1 2007-02-04 Kazu Hirata <kazu@codesourcery.com>
3 * config/h8300/h8300.c, config/h8300/h8300.h,
4 config/h8300/h8300.md: Follow spelling conventions.
6 2007-02-03 Uros Bizjak <ubizjak@gmail.com>
9 * combine.c (try_combine): Do not substitute source operand
10 with constants wider than 2 * HOST_BITS_PER_WIDE_INT.
12 2007-02-03 Jan Hubicka <jh@suse.cz>
15 * value-prof.c (stringop_block_profile): Fix handling of size counter;
16 do not divide by zero for never executed counters.
17 (tree_find_values_to_profile): Fix counters.
18 * gcov-ui.h (GCOV_COUNTER_AVERAGE, GCOV_COUNTER_IOR): Fix comments.
20 2007-02-03 Ian Lance Taylor <iant@google.com>
22 * lower-subreg.c (simple_move_operand): New static function,
23 broken out of simple_move. Reject LABEL_REF, SYMBOL_REF, and HIGH
25 (simple_move): Call simple_move_operand.
26 (find_decomposable_subregs): Add special handling of MEMs.
27 (can_decompose_p): Rename from cannot_decompose_p. Reverse
28 meaning of return value. If we see a hard register, test whether
29 it can store a word_mode value. Change all callers.
31 2007-02-03 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
33 * pa.md (addvdi3, addvsi3, subvdi3, subvsi3, negvdi2, negvsi2): New
34 ftrapv insns and expanders.
35 (subdi3): Change define_expand operand 1 to arith11_operand, and
36 operand 2 to reg_or_0_operand. Change constraints of 64-bit insn
37 pattern to handle reg_or_0 operands. Revise 32-bit insn pattern to
38 handle 11-bit constants and reg_or_0 operands in operands 1 and 2,
42 * varasm.c (notice_global_symbol): Treat global objects as weak when
45 2007-02-03 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
47 * emit-rtl.c (dconstpi): Delete.
49 (init_emit_once): Delete dconstpi and init dconstsqrt2.
50 * real.h (dconstpi): Delete.
52 * builtins.c (fold_builtin_cabs): Use dconstsqrt2.
53 (fold_builtin_hypot): Likewise.
55 2007-02-03 Tom Tromey <tromey@redhat.com>
58 * gcc.c (cpp_unique_options): Any of -ggdb3, -gstabs3,
59 -gcoff3, -gxcoff3, -gvms3 implies -dD.
61 2007-02-03 Kazu Hirata <kazu@codesourcery.com>
63 * c-decl.c, config/avr/avr.c, config/avr/avr.h,
64 config/m68k/m68k.c, config/m68k/netbsd-elf.h,
65 config/mn10300/mn10300.c, config/pdp11/pdp11.h,
66 config/rs6000/cell.md, config/rs6000/darwin.h,
67 config/sh/sh.md, config/sh/sh4-300.md, config/spu/spu.c,
68 config/spu/spu.md, cselib.c, expr.c, haifa-sched.c, hwint.h,
69 jump.c, reload.c, sched-deps.c, sched-int.h, tree-inline.c,
70 tree-profile.c, tree-ssa-live.h, tree-vrp.c: Fix comment
71 typos. Follow spelling conventions.
72 * doc/invoke.texi: Follow spelling conventions.
74 2007-02-03 Roger Sayle <roger@eyesopen.com>
76 * simplify-rtx.c (simplify_relational_operation_1): Implement some
77 canonicalization transformations that attempt to simplify integer
78 constant comparisons to become comparisons against zero.
80 2007-02-03 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
82 * builtins.c (fold_builtin_cabs): Fold cabs(x+xi) into
84 * fold-const.c (fold_binary): Fix comment typos. Fold complex
85 (x,0)-(0,y) into (x,-y). Likewise (0,y)-(x,0) into (-x,y).
87 2007-02-02 Mike Stump <mrs@apple.com>
89 * config/darwin9.h (DARWIN_LINKER_GENERATES_ISLANDS): Add.
90 * config/rs6000/rs6000.c (DARWIN_GENERATE_ISLANDS): Add.
91 (output_call): Use DARWIN_GENERATE_ISLANDS to decide when to
92 generate a branch island.
94 2007-02-02 Bob Wilson <bob.wilson@acm.org>
96 * config/xtensa/xtensa.c (smalloffset_mem_p): Use BASE_REG_P.
97 (xtensa_legitimate_address_p): New.
98 (xtensa_legitimize_address): New.
99 (xtensa_output_addr_const_extra): New.
100 * config/xtensa/xtensa.h (REG_OK_STRICT_FLAG): Define.
102 (REG_OK_FOR_BASE_P): Use BASE_REG_P.
103 (GO_IF_LEGITIMATE_ADDRESS): Move code to xtensa_legitimate_address_p.
104 (LEGITIMIZE_ADDRESS): Move code to xtensa_legitimize_address.
105 (OUTPUT_ADDR_CONST_EXTRA): Move code to xtensa_output_addr_const_extra.
106 * config/xtensa/xtensa-protos.h (xtensa_legitimate_address_p): New.
107 (xtensa_legitimize_address): New.
108 (xtensa_output_addr_const_extra): New.
110 2007-02-02 Steve Ellcey <sje@cup.hp.com>
112 * config/ia64/ia64.c (ia64_print_operand): Fix compare strings.
114 2007-02-02 Ian Lance Taylor <iant@google.com>
116 * expmed.c (expand_divmod): Add comment.
118 2007-02-02 Kazu Hirata <kazu@codesourcery.com>
120 * emit-rtl.c (renumber_insns): Remove.
121 * flags.h: Remove the extern for flag_renumber_insns.
122 * rtl.h: Remove the prototype for renumber_insns.
123 * toplev.c (flag_renumber_insns): Remove.
125 2007-02-02 Hui-May Chang <hm.chang@apple.com>
127 Revert for x86 darwin:
128 2005-06-19 Uros Bizjak <uros@kss-loka.si>
130 * config/i386/i386.c (ix86_function_arg_regno_p): Put back the
131 code before the following patch under TARGET_MACHO.
132 (ix86_function_value_regno_p): Likewise.
134 2007-02-02 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
136 * fold-const.c (negate_expr_p, fold_negate_expr): Handle
139 2007-02-02 Maxim Kuvyrkov <mkuvyrkov@ispras.ru>
142 * config/ia64/ia64.c (ia64_speculate_insn): Restrict to memory loads to
143 general or fp registers. Add comments.
144 * config/ia64/ia64.md (reg_pred_prefix): Add comment.
146 2007-02-02 Paolo Bonzini <bonzini@gnu.org>
148 * pointer-set.c (insert_aux): Only return insertion slot.
149 (pointer_set_insert): Adjust.
150 (pointer_set_traverse, struct pointer_map_t, pointer_map_create,
151 pointer_map_destroy, pointer_map_insert, pointer_map_contains,
152 pointer_map_traverse): New.
153 * pointer-set.h (pointer_set_traverse, struct pointer_map_t,
154 pointer_map_create, pointer_map_destroy, pointer_map_insert,
155 pointer_map_contains, pointer_map_traverse): Declare.
157 2007-02-02 Jakub Jelinek <jakub@redhat.com>
160 * builtins.c (fold_builtin_sprintf): Do not attempt to optimize
161 sprintf (str, "%s"). Do not optimize sprintf (str, "nopercent", p++).
163 2007-02-02 Maxim Kuvyrkov <mkuvyrkov@ispras.ru>
165 * sched-int.h (ds_to_dk, dk_to_ds): Declare functions.
167 (struct _dep): New type.
168 (dep_t): New typedef.
169 (DEP_PRO, DEP_CON, DEP_KIND): New access macros.
170 (DEP_STATUS): New access macro. The macro with the same name was
171 renamed to DEP_LINK_STATUS.
172 (dep_init): Declare function
174 (struct _dep_link): New type.
175 (dep_link_t): New typedef.
176 (DEP_LINK_NODE, DEP_LINK_NEXT, DEP_LINK_PREV_NEXTP): New access macros.
177 (DEP_LINK_DEP, DEP_LINK_PRO, DEP_LINK_CON, DEP_LINK_KIND): New macros.
178 (DEP_LINK_STATUS): New macro.
179 (debug_dep_links): New debug function.
181 (struct _deps_list): New type.
182 (deps_list_t): New typedef.
183 (DEPS_LIST_FIRST): New access macro.
184 (FOR_EACH_DEP_LINK): New cycle macro.
185 (create_deps_list, free_deps_list, delete_deps_list): Declare
187 (deps_list_empty_p, debug_deps_list, add_back_dep_to_deps_list): Ditto.
188 (find_link_by_pro_in_deps_list, find_link_by_con_in_deps_list): Ditto.
189 (copy_deps_list_change_con): Ditto.
191 (move_dep_link): Declare function.
193 (struct _dep_node): New type.
194 (dep_node_t): New typedef.
195 (DEP_NODE_BACK, DEP_NODE_DEP, DEP_NODE_FORW): New access macros.
197 (struct haifa_insn_data.back_deps): New field to hold backward
198 dependencies of the insn.
199 (struct haifa_insn_data.depend): Rename to forw_deps. Change its type
201 (struct haifa_insn_data.resolved_deps): Rename to resolved_back_deps.
202 Change its type to deps_list_t.
203 (INSN_BACK_DEPS): New access macro to use instead of LOG_LINKS.
204 (INSN_DEPEND): Rename to INSN_FORW_DEPS.
205 (RESOLVED_DEPS): Rename to INSN_RESOLVED_BACK_DEPS.
207 (INSN_COST): Move to haifa-sched.c. Use insn_cost () instead.
209 (DEP_STATUS): Rename to DEP_LINK_STATUS. Fix typo in the comment.
211 (add_forw_dep, delete_back_forw_dep, insn_cost): Update declaration and
215 * sched-deps.c (CHECK): New macro to (en/dis)able sanity checks.
216 (ds_to_dk, dk_to_ds): New functions.
218 (init_dep_1): New static function.
219 (init_dep): New function.
220 (copy_dep): New static function.
222 (dep_link_consistent_p, attach_dep_link, add_to_deps_list): New static
224 (detach_dep_link): New static function.
225 (move_dep_link): New function.
227 (dep_links_consistent_p, dump_dep_links): New static functions.
228 (debug_dep_links): New debugging function.
230 (deps_obstack, dl_obstack, dn_obstack): New static variables.
232 (alloc_deps_list, init_deps_list): New static functions.
233 (create_deps_list): New function.
234 (clear_deps_list): New static function.
235 (free_deps_list, delete_deps_list, deps_list_empty_p): New functions.
236 (deps_list_consistent_p, dump_deps_list): New static functions.
237 (debug_deps_list): New function.
238 (add_back_dep_to_deps_list, find_link_by_pro_in_deps_list): New
240 (find_link_by_con_in_deps_list, copy_deps_list_change_con): Ditto.
242 (maybe_add_or_update_back_dep_1, add_or_update_back_dep_1): Update to
243 use new scheduler dependencies lists.
244 (add_back_dep, delete_all_dependences, fixup_sched_groups): Ditto.
245 (sched_analyze): Ditto. Initialize dependencies lists.
246 (add_forw_dep, compute_forward_dependences): Update to use new
247 scheduler dependencies lists.
249 (init_dependency_caches): Init deps_obstack.
250 (free_dependency_caches): Free deps_obstack.
252 (adjust_add_sorted_back_dep, adjust_back_add_forw_dep): Update to use
253 new scheduler dependencies lists.
254 (delete_forw_dep, add_or_update_back_forw_dep): Ditto.
255 (add_back_forw_dep, delete_back_forw_dep): Ditto.
257 * sched-rgn.c (set_spec_fed, find_conditional_protection, is_pfree):
258 Update to use new scheduler dependencies lists.
259 (is_conditionally_protected, is_prisky, add_branch_dependences): Ditto.
260 (debug_dependencies): Ditto.
261 (schedule_region): Update comments.
263 * sched-ebb.c (earliest_block_with_similiar_load): Update to use new
264 scheduler dependencies lists.
265 (schedule_ebb): Update comments.
267 * rtl.def (DEPS_LIST): Remove.
269 * lists.c (unused_deps_list): Remove.
270 (free_list): Update assertions.
272 (alloc_DEPS_LIST, free_DEPS_LIST_list, free_DEPS_LIST_node): Remove.
273 (remove_free_DEPS_LIST_elem, copy_DEPS_LIST_list): Ditto.
275 * rtl.h (free_DEPS_LIST_list, alloc_DEPS_LIST): Remove declarations.
276 (remove_free_DEPS_LIST_elem, copy_DEPS_LIST_list): Ditto.
278 * haifa-sched.c (comments): Update.
279 (insn_cost1): Remove. Inline the code into insn_cost ().
280 (insn_cost): Update to use new scheduler dependencies lists. Move
281 processing of the dependency cost to dep_cost ().
282 (dep_cost): New function. Use it instead of insn_cost () when
283 evaluating cost of the dependency. Use compatible interface to
284 interact with the target.
285 (priority): Update to use new scheduler dependencies lists.
286 (rank_for_schedule): Ditto. Optimize heuristic that prefers the insn
287 with greater number of insns that depend on the insn.
288 (schedule_insn): Update to use new scheduler dependencies lists. Add
289 code to free backward dependencies lists. Inline and optimize code
290 from resolve_dep () - see PR28071.
291 (ok_for_early_queue_removal): Update to use new scheduler dependencies
292 lists. Update call to targetm.sched.is_costly_dependence hook.
294 (fix_inter_tick, try_ready, fix_tick_ready): Update to use new
295 scheduler dependencies lists.
297 (resolve_dep): Remove. Move the logic to schedule_insn ().
298 (init_h_i_d): Initialize dependencies lists.
300 (process_insn_depend_be_in_spec): Rename to
301 process_insn_forw_deps_be_in_spec. Update to use new scheduler
303 (add_to_speculative_block, create_check_block_twin, fix_recovery_deps):
304 Update to use new scheduler dependencies lists.
305 (clear_priorities, calc_priorities, add_jump_dependencies): Ditto.
307 * ddg.c (create_ddg_dependence, create_ddg_dep_no_link): Update to use
308 new scheduler dependencies lists.
309 (build_intra_loop_deps): Ditto.
311 * target.h (struct _dep): Declare to use in
312 gcc_target.sched.is_costly_dependence.
313 (struct gcc_target.sched.adjust_cost): Fix typo.
314 (struct gcc_target.sched.is_costly_dependence): Change signature to use
315 single dep_t parameter instead of an equivalent triad.
316 (struct gcc_target.sched.adjust_cost_2): Remove.
318 * target-def.h (TARGET_SCHED_ADJUST_COST_2): Remove.
320 * reg-notes.def (DEP_TRUE, DEP_OUTPUT, DEP_ANTI): Update comments.
322 * doc/tm.texi (TARGET_SCHED_IS_COSTLY_DEPENDENCE): Update
324 (TARGET_SCHED_ADJUST_COST_2): Remove documentation.
326 * doc/rtl.texi (LOG_LINKS): Remove part about instruction scheduler.
327 (REG_DEP_TRUE): Document.
329 * config/ia64/ia64.c (ia64_adjust_cost_2): Rename to ia64_adjust_cost.
330 Change signature to correspond to the targetm.sched.adjust_cost hook.
331 Update use in TARGET_SCHED_ADJUST_COST_2.
332 (TARGET_SCHED_ADJUST_COST_2): Rename to TARGET_SCHED_ADJUST_COST.
333 (ia64_dependencies_evaluation_hook, ia64_dfa_new_cycle): Update to use
334 new scheduler dependencies lists.
335 (ia64_gen_check): Ditto.
337 * config/mips/mips.c (vr4130_swap_insns_p): Update to use new scheduler
340 * config/rs6000/rs6000.c (rs6000_is_costly_dependence): Change
341 signature to correspond to the targetm.sched.is_costly_dependence hook.
342 (is_costly_group): Update to use new scheduler dependencies lists.
344 * config/spu/spu.c (spu_sched_adjust_cost): Use insn_cost () function
345 instead of INSN_COST () macro.
347 2007-02-01 Ian Lance Taylor <iant@google.com>
349 * lower-subreg.c (resolve_clobber): Handle a subreg of a concatn.
351 2007-02-01 Guy Martin <gmsoft@gentoo.org>
353 * pa.md (tp_load): Correct mfctl instruction syntax.
355 2007-02-01 Geoffrey Keating <geoffk@apple.com>
357 * config/rs6000/rs6000.c (rs6000_stack_info): Correct
358 altivec_padding_size calculation on AIX. Improve comment, add
359 assert to verify that it's right.
361 * config/rs6000/darwin.h (MD_UNWIND_SUPPORT): Don't define for
364 2007-2-01 Seongbae Park <seongbae.park@gmail.com>
367 * global.c (global_alloc): Add mising initialization of
370 2007-02-01 Roger Sayle <roger@eyesopen.com>
372 * alias.c (init_alias_analysis): Correct whitespace.
373 * bb-reorder.c (fix_edges_for_rarely_executed_code,
374 partition_hot_cold_basic_blocks): Likewise.
375 * builtins.c (expand_builtin_printf, expand_builtin_fprintf,
376 expand_builtin_sprintf, fold_builtin_carg, fold_builtin_sprintf,
377 maybe_emit_sprintf_chk_warning, fold_builtin_sprintf_chk,
378 fold_builtin_snprintf_chk, fold_builtin_printf,
379 fold_builtin_fprintf, do_mpfr_ckconv, do_mpfr_arg1, do_mpfr_arg2,
380 do_mpfr_arg3, do_mpfr_sincos): Likewise.
381 * cfgcleanup.c (cleanup_cfg): Likewise.
382 * cfgexpand.c (tree_expand_cfg): Likewise.
383 * fold-const.c (fold_binary) <RDIV_EXPR>: Likewise.
384 * function.c (get_next_funcdef_no): Likewise.
385 * gengtype.c (main): Likewise.
386 * genmodes.c (main): Likewise.
387 * gcse.c (bypass_conditional_jumps, print_ldst_list): Likewise.
388 * haifa-sched.c (schedule_block, extend_h_i_d): Likewise.
389 * ifcvt.c (noce_emit_move_insn): Likewise.
390 * modulo-sched.c (generate_prolog_epilog, sms_schedule_by_order):
392 * stor-layout.c (get_best_mode): Likewise.
393 * tree-ssa-loop-niter.c (get_val_for): Likewise.
394 * tree-ssa-structalias.c (get_varinfo, get_varinfo_fc,
395 scc_visit, do_ds_constraint, do_complex_constraint, label_visit,
396 perform_var_substitution, solve_graph): Likewise.
397 * tree-vrp.c (vrp_finalize): Likewise.
399 2007-02-01 Ian Lance Taylor <iant@google.com>
401 * lower-subreg.c (simplify_gen_subreg_concatn): If we ask for the
402 high part of a paradoxical subreg, return a constant zero.
404 2007-02-01 Zdenek Dvorak <dvorakz@suse.cz>
406 * toplev.c (lang_dependent_init): Call init_set_costs.
407 * loop-init.c (loop_optimizer_init): Do not call init_set_costs.
409 2007-02-01 Richard Guenther <rguenther@suse.de>
412 * fold-const.c (fold_negate_expr): Allow negating a
413 constant if overflow does not change.
415 2007-02-01 Ralf Wildenhues <Ralf.Wildenhues@gmx.de>
417 * doc/c-tree.texi (Expression trees): Improve markup.
418 * doc/tm.texi (Register Classes, Addressing Modes)
419 (Floating Point): Fix spacing after abbreviations. Fix some
422 2007-02-01 Ben Elliston <bje@au.ibm.com>
424 * doc/invoke.texi: Replace "bugfix" with "bug fix" throughout.
425 * doc/contrib.texi: Likewise.
426 * doc/install.texi: Likewise.
428 2007-01-31 Richard Henderson <rth@redhat.com>
429 Ian Lance Taylor <iant@google.com>
431 * lower-subreg.c: New file.
432 * rtl.def (CONCATN): Define.
433 * passes.c (init_optimization_passes): Add pass_lower_subreg and
435 * emit-rtl.c (update_reg_offset): New static function, broken out
436 of gen_rtx_REG_offset.
437 (gen_rtx_REG_offset): Call update_reg_offset.
438 (gen_reg_rtx_offset): New function.
439 * regclass.c: Revert patch of 2006-03-05, restoring
441 (clear_reg_info_regno): New function.
442 * dwarf2out.c (concatn_loc_descriptor): New static function.
443 (loc_descriptor): Handle CONCATN.
444 * common.opt (fsplit_wide_types): New option.
445 * opts.c (decode_options): Set flag_split_wide_types when
447 * timevar.def (TV_LOWER_SUBREG): Define.
448 * rtl.h (gen_reg_rtx_offset): Declare.
449 (reg_scan_update): Declare.
450 * regs.h (clear_reg_info_regno): Declare.
451 * tree-pass.h (pass_lower_subreg): Declare.
452 (pass_lower_subreg2): Declare.
453 * doc/invoke.texi (Option Summary): List -fno-split-wide-types.
454 (Optimize Options): Add -fsplit-wide-types to -O1 list. Document
456 * doc/rtl.texi (Regs and Memory): Document concat and concatn.
457 * Makefile.in (OBJS-common): Add lower-subreg.o.
458 (lower-subreg.o): New target.
460 2007-01-31 Kazu Hirata <kazu@codesourcery.com>
462 * config/sh/sh.h (HAVE_SECONDARY_RELOADS): Remove.
464 2007-01-31 Anatoly Sokolov <aesok@post.ru>
467 * config/avr/avr.c (DWARF2_ADDR_SIZE): Define.
469 2007-01-31 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
472 * builtins.c (fold_builtin_sqrt): Use MPFR for constant args.
474 2007-01-31 Zdenek Dvorak <dvorakz@suse.cz>
476 * cfgloop.h: Include vec-prim.h.
477 (enum li_flags): Remove LI_ONLY_OLD.
478 (loop_iterator): Changed.
479 (fel_next, fel_init): Iterate over loop tree.
480 (FOR_EACH_LOOP_BREAK): New macro.
481 * loop-unswitch.c (unswitch_loops): Do not pass LI_ONLY_OLD to
483 * tree-ssa-loop-unswitch.c (tree_ssa_unswitch_loops): Ditto.
484 * modulo-sched.c (sms_schedule): Ditto.
485 * tree-vectorizer.c (vectorize_loops): Ditto.
486 * doc/loop.texi: Update information on loop numbering and behavior of
487 FOR_EACH_LOOP wrto new loops.
488 * tree-scalar-evolution.c (compute_overall_effect_of_inner_loop,
489 add_to_evolution_1): Test nestedness of loops instead of comparing
491 * tree-chrec.c (chrec_fold_plus_poly_poly,
492 chrec_fold_multiply_poly_poly, chrec_evaluate,
493 hide_evolution_in_other_loops_than_loop, chrec_component_in_loop_num,
494 reset_evolution_in_loop): Ditto.
495 * Makefile.in (CFGLOOP_H): Add vecprim.h dependency.
497 2007-01-31 Dirk Mueller <dmueller@suse.de>
499 * c-common.c (warn_about_parentheses): Separate warning about
500 un-parenthized sequence of comparison operators from the one
501 which is supposed to warn about x <= y <= z.
503 2007-01-31 Uros Bizjak <ubizjak@gmail.com>
505 * optabs.h (enum optab_index): Add new OTI_isinf.
506 (isinf_optab): Define corresponding macro.
507 * optabs.c (init_optabs): Initialize isinf_optab.
508 * genopinit.c (optabs): Implement isinf_optab using isinf?f2
510 * builtins.c (mathfn_built_in): Handle BUILT_IN_ISINF{,F,L}.
511 (expand_builtin_interclass_mathfn): Expand BUILT_IN_ISINF{,F,L}
513 (expand_builtin): Expand BUILT_IN_ISINF{,F,L} using
514 expand_builtin_interclass_mathfn.
515 * reg_stack.c (subst_stack_regs_pat): Handle UNSPEC_FXAM.
516 * config/i386/i386.md (UNSPEC_FXAM): New constant.
517 (fxam<mode>2_i387): New insn pattern.
518 (isinf<mode>2) New expander to implement isinf, isinff and isinfl
519 built-in functions as x87 inline asm.
521 2007-01-31 Kazu Hirata <kazu@codesourcery.com>
523 * gcc/config/arm/unwind-arm.h (_sleb128_t, _uleb128_t): New.
525 2007-01-30 Eric Christopher <echristo@apple.com>
527 * config.gcc: Add geode.
529 2007-01-31 Kazu Hirata <kazu@codesourcery.com>
531 * cgraphunit.c, config/arm/arm.c, config/m68k/m68k.c,
532 ipa-inline.c, tree-profile.c, tree-ssa-live.c,
533 tree-ssa-math-opts.c, tree-ssanames.c, tree-vect-analyze.c,
534 value-prof.c: Fix comment typos.
536 2007-01-30 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
539 * doc/invoke.texi (Wpointer-arith): Document warning.
541 2007-01-30 Janis Johnson
543 * doc/extend.texi (Decimal Floating Types): Remove decfloat.h from
544 the list of discrepancies from the draft TR.
546 2007-01-30 Dirk Mueller <dmueller@suse.de>
549 * doc/invoke.texi (-Wreturn-type): Update description to
552 2007-01-30 Richard Sandiford <richard@codesourcery.com>
554 * cfgrtl.c (try_redirect_by_replacing_jump): Check only_sets_cc0_p.
