1 2007-02-01 Ralf Wildenhues <Ralf.Wildenhues@gmx.de>
3 * doc/c-tree.texi (Expression trees): Improve markup.
4 * doc/tm.texi (Register Classes, Addressing Modes)
5 (Floating Point): Fix spacing after abbreviations. Fix some
8 2007-02-01 Ben Elliston <bje@au.ibm.com>
10 * doc/invoke.texi: Replace "bugfix" with "bug fix" throughout.
11 * doc/contrib.texi: Likewise.
12 * doc/install.texi: Likewise.
14 2007-01-31 Richard Henderson <rth@redhat.com>
15 Ian Lance Taylor <iant@google.com>
17 * lower-subreg.c: New file.
18 * rtl.def (CONCATN): Define.
19 * passes.c (init_optimization_passes): Add pass_lower_subreg and
21 * emit-rtl.c (update_reg_offset): New static function, broken out
22 of gen_rtx_REG_offset.
23 (gen_rtx_REG_offset): Call update_reg_offset.
24 (gen_reg_rtx_offset): New function.
25 * regclass.c: Revert patch of 2006-03-05, restoring
27 (clear_reg_info_regno): New function.
28 * dwarf2out.c (concatn_loc_descriptor): New static function.
29 (loc_descriptor): Handle CONCATN.
30 * common.opt (fsplit_wide_types): New option.
31 * opts.c (decode_options): Set flag_split_wide_types when
33 * timevar.def (TV_LOWER_SUBREG): Define.
34 * rtl.h (gen_reg_rtx_offset): Declare.
35 (reg_scan_update): Declare.
36 * regs.h (clear_reg_info_regno): Declare.
37 * tree-pass.h (pass_lower_subreg): Declare.
38 (pass_lower_subreg2): Declare.
39 * doc/invoke.texi (Option Summary): List -fno-split-wide-types.
40 (Optimize Options): Add -fsplit-wide-types to -O1 list. Document
42 * doc/rtl.texi (Regs and Memory): Document concat and concatn.
43 * Makefile.in (OBJS-common): Add lower-subreg.o.
44 (lower-subreg.o): New target.
46 2007-01-31 Kazu Hirata <kazu@codesourcery.com>
48 * config/sh/sh.h (HAVE_SECONDARY_RELOADS): Remove.
50 2007-01-31 Anatoly Sokolov <aesok@post.ru>
53 * config/avr/avr.c (DWARF2_ADDR_SIZE): Define.
55 2007-01-31 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
58 * builtins.c (fold_builtin_sqrt): Use MPFR for constant args.
60 2007-01-31 Zdenek Dvorak <dvorakz@suse.cz>
62 * cfgloop.h: Include vec-prim.h.
63 (enum li_flags): Remove LI_ONLY_OLD.
64 (loop_iterator): Changed.
65 (fel_next, fel_init): Iterate over loop tree.
66 (FOR_EACH_LOOP_BREAK): New macro.
67 * loop-unswitch.c (unswitch_loops): Do not pass LI_ONLY_OLD to
69 * tree-ssa-loop-unswitch.c (tree_ssa_unswitch_loops): Ditto.
70 * modulo-sched.c (sms_schedule): Ditto.
71 * tree-vectorizer.c (vectorize_loops): Ditto.
72 * doc/loop.texi: Update information on loop numbering and behavior of
73 FOR_EACH_LOOP wrto new loops.
74 * tree-scalar-evolution.c (compute_overall_effect_of_inner_loop,
75 add_to_evolution_1): Test nestedness of loops instead of comparing
77 * tree-chrec.c (chrec_fold_plus_poly_poly,
78 chrec_fold_multiply_poly_poly, chrec_evaluate,
79 hide_evolution_in_other_loops_than_loop, chrec_component_in_loop_num,
80 reset_evolution_in_loop): Ditto.
81 * Makefile.in (CFGLOOP_H): Add vecprim.h dependency.
83 2007-01-31 Dirk Mueller <dmueller@suse.de>
85 * c-common.c (warn_about_parentheses): Separate warning about
86 un-parenthized sequence of comparison operators from the one
87 which is supposed to warn about x <= y <= z.
89 2007-01-31 Uros Bizjak <ubizjak@gmail.com>
91 * optabs.h (enum optab_index): Add new OTI_isinf.
92 (isinf_optab): Define corresponding macro.
93 * optabs.c (init_optabs): Initialize isinf_optab.
94 * genopinit.c (optabs): Implement isinf_optab using isinf?f2
96 * builtins.c (mathfn_built_in): Handle BUILT_IN_ISINF{,F,L}.
97 (expand_builtin_interclass_mathfn): Expand BUILT_IN_ISINF{,F,L}
99 (expand_builtin): Expand BUILT_IN_ISINF{,F,L} using
100 expand_builtin_interclass_mathfn.
101 * reg_stack.c (subst_stack_regs_pat): Handle UNSPEC_FXAM.
102 * config/i386/i386.md (UNSPEC_FXAM): New constant.
103 (fxam<mode>2_i387): New insn pattern.
