1 2007-02-02 Ian Lance Taylor <iant@google.com>
3 * expmed.c (expand_divmod): Add comment.
5 2007-02-02 Kazu Hirata <kazu@codesourcery.com>
7 * emit-rtl.c (renumber_insns): Remove.
8 * flags.h: Remove the extern for flag_renumber_insns.
9 * rtl.h: Remove the prototype for renumber_insns.
10 * toplev.c (flag_renumber_insns): Remove.
12 2007-02-02 Hui-May Chang <hm.chang@apple.com>
14 Revert for x86 darwin:
15 2005-06-19 Uros Bizjak <uros@kss-loka.si>
17 * config/i386/i386.c (ix86_function_arg_regno_p): Put back the
18 code before the following patch under TARGET_MACHO.
19 (ix86_function_value_regno_p): Likewise.
21 2007-02-02 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
23 * fold-const.c (negate_expr_p, fold_negate_expr): Handle
26 2007-02-02 Maxim Kuvyrkov <mkuvyrkov@ispras.ru>
29 * config/ia64/ia64.c (ia64_speculate_insn): Restrict to memory loads to
30 general or fp registers. Add comments.
31 * config/ia64/ia64.md (reg_pred_prefix): Add comment.
33 2007-02-02 Paolo Bonzini <bonzini@gnu.org>
35 * pointer-set.c (insert_aux): Only return insertion slot.
36 (pointer_set_insert): Adjust.
37 (pointer_set_traverse, struct pointer_map_t, pointer_map_create,
38 pointer_map_destroy, pointer_map_insert, pointer_map_contains,
39 pointer_map_traverse): New.
40 * pointer-set.h (pointer_set_traverse, struct pointer_map_t,
41 pointer_map_create, pointer_map_destroy, pointer_map_insert,
42 pointer_map_contains, pointer_map_traverse): Declare.
44 2007-02-02 Jakub Jelinek <jakub@redhat.com>
47 * builtins.c (fold_builtin_sprintf): Do not attempt to optimize
48 sprintf (str, "%s"). Do not optimize sprintf (str, "nopercent", p++).
50 2007-02-02 Maxim Kuvyrkov <mkuvyrkov@ispras.ru>
52 * sched-int.h (ds_to_dk, dk_to_ds): Declare functions.
54 (struct _dep): New type.
56 (DEP_PRO, DEP_CON, DEP_KIND): New access macros.
57 (DEP_STATUS): New access macro. The macro with the same name was
58 renamed to DEP_LINK_STATUS.
59 (dep_init): Declare function
61 (struct _dep_link): New type.
62 (dep_link_t): New typedef.
63 (DEP_LINK_NODE, DEP_LINK_NEXT, DEP_LINK_PREV_NEXTP): New access macros.
64 (DEP_LINK_DEP, DEP_LINK_PRO, DEP_LINK_CON, DEP_LINK_KIND): New macros.
65 (DEP_LINK_STATUS): New macro.
66 (debug_dep_links): New debug function.
68 (struct _deps_list): New type.
69 (deps_list_t): New typedef.
70 (DEPS_LIST_FIRST): New access macro.
71 (FOR_EACH_DEP_LINK): New cycle macro.
72 (create_deps_list, free_deps_list, delete_deps_list): Declare
74 (deps_list_empty_p, debug_deps_list, add_back_dep_to_deps_list): Ditto.
75 (find_link_by_pro_in_deps_list, find_link_by_con_in_deps_list): Ditto.
76 (copy_deps_list_change_con): Ditto.
78 (move_dep_link): Declare function.
80 (struct _dep_node): New type.
81 (dep_node_t): New typedef.
82 (DEP_NODE_BACK, DEP_NODE_DEP, DEP_NODE_FORW): New access macros.
84 (struct haifa_insn_data.back_deps): New field to hold backward
85 dependencies of the insn.
86 (struct haifa_insn_data.depend): Rename to forw_deps. Change its type
88 (struct haifa_insn_data.resolved_deps): Rename to resolved_back_deps.
89 Change its type to deps_list_t.
90 (INSN_BACK_DEPS): New access macro to use instead of LOG_LINKS.
91 (INSN_DEPEND): Rename to INSN_FORW_DEPS.
92 (RESOLVED_DEPS): Rename to INSN_RESOLVED_BACK_DEPS.
94 (INSN_COST): Move to haifa-sched.c. Use insn_cost () instead.
96 (DEP_STATUS): Rename to DEP_LINK_STATUS. Fix typo in the comment.
98 (add_forw_dep, delete_back_forw_dep, insn_cost): Update declaration and
102 * sched-deps.c (CHECK): New macro to (en/dis)able sanity checks.
103 (ds_to_dk, dk_to_ds): New functions.
105 (init_dep_1): New static function.
106 (init_dep): New function.
107 (copy_dep): New static function.
109 (dep_link_consistent_p, attach_dep_link, add_to_deps_list): New static
111 (detach_dep_link): New static function.
112 (move_dep_link): New function.
114 (dep_links_consistent_p, dump_dep_links): New static functions.
115 (debug_dep_links): New debugging function.
117 (deps_obstack, dl_obstack, dn_obstack): New static variables.
119 (alloc_deps_list, init_deps_list): New static functions.
120 (create_deps_list): New function.
121 (clear_deps_list): New static function.
122 (free_deps_list, delete_deps_list, deps_list_empty_p): New functions.
123 (deps_list_consistent_p, dump_deps_list): New static functions.
124 (debug_deps_list): New function.
125 (add_back_dep_to_deps_list, find_link_by_pro_in_deps_list): New
127 (find_link_by_con_in_deps_list, copy_deps_list_change_con): Ditto.
129 (maybe_add_or_update_back_dep_1, add_or_update_back_dep_1): Update to
130 use new scheduler dependencies lists.
131 (add_back_dep, delete_all_dependences, fixup_sched_groups): Ditto.
132 (sched_analyze): Ditto. Initialize dependencies lists.
133 (add_forw_dep, compute_forward_dependences): Update to use new
134 scheduler dependencies lists.
136 (init_dependency_caches): Init deps_obstack.
137 (free_dependency_caches): Free deps_obstack.
139 (adjust_add_sorted_back_dep, adjust_back_add_forw_dep): Update to use
140 new scheduler dependencies lists.
141 (delete_forw_dep, add_or_update_back_forw_dep): Ditto.
142 (add_back_forw_dep, delete_back_forw_dep): Ditto.
144 * sched-rgn.c (set_spec_fed, find_conditional_protection, is_pfree):
145 Update to use new scheduler dependencies lists.
146 (is_conditionally_protected, is_prisky, add_branch_dependences): Ditto.
147 (debug_dependencies): Ditto.
148 (schedule_region): Update comments.
150 * sched-ebb.c (earliest_block_with_similiar_load): Update to use new
151 scheduler dependencies lists.
152 (schedule_ebb): Update comments.
154 * rtl.def (DEPS_LIST): Remove.
156 * lists.c (unused_deps_list): Remove.
157 (free_list): Update assertions.
159 (alloc_DEPS_LIST, free_DEPS_LIST_list, free_DEPS_LIST_node): Remove.
160 (remove_free_DEPS_LIST_elem, copy_DEPS_LIST_list): Ditto.
162 * rtl.h (free_DEPS_LIST_list, alloc_DEPS_LIST): Remove declarations.
163 (remove_free_DEPS_LIST_elem, copy_DEPS_LIST_list): Ditto.
165 * haifa-sched.c (comments): Update.
166 (insn_cost1): Remove. Inline the code into insn_cost ().
167 (insn_cost): Update to use new scheduler dependencies lists. Move
168 processing of the dependency cost to dep_cost ().
169 (dep_cost): New function. Use it instead of insn_cost () when
170 evaluating cost of the dependency. Use compatible interface to
171 interact with the target.
172 (priority): Update to use new scheduler dependencies lists.
173 (rank_for_schedule): Ditto. Optimize heuristic that prefers the insn
174 with greater number of insns that depend on the insn.
175 (schedule_insn): Update to use new scheduler dependencies lists. Add
176 code to free backward dependencies lists. Inline and optimize code
177 from resolve_dep () - see PR28071.
178 (ok_for_early_queue_removal): Update to use new scheduler dependencies
179 lists. Update call to targetm.sched.is_costly_dependence hook.
181 (fix_inter_tick, try_ready, fix_tick_ready): Update to use new
182 scheduler dependencies lists.
184 (resolve_dep): Remove. Move the logic to schedule_insn ().
185 (init_h_i_d): Initialize dependencies lists.
187 (process_insn_depend_be_in_spec): Rename to
188 process_insn_forw_deps_be_in_spec. Update to use new scheduler
190 (add_to_speculative_block, create_check_block_twin, fix_recovery_deps):
191 Update to use new scheduler dependencies lists.
192 (clear_priorities, calc_priorities, add_jump_dependencies): Ditto.
194 * ddg.c (create_ddg_dependence, create_ddg_dep_no_link): Update to use
195 new scheduler dependencies lists.
196 (build_intra_loop_deps): Ditto.
198 * target.h (struct _dep): Declare to use in
199 gcc_target.sched.is_costly_dependence.
200 (struct gcc_target.sched.adjust_cost): Fix typo.
201 (struct gcc_target.sched.is_costly_dependence): Change signature to use
202 single dep_t parameter instead of an equivalent triad.
203 (struct gcc_target.sched.adjust_cost_2): Remove.
205 * target-def.h (TARGET_SCHED_ADJUST_COST_2): Remove.
207 * reg-notes.def (DEP_TRUE, DEP_OUTPUT, DEP_ANTI): Update comments.
209 * doc/tm.texi (TARGET_SCHED_IS_COSTLY_DEPENDENCE): Update
211 (TARGET_SCHED_ADJUST_COST_2): Remove documentation.
213 * doc/rtl.texi (LOG_LINKS): Remove part about instruction scheduler.
214 (REG_DEP_TRUE): Document.
216 * config/ia64/ia64.c (ia64_adjust_cost_2): Rename to ia64_adjust_cost.
217 Change signature to correspond to the targetm.sched.adjust_cost hook.
218 Update use in TARGET_SCHED_ADJUST_COST_2.
219 (TARGET_SCHED_ADJUST_COST_2): Rename to TARGET_SCHED_ADJUST_COST.
220 (ia64_dependencies_evaluation_hook, ia64_dfa_new_cycle): Update to use
221 new scheduler dependencies lists.
222 (ia64_gen_check): Ditto.
224 * config/mips/mips.c (vr4130_swap_insns_p): Update to use new scheduler
227 * config/rs6000/rs6000.c (rs6000_is_costly_dependence): Change
228 signature to correspond to the targetm.sched.is_costly_dependence hook.
229 (is_costly_group): Update to use new scheduler dependencies lists.
231 * config/spu/spu.c (spu_sched_adjust_cost): Use insn_cost () function
232 instead of INSN_COST () macro.
234 2007-02-01 Ian Lance Taylor <iant@google.com>
236 * lower-subreg.c (resolve_clobber): Handle a subreg of a concatn.
238 2007-02-01 Guy Martin <gmsoft@gentoo.org>
240 * pa.md (tp_load): Correct mfctl instruction syntax.
242 2007-02-01 Geoffrey Keating <geoffk@apple.com>
244 * config/rs6000/rs6000.c (rs6000_stack_info): Correct
245 altivec_padding_size calculation on AIX. Improve comment, add
246 assert to verify that it's right.
248 * config/rs6000/darwin.h (MD_UNWIND_SUPPORT): Don't define for
251 2007-2-01 Seongbae Park <seongbae.park@gmail.com>
254 * global.c (global_alloc): Add mising initialization of
257 2007-02-01 Roger Sayle <roger@eyesopen.com>
259 * alias.c (init_alias_analysis): Correct whitespace.
260 * bb-reorder.c (fix_edges_for_rarely_executed_code,
261 partition_hot_cold_basic_blocks): Likewise.
262 * builtins.c (expand_builtin_printf, expand_builtin_fprintf,
263 expand_builtin_sprintf, fold_builtin_carg, fold_builtin_sprintf,
264 maybe_emit_sprintf_chk_warning, fold_builtin_sprintf_chk,
265 fold_builtin_snprintf_chk, fold_builtin_printf,
266 fold_builtin_fprintf, do_mpfr_ckconv, do_mpfr_arg1, do_mpfr_arg2,
267 do_mpfr_arg3, do_mpfr_sincos): Likewise.
268 * cfgcleanup.c (cleanup_cfg): Likewise.
269 * cfgexpand.c (tree_expand_cfg): Likewise.
270 * fold-const.c (fold_binary) <RDIV_EXPR>: Likewise.
271 * function.c (get_next_funcdef_no): Likewise.
272 * gengtype.c (main): Likewise.
273 * genmodes.c (main): Likewise.
274 * gcse.c (bypass_conditional_jumps, print_ldst_list): Likewise.
275 * haifa-sched.c (schedule_block, extend_h_i_d): Likewise.
276 * ifcvt.c (noce_emit_move_insn): Likewise.
277 * modulo-sched.c (generate_prolog_epilog, sms_schedule_by_order):
279 * stor-layout.c (get_best_mode): Likewise.
280 * tree-ssa-loop-niter.c (get_val_for): Likewise.
281 * tree-ssa-structalias.c (get_varinfo, get_varinfo_fc,
282 scc_visit, do_ds_constraint, do_complex_constraint, label_visit,
283 perform_var_substitution, solve_graph): Likewise.
284 * tree-vrp.c (vrp_finalize): Likewise.
286 2007-02-01 Ian Lance Taylor <iant@google.com>
288 * lower-subreg.c (simplify_gen_subreg_concatn): If we ask for the
289 high part of a paradoxical subreg, return a constant zero.
291 2007-02-01 Zdenek Dvorak <dvorakz@suse.cz>
293 * toplev.c (lang_dependent_init): Call init_set_costs.
294 * loop-init.c (loop_optimizer_init): Do not call init_set_costs.
296 2007-02-01 Richard Guenther <rguenther@suse.de>
299 * fold-const.c (fold_negate_expr): Allow negating a
300 constant if overflow does not change.
302 2007-02-01 Ralf Wildenhues <Ralf.Wildenhues@gmx.de>
304 * doc/c-tree.texi (Expression trees): Improve markup.
305 * doc/tm.texi (Register Classes, Addressing Modes)
306 (Floating Point): Fix spacing after abbreviations. Fix some
309 2007-02-01 Ben Elliston <bje@au.ibm.com>
311 * doc/invoke.texi: Replace "bugfix" with "bug fix" throughout.
312 * doc/contrib.texi: Likewise.
313 * doc/install.texi: Likewise.
315 2007-01-31 Richard Henderson <rth@redhat.com>
316 Ian Lance Taylor <iant@google.com>
318 * lower-subreg.c: New file.
319 * rtl.def (CONCATN): Define.
320 * passes.c (init_optimization_passes): Add pass_lower_subreg and
322 * emit-rtl.c (update_reg_offset): New static function, broken out
323 of gen_rtx_REG_offset.
324 (gen_rtx_REG_offset): Call update_reg_offset.
325 (gen_reg_rtx_offset): New function.
326 * regclass.c: Revert patch of 2006-03-05, restoring
328 (clear_reg_info_regno): New function.
329 * dwarf2out.c (concatn_loc_descriptor): New static function.
330 (loc_descriptor): Handle CONCATN.
331 * common.opt (fsplit_wide_types): New option.
332 * opts.c (decode_options): Set flag_split_wide_types when
334 * timevar.def (TV_LOWER_SUBREG): Define.
335 * rtl.h (gen_reg_rtx_offset): Declare.
336 (reg_scan_update): Declare.
337 * regs.h (clear_reg_info_regno): Declare.
338 * tree-pass.h (pass_lower_subreg): Declare.
339 (pass_lower_subreg2): Declare.
340 * doc/invoke.texi (Option Summary): List -fno-split-wide-types.
341 (Optimize Options): Add -fsplit-wide-types to -O1 list. Document
343 * doc/rtl.texi (Regs and Memory): Document concat and concatn.
344 * Makefile.in (OBJS-common): Add lower-subreg.o.
345 (lower-subreg.o): New target.
347 2007-01-31 Kazu Hirata <kazu@codesourcery.com>
349 * config/sh/sh.h (HAVE_SECONDARY_RELOADS): Remove.
351 2007-01-31 Anatoly Sokolov <aesok@post.ru>
354 * config/avr/avr.c (DWARF2_ADDR_SIZE): Define.
356 2007-01-31 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
359 * builtins.c (fold_builtin_sqrt): Use MPFR for constant args.
361 2007-01-31 Zdenek Dvorak <dvorakz@suse.cz>
363 * cfgloop.h: Include vec-prim.h.
364 (enum li_flags): Remove LI_ONLY_OLD.
365 (loop_iterator): Changed.
366 (fel_next, fel_init): Iterate over loop tree.
367 (FOR_EACH_LOOP_BREAK): New macro.
368 * loop-unswitch.c (unswitch_loops): Do not pass LI_ONLY_OLD to
370 * tree-ssa-loop-unswitch.c (tree_ssa_unswitch_loops): Ditto.
371 * modulo-sched.c (sms_schedule): Ditto.
372 * tree-vectorizer.c (vectorize_loops): Ditto.
373 * doc/loop.texi: Update information on loop numbering and behavior of
374 FOR_EACH_LOOP wrto new loops.
375 * tree-scalar-evolution.c (compute_overall_effect_of_inner_loop,
376 add_to_evolution_1): Test nestedness of loops instead of comparing
378 * tree-chrec.c (chrec_fold_plus_poly_poly,
379 chrec_fold_multiply_poly_poly, chrec_evaluate,
380 hide_evolution_in_other_loops_than_loop, chrec_component_in_loop_num,
381 reset_evolution_in_loop): Ditto.
382 * Makefile.in (CFGLOOP_H): Add vecprim.h dependency.
384 2007-01-31 Dirk Mueller <dmueller@suse.de>
386 * c-common.c (warn_about_parentheses): Separate warning about
387 un-parenthized sequence of comparison operators from the one
388 which is supposed to warn about x <= y <= z.
390 2007-01-31 Uros Bizjak <ubizjak@gmail.com>
392 * optabs.h (enum optab_index): Add new OTI_isinf.
393 (isinf_optab): Define corresponding macro.
394 * optabs.c (init_optabs): Initialize isinf_optab.
395 * genopinit.c (optabs): Implement isinf_optab using isinf?f2
397 * builtins.c (mathfn_built_in): Handle BUILT_IN_ISINF{,F,L}.
398 (expand_builtin_interclass_mathfn): Expand BUILT_IN_ISINF{,F,L}
400 (expand_builtin): Expand BUILT_IN_ISINF{,F,L} using
401 expand_builtin_interclass_mathfn.
402 * reg_stack.c (subst_stack_regs_pat): Handle UNSPEC_FXAM.
403 * config/i386/i386.md (UNSPEC_FXAM): New constant.
404 (fxam<mode>2_i387): New insn pattern.
405 (isinf<mode>2) New expander to implement isinf, isinff and isinfl
406 built-in functions as x87 inline asm.
408 2007-01-31 Kazu Hirata <kazu@codesourcery.com>
410 * gcc/config/arm/unwind-arm.h (_sleb128_t, _uleb128_t): New.
412 2007-01-30 Eric Christopher <echristo@apple.com>
414 * config.gcc: Add geode.
416 2007-01-31 Kazu Hirata <kazu@codesourcery.com>
418 * cgraphunit.c, config/arm/arm.c, config/m68k/m68k.c,
419 ipa-inline.c, tree-profile.c, tree-ssa-live.c,
420 tree-ssa-math-opts.c, tree-ssanames.c, tree-vect-analyze.c,
421 value-prof.c: Fix comment typos.
423 2007-01-30 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
426 * doc/invoke.texi (Wpointer-arith): Document warning.
428 2007-01-30 Janis Johnson
430 * doc/extend.texi (Decimal Floating Types): Remove decfloat.h from
431 the list of discrepancies from the draft TR.
433 2007-01-30 Dirk Mueller <dmueller@suse.de>
436 * doc/invoke.texi (-Wreturn-type): Update description to
439 2007-01-30 Richard Sandiford <richard@codesourcery.com>
441 * cfgrtl.c (try_redirect_by_replacing_jump): Check only_sets_cc0_p.
443 2007-01-30 Uros Bizjak <ubizjak@gmail.com>
445 * builtins.c (expand_builtin_int_interclass_roundingfn): New function
446 to handle optabs that operate on floating point input argument and
447 output to integer output.
448 (expand_builtin_mathfn) [BUILT_IN_ILOGB]: Move from here ...
449 (expand_builtin_interclass_mathfn) [BUILT_IN_ILOGB]: ... to here.
450 (expand_builtin): Expand BUILT_IN_ILOGB{,F,L} using
451 expand_builtin_interclass_mathfn ().
452 * config/i386/i386.md (fxtractxf3_i387): Rename from *fxtractxf3_i387.
454 (ilogbxf2, ilogb<mode>2): New expanders to implement ilogb, ilogbf and
455 ilogbl built-in functions as x87 intrinsics.
457 2007-01-30 Richard Guenther <rguenther@suse.de>
460 * dwarf2out.c (reference_to_unused): Query varpool if the
463 2007-01-30 Richard Guenther <rguenther@suse.de>
466 * passes.c (execute_one_pass): Reset in_gimple_form to not
467 confuse non-unit-at-a-time mode.
469 2007-01-29 Roger Sayle <roger@eyesopen.com>
470 Richard Guenther <rguenther@suse.de>
472 * fold-const.c (round_up): Make HIGH an unsigned HOST_WIDE_INT to
473 avoid undefined behaviour on overflow. Use force_fit_type_double
474 to construct the constant with the specified TREE_OVERFLOW.
476 2007-01-29 Janis Johnson <janis187@us.ibm.com>
478 * config/dfp-bit.c: Add parameterized support for fp exceptions.
479 * config/dfp-bit.h: Ditto.
481 2007-01-29 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
483 * c-decl.c (pop_scope): Replace warnings with call to
484 warn_for_unused_label.
485 * c-common.h (warn_for_unused_label): Declare.
486 * c-common.c (warn_for_unused_label): Define.
