instead of node uids.
8085 2013-05-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
8087 * config/s390/s390.c (s390_register_move_cost): Don't impose the
8088 FPR<->GPR move cost penalty if ldgr/lgdr can be used.
8090 2013-05-15 Richard Biener <rguenther@suse.de>
8092 PR tree-optimization/57275
8093 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Fix
8094 return value for fail to do runtime alias checks for gather loads.
8096 2013-05-15 Jan Hubicka <jh@suse.cz>
8100 * lto-symtab.c (lto_symtab_symbol_p): Add external symbol;
8101 weakrefs are not external.
8102 (lto_symtab_merge_decls): Fix thinko when dealing with
8103 non-lto_symtab decls.
8104 (lto_symtab_merge_cgraph_nodes): Use lto_symtab_symbol_p.
8105 (lto_symtab_prevailing_decl): Get int sync with lto_symtab_symbol_p.
8106 * varpool.c (dump_varpool_node): Dump more flags.
8108 2013-05-15 Ganesh Gopalasubramanian <Ganesh.Gopalasubramanian@amd.com>
8110 * config/i386/i386.c (processor_alias_table): Add instruction
8111 FSGSBASE for AMD bdver3 architecture.
8113 2013-05-14 Jakub Jelinek <jakub@redhat.com>
8115 * tree.c (warn_deprecated_use): Print file:line using locus color.
8116 * diagnostic.c (diagnostic_report_current_module): Print file:line
8117 and file:line:column using locus color.
8119 2013-05-14 Mike Stump <mikestump@comcast.net>
8121 * gdbinit.in: Add __null.
8123 2013-05-14 Mike Stump <mikestump@comcast.net>
8125 * recog.h: Rename struct recog_data to Recog_data.
8126 * recog.c: Likewise.
8127 * reload.c (can_reload_into): Likewise.
8128 * config/picochip/picochip.c: Likewise.
8130 2013-05-14 Mike Stump <mikestump@comcast.net>
8132 * web.c (union_match_dups): Also check DF_REF_REAL_LOC.
8134 2013-05-14 Steven Bosscher <steven@gcc.gnu.org>
8136 * resource.h (struct resources): Remove unch_memory member.
8137 (CLEAR_RESOURCE): Don't clear unch_memory.
8138 * resource.c (mark_referenced_resources): Don't set it.
8139 (mark_set_resources): Likewise.
8140 (mark_target_live_regs): Don't clear it.
8141 (init_resource_info): Likewise.
8142 * reorg.c (resource_conflicts_p): Don't compare it.
8143 (redundant_insn): Don't set it.
8145 * rtl.h (next_label, skip_consecutive_labels, link_cc0_insns):
8147 * emit-rtl.c (next_label): Remove unused function.
8148 (skip_consecutive_labels, link_cc0_insns): Move to ...
8149 * reorg.c (skip_consecutive_labels, link_cc0_insns): ... here, the
8150 only place where these functions are used, and make them static.
8152 2013-05-14 Marc Glisse <marc.glisse@inria.fr>
8154 * fold-const.c (fold_negate_expr): Handle vectors.
8155 (fold_truth_not_expr): Make it static.
8156 (fold_invert_truthvalue): New static function.
8157 (invert_truthvalue_loc): Handle vectors. Do not call
8158 fold_truth_not_expr directly.
8159 (fold_unary_loc) <BIT_NOT_EXPR>: Handle comparisons.
8160 <TRUTH_NOT_EXPR>: Do not cast to boolean.
8161 (fold_comparison): Handle vector constants.
8162 (fold_binary_loc) <TRUTH_XOR_EXPR>: Remove redundant code.
8163 (fold_ternary_loc) <VEC_COND_EXPR>: Adapt more COND_EXPR optimizations.
8164 * tree.h (fold_truth_not_expr): Remove declaration.
8166 2013-05-14 James Greenhalgh <james.greenhalgh@arm.com>
8168 * config/aarch64/aarch64-simd.md
8169 (aarch64_vcond_internal<mode>): Rename to...
8170 (aarch64_vcond_internal<mode><mode>): ...This, for integer modes.
8171 (aarch64_vcond_internal<VDQF_COND:mode><VDQF:mode>): ...This for
8172 float modes. Clarify all iterator modes.
8173 (vcond<mode><mode>): Use new name for vcond expanders.
8174 (vcond<v_cmp_result><mode>): Likewise.
8175 (vcondu<mode><mode>: Likewise.
8176 * config/aarch64/iterators.md (VDQF_COND): New.
8178 2013-05-14 Marc Glisse <marc.glisse@inria.fr>
8181 * fold-const.c (fold_binary_loc) <shift>: Use an unsigned
8182 variable for the shift amount. Check that we shift by non-negative
8185 2013-05-14 Chung-Lin Tang <cltang@codesourcery.com>
8188 * config/arm/arm.h (EPILOGUE_USES): Only return true
8189 for LR_REGNUM after epilogue_completed.
8191 2013-05-14 Joern Rennecke <joern.rennecke@embecosm.com>
8193 * config/avr/avr.c (avr_encode_section_info): Bail out if the type
8196 2013-05-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
8199 * configure.ac (gcc_cv_ld_as_needed): Disable before Solaris 11
8200 and Solaris 11+/x86 with gld.
8201 * configure: Regenerate.
8203 2013-05-14 Jakub Jelinek <jakub@redhat.com>
8205 * expmed.c (expand_shift_1): Canonicalize rotates by
8206 constant bitsize / 2 to bitsize - 1.
8207 * simplify-rtx.c (simplify_binary_operation_1) <case ROTATE,
8208 case ROTATERT>: Likewise.
8211 2013-05-10 Jakub Jelinek <jakub@redhat.com>
8213 * config/i386/i386.md (rotateinv): New code attr.
8214 (*<rotate_insn><mode>3_1, *<rotate_insn>si3_1_zext,
8215 *<rotate_insn>qi3_1_slp): Emit rorl %eax instead of
8216 roll $31, %eax, etc.
8218 2013-05-14 Richard Biener <rguenther@suse.de>
8221 * tree-eh.c (sink_clobbers): Give up for successors with
8222 multiple predecessors and no virtual uses.
8224 2013-05-14 Eric Botcazou <ebotcazou@adacore.com>
8226 * config/sparc/sp64-elf.h (CPP_SUBTARGET_SPEC): Delete.
8227 * config/sparc/openbsd64.h (CPP_SUBTARGET_SPEC): Likewise.
8229 2013-05-14 Jakub Jelinek <jakub@redhat.com>
8232 * expr.c (expand_expr_real_2) <case WIDEN_MULT_EXPR>: Handle
8233 the case when both op0 and op1 have VOIDmode.
8235 2013-05-14 Kaushik Phatak <kaushik.phatak@kpitcummins.com>
8237 * config/rl78/rl78.md(mulsi3_g13): Add additional 'nop' required
8238 in multiply-accumulate mode.
8240 2013-05-13 Guozhi Wei <carrot@google.com>
8242 * dwarf2asm.c (dw2_output_indirect_constant_1): Mark new decl STATIC.
8244 2013-05-13 Kai Tietz <ktietz@redhat.com>
8247 * config/i386/cygming.h (TARGET_PECOFF): Define as true.
8248 * config/i386/i386.h (TARGET_PECOFF): Define by default as false.
8249 (PIC_OFFSET_TABLE_REGNUM): Use TARGET_PECOFF.
8250 * config/i386/i386.c (ix86_option_override_internal): Likewise.
8251 (ix86_expand_prologue): Likewise.
8252 (ix86_expand_split_stack_prologue): Likewise.
8253 (legitimate_pic_address_disp_p): Likewise.
8254 (legitimize_pic_address): Likewise.
8255 (legitimize_tls_address): Likewise.
8256 (legitimize_pe_coff_symbol): Likewise.
8257 (output_pic_addr_const): Likewise.
8258 (construct_plt_address): Likewise.
8259 (ix86_expand_call): Likewise.
8260 (x86_output_mi_thunk): Likewise.
8261 (x86_function_profiler): Likewise.
8263 2013-05-13 Sofiane Naci <sofiane.naci@arm.com>
8265 * config/aarch64/aarch64-simd.md (aarch64_simd_mov<mode>): Group
8266 similar switch cases.
8267 (aarch64_simd_mov): Rename to aarch64_split_simd_mov. Update.
8268 (aarch64_simd_mov_to_<mode>low): Delete.
8269 (aarch64_simd_mov_to_<mode>high): Delete.
8270 (move_lo_quad_<mode>): Add w<-r alternative.
8271 (aarch64_simd_move_hi_quad_<mode>): Likewise.
8272 (aarch64_simd_mov_from_*): Update type attribute.
8273 * config/aarch64/aarch64.c (aarch64_split_simd_move): Refacror switch
8276 2013-05-13 Jan Hubicka <jh@suse.cz>
8278 * mode-switching.c (optimize_mode_switching): Set correct RTL profile.
8279 * config/i386/i386.c (ix86_compute_frame_layout,
8280 ix86_expand_epilogue, emit_i387_cw_initialization,
8281 ix86_expand_vector_move_misalign, ix86_fp_comparison_strategy,
8282 ix86_local_alignment): Fix use of size/speed predicates.
8284 2013-05-13 Jakub Jelinek <jakub@redhat.com>
8286 PR tree-optimization/45216
8287 PR tree-optimization/57157
8288 * tree-ssa-forwprop.c (simplify_rotate): Only recognize
8289 the (-Y) & (B - 1) variant if OP is |.
8290 * expmed.c (expand_shift_1): For rotations by const0_rtx just
8291 return shifted. Use (-op1) & (prec - 1) as other_amount
8292 instead of prec - op1.
8294 2013-05-13 Martin Jambor <mjambor@suse.cz>
8297 * ipa-prop.h (IPA_UNDESCRIBED_USE): New macro.
8298 (ipa_constant_data): New type.
8299 (ipa_jump_func): Use ipa_constant_data to hold information about
8300 constant jump functions.
8301 (ipa_get_jf_constant): Adjust to jump function type changes.
8302 (ipa_get_jf_constant_rdesc): New function.
8303 (ipa_param_descriptor): New field controlled_uses.
8304 (ipa_get_controlled_uses): New function.
8305 (ipa_set_controlled_uses): Likewise.
8306 * ipa-ref.h (ipa_find_reference): Declare.
8307 * ipa-prop.c (ipa_cst_ref_desc): New type.
8308 (ipa_print_node_jump_functions_for_edge): Adjust for jump function type
8310 (ipa_set_jf_constant): Likewise. Also create reference descriptions.
8311 New parameter cs. Adjust all callers.
8312 (ipa_analyze_params_uses): Detect uncontrolled and controlled uses.
8313 (remove_described_reference): New function.
8314 (jfunc_rdesc_usable): Likewise.
8315 (try_make_edge_direct_simple_call): Decrement controlled use count,
8316 attempt to remove reference if it hits zero.
8317 (combine_controlled_uses_counters): New function.
8318 (propagate_controlled_uses): Likewise.
8319 (ipa_propagate_indirect_call_infos): Call propagate_controlled_uses.
8320 (ipa_edge_duplication_hook): Duplicate reference descriptions.
8321 (ipa_print_node_params): Print described use counter.
8322 (ipa_write_jump_function): Adjust to jump function type changes.
8323 (ipa_read_jump_function): New parameter CS, pass it to
8324 ipa_set_jf_constant. Adjust caller.
8325 (ipa_write_node_info): Stream controlled use count
8326 (ipa_read_node_info): Likewise.
8327 * cgraph.c (cgraph_mark_address_taken_node): Bail out instead of
8329 * ipa-cp.c (ipcp_discover_new_direct_edges): Decrement controlled use
8330 count. Remove cloning-added reference if it reaches zero.
8331 * ipa-ref.c (ipa_find_reference): New function.
8333 2013-05-13 Ganesh Gopalasubramanian <Ganesh.Gopalasubramanian@amd.com>
8335 * config/i386/i386.c (processor_target_table): Modified default
8336 alignment values for AMD BD and BT architectures.
8338 2013-05-13 Marc Glisse <marc.glisse@inria.fr>
8340 * tree-vect-generic.c (uniform_vector_p): Move ...
8341 * tree.c (uniform_vector_p): ... here.
8342 * tree.h (uniform_vector_p): Declare it.
8343 * fold-const.c (fold_binary_loc) <shift>: Turn the second argument
8346 2013-05-13 Jakub Jelinek <jakub@redhat.com>
8348 PR tree-optimization/57230
8349 * tree-ssa-strlen.c (handle_char_store): Record length for
8350 array store from STRING_CST.
8352 PR tree-optimization/57230
8353 * tree-ssa-strlen.c (handle_char_store): Add missing integer_zerop
8356 2013-05-12 Joern Rennecke <joern.rennecke@embecosm.com>
8358 * config/epiphany/epiphany.c (epiphany_init): Check size of
8359 NUM_MODES_FOR_MODE_SWITCHING.
8360 (epiphany_expand_prologue):
8361 Remove CONFIG_REGNUM initial value handling code.
8362 (epiphany_optimize_mode_switching): Handle EPIPHANY_MSW_ENTITY_CONFIG.
8363 (epiphany_mode_needed, epiphany_mode_entry_exit): Likewise.
8364 (emit_set_fp_mode, epiphany_mode_after): Likewise.
8365 (epiphany_mode_needed) <Handle EPIPHANY_MSW_ENTITY_AND>:
8366 Don't return 1 for FP_MODE_NONE.
8367 * config/epiphany/epiphany.h (NUM_MODES_FOR_MODE_SWITCHING):
8368 Add value for EPIPHANY_MSW_ENTITY_CONFIG.
8369 (EPIPHANY_MSW_ENTITY_CONFIG, EPIPHANY_MSW_ENTITY_NUM): Define.
8370 * config/epiphany/epiphany.md (save_config): New pattern.
8372 2013-05-12 Uros Bizjak <ubizjak@gmail.com>
8374 * config/i386/i386.md (*zero_extendsidi2): Add *x->?r alternative.
8376 2013-05-10 Uros Bizjak <ubizjak@gmail.com>
8378 * config/i386/i386.md (memory): Handle sseishft1.
8379 * config/i386/sse.md (*vec_extractv4si): Remove memory attribute.
8380 (*vec_extractv2di_1): Ditto.
8382 2013-05-10 Vladimir Makarov <vmakarov@redhat.com>
8384 * lra-assigns.c (find_hard_regno_for): Add 1 to the cost of call
8387 2013-05-10 Sebastian Huber <sebastian.huber@embedded-brains.de>
8389 * config/arm/t-rtems-eabi: Remove mthumb/march=armv7 multilib.
8390 Add mthumb/march=armv7-a multilib.
8391 Add mthumb/march=armv7-r multilib.
8392 Add mthumb/march=armv7-a/mfpu=neon/mfloat-abi=hard multilib.
8394 2013-05-10 Ralf Corsépius <ralf.corsepius@rtems.org>
8396 * config/v850/t-rtems: Add more multilibs.
8398 2013-05-10 Richard Biener <rguenther@suse.de>
8400 PR tree-optimization/57214
8401 * tree-ssa-loop-ivcanon.c (propagate_constants_for_unrolling): Do
8402 not propagate from SSA names that occur in abnormal PHI nodes.
8404 2013-05-10 Marc Glisse <marc.glisse@inria.fr>
8406 * stor-layout.c (element_precision): New function.
8407 * machmode.h (element_precision): Declare it.
8408 * tree.c (build_minus_one_cst): New function.
8409 (element_precision): Likewise.
8410 * tree.h (build_minus_one_cst): Declare new function.
8411 (element_precision): Likewise.
8412 * fold-const.c (operand_equal_p): Use element_precision.
8413 (fold_binary_loc): Handle vector types.
8414 * convert.c (convert_to_integer): Use element_precision.
8415 * gimple.c (iterative_hash_canonical_type): Handle complex and vectors
8418 2013-05-10 Richard Sandiford <rdsandiford@googlemail.com>
8420 * config/mips/mips-protos.h (m16_uimm3_b, m16_simm4_1, m16_nsimm4_1)
8421 (m16_simm5_1, m16_nsimm5_1, m16_uimm5_4, m16_nuimm5_4, m16_simm8_1)
8422 (m16_nsimm8_1, m16_uimm8_1, m16_nuimm8_1, m16_uimm8_m1_1, m16_uimm8_4)
8423 (m16_nuimm8_4, m16_simm8_8, m16_nsimm8_8): Delete.
8424 * config/mips/mips.c (m16_check_op, m16_uimm3_b, m16_simm4_1)
8425 (m16_nsimm4_1, m16_simm5_1, m16_nsimm5_1, m16_uimm5_4, m16_nuimm5_4)
8426 (m16_simm8_1, m16_nsimm8_1, m16_uimm8_1, m16_nuimm8_1, m16_uimm8_m1_1)
8427 (m16_uimm8_4, m16_nuimm8_4, m16_simm8_8, m16_nsimm8_8): Delete.
8428 * config/mips/constraints.md (Udb8, Usb5, Usb8, Usd8, Uub8, Uuw5)
8429 (Uuw8): New constraints.
8430 (Usb4): Move into alphabetical order.
8431 * config/mips/predicates.md (db8_operand, sb5_operand, sb8_operand)
8432 (sd8_operand, ub8_operand, uw8_operand): New predicates.
8433 * config/mips/mips.md (*xor<mode>3, *xor<mode>3_mips16): Name
8434 previously unnamed patterns.
8435 (*add<mode>3_mips16, *xor<mode>3_mips16, *<optab>si3_mips16)
8436 (*ashldi3_mips16, *ashrdi3_mips16, *lshrdi3_mips16)
8437 (*slt<u>_<GPR:mode><GPR2:mode>_mips16)
8438 (*sle<u>_<GPR:mode><GPR2:mode>_mips16): Use constraints instead
8439 of set_attr_alternative/if_then_else. Use extended_mips16 instead
8440 of specific lengths.
8442 2013-05-10 Jakub Jelinek <jakub@redhat.com>
8444 * config/i386/i386.md (rotateinv): New code attr.
8445 (*<rotate_insn><mode>3_1, *<rotate_insn>si3_1_zext,
8446 *<rotate_insn>qi3_1_slp): Emit rorl %eax instead of
8447 roll $31, %eax, etc.
8449 PR tree-optimization/45216
8450 PR tree-optimization/57157
8451 * tree-ssa-forwprop.c (simplify_rotate): New function.
8452 (ssa_forward_propagate_and_combine): Call it.
8454 2013-05-10 Richard Biener <rguenther@suse.de>
8456 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Do not
8457 disable peeling when we version for aliasing.
8458 (vector_alignment_reachable_p): Honor explicit user alignment.
8459 (vect_supportable_dr_alignment): Likewise.
8460 * tree-vect-loop-manip.c (vect_can_advance_ivs_p): Use
8461 STMT_VINFO_LOOP_PHI_EVOLUTION_PART instead of recomputing it.
8462 * tree-vect-loop.c (vect_transform_loop): First apply versioning,
8463 then peeling to arrange for the cost-model check to come first.
8465 2013-05-10 Alan Modra <amodra@gmail.com>
8467 * configure.ac (HAVE_AS_TLS): Swap powerpc64 and powerpc cases.
8468 (HAVE_LD_LARGE_TOC): Don't mention AIX in help text.
8469 * configure: Regenerate.
8471 2013-05-10 Alan Modra <amodra@gmail.com>
8474 * varasm.c (default_elf_select_section): Move !DECL_P check..
8475 (get_named_section): ..to here before calling get_section_name.
8477 (default_section_type_flags): Add DECL_P check.
8478 * config/i386/winnt.c (i386_pe_section_type_flags): Likewise.
8479 * config/rs6000/rs6000.c (rs6000_xcoff_section_type_flags): Likewise.
8481 2013-05-09 Joern Rennecke <joern.rennecke@embecosm.com>
8483 * config/epiphany/epiphany.c (epiphany_expand_prologue):
8484 When using gen_stack_adjust_str with a register offset, add a
8485 REG_FRAME_RELATED_EXPR note.
8487 2013-05-09 Uros Bizjak <ubizjak@gmail.com>
8489 * config/i386/sse.md (*vec_extractv4si_0_zext): New pattern.
8490 (*vec_extractv4si_zext_mem): Ditto.
8491 (*vec_extractv2di): Add 0->x and x->x alternatives.
8492 * config/i386/mmx.md (*vec_extractv2si_zext_mem): New pattern.
8493 * config/i386/i386.md (*zero_extendsidi2): Add *Yj->?r alternative.
8495 2013-05-09 Jason Merrill <jason@redhat.com>
8497 N3639 C++1y VLA support
8498 * gimplify.c (gimplify_vla_decl): Don't touch an existing
8501 * tree.c (build_constructor_va): New.
8502 * tree.h: Declare it.
8504 2013-05-09 Martin Jambor <mjambor@suse.cz>
8507 * gimple-fold.c (canonicalize_constructor_val): Call
8508 cgraph_get_create_real_symbol_node instead of cgraph_get_create_node.
8510 2013-05-09 Jan Hubicka <jh@suse.cz>
8511 Richard Biener <rguenther@suse.de>
8514 * symtab.c (symtab_make_decl_local): Do not add private names.
8516 2013-05-09 Jan Hubicka <jh@suse.cz>
8519 * symtab.c (insert_to_assembler_name_hash): Handle clones.
8520 (unlink_from_assembler_name_hash): Likewise.
8521 (symtab_prevail_in_asm_name_hash, symtab_register_node,
8522 symtab_unregister_node, symtab_initialize_asm_name_hash,
8523 change_decl_assembler_name): Update.
8525 2013-05-09 Sofiane Naci <sofiane.naci@arm.com>
8527 * config/aarch64/aarch64.md: New movtf split.
8528 (*movtf_aarch64): Update.
8529 (aarch64_movdi_tilow): Handle TF modes and rename to
8530 aarch64_movdi_<mode>low.
8531 (aarch64_movdi_tihigh): Handle TF modes and rename to
8532 aarch64_movdi_<mode>high
8533 (aarch64_movtihigh_di): Handle TF modes and rename to
8534 aarch64_mov<mode>high_di
8535 (aarch64_movtilow_di): Handle TF modes and rename to
8536 aarch64_mov<mode>low_di
8537 (aarch64_movtilow_tilow): Remove spurious whitespace.
8538 * config/aarch64/aarch64.c (aarch64_split_128bit_move): Handle TFmode
8540 (aarch64_print_operand): Update.
8542 2013-05-09 Alan Modra <amodra@gmail.com>
8544 * configure.ac (HAVE_AS_TLS): Enable tests for powerpcle and
8546 * configure: Regenerate.
8548 2013-05-08 Uros Bizjak <ubizjak@gmail.com>
8550 * config/i386/mmx.md (*vec_extract* splitters): Simplify post-reload
8551 splitter preparation statements.
8552 * config/i386/sse.md (*vec_extract* splitters): Ditto.
8553 (*avx_vperm_broadcast_<mode>): Use adjust_address instead of
8556 2013-05-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8558 * gimple-ssa-strength-reduction.c (count_candidates): Change
8559 return value to int.
8560 (analyze_candidates_and_replace): Change type of length to int.
8562 2013-05-08 Uros Bizjak <ubizjak@gmail.com>
8564 * config/i386/sse.md (PEXTR_MODE, PEXTR_MODEx): Remove.
8565 (*vec_extract<mode>): Use VI12_128 mode iterator.
8566 (*vec_extract<mode>_mem): Ditto.
8567 (*vec_extract*_mem splitters): Merge splitters using VI_128 mode
8570 2013-05-08 Diego Novillo <dnovillo@google.com>
8575 2012-08-17 Diego Novillo <dnovillo@google.com>
8578 * configure.ac: Add libintl.h to AC_CHECK_HEADERS list.
8579 * config.in: Regenerate.
8580 * configure: Regenerate.
8581 * intl.h: Always include libintl.h if HAVE_LIBINTL_H is set.
8583 2013-05-08 Jan Hubicka <jh@suse.cz>
8586 * cgraph.c (cgraph_make_node_local_1): Se unique_name.
8587 * cgraph.h (symtab_node_base): Add unique_name.
8588 * lto-cgraph.c (lto_output_node, lto_output_varpool_node,
8589 input_overwrite_node, input_varpool_node): Stream unique_name.
8590 * cgraphclones.c (cgraph_create_virtual_clone,
8591 cgraph_function_versioning): Set unique_name.
8592 * ipa.c (function_and_variable_visibility): Set unique_name.
8594 2013-05-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8596 * gimple-ssa-strength-reduction.c (find_phi_def): Revert former "fix."
8597 (alloc_cand_and_find_basis): Restrict conditional candidate
8598 processing to CAND_MULTs.
8600 2013-05-08 Jan Hubicka <jh@suse.cz>
8603 lto-symtab.c (lto_symtab_symbol_p): New function.
8604 (lto_symtab_resolve_can_prevail_p, lto_symtab_resolve_symbols,
8605 lto_symtab_resolve_symbols, lto_symtab_merge_decls_2,
8606 lto_symtab_merge_decls_1, lto_symtab_merge_cgraph_nodes_1):
8607 Skip static symbols.
8609 2013-05-08 Paolo Carlini <paolo.carlini@oracle.com>
8611 PR tree-optimization/57200
8612 * tree-ssa-loop-niter.c (do_warn_aggressive_loop_optimizations):
8613 Only call inform if the preceding warning_at returns true.
8615 2013-05-07 Han Shen <shenhan@google.com>
8617 * cfgexpand.c (record_or_union_type_has_array_p): New function.
8618 (expand_used_vars): Add logic handling '-fstack-protector-strong'.
8619 * common.opt (fstack-protector-strong): New option.
8620 * doc/cpp.texi (__SSP_STRONG__): New builtin "__SSP_STRONG__".
8621 * doc/invoke.texi (Optimization Options): Document
8622 "-fstack-protector-strong".
8623 * gcc.c (LINK_SSP_SPEC): Add 'fstack-protector-strong'.
8625 2013-05-06 Steven Bosscher <steven@gcc.gnu.org>
8627 * config/mips/mips.c (mips_machine_reorg2): Return 0.
8629 2013-05-07 Vladimir Makarov <vmakarov@redhat.com>
8631 * ira.c (update_equiv_regs): Add insn having equiv memory even if
8632 it is not lhs of the insn.
8633 (setup_reg_equiv): Remove insn having equiv memory which it is not
8635 * lra-constraints.c (process_address): Try to improve generation
8636 code for address base + disp.
8637 (lra_constraints): Make correct the code for checking insn setting
8638 up backward equivalence. Remove insn only if it is in the init
8640 * lra-eliminations.c (update_reg_eliminate): Change return value.
8641 (lra_eliminate): Use the result.
8643 2013-05-07 Uros Bizjak <ubizjak@gmail.com>
8645 * config/i386/sse.md (ssescalarnummask): New mode attribute.
8646 (PEXTR_MODE, PEXTR_MODEx): New mode iterators.
8647 (*vec_extract<mode>): Merge from *sse4_1_pextrb_memory and
8648 *sse4_1_pextrw_memory using PEXTR_MODE mode iterator. Handle
8649 register target operands.
8650 (*vec_extractv8hi_sse2): New pattern.
8651 (*vec_extractv16qi_zext): Rename from *sse4_1_pextrb_<mode>.
8652 (*vec_extractv8hi_zext): Rename from *sse2_pextrw_<mode>.
8653 (*vec_extract<mode>_mem): New insn and split pattern.
8655 2013-05-07 Christophe Lyon <christophe.lyon@linaro.org>
8657 * config/arm/arm.c (arm_asan_shadow_offset): New function.
8658 (TARGET_ASAN_SHADOW_OFFSET): Define.
8659 * config/arm/linux-eabi.h (ASAN_CC1_SPEC): Define.
8660 (LINUX_OR_ANDROID_CC): Add ASAN_CC1_SPEC.
8662 2013-05-07 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8664 * gimple-ssa-strength-reduction.c (MAX_INCR_VEC_LEN): New constant.
8665 (incr_vec_index): Return -1 if increment not found.
8666 (create_add_on_incoming_edge): Assert if increment not found.
8667 (record_increment): Limit number of increments recorded.
8668 (all_phi_incrs_profitable): Return false if an increment not found.
8669 (replace_profitable_candidates): Don't process increments that were
8671 (analyze_candidates_and_replace): Limit size of incr_vec.
8673 2013-05-07 Richard Biener <rguenther@suse.de>
8675 * calls.c (special_function_p): setjmp-like functions are leaf.
8676 * builtins.def (BUILT_IN_SETJMP): setjmp is leaf.
8677 * tree-inline.c (update_ssa_across_abnormal_edges): Remove assert.
8679 2013-05-07 Sofiane Naci <sofiane.naci@arm.com>
8681 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>): call splitter.
8682 (aarch64_simd_mov<mode>): New expander.
8683 (aarch64_simd_mov_to_<mode>low): New instruction pattern.
8684 (aarch64_simd_mov_to_<mode>high): Likewise.
8685 (aarch64_simd_mov_from_<mode>low): Likewise.
8686 (aarch64_simd_mov_from_<mode>high): Likewise.
8687 (aarch64_dup_lane<mode>): Update.
8688 (aarch64_dup_lanedi): New instruction pattern.
8689 * config/aarch64/aarch64-protos.h (aarch64_split_simd_move): New prototype.
8690 * config/aarch64/aarch64.c (aarch64_split_simd_move): New function.
8692 2013-05-07 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8694 * gimple-ssa-strength-reduction.c (lazy_create_slsr_reg): Remove.
8695 (replace_mult_candidate): Remove unnecessary argument; remove
8696 unnecessary parameter from call to introduce_cast_before_cand.
8697 (replace_unconditional_candidate): Remove unnecessary parameter
8698 from call to replace_mult_candidate.
8699 (replace_conditional_candidate): Likewise.
8700 (insert_initializers): Use make_temp_ssa_name.
8701 (introduce_cast_before_cand): Remove unnecessary argument; use
8703 (replace_one_candidate): Remove unnecessary argument; remove
8704 unnecessary parameter from calls to introduce_cast_before_cand.
8705 (replace_profitable_candidates): Remove unnecessary parameters
8706 from calls to replace_one_candidate.
8708 2013-05-07 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8710 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
8711 phi def as possibly hiding a basis for a CAND_ADD whose operands
8712 have been commuted in the analysis.
8713 (alloc_cand_and_find_basis): Add parms to call to find_phi_def.
8715 2013-05-07 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
8717 * config/aarch64/aarch64.md
8718 (cmp_swp_<optab><ALLX:mode>_shft_<GPI:mode>): Restrict the
8719 shift value between 0-4.
8721 2013-05-07 Richard Biener <rguenther@suse.de>
8723 * double-int.h (rshift): New overload.
8724 * double-int.c (rshift): New function.
8725 * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Optimize.
8726 (create_reference_ops_from_ref): Remove.
8727 (vn_reference_insert): Use shared ops for constructing the
8728 reference and copy it.
8730 2013-05-07 Richard Biener <rguenther@suse.de>
8733 * tree-eh.c (sink_clobbers): Properly propagate
8734 SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
8736 2013-05-07 Jakub Jelinek <jakub@redhat.com>
8738 PR tree-optimization/57149
8739 * tree-ssa-uninit.c (uninit_undefined_value_p): New inline.
8740 (can_skip_redundant_opnd, compute_uninit_opnds_pos,
8741 collect_phi_def_edges, execute_late_warn_uninitialized): Use
8742 uninit_undefined_value_p instead of ssa_undefined_value_p.
8745 * expr.c (expand_expr_addr_expr_1): Handle COMPOUND_LITERAL_EXPR
8746 for modifier == EXPAND_INITIALIZER.
8748 2013-05-07 Anton Blanchard <anton@samba.org>
8750 * configure.ac (HAVE_LD_LARGE_TOC): Use correct linker emulation
8751 for powerpc64 little endian.
8752 * configure: Regenerate.
8754 2013-05-06 Graham Stott <grahams@btinternet.com>
8756 * expmed.c (init_expmed_rtl): Remove unused fields reg_fld, plus_fld,
8757 mult_fld, sdiv_fld1, udiv_fld1, sdiv_32_fld1, smod_32_fld1,
8758 wide_mult_fld1, wide_lshr_fld1, shift_fld1, shift_mult_fld1,
8759 shift_add_fld1, shift_sub0_fld1, shift_sub1_fld1.
8761 2013-05-06 Graham Stott <grahams@btinternet.com>
8763 * gensupport.c (add_predicate_code): Also exclude SCRATCH from rtx
8764 codes which allow non-lvalues.
8766 2013-05-06 Marc Glisse <marc.glisse@inria.fr>
8768 * tree.c (integer_all_onesp) <COMPLEX_CST>: Test that both
8769 components are all 1s.
8770 (integer_minus_onep): New function.
8771 * tree.h (integer_minus_onep): Declare it.
8772 * fold-const.c (fold_binary_loc) <MULT_EXPR>: Test
8773 integer_minus_onep instead of integer_all_onesp.
8775 2013-05-06 Oleg Endo <olegendo@gcc.gnu.org>
8778 * config/sh/sh.md (*cmp_div0s_0, *cmp_div0s_1, *movsicc_div0s): Add
8779 variations of these patterns.
8781 2013-05-06 Uros Bizjak <ubizjak@gmail.com>
8783 * config/i386/i386.md (isa): Add x64_sse4 member.
8784 (enabled): Handle x64_sse4.
8785 (*movdi_internal): Add *x->?r alternative to emit pextrq $0,%xmm,%reg
8786 instruction for 64bit SSE4_1 targets. Update insn attributes.
8787 (*movsi_internal): Add *x->?r alternative to emit pextrd $0,%xmm,%reg
8788 instruction for SSE4_1 targets. Update insn attributes.
8789 * config/i386/sse.md (*vec_extract<ssevecmodelower>_0): Merge
8790 with *sse4_1_pextrd and *sse4_1_pextrq having const_0 selector.
8791 (*vec_extractv2di_1): Merge with *sse4_1_pextrq having
8793 (*vec_extractv4si): Rename from *sse4_1_pextrd.
8794 (*vec_extractv4si_zext): Rename from *sse4_1_pextrd_zext.
8795 (*vec_extract<ssevecmodelower>_0 splitters): Merge splitters together.
8797 2013-05-06 Oleg Endo <olegendo@gcc.gnu.org>
8800 * config/sh/sh.md (tstsi_t_zero_extract_eq): Use QIHISIDI mode iterator.
8802 2013-05-06 Maxim Kuznetsov <maks.kuznetsov@gmail.com>
8804 * final.c (do_assembler_dialects): Don't handle curly braces and
8805 vertical bar escaped by % as dialect delimiters.
8806 (output_asm_insn): Print curly braces and vertical bar if escaped
8807 by % and ASSEMBLER_DIALECT defined.
8808 * doc/tm.texi.in (ASSEMBLER_DIALECT): Document new standard escapes.
8809 * doc/tm.texi: Regenerated.
8811 2013-05-06 Steven Bosscher <steven@gcc.gnu.org>
8813 * config/mips/mips.c: Include tree-pass.h.
8814 (mips_reorg): Split in pre- and post-dbr_schedule parts.
8815 (mips_machine_reorg2): Move mips_reorg post-dbr_schedule parts here.
8816 (pass_mips_machine_reorg2): New machine specific pass.
8817 (insert_pass_mips_machine_reorg2): New pass plugin definition.
8818 (mips_option_override): Register the new pass.
8819 * rtl.h (cleanup_barriers): Remove prototype.
8820 (dbr_schedule): Likewise.
8821 * jump.c (cleanup_barriers): Make static.
8822 * reorg.c (dbr_schedule): Likewise.
8824 2013-05-06 Richard Biener <rguenther@suse.de>
8826 PR tree-optimization/57185
8827 * tree-parloops.c (add_field_for_reduction): Handle anonymous
8830 2013-05-06 Uros Bizjak <ubizjak@gmail.com>
8833 * config/i386/i386.c (add_parameter_dependencies): Add dependence
8834 between "first_arg" and "insn", not "last" and "insn".
8836 2013-05-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8838 * gimple-ssa-strength-reduction.c (slsr_process_phi): Re-enable.
8839 (find_candidates_in_block): Re-enable slsr_process_phi.
8840 (create_phi_basis): Fix double counting of candidate adjustment.
8842 2013-05-06 Richard Biener <rguenther@suse.de>
8845 * tree-cfg.c (gimple_purge_dead_abnormal_call_edges): If
8846 the edge is also fallthru, preserve it and just clear the
8848 * tree-cfgcleanup.c (remove_fallthru_edge): If the edge is
8849 also complex, preserve that and just clear the fallthru flag.
8850 * tree-inline.c (update_ssa_across_abnormal_edges): Also
8851 update virtual operands.
8853 2013-05-06 Alan Modra <amodra@gmail.com>
8855 * config/rs6000/linux.h (DEFAULT_ASM_ENDIAN): Define.
8856 (LINK_OS_LINUX_EMUL): Use ENDIAN_SELECT.
8857 * config/rs6000/linux64.h (DEFAULT_ASM_ENDIAN): Define.
8858 * config/rs6000/sysv4le.h (DEFAULT_ASM_ENDIAN): Define.
8859 (LINK_TARGET_SPEC): Use ENDIAN_SELECT.
8860 * config/rs6000/sysv4.h (DEFAULT_ASM_ENDIAN): Define as -mbig.
8862 2013-05-06 Alan Modra <amodra@gmail.com>
8864 * config/rs6000/sysv4.h (ENDIAN_SELECT): Define, extracted from
8865 (ASM_SPEC): ..here. Emit DEFAULT_ASM_ENDIAN too.
8866 (DEFAULT_ASM_ENDIAN): Define.
8867 (CC1_SPEC, LINK_TARGET_SPEC): Use ENDIAN_SELECT.
8868 * config/rs6000/linux64.h (ASM_SPEC32): Remove endian options.
8869 Update -K PIC clause from sysv4.h.
8870 (ASM_SPEC_COMMON): Use ENDIAN_SELECT.
8871 (LINK_OS_LINUX_EMUL32, LINK_OS_LINUX_EMUL64): Likewise.
8873 2013-05-06 Alan Modra <amodra@gmail.com>
8875 * config/rs6000/rs6000.md (bswapdi 2nd splitter): Don't swap words
8876 twice for little-endian.
8877 (ashrdi3_no_power, ashrdi3): Support little-endian.
8879 2013-05-06 Oleg Endo <olegendo@gcc.gnu.org>
8882 * config/sh/sh.c (sh_rtx_costs): Handle SMIN and SMAX cases.
8883 * config/sh/sh.md (*clips, uminsi3, *clipu, clipu_one): New insns and
8885 * config/sh/iterators.md (SMIN_SMAX): New code iterator.
8886 * config/sh/predicates.md (arith_reg_or_0_or_1_operand,
8887 clips_min_const_int, clips_max_const_int, clipu_max_const_int):
8890 2013-05-05 Steven Bosscher <steven@gcc.gnu.org>
8891 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
8893 * config.gcc (hppa*-*-*): Remove MASK_BIG_SWITCH from CPU default.
8894 * config/pa/pa.opt: Make mbig-switch a no-op.
8895 * config/pa/pa.h (TARGET_DEFAULT): Remove MASK_BIG_SWITCH.
8896 (CASE_VECTOR_MODE): Always return SImode.
8897 (ASM_OUTPUT_ADDR_VEC_ELT, ASM_OUTPUT_ADDR_DIFF_ELT): Remove code
8898 for the !TARGET_BIG_SWITCH case.
8899 * config/pa/pa-linux.h: Likewise.
8900 * config/pa/pa-openbsd.h: Likewise.
8901 * config/pa/pa-hpux.h: Define TARGET_DEFAULT to 0.
8902 * config/pa/pa.md (short_jump): Remove define_insn.
8903 (casesi): Remove code for the !TARGET_BIG_SWITCH case.
8904 (casesi0): Remove define_insn.
8905 (type): Remove btable_branch.
8906 (pa_combine_type): Likewise.
8907 (in_nullified_branch_delay): Likewise.
8908 (in_call_delay): Likewise.
8909 (define_delay): Likewise.
8910 (define_insn_reservation "Z3"): Likewise.
8911 (define_insn_reservation "Z4"): Likewise.
8912 * config/pa/pa.c (pa_reorg): Remove code for !TARGET_BIG_SWITCH.
8913 (pa_adjust_insn_length): Remove adjustment for btable branches.
8914 * doc/invoke.texi (HPPA Options): Delete documentation for mbig-switch
8917 2013-05-05 Uros Bizjak <ubizjak@gmail.com>
8919 * config/i386/sse.md (*vec_extract<ssevecmodelower>_0): Merge
8920 from sse2_stored and *sse2_storeq_rex64 using SWI48 mode iterator.
8921 Add m->r,x alternatives.
8922 (*vec_extract<ssevecmodelower>_0 splitters): Merge V2DI and V4SI
8923 splitters using SWI48x mode iterator.
8924 (*vec_extract_v2di_0_sse): Rename from *sse2_storeq. Disable for
8925 TARGET_64BIT. Add m->x alternative.
8926 (*vec_extractv4si_mem): Rename from *vec_ext_v4si_mem.
8927 Add o->x alternative. Enable for TARGET_SSE.
8928 (sse_storeq): Remove expander.
8929 (*vec_extractv2di_1): Enable for TARGET_SSE. Split alternatives
8930 with memory input operand.
8931 (*vec_extractv2di_1 splitter): New.
8932 (*vec_extractv4sf_mem): Rename from *vec_extract_v4sf_mem.
8933 * config/i386/i386.md (ssevecmodelower): New mode attribute.
8935 2013-05-04 Segher Boessenkool <segher@kernel.crashing.org>
8937 * config/rs6000/rs6000.c (INT_P): Reformat. Delete obsolete comment.
8938 (INT_LOWPART): Delete.
8939 (extract_MB): Adjust.
8940 (extract_ME): Adjust.
8941 (print_operand): Adjust.
8943 2013-05-04 Segher Boessenkool <segher@kernel.crashing.org>
8945 * config/rs6000/predicates.md (reg_or_add_cint_operand,
8946 reg_or_sub_cint_operand): Delete "HOST_BITS_PER_WIDE_INT == 32" case.
8947 (reg_or_logical_cint_operand, easy_fp_constant,
8948 logical_const_operand): Delete "CONST_DOUBLE" case.
8949 * config/rs6000/rs6000.c (num_insns_constant_wide): Delete
8950 "HOST_BITS_PER_WIDE_INT == 64" test.
8951 (num_insns_constant): Ditto. Delete CONST_DOUBLE DImode/VOIDmode case.
8952 (build_mask64_2_operands): Delete "HOST_BITS_PER_WIDE_INT >= 64" test.
8953 (rs6000_emit_set_const): Delete CONST_DOUBLE case.
8954 (rs6000_emit_set_long_const): Delete "HOST_BITS_PER_WIDE_INT >= 64"
8956 (includes_rldic_lshift_p, includes_rldicr_lshift_p): Delete
8957 CONST_DOUBLE DImode/VOIDmode case.
8958 (INT_P, INT_LOWPART): Delete CONST_DOUBLE case.
8959 (print_operand): Delete "HOST_BITS_PER_WIDE_INT == 32" case. Delete
8960 CONST_DOUBLE VOIDmode case.
8961 (output_toc): Delete "HOST_BITS_PER_WIDE_INT == 32" case.
8962 (rs6000_rtx_costs): Delete CONST_DOUBLE DImode/VOIDmode case.
8963 * config/rs6000/rs6000.md (iordi3, xordi3, splitter for these):
8964 Delete CONST_DOUBLE case.
8965 (splitters for mov FMOVE64 const_double): Delete
8966 "HOST_BITS_PER_WIDE_INT == 32" case. Delete
8967 "HOST_BITS_PER_WIDE_INT >= 64" test.
8968 (splitter for mov DI const_int): Delete "HOST_BITS_PER_WIDE_INT == 32"
8970 (mov DI const_double): Delete.
8972 2013-05-04 Jakub Jelinek <jakub@redhat.com>
8974 * combine.c (combine_simplify_rtx) <case SUBREG>: If nonzero_bits
8975 on op shows all bits zero in mode of a lowpart subreg, return zero.
8977 2013-05-03 Michael Meissner <meissner@linux.vnet.ibm.com>
8980 * config/rs6000/rs6000.h (HARD_REGNO_CALLER_SAVE_MODE): Use DFmode
8981 to save TFmode registers and DImode to save TImode registers for
8982 caller save operations.
8983 (HARD_REGNO_CALL_PART_CLOBBERED): TFmode and TDmode do not need to
8984 mark being partially clobbered since they only use the first
8987 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): TFmode
8988 and TDmode only use the upper 64-bits of each VSX register.
8990 2013-05-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8992 * gimple-ssa-strength-reduction.c (slsr_process_phi): Disable.
8993 (find_candidates_in_block): Disable slsr_process_phi.
8995 2013-05-03 Guozhi Wei <carrot@google.com>
8997 * coverage.c (coverage_obj_init): Move the construction of gcov
8999 (build_init_ctor): ... here.
9001 2013-05-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9003 * gimple-ssa-strength-reduction.c (cand_kind): Add CAND_PHI.
9004 (slsr_cand_d): Redefine def_phi.
9005 (stride_status, phi_adjust_status, count_phis_status): New enums.
9006 (find_phi_def): New.
9007 (find_basis_for_base_expr): New.
9008 (find_basis_for_candidate): Handle hidden bases.
9009 (alloc_cand_and_find_basis): Handle phi candidates.
9010 (slsr_process_phi): New.
9011 (create_mul_ssa_cand): Exclude phi base candidates; use integer_onep.
9012 (create_mul_imm_cand): Likewise.
9013 (create_add_ssa_cand): Exclude phi base candidates.
9014 (create_add_imm_cand): Likewise.
9015 (slsr_process_cast): Likewise.
9016 (slsr_process_copy): Likewise.
9017 (find_candidates_in_block): Handle phi candidates.
9018 (dump_candidate): Likewise.
9019 (unconditional_cands): Delete.
9020 (unconditional_cands_with_known_stride_p): Delete.
9021 (phi_dependent_cand_p): New.
9022 (cand_increment): Handle phi-dependent candidates.
9023 (replace_dependent): Delete.
9024 (replace_mult_candidate): New.
9025 (replace_unconditional_candidate): New.
9026 (incr_vec_index): Move to avoid forward reference.
9027 (create_add_on_incoming_edge): New.
9028 (create_phi_basis): New.
9029 (replace_dependents): Delete.
9030 (replace_conditional_candidate): New.
9031 (phi_add_costs): New.
9032 (replace_uncond_cands_and_profitable_phis): New.
9033 (record_increment): Handle phi adjustments.
9034 (record_phi_increments): New.
9035 (record_increments): Handle phi adjustments.
9036 (phi_incr_cost): New.
9037 (lowest_cost_path): Handle phis.
9038 (total_savings): Likewise.
9039 (analyze_increments): Likewise.
9040 (ncd_with_phi): New.
9041 (ncd_of_cand_and_phis): New.
9042 (nearest_common_dominator_for_cands): Handle phi increments.
9043 (all_phi_incrs_profitable): New.
9044 (replace_profitable_candidates): Handle phi-dependent candidates.
9045 (analyze_candidates_and_replace): Likewise.
9047 2013-05-03 Teresa Johnson <tejohnson@google.com>
9050 * sched-rgn.c (compute_dom_prob_ps): Ensure accumulated probabilities
9051 do not exceed REG_BR_PROB_BASE.
9053 2013-05-03 Jeff Law <law@redhat.com>
9055 PR tree-optimization/57144
9056 * tree-vrp.c (simplify_cond_using_ranges): Verify the constant
9057 operand of the condition will bit into the new type when eliminating
9058 a cast feeding a condition.
9060 2013-05-03 Jakub Jelinek <jakub@redhat.com>
9062 PR rtl-optimization/57130
9063 * combine.c (make_compound_operation) <case SUBREG>: Pass SET instead
9064 of COMPARE as in_code to the recursive call if needed.
9066 2013-05-03 Uros Bizjak <ubizjak@gmail.com>
9068 * config/i386/i386.md (isa): Add x64_sse4_noavx and x64_avx members.
9069 (enabled): Handle new members.
9070 * config/i386/sse.md (*vec_concatv2si): Merge from
9071 *vec_concatv2si_sse2 and vec_concatv2si_sse.
9072 (vec_concatv2di): Merge with *vec_concatv2di_rex64.
9074 2013-05-03 Joern Rennecke <joern.rennecke@embecosm.com>
9076 PR tree-optimization/57027
9077 * tree-ssa-math-opts.c (convert_mult_to_fma): When checking
9078 for fnms opportunity, check we got the prerequisite kind
9079 of tree / gimple before using accessor functions.
9081 2013-05-03 Richard Biener <rguenther@suse.de>
9083 * double-int.h (lshift): New overload without precision
9085 (operator *=, operator +=, operator -=): Move ...
9086 * double-int.c (operator *=, operator +=, operator -=): ... here
9087 and implement more efficiently.
9088 (mul_double_with_sign): Remove.
9089 (lshift_double): Adjust to take unsinged shift argument, push
9090 dispatching code to callers.
9091 (mul_double_wide_with_sign): Add early out for callers that
9092 are not interested in high parts or overflow.
9093 (lshift): New function.
9094 (lshift, rshift, alshift, arshift, llshift, lrshift): Add
9096 (lrotate, rrotate): Use logical shifts.
9097 * expr.c (get_inner_reference): Use lshift.
9098 * fixed-value.c (do_fixed_divide): Likewise.
9099 * tree-dfa.c (get_ref_base_and_extent): Likewise.
9100 * tree-ssa-alias.c (indirect_ref_may_alias_decl_p): Likewise.
9101 (indirect_refs_may_alias_p): Likewise.
9102 (stmt_kills_ref_p_1): Likewise.
9104 2013-05-03 Vidya Praveen <vidyapraveen@arm.com>
9106 * config/aarch64/aarch64-simd.md (simd_fabd): Correct the description.
9108 2013-05-03 Vidya Praveen <vidyapraveen@arm.com>
9110 * config/aarch64/aarch64-simd.md (*fabd_scalar<mode>3): Support
9111 scalar form of FABD instruction.
9113 2013-05-02 Vladimir Makarov <vmakarov@redhat.com>
9115 * lra-constraints.c (process_alt_operands): Add checking alt
9116 number to choose the best alternative.
9118 2013-05-02 Richard Biener <rguenther@suse.de>
9120 * tree-eh.c (cleanup_empty_eh_merge_phis): Remove rename_virts
9121 bitmap and its handling.
9122 (pass_cleanup_eh): Set todo_flags_finish to TODO_verify_ssa.
9124 2013-05-02 Richard Biener <rguenther@suse.de>
9127 * tree-inline.c (copy_loops): Properly handle removed loops.
9128 (copy_cfg_body): Mark destination loops for fixup if source
9131 2013-05-02 Greta Yorsh <Greta.Yorsh@arm.com>
9134 * config/arm/arm.c (arm_expand_epilogue): Check really_return before
9135 generating simple_return for naked functions.
9137 2013-05-02 Martin Jambor <mjambor@suse.cz>
9140 * ipa-prop.h (ipa_agg_replacement_value): New flag by_ref.
9141 * ipa-cp.c (ipa_get_indirect_edge_target_1): Also check that by_ref
9143 (find_aggregate_values_for_callers_subset): Fill in the by_ref flag of
9144 ipa_agg_replacement_value structures.
9145 (known_aggs_to_agg_replacement_list): Likewise.
9146 * ipa-prop.c (write_agg_replacement_chain): Stream by_ref flag.
9147 (read_agg_replacement_chain): Likewise.
9148 (ipcp_transform_function): Also check that by_ref flags match.
9150 2013-05-02 Richard Biener <rguenther@suse.de>
9152 * graphds.h (struct graph): Add obstack member.
9153 * graphds.c (new_graph): Initialize obstack and allocate
9155 (add_edge): Allocate edge from the obstack.
9156 (free_graph): Free the obstack instead of all edges and vertices.
9158 2013-05-02 Teresa Johnson <tejohnson@google.com>
9160 * loop-unswitch.c (unswitch_loop): Use helper routines with rounding
9162 * cfg.c (update_bb_profile_for_threading): Ditto.
9163 * tree-inline.c (copy_bb): Ditto.
9164 (copy_edges_for_bb): Ditto.
9165 (initialize_cfun): Ditto.
9166 (copy_cfg_body): Ditto.
9167 (expand_call_inline): Ditto.
9168 * ipa-inline-analysis.c (estimate_edge_size_and_time): Ditto.
9169 (estimate_node_size_and_time): Ditto.
9170 (inline_merge_summary): Ditto.
9171 * cgraphclones.c (cgraph_clone_edge): Ditto.
9172 (cgraph_clone_node): Ditto.
9173 * sched-rgn.c (compute_dom_prob_ps): Ditto.
9174 (compute_trg_info): Ditto.
9176 2013-05-02 Ian Bolton <ian.bolton@arm.com>
9178 * config/aarch64/aarch64.md (movsi_aarch64): Only allow to/from
9179 S reg when fp attribute set.
9180 (movdi_aarch64): Only allow to/from D reg when fp attribute set.
9182 2013-05-02 Ian Bolton <ian.bolton@arm.com>
9184 * config/aarch64/aarch64.md (*and_one_cmpl<mode>3_compare0):
9186 (*and_one_cmplsi3_compare0_uxtw): Likewise.
9187 (*and_one_cmpl_<SHIFT:optab><mode>3_compare0): Likewise.
9188 (*and_one_cmpl_<SHIFT:optab>si3_compare0_uxtw): Likewise.
9190 2013-05-02 Richard Biener <rguenther@suse.de>
9192 * tree-scalar-evolution.c (scev_info_hasher): Remove.
9193 (struct instantiate_cache_entry): New type.
9194 (struct instantiate_cache_entry_hasher): New hashtable descriptor.
9195 (struct instantiate_cache_type): New type.
9196 (set_instantiated_value, get_instantiated_value): Remove.
9197 (get_instantiated_value_entry): New function.
9198 (instantiate_scev_name): Use the new cache and adjust.
9199 (instantiate_scev_poly): Adjust.
9200 (instantiate_scev_binary): Likewise.
9201 (instantiate_array_ref): Likewise.
9202 (instantiate_scev_convert): Likewise.
9203 (instantiate_scev_not): Likewise.
9204 (instantiate_scev_3): Likewise.
9205 (instantiate_scev_2): Likewise.
9206 (instantiate_scev_r): Likewise.
9207 (instantiate_scev): Likewise.
9208 (resolve_mixers): Likewise.
9210 2013-05-01 Vladimir Makarov <vmakarov@redhat.com>
9213 * lra-constraints.c (best_small_class_operands_num): Remove.
9214 (process_alt_operands): Remove small_class_operands_num. Take
9215 small classes operands into losers and only if the operand is not
9216 matched. Modify debugging output.
9217 (curr_insn_transform): Remove best_small_class_operands_num.
9220 2013-05-01 James Greenhalgh <james.greenhalgh@arm.com>
9222 * config/aarch64/aarch64-builtins.c
9223 (aarch64_gimple_fold_builtin.c): Fold more modes for reduc_splus_.
9224 * config/aarch64/aarch64-simd-builtins.def
9225 (reduc_splus_): Add new modes.
9226 (reduc_uplus_): New.
9227 * config/aarch64/aarch64-simd.md (aarch64_addvv4sf): Remove.
9228 (reduc_uplus_v4sf): Likewise.
9229 (reduc_splus_v4sf): Likewise.
9230 (aarch64_addv<mode>): Likewise.
9231 (reduc_uplus_<mode>): Likewise.
9232 (reduc_splus_<mode>): Likewise.
9233 (aarch64_addvv2di): Likewise.
9234 (reduc_uplus_v2di): Likewise.
9235 (reduc_splus_v2di): Likewise.
9236 (aarch64_addvv2si): Likewise.
9237 (reduc_uplus_v2si): Likewise.
9238 (reduc_splus_v2si): Likewise.
9239 (reduc_<sur>plus_<mode>): New.
9240 (reduc_<sur>plus_v2di): Likewise.
9241 (reduc_<sur>plus_v2si): Likewise.
9242 (reduc_<sur>plus_v4sf): Likewise.
9243 (aarch64_addpv4sf): Likewise.
9244 * config/aarch64/arm_neon.h
9245 (vaddv<q>_<s,u,f><8, 16, 32, 64): Rewrite using builtins.
9246 * config/aarch64/iterators.md (unspec): Remove UNSPEC_ADDV,
9247 add UNSPEC_SADDV, UNSPEC_UADDV.
9249 (sur): Add UNSPEC_SADDV, UNSPEC_UADDV.
9251 2013-05-01 James Greenhalgh <james.greenhalgh@arm.com>
9253 * config/aarch64/arm_neon.h
9254 (v<max,min><nm><q><v>_<sfu><8, 16, 32, 64>): Rewrite using builtins.
9256 2013-05-01 James Greenhalgh <james.greenhalgh@arm.com>
9258 * config/aarch64/aarch64-builtins
9259 (aarch64_gimple_fold_builtin): Fold reduc_<su><maxmin>_ builtins.
9261 2013-05-01 James Greenhalgh <james.greenhalgh@arm.com>
9263 * config/aarch64/aarch64-simd-builtins.def
9265 (reduc_smin_): Likewise.
9266 (reduc_umax_): Likewise.
9267 (reduc_umin_): Likewise.
9268 (reduc_smax_nan_): Likewise.
9269 (reduc_smin_nan_): Likewise.
9272 (smax): Update for V2SF, V4SF and V2DF modes.
9275 (smin_nan): Likewise.
9276 * config/aarch64/aarch64-simd.md (<maxmin><mode>3): Rename to...
9277 (<su><maxmin><mode>3): ...This, refactor.
9278 (s<maxmin><mode>3): New.
9279 (<maxmin_uns><mode>3): Likewise.
9280 (reduc_<maxmin_uns>_<mode>): Refactor.
9281 (reduc_<maxmin_uns>_v4sf): Likewise.
9282 (reduc_<maxmin_uns>_v2si): Likewise.
9283 (aarch64_<fmaxmin><mode>: Remove.
9284 * config/aarch64/arm_neon.h (vmax<q>_f<32,64>): Rewrite to use
9286 (vmin<q>_f<32,64>): Likewise.
9287 * config/iterators.md (unspec): Add UNSPEC_FMAXNMV, UNSPEC_FMINNMV.
9289 (su): Add mappings for smax, smin, umax, umin.
9291 (FMAXMINV): Add UNSPEC_FMAXNMV, UNSPEC_FMINNMV.
9292 (FMAXMIN): Rename as...
9293 (FMAXMIN_UNS): ...This.
9295 (fmaxminv): Likewise.
9296 (fmaxmin): Likewise.
9298 (maxmin_uns_op): Likewise.