104 (isinf<mode>2) New expander to implement isinf, isinff and isinfl
105 built-in functions as x87 inline asm.
107 2007-01-31 Kazu Hirata <kazu@codesourcery.com>
109 * gcc/config/arm/unwind-arm.h (_sleb128_t, _uleb128_t): New.
111 2007-01-30 Eric Christopher <echristo@apple.com>
113 * config.gcc: Add geode.
115 2007-01-31 Kazu Hirata <kazu@codesourcery.com>
117 * cgraphunit.c, config/arm/arm.c, config/m68k/m68k.c,
118 ipa-inline.c, tree-profile.c, tree-ssa-live.c,
119 tree-ssa-math-opts.c, tree-ssanames.c, tree-vect-analyze.c,
120 value-prof.c: Fix comment typos.
122 2007-01-30 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
125 * doc/invoke.texi (Wpointer-arith): Document warning.
127 2007-01-30 Janis Johnson
129 * doc/extend.texi (Decimal Floating Types): Remove decfloat.h from
130 the list of discrepancies from the draft TR.
132 2007-01-30 Dirk Mueller <dmueller@suse.de>
135 * doc/invoke.texi (-Wreturn-type): Update description to
138 2007-01-30 Richard Sandiford <richard@codesourcery.com>
140 * cfgrtl.c (try_redirect_by_replacing_jump): Check only_sets_cc0_p.
142 2007-01-30 Uros Bizjak <ubizjak@gmail.com>
144 * builtins.c (expand_builtin_int_interclass_roundingfn): New function
145 to handle optabs that operate on floating point input argument and
146 output to integer output.
147 (expand_builtin_mathfn) [BUILT_IN_ILOGB]: Move from here ...
148 (expand_builtin_interclass_mathfn) [BUILT_IN_ILOGB]: ... to here.
149 (expand_builtin): Expand BUILT_IN_ILOGB{,F,L} using
150 expand_builtin_interclass_mathfn ().
151 * config/i386/i386.md (fxtractxf3_i387): Rename from *fxtractxf3_i387.
153 (ilogbxf2, ilogb<mode>2): New expanders to implement ilogb, ilogbf and
154 ilogbl built-in functions as x87 intrinsics.
156 2007-01-30 Richard Guenther <rguenther@suse.de>
159 * dwarf2out.c (reference_to_unused): Query varpool if the
162 2007-01-30 Richard Guenther <rguenther@suse.de>
165 * passes.c (execute_one_pass): Reset in_gimple_form to not
166 confuse non-unit-at-a-time mode.
168 2007-01-29 Roger Sayle <roger@eyesopen.com>
169 Richard Guenther <rguenther@suse.de>
171 * fold-const.c (round_up): Make HIGH an unsigned HOST_WIDE_INT to
172 avoid undefined behaviour on overflow. Use force_fit_type_double
173 to construct the constant with the specified TREE_OVERFLOW.
175 2007-01-29 Janis Johnson <janis187@us.ibm.com>
177 * config/dfp-bit.c: Add parameterized support for fp exceptions.
178 * config/dfp-bit.h: Ditto.
180 2007-01-29 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
182 * c-decl.c (pop_scope): Replace warnings with call to
183 warn_for_unused_label.
184 * c-common.h (warn_for_unused_label): Declare.
185 * c-common.c (warn_for_unused_label): Define.
187 2007-01-29 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
189 * tree-optimize.c (update_inlined_to_pointers): Delete unused
192 2007-01-29 Janis Johnson <janis187@us.ibm.com>
194 * Makefile.in (USER_H): Remove decfloat.h.
195 * ginclude/decfloat.h: Delete, moving contents to ...
196 * ginclude/float.h: Add support for decimal floating point,
197 guarded by __STDC_WANT_DEC_FP__.
199 2007-01-29 Mike Stump <mrs@apple.com>
201 * doc/gccint.texi (Top): Rename Loop Representation to Loop
202 Analysis and Representation to resolve case insensitive conflict.
203 * doc/loop.texi (Loop Analysis and Representation): Likewise.
205 2007-01-28 Daniel Berlin <dberlin@dberlin.org>
207 * tree.h (struct tree_memory_tag): Add aliases member.
208 (MTAG_ALIASES): New macro.
209 * tree-ssa-alias.c (alias_bitmap_obstack): New variable.
210 (add_may_alias): Remove pointer-set. Update for may_aliases being
212 (mark_aliases_call_clobbered): Update for may_aliases being a
214 (compute_tag_properties): Ditto.
215 (create_partition_for): Ditto.
216 (compute_memory_partitions): Ditto.
217 (dump_may_aliases_for): Ditto.
218 (is_aliased_with): Ditto.
219 (add_may_alias_for_new_tag): Ditto.
220 (rewrite_alias_set_for): Rewrite for may_aliases being a bitmap.
221 (compute_is_aliased): New function.
222 (compute_may_aliases): Call compute_is_aliased).
223 (init_alias_info): Initialize alias_bitmap_obstack.
224 (union_alias_set_into): New function.
225 (compute_flow_sensitive_aliasing): Use union_aliases_into.