488 2007-01-29 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
490 * tree-optimize.c (update_inlined_to_pointers): Delete unused
493 2007-01-29 Janis Johnson <janis187@us.ibm.com>
495 * Makefile.in (USER_H): Remove decfloat.h.
496 * ginclude/decfloat.h: Delete, moving contents to ...
497 * ginclude/float.h: Add support for decimal floating point,
498 guarded by __STDC_WANT_DEC_FP__.
500 2007-01-29 Mike Stump <mrs@apple.com>
502 * doc/gccint.texi (Top): Rename Loop Representation to Loop
503 Analysis and Representation to resolve case insensitive conflict.
504 * doc/loop.texi (Loop Analysis and Representation): Likewise.
506 2007-01-28 Daniel Berlin <dberlin@dberlin.org>
508 * tree.h (struct tree_memory_tag): Add aliases member.
509 (MTAG_ALIASES): New macro.
510 * tree-ssa-alias.c (alias_bitmap_obstack): New variable.
511 (add_may_alias): Remove pointer-set. Update for may_aliases being
513 (mark_aliases_call_clobbered): Update for may_aliases being a
515 (compute_tag_properties): Ditto.
516 (create_partition_for): Ditto.
517 (compute_memory_partitions): Ditto.
518 (dump_may_aliases_for): Ditto.
519 (is_aliased_with): Ditto.
520 (add_may_alias_for_new_tag): Ditto.
521 (rewrite_alias_set_for): Rewrite for may_aliases being a bitmap.
522 (compute_is_aliased): New function.
523 (compute_may_aliases): Call compute_is_aliased).
524 (init_alias_info): Initialize alias_bitmap_obstack.
525 (union_alias_set_into): New function.
526 (compute_flow_sensitive_aliasing): Use union_aliases_into.
527 (have_common_aliases_p): Rewrite to take two bitmaps and use
529 (compute_flow_insensitive_aliasing): Stop using pointer-sets.
531 (finalize_ref_all_pointers): Update for add_may_alias changes.
532 (new_type_alias): Ditto.
533 * tree-flow-inline.h (may_aliases): Return a bitmap.
534 * tree-dfa.c (dump_variable): Check for MTAG_P'ness.
535 * tree-ssa.c (verify_flow_insensitive_alias_info): Update for
536 may_aliases being a bitmap.
537 * tree-flow.h (struct var_ann_d): Remove may_aliases member.
538 may_aliases now returns a bitmap.
539 * tree-ssa-structalias.c (merge_smts_into): Update for may_aliases
541 * tree-ssa-operands.c (add_virtual_operand): Update for
542 may_aliases being a bitmap.
544 2007-01-29 Daniel Berlin <dberlin@dberlin.org>
546 PR tree-optimization/30630
547 * tree-ssa-structalias.c (do_complex_constraint): Mark correct
550 2007-01-29 Simon Martin <simartin@users.sourceforge.net>
553 * gimplify.c (gimplify_target_expr): Make sure that the TARGET_EXPR is
554 expanded only once even if an error occurs.
556 2007-01-29 Ben Elliston <bje@au.ibm.com>
558 * gcov-io.h (__gcov_indirect_call_profiler): Declare.
559 (__gcov_average_profiler): Likewise.
560 (__gcov_ior_profiler): Likewise.
561 (__gcov_merge_ior): Likewise.
563 2007-01-28 Jan Hubicka <jh@suse.cz>
565 * builtins.c (expand_builtin_memset): Fix typo in my last patch.
566 * value-prof.c (stringop_block_profile): Likewise.
568 2007-01-28 Jan Hubicka <jh@suse.cz>
570 * expr.c (emit_block_move_via_movmem, emit_block_move_via_libcall): Add
571 variant handling histograms; add wrapper.
572 (clear_storage_via_libcall): Export.
573 (emit_block_move_hints): Break out from ...; add histograms.
574 (emit_block_move): ... this one.
575 (clear_storage_hints): Break out from ...; add histograms.
576 (clear_storage): ... this one.
577 (set_storage_via_memset): Handle histogram.
578 * expr.h (emit_block_move_via_libcall, emit_block_move_hints): Declare.
579 (clear_storage_hints, clear_storage_via_libcall): Declare.
580 (set_storage_via_setmem): Update prototype.
581 * doc/md.texi (movmem, setmem): Document new arguments.
583 * value-prof.c (dump_histogram_value, tree_find_values_to_profile): Add
585 (stringop_block_profile): New global function.
586 (tree_stringops_values_to_profile): Profile block size and alignment.
587 * value-prof.h (enum hist_type): add HIST_TYPE_AVERAGE and
589 (struct profile_hooks): Add gen_average_profiler and gen_ior_profiler.
590 (stringop_block_profile): Declare.
591 * builtins.c: Include value-prof.h.
592 (expand_builtin_memcpy, expand_builtin_memset): Pass block profile.
593 * gcov-ui.h (GCOV_COUNTER_NAMES): Add new counter.
594 (GCOV_COUNTER_AVERAGE, GCOV_COUNTER_IOR): New constants.
595 (GCOV_COUNTERS, GCOV_LAST_VALUE_COUNTER): Update.
596 * profile.c (instrument_values): Add new counters.
597 * cfgexpand.c (expand_gimple_basic_block): Propagate histograms to
599 * tree-profile.c (tree_average_profiler_fn, tree_ior_profiler_fn): New.
600 (tree_init_edge_profiler): Build new profilers.
601 (tree_gen_average_profiler, tree_gen_ior_profiler): New.
602 (pass_tree_profile): Add dump.
603 (tree_profile_hooks): Update.
604 * Makefile.in (LIBGCOV): Add new constants.
605 * libgcov.c (__gcov_merge_ior, __gcov_average_profiler,
606 __gcov_ior_profiler): New.
607 * i386.md (movmem/setmem expanders): Add new optional arguments.
609 2007-01-28 David Edelsohn <edelsohn@gnu.org>
611 * doc/md.texi (Standard Pattern Names): Document blockage pattern.
613 2007-01-28 Zdenek Dvorak <dvorakz@suse.cz>
615 * tree-ssa-loop-unswitch.c: Include tree-inline.h.
616 (tree_unswitch_single_loop): Pass eni_size_weights to
618 * tree-ssa-loop-manip.c: Include tree-inline.h.
619 (can_unroll_loop_p): Pass eni_size_weights to
621 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p):
622 Pass eni_size_weights to estimate_num_insns.
623 * tree.h (init_inline_once): Export.
624 * toplev.c (backend_init): Call init_inline_once.
625 * cgraphunit.c (cgraph_process_new_functions,
626 cgraph_analyze_function): Pass eni_inlining_weights to
628 * ipa-inline.c (compute_inline_parameters): Ditto.
629 * tree-ssa-loop-ivcanon.c (tree_num_loop_insns): Pass weights
630 to estimate_num_insns.
631 (try_unroll_loop_completely): Pass eni_size_weights to
633 * tree-eh.c (decide_copy_try_finally): Pass eni_size_weights
634 ot estimate_num_insns.
635 * tree-ssa-loop-prefetch.c: Include tree-inline.h.
636 (loop_prefetch_arrays): Pass eni_time_weights to tree_num_loop_insns.
637 * tree-inline.c (eni_inlining_weights, eni_size_weights,
638 eni_time_weights): New variables.
639 (init_inline_once): Initialize them.
640 (struct eni_data): Mew.
641 (estimate_num_insns_1, estimate_num_insns): Use weights.
642 * tree-inline.h (struct eni_weights_d): New.
643 (eni_inlining_weights, eni_size_weights, eni_time_weights): Declare.
644 (estimate_num_insns): Declaration changed.
645 * cfgloop.h (tree_num_loop_insns): Declaration changed.
646 * Makefile.in (tree-ssa-loop-unswitch.o, tree-ssa-loop-prefetch.o,
647 tree-ssa-loop-manip.o): Add TREE_INLINE_H dependency.
649 2007-01-28 Zdenek Dvorak <dvorakz@suse.cz>
651 * tree-data-ref.c (conflict_fn): Assert that the number of affine
652 relations in the conflict function is valid.
654 2007-01-27 Ian Lance Taylor <iant@google.com>
656 * common.opt: Add fstrict-overflow.
657 * opts.c (decode_options): Set flag_strict_overflow if -O2.
658 * flags.h (TYPE_OVERFLOW_WRAPS): Define.
659 (TYPE_OVERFLOW_UNDEFINED): Define.
660 (TYPE_OVERFLOW_TRAPS): Define. This replaces TYPE_TRAP_SIGNED.
662 * tree.h (TYPE_TRAP_SIGNED): Don't define.
663 * fold-const.c (negate_expr_p): Use TYPE_OVERFLOW_UNDEFINED.
664 (fold_negate_expr): Likewise.
665 (make_range): Likewise.
666 (extract_muldiv_1): Likewise.
667 (maybe_canonicalize_comparison): Likewise.
668 (fold_comparison): Likewise.
669 (fold_binary): Likewise.
670 (tree_expr_nonnegative_p): Likewise.
671 (tree_expr_nonzero_p): Likewise.
672 * tree-vrp.c (compare_values): Likewise.
673 (extract_range_from_binary_expr): Likewise.
674 (extract_range_from_unary_expr): Likewise.
675 * tree-ssa-loop-niter.c (infer_loop_bounds_from_signedness):
677 (nowrap_type_p): Likewise.
678 * tree-scalar-evolution.c (simple_iv): Likewise.
679 * fold-const.c (negate_expr_p): Use TYPE_OVERFLOW_WRAPS.
680 (build_range_check): Likewise.
681 (extract_muldiv_1): Likewise.
682 (fold_comparison): Likewise.
683 * tree-vrp.c (vrp_int_const_binop): Likewise.
684 (extract_range_from_unary_expr): Likewise.
685 * convert.c (convert_to_integer): Likewise.
686 * fold-const.c (fold_negate_expr): Use TYPE_OVERFLOW_TRAPS.
687 (fold_comparison): Likewise.
688 (fold_binary): Likewise.
689 * optabs.c (optab_for_tree_code): Likewise.
690 * tree-vectorizer.c (vect_is_simple_reduction): Likewise.
691 * simplify-rtx.c (simplify_const_relational_operation): Check
692 flag_strict_overflow and flag_trapv.
693 (simplify_const_relational_operation): Likewise.
694 * doc/invoke.texi (Option Summary): Mention -fstrict-overflow.
695 (Optimize Options): Add -fstrict-overflow to -O2 list. Document
698 2007-01-27 Roger Sayle <roger@eyesopen.com>
700 * tree.c (tree_fold_gcd): Delete.
701 * tree.h (tree_fold_gcd): Remove prototype.
702 * tree-data-ref.c (tree_fold_divides_p): Don't use tree_fold_gcd to
703 test whether one constant integer is a multiple of another. Instead
704 call int_const_binop with TRUNC_MOD_EXPR and test for a zero result.
705 * fold-const.c (multiple_of_p): We've determined both TOP and
706 BOTTOM are integer constants so we can call int_const_binop directly
707 instead of the more generic const_binop.
709 2007-01-27 Roger Sayle <roger@eyesopen.com>
711 * fold-const.c (size_binop): In the fast-paths for X+0, 0+X, X-0 and
712 1*X check that the constant hasn't overflowed, to preserve the
714 (round_up): Provide an efficient implementation when rouding-up an
715 INTEGER_CST to a power-of-two.
717 2007-01-28 Ralf Wildenhues <Ralf.Wildenhues@gmx.de>
719 * doc/sourcebuild.texi: Add comma for clarity.
720 * doc/extend.texi: Fix some typos.
721 * doc/passes.texi: Likewise.
722 * doc/cppinternals.texi: Likewise.
723 * doc/c-tree.texi: Likewise.
724 * doc/tree-ssa.texi: Likewise.
725 * doc/install.texi: Likewise.
727 2007-01-27 Jan Hubicka <jh@suse.cz>
729 * tree-sra.c (sra_walk_function): Don't rely on aliases being build.
730 (pass_sra): Do not require alias information.
731 * passes.c (init_optimization_passes): Add SRA
733 2007-01-27 Steven Bosscher <steven@gcc.gnu.org>
735 * tracer.c (rest_of_handle_tracer): We already cleaned
736 up the CFG in tracer() so don't do it here again.
737 * cfgcleanup.c (rest_of_handle_jump2): Don't repeat
738 cleanup_cfg here, either. And don't call renumber_insns.
740 * cfgrtl.c (rtl_verify_flow_info_1): Don't verify that BB_END
741 and BB_HEAD are in the insn stream here. Instead make sure
742 that BB_INSN is valid on all insns. Also, do check here that
743 there are no pending branch predictions...
744 (rtl_verify_flow_info): ...instead of doing it here. Checks
745 for BB_END and BB_HEAD moved from rtl_verify_flow_info_1 to
748 2007-01-26 Roger Sayle <roger@eyesopen.com>
750 * config/i386/i386.c (ix86_swap_binary_operands_p): New helper
751 function to simplify/factorize operand order canonicalization.
752 (ix86_fixup_binary_operands): Reorganize using the above function.
753 (ix86_binary_operator_ok): Likewise.
755 2007-01-27 Jakub Jelinek <jakub@redhat.com>
757 * genattrtab.c (struct attr_value_list, insn_code_values): Move to
758 file scope from optimize_attrs.
759 (simplify_test_exp): If insn_code_values is not NULL, use it to speed
761 (optimize_attrs): Clear insn_code_values after freeing it.
763 2007-01-26 Zdenek Dvorak <dvorakz@suse.cz>
765 * tree-ssa-address.c (create_mem_ref): Remove ", bsi" from
766 a parts.base assignment.
768 2007-01-26 Zdenek Dvorak <dvorakz@suse.cz>
770 * tree-data-ref.c (dump_subscript): Use dump_conflict_function.
771 (compute_subscript_distance, initialize_data_dependence_relation,
772 finalize_ddr_dependent, analyze_ziv_subscript,
773 analyze_siv_subscript_cst_affine,
774 compute_overlap_steps_for_affine_univar,
775 compute_overlap_steps_for_affine_1_2, analyze_subscript_affine_affine,
776 analyze_siv_subscript, analyze_miv_subscript,
777 analyze_overlapping_iterations, subscript_dependence_tester_1,
778 compute_self_dependence, free_dependence_relation): Work
779 with affine_fn instead of chrecs.
780 (dump_affine_function, dump_conflict_function, affine_function_equal_p,
781 common_affine_function, affine_function_base,
782 affine_function_constant_p, affine_fn_op, affine_fn_plus,
783 affine_fn_minus, affine_fn_free, conflict_fn_not_known,
784 conflict_fn_no_dependence, free_conflict_function, free_subscripts,
785 conflict_fn, affine_fn_cst, affine_fn_univar): New functions.
786 (all_chrecs_equal_p): Removed.
787 * tree-data-ref.h (affine_fn, conflict_function): New types.
788 (struct subscript): Change type of conflicting_iterations_in_a
789 and conflicting_iterations_in_b.
791 2007-01-26 Steve Ellcey <sje@cup.hp.com>
794 * config/pa/pa.h (TARGET_HPUX_11): New.
795 * config/pa/pa-hpux11.h (TARGET_HPUX_11): New.
796 * config/pa/pa.c (pa_init_builtins): Use TARGET_HPUX_11.
798 2007-01-26 Daniel Berlin <dberlin@dberlin.org>
799 Richard Guenther <rguenther@suse.de>
801 * tree-ssa-structalias.c (solve_graph): Handle case
802 we merged the variable to another.
804 2007-01-25 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
806 * builtins.c (fold_builtin_1): Treat ccos and ccosh as 'even'
809 * fold-const.c (negate_mathfn_p): Treat casin, casinh, catan,
810 catanh, cproj, csin, csinh, ctan and ctanh as 'odd' functions.
812 2007-01-25 DJ Delorie <dj@redhat.com>
814 * config/m32c/m32c.c (m32c_cannot_change_mode_class): We don't
815 allow changes to modes which don't fit in those registers.
817 * reload1.c (choose_reload_regs): Check for invalid subregs before
818 computing their locations, not after.
820 2007-01-25 Geoffrey Keating <geoffk@apple.com>
823 * config/rs6000/rs6000.c (first_altivec_reg_to_save): On Darwin,
824 save Altivec registers in an eh_return function.
825 (compute_vrsave_mask): Likewise.
826 (rs6000_stack_info): Correct AIX/Darwin stack alignment computation
827 for saving Altivec registers.
828 (rs6000_emit_prologue): Don't allocate stack twice in
829 eh_return function. Correct expected value of altivec_save_offset
830 when using save_world. Describe save of R0 to stack when using
831 save_world. Describe stack pointer adjustment when using
832 save_world. Remove duplicated eh_return parameter register saving.
833 Update sp_offset variable after save_world.
834 * config/rs6000/t-darwin (LIB2FUNCS_STATIC_EXTRA): Remove
836 (LIB2FUNCS_EXTRA): Add darwin-world.asm.
837 * config/rs6000/darwin.h (SUBTARGET_OVERRIDE_OPTIONS): -m64
840 2007-01-25 Steve Ellcey <sje@cup.hp.com>
842 * config.gcc (ia64*-*-hpux*): Make posix threads the default.
844 2007-01-25 Steve Ellcey <sje@cup.hp.com>
847 * config/pa/pa.c (pa_init_builtins): Set asm names for finite routines.
848 * config/ia64/ia64.c (ia64_init_builtins): Ditto.
850 2007-01-25 Richard Guenther <rguenther@suse.de>
852 * doc/invoke.texi (-Wcoverage-mismatch): Document.
853 * common.opt (-Wcoverage-mismatch): New warning option.
854 * coverage.c (get_coverage_counts): Ignore coverage mismatch
855 if -Wcoverage-mismatch is given.
857 2007-01-25 Razya Ladelsky <razya@il.ibm.com>
859 * ipa-cp.c (ipcp_insert_stage, ipcp_driver): Support for SSA.
860 (ipcp_driver): Change to static definition.
861 Add dumping of the ifunctions.
862 (constant_val_insert): Remove unused parameter. Support for SSA.
863 (ipcp_propagate_const): Support for SSA.
864 (ipcp_profile_bb_print): Print only analyzed nodes.
865 (ipcp_replace_map_create): Remove support for Fortran constant
867 * ipa-prop.c (ipa_method_modify_stmt,
868 ipa_callsite_compute_param): Support for SSA.
869 * ipa-prop.h (ipcp_driver): Remove declaration.
870 (IS_VALID_TREE_MAP_INDEX): Add define.
872 2007-01-24 Geoffrey Keating <geoffk@apple.com>
874 * unwind-dw2.c (execute_stack_op): Handle DW_OP_swap.
876 2007-01-24 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
879 * builtins.c (fold_builtin_cabs): Use MPFR to evaluate a
880 constant argument to cabs and do it without checking for
881 -funsafe-math-optimizations.
883 2007-01-24 Douglas Gregor <dgregor@osl.iu.edu>
885 * c-common.h (RID_FIRST_CXX0X): New.
886 (RID_LAST_CXX0X): New.
887 * c-opts.c (c_common_handle_option): -Wc++0x-compat is triggered
889 * c.opt (Wc++0x-compat): New.
890 * doc/invoke.texi (-Wc++0x-compat): Document.
892 2007-01-24 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
894 * builtins.c (fold_builtin_carg): New.
895 (fold_builtin_1): Use it.
897 2007-01-24 Jan Hubicka <jh@suse.cz>
899 * ipa-inline.c (cgraph_decide_inlining): Initialize initial_insns.
901 * ipa-inline.c (initial_insns, max_insns): Delete.
902 (compute_max_insns): New function.
903 (cgraph_decide_inlining_of_small_function): Use it; take minimal amount
904 of insns as base for code growth.
905 (cgraph_decide_inlining): Make initial_insns local; do not compute
907 * params.def (PARAM_INLINE_UNIT_GROWTH): Set to 60.
908 * doc/invoke.texi (inline-unit-growth): Update docs.
910 2007-01-24 Jakub Jelinek <jakub@redhat.com>
912 * config/i386/i386.h (x86_cmpxchg16b): Remove const.
913 (TARGET_CMPXCHG16B): Define to x86_cmpxchg16b.
914 * config/i386/i386.c (x86_cmpxchg16b): Remove const.
915 (override_options): Add PTA_CX16 flag. Set x86_cmpxchg16b
916 for CPUs that have PTA_CX16 set.
919 * gimplify.c (omp_check_private): New function.
920 (gimplify_scan_omp_clauses): Use it for
921 firstprivate/lastprivate/reduction.
924 * gimplify.c (omp_add_variable): Don't call omp_notice_variable
925 on TYPE_SIZE_UNIT for GOVD_LOCAL VLAs.
928 * omp-low.c (lower_omp_for_lastprivate): Add dlist argument.
929 If lower_lastprivate_clauses emits some statements, append them
930 to dlist rather than body_p and to body_p append an initializer.
931 (lower_omp_for): Adjust caller.
933 2007-01-24 Steve Ellcey <sje@cup.hp.com>
935 * target.h (globalize_decl_name): New.
936 * target-def.h (TARGET_ASM_GLOBALIZE_DECL_NAME): New.
937 * output.h (default_globalize_decl_name): New.
938 * varasm.c (asm_output_bss): Use globalize_decl_name instead of
940 (globalize_decl): Ditto.
941 (default_globalize_decl_name): New.
942 * config/ia64/ia64.c (ia64_globalize_decl_name): New.
943 (ia64_handle_version_id_attribute): New.
944 (TARGET_ASM_GLOBALIZE_DECL_NAME): New.
945 (ia64_asm_output_external): Use globalize_decl_name instead
947 * doc/extend.texi (version_id): New pragma.
948 * doc/tm.texi (ARGET_ASM_GLOBALIZE_DECL_NAME): New target hook.
950 2007-01-24 Andreas Krebbel <krebbel1@de.ibm.com>
952 * unwind-dw2-fde.c (get_cie_encoding): Replaced _Unwind_Word with
953 _uleb128_t and _Unwind_SWord with _sleb128_t.
954 * unwind-dw2.c (extract_cie_info, execute_stack_op, execute_cfa_program,
955 uw_frame_state_for, uw_update_context_1): Likewise.
956 * unwind-c.c (parse_lsda_header, PERSONALITY_FUNCTION): Likewise.
