| blob | fbf14560fd3dcdb0fcdf08a64e2e07a7b3f59b32 |
1 2007-02-03 Uros Bizjak <ubizjak@gmail.com>
3 PR middle-end/30667
4 * combine.c (try_combine): Do not substitute source operand
5 with constants wider than 2 * HOST_BITS_PER_WIDE_INT.
7 2007-02-03 Jan Hubicka <jh@suse.cz>
9 PR gcov-profile/30650
10 * value-prof.c (stringop_block_profile): Fix handling of size counter;
11 do not divide by zero for never executed counters.
12 (tree_find_values_to_profile): Fix counters.
13 * gcov-ui.h (GCOV_COUNTER_AVERAGE, GCOV_COUNTER_IOR): Fix comments.
15 2007-02-03 Ian Lance Taylor <iant@google.com>
17 * lower-subreg.c (simple_move_operand): New static function,
18 broken out of simple_move. Reject LABEL_REF, SYMBOL_REF, and HIGH
19 operands.
20 (simple_move): Call simple_move_operand.
21 (find_decomposable_subregs): Add special handling of MEMs.
22 (can_decompose_p): Rename from cannot_decompose_p. Reverse
23 meaning of return value. If we see a hard register, test whether
24 it can store a word_mode value. Change all callers.
26 2007-02-03 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
28 * pa.md (addvdi3, addvsi3, subvdi3, subvsi3, negvdi2, negvsi2): New
29 ftrapv insns and expanders.
30 (subdi3): Change define_expand operand 1 to arith11_operand, and
31 operand 2 to reg_or_0_operand. Change constraints of 64-bit insn
32 pattern to handle reg_or_0 operands. Revise 32-bit insn pattern to
33 handle 11-bit constants and reg_or_0 operands in operands 1 and 2,
34 respectively.
36 PR middle-end/30174
37 * varasm.c (notice_global_symbol): Treat global objects as weak when
38 flag_shlib is true.
40 2007-02-03 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
42 * emit-rtl.c (dconstpi): Delete.
43 (dconstsqrt2): New.
44 (init_emit_once): Delete dconstpi and init dconstsqrt2.
45 * real.h (dconstpi): Delete.
46 (dconstsqrt2): New.
47 * builtins.c (fold_builtin_cabs): Use dconstsqrt2.
48 (fold_builtin_hypot): Likewise.
50 2007-02-03 Tom Tromey <tromey@redhat.com>
52 PR driver/30246
53 * gcc.c (cpp_unique_options): Any of -ggdb3, -gstabs3,
54 -gcoff3, -gxcoff3, -gvms3 implies -dD.
56 2007-02-03 Kazu Hirata <kazu@codesourcery.com>
58 * c-decl.c, config/avr/avr.c, config/avr/avr.h,
59 config/m68k/m68k.c, config/m68k/netbsd-elf.h,
60 config/mn10300/mn10300.c, config/pdp11/pdp11.h,
61 config/rs6000/cell.md, config/rs6000/darwin.h,
62 config/sh/sh.md, config/sh/sh4-300.md, config/spu/spu.c,
63 config/spu/spu.md, cselib.c, expr.c, haifa-sched.c, hwint.h,
64 jump.c, reload.c, sched-deps.c, sched-int.h, tree-inline.c,
65 tree-profile.c, tree-ssa-live.h, tree-vrp.c: Fix comment
66 typos. Follow spelling conventions.
67 * doc/invoke.texi: Follow spelling conventions.
69 2007-02-03 Roger Sayle <roger@eyesopen.com>
71 * simplify-rtx.c (simplify_relational_operation_1): Implement some
72 canonicalization transformations that attempt to simplify integer
73 constant comparisons to become comparisons against zero.
75 2007-02-03 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
77 * builtins.c (fold_builtin_cabs): Fold cabs(x+xi) into
78 fabs(x)*sqrt(2).
79 * fold-const.c (fold_binary): Fix comment typos. Fold complex
80 (x,0)-(0,y) into (x,-y). Likewise (0,y)-(x,0) into (-x,y).
82 2007-02-02 Mike Stump <mrs@apple.com>
84 * config/darwin9.h (DARWIN_LINKER_GENERATES_ISLANDS): Add.
85 * config/rs6000/rs6000.c (DARWIN_GENERATE_ISLANDS): Add.
86 (output_call): Use DARWIN_GENERATE_ISLANDS to decide when to
87 generate a branch island.
89 2007-02-02 Bob Wilson <bob.wilson@acm.org>
91 * config/xtensa/xtensa.c (smalloffset_mem_p): Use BASE_REG_P.
92 (xtensa_legitimate_address_p): New.
93 (xtensa_legitimize_address): New.
94 (xtensa_output_addr_const_extra): New.
95 * config/xtensa/xtensa.h (REG_OK_STRICT_FLAG): Define.
96 (BASE_REG_P): New.
97 (REG_OK_FOR_BASE_P): Use BASE_REG_P.
98 (GO_IF_LEGITIMATE_ADDRESS): Move code to xtensa_legitimate_address_p.
99 (LEGITIMIZE_ADDRESS): Move code to xtensa_legitimize_address.
100 (OUTPUT_ADDR_CONST_EXTRA): Move code to xtensa_output_addr_const_extra.
101 * config/xtensa/xtensa-protos.h (xtensa_legitimate_address_p): New.
102 (xtensa_legitimize_address): New.
103 (xtensa_output_addr_const_extra): New.
105 2007-02-02 Steve Ellcey <sje@cup.hp.com>
107 * config/ia64/ia64.c (ia64_print_operand): Fix compare strings.
109 2007-02-02 Ian Lance Taylor <iant@google.com>
111 * expmed.c (expand_divmod): Add comment.
113 2007-02-02 Kazu Hirata <kazu@codesourcery.com>
115 * emit-rtl.c (renumber_insns): Remove.
116 * flags.h: Remove the extern for flag_renumber_insns.
117 * rtl.h: Remove the prototype for renumber_insns.
118 * toplev.c (flag_renumber_insns): Remove.
120 2007-02-02 Hui-May Chang <hm.chang@apple.com>
122 Revert for x86 darwin:
123 2005-06-19 Uros Bizjak <uros@kss-loka.si>
125 * config/i386/i386.c (ix86_function_arg_regno_p): Put back the
126 code before the following patch under TARGET_MACHO.
127 (ix86_function_value_regno_p): Likewise.
129 2007-02-02 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
131 * fold-const.c (negate_expr_p, fold_negate_expr): Handle
132 COMPLEX_EXPR.
134 2007-02-02 Maxim Kuvyrkov <mkuvyrkov@ispras.ru>
136 PR target/29682
137 * config/ia64/ia64.c (ia64_speculate_insn): Restrict to memory loads to
138 general or fp registers. Add comments.
139 * config/ia64/ia64.md (reg_pred_prefix): Add comment.
141 2007-02-02 Paolo Bonzini <bonzini@gnu.org>
143 * pointer-set.c (insert_aux): Only return insertion slot.
144 (pointer_set_insert): Adjust.
145 (pointer_set_traverse, struct pointer_map_t, pointer_map_create,
146 pointer_map_destroy, pointer_map_insert, pointer_map_contains,
147 pointer_map_traverse): New.
148 * pointer-set.h (pointer_set_traverse, struct pointer_map_t,
149 pointer_map_create, pointer_map_destroy, pointer_map_insert,
150 pointer_map_contains, pointer_map_traverse): Declare.
152 2007-02-02 Jakub Jelinek <jakub@redhat.com>
154 PR middle-end/30473
155 * builtins.c (fold_builtin_sprintf): Do not attempt to optimize
156 sprintf (str, "%s"). Do not optimize sprintf (str, "nopercent", p++).
158 2007-02-02 Maxim Kuvyrkov <mkuvyrkov@ispras.ru>
160 * sched-int.h (ds_to_dk, dk_to_ds): Declare functions.
162 (struct _dep): New type.
163 (dep_t): New typedef.
164 (DEP_PRO, DEP_CON, DEP_KIND): New access macros.
165 (DEP_STATUS): New access macro. The macro with the same name was
166 renamed to DEP_LINK_STATUS.
167 (dep_init): Declare function
169 (struct _dep_link): New type.
170 (dep_link_t): New typedef.
171 (DEP_LINK_NODE, DEP_LINK_NEXT, DEP_LINK_PREV_NEXTP): New access macros.
172 (DEP_LINK_DEP, DEP_LINK_PRO, DEP_LINK_CON, DEP_LINK_KIND): New macros.
173 (DEP_LINK_STATUS): New macro.
174 (debug_dep_links): New debug function.
176 (struct _deps_list): New type.
177 (deps_list_t): New typedef.
178 (DEPS_LIST_FIRST): New access macro.
179 (FOR_EACH_DEP_LINK): New cycle macro.
180 (create_deps_list, free_deps_list, delete_deps_list): Declare
181 functions.
182 (deps_list_empty_p, debug_deps_list, add_back_dep_to_deps_list): Ditto.
183 (find_link_by_pro_in_deps_list, find_link_by_con_in_deps_list): Ditto.
184 (copy_deps_list_change_con): Ditto.
186 (move_dep_link): Declare function.
188 (struct _dep_node): New type.
189 (dep_node_t): New typedef.
190 (DEP_NODE_BACK, DEP_NODE_DEP, DEP_NODE_FORW): New access macros.
192 (struct haifa_insn_data.back_deps): New field to hold backward
193 dependencies of the insn.
194 (struct haifa_insn_data.depend): Rename to forw_deps. Change its type
195 to deps_list_t.
196 (struct haifa_insn_data.resolved_deps): Rename to resolved_back_deps.
197 Change its type to deps_list_t.
198 (INSN_BACK_DEPS): New access macro to use instead of LOG_LINKS.
199 (INSN_DEPEND): Rename to INSN_FORW_DEPS.
200 (RESOLVED_DEPS): Rename to INSN_RESOLVED_BACK_DEPS.
202 (INSN_COST): Move to haifa-sched.c. Use insn_cost () instead.
204 (DEP_STATUS): Rename to DEP_LINK_STATUS. Fix typo in the comment.
206 (add_forw_dep, delete_back_forw_dep, insn_cost): Update declaration and
207 all callers.
208 (dep_cost): Declare.
210 * sched-deps.c (CHECK): New macro to (en/dis)able sanity checks.
211 (ds_to_dk, dk_to_ds): New functions.
213 (init_dep_1): New static function.
214 (init_dep): New function.
215 (copy_dep): New static function.
217 (dep_link_consistent_p, attach_dep_link, add_to_deps_list): New static
218 functions.
219 (detach_dep_link): New static function.
220 (move_dep_link): New function.
222 (dep_links_consistent_p, dump_dep_links): New static functions.
223 (debug_dep_links): New debugging function.
225 (deps_obstack, dl_obstack, dn_obstack): New static variables.
227 (alloc_deps_list, init_deps_list): New static functions.
228 (create_deps_list): New function.
229 (clear_deps_list): New static function.
230 (free_deps_list, delete_deps_list, deps_list_empty_p): New functions.
231 (deps_list_consistent_p, dump_deps_list): New static functions.
232 (debug_deps_list): New function.
233 (add_back_dep_to_deps_list, find_link_by_pro_in_deps_list): New
234 functions.
235 (find_link_by_con_in_deps_list, copy_deps_list_change_con): Ditto.
237 (maybe_add_or_update_back_dep_1, add_or_update_back_dep_1): Update to
238 use new scheduler dependencies lists.
239 (add_back_dep, delete_all_dependences, fixup_sched_groups): Ditto.
240 (sched_analyze): Ditto. Initialize dependencies lists.
241 (add_forw_dep, compute_forward_dependences): Update to use new
242 scheduler dependencies lists.
244 (init_dependency_caches): Init deps_obstack.
245 (free_dependency_caches): Free deps_obstack.
247 (adjust_add_sorted_back_dep, adjust_back_add_forw_dep): Update to use
248 new scheduler dependencies lists.
249 (delete_forw_dep, add_or_update_back_forw_dep): Ditto.
250 (add_back_forw_dep, delete_back_forw_dep): Ditto.
252 * sched-rgn.c (set_spec_fed, find_conditional_protection, is_pfree):
253 Update to use new scheduler dependencies lists.
254 (is_conditionally_protected, is_prisky, add_branch_dependences): Ditto.
255 (debug_dependencies): Ditto.
256 (schedule_region): Update comments.
258 * sched-ebb.c (earliest_block_with_similiar_load): Update to use new
259 scheduler dependencies lists.
260 (schedule_ebb): Update comments.
262 * rtl.def (DEPS_LIST): Remove.
264 * lists.c (unused_deps_list): Remove.
265 (free_list): Update assertions.
267 (alloc_DEPS_LIST, free_DEPS_LIST_list, free_DEPS_LIST_node): Remove.
268 (remove_free_DEPS_LIST_elem, copy_DEPS_LIST_list): Ditto.
270 * rtl.h (free_DEPS_LIST_list, alloc_DEPS_LIST): Remove declarations.
271 (remove_free_DEPS_LIST_elem, copy_DEPS_LIST_list): Ditto.
273 * haifa-sched.c (comments): Update.
274 (insn_cost1): Remove. Inline the code into insn_cost ().
275 (insn_cost): Update to use new scheduler dependencies lists. Move
276 processing of the dependency cost to dep_cost ().
277 (dep_cost): New function. Use it instead of insn_cost () when
278 evaluating cost of the dependency. Use compatible interface to
279 interact with the target.
280 (priority): Update to use new scheduler dependencies lists.
281 (rank_for_schedule): Ditto. Optimize heuristic that prefers the insn
282 with greater number of insns that depend on the insn.
283 (schedule_insn): Update to use new scheduler dependencies lists. Add
284 code to free backward dependencies lists. Inline and optimize code
285 from resolve_dep () - see PR28071.
286 (ok_for_early_queue_removal): Update to use new scheduler dependencies
287 lists. Update call to targetm.sched.is_costly_dependence hook.
289 (fix_inter_tick, try_ready, fix_tick_ready): Update to use new
290 scheduler dependencies lists.
292 (resolve_dep): Remove. Move the logic to schedule_insn ().
293 (init_h_i_d): Initialize dependencies lists.
295 (process_insn_depend_be_in_spec): Rename to
296 process_insn_forw_deps_be_in_spec. Update to use new scheduler
297 dependencies lists.
298 (add_to_speculative_block, create_check_block_twin, fix_recovery_deps):
299 Update to use new scheduler dependencies lists.
300 (clear_priorities, calc_priorities, add_jump_dependencies): Ditto.
302 * ddg.c (create_ddg_dependence, create_ddg_dep_no_link): Update to use
303 new scheduler dependencies lists.
304 (build_intra_loop_deps): Ditto.
306 * target.h (struct _dep): Declare to use in
307 gcc_target.sched.is_costly_dependence.
308 (struct gcc_target.sched.adjust_cost): Fix typo.
309 (struct gcc_target.sched.is_costly_dependence): Change signature to use
310 single dep_t parameter instead of an equivalent triad.
311 (struct gcc_target.sched.adjust_cost_2): Remove.
313 * target-def.h (TARGET_SCHED_ADJUST_COST_2): Remove.
315 * reg-notes.def (DEP_TRUE, DEP_OUTPUT, DEP_ANTI): Update comments.
317 * doc/tm.texi (TARGET_SCHED_IS_COSTLY_DEPENDENCE): Update
318 documentation.
319 (TARGET_SCHED_ADJUST_COST_2): Remove documentation.
321 * doc/rtl.texi (LOG_LINKS): Remove part about instruction scheduler.
322 (REG_DEP_TRUE): Document.
324 * config/ia64/ia64.c (ia64_adjust_cost_2): Rename to ia64_adjust_cost.
325 Change signature to correspond to the targetm.sched.adjust_cost hook.
326 Update use in TARGET_SCHED_ADJUST_COST_2.
327 (TARGET_SCHED_ADJUST_COST_2): Rename to TARGET_SCHED_ADJUST_COST.
328 (ia64_dependencies_evaluation_hook, ia64_dfa_new_cycle): Update to use
329 new scheduler dependencies lists.
330 (ia64_gen_check): Ditto.
332 * config/mips/mips.c (vr4130_swap_insns_p): Update to use new scheduler
333 dependencies lists.
335 * config/rs6000/rs6000.c (rs6000_is_costly_dependence): Change
336 signature to correspond to the targetm.sched.is_costly_dependence hook.
337 (is_costly_group): Update to use new scheduler dependencies lists.
339 * config/spu/spu.c (spu_sched_adjust_cost): Use insn_cost () function
340 instead of INSN_COST () macro.
342 2007-02-01 Ian Lance Taylor <iant@google.com>
344 * lower-subreg.c (resolve_clobber): Handle a subreg of a concatn.
346 2007-02-01 Guy Martin <gmsoft@gentoo.org>
348 * pa.md (tp_load): Correct mfctl instruction syntax.
350 2007-02-01 Geoffrey Keating <geoffk@apple.com>
352 * config/rs6000/rs6000.c (rs6000_stack_info): Correct
353 altivec_padding_size calculation on AIX. Improve comment, add
354 assert to verify that it's right.
356 * config/rs6000/darwin.h (MD_UNWIND_SUPPORT): Don't define for
357 64-bit.
359 2007-2-01 Seongbae Park <seongbae.park@gmail.com>
361 PR inline-asm/28686
362 * global.c (global_alloc): Add mising initialization of
363 ELIMINABLE_REGSET.
365 2007-02-01 Roger Sayle <roger@eyesopen.com>
367 * alias.c (init_alias_analysis): Correct whitespace.
368 * bb-reorder.c (fix_edges_for_rarely_executed_code,
369 partition_hot_cold_basic_blocks): Likewise.
370 * builtins.c (expand_builtin_printf, expand_builtin_fprintf,
371 expand_builtin_sprintf, fold_builtin_carg, fold_builtin_sprintf,
372 maybe_emit_sprintf_chk_warning, fold_builtin_sprintf_chk,
373 fold_builtin_snprintf_chk, fold_builtin_printf,
374 fold_builtin_fprintf, do_mpfr_ckconv, do_mpfr_arg1, do_mpfr_arg2,
375 do_mpfr_arg3, do_mpfr_sincos): Likewise.
376 * cfgcleanup.c (cleanup_cfg): Likewise.
377 * cfgexpand.c (tree_expand_cfg): Likewise.
378 * fold-const.c (fold_binary) <RDIV_EXPR>: Likewise.
379 * function.c (get_next_funcdef_no): Likewise.
380 * gengtype.c (main): Likewise.
381 * genmodes.c (main): Likewise.
382 * gcse.c (bypass_conditional_jumps, print_ldst_list): Likewise.
383 * haifa-sched.c (schedule_block, extend_h_i_d): Likewise.
384 * ifcvt.c (noce_emit_move_insn): Likewise.
385 * modulo-sched.c (generate_prolog_epilog, sms_schedule_by_order):
386 Likewise.
387 * stor-layout.c (get_best_mode): Likewise.
388 * tree-ssa-loop-niter.c (get_val_for): Likewise.
389 * tree-ssa-structalias.c (get_varinfo, get_varinfo_fc,
390 scc_visit, do_ds_constraint, do_complex_constraint, label_visit,
391 perform_var_substitution, solve_graph): Likewise.
392 * tree-vrp.c (vrp_finalize): Likewise.
394 2007-02-01 Ian Lance Taylor <iant@google.com>
396 * lower-subreg.c (simplify_gen_subreg_concatn): If we ask for the
397 high part of a paradoxical subreg, return a constant zero.
399 2007-02-01 Zdenek Dvorak <dvorakz@suse.cz>
401 * toplev.c (lang_dependent_init): Call init_set_costs.
402 * loop-init.c (loop_optimizer_init): Do not call init_set_costs.
404 2007-02-01 Richard Guenther <rguenther@suse.de>
406 PR middle-end/30656
407 * fold-const.c (fold_negate_expr): Allow negating a
408 constant if overflow does not change.
410 2007-02-01 Ralf Wildenhues <Ralf.Wildenhues@gmx.de>
412 * doc/c-tree.texi (Expression trees): Improve markup.
413 * doc/tm.texi (Register Classes, Addressing Modes)
414 (Floating Point): Fix spacing after abbreviations. Fix some
415 typos.
417 2007-02-01 Ben Elliston <bje@au.ibm.com>
419 * doc/invoke.texi: Replace "bugfix" with "bug fix" throughout.
420 * doc/contrib.texi: Likewise.
421 * doc/install.texi: Likewise.
423 2007-01-31 Richard Henderson <rth@redhat.com>
424 Ian Lance Taylor <iant@google.com>
426 * lower-subreg.c: New file.
427 * rtl.def (CONCATN): Define.
428 * passes.c (init_optimization_passes): Add pass_lower_subreg and
429 pass_lower_subreg2.
430 * emit-rtl.c (update_reg_offset): New static function, broken out
431 of gen_rtx_REG_offset.
432 (gen_rtx_REG_offset): Call update_reg_offset.
433 (gen_reg_rtx_offset): New function.
434 * regclass.c: Revert patch of 2006-03-05, restoring
435 reg_scan_update.
436 (clear_reg_info_regno): New function.
437 * dwarf2out.c (concatn_loc_descriptor): New static function.
438 (loc_descriptor): Handle CONCATN.
439 * common.opt (fsplit_wide_types): New option.
440 * opts.c (decode_options): Set flag_split_wide_types when
441 optimizing.
442 * timevar.def (TV_LOWER_SUBREG): Define.
443 * rtl.h (gen_reg_rtx_offset): Declare.
444 (reg_scan_update): Declare.
445 * regs.h (clear_reg_info_regno): Declare.
446 * tree-pass.h (pass_lower_subreg): Declare.
447 (pass_lower_subreg2): Declare.
448 * doc/invoke.texi (Option Summary): List -fno-split-wide-types.
449 (Optimize Options): Add -fsplit-wide-types to -O1 list. Document
450 -fsplit-wide-types.
451 * doc/rtl.texi (Regs and Memory): Document concat and concatn.
452 * Makefile.in (OBJS-common): Add lower-subreg.o.
453 (lower-subreg.o): New target.
455 2007-01-31 Kazu Hirata <kazu@codesourcery.com>
457 * config/sh/sh.h (HAVE_SECONDARY_RELOADS): Remove.
459 2007-01-31 Anatoly Sokolov <aesok@post.ru>
461 PR target/19087
462 * config/avr/avr.c (DWARF2_ADDR_SIZE): Define.
464 2007-01-31 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
466 PR middle-end/29335
467 * builtins.c (fold_builtin_sqrt): Use MPFR for constant args.
469 2007-01-31 Zdenek Dvorak <dvorakz@suse.cz>
471 * cfgloop.h: Include vec-prim.h.
472 (enum li_flags): Remove LI_ONLY_OLD.
473 (loop_iterator): Changed.
474 (fel_next, fel_init): Iterate over loop tree.
475 (FOR_EACH_LOOP_BREAK): New macro.
476 * loop-unswitch.c (unswitch_loops): Do not pass LI_ONLY_OLD to
477 FOR_EACH_LOOP.
478 * tree-ssa-loop-unswitch.c (tree_ssa_unswitch_loops): Ditto.
479 * modulo-sched.c (sms_schedule): Ditto.
480 * tree-vectorizer.c (vectorize_loops): Ditto.
481 * doc/loop.texi: Update information on loop numbering and behavior of
482 FOR_EACH_LOOP wrto new loops.
483 * tree-scalar-evolution.c (compute_overall_effect_of_inner_loop,
484 add_to_evolution_1): Test nestedness of loops instead of comparing
485 their numbers.
486 * tree-chrec.c (chrec_fold_plus_poly_poly,
487 chrec_fold_multiply_poly_poly, chrec_evaluate,
488 hide_evolution_in_other_loops_than_loop, chrec_component_in_loop_num,
489 reset_evolution_in_loop): Ditto.
490 * Makefile.in (CFGLOOP_H): Add vecprim.h dependency.
492 2007-01-31 Dirk Mueller <dmueller@suse.de>
494 * c-common.c (warn_about_parentheses): Separate warning about
495 un-parenthized sequence of comparison operators from the one
496 which is supposed to warn about x <= y <= z.
498 2007-01-31 Uros Bizjak <ubizjak@gmail.com>
500 * optabs.h (enum optab_index): Add new OTI_isinf.
501 (isinf_optab): Define corresponding macro.
502 * optabs.c (init_optabs): Initialize isinf_optab.
503 * genopinit.c (optabs): Implement isinf_optab using isinf?f2
504 patterns.
505 * builtins.c (mathfn_built_in): Handle BUILT_IN_ISINF{,F,L}.
506 (expand_builtin_interclass_mathfn): Expand BUILT_IN_ISINF{,F,L}
507 using isinf_optab.
508 (expand_builtin): Expand BUILT_IN_ISINF{,F,L} using
509 expand_builtin_interclass_mathfn.
510 * reg_stack.c (subst_stack_regs_pat): Handle UNSPEC_FXAM.
511 * config/i386/i386.md (UNSPEC_FXAM): New constant.
512 (fxam<mode>2_i387): New insn pattern.
513 (isinf<mode>2) New expander to implement isinf, isinff and isinfl
514 built-in functions as x87 inline asm.
516 2007-01-31 Kazu Hirata <kazu@codesourcery.com>
518 * gcc/config/arm/unwind-arm.h (_sleb128_t, _uleb128_t): New.
520 2007-01-30 Eric Christopher <echristo@apple.com>
522 * config.gcc: Add geode.
524 2007-01-31 Kazu Hirata <kazu@codesourcery.com>
526 * cgraphunit.c, config/arm/arm.c, config/m68k/m68k.c,
527 ipa-inline.c, tree-profile.c, tree-ssa-live.c,
528 tree-ssa-math-opts.c, tree-ssanames.c, tree-vect-analyze.c,
529 value-prof.c: Fix comment typos.
531 2007-01-30 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
533 PR c++/24745
534 * doc/invoke.texi (Wpointer-arith): Document warning.
536 2007-01-30 Janis Johnson
538 * doc/extend.texi (Decimal Floating Types): Remove decfloat.h from
539 the list of discrepancies from the draft TR.