226 (have_common_aliases_p): Rewrite to take two bitmaps and use
228 (compute_flow_insensitive_aliasing): Stop using pointer-sets.
230 (finalize_ref_all_pointers): Update for add_may_alias changes.
231 (new_type_alias): Ditto.
232 * tree-flow-inline.h (may_aliases): Return a bitmap.
233 * tree-dfa.c (dump_variable): Check for MTAG_P'ness.
234 * tree-ssa.c (verify_flow_insensitive_alias_info): Update for
235 may_aliases being a bitmap.
236 * tree-flow.h (struct var_ann_d): Remove may_aliases member.
237 may_aliases now returns a bitmap.
238 * tree-ssa-structalias.c (merge_smts_into): Update for may_aliases
240 * tree-ssa-operands.c (add_virtual_operand): Update for
241 may_aliases being a bitmap.
243 2007-01-29 Daniel Berlin <dberlin@dberlin.org>
245 PR tree-optimization/30630
246 * tree-ssa-structalias.c (do_complex_constraint): Mark correct
249 2007-01-29 Simon Martin <simartin@users.sourceforge.net>
252 * gimplify.c (gimplify_target_expr): Make sure that the TARGET_EXPR is
253 expanded only once even if an error occurs.
255 2007-01-29 Ben Elliston <bje@au.ibm.com>
257 * gcov-io.h (__gcov_indirect_call_profiler): Declare.
258 (__gcov_average_profiler): Likewise.
259 (__gcov_ior_profiler): Likewise.
260 (__gcov_merge_ior): Likewise.
262 2007-01-28 Jan Hubicka <jh@suse.cz>
264 * builtins.c (expand_builtin_memset): Fix typo in my last patch.
265 * value-prof.c (stringop_block_profile): Likewise.
267 2007-01-28 Jan Hubicka <jh@suse.cz>
269 * expr.c (emit_block_move_via_movmem, emit_block_move_via_libcall): Add
270 variant handling histograms; add wrapper.
271 (clear_storage_via_libcall): Export.
272 (emit_block_move_hints): Break out from ...; add histograms.
273 (emit_block_move): ... this one.
274 (clear_storage_hints): Break out from ...; add histograms.
275 (clear_storage): ... this one.
276 (set_storage_via_memset): Handle histogram.
277 * expr.h (emit_block_move_via_libcall, emit_block_move_hints): Declare.
278 (clear_storage_hints, clear_storage_via_libcall): Declare.
279 (set_storage_via_setmem): Update prototype.
280 * doc/md.texi (movmem, setmem): Document new arguments.
282 * value-prof.c (dump_histogram_value, tree_find_values_to_profile): Add
284 (stringop_block_profile): New global function.
285 (tree_stringops_values_to_profile): Profile block size and alignment.
286 * value-prof.h (enum hist_type): add HIST_TYPE_AVERAGE and
288 (struct profile_hooks): Add gen_average_profiler and gen_ior_profiler.
289 (stringop_block_profile): Declare.
290 * builtins.c: Include value-prof.h.
291 (expand_builtin_memcpy, expand_builtin_memset): Pass block profile.
292 * gcov-ui.h (GCOV_COUNTER_NAMES): Add new counter.
293 (GCOV_COUNTER_AVERAGE, GCOV_COUNTER_IOR): New constants.
294 (GCOV_COUNTERS, GCOV_LAST_VALUE_COUNTER): Update.
295 * profile.c (instrument_values): Add new counters.
296 * cfgexpand.c (expand_gimple_basic_block): Propagate histograms to
298 * tree-profile.c (tree_average_profiler_fn, tree_ior_profiler_fn): New.
299 (tree_init_edge_profiler): Build new profilers.
300 (tree_gen_average_profiler, tree_gen_ior_profiler): New.
301 (pass_tree_profile): Add dump.
302 (tree_profile_hooks): Update.
303 * Makefile.in (LIBGCOV): Add new constants.
304 * libgcov.c (__gcov_merge_ior, __gcov_average_profiler,
305 __gcov_ior_profiler): New.
306 * i386.md (movmem/setmem expanders): Add new optional arguments.
308 2007-01-28 David Edelsohn <edelsohn@gnu.org>
310 * doc/md.texi (Standard Pattern Names): Document blockage pattern.
312 2007-01-28 Zdenek Dvorak <dvorakz@suse.cz>
314 * tree-ssa-loop-unswitch.c: Include tree-inline.h.
315 (tree_unswitch_single_loop): Pass eni_size_weights to
317 * tree-ssa-loop-manip.c: Include tree-inline.h.
318 (can_unroll_loop_p): Pass eni_size_weights to
320 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p):
321 Pass eni_size_weights to estimate_num_insns.
322 * tree.h (init_inline_once): Export.
323 * toplev.c (backend_init): Call init_inline_once.
324 * cgraphunit.c (cgraph_process_new_functions,
325 cgraph_analyze_function): Pass eni_inlining_weights to
327 * ipa-inline.c (compute_inline_parameters): Ditto.