957 * unwind-pe.h (read_uleb128, read_sleb128,
958 read_encoded_value_with_base): Likewise.
959 * unwind-generic.h: Define _sleb128_t and _uleb128_t types.
961 2007-01-24 Richard Guenther <rguenther@suse.de>
963 * builtins.c (expand_builtin_cexpi): Get the fndecl
964 for cexp in the correct way.
966 2007-01-24 Jan Hubicka <jh@suse.cz>
968 * tree-ssa-dce.c (eliminate_unnecesary_stmts): Remove dead LHS of calls.
970 2007-01-24 Andreas Krebbel <krebbel1@de.ibm.com>
972 * c-cppbuiltin.c (builtin_define_type_sizeof): New function.
973 (c_cpp_builtins): New builtin macros: __SIZEOF_INT__, __SIZEOF_LONG__,
974 __SIZEOF_LONG_LONG__, __SIZEOF_SHORT__, __SIZEOF_POINTER__,
975 __SIZEOF_FLOAT__, __SIZEOF_DOUBLE__, __SIZEOF_LONG_DOUBLE__,
976 __SIZEOF_SIZE_T__, __SIZEOF_WCHAR_T__, __SIZEOF_WINT_T__ and
977 __SIZEOF_PTRDIFF_T__.
978 * doc/cpp.texi: Documentation for the new builtin macros added.
980 2007-01-24 Uros Bizjak <ubizjak@gmail.com>
982 * config/i386/i386.md (tanxf2, tan<mode>2, atan<mode>2, log<mode>2,
983 log10<mode>2, log2<mode>2, expxf2, exp10xf2, exp2xf2): Use op2
984 instead of operands[2] to avoid access past the end of array.
986 2007-01-24 Richard Sandiford <richard@codesourcery.com>
988 * reload1.c (emit_reload_insns): Pass the reload register
989 for a non-spill output reload through forget_old_reloads_1.
991 2007-01-23 Joseph Myers <joseph@codesourcery.com>
993 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Define
994 _SOFT_DOUBLE if doubles use software floating-point.
995 * config/rs6000/libgcc-ppc-glibc.ver: Export additional long
996 double functions if _SOFT_DOUBLE, not _SOFT_FLOAT.
997 * config/rs6000/darwin-ldouble.c: Also compile functions for
998 hard-float without FPRs. Use fmsub function for all __NO_FPRS__
999 cases. Compile extra functions if _SOFT_DOUBLE, not _SOFT_FLOAT.
1000 * config/rs6000/linuxspe.h (SUBSUBTARGET_OVERRIDE_OPTIONS): Remove
1001 commented-out long double override.
1002 (CPP_LONGDOUBLE_DEFAULT_SPEC): Likewise.
1003 * config/rs6000/eabispe.h: Likewise.
1004 * config/rs6000/rs6000.c (rs6000_override_options): Don't override
1005 long double for non-SPE.
1006 (rs6000_handle_option): Likewise.
1007 (invalid_e500_subreg): Disallow more subregs involding DImode,
1008 DFmode, TImode or TFmode.
1009 (rs6000_legitimate_offset_address_p): Check TFmode offsets for
1011 (legitimate_lo_sum_address_p): Also check for TFmode for E500
1013 (rs6000_legitimize_address): Also handle TFmode for E500 double.
1014 (rs6000_legitimize_reload_address): Also handle TFmode for E500
1016 (rs6000_legitimate_address): Also check for TFmode for E500
1018 (rs6000_emit_move): Use DFmode subregs of TFmode for E500 double.
1019 (spe_build_register_parallel): Handle TFmode and TCmode.
1020 (rs6000_spe_function_arg): Handle TFmode and TCmode for E500
1022 (function_arg): Handle TFmode and TCmode for E500 double.
1023 (rs6000_init_libfuncs): Initialize extra libfuncs for soft double
1025 (print_operand): Handle TFmode and TImode for %y.
1026 (rs6000_generate_compare): Handle TFmode comparisons for E500
1028 (spe_func_has_64bit_regs_p): Check for TFmode for E500 double.
1029 (rs6000_function_value): Handle TFmode and TCmode for E500 double.
1030 (rs6000_libcall_value): Handle TFmode and TCmode for E500 double.
1031 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Check for
1032 TFmode for E500 double.
1033 * config/rs6000/rs6000.md (FP): Allow TF for E500 double.
1034 (floatsidf2): Enable for E500 double.
1035 (movtf_softfloat): Use rs6000_nonimmediate_operand.
1036 (extenddftf2): Change to extenddftf2_fprs.
1037 (extenddftf2): Call gen_spe_extenddftf2 or gen_extenddftf2_fprs
1038 depending on TARGET_E500_DOUBLE.
1039 (extendsftf2): Enable for E500 double.
1040 (trunctfdf2): Enable for E500 double.
1041 (trunctfsf2): Change to trunctfsf2_fprs.
1042 (trunctfsf2): Call gen_spe_trunctfsf2 or gen_trunctfsf2_fprs
1043 depending on TARGET_E500_DOUBLE.
1044 (floatsitf2): Enable for E500 double.
1045 (fix_trunctfsi2): Change to fix_trunctfsi2_fprs.
1046 (fix_trunctfsi2): Call gen_spe_fix_trunctfsi2 or
1047 gen_fix_trunctfsi2_fprs depending on TARGET_E500_DOUBLE.
1048 (negtf2): Change to negtf2_internal.
1049 (negtf2): New expander.
1050 (abstf2): Enable for E500 double. Call gen_spe_abstf2_tst,
1051 gen_spe_abstf2_cmp or gen_abstf2_internal depending on
1052 TARGET_E500_DOUBLE and flag_unsafe_math_optimizations.
1053 (movdi_internal32): Use rs6000_nonimmediate_operand.
1054 (unnamed splitter): Likewise.
1055 * config/rs6000/spe.md (CMPTFEQ_GPR, TSTTFEQ_GPR, CMPTFGT_GPR,
1056 TSTTFGT_GPR, CMPTFLT_GPR, TSTTFLT_GPR): New unspecs.
1057 (SPE64TF, DITI): New mode macros.
1058 (frob_df_di): Change to frob_<SPE64:mode>_<DITI:mode>; allow more
1061 (frob_<mode>_di_2): New.
1062 (frob_tf_di_8_2): New.
1063 (frob_di_df): Change to frob_di_<mode>; allow more modes.
1065 (frob_di_df_2): Change to frob_<DITI:mode>_<SPE64:mode>_2; allow
1067 (frob_ti_<mode>_8_2): New.
1068 (frob_ti_tf_2): New.
1069 (mov_si<mode>_e500_subreg0, mov_si<mode>_e500_subreg0_2,
1070 mov_si<mode>_e500_subreg4, mov_si<mode>_e500_subreg4_2): Allow
1072 (mov_sitf_e500_subreg8, mov_sitf_e500_subreg8_2,
1073 mov_sitf_e500_subreg12, mov_sitf_e500_subreg12_2): New.
1074 (spe_trunctfdf2_internal1, spe_trunctfsf2, spe_extenddftf2,
1075 spe_fix_trunctfsi2, spe_fix_trunctfsi2_internal,
1076 spe_negtf2_internal, spe_abstf2_cmp, spe_abstf2_tst): New.
1077 (cmptfeq_gpr, tsttfeq_gpr, cmptfgt_gpr, tsttfgt_gpr, cmptflt_gpr,
1080 2007-01-23 Ian Lance Taylor <iant@google.com>
1082 * Makefile.in (OBJS-common): Reformat, alphabetize, but put
1084 (OBJS-archive): Reformat, alphabetize.
1085 (OBJS): Change out_object_file to OBJS-md.
1087 2007-01-23 Uros Bizjak <ubizjak@gmail.com>
1089 * config/i386/i386.md: Use REG_P, MEM_P, CONST_INT_P, LABEL_P,
1090 JUMP_P and CALL_P predicates where applicable.
1091 * config/i386/i386.c: Ditto.
1092 * config/i386/i386.md: Ditto.
1093 * config/i386/mmx.md: Ditto.
1094 * config/i386/predicates.md: Ditto.
1096 2007-01-22 Andreas Schwab <schwab@suse.de>
1098 * config/m68k/m68k.h: Fix comment.
1100 2007-01-22 Jan Hubicka <jh@suse.cz>
1102 * passes.c (init_optimization_passes): Do not rerun
1103 pass_early_warn_uninitialized.
1105 2007-01-22 Richard Guenther <rguenther@suse.de>
1107 PR tree-optimization/30038
1108 * tree-ssa-math-opts.c (maybe_record_sincos): New static helper
1110 (execute_cse_sincos_1): Likewise.
1111 (execute_cse_sincos): Likewise.
1112 (gate_cse_sincos): Likewise.
1113 (pass_cse_sincos): New pass CSEing sin() and cos() calls using
1114 the cexpi() canonicalization of sincos().
1115 * tree-pass.h (pass_cse_sincos): Declare.
1116 * passes.c (init_optimization_passes): New pass pas_cse_sincos.
1118 2007-01-21 Eric Botcazou <ebotcazou@libertysurf.fr>
1120 PR rtl-optimization/29329
1121 * combine.c (replaced_rhs_insn): Rename to i2mod.
1122 (replaced_rhs_value): Rename to i2mod_new_rhs.
1123 (i2mod_old_rhs): New global variable.
1124 (combine_instructions): Adjust for above change. Save a copy of
1125 the old RHS into i2mod_old_rhs when the contents of a REG_EQUAL
1126 note are substituted in the second instruction.
1127 (distribute_notes) <REG_DEAD>: Adjust for above change. Do not
1128 ditch the note if it pertains to the second eliminated register
1129 and this register is mentioned in i2mod_old_rhs.
1132 2006-09-12 Eric Botcazou <ebotcazou@libertysurf.fr>
1134 * combine.c (distribute_notes) <REG_DEAD>: Do not consider SETs past
1135 the insn to which the note was originally attached.
1137 2007-01-21 Jan Hubicka <jh@suse.cz>
1139 * ipa-inline.c (inlining_mode): Comment, move up.
1140 (cgraph_decide_inlining_incrementally): Do not perform inlining
1141 itself; fix handling of flattening of self recursive functions.
1142 (cgraph_find_cycles): Remove.
1143 (cgraph_flatten_node): Remove.
1144 (cgraph_decide_inlining): Use incremental inliner to handle flattening.
1145 (try_inline): New function.
1146 (cgraph_early_inlining): Update call of
1147 cgraph_decide_inlining_incrementally. Apply inlining here.
1148 (apply_inline): Update call of cgraph_decide_inlining_incrementally.
1150 2007-01-21 Dirk Mueller <dmueller@suse.de>
1153 * tree-vrp.c (check_array_bounds): do not warn
1154 about ADDR_EXPR's of ARRAY_REF's which are immediately
1155 used in binary expressions.
1157 2007-01-21 Ira Rosen <irar@il.ibm.com>
1159 * tree-vectorizer.h (struct _stmt_vec_info): Add new field
1160 read_write_dep and macros for its access.
1161 * tree-vectorizer.c (new_stmt_vec_info): Initialize the new field.
1162 * tree-vect-analyze.c (vect_analyze_data_ref_dependence): Remove
1163 argument, call vect_check_interleaving for every independent pair of
1164 data-refs. Mark loads that access the same memory location as a store
1166 (vect_check_dependences): Remove.
1167 (vect_analyze_data_ref_dependences): Remove vect_check_dependences
1168 call, fix the call to vect_analyze_data_ref_dependence.
1169 (vect_analyze_data_ref_access): For statements that access the same
1170 data-ref, check that they are not stores; for loads, check that there
1171 is no store that access the same location.
1173 2007-01-20 Roger Sayle <roger@eyesopen.com>
1174 Joseph Myers <joseph@codesourcery.com>
1176 * doc/invoke.texi (-fdump-rtl-combine): Fix under/overfull hbox.
1177 (-fdump-rtl-stack): Likewise.
1178 (-fno-signed-zeros): Use @minus{} for a minus sign. Correct typo.
1179 (-mcheck-zero-division, -mcpu): Fix under/overfull hbox.
1180 (-mpt-fixed): Use @minus{} for minus sign.
1181 (Using Precompiled Headers): Fix under/overfull hbox.
1183 2007-01-20 Jan Hubicka <jh@suse.cz>
1185 * tree-flow.h (struct stmt_ann_d): Move references_memory to proper
1186 place within annotation.
1188 2007-01-20 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
1190 * pa.c (output_move_double): Change array size of xoperands to 4.
1192 2007-01-20 Jan Hubicka <jh@suse.cz>
1194 * tree-tailcall.c (adjust_return_value): Do not use RESULT_DECL
1197 2007-01-19 Ian Lance Taylor <iant@google.com>
1199 * expmed.c (expand_divmod) [TRUNC_MOD_EXPR, TRUNC_DIV_EXPR]: Cast
1200 constant to unsigned HOST_WIDE_INT before negating.
1202 2007-01-19 Ian Lance Taylor <iant@google.com>
1204 * tree-ssa-operands.h (struct vuse_vec_d): Change num_vuse field
1206 (VUSE_VECT_ELEMENT) [ENABLE_CHECKING]: Use unsigned comparison.
1207 (VUSE_ELEMENT_PTR) [ENABLE_CHECKING]: Likewise.
1208 (SET_VUSE_VECT_ELEMENT) [ENABLE_CHECKING]: Likewise.
1209 (SET_VUSE_ELEMENT_VAR) [ENABLE_CHECKING]: Likewise.
1210 (SET_VUSE_ELEMENT_PTR) [ENABLE_CHECKING]: Likewise.
1211 (realloc_vdef, realloc_vuse): Change second parameter to
1213 (ssa_operand_iterator_d): Change vuse_index and mayuse_index
1215 * tree-ssa-operands.c (realloc_vop): Change num_elem parameter to
1216 unsigned. Change x and lim locals to unsigned.
1217 (realloc_vdef, realloc_vuse): Change num_elem parameter to
1219 (finalize_ssa_vuse_ops): Change old_i local to unsigned.
1220 (copy_virtual_operands): Change i and n locals to unsigned.
1222 2007-01-19 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
1225 * builtins.c (fold_builtin_1): Handle builtin fdim.
1227 2007-01-20 Jan Hubicka <jh@suse.cz>
1229 * tree-ssa.c (init_tree_ssa): Do not call init_alias_heapvars.
1230 * tree-ssa-structalias.c (compute_points_to_sets): Do call
1231 init_alias_heapvars.
1232 (init_alias_heapvars): Initialize only when not already initialized.
1233 (delete_alias_heapvars): Set heapvar_for_stmt to NULL.
1235 2007-01-19 Roger Sayle <roger@eyesopen.com>
1237 * common.opt (fsigned-zeros): New command line option.
1238 * flags.h (HONOR_SIGNED_ZEROS): Control via flag_signed_zeros instead
1239 of flag_unsafe_math_optimizations.
1240 * opts.c (set_fast_math_flags): The -ffast-math command line option
1241 implies -fno-signed-zeros.
1242 (fast_math_flags_set_p): Likewise.
1244 * doc/invoke.texi: Document new -fno-signed-zeros option, and update
1245 the documentation of -ffast-math appropriately. Wrap long lines.
1247 2007-01-19 Steve Ellcey <sje@cup.hp.com>
1249 * system.h (ASM_MAKE_LABEL_LINKONCE): Poison.
1250 * varasm.c (globalize_decl): Remove ASM_MAKE_LABEL_LINKONCE ifdef.
1252 2007-01-19 Tomas Bily <tbily@suse.cz>
1254 * cgraphunit.c (cgraph_finalize_function): Updating of pid
1256 (tree_init_ic_make_global_vars): New function
1257 (tree_init_edge_profiler): call of tree_init_ic_make_global_vars
1258 (tree_gen_ic_profiler): New function
1259 (tree_gen_ic_func_profiler): New function
1260 (tree_profiling): Added calling of tree_gen_ic_func_profiler
1261 (tree_profile_hooks): Added hook for indirec/virtual calls
1262 * value-prof.c (tree_find_values_to_profile): New case for
1264 (tree_values_to_profile): Call for determining indirect/virtual
1266 (tree_indirect_call_to_profile): New function
1267 (tree_ic_transform): New function
1268 (tree_ic): New function
1269 (find_func_by_pid): New function
1270 (init_pid_map): New function
1271 (tree_value_profile_transformations): Added check for
1272 indirect/virtual call transformation
1273 * value-prof.h (enum hist_type): New counter type for
1274 indirect/virtual calls
1275 (profile_hooks): Added new hook for profiling indirect/virtual
1277 * profile.c (instrument_values): New case for indirect/virtual
1279 * gcov-io.h (GCOV_LAST_VALUE_COUNTER): Changed to 6
1280 (GCOV_COUNTER_V_INDIR): New counter type
1281 (GCOV_COUNTER_NAMES): New name of counter "indirect" added
1282 (GCOV_MERGE_FUNCTIONS): New merge function for indirect/virtual
1284 * cgraph.c: Definition of cgraph_max_pid
1285 (cgraph_create_node): Default init of pid attribute
1286 * cgraph.h: Declaration of cgraph_max_pid
1287 (struct cgraph_node): Added pid attribute
1288 * libgcov.c (__gcov_indirect_call_profiler): New function
1289 (__gcov_one_value_profiler_body): New function
1290 (__gcov_one_value_profiler): Body was moved to
1291 __gcov_one_value_profiler_body and calls it
1293 2007-01-19 Basile Starynkevitch <basile@starynkevitch.net>
1295 * doc/gty.texi (Options): Document the mark_hook option to GTY.
1296 * gengtype.c (write_types_data, write_func_for_structure,
1297 write_types, ggc_wtd, pch_wtd): Add skip_hooks to
1298 write_types_data, ggc_wtd, pch_wtd for processing mark_hook.
1299 (walk_type, write_func_for_structure): Generate the mark_hook if
1302 2007-01-19 Jan Hubicka <jh@suse.cz>
1304 * ipa-inline.c (cgraph_decide_inlining_incrementally): Instead of
1305 'early' argument take inlining mode argument specifying whether to
1306 inline for size/speeed or all functions; add support for flattening;
1308 (cgraph_early_inlining): Update call of decide_inlining_incrementally.
1310 2007-01-19 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
1313 * toplev.c (warn_deprecated_use): Use %qD instead of %qs to print
1314 the name of the declared identifier.
1316 2007-01-19 Dirk Mueller <dmueller@suse.de>
1318 * config/i386.h (CONDITIONAL_REGISTER_USAGE): Store
1319 result of PIC_OFFSET_TABLE_REGNUM in temporary variable to avoid
1320 duplicate evaluation.
1322 2007-01-19 Uros Bizjak <ubizjak@gmail.com>
1324 * config/i386/i386.md (acos<mode>2): Rename from acossf2 and acosdf2.
1325 Macroize expander using X87MODEF12 mode macro. Extend operand 1
1326 to XFMode, use acosxf2 and truncate result to requested mode.
1327 Use SSE_FLOAT_MODE_P to disable patterns for SSE math.
1328 (asin<mode>2): Similarly, with asin expanders.
1329 (*fscalexf4_i387): Rename from *fscalexf4.
1330 (expNcorexf3): New expander.
1331 (expxf2, exp10xf2, exp2xf2): Use expNcorexf3 expander.
1332 (exp<mode>2): Rename from expsf2 and expdf2. Macroize expander using
1333 X87MODEF12 mode macro. Extend operand 1 to XFMode, use expxf2 and
1334 truncate result to requested mode. Use SSE_FLOAT_MODE_P to disable
1335 patterns for SSE math.
1336 (exp10<mode>2): Similarly, with exp10 expanders.
1337 (exp2<mode>2): Similarly, with exp2 expanders.
1338 (expm1<mode>2): Similarly, with expm1 expanders.
1339 (ldexp<mode>3): Similarly, with ldexp expanders.
1340 (log<mode>2, log10<mode>2, log2<mode>2, log1p<mode>2, logb<mode>2):
1341 Use gen_truncxf<mode>2_i387_noop to truncate result.
1343 2007-01-19 Richard Sandiford <richard@codesourcery.com>
1345 * config/m68k/m68k.h (PREFERRED_STACK_BOUNDARY): Define to 32
1346 for ColdFire targets.
1348 2007-01-19 Nathan Sidwell <nathan@codesourcery.com>
1349 Richard Sandiford <richard@codesourcery.com>
1351 * config/m68k/m68k.h (M68K_STATIC_CHAIN_REG_NAME): New macro.
1352 (INITIALIZE_TRAMPOLINE): Use STATIC_CHAIN_REGNUM.
1353 (__transfer_from_trampoline): Use M68K_STATIC_CHAIN_REG_NAME.
1354 * config/m68k/m68kelf.h (STATIC_CHAIN_REGNUM): Override.
1355 (M68K_STATIC_CHAIN_REG_NAME): Likewise.
1356 * config/m68k/netbsd-elf.h (M68K_STATIC_CHAIN_REG_NAME): Likewise.
1358 2007-01-19 Richard Sandiford <richard@codesourcery.com>
1360 * config/m68k/m68k.md (adddi_dilshr32): Rename to...
1361 (*adddi_dilshr32): ...this. Fix formatting. Remove commented-out
1362 non-canonical pattern. Restrict to !TARGET_COLDFIRE.
1363 (*adddi_dilshr32_cf): New pattern.
1364 (adddi3, subdi3): Remove first alternatives.
1366 2007-01-19 Richard Sandiford <richard@codesourcery.com>
1368 * config/m68k/m68k.c (notice_update_cc): If an SFmode move is
1369 implemented using move.l, do not use its cc result for floating-point
1372 2007-01-19 Richard Sandiford <richard@codesourcery.com>
1374 * config/m68k/m68k.h (EXTRA_CONSTRAINT): Stop the 'T' constraint
1375 from accepting 's' constraints if flag_pic.
1377 2007-01-19 Richard Sandiford <richard@codesourcery.com>
1379 * config/m68k/m68k.md (bordered, bunordered, buneq, bunge, bungt)
1380 (bunle, bunlt, bltgt, bordered_rev, bunordered_rev, buneq_rev)
1381 (bunge_rev, bungt_rev, bunle_rev, bunlt_rev, bltgt_rev): Change
1382 condition from TARGET_68881 to TARGET_HARD_FLOAT.