9300 2013-05-01 James Greenhalgh <james.greenhalgh@arm.com>
9302 * config/aarch64/arm_neon.h
9303 (vac<ge, gt><sd>_f<32, 64>): Rename to...
9304 (vca<ge, gt><sd>_f<32, 64>): ...this, reimpliment in C.
9305 (vca<ge, gt, lt, le><q>_f<32, 64>): Reimpliment in C.
9307 2013-05-01 James Greenhalgh <james.greenhalgh@arm.com>
9309 * config/aarch64/aarch64-simd.md (*aarch64_fac<optab><mode>): New.
9310 * config/aarch64/iterators.md (FAC_COMPARISONS): New.
9312 2013-05-01 James Greenhalgh <james.greenhalgh@arm.com>
9314 * config/aarch64/aarch64-simd.md
9315 (vcond<mode>_internal): Handle special cases for constant masks.
9316 (vcond<mode><mode>): Allow nonmemory_operands for outcome vectors.
9317 (vcondu<mode><mode>): Likewise.
9318 (vcond<v_cmp_result><mode>): New.
9320 2013-05-01 James Greenhalgh <james.greenhalgh@arm.com>
9322 * config/aarch64/aarch64-builtins.c (BUILTIN_VALLDI): Define.
9323 (aarch64_fold_builtin): Add folding for cm<eq,ge,gt,tst>.
9324 * config/aarch64/aarch64-simd-builtins.def
9325 (cmeq): Update to BUILTIN_VALLDI.
9330 * config/aarch64/arm_neon.h
9331 (vc<eq, lt, le, gt, ge, tst><z><qsd>_<fpsu><8,16,32,64>): Remap
9332 to builtins or C as appropriate.
9334 2013-05-01 James Greenhalgh <james.greenhalgh@arm.com>
9336 * config/aarch64/aarch64-simd-builtins.def (cmhs): Rename to...
9338 (cmhi): Rename to...
9340 * config/aarch64/aarch64-simd.md
9341 (simd_mode): Add SF.
9342 (aarch64_vcond_internal): Use new names for unsigned comparison insns.
9343 (aarch64_cm<optab><mode>): Rewrite to not use UNSPECs.
9344 * config/aarch64/aarch64.md (*cstore<mode>_neg): Rename to...
9345 (cstore<mode>_neg): ...This.
9346 * config/aarch64/iterators.md
9348 (unspec): Remove UNSPEC_CM<EQ, LE, LT, GE, GT, HS, HI, TST>.
9350 (UCOMPARISONS): Likewise.
9351 (optab): Add missing comparisons.
9359 (V_cmp_result): Add DF, SF modes.
9360 (v_cmp_result): Likewise.
9363 * config/aarch64/predicates.md (aarch64_reg_or_fp_zero): New.
9365 2013-05-01 Greta Yorsh <Greta.Yorsh@arm.com>
9367 * config/arm/thumb2.md (thumb2_smaxsi3,thumb2_sminsi3): Convert
9368 define_insn to define_insn_and_split.
9369 (thumb32_umaxsi3,thumb2_uminsi3): Likewise.
9370 (thumb2_negdi2,thumb2_abssi2,thumb2_neg_abssi2): Likewise.
9371 (thumb2_mov_scc,thumb2_mov_negscc,thumb2_mov_notscc): Likewise.
9372 (thumb2_movsicc_insn,thumb2_and_scc,thumb2_ior_scc): Likewise.
9373 (thumb2_negscc): Likewise.
9375 2013-04-30 Greta Yorsh <Greta.Yorsh@arm.com>
9377 * config/arm/thumb2.md (thumb2_incscc, thumb2_decscc): Delete.
9379 2013-04-30 Greta Yorsh <Greta.Yorsh@arm.com>
9381 * config/arm/thumb2.md: Remove trailing whitespaces.
9383 2013-04-30 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
9385 * explow.c (plus_constant): Pass "mode" to immed_double_int_const.
9386 Use gen_int_mode rather than GEN_INT.
9388 2013-04-30 H.J. Lu <hongjiu.lu@intel.com>
9390 * value-prof.c (stream_in_histogram_value): Remove the strayed
9393 2013-04-30 Richard Biener <rguenther@suse.de>
9396 * cfghooks.c (split_edge): Properly check for the loop latch edge.
9398 2013-04-30 Richard Biener <rguenther@suse.de>
9401 * tree-eh.c (sink_clobbers): Preserve virtual SSA form.
9403 2013-04-30 Andrey Belevantsev <abel@ispras.ru>
9405 PR rtl-optimization/56957
9406 PR rtl-optimization/57105
9407 * sel-sched.c (move_op_orig_expr_found): Remove insn_emitted
9408 variable. Use just INSN_UID for determining whether an insn
9409 should be only disconnected from the insn stream.
9410 * sel-sched-ir.h (EXPR_WAS_CHANGED): Remove.
9412 2013-04-30 Jakub Jelinek <jakub@redhat.com>
9414 PR tree-optimization/57104
9415 * tsan.c (instrument_expr): Don't instrument accesses to
9416 DECL_HARD_REGISTER VAR_DECLs.
9418 2013-04-30 Richard Biener <rguenther@suse.de>
9420 * function.h (loops_for_fn): New inline function.
9421 (set_loops_for_fn): Likewise.
9422 * cfgloop.h (place_new_loop): Add struct function parameter.
9423 (get_loop): Likewise.
9424 (get_loops): Likewise.
9425 (number_of_loops): Likewise.
9427 (fel_init): Likewise.
9428 * cfg.c (get_loop_copy): Adjust.
9429 * cfgloop.c (flow_loops_dump): Likewise.
9430 (record_loop_exits): Likewise.
9431 (verify_loop_structure): Likewise.
9432 * cfgloopanal.c (mark_irreducible_loops): Likewise.
9433 (estimate_reg_pressure_cost): Likewise.
9434 (mark_loop_exit_edges): Likewise.
9435 * cfgloopmanip.c (place_new_loop): Likewise.
9436 (add_loop): Likewise.
9437 (duplicate_loop): Likewise.
9438 * graph.c (draw_cfg_nodes): Likewise.
9439 * graphite-clast-to-gimple.c (translate_clast_user): Likewise.
9440 * graphite-sese-to-poly.c (build_scop_scattering): Likewise.
9441 (extract_affine_chrec): Likewise.
9442 (build_scop_iteration_domain): Likewise.
9443 * graphite.c (graphite_initialize): Likewise.
9444 * ira-build.c (create_loop_tree_nodes): Likewise.
9445 (more_one_region_p): Likewise.
9446 (rebuild_regno_allocno_maps): Likewise.
9447 (mark_loops_for_removal): Likewise.
9448 (mark_all_loops_for_removal): Likewise.
9449 (remove_unnecessary_regions): Likewise.
9450 (ira_build): Likewise.
9451 * ira-emit.c (setup_entered_from_non_parent_p): Likewise.
9452 * loop-init.c (fix_loop_structure): Likewise.
9453 (gate_rtl_move_loop_invariants): Likewise.
9454 (gate_rtl_unswitch): Likewise.
9455 (gate_rtl_unroll_and_peel_loops): Likewise.
9456 (rtl_doloop): Likewise.
9457 * lto-streamer-in.c (input_cfg): Likewise.
9458 * lto-streamer-out.c (output_cfg): Likewise.
9459 * modulo-sched.c (sms_schedule): Likewise.
9460 * predict.c (tree_estimate_probability): Likewise.
9461 (tree_estimate_probability_driver): Likewise.
9462 (estimate_loops): Likewise.
9463 * tree-cfg.c (fixup_loop_arrays_after_move): Likewise.
9464 (move_sese_region_to_fn): Likewise.
9465 (debug_loop_num): Likewise.
9466 * tree-chrec.c (chrec_evaluate): Likewise.
9467 (hide_evolution_in_other_loops_than_loop): Likewise.
9468 (chrec_component_in_loop_num): Likewise.
9469 (reset_evolution_in_loop): Likewise.
9470 (evolution_function_is_invariant_rec_p): Likewise.
9471 * tree-if-conv.c (main_tree_if_conversion): Likewise.
9472 * tree-inline.c (copy_loops): Likewise.
9473 (copy_cfg_body): Likewise.
9474 (tree_function_versioning): Likewise.
9475 * tree-loop-distribution.c (rdg_flag_loop_exits): Likewise.
9476 * tree-scalar-evolution.c (chrec_contains_symbols_defined_in_loop):
9478 (add_to_evolution_1): Likewise.
9479 (scev_const_prop): Likewise.
9480 * tree-scalar-evolution.h (get_chrec_loop): Likewise.
9481 * tree-ssa-loop-ch.c (copy_loop_headers): Likewise.
9482 * tree-ssa-loop-im.c (analyze_memory_references): Likewise.
9483 (tree_ssa_lim_initialize): Likewise.
9484 * tree-ssa-loop-manip.c (rewrite_into_loop_closed_ssa): Likewise.
9485 (verify_loop_closed_ssa): Likewise.
9486 * tree-ssa-loop.c (tree_ssa_loop_init): Likewise.
9487 (tree_ssa_loop_im): Likewise.
9488 (tree_ssa_loop_unswitch): Likewise.
9489 (tree_vectorize): Likewise.
9490 (check_data_deps): Likewise.
9491 (tree_ssa_loop_ivcanon): Likewise.
9492 (tree_ssa_loop_bounds): Likewise.
9493 (tree_complete_unroll): Likewise.
9494 (tree_complete_unroll_inner): Likewise.
9495 (tree_parallelize_loops): Likewise.
9496 (tree_ssa_loop_prefetch): Likewise.
9497 (tree_ssa_loop_ivopts): Likewise.
9498 * tree-ssa.c (execute_update_addresses_taken): Liekwise.
9499 * tree-vectorizer.c (vectorize_loops): Likewise.
9501 2013-04-29 Mike Frysinger <vapier@gentoo.org>
9503 * config/arm/bpabi.h (EABI_LINK_SPEC): Define.
9504 (BPABI_LINK_SPEC): Use new EABI_LINK_SPEC.
9505 * config/arm/linux-eabi.h (LINK_SPEC): Replace BE8_LINK_SPEC
9506 with EABI_LINK_SPEC.
9508 2013-04-29 Uros Bizjak <ubizjak@gmail.com>
9511 * config/i386/i386.md (*zero_extendsidi2): Add "!" to m->?*y
9514 2013-04-29 Vladimir Makarov <vmakarov@redhat.com>
9517 * lra-constraints.c (process_alt_operands): Discourage a bit more
9518 using memory for pseudos. Print cost dump for alternatives.
9519 Modify cost values for conflicts with early clobbers.
9520 (curr_insn_transform): Spill pseudos reassigned to NO_REGS.
9522 2013-04-29 Uros Bizjak <ubizjak@gmail.com>
9525 * config/i386/i386.c (ix86_expand_vec_perm): Validize constant memory.
9527 2013-04-29 Ian Bolton <ian.bolton@arm.com>
9529 * config/aarch64/aarch64.md (movsi_aarch64): Support LDR/STR
9531 (movdi_aarch64): Support LDR/STR from/to D register.
9533 2013-04-29 Ian Bolton <ian.bolton@arm.com>
9535 * common/config/aarch64/aarch64-common.c: Enable REE pass at O2
9536 or higher by default.
9538 2013-04-29 Richard Biener <rguenther@suse.de>
9541 * tree-inline.c (copy_edges_for_bb): Still split the bbs,
9542 even if not adding abnormal edges for calls that can make
9545 2013-04-29 Richard Biener <rguenther@suse.de>
9548 * tree-cfg.c (move_stmt_op): Fix condition under which to update
9550 (move_stmt_r): Remove redundant checking.
9552 2013-04-29 Teresa Johnson <tejohnson@google.com>
9555 * basic-block.h (apply_scale): New function.
9556 (apply_probability): Use apply_scale.
9557 * gimple-streamer-in.c (input_bb): Ditto.
9558 * lto-streamer-in.c (input_cfg): Ditto.
9559 * lto-cgraph.c (merge_profile_summaries): Ditto.
9560 * tree-optimize.c (execute_fixup_cfg): Ditto.
9561 * tree-inline.c (copy_bb): Update comment to use apply_scale.
9562 (copy_edges_for_bb): Ditto.
9563 (copy_cfg_body): Ditto.
9565 2013-04-29 Tom de Vries <tom@codesourcery.com>
9567 * tree-ssa-tail-merge.c (find_same_succ_bb): Skip loop latch bbs.
9568 (replace_block_by): Don't set LOOPS_NEED_FIXUP.
9569 (tail_merge_optimize): Handle current_loops == NULL.
9571 2013-04-26 Jeff Law <law@redhat.com>
9573 * tree-vrp.c (range_fits_type_p): Move to earlier point in file.
9574 (simplify_cond_using_ranges): Generalize code to simplify
9575 COND_EXPRs where one argument is a constant and the other
9576 is an SSA_NAME created by an integral type conversion.
9578 2013-04-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9580 * config/arm/arm.md (store_minmaxsi): Use only when
9581 optimize_insn_for_size_p.
9583 2013-04-29 Christian Bruel <christian.bruel@st.com>
9586 * sh.md (tstsi_t_zero_extract_eq): Set mode for operand 0.
9588 2013-04-29 Richard Biener <rguenther@suse.de>
9591 * omp-low.c (expand_omp_taskreg): If the parent function had a broken
9592 loop tree make sure to schedule a fixup for the child as well.
9593 (expand_omp_for_generic): Properly add loops.
9594 (expand_omp_for_static_nochunk): Likewise.
9595 (expand_omp_for_static_chunk): Likewise.
9596 (expand_omp_for): For the degenerate case fixup loops.
9597 (expand_omp_sections): Fix default bb placement in loops.
9598 (expand_omp_atomic_pipeline): Properly add loops.
9600 2013-04-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9602 * predict.c: Fix typo in comment above #define PROB_VERY_UNLIKELY.
9604 2013-04-29 Tom de Vries <tom@codesourcery.com>
9606 * tree-ssa-tail-merge.c: Update header comment.
9608 2013-04-29 James Greenhalgh <james.greenhalgh@arm.com>
9610 * config/aarch64/arm_neon.h
9611 (vcvt<sd>_f<32,64>_s<32,64>): Rewrite in C.
9612 (vcvt<q>_f<32,64>_s<32,64>): Rewrite using builtins.
9613 (vcvt_<high_>_f<32,64>_f<32,64>): Likewise.
9614 (vcvt<qsd>_<su><32,64>_f<32,64>): Likewise.
9615 (vcvta<qsd>_<su><32,64>_f<32,64>): Likewise.
9616 (vcvtm<qsd>_<su><32,64>_f<32,64>): Likewise.
9617 (vcvtn<qsd>_<su><32,64>_f<32,64>): Likewise.
9618 (vcvtp<qsd>_<su><32,64>_f<32,64>): Likewise.
9620 2013-04-29 James Greenhalgh <james.greenhalgh@arm.com>
9622 * config/aarch64/aarch64-simd.md
9623 (<optab><VDQF:mode><fcvt_target>2): New, maps to fix, fixuns.
9624 (<fix_trunc_optab><VDQF:mode><fcvt_target>2): New, maps to
9625 fix_trunc, fixuns_trunc.
9626 (ftrunc<VDQF:mode>2): New.
9627 * config/aarch64/iterators.md (optab): Add fix, fixuns.
9628 (fix_trunc_optab): New.
9630 2013-04-29 James Greenhalgh <james.greenhalgh@arm.com>
9632 * config/aarch64/aarch64-builtins.c
9633 (aarch64_builtin_vectorized_function): Vectorize over ifloorf,
9634 iceilf, lround, iroundf.
9636 2013-04-29 Uros Bizjak <ubizjak@gmail.com>
9639 * config/i386/i386.h (enum ix86_tune_indices)
9640 <X86_TUNE_INTER_UNIT_MOVES_TO_VEC, X86_TUNE_INTER_UNIT_MOVES_FROM_VEC>:
9641 New, split from X86_TUNE_INTER_UNIT_MOVES.
9642 <X86_TUNE_INTER_UNIT_MOVES>: Remove.
9643 (TARGET_INTER_UNIT_MOVES_TO_VEC): New define.
9644 (TARGET_INTER_UNIT_MOVES_FROM_VEC): Ditto.
9645 (TARGET_INTER_UNIT_MOVES): Remove.
9646 * config/i386/i386.c (initial_ix86_tune_features): Update.
9647 Disable X86_TUNE_INTER_UNIT_MOVES_FROM_VEC for m_ATHLON_K8 only.
9648 (ix86_expand_convert_uns_didf_sse): Use
9649 TARGET_INTER_UNIT_MOVES_TO_VEC instead of TARGET_INTER_UNIT_MOVES.
9650 (ix86_expand_vector_init_one_nonzero): Ditto.
9651 (ix86_expand_vector_init_interleave): Ditto.
9652 (inline_secondary_memory_needed): Return true for moves from SSE class
9653 registers for !TARGET_INTER_UNIT_MOVES_FROM_VEC targets and for moves
9654 to SSE class registers for !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
9655 * config/i386/constraints.md (Yi, Ym): Depend on
9656 TARGET_INTER_UNIT_MOVES_TO_VEC.
9657 (Yj, Yn): New constraints.
9658 * config/i386/i386.md (*movdi_internal): Change constraints of
9659 operand 1 from Yi to Yj and from Ym to Yn.
9660 (*movsi_internal): Ditto.
9661 (*movdf_internal): Ditto.
9662 (*movsf_internal): Ditto.
9663 (*float<SWI48x:mode><X87MODEF:mode>2_1): Use
9664 TARGET_INTER_UNIT_MOVES_TO_VEC instead of TARGET_INTER_UNIT_MOVES.
9665 (*float<SWI48x:mode><X87MODEF:mode>2_1 splitters): Ditto.
9666 (floatdi<X87MODEF:mode>2_i387_with_xmm): Ditto.
9667 (floatdi<X87MODEF:mode>2_i387_with_xmm splitters): Ditto.
9668 * config/i386/sse.md (movdi_to_sse): Ditto.
9669 (sse2_stored): Change constraint of operand 1 from Yi to Yj.
9670 Use TARGET_INTER_UNIT_MOVES_FROM_VEC instead of
9671 TARGET_INTER_UNIT_MOVES.
9672 (sse_storeq_rex64): Change constraint of operand 1 from Yi to Yj.
9673 (sse_storeq_rex64 splitter): Use TARGET_INTER_UNIT_MOVES_FROM_VEC
9674 instead of TARGET_INTER_UNIT_MOVES.
9675 * config/i386/mmx.md (*mov<mode>_internal): Change constraint of
9676 operand 1 from Yi to Yj and from Ym to Yn.
9678 2013-04-29 James Greenhalgh <james.greenhalgh@arm.com>
9680 * config/aarch64/aarch64-simd-builtins.def (vec_unpacks_hi_): New.
9681 (float_truncate_hi_): Likewise.
9682 (float_extend_lo_): Likewise.
9683 (float_truncate_lo_): Likewise.
9684 * config/aarch64/aarch64-simd.md (vec_unpacks_lo_v4sf): New.
9685 (aarch64_float_extend_lo_v2df): Likewise.
9686 (vec_unpacks_hi_v4sf): Likewise.
9687 (aarch64_float_truncate_lo_v2sf): Likewise.
9688 (aarch64_float_truncate_hi_v4sf): Likewise.
9689 (vec_pack_trunc_v2df): Likewise.
9690 (vec_pack_trunc_df): Likewise.
9692 2013-04-29 James Greenhalgh <james.greenhalgh@arm.com>
9694 * config/aarch64/aarch64-builtins.c
9695 (aarch64_fold_builtin): Fold float conversions.
9696 * config/aarch64/aarch64-simd-builtins.def
9697 (floatv2si, floatv4si, floatv2di): New.
9698 (floatunsv2si, floatunsv4si, floatunsv2di): Likewise.
9699 * config/aarch64/aarch64-simd.md
9700 (<optab><fcvt_target><VDQF:mode>2): New, expands to float and floatuns.
9701 * config/aarch64/iterators.md (FLOATUORS): New.
9702 (optab): Add float, floatuns.
9703 (su_optab): Likewise.
9705 2013-04-29 James Greenhalgh <james.greenhalgh@arm.com>
9707 * config/aarch64/aarch64-builtins.c
9708 (aarch64_builtin_vectorized_function): Use new names for
9710 * config/aarch64/aarch64-simd-builtins.def (fcvtzs): Split as...
9711 (lbtruncv2sf, lbtruncv4sf, lbtruncv2df): ...This.
9712 (fcvtzu): Split as...
9713 (lbtruncuv2sf, lbtruncuv4sf, lbtruncuv2df): ...This.
9714 (fcvtas): Split as...
9715 (lroundv2sf, lroundv4sf, lroundv2df, lroundsf, lrounddf): ...This.
9716 (fcvtau): Split as...
9717 (lrounduv2sf, lrounduv4sf, lrounduv2df, lroundusf, lroundudf): ...This.
9718 (fcvtps): Split as...
9719 (lceilv2sf, lceilv4sf, lceilv2df): ...This.
9720 (fcvtpu): Split as...
9721 (lceiluv2sf, lceiluv4sf, lceiluv2df, lceilusf, lceiludf): ...This.
9722 (fcvtms): Split as...
9723 (lfloorv2sf, lfloorv4sf, lfloorv2df): ...This.
9724 (fcvtmu): Split as...
9725 (lflooruv2sf, lflooruv4sf, lflooruv2df, lfloorusf, lfloorudf): ...This.
9726 (lfrintnv2sf, lfrintnv4sf, lfrintnv2df, lfrintnsf, lfrintndf): New.
9727 (lfrintnuv2sf, lfrintnuv4sf, lfrintnuv2df): Likewise.
9728 (lfrintnusf, lfrintnudf): Likewise.
9729 * config/aarch64/aarch64-simd.md
9730 (l<fcvt_pattern><su_optab><fcvt_target><VDQF:mode>2): Convert to
9732 (aarch64_fcvt<frint_suffix><su><mode>): Remove.
9733 * config/aarch64/iterators.md (FCVT): Include UNSPEC_FRINTN.
9734 (fcvt_pattern): Likewise.
9736 2013-04-29 James Greenhalgh <james.greenhalgh@arm.com>
9738 * config/aarch64/aarch64-simd.md
9739 (l<fcvt_pattern><su_optab><fcvt_target><VDQF:mode>2): Rename to...
9740 (l<fcvt_pattern><su_optab><VDQF:mode><fcvt_target>2): ... This.
9742 2013-04-29 James Greenhalgh <james.greenhalgh@arm.com>
9744 * config/aarch64/arm_neon.h (vrndq<a,m,n,p>_f<32, 64>): Rename to...
9745 (vrnd<a,m,n,p>q_f<32, 64>): ...This, implement using builtin.
9746 (vrnd<a,m,n,p>_f32): Implement using builtins.
9747 (vrnd<i,x><q>_f<32, 64>): New.
9749 2013-04-29 James Greenhalgh <james.greenhalgh@arm.com>
9751 * config/aarch64/aarch64-builtins.c
9752 (aarch64_builtin_vectorized_function): Fold to standard pattern names.
9753 * config/aarch64/aarch64-simd-builtins.def (frintn): New.
9754 (frintz): Rename to...
9756 (frintp): Rename to...
9758 (frintm): Rename to...
9760 (frinti): Rename to...
9761 (nearbyint): ...this.
9762 (frintx): Rename to...
9764 (frinta): Rename to...
9766 * config/aarch64/aarch64-simd.md
9767 (aarch64_frint<frint_suffix><mode>): Delete.
9768 (<frint_pattern><mode>2): Convert to insn.
9769 * config/aarch64/aarch64.md (unspec): Add UNSPEC_FRINTN.
9770 * config/aarch64/iterators.md (FRINT): Add UNSPEC_FRINTN.
9771 (frint_pattern): Likewise.
9772 (frint_suffix): Likewise.
9774 2013-04-29 Richard Biener <rguenther@suse.de>
9776 PR tree-optimization/57081
9777 * loop-init.c: Include tree-flow.h.
9778 (loop_optimizer_finalize): Free number of iteration estimates.
9779 * Makefile.in (loop-init.o): Add $(TREE_FLOW_H) dependency.
9781 2013-04-29 Jakub Jelinek <jakub@redhat.com>
9783 PR tree-optimization/57083
9784 * tree-vrp.c (extract_range_from_binary_expr_1): For LSHIFT_EXPR with
9785 non-singleton shift count range, zero extend low_bound for uns case.
9787 * config/i386/predicates.md (general_vector_operand): New predicate.
9788 * config/i386/i386.c (const_vector_equal_evenodd_p): New function.
9789 (ix86_expand_mul_widen_evenodd): Force op1 resp. op2 into register
9790 if they aren't nonimmediate operands. If their original values
9791 satisfy const_vector_equal_evenodd_p, don't shift them.
9792 * config/i386/sse.md (mul<mode>3): Use general_vector_operand
9793 predicates. For the SSE4.1 case force operands[{1,2}] into registers
9794 if not nonimmediate_operand.
9795 (vec_widen_smult_even_v4si): Use nonimmediate_operand predicates
9796 instead of register_operand.
9797 (vec_widen_<s>mult_odd_<mode>): Use general_vector_operand predicates.
9799 2013-04-28 Eric Botcazou <ebotcazou@adacore.com>
9801 * stor-layout.c (finalize_size_functions): Allocate a structure and
9802 reset cfun before dumping the functions.
9804 2013-04-27 Jakub Jelinek <jakub@redhat.com>
9806 * config/i386/i386.c (ix86_expand_call): Make cregs_size unsigned.
9809 * config/i386/i386.c (ix86_expand_mul_widen_evenodd): Don't
9810 use xop_pmacsdqh if uns_p.
9811 * config/i386/sse.md (xop_rotr<mode>3): Fix up computation of
9812 the immediate rotate count.
9814 2013-04-26 Vladimir Makarov <vmakarov@redhat.com>
9816 * rtl.h (struct rtx_def): Add comment for field jump.
9817 (LRA_SUBREG_P): New macro.
9818 * recog.c (register_operand): Check LRA_SUBREG_P.
9819 * lra.c (lra): Add note at the end of RTL code. Align non-empty
9821 * lra-spills.c (lra_spill): Align stack after spilling pseudos.
9822 (lra_final_code_change): Skip subreg change for operators.
9823 * lra-eliminations.c (eliminate_regs_in_insn): Make return earlier
9824 if there are no operand changes.
9825 * lra-constraints.c (curr_insn_set): New.
9826 (match_reload): Set LRA_SUBREG_P.
9827 (emit_spill_move): Ditto.
9828 (check_and_process_move): Use curr_insn_set. Process only single
9829 set insns. Don't initialize sec_mem_p and change_p.
9830 (simplify_operand_subreg): Use LRA_SUBREG_P.
9831 (reg_in_class_p): New function.
9832 (process_alt_operands): Use it. Use #if HAVE_ATTR_enabled instead
9833 of #ifdef. Add code to remove cycling.
9834 (process_address): Check EXTRA_CONSTRAINT_STR. Process even if
9835 non-null disp. Reload inner instead of disp when base and index
9836 are null. Try to put lo_sum into register.
9837 (EBB_PROBABILITY_CUTOFF): Redefine probability in percents.
9838 (check_and_process_move): Move code for move cost check to
9839 simple_move_p. Remove equiv_substitution.
9840 (simple_move_p): New function.
9841 (curr_insn_transform): Initialize sec_mem_p and change_p. Set up
9842 curr_insn_set. Call check_and_process_move only for single set
9843 insns. Use the new function. Move call of check_and_process_move
9844 after operand equiv substitution and address process.
9846 2013-04-26 Jakub Jelinek <jakub@redhat.com>
9849 * tree-ssa-uninit.c (compute_uninit_opnds_pos): In functions
9850 with nonlocal goto receivers or returns twice calls, ignore
9851 unininitialized values from abnormal edges to nl goto receiver
9852 or returns twice call.
9854 2013-04-26 Jakub Jelinek <jakub@redhat.com>
9856 PR tree-optimization/57051
9857 * fold-const.c (const_binop): Handle VEC_LSHIFT_EXPR
9858 and VEC_RSHIFT_EXPR if shift count is a multiple of element
9861 2013-04-26 Richard Biener <rguenther@suse.de>
9863 * omp-low.c (finalize_task_copyfn): Do not drop PROP_loops.
9864 (expand_omp_taskreg): Likewise. Mark loops for fixup.
9865 * tree-cfg.c (move_block_to_fn): Remap loop fathers.
9866 (fixup_loop_arrays_after_move): New function.
9867 (move_sese_region_to_fn): Properly outline the loop tree parts
9870 2013-04-26 Uros Bizjak <ubizjak@gmail.com>
9872 * config/i386/i386.md (type, unit): Fix long lines.
9874 2013-04-26 Richard Biener <rguenther@suse.de>
9876 * Makefile.in (lto-streamer-in.o): Add $(CFGLOOP_H) dependency.
9877 (lto-streamer-out.o): Likewise.
9878 * cfgloop.c (init_loops_structure): Export, add struct function
9879 argument and adjust.
9880 (flow_loops_find): Adjust.
9881 * cfgloop.h (enum loop_estimation): Add EST_LAST.
9882 (init_loops_structure): Declare.
9883 * lto-streamer-in.c: Include cfgloop.h.
9884 (input_cfg): Input the loop tree.
9885 * lto-streamer-out.c: Include cfgloop.h.
9886 (output_cfg): Output the loop tree.
9887 (output_struct_function_base): Do not drop PROP_loops.
9889 2013-03-26 Richard Biener <rguenther@suse.de>
9891 * tree-cfg.c (execute_build_cfg): Build the loop tree.
9892 (pass_build_cfg): Provide PROP_loops.
9893 (move_sese_region_to_fn): Remove loops that are outlined into fn
9895 * tree-inline.c: Include cfgloop.h.
9896 (initialize_cfun): Do not drop PROP_loops.
9897 (copy_loops): New function.
9898 (copy_cfg_body): Copy loop structure.
9899 (tree_function_versioning): Initialize destination loop tree.
9900 * tree-ssa-loop.c (pass_tree_loop_init): Do not provide PROP_loops.
9901 (pass_parallelize_loops): Do IL verification.
9902 * loop-init.c (loop_optimizer_init): Fixup loops if required.
9903 * tree-optimize.c (execute_fixup_cfg): If we need to cleanup
9904 the CFG make sure we fixup loops as well.
9905 * tree-ssa-tail-merge.c: Include cfgloop.h.
9906 (replace_block_by): When merging loop latches mark loops for fixup.
9907 * lto-streamer-out.c (output_struct_function_base): Drop
9909 * tree-ssa-phiopt.c: Include tree-scalar-evolution.h.
9910 (tree_ssa_cs_elim): Initialize the loop optimizer and SCEV.
9911 * ipa-split.c: Include cfgloop.h.
9912 (split_function): Add the new return block to the loop tree root.
9913 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Return
9914 whether we have removed the forwarder block.
9915 (merge_phi_nodes): If we removed a forwarder mark loops for fixup.
9916 * cfgloop.h (place_new_loop): Declare.
9917 * cfgloopmanip.c (place_new_loop): Export.
9918 * Makefile.in (asan.o): Add $(CFGLOOP_H) dependency.
9919 (tree-switch-conversion.o): Likewise.
9920 (tree-complex.o): Likewise.
9921 (tree-inline.o): Likewise.
9922 (tree-ssa-tailmerge.o): Likewise.
9923 (ipa-split.o): Likewise.
9924 (tree-ssa-phiopt.o): Add $(SCEV_H) dependency.
9925 (tree-ssa-copy.o): Likewise.
9926 * tree-switch-conversion.c: Include cfgloop.h
9927 (process_switch): If we emit a bit-test cascade, schedule loops
9929 * tree-complex.c: Include cfgloop.h.
9930 (expand_complex_div_wide): Properly add new basic-blocks to loops.
9931 * asan.c: Include cfgloop.h.
9932 (create_cond_insert_point): Properly add new basic-blocks to
9933 loops, schedule loop fixup.
9934 * cfgloop.c (verify_loop_structure): Check that looks are not
9936 * omp-low.c (expand_parallel_call): Properly add new basic-blocks
9938 (expand_omp_for_generic): Likewise.
9939 (expand_omp_sections): Likewise.
9940 (expand_omp_atomic_pipeline): Schedule loops for fixup.
9941 * tree-ssa-copy.c: Include tree-scalar-evolution.h.
9942 (fini_copy_prop): Disable DCE in substitute_and_fold if SCEV
9943 is initialized, not when loops are present.
9944 * tree-parloops.c (parallelize_loops): Remove checking here.
9945 * passes.c (init_optimization_passes): Schedule a copy-propagation
9946 pass before complete unrolling of inner loops.
9948 2013-04-26 Jakub Jelinek <jakub@redhat.com>
9950 * Makefile.in (toplev.o): Depend on diagnostic-color.h.
9951 * diagnostic-color.c (should_colorize): Remove _WIN32 version.
9952 (colorize_init): Add argument to _WIN32 version.
9953 * toplev.c: Include diagnostic-color.h.
9954 (process_options): Default to -fdiagnostics-color=auto if
9955 GCC_COLORS env var is in the environment.
9956 * common.opt (fdiagnostics-color=): Add Var and Init.
9957 * doc/invoke.texi (-fdiagnostics-color=): Document that if GCC_COLORS
9958 env var is in the environment, the default is auto rather than never.
9960 * diagnostic.h (file_name_as_prefix): Add context argument.
9961 * diagnostic.c (file_name_as_prefix): Likewise. Colorize
9962 the string as locus.
9963 * langhooks.c (lhd_print_error_function): Adjust caller.
9965 2013-04-25 Lawrence Crowl <crowl@google.com>
9967 * var-tracking.c (shared_hash_def::htab):
9968 Change type to hash_table. Update dependent calls and types.
9970 2013-04-25 Lawrence Crowl <crowl@google.com>
9972 * Makefile.in: Update as needed below.
9974 * alloc-pool.c (static hash_table <alloc_pool_hasher> alloc_pool_hash):
9975 Move declaration to after the type's method definitons.
9977 * attribs.c (htab_t scoped_attributes::attribute_hash):
9978 Change type to hash_table. Update dependent calls and types.
9980 * bitmap.c (htab_t bitmap_desc_hash):
9981 Change type to hash_table. Update dependent calls and types.
9983 * cselib.c (htab_t cselib_hash_table):
9984 Change type to hash_table. Update dependent calls and types.
9986 * data-streamer.h (struct string_slot): Move to lto-streamer.h.
9987 (hash_string_slot_node): Move implementation into lto-streamer.h
9988 struct string_slot_hasher.
9989 (eq_string_slot_node): Likewise.
9991 * data-streamer-out.c: Update output_block::string_hash_table
9992 dependent calls and types.
9994 * dwarf2cfi.c (htab_t trace_index):
9995 Change type to hash_table. Update dependent calls and types.
9997 * dwarf2out.c (htab_t break_out_includes::cu_hash_table):
9998 Change type to hash_table. Update dependent calls and types.
9999 (htab_t copy_decls_for_unworthy_types::decl_table): Likewise.
10000 (htab_t optimize_external_refs::map): Likewise.
10001 (htab_t output_comp_unit::extern_map): Likewise.
10002 (htab_t output_comdat_type_unit::extern_map): Likewise.
10003 (htab_t output_macinfo::macinfo_htab): Likewise.
10004 (htab_t optimize_location_lists::htab): Likewise.
10005 (htab_t dwarf2out_finish::comdat_type_table): Likewise.
10007 * except.c (htab_t ehspec_hash_type):
10008 Change type to hash_table. Update dependent calls and types.
10009 (assign_filter_values::ttypes): Likewise.
10010 (assign_filter_values::ehspec): Likewise.
10011 (sjlj_assign_call_site_values::ar_hash): Likewise.
10012 (convert_to_eh_region_ranges::ar_hash): Likewise.
10014 * gcse.c (htab_t pre_ldst_table):
10015 Change type to hash_table. Update dependent calls and types.
10017 * ggc-common.c (htab_t saving_htab):
10018 Change type to hash_table. Update dependent calls and types.
10019 (htab_t loc_hash): Likewise.
10020 (htab_t ptr_hash): Likewise.
10021 (call_count): Rename ggc_call_count.
10022 (call_alloc): Rename ggc_call_alloc.
10023 (loc_descriptor): Rename make_loc_descriptor.
10024 (add_statistics): Rename ggc_add_statistics.
10026 * ggc-common.c (saving_htab):
10027 Change type to hash_table. Update dependent calls and types.
10029 * gimple.h (struct gimplify_ctx): Move to gimplify-ctx.h.
10030 (push_gimplify_context): Likewise.
10031 (pop_gimplify_context): Likewise.
10032 (struct gimple_temp_hash_elt): Added.
10033 (struct gimplify_hasher): Likewise.
10034 (struct gimplify_ctx.temp_htab):
10035 Change type to hash_table. Update dependent calls and types.
10037 * gimple-fold.c: Include gimplify-ctx.h.
10039 * gimple-ssa-strength-reduction.c (htab_t base_cand_map):
10040 Change type to hash_table. Update dependent calls and types.
10041 (base_cand_dump_callback): Rename to ssa_base_cand_dump_callback to
10042 avoid potential global name collision.
10044 * gimplify.c: Include gimplify-ctx.h.
10045 (struct gimple_temp_hash_elt): Move to gimplify-ctx.h.
10046 (htab_t gimplify_ctx::temp_htab):
10047 Update dependent calls and types for new type hash_table.
10048 (gimple_tree_hash): Move into gimplify_hasher in gimplify-ctx.h.
10049 (gimple_tree_eq): Move into gimplify_hasher in gimplify-ctx.h.
10051 * gimplify-ctx.h: New.
10052 (struct gimple_temp_hash_elt): Move from gimplify.c.
10053 (class gimplify_hasher): New.
10054 (struct gimplify_ctx): Move from gimple.h.
10055 (htab_t gimplify_ctx::temp_htab):
10056 Change type to hash_table. Update dependent calls and types.
10058 * graphite-clast-to-gimple.c: Include graphite-htab.h.
10059 (htab_t ivs_params::newivs_index):
10060 Change type to hash_table. Update dependent calls and types.
10061 (htab_t ivs_params::params_index): Likewise.
10062 (htab_t print_generated_program::params_index): Likewise.
10063 (htab_t gloog::newivs_index): Likewise.
10064 (htab_t gloog::params_index): Likewise.
10066 * graphite.c: Include graphite-htab.h.
10067 4htab_t graphite_transform_loops::bb_pbb_mapping):
10068 Change type to hash_table. Update dependent calls and types.
10070 * graphite-clast-to-gimple.h: (extern gloog) Move to graphite-htab.h.
10071 (bb_pbb_map_hash): Fold into bb_pbb_htab_type in graphite-htab.h.
10072 (eq_bb_pbb_map): Fold into bb_pbb_htab_type in graphite-htab.h.
10074 * graphite-dependences.c: Include graphite-htab.h.
10075 (loop_is_parallel_p): Change hash table type of parameter.
10077 * graphite-htab.h: New.
10078 (typedef hash_table <bb_pbb_hasher> bb_pbb_htab_type): New.
10079 (extern find_pbb_via_hash): Move from graphite-poly.h.
10080 (extern loop_is_parallel_p): Move from graphite-poly.h.
10081 (extern get_loop_body_pbbs): Move from graphite-poly.h.
10083 * graphite-poly.h (extern find_pbb_via_hash): Move to graphite-htab.h.
10084 (extern loop_is_parallel_p): Move to graphite-htab.h.
10085 (extern get_loop_body_pbbs): Move to graphite-htab.h.
10087 * haifa-sched.c (htab_t delay_htab):
10088 Change type to hash_table. Update dependent calls and types.
10089 (htab_t delay_htab_i2): Likewise.
10091 * ira-color.c (htab_t allocno_hard_regs_htab):
10092 Change type to hash_table. Update dependent calls and types.
10094 * ira-costs.c (htab_t cost_classes_htab):
10095 Change type to hash_table. Update dependent calls and types.
10097 * loop-invariant.c (htab_t merge_identical_invariants::eq):
10098 Change type to hash_table. Update dependent calls and types.
10100 * loop-iv.c (htab_t bivs):
10101 Change type to hash_table. Update dependent calls and types.
10103 * loop-unroll.c (htab_t opt_info::insns_to_split):
10104 Change type to hash_table. Update dependent calls and types.
10105 (htab_t opt_info::insns_with_var_to_expand): Likewise.
10107 * lto-streamer.h (struct string_slot): Move from data-streamer.h
10108 (struct string_slot_hasher): New.
10109 (htab_t output_block::string_hash_table):
10110 Change type to hash_table. Update dependent calls and types.
10112 * lto-streamer-in.c (freeing_string_slot_hasher): New.
10113 (htab_t file_name_hash_table):
10114 Change type to hash_table. Update dependent calls and types.
10116 * lto-streamer-out.c: Update output_block::string_hash_table dependent
10119 * lto-streamer.c (htab_t tree_htab):
10120 Change type to hash_table. Update dependent calls and types.
10122 * omp-low.c: Include gimplify-ctx.h.
10124 * passes.c (htab_t name_to_pass_map):
10125 Change type to hash_table. Update dependent calls and types.
10126 (pass_traverse): Rename to passes_pass_traverse.
10128 * plugin.c (htab_t event_tab):
10129 Change type to hash_table. Update dependent calls and types.
10131 * postreload-gcse.c (htab_t expr_table):
10132 Change type to hash_table. Update dependent calls and types.
10133 (dump_hash_table_entry): Rename dump_expr_hash_table_entry.
10135 * sese.c (debug_rename_map_1): Make extern.
10136 (htab_t copy_bb_and_scalar_dependences::rename_map):
10137 Change type to hash_table. Update dependent calls and types.
10139 * sese.h (extern debug_rename_map): Move to .c file.
10141 * store-motion.c (htab_t store_motion_mems_table):
10142 Change type to hash_table. Update dependent calls and types.
10144 * trans-mem.c (htab_t tm_new_mem_hash):
10145 Change type to hash_table. Update dependent calls and types.
10147 * tree-browser.c (htab_t TB_up_ht):
10148 Change type to hash_table. Update dependent calls and types.
10150 * tree-cfg.c (htab_t discriminator_per_locus):
10151 Change type to hash_table. Update dependent calls and types.
10153 * tree-complex.c: Include tree-hasher.h
10154 (htab_t complex_variable_components):
10155 Change type to hash_table. Update dependent calls and types.
10157 * tree-eh.c (htab_t finally_tree):
10158 Change type to hash_table. Update dependent calls and types.
10160 * tree-flow.h (extern int_tree_map_hash): Moved into tree-hasher
10161 struct int_tree_hasher.
10162 (extern int_tree_map_eq): Likewise.
10163 (uid_decl_map_hash): Removed.
10164 (extern decl_tree_map_eq): Likewise.
10166 * tree-hasher.h: New.
10167 (struct int_tree_hasher): New.
10168 (typedef int_tree_htab_type): New.
10170 * tree-inline.c: Include gimplify-ctx.h.
10172 * tree-mudflap.c: Include gimplify-ctx.h.
10174 * tree-parloops.c: Include tree-hasher.h.
10175 (htab_t eliminate_local_variables_stmt::decl_address):
10176 Change type to hash_table. Update dependent calls and types.
10177 (htab_t separate_decls_in_region::decl_copies): Likewise.
10179 * tree-scalar-evolution.c (htab_t resolve_mixers::cache):
10180 Change type to hash_table. Update dependent calls and types.
10182 * tree-sra.c (candidates):
10183 Change type to hash_table. Update dependent calls and types.
10185 * tree-ssa.c (int_tree_map_eq): Moved into struct int_tree_hasher
10187 (int_tree_map_hash): Likewise.
10189 * tree-ssa-dom.c (htab_t avail_exprs):
10190 Change type to hash_table. Update dependent calls and types.
10192 * tree-ssa-live.c (var_map_base_init::tree_to_index):
10193 Change type to hash_table. Update dependent calls and types.
10195 * tree-ssa-loop-ivopts.c (struct ivopts_data.inv_expr_tab):
10196 Change type to hash_table. Update dependent calls and types.
10198 * tree-ssa-phiopt.c (seen_ssa_names):
10199 Change type to hash_table. Update dependent calls and types.
10201 * tree-ssa-strlen.c (decl_to_stridxlist_htab):
10202 Change type to hash_table. Update dependent calls and types.
10204 * tree-ssa-uncprop.c (equiv):
10205 Change type to hash_table. Update dependent calls and types.
10207 2013-04-25 Jakub Jelinek <jakub@redhat.com>
10209 PR rtl-optimization/57003
10210 * regcprop.c (copyprop_hardreg_forward_1): If ksvd.ignore_set_reg,
10211 call note_stores with kill_clobbered_value callback again after
10212 killing regs_invalidated_by_call.
10214 2013-04-25 James Greenhalgh <james.greenhalgh@arm.com>
10216 * config/aarch64/aarch64-simd.md
10217 (aarch64_simd_bsl<mode>_internal): Rewrite RTL to not use UNSPEC_BSL.
10218 (aarch64_simd_bsl<mode>): Likewise.
10219 * config/aarch64/iterators.md (unspec): Remove UNSPEC_BSL.
10221 2013-04-25 Marek Polacek <polacek@redhat.com>
10223 PR tree-optimization/57066
10224 * builtins.c (fold_builtin_logb): Return +Inf for -Inf.
10226 2013-04-25 James Greenhalgh <jame.greenhalgh@arm.com>
10228 * config/aarch64/aarch64-simd.md (neg<mode>2): Use VDQ iterator.
10230 2013-04-25 James Greenhalgh <james.greenhalgh@arm.com>
10232 * config/aarch64/aarch64-builtins.c
10233 (aarch64_fold_builtin): New.
10234 * config/aarch64/aarch64-protos.h (aarch64_fold_builtin): New.
10235 * config/aarch64/aarch64.c (TARGET_FOLD_BUILTIN): Define.
10236 * config/aarch64/aarch64-simd-builtins.def (abs): New.
10237 * config/aarch64/arm_neon.h
10238 (vabs<q>_<f32, 64>): Implement using __builtin_aarch64_fabs.
10240 2013-04-25 James Greenhalgh <james.greenhalgh@arm.com>
10241 Tejas Belagod <tejas.belagod@arm.com>
10243 * config/aarch64/aarch64-builtins.c
10244 (aarch64_gimple_fold_builtin): New.
10245 * config/aarch64/aarch64-protos.h (aarch64_gimple_fold_builtin): New.
10246 * config/aarch64/aarch64-simd-builtins.def (addv): New.
10247 * config/aarch64/aarch64-simd.md (addpv4sf): New.
10248 (addvv4sf): Update.
10249 * config/aarch64/aarch64.c (TARGET_GIMPLE_FOLD_BUILTIN): Define.
10251 2013-04-25 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
10253 * config/aarch64/aarch64.md
10254 (*cmp_swp_<optab><ALLX:mode>_shft_<GPI:mode>): New pattern.
10256 2013-04-25 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
10258 * config/aarch64/aarch64.md (*ngc<mode>): New pattern.
10259 (*ngcsi_uxtw): New pattern.
10261 2013-04-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10262 Julian Brown <julian@codesourcery.com>
10264 * config/arm/arm.c (neon_builtin_type_mode): Add T_V4HF.
10265 (TB_DREG): Add T_V4HF.
10266 (v4hf_UP): New macro.
10267 (neon_itype): Add NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
10268 (arm_init_neon_builtins): Handle NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
10269 Handle initialisation of V4HF. Adjust initialisation of reinterpret
10271 (arm_expand_neon_builtin): Handle NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
10272 (arm_vector_mode_supported_p): Handle V4HF.
10273 (arm_mangle_map): Handle V4HFmode.
10274 * config/arm/arm.h (VALID_NEON_DREG_MODE): Add V4HF.
10275 * config/arm/arm_neon_builtins.def: Add entries for
10276 vcvtv4hfv4sf, vcvtv4sfv4hf.
10277 * config/arm/neon.md (neon_vcvtv4sfv4hf): New pattern.
10278 (neon_vcvtv4hfv4sf): Likewise.
10279 * config/arm/neon-gen.ml: Handle half-precision floating point
10281 * config/arm/neon-testgen.ml: Handle Requires_FP_bit feature.
10282 * config/arm/arm_neon.h: Regenerate.
10283 * config/arm/neon.ml (type elts): Add F16.
10284 (type vectype): Add T_float16x4, T_floatHF.
10285 (type vecmode): Add V4HF.
10286 (type features): Add Requires_FP_bit feature.
10287 (elt_width): Handle F16.
10288 (elt_class): Likewise.
10289 (elt_of_class_width): Likewise.
10290 (mode_of_elt): Refactor.
10291 (type_for_elt): Handle F16, fix error messages.
10292 (vectype_size): Handle T_float16x4.
10293 (vcvt_sh): New function.
10294 (ops): Add entries for vcvt_f16_f32, vcvt_f32_f16.
10295 (string_of_vectype): Handle T_floatHF, T_float16, T_float16x4.
10296 (string_of_mode): Handle V4HF.
10297 * doc/arm-neon-intrinsics.texi: Regenerate.
10299 2013-04-25 James Greenhalgh <james.greenhalgh@arm.com>
10301 * config/aarch64/aarch64.c (aarch64_print_operand): Fix asm_fprintf
10302 format specifier in 'X' case.
10304 2013-04-25 Alan Modra <amodra@gmail.com>
10307 * config/rs6000/rs6000.md (rotlsi3_internal7): Rename to
10308 rotlsi3_internal7le and condition on !BYTES_BIG_ENDIAN.
10309 (rotlsi3_internal8be): New BYTES_BIG_ENDIAN insn.
10310 Repeat for many other rotate/shift and mask patterns using subregs.
10311 Name lshiftrt insns.
10312 (ashrdisi3_noppc64): Rename to ashrdisi3_noppc64be and condition
10313 on WORDS_BIG_ENDIAN.
10315 2013-04-25 Alan Modra <amodra@gmail.com>
10317 * config.gcc: Support little-endian powerpc-linux targets.
10318 * config/rs6000/linux.h (LINK_OS_LINUX_EMUL): Define.
10319 (LINK_OS_LINUX_SPEC): Define.
10320 * config/rs6000/linuxspe.h (TARGET_DEFAULT):
10321 Preserve MASK_LITTLE_ENDIAN.
10322 * config/rs6000/default64.h (TARGET_DEFAULT): Likewise.
10323 * config/rs6000/linuxaltivec.h (TARGET_DEFAULT): Likewise.
10324 * config/rs6000/linux64.h (OPTION_LITTLE_ENDIAN): Don't zero.
10325 (LINK_OS_LINUX_EMUL32, LINK_OS_LINUX_EMUL64): Define.
10326 (LINK_OS_LINUX_SPEC32, LINK_OS_LINUX_SPEC64): Use above.
10327 * config/rs6000/rs6000.c (output_toc): Don't use .tc for TARGET_ELF.
10328 Correct fp word order for little-endian. Don't shift toc entries
10329 smaller than a word for little-endian.
10330 * config/rs6000/rs6000.md (bswaphi2, bswapsi2 split): Comment.
10331 (bswapdi2 splits): Correct low-part subreg for little-endian.
10332 Remove wrong BYTES_BIG_ENDIAN tests, and rename vars to remove
10333 low/high where such is correct only for be.
10334 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Allow
10335 little-endian for -mcall-aixdesc.
10337 2013-04-25 Alan Modra <amodra@gmail.com>
10339 * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
10340 replace_equiv_address_nv.
10342 2013-04-25 Alan Modra <amodra@gmail.com>
10344 * config/rs6000/rs6000.c (rs6000_emit_set_long_const): Tidy.
10346 2013-04-24 Vladimir Makarov <vmakarov@redhat.com>
10349 2013-04-24 Vladimir Makarov <vmakarov@redhat.com>
10350 * rtl.h (struct rtx_def): ...
10352 2013-04-24 Vladimir Makarov <vmakarov@redhat.com>
10354 PR rtl-optimizations/57046
10355 * lra-constraints (split_reg): Set up lra_risky_transformations_p
10356 for multi-reg splits.
10358 2013-04-24 H.J. Lu <hongjiu.lu@intel.com>
10360 * config/i386/x86-64.h (ASM_SPEC): Support -mx32.
10362 2013-04-24 Sterling Augustine <saugustine@google.com>
10364 * dwarf2out.c (skeleton_debug_str_hash, add_skeleton_AT_string)
10365 (comp_dir_string, debug_str_dwo_section): New.
10366 (DEBUG_STR_DWO_SECTION): Rename to ...
10367 (DEBUG_DWO_STR_SECTION): ... this.
10368 (DEBUG_NORM_STR_SECTION): Delete.
10369 (DEBUG_STR_SECTION, DEBUG_STR_SECTION_FLAGS): Edit definitions.
10370 (DEBUG_STR_DWO_SECTION_FLAGS): New.
10371 (find_AT_string): Move most logic to ...
10372 (find_AT_string_in_table): ... here. New.
10373 (add_top_level_skeleton_die_attrs): Call comp_dir_string and
10374 add_skeleton_AT_string. Delete logic.
10375 (output_skeleton_debug_sections): Remove call to
10376 add_top_level_skeleton_die_attrs.
10377 (add_comp_dir_attribute): Move logic to comp_dir_string.
10378 (dwarf2out_init): Initialize debug_str_dwo_section.
10379 (output_indirect_string): Call find_string_form.
10380 (output_indirect_strings): Rewrite.
10381 (prune_unused_types): Empty skeleton_debug_str_hash.
10382 Call get_skeleton_type_unit and add_top_level_skeleton_die_attrs.
10383 (dwarf2out_finish): Call output_indirect_strings.
10385 2013-04-24 Paolo Carlini <paolo.carlini@oracle.com>
10387 * doc/cpp.texi: Remove __GXX_EXPERIMENTAL_CXX1Y__.
10389 2013-04-24 Vladimir Makarov <vmakarov@redhat.com>
10391 * rtl.h (struct rtx_def): Add comment for field jump.
10392 (LRA_SUBREG_P): New macro.
10393 * recog.c (register_operand): Check LRA_SUBREG_P.
10394 * lra.c (lra): Add note at the end of RTL code. Align non-empty
10396 * lra-spills.c (lra_spill): Align stack after spilling pseudos.
10397 (lra_final_code_change): Skip subreg change for operators.
10398 * lra-eliminations.c (eliminate_regs_in_insn): Make return earlier
10399 if there are no operand changes.
10400 * lra-constraints.c (curr_insn_set): New.
10401 (match_reload): Set LRA_SUBREG_P.
10402 (emit_spill_move): Ditto.
10403 (check_and_process_move): Use curr_insn_set. Process only single
10404 set insns. Don't initialize sec_mem_p and change_p.
10405 (simplify_operand_subreg): Use LRA_SUBREG_P.
10406 (reg_in_class_p): New function.
10407 (process_alt_operands): Use it. Use #if HAVE_ATTR_enabled instead
10408 of #ifdef. Add code to remove cycling.
10409 (process_address): Check EXTRA_CONSTRAINT_STR. Process even if
10410 non-null disp. Reload inner instead of disp when base and index
10411 are null. Try to put lo_sum into register.
10412 (EBB_PROBABILITY_CUTOFF): Redefine probability in percents.
10413 (check_and_process_move): Move code for move cost check to
10414 simple_move_p. Remove equiv_substitution.
10415 (simple_move_p): New function.
10416 (curr_insn_transform): Initialize sec_mem_p and change_p. Set up
10417 curr_insn_set. Call check_and_process_move only for single set
10418 insns. Use the new function. Move call of check_and_process_move
10419 after operand equiv substitution and address process.
10421 2013-04-24 James Greenhalgh <james.greenhalgh@arm.com>
10423 * config/aarch64/arm_neon.h (vld1<q>_lane*): Fix constraints.
10424 (vld1<q>_dup_<sufp><8, 16, 32, 64>): Likewise.
10425 (vld1<q>_<sufp><8, 16, 32, 64>): Likewise.
10427 2013-04-24 Paolo Carlini <paolo.carlini@oracle.com>
10429 * doc/cpp.texi: Document __GXX_EXPERIMENTAL_CXX1Y__.
10431 2013-04-24 Marek Polacek <polacek@redhat.com>
10433 * tree-scalar-evolution.h (analyze_scalar_evolution): Remove.
10434 * tree-scalar-evolution.c (get_exit_conditions_rec): Likewise.
10435 (select_loops_exit_conditions): Likewise.
10436 (number_of_iterations_for_all_loops): Likewise.
10437 (analyze_scalar_evolution_for_all_loop_phi_nodes): Likewise.
10438 (scev_analysis): Likewise.
10440 2013-04-02 Catherine Moore <clm@codesourcery.com>
10441 Chao-ying Fu <fu@mips.com>
10443 * config/mips/micromips.md (jraddiusp): New pattern.
10444 * config/mips/mips.c (mips_expand_epilogue): Use the JRADDIUSP
10445 instruction if possible.
10447 2013-04-24 Alan Modra <amodra@gmail.com>
10449 * config/rs6000/driver-rs6000.c (elf_dcachebsize): Fix comment pasto.
10451 2013-04-24 Julian Brown <julian@codesourcery.com>
10452 Chung-Lin Tang <cltang@codesourcery.com>
10454 * dwarf2out.c (gen_enumeration_type_die): Fix HOST_BITS_PER_WIDE_INT
10455 dependency behavior in enumeration type DIE generation. Add TODO note
10456 to comments about future DW_FORM_sdata/udata re-work of related code.
10458 2013-04-23 Lawrence Crowl <crowl@google.com>
10460 * Makefile.in: Update as needed below.
10462 * hash-table.h (class hash_table):
10463 Correct many methods with parameter types compare_type to the correct
10464 value_type. (Correct code was unlikely to notice the change.)
10465 (hash_table::elements_with_deleted) New.
10466 (class hashtable::iterator): New.
10467 (hashtable::begin()): New.
10468 (hashtable::end()): New.
10469 (FOR_EACH_HASH_TABLE_ELEMENT): New.
10471 * statistics.c (statistics_hashes):
10472 Change type to hash_table. Update dependent calls and types.
10474 * tree-into-ssa.c (var_infos):
10475 Change type to hash_table. Update dependent calls and types.
10477 * tree-ssa-coalesce.c (struct coalesce_list_d.list):
10478 Change type to hash_table. Update dependent calls and types.
10480 * tree-ssa-loop-im.c (struct mem_ref.refs):
10481 Change type to hash_table. Update dependent calls and types.
10483 * tree-ssa-reassoc.c (undistribute_ops_list::ctable):
10484 Change type to hash_table. Update dependent calls and types.
10486 * tree-ssa-sccvn.c (vn_tables_s::nary):
10487 Change type to hash_table. Update dependent calls and types.
10488 (vn_tables_s::phis): Likewise.
10489 (vn_tables_s::references): Likewise.