328 * tree-ssa-loop-ivcanon.c (tree_num_loop_insns): Pass weights
329 to estimate_num_insns.
330 (try_unroll_loop_completely): Pass eni_size_weights to
332 * tree-eh.c (decide_copy_try_finally): Pass eni_size_weights
333 ot estimate_num_insns.
334 * tree-ssa-loop-prefetch.c: Include tree-inline.h.
335 (loop_prefetch_arrays): Pass eni_time_weights to tree_num_loop_insns.
336 * tree-inline.c (eni_inlining_weights, eni_size_weights,
337 eni_time_weights): New variables.
338 (init_inline_once): Initialize them.
339 (struct eni_data): Mew.
340 (estimate_num_insns_1, estimate_num_insns): Use weights.
341 * tree-inline.h (struct eni_weights_d): New.
342 (eni_inlining_weights, eni_size_weights, eni_time_weights): Declare.
343 (estimate_num_insns): Declaration changed.
344 * cfgloop.h (tree_num_loop_insns): Declaration changed.
345 * Makefile.in (tree-ssa-loop-unswitch.o, tree-ssa-loop-prefetch.o,
346 tree-ssa-loop-manip.o): Add TREE_INLINE_H dependency.
348 2007-01-28 Zdenek Dvorak <dvorakz@suse.cz>
350 * tree-data-ref.c (conflict_fn): Assert that the number of affine
351 relations in the conflict function is valid.
353 2007-01-27 Ian Lance Taylor <iant@google.com>
355 * common.opt: Add fstrict-overflow.
356 * opts.c (decode_options): Set flag_strict_overflow if -O2.
357 * flags.h (TYPE_OVERFLOW_WRAPS): Define.
358 (TYPE_OVERFLOW_UNDEFINED): Define.
359 (TYPE_OVERFLOW_TRAPS): Define. This replaces TYPE_TRAP_SIGNED.
361 * tree.h (TYPE_TRAP_SIGNED): Don't define.
362 * fold-const.c (negate_expr_p): Use TYPE_OVERFLOW_UNDEFINED.
363 (fold_negate_expr): Likewise.
364 (make_range): Likewise.
365 (extract_muldiv_1): Likewise.
366 (maybe_canonicalize_comparison): Likewise.
367 (fold_comparison): Likewise.
368 (fold_binary): Likewise.
369 (tree_expr_nonnegative_p): Likewise.
370 (tree_expr_nonzero_p): Likewise.
371 * tree-vrp.c (compare_values): Likewise.
372 (extract_range_from_binary_expr): Likewise.
373 (extract_range_from_unary_expr): Likewise.
374 * tree-ssa-loop-niter.c (infer_loop_bounds_from_signedness):
376 (nowrap_type_p): Likewise.
377 * tree-scalar-evolution.c (simple_iv): Likewise.
378 * fold-const.c (negate_expr_p): Use TYPE_OVERFLOW_WRAPS.
379 (build_range_check): Likewise.
380 (extract_muldiv_1): Likewise.
381 (fold_comparison): Likewise.
382 * tree-vrp.c (vrp_int_const_binop): Likewise.
383 (extract_range_from_unary_expr): Likewise.
384 * convert.c (convert_to_integer): Likewise.
385 * fold-const.c (fold_negate_expr): Use TYPE_OVERFLOW_TRAPS.
386 (fold_comparison): Likewise.
387 (fold_binary): Likewise.
388 * optabs.c (optab_for_tree_code): Likewise.
389 * tree-vectorizer.c (vect_is_simple_reduction): Likewise.
390 * simplify-rtx.c (simplify_const_relational_operation): Check
391 flag_strict_overflow and flag_trapv.
392 (simplify_const_relational_operation): Likewise.
393 * doc/invoke.texi (Option Summary): Mention -fstrict-overflow.
394 (Optimize Options): Add -fstrict-overflow to -O2 list. Document
397 2007-01-27 Roger Sayle <roger@eyesopen.com>
399 * tree.c (tree_fold_gcd): Delete.
400 * tree.h (tree_fold_gcd): Remove prototype.
401 * tree-data-ref.c (tree_fold_divides_p): Don't use tree_fold_gcd to
402 test whether one constant integer is a multiple of another. Instead
403 call int_const_binop with TRUNC_MOD_EXPR and test for a zero result.
404 * fold-const.c (multiple_of_p): We've determined both TOP and
405 BOTTOM are integer constants so we can call int_const_binop directly
406 instead of the more generic const_binop.
408 2007-01-27 Roger Sayle <roger@eyesopen.com>
410 * fold-const.c (size_binop): In the fast-paths for X+0, 0+X, X-0 and
411 1*X check that the constant hasn't overflowed, to preserve the
413 (round_up): Provide an efficient implementation when rouding-up an
414 INTEGER_CST to a power-of-two.
416 2007-01-28 Ralf Wildenhues <Ralf.Wildenhues@gmx.de>
418 * doc/sourcebuild.texi: Add comma for clarity.
419 * doc/extend.texi: Fix some typos.
420 * doc/passes.texi: Likewise.
421 * doc/cppinternals.texi: Likewise.