556 2007-01-30 Uros Bizjak <ubizjak@gmail.com>
558 * builtins.c (expand_builtin_int_interclass_roundingfn): New function
559 to handle optabs that operate on floating point input argument and
560 output to integer output.
561 (expand_builtin_mathfn) [BUILT_IN_ILOGB]: Move from here ...
562 (expand_builtin_interclass_mathfn) [BUILT_IN_ILOGB]: ... to here.
563 (expand_builtin): Expand BUILT_IN_ILOGB{,F,L} using
564 expand_builtin_interclass_mathfn ().
565 * config/i386/i386.md (fxtractxf3_i387): Rename from *fxtractxf3_i387.
567 (ilogbxf2, ilogb<mode>2): New expanders to implement ilogb, ilogbf and
568 ilogbl built-in functions as x87 intrinsics.
570 2007-01-30 Richard Guenther <rguenther@suse.de>
573 * dwarf2out.c (reference_to_unused): Query varpool if the
576 2007-01-30 Richard Guenther <rguenther@suse.de>
579 * passes.c (execute_one_pass): Reset in_gimple_form to not
580 confuse non-unit-at-a-time mode.
582 2007-01-29 Roger Sayle <roger@eyesopen.com>
583 Richard Guenther <rguenther@suse.de>
585 * fold-const.c (round_up): Make HIGH an unsigned HOST_WIDE_INT to
586 avoid undefined behaviour on overflow. Use force_fit_type_double
587 to construct the constant with the specified TREE_OVERFLOW.
589 2007-01-29 Janis Johnson <janis187@us.ibm.com>
591 * config/dfp-bit.c: Add parameterized support for fp exceptions.
592 * config/dfp-bit.h: Ditto.
594 2007-01-29 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
596 * c-decl.c (pop_scope): Replace warnings with call to
597 warn_for_unused_label.
598 * c-common.h (warn_for_unused_label): Declare.
599 * c-common.c (warn_for_unused_label): Define.
601 2007-01-29 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
603 * tree-optimize.c (update_inlined_to_pointers): Delete unused
606 2007-01-29 Janis Johnson <janis187@us.ibm.com>
608 * Makefile.in (USER_H): Remove decfloat.h.
609 * ginclude/decfloat.h: Delete, moving contents to ...
610 * ginclude/float.h: Add support for decimal floating point,
611 guarded by __STDC_WANT_DEC_FP__.
613 2007-01-29 Mike Stump <mrs@apple.com>
615 * doc/gccint.texi (Top): Rename Loop Representation to Loop
616 Analysis and Representation to resolve case insensitive conflict.
617 * doc/loop.texi (Loop Analysis and Representation): Likewise.
619 2007-01-28 Daniel Berlin <dberlin@dberlin.org>
621 * tree.h (struct tree_memory_tag): Add aliases member.
622 (MTAG_ALIASES): New macro.
623 * tree-ssa-alias.c (alias_bitmap_obstack): New variable.
624 (add_may_alias): Remove pointer-set. Update for may_aliases being
626 (mark_aliases_call_clobbered): Update for may_aliases being a
628 (compute_tag_properties): Ditto.
629 (create_partition_for): Ditto.
630 (compute_memory_partitions): Ditto.
631 (dump_may_aliases_for): Ditto.
632 (is_aliased_with): Ditto.
633 (add_may_alias_for_new_tag): Ditto.
634 (rewrite_alias_set_for): Rewrite for may_aliases being a bitmap.
635 (compute_is_aliased): New function.
636 (compute_may_aliases): Call compute_is_aliased).
637 (init_alias_info): Initialize alias_bitmap_obstack.
638 (union_alias_set_into): New function.
639 (compute_flow_sensitive_aliasing): Use union_aliases_into.
640 (have_common_aliases_p): Rewrite to take two bitmaps and use
642 (compute_flow_insensitive_aliasing): Stop using pointer-sets.
644 (finalize_ref_all_pointers): Update for add_may_alias changes.
645 (new_type_alias): Ditto.
646 * tree-flow-inline.h (may_aliases): Return a bitmap.
647 * tree-dfa.c (dump_variable): Check for MTAG_P'ness.
648 * tree-ssa.c (verify_flow_insensitive_alias_info): Update for
649 may_aliases being a bitmap.
650 * tree-flow.h (struct var_ann_d): Remove may_aliases member.
651 may_aliases now returns a bitmap.
652 * tree-ssa-structalias.c (merge_smts_into): Update for may_aliases
654 * tree-ssa-operands.c (add_virtual_operand): Update for
655 may_aliases being a bitmap.
657 2007-01-29 Daniel Berlin <dberlin@dberlin.org>
659 PR tree-optimization/30630
660 * tree-ssa-structalias.c (do_complex_constraint): Mark correct
663 2007-01-29 Simon Martin <simartin@users.sourceforge.net>
666 * gimplify.c (gimplify_target_expr): Make sure that the TARGET_EXPR is
667 expanded only once even if an error occurs.
669 2007-01-29 Ben Elliston <bje@au.ibm.com>
671 * gcov-io.h (__gcov_indirect_call_profiler): Declare.
672 (__gcov_average_profiler): Likewise.
673 (__gcov_ior_profiler): Likewise.
674 (__gcov_merge_ior): Likewise.
676 2007-01-28 Jan Hubicka <jh@suse.cz>
678 * builtins.c (expand_builtin_memset): Fix typo in my last patch.
679 * value-prof.c (stringop_block_profile): Likewise.
681 2007-01-28 Jan Hubicka <jh@suse.cz>
683 * expr.c (emit_block_move_via_movmem, emit_block_move_via_libcall): Add
684 variant handling histograms; add wrapper.
685 (clear_storage_via_libcall): Export.
686 (emit_block_move_hints): Break out from ...; add histograms.
687 (emit_block_move): ... this one.
688 (clear_storage_hints): Break out from ...; add histograms.
689 (clear_storage): ... this one.
690 (set_storage_via_memset): Handle histogram.
691 * expr.h (emit_block_move_via_libcall, emit_block_move_hints): Declare.
692 (clear_storage_hints, clear_storage_via_libcall): Declare.
693 (set_storage_via_setmem): Update prototype.
694 * doc/md.texi (movmem, setmem): Document new arguments.
696 * value-prof.c (dump_histogram_value, tree_find_values_to_profile): Add
698 (stringop_block_profile): New global function.
699 (tree_stringops_values_to_profile): Profile block size and alignment.
700 * value-prof.h (enum hist_type): add HIST_TYPE_AVERAGE and
702 (struct profile_hooks): Add gen_average_profiler and gen_ior_profiler.
703 (stringop_block_profile): Declare.
704 * builtins.c: Include value-prof.h.
705 (expand_builtin_memcpy, expand_builtin_memset): Pass block profile.
706 * gcov-ui.h (GCOV_COUNTER_NAMES): Add new counter.
707 (GCOV_COUNTER_AVERAGE, GCOV_COUNTER_IOR): New constants.
708 (GCOV_COUNTERS, GCOV_LAST_VALUE_COUNTER): Update.
709 * profile.c (instrument_values): Add new counters.
710 * cfgexpand.c (expand_gimple_basic_block): Propagate histograms to
712 * tree-profile.c (tree_average_profiler_fn, tree_ior_profiler_fn): New.
713 (tree_init_edge_profiler): Build new profilers.
714 (tree_gen_average_profiler, tree_gen_ior_profiler): New.
715 (pass_tree_profile): Add dump.
716 (tree_profile_hooks): Update.
717 * Makefile.in (LIBGCOV): Add new constants.
718 * libgcov.c (__gcov_merge_ior, __gcov_average_profiler,
719 __gcov_ior_profiler): New.
720 * i386.md (movmem/setmem expanders): Add new optional arguments.
722 2007-01-28 David Edelsohn <edelsohn@gnu.org>
724 * doc/md.texi (Standard Pattern Names): Document blockage pattern.
726 2007-01-28 Zdenek Dvorak <dvorakz@suse.cz>
728 * tree-ssa-loop-unswitch.c: Include tree-inline.h.
729 (tree_unswitch_single_loop): Pass eni_size_weights to
731 * tree-ssa-loop-manip.c: Include tree-inline.h.
732 (can_unroll_loop_p): Pass eni_size_weights to
734 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p):
735 Pass eni_size_weights to estimate_num_insns.
736 * tree.h (init_inline_once): Export.
737 * toplev.c (backend_init): Call init_inline_once.
738 * cgraphunit.c (cgraph_process_new_functions,
739 cgraph_analyze_function): Pass eni_inlining_weights to
741 * ipa-inline.c (compute_inline_parameters): Ditto.
742 * tree-ssa-loop-ivcanon.c (tree_num_loop_insns): Pass weights
743 to estimate_num_insns.
744 (try_unroll_loop_completely): Pass eni_size_weights to
746 * tree-eh.c (decide_copy_try_finally): Pass eni_size_weights
747 ot estimate_num_insns.
748 * tree-ssa-loop-prefetch.c: Include tree-inline.h.
749 (loop_prefetch_arrays): Pass eni_time_weights to tree_num_loop_insns.
750 * tree-inline.c (eni_inlining_weights, eni_size_weights,
751 eni_time_weights): New variables.
752 (init_inline_once): Initialize them.
753 (struct eni_data): Mew.
754 (estimate_num_insns_1, estimate_num_insns): Use weights.
755 * tree-inline.h (struct eni_weights_d): New.
756 (eni_inlining_weights, eni_size_weights, eni_time_weights): Declare.
757 (estimate_num_insns): Declaration changed.
758 * cfgloop.h (tree_num_loop_insns): Declaration changed.
759 * Makefile.in (tree-ssa-loop-unswitch.o, tree-ssa-loop-prefetch.o,
760 tree-ssa-loop-manip.o): Add TREE_INLINE_H dependency.
762 2007-01-28 Zdenek Dvorak <dvorakz@suse.cz>
764 * tree-data-ref.c (conflict_fn): Assert that the number of affine
765 relations in the conflict function is valid.
767 2007-01-27 Ian Lance Taylor <iant@google.com>
769 * common.opt: Add fstrict-overflow.
770 * opts.c (decode_options): Set flag_strict_overflow if -O2.
771 * flags.h (TYPE_OVERFLOW_WRAPS): Define.
772 (TYPE_OVERFLOW_UNDEFINED): Define.
773 (TYPE_OVERFLOW_TRAPS): Define. This replaces TYPE_TRAP_SIGNED.
775 * tree.h (TYPE_TRAP_SIGNED): Don't define.
776 * fold-const.c (negate_expr_p): Use TYPE_OVERFLOW_UNDEFINED.
777 (fold_negate_expr): Likewise.
778 (make_range): Likewise.
779 (extract_muldiv_1): Likewise.
780 (maybe_canonicalize_comparison): Likewise.
781 (fold_comparison): Likewise.
782 (fold_binary): Likewise.
783 (tree_expr_nonnegative_p): Likewise.
784 (tree_expr_nonzero_p): Likewise.
785 * tree-vrp.c (compare_values): Likewise.
786 (extract_range_from_binary_expr): Likewise.
787 (extract_range_from_unary_expr): Likewise.
788 * tree-ssa-loop-niter.c (infer_loop_bounds_from_signedness):
790 (nowrap_type_p): Likewise.
791 * tree-scalar-evolution.c (simple_iv): Likewise.
792 * fold-const.c (negate_expr_p): Use TYPE_OVERFLOW_WRAPS.
793 (build_range_check): Likewise.
794 (extract_muldiv_1): Likewise.
795 (fold_comparison): Likewise.
796 * tree-vrp.c (vrp_int_const_binop): Likewise.
797 (extract_range_from_unary_expr): Likewise.
798 * convert.c (convert_to_integer): Likewise.
799 * fold-const.c (fold_negate_expr): Use TYPE_OVERFLOW_TRAPS.
800 (fold_comparison): Likewise.
801 (fold_binary): Likewise.
802 * optabs.c (optab_for_tree_code): Likewise.
803 * tree-vectorizer.c (vect_is_simple_reduction): Likewise.
804 * simplify-rtx.c (simplify_const_relational_operation): Check
805 flag_strict_overflow and flag_trapv.
806 (simplify_const_relational_operation): Likewise.
807 * doc/invoke.texi (Option Summary): Mention -fstrict-overflow.
808 (Optimize Options): Add -fstrict-overflow to -O2 list. Document
811 2007-01-27 Roger Sayle <roger@eyesopen.com>
813 * tree.c (tree_fold_gcd): Delete.
814 * tree.h (tree_fold_gcd): Remove prototype.
815 * tree-data-ref.c (tree_fold_divides_p): Don't use tree_fold_gcd to
816 test whether one constant integer is a multiple of another. Instead
817 call int_const_binop with TRUNC_MOD_EXPR and test for a zero result.
818 * fold-const.c (multiple_of_p): We've determined both TOP and
819 BOTTOM are integer constants so we can call int_const_binop directly
820 instead of the more generic const_binop.
822 2007-01-27 Roger Sayle <roger@eyesopen.com>
824 * fold-const.c (size_binop): In the fast-paths for X+0, 0+X, X-0 and
825 1*X check that the constant hasn't overflowed, to preserve the
827 (round_up): Provide an efficient implementation when rouding-up an
828 INTEGER_CST to a power-of-two.
830 2007-01-28 Ralf Wildenhues <Ralf.Wildenhues@gmx.de>
832 * doc/sourcebuild.texi: Add comma for clarity.
833 * doc/extend.texi: Fix some typos.
834 * doc/passes.texi: Likewise.
835 * doc/cppinternals.texi: Likewise.
836 * doc/c-tree.texi: Likewise.
837 * doc/tree-ssa.texi: Likewise.
838 * doc/install.texi: Likewise.
840 2007-01-27 Jan Hubicka <jh@suse.cz>
842 * tree-sra.c (sra_walk_function): Don't rely on aliases being build.
843 (pass_sra): Do not require alias information.
844 * passes.c (init_optimization_passes): Add SRA
846 2007-01-27 Steven Bosscher <steven@gcc.gnu.org>
848 * tracer.c (rest_of_handle_tracer): We already cleaned
849 up the CFG in tracer() so don't do it here again.
850 * cfgcleanup.c (rest_of_handle_jump2): Don't repeat
851 cleanup_cfg here, either. And don't call renumber_insns.
853 * cfgrtl.c (rtl_verify_flow_info_1): Don't verify that BB_END
854 and BB_HEAD are in the insn stream here. Instead make sure
855 that BB_INSN is valid on all insns. Also, do check here that
856 there are no pending branch predictions...
857 (rtl_verify_flow_info): ...instead of doing it here. Checks
858 for BB_END and BB_HEAD moved from rtl_verify_flow_info_1 to
861 2007-01-26 Roger Sayle <roger@eyesopen.com>
863 * config/i386/i386.c (ix86_swap_binary_operands_p): New helper
864 function to simplify/factorize operand order canonicalization.
865 (ix86_fixup_binary_operands): Reorganize using the above function.
866 (ix86_binary_operator_ok): Likewise.
868 2007-01-27 Jakub Jelinek <jakub@redhat.com>
870 * genattrtab.c (struct attr_value_list, insn_code_values): Move to
871 file scope from optimize_attrs.
872 (simplify_test_exp): If insn_code_values is not NULL, use it to speed
874 (optimize_attrs): Clear insn_code_values after freeing it.
876 2007-01-26 Zdenek Dvorak <dvorakz@suse.cz>
878 * tree-ssa-address.c (create_mem_ref): Remove ", bsi" from
879 a parts.base assignment.
881 2007-01-26 Zdenek Dvorak <dvorakz@suse.cz>
883 * tree-data-ref.c (dump_subscript): Use dump_conflict_function.
884 (compute_subscript_distance, initialize_data_dependence_relation,
885 finalize_ddr_dependent, analyze_ziv_subscript,
886 analyze_siv_subscript_cst_affine,
887 compute_overlap_steps_for_affine_univar,
888 compute_overlap_steps_for_affine_1_2, analyze_subscript_affine_affine,
889 analyze_siv_subscript, analyze_miv_subscript,
890 analyze_overlapping_iterations, subscript_dependence_tester_1,
891 compute_self_dependence, free_dependence_relation): Work
892 with affine_fn instead of chrecs.
893 (dump_affine_function, dump_conflict_function, affine_function_equal_p,
894 common_affine_function, affine_function_base,
895 affine_function_constant_p, affine_fn_op, affine_fn_plus,
896 affine_fn_minus, affine_fn_free, conflict_fn_not_known,
897 conflict_fn_no_dependence, free_conflict_function, free_subscripts,
898 conflict_fn, affine_fn_cst, affine_fn_univar): New functions.
899 (all_chrecs_equal_p): Removed.
900 * tree-data-ref.h (affine_fn, conflict_function): New types.
901 (struct subscript): Change type of conflicting_iterations_in_a
902 and conflicting_iterations_in_b.
904 2007-01-26 Steve Ellcey <sje@cup.hp.com>
907 * config/pa/pa.h (TARGET_HPUX_11): New.
908 * config/pa/pa-hpux11.h (TARGET_HPUX_11): New.
909 * config/pa/pa.c (pa_init_builtins): Use TARGET_HPUX_11.
911 2007-01-26 Daniel Berlin <dberlin@dberlin.org>
912 Richard Guenther <rguenther@suse.de>
914 * tree-ssa-structalias.c (solve_graph): Handle case
915 we merged the variable to another.
917 2007-01-25 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
919 * builtins.c (fold_builtin_1): Treat ccos and ccosh as 'even'
922 * fold-const.c (negate_mathfn_p): Treat casin, casinh, catan,
923 catanh, cproj, csin, csinh, ctan and ctanh as 'odd' functions.
925 2007-01-25 DJ Delorie <dj@redhat.com>
927 * config/m32c/m32c.c (m32c_cannot_change_mode_class): We don't
928 allow changes to modes which don't fit in those registers.
930 * reload1.c (choose_reload_regs): Check for invalid subregs before
931 computing their locations, not after.
933 2007-01-25 Geoffrey Keating <geoffk@apple.com>
936 * config/rs6000/rs6000.c (first_altivec_reg_to_save): On Darwin,
937 save Altivec registers in an eh_return function.
938 (compute_vrsave_mask): Likewise.
939 (rs6000_stack_info): Correct AIX/Darwin stack alignment computation
940 for saving Altivec registers.
941 (rs6000_emit_prologue): Don't allocate stack twice in
942 eh_return function. Correct expected value of altivec_save_offset
943 when using save_world. Describe save of R0 to stack when using
944 save_world. Describe stack pointer adjustment when using
945 save_world. Remove duplicated eh_return parameter register saving.
946 Update sp_offset variable after save_world.
947 * config/rs6000/t-darwin (LIB2FUNCS_STATIC_EXTRA): Remove
949 (LIB2FUNCS_EXTRA): Add darwin-world.asm.
950 * config/rs6000/darwin.h (SUBTARGET_OVERRIDE_OPTIONS): -m64
953 2007-01-25 Steve Ellcey <sje@cup.hp.com>
955 * config.gcc (ia64*-*-hpux*): Make posix threads the default.
957 2007-01-25 Steve Ellcey <sje@cup.hp.com>
960 * config/pa/pa.c (pa_init_builtins): Set asm names for finite routines.
961 * config/ia64/ia64.c (ia64_init_builtins): Ditto.
963 2007-01-25 Richard Guenther <rguenther@suse.de>
965 * doc/invoke.texi (-Wcoverage-mismatch): Document.
966 * common.opt (-Wcoverage-mismatch): New warning option.
967 * coverage.c (get_coverage_counts): Ignore coverage mismatch
968 if -Wcoverage-mismatch is given.
970 2007-01-25 Razya Ladelsky <razya@il.ibm.com>
972 * ipa-cp.c (ipcp_insert_stage, ipcp_driver): Support for SSA.
973 (ipcp_driver): Change to static definition.
974 Add dumping of the ifunctions.
975 (constant_val_insert): Remove unused parameter. Support for SSA.
976 (ipcp_propagate_const): Support for SSA.
977 (ipcp_profile_bb_print): Print only analyzed nodes.
978 (ipcp_replace_map_create): Remove support for Fortran constant
980 * ipa-prop.c (ipa_method_modify_stmt,
981 ipa_callsite_compute_param): Support for SSA.
982 * ipa-prop.h (ipcp_driver): Remove declaration.
983 (IS_VALID_TREE_MAP_INDEX): Add define.
985 2007-01-24 Geoffrey Keating <geoffk@apple.com>
987 * unwind-dw2.c (execute_stack_op): Handle DW_OP_swap.
989 2007-01-24 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
992 * builtins.c (fold_builtin_cabs): Use MPFR to evaluate a
993 constant argument to cabs and do it without checking for
994 -funsafe-math-optimizations.
996 2007-01-24 Douglas Gregor <dgregor@osl.iu.edu>
998 * c-common.h (RID_FIRST_CXX0X): New.
999 (RID_LAST_CXX0X): New.
1000 * c-opts.c (c_common_handle_option): -Wc++0x-compat is triggered
1002 * c.opt (Wc++0x-compat): New.
1003 * doc/invoke.texi (-Wc++0x-compat): Document.
1005 2007-01-24 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
1007 * builtins.c (fold_builtin_carg): New.
1008 (fold_builtin_1): Use it.
1010 2007-01-24 Jan Hubicka <jh@suse.cz>
1012 * ipa-inline.c (cgraph_decide_inlining): Initialize initial_insns.
1014 * ipa-inline.c (initial_insns, max_insns): Delete.
1015 (compute_max_insns): New function.
1016 (cgraph_decide_inlining_of_small_function): Use it; take minimal amount
1017 of insns as base for code growth.
1018 (cgraph_decide_inlining): Make initial_insns local; do not compute
1020 * params.def (PARAM_INLINE_UNIT_GROWTH): Set to 60.
1021 * doc/invoke.texi (inline-unit-growth): Update docs.
1023 2007-01-24 Jakub Jelinek <jakub@redhat.com>
1025 * config/i386/i386.h (x86_cmpxchg16b): Remove const.
1026 (TARGET_CMPXCHG16B): Define to x86_cmpxchg16b.
1027 * config/i386/i386.c (x86_cmpxchg16b): Remove const.
1028 (override_options): Add PTA_CX16 flag. Set x86_cmpxchg16b
1029 for CPUs that have PTA_CX16 set.
1032 * gimplify.c (omp_check_private): New function.
1033 (gimplify_scan_omp_clauses): Use it for
1034 firstprivate/lastprivate/reduction.
1037 * gimplify.c (omp_add_variable): Don't call omp_notice_variable
1038 on TYPE_SIZE_UNIT for GOVD_LOCAL VLAs.
1041 * omp-low.c (lower_omp_for_lastprivate): Add dlist argument.
1042 If lower_lastprivate_clauses emits some statements, append them
1043 to dlist rather than body_p and to body_p append an initializer.
1044 (lower_omp_for): Adjust caller.
1046 2007-01-24 Steve Ellcey <sje@cup.hp.com>
1048 * target.h (globalize_decl_name): New.
1049 * target-def.h (TARGET_ASM_GLOBALIZE_DECL_NAME): New.
1050 * output.h (default_globalize_decl_name): New.
1051 * varasm.c (asm_output_bss): Use globalize_decl_name instead of
1053 (globalize_decl): Ditto.
1054 (default_globalize_decl_name): New.
1055 * config/ia64/ia64.c (ia64_globalize_decl_name): New.
1056 (ia64_handle_version_id_attribute): New.
1057 (TARGET_ASM_GLOBALIZE_DECL_NAME): New.
1058 (ia64_asm_output_external): Use globalize_decl_name instead
1060 * doc/extend.texi (version_id): New pragma.
1061 * doc/tm.texi (ARGET_ASM_GLOBALIZE_DECL_NAME): New target hook.
1063 2007-01-24 Andreas Krebbel <krebbel1@de.ibm.com>
1065 * unwind-dw2-fde.c (get_cie_encoding): Replaced _Unwind_Word with
1066 _uleb128_t and _Unwind_SWord with _sleb128_t.
1067 * unwind-dw2.c (extract_cie_info, execute_stack_op, execute_cfa_program,
1068 uw_frame_state_for, uw_update_context_1): Likewise.
1069 * unwind-c.c (parse_lsda_header, PERSONALITY_FUNCTION): Likewise.
1070 * unwind-pe.h (read_uleb128, read_sleb128,
1071 read_encoded_value_with_base): Likewise.
1072 * unwind-generic.h: Define _sleb128_t and _uleb128_t types.
1074 2007-01-24 Richard Guenther <rguenther@suse.de>
1076 * builtins.c (expand_builtin_cexpi): Get the fndecl
1077 for cexp in the correct way.
1079 2007-01-24 Jan Hubicka <jh@suse.cz>
1081 * tree-ssa-dce.c (eliminate_unnecesary_stmts): Remove dead LHS of calls.
1083 2007-01-24 Andreas Krebbel <krebbel1@de.ibm.com>
1085 * c-cppbuiltin.c (builtin_define_type_sizeof): New function.