1384 2007-01-19 Sandra Loosemore <sandra@codesourcery.com>
1386 * longlong.h (count_leading_zeros, COUNT_LEADING_ZEROS_0): Add
1387 ColdFire alternatives.
1388 * config/m68k/m68k.h (CLZ_DEFINED_VALUE_AT_ZERO): New macro.
1389 * config/m68k/m68k.md (clzsi2): Define for ColdFire
1390 architectures that support ff1 instruction.
1392 2007-01-19 Richard Sandiford <richard@codesourcery.com>
1393 Julian Brown <julian@codesourcery.com>
1395 * config/m68k/m68k.h (CONST_OK_FOR_LETTER_P): Add an 'R' case.
1396 * config/m68k/m68k.md (*movsi_cfv4): Fold into...
1397 (*movsi_cf): ...here. Remove unnecessary 'R' from 'Rg'.
1400 2007-01-19 Richard Sandiford <richard@codesourcery.com>
1402 * config/m68k/m68k-protos.h (valid_mov3q_const): Take a HOST_WIDE_INT
1404 (output_move_const_into_data_reg, output_move_simode_const): Delete.
1405 * config/m68k/m68k.c (const_method, const_int_cost): Take a
1406 HOST_WIDE_INT instead of an rtx.
1407 (m68k_rtx_costs): Update call accordingly.
1408 (output_move_const_into_data_reg): Likewise. Fix formatting.
1409 (valid_mov3q_const): Take a HOST_WIDE_INT instead of an rtx.
1411 (output_move_simode_const): Update calls after above changes.
1412 Rework to use automatic variables and predicates like MEM_P.
1413 * config/m68k/m68k.md (pushexthisi_const): Update call to
1416 2007-01-19 Dirk Mueller <dmueller@suse.de>
1418 * tree-ssa-alias.c (perform_var_substitution): Fix typo
1421 2007-01-19 Richard Guenther <rguenther@suse.de>
1423 * builtins.c (expand_builtin_cexpi): Fall back to expanding
1424 via cexp in case sincos is not available.
1426 2007-01-19 Richard Guenther <rguenther@suse.de>
1428 * doc/tm.texi (TARGET_HAS_SINCOS): Document new target macro.
1429 * defaults.h (TARGET_HAS_SINCOS): Default to off.
1430 * config/linux.h (TARGET_HAS_SINCOS): Set to on if we have glibc.
1431 * config/alpha/linux.h (TARGET_HAS_SINCOS): Likewise.
1432 * config/sparc/linux.h (TARGET_HAS_SINCOS): Likewise.
1433 * config/sparc/linux64.h (TARGET_HAS_SINCOS): Likewise.
1434 * config/rs6000/linux.h (TARGET_HAS_SINCOS): Likewise.
1435 * config/rs6000/linux64.h (TARGET_HAS_SINCOS): Likewise.
1437 2007-01-19 Uros Bizjak <ubizjak@gmail.com>
1439 * config/i386/i386.md (*fpatanxf3_i387, fpatan_extend<mode>xf3_i387):
1441 (atan2sf3_1, atan2df3_1, atan2xf3_1): Remove insn patterns.
1442 (atan2xf3): Directly generate RTL pattern.
1443 (atan2<mode>3): Rename from atan2sf3 and atan2df3 and macroize insn
1444 patterns using X87MODEF12 mode macro. Use fpatan_extend<mode>xf3_i387
1445 and truncate result to requested mode. Use SSE_FLOAT_MODE_P to
1446 disable patterns for SSE math.
1447 (atan<mode>2): Rename from atansf2 and atandf2 and macroize insn
1448 patterns using X87MODEF12 mode macro. Use fpatan_extend<mode>xf3_i387
1449 and truncate result to requested mode. Use SSE_FLOAT_MODE_P to
1450 disable patterns for SSE math.
1452 2007-01-19 Alexandre Oliva <aoliva@redhat.com>
1454 * libgcc-std.ver: Fix typo in %inherit for GCC_4.3.0.
1456 2007-01-18 Roger Sayle <roger@eyesopen.com>
1458 * fold-const.c (fold_unary) <VIEW_CONVERT_EXPR>: Optimize away a
1459 VIEW_CONVERT_EXPR to the same type as it's operand.
1461 2007-01-18 David Edelsohn <edelsohn@gnu.org>
1463 * config/rs6000/darwin-ldouble.c: Only build _SOFT_FLOAT if
1464 configured for long double 128.
1466 2007-01-18 Mike Stump <mrs@apple.com>
1468 * config/rs6000/rs6000.c (rs6000_emit_vector_compare): Fix build
1471 2007-01-18 Michael Meissner <michael.meissner@amd.com>
1473 * i386.c (ix86_compute_frame_layout): Make fprintf's in #if 0 code
1476 2007-01-18 Jan Hubicka <jh@suse.cz>
1478 * tree-ssa-operands.c (vop_free_bucket_size): Never return value
1479 greater than NUM_VOP_FREE_BUCKETS.
1481 2007-01-18 Daniel Berlin <dberlin@dberlin.org>
1483 * tree-ssa-structalias.c: Update comments.
1484 (ptabitmap_obstack): Removed.
1486 (oldpta_obstack): Ditto.
1487 (stats): Add a few members.
1488 (struct variable_info): Remove node, complex, address_taken, and
1489 indirect_target members. Add oldsolution member.
1490 (new_var_info): Do not initialize removed members.
1491 (constraint_expr_type): Remove INCLUDES.
1492 (constraint_graph): Add size, implicit_preds, rep,
1493 indirect_cycles, eq_rep, label, direct_nodes, and complex members.
1494 (FIRST_REF_NODE): New macro.
1495 (LAST_REF_NODE): Ditto.
1496 (FIRST_ADDR_NODE): Ditto.
1497 (find): New function.
1499 (dump_constraint): Do not handle INCLUDES.
1500 (insert_into_complex): Do not insert duplicate constraints.
1501 (condense_varmap_nodes): Renamed and rewritten into ...
1502 (merge_node_constraints): This. Also fix bug in handling of
1503 offseted copy constraints.
1504 (clear_edges_for_node): No longer need to deal with preds at all,
1505 or removing associated preds/succs.
1506 (merge_graph_nodes): Deal with indirect_cycles.
1507 Don't deal with predecessors.
1508 (add_implicit_graph_edge): New function.
1509 (add_pred_graph_edge): Ditto.
1510 (add_graph_edge): Don't deal with predecessors.
1511 (build_constraint_graph): Removed.
1512 (build_pred_graph): New function.
1513 (build_succ_graph): Ditto.
1514 (struct scc_info): Removed in_component. Added roots, dfs, and
1515 node_mapping. Remove visited_index, unification_queue.
1516 (scc_visit): Deal with union-find we do now.
1517 Deal with cycles with REF nodes.
1518 (collapse_nodes): Renamed and rewritten to ...
1519 (unify_nodes): This.
1520 (process_unification_queue): Removed.
1521 (topo_visit): Cleanup
1522 (do_da_constraint): Use find.
1523 (do_sd_constraint): Ditto.
1524 (do_ds_constraint): Ditto.
1525 (do_complex_constraint): Ditto.
1526 (init_scc_info): Update for removed and added members.
1527 (find_and_collapse_graph_cycles): Renamed and rewritten into ...
1528 (find_indirect_cycles): This.
1529 (equivalence_class): New variable.
1530 (label_visit): New function.
1531 (perform_variable_substitution): Rewritten.
1532 (free_var_substitution_info): New function.
1533 (find_equivalent_node): Ditto.
1534 (move_complex_constraints): Ditto.
1535 (eliminate_indirect_cycles): Ditto.
1536 (solve_graph): Only propagate changed bits.
1537 Use indirect cycle elimination.
1539 (tree_id_t): Rename to tree_vi_t, delete id member, add vi member.
1540 (tree_id_eq): Renamed to ...
1541 (tree_vi_eq): This. Update for member change
1542 (insert_id_for_tree): Renamed and rewritten to ...
1543 (insert_vi_for_tree): This.
1544 (lookup_id_for_tree): Renamed and rewritten to ...
1545 (lookup_vi_for_tree): This.
1546 (get_id_for_tree): Renamed and rewritten to ...
1547 (get_vi_for_tree): Ditto.
1548 (get_constraint_exp_from_ssa_var): Update to use get_vi_for_tree.
1549 (process_constraint): Don't handle INCLUDES.
1550 Remove special ADDRESSOF case.
1551 (find_func_aliases): Rewrite to use vi functions instead of id
1553 (create_function_info_for): Ditto.
1554 (create_variable_info_for): Ditto.
1555 (intra_create_variable_infos): Ditto.
1556 (merge_smts_into): Ditto.
1557 (find_what_p_points_to): Ditto.
1558 (init_base_vars): Ditto.
1559 (init_alias_vars): Ditto.
1560 (remove_preds_and_fake_succs): New function.
1561 (dump_sa_points_to_info): Dump new stats.
1562 (dump_solution_for_var): Use find.
1563 (set_used_smts): Fix formatting.
1564 (compute_points_to_sets): Updated for new functions.
1565 (ipa_pta_execute): Ditto.
1567 2007-01-18 Kazu Hirata <kazu@codesourcery.com>
1568 Richard Sandiford <richard@codesourcery.com>
1570 * doc/tm.texi (TARGET_FUNCTION_VALUE): Expand documentation of
1572 * calls.c (expand_call): If the return value is a PARALLEL,
1573 extract its first member.
1574 * config/m68k/linux.h (FUNCTION_EXTRA_EPILOGUE): Remove.
1575 * config/m68k/m68k.c (m68k_output_function_epilogue): Don't
1576 use FUNCTION_EXTRA_EPILOGUE.
1577 (m68k_function_value): Return a PARALLEL if the return value
1578 is of a pointer type.
1579 * config/m68k/netbsd-elf.h (current_function_returns_pointer)
1580 (FUNCTION_EXTRA_EPILOGUE): Remove.
1581 * config/m68k/m68k.md (D0_REG): New constant.
1583 2007-01-18 Kazu Hirata <kazu@codesourcery.com>
1585 * config/m68k/m68k.c (m68k_output_function_epilogue): Don't
1586 output a NOP for empty epilogues.
1588 2007-01-18 Richard Sandiford <richard@codesourcery.com>
1590 * config/m68k/m68k.c (m68k_use_return_insn): Update comments
1591 before function. Extend register save check to include all
1592 registers, not just integer ones.
1594 2007-01-18 Kazu Hirata <kazu@codesourcery.com>
1596 * config/m68k/m68k-protos.h (use_return_insn): Rename to...
1597 (m68k_use_return_insn): ...this.
1598 * config/m68k/m68k.h (USE_RETURN_INSN): Delete.
1599 * config/m68k/m68k.c (use_return_insn): Rename to...
1600 (m68k_use_return_insn): ...this.
1601 * config/m68k/m68k.md (return): Use m68k_use_return_insn instead
1604 2007-01-18 Nathan Sidwell <nathan@codesourcery.com>
1606 * config/m68k/fpgnulib.c (__truncdfsf2): Implement round to
1607 nearest even, fix denormal rounding overflow.
1609 2007-01-18 Richard Sandiford <richard@codesourcery.com>
1611 * config/m68k/m68k.md (movsf_cf_hard): Use fsmove instead of
1612 f%$smove and f%$move.
1613 (movdf_cf_hard): Use fdmove for cases 0 and 3 and fmove for case 1.
1614 (extendsfdf2_cf): Use fdmove instead of f%&move.
1615 (truncdfsf2_cf): Use fsmove instead of f%$smove.
1616 (add<mode>3_cf, sub<mode>3_cf): Use <FP:prec> instead of <FP:round>.
1618 2007-01-18 Richard Sandiford <richard@codesourcery.com>
1620 * config/m68k/m68k.md (movdf_cf_hard): Use output_move_double for
1623 2007-01-18 Richard Sandiford <richard@codesourcery.com>
1625 * real.h (real_format): Add a canonical_nan_lsbs_set field.
1626 (coldfire_single_format): Declare.
1627 (coldfire_double_format): Likewise.
1628 * real.c (encode_ieee_single): Use canonical_nan_lsbs_set instead
1629 of qnan_msb_set to determine the lower bits of a canonical
1631 (encode_ieee_double): Likewise.
1632 (encode_ieee_quad): Likewise.
1633 (ieee_single_format): Initialize canonical_nan_lsbs_set.
1634 (mips_single_format): Likewise.
1635 (ieee_double_format): Likewise.
1636 (mips_double_format): Likewise.
1637 (ieee_extended_motorola_format): Likewise.
1638 (ieee_extended_intel_96_format): Likewise.
1639 (ieee_extended_intel_128_format): Likewise.
1640 (ieee_extended_intel_96_round_53_format): Likewise.
1641 (ibm_extended_format): Likewise.
1642 (mips_extended_format): Likewise.
1643 (ieee_quad_format): Likewise.
1644 (mips_quad_format): Likewise.
1645 (vax_f_format): Likewise.
1646 (vax_d_format): Likewise.
1647 (vax_g_format): Likewise.
1648 (i370_single_format): Likewise.
1649 (i370_double_format): Likewise.
1650 (decimal_single_format): Likewise.
1651 (decimal_double_format): Likewise.
1652 (decimal_quad_format): Likewise.
1653 (c4x_single_format): Likewise.
1654 (c4x_extended_format): Likewise.
1655 (real_internal_format): Likewise.
1656 (coldfire_single_format): New real_format.
1657 (coldfire_double_format): Likewise.
1658 * config/pdp11/pdp11.c (pdp11_f_format): Initialize
1659 canonical_nan_lsbs_set.
1660 (pdp11_d_format): Likewise.
1661 * config/m68k/m68k.c (override_options): Override REAL_FORMAT_MODE
1662 if TARGET_COLDFIRE_CPU.
1664 2007-01-18 Richard Sandiford <richard@codesourcery.com>
1666 * config/m68k/m68k-protos.h (m68k_output_pic_call): Delete.
1667 (output_call, m68k_legitimize_call_address): Declare.
1668 * config/m68k/m68k.h (EXTRA_CONSTRAINT): Remove unnecessary
1669 parenthesees. Add support for a 'W' constraint.
1670 (LEGITIMATE_PIC_OPERAND_P): Remove SYMBOL_REF_FLAG handling.
1671 (PRINT_OPERAND_PUNCT_VALID_P): Remove comment about 'o'.
1672 (m68k_symbolic_call, m68k_symbolic_jump): Declare.
1673 * config/m68k/m68k.c (m68k_symbolic_call, m68k_symbolic_jump): New
1675 (override_options): Initialize them. Do not set flag_no_function_cse
1676 for TARGET_ID_SHARED_LIBRARY.
1677 (m68k_output_pic_call): Delete.
1678 (m68k_legitimize_call_address): New function.
1679 (print_operand): Remove the %o prefix. Handle the %p prefix.
1680 (output_call): New function.
1681 (m68k_output_mi_thunk): Use m68k_symbolic_jump. Always load the
1682 target address from the GOT if symbolic jumps are not allowed.
1683 * config/m68k/m68k.md (call, general_operand): Do not set
1684 SYMBOL_REF_FLAG. Use m68k_legitimize_call_address instead.
1685 Merge separate flag_pic and !flag_pic define_insns into...
1686 (*call, *call_value): ...these new patterns. Match the address
1687 rather than the containing MEM and require it to be a call_operand.
1688 Use output_call to generate the asm template.
1689 * config/m68k/predicates.md (const_call_operand): New predicate.
1690 (call_operand): Likewise.
1692 2007-01-18 Nathan Sidwell <nathan@codesourcery.com>
1694 * config/m68k/m68k.h (REGISTER_MOVE_COST): Simplify definition.
1695 (STACK_GROWS_DOWNWARD): Define to 1.
1696 (FUNCTION_VALUE, LIBCALL_VALUE, FUNCTION_VALUE_REGNO_P): Equivocate
1697 comments, emphasizing that these values are only defaults.
1698 * config/m68k/linux.h (LINK_SPEC): Fix formatting in #undef.
1699 * config/m68k/m68k.c (const_method): Remove trailing whitespace.
1701 2007-01-18 Richard Sandiford <richard@codesourcery.com>
1703 * config/m68k/m68k.md (cmpsi): Remove outdated flag_pic handling.
1705 2007-01-18 Kazu Hirata <kazu@codesourcery.com>
1706 Richard Sandiford <richard@codesourcery.com>
1708 * config/m68k/m68k.h (DATA_REGNO_P, ADDRESS_REGNO_P, INT_REGNO_P)
1709 (FP_REGNO_P): New macros.
1710 (REGNO_OK_FOR_INDEX_P, REGNO_OK_FOR_BASE_P, REGNO_OK_FOR_DATA_P)
1711 (REGNO_OK_FOR_FP_P, REG_OK_FOR_INDEX_P, REG_OK_FOR_BASE_P): Use them.
1712 (EH_RETURN_STACKADJ_RTX): Use A0_REG.
1713 * config/m68k/m68k.c (m68k_regno_mode_ok): Use the new REGNO macros.
1715 2007-01-18 Nathan Sidwell <nathan@codesourcery.com>
1717 * config.gcc (m68k-*-aout*, m68k-*-coff*, m68020-*-elf*, m68k-*-elf*)
1718 (m68k-*-uclinuxoldabi, m68k-*-uclinux*, m68k-*-rtems*): Add t-floatlib
1720 * config/m68k/t-floatlib: New file, extracting common code from...
1721 * config/m68k/t-m68kbare, config/m68k/t-m68kelf,
1722 * config/m68k/t-uclinux: Here.
1723 * config/m68k/fpgnulib.c: Do not compile extendeed precision
1724 routines on ColdFire targets.
1726 2007-01-18 Nathan Sidwell <nathan@codesourcery.com>
1728 * config.gcc (m68k-*-aout*, m68k-*-coff*, m68020-*-elf*, m68k-*-elf*)
1729 (m68010-*-netbsdelf*, m68k*-*-netbsdelf*, m68k*-*-openbsd*)
1730 (m68k-*-uclinuxoldabi, m68k-*-uclinux*, m68k-*-linux*)
1731 (m68k-*-rtems*): Use tm_file rather than m68k/m68k.h and
1732 explicitly set MOTOROLA to 1.
1733 * config/m68k/m68k.h (MOTOROLA): Simplify definition accordingly.
1735 2007-01-18 Richard Sandiford <richard@codesourcery.com>
1736 Nathan Sidwell <nathan@codesourcery.com>
1738 * config/m68k/m68k.h (PCC_STATIC_STRUCT_RETURN): Delete.
1739 (ASM_OUTPUT_REG_PUSH, ASM_OUTPUT_REG_POP): Add MOTOROLA cases.
1740 * config/m68k/coff.h (REGISTER_PREFIX_MD): Delete.
1741 * config/m68k/m68020-elf.h (LIB_SPEC): Delete.
1742 * config/m68k/m68k-none.h (CC1_SPEC, CPP_SUBTARGET_SPEC): Delete.
1743 * config/m68k/m68kelf.h (IMMEDIATE_PREFIX, REGISTER_PREFIX_MD)
1744 (ASM_OUTPUT_REG_PUSH, ASM_OUTPUT_REG_POP): Delete.
1745 (NO_DOLLAR_IN_LABEL, PCC_STATIC_STRUCT_RETURN): Don't undefine.
1746 (BSS_ASM_OP, ASM_OUTPUT_SKIP, ASM_OUTPUT_ASCII): Delete.
1747 * config/m68k/m68kemb.h (PCC_STATIC_STRUCT_RETURN): Don't undefine.
1748 * config/m68k/linux.h (SIZE_TYPE, PTRDIFF_TYPE, WCHAR_TYPE)
1749 (WCHAR_TYPE_SIZE, TARGET_OBJFMT_CPP_BUILTINS): Delete.
1750 (TARGET_OS_CPP_BUILTINS): Don't define mc68000 and mc68020 here.
1751 (DBX_REGISTER_NUMBER): Delete.
1752 * config/m68k/netbsd-elf.h (IMMEDIATE_PREFIX): Delete.
1753 (PCC_STATIC_STRUCT_RETURN): Don't undefine.
1754 * config/m68k/openbsd.h (PCC_STATIC_STRUCT_RETURN): Define.
1756 2007-01-18 Richard Sandiford <richard@codesourcery.com>
1758 * config.gcc (m68k-*-uclinux*): Add flat.h to $tm_file.
1759 * config/flat.h: New file.
1760 * crtstuff.c (USE_PT_GNU_EH_FRAME): Don't define if
1762 * config/m68k/m68k.h (ASM_PREFERRED_EH_DATA_FORMAT): Do not use
1763 indirect references for -msep-data or -mid-shared-library.
1764 Do not use PC-relative code addresses either.
1765 * config/m68k/m68k.c (override_options): Restrict -fPIC error
1767 * config/m68k/uclinux.h (STARTFILE_SPEC): Define. Use Scrt1.o
1768 for shared libraries and crt1.o for executables. Use crti.o and
1770 (ENDFILE_SPEC): Use crtend.o and crtn.o.
1771 (LIB_SPEC): Suppress -Rlibc.gdb if -static-libc is given.
1772 Do not add -elf2flt or -shared-lib-id options here.
1773 (LINK_SPEC): Define. Pass -elf2flt if no -elf2flt option is given.
1774 Pass -shared-lib-id if -mid-shared-library, taking the library
1775 identifier from -mshared-library-id if given, otherwise
1777 (EH_FRAME_IN_DATA_SECTION): Do not undefine.
1778 (INIT_SECTION_ASM_OP, FINI_SECTION_ASM_OP): Likewise.
1779 (TARGET_OS_CPP_BUILTINS): Define __GXX_MERGED_TYPEINFO_NAMES=0
1780 and __GXX_TYPEINFO_EQUALITY_INLINE=0 if -mid-shared-library.
1781 (DRIVER_SELF_SPECS): Map unadorned PIC options to -msep-data.
1782 * config/m68k/t-uclinux (EXTRA_MULTILIB_PARTS): Add crtbegin.o
1784 * config/m68k/lb1sf68.asm (PICCALL): Use an lea and pc-relative
1785 jump sequence for ISA A and ISA A+.
