1 2007-02-03 Ian Lance Taylor <iant@google.com>
3 * lower-subreg.c (simple_move_operand): New static function,
4 broken out of simple_move. Reject LABEL_REF, SYMBOL_REF, and HIGH
6 (simple_move): Call simple_move_operand.
7 (find_decomposable_subregs): Add special handling of MEMs.
8 (can_decompose_p): Rename from cannot_decompose_p. Reverse
9 meaning of return value. If we see a hard register, test whether
10 it can store a word_mode value. Change all callers.
12 2007-02-03 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
14 * pa.md (addvdi3, addvsi3, subvdi3, subvsi3, negvdi2, negvsi2): New
15 ftrapv insns and expanders.
16 (subdi3): Change define_expand operand 1 to arith11_operand, and
17 operand 2 to reg_or_0_operand. Change constraints of 64-bit insn
18 pattern to handle reg_or_0 operands. Revise 32-bit insn pattern to
19 handle 11-bit constants and reg_or_0 operands in operands 1 and 2,
23 * varasm.c (notice_global_symbol): Treat global objects as weak when
26 2007-02-03 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
28 * emit-rtl.c (dconstpi): Delete.
30 (init_emit_once): Delete dconstpi and init dconstsqrt2.
31 * real.h (dconstpi): Delete.
33 * builtins.c (fold_builtin_cabs): Use dconstsqrt2.
34 (fold_builtin_hypot): Likewise.
36 2007-02-03 Tom Tromey <tromey@redhat.com>
39 * gcc.c (cpp_unique_options): Any of -ggdb3, -gstabs3,
40 -gcoff3, -gxcoff3, -gvms3 implies -dD.
42 2007-02-03 Kazu Hirata <kazu@codesourcery.com>
44 * c-decl.c, config/avr/avr.c, config/avr/avr.h,
45 config/m68k/m68k.c, config/m68k/netbsd-elf.h,
46 config/mn10300/mn10300.c, config/pdp11/pdp11.h,
47 config/rs6000/cell.md, config/rs6000/darwin.h,
48 config/sh/sh.md, config/sh/sh4-300.md, config/spu/spu.c,
49 config/spu/spu.md, cselib.c, expr.c, haifa-sched.c, hwint.h,
50 jump.c, reload.c, sched-deps.c, sched-int.h, tree-inline.c,
51 tree-profile.c, tree-ssa-live.h, tree-vrp.c: Fix comment
52 typos. Follow spelling conventions.
53 * doc/invoke.texi: Follow spelling conventions.
55 2007-02-03 Roger Sayle <roger@eyesopen.com>
57 * simplify-rtx.c (simplify_relational_operation_1): Implement some
58 canonicalization transformations that attempt to simplify integer
59 constant comparisons to become comparisons against zero.
61 2007-02-03 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
63 * builtins.c (fold_builtin_cabs): Fold cabs(x+xi) into
65 * fold-const.c (fold_binary): Fix comment typos. Fold complex
66 (x,0)-(0,y) into (x,-y). Likewise (0,y)-(x,0) into (-x,y).
68 2007-02-02 Mike Stump <mrs@apple.com>
70 * config/darwin9.h (DARWIN_LINKER_GENERATES_ISLANDS): Add.
71 * config/rs6000/rs6000.c (DARWIN_GENERATE_ISLANDS): Add.
72 (output_call): Use DARWIN_GENERATE_ISLANDS to decide when to
73 generate a branch island.
75 2007-02-02 Bob Wilson <bob.wilson@acm.org>
77 * config/xtensa/xtensa.c (smalloffset_mem_p): Use BASE_REG_P.
78 (xtensa_legitimate_address_p): New.
79 (xtensa_legitimize_address): New.
80 (xtensa_output_addr_const_extra): New.
81 * config/xtensa/xtensa.h (REG_OK_STRICT_FLAG): Define.
83 (REG_OK_FOR_BASE_P): Use BASE_REG_P.
84 (GO_IF_LEGITIMATE_ADDRESS): Move code to xtensa_legitimate_address_p.
85 (LEGITIMIZE_ADDRESS): Move code to xtensa_legitimize_address.
86 (OUTPUT_ADDR_CONST_EXTRA): Move code to xtensa_output_addr_const_extra.
87 * config/xtensa/xtensa-protos.h (xtensa_legitimate_address_p): New.
88 (xtensa_legitimize_address): New.
89 (xtensa_output_addr_const_extra): New.
91 2007-02-02 Steve Ellcey <sje@cup.hp.com>
93 * config/ia64/ia64.c (ia64_print_operand): Fix compare strings.
95 2007-02-02 Ian Lance Taylor <iant@google.com>
97 * expmed.c (expand_divmod): Add comment.
99 2007-02-02 Kazu Hirata <kazu@codesourcery.com>
101 * emit-rtl.c (renumber_insns): Remove.
102 * flags.h: Remove the extern for flag_renumber_insns.
103 * rtl.h: Remove the prototype for renumber_insns.
104 * toplev.c (flag_renumber_insns): Remove.
106 2007-02-02 Hui-May Chang <hm.chang@apple.com>
108 Revert for x86 darwin:
109 2005-06-19 Uros Bizjak <uros@kss-loka.si>
111 * config/i386/i386.c (ix86_function_arg_regno_p): Put back the
112 code before the following patch under TARGET_MACHO.
113 (ix86_function_value_regno_p): Likewise.
115 2007-02-02 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
117 * fold-const.c (negate_expr_p, fold_negate_expr): Handle
120 2007-02-02 Maxim Kuvyrkov <mkuvyrkov@ispras.ru>
123 * config/ia64/ia64.c (ia64_speculate_insn): Restrict to memory loads to
124 general or fp registers. Add comments.
125 * config/ia64/ia64.md (reg_pred_prefix): Add comment.
127 2007-02-02 Paolo Bonzini <bonzini@gnu.org>
129 * pointer-set.c (insert_aux): Only return insertion slot.
130 (pointer_set_insert): Adjust.
131 (pointer_set_traverse, struct pointer_map_t, pointer_map_create,
132 pointer_map_destroy, pointer_map_insert, pointer_map_contains,
133 pointer_map_traverse): New.
134 * pointer-set.h (pointer_set_traverse, struct pointer_map_t,
135 pointer_map_create, pointer_map_destroy, pointer_map_insert,
136 pointer_map_contains, pointer_map_traverse): Declare.
138 2007-02-02 Jakub Jelinek <jakub@redhat.com>
141 * builtins.c (fold_builtin_sprintf): Do not attempt to optimize
142 sprintf (str, "%s"). Do not optimize sprintf (str, "nopercent", p++).
144 2007-02-02 Maxim Kuvyrkov <mkuvyrkov@ispras.ru>
146 * sched-int.h (ds_to_dk, dk_to_ds): Declare functions.
148 (struct _dep): New type.
149 (dep_t): New typedef.
150 (DEP_PRO, DEP_CON, DEP_KIND): New access macros.
151 (DEP_STATUS): New access macro. The macro with the same name was
152 renamed to DEP_LINK_STATUS.
153 (dep_init): Declare function
155 (struct _dep_link): New type.
156 (dep_link_t): New typedef.
157 (DEP_LINK_NODE, DEP_LINK_NEXT, DEP_LINK_PREV_NEXTP): New access macros.
158 (DEP_LINK_DEP, DEP_LINK_PRO, DEP_LINK_CON, DEP_LINK_KIND): New macros.
159 (DEP_LINK_STATUS): New macro.
160 (debug_dep_links): New debug function.
162 (struct _deps_list): New type.
163 (deps_list_t): New typedef.
164 (DEPS_LIST_FIRST): New access macro.
165 (FOR_EACH_DEP_LINK): New cycle macro.
166 (create_deps_list, free_deps_list, delete_deps_list): Declare
168 (deps_list_empty_p, debug_deps_list, add_back_dep_to_deps_list): Ditto.
169 (find_link_by_pro_in_deps_list, find_link_by_con_in_deps_list): Ditto.
170 (copy_deps_list_change_con): Ditto.
172 (move_dep_link): Declare function.
174 (struct _dep_node): New type.
175 (dep_node_t): New typedef.
176 (DEP_NODE_BACK, DEP_NODE_DEP, DEP_NODE_FORW): New access macros.
178 (struct haifa_insn_data.back_deps): New field to hold backward
179 dependencies of the insn.
180 (struct haifa_insn_data.depend): Rename to forw_deps. Change its type
182 (struct haifa_insn_data.resolved_deps): Rename to resolved_back_deps.
183 Change its type to deps_list_t.
184 (INSN_BACK_DEPS): New access macro to use instead of LOG_LINKS.
185 (INSN_DEPEND): Rename to INSN_FORW_DEPS.
186 (RESOLVED_DEPS): Rename to INSN_RESOLVED_BACK_DEPS.
188 (INSN_COST): Move to haifa-sched.c. Use insn_cost () instead.
190 (DEP_STATUS): Rename to DEP_LINK_STATUS. Fix typo in the comment.
192 (add_forw_dep, delete_back_forw_dep, insn_cost): Update declaration and
196 * sched-deps.c (CHECK): New macro to (en/dis)able sanity checks.
197 (ds_to_dk, dk_to_ds): New functions.
199 (init_dep_1): New static function.
200 (init_dep): New function.
201 (copy_dep): New static function.
203 (dep_link_consistent_p, attach_dep_link, add_to_deps_list): New static
205 (detach_dep_link): New static function.
206 (move_dep_link): New function.
208 (dep_links_consistent_p, dump_dep_links): New static functions.
209 (debug_dep_links): New debugging function.
211 (deps_obstack, dl_obstack, dn_obstack): New static variables.
213 (alloc_deps_list, init_deps_list): New static functions.
214 (create_deps_list): New function.
215 (clear_deps_list): New static function.
216 (free_deps_list, delete_deps_list, deps_list_empty_p): New functions.
217 (deps_list_consistent_p, dump_deps_list): New static functions.
218 (debug_deps_list): New function.
219 (add_back_dep_to_deps_list, find_link_by_pro_in_deps_list): New
221 (find_link_by_con_in_deps_list, copy_deps_list_change_con): Ditto.
223 (maybe_add_or_update_back_dep_1, add_or_update_back_dep_1): Update to
224 use new scheduler dependencies lists.
225 (add_back_dep, delete_all_dependences, fixup_sched_groups): Ditto.
226 (sched_analyze): Ditto. Initialize dependencies lists.
227 (add_forw_dep, compute_forward_dependences): Update to use new
228 scheduler dependencies lists.
230 (init_dependency_caches): Init deps_obstack.
231 (free_dependency_caches): Free deps_obstack.
233 (adjust_add_sorted_back_dep, adjust_back_add_forw_dep): Update to use
234 new scheduler dependencies lists.
235 (delete_forw_dep, add_or_update_back_forw_dep): Ditto.
236 (add_back_forw_dep, delete_back_forw_dep): Ditto.
238 * sched-rgn.c (set_spec_fed, find_conditional_protection, is_pfree):
239 Update to use new scheduler dependencies lists.
240 (is_conditionally_protected, is_prisky, add_branch_dependences): Ditto.
241 (debug_dependencies): Ditto.
242 (schedule_region): Update comments.
244 * sched-ebb.c (earliest_block_with_similiar_load): Update to use new
245 scheduler dependencies lists.
246 (schedule_ebb): Update comments.
248 * rtl.def (DEPS_LIST): Remove.
250 * lists.c (unused_deps_list): Remove.
251 (free_list): Update assertions.
253 (alloc_DEPS_LIST, free_DEPS_LIST_list, free_DEPS_LIST_node): Remove.
254 (remove_free_DEPS_LIST_elem, copy_DEPS_LIST_list): Ditto.
256 * rtl.h (free_DEPS_LIST_list, alloc_DEPS_LIST): Remove declarations.
257 (remove_free_DEPS_LIST_elem, copy_DEPS_LIST_list): Ditto.
259 * haifa-sched.c (comments): Update.
260 (insn_cost1): Remove. Inline the code into insn_cost ().
261 (insn_cost): Update to use new scheduler dependencies lists. Move
262 processing of the dependency cost to dep_cost ().
263 (dep_cost): New function. Use it instead of insn_cost () when
264 evaluating cost of the dependency. Use compatible interface to
265 interact with the target.
266 (priority): Update to use new scheduler dependencies lists.
267 (rank_for_schedule): Ditto. Optimize heuristic that prefers the insn
268 with greater number of insns that depend on the insn.
269 (schedule_insn): Update to use new scheduler dependencies lists. Add
270 code to free backward dependencies lists. Inline and optimize code
271 from resolve_dep () - see PR28071.
272 (ok_for_early_queue_removal): Update to use new scheduler dependencies
273 lists. Update call to targetm.sched.is_costly_dependence hook.
275 (fix_inter_tick, try_ready, fix_tick_ready): Update to use new
276 scheduler dependencies lists.
278 (resolve_dep): Remove. Move the logic to schedule_insn ().
279 (init_h_i_d): Initialize dependencies lists.
281 (process_insn_depend_be_in_spec): Rename to
282 process_insn_forw_deps_be_in_spec. Update to use new scheduler
284 (add_to_speculative_block, create_check_block_twin, fix_recovery_deps):
285 Update to use new scheduler dependencies lists.
286 (clear_priorities, calc_priorities, add_jump_dependencies): Ditto.
288 * ddg.c (create_ddg_dependence, create_ddg_dep_no_link): Update to use
289 new scheduler dependencies lists.
290 (build_intra_loop_deps): Ditto.
292 * target.h (struct _dep): Declare to use in
293 gcc_target.sched.is_costly_dependence.
294 (struct gcc_target.sched.adjust_cost): Fix typo.
295 (struct gcc_target.sched.is_costly_dependence): Change signature to use
296 single dep_t parameter instead of an equivalent triad.
297 (struct gcc_target.sched.adjust_cost_2): Remove.
299 * target-def.h (TARGET_SCHED_ADJUST_COST_2): Remove.
301 * reg-notes.def (DEP_TRUE, DEP_OUTPUT, DEP_ANTI): Update comments.
303 * doc/tm.texi (TARGET_SCHED_IS_COSTLY_DEPENDENCE): Update
305 (TARGET_SCHED_ADJUST_COST_2): Remove documentation.
307 * doc/rtl.texi (LOG_LINKS): Remove part about instruction scheduler.
308 (REG_DEP_TRUE): Document.
310 * config/ia64/ia64.c (ia64_adjust_cost_2): Rename to ia64_adjust_cost.
311 Change signature to correspond to the targetm.sched.adjust_cost hook.
312 Update use in TARGET_SCHED_ADJUST_COST_2.
313 (TARGET_SCHED_ADJUST_COST_2): Rename to TARGET_SCHED_ADJUST_COST.
314 (ia64_dependencies_evaluation_hook, ia64_dfa_new_cycle): Update to use
315 new scheduler dependencies lists.
316 (ia64_gen_check): Ditto.
318 * config/mips/mips.c (vr4130_swap_insns_p): Update to use new scheduler
321 * config/rs6000/rs6000.c (rs6000_is_costly_dependence): Change
322 signature to correspond to the targetm.sched.is_costly_dependence hook.
323 (is_costly_group): Update to use new scheduler dependencies lists.
325 * config/spu/spu.c (spu_sched_adjust_cost): Use insn_cost () function
326 instead of INSN_COST () macro.
328 2007-02-01 Ian Lance Taylor <iant@google.com>
330 * lower-subreg.c (resolve_clobber): Handle a subreg of a concatn.
332 2007-02-01 Guy Martin <gmsoft@gentoo.org>
334 * pa.md (tp_load): Correct mfctl instruction syntax.
336 2007-02-01 Geoffrey Keating <geoffk@apple.com>
338 * config/rs6000/rs6000.c (rs6000_stack_info): Correct
339 altivec_padding_size calculation on AIX. Improve comment, add
340 assert to verify that it's right.
342 * config/rs6000/darwin.h (MD_UNWIND_SUPPORT): Don't define for
345 2007-2-01 Seongbae Park <seongbae.park@gmail.com>
348 * global.c (global_alloc): Add mising initialization of
351 2007-02-01 Roger Sayle <roger@eyesopen.com>
353 * alias.c (init_alias_analysis): Correct whitespace.
354 * bb-reorder.c (fix_edges_for_rarely_executed_code,
355 partition_hot_cold_basic_blocks): Likewise.
356 * builtins.c (expand_builtin_printf, expand_builtin_fprintf,
357 expand_builtin_sprintf, fold_builtin_carg, fold_builtin_sprintf,
358 maybe_emit_sprintf_chk_warning, fold_builtin_sprintf_chk,
359 fold_builtin_snprintf_chk, fold_builtin_printf,
360 fold_builtin_fprintf, do_mpfr_ckconv, do_mpfr_arg1, do_mpfr_arg2,
361 do_mpfr_arg3, do_mpfr_sincos): Likewise.
362 * cfgcleanup.c (cleanup_cfg): Likewise.
363 * cfgexpand.c (tree_expand_cfg): Likewise.
364 * fold-const.c (fold_binary) <RDIV_EXPR>: Likewise.
365 * function.c (get_next_funcdef_no): Likewise.
366 * gengtype.c (main): Likewise.
367 * genmodes.c (main): Likewise.
368 * gcse.c (bypass_conditional_jumps, print_ldst_list): Likewise.
369 * haifa-sched.c (schedule_block, extend_h_i_d): Likewise.
370 * ifcvt.c (noce_emit_move_insn): Likewise.
371 * modulo-sched.c (generate_prolog_epilog, sms_schedule_by_order):
373 * stor-layout.c (get_best_mode): Likewise.
374 * tree-ssa-loop-niter.c (get_val_for): Likewise.
375 * tree-ssa-structalias.c (get_varinfo, get_varinfo_fc,
376 scc_visit, do_ds_constraint, do_complex_constraint, label_visit,
377 perform_var_substitution, solve_graph): Likewise.
378 * tree-vrp.c (vrp_finalize): Likewise.
380 2007-02-01 Ian Lance Taylor <iant@google.com>
382 * lower-subreg.c (simplify_gen_subreg_concatn): If we ask for the
383 high part of a paradoxical subreg, return a constant zero.
385 2007-02-01 Zdenek Dvorak <dvorakz@suse.cz>
387 * toplev.c (lang_dependent_init): Call init_set_costs.
388 * loop-init.c (loop_optimizer_init): Do not call init_set_costs.
390 2007-02-01 Richard Guenther <rguenther@suse.de>
393 * fold-const.c (fold_negate_expr): Allow negating a
394 constant if overflow does not change.
396 2007-02-01 Ralf Wildenhues <Ralf.Wildenhues@gmx.de>
398 * doc/c-tree.texi (Expression trees): Improve markup.
399 * doc/tm.texi (Register Classes, Addressing Modes)
400 (Floating Point): Fix spacing after abbreviations. Fix some
403 2007-02-01 Ben Elliston <bje@au.ibm.com>
405 * doc/invoke.texi: Replace "bugfix" with "bug fix" throughout.
406 * doc/contrib.texi: Likewise.
407 * doc/install.texi: Likewise.
409 2007-01-31 Richard Henderson <rth@redhat.com>
410 Ian Lance Taylor <iant@google.com>
412 * lower-subreg.c: New file.
413 * rtl.def (CONCATN): Define.
414 * passes.c (init_optimization_passes): Add pass_lower_subreg and
416 * emit-rtl.c (update_reg_offset): New static function, broken out
417 of gen_rtx_REG_offset.
418 (gen_rtx_REG_offset): Call update_reg_offset.
419 (gen_reg_rtx_offset): New function.
420 * regclass.c: Revert patch of 2006-03-05, restoring
422 (clear_reg_info_regno): New function.
423 * dwarf2out.c (concatn_loc_descriptor): New static function.
424 (loc_descriptor): Handle CONCATN.
425 * common.opt (fsplit_wide_types): New option.
426 * opts.c (decode_options): Set flag_split_wide_types when
428 * timevar.def (TV_LOWER_SUBREG): Define.
429 * rtl.h (gen_reg_rtx_offset): Declare.
430 (reg_scan_update): Declare.
431 * regs.h (clear_reg_info_regno): Declare.
432 * tree-pass.h (pass_lower_subreg): Declare.
433 (pass_lower_subreg2): Declare.
434 * doc/invoke.texi (Option Summary): List -fno-split-wide-types.
435 (Optimize Options): Add -fsplit-wide-types to -O1 list. Document
437 * doc/rtl.texi (Regs and Memory): Document concat and concatn.
438 * Makefile.in (OBJS-common): Add lower-subreg.o.
439 (lower-subreg.o): New target.
441 2007-01-31 Kazu Hirata <kazu@codesourcery.com>
443 * config/sh/sh.h (HAVE_SECONDARY_RELOADS): Remove.
445 2007-01-31 Anatoly Sokolov <aesok@post.ru>
448 * config/avr/avr.c (DWARF2_ADDR_SIZE): Define.
450 2007-01-31 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
453 * builtins.c (fold_builtin_sqrt): Use MPFR for constant args.
455 2007-01-31 Zdenek Dvorak <dvorakz@suse.cz>
457 * cfgloop.h: Include vec-prim.h.
458 (enum li_flags): Remove LI_ONLY_OLD.
459 (loop_iterator): Changed.
460 (fel_next, fel_init): Iterate over loop tree.
461 (FOR_EACH_LOOP_BREAK): New macro.
462 * loop-unswitch.c (unswitch_loops): Do not pass LI_ONLY_OLD to
464 * tree-ssa-loop-unswitch.c (tree_ssa_unswitch_loops): Ditto.
465 * modulo-sched.c (sms_schedule): Ditto.
466 * tree-vectorizer.c (vectorize_loops): Ditto.