1086 (c_cpp_builtins): New builtin macros: __SIZEOF_INT__, __SIZEOF_LONG__,
1087 __SIZEOF_LONG_LONG__, __SIZEOF_SHORT__, __SIZEOF_POINTER__,
1088 __SIZEOF_FLOAT__, __SIZEOF_DOUBLE__, __SIZEOF_LONG_DOUBLE__,
1089 __SIZEOF_SIZE_T__, __SIZEOF_WCHAR_T__, __SIZEOF_WINT_T__ and
1090 __SIZEOF_PTRDIFF_T__.
1091 * doc/cpp.texi: Documentation for the new builtin macros added.
1093 2007-01-24 Uros Bizjak <ubizjak@gmail.com>
1095 * config/i386/i386.md (tanxf2, tan<mode>2, atan<mode>2, log<mode>2,
1096 log10<mode>2, log2<mode>2, expxf2, exp10xf2, exp2xf2): Use op2
1097 instead of operands[2] to avoid access past the end of array.
1099 2007-01-24 Richard Sandiford <richard@codesourcery.com>
1101 * reload1.c (emit_reload_insns): Pass the reload register
1102 for a non-spill output reload through forget_old_reloads_1.
1104 2007-01-23 Joseph Myers <joseph@codesourcery.com>
1106 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Define
1107 _SOFT_DOUBLE if doubles use software floating-point.
1108 * config/rs6000/libgcc-ppc-glibc.ver: Export additional long
1109 double functions if _SOFT_DOUBLE, not _SOFT_FLOAT.
1110 * config/rs6000/darwin-ldouble.c: Also compile functions for
1111 hard-float without FPRs. Use fmsub function for all __NO_FPRS__
1112 cases. Compile extra functions if _SOFT_DOUBLE, not _SOFT_FLOAT.
1113 * config/rs6000/linuxspe.h (SUBSUBTARGET_OVERRIDE_OPTIONS): Remove
1114 commented-out long double override.
1115 (CPP_LONGDOUBLE_DEFAULT_SPEC): Likewise.
1116 * config/rs6000/eabispe.h: Likewise.
1117 * config/rs6000/rs6000.c (rs6000_override_options): Don't override
1118 long double for non-SPE.
1119 (rs6000_handle_option): Likewise.
1120 (invalid_e500_subreg): Disallow more subregs involding DImode,
1121 DFmode, TImode or TFmode.
1122 (rs6000_legitimate_offset_address_p): Check TFmode offsets for
1124 (legitimate_lo_sum_address_p): Also check for TFmode for E500
1126 (rs6000_legitimize_address): Also handle TFmode for E500 double.
1127 (rs6000_legitimize_reload_address): Also handle TFmode for E500
1129 (rs6000_legitimate_address): Also check for TFmode for E500
1131 (rs6000_emit_move): Use DFmode subregs of TFmode for E500 double.
1132 (spe_build_register_parallel): Handle TFmode and TCmode.
1133 (rs6000_spe_function_arg): Handle TFmode and TCmode for E500
1135 (function_arg): Handle TFmode and TCmode for E500 double.
1136 (rs6000_init_libfuncs): Initialize extra libfuncs for soft double
1138 (print_operand): Handle TFmode and TImode for %y.
1139 (rs6000_generate_compare): Handle TFmode comparisons for E500
1141 (spe_func_has_64bit_regs_p): Check for TFmode for E500 double.
1142 (rs6000_function_value): Handle TFmode and TCmode for E500 double.
1143 (rs6000_libcall_value): Handle TFmode and TCmode for E500 double.
1144 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Check for
1145 TFmode for E500 double.
1146 * config/rs6000/rs6000.md (FP): Allow TF for E500 double.
1147 (floatsidf2): Enable for E500 double.
1148 (movtf_softfloat): Use rs6000_nonimmediate_operand.
1149 (extenddftf2): Change to extenddftf2_fprs.
1150 (extenddftf2): Call gen_spe_extenddftf2 or gen_extenddftf2_fprs
1151 depending on TARGET_E500_DOUBLE.
1152 (extendsftf2): Enable for E500 double.
1153 (trunctfdf2): Enable for E500 double.
1154 (trunctfsf2): Change to trunctfsf2_fprs.
1155 (trunctfsf2): Call gen_spe_trunctfsf2 or gen_trunctfsf2_fprs
1156 depending on TARGET_E500_DOUBLE.
1157 (floatsitf2): Enable for E500 double.
1158 (fix_trunctfsi2): Change to fix_trunctfsi2_fprs.
1159 (fix_trunctfsi2): Call gen_spe_fix_trunctfsi2 or
1160 gen_fix_trunctfsi2_fprs depending on TARGET_E500_DOUBLE.
1161 (negtf2): Change to negtf2_internal.
1162 (negtf2): New expander.
1163 (abstf2): Enable for E500 double. Call gen_spe_abstf2_tst,
1164 gen_spe_abstf2_cmp or gen_abstf2_internal depending on
1165 TARGET_E500_DOUBLE and flag_unsafe_math_optimizations.
1166 (movdi_internal32): Use rs6000_nonimmediate_operand.
1167 (unnamed splitter): Likewise.
1168 * config/rs6000/spe.md (CMPTFEQ_GPR, TSTTFEQ_GPR, CMPTFGT_GPR,
1169 TSTTFGT_GPR, CMPTFLT_GPR, TSTTFLT_GPR): New unspecs.
1170 (SPE64TF, DITI): New mode macros.
1171 (frob_df_di): Change to frob_<SPE64:mode>_<DITI:mode>; allow more
1174 (frob_<mode>_di_2): New.
1175 (frob_tf_di_8_2): New.
1176 (frob_di_df): Change to frob_di_<mode>; allow more modes.
1178 (frob_di_df_2): Change to frob_<DITI:mode>_<SPE64:mode>_2; allow
1180 (frob_ti_<mode>_8_2): New.
1181 (frob_ti_tf_2): New.
1182 (mov_si<mode>_e500_subreg0, mov_si<mode>_e500_subreg0_2,
1183 mov_si<mode>_e500_subreg4, mov_si<mode>_e500_subreg4_2): Allow
1185 (mov_sitf_e500_subreg8, mov_sitf_e500_subreg8_2,
1186 mov_sitf_e500_subreg12, mov_sitf_e500_subreg12_2): New.
1187 (spe_trunctfdf2_internal1, spe_trunctfsf2, spe_extenddftf2,
1188 spe_fix_trunctfsi2, spe_fix_trunctfsi2_internal,
1189 spe_negtf2_internal, spe_abstf2_cmp, spe_abstf2_tst): New.
1190 (cmptfeq_gpr, tsttfeq_gpr, cmptfgt_gpr, tsttfgt_gpr, cmptflt_gpr,
1193 2007-01-23 Ian Lance Taylor <iant@google.com>
1195 * Makefile.in (OBJS-common): Reformat, alphabetize, but put
1197 (OBJS-archive): Reformat, alphabetize.
1198 (OBJS): Change out_object_file to OBJS-md.
1200 2007-01-23 Uros Bizjak <ubizjak@gmail.com>
1202 * config/i386/i386.md: Use REG_P, MEM_P, CONST_INT_P, LABEL_P,
1203 JUMP_P and CALL_P predicates where applicable.
1204 * config/i386/i386.c: Ditto.
1205 * config/i386/i386.md: Ditto.
1206 * config/i386/mmx.md: Ditto.
1207 * config/i386/predicates.md: Ditto.
1209 2007-01-22 Andreas Schwab <schwab@suse.de>
1211 * config/m68k/m68k.h: Fix comment.
1213 2007-01-22 Jan Hubicka <jh@suse.cz>
1215 * passes.c (init_optimization_passes): Do not rerun
1216 pass_early_warn_uninitialized.
1218 2007-01-22 Richard Guenther <rguenther@suse.de>
1220 PR tree-optimization/30038
1221 * tree-ssa-math-opts.c (maybe_record_sincos): New static helper
1223 (execute_cse_sincos_1): Likewise.
1224 (execute_cse_sincos): Likewise.
1225 (gate_cse_sincos): Likewise.
1226 (pass_cse_sincos): New pass CSEing sin() and cos() calls using
1227 the cexpi() canonicalization of sincos().
1228 * tree-pass.h (pass_cse_sincos): Declare.
1229 * passes.c (init_optimization_passes): New pass pas_cse_sincos.
1231 2007-01-21 Eric Botcazou <ebotcazou@libertysurf.fr>
1233 PR rtl-optimization/29329
1234 * combine.c (replaced_rhs_insn): Rename to i2mod.
1235 (replaced_rhs_value): Rename to i2mod_new_rhs.
1236 (i2mod_old_rhs): New global variable.
1237 (combine_instructions): Adjust for above change. Save a copy of
1238 the old RHS into i2mod_old_rhs when the contents of a REG_EQUAL
1239 note are substituted in the second instruction.
1240 (distribute_notes) <REG_DEAD>: Adjust for above change. Do not
1241 ditch the note if it pertains to the second eliminated register
1242 and this register is mentioned in i2mod_old_rhs.
1245 2006-09-12 Eric Botcazou <ebotcazou@libertysurf.fr>
1247 * combine.c (distribute_notes) <REG_DEAD>: Do not consider SETs past
1248 the insn to which the note was originally attached.
1250 2007-01-21 Jan Hubicka <jh@suse.cz>
1252 * ipa-inline.c (inlining_mode): Comment, move up.
1253 (cgraph_decide_inlining_incrementally): Do not perform inlining
1254 itself; fix handling of flattening of self recursive functions.
1255 (cgraph_find_cycles): Remove.
1256 (cgraph_flatten_node): Remove.
1257 (cgraph_decide_inlining): Use incremental inliner to handle flattening.
1258 (try_inline): New function.
1259 (cgraph_early_inlining): Update call of
1260 cgraph_decide_inlining_incrementally. Apply inlining here.
1261 (apply_inline): Update call of cgraph_decide_inlining_incrementally.
1263 2007-01-21 Dirk Mueller <dmueller@suse.de>
1266 * tree-vrp.c (check_array_bounds): do not warn
1267 about ADDR_EXPR's of ARRAY_REF's which are immediately
1268 used in binary expressions.
1270 2007-01-21 Ira Rosen <irar@il.ibm.com>
1272 * tree-vectorizer.h (struct _stmt_vec_info): Add new field
1273 read_write_dep and macros for its access.
1274 * tree-vectorizer.c (new_stmt_vec_info): Initialize the new field.
1275 * tree-vect-analyze.c (vect_analyze_data_ref_dependence): Remove
1276 argument, call vect_check_interleaving for every independent pair of
1277 data-refs. Mark loads that access the same memory location as a store
1279 (vect_check_dependences): Remove.
1280 (vect_analyze_data_ref_dependences): Remove vect_check_dependences
1281 call, fix the call to vect_analyze_data_ref_dependence.
1282 (vect_analyze_data_ref_access): For statements that access the same
1283 data-ref, check that they are not stores; for loads, check that there
1284 is no store that access the same location.
1286 2007-01-20 Roger Sayle <roger@eyesopen.com>
1287 Joseph Myers <joseph@codesourcery.com>
1289 * doc/invoke.texi (-fdump-rtl-combine): Fix under/overfull hbox.
1290 (-fdump-rtl-stack): Likewise.
1291 (-fno-signed-zeros): Use @minus{} for a minus sign. Correct typo.
1292 (-mcheck-zero-division, -mcpu): Fix under/overfull hbox.
1293 (-mpt-fixed): Use @minus{} for minus sign.
1294 (Using Precompiled Headers): Fix under/overfull hbox.
1296 2007-01-20 Jan Hubicka <jh@suse.cz>
1298 * tree-flow.h (struct stmt_ann_d): Move references_memory to proper
1299 place within annotation.
1301 2007-01-20 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
1303 * pa.c (output_move_double): Change array size of xoperands to 4.
1305 2007-01-20 Jan Hubicka <jh@suse.cz>
1307 * tree-tailcall.c (adjust_return_value): Do not use RESULT_DECL
1310 2007-01-19 Ian Lance Taylor <iant@google.com>
1312 * expmed.c (expand_divmod) [TRUNC_MOD_EXPR, TRUNC_DIV_EXPR]: Cast
1313 constant to unsigned HOST_WIDE_INT before negating.
1315 2007-01-19 Ian Lance Taylor <iant@google.com>
1317 * tree-ssa-operands.h (struct vuse_vec_d): Change num_vuse field
1319 (VUSE_VECT_ELEMENT) [ENABLE_CHECKING]: Use unsigned comparison.
1320 (VUSE_ELEMENT_PTR) [ENABLE_CHECKING]: Likewise.
1321 (SET_VUSE_VECT_ELEMENT) [ENABLE_CHECKING]: Likewise.
1322 (SET_VUSE_ELEMENT_VAR) [ENABLE_CHECKING]: Likewise.
1323 (SET_VUSE_ELEMENT_PTR) [ENABLE_CHECKING]: Likewise.
1324 (realloc_vdef, realloc_vuse): Change second parameter to
1326 (ssa_operand_iterator_d): Change vuse_index and mayuse_index
1328 * tree-ssa-operands.c (realloc_vop): Change num_elem parameter to
1329 unsigned. Change x and lim locals to unsigned.
1330 (realloc_vdef, realloc_vuse): Change num_elem parameter to
1332 (finalize_ssa_vuse_ops): Change old_i local to unsigned.
1333 (copy_virtual_operands): Change i and n locals to unsigned.
1335 2007-01-19 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
1338 * builtins.c (fold_builtin_1): Handle builtin fdim.
1340 2007-01-20 Jan Hubicka <jh@suse.cz>
1342 * tree-ssa.c (init_tree_ssa): Do not call init_alias_heapvars.
1343 * tree-ssa-structalias.c (compute_points_to_sets): Do call
1344 init_alias_heapvars.
1345 (init_alias_heapvars): Initialize only when not already initialized.
1346 (delete_alias_heapvars): Set heapvar_for_stmt to NULL.
1348 2007-01-19 Roger Sayle <roger@eyesopen.com>
1350 * common.opt (fsigned-zeros): New command line option.
1351 * flags.h (HONOR_SIGNED_ZEROS): Control via flag_signed_zeros instead
1352 of flag_unsafe_math_optimizations.
1353 * opts.c (set_fast_math_flags): The -ffast-math command line option
1354 implies -fno-signed-zeros.
1355 (fast_math_flags_set_p): Likewise.
1357 * doc/invoke.texi: Document new -fno-signed-zeros option, and update
1358 the documentation of -ffast-math appropriately. Wrap long lines.
1360 2007-01-19 Steve Ellcey <sje@cup.hp.com>
1362 * system.h (ASM_MAKE_LABEL_LINKONCE): Poison.
1363 * varasm.c (globalize_decl): Remove ASM_MAKE_LABEL_LINKONCE ifdef.
1365 2007-01-19 Tomas Bily <tbily@suse.cz>
1367 * cgraphunit.c (cgraph_finalize_function): Updating of pid
1369 (tree_init_ic_make_global_vars): New function
1370 (tree_init_edge_profiler): call of tree_init_ic_make_global_vars
1371 (tree_gen_ic_profiler): New function
1372 (tree_gen_ic_func_profiler): New function
1373 (tree_profiling): Added calling of tree_gen_ic_func_profiler
1374 (tree_profile_hooks): Added hook for indirec/virtual calls
1375 * value-prof.c (tree_find_values_to_profile): New case for
1377 (tree_values_to_profile): Call for determining indirect/virtual
1379 (tree_indirect_call_to_profile): New function
1380 (tree_ic_transform): New function
1381 (tree_ic): New function
1382 (find_func_by_pid): New function
1383 (init_pid_map): New function
1384 (tree_value_profile_transformations): Added check for
1385 indirect/virtual call transformation
1386 * value-prof.h (enum hist_type): New counter type for
1387 indirect/virtual calls
1388 (profile_hooks): Added new hook for profiling indirect/virtual
1390 * profile.c (instrument_values): New case for indirect/virtual
1392 * gcov-io.h (GCOV_LAST_VALUE_COUNTER): Changed to 6
1393 (GCOV_COUNTER_V_INDIR): New counter type
1394 (GCOV_COUNTER_NAMES): New name of counter "indirect" added
1395 (GCOV_MERGE_FUNCTIONS): New merge function for indirect/virtual
1397 * cgraph.c: Definition of cgraph_max_pid
1398 (cgraph_create_node): Default init of pid attribute
1399 * cgraph.h: Declaration of cgraph_max_pid
1400 (struct cgraph_node): Added pid attribute
1401 * libgcov.c (__gcov_indirect_call_profiler): New function
1402 (__gcov_one_value_profiler_body): New function
1403 (__gcov_one_value_profiler): Body was moved to
1404 __gcov_one_value_profiler_body and calls it
1406 2007-01-19 Basile Starynkevitch <basile@starynkevitch.net>
1408 * doc/gty.texi (Options): Document the mark_hook option to GTY.
1409 * gengtype.c (write_types_data, write_func_for_structure,
1410 write_types, ggc_wtd, pch_wtd): Add skip_hooks to
1411 write_types_data, ggc_wtd, pch_wtd for processing mark_hook.
1412 (walk_type, write_func_for_structure): Generate the mark_hook if
1415 2007-01-19 Jan Hubicka <jh@suse.cz>
1417 * ipa-inline.c (cgraph_decide_inlining_incrementally): Instead of
1418 'early' argument take inlining mode argument specifying whether to
1419 inline for size/speeed or all functions; add support for flattening;
1421 (cgraph_early_inlining): Update call of decide_inlining_incrementally.
1423 2007-01-19 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
1426 * toplev.c (warn_deprecated_use): Use %qD instead of %qs to print
1427 the name of the declared identifier.
1429 2007-01-19 Dirk Mueller <dmueller@suse.de>
1431 * config/i386.h (CONDITIONAL_REGISTER_USAGE): Store
1432 result of PIC_OFFSET_TABLE_REGNUM in temporary variable to avoid
1433 duplicate evaluation.
1435 2007-01-19 Uros Bizjak <ubizjak@gmail.com>
1437 * config/i386/i386.md (acos<mode>2): Rename from acossf2 and acosdf2.
1438 Macroize expander using X87MODEF12 mode macro. Extend operand 1
1439 to XFMode, use acosxf2 and truncate result to requested mode.
1440 Use SSE_FLOAT_MODE_P to disable patterns for SSE math.
1441 (asin<mode>2): Similarly, with asin expanders.
1442 (*fscalexf4_i387): Rename from *fscalexf4.
1443 (expNcorexf3): New expander.
1444 (expxf2, exp10xf2, exp2xf2): Use expNcorexf3 expander.
1445 (exp<mode>2): Rename from expsf2 and expdf2. Macroize expander using
1446 X87MODEF12 mode macro. Extend operand 1 to XFMode, use expxf2 and
1447 truncate result to requested mode. Use SSE_FLOAT_MODE_P to disable
1448 patterns for SSE math.
1449 (exp10<mode>2): Similarly, with exp10 expanders.
1450 (exp2<mode>2): Similarly, with exp2 expanders.
1451 (expm1<mode>2): Similarly, with expm1 expanders.
1452 (ldexp<mode>3): Similarly, with ldexp expanders.
1453 (log<mode>2, log10<mode>2, log2<mode>2, log1p<mode>2, logb<mode>2):
1454 Use gen_truncxf<mode>2_i387_noop to truncate result.
1456 2007-01-19 Richard Sandiford <richard@codesourcery.com>
1458 * config/m68k/m68k.h (PREFERRED_STACK_BOUNDARY): Define to 32
1459 for ColdFire targets.
1461 2007-01-19 Nathan Sidwell <nathan@codesourcery.com>
1462 Richard Sandiford <richard@codesourcery.com>
1464 * config/m68k/m68k.h (M68K_STATIC_CHAIN_REG_NAME): New macro.
1465 (INITIALIZE_TRAMPOLINE): Use STATIC_CHAIN_REGNUM.
1466 (__transfer_from_trampoline): Use M68K_STATIC_CHAIN_REG_NAME.
1467 * config/m68k/m68kelf.h (STATIC_CHAIN_REGNUM): Override.
1468 (M68K_STATIC_CHAIN_REG_NAME): Likewise.
1469 * config/m68k/netbsd-elf.h (M68K_STATIC_CHAIN_REG_NAME): Likewise.
1471 2007-01-19 Richard Sandiford <richard@codesourcery.com>
1473 * config/m68k/m68k.md (adddi_dilshr32): Rename to...
1474 (*adddi_dilshr32): ...this. Fix formatting. Remove commented-out
1475 non-canonical pattern. Restrict to !TARGET_COLDFIRE.
1476 (*adddi_dilshr32_cf): New pattern.
1477 (adddi3, subdi3): Remove first alternatives.
1479 2007-01-19 Richard Sandiford <richard@codesourcery.com>
1481 * config/m68k/m68k.c (notice_update_cc): If an SFmode move is
1482 implemented using move.l, do not use its cc result for floating-point
1485 2007-01-19 Richard Sandiford <richard@codesourcery.com>
1487 * config/m68k/m68k.h (EXTRA_CONSTRAINT): Stop the 'T' constraint
1488 from accepting 's' constraints if flag_pic.
1490 2007-01-19 Richard Sandiford <richard@codesourcery.com>
1492 * config/m68k/m68k.md (bordered, bunordered, buneq, bunge, bungt)
1493 (bunle, bunlt, bltgt, bordered_rev, bunordered_rev, buneq_rev)
1494 (bunge_rev, bungt_rev, bunle_rev, bunlt_rev, bltgt_rev): Change
1495 condition from TARGET_68881 to TARGET_HARD_FLOAT.
1497 2007-01-19 Sandra Loosemore <sandra@codesourcery.com>
1499 * longlong.h (count_leading_zeros, COUNT_LEADING_ZEROS_0): Add
1500 ColdFire alternatives.
1501 * config/m68k/m68k.h (CLZ_DEFINED_VALUE_AT_ZERO): New macro.
1502 * config/m68k/m68k.md (clzsi2): Define for ColdFire
1503 architectures that support ff1 instruction.
1505 2007-01-19 Richard Sandiford <richard@codesourcery.com>
1506 Julian Brown <julian@codesourcery.com>
1508 * config/m68k/m68k.h (CONST_OK_FOR_LETTER_P): Add an 'R' case.
1509 * config/m68k/m68k.md (*movsi_cfv4): Fold into...
1510 (*movsi_cf): ...here. Remove unnecessary 'R' from 'Rg'.
1513 2007-01-19 Richard Sandiford <richard@codesourcery.com>
1515 * config/m68k/m68k-protos.h (valid_mov3q_const): Take a HOST_WIDE_INT
1517 (output_move_const_into_data_reg, output_move_simode_const): Delete.
1518 * config/m68k/m68k.c (const_method, const_int_cost): Take a
1519 HOST_WIDE_INT instead of an rtx.
1520 (m68k_rtx_costs): Update call accordingly.
1521 (output_move_const_into_data_reg): Likewise. Fix formatting.
1522 (valid_mov3q_const): Take a HOST_WIDE_INT instead of an rtx.
1524 (output_move_simode_const): Update calls after above changes.
1525 Rework to use automatic variables and predicates like MEM_P.
1526 * config/m68k/m68k.md (pushexthisi_const): Update call to
1529 2007-01-19 Dirk Mueller <dmueller@suse.de>
1531 * tree-ssa-alias.c (perform_var_substitution): Fix typo
1534 2007-01-19 Richard Guenther <rguenther@suse.de>
1536 * builtins.c (expand_builtin_cexpi): Fall back to expanding
1537 via cexp in case sincos is not available.
1539 2007-01-19 Richard Guenther <rguenther@suse.de>
1541 * doc/tm.texi (TARGET_HAS_SINCOS): Document new target macro.
1542 * defaults.h (TARGET_HAS_SINCOS): Default to off.
1543 * config/linux.h (TARGET_HAS_SINCOS): Set to on if we have glibc.
1544 * config/alpha/linux.h (TARGET_HAS_SINCOS): Likewise.
1545 * config/sparc/linux.h (TARGET_HAS_SINCOS): Likewise.
1546 * config/sparc/linux64.h (TARGET_HAS_SINCOS): Likewise.
1547 * config/rs6000/linux.h (TARGET_HAS_SINCOS): Likewise.
1548 * config/rs6000/linux64.h (TARGET_HAS_SINCOS): Likewise.
1550 2007-01-19 Uros Bizjak <ubizjak@gmail.com>
1552 * config/i386/i386.md (*fpatanxf3_i387, fpatan_extend<mode>xf3_i387):
1554 (atan2sf3_1, atan2df3_1, atan2xf3_1): Remove insn patterns.
1555 (atan2xf3): Directly generate RTL pattern.
1556 (atan2<mode>3): Rename from atan2sf3 and atan2df3 and macroize insn
1557 patterns using X87MODEF12 mode macro. Use fpatan_extend<mode>xf3_i387
1558 and truncate result to requested mode. Use SSE_FLOAT_MODE_P to
1559 disable patterns for SSE math.
1560 (atan<mode>2): Rename from atansf2 and atandf2 and macroize insn
1561 patterns using X87MODEF12 mode macro. Use fpatan_extend<mode>xf3_i387
1562 and truncate result to requested mode. Use SSE_FLOAT_MODE_P to
1563 disable patterns for SSE math.
1565 2007-01-19 Alexandre Oliva <aoliva@redhat.com>
1567 * libgcc-std.ver: Fix typo in %inherit for GCC_4.3.0.