541 2007-01-30 Dirk Mueller <dmueller@suse.de>
543 PR c++/30601
544 * doc/invoke.texi (-Wreturn-type): Update description to
545 match new behavior.
547 2007-01-30 Richard Sandiford <richard@codesourcery.com>
549 * cfgrtl.c (try_redirect_by_replacing_jump): Check only_sets_cc0_p.
551 2007-01-30 Uros Bizjak <ubizjak@gmail.com>
553 * builtins.c (expand_builtin_int_interclass_roundingfn): New function
554 to handle optabs that operate on floating point input argument and
555 output to integer output.
556 (expand_builtin_mathfn) [BUILT_IN_ILOGB]: Move from here ...
557 (expand_builtin_interclass_mathfn) [BUILT_IN_ILOGB]: ... to here.
558 (expand_builtin): Expand BUILT_IN_ILOGB{,F,L} using
559 expand_builtin_interclass_mathfn ().
560 * config/i386/i386.md (fxtractxf3_i387): Rename from *fxtractxf3_i387.
561 (ilogbsi2): Remove.
562 (ilogbxf2, ilogb<mode>2): New expanders to implement ilogb, ilogbf and
563 ilogbl built-in functions as x87 intrinsics.
565 2007-01-30 Richard Guenther <rguenther@suse.de>
567 PR middle-end/27657
568 * dwarf2out.c (reference_to_unused): Query varpool if the
569 variable was output.
571 2007-01-30 Richard Guenther <rguenther@suse.de>
573 PR middle-end/30313
574 * passes.c (execute_one_pass): Reset in_gimple_form to not
575 confuse non-unit-at-a-time mode.
577 2007-01-29 Roger Sayle <roger@eyesopen.com>
578 Richard Guenther <rguenther@suse.de>
580 * fold-const.c (round_up): Make HIGH an unsigned HOST_WIDE_INT to
581 avoid undefined behaviour on overflow. Use force_fit_type_double
582 to construct the constant with the specified TREE_OVERFLOW.
584 2007-01-29 Janis Johnson <janis187@us.ibm.com>
586 * config/dfp-bit.c: Add parameterized support for fp exceptions.
587 * config/dfp-bit.h: Ditto.
589 2007-01-29 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
591 * c-decl.c (pop_scope): Replace warnings with call to
592 warn_for_unused_label.
593 * c-common.h (warn_for_unused_label): Declare.
594 * c-common.c (warn_for_unused_label): Define.
596 2007-01-29 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
598 * tree-optimize.c (update_inlined_to_pointers): Delete unused
599 function.
601 2007-01-29 Janis Johnson <janis187@us.ibm.com>
603 * Makefile.in (USER_H): Remove decfloat.h.
604 * ginclude/decfloat.h: Delete, moving contents to ...
605 * ginclude/float.h: Add support for decimal floating point,
606 guarded by __STDC_WANT_DEC_FP__.
608 2007-01-29 Mike Stump <mrs@apple.com>
610 * doc/gccint.texi (Top): Rename Loop Representation to Loop
611 Analysis and Representation to resolve case insensitive conflict.
612 * doc/loop.texi (Loop Analysis and Representation): Likewise.
614 2007-01-28 Daniel Berlin <dberlin@dberlin.org>
616 * tree.h (struct tree_memory_tag): Add aliases member.
617 (MTAG_ALIASES): New macro.
618 * tree-ssa-alias.c (alias_bitmap_obstack): New variable.
619 (add_may_alias): Remove pointer-set. Update for may_aliases being
620 a bitmap.
621 (mark_aliases_call_clobbered): Update for may_aliases being a
622 bitmap.
623 (compute_tag_properties): Ditto.
624 (create_partition_for): Ditto.
625 (compute_memory_partitions): Ditto.
626 (dump_may_aliases_for): Ditto.
627 (is_aliased_with): Ditto.
628 (add_may_alias_for_new_tag): Ditto.
629 (rewrite_alias_set_for): Rewrite for may_aliases being a bitmap.
630 (compute_is_aliased): New function.
631 (compute_may_aliases): Call compute_is_aliased).
632 (init_alias_info): Initialize alias_bitmap_obstack.
633 (union_alias_set_into): New function.
634 (compute_flow_sensitive_aliasing): Use union_aliases_into.
635 (have_common_aliases_p): Rewrite to take two bitmaps and use
636 intersection.
637 (compute_flow_insensitive_aliasing): Stop using pointer-sets.
638 Update for bitmaps.
639 (finalize_ref_all_pointers): Update for add_may_alias changes.
640 (new_type_alias): Ditto.
641 * tree-flow-inline.h (may_aliases): Return a bitmap.
642 * tree-dfa.c (dump_variable): Check for MTAG_P'ness.
643 * tree-ssa.c (verify_flow_insensitive_alias_info): Update for
644 may_aliases being a bitmap.
645 * tree-flow.h (struct var_ann_d): Remove may_aliases member.
646 may_aliases now returns a bitmap.
647 * tree-ssa-structalias.c (merge_smts_into): Update for may_aliases
648 being a bitmap.
649 * tree-ssa-operands.c (add_virtual_operand): Update for
650 may_aliases being a bitmap.
652 2007-01-29 Daniel Berlin <dberlin@dberlin.org>
654 PR tree-optimization/30630
655 * tree-ssa-structalias.c (do_complex_constraint): Mark correct
656 variable as changed.
658 2007-01-29 Simon Martin <simartin@users.sourceforge.net>
660 PR c++/28266
661 * gimplify.c (gimplify_target_expr): Make sure that the TARGET_EXPR is
662 expanded only once even if an error occurs.
664 2007-01-29 Ben Elliston <bje@au.ibm.com>
666 * gcov-io.h (__gcov_indirect_call_profiler): Declare.
667 (__gcov_average_profiler): Likewise.
668 (__gcov_ior_profiler): Likewise.
669 (__gcov_merge_ior): Likewise.
671 2007-01-28 Jan Hubicka <jh@suse.cz>
673 * builtins.c (expand_builtin_memset): Fix typo in my last patch.
674 * value-prof.c (stringop_block_profile): Likewise.
676 2007-01-28 Jan Hubicka <jh@suse.cz>
678 * expr.c (emit_block_move_via_movmem, emit_block_move_via_libcall): Add
679 variant handling histograms; add wrapper.
680 (clear_storage_via_libcall): Export.
681 (emit_block_move_hints): Break out from ...; add histograms.
682 (emit_block_move): ... this one.
683 (clear_storage_hints): Break out from ...; add histograms.
684 (clear_storage): ... this one.
685 (set_storage_via_memset): Handle histogram.
686 * expr.h (emit_block_move_via_libcall, emit_block_move_hints): Declare.
687 (clear_storage_hints, clear_storage_via_libcall): Declare.
688 (set_storage_via_setmem): Update prototype.
689 * doc/md.texi (movmem, setmem): Document new arguments.
691 * value-prof.c (dump_histogram_value, tree_find_values_to_profile): Add
692 new histograms.
693 (stringop_block_profile): New global function.
694 (tree_stringops_values_to_profile): Profile block size and alignment.
695 * value-prof.h (enum hist_type): add HIST_TYPE_AVERAGE and
696 HIST_TYPE_IOR.
697 (struct profile_hooks): Add gen_average_profiler and gen_ior_profiler.
698 (stringop_block_profile): Declare.
699 * builtins.c: Include value-prof.h.
700 (expand_builtin_memcpy, expand_builtin_memset): Pass block profile.
701 * gcov-ui.h (GCOV_COUNTER_NAMES): Add new counter.
702 (GCOV_COUNTER_AVERAGE, GCOV_COUNTER_IOR): New constants.
703 (GCOV_COUNTERS, GCOV_LAST_VALUE_COUNTER): Update.
704 * profile.c (instrument_values): Add new counters.
705 * cfgexpand.c (expand_gimple_basic_block): Propagate histograms to
706 calls.
707 * tree-profile.c (tree_average_profiler_fn, tree_ior_profiler_fn): New.
708 (tree_init_edge_profiler): Build new profilers.
709 (tree_gen_average_profiler, tree_gen_ior_profiler): New.
710 (pass_tree_profile): Add dump.
711 (tree_profile_hooks): Update.
712 * Makefile.in (LIBGCOV): Add new constants.
713 * libgcov.c (__gcov_merge_ior, __gcov_average_profiler,
714 __gcov_ior_profiler): New.
715 * i386.md (movmem/setmem expanders): Add new optional arguments.
717 2007-01-28 David Edelsohn <edelsohn@gnu.org>
719 * doc/md.texi (Standard Pattern Names): Document blockage pattern.
721 2007-01-28 Zdenek Dvorak <dvorakz@suse.cz>
723 * tree-ssa-loop-unswitch.c: Include tree-inline.h.
724 (tree_unswitch_single_loop): Pass eni_size_weights to
725 tree_num_loop_insns.
726 * tree-ssa-loop-manip.c: Include tree-inline.h.
727 (can_unroll_loop_p): Pass eni_size_weights to
728 tree_num_loop_insns.
729 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p):
730 Pass eni_size_weights to estimate_num_insns.
731 * tree.h (init_inline_once): Export.
732 * toplev.c (backend_init): Call init_inline_once.
733 * cgraphunit.c (cgraph_process_new_functions,
734 cgraph_analyze_function): Pass eni_inlining_weights to
735 estimate_num_insns.
736 * ipa-inline.c (compute_inline_parameters): Ditto.
737 * tree-ssa-loop-ivcanon.c (tree_num_loop_insns): Pass weights
738 to estimate_num_insns.
739 (try_unroll_loop_completely): Pass eni_size_weights to
740 tree_num_loop_insns.
741 * tree-eh.c (decide_copy_try_finally): Pass eni_size_weights
742 ot estimate_num_insns.
743 * tree-ssa-loop-prefetch.c: Include tree-inline.h.
744 (loop_prefetch_arrays): Pass eni_time_weights to tree_num_loop_insns.
745 * tree-inline.c (eni_inlining_weights, eni_size_weights,
746 eni_time_weights): New variables.
747 (init_inline_once): Initialize them.
748 (struct eni_data): Mew.
749 (estimate_num_insns_1, estimate_num_insns): Use weights.
750 * tree-inline.h (struct eni_weights_d): New.
751 (eni_inlining_weights, eni_size_weights, eni_time_weights): Declare.
752 (estimate_num_insns): Declaration changed.
753 * cfgloop.h (tree_num_loop_insns): Declaration changed.
754 * Makefile.in (tree-ssa-loop-unswitch.o, tree-ssa-loop-prefetch.o,
755 tree-ssa-loop-manip.o): Add TREE_INLINE_H dependency.
757 2007-01-28 Zdenek Dvorak <dvorakz@suse.cz>
759 * tree-data-ref.c (conflict_fn): Assert that the number of affine
760 relations in the conflict function is valid.
762 2007-01-27 Ian Lance Taylor <iant@google.com>
764 * common.opt: Add fstrict-overflow.
765 * opts.c (decode_options): Set flag_strict_overflow if -O2.
766 * flags.h (TYPE_OVERFLOW_WRAPS): Define.
767 (TYPE_OVERFLOW_UNDEFINED): Define.
768 (TYPE_OVERFLOW_TRAPS): Define. This replaces TYPE_TRAP_SIGNED.
769 Replace all uses.
770 * tree.h (TYPE_TRAP_SIGNED): Don't define.
771 * fold-const.c (negate_expr_p): Use TYPE_OVERFLOW_UNDEFINED.
772 (fold_negate_expr): Likewise.
773 (make_range): Likewise.
774 (extract_muldiv_1): Likewise.
775 (maybe_canonicalize_comparison): Likewise.
776 (fold_comparison): Likewise.
777 (fold_binary): Likewise.
778 (tree_expr_nonnegative_p): Likewise.
779 (tree_expr_nonzero_p): Likewise.
780 * tree-vrp.c (compare_values): Likewise.
781 (extract_range_from_binary_expr): Likewise.
782 (extract_range_from_unary_expr): Likewise.
783 * tree-ssa-loop-niter.c (infer_loop_bounds_from_signedness):
784 Likewise.
785 (nowrap_type_p): Likewise.
786 * tree-scalar-evolution.c (simple_iv): Likewise.
787 * fold-const.c (negate_expr_p): Use TYPE_OVERFLOW_WRAPS.
788 (build_range_check): Likewise.
789 (extract_muldiv_1): Likewise.
790 (fold_comparison): Likewise.
791 * tree-vrp.c (vrp_int_const_binop): Likewise.
792 (extract_range_from_unary_expr): Likewise.
793 * convert.c (convert_to_integer): Likewise.
794 * fold-const.c (fold_negate_expr): Use TYPE_OVERFLOW_TRAPS.
795 (fold_comparison): Likewise.
796 (fold_binary): Likewise.
797 * optabs.c (optab_for_tree_code): Likewise.
798 * tree-vectorizer.c (vect_is_simple_reduction): Likewise.
799 * simplify-rtx.c (simplify_const_relational_operation): Check
800 flag_strict_overflow and flag_trapv.
801 (simplify_const_relational_operation): Likewise.
802 * doc/invoke.texi (Option Summary): Mention -fstrict-overflow.
803 (Optimize Options): Add -fstrict-overflow to -O2 list. Document
804 -fstrict-overflow.
806 2007-01-27 Roger Sayle <roger@eyesopen.com>
808 * tree.c (tree_fold_gcd): Delete.
809 * tree.h (tree_fold_gcd): Remove prototype.
810 * tree-data-ref.c (tree_fold_divides_p): Don't use tree_fold_gcd to
811 test whether one constant integer is a multiple of another. Instead
812 call int_const_binop with TRUNC_MOD_EXPR and test for a zero result.
813 * fold-const.c (multiple_of_p): We've determined both TOP and
814 BOTTOM are integer constants so we can call int_const_binop directly
815 instead of the more generic const_binop.
817 2007-01-27 Roger Sayle <roger@eyesopen.com>
819 * fold-const.c (size_binop): In the fast-paths for X+0, 0+X, X-0 and
820 1*X check that the constant hasn't overflowed, to preserve the
821 TREE_OVERFLOW bit.
822 (round_up): Provide an efficient implementation when rouding-up an
823 INTEGER_CST to a power-of-two.
825 2007-01-28 Ralf Wildenhues <Ralf.Wildenhues@gmx.de>
827 * doc/sourcebuild.texi: Add comma for clarity.
828 * doc/extend.texi: Fix some typos.
829 * doc/passes.texi: Likewise.
830 * doc/cppinternals.texi: Likewise.
831 * doc/c-tree.texi: Likewise.
832 * doc/tree-ssa.texi: Likewise.
833 * doc/install.texi: Likewise.
835 2007-01-27 Jan Hubicka <jh@suse.cz>
837 * tree-sra.c (sra_walk_function): Don't rely on aliases being build.
838 (pass_sra): Do not require alias information.
839 * passes.c (init_optimization_passes): Add SRA
841 2007-01-27 Steven Bosscher <steven@gcc.gnu.org>
843 * tracer.c (rest_of_handle_tracer): We already cleaned
844 up the CFG in tracer() so don't do it here again.
845 * cfgcleanup.c (rest_of_handle_jump2): Don't repeat
846 cleanup_cfg here, either. And don't call renumber_insns.
848 * cfgrtl.c (rtl_verify_flow_info_1): Don't verify that BB_END
849 and BB_HEAD are in the insn stream here. Instead make sure
850 that BB_INSN is valid on all insns. Also, do check here that
851 there are no pending branch predictions...
852 (rtl_verify_flow_info): ...instead of doing it here. Checks
853 for BB_END and BB_HEAD moved from rtl_verify_flow_info_1 to
854 here.
856 2007-01-26 Roger Sayle <roger@eyesopen.com>
858 * config/i386/i386.c (ix86_swap_binary_operands_p): New helper
859 function to simplify/factorize operand order canonicalization.
860 (ix86_fixup_binary_operands): Reorganize using the above function.
861 (ix86_binary_operator_ok): Likewise.
863 2007-01-27 Jakub Jelinek <jakub@redhat.com>
865 * genattrtab.c (struct attr_value_list, insn_code_values): Move to
866 file scope from optimize_attrs.
867 (simplify_test_exp): If insn_code_values is not NULL, use it to speed
868 up search.
869 (optimize_attrs): Clear insn_code_values after freeing it.
871 2007-01-26 Zdenek Dvorak <dvorakz@suse.cz>
873 * tree-ssa-address.c (create_mem_ref): Remove ", bsi" from
874 a parts.base assignment.
876 2007-01-26 Zdenek Dvorak <dvorakz@suse.cz>
878 * tree-data-ref.c (dump_subscript): Use dump_conflict_function.
879 (compute_subscript_distance, initialize_data_dependence_relation,
880 finalize_ddr_dependent, analyze_ziv_subscript,
881 analyze_siv_subscript_cst_affine,
882 compute_overlap_steps_for_affine_univar,
883 compute_overlap_steps_for_affine_1_2, analyze_subscript_affine_affine,
884 analyze_siv_subscript, analyze_miv_subscript,
885 analyze_overlapping_iterations, subscript_dependence_tester_1,
886 compute_self_dependence, free_dependence_relation): Work
887 with affine_fn instead of chrecs.
888 (dump_affine_function, dump_conflict_function, affine_function_equal_p,
889 common_affine_function, affine_function_base,
890 affine_function_constant_p, affine_fn_op, affine_fn_plus,
891 affine_fn_minus, affine_fn_free, conflict_fn_not_known,
892 conflict_fn_no_dependence, free_conflict_function, free_subscripts,
893 conflict_fn, affine_fn_cst, affine_fn_univar): New functions.
894 (all_chrecs_equal_p): Removed.
895 * tree-data-ref.h (affine_fn, conflict_function): New types.
896 (struct subscript): Change type of conflicting_iterations_in_a
897 and conflicting_iterations_in_b.
899 2007-01-26 Steve Ellcey <sje@cup.hp.com>
901 PR other/30182
902 * config/pa/pa.h (TARGET_HPUX_11): New.
903 * config/pa/pa-hpux11.h (TARGET_HPUX_11): New.
904 * config/pa/pa.c (pa_init_builtins): Use TARGET_HPUX_11.
906 2007-01-26 Daniel Berlin <dberlin@dberlin.org>
907 Richard Guenther <rguenther@suse.de>
909 * tree-ssa-structalias.c (solve_graph): Handle case
910 we merged the variable to another.
912 2007-01-25 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
914 * builtins.c (fold_builtin_1): Treat ccos and ccosh as 'even'
915 functions.
917 * fold-const.c (negate_mathfn_p): Treat casin, casinh, catan,
918 catanh, cproj, csin, csinh, ctan and ctanh as 'odd' functions.
920 2007-01-25 DJ Delorie <dj@redhat.com>
922 * config/m32c/m32c.c (m32c_cannot_change_mode_class): We don't
923 allow changes to modes which don't fit in those registers.
925 * reload1.c (choose_reload_regs): Check for invalid subregs before
926 computing their locations, not after.
928 2007-01-25 Geoffrey Keating <geoffk@apple.com>
930 PR 25127
931 * config/rs6000/rs6000.c (first_altivec_reg_to_save): On Darwin,
932 save Altivec registers in an eh_return function.
933 (compute_vrsave_mask): Likewise.
934 (rs6000_stack_info): Correct AIX/Darwin stack alignment computation
935 for saving Altivec registers.
936 (rs6000_emit_prologue): Don't allocate stack twice in
937 eh_return function. Correct expected value of altivec_save_offset
938 when using save_world. Describe save of R0 to stack when using
939 save_world. Describe stack pointer adjustment when using
940 save_world. Remove duplicated eh_return parameter register saving.
941 Update sp_offset variable after save_world.
942 * config/rs6000/t-darwin (LIB2FUNCS_STATIC_EXTRA): Remove
943 darwin-world.asm.
944 (LIB2FUNCS_EXTRA): Add darwin-world.asm.
945 * config/rs6000/darwin.h (SUBTARGET_OVERRIDE_OPTIONS): -m64
946 implies Altivec.
948 2007-01-25 Steve Ellcey <sje@cup.hp.com>
950 * config.gcc (ia64*-*-hpux*): Make posix threads the default.
952 2007-01-25 Steve Ellcey <sje@cup.hp.com>
954 PR other/30182
955 * config/pa/pa.c (pa_init_builtins): Set asm names for finite routines.
956 * config/ia64/ia64.c (ia64_init_builtins): Ditto.
958 2007-01-25 Richard Guenther <rguenther@suse.de>
960 * doc/invoke.texi (-Wcoverage-mismatch): Document.
961 * common.opt (-Wcoverage-mismatch): New warning option.
962 * coverage.c (get_coverage_counts): Ignore coverage mismatch
963 if -Wcoverage-mismatch is given.
965 2007-01-25 Razya Ladelsky <razya@il.ibm.com>
967 * ipa-cp.c (ipcp_insert_stage, ipcp_driver): Support for SSA.
968 (ipcp_driver): Change to static definition.
969 Add dumping of the ifunctions.
970 (constant_val_insert): Remove unused parameter. Support for SSA.
971 (ipcp_propagate_const): Support for SSA.
972 (ipcp_profile_bb_print): Print only analyzed nodes.
973 (ipcp_replace_map_create): Remove support for Fortran constant
974 for now.
975 * ipa-prop.c (ipa_method_modify_stmt,
976 ipa_callsite_compute_param): Support for SSA.
977 * ipa-prop.h (ipcp_driver): Remove declaration.
978 (IS_VALID_TREE_MAP_INDEX): Add define.
980 2007-01-24 Geoffrey Keating <geoffk@apple.com>
982 * unwind-dw2.c (execute_stack_op): Handle DW_OP_swap.
984 2007-01-24 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
986 PR middle-end/30447
987 * builtins.c (fold_builtin_cabs): Use MPFR to evaluate a
988 constant argument to cabs and do it without checking for
989 -funsafe-math-optimizations.
991 2007-01-24 Douglas Gregor <dgregor@osl.iu.edu>
993 * c-common.h (RID_FIRST_CXX0X): New.
994 (RID_LAST_CXX0X): New.
995 * c-opts.c (c_common_handle_option): -Wc++0x-compat is triggered
996 by -Wall.
997 * c.opt (Wc++0x-compat): New.
998 * doc/invoke.texi (-Wc++0x-compat): Document.
1000 2007-01-24 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
1002 * builtins.c (fold_builtin_carg): New.
1003 (fold_builtin_1): Use it.
1005 2007-01-24 Jan Hubicka <jh@suse.cz>
1007 * ipa-inline.c (cgraph_decide_inlining): Initialize initial_insns.
1009 * ipa-inline.c (initial_insns, max_insns): Delete.
1010 (compute_max_insns): New function.
1011 (cgraph_decide_inlining_of_small_function): Use it; take minimal amount
1012 of insns as base for code growth.
1013 (cgraph_decide_inlining): Make initial_insns local; do not compute
1014 max_insns.
1015 * params.def (PARAM_INLINE_UNIT_GROWTH): Set to 60.
1016 * doc/invoke.texi (inline-unit-growth): Update docs.
1018 2007-01-24 Jakub Jelinek <jakub@redhat.com>
1020 * config/i386/i386.h (x86_cmpxchg16b): Remove const.
1021 (TARGET_CMPXCHG16B): Define to x86_cmpxchg16b.
1022 * config/i386/i386.c (x86_cmpxchg16b): Remove const.
1023 (override_options): Add PTA_CX16 flag. Set x86_cmpxchg16b
1024 for CPUs that have PTA_CX16 set.
1026 PR middle-end/27416
1027 * gimplify.c (omp_check_private): New function.
1028 (gimplify_scan_omp_clauses): Use it for
1029 firstprivate/lastprivate/reduction.
1031 PR middle-end/30494
1032 * gimplify.c (omp_add_variable): Don't call omp_notice_variable
1033 on TYPE_SIZE_UNIT for GOVD_LOCAL VLAs.
1035 PR middle-end/30421
1036 * omp-low.c (lower_omp_for_lastprivate): Add dlist argument.
1037 If lower_lastprivate_clauses emits some statements, append them
1038 to dlist rather than body_p and to body_p append an initializer.
1039 (lower_omp_for): Adjust caller.
1041 2007-01-24 Steve Ellcey <sje@cup.hp.com>
1043 * target.h (globalize_decl_name): New.
1044 * target-def.h (TARGET_ASM_GLOBALIZE_DECL_NAME): New.
1045 * output.h (default_globalize_decl_name): New.
1046 * varasm.c (asm_output_bss): Use globalize_decl_name instead of
1047 globalize_label.
1048 (globalize_decl): Ditto.
1049 (default_globalize_decl_name): New.
1050 * config/ia64/ia64.c (ia64_globalize_decl_name): New.
1051 (ia64_handle_version_id_attribute): New.
1052 (TARGET_ASM_GLOBALIZE_DECL_NAME): New.
1053 (ia64_asm_output_external): Use globalize_decl_name instead
1054 of globalize_label.
1055 * doc/extend.texi (version_id): New pragma.
1056 * doc/tm.texi (ARGET_ASM_GLOBALIZE_DECL_NAME): New target hook.
1058 2007-01-24 Andreas Krebbel <krebbel1@de.ibm.com>
1060 * unwind-dw2-fde.c (get_cie_encoding): Replaced _Unwind_Word with
1061 _uleb128_t and _Unwind_SWord with _sleb128_t.
1062 * unwind-dw2.c (extract_cie_info, execute_stack_op, execute_cfa_program,
1063 uw_frame_state_for, uw_update_context_1): Likewise.
1064 * unwind-c.c (parse_lsda_header, PERSONALITY_FUNCTION): Likewise.
1065 * unwind-pe.h (read_uleb128, read_sleb128,
1066 read_encoded_value_with_base): Likewise.
1067 * unwind-generic.h: Define _sleb128_t and _uleb128_t types.
1069 2007-01-24 Richard Guenther <rguenther@suse.de>
1071 * builtins.c (expand_builtin_cexpi): Get the fndecl
1072 for cexp in the correct way.
1074 2007-01-24 Jan Hubicka <jh@suse.cz>
1076 * tree-ssa-dce.c (eliminate_unnecesary_stmts): Remove dead LHS of calls.