467 * doc/loop.texi: Update information on loop numbering and behavior of
468 FOR_EACH_LOOP wrto new loops.
469 * tree-scalar-evolution.c (compute_overall_effect_of_inner_loop,
470 add_to_evolution_1): Test nestedness of loops instead of comparing
472 * tree-chrec.c (chrec_fold_plus_poly_poly,
473 chrec_fold_multiply_poly_poly, chrec_evaluate,
474 hide_evolution_in_other_loops_than_loop, chrec_component_in_loop_num,
475 reset_evolution_in_loop): Ditto.
476 * Makefile.in (CFGLOOP_H): Add vecprim.h dependency.
478 2007-01-31 Dirk Mueller <dmueller@suse.de>
480 * c-common.c (warn_about_parentheses): Separate warning about
481 un-parenthized sequence of comparison operators from the one
482 which is supposed to warn about x <= y <= z.
484 2007-01-31 Uros Bizjak <ubizjak@gmail.com>
486 * optabs.h (enum optab_index): Add new OTI_isinf.
487 (isinf_optab): Define corresponding macro.
488 * optabs.c (init_optabs): Initialize isinf_optab.
489 * genopinit.c (optabs): Implement isinf_optab using isinf?f2
491 * builtins.c (mathfn_built_in): Handle BUILT_IN_ISINF{,F,L}.
492 (expand_builtin_interclass_mathfn): Expand BUILT_IN_ISINF{,F,L}
494 (expand_builtin): Expand BUILT_IN_ISINF{,F,L} using
495 expand_builtin_interclass_mathfn.
496 * reg_stack.c (subst_stack_regs_pat): Handle UNSPEC_FXAM.
497 * config/i386/i386.md (UNSPEC_FXAM): New constant.
498 (fxam<mode>2_i387): New insn pattern.
499 (isinf<mode>2) New expander to implement isinf, isinff and isinfl
500 built-in functions as x87 inline asm.
502 2007-01-31 Kazu Hirata <kazu@codesourcery.com>
504 * gcc/config/arm/unwind-arm.h (_sleb128_t, _uleb128_t): New.
506 2007-01-30 Eric Christopher <echristo@apple.com>
508 * config.gcc: Add geode.
510 2007-01-31 Kazu Hirata <kazu@codesourcery.com>
512 * cgraphunit.c, config/arm/arm.c, config/m68k/m68k.c,
513 ipa-inline.c, tree-profile.c, tree-ssa-live.c,
514 tree-ssa-math-opts.c, tree-ssanames.c, tree-vect-analyze.c,
515 value-prof.c: Fix comment typos.
517 2007-01-30 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
520 * doc/invoke.texi (Wpointer-arith): Document warning.
522 2007-01-30 Janis Johnson
524 * doc/extend.texi (Decimal Floating Types): Remove decfloat.h from
525 the list of discrepancies from the draft TR.
527 2007-01-30 Dirk Mueller <dmueller@suse.de>
530 * doc/invoke.texi (-Wreturn-type): Update description to
533 2007-01-30 Richard Sandiford <richard@codesourcery.com>
535 * cfgrtl.c (try_redirect_by_replacing_jump): Check only_sets_cc0_p.
537 2007-01-30 Uros Bizjak <ubizjak@gmail.com>
539 * builtins.c (expand_builtin_int_interclass_roundingfn): New function
540 to handle optabs that operate on floating point input argument and
541 output to integer output.
542 (expand_builtin_mathfn) [BUILT_IN_ILOGB]: Move from here ...
543 (expand_builtin_interclass_mathfn) [BUILT_IN_ILOGB]: ... to here.
544 (expand_builtin): Expand BUILT_IN_ILOGB{,F,L} using
545 expand_builtin_interclass_mathfn ().
546 * config/i386/i386.md (fxtractxf3_i387): Rename from *fxtractxf3_i387.
548 (ilogbxf2, ilogb<mode>2): New expanders to implement ilogb, ilogbf and
549 ilogbl built-in functions as x87 intrinsics.
551 2007-01-30 Richard Guenther <rguenther@suse.de>
554 * dwarf2out.c (reference_to_unused): Query varpool if the
557 2007-01-30 Richard Guenther <rguenther@suse.de>
560 * passes.c (execute_one_pass): Reset in_gimple_form to not
561 confuse non-unit-at-a-time mode.
563 2007-01-29 Roger Sayle <roger@eyesopen.com>
564 Richard Guenther <rguenther@suse.de>
566 * fold-const.c (round_up): Make HIGH an unsigned HOST_WIDE_INT to
567 avoid undefined behaviour on overflow. Use force_fit_type_double
568 to construct the constant with the specified TREE_OVERFLOW.
570 2007-01-29 Janis Johnson <janis187@us.ibm.com>
572 * config/dfp-bit.c: Add parameterized support for fp exceptions.
573 * config/dfp-bit.h: Ditto.
575 2007-01-29 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
577 * c-decl.c (pop_scope): Replace warnings with call to
578 warn_for_unused_label.
579 * c-common.h (warn_for_unused_label): Declare.
580 * c-common.c (warn_for_unused_label): Define.
582 2007-01-29 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
584 * tree-optimize.c (update_inlined_to_pointers): Delete unused
587 2007-01-29 Janis Johnson <janis187@us.ibm.com>
589 * Makefile.in (USER_H): Remove decfloat.h.
590 * ginclude/decfloat.h: Delete, moving contents to ...
591 * ginclude/float.h: Add support for decimal floating point,
592 guarded by __STDC_WANT_DEC_FP__.
594 2007-01-29 Mike Stump <mrs@apple.com>
596 * doc/gccint.texi (Top): Rename Loop Representation to Loop
597 Analysis and Representation to resolve case insensitive conflict.
598 * doc/loop.texi (Loop Analysis and Representation): Likewise.
600 2007-01-28 Daniel Berlin <dberlin@dberlin.org>
602 * tree.h (struct tree_memory_tag): Add aliases member.
603 (MTAG_ALIASES): New macro.
604 * tree-ssa-alias.c (alias_bitmap_obstack): New variable.
605 (add_may_alias): Remove pointer-set. Update for may_aliases being
607 (mark_aliases_call_clobbered): Update for may_aliases being a
609 (compute_tag_properties): Ditto.
610 (create_partition_for): Ditto.
611 (compute_memory_partitions): Ditto.
612 (dump_may_aliases_for): Ditto.
613 (is_aliased_with): Ditto.
614 (add_may_alias_for_new_tag): Ditto.
615 (rewrite_alias_set_for): Rewrite for may_aliases being a bitmap.
616 (compute_is_aliased): New function.
617 (compute_may_aliases): Call compute_is_aliased).
618 (init_alias_info): Initialize alias_bitmap_obstack.
619 (union_alias_set_into): New function.
620 (compute_flow_sensitive_aliasing): Use union_aliases_into.
621 (have_common_aliases_p): Rewrite to take two bitmaps and use
623 (compute_flow_insensitive_aliasing): Stop using pointer-sets.
625 (finalize_ref_all_pointers): Update for add_may_alias changes.
626 (new_type_alias): Ditto.
627 * tree-flow-inline.h (may_aliases): Return a bitmap.
628 * tree-dfa.c (dump_variable): Check for MTAG_P'ness.
629 * tree-ssa.c (verify_flow_insensitive_alias_info): Update for
630 may_aliases being a bitmap.
631 * tree-flow.h (struct var_ann_d): Remove may_aliases member.
632 may_aliases now returns a bitmap.
633 * tree-ssa-structalias.c (merge_smts_into): Update for may_aliases
635 * tree-ssa-operands.c (add_virtual_operand): Update for
636 may_aliases being a bitmap.
638 2007-01-29 Daniel Berlin <dberlin@dberlin.org>
640 PR tree-optimization/30630
641 * tree-ssa-structalias.c (do_complex_constraint): Mark correct
644 2007-01-29 Simon Martin <simartin@users.sourceforge.net>
647 * gimplify.c (gimplify_target_expr): Make sure that the TARGET_EXPR is
648 expanded only once even if an error occurs.
650 2007-01-29 Ben Elliston <bje@au.ibm.com>
652 * gcov-io.h (__gcov_indirect_call_profiler): Declare.
653 (__gcov_average_profiler): Likewise.
654 (__gcov_ior_profiler): Likewise.
655 (__gcov_merge_ior): Likewise.
657 2007-01-28 Jan Hubicka <jh@suse.cz>
659 * builtins.c (expand_builtin_memset): Fix typo in my last patch.
660 * value-prof.c (stringop_block_profile): Likewise.
662 2007-01-28 Jan Hubicka <jh@suse.cz>
664 * expr.c (emit_block_move_via_movmem, emit_block_move_via_libcall): Add
665 variant handling histograms; add wrapper.
666 (clear_storage_via_libcall): Export.
667 (emit_block_move_hints): Break out from ...; add histograms.
668 (emit_block_move): ... this one.
669 (clear_storage_hints): Break out from ...; add histograms.
670 (clear_storage): ... this one.
671 (set_storage_via_memset): Handle histogram.
672 * expr.h (emit_block_move_via_libcall, emit_block_move_hints): Declare.
673 (clear_storage_hints, clear_storage_via_libcall): Declare.
674 (set_storage_via_setmem): Update prototype.
675 * doc/md.texi (movmem, setmem): Document new arguments.
677 * value-prof.c (dump_histogram_value, tree_find_values_to_profile): Add
679 (stringop_block_profile): New global function.
680 (tree_stringops_values_to_profile): Profile block size and alignment.
681 * value-prof.h (enum hist_type): add HIST_TYPE_AVERAGE and
683 (struct profile_hooks): Add gen_average_profiler and gen_ior_profiler.
684 (stringop_block_profile): Declare.
685 * builtins.c: Include value-prof.h.
686 (expand_builtin_memcpy, expand_builtin_memset): Pass block profile.
687 * gcov-ui.h (GCOV_COUNTER_NAMES): Add new counter.
688 (GCOV_COUNTER_AVERAGE, GCOV_COUNTER_IOR): New constants.
689 (GCOV_COUNTERS, GCOV_LAST_VALUE_COUNTER): Update.
690 * profile.c (instrument_values): Add new counters.
691 * cfgexpand.c (expand_gimple_basic_block): Propagate histograms to
693 * tree-profile.c (tree_average_profiler_fn, tree_ior_profiler_fn): New.
694 (tree_init_edge_profiler): Build new profilers.
695 (tree_gen_average_profiler, tree_gen_ior_profiler): New.
696 (pass_tree_profile): Add dump.
697 (tree_profile_hooks): Update.
698 * Makefile.in (LIBGCOV): Add new constants.
699 * libgcov.c (__gcov_merge_ior, __gcov_average_profiler,
700 __gcov_ior_profiler): New.
701 * i386.md (movmem/setmem expanders): Add new optional arguments.
703 2007-01-28 David Edelsohn <edelsohn@gnu.org>
705 * doc/md.texi (Standard Pattern Names): Document blockage pattern.
707 2007-01-28 Zdenek Dvorak <dvorakz@suse.cz>
709 * tree-ssa-loop-unswitch.c: Include tree-inline.h.
710 (tree_unswitch_single_loop): Pass eni_size_weights to
712 * tree-ssa-loop-manip.c: Include tree-inline.h.
713 (can_unroll_loop_p): Pass eni_size_weights to
715 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p):
716 Pass eni_size_weights to estimate_num_insns.
717 * tree.h (init_inline_once): Export.
718 * toplev.c (backend_init): Call init_inline_once.
719 * cgraphunit.c (cgraph_process_new_functions,
720 cgraph_analyze_function): Pass eni_inlining_weights to
722 * ipa-inline.c (compute_inline_parameters): Ditto.
723 * tree-ssa-loop-ivcanon.c (tree_num_loop_insns): Pass weights
724 to estimate_num_insns.
725 (try_unroll_loop_completely): Pass eni_size_weights to
727 * tree-eh.c (decide_copy_try_finally): Pass eni_size_weights
728 ot estimate_num_insns.
729 * tree-ssa-loop-prefetch.c: Include tree-inline.h.
730 (loop_prefetch_arrays): Pass eni_time_weights to tree_num_loop_insns.
731 * tree-inline.c (eni_inlining_weights, eni_size_weights,
732 eni_time_weights): New variables.
733 (init_inline_once): Initialize them.
734 (struct eni_data): Mew.
735 (estimate_num_insns_1, estimate_num_insns): Use weights.
736 * tree-inline.h (struct eni_weights_d): New.
737 (eni_inlining_weights, eni_size_weights, eni_time_weights): Declare.
738 (estimate_num_insns): Declaration changed.
739 * cfgloop.h (tree_num_loop_insns): Declaration changed.
740 * Makefile.in (tree-ssa-loop-unswitch.o, tree-ssa-loop-prefetch.o,
741 tree-ssa-loop-manip.o): Add TREE_INLINE_H dependency.
743 2007-01-28 Zdenek Dvorak <dvorakz@suse.cz>
745 * tree-data-ref.c (conflict_fn): Assert that the number of affine
746 relations in the conflict function is valid.
748 2007-01-27 Ian Lance Taylor <iant@google.com>
750 * common.opt: Add fstrict-overflow.
751 * opts.c (decode_options): Set flag_strict_overflow if -O2.
752 * flags.h (TYPE_OVERFLOW_WRAPS): Define.
753 (TYPE_OVERFLOW_UNDEFINED): Define.
754 (TYPE_OVERFLOW_TRAPS): Define. This replaces TYPE_TRAP_SIGNED.
756 * tree.h (TYPE_TRAP_SIGNED): Don't define.
757 * fold-const.c (negate_expr_p): Use TYPE_OVERFLOW_UNDEFINED.
758 (fold_negate_expr): Likewise.
759 (make_range): Likewise.
760 (extract_muldiv_1): Likewise.
761 (maybe_canonicalize_comparison): Likewise.
762 (fold_comparison): Likewise.
763 (fold_binary): Likewise.
764 (tree_expr_nonnegative_p): Likewise.
765 (tree_expr_nonzero_p): Likewise.
766 * tree-vrp.c (compare_values): Likewise.
767 (extract_range_from_binary_expr): Likewise.
768 (extract_range_from_unary_expr): Likewise.
769 * tree-ssa-loop-niter.c (infer_loop_bounds_from_signedness):
771 (nowrap_type_p): Likewise.
772 * tree-scalar-evolution.c (simple_iv): Likewise.
773 * fold-const.c (negate_expr_p): Use TYPE_OVERFLOW_WRAPS.
774 (build_range_check): Likewise.
775 (extract_muldiv_1): Likewise.
776 (fold_comparison): Likewise.
777 * tree-vrp.c (vrp_int_const_binop): Likewise.
778 (extract_range_from_unary_expr): Likewise.
779 * convert.c (convert_to_integer): Likewise.
780 * fold-const.c (fold_negate_expr): Use TYPE_OVERFLOW_TRAPS.
781 (fold_comparison): Likewise.
782 (fold_binary): Likewise.
783 * optabs.c (optab_for_tree_code): Likewise.
784 * tree-vectorizer.c (vect_is_simple_reduction): Likewise.
785 * simplify-rtx.c (simplify_const_relational_operation): Check
786 flag_strict_overflow and flag_trapv.
787 (simplify_const_relational_operation): Likewise.
788 * doc/invoke.texi (Option Summary): Mention -fstrict-overflow.
789 (Optimize Options): Add -fstrict-overflow to -O2 list. Document
792 2007-01-27 Roger Sayle <roger@eyesopen.com>
794 * tree.c (tree_fold_gcd): Delete.
795 * tree.h (tree_fold_gcd): Remove prototype.
796 * tree-data-ref.c (tree_fold_divides_p): Don't use tree_fold_gcd to
797 test whether one constant integer is a multiple of another. Instead
798 call int_const_binop with TRUNC_MOD_EXPR and test for a zero result.
799 * fold-const.c (multiple_of_p): We've determined both TOP and
800 BOTTOM are integer constants so we can call int_const_binop directly
801 instead of the more generic const_binop.
803 2007-01-27 Roger Sayle <roger@eyesopen.com>
805 * fold-const.c (size_binop): In the fast-paths for X+0, 0+X, X-0 and
806 1*X check that the constant hasn't overflowed, to preserve the
808 (round_up): Provide an efficient implementation when rouding-up an
809 INTEGER_CST to a power-of-two.
811 2007-01-28 Ralf Wildenhues <Ralf.Wildenhues@gmx.de>
813 * doc/sourcebuild.texi: Add comma for clarity.
814 * doc/extend.texi: Fix some typos.
815 * doc/passes.texi: Likewise.
816 * doc/cppinternals.texi: Likewise.
817 * doc/c-tree.texi: Likewise.
818 * doc/tree-ssa.texi: Likewise.
819 * doc/install.texi: Likewise.
821 2007-01-27 Jan Hubicka <jh@suse.cz>
823 * tree-sra.c (sra_walk_function): Don't rely on aliases being build.
824 (pass_sra): Do not require alias information.
825 * passes.c (init_optimization_passes): Add SRA
827 2007-01-27 Steven Bosscher <steven@gcc.gnu.org>
829 * tracer.c (rest_of_handle_tracer): We already cleaned
830 up the CFG in tracer() so don't do it here again.
831 * cfgcleanup.c (rest_of_handle_jump2): Don't repeat
832 cleanup_cfg here, either. And don't call renumber_insns.
834 * cfgrtl.c (rtl_verify_flow_info_1): Don't verify that BB_END
835 and BB_HEAD are in the insn stream here. Instead make sure
836 that BB_INSN is valid on all insns. Also, do check here that
837 there are no pending branch predictions...
838 (rtl_verify_flow_info): ...instead of doing it here. Checks
839 for BB_END and BB_HEAD moved from rtl_verify_flow_info_1 to
842 2007-01-26 Roger Sayle <roger@eyesopen.com>
844 * config/i386/i386.c (ix86_swap_binary_operands_p): New helper
845 function to simplify/factorize operand order canonicalization.
846 (ix86_fixup_binary_operands): Reorganize using the above function.
847 (ix86_binary_operator_ok): Likewise.
849 2007-01-27 Jakub Jelinek <jakub@redhat.com>
851 * genattrtab.c (struct attr_value_list, insn_code_values): Move to
852 file scope from optimize_attrs.
853 (simplify_test_exp): If insn_code_values is not NULL, use it to speed
855 (optimize_attrs): Clear insn_code_values after freeing it.
857 2007-01-26 Zdenek Dvorak <dvorakz@suse.cz>
859 * tree-ssa-address.c (create_mem_ref): Remove ", bsi" from
860 a parts.base assignment.
862 2007-01-26 Zdenek Dvorak <dvorakz@suse.cz>
864 * tree-data-ref.c (dump_subscript): Use dump_conflict_function.
865 (compute_subscript_distance, initialize_data_dependence_relation,
866 finalize_ddr_dependent, analyze_ziv_subscript,
867 analyze_siv_subscript_cst_affine,
868 compute_overlap_steps_for_affine_univar,
869 compute_overlap_steps_for_affine_1_2, analyze_subscript_affine_affine,
870 analyze_siv_subscript, analyze_miv_subscript,
871 analyze_overlapping_iterations, subscript_dependence_tester_1,
872 compute_self_dependence, free_dependence_relation): Work
873 with affine_fn instead of chrecs.
874 (dump_affine_function, dump_conflict_function, affine_function_equal_p,
875 common_affine_function, affine_function_base,
876 affine_function_constant_p, affine_fn_op, affine_fn_plus,
877 affine_fn_minus, affine_fn_free, conflict_fn_not_known,
878 conflict_fn_no_dependence, free_conflict_function, free_subscripts,
879 conflict_fn, affine_fn_cst, affine_fn_univar): New functions.
880 (all_chrecs_equal_p): Removed.
881 * tree-data-ref.h (affine_fn, conflict_function): New types.
882 (struct subscript): Change type of conflicting_iterations_in_a
883 and conflicting_iterations_in_b.
885 2007-01-26 Steve Ellcey <sje@cup.hp.com>
888 * config/pa/pa.h (TARGET_HPUX_11): New.
889 * config/pa/pa-hpux11.h (TARGET_HPUX_11): New.
890 * config/pa/pa.c (pa_init_builtins): Use TARGET_HPUX_11.
892 2007-01-26 Daniel Berlin <dberlin@dberlin.org>
893 Richard Guenther <rguenther@suse.de>
895 * tree-ssa-structalias.c (solve_graph): Handle case
896 we merged the variable to another.
898 2007-01-25 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
900 * builtins.c (fold_builtin_1): Treat ccos and ccosh as 'even'
903 * fold-const.c (negate_mathfn_p): Treat casin, casinh, catan,
904 catanh, cproj, csin, csinh, ctan and ctanh as 'odd' functions.
906 2007-01-25 DJ Delorie <dj@redhat.com>
908 * config/m32c/m32c.c (m32c_cannot_change_mode_class): We don't
909 allow changes to modes which don't fit in those registers.
911 * reload1.c (choose_reload_regs): Check for invalid subregs before
912 computing their locations, not after.
914 2007-01-25 Geoffrey Keating <geoffk@apple.com>
917 * config/rs6000/rs6000.c (first_altivec_reg_to_save): On Darwin,
918 save Altivec registers in an eh_return function.
919 (compute_vrsave_mask): Likewise.
920 (rs6000_stack_info): Correct AIX/Darwin stack alignment computation
921 for saving Altivec registers.
922 (rs6000_emit_prologue): Don't allocate stack twice in
923 eh_return function. Correct expected value of altivec_save_offset
924 when using save_world. Describe save of R0 to stack when using
925 save_world. Describe stack pointer adjustment when using
926 save_world. Remove duplicated eh_return parameter register saving.
927 Update sp_offset variable after save_world.
928 * config/rs6000/t-darwin (LIB2FUNCS_STATIC_EXTRA): Remove
930 (LIB2FUNCS_EXTRA): Add darwin-world.asm.
931 * config/rs6000/darwin.h (SUBTARGET_OVERRIDE_OPTIONS): -m64
934 2007-01-25 Steve Ellcey <sje@cup.hp.com>
936 * config.gcc (ia64*-*-hpux*): Make posix threads the default.