422 * doc/c-tree.texi: Likewise.
423 * doc/tree-ssa.texi: Likewise.
424 * doc/install.texi: Likewise.
426 2007-01-27 Jan Hubicka <jh@suse.cz>
428 * tree-sra.c (sra_walk_function): Don't rely on aliases being build.
429 (pass_sra): Do not require alias information.
430 * passes.c (init_optimization_passes): Add SRA
432 2007-01-27 Steven Bosscher <steven@gcc.gnu.org>
434 * tracer.c (rest_of_handle_tracer): We already cleaned
435 up the CFG in tracer() so don't do it here again.
436 * cfgcleanup.c (rest_of_handle_jump2): Don't repeat
437 cleanup_cfg here, either. And don't call renumber_insns.
439 * cfgrtl.c (rtl_verify_flow_info_1): Don't verify that BB_END
440 and BB_HEAD are in the insn stream here. Instead make sure
441 that BB_INSN is valid on all insns. Also, do check here that
442 there are no pending branch predictions...
443 (rtl_verify_flow_info): ...instead of doing it here. Checks
444 for BB_END and BB_HEAD moved from rtl_verify_flow_info_1 to
447 2007-01-26 Roger Sayle <roger@eyesopen.com>
449 * config/i386/i386.c (ix86_swap_binary_operands_p): New helper
450 function to simplify/factorize operand order canonicalization.
451 (ix86_fixup_binary_operands): Reorganize using the above function.
452 (ix86_binary_operator_ok): Likewise.
454 2007-01-27 Jakub Jelinek <jakub@redhat.com>
456 * genattrtab.c (struct attr_value_list, insn_code_values): Move to
457 file scope from optimize_attrs.
458 (simplify_test_exp): If insn_code_values is not NULL, use it to speed
460 (optimize_attrs): Clear insn_code_values after freeing it.
462 2007-01-26 Zdenek Dvorak <dvorakz@suse.cz>
464 * tree-ssa-address.c (create_mem_ref): Remove ", bsi" from
465 a parts.base assignment.
467 2007-01-26 Zdenek Dvorak <dvorakz@suse.cz>
469 * tree-data-ref.c (dump_subscript): Use dump_conflict_function.
470 (compute_subscript_distance, initialize_data_dependence_relation,
471 finalize_ddr_dependent, analyze_ziv_subscript,
472 analyze_siv_subscript_cst_affine,
473 compute_overlap_steps_for_affine_univar,
474 compute_overlap_steps_for_affine_1_2, analyze_subscript_affine_affine,
475 analyze_siv_subscript, analyze_miv_subscript,
476 analyze_overlapping_iterations, subscript_dependence_tester_1,
477 compute_self_dependence, free_dependence_relation): Work
478 with affine_fn instead of chrecs.
479 (dump_affine_function, dump_conflict_function, affine_function_equal_p,
480 common_affine_function, affine_function_base,
481 affine_function_constant_p, affine_fn_op, affine_fn_plus,
482 affine_fn_minus, affine_fn_free, conflict_fn_not_known,
483 conflict_fn_no_dependence, free_conflict_function, free_subscripts,
484 conflict_fn, affine_fn_cst, affine_fn_univar): New functions.
485 (all_chrecs_equal_p): Removed.
486 * tree-data-ref.h (affine_fn, conflict_function): New types.
487 (struct subscript): Change type of conflicting_iterations_in_a
488 and conflicting_iterations_in_b.
490 2007-01-26 Steve Ellcey <sje@cup.hp.com>
493 * config/pa/pa.h (TARGET_HPUX_11): New.
494 * config/pa/pa-hpux11.h (TARGET_HPUX_11): New.
495 * config/pa/pa.c (pa_init_builtins): Use TARGET_HPUX_11.
497 2007-01-26 Daniel Berlin <dberlin@dberlin.org>
498 Richard Guenther <rguenther@suse.de>
500 * tree-ssa-structalias.c (solve_graph): Handle case
501 we merged the variable to another.
503 2007-01-25 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
505 * builtins.c (fold_builtin_1): Treat ccos and ccosh as 'even'
508 * fold-const.c (negate_mathfn_p): Treat casin, casinh, catan,
509 catanh, cproj, csin, csinh, ctan and ctanh as 'odd' functions.
511 2007-01-25 DJ Delorie <dj@redhat.com>
513 * config/m32c/m32c.c (m32c_cannot_change_mode_class): We don't
514 allow changes to modes which don't fit in those registers.
516 * reload1.c (choose_reload_regs): Check for invalid subregs before
517 computing their locations, not after.
519 2007-01-25 Geoffrey Keating <geoffk@apple.com>
522 * config/rs6000/rs6000.c (first_altivec_reg_to_save): On Darwin,
523 save Altivec registers in an eh_return function.
524 (compute_vrsave_mask): Likewise.
525 (rs6000_stack_info): Correct AIX/Darwin stack alignment computation
526 for saving Altivec registers.