10491 * tree-ssa-sccvn.h (vn_nary_op_eq): Update parameter and return types.
10492 (vn_reference_eq): Update parameter and return types.
10494 * tree-ssa-structalias.c (pointer_equiv_class_table):
10495 Change type to hash_table. Update dependent calls and types.
10496 (location_equiv_class_table): Likewise.
10498 * tree-vect-data-refs.c: Consequential changes for making
10499 peeling a hash_table.
10501 * tree-vect-loop.c (new_loop_vec_info): Dependent hash_table update.
10502 (destroy_loop_vec_info): Dependent hash_table update.
10504 * tree-vectorizer.h (peeling_htab):
10505 Change type to hash_table. Update dependent calls and types.
10507 2013-04-23 Shiva Chen <shiva0217@gmail.com>
10509 * lra-assigns.c (find_hard_regno_for): Use lra_reg_val_equal_p
10510 to check the register content is equal or not.
10511 * lra-constraints.c (match_reload): Use lra_assign_reg_val
10512 to assign register content record.
10513 * lra-eliminations.c (update_reg_eliminate): Use
10514 lra_update_reg_val_offset to update register content offset.
10515 * lra-int.h (struct lra_reg): Add offset member.
10516 (lra_reg_val_equal_p): New static inline function.
10517 (lra_update_reg_val_offset): New static inline function.
10518 (lra_assign_reg_val): New static inline function.
10519 * lra.c (lra_create_new_reg): Use lra_assign_reg_val
10520 to assign register content record.
10521 (initialize_lra_reg_info_element): Initial offset to zero.
10523 2013-04-23 Catherine Moore <clm@codesourcery.com>
10525 * config/mips/mips.md (*movhi_internal, *movqi_internal): New
10526 operands. Record compression.
10528 2013-04-23 Xinliang David Li <davidxl@google.com>
10530 * cfghhooks.c (dump_bb_for_graph): Support 'slim' graph dump.
10532 2013-04-23 Richard Biener <rguenther@suse.de>
10534 PR middle-end/57036
10535 * tree-inline.c (copy_edges_for_bb): Add can_make_abnormal_goto
10536 parameter, only add abnormal goto edges from the copied body
10537 if the call could perform abnormal gotos.
10538 (copy_cfg_body): Adjust.
10540 2013-04-23 Sofiane Naci <sofiane.naci@arm.com>
10542 * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add simd attribute.
10544 2013-04-23 Andreas Schwab <schwab@linux-m68k.org>
10546 * coretypes.h (gimple_stmt_iterator): Add struct to make
10549 2013-04-23 Richard Biener <rguenther@suse.de>
10551 PR tree-optimization/57026
10552 * tree-vrp.c (simplify_conversion_using_ranges): Do not propagate
10553 from SSA names occuring in abnormal PHI nodes.
10555 2013-04-22 Andi Kleen <ak@linux.intel.com>
10557 * lto/lto.c (print_lto_report_1): Fix LTO report names.
10559 2013-04-22 Andi Kleen <ak@linux.intel.com>
10561 * lto/lto.c (print_lto_report_1): Declare early.
10562 (read_cgraph_and_symbols): Call print_lto_report_1 early.
10564 2013-04-22 Andi Kleen <ak@linux.intel.com>
10566 * common.opt (-flto-report-wpa): Add.
10567 * doc/invoke.texi (-flto-report-wpa): Add.
10568 * lto/lto.c (do_whole_program_analysis): Check for lto-report-wpa.
10571 2013-04-22 Xinliang David Li <davidxl@google.com>
10573 * graph.c (draw_cfg_node_succ_edges): Add branch probility as label.
10574 * cfghhooks.c (dump_bb_for_graph): Dump profile count and frquency.
10575 * Makefile.in: New dependency
10577 David Daney <ddaney.cavm@gmail.com>
10579 * configure.ac (gcc_cv_as_micromips_support): Use the
10580 --fatal-warnings option.
10581 * configure: Regenerate.
10583 2013-04-22 Marek Polacek <polacek@redhat.com>
10586 * tsan.c (instrument_expr): Don't instrument expression
10587 in case its size is zero.
10589 2013-04-22 Uros Bizjak <ubizjak@gmail.com>
10593 2013-03-17 Uros Bizjak <ubizjak@gmail.com>
10595 * config/alpha/alpha.c (TARGET_LRA_P): New define.
10597 2013-04-22 James Greenhalgh <james.greenhalgh@arm.com>
10599 * coretypes.h (gimple_stmt_iterator_d): Forward declare.
10600 (gimple_stmt_iterator): New typedef.
10601 * gimple.h (gimple_stmt_iterator): Rename to...
10602 (gimple_stmt_iterator_d): ... This.
10603 * doc/tm.texi.in (TARGET_FOLD_BUILTIN): Detail restriction that
10604 trees be valid for GIMPLE and GENERIC.
10605 (TARGET_GIMPLE_FOLD_BUILTIN): New.
10606 * gimple-fold.c (gimple_fold_call): Call target hook
10607 gimple_fold_builtin.
10608 * hooks.c (hook_bool_gsiptr_false): New.
10609 * hooks.h (hook_bool_gsiptr_false): New.
10610 * target.def (fold_stmt): New.
10611 * doc/tm.texi: Regenerate.
10613 2013-04-22 Vladimir Makarov <vmakarov@redhat.com>
10616 * lra-eliminations.c (mark_not_eliminable): Prevent elimination of
10617 a set sp if no stack realignment.
10619 2013-04-22 Nick Clifton <nickc@redhat.com>
10621 * config.gcc (tilegx-linux): Extend extra_objs rather than
10623 (tilepro-linux): Likewise.
10625 2013-04-22 James Greenhalgh <james.greenhalgh@arm.com>
10627 * config/aarch64/aarch64-builtins.c
10629 (CF0, CF1, CF2, CF3, CF4, CF10): New.
10630 (VAR<1-12>): Add MAP parameter.
10631 (BUILTIN_*): Likewise.
10632 * config/aarch64/aarch64-simd-builtins.def: Set MAP parameter.
10633 * config/aarch64/aarch64-simd.md (aarch64_sshl_n<mode>): Remove.
10634 (aarch64_ushl_n<mode>): Likewise.
10635 (aarch64_sshr_n<mode>): Likewise.
10636 (aarch64_ushr_n<mode>): Likewise.
10637 (aarch64_<maxmin><mode>): Likewise.
10638 (aarch64_sqrt<mode>): Likewise.
10639 * config/aarch64/arm_neon.h (vshl<q>_n_*): Use new builtin names.
10640 (vshr<q>_n_*): Likewise.
10642 2013-04-22 James Greenhalgh <james.greenhalgh@arm.com>
10644 * config/aarch64/aarch64-builtins.c
10645 (aarch64_simd_builtin_type_mode): Handle SF types.
10647 (BUILTIN_GPF): Define.
10648 (aarch64_init_simd_builtins): Handle SF types.
10649 * config/aarch64/aarch64-simd-builtins.def (frecpe): Add support.
10650 (frecps): Likewise.
10651 (frecpx): Likewise.
10652 * config/aarch64/aarch64-simd.md
10653 (simd_types): Update simd_frcp<esx> to simd_frecp<esx>.
10654 (aarch64_frecpe<mode>): New.
10655 (aarch64_frecps<mode>): Likewise.
10656 * config/aarch64/aarch64.md (unspec): Add UNSPEC_FRECP<ESX>.
10657 (v8type): Add frecp<esx>.
10658 (aarch64_frecp<FRECP:frecp_suffix><mode>): New.
10659 (aarch64_frecps<mode>): Likewise.
10660 * config/aarch64/iterators.md (FRECP): New.
10661 (frecp_suffix): Likewise.
10662 * config/aarch64/arm_neon.h
10663 (vrecp<esx><qsd>_<fd><32, 64>): Convert to using builtins.
10665 2013-04-22 Christian Bruel <christian.bruel@st.com>
10668 * config/sh/sh.h (enum reg_class): Remove DF_HI_REGS.
10669 (REG_CLASS_NAMES): Idem.
10670 (REG_CLASS_CONTENTS): Idem.
10671 (REGCLASS_HAS_FP_REG): Idem.
10672 * config/sh/sh.c (sh_cannot_change_mode_class): Idem.
10673 (sh_conditional_register_usage): Idem.
10675 2013-04-21 Jeff Law <law@redhat.com>
10677 * tree-ssa-forwprop.c (simplify_conversion_from_bitmask): New function.
10678 (ssa_forward_propagate_and_combine): Use it.
10680 2013-04-19 Vladimir Makarov <vmakarov@redhat.com>
10682 * lra.c: Update the flow chart diagram.
10684 2013-04-19 Vladimir Makarov <vmakarov@redhat.com>
10686 PR rtl-optimization/56847
10687 * lra-constraints.c (process_alt_operands): Discourage alternative
10688 with non-matche doffsettable memory constraint fro memory with
10691 2013-04-19 Richard Biener <rguenther@suse.de>
10693 PR tree-optimization/56982
10694 * builtins.def (BUILT_IN_LONGJMP): longjmp is not a leaf
10696 * gimplify.c (gimplify_call_expr): Notice special calls.
10697 (gimplify_modify_expr): Likewise.
10698 * tree-cfg.c (make_abnormal_goto_edges): Handle setjmp-like
10699 abnormal control flow receivers.
10700 (call_can_make_abnormal_goto): Handle cfun->calls_setjmp
10701 in the same way as cfun->has_nonlocal_labels.
10702 (gimple_purge_dead_abnormal_call_edges): Likewise.
10703 (stmt_starts_bb_p): Make setjmp-like abnormal control flow
10704 receivers start a basic-block.
10706 2013-04-19 Richard Biener <rguenther@suse.de>
10708 * tree-vectorizer.h (struct _slp_instance): Move load_permutation
10710 (struct _slp_tree): ... here. Make it a vector of unsigned ints.
10711 (SLP_INSTANCE_LOAD_PERMUTATION): Remove.
10712 (SLP_TREE_LOAD_PERMUTATION): Add.
10713 (vect_transform_slp_perm_load): Adjust prototype.
10714 * tree-vect-slp.c (vect_free_slp_tree): Adjust.
10715 (vect_free_slp_instance): Likewise.
10716 (vect_create_new_slp_node): Likewise.
10717 (vect_supported_slp_permutation_p): Remove.
10718 (vect_slp_rearrange_stmts): Adjust.
10719 (vect_supported_load_permutation_p): Likewise. Inline
10720 vect_supported_slp_permutation_p here.
10721 (vect_analyze_slp_instance): Compute load permutations per
10722 slp node instead of per instance.
10723 (vect_get_slp_defs): Adjust.
10724 (vect_transform_slp_perm_load): Likewise.
10725 (vect_schedule_slp_instance): Remove redundant code.
10726 (vect_schedule_slp): Remove hack for PR56270, add it ...
10727 * tree-vect-stmts.c (vectorizable_load): ... here, do not
10728 CSE loads for SLP. Adjust.
10730 2013-04-19 Greta Yorsh <Greta.Yorsh@arm.com>
10732 * config/arm/arm.c (load_multiple_sequence, ldm_stm_operation_p): Fix
10733 spelling in two comments.
10735 2013-04-19 Greta Yorsh <Greta.Yorsh@arm.com>
10738 * config/arm/arm.c (load_multiple_sequence): Require SP
10739 as base register for loads if SP is in the register list.
10741 2013-04-19 Martin Jambor <mjambor@suse.cz>
10743 PR tree-optimization/56718
10744 * ipa-cp.c (ipa_value_from_known_type_jfunc): Moved...
10745 * ipa-prop.c (ipa_binfo_from_known_type_jfunc): ...here, renamed
10746 and made public. Adjusted all callers.
10747 (ipa_intraprocedural_devirtualization): New function.
10748 * ipa-prop.h (ipa_binfo_from_known_type_jfunc): Declare.
10749 (ipa_intraprocedural_devirtualization): Likewise.
10750 * Makefile.in (tree-ssa-pre.o): Add ipa-prop.h to dependencies.
10752 2013-04-19 Richard Biener <rguenther@suse.de>
10754 PR tree-optimization/57000
10755 * tree-ssa-reassoc.c (pass_reassoc): Add TODO_update_ssa_only_virtuals.
10757 2013-04-19 Terry Guo <terry.guo@arm.com>
10759 * config/arm/cortex-m4-fpu.md (cortex_m4_v): Delete cpu unit.
10761 (cortex_m4_v_a, cortex_m4_v_b): ... new cpu units.
10762 (cortex_m4_v, cortex_m4_exa_va, cortex_m4_exb_vb): New reservations.
10763 (cortex_m4_fmacs): Use new reservations.
10764 (cortex_m4_f_load, cortex_m4_f_store): Likewise.
10766 2013-04-18 Vladimir Makarov <vmakarov@redhat.com>
10768 PR rtl-optimization/56999
10769 * lra-coalesce.c (coalescable_pseudo_p): Remove 2nd parameter and
10771 (lra_coalesce): Remove split_origin_bitmap and related code.
10772 * lra.c (lra): Coalesce after undoing inheritance. Recreate live
10773 ranges if necessary.
10775 2013-04-18 Uros Bizjak <ubizjak@gmail.com>
10777 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers):
10779 (ix86_expand_call): Remove clobbered_registers array and use
10780 x86_64_ms_sysv_extra_clobbered_registers instead.
10781 * config/i386/i386.h (x86_64_ms_sysv_extra_clobbered_registers):
10783 * config/i386/predicates.md (call_rex64_ms_sysv_operation): New
10785 * config/i386/i386.md (*call_rex64_ms_sysv): Use
10786 call_rex64_ms_sysv_operation predicate. Remove explicit clobbers.
10787 (*call_value_rex64_ms_sysv): Ditto.
10789 2013-04-18 Cary Coutant <ccoutant@google.com>
10791 * dwarf2out.c (output_pubnames): Check die_perennial_p of
10792 parent instead of die_mark.
10794 2013-04-18 Diego Novillo <dnovillo@google.com>
10796 * gimple.c (create_gimple_tmp): New.
10797 (get_expr_type): New.
10798 (build_assign): New.
10799 (build_type_cast): New.
10800 * gimple.h (enum ssa_mode): Define.
10801 (gimple_seq_set_location): New.
10802 * asan.c (build_check_stmt): Change some gimple_build_* calls
10803 to use build_assign and build_type_cast.
10805 2013-04-18 Richard Biener <rguenther@suse.de>
10807 * tree-vect-data-refs.c (vect_analyze_group_access): Properly
10808 handle negative step. Remove redundant checks.
10809 (vect_create_data_ref_ptr): Avoid ICEs with non-constant steps.
10810 * tree-vect-stmts.c (vectorizable_load): Instead of asserting
10811 for negative step and grouped loads fail to vectorize.
10813 2013-04-18 Steven Bosscher <steven@gcc.gnu.org>
10815 * emit-rtl.c (reset_insn_used_flags): New function.
10816 (reset_all_used_flags): Use it.
10817 (verify_insn_sharing): New function.
10818 (verify_rtl_sharing): Fix verification for SEQUENCEs.
10820 2013-04-18 Jakub Jelinek <jakub@redhat.com>
10822 PR tree-optimization/56984
10823 * tree-vrp.c (register_edge_assert_for_2): For (x >> M) < N
10824 and (x >> M) >= N don't register any assertion if N << M is the
10827 2013-04-18 Steven Bosscher <steven@gcc.gnu.org>
10829 * lower-subreg.c (resolve_simple_move): If called self-recursive,
10830 do not delete_insn insns that have not yet been emitted, only
10831 unlink them with remove_insn.
10832 * df-scan.c (df_insn_delete): Revert r197492.
10834 2013-04-17 Steven Bosscher <steven@gcc.gnu.org>
10836 * emit-rtl.c (link_insn_into_chain): Handle chaining of SEQUENCEs.
10837 * reorg.c (emit_delay_sequence): Simplify with emit-rtl API.
10839 2013-04-17 Greta Yorsh <Greta.Yorsh@arm.com>
10841 * config/arm/arm.md (movsicc_insn): Convert define_insn into
10842 define_insn_and_split.
10843 (and_scc,ior_scc,negscc): Likewise.
10844 (cmpsi2_addneg, subsi3_compare): Convert to named patterns.
10846 2013-04-17 Greta Yorsh <Greta.Yorsh@arm.com>
10848 * config/arm/arm.c (use_return_insn): Return 0 for targets that
10849 can benefit from using a sequence of LDRD instructions in epilogue
10850 instead of a single LDM instruction.
10852 2013-04-17 Manuel López-Ibáñez <manu@gcc.gnu.org>
10855 * doc/extend.texi: Fix typo.
10857 2013-04-17 Richard Biener <rguenther@suse.de>
10859 * tree-vect-slp.c (vect_build_slp_tree_1): Split out from ...
10860 (vect_build_slp_tree): ... here.
10861 (vect_build_slp_tree_1): Compute which stmts of the SLP group
10862 match. Remove special-casing of mismatched complex loads.
10863 (vect_build_slp_tree): Based on the result from vect_build_slp_tree_1
10864 re-try the match with swapped commutative operands.
10865 (vect_supported_load_permutation_p): Remove special-casing of
10866 mismatched complex loads.
10867 (vect_analyze_slp_instance): Adjust.
10869 2013-04-17 Richard Biener <rguenther@suse.de>
10871 PR rtl-optimization/56921
10872 * cfgloop.h (struct loop): Add simple_loop_desc member.
10873 (struct niter_desc): Mark with GTY(()).
10874 (simple_loop_desc): Do not use aux field but simple_loop_desc.
10875 * loop-iv.c (get_simple_loop_desc): Likewise.
10876 (free_simple_loop_desc): Likewise.
10879 2013-04-16 Richard Biener <rguenther@suse.de>
10881 PR rtl-optimization/56921
10882 * loop-init.c (pass_rtl_move_loop_invariants): Add
10883 TODO_do_not_ggc_collect to todo_flags_finish.
10884 (pass_rtl_unswitch): Same.
10885 (pass_rtl_unroll_and_peel_loops): Same.
10886 (pass_rtl_doloop): Same.
10888 2013-04-17 Eric Botcazou <ebotcazou@adacore.com>
10890 * tree-ssa-alias.c (nonoverlapping_component_refs_of_decl_p): New.
10891 (decl_refs_may_alias_p): Add REF1 and REF2 parameters.
10892 Use nonoverlapping_component_refs_of_decl_p to disambiguate component
10894 (refs_may_alias_p_1): Adjust call to decl_refs_may_alias_p.
10895 * tree-streamer.c (record_common_node): Adjust reference in comment.
10897 2013-04-17 Terry Guo <terry.guo@arm.com>
10899 * config/arm/cortex-m4.md: Add a new bypass.
10901 2013-04-16 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
10903 * config/aarch64/aarch64.md (*adds_<optab><mode>_multp2):
10905 (*subs_<optab><mode>_multp2): New pattern.
10906 (*adds_<optab><ALLX:mode>_<GPI:mode>): New pattern.
10907 (*subs_<optab><ALLX:mode>_<GPI:mode>): New pattern.
10909 2013-04-16 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
10911 * config/aarch64/aarch64.md (*adds_mul_imm_<mode>): New pattern.
10912 (*subs_mul_imm_<mode>): New pattern.
10914 2013-04-16 David Edelsohn <dje.gcc@gmail.com>
10917 * config/rs6000/vsx.md (vsx_mov<mode>): Add j->r alternative.
10918 (vsx_movti_64bit): Change j->wa to O->wa. Add n->r alternative.
10919 (vsx_movti_32bit): Change j->wa to O->wa.
10921 2013-04-16 Richard Biener <rguenther@suse.de>
10923 PR rtl-optimization/56921
10924 * loop-init.c (pass_rtl_move_loop_invariants): Add
10925 TODO_do_not_ggc_collect to todo_flags_finish.
10926 (pass_rtl_unswitch): Same.
10927 (pass_rtl_unroll_and_peel_loops): Same.
10928 (pass_rtl_doloop): Same.
10930 2013-04-16 Greta Yorsh <Greta.Yorsh@arm.com>
10932 * config/arm/arm.c (emit_multi_reg_push): New declaration
10933 for an existing function.
10934 (arm_emit_strd_push): New function.
10935 (arm_expand_prologue): Used here.
10936 (arm_emit_ldrd_pop): New function.
10937 (arm_expand_epilogue): Used here.
10938 (arm_get_frame_offsets): Update condition.
10939 (arm_emit_multi_reg_pop): Add a special case for load of a single
10940 register with writeback.
10942 2013-04-16 Uros Bizjak <ubizjak@gmail.com>
10944 * doc/invoke.texi (i386 Option): Reword -mstack-protector-guard
10947 2013-04-16 Richard Biener <rguenther@suse.de>
10949 PR tree-optimization/56756
10950 * tree-ssa-loop-im.c (struct first_mem_ref_loc_1): New functor.
10951 (first_mem_ref_loc): New.
10952 (execute_sm): Place the load temporarily before a previous
10953 access instead of in the latch edge to ensure its SSA dependencies
10954 are defined at points dominating the load.
10956 2013-04-16 Steven Bosscher <steven@gcc.gnu.org>
10958 * cfgrtl.c (cfg_layout_merge_blocks): Revert r184005, implement
10959 correct fix by moving header and footer insn to the footer of
10960 the merged basic block. Clear BB_END of the merged-away block.
10962 PR middle-end/43631
10963 * emit-rtl.c (make_note_raw): New function.
10964 (link_insn_into_chain): New static inline function.
10965 (add_insn): Use it.
10966 (add_insn_before, add_insn_after): Factor insn chain linking code...
10967 (add_insn_before_nobb, add_insn_after_nobb): ...here, new functions
10968 using link_insn_into_chain.
10969 (note_outside_basic_block_p): New helper function for emit_note_after
10970 and emit_note_before.
10971 (emit_note_after): Use nobb variant of add_insn_after if the note
10972 should not be contained in a basic block.
10973 (emit_note_before): Use nobb variant of add_insn_before if the note
10974 should not be contained in a basic block.
10975 (emit_note_copy): Use make_note_raw.
10976 (emit_note): Likewise.
10977 * bb-reorder.c (insert_section_boundary_note): Remove hack to set
10978 BLOCK_FOR_INSN to NULL manually for NOTE_INSN_SWITCH_TEXT_SECTIONS.
10979 * jump.c (cleanup_barriers): Use reorder_insns_nobb to avoid making
10980 the moved barrier the tail of the basic block it follows.
10981 * var-tracking.c (pass_variable_tracking): Add TODO_verify_flow.
10983 2013-04-15 Jakub Jelinek <jakub@redhat.com>
10985 PR tree-optimization/56962
10986 * gimple-ssa-strength-reduction.c (record_increment): Only set
10987 initializer if gimple_assign_rhs_code is {,POINTER_}PLUS_EXPR and
10988 either rhs1 or rhs2 is equal to c->base_expr.
10990 2013-04-15 Richard Biener <rguenther@suse.de>
10992 PR tree-optimization/56933
10993 * tree-vectorizer.h (struct _stmt_vec_info): Remove read_write_dep
10995 (GROUP_READ_WRITE_DEPENDENCE): Remove.
10996 (STMT_VINFO_GROUP_READ_WRITE_DEPENDENCE): Likewise.
10997 * tree-vect-data-refs.c (vect_analyze_group_access): Move
10998 dependence check ...
10999 vect_analyze_data_ref_dependence (vect_analyze_data_ref_dependence):
11001 * tree-vect-stmts.c (new_stmt_vec_info): Do not initialize
11002 GROUP_READ_WRITE_DEPENDENCE.
11004 2013-04-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
11006 * emit-rtl.c (reset_all_used_flags): New function.
11007 (verify_rtl_sharing): Call reset_all_used_flags before and after
11008 performing the checks.
11010 2013-04-15 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11012 * config/arm/arm.c (const_ok_for_dimode_op): Handle AND case.
11013 * config/arm/arm.md (*anddi3_insn): Change to insn_and_split.
11014 * config/arm/constraints.md (De): New constraint.
11015 * config/arm/neon.md (anddi3_neon): Delete.
11016 (neon_vand<mode>): Expand to standard anddi3 pattern.
11017 * config/arm/predicates.md (imm_for_neon_inv_logic_operand):
11018 Move earlier in the file.
11019 (neon_inv_logic_op2): Likewise.
11020 (arm_anddi_operand_neon): New predicate.
11022 2013-04-15 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
11024 * configure.ac (gcc_cv_ld_as_needed): Set
11025 gcc_cv_ld_as_needed_option, gcc_cv_no_as_needed_option.
11026 Use -z ignore, -z record on *-*-solaris2*.
11027 (HAVE_LD_AS_NEEDED): Update comment.
11028 (LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION): Define.
11029 * configure: Regenerate.
11030 * config.in: Regenerate.
11031 * gcc.c (init_gcc_specs) [USE_LD_AS_NEEDED]: Use
11032 LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION.
11033 * config/sol2.h [HAVE_LD_AS_NEEDED] (USE_LD_AS_NEEDED): Define.
11034 * doc/tm.texi.in (USE_LD_AS_NEEDED): Allow for --as-needed
11035 equivalents. Fix markup.
11036 * doc/tm.texi: Regenerate.
11038 2013-04-15 Andrew Hsieh <andrewhsieh.google.com>
11040 * config/i386/i386.opt: New option mstack-protector-guard=.
11041 * config/i386/i386-opts.h: Add enum stack_protector_guard.
11042 * config/i386/i386.h: Define TARGET_SSP_GLOBAL_GUARD and
11043 TARGET_SSP_TLS_GUARD.
11044 * config/i386/i386.c (ix86_option_override_internal): Set
11045 ix86_stack_protector_guard.
11046 * config/i386/i386.md (stack_protect_set): Enable for
11047 TARGET_SSP_TLS_GUARD only.
11048 (stack_protect_set_<mode>): Ditto.
11049 (stack_protect_test): Ditto.
11050 (stack_protect_test_<mode>): Ditto.
11051 * doc/invoke.texi (i386 Option): Document.
11053 2013-04-15 Eric Botcazou <ebotcazou@adacore.com>
11056 * config/sparc/sparc.c (enum sparc_mode_class): Add H_MODE value.
11057 (S_MODES): Set H_MODE bit.
11058 (SF_MODES): Set only S_MODE and SF_MODE bits.
11059 (DF_MODES): Set SF_MODES and only D_MODE and DF_MODE bits.
11060 (sparc_init_modes) <MODE_INT>: Set H_MODE bit for sub-word modes.
11061 <MODE_VECTOR_INT>: Do not set SF_MODE for sub-word modes.
11062 <MODE_FLOAT>: Likewise.
11064 2013-04-15 Joey Ye <joey.ye@arm.com>
11066 * config/arm/arm.c (thumb_far_jump_used_p): Fix typo in comments.
11068 2013-04-15 Joey Ye <joey.ye@arm.com>
11070 * config/arm/arm.c (thumb1_final_prescan_insn): Assert lr save
11072 (thumb_far_jump_used_p): Count instruction size and set
11075 2013-04-14 Eric Botcazou <ebotcazou@adacore.com>
11077 * reorg.c (fill_simple_delay_slots): Reindent block of code.
11078 * resource.c (mark_target_live_regs): Reformat conditional block.
11080 2013-04-13 Steven Bosscher <steven@gcc.gnu.org>
11082 * sched-deps.c (deps_analyze_insn): Do not check for EH_REGION insn
11083 notes, they are emitted only just before final.
11084 * sched-int.h: Include insn-attr.h before checking INSN_SCHEDULING.
11086 2013-04-13 Steven Bosscher <steven@gcc.gnu.org>
11088 * emit-rtl.c (remove_insn): Do not call df_insn_delete here.
11089 * cfgrtl.c (delete_insn): Call it here instead.
11090 * lra-spills.c (lra_final_code_change): Use delete_insn.
11091 * haifa-sched.c (sched_remove_insn): Likewise.
11092 * sel-sched-ir.c (return_nop_to_pool): Clear INSN_DELETED_P for nops
11093 returning to the nop pool.
11094 (sel_remove_insn): Simplify the only_disconnect case via remove_insn,
11095 use delete_insn for definitive removal. Clear BLOCK_FOR_INSN.
11097 2013-04-12 Steven Bosscher <steven@gcc.gnu.org>
11099 * doc/tm.texi.in (LOOP_ALIGN): Remove loop note references.
11100 * doc/tm.texi: Regenerated.
11102 2013-04-12 Uros Bizjak <ubizjak@gmail.com>
11104 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use ANY_QI_REGNO_P in
11107 2013-04-12 Steven Bosscher <steven@gcc.gnu.org>
11109 * df-core.c (df_find_def): Compare register numbers.
11110 (df_find_use): Likewise.
11112 2013-04-12 Vladimir Makarov <vmakarov@redhat.com>
11115 * config/i386/i386.c (ix86_hard_regno_mode_ok): Add
11116 lra_in_progress for return.
11118 2013-04-12 Greta Yorsh <Greta.Yorsh@arm.com>
11120 * config/arm/arm.md (mov_scc,mov_negscc,mov_notscc): Convert
11121 define_insn into define_insn_and_split and emit movsicc patterns.
11123 2013-04-12 Greta Yorsh <Greta.Yorsh@arm.com>
11125 * config/arm/arm.c (gen_operands_ldrd_strd): Initialize "base".
11127 2013-04-12 Richard Biener <rguenther@suse.de>
11129 * tree-pass.h (TODO_do_not_ggc_collect): New.
11130 * passes.c (execute_one_ipa_transform_pass): Honor
11131 TODO_do_not_ggc_collect.
11132 (execute_one_pass): Likewise.
11135 2013-04-10 Richard Biener <rguenther@suse.de>
11137 * passes.c (init_optimization_passes): Remove reload pass.
11138 * ira.c (do_reload): Merge into ...
11140 (rest_of_handle_reload): Remove.
11141 (pass_reload): Likewise.
11142 * config/i386/i386.c (ix86_option_override): Refer to ira instead
11143 of reload for vzeroupper pass placement.
11145 2013-04-12 Jakub Jelinek <jakub@redhat.com>
11147 PR tree-optimization/56918
11148 PR tree-optimization/56920
11149 * fold-const.c (int_const_binop_1): Use op1.mul_with_sign (op2, ...)
11150 instead of op1 - op2. Pass 2 * TYPE_PRECISION (type) as second
11151 argument to rshift method. For 2 * HOST_BITS_PER_WIDE_INT precision
11152 use wide_mul_with_sign method.
11154 2013-04-12 Richard Biener <rguenther@suse.de>
11156 * gimple.c (is_gimple_constant): Vector CONSTRUCTORs should
11157 not be considered a gimple constant.
11159 2013-04-12 Marc Glisse <marc.glisse@inria.fr>
11161 * fold-const.c (const_binop): Handle vector shifts by a scalar.
11162 (fold_binary_loc): Call const_binop also for mixed vector-scalar
11165 2013-04-12 Manuel López-Ibáñez <manu@gcc.gnu.org>
11166 Jakub Jelinek <jakub@redhat.com>
11168 * opts.c: Include diagnostic-color.h.
11169 (common_handle_option): Handle OPT_fdiagnostics_color_.
11170 * Makefile.in (OBJS-libcommon): Add diagnostic-color.o.
11171 (diagnostic.o, opts.o, pretty-print.o): Depend on diagnostic-color.h.
11172 (diagnostic-color.o): New.
11173 * common.opt (fdiagnostics-color, fdiagnostics-color=): New options.
11174 (diagnostic_color_rule): New enum.
11175 * dwarf2out.c (gen_producer_string): Don't print -fdiagnostics-color*.
11176 * langhooks.c (lhd_print_error_function): Add %r "locus" and %R around
11177 the location string.
11178 * diagnostic.def: Add 3rd argument to DEFINE_DIAGNOSTIC_KIND macros,
11179 either NULL, or color kind.
11180 * diagnostic-color.c: New file.
11181 * diagnostic-color.h: New file.
11182 * diagnostic-core.h (DEFINE_DIAGNOSTIC_KIND): Adjust macro for 3
11184 * doc/invoke.texi (-fdiagnostics-color): Document.
11185 * pretty-print.h (pp_show_color): Define.
11186 (struct pretty_print_info): Add show_color field.
11187 * diagnostic.c: Include diagnostic-color.h.
11188 (diagnostic_build_prefix): Adjust for 3 argument DEFINE_DIAGNOSTIC_KIND
11189 macros. Colorize error:, warning: etc. strings and also the location
11191 (diagnostic_show_locus): Colorize the caret line.
11192 * pretty-print.c: Include diagnostic-color.h.
11193 (pp_base_format): Handle %r and %R format specifiers. Colorize strings
11194 inside of %< %> quotes or quoted through q format modifier.
11196 2013-04-12 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
11198 * ifcvt.c (end_ifcvt_sequence): Mark a and b for unsharing as well.
11200 2013-04-11 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
11202 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Allow NEG
11203 code in CC_NZ mode.
11204 * config/aarch64/aarch64.md (*neg_<shift><mode>3_compare0): New
11207 2013-04-11 Marek Polacek <polacek@redhat.com>
11209 PR tree-optimization/48184
11210 * params.def (PARAM_ALIGN_THRESHOLD): Increase the minimum value to 1.
11212 2013-04-11 Eric Botcazou <ebotcazou@adacore.com>
11214 * stor-layout.c (skip_simple_constant_arithmetic): Move to...
11215 * tree.c (skip_simple_constant_arithmetic): ...here and make public.
11216 (skip_simple_arithmetic): Tidy up.
11217 * tree.h (skip_simple_constant_arithmetic): Declare.
11219 2013-04-11 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
11221 * config/aarch64/aarch64.h (REVERSIBLE_CC_MODE): Define.
11223 2013-04-11 Richard Biener <rguenther@suse.de>
11225 * tree-vect-loop.c (get_initial_def_for_induction): Properly
11226 generate vector constants.
11228 2013-04-11 Richard Biener <rguenther@suse.de>
11230 PR tree-optimization/56878
11231 * tree-flow.h (outermost_invariant_loop_for_expr): Declare.
11232 * tree-ssa-loop-ivopts.c (outermost_invariant_loop_for_expr):
11234 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
11235 Prefer to align the DR with the most invariant base address.
11237 2013-04-11 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
11239 * opts.c (common_handle_option): Fix formatting and add FALLTHRU
11242 2013-04-11 James Greenhalgh <james.greenhalgh@arm.com>
11244 * config/aarch64/aarch64-simd.md (aarch64_vcond_internal): Fix
11245 floating-point vector comparisons against 0.
11247 2013-04-11 Jakub Jelinek <jakub@redhat.com>
11249 PR tree-optimization/56899
11250 * fold-const.c (extract_muldiv_1): Apply distributive law
11251 only if TYPE_OVERFLOW_WRAPS (ctype).
11253 2013-04-11 Bin Cheng <bin.cheng@arm.com>
11256 * ira-costs.c (scan_one_insn): Check whether the source rtx of
11257 loading has side effect.
11259 2013-04-10 Steven Bosscher <steven@gcc.gnu.org>
11261 * config/sparc/sparc.c: Include tree-pass.h.
11262 (TARGET_MACHINE_DEPENDENT_REORG): Do not redefine.
11263 (sparc_reorg): Rename to sparc_do_work_around_errata. Move to
11264 head of file. Change return type. Split off gate function.
11265 (sparc_gate_work_around_errata): New function.
11266 (pass_work_around_errata): New pass definition.
11267 (insert_pass_work_around_errata) New pass insert definition to
11268 insert pass_work_around_errata just after delayed-branch scheduling.
11269 (sparc_option_override): Insert the pass.
11270 * config/sparc/t-sparc (sparc.o): Add TREE_PASS_H dependence.
11272 2013-04-10 David S. Miller <davem@davemloft.net>
11274 * config/sparc/sparc.h (ASM_CPU_SPEC): Pass -Av8 if -mcpu=supersparc
11275 or -mcpu=hypersparc.
11277 * target.def (cstore_mode): New hook.
11278 * target.h: Include insn-codes.h
11279 * targhooks.c: Likewise.
11280 (default_cstore_mode): New function.
11281 * targhooks.h: Declare it.
11282 * doc/tm.texi.in: New hook slot for TARGET_CSTORE_MODE.
11283 * doc/tm.texi: Rebuild.
11284 * expmed.c (emit_cstore): Obtain cstore boolean result mode using
11285 target hook, rather than inspecting the insn_data.
11286 * config/sparc/sparc.c (sparc_cstore_mode): New function.
11287 (TARGET_CSTORE_MODE): Redefine.
11288 (emit_scc_insn): When TARGET_ARCH64, emit new 64-bit boolean
11290 * config/sparc/predicates.md (cstore_result_operand): New special
11292 * config/sparc/sparc.md (cstoresi4, cstoredi4, cstore<F:mode>4):
11293 Use it for operand 0.
11294 (*seqsi_special): Rewrite using 'P' mode iterator on operand 0.
11295 (*snesi_special): Likewise.
11296 (*snesi_zero): Likewise.
11297 (*seqsi_zero): Likewise.
11298 (*sltu_insn): Likewise.
11299 (*sgeu_insn): Likewise.
11300 (*seqdi_special): Make operand 0 and comparison operation be of
11302 (*snedi_special): Likewise.
11303 (*snedi_special_vis3): Likewise.
11304 (*neg_snesi_zero): Rename to *neg_snesisi_zero.
11305 (*neg_snesi_sign_extend): Rename to *neg_snesidi_zero.
11306 (*snesi_zero_extend): Delete, covered by 'P' mode iterator.
11307 (*neg_seqsi_zero): Rename to *neg_seqsisi_zero.
11308 (*neg_seqsi_sign_extend): Rename to *neg_seqsidi_zero.
11309 (*seqsi_zero_extend): Delete, covered by 'P' mode iterator.
11310 (*sltu_extend_sp64): Likewise.
11311 (*neg_sltu_insn): Rename to *neg_sltusi_insn.
11312 (*neg_sltu_extend_sp64): Rename to *neg_sltudi_insn.
11313 (*sgeu_extend_sp64): Delete, covered by 'P' mode iterator.
11314 (*neg_sgeu_insn): Rename to *neg_sgeusi_insn.
11315 (*neg_sgeu_extend_sp64): Rename to *neg_sgeudi_insn.
11317 2013-04-10 Yufeng Zhang <yufeng.zhang@arm.com>
11319 * config/aarch64/aarch64.c (aarch64_print_extension): New function.
11320 (aarch64_start_file): Use the new function.
11322 2013-04-10 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
11323 Jason Merrill <jason@redhat.com>
11325 * common.opt: Add -gdwarf.
11326 * opts.c (common_handle_option): Handle it.
11327 * gcc.c (ASM_DEBUG_SPEC): Don't expect "-2" for DWARF.
11329 2013-04-10 Richard Biener <rguenther@suse.de>
11331 * passes.c (execute_todo): Do not call ggc_collect conditional here.
11332 (execute_one_ipa_transform_pass): But unconditionally here.
11333 (execute_one_pass): And here.
11334 (init_optimization_passes): Remove reload pass.
11335 * tree-pass.h (TODO_ggc_collect): Remove.
11336 (pass_reload): Likewise.
11337 * ira.c (do_reload): Merge into ...
11339 (rest_of_handle_reload): Remove.
11340 (pass_reload): Likewise.
11341 * config/i386/i386.c (ix86_option_override): Refer to ira instead
11342 of reload for vzeroupper pass placement.
11343 * <everywhere>: Remove TODO_ggc_collect from todo_flags_start
11344 and todo_flags_finish of all passes.
11346 2013-04-10 Richard Biener <rguenther@suse.de>
11348 * tree-vectorizer.h (struct _slp_oprnd_info): Remove
11349 first_const_oprnd field, rename first_def_type to first_op_type.
11350 * tree-vect-slp.c (vect_create_oprnd_info): Adjust.
11351 (vect_get_and_check_slp_defs): Always use the type of the
11352 operand. Allow mixed vect_external_def, vect_constant_def types.
11353 (vect_get_constant_vectors): Handle mixed vect_external_def,
11354 vect_constant_def types.
11356 2013-04-10 Joern Rennecke <joern.rennecke@embecosm.com>
11358 PR tree-optimization/55524
11359 * tree-ssa-math-opts.c
11360 (convert_mult_to_fma): Don't use an fms construct
11361 when we don't have an fms operation, but fnma, and it looks
11362 likely that we'll be able to use the latter.
11364 2013-04-10 Zhouyi Zhou <yizhouzhou@ict.ac.cn>
11366 * cif-code.def (OVERWRITABLE): Correct the comment for overwritable
11368 * ipa-inline.c (can_inline_edge_p): Let dump mechanism report the
11369 inline fail caused by overwritable functions.
11371 2013-04-10 Chung-Ju Wu <jasonwucj@gmail.com>
11373 * combine.c (simplify_compare_const): Use GET_MODE_MASK to filter out
11374 unnecessary bits in the constant power of two case.
11376 2013-04-10 Richard Biener <rguenther@suse.de>
11378 * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove
11379 broken code swapping operands.
11380 (vect_build_slp_tree): Do not compute load permutations here.
11381 (vect_analyze_slp_instance): Compute load permutations here,
11382 after building the SLP tree.
11384 2013-04-09 Christian Bruel <christian.bruel@st.com>
11386 * config/sh/sh.md (barrier_align): Use next/prev_active_insn instead
11387 of next/prev_real_insn.
11389 2013-04-09 Jan Hubicka <jh@suse.cz>
11391 * ipa.c (cgraph_externally_visible_p, varpool_externally_visible_p):
11392 Drop aliased parameter.
11393 (function_and_variable_visibility): Do not handle alias pairs.
11394 * cgraph.c (varpool_externally_visible_p): Update prototype.
11395 * varpool.c (varpool_add_new_variable): Update.
11397 2013-04-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11399 * config/arm/arm.md (minmax_arithsi_non_canon): New pattern.
11401 2013-04-09 Steven Bosscher <steven@gcc.gnu.org>
11403 * sched-vis.c (print_pattern): Print SEQUENCE of insns as insns.
11405 * config/sparc/sparc.md: Use define_c_enum for "unspec" and "unspecv".
11407 2013-04-09 Marek Polacek <polacek@redhat.com>
11409 PR tree-optimization/48762
11410 * params.def (PARAM_MAX_CSE_INSNS): Increase the minimum value to 1.
11412 2013-04-09 Richard Biener <rguenther@suse.de>
11414 * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove code
11416 (vect_build_slp_tree): Likewise.
11417 (vect_analyze_slp_cost_1, vect_analyze_slp_cost): New functions
11418 calculating the cost of a SLP instance.
11419 (vect_analyze_slp_instance): Use it from here, after building
11422 2013-04-09 Jakub Jelinek <jakub@redhat.com>
11424 PR middle-end/56883
11425 * omp-low.c (expand_omp_for_generic, expand_omp_for_static_nochunk,
11426 expand_omp_for_static_chunk): Use simple_p = true in
11427 force_gimple_operand_gsi calls when assigning to addressable decls.
11429 2013-04-09 Jeff Law <law@redhat.com>
11431 * tree-vrp.c (simplify_cond_using_ranges): Simplify test of boolean
11432 when the boolean was created by converting a wider object which
11433 had a boolean range.
11435 2013-04-09 Richard Biener <rguenther@suse.de>
11437 * tree-vectorizer.h (slp_void_p): Remove.
11438 (slp_tree): Typedef before _slp_tree declaration.
11439 (struct _slp_tree): Use a vector of slp_tree as children.
11440 (vect_get_place_in_interleaving_chain): Remove.
11441 * tree-vect-data-refs.c (vect_get_place_in_interleaving_chain):
11443 * tree-vect-slp.c (vect_get_place_in_interleaving_chain): ... here
11445 (vect_free_slp_tree, vect_print_slp_tree, vect_mark_slp_stmts,
11446 vect_mark_slp_stmts_relevant, vect_slp_rearrange_stmts,
11447 vect_detect_hybrid_slp_stmts, vect_slp_analyze_node_operations,
11448 vect_schedule_slp_instance, vect_remove_slp_scalar_calls):
11449 Use slp_node instead of slp_void_p and adjust.
11451 2013-04-09 Richard Biener <rguenther@suse.de>
11453 * tree-ssa-loop-manip.c (rewrite_into_loop_closed_ssa): Avoid
11454 work that is not necessary.
11456 2013-04-09 Jakub Jelinek <jakub@redhat.com>
11458 PR tree-optimization/56854
11459 * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Don't
11460 forward into clobber stmts if it would change MEM_REF lhs into
11463 2013-04-09 Maxim Kuvyrkov <maxim@kugelworks.com>
11465 * tree.c (type_hash_lookup, type_hash_add): Make static.
11466 * tree.h (type_hash_lookup, type_hash_add): Remove global declarations.
11468 2013-04-09 Richard Biener <rguenther@suse.de>
11470 * tree.h (unsave_expr_now): Remove.
11471 * tree-inline.c (mark_local_for_remap_r): Remove.
11472 (unsave_expr_1): Likewise.
11473 (unsave_r): Likewise.
11474 (unsave_expr_now): Likewise.
11475 * tree-ssa-copy.c (replace_exp_1): Use unshare_expr.
11476 (propagate_tree_value): Likewise.
11478 2013-04-08 Steven Bosscher <steven@gcc.gnu.org>
11480 * doc/rtl.texi (sequence): Rewrite documentation to match the
11481 current use of SEQUENCE rtl objects.
11482 * rtl.def (SEQUENCE): Likewise.
11484 * doc/rtl.texi (NOTE_INSN_EH_REGION_BEG, NOTE_INSN_EH_REGION_END):
11485 Update documentation.
11486 (NOTE_INSN_LOOP_BEG, NOTE_INSN_LOOP_END, NOTE_INSN_LOOP_CONT,
11487 NOTE_INSN_LOOP_VTOP): Remove documentation for non-existing notes.
11489 * reg-notes.def (REG_EH_CONTEXT): Remove unused note.
11491 2013-04-08 Teresa Johnson <tejohnson@google.com>
11493 * basic-block.h (GCOV_COMPUTE_SCALE): Define.
11494 * ipa-inline-analysis.c (param_change_prob): Use helper rounding divide
11496 (estimate_edge_size_and_time): Add comment to suggest using rounding
11498 (estimate_node_size_and_time): Ditto.
11499 (remap_edge_change_prob): Use helper rounding divide methods.
11500 * value-prof.c (gimple_divmod_fixed_value_transform): Ditto.
11501 (gimple_mod_pow2_value_transform): Ditto.
11502 (gimple_mod_subtract_transform): Ditto.
11503 (gimple_ic_transform): Ditto.
11504 (gimple_stringops_transform): Ditto.
11505 * stmt.c (conditional_probability): Ditto.
11506 (emit_case_dispatch_table): Ditto.
11507 * lto-cgraph.c (merge_profile_summaries): Ditto.
11508 * tree-optimize.c (execute_fixup_cfg): Ditto.
11509 * cfgcleanup.c (try_forward_edges): Ditto.
11510 * cfgloopmanip.c (scale_loop_profile): Ditto.
11512 (duplicate_loop_to_header_edge): Ditto.
11513 (lv_adjust_loop_entry_edge): Ditto.
11514 * tree-vect-loop.c (vect_transform_loop): Ditto.
11515 * profile.c (compute_branch_probabilities): Ditto.
11516 * cfgbuild.c (compute_outgoing_frequencies): Ditto.
11517 * lto-streamer-in.c (input_cfg): Ditto.
11518 * gimple-streamer-in.c (input_bb): Ditto.
11519 * ipa-cp.c (update_profiling_info): Ditto.
11520 (update_specialized_profile): Ditto.
11521 * tree-vect-loop-manip.c (slpeel_tree_peel_loop_to_edge): Ditto.
11522 * cfg.c (update_bb_profile_for_threading): Add comment to suggest using
11524 * sched-rgn.c (compute_dom_prob_ps): Ditto.
11525 (compute_trg_info): Ditto.
11526 * cfgrtl.c (force_nonfallthru_and_redirect): Ditto.
11527 (purge_dead_edges): Ditto.
11528 * loop-unswitch.c (unswitch_loop): Ditto.
11529 * cgraphclones.c (cgraph_clone_edge): Ditto.
11530 (cgraph_clone_node): Ditto.
11531 * tree-inline.c (copy_bb): Ditto.
11532 (copy_edges_for_bb): Ditto.
11533 (initialize_cfun): Ditto.
11534 (copy_cfg_body): Ditto.
11535 (expand_call_inline): Ditto.
11537 2013-04-08 Kai Tietz <ktietz@redhat.com>
11539 * config/i386/cygwin.h (EXTRA_OS_CPP_BUILTINS): Replaced
11540 TARGET_CYGWIN64 by TARGET_64BIT.
11542 2013-04-08 Joern Rennecke <joern.rennecke@embecosm.com>
11544 * config/epiphany/epiphany.md (GPR_1): New constant.
11545 (define_expand "mov<mode>cc): FAIL if gen_compare_reg returned 0.
11546 * config/epiphany/epiphany.c (gen_compare_reg):
11547 For flag_finite_math_only, avoid swapping operands when r0 and/or r1
11548 is already in place.
11549 Use GPR_0 / GPR_1 instead of 0/1 for r0/r1 register numbers.
11550 Don't require being called during rtl expansion; If y operlaps r0,
11552 (epiphany_compute_frame_size, epiphany_expand_prologue): Use GPR_1.
11553 (epiphany_expand_epilogue): Likewise.
11555 * config/epiphany/epiphany.c (epiphany_select_cc_mode):
11556 Don't use CC_FPmode for ORDERED / UNORDERED.
11557 * config/epiphany/epiphany.md (cmpsf_ord): Make pattern unconditional.
11559 * config/epiphany/constraints.md (CnL): New constraint.
11560 * config/epiphany/epiphany.md (addsi3_i): Add r/r/CnL alternative.
11561 * config/epiphany/predicates.md (add_operand): Allow 1024.
11563 * config/epiphany/epiphany.md (logical_op): New code iterator.
11564 (op_mnc): New code attribute.
11565 (<op_mnc>_f, mov_f, cstoresi4): New patterns.
11566 (mov_f+1, mov_f+2): New peephole2 patterns.
11568 * config/epiphany/epiphany.md (mov_f+2): New peephole2 pattern.
11569 (cstoresi4): Also allow re-use of zero result when doing a NE
11570 comparison to a non-zero operand.
11571 Use (clobber (scratch)) for first insn if the gpr output is not needed.
11573 * config/epiphany/epiphany.md (<insn_opname>v2si3):
11574 Use gen_addsi3_i / gen_subsi3_i.
11576 2013-04-08 Jakub Jelinek <jakub@redhat.com>
11580 * tree-eh.c (optimize_clobbers): Only remove clobbers if bb doesn't
11581 contain anything but clobbers, at most one __builtin_stack_restore,
11582 optionally debug stmts and final resx, and if it has at least one
11583 incoming EH edge. Don't check for SSA_NAME on LHS of a clobber.
11584 (sink_clobbers): Don't check for SSA_NAME on LHS of a clobber.
11585 Instead of moving clobbers with MEM_REF LHS with SSA_NAME address
11586 which isn't defaut definition, remove them.
11587 (unsplit_eh, cleanup_empty_eh): Use single_{pred,succ}_{p,edge}
11588 instead of EDGE_COUNT comparisons or EDGE_{PRED,SUCC}.
11589 * tree-ssa-ccp.c (execute_fold_all_builtins): Remove clobbers
11590 with MEM_REF LHS with SSA_NAME address.
11592 2013-04-08 Jeff Law <law@redhat.com>
11594 * gimple.c (canonicalize_cond_expr_cond): Rewrite x ^ y into x != y.
11596 2013-04-08 Richard Biener <rguenther@suse.de>
11598 * gimple-pretty-print.c (debug_gimple_stmt): Do not print
11600 * tree-vect-loop.c (vect_determine_vectorization_factor): Dump
11601 determined vector type.
11602 (vect_analyze_data_refs): Likewise.
11603 (vect_get_new_vect_var): Adjust.
11604 (vect_create_destination_var): Preserve SSA name versions.
11605 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Do
11606 not dump anything here.
11608 2013-04-08 Joern Rennecke <joern.rennecke@embecosm.com>
11610 * config/epiphany/epiphany.h (struct GTY (()) machine_function):
11611 Add member lr_slot_known.
11612 * config/epiphany/epiphany.md (reload_insi_ra): Compute lr_slot_offs
11614 * config/epiphany/epiphany.c (epiphany_compute_frame_size):
11615 Remove code that sets lr_slot_offset according to what a previous
11616 version of epiphany_emit_save_restore used to do.
11617 (epiphany_emit_save_restore): When doing an lr save or restore,
11618 set/verify lr_slot_known and lr_slot_offset.
11620 2013-04-08 Xinyu Qi <xyqi@marvell.com>
11623 * config/arm/arm.h (REG_CLASS_CONTENTS): Include IWMMXT_GR_REGS
11626 2013-04-08 Richard Biener <rguenther@suse.de>
11628 * alias.c (find_base_term): Fix thinko in previous change.
11630 2013-04-08 Jakub Jelinek <jakub@redhat.com>
11632 * tree-loop-distribution.c (const_with_all_bytes_same): New function.
11633 (generate_memset_builtin): Only handle integer_all_onesp as -1 val if
11634 TYPE_PRECISION is equal to mode bitsize. Use const_with_all_bytes_same
11635 if possible to compute val.
11636 (classify_partition): Verify CONSTRUCTOR doesn't have any elts.
11637 For QImode integers don't require anything about precision. Use
11638 const_with_all_bytes_same to find out if the constant doesn't have
11639 repeated bytes in it.
11641 2013-04-08 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
11643 * config/s390/s390.c (s390_expand_insv): Only accept insertions
11646 2013-04-08 Marek Polacek <polacek@redhat.com>
11648 PR rtl-optimization/48182
11649 * params.def (PARAM_MIN_CROSSJUMP_INSNS): Increase the minimum
11652 2013-04-06 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
11655 * config/pa/pa.c (legitimize_pic_address): Before incrementing label
11656 nuses, make sure we have a label.
11658 2013-04-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11661 * config/rs6000/rs6000.c (rs6000_emit_swdiv_high_precision): Remove.
11662 (rs6000_emit_swdiv_low_precision): Remove.
11663 (rs6000_emit_swdiv): Rewrite to handle between one and four
11664 iterations of Newton-Raphson generally; modify required number of
11665 iterations for some cases.
11666 * config/rs6000/rs6000.h (RS6000_RECIP_HIGH_PRECISION_P): Remove.
11668 2013-04-05 Steven Bosscher <steven@gcc.gnu.org>
11670 * bb-reorder.c (fix_crossing_unconditional_branches): Remove a
11671 set-but-unused variable.
11673 * cgraph.c (cgraph_release_function_body): Clear cfun->cfg to make
11674 basic blocks of released function bodies garbage-collectable.
11676 * ree.c (find_and_remove_re): Do not call df_finish_pass here.
11677 (struct rtl_opt_pass): Add TODO_df_finish.
11679 * rtl.def (DEFINE_SUBST, DEFINE_SUBST_ATTR): Add documentation.
11681 2013-04-05 Greta Yorsh <Greta.Yorsh@arm.com>
11683 * config/arm/constraints.md (q): New constraint.
11684 * config/arm/ldrdstrd.md: New file.
11685 * config/arm/arm.md (ldrdstrd.md) New include.
11686 (arm_movdi): Use "q" instead of "r" constraint
11687 for double-word memory access.
11688 (movdf_soft_insn): Likewise.
11689 * config/arm/vfp.md (movdi_vfp): Likewise.
11690 * config/arm/t-arm (MD_INCLUDES): Add ldrdstrd.md.
11691 * config/arm/arm-protos.h (gen_operands_ldrd_strd): New declaration.
11692 * config/arm/arm.c (gen_operands_ldrd_strd): New function.
11693 (mem_ok_for_ldrd_strd): Likewise.
11694 (output_move_double): Update assertion.
11696 2013-04-05 Greta Yorsh <Greta.Yorsh@arm.com>
11698 * config/arm/arm.md: Comment on splitting Thumb1 patterns.
11700 2013-04-05 Greta Yorsh <Greta.Yorsh@arm.com>
11702 * config/arm/arm.md (arm_smax_insn): Convert define_insn into
11703 define_insn_and_split.
11704 (arm_smin_insn,arm_umaxsi3,arm_uminsi3): Likewise.
11706 2013-04-05 Greta Yorsh <Greta.Yorsh@arm.com>
11708 * config/arm/arm.md (arm_ashldi3_1bit): Convert define_insn into
11709 define_insn_and_split.
11710 (arm_ashrdi3_1bit,arm_lshrdi3_1bit): Likewise.
11711 (shiftsi3_compare): New pattern.
11712 (rrx): New pattern.
11713 * config/arm/unspecs.md (UNSPEC_RRX): New.
11715 2013-04-05 Greta Yorsh <Greta.Yorsh@arm.com>
11717 * config/arm/arm.md (negdi_extendsidi): New pattern.
11718 (negdi_zero_extendsidi): Likewise.
11720 2013-04-05 Greta Yorsh <Greta.Yorsh@arm.com>
11722 * config/arm/arm.md (andsi_iorsi3_notsi): Convert define_insn into
11723 define_insn_and_split.
11724 (arm_negdi2,arm_abssi2,arm_neg_abssi2): Likewise.
11725 (arm_cmpdi_insn,arm_cmpdi_unsigned): Likewise.
11727 2013-04-05 Greta Yorsh <Greta.Yorsh@arm.com>
11729 * config/arm/arm.md (arm_subdi3): Convert define_insn into
11730 define_insn_and_split.
11731 (subdi_di_zesidi,subdi_di_sesidi): Likewise.
11732 (subdi_zesidi_di,subdi_sesidi_di,subdi_zesidi_zesidi): Likewise.
11734 2013-04-05 Greta Yorsh <Greta.Yorsh@arm.com>
11736 * config/arm/arm.md (subsi3_carryin): New pattern.
11737 (subsi3_carryin_const): Likewise.
11738 (subsi3_carryin_compare,subsi3_carryin_compare_const): Likewise.
11739 (subsi3_carryin_shift,rsbsi3_carryin_shift): Likewise.
11741 2013-04-05 Greta Yorsh <Greta.Yorsh@arm.com>
11743 * config/arm/arm.md (incscc,arm_incscc,decscc,arm_decscc): Delete.
11745 2013-04-05 Greta Yorsh <Greta.Yorsh@arm.com>
11747 * config/arm/arm.md (addsi3_carryin_<optab>): Set attribute predicable.
11748 (addsi3_carryin_alt2_<optab>,addsi3_carryin_shift_<optab>): Likewise.
11750 2013-04-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11752 * config/arm/arm.c (arm_expand_builtin): Change fcode
11753 type to unsigned int.
11755 2013-04-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
11757 * doc/invoke.texi (ARM Options): Document cortex-a53 support.