938 2007-01-25 Steve Ellcey <sje@cup.hp.com>
941 * config/pa/pa.c (pa_init_builtins): Set asm names for finite routines.
942 * config/ia64/ia64.c (ia64_init_builtins): Ditto.
944 2007-01-25 Richard Guenther <rguenther@suse.de>
946 * doc/invoke.texi (-Wcoverage-mismatch): Document.
947 * common.opt (-Wcoverage-mismatch): New warning option.
948 * coverage.c (get_coverage_counts): Ignore coverage mismatch
949 if -Wcoverage-mismatch is given.
951 2007-01-25 Razya Ladelsky <razya@il.ibm.com>
953 * ipa-cp.c (ipcp_insert_stage, ipcp_driver): Support for SSA.
954 (ipcp_driver): Change to static definition.
955 Add dumping of the ifunctions.
956 (constant_val_insert): Remove unused parameter. Support for SSA.
957 (ipcp_propagate_const): Support for SSA.
958 (ipcp_profile_bb_print): Print only analyzed nodes.
959 (ipcp_replace_map_create): Remove support for Fortran constant
961 * ipa-prop.c (ipa_method_modify_stmt,
962 ipa_callsite_compute_param): Support for SSA.
963 * ipa-prop.h (ipcp_driver): Remove declaration.
964 (IS_VALID_TREE_MAP_INDEX): Add define.
966 2007-01-24 Geoffrey Keating <geoffk@apple.com>
968 * unwind-dw2.c (execute_stack_op): Handle DW_OP_swap.
970 2007-01-24 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
973 * builtins.c (fold_builtin_cabs): Use MPFR to evaluate a
974 constant argument to cabs and do it without checking for
975 -funsafe-math-optimizations.
977 2007-01-24 Douglas Gregor <dgregor@osl.iu.edu>
979 * c-common.h (RID_FIRST_CXX0X): New.
980 (RID_LAST_CXX0X): New.
981 * c-opts.c (c_common_handle_option): -Wc++0x-compat is triggered
983 * c.opt (Wc++0x-compat): New.
984 * doc/invoke.texi (-Wc++0x-compat): Document.
986 2007-01-24 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
988 * builtins.c (fold_builtin_carg): New.
989 (fold_builtin_1): Use it.
991 2007-01-24 Jan Hubicka <jh@suse.cz>
993 * ipa-inline.c (cgraph_decide_inlining): Initialize initial_insns.
995 * ipa-inline.c (initial_insns, max_insns): Delete.
996 (compute_max_insns): New function.
997 (cgraph_decide_inlining_of_small_function): Use it; take minimal amount
998 of insns as base for code growth.
999 (cgraph_decide_inlining): Make initial_insns local; do not compute
1001 * params.def (PARAM_INLINE_UNIT_GROWTH): Set to 60.
1002 * doc/invoke.texi (inline-unit-growth): Update docs.
1004 2007-01-24 Jakub Jelinek <jakub@redhat.com>
1006 * config/i386/i386.h (x86_cmpxchg16b): Remove const.
1007 (TARGET_CMPXCHG16B): Define to x86_cmpxchg16b.
1008 * config/i386/i386.c (x86_cmpxchg16b): Remove const.
1009 (override_options): Add PTA_CX16 flag. Set x86_cmpxchg16b
1010 for CPUs that have PTA_CX16 set.
1013 * gimplify.c (omp_check_private): New function.
1014 (gimplify_scan_omp_clauses): Use it for
1015 firstprivate/lastprivate/reduction.
1018 * gimplify.c (omp_add_variable): Don't call omp_notice_variable
1019 on TYPE_SIZE_UNIT for GOVD_LOCAL VLAs.
1022 * omp-low.c (lower_omp_for_lastprivate): Add dlist argument.
1023 If lower_lastprivate_clauses emits some statements, append them
1024 to dlist rather than body_p and to body_p append an initializer.
1025 (lower_omp_for): Adjust caller.
1027 2007-01-24 Steve Ellcey <sje@cup.hp.com>
1029 * target.h (globalize_decl_name): New.
1030 * target-def.h (TARGET_ASM_GLOBALIZE_DECL_NAME): New.
1031 * output.h (default_globalize_decl_name): New.
1032 * varasm.c (asm_output_bss): Use globalize_decl_name instead of
1034 (globalize_decl): Ditto.
1035 (default_globalize_decl_name): New.
1036 * config/ia64/ia64.c (ia64_globalize_decl_name): New.
1037 (ia64_handle_version_id_attribute): New.
1038 (TARGET_ASM_GLOBALIZE_DECL_NAME): New.
1039 (ia64_asm_output_external): Use globalize_decl_name instead
1041 * doc/extend.texi (version_id): New pragma.
1042 * doc/tm.texi (ARGET_ASM_GLOBALIZE_DECL_NAME): New target hook.
1044 2007-01-24 Andreas Krebbel <krebbel1@de.ibm.com>
1046 * unwind-dw2-fde.c (get_cie_encoding): Replaced _Unwind_Word with
1047 _uleb128_t and _Unwind_SWord with _sleb128_t.
1048 * unwind-dw2.c (extract_cie_info, execute_stack_op, execute_cfa_program,
1049 uw_frame_state_for, uw_update_context_1): Likewise.
1050 * unwind-c.c (parse_lsda_header, PERSONALITY_FUNCTION): Likewise.
1051 * unwind-pe.h (read_uleb128, read_sleb128,
1052 read_encoded_value_with_base): Likewise.
1053 * unwind-generic.h: Define _sleb128_t and _uleb128_t types.
1055 2007-01-24 Richard Guenther <rguenther@suse.de>
1057 * builtins.c (expand_builtin_cexpi): Get the fndecl
1058 for cexp in the correct way.
1060 2007-01-24 Jan Hubicka <jh@suse.cz>
1062 * tree-ssa-dce.c (eliminate_unnecesary_stmts): Remove dead LHS of calls.
1064 2007-01-24 Andreas Krebbel <krebbel1@de.ibm.com>
1066 * c-cppbuiltin.c (builtin_define_type_sizeof): New function.
1067 (c_cpp_builtins): New builtin macros: __SIZEOF_INT__, __SIZEOF_LONG__,
1068 __SIZEOF_LONG_LONG__, __SIZEOF_SHORT__, __SIZEOF_POINTER__,
1069 __SIZEOF_FLOAT__, __SIZEOF_DOUBLE__, __SIZEOF_LONG_DOUBLE__,
1070 __SIZEOF_SIZE_T__, __SIZEOF_WCHAR_T__, __SIZEOF_WINT_T__ and
1071 __SIZEOF_PTRDIFF_T__.
1072 * doc/cpp.texi: Documentation for the new builtin macros added.
1074 2007-01-24 Uros Bizjak <ubizjak@gmail.com>
1076 * config/i386/i386.md (tanxf2, tan<mode>2, atan<mode>2, log<mode>2,
1077 log10<mode>2, log2<mode>2, expxf2, exp10xf2, exp2xf2): Use op2
1078 instead of operands[2] to avoid access past the end of array.
1080 2007-01-24 Richard Sandiford <richard@codesourcery.com>
1082 * reload1.c (emit_reload_insns): Pass the reload register
1083 for a non-spill output reload through forget_old_reloads_1.
1085 2007-01-23 Joseph Myers <joseph@codesourcery.com>
1087 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Define
1088 _SOFT_DOUBLE if doubles use software floating-point.
1089 * config/rs6000/libgcc-ppc-glibc.ver: Export additional long
1090 double functions if _SOFT_DOUBLE, not _SOFT_FLOAT.
1091 * config/rs6000/darwin-ldouble.c: Also compile functions for
1092 hard-float without FPRs. Use fmsub function for all __NO_FPRS__
1093 cases. Compile extra functions if _SOFT_DOUBLE, not _SOFT_FLOAT.
1094 * config/rs6000/linuxspe.h (SUBSUBTARGET_OVERRIDE_OPTIONS): Remove
1095 commented-out long double override.
1096 (CPP_LONGDOUBLE_DEFAULT_SPEC): Likewise.
1097 * config/rs6000/eabispe.h: Likewise.
1098 * config/rs6000/rs6000.c (rs6000_override_options): Don't override
1099 long double for non-SPE.
1100 (rs6000_handle_option): Likewise.
1101 (invalid_e500_subreg): Disallow more subregs involding DImode,
1102 DFmode, TImode or TFmode.
1103 (rs6000_legitimate_offset_address_p): Check TFmode offsets for
1105 (legitimate_lo_sum_address_p): Also check for TFmode for E500
1107 (rs6000_legitimize_address): Also handle TFmode for E500 double.
1108 (rs6000_legitimize_reload_address): Also handle TFmode for E500
1110 (rs6000_legitimate_address): Also check for TFmode for E500
1112 (rs6000_emit_move): Use DFmode subregs of TFmode for E500 double.
1113 (spe_build_register_parallel): Handle TFmode and TCmode.
1114 (rs6000_spe_function_arg): Handle TFmode and TCmode for E500
1116 (function_arg): Handle TFmode and TCmode for E500 double.
1117 (rs6000_init_libfuncs): Initialize extra libfuncs for soft double
1119 (print_operand): Handle TFmode and TImode for %y.
1120 (rs6000_generate_compare): Handle TFmode comparisons for E500
1122 (spe_func_has_64bit_regs_p): Check for TFmode for E500 double.
1123 (rs6000_function_value): Handle TFmode and TCmode for E500 double.
1124 (rs6000_libcall_value): Handle TFmode and TCmode for E500 double.
1125 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Check for
1126 TFmode for E500 double.
1127 * config/rs6000/rs6000.md (FP): Allow TF for E500 double.
1128 (floatsidf2): Enable for E500 double.
1129 (movtf_softfloat): Use rs6000_nonimmediate_operand.
1130 (extenddftf2): Change to extenddftf2_fprs.
1131 (extenddftf2): Call gen_spe_extenddftf2 or gen_extenddftf2_fprs
1132 depending on TARGET_E500_DOUBLE.
1133 (extendsftf2): Enable for E500 double.
1134 (trunctfdf2): Enable for E500 double.
1135 (trunctfsf2): Change to trunctfsf2_fprs.
1136 (trunctfsf2): Call gen_spe_trunctfsf2 or gen_trunctfsf2_fprs
1137 depending on TARGET_E500_DOUBLE.
1138 (floatsitf2): Enable for E500 double.
1139 (fix_trunctfsi2): Change to fix_trunctfsi2_fprs.
1140 (fix_trunctfsi2): Call gen_spe_fix_trunctfsi2 or
1141 gen_fix_trunctfsi2_fprs depending on TARGET_E500_DOUBLE.
1142 (negtf2): Change to negtf2_internal.
1143 (negtf2): New expander.
1144 (abstf2): Enable for E500 double. Call gen_spe_abstf2_tst,
1145 gen_spe_abstf2_cmp or gen_abstf2_internal depending on
1146 TARGET_E500_DOUBLE and flag_unsafe_math_optimizations.
1147 (movdi_internal32): Use rs6000_nonimmediate_operand.
1148 (unnamed splitter): Likewise.
1149 * config/rs6000/spe.md (CMPTFEQ_GPR, TSTTFEQ_GPR, CMPTFGT_GPR,
1150 TSTTFGT_GPR, CMPTFLT_GPR, TSTTFLT_GPR): New unspecs.
1151 (SPE64TF, DITI): New mode macros.
1152 (frob_df_di): Change to frob_<SPE64:mode>_<DITI:mode>; allow more
1155 (frob_<mode>_di_2): New.
1156 (frob_tf_di_8_2): New.
1157 (frob_di_df): Change to frob_di_<mode>; allow more modes.
1159 (frob_di_df_2): Change to frob_<DITI:mode>_<SPE64:mode>_2; allow
1161 (frob_ti_<mode>_8_2): New.
1162 (frob_ti_tf_2): New.
1163 (mov_si<mode>_e500_subreg0, mov_si<mode>_e500_subreg0_2,
1164 mov_si<mode>_e500_subreg4, mov_si<mode>_e500_subreg4_2): Allow
1166 (mov_sitf_e500_subreg8, mov_sitf_e500_subreg8_2,
1167 mov_sitf_e500_subreg12, mov_sitf_e500_subreg12_2): New.
1168 (spe_trunctfdf2_internal1, spe_trunctfsf2, spe_extenddftf2,
1169 spe_fix_trunctfsi2, spe_fix_trunctfsi2_internal,
1170 spe_negtf2_internal, spe_abstf2_cmp, spe_abstf2_tst): New.
1171 (cmptfeq_gpr, tsttfeq_gpr, cmptfgt_gpr, tsttfgt_gpr, cmptflt_gpr,
1174 2007-01-23 Ian Lance Taylor <iant@google.com>
1176 * Makefile.in (OBJS-common): Reformat, alphabetize, but put
1178 (OBJS-archive): Reformat, alphabetize.
1179 (OBJS): Change out_object_file to OBJS-md.
1181 2007-01-23 Uros Bizjak <ubizjak@gmail.com>
1183 * config/i386/i386.md: Use REG_P, MEM_P, CONST_INT_P, LABEL_P,
1184 JUMP_P and CALL_P predicates where applicable.
1185 * config/i386/i386.c: Ditto.
1186 * config/i386/i386.md: Ditto.
1187 * config/i386/mmx.md: Ditto.
1188 * config/i386/predicates.md: Ditto.
1190 2007-01-22 Andreas Schwab <schwab@suse.de>
1192 * config/m68k/m68k.h: Fix comment.
1194 2007-01-22 Jan Hubicka <jh@suse.cz>
1196 * passes.c (init_optimization_passes): Do not rerun
1197 pass_early_warn_uninitialized.
1199 2007-01-22 Richard Guenther <rguenther@suse.de>
1201 PR tree-optimization/30038
1202 * tree-ssa-math-opts.c (maybe_record_sincos): New static helper
1204 (execute_cse_sincos_1): Likewise.
1205 (execute_cse_sincos): Likewise.
1206 (gate_cse_sincos): Likewise.
1207 (pass_cse_sincos): New pass CSEing sin() and cos() calls using
1208 the cexpi() canonicalization of sincos().
1209 * tree-pass.h (pass_cse_sincos): Declare.
1210 * passes.c (init_optimization_passes): New pass pas_cse_sincos.
1212 2007-01-21 Eric Botcazou <ebotcazou@libertysurf.fr>
1214 PR rtl-optimization/29329
1215 * combine.c (replaced_rhs_insn): Rename to i2mod.
1216 (replaced_rhs_value): Rename to i2mod_new_rhs.
1217 (i2mod_old_rhs): New global variable.
1218 (combine_instructions): Adjust for above change. Save a copy of
1219 the old RHS into i2mod_old_rhs when the contents of a REG_EQUAL
1220 note are substituted in the second instruction.
1221 (distribute_notes) <REG_DEAD>: Adjust for above change. Do not
1222 ditch the note if it pertains to the second eliminated register
1223 and this register is mentioned in i2mod_old_rhs.
1226 2006-09-12 Eric Botcazou <ebotcazou@libertysurf.fr>
1228 * combine.c (distribute_notes) <REG_DEAD>: Do not consider SETs past
1229 the insn to which the note was originally attached.
1231 2007-01-21 Jan Hubicka <jh@suse.cz>
1233 * ipa-inline.c (inlining_mode): Comment, move up.
1234 (cgraph_decide_inlining_incrementally): Do not perform inlining
1235 itself; fix handling of flattening of self recursive functions.
1236 (cgraph_find_cycles): Remove.
1237 (cgraph_flatten_node): Remove.
1238 (cgraph_decide_inlining): Use incremental inliner to handle flattening.
1239 (try_inline): New function.
1240 (cgraph_early_inlining): Update call of
1241 cgraph_decide_inlining_incrementally. Apply inlining here.
1242 (apply_inline): Update call of cgraph_decide_inlining_incrementally.
1244 2007-01-21 Dirk Mueller <dmueller@suse.de>
1247 * tree-vrp.c (check_array_bounds): do not warn
1248 about ADDR_EXPR's of ARRAY_REF's which are immediately
1249 used in binary expressions.
1251 2007-01-21 Ira Rosen <irar@il.ibm.com>
1253 * tree-vectorizer.h (struct _stmt_vec_info): Add new field
1254 read_write_dep and macros for its access.
1255 * tree-vectorizer.c (new_stmt_vec_info): Initialize the new field.
1256 * tree-vect-analyze.c (vect_analyze_data_ref_dependence): Remove
1257 argument, call vect_check_interleaving for every independent pair of
1258 data-refs. Mark loads that access the same memory location as a store
1260 (vect_check_dependences): Remove.
1261 (vect_analyze_data_ref_dependences): Remove vect_check_dependences
1262 call, fix the call to vect_analyze_data_ref_dependence.
1263 (vect_analyze_data_ref_access): For statements that access the same
1264 data-ref, check that they are not stores; for loads, check that there
1265 is no store that access the same location.
1267 2007-01-20 Roger Sayle <roger@eyesopen.com>
1268 Joseph Myers <joseph@codesourcery.com>
1270 * doc/invoke.texi (-fdump-rtl-combine): Fix under/overfull hbox.
1271 (-fdump-rtl-stack): Likewise.
1272 (-fno-signed-zeros): Use @minus{} for a minus sign. Correct typo.
1273 (-mcheck-zero-division, -mcpu): Fix under/overfull hbox.
1274 (-mpt-fixed): Use @minus{} for minus sign.
1275 (Using Precompiled Headers): Fix under/overfull hbox.
1277 2007-01-20 Jan Hubicka <jh@suse.cz>
1279 * tree-flow.h (struct stmt_ann_d): Move references_memory to proper
1280 place within annotation.
1282 2007-01-20 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
1284 * pa.c (output_move_double): Change array size of xoperands to 4.
1286 2007-01-20 Jan Hubicka <jh@suse.cz>
1288 * tree-tailcall.c (adjust_return_value): Do not use RESULT_DECL
1291 2007-01-19 Ian Lance Taylor <iant@google.com>
1293 * expmed.c (expand_divmod) [TRUNC_MOD_EXPR, TRUNC_DIV_EXPR]: Cast
1294 constant to unsigned HOST_WIDE_INT before negating.
1296 2007-01-19 Ian Lance Taylor <iant@google.com>
1298 * tree-ssa-operands.h (struct vuse_vec_d): Change num_vuse field
1300 (VUSE_VECT_ELEMENT) [ENABLE_CHECKING]: Use unsigned comparison.
1301 (VUSE_ELEMENT_PTR) [ENABLE_CHECKING]: Likewise.
1302 (SET_VUSE_VECT_ELEMENT) [ENABLE_CHECKING]: Likewise.
1303 (SET_VUSE_ELEMENT_VAR) [ENABLE_CHECKING]: Likewise.
1304 (SET_VUSE_ELEMENT_PTR) [ENABLE_CHECKING]: Likewise.
1305 (realloc_vdef, realloc_vuse): Change second parameter to
1307 (ssa_operand_iterator_d): Change vuse_index and mayuse_index
1309 * tree-ssa-operands.c (realloc_vop): Change num_elem parameter to
1310 unsigned. Change x and lim locals to unsigned.
1311 (realloc_vdef, realloc_vuse): Change num_elem parameter to
1313 (finalize_ssa_vuse_ops): Change old_i local to unsigned.
1314 (copy_virtual_operands): Change i and n locals to unsigned.
1316 2007-01-19 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
1319 * builtins.c (fold_builtin_1): Handle builtin fdim.
1321 2007-01-20 Jan Hubicka <jh@suse.cz>
1323 * tree-ssa.c (init_tree_ssa): Do not call init_alias_heapvars.
1324 * tree-ssa-structalias.c (compute_points_to_sets): Do call
1325 init_alias_heapvars.
1326 (init_alias_heapvars): Initialize only when not already initialized.
1327 (delete_alias_heapvars): Set heapvar_for_stmt to NULL.
1329 2007-01-19 Roger Sayle <roger@eyesopen.com>
1331 * common.opt (fsigned-zeros): New command line option.
1332 * flags.h (HONOR_SIGNED_ZEROS): Control via flag_signed_zeros instead
1333 of flag_unsafe_math_optimizations.
1334 * opts.c (set_fast_math_flags): The -ffast-math command line option
1335 implies -fno-signed-zeros.
1336 (fast_math_flags_set_p): Likewise.
1338 * doc/invoke.texi: Document new -fno-signed-zeros option, and update
1339 the documentation of -ffast-math appropriately. Wrap long lines.
1341 2007-01-19 Steve Ellcey <sje@cup.hp.com>
1343 * system.h (ASM_MAKE_LABEL_LINKONCE): Poison.
1344 * varasm.c (globalize_decl): Remove ASM_MAKE_LABEL_LINKONCE ifdef.
1346 2007-01-19 Tomas Bily <tbily@suse.cz>
1348 * cgraphunit.c (cgraph_finalize_function): Updating of pid
1350 (tree_init_ic_make_global_vars): New function
1351 (tree_init_edge_profiler): call of tree_init_ic_make_global_vars
1352 (tree_gen_ic_profiler): New function
1353 (tree_gen_ic_func_profiler): New function
1354 (tree_profiling): Added calling of tree_gen_ic_func_profiler
1355 (tree_profile_hooks): Added hook for indirec/virtual calls
1356 * value-prof.c (tree_find_values_to_profile): New case for
1358 (tree_values_to_profile): Call for determining indirect/virtual
1360 (tree_indirect_call_to_profile): New function
1361 (tree_ic_transform): New function
1362 (tree_ic): New function
1363 (find_func_by_pid): New function
1364 (init_pid_map): New function
1365 (tree_value_profile_transformations): Added check for
1366 indirect/virtual call transformation
1367 * value-prof.h (enum hist_type): New counter type for
1368 indirect/virtual calls
1369 (profile_hooks): Added new hook for profiling indirect/virtual
1371 * profile.c (instrument_values): New case for indirect/virtual
1373 * gcov-io.h (GCOV_LAST_VALUE_COUNTER): Changed to 6
1374 (GCOV_COUNTER_V_INDIR): New counter type
1375 (GCOV_COUNTER_NAMES): New name of counter "indirect" added
1376 (GCOV_MERGE_FUNCTIONS): New merge function for indirect/virtual
1378 * cgraph.c: Definition of cgraph_max_pid
1379 (cgraph_create_node): Default init of pid attribute
1380 * cgraph.h: Declaration of cgraph_max_pid
1381 (struct cgraph_node): Added pid attribute
1382 * libgcov.c (__gcov_indirect_call_profiler): New function
1383 (__gcov_one_value_profiler_body): New function
1384 (__gcov_one_value_profiler): Body was moved to
1385 __gcov_one_value_profiler_body and calls it
1387 2007-01-19 Basile Starynkevitch <basile@starynkevitch.net>
1389 * doc/gty.texi (Options): Document the mark_hook option to GTY.