1569 2007-01-18 Roger Sayle <roger@eyesopen.com>
1571 * fold-const.c (fold_unary) <VIEW_CONVERT_EXPR>: Optimize away a
1572 VIEW_CONVERT_EXPR to the same type as it's operand.
1574 2007-01-18 David Edelsohn <edelsohn@gnu.org>
1576 * config/rs6000/darwin-ldouble.c: Only build _SOFT_FLOAT if
1577 configured for long double 128.
1579 2007-01-18 Mike Stump <mrs@apple.com>
1581 * config/rs6000/rs6000.c (rs6000_emit_vector_compare): Fix build
1584 2007-01-18 Michael Meissner <michael.meissner@amd.com>
1586 * i386.c (ix86_compute_frame_layout): Make fprintf's in #if 0 code
1589 2007-01-18 Jan Hubicka <jh@suse.cz>
1591 * tree-ssa-operands.c (vop_free_bucket_size): Never return value
1592 greater than NUM_VOP_FREE_BUCKETS.
1594 2007-01-18 Daniel Berlin <dberlin@dberlin.org>
1596 * tree-ssa-structalias.c: Update comments.
1597 (ptabitmap_obstack): Removed.
1599 (oldpta_obstack): Ditto.
1600 (stats): Add a few members.
1601 (struct variable_info): Remove node, complex, address_taken, and
1602 indirect_target members. Add oldsolution member.
1603 (new_var_info): Do not initialize removed members.
1604 (constraint_expr_type): Remove INCLUDES.
1605 (constraint_graph): Add size, implicit_preds, rep,
1606 indirect_cycles, eq_rep, label, direct_nodes, and complex members.
1607 (FIRST_REF_NODE): New macro.
1608 (LAST_REF_NODE): Ditto.
1609 (FIRST_ADDR_NODE): Ditto.
1610 (find): New function.
1612 (dump_constraint): Do not handle INCLUDES.
1613 (insert_into_complex): Do not insert duplicate constraints.
1614 (condense_varmap_nodes): Renamed and rewritten into ...
1615 (merge_node_constraints): This. Also fix bug in handling of
1616 offseted copy constraints.
1617 (clear_edges_for_node): No longer need to deal with preds at all,
1618 or removing associated preds/succs.
1619 (merge_graph_nodes): Deal with indirect_cycles.
1620 Don't deal with predecessors.
1621 (add_implicit_graph_edge): New function.
1622 (add_pred_graph_edge): Ditto.
1623 (add_graph_edge): Don't deal with predecessors.
1624 (build_constraint_graph): Removed.
1625 (build_pred_graph): New function.
1626 (build_succ_graph): Ditto.
1627 (struct scc_info): Removed in_component. Added roots, dfs, and
1628 node_mapping. Remove visited_index, unification_queue.
1629 (scc_visit): Deal with union-find we do now.
1630 Deal with cycles with REF nodes.
1631 (collapse_nodes): Renamed and rewritten to ...
1632 (unify_nodes): This.
1633 (process_unification_queue): Removed.
1634 (topo_visit): Cleanup
1635 (do_da_constraint): Use find.
1636 (do_sd_constraint): Ditto.
1637 (do_ds_constraint): Ditto.
1638 (do_complex_constraint): Ditto.
1639 (init_scc_info): Update for removed and added members.
1640 (find_and_collapse_graph_cycles): Renamed and rewritten into ...
1641 (find_indirect_cycles): This.
1642 (equivalence_class): New variable.
1643 (label_visit): New function.
1644 (perform_variable_substitution): Rewritten.
1645 (free_var_substitution_info): New function.
1646 (find_equivalent_node): Ditto.
1647 (move_complex_constraints): Ditto.
1648 (eliminate_indirect_cycles): Ditto.
1649 (solve_graph): Only propagate changed bits.
1650 Use indirect cycle elimination.
1652 (tree_id_t): Rename to tree_vi_t, delete id member, add vi member.
1653 (tree_id_eq): Renamed to ...
1654 (tree_vi_eq): This. Update for member change
1655 (insert_id_for_tree): Renamed and rewritten to ...
1656 (insert_vi_for_tree): This.
1657 (lookup_id_for_tree): Renamed and rewritten to ...
1658 (lookup_vi_for_tree): This.
1659 (get_id_for_tree): Renamed and rewritten to ...
1660 (get_vi_for_tree): Ditto.
1661 (get_constraint_exp_from_ssa_var): Update to use get_vi_for_tree.
1662 (process_constraint): Don't handle INCLUDES.
1663 Remove special ADDRESSOF case.
1664 (find_func_aliases): Rewrite to use vi functions instead of id
1666 (create_function_info_for): Ditto.
1667 (create_variable_info_for): Ditto.
1668 (intra_create_variable_infos): Ditto.
1669 (merge_smts_into): Ditto.
1670 (find_what_p_points_to): Ditto.
1671 (init_base_vars): Ditto.
1672 (init_alias_vars): Ditto.
1673 (remove_preds_and_fake_succs): New function.
1674 (dump_sa_points_to_info): Dump new stats.
1675 (dump_solution_for_var): Use find.
1676 (set_used_smts): Fix formatting.
1677 (compute_points_to_sets): Updated for new functions.
1678 (ipa_pta_execute): Ditto.
1680 2007-01-18 Kazu Hirata <kazu@codesourcery.com>
1681 Richard Sandiford <richard@codesourcery.com>
1683 * doc/tm.texi (TARGET_FUNCTION_VALUE): Expand documentation of
1685 * calls.c (expand_call): If the return value is a PARALLEL,
1686 extract its first member.
1687 * config/m68k/linux.h (FUNCTION_EXTRA_EPILOGUE): Remove.
1688 * config/m68k/m68k.c (m68k_output_function_epilogue): Don't
1689 use FUNCTION_EXTRA_EPILOGUE.
1690 (m68k_function_value): Return a PARALLEL if the return value
1691 is of a pointer type.
1692 * config/m68k/netbsd-elf.h (current_function_returns_pointer)
1693 (FUNCTION_EXTRA_EPILOGUE): Remove.
1694 * config/m68k/m68k.md (D0_REG): New constant.
1696 2007-01-18 Kazu Hirata <kazu@codesourcery.com>
1698 * config/m68k/m68k.c (m68k_output_function_epilogue): Don't
1699 output a NOP for empty epilogues.
1701 2007-01-18 Richard Sandiford <richard@codesourcery.com>
1703 * config/m68k/m68k.c (m68k_use_return_insn): Update comments
1704 before function. Extend register save check to include all
1705 registers, not just integer ones.
1707 2007-01-18 Kazu Hirata <kazu@codesourcery.com>
1709 * config/m68k/m68k-protos.h (use_return_insn): Rename to...
1710 (m68k_use_return_insn): ...this.
1711 * config/m68k/m68k.h (USE_RETURN_INSN): Delete.
1712 * config/m68k/m68k.c (use_return_insn): Rename to...
1713 (m68k_use_return_insn): ...this.
1714 * config/m68k/m68k.md (return): Use m68k_use_return_insn instead
1717 2007-01-18 Nathan Sidwell <nathan@codesourcery.com>
1719 * config/m68k/fpgnulib.c (__truncdfsf2): Implement round to
1720 nearest even, fix denormal rounding overflow.
1722 2007-01-18 Richard Sandiford <richard@codesourcery.com>
1724 * config/m68k/m68k.md (movsf_cf_hard): Use fsmove instead of
1725 f%$smove and f%$move.
1726 (movdf_cf_hard): Use fdmove for cases 0 and 3 and fmove for case 1.
1727 (extendsfdf2_cf): Use fdmove instead of f%&move.
1728 (truncdfsf2_cf): Use fsmove instead of f%$smove.
1729 (add<mode>3_cf, sub<mode>3_cf): Use <FP:prec> instead of <FP:round>.
1731 2007-01-18 Richard Sandiford <richard@codesourcery.com>
1733 * config/m68k/m68k.md (movdf_cf_hard): Use output_move_double for
1736 2007-01-18 Richard Sandiford <richard@codesourcery.com>
1738 * real.h (real_format): Add a canonical_nan_lsbs_set field.
1739 (coldfire_single_format): Declare.
1740 (coldfire_double_format): Likewise.
1741 * real.c (encode_ieee_single): Use canonical_nan_lsbs_set instead
1742 of qnan_msb_set to determine the lower bits of a canonical
1744 (encode_ieee_double): Likewise.
1745 (encode_ieee_quad): Likewise.
1746 (ieee_single_format): Initialize canonical_nan_lsbs_set.
1747 (mips_single_format): Likewise.
1748 (ieee_double_format): Likewise.
1749 (mips_double_format): Likewise.
1750 (ieee_extended_motorola_format): Likewise.
1751 (ieee_extended_intel_96_format): Likewise.
1752 (ieee_extended_intel_128_format): Likewise.
1753 (ieee_extended_intel_96_round_53_format): Likewise.
1754 (ibm_extended_format): Likewise.
1755 (mips_extended_format): Likewise.
1756 (ieee_quad_format): Likewise.
1757 (mips_quad_format): Likewise.
1758 (vax_f_format): Likewise.
1759 (vax_d_format): Likewise.
1760 (vax_g_format): Likewise.
1761 (i370_single_format): Likewise.
1762 (i370_double_format): Likewise.
1763 (decimal_single_format): Likewise.
1764 (decimal_double_format): Likewise.
1765 (decimal_quad_format): Likewise.
1766 (c4x_single_format): Likewise.
1767 (c4x_extended_format): Likewise.
1768 (real_internal_format): Likewise.
1769 (coldfire_single_format): New real_format.
1770 (coldfire_double_format): Likewise.
1771 * config/pdp11/pdp11.c (pdp11_f_format): Initialize
1772 canonical_nan_lsbs_set.
1773 (pdp11_d_format): Likewise.
1774 * config/m68k/m68k.c (override_options): Override REAL_FORMAT_MODE
1775 if TARGET_COLDFIRE_CPU.
1777 2007-01-18 Richard Sandiford <richard@codesourcery.com>
1779 * config/m68k/m68k-protos.h (m68k_output_pic_call): Delete.
1780 (output_call, m68k_legitimize_call_address): Declare.
1781 * config/m68k/m68k.h (EXTRA_CONSTRAINT): Remove unnecessary
1782 parenthesees. Add support for a 'W' constraint.
1783 (LEGITIMATE_PIC_OPERAND_P): Remove SYMBOL_REF_FLAG handling.
1784 (PRINT_OPERAND_PUNCT_VALID_P): Remove comment about 'o'.
1785 (m68k_symbolic_call, m68k_symbolic_jump): Declare.
1786 * config/m68k/m68k.c (m68k_symbolic_call, m68k_symbolic_jump): New
1788 (override_options): Initialize them. Do not set flag_no_function_cse
1789 for TARGET_ID_SHARED_LIBRARY.
1790 (m68k_output_pic_call): Delete.
1791 (m68k_legitimize_call_address): New function.
1792 (print_operand): Remove the %o prefix. Handle the %p prefix.
1793 (output_call): New function.
1794 (m68k_output_mi_thunk): Use m68k_symbolic_jump. Always load the
1795 target address from the GOT if symbolic jumps are not allowed.
1796 * config/m68k/m68k.md (call, general_operand): Do not set
1797 SYMBOL_REF_FLAG. Use m68k_legitimize_call_address instead.
1798 Merge separate flag_pic and !flag_pic define_insns into...
1799 (*call, *call_value): ...these new patterns. Match the address
1800 rather than the containing MEM and require it to be a call_operand.
1801 Use output_call to generate the asm template.
1802 * config/m68k/predicates.md (const_call_operand): New predicate.
1803 (call_operand): Likewise.
1805 2007-01-18 Nathan Sidwell <nathan@codesourcery.com>
1807 * config/m68k/m68k.h (REGISTER_MOVE_COST): Simplify definition.
1808 (STACK_GROWS_DOWNWARD): Define to 1.
1809 (FUNCTION_VALUE, LIBCALL_VALUE, FUNCTION_VALUE_REGNO_P): Equivocate
1810 comments, emphasizing that these values are only defaults.
1811 * config/m68k/linux.h (LINK_SPEC): Fix formatting in #undef.
1812 * config/m68k/m68k.c (const_method): Remove trailing whitespace.
1814 2007-01-18 Richard Sandiford <richard@codesourcery.com>
1816 * config/m68k/m68k.md (cmpsi): Remove outdated flag_pic handling.
1818 2007-01-18 Kazu Hirata <kazu@codesourcery.com>
1819 Richard Sandiford <richard@codesourcery.com>
1821 * config/m68k/m68k.h (DATA_REGNO_P, ADDRESS_REGNO_P, INT_REGNO_P)
1822 (FP_REGNO_P): New macros.
1823 (REGNO_OK_FOR_INDEX_P, REGNO_OK_FOR_BASE_P, REGNO_OK_FOR_DATA_P)
1824 (REGNO_OK_FOR_FP_P, REG_OK_FOR_INDEX_P, REG_OK_FOR_BASE_P): Use them.
1825 (EH_RETURN_STACKADJ_RTX): Use A0_REG.
1826 * config/m68k/m68k.c (m68k_regno_mode_ok): Use the new REGNO macros.
1828 2007-01-18 Nathan Sidwell <nathan@codesourcery.com>
1830 * config.gcc (m68k-*-aout*, m68k-*-coff*, m68020-*-elf*, m68k-*-elf*)
1831 (m68k-*-uclinuxoldabi, m68k-*-uclinux*, m68k-*-rtems*): Add t-floatlib
1833 * config/m68k/t-floatlib: New file, extracting common code from...
1834 * config/m68k/t-m68kbare, config/m68k/t-m68kelf,
1835 * config/m68k/t-uclinux: Here.
1836 * config/m68k/fpgnulib.c: Do not compile extendeed precision
1837 routines on ColdFire targets.
1839 2007-01-18 Nathan Sidwell <nathan@codesourcery.com>
1841 * config.gcc (m68k-*-aout*, m68k-*-coff*, m68020-*-elf*, m68k-*-elf*)
1842 (m68010-*-netbsdelf*, m68k*-*-netbsdelf*, m68k*-*-openbsd*)
1843 (m68k-*-uclinuxoldabi, m68k-*-uclinux*, m68k-*-linux*)
1844 (m68k-*-rtems*): Use tm_file rather than m68k/m68k.h and
1845 explicitly set MOTOROLA to 1.
1846 * config/m68k/m68k.h (MOTOROLA): Simplify definition accordingly.
1848 2007-01-18 Richard Sandiford <richard@codesourcery.com>
1849 Nathan Sidwell <nathan@codesourcery.com>
1851 * config/m68k/m68k.h (PCC_STATIC_STRUCT_RETURN): Delete.
1852 (ASM_OUTPUT_REG_PUSH, ASM_OUTPUT_REG_POP): Add MOTOROLA cases.
1853 * config/m68k/coff.h (REGISTER_PREFIX_MD): Delete.
1854 * config/m68k/m68020-elf.h (LIB_SPEC): Delete.
1855 * config/m68k/m68k-none.h (CC1_SPEC, CPP_SUBTARGET_SPEC): Delete.
1856 * config/m68k/m68kelf.h (IMMEDIATE_PREFIX, REGISTER_PREFIX_MD)
1857 (ASM_OUTPUT_REG_PUSH, ASM_OUTPUT_REG_POP): Delete.
1858 (NO_DOLLAR_IN_LABEL, PCC_STATIC_STRUCT_RETURN): Don't undefine.
1859 (BSS_ASM_OP, ASM_OUTPUT_SKIP, ASM_OUTPUT_ASCII): Delete.
1860 * config/m68k/m68kemb.h (PCC_STATIC_STRUCT_RETURN): Don't undefine.
1861 * config/m68k/linux.h (SIZE_TYPE, PTRDIFF_TYPE, WCHAR_TYPE)
1862 (WCHAR_TYPE_SIZE, TARGET_OBJFMT_CPP_BUILTINS): Delete.
1863 (TARGET_OS_CPP_BUILTINS): Don't define mc68000 and mc68020 here.
1864 (DBX_REGISTER_NUMBER): Delete.
1865 * config/m68k/netbsd-elf.h (IMMEDIATE_PREFIX): Delete.
1866 (PCC_STATIC_STRUCT_RETURN): Don't undefine.
1867 * config/m68k/openbsd.h (PCC_STATIC_STRUCT_RETURN): Define.
1869 2007-01-18 Richard Sandiford <richard@codesourcery.com>
1871 * config.gcc (m68k-*-uclinux*): Add flat.h to $tm_file.
1872 * config/flat.h: New file.
1873 * crtstuff.c (USE_PT_GNU_EH_FRAME): Don't define if
1875 * config/m68k/m68k.h (ASM_PREFERRED_EH_DATA_FORMAT): Do not use
1876 indirect references for -msep-data or -mid-shared-library.
1877 Do not use PC-relative code addresses either.
1878 * config/m68k/m68k.c (override_options): Restrict -fPIC error
1880 * config/m68k/uclinux.h (STARTFILE_SPEC): Define. Use Scrt1.o
1881 for shared libraries and crt1.o for executables. Use crti.o and
1883 (ENDFILE_SPEC): Use crtend.o and crtn.o.
1884 (LIB_SPEC): Suppress -Rlibc.gdb if -static-libc is given.
1885 Do not add -elf2flt or -shared-lib-id options here.
1886 (LINK_SPEC): Define. Pass -elf2flt if no -elf2flt option is given.
1887 Pass -shared-lib-id if -mid-shared-library, taking the library
1888 identifier from -mshared-library-id if given, otherwise
1890 (EH_FRAME_IN_DATA_SECTION): Do not undefine.
1891 (INIT_SECTION_ASM_OP, FINI_SECTION_ASM_OP): Likewise.
1892 (TARGET_OS_CPP_BUILTINS): Define __GXX_MERGED_TYPEINFO_NAMES=0
1893 and __GXX_TYPEINFO_EQUALITY_INLINE=0 if -mid-shared-library.
1894 (DRIVER_SELF_SPECS): Map unadorned PIC options to -msep-data.
1895 * config/m68k/t-uclinux (EXTRA_MULTILIB_PARTS): Add crtbegin.o
1897 * config/m68k/lb1sf68.asm (PICCALL): Use an lea and pc-relative
1898 jump sequence for ISA A and ISA A+.
1899 (PICJUMP): Likewise.
1901 2007-01-18 Richard Sandiford <richard@codesourcery.com>
1903 * config.gcc (m68k-*-uclinux*): Base the port on the common
1904 and m68k GNU/Linux files rather than on the generic ELF ones.
1905 * config/m68k/uclinux.h (TARGET_VERSION): Override.
1906 (TARGET_OS_CPP_BUILTINS): Use LINUX_TARGET_OS_CPP_BUILTINS.
1908 2007-01-18 Julian Brown <julian@codesourcery.com>
1909 Richard Sandiford <richard@codesourcery.com>
1911 * config/m68k/m68k.h (LONG_DOUBLE_TYPE_SIZE): Make 64-bit on ColdFire.
1912 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
1913 * config/m68k/netbsd-elf.h (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Undefine
1915 * config/m68k/uclinux-oldabi.h (LONG_DOUBLE_TYPE_SIZE): Redefine to
1917 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
1919 2007-01-18 Richard Sandiford <richard@codesourcery.com>
1921 * doc/install.texi: Document m68k-uclinuxoldabi.
1922 * config.gcc (m68k-*-uclinuxoldabi): New configuration.
1923 * config/m68k/uclinux-oldabi.h: New file, copied from
1924 config/m68k/uclinux.h.
1926 2007-01-18 Kaz Kojima <kkojima@rr.iij4u.or.jp>
1928 * config/m32r/m32r-protos.h (m32r_expand_epilogue): Declare it.
1929 * config/m32r/m32r.c (m32r_setup_incoming_varargs): Use gen_frame_mem.
1930 (m32r_compute_frame_size): Use unsigned for regno.
1931 (m32r_reload_lr): Use gen_frame_mem.
1933 (m32r_output_function_epilogue): Don't output the function epilogue
1935 (m32r_expand_epilogue): New.
1936 (direct_return): Return false if the function has the interrupt
1938 (m32r_hard_regno_rename_ok): Remove code for the textual epilogue.
1939 * config/m32r/m32r.md (epilogue): New expander.
1940 (return_lr, return_rte): New insns.
1941 (return): Make it expander.
1942 (return_normal): New expander.
1944 2007-01-18 Josh Conner <jconner@apple.com>
1947 * config/rs6000/rs6000.c (rs6000_emit_vector_compare): Add
1948 support for UNLE, UNLT, UNGE, and UNGT.
1950 2007-01-18 Jan Hubicka <jh@suse.cz>
1952 * tree-vrp.c (finalize_jump_threads): Do not call cleanup_cfg by hand.
1953 * tree-tailcall (add_virtual_phis): Likewise.
1954 (optimize_tail_call): Return TODOs.
1955 (execute_tail_calls): Return TODOs.
1956 * tree-ssa-ccp (execute_fold_all_builtins): Do cleanup_cfg via TODO.
1957 * tree-cfgcleanup.c (cleanup_tree_cfg_loop): Return if something
1959 * tree-ssa-forwprop.c (tree_ssa_forward_propagate_single_use_value):
1960 Cleanup cfg using TODO.
1961 * tree-flow.h (cleanup_tree_cfg_loop): Update prototype.
1962 * passes.c (execute_function_todo): When cleanup did something, remove
1964 * tree-cfg.c (pass_build_cfg): Add cleanup_cfg TODO.
1965 (make_edges): Don't cleanup_cfg.
1967 2007-01-18 Uros Bizjak <ubizjak@gmail.com>
1969 * reg-stack.c (subst_stack_regs_pat) [UNSPEC_SINCOS_COS,
1970 UNSPEC_XTRACT_FRACT]: Use generic code for instructions that
1971 operate on the top of stack.
1972 [UNSPEC_SINCOS_SIN, UNSPEC_XTRACT_EXP, UNSPEC_TAN]: Rewrite
1973 register handling of instructions that output to the second
1975 [UNSPEC_TAN_ONE, UNSPEC_TAN_TAN]: Remove.
1976 (move_for_stack_reg): Special-case check for dead destination
1977 stack slot for constant load of 1.0 inside UNSPEC_TAN.
1979 * config/i386/i386.md (UNSPEC_TAN): New constant.
1980 (UNSPEC_TAN_ONE, UNSPEC_TAN_TAN): Remove.
1981 (fptanxf4_i387, fptan_extend<mode>xf4_i387): New patterns
1982 to correctly model move of constant 1.0 to top stack slot.
1983 (*tandf3_1, *tansf3_1, *tanxf3_1): Remove insn patterns.
1984 (unnamed peephole2 pattern): Remove corresponding peephole2
1985 pattern that optimizes tan insn and loading of constant 1.0.
1986 (tanxf2): Use fptanxf4_i387.
1987 (tan<mode>2): Rename from tansf2 and tandf2 and macroize insn
1988 patterns using X87MODEF12 mode macro. Use fptan_extend<mode>xf4_i387
1989 and truncate result to requested mode. Use SSE_FLOAT_MODE_P to
1990 disable patterns for SSE math.
1991 (sincos<mode>3): Use truncxf<mode>2_i387_noop for truncation.
1992 (fyl2x_extend<mode>xf3_i387): Use X87MODEF12 for operand 1.
1994 2007-01-18 Dirk Mueller <dmueller@suse.de>
1995 Richard Guenther <rguenther@suse.de>
1998 * doc/invoke.texi (Warray-bounds): Document -Warray-bounds.
1999 * common.opt (Warray-bounds): Add new warning option.
2000 * c-opts.c (c_common_handle_option): Define -Warray-bounds
2002 * Makefile.in: make tree-vrp.o depend on toplev.h
2003 * tree-vrp.c (vrp_finalize): Call check_array_refs if -Warray-bounds
2005 (check_array_refs, check_array_bounds, check_array_ref): New.
2007 2007-01-18 Jan Hubicka <jh@suse.cz>
2009 * tree-ssa-ccp.c (ccp_finalize): Return if something changed.
2010 (execute_ssa_ccp): Return flags conditionally.
2011 * tree-ssa-propagate.c (substitue_and_fold): Return if something was
2013 * tree-ssa-propagate.h (substitute_and_fold): Update prototype.
2015 2007-01-18 Steven Bosscher <steven@gcc.gnu.org>
2017 * cfgcleanup.c (cleanup_cfg): Detect cfglayout mode and set
2018 the CLEANUP_CFGLAYOUT flag when in cfglayout mode.
2020 * Makefile.c (GTFILES): Add cfglayout.h.
2021 * gengtype.c (open_base_files): Likewise.
2022 * cfglayout.c (cfg_layout_function_footer,
2023 cfg_layout_function_header) Reindent to make gengtype happy.
2024 * cfglayout.h (cfg_layout_function_footer,
2025 cfg_layout_function_header): Add GTY(()) marker.
2027 * ifcvt.c (noce_try_sign_mask): Make sure INSN_B is non-null.
2029 2007-01-18 Ben Elliston <bje@au.ibm.com>
2031 * genautomata.c (write_automata): Include xstrerror output in the
2032 error message if writing the DFA description file fails.
2034 2007-01-17 H.J. Lu <hongjiu.lu@intel.com>
2036 * config/mips/mips-protos.h (mips_output_external): Make it
2038 * config/mips/iris.h (TARGET_ASM_EXTERNAL_LIBCALL): Removed.
2039 * config/mips/mips.c (irix_output_external_libcall): Likewise.