1078 2007-01-24 Andreas Krebbel <krebbel1@de.ibm.com>
1080 * c-cppbuiltin.c (builtin_define_type_sizeof): New function.
1081 (c_cpp_builtins): New builtin macros: __SIZEOF_INT__, __SIZEOF_LONG__,
1082 __SIZEOF_LONG_LONG__, __SIZEOF_SHORT__, __SIZEOF_POINTER__,
1083 __SIZEOF_FLOAT__, __SIZEOF_DOUBLE__, __SIZEOF_LONG_DOUBLE__,
1084 __SIZEOF_SIZE_T__, __SIZEOF_WCHAR_T__, __SIZEOF_WINT_T__ and
1085 __SIZEOF_PTRDIFF_T__.
1086 * doc/cpp.texi: Documentation for the new builtin macros added.
1088 2007-01-24 Uros Bizjak <ubizjak@gmail.com>
1090 * config/i386/i386.md (tanxf2, tan<mode>2, atan<mode>2, log<mode>2,
1091 log10<mode>2, log2<mode>2, expxf2, exp10xf2, exp2xf2): Use op2
1092 instead of operands[2] to avoid access past the end of array.
1094 2007-01-24 Richard Sandiford <richard@codesourcery.com>
1096 * reload1.c (emit_reload_insns): Pass the reload register
1097 for a non-spill output reload through forget_old_reloads_1.
1099 2007-01-23 Joseph Myers <joseph@codesourcery.com>
1101 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Define
1102 _SOFT_DOUBLE if doubles use software floating-point.
1103 * config/rs6000/libgcc-ppc-glibc.ver: Export additional long
1104 double functions if _SOFT_DOUBLE, not _SOFT_FLOAT.
1105 * config/rs6000/darwin-ldouble.c: Also compile functions for
1106 hard-float without FPRs. Use fmsub function for all __NO_FPRS__
1107 cases. Compile extra functions if _SOFT_DOUBLE, not _SOFT_FLOAT.
1108 * config/rs6000/linuxspe.h (SUBSUBTARGET_OVERRIDE_OPTIONS): Remove
1109 commented-out long double override.
1110 (CPP_LONGDOUBLE_DEFAULT_SPEC): Likewise.
1111 * config/rs6000/eabispe.h: Likewise.
1112 * config/rs6000/rs6000.c (rs6000_override_options): Don't override
1113 long double for non-SPE.
1114 (rs6000_handle_option): Likewise.
1115 (invalid_e500_subreg): Disallow more subregs involding DImode,
1116 DFmode, TImode or TFmode.
1117 (rs6000_legitimate_offset_address_p): Check TFmode offsets for
1118 E500 double.
1119 (legitimate_lo_sum_address_p): Also check for TFmode for E500
1120 double.
1121 (rs6000_legitimize_address): Also handle TFmode for E500 double.
1122 (rs6000_legitimize_reload_address): Also handle TFmode for E500
1123 double.
1124 (rs6000_legitimate_address): Also check for TFmode for E500
1125 double.
1126 (rs6000_emit_move): Use DFmode subregs of TFmode for E500 double.
1127 (spe_build_register_parallel): Handle TFmode and TCmode.
1128 (rs6000_spe_function_arg): Handle TFmode and TCmode for E500
1129 double.
1130 (function_arg): Handle TFmode and TCmode for E500 double.
1131 (rs6000_init_libfuncs): Initialize extra libfuncs for soft double
1132 in general.
1133 (print_operand): Handle TFmode and TImode for %y.
1134 (rs6000_generate_compare): Handle TFmode comparisons for E500
1135 double.
1136 (spe_func_has_64bit_regs_p): Check for TFmode for E500 double.
1137 (rs6000_function_value): Handle TFmode and TCmode for E500 double.
1138 (rs6000_libcall_value): Handle TFmode and TCmode for E500 double.
1139 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Check for
1140 TFmode for E500 double.
1141 * config/rs6000/rs6000.md (FP): Allow TF for E500 double.
1142 (floatsidf2): Enable for E500 double.
1143 (movtf_softfloat): Use rs6000_nonimmediate_operand.
1144 (extenddftf2): Change to extenddftf2_fprs.
1145 (extenddftf2): Call gen_spe_extenddftf2 or gen_extenddftf2_fprs
1146 depending on TARGET_E500_DOUBLE.
1147 (extendsftf2): Enable for E500 double.
1148 (trunctfdf2): Enable for E500 double.
1149 (trunctfsf2): Change to trunctfsf2_fprs.
1150 (trunctfsf2): Call gen_spe_trunctfsf2 or gen_trunctfsf2_fprs
1151 depending on TARGET_E500_DOUBLE.
1152 (floatsitf2): Enable for E500 double.
1153 (fix_trunctfsi2): Change to fix_trunctfsi2_fprs.
1154 (fix_trunctfsi2): Call gen_spe_fix_trunctfsi2 or
1155 gen_fix_trunctfsi2_fprs depending on TARGET_E500_DOUBLE.
1156 (negtf2): Change to negtf2_internal.
1157 (negtf2): New expander.
1158 (abstf2): Enable for E500 double. Call gen_spe_abstf2_tst,
1159 gen_spe_abstf2_cmp or gen_abstf2_internal depending on
1160 TARGET_E500_DOUBLE and flag_unsafe_math_optimizations.
1161 (movdi_internal32): Use rs6000_nonimmediate_operand.
1162 (unnamed splitter): Likewise.
1163 * config/rs6000/spe.md (CMPTFEQ_GPR, TSTTFEQ_GPR, CMPTFGT_GPR,
1164 TSTTFGT_GPR, CMPTFLT_GPR, TSTTFLT_GPR): New unspecs.
1165 (SPE64TF, DITI): New mode macros.
1166 (frob_df_di): Change to frob_<SPE64:mode>_<DITI:mode>; allow more
1167 modes.
1168 (frob_tf_ti): New.
1169 (frob_<mode>_di_2): New.
1170 (frob_tf_di_8_2): New.
1171 (frob_di_df): Change to frob_di_<mode>; allow more modes.
1172 (frob_ti_tf): New.
1173 (frob_di_df_2): Change to frob_<DITI:mode>_<SPE64:mode>_2; allow
1174 more modes.
1175 (frob_ti_<mode>_8_2): New.
1176 (frob_ti_tf_2): New.
1177 (mov_si<mode>_e500_subreg0, mov_si<mode>_e500_subreg0_2,
1178 mov_si<mode>_e500_subreg4, mov_si<mode>_e500_subreg4_2): Allow
1179 TFmode.
1180 (mov_sitf_e500_subreg8, mov_sitf_e500_subreg8_2,
1181 mov_sitf_e500_subreg12, mov_sitf_e500_subreg12_2): New.
1182 (spe_trunctfdf2_internal1, spe_trunctfsf2, spe_extenddftf2,
1183 spe_fix_trunctfsi2, spe_fix_trunctfsi2_internal,
1184 spe_negtf2_internal, spe_abstf2_cmp, spe_abstf2_tst): New.
1185 (cmptfeq_gpr, tsttfeq_gpr, cmptfgt_gpr, tsttfgt_gpr, cmptflt_gpr,
1186 tsttflt_gp): New.
1188 2007-01-23 Ian Lance Taylor <iant@google.com>
1190 * Makefile.in (OBJS-common): Reformat, alphabetize, but put
1191 insn-*.o first.
1192 (OBJS-archive): Reformat, alphabetize.
1193 (OBJS): Change out_object_file to OBJS-md.
1195 2007-01-23 Uros Bizjak <ubizjak@gmail.com>
1197 * config/i386/i386.md: Use REG_P, MEM_P, CONST_INT_P, LABEL_P,
1198 JUMP_P and CALL_P predicates where applicable.
1199 * config/i386/i386.c: Ditto.
1200 * config/i386/i386.md: Ditto.
1201 * config/i386/mmx.md: Ditto.
1202 * config/i386/predicates.md: Ditto.
1204 2007-01-22 Andreas Schwab <schwab@suse.de>
1206 * config/m68k/m68k.h: Fix comment.
1208 2007-01-22 Jan Hubicka <jh@suse.cz>
1210 * passes.c (init_optimization_passes): Do not rerun
1211 pass_early_warn_uninitialized.
1213 2007-01-22 Richard Guenther <rguenther@suse.de>
1215 PR tree-optimization/30038
1216 * tree-ssa-math-opts.c (maybe_record_sincos): New static helper
1217 function.
1218 (execute_cse_sincos_1): Likewise.
1219 (execute_cse_sincos): Likewise.
1220 (gate_cse_sincos): Likewise.
1221 (pass_cse_sincos): New pass CSEing sin() and cos() calls using
1222 the cexpi() canonicalization of sincos().
1223 * tree-pass.h (pass_cse_sincos): Declare.
1224 * passes.c (init_optimization_passes): New pass pas_cse_sincos.
1226 2007-01-21 Eric Botcazou <ebotcazou@libertysurf.fr>
1228 PR rtl-optimization/29329
1229 * combine.c (replaced_rhs_insn): Rename to i2mod.
1230 (replaced_rhs_value): Rename to i2mod_new_rhs.
1231 (i2mod_old_rhs): New global variable.
1232 (combine_instructions): Adjust for above change. Save a copy of
1233 the old RHS into i2mod_old_rhs when the contents of a REG_EQUAL
1234 note are substituted in the second instruction.
1235 (distribute_notes) <REG_DEAD>: Adjust for above change. Do not
1236 ditch the note if it pertains to the second eliminated register
1237 and this register is mentioned in i2mod_old_rhs.
1239 Revert:
1240 2006-09-12 Eric Botcazou <ebotcazou@libertysurf.fr>
1242 * combine.c (distribute_notes) <REG_DEAD>: Do not consider SETs past
1243 the insn to which the note was originally attached.
1245 2007-01-21 Jan Hubicka <jh@suse.cz>
1247 * ipa-inline.c (inlining_mode): Comment, move up.
1248 (cgraph_decide_inlining_incrementally): Do not perform inlining
1249 itself; fix handling of flattening of self recursive functions.
1250 (cgraph_find_cycles): Remove.
1251 (cgraph_flatten_node): Remove.
1252 (cgraph_decide_inlining): Use incremental inliner to handle flattening.
1253 (try_inline): New function.
1254 (cgraph_early_inlining): Update call of
1255 cgraph_decide_inlining_incrementally. Apply inlining here.
1256 (apply_inline): Update call of cgraph_decide_inlining_incrementally.
1258 2007-01-21 Dirk Mueller <dmueller@suse.de>
1260 PR bootstrap/30511
1261 * tree-vrp.c (check_array_bounds): do not warn
1262 about ADDR_EXPR's of ARRAY_REF's which are immediately
1263 used in binary expressions.
1265 2007-01-21 Ira Rosen <irar@il.ibm.com>
1267 * tree-vectorizer.h (struct _stmt_vec_info): Add new field
1268 read_write_dep and macros for its access.
1269 * tree-vectorizer.c (new_stmt_vec_info): Initialize the new field.
1270 * tree-vect-analyze.c (vect_analyze_data_ref_dependence): Remove
1271 argument, call vect_check_interleaving for every independent pair of
1272 data-refs. Mark loads that access the same memory location as a store
1273 in the loop.
1274 (vect_check_dependences): Remove.
1275 (vect_analyze_data_ref_dependences): Remove vect_check_dependences
1276 call, fix the call to vect_analyze_data_ref_dependence.
1277 (vect_analyze_data_ref_access): For statements that access the same
1278 data-ref, check that they are not stores; for loads, check that there
1279 is no store that access the same location.
1281 2007-01-20 Roger Sayle <roger@eyesopen.com>
1282 Joseph Myers <joseph@codesourcery.com>
1284 * doc/invoke.texi (-fdump-rtl-combine): Fix under/overfull hbox.
1285 (-fdump-rtl-stack): Likewise.
1286 (-fno-signed-zeros): Use @minus{} for a minus sign. Correct typo.
1287 (-mcheck-zero-division, -mcpu): Fix under/overfull hbox.
1288 (-mpt-fixed): Use @minus{} for minus sign.
1289 (Using Precompiled Headers): Fix under/overfull hbox.
1291 2007-01-20 Jan Hubicka <jh@suse.cz>
1293 * tree-flow.h (struct stmt_ann_d): Move references_memory to proper
1294 place within annotation.
1296 2007-01-20 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
1298 * pa.c (output_move_double): Change array size of xoperands to 4.
1300 2007-01-20 Jan Hubicka <jh@suse.cz>
1302 * tree-tailcall.c (adjust_return_value): Do not use RESULT_DECL
1303 as temporary.
1305 2007-01-19 Ian Lance Taylor <iant@google.com>
1307 * expmed.c (expand_divmod) [TRUNC_MOD_EXPR, TRUNC_DIV_EXPR]: Cast
1308 constant to unsigned HOST_WIDE_INT before negating.
1310 2007-01-19 Ian Lance Taylor <iant@google.com>
1312 * tree-ssa-operands.h (struct vuse_vec_d): Change num_vuse field
1313 to unsigned.
1314 (VUSE_VECT_ELEMENT) [ENABLE_CHECKING]: Use unsigned comparison.
1315 (VUSE_ELEMENT_PTR) [ENABLE_CHECKING]: Likewise.
1316 (SET_VUSE_VECT_ELEMENT) [ENABLE_CHECKING]: Likewise.
1317 (SET_VUSE_ELEMENT_VAR) [ENABLE_CHECKING]: Likewise.
1318 (SET_VUSE_ELEMENT_PTR) [ENABLE_CHECKING]: Likewise.
1319 (realloc_vdef, realloc_vuse): Change second parameter to
1320 unsigned.
1321 (ssa_operand_iterator_d): Change vuse_index and mayuse_index
1322 fields to unsigned.
1323 * tree-ssa-operands.c (realloc_vop): Change num_elem parameter to
1324 unsigned. Change x and lim locals to unsigned.
1325 (realloc_vdef, realloc_vuse): Change num_elem parameter to
1326 unsigned.
1327 (finalize_ssa_vuse_ops): Change old_i local to unsigned.
1328 (copy_virtual_operands): Change i and n locals to unsigned.
1330 2007-01-19 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
1332 PR middle-end/29335
1333 * builtins.c (fold_builtin_1): Handle builtin fdim.
1335 2007-01-20 Jan Hubicka <jh@suse.cz>
1337 * tree-ssa.c (init_tree_ssa): Do not call init_alias_heapvars.
1338 * tree-ssa-structalias.c (compute_points_to_sets): Do call
1339 init_alias_heapvars.
1340 (init_alias_heapvars): Initialize only when not already initialized.
1341 (delete_alias_heapvars): Set heapvar_for_stmt to NULL.
1343 2007-01-19 Roger Sayle <roger@eyesopen.com>
1345 * common.opt (fsigned-zeros): New command line option.
1346 * flags.h (HONOR_SIGNED_ZEROS): Control via flag_signed_zeros instead
1347 of flag_unsafe_math_optimizations.
1348 * opts.c (set_fast_math_flags): The -ffast-math command line option
1349 implies -fno-signed-zeros.
1350 (fast_math_flags_set_p): Likewise.
1352 * doc/invoke.texi: Document new -fno-signed-zeros option, and update
1353 the documentation of -ffast-math appropriately. Wrap long lines.
1355 2007-01-19 Steve Ellcey <sje@cup.hp.com>
1357 * system.h (ASM_MAKE_LABEL_LINKONCE): Poison.
1358 * varasm.c (globalize_decl): Remove ASM_MAKE_LABEL_LINKONCE ifdef.
1360 2007-01-19 Tomas Bily <tbily@suse.cz>
1362 * cgraphunit.c (cgraph_finalize_function): Updating of pid
1363 * tree-profile.c:
1364 (tree_init_ic_make_global_vars): New function
1365 (tree_init_edge_profiler): call of tree_init_ic_make_global_vars
1366 (tree_gen_ic_profiler): New function
1367 (tree_gen_ic_func_profiler): New function
1368 (tree_profiling): Added calling of tree_gen_ic_func_profiler
1369 (tree_profile_hooks): Added hook for indirec/virtual calls
1370 * value-prof.c (tree_find_values_to_profile): New case for
1371 indirect calls
1372 (tree_values_to_profile): Call for determining indirect/virtual
1373 counters
1374 (tree_indirect_call_to_profile): New function
1375 (tree_ic_transform): New function
1376 (tree_ic): New function
1377 (find_func_by_pid): New function
1378 (init_pid_map): New function
1379 (tree_value_profile_transformations): Added check for
1380 indirect/virtual call transformation
1381 * value-prof.h (enum hist_type): New counter type for
1382 indirect/virtual calls
1383 (profile_hooks): Added new hook for profiling indirect/virtual
1384 calls
1385 * profile.c (instrument_values): New case for indirect/virtual
1386 call added
1387 * gcov-io.h (GCOV_LAST_VALUE_COUNTER): Changed to 6
1388 (GCOV_COUNTER_V_INDIR): New counter type
1389 (GCOV_COUNTER_NAMES): New name of counter "indirect" added
1390 (GCOV_MERGE_FUNCTIONS): New merge function for indirect/virtual
1391 call added
1392 * cgraph.c: Definition of cgraph_max_pid
1393 (cgraph_create_node): Default init of pid attribute
1394 * cgraph.h: Declaration of cgraph_max_pid
1395 (struct cgraph_node): Added pid attribute
1396 * libgcov.c (__gcov_indirect_call_profiler): New function
1397 (__gcov_one_value_profiler_body): New function
1398 (__gcov_one_value_profiler): Body was moved to
1399 __gcov_one_value_profiler_body and calls it
1401 2007-01-19 Basile Starynkevitch <basile@starynkevitch.net>
1403 * doc/gty.texi (Options): Document the mark_hook option to GTY.
1404 * gengtype.c (write_types_data, write_func_for_structure,
1405 write_types, ggc_wtd, pch_wtd): Add skip_hooks to
1406 write_types_data, ggc_wtd, pch_wtd for processing mark_hook.
1407 (walk_type, write_func_for_structure): Generate the mark_hook if
1408 needed.
1410 2007-01-19 Jan Hubicka <jh@suse.cz>
1412 * ipa-inline.c (cgraph_decide_inlining_incrementally): Instead of
1413 'early' argument take inlining mode argument specifying whether to
1414 inline for size/speeed or all functions; add support for flattening;
1415 improve dumpting.
1416 (cgraph_early_inlining): Update call of decide_inlining_incrementally.
1418 2007-01-19 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
1420 PR c++/17947
1421 * toplev.c (warn_deprecated_use): Use %qD instead of %qs to print
1422 the name of the declared identifier.
1424 2007-01-19 Dirk Mueller <dmueller@suse.de>
1426 * config/i386.h (CONDITIONAL_REGISTER_USAGE): Store
1427 result of PIC_OFFSET_TABLE_REGNUM in temporary variable to avoid
1428 duplicate evaluation.
1430 2007-01-19 Uros Bizjak <ubizjak@gmail.com>
1432 * config/i386/i386.md (acos<mode>2): Rename from acossf2 and acosdf2.
1433 Macroize expander using X87MODEF12 mode macro. Extend operand 1
1434 to XFMode, use acosxf2 and truncate result to requested mode.
1435 Use SSE_FLOAT_MODE_P to disable patterns for SSE math.
1436 (asin<mode>2): Similarly, with asin expanders.
1437 (*fscalexf4_i387): Rename from *fscalexf4.
1438 (expNcorexf3): New expander.
1439 (expxf2, exp10xf2, exp2xf2): Use expNcorexf3 expander.
1440 (exp<mode>2): Rename from expsf2 and expdf2. Macroize expander using
1441 X87MODEF12 mode macro. Extend operand 1 to XFMode, use expxf2 and
1442 truncate result to requested mode. Use SSE_FLOAT_MODE_P to disable
1443 patterns for SSE math.
1444 (exp10<mode>2): Similarly, with exp10 expanders.
1445 (exp2<mode>2): Similarly, with exp2 expanders.
1446 (expm1<mode>2): Similarly, with expm1 expanders.
1447 (ldexp<mode>3): Similarly, with ldexp expanders.
1448 (log<mode>2, log10<mode>2, log2<mode>2, log1p<mode>2, logb<mode>2):
1449 Use gen_truncxf<mode>2_i387_noop to truncate result.
1451 2007-01-19 Richard Sandiford <richard@codesourcery.com>
1453 * config/m68k/m68k.h (PREFERRED_STACK_BOUNDARY): Define to 32
1454 for ColdFire targets.
1456 2007-01-19 Nathan Sidwell <nathan@codesourcery.com>
1457 Richard Sandiford <richard@codesourcery.com>
1459 * config/m68k/m68k.h (M68K_STATIC_CHAIN_REG_NAME): New macro.
1460 (INITIALIZE_TRAMPOLINE): Use STATIC_CHAIN_REGNUM.
1461 (__transfer_from_trampoline): Use M68K_STATIC_CHAIN_REG_NAME.
1462 * config/m68k/m68kelf.h (STATIC_CHAIN_REGNUM): Override.
1463 (M68K_STATIC_CHAIN_REG_NAME): Likewise.
1464 * config/m68k/netbsd-elf.h (M68K_STATIC_CHAIN_REG_NAME): Likewise.
1466 2007-01-19 Richard Sandiford <richard@codesourcery.com>
1468 * config/m68k/m68k.md (adddi_dilshr32): Rename to...
1469 (*adddi_dilshr32): ...this. Fix formatting. Remove commented-out
1470 non-canonical pattern. Restrict to !TARGET_COLDFIRE.
1471 (*adddi_dilshr32_cf): New pattern.
1472 (adddi3, subdi3): Remove first alternatives.
1474 2007-01-19 Richard Sandiford <richard@codesourcery.com>
1476 * config/m68k/m68k.c (notice_update_cc): If an SFmode move is
1477 implemented using move.l, do not use its cc result for floating-point
1478 comparisons.
1480 2007-01-19 Richard Sandiford <richard@codesourcery.com>
1482 * config/m68k/m68k.h (EXTRA_CONSTRAINT): Stop the 'T' constraint
1483 from accepting 's' constraints if flag_pic.
1485 2007-01-19 Richard Sandiford <richard@codesourcery.com>
1487 * config/m68k/m68k.md (bordered, bunordered, buneq, bunge, bungt)
1488 (bunle, bunlt, bltgt, bordered_rev, bunordered_rev, buneq_rev)
1489 (bunge_rev, bungt_rev, bunle_rev, bunlt_rev, bltgt_rev): Change
1490 condition from TARGET_68881 to TARGET_HARD_FLOAT.
1492 2007-01-19 Sandra Loosemore <sandra@codesourcery.com>
1494 * longlong.h (count_leading_zeros, COUNT_LEADING_ZEROS_0): Add
1495 ColdFire alternatives.
1496 * config/m68k/m68k.h (CLZ_DEFINED_VALUE_AT_ZERO): New macro.
1497 * config/m68k/m68k.md (clzsi2): Define for ColdFire
1498 architectures that support ff1 instruction.
1500 2007-01-19 Richard Sandiford <richard@codesourcery.com>
1501 Julian Brown <julian@codesourcery.com>
1503 * config/m68k/m68k.h (CONST_OK_FOR_LETTER_P): Add an 'R' case.
1504 * config/m68k/m68k.md (*movsi_cfv4): Fold into...
1505 (*movsi_cf): ...here. Remove unnecessary 'R' from 'Rg'.
1506 Add commentary.
1508 2007-01-19 Richard Sandiford <richard@codesourcery.com>
1510 * config/m68k/m68k-protos.h (valid_mov3q_const): Take a HOST_WIDE_INT
1511 and return a bool.
1512 (output_move_const_into_data_reg, output_move_simode_const): Delete.
1513 * config/m68k/m68k.c (const_method, const_int_cost): Take a
1514 HOST_WIDE_INT instead of an rtx.
1515 (m68k_rtx_costs): Update call accordingly.
1516 (output_move_const_into_data_reg): Likewise. Fix formatting.
1517 (valid_mov3q_const): Take a HOST_WIDE_INT instead of an rtx.
1518 Return a bool.
1519 (output_move_simode_const): Update calls after above changes.
1520 Rework to use automatic variables and predicates like MEM_P.
1521 * config/m68k/m68k.md (pushexthisi_const): Update call to
1522 valid_mov3q_const.
1524 2007-01-19 Dirk Mueller <dmueller@suse.de>
1526 * tree-ssa-alias.c (perform_var_substitution): Fix typo
1527 in dump_flags test.
1529 2007-01-19 Richard Guenther <rguenther@suse.de>
1531 * builtins.c (expand_builtin_cexpi): Fall back to expanding
1532 via cexp in case sincos is not available.
1534 2007-01-19 Richard Guenther <rguenther@suse.de>
1536 * doc/tm.texi (TARGET_HAS_SINCOS): Document new target macro.
1537 * defaults.h (TARGET_HAS_SINCOS): Default to off.
1538 * config/linux.h (TARGET_HAS_SINCOS): Set to on if we have glibc.
1539 * config/alpha/linux.h (TARGET_HAS_SINCOS): Likewise.
1540 * config/sparc/linux.h (TARGET_HAS_SINCOS): Likewise.
1541 * config/sparc/linux64.h (TARGET_HAS_SINCOS): Likewise.
1542 * config/rs6000/linux.h (TARGET_HAS_SINCOS): Likewise.
1543 * config/rs6000/linux64.h (TARGET_HAS_SINCOS): Likewise.
1545 2007-01-19 Uros Bizjak <ubizjak@gmail.com>
1547 * config/i386/i386.md (*fpatanxf3_i387, fpatan_extend<mode>xf3_i387):
1548 New insn patterns.
1549 (atan2sf3_1, atan2df3_1, atan2xf3_1): Remove insn patterns.
1550 (atan2xf3): Directly generate RTL pattern.
1551 (atan2<mode>3): Rename from atan2sf3 and atan2df3 and macroize insn
1552 patterns using X87MODEF12 mode macro. Use fpatan_extend<mode>xf3_i387
1553 and truncate result to requested mode. Use SSE_FLOAT_MODE_P to
1554 disable patterns for SSE math.
1555 (atan<mode>2): Rename from atansf2 and atandf2 and macroize insn
1556 patterns using X87MODEF12 mode macro. Use fpatan_extend<mode>xf3_i387
1557 and truncate result to requested mode. Use SSE_FLOAT_MODE_P to
1558 disable patterns for SSE math.