1390 * gengtype.c (write_types_data, write_func_for_structure,
1391 write_types, ggc_wtd, pch_wtd): Add skip_hooks to
1392 write_types_data, ggc_wtd, pch_wtd for processing mark_hook.
1393 (walk_type, write_func_for_structure): Generate the mark_hook if
1396 2007-01-19 Jan Hubicka <jh@suse.cz>
1398 * ipa-inline.c (cgraph_decide_inlining_incrementally): Instead of
1399 'early' argument take inlining mode argument specifying whether to
1400 inline for size/speeed or all functions; add support for flattening;
1402 (cgraph_early_inlining): Update call of decide_inlining_incrementally.
1404 2007-01-19 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
1407 * toplev.c (warn_deprecated_use): Use %qD instead of %qs to print
1408 the name of the declared identifier.
1410 2007-01-19 Dirk Mueller <dmueller@suse.de>
1412 * config/i386.h (CONDITIONAL_REGISTER_USAGE): Store
1413 result of PIC_OFFSET_TABLE_REGNUM in temporary variable to avoid
1414 duplicate evaluation.
1416 2007-01-19 Uros Bizjak <ubizjak@gmail.com>
1418 * config/i386/i386.md (acos<mode>2): Rename from acossf2 and acosdf2.
1419 Macroize expander using X87MODEF12 mode macro. Extend operand 1
1420 to XFMode, use acosxf2 and truncate result to requested mode.
1421 Use SSE_FLOAT_MODE_P to disable patterns for SSE math.
1422 (asin<mode>2): Similarly, with asin expanders.
1423 (*fscalexf4_i387): Rename from *fscalexf4.
1424 (expNcorexf3): New expander.
1425 (expxf2, exp10xf2, exp2xf2): Use expNcorexf3 expander.
1426 (exp<mode>2): Rename from expsf2 and expdf2. Macroize expander using
1427 X87MODEF12 mode macro. Extend operand 1 to XFMode, use expxf2 and
1428 truncate result to requested mode. Use SSE_FLOAT_MODE_P to disable
1429 patterns for SSE math.
1430 (exp10<mode>2): Similarly, with exp10 expanders.
1431 (exp2<mode>2): Similarly, with exp2 expanders.
1432 (expm1<mode>2): Similarly, with expm1 expanders.
1433 (ldexp<mode>3): Similarly, with ldexp expanders.
1434 (log<mode>2, log10<mode>2, log2<mode>2, log1p<mode>2, logb<mode>2):
1435 Use gen_truncxf<mode>2_i387_noop to truncate result.
1437 2007-01-19 Richard Sandiford <richard@codesourcery.com>
1439 * config/m68k/m68k.h (PREFERRED_STACK_BOUNDARY): Define to 32
1440 for ColdFire targets.
1442 2007-01-19 Nathan Sidwell <nathan@codesourcery.com>
1443 Richard Sandiford <richard@codesourcery.com>
1445 * config/m68k/m68k.h (M68K_STATIC_CHAIN_REG_NAME): New macro.
1446 (INITIALIZE_TRAMPOLINE): Use STATIC_CHAIN_REGNUM.
1447 (__transfer_from_trampoline): Use M68K_STATIC_CHAIN_REG_NAME.
1448 * config/m68k/m68kelf.h (STATIC_CHAIN_REGNUM): Override.
1449 (M68K_STATIC_CHAIN_REG_NAME): Likewise.
1450 * config/m68k/netbsd-elf.h (M68K_STATIC_CHAIN_REG_NAME): Likewise.
1452 2007-01-19 Richard Sandiford <richard@codesourcery.com>
1454 * config/m68k/m68k.md (adddi_dilshr32): Rename to...
1455 (*adddi_dilshr32): ...this. Fix formatting. Remove commented-out
1456 non-canonical pattern. Restrict to !TARGET_COLDFIRE.
1457 (*adddi_dilshr32_cf): New pattern.
1458 (adddi3, subdi3): Remove first alternatives.
1460 2007-01-19 Richard Sandiford <richard@codesourcery.com>
1462 * config/m68k/m68k.c (notice_update_cc): If an SFmode move is
1463 implemented using move.l, do not use its cc result for floating-point
1466 2007-01-19 Richard Sandiford <richard@codesourcery.com>
1468 * config/m68k/m68k.h (EXTRA_CONSTRAINT): Stop the 'T' constraint
1469 from accepting 's' constraints if flag_pic.
1471 2007-01-19 Richard Sandiford <richard@codesourcery.com>
1473 * config/m68k/m68k.md (bordered, bunordered, buneq, bunge, bungt)
1474 (bunle, bunlt, bltgt, bordered_rev, bunordered_rev, buneq_rev)
1475 (bunge_rev, bungt_rev, bunle_rev, bunlt_rev, bltgt_rev): Change
1476 condition from TARGET_68881 to TARGET_HARD_FLOAT.
1478 2007-01-19 Sandra Loosemore <sandra@codesourcery.com>
1480 * longlong.h (count_leading_zeros, COUNT_LEADING_ZEROS_0): Add
1481 ColdFire alternatives.
1482 * config/m68k/m68k.h (CLZ_DEFINED_VALUE_AT_ZERO): New macro.
1483 * config/m68k/m68k.md (clzsi2): Define for ColdFire
1484 architectures that support ff1 instruction.
1486 2007-01-19 Richard Sandiford <richard@codesourcery.com>
1487 Julian Brown <julian@codesourcery.com>
1489 * config/m68k/m68k.h (CONST_OK_FOR_LETTER_P): Add an 'R' case.
1490 * config/m68k/m68k.md (*movsi_cfv4): Fold into...
1491 (*movsi_cf): ...here. Remove unnecessary 'R' from 'Rg'.
1494 2007-01-19 Richard Sandiford <richard@codesourcery.com>
1496 * config/m68k/m68k-protos.h (valid_mov3q_const): Take a HOST_WIDE_INT
1498 (output_move_const_into_data_reg, output_move_simode_const): Delete.
1499 * config/m68k/m68k.c (const_method, const_int_cost): Take a
1500 HOST_WIDE_INT instead of an rtx.
1501 (m68k_rtx_costs): Update call accordingly.
1502 (output_move_const_into_data_reg): Likewise. Fix formatting.
1503 (valid_mov3q_const): Take a HOST_WIDE_INT instead of an rtx.
1505 (output_move_simode_const): Update calls after above changes.
1506 Rework to use automatic variables and predicates like MEM_P.
1507 * config/m68k/m68k.md (pushexthisi_const): Update call to
1510 2007-01-19 Dirk Mueller <dmueller@suse.de>
1512 * tree-ssa-alias.c (perform_var_substitution): Fix typo
1515 2007-01-19 Richard Guenther <rguenther@suse.de>
1517 * builtins.c (expand_builtin_cexpi): Fall back to expanding
1518 via cexp in case sincos is not available.
1520 2007-01-19 Richard Guenther <rguenther@suse.de>
1522 * doc/tm.texi (TARGET_HAS_SINCOS): Document new target macro.
1523 * defaults.h (TARGET_HAS_SINCOS): Default to off.
1524 * config/linux.h (TARGET_HAS_SINCOS): Set to on if we have glibc.
1525 * config/alpha/linux.h (TARGET_HAS_SINCOS): Likewise.
1526 * config/sparc/linux.h (TARGET_HAS_SINCOS): Likewise.
1527 * config/sparc/linux64.h (TARGET_HAS_SINCOS): Likewise.
1528 * config/rs6000/linux.h (TARGET_HAS_SINCOS): Likewise.
1529 * config/rs6000/linux64.h (TARGET_HAS_SINCOS): Likewise.
1531 2007-01-19 Uros Bizjak <ubizjak@gmail.com>
1533 * config/i386/i386.md (*fpatanxf3_i387, fpatan_extend<mode>xf3_i387):
1535 (atan2sf3_1, atan2df3_1, atan2xf3_1): Remove insn patterns.
1536 (atan2xf3): Directly generate RTL pattern.
1537 (atan2<mode>3): Rename from atan2sf3 and atan2df3 and macroize insn
1538 patterns using X87MODEF12 mode macro. Use fpatan_extend<mode>xf3_i387
1539 and truncate result to requested mode. Use SSE_FLOAT_MODE_P to
1540 disable patterns for SSE math.
1541 (atan<mode>2): Rename from atansf2 and atandf2 and macroize insn
1542 patterns using X87MODEF12 mode macro. Use fpatan_extend<mode>xf3_i387
1543 and truncate result to requested mode. Use SSE_FLOAT_MODE_P to
1544 disable patterns for SSE math.
1546 2007-01-19 Alexandre Oliva <aoliva@redhat.com>
1548 * libgcc-std.ver: Fix typo in %inherit for GCC_4.3.0.
1550 2007-01-18 Roger Sayle <roger@eyesopen.com>
1552 * fold-const.c (fold_unary) <VIEW_CONVERT_EXPR>: Optimize away a
1553 VIEW_CONVERT_EXPR to the same type as it's operand.
1555 2007-01-18 David Edelsohn <edelsohn@gnu.org>
1557 * config/rs6000/darwin-ldouble.c: Only build _SOFT_FLOAT if
1558 configured for long double 128.
1560 2007-01-18 Mike Stump <mrs@apple.com>
1562 * config/rs6000/rs6000.c (rs6000_emit_vector_compare): Fix build
1565 2007-01-18 Michael Meissner <michael.meissner@amd.com>
1567 * i386.c (ix86_compute_frame_layout): Make fprintf's in #if 0 code
1570 2007-01-18 Jan Hubicka <jh@suse.cz>
1572 * tree-ssa-operands.c (vop_free_bucket_size): Never return value
1573 greater than NUM_VOP_FREE_BUCKETS.
1575 2007-01-18 Daniel Berlin <dberlin@dberlin.org>
1577 * tree-ssa-structalias.c: Update comments.
1578 (ptabitmap_obstack): Removed.
1580 (oldpta_obstack): Ditto.
1581 (stats): Add a few members.
1582 (struct variable_info): Remove node, complex, address_taken, and
1583 indirect_target members. Add oldsolution member.
1584 (new_var_info): Do not initialize removed members.
1585 (constraint_expr_type): Remove INCLUDES.
1586 (constraint_graph): Add size, implicit_preds, rep,
1587 indirect_cycles, eq_rep, label, direct_nodes, and complex members.
1588 (FIRST_REF_NODE): New macro.
1589 (LAST_REF_NODE): Ditto.
1590 (FIRST_ADDR_NODE): Ditto.
1591 (find): New function.
1593 (dump_constraint): Do not handle INCLUDES.
1594 (insert_into_complex): Do not insert duplicate constraints.
1595 (condense_varmap_nodes): Renamed and rewritten into ...
1596 (merge_node_constraints): This. Also fix bug in handling of
1597 offseted copy constraints.
1598 (clear_edges_for_node): No longer need to deal with preds at all,
1599 or removing associated preds/succs.
1600 (merge_graph_nodes): Deal with indirect_cycles.
1601 Don't deal with predecessors.
1602 (add_implicit_graph_edge): New function.
1603 (add_pred_graph_edge): Ditto.
1604 (add_graph_edge): Don't deal with predecessors.
1605 (build_constraint_graph): Removed.
1606 (build_pred_graph): New function.
1607 (build_succ_graph): Ditto.
1608 (struct scc_info): Removed in_component. Added roots, dfs, and
1609 node_mapping. Remove visited_index, unification_queue.
1610 (scc_visit): Deal with union-find we do now.
1611 Deal with cycles with REF nodes.
1612 (collapse_nodes): Renamed and rewritten to ...
1613 (unify_nodes): This.
1614 (process_unification_queue): Removed.
1615 (topo_visit): Cleanup
1616 (do_da_constraint): Use find.
1617 (do_sd_constraint): Ditto.
1618 (do_ds_constraint): Ditto.
1619 (do_complex_constraint): Ditto.
1620 (init_scc_info): Update for removed and added members.
1621 (find_and_collapse_graph_cycles): Renamed and rewritten into ...
1622 (find_indirect_cycles): This.
1623 (equivalence_class): New variable.
1624 (label_visit): New function.
1625 (perform_variable_substitution): Rewritten.
1626 (free_var_substitution_info): New function.
1627 (find_equivalent_node): Ditto.
1628 (move_complex_constraints): Ditto.
1629 (eliminate_indirect_cycles): Ditto.
1630 (solve_graph): Only propagate changed bits.
1631 Use indirect cycle elimination.
1633 (tree_id_t): Rename to tree_vi_t, delete id member, add vi member.
1634 (tree_id_eq): Renamed to ...
1635 (tree_vi_eq): This. Update for member change
1636 (insert_id_for_tree): Renamed and rewritten to ...
1637 (insert_vi_for_tree): This.
1638 (lookup_id_for_tree): Renamed and rewritten to ...
1639 (lookup_vi_for_tree): This.
1640 (get_id_for_tree): Renamed and rewritten to ...
1641 (get_vi_for_tree): Ditto.
1642 (get_constraint_exp_from_ssa_var): Update to use get_vi_for_tree.
1643 (process_constraint): Don't handle INCLUDES.
1644 Remove special ADDRESSOF case.
1645 (find_func_aliases): Rewrite to use vi functions instead of id
1647 (create_function_info_for): Ditto.
1648 (create_variable_info_for): Ditto.
1649 (intra_create_variable_infos): Ditto.
1650 (merge_smts_into): Ditto.
1651 (find_what_p_points_to): Ditto.
1652 (init_base_vars): Ditto.
1653 (init_alias_vars): Ditto.
1654 (remove_preds_and_fake_succs): New function.
1655 (dump_sa_points_to_info): Dump new stats.
1656 (dump_solution_for_var): Use find.
1657 (set_used_smts): Fix formatting.
1658 (compute_points_to_sets): Updated for new functions.
1659 (ipa_pta_execute): Ditto.
1661 2007-01-18 Kazu Hirata <kazu@codesourcery.com>
1662 Richard Sandiford <richard@codesourcery.com>
1664 * doc/tm.texi (TARGET_FUNCTION_VALUE): Expand documentation of
1666 * calls.c (expand_call): If the return value is a PARALLEL,
1667 extract its first member.
1668 * config/m68k/linux.h (FUNCTION_EXTRA_EPILOGUE): Remove.
1669 * config/m68k/m68k.c (m68k_output_function_epilogue): Don't
1670 use FUNCTION_EXTRA_EPILOGUE.
1671 (m68k_function_value): Return a PARALLEL if the return value
1672 is of a pointer type.
1673 * config/m68k/netbsd-elf.h (current_function_returns_pointer)
1674 (FUNCTION_EXTRA_EPILOGUE): Remove.
1675 * config/m68k/m68k.md (D0_REG): New constant.
1677 2007-01-18 Kazu Hirata <kazu@codesourcery.com>
1679 * config/m68k/m68k.c (m68k_output_function_epilogue): Don't
1680 output a NOP for empty epilogues.
1682 2007-01-18 Richard Sandiford <richard@codesourcery.com>
1684 * config/m68k/m68k.c (m68k_use_return_insn): Update comments
1685 before function. Extend register save check to include all
1686 registers, not just integer ones.
1688 2007-01-18 Kazu Hirata <kazu@codesourcery.com>
1690 * config/m68k/m68k-protos.h (use_return_insn): Rename to...
1691 (m68k_use_return_insn): ...this.
1692 * config/m68k/m68k.h (USE_RETURN_INSN): Delete.
1693 * config/m68k/m68k.c (use_return_insn): Rename to...
1694 (m68k_use_return_insn): ...this.
1695 * config/m68k/m68k.md (return): Use m68k_use_return_insn instead
1698 2007-01-18 Nathan Sidwell <nathan@codesourcery.com>
1700 * config/m68k/fpgnulib.c (__truncdfsf2): Implement round to
1701 nearest even, fix denormal rounding overflow.
1703 2007-01-18 Richard Sandiford <richard@codesourcery.com>
1705 * config/m68k/m68k.md (movsf_cf_hard): Use fsmove instead of
1706 f%$smove and f%$move.
1707 (movdf_cf_hard): Use fdmove for cases 0 and 3 and fmove for case 1.
1708 (extendsfdf2_cf): Use fdmove instead of f%&move.
1709 (truncdfsf2_cf): Use fsmove instead of f%$smove.
1710 (add<mode>3_cf, sub<mode>3_cf): Use <FP:prec> instead of <FP:round>.
1712 2007-01-18 Richard Sandiford <richard@codesourcery.com>
1714 * config/m68k/m68k.md (movdf_cf_hard): Use output_move_double for
1717 2007-01-18 Richard Sandiford <richard@codesourcery.com>
1719 * real.h (real_format): Add a canonical_nan_lsbs_set field.
1720 (coldfire_single_format): Declare.
1721 (coldfire_double_format): Likewise.
1722 * real.c (encode_ieee_single): Use canonical_nan_lsbs_set instead
1723 of qnan_msb_set to determine the lower bits of a canonical
1725 (encode_ieee_double): Likewise.
1726 (encode_ieee_quad): Likewise.
1727 (ieee_single_format): Initialize canonical_nan_lsbs_set.
1728 (mips_single_format): Likewise.
1729 (ieee_double_format): Likewise.
1730 (mips_double_format): Likewise.
1731 (ieee_extended_motorola_format): Likewise.
1732 (ieee_extended_intel_96_format): Likewise.
1733 (ieee_extended_intel_128_format): Likewise.
1734 (ieee_extended_intel_96_round_53_format): Likewise.
1735 (ibm_extended_format): Likewise.
1736 (mips_extended_format): Likewise.
1737 (ieee_quad_format): Likewise.
1738 (mips_quad_format): Likewise.
1739 (vax_f_format): Likewise.
1740 (vax_d_format): Likewise.
1741 (vax_g_format): Likewise.
1742 (i370_single_format): Likewise.
1743 (i370_double_format): Likewise.
1744 (decimal_single_format): Likewise.
1745 (decimal_double_format): Likewise.
1746 (decimal_quad_format): Likewise.
1747 (c4x_single_format): Likewise.
1748 (c4x_extended_format): Likewise.
1749 (real_internal_format): Likewise.
1750 (coldfire_single_format): New real_format.
1751 (coldfire_double_format): Likewise.
1752 * config/pdp11/pdp11.c (pdp11_f_format): Initialize
1753 canonical_nan_lsbs_set.
1754 (pdp11_d_format): Likewise.
1755 * config/m68k/m68k.c (override_options): Override REAL_FORMAT_MODE
1756 if TARGET_COLDFIRE_CPU.
1758 2007-01-18 Richard Sandiford <richard@codesourcery.com>
1760 * config/m68k/m68k-protos.h (m68k_output_pic_call): Delete.
1761 (output_call, m68k_legitimize_call_address): Declare.
1762 * config/m68k/m68k.h (EXTRA_CONSTRAINT): Remove unnecessary
1763 parenthesees. Add support for a 'W' constraint.
1764 (LEGITIMATE_PIC_OPERAND_P): Remove SYMBOL_REF_FLAG handling.
1765 (PRINT_OPERAND_PUNCT_VALID_P): Remove comment about 'o'.
1766 (m68k_symbolic_call, m68k_symbolic_jump): Declare.
1767 * config/m68k/m68k.c (m68k_symbolic_call, m68k_symbolic_jump): New
1769 (override_options): Initialize them. Do not set flag_no_function_cse
1770 for TARGET_ID_SHARED_LIBRARY.
1771 (m68k_output_pic_call): Delete.
1772 (m68k_legitimize_call_address): New function.
1773 (print_operand): Remove the %o prefix. Handle the %p prefix.
1774 (output_call): New function.
1775 (m68k_output_mi_thunk): Use m68k_symbolic_jump. Always load the
1776 target address from the GOT if symbolic jumps are not allowed.
1777 * config/m68k/m68k.md (call, general_operand): Do not set
1778 SYMBOL_REF_FLAG. Use m68k_legitimize_call_address instead.
1779 Merge separate flag_pic and !flag_pic define_insns into...
1780 (*call, *call_value): ...these new patterns. Match the address
1781 rather than the containing MEM and require it to be a call_operand.
1782 Use output_call to generate the asm template.
1783 * config/m68k/predicates.md (const_call_operand): New predicate.
1784 (call_operand): Likewise.
1786 2007-01-18 Nathan Sidwell <nathan@codesourcery.com>
1788 * config/m68k/m68k.h (REGISTER_MOVE_COST): Simplify definition.
1789 (STACK_GROWS_DOWNWARD): Define to 1.
1790 (FUNCTION_VALUE, LIBCALL_VALUE, FUNCTION_VALUE_REGNO_P): Equivocate
1791 comments, emphasizing that these values are only defaults.
1792 * config/m68k/linux.h (LINK_SPEC): Fix formatting in #undef.
1793 * config/m68k/m68k.c (const_method): Remove trailing whitespace.
1795 2007-01-18 Richard Sandiford <richard@codesourcery.com>
1797 * config/m68k/m68k.md (cmpsi): Remove outdated flag_pic handling.