11759 2013-04-04 Ian Lance Taylor <iant@google.com>
11761 * doc/standards.texi (Standards): The Go frontend supports the Go 1
11764 2013-04-04 Steven Bosscher <steven@gcc.gnu.org>
11766 PR middle-end/56729
11767 * df-scan.c (df_insn_delete): Disable failing assert.
11769 2013-04-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11771 * config/arm/arm-protos.h (arm_builtin_vectorized_function):
11772 New function prototype.
11773 * config/arm/arm.c (TARGET_VECTORIZE_BUILTINS): Define.
11774 (TARGET_VECTORIZE_BUILTIN_VECTORIZED_FUNCTION): Likewise.
11775 (arm_builtin_vectorized_function): New function.
11777 2013-04-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11779 * config/arm/arm_neon_builtins.def: New file.
11780 * config/arm/arm.c (neon_builtin_data): Move contents to
11781 arm_neon_builtins.def.
11782 (enum arm_builtins): Include neon builtin definitions.
11783 (ARM_BUILTIN_NEON_BASE): Move from enum to macro.
11784 * config/arm/t-arm (arm.o): Add dependency on arm_neon_builtins.def.
11786 2013-04-04 Marek Polacek <polacek@redhat.com>
11788 PR tree-optimization/48186
11789 * predict.c (maybe_hot_frequency_p): Return false if
11790 HOT_BB_FREQUENCY_FRACTION is 0.
11791 (cgraph_maybe_hot_edge_p): Likewise.
11793 2013-04-04 Richard Biener <rguenther@suse.de>
11795 PR tree-optimization/56826
11796 * tree-vect-slp.c (vect_build_slp_tree): Compute ncopies
11799 2013-04-04 Richard Biener <rguenther@suse.de>
11801 PR tree-optimization/56213
11802 * tree-vect-data-refs.c (vect_check_strided_load): Remove.
11803 (vect_analyze_data_refs): Allow all non-nested loads as strided loads.
11805 2013-04-04 Richard Biener <rguenther@suse.de>
11807 PR tree-optimization/56837
11808 * tree-loop-distribution.c (classify_partition): For non-zero
11809 values require that the value has the same precision as its
11810 mode to be useful as memset value.
11812 2013-04-03 Nick Clifton <nickc@redhat.com>
11814 * config/v850/v850e3v5.md (fmasf4): Use fmaf.s on E3V5 architectures.
11815 (fmssf4): Use fmsf.s on E3V5 architectures.
11816 (fnmasf4): Use fnmaf.s on E3V5 architectures.
11817 (fnmssf4): Use fnmsf.s on E3V5 architectures.
11819 2013-04-03 Jeff Law <law@redhat.com>
11821 * Makefile.in (lra-constraints.o): Depend on $(OPTABS_H).
11822 (lra-eliminations.o): Likewise.
11824 2013-04-03 Teresa Johnson <tejohnson@google.com>
11826 * gcov-io.c (compute_working_sets): Moved most of body of old
11827 compute_working_sets here from profile.c.
11828 * gcov-io.h (NUM_GCOV_WORKING_SETS): Moved here from profile.c.
11829 (gcov_working_set_t): Moved typedef here from basic-block.h
11830 (compute_working_set): Declare.
11831 * profile.c (NUM_GCOV_WORKING_SETS): Moved to gcov-io.h.
11832 (get_working_sets): Renamed from compute_working_set,
11833 replace most of body with call to new compute_working_sets.
11834 (get_exec_counts): Replace call to compute_working_sets
11835 to get_working_sets.
11836 * profile.h (get_working_sets): Renamed from compute_working_set.
11837 * lto-cgraph.c (input_symtab): Replace call to compute_working_sets
11838 to get_working_sets.
11839 * basic-block.h (gcov_working_set_t): Moved to gcov-io.h.
11840 * gcov-dump.c (dump_working_sets): New function.
11842 2013-04-03 Kenneth Zadeck <zadeck@naturalbridge.com>
11844 * hwint.c (sext_hwi, zext_hwi): New functions.
11845 * hwint.h (HOST_BITS_PER_HALF_WIDE_INT, HOST_HALF_WIDE_INT,
11846 HOST_HALF_WIDE_INT_PRINT, HOST_HALF_WIDE_INT_PRINT_C,
11847 HOST_HALF_WIDE_INT_PRINT_DEC, HOST_HALF_WIDE_INT_PRINT_DEC_C,
11848 HOST_HALF_WIDE_INT_PRINT_UNSIGNED, HOST_HALF_WIDE_INT_PRINT_HEX,
11849 HOST_HALF_WIDE_INT_PRINT_HEX_PURE): New symbols.
11850 (sext_hwi, zext_hwi): New functions.
11852 2013-04-03 Jeff Law <law@redhat.com>
11854 PR tree-optimization/56799
11855 * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Bring
11856 back test for widening conversion erroneously dropped in prior change.
11858 2013-04-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11861 * config/aarch64/aarch64.c (is_jump_table): Use next_active_insn
11862 instead of next_real_insn.
11864 2013-04-03 Marek Polacek <polacek@redhat.com>
11867 * tsan.c (instrument_func_exit): Allow BUILT_IN_RETURN functions.
11869 2013-04-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11872 * config/arm/arm.c (is_jump_table): Use next_active_insn instead of
11874 (thumb1_output_casesi): Likewise.
11875 (thumb2_output_casesi): Likewise.
11877 2013-04-03 Richard Biener <rguenther@suse.de>
11879 PR tree-optimization/56817
11880 * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely):
11882 (tree_unroll_loops_completely_1): ... new function to manually
11883 walk the loop tree, properly defering outer loops of unrolled
11884 loops to later iterations.
11886 2013-04-03 Marc Glisse <marc.glisse@inria.fr>
11888 * tree-vect-stmts.c (vectorizable_store): Accept BIT_FIELD_REF.
11889 (vectorizable_load): Likewise.
11890 * tree-vect-slp.c (vect_build_slp_tree): Likewise.
11891 * tree-vect-data-refs.c (vect_create_data_ref_ptr): Handle VECTOR_TYPE.
11893 2013-04-03 Marc Glisse <marc.glisse@inria.fr>
11895 * tree-flow-inline.h (get_addr_base_and_unit_offset_1): Handle
11898 2013-04-03 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
11900 * config/spu/spu.c (emit_nop_for_insn): Handle JUMP_TABLE_DATA.
11902 2013-04-03 Bin Cheng <bin.cheng@arm.com>
11904 * rtl.h (AUTO_INC_DEC): Fix typo of HAVE_POST_MODIFY_DISP.
11906 2013-04-03 Marc Glisse <marc.glisse@inria.fr>
11908 PR tree-optimization/56790
11909 * fold-const.c (fold_ternary_loc) <VEC_COND_EXPR>: Add constant
11912 2013-04-03 Marc Glisse <marc.glisse@inria.fr>
11914 * simplify-rtx.c (simplify_binary_operation_1) <VEC_SELECT>:
11916 (simplify_ternary_operation) <VEC_MERGE>: Use unsigned HOST_WIDE_INT
11917 for masks. Test for side effects. Handle nested VEC_MERGE. Handle
11920 2013-04-03 Jakub Jelinek <jakub@redhat.com>
11923 * tree.h (force_folding_builtin_constant_p): New decl.
11924 * builtins.c (force_folding_builtin_constant_p): New variable.
11925 (fold_builtin_constant_p): Fold immediately also if
11926 force_folding_builtin_constant_p.
11928 2013-04-03 Richard Biener <rguenther@suse.de>
11930 PR tree-optimization/56812
11931 * tree-vect-data-refs.c (vect_slp_analyze_data_ref_dependence):
11932 DRs of the same interleaving chain are independent.
11934 2013-04-02 Jason Merrill <jason@redhat.com>
11936 * gdbinit.in (pbb): Use debug fn.
11938 2013-04-02 Lawrence Crowl <crowl@google.com>
11940 * sese.h (struct ivtype_map_elt_s): Remove unused.
11941 (extern debug_ivtype_map): Remove unused.
11942 (extern eq_ivtype_map_elts): Remove unused.
11943 * sese.c (debug_ivtype_map): Removed unused.
11944 (debug_ivtype_map_1): Removed unused.
11945 (debug_ivtype_elt): Remove unused.
11946 (eq_ivtype_map_elts): Remove unused.
11949 2013-04-02 Kai Tietz <ktietz@redhat.com>
11952 * config/i386/cygming.h (SUB_TARGET_RECORD_STUB): New sub-target macro.
11953 * config/i386/i386-protos.h (i386_pe_record_stub): Add new prototype.
11954 * config/i386/i386.c (legitimize_pe_coff_extern_decl): New static
11956 (legitimize_pe_coff_symbol): Likewise.
11957 (is_imported_p): New helper-function.
11958 (ix86_option_override_internal): Make MEDIUM_PIC the default code-model
11959 for Windows x64 targets.
11960 (ix86_expand_prologue): Optimize for pe-coff targets.
11961 (ix86_expand_split_stack_prologue): Adjust for pe-coff targets.
11962 (legitimate_pic_address_disp_p): Adjust for x64 pe-coff to support
11963 medium/large code-model.
11964 (legitimize_pic_address): Likewise.
11965 (legitimize_tls_address): Likewise.
11966 (ix86_expand_call): Likewise.
11967 (x86_output_mi_thunk): Likewise.
11968 (get_dllimport_decl): Add new beimport argument.
11969 (construct_plt_address): Don't assert for x64 pe-coff targets.
11970 * config/i386/i386.h (PIC_OFFSET_TABLE_REGNUM): Adjust for x64 pe-coff
11972 (SYMBOL_FLAG_STUBVAR): New macro.
11973 (SYMBOL_REF_STUBVAR_P): Likewise.
11974 * config/i386/winnt.c (stub_list): New structure.
11975 (stub_head): New local variable.
11976 (i386_pe_record_stub): New function.
11977 (i386_pe_file_end): Emit refptr-stubs.
11979 2013-04-02 Jakub Jelinek <jakub@redhat.com>
11981 PR rtl-optimization/56745
11982 * ifcvt.c (cond_exec_find_if_block): Don't try to optimize
11983 if then_bb has no successors and else_bb is EXIT_BLOCK_PTR.
11986 * tree-ssa-alias.c (stmt_kills_ref_p_1): If base != ref->base
11987 and both of them are MEM_REFs, just compare first argument for
11988 equality and attempt to deal even with differing offsets.
11991 * tree-cfg.c (verify_gimple_assign_single): Allow lhs
11992 of gimple_clobber_p to be MEM_REF.
11993 * gimplify.c (gimplify_modify_expr): Gimplify *to_p of
11994 an assignment from TREE_CLOBBER_P. Allow it to be MEM_REF
11995 after gimplification.
11996 * asan.c (get_mem_ref_of_assignment): Don't instrument
11997 gimple_clobber_p stmts.
11998 * tree-ssa-dse.c (dse_optimize_stmt): Allow DSE of
11999 gimple_clobber_p stmt if they have MEM_REF lhs and
12000 are dead because of another gimple_clobber_p stmt.
12001 * tree-ssa-live.c (clear_unused_block_pointer): Treat
12002 gimple_clobber_p stmts like debug stmts.
12003 (remove_unused_locals): Remove clobbers with MEM_REF lhs
12004 that refer to unused VAR_DECLs or uninitialized values.
12005 * tree-sra.c (sra_ipa_reset_debug_stmts): Also remove
12006 gimple_clobber_p stmts if they refer to removed parameters.
12007 (get_repl_default_def_ssa_name, sra_ipa_modify_expr): Fix up
12010 2013-04-02 Uros Bizjak <ubizjak@gmail.com>
12012 * config/i386/i386.md (*testqi_ext_3): Merge with *testqi_ext_3_rex64
12013 using SWI48 mode attribute.
12015 2013-04-02 Wei Mi <wmi@google.com>
12017 * config/i386/i386.c (ix86_rtx_costs): Set proper rtx cost for
12018 ashl<mode>3_mask, *<shift_insn><mode>3_mask and
12019 *<rotate_insn><mode>3_mask in i386.md.
12021 2013-04-02 Alexander Ivchenko <alexander.ivchenko@intel.com>
12023 * config.gcc (arm*-*-linux-*): Remove duplicate t-linux-android.
12025 2013-04-02 Richard Biener <rguenther@suse.de>
12027 PR tree-optimization/56778
12028 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence):
12029 Runtime alias tests are not supported for gather loads.
12030 * tree-vect-loop-manip.c (vect_loop_versioning): Insert
12031 stmts referenced from SSA operands before updating SSA form.
12033 2013-04-02 Ian Caulfield <ian.caulfield@arm.com>
12034 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
12036 * config/arm/arm-arches.def (armv8-a): Default to cortex-a53.
12037 * config/arm/t-arm (MD_INCLUDES): Depend on cortex-a53.md.
12038 * config/arm/cortex-a53.md: New file.
12039 * config/arm/bpabi.h (BE8_LINK_SPEC): Handle cortex-a53.
12040 * config/arm/arm.md (generic_sched, generic_vfp): Handle cortex-a53.
12041 * config/arm/arm.c (arm_issue_rate): Likewise.
12042 * config/arm/arm-tune.md: Regenerate
12043 * config/arm/arm-tables.opt: Regenerate.
12044 * config/arm/arm-cores.def: Add cortex-a53.
12046 2013-04-02 Zhenqiang Chen <zhenqiang.chen@arm.com>
12048 * config/arm/uclinux-elf.h: Add %L to LINK_GCC_C_SEQUENCE_SPEC for
12051 2013-04-02 Sofiane Naci <sofiane.naci@arm.com>
12053 * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add variants for
12054 scalar load/store operations using B/H registers.
12055 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64): Likewise.
12057 2013-04-02 Sofiane Naci <sofiane.naci@arm.com>
12059 * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add alternatives for
12061 * config/aarch64/aarch64.c
12062 (aarch64_simd_scalar_immediate_valid_for_move): New.
12063 * config/aarch64/aarch64-protos.h
12064 (aarch64_simd_scalar_immediate_valid_for_move): New.
12065 * config/aarch64/constraints.md (Dh, Dq): New.
12066 * config/aarch64/iterators.md (hq): New.
12068 2013-04-02 Eric Botcazou <ebotcazou@adacore.com>
12070 * reorg.c (get_branch_condition): Deal with conditional returns.
12071 (fill_simple_delay_slots): Remove dead code dealing with jumps.
12073 2013-04-01 Wei Mi <wmi@google.com>
12075 * config/i386/i386.md (*ashl<mode>3_mask): Rewrite as define_insn.
12076 Truncate operand 2 using %b asm operand modifier.
12077 (*<shift_insn><mode>3_mask): Ditto.
12078 (*<rotate_insn><mode>3_mask): Ditto.
12080 2013-04-01 Steven Bosscher <steven@gcc.gnu.org>
12082 PR middle-end/56798
12083 * cfgbuild.c (inside_basic_block_p): Restore check broken at r197234.
12085 2013-03-31 Kaz Kojima <kkojima@gcc.gnu.org>
12087 * config/sh/sh.md (casesi_worker_1): Use next_active_insn instead
12089 (casesi_worker_2, casesi_shift_media, casesi_load_media): Likewise.
12091 2013-03-30 Lawrence Crowl <crowl@google.com>
12093 * dse.c (clear_alias_sets): Remove never set.
12094 (disqualified_clear_alias_sets): Remove never set.
12095 (clear_alias_mode_pool): Remove never set.
12096 (dse_step0): Remove condition that is never true.
12097 (canon_address): Remove condition that is never true.
12098 (dse_step7): Remove condition that is never true.
12099 (rest_of_handle_dse): Remove condition that is never true.
12100 (rest_of_handle_dse::did_global): Remove never read from above.
12101 (dse_step2_spill): Remove never called from above.
12102 (dse_step5_spill): Remove never called from above.
12104 2013-03-30 Steven Bosscher <steven@gcc.gnu.org>
12106 * doc/md.texi (Standard Names) <casesi>: Update documentation for
12107 JUMP_TABLE_DATA changes.
12108 * doc/tm.texi.in (Dispatch Tables) <ASM_OUTPUT_CASE_LABEL>: Likewise.
12109 * doc/rtl.texi (Flags) <INSN_DELETED_P, SCHED_GROUP_P>: Likewise.
12110 (Insns) <jump_table_data>: New entry.
12111 * doc/tm.texi: Regenerate.
12113 * cfgrtl.c (fixup_reorder_chain): Do not emit barriers to BB_FOOTER.
12115 * postreload-gcse.c (bb_has_well_behaved_predecessors): Correct test
12116 for table jump at the end of a basic block using tablejump_p.
12117 * targhooks.c (default_invalid_within_doloop): Likewise.
12118 * config/rs6000/rs6000.c (TARGET_INVALID_WITHIN_DOLOOP): Remove
12119 target hook implementation that is identical to the default hook.
12120 (rs6000_invalid_within_doloop): Remove.
12122 * bb-reorder.c (fix_crossing_unconditional_branches): Remove set but
12123 unused variable from tablejump_p call.
12125 * rtl.def (JUMP_TABLE_DATA): New RTX_INSN object.
12126 * rtl.h (RTX_PREV, RTX_NEXT): Adjust for new JUMP_TABLE_DATA.
12127 (INSN_DELETED_P): Likewise.
12128 (emit_jump_table_data): New prototype.
12129 * gengtype.c (adjust_field_rtx_def): Handle JUMP_TABLE_DATA fields
12130 after 4th as unused.
12131 * print-rtl.c (print_rtl): Handle JUMP_TABLE_DATA.
12132 * sched-vis.c (print_insn): Likewise.
12133 * emit-rtl.c (active_insn_p): Consider JUMP_TABLE_DATA an active
12134 insn for compatibility with back ends that use next_active_insn to
12135 identify jump table data.
12136 (set_insn_deleted): Remove no longer useful JUMP_TABLE_DATA_P check.
12137 (remove_insn): Likewise.
12138 (emit_insn): Do not accept JUMP_TABLE_DATA objects in insn chains
12140 (emit_debug_insn, emit_jump_insn, emit_call_insn, emit_label): Idem.
12141 (emit_jump_table_data): New function.
12143 * cfgbuild.c (inside_basic_block_p): A JUMP_INSN is always inside a
12144 basic block, a JUMP_TABLE_DATA never is.
12145 (control_flow_insn_p): JUMP_TABLE_DATA is not a control flow insn.
12146 * cfgrtl.c (duplicate_insn_chain): Split handling of JUMP_TABLE_DATA
12147 off from code handling real insns.
12148 * final.c (get_attr_length_1): Simplify for JUMP_INSNs.
12149 * function.c (instantiate_virtual_regs): Remove JUMP_TABLE_DATA_P
12150 test, now redundant because JUMP_TABLE_DATA is not an INSN_P insn.
12151 * gcse.c (insert_insn_end_basic_block): Likewise, JUMP_TABLE_DATA_P
12152 is not a NONDEBUG_INSN_P.
12153 * ira-costs.c (scan_one_insn): Likewise.
12154 * jump.c (mark_all_labels): Likewise.
12155 (mark_jump_label_1): Likewise.
12156 * lra-eliminations.c (eliminate_regs_in_insn): Likewise.
12157 * lra.c (get_insn_freq): Expect all insns reaching here to be in
12159 (check_rtl): Remove JUMP_TABLE_DATA_P test, not a NONDEBUG_INSN_P insn.
12160 * predict.c (expensive_function_p): Use FOR_BB_INSNS.
12161 * reload1.c (calculate_needs_all_insns): Call set_label_offsets for
12162 JUMP_TABLE_DATA_P insns.
12163 (calculate_elim_costs_all_insns): Likewise.
12164 (set_label_offsets): Recurse on the PATTERN of JUMP_TABLE_DATA insns.
12165 (elimination_costs_in_insn): Remove redundant JUMP_TABLE_DATA_P test.
12166 (delete_output_reload): Code style fixups.
12167 * reorg.c (dbr_schedule): Move JUMP_TABLE_DATA_P up to avoid setting
12168 insn flags on this non-insn.
12169 * sched-rgn.c (add_branch_dependences): Treat JUMP_TABLE_DATA insns
12170 as scheduling barriers, for pre-change compatibility.
12171 * stmt.c (emit_case_dispatch_table): Emit jump table data not as
12172 JUMP_INSN objects but instead as JUMP_TABLE_DATA objects.
12174 * config/alpha/alpha.c (alpha_does_function_need_gp): Remove
12175 redundant JUMP_TABLE_DATA_P test.
12176 * config/arm/arm.c (thumb_far_jump_used_p): Likewise.
12177 * config/frv/frv.c (frv_function_contains_far_jump): Likewise.
12178 (frv_for_each_packet): Likewise.
12179 * config/i386/i386.c (min_insn_size): Likewise.
12180 (ix86_avoid_jump_mispredicts): Likewise.
12181 * config/m32r/m32r.c (m32r_is_insn): Likewise.
12182 * config/mep/mep.c (mep_reorg_erepeat): Likewise.
12183 * config/mips/mips.c (USEFUL_INSN_P): Likewise.
12184 (mips16_insn_length): Robustify.
12185 (mips_has_long_branch_p): Remove redundant JUMP_TABLE_DATA_P test.
12186 (mips16_split_long_branches): Likewise.
12187 * config/pa/pa.c (pa_combine_instructions): Likewise.
12188 * config/rs6000/rs6000.c (get_next_active_insn): Treat
12189 JUMP_TABLE_DATA objects as active insns, like in active_insn_p.
12190 * config/s390/s390.c (s390_chunkify_start): Treat JUMP_TABLE_DATA
12191 as contributing to pool range lengths.
12192 * config/sh/sh.c (find_barrier): Restore check for ADDR_DIFF_VEC.
12193 Remove redundant JUMP_TABLE_DATA_P test.
12194 (sh_loop_align): Likewise.
12195 (split_branches): Likewise.
12196 (sh_insn_length_adjustment): Likewise.
12197 * config/spu/spu.c (get_branch_target): Likewise.
12199 2013-03-29 Jan Hubicka <jh@suse.cz>
12201 * lto-cgraph.c (output_profile_summary, input_profile_summary): Use
12202 gcov streaming; stream hot bb threshold to ltrans.
12203 * predict.c (get_hot_bb_threshold): Break out from ....
12204 (maybe_hot_count_p): ... here.
12205 (set_hot_bb_threshold): New function.
12206 * lto-section-in.c (lto_section_name): Add profile.
12207 * profile.h (get_hot_bb_threshold, set_hot_bb_threshold): Declare.
12208 * ipa.c: Include hash-table.h, tree-inline.h, profile.h, lto-streamer.h
12209 and data-streamer.h
12210 (histogram_entry): New structure.
12211 (histogram, histogram_pool): New global vars.
12212 (histogram_hash): New structure.
12213 (histogram_hash::hash): New method.
12214 (histogram_hash::equal): Likewise.
12215 (account_time_size): New function.
12216 (cmp_counts): New function.
12217 (dump_histogram): New function.
12218 (ipa_profile_generate_summary): New function.
12219 (ipa_profile_write_summary): New function.
12220 (ipa_profile_read_summary): New function.
12221 (ipa_profile): Decide on threshold.
12222 (pass_ipa_profile): Add ipa_profile_write_summary and
12223 ipa_profile_read_summary.
12224 * Makefile.in (ipa.o): Update dependencies.
12225 * lto-streamer.h (LTO_section_ipa_profile): New section.
12227 2013-03-29 Gabriel Dos Reis <gdr@integrable-solutions.net>
12229 * tree.h (VAR_P): New.
12231 2013-03-29 Paolo Carlini <paolo.carlini@oracle.com>
12234 * doc/invoke.texi ([-fwhole-program]): Fix typo.
12236 2013-03-29 Steven Bosscher <steven@gcc.gnu.org>
12238 * cfgbuild.c (inside_basic_block_p): Use JUMP_TABLE_DATA_P in lieu
12239 of tests for JUMP_P and a ADDR_DIFF_VEC or ADDR_VEC pattern.
12240 (control_flow_insn_p): Likewise.
12241 * cfgrtl.c (duplicate_insn_chain): Likewise.
12242 * final.c (get_attr_length_1): Likewise.
12243 (shorten_branches): Likewise.
12244 (final_scan_insn): Likewise.
12245 * function.c (instantiate_virtual_regs): Likewise.
12246 * gcse.c (insert_insn_end_basic_block): Likewise.
12247 * ira-costs.c (scan_one_insn): Likewise.
12248 * lra-eliminations.c (eliminate_regs_in_insn): Likewise.
12249 * lra.c (check_rtl): Likewise.
12250 * reload1.c (elimination_costs_in_insn): Likewise.
12251 * reorg.c (follow_jumps): Likewise.
12253 * config/arm/arm.c (is_jump_table): Use JUMP_TABLE_DATA_P in lieu
12254 of tests for JUMP_P and a ADDR_DIFF_VEC or ADDR_VEC pattern.
12255 (thumb_far_jump_used_p): Likewise.
12256 * config/bfin/bfin.c (workaround_rts_anomaly): Likewise.
12257 (workaround_speculation): Likewise.
12258 (add_sched_insns_for_speculation): Likewise.
12259 * config/c6x/c6x.c (reorg_emit_nops): Likewise.
12260 * config/frv/frv.c (frv_function_contains_far_jump): Likewise.
12261 (frv_for_each_packet): Likewise.
12262 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Likewise.
12263 * config/ia64/ia64.c (emit_all_insn_group_barriers): Likewise.
12264 (final_emit_insn_group_barriers): Likewise.
12265 * config/m32r/m32r.c (m32r_is_insn): Likewise.
12266 * config/mips/mips.c (USEFUL_INSN_P): Likewise.
12267 (mips16_insn_length): Likewise.
12268 * config/pa/pa.c (pa_reorg): Likewise.
12269 (pa_combine_instructions): Likewise.
12270 * config/rs6000/rs6000.c (rs6000_invalid_within_doloop): Likewise.
12271 * config/sh/sh.c (fixup_addr_diff_vecs): Likewise.
12272 (sh_reorg): Likewise.
12273 (split_branches): Likewise.
12274 * config/spu/spu.c (get_branch_target): Likewise.
12276 * config/s390/s390.c (s390_chunkify_start): Simplify logic using
12279 2013-03-29 Kirill Yukhin <kirill.yukhin@intel.com>
12281 * config/i386/avx2intrin.h (_mm256_broadcastsi128_si256):
12282 Fix declaration name.
12284 2013-03-28 Lawrence Crowl <crowl@google.com>
12286 * graphds.h (struct graph.indicies): Remove unused.
12287 * graphite-poly.h (struct graph.original_pddrs): Remove unused.
12288 (SCOP_ORIGINAL_PDDRS): Remove unused.
12289 * sese.h (extern insert_loop_close_phis): Removed unused.
12290 (extern insert_guard_phis): Removed unused.
12291 (extern ivtype_map_elt_info): Removed unused.
12292 (new_ivtype_map_elt): Removed unused.
12293 * sese.c (ivtype_map_elt_info): Removed unused.
12295 2013-03-28 Lawrence Crowl <crowl@google.com>
12297 * Makefile.in: Add several missing include dependences.
12299 (test-dump.o): New. This object is not added to any executable,
12300 but is present for ad-hoc testing.
12302 (debug (const bitmap_head_def &)): New.
12303 (debug (const bitmap_head_def *)): New.
12305 (extern debug (const bitmap_head_def &)): New.
12306 (extern debug (const bitmap_head_def *)): New.
12308 (debug (edge_def &)): New.
12309 (debug (edge_def *)): New.
12311 (debug (basic_block_def &)): New.
12312 (debug (basic_block_def *)): New.
12314 (dump_node (const_tree, int, FILE *)): Correct source file.
12316 (debug (die_struct &)): New.
12317 (debug (die_struct *)): New.
12319 (extern debug (die_struct &)): New.
12320 (extern debug (die_struct *)): New.
12321 * gimple-pretty-print.c
12322 (debug (gimple_statement_d &)): New.
12323 (debug (gimple_statement_d *)): New.
12324 * gimple-pretty-print.h
12325 (extern debug (gimple_statement_d &)): New.
12326 (extern debug (gimple_statement_d *)): New.
12328 (debug (ira_allocno_copy &)): New.
12329 (debug (ira_allocno_copy *)): New.
12330 (debug (ira_allocno &)): New.
12331 (debug (ira_allocno *)): New.
12333 (extern debug (ira_allocno_copy &)): New.
12334 (extern debug (ira_allocno_copy *)): New.
12335 (extern debug (ira_allocno &)): New.
12336 (extern debug (ira_allocno *)): New.
12338 (debug (live_range &)): New.
12339 (debug (live_range *)): New.
12341 (debug (lra_live_range &)): New.
12342 (debug (lra_live_range *)): New.
12344 (debug (lra_live_range &)): New.
12345 (debug (lra_live_range *)): New.
12347 (debug (omega_pb_d &)): New.
12348 (debug (omega_pb_d *)): New.
12350 (extern debug (omega_pb_d &)): New.
12351 (extern debug (omega_pb_d *)): New.
12353 (debug (const rtx_def &)): New.
12354 (debug (const rtx_def *)): New.
12356 (debug_tree (tree): Move within file.
12357 (debug_raw (const tree_node &)): New.
12358 (debug_raw (const tree_node *)): New.
12359 (dump_tree_via_hooks (const tree_node *, int)): New.
12360 (debug (const tree_node &)): New.
12361 (debug (const tree_node *)): New.
12362 (debug_verbose (const tree_node &)): New.
12363 (debug_verbose (const tree_node *)): New.
12364 (debug_head (const tree_node &)): New.
12365 (debug_head (const tree_node *)): New.
12366 (debug_body (const tree_node &)): New.
12367 (debug_body (const tree_node *)): New.
12368 (debug_vec_tree (tree): Move and reimplement in terms of dump.
12369 (debug (vec<tree, va_gc> &)): New.
12370 (debug (vec<tree, va_gc> *)): New.
12372 (extern debug (const rtx_def &)): New.
12373 (extern debug (const rtx_def *)): New.
12375 (debug_raw (simple_bitmap_def &)): New.
12376 (debug_raw (simple_bitmap_def *)): New.
12377 (debug (simple_bitmap_def &)): New.
12378 (debug (simple_bitmap_def *)): New.
12380 (extern debug (simple_bitmap_def &)): New.
12381 (extern debug (simple_bitmap_def *)): New.
12382 (extern debug_raw (simple_bitmap_def &)): New.
12383 (extern debug_raw (simple_bitmap_def *)): New.
12385 (debug (vinsn_def &)): New.
12386 (debug (vinsn_def *)): New.
12387 (debug_verbose (vinsn_def &)): New.
12388 (debug_verbose (vinsn_def *)): New.
12389 (debug (expr_def &)): New.
12390 (debug (expr_def *)): New.
12391 (debug_verbose (expr_def &)): New.
12392 (debug_verbose (expr_def *)): New.
12393 (debug (vec<rtx> &)): New.
12394 (debug (vec<rtx> *)): New.
12396 (extern debug (vinsn_def &)): New.
12397 (extern debug (vinsn_def *)): New.
12398 (extern debug_verbose (vinsn_def &)): New.
12399 (extern debug_verbose (vinsn_def *)): New.
12400 (extern debug (expr_def &)): New.
12401 (extern debug (expr_def *)): New.
12402 (extern debug_verbose (expr_def &)): New.
12403 (extern debug_verbose (expr_def *)): New.
12404 (extern debug (vec<rtx> &)): New.
12405 (extern debug (vec<rtx> *)): New.
12407 (_list_iter_cond_expr): Make inline instead of static.
12409 (debug (sreal &)): New.
12410 (debug (sreal *)): New.
12412 (extern debug (sreal &)): New.
12413 (extern debug (sreal *)): New.
12415 (extern debug_raw (const tree_node &)): New.
12416 (extern debug_raw (const tree_node *)): New.
12417 (extern debug (const tree_node &)): New.
12418 (extern debug (const tree_node *)): New.
12419 (extern debug_verbose (const tree_node &)): New.
12420 (extern debug_verbose (const tree_node *)): New.
12421 (extern debug_head (const tree_node &)): New.
12422 (extern debug_head (const tree_node *)): New.
12423 (extern debug_body (const tree_node &)): New.
12424 (extern debug_body (const tree_node *)): New.
12425 (extern debug (vec<tree, va_gc> &)): New.
12426 (extern debug (vec<tree, va_gc> *)): New.
12428 (debug (struct loop &)): New.
12429 (debug (struct loop *)): New.
12430 (debug_verbose (struct loop &)): New.
12431 (debug_verbose (struct loop *)): New.
12432 * tree-dump.c: Add header dependence.
12434 (extern debug (struct loop &)): New.
12435 (extern debug (struct loop *)): New.
12436 (extern debug_verbose (struct loop &)): New.
12437 (extern debug_verbose (struct loop *)): New.
12439 (debug (data_reference &)): New.
12440 (debug (data_reference *)): New.
12441 (debug (vec<data_reference_p> &)): New.
12442 (debug (vec<data_reference_p> *)): New.
12443 (debug (vec<ddr_p> &)): New.
12444 (debug (vec<ddr_p> *)): New.
12446 (extern debug (data_reference &)): New.
12447 (extern debug (data_reference *)): New.
12448 (extern debug (vec<data_reference_p> &)): New.
12449 (extern debug (vec<data_reference_p> *)): New.
12450 (extern debug (vec<ddr_p> &)): New.
12451 (extern debug (vec<ddr_p> *)): New.
12453 (debug (pt_solution &)): New.
12454 (debug (pt_solution *)): New.
12456 (extern debug (pt_solution &)): New.
12457 (extern debug (pt_solution *)): New.
12459 (debug (_var_map &)): New.
12460 (debug (_var_map *)): New.
12461 (debug (tree_live_info_d &)): New.
12462 (debug (tree_live_info_d *)): New.
12464 (extern debug (_var_map &)): New.
12465 (extern debug (_var_map *)): New.
12466 (extern debug (tree_live_info_d &)): New.
12467 (extern debug (tree_live_info_d *)): New.
12469 2013-03-28 Jan Hubicka <jh@suse.cz>
12471 * lto-cgraph.c (merge_profile_summaries): Fix overflows.
12473 2013-03-28 Ian Bolton <ian.bolton@arm.com>
12475 * config/aarch64/aarch64.md (aarch64_can_eliminate): Keep frame
12476 record only when desired or required.
12478 2013-03-28 Uros Bizjak <ubizjak@gmail.com>
12480 * config/i386/i386.md (*vec_extract2vdi_1): Merge with
12481 *vec_extractv2di_1_rex64. Use x64 isa attribute.
12483 2013-03-28 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
12485 * config/aarch64/aarch64.md (*and<mode>3_compare0): New pattern.
12486 (*andsi3_compare0_uxtw): New pattern.
12487 (*and_<SHIFT:optab><mode>3_compare0): New pattern.
12488 (*and_<SHIFT:optab>si3_compare0_uxtw): New pattern.
12490 2013-03-28 Jan Hubicka <jh@suse.cz>
12492 * data-streamer-in.c (streamer_read_gcov_count): New function.
12493 * gimple-streamer-out.c: Include value-prof.h.
12494 (output_gimple_stmt): Output histogram.
12495 (output_bb): Use streamer_write_gcov_count.
12496 * value-prof.c: Include data-streamer.h
12497 (dump_histogram_value): Add HIST_TYPE_MAX.
12498 (stream_out_histogram_value): New function.
12499 (stream_in_histogram_value): New function.
12500 * value-prof.h (enum hist_type): Add HIST_TYPE_MAX.
12501 (stream_out_histogram_value, stream_in_histogram_value): Declare.
12502 * data-streamer-out.c (streamer_write_gcov_count): New function.
12503 (streamer_write_gcov_count_stream): New function.
12504 * lto-cgraph.c (lto_output_edge): Update counter streaming.
12505 (lto_output_node): Likewise.
12506 (input_node, input_edge): Likewise.
12507 * lto-streamer-out.c (output_cfg): Update streaming.
12508 * lto-streamer-in.c (input_cfg): Likewise.
12509 * data-streamer.h (streamer_write_gcov_count,
12510 streamer_write_gcov_count_stream, streamer_read_gcov_count): Declare.
12511 * gimple-streamer-in.c: Include value-prof.h
12512 (input_gimple_stmt): Input histograms.
12513 (input_bb): Update profile streaming.
12515 2013-03-28 Kenneth Zadeck <zadeck@naturalbridge.com>
12517 * genmodes.c (emit_max_int): New function.
12518 (emit_insn_modes_h): Added call to emit_max_function.
12519 * doc/rtl.texi (MAX_BITSIZE_MODE_ANY_INT, MAX_BITSIZE_MODE_ANY_MODE):
12521 * machmode.def: Fixed comment.
12523 2013-03-28 Kenneth Zadeck <zadeck@naturalbridge.com>
12525 * combine.c (try_combine): Removed useless assert.
12526 * cselib.c (rtx_equal_for_cselib_1): Removed unnecessary parens.
12528 2013-03-28 Marek Polacek <polacek@redhat.com>
12529 Richard Biener <rguenther@suse.de>
12531 PR tree-optimization/56695
12532 * tree-vect-stmts.c (vectorizable_condition): Unconditionally
12533 build signed result of a vector comparison.
12534 * tree-cfg.c (verify_gimple_comparison): Check that a result
12535 of a vector comparison has signed type.
12537 2013-03-28 Richard Biener <rguenther@suse.de>
12539 PR tree-optimization/37021
12540 * tree-vect-slp.c (vect_build_slp_tree): When not unrolling
12541 do not restrict gaps between groups.
12542 * tree-vect-stmts.c (vectorizable_load): Properly account for
12543 a gap between groups.
12545 2013-03-28 Eric Botcazou <ebotcazou@adacore.com>
12547 * toplev.c (process_options): Do not disable -fomit-frame-pointer on a
12548 general basis if unwind info is requested and ACCUMULATE_OUTGOING_ARGS
12551 2013-03-27 Gerald Pfeifer <gerald@pfeifer.com>
12553 * doc/invoke.texi (AVR Options): Tweak link for AVR-LibC user manual.
12554 * doc/extend.texi (Named Address Spaces): Ditto.
12555 (Variable Attributes): Ditto.
12557 2013-03-27 Kai Tietz <ktietz@redhat.com>
12559 * config.build: Add support for cygwin x64 target.
12560 * config.gcc: Likewise.
12561 * config.host: Likewise.
12562 * configure.ac: Likewise
12563 * configure: Regenerated.
12565 2013-03-27 Kai Tietz <ktietz@redhat.com>
12567 * config/i386/cygwin-stdint.h: Add support for cygwin x64 target.
12568 * config/i386/t-cygwin-w64: New file.
12569 * config/i386/cygwin-w64.h: New file.
12570 * config/i386/cygwin.h (EXTRA_OS_CPP_BUILTINS): Extend
12571 and add support for x64-cygwin target.
12572 (CPP_SPEC): Likewise.
12573 (CXX_WRAP_SPEC_LIST): Undefine before define.
12574 (LIBGCJ_SONAME): Use 15 as version.
12576 2013-03-27 Richard Biener <rguenther@suse.de>
12578 PR tree-optimization/56716
12579 * tree-ssa-structalias.c (perform_var_substitution): Adjust
12580 dumping for ref nodes.
12582 2013-03-27 Martin Jambor <mjambor@suse.cz>
12584 PR tree-optimization/55334
12585 * ipa-cp.c (initialize_node_lattices): Allow IPA-CP through and to
12586 restricted pointers to arrays.
12588 2013-03-27 Gabriel Dos Reis <gdr@integrable-solutions.net>
12590 * Makefile.in (.SUFFIXES): Add .cc.
12591 (.c.o): Apply same recipe for implicit rule .cc.o.
12593 2013-03-27 Richard Biener <rguenther@suse.de>
12595 PR tree-optimization/37021
12596 * tree-vect-data-refs.c (vect_check_strided_load): Allow
12597 REALPART/IMAGPART_EXPRs around the supported refs.
12598 * tree-ssa-structalias.c (find_func_aliases): Assume that
12599 floating-point values are not used to transfer pointers.
12601 2013-03-27 Alexander Ivchenko <alexander.ivchenko@intel.com>
12603 * target.def (TARGET_HAS_IFUNC_P): New target hook.
12604 * doc/tm.texi.in (TARGET_HAS_IFUNC_P): New.
12605 * doc/tm.texi: Regenerate.
12606 * targhooks.h (default_has_ifunc_p): New.
12607 * targhooks.c (default_has_ifunc_p): Ditto.
12608 * config/linux-protos.h: New file.
12609 * config/linux-android.h (TARGET_HAS_IFUNC_P): Using version of
12610 this hook for linux which disables support of indirect functions in
12612 * config/linux-android.c: New file.
12613 * config/t-linux-android.c: Ditto.
12614 * config.gcc: Added new object file linux-android.o.
12615 * config/i386/i386.c (ix86_get_function_versions_dispatcher):
12616 Using TARGET_HAS_IFUNC hook instead of HAVE_GNU_INDIRECT_FUNCTION.
12617 * varasm.c (do_assemble_alias): Likewise.
12618 * configure.ac: Define HAVE_GNU_INDIRECT_FUNCTION as zero if the target
12619 doesn't support indirect functions.
12620 * configure: Regenerate.
12622 2013-03-27 Bin Cheng <bin.cheng@arm.com>
12625 * config/arm/arm.c (thumb1_rtx_costs, thumb1_size_rtx_costs): Fix
12626 rtx costs for SET/ASHIFT/ASHIFTRT/LSHIFTRT/ROTATERT patterns with
12629 2013-03-27 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
12631 * config/s390/s390.h (TARGET_FLT_EVAL_METHOD): Define.
12633 2013-03-27 Terry Guo <terry.guo@arm.com>
12635 * config/arm/arm-cores.def: Added core cortex-r7.
12636 * config/arm/arm-tune.md: Regenerated.
12637 * config/arm/arm-tables.opt: Regenerated.
12638 * doc/invoke.texi: Added entry for core cortex-r7.
12640 2013-03-27 Walter Lee <walt@tilera.com>
12642 * config/tilegx/tilegx.c (tilegx_expand_prologue): Avoid
12643 double-decrement of next_scratch_regno.
12645 2013-03-27 Walter Lee <walt@tilera.com>
12647 * config/tilegx/tilegx.md (insn_v1mulu): Fix predicates on
12649 (insn_v1mulus): Ditto.
12650 (insn_v2muls): Ditto.
12652 2013-03-27 Walter Lee <walt@tilera.com>
12654 * config/tilegx/tilegx.h (ASM_OUTPUT_ADDR_VEC_ELT): Delete extra tab.
12655 (ASM_OUTPUT_ADDR_DIFF_ELT): Ditto.
12657 2013-03-27 Walter Lee <walt@tilera.com>
12659 * config/tilegx/tilegx.md (*sibcall_insn): Fix type atribute for jr.
12660 (*sibcall_value): Ditto.
12662 2013-03-27 Walter Lee <walt@tilera.com>
12664 * config/tilegx/tilegx.md (insn_mnz_<mode>): Replaced by ...
12665 (insn_mnz_v8qi): ... this ...
12666 (insn_mnz_v4hi): ... and this. Replace (const_int 0) with the
12668 (insn_v<n>mnz): Replaced by ...
12669 (insn_v1mnz): ... this ...
12670 (insn_v2mnz): ... and this. Replace (const_int 0) with the vector
12672 (insn_mz_<mode>): Replaced by ...
12673 (insn_mz_v8qi): ... this ...
12674 (insn_mz_v4hi): ... and this. Replace (const_int 0) with the
12676 (insn_v<n>mz): Replaced by ...
12677 (insn_v1mz): ... this ...
12678 (insn_v2mz): ... and this. Replace (const_int 0) with the vector
12681 2013-03-26 Eric Botcazou <ebotcazou@adacore.com>
12683 * doc/invoke.texi (SPARC options): Remove -mlittle-endian.
12685 2013-03-26 Roland McGrath <mcgrathr@google.com>
12687 * config/arm/arm.c (arm_print_operand: case 'w'): Use fputs rather
12688 than fprintf with a non-constant, non-format string.
12690 2013-03-26 Uros Bizjak <ubizjak@gmail.com>
12692 * config/i386/i386.md (*cmpqi_ext_1): Merge with *cmpqi_ext_1_rex64
12693 using nox64 isa attribute. Use nonimmediate_x86nomem_operand as
12694 operand 0 predicate.
12695 (*cmpqi_ext_3): Merge with *cmpqi_ext_3_rex64 using nox64 isa
12696 attribute. Use general_x64nomem_operand as operand 1 predicate.
12697 (*movqi_extv_1): Merge with *movqi_extv_1_rex64 using nox64 isa
12698 attribute. Use nonimmediate_x64nomem_operand as operand 0 predicate.
12699 (*movqi_extzv_2): Merge with *movqi_extzv_2_rex64 using nox64 isa
12700 attribute. Use nonimmediate_x64nomem_operand as operand 0 predicate.
12701 (mov<mode>_insv_1): Remove expander. Merge insn with
12702 movsi_insv_1 using SWI48 mode iterator and nox64 isa attribute.
12703 Use general_x64nomem_operand as operand 1 predicate.
12704 (addqi_ext_1): Merge with *addqi_ext_1_rex64 using nox64 isa attribute.
12705 (*testqi_ext_1): Merge with *testqi_ext_1_rex64 using nox64 isa
12706 attribute. Use nonimmediate_x64nomem_operand as operand 1 predicate.
12707 (*andqi_ext_1): Merge with *andqi_ext_1_rex64 using nox64 isa
12708 attribute. Use nonimmediate_x64nomem_operand as operand 2 predicate.
12709 (*<code>qi_ext_1): Merge with *<code>qi_ext_1_rex64 using nox64 isa
12710 attribute. Use nonimmediate_x64nomem_operand as operand 1 predicate.
12711 (*xorqi_cc_ext_1): Merge with *xorqi_cc_ext_1_rex64 using nox64
12712 isa attribute. Use general_x64nomem_operand as operand 2 predicate.
12713 * config/i386/predicates.md (nonimmediate_x64nomem_operand): New.
12714 (general_x64nomem_operand): Ditto.
12716 2013-03-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
12718 * config/rtems.opt: Add -pthread option.
12720 2013-03-26 Richard Biener <rguenther@suse.de>
12722 * alias.c (find_base_term): Avoid redundant and not used recursion.
12723 (base_alias_check): Get the initial base term from the caller.
12724 (true_dependence_1): Compute and pass base terms to base_alias_check.
12725 (write_dependence_p): Likewise.
12726 (may_alias_p): Likewise.
12728 2013-03-26 Sofiane Naci <sofiane.naci@arm.com>
12730 * config/aarch64/aarch64.c (aarch64_classify_address): Support
12731 PC-relative load in SI modes and above only.
12733 2013-03-26 Xinyu Qi <xyqi@marvell.com>
12735 * config/arm/arm.h (FIRST_IWMMXT_GR_REGNUM): Add comment.
12736 * config/arm/iwmmxt.md (WCGR0): Update.
12737 (WCGR1, WCGR2, WCGR3): Likewise.
12739 2013-03-26 Uros Bizjak <ubizjak@gmail.com>
12741 * config/i386/i386.md (*movdfcc_1): Merge with *movdfcc_1_rex64.
12742 Use x64 and nox64 isa attributes.
12744 2013-03-26 Richard Biener <rguenther@suse.de>
12746 * emit-rtl.c (set_mem_attributes_minus_bitpos): Remove
12747 alignment computations and rely on get_object_alignment_1
12748 for the !TYPE_P case.
12749 Commonize DECL/COMPONENT_REF handling in the ARRAY_REF path.
12751 2013-03-26 Walter Lee <walt@tilera.com>
12753 * config/tilegx/tilegx.h (PROFILE_BEFORE_PROLOGUE): Define.
12754 * config/tilegx/tilepro.h (PROFILE_BEFORE_PROLOGUE): Define.
12756 2013-03-25 Jeff Law <law@redhat.com>
12758 * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Add missing
12759 check for INTEGRAL_TYPE_P that was missing due to checking in
12760 wrong version of prior patch.
12762 2013-03-25 Walter Lee <walt@tilera.com>
12764 * config/tilegx/tilegx-builtins.h (enum tilegx_builtin): Add
12765 TILEGX_INSN_SHUFFLEBYTES1.
12766 * config/tilegx/tilegx.c (tilegx_builtin_info): Add entry for
12768 (tilegx_builtins): Ditto.
12769 * config/tilegx/tilegx.md (insn_shufflebytes1): New pattern.
12771 2013-03-25 Walter Lee <walt@tilera.com>
12773 * config/tilegx/tilegx.md (floatsisf2): New pattern.
12774 (floatunssisf2): New pattern.
12775 (floatsidf2): New pattern.
12776 (floatunssidf2): New pattern.
12778 2013-03-25 Walter Lee <walt@tilera.com>
12780 * config/tilegx/tilegx.c (expand_set_cint64_one_inst): Inline
12781 tests for constraint J, K, N, P.
12783 2013-03-25 Walter Lee <walt@tilera.com>
12785 * config/tilegx/tilegx.c (tilegx_asm_preferred_eh_data_format):
12786 Use indirect/pcrel encoding.
12787 * config/tilepro/tilepro.c (tilepro_asm_preferred_eh_data_format):
12790 2013-03-25 Steve Ellcey <sellcey@mips.com>
12792 * config/mips/mmips-cpus.def (74kc, 74kf2_1, 74kf, 74kf, 74kf1_1,
12793 74kfx, 74kx, 74kf3_2): Add PTF_AVOID_IMADD.
12794 * config/mips/mips.c (mips_option_override): Set IMADD default.
12795 * config/mips/mips.h (PTF_AVOID_IMADD): New.
12796 (ISA_HAS_MADD_MSUB): Remove MIPS16 check.
12797 (GENERATE_MADD_MSUB): Remove TUNE_74K check, add MIPS16 check.
12798 * config/mips/mips.md (mimadd): New flag for integer madd/msub.
12799 * doc/invoke.texi (-mimadd/-mno-imadd): New.
12801 2013-03-25 Jeff Law <law@redhat.com>
12803 * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Rework
12804 slightly to avoid creating and folding useless trees. Simplify
12805 slightly by restricting to INTEGER_CSTs and using int_fits_type_p.
12807 2013-03-25 Uros Bizjak <ubizjak@gmail.com>
12809 * config/i386/i386.md (*zero_extendsidi2): Merge with
12810 *zero_extendsidi2_rex64. Use x64 and nox64 isa attributes.
12811 * config/i386/predicates.md (x86_64_zext_operand): Rename from
12812 x86_64_zext_general_operand. Use nonimmediate_operand on 32bit
12813 targets. Clarify comment.
12815 2013-03-25 Martin Jambor <mjambor@suse.cz>
12817 * ipa-prop.c (ipa_write_jump_function): Stream simple and aritmetic
12818 pass-through jump functions differently.
12819 (ipa_read_jump_function): Likewise. Also use setter functions to set
12822 2013-03-25 Martin Jambor <mjambor@suse.cz>
12824 * ipa-cp.c (ipa_get_indirect_edge_target): Renamed to
12825 ipa_get_indirect_edge_target_1, added parameter agg_reps and ability to
12827 (ipa_get_indirect_edge_target): New function.
12828 (devirtualization_time_bonus): New parameter known_aggs, pass it to
12829 ipa_get_indirect_edge_target. Update all callers.
12830 (ipcp_discover_new_direct_edges): New parameter aggvals. Pass it to
12831 ipa_get_indirect_edge_target_1 instead of calling
12832 ipa_get_indirect_edge_target.
12833 (create_specialized_node): Pass aggvlas to
12834 ipcp_discover_new_direct_edges.
12836 2013-03-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12838 * config/arm/arm.md (f_sels, f_seld): New types.
12839 (*cmov<mode>): New pattern.
12840 * config/arm/predicates.md (arm_vsel_comparison_operator): New
12843 2013-03-25 Kai Tietz <ktietz@redhat.com>
12845 * config/i386/xm-mingw32.h (__USE_MINGW_ANSI_STDIO): Enable
12846 POSIX-printf for mingw-hosted builds.
12848 2013-03-25 Richard Biener <rguenther@suse.de>
12850 PR middle-end/56694
12851 * tree-eh.c (lower_eh_must_not_throw): Strip BLOCKs from the
12852 must-not-throw stmt location.
12854 2013-03-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12856 * config/arm/arm.c (arm_emit_load_exclusive): Add acq parameter.
12857 Emit load-acquire versions when acq is true.
12858 (arm_emit_store_exclusive): Add rel parameter.
12859 Emit store-release versions when rel is true.
12860 (arm_split_compare_and_swap): Use acquire-release instructions
12862 of barriers when appropriate.
12863 (arm_split_atomic_op): Likewise.
12864 * config/arm/arm.h (TARGET_HAVE_LDACQ): New macro.
12865 * config/arm/unspecs.md (VUNSPEC_LAX): New unspec.
12866 (VUNSPEC_SLX): Likewise.
12867 (VUNSPEC_LDA): Likewise.
12868 (VUNSPEC_STL): Likewise.
12869 * config/arm/sync.md (atomic_load<mode>): New pattern.
12870 (atomic_store<mode>): Likewise.
12871 (arm_load_acquire_exclusive<mode>): Likewise.
12872 (arm_load_acquire_exclusivesi): Likewise.
12873 (arm_load_acquire_exclusivedi): Likewise.
12874 (arm_store_release_exclusive<mode>): Likewise.
12876 2013-03-25 Catherine Moore <clm@codesourcery.com>
12878 * config/mips/constraints.md (u, Udb7 Uead, Uean, Uesp, Uib3,
12879 Uuw6, Usb4, ZS, ZT, ZU, ZV, ZW): New constraints.
12880 * config/mip/predicates.md (lwsp_swsp_operand,
12881 lw16_sw16_operand, lhu16_sh16_operand, lbu16_operand,
12882 sb16_operand, db4_operand, db7_operand, ib3_operand,
12883 sb4_operand, ub4_operand, uh4_operand, uw4_operand,
12884 uw5_operand, uw6_operand, addiur2_operand, addiusp_operand,
12885 andi16_operand): New predicates.
12886 * config/mips/mips.md (compression): New attribute.
12887 (enabled): New attribute.
12888 (length): Consider compression in computing length.
12889 (shift_compression): New code attribute.
12890 (*add<mode>3): New operands. Record compression.
12891 (sub<mode>3): Likewise.
12892 (one_cmpl<mode>2): Likewise.
12893 (*and<mode>3): Likewise.
12894 (*ior<mode>3): Likewise.
12895 (unnamed pattern for xor): Likewise.
12896 (*zero_extend<SHORT:mode><GPR:mode>2): Likewise.
12897 (*<optab><mode>3): Likewise.
12898 (*mov<mode>_internal: Likewise.
12899 * config/mips/mips-protos.h (mips_signed_immediate_p): New.
12900 (mips_unsigned_immediate_p): New.
12901 (umips_lwsp_swsp_address_p): New.
12902 (m16_based_address_p): New.
12903 * config/mips/mips-protos.h (mips_signed_immediate_p): New prototype.
12904 (mips_unsigned_immediate_p): New prototype.
12905 (lwsp_swsp_address_p): New prototype.
12906 (m16_based_address_p): New prototype.
12907 * config/mips/mips.c (mips_unsigned_immediate_p): New function.
12908 (mips_signed_immediate_p): New function.
12909 (m16_based_address_p): New function.
12910 (lwsp_swsp_address_p): New function.
12911 (mips_print_operand_punctuation): Recognize short delay slot insns
12912 for microMIPS.add<mode>3"
12914 2013-03-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12917 * config/arm/iterators.md (v_cmp_result): New mode attribute.
12918 * config/arm/neon.md (vcond<mode><mode>): Handle unordered cases.
12920 2013-03-25 Richard Biener <rguenther@suse.de>
12922 PR tree-optimization/56689
12923 * tree-vrp.c (execute_vrp): Mark loops for fixup if we removed
12926 2013-03-25 Richard Biener <rguenther@suse.de>
12928 * tree-ssa-loop-im.c (struct mem_ref): Use bitmap_head instead
12930 (memory_references): Likewise.
12931 (outermost_indep_loop, mem_ref_alloc, mark_ref_stored,
12932 gather_mem_refs_stmt, record_dep_loop, ref_indep_loop_p_1,
12933 ref_indep_loop_p_2, find_refs_for_sm): Adjust.
12934 (gather_mem_refs_in_loops): Fold into ...
12935 (analyze_memory_references): ... this. Move initialization
12936 to tree_ssa_lim_initialize.
12937 (fill_always_executed_in): Rename to ...
12938 (fill_always_executed_in_1): ... this.
12939 (fill_always_executed_in): Move contains_call computation to
12940 this new function from ...
12941 (tree_ssa_lim_initialize): ... here.
12942 (tree_ssa_lim): Call fill_always_executed_in.
12944 2013-03-25 Eric Botcazou <ebotcazou@adacore.com>
12946 * postreload.c (reload_combine): Fix code detecting returns.
12948 2013-03-25 Eric Botcazou <ebotcazou@adacore.com>
12950 * function.c (emit_use_return_register_into_block): On cc0 targets,
12951 do not emit the sequence between cc0 setter and user.
12953 2013-03-25 Kai Tietz <ktietz@redhat.com>
12955 * config/i386/predicates.md (local_symbolic_operand): Interpret
12956 dll-imported symbols as none-local.
12958 2013-03-25 Richard Biener <rguenther@suse.de>
12960 * tree-ssa-loop-im.c (struct depend): Remove.
12961 (struct lim_aux_data): Make depends a vec of gimples.
12962 (free_lim_aux_data): Adjust.
12963 (add_dependency): Likewise.
12964 (set_level): Likewise.
12966 2013-03-25 Richard Biener <rguenther@suse.de>
12968 PR middle-end/56434
12969 * calls.c (expand_call): Use MALLOC_ABI_ALIGNMENT to annotate
12970 the pointer returned by calls with ECF_MALLOC set.
12972 2013-03-24 Uros Bizjak <ubizjak@gmail.com>
12974 * config/i386/mmx.md (mov<mode>): Add ?!Ym,r and r,?!Ym alternatives.
12976 2013-03-24 Uros Bizjak <ubizjak@gmail.com>
12978 * config/i386/mmx.md (mov<mode>): Merge with movv2sf expander
12979 using MMXMODE mode iterator.
12980 (*move<mode>_internal): Merge with *movv2sf_internal and
12981 *movv2sf_internal_rex64 using MMXMODE mode iterator.
12983 2013-03-23 Steven Bosscher <steven@gcc.gnu.org>
12985 * gcse.c (oprs_unchanged_p): Respect flag_gcse_lm.
12986 (record_last_mem_set_info): Likewise.
12988 * df-core.c (rest_of_handle_df_initialize): Use XCNEWVEC instead
12989 of XNEWVEC followed by memset.