1560 2007-01-19 Alexandre Oliva <aoliva@redhat.com>
1562 * libgcc-std.ver: Fix typo in %inherit for GCC_4.3.0.
1564 2007-01-18 Roger Sayle <roger@eyesopen.com>
1566 * fold-const.c (fold_unary) <VIEW_CONVERT_EXPR>: Optimize away a
1567 VIEW_CONVERT_EXPR to the same type as it's operand.
1569 2007-01-18 David Edelsohn <edelsohn@gnu.org>
1571 * config/rs6000/darwin-ldouble.c: Only build _SOFT_FLOAT if
1572 configured for long double 128.
1574 2007-01-18 Mike Stump <mrs@apple.com>
1576 * config/rs6000/rs6000.c (rs6000_emit_vector_compare): Fix build
1577 error.
1579 2007-01-18 Michael Meissner <michael.meissner@amd.com>
1581 * i386.c (ix86_compute_frame_layout): Make fprintf's in #if 0 code
1582 type correct.
1584 2007-01-18 Jan Hubicka <jh@suse.cz>
1586 * tree-ssa-operands.c (vop_free_bucket_size): Never return value
1587 greater than NUM_VOP_FREE_BUCKETS.
1589 2007-01-18 Daniel Berlin <dberlin@dberlin.org>
1591 * tree-ssa-structalias.c: Update comments.
1592 (ptabitmap_obstack): Removed.
1593 (pta_obstack): New.
1594 (oldpta_obstack): Ditto.
1595 (stats): Add a few members.
1596 (struct variable_info): Remove node, complex, address_taken, and
1597 indirect_target members. Add oldsolution member.
1598 (new_var_info): Do not initialize removed members.
1599 (constraint_expr_type): Remove INCLUDES.
1600 (constraint_graph): Add size, implicit_preds, rep,
1601 indirect_cycles, eq_rep, label, direct_nodes, and complex members.
1602 (FIRST_REF_NODE): New macro.
1603 (LAST_REF_NODE): Ditto.
1604 (FIRST_ADDR_NODE): Ditto.
1605 (find): New function.
1606 (unite): Ditto.
1607 (dump_constraint): Do not handle INCLUDES.
1608 (insert_into_complex): Do not insert duplicate constraints.
1609 (condense_varmap_nodes): Renamed and rewritten into ...
1610 (merge_node_constraints): This. Also fix bug in handling of
1611 offseted copy constraints.
1612 (clear_edges_for_node): No longer need to deal with preds at all,
1613 or removing associated preds/succs.
1614 (merge_graph_nodes): Deal with indirect_cycles.
1615 Don't deal with predecessors.
1616 (add_implicit_graph_edge): New function.
1617 (add_pred_graph_edge): Ditto.
1618 (add_graph_edge): Don't deal with predecessors.
1619 (build_constraint_graph): Removed.
1620 (build_pred_graph): New function.
1621 (build_succ_graph): Ditto.
1622 (struct scc_info): Removed in_component. Added roots, dfs, and
1623 node_mapping. Remove visited_index, unification_queue.
1624 (scc_visit): Deal with union-find we do now.
1625 Deal with cycles with REF nodes.
1626 (collapse_nodes): Renamed and rewritten to ...
1627 (unify_nodes): This.
1628 (process_unification_queue): Removed.
1629 (topo_visit): Cleanup
1630 (do_da_constraint): Use find.
1631 (do_sd_constraint): Ditto.
1632 (do_ds_constraint): Ditto.
1633 (do_complex_constraint): Ditto.
1634 (init_scc_info): Update for removed and added members.
1635 (find_and_collapse_graph_cycles): Renamed and rewritten into ...
1636 (find_indirect_cycles): This.
1637 (equivalence_class): New variable.
1638 (label_visit): New function.
1639 (perform_variable_substitution): Rewritten.
1640 (free_var_substitution_info): New function.
1641 (find_equivalent_node): Ditto.
1642 (move_complex_constraints): Ditto.
1643 (eliminate_indirect_cycles): Ditto.
1644 (solve_graph): Only propagate changed bits.
1645 Use indirect cycle elimination.
1646 Use find.
1647 (tree_id_t): Rename to tree_vi_t, delete id member, add vi member.
1648 (tree_id_eq): Renamed to ...
1649 (tree_vi_eq): This. Update for member change
1650 (insert_id_for_tree): Renamed and rewritten to ...
1651 (insert_vi_for_tree): This.
1652 (lookup_id_for_tree): Renamed and rewritten to ...
1653 (lookup_vi_for_tree): This.
1654 (get_id_for_tree): Renamed and rewritten to ...
1655 (get_vi_for_tree): Ditto.
1656 (get_constraint_exp_from_ssa_var): Update to use get_vi_for_tree.
1657 (process_constraint): Don't handle INCLUDES.
1658 Remove special ADDRESSOF case.
1659 (find_func_aliases): Rewrite to use vi functions instead of id
1660 ones.
1661 (create_function_info_for): Ditto.
1662 (create_variable_info_for): Ditto.
1663 (intra_create_variable_infos): Ditto.
1664 (merge_smts_into): Ditto.
1665 (find_what_p_points_to): Ditto.
1666 (init_base_vars): Ditto.
1667 (init_alias_vars): Ditto.
1668 (remove_preds_and_fake_succs): New function.
1669 (dump_sa_points_to_info): Dump new stats.
1670 (dump_solution_for_var): Use find.
1671 (set_used_smts): Fix formatting.
1672 (compute_points_to_sets): Updated for new functions.
1673 (ipa_pta_execute): Ditto.
1675 2007-01-18 Kazu Hirata <kazu@codesourcery.com>
1676 Richard Sandiford <richard@codesourcery.com>
1678 * doc/tm.texi (TARGET_FUNCTION_VALUE): Expand documentation of
1679 parallels.
1680 * calls.c (expand_call): If the return value is a PARALLEL,
1681 extract its first member.
1682 * config/m68k/linux.h (FUNCTION_EXTRA_EPILOGUE): Remove.
1683 * config/m68k/m68k.c (m68k_output_function_epilogue): Don't
1684 use FUNCTION_EXTRA_EPILOGUE.
1685 (m68k_function_value): Return a PARALLEL if the return value
1686 is of a pointer type.
1687 * config/m68k/netbsd-elf.h (current_function_returns_pointer)
1688 (FUNCTION_EXTRA_EPILOGUE): Remove.
1689 * config/m68k/m68k.md (D0_REG): New constant.
1691 2007-01-18 Kazu Hirata <kazu@codesourcery.com>
1693 * config/m68k/m68k.c (m68k_output_function_epilogue): Don't
1694 output a NOP for empty epilogues.
1696 2007-01-18 Richard Sandiford <richard@codesourcery.com>
1698 * config/m68k/m68k.c (m68k_use_return_insn): Update comments
1699 before function. Extend register save check to include all
1700 registers, not just integer ones.
1702 2007-01-18 Kazu Hirata <kazu@codesourcery.com>
1704 * config/m68k/m68k-protos.h (use_return_insn): Rename to...
1705 (m68k_use_return_insn): ...this.
1706 * config/m68k/m68k.h (USE_RETURN_INSN): Delete.
1707 * config/m68k/m68k.c (use_return_insn): Rename to...
1708 (m68k_use_return_insn): ...this.
1709 * config/m68k/m68k.md (return): Use m68k_use_return_insn instead
1710 of USE_RETURN_INSN.
1712 2007-01-18 Nathan Sidwell <nathan@codesourcery.com>
1714 * config/m68k/fpgnulib.c (__truncdfsf2): Implement round to
1715 nearest even, fix denormal rounding overflow.
1717 2007-01-18 Richard Sandiford <richard@codesourcery.com>
1719 * config/m68k/m68k.md (movsf_cf_hard): Use fsmove instead of
1720 f%$smove and f%$move.
1721 (movdf_cf_hard): Use fdmove for cases 0 and 3 and fmove for case 1.
1722 (extendsfdf2_cf): Use fdmove instead of f%&move.
1723 (truncdfsf2_cf): Use fsmove instead of f%$smove.
1724 (add<mode>3_cf, sub<mode>3_cf): Use <FP:prec> instead of <FP:round>.
1726 2007-01-18 Richard Sandiford <richard@codesourcery.com>
1728 * config/m68k/m68k.md (movdf_cf_hard): Use output_move_double for
1729 GPR<-GPR moves.
1731 2007-01-18 Richard Sandiford <richard@codesourcery.com>
1733 * real.h (real_format): Add a canonical_nan_lsbs_set field.
1734 (coldfire_single_format): Declare.
1735 (coldfire_double_format): Likewise.
1736 * real.c (encode_ieee_single): Use canonical_nan_lsbs_set instead
1737 of qnan_msb_set to determine the lower bits of a canonical
1738 NaN significand.
1739 (encode_ieee_double): Likewise.
1740 (encode_ieee_quad): Likewise.
1741 (ieee_single_format): Initialize canonical_nan_lsbs_set.
1742 (mips_single_format): Likewise.
1743 (ieee_double_format): Likewise.
1744 (mips_double_format): Likewise.
1745 (ieee_extended_motorola_format): Likewise.
1746 (ieee_extended_intel_96_format): Likewise.
1747 (ieee_extended_intel_128_format): Likewise.
1748 (ieee_extended_intel_96_round_53_format): Likewise.
1749 (ibm_extended_format): Likewise.
1750 (mips_extended_format): Likewise.
1751 (ieee_quad_format): Likewise.
1752 (mips_quad_format): Likewise.
1753 (vax_f_format): Likewise.
1754 (vax_d_format): Likewise.
1755 (vax_g_format): Likewise.
1756 (i370_single_format): Likewise.
1757 (i370_double_format): Likewise.
1758 (decimal_single_format): Likewise.
1759 (decimal_double_format): Likewise.
1760 (decimal_quad_format): Likewise.
1761 (c4x_single_format): Likewise.
1762 (c4x_extended_format): Likewise.
1763 (real_internal_format): Likewise.
1764 (coldfire_single_format): New real_format.
1765 (coldfire_double_format): Likewise.
1766 * config/pdp11/pdp11.c (pdp11_f_format): Initialize
1767 canonical_nan_lsbs_set.
1768 (pdp11_d_format): Likewise.
1769 * config/m68k/m68k.c (override_options): Override REAL_FORMAT_MODE
1770 if TARGET_COLDFIRE_CPU.
1772 2007-01-18 Richard Sandiford <richard@codesourcery.com>
1774 * config/m68k/m68k-protos.h (m68k_output_pic_call): Delete.
1775 (output_call, m68k_legitimize_call_address): Declare.
1776 * config/m68k/m68k.h (EXTRA_CONSTRAINT): Remove unnecessary
1777 parenthesees. Add support for a 'W' constraint.
1778 (LEGITIMATE_PIC_OPERAND_P): Remove SYMBOL_REF_FLAG handling.
1779 (PRINT_OPERAND_PUNCT_VALID_P): Remove comment about 'o'.
1780 (m68k_symbolic_call, m68k_symbolic_jump): Declare.
1781 * config/m68k/m68k.c (m68k_symbolic_call, m68k_symbolic_jump): New
1782 variables.
1783 (override_options): Initialize them. Do not set flag_no_function_cse
1784 for TARGET_ID_SHARED_LIBRARY.
1785 (m68k_output_pic_call): Delete.
1786 (m68k_legitimize_call_address): New function.
1787 (print_operand): Remove the %o prefix. Handle the %p prefix.
1788 (output_call): New function.
1789 (m68k_output_mi_thunk): Use m68k_symbolic_jump. Always load the
1790 target address from the GOT if symbolic jumps are not allowed.
1791 * config/m68k/m68k.md (call, general_operand): Do not set
1792 SYMBOL_REF_FLAG. Use m68k_legitimize_call_address instead.
1793 Merge separate flag_pic and !flag_pic define_insns into...
1794 (*call, *call_value): ...these new patterns. Match the address
1795 rather than the containing MEM and require it to be a call_operand.
1796 Use output_call to generate the asm template.
1797 * config/m68k/predicates.md (const_call_operand): New predicate.
1798 (call_operand): Likewise.
1800 2007-01-18 Nathan Sidwell <nathan@codesourcery.com>
1802 * config/m68k/m68k.h (REGISTER_MOVE_COST): Simplify definition.
1803 (STACK_GROWS_DOWNWARD): Define to 1.
1804 (FUNCTION_VALUE, LIBCALL_VALUE, FUNCTION_VALUE_REGNO_P): Equivocate
1805 comments, emphasizing that these values are only defaults.
1806 * config/m68k/linux.h (LINK_SPEC): Fix formatting in #undef.
1807 * config/m68k/m68k.c (const_method): Remove trailing whitespace.
1809 2007-01-18 Richard Sandiford <richard@codesourcery.com>
1811 * config/m68k/m68k.md (cmpsi): Remove outdated flag_pic handling.
1813 2007-01-18 Kazu Hirata <kazu@codesourcery.com>
1814 Richard Sandiford <richard@codesourcery.com>
1816 * config/m68k/m68k.h (DATA_REGNO_P, ADDRESS_REGNO_P, INT_REGNO_P)
1817 (FP_REGNO_P): New macros.
1818 (REGNO_OK_FOR_INDEX_P, REGNO_OK_FOR_BASE_P, REGNO_OK_FOR_DATA_P)
1819 (REGNO_OK_FOR_FP_P, REG_OK_FOR_INDEX_P, REG_OK_FOR_BASE_P): Use them.
1820 (EH_RETURN_STACKADJ_RTX): Use A0_REG.
1821 * config/m68k/m68k.c (m68k_regno_mode_ok): Use the new REGNO macros.
1823 2007-01-18 Nathan Sidwell <nathan@codesourcery.com>
1825 * config.gcc (m68k-*-aout*, m68k-*-coff*, m68020-*-elf*, m68k-*-elf*)
1826 (m68k-*-uclinuxoldabi, m68k-*-uclinux*, m68k-*-rtems*): Add t-floatlib
1827 to $tmake_file.
1828 * config/m68k/t-floatlib: New file, extracting common code from...
1829 * config/m68k/t-m68kbare, config/m68k/t-m68kelf,
1830 * config/m68k/t-uclinux: Here.
1831 * config/m68k/fpgnulib.c: Do not compile extendeed precision
1832 routines on ColdFire targets.
1834 2007-01-18 Nathan Sidwell <nathan@codesourcery.com>
1836 * config.gcc (m68k-*-aout*, m68k-*-coff*, m68020-*-elf*, m68k-*-elf*)
1837 (m68010-*-netbsdelf*, m68k*-*-netbsdelf*, m68k*-*-openbsd*)
1838 (m68k-*-uclinuxoldabi, m68k-*-uclinux*, m68k-*-linux*)
1839 (m68k-*-rtems*): Use tm_file rather than m68k/m68k.h and
1840 explicitly set MOTOROLA to 1.
1841 * config/m68k/m68k.h (MOTOROLA): Simplify definition accordingly.
1843 2007-01-18 Richard Sandiford <richard@codesourcery.com>
1844 Nathan Sidwell <nathan@codesourcery.com>
1846 * config/m68k/m68k.h (PCC_STATIC_STRUCT_RETURN): Delete.
1847 (ASM_OUTPUT_REG_PUSH, ASM_OUTPUT_REG_POP): Add MOTOROLA cases.
1848 * config/m68k/coff.h (REGISTER_PREFIX_MD): Delete.
1849 * config/m68k/m68020-elf.h (LIB_SPEC): Delete.
1850 * config/m68k/m68k-none.h (CC1_SPEC, CPP_SUBTARGET_SPEC): Delete.
1851 * config/m68k/m68kelf.h (IMMEDIATE_PREFIX, REGISTER_PREFIX_MD)
1852 (ASM_OUTPUT_REG_PUSH, ASM_OUTPUT_REG_POP): Delete.
1853 (NO_DOLLAR_IN_LABEL, PCC_STATIC_STRUCT_RETURN): Don't undefine.
1854 (BSS_ASM_OP, ASM_OUTPUT_SKIP, ASM_OUTPUT_ASCII): Delete.
1855 * config/m68k/m68kemb.h (PCC_STATIC_STRUCT_RETURN): Don't undefine.
1856 * config/m68k/linux.h (SIZE_TYPE, PTRDIFF_TYPE, WCHAR_TYPE)
1857 (WCHAR_TYPE_SIZE, TARGET_OBJFMT_CPP_BUILTINS): Delete.
1858 (TARGET_OS_CPP_BUILTINS): Don't define mc68000 and mc68020 here.
1859 (DBX_REGISTER_NUMBER): Delete.
1860 * config/m68k/netbsd-elf.h (IMMEDIATE_PREFIX): Delete.
1861 (PCC_STATIC_STRUCT_RETURN): Don't undefine.
1862 * config/m68k/openbsd.h (PCC_STATIC_STRUCT_RETURN): Define.
1864 2007-01-18 Richard Sandiford <richard@codesourcery.com>
1866 * config.gcc (m68k-*-uclinux*): Add flat.h to $tm_file.
1867 * config/flat.h: New file.
1868 * crtstuff.c (USE_PT_GNU_EH_FRAME): Don't define if
1869 OBJECT_FORMAT_FLAT.
1870 * config/m68k/m68k.h (ASM_PREFERRED_EH_DATA_FORMAT): Do not use
1871 indirect references for -msep-data or -mid-shared-library.
1872 Do not use PC-relative code addresses either.
1873 * config/m68k/m68k.c (override_options): Restrict -fPIC error
1874 to -mpcrel.
1875 * config/m68k/uclinux.h (STARTFILE_SPEC): Define. Use Scrt1.o
1876 for shared libraries and crt1.o for executables. Use crti.o and
1877 crtbegin.o.
1878 (ENDFILE_SPEC): Use crtend.o and crtn.o.
1879 (LIB_SPEC): Suppress -Rlibc.gdb if -static-libc is given.
1880 Do not add -elf2flt or -shared-lib-id options here.
1881 (LINK_SPEC): Define. Pass -elf2flt if no -elf2flt option is given.
1882 Pass -shared-lib-id if -mid-shared-library, taking the library
1883 identifier from -mshared-library-id if given, otherwise
1884 defaulting to 0.
1885 (EH_FRAME_IN_DATA_SECTION): Do not undefine.
1886 (INIT_SECTION_ASM_OP, FINI_SECTION_ASM_OP): Likewise.
1887 (TARGET_OS_CPP_BUILTINS): Define __GXX_MERGED_TYPEINFO_NAMES=0
1888 and __GXX_TYPEINFO_EQUALITY_INLINE=0 if -mid-shared-library.
1889 (DRIVER_SELF_SPECS): Map unadorned PIC options to -msep-data.
1890 * config/m68k/t-uclinux (EXTRA_MULTILIB_PARTS): Add crtbegin.o
1891 and crtend.o.
1892 * config/m68k/lb1sf68.asm (PICCALL): Use an lea and pc-relative
1893 jump sequence for ISA A and ISA A+.
1894 (PICJUMP): Likewise.
1896 2007-01-18 Richard Sandiford <richard@codesourcery.com>
1898 * config.gcc (m68k-*-uclinux*): Base the port on the common
1899 and m68k GNU/Linux files rather than on the generic ELF ones.
1900 * config/m68k/uclinux.h (TARGET_VERSION): Override.
1901 (TARGET_OS_CPP_BUILTINS): Use LINUX_TARGET_OS_CPP_BUILTINS.
1903 2007-01-18 Julian Brown <julian@codesourcery.com>
1904 Richard Sandiford <richard@codesourcery.com>
1906 * config/m68k/m68k.h (LONG_DOUBLE_TYPE_SIZE): Make 64-bit on ColdFire.
1907 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
1908 * config/m68k/netbsd-elf.h (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Undefine
1909 before redefining.
1910 * config/m68k/uclinux-oldabi.h (LONG_DOUBLE_TYPE_SIZE): Redefine to
1911 80 unconditionally.
1912 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
1914 2007-01-18 Richard Sandiford <richard@codesourcery.com>
1916 * doc/install.texi: Document m68k-uclinuxoldabi.
1917 * config.gcc (m68k-*-uclinuxoldabi): New configuration.
1918 * config/m68k/uclinux-oldabi.h: New file, copied from
1919 config/m68k/uclinux.h.
1921 2007-01-18 Kaz Kojima <kkojima@rr.iij4u.or.jp>
1923 * config/m32r/m32r-protos.h (m32r_expand_epilogue): Declare it.
1924 * config/m32r/m32r.c (m32r_setup_incoming_varargs): Use gen_frame_mem.
1925 (m32r_compute_frame_size): Use unsigned for regno.
1926 (m32r_reload_lr): Use gen_frame_mem.
1927 (pop): New.
1928 (m32r_output_function_epilogue): Don't output the function epilogue
1929 textually here.
1930 (m32r_expand_epilogue): New.
1931 (direct_return): Return false if the function has the interrupt
1932 attribute.
1933 (m32r_hard_regno_rename_ok): Remove code for the textual epilogue.
1934 * config/m32r/m32r.md (epilogue): New expander.
1935 (return_lr, return_rte): New insns.
1936 (return): Make it expander.
1937 (return_normal): New expander.
1939 2007-01-18 Josh Conner <jconner@apple.com>
1941 PR target/30485
1942 * config/rs6000/rs6000.c (rs6000_emit_vector_compare): Add
1943 support for UNLE, UNLT, UNGE, and UNGT.
1945 2007-01-18 Jan Hubicka <jh@suse.cz>
1947 * tree-vrp.c (finalize_jump_threads): Do not call cleanup_cfg by hand.
1948 * tree-tailcall (add_virtual_phis): Likewise.
1949 (optimize_tail_call): Return TODOs.
1950 (execute_tail_calls): Return TODOs.
1951 * tree-ssa-ccp (execute_fold_all_builtins): Do cleanup_cfg via TODO.
1952 * tree-cfgcleanup.c (cleanup_tree_cfg_loop): Return if something
1953 changed.
1954 * tree-ssa-forwprop.c (tree_ssa_forward_propagate_single_use_value):
1955 Cleanup cfg using TODO.
1956 * tree-flow.h (cleanup_tree_cfg_loop): Update prototype.
1957 * passes.c (execute_function_todo): When cleanup did something, remove
1958 unused locals.
1959 * tree-cfg.c (pass_build_cfg): Add cleanup_cfg TODO.
1960 (make_edges): Don't cleanup_cfg.
1962 2007-01-18 Uros Bizjak <ubizjak@gmail.com>
1964 * reg-stack.c (subst_stack_regs_pat) [UNSPEC_SINCOS_COS,
1965 UNSPEC_XTRACT_FRACT]: Use generic code for instructions that
1966 operate on the top of stack.
1967 [UNSPEC_SINCOS_SIN, UNSPEC_XTRACT_EXP, UNSPEC_TAN]: Rewrite
1968 register handling of instructions that output to the second
1969 stack slot.
1970 [UNSPEC_TAN_ONE, UNSPEC_TAN_TAN]: Remove.
1971 (move_for_stack_reg): Special-case check for dead destination
1972 stack slot for constant load of 1.0 inside UNSPEC_TAN.
1974 * config/i386/i386.md (UNSPEC_TAN): New constant.
1975 (UNSPEC_TAN_ONE, UNSPEC_TAN_TAN): Remove.
1976 (fptanxf4_i387, fptan_extend<mode>xf4_i387): New patterns
1977 to correctly model move of constant 1.0 to top stack slot.
1978 (*tandf3_1, *tansf3_1, *tanxf3_1): Remove insn patterns.
1979 (unnamed peephole2 pattern): Remove corresponding peephole2
1980 pattern that optimizes tan insn and loading of constant 1.0.
1981 (tanxf2): Use fptanxf4_i387.
1982 (tan<mode>2): Rename from tansf2 and tandf2 and macroize insn
1983 patterns using X87MODEF12 mode macro. Use fptan_extend<mode>xf4_i387
1984 and truncate result to requested mode. Use SSE_FLOAT_MODE_P to
1985 disable patterns for SSE math.
1986 (sincos<mode>3): Use truncxf<mode>2_i387_noop for truncation.
1987 (fyl2x_extend<mode>xf3_i387): Use X87MODEF12 for operand 1.
1989 2007-01-18 Dirk Mueller <dmueller@suse.de>
1990 Richard Guenther <rguenther@suse.de>
1992 PR diagnostic/8268
1993 * doc/invoke.texi (Warray-bounds): Document -Warray-bounds.
1994 * common.opt (Warray-bounds): Add new warning option.
1995 * c-opts.c (c_common_handle_option): Define -Warray-bounds
1996 if -Wall is given.
1997 * Makefile.in: make tree-vrp.o depend on toplev.h
1998 * tree-vrp.c (vrp_finalize): Call check_array_refs if -Warray-bounds
1999 is enabled.
2000 (check_array_refs, check_array_bounds, check_array_ref): New.
2002 2007-01-18 Jan Hubicka <jh@suse.cz>
2004 * tree-ssa-ccp.c (ccp_finalize): Return if something changed.
2005 (execute_ssa_ccp): Return flags conditionally.
2006 * tree-ssa-propagate.c (substitue_and_fold): Return if something was
2007 changed.
2008 * tree-ssa-propagate.h (substitute_and_fold): Update prototype.
2010 2007-01-18 Steven Bosscher <steven@gcc.gnu.org>
2012 * cfgcleanup.c (cleanup_cfg): Detect cfglayout mode and set
2013 the CLEANUP_CFGLAYOUT flag when in cfglayout mode.
2015 * Makefile.c (GTFILES): Add cfglayout.h.
2016 * gengtype.c (open_base_files): Likewise.
2017 * cfglayout.c (cfg_layout_function_footer,
2018 cfg_layout_function_header) Reindent to make gengtype happy.
2019 * cfglayout.h (cfg_layout_function_footer,
2020 cfg_layout_function_header): Add GTY(()) marker.
2022 * ifcvt.c (noce_try_sign_mask): Make sure INSN_B is non-null.
2024 2007-01-18 Ben Elliston <bje@au.ibm.com>
2026 * genautomata.c (write_automata): Include xstrerror output in the
2027 error message if writing the DFA description file fails.