1799 2007-01-18 Kazu Hirata <kazu@codesourcery.com>
1800 Richard Sandiford <richard@codesourcery.com>
1802 * config/m68k/m68k.h (DATA_REGNO_P, ADDRESS_REGNO_P, INT_REGNO_P)
1803 (FP_REGNO_P): New macros.
1804 (REGNO_OK_FOR_INDEX_P, REGNO_OK_FOR_BASE_P, REGNO_OK_FOR_DATA_P)
1805 (REGNO_OK_FOR_FP_P, REG_OK_FOR_INDEX_P, REG_OK_FOR_BASE_P): Use them.
1806 (EH_RETURN_STACKADJ_RTX): Use A0_REG.
1807 * config/m68k/m68k.c (m68k_regno_mode_ok): Use the new REGNO macros.
1809 2007-01-18 Nathan Sidwell <nathan@codesourcery.com>
1811 * config.gcc (m68k-*-aout*, m68k-*-coff*, m68020-*-elf*, m68k-*-elf*)
1812 (m68k-*-uclinuxoldabi, m68k-*-uclinux*, m68k-*-rtems*): Add t-floatlib
1814 * config/m68k/t-floatlib: New file, extracting common code from...
1815 * config/m68k/t-m68kbare, config/m68k/t-m68kelf,
1816 * config/m68k/t-uclinux: Here.
1817 * config/m68k/fpgnulib.c: Do not compile extendeed precision
1818 routines on ColdFire targets.
1820 2007-01-18 Nathan Sidwell <nathan@codesourcery.com>
1822 * config.gcc (m68k-*-aout*, m68k-*-coff*, m68020-*-elf*, m68k-*-elf*)
1823 (m68010-*-netbsdelf*, m68k*-*-netbsdelf*, m68k*-*-openbsd*)
1824 (m68k-*-uclinuxoldabi, m68k-*-uclinux*, m68k-*-linux*)
1825 (m68k-*-rtems*): Use tm_file rather than m68k/m68k.h and
1826 explicitly set MOTOROLA to 1.
1827 * config/m68k/m68k.h (MOTOROLA): Simplify definition accordingly.
1829 2007-01-18 Richard Sandiford <richard@codesourcery.com>
1830 Nathan Sidwell <nathan@codesourcery.com>
1832 * config/m68k/m68k.h (PCC_STATIC_STRUCT_RETURN): Delete.
1833 (ASM_OUTPUT_REG_PUSH, ASM_OUTPUT_REG_POP): Add MOTOROLA cases.
1834 * config/m68k/coff.h (REGISTER_PREFIX_MD): Delete.
1835 * config/m68k/m68020-elf.h (LIB_SPEC): Delete.
1836 * config/m68k/m68k-none.h (CC1_SPEC, CPP_SUBTARGET_SPEC): Delete.
1837 * config/m68k/m68kelf.h (IMMEDIATE_PREFIX, REGISTER_PREFIX_MD)
1838 (ASM_OUTPUT_REG_PUSH, ASM_OUTPUT_REG_POP): Delete.
1839 (NO_DOLLAR_IN_LABEL, PCC_STATIC_STRUCT_RETURN): Don't undefine.
1840 (BSS_ASM_OP, ASM_OUTPUT_SKIP, ASM_OUTPUT_ASCII): Delete.
1841 * config/m68k/m68kemb.h (PCC_STATIC_STRUCT_RETURN): Don't undefine.
1842 * config/m68k/linux.h (SIZE_TYPE, PTRDIFF_TYPE, WCHAR_TYPE)
1843 (WCHAR_TYPE_SIZE, TARGET_OBJFMT_CPP_BUILTINS): Delete.
1844 (TARGET_OS_CPP_BUILTINS): Don't define mc68000 and mc68020 here.
1845 (DBX_REGISTER_NUMBER): Delete.
1846 * config/m68k/netbsd-elf.h (IMMEDIATE_PREFIX): Delete.
1847 (PCC_STATIC_STRUCT_RETURN): Don't undefine.
1848 * config/m68k/openbsd.h (PCC_STATIC_STRUCT_RETURN): Define.
1850 2007-01-18 Richard Sandiford <richard@codesourcery.com>
1852 * config.gcc (m68k-*-uclinux*): Add flat.h to $tm_file.
1853 * config/flat.h: New file.
1854 * crtstuff.c (USE_PT_GNU_EH_FRAME): Don't define if
1856 * config/m68k/m68k.h (ASM_PREFERRED_EH_DATA_FORMAT): Do not use
1857 indirect references for -msep-data or -mid-shared-library.
1858 Do not use PC-relative code addresses either.
1859 * config/m68k/m68k.c (override_options): Restrict -fPIC error
1861 * config/m68k/uclinux.h (STARTFILE_SPEC): Define. Use Scrt1.o
1862 for shared libraries and crt1.o for executables. Use crti.o and
1864 (ENDFILE_SPEC): Use crtend.o and crtn.o.
1865 (LIB_SPEC): Suppress -Rlibc.gdb if -static-libc is given.
1866 Do not add -elf2flt or -shared-lib-id options here.
1867 (LINK_SPEC): Define. Pass -elf2flt if no -elf2flt option is given.
1868 Pass -shared-lib-id if -mid-shared-library, taking the library
1869 identifier from -mshared-library-id if given, otherwise
1871 (EH_FRAME_IN_DATA_SECTION): Do not undefine.
1872 (INIT_SECTION_ASM_OP, FINI_SECTION_ASM_OP): Likewise.
1873 (TARGET_OS_CPP_BUILTINS): Define __GXX_MERGED_TYPEINFO_NAMES=0
1874 and __GXX_TYPEINFO_EQUALITY_INLINE=0 if -mid-shared-library.
1875 (DRIVER_SELF_SPECS): Map unadorned PIC options to -msep-data.
1876 * config/m68k/t-uclinux (EXTRA_MULTILIB_PARTS): Add crtbegin.o
1878 * config/m68k/lb1sf68.asm (PICCALL): Use an lea and pc-relative
1879 jump sequence for ISA A and ISA A+.
1880 (PICJUMP): Likewise.
1882 2007-01-18 Richard Sandiford <richard@codesourcery.com>
1884 * config.gcc (m68k-*-uclinux*): Base the port on the common
1885 and m68k GNU/Linux files rather than on the generic ELF ones.
1886 * config/m68k/uclinux.h (TARGET_VERSION): Override.
1887 (TARGET_OS_CPP_BUILTINS): Use LINUX_TARGET_OS_CPP_BUILTINS.
1889 2007-01-18 Julian Brown <julian@codesourcery.com>
1890 Richard Sandiford <richard@codesourcery.com>
1892 * config/m68k/m68k.h (LONG_DOUBLE_TYPE_SIZE): Make 64-bit on ColdFire.
1893 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
1894 * config/m68k/netbsd-elf.h (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Undefine
1896 * config/m68k/uclinux-oldabi.h (LONG_DOUBLE_TYPE_SIZE): Redefine to
1898 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
1900 2007-01-18 Richard Sandiford <richard@codesourcery.com>
1902 * doc/install.texi: Document m68k-uclinuxoldabi.
1903 * config.gcc (m68k-*-uclinuxoldabi): New configuration.
1904 * config/m68k/uclinux-oldabi.h: New file, copied from
1905 config/m68k/uclinux.h.
1907 2007-01-18 Kaz Kojima <kkojima@rr.iij4u.or.jp>
1909 * config/m32r/m32r-protos.h (m32r_expand_epilogue): Declare it.
1910 * config/m32r/m32r.c (m32r_setup_incoming_varargs): Use gen_frame_mem.
1911 (m32r_compute_frame_size): Use unsigned for regno.
1912 (m32r_reload_lr): Use gen_frame_mem.
1914 (m32r_output_function_epilogue): Don't output the function epilogue
1916 (m32r_expand_epilogue): New.
1917 (direct_return): Return false if the function has the interrupt
1919 (m32r_hard_regno_rename_ok): Remove code for the textual epilogue.
1920 * config/m32r/m32r.md (epilogue): New expander.
1921 (return_lr, return_rte): New insns.
1922 (return): Make it expander.
1923 (return_normal): New expander.
1925 2007-01-18 Josh Conner <jconner@apple.com>
1928 * config/rs6000/rs6000.c (rs6000_emit_vector_compare): Add
1929 support for UNLE, UNLT, UNGE, and UNGT.
1931 2007-01-18 Jan Hubicka <jh@suse.cz>
1933 * tree-vrp.c (finalize_jump_threads): Do not call cleanup_cfg by hand.
1934 * tree-tailcall (add_virtual_phis): Likewise.
1935 (optimize_tail_call): Return TODOs.
1936 (execute_tail_calls): Return TODOs.
1937 * tree-ssa-ccp (execute_fold_all_builtins): Do cleanup_cfg via TODO.
1938 * tree-cfgcleanup.c (cleanup_tree_cfg_loop): Return if something
1940 * tree-ssa-forwprop.c (tree_ssa_forward_propagate_single_use_value):
1941 Cleanup cfg using TODO.
1942 * tree-flow.h (cleanup_tree_cfg_loop): Update prototype.
1943 * passes.c (execute_function_todo): When cleanup did something, remove
1945 * tree-cfg.c (pass_build_cfg): Add cleanup_cfg TODO.
1946 (make_edges): Don't cleanup_cfg.
1948 2007-01-18 Uros Bizjak <ubizjak@gmail.com>
1950 * reg-stack.c (subst_stack_regs_pat) [UNSPEC_SINCOS_COS,
1951 UNSPEC_XTRACT_FRACT]: Use generic code for instructions that
1952 operate on the top of stack.
1953 [UNSPEC_SINCOS_SIN, UNSPEC_XTRACT_EXP, UNSPEC_TAN]: Rewrite
1954 register handling of instructions that output to the second
1956 [UNSPEC_TAN_ONE, UNSPEC_TAN_TAN]: Remove.
1957 (move_for_stack_reg): Special-case check for dead destination
1958 stack slot for constant load of 1.0 inside UNSPEC_TAN.
1960 * config/i386/i386.md (UNSPEC_TAN): New constant.
1961 (UNSPEC_TAN_ONE, UNSPEC_TAN_TAN): Remove.
1962 (fptanxf4_i387, fptan_extend<mode>xf4_i387): New patterns
1963 to correctly model move of constant 1.0 to top stack slot.
1964 (*tandf3_1, *tansf3_1, *tanxf3_1): Remove insn patterns.
1965 (unnamed peephole2 pattern): Remove corresponding peephole2
1966 pattern that optimizes tan insn and loading of constant 1.0.
1967 (tanxf2): Use fptanxf4_i387.
1968 (tan<mode>2): Rename from tansf2 and tandf2 and macroize insn
1969 patterns using X87MODEF12 mode macro. Use fptan_extend<mode>xf4_i387
1970 and truncate result to requested mode. Use SSE_FLOAT_MODE_P to
1971 disable patterns for SSE math.
1972 (sincos<mode>3): Use truncxf<mode>2_i387_noop for truncation.
1973 (fyl2x_extend<mode>xf3_i387): Use X87MODEF12 for operand 1.
1975 2007-01-18 Dirk Mueller <dmueller@suse.de>
1976 Richard Guenther <rguenther@suse.de>
1979 * doc/invoke.texi (Warray-bounds): Document -Warray-bounds.
1980 * common.opt (Warray-bounds): Add new warning option.
1981 * c-opts.c (c_common_handle_option): Define -Warray-bounds
1983 * Makefile.in: make tree-vrp.o depend on toplev.h
1984 * tree-vrp.c (vrp_finalize): Call check_array_refs if -Warray-bounds
1986 (check_array_refs, check_array_bounds, check_array_ref): New.
1988 2007-01-18 Jan Hubicka <jh@suse.cz>
1990 * tree-ssa-ccp.c (ccp_finalize): Return if something changed.
1991 (execute_ssa_ccp): Return flags conditionally.
1992 * tree-ssa-propagate.c (substitue_and_fold): Return if something was
1994 * tree-ssa-propagate.h (substitute_and_fold): Update prototype.
1996 2007-01-18 Steven Bosscher <steven@gcc.gnu.org>
1998 * cfgcleanup.c (cleanup_cfg): Detect cfglayout mode and set
1999 the CLEANUP_CFGLAYOUT flag when in cfglayout mode.
2001 * Makefile.c (GTFILES): Add cfglayout.h.
2002 * gengtype.c (open_base_files): Likewise.
2003 * cfglayout.c (cfg_layout_function_footer,
2004 cfg_layout_function_header) Reindent to make gengtype happy.
2005 * cfglayout.h (cfg_layout_function_footer,
2006 cfg_layout_function_header): Add GTY(()) marker.
2008 * ifcvt.c (noce_try_sign_mask): Make sure INSN_B is non-null.
2010 2007-01-18 Ben Elliston <bje@au.ibm.com>
2012 * genautomata.c (write_automata): Include xstrerror output in the
2013 error message if writing the DFA description file fails.
2015 2007-01-17 H.J. Lu <hongjiu.lu@intel.com>
2017 * config/mips/mips-protos.h (mips_output_external): Make it
2019 * config/mips/iris.h (TARGET_ASM_EXTERNAL_LIBCALL): Removed.
2020 * config/mips/mips.c (irix_output_external_libcall): Likewise.
2021 (extern_list): Likewise.
2022 (extern_head): Likewise.
2023 (TARGET_ASM_FILE_END): Likewise.
2024 (mips_file_end): Likewise.
2025 (mips_output_external): Rewritten.
2027 2007-01-18 Ben Elliston <bje@au.ibm.com>
2029 * genpreds.c (write_insn_preds_c): Only write out the function
2030 body for regclass_for_constraint if we have register constraints.
2032 2007-01-17 Tom Tromey <tromey@redhat.com>
2034 * doc/sourcebuild.texi (libgcj Tests): Use sourceware.org.
2035 * doc/install.texi (Testing): Use sourceware.org.
2036 (Binaries): Likewise.
2037 (Specific): Likewise.
2038 * doc/contrib.texi (Contributors): Use sourceware.org.
2040 2007-01-17 Anatoly Sokolov <aesok@post.ru>
2042 * config/avr/avr.h (AVR_HAVE_LPMX): New macro.
2043 (AVR_ENHANCED): Rename to ...
2044 (AVR_HAVE_MUL): ... new.
2045 (avr_enhanced_p): Rename to ...
2046 (avr_have_mul_p): ... new.
2047 (TARGET_CPU_CPP_BUILTINS): Use 'avr_have_mul_p' instead of
2048 'avr_enhanced_p' for "__AVR_ENHANCED__". Define "__AVR_HAVE_MUL__".
2049 * config/avr/avr.c (avr_enhanced_p): Rename to ...
2050 (avr_have_mul_p): ... new.
2051 (base_arch_s): Rename 'enhanced' to 'have_mul'.
2052 (avr_override_options): Use 'avr_have_mul_p' and 'have_mul' instead of
2053 'avr_enhanced_p' and 'enhanced'.
2054 (ashlhi3_out, ashrhi3_out, lshrhi3_out, avr_rtx_costs): Use
2055 AVR_HAVE_MUL instead of AVR_ENHANCED.
2056 * avr.md (*tablejump_enh): Use AVR_HAVE_LPMX instead of AVR_ENHANCED.
2057 (mulqi3, *mulqi3_enh, *mulqi3_call, mulqihi3, umulqihi3, mulhi3,
2058 *mulhi3_enh, *mulhi3_call, mulsi3, *mulsi3_call): Use AVR_HAVE_MUL
2059 instead of AVR_ENHANCED.
2060 (*tablejump_enh): Use AVR_HAVE_LPMX instead of AVR_ENHANCED.
2061 * libgcc.S: Use __AVR_HAVE_MUL__ instead of __AVR_ENHANCED__.
2062 (__tablejump__): Use __AVR_HAVE_LPMX__ instead of __AVR_ENHANCED__.
2064 2007-01-17 Ian Lance Taylor <iant@google.com>
2066 * vec.h (VEC_reserve_exact): Define.
2067 (vec_gc_p_reserve_exact): Declare.
2068 (vec_gc_o_reserve_exact): Declare.
2069 (vec_heap_p_reserve_exact): Declare.
2070 (vec_heap_o_reserve_exact): Declare.
2071 (VEC_OP (T,A,reserve_exact)): New static inline function, three
2073 (VEC_OP (T,A,reserve)) [all versions]: Remove handling of
2075 (VEC_OP (T,A,alloc)) [all versions]: Call ...reserve_exact.
2076 (VEC_OP (T,A,copy)) [all versions]: Likewise.
2077 (VEC_OP (T,a,safe_grow)) [all versions]: Likewise.
2078 * vec.c (calculate_allocation): Add exact parameter. Change all
2080 (vec_gc_o_reserve_1): New static function, from vec_gc_o_reserve.
2081 (vec_gc_p_reserve, vec_gc_o_reserve): Call vec_gc_o_reserve_1.
2082 (vec_gc_p_reserve_exact, vec_gc_o_reserve_exact): New functions.
2083 (vec_heap_o_reserve_1): New static function, from vec_heap_o_reserve.
2084 (vec_heap_p_reserve, vec_heap_o_reserve): Call vec_heap_o_reserve_1.
2085 (vec_heap_p_reserve_exact): New function.
2086 (vec_heap_o_reserve_exact): New function.
2088 2007-01-17 Jan Hubicka <jh@suse.cz>
2090 * ipa-type-escape.c (look_for_casts): Revamp using handled_component_p.
2092 2007-01-17 Eric Christopher <echristo@apple.com>
2094 * config.gcc: Support core2 processor.
2096 2007-01-16 Jan Hubicka <jh@suse.cz>
2098 * tree-ssanames.c (release_dead_ssa_names): Instead of ggc_freeing
2099 the names, just unlink the chain so we don't crash on dangling pointers
2102 2007-01-16 Jan Hubicka <jh@suse.cz>
2104 * cgraph.h (cgraph_decide_inlining_incrementally): Kill.
2105 * tree-pass.h: Reorder to make IPA passes appear toegher.
2106 (pass_early_inline, pass_inline_parameters, pass_apply_inline): Declare.
2107 * cgraphunit.c (cgraph_finalize_function): Do not compute inling
2108 parameters, do not call early inliner.
2109 * ipa-inline.c: Update comments. Include tree-flow.h
2110 (cgraph_decide_inlining): Do not compute inlining parameters.
2111 (cgraph_decide_inlining_incrementally): Return TODOs; assume to
2112 be called with function context set up.
2113 (pass_ipa_inline): Remove unreachable functions before pass.
2114 (cgraph_early_inlining): Simplify assuming to be called from the
2116 (pass_early_inline): New pass.
2117 (cgraph_gate_ipa_early_inlining): New gate.
2118 (pass_ipa_early_inline): Turn into simple wrapper.
2119 (compute_inline_parameters): New function.
2120 (gate_inline_passes): New gate.
2121 (pass_inline_parameters): New pass.
2122 (apply_inline): Move here from tree-optimize.c
2123 (pass_apply_inline): New pass.
2124 * ipa.c (cgraph_remove_unreachable_nodes): Verify cgraph after
2126 * tree-inline.c (optimize_inline_calls): Return TODOs rather than
2128 (tree_function_versioning): Do not allocate dummy struct function.
2129 * tree-inline.h (optimize_inline_calls): Update prototype.
2130 * tree-optimize.c (execute_fixup_cfg): Export.
2131 (pass_fixup_cfg): Remove
2132 (tree_rest_of_compilation): Do not apply inlines.
2133 * tree-flow.h (execute_fixup_cfg): Declare.
2134 * Makefile.in (gt-passes.c): New.
2135 * passes.c: Include gt-passes.h
2136 (init_optimization_passes): New passes.
2137 (nnodes, order): New static vars.
2138 (do_per_function_toporder): New function.
2139 (execute_one_pass): Dump current pass here.
2140 (execute_ipa_pass_list): Don't dump current pass here.
2142 2007-01-16 Janis Johnson <janis187@us.ibm.com>
2144 * config/dfp-bit.c (dfp_compare_op): Return separate value for NaN.
2145 (DFP_NE, DFP_LE, DFP_GE): Return false for NaN.
2147 2007-01-16 David Edelsohn <edelsohn@gnu.org>
2149 * config/rs6000/darwin-ldouble.c: Build file for SOFT_FLOAT.
2150 (strong_alias): Define.
2151 (__gcc_qmul): Provide non-FMA for soft-float.
2157 (__gcc_qunord): New.
2167 * config/rs6000/rs6000.c (rs6000_init_libfuncs): Initialize
2168 soft-float functions.
2169 * config/rs6000/libgcc-ppc-glibc.ver: Version soft-float symbols.
2170 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Do not warn
2171 about long double soft float.
2173 2007-01-16 Dorit Nuzman <dorit@il.ibm.com>
2174 Tehila Meyzels <tehila@il.ibm.com>
2176 * tree-vectorizer.h (is_pattern_stmt_p): New.
2177 * tree-vect-analyze.c (vect_determine_vectorization_factor): Fix
2178 formatting (tabs instead of spaces). Cleanup and clarify setting
2179 of STMT_VINFO_VECTYPE. Call is_pattern_stmt_p.
2180 * tree-vect-transform.c (vect_get_vec_def_for_operand): Fix typo.
2181 (vectorizable_type_demotion): Check that types are integral.
2182 (vectorizable_type_promotion): Likewise.
2183 (vectorizable_store): Fix typo. Eliminate new-line at end of
2186 2007-01-16 Jan Hubicka <jh@suse.cz>
2188 * tree-ssanames.c (release_dead_ssa_names): Remove invalidated
2191 2007-01-15 Eric Christopher <echristo@apple.com>
2193 * ifcvt.c: Include vec.h, vecprim.h.
2194 (check_cond_move_block): New argument regs.
2195 Reorganize. Add registers used to regs.
2196 (cond_move_process_if_block): Use regs set above as
2199 2007-01-15 Eric Christopher <echristo@apple.com>
2201 * config/darwin.h: Update copyright.