2040 (extern_list): Likewise.
2041 (extern_head): Likewise.
2042 (TARGET_ASM_FILE_END): Likewise.
2043 (mips_file_end): Likewise.
2044 (mips_output_external): Rewritten.
2046 2007-01-18 Ben Elliston <bje@au.ibm.com>
2048 * genpreds.c (write_insn_preds_c): Only write out the function
2049 body for regclass_for_constraint if we have register constraints.
2051 2007-01-17 Tom Tromey <tromey@redhat.com>
2053 * doc/sourcebuild.texi (libgcj Tests): Use sourceware.org.
2054 * doc/install.texi (Testing): Use sourceware.org.
2055 (Binaries): Likewise.
2056 (Specific): Likewise.
2057 * doc/contrib.texi (Contributors): Use sourceware.org.
2059 2007-01-17 Anatoly Sokolov <aesok@post.ru>
2061 * config/avr/avr.h (AVR_HAVE_LPMX): New macro.
2062 (AVR_ENHANCED): Rename to ...
2063 (AVR_HAVE_MUL): ... new.
2064 (avr_enhanced_p): Rename to ...
2065 (avr_have_mul_p): ... new.
2066 (TARGET_CPU_CPP_BUILTINS): Use 'avr_have_mul_p' instead of
2067 'avr_enhanced_p' for "__AVR_ENHANCED__". Define "__AVR_HAVE_MUL__".
2068 * config/avr/avr.c (avr_enhanced_p): Rename to ...
2069 (avr_have_mul_p): ... new.
2070 (base_arch_s): Rename 'enhanced' to 'have_mul'.
2071 (avr_override_options): Use 'avr_have_mul_p' and 'have_mul' instead of
2072 'avr_enhanced_p' and 'enhanced'.
2073 (ashlhi3_out, ashrhi3_out, lshrhi3_out, avr_rtx_costs): Use
2074 AVR_HAVE_MUL instead of AVR_ENHANCED.
2075 * avr.md (*tablejump_enh): Use AVR_HAVE_LPMX instead of AVR_ENHANCED.
2076 (mulqi3, *mulqi3_enh, *mulqi3_call, mulqihi3, umulqihi3, mulhi3,
2077 *mulhi3_enh, *mulhi3_call, mulsi3, *mulsi3_call): Use AVR_HAVE_MUL
2078 instead of AVR_ENHANCED.
2079 (*tablejump_enh): Use AVR_HAVE_LPMX instead of AVR_ENHANCED.
2080 * libgcc.S: Use __AVR_HAVE_MUL__ instead of __AVR_ENHANCED__.
2081 (__tablejump__): Use __AVR_HAVE_LPMX__ instead of __AVR_ENHANCED__.
2083 2007-01-17 Ian Lance Taylor <iant@google.com>
2085 * vec.h (VEC_reserve_exact): Define.
2086 (vec_gc_p_reserve_exact): Declare.
2087 (vec_gc_o_reserve_exact): Declare.
2088 (vec_heap_p_reserve_exact): Declare.
2089 (vec_heap_o_reserve_exact): Declare.
2090 (VEC_OP (T,A,reserve_exact)): New static inline function, three
2092 (VEC_OP (T,A,reserve)) [all versions]: Remove handling of
2094 (VEC_OP (T,A,alloc)) [all versions]: Call ...reserve_exact.
2095 (VEC_OP (T,A,copy)) [all versions]: Likewise.
2096 (VEC_OP (T,a,safe_grow)) [all versions]: Likewise.
2097 * vec.c (calculate_allocation): Add exact parameter. Change all
2099 (vec_gc_o_reserve_1): New static function, from vec_gc_o_reserve.
2100 (vec_gc_p_reserve, vec_gc_o_reserve): Call vec_gc_o_reserve_1.
2101 (vec_gc_p_reserve_exact, vec_gc_o_reserve_exact): New functions.
2102 (vec_heap_o_reserve_1): New static function, from vec_heap_o_reserve.
2103 (vec_heap_p_reserve, vec_heap_o_reserve): Call vec_heap_o_reserve_1.
2104 (vec_heap_p_reserve_exact): New function.
2105 (vec_heap_o_reserve_exact): New function.
2107 2007-01-17 Jan Hubicka <jh@suse.cz>
2109 * ipa-type-escape.c (look_for_casts): Revamp using handled_component_p.
2111 2007-01-17 Eric Christopher <echristo@apple.com>
2113 * config.gcc: Support core2 processor.
2115 2007-01-16 Jan Hubicka <jh@suse.cz>
2117 * tree-ssanames.c (release_dead_ssa_names): Instead of ggc_freeing
2118 the names, just unlink the chain so we don't crash on dangling pointers
2121 2007-01-16 Jan Hubicka <jh@suse.cz>
2123 * cgraph.h (cgraph_decide_inlining_incrementally): Kill.
2124 * tree-pass.h: Reorder to make IPA passes appear toegher.
2125 (pass_early_inline, pass_inline_parameters, pass_apply_inline): Declare.
2126 * cgraphunit.c (cgraph_finalize_function): Do not compute inling
2127 parameters, do not call early inliner.
2128 * ipa-inline.c: Update comments. Include tree-flow.h
2129 (cgraph_decide_inlining): Do not compute inlining parameters.
2130 (cgraph_decide_inlining_incrementally): Return TODOs; assume to
2131 be called with function context set up.
2132 (pass_ipa_inline): Remove unreachable functions before pass.
2133 (cgraph_early_inlining): Simplify assuming to be called from the
2135 (pass_early_inline): New pass.
2136 (cgraph_gate_ipa_early_inlining): New gate.
2137 (pass_ipa_early_inline): Turn into simple wrapper.
2138 (compute_inline_parameters): New function.
2139 (gate_inline_passes): New gate.
2140 (pass_inline_parameters): New pass.
2141 (apply_inline): Move here from tree-optimize.c
2142 (pass_apply_inline): New pass.
2143 * ipa.c (cgraph_remove_unreachable_nodes): Verify cgraph after
2145 * tree-inline.c (optimize_inline_calls): Return TODOs rather than
2147 (tree_function_versioning): Do not allocate dummy struct function.
2148 * tree-inline.h (optimize_inline_calls): Update prototype.
2149 * tree-optimize.c (execute_fixup_cfg): Export.
2150 (pass_fixup_cfg): Remove
2151 (tree_rest_of_compilation): Do not apply inlines.
2152 * tree-flow.h (execute_fixup_cfg): Declare.
2153 * Makefile.in (gt-passes.c): New.
2154 * passes.c: Include gt-passes.h
2155 (init_optimization_passes): New passes.
2156 (nnodes, order): New static vars.
2157 (do_per_function_toporder): New function.
2158 (execute_one_pass): Dump current pass here.
2159 (execute_ipa_pass_list): Don't dump current pass here.
2161 2007-01-16 Janis Johnson <janis187@us.ibm.com>
2163 * config/dfp-bit.c (dfp_compare_op): Return separate value for NaN.
2164 (DFP_NE, DFP_LE, DFP_GE): Return false for NaN.
2166 2007-01-16 David Edelsohn <edelsohn@gnu.org>
2168 * config/rs6000/darwin-ldouble.c: Build file for SOFT_FLOAT.
2169 (strong_alias): Define.
2170 (__gcc_qmul): Provide non-FMA for soft-float.
2176 (__gcc_qunord): New.
2186 * config/rs6000/rs6000.c (rs6000_init_libfuncs): Initialize
2187 soft-float functions.
2188 * config/rs6000/libgcc-ppc-glibc.ver: Version soft-float symbols.
2189 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Do not warn
2190 about long double soft float.
2192 2007-01-16 Dorit Nuzman <dorit@il.ibm.com>
2193 Tehila Meyzels <tehila@il.ibm.com>
2195 * tree-vectorizer.h (is_pattern_stmt_p): New.
2196 * tree-vect-analyze.c (vect_determine_vectorization_factor): Fix
2197 formatting (tabs instead of spaces). Cleanup and clarify setting
2198 of STMT_VINFO_VECTYPE. Call is_pattern_stmt_p.
2199 * tree-vect-transform.c (vect_get_vec_def_for_operand): Fix typo.
2200 (vectorizable_type_demotion): Check that types are integral.
2201 (vectorizable_type_promotion): Likewise.
2202 (vectorizable_store): Fix typo. Eliminate new-line at end of
2205 2007-01-16 Jan Hubicka <jh@suse.cz>
2207 * tree-ssanames.c (release_dead_ssa_names): Remove invalidated
2210 2007-01-15 Eric Christopher <echristo@apple.com>
2212 * ifcvt.c: Include vec.h, vecprim.h.
2213 (check_cond_move_block): New argument regs.
2214 Reorganize. Add registers used to regs.
2215 (cond_move_process_if_block): Use regs set above as
2218 2007-01-15 Eric Christopher <echristo@apple.com>
2220 * config/darwin.h: Update copyright.
2221 (TARGET_OPTION_TRANSLATE_TABLE): Add umbrella.
2222 (LINK_COMMAND_SPEC): Add -u.
2223 (LINK_SPEC): Fix umbrella for above.
2225 2007-01-15 Joseph S. Myers <joseph@codesourcery.com>
2227 * config/soft-fp/op-common.h, config/soft-fp/op-4.h: Update from
2230 2007-01-15 Tom Tromey <tromey@redhat.com>
2232 * doc/sourcebuild.texi (libgcj Tests): Don't mention jacks.
2233 * doc/install.texi (Testing): Don't mention jacks.
2234 (Configuration): Document --enable-java-maintainer-mode. Move
2235 --with-java-home to libgcj-specific section. Document
2237 (Prerequisites): Mention --enable-java-maintainer-mode, ecj1.
2239 2007-01-15 Jan Hubicka <jh@suse.cz>
2241 * tree-ssa-dce.c (DCE_TODOs): New.
2242 (propagate_necessity): Return if something changed.
2243 (eliminate_unnecessary_stmts): Likewise.
2244 (perform_tree_ssa_dce): Return TODO flags when needed.
2245 (pass_dce, pass_dce_loop, pass_cd_dce): Remove TODO flags.
2247 2007-01-15 Uros Bizjak <ubizjak@gmail.com>
2249 * config/i386/i386.md (fyl2xxf3_i387): Rename from fyl2x_xf3.
2250 (fyl2x_extend<mode>xf3_i387): New insn pattern.
2251 (log<mode>2): Rename from logsf2 and logdf2 and macroize insn
2252 patterns using X87MODEF12 mode macro. Extend operand 1
2253 to XFmode. Use SSE_FLOAT_MODE_P to disable patterns for SSE math.
2254 (log10<mode>2): Ditto.
2255 (log2<mode>2): Ditto.
2256 (log1p<mode>2): Ditto.
2257 (logb<mode>2): Ditto.
2258 (fyl2xp1xf3_i387): Rename from fyl2xp1_xf3.
2259 (fyl2xp1_extend<mode>xf3_i387): New insn pattern.
2260 (*fxtractxf3_i387): Rename from *fxtractxf3.
2261 (fxtract_extend<mode>xf3_i387): New insn pattern.
2262 (ilogbsi2): Use match_dup 3, not match_operand:XF 3.
2263 * config/i386/i386.c (ix86_emit_i387_log1p): Use gen_fyl2xp1xf3_i387()
2264 and gen_fyl2xxf3_i387().
2266 2007-01-14 Zdenek Dvorak <dvorakz@suse.cz>
2268 * loop-unswitch.c (unswitch_loop): Do not call fix_loop_placement.
2269 * cfgloopmanip.c (fix_loop_placement): Made static. Use
2270 get_loop_exit_edges. Changed return type to bool.
2271 * cfgloop.h (fix_loop_placement): Declaration removed.
2273 2007-01-14 Dorit Nuzman <dorit@il.ibm.com>
2275 * param.h (MIN_VECT_LOOP_BOUND): New.
2276 * params.def (MIN_VECT_LOOP_BOUND): New.
2277 * tree-vectorizer.c (slpeel_tree_peel_loop_to_edge): Takes another
2278 argument - minimum threshold for number of iterations.
2279 * tree-vectorizer.h (slpeel_tree_peel_loop_to_edge): Add another
2280 argument to declaration.
2281 * tree-vect-analyze.c (vect_analyze_operations): Check value of
2282 MIN_VECT_LOOP_BOUND.
2283 * tree-vect-transform.c (vect_do_peeling_for_loop_bound): Call
2284 slpeel_tree_peel_loop_to_edge with additional argument.
2285 (vect_do_peeling_for_alignment): Likewise.
2286 * doc/invoke.texi (min-vect-loop-bound): Document new param option.
2288 2007-01-14 Uros Bizjak <ubizjak@gmail.com>
2291 * config/i386/i386.c (print_operand) ['z']: Output 'b' for
2294 2007-01-14 Jan Hubicka <jh@suse.cz>
2296 * tree-dfa.c (remove_referenced_var): New function.
2297 * tree-ssa-live.c (remove_unused_locals): Walk referenced vars and
2298 prune referenced vars list too.
2299 * tree-flow.h (remove_referenced_var): Declare.
2301 2007-01-14 Jan Hubicka <jh@suse.cz>
2303 * tree-eh.c (add_stmt_to_eh_region_fn): Do not add call_exprs
2305 (remove_stmt_from_eh_region_fn): Do not remove call_exprs.
2306 (verify_eh_throw_stmt_node, verify_eh_throw_table_statements): Kill.
2307 * except.h (verify_eh_throw_table_statements): Kill prototype.
2308 * cfgexpand.c (expand_gimple_basic_block): Propagate Eh regions
2309 into call exrepssions.
2310 * tree-optimize.c (execute_free_cfg_annotatiosn): Do not call
2312 * tree-cfg.c: Include pointer-set.h.
2313 (verify_node_sharing): Work on pointer set.
2314 (verify_eh_throw_stmt_node): New.
2315 (verify_stmts): Use pointers sets, verify throw_stmt.
2317 2007-01-13 Zdenek Dvorak <dvorakz@suse.cz>
2319 * ipa-reference.c (analyze_function): Consider also addresses taken
2322 2007-01-12 Roger Sayle <roger@eyesopen.com>
2324 * c-typeck.c (null_pointer_constant_p): Replace use of
2325 TREE_CONSTANT_OVERFLOW with TREE_OVERFLOW.
2326 (build_c_cast): Likewise.
2328 2007-01-12 Roger Sayle <roger@eyesopen.com>
2330 * tree.h (force_fit_type_double): Remove unused final argument.
2331 * c-common.c (constant_expression_warning): Replace use of
2332 TREE_CONSTANT_OVERFLOW with TREE_OVERFLOW.
2333 (convert_and_check): Likewise.
2334 (shorten_compare): Update call to force_fit_type_double.
2335 (c_common_truthvalue_conversion) <INTEGER_CST>: Use integer_zerop.
2336 * convert.c (convert_to_pointer): Update call to
2337 force_fit_type_double.
2338 * fold-const.c (force_fit_type_double): Remove overflowed_const
2340 (int_const_binop, fold_convert_const_int_from_int,
2341 fold_convert_const_int_from_real, fold_div_compare,
2342 fold_sign_changed_comparison, fold_unary, fold_negate_const,
2343 fold_abs_const, fold_not_const): Remove the final argument from
2344 calls to force_fit_type_double.
2346 2007-01-12 Andrew Pinski <andrew_pinski@playstation.sony.com>
2348 * configure.ac: Set insn to "nop" for spu-*-* also.
2349 * configure: Regenerate.
2351 2007-01-12 Olga Golovanevsky <olga@il.ibm.com>
2353 * builtins.def : Add BUILT_IN_FREE.
2355 2007-01-12 Jan Hubicka <jh@suse.cz>
2357 PR tree-optimization/30443
2358 * tree-inline.c (tree_function_versioning): Do not optimize when
2359 cloning for inlining.
2361 2007-01-12 Zdenek Dvorak <dvorakz@suse.cz>
2363 * doc/loop.texi: Document recording of loop exits.
2364 * cfgloopmanip.c (loopify, duplicate_loop): Use alloc_loop.
2365 (update_single_exits_after_duplication,
2366 update_single_exit_for_duplicated_loop,
2367 update_single_exit_for_duplicated_loops): Removed.
2368 (duplicate_loop_to_header_edge): Do not call
2369 update_single_exits_after_duplication and
2370 update_single_exit_for_duplicated_loops.
2371 (loop_version): Do not update single_exit information.
2372 (fix_loop_structure): Use record_loop_exits instead of
2373 mark_single_exit_loops.
2374 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update
2375 the lists of loop exits.
2376 * cfghooks.c (redirect_edge_and_branch, redirect_edge_and_branch_force,
2377 split_edge, merge_blocks): Update the lists of loop exits.
2378 * modulo-sched.c (sms_schedule): Pass LOOPS_HAVE_RECORDED_EXITS to
2379 loop_optimizer_init.
2380 * loop-init.c (loop_optimizer_init): Call record_loop_exits instead
2381 of mark_single_exit_loops.
2382 (loop_optimizer_finalize): Call release_recorded_exits.
2383 * tree-ssa-loop.c (tree_loop_optimizer_init): Pass
2384 LOOPS_HAVE_RECORDED_EXITS to loop_optimizer_init.
2385 * tree-vectorizer.c (slpeel_tree_duplicate_loop_to_edge_cfg): Do not
2386 update single exit information.
2387 * lambda-code.c (perfect_nestify): Ditto.
2388 * cfgloop.c (flow_loop_free): Destroy the list of exits of the loop.
2389 (mark_single_exit_loops): Removed.
2390 (alloc_loop, loop_exit_hash, loop_exit_eq, loop_exit_free,
2391 get_exit_descriptions, rescan_loop_exit, record_loop_exits,
2392 dump_recorded_exit, dump_recorded_exits, release_recorded_exits): New
2394 (get_loop_exit_edges, single_exit): Use recorded exit lists.
2395 (add_bb_to_loop, remove_bb_from_loops): Update the lists of loop exits.
2396 (verify_loop_structure): Verify consistency of the exit lists.
2397 (flow_loops_find): Use alloc_loop. Initialize exits hash.
2398 (set_single_exit): Removed.
2399 * cfgloop.h (struct loop_exit): New function.
2400 (struct loop): single_exit_ field replaced by exits field.
2401 (LOOPS_HAVE_MARKED_SINGLE_EXITS): Replaced by LOOPS_HAVE_RECORDED_EXITS.
2402 (struct loops): Added exits hash.
2403 (mark_single_exit_loops, set_single_exit): Declaration removed.
2404 (release_recorded_exits, record_loop_exits, rescan_loop_exit): Declare.
2406 2007-01-12 Richard Sandiford <richard@codesourcery.com>
2408 * doc/invoke.texi: Avoid use of @headitem.
2410 2007-01-12 Richard Sandiford <richard@codesourcery.com>
2412 * cse.c (cse_insn): Move HAVE_CC0 code after declarations.
2414 2007-01-12 Richard Sandiford <richard@codesourcery.com>
2416 * doc/install.texi: Fix m68k-*-* anchor and add m68k-*-* to the
2419 2007-01-12 Nathan Sidwell <nathan@codesourcery.com>
2420 Richard Sandiford <richard@codesourcery.com>
2422 * doc/invoke.texi: Document -mno-bitfield, -mno-rtd and -mno-short.
2423 * config/m68k/m68k.opt: Resort options.
2424 (mbitfield, mrtd, mshort): Remove RejectNegative properties.
2426 2007-01-12 Nathan Sidwell <nathan@codesourcery.com>
2427 Richard Sandiford <richard@codesourcery.com>
2429 * doc/invoke.texi: Document the macros that are defined by
2430 m68k's -mtune and -mhard-float options.
2431 * config/m68k/m68k-protos.h (m68k_cpp_cpu_ident) Declare.
2432 (m68k_cpp_cpu_family): Likewise.
2433 * config/m68k/m68k.h (TARGET_CPU_CPP_BUILTINS): Add a full set
2434 of __ucfv*__ macros. Define __mcffpu__ if generating code for
2435 ColdFire FPUs. Define __mcf_cpu_* and __mcf_family_* macros.
2436 * config/m68k/m68k.c (m68k_cpp_cpu_ident): New function.
2437 (m68k_cpp_cpu_family): Likewise.
2439 2007-01-12 Richard Sandiford <richard@codesourcery.com>
2441 * config/m68k/m68k.h (TARGET_CPU_CPP_BUILTINS): Treat all mc68*
2442 macros besides mc68000 as tuning macros. Use a switch statement
2443 to set them and mcpu32.
2445 2007-01-12 Julian Brown <julian@codesourcery.com>
2447 * config/m68k/m68k.h: Use TARGET_68040 instead of TARGET_68040_ONLY.
2448 (TARGET_68040_ONLY): Rename to...
2449 (TARGET_68040): ...this.
2450 * config/m68k/m68k.c: Use TARGET_68040 instead of TARGET_68040_ONLY.
2451 * config/m68k/m68k.md: Likewise.
2453 2007-01-12 Julian Brown <julian@codesourcery.com>
2454 Nathan Sidwell <nathan@codesourcery.com>
2455 Richard Sandiford <richard@codesourcery.com>
2457 * config.gcc (m680[012]0-*-*, m68k*-*-*): Set m68k_cpu_ident to
2458 the -mcpu= argument associated with the --with-cpu setting.
2459 Define M68K_DEFAULT_TUNE to the default -mtune= option,
2460 if different from the one implied by the -mcpu setting.
2461 Accept --with-cpu=FOO if FOO is listed in m68k-devices.def,
2462 using mcpu=FOO as the default CPU option. Set target_cpu_default2.
2463 * doc/invoke.texi: Mention ColdFire in the introduction to the
2464 m68k options. Document the new -march, -mcpu, -mtune, -mdiv,
2465 -mno-div and -mhard-float options. Make -m68881 a synonym for
2466 -mhard-float. Document the previously-undocumented -m5206e,
2467 -m528x, -m5307 and -m5407 options. Tweak the existing option
2468 documentation for consistency.
2469 * doc/install.texi: Mention new --with-cpu arguments.
2470 * config/m68k/m68k.h (OPTION_DEFAULT_SPECS): Only use the
2471 default CPU if neither -mcpu nor -march are specified.
2472 (ASM_CPU_SPEC): Pass down -mcpu and -march options.
2473 (TARGET_CPU_CPP_BUILTINS): Set __mcfisa*__ macros from
2474 TARGET_ISA*. Set the legacy __mcf*__ cpu macros in the same way,
2475 using m68k_tune to decide between families that implement the
2476 same ISA. Use m68k_tune to set __mcfv4e__.
2477 (FL_BITFIELD, FL_68881, FL_COLDFIRE, FL_CF_HWDIV, FL_CF_MAC)
2478 (FL_CF_EMAC, FL_CF_EMAC_B, FL_CF_USP, FL_CF_FPU, FL_ISA_68000)
2479 (FL_ISA_68010, FL_ISA_68020, FL_ISA_68040, FL_ISA_A, FL_ISA_B)
2480 (FL_ISA_C, FL_ISA_MMU): New macros.
2481 (MASK_COLDFIRE): Delete.
2482 (TARGET_68010, TARGET_68020, TARGET_68040_ONLY, TARGET_COLDFIRE)
2483 (TARGET_ISAB): Redefine in terms of m68k_cpu_flags.
2484 (TARGET_68881, TARGET_COLDFIRE_FPU): Redefine in terms of m68k_fpu.
2485 (TARGET_HARD_FLOAT): Do not define here.
2486 (TARGET_ISAAPLUS, TARGET_ISAC): New macros.
2487 (TUNE_68000): New macro.
2488 (TUNE_68000_10): Redefine in terms of TUNE_68000 and TUNE_68010.
2489 (TUNE_68010, TUNE_68030, TUNE_68040, TUNE_68060, TUNE_CPU32)
2490 (TUNE_CFV2): Redefine in terms of m68k_tune.
2491 (uarch_type, target_device, fpu_type): New enums.
2492 (m68k_cpu, m68k_tune, m68k_fpu, m68k_cpu_flags): Declare.
2493 * config/m68k/m68k.c (TARGET_DEFAULT): Remove MASK_68881.
2494 (FL_FOR_isa_00, FL_FOR_isa_10, FL_FOR_isa_20, FL_FOR_isa_40)
2495 (FL_FOR_isa_cpu32, FL_FOR_isa_a, FL_FOR_isa_aplus, FL_FOR_isa_b)
2496 (FL_FOR_isa_c): New macros.
2497 (m68k_isa): New enum.
2498 (m68k_target_selection): New structure.
2499 (all_devices, all_isas, all_microarchs): New tables.
2500 (m68k_cpu_entry, m68k_arch_entry, m68k_tune_entry, m68k_cpu)
2501 (m68k_tune, m68k_fpu, m68k_cpu_flags): New variables.
2502 (MASK_ALL_CPU_BITS): Delete.
2503 (m68k_find_selection): New function.
2504 (m68k_handle_option): Handle -mcpu=, -march= and -mtune=.
2505 Map the legacy target options to a combination of the new ones.
2506 (override_options): Set m68k_cpu, m68k_tune, m68k_fpu and
2507 m68k_cpu_flags. Handle M68K_DEFAULT_TUNE. Use m68k_cpu_flags
2508 to derive default MASK_BITFIELD, MASK_CF_HWDIV and MASK_HARD_FLOAT
2510 * config/m68k/m68k.opt (m5200, m5206e, m528x, m5307, m5407, mcfv4e)
2511 (m68010, m68020, m68020-40, m68020-60, m68030, m68040): Remove Mask
2513 (m68881, msoft-float): Change mask from 68881 to HARD_FLOAT.