2029 2007-01-17 H.J. Lu <hongjiu.lu@intel.com>
2031 * config/mips/mips-protos.h (mips_output_external): Make it
2032 return void.
2033 * config/mips/iris.h (TARGET_ASM_EXTERNAL_LIBCALL): Removed.
2034 * config/mips/mips.c (irix_output_external_libcall): Likewise.
2035 (extern_list): Likewise.
2036 (extern_head): Likewise.
2037 (TARGET_ASM_FILE_END): Likewise.
2038 (mips_file_end): Likewise.
2039 (mips_output_external): Rewritten.
2041 2007-01-18 Ben Elliston <bje@au.ibm.com>
2043 * genpreds.c (write_insn_preds_c): Only write out the function
2044 body for regclass_for_constraint if we have register constraints.
2046 2007-01-17 Tom Tromey <tromey@redhat.com>
2048 * doc/sourcebuild.texi (libgcj Tests): Use sourceware.org.
2049 * doc/install.texi (Testing): Use sourceware.org.
2050 (Binaries): Likewise.
2051 (Specific): Likewise.
2052 * doc/contrib.texi (Contributors): Use sourceware.org.
2054 2007-01-17 Anatoly Sokolov <aesok@post.ru>
2056 * config/avr/avr.h (AVR_HAVE_LPMX): New macro.
2057 (AVR_ENHANCED): Rename to ...
2058 (AVR_HAVE_MUL): ... new.
2059 (avr_enhanced_p): Rename to ...
2060 (avr_have_mul_p): ... new.
2061 (TARGET_CPU_CPP_BUILTINS): Use 'avr_have_mul_p' instead of
2062 'avr_enhanced_p' for "__AVR_ENHANCED__". Define "__AVR_HAVE_MUL__".
2063 * config/avr/avr.c (avr_enhanced_p): Rename to ...
2064 (avr_have_mul_p): ... new.
2065 (base_arch_s): Rename 'enhanced' to 'have_mul'.
2066 (avr_override_options): Use 'avr_have_mul_p' and 'have_mul' instead of
2067 'avr_enhanced_p' and 'enhanced'.
2068 (ashlhi3_out, ashrhi3_out, lshrhi3_out, avr_rtx_costs): Use
2069 AVR_HAVE_MUL instead of AVR_ENHANCED.
2070 * avr.md (*tablejump_enh): Use AVR_HAVE_LPMX instead of AVR_ENHANCED.
2071 (mulqi3, *mulqi3_enh, *mulqi3_call, mulqihi3, umulqihi3, mulhi3,
2072 *mulhi3_enh, *mulhi3_call, mulsi3, *mulsi3_call): Use AVR_HAVE_MUL
2073 instead of AVR_ENHANCED.
2074 (*tablejump_enh): Use AVR_HAVE_LPMX instead of AVR_ENHANCED.
2075 * libgcc.S: Use __AVR_HAVE_MUL__ instead of __AVR_ENHANCED__.
2076 (__tablejump__): Use __AVR_HAVE_LPMX__ instead of __AVR_ENHANCED__.
2078 2007-01-17 Ian Lance Taylor <iant@google.com>
2080 * vec.h (VEC_reserve_exact): Define.
2081 (vec_gc_p_reserve_exact): Declare.
2082 (vec_gc_o_reserve_exact): Declare.
2083 (vec_heap_p_reserve_exact): Declare.
2084 (vec_heap_o_reserve_exact): Declare.
2085 (VEC_OP (T,A,reserve_exact)): New static inline function, three
2086 versions.
2087 (VEC_OP (T,A,reserve)) [all versions]: Remove handling of
2088 negative parameter.
2089 (VEC_OP (T,A,alloc)) [all versions]: Call ...reserve_exact.
2090 (VEC_OP (T,A,copy)) [all versions]: Likewise.
2091 (VEC_OP (T,a,safe_grow)) [all versions]: Likewise.
2092 * vec.c (calculate_allocation): Add exact parameter. Change all
2093 callers.
2094 (vec_gc_o_reserve_1): New static function, from vec_gc_o_reserve.
2095 (vec_gc_p_reserve, vec_gc_o_reserve): Call vec_gc_o_reserve_1.
2096 (vec_gc_p_reserve_exact, vec_gc_o_reserve_exact): New functions.
2097 (vec_heap_o_reserve_1): New static function, from vec_heap_o_reserve.
2098 (vec_heap_p_reserve, vec_heap_o_reserve): Call vec_heap_o_reserve_1.
2099 (vec_heap_p_reserve_exact): New function.
2100 (vec_heap_o_reserve_exact): New function.
2102 2007-01-17 Jan Hubicka <jh@suse.cz>
2104 * ipa-type-escape.c (look_for_casts): Revamp using handled_component_p.
2106 2007-01-17 Eric Christopher <echristo@apple.com>
2108 * config.gcc: Support core2 processor.
2110 2007-01-16 Jan Hubicka <jh@suse.cz>
2112 * tree-ssanames.c (release_dead_ssa_names): Instead of ggc_freeing
2113 the names, just unlink the chain so we don't crash on dangling pointers
2114 to dead SSA names.
2116 2007-01-16 Jan Hubicka <jh@suse.cz>
2118 * cgraph.h (cgraph_decide_inlining_incrementally): Kill.
2119 * tree-pass.h: Reorder to make IPA passes appear toegher.
2120 (pass_early_inline, pass_inline_parameters, pass_apply_inline): Declare.
2121 * cgraphunit.c (cgraph_finalize_function): Do not compute inling
2122 parameters, do not call early inliner.
2123 * ipa-inline.c: Update comments. Include tree-flow.h
2124 (cgraph_decide_inlining): Do not compute inlining parameters.
2125 (cgraph_decide_inlining_incrementally): Return TODOs; assume to
2126 be called with function context set up.
2127 (pass_ipa_inline): Remove unreachable functions before pass.
2128 (cgraph_early_inlining): Simplify assuming to be called from the
2129 PM as local pass.
2130 (pass_early_inline): New pass.
2131 (cgraph_gate_ipa_early_inlining): New gate.
2132 (pass_ipa_early_inline): Turn into simple wrapper.
2133 (compute_inline_parameters): New function.
2134 (gate_inline_passes): New gate.
2135 (pass_inline_parameters): New pass.
2136 (apply_inline): Move here from tree-optimize.c
2137 (pass_apply_inline): New pass.
2138 * ipa.c (cgraph_remove_unreachable_nodes): Verify cgraph after
2139 transforming.
2140 * tree-inline.c (optimize_inline_calls): Return TODOs rather than
2141 doing them by hand.
2142 (tree_function_versioning): Do not allocate dummy struct function.
2143 * tree-inline.h (optimize_inline_calls): Update prototype.
2144 * tree-optimize.c (execute_fixup_cfg): Export.
2145 (pass_fixup_cfg): Remove
2146 (tree_rest_of_compilation): Do not apply inlines.
2147 * tree-flow.h (execute_fixup_cfg): Declare.
2148 * Makefile.in (gt-passes.c): New.
2149 * passes.c: Include gt-passes.h
2150 (init_optimization_passes): New passes.
2151 (nnodes, order): New static vars.
2152 (do_per_function_toporder): New function.
2153 (execute_one_pass): Dump current pass here.
2154 (execute_ipa_pass_list): Don't dump current pass here.
2156 2007-01-16 Janis Johnson <janis187@us.ibm.com>
2158 * config/dfp-bit.c (dfp_compare_op): Return separate value for NaN.
2159 (DFP_NE, DFP_LE, DFP_GE): Return false for NaN.
2161 2007-01-16 David Edelsohn <edelsohn@gnu.org>
2163 * config/rs6000/darwin-ldouble.c: Build file for SOFT_FLOAT.
2164 (strong_alias): Define.
2165 (__gcc_qmul): Provide non-FMA for soft-float.
2166 (__gcc_qdiv): Same.
2167 (__gcc_qneg): New.
2168 (__gcc_qeq): New.
2169 (__gcc_qle): New.
2170 (__gcc_qge): New.
2171 (__gcc_qunord): New.
2172 (__gcc_stoq): New.
2173 (__gcc_dtoq): New.
2174 (__gcc_qtos): New.
2175 (__gcc_qtod): New.
2176 (__gcc_qtoi): New.
2177 (__gcc_qtou): New.
2178 (__gcc_itoq): New.
2179 (__gcc_utoq): New.
2180 (fmsub): New.
2181 * config/rs6000/rs6000.c (rs6000_init_libfuncs): Initialize
2182 soft-float functions.
2183 * config/rs6000/libgcc-ppc-glibc.ver: Version soft-float symbols.
2184 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Do not warn
2185 about long double soft float.
2187 2007-01-16 Dorit Nuzman <dorit@il.ibm.com>
2188 Tehila Meyzels <tehila@il.ibm.com>
2190 * tree-vectorizer.h (is_pattern_stmt_p): New.
2191 * tree-vect-analyze.c (vect_determine_vectorization_factor): Fix
2192 formatting (tabs instead of spaces). Cleanup and clarify setting
2193 of STMT_VINFO_VECTYPE. Call is_pattern_stmt_p.
2194 * tree-vect-transform.c (vect_get_vec_def_for_operand): Fix typo.
2195 (vectorizable_type_demotion): Check that types are integral.
2196 (vectorizable_type_promotion): Likewise.
2197 (vectorizable_store): Fix typo. Eliminate new-line at end of
2198 comments.
2200 2007-01-16 Jan Hubicka <jh@suse.cz>
2202 * tree-ssanames.c (release_dead_ssa_names): Remove invalidated
2203 cgraph edges too.
2205 2007-01-15 Eric Christopher <echristo@apple.com>
2207 * ifcvt.c: Include vec.h, vecprim.h.
2208 (check_cond_move_block): New argument regs.
2209 Reorganize. Add registers used to regs.
2210 (cond_move_process_if_block): Use regs set above as
2211 loop bounds.
2213 2007-01-15 Eric Christopher <echristo@apple.com>
2215 * config/darwin.h: Update copyright.
2216 (TARGET_OPTION_TRANSLATE_TABLE): Add umbrella.
2217 (LINK_COMMAND_SPEC): Add -u.
2218 (LINK_SPEC): Fix umbrella for above.
2220 2007-01-15 Joseph S. Myers <joseph@codesourcery.com>
2222 * config/soft-fp/op-common.h, config/soft-fp/op-4.h: Update from
2223 glibc CVS.
2225 2007-01-15 Tom Tromey <tromey@redhat.com>
2227 * doc/sourcebuild.texi (libgcj Tests): Don't mention jacks.
2228 * doc/install.texi (Testing): Don't mention jacks.
2229 (Configuration): Document --enable-java-maintainer-mode. Move
2230 --with-java-home to libgcj-specific section. Document
2231 --with-ecj-jar.
2232 (Prerequisites): Mention --enable-java-maintainer-mode, ecj1.
2234 2007-01-15 Jan Hubicka <jh@suse.cz>
2236 * tree-ssa-dce.c (DCE_TODOs): New.
2237 (propagate_necessity): Return if something changed.
2238 (eliminate_unnecessary_stmts): Likewise.
2239 (perform_tree_ssa_dce): Return TODO flags when needed.
2240 (pass_dce, pass_dce_loop, pass_cd_dce): Remove TODO flags.
2242 2007-01-15 Uros Bizjak <ubizjak@gmail.com>
2244 * config/i386/i386.md (fyl2xxf3_i387): Rename from fyl2x_xf3.
2245 (fyl2x_extend<mode>xf3_i387): New insn pattern.
2246 (log<mode>2): Rename from logsf2 and logdf2 and macroize insn
2247 patterns using X87MODEF12 mode macro. Extend operand 1
2248 to XFmode. Use SSE_FLOAT_MODE_P to disable patterns for SSE math.
2249 (log10<mode>2): Ditto.
2250 (log2<mode>2): Ditto.
2251 (log1p<mode>2): Ditto.
2252 (logb<mode>2): Ditto.
2253 (fyl2xp1xf3_i387): Rename from fyl2xp1_xf3.
2254 (fyl2xp1_extend<mode>xf3_i387): New insn pattern.
2255 (*fxtractxf3_i387): Rename from *fxtractxf3.
2256 (fxtract_extend<mode>xf3_i387): New insn pattern.
2257 (ilogbsi2): Use match_dup 3, not match_operand:XF 3.
2258 * config/i386/i386.c (ix86_emit_i387_log1p): Use gen_fyl2xp1xf3_i387()
2259 and gen_fyl2xxf3_i387().
2261 2007-01-14 Zdenek Dvorak <dvorakz@suse.cz>
2263 * loop-unswitch.c (unswitch_loop): Do not call fix_loop_placement.
2264 * cfgloopmanip.c (fix_loop_placement): Made static. Use
2265 get_loop_exit_edges. Changed return type to bool.
2266 * cfgloop.h (fix_loop_placement): Declaration removed.
2268 2007-01-14 Dorit Nuzman <dorit@il.ibm.com>
2270 * param.h (MIN_VECT_LOOP_BOUND): New.
2271 * params.def (MIN_VECT_LOOP_BOUND): New.
2272 * tree-vectorizer.c (slpeel_tree_peel_loop_to_edge): Takes another
2273 argument - minimum threshold for number of iterations.
2274 * tree-vectorizer.h (slpeel_tree_peel_loop_to_edge): Add another
2275 argument to declaration.
2276 * tree-vect-analyze.c (vect_analyze_operations): Check value of
2277 MIN_VECT_LOOP_BOUND.
2278 * tree-vect-transform.c (vect_do_peeling_for_loop_bound): Call
2279 slpeel_tree_peel_loop_to_edge with additional argument.
2280 (vect_do_peeling_for_alignment): Likewise.
2281 * doc/invoke.texi (min-vect-loop-bound): Document new param option.
2283 2007-01-14 Uros Bizjak <ubizjak@gmail.com>
2285 PR target/30413
2286 * config/i386/i386.c (print_operand) ['z']: Output 'b' for
2287 operands of size 1.
2289 2007-01-14 Jan Hubicka <jh@suse.cz>
2291 * tree-dfa.c (remove_referenced_var): New function.
2292 * tree-ssa-live.c (remove_unused_locals): Walk referenced vars and
2293 prune referenced vars list too.
2294 * tree-flow.h (remove_referenced_var): Declare.
2296 2007-01-14 Jan Hubicka <jh@suse.cz>
2298 * tree-eh.c (add_stmt_to_eh_region_fn): Do not add call_exprs
2299 separately.
2300 (remove_stmt_from_eh_region_fn): Do not remove call_exprs.
2301 (verify_eh_throw_stmt_node, verify_eh_throw_table_statements): Kill.
2302 * except.h (verify_eh_throw_table_statements): Kill prototype.
2303 * cfgexpand.c (expand_gimple_basic_block): Propagate Eh regions
2304 into call exrepssions.
2305 * tree-optimize.c (execute_free_cfg_annotatiosn): Do not call
2306 eh trhow verifier.
2307 * tree-cfg.c: Include pointer-set.h.
2308 (verify_node_sharing): Work on pointer set.
2309 (verify_eh_throw_stmt_node): New.
2310 (verify_stmts): Use pointers sets, verify throw_stmt.
2312 2007-01-13 Zdenek Dvorak <dvorakz@suse.cz>
2314 * ipa-reference.c (analyze_function): Consider also addresses taken
2315 in phi nodes.
2317 2007-01-12 Roger Sayle <roger@eyesopen.com>
2319 * c-typeck.c (null_pointer_constant_p): Replace use of
2320 TREE_CONSTANT_OVERFLOW with TREE_OVERFLOW.
2321 (build_c_cast): Likewise.
2323 2007-01-12 Roger Sayle <roger@eyesopen.com>
2325 * tree.h (force_fit_type_double): Remove unused final argument.
2326 * c-common.c (constant_expression_warning): Replace use of
2327 TREE_CONSTANT_OVERFLOW with TREE_OVERFLOW.
2328 (convert_and_check): Likewise.
2329 (shorten_compare): Update call to force_fit_type_double.
2330 (c_common_truthvalue_conversion) <INTEGER_CST>: Use integer_zerop.
2331 * convert.c (convert_to_pointer): Update call to
2332 force_fit_type_double.
2333 * fold-const.c (force_fit_type_double): Remove overflowed_const
2334 argument.
2335 (int_const_binop, fold_convert_const_int_from_int,
2336 fold_convert_const_int_from_real, fold_div_compare,
2337 fold_sign_changed_comparison, fold_unary, fold_negate_const,
2338 fold_abs_const, fold_not_const): Remove the final argument from
2339 calls to force_fit_type_double.
2341 2007-01-12 Andrew Pinski <andrew_pinski@playstation.sony.com>
2343 * configure.ac: Set insn to "nop" for spu-*-* also.
2344 * configure: Regenerate.
2346 2007-01-12 Olga Golovanevsky <olga@il.ibm.com>
2348 * builtins.def : Add BUILT_IN_FREE.
2350 2007-01-12 Jan Hubicka <jh@suse.cz>
2352 PR tree-optimization/30443
2353 * tree-inline.c (tree_function_versioning): Do not optimize when
2354 cloning for inlining.
2356 2007-01-12 Zdenek Dvorak <dvorakz@suse.cz>
2358 * doc/loop.texi: Document recording of loop exits.
2359 * cfgloopmanip.c (loopify, duplicate_loop): Use alloc_loop.
2360 (update_single_exits_after_duplication,
2361 update_single_exit_for_duplicated_loop,
2362 update_single_exit_for_duplicated_loops): Removed.
2363 (duplicate_loop_to_header_edge): Do not call
2364 update_single_exits_after_duplication and
2365 update_single_exit_for_duplicated_loops.
2366 (loop_version): Do not update single_exit information.
2367 (fix_loop_structure): Use record_loop_exits instead of
2368 mark_single_exit_loops.
2369 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update
2370 the lists of loop exits.
2371 * cfghooks.c (redirect_edge_and_branch, redirect_edge_and_branch_force,
2372 split_edge, merge_blocks): Update the lists of loop exits.
2373 * modulo-sched.c (sms_schedule): Pass LOOPS_HAVE_RECORDED_EXITS to
2374 loop_optimizer_init.
2375 * loop-init.c (loop_optimizer_init): Call record_loop_exits instead
2376 of mark_single_exit_loops.
2377 (loop_optimizer_finalize): Call release_recorded_exits.
2378 * tree-ssa-loop.c (tree_loop_optimizer_init): Pass
2379 LOOPS_HAVE_RECORDED_EXITS to loop_optimizer_init.
2380 * tree-vectorizer.c (slpeel_tree_duplicate_loop_to_edge_cfg): Do not
2381 update single exit information.
2382 * lambda-code.c (perfect_nestify): Ditto.
2383 * cfgloop.c (flow_loop_free): Destroy the list of exits of the loop.
2384 (mark_single_exit_loops): Removed.
2385 (alloc_loop, loop_exit_hash, loop_exit_eq, loop_exit_free,
2386 get_exit_descriptions, rescan_loop_exit, record_loop_exits,
2387 dump_recorded_exit, dump_recorded_exits, release_recorded_exits): New
2388 functions.
2389 (get_loop_exit_edges, single_exit): Use recorded exit lists.
2390 (add_bb_to_loop, remove_bb_from_loops): Update the lists of loop exits.
2391 (verify_loop_structure): Verify consistency of the exit lists.
2392 (flow_loops_find): Use alloc_loop. Initialize exits hash.
2393 (set_single_exit): Removed.
2394 * cfgloop.h (struct loop_exit): New function.
2395 (struct loop): single_exit_ field replaced by exits field.
2396 (LOOPS_HAVE_MARKED_SINGLE_EXITS): Replaced by LOOPS_HAVE_RECORDED_EXITS.
2397 (struct loops): Added exits hash.
2398 (mark_single_exit_loops, set_single_exit): Declaration removed.
2399 (release_recorded_exits, record_loop_exits, rescan_loop_exit): Declare.
2401 2007-01-12 Richard Sandiford <richard@codesourcery.com>
2403 * doc/invoke.texi: Avoid use of @headitem.
2405 2007-01-12 Richard Sandiford <richard@codesourcery.com>
2407 * cse.c (cse_insn): Move HAVE_CC0 code after declarations.
2409 2007-01-12 Richard Sandiford <richard@codesourcery.com>
2411 * doc/install.texi: Fix m68k-*-* anchor and add m68k-*-* to the
2412 list of targets.
2414 2007-01-12 Nathan Sidwell <nathan@codesourcery.com>
2415 Richard Sandiford <richard@codesourcery.com>
2417 * doc/invoke.texi: Document -mno-bitfield, -mno-rtd and -mno-short.
2418 * config/m68k/m68k.opt: Resort options.
2419 (mbitfield, mrtd, mshort): Remove RejectNegative properties.
2421 2007-01-12 Nathan Sidwell <nathan@codesourcery.com>
2422 Richard Sandiford <richard@codesourcery.com>
2424 * doc/invoke.texi: Document the macros that are defined by
2425 m68k's -mtune and -mhard-float options.
2426 * config/m68k/m68k-protos.h (m68k_cpp_cpu_ident) Declare.
2427 (m68k_cpp_cpu_family): Likewise.
2428 * config/m68k/m68k.h (TARGET_CPU_CPP_BUILTINS): Add a full set
2429 of __ucfv*__ macros. Define __mcffpu__ if generating code for
2430 ColdFire FPUs. Define __mcf_cpu_* and __mcf_family_* macros.
2431 * config/m68k/m68k.c (m68k_cpp_cpu_ident): New function.
2432 (m68k_cpp_cpu_family): Likewise.
2434 2007-01-12 Richard Sandiford <richard@codesourcery.com>
2436 * config/m68k/m68k.h (TARGET_CPU_CPP_BUILTINS): Treat all mc68*
2437 macros besides mc68000 as tuning macros. Use a switch statement
2438 to set them and mcpu32.
2440 2007-01-12 Julian Brown <julian@codesourcery.com>
2442 * config/m68k/m68k.h: Use TARGET_68040 instead of TARGET_68040_ONLY.
2443 (TARGET_68040_ONLY): Rename to...
2444 (TARGET_68040): ...this.
2445 * config/m68k/m68k.c: Use TARGET_68040 instead of TARGET_68040_ONLY.
2446 * config/m68k/m68k.md: Likewise.
2448 2007-01-12 Julian Brown <julian@codesourcery.com>
2449 Nathan Sidwell <nathan@codesourcery.com>
2450 Richard Sandiford <richard@codesourcery.com>
2452 * config.gcc (m680[012]0-*-*, m68k*-*-*): Set m68k_cpu_ident to
2453 the -mcpu= argument associated with the --with-cpu setting.
2454 Define M68K_DEFAULT_TUNE to the default -mtune= option,
2455 if different from the one implied by the -mcpu setting.
2456 Accept --with-cpu=FOO if FOO is listed in m68k-devices.def,
2457 using mcpu=FOO as the default CPU option. Set target_cpu_default2.
2458 * doc/invoke.texi: Mention ColdFire in the introduction to the
2459 m68k options. Document the new -march, -mcpu, -mtune, -mdiv,
2460 -mno-div and -mhard-float options. Make -m68881 a synonym for
2461 -mhard-float. Document the previously-undocumented -m5206e,
2462 -m528x, -m5307 and -m5407 options. Tweak the existing option
2463 documentation for consistency.
2464 * doc/install.texi: Mention new --with-cpu arguments.
2465 * config/m68k/m68k.h (OPTION_DEFAULT_SPECS): Only use the
2466 default CPU if neither -mcpu nor -march are specified.
2467 (ASM_CPU_SPEC): Pass down -mcpu and -march options.
2468 (TARGET_CPU_CPP_BUILTINS): Set __mcfisa*__ macros from
2469 TARGET_ISA*. Set the legacy __mcf*__ cpu macros in the same way,
2470 using m68k_tune to decide between families that implement the
2471 same ISA. Use m68k_tune to set __mcfv4e__.
2472 (FL_BITFIELD, FL_68881, FL_COLDFIRE, FL_CF_HWDIV, FL_CF_MAC)
2473 (FL_CF_EMAC, FL_CF_EMAC_B, FL_CF_USP, FL_CF_FPU, FL_ISA_68000)
2474 (FL_ISA_68010, FL_ISA_68020, FL_ISA_68040, FL_ISA_A, FL_ISA_B)
2475 (FL_ISA_C, FL_ISA_MMU): New macros.
2476 (MASK_COLDFIRE): Delete.
2477 (TARGET_68010, TARGET_68020, TARGET_68040_ONLY, TARGET_COLDFIRE)
2478 (TARGET_ISAB): Redefine in terms of m68k_cpu_flags.
2479 (TARGET_68881, TARGET_COLDFIRE_FPU): Redefine in terms of m68k_fpu.
2480 (TARGET_HARD_FLOAT): Do not define here.
2481 (TARGET_ISAAPLUS, TARGET_ISAC): New macros.
2482 (TUNE_68000): New macro.
2483 (TUNE_68000_10): Redefine in terms of TUNE_68000 and TUNE_68010.
2484 (TUNE_68010, TUNE_68030, TUNE_68040, TUNE_68060, TUNE_CPU32)
2485 (TUNE_CFV2): Redefine in terms of m68k_tune.
2486 (uarch_type, target_device, fpu_type): New enums.
2487 (m68k_cpu, m68k_tune, m68k_fpu, m68k_cpu_flags): Declare.
2488 * config/m68k/m68k.c (TARGET_DEFAULT): Remove MASK_68881.
2489 (FL_FOR_isa_00, FL_FOR_isa_10, FL_FOR_isa_20, FL_FOR_isa_40)
2490 (FL_FOR_isa_cpu32, FL_FOR_isa_a, FL_FOR_isa_aplus, FL_FOR_isa_b)
2491 (FL_FOR_isa_c): New macros.
2492 (m68k_isa): New enum.
2493 (m68k_target_selection): New structure.
2494 (all_devices, all_isas, all_microarchs): New tables.
2495 (m68k_cpu_entry, m68k_arch_entry, m68k_tune_entry, m68k_cpu)
2496 (m68k_tune, m68k_fpu, m68k_cpu_flags): New variables.
2497 (MASK_ALL_CPU_BITS): Delete.