527 (rs6000_emit_prologue): Don't allocate stack twice in
528 eh_return function. Correct expected value of altivec_save_offset
529 when using save_world. Describe save of R0 to stack when using
530 save_world. Describe stack pointer adjustment when using
531 save_world. Remove duplicated eh_return parameter register saving.
532 Update sp_offset variable after save_world.
533 * config/rs6000/t-darwin (LIB2FUNCS_STATIC_EXTRA): Remove
535 (LIB2FUNCS_EXTRA): Add darwin-world.asm.
536 * config/rs6000/darwin.h (SUBTARGET_OVERRIDE_OPTIONS): -m64
539 2007-01-25 Steve Ellcey <sje@cup.hp.com>
541 * config.gcc (ia64*-*-hpux*): Make posix threads the default.
543 2007-01-25 Steve Ellcey <sje@cup.hp.com>
546 * config/pa/pa.c (pa_init_builtins): Set asm names for finite routines.
547 * config/ia64/ia64.c (ia64_init_builtins): Ditto.
549 2007-01-25 Richard Guenther <rguenther@suse.de>
551 * doc/invoke.texi (-Wcoverage-mismatch): Document.
552 * common.opt (-Wcoverage-mismatch): New warning option.
553 * coverage.c (get_coverage_counts): Ignore coverage mismatch
554 if -Wcoverage-mismatch is given.
556 2007-01-25 Razya Ladelsky <razya@il.ibm.com>
558 * ipa-cp.c (ipcp_insert_stage, ipcp_driver): Support for SSA.
559 (ipcp_driver): Change to static definition.
560 Add dumping of the ifunctions.
561 (constant_val_insert): Remove unused parameter. Support for SSA.
562 (ipcp_propagate_const): Support for SSA.
563 (ipcp_profile_bb_print): Print only analyzed nodes.
564 (ipcp_replace_map_create): Remove support for Fortran constant
566 * ipa-prop.c (ipa_method_modify_stmt,
567 ipa_callsite_compute_param): Support for SSA.
568 * ipa-prop.h (ipcp_driver): Remove declaration.
569 (IS_VALID_TREE_MAP_INDEX): Add define.
571 2007-01-24 Geoffrey Keating <geoffk@apple.com>
573 * unwind-dw2.c (execute_stack_op): Handle DW_OP_swap.
575 2007-01-24 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
578 * builtins.c (fold_builtin_cabs): Use MPFR to evaluate a
579 constant argument to cabs and do it without checking for
580 -funsafe-math-optimizations.
582 2007-01-24 Douglas Gregor <dgregor@osl.iu.edu>
584 * c-common.h (RID_FIRST_CXX0X): New.
585 (RID_LAST_CXX0X): New.
586 * c-opts.c (c_common_handle_option): -Wc++0x-compat is triggered
588 * c.opt (Wc++0x-compat): New.
589 * doc/invoke.texi (-Wc++0x-compat): Document.
591 2007-01-24 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
593 * builtins.c (fold_builtin_carg): New.
594 (fold_builtin_1): Use it.
596 2007-01-24 Jan Hubicka <jh@suse.cz>
598 * ipa-inline.c (cgraph_decide_inlining): Initialize initial_insns.
600 * ipa-inline.c (initial_insns, max_insns): Delete.
601 (compute_max_insns): New function.
602 (cgraph_decide_inlining_of_small_function): Use it; take minimal amount
603 of insns as base for code growth.
604 (cgraph_decide_inlining): Make initial_insns local; do not compute
606 * params.def (PARAM_INLINE_UNIT_GROWTH): Set to 60.
607 * doc/invoke.texi (inline-unit-growth): Update docs.
609 2007-01-24 Jakub Jelinek <jakub@redhat.com>
611 * config/i386/i386.h (x86_cmpxchg16b): Remove const.
612 (TARGET_CMPXCHG16B): Define to x86_cmpxchg16b.
613 * config/i386/i386.c (x86_cmpxchg16b): Remove const.
614 (override_options): Add PTA_CX16 flag. Set x86_cmpxchg16b
615 for CPUs that have PTA_CX16 set.
618 * gimplify.c (omp_check_private): New function.
619 (gimplify_scan_omp_clauses): Use it for
620 firstprivate/lastprivate/reduction.
623 * gimplify.c (omp_add_variable): Don't call omp_notice_variable
624 on TYPE_SIZE_UNIT for GOVD_LOCAL VLAs.
627 * omp-low.c (lower_omp_for_lastprivate): Add dlist argument.
628 If lower_lastprivate_clauses emits some statements, append them
629 to dlist rather than body_p and to body_p append an initializer.
630 (lower_omp_for): Adjust caller.
632 2007-01-24 Steve Ellcey <sje@cup.hp.com>
634 * target.h (globalize_decl_name): New.
635 * target-def.h (TARGET_ASM_GLOBALIZE_DECL_NAME): New.
636 * output.h (default_globalize_decl_name): New.
637 * varasm.c (asm_output_bss): Use globalize_decl_name instead of
639 (globalize_decl): Ditto.
640 (default_globalize_decl_name): New.
641 * config/ia64/ia64.c (ia64_globalize_decl_name): New.
642 (ia64_handle_version_id_attribute): New.