2514 (march=, mcpu=, mdiv, mhard-float, mtune=): New options.
2515 * config/m68k/m68k-devices.def: New file.
2517 2007-01-12 Richard Sandiford <richard@codesourcery.com>
2518 Nathan Sidwell <nathan@codesourcery.com>
2520 * config/m68k/m68k.h (ASM_CPU_SPEC, ASM_SPEC, EXTRA_SPECS)
2521 (SUBTARGET_EXTRA_SPECS): New macros.
2522 * config/m68k/linux.h (ASM_SPEC): Remove CPU flags;
2523 use %(asm_cpu_spec) instead.
2524 * config/m68k/m68k-none.h (ASM_SPEC): Likewise.
2525 * config/m68k/openbsd.h (ASM_SPEC): Likewise.
2526 * config/m68k/netbsd-elf.h (ASM_SPEC): Likewise.
2527 (EXTRA_SPECS): Rename to...
2528 (SUBTARGET_EXTRA_SPECS): ...this.
2530 2007-01-12 Nathan Sidwell <nathan@codesourcery.com>
2531 Richard Sandiford <richard@codesourcery.com>
2532 Julian Brown <julian@codesourcery.com>
2534 * config.gcc (m68k-*-aout*, m68k-*-coff*, m68020-*-elf*, m68k-*-elf*)
2535 (m68k-*-uclinux*, m68k-*-linux*, m68k-*-rtems*): Set default_m68k_cpu
2536 to the configuration's default CPU.
2537 (m68010-*-netbsdelf*, m68k*-*-netbsdelf*, m68k*-*-openbsd*): Likewise.
2538 Remove default masks.
2539 (m680[012]0-*-*): Set the default with_cpu to the first part of
2541 (m68k*-*-*): Set the default with_cpu to m$default_m68k_cpu.
2542 (m68k*-*-linux): Extend the --with-cpu handling to...
2543 (m680[012]0-*-*, m68k*-*-*): ...these configurations. Allow m68000
2544 and m68010. Don't set target_cpu_default2.
2545 * doc/install.texi: Document --with-cpu for m68k.
2546 * config/m68k/m68k.h (OPTION_DEFAULT_SPECS): Define.
2547 * config/m68k/m68k-none.h (TARGET_CPU_DEFAULT, M68K_CPU_m68k)
2548 (M68K_CPU_m68000, M68K_CPU_m68010, M68K_CPU_m68020, M68K_CPU_m68030)
2549 (M68K_CPU_m68040, M68K_CPU_m68302, M68K_CPU_m68332, TARGET_DEFAULT)
2550 (ASM_CPU_DEFAULT_SPEC, CC1_CPU_DEFAULT_SPEC): Delete.
2551 (ASM_SPEC): Remove use of %(asm_cpu_default).
2552 (EXTRA_SPECS, SUBTARGET_EXTRA_SPECS, MULTILIB_DEFAULTS): Delete.
2553 * config/m68k/linux.h (TARGET_DEFAULT): Delete.
2554 (CPP_SPEC): Merge definitions. Do not handle __HAVE_68881__ here.
2555 * config/m68k/netbsd-elf.h (TARGET_OS_CPP_BUILTINS): Define
2556 __HAVE_FPU__ if TARGET_HARD_FLOAT.
2557 (TARGET_DEFAULT): Delete.
2558 (EXTRA_SPECS): Delete cpp_cpu_default_spec, cpp_cpu_spec,
2559 cpp_fpu_spec, asm_default_spec and netbsd_cpp_spec.
2560 (CPP_CPU_SPEC): Delete.
2561 (TARGET_VERSION): Merge definitions, using TARGET_68010 to pick
2562 the appropriate string.
2563 (CPP_CPU_DEFAULT_SPEC, ASM_DEFAULT_SPEC, CPP_FPU_SPEC): Delete.
2564 (CPP_SPEC): Define to NETBSD_CPP_SPEC.
2565 (ASM_SPEC): Don't use %(asm_default_spec).
2566 * config/m68k/m68k.c (TARGET_DEFAULT_TARGET_FLAGS): Remove
2567 TARGET_DEFAULT and add MASK_68881.
2568 * config/m68k/m68k.md: Remove mention of TARGET_DEFAULT from comments.
2570 2007-01-12 Richard Sandiford <richard@codesourcery.com>
2572 * config.gcc (m68010-*-netbsdelf*): Add MASK_68010.
2573 (m68k*-*-netbsdelf*, m68k*-*-openbsd*, m68k*-linux*): Add
2574 MASK_68010 alongside MASK_68020.
2575 * doc/invoke.texi: Document -m68010.
2576 * config/m68k/m68k.opt (m68010): New.
2577 * config/m68k/m68k.h (TARGET_CPU_CPP_BUILTINS): Define mc68010
2579 (TUNE_68010): New macro.
2580 * config/m68k/m68k-none.h (M68K_CPU_m68k, M68K_CPU_m68010)
2581 (M68K_CPU_m68020, M68K_CPU_m68030, M68K_CPU_m68040)
2582 (M68K_CPU_m68332): Add MASK_68010.
2583 * config/m68k/linux.h (TARGET_DEFAULT): Add MASK_68010 to
2584 fallback definition.
2585 * config/m68k/netbsd-elf.h (CPP_CPU_SPEC): Remove now-redundant
2587 * config/m68k/m68k.c (MASK_ALL_CPU_BITS): Add MASK_68010.
2588 (m68k_handle_option): Handle OPT_m68010. Add MASK_68010
2589 to all entries that use MASK_68020.
2590 (output_move_simode_const, output_move_himode, output_move_qimode)
2591 (output_move_stricthi, output_move_strictqi): Use TARGET_68010
2592 instead of TARGET_68020 to select clr behavior. Remove comment
2593 about there being no TARGET_68010.
2594 * config/m68k/m68k.md: Likewise throughout.
2596 2007-01-12 Julian Brown <julian@codesourcery.com>
2598 * config/m68k/m68k.h (TARGET_ISAB): New macro.
2599 * config/m68k/m68k.c: Use TARGET_ISAB rather than TARGET_CFV4.
2600 * config/m68k/m68k.md: Likewise.
2602 2007-01-12 Julian Brown <julian@codesourcery.com>
2604 * config/m68k/m68k.h (LEGITIMATE_INDEX_P, LEGITIMIZE_ADDRESS): Use
2605 TARGET_COLDFIRE_FPU instead of TARGET_CFV4E.
2607 2007-01-12 Julian Brown <julian@codesourcery.com>
2609 * config/m68k/m68k.h (TUNE_68040_60): New macro.
2610 * config/m68k/m68k.c (standard_68881_constant_p): Use it.
2611 * config/m68k/m68k.md: Likewise.
2613 2007-01-12 Julian Brown <julian@codesourcery.com>
2614 Richard Sandiford <richard@codesourcery.com>
2616 * config/m68k/m68k.h (TARGET_CPU_CPP_BUILTINS): Use TUNE_68030
2617 instead of TARGET_68030, TUNE_68040 instead of TARGET_68040,
2618 TUNE_68060 instead of TARGET_68060 and TUNE_CPU32 instead of
2620 (TARGET_CPU32): Rename to...
2621 (TUNE_CPU32): ...this.
2622 (TUNE_68000_10, TUNE_68030, TUNE_68040, TUNE_68060)
2623 (TUNE_CFV2): New macros.
2624 * config/m68k/netbsd-elf.h (LONG_DOUBLE_TYPE_SIZE): Simplify;
2625 remove conditions that are implied by TARGET_68020.
2626 * config/m68k/m68k.c (m68k_output_function_prologue): Use TUNE_68040
2627 instead of TARGET_68040 and TUNE_CPU32 instead of TARGET_CPU32.
2628 (m68k_output_function_epilogue): Likewise.
2629 (m68k_rtx_costs): Likewise. Use TUNE_68060 instead of TARGET_68060
2630 and TUNE_CFV2 instead of TARGET_5200. Use TUNE_68000_10 instead of
2631 "!TARGET_68020 && !TARGET_COLDFIRE" to choose between 68000 and
2632 non-68000 timings. Refactor multiplication and division costs.
2633 (output_addsi3): Use TUNE_68040 instead of TARGET_68040 and
2634 TUNE_CPU32 instead of TARGET_CPU32.
2635 (standard_68881_constant_p): Use TUNE_68040 instead of TARGET_68040
2636 and TUNE_68060 instead of TARGET_68060.
2637 * config/m68k/m68k.md: Use TUNE_68040 instead of TARGET_68040,
2638 TUNE_68060 instead of TARGET_68060, and TUNE_CPU32 instead of
2640 (movsi_const0): Use TUNE_68000_10 rather than "!TARGET_68020
2641 && !TARGET_COLDFIRE" to choose between moveq and clr.
2642 Likewise in the unnamed movsf pattern.
2643 (ashlsi_17_24, lshrsi_17_24): Guard with TUNE_68000_10 rather than
2644 "!TARGET_68020 && !TARGET_COLDFIRE". Likewise the unnamed
2647 2007-01-12 Richard Sandiford <richard@codesourcery.com>
2649 * config/m68k/m68k.h (TARGET_CPU_CPP_BUILTINS): Increase amount
2650 of tabbing before backslashes.
2652 2007-01-11 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
2654 * pa-linux.h (ASM_OUTPUT_INTERNAL_LABEL): Undefine.
2655 * pa.h (ASM_OUTPUT_LABEL): Output colon when using GAS.
2656 (ASM_OUTPUT_INTERNAL_LABEL): Define.
2658 2007-01-11 Zdenek Dvorak <dvorakz@suse.cz>
2660 * tree-ssa-loop-ivopts.c (extract_cond_operands): Split from
2661 find_interesting_uses_cond.
2662 (find_interesting_uses_cond): Use extract_cond_operands.
2663 (rewrite_use_compare): Use extract_cond_operands and
2664 force_gimple_operand_bsi. Do not call update_stmt.
2665 (determine_use_iv_cost_condition): Use extract_cond_operands.
2666 Return cheaper of using original bound and changing the exit bound.
2668 2007-01-11 Zdenek Dvorak <dvorakz@suse.cz>
2670 PR tree-optimization/29516
2671 * tree-ssa-address.c (tree_mem_ref_addr, add_to_parts,
2672 most_expensive_mult_to_index, addr_to_parts,
2673 create_mem_ref, maybe_fold_tmr): Make the type of
2674 fields of TARGET_MEM_REF sizetype.
2675 (move_fixed_address_to_symbol, move_pointer_to_base):
2677 * tree.def (TARGET_MEM_REF): Add comment on types of
2680 2007-01-11 Joseph Myers <joseph@codesourcery.com>
2682 * c-common.c (vector_types_convertible_p): Treat opaque types as
2683 always convertible if they have the same size, but not otherwise.
2685 2007-01-11 Steven Bosscher <steven@gcc.gnu.org>
2687 * ifcvt.c (struct noce_if_info): Add comments to the fields.
2688 Remove the b_unconditional field.
2689 (noce_try_sign_mask): Do not look at b_unconditional.
2690 (noce_process_if_block): Do not use merge_if_blocks. Update
2691 the CFG here. Do not set b_unconditional.
2692 (cond_move_process_if_block): Likewise.
2693 (find_cond_trap): Likewise.
2694 (check_cond_move_block): Require simple jump insns at the end
2697 2007-01-11 Jan Hubicka <jh@suse.cz>
2699 PR tree-optimization/1046
2700 * tree-tailcall.c (suitable_for_tail_call_opt_p): Use TREE_ADDRESSABLE
2701 when alias info is not ready.
2702 (pass_tail_recursion): Do not require aliasing.
2703 * tree-ssa-copyrename.c (pass_rename_ssa_cop): Likewise.
2704 * tree-ssa-ccp.c (pass_ccp, pass_fold_builtins): Likewise.
2705 * tree-ssa-copy.c (pass_copy_prop): Likewise.
2706 * tree-ssa-forwprop.c (pass_forwprop): Likewise.
2707 * tree-ssa-dce.c (pass_dce, pass_dce_loop, pass_cd_dce): Likewise.
2708 * passes.c (init_optimization_passes): Execute rename_ssa_copies,
2709 ccp, forwprop, copy_prop, merge_phi, copy_prop, dce and tail recursion
2711 * tree-ssa-operands.c (add_virtual_operand, get_indirect_ref_operand):
2712 When aliasing is not build, mark statement as volatile.
2714 2007-01-11 Tom Tromey <tromey@redhat.com>
2716 PR preprocessor/15185, PR preprocessor/20989:
2717 * doc/cppopts.texi <-MT>: Update description of algorithm for
2718 computing default target.
2719 <-M, -MD>: Reword "basename" text.
2721 2007-01-11 Roger Sayle <roger@eyesopen.com>
2723 * builtins.c (expand_builtin_pow, expand_builtin_powi,
2724 fold_builtin_cabs, fold_builtin_sqrt, fold_builtin_trunc,
2725 fold_builtin_floor, fold_builtin_ceil, fold_builtin_round,
2726 fold_builtin_int_int_roundingfn, fold_builtin_bitop,
2727 fold_builtin_bswap, real_constp, fold_builtin_pow,
2728 fold_builtin_powi, fold_builtin_signbit, fold_builtin_copysign,
2729 do_mpfr_arg1, do_mpfr_arg2, do_mpfr_arg3, do_mpfr_sincos): Replace
2730 uses of the macro TREE_CONSTANT_OVERFLOW with TREE_OVERFLOW.
2731 * convert.c (convert_to_pointer): Likewise.
2732 * expr.c (highest_pow2_factor, expand_expr_real_1): Likewise.
2733 * fold-const.c (force_fit_type, fold_negate_expr, int_const_binop,
2734 const_binop, fold_convert_const_int_from_int,
2735 fold_convert_const_int_from_real,
2736 fold_convert_const_real_from_real, sign_bit_p,
2737 optimize_minmax_comparison, extract_muldiv_1, fold_div_compare,
2738 fold_sign_changed_comparison, fold_unary, fold_comparison,
2739 fold_binary, multiple_of_p, tree_Expr_non_zero_p,
2740 fold_negate_const, fold_abs_const, fold_not_const): Likewise.
2741 * print-tree.c (print_node_brief, print_node): Likewise.
2742 * stor-layout.c (place_field, layout_type): Likewise.
2743 * tree-chrec.c (keep_cast): Likewise.
2744 * tree.c (build_vector, build_real, build_real_from_int_cst,
2745 build_complex): Likewise.
2747 2007-01-11 Roger Sayle <roger@eyesopen.com>
2749 * tree.h (TREE_CONSTANT_OVERFLOW): Obsolete. For the time being,
2750 treat TREE_CONSTANT_OVERFLOW as a synonym of TREE_OVERFLOW.
2752 2007-01-11 Paolo Bonzini <bonzini@gnu.org>
2754 * configure.ac (strict1_warn): Rename to strict_warn.
2755 (WERROR, --enable-werror, symlink hacks, stage1_cflags,
2756 cc_set_by_configure, quoted_cc_set_by_configure,
2757 stage_prefix_set_by_configure, quoted_stage_prefix_set_by_configure,
2758 all_boot_languages, all_stagestuff): Remove.
2759 (target_list): Remove bootstrap targets.
2760 * Makefile.in (quickstrap): Unconditionally make a synonym of all.
2761 (BOOT_LANGUAGES, STAGE1_CFLAGS, STAGE1_CHECKING,
2762 REMAKEFLAGS, FLAGS_TO_PASS, PREPEND_DOTDOT_TO_RELATIVE_PATHS,
2763 SUBDIR_FLAGS_TO_PASS, WERROR_FLAGS, STRICT2_WARN, LANG_STAGESTUFF,
2764 VOL_FILES, POSTSTAGE1_FLAGS_TO_PASS, STAGE2_FLAGS_TO_PASS,
2765 STAGEPROFILE_FLAGS_TO_PASS, STAGEFEEDBACK_FLAGS_TO_PASS, stage1_build,
2766 stage1_copy, stage2_build, stage2_copy, stageprofile_build,
2767 stageprofile_copy, stage3_build, stage3_copy, stagefeedback_build,
2768 stagefeedback_copy, stage4_build, clean_s1, clean_sw, bootstrap,
2769 bootstrap-lean, bootstrap2, bootstrap2-lean, bootstrap3,
2770 bootstrap3-lean, bootstrap4, bootstrap4-lean, unstage1, unstage2,
2771 unstage3, unstage4, unstageprofile, unstagefeedback, restage, restage2,
2772 restage3, restage4, restageprofile, restagefeedback, bubbleestrap,
2773 cleanstrap, unstrap, restrap, *compare, *compare3, *compare4,
2774 *compare-lean, *compare3-lean, *compare4-lean, stage1-start, stage1,
2775 stage2-start, stage2, stage3-start, stage3, stage4-start, stage4,
2776 stageprofile-start, stageprofile, stagefeedback-start, stagefeedback,
2777 risky-stage1, risky-stage2, risky-stage3, risky-stage4): Remove.
2778 (ORDINARY_FLAGS_TO_PASS): Rename to FLAGS_TO_PASS.
2779 (STAGECOPYSTUFF, STAGEMOVESTUFF): Consolidate into MOSTLYCLEANFILES.
2780 (mostlyclean): Adjust.
2781 (clean, distclean): Don't mention bootstrap stuff.
2782 * configure: Regenerate.
2783 * ada/config-lang.in, cp/config-lang.in, forttran/config-lang.in,
2784 java/config-lang.in, objc/config-lang.in, objcp/config-lang.in,
2785 treelang/config-lang.in (stagestuff): Remove.
2786 * doc/sourcebuild.texi (stage1, stage2, stage3, stage4,
2787 stageprofile, stagefeedback, stagestuff): Remove mention.
2789 2007-01-11 Nick Clifton <nickc@redhat.com>
2791 * config/mcore/predicates.md (mcore_general_movesrc_operand):
2793 (mcore_general_movdst_operand): Do not accept CONST_INTs.
2794 (mcore_arith_K_S_operand): Run the test for the S constraint not
2795 the test for the M constraint.
2796 (mcore_addsub_operand): Do not accept integer values that are
2797 larger than 32 bits.
2798 * config/mcore/mcore.md: Remove unused constraints from split.
2799 (andsi3): Use HOST_WIDE_INT instead of int to hold an INTVAL.
2801 (allocate_stack): Likewise.
2802 * config/mcore/mcore.c (mcore_print_operand): Restrict output of P
2803 operands to 32 bits.
2804 (mcore_const_costs): Use HOST_WIDE_INT instead of int to hold an
2806 (mcore_and_cost, mcore_modify_comparison, const_ok_for_mcore,
2807 mcore_const_ok_for_inline, mcore_const_trick_uses_not,
2808 try_constant_tricks, mcore_num_ones, mcore_num_zeros,
2809 mcore_output_bclri, mcore_output_andn, output_inline_const,
2810 mcore_output_move, mcore_output_movedouble): Likewise.
2811 (mcore_output_cmov): Use CONST_OK_FOR_M and CONST_OK_FOR_N.
2812 (output_inline_const): Likewise.
2813 (output_inline_const): Fix format strings used in sprintf
2815 * config/mcore/mcore-protos.h: Update prototypes for changed
2816 functions in mcore.c.
2817 * config/mcore/mcore.h (CONST_OK_FOR_I): Cast values to
2818 HOST_WIDE_INT and not int.
2819 (CONST_OK_FOR_J, CONST_OK_FOR_K, CONST_OK_FOR_L, CONST_OK_FOR_M,
2820 CONST_OK_FOR_N): Likewise.
2821 (LEGITIMATE_CONSTANT_P): Also check CONSTANT_P.
2822 (GO_IF_LEGITIMATE_INDEX): Use HOST_WIDE_INT instead of int to hold
2825 2007-01-10 Jan Hubicka <jh@suse.cz>
2827 * tree-vrp.c (remove_range_assertions): Release defs.
2828 * tree-ssa-loop-ivopts.c (rmeove_statement): Likewise.
2829 * tree-ssa-dom.c (remove_stmt_or_phi): Likewise.
2831 2007-01-10 Paul Brook <paul@codesourcery.com>
2833 * config/arm/arm.c (arm_rtx_costs_1): Handle mutiply-subtract.
2834 * config/arm/arm.md (mulsi3subsi): New insn.
2836 2007-01-10 Zdenek Dvorak <dvorakz@suse.cz>
2838 * tree-ssa-loop-manip.c (tree_unroll_loop): Make it a wrapper over ...
2839 (tree_transform_and_unroll_loop): New.
2840 * tree-flow.h (transform_callback, tree_transform_and_unroll_loop):
2843 2007-01-10 Robert Kennedy <jimbob@google.com>
2845 * fold-const.c (fold_comparison): Fold comparisons like (x *
2846 1000 < 0) to (x < 0).
2848 2007-01-10 Ian Lance Taylor <iant@google.com>
2850 * tree-pretty-print.c (dump_generic_node): Print parentheses when
2851 operands have the same priority.
2853 2007-01-10 Tom Tromey <tromey@redhat.com>
2855 * fold-const.c (fold_truthop): Don't check can_use_bit_fields_p.
2856 (fold_binary): Likewise.
2857 * langhooks.c (lhd_can_use_bit_fields_p): Removed.
2858 * langhooks-def.h (lhd_can_use_bit_fields_p): Removed.
2859 (LANG_HOOKS_CAN_USE_BIT_FIELDS_P): Removed.
2860 (LANG_HOOKS_INITIALIZER): Remove LANG_HOOKS_CAN_USE_BIT_FIELDS_P.
2861 * langhooks.h (struct lang_hooks): Removed field
2862 'can_use_bit_fields_p'.
2864 2007-01-10 Ralf Corsépius <ralf.corsepius@rtems.org>
2866 * config/bfin/t-bfin, config/bfin/t-bfin-elf: Remove GCC_CFLAGS.
2868 2007-01-10 Razya Ladelsky <razya@il.ibm.com>
2870 * function.c (get_last_funcdef_no): New function.
2871 * function.h (get_last_funcdef_no): Declare.
2872 * tree-inline.c (initialize_cfun): Add initialization.
2873 (tree_function_versioning): Cleanup.
2875 2007-01-10 Jan Hubicka <jh@suse.cz>
2877 * tree-inline.c (setup_one_parameter): Do not propagate into abnormal
2880 2007-01-10 Sa Liu <saliu@de.ibm.com>
2881 Ben Elliston <bje@au.ibm.com>
2883 * spu.h (STACK_SAVE_AREA): Use VOIDmode for SAVE_FUNCTION, SImode
2884 for SAVE_NONLOCAL and Pmode for any other save level.
2885 * spu-protos.h (spu_restore_stack_block): Declare.
2886 * spu.md (save_stack_block): Remove.
2887 (restore_stack_block): Call spu_restore_stack_block.
2888 * spu.c (spu_restore_stack_block): New function.
2889 (spu_expand_epilogue): Remove old comment.
2891 2007-01-09 Zdenek Dvorak <dvorakz@suse.cz>
2893 PR tree-optimization/30322
2894 * tree-ssa-loop-ivopts.c (fold_affine_expr, iv_value): Removed.
2895 (cand_value_at): Return the value as aff_tree.
2896 (may_eliminate_iv): Convert the bound from aff_tree to tree.
2897 * tree-affine.c (aff_combination_add_cst, aff_combination_add_product,
2898 aff_combination_mult): New functions.
2899 (aff_combination_add): Use aff_combination_add_cst.
2900 (aff_combination_convert): Allow conversions to a wider type.
2901 (tree_to_aff_combination): Handle BIT_NOT_EXPR.
2902 * tree-affine.h (aff_combination_mult): Declare.
2904 2007-01-09 Carlos O'Donell <carlos@codesourcery.com>
2906 * doc/tm.texi: Update documentation to reflect reality of exec
2907 and start file search behaviours. Update copyright year.
2908 * doc/invoke.texi: Explain how GCC_EXEC_PREFIX is used to find
2909 header file directories.
2911 2007-01-09 Uros Bizjak <ubizjak@gmail.com>
2913 * config/i386/i386.md (*sinxf2): Rename to *sinxf2_i387.
2914 (*cosxf2): Rename to cosxf2_i387.
2915 (*sindf2, *sinsf2): Extend operand 1 to XFmode. Macroize patterns
2916 using X87MODEF12 mode macro. Rename patterns to
2917 *sin_extend<mode>xf2_i387. Use SSE_FLOAT_MODE_P to disable patterns
2919 (*cosdf2, *cossf2): Ditto.
2920 (sincosdf3, sincossf3): Ditto. Rewrite corresponding splitters
2921 to match extended input operands.
2922 (sincos<mode>3): New expander.
2923 (*sinextendsfdf2, *cosextendsfdf2, *sincosextendsfdf3): Remove
2924 insn patterns and corresponding splitters.
2926 2007-01-09 Kaz Kojima <kkojima@gcc.gnu.org>
2928 * config/sh/t-linux (TARGET_LIBGCC2_CFLAGS): Delete.
2929 (SHLIB_MAPFILES, SHLIB_LINK, SHLIB_INSTALL): Likewise.
2931 2007-01-09 Nicolas Pitre <nico@cam.org>
2934 * arm/ieee754-df.S (Lad_s): Also test the low word of X for zero.