2498 (m68k_find_selection): New function.
2499 (m68k_handle_option): Handle -mcpu=, -march= and -mtune=.
2500 Map the legacy target options to a combination of the new ones.
2501 (override_options): Set m68k_cpu, m68k_tune, m68k_fpu and
2502 m68k_cpu_flags. Handle M68K_DEFAULT_TUNE. Use m68k_cpu_flags
2503 to derive default MASK_BITFIELD, MASK_CF_HWDIV and MASK_HARD_FLOAT
2504 settings.
2505 * config/m68k/m68k.opt (m5200, m5206e, m528x, m5307, m5407, mcfv4e)
2506 (m68010, m68020, m68020-40, m68020-60, m68030, m68040): Remove Mask
2507 properties.
2508 (m68881, msoft-float): Change mask from 68881 to HARD_FLOAT.
2509 (march=, mcpu=, mdiv, mhard-float, mtune=): New options.
2510 * config/m68k/m68k-devices.def: New file.
2512 2007-01-12 Richard Sandiford <richard@codesourcery.com>
2513 Nathan Sidwell <nathan@codesourcery.com>
2515 * config/m68k/m68k.h (ASM_CPU_SPEC, ASM_SPEC, EXTRA_SPECS)
2516 (SUBTARGET_EXTRA_SPECS): New macros.
2517 * config/m68k/linux.h (ASM_SPEC): Remove CPU flags;
2518 use %(asm_cpu_spec) instead.
2519 * config/m68k/m68k-none.h (ASM_SPEC): Likewise.
2520 * config/m68k/openbsd.h (ASM_SPEC): Likewise.
2521 * config/m68k/netbsd-elf.h (ASM_SPEC): Likewise.
2522 (EXTRA_SPECS): Rename to...
2523 (SUBTARGET_EXTRA_SPECS): ...this.
2525 2007-01-12 Nathan Sidwell <nathan@codesourcery.com>
2526 Richard Sandiford <richard@codesourcery.com>
2527 Julian Brown <julian@codesourcery.com>
2529 * config.gcc (m68k-*-aout*, m68k-*-coff*, m68020-*-elf*, m68k-*-elf*)
2530 (m68k-*-uclinux*, m68k-*-linux*, m68k-*-rtems*): Set default_m68k_cpu
2531 to the configuration's default CPU.
2532 (m68010-*-netbsdelf*, m68k*-*-netbsdelf*, m68k*-*-openbsd*): Likewise.
2533 Remove default masks.
2534 (m680[012]0-*-*): Set the default with_cpu to the first part of
2535 the target name.
2536 (m68k*-*-*): Set the default with_cpu to m$default_m68k_cpu.
2537 (m68k*-*-linux): Extend the --with-cpu handling to...
2538 (m680[012]0-*-*, m68k*-*-*): ...these configurations. Allow m68000
2539 and m68010. Don't set target_cpu_default2.
2540 * doc/install.texi: Document --with-cpu for m68k.
2541 * config/m68k/m68k.h (OPTION_DEFAULT_SPECS): Define.
2542 * config/m68k/m68k-none.h (TARGET_CPU_DEFAULT, M68K_CPU_m68k)
2543 (M68K_CPU_m68000, M68K_CPU_m68010, M68K_CPU_m68020, M68K_CPU_m68030)
2544 (M68K_CPU_m68040, M68K_CPU_m68302, M68K_CPU_m68332, TARGET_DEFAULT)
2545 (ASM_CPU_DEFAULT_SPEC, CC1_CPU_DEFAULT_SPEC): Delete.
2546 (ASM_SPEC): Remove use of %(asm_cpu_default).
2547 (EXTRA_SPECS, SUBTARGET_EXTRA_SPECS, MULTILIB_DEFAULTS): Delete.
2548 * config/m68k/linux.h (TARGET_DEFAULT): Delete.
2549 (CPP_SPEC): Merge definitions. Do not handle __HAVE_68881__ here.
2550 * config/m68k/netbsd-elf.h (TARGET_OS_CPP_BUILTINS): Define
2551 __HAVE_FPU__ if TARGET_HARD_FLOAT.
2552 (TARGET_DEFAULT): Delete.
2553 (EXTRA_SPECS): Delete cpp_cpu_default_spec, cpp_cpu_spec,
2554 cpp_fpu_spec, asm_default_spec and netbsd_cpp_spec.
2555 (CPP_CPU_SPEC): Delete.
2556 (TARGET_VERSION): Merge definitions, using TARGET_68010 to pick
2557 the appropriate string.
2558 (CPP_CPU_DEFAULT_SPEC, ASM_DEFAULT_SPEC, CPP_FPU_SPEC): Delete.
2559 (CPP_SPEC): Define to NETBSD_CPP_SPEC.
2560 (ASM_SPEC): Don't use %(asm_default_spec).
2561 * config/m68k/m68k.c (TARGET_DEFAULT_TARGET_FLAGS): Remove
2562 TARGET_DEFAULT and add MASK_68881.
2563 * config/m68k/m68k.md: Remove mention of TARGET_DEFAULT from comments.
2565 2007-01-12 Richard Sandiford <richard@codesourcery.com>
2567 * config.gcc (m68010-*-netbsdelf*): Add MASK_68010.
2568 (m68k*-*-netbsdelf*, m68k*-*-openbsd*, m68k*-linux*): Add
2569 MASK_68010 alongside MASK_68020.
2570 * doc/invoke.texi: Document -m68010.
2571 * config/m68k/m68k.opt (m68010): New.
2572 * config/m68k/m68k.h (TARGET_CPU_CPP_BUILTINS): Define mc68010
2573 if TUNE_68010.
2574 (TUNE_68010): New macro.
2575 * config/m68k/m68k-none.h (M68K_CPU_m68k, M68K_CPU_m68010)
2576 (M68K_CPU_m68020, M68K_CPU_m68030, M68K_CPU_m68040)
2577 (M68K_CPU_m68332): Add MASK_68010.
2578 * config/m68k/linux.h (TARGET_DEFAULT): Add MASK_68010 to
2579 fallback definition.
2580 * config/m68k/netbsd-elf.h (CPP_CPU_SPEC): Remove now-redundant
2581 defines.
2582 * config/m68k/m68k.c (MASK_ALL_CPU_BITS): Add MASK_68010.
2583 (m68k_handle_option): Handle OPT_m68010. Add MASK_68010
2584 to all entries that use MASK_68020.
2585 (output_move_simode_const, output_move_himode, output_move_qimode)
2586 (output_move_stricthi, output_move_strictqi): Use TARGET_68010
2587 instead of TARGET_68020 to select clr behavior. Remove comment
2588 about there being no TARGET_68010.
2589 * config/m68k/m68k.md: Likewise throughout.
2591 2007-01-12 Julian Brown <julian@codesourcery.com>
2593 * config/m68k/m68k.h (TARGET_ISAB): New macro.
2594 * config/m68k/m68k.c: Use TARGET_ISAB rather than TARGET_CFV4.
2595 * config/m68k/m68k.md: Likewise.
2597 2007-01-12 Julian Brown <julian@codesourcery.com>
2599 * config/m68k/m68k.h (LEGITIMATE_INDEX_P, LEGITIMIZE_ADDRESS): Use
2600 TARGET_COLDFIRE_FPU instead of TARGET_CFV4E.
2602 2007-01-12 Julian Brown <julian@codesourcery.com>
2604 * config/m68k/m68k.h (TUNE_68040_60): New macro.
2605 * config/m68k/m68k.c (standard_68881_constant_p): Use it.
2606 * config/m68k/m68k.md: Likewise.
2608 2007-01-12 Julian Brown <julian@codesourcery.com>
2609 Richard Sandiford <richard@codesourcery.com>
2611 * config/m68k/m68k.h (TARGET_CPU_CPP_BUILTINS): Use TUNE_68030
2612 instead of TARGET_68030, TUNE_68040 instead of TARGET_68040,
2613 TUNE_68060 instead of TARGET_68060 and TUNE_CPU32 instead of
2614 TARGET_CPU32.
2615 (TARGET_CPU32): Rename to...
2616 (TUNE_CPU32): ...this.
2617 (TUNE_68000_10, TUNE_68030, TUNE_68040, TUNE_68060)
2618 (TUNE_CFV2): New macros.
2619 * config/m68k/netbsd-elf.h (LONG_DOUBLE_TYPE_SIZE): Simplify;
2620 remove conditions that are implied by TARGET_68020.
2621 * config/m68k/m68k.c (m68k_output_function_prologue): Use TUNE_68040
2622 instead of TARGET_68040 and TUNE_CPU32 instead of TARGET_CPU32.
2623 (m68k_output_function_epilogue): Likewise.
2624 (m68k_rtx_costs): Likewise. Use TUNE_68060 instead of TARGET_68060
2625 and TUNE_CFV2 instead of TARGET_5200. Use TUNE_68000_10 instead of
2626 "!TARGET_68020 && !TARGET_COLDFIRE" to choose between 68000 and
2627 non-68000 timings. Refactor multiplication and division costs.
2628 (output_addsi3): Use TUNE_68040 instead of TARGET_68040 and
2629 TUNE_CPU32 instead of TARGET_CPU32.
2630 (standard_68881_constant_p): Use TUNE_68040 instead of TARGET_68040
2631 and TUNE_68060 instead of TARGET_68060.
2632 * config/m68k/m68k.md: Use TUNE_68040 instead of TARGET_68040,
2633 TUNE_68060 instead of TARGET_68060, and TUNE_CPU32 instead of
2634 TARGET_CPU32.
2635 (movsi_const0): Use TUNE_68000_10 rather than "!TARGET_68020
2636 && !TARGET_COLDFIRE" to choose between moveq and clr.
2637 Likewise in the unnamed movsf pattern.
2638 (ashlsi_17_24, lshrsi_17_24): Guard with TUNE_68000_10 rather than
2639 "!TARGET_68020 && !TARGET_COLDFIRE". Likewise the unnamed
2640 ashiftrt pattern.
2642 2007-01-12 Richard Sandiford <richard@codesourcery.com>
2644 * config/m68k/m68k.h (TARGET_CPU_CPP_BUILTINS): Increase amount
2645 of tabbing before backslashes.
2647 2007-01-11 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
2649 * pa-linux.h (ASM_OUTPUT_INTERNAL_LABEL): Undefine.
2650 * pa.h (ASM_OUTPUT_LABEL): Output colon when using GAS.
2651 (ASM_OUTPUT_INTERNAL_LABEL): Define.
2653 2007-01-11 Zdenek Dvorak <dvorakz@suse.cz>
2655 * tree-ssa-loop-ivopts.c (extract_cond_operands): Split from
2656 find_interesting_uses_cond.
2657 (find_interesting_uses_cond): Use extract_cond_operands.
2658 (rewrite_use_compare): Use extract_cond_operands and
2659 force_gimple_operand_bsi. Do not call update_stmt.
2660 (determine_use_iv_cost_condition): Use extract_cond_operands.
2661 Return cheaper of using original bound and changing the exit bound.
2663 2007-01-11 Zdenek Dvorak <dvorakz@suse.cz>
2665 PR tree-optimization/29516
2666 * tree-ssa-address.c (tree_mem_ref_addr, add_to_parts,
2667 most_expensive_mult_to_index, addr_to_parts,
2668 create_mem_ref, maybe_fold_tmr): Make the type of
2669 fields of TARGET_MEM_REF sizetype.
2670 (move_fixed_address_to_symbol, move_pointer_to_base):
2671 New functions.
2672 * tree.def (TARGET_MEM_REF): Add comment on types of
2673 the operands.
2675 2007-01-11 Joseph Myers <joseph@codesourcery.com>
2677 * c-common.c (vector_types_convertible_p): Treat opaque types as
2678 always convertible if they have the same size, but not otherwise.
2680 2007-01-11 Steven Bosscher <steven@gcc.gnu.org>
2682 * ifcvt.c (struct noce_if_info): Add comments to the fields.
2683 Remove the b_unconditional field.
2684 (noce_try_sign_mask): Do not look at b_unconditional.
2685 (noce_process_if_block): Do not use merge_if_blocks. Update
2686 the CFG here. Do not set b_unconditional.
2687 (cond_move_process_if_block): Likewise.
2688 (find_cond_trap): Likewise.
2689 (check_cond_move_block): Require simple jump insns at the end
2690 of the basic block.
2692 2007-01-11 Jan Hubicka <jh@suse.cz>
2694 PR tree-optimization/1046
2695 * tree-tailcall.c (suitable_for_tail_call_opt_p): Use TREE_ADDRESSABLE
2696 when alias info is not ready.
2697 (pass_tail_recursion): Do not require aliasing.
2698 * tree-ssa-copyrename.c (pass_rename_ssa_cop): Likewise.
2699 * tree-ssa-ccp.c (pass_ccp, pass_fold_builtins): Likewise.
2700 * tree-ssa-copy.c (pass_copy_prop): Likewise.
2701 * tree-ssa-forwprop.c (pass_forwprop): Likewise.
2702 * tree-ssa-dce.c (pass_dce, pass_dce_loop, pass_cd_dce): Likewise.
2703 * passes.c (init_optimization_passes): Execute rename_ssa_copies,
2704 ccp, forwprop, copy_prop, merge_phi, copy_prop, dce and tail recursion
2705 before inlining.
2706 * tree-ssa-operands.c (add_virtual_operand, get_indirect_ref_operand):
2707 When aliasing is not build, mark statement as volatile.
2709 2007-01-11 Tom Tromey <tromey@redhat.com>
2711 PR preprocessor/15185, PR preprocessor/20989:
2712 * doc/cppopts.texi <-MT>: Update description of algorithm for
2713 computing default target.
2714 <-M, -MD>: Reword "basename" text.
2716 2007-01-11 Roger Sayle <roger@eyesopen.com>
2718 * builtins.c (expand_builtin_pow, expand_builtin_powi,
2719 fold_builtin_cabs, fold_builtin_sqrt, fold_builtin_trunc,
2720 fold_builtin_floor, fold_builtin_ceil, fold_builtin_round,
2721 fold_builtin_int_int_roundingfn, fold_builtin_bitop,
2722 fold_builtin_bswap, real_constp, fold_builtin_pow,
2723 fold_builtin_powi, fold_builtin_signbit, fold_builtin_copysign,
2724 do_mpfr_arg1, do_mpfr_arg2, do_mpfr_arg3, do_mpfr_sincos): Replace
2725 uses of the macro TREE_CONSTANT_OVERFLOW with TREE_OVERFLOW.
2726 * convert.c (convert_to_pointer): Likewise.
2727 * expr.c (highest_pow2_factor, expand_expr_real_1): Likewise.
2728 * fold-const.c (force_fit_type, fold_negate_expr, int_const_binop,
2729 const_binop, fold_convert_const_int_from_int,
2730 fold_convert_const_int_from_real,
2731 fold_convert_const_real_from_real, sign_bit_p,
2732 optimize_minmax_comparison, extract_muldiv_1, fold_div_compare,
2733 fold_sign_changed_comparison, fold_unary, fold_comparison,
2734 fold_binary, multiple_of_p, tree_Expr_non_zero_p,
2735 fold_negate_const, fold_abs_const, fold_not_const): Likewise.
2736 * print-tree.c (print_node_brief, print_node): Likewise.
2737 * stor-layout.c (place_field, layout_type): Likewise.
2738 * tree-chrec.c (keep_cast): Likewise.
2739 * tree.c (build_vector, build_real, build_real_from_int_cst,
2740 build_complex): Likewise.
2742 2007-01-11 Roger Sayle <roger@eyesopen.com>
2744 * tree.h (TREE_CONSTANT_OVERFLOW): Obsolete. For the time being,
2745 treat TREE_CONSTANT_OVERFLOW as a synonym of TREE_OVERFLOW.
2747 2007-01-11 Paolo Bonzini <bonzini@gnu.org>
2749 * configure.ac (strict1_warn): Rename to strict_warn.
2750 (WERROR, --enable-werror, symlink hacks, stage1_cflags,
2751 cc_set_by_configure, quoted_cc_set_by_configure,
2752 stage_prefix_set_by_configure, quoted_stage_prefix_set_by_configure,
2753 all_boot_languages, all_stagestuff): Remove.
2754 (target_list): Remove bootstrap targets.
2755 * Makefile.in (quickstrap): Unconditionally make a synonym of all.
2756 (BOOT_LANGUAGES, STAGE1_CFLAGS, STAGE1_CHECKING,
2757 REMAKEFLAGS, FLAGS_TO_PASS, PREPEND_DOTDOT_TO_RELATIVE_PATHS,
2758 SUBDIR_FLAGS_TO_PASS, WERROR_FLAGS, STRICT2_WARN, LANG_STAGESTUFF,
2759 VOL_FILES, POSTSTAGE1_FLAGS_TO_PASS, STAGE2_FLAGS_TO_PASS,
2760 STAGEPROFILE_FLAGS_TO_PASS, STAGEFEEDBACK_FLAGS_TO_PASS, stage1_build,
2761 stage1_copy, stage2_build, stage2_copy, stageprofile_build,
2762 stageprofile_copy, stage3_build, stage3_copy, stagefeedback_build,
2763 stagefeedback_copy, stage4_build, clean_s1, clean_sw, bootstrap,
2764 bootstrap-lean, bootstrap2, bootstrap2-lean, bootstrap3,
2765 bootstrap3-lean, bootstrap4, bootstrap4-lean, unstage1, unstage2,
2766 unstage3, unstage4, unstageprofile, unstagefeedback, restage, restage2,
2767 restage3, restage4, restageprofile, restagefeedback, bubbleestrap,
2768 cleanstrap, unstrap, restrap, *compare, *compare3, *compare4,
2769 *compare-lean, *compare3-lean, *compare4-lean, stage1-start, stage1,
2770 stage2-start, stage2, stage3-start, stage3, stage4-start, stage4,
2771 stageprofile-start, stageprofile, stagefeedback-start, stagefeedback,
2772 risky-stage1, risky-stage2, risky-stage3, risky-stage4): Remove.
2773 (ORDINARY_FLAGS_TO_PASS): Rename to FLAGS_TO_PASS.
2774 (STAGECOPYSTUFF, STAGEMOVESTUFF): Consolidate into MOSTLYCLEANFILES.
2775 (mostlyclean): Adjust.
2776 (clean, distclean): Don't mention bootstrap stuff.
2777 * configure: Regenerate.
2778 * ada/config-lang.in, cp/config-lang.in, forttran/config-lang.in,
2779 java/config-lang.in, objc/config-lang.in, objcp/config-lang.in,
2780 treelang/config-lang.in (stagestuff): Remove.
2781 * doc/sourcebuild.texi (stage1, stage2, stage3, stage4,
2782 stageprofile, stagefeedback, stagestuff): Remove mention.
2784 2007-01-11 Nick Clifton <nickc@redhat.com>
2786 * config/mcore/predicates.md (mcore_general_movesrc_operand):
2787 Accept CONSTs.
2788 (mcore_general_movdst_operand): Do not accept CONST_INTs.
2789 (mcore_arith_K_S_operand): Run the test for the S constraint not
2790 the test for the M constraint.
2791 (mcore_addsub_operand): Do not accept integer values that are
2792 larger than 32 bits.
2793 * config/mcore/mcore.md: Remove unused constraints from split.
2794 (andsi3): Use HOST_WIDE_INT instead of int to hold an INTVAL.
2795 (addsi3): Likewise.
2796 (allocate_stack): Likewise.
2797 * config/mcore/mcore.c (mcore_print_operand): Restrict output of P
2798 operands to 32 bits.
2799 (mcore_const_costs): Use HOST_WIDE_INT instead of int to hold an
2800 INTVAL.
2801 (mcore_and_cost, mcore_modify_comparison, const_ok_for_mcore,
2802 mcore_const_ok_for_inline, mcore_const_trick_uses_not,
2803 try_constant_tricks, mcore_num_ones, mcore_num_zeros,
2804 mcore_output_bclri, mcore_output_andn, output_inline_const,
2805 mcore_output_move, mcore_output_movedouble): Likewise.
2806 (mcore_output_cmov): Use CONST_OK_FOR_M and CONST_OK_FOR_N.
2807 (output_inline_const): Likewise.
2808 (output_inline_const): Fix format strings used in sprintf
2809 statements.
2810 * config/mcore/mcore-protos.h: Update prototypes for changed
2811 functions in mcore.c.
2812 * config/mcore/mcore.h (CONST_OK_FOR_I): Cast values to
2813 HOST_WIDE_INT and not int.
2814 (CONST_OK_FOR_J, CONST_OK_FOR_K, CONST_OK_FOR_L, CONST_OK_FOR_M,
2815 CONST_OK_FOR_N): Likewise.
2816 (LEGITIMATE_CONSTANT_P): Also check CONSTANT_P.
2817 (GO_IF_LEGITIMATE_INDEX): Use HOST_WIDE_INT instead of int to hold
2818 an INTVAL.
2820 2007-01-10 Jan Hubicka <jh@suse.cz>
2822 * tree-vrp.c (remove_range_assertions): Release defs.
2823 * tree-ssa-loop-ivopts.c (rmeove_statement): Likewise.
2824 * tree-ssa-dom.c (remove_stmt_or_phi): Likewise.
2826 2007-01-10 Paul Brook <paul@codesourcery.com>
2828 * config/arm/arm.c (arm_rtx_costs_1): Handle mutiply-subtract.
2829 * config/arm/arm.md (mulsi3subsi): New insn.
2831 2007-01-10 Zdenek Dvorak <dvorakz@suse.cz>
2833 * tree-ssa-loop-manip.c (tree_unroll_loop): Make it a wrapper over ...
2834 (tree_transform_and_unroll_loop): New.
2835 * tree-flow.h (transform_callback, tree_transform_and_unroll_loop):
2836 Declare.
2838 2007-01-10 Robert Kennedy <jimbob@google.com>
2840 * fold-const.c (fold_comparison): Fold comparisons like (x *
2841 1000 < 0) to (x < 0).
2843 2007-01-10 Ian Lance Taylor <iant@google.com>
2845 * tree-pretty-print.c (dump_generic_node): Print parentheses when
2846 operands have the same priority.
2848 2007-01-10 Tom Tromey <tromey@redhat.com>
2850 * fold-const.c (fold_truthop): Don't check can_use_bit_fields_p.
2851 (fold_binary): Likewise.
2852 * langhooks.c (lhd_can_use_bit_fields_p): Removed.
2853 * langhooks-def.h (lhd_can_use_bit_fields_p): Removed.
2854 (LANG_HOOKS_CAN_USE_BIT_FIELDS_P): Removed.
2855 (LANG_HOOKS_INITIALIZER): Remove LANG_HOOKS_CAN_USE_BIT_FIELDS_P.
2856 * langhooks.h (struct lang_hooks): Removed field
2857 'can_use_bit_fields_p'.
2859 2007-01-10 Ralf Corsépius <ralf.corsepius@rtems.org>
2861 * config/bfin/t-bfin, config/bfin/t-bfin-elf: Remove GCC_CFLAGS.
2863 2007-01-10 Razya Ladelsky <razya@il.ibm.com>
2865 * function.c (get_last_funcdef_no): New function.
2866 * function.h (get_last_funcdef_no): Declare.
2867 * tree-inline.c (initialize_cfun): Add initialization.
2868 (tree_function_versioning): Cleanup.
2870 2007-01-10 Jan Hubicka <jh@suse.cz>
2872 * tree-inline.c (setup_one_parameter): Do not propagate into abnormal
2873 PHIs.
2875 2007-01-10 Sa Liu <saliu@de.ibm.com>
2876 Ben Elliston <bje@au.ibm.com>
2878 * spu.h (STACK_SAVE_AREA): Use VOIDmode for SAVE_FUNCTION, SImode
2879 for SAVE_NONLOCAL and Pmode for any other save level.
2880 * spu-protos.h (spu_restore_stack_block): Declare.
2881 * spu.md (save_stack_block): Remove.
2882 (restore_stack_block): Call spu_restore_stack_block.
2883 * spu.c (spu_restore_stack_block): New function.
2884 (spu_expand_epilogue): Remove old comment.
2886 2007-01-09 Zdenek Dvorak <dvorakz@suse.cz>
2888 PR tree-optimization/30322
2889 * tree-ssa-loop-ivopts.c (fold_affine_expr, iv_value): Removed.
2890 (cand_value_at): Return the value as aff_tree.
2891 (may_eliminate_iv): Convert the bound from aff_tree to tree.
2892 * tree-affine.c (aff_combination_add_cst, aff_combination_add_product,
2893 aff_combination_mult): New functions.
2894 (aff_combination_add): Use aff_combination_add_cst.
2895 (aff_combination_convert): Allow conversions to a wider type.
2896 (tree_to_aff_combination): Handle BIT_NOT_EXPR.
2897 * tree-affine.h (aff_combination_mult): Declare.
2899 2007-01-09 Carlos O'Donell <carlos@codesourcery.com>
2901 * doc/tm.texi: Update documentation to reflect reality of exec
2902 and start file search behaviours. Update copyright year.
2903 * doc/invoke.texi: Explain how GCC_EXEC_PREFIX is used to find
2904 header file directories.
2906 2007-01-09 Uros Bizjak <ubizjak@gmail.com>
2908 * config/i386/i386.md (*sinxf2): Rename to *sinxf2_i387.
2909 (*cosxf2): Rename to cosxf2_i387.
2910 (*sindf2, *sinsf2): Extend operand 1 to XFmode. Macroize patterns
2911 using X87MODEF12 mode macro. Rename patterns to
2912 *sin_extend<mode>xf2_i387. Use SSE_FLOAT_MODE_P to disable patterns
2913 for SSE math.
2914 (*cosdf2, *cossf2): Ditto.
2915 (sincosdf3, sincossf3): Ditto. Rewrite corresponding splitters
2916 to match extended input operands.
2917 (sincos<mode>3): New expander.
2918 (*sinextendsfdf2, *cosextendsfdf2, *sincosextendsfdf3): Remove
2919 insn patterns and corresponding splitters.
2921 2007-01-09 Kaz Kojima <kkojima@gcc.gnu.org>
2923 * config/sh/t-linux (TARGET_LIBGCC2_CFLAGS): Delete.