1786 (PICJUMP): Likewise.
1788 2007-01-18 Richard Sandiford <richard@codesourcery.com>
1790 * config.gcc (m68k-*-uclinux*): Base the port on the common
1791 and m68k GNU/Linux files rather than on the generic ELF ones.
1792 * config/m68k/uclinux.h (TARGET_VERSION): Override.
1793 (TARGET_OS_CPP_BUILTINS): Use LINUX_TARGET_OS_CPP_BUILTINS.
1795 2007-01-18 Julian Brown <julian@codesourcery.com>
1796 Richard Sandiford <richard@codesourcery.com>
1798 * config/m68k/m68k.h (LONG_DOUBLE_TYPE_SIZE): Make 64-bit on ColdFire.
1799 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
1800 * config/m68k/netbsd-elf.h (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Undefine
1802 * config/m68k/uclinux-oldabi.h (LONG_DOUBLE_TYPE_SIZE): Redefine to
1804 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
1806 2007-01-18 Richard Sandiford <richard@codesourcery.com>
1808 * doc/install.texi: Document m68k-uclinuxoldabi.
1809 * config.gcc (m68k-*-uclinuxoldabi): New configuration.
1810 * config/m68k/uclinux-oldabi.h: New file, copied from
1811 config/m68k/uclinux.h.
1813 2007-01-18 Kaz Kojima <kkojima@rr.iij4u.or.jp>
1815 * config/m32r/m32r-protos.h (m32r_expand_epilogue): Declare it.
1816 * config/m32r/m32r.c (m32r_setup_incoming_varargs): Use gen_frame_mem.
1817 (m32r_compute_frame_size): Use unsigned for regno.
1818 (m32r_reload_lr): Use gen_frame_mem.
1820 (m32r_output_function_epilogue): Don't output the function epilogue
1822 (m32r_expand_epilogue): New.
1823 (direct_return): Return false if the function has the interrupt
1825 (m32r_hard_regno_rename_ok): Remove code for the textual epilogue.
1826 * config/m32r/m32r.md (epilogue): New expander.
1827 (return_lr, return_rte): New insns.
1828 (return): Make it expander.
1829 (return_normal): New expander.
1831 2007-01-18 Josh Conner <jconner@apple.com>
1834 * config/rs6000/rs6000.c (rs6000_emit_vector_compare): Add
1835 support for UNLE, UNLT, UNGE, and UNGT.
1837 2007-01-18 Jan Hubicka <jh@suse.cz>
1839 * tree-vrp.c (finalize_jump_threads): Do not call cleanup_cfg by hand.
1840 * tree-tailcall (add_virtual_phis): Likewise.
1841 (optimize_tail_call): Return TODOs.
1842 (execute_tail_calls): Return TODOs.
1843 * tree-ssa-ccp (execute_fold_all_builtins): Do cleanup_cfg via TODO.
1844 * tree-cfgcleanup.c (cleanup_tree_cfg_loop): Return if something
1846 * tree-ssa-forwprop.c (tree_ssa_forward_propagate_single_use_value):
1847 Cleanup cfg using TODO.
1848 * tree-flow.h (cleanup_tree_cfg_loop): Update prototype.
1849 * passes.c (execute_function_todo): When cleanup did something, remove
1851 * tree-cfg.c (pass_build_cfg): Add cleanup_cfg TODO.
1852 (make_edges): Don't cleanup_cfg.
1854 2007-01-18 Uros Bizjak <ubizjak@gmail.com>
1856 * reg-stack.c (subst_stack_regs_pat) [UNSPEC_SINCOS_COS,
1857 UNSPEC_XTRACT_FRACT]: Use generic code for instructions that
1858 operate on the top of stack.
1859 [UNSPEC_SINCOS_SIN, UNSPEC_XTRACT_EXP, UNSPEC_TAN]: Rewrite
1860 register handling of instructions that output to the second
1862 [UNSPEC_TAN_ONE, UNSPEC_TAN_TAN]: Remove.
1863 (move_for_stack_reg): Special-case check for dead destination
1864 stack slot for constant load of 1.0 inside UNSPEC_TAN.
1866 * config/i386/i386.md (UNSPEC_TAN): New constant.
1867 (UNSPEC_TAN_ONE, UNSPEC_TAN_TAN): Remove.
1868 (fptanxf4_i387, fptan_extend<mode>xf4_i387): New patterns
1869 to correctly model move of constant 1.0 to top stack slot.
1870 (*tandf3_1, *tansf3_1, *tanxf3_1): Remove insn patterns.
1871 (unnamed peephole2 pattern): Remove corresponding peephole2
1872 pattern that optimizes tan insn and loading of constant 1.0.
1873 (tanxf2): Use fptanxf4_i387.
1874 (tan<mode>2): Rename from tansf2 and tandf2 and macroize insn
1875 patterns using X87MODEF12 mode macro. Use fptan_extend<mode>xf4_i387
1876 and truncate result to requested mode. Use SSE_FLOAT_MODE_P to
1877 disable patterns for SSE math.
1878 (sincos<mode>3): Use truncxf<mode>2_i387_noop for truncation.
1879 (fyl2x_extend<mode>xf3_i387): Use X87MODEF12 for operand 1.
1881 2007-01-18 Dirk Mueller <dmueller@suse.de>
1882 Richard Guenther <rguenther@suse.de>
1885 * doc/invoke.texi (Warray-bounds): Document -Warray-bounds.
1886 * common.opt (Warray-bounds): Add new warning option.
1887 * c-opts.c (c_common_handle_option): Define -Warray-bounds
1889 * Makefile.in: make tree-vrp.o depend on toplev.h
1890 * tree-vrp.c (vrp_finalize): Call check_array_refs if -Warray-bounds
1892 (check_array_refs, check_array_bounds, check_array_ref): New.
1894 2007-01-18 Jan Hubicka <jh@suse.cz>
1896 * tree-ssa-ccp.c (ccp_finalize): Return if something changed.
1897 (execute_ssa_ccp): Return flags conditionally.
1898 * tree-ssa-propagate.c (substitue_and_fold): Return if something was
1900 * tree-ssa-propagate.h (substitute_and_fold): Update prototype.
1902 2007-01-18 Steven Bosscher <steven@gcc.gnu.org>
1904 * cfgcleanup.c (cleanup_cfg): Detect cfglayout mode and set
1905 the CLEANUP_CFGLAYOUT flag when in cfglayout mode.
1907 * Makefile.c (GTFILES): Add cfglayout.h.
1908 * gengtype.c (open_base_files): Likewise.
1909 * cfglayout.c (cfg_layout_function_footer,
1910 cfg_layout_function_header) Reindent to make gengtype happy.
1911 * cfglayout.h (cfg_layout_function_footer,
1912 cfg_layout_function_header): Add GTY(()) marker.
1914 * ifcvt.c (noce_try_sign_mask): Make sure INSN_B is non-null.
1916 2007-01-18 Ben Elliston <bje@au.ibm.com>
1918 * genautomata.c (write_automata): Include xstrerror output in the
1919 error message if writing the DFA description file fails.
1921 2007-01-17 H.J. Lu <hongjiu.lu@intel.com>
1923 * config/mips/mips-protos.h (mips_output_external): Make it
1925 * config/mips/iris.h (TARGET_ASM_EXTERNAL_LIBCALL): Removed.
1926 * config/mips/mips.c (irix_output_external_libcall): Likewise.
1927 (extern_list): Likewise.
1928 (extern_head): Likewise.
1929 (TARGET_ASM_FILE_END): Likewise.
1930 (mips_file_end): Likewise.
1931 (mips_output_external): Rewritten.
1933 2007-01-18 Ben Elliston <bje@au.ibm.com>
1935 * genpreds.c (write_insn_preds_c): Only write out the function
1936 body for regclass_for_constraint if we have register constraints.
1938 2007-01-17 Tom Tromey <tromey@redhat.com>
1940 * doc/sourcebuild.texi (libgcj Tests): Use sourceware.org.
1941 * doc/install.texi (Testing): Use sourceware.org.
1942 (Binaries): Likewise.
1943 (Specific): Likewise.
1944 * doc/contrib.texi (Contributors): Use sourceware.org.
1946 2007-01-17 Anatoly Sokolov <aesok@post.ru>
1948 * config/avr/avr.h (AVR_HAVE_LPMX): New macro.
1949 (AVR_ENHANCED): Rename to ...
1950 (AVR_HAVE_MUL): ... new.
1951 (avr_enhanced_p): Rename to ...
1952 (avr_have_mul_p): ... new.
1953 (TARGET_CPU_CPP_BUILTINS): Use 'avr_have_mul_p' instead of
1954 'avr_enhanced_p' for "__AVR_ENHANCED__". Define "__AVR_HAVE_MUL__".
1955 * config/avr/avr.c (avr_enhanced_p): Rename to ...
1956 (avr_have_mul_p): ... new.
1957 (base_arch_s): Rename 'enhanced' to 'have_mul'.
1958 (avr_override_options): Use 'avr_have_mul_p' and 'have_mul' instead of
1959 'avr_enhanced_p' and 'enhanced'.
1960 (ashlhi3_out, ashrhi3_out, lshrhi3_out, avr_rtx_costs): Use
1961 AVR_HAVE_MUL instead of AVR_ENHANCED.
1962 * avr.md (*tablejump_enh): Use AVR_HAVE_LPMX instead of AVR_ENHANCED.
1963 (mulqi3, *mulqi3_enh, *mulqi3_call, mulqihi3, umulqihi3, mulhi3,
1964 *mulhi3_enh, *mulhi3_call, mulsi3, *mulsi3_call): Use AVR_HAVE_MUL
1965 instead of AVR_ENHANCED.
1966 (*tablejump_enh): Use AVR_HAVE_LPMX instead of AVR_ENHANCED.
1967 * libgcc.S: Use __AVR_HAVE_MUL__ instead of __AVR_ENHANCED__.
1968 (__tablejump__): Use __AVR_HAVE_LPMX__ instead of __AVR_ENHANCED__.
1970 2007-01-17 Ian Lance Taylor <iant@google.com>
1972 * vec.h (VEC_reserve_exact): Define.
1973 (vec_gc_p_reserve_exact): Declare.
1974 (vec_gc_o_reserve_exact): Declare.
1975 (vec_heap_p_reserve_exact): Declare.
1976 (vec_heap_o_reserve_exact): Declare.
1977 (VEC_OP (T,A,reserve_exact)): New static inline function, three
1979 (VEC_OP (T,A,reserve)) [all versions]: Remove handling of
1981 (VEC_OP (T,A,alloc)) [all versions]: Call ...reserve_exact.
1982 (VEC_OP (T,A,copy)) [all versions]: Likewise.
1983 (VEC_OP (T,a,safe_grow)) [all versions]: Likewise.
1984 * vec.c (calculate_allocation): Add exact parameter. Change all
1986 (vec_gc_o_reserve_1): New static function, from vec_gc_o_reserve.
1987 (vec_gc_p_reserve, vec_gc_o_reserve): Call vec_gc_o_reserve_1.
1988 (vec_gc_p_reserve_exact, vec_gc_o_reserve_exact): New functions.
1989 (vec_heap_o_reserve_1): New static function, from vec_heap_o_reserve.
1990 (vec_heap_p_reserve, vec_heap_o_reserve): Call vec_heap_o_reserve_1.
1991 (vec_heap_p_reserve_exact): New function.
1992 (vec_heap_o_reserve_exact): New function.
1994 2007-01-17 Jan Hubicka <jh@suse.cz>
1996 * ipa-type-escape.c (look_for_casts): Revamp using handled_component_p.
1998 2007-01-17 Eric Christopher <echristo@apple.com>
2000 * config.gcc: Support core2 processor.
2002 2007-01-16 Jan Hubicka <jh@suse.cz>
2004 * tree-ssanames.c (release_dead_ssa_names): Instead of ggc_freeing
2005 the names, just unlink the chain so we don't crash on dangling pointers
2008 2007-01-16 Jan Hubicka <jh@suse.cz>
2010 * cgraph.h (cgraph_decide_inlining_incrementally): Kill.
2011 * tree-pass.h: Reorder to make IPA passes appear toegher.
2012 (pass_early_inline, pass_inline_parameters, pass_apply_inline): Declare.
2013 * cgraphunit.c (cgraph_finalize_function): Do not compute inling
2014 parameters, do not call early inliner.
2015 * ipa-inline.c: Update comments. Include tree-flow.h
2016 (cgraph_decide_inlining): Do not compute inlining parameters.
2017 (cgraph_decide_inlining_incrementally): Return TODOs; assume to
2018 be called with function context set up.
2019 (pass_ipa_inline): Remove unreachable functions before pass.
2020 (cgraph_early_inlining): Simplify assuming to be called from the
2022 (pass_early_inline): New pass.
2023 (cgraph_gate_ipa_early_inlining): New gate.
2024 (pass_ipa_early_inline): Turn into simple wrapper.
2025 (compute_inline_parameters): New function.
2026 (gate_inline_passes): New gate.
2027 (pass_inline_parameters): New pass.
2028 (apply_inline): Move here from tree-optimize.c
2029 (pass_apply_inline): New pass.
2030 * ipa.c (cgraph_remove_unreachable_nodes): Verify cgraph after
2032 * tree-inline.c (optimize_inline_calls): Return TODOs rather than
2034 (tree_function_versioning): Do not allocate dummy struct function.
2035 * tree-inline.h (optimize_inline_calls): Update prototype.
2036 * tree-optimize.c (execute_fixup_cfg): Export.
2037 (pass_fixup_cfg): Remove
2038 (tree_rest_of_compilation): Do not apply inlines.
2039 * tree-flow.h (execute_fixup_cfg): Declare.
2040 * Makefile.in (gt-passes.c): New.
2041 * passes.c: Include gt-passes.h
2042 (init_optimization_passes): New passes.
2043 (nnodes, order): New static vars.
2044 (do_per_function_toporder): New function.
2045 (execute_one_pass): Dump current pass here.
2046 (execute_ipa_pass_list): Don't dump current pass here.
2048 2007-01-16 Janis Johnson <janis187@us.ibm.com>
2050 * config/dfp-bit.c (dfp_compare_op): Return separate value for NaN.
2051 (DFP_NE, DFP_LE, DFP_GE): Return false for NaN.
2053 2007-01-16 David Edelsohn <edelsohn@gnu.org>
2055 * config/rs6000/darwin-ldouble.c: Build file for SOFT_FLOAT.
2056 (strong_alias): Define.
2057 (__gcc_qmul): Provide non-FMA for soft-float.
2063 (__gcc_qunord): New.
2073 * config/rs6000/rs6000.c (rs6000_init_libfuncs): Initialize
2074 soft-float functions.
2075 * config/rs6000/libgcc-ppc-glibc.ver: Version soft-float symbols.
2076 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Do not warn
2077 about long double soft float.
2079 2007-01-16 Dorit Nuzman <dorit@il.ibm.com>
2080 Tehila Meyzels <tehila@il.ibm.com>
2082 * tree-vectorizer.h (is_pattern_stmt_p): New.
2083 * tree-vect-analyze.c (vect_determine_vectorization_factor): Fix
2084 formatting (tabs instead of spaces). Cleanup and clarify setting
2085 of STMT_VINFO_VECTYPE. Call is_pattern_stmt_p.
2086 * tree-vect-transform.c (vect_get_vec_def_for_operand): Fix typo.
2087 (vectorizable_type_demotion): Check that types are integral.
2088 (vectorizable_type_promotion): Likewise.
2089 (vectorizable_store): Fix typo. Eliminate new-line at end of
2092 2007-01-16 Jan Hubicka <jh@suse.cz>
2094 * tree-ssanames.c (release_dead_ssa_names): Remove invalidated
2097 2007-01-15 Eric Christopher <echristo@apple.com>
2099 * ifcvt.c: Include vec.h, vecprim.h.
2100 (check_cond_move_block): New argument regs.
2101 Reorganize. Add registers used to regs.
2102 (cond_move_process_if_block): Use regs set above as
2105 2007-01-15 Eric Christopher <echristo@apple.com>
2107 * config/darwin.h: Update copyright.
2108 (TARGET_OPTION_TRANSLATE_TABLE): Add umbrella.
2109 (LINK_COMMAND_SPEC): Add -u.
2110 (LINK_SPEC): Fix umbrella for above.
2112 2007-01-15 Joseph S. Myers <joseph@codesourcery.com>
2114 * config/soft-fp/op-common.h, config/soft-fp/op-4.h: Update from
2117 2007-01-15 Tom Tromey <tromey@redhat.com>
2119 * doc/sourcebuild.texi (libgcj Tests): Don't mention jacks.
2120 * doc/install.texi (Testing): Don't mention jacks.
2121 (Configuration): Document --enable-java-maintainer-mode. Move
2122 --with-java-home to libgcj-specific section. Document
2124 (Prerequisites): Mention --enable-java-maintainer-mode, ecj1.
2126 2007-01-15 Jan Hubicka <jh@suse.cz>
2128 * tree-ssa-dce.c (DCE_TODOs): New.
2129 (propagate_necessity): Return if something changed.
2130 (eliminate_unnecessary_stmts): Likewise.
2131 (perform_tree_ssa_dce): Return TODO flags when needed.
2132 (pass_dce, pass_dce_loop, pass_cd_dce): Remove TODO flags.
2134 2007-01-15 Uros Bizjak <ubizjak@gmail.com>
2136 * config/i386/i386.md (fyl2xxf3_i387): Rename from fyl2x_xf3.
2137 (fyl2x_extend<mode>xf3_i387): New insn pattern.
2138 (log<mode>2): Rename from logsf2 and logdf2 and macroize insn
2139 patterns using X87MODEF12 mode macro. Extend operand 1
2140 to XFmode. Use SSE_FLOAT_MODE_P to disable patterns for SSE math.
2141 (log10<mode>2): Ditto.
2142 (log2<mode>2): Ditto.
2143 (log1p<mode>2): Ditto.
2144 (logb<mode>2): Ditto.
2145 (fyl2xp1xf3_i387): Rename from fyl2xp1_xf3.
2146 (fyl2xp1_extend<mode>xf3_i387): New insn pattern.
2147 (*fxtractxf3_i387): Rename from *fxtractxf3.
2148 (fxtract_extend<mode>xf3_i387): New insn pattern.
2149 (ilogbsi2): Use match_dup 3, not match_operand:XF 3.
2150 * config/i386/i386.c (ix86_emit_i387_log1p): Use gen_fyl2xp1xf3_i387()
2151 and gen_fyl2xxf3_i387().
2153 2007-01-14 Zdenek Dvorak <dvorakz@suse.cz>
2155 * loop-unswitch.c (unswitch_loop): Do not call fix_loop_placement.
2156 * cfgloopmanip.c (fix_loop_placement): Made static. Use
2157 get_loop_exit_edges. Changed return type to bool.
2158 * cfgloop.h (fix_loop_placement): Declaration removed.
2160 2007-01-14 Dorit Nuzman <dorit@il.ibm.com>
2162 * param.h (MIN_VECT_LOOP_BOUND): New.
2163 * params.def (MIN_VECT_LOOP_BOUND): New.
2164 * tree-vectorizer.c (slpeel_tree_peel_loop_to_edge): Takes another
2165 argument - minimum threshold for number of iterations.
2166 * tree-vectorizer.h (slpeel_tree_peel_loop_to_edge): Add another
2167 argument to declaration.
2168 * tree-vect-analyze.c (vect_analyze_operations): Check value of
2169 MIN_VECT_LOOP_BOUND.
2170 * tree-vect-transform.c (vect_do_peeling_for_loop_bound): Call
2171 slpeel_tree_peel_loop_to_edge with additional argument.
2172 (vect_do_peeling_for_alignment): Likewise.
2173 * doc/invoke.texi (min-vect-loop-bound): Document new param option.
2175 2007-01-14 Uros Bizjak <ubizjak@gmail.com>
2178 * config/i386/i386.c (print_operand) ['z']: Output 'b' for
2181 2007-01-14 Jan Hubicka <jh@suse.cz>
2183 * tree-dfa.c (remove_referenced_var): New function.
2184 * tree-ssa-live.c (remove_unused_locals): Walk referenced vars and
2185 prune referenced vars list too.
2186 * tree-flow.h (remove_referenced_var): Declare.
2188 2007-01-14 Jan Hubicka <jh@suse.cz>
2190 * tree-eh.c (add_stmt_to_eh_region_fn): Do not add call_exprs
2192 (remove_stmt_from_eh_region_fn): Do not remove call_exprs.
2193 (verify_eh_throw_stmt_node, verify_eh_throw_table_statements): Kill.
2194 * except.h (verify_eh_throw_table_statements): Kill prototype.
2195 * cfgexpand.c (expand_gimple_basic_block): Propagate Eh regions
2196 into call exrepssions.
2197 * tree-optimize.c (execute_free_cfg_annotatiosn): Do not call
2199 * tree-cfg.c: Include pointer-set.h.
2200 (verify_node_sharing): Work on pointer set.
2201 (verify_eh_throw_stmt_node): New.
2202 (verify_stmts): Use pointers sets, verify throw_stmt.
2204 2007-01-13 Zdenek Dvorak <dvorakz@suse.cz>
2206 * ipa-reference.c (analyze_function): Consider also addresses taken
2209 2007-01-12 Roger Sayle <roger@eyesopen.com>
2211 * c-typeck.c (null_pointer_constant_p): Replace use of
2212 TREE_CONSTANT_OVERFLOW with TREE_OVERFLOW.
2213 (build_c_cast): Likewise.
2215 2007-01-12 Roger Sayle <roger@eyesopen.com>
2217 * tree.h (force_fit_type_double): Remove unused final argument.
2218 * c-common.c (constant_expression_warning): Replace use of
2219 TREE_CONSTANT_OVERFLOW with TREE_OVERFLOW.
2220 (convert_and_check): Likewise.
2221 (shorten_compare): Update call to force_fit_type_double.
2222 (c_common_truthvalue_conversion) <INTEGER_CST>: Use integer_zerop.
2223 * convert.c (convert_to_pointer): Update call to
2224 force_fit_type_double.