2936 2007-01-08 Geoffrey Keating <geoffk@apple.com>
2938 * target.h (struct gcc_target): New field library_rtti_comdat.
2939 * target-def.h (TARGET_CXX_LIBRARY_RTTI_COMDAT): New.
2940 (TARGET_CXX): Add TARGET_CXX_LIBRARY_RTTI_COMDAT.
2941 * doc/tm.texi (C++ ABI): Document TARGET_CXX_LIBRARY_RTTI_COMDAT.
2942 * config/darwin.h (TARGET_CXX_LIBRARY_RTTI_COMDAT): Define.
2944 2007-01-08 Geoffrey Keating <geoffk@apple.com>
2946 * doc/invoke.texi (Optimize Options): Correct description of -O0.
2948 2007-01-08 Richard Guenther <rguenther@suse.de>
2950 * tree.h (force_fit_type_double): Export.
2951 (force_fit_type): Remove.
2952 * fold-const.c (force_fit_type_double): New function.
2953 (force_fit_type): Remove.
2954 (int_const_binop): Use it.
2955 (fold_convert_const_int_from_int): Likewise.
2956 (fold_convert_const_int_from_real): Likewise.
2957 (fold_div_compare): Likewise.
2958 (fold_sign_changed_comparison): Likewise.
2959 (fold_unary): Likewise.
2960 (fold_negate_const): Likewise.
2961 (fold_abs_const): Likewise.
2962 (fold_not_const): Likewise.
2963 * c-common.c (shorten_compare): Use force_fit_type_double.
2964 * convert.c (convert_to_pointer): Likewise.
2966 2007-01-08 Richard Guenther <rguenther@suse.de>
2968 * tree.h (build_int_cst_wide_type): Export.
2969 * tree.c (build_int_cst_wide_type): New function.
2970 (build_int_cst_wide): Fix comment.
2971 * builtins.c (fold_builtin_object_size): Use build_int_cst
2972 to build -1 or 0 of the correct type. Use fit_double_type
2973 to check for overflow.
2974 * fold-const.c (optimize_bit_field_compare): Use build_int_cst_type
2976 (decode_field_reference): Likewise.
2977 (all_ones_mask_p): Likewise.
2978 (native_interpret_int): Use build_int_cst_wide_type.
2979 (fold_binary): Use build_int_cst_type to build an all-ones
2981 * stor-layout.c (set_sizetype): Use build_int_cst_wide_type.
2983 2007-01-08 Daniel Jacobowitz <dan@codesourcery.com>
2985 * config/pa/t-pa64 (libgcc_stub.a): Use $(T).
2987 2007-01-09 Ben Elliston <bje@au.ibm.com>
2989 * genautomata.c (STATS_OPTION): New option.
2990 (stats_flag): New flag.
2991 (gen_automata_option): Handle it.
2992 (initiate_automaton_gen): Ditto.
2993 (write_automata): Output statistics only if stats_flag is
2994 set. Likewise, output time statistics only if time_flag is set.
2995 * doc/md.texi (Processor pipeline description): Document new flag.
2997 2007-01-08 Richard Guenther <rguenther@suse.de>
2999 * builtins.c (fold_builtin_int_roundingfn): Use fit_double_type.
3000 * tree.c (build_int_cst_type): Likewise.
3001 (size_in_bytes): Don't call force_fit_type on the result.
3002 (int_fits_type_p): Use fit_double_type.
3003 * fold-const.c (fit_double_type): New function.
3004 (force_fit_type): Use it.
3005 * tree.h (fit_double_type): Export.
3007 2007-01-08 Jan Hubicka <jh@suse.cz>
3009 * tree-vectorizer.c (gate_increase_alignment): Fix return type.
3010 * ipa.c (function_and_variable_visibility): Fix return type.
3012 2007-01-08 Richard Guenther <rguenther@suse.de>
3014 * tree-ssa-ccp.c (maybe_fold_offset_to_array_ref): Use type
3015 of offset to build the index.
3016 * tree-pretty-print.c (dump_generic_node): Don't build negated
3017 const just for printing.
3018 * c-pretty-print.c (pp_c_integer_constant): Likewise.
3019 * builtins.c (fold_builtin_int_roundingfn): Check if result
3020 fits the type by using force_fit_type and comparing the result.
3021 * predict.c (predict_loops): Use compare_tree_int for comparison.
3022 * tree.c (build_int_cst): Fall back to integer_type_node for
3024 (build_int_cst_wide): Assert type is non-null.
3026 2007-01-08 Roberto Costa <roberto.costa@st.com>
3028 * tree-vrp.c (extract_range_from_cond_expr): New.
3029 (extract_range_from_expr): Handle COND_EXPR nodes used as expressions.
3030 * tree-ssa-ccp.c (get_maxval_strlen): Handle COND_EXPR nodes used
3032 (fold_stmt): Bug fix, avoid infinite recursion when folding COND_EXPRs.
3033 * tree-ssa-forwprop.c (simplify_cond, forward_propagate_into_cond,
3034 tree_ssa_forward_propagate_single_use_vars): Handle COND_EXPR nodes
3035 used as expressions.
3036 * tree-object-size.c (cond_expr_object_size): New.
3037 (collect_object_sizes_for): Handle COND_EXPR nodes used as expressions.
3039 2007-01-08 Jan Hubicka <jh@suse.cz>
3041 * tree-ssa-forwprop.c (forward_propagate_into_cond,
3042 tree_ssa_forward_propagate_single_use_va): Release defs of propagated
3045 2007-01-08 Richard Guenther <rguenther@suse.de>
3047 PR tree-optimization/23603
3048 * tree-vrp.c (set_value_range_to_truthvalue): New function.
3049 (extract_range_from_binary): Fall back to truthvalue instead of
3050 varying for TRUTH_*_EXPR.
3051 (extract_range_from_comparison): Fall back to truthvalue instead of
3053 (vrp_visit_phi_node): Don't adjust new range bounds to +INF/-INF
3054 if all visited PHI values were constant.
3056 2007-01-08 Jan Hubicka <jh@suse.cz>
3058 * cgraphunit.c (cgraph_process_new_functions): Reset reachable flag.
3059 (cgraph_analyze_function): break out from ...
3060 (cgraph_finalize_compilation_unit): ... here.
3061 (cgraph_expand_function): Remove forgoten commented out line.
3062 (cgraph_optimize): Analyze functions.
3064 2007-01-08 Jan Hubicka <jh@suse.cz>
3066 * tree-pas.h (TODO_remove_function): New flag.
3067 (TODO_update*): Renumber.
3068 (pass_ipa_increase_alignment,
3069 pass_ipa_function_and_variable_visibility): New passes.
3070 * cgraphunit.c (cgraph_increase_alignment): Move to tree-vectorizer.c
3071 (cgraph_function_and_variable_visibility): Move to ipa.c
3072 (cgraph_optimize): Don't call cgraph_function_and_variable_visibility,
3073 cgraph_increase_alignment.
3074 * ipa-inline.c (cgraph_decide_inlining): Don't push timevar.
3075 (cgraph_decide_inlining_incrementally): Push TV_INTEGRATION before
3076 calling tree-inline.
3077 (cgraph_early_inlining): Do not call cgraph_remove_unreachable_nodes.
3078 (pass_ipa_inline, pass_early_ipa_inlining): Set TODO_remove_functions
3079 * tree-vectorizer.c (increase_alignment): Move here from cgraphunit.c
3080 (gate_increase_alignment): New function.
3081 (pass_ipa_increase_alignment): New pass.
3082 * ipa.c: Inline tree-pass.h and timevar.h
3083 (function_and_variable_visibility): Move here from cgraphunit.c
3084 * tree-optimize.c (pass_early_local_passes): Add TODO_remove_functions.
3085 * passes.c (init_optimization_passes): Add the two new passes.
3086 (execute_todo): Handle cgraph_remove_functions.
3088 2007-01-08 Nick Clifton <nickc@redhat.com>
3090 * config/frv/predicates.md (reg_or_0_operand): Accept
3093 2007-01-08 Ralf Corsépius <ralf.corsepius@rtems.org>
3095 * config/bfin/rtems.h, config/bfin/t-rtems: New.
3096 * config.gcc: Add bfin*-rtems*.
3098 2007-01-08 Mark Shinwell <shinwell@codesourcery.com>
3100 * c.opt: Add -flax-vector-conversions.
3101 * c-typeck.c (convert_for_assignment): Pass flag to
3102 vector_types_convertible_p to allow emission of note.
3103 (digest_init): Likewise.
3104 * c-opts.c: Handle -flax-vector-conversions.
3105 * c-common.c (flag_lax_vector_conversions): New.
3106 (vector_types_convertible_p): Unless -flax-vector conversions
3107 has been passed, disallow conversions between vectors with
3108 differing numbers of subparts and/or element types. If such
3109 a conversion is disallowed, possibly emit a note on the first
3110 occasion only to inform the user of -flax-vector-conversions.
3111 The new last argument specifies this.
3112 * c-common.h (flag_lax_vector_conversions): New.
3113 (vector_types_convertible_p): Add extra argument.
3114 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use
3115 char_type_node for V*QI type vectors.
3116 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
3117 Update to satisfy new typechecking rules.
3118 * config/rs6000/altivec.h (vec_cmple): Use vec_cmpge, for both
3120 * doc/invoke.texi (C Dialect Options): Document
3121 -flax-vector-conversions.
3123 2007-01-08 Mark Shinwell <shinwell@codesourcery.com>
3125 PR tree-optimization/29877
3126 * tree-ssa-ter.c (is_replaceable_p): Deem assignments with
3127 a register variable on the RHS to not be replaceable.
3129 2007-01-08 Chen Liqin <liqin@sunnorth.com.cn>
3130 * config/score/t-score-elf (MULTILIB_OPTIONS): Change.
3131 * config/score/predicates.md (const_uimm5, sr0_operand, const_simm12,
3132 const_simm15, const_pow2, const_npow2): Added.
3133 * config/score/misc.md (insv, extv, extzv, movmemsi,
3134 move_lbu_a/b, mov_lhu_a/b etc): Added and fix some bug.
3135 * config/score/score.c (score_address_cost, score_select_cc_mode):
3137 Change CONST_OK_FOR_LETTER_P/EXTRA_CONSTRAINT define.
3138 Update score_rtx_costs for MACRO TARGET_RTX_COSTS.
3139 Update score_print_operand.
3140 * config/score/score.h (DATA_ALIGNMENT, SELECT_CC_MODE): Added.
3141 Adjust register allocate order and update some macro define.
3142 * config/score/score-mdaux.c (mdx_unaligned_load, mdx_unsigned_store,
3143 mdx_block_move_straight, mdx_block_move_loop_head,
3144 mdx_block_move_loop_body, mdx_block_move_loop_foot, mdx_block_move_loop,
3145 mdx_block_move): Added.
3146 (mdx_movsicc, mdp_select_add_imm, mdp_select, mds_zero_extract_andi,
3147 mdp_limm): Updated and fix some bug and typo.
3148 * config/score/score.md (movqi/hi/si, add/sub/zero/ext): Updated.
3149 (movsf, movdf, doloop_end): Added.
3151 2007-01-08 Kazu Hirata <kazu@codesourcery.com>
3153 * config/arm/arm.c, config/arm/arm.h, config/arm/arm.md,
3154 config/arm/thumb2.md: Fix comment typos.
3155 * doc/extend.texi: Fix a typo.
3157 2007-01-07 Eric Christopher <echristo@apple.com>
3159 * configure.ac: Check for __stack_chk_fail for darwin.
3160 * configure: Regenerate.
3162 2007-01-07 Richard Guenther <rguenther@suse.de>
3164 * tree-vrp.c (extract_range_from_assert): CSE calls to
3165 compare_values where possible.
3166 (extract_range_from_unary_expr): Likewise.
3168 2007-01-07 Anatoly Sokolov <aesok@post.ru>
3170 * config/avr/avr-protos.h (call_insn_operand): Delete prototype.
3171 * config/avr/avr.c (call_insn_operand): Delete function.
3172 * config/avr/avr.md (*pushqi, *pushhi, *pushsi, *pushsf): Use REG_SP
3173 instead of register number. Use predicates.
3174 * config/avr/predicates.md (const0_operand, reg_or_0_operand,
3175 call_insn_operand): Add.
3177 2007-01-06 Jan Hubicka <jh@suse.cz>
3179 * tree-pass.h (pass_build_cgraph_edges): Declare.
3180 * cgraphunit.c (record_refernece): Move to cgraphbuild.c
3181 (visited_nodes): Remove.
3182 (cgraph_create_edges): Move to cgraphbuild.c; rename to
3183 build_cgrpah_edges; make visited_nodes local.
3184 (cgraph_process_new_functions): DO not call initialize_inline_failed.
3185 (record_references_in_initializer): Move to cgraphbuild.c
3186 (initialize_inline_failed, rebuild_cgraph_edges,
3187 pass_rebuild_cgraph_edges): Move to cgraphbuild.c.
3188 (verify_cgraph_node): Make visited_nodes local.
3189 (cgraph_analyze_function): Do not call cgraph_create_edges and
3190 initialize_inline_failed.
3191 (cgraph_expand_function): Do not call cgraph_lower_function;
3192 assert that function is already lowered.
3193 * Makefile.in (cgraphbuild.o): New.
3194 * passes.c (init_optimization_passes): Add pass_build_cgraph_edges
3195 at the end of lowering passes.
3197 2007-01-06 Steven Bosscher <steven@gcc.gnu.org>
3199 * ifcvt.c (cond_move_convert_if_block): New function, code
3200 factored out from...
3201 (cond_move_process_if_block): ...here. Call the new function
3202 on the THEN and ELSE blocks.
3203 (merge_if_block): Do not copy global_live_at_end, merge_blocks
3204 already takes care of this.
3206 2007-01-05 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
3209 * tree.h (TREE_OVERFLOW_P): New.
3210 * c-typeck.c (parser_build_unary_op): Warn only if result
3211 overflowed and operands did not.
3212 (parser_build_binary_op): Likewise.
3213 (convert_for_assignment): Remove redundant overflow_warning.
3214 * c-common.c (overflow_warning): Don't check or set TREE_OVERFLOW.
3216 2007-01-05 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
3218 * c-typeck.c (store_init_value): Split over two lines to follow
3219 the GNU coding style.
3221 2007-01-05 Benjamin Kosnik <bkoz@redhat.com>
3223 * c-cppbuiltin.c (c_cpp_builtins): __GXX_EXPERIMENTAL_CPP0X__ to
3224 __GXX_EXPERIMENTAL_CXX0X__.
3225 * doc/cpp.texi: Same.
3227 2007-01-05 Richard Guenther <rguenther@suse.de>
3230 * tree.c (get_narrower): Do not construct COMPONENT_REFs
3231 with mismatched types. Instead explicitly build a
3232 conversion NOP_EXPR.
3234 2007-01-05 Ian Lance Taylor <iant@google.com>
3236 * c-common.c (decl_with_nonnull_addr_p): New function.
3237 (c_common_truthvalue_conversion): Call it.
3238 * c-typeck.c (build_binary_op): Likewise.
3239 * c-common.h (decl_with_nonnull_addr_p): Declare.
3241 2007-01-05 Jakub Jelinek <jakub@redhat.com>
3244 * libgcc2.c (__divdc3): Compare c and d against 0.0 instead of
3247 2007-01-05 Joel Brobecker <brobecker@adacore.com>
3249 * doc/install.texi (Final install): Document the fact that
3250 the GNAT runtime should not be stripped.
3252 2007-01-04 Jan Hubicka <jh@suse.cz>
3254 * tree-inline.c (fold_marked_statements): Update operand caches
3255 and EH after folding
3257 2007-01-04 Ian Lance Taylor <iant@google.com>
3259 * c-common.c (check_function_nonnull): Whitespace fix.
3261 2007-01-04 Jan Hubicka <jh@suse.cz>
3263 * tree-optimize.c (execute_fixup_cfg): Correct previously mistakely
3264 comitted older version of patch.
3265 (pass_fixup_cfg): Add TODOs to verify flow and statements, dump
3266 function, celanup cfg and collect garbage.
3268 2007-01-04 Mike Stump <mrs@apple.com>
3270 * Makefile.in (mostlyclean): Don't remove libgcc anymore.
3273 2007-01-04 Eric Christopher <echristo@apple.com>
3275 * libgcc2.c (__bswapsi2): Use SItype.
3276 (__bswapdi2): Use DItype.
3277 * libgcc2.h: Update for above.
3279 2007-01-04 Paul Brook <paul@codesourcery.com>
3281 * config/arm/arm.md (arm_mulsi3, thumb_mulsi3, mulsi3_compare0,
3282 mulsi_compare0_scratch, mulsi3addsi, mulsi3addsi_compare0,
3283 mulsi3addsi_compare0_scratch, mulsidi3adddi, mulsidi3,
3284 umulsidi3, umulsidi3adddi, smulsi3_highpart,
3285 umulsi3_highpart): Make conditional on !arm_arch6.
3286 (arm_mulsi3_v6, thumb_mulsi3_v6, mulsi3_compare0_v6,
3287 mulsi_compare0_scratch_v6, mulsi3addsi_v6, mulsi3addsi_compare0_v6,
3288 mulsi3addsi_compare0_scratch_v6, mulsidi3adddi_v6, mulsidi3_v6,
3289 umulsidi3_v6, umulsidi3adddi_v6, smulsi3_highpart_v6,
3290 umulsi3_highpart_v6): New insns.
3292 2007-01-04 Roger Sayle <roger@eyesopen.com>
3294 * fold-const.c (fold_convert): When casting an expression to void,
3295 fold_ignored_result may discover a GIMPLE_MODIFY_STMT which doesn't
3296 have a type. Instead of attempting to build a NOP_EXPR, return
3297 these "special" trees directly.
3299 2007-01-04 Joseph Myers <joseph@codesourcery.com>
3301 * config/rs6000/rs6000.c (rs6000_rtx_costs): Make adjustment for
3302 MULT inside MINUS as either argument. Use rs6000_cost->dmul -
3303 rs6000_cost->fp not 0 as adjustment for outer NEG.
3305 2007-01-04 Jan Hubicka <jh@suse.cz>
3307 * cgraph.c (cgraph_release_function_body): New function.
3308 (cgraph_remove_node): Use it.
3309 * cgraph.h (cgraph_release_function_body): Declare.
3310 * cgraphunit.c (cgraph_expand_function): Use it.
3311 * ipa.c (cgraph_remove_unreahchable_nodes): Use it.
3312 * tree-ssa.c (delete_tree_ssa): Allow to be called before aliasing
3313 is initialized and while compilation of other function is running.
3314 * tree-optimize.c (execute_free_cfg_annotations): Move code to clear
3315 statement CFG annotations from here to ...
3316 * tree-cfg.c (delete_tree_cfg_annotations): ... here.
3318 2007-01-04 Zdenek Dvorak <dvorakz@suse.cz>
3320 * cfgloop.h (enum li_flags): Make the constants powers of two.
3322 2007-01-04 Jan Hubicka <jh@suse.cz>
3324 * tree-inline.c (copy_bb): Insert new statements to statements_to_fold
3326 (fold_marked_statements): New function.
3327 (optimize_inline_calls, tree_function_versioning): Fold new statements.
3328 * tree-inline.h (copy_body_data): Add statements_to_fold.
3330 2007-01-03 Daniel Jacobowitz <dan@codesourcery.com>
3332 * config.gcc: Mention libgcc/config.host.
3333 * Makefile.in: Update comments mentioning libgcc.
3334 (LIBGCC, INSTALL_LIBGCC, GCC_PARTS, mklibgcc): Delete.
3335 (all.cross, start.encap, rest.encap, rest.cross): Update
3336 dependencies for libgcc move.
3337 (libgcc.mk, LIBGCC_DEPS, libgcov.a, libgcc.a, stmp-multilib)
3338 (clean-target, clean-target-libgcc): Delete.
3339 (srcdirify, GCC_EXTRA_PARTS): New macros.
3340 (libgcc-support, libgcc.mvars): New rules.
3341 (distclean): Remove mention of mklibgcc.
3342 (install): Don't reference INSTALL_LIBGCC.
3343 (install-common): Don't reference EXTRA_PARTS.
3344 (install-libgcc, install-multilib): Delete rules.
3345 * mklibgcc.in: Delete file.
3346 * doc/configfiles.texi: Don't mention mklibgcc.
3348 * config/i386/t-darwin (SHLIB_VERPFX): Delete (moved to libgcc).
3349 * config/i386/t-darwin64 (SHLIB_VERPFX): Likewise.
3350 * config/rs6000/t-darwin (SHLIB_VERPFX): Likewise.
3351 * config/rs6000/t-ppccomm (TARGET_LIBGCC2_CFLAGS, SHLIB_MAPFILES)
3352 (mklibgcc, ldblspecs): Likewise.
3354 * config/i386/t-nwld (libgcc.def, libc.def, libpcre.def)
3355 (posixpre.def): Use $(T).
3356 (SHLIB_EXT, SHLIB_NAME, SHLIB_SLIBDIR_QUAL, SHLIB_DEF, SHLIB_MAP)
3357 (SHLIB_SRC, SHLIB_INSTALL): Delete.
3358 (SHLIB_LINK): Make dummy.
3359 * config/t-slibgcc-darwin: Delete contents except for dummy SHLIB_LINK.
3361 * config/frv/t-linux (EXTRA_MULTILIB_PARTS): Clear.
3363 * config/alpha/t-crtfm: Use $(T) in rules for EXTRA_PARTS.
3364 * config/alpha/t-vms, config/alpha/t-vms64, config/fr30/t-fr30,
3365 config/i386/t-rtems-i386, config/ia64/t-ia64, config/rs6000/t-beos,
3366 config/rs6000/t-newas, config/sparc/t-elf: Likewise.
3368 * configure.ac (all_outputs): Remove mklibgcc.
3369 * configure: Regenerated.
3371 2007-01-03 Josh Conner <jconner@apple.com>
3374 * calls.c (compute_argument_addresses): Set stack and stack_slot
3375 for partial args, too.
3376 (store_one_arg): Use locate.size.constant for the size when
3377 generating a save_area.
3379 2007-01-03 Robert Kennedy <jimbob@google.com>
3381 * tree-cfg.c (tree_merge_blocks): Release SSA_NAME phi results
3382 whose definitions are deleted due to basic block merging.
3384 2007-01-03 Paul Brook <paul@codesourcery.com>
3387 * config/arm/arm.c (output_return_instruction): Pop PC in interrupt
3389 (use_return_insn): Return 0 for Thumb interrupt functions.
3390 (print_multi_reg): Add rfe argument for IRQ returns.
3391 (arm_output_epilogue): Pop interrupt return address directly into PC.
3392 (arm_expand_prologue): Only adjust IRQ return address in Arm mode.
3394 2007-01-03 Paul Brook <paul@codesourcery.com>
3396 Merge from sourcerygxx-4_1.
3397 * config/arm/thumb2.md: New file.
3398 * config/arm/elf.h (JUMP_TABLES_IN_TEXT_SECTION): Return True for
3400 * config/arm/coff.h (JUMP_TABLES_IN_TEXT_SECTION): Ditto.
3401 * config/arm/aout.h (ASM_OUTPUT_ADDR_VEC_ELT): Add !Thumb-2 assertion.
3402 (ASM_OUTPUT_ADDR_DIFF_ELT): Output Thumb-2 jump tables.
3403 * config/arm/aof.h (ASM_OUTPUT_ADDR_DIFF_ELT): Output Thumb-2 jump
3405 (ASM_OUTPUT_ADDR_VEC_ELT): Add !Thumb-2 assertion.
3406 * config/arm/ieee754-df.S: Use macros for Thumb-2/Unified asm
3408 * config/arm/ieee754-sf.S: Ditto.
3409 * config/arm/arm.c (thumb_base_register_rtx_p): Rename...
3410 (thumb1_base_register_rtx_p): ... to this.
3411 (thumb_index_register_rtx_p): Rename...
3412 (thumb1_index_register_rtx_p): ... to this.
3413 (thumb_output_function_prologue): Rename...
3414 (thumb1_output_function_prologue): ... to this.
3415 (thumb_legitimate_address_p): Rename...
3416 (thumb1_legitimate_address_p): ... to this.
3417 (thumb_rtx_costs): Rename...
3418 (thumb1_rtx_costs): ... to this.
3419 (thumb_compute_save_reg_mask): Rename...
3420 (thumb1_compute_save_reg_mask): ... to this.
3421 (thumb_final_prescan_insn): Rename...
3422 (thumb1_final_prescan_insn): ... to this.
3423 (thumb_expand_epilogue): Rename...
3424 (thumb1_expand_epilogue): ... to this.
3425 (arm_unwind_emit_stm): Rename...
3426 (arm_unwind_emit_sequence): ... to this.
3427 (thumb2_legitimate_index_p, thumb2_legitimate_address_p,
3428 thumb1_compute_save_reg_mask, arm_dwarf_handle_frame_unspec,
3429 thumb2_index_mul_operand, output_move_vfp, arm_shift_nmem,
3430 arm_save_coproc_regs, thumb_set_frame_pointer, arm_print_condition,
3431 thumb2_final_prescan_insn, thumb2_asm_output_opcode, arm_output_shift,
3432 thumb2_output_casesi): New functions.