12990 (df_worklist_dataflow): Use XNEWVEC instead of xmalloc with a cast.
12992 2013-03-23 Steven Bosscher <steven@gcc.gnu.org>
12994 * config/avr/avr.c, config/bfin/bfin.c, config/c6x/c6x.c,
12995 config/epiphany/epiphany.c, config/frv/frv.c, config/ia64/ia64.c,
12996 config/iq2000/iq2000.c, config/mcore/mcore.c, config/mep/mep.c,
12997 config/mmix/mmix.c, config/pa/pa.c, config/rs6000/rs6000.c,
12998 config/s390/s390.c, config/sparc/sparc.c, config/spu/spu.c,
12999 config/stormy16/stormy16.c, config/v850/v850.c, config/xtensa/xtensa.c,
13000 dwarf2out.c, hw-doloop.c, resource.c, rtl.h : Where applicable, use
13001 the predicates NOTE_P, NONJUMP_INSN_P, JUMP_P, CALL_P, LABEL_P, and
13002 BARRIER_P instead of GET_CODE.
13004 2013-03-23 Eric Botcazou <ebotcazou@adacore.com>
13006 * config/sparc/sparc.c (sparc_emit_probe_stack_range): Fix small
13007 inaccuracy in the probing code.
13009 * config/sparc/sparc.md (ctrapsi4): Add predicate for operand #3.
13010 (ctrapdi4): Likewise.
13012 2013-03-23 Eric Botcazou <ebotcazou@adacore.com>
13014 * calls.c (expand_call): Add missing guard to code handling return
13015 of non-BLKmode structures in MSB.
13016 * function.c (expand_function_end): Likewise.
13018 2013-03-23 Eric Botcazou <ebotcazou@adacore.com>
13020 * combine.c (try_combine): Adjust comment. Do not add the set of
13021 insn #0 if the destination indirectly is set or dies in insn #2.
13022 Tidy up code to distribute a new note.
13024 2013-03-22 Uros Bizjak <ubizjak@gmail.com>
13026 * config/i386/i386.md (*movdi_internal): Set prefix_rex attribute
13027 also for alternatives 16 and 17.
13029 2013-03-22 Uros Bizjak <ubizjak@gmail.com>
13031 * config/i386/sse.md (*mov<mode>_internal): Merge with
13032 *mov<mode>_internal_rex64. Use x64 and nox64 isa attributes.
13033 Emit insn template depending on type attribute. Use
13034 HAVE_AS_IX86_INTERUNIT_MOVQ to handle broken assemblers that require
13035 movd instead of movq mnemonic for interunit moves. Rewrite mode
13036 attribute calculation. Remove unit attribute calculation.
13037 Set prefix attribute to maybe_vex for sselog1 and ssemov types.
13038 Set prefix_data16 attribute for DImode ssemov types.
13039 Use Ym instead of y for SSE-MMX conversion alternatives.
13040 Reorder operand constraints.
13042 2013-03-22 Steven Bosscher <steven@gcc.gnu.org>
13044 * df.h (df_insn_delete): Adjust prototype.
13045 * emit-rtl.c (remove_insn): Pass a basic block to df_insn_delete
13046 and let it decide whether mark the basic block dirty.
13047 (set_insn_deleted): Only pass INSN_P insns to df_insn_delete.
13048 * df-scan.c (df_insn_info_delete): New helper function, split
13049 off from df_insn_delete.
13050 (df_scan_free_bb_info): Use it.
13051 (df_insn_rescan, df_insn_rescan_all, df_process_deferred_rescans):
13053 (df_insn_delete): Likewise. Take insn rtx as argument. Verify
13054 that the insn is actually an insn and it has a non-NULL basic block.
13055 Do not mark basic block dirty if only deleting a DEBUG_INSN.
13057 2013-03-22 Richard Biener <rguenther@suse.de>
13059 * tree-ssa-loop-im.c (struct mem_ref): Remove indep_ref and
13061 (mem_ref_alloc): Do not allocate them.
13062 (refs_independent_p): Do not query or maintain a cache.
13064 2013-03-22 Richard Biener <rguenther@suse.de>
13066 * tree-ssa-loop-im.c (memory_references): Drop all_refs_in_loop.
13067 (gather_mem_refs_in_loops): Do not compute it.
13068 (analyze_memory_references): Do not allocate it.
13069 (tree_ssa_lim_finalize): Do not free it.
13070 (for_all_locs_in_loop): Do not query all_refs_in_loop.
13072 2013-03-22 Richard Biener <rguenther@suse.de>
13074 * is-a.h (as_a): Use gcc_checking_assert.
13076 2013-03-22 Ian Bolton <ian.bolton@arm.com>
13078 * config/aarch64/aarch64.c (aarch64_print_operand): New
13079 format specifier for printing a constant in hex.
13080 * config/aarch64/aarch64.md (insv_imm<mode>): Use the X
13081 format specifier for printing second operand.
13083 2013-03-22 Richard Biener <rguenther@suse.de>
13085 * tree-ssa-loop-im.c (memory_references): Add refs_stored_in_loop
13087 (gather_mem_refs_in_loops): Perform store accumulation here.
13088 (create_vop_ref_mapping_loop): Remove.
13089 (create_vop_ref_mapping): Likewise.
13090 (analyze_memory_references): Initialize refs_stored_in_loop.
13091 (LOOP_DEP_BIT): New define to map to bits in (in)dep_loop bitmaps.
13092 (record_indep_loop): Remove.
13093 (record_dep_loop): New function.
13094 (ref_indep_loop_p_1): Adjust to only walk over references
13095 in the loop, not its subloops.
13096 (ref_indep_loop_p): Rename to ...
13097 (ref_indep_loop_p_2): ... this and recurse over the loop tree,
13098 maintaining a more fine-grained cache.
13099 (ref_indep_loop_p): Wrap ref_indep_loop_p_2.
13100 (tree_ssa_lim_finalize): Free refs_stored_in_loop.
13102 2013-03-22 Richard Biener <rguenther@suse.de>
13104 * tree-ssa-loop-im.c (struct mem_ref_locs): Remove.
13105 (struct mem_ref): Make accesses_in_loop a vec of a vec of
13106 aggregate mem_ref_loc.
13107 (free_mem_ref_locs): Inline into ...
13108 (memref_free): ... this and adjust.
13109 (mem_ref_alloc): Adjust.
13110 (mem_ref_locs_alloc): Remove.
13111 (record_mem_ref_loc): Adjust.
13112 (get_all_locs_in_loop): Rewrite into ...
13113 (for_all_locs_in_loop): ... this iterator.
13114 (rewrite_mem_ref_loc): New functor.
13115 (rewrite_mem_refs): Use for_all_locs_in_loop.
13116 (sm_set_flag_if_changed): New functor.
13117 (execute_sm_if_changed_flag_set): Use for_all_locs_in_loop.
13118 (ref_always_accessed): New functor.
13119 (ref_always_accessed_p): Use for_all_locs_in_loop.
13121 2013-03-21 Marc Glisse <marc.glisse@inria.fr>
13123 * tree-pass.h (PROP_gimple_lvec): New.
13124 * passes.c (dump_properties): Handle PROP_gimple_lvec.
13125 (init_optimization_passes): Move pass_lower_vector.
13126 * tree-vect-generic.c (gate_expand_vector_operations_ssa): Test
13128 (pass_lower_vector): Provide PROP_gimple_lvec.
13129 (pass_lower_vector_ssa): Likewise.
13130 * cfgexpand.c (pass_expand): Require PROP_gimple_lvec.
13132 2013-03-21 Mark Wielaard <mjw@redhat.com>
13134 * dwarf2out.c (size_of_aranges): Skip DECL_IGNORED_P functions.
13136 2013-03-21 Uros Bizjak <ubizjak@gmail.com>
13138 * config/i386/i386.md (*movdi_internal): Disparage slightly
13139 all MMX moves to/from memory. Use Yi instead of x for SSE-MMX
13140 conversion alternatives.
13142 2013-03-21 Jakub Jelinek <jakub@redhat.com>
13144 PR middle-end/48087
13145 * diagnostic.def (DK_WERROR): New kind.
13146 * diagnostic.h (werrorcount): Define.
13147 * diagnostic.c (diagnostic_report_diagnostic): For DK_WARNING
13148 promoted to DK_ERROR, increment DK_WERROR counter instead of
13150 * toplev.c (toplev_main): Call print_ignored_options even if
13151 just werrorcount is non-zero. Exit with FATAL_EXIT_CODE
13152 even if just werrorcount is non-zero.
13155 * dwarf2out.c (tree_add_const_value_attribute): Call ggc_free (array)
13157 (resolve_one_addr): Fail if referenced STRING_CST hasn't been written.
13158 (string_cst_pool_decl): New function.
13159 (optimize_one_addr_into_implicit_ptr): New function.
13160 (resolve_addr_in_expr): Optimize DWARF location expression
13161 DW_OP_addr DW_OP_stack_value where DW_OP_addr refers to some variable
13162 which doesn't live in memory, but has DW_AT_location or
13163 DW_AT_const_value, or refers to a string literal, into
13164 DW_OP_GNU_implicit_pointer.
13165 (optimize_location_into_implicit_ptr): New function.
13166 (resolve_addr): If removing DW_AT_location of a variable because
13167 it was DW_OP_addr of address of the variable, but the variable doesn't
13168 live in memory, try to emit const value attribute for the initializer.
13170 2013-03-21 Marc Glisse <marc.glisse@inria.fr>
13172 * tree.h (VECTOR_TYPE_P): New macro.
13173 (VECTOR_INTEGER_TYPE_P, VECTOR_FLOAT_TYPE_P, FLOAT_TYPE_P,
13174 TYPE_MODE): Use it.
13175 * fold-const.c (fold_cond_expr_with_comparison): Use build_zero_cst.
13176 VEC_COND_EXPR cannot be lvalues.
13177 (fold_ternary_loc) <VEC_COND_EXPR>: Merge with the COND_EXPR case.
13179 2013-03-21 Marc Glisse <marc.glisse@inria.fr>
13181 * simplify-rtx.c (simplify_binary_operation_1) <VEC_CONCAT>:
13182 Restrict the transformation to equal modes.
13184 2013-03-21 Richard Biener <rguenther@suse.de>
13186 PR tree-optimization/39326
13187 * tree-ssa-loop-im.c (UNANALYZABLE_MEM_ID): New define.
13188 (MEM_ANALYZABLE): Adjust.
13189 (record_mem_ref_loc): Move bitmap ops ...
13190 (gather_mem_refs_stmt): ... here. Use the shared mem-ref for
13191 unanalyzable refs, do not record locations for it.
13192 (analyze_memory_references): Allocate ref zero as shared
13194 (refs_independent_p): Do not test for unanalyzed mems here.
13195 (ref_indep_loop_p_1): Special-case disambiguation against
13196 the unanalyzed ref.
13197 (ref_indep_loop_p): Assert we are not queried for the unanalyzed mem.
13199 2013-03-21 Christophe Lyon <christophe.lyon@linaro.org>
13201 * config/arm/arm-protos.h (tune_params): Add
13202 prefer_neon_for_64bits field.
13203 * config/arm/arm.c (prefer_neon_for_64bits): New variable.
13204 (arm_slowmul_tune): Default prefer_neon_for_64bits to false.
13205 (arm_fastmul_tune, arm_strongarm_tune, arm_xscale_tune): Ditto.
13206 (arm_9e_tune, arm_v6t2_tune, arm_cortex_tune): Ditto.
13207 (arm_cortex_a15_tune, arm_cortex_a5_tune): Ditto.
13208 (arm_cortex_a9_tune, arm_v6m_tune, arm_fa726te_tune): Ditto.
13209 (arm_option_override): Handle -mneon-for-64bits new option.
13210 * config/arm/arm.h (TARGET_PREFER_NEON_64BITS): New macro.
13211 (prefer_neon_for_64bits): Declare new variable.
13212 * config/arm/arm.md (arch): Rename neon_onlya8 and neon_nota8 to
13213 avoid_neon_for_64bits and neon_for_64bits. Remove onlya8 and nota8.
13214 (arch_enabled): Handle new arch types. Remove support for onlya8
13216 (one_cmpldi2): Use new arch names.
13217 (zero_extend<mode>di2, extend<mode>di2): Ditto.
13218 * config/arm/arm.opt (mneon-for-64bits): Add option.
13219 * config/arm/neon.md (adddi3_neon, subdi3_neon, iordi3_neon)
13220 (anddi3_neon, xordi3_neon, ashldi3_neon, <shift>di3_neon): Use
13221 neon_for_64bits instead of nota8 and avoid_neon_for_64bits instead
13223 * doc/invoke.texi (-mneon-for-64bits): Document.
13225 2013-03-21 Richard Biener <rguenther@suse.de>
13227 PR tree-optimization/39326
13228 * tree-ssa-loop-im.c (bb_loop_postorder): New global static.
13229 (sort_bbs_in_loop_postorder_cmp): New function.
13230 (gather_mem_refs_in_loops): Assign mem-ref IDs in loop postorder.
13232 2013-03-21 Richard Biener <rguenther@suse.de>
13234 * tree-vect-data-refs.c (vect_update_interleaving_chain): Remove.
13235 (vect_insert_into_interleaving_chain): Likewise.
13236 (vect_drs_dependent_in_basic_block): Inline ...
13237 (vect_slp_analyze_data_ref_dependence): ... here. New function,
13239 (vect_analyze_data_ref_dependence): ... here. Simplify.
13240 (vect_check_interleaving): Simplify.
13241 (vect_analyze_data_ref_dependences): Likewise. Split out ...
13242 (vect_slp_analyze_data_ref_dependences): ... this new function.
13243 (dr_group_sort_cmp): New function.
13244 (vect_analyze_data_ref_accesses): Compute data-reference groups
13245 here instead of in vect_analyze_data_ref_dependence. Use
13246 a more efficient algorithm.
13247 * tree-vect-slp.c (vect_slp_analyze_bb_1): Use
13248 vect_slp_analyze_data_ref_dependences. Call
13249 vect_analyze_data_ref_accesses earlier.
13250 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
13251 * tree-vectorizer.h (vect_analyze_data_ref_dependences): Adjust.
13252 (vect_slp_analyze_data_ref_dependences): New prototype.
13254 2013-03-21 Richard Biener <rguenther@suse.de>
13256 * tree-ssa-loop-im.c (can_sm_ref_p): Do not test whether
13257 ref is stored in the loop.
13258 (find_refs_for_sm): Walk only over all stores.
13259 (store_motion_loop): Allocate from lim_bitmap_obstack.
13260 (store_motion): Likewise.
13262 2013-03-21 Richard Biener <rguenther@suse.de>
13264 * tree-vect-loop-manip.c (slpeel_tree_peel_loop_to_edge):
13265 Update virtual SSA form.
13267 2013-03-21 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
13269 * configure.ac (gcc_cv_ld_eh_frame_ciev3): New test.
13270 * configure: Regenerate.
13271 * config.in: Regenerate.
13272 * config/sol2.c (solaris_override_options): Only enforce DWARF 2
13273 if !HAVE_LD_EH_FRAME_CIEV3.
13275 2013-03-21 Richard Biener <rguenther@suse.de>
13277 * tree-cfg.c (verify_expr_no_block): New function.
13278 (verify_expr_location_1): Verify that neither DECL_DEBUG_EXPR
13279 nor DECL_VALUE_EXPR have locations with associated blocks.
13280 * tree-ssa-live.c (clear_unused_block_pointer_1): Remove.
13281 (clear_unused_block_pointer): Remove code dealing with
13282 blocks in DECL_DEBUG_EXPR locations.
13284 2013-03-21 Richard Biener <rguenther@suse.de>
13286 * tree.h (DECL_DEBUG_EXPR_IS_FROM): Rename to ...
13287 (DECL_HAS_DEBUG_EXPR_P): ... this. Guard properly.
13288 * tree.c (copy_node_stat): Do not copy DECL_HAS_DEBUG_EXPR_P.
13289 * dwarf2out.c (add_var_loc_to_decl): Use DECL_HAS_DEBUG_EXPR_P
13290 instead of DECL_DEBUG_EXPR_IS_FROM.
13291 * gimplify.c (gimplify_modify_expr): Likewise.
13292 * tree-cfg.c (verify_expr_location_1): Likewise.
13293 * tree-complex.c (create_one_component_var): Likewise.
13294 * tree-sra.c (create_access_replacement): Likewise.
13295 * tree-ssa-live.c (clear_unused_block_pointer_1): Likewise.
13296 (clear_unused_block_pointer): Likewise.
13297 * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Likewise.
13298 * tree-streamer-out.c (pack_ts_decl_common_value_fields): Likewise.
13299 * var-tracking.c (var_debug_decl): Likewise.
13300 (track_expr_p): Likewise.
13301 * tree-inline.c (add_local_variables): Likewise. Set
13302 DECL_HAS_DEBUG_EXPR_P after copying it.
13303 * tree-diagnostic.c (default_tree_printer): Use DECL_HAS_DEBUG_EXPR_P
13304 instead of DECL_DEBUG_EXPR_IS_FROM. Guard properly.
13306 2013-03-21 Uros Bizjak <ubizjak@gmail.com>
13309 * configure.ac (HAVE_AS_IX86_INTERUNIT_MOVQ): New test.
13310 * configure: Regenerate.
13311 * config.in: Regenerate.
13312 * config/i386/i386.md (*movdf_internal): Use
13313 HAVE_AS_IX86_INTERUNIT_MOVQ to handle broken assemblers that require
13314 movd instead of movq mnemonic for interunit moves.
13315 (*movdi_internal): Ditto.
13317 2013-03-21 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
13319 * config/aarch64/aarch64-simd.md (simd_fabd): New Attribute.
13320 (abd<mode>_3): New pattern.
13321 (aba<mode>_3): New pattern.
13322 (fabd<mode>_3): New pattern.
13324 2013-03-21 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
13326 * config/aarch64/aarch64-elf.h (REGISTER_PREFIX): Remove.
13327 * config/aarch64/aarch64.c (aarch64_print_operand): Remove all
13328 occurrence of REGISTER_PREFIX as its empty string.
13330 2013-03-20 Jeff Law <law@redhat.com>
13332 * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Record
13333 addititional equivalences for equality comparisons between an SSA_NAME
13334 and a constant where the SSA_NAME was set from a widening conversion.
13336 2013-03-20 Walter Lee <walt@tilera.com>
13338 * config/tilegx/sync.md (atomic_test_and_set): New pattern.
13340 2013-03-20 Uros Bizjak <ubizjak@gmail.com>
13342 * config/i386/i386.md (*movoi_internal_avx): Emit insn template
13343 depending on type attribute.
13344 (*movti_internal): Ditto.
13345 (*movtf_internal): Ditto.
13346 (*movxf_internal): Ditto.
13347 (*movdf_internal): Ditto.
13348 (*movsf_internal): Ditto.
13350 2013-03-20 Uros Bizjak <ubizjak@gmail.com>
13352 * config/i386/i386.md (*movti_internal): Set prefix attribute to
13353 maybe_vex for sselog1 and ssemov types.
13354 (*movdi_internal): Reorder operand constraints.
13355 (*movsi_internal): Ditto. Set prefix attribute to
13356 maybe_vex for sselog1 and ssemov types.
13357 (*movtf_internal): Set prefix attribute to maybe_vex
13358 for sselog1 and ssemov types.
13359 (*movdf_internal): Ditto. Set prefix_data16 attribute for
13360 DImode ssemov types. Reorder operand constraints.
13361 (*movsf_internal): Set type of alternatives 3,4 to imov. Set prefix
13362 attribute to maybe_vex for sselog1 and ssemov types. Set prefix_data16
13363 attribute for SImode ssemov types. Reorder operand constraints.
13365 2013-03-20 Martin Jambor <mjambor@suse.cz>
13367 * params.def (PARAM_IPA_CP_ARRAY_INDEX_HINT_BONUS): New parameter.
13368 * ipa-cp.c (hint_time_bonus): Add abonus for known array indices.
13370 2013-03-20 Pat Haugen <pthaugen@us.ibm.com>
13372 * config/rs6000/predicates.md (indexed_address, update_address_mem
13373 update_indexed_address_mem): New predicates.
13374 * config/rs6000/vsx.md (vsx_extract_<mode>_zero): Set correct "type"
13375 attribute for load/store instructions.
13376 * config/rs6000/dfp.md (movsd_store): Likewise.
13377 (movsd_load): Likewise.
13378 * config/rs6000/rs6000.md (zero_extend<mode>di2_internal1): Likewise.
13379 (unnamed HI->DI extend define_insn): Likewise.
13380 (unnamed SI->DI extend define_insn): Likewise.
13381 (unnamed QI->SI extend define_insn): Likewise.
13382 (unnamed QI->HI extend define_insn): Likewise.
13383 (unnamed HI->SI extend define_insn): Likewise.
13384 (unnamed HI->SI extend define_insn): Likewise.
13385 (extendsfdf2_fpr): Likewise.
13386 (movsi_internal1): Likewise.
13387 (movsi_internal1_single): Likewise.
13388 (movhi_internal): Likewise.
13389 (movqi_internal): Likewise.
13390 (movcc_internal1): Correct mnemonic for stw insn. Set correct "type"
13391 attribute for load/store instructions.
13392 (mov<mode>_hardfloat): Set correct "type" attribute for load/store
13394 (mov<mode>_softfloat): Likewise.
13395 (mov<mode>_hardfloat32): Likewise.
13396 (mov<mode>_hardfloat64): Likewise.
13397 (mov<mode>_softfloat64): Likewise.
13398 (movdi_internal32): Likewise.
13399 (movdi_internal64): Likewise.
13400 (probe_stack_<mode>): Likewise.
13402 2013-03-20 Michael Meissner <meissner@linux.vnet.ibm.com>
13404 * config/rs6000/vector.md (VEC_R): Add 32-bit integer, binary
13405 floating point, and decimal floating point to reload iterator.
13407 * config/rs6000/constraints.md (wl constraint): New constraints to
13408 return FLOAT_REGS if certain options are used to reduce the number
13409 of separate patterns that exist in the file.
13410 (wx constraint): Likewise.
13411 (wz constraint): Likewise.
13413 * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
13414 -mdebug=reg, print wg, wl, wx, and wz constraints.
13415 (rs6000_init_hard_regno_mode_ok): Initialize new constraints.
13416 Initialize the reload functions for 64-bit binary/decimal floating
13418 (reg_offset_addressing_ok_p): If we are on a power7 or later, use
13419 LFIWZX and STFIWX to load/store 32-bit decimal types, and don't
13420 create the buffer on the stack to overcome not having a 32-bit
13422 (rs6000_emit_move): Likewise.
13423 (rs6000_secondary_memory_needed_rtx): Likewise.
13424 (rs6000_alloc_sdmode_stack_slot): Likewise.
13425 (rs6000_preferred_reload_class): On VSX, we can create SFmode 0.0f
13426 via xxlxor, just like DFmode 0.0.
13428 * config/rs6000/rs6000.h (TARGET_NO_SDMODE_STACK): New macro,
13429 define as 1 if we are running on a power7 or newer.
13430 (enum r6000_reg_class_enum): Add new constraints.
13432 * config/rs6000/dfp.md (movsd): Delete, combine with binary
13433 floating point moves in rs6000.md. Combine power6x (mfpgpr) moves
13434 with other moves by using conditional constraits (wg). Use LFIWZX
13435 and STFIWX for loading SDmode on power7. Use xxlxor to create 0.0f.
13436 (movsd splitter): Likewise.
13437 (movsd_hardfloat): Likewise.
13438 (movsd_softfloat): Likewise.
13440 * config/rs6000/rs6000.md (FMOVE32): New iterators to combine
13441 binary and decimal floating point moves.
13442 (fmove_ok): New attributes to combine binary and decimal floating
13443 point moves, and to combine power6x (mfpgpr) moves along normal
13445 (real_value_to_target): Likewise.
13446 (f32_lr): Likewise.
13447 (f32_lm): Likewise.
13448 (f32_li): Likewise.
13449 (f32_sr): Likewise.
13450 (f32_sm): Likewise.
13451 (f32_si): Likewise.
13452 (movsf): Combine binary and decimal floating point moves. Combine
13453 power6x (mfpgpr) moves with other moves by using conditional
13454 constraits (wg). Use LFIWZX and STFIWX for loading SDmode on power7.
13455 (mov<mode> for SFmode/SDmode); Likewise.
13456 (SFmode/SDmode splitters): Likewise.
13457 (movsf_hardfloat): Likewise.
13458 (mov<mode>_hardfloat for SFmode/SDmode): Likewise.
13459 (movsf_softfloat): Likewise.
13460 (mov<mode>_softfloat for SFmode/SDmode): Likewise.
13462 * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wl,
13463 wx and wz constraints.
13465 * config/rs6000/constraints.md (wg constraint): New constraint to
13466 return FLOAT_REGS if -mmfpgpr (power6x) was used.
13468 * config/rs6000/rs6000.h (enum r6000_reg_class_enum): Add wg
13471 * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
13472 -mdebug=reg, print wg, wl, wx, and wz constraints.
13473 (rs6000_init_hard_regno_mode_ok): Initialize new constraints.
13474 Initialize the reload functions for 64-bit binary/decimal floating
13476 (reg_offset_addressing_ok_p): If we are on a power7 or later, use
13477 LFIWZX and STFIWX to load/store 32-bit decimal types, and don't
13478 create the buffer on the stack to overcome not having a 32-bit
13480 (rs6000_emit_move): Likewise.
13481 (rs6000_secondary_memory_needed_rtx): Likewise.
13482 (rs6000_alloc_sdmode_stack_slot): Likewise.
13483 (rs6000_preferred_reload_class): On VSX, we can create SFmode 0.0f
13484 via xxlxor, just like DFmode 0.0.
13486 * config/rs6000/dfp.md (movdd): Delete, combine with binary
13487 floating point moves in rs6000.md. Combine power6x (mfpgpr) moves
13488 with other moves by using conditional constraits (wg). Use LFIWZX
13489 and STFIWX for loading SDmode on power7.
13490 (movdd splitters): Likewise.
13491 (movdd_hardfloat32): Likewise.
13492 (movdd_softfloat32): Likewise.
13493 (movdd_hardfloat64_mfpgpr): Likewise.
13494 (movdd_hardfloat64): Likewise.
13495 (movdd_softfloat64): Likewise.
13497 * config/rs6000/rs6000.md (FMOVE64): New iterators to combine
13498 64-bit binary and decimal floating point moves.
13499 (FMOVE64X): Likewise.
13500 (movdf): Combine 64-bit binary and decimal floating point moves.
13501 Combine power6x (mfpgpr) moves with other moves by using
13502 conditional constraits (wg).
13503 (mov<mode> for DFmode/DDmode): Likewise.
13504 (DFmode/DDmode splitters): Likewise.
13505 (movdf_hardfloat32): Likewise.
13506 (mov<mode>_hardfloat32 for DFmode/DDmode): Likewise.
13507 (movdf_softfloat32): Likewise.
13508 (movdf_hardfloat64_mfpgpr): Likewise.
13509 (movdf_hardfloat64): Likewise.
13510 (mov<mode>_hardfloat64 for DFmode/DDmode): Likewise.
13511 (movdf_softfloat64): Likewise.
13512 (mov<mode>_softfloat64 for DFmode/DDmode): Likewise.
13513 (reload_<mode>_load): Move to later in the file so they aren't in
13514 the middle of the floating point move insns.
13515 (reload_<mode>_store): Likewise.
13517 * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wg
13520 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print out wg
13521 constraint if -mdebug=reg.
13522 (rs6000_initi_hard_regno_mode_ok): Enable wg constraint if -mfpgpr.
13523 Enable using dd reload support if needed.
13525 * config/rs6000/dfp.md (movtd): Delete, combine with 128-bit
13526 binary and decimal floating point moves in rs6000.md.
13527 (movtd_internal): Likewise.
13529 * config/rs6000/rs6000.md (FMOVE128): Combine 128-bit binary and
13530 decimal floating point moves.
13532 (movtf_internal): Likewise.
13533 (mov<mode>_internal, TDmode/TFmode): Likewise.
13534 (movtf_softfloat): Likewise.
13535 (mov<mode>_softfloat, TDmode/TFmode): Likewise.
13537 * config/rs6000/rs6000.md (movdi_mfpgpr): Delete, combine with
13538 movdi_internal64, using wg constraint for move direct operations.
13539 (movdi_internal64): Likewise.
13541 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print
13542 MODES_TIEABLE_P for selected modes. Print the numerical value of
13543 the various virtual registers. Use GPR/FPR first/last values,
13544 instead of hard coding the register numbers. Print which modes
13545 have reload functions registered.
13546 (rs6000_option_override_internal): If -mdebug=reg, trace the options
13547 settings before/after setting cpu, target and subtarget settings.
13548 (rs6000_secondary_reload_trace): Improve the RTL dump for -mdebug=addr
13549 and for secondary reload failures in rs6000_secondary_reload_inner.
13550 (rs6000_secondary_reload_fail): Likewise.
13551 (rs6000_secondary_reload_inner): Likewise.
13553 * config/rs6000/rs6000.md (FIRST_GPR_REGNO): Add convenience
13554 macros for first/last GPR and FPR registers.
13555 (LAST_GPR_REGNO): Likewise.
13556 (FIRST_FPR_REGNO): Likewise.
13557 (LAST_FPR_REGNO): Likewise.
13559 * config/rs6000/vector.md (mul<mode>3): Use the combined macro
13560 VECTOR_UNIT_ALTIVEC_OR_VSX_P instead of separate calls to
13561 VECTOR_UNIT_ALTIVEC_P and VECTOR_UNIT_VSX_P.
13562 (vcond<mode><mode>): Likewise.
13563 (vcondu<mode><mode>): Likewise.
13564 (vector_gtu<mode>): Likewise.
13565 (vector_gte<mode>): Likewise.
13566 (xor<mode>3): Don't allow logical operations on TImode in 32-bit
13567 to prevent the compiler from converting DImode operations to TImode.
13568 (ior<mode>3): Likewise.
13569 (and<mode>3): Likewise.
13570 (one_cmpl<mode>2): Likewise.
13571 (nor<mode>3): Likewise.
13572 (andc<mode>3): Likewise.
13574 * config/rs6000/constraints.md (wt constraint): New constraint
13575 that returns VSX_REGS if TImode is allowed in VSX registers.
13577 * config/rs6000/predicates.md (easy_fp_constant): 0.0f is an easy
13578 constant under VSX.
13580 * config/rs6000/rs6000-modes.def (PTImode): Define, PTImode is
13581 similar to TImode, but it is restricted to being in the GPRs.
13583 * config/rs6000/rs6000.opt (-mvsx-timode): New switch to allow
13584 TImode to occupy a single VSX register.
13586 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Default to
13587 -mvsx-timode for power7/power8.
13588 (power7 cpu): Likewise.
13589 (power8 cpu): Likewise.
13591 * config/rs6000/rs6000.c (rs6000_hard_regno_nregs_internal): Make
13592 sure that TFmode/TDmode take up two registers if they are ever
13593 allowed in the upper VSX registers.
13594 (rs6000_hard_regno_mode_ok): If -mvsx-timode, allow TImode in VSX
13596 (rs6000_init_hard_regno_mode_ok): Likewise.
13597 (rs6000_debug_reg_global): Add debugging for PTImode and wt
13598 constraint. Print if LRA is turned on.
13599 (rs6000_option_override_internal): Give an error if -mvsx-timode
13600 and VSX is not enabled.
13601 (invalid_e500_subreg): Handle PTImode, restricting it to GPRs. If
13602 -mvsx-timode, restrict TImode to reg+reg addressing, and PTImode
13603 to reg+offset addressing. Use PTImode when checking offset
13604 addresses for validity.
13605 (reg_offset_addressing_ok_p): Likewise.
13606 (rs6000_legitimate_offset_address_p): Likewise.
13607 (rs6000_legitimize_address): Likewise.
13608 (rs6000_legitimize_reload_address): Likewise.
13609 (rs6000_legitimate_address_p): Likewise.
13610 (rs6000_eliminate_indexed_memrefs): Likewise.
13611 (rs6000_emit_move): Likewise.
13612 (rs6000_secondary_reload): Likewise.
13613 (rs6000_secondary_reload_inner): Handle PTImode. Allow 64-bit
13614 reloads to fpr registers to continue to use reg+offset addressing,
13615 but 64-bit reloads to altivec registers need reg+reg addressing.
13616 Drop test for PRE_MODIFY, since VSX loads/stores no longer support
13617 it. Treat LO_SUM like a PLUS operation.
13618 (rs6000_secondary_reload_class): If type is 64-bit, prefer to use
13619 FLOAT_REGS instead of VSX_RGS to allow use of reg+offset addressing.
13620 (rs6000_cannot_change_mode_class): Do not allow TImode in VSX
13621 registers to share a register with a smaller sized type, since VSX
13622 puts scalars in the upper 64-bits.
13623 (print_operand): Add support for PTImode.
13624 (rs6000_register_move_cost): Use VECTOR_MEM_VSX_P instead of
13625 VECTOR_UNIT_VSX_P to catch types that can be loaded in VSX
13626 registers, but don't have arithmetic support.
13627 (rs6000_memory_move_cost): Add test for VSX.
13628 (rs6000_opt_masks): Add -mvsx-timode.
13630 * config/rs6000/vsx.md (VSm): Change to use 64-bit aligned moves
13633 (VSr): Use wt constraint for TImode.
13634 (VSv): Drop TImode support.
13635 (vsx_movti): Delete, replace with versions for 32-bit and 64-bit.
13636 (vsx_movti_64bit): Likewise.
13637 (vsx_movti_32bit): Likewise.
13638 (vec_store_<mode>): Use VSX iterator instead of vector iterator.
13639 (vsx_and<mode>3): Delete use of '?' constraint on inputs, just put
13640 one '?' on the appropriate output constraint. Do not allow TImode
13641 logical operations on 32-bit systems.
13642 (vsx_ior<mode>3): Likewise.
13643 (vsx_xor<mode>3): Likewise.
13644 (vsx_one_cmpl<mode>2): Likewise.
13645 (vsx_nor<mode>3): Likewise.
13646 (vsx_andc<mode>3): Likewise.
13647 (vsx_concat_<mode>): Likewise.
13648 (vsx_xxpermdi_<mode>): Fix thinko for non V2DF/V2DI modes.
13650 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Map from
13651 OPTION_MASK_VSX_TIMODE.
13652 (enum rs6000_reg_class_enum): Add RS6000_CONSTRAINT_wt.
13653 (STACK_SAVEAREA_MODE): Use PTImode instead of TImode.
13655 * config/rs6000/rs6000.md (INT mode attribute): Add PTImode.
13656 (TI2 iterator): New iterator for TImode, PTImode.
13657 (wd mode attribute): Add values for vector types.
13658 (movti_string): Replace TI move operations with operations for TImode
13659 and PTImode. Add support for TImode being allowed in VSX registers.
13660 (mov<mode>_string, TImode/PTImode): Likewise.
13661 (movti_ppc64): Likewise.
13662 (mov<mode>_ppc64, TImode/PTImode): Likewise.
13663 (TI mode splitters): Likewise.
13665 * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wt
13668 2013-03-20 Marc Glisse <marc.glisse@inria.fr>
13670 PR tree-optimization/56355
13671 * fold-const.c (tree_binary_nonnegative_warnv_p) <MULT_EXPR>:
13672 Also handle integers with undefined overflow.
13674 2013-03-20 Catherine Moore <clm@codesourcery.com>
13675 Maciej W. Rozycki <macro@codesourcery.com>
13676 Tom de Vries <tom@codesourcery.com>
13677 Nathan Sidwell <nathan@codesourcery.com>
13678 Iain Sandoe <iain@codesourcery.com>
13679 Nathan Froyd <froydnj@codesourcery.com>
13680 Chao-ying Fu <fu@mips.com>
13682 * doc/extend.texi: (micromips, nomicromips, nocompression):
13683 Document new function attributes.
13684 * doc/invoke.texi (minterlink-compressed, mmicromips,
13685 m14k, m14ke, m14kec): Document new options.
13686 (minterlink-mips16): Update documentation.
13687 * doc/md.texi (ZC, ZD): Document new constraints.
13688 * configure.ac (gcc_cv_as_micromips): Check if linker
13689 supports the .set micromips directive.
13690 * configure: Regenerate.
13691 * config.in: Regenerate.
13692 * config/mips/mips-tables.opt: Regenerate.
13693 * config/mips/micromips.md: New file.
13694 * constraints.md (ZC, ZD): New constraints.
13695 * config/mips/predicates.md (movep_src_register): New predicate.
13696 (movep_src_operand): New predicate.
13697 (non_volatile_mem_operand): New predicate.
13698 * config/mips/mips.md (multimem): New type.
13699 (length): Differentiate between 17-bit and 18-bit branch offsets.
13700 (MOVEP1, MOVEP2): New mode iterator.
13701 (mov_<load>l): Use ZC constraint.
13702 (mov_<load>r): Likewise.
13703 (mov_<store>l): Likewise.
13704 (mov_<store>r): Likewise.
13705 (*branch_equality<mode>_inverted): Add microMIPS support.
13706 (*branch_equality<mode>): Likewise.
13707 (*jump_absolute): Likewise.
13708 (indirect_jump_<mode>): Likewise.
13709 (tablejump_<mode>): Likewise.
13710 (<optab>_internal): Likewise.
13711 (sibcall_internal): Likewise.
13712 (sibcall_value_internal): Likewise.
13713 (prefetch): Use constraint ZD.
13714 * config/mips/mips.opt (minterlink-compressed): New option.
13715 (minterlink-mips16): Now an alias for minterlink-compressed.
13716 (mmicromips): New option.
13717 * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
13718 (compare_and_swap_12): Likewise.
13719 (sync_add<mode>): Likewise.
13720 (sync_<optab>_12): Likewise.
13721 (sync_old_<optab>_12): Likewise.
13722 (sync_new_<optab>_12): Likewise.
13723 (sync_nand_12): Likewise.
13724 (sync_old_nand_12): Likewise.
13725 (sync_new_nand_12): Likewise.
13726 (sync_sub<mode>): Likewise.
13727 (sync_old_add<mode>): Likewise.
13728 (sync_old_sub<mode>): Likewise.
13729 (sync_new_add<mode>): Likewise.
13730 (sync_new_sub<mode>): Likewise.
13731 (sync_<optab><mode>): Likewise.
13732 (sync_old_<optab><mode>): Likewise.
13733 (sync_new_<optab><mode>): Likewise.
13734 (sync_nand<mode>): Likewise.
13735 (sync_old_nand<mode>): Likewise.
13736 (sync_new_nand<mode>): Likewise.
13737 (sync_lock_test_and_set<mode>): Likewise.
13738 (test_and_set_12): Likewise.
13739 (atomic_compare_and_swap<mode>): Likewise.
13740 (atomic_exchange<mode>_llsc): Likewise.
13741 (atomic_fetch_add<mode>_llsc): Likewise.
13742 * config/mips/mips-cpus.def (m14kc, m14k): New processors.
13743 * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
13744 (umips_save_restore_pattern_p): Likewise.
13745 (umips_load_store_pair_p): Likewise.
13746 (umips_output_load_store_pair): Likewise.
13747 (umips_movep_target_p): Likewise.
13748 (umips_12bit_offset_address_p): Likewise.
13749 * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
13750 (mips_base_mips16): Rename this...
13751 (mips_base_compression_flags): ...to this. Update all uses.
13752 (mips_attribute_table): Add micromips, nomicromips and nocompression.
13753 (mips_mips16_decl_p): Delete.
13754 (mips_nomips16_decl_p): Delete.
13755 (mips_get_compress_on_flags): New function.
13756 (mips_get_compress_off_flags): New function.
13757 (mips_get_compress_mode): New function.
13758 (mips_get_compress_on_name): New function.
13759 (mips_get_compress_off_name): New function.
13760 (mips_insert_attributes): Support multiple compression types.
13761 (mips_merge_decl_attributes): Likewise.
13762 (umips_12bit_offset_address_p): New function.
13763 (mips_start_function_definition): Emit .set micromips directive.
13764 (mips_call_may_need_jalx_p): New function.
13765 (mips_function_ok_for_sibcall): Add microMIPS support.
13766 (mips_print_operand_punctuation): Support short delay slots and
13768 (umips_swm_mask, umips_swm_encoding): New.
13769 (umips_build_save_restore): New function.
13770 (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
13771 (was_mips16_p): Remove.
13772 (old_compression_mode): New.
13773 (mips_set_compression_mode): New function.
13774 (mips_set_current_function): Add microMIPS support.
13775 (mips_option_override): Likewise.
13776 (umips_save_restore_pattern_p): New function.
13777 (umips_output_save_restore): New function.
13778 (umips_load_store_pair_p_1): New function.
13779 (umips_load_store_pair_p): New function.
13780 (umips_output_load_store_pair_1): New function.
13781 (umips_output_load_store_pair): New function.
13782 (umips_movep_target_p) New function.
13783 (mips_prepare_pch_save): Add microMIPS support.
13784 * config/mips/mips.h (TARGET_COMPRESSION): New.
13785 (TARGET_CPU_CPP_BUILTINS): Update macro
13786 to use new compression flags and to support microMIPS.
13787 (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
13788 (MIPS_ARCH_FLOAT_SPEC): Likewise.
13789 (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
13790 (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
13791 (ASM_SPEC): Support mmicromips and mno-micromips.
13792 (M16STORE_REG_P): New macro.
13793 (MIPS_CALL): Support TARGET_MICROMIPS.
13794 (MICROMIPS_J): New macro.
13795 (mips_base_mips16): Rename this...
13796 (mips_base_compression_flags): ...to this.
13797 (UMIPS_12BIT_OFFSET_P): New macro.
13798 * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
13799 (MULTILIB_DIRNAMES): Likewise.
13800 2013-03-20 Richard Biener <rguenther@suse.de>
13802 PR tree-optimization/56661
13803 * tree-ssa-sccvn.c (visit_use): Only value-number calls if
13804 the result does not have to be distinct.
13806 2013-03-20 Richard Biener <rguenther@suse.de>
13808 * tree-inline.c (copy_tree_body_r): Sync MEM_REF code with
13811 2013-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13812 Steven Bosscher <steven@gcc.gnu.org>
13814 PR rtl-optimization/56605
13815 * loop-iv.c (implies_p): Handle equal RTXs and subregs.
13817 2013-03-20 Uros Bizjak <ubizjak@gmail.com>
13820 * config/i386/i386.md (*movdi_internal): Handle broken assemblers
13821 that require movd instead of movq.
13823 2013-03-20 Richard Biener <rguenther@suse.de>
13825 * tree-ssa-structalias.c (struct variable_info): Add pointer
13826 to the first field of an aggregate with sub-vars. Make
13827 this and the pointer to the next subfield its ID.
13828 (vi_next): New function.
13829 (nothing_id, anything_id, readonly_id, escaped_id, nonlocal_id,
13830 storedanything_id, integer_id): Increment by one.
13831 (new_var_info, get_call_vi, lookup_call_clobber_vi,
13832 get_call_clobber_vi): Adjust.
13833 (solution_set_expand): Simplify and speedup.
13834 (solution_set_add): Inline into ...
13835 (set_union_with_increment): ... this. Adjust accordingly.
13836 (do_sd_constraint): Likewise.
13837 (do_ds_constraint): Likewise.
13838 (do_complex_constraint): Simplify.
13839 (build_pred_graph): Adjust.
13840 (solve_graph): Likewise. Simplify and speedup.
13841 (get_constraint_for_ssa_var, get_constraint_for_ptr_offset,
13842 get_constraint_for_component_ref, get_constraint_for_1,
13843 first_vi_for_offset, first_or_preceding_vi_for_offset,
13844 create_function_info_for, create_variable_info_for_1,
13845 create_variable_info_for, intra_create_variable_infos): Adjust.
13846 (init_base_vars): Push NULL for ID zero.
13847 (compute_points_to_sets): Adjust.
13849 2013-03-20 Richard Biener <rguenther@suse.de>
13851 * cfgloop.c (verify_loop_structure): Streamline and avoid
13852 ICEing on corrupt loop tree.
13853 * graph.c (draw_cfg_nodes_for_loop): Avoid ICEing on corrupt
13856 2013-03-20 Richard Biener <rguenther@suse.de>
13858 * tree-vect-loop-manip.c (slpeel_can_duplicate_loop_p): Do not
13859 check whether an SSA update is needed.
13861 2013-03-20 Richard Sandiford <rdsandiford@googlemail.com>
13863 * config/mips/constraints.md (T): Rename to...
13867 * config/mips/mips.md (*movdi_64bit, *movdi_64bit_mips16)
13868 (*mov<mode>_internal, *mov<mode>_mips16): Update accordingly.
13870 2013-03-19 Ian Bolton <ian.bolton@arm.com>
13872 * config/aarch64/aarch64.md (*sub<mode>3_carryin): New pattern.
13873 (*subsi3_carryin_uxtw): Likewise.
13875 2013-03-19 Ian Bolton <ian.bolton@arm.com>
13877 * config/aarch64/aarch64.md (*ror<mode>3_insn): New pattern.
13878 (*rorsi3_insn_uxtw): Likewise.
13880 2013-03-19 Ian Bolton <ian.bolton@arm.com>
13882 * config/aarch64/aarch64.md (*extr<mode>5_insn): New pattern.
13883 (*extrsi5_insn_uxtw): Likewise.
13885 2013-03-19 Richard Biener <rguenther@suse.de>
13887 PR tree-optimization/56273
13888 * passes.c (init_optimization_passes): Move second VRP after DOM.
13890 2013-03-19 Uros Bizjak <ubizjak@gmail.com>
13892 * config/i386/i386.md (*movti_internal): Merge from
13893 *movti_internal_rex64 and *movti_internal_sse. Use x64 isa attribute.
13894 (*movdi_internal): Merge with *movdi_internal_rex64. Use x64 and
13895 nox64 isa attributes.
13897 2013-03-18 Richard Biener <rguenther@suse.de>
13899 * tree-ssa-structalias.c (find): Use gcc_checking_assert.
13901 (merge_node_constraints): Likewise.
13902 (build_succ_graph): Likewise.
13903 (valid_graph_edge): Inline into single caller.
13904 (unify_nodes): Likewise. Use bitmap_set_bit return value
13906 (scc_visit): Fix formatting and variable use.
13907 (do_sd_constraint): Use gcc_checking_assert.
13908 (do_ds_constraint): Likewise.
13909 (do_complex_constraint): Likewise.
13910 (condense_visit): Likewise. Cleanup.
13911 (dump_pred_graph): New function.
13912 (perform_var_substitution): Dump the pred-graph before
13913 variable substitution.
13914 (find_equivalent_node): Use gcc_checking_assert.
13915 (rewrite_constraints): Guard checking loop with ENABLE_CHECKING.
13917 2013-03-18 Richard Biener <rguenther@suse.de>
13919 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
13920 Remove cond_expr_stmt_list argument and do not gimplify the
13922 (vect_loop_versioning): Adjust.
13923 * tree-vect-data-refs.c (vect_create_addr_base_for_vector_ref):
13924 Cleanup to use less temporaries.
13925 (vect_create_data_ref_ptr): Cleanup.
13927 2013-03-18 Jakub Jelinek <jakub@redhat.com>
13929 PR tree-optimization/56635
13930 * fold-const.c (operand_equal_p): For MEM_REF and TARGET_MEM_REF,
13931 require types_compatible_p types.
13933 2013-03-18 Nick Clifton <nickc@redhat.com>
13935 * config/stormy16/stormy16.c (xstormy16_expand_prologue): Remove
13936 spurious backslash.
13938 * config/mn10300/mn10300.c (mn10300_get_live_callee_saved_regs):
13939 Add missing line to comment describing function.
13941 2013-03-18 Richard Biener <rguenther@suse.de>
13943 PR tree-optimization/56210
13944 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
13945 Handle string / character search functions.
13946 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
13948 2013-03-18 Richard Biener <rguenther@suse.de>
13950 PR middle-end/56483
13951 * cfgexpand.c (expand_gimple_cond): Inline gimple_cond_single_var_p
13952 and implement properly.
13953 * gimple.h (gimple_cond_single_var_p): Remove.
13955 2013-03-18 Richard Biener <rguenther@suse.de>
13957 * tree-data-ref.h (find_data_references_in_loop): Declare.
13958 * tree-data-ref.c (get_references_in_stmt): Use a stack
13959 vector pre-allocated in the callers.
13960 (find_data_references_in_stmt): Adjust.
13961 (graphite_find_data_references_in_stmt): Likewise.
13962 (create_rdg_vertices): Likewise.
13963 (find_data_references_in_loop): Export.
13964 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
13965 Compute dependences here...
13966 (vect_analyze_data_refs): ...not here. When we encounter
13967 a non-vectorizable data reference in basic-block vectorization
13968 truncate the data reference vector. Do not bother to
13969 fixup data-dependence information for gather loads.
13970 * tree-vect-slp.c (vect_slp_analyze_bb_1): Check the number
13971 of data references, as reported.
13973 2013-03-18 Richard Biener <rguenther@suse.de>
13975 PR tree-optimization/3713
13976 * tree-ssa-sccvn.c (visit_copy): Simplify. Always propagate
13977 has_constants and expr.
13978 (stmt_has_constants): Properly valueize SSA names when deciding
13979 whether the stmt has constants.
13981 2013-03-18 Richard Biener <rguenther@suse.de>
13983 * tree-ssa-loop-manip.c (find_uses_to_rename): Do not scan the
13984 whole function when there is nothing to do.
13985 * tree-ssa-loop.c (pass_vectorize): Remove TODO_update_ssa.
13986 * tree-vectorizer.c (vectorize_loops): Update virtual and
13987 loop-closed SSA once.
13988 * tree-vect-loop.c (vect_transform_loop): Do not update SSA here.
13990 2013-03-18 Richard Biener <rguenther@suse.de>
13992 PR middle-end/56113
13993 * domwalk.c (bb_postorder): New global static.
13994 (cmp_bb_postorder): New function.
13995 (walk_dominator_tree): Replace scheme imposing an order for
13996 visiting dominator sons by one sorting them at the time they
13997 are pushed on the stack.
13999 2013-03-18 Richard Biener <rguenther@suse.de>
14001 PR tree-optimization/39326
14002 * tree-ssa-loop-im.c (refs_independent_p): Exploit symmetry.
14003 (struct mem_ref): Replace mem member with ao_ref typed member.
14004 (MEM_ANALYZABLE): Adjust.
14005 (memref_eq): Likewise.
14006 (mem_ref_alloc): Likewise.
14007 (gather_mem_refs_stmt): Likewise.
14008 (mem_refs_may_alias_p): Use the ao_ref to query the alias oracle.
14009 (execute_sm_if_changed_flag_set): Adjust.
14010 (execute_sm): Likewise.
14011 (ref_always_accessed_p): Likewise.
14012 (refs_independent_p): Likewise.
14013 (can_sm_ref_p): Likewise.
14015 2013-03-18 Jakub Jelinek <jakub@redhat.com>
14018 * tree.c (tree_int_cst_min_precision): For integer_zerop (value)
14019 return 1 even for !unsignedp.
14021 2013-03-17 Uros Bizjak <ubizjak@gmail.com>
14023 * config/i386/i386.md (isa): Add x64 and nox64.
14024 (enabled): Define x64 for TARGET_64BIT and nox64 for !TARGET_64BIT.
14025 (*pushtf): Enable *roF alternative for x64 isa only.
14026 (*pushxf): Merge with *pushxf_nointeger. Use Yx*r constraint. Set
14027 mode attribute of integer alternatives to DImode for TARGET_64BIT.
14028 (*pushdf): Merge with *pushdf_rex64. Use x64 and nox64 isa attributes.
14029 (*movtf_internal): Merge from *movtf_internal_rex64 and
14030 *movtf_internal_sse. Use x64 and nox64 isa attributes.
14031 (*movxf_internal): Merge with *movxf_internal_rex64. Use x64 and
14032 nox64 isa attributes.
14033 (*movdf_internal): Merge with *movdf_internal_rex64. Use x64 and
14034 nox64 isa attributes.
14035 * config/i386/constraints.md (Yd): Do not set for TARGET_64BIT.
14037 2013-03-17 Uros Bizjak <ubizjak@gmail.com>
14039 * config/alpha/alpha.c (TARGET_LRA_P): New define.
14041 2013-03-17 Jakub Jelinek <jakub@redhat.com>
14044 * config/arm/arm.h (REG_CLASS_NAMES): Add "SFP_REG" and "AFP_REG"
14045 class names. Remove trailing comma after "ALL_REGS".
14047 2013-03-16 Jan Hubicka <jh@suse.cz>
14049 * cgraph.h (cgraph_get_create_real_symbol_node): Declare.
14050 * cgraph.c (cgraph_get_create_real_symbol_node): New function.
14051 * cgrpahbuild.c: Use cgraph_get_create_real_symbol_node instead
14052 of cgraph_get_create_node.
14053 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
14055 2013-03-16 Jason Merrill <jason@redhat.com>
14058 * dwarf2out.c (gen_generic_params_dies): Indicate default arguments
14059 with DW_AT_default_value.
14061 2013-03-16 Jakub Jelinek <jakub@redhat.com>
14063 * BASE-VER: Set to 4.9.0.
14065 2013-03-14 Andi Kleen <ak@linux.intel.com>
14068 * doc/extend.texi: Document __ATOMIC_HLE_ACQUIRE,
14069 __ATOMIC_HLE_RELEASE. Document __builtin_ia32 TSX intrincs.
14070 Document _x* TSX intrinsics.
14072 2013-03-14 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
14073 David Holsgrove <david.holsgrove@xilinx.com>
14075 * configure.ac: Add MicroBlaze TLS support detection.
14076 * configure: Regenerate.
14077 * config/microblaze/microblaze-protos.h
14078 (microblaze_cannot_force_const_mem, microblaze_tls_referenced_p,
14079 symbol_mentioned_p, label_mentioned_p): Add prototypes.
14080 * config/microblaze/microblaze.c (microblaze_address_type): Add
14081 ADDRESS_TLS and tls_reloc address types.
14082 (microblaze_address_info): Add tls_reloc.
14083 (TARGET_HAVE_TLS): Define.
14084 (get_tls_get_addr, microblaze_tls_symbol_p, microblaze_tls_operand_p_1,
14085 microblaze_tls_referenced_p, microblaze_cannot_force_const_mem,
14086 symbol_mentioned_p, label_mentioned_p, tls_mentioned_p,
14087 load_tls_operand, microblaze_call_tls_get_addr,
14088 microblaze_legitimize_tls_address): New functions.
14089 (microblaze_classify_unspec): Handle UNSPEC_TLS.
14090 (get_base_reg): Use microblaze_tls_symbol_p.
14091 (microblaze_classify_address): Handle TLS.
14092 (microblaze_legitimate_pic_operand): Use symbol_mentioned_p,
14093 label_mentioned_p and microblaze_tls_referenced_p.
14094 (microblaze_legitimize_address): Handle TLS.
14095 (microblaze_address_insns): Handle ADDRESS_TLS.
14096 (pic_address_needs_scratch): Handle TLS.
14097 (print_operand_address): Handle TLS.
14098 (microblaze_expand_prologue): Check TLS_NEEDS_GOT.
14099 (microblaze_expand_move): Handle TLS.
14100 (microblaze_legitimate_constant_p): Check
14101 microblaze_cannot_force_const_mem and microblaze_tls_symbol_p.
14102 (TARGET_CANNOT_FORCE_CONST_MEM): Define.
14103 * config/microblaze/microblaze.h (TLS_NEEDS_GOT): Define
14104 (PIC_OFFSET_TABLE_REGNUM): Set.
14105 * config/microblaze/linux.h (TLS_NEEDS_GOT): Define.
14106 * config/microblaze/microblaze.md (UNSPEC_TLS): Define.
14107 (addsi3, movsi_internal2, movdf_internal): Update constraints
14108 * config/microblaze/predicates.md (arith_plus_operand): Define
14109 (move_operand): Redefine as move_src_operand,
14110 check microblaze_tls_referenced_p.
14112 2013-03-14 Ian Bolton <ian.bolton@arm.com>
14114 * config/aarch64/aarch64.md: (*and<mode>3nr_compare0): Use CC_NZ.
14115 (*and_<SHIFT:optab><mode>3nr_compare0): Likewise.
14117 2013-03-14 Ian Bolton <ian.bolton@arm.com>
14119 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return correct
14122 2013-03-14 Jakub Jelinek <jakub@redhat.com>
14124 PR tree-optimization/53265
14125 * common.opt (Waggressive-loop-optimizations): New option.
14126 * tree-ssa-loop-niter.c: Include tree-pass.h.
14127 (do_warn_aggressive_loop_optimizations): New function.
14128 (record_estimate): Call it. Don't add !is_exit bounds to loop->bounds
14129 if number_of_latch_executions returned constant.
14130 (estimate_numbers_of_iterations_loop): Call number_of_latch_executions
14131 early. If number_of_latch_executions returned constant, set
14132 nb_iterations_upper_bound back to it.
14133 * cfgloop.h (struct loop): Add warned_aggressive_loop_optimizations
14135 * Makefile.in (tree-ssa-loop-niter.o): Depend on $(TREE_PASS_H).
14136 * doc/invoke.texi (-Wno-aggressive-loop-optimizations): Document.
14138 * config/aarch64/t-aarch64-linux (MULTARCH_DIRNAME): Remove.
14139 (MULTILIB_OSDIRNAMES): Set.
14140 * genmultilib: If defaultosdirname doesn't start with :: , set
14141 defaultosdirname2 instead, clear it and emit two . multilib_raw
14142 entries instead of just one.
14144 2013-03-14 Kaz Kojima <kkojima@gcc.gnu.org>
14146 * config/sh/linux.h (TARGET_DEFAULT): Remove MASK_USERMODE.
14147 (SUBTARGET_OVERRIDE_OPTIONS): Set TARGET_USERMODE as default.
14148 * config/sh/netbsd-elf.h (TARGET_DEFAULT): Remove MASK_USERMODE.
14149 (SUBTARGET_OVERRIDE_OPTIONS): New.
14151 2013-03-13 Oleg Endo <olegendo@gcc.gnu.org>
14154 * config/sh/sh.opt (FPU_SINGLE_ONLY): New mask.
14155 (musermode): Convert to Var(TARGET_USERMODE).