2924 (SHLIB_MAPFILES, SHLIB_LINK, SHLIB_INSTALL): Likewise.
2926 2007-01-09 Nicolas Pitre <nico@cam.org>
2928 PR target/30173
2929 * arm/ieee754-df.S (Lad_s): Also test the low word of X for zero.
2931 2007-01-08 Geoffrey Keating <geoffk@apple.com>
2933 * target.h (struct gcc_target): New field library_rtti_comdat.
2934 * target-def.h (TARGET_CXX_LIBRARY_RTTI_COMDAT): New.
2935 (TARGET_CXX): Add TARGET_CXX_LIBRARY_RTTI_COMDAT.
2936 * doc/tm.texi (C++ ABI): Document TARGET_CXX_LIBRARY_RTTI_COMDAT.
2937 * config/darwin.h (TARGET_CXX_LIBRARY_RTTI_COMDAT): Define.
2939 2007-01-08 Geoffrey Keating <geoffk@apple.com>
2941 * doc/invoke.texi (Optimize Options): Correct description of -O0.
2943 2007-01-08 Richard Guenther <rguenther@suse.de>
2945 * tree.h (force_fit_type_double): Export.
2946 (force_fit_type): Remove.
2947 * fold-const.c (force_fit_type_double): New function.
2948 (force_fit_type): Remove.
2949 (int_const_binop): Use it.
2950 (fold_convert_const_int_from_int): Likewise.
2951 (fold_convert_const_int_from_real): Likewise.
2952 (fold_div_compare): Likewise.
2953 (fold_sign_changed_comparison): Likewise.
2954 (fold_unary): Likewise.
2955 (fold_negate_const): Likewise.
2956 (fold_abs_const): Likewise.
2957 (fold_not_const): Likewise.
2958 * c-common.c (shorten_compare): Use force_fit_type_double.
2959 * convert.c (convert_to_pointer): Likewise.
2961 2007-01-08 Richard Guenther <rguenther@suse.de>
2963 * tree.h (build_int_cst_wide_type): Export.
2964 * tree.c (build_int_cst_wide_type): New function.
2965 (build_int_cst_wide): Fix comment.
2966 * builtins.c (fold_builtin_object_size): Use build_int_cst
2967 to build -1 or 0 of the correct type. Use fit_double_type
2968 to check for overflow.
2969 * fold-const.c (optimize_bit_field_compare): Use build_int_cst_type
2970 to build the mask.
2971 (decode_field_reference): Likewise.
2972 (all_ones_mask_p): Likewise.
2973 (native_interpret_int): Use build_int_cst_wide_type.
2974 (fold_binary): Use build_int_cst_type to build an all-ones
2975 value.
2976 * stor-layout.c (set_sizetype): Use build_int_cst_wide_type.
2978 2007-01-08 Daniel Jacobowitz <dan@codesourcery.com>
2980 * config/pa/t-pa64 (libgcc_stub.a): Use $(T).
2982 2007-01-09 Ben Elliston <bje@au.ibm.com>
2984 * genautomata.c (STATS_OPTION): New option.
2985 (stats_flag): New flag.
2986 (gen_automata_option): Handle it.
2987 (initiate_automaton_gen): Ditto.
2988 (write_automata): Output statistics only if stats_flag is
2989 set. Likewise, output time statistics only if time_flag is set.
2990 * doc/md.texi (Processor pipeline description): Document new flag.
2992 2007-01-08 Richard Guenther <rguenther@suse.de>
2994 * builtins.c (fold_builtin_int_roundingfn): Use fit_double_type.
2995 * tree.c (build_int_cst_type): Likewise.
2996 (size_in_bytes): Don't call force_fit_type on the result.
2997 (int_fits_type_p): Use fit_double_type.
2998 * fold-const.c (fit_double_type): New function.
2999 (force_fit_type): Use it.
3000 * tree.h (fit_double_type): Export.
3002 2007-01-08 Jan Hubicka <jh@suse.cz>
3004 * tree-vectorizer.c (gate_increase_alignment): Fix return type.
3005 * ipa.c (function_and_variable_visibility): Fix return type.
3007 2007-01-08 Richard Guenther <rguenther@suse.de>
3009 * tree-ssa-ccp.c (maybe_fold_offset_to_array_ref): Use type
3010 of offset to build the index.
3011 * tree-pretty-print.c (dump_generic_node): Don't build negated
3012 const just for printing.
3013 * c-pretty-print.c (pp_c_integer_constant): Likewise.
3014 * builtins.c (fold_builtin_int_roundingfn): Check if result
3015 fits the type by using force_fit_type and comparing the result.
3016 * predict.c (predict_loops): Use compare_tree_int for comparison.
3017 * tree.c (build_int_cst): Fall back to integer_type_node for
3018 NULL_TREE type.
3019 (build_int_cst_wide): Assert type is non-null.
3021 2007-01-08 Roberto Costa <roberto.costa@st.com>
3023 * tree-vrp.c (extract_range_from_cond_expr): New.
3024 (extract_range_from_expr): Handle COND_EXPR nodes used as expressions.
3025 * tree-ssa-ccp.c (get_maxval_strlen): Handle COND_EXPR nodes used
3026 as expressions.
3027 (fold_stmt): Bug fix, avoid infinite recursion when folding COND_EXPRs.
3028 * tree-ssa-forwprop.c (simplify_cond, forward_propagate_into_cond,
3029 tree_ssa_forward_propagate_single_use_vars): Handle COND_EXPR nodes
3030 used as expressions.
3031 * tree-object-size.c (cond_expr_object_size): New.
3032 (collect_object_sizes_for): Handle COND_EXPR nodes used as expressions.
3034 2007-01-08 Jan Hubicka <jh@suse.cz>
3036 * tree-ssa-forwprop.c (forward_propagate_into_cond,
3037 tree_ssa_forward_propagate_single_use_va): Release defs of propagated
3038 statement.
3040 2007-01-08 Richard Guenther <rguenther@suse.de>
3042 PR tree-optimization/23603
3043 * tree-vrp.c (set_value_range_to_truthvalue): New function.
3044 (extract_range_from_binary): Fall back to truthvalue instead of
3045 varying for TRUTH_*_EXPR.
3046 (extract_range_from_comparison): Fall back to truthvalue instead of
3047 varying.
3048 (vrp_visit_phi_node): Don't adjust new range bounds to +INF/-INF
3049 if all visited PHI values were constant.
3051 2007-01-08 Jan Hubicka <jh@suse.cz>
3053 * cgraphunit.c (cgraph_process_new_functions): Reset reachable flag.
3054 (cgraph_analyze_function): break out from ...
3055 (cgraph_finalize_compilation_unit): ... here.
3056 (cgraph_expand_function): Remove forgoten commented out line.
3057 (cgraph_optimize): Analyze functions.
3059 2007-01-08 Jan Hubicka <jh@suse.cz>
3061 * tree-pas.h (TODO_remove_function): New flag.
3062 (TODO_update*): Renumber.
3063 (pass_ipa_increase_alignment,
3064 pass_ipa_function_and_variable_visibility): New passes.
3065 * cgraphunit.c (cgraph_increase_alignment): Move to tree-vectorizer.c
3066 (cgraph_function_and_variable_visibility): Move to ipa.c
3067 (cgraph_optimize): Don't call cgraph_function_and_variable_visibility,
3068 cgraph_increase_alignment.
3069 * ipa-inline.c (cgraph_decide_inlining): Don't push timevar.
3070 (cgraph_decide_inlining_incrementally): Push TV_INTEGRATION before
3071 calling tree-inline.
3072 (cgraph_early_inlining): Do not call cgraph_remove_unreachable_nodes.
3073 (pass_ipa_inline, pass_early_ipa_inlining): Set TODO_remove_functions
3074 * tree-vectorizer.c (increase_alignment): Move here from cgraphunit.c
3075 (gate_increase_alignment): New function.
3076 (pass_ipa_increase_alignment): New pass.
3077 * ipa.c: Inline tree-pass.h and timevar.h
3078 (function_and_variable_visibility): Move here from cgraphunit.c
3079 * tree-optimize.c (pass_early_local_passes): Add TODO_remove_functions.
3080 * passes.c (init_optimization_passes): Add the two new passes.
3081 (execute_todo): Handle cgraph_remove_functions.
3083 2007-01-08 Nick Clifton <nickc@redhat.com>
3085 * config/frv/predicates.md (reg_or_0_operand): Accept
3086 CONST_DOUBLEs.
3088 2007-01-08 Ralf Corsépius <ralf.corsepius@rtems.org>
3090 * config/bfin/rtems.h, config/bfin/t-rtems: New.
3091 * config.gcc: Add bfin*-rtems*.
3093 2007-01-08 Mark Shinwell <shinwell@codesourcery.com>
3095 * c.opt: Add -flax-vector-conversions.
3096 * c-typeck.c (convert_for_assignment): Pass flag to
3097 vector_types_convertible_p to allow emission of note.
3098 (digest_init): Likewise.
3099 * c-opts.c: Handle -flax-vector-conversions.
3100 * c-common.c (flag_lax_vector_conversions): New.
3101 (vector_types_convertible_p): Unless -flax-vector conversions
3102 has been passed, disallow conversions between vectors with
3103 differing numbers of subparts and/or element types. If such
3104 a conversion is disallowed, possibly emit a note on the first
3105 occasion only to inform the user of -flax-vector-conversions.
3106 The new last argument specifies this.
3107 * c-common.h (flag_lax_vector_conversions): New.
3108 (vector_types_convertible_p): Add extra argument.
3109 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use
3110 char_type_node for V*QI type vectors.
3111 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
3112 Update to satisfy new typechecking rules.
3113 * config/rs6000/altivec.h (vec_cmple): Use vec_cmpge, for both
3114 C and C++ variants.
3115 * doc/invoke.texi (C Dialect Options): Document
3116 -flax-vector-conversions.
3118 2007-01-08 Mark Shinwell <shinwell@codesourcery.com>
3120 PR tree-optimization/29877
3121 * tree-ssa-ter.c (is_replaceable_p): Deem assignments with
3122 a register variable on the RHS to not be replaceable.
3124 2007-01-08 Chen Liqin <liqin@sunnorth.com.cn>
3125 * config/score/t-score-elf (MULTILIB_OPTIONS): Change.
3126 * config/score/predicates.md (const_uimm5, sr0_operand, const_simm12,
3127 const_simm15, const_pow2, const_npow2): Added.
3128 * config/score/misc.md (insv, extv, extzv, movmemsi,
3129 move_lbu_a/b, mov_lhu_a/b etc): Added and fix some bug.
3130 * config/score/score.c (score_address_cost, score_select_cc_mode):
3131 Added.
3132 Change CONST_OK_FOR_LETTER_P/EXTRA_CONSTRAINT define.
3133 Update score_rtx_costs for MACRO TARGET_RTX_COSTS.
3134 Update score_print_operand.
3135 * config/score/score.h (DATA_ALIGNMENT, SELECT_CC_MODE): Added.
3136 Adjust register allocate order and update some macro define.
3137 * config/score/score-mdaux.c (mdx_unaligned_load, mdx_unsigned_store,
3138 mdx_block_move_straight, mdx_block_move_loop_head,
3139 mdx_block_move_loop_body, mdx_block_move_loop_foot, mdx_block_move_loop,
3140 mdx_block_move): Added.
3141 (mdx_movsicc, mdp_select_add_imm, mdp_select, mds_zero_extract_andi,
3142 mdp_limm): Updated and fix some bug and typo.
3143 * config/score/score.md (movqi/hi/si, add/sub/zero/ext): Updated.
3144 (movsf, movdf, doloop_end): Added.
3146 2007-01-08 Kazu Hirata <kazu@codesourcery.com>
3148 * config/arm/arm.c, config/arm/arm.h, config/arm/arm.md,
3149 config/arm/thumb2.md: Fix comment typos.
3150 * doc/extend.texi: Fix a typo.
3152 2007-01-07 Eric Christopher <echristo@apple.com>
3154 * configure.ac: Check for __stack_chk_fail for darwin.
3155 * configure: Regenerate.
3157 2007-01-07 Richard Guenther <rguenther@suse.de>
3159 * tree-vrp.c (extract_range_from_assert): CSE calls to
3160 compare_values where possible.
3161 (extract_range_from_unary_expr): Likewise.
3163 2007-01-07 Anatoly Sokolov <aesok@post.ru>
3165 * config/avr/avr-protos.h (call_insn_operand): Delete prototype.
3166 * config/avr/avr.c (call_insn_operand): Delete function.
3167 * config/avr/avr.md (*pushqi, *pushhi, *pushsi, *pushsf): Use REG_SP
3168 instead of register number. Use predicates.
3169 * config/avr/predicates.md (const0_operand, reg_or_0_operand,
3170 call_insn_operand): Add.
3172 2007-01-06 Jan Hubicka <jh@suse.cz>
3174 * tree-pass.h (pass_build_cgraph_edges): Declare.
3175 * cgraphunit.c (record_refernece): Move to cgraphbuild.c
3176 (visited_nodes): Remove.
3177 (cgraph_create_edges): Move to cgraphbuild.c; rename to
3178 build_cgrpah_edges; make visited_nodes local.
3179 (cgraph_process_new_functions): DO not call initialize_inline_failed.
3180 (record_references_in_initializer): Move to cgraphbuild.c
3181 (initialize_inline_failed, rebuild_cgraph_edges,
3182 pass_rebuild_cgraph_edges): Move to cgraphbuild.c.
3183 (verify_cgraph_node): Make visited_nodes local.
3184 (cgraph_analyze_function): Do not call cgraph_create_edges and
3185 initialize_inline_failed.
3186 (cgraph_expand_function): Do not call cgraph_lower_function;
3187 assert that function is already lowered.
3188 * Makefile.in (cgraphbuild.o): New.
3189 * passes.c (init_optimization_passes): Add pass_build_cgraph_edges
3190 at the end of lowering passes.
3192 2007-01-06 Steven Bosscher <steven@gcc.gnu.org>
3194 * ifcvt.c (cond_move_convert_if_block): New function, code
3195 factored out from...
3196 (cond_move_process_if_block): ...here. Call the new function
3197 on the THEN and ELSE blocks.
3198 (merge_if_block): Do not copy global_live_at_end, merge_blocks
3199 already takes care of this.
3201 2007-01-05 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
3203 PR c/19978
3204 * tree.h (TREE_OVERFLOW_P): New.
3205 * c-typeck.c (parser_build_unary_op): Warn only if result
3206 overflowed and operands did not.
3207 (parser_build_binary_op): Likewise.
3208 (convert_for_assignment): Remove redundant overflow_warning.
3209 * c-common.c (overflow_warning): Don't check or set TREE_OVERFLOW.
3211 2007-01-05 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
3213 * c-typeck.c (store_init_value): Split over two lines to follow
3214 the GNU coding style.
3216 2007-01-05 Benjamin Kosnik <bkoz@redhat.com>
3218 * c-cppbuiltin.c (c_cpp_builtins): __GXX_EXPERIMENTAL_CPP0X__ to
3219 __GXX_EXPERIMENTAL_CXX0X__.
3220 * doc/cpp.texi: Same.
3222 2007-01-05 Richard Guenther <rguenther@suse.de>
3224 PR middle-end/27826
3225 * tree.c (get_narrower): Do not construct COMPONENT_REFs
3226 with mismatched types. Instead explicitly build a
3227 conversion NOP_EXPR.
3229 2007-01-05 Ian Lance Taylor <iant@google.com>
3231 * c-common.c (decl_with_nonnull_addr_p): New function.
3232 (c_common_truthvalue_conversion): Call it.
3233 * c-typeck.c (build_binary_op): Likewise.
3234 * c-common.h (decl_with_nonnull_addr_p): Declare.
3236 2007-01-05 Jakub Jelinek <jakub@redhat.com>
3238 PR c/30360
3239 * libgcc2.c (__divdc3): Compare c and d against 0.0 instead of
3240 denom against 0.0.
3242 2007-01-05 Joel Brobecker <brobecker@adacore.com>
3244 * doc/install.texi (Final install): Document the fact that
3245 the GNAT runtime should not be stripped.
3247 2007-01-04 Jan Hubicka <jh@suse.cz>
3249 * tree-inline.c (fold_marked_statements): Update operand caches
3250 and EH after folding
3252 2007-01-04 Ian Lance Taylor <iant@google.com>
3254 * c-common.c (check_function_nonnull): Whitespace fix.
3256 2007-01-04 Jan Hubicka <jh@suse.cz>
3258 * tree-optimize.c (execute_fixup_cfg): Correct previously mistakely
3259 comitted older version of patch.
3260 (pass_fixup_cfg): Add TODOs to verify flow and statements, dump
3261 function, celanup cfg and collect garbage.
3263 2007-01-04 Mike Stump <mrs@apple.com>
3265 * Makefile.in (mostlyclean): Don't remove libgcc anymore.
3266 (clean): Likewise.
3268 2007-01-04 Eric Christopher <echristo@apple.com>
3270 * libgcc2.c (__bswapsi2): Use SItype.
3271 (__bswapdi2): Use DItype.
3272 * libgcc2.h: Update for above.
3274 2007-01-04 Paul Brook <paul@codesourcery.com>
3276 * config/arm/arm.md (arm_mulsi3, thumb_mulsi3, mulsi3_compare0,
3277 mulsi_compare0_scratch, mulsi3addsi, mulsi3addsi_compare0,
3278 mulsi3addsi_compare0_scratch, mulsidi3adddi, mulsidi3,
3279 umulsidi3, umulsidi3adddi, smulsi3_highpart,
3280 umulsi3_highpart): Make conditional on !arm_arch6.
3281 (arm_mulsi3_v6, thumb_mulsi3_v6, mulsi3_compare0_v6,
3282 mulsi_compare0_scratch_v6, mulsi3addsi_v6, mulsi3addsi_compare0_v6,
3283 mulsi3addsi_compare0_scratch_v6, mulsidi3adddi_v6, mulsidi3_v6,
3284 umulsidi3_v6, umulsidi3adddi_v6, smulsi3_highpart_v6,
3285 umulsi3_highpart_v6): New insns.
3287 2007-01-04 Roger Sayle <roger@eyesopen.com>
3289 * fold-const.c (fold_convert): When casting an expression to void,
3290 fold_ignored_result may discover a GIMPLE_MODIFY_STMT which doesn't
3291 have a type. Instead of attempting to build a NOP_EXPR, return
3292 these "special" trees directly.
3294 2007-01-04 Joseph Myers <joseph@codesourcery.com>
3296 * config/rs6000/rs6000.c (rs6000_rtx_costs): Make adjustment for
3297 MULT inside MINUS as either argument. Use rs6000_cost->dmul -
3298 rs6000_cost->fp not 0 as adjustment for outer NEG.
3300 2007-01-04 Jan Hubicka <jh@suse.cz>
3302 * cgraph.c (cgraph_release_function_body): New function.
3303 (cgraph_remove_node): Use it.
3304 * cgraph.h (cgraph_release_function_body): Declare.
3305 * cgraphunit.c (cgraph_expand_function): Use it.
3306 * ipa.c (cgraph_remove_unreahchable_nodes): Use it.
3307 * tree-ssa.c (delete_tree_ssa): Allow to be called before aliasing
3308 is initialized and while compilation of other function is running.
3309 * tree-optimize.c (execute_free_cfg_annotations): Move code to clear
3310 statement CFG annotations from here to ...
3311 * tree-cfg.c (delete_tree_cfg_annotations): ... here.
3313 2007-01-04 Zdenek Dvorak <dvorakz@suse.cz>
3315 * cfgloop.h (enum li_flags): Make the constants powers of two.
3317 2007-01-04 Jan Hubicka <jh@suse.cz>
3319 * tree-inline.c (copy_bb): Insert new statements to statements_to_fold
3320 set.
3321 (fold_marked_statements): New function.
3322 (optimize_inline_calls, tree_function_versioning): Fold new statements.
3323 * tree-inline.h (copy_body_data): Add statements_to_fold.
3325 2007-01-03 Daniel Jacobowitz <dan@codesourcery.com>
3327 * config.gcc: Mention libgcc/config.host.
3328 * Makefile.in: Update comments mentioning libgcc.
3329 (LIBGCC, INSTALL_LIBGCC, GCC_PARTS, mklibgcc): Delete.
3330 (all.cross, start.encap, rest.encap, rest.cross): Update
3331 dependencies for libgcc move.
3332 (libgcc.mk, LIBGCC_DEPS, libgcov.a, libgcc.a, stmp-multilib)
3333 (clean-target, clean-target-libgcc): Delete.
3334 (srcdirify, GCC_EXTRA_PARTS): New macros.
3335 (libgcc-support, libgcc.mvars): New rules.
3336 (distclean): Remove mention of mklibgcc.
3337 (install): Don't reference INSTALL_LIBGCC.
3338 (install-common): Don't reference EXTRA_PARTS.
3339 (install-libgcc, install-multilib): Delete rules.
3340 * mklibgcc.in: Delete file.
3341 * doc/configfiles.texi: Don't mention mklibgcc.
3343 * config/i386/t-darwin (SHLIB_VERPFX): Delete (moved to libgcc).
3344 * config/i386/t-darwin64 (SHLIB_VERPFX): Likewise.
3345 * config/rs6000/t-darwin (SHLIB_VERPFX): Likewise.
3346 * config/rs6000/t-ppccomm (TARGET_LIBGCC2_CFLAGS, SHLIB_MAPFILES)
3347 (mklibgcc, ldblspecs): Likewise.
3349 * config/i386/t-nwld (libgcc.def, libc.def, libpcre.def)
3350 (posixpre.def): Use $(T).
3351 (SHLIB_EXT, SHLIB_NAME, SHLIB_SLIBDIR_QUAL, SHLIB_DEF, SHLIB_MAP)
3352 (SHLIB_SRC, SHLIB_INSTALL): Delete.
3353 (SHLIB_LINK): Make dummy.
3354 * config/t-slibgcc-darwin: Delete contents except for dummy SHLIB_LINK.
3356 * config/frv/t-linux (EXTRA_MULTILIB_PARTS): Clear.
3358 * config/alpha/t-crtfm: Use $(T) in rules for EXTRA_PARTS.
3359 * config/alpha/t-vms, config/alpha/t-vms64, config/fr30/t-fr30,
3360 config/i386/t-rtems-i386, config/ia64/t-ia64, config/rs6000/t-beos,
3361 config/rs6000/t-newas, config/sparc/t-elf: Likewise.
3363 * configure.ac (all_outputs): Remove mklibgcc.
3364 * configure: Regenerated.
3366 2007-01-03 Josh Conner <jconner@apple.com>
3368 PR middle-end/29683
3369 * calls.c (compute_argument_addresses): Set stack and stack_slot
3370 for partial args, too.
3371 (store_one_arg): Use locate.size.constant for the size when
3372 generating a save_area.
3374 2007-01-03 Robert Kennedy <jimbob@google.com>
3376 * tree-cfg.c (tree_merge_blocks): Release SSA_NAME phi results
3377 whose definitions are deleted due to basic block merging.
3379 2007-01-03 Paul Brook <paul@codesourcery.com>
3381 PR target/16634
3382 * config/arm/arm.c (output_return_instruction): Pop PC in interrupt
3383 functions.
3384 (use_return_insn): Return 0 for Thumb interrupt functions.
3385 (print_multi_reg): Add rfe argument for IRQ returns.
3386 (arm_output_epilogue): Pop interrupt return address directly into PC.
3387 (arm_expand_prologue): Only adjust IRQ return address in Arm mode.
3389 2007-01-03 Paul Brook <paul@codesourcery.com>
3391 Merge from sourcerygxx-4_1.
3392 * config/arm/thumb2.md: New file.
3393 * config/arm/elf.h (JUMP_TABLES_IN_TEXT_SECTION): Return True for
3394 Thumb-2.
3395 * config/arm/coff.h (JUMP_TABLES_IN_TEXT_SECTION): Ditto.
3396 * config/arm/aout.h (ASM_OUTPUT_ADDR_VEC_ELT): Add !Thumb-2 assertion.
3397 (ASM_OUTPUT_ADDR_DIFF_ELT): Output Thumb-2 jump tables.
3398 * config/arm/aof.h (ASM_OUTPUT_ADDR_DIFF_ELT): Output Thumb-2 jump
3399 tables.
3400 (ASM_OUTPUT_ADDR_VEC_ELT): Add !Thumb-2 assertion.
3401 * config/arm/ieee754-df.S: Use macros for Thumb-2/Unified asm
3402 comptibility.
3403 * config/arm/ieee754-sf.S: Ditto.
3404 * config/arm/arm.c (thumb_base_register_rtx_p): Rename...
3405 (thumb1_base_register_rtx_p): ... to this.
3406 (thumb_index_register_rtx_p): Rename...
3407 (thumb1_index_register_rtx_p): ... to this.
3408 (thumb_output_function_prologue): Rename...
3409 (thumb1_output_function_prologue): ... to this.
3410 (thumb_legitimate_address_p): Rename...
3411 (thumb1_legitimate_address_p): ... to this.
3412 (thumb_rtx_costs): Rename...
3413 (thumb1_rtx_costs): ... to this.
3414 (thumb_compute_save_reg_mask): Rename...
3415 (thumb1_compute_save_reg_mask): ... to this.
3416 (thumb_final_prescan_insn): Rename...
3417 (thumb1_final_prescan_insn): ... to this.
3418 (thumb_expand_epilogue): Rename...
3419 (thumb1_expand_epilogue): ... to this.
3420 (arm_unwind_emit_stm): Rename...
3421 (arm_unwind_emit_sequence): ... to this.
3422 (thumb2_legitimate_index_p, thumb2_legitimate_address_p,
3423 thumb1_compute_save_reg_mask, arm_dwarf_handle_frame_unspec,
3424 thumb2_index_mul_operand, output_move_vfp, arm_shift_nmem,
3425 arm_save_coproc_regs, thumb_set_frame_pointer, arm_print_condition,
3426 thumb2_final_prescan_insn, thumb2_asm_output_opcode, arm_output_shift,
3427 thumb2_output_casesi): New functions.
3428 (TARGET_DWARF_HANDLE_FRAME_UNSPEC): Define.