3433 (TARGET_DWARF_HANDLE_FRAME_UNSPEC): Define.
3434 (FL_THUMB2, FL_NOTM, FL_DIV, FL_FOR_ARCH6T2, FL_FOR_ARCH7,
3435 FL_FOR_ARCH7A, FL_FOR_ARCH7R, FL_FOR_ARCH7M, ARM_LSL_NAME,
3436 THUMB2_WORK_REGS): Define.
3437 (arm_arch_notm, arm_arch_thumb2, arm_arch_hwdiv, arm_condexec_count,
3438 arm_condexec_mask, arm_condexec_masklen)): New variables.
3439 (all_architectures): Add armv6t2, armv7, armv7a, armv7r and armv7m.
3440 (arm_override_options): Check new CPU capabilities.
3441 Set new architecture flag variables.
3442 (arm_isr_value): Handle v7m interrupt functions.
3443 (user_return_insn): Return 0 for v7m interrupt functions. Handle
3445 (const_ok_for_arm): Handle Thumb-2 constants.
3446 (arm_gen_constant): Ditto. Use movw when available.
3447 (arm_function_ok_for_sibcall): Return false for v7m interrupt
3449 (legitimize_pic_address, arm_call_tls_get_addr): Handle Thumb-2.
3450 (thumb_find_work_register, arm_load_pic_register,
3451 legitimize_tls_address, arm_address_cost, load_multiple_sequence,
3452 emit_ldm_seq, emit_stm_seq, arm_select_cc_mode, get_jump_table_size,
3453 print_multi_reg, output_mov_long_double_fpa_from_arm,
3454 output_mov_long_double_arm_from_fpa, output_mov_double_fpa_from_arm,
3455 output_mov_double_fpa_from_arm, output_move_double,
3456 arm_compute_save_reg_mask, arm_compute_save_reg0_reg12_mask,
3457 output_return_instruction, arm_output_function_prologue,
3458 arm_output_epilogue, arm_get_frame_offsets, arm_regno_class,
3459 arm_output_mi_thunk, thumb_set_return_address): Ditto.
3460 (arm_expand_prologue): Handle Thumb-2. Use arm_save_coproc_regs.
3461 (arm_coproc_mem_operand): Allow POST_INC/PRE_DEC.
3462 (arithmetic_instr, shift_op): Use arm_shift_nmem.
3463 (arm_print_operand): Use arm_print_condition. Handle '(', ')', '.',
3465 (arm_final_prescan_insn): Use extract_constrain_insn_cached.
3466 (thumb_expand_prologue): Use thumb_set_frame_pointer.
3467 (arm_file_start): Output directive for unified syntax.
3468 (arm_unwind_emit_set): Handle stack alignment instruction.
3469 * config/arm/lib1funcs.asm: Remove default for __ARM_ARCH__.
3470 Add v6t2, v7, v7a, v7r and v7m.
3471 (RETLDM): Add Thumb-2 code.
3472 (do_it, shift1, do_push, do_pop, COND, THUMB_SYNTAX): New macros.
3473 * config/arm/arm.h (TARGET_CPU_CPP_BUILTINS): Define __thumb2__.
3474 (TARGET_THUMB1, TARGET_32BIT, TARGET_THUMB2, TARGET_DSP_MULTIPLY,
3475 TARGET_INT_SIMD, TARGET_UNIFIED_ASM, ARM_FT_STACKALIGN, IS_STACKALIGN,
3476 THUMB2_TRAMPOLINE_TEMPLATE, TRAMPOLINE_ADJUST_ADDRESS,
3477 ASM_OUTPUT_OPCODE, THUMB2_GO_IF_LEGITIMATE_ADDRESS,
3478 THUMB2_LEGITIMIZE_ADDRESS, CASE_VECTOR_PC_RELATIVE,
3479 CASE_VECTOR_SHORTEN_MODE, ADDR_VEC_ALIGN, ASM_OUTPUT_CASE_END,
3480 ADJUST_INSN_LENGTH): Define.
3481 (TARGET_REALLY_IWMMXT, TARGET_IWMMXT_ABI, CONDITIONAL_REGISTER_USAGE,
3482 STATIC_CHAIN_REGNUM, HARD_REGNO_NREGS, INDEX_REG_CLASS,
3483 BASE_REG_CLASS, MODE_BASE_REG_CLASS, SMALL_REGISTER_CLASSES,
3484 PREFERRED_RELOAD_CLASS, SECONDARY_OUTPUT_RELOAD_CLASS,
3485 SECONDARY_INPUT_RELOAD_CLASS, LIBCALL_VALUE, FUNCTION_VALUE_REGNO_P,
3486 TRAMPOLINE_SIZE, INITIALIZE_TRAMPOLINE, HAVE_PRE_INCREMENT,
3487 HAVE_POST_DECREMENT, HAVE_PRE_DECREMENT, HAVE_PRE_MODIFY_DISP,
3488 HAVE_POST_MODIFY_DISP, HAVE_PRE_MODIFY_REG, HAVE_POST_MODIFY_REG,
3489 REGNO_MODE_OK_FOR_BASE_P, LEGITIMATE_CONSTANT_P,
3490 REG_MODE_OK_FOR_BASE_P, REG_OK_FOR_INDEX_P, GO_IF_LEGITIMATE_ADDRESS,
3491 LEGITIMIZE_ADDRESS, THUMB2_LEGITIMIZE_ADDRESS,
3492 GO_IF_MODE_DEPENDENT_ADDRESS, MEMORY_MOVE_COST, BRANCH_COST,
3493 ASM_APP_OFF, ASM_OUTPUT_CASE_LABEL, ARM_DECLARE_FUNCTION_NAME,
3494 FINAL_PRESCAN_INSN, PRINT_OPERAND_PUNCT_VALID_P,
3495 PRINT_OPERAND_ADDRESS): Adjust for Thumb-2.
3496 (arm_arch_notm, arm_arch_thumb2, arm_arch_hwdiv): New declarations.
3497 * config/arm/arm-cores.def: Add arm1156t2-s, cortex-a8, cortex-r4 and
3499 * config/arm/arm-tune.md: Regenerate.
3500 * config/arm/arm-protos.h: Update prototypes.
3501 * config/arm/vfp.md: Enable patterns for Thumb-2.
3502 (arm_movsi_vfp): Add movw alternative. Use output_move_vfp.
3503 (arm_movdi_vfp, movsf_vfp, movdf_vfp): Use output_move_vfp.
3504 (thumb2_movsi_vfp, thumb2_movdi_vfp, thumb2_movsf_vfp,
3505 thumb2_movdf_vfp, thumb2_movsfcc_vfp, thumb2_movdfcc_vfp): New.
3506 * config/arm/libunwind.S: Add Thumb-2 code.
3507 * config/arm/constraints.md: Update include Thumb-2.
3508 * config/arm/ieee754-sf.S: Add Thumb-2/Unified asm support.
3509 * config/arm/ieee754-df.S: Ditto.
3510 * config/arm/bpabi.S: Ditto.
3511 * config/arm/t-arm (MD_INCLUDES): Add thumb2.md.
3512 * config/arm/predicates.md (low_register_operand,
3513 low_reg_or_int_operand, thumb_16bit_operator): New.
3514 (thumb_cmp_operand, thumb_cmpneg_operand): Rename...
3515 (thumb1_cmp_operand, thumb1_cmpneg_operand): ... to this.
3516 * config/arm/t-arm-elf: Add armv7 multilib.
3517 * config/arm/arm.md: Update patterns for Thumb-2 and Unified asm.
3519 (UNSPEC_STACK_ALIGN, ce_count): New.
3520 (arm_incscc, arm_decscc, arm_umaxsi3, arm_uminsi3,
3521 arm_zero_extendsidi2, arm_zero_extendqidi2): New
3523 * config/arm/fpa.md: Update patterns for Thumb-2 and Unified asm.
3524 (thumb2_movsf_fpa, thumb2_movdf_fpa, thumb2_movxf_fpa,
3525 thumb2_movsfcc_fpa, thumb2_movdfcc_fpa): New insns.
3526 * config/arm/cirrus.md: Update patterns for Thumb-2 and Unified asm.
3527 (cirrus_thumb2_movdi, cirrus_thumb2_movsi_insn,
3528 thumb2_cirrus_movsf_hard_insn, thumb2_cirrus_movdf_hard_insn): New
3530 * doc/extend.texi: Document ARMv7-M interrupt functions.
3531 * doc/invoke.texi: Document Thumb-2 new cores+architectures.
3533 2007-01-03 Jakub Jelinek <jakub@redhat.com>
3535 * unwind-dw2.c (SIGNAL_FRAME_BIT, EXTENDED_CONTEXT_BIT): Define.
3536 (struct _Unwind_Context): Rename args_size to flags, remove
3537 signal_frame field, add a new args_size field and version field.
3538 (_Unwind_IsSignalFrame, _Unwind_SetSignalFrame,
3539 _Unwind_IsExtendedContext): New inline functions.
3540 (_Unwind_GetGR, _Unwind_SetGR, _Unwind_GetGRPtr, _Unwind_SetGRPtr):
3541 Assume by_value array is only present if _Unwind_IsExtendedContext.
3542 (_Unwind_GetIPInfo, execute_cfa_program, uw_frame_state_for): Use
3543 _Unwind_IsSignalFrame.
3544 (__frame_state_for): Initialize context.flags to EXTENDED_CONTEXT_BIT.
3545 (uw_update_context_1): Use _Unwind_SetSignalFrame.
3546 (uw_init_context_1): Initialize context->flags to
3547 EXTENDED_CONTEXT_BIT.
3548 * config/rs6000/linux-unwind.h (frob_update_context): Use
3549 _Unwind_SetSignalFrame.
3551 2007-01-03 Andrew Pinski <andrew_pinski@playstation.sony.com>
3554 * gimplify.c (gimplify_modify_expr_complex_part): Move below
3555 tree_to_gimple_tuple. Call tree_to_gimple_tuple when we need
3558 2007-01-03 Kazu Hirata <kazu@codesourcery.com>
3560 * config/i386/i386.h (NON_STACK_REG_P, REGNO_OK_FOR_SIREG_P,
3561 REGNO_OK_FOR_DIREG_P, REWRITE_ADDRESS, ASM_OPERAND_LETTER,
3562 RET, AT_SP): Remove.
3563 * config/i386/i386.md (*sse_prologue_save_insn): Use return
3566 * alias.c (init_alias_analysis): Use VEC_safe_grow_cleared.
3567 * cfgbuild.c (find_basic_blocks): Likewise.
3568 * cfgrtl.c (rtl_create_basic_block): Likewise.
3569 * function.c (temp_slots_at_level): Likewise.
3570 * reg-stack.c (stack_regs_mentioned): Likewise.
3571 * regclass.c (allocate_reg_info): Likewise.
3572 * tree-cfg.c (init_empty_tree_cfg, build_tree_cfg, create_bb,
3573 set_bb_for_stmt, move_block_to_fn): Likewise.
3574 * tree-complex.c (tree_lower_complex): Likewise.
3575 * vec.h (VEC_safe_grow_cleared): New.
3577 * cgraphunit.c, tree-ssa-alias.c: Fix comment typos.
3579 2007-01-03 Zdenek Dvorak <dvorakz@suse.cz>
3581 * loop-unswitch.c (unswitch_loop): Pass probabilities to loopify.
3582 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Pass probabilities
3584 * cfgloopmanip.c (scale_loop_frequencies): Export.
3585 (loopify): Scale the frequencies by prescribed coefficients.
3586 (set_zero_probability): New function.
3587 (duplicate_loop_to_header_edge): Improve updating of frequencies.
3588 (lv_adjust_loop_entry_edge, loop_version): Set probabilities
3589 and frequencies according to arguments.
3590 * tree-ssa-loop-manip.c (tree_unroll_loop): Set probabilities
3592 * cfg.c (scale_bbs_frequencies_int): Allow scaling the frequencies up.
3593 * modulo-sched.c (sms_schedule): Set probabilities for entering
3594 versioned loop correctly.
3595 * tree-vect-transform.c (vect_transform_loop): Ditto.
3596 * cfgloop.h (loopify, loop_version): Declaration changed.
3597 (scale_loop_frequencies): Declared.
3599 2007-01-02 Jan Hubicka <jh@suse.cz>
3601 * cgraph.c: Include tree-flow.h
3602 (cgraph_add_new-function): Handle IPA_SSA mode; execute
3604 * cgraph.h (enum cgraph_state): Add CGRAPH_STATE_IPA_SSA.
3605 * tree-pass.h (pass_all_early_optimizations): Declare.
3606 * cgraphunit.c (cgraph_process_new_functions): Add IPA_SSA; execute
3608 (cgraph_analyze_function): Do early_local_passes.
3609 * tree-mudflap.c (mf_decl_cache_locals, mf_build_check_statement_for):
3610 Do not add referenced vars.
3611 * tree-optimize.c (gate_all_optimizations): Do not execute when not in
3613 (gate_all_early_local_passes): New gate.
3614 (pass_early_local_passes): Use new gate.
3615 (execute_early_local_optimizations): New functions.
3616 (gate_all_early_optimizations): New gate.
3617 (pass_all_early_optimizations): New pass.
3618 (execute_free_datastructures): Free SSA only when initialized.
3619 (gate_init_datastructures): Init only when optimizing.
3620 (tree_lowering_passes): Do early local passes when called late.
3621 * tree-profile.c (do_tree_profiling): Don't profile functions added
3623 (do_early_tree_profiling, pass_early_tree_profile): Kill.
3624 * tree-cfg.c (update_modified_stmts): Do not update when operands are
3626 * passes.c (init_optimizations_passes): Reorder so we go into SSA
3627 during early_local_passes.
3628 * Makefile.in (cgraph.o): Add dependency on tree-flow.h.
3631 2007-01-02 Carlos O'Donell <carlos@codesourcery.com>
3633 * Makefile.in: Update copyright year.
3635 2007-01-02 Carlos O'Donell <carlos@codesourcery.com>
3637 * Makefile.in: Export GCC_EXEC_PREFIX before calling $(RUNTEST)
3638 in $(lang_checks) and check-consistency targets.
3640 2007-01-02 Jan Hubicka <jh@suse.cz>
3642 * tree-mudflap.c (mf_decl_cache_locals, mf_build_check_statement_for):
3643 Do not add referenced vars.
3644 * tree-cfg.c (update_modified_stmts): Do not update when SSA operands
3646 * passes.c (init_optimization_passes): Put mudflap_2 after
3647 free_datastructures.
3649 2007-01-02 Jan Hubicka <jh@suse.cz>
3651 * tree-optimize (execute_fixup_cfg): Set after_inlining flag.
3652 Set NOTHROW flag on call statements proved to be nothrow.
3653 Update statement of local calls so new pure/const functions are
3654 updated. Update_ssa when in ssa form. Mark PHI nodes of nonlocal
3656 (tree_rest_of_compilation): Register hooks and initialize bitmap
3657 early. Do not set after_inlining flag.
3659 2007-01-02 Steve Ellcey <sje@cup.hp.com>
3661 * sbitmap.c (HOST_BITS_PER_LONG_LONG): Change to
3662 HOST_BITS_PER_LONGLONG
3664 2007-01-02 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
3667 * c-typeck.c (store_init_value): Don't emit pedantic overflow
3668 warning for non-static initializers.
3670 2007-01-02 Steven Bosscher <steven@gcc.gnu.org>
3672 * config/alpha/alpha.md, arm/arm.c, darwin.c, frv/frv.md,
3673 m32r/m32r.c, m32r/m32r.c, mn10300/mn10300.md, pa/pa.c,
3674 rs6000/rs6000.c, s390/s390.md, sh/sh.md, sparc/sparc.c:
3675 Always use set_unique_reg_note to add REG_EQUAL notes.
3677 2007-01-02 Kazu Hirata <kazu@codesourcery.com>
3680 2007-01-02 Kazu Hirata <kazu@codesourcery.com>
3682 * alias.c (init_alias_analysis): Use VEC_safe_grow_cleared.
3683 * cfgbuild.c (find_basic_blocks): Likewise.
3684 * cfgrtl.c (rtl_create_basic_block): Likewise.
3685 * function.c (temp_slots_at_level): Likewise.
3686 * reg-stack.c (stack_regs_mentioned): Likewise.
3687 * regclass.c (allocate_reg_info): Likewise.
3688 * tree-cfg.c (init_empty_tree_cfg, build_tree_cfg, create_bb,
3689 set_bb_for_stmt, move_block_to_fn): Likewise.
3690 * tree-complex.c (tree_lower_complex): Likewise.
3691 * vec.h (VEC_safe_grow_cleared): New.
3693 2007-01-02 Ian Lance Taylor <iant@google.com>
3695 * c-common.c (c_common_truthvalue_conversion): When warning about
3696 using an assignment as a truth value, set TREE_NO_WARNING.
3698 2007-01-02 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
3701 * c.opt (Wold-style-declaration): New.
3702 * doc/invoke.texi (C-only Warning Options): New.
3703 (Wold-style-declaration): Document it.
3704 (Wextra): Enabled by -Wextra.
3705 * c-opts.c (c_common_post_options): Enabled by -Wextra.
3706 * c-decl.c (declspecs_add_scspec): Replace -Wextra with
3707 -Wold-style-declaration.
3709 2007-01-02 Kazu Hirata <kazu@codesourcery.com>
3711 * alias.c (init_alias_analysis): Use VEC_safe_grow_cleared.
3712 * cfgbuild.c (find_basic_blocks): Likewise.
3713 * cfgrtl.c (rtl_create_basic_block): Likewise.
3714 * function.c (temp_slots_at_level): Likewise.
3715 * reg-stack.c (stack_regs_mentioned): Likewise.
3716 * regclass.c (allocate_reg_info): Likewise.
3717 * tree-cfg.c (init_empty_tree_cfg, build_tree_cfg, create_bb,
3718 set_bb_for_stmt, move_block_to_fn): Likewise.
3719 * tree-complex.c (tree_lower_complex): Likewise.
3720 * vec.h (VEC_safe_grow_cleared): New.
3722 2007-01-02 Douglas Gregor <doug.gregor@gmail.com>
3724 * c-common.c (c_common_nodes_and_builtins): Since variants of
3725 void_type_node get built before it is given a name, we need to
3726 give those variants the name, too.
3727 (complete_array_type): We need to work with the canonical main
3728 type of the array, from which we will build the qualified version.
3729 * params.def (PARAM_VERIFY_CANONICAL_TYPES): New.
3730 * print-tree.c (print_node): Display canonical type information
3732 * stor-layout.c (layout_type): When we don't know the
3733 alignment of a type for which we're building an array, we end up
3734 guessing wrong, so make the type require structural equality.
3735 * tree.c (make_node_stat): When we build a new type, it is its
3737 (build_type_attribute_qual_variant): When building an attribute
3738 variant, its canonical type is the non-attribute variant. However,
3739 if the attributes are target-dependent and they differ, we need to
3740 use structural equality checks for this type.
3741 (build_qualified_type): A qualified type is not equivalent to its
3742 unqualified variant; set the canonical type appropriately.
3743 (build_distinct_type_copy): When building a distinct type from
3744 another type, the new type is its own canonical type.
3745 (build_variant_type_copy): When building a new type variant, we
3746 assume that it is equivalent to the original type.
3747 (build_pointer_type_for_mode): When building a pointer type, also
3748 build a canonical type pointer.
3749 (build_reference_type_for_mode): When building a reference type,
3750 also build a canonical type reference.
3751 (build_index_type): When we can't hash an index type (e.g.,
3752 because its maximum value is negative), the index type requires
3753 structural equality tests.
3754 (build_array_type): Build the canonical form of an array type.
3755 (build_function_type): Function types require structural equality,
3756 because they contain default arguments, attributes, etc.
3757 (build_method_type_directly): Ditto for method types.
3758 (build_offset_type): Build the canonical offset type.
3759 (build_complex_type): Build the canonical vector type.
3760 (make_vector_type): Build the canonical vector type.
3761 * tree.h (TYPE_CANONICAL): New.
3762 (TYPE_STRUCTURAL_EQUALITY_P): New.
3763 (SET_TYPE_STRUCTURAL_EQUALITY): New.
3764 (struct tree_type): Added "canonical" field.
3765 * params.h (VERIFY_CANONICAL_TYPES): New.
3766 * doc/c-tree.texi (TYPE_CANONICAL): Document.
3767 (TYPE_STRUCTURAL_EQUALITY_P): Document.
3768 (SET_TYPE_STRUCTURAL_EQUALITY): Document.
3769 * doc/invoke.texi (verify-canonical-types): Document --param
3770 parameter for verifying canonical types.
3772 2007-01-02 Joseph Myers <joseph@codesourcery.com>
3774 * config.gcc (powerpc-*-eabispe*, powerpc-*-eabisimaltivec*,
3775 powerpc-*-eabisim*, powerpc-*-eabialtivec*, powerpc-*-eabi*,
3776 powerpc-*-rtems*, powerpc-wrs-vxworks, powerpc-wrs-vxworksae,
3777 powerpcle-*-eabisim*, powerpcle-*-eabi*): Add rs6000/e500.h to
3779 * config/rs6000/e500.h: New.
3780 * config/rs6000/eabi.h (TARGET_SPE_ABI, TARGET_SPE, TARGET_E500,
3781 TARGET_ISEL, TARGET_FPRS, TARGET_E500_SINGLE, TARGET_E500_DOUBLE):
3783 * config/rs6000/linuxspe.h (TARGET_SPE_ABI, TARGET_SPE,
3784 TARGET_E500, TARGET_ISEL, TARGET_FPRS, TARGET_E500_SINGLE,
3785 TARGET_E500_DOUBLE): Remove.
3786 * config/rs6000/vxworks.h (TARGET_SPE_ABI, TARGET_SPE,
3787 TARGET_E500, TARGET_ISEL, TARGET_FPRS): Remove.
3788 * config/rs6000/rs6000.h (CHECK_E500_OPTIONS): Define.
3789 * config/rs6000/rs6000.c (rs6000_override_options): Use
3792 2007-01-02 Joseph Myers <joseph@codesourcery.com>
3794 * config/rs6000/rs6000.c (print_operand): Check (TARGET_SPE ||
3795 TARGET_E500_DOUBLE), not TARGET_E500, for %y.
3796 (rs6000_generate_compare, rs6000_emit_sCOND, output_cbranch,
3797 rs6000_emit_cmove): Don't check TARGET_E500.
3798 * config/rs6000/rs6000.md (bunordered, bordered, sunordered,
3799 sordered): Don't check TARGET_E500.
3801 2007-01-01 Eric Christopher <echristo@apple.com>
3803 * config/mips/mips.c (mips_regno_mode_ok_for_base_p): Use
3804 HARD_REGISTER_NUM_P.
3806 2007-01-01 Roger Sayle <roger@eyesopen.com>
3808 * fold-const.c (fold_binary) <EQ_EXPR>: Fold "(X^C1) eq/ne C2" into
3809 "X eq/ne (C1^C2)". Fold "(X^Z) eq/ne (Y^Z)" as "X eq/ne Y" when Z
3810 has no side-effects. Fold "(X^C1) eq/ne (Y^C2)" as "(X^(C1^C2))
3813 2007-01-01 Mike Stump <mrs@apple.com>
3815 * configure.ac: Remove support for building with Apple's gcc-3.1.
3817 2007-01-02 Joseph Myers <joseph@codesourcery.com>
3820 * caller-save.c (add_stored_regs): Only handle SUBREGs if inner
3821 REG is a hard register. Do not modify REG before calling
3823 * rtlanal.c (subreg_get_info): Don't assert size of XMODE is a
3824 multiple of the size of YMODE for certain lowpart cases.
3826 2007-01-01 Andrew Pinski <pinskia@gmail.com>
3829 * gimplify (voidify_wrapper_expr): Update for
3832 2007-01-01 Andreas Schwab <schwab@suse.de>
3835 * config/ia64/ia64.c (ia64_compute_frame_size): Account space for
3836 save of BR0 in extra_spill_size instead of spill_size.
3837 (ia64_expand_prologue): Save BR0 outside of the gr/br/fr spill
3839 (ia64_expand_epilogue): Restore BR0 from its new location.
3841 2007-01-01 Andrew Pinski <pinskia@gmail.com>
3843 * gimplify.c (gimplify_init_constructor <case VECTOR_TYPE>):
3844 Use a temporary variable if the left hand side is not a gimple
3847 2007-01-01 Andrew Pinski <pinskia@gmail.com>
3849 * gimplify.c (gimplify_return_expr): Make the temporary variable
3850 for the return expression, a gimple register variable.
3852 2007-01-01 Jan Hubicka <jh@suse.cz>
3854 * emit-rtl.c (emit_copy_of_insn_after): Do not call copy_insn_1 for
3857 2007-01-01 Mike Stump <mrs@apple.com>
3859 * configure.ac (HAVE_GAS_LITERAL16): Add autoconf check for
3861 * config/darwin.c (machopic_select_rtx_section): Use
3863 (darwin_mergeable_constant_section): Likewise.
3864 * configure: Regenerate.
3865 * config.in: Regenerate.
3867 2007-01-01 Jan Hubicka <jh@suse.cz>
3868 Andrew Pinski <pinskia@gmail.com>
3870 * cgraphunit.c (cgraph_optimize): Call cgraph_add_new_functions
3871 before starting IPA passes.