14156 * config/sh/sh.h (SELECT_SH2A_SINGLE_ONLY, SELECT_SH4_SINGLE_ONLY,
14157 MASK_ARCH): Add MASK_FPU_SINGLE_ONLY.
14158 * config/sh/sh.c (sh_option_override): Use
14159 TARGET_FPU_DOUBLE || TARGET_FPU_SINGLE_ONLY for call-fp case.
14160 * config/sh/sh.md (udivsi3_i1, divsi3_i1): Remove ! TARGET_SH4
14162 (udivsi3_i4, divsi3_i4): Use TARGET_FPU_DOUBLE condition instead of
14164 (udivsi3_i4_single, divsi3_i4_single): Use
14165 TARGET_FPU_SINGLE_ONLY || TARGET_FPU_DOUBLE instead of TARGET_HARD_SH4.
14167 2013-03-13 Dave Korn <dave.korn.cygwin@....>
14169 * config/i386/cygwin.h (SHARED_LIBGCC_SPEC): Make shared libgcc the
14172 2013-03-13 Richard Biener <rguenther@suse.de>
14174 PR tree-optimization/56608
14175 * tree-vect-slp.c (vect_schedule_slp): Do not remove scalar
14176 calls when vectorizing basic-blocks.
14178 2013-03-13 Jakub Jelinek <jakub@redhat.com>
14181 * config.gcc: On arm, mips, sh and sparc add vxworks-dummy.h to
14184 2013-03-12 Jakub Jelinek <jakub@redhat.com>
14186 * doc/invoke.texi (-Waddr-space-convert): Move into the table earlier.
14188 2013-03-11 Jan Hubicka <jh@suse.cz>
14191 * lto-streamer-out.c (output_symbol_p): Skip references from
14192 constructors of external variables.
14194 2013-03-11 Jan Hubicka <jh@suse.cz>
14196 PR middle-end/56571
14197 * valtrack.c (cleanup_auto_inc_dec): Unshare clobbers originating
14199 * emit-rtl.c (verify_rtx_sharing): Likewise.
14200 (copy_insn_1): Likewise.
14201 * rtl.c (copy_rtx): Likewise.
14203 2013-03-11 Georg-Johann Lay <avr@gjlay.de>
14206 * config/avr/avr.c (avr_print_operand): Add space after '%c' in
14207 output_operand_lossage message.
14209 2013-03-11 Richard Earnshaw <rearnsha@arm.com>
14212 * arm.c (shift_op): Validate RTL pattern on the fly.
14213 (arm_print_operand, case 'S'): Don't use shift_operator to validate
14216 2013-03-10 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
14219 * config/pa/pa.md (call_value): Check for calls to powf and direct to
14220 new call patterns that clobber %fr12.
14221 (call_val_powf, call_val_powf_pic, call_val_powf_64bit): New insn,
14222 split and postreload patterns.
14223 * config/pa/pa.c (pa_conditional_register_usage): Revert marking
14224 registers %fr12 and %fr12R as call used.
14226 2013-03-09 Steven Bosscher <steven@gcc.gnu.org>
14228 * dse.c (delete_dead_store_insn): Respect TDF_DETAILS.
14229 (canon_address, record_store, replace_read, check_mem_read_rtx,
14230 scan_insn, dse_step1, dse_step2_init, dse_step2_spill,
14231 dse_step4, dse_step5_nospill, dse_step5_spill, dse_step6,
14232 rest_of_handle_dse): Likewise.
14234 2013-03-09 Richard Sandiford <rdsandiford@googlemail.com>
14236 PR middle-end/56524
14237 * tree.h (tree_optimization_option): Rename target_optabs to optabs.
14239 (TREE_OPTIMIZATION_OPTABS): Update after previous field change.
14240 (TREE_OPTIMIZATION_BASE_OPTABS): New macro.
14241 (save_optabs_if_changed): Replace with...
14242 (init_tree_optimization_optabs): ...this.
14243 * optabs.c (save_optabs_if_changed): Rename to...
14244 (init_tree_optimization_optabs): ...this. Take the optimization node
14245 as argument. Do nothing if the base optabs are already correct.
14246 Reuse the existing TREE_OPTIMIZATION_OPTABS memory if we need
14247 to recompute optabs.
14248 * function.h (function): Remove optabs field.
14249 * function.c (invoke_set_current_function_hook): Call
14250 init_tree_optimization_optabs. Use the result to initialize
14253 2013-02-27 Aldy Hernandez <aldyh@redhat.com>
14255 * trans-mem.c (expand_transaction): Do not set PR_INSTRUMENTEDCODE
14256 if GTMA_HAS_NO_INSTRUMENTATION.
14257 (generate_tm_state): Keep GTMA_HAS_NO_INSTRUMENTATION bit.
14258 (ipa_tm_transform_transaction): Set GTMA_HAS_NO_INSTRUMENTATION.
14259 * gimple.h (GTMA_HAS_NO_INSTRUMENTATION): Define.
14260 * gimple-pretty-print.c (dump_gimple_transaction): Handle
14261 GTMA_HAS_NO_INSTRUMENTATION.
14263 2013-03-08 Jakub Jelinek <jakub@redhat.com>
14265 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Don't link against
14268 2013-03-08 Marek Polacek <polacek@redhat.com>
14269 Jakub Jelinek <jakub@redhat.com>
14271 PR tree-optimization/56478
14272 * predict.c (is_comparison_with_loop_invariant_p): Change the
14273 type of loop_step to tree.
14274 (predict_loops): Adjust.
14275 (predict_iv_comparison): Perform the computations on double_ints.
14277 2013-03-08 Richard Biener <rguenther@suse.de>
14279 PR tree-optimization/56570
14280 * tree-cfg.c (verify_expr_location_1): Verify locations for
14282 * tree-sra.c (create_access_replacement): Strip locations
14283 from DECL_DEBUG_EXPRs.
14285 2013-03-08 Richard Biener <rguenther@suse.de>
14287 * tree-inline.c (expand_call_inline): Do not associate
14288 a BLOCK with the location in BLOCK_SOURCE_LOCATION.
14289 * tree-cfg.c (verify_location): Verify BLOCK_SOURCE_LOCATION.
14291 2013-03-08 Richard Biener <rguenther@suse.de>
14293 * tree-ssa-ter.c (is_replaceable_p): Do not TER across location
14294 or block changes with -Og. Fix for location / block encoding
14295 changes and PHI arguments with locations.
14297 2013-03-07 Steven Bosscher <steven@gcc.gnu.org>
14299 * bitmap.c (struct bitmap_descriptor_d): Use unsigned HOST_WIDEST_INT
14301 (struct output_info): Likewise.
14302 (register_overhead): Remove bad gcc_assert.
14303 (bitmap_find_bit): If there is only a single bitmap element, do not
14304 count a miss as a search.
14305 (print_statistics): Update for counter type changes.
14306 (dump_bitmap_statistics): Likewise. Print headers such that they
14307 are properly lined up with the printed counters.
14309 2013-03-07 Jakub Jelinek <jakub@redhat.com>
14311 PR tree-optimization/56559
14312 * tree-ssa-reassoc.c (zero_one_operation): When looking at rhs2,
14313 check that it has only a single use.
14315 2013-03-07 Richard Biener <rguenther@suse.de>
14317 * doc/invoke.texi (fwhole-program): Discourage use in combination
14320 2013-03-06 Jakub Jelinek <jakub@redhat.com>
14322 * config/arm/t-arm (TM_H, OPTIONS_H_EXTRA): Add arm-cores.def.
14324 PR tree-optimization/56539
14325 * tree-tailcall.c (adjust_return_value_with_ops): Use GSI_SAME_STMT
14326 instead of GSI_CONTINUE_LINKING as last argument to
14327 force_gimple_operand_gsi. Adjust function comment.
14329 * config/aarch64/t-aarch64 (TM_H, OPTIONS_H_EXTRA): Add
14332 PR middle-end/56548
14333 * expr.c (expand_cond_expr_using_cmove): When expanding cmove in
14334 promoted mode, convert the result back to the original mode.
14336 2013-03-06 Richard Biener <rguenther@suse.de>
14338 PR middle-end/56294
14339 * tree-into-ssa.c (insert_phi_nodes_for): Add dumping.
14340 (insert_updated_phi_nodes_compare_uids): New function.
14341 (update_ssa): Sort symbols_to_rename after UID before
14342 traversing it to insert PHI nodes.
14344 2013-03-06 Richard Biener <rguenther@suse.de>
14346 PR middle-end/50494
14347 * tree-vect-data-refs.c (vect_can_force_dr_alignment_p):
14348 Do not adjust alignment of DECL_IN_CONSTANT_POOL decls.
14351 2013-02-13 Richard Biener <rguenther@suse.de>
14354 * varasm.c (output_constant_def_1): Get the decl representing
14355 the constant as argument.
14356 (output_constant_def): Wrap output_constant_def_1.
14357 (make_decl_rtl): Use output_constant_def_1 with the decl
14358 representing the constant.
14359 (build_constant_desc): Optionally re-use a decl already
14360 representing the constant.
14361 (tree_output_constant_def): Adjust.
14363 2013-03-06 Joey Ye <joey.ye@arm.com>
14366 * gcc.c (convert_white_space): New function.
14367 (main): Handles white space in function name.
14369 2013-03-06 Oleg Endo <olegendo@gcc.gnu.org>
14372 * config/sh/sh.c (sh_option_override): Check for TARGET_DYNSHIFT
14373 instead of TARGET_SH2 for call-table case. Do not set sh_div_strategy
14374 to SH_DIV_CALL_TABLE for TARGET_SH2.
14375 * config.gcc (sh_multilibs): Add m2 and m2a to sh*-*-linux* multilib
14377 * doc/invoke.texi (SH options): Document mdiv= call-div1, call-fp,
14378 call-table options.
14380 2013-03-05 Sterling Augustine <saugustine@google.com>
14381 Cary Coutant <ccoutant@google.com>
14384 * dwarf2out.c (resolve_addr): Don't call
14385 remove_loc_list_addr_table_entries a second time for the same
14388 2013-03-05 Jakub Jelinek <jakub@redhat.com>
14391 * cfgexpand.c (expand_debug_parm_decl): Call copy_rtx on incoming.
14392 (avoid_complex_debug_insns): New function.
14393 (expand_debug_locations): Call it.
14395 PR rtl-optimization/56484
14396 * ifcvt.c (noce_process_if_block): If else_bb is NULL, avoid extending
14397 lifetimes of hard registers on small register class machines.
14399 2013-03-05 David Holsgrove <david.holsgrove@xilinx.com>
14401 * config/microblaze/microblaze-protos.h: Rename
14402 microblaze_is_interrupt_handler to microblaze_is_interrupt_variant.
14403 * config/microblaze/microblaze.c (microblaze_attribute_table): Add
14405 (microblaze_fast_interrupt_function_p): New function.
14406 (microblaze_is_interrupt_handler): Rename to
14407 microblaze_is_interrupt_variant and add fast_interrupt check.
14408 (microblaze_must_save_register): Use microblaze_is_interrupt_variant.
14409 (save_restore_insns): Likewise.
14410 (compute_frame_size): Likewise.
14411 (microblaze_function_prologue): Add FAST_INTERRUPT_NAME.
14412 (microblaze_globalize_label): Likewise.
14413 * config/microblaze/microblaze.h: Define FAST_INTERRUPT_NAME.
14414 * config/microblaze/microblaze.md: Use wrapper
14415 microblaze_is_interrupt_variant.
14417 2013-03-05 Kai Tietz <ktietz@redhat.com>
14419 * sdbout.c (sdbout_one_type): Switch to current function's section
14420 supporting cold/hot.
14422 2013-03-05 David Holsgrove <david.holsgrove@xilinx.com>
14424 * doc/invoke.texi (MicroBlaze): Add -mbig-endian, -mlittle-endian,
14427 2013-03-05 Jakub Jelinek <jakub@redhat.com>
14429 PR middle-end/56461
14430 * ggc-common.c (gt_pch_save): For ENABLE_VALGRIND_CHECKING,
14431 if VALGRIND_GET_VBITS is defined, temporarily make object
14432 memory all defined, and restore previous valgrind addressability
14433 and definability afterwards. Free this_object at the end.
14435 PR middle-end/56461
14436 * lra.c (lra): Call lra_clear_live_ranges if live_p,
14437 right before calling lra_create_live_ranges, also call it
14438 when clearing live_p. Only call lra_clear_live_ranges
14439 at the end if live_p.
14441 PR middle-end/56461
14442 * sched-deps.c (delete_dep_node): Free DEP_REPLACE.
14444 2013-03-05 Richard Biener <rguenther@suse.de>
14446 PR tree-optimization/56521
14447 * tree-ssa-sccvn.c (set_value_id_for_result): Always initialize
14450 2013-03-05 Steven Bosscher <steven@gcc.gnu.org>
14453 * except.h (remove_unreachable_eh_regions): New prototype.
14454 * except.c (remove_eh_handler_splicer): New function, split out
14455 of remove_eh_handler.
14456 (remove_eh_handler): Use remove_eh_handler_splicer. Add comment
14457 warning about running it on many EH regions one at a time.
14458 (remove_unreachable_eh_regions_worker): New function, walk the
14459 EH tree in depth-first order and remove non-marked regions.
14460 (remove_unreachable_eh_regions): New function.
14461 * tree-eh.c (mark_reachable_handlers): New function, split out
14462 from remove_unreachable_handlers.
14463 (remove_unreachable_handlers): Use mark_reachable_handlers and
14464 remove_unreachable_eh_regions.
14465 (remove_unreachable_handlers_no_lp): Use mark_reachable_handlers
14466 and remove_unreachable_eh_regions.
14468 2013-03-05 Richard Biener <rguenther@suse.de>
14470 PR middle-end/56525
14471 * loop-init.c (fix_loop_structure): Remove loops in two stages,
14472 not freeing them until the end.
14474 2013-03-05 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
14476 * config/s390/s390.h: Define DWARF2_ASM_LINE_DEBUG_INFO.
14478 2013-03-05 Richard Biener <rguenther@suse.de>
14480 PR tree-optimization/56270
14481 * tree-vect-slp.c (vect_schedule_slp): Clear vectorized stmts
14482 of loads after scheduling an SLP instance.
14484 2013-03-05 Jakub Jelinek <jakub@redhat.com>
14486 * Makefile.in (dg_target_exps): Add aarch64.exp, epiphany.exp and
14488 (check_gcc_parallelize): Run guality.exp as a separate job from
14489 vect.exp with unsorted.exp and $(dg_target_exps) separately from
14490 struct-layout-1.exp with stackalign.exp.
14492 * alias.c (init_alias_analysis): Clear reg_known_equiv_p bitmap.
14494 PR middle-end/56461
14495 * tree-vect-slp.c (vect_supported_load_permutation_p): Free
14496 load_index sbitmap even if some bit in it isn't set.
14498 PR middle-end/56461
14499 * tree-ssa-loop-niter.c (bb_queue): Remove typedef.
14500 (discover_iteration_bound_by_body_walk): Change queues to
14501 vec<vec<basic_block> > and queue to vec<basic_block>. Fix up
14502 spelling in comment. Call safe_push on queues[bound_index] directly.
14503 Release queues[queue_index] in every iteration unconditionally.
14504 Release bounds vector.
14506 PR middle-end/56461
14507 * tree-vect-stmts.c (free_stmt_vec_info_vec): Call
14508 free_stmt_vec_info on any left-over stmt_vec_info in the vector.
14509 * tree-vect-loop.c (vect_create_epilog_for_reduction): Release
14512 2013-03-05 Richard Biener <rguenther@suse.de>
14515 * tree-inline.c (remap_blocks_to_null): New function.
14516 (expand_call_inline): When expanding a call stmt without
14517 an associated block inline remap all callee blocks to NULL.
14519 2013-03-05 Jakub Jelinek <jakub@redhat.com>
14521 PR rtl-optimization/56494
14522 * simplify-rtx.c (simplify_truncation): If C is narrower than A,
14523 optimize (truncate:A (subreg:B (truncate:C X) 0)) into
14524 (subreg:A (truncate:C X) 0) instead of (truncate:A X).
14526 PR middle-end/56461
14527 * sel-sched-ir.c (free_sched_pools): Release
14528 succs_info_pool.stack[succs_info_pool.max_top] vectors too
14529 if succs_info_pool.max_top isn't -1.
14532 * opts.c (opts_obstack, opts_concat): Moved to...
14533 * opts-common.c (opts_obstack, opts_concat): ... here.
14535 2013-03-04 Jakub Jelinek <jakub@redhat.com>
14537 PR middle-end/56461
14538 * diagnostic.c (diagnostic_append_note): Save and restore old prefix.
14540 2013-03-04 Martin Jambor <mjambor@suse.cz>
14542 * tree-dfa.c (get_or_create_ssa_default_def): Use parameter fn in
14543 all appropriate places.
14545 2013-01-04 Eric Botcazou <ebotcazou@adacore.com>
14547 PR tree-optimization/56424
14548 * ipa-split.c (split_function): Do not set the RSO flag if result is
14549 not by reference and its type is a register type.
14551 2013-03-04 David Holsgrove <david.holsgrove@xilinx.com>
14553 * config/microblaze/microblaze.c (microblaze_valid_pic_const): New
14554 (microblaze_legitimate_pic_operand): Likewise
14555 * config/microblaze/microblaze.h (LEGITIMATE_PIC_OPERAND_P): calls
14556 new function microblaze_legitimate_pic_operand
14557 * config/microblaze/microblaze-protos.h
14558 (microblaze_legitimate_pic_operand): Declare.
14560 2013-03-04 Edgar E. Iglesias <edgar.iglesias@gmail.com>
14562 * config/microblaze/predicates.md (call_insn_simple_operand):
14563 New predicate for supported rtx code types.
14564 * config/microblaze/microblaze.md (call_internal1): Use
14565 call_insn_simple_operand predicate.
14567 2013-03-04 Jakub Jelinek <jakub@redhat.com>
14569 PR middle-end/56461
14570 * tree-loop-distribution.c (ldist_gen): Call partition_free after each
14571 partitions.ordered_remove.
14573 PR middle-end/56461
14574 * tree-vect-stmts.c (vectorizable_conversion): Don't call
14575 vec_oprnds0.create (1) for modifier == NONE.
14577 PR middle-end/56461
14578 * tree-vect-stmts.c (vectorizable_shift): Don't call create methods
14579 on vec_oprnds0 or vec_oprnds1 before loop, only call it on
14580 vec_oprnds1 right before pushing anything to it for
14583 PR middle-end/56461
14584 * tree-vect-loop.c (destroy_loop_vec_info): For !clean_stmts, just
14585 set nbbs to 0 instead of having separate code path.
14586 (vect_analyze_loop_form): Call destroy_loop_vec_info with true
14587 instead of false as last argument if returning NULL.
14589 2013-03-03 Sandra Loosemore <sandra@codesourcery.com>
14591 * target.def (TARGET_OPTION_VALID_ATTRIBUTE_P): Update comments;
14592 the attribute is now called "target" instead of "option".
14593 (TARGET_OPTION_PRAGMA_PARSE): Likewise, for the pragma.
14594 * doc/tm.texi.in (Target Attributes): Likewise document the correct
14595 attribute/pragma name for TARGET_OPTION_VALID_P and
14596 TARGET_OPTION_PRAGMA_PARSE. Also copy-edit and correct markup.
14597 * doc/tm.texi: Regenerated.
14599 2013-03-02 David Holsgrove <david.holsgrove@xilinx.com>
14601 * config/microblaze/microblaze.c:
14602 Check mcpu, pcmp requirement and set TARGET_REORDER to 0 if not met.
14603 * config/microblaze/microblaze.h: Add -mxl-reorder to
14605 * config/microblaze/microblaze.md: New bswapsi2 and bswaphi2.
14606 instructions emitted if TARGET_REORDER.
14607 * config/microblaze/microblaze.opt: New option -mxl-reorder set to 1
14608 or 0 for -m/-mno case, but initialises as 2 to detect default use case
14611 2013-03-01 Xinliang David Li <davidxl@google.com>
14613 * tree-ssa-uninit.c (compute_control_dep_chain): Limit post-dom
14616 2013-03-01 Jakub Jelinek <jakub@redhat.com>
14618 PR middle-end/56461
14619 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Release path
14620 vector even when returning true. Fix up function comment formatting.
14622 PR middle-end/56461
14623 * ira-build.c (ira_loop_nodes_count): New variable.
14624 (create_loop_tree_nodes): Initialize it.
14625 (finish_loop_tree_nodes): Use it instead of looking at current_loops.
14627 PR middle-end/56461
14628 * tree-vect-data-refs.c (vect_permute_store_chain): Avoid using copy
14629 method on dr_chain and result_chain.
14630 * tree-vect-stmts.c (vectorizable_store): Only call
14631 result_chain.create if j == 0.
14633 PR middle-end/56461
14634 * tree-vect-stmts.c (vect_create_vectorized_promotion_stmts): Call
14635 vec_oprnds0->release (); rather than vec_oprnds0->truncate (0)
14636 before overwriting it.
14638 2013-03-01 Tobias Burnus <burnus@net-b.de>
14640 * doc/extended.texi (C Extensions): Change order in @menu
14642 (Other MIPS Built-in Functions): Move last MIPS entry before
14643 "picoChip Built-in Functions".
14644 (SH Built-in Functions): Move after RX Built-in Functions.
14645 * doc/gcc.texi (Introduction): Change order in @menu
14647 * doc/md.texi (Constraints): Ditto.
14648 * gty.texi (Type Information): Ditto.
14649 (User-provided marking routines for template types): Make
14651 * doc/invoke.texi (AArch64 Options): Move before
14652 "Adapteva Epiphany Options".
14654 2013-02-28 Konstantin Serebryany <konstantin.s.serebryany@gmail.com>
14655 Jakub Jelinek <jakub@redhat.com>
14658 * asan.c (gate_asan): Lookup no_sanitize_address instead of
14659 no_address_safety_analysis attribute.
14660 * doc/extend.texi (no_address_safety_attribute): Rename to
14661 no_sanitize_address attribute, mention no_address_safety_analysis
14662 attribute as deprecated alias.
14664 2013-02-28 Jakub Jelinek <jakub@redhat.com>
14666 PR middle-end/56461
14667 * tree-vectorizer.h (vect_get_slp_defs): Change 3rd argument
14668 type to vec<vec<tree> > *.
14669 * tree-vect-slp.c (vect_get_slp_defs): Likewise. Change vec_defs
14670 to be vec<tree> instead of vec<tree> *, set vec_defs
14671 to vNULL and call vec_defs.create (number_of_vects), adjust other
14673 * tree-vect-stmts.c (vect_get_vec_defs, vectorizable_call,
14674 vectorizable_condition): Adjust vect_get_slp_defs callers.
14676 2013-02-28 James Greenhalgh <james.greenhalgh@arm.com>
14678 * config/aarch64/aarch64.c
14679 (aarch64_float_const_representable): Remove unused variable.
14681 2013-02-28 James Greenhalgh <james.greenhalgh@arm.com>
14683 * config/aarch64/aarch64.c (aarch64_mangle_type): Make static.
14685 2013-02-28 James Greenhalgh <james.greenhalgh@arm.com>
14687 * config/aarch64/aarch64-builtins.c
14688 (aarch64_init_simd_builtins): Make static.
14690 2013-02-28 James Greenhalgh <james.greenhalgh@arm.com>
14692 * config/aarch64/aarch64.c
14693 (aarch64_simd_make_constant): Make static.
14695 2013-02-28 Martin Jambor <mjambor@suse.cz>
14697 * tree-sra.c (load_assign_lhs_subreplacements): Do not put replacements
14698 with no initialization to the RHS of debug statements.
14700 2013-02-28 Martin Jambor <mjambor@suse.cz>
14702 PR tree-optimization/56294
14703 * tree-sra.c (analyze_access_subtree): Create replacement declarations.
14705 (get_access_replacement): Do not call create_access_replacement.
14706 Assert a replacement exists.
14707 (get_repl_default_def_ssa_name): Create the replacement declaration
14710 2013-02-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
14712 * config/arm/arm.c (arm_output_mi_thunk): Call final_start_function and
14713 final_end_function.
14715 2013-02-28 Marek Polacek <polacek@redhat.com>
14717 PR rtl-optimization/56466
14718 * loop-unroll.c (unroll_and_peel_loops): Call fix_loop_structure
14719 if we're changing a loop.
14720 (peel_loops_completely): Likewise.
14722 2013-02-28 Paolo Carlini <paolo.carlini@oracle.com>
14725 * doc/invoke.texi ([-Wctor-dtor-privacy]): Complete.
14727 2013-02-28 Georg-Johann Lay <avr@gjlay.de>
14730 * config/avr/avr.c (avr_init_builtins): Use 'n' instead of empty
14731 macro parameters with: FX_FTYPE_FX, FX_FTYPE_FX_INT, INT_FTYPE_FX,
14732 INTX_FTYPE_FX, FX_FTYPE_INTX.
14733 * config/avr/builtins.def: Adjust respective DEF_BUILTIN.
14735 2013-02-28 Georg-Johann Lay <avr@gjlay.de>
14737 * avr/avr-mcus.def (ata5272, ata5505, attiny1634, ata6285)
14738 (ata6286, atmega8a, atmega48pa, ata5790, ata5790n, ata5795)
14739 (atmega164pa, atmega165pa, atmega168pa, atmega16hva, atmega16hvb)
14740 (atmega16hvbrevb, atmega16m1, atmega16u4, atmega26hvg, atmega32a)
14741 (atmega32a, atmega3250pa, atmega3290pa, atmega32c1, atmega32m1)
14742 (atmega32u4, atmega32u6, atmega64a, atmega6490a, atmega6490p)
14743 (atmega64c1, atmega64m1, atmega64rfa2, atmega64rfr2, atmega32hvb)
14744 (atmega32hvbrevb, atmega16hva2, atmega48hvf, at90pwm161)
14745 (atmega128a, atmega1284, atmxt112sl, atmxt224, atmxt224e)
14746 (atmxt336s, atxmega16a4u, atxmega16c4, atxmega32a4u, atxmega32c4)
14747 (atxmega32e5, atxmega64a3u, atxmega64a4u, atxmega64b1, atxmega64b3)
14748 (atxmega64c3, atxmega64d4, atxmega128a3u, atxmega128b1)
14749 (atxmega128b3, atxmega128c3, atxmega128d4, atmxt540s, atmxt540sreva)
14750 (atxmega192a3u, atxmega192c3, atxmega256a3u, atxmega256c3)
14751 (atxmega384c3, atxmega384d3, atxmega128a4u): New AVR_MCU.
14752 (avrxmega6): Increase max flash segments from 5 to 6.
14753 * config/avr/t-multilib: Regenerate.
14754 * config/avr/avr-tables.opt: Regenerate.
14755 * doc/avr-mmcu.texi: Regenerate.
14757 2013-02-28 Georg-Johann Lay <avr@gjlay.de>
14759 * config/avr/avr.h (device_to_arch): Rename to device_to_ld.
14760 (avr_device_to_arch): Rename to avr_device_to_ld.
14761 (avr_device_to_as): New prototype.
14762 (EXTRA_SPEC_FUNCTIONS): Add device_to_as.
14763 (ASM_SPEC): Use device_to_as to get -mmcu= and -mno-skip-bug=.
14764 * config/avr/driver-avr.c (avr_device_to_as): New.
14765 (avr_device_to_arch): Rename to avr_device_to_ld.
14767 2013-02-27 Jakub Jelinek <jakub@redhat.com>
14769 PR middle-end/56461
14770 * tree-vect-data-refs.c (vect_permute_load_chain): Avoid using copy
14771 method on dr_chain and result_chain.
14773 PR middle-end/56461
14774 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Call
14775 pointer_set_destroy on not_executed_last_iteration.
14777 PR middle-end/56461
14778 * tree-vect-loop.c (vectorizable_reduction): Release vect_defs vector.
14780 PR middle-end/56461
14781 * ipa-pure-const.c (propagate): Use FOR_EACH_FUNCTION instead of
14782 FOR_EACH_DEFINED_FUNCTION when freeing state.
14784 PR middle-end/56461
14785 * df-scan.c (df_insn_delete): Use df_scan_free_mws_vec before
14787 (df_insn_rescan_debug_internal): Use df_scan_free_mws_vec before
14790 PR middle-end/56461
14791 * ipa-cp.c (decide_whether_version_node): Call vec_free on
14792 known_aggs[i].items and release known_aggs vector.
14794 PR middle-end/56461
14795 * ipa-reference.c (propagate): Free node_info even for alias nodes.
14797 2013-02-27 Edgar E. Iglesias <edgar.iglesias@gmail.com>
14799 * config/microblaze/microblaze.c (microblaze_emit_compare):
14800 Use xor for EQ/NE comparisions.
14801 * config/microblaze/microblaze.md (cstoresf4): Add constraints
14802 (cbranchsf4): Adjust operator to comparison_operator.
14804 2013-02-27 Jakub Jelinek <jakub@redhat.com>
14806 PR middle-end/56461
14807 * tree-flow.h (edge_var_map_vector): Change into va_heap, vl_embed
14809 * tree-ssa.c (redirect_edge_var_map_add): Use vec_safe_reserve and
14810 vec_safe_push, always update *slot.
14811 (redirect_edge_var_map_clear): Use vec_free.
14812 (redirect_edge_var_map_dup): Use vec_safe_copy and vec_safe_reserve.
14813 (free_var_map_entry): Use vec_free.
14814 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Use
14815 FOR_EACH_VEC_SAFE_ELT instead of FOR_EACH_VEC_ELT.
14817 2013-02-27 Andrey Belevantsev <abel@ispras.ru>
14819 PR middle-end/45472
14820 * sel-sched-ir.c (merge_expr): Also change vinsn of merged expr
14821 when the may_trap_p bit of the exprs being merged differs.
14822 Reorder tests for speculativeness in the logical and operator.
14824 2013-02-27 Jakub Jelinek <jakub@redhat.com>
14826 * incpath.c (add_standard_paths): Use reconcat instead of concat
14827 where appropriate and avoid leaking memory.
14829 * opts.h: Include obstack.h.
14830 (opts_concat): New prototype.
14831 (opts_obstack): New declaration.
14832 * opts.c (opts_concat): New function.
14833 (opts_obstack): New variable.
14834 (init_options_struct): Call gcc_init_obstack on opts_obstack.
14835 (finish_options): Use opts_concat instead of concat
14836 and XOBNEWVEC instead of XNEWVEC.
14837 * opts-common.c (generate_canonical_option, decode_cmdline_option,
14838 generate_option): Likewise.
14839 * Makefile.in (OPTS_H): Depend on $(OBSTACK_H).
14840 * lto-wrapper.c (main): Call gcc_init_obstack on opts_obstack.
14843 * stmt.c (expand_switch_as_decision_tree_p): If flag_pic
14844 and ASM_OUTPUT_ADDR_DIFF_ELT isn't defined, return true.
14846 2013-02-26 Jakub Jelinek <jakub@redhat.com>
14848 PR middle-end/56461
14849 * lra-spills.c (lra_spill): Free spill_hard_reg at the end.
14851 2013-02-26 Joern Rennecke <joern.rennecke@embecosm.com>
14853 * config/arm/arm.c (const_ok_for_dimode_op): Back out last change.
14854 (arm_block_move_unaligned_straight): Likewise.
14855 (arm_adjust_block_mem): Likewise.
14857 2013-02-26 Joern Rennecke <joern.rennecke@embecosm.com>
14860 * config/lm32/lm32.c (gen_int_relational): Remove unused variables
14861 temp, cond and label.
14862 * config/lm32/lm32.md (ashlsi3): Remove unused variable one.
14865 * config/c6x/c6x.c (dbx_register_map): Change to unsigned.
14866 * config/c6x/c6x.h (dbx_register_map): Update declaration.
14869 * config/cr16/cr16-protos.h: Move end of RTX_CODE guard below end
14870 of prologue/epilogue functions.
14873 * config/tilegx/tilegx.c (tilegx_expand_prologue):
14874 Remove unused variable cfa_offset.
14875 * config/tilepro/tilepro.c (tilepro_expand_prologue): Likewise.
14878 * config/mn10300/mn10300.c (mn10300_expand_epilogue): Avoid offset
14879 type promotion to unsigned.
14882 * config/arm/arm.c (const_ok_for_dimode_op): Make code consistent
14883 for HOST_WIDE_INT of 32 bit / same size as int.
14884 (arm_block_move_unaligned_straight): Likewise.
14885 (arm_adjust_block_mem): Likewise.
14888 * config/mep/t-mep (mep-pragma.o): Use ALL_COMPILERFLAGS instead of
14891 2013-02-26 Marek Polacek <polacek@redhat.com>
14893 PR tree-optimization/56426
14894 * tree-ssa-loop.c (tree_ssa_loop_init): Always call scev_initialize.
14896 2013-02-26 Richard Biener <rguenther@suse.de>
14899 * config/mn10300/mn10300.c (mn10300_scan_for_setlb_lcc): Remove
14900 unused variable loops.
14902 2013-02-26 Jakub Jelinek <jakub@redhat.com>
14904 PR tree-optimization/56448
14905 * fold-const.c (operand_equal_p) <case tcc_reference>: Don't look at
14906 TREE_SIDE_EFFECTS if flags contain OEP_CONSTANT_ADDRESS_OF.
14907 Clear OEP_CONSTANT_ADDRESS_OF from flags before recursing on second or
14908 later operands of the references, or even first operand for
14909 INDIRECT_REF, TARGET_MEM_REF or MEM_REF.
14911 PR tree-optimization/56443
14912 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): For
14913 overaligned types, pass TYPE_UNSIGNED (scalar_type) as second argument
14914 to type_for_mode langhook.
14916 2013-02-25 Matt Turner <mattst88@gmail.com>
14918 * doc/invoke.texi: Document r4700.
14920 2013-02-25 Richard Biener <rguenther@suse.de>
14922 PR tree-optimization/56175
14923 * tree-ssa-forwprop.c (hoist_conversion_for_bitop_p): New predicate,
14925 (simplify_bitwise_binary): ... here. Also guard the conversion
14926 of (type) X op CST to (type) (X op ((type-x) CST)) with it.
14928 2013-02-25 Catherine Moore <clm@codesourcery.com>
14931 2013-02-24 Catherine Moore <clm@codesourcery.com>
14932 Maciej W. Rozycki <macro@codesourcery.com>
14933 Tom de Vries <tom@codesourcery.com>
14934 Nathan Sidwell <nathan@codesourcery.com>
14935 Iain Sandoe <iain@codesourcery.com>
14936 Nathan Froyd <froydnj@codesourcery.com>
14937 Chao-ying Fu <fu@mips.com>
14939 * doc/extend.texi: (micromips, nomicromips, nocompression):
14940 Document new function attributes.
14941 * doc/invoke.texi (minterlink-compressed, mmicromips,
14942 m14k, m14ke, m14kec): Document new options.
14943 (minterlink-mips16): Update documentation.
14944 * doc/md.texi (ZC, ZD): Document new constraints.
14945 * configure.ac (gcc_cv_as_micromips): Check if linker
14946 supports the .set micromips directive.
14947 * configure: Regenerate.
14948 * config.in: Regenerate.
14949 * config/mips/mips-tables.opt: Regenerate.
14950 * config/mips/micromips.md: New file.
14951 * constraints.md (ZC, AD): New constraints.
14952 * config/mips/predicates.md (movep_src_register): New predicate.
14953 (movep_src_operand): New predicate.
14954 (non_volatile_mem_operand): New predicate.
14955 * config/mips/mips.md (multimem): New type.
14956 (length): Differentiate between 17-bit and 18-bit branch offsets.
14957 (MOVEP1, MOVEP2): New mode iterator.
14958 (mov_<load>l): Use ZC constraint.
14959 (mov_<load>r): Likewise.
14960 (mov_<store>l): Likewise.
14961 (mov_<store>r): Likewise.
14962 (*branch_equality<mode>_inverted): Add microMIPS support.
14963 (*branch_equality<mode>): Likewise.
14964 (*jump_absolute): Likewise.
14965 (indirect_jump_<mode>): Likewise.
14966 (tablejump_<mode>): Likewise.
14967 (<optab>_internal): Likewise.
14968 (sibcall_internal): Likewise.
14969 (sibcall_value_internal): Likewise.
14970 (prefetch): Use constraint ZD.
14971 * config/mips/mips.opt (minterlink-compressed): New option.
14972 (minterlink-mips16): Now an alias for minterlink-compressed.
14973 (mmicromips): New option.
14974 * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
14975 (compare_and_swap_12): Likewise.
14976 (sync_add<mode>): Likewise.
14977 (sync_<optab>_12): Likewise.
14978 (sync_old_<optab>_12): Likewise.
14979 (sync_new_<optab>_12): Likewise.
14980 (sync_nand_12): Likewise.
14981 (sync_old_nand_12): Likewise.
14982 (sync_new_nand_12): Likewise.
14983 (sync_sub<mode>): Likewise.
14984 (sync_old_add<mode>): Likewise.
14985 (sync_old_sub<mode>): Likewise.
14986 (sync_new_add<mode>): Likewise.
14987 (sync_new_sub<mode>): Likewise.
14988 (sync_<optab><mode>): Likewise.
14989 (sync_old_<optab><mode>): Likewise.
14990 (sync_new_<optab><mode>): Likewise.
14991 (sync_nand<mode>): Likewise.
14992 (sync_old_nand<mode>): Likewise.
14993 (sync_new_nand<mode>): Likewise.
14994 (sync_lock_test_and_set<mode>): Likewise.
14995 (test_and_set_12): Likewise.
14996 (atomic_compare_and_swap<mode>): Likewise.
14997 (atomic_exchange<mode>_llsc): Likewise.
14998 (atomic_fetch_add<mode>_llsc): Likewise.
14999 * config/mips/mips-cpus.def (m14kc, m14k): New processors.
15000 * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
15001 (umips_save_restore_pattern_p): Likewise.
15002 (umips_load_store_pair_p): Likewise.
15003 (umips_output_load_store_pair): Likewise.
15004 (umips_movep_target_p): Likewise.
15005 (umips_12bit_offset_address_p): Likewise.
15006 * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
15007 (mips_base_mips16): Rename this...
15008 (mips_base_compression_flags): ...to this. Update all uses.
15009 (mips_attribute_table): Add micromips, nomicromips and nocompression.
15010 (mips_mips16_decl_p): Delete.
15011 (mips_nomips16_decl_p): Delete.
15012 (mips_get_compress_on_flags): New function.
15013 (mips_get_compress_off_flags): New function.
15014 (mips_get_compress_mode): New function.
15015 (mips_get_compress_on_name): New function.
15016 (mips_get_compress_off_name): New function.
15017 (mips_insert_attributes): Support multiple compression types.
15018 (mips_merge_decl_attributes): Likewise.
15019 (umips_12bit_offset_address_p): New function.
15020 (mips_start_function_definition): Emit .set micromips directive.
15021 (mips_call_may_need_jalx_p): New function.
15022 (mips_function_ok_for_sibcall): Add microMIPS support.
15023 (mips_print_operand_punctuation): Support short delay slots and
15025 (umips_swm_mask, umips_swm_encoding): New.
15026 (umips_build_save_restore): New function.
15027 (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
15028 (was_mips16_p): Remove.
15029 (old_compression_mode): New.
15030 (mips_set_compression_mode): New function.
15031 (mips_set_current_function): Add microMIPS support.
15032 (mips_option_override): Likewise.
15033 (umips_save_restore_pattern_p): New function.
15034 (umips_output_save_restore): New function.
15035 (umips_load_store_pair_p_1): New function.
15036 (umips_load_store_pair_p): New function.
15037 (umips_output_load_store_pair_1): New function.
15038 (umips_output_load_store_pair): New function.
15039 (umips_movep_target_p) New function.
15040 (mips_prepare_pch_save): Add microMIPS support.
15041 * config/mips/mips.h (TARGET_COMPRESSION): New.
15042 (TARGET_CPU_CPP_BUILTINS): Update macro
15043 to use new compression flags and to support microMIPS.
15044 (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
15045 (MIPS_ARCH_FLOAT_SPEC): Likewise.
15046 (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
15047 (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
15048 (ASM_SPEC): Support mmicromips and mno-micromips.
15049 (M16STORE_REG_P): New macro.
15050 (MIPS_CALL): Support TARGET_MICROMIPS.
15051 (MICROMIPS_J): New macro.
15052 (mips_base_mips16): Rename this...
15053 (mips_base_compression_flags): ...to this.
15054 (UMIPS_12BIT_OFFSET_P): New macro.
15055 * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
15056 (MULTILIB_DIRNAMES): Likewise.
15058 2013-02-25 Tom de Vries <tom@codesourcery.com>
15060 PR rtl-optimization/56131
15061 * insn-notes.def (INSN_NOTE_BASIC_BLOCK): Update comment.
15062 * cfgrtl.c (delete_insn): Don't reorder NOTE_INSN_DELETED_LABEL and
15063 NOTE_INSN_BASIC_BLOCK if BLOCK_FOR_INSN == NULL.
15065 2013-02-25 Tobias Burnus <burnus@net-b.de>
15067 * doc/invoke.texi (-fsanitize=): Move from optimization
15068 to debugging options.
15070 2013-02-25 Andrey Belevantsev <abel@ispras.ru>
15072 * sched-deps.c (sched_analyze_insn): Fix typo in comment.
15074 2013-02-25 Andrey Belevantsev <abel@ispras.ru>
15075 Alexander Monakov <amonakov@ispras.ru>
15077 PR middle-end/56077
15078 * sched-deps.c (sched_analyze_insn): When reg_pending_barrier,
15079 flush pending lists also on non-jumps. Adjust comment.
15081 2013-02-24 Catherine Moore <clm@codesourcery.com>
15082 Maciej W. Rozycki <macro@codesourcery.com>
15083 Tom de Vries <tom@codesourcery.com>
15084 Nathan Sidwell <nathan@codesourcery.com>
15085 Iain Sandoe <iain@codesourcery.com>
15086 Nathan Froyd <froydnj@codesourcery.com>
15087 Chao-ying Fu <fu@mips.com>
15089 * doc/extend.texi: (micromips, nomicromips, nocompression):
15090 Document new function attributes.
15091 * doc/invoke.texi (minterlink-compressed, mmicromips,
15092 m14k, m14ke, m14kec): Document new options.
15093 (minterlink-mips16): Update documentation.
15094 * doc/md.texi (ZC, ZD): Document new constraints.
15095 * configure.ac (gcc_cv_as_micromips): Check if linker
15096 supports the .set micromips directive.
15097 * configure: Regenerate.
15098 * config.in: Regenerate.
15099 * config/mips/mips-tables.opt: Regenerate.
15100 * config/mips/micromips.md: New file.
15101 * constraints.md (ZC, AD): New constraints.
15102 * config/mips/predicates.md (movep_src_register): New predicate.
15103 (movep_src_operand): New predicate.
15104 (non_volatile_mem_operand): New predicate.
15105 * config/mips/mips.md (multimem): New type.
15106 (length): Differentiate between 17-bit and 18-bit branch offsets.
15107 (MOVEP1, MOVEP2): New mode iterator.
15108 (mov_<load>l): Use ZC constraint.
15109 (mov_<load>r): Likewise.
15110 (mov_<store>l): Likewise.
15111 (mov_<store>r): Likewise.
15112 (*branch_equality<mode>_inverted): Add microMIPS support.
15113 (*branch_equality<mode>): Likewise.
15114 (*jump_absolute): Likewise.
15115 (indirect_jump_<mode>): Likewise.
15116 (tablejump_<mode>): Likewise.
15117 (<optab>_internal): Likewise.
15118 (sibcall_internal): Likewise.
15119 (sibcall_value_internal): Likewise.
15120 (prefetch): Use constraint ZD.
15121 * config/mips/mips.opt (minterlink-compressed): New option.
15122 (minterlink-mips16): Now an alias for minterlink-compressed.
15123 (mmicromips): New option.
15124 * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
15125 (compare_and_swap_12): Likewise.
15126 (sync_add<mode>): Likewise.
15127 (sync_<optab>_12): Likewise.
15128 (sync_old_<optab>_12): Likewise.
15129 (sync_new_<optab>_12): Likewise.
15130 (sync_nand_12): Likewise.
15131 (sync_old_nand_12): Likewise.
15132 (sync_new_nand_12): Likewise.
15133 (sync_sub<mode>): Likewise.
15134 (sync_old_add<mode>): Likewise.
15135 (sync_old_sub<mode>): Likewise.
15136 (sync_new_add<mode>): Likewise.
15137 (sync_new_sub<mode>): Likewise.
15138 (sync_<optab><mode>): Likewise.
15139 (sync_old_<optab><mode>): Likewise.
15140 (sync_new_<optab><mode>): Likewise.
15141 (sync_nand<mode>): Likewise.
15142 (sync_old_nand<mode>): Likewise.
15143 (sync_new_nand<mode>): Likewise.
15144 (sync_lock_test_and_set<mode>): Likewise.
15145 (test_and_set_12): Likewise.
15146 (atomic_compare_and_swap<mode>): Likewise.
15147 (atomic_exchange<mode>_llsc): Likewise.
15148 (atomic_fetch_add<mode>_llsc): Likewise.
15149 * config/mips/mips-cpus.def (m14kc, m14k): New processors.
15150 * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
15151 (umips_save_restore_pattern_p): Likewise.
15152 (umips_load_store_pair_p): Likewise.
15153 (umips_output_load_store_pair): Likewise.
15154 (umips_movep_target_p): Likewise.
15155 (umips_12bit_offset_address_p): Likewise.
15156 * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
15157 (mips_base_mips16): Rename this...
15158 (mips_base_compression_flags): ...to this. Update all uses.
15159 (mips_attribute_table): Add micromips, nomicromips and nocompression.
15160 (mips_mips16_decl_p): Delete.
15161 (mips_nomips16_decl_p): Delete.
15162 (mips_get_compress_on_flags): New function.
15163 (mips_get_compress_off_flags): New function.
15164 (mips_get_compress_mode): New function.
15165 (mips_get_compress_on_name): New function.
15166 (mips_get_compress_off_name): New function.
15167 (mips_insert_attributes): Support multiple compression types.
15168 (mips_merge_decl_attributes): Likewise.
15169 (umips_12bit_offset_address_p): New function.
15170 (mips_start_function_definition): Emit .set micromips directive.
15171 (mips_call_may_need_jalx_p): New function.
15172 (mips_function_ok_for_sibcall): Add microMIPS support.
15173 (mips_print_operand_punctuation): Support short delay slots and
15175 (umips_swm_mask, umips_swm_encoding): New.
15176 (umips_build_save_restore): New function.
15177 (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
15178 (was_mips16_p): Remove.
15179 (old_compression_mode): New.
15180 (mips_set_compression_mode): New function.
15181 (mips_set_current_function): Add microMIPS support.
15182 (mips_option_override): Likewise.
15183 (umips_save_restore_pattern_p): New function.
15184 (umips_output_save_restore): New function.
15185 (umips_load_store_pair_p_1): New function.
15186 (umips_load_store_pair_p): New function.
15187 (umips_output_load_store_pair_1): New function.
15188 (umips_output_load_store_pair): New function.
15189 (umips_movep_target_p) New function.
15190 (mips_prepare_pch_save): Add microMIPS support.
15191 * config/mips/mips.h (TARGET_COMPRESSION): New.
15192 (TARGET_CPU_CPP_BUILTINS): Update macro
15193 to use new compression flags and to support microMIPS.
15194 (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
15195 (MIPS_ARCH_FLOAT_SPEC): Likewise.
15196 (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
15197 (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
15198 (ASM_SPEC): Support mmicromips and mno-micromips.
15199 (M16STORE_REG_P): New macro.
15200 (MIPS_CALL): Support TARGET_MICROMIPS.
15201 (MICROMIPS_J): New macro.
15202 (mips_base_mips16): Rename this...
15203 (mips_base_compression_flags): ...to this.
15204 (UMIPS_12BIT_OFFSET_P): New macro.
15205 * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
15206 (MULTILIB_DIRNAMES): Likewise.
15208 2013-02-24 Jakub Jelinek <jakub@redhat.com>
15211 * target-globals.c (save_target_globals): For init_reg_sets and
15212 target_reinit remporarily set this_fn_optabs to this_target_optabs.
15214 2013-02-22 James Grennahlgh <james.greenhalgh@arm.com>
15216 * config/aarch64/aarch64-simd-builtins.def: Add copyright header.
15217 * config/aarch64/t-aarch64
15218 (aarch64-builtins.o): Depend on aarch64-simd-builtins.def.
15220 2013-02-22 Vladimir Makarov <vmakarov@redhat.com>
15222 PR inline-asm/56148
15223 * lra-constraints.c (process_alt_operands): Reload operand
15224 conflicting with earlier clobber only if no more other conflicting
15227 2013-02-22 Jakub Jelinek <jakub@redhat.com>
15230 * config/gnu-user.h (LIBASAN_EARLY_SPEC): Link in libasan_preinit.o
15231 if not linking a shared library.
15233 2013-02-22 Seth LaForge <sethml@google.com>
15235 * config.gcc (arm*-*-eabi*): Treat arm*eb as big-endian.
15237 2013-02-22 Greta Yorsh <Greta.Yorsh@arm.com>
15239 * config/arm/arm.md (split for extendsidi): Update condition.
15240 (zero_extend<mode>di2,extend<mode>di2): Add an alternative.
15241 * config/arm/iterators.md (qhs_extenddi_cstr): Likewise.
15242 (qhs_zextenddi_cstr): Likewise.
15244 2013-02-21 Jakub Jelinek <jakub@redhat.com>
15246 PR middle-end/56420
15247 * expmed.c (EXACT_POWER_OF_2_OR_ZERO_P): Do subtraction in uhwi, to
15248 avoid signed wrapping.
15249 (expand_mult): Handle properly multiplication by
15250 ((dword_type) -1) << (BITS_PER_WORD - 1). Improve multiplication by
15251 ((dword_type) 1) << (BITS_PER_WORD - 1). Avoid undefined behavior
15252 in the compiler if coeff is HOST_WIDE_INT_MIN.
15253 (expand_divmod): Don't make ext_op1 static, change it's type to
15254 uhwi. Avoid undefined behavior in -INTVAL (op1).
15256 PR rtl-optimization/50339
15257 * lower-subreg.h (struct lower_subreg_choices): Add splitting_ashiftrt
15259 * lower-subreg.c (compute_splitting_shift): Handle ASHIFTRT.
15260 (compute_costs): Call compute_splitting_shift also for ASHIFTRT
15261 into splitting_ashiftrt field.
15262 (find_decomposable_shift_zext, resolve_shift_zext): Handle also
15264 (dump_choices): Fix up printing LSHIFTRT choices, print ASHIFTRT
15267 2013-02-20 Aldy Hernandez <aldyh@redhat.com>
15269 PR middle-end/56108
15270 * trans-mem.c (execute_tm_mark): Do not expand transactions that
15271 are sure to go irrevocable.
15273 2013-02-21 Hans-Peter Nilsson <hp@axis.com>
15275 * doc/rtl.texi (vec_concat, vec_duplicate): Mention that
15276 scalars are valid operands.
15278 2013-02-21 Martin Jambor <mjambor@suse.cz>
15280 PR tree-optimization/56310
15281 * ipa-cp.c (agg_replacements_to_vector): New parameter index, copy
15282 only matching indices and non-negative final offsets.
15283 (intersect_aggregates_with_edge): Pass src_idx to
15284 agg_replacements_to_vector. Pass src_idx insstead of index to
15285 intersect_with_agg_replacements.
15287 2013-02-21 Martin Jambor <mjambor@suse.cz>
15289 * ipa-cp.c (good_cloning_opportunity_p): Dump the real threshold
15290 instead of hard-wired defaults.
15292 2013-02-21 Maciej W. Rozycki <macro@codesourcery.com>
15294 * doc/invoke.texi (MIPS Options): Update documentation of the
15295 floating-point multiply-accumulate instruction restrictions.
15297 2013-02-21 Kostya Serebryany <kcc@google.com>
15299 * config/i386/i386.c (ix86_asan_shadow_offset): Use 0x7fff8000 as
15300 asan_shadow_offset on x86_64 linux.
15302 2013-02-21 Richard Biener <rguenther@suse.de>
15304 PR tree-optimization/56415
15306 2013-02-11 Richard Biener <rguenther@suse.de>
15308 PR tree-optimization/56273
15309 * tree-vrp.c (simplify_cond_using_ranges): Disable for the
15312 2013-02-21 Jakub Jelinek <jakub@redhat.com>
15315 * doc/invoke.texi (-fdump-rtl-pro_and_epilogue): Use @item
15318 PR inline-asm/56405
15319 * expr.c (expand_expr_real_1) <case TARGET_MEM_REF, MEM_REF>: Don't
15320 use movmisalign or extract_bit_field for EXPAND_MEMORY modifier.
15322 2013-02-20 Jan Hubicka <jh@suse.cz>
15324 PR tree-optimization/56265
15325 * ipa-prop.c (ipa_make_edge_direct_to_target): Fixup callgraph
15326 when target is referenced for first time.
15328 2013-02-20 Richard Biener <rguenther@suse.de>
15330 * tree-call-cdce.c (tree_call_cdce): Do not remove unused locals.
15331 * tree-ssa-forwprop.c (ssa_forward_propagate_and_combine): Likewise.
15332 * tree-ssa-dce.c (perform_tree_ssa_dce): Likewise.
15333 * tree-ssa-copyrename.c (copy_rename_partition_coalesce): Do
15334 not return anything.
15335 (rename_ssa_copies): Do not remove unused locals.
15336 * tree-ssa-ccp.c (do_ssa_ccp): Likewise.
15337 * tree-ssanames.c (pass_release_ssa_names): Remove unused locals first.
15338 * passes.c (execute_function_todo): Do not schedule unused locals
15339 removal if cleanup_tree_cfg did something.
15340 * tree-ssa-live.c (remove_unused_locals): Dump statistics
15341 about the number of removed locals.
15343 2013-02-20 Richard Biener <rguenther@suse.de>
15345 PR tree-optimization/56398
15346 * tree-vect-loop-manip.c (adjust_debug_stmts): Skip SSA default defs.
15348 2013-02-20 Martin Jambor <mjambor@suse.cz>
15350 PR tree-optimization/55334
15351 * ipa-cp.c (initialize_node_lattices): Disable IPA-CP through and to
15352 restricted pointers to arrays.
15354 2013-02-20 Richard Biener <rguenther@suse.de>
15355 Jakub Jelinek <jakub@redhat.com>
15357 PR tree-optimization/56396
15358 * tree-ssa-ccp.c (n_const_val): New static variable.
15359 (get_value): Return NULL for SSA names we don't have a lattice
15361 (ccp_initialize): Initialize n_const_val.
15362 * tree-ssa-copy.c (n_copy_of): New static variable.
15363 (init_copy_prop): Initialize n_copy_of.
15364 (get_value): Return NULL_TREE for SSA names we don't have a
15367 2013-02-20 Martin Jambor <mjambor@suse.cz>
15369 * ipa-cp.c (initialize_node_lattices): Fix dumping condition.
15371 2013-02-20 Richard Biener <rguenther@suse.de>
15373 * genpreds.c (write_lookup_constraint): Do not compare first
15374 letter of the constraint again.
15376 2013-02-20 Richard Biener <rguenther@suse.de>
15378 * tree-ssa-loop-ivopts.c (alloc_use_cost_map): Use bitmap_count_bits
15380 (get_use_iv_cost): Terminate hashtable walk when coming across
15383 2013-02-20 Igor Zamyatin <igor.zamyatin@intel.com>
15385 * config/i386/i386.c (initial_ix86_tune_features): Turn on fp
15386 reassociation for avx2 targets.
15388 2012-02-19 Edgar E. Iglesias <edgar.iglesias@gmail.com>
15390 * config/microblaze/microblaze.c: microblaze_has_clz = 0
15391 Add version check for v8.10.a to enable microblaze_has_clz
15392 * config/microblaze/microblaze.h: Add TARGET_HAS_CLZ as combined
15393 version and TARGET_PATTERN_COMPARE check
15394 * config/microblaze/microblaze.md: New clzsi2 instruction
15396 2012-02-19 Edgar E. Iglesias <edgar.iglesias@gmail.com>
15398 * config/microblaze/microblaze.md (call_value_intern): Check symbol is
15399 function before branching.
15401 2012-02-19 Andrey Belevantsev <abel@ispras.ru>
15403 * sel-sched-dump.c (dump_insn_rtx_flags): Explicitly set
15405 (dump_insn_rtx_1): Pass PATTERN (insn) to str_pattern_slim.
15407 2012-02-19 Andrey Belevantsev <abel@ispras.ru>
15409 PR middle-end/55889
15410 * sel-sched.c: Include ira.h.
15411 (implicit_clobber_conflict_p): New function.
15412 (moveup_expr): Use it.
15413 * Makefile.in (sel-sched.o): Depend on ira.h.
15415 2013-02-19 Richard Biener <rguenther@suse.de>
15417 PR tree-optimization/56384
15418 * tree-ssa-sccvn.h (struct vn_phi_s): Add type member.
15419 (vn_hash_type): Split out from ...
15420 (vn_hash_constant_with_type): ... here.
15421 * tree-ssa-sccvn.c (vn_phi_compute_hash): Use vn_hash_type.
15422 (vn_phi_eq): Compare types from vn_phi_s structure.
15423 (vn_phi_lookup): Populate vn_phi_s type.
15424 (vn_phi_insert): Likewise.
15426 2013-02-19 Jakub Jelinek <jakub@redhat.com>
15428 PR tree-optimization/56350
15429 * tree-vect-loop.c (vectorizable_reduction): If orig_stmt, return false
15430 if haven't found reduction or nested cycle operand, rather than
15431 asserting we must find it.
15433 PR tree-optimization/56381
15434 * tree-ssa-pre.c (create_expression_by_pieces): Fix up last argument
15437 2013-02-18 Aldy Hernandez <aldyh@redhat.com>
15438 Jakub Jelinek <jakub@redhat.com>
15441 * genopinit.c (raw_optab_handler): Use this_fn_optabs.
15442 (swap_optab_enable): Same.
15443 (init_all_optabs): Use argument instead of global.
15444 * tree.h (struct tree_optimization_option): New field target_optabs.
15445 * expr.h (init_all_optabs): Add argument to prototype.
15446 (TREE_OPTIMIZATION_OPTABS): New.
15447 (save_optabs_if_changed): Protoize.
15448 * optabs.h: Declare this_fn_optabs.
15449 * optabs.c (save_optabs_if_changed): New.
15450 Declare this_fn_optabs.
15451 (init_optabs): Add argument to init_all_optabs() call.
15452 * function.c (invoke_set_current_function_hook): Handle per
15454 * function.h (struct function): New field optabs.
15455 * config/mips/mips.c (mips_set_mips16_mode): Handle when
15456 optimization_current_node has changed.
15457 * target-globals.h (save_target_globals_default_opts): Protoize.
15458 * target-globals.c (save_target_globals_default_opts): New.
15460 2013-02-18 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
15463 * config/pa/pa.c (pa_conditional_register_usage): On HP-UX, mark
15464 registers %fr12 and %fr12R as call used.