2202 (TARGET_OPTION_TRANSLATE_TABLE): Add umbrella.
2203 (LINK_COMMAND_SPEC): Add -u.
2204 (LINK_SPEC): Fix umbrella for above.
2206 2007-01-15 Joseph S. Myers <joseph@codesourcery.com>
2208 * config/soft-fp/op-common.h, config/soft-fp/op-4.h: Update from
2211 2007-01-15 Tom Tromey <tromey@redhat.com>
2213 * doc/sourcebuild.texi (libgcj Tests): Don't mention jacks.
2214 * doc/install.texi (Testing): Don't mention jacks.
2215 (Configuration): Document --enable-java-maintainer-mode. Move
2216 --with-java-home to libgcj-specific section. Document
2218 (Prerequisites): Mention --enable-java-maintainer-mode, ecj1.
2220 2007-01-15 Jan Hubicka <jh@suse.cz>
2222 * tree-ssa-dce.c (DCE_TODOs): New.
2223 (propagate_necessity): Return if something changed.
2224 (eliminate_unnecessary_stmts): Likewise.
2225 (perform_tree_ssa_dce): Return TODO flags when needed.
2226 (pass_dce, pass_dce_loop, pass_cd_dce): Remove TODO flags.
2228 2007-01-15 Uros Bizjak <ubizjak@gmail.com>
2230 * config/i386/i386.md (fyl2xxf3_i387): Rename from fyl2x_xf3.
2231 (fyl2x_extend<mode>xf3_i387): New insn pattern.
2232 (log<mode>2): Rename from logsf2 and logdf2 and macroize insn
2233 patterns using X87MODEF12 mode macro. Extend operand 1
2234 to XFmode. Use SSE_FLOAT_MODE_P to disable patterns for SSE math.
2235 (log10<mode>2): Ditto.
2236 (log2<mode>2): Ditto.
2237 (log1p<mode>2): Ditto.
2238 (logb<mode>2): Ditto.
2239 (fyl2xp1xf3_i387): Rename from fyl2xp1_xf3.
2240 (fyl2xp1_extend<mode>xf3_i387): New insn pattern.
2241 (*fxtractxf3_i387): Rename from *fxtractxf3.
2242 (fxtract_extend<mode>xf3_i387): New insn pattern.
2243 (ilogbsi2): Use match_dup 3, not match_operand:XF 3.
2244 * config/i386/i386.c (ix86_emit_i387_log1p): Use gen_fyl2xp1xf3_i387()
2245 and gen_fyl2xxf3_i387().
2247 2007-01-14 Zdenek Dvorak <dvorakz@suse.cz>
2249 * loop-unswitch.c (unswitch_loop): Do not call fix_loop_placement.
2250 * cfgloopmanip.c (fix_loop_placement): Made static. Use
2251 get_loop_exit_edges. Changed return type to bool.
2252 * cfgloop.h (fix_loop_placement): Declaration removed.
2254 2007-01-14 Dorit Nuzman <dorit@il.ibm.com>
2256 * param.h (MIN_VECT_LOOP_BOUND): New.
2257 * params.def (MIN_VECT_LOOP_BOUND): New.
2258 * tree-vectorizer.c (slpeel_tree_peel_loop_to_edge): Takes another
2259 argument - minimum threshold for number of iterations.
2260 * tree-vectorizer.h (slpeel_tree_peel_loop_to_edge): Add another
2261 argument to declaration.
2262 * tree-vect-analyze.c (vect_analyze_operations): Check value of
2263 MIN_VECT_LOOP_BOUND.
2264 * tree-vect-transform.c (vect_do_peeling_for_loop_bound): Call
2265 slpeel_tree_peel_loop_to_edge with additional argument.
2266 (vect_do_peeling_for_alignment): Likewise.
2267 * doc/invoke.texi (min-vect-loop-bound): Document new param option.
2269 2007-01-14 Uros Bizjak <ubizjak@gmail.com>
2272 * config/i386/i386.c (print_operand) ['z']: Output 'b' for
2275 2007-01-14 Jan Hubicka <jh@suse.cz>
2277 * tree-dfa.c (remove_referenced_var): New function.
2278 * tree-ssa-live.c (remove_unused_locals): Walk referenced vars and
2279 prune referenced vars list too.
2280 * tree-flow.h (remove_referenced_var): Declare.
2282 2007-01-14 Jan Hubicka <jh@suse.cz>
2284 * tree-eh.c (add_stmt_to_eh_region_fn): Do not add call_exprs
2286 (remove_stmt_from_eh_region_fn): Do not remove call_exprs.
2287 (verify_eh_throw_stmt_node, verify_eh_throw_table_statements): Kill.
2288 * except.h (verify_eh_throw_table_statements): Kill prototype.
2289 * cfgexpand.c (expand_gimple_basic_block): Propagate Eh regions
2290 into call exrepssions.
2291 * tree-optimize.c (execute_free_cfg_annotatiosn): Do not call
2293 * tree-cfg.c: Include pointer-set.h.
2294 (verify_node_sharing): Work on pointer set.
2295 (verify_eh_throw_stmt_node): New.
2296 (verify_stmts): Use pointers sets, verify throw_stmt.
2298 2007-01-13 Zdenek Dvorak <dvorakz@suse.cz>
2300 * ipa-reference.c (analyze_function): Consider also addresses taken
2303 2007-01-12 Roger Sayle <roger@eyesopen.com>
2305 * c-typeck.c (null_pointer_constant_p): Replace use of
2306 TREE_CONSTANT_OVERFLOW with TREE_OVERFLOW.
2307 (build_c_cast): Likewise.
2309 2007-01-12 Roger Sayle <roger@eyesopen.com>
2311 * tree.h (force_fit_type_double): Remove unused final argument.
2312 * c-common.c (constant_expression_warning): Replace use of
2313 TREE_CONSTANT_OVERFLOW with TREE_OVERFLOW.
2314 (convert_and_check): Likewise.
2315 (shorten_compare): Update call to force_fit_type_double.
2316 (c_common_truthvalue_conversion) <INTEGER_CST>: Use integer_zerop.
2317 * convert.c (convert_to_pointer): Update call to
2318 force_fit_type_double.
2319 * fold-const.c (force_fit_type_double): Remove overflowed_const
2321 (int_const_binop, fold_convert_const_int_from_int,
2322 fold_convert_const_int_from_real, fold_div_compare,
2323 fold_sign_changed_comparison, fold_unary, fold_negate_const,
2324 fold_abs_const, fold_not_const): Remove the final argument from
2325 calls to force_fit_type_double.
2327 2007-01-12 Andrew Pinski <andrew_pinski@playstation.sony.com>
2329 * configure.ac: Set insn to "nop" for spu-*-* also.
2330 * configure: Regenerate.
2332 2007-01-12 Olga Golovanevsky <olga@il.ibm.com>
2334 * builtins.def : Add BUILT_IN_FREE.
2336 2007-01-12 Jan Hubicka <jh@suse.cz>
2338 PR tree-optimization/30443
2339 * tree-inline.c (tree_function_versioning): Do not optimize when
2340 cloning for inlining.
2342 2007-01-12 Zdenek Dvorak <dvorakz@suse.cz>
2344 * doc/loop.texi: Document recording of loop exits.
2345 * cfgloopmanip.c (loopify, duplicate_loop): Use alloc_loop.
2346 (update_single_exits_after_duplication,
2347 update_single_exit_for_duplicated_loop,
2348 update_single_exit_for_duplicated_loops): Removed.
2349 (duplicate_loop_to_header_edge): Do not call
2350 update_single_exits_after_duplication and
2351 update_single_exit_for_duplicated_loops.
2352 (loop_version): Do not update single_exit information.
2353 (fix_loop_structure): Use record_loop_exits instead of
2354 mark_single_exit_loops.
2355 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update
2356 the lists of loop exits.
2357 * cfghooks.c (redirect_edge_and_branch, redirect_edge_and_branch_force,
2358 split_edge, merge_blocks): Update the lists of loop exits.
2359 * modulo-sched.c (sms_schedule): Pass LOOPS_HAVE_RECORDED_EXITS to
2360 loop_optimizer_init.
2361 * loop-init.c (loop_optimizer_init): Call record_loop_exits instead
2362 of mark_single_exit_loops.
2363 (loop_optimizer_finalize): Call release_recorded_exits.
2364 * tree-ssa-loop.c (tree_loop_optimizer_init): Pass
2365 LOOPS_HAVE_RECORDED_EXITS to loop_optimizer_init.
2366 * tree-vectorizer.c (slpeel_tree_duplicate_loop_to_edge_cfg): Do not
2367 update single exit information.
2368 * lambda-code.c (perfect_nestify): Ditto.
2369 * cfgloop.c (flow_loop_free): Destroy the list of exits of the loop.
2370 (mark_single_exit_loops): Removed.
2371 (alloc_loop, loop_exit_hash, loop_exit_eq, loop_exit_free,
2372 get_exit_descriptions, rescan_loop_exit, record_loop_exits,
2373 dump_recorded_exit, dump_recorded_exits, release_recorded_exits): New
2375 (get_loop_exit_edges, single_exit): Use recorded exit lists.
2376 (add_bb_to_loop, remove_bb_from_loops): Update the lists of loop exits.
2377 (verify_loop_structure): Verify consistency of the exit lists.
2378 (flow_loops_find): Use alloc_loop. Initialize exits hash.
2379 (set_single_exit): Removed.
2380 * cfgloop.h (struct loop_exit): New function.
2381 (struct loop): single_exit_ field replaced by exits field.
2382 (LOOPS_HAVE_MARKED_SINGLE_EXITS): Replaced by LOOPS_HAVE_RECORDED_EXITS.
2383 (struct loops): Added exits hash.
2384 (mark_single_exit_loops, set_single_exit): Declaration removed.
2385 (release_recorded_exits, record_loop_exits, rescan_loop_exit): Declare.
2387 2007-01-12 Richard Sandiford <richard@codesourcery.com>
2389 * doc/invoke.texi: Avoid use of @headitem.
2391 2007-01-12 Richard Sandiford <richard@codesourcery.com>
2393 * cse.c (cse_insn): Move HAVE_CC0 code after declarations.
2395 2007-01-12 Richard Sandiford <richard@codesourcery.com>
2397 * doc/install.texi: Fix m68k-*-* anchor and add m68k-*-* to the
2400 2007-01-12 Nathan Sidwell <nathan@codesourcery.com>
2401 Richard Sandiford <richard@codesourcery.com>
2403 * doc/invoke.texi: Document -mno-bitfield, -mno-rtd and -mno-short.
2404 * config/m68k/m68k.opt: Resort options.
2405 (mbitfield, mrtd, mshort): Remove RejectNegative properties.
2407 2007-01-12 Nathan Sidwell <nathan@codesourcery.com>
2408 Richard Sandiford <richard@codesourcery.com>
2410 * doc/invoke.texi: Document the macros that are defined by
2411 m68k's -mtune and -mhard-float options.
2412 * config/m68k/m68k-protos.h (m68k_cpp_cpu_ident) Declare.
2413 (m68k_cpp_cpu_family): Likewise.
2414 * config/m68k/m68k.h (TARGET_CPU_CPP_BUILTINS): Add a full set
2415 of __ucfv*__ macros. Define __mcffpu__ if generating code for
2416 ColdFire FPUs. Define __mcf_cpu_* and __mcf_family_* macros.
2417 * config/m68k/m68k.c (m68k_cpp_cpu_ident): New function.
2418 (m68k_cpp_cpu_family): Likewise.
2420 2007-01-12 Richard Sandiford <richard@codesourcery.com>
2422 * config/m68k/m68k.h (TARGET_CPU_CPP_BUILTINS): Treat all mc68*
2423 macros besides mc68000 as tuning macros. Use a switch statement
2424 to set them and mcpu32.
2426 2007-01-12 Julian Brown <julian@codesourcery.com>
2428 * config/m68k/m68k.h: Use TARGET_68040 instead of TARGET_68040_ONLY.
2429 (TARGET_68040_ONLY): Rename to...
2430 (TARGET_68040): ...this.
2431 * config/m68k/m68k.c: Use TARGET_68040 instead of TARGET_68040_ONLY.
2432 * config/m68k/m68k.md: Likewise.
2434 2007-01-12 Julian Brown <julian@codesourcery.com>
2435 Nathan Sidwell <nathan@codesourcery.com>
2436 Richard Sandiford <richard@codesourcery.com>
2438 * config.gcc (m680[012]0-*-*, m68k*-*-*): Set m68k_cpu_ident to
2439 the -mcpu= argument associated with the --with-cpu setting.
2440 Define M68K_DEFAULT_TUNE to the default -mtune= option,
2441 if different from the one implied by the -mcpu setting.
2442 Accept --with-cpu=FOO if FOO is listed in m68k-devices.def,
2443 using mcpu=FOO as the default CPU option. Set target_cpu_default2.
2444 * doc/invoke.texi: Mention ColdFire in the introduction to the
2445 m68k options. Document the new -march, -mcpu, -mtune, -mdiv,
2446 -mno-div and -mhard-float options. Make -m68881 a synonym for
2447 -mhard-float. Document the previously-undocumented -m5206e,
2448 -m528x, -m5307 and -m5407 options. Tweak the existing option
2449 documentation for consistency.
2450 * doc/install.texi: Mention new --with-cpu arguments.
2451 * config/m68k/m68k.h (OPTION_DEFAULT_SPECS): Only use the
2452 default CPU if neither -mcpu nor -march are specified.
2453 (ASM_CPU_SPEC): Pass down -mcpu and -march options.
2454 (TARGET_CPU_CPP_BUILTINS): Set __mcfisa*__ macros from
2455 TARGET_ISA*. Set the legacy __mcf*__ cpu macros in the same way,
2456 using m68k_tune to decide between families that implement the
2457 same ISA. Use m68k_tune to set __mcfv4e__.
2458 (FL_BITFIELD, FL_68881, FL_COLDFIRE, FL_CF_HWDIV, FL_CF_MAC)
2459 (FL_CF_EMAC, FL_CF_EMAC_B, FL_CF_USP, FL_CF_FPU, FL_ISA_68000)
2460 (FL_ISA_68010, FL_ISA_68020, FL_ISA_68040, FL_ISA_A, FL_ISA_B)
2461 (FL_ISA_C, FL_ISA_MMU): New macros.
2462 (MASK_COLDFIRE): Delete.
2463 (TARGET_68010, TARGET_68020, TARGET_68040_ONLY, TARGET_COLDFIRE)
2464 (TARGET_ISAB): Redefine in terms of m68k_cpu_flags.
2465 (TARGET_68881, TARGET_COLDFIRE_FPU): Redefine in terms of m68k_fpu.
2466 (TARGET_HARD_FLOAT): Do not define here.
2467 (TARGET_ISAAPLUS, TARGET_ISAC): New macros.
2468 (TUNE_68000): New macro.
2469 (TUNE_68000_10): Redefine in terms of TUNE_68000 and TUNE_68010.
2470 (TUNE_68010, TUNE_68030, TUNE_68040, TUNE_68060, TUNE_CPU32)
2471 (TUNE_CFV2): Redefine in terms of m68k_tune.
2472 (uarch_type, target_device, fpu_type): New enums.
2473 (m68k_cpu, m68k_tune, m68k_fpu, m68k_cpu_flags): Declare.
2474 * config/m68k/m68k.c (TARGET_DEFAULT): Remove MASK_68881.
2475 (FL_FOR_isa_00, FL_FOR_isa_10, FL_FOR_isa_20, FL_FOR_isa_40)
2476 (FL_FOR_isa_cpu32, FL_FOR_isa_a, FL_FOR_isa_aplus, FL_FOR_isa_b)
2477 (FL_FOR_isa_c): New macros.
2478 (m68k_isa): New enum.
2479 (m68k_target_selection): New structure.
2480 (all_devices, all_isas, all_microarchs): New tables.
2481 (m68k_cpu_entry, m68k_arch_entry, m68k_tune_entry, m68k_cpu)
2482 (m68k_tune, m68k_fpu, m68k_cpu_flags): New variables.
2483 (MASK_ALL_CPU_BITS): Delete.
2484 (m68k_find_selection): New function.
2485 (m68k_handle_option): Handle -mcpu=, -march= and -mtune=.
2486 Map the legacy target options to a combination of the new ones.
2487 (override_options): Set m68k_cpu, m68k_tune, m68k_fpu and
2488 m68k_cpu_flags. Handle M68K_DEFAULT_TUNE. Use m68k_cpu_flags
2489 to derive default MASK_BITFIELD, MASK_CF_HWDIV and MASK_HARD_FLOAT
2491 * config/m68k/m68k.opt (m5200, m5206e, m528x, m5307, m5407, mcfv4e)
2492 (m68010, m68020, m68020-40, m68020-60, m68030, m68040): Remove Mask
2494 (m68881, msoft-float): Change mask from 68881 to HARD_FLOAT.
2495 (march=, mcpu=, mdiv, mhard-float, mtune=): New options.
2496 * config/m68k/m68k-devices.def: New file.
2498 2007-01-12 Richard Sandiford <richard@codesourcery.com>
2499 Nathan Sidwell <nathan@codesourcery.com>
2501 * config/m68k/m68k.h (ASM_CPU_SPEC, ASM_SPEC, EXTRA_SPECS)
2502 (SUBTARGET_EXTRA_SPECS): New macros.
2503 * config/m68k/linux.h (ASM_SPEC): Remove CPU flags;
2504 use %(asm_cpu_spec) instead.
2505 * config/m68k/m68k-none.h (ASM_SPEC): Likewise.
2506 * config/m68k/openbsd.h (ASM_SPEC): Likewise.
2507 * config/m68k/netbsd-elf.h (ASM_SPEC): Likewise.
2508 (EXTRA_SPECS): Rename to...
2509 (SUBTARGET_EXTRA_SPECS): ...this.
2511 2007-01-12 Nathan Sidwell <nathan@codesourcery.com>
2512 Richard Sandiford <richard@codesourcery.com>
2513 Julian Brown <julian@codesourcery.com>
2515 * config.gcc (m68k-*-aout*, m68k-*-coff*, m68020-*-elf*, m68k-*-elf*)
2516 (m68k-*-uclinux*, m68k-*-linux*, m68k-*-rtems*): Set default_m68k_cpu
2517 to the configuration's default CPU.
2518 (m68010-*-netbsdelf*, m68k*-*-netbsdelf*, m68k*-*-openbsd*): Likewise.
2519 Remove default masks.
2520 (m680[012]0-*-*): Set the default with_cpu to the first part of
2522 (m68k*-*-*): Set the default with_cpu to m$default_m68k_cpu.
2523 (m68k*-*-linux): Extend the --with-cpu handling to...
2524 (m680[012]0-*-*, m68k*-*-*): ...these configurations. Allow m68000
2525 and m68010. Don't set target_cpu_default2.
2526 * doc/install.texi: Document --with-cpu for m68k.
2527 * config/m68k/m68k.h (OPTION_DEFAULT_SPECS): Define.
2528 * config/m68k/m68k-none.h (TARGET_CPU_DEFAULT, M68K_CPU_m68k)
2529 (M68K_CPU_m68000, M68K_CPU_m68010, M68K_CPU_m68020, M68K_CPU_m68030)
2530 (M68K_CPU_m68040, M68K_CPU_m68302, M68K_CPU_m68332, TARGET_DEFAULT)
2531 (ASM_CPU_DEFAULT_SPEC, CC1_CPU_DEFAULT_SPEC): Delete.
2532 (ASM_SPEC): Remove use of %(asm_cpu_default).
2533 (EXTRA_SPECS, SUBTARGET_EXTRA_SPECS, MULTILIB_DEFAULTS): Delete.
2534 * config/m68k/linux.h (TARGET_DEFAULT): Delete.
2535 (CPP_SPEC): Merge definitions. Do not handle __HAVE_68881__ here.
2536 * config/m68k/netbsd-elf.h (TARGET_OS_CPP_BUILTINS): Define
2537 __HAVE_FPU__ if TARGET_HARD_FLOAT.
2538 (TARGET_DEFAULT): Delete.
2539 (EXTRA_SPECS): Delete cpp_cpu_default_spec, cpp_cpu_spec,
2540 cpp_fpu_spec, asm_default_spec and netbsd_cpp_spec.
2541 (CPP_CPU_SPEC): Delete.
2542 (TARGET_VERSION): Merge definitions, using TARGET_68010 to pick
2543 the appropriate string.
2544 (CPP_CPU_DEFAULT_SPEC, ASM_DEFAULT_SPEC, CPP_FPU_SPEC): Delete.
2545 (CPP_SPEC): Define to NETBSD_CPP_SPEC.
2546 (ASM_SPEC): Don't use %(asm_default_spec).
2547 * config/m68k/m68k.c (TARGET_DEFAULT_TARGET_FLAGS): Remove
2548 TARGET_DEFAULT and add MASK_68881.
2549 * config/m68k/m68k.md: Remove mention of TARGET_DEFAULT from comments.
2551 2007-01-12 Richard Sandiford <richard@codesourcery.com>
2553 * config.gcc (m68010-*-netbsdelf*): Add MASK_68010.
2554 (m68k*-*-netbsdelf*, m68k*-*-openbsd*, m68k*-linux*): Add
2555 MASK_68010 alongside MASK_68020.
2556 * doc/invoke.texi: Document -m68010.
2557 * config/m68k/m68k.opt (m68010): New.
2558 * config/m68k/m68k.h (TARGET_CPU_CPP_BUILTINS): Define mc68010
2560 (TUNE_68010): New macro.
2561 * config/m68k/m68k-none.h (M68K_CPU_m68k, M68K_CPU_m68010)
2562 (M68K_CPU_m68020, M68K_CPU_m68030, M68K_CPU_m68040)
2563 (M68K_CPU_m68332): Add MASK_68010.
2564 * config/m68k/linux.h (TARGET_DEFAULT): Add MASK_68010 to
2565 fallback definition.