643 (TARGET_ASM_GLOBALIZE_DECL_NAME): New.
644 (ia64_asm_output_external): Use globalize_decl_name instead
646 * doc/extend.texi (version_id): New pragma.
647 * doc/tm.texi (ARGET_ASM_GLOBALIZE_DECL_NAME): New target hook.
649 2007-01-24 Andreas Krebbel <krebbel1@de.ibm.com>
651 * unwind-dw2-fde.c (get_cie_encoding): Replaced _Unwind_Word with
652 _uleb128_t and _Unwind_SWord with _sleb128_t.
653 * unwind-dw2.c (extract_cie_info, execute_stack_op, execute_cfa_program,
654 uw_frame_state_for, uw_update_context_1): Likewise.
655 * unwind-c.c (parse_lsda_header, PERSONALITY_FUNCTION): Likewise.
656 * unwind-pe.h (read_uleb128, read_sleb128,
657 read_encoded_value_with_base): Likewise.
658 * unwind-generic.h: Define _sleb128_t and _uleb128_t types.
660 2007-01-24 Richard Guenther <rguenther@suse.de>
662 * builtins.c (expand_builtin_cexpi): Get the fndecl
663 for cexp in the correct way.
665 2007-01-24 Jan Hubicka <jh@suse.cz>
667 * tree-ssa-dce.c (eliminate_unnecesary_stmts): Remove dead LHS of calls.
669 2007-01-24 Andreas Krebbel <krebbel1@de.ibm.com>
671 * c-cppbuiltin.c (builtin_define_type_sizeof): New function.
672 (c_cpp_builtins): New builtin macros: __SIZEOF_INT__, __SIZEOF_LONG__,
673 __SIZEOF_LONG_LONG__, __SIZEOF_SHORT__, __SIZEOF_POINTER__,
674 __SIZEOF_FLOAT__, __SIZEOF_DOUBLE__, __SIZEOF_LONG_DOUBLE__,
675 __SIZEOF_SIZE_T__, __SIZEOF_WCHAR_T__, __SIZEOF_WINT_T__ and
676 __SIZEOF_PTRDIFF_T__.
677 * doc/cpp.texi: Documentation for the new builtin macros added.
679 2007-01-24 Uros Bizjak <ubizjak@gmail.com>
681 * config/i386/i386.md (tanxf2, tan<mode>2, atan<mode>2, log<mode>2,
682 log10<mode>2, log2<mode>2, expxf2, exp10xf2, exp2xf2): Use op2
683 instead of operands[2] to avoid access past the end of array.
685 2007-01-24 Richard Sandiford <richard@codesourcery.com>
687 * reload1.c (emit_reload_insns): Pass the reload register
688 for a non-spill output reload through forget_old_reloads_1.
690 2007-01-23 Joseph Myers <joseph@codesourcery.com>
692 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Define
693 _SOFT_DOUBLE if doubles use software floating-point.
694 * config/rs6000/libgcc-ppc-glibc.ver: Export additional long
695 double functions if _SOFT_DOUBLE, not _SOFT_FLOAT.
696 * config/rs6000/darwin-ldouble.c: Also compile functions for
697 hard-float without FPRs. Use fmsub function for all __NO_FPRS__
698 cases. Compile extra functions if _SOFT_DOUBLE, not _SOFT_FLOAT.
699 * config/rs6000/linuxspe.h (SUBSUBTARGET_OVERRIDE_OPTIONS): Remove
700 commented-out long double override.
701 (CPP_LONGDOUBLE_DEFAULT_SPEC): Likewise.
702 * config/rs6000/eabispe.h: Likewise.
703 * config/rs6000/rs6000.c (rs6000_override_options): Don't override
704 long double for non-SPE.
705 (rs6000_handle_option): Likewise.
706 (invalid_e500_subreg): Disallow more subregs involding DImode,
707 DFmode, TImode or TFmode.
708 (rs6000_legitimate_offset_address_p): Check TFmode offsets for
710 (legitimate_lo_sum_address_p): Also check for TFmode for E500
712 (rs6000_legitimize_address): Also handle TFmode for E500 double.
713 (rs6000_legitimize_reload_address): Also handle TFmode for E500
715 (rs6000_legitimate_address): Also check for TFmode for E500
717 (rs6000_emit_move): Use DFmode subregs of TFmode for E500 double.
718 (spe_build_register_parallel): Handle TFmode and TCmode.
719 (rs6000_spe_function_arg): Handle TFmode and TCmode for E500
721 (function_arg): Handle TFmode and TCmode for E500 double.
722 (rs6000_init_libfuncs): Initialize extra libfuncs for soft double
724 (print_operand): Handle TFmode and TImode for %y.
725 (rs6000_generate_compare): Handle TFmode comparisons for E500
727 (spe_func_has_64bit_regs_p): Check for TFmode for E500 double.
728 (rs6000_function_value): Handle TFmode and TCmode for E500 double.
729 (rs6000_libcall_value): Handle TFmode and TCmode for E500 double.