2225 * fold-const.c (force_fit_type_double): Remove overflowed_const
2227 (int_const_binop, fold_convert_const_int_from_int,
2228 fold_convert_const_int_from_real, fold_div_compare,
2229 fold_sign_changed_comparison, fold_unary, fold_negate_const,
2230 fold_abs_const, fold_not_const): Remove the final argument from
2231 calls to force_fit_type_double.
2233 2007-01-12 Andrew Pinski <andrew_pinski@playstation.sony.com>
2235 * configure.ac: Set insn to "nop" for spu-*-* also.
2236 * configure: Regenerate.
2238 2007-01-12 Olga Golovanevsky <olga@il.ibm.com>
2240 * builtins.def : Add BUILT_IN_FREE.
2242 2007-01-12 Jan Hubicka <jh@suse.cz>
2244 PR tree-optimization/30443
2245 * tree-inline.c (tree_function_versioning): Do not optimize when
2246 cloning for inlining.
2248 2007-01-12 Zdenek Dvorak <dvorakz@suse.cz>
2250 * doc/loop.texi: Document recording of loop exits.
2251 * cfgloopmanip.c (loopify, duplicate_loop): Use alloc_loop.
2252 (update_single_exits_after_duplication,
2253 update_single_exit_for_duplicated_loop,
2254 update_single_exit_for_duplicated_loops): Removed.
2255 (duplicate_loop_to_header_edge): Do not call
2256 update_single_exits_after_duplication and
2257 update_single_exit_for_duplicated_loops.
2258 (loop_version): Do not update single_exit information.
2259 (fix_loop_structure): Use record_loop_exits instead of
2260 mark_single_exit_loops.
2261 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update
2262 the lists of loop exits.
2263 * cfghooks.c (redirect_edge_and_branch, redirect_edge_and_branch_force,
2264 split_edge, merge_blocks): Update the lists of loop exits.
2265 * modulo-sched.c (sms_schedule): Pass LOOPS_HAVE_RECORDED_EXITS to
2266 loop_optimizer_init.
2267 * loop-init.c (loop_optimizer_init): Call record_loop_exits instead
2268 of mark_single_exit_loops.
2269 (loop_optimizer_finalize): Call release_recorded_exits.
2270 * tree-ssa-loop.c (tree_loop_optimizer_init): Pass
2271 LOOPS_HAVE_RECORDED_EXITS to loop_optimizer_init.
2272 * tree-vectorizer.c (slpeel_tree_duplicate_loop_to_edge_cfg): Do not
2273 update single exit information.
2274 * lambda-code.c (perfect_nestify): Ditto.
2275 * cfgloop.c (flow_loop_free): Destroy the list of exits of the loop.
2276 (mark_single_exit_loops): Removed.
2277 (alloc_loop, loop_exit_hash, loop_exit_eq, loop_exit_free,
2278 get_exit_descriptions, rescan_loop_exit, record_loop_exits,
2279 dump_recorded_exit, dump_recorded_exits, release_recorded_exits): New
2281 (get_loop_exit_edges, single_exit): Use recorded exit lists.
2282 (add_bb_to_loop, remove_bb_from_loops): Update the lists of loop exits.
2283 (verify_loop_structure): Verify consistency of the exit lists.
2284 (flow_loops_find): Use alloc_loop. Initialize exits hash.
2285 (set_single_exit): Removed.
2286 * cfgloop.h (struct loop_exit): New function.
2287 (struct loop): single_exit_ field replaced by exits field.
2288 (LOOPS_HAVE_MARKED_SINGLE_EXITS): Replaced by LOOPS_HAVE_RECORDED_EXITS.
2289 (struct loops): Added exits hash.
2290 (mark_single_exit_loops, set_single_exit): Declaration removed.
2291 (release_recorded_exits, record_loop_exits, rescan_loop_exit): Declare.
2293 2007-01-12 Richard Sandiford <richard@codesourcery.com>
2295 * doc/invoke.texi: Avoid use of @headitem.
2297 2007-01-12 Richard Sandiford <richard@codesourcery.com>
2299 * cse.c (cse_insn): Move HAVE_CC0 code after declarations.
2301 2007-01-12 Richard Sandiford <richard@codesourcery.com>
2303 * doc/install.texi: Fix m68k-*-* anchor and add m68k-*-* to the
2306 2007-01-12 Nathan Sidwell <nathan@codesourcery.com>
2307 Richard Sandiford <richard@codesourcery.com>
2309 * doc/invoke.texi: Document -mno-bitfield, -mno-rtd and -mno-short.
2310 * config/m68k/m68k.opt: Resort options.
2311 (mbitfield, mrtd, mshort): Remove RejectNegative properties.
2313 2007-01-12 Nathan Sidwell <nathan@codesourcery.com>
2314 Richard Sandiford <richard@codesourcery.com>
2316 * doc/invoke.texi: Document the macros that are defined by
2317 m68k's -mtune and -mhard-float options.
2318 * config/m68k/m68k-protos.h (m68k_cpp_cpu_ident) Declare.
2319 (m68k_cpp_cpu_family): Likewise.
2320 * config/m68k/m68k.h (TARGET_CPU_CPP_BUILTINS): Add a full set
2321 of __ucfv*__ macros. Define __mcffpu__ if generating code for
2322 ColdFire FPUs. Define __mcf_cpu_* and __mcf_family_* macros.
2323 * config/m68k/m68k.c (m68k_cpp_cpu_ident): New function.
2324 (m68k_cpp_cpu_family): Likewise.
2326 2007-01-12 Richard Sandiford <richard@codesourcery.com>
2328 * config/m68k/m68k.h (TARGET_CPU_CPP_BUILTINS): Treat all mc68*
2329 macros besides mc68000 as tuning macros. Use a switch statement
2330 to set them and mcpu32.
2332 2007-01-12 Julian Brown <julian@codesourcery.com>
2334 * config/m68k/m68k.h: Use TARGET_68040 instead of TARGET_68040_ONLY.
2335 (TARGET_68040_ONLY): Rename to...
2336 (TARGET_68040): ...this.
2337 * config/m68k/m68k.c: Use TARGET_68040 instead of TARGET_68040_ONLY.
2338 * config/m68k/m68k.md: Likewise.
2340 2007-01-12 Julian Brown <julian@codesourcery.com>
2341 Nathan Sidwell <nathan@codesourcery.com>
2342 Richard Sandiford <richard@codesourcery.com>
2344 * config.gcc (m680[012]0-*-*, m68k*-*-*): Set m68k_cpu_ident to
2345 the -mcpu= argument associated with the --with-cpu setting.
2346 Define M68K_DEFAULT_TUNE to the default -mtune= option,
2347 if different from the one implied by the -mcpu setting.
2348 Accept --with-cpu=FOO if FOO is listed in m68k-devices.def,
2349 using mcpu=FOO as the default CPU option. Set target_cpu_default2.
2350 * doc/invoke.texi: Mention ColdFire in the introduction to the
2351 m68k options. Document the new -march, -mcpu, -mtune, -mdiv,
2352 -mno-div and -mhard-float options. Make -m68881 a synonym for
2353 -mhard-float. Document the previously-undocumented -m5206e,
2354 -m528x, -m5307 and -m5407 options. Tweak the existing option
2355 documentation for consistency.
2356 * doc/install.texi: Mention new --with-cpu arguments.
2357 * config/m68k/m68k.h (OPTION_DEFAULT_SPECS): Only use the
2358 default CPU if neither -mcpu nor -march are specified.
2359 (ASM_CPU_SPEC): Pass down -mcpu and -march options.
2360 (TARGET_CPU_CPP_BUILTINS): Set __mcfisa*__ macros from
2361 TARGET_ISA*. Set the legacy __mcf*__ cpu macros in the same way,
2362 using m68k_tune to decide between families that implement the
2363 same ISA. Use m68k_tune to set __mcfv4e__.
2364 (FL_BITFIELD, FL_68881, FL_COLDFIRE, FL_CF_HWDIV, FL_CF_MAC)
2365 (FL_CF_EMAC, FL_CF_EMAC_B, FL_CF_USP, FL_CF_FPU, FL_ISA_68000)
2366 (FL_ISA_68010, FL_ISA_68020, FL_ISA_68040, FL_ISA_A, FL_ISA_B)
2367 (FL_ISA_C, FL_ISA_MMU): New macros.
2368 (MASK_COLDFIRE): Delete.
2369 (TARGET_68010, TARGET_68020, TARGET_68040_ONLY, TARGET_COLDFIRE)
2370 (TARGET_ISAB): Redefine in terms of m68k_cpu_flags.
2371 (TARGET_68881, TARGET_COLDFIRE_FPU): Redefine in terms of m68k_fpu.
2372 (TARGET_HARD_FLOAT): Do not define here.
2373 (TARGET_ISAAPLUS, TARGET_ISAC): New macros.
2374 (TUNE_68000): New macro.
2375 (TUNE_68000_10): Redefine in terms of TUNE_68000 and TUNE_68010.
2376 (TUNE_68010, TUNE_68030, TUNE_68040, TUNE_68060, TUNE_CPU32)
2377 (TUNE_CFV2): Redefine in terms of m68k_tune.
2378 (uarch_type, target_device, fpu_type): New enums.
2379 (m68k_cpu, m68k_tune, m68k_fpu, m68k_cpu_flags): Declare.
2380 * config/m68k/m68k.c (TARGET_DEFAULT): Remove MASK_68881.
2381 (FL_FOR_isa_00, FL_FOR_isa_10, FL_FOR_isa_20, FL_FOR_isa_40)
2382 (FL_FOR_isa_cpu32, FL_FOR_isa_a, FL_FOR_isa_aplus, FL_FOR_isa_b)
2383 (FL_FOR_isa_c): New macros.
2384 (m68k_isa): New enum.
2385 (m68k_target_selection): New structure.
2386 (all_devices, all_isas, all_microarchs): New tables.
2387 (m68k_cpu_entry, m68k_arch_entry, m68k_tune_entry, m68k_cpu)
2388 (m68k_tune, m68k_fpu, m68k_cpu_flags): New variables.
2389 (MASK_ALL_CPU_BITS): Delete.
2390 (m68k_find_selection): New function.
2391 (m68k_handle_option): Handle -mcpu=, -march= and -mtune=.
2392 Map the legacy target options to a combination of the new ones.
2393 (override_options): Set m68k_cpu, m68k_tune, m68k_fpu and
2394 m68k_cpu_flags. Handle M68K_DEFAULT_TUNE. Use m68k_cpu_flags
2395 to derive default MASK_BITFIELD, MASK_CF_HWDIV and MASK_HARD_FLOAT
2397 * config/m68k/m68k.opt (m5200, m5206e, m528x, m5307, m5407, mcfv4e)
2398 (m68010, m68020, m68020-40, m68020-60, m68030, m68040): Remove Mask
2400 (m68881, msoft-float): Change mask from 68881 to HARD_FLOAT.
2401 (march=, mcpu=, mdiv, mhard-float, mtune=): New options.
2402 * config/m68k/m68k-devices.def: New file.
2404 2007-01-12 Richard Sandiford <richard@codesourcery.com>
2405 Nathan Sidwell <nathan@codesourcery.com>
2407 * config/m68k/m68k.h (ASM_CPU_SPEC, ASM_SPEC, EXTRA_SPECS)
2408 (SUBTARGET_EXTRA_SPECS): New macros.
2409 * config/m68k/linux.h (ASM_SPEC): Remove CPU flags;
2410 use %(asm_cpu_spec) instead.
2411 * config/m68k/m68k-none.h (ASM_SPEC): Likewise.
2412 * config/m68k/openbsd.h (ASM_SPEC): Likewise.
2413 * config/m68k/netbsd-elf.h (ASM_SPEC): Likewise.
2414 (EXTRA_SPECS): Rename to...
2415 (SUBTARGET_EXTRA_SPECS): ...this.
2417 2007-01-12 Nathan Sidwell <nathan@codesourcery.com>
2418 Richard Sandiford <richard@codesourcery.com>
2419 Julian Brown <julian@codesourcery.com>
2421 * config.gcc (m68k-*-aout*, m68k-*-coff*, m68020-*-elf*, m68k-*-elf*)
2422 (m68k-*-uclinux*, m68k-*-linux*, m68k-*-rtems*): Set default_m68k_cpu
2423 to the configuration's default CPU.
2424 (m68010-*-netbsdelf*, m68k*-*-netbsdelf*, m68k*-*-openbsd*): Likewise.
2425 Remove default masks.
2426 (m680[012]0-*-*): Set the default with_cpu to the first part of
2428 (m68k*-*-*): Set the default with_cpu to m$default_m68k_cpu.
2429 (m68k*-*-linux): Extend the --with-cpu handling to...
2430 (m680[012]0-*-*, m68k*-*-*): ...these configurations. Allow m68000
2431 and m68010. Don't set target_cpu_default2.
2432 * doc/install.texi: Document --with-cpu for m68k.
2433 * config/m68k/m68k.h (OPTION_DEFAULT_SPECS): Define.
2434 * config/m68k/m68k-none.h (TARGET_CPU_DEFAULT, M68K_CPU_m68k)
2435 (M68K_CPU_m68000, M68K_CPU_m68010, M68K_CPU_m68020, M68K_CPU_m68030)
2436 (M68K_CPU_m68040, M68K_CPU_m68302, M68K_CPU_m68332, TARGET_DEFAULT)
2437 (ASM_CPU_DEFAULT_SPEC, CC1_CPU_DEFAULT_SPEC): Delete.
2438 (ASM_SPEC): Remove use of %(asm_cpu_default).
2439 (EXTRA_SPECS, SUBTARGET_EXTRA_SPECS, MULTILIB_DEFAULTS): Delete.
2440 * config/m68k/linux.h (TARGET_DEFAULT): Delete.
2441 (CPP_SPEC): Merge definitions. Do not handle __HAVE_68881__ here.
2442 * config/m68k/netbsd-elf.h (TARGET_OS_CPP_BUILTINS): Define
2443 __HAVE_FPU__ if TARGET_HARD_FLOAT.
2444 (TARGET_DEFAULT): Delete.
2445 (EXTRA_SPECS): Delete cpp_cpu_default_spec, cpp_cpu_spec,
2446 cpp_fpu_spec, asm_default_spec and netbsd_cpp_spec.
2447 (CPP_CPU_SPEC): Delete.
2448 (TARGET_VERSION): Merge definitions, using TARGET_68010 to pick
2449 the appropriate string.
2450 (CPP_CPU_DEFAULT_SPEC, ASM_DEFAULT_SPEC, CPP_FPU_SPEC): Delete.
2451 (CPP_SPEC): Define to NETBSD_CPP_SPEC.
2452 (ASM_SPEC): Don't use %(asm_default_spec).
2453 * config/m68k/m68k.c (TARGET_DEFAULT_TARGET_FLAGS): Remove
2454 TARGET_DEFAULT and add MASK_68881.
2455 * config/m68k/m68k.md: Remove mention of TARGET_DEFAULT from comments.
2457 2007-01-12 Richard Sandiford <richard@codesourcery.com>
2459 * config.gcc (m68010-*-netbsdelf*): Add MASK_68010.
2460 (m68k*-*-netbsdelf*, m68k*-*-openbsd*, m68k*-linux*): Add
2461 MASK_68010 alongside MASK_68020.
2462 * doc/invoke.texi: Document -m68010.
2463 * config/m68k/m68k.opt (m68010): New.
2464 * config/m68k/m68k.h (TARGET_CPU_CPP_BUILTINS): Define mc68010
2466 (TUNE_68010): New macro.
2467 * config/m68k/m68k-none.h (M68K_CPU_m68k, M68K_CPU_m68010)
2468 (M68K_CPU_m68020, M68K_CPU_m68030, M68K_CPU_m68040)
2469 (M68K_CPU_m68332): Add MASK_68010.
2470 * config/m68k/linux.h (TARGET_DEFAULT): Add MASK_68010 to
2471 fallback definition.
2472 * config/m68k/netbsd-elf.h (CPP_CPU_SPEC): Remove now-redundant
2474 * config/m68k/m68k.c (MASK_ALL_CPU_BITS): Add MASK_68010.
2475 (m68k_handle_option): Handle OPT_m68010. Add MASK_68010
2476 to all entries that use MASK_68020.
2477 (output_move_simode_const, output_move_himode, output_move_qimode)
2478 (output_move_stricthi, output_move_strictqi): Use TARGET_68010
2479 instead of TARGET_68020 to select clr behavior. Remove comment
2480 about there being no TARGET_68010.
2481 * config/m68k/m68k.md: Likewise throughout.
2483 2007-01-12 Julian Brown <julian@codesourcery.com>
2485 * config/m68k/m68k.h (TARGET_ISAB): New macro.
2486 * config/m68k/m68k.c: Use TARGET_ISAB rather than TARGET_CFV4.
2487 * config/m68k/m68k.md: Likewise.
2489 2007-01-12 Julian Brown <julian@codesourcery.com>
2491 * config/m68k/m68k.h (LEGITIMATE_INDEX_P, LEGITIMIZE_ADDRESS): Use
2492 TARGET_COLDFIRE_FPU instead of TARGET_CFV4E.
2494 2007-01-12 Julian Brown <julian@codesourcery.com>
2496 * config/m68k/m68k.h (TUNE_68040_60): New macro.
2497 * config/m68k/m68k.c (standard_68881_constant_p): Use it.
2498 * config/m68k/m68k.md: Likewise.
2500 2007-01-12 Julian Brown <julian@codesourcery.com>
2501 Richard Sandiford <richard@codesourcery.com>
2503 * config/m68k/m68k.h (TARGET_CPU_CPP_BUILTINS): Use TUNE_68030
2504 instead of TARGET_68030, TUNE_68040 instead of TARGET_68040,
2505 TUNE_68060 instead of TARGET_68060 and TUNE_CPU32 instead of
2507 (TARGET_CPU32): Rename to...
2508 (TUNE_CPU32): ...this.
2509 (TUNE_68000_10, TUNE_68030, TUNE_68040, TUNE_68060)
2510 (TUNE_CFV2): New macros.
2511 * config/m68k/netbsd-elf.h (LONG_DOUBLE_TYPE_SIZE): Simplify;
2512 remove conditions that are implied by TARGET_68020.
2513 * config/m68k/m68k.c (m68k_output_function_prologue): Use TUNE_68040
2514 instead of TARGET_68040 and TUNE_CPU32 instead of TARGET_CPU32.
2515 (m68k_output_function_epilogue): Likewise.
2516 (m68k_rtx_costs): Likewise. Use TUNE_68060 instead of TARGET_68060
2517 and TUNE_CFV2 instead of TARGET_5200. Use TUNE_68000_10 instead of
2518 "!TARGET_68020 && !TARGET_COLDFIRE" to choose between 68000 and
2519 non-68000 timings. Refactor multiplication and division costs.
2520 (output_addsi3): Use TUNE_68040 instead of TARGET_68040 and
2521 TUNE_CPU32 instead of TARGET_CPU32.
2522 (standard_68881_constant_p): Use TUNE_68040 instead of TARGET_68040
2523 and TUNE_68060 instead of TARGET_68060.
2524 * config/m68k/m68k.md: Use TUNE_68040 instead of TARGET_68040,
2525 TUNE_68060 instead of TARGET_68060, and TUNE_CPU32 instead of
2527 (movsi_const0): Use TUNE_68000_10 rather than "!TARGET_68020
2528 && !TARGET_COLDFIRE" to choose between moveq and clr.
2529 Likewise in the unnamed movsf pattern.
2530 (ashlsi_17_24, lshrsi_17_24): Guard with TUNE_68000_10 rather than
2531 "!TARGET_68020 && !TARGET_COLDFIRE". Likewise the unnamed
2534 2007-01-12 Richard Sandiford <richard@codesourcery.com>
2536 * config/m68k/m68k.h (TARGET_CPU_CPP_BUILTINS): Increase amount
2537 of tabbing before backslashes.
2539 2007-01-11 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
2541 * pa-linux.h (ASM_OUTPUT_INTERNAL_LABEL): Undefine.
2542 * pa.h (ASM_OUTPUT_LABEL): Output colon when using GAS.
2543 (ASM_OUTPUT_INTERNAL_LABEL): Define.
2545 2007-01-11 Zdenek Dvorak <dvorakz@suse.cz>
2547 * tree-ssa-loop-ivopts.c (extract_cond_operands): Split from
2548 find_interesting_uses_cond.
2549 (find_interesting_uses_cond): Use extract_cond_operands.
2550 (rewrite_use_compare): Use extract_cond_operands and
2551 force_gimple_operand_bsi. Do not call update_stmt.
2552 (determine_use_iv_cost_condition): Use extract_cond_operands.
2553 Return cheaper of using original bound and changing the exit bound.
2555 2007-01-11 Zdenek Dvorak <dvorakz@suse.cz>
2557 PR tree-optimization/29516
2558 * tree-ssa-address.c (tree_mem_ref_addr, add_to_parts,
2559 most_expensive_mult_to_index, addr_to_parts,
2560 create_mem_ref, maybe_fold_tmr): Make the type of
2561 fields of TARGET_MEM_REF sizetype.
2562 (move_fixed_address_to_symbol, move_pointer_to_base):
2564 * tree.def (TARGET_MEM_REF): Add comment on types of
2567 2007-01-11 Joseph Myers <joseph@codesourcery.com>
2569 * c-common.c (vector_types_convertible_p): Treat opaque types as
2570 always convertible if they have the same size, but not otherwise.
2572 2007-01-11 Steven Bosscher <steven@gcc.gnu.org>
2574 * ifcvt.c (struct noce_if_info): Add comments to the fields.
2575 Remove the b_unconditional field.
2576 (noce_try_sign_mask): Do not look at b_unconditional.
2577 (noce_process_if_block): Do not use merge_if_blocks. Update
2578 the CFG here. Do not set b_unconditional.
2579 (cond_move_process_if_block): Likewise.
2580 (find_cond_trap): Likewise.
2581 (check_cond_move_block): Require simple jump insns at the end
2584 2007-01-11 Jan Hubicka <jh@suse.cz>
2586 PR tree-optimization/1046
2587 * tree-tailcall.c (suitable_for_tail_call_opt_p): Use TREE_ADDRESSABLE
2588 when alias info is not ready.
2589 (pass_tail_recursion): Do not require aliasing.