15467 * config/pa/predicates.md (base14_operand): Except for BLKmode, QImode
15468 and HImode, require all displacements to be an integer multiple of
15470 * config/pa/pa.c (pa_legitimate_address_p): For REG+BASE addresses,
15471 only allow QImode and HImode when reload is in progress and strict is
15472 true. Likewise for symbolic addresses. Use base14_operand to check
15473 displacements in REG+BASE addresses.
15475 2013-02-18 Richard Biener <rguenther@suse.de>
15477 PR tree-optimization/56366
15478 * tree-vect-loop.c (get_initial_def_for_induction): Properly
15479 handle sign-conversion of outer-loop initial induction value.
15481 2013-02-18 Richard Biener <rguenther@suse.de>
15483 PR middle-end/56349
15484 * cfghooks.c (merge_blocks): If we merge a latch into another
15485 block adjust references to it.
15486 * cfgloop.c (flow_loops_find): Reset latch before recomputing it.
15487 (verify_loop_structure): Verify that a recorded latch is in fact
15490 2013-02-18 Richard Biener <rguenther@suse.de>
15492 PR tree-optimization/56321
15493 * tree-ssa-reassoc.c (propagate_op_to_single_use): Properly
15494 order SSA name release and virtual operand unlinking.
15496 2013-02-17 Edgar E. Iglesias <edgar.iglesias@gmail.com>
15498 * config/microblaze/microblaze.md (save_stack_block): Define.
15499 (restore_stack_block): Likewise.
15501 2013-02-16 Edgar E. Iglesias <edgar.iglesias@gmail.com>
15503 * config/microblaze/linux.h (TARGET_SUPPORTS_PIC): Define as 1.
15504 * config/microblaze/microblaze.h (TARGET_SUPPORTS_PIC): Define as 1.
15505 * config/microblaze/microblaze.c (microblaze_option_override):
15506 Bail out early for PIC modes when target does not support PIC.
15508 2013-02-16 Edgar E. Iglesias <edgar.iglesias@gmail.com>
15510 * config/microblaze/microblaze.c (microblaze_asm_trampoline_template):
15511 Replace with a microblaze version.
15512 (microblaze_trampoline_init): Adapt for microblaze.
15513 * config/microblaze/microblaze.h (TRAMPOLINE_SIZE): Adapt for
15516 2013-02-16 Jakub Jelinek <jakub@redhat.com>
15517 Dodji Seketeli <dodji@redhat.com>
15520 * asan.c (get_mem_refs_of_builtin_call): White space and style cleanup.
15521 (instrument_mem_region_access): Do not forget to always put
15522 instrumentation of the of 'base' and 'base + len' in a "if (len !=
15523 0) statement, even for cases where either 'base' or 'base + len'
15524 are not instrumented -- because they have been previously
15525 instrumented. Simplify the logic by putting all the statements
15526 instrument 'base + len' inside a sequence, and then insert that
15527 sequence right before the current insertion point. Then, to
15528 instrument 'base + len', just get an iterator on that statement.
15529 And do not forget to update the pointer to iterator the function
15530 received as argument.
15532 2013-02-15 Vladimir Makarov <vmakarov@redhat.com>
15534 PR rtl-optimization/56348
15535 * lra-assigns.c (reload_pseudo_compare_func): Prefer bigger pseudos.
15537 2013-02-15 Steven Bosscher <steven@gcc.gnu.org>
15539 * graph.c (start_graph_dump): Print dumpfile base as digraph label.
15540 (clean_graph_dump_file): Pass base to start_graph_dump.
15542 2013-02-14 Richard Henderson <rth@redhat.com>
15545 * lower-subreg.c (simple_move): Check dest mode instead of src mode.
15547 2013-02-14 Steven Bosscher <steven@gcc.gnu.org>
15549 * collect2-aix.h: Define F_LOADONLY.
15551 2013-02-14 Richard Biener <rguenther@suse.de>
15554 * varasm.c (output_constant_def_1): Get the decl representing
15555 the constant as argument.
15556 (output_constant_def): Wrap output_constant_def_1.
15557 (make_decl_rtl): Use output_constant_def_1 with the decl
15558 representing the constant.
15559 (build_constant_desc): Optionally re-use a decl already
15560 representing the constant.
15561 (tree_output_constant_def): Adjust.
15563 2013-02-14 Dodji Seketeli <dodji@redhat.com>
15566 * asan.c (instrument_builtin_call): Really put the length of the
15567 second source argument into src1_len.
15569 2013-02-13 Jakub Jelinek <jakub@redhat.com>
15571 * asan.c (create_cond_insert_point): Add create_then_fallthru_edge
15572 argument. If it is false, don't create edge from then_bb to
15574 (insert_if_then_before_iter): Pass true to it.
15575 (build_check_stmt): Pass false to it.
15576 (transform_statements): Flush hash table only on extended basic
15577 block boundaries, rather than at the beginning of every bb.
15578 Don't flush hash table on nonfreeing_call_p calls.
15579 * tree-flow.h (nonfreeing_call_p): New prototype.
15580 * tree-ssa-phiopt.c (nonfreeing_call_p): No longer static.
15582 2013-02-13 David S. Miller <davem@davemloft.net>
15584 * expmed.c (expand_shift_1): Only strip scalar integer subregs.
15586 2013-02-13 Vladimir Makarov <vmakarov@redhat.com>
15589 * ira.c (max_regno_before_ira): Move from ...
15591 (fix_reg_equiv_init): Use max_regno_before_ira instead of
15594 2013-02-13 Jakub Jelinek <jakub@redhat.com>
15596 * config/i386/i386.c (ix86_asan_shadow_offset): Revert last change.
15598 2013-02-13 Richard Biener <rguenther@suse.de>
15601 * gimple-streamer-out.c (output_gimple_stmt): Undo wrapping
15602 globals in MEM_REFs.
15604 2013-02-13 Richard Biener <rguenther@suse.de>
15606 * loop-init.c (loop_optimizer_init): Clear loop state when
15607 re-initializing preserved loops.
15608 * loop-unswitch.c (unswitch_single_loop): Return whether
15609 we unswitched the loop. Do not verify loop state here.
15610 (unswitch_loops): When we unswitched a loop discover new loops.
15612 2013-02-13 Kostya Serebryany <kcc@google.com>
15614 * config/i386/i386.c: Use 0x7fff8000 as asan_shadow_offset
15616 * sanitizer.def: Rename __asan_init to __asan_init_v1.
15618 2013-02-12 Dodji Seketeli <dodji@redhat.com>
15620 Avoid instrumenting duplicated memory access in the same basic block
15621 * Makefile.in (asan.o): Add new dependency on hash-table.h
15622 * asan.c (struct asan_mem_ref, struct mem_ref_hasher): New types.
15623 (asan_mem_ref_init, asan_mem_ref_get_end, get_mem_ref_hash_table)
15624 (has_stmt_been_instrumented_p, empty_mem_ref_hash_table)
15625 (free_mem_ref_resources, has_mem_ref_been_instrumented)
15626 (has_stmt_been_instrumented_p, update_mem_ref_hash_table)
15627 (get_mem_ref_of_assignment): New functions.
15628 (get_mem_refs_of_builtin_call): Extract from
15629 instrument_builtin_call and tweak a little bit to make it fit with
15631 (instrument_builtin_call): Use the new
15632 get_mem_refs_of_builtin_call. Use gimple_call_builtin_p instead
15633 of is_gimple_builtin_call.
15634 (instrument_derefs, instrument_mem_region_access): Insert the
15635 instrumented memory reference into the hash table.
15636 (maybe_instrument_assignment): Renamed instrument_assignment into
15637 this, and change it to advance the iterator when instrumentation
15638 actually happened and return true in that case. This makes it
15639 homogeneous with maybe_instrument_assignment, and thus give a
15640 chance to callers to be more 'regular'.
15641 (transform_statements): Clear the memory reference hash table
15642 whenever we enter a new BB, when we cross a function call, or when
15643 we are done transforming statements. Use
15644 maybe_instrument_assignment instead of instrumentation. No more
15645 need to special case maybe_instrument_assignment and advance the
15646 iterator after calling it; it's now handled just like
15647 maybe_instrument_call. Update comment.
15649 2013-02-13 Richard Biener <rguenther@suse.de>
15651 * config/mn10300/mn10300.c (mn10300_scan_for_setlb_lcc):
15652 Fix loop discovery code.
15654 2013-02-12 Vladimir Makarov <vmakarov@redhat.com>
15656 PR inline-asm/56148
15657 * lra-constraints.c (process_alt_operands): Match early clobber
15658 operand with itself. Check conflicts with earlyclobber only if
15659 the operand is not reloaded. Prefer to reload conflicting operand
15660 if earlyclobber and matching operands are the same.
15662 2013-02-12 Richard Biener <rguenther@suse.de>
15665 * lto-streamer-out.c (write_symbol): Do not output symbols
15666 for hard register variables.
15668 2013-02-12 Georg-Johann Lay <avr@gjlay.de>
15671 * config/avr/avr-dimode.md (umulsidi3, mulsidi3): New expanders.
15672 (umulsidi3_insn, mulsidi3_insn): New insns.
15674 2013-02-12 Christophe Lyon <christophe.lyon@linaro.org>
15676 * config/arm/arm-protos.h (struct cpu_vec_costs): New struct type.
15677 (struct tune_params): Add vec_costs field.
15678 * config/arm/arm.c (arm_builtin_vectorization_cost)
15679 (arm_add_stmt_cost): New functions.
15680 (TARGET_VECTORIZE_BUILTIN_VECTORIZATION_COST)
15681 (TARGET_VECTORIZE_ADD_STMT_COST): Define.
15682 (arm_default_vec_cost): New struct of type cpu_vec_costs.
15683 (arm_slowmul_tune, arm_fastmul_tune, arm_strongarm_tune)
15684 (arm_xscale_tune, arm_9e_tune, arm_v6t2_tune, arm_cortex_tune)
15685 (arm_cortex_a15_tune, arm_cortex_a5_tune, arm_cortex_a9_tune)
15686 (arm_v6m_tune, arm_fa726te_tune): Define new vec_costs field.
15688 2013-02-12 Richard Biener <rguenther@suse.de>
15691 * gimple-streamer-in.c (input_gimple_stmt): Strip MEM_REFs off
15692 decls again if possible.
15694 2013-02-12 Richard Biener <rguenther@suse.de>
15696 PR middle-end/56288
15697 * tree-ssa.c (verify_ssa_name): Fix check, move
15698 SSA_NAME_IN_FREE_LIST check up.
15700 2013-02-12 Jakub Jelinek <jakub@redhat.com>
15701 Steven Bosscher <steven@gcc.gnu.org>
15703 PR rtl-optimization/56151
15704 * optabs.c (add_equal_note): Don't return 0 if target is a MEM,
15705 equal to op0 or op1, and last_insn pattern is CODE operation
15706 with MEM dest and one of the operands matches that MEM.
15708 2013-02-11 Sriraman Tallam <tmsriramgoogle.com>
15710 * doc/extend.texi: Document Function Multiversioning and "default"
15711 parameter string to target attribute.
15712 * config/i386/i386.c (get_builtin_code_for_version): Return 0 if
15713 target attribute parameter is "default".
15714 (ix86_compare_version_priority): Remove checks for target attribute.
15715 (ix86_mangle_function_version_assembler_name): Change error to sorry.
15716 Remove check for target attribute equal to NULL. Add assert.
15717 (ix86_generate_version_dispatcher_body): Change error to sorry.
15719 2013-02-11 Iain Sandoe <iain@codesourcery.com>
15720 Jack Howarth <howarth@bromo.med.uc.edu>
15721 Patrick Marlier <patrick.marlier@gmail.com>
15724 * config/darwin.h: Replace ENDFILE_SPEC with TM_DESTRUCTOR and
15725 define ENDFILE_SPEC as TM_DESTRUCTOR.
15726 * config/i386/darwin.h (ENDFILE_SPEC): Use TM_DESTRUCTOR.
15728 2013-02-11 Alexander Potapenko <glider@google.com>
15729 Jack Howarth <howarth@bromo.med.uc.edu>
15730 Jakub Jelinek <jakub@redhat.com>
15733 * config/darwin.c (cdtor_record): Rename ctor_record.
15734 (sort_cdtor_records): Rename sort_ctor_records.
15735 (finalize_dtors): New routine to sort destructors by
15736 priority before use in assemble_integer.
15737 (machopic_asm_out_destructor): Use finalize_dtors if needed.
15739 2013-02-11 Uros Bizjak <ubizjak@gmail.com>
15741 PR rtl-optimization/56275
15742 * simplify-rtx.c (avoid_constant_pool_reference): Check that
15743 offset is non-negative and less than cmode size before
15744 calling simplify_subreg.
15746 2013-02-11 Richard Biener <rguenther@suse.de>
15748 PR tree-optimization/56264
15749 * cfgloop.h (fix_loop_structure): Adjust prototype.
15750 * loop-init.c (fix_loop_structure): Return the number of
15751 newly discovered loops.
15752 * tree-cfgcleanup.c (repair_loop_structures): When new loops
15753 are discovered, do a full loop-closed SSA rewrite.
15755 2013-02-11 Richard Biener <rguenther@suse.de>
15757 PR tree-optimization/56273
15758 * tree-vrp.c (simplify_cond_using_ranges): Disable for the
15760 (check_array_ref): Fix missing newline in dumps.
15761 (search_for_addr_array): Likewise.
15763 2013-02-09 David Edelsohn <dje.gcc@gmail.com>
15765 * config/rs6000/aix61.h (OS_MISSING_ALTIVEC): Undefine.
15767 2013-02-09 Jakub Jelinek <jakub@redhat.com>
15770 * config/rs6000/rs6000.h (ASSEMBLER_DIALECT): Define.
15772 2013-02-08 Vladimir Makarov <vmakarov@redhat.com>
15774 PR rtl-optimization/56246
15775 * lra-constraints.c (simplify_operand_subreg): Try to reuse
15777 * lra.c (lra): Clear lra_optional_reload_pseudos only when all
15778 constraints are satisfied.
15780 2013-02-08 Jeff Law <law@redhat.com>
15783 * emit-rtl.c (reg_is_parm_p): New function.
15784 * regs.h (reg_is_parm_p): New prototype.
15785 * ira-conflicts.c (ira_build_conflicts): Allow parameters in
15786 callee-clobbered registers.
15788 2013-02-08 Michael Meissner <meissner@linux.vnet.ibm.com>
15791 * config/rs6000/rs6000.c (rs6000_builtin_vectorized_libmass):
15792 If there is no implicit builtin declaration, just return NULL.
15794 2013-02-08 Uros Bizjak <ubizjak@gmail.com>
15796 * config/i386/sse.md (FMAMODEM): New mode iterator.
15797 (fma<mode>4, fms<mode>4, fnma<mode>4, fnms<mode>4): Use FMAMODEM
15798 mode iterator. Do not use TARGET_SSE_MATH in insn constraint.
15800 2013-02-08 Uros Bizjak <ubizjak@gmail.com>
15802 * config/i386/gnu-user.h (TARGET_CAN_SPLIT_STACK): Define only
15803 when HAVE_GAS_CFI_PERSONALITY_DIRECTIVE is set.
15804 * config/i386/gnu-user64.h (TARGET_CAN_SPLIT_STACK): Ditto.
15806 2013-02-08 Edgar E. Iglesias <edgar.iglesias@gmail.com>
15808 * config.gcc (microblaze*-linux*): Add TARGET_BIG_ENDIAN_DEFAULT.
15809 (microblaze*-*-elf): Likewise.
15810 * config/microblaze/linux.h: Add -mbig-endian / -mlittle-endian to
15812 * config/microblaze/microblaze-c.c: Add builtin defines for
15813 _LITTLE_ENDIAN and _BIG_ENDIAN.
15814 * config/microblaze/microblaze.h: Add TARGET_ENDIAN_DEFAULT and
15815 add to TARGET_DEFAULT flags.
15816 Expand ASM_SPEC and LINK_SPEC.
15817 Update BYTES_BIG_ENDIAN and WORDS_BIG_ENDIAN.
15818 * config/microblaze/microblaze.md: Update extendsidi2 and
15819 movdi_internal instructions to use low-order / high-order reg
15821 * config/microblaze/microblaze.opt: Add mbig-endian and mlittle-endian
15822 options and inversemask / mask of LITTLE_ENDIAN.
15823 * config/microblaze/t-microblaze: Expand multilib options to
15824 include mlittle-endian (le) and update exceptions patterns.
15826 2013-02-08 Jakub Jelinek <jakub@redhat.com>
15828 PR rtl-optimization/56195
15829 * lra-constraints.c (get_reload_reg): Don't reuse regs
15830 if they have smaller mode than requested, if they have
15831 wider mode than requested, try to return a SUBREG.
15833 PR tree-optimization/56250
15834 * fold-const.c (extract_muldiv_1) <case NEGATE_EXPR>: Don't optimize
15835 if type is unsigned and code isn't MULT_EXPR.
15837 2013-02-08 Georg-Johann Lay <avr@gjlay.de>
15839 PR tree-optimization/56064
15840 * fixed-value.c (fixed_from_double_int): Sign/zero extend payload
15841 bits according to mode.
15842 * fixed-value.h (fixed_from_double_int)
15843 (const_fixed_from_double_int): Adjust comments.
15845 2013-02-08 Richard Biener <rguenther@suse.de>
15848 * lto-streamer.h (struct data_in): Remove current_file, current_line
15849 and current_col members.
15850 * lto-streamer-out.c (lto_output_location): Stream changed bits
15851 en-block for efficiency.
15852 * lto-streamer-in.c (clear_line_info): Remove.
15853 (lto_input_location): Cache current file, line and column
15854 globally via local statics. Read changed bits en-block.
15855 (input_function): Do not call clear_line_info.
15856 (lto_read_body): Likewise.
15857 (lto_input_toplevel_asms): Likewise.
15859 2013-02-08 Michael Matz <matz@suse.de>
15861 PR tree-optimization/52448
15862 * tree-ssa-phiopt.c (struct name_to_bb): Add phase member.
15863 (nt_call_phase): New static.
15864 (add_or_mark_expr): Only mark accesses with newer phase than any
15866 (nonfreeing_call_p): New.
15867 (nt_init_block): Update nt_call_phase, mark blocks as visited.
15868 (nt_fini_block): Keep blocks marked as visited.
15869 (get_non_trapping): Initialize nt_call_phase, and reset aux pointer.
15871 2013-02-08 Richard Biener <rguenther@suse.de>
15873 * ira.c (ira): Free broken dominator information.
15875 2013-02-08 Uros Bizjak <ubizjak@gmail.com>
15877 * config/i386/i386.c (ix86_spill_class): Use INTEGER_CLASS_P macro.
15879 2013-02-08 Marek Polacek <polacek@redhat.com>
15881 * cfgloop.c (verify_loop_structure): Add more checking of headers.
15883 2013-02-08 Richard Biener <rguenther@suse.de>
15885 PR middle-end/56181
15886 * cfgloop.h (flow_loops_find): Adjust.
15887 (bb_loop_header_p): Declare.
15888 * cfgloop.c (bb_loop_header_p): New function split out from ...
15889 (flow_loops_find): ... here. Adjust function signature,
15890 support incremental loop structure update.
15891 (verify_loop_structure): Cleanup. Verify a loop is a loop.
15892 * cfgloopmanip.c (fix_loop_structure): Move ...
15893 * loop-init.c (fix_loop_structure): ... here.
15894 (apply_loop_flags): Split out from ...
15895 (loop_optimizer_init): ... here.
15896 (fix_loop_structure): Use apply_loop_flags. Use flow_loops_find
15897 in incremental mode, only remove dead loops here.
15899 2013-02-08 Georg-Johann Lay <avr@gjlay.de>
15902 * config/avr/avr.md (unspec) <UNSPEC_ROUND>: Add.
15903 * config/avr/avr-fixed.md (ALL4QA, ALL124QA): New mode iterators.
15904 (round<mode>3, round<mode>3_const): New expanders for fixed-mode.
15905 (*round<mode>3.libgcc): New insns for fixed-modes.
15906 * config/avr/builtins.def (ABSxx): Use a non-NULL LIBNAME.
15907 (ROUNDxx, COUNTLSxx, BITSxx, xxBITS): New DEF_BUILTINs.
15908 (ROUNDFX, COUNTLSFX, ABSFX): New DEF_BUILTINs.
15909 * config/avr/stdfix.h (absFX, bitsFX, FXbits): Remove inline
15910 implementations. Define to __builtin_avr_absFX,
15911 __builtin_avr_bitsFX, __builtin_avr_FXbits, respectively.
15912 (roundFX, countlsFX): Define to __builtin_avr_roundFX,
15913 __builtin_avr_countlsFX, respectively.
15914 * config/avr/avr-c.c (target.h): Include it.
15915 (enum avr_builtin_id): New enum.
15916 (avr_resolve_overloaded_builtin): New static function.
15917 (avr_register_target_pragmas): Use it to set
15918 targetm.resolve_overloaded_builtin.
15919 * config/avr/avr.c (avr_init_builtins): Supply myriads of local
15920 tree nodes used by DEF_BUILTIN.
15921 (avr_expand_builtin) <AVR_BUILTIN_ROUNDxx>: Sanity-check them.
15922 (avr_fold_builtin) <AVR_BUILTIN_BITSxx>: Fold to VIEW_COVERT_EXPR.
15923 <AVR_BUILTIN_xxBITS>: Same.
15925 2013-02-08 Richard Biener <rguenther@suse.de>
15927 * cfgloop.c (verify_loop_structure): Properly handle
15928 a loop exiting to another loop header.
15929 * ira-int.h (ira_loops): Remove.
15930 * ira.c (ira_loops): Remove.
15931 (ira): Use loop_optimizer_init and loop_optimizer_finalize.
15932 (do_reload): Use loop_optimizer_finalize.
15933 * ira-build.c (create_loop_tree_nodes): Use get_loops and
15934 number_of_loops to access the loop tree.
15935 (more_one_region_p): Likewise.
15936 (finish_loop_tree_nodes): Likewise.
15937 (rebuild_regno_allocno_maps): Likewise.
15938 (mark_loops_for_removal): Likewise.
15939 (mark_all_loops_for_removal): Likewise.
15940 (remove_unnecessary_regions): Likewise.
15941 (ira_build): Likewise.
15942 * ira-emit.c (setup_entered_from_non_parent_p): Likewise.
15944 2013-02-08 Richard Biener <rguenther@suse.de>
15946 * Makefile.in (tree-tailcall.o): Add $(CFGLOOP_H) dependency.
15947 * ipa-pure-const.c (analyze_function): Avoid calling
15948 mark_irreducible_loops twice.
15949 * tree-tailcall.c (tree_optimize_tail_calls_1): Mark loops for fixup.
15951 2013-02-07 David S. Miller <davem@davemloft.net>
15953 * dwarf2out.c (based_loc_descr): Perform leaf register remapping
15955 * var-tracking.c (vt_add_function_parameter): Test the presence of
15956 HAVE_window_save properly and do not remap argument registers when
15957 we have a leaf function.
15959 2013-02-07 Uros Bizjak <ubizjak@gmail.com>
15962 * ggc-page.c (ggc_print_statistics): Use HOST_LONG_LONG_FORMAT
15964 * config/i386/i386.c (ix86_print_operand): Ditto.
15966 2013-02-07 Vladimir Makarov <vmakarov@redhat.com>
15968 * lra-constraints.c (process_alt_operands): Fix recently added comment.
15970 2013-02-07 Vladimir Makarov <vmakarov@redhat.com>
15972 PR rtl-optimization/56225
15973 * lra-constraints.c (process_alt_operands): Check that reload hard
15974 reg can hold value for strict_low_part.
15976 2013-02-07 Jakub Jelinek <jakub@redhat.com>
15979 * dwarf2out.c (dwarf2_debug_hooks): Set end_function hook to
15980 dwarf2out_end_function.
15981 (in_first_function_p, maybe_at_text_label_p,
15982 first_loclabel_num_not_at_text_label): New variables.
15983 (dwarf2out_var_location): In the first function find out
15984 lowest loclabel_num N where .LVLN is known not to be equal to .Ltext0.
15985 (find_empty_loc_ranges_at_text_label, dwarf2out_end_function): New
15988 2013-02-07 Eric Botcazou <ebotcazou@adacore.com>
15990 PR rtl-optimization/56178
15991 * cse.c (cse_insn): Do not create a REG_EQUAL note if the source is a
15992 SUBREG of a register. Tidy up related block of code.
15993 * fwprop.c (forward_propagate_and_simplify): Do not create a REG_EQUAL
15994 note if the source is a register or a SUBREG of a register.
15996 2013-02-07 Jakub Jelinek <jakub@redhat.com>
15999 * config/rs6000/rs6000.md (ptrm): New mode attr.
16000 (call_indirect_aix<ptrsize>, call_indirect_aix<ptrsize>_nor11,
16001 call_value_indirect_aix<pttrsize>,
16002 call_value_indirect_aix<pttrsize>_nor11): Use <ptrm> instead of
16005 2013-02-07 Michael Haubenwallner <michael.haubenwallner@salomon.at>
16007 * collect2.c (main): Set aix64_flag for -G and -bsvr4 too, disable
16008 if -bnortl. Convert to strcmp and strncmp.
16010 2013-02-07 Alan Modra <amodra@gmail.com>
16013 * config/rs6000/rs6000.c (mem_operand_gpr): Check that LO_SUM
16014 addresses won't wrap when offsetting.
16015 (rs6000_secondary_reload): Provide secondary reloads needed for
16016 wrapping LO_SUM addresses.
16018 2013-02-06 Thomas Schwinge <thomas@codesourcery.com>
16020 * config/gnu.h (GNU_USER_TARGET_OS_CPP_BUILTINS): Never define
16021 MACH, just __MACH__.
16023 2013-02-06 Richard Biener <rguenther@suse.de>
16025 * tracer.c (tracer): Mark loops with LOOPS_NEED_FIXUP
16026 instead of calling fix_loop_structure.
16028 2013-02-06 Jakub Jelinek <jakub@redhat.com>
16030 PR middle-end/56217
16031 * omp-low.c (use_pointer_for_field): Return false if
16032 lower_send_shared_vars doesn't generate any copy-out code.
16034 2013-02-06 Tom de Vries <tom@codesourcery.com>
16036 PR rtl-optimization/56131
16037 * cfgrtl.c (delete_insn): Use NOTE_BASIC_BLOCK instead of BLOCK_FOR_INSN
16038 to get the bb of a NOTE_INSN_BASIC_BLOCK. Handle the case that the bb
16039 of the label is NULL. Add comment.
16041 2013-02-05 Jakub Jelinek <jakub@redhat.com>
16043 * tree.h (struct tree_decl_with_vis): Remove thread_local field.
16046 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Define.
16047 (STATIC_LIBTSAN_LIBS): Likewise.
16048 * gcc.c (ADD_STATIC_LIBTSAN_LIBS, LIBTSAN_EARLY_SPEC): Define.
16049 (LIBTSAN_SPEC): Add ADD_STATIC_LIBTSAN_LIBS, if LIBTSAN_EARLY_SPEC
16050 is defined, don't add anything else beyond that.
16051 (SANITIZER_EARLY_SPEC, SANITIZER_SPEC): Define.
16052 (LINK_COMMAND_SPEC): Use them.
16054 PR tree-optimization/56205
16055 * tree-stdarg.c (check_all_va_list_escapes): Return true if
16056 there are any PHI nodes that set non-va_list_escape_vars SSA_NAME
16057 and some va_list_escape_vars SSA_NAME appears in some PHI argument.
16059 2013-02-05 Richard Biener <rguenther@suse.de>
16061 PR tree-optimization/53342
16062 PR tree-optimization/53185
16063 * tree-vectorizer.h (vect_check_strided_load): Remove.
16064 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Do
16065 not disallow peeling for vectorized strided loads.
16066 (vect_check_strided_load): Make static and simplify.
16067 (vect_analyze_data_refs): Adjust.
16068 * tree-vect-stmts.c (vectorizable_load): Handle peeled loops
16069 correctly when vectorizing strided loads.
16071 2013-02-05 Richard Biener <rguenther@suse.de>
16073 * doc/install.texi: Refer to ISL, not PPL.
16075 2013-02-05 Jan Hubicka <jh@suse.cz>
16077 PR tree-optimization/55789
16078 * params.def (PARAM_EARLY_INLINER_MAX_ITERATIONS): Drop to 1.
16080 2013-02-05 Jan Hubicka <jh@suse.cz>
16082 PR tree-optimization/55789
16083 * cgraphclones.c (cgraph_remove_node_and_inline_clones): Remove
16084 the dead call anyway.
16086 2013-02-05 Eric Botcazou <ebotcazou@adacore.com>
16089 * config/gnu-user.h (LIBASAN_EARLY_SPEC): Add missing guard.
16091 2013-02-04 Alexander Potapenko <glider@google.com>
16092 Jack Howarth <howarth@bromo.med.uc.edu>
16093 Jakub Jelinek <jakub@redhat.com>
16096 * config/darwin.c (sort_ctor_records): Stabilized qsort
16097 on constructor priority by using original position.
16098 (finalize_ctors): New routine to sort constructors by
16099 priority before use in assemble_integer.
16100 (machopic_asm_out_constructor): Use finalize_ctors if needed.
16102 2013-02-04 Jakub Jelinek <jakub@redhat.com>
16105 * config/i386/winnt.c (i386_pe_assemble_visibility): Don't warn
16106 about visibility on artificial decls.
16107 * config/sol2.c (solaris_assemble_visibility): Likewise.
16109 2013-02-04 Kai Tietz <ktietz@redhat.com>
16112 * config/i386/i386.c (function_value_ms_64): Add additional valtype
16113 argument and improve checking of return-argument types for 16-byte
16115 (ix86_function_value_1): Add additional valtype argument on call
16116 of function_value_64.
16117 (return_in_memory_ms_64): Sync 16-byte sized mode handling with
16118 handling infunction_value_64 function.
16120 2013-02-04 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
16122 * reload.c (subst_reloads): Fix DEBUG_RELOAD build issue.
16124 2013-02-04 Richard Biener <rguenther@suse.de>
16126 PR tree-optimization/56188
16127 * tree-ssa-structalias.c (label_visit): Consider case with
16128 initially non-empty points-to set.
16129 (perform_var_substitution): Dump node mapping and clean up.
16131 2013-02-04 Richard Guenther <rguenther@suse.de>
16134 * lto-symtab.c (lto_symtab_merge_decls_1): Make non-builtin
16135 node prevail as last resort.
16136 (lto_symtab_merge_decls): Remove guard on LTRANS here.
16137 (lto_symtab_prevailing_decl): Builtins are their own prevailing decl.
16139 2013-02-04 Richard Biener <rguenther@suse.de>
16141 PR tree-optimization/56113
16142 * tree-ssa-structalias.c (equiv_class_lookup, equiv_class_add):
16144 (equiv_class_lookup_or_add): ... this.
16145 (label_visit): Adjust and fix error in previous patch.
16146 (perform_var_substitution): Adjust.
16148 2013-02-03 Oleg Endo <olegendo@gcc.gnu.org>
16150 * config/sh/divtab.c: Fix formatting and comments throughout the file.
16151 * config/sh/sh4-300.md: Likewise.
16152 * config/sh/sh4a.md: Likewise.
16153 * config/sh/constraints.md: Likewise.
16154 * config/sh/sh.md: Likewise.
16155 * config/sh/netbsd-elf.h: Likewise.
16156 * config/sh/predicates.md: Likewise.
16157 * config/sh/sh-protos.h: Likewise.
16158 * config/sh/ushmedia.h: Likewise.
16159 * config/sh/linux.h: Likewise.
16160 * config/sh/sh.c: Likewise.
16161 * config/sh/superh.h: Likewise.
16162 * config/sh/elf.h: Likewise.
16163 * config/sh/sh4.md: Likewise.
16164 * config/sh/sh.h: Likewise.
16166 2013-02-03 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
16168 * config/pa/constraints.md: Adjust unused letters. Change "T"
16169 constraint to match_test floating_point_store_memory_operand().
16170 * config/pa/predicates.md (reg_plus_base_memory_operand): New.
16171 (base14_operand): New.
16172 (floating_point_store_memory_operand): New.
16173 (integer_store_memory_operand): Revise to use base14_operand and
16174 reg_plus_base_memory_operand.
16175 (move_dest_operand): Allow symbolic_memory_operands.
16176 (symbolic_memory_operand): Check for LO_SOM.
16177 (symbolic_operand): Change default case to break.
16178 * config/pa/pa.md: Remove unamed DFmode and SFmode patterns to force
16179 CONST_DOUBLE values to be reloaded by putting them into memory when
16180 the destination is a floating point register.
16181 (movdf): Remove code to handle CONST_DOUBLE.
16183 (reload_indf_r1): New.
16184 (reload_insf_r1): New.
16185 Consistently use "Q" and "T" constraints with integer and floating
16186 point move instructions, respectively.
16187 (movdi): Remove FAIL.
16188 Change predicate for source operand unamed DImode move from
16189 general_operand to move_src_operand.
16190 (umulsidi3): Change predicate for destination operand to
16192 Likewise for similar unamed patterns.
16193 * config/pa/pa-protos.h (pa_legitimize_reload_address): Declare.
16194 * config/pa/pa.c (pa_symbolic_expression_p): Remove extra parenthesis.
16195 (hppa_legitimize_address): Simplify mask calculation.
16196 (pa_emit_move_sequence): Revised handling of secondary reloads from
16197 REG+D addresses for floating point loads and stores. Directly handle
16198 loading CONST0_RTX (mode) to a floating point register.
16199 (pa_secondary_reload): Handle reloading DF and SFmode constant values
16200 to floating point registers. Don't restrict secondary reloads to
16201 floating point registers to integer modes. Revise some comments and
16203 (TARGET_LEGITIMATE_ADDRESS_P): Define.
16204 (pa_legitimate_address_p): New.
16205 (pa_legitimize_reload_address): New.
16206 * config/pa/pa.h (STRICT_REG_OK_FOR_INDEX_P): New.
16207 (STRICT_REG_OK_FOR_BASE_P): New.
16208 (GO_IF_LEGITIMATE_ADDRESS): Delete. Update some related comments.
16209 (LEGITIMIZE_RELOAD_ADDRESS): Revise to use pa_legitimize_reload_address.
16211 2013-02-03 David Edelsohn <dje.gcc@gmail.com>
16212 Andrew Dixie <andrewd@gentrack.com>
16214 * collect2.c (GCC_CHECK_HDR): Do not scan objects with F_LOADONLY
16217 2013-02-03 Richard Sandiford <rdsandiford@googlemail.com>
16219 * expmed.c (extract_bit_field_1): Pass the full width of the
16220 structure to get_best_reg_extraction_insn.
16222 2013-02-01 David Edelsohn <dje.gcc@gmail.com>
16225 * configure.ac (use_cxa_atexit): Add AIX.
16226 * configure: Regenerate.
16228 * config/rs6000/aix61.h (STARTFILE_SPEC): Add crtcxa.o.
16230 2013-02-01 Jakub Jelinek <jakub@redhat.com>
16233 * final.c (need_profile_function): New variable.
16234 (final_start_function): Drop ATTRIBUTE_UNUSED from first argument.
16235 If first of NOTE_INSN_BASIC_BLOCK or NOTE_INSN_FUNCTION_BEG
16236 is only preceeded by NOTE_INSN_VAR_LOCATION or NOTE_INSN_DELETED
16237 notes, targetm.asm_out.function_prologue doesn't emit anything,
16238 HAVE_prologue and profiler should be emitted before prologue,
16239 set need_profile_function instead of emitting it.
16240 (final_scan_insn): If need_profile_function, emit
16241 profile_function on the first NOTE_INSN_BASIC_BLOCK or
16242 NOTE_INSN_FUNCTION_BEG note.
16244 2013-02-01 Richard Henderson <rth@redhat.com>
16246 * config/rs6000/rs6000.md (smulditi3): New.
16249 * config/alpha/alpha.md (umulditi3): New.
16251 2013-02-01 David Edelsohn <dje.gcc@gmail.com>
16253 * config/rs6000/xcoff.h (ASM_OUTPUT_ALIGNED_COMMON): Use floor_log2.
16254 (ASM_OUTPUT_ALIGNED_LOCAL): New.
16256 2013-02-01 Richard Biener <rguenther@suse.de>
16258 PR tree-optimization/56113
16259 * tree-ssa-structalias.c (label_visit): Reduce work for
16260 single-predecessor nodes.
16262 2013-02-01 Eric Botcazou <ebotcazou@adacore.com>
16264 * fold-const.c (make_range_step) <TRUTH_NOT_EXPR>: Bail out if the
16265 range isn't testing for zero.
16267 2013-01-31 Steven Bosscher <steven@gcc.gnu.org>
16269 PR middle-end/56113
16270 * fwprop.c (fwprop_init): Set up loops without CFG modifications.
16272 2013-01-31 Hiroyuki Ono <hiroyuki.ono.jc@renesas.com>
16273 Nick Clifton <nickc@redhat.com>
16275 * config/v850/constraints.md (Q): Define as a memory constraint.
16276 * config/v850/predicates.md (label_ref_operand): New predicate.
16277 (e3v5_shift_operand): New predicate.
16278 (ior_operator): New predicate.
16279 * config/v850/t-v850: Add e3v5 multilib.
16280 * config/v850/v850-protos.h (v850_adjust_insn_length): Prototype.
16281 (v850_gen_movdi): Prototype.
16282 * config/v850/v850.c: Add support for e3v5 architecture.
16283 Rename all uses of TARGET_V850E || TARGET_V850E2_ALL to
16285 (construct_save_jarl): Add e3v5 long JARL support.
16286 (v850_adjust_insn_length): New function. Adjust length of call
16287 insns when using e3v5 instructions.
16288 (v850_gen_movdi): New function: Generate instructions to move a
16290 * config/v850/v850.h (TARGET_CPU_v850e3v5): Define.
16291 (CPP_SPEC): Define __v850e3v5__ as appropriate.
16292 (TARGET_USE_FPU): Enable for e3v5.
16293 (CONST_OK_FOR_W): New macro.
16294 (ADJUST_INSN_LENGTH): Define.
16295 * config/v850/v850.md (UNSPEC_LOOP): Define.
16296 (attr cpu): Add v850e3v5.
16297 Rename all uses of TARGET_V850E2 to TARGET_V850E2V3_UP.
16298 (movdi): New pattern.
16299 (movdi_internal): New pattern.
16300 (cbranchsf4): Conditionalize on TARGET_USE_FPU.
16301 (cbranchdf4): Conditionalize on TARGET_USE_FPU.
16302 (cstoresf4): Likewise.
16303 (cstoredf4): Likewise.
16304 (insv): New pattern.
16305 (rotlso3_a): New pattern.
16306 (rotlsi3_b): New pattern
16307 (rotlsi3_v850e3v5): New pattern.
16308 (doloop_begin): New pattern.
16309 (fix_loop_counter): New pattern.
16310 (doloop_end): New pattern.
16311 (branch_normal): Add e3v5 long branch support.
16312 (branch_invert): Likewise.
16313 (branch_z_normal): Likewise.
16314 (branch_z_invert): Likewise.
16315 (branch_nz_normal): Likewise.
16316 (branch_nz_invert): Likewise.
16317 (call_internal_short): Add e3v5 register-indirect JARL support.
16318 (call_internal_long): Likewise.
16319 (call_value_internal_short): Likewise.
16320 (call_value_internal_long): Likewise.
16321 * config/v850/v850.opt (mv850e3v5, mv850e2v4): New options.
16322 (mloop): New option.
16323 * config.gcc: Add support for configuring v840e3v5 target.
16324 * doc/invoke.texi: Document new v850 specific command line options.
16326 2013-01-31 Paul Koning <ni1d@arrl.net>
16330 * dwarf2out.c (prune_unused_types_mark): Mark all of parent's
16331 children if parent is a class.
16332 (prune_unused_types_prune): Don't add DW_AT_declaration.
16334 2013-01-31 Richard Biener <rguenther@suse.de>
16336 PR tree-optimization/56157
16337 * tree-vect-slp.c (vect_get_slp_defs): More thoroughly try to
16338 match up operand with SLP child.
16340 2013-01-31 Jason Merrill <jason@redhat.com>
16343 * dwarf2out.c (gen_struct_or_union_type_die): Always schedule template
16344 parameters the first time.
16345 (gen_scheduled_generic_parms_dies): Check completeness here.
16347 2013-01-31 Richard Biener <rguenther@suse.de>
16349 PR middle-end/53073
16350 * common.opt (faggressive-loop-optimizations): New flag,
16351 enabled by default.
16352 * doc/invoke.texi (faggressive-loop-optimizations): Document.
16353 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Guard
16354 infer_loop_bounds_from_undefined by it.
16356 2013-01-31 Richard Biener <rguenther@suse.de>
16358 PR tree-optimization/56150
16359 * tree-ssa-loop-manip.c (find_uses_to_rename_stmt): Do not
16360 visit virtual operands.
16361 (find_uses_to_rename_bb): Likewise.
16363 2013-01-31 Richard Biener <rguenther@suse.de>
16365 PR tree-optimization/56150
16366 * tree-ssa-tail-merge.c (gimple_equal_p): Properly handle
16367 mixed store non-store stmts.
16369 2013-01-30 Jakub Jelinek <jakub@redhat.com>
16372 * gcc.c (LIBASAN_SPEC): Define just to ADD_STATIC_LIBASAN_LIBS if
16373 LIBASAN_EARLY_SPEC is defined.
16374 (LIBASAN_EARLY_SPEC): Define to empty string if not already defined.
16375 (LINK_COMMAND_SPEC): Add LIBASAN_EARLY_SPEC for -fsanitize=address,
16377 * config/gnu-user.h (LIBASAN_EARLY_SPEC): Define.
16380 * config/i386/i386.c (ix86_valid_target_attribute_inner_p): Diagnose
16381 invalid args instead of ICEing on it.
16382 (ix86_valid_target_attribute_tree): Return error_mark_node if
16383 ix86_valid_target_attribute_inner_p failed.
16384 (ix86_valid_target_attribute_p): Return false only if
16385 ix86_valid_target_attribute_tree returned error_mark_node. Allow
16386 target("default") attribute.
16387 (sorted_attr_string): Change argument from const char * to tree,
16388 merge in all target attribute arguments rather than just one.
16389 Formatting fix. Use XNEWVEC instead of xmalloc and XDELETEVEC
16390 instead of free. Avoid using strcat.
16391 (ix86_mangle_function_version_assembler_name): Mangle
16392 target("default") as if no target attribute is present. Adjust
16393 sorted_attr_string caller. Avoid leaking memory. Use XNEWVEC
16394 instead of xmalloc and XDELETEVEC instead of free.
16395 (ix86_function_versions): Don't return true if one of the decls
16396 doesn't have target attribute. If they don't and one of the decls
16397 is DECL_FUNCTION_VERSIONED, report an error. Adjust
16398 sorted_attr_string caller. Use XDELETEVEC instead of free.
16399 (ix86_supports_function_versions): Remove.
16400 (make_name): Fix up formatting.
16401 (make_dispatcher_decl): Remove resolver_name and its initialization.
16402 Avoid leaking memory.
16403 (is_function_default_version): Return true if there is
16404 target("default") attribute rather than no target attribute at all.
16405 (make_resolver_func): Avoid leaking memory.
16406 (ix86_generate_version_dispatcher_body): Likewise.
16407 (TARGET_OPTION_SUPPORTS_FUNCTION_VERSIONS): Remove.
16408 * target.def (supports_function_versions): Remove.
16409 * doc/tm.texi.in (SUPPORTS_FUNCTION_VERSIONS): Remove.
16410 * doc/tm.texi: Regenerated.
16412 2013-01-30 Vladimir Makarov <vmakarov@redhat.com>
16414 PR rtl-optimization/56144
16415 * lra-constraints.c (get_reload_reg): Don't reuse reload pseudo
16416 for values with side effects.
16418 2013-01-30 Richard Biener <rguenther@suse.de>
16420 * sparseset.h (sparseset_bit_p): Use gcc_checking_assert.
16421 (sparseset_pop): Likewise.
16422 * cfganal.c (compute_idf): Likewise. Increase work-stack size
16423 to be able to use quick_push in the worker loop.
16425 2013-01-30 Marek Polacek <polacek@redhat.com>
16427 * cfgcleanup.c (cleanup_cfg): Don't mark affected BBs.
16429 2013-01-30 Richard Biener <rguenther@suse.de>
16432 * lto-symtab.c (lto_symtab_merge_decls_1): Guard DECL_BUILT_IN check.
16434 2013-01-30 Georg-Johann Lay <avr@gjlay.de>
16436 PR tree-optimization/56064
16437 * fixed-value.c (fixed_from_double_int): New function.
16438 * fixed-value.h (fixed_from_double_int): New prototype.
16439 (const_fixed_from_double_int): New static inline function.
16440 * fold-const.c (native_interpret_fixed): New static function.
16441 (native_interpret_expr) <FIXED_POINT_TYPE>: Use it.
16442 (can_native_interpret_type_p) <FIXED_POINT_TYPE>: Return true.
16443 (native_encode_fixed): New static function.
16444 (native_encode_expr) <FIXED_CST>: Use it.
16445 (native_interpret_int): Move double_int worker code to...
16446 * double-int.c (double_int::from_buffer): ...this new static method.
16447 * double-int.h (double_int::from_buffer): Prototype it.
16449 2013-01-30 Richard Biener <rguenther@suse.de>
16451 * tree-ssa-structalias.c (final_solutions, final_solutions_obstack):
16452 New pointer-map and obstack.
16453 (init_alias_vars): Allocate pointer-map and obstack.
16454 (delete_points_to_sets): Free them.
16455 (find_what_var_points_to): Cache result.
16456 (find_what_p_points_to): Adjust for changed interface of
16457 find_what_var_points_to.
16458 (compute_points_to_sets): Likewise.
16459 (ipa_pta_execute): Likewise.
16461 2013-01-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
16463 * configure.ac (HAVE_AS_SPARC_NOBITS): New test.
16464 * configure: Regenerate.
16465 * config.in: Regenerate.
16466 * config/sparc/sparc.c (sparc_solaris_elf_asm_named_section): Emit
16467 #nobits/#progbits if supported.
16469 2013-01-29 Oleg Endo <olegendo@gcc.gnu.org>
16472 * config/sh/sh.md (bclr_m2a, bset_m2a, bst_m2a, bld_m2a, bldsign_m2a,
16473 bld_reg, *bld_regqi, band_m2a, bandreg_m2a, bor_m2a, borreg_m2a,
16474 bxor_m2a, bxorreg_m2a): Add satisfies_constraint_K03 condition.
16476 2013-01-29 Greta Yorsh <Greta.Yorsh@arm.com>
16478 * config/arm/cortex-a7.md (cortex_a7_neon, cortex_a7_all): Remove.
16479 (cortex_a7_idiv): Use cortex_a7_both instead of cortex_a7_all.
16481 2013-01-29 Greta Yorsh <Greta.Yorsh@arm.com>
16483 * config/arm/arm.c (cortexa7_younger): Return true for TYPE_CALL.
16484 * config/arm/cortex-a7.md (cortex_a7_call): Update required units.
16486 2013-01-29 Greta Yorsh <Greta.Yorsh@arm.com>
16488 * config/arm/arm-protos.h (arm_mac_accumulator_is_result): New
16490 * config/arm/arm.c (arm_mac_accumulator_is_result): New function.
16491 * config/arm/cortex-a7.md: New bypasses using
16492 arm_mac_accumulator_is_result.
16494 2013-01-29 Greta Yorsh <Greta.Yorsh@arm.com>
16496 * config/arm/cortex-a7.md (cortex_a7_neon_mul): New reservation.
16497 (cortex_a7_neon_mla): Likewise.
16498 (cortex_a7_fpfmad): New reservation.
16499 (cortex_a7_fpmacs): Use ffmas and update required units.
16500 (cortex_a7_fpmuld): Update required units and latency.
16501 (cortex_a7_fpmacd): Likewise.
16502 (cortex_a7_fdivs, cortex_a7_fdivd): Likewise.
16503 (cortex_a7_neon). Likewise.
16504 (bypass) Update participating units.
16506 2013-01-29 Greta Yorsh <Greta.Yorsh@arm.com>
16508 * config/arm/arm.md (type): Add ffmas and ffmad to "type" attribute.
16509 * config/arm/vfp.md (fma,fmsub,fnmsub,fnmadd): Change type
16511 * config/arm/vfp11.md (vfp_farith): Use ffmas.
16512 (vfp_fmul): Use ffmad.
16513 * config/arm/cortex-r4f.md (cortex_r4_fmacs): Use ffmas.
16514 (cortex_r4_fmacd): Use ffmad.
16515 * config/arm/cortex-m4-fpu.md (cortex_m4_fmacs): Use ffmas.
16516 * config/arm/cortex-a9.md (cortex_a9_fmacs): Use ffmas.
16517 (cortex_a9_fmacd): Use ffmad.
16518 * config/arm/cortex-a8-neon.md (cortex_a8_vfp_macs): Use ffmas.
16519 (cortex_a8_vfp_macd): Use ffmad.
16520 * config/arm/cortex-a5.md (cortex_a5_fpmacs): Use ffmas.
16521 (cortex_a5_fpmacd): Use ffmad.
16522 * config/arm/cortex-a15-neon.md (cortex_a15_vfp_macs) Use ffmas.
16523 (cortex_a15_vfp_macd): Use ffmad.
16524 * config/arm/arm1020e.md (v10_fmul): Use ffmas and ffmad.
16526 2013-01-29 Jason Merrill <jason@redhat.com>
16529 * varasm.c (default_assemble_visibility): Don't warn about
16530 visibility on artificial decls.
16532 2013-01-29 Richard Biener <rguenther@suse.de>
16534 PR tree-optimization/56113
16535 * tree-ssa-structalias.c (equiv_class_lookup): Also return
16537 (label_visit): Free duplicate bitmaps and record the leader instead.
16538 (perform_var_substitution): Adjust.
16540 2013-01-29 Richard Biener <rguenther@suse.de>
16542 PR tree-optimization/55270
16543 * tree-ssa-dom.c (eliminate_degenerate_phis): If we changed
16544 the CFG, schedule loops for fixup.
16546 2013-01-29 Nick Clifton <nickc@redhat.com>
16548 * config/rl78/rl78.c (rl78_regno_mode_code_ok_for_base_p): Allow
16551 2013-01-28 Leif Ekblad <leif@rdos.net>
16553 * config.gcc (i[34567]86-*-rdos*, x86_64-*-rdos*): New targets.
16554 * config/i386/i386.h (TARGET_RDOS): New macro.
16555 (DEFAULT_LARGE_SECTION_THRESHOLD): New macro.
16556 * config/i386/i386.c (ix86_option_override_internal): For 64bit
16557 TARGET_RDOS, set ix86_cmodel to CM_MEDIUM_PIC and flag_pic to 1.
16558 * config/i386/i386.opt (mlarge-data-threshold): Initialize to
16559 DEFAULT_LARGE_SECTION_THRESHOLD.
16560 * config/i386/i386.md (R14_REG, R15_REG): New constants.
16561 * config/i386/rdos.h: New file.
16562 * config/i386/rdos64.h: New file.
16564 2013-01-28 Bernd Schmidt <bernds@codesourcery.com>
16567 * reload.c (find_valid_class_1): Use in_hard_reg_set_p instead of
16570 2013-01-28 Jakub Jelinek <jakub@redhat.com>
16572 PR rtl-optimization/56117
16573 * sched-deps.c (sched_analyze_2) <case PREFETCH>: For use_cselib
16574 call cselib_lookup_from_insn on the MEM before calling
16575 add_insn_mem_dependence.
16577 2013-01-28 Richard Biener <rguenther@suse.de>
16579 * tree-inline.c (remap_gimple_stmt): Do not assing a BLOCK
16580 to a stmt that didn't have one.
16581 (copy_phis_for_bb): Likewise for PHI arguments.
16582 (copy_debug_stmt): Likewise for debug stmts.
16584 2013-01-28 Richard Biener <rguenther@suse.de>
16586 PR tree-optimization/56034
16587 * tree-loop-distribution.c (enum partition_kind): Add PKIND_REDUCTION.
16588 (partition_builtin_p): Adjust.
16589 (generate_code_for_partition): Handle PKIND_REDUCTION. Assert
16590 it is the last partition.
16591 (rdg_flag_uses): Check SSA_NAME_IS_DEFAULT_DEF before looking
16592 up the vertex for the definition.
16593 (classify_partition): Classify whether a partition is a
16594 PKIND_REDUCTION, thus has uses outside of the loop.
16595 (ldist_gen): Inherit PKIND_REDUCTION when merging partitions.
16596 Merge all PKIND_REDUCTION partitions into the last partition.
16597 (tree_loop_distribution): Seed partitions from reductions as well.
16599 2013-01-28 Jakub Jelinek <jakub@redhat.com>
16601 PR tree-optimization/56125
16602 * tree-ssa-math-opts.c (gimple_expand_builtin_pow): Don't optimize
16603 pow(x,c) into sqrt(x) * powi(x, n/2) or
16604 1.0 / (sqrt(x) * powi(x, abs(n/2))) if c is an integer or when
16605 optimizing for size.
16606 Don't optimize pow(x,c) into powi(x, n/3) * powi(cbrt(x), n%3) or
16607 1.0 / (powi(x, abs(n)/3) * powi(cbrt(x), abs(n)%3)) if 2c is an
16610 PR tree-optimization/56094
16611 * gimplify.c (force_gimple_operand_1): Temporarily set input_location
16612 to UNKNOWN_LOCATION while gimplifying expr.
16614 2013-01-27 Uros Bizjak <ubizjak@gmail.com>
16617 * config/i386/i386.md (*movabs<mode>_1): Add square brackets around
16618 operand 0 in movabs insn template for -masm=intel asm alternative.
16619 (*movabs<mode>_2): Ditto for operand 1.
16621 2013-01-26 David Holsgrove <david.holsgrove@xilinx.com>
16624 * config.gcc (microblaze*-linux*): Add tmake_file to allow building
16627 2013-01-26 Edgar E. Iglesias <edgar.iglesias@gmail.com>
16629 * config.gcc (microblaze*-*-*): Rename microblaze*-*-elf, update
16632 2013-01-25 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
16634 * config/aarch64/aarch64.c (TARGET_FIXED_CONDITION_CODE_REGS):
16635 Undef to avoid warning.
16637 2013-01-25 Michael Haubenwallner <michael.haubenwallner@salomon.at>
16639 * configure.ac (gcc_cv_ld_static_dynamic): Define for AIX native ld.
16640 * configure: Regenerate.
16642 2013-01-25 Jakub Jelinek <jakub@redhat.com>
16644 PR tree-optimization/56098
16645 * tree-ssa-phiopt.c (nt_init_block): Don't call add_or_mark_expr
16646 for stmts with volatile ops.
16647 (cond_store_replacement): Don't optimize if assign has volatile ops.
16648 (cond_if_else_store_replacement_1): Don't optimize if either
16649 then_assign or else_assign have volatile ops.
16650 (hoist_adjacent_loads): Don't optimize if either def1 or def2 have
16653 2013-01-25 Georg-Johann Lay <avr@gjlay.de>
16655 * doc/invoke.texi (AVR Built-in Macros): Document __XMEGA__.
16657 2013-01-25 Georg-Johann Lay <avr@gjlay.de>
16659 * doc/extend.texi (Example of asm with clobbered asm reg): Fix
16660 missing ':' in asm example.
16662 2013-01-25 Tejas Belagod <tejas.belagod@arm.com>
16664 * config/aarch64/aarch64-simd-builtins.def: Separate sq<r>dmulh_lane
16665 entries into lane and laneq entries.
16666 * config/aarch64/aarch64-simd.md (aarch64_sq<r>dmulh_lane<mode>):
16667 Remove AdvSIMD scalar modes.
16668 (aarch64_sq<r>dmulh_laneq<mode>): New.
16669 (aarch64_sq<r>dmulh_lane<mode>): New RTL pattern for Scalar AdvSIMD
16671 * config/aarch64/arm_neon.h: Fix all the vq<r>dmulh_lane* intrinsics'
16672 builtin implementations to relfect changes in RTL in aarch64-simd.md.
16673 * config/aarch64/iterators.md (VCOND): New.
16676 2013-01-25 Georg-Johann Lay <avr@gjlay.de>
16679 * config/avr/builtins.def (DEF_BUILTIN): Add LIBNAME argument.
16680 Add NULL LIBNAME argument to existing definitions.
16681 (ABSHR, ABSR, ABSLR, ABSLLR, ABSHK, ABSK, ABSLK, ABSLLK): New.
16682 * config/avr/avr-c.c (DEF_BUILTIN): Add LIBNAME argument.
16683 * config/avr/avr.c (DEF_BUILTIN): Same.
16684 (avr_init_builtins): Pass down LIBNAME to add_builtin_function.
16685 (avr_expand_builtin): Expand to a vanilla call if a libgcc
16686 implementation is available (DECL_ASSEMBLER_NAME is set).
16687 (avr_fold_absfx): New static function.
16688 (avr_fold_builtin): Use it to handle: AVR_BUILTIN_ABSHR,
16689 AVR_BUILTIN_ABSR, AVR_BUILTIN_ABSLR, AVR_BUILTIN_ABSLLR,
16690 AVR_BUILTIN_ABSHK, AVR_BUILTIN_ABSK, AVR_BUILTIN_ABSLK,
16691 AVR_BUILTIN_ABSLLK.
16692 * config/avr/stdfix.h (abshr, absr, abslr, absllr)
16693 (abshk, absk, abslk, absllk): Provide as static inline functions.
16695 2013-01-25 Marek Polacek <polacek@redhat.com>
16697 PR tree-optimization/56035
16698 * cfgloopmanip.c (fix_loop_structure): Remove redundant condition.
16700 2012-01-24 Uros Bizjak <ubizjak@gmail.com>
16702 * config/i386/i386.md (*movti_internal_rex64): Add (o,e) alternative.
16703 (*movtf_internal_rex64): Add (!o,C) alternative
16704 (*movxf_internal_rex64): Ditto.
16705 (*movdf_internal_rex64): Add (?r,C) and (?m,C) alternatives.
16707 2013-01-24 Shenghou Ma <minux.ma@gmail.com>
16709 * doc/invoke.texi: fix typo.
16710 * doc/objc.texi: fix typo.
16712 2013-01-24 Richard Sandiford <rdsandiford@googlemail.com>
16714 * config/mips/mips.md (*and<mode>3_mips16): Use the "W" constraint
16715 for the first two alternatives.
16717 2013-01-24 Diego Novillo <dnovillo@google.com>
16719 * Makefile.in (GGC): Remove. Replace all instances with ggc-page.o.
16720 (ggc-zone.o): Remove.