2566 * config/m68k/netbsd-elf.h (CPP_CPU_SPEC): Remove now-redundant
2568 * config/m68k/m68k.c (MASK_ALL_CPU_BITS): Add MASK_68010.
2569 (m68k_handle_option): Handle OPT_m68010. Add MASK_68010
2570 to all entries that use MASK_68020.
2571 (output_move_simode_const, output_move_himode, output_move_qimode)
2572 (output_move_stricthi, output_move_strictqi): Use TARGET_68010
2573 instead of TARGET_68020 to select clr behavior. Remove comment
2574 about there being no TARGET_68010.
2575 * config/m68k/m68k.md: Likewise throughout.
2577 2007-01-12 Julian Brown <julian@codesourcery.com>
2579 * config/m68k/m68k.h (TARGET_ISAB): New macro.
2580 * config/m68k/m68k.c: Use TARGET_ISAB rather than TARGET_CFV4.
2581 * config/m68k/m68k.md: Likewise.
2583 2007-01-12 Julian Brown <julian@codesourcery.com>
2585 * config/m68k/m68k.h (LEGITIMATE_INDEX_P, LEGITIMIZE_ADDRESS): Use
2586 TARGET_COLDFIRE_FPU instead of TARGET_CFV4E.
2588 2007-01-12 Julian Brown <julian@codesourcery.com>
2590 * config/m68k/m68k.h (TUNE_68040_60): New macro.
2591 * config/m68k/m68k.c (standard_68881_constant_p): Use it.
2592 * config/m68k/m68k.md: Likewise.
2594 2007-01-12 Julian Brown <julian@codesourcery.com>
2595 Richard Sandiford <richard@codesourcery.com>
2597 * config/m68k/m68k.h (TARGET_CPU_CPP_BUILTINS): Use TUNE_68030
2598 instead of TARGET_68030, TUNE_68040 instead of TARGET_68040,
2599 TUNE_68060 instead of TARGET_68060 and TUNE_CPU32 instead of
2601 (TARGET_CPU32): Rename to...
2602 (TUNE_CPU32): ...this.
2603 (TUNE_68000_10, TUNE_68030, TUNE_68040, TUNE_68060)
2604 (TUNE_CFV2): New macros.
2605 * config/m68k/netbsd-elf.h (LONG_DOUBLE_TYPE_SIZE): Simplify;
2606 remove conditions that are implied by TARGET_68020.
2607 * config/m68k/m68k.c (m68k_output_function_prologue): Use TUNE_68040
2608 instead of TARGET_68040 and TUNE_CPU32 instead of TARGET_CPU32.
2609 (m68k_output_function_epilogue): Likewise.
2610 (m68k_rtx_costs): Likewise. Use TUNE_68060 instead of TARGET_68060
2611 and TUNE_CFV2 instead of TARGET_5200. Use TUNE_68000_10 instead of
2612 "!TARGET_68020 && !TARGET_COLDFIRE" to choose between 68000 and
2613 non-68000 timings. Refactor multiplication and division costs.
2614 (output_addsi3): Use TUNE_68040 instead of TARGET_68040 and
2615 TUNE_CPU32 instead of TARGET_CPU32.
2616 (standard_68881_constant_p): Use TUNE_68040 instead of TARGET_68040
2617 and TUNE_68060 instead of TARGET_68060.
2618 * config/m68k/m68k.md: Use TUNE_68040 instead of TARGET_68040,
2619 TUNE_68060 instead of TARGET_68060, and TUNE_CPU32 instead of
2621 (movsi_const0): Use TUNE_68000_10 rather than "!TARGET_68020
2622 && !TARGET_COLDFIRE" to choose between moveq and clr.
2623 Likewise in the unnamed movsf pattern.
2624 (ashlsi_17_24, lshrsi_17_24): Guard with TUNE_68000_10 rather than
2625 "!TARGET_68020 && !TARGET_COLDFIRE". Likewise the unnamed
2628 2007-01-12 Richard Sandiford <richard@codesourcery.com>
2630 * config/m68k/m68k.h (TARGET_CPU_CPP_BUILTINS): Increase amount
2631 of tabbing before backslashes.
2633 2007-01-11 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
2635 * pa-linux.h (ASM_OUTPUT_INTERNAL_LABEL): Undefine.
2636 * pa.h (ASM_OUTPUT_LABEL): Output colon when using GAS.
2637 (ASM_OUTPUT_INTERNAL_LABEL): Define.
2639 2007-01-11 Zdenek Dvorak <dvorakz@suse.cz>
2641 * tree-ssa-loop-ivopts.c (extract_cond_operands): Split from
2642 find_interesting_uses_cond.
2643 (find_interesting_uses_cond): Use extract_cond_operands.
2644 (rewrite_use_compare): Use extract_cond_operands and
2645 force_gimple_operand_bsi. Do not call update_stmt.
2646 (determine_use_iv_cost_condition): Use extract_cond_operands.
2647 Return cheaper of using original bound and changing the exit bound.
2649 2007-01-11 Zdenek Dvorak <dvorakz@suse.cz>
2651 PR tree-optimization/29516
2652 * tree-ssa-address.c (tree_mem_ref_addr, add_to_parts,
2653 most_expensive_mult_to_index, addr_to_parts,
2654 create_mem_ref, maybe_fold_tmr): Make the type of
2655 fields of TARGET_MEM_REF sizetype.
2656 (move_fixed_address_to_symbol, move_pointer_to_base):
2658 * tree.def (TARGET_MEM_REF): Add comment on types of
2661 2007-01-11 Joseph Myers <joseph@codesourcery.com>
2663 * c-common.c (vector_types_convertible_p): Treat opaque types as
2664 always convertible if they have the same size, but not otherwise.
2666 2007-01-11 Steven Bosscher <steven@gcc.gnu.org>
2668 * ifcvt.c (struct noce_if_info): Add comments to the fields.
2669 Remove the b_unconditional field.
2670 (noce_try_sign_mask): Do not look at b_unconditional.
2671 (noce_process_if_block): Do not use merge_if_blocks. Update
2672 the CFG here. Do not set b_unconditional.
2673 (cond_move_process_if_block): Likewise.
2674 (find_cond_trap): Likewise.
2675 (check_cond_move_block): Require simple jump insns at the end
2678 2007-01-11 Jan Hubicka <jh@suse.cz>
2680 PR tree-optimization/1046
2681 * tree-tailcall.c (suitable_for_tail_call_opt_p): Use TREE_ADDRESSABLE
2682 when alias info is not ready.
2683 (pass_tail_recursion): Do not require aliasing.
2684 * tree-ssa-copyrename.c (pass_rename_ssa_cop): Likewise.
2685 * tree-ssa-ccp.c (pass_ccp, pass_fold_builtins): Likewise.
2686 * tree-ssa-copy.c (pass_copy_prop): Likewise.
2687 * tree-ssa-forwprop.c (pass_forwprop): Likewise.
2688 * tree-ssa-dce.c (pass_dce, pass_dce_loop, pass_cd_dce): Likewise.
2689 * passes.c (init_optimization_passes): Execute rename_ssa_copies,
2690 ccp, forwprop, copy_prop, merge_phi, copy_prop, dce and tail recursion
2692 * tree-ssa-operands.c (add_virtual_operand, get_indirect_ref_operand):
2693 When aliasing is not build, mark statement as volatile.
2695 2007-01-11 Tom Tromey <tromey@redhat.com>
2697 PR preprocessor/15185, PR preprocessor/20989:
2698 * doc/cppopts.texi <-MT>: Update description of algorithm for
2699 computing default target.
2700 <-M, -MD>: Reword "basename" text.
2702 2007-01-11 Roger Sayle <roger@eyesopen.com>
2704 * builtins.c (expand_builtin_pow, expand_builtin_powi,
2705 fold_builtin_cabs, fold_builtin_sqrt, fold_builtin_trunc,
2706 fold_builtin_floor, fold_builtin_ceil, fold_builtin_round,
2707 fold_builtin_int_int_roundingfn, fold_builtin_bitop,
2708 fold_builtin_bswap, real_constp, fold_builtin_pow,
2709 fold_builtin_powi, fold_builtin_signbit, fold_builtin_copysign,
2710 do_mpfr_arg1, do_mpfr_arg2, do_mpfr_arg3, do_mpfr_sincos): Replace
2711 uses of the macro TREE_CONSTANT_OVERFLOW with TREE_OVERFLOW.
2712 * convert.c (convert_to_pointer): Likewise.
2713 * expr.c (highest_pow2_factor, expand_expr_real_1): Likewise.
2714 * fold-const.c (force_fit_type, fold_negate_expr, int_const_binop,
2715 const_binop, fold_convert_const_int_from_int,
2716 fold_convert_const_int_from_real,
2717 fold_convert_const_real_from_real, sign_bit_p,
2718 optimize_minmax_comparison, extract_muldiv_1, fold_div_compare,
2719 fold_sign_changed_comparison, fold_unary, fold_comparison,
2720 fold_binary, multiple_of_p, tree_Expr_non_zero_p,
2721 fold_negate_const, fold_abs_const, fold_not_const): Likewise.
2722 * print-tree.c (print_node_brief, print_node): Likewise.
2723 * stor-layout.c (place_field, layout_type): Likewise.
2724 * tree-chrec.c (keep_cast): Likewise.
2725 * tree.c (build_vector, build_real, build_real_from_int_cst,
2726 build_complex): Likewise.
2728 2007-01-11 Roger Sayle <roger@eyesopen.com>
2730 * tree.h (TREE_CONSTANT_OVERFLOW): Obsolete. For the time being,
2731 treat TREE_CONSTANT_OVERFLOW as a synonym of TREE_OVERFLOW.
2733 2007-01-11 Paolo Bonzini <bonzini@gnu.org>
2735 * configure.ac (strict1_warn): Rename to strict_warn.
2736 (WERROR, --enable-werror, symlink hacks, stage1_cflags,
2737 cc_set_by_configure, quoted_cc_set_by_configure,
2738 stage_prefix_set_by_configure, quoted_stage_prefix_set_by_configure,
2739 all_boot_languages, all_stagestuff): Remove.
2740 (target_list): Remove bootstrap targets.
2741 * Makefile.in (quickstrap): Unconditionally make a synonym of all.
2742 (BOOT_LANGUAGES, STAGE1_CFLAGS, STAGE1_CHECKING,
2743 REMAKEFLAGS, FLAGS_TO_PASS, PREPEND_DOTDOT_TO_RELATIVE_PATHS,
2744 SUBDIR_FLAGS_TO_PASS, WERROR_FLAGS, STRICT2_WARN, LANG_STAGESTUFF,
2745 VOL_FILES, POSTSTAGE1_FLAGS_TO_PASS, STAGE2_FLAGS_TO_PASS,
2746 STAGEPROFILE_FLAGS_TO_PASS, STAGEFEEDBACK_FLAGS_TO_PASS, stage1_build,
2747 stage1_copy, stage2_build, stage2_copy, stageprofile_build,
2748 stageprofile_copy, stage3_build, stage3_copy, stagefeedback_build,
2749 stagefeedback_copy, stage4_build, clean_s1, clean_sw, bootstrap,
2750 bootstrap-lean, bootstrap2, bootstrap2-lean, bootstrap3,
2751 bootstrap3-lean, bootstrap4, bootstrap4-lean, unstage1, unstage2,
2752 unstage3, unstage4, unstageprofile, unstagefeedback, restage, restage2,
2753 restage3, restage4, restageprofile, restagefeedback, bubbleestrap,
2754 cleanstrap, unstrap, restrap, *compare, *compare3, *compare4,
2755 *compare-lean, *compare3-lean, *compare4-lean, stage1-start, stage1,
2756 stage2-start, stage2, stage3-start, stage3, stage4-start, stage4,
2757 stageprofile-start, stageprofile, stagefeedback-start, stagefeedback,
2758 risky-stage1, risky-stage2, risky-stage3, risky-stage4): Remove.
2759 (ORDINARY_FLAGS_TO_PASS): Rename to FLAGS_TO_PASS.
2760 (STAGECOPYSTUFF, STAGEMOVESTUFF): Consolidate into MOSTLYCLEANFILES.
2761 (mostlyclean): Adjust.
2762 (clean, distclean): Don't mention bootstrap stuff.
2763 * configure: Regenerate.
2764 * ada/config-lang.in, cp/config-lang.in, forttran/config-lang.in,
2765 java/config-lang.in, objc/config-lang.in, objcp/config-lang.in,
2766 treelang/config-lang.in (stagestuff): Remove.
2767 * doc/sourcebuild.texi (stage1, stage2, stage3, stage4,
2768 stageprofile, stagefeedback, stagestuff): Remove mention.
2770 2007-01-11 Nick Clifton <nickc@redhat.com>
2772 * config/mcore/predicates.md (mcore_general_movesrc_operand):
2774 (mcore_general_movdst_operand): Do not accept CONST_INTs.
2775 (mcore_arith_K_S_operand): Run the test for the S constraint not
2776 the test for the M constraint.
2777 (mcore_addsub_operand): Do not accept integer values that are
2778 larger than 32 bits.
2779 * config/mcore/mcore.md: Remove unused constraints from split.
2780 (andsi3): Use HOST_WIDE_INT instead of int to hold an INTVAL.
2782 (allocate_stack): Likewise.
2783 * config/mcore/mcore.c (mcore_print_operand): Restrict output of P
2784 operands to 32 bits.
2785 (mcore_const_costs): Use HOST_WIDE_INT instead of int to hold an
2787 (mcore_and_cost, mcore_modify_comparison, const_ok_for_mcore,
2788 mcore_const_ok_for_inline, mcore_const_trick_uses_not,
2789 try_constant_tricks, mcore_num_ones, mcore_num_zeros,
2790 mcore_output_bclri, mcore_output_andn, output_inline_const,
2791 mcore_output_move, mcore_output_movedouble): Likewise.
2792 (mcore_output_cmov): Use CONST_OK_FOR_M and CONST_OK_FOR_N.
2793 (output_inline_const): Likewise.
2794 (output_inline_const): Fix format strings used in sprintf
2796 * config/mcore/mcore-protos.h: Update prototypes for changed
2797 functions in mcore.c.
2798 * config/mcore/mcore.h (CONST_OK_FOR_I): Cast values to
2799 HOST_WIDE_INT and not int.
2800 (CONST_OK_FOR_J, CONST_OK_FOR_K, CONST_OK_FOR_L, CONST_OK_FOR_M,
2801 CONST_OK_FOR_N): Likewise.
2802 (LEGITIMATE_CONSTANT_P): Also check CONSTANT_P.
2803 (GO_IF_LEGITIMATE_INDEX): Use HOST_WIDE_INT instead of int to hold
2806 2007-01-10 Jan Hubicka <jh@suse.cz>
2808 * tree-vrp.c (remove_range_assertions): Release defs.
2809 * tree-ssa-loop-ivopts.c (rmeove_statement): Likewise.
2810 * tree-ssa-dom.c (remove_stmt_or_phi): Likewise.
2812 2007-01-10 Paul Brook <paul@codesourcery.com>
2814 * config/arm/arm.c (arm_rtx_costs_1): Handle mutiply-subtract.
2815 * config/arm/arm.md (mulsi3subsi): New insn.
2817 2007-01-10 Zdenek Dvorak <dvorakz@suse.cz>
2819 * tree-ssa-loop-manip.c (tree_unroll_loop): Make it a wrapper over ...
2820 (tree_transform_and_unroll_loop): New.
2821 * tree-flow.h (transform_callback, tree_transform_and_unroll_loop):
2824 2007-01-10 Robert Kennedy <jimbob@google.com>
2826 * fold-const.c (fold_comparison): Fold comparisons like (x *
2827 1000 < 0) to (x < 0).
2829 2007-01-10 Ian Lance Taylor <iant@google.com>
2831 * tree-pretty-print.c (dump_generic_node): Print parentheses when
2832 operands have the same priority.
2834 2007-01-10 Tom Tromey <tromey@redhat.com>
2836 * fold-const.c (fold_truthop): Don't check can_use_bit_fields_p.
2837 (fold_binary): Likewise.
2838 * langhooks.c (lhd_can_use_bit_fields_p): Removed.
2839 * langhooks-def.h (lhd_can_use_bit_fields_p): Removed.
2840 (LANG_HOOKS_CAN_USE_BIT_FIELDS_P): Removed.
2841 (LANG_HOOKS_INITIALIZER): Remove LANG_HOOKS_CAN_USE_BIT_FIELDS_P.
2842 * langhooks.h (struct lang_hooks): Removed field
2843 'can_use_bit_fields_p'.
2845 2007-01-10 Ralf Corsépius <ralf.corsepius@rtems.org>
2847 * config/bfin/t-bfin, config/bfin/t-bfin-elf: Remove GCC_CFLAGS.
2849 2007-01-10 Razya Ladelsky <razya@il.ibm.com>
2851 * function.c (get_last_funcdef_no): New function.
2852 * function.h (get_last_funcdef_no): Declare.
2853 * tree-inline.c (initialize_cfun): Add initialization.
2854 (tree_function_versioning): Cleanup.
2856 2007-01-10 Jan Hubicka <jh@suse.cz>
2858 * tree-inline.c (setup_one_parameter): Do not propagate into abnormal
2861 2007-01-10 Sa Liu <saliu@de.ibm.com>
2862 Ben Elliston <bje@au.ibm.com>
2864 * spu.h (STACK_SAVE_AREA): Use VOIDmode for SAVE_FUNCTION, SImode
2865 for SAVE_NONLOCAL and Pmode for any other save level.
2866 * spu-protos.h (spu_restore_stack_block): Declare.
2867 * spu.md (save_stack_block): Remove.
2868 (restore_stack_block): Call spu_restore_stack_block.
2869 * spu.c (spu_restore_stack_block): New function.
2870 (spu_expand_epilogue): Remove old comment.
2872 2007-01-09 Zdenek Dvorak <dvorakz@suse.cz>
2874 PR tree-optimization/30322
2875 * tree-ssa-loop-ivopts.c (fold_affine_expr, iv_value): Removed.
2876 (cand_value_at): Return the value as aff_tree.
2877 (may_eliminate_iv): Convert the bound from aff_tree to tree.
2878 * tree-affine.c (aff_combination_add_cst, aff_combination_add_product,
2879 aff_combination_mult): New functions.
2880 (aff_combination_add): Use aff_combination_add_cst.
2881 (aff_combination_convert): Allow conversions to a wider type.
2882 (tree_to_aff_combination): Handle BIT_NOT_EXPR.
2883 * tree-affine.h (aff_combination_mult): Declare.
2885 2007-01-09 Carlos O'Donell <carlos@codesourcery.com>
2887 * doc/tm.texi: Update documentation to reflect reality of exec
2888 and start file search behaviours. Update copyright year.
2889 * doc/invoke.texi: Explain how GCC_EXEC_PREFIX is used to find
2890 header file directories.
2892 2007-01-09 Uros Bizjak <ubizjak@gmail.com>
2894 * config/i386/i386.md (*sinxf2): Rename to *sinxf2_i387.
2895 (*cosxf2): Rename to cosxf2_i387.
2896 (*sindf2, *sinsf2): Extend operand 1 to XFmode. Macroize patterns
2897 using X87MODEF12 mode macro. Rename patterns to
2898 *sin_extend<mode>xf2_i387. Use SSE_FLOAT_MODE_P to disable patterns
2900 (*cosdf2, *cossf2): Ditto.
2901 (sincosdf3, sincossf3): Ditto. Rewrite corresponding splitters
2902 to match extended input operands.
2903 (sincos<mode>3): New expander.
2904 (*sinextendsfdf2, *cosextendsfdf2, *sincosextendsfdf3): Remove
2905 insn patterns and corresponding splitters.
2907 2007-01-09 Kaz Kojima <kkojima@gcc.gnu.org>
2909 * config/sh/t-linux (TARGET_LIBGCC2_CFLAGS): Delete.
2910 (SHLIB_MAPFILES, SHLIB_LINK, SHLIB_INSTALL): Likewise.
2912 2007-01-09 Nicolas Pitre <nico@cam.org>
2915 * arm/ieee754-df.S (Lad_s): Also test the low word of X for zero.
2917 2007-01-08 Geoffrey Keating <geoffk@apple.com>
2919 * target.h (struct gcc_target): New field library_rtti_comdat.
2920 * target-def.h (TARGET_CXX_LIBRARY_RTTI_COMDAT): New.
2921 (TARGET_CXX): Add TARGET_CXX_LIBRARY_RTTI_COMDAT.
2922 * doc/tm.texi (C++ ABI): Document TARGET_CXX_LIBRARY_RTTI_COMDAT.
2923 * config/darwin.h (TARGET_CXX_LIBRARY_RTTI_COMDAT): Define.
2925 2007-01-08 Geoffrey Keating <geoffk@apple.com>
2927 * doc/invoke.texi (Optimize Options): Correct description of -O0.
2929 2007-01-08 Richard Guenther <rguenther@suse.de>
2931 * tree.h (force_fit_type_double): Export.
2932 (force_fit_type): Remove.
2933 * fold-const.c (force_fit_type_double): New function.
2934 (force_fit_type): Remove.
2935 (int_const_binop): Use it.
2936 (fold_convert_const_int_from_int): Likewise.
2937 (fold_convert_const_int_from_real): Likewise.
2938 (fold_div_compare): Likewise.
2939 (fold_sign_changed_comparison): Likewise.
2940 (fold_unary): Likewise.
2941 (fold_negate_const): Likewise.
2942 (fold_abs_const): Likewise.
2943 (fold_not_const): Likewise.
2944 * c-common.c (shorten_compare): Use force_fit_type_double.
2945 * convert.c (convert_to_pointer): Likewise.
2947 2007-01-08 Richard Guenther <rguenther@suse.de>
2949 * tree.h (build_int_cst_wide_type): Export.
2950 * tree.c (build_int_cst_wide_type): New function.