2590 * tree-ssa-copyrename.c (pass_rename_ssa_cop): Likewise.
2591 * tree-ssa-ccp.c (pass_ccp, pass_fold_builtins): Likewise.
2592 * tree-ssa-copy.c (pass_copy_prop): Likewise.
2593 * tree-ssa-forwprop.c (pass_forwprop): Likewise.
2594 * tree-ssa-dce.c (pass_dce, pass_dce_loop, pass_cd_dce): Likewise.
2595 * passes.c (init_optimization_passes): Execute rename_ssa_copies,
2596 ccp, forwprop, copy_prop, merge_phi, copy_prop, dce and tail recursion
2598 * tree-ssa-operands.c (add_virtual_operand, get_indirect_ref_operand):
2599 When aliasing is not build, mark statement as volatile.
2601 2007-01-11 Tom Tromey <tromey@redhat.com>
2603 PR preprocessor/15185, PR preprocessor/20989:
2604 * doc/cppopts.texi <-MT>: Update description of algorithm for
2605 computing default target.
2606 <-M, -MD>: Reword "basename" text.
2608 2007-01-11 Roger Sayle <roger@eyesopen.com>
2610 * builtins.c (expand_builtin_pow, expand_builtin_powi,
2611 fold_builtin_cabs, fold_builtin_sqrt, fold_builtin_trunc,
2612 fold_builtin_floor, fold_builtin_ceil, fold_builtin_round,
2613 fold_builtin_int_int_roundingfn, fold_builtin_bitop,
2614 fold_builtin_bswap, real_constp, fold_builtin_pow,
2615 fold_builtin_powi, fold_builtin_signbit, fold_builtin_copysign,
2616 do_mpfr_arg1, do_mpfr_arg2, do_mpfr_arg3, do_mpfr_sincos): Replace
2617 uses of the macro TREE_CONSTANT_OVERFLOW with TREE_OVERFLOW.
2618 * convert.c (convert_to_pointer): Likewise.
2619 * expr.c (highest_pow2_factor, expand_expr_real_1): Likewise.
2620 * fold-const.c (force_fit_type, fold_negate_expr, int_const_binop,
2621 const_binop, fold_convert_const_int_from_int,
2622 fold_convert_const_int_from_real,
2623 fold_convert_const_real_from_real, sign_bit_p,
2624 optimize_minmax_comparison, extract_muldiv_1, fold_div_compare,
2625 fold_sign_changed_comparison, fold_unary, fold_comparison,
2626 fold_binary, multiple_of_p, tree_Expr_non_zero_p,
2627 fold_negate_const, fold_abs_const, fold_not_const): Likewise.
2628 * print-tree.c (print_node_brief, print_node): Likewise.
2629 * stor-layout.c (place_field, layout_type): Likewise.
2630 * tree-chrec.c (keep_cast): Likewise.
2631 * tree.c (build_vector, build_real, build_real_from_int_cst,
2632 build_complex): Likewise.
2634 2007-01-11 Roger Sayle <roger@eyesopen.com>
2636 * tree.h (TREE_CONSTANT_OVERFLOW): Obsolete. For the time being,
2637 treat TREE_CONSTANT_OVERFLOW as a synonym of TREE_OVERFLOW.
2639 2007-01-11 Paolo Bonzini <bonzini@gnu.org>
2641 * configure.ac (strict1_warn): Rename to strict_warn.
2642 (WERROR, --enable-werror, symlink hacks, stage1_cflags,
2643 cc_set_by_configure, quoted_cc_set_by_configure,
2644 stage_prefix_set_by_configure, quoted_stage_prefix_set_by_configure,
2645 all_boot_languages, all_stagestuff): Remove.
2646 (target_list): Remove bootstrap targets.
2647 * Makefile.in (quickstrap): Unconditionally make a synonym of all.
2648 (BOOT_LANGUAGES, STAGE1_CFLAGS, STAGE1_CHECKING,
2649 REMAKEFLAGS, FLAGS_TO_PASS, PREPEND_DOTDOT_TO_RELATIVE_PATHS,
2650 SUBDIR_FLAGS_TO_PASS, WERROR_FLAGS, STRICT2_WARN, LANG_STAGESTUFF,
2651 VOL_FILES, POSTSTAGE1_FLAGS_TO_PASS, STAGE2_FLAGS_TO_PASS,
2652 STAGEPROFILE_FLAGS_TO_PASS, STAGEFEEDBACK_FLAGS_TO_PASS, stage1_build,
2653 stage1_copy, stage2_build, stage2_copy, stageprofile_build,
2654 stageprofile_copy, stage3_build, stage3_copy, stagefeedback_build,
2655 stagefeedback_copy, stage4_build, clean_s1, clean_sw, bootstrap,
2656 bootstrap-lean, bootstrap2, bootstrap2-lean, bootstrap3,
2657 bootstrap3-lean, bootstrap4, bootstrap4-lean, unstage1, unstage2,
2658 unstage3, unstage4, unstageprofile, unstagefeedback, restage, restage2,
2659 restage3, restage4, restageprofile, restagefeedback, bubbleestrap,
2660 cleanstrap, unstrap, restrap, *compare, *compare3, *compare4,
2661 *compare-lean, *compare3-lean, *compare4-lean, stage1-start, stage1,
2662 stage2-start, stage2, stage3-start, stage3, stage4-start, stage4,
2663 stageprofile-start, stageprofile, stagefeedback-start, stagefeedback,
2664 risky-stage1, risky-stage2, risky-stage3, risky-stage4): Remove.
2665 (ORDINARY_FLAGS_TO_PASS): Rename to FLAGS_TO_PASS.
2666 (STAGECOPYSTUFF, STAGEMOVESTUFF): Consolidate into MOSTLYCLEANFILES.
2667 (mostlyclean): Adjust.
2668 (clean, distclean): Don't mention bootstrap stuff.
2669 * configure: Regenerate.
2670 * ada/config-lang.in, cp/config-lang.in, forttran/config-lang.in,
2671 java/config-lang.in, objc/config-lang.in, objcp/config-lang.in,
2672 treelang/config-lang.in (stagestuff): Remove.
2673 * doc/sourcebuild.texi (stage1, stage2, stage3, stage4,
2674 stageprofile, stagefeedback, stagestuff): Remove mention.
2676 2007-01-11 Nick Clifton <nickc@redhat.com>
2678 * config/mcore/predicates.md (mcore_general_movesrc_operand):
2680 (mcore_general_movdst_operand): Do not accept CONST_INTs.
2681 (mcore_arith_K_S_operand): Run the test for the S constraint not
2682 the test for the M constraint.
2683 (mcore_addsub_operand): Do not accept integer values that are
2684 larger than 32 bits.
2685 * config/mcore/mcore.md: Remove unused constraints from split.
2686 (andsi3): Use HOST_WIDE_INT instead of int to hold an INTVAL.
2688 (allocate_stack): Likewise.
2689 * config/mcore/mcore.c (mcore_print_operand): Restrict output of P
2690 operands to 32 bits.
2691 (mcore_const_costs): Use HOST_WIDE_INT instead of int to hold an
2693 (mcore_and_cost, mcore_modify_comparison, const_ok_for_mcore,
2694 mcore_const_ok_for_inline, mcore_const_trick_uses_not,
2695 try_constant_tricks, mcore_num_ones, mcore_num_zeros,
2696 mcore_output_bclri, mcore_output_andn, output_inline_const,
2697 mcore_output_move, mcore_output_movedouble): Likewise.
2698 (mcore_output_cmov): Use CONST_OK_FOR_M and CONST_OK_FOR_N.
2699 (output_inline_const): Likewise.
2700 (output_inline_const): Fix format strings used in sprintf
2702 * config/mcore/mcore-protos.h: Update prototypes for changed
2703 functions in mcore.c.
2704 * config/mcore/mcore.h (CONST_OK_FOR_I): Cast values to
2705 HOST_WIDE_INT and not int.
2706 (CONST_OK_FOR_J, CONST_OK_FOR_K, CONST_OK_FOR_L, CONST_OK_FOR_M,
2707 CONST_OK_FOR_N): Likewise.
2708 (LEGITIMATE_CONSTANT_P): Also check CONSTANT_P.
2709 (GO_IF_LEGITIMATE_INDEX): Use HOST_WIDE_INT instead of int to hold
2712 2007-01-10 Jan Hubicka <jh@suse.cz>
2714 * tree-vrp.c (remove_range_assertions): Release defs.
2715 * tree-ssa-loop-ivopts.c (rmeove_statement): Likewise.
2716 * tree-ssa-dom.c (remove_stmt_or_phi): Likewise.
2718 2007-01-10 Paul Brook <paul@codesourcery.com>
2720 * config/arm/arm.c (arm_rtx_costs_1): Handle mutiply-subtract.
2721 * config/arm/arm.md (mulsi3subsi): New insn.
2723 2007-01-10 Zdenek Dvorak <dvorakz@suse.cz>
2725 * tree-ssa-loop-manip.c (tree_unroll_loop): Make it a wrapper over ...
2726 (tree_transform_and_unroll_loop): New.
2727 * tree-flow.h (transform_callback, tree_transform_and_unroll_loop):
2730 2007-01-10 Robert Kennedy <jimbob@google.com>
2732 * fold-const.c (fold_comparison): Fold comparisons like (x *
2733 1000 < 0) to (x < 0).
2735 2007-01-10 Ian Lance Taylor <iant@google.com>
2737 * tree-pretty-print.c (dump_generic_node): Print parentheses when
2738 operands have the same priority.
2740 2007-01-10 Tom Tromey <tromey@redhat.com>
2742 * fold-const.c (fold_truthop): Don't check can_use_bit_fields_p.
2743 (fold_binary): Likewise.
2744 * langhooks.c (lhd_can_use_bit_fields_p): Removed.
2745 * langhooks-def.h (lhd_can_use_bit_fields_p): Removed.
2746 (LANG_HOOKS_CAN_USE_BIT_FIELDS_P): Removed.
2747 (LANG_HOOKS_INITIALIZER): Remove LANG_HOOKS_CAN_USE_BIT_FIELDS_P.
2748 * langhooks.h (struct lang_hooks): Removed field
2749 'can_use_bit_fields_p'.
2751 2007-01-10 Ralf Corsépius <ralf.corsepius@rtems.org>
2753 * config/bfin/t-bfin, config/bfin/t-bfin-elf: Remove GCC_CFLAGS.
2755 2007-01-10 Razya Ladelsky <razya@il.ibm.com>
2757 * function.c (get_last_funcdef_no): New function.
2758 * function.h (get_last_funcdef_no): Declare.
2759 * tree-inline.c (initialize_cfun): Add initialization.
2760 (tree_function_versioning): Cleanup.
2762 2007-01-10 Jan Hubicka <jh@suse.cz>
2764 * tree-inline.c (setup_one_parameter): Do not propagate into abnormal
2767 2007-01-10 Sa Liu <saliu@de.ibm.com>
2768 Ben Elliston <bje@au.ibm.com>
2770 * spu.h (STACK_SAVE_AREA): Use VOIDmode for SAVE_FUNCTION, SImode
2771 for SAVE_NONLOCAL and Pmode for any other save level.
2772 * spu-protos.h (spu_restore_stack_block): Declare.
2773 * spu.md (save_stack_block): Remove.
2774 (restore_stack_block): Call spu_restore_stack_block.
2775 * spu.c (spu_restore_stack_block): New function.
2776 (spu_expand_epilogue): Remove old comment.
2778 2007-01-09 Zdenek Dvorak <dvorakz@suse.cz>
2780 PR tree-optimization/30322
2781 * tree-ssa-loop-ivopts.c (fold_affine_expr, iv_value): Removed.
2782 (cand_value_at): Return the value as aff_tree.
2783 (may_eliminate_iv): Convert the bound from aff_tree to tree.
2784 * tree-affine.c (aff_combination_add_cst, aff_combination_add_product,
2785 aff_combination_mult): New functions.
2786 (aff_combination_add): Use aff_combination_add_cst.
2787 (aff_combination_convert): Allow conversions to a wider type.
2788 (tree_to_aff_combination): Handle BIT_NOT_EXPR.
2789 * tree-affine.h (aff_combination_mult): Declare.
2791 2007-01-09 Carlos O'Donell <carlos@codesourcery.com>
2793 * doc/tm.texi: Update documentation to reflect reality of exec
2794 and start file search behaviours. Update copyright year.
2795 * doc/invoke.texi: Explain how GCC_EXEC_PREFIX is used to find
2796 header file directories.
2798 2007-01-09 Uros Bizjak <ubizjak@gmail.com>
2800 * config/i386/i386.md (*sinxf2): Rename to *sinxf2_i387.
2801 (*cosxf2): Rename to cosxf2_i387.
2802 (*sindf2, *sinsf2): Extend operand 1 to XFmode. Macroize patterns
2803 using X87MODEF12 mode macro. Rename patterns to
2804 *sin_extend<mode>xf2_i387. Use SSE_FLOAT_MODE_P to disable patterns
2806 (*cosdf2, *cossf2): Ditto.
2807 (sincosdf3, sincossf3): Ditto. Rewrite corresponding splitters
2808 to match extended input operands.
2809 (sincos<mode>3): New expander.
2810 (*sinextendsfdf2, *cosextendsfdf2, *sincosextendsfdf3): Remove
2811 insn patterns and corresponding splitters.
2813 2007-01-09 Kaz Kojima <kkojima@gcc.gnu.org>
2815 * config/sh/t-linux (TARGET_LIBGCC2_CFLAGS): Delete.
2816 (SHLIB_MAPFILES, SHLIB_LINK, SHLIB_INSTALL): Likewise.
2818 2007-01-09 Nicolas Pitre <nico@cam.org>
2821 * arm/ieee754-df.S (Lad_s): Also test the low word of X for zero.
2823 2007-01-08 Geoffrey Keating <geoffk@apple.com>
2825 * target.h (struct gcc_target): New field library_rtti_comdat.
2826 * target-def.h (TARGET_CXX_LIBRARY_RTTI_COMDAT): New.
2827 (TARGET_CXX): Add TARGET_CXX_LIBRARY_RTTI_COMDAT.
2828 * doc/tm.texi (C++ ABI): Document TARGET_CXX_LIBRARY_RTTI_COMDAT.
2829 * config/darwin.h (TARGET_CXX_LIBRARY_RTTI_COMDAT): Define.
2831 2007-01-08 Geoffrey Keating <geoffk@apple.com>
2833 * doc/invoke.texi (Optimize Options): Correct description of -O0.
2835 2007-01-08 Richard Guenther <rguenther@suse.de>
2837 * tree.h (force_fit_type_double): Export.
2838 (force_fit_type): Remove.
2839 * fold-const.c (force_fit_type_double): New function.
2840 (force_fit_type): Remove.
2841 (int_const_binop): Use it.
2842 (fold_convert_const_int_from_int): Likewise.
2843 (fold_convert_const_int_from_real): Likewise.
2844 (fold_div_compare): Likewise.
2845 (fold_sign_changed_comparison): Likewise.
2846 (fold_unary): Likewise.
2847 (fold_negate_const): Likewise.
2848 (fold_abs_const): Likewise.
2849 (fold_not_const): Likewise.
2850 * c-common.c (shorten_compare): Use force_fit_type_double.
2851 * convert.c (convert_to_pointer): Likewise.
2853 2007-01-08 Richard Guenther <rguenther@suse.de>
2855 * tree.h (build_int_cst_wide_type): Export.
2856 * tree.c (build_int_cst_wide_type): New function.
2857 (build_int_cst_wide): Fix comment.
2858 * builtins.c (fold_builtin_object_size): Use build_int_cst
2859 to build -1 or 0 of the correct type. Use fit_double_type
2860 to check for overflow.
2861 * fold-const.c (optimize_bit_field_compare): Use build_int_cst_type
2863 (decode_field_reference): Likewise.
2864 (all_ones_mask_p): Likewise.
2865 (native_interpret_int): Use build_int_cst_wide_type.
2866 (fold_binary): Use build_int_cst_type to build an all-ones
2868 * stor-layout.c (set_sizetype): Use build_int_cst_wide_type.
2870 2007-01-08 Daniel Jacobowitz <dan@codesourcery.com>
2872 * config/pa/t-pa64 (libgcc_stub.a): Use $(T).
2874 2007-01-09 Ben Elliston <bje@au.ibm.com>
2876 * genautomata.c (STATS_OPTION): New option.
2877 (stats_flag): New flag.
2878 (gen_automata_option): Handle it.
2879 (initiate_automaton_gen): Ditto.
2880 (write_automata): Output statistics only if stats_flag is
2881 set. Likewise, output time statistics only if time_flag is set.
2882 * doc/md.texi (Processor pipeline description): Document new flag.
2884 2007-01-08 Richard Guenther <rguenther@suse.de>
2886 * builtins.c (fold_builtin_int_roundingfn): Use fit_double_type.
2887 * tree.c (build_int_cst_type): Likewise.
2888 (size_in_bytes): Don't call force_fit_type on the result.
2889 (int_fits_type_p): Use fit_double_type.
2890 * fold-const.c (fit_double_type): New function.
2891 (force_fit_type): Use it.
2892 * tree.h (fit_double_type): Export.
2894 2007-01-08 Jan Hubicka <jh@suse.cz>
2896 * tree-vectorizer.c (gate_increase_alignment): Fix return type.
2897 * ipa.c (function_and_variable_visibility): Fix return type.
2899 2007-01-08 Richard Guenther <rguenther@suse.de>
2901 * tree-ssa-ccp.c (maybe_fold_offset_to_array_ref): Use type
2902 of offset to build the index.
2903 * tree-pretty-print.c (dump_generic_node): Don't build negated
2904 const just for printing.
2905 * c-pretty-print.c (pp_c_integer_constant): Likewise.
2906 * builtins.c (fold_builtin_int_roundingfn): Check if result
2907 fits the type by using force_fit_type and comparing the result.
2908 * predict.c (predict_loops): Use compare_tree_int for comparison.
2909 * tree.c (build_int_cst): Fall back to integer_type_node for
2911 (build_int_cst_wide): Assert type is non-null.
2913 2007-01-08 Roberto Costa <roberto.costa@st.com>
2915 * tree-vrp.c (extract_range_from_cond_expr): New.
2916 (extract_range_from_expr): Handle COND_EXPR nodes used as expressions.
2917 * tree-ssa-ccp.c (get_maxval_strlen): Handle COND_EXPR nodes used
2919 (fold_stmt): Bug fix, avoid infinite recursion when folding COND_EXPRs.
2920 * tree-ssa-forwprop.c (simplify_cond, forward_propagate_into_cond,
2921 tree_ssa_forward_propagate_single_use_vars): Handle COND_EXPR nodes
2922 used as expressions.
2923 * tree-object-size.c (cond_expr_object_size): New.
2924 (collect_object_sizes_for): Handle COND_EXPR nodes used as expressions.
2926 2007-01-08 Jan Hubicka <jh@suse.cz>
2928 * tree-ssa-forwprop.c (forward_propagate_into_cond,
2929 tree_ssa_forward_propagate_single_use_va): Release defs of propagated
2932 2007-01-08 Richard Guenther <rguenther@suse.de>
2934 PR tree-optimization/23603
2935 * tree-vrp.c (set_value_range_to_truthvalue): New function.
2936 (extract_range_from_binary): Fall back to truthvalue instead of
2937 varying for TRUTH_*_EXPR.
2938 (extract_range_from_comparison): Fall back to truthvalue instead of
2940 (vrp_visit_phi_node): Don't adjust new range bounds to +INF/-INF
2941 if all visited PHI values were constant.
2943 2007-01-08 Jan Hubicka <jh@suse.cz>
2945 * cgraphunit.c (cgraph_process_new_functions): Reset reachable flag.
2946 (cgraph_analyze_function): break out from ...
2947 (cgraph_finalize_compilation_unit): ... here.
2948 (cgraph_expand_function): Remove forgoten commented out line.
2949 (cgraph_optimize): Analyze functions.
2951 2007-01-08 Jan Hubicka <jh@suse.cz>
2953 * tree-pas.h (TODO_remove_function): New flag.
2954 (TODO_update*): Renumber.
2955 (pass_ipa_increase_alignment,
2956 pass_ipa_function_and_variable_visibility): New passes.
2957 * cgraphunit.c (cgraph_increase_alignment): Move to tree-vectorizer.c
2958 (cgraph_function_and_variable_visibility): Move to ipa.c
2959 (cgraph_optimize): Don't call cgraph_function_and_variable_visibility,
2960 cgraph_increase_alignment.
2961 * ipa-inline.c (cgraph_decide_inlining): Don't push timevar.
2962 (cgraph_decide_inlining_incrementally): Push TV_INTEGRATION before
2963 calling tree-inline.
2964 (cgraph_early_inlining): Do not call cgraph_remove_unreachable_nodes.
2965 (pass_ipa_inline, pass_early_ipa_inlining): Set TODO_remove_functions
2966 * tree-vectorizer.c (increase_alignment): Move here from cgraphunit.c
2967 (gate_increase_alignment): New function.
2968 (pass_ipa_increase_alignment): New pass.
2969 * ipa.c: Inline tree-pass.h and timevar.h
2970 (function_and_variable_visibility): Move here from cgraphunit.c
2971 * tree-optimize.c (pass_early_local_passes): Add TODO_remove_functions.
2972 * passes.c (init_optimization_passes): Add the two new passes.
2973 (execute_todo): Handle cgraph_remove_functions.
2975 2007-01-08 Nick Clifton <nickc@redhat.com>
2977 * config/frv/predicates.md (reg_or_0_operand): Accept
2980 2007-01-08 Ralf Corsépius <ralf.corsepius@rtems.org>
2982 * config/bfin/rtems.h, config/bfin/t-rtems: New.
2983 * config.gcc: Add bfin*-rtems*.
2985 2007-01-08 Mark Shinwell <shinwell@codesourcery.com>
2987 * c.opt: Add -flax-vector-conversions.
2988 * c-typeck.c (convert_for_assignment): Pass flag to
2989 vector_types_convertible_p to allow emission of note.
2990 (digest_init): Likewise.
2991 * c-opts.c: Handle -flax-vector-conversions.
2992 * c-common.c (flag_lax_vector_conversions): New.