16721 * configure.ac: Remove option --with-gc.
16722 * configure: Re-generate.
16723 * doc/install.texi: Remove documentation for --with-gc.
16724 * gengtype.c (write_enum_defn): Remove. Update all users.
16725 (write_Types_process_field): Remove generation of gt_e_* argument.
16726 (output_type_enum): Remove. Update all users.
16727 (write_enum_defn): Remove. Update all users.
16728 (enum alloc_zone): Remove. Update all users.
16729 (write_splay_tree_allocator_def): Remove generation of gt_e_* argument.
16730 * ggc-common.c (ggc_splay_alloc): Remove first argument.
16731 Update all callers.
16732 (struct ptr_data): Remove field TYPE. Update all users.
16733 (gt_pch_note_object): Remove argument TYPE. Update all users.
16734 * ggc-internal.h (ggc_pch_alloc_object): Remove last argument.
16736 * ggc-none.c (ggc_alloc_typed_stat): Remove.
16737 (struct alloc_zone): Remove.
16738 (ggc_internal_alloc_zone_stat): Remove.
16739 (ggc_internal_cleared_alloc_zone_stat): Remove.
16740 * ggc-page.c (ggc_alloc_typed_stat): Remove.
16741 (ggc_pch_count_object): Remove last argument. Update all users.
16742 (ggc_pch_alloc_object): Remove last argument. Update all users.
16743 (struct alloc_zone): Remove.
16744 * ggc-zone.c: Remove.
16745 * ggc.h (gt_pch_note_object): Remove last argument. Update all users.
16746 (struct alloc_zone): Remove.
16747 (ggc_alloc_typed_stat): Remove.
16748 (ggc_alloc_typed): Remove.
16749 (ggc_splay_alloc): Remove first argument.
16750 (rtl_zone): Remove. Update all users.
16751 (tree_zone): Remove. Update all users.
16752 (tree_id_zone): Remove. Update all users.
16753 (ggc_internal_zone_alloc_stat): Remove. Update all users.
16754 (ggc_internal_zone_cleared_alloc_stat): Remove. Update all users.
16755 (ggc_internal_zone_vec_alloc_stat): Remove. Update all users.
16756 * tree-ssanames.c: Remove references to zone allocator in comments.
16758 2013-01-24 Georg-Johann Lay <avr@gjlay.de>
16760 * config/avr/avr.c (avr_out_fract): Make register numbers that
16761 might be outside of source operand signed.
16763 2013-01-24 Uros Bizjak <ubizjak@gmail.com>
16765 * config/i386/constraints.md (Yf): New constraint.
16766 * config/i386/i386.md (*movdf_internal_rex64): Use Yf*f instead
16767 of f constraint to conditionaly disable x87 register preferences.
16768 (*movdf_internal): Ditto.
16769 (*movsf_internal): Ditto.
16771 2013-01-24 Steven Bosscher <steven@gcc.gnu.org>
16773 PR inline-asm/55934
16774 * lra-assigns.c (assign_by_spills): Throw away the pattern of asms
16775 that have operands with impossible constraints.
16776 Add a FIXME for a speed-up opportunity.
16777 * lra-constraints.c (process_alt_operands): Verify that a class
16778 selected from constraints on asms is valid for the operand mode.
16779 (curr_insn_transform): Remove incorrect comment.
16781 2013-01-23 David Edelsohn <dje.gcc@gmail.com>
16783 * config/rs6000/rs6000.c (rs6000_delegitimize_address): Check that
16784 TOC operand is a valid symbol ref in the constant pool.
16786 2013-01-23 Edgar E. Iglesias <edgar.iglesias@gmail.com>
16788 * config/microblaze/linux.h: Add TARGET_OS_CPP_BUILTINS
16790 2013-01-23 Georg-Johann Lay <avr@gjlay.de>
16793 * config/avr/stdfix.h: New file.
16794 * t-avr (stdfix-gcc.h): New rule to build it.
16795 (EXTRA_HEADERS): Set it to install stdfix.h, stdfix-gcc.h.
16797 2013-01-23 Kostya Serebryany <kcc@google.com>
16799 * config/darwin.h: remove dependency on
16800 CoreFoundation (asan on Mac OS).
16802 2013-01-23 Jakub Jelinek <jakub@redhat.com>
16805 * config/arm/arm.md (cbranchdi4, cstoredi4): Use s_register_operand
16806 instead of cmpdi_operand for first comparison operand.
16807 Don't assert that comparison operands aren't both constants.
16809 2013-01-22 Jonathan Wakely <jwakely.gcc@gmail.com>
16811 * doc/install.texi (Downloading the Source): Update references to
16812 downloading separate components.
16814 2013-01-22 Jonathan Wakely <jwakely.gcc@gmail.com>
16816 * doc/extend.texi (__int128): Improve grammar.
16818 2013-01-22 Uros Bizjak <ubizjak@gmail.com>
16821 * config/i386/i386.md (*movti_internal_rex64): Change (o,riF)
16822 alternative to (o,r).
16823 (*movdi_internal_rex64): Remove (!o,n) alternative.
16824 (DImode immediate->memory splitter): Remove.
16825 (DImode immediate->memory peephole2): Remove.
16826 (movtf): Enable for TARGET_64BIT || TARGET_SSE.
16827 (*movtf_internal_rex64): Rename from *movtf_internal. Change (!o,F*r)
16828 alternative to (!o,*r).
16829 (*movtf_internal_sse): New pattern.
16830 (*movxf_internal_rex64): New pattern.
16831 (*movxf_internal): Disable for TARGET_64BIT.
16832 (*movdf_internal_rex64): Remove (!o,F) alternative.
16834 2013-01-22 Jakub Jelinek <jakub@redhat.com>
16836 PR middle-end/56074
16837 * dumpfile.c (dump_loc): Only print loc if LOCATION_LOCUS (loc)
16838 isn't UNKNOWN_LOCATION nor BUILTINS_LOCATION.
16839 * tree-vect-loop-manip.c (find_loop_location): Also ignore
16840 stmt locations where LOCATION_LOCUS of the stmt location is
16841 UNKNOWN_LOCATION or BUILTINS_LOCATION.
16844 * config/i386/i386.md (UNSPEC_STOS): New.
16845 (strset_singleop, *strsetdi_rex_1, *strsetsi_1, *strsethi_1,
16846 *strsetqi_1): Add UNSPEC_STOS.
16848 2013-01-22 Paolo Carlini <paolo.carlini@oracle.com>
16851 * doc/invoke.texi: Remove left over -Wsynth example.
16853 2013-01-21 Jakub Jelinek <jakub@redhat.com>
16855 PR tree-optimization/56051
16856 * fold-const.c (fold_binary_loc): Don't fold
16857 X < (cast) (1 << Y) into (X >> Y) != 0 if cast is either
16858 a narrowing conversion, or widening conversion from signed
16861 2013-01-21 Uros Bizjak <ubizjak@gmail.com>
16863 PR rtl-optimization/56023
16864 * haifa-sched.c (fix_inter_tick): Do not update ticks of instructions,
16865 dependent on debug instruction.
16867 2013-01-21 Martin Jambor <mjambor@suse.cz>
16869 PR middle-end/56022
16870 * function.c (allocate_struct_function): Call
16871 invoke_set_current_function_hook earlier.
16873 2013-01-21 Jakub Jelinek <jakub@redhat.com>
16875 * reload1.c (init_reload): Only initialize reload_obstack
16876 during the first call.
16878 2013-01-21 Marek Polacek <polacek@redhat.com>
16880 * cfgloop.c (verify_loop_structure): Fix up grammar.
16882 2013-01-21 Yi-Hsiu Hsu <ahsu@marvell.com>
16884 * config/arm/marvell-pj4.md (pj4_shift_conds, pj4_alu_shift,
16885 pj4_alu_shift_conds, pj4_shift): Handle simple_alu_shift.
16887 2013-01-21 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
16890 * config/arm/marvell-pj4.md: Update copyright year.
16891 Fix up use of alu to alu_reg and simple_alu_imm.
16893 2013-01-21 Uros Bizjak <ubizjak@gmail.com>
16895 * config/i386/i386.md (enabled): Do not disable fma4 for TARGET_FMA.
16897 2013-01-20 Vladimir Makarov <vmakarov@redhat.com>
16900 * lra-constraints.c (curr_insn_transform): Don't reuse original
16901 insn for secondary memory move when memory mode should be different.
16903 2013-01-20 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
16905 * config/pa/pa.md (atomic_loaddi, atomic_loaddi_1, atomic_storedi,
16906 atomic_storedi_1): New patterns.
16908 2013-01-20 Venkataramanan Kumar <venkataramanan.kumar@amd.com>
16910 btver2 pipeline descriptions.
16911 * config/i386/i386.c: Enable CPU_BTVER2 to use btver2 pipeline
16913 * config/i386/i386.md (btver2_decode): New type attributes.
16914 * config/i386/sse.md (btver2_decode, btver2_sse_attr): New
16916 * config/i386/btver2.md: New file describing btver2 pipelines.
16918 2013-01-19 Andrew Pinski <apinski@cavium.com>
16920 PR tree-optimization/52631
16921 * tree-ssa-sccvn (visit_use): Before looking up the original
16922 statement, try looking up the simplified expression.
16924 2013-01-19 Anthony Green <green@moxielogic.com>
16926 * config/moxie/moxie.c (moxie_expand_prologue): Set
16927 current_function_static_stack_size.
16929 2013-01-18 Jakub Jelinek <jakub@redhat.com>
16931 PR tree-optimization/56029
16932 * tree-phinodes.c (reserve_phi_args_for_new_edge): Set
16933 gimple_phi_arg_location for the new arg to UNKNOWN_LOCATION.
16935 2013-01-18 Sharad Singhai <singhai@google.com>
16937 PR tree-optimization/55995
16938 * dumpfile.c (dump_loc): Print location only if available.
16939 * tree-vectorizer.c (increase_alignment): Intialize vect_location.
16941 2013-01-18 Vladimir Makarov <vmakarov@redhat.com>
16944 * lra-constraints.c (curr_insn_transform): Reuse original insn for
16945 secondary memory move.
16946 (inherit_reload_reg): Use rclass instead of cl for
16947 check_secondary_memory_needed_p.
16949 2013-01-18 Jakub Jelinek <jakub@redhat.com>
16951 PR middle-end/56015
16952 * expr.c (expand_expr_real_2) <case COMPLEX_EXPR>: Handle
16953 the case where writing real complex part of target modifies op1.
16955 2013-01-18 James Greenhalgh <james.greenhalgh@arm.com>
16957 * config/aarch64/aarch64-simd.md
16958 (aarch64_vcond_internal<mode>): Handle unordered cases.
16959 * config/aarch64/iterators.md (v_cmp_result): New.
16961 2013-01-18 Yi-Hsiu Hsu <ahsu@marvell.com>
16962 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
16964 * config/arm/marvell-pj4.md: New file.
16965 * config/arm/arm.c (arm_issue_rate): Add marvell_pj4.
16966 * config/arm/arm.md (generic_sched): Add marvell_pj4.
16967 (generic_vfp): Likewise.
16968 * config/arm/arm-cores.def: Add marvell-pj4.
16969 * config/arm/arm-tune.md: Regenerate.
16970 * config/arm/arm-tables.opt: Regenerate.
16971 * config/arm/bpabi.h (BE8_LINK_SPEC): Add marvell_pj4.
16972 * doc/invoke.texi: Document marvell-pj4.
16974 2013-01-18 Tejas Belagod <tejas.belagod@arm.com>
16976 * config/aarch64/arm_neon.h: Map scalar types to standard types.
16978 2013-01-18 Alexandre Oliva <aoliva@redhat.com>
16983 * var-tracking.c (negative_power_of_two_p): New.
16984 (global_get_addr_cache, local_get_addr_cache): New.
16985 (get_addr_from_global_cache, get_addr_from_local_cache): New.
16986 (vt_canonicalize_addr): Rewrite using the above. Adjust the
16988 (vt_stack_offset_p): Remove.
16989 (vt_canon_true_dep): Always canonicalize loc's address.
16990 (clobber_overlapping_mems): Make sure we have a MEM.
16991 (local_get_addr_clear_given_value): New.
16992 (val_reset): Clear local cached entries.
16993 (compute_bb_dataflow): Create and release the local cache.
16994 Disable duplicate MEMs clobbering.
16995 (emit_notes_in_bb): Clobber MEMs likewise.
16996 (vt_emit_notes): Create and release the local cache.
16997 (vt_initialize, vt_finalize): Create and release the global
16998 cache, respectively.
16999 * alias.c (rtx_equal_for_memref_p): Compare operands of ENTRY_VALUEs.
17001 2013-01-18 Alexandre Oliva <aoliva@redhat.com>
17003 PR libmudflap/53359
17004 * tree-mudflap.c (mudflap_finish_file): Skip deferred decls
17005 not found in the symtab.
17007 2013-01-18 Alexandre Oliva <aoliva@redhat.com>
17010 PR rtl-optimization/55547
17011 PR rtl-optimization/53827
17014 * alias.c (offset_overlap_p): New, factored out of...
17015 (memrefs_conflict_p): ... this. Use absolute sizes. Retain
17016 the conservative special case for symbolic constants. Don't
17017 adjust zero sizes on alignment.
17019 2013-01-18 Bernd Schmidt <bernds@codesourcery.com>
17021 PR rtl-optimization/52573
17022 * regrename.c (build_def_use): Ignore REG_DEAD notes if there is a
17023 REG_UNUSED for the same register.
17025 2013-01-17 Richard Biener <rguenther@suse.de>
17026 Marek Polacek <polacek@redhat.com>
17028 PR rtl-optimization/55833
17029 * loop-unswitch.c (unswitch_loops): Move loop verification...
17030 (unswitch_single_loop): ...here. Call mark_irreducible_loops.
17031 * cfgloopmanip.c (fix_loop_placement): Add IRRED_INVALIDATED parameter.
17032 Set it to true when we're removing a loop from hierarchy tree in
17033 an irreducible region.
17034 (fix_bb_placements): Adjust caller.
17035 (fix_loop_placements): Likewise.
17037 2013-01-17 Georg-Johann Lay <avr@gjlay.de>
17039 * config/avr/builtins.def (DEF_BUILTIN): Factor out
17040 "__builtin_avr_" from NAME, turn NAME to an uppercase identifier.
17041 Factor out 'CODE_FOR_' from ICODE, use 'nothing' instead of '-1'.
17042 Remove ID. Adjust comments.
17043 * config/avr/avr-c.c (avr_builtin_name): Remove.
17044 (avr_cpu_cpp_builtins): Use DEF_BUILTIN instead of for-loop.
17045 * config/avr/avr.c (avr_tolower): New static function.
17046 (DEF_BUILTIN): Remove parameter ID. Prefix ICODE by 'CODE_FOR_'.
17047 Stringify NAME, prefix it with "__builtin_avr_" and lowercase it.
17048 (avr_expand_builtin): Assert insn_code != CODE_FOR_nothing for
17051 2013-01-17 Jan Hubicka <jh@suse.cz>
17053 PR tree-optimization/55273
17054 * loop-iv.c (iv_number_of_iterations): Consider zero iteration case.
17056 2013-01-17 Uros Bizjak <ubizjak@gmail.com>
17059 * config/i386/sync.md (atomic_store<mode>): Always generate SWImode
17060 store through atomic_store<mode>_1.
17061 (atomic_store<mode>_1): Macroize insn using SWI mode iterator.
17063 2013-01-17 Martin Jambor <mjambor@suse.cz>
17065 PR tree-optimizations/55264
17066 * ipa-inline-transform.c (can_remove_node_now_p_1): Never return true
17067 for virtual methods.
17068 * ipa.c (symtab_remove_unreachable_nodes): Never return true for
17069 virtual methods before inlining is over.
17070 * cgraph.h (cgraph_only_called_directly_or_aliased_p): Return false for
17072 * cgraphclones.c (cgraph_create_virtual_clone): Mark clones as
17075 2013-01-16 Vladimir Makarov <vmakarov@redhat.com>
17077 PR rtl-optimization/56005
17078 * sched-deps.c (sched_analyze_2): Check deps->readonly for adding
17079 pending reads for prefetch.
17081 2013-01-16 Ian Bolton <ian.bolton@arm.com>
17083 * config/aarch64/aarch64.md
17084 (*cstoresi_neg_uxtw): New pattern.
17085 (*cmovsi_insn_uxtw): New pattern.
17086 (*<optab>si3_uxtw): New pattern.
17087 (*<LOGICAL:optab>_<SHIFT:optab>si3_uxtw): New pattern.
17088 (*<optab>si3_insn_uxtw): New pattern.
17089 (*bswapsi2_uxtw): New pattern.
17091 2013-01-16 Richard Biener <rguenther@suse.de>
17093 * tree-inline.c (tree_function_versioning): Remove set but
17094 never used variable.
17096 2013-01-16 Richard Biener <rguenther@suse.de>
17098 PR tree-optimization/55964
17099 * tree-flow.h (rename_variables_in_loop): Remove.
17100 (rename_variables_in_bb): Likewise.
17101 * tree-loop-distribution.c (update_phis_for_loop_copy): Remove.
17102 (copy_loop_before): Adjust and delete update-ssa status.
17103 * tree-vect-loop-manip.c (rename_variables_in_bb): Make static.
17104 (rename_variables_in_bb): Likewise. Properly walk over predecessors.
17105 (rename_variables_in_loop): Remove.
17106 (slpeel_update_phis_for_duplicate_loop): Likewise.
17107 (slpeel_tree_duplicate_loop_to_edge_cfg): Handle nested loops,
17108 use available cfg machinery instead of duplicating it.
17109 Update PHI nodes and perform poor-mans SSA update here.
17110 (slpeel_tree_peel_loop_to_edge): Adjust.
17112 2013-01-16 Richard Biener <rguenther@suse.de>
17114 PR tree-optimization/54767
17115 PR tree-optimization/53465
17116 * tree-vrp.c (vrp_meet_1): Revert original fix for PR53465.
17117 (vrp_visit_phi_node): For PHI arguments coming via backedges
17118 drop all symbolical range information.
17119 (execute_vrp): Compute backedges.
17121 2013-01-16 Richard Biener <rguenther@suse.de>
17123 * doc/install.texi: Update CLooG and ISL requirements to
17126 2013-01-16 Christian Bruel <christian.bruel@st.com>
17129 * config/sh/sh.c (sh_expand_prologue): Postpone new_stack mem symbol.
17130 (broken_move): Handle UNSPECV_SP_SWITCH_B.
17131 * config/sh/sh.md (sp_switch_1): Use set (reg:SI SP_REG).
17133 2013-01-16 DJ Delorie <dj@redhat.com>
17135 * config/sh/sh.md (UNSPECV_SP_SWITCH_B): New.
17136 (UNSPECV_SP_SWITCH_E): New.
17137 (sp_switch_1): Change to an unspec.
17138 (sp_switch_2): Change to an unspec. Don't use post-inc when we
17141 2013-01-16 Uros Bizjak <ubizjak@gmail.com>
17143 * emit-rtl.c (need_atomic_barrier_p): Mask memory model argument
17144 with MEMMODEL_MASK before comparing with MEMMODEL_* memory types.
17145 * optabs.c (maybe_emit_sync_lock_test_and_set): Ditto.
17146 (expand_mem_thread_fence): Ditto.
17147 (expand_mem_signal_fence): Ditto.
17148 (expand_atomic_load): Ditto.
17149 (expand_atomic_store): Ditto.
17151 2013-01-16 Alexandre Oliva <aoliva@redhat.com>
17153 PR rtl-optimization/55547
17154 PR rtl-optimization/53827
17157 * alias.c (memrefs_conflict_p): Set sizes to negative after
17160 2013-01-15 Jakub Jelinek <jakub@redhat.com>
17163 * function.c (thread_prologue_and_epilogue_insns): Always
17164 add crtl->drap_reg to set_up_by_prologue.set, even if
17165 stack_realign_drap is false.
17167 2013-01-15 Jan-Benedict Glaw <jbglaw@lug-owl.de>
17169 * config/vax/vax.md (add<mode>3, sub<mode>3, mul<mode>3, div<mode>3,
17170 and<mode>3, *and<mode>_const_int, ior<mode>3, xor<mode>3, ashrsi3,
17171 *call): Fix indention.
17173 2013-01-15 Tom de Vries <tom@codesourcery.com>
17176 * optabs.c (widen_operand): Use gen_lowpart instead of gen_rtx_SUBREG.
17179 2013-01-15 Vladimir Makarov <vmakarov@redhat.com>
17181 PR rtl-optimization/55153
17182 * sched-deps.c (sched_analyze_2): Add pending reads for prefetch.
17184 2013-01-15 Martin Jambor <mjambor@suse.cz>
17186 PR tree-optimization/55920
17187 * tree-sra.c (analyze_access_subtree): Do not mark non-removable
17188 accesses as grp_to_be_debug_replaced.
17190 2013-01-15 Jakub Jelinek <jakub@redhat.com>
17192 PR tree-optimization/55920
17193 * tree-sra.c (sra_modify_assign): If for lacc->grp_to_be_debug_replaced
17194 there is non-useless type conversion needed from debug rhs to lhs,
17195 use build_debug_ref_for_model and/or VIEW_CONVERT_EXPR.
17197 2013-01-15 Joseph Myers <joseph@codesourcery.com>
17198 Mikael Pettersson <mikpe@it.uu.se>
17201 * config/arm/arm.h (ADDR_VEC_ALIGN): Align SImode jump tables for
17203 (ASM_OUTPUT_CASE_LABEL): Remove.
17204 (ASM_OUTPUT_BEFORE_CASE_LABEL): Define to empty.
17205 * final.c (shorten_branches): Update alignment of labels before
17206 jump tables if CASE_VECTOR_SHORTEN_MODE.
17208 2013-01-15 Richard Biener <rguenther@suse.de>
17211 * system.h: Do not include gmp.h for building host tools.
17213 2013-01-15 Richard Biener <rguenther@suse.de>
17215 PR middle-end/55882
17216 * emit-rtl.c (set_mem_attributes_minus_bitpos): Correctly
17217 account for bitpos when computing alignment.
17219 2013-01-15 Vladimir Yakovlev <vladimir.b.yakovlev@intel.com>
17221 * config/i386/i386-c.c (ix86_target_macros_internal): New case.
17222 (ix86_target_macros_internal): Likewise.
17224 * config/i386/i386.c (m_CORE2I7): Removed.
17225 (m_CORE_HASWELL): New macro.
17226 (m_CORE_ALL): Likewise.
17227 (initial_ix86_tune_features): m_CORE2I7 is replaced by m_CORE_ALL.
17228 (initial_ix86_arch_features): Likewise.
17229 (processor_target_table): Initializations for Core avx2.
17230 (cpu_names): New names "core-avx2".
17231 (ix86_option_override_internal): Changed PROCESSOR_COREI7 by
17232 PROCESSOR_CORE_HASWELL.
17233 (ix86_issue_rate): New case.
17234 (ia32_multipass_dfa_lookahead): Likewise.
17235 (ix86_sched_init_global): Likewise.
17237 * config/i386/i386.h (TARGET_HASWELL): New macro.
17238 (target_cpu_default): New TARGET_CPU_DEFAULT_haswell.
17239 (processor_type): New PROCESSOR_HASWELL.
17241 2013-01-15 Jakub Jelinek <jakub@redhat.com>
17243 PR tree-optimization/55955
17244 * tree-vect-loop.c (vectorizable_reduction): Give up early on
17245 *SHIFT_EXPR and *ROTATE_EXPR codes.
17247 PR tree-optimization/48766
17248 * opts.c (common_handle_option): For -fwrapv disable -ftrapv, for
17249 -ftrapv disable -fwrapv.
17251 2013-01-14 Georg-Johann Lay <avr@gjlay.de>
17254 * config/avr/avr-c.c (avr_cpu_cpp_builtins): Define __FLASH
17255 etc. to 1 and not to __flash.
17256 Use LL suffix for __INT24_MAX__ with -mint8.
17257 Use ULL suffix for __UINT24_MAX__ with -mint8.
17259 2013-01-14 Georg-Johann Lay <avr@gjlay.de>
17261 * config/avr/avr-arch.h
17262 (struct base_arch_s): Use typedef avr_arch_t instead.
17263 (struct arch_info_s): Use typedef avr_arch_info_t instead.
17264 (struct mcu_type_s): Use typedef avr_mcu_t instead.
17265 * config/avr/avr.c: Same.
17266 * config/avr/avr-devices.c: Same.
17267 * config/avr/driver-avr.c: Same.
17268 * config/avr/gen-avr-mmcu-texi.c: Same.
17269 * config/avr/avr-mcus.def: Adjust comment.
17271 2013-01-14 Tejas Belagod <tejas.belagod@arm.com>
17273 * config/aarch64/aarch64-simd.md (*aarch64_simd_ld1r<mode>): New.
17274 * config/aarch64/iterators.md (VALLDI): New.
17276 2013-01-14 Uros Bizjak <ubizjak@gmail.com>
17277 Andi Kleen <ak@linux.intel.com>
17280 * config/i386/sync.md (atomic_store<mode>_1): New pattern.
17281 (atomic_store<mode>): Call atomic_store<mode>_1 for IX86_HLE_RELEASE
17284 2013-01-14 Georg-Johann Lay <avr@gjlay.de>
17286 * config/avr/avr-stdint.h: Remove trailing blanks.
17287 * config/avr/avr-log.h: Same.
17288 * config/avr/avr-arch.h: Same.
17289 * config/avr/avr-devices.c: Same.
17290 * config/avr/avr-dimode.md: Same.
17291 * config/avr/predicates.md: Same.
17292 * config/avr/avr-c.c: Same. And fix typo.
17294 * config/avr/avr-protos.h: Same. And:
17295 (function_arg_regno_p): Rename to avr_function_arg_regno_p.
17296 (init_cumulative_args): Rename to avr_init_cumulative_args.
17297 (expand_prologue): Rename to avr_expand_prologue.
17298 (expand_epilogue): Rename to avr_expand_epilogue.
17299 (adjust_insn_length): Rename to avr_adjust_insn_length.
17300 (notice_update_cc): Rename to avr_notice_update_cc.
17301 (final_prescan_insn): Rename to avr_final_prescan_insn.
17302 * config/avr/avr.c: Same.
17303 * config/avr/avr.h: Same.
17304 * config/avr/avr.md: Remove trailing blanks.
17305 (prologue): Use avr_expand_prologue.
17306 (epilogue, sibcall_epilogue): Use avr_expand_epilogue.
17308 2013-01-14 Richard Biener <rguenther@suse.de>
17310 * tree-cfg.c (verify_expr_location, verify_expr_location_1,
17311 verify_location, collect_subblocks): New functions.
17312 (verify_gimple_in_cfg): Verify that locations only reference
17313 BLOCKs in the functions BLOCK tree.
17315 2013-01-14 Richard Biener <rguenther@suse.de>
17317 * tree-cfgcleanup.c (remove_forwarder_block): Unshare propagated
17319 * graphite-sese-to-poly.c (insert_out_of_ssa_copy): Properly
17321 (insert_out_of_ssa_copy_on_edge): Likewise.
17322 (rewrite_close_phi_out_of_ssa): Likewise.
17323 * tree-ssa.c (insert_debug_temp_for_var_def): Properly unshare
17325 * tree-ssa-pre.c (insert_into_preds_of_block): Properly unshare
17326 propagated constants.
17327 * tree-cfg.c (tree_node_can_be_shared): Handled component-refs
17330 2013-01-14 Georg-Johann Lay <avr@gjlay.de>
17332 * config/avr/avr-modes.def: Add GPL copyright notice.
17334 2013-01-13 Uros Bizjak <ubizjak@gmail.com>
17336 * config/i386/sync.md (mem_thread_fence): Mask operands[0] with
17337 MEMMODEL_MASK to determine memory model.
17338 (atomic_store<mode>): Ditto from operands[2].
17339 * config/i386/i386.c (ix86_memmodel_check): Declare "strong" as bool.
17341 2013-01-13 Jakub Jelinek <jakub@redhat.com>
17344 * gimple-fold.c (get_symbol_constant_value): Call unshare_expr.
17345 (fold_gimple_assign): Don't call unshare_expr here.
17346 (fold_ctor_reference): Call unshare_expr.
17348 2013-01-13 Terry Guo <terry.guo@arm.com>
17350 * Makefile.in (s-mlib): New argument MULTILIB_REUSE.
17351 * doc/fragments.texi: Document MULTILIB_REUSE.
17352 * gcc.c (multilib_reuse): New internal spec.
17353 (set_multilib_dir): Also search multilib from multilib_reuse.
17354 * genmultilib (tmpmultilib3): Refactor code.
17355 (tmpmultilib4): Ditto.
17356 (multilib_reuse): New multilib argument.
17358 2013-01-13 Richard Sandiford <rdsandiford@googlemail.com>
17360 * Makefile.in: Update copyright.
17362 2013-01-12 Tom de Vries <tom@codesourcery.com>
17364 PR middle-end/55890
17365 * calls.c (expand_call): Check if arg_nr is valid.
17367 2013-01-11 Michael Meissner <meissner@linux.vnet.ibm.com>
17369 * doc/extend.texi (X86 Built-in Functions): Add whitespace in
17370 __builtin_ia32_paddb256 and __builtin_ia32_pavgb256
17371 documentation. Add missing '__' in front of
17372 __builtin_ia32_packssdw256.
17374 2013-01-11 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
17377 * config/s390/s390.c (s390_preferred_reload_class): Do not return
17378 NO_REGS for larl operands.
17379 (s390_reload_larl_operand): Use s390_load_address instead of
17382 2013-01-11 Richard Biener <rguenther@suse.de>
17384 * tree-cfg.c (verify_node_sharing_1): Split out from ...
17385 (verify_node_sharing): ... here.
17386 (verify_gimple_in_cfg): Use verify_node_sharing_1 for walk_tree.
17388 2013-01-11 Eric Botcazou <ebotcazou@adacore.com>
17390 * configure.ac (Tree checking): Set TREECHECKING to yes if enabled.
17391 Substitute TREECHECKING.
17392 * configure: Regenerate.
17393 * Makefile.in (TREECHECKING): New.
17395 2013-01-11 Richard Guenther <rguenther@suse.de>
17397 PR tree-optimization/44061
17398 * tree-vrp.c (extract_range_basic): Compute zero as
17399 value-range for __builtin_constant_p of function parameters.
17401 2013-01-10 Richard Sandiford <rdsandiford@googlemail.com>
17403 Update copyright years.
17405 2013-01-10 Vladimir Makarov <vmakarov@redhat.com>
17407 PR rtl-optimization/55672
17408 * lra-eliminations.c (mark_not_eliminable): Permit addition with
17409 const to be eliminable.
17411 2013-01-10 David Edelsohn <dje.gcc@gmail.com>
17413 * configure.ac (HAVE_AS_TLS): Add check for powerpc-ibm-aix.
17414 * configure: Regenerate.
17416 2013-01-10 Richard Biener <rguenther@suse.de>
17418 * builtins.c (expand_builtin_init_trampoline): Use set_mem_attributes.
17420 2013-01-10 Richard Biener <rguenther@suse.de>
17423 * tree-into-ssa.c (rewrite_add_phi_arguments): Do not set
17424 locations for virtual PHI arguments.
17425 (rewrite_update_phi_arguments): Likewise.
17427 2013-01-10 Joel Sherrill <joel.sherrill@OARcorp.com>
17429 * config/v850/rtems.h (ASM_SPEC): Pass -m8byte-align and -mgcc-abi
17432 2013-01-10 Jakub Jelinek <jakub@redhat.com>
17434 PR tree-optimization/55921
17435 * tree-complex.c (expand_complex_asm): New function.
17436 (expand_complex_operations_1): Call it for GIMPLE_ASM.
17438 2013-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
17441 * config/s390/s390.c (s390_symref_operand_p)
17442 (s390_loadrelative_operand_p): Merge the two functions.
17443 (s390_check_qrst_address, print_operand_address): Add parameters
17444 to s390_loadrelative_operand_p invokation.
17445 (s390_check_symref_alignment): Use s390_loadrelative_operand_p.
17446 (s390_reload_larl_operand, s390_secondary_reload): Use
17447 s390_loadrelative_operand_p instead of s390_symref_operand_p.
17448 (legitimize_pic_address): Handle @GOTENT and @PLT + addend.
17450 2013-01-09 Mike Stump <mikestump@comcast.net>
17452 * dse.c (record_store): Remove unnecessary assert.
17454 2013-01-09 Jan Hubicka <jh@suse.cz>
17456 PR tree-optimization/55569
17457 * cfgloopmanip.c (scale_loop_profile): Make ITERATION_BOUND gcov_type.
17458 * cfgloop.h (scale_loop_profile): Likewise.
17460 2013-01-09 Jan Hubicka <jh@suse.cz>
17463 * ipa-inline.c (ipa_inline): Remove extern inlines and virtual
17465 * cgraphclones.c (cgraph_clone_node): Cpoy also LTO file data.
17467 2013-01-09 Richard Sandiford <rdsandiford@googlemail.com>
17469 PR middle-end/55114
17470 * expr.h (maybe_emit_group_store): Declare.
17471 * expr.c (maybe_emit_group_store): New function.
17472 * builtins.c (expand_builtin_int_roundingfn): Call it.
17473 (expand_builtin_int_roundingfn_2): Likewise.
17475 2013-01-09 Vladimir Makarov <vmakarov@redhat.com>
17477 PR rtl-optimization/55829
17478 * lra-constraints.c (match_reload): Add code for absent output.
17479 (curr_insn_transform): Add code for reloads of matched inputs
17482 2013-01-09 Uros Bizjak <ubizjak@gmail.com>
17484 * config/i386/sse.md (*vec_interleave_highv2df): Change mode
17485 attribute of movddup insn to DF.
17486 (*vec_interleave_lowv2df): Ditto.
17487 (vec_dupv2df): Ditto.
17489 2013-01-09 Jan Hubicka <jh@suse.cz>
17491 PR tree-optimiation/55875
17492 * tree-ssa-loop-niter.c (number_of_iterations_cond): Add
17493 EVERY_ITERATION parameter.
17494 (number_of_iterations_exit): Check if exit is executed every iteration.
17495 (idx_infer_loop_bounds): Similarly here.
17496 (n_of_executions_at_most): Simplify
17497 to only test for cases where statement is dominated by the
17498 particular bound; handle correctly the "postdominance" test.
17499 (scev_probably_wraps_p): Use max loop iterations info
17500 as a global bound first.
17502 2013-01-09 Nguyen Duy Dat <dat.nguyen.yn@rvc.renesas.com>
17503 Nick Clifton <nickc@redhat.com>
17505 * config/v850/v850.md (cbranchsf4): New pattern.
17506 (cstoresf4): New pattern.
17507 (cbranchdf4): New pattern.
17508 (cstoredf4): New pattern.
17509 (movsicc): Disallow floating point comparisons.
17510 (cmpsf_le_insn): Fix order of operators.
17511 (cmpsf_lt_insn): Likewise.
17512 (cmpsf_eq_insn): Likewise.
17513 (cmpdf_le_insn): Likewise.
17514 (cmpdf_lt_insn): Likewise.
17515 (cmpdf_eq_insn): Likewise.
17516 (cmpsf_ge_insn): Use LE comparison.
17517 (cmpdf_ge_insn): Likewise.
17518 (cmpsf_gt_insn): Use LT comparison.
17519 (cmpdf_gt_insn): Likewise.
17520 (cmpsf_ne_insn): Delete pattern.
17521 (cmpdf_ne_insn): Delete pattern.
17522 * config/v850/v850.c (v850_gen_float_compare): Use
17523 gen_cmpdf_eq_insn for NE comparison.
17524 (v850_float_z_comparison_operator)
17525 (v850_float_nz_comparison_operator): Move from here ...
17526 * config/v850/predicates.md: ... to here. Move GT and GE
17527 comparisons into v850_float_z_comparison_operator.
17528 * config/v850/v850-protos.h (v850_float_z_comparison_operator):
17530 (v850_float_nz_comparison_operator): Likewise.
17532 2013-01-09 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
17534 * config/pa/pa.c (pa_emit_move_sequence): Replace calls to gen_insv
17535 with calls to gen_insvsi/gen_insvdi.
17537 2013-01-09 Venkataramanan Kumar <venkataramanan.kumar@amd.com>
17539 * config/i386/i386.c (initial_ix86_tune_features): Set up
17540 X86_TUNE_AVX128_OPTIMAL for m_BTVER2.
17542 2013-01-09 Steven Bosscher <steven@gcc.gnu.org>
17543 Jakub Jelinek <jakub@redhat.com>
17545 PR tree-optimization/48189
17546 * predict.c (predict_loops): If max is 0, don't call compare_tree_int.
17547 If nitercst is 0, don't predict the exit edge.
17549 2013-01-08 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
17551 * config/aarch64/aarch64.c (aarch64_print_operand): Replace %r
17552 in asm_fprintf with reg_names.
17553 (aarch64_print_operand_address): Likewise.
17554 (aarch64_return_addr): Likewise.
17555 * config/aarch64/aarch64.h (ASM_FPRINTF_EXTENSIONS): Remove.
17557 2013-01-08 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
17559 * config/pa/pa.h (VAL_U6_BITS_P): Define.
17560 (INT_U6_BITS): Likewise.
17561 * config/pa/predicates.md (uint6_operand): New predicate.
17562 (shift5_operand, shift6_operand): Likewise.
17563 * config/pa/pa.md (lshrsi3, rotrsi3): Use shift5_operand instead of
17565 (lshrdi3): Use shift6_operand.
17566 (shrpsi4, shrpdi4): New insn patterns.
17567 (extzv): Delete expander.
17568 (extzvsi, extzvdi): New expanders. Use uint5_operand and uint6_operand
17569 predicates in unamed zero extract patterns. Tighten common constraint.
17570 (extv): Delete expander.
17571 (extvsi, extvdi): New expanders. Use uint5_operand and uint6_operand
17572 predicates in unamed sign extract patterns. Tighten common constraint.
17573 (insv): Delete expander.
17574 (insvsi, insvdi): New expanders. Use uint5_operand and uint6_operand
17575 predicates in unamed insert patterns. Tighten common constraint.
17576 Change uint32_operand predicate to uint6_operand predicate in unamed
17577 DImode pattern to insert constant values of type 1...1xxxx.
17579 2013-01-04 Jan Hubicka <jh@suse.cz>
17581 PR tree-optimization/55823
17582 * ipa-prop.c (update_indirect_edges_after_inlining): Fix ordering
17585 2013-01-08 Jakub Jelinek <jakub@redhat.com>
17586 Uros Bizjak <ubizjak@gmail.com>
17588 PR rtl-optimization/55845
17589 * df-problems.c (can_move_insns_across): Stop scanning at
17590 volatile_insn_p source instruction or give up if
17591 across_from .. across_to range contains any volatile_insn_p
17594 2013-01-08 Tejas Belagod <tejas.belagod@arm.com>
17596 * config/aarch64/aarch64-simd.md (vec_init<mode>): New.
17597 * config/aarch64/aarch64-protos.h (aarch64_expand_vector_init):
17599 * config/aarch64/aarch64.c (aarch64_simd_dup_constant,
17600 aarch64_simd_make_constant, aarch64_expand_vector_init): New.
17602 2013-01-08 Jakub Jelinek <jakub@redhat.com>
17605 * asan.c (asan_clear_shadow): New function.
17606 (asan_emit_stack_protection): Use it.
17608 2013-01-08 Tejas Belagod <tejas.belagod@arm.com>
17610 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_<su>mult_lo_<mode>,
17611 aarch64_simd_vec_<su>mult_hi_<mode>): Separate instruction and operand
17612 with tab instead of space.
17614 2013-01-08 Nick Clifton <nickc@redhat.com>
17616 * config/rl78/rl78.c (rl78_expand_prologue): Always select
17617 register bank 0 at the start of an interrupt handler.
17618 * config/rl78/rl78.md (mulsi3_g13): Correct values for MDBL and
17621 2013-01-08 James Greenhalgh <james.greenhalgh@arm.com>
17623 * config/aarch64/aarch64-simd.md
17624 (aarch64_simd_bsl<mode>_internal): Add floating-point modes.
17625 (aarch64_simd_bsl): Likewise.
17626 (aarch64_vcond_internal<mode>): Likewise.
17627 (vcond<mode><mode>): Likewise.
17628 (aarch64_cm<cmp><mode>): Fix constraints, add new modes.
17629 * config/aarch64/iterators.md (V_cmp_result): Add V2DF.
17631 2013-01-08 James Greenhalgh <james.greenhalgh@arm.com>
17633 * config/aarch64/aarch64-builtins.c
17634 (aarch64_builtin_vectorized_function): Handle sqrt, sqrtf.
17636 2013-01-08 Martin Jambor <mjambor@suse.cz>
17639 * tree-sra.c (analyze_access_subtree): Return true also after
17640 potentially creating a debug-only replacement.
17642 2013-01-08 Jakub Jelinek <jakub@redhat.com>
17644 PR middle-end/55890
17645 * tree-ssa-ccp.c (evaluate_stmt): Use gimple_call_builtin_p.
17647 PR tree-optimization/54120
17648 * tree-vrp.c (range_fits_type_p): Don't allow
17649 src_precision < precision from signed vr to unsigned_p
17650 if vr->min or vr->max is negative.
17651 (simplify_float_conversion_using_ranges): Test can_float_p
17652 against CODE_FOR_nothing.
17654 2013-01-08 Jakub Jelinek <jakub@redhat.com>
17655 Richard Biener <rguenther@suse.de>
17657 PR middle-end/55851
17658 * fold-const.c (int_binop_types_match_p): Allow all INTEGRAL_TYPE_P
17659 types instead of just INTEGER_TYPE types.
17661 2013-01-07 Mark Kettenis <kettenis@openbsd.org>
17663 * config/i386/openbsdelf.h (LIBGCC2_HAS_TF_MODE, LIBGCC2_TF_CEXT,
17666 2013-01-07 Steve Ellcey <sellcey@mips.com>
17669 * config/mips/mips.opt: Change mad to mmad to match documentation.
17671 2013-01-07 Georg-Johann Lay <avr@gjlay.de>
17674 * doc/extend.texi (AVR Named Address Spaces): __memx goes into
17675 .progmemx.data now.
17677 2013-01-07 Georg-Johann Lay <avr@gjlay.de>
17680 * config/avr/avr.h (ADDR_SPACE_COUNT): New enum.
17681 (avr_addrspace_t): Add .section_name field.
17682 * config/avr/avr.c (progmem_section): Use ADDR_SPACE_COUNT as
17684 (avr_addrspace): Same. Initialize .section_name. Remove last
17685 NULL entry. Put __memx into .progmemx.data.
17686 (progmem_section_prefix): Remove.
17687 (avr_asm_init_sections): No need to initialize progmem_section.
17688 (avr_asm_named_section): Use avr_addrspace[].section_name to get
17689 section name prefix.
17690 (avr_asm_select_section): Ditto. And use get_unnamed_section to
17691 retrieve the progmem section.
17692 * avr-c.c (avr_cpu_cpp_builtins): Use ADDR_SPACE_COUNT as loop
17693 boundary to run over avr_addrspace[].
17694 (avr_register_target_pragmas): Ditto.
17696 2013-01-06 Jakub Jelinek <jakub@redhat.com>
17698 * varasm.c (output_constant_def_contents): For asan_protect_global
17699 protected strings, adjust DECL_ALIGN if needed, before testing for
17701 (place_block_symbol): Adjust size for asan protected STRING_CSTs if
17702 TREE_CONSTANT_POOL_ADDRESS_P. Increase alignment for asan protected
17704 (output_object_block): For asan protected decls, emit asan padding
17705 after their contents.
17706 * asan.c (asan_protect_global): Don't check TREE_ASM_WRITTEN here.
17707 (asan_finish_file): Test it here instead.
17709 2013-01-07 Nick Clifton <nickc@redhat.com>
17710 Matthias Klose <doko@debian.org>
17711 Doug Kwan <dougkwan@google.com>
17712 H.J. Lu <hongjiu.lu@intel.com>
17715 * collect2.c (main): Support -fuse-ld=bfd and -fuse-ld=gold.
17717 * common.opt: Add fuse-ld=bfd and fuse-ld=gold.
17719 * gcc.c (LINK_COMMAND_SPEC): Pass -fuse-ld=* to collect2.
17721 * opts.c (comman_handle_option): Ignore -fuse-ld=bfd and -fuse-ld=gold.
17723 * doc/invoke.texi: Document -fuse-ld=bfd and -fuse-ld=gold.
17725 2013-01-07 Georg-Johann Lay <avr@gjlay.de>
17728 * doc/install.texi (Cross-Compiler-Specific Options): Document
17731 2013-01-07 Tejas Belagod <tejas.belagod@arm.com>
17733 * config/aarch64/arm_neon.h (vmovn_high_is16, vmovn_high_s32,
17734 vmovn_high_s64, vmovn_high_u16, vmovn_high_u32, vmovn_high_u64,
17735 vqmovn_high_s16, vqmovn_high_s32, vqmovn_high_s64, vqmovn_high_u16,
17736 vqmovn_high_u32, vqmovn_high_u64, vqmovun_high_s16, vqmovun_high_s32,
17737 vqmovun_high_s64): Fix source operand number and update copyright.
17739 2013-01-07 Richard Biener <rguenther@suse.de>
17741 PR middle-end/55890
17742 * gimple.h (gimple_call_builtin_p): New overload.
17743 * gimple.c (validate_call): New function.
17744 (gimple_call_builtin_p): Likewise.
17745 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
17746 Use gimple_call_builtin_p.
17747 (find_func_clobbers): Likewise.
17748 * tree-ssa-strlen.c (adjust_last_stmt): Likewise.
17749 (strlen_optimize_stmt): Likewise.
17751 2013-01-07 James Greenhalgh <james.greenhalgh@arm.com>
17753 * config/aarch64/arm_neon.h (vld1_dup_*): Make argument const.
17754 (vld1q_dup_*): Likewise.
17755 (vld1_*): Likewise.
17756 (vld1q_*): Likewise.
17757 (vld1_lane_*): Likewise.
17758 (vld1q_lane_*): Likewise.
17760 2013-01-07 Richard Biener <rguenther@suse.de>
17762 * lto-streamer.h (LTO_minor_version): Bump to 2.
17764 2013-01-07 James Greenhalgh <james.greenhalgh@arm.com>
17766 * config/aarch64/aarch64-protos.h
17767 (aarch64_const_double_zero_rtx_p): Rename to...
17768 (aarch64_float_const_zero_rtx_p): ...this.
17769 (aarch64_float_const_representable_p): New.
17770 (aarch64_output_simd_mov_immediate): Likewise.
17771 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>): Refactor
17772 move immediate case.
17773 * config/aarch64/aarch64.c
17774 (aarch64_const_double_zero_rtx_p): Rename to...
17775 (aarch64_float_const_zero_rtx_p): ...this.
17776 (aarch64_print_operand): Allow printing of new constants.
17777 (aarch64_valid_floating_const): New.
17778 (aarch64_legitimate_constant_p): Check for valid floating-point
17780 (aarch64_simd_valid_immediate): Likewise.
17781 (aarch64_vect_float_const_representable_p): New.
17782 (aarch64_float_const_representable_p): Likewise.
17783 (aarch64_simd_imm_zero_p): Also allow for floating-point 0.0.
17784 (aarch64_output_simd_mov_immediate): New.
17785 * config/aarch64/aarch64.md (*movsf_aarch64): Add new alternative.
17786 (*movdf_aarch64): Likewise.
17787 * config/aarch64/constraints.md (Ufc): New.
17788 (Y): call aarch64_float_const_zero_rtx.
17789 * config/aarch64/predicates.md (aarch64_fp_compare_operand): New.
17791 2013-01-07 Richard Biener <rguenther@suse.de>
17793 PR tree-optimization/55888
17794 PR tree-optimization/55862
17795 * tree-ssa-pre.c (phi_translate_1): Revert previous change.
17796 (valid_in_sets): Check if a NAME has a leader in AVAIL_OUT,
17797 not if it is contained therein.
17799 2013-01-07 Georg-Johann Lay <avr@gjlay.de>
17801 * config/avr/t-avr: Typo.
17803 2013-01-07 Georg-Johann Lay <avr@gjlay.de>
17806 * config/avr/t-avr: Don't automatically rebuild
17807 $(srcdir)/config/avr/t-multilib
17808 $(srcdir)/config/avr/avr-tables.opt
17809 $(srcdir)/doc/avr-mmcu.texi
17810 (avr-mcus): New phony target to build them on request.
17811 (s-avr-mlib, s-avr-mmcu-texi): Remove.
17812 * avr/avr-mcus.def: Adjust comments.
17814 2013-01-07 Uros Bizjak <ubizjak@gmail.com>
17816 * config/i386/i386.c (DEFAULT_PCC_STRUCT_RETURN): Remove.
17818 2013-01-06 Richard Sandiford <rdsandiford@googlemail.com>
17820 * file-find.c, file-find.h, realmpfr.c: Add FSF as copyright holder.
17822 2013-01-06 Richard Sandiford <rdsandiford@googlemail.com>
17824 * config/tilepro/gen-mul-tables.cc: Put copyright on one line.
17826 2013-01-05 David Edelsohn <dje.gcc@gmail.com>
17828 * config/rs6000/aix53.h (LIB_SPEC): Add -lpthreads when compiling
17829 to generate profiling.
17830 * config/rs6000/aix64.h (LIB_SPEC): Same.
17832 2013-01-04 Andrew Pinski <apinski@cavium.com>
17834 * config/aarch64/aarch64.c (aarch64_fixed_condition_code_regs):
17836 (TARGET_FIXED_CONDITION_CODE_REGS): Define.
17838 2013-01-04 Uros Bizjak <ubizjak@gmail.com>
17840 * config/i386/i386.c (ix86_legitimize_address): Call convert_to_mode
17842 (ix86_expand_move): Ditto.
17843 (ix86_zero_extend_to_Pmode): Ditto.
17844 (ix86_expand_call): Ditto.
17845 (ix86_expand_special_args_builtin): Ditto.
17846 (ix86_expand_builtin): Ditto.
17848 2013-01-04 Richard Biener <rguenther@suse.de>
17850 PR tree-optimization/55862
17851 * tree-ssa-pre.c (phi_translate_1): Valueize SSA names after
17852 translating them through PHI nodes.
17854 2013-01-04 Martin Jambor <mjambor@suse.cz>
17856 PR tree-optimization/55755
17857 * tree-sra.c (sra_modify_assign): Do not check that an access has no
17858 children when trying to avoid producing a VIEW_CONVERT_EXPR.
17860 2013-01-04 Marek Polacek <polacek@redhat.com>
17862 PR middle-end/55859
17863 * opts.c (default_options_optimization): Clarify error message.
17865 2013-01-04 Richard Biener <rguenther@suse.de>
17867 PR middle-end/55863
17868 * fold-const.c (split_tree): Undo -X - 1 to ~X folding for
17871 2013-01-03 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
17874 * config/pa/pa.md (movsi): Revert previous change.
17875 * config/pa/pa.c (pa_legitimate_constant_p): Reject all TLS symbol
17878 2013-01-03 Richard Henderson <rth@redhat.com>
17880 * config/i386/i386.c (ix86_expand_move): Always assign to op1
17881 after eliminating TLS symbols.
17883 2013-01-03 Marc Glisse <marc.glisse@inria.fr>
17886 * graphite-interchange.c (pdr_stride_in_loop): Use gmp_fprintf.
17887 * graphite-poly.c (debug_gmp_value): Likewise.
17889 2013-01-03 Uros Bizjak <ubizjak@gmail.com>
17892 * config/i386/i386-c.c (ix86_target_macros_internal): Depending on
17893 selected code model, define __code_mode_small__, __code_model_medium__,
17894 __code_model_large__, __code_model_32__ or __code_model_kernel__.
17895 * config/i386/cpuid.h (__cpuid, __cpuid_count) [__i386__]: Prefix
17896 xchg temporary register with %k. Declare temporary register as
17898 [__x86_64__]: For medium and large code models, preserve %rbx register.
17900 2013-01-03 Richard Biener <rguenther@suse.de>
17902 * tree-data-ref.c (dump_conflict_function): Use less vertical spacing.
17903 (dump_subscript): Adjust.
17904 (finalize_ddr_dependent): Do not dump redundant info.
17905 (analyze_siv_subscript): Adjust.
17906 (subscript_dependence_tester): Likewise.
17907 (compute_affine_dependence): Likewise.
17909 2013-01-03 Richard Biener <rguenther@suse.de>
17912 2013-01-03 Richard Biener <rguenther@suse.de>
17914 PR tree-optimization/55857
17915 * tree-vect-stmts.c (vectorizable_load): Do not setup
17916 re-alignment for invariant loads.
17918 2013-01-02 Richard Biener <rguenther@suse.de>
17920 * tree-vect-stmts.c (vectorizable_load): When vectorizing an
17921 invariant load do not generate a vector load from the scalar location.
17923 2013-01-03 Richard Biener <rguenther@suse.de>
17925 * tree-vect-loop.c (vect_analyze_loop_form): Clarify reason
17926 for not vectorizing.
17927 * tree-vect-data-refs.c (vect_create_addr_base_for_vector_ref): Do
17928 not build INDIRECT_REFs, call get_name once only.
17929 (vect_create_data_ref_ptr): Likewise. Dump base object kind
17930 based on DR_BASE_OBJECT, not DR_BASE_ADDRESS.
17932 2013-01-03 Richard Biener <rguenther@suse.de>
17934 PR tree-optimization/55857
17935 * tree-vect-stmts.c (vectorizable_load): Do not setup
17936 re-alignment for invariant loads.
17938 2013-01-03 Richard Biener <rguenther@suse.de>
17941 * lto-symtab.c (lto_symtab_merge_decls_1): As last resort, always
17942 prefer a built-in decl.
17944 2013-01-03 Jakub Jelinek <jakub@redhat.com>
17946 * gcc.c (process_command): Update copyright notice dates.
17947 * gcov.c (print_version): Likewise.
17948 * gcov-dump.c (print_version): Likewise.
17950 PR rtl-optimization/55838
17951 * loop-iv.c (iv_number_of_iterations): Call lowpart_subreg on
17952 iv0.step, iv1.step and step.
17954 2013-01-03 Jakub Jelinek <jakub@redhat.com>
17955 Marc Glisse <marc.glisse@inria.fr>
17957 PR tree-optimization/55832
17958 * fold-const.c (fold_binary_loc): For ABS_EXPR<x> >= 0 and
17959 ABS_EXPR<x> < 0 folding use constant_boolean_node instead of
17960 integer_{one,zero}_node.
17962 2013-01-03 Jakub Jelinek <jakub@redhat.com>
17965 * params.def (PARAM_MAX_VARTRACK_REVERSE_OP_SIZE): New param.
17966 * var-tracking.c (reverse_op): Don't add reverse ops to
17967 VALUEs that have already
17968 PARAM_VALUE (PARAM_MAX_VARTRACK_REVERSE_OP_SIZE) or longer locs list.
17970 2013-01-02 Gerald Pfeifer <gerald@pfeifer.com>
17972 * doc/contrib.texi: Note years as release manager for Mark Mitchell.
17974 2013-01-02 Teresa Johnson <tejohnson@google.com>
17976 * dumpfile.c (dump_loc): Print filename with location.
17977 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Use
17978 new location_t parameter to emit complete unroll message with
17979 new dump framework.
17980 (canonicalize_loop_induction_variables): Compute loops location
17981 and pass to try_unroll_loop_completely.
17982 * loop-unroll.c (report_unroll_peel): New function.
17983 (peel_loops_completely): Use new dump format with location
17984 for main dumpfile message, and invoke report_unroll_peel on success.
17985 (decide_unrolling_and_peeling): Ditto.
17986 (decide_peel_once_rolling): Remove old dumpfile message subsumed
17987 by report_unroll_peel.
17988 (decide_peel_completely): Ditto.
17989 (decide_unroll_constant_iterations): Ditto.
17990 (decide_unroll_runtime_iterations): Ditto.
17991 (decide_peel_simple): Ditto.
17992 (decide_unroll_stupid): Ditto.
17993 * cfgloop.c (get_loop_location): New function.
17994 * cfgloop.h (get_loop_location): Declare.
17996 2013-01-02 Sriraman Tallam <tmsriram@google.com>
17998 * config/i386/i386.c (fold_builtin_cpu): Remove unnecessary checks for
18001 2013-01-02 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
18003 PR middle-end/55198
18004 * expr.c (expand_expr_real_1): Don't use bitfield extraction for non
18005 BLKmode objects when EXPAND_MEMORY is specified.
18007 2013-01-02 Sriraman Tallam <tmsriram@google.com>
18009 * config/i386/i386.c (ix86_get_function_versions_dispatcher): Fix bug
18011 (fold_builtin_cpu): Do not share cpu model decls across statements.
18013 2013-01-02 Jason Merrill <jason@redhat.com>
18016 * tree.c (build_array_type_1): Revert earlier change.
18018 2013-01-02 Yufeng Zhang <yufeng.zhang@arm.com>
18020 * config/aarch64/aarch64-cores.def: Add entries for "cortex-a53" and
18022 * config/aarch64/aarch64-tune.md: Re-generate.
18024 2013-01-02 Richard Biener <rguenther@suse.de>
18026 * tree-vect-stmts.c (vectorizable_load): When vectorizing an
18027 invariant load do not generate a vector load from the scalar location.
18029 2013-01-02 Richard Biener <rguenther@suse.de>
18032 * configure.ac: Add $GMPINC to CFLAGS/CXXFLAGS.
18033 * configure: Regenerate.
18035 2013-01-02 Richard Sandiford <rdsandiford@googlemail.com>
18037 * builtins.c (expand_builtin_mathfn, expand_builtin_mathfn_2)
18038 (expand_builtin_mathfn_ternary, expand_builtin_mathfn_3)
18039 (expand_builtin_int_roundingfn_2): Keep the original target around
18040 for the fallback case.
18042 2013-01-02 Richard Sandiford <rdsandiford@googlemail.com>
18044 * tree-vrp.c (range_fits_type_p): Require the MSB of the double_int
18045 to be clear for sign changes.
18047 2013-01-01 Jan Hubicka <jh@suse.cz>
18049 * ipa-inline-analysis.c: Fix formatting.
18051 2013-01-01 Jakub Jelinek <jakub@redhat.com>
18053 PR tree-optimization/55831
18054 * tree-vect-loop.c (get_initial_def_for_induction): Use
18055 gsi_after_labels instead of gsi_start_bb.
18057 Copyright (C) 2013 Free Software Foundation, Inc.
18059 Copying and distribution of this file, with or without modification,
18060 are permitted in any medium without royalty provided the copyright
18061 notice and this notice are preserved.