2951 (build_int_cst_wide): Fix comment.
2952 * builtins.c (fold_builtin_object_size): Use build_int_cst
2953 to build -1 or 0 of the correct type. Use fit_double_type
2954 to check for overflow.
2955 * fold-const.c (optimize_bit_field_compare): Use build_int_cst_type
2957 (decode_field_reference): Likewise.
2958 (all_ones_mask_p): Likewise.
2959 (native_interpret_int): Use build_int_cst_wide_type.
2960 (fold_binary): Use build_int_cst_type to build an all-ones
2962 * stor-layout.c (set_sizetype): Use build_int_cst_wide_type.
2964 2007-01-08 Daniel Jacobowitz <dan@codesourcery.com>
2966 * config/pa/t-pa64 (libgcc_stub.a): Use $(T).
2968 2007-01-09 Ben Elliston <bje@au.ibm.com>
2970 * genautomata.c (STATS_OPTION): New option.
2971 (stats_flag): New flag.
2972 (gen_automata_option): Handle it.
2973 (initiate_automaton_gen): Ditto.
2974 (write_automata): Output statistics only if stats_flag is
2975 set. Likewise, output time statistics only if time_flag is set.
2976 * doc/md.texi (Processor pipeline description): Document new flag.
2978 2007-01-08 Richard Guenther <rguenther@suse.de>
2980 * builtins.c (fold_builtin_int_roundingfn): Use fit_double_type.
2981 * tree.c (build_int_cst_type): Likewise.
2982 (size_in_bytes): Don't call force_fit_type on the result.
2983 (int_fits_type_p): Use fit_double_type.
2984 * fold-const.c (fit_double_type): New function.
2985 (force_fit_type): Use it.
2986 * tree.h (fit_double_type): Export.
2988 2007-01-08 Jan Hubicka <jh@suse.cz>
2990 * tree-vectorizer.c (gate_increase_alignment): Fix return type.
2991 * ipa.c (function_and_variable_visibility): Fix return type.
2993 2007-01-08 Richard Guenther <rguenther@suse.de>
2995 * tree-ssa-ccp.c (maybe_fold_offset_to_array_ref): Use type
2996 of offset to build the index.
2997 * tree-pretty-print.c (dump_generic_node): Don't build negated
2998 const just for printing.
2999 * c-pretty-print.c (pp_c_integer_constant): Likewise.
3000 * builtins.c (fold_builtin_int_roundingfn): Check if result
3001 fits the type by using force_fit_type and comparing the result.
3002 * predict.c (predict_loops): Use compare_tree_int for comparison.
3003 * tree.c (build_int_cst): Fall back to integer_type_node for
3005 (build_int_cst_wide): Assert type is non-null.
3007 2007-01-08 Roberto Costa <roberto.costa@st.com>
3009 * tree-vrp.c (extract_range_from_cond_expr): New.
3010 (extract_range_from_expr): Handle COND_EXPR nodes used as expressions.
3011 * tree-ssa-ccp.c (get_maxval_strlen): Handle COND_EXPR nodes used
3013 (fold_stmt): Bug fix, avoid infinite recursion when folding COND_EXPRs.
3014 * tree-ssa-forwprop.c (simplify_cond, forward_propagate_into_cond,
3015 tree_ssa_forward_propagate_single_use_vars): Handle COND_EXPR nodes
3016 used as expressions.
3017 * tree-object-size.c (cond_expr_object_size): New.
3018 (collect_object_sizes_for): Handle COND_EXPR nodes used as expressions.
3020 2007-01-08 Jan Hubicka <jh@suse.cz>
3022 * tree-ssa-forwprop.c (forward_propagate_into_cond,
3023 tree_ssa_forward_propagate_single_use_va): Release defs of propagated
3026 2007-01-08 Richard Guenther <rguenther@suse.de>
3028 PR tree-optimization/23603
3029 * tree-vrp.c (set_value_range_to_truthvalue): New function.
3030 (extract_range_from_binary): Fall back to truthvalue instead of
3031 varying for TRUTH_*_EXPR.
3032 (extract_range_from_comparison): Fall back to truthvalue instead of
3034 (vrp_visit_phi_node): Don't adjust new range bounds to +INF/-INF
3035 if all visited PHI values were constant.
3037 2007-01-08 Jan Hubicka <jh@suse.cz>
3039 * cgraphunit.c (cgraph_process_new_functions): Reset reachable flag.
3040 (cgraph_analyze_function): break out from ...
3041 (cgraph_finalize_compilation_unit): ... here.
3042 (cgraph_expand_function): Remove forgoten commented out line.
3043 (cgraph_optimize): Analyze functions.
3045 2007-01-08 Jan Hubicka <jh@suse.cz>
3047 * tree-pas.h (TODO_remove_function): New flag.
3048 (TODO_update*): Renumber.
3049 (pass_ipa_increase_alignment,
3050 pass_ipa_function_and_variable_visibility): New passes.
3051 * cgraphunit.c (cgraph_increase_alignment): Move to tree-vectorizer.c
3052 (cgraph_function_and_variable_visibility): Move to ipa.c
3053 (cgraph_optimize): Don't call cgraph_function_and_variable_visibility,
3054 cgraph_increase_alignment.
3055 * ipa-inline.c (cgraph_decide_inlining): Don't push timevar.
3056 (cgraph_decide_inlining_incrementally): Push TV_INTEGRATION before
3057 calling tree-inline.
3058 (cgraph_early_inlining): Do not call cgraph_remove_unreachable_nodes.
3059 (pass_ipa_inline, pass_early_ipa_inlining): Set TODO_remove_functions
3060 * tree-vectorizer.c (increase_alignment): Move here from cgraphunit.c
3061 (gate_increase_alignment): New function.
3062 (pass_ipa_increase_alignment): New pass.
3063 * ipa.c: Inline tree-pass.h and timevar.h
3064 (function_and_variable_visibility): Move here from cgraphunit.c
3065 * tree-optimize.c (pass_early_local_passes): Add TODO_remove_functions.
3066 * passes.c (init_optimization_passes): Add the two new passes.
3067 (execute_todo): Handle cgraph_remove_functions.
3069 2007-01-08 Nick Clifton <nickc@redhat.com>
3071 * config/frv/predicates.md (reg_or_0_operand): Accept
3074 2007-01-08 Ralf Corsépius <ralf.corsepius@rtems.org>
3076 * config/bfin/rtems.h, config/bfin/t-rtems: New.
3077 * config.gcc: Add bfin*-rtems*.
3079 2007-01-08 Mark Shinwell <shinwell@codesourcery.com>
3081 * c.opt: Add -flax-vector-conversions.
3082 * c-typeck.c (convert_for_assignment): Pass flag to
3083 vector_types_convertible_p to allow emission of note.
3084 (digest_init): Likewise.
3085 * c-opts.c: Handle -flax-vector-conversions.
3086 * c-common.c (flag_lax_vector_conversions): New.
3087 (vector_types_convertible_p): Unless -flax-vector conversions
3088 has been passed, disallow conversions between vectors with
3089 differing numbers of subparts and/or element types. If such
3090 a conversion is disallowed, possibly emit a note on the first
3091 occasion only to inform the user of -flax-vector-conversions.
3092 The new last argument specifies this.
3093 * c-common.h (flag_lax_vector_conversions): New.
3094 (vector_types_convertible_p): Add extra argument.
3095 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use
3096 char_type_node for V*QI type vectors.
3097 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
3098 Update to satisfy new typechecking rules.
3099 * config/rs6000/altivec.h (vec_cmple): Use vec_cmpge, for both
3101 * doc/invoke.texi (C Dialect Options): Document
3102 -flax-vector-conversions.
3104 2007-01-08 Mark Shinwell <shinwell@codesourcery.com>
3106 PR tree-optimization/29877
3107 * tree-ssa-ter.c (is_replaceable_p): Deem assignments with
3108 a register variable on the RHS to not be replaceable.
3110 2007-01-08 Chen Liqin <liqin@sunnorth.com.cn>
3111 * config/score/t-score-elf (MULTILIB_OPTIONS): Change.
3112 * config/score/predicates.md (const_uimm5, sr0_operand, const_simm12,
3113 const_simm15, const_pow2, const_npow2): Added.
3114 * config/score/misc.md (insv, extv, extzv, movmemsi,
3115 move_lbu_a/b, mov_lhu_a/b etc): Added and fix some bug.
3116 * config/score/score.c (score_address_cost, score_select_cc_mode):
3118 Change CONST_OK_FOR_LETTER_P/EXTRA_CONSTRAINT define.
3119 Update score_rtx_costs for MACRO TARGET_RTX_COSTS.
3120 Update score_print_operand.
3121 * config/score/score.h (DATA_ALIGNMENT, SELECT_CC_MODE): Added.
3122 Adjust register allocate order and update some macro define.
3123 * config/score/score-mdaux.c (mdx_unaligned_load, mdx_unsigned_store,
3124 mdx_block_move_straight, mdx_block_move_loop_head,
3125 mdx_block_move_loop_body, mdx_block_move_loop_foot, mdx_block_move_loop,
3126 mdx_block_move): Added.
3127 (mdx_movsicc, mdp_select_add_imm, mdp_select, mds_zero_extract_andi,
3128 mdp_limm): Updated and fix some bug and typo.
3129 * config/score/score.md (movqi/hi/si, add/sub/zero/ext): Updated.
3130 (movsf, movdf, doloop_end): Added.
3132 2007-01-08 Kazu Hirata <kazu@codesourcery.com>
3134 * config/arm/arm.c, config/arm/arm.h, config/arm/arm.md,
3135 config/arm/thumb2.md: Fix comment typos.
3136 * doc/extend.texi: Fix a typo.
3138 2007-01-07 Eric Christopher <echristo@apple.com>
3140 * configure.ac: Check for __stack_chk_fail for darwin.
3141 * configure: Regenerate.
3143 2007-01-07 Richard Guenther <rguenther@suse.de>
3145 * tree-vrp.c (extract_range_from_assert): CSE calls to
3146 compare_values where possible.
3147 (extract_range_from_unary_expr): Likewise.
3149 2007-01-07 Anatoly Sokolov <aesok@post.ru>
3151 * config/avr/avr-protos.h (call_insn_operand): Delete prototype.
3152 * config/avr/avr.c (call_insn_operand): Delete function.
3153 * config/avr/avr.md (*pushqi, *pushhi, *pushsi, *pushsf): Use REG_SP
3154 instead of register number. Use predicates.
3155 * config/avr/predicates.md (const0_operand, reg_or_0_operand,
3156 call_insn_operand): Add.
3158 2007-01-06 Jan Hubicka <jh@suse.cz>
3160 * tree-pass.h (pass_build_cgraph_edges): Declare.
3161 * cgraphunit.c (record_refernece): Move to cgraphbuild.c
3162 (visited_nodes): Remove.
3163 (cgraph_create_edges): Move to cgraphbuild.c; rename to
3164 build_cgrpah_edges; make visited_nodes local.
3165 (cgraph_process_new_functions): DO not call initialize_inline_failed.
3166 (record_references_in_initializer): Move to cgraphbuild.c
3167 (initialize_inline_failed, rebuild_cgraph_edges,
3168 pass_rebuild_cgraph_edges): Move to cgraphbuild.c.
3169 (verify_cgraph_node): Make visited_nodes local.
3170 (cgraph_analyze_function): Do not call cgraph_create_edges and
3171 initialize_inline_failed.
3172 (cgraph_expand_function): Do not call cgraph_lower_function;
3173 assert that function is already lowered.
3174 * Makefile.in (cgraphbuild.o): New.
3175 * passes.c (init_optimization_passes): Add pass_build_cgraph_edges
3176 at the end of lowering passes.
3178 2007-01-06 Steven Bosscher <steven@gcc.gnu.org>
3180 * ifcvt.c (cond_move_convert_if_block): New function, code
3181 factored out from...
3182 (cond_move_process_if_block): ...here. Call the new function
3183 on the THEN and ELSE blocks.
3184 (merge_if_block): Do not copy global_live_at_end, merge_blocks
3185 already takes care of this.
3187 2007-01-05 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
3190 * tree.h (TREE_OVERFLOW_P): New.
3191 * c-typeck.c (parser_build_unary_op): Warn only if result
3192 overflowed and operands did not.
3193 (parser_build_binary_op): Likewise.
3194 (convert_for_assignment): Remove redundant overflow_warning.
3195 * c-common.c (overflow_warning): Don't check or set TREE_OVERFLOW.
3197 2007-01-05 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
3199 * c-typeck.c (store_init_value): Split over two lines to follow
3200 the GNU coding style.
3202 2007-01-05 Benjamin Kosnik <bkoz@redhat.com>
3204 * c-cppbuiltin.c (c_cpp_builtins): __GXX_EXPERIMENTAL_CPP0X__ to
3205 __GXX_EXPERIMENTAL_CXX0X__.
3206 * doc/cpp.texi: Same.
3208 2007-01-05 Richard Guenther <rguenther@suse.de>
3211 * tree.c (get_narrower): Do not construct COMPONENT_REFs
3212 with mismatched types. Instead explicitly build a
3213 conversion NOP_EXPR.
3215 2007-01-05 Ian Lance Taylor <iant@google.com>
3217 * c-common.c (decl_with_nonnull_addr_p): New function.
3218 (c_common_truthvalue_conversion): Call it.
3219 * c-typeck.c (build_binary_op): Likewise.
3220 * c-common.h (decl_with_nonnull_addr_p): Declare.
3222 2007-01-05 Jakub Jelinek <jakub@redhat.com>
3225 * libgcc2.c (__divdc3): Compare c and d against 0.0 instead of
3228 2007-01-05 Joel Brobecker <brobecker@adacore.com>
3230 * doc/install.texi (Final install): Document the fact that
3231 the GNAT runtime should not be stripped.
3233 2007-01-04 Jan Hubicka <jh@suse.cz>
3235 * tree-inline.c (fold_marked_statements): Update operand caches
3236 and EH after folding
3238 2007-01-04 Ian Lance Taylor <iant@google.com>
3240 * c-common.c (check_function_nonnull): Whitespace fix.
3242 2007-01-04 Jan Hubicka <jh@suse.cz>
3244 * tree-optimize.c (execute_fixup_cfg): Correct previously mistakely
3245 comitted older version of patch.
3246 (pass_fixup_cfg): Add TODOs to verify flow and statements, dump
3247 function, celanup cfg and collect garbage.
3249 2007-01-04 Mike Stump <mrs@apple.com>
3251 * Makefile.in (mostlyclean): Don't remove libgcc anymore.
3254 2007-01-04 Eric Christopher <echristo@apple.com>
3256 * libgcc2.c (__bswapsi2): Use SItype.
3257 (__bswapdi2): Use DItype.
3258 * libgcc2.h: Update for above.
3260 2007-01-04 Paul Brook <paul@codesourcery.com>
3262 * config/arm/arm.md (arm_mulsi3, thumb_mulsi3, mulsi3_compare0,
3263 mulsi_compare0_scratch, mulsi3addsi, mulsi3addsi_compare0,
3264 mulsi3addsi_compare0_scratch, mulsidi3adddi, mulsidi3,
3265 umulsidi3, umulsidi3adddi, smulsi3_highpart,
3266 umulsi3_highpart): Make conditional on !arm_arch6.
3267 (arm_mulsi3_v6, thumb_mulsi3_v6, mulsi3_compare0_v6,
3268 mulsi_compare0_scratch_v6, mulsi3addsi_v6, mulsi3addsi_compare0_v6,
3269 mulsi3addsi_compare0_scratch_v6, mulsidi3adddi_v6, mulsidi3_v6,
3270 umulsidi3_v6, umulsidi3adddi_v6, smulsi3_highpart_v6,
3271 umulsi3_highpart_v6): New insns.
3273 2007-01-04 Roger Sayle <roger@eyesopen.com>
3275 * fold-const.c (fold_convert): When casting an expression to void,
3276 fold_ignored_result may discover a GIMPLE_MODIFY_STMT which doesn't
3277 have a type. Instead of attempting to build a NOP_EXPR, return
3278 these "special" trees directly.
3280 2007-01-04 Joseph Myers <joseph@codesourcery.com>
3282 * config/rs6000/rs6000.c (rs6000_rtx_costs): Make adjustment for
3283 MULT inside MINUS as either argument. Use rs6000_cost->dmul -
3284 rs6000_cost->fp not 0 as adjustment for outer NEG.
3286 2007-01-04 Jan Hubicka <jh@suse.cz>
3288 * cgraph.c (cgraph_release_function_body): New function.
3289 (cgraph_remove_node): Use it.
3290 * cgraph.h (cgraph_release_function_body): Declare.
3291 * cgraphunit.c (cgraph_expand_function): Use it.
3292 * ipa.c (cgraph_remove_unreahchable_nodes): Use it.
3293 * tree-ssa.c (delete_tree_ssa): Allow to be called before aliasing
3294 is initialized and while compilation of other function is running.
3295 * tree-optimize.c (execute_free_cfg_annotations): Move code to clear
3296 statement CFG annotations from here to ...
3297 * tree-cfg.c (delete_tree_cfg_annotations): ... here.
3299 2007-01-04 Zdenek Dvorak <dvorakz@suse.cz>
3301 * cfgloop.h (enum li_flags): Make the constants powers of two.
3303 2007-01-04 Jan Hubicka <jh@suse.cz>
3305 * tree-inline.c (copy_bb): Insert new statements to statements_to_fold
3307 (fold_marked_statements): New function.
3308 (optimize_inline_calls, tree_function_versioning): Fold new statements.
3309 * tree-inline.h (copy_body_data): Add statements_to_fold.
3311 2007-01-03 Daniel Jacobowitz <dan@codesourcery.com>
3313 * config.gcc: Mention libgcc/config.host.
3314 * Makefile.in: Update comments mentioning libgcc.
3315 (LIBGCC, INSTALL_LIBGCC, GCC_PARTS, mklibgcc): Delete.
3316 (all.cross, start.encap, rest.encap, rest.cross): Update
3317 dependencies for libgcc move.
3318 (libgcc.mk, LIBGCC_DEPS, libgcov.a, libgcc.a, stmp-multilib)
3319 (clean-target, clean-target-libgcc): Delete.
3320 (srcdirify, GCC_EXTRA_PARTS): New macros.
3321 (libgcc-support, libgcc.mvars): New rules.
3322 (distclean): Remove mention of mklibgcc.
3323 (install): Don't reference INSTALL_LIBGCC.
3324 (install-common): Don't reference EXTRA_PARTS.
3325 (install-libgcc, install-multilib): Delete rules.
3326 * mklibgcc.in: Delete file.
3327 * doc/configfiles.texi: Don't mention mklibgcc.
3329 * config/i386/t-darwin (SHLIB_VERPFX): Delete (moved to libgcc).
3330 * config/i386/t-darwin64 (SHLIB_VERPFX): Likewise.
3331 * config/rs6000/t-darwin (SHLIB_VERPFX): Likewise.
3332 * config/rs6000/t-ppccomm (TARGET_LIBGCC2_CFLAGS, SHLIB_MAPFILES)
3333 (mklibgcc, ldblspecs): Likewise.
3335 * config/i386/t-nwld (libgcc.def, libc.def, libpcre.def)
3336 (posixpre.def): Use $(T).
3337 (SHLIB_EXT, SHLIB_NAME, SHLIB_SLIBDIR_QUAL, SHLIB_DEF, SHLIB_MAP)
3338 (SHLIB_SRC, SHLIB_INSTALL): Delete.
3339 (SHLIB_LINK): Make dummy.
3340 * config/t-slibgcc-darwin: Delete contents except for dummy SHLIB_LINK.
3342 * config/frv/t-linux (EXTRA_MULTILIB_PARTS): Clear.
3344 * config/alpha/t-crtfm: Use $(T) in rules for EXTRA_PARTS.
3345 * config/alpha/t-vms, config/alpha/t-vms64, config/fr30/t-fr30,
3346 config/i386/t-rtems-i386, config/ia64/t-ia64, config/rs6000/t-beos,
3347 config/rs6000/t-newas, config/sparc/t-elf: Likewise.
3349 * configure.ac (all_outputs): Remove mklibgcc.
3350 * configure: Regenerated.
3352 2007-01-03 Josh Conner <jconner@apple.com>
3355 * calls.c (compute_argument_addresses): Set stack and stack_slot
3356 for partial args, too.
3357 (store_one_arg): Use locate.size.constant for the size when
3358 generating a save_area.
3360 2007-01-03 Robert Kennedy <jimbob@google.com>
3362 * tree-cfg.c (tree_merge_blocks): Release SSA_NAME phi results
3363 whose definitions are deleted due to basic block merging.
3365 2007-01-03 Paul Brook <paul@codesourcery.com>
3368 * config/arm/arm.c (output_return_instruction): Pop PC in interrupt
3370 (use_return_insn): Return 0 for Thumb interrupt functions.
3371 (print_multi_reg): Add rfe argument for IRQ returns.
3372 (arm_output_epilogue): Pop interrupt return address directly into PC.
3373 (arm_expand_prologue): Only adjust IRQ return address in Arm mode.
3375 2007-01-03 Paul Brook <paul@codesourcery.com>
3377 Merge from sourcerygxx-4_1.
3378 * config/arm/thumb2.md: New file.
3379 * config/arm/elf.h (JUMP_TABLES_IN_TEXT_SECTION): Return True for
3381 * config/arm/coff.h (JUMP_TABLES_IN_TEXT_SECTION): Ditto.
3382 * config/arm/aout.h (ASM_OUTPUT_ADDR_VEC_ELT): Add !Thumb-2 assertion.
3383 (ASM_OUTPUT_ADDR_DIFF_ELT): Output Thumb-2 jump tables.
3384 * config/arm/aof.h (ASM_OUTPUT_ADDR_DIFF_ELT): Output Thumb-2 jump
3386 (ASM_OUTPUT_ADDR_VEC_ELT): Add !Thumb-2 assertion.