2993 (vector_types_convertible_p): Unless -flax-vector conversions
2994 has been passed, disallow conversions between vectors with
2995 differing numbers of subparts and/or element types. If such
2996 a conversion is disallowed, possibly emit a note on the first
2997 occasion only to inform the user of -flax-vector-conversions.
2998 The new last argument specifies this.
2999 * c-common.h (flag_lax_vector_conversions): New.
3000 (vector_types_convertible_p): Add extra argument.
3001 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use
3002 char_type_node for V*QI type vectors.
3003 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
3004 Update to satisfy new typechecking rules.
3005 * config/rs6000/altivec.h (vec_cmple): Use vec_cmpge, for both
3007 * doc/invoke.texi (C Dialect Options): Document
3008 -flax-vector-conversions.
3010 2007-01-08 Mark Shinwell <shinwell@codesourcery.com>
3012 PR tree-optimization/29877
3013 * tree-ssa-ter.c (is_replaceable_p): Deem assignments with
3014 a register variable on the RHS to not be replaceable.
3016 2007-01-08 Chen Liqin <liqin@sunnorth.com.cn>
3017 * config/score/t-score-elf (MULTILIB_OPTIONS): Change.
3018 * config/score/predicates.md (const_uimm5, sr0_operand, const_simm12,
3019 const_simm15, const_pow2, const_npow2): Added.
3020 * config/score/misc.md (insv, extv, extzv, movmemsi,
3021 move_lbu_a/b, mov_lhu_a/b etc): Added and fix some bug.
3022 * config/score/score.c (score_address_cost, score_select_cc_mode):
3024 Change CONST_OK_FOR_LETTER_P/EXTRA_CONSTRAINT define.
3025 Update score_rtx_costs for MACRO TARGET_RTX_COSTS.
3026 Update score_print_operand.
3027 * config/score/score.h (DATA_ALIGNMENT, SELECT_CC_MODE): Added.
3028 Adjust register allocate order and update some macro define.
3029 * config/score/score-mdaux.c (mdx_unaligned_load, mdx_unsigned_store,
3030 mdx_block_move_straight, mdx_block_move_loop_head,
3031 mdx_block_move_loop_body, mdx_block_move_loop_foot, mdx_block_move_loop,
3032 mdx_block_move): Added.
3033 (mdx_movsicc, mdp_select_add_imm, mdp_select, mds_zero_extract_andi,
3034 mdp_limm): Updated and fix some bug and typo.
3035 * config/score/score.md (movqi/hi/si, add/sub/zero/ext): Updated.
3036 (movsf, movdf, doloop_end): Added.
3038 2007-01-08 Kazu Hirata <kazu@codesourcery.com>
3040 * config/arm/arm.c, config/arm/arm.h, config/arm/arm.md,
3041 config/arm/thumb2.md: Fix comment typos.
3042 * doc/extend.texi: Fix a typo.
3044 2007-01-07 Eric Christopher <echristo@apple.com>
3046 * configure.ac: Check for __stack_chk_fail for darwin.
3047 * configure: Regenerate.
3049 2007-01-07 Richard Guenther <rguenther@suse.de>
3051 * tree-vrp.c (extract_range_from_assert): CSE calls to
3052 compare_values where possible.
3053 (extract_range_from_unary_expr): Likewise.
3055 2007-01-07 Anatoly Sokolov <aesok@post.ru>
3057 * config/avr/avr-protos.h (call_insn_operand): Delete prototype.
3058 * config/avr/avr.c (call_insn_operand): Delete function.
3059 * config/avr/avr.md (*pushqi, *pushhi, *pushsi, *pushsf): Use REG_SP
3060 instead of register number. Use predicates.
3061 * config/avr/predicates.md (const0_operand, reg_or_0_operand,
3062 call_insn_operand): Add.
3064 2007-01-06 Jan Hubicka <jh@suse.cz>
3066 * tree-pass.h (pass_build_cgraph_edges): Declare.
3067 * cgraphunit.c (record_refernece): Move to cgraphbuild.c
3068 (visited_nodes): Remove.
3069 (cgraph_create_edges): Move to cgraphbuild.c; rename to
3070 build_cgrpah_edges; make visited_nodes local.
3071 (cgraph_process_new_functions): DO not call initialize_inline_failed.
3072 (record_references_in_initializer): Move to cgraphbuild.c
3073 (initialize_inline_failed, rebuild_cgraph_edges,
3074 pass_rebuild_cgraph_edges): Move to cgraphbuild.c.
3075 (verify_cgraph_node): Make visited_nodes local.
3076 (cgraph_analyze_function): Do not call cgraph_create_edges and
3077 initialize_inline_failed.
3078 (cgraph_expand_function): Do not call cgraph_lower_function;
3079 assert that function is already lowered.
3080 * Makefile.in (cgraphbuild.o): New.
3081 * passes.c (init_optimization_passes): Add pass_build_cgraph_edges
3082 at the end of lowering passes.
3084 2007-01-06 Steven Bosscher <steven@gcc.gnu.org>
3086 * ifcvt.c (cond_move_convert_if_block): New function, code
3087 factored out from...
3088 (cond_move_process_if_block): ...here. Call the new function
3089 on the THEN and ELSE blocks.
3090 (merge_if_block): Do not copy global_live_at_end, merge_blocks
3091 already takes care of this.
3093 2007-01-05 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
3096 * tree.h (TREE_OVERFLOW_P): New.
3097 * c-typeck.c (parser_build_unary_op): Warn only if result
3098 overflowed and operands did not.
3099 (parser_build_binary_op): Likewise.
3100 (convert_for_assignment): Remove redundant overflow_warning.
3101 * c-common.c (overflow_warning): Don't check or set TREE_OVERFLOW.
3103 2007-01-05 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
3105 * c-typeck.c (store_init_value): Split over two lines to follow
3106 the GNU coding style.
3108 2007-01-05 Benjamin Kosnik <bkoz@redhat.com>
3110 * c-cppbuiltin.c (c_cpp_builtins): __GXX_EXPERIMENTAL_CPP0X__ to
3111 __GXX_EXPERIMENTAL_CXX0X__.
3112 * doc/cpp.texi: Same.
3114 2007-01-05 Richard Guenther <rguenther@suse.de>
3117 * tree.c (get_narrower): Do not construct COMPONENT_REFs
3118 with mismatched types. Instead explicitly build a
3119 conversion NOP_EXPR.
3121 2007-01-05 Ian Lance Taylor <iant@google.com>
3123 * c-common.c (decl_with_nonnull_addr_p): New function.
3124 (c_common_truthvalue_conversion): Call it.
3125 * c-typeck.c (build_binary_op): Likewise.
3126 * c-common.h (decl_with_nonnull_addr_p): Declare.
3128 2007-01-05 Jakub Jelinek <jakub@redhat.com>
3131 * libgcc2.c (__divdc3): Compare c and d against 0.0 instead of
3134 2007-01-05 Joel Brobecker <brobecker@adacore.com>
3136 * doc/install.texi (Final install): Document the fact that
3137 the GNAT runtime should not be stripped.
3139 2007-01-04 Jan Hubicka <jh@suse.cz>
3141 * tree-inline.c (fold_marked_statements): Update operand caches
3142 and EH after folding
3144 2007-01-04 Ian Lance Taylor <iant@google.com>
3146 * c-common.c (check_function_nonnull): Whitespace fix.
3148 2007-01-04 Jan Hubicka <jh@suse.cz>
3150 * tree-optimize.c (execute_fixup_cfg): Correct previously mistakely
3151 comitted older version of patch.
3152 (pass_fixup_cfg): Add TODOs to verify flow and statements, dump
3153 function, celanup cfg and collect garbage.
3155 2007-01-04 Mike Stump <mrs@apple.com>
3157 * Makefile.in (mostlyclean): Don't remove libgcc anymore.
3160 2007-01-04 Eric Christopher <echristo@apple.com>
3162 * libgcc2.c (__bswapsi2): Use SItype.
3163 (__bswapdi2): Use DItype.
3164 * libgcc2.h: Update for above.
3166 2007-01-04 Paul Brook <paul@codesourcery.com>
3168 * config/arm/arm.md (arm_mulsi3, thumb_mulsi3, mulsi3_compare0,
3169 mulsi_compare0_scratch, mulsi3addsi, mulsi3addsi_compare0,
3170 mulsi3addsi_compare0_scratch, mulsidi3adddi, mulsidi3,
3171 umulsidi3, umulsidi3adddi, smulsi3_highpart,
3172 umulsi3_highpart): Make conditional on !arm_arch6.
3173 (arm_mulsi3_v6, thumb_mulsi3_v6, mulsi3_compare0_v6,
3174 mulsi_compare0_scratch_v6, mulsi3addsi_v6, mulsi3addsi_compare0_v6,
3175 mulsi3addsi_compare0_scratch_v6, mulsidi3adddi_v6, mulsidi3_v6,
3176 umulsidi3_v6, umulsidi3adddi_v6, smulsi3_highpart_v6,
3177 umulsi3_highpart_v6): New insns.
3179 2007-01-04 Roger Sayle <roger@eyesopen.com>
3181 * fold-const.c (fold_convert): When casting an expression to void,
3182 fold_ignored_result may discover a GIMPLE_MODIFY_STMT which doesn't
3183 have a type. Instead of attempting to build a NOP_EXPR, return
3184 these "special" trees directly.
3186 2007-01-04 Joseph Myers <joseph@codesourcery.com>
3188 * config/rs6000/rs6000.c (rs6000_rtx_costs): Make adjustment for
3189 MULT inside MINUS as either argument. Use rs6000_cost->dmul -
3190 rs6000_cost->fp not 0 as adjustment for outer NEG.
3192 2007-01-04 Jan Hubicka <jh@suse.cz>
3194 * cgraph.c (cgraph_release_function_body): New function.
3195 (cgraph_remove_node): Use it.
3196 * cgraph.h (cgraph_release_function_body): Declare.
3197 * cgraphunit.c (cgraph_expand_function): Use it.
3198 * ipa.c (cgraph_remove_unreahchable_nodes): Use it.
3199 * tree-ssa.c (delete_tree_ssa): Allow to be called before aliasing
3200 is initialized and while compilation of other function is running.
3201 * tree-optimize.c (execute_free_cfg_annotations): Move code to clear
3202 statement CFG annotations from here to ...
3203 * tree-cfg.c (delete_tree_cfg_annotations): ... here.
3205 2007-01-04 Zdenek Dvorak <dvorakz@suse.cz>
3207 * cfgloop.h (enum li_flags): Make the constants powers of two.
3209 2007-01-04 Jan Hubicka <jh@suse.cz>
3211 * tree-inline.c (copy_bb): Insert new statements to statements_to_fold
3213 (fold_marked_statements): New function.
3214 (optimize_inline_calls, tree_function_versioning): Fold new statements.
3215 * tree-inline.h (copy_body_data): Add statements_to_fold.
3217 2007-01-03 Daniel Jacobowitz <dan@codesourcery.com>
3219 * config.gcc: Mention libgcc/config.host.
3220 * Makefile.in: Update comments mentioning libgcc.
3221 (LIBGCC, INSTALL_LIBGCC, GCC_PARTS, mklibgcc): Delete.
3222 (all.cross, start.encap, rest.encap, rest.cross): Update
3223 dependencies for libgcc move.
3224 (libgcc.mk, LIBGCC_DEPS, libgcov.a, libgcc.a, stmp-multilib)
3225 (clean-target, clean-target-libgcc): Delete.
3226 (srcdirify, GCC_EXTRA_PARTS): New macros.
3227 (libgcc-support, libgcc.mvars): New rules.
3228 (distclean): Remove mention of mklibgcc.
3229 (install): Don't reference INSTALL_LIBGCC.
3230 (install-common): Don't reference EXTRA_PARTS.
3231 (install-libgcc, install-multilib): Delete rules.
3232 * mklibgcc.in: Delete file.
3233 * doc/configfiles.texi: Don't mention mklibgcc.
3235 * config/i386/t-darwin (SHLIB_VERPFX): Delete (moved to libgcc).
3236 * config/i386/t-darwin64 (SHLIB_VERPFX): Likewise.
3237 * config/rs6000/t-darwin (SHLIB_VERPFX): Likewise.
3238 * config/rs6000/t-ppccomm (TARGET_LIBGCC2_CFLAGS, SHLIB_MAPFILES)
3239 (mklibgcc, ldblspecs): Likewise.
3241 * config/i386/t-nwld (libgcc.def, libc.def, libpcre.def)
3242 (posixpre.def): Use $(T).
3243 (SHLIB_EXT, SHLIB_NAME, SHLIB_SLIBDIR_QUAL, SHLIB_DEF, SHLIB_MAP)
3244 (SHLIB_SRC, SHLIB_INSTALL): Delete.
3245 (SHLIB_LINK): Make dummy.
3246 * config/t-slibgcc-darwin: Delete contents except for dummy SHLIB_LINK.
3248 * config/frv/t-linux (EXTRA_MULTILIB_PARTS): Clear.
3250 * config/alpha/t-crtfm: Use $(T) in rules for EXTRA_PARTS.
3251 * config/alpha/t-vms, config/alpha/t-vms64, config/fr30/t-fr30,
3252 config/i386/t-rtems-i386, config/ia64/t-ia64, config/rs6000/t-beos,
3253 config/rs6000/t-newas, config/sparc/t-elf: Likewise.
3255 * configure.ac (all_outputs): Remove mklibgcc.
3256 * configure: Regenerated.
3258 2007-01-03 Josh Conner <jconner@apple.com>
3261 * calls.c (compute_argument_addresses): Set stack and stack_slot
3262 for partial args, too.
3263 (store_one_arg): Use locate.size.constant for the size when
3264 generating a save_area.
3266 2007-01-03 Robert Kennedy <jimbob@google.com>
3268 * tree-cfg.c (tree_merge_blocks): Release SSA_NAME phi results
3269 whose definitions are deleted due to basic block merging.
3271 2007-01-03 Paul Brook <paul@codesourcery.com>
3274 * config/arm/arm.c (output_return_instruction): Pop PC in interrupt
3276 (use_return_insn): Return 0 for Thumb interrupt functions.
3277 (print_multi_reg): Add rfe argument for IRQ returns.
3278 (arm_output_epilogue): Pop interrupt return address directly into PC.
3279 (arm_expand_prologue): Only adjust IRQ return address in Arm mode.
3281 2007-01-03 Paul Brook <paul@codesourcery.com>
3283 Merge from sourcerygxx-4_1.
3284 * config/arm/thumb2.md: New file.
3285 * config/arm/elf.h (JUMP_TABLES_IN_TEXT_SECTION): Return True for
3287 * config/arm/coff.h (JUMP_TABLES_IN_TEXT_SECTION): Ditto.
3288 * config/arm/aout.h (ASM_OUTPUT_ADDR_VEC_ELT): Add !Thumb-2 assertion.
3289 (ASM_OUTPUT_ADDR_DIFF_ELT): Output Thumb-2 jump tables.
3290 * config/arm/aof.h (ASM_OUTPUT_ADDR_DIFF_ELT): Output Thumb-2 jump
3292 (ASM_OUTPUT_ADDR_VEC_ELT): Add !Thumb-2 assertion.
3293 * config/arm/ieee754-df.S: Use macros for Thumb-2/Unified asm
3295 * config/arm/ieee754-sf.S: Ditto.
3296 * config/arm/arm.c (thumb_base_register_rtx_p): Rename...
3297 (thumb1_base_register_rtx_p): ... to this.
3298 (thumb_index_register_rtx_p): Rename...
3299 (thumb1_index_register_rtx_p): ... to this.
3300 (thumb_output_function_prologue): Rename...
3301 (thumb1_output_function_prologue): ... to this.
3302 (thumb_legitimate_address_p): Rename...
3303 (thumb1_legitimate_address_p): ... to this.
3304 (thumb_rtx_costs): Rename...
3305 (thumb1_rtx_costs): ... to this.
3306 (thumb_compute_save_reg_mask): Rename...
3307 (thumb1_compute_save_reg_mask): ... to this.
3308 (thumb_final_prescan_insn): Rename...
3309 (thumb1_final_prescan_insn): ... to this.
3310 (thumb_expand_epilogue): Rename...
3311 (thumb1_expand_epilogue): ... to this.
3312 (arm_unwind_emit_stm): Rename...
3313 (arm_unwind_emit_sequence): ... to this.
3314 (thumb2_legitimate_index_p, thumb2_legitimate_address_p,
3315 thumb1_compute_save_reg_mask, arm_dwarf_handle_frame_unspec,
3316 thumb2_index_mul_operand, output_move_vfp, arm_shift_nmem,
3317 arm_save_coproc_regs, thumb_set_frame_pointer, arm_print_condition,
3318 thumb2_final_prescan_insn, thumb2_asm_output_opcode, arm_output_shift,
3319 thumb2_output_casesi): New functions.
3320 (TARGET_DWARF_HANDLE_FRAME_UNSPEC): Define.
3321 (FL_THUMB2, FL_NOTM, FL_DIV, FL_FOR_ARCH6T2, FL_FOR_ARCH7,
3322 FL_FOR_ARCH7A, FL_FOR_ARCH7R, FL_FOR_ARCH7M, ARM_LSL_NAME,
3323 THUMB2_WORK_REGS): Define.
3324 (arm_arch_notm, arm_arch_thumb2, arm_arch_hwdiv, arm_condexec_count,
3325 arm_condexec_mask, arm_condexec_masklen)): New variables.
3326 (all_architectures): Add armv6t2, armv7, armv7a, armv7r and armv7m.
3327 (arm_override_options): Check new CPU capabilities.
3328 Set new architecture flag variables.
3329 (arm_isr_value): Handle v7m interrupt functions.
3330 (user_return_insn): Return 0 for v7m interrupt functions. Handle
3332 (const_ok_for_arm): Handle Thumb-2 constants.
3333 (arm_gen_constant): Ditto. Use movw when available.
3334 (arm_function_ok_for_sibcall): Return false for v7m interrupt
3336 (legitimize_pic_address, arm_call_tls_get_addr): Handle Thumb-2.
3337 (thumb_find_work_register, arm_load_pic_register,
3338 legitimize_tls_address, arm_address_cost, load_multiple_sequence,
3339 emit_ldm_seq, emit_stm_seq, arm_select_cc_mode, get_jump_table_size,
3340 print_multi_reg, output_mov_long_double_fpa_from_arm,
3341 output_mov_long_double_arm_from_fpa, output_mov_double_fpa_from_arm,
3342 output_mov_double_fpa_from_arm, output_move_double,
3343 arm_compute_save_reg_mask, arm_compute_save_reg0_reg12_mask,
3344 output_return_instruction, arm_output_function_prologue,
3345 arm_output_epilogue, arm_get_frame_offsets, arm_regno_class,
3346 arm_output_mi_thunk, thumb_set_return_address): Ditto.
3347 (arm_expand_prologue): Handle Thumb-2. Use arm_save_coproc_regs.
3348 (arm_coproc_mem_operand): Allow POST_INC/PRE_DEC.
3349 (arithmetic_instr, shift_op): Use arm_shift_nmem.
3350 (arm_print_operand): Use arm_print_condition. Handle '(', ')', '.',
3352 (arm_final_prescan_insn): Use extract_constrain_insn_cached.
3353 (thumb_expand_prologue): Use thumb_set_frame_pointer.
3354 (arm_file_start): Output directive for unified syntax.
3355 (arm_unwind_emit_set): Handle stack alignment instruction.
3356 * config/arm/lib1funcs.asm: Remove default for __ARM_ARCH__.
3357 Add v6t2, v7, v7a, v7r and v7m.
3358 (RETLDM): Add Thumb-2 code.
3359 (do_it, shift1, do_push, do_pop, COND, THUMB_SYNTAX): New macros.
3360 * config/arm/arm.h (TARGET_CPU_CPP_BUILTINS): Define __thumb2__.
3361 (TARGET_THUMB1, TARGET_32BIT, TARGET_THUMB2, TARGET_DSP_MULTIPLY,
3362 TARGET_INT_SIMD, TARGET_UNIFIED_ASM, ARM_FT_STACKALIGN, IS_STACKALIGN,
3363 THUMB2_TRAMPOLINE_TEMPLATE, TRAMPOLINE_ADJUST_ADDRESS,
3364 ASM_OUTPUT_OPCODE, THUMB2_GO_IF_LEGITIMATE_ADDRESS,
3365 THUMB2_LEGITIMIZE_ADDRESS, CASE_VECTOR_PC_RELATIVE,
3366 CASE_VECTOR_SHORTEN_MODE, ADDR_VEC_ALIGN, ASM_OUTPUT_CASE_END,
3367 ADJUST_INSN_LENGTH): Define.
3368 (TARGET_REALLY_IWMMXT, TARGET_IWMMXT_ABI, CONDITIONAL_REGISTER_USAGE,
3369 STATIC_CHAIN_REGNUM, HARD_REGNO_NREGS, INDEX_REG_CLASS,
3370 BASE_REG_CLASS, MODE_BASE_REG_CLASS, SMALL_REGISTER_CLASSES,
3371 PREFERRED_RELOAD_CLASS, SECONDARY_OUTPUT_RELOAD_CLASS,
3372 SECONDARY_INPUT_RELOAD_CLASS, LIBCALL_VALUE, FUNCTION_VALUE_REGNO_P,
3373 TRAMPOLINE_SIZE, INITIALIZE_TRAMPOLINE, HAVE_PRE_INCREMENT,
3374 HAVE_POST_DECREMENT, HAVE_PRE_DECREMENT, HAVE_PRE_MODIFY_DISP,
3375 HAVE_POST_MODIFY_DISP, HAVE_PRE_MODIFY_REG, HAVE_POST_MODIFY_REG,
3376 REGNO_MODE_OK_FOR_BASE_P, LEGITIMATE_CONSTANT_P,
3377 REG_MODE_OK_FOR_BASE_P, REG_OK_FOR_INDEX_P, GO_IF_LEGITIMATE_ADDRESS,
3378 LEGITIMIZE_ADDRESS, THUMB2_LEGITIMIZE_ADDRESS,
3379 GO_IF_MODE_DEPENDENT_ADDRESS, MEMORY_MOVE_COST, BRANCH_COST,