3429 (FL_THUMB2, FL_NOTM, FL_DIV, FL_FOR_ARCH6T2, FL_FOR_ARCH7,
3430 FL_FOR_ARCH7A, FL_FOR_ARCH7R, FL_FOR_ARCH7M, ARM_LSL_NAME,
3431 THUMB2_WORK_REGS): Define.
3432 (arm_arch_notm, arm_arch_thumb2, arm_arch_hwdiv, arm_condexec_count,
3433 arm_condexec_mask, arm_condexec_masklen)): New variables.
3434 (all_architectures): Add armv6t2, armv7, armv7a, armv7r and armv7m.
3435 (arm_override_options): Check new CPU capabilities.
3436 Set new architecture flag variables.
3437 (arm_isr_value): Handle v7m interrupt functions.
3438 (user_return_insn): Return 0 for v7m interrupt functions. Handle
3439 Thumb-2.
3440 (const_ok_for_arm): Handle Thumb-2 constants.
3441 (arm_gen_constant): Ditto. Use movw when available.
3442 (arm_function_ok_for_sibcall): Return false for v7m interrupt
3443 functions.
3444 (legitimize_pic_address, arm_call_tls_get_addr): Handle Thumb-2.
3445 (thumb_find_work_register, arm_load_pic_register,
3446 legitimize_tls_address, arm_address_cost, load_multiple_sequence,
3447 emit_ldm_seq, emit_stm_seq, arm_select_cc_mode, get_jump_table_size,
3448 print_multi_reg, output_mov_long_double_fpa_from_arm,
3449 output_mov_long_double_arm_from_fpa, output_mov_double_fpa_from_arm,
3450 output_mov_double_fpa_from_arm, output_move_double,
3451 arm_compute_save_reg_mask, arm_compute_save_reg0_reg12_mask,
3452 output_return_instruction, arm_output_function_prologue,
3453 arm_output_epilogue, arm_get_frame_offsets, arm_regno_class,
3454 arm_output_mi_thunk, thumb_set_return_address): Ditto.
3455 (arm_expand_prologue): Handle Thumb-2. Use arm_save_coproc_regs.
3456 (arm_coproc_mem_operand): Allow POST_INC/PRE_DEC.
3457 (arithmetic_instr, shift_op): Use arm_shift_nmem.
3458 (arm_print_operand): Use arm_print_condition. Handle '(', ')', '.',
3459 '!' and 'L'.
3460 (arm_final_prescan_insn): Use extract_constrain_insn_cached.
3461 (thumb_expand_prologue): Use thumb_set_frame_pointer.
3462 (arm_file_start): Output directive for unified syntax.
3463 (arm_unwind_emit_set): Handle stack alignment instruction.
3464 * config/arm/lib1funcs.asm: Remove default for __ARM_ARCH__.
3465 Add v6t2, v7, v7a, v7r and v7m.
3466 (RETLDM): Add Thumb-2 code.
3467 (do_it, shift1, do_push, do_pop, COND, THUMB_SYNTAX): New macros.
3468 * config/arm/arm.h (TARGET_CPU_CPP_BUILTINS): Define __thumb2__.
3469 (TARGET_THUMB1, TARGET_32BIT, TARGET_THUMB2, TARGET_DSP_MULTIPLY,
3470 TARGET_INT_SIMD, TARGET_UNIFIED_ASM, ARM_FT_STACKALIGN, IS_STACKALIGN,
3471 THUMB2_TRAMPOLINE_TEMPLATE, TRAMPOLINE_ADJUST_ADDRESS,
3472 ASM_OUTPUT_OPCODE, THUMB2_GO_IF_LEGITIMATE_ADDRESS,
3473 THUMB2_LEGITIMIZE_ADDRESS, CASE_VECTOR_PC_RELATIVE,
3474 CASE_VECTOR_SHORTEN_MODE, ADDR_VEC_ALIGN, ASM_OUTPUT_CASE_END,
3475 ADJUST_INSN_LENGTH): Define.
3476 (TARGET_REALLY_IWMMXT, TARGET_IWMMXT_ABI, CONDITIONAL_REGISTER_USAGE,
3477 STATIC_CHAIN_REGNUM, HARD_REGNO_NREGS, INDEX_REG_CLASS,
3478 BASE_REG_CLASS, MODE_BASE_REG_CLASS, SMALL_REGISTER_CLASSES,
3479 PREFERRED_RELOAD_CLASS, SECONDARY_OUTPUT_RELOAD_CLASS,
3480 SECONDARY_INPUT_RELOAD_CLASS, LIBCALL_VALUE, FUNCTION_VALUE_REGNO_P,
3481 TRAMPOLINE_SIZE, INITIALIZE_TRAMPOLINE, HAVE_PRE_INCREMENT,
3482 HAVE_POST_DECREMENT, HAVE_PRE_DECREMENT, HAVE_PRE_MODIFY_DISP,
3483 HAVE_POST_MODIFY_DISP, HAVE_PRE_MODIFY_REG, HAVE_POST_MODIFY_REG,
3484 REGNO_MODE_OK_FOR_BASE_P, LEGITIMATE_CONSTANT_P,
3485 REG_MODE_OK_FOR_BASE_P, REG_OK_FOR_INDEX_P, GO_IF_LEGITIMATE_ADDRESS,
3486 LEGITIMIZE_ADDRESS, THUMB2_LEGITIMIZE_ADDRESS,
3487 GO_IF_MODE_DEPENDENT_ADDRESS, MEMORY_MOVE_COST, BRANCH_COST,
3488 ASM_APP_OFF, ASM_OUTPUT_CASE_LABEL, ARM_DECLARE_FUNCTION_NAME,
3489 FINAL_PRESCAN_INSN, PRINT_OPERAND_PUNCT_VALID_P,
3490 PRINT_OPERAND_ADDRESS): Adjust for Thumb-2.
3491 (arm_arch_notm, arm_arch_thumb2, arm_arch_hwdiv): New declarations.
3492 * config/arm/arm-cores.def: Add arm1156t2-s, cortex-a8, cortex-r4 and
3493 cortex-m3.
3494 * config/arm/arm-tune.md: Regenerate.
3495 * config/arm/arm-protos.h: Update prototypes.
3496 * config/arm/vfp.md: Enable patterns for Thumb-2.
3497 (arm_movsi_vfp): Add movw alternative. Use output_move_vfp.
3498 (arm_movdi_vfp, movsf_vfp, movdf_vfp): Use output_move_vfp.
3499 (thumb2_movsi_vfp, thumb2_movdi_vfp, thumb2_movsf_vfp,
3500 thumb2_movdf_vfp, thumb2_movsfcc_vfp, thumb2_movdfcc_vfp): New.
3501 * config/arm/libunwind.S: Add Thumb-2 code.
3502 * config/arm/constraints.md: Update include Thumb-2.
3503 * config/arm/ieee754-sf.S: Add Thumb-2/Unified asm support.
3504 * config/arm/ieee754-df.S: Ditto.
3505 * config/arm/bpabi.S: Ditto.
3506 * config/arm/t-arm (MD_INCLUDES): Add thumb2.md.
3507 * config/arm/predicates.md (low_register_operand,
3508 low_reg_or_int_operand, thumb_16bit_operator): New.
3509 (thumb_cmp_operand, thumb_cmpneg_operand): Rename...
3510 (thumb1_cmp_operand, thumb1_cmpneg_operand): ... to this.
3511 * config/arm/t-arm-elf: Add armv7 multilib.
3512 * config/arm/arm.md: Update patterns for Thumb-2 and Unified asm.
3513 Include thumb2.md.
3514 (UNSPEC_STACK_ALIGN, ce_count): New.
3515 (arm_incscc, arm_decscc, arm_umaxsi3, arm_uminsi3,
3516 arm_zero_extendsidi2, arm_zero_extendqidi2): New
3517 insns/expanders.
3518 * config/arm/fpa.md: Update patterns for Thumb-2 and Unified asm.
3519 (thumb2_movsf_fpa, thumb2_movdf_fpa, thumb2_movxf_fpa,
3520 thumb2_movsfcc_fpa, thumb2_movdfcc_fpa): New insns.
3521 * config/arm/cirrus.md: Update patterns for Thumb-2 and Unified asm.
3522 (cirrus_thumb2_movdi, cirrus_thumb2_movsi_insn,
3523 thumb2_cirrus_movsf_hard_insn, thumb2_cirrus_movdf_hard_insn): New
3524 insns.
3525 * doc/extend.texi: Document ARMv7-M interrupt functions.
3526 * doc/invoke.texi: Document Thumb-2 new cores+architectures.
3528 2007-01-03 Jakub Jelinek <jakub@redhat.com>
3530 * unwind-dw2.c (SIGNAL_FRAME_BIT, EXTENDED_CONTEXT_BIT): Define.
3531 (struct _Unwind_Context): Rename args_size to flags, remove
3532 signal_frame field, add a new args_size field and version field.
3533 (_Unwind_IsSignalFrame, _Unwind_SetSignalFrame,
3534 _Unwind_IsExtendedContext): New inline functions.
3535 (_Unwind_GetGR, _Unwind_SetGR, _Unwind_GetGRPtr, _Unwind_SetGRPtr):
3536 Assume by_value array is only present if _Unwind_IsExtendedContext.
3537 (_Unwind_GetIPInfo, execute_cfa_program, uw_frame_state_for): Use
3538 _Unwind_IsSignalFrame.
3539 (__frame_state_for): Initialize context.flags to EXTENDED_CONTEXT_BIT.
3540 (uw_update_context_1): Use _Unwind_SetSignalFrame.
3541 (uw_init_context_1): Initialize context->flags to
3542 EXTENDED_CONTEXT_BIT.
3543 * config/rs6000/linux-unwind.h (frob_update_context): Use
3544 _Unwind_SetSignalFrame.
3546 2007-01-03 Andrew Pinski <andrew_pinski@playstation.sony.com>
3548 PR middle-end/30353
3549 * gimplify.c (gimplify_modify_expr_complex_part): Move below
3550 tree_to_gimple_tuple. Call tree_to_gimple_tuple when we need
3551 the value.
3553 2007-01-03 Kazu Hirata <kazu@codesourcery.com>
3555 * config/i386/i386.h (NON_STACK_REG_P, REGNO_OK_FOR_SIREG_P,
3556 REGNO_OK_FOR_DIREG_P, REWRITE_ADDRESS, ASM_OPERAND_LETTER,
3557 RET, AT_SP): Remove.
3558 * config/i386/i386.md (*sse_prologue_save_insn): Use return
3559 instead of RET.
3561 * alias.c (init_alias_analysis): Use VEC_safe_grow_cleared.
3562 * cfgbuild.c (find_basic_blocks): Likewise.
3563 * cfgrtl.c (rtl_create_basic_block): Likewise.
3564 * function.c (temp_slots_at_level): Likewise.
3565 * reg-stack.c (stack_regs_mentioned): Likewise.
3566 * regclass.c (allocate_reg_info): Likewise.
3567 * tree-cfg.c (init_empty_tree_cfg, build_tree_cfg, create_bb,
3568 set_bb_for_stmt, move_block_to_fn): Likewise.
3569 * tree-complex.c (tree_lower_complex): Likewise.
3570 * vec.h (VEC_safe_grow_cleared): New.
3572 * cgraphunit.c, tree-ssa-alias.c: Fix comment typos.
3574 2007-01-03 Zdenek Dvorak <dvorakz@suse.cz>
3576 * loop-unswitch.c (unswitch_loop): Pass probabilities to loopify.
3577 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Pass probabilities
3578 to loop_version.
3579 * cfgloopmanip.c (scale_loop_frequencies): Export.
3580 (loopify): Scale the frequencies by prescribed coefficients.
3581 (set_zero_probability): New function.
3582 (duplicate_loop_to_header_edge): Improve updating of frequencies.
3583 (lv_adjust_loop_entry_edge, loop_version): Set probabilities
3584 and frequencies according to arguments.
3585 * tree-ssa-loop-manip.c (tree_unroll_loop): Set probabilities
3586 correctly.
3587 * cfg.c (scale_bbs_frequencies_int): Allow scaling the frequencies up.
3588 * modulo-sched.c (sms_schedule): Set probabilities for entering
3589 versioned loop correctly.
3590 * tree-vect-transform.c (vect_transform_loop): Ditto.
3591 * cfgloop.h (loopify, loop_version): Declaration changed.
3592 (scale_loop_frequencies): Declared.
3594 2007-01-02 Jan Hubicka <jh@suse.cz>
3596 * cgraph.c: Include tree-flow.h
3597 (cgraph_add_new-function): Handle IPA_SSA mode; execute
3598 early_local_passes.
3599 * cgraph.h (enum cgraph_state): Add CGRAPH_STATE_IPA_SSA.
3600 * tree-pass.h (pass_all_early_optimizations): Declare.
3601 * cgraphunit.c (cgraph_process_new_functions): Add IPA_SSA; execute
3602 early_local_passes.
3603 (cgraph_analyze_function): Do early_local_passes.
3604 * tree-mudflap.c (mf_decl_cache_locals, mf_build_check_statement_for):
3605 Do not add referenced vars.
3606 * tree-optimize.c (gate_all_optimizations): Do not execute when not in
3607 SSA form.
3608 (gate_all_early_local_passes): New gate.
3609 (pass_early_local_passes): Use new gate.
3610 (execute_early_local_optimizations): New functions.
3611 (gate_all_early_optimizations): New gate.
3612 (pass_all_early_optimizations): New pass.
3613 (execute_free_datastructures): Free SSA only when initialized.
3614 (gate_init_datastructures): Init only when optimizing.
3615 (tree_lowering_passes): Do early local passes when called late.
3616 * tree-profile.c (do_tree_profiling): Don't profile functions added
3617 late.
3618 (do_early_tree_profiling, pass_early_tree_profile): Kill.
3619 * tree-cfg.c (update_modified_stmts): Do not update when operands are
3620 not active.
3621 * passes.c (init_optimizations_passes): Reorder so we go into SSA
3622 during early_local_passes.
3623 * Makefile.in (cgraph.o): Add dependency on tree-flow.h.
3626 2007-01-02 Carlos O'Donell <carlos@codesourcery.com>
3628 * Makefile.in: Update copyright year.
3630 2007-01-02 Carlos O'Donell <carlos@codesourcery.com>
3632 * Makefile.in: Export GCC_EXEC_PREFIX before calling $(RUNTEST)
3633 in $(lang_checks) and check-consistency targets.
3635 2007-01-02 Jan Hubicka <jh@suse.cz>
3637 * tree-mudflap.c (mf_decl_cache_locals, mf_build_check_statement_for):
3638 Do not add referenced vars.
3639 * tree-cfg.c (update_modified_stmts): Do not update when SSA operands
3640 are not active.
3641 * passes.c (init_optimization_passes): Put mudflap_2 after
3642 free_datastructures.
3644 2007-01-02 Jan Hubicka <jh@suse.cz>
3646 * tree-optimize (execute_fixup_cfg): Set after_inlining flag.
3647 Set NOTHROW flag on call statements proved to be nothrow.
3648 Update statement of local calls so new pure/const functions are
3649 updated. Update_ssa when in ssa form. Mark PHI nodes of nonlocal
3650 goto receivers.
3651 (tree_rest_of_compilation): Register hooks and initialize bitmap
3652 early. Do not set after_inlining flag.
3654 2007-01-02 Steve Ellcey <sje@cup.hp.com>
3656 * sbitmap.c (HOST_BITS_PER_LONG_LONG): Change to
3657 HOST_BITS_PER_LONGLONG
3659 2007-01-02 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
3661 PR c/19977
3662 * c-typeck.c (store_init_value): Don't emit pedantic overflow
3663 warning for non-static initializers.
3665 2007-01-02 Steven Bosscher <steven@gcc.gnu.org>
3667 * config/alpha/alpha.md, arm/arm.c, darwin.c, frv/frv.md,
3668 m32r/m32r.c, m32r/m32r.c, mn10300/mn10300.md, pa/pa.c,
3669 rs6000/rs6000.c, s390/s390.md, sh/sh.md, sparc/sparc.c:
3670 Always use set_unique_reg_note to add REG_EQUAL notes.
3672 2007-01-02 Kazu Hirata <kazu@codesourcery.com>
3674 Revert:
3675 2007-01-02 Kazu Hirata <kazu@codesourcery.com>
3677 * alias.c (init_alias_analysis): Use VEC_safe_grow_cleared.
3678 * cfgbuild.c (find_basic_blocks): Likewise.
3679 * cfgrtl.c (rtl_create_basic_block): Likewise.
3680 * function.c (temp_slots_at_level): Likewise.
3681 * reg-stack.c (stack_regs_mentioned): Likewise.
3682 * regclass.c (allocate_reg_info): Likewise.
3683 * tree-cfg.c (init_empty_tree_cfg, build_tree_cfg, create_bb,
3684 set_bb_for_stmt, move_block_to_fn): Likewise.
3685 * tree-complex.c (tree_lower_complex): Likewise.
3686 * vec.h (VEC_safe_grow_cleared): New.
3688 2007-01-02 Ian Lance Taylor <iant@google.com>
3690 * c-common.c (c_common_truthvalue_conversion): When warning about
3691 using an assignment as a truth value, set TREE_NO_WARNING.
3693 2007-01-02 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
3695 PR middle-end/7651
3696 * c.opt (Wold-style-declaration): New.
3697 * doc/invoke.texi (C-only Warning Options): New.
3698 (Wold-style-declaration): Document it.
3699 (Wextra): Enabled by -Wextra.
3700 * c-opts.c (c_common_post_options): Enabled by -Wextra.
3701 * c-decl.c (declspecs_add_scspec): Replace -Wextra with
3702 -Wold-style-declaration.
3704 2007-01-02 Kazu Hirata <kazu@codesourcery.com>
3706 * alias.c (init_alias_analysis): Use VEC_safe_grow_cleared.
3707 * cfgbuild.c (find_basic_blocks): Likewise.
3708 * cfgrtl.c (rtl_create_basic_block): Likewise.
3709 * function.c (temp_slots_at_level): Likewise.
3710 * reg-stack.c (stack_regs_mentioned): Likewise.
3711 * regclass.c (allocate_reg_info): Likewise.
3712 * tree-cfg.c (init_empty_tree_cfg, build_tree_cfg, create_bb,
3713 set_bb_for_stmt, move_block_to_fn): Likewise.
3714 * tree-complex.c (tree_lower_complex): Likewise.
3715 * vec.h (VEC_safe_grow_cleared): New.
3717 2007-01-02 Douglas Gregor <doug.gregor@gmail.com>
3719 * c-common.c (c_common_nodes_and_builtins): Since variants of
3720 void_type_node get built before it is given a name, we need to
3721 give those variants the name, too.
3722 (complete_array_type): We need to work with the canonical main
3723 type of the array, from which we will build the qualified version.
3724 * params.def (PARAM_VERIFY_CANONICAL_TYPES): New.
3725 * print-tree.c (print_node): Display canonical type information
3726 for each type.
3727 * stor-layout.c (layout_type): When we don't know the
3728 alignment of a type for which we're building an array, we end up
3729 guessing wrong, so make the type require structural equality.
3730 * tree.c (make_node_stat): When we build a new type, it is its
3731 own canonical type.
3732 (build_type_attribute_qual_variant): When building an attribute
3733 variant, its canonical type is the non-attribute variant. However,
3734 if the attributes are target-dependent and they differ, we need to
3735 use structural equality checks for this type.
3736 (build_qualified_type): A qualified type is not equivalent to its
3737 unqualified variant; set the canonical type appropriately.
3738 (build_distinct_type_copy): When building a distinct type from
3739 another type, the new type is its own canonical type.
3740 (build_variant_type_copy): When building a new type variant, we
3741 assume that it is equivalent to the original type.
3742 (build_pointer_type_for_mode): When building a pointer type, also
3743 build a canonical type pointer.
3744 (build_reference_type_for_mode): When building a reference type,
3745 also build a canonical type reference.
3746 (build_index_type): When we can't hash an index type (e.g.,
3747 because its maximum value is negative), the index type requires
3748 structural equality tests.
3749 (build_array_type): Build the canonical form of an array type.
3750 (build_function_type): Function types require structural equality,
3751 because they contain default arguments, attributes, etc.
3752 (build_method_type_directly): Ditto for method types.
3753 (build_offset_type): Build the canonical offset type.
3754 (build_complex_type): Build the canonical vector type.
3755 (make_vector_type): Build the canonical vector type.
3756 * tree.h (TYPE_CANONICAL): New.
3757 (TYPE_STRUCTURAL_EQUALITY_P): New.
3758 (SET_TYPE_STRUCTURAL_EQUALITY): New.
3759 (struct tree_type): Added "canonical" field.
3760 * params.h (VERIFY_CANONICAL_TYPES): New.
3761 * doc/c-tree.texi (TYPE_CANONICAL): Document.
3762 (TYPE_STRUCTURAL_EQUALITY_P): Document.
3763 (SET_TYPE_STRUCTURAL_EQUALITY): Document.
3764 * doc/invoke.texi (verify-canonical-types): Document --param
3765 parameter for verifying canonical types.
3767 2007-01-02 Joseph Myers <joseph@codesourcery.com>
3769 * config.gcc (powerpc-*-eabispe*, powerpc-*-eabisimaltivec*,
3770 powerpc-*-eabisim*, powerpc-*-eabialtivec*, powerpc-*-eabi*,
3771 powerpc-*-rtems*, powerpc-wrs-vxworks, powerpc-wrs-vxworksae,
3772 powerpcle-*-eabisim*, powerpcle-*-eabi*): Add rs6000/e500.h to
3773 tm_file.
3774 * config/rs6000/e500.h: New.
3775 * config/rs6000/eabi.h (TARGET_SPE_ABI, TARGET_SPE, TARGET_E500,
3776 TARGET_ISEL, TARGET_FPRS, TARGET_E500_SINGLE, TARGET_E500_DOUBLE):
3777 Remove.
3778 * config/rs6000/linuxspe.h (TARGET_SPE_ABI, TARGET_SPE,
3779 TARGET_E500, TARGET_ISEL, TARGET_FPRS, TARGET_E500_SINGLE,
3780 TARGET_E500_DOUBLE): Remove.
3781 * config/rs6000/vxworks.h (TARGET_SPE_ABI, TARGET_SPE,
3782 TARGET_E500, TARGET_ISEL, TARGET_FPRS): Remove.
3783 * config/rs6000/rs6000.h (CHECK_E500_OPTIONS): Define.
3784 * config/rs6000/rs6000.c (rs6000_override_options): Use
3785 CHECK_E500_OPTIONS.
3787 2007-01-02 Joseph Myers <joseph@codesourcery.com>
3789 * config/rs6000/rs6000.c (print_operand): Check (TARGET_SPE ||
3790 TARGET_E500_DOUBLE), not TARGET_E500, for %y.
3791 (rs6000_generate_compare, rs6000_emit_sCOND, output_cbranch,
3792 rs6000_emit_cmove): Don't check TARGET_E500.
3793 * config/rs6000/rs6000.md (bunordered, bordered, sunordered,
3794 sordered): Don't check TARGET_E500.
3796 2007-01-01 Eric Christopher <echristo@apple.com>
3798 * config/mips/mips.c (mips_regno_mode_ok_for_base_p): Use
3799 HARD_REGISTER_NUM_P.
3801 2007-01-01 Roger Sayle <roger@eyesopen.com>
3803 * fold-const.c (fold_binary) <EQ_EXPR>: Fold "(X^C1) eq/ne C2" into
3804 "X eq/ne (C1^C2)". Fold "(X^Z) eq/ne (Y^Z)" as "X eq/ne Y" when Z
3805 has no side-effects. Fold "(X^C1) eq/ne (Y^C2)" as "(X^(C1^C2))
3806 eq/ne Y".
3808 2007-01-01 Mike Stump <mrs@apple.com>
3810 * configure.ac: Remove support for building with Apple's gcc-3.1.
3812 2007-01-02 Joseph Myers <joseph@codesourcery.com>
3814 PR middle-end/30311
3815 * caller-save.c (add_stored_regs): Only handle SUBREGs if inner
3816 REG is a hard register. Do not modify REG before calling
3817 subreg_nregs.
3818 * rtlanal.c (subreg_get_info): Don't assert size of XMODE is a
3819 multiple of the size of YMODE for certain lowpart cases.
3821 2007-01-01 Andrew Pinski <pinskia@gmail.com>
3823 PR middle-end/30253
3824 * gimplify (voidify_wrapper_expr): Update for
3825 GIMPLE_MODIFY_STMT.
3827 2007-01-01 Andreas Schwab <schwab@suse.de>
3829 PR target/29166
3830 * config/ia64/ia64.c (ia64_compute_frame_size): Account space for
3831 save of BR0 in extra_spill_size instead of spill_size.
3832 (ia64_expand_prologue): Save BR0 outside of the gr/br/fr spill
3833 area.
3834 (ia64_expand_epilogue): Restore BR0 from its new location.
3836 2007-01-01 Andrew Pinski <pinskia@gmail.com>
3838 * gimplify.c (gimplify_init_constructor <case VECTOR_TYPE>):
3839 Use a temporary variable if the left hand side is not a gimple
3840 register.
3842 2007-01-01 Andrew Pinski <pinskia@gmail.com>
3844 * gimplify.c (gimplify_return_expr): Make the temporary variable
3845 for the return expression, a gimple register variable.
3847 2007-01-01 Jan Hubicka <jh@suse.cz>
3849 * emit-rtl.c (emit_copy_of_insn_after): Do not call copy_insn_1 for
3850 INSN_LIST.
3852 2007-01-01 Mike Stump <mrs@apple.com>
3854 * configure.ac (HAVE_GAS_LITERAL16): Add autoconf check for
3855 .literal16.
3856 * config/darwin.c (machopic_select_rtx_section): Use
3857 HAVE_GAS_LITERAL16.
3858 (darwin_mergeable_constant_section): Likewise.
3859 * configure: Regenerate.
3860 * config.in: Regenerate.
3862 2007-01-01 Jan Hubicka <jh@suse.cz>
3863 Andrew Pinski <pinskia@gmail.com>
3865 * cgraphunit.c (cgraph_optimize): Call cgraph_add_new_functions
3866 before starting IPA passes.