730 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Check for
731 TFmode for E500 double.
732 * config/rs6000/rs6000.md (FP): Allow TF for E500 double.
733 (floatsidf2): Enable for E500 double.
734 (movtf_softfloat): Use rs6000_nonimmediate_operand.
735 (extenddftf2): Change to extenddftf2_fprs.
736 (extenddftf2): Call gen_spe_extenddftf2 or gen_extenddftf2_fprs
737 depending on TARGET_E500_DOUBLE.
738 (extendsftf2): Enable for E500 double.
739 (trunctfdf2): Enable for E500 double.
740 (trunctfsf2): Change to trunctfsf2_fprs.
741 (trunctfsf2): Call gen_spe_trunctfsf2 or gen_trunctfsf2_fprs
742 depending on TARGET_E500_DOUBLE.
743 (floatsitf2): Enable for E500 double.
744 (fix_trunctfsi2): Change to fix_trunctfsi2_fprs.
745 (fix_trunctfsi2): Call gen_spe_fix_trunctfsi2 or
746 gen_fix_trunctfsi2_fprs depending on TARGET_E500_DOUBLE.
747 (negtf2): Change to negtf2_internal.
748 (negtf2): New expander.
749 (abstf2): Enable for E500 double. Call gen_spe_abstf2_tst,
750 gen_spe_abstf2_cmp or gen_abstf2_internal depending on
751 TARGET_E500_DOUBLE and flag_unsafe_math_optimizations.
752 (movdi_internal32): Use rs6000_nonimmediate_operand.
753 (unnamed splitter): Likewise.
754 * config/rs6000/spe.md (CMPTFEQ_GPR, TSTTFEQ_GPR, CMPTFGT_GPR,
755 TSTTFGT_GPR, CMPTFLT_GPR, TSTTFLT_GPR): New unspecs.
756 (SPE64TF, DITI): New mode macros.
757 (frob_df_di): Change to frob_<SPE64:mode>_<DITI:mode>; allow more
760 (frob_<mode>_di_2): New.
761 (frob_tf_di_8_2): New.
762 (frob_di_df): Change to frob_di_<mode>; allow more modes.
764 (frob_di_df_2): Change to frob_<DITI:mode>_<SPE64:mode>_2; allow
766 (frob_ti_<mode>_8_2): New.
768 (mov_si<mode>_e500_subreg0, mov_si<mode>_e500_subreg0_2,
769 mov_si<mode>_e500_subreg4, mov_si<mode>_e500_subreg4_2): Allow
771 (mov_sitf_e500_subreg8, mov_sitf_e500_subreg8_2,
772 mov_sitf_e500_subreg12, mov_sitf_e500_subreg12_2): New.
773 (spe_trunctfdf2_internal1, spe_trunctfsf2, spe_extenddftf2,
774 spe_fix_trunctfsi2, spe_fix_trunctfsi2_internal,
775 spe_negtf2_internal, spe_abstf2_cmp, spe_abstf2_tst): New.
776 (cmptfeq_gpr, tsttfeq_gpr, cmptfgt_gpr, tsttfgt_gpr, cmptflt_gpr,
779 2007-01-23 Ian Lance Taylor <iant@google.com>
781 * Makefile.in (OBJS-common): Reformat, alphabetize, but put
783 (OBJS-archive): Reformat, alphabetize.
784 (OBJS): Change out_object_file to OBJS-md.
786 2007-01-23 Uros Bizjak <ubizjak@gmail.com>
788 * config/i386/i386.md: Use REG_P, MEM_P, CONST_INT_P, LABEL_P,
789 JUMP_P and CALL_P predicates where applicable.
790 * config/i386/i386.c: Ditto.
791 * config/i386/i386.md: Ditto.
792 * config/i386/mmx.md: Ditto.
793 * config/i386/predicates.md: Ditto.
795 2007-01-22 Andreas Schwab <schwab@suse.de>
797 * config/m68k/m68k.h: Fix comment.
799 2007-01-22 Jan Hubicka <jh@suse.cz>
801 * passes.c (init_optimization_passes): Do not rerun
802 pass_early_warn_uninitialized.
804 2007-01-22 Richard Guenther <rguenther@suse.de>
806 PR tree-optimization/30038
807 * tree-ssa-math-opts.c (maybe_record_sincos): New static helper
809 (execute_cse_sincos_1): Likewise.
810 (execute_cse_sincos): Likewise.
811 (gate_cse_sincos): Likewise.
812 (pass_cse_sincos): New pass CSEing sin() and cos() calls using
813 the cexpi() canonicalization of sincos().
814 * tree-pass.h (pass_cse_sincos): Declare.
815 * passes.c (init_optimization_passes): New pass pas_cse_sincos.
817 2007-01-21 Eric Botcazou <ebotcazou@libertysurf.fr>
819 PR rtl-optimization/29329
820 * combine.c (replaced_rhs_insn): Rename to i2mod.
821 (replaced_rhs_value): Rename to i2mod_new_rhs.
822 (i2mod_old_rhs): New global variable.
823 (combine_instructions): Adjust for above change. Save a copy of
824 the old RHS into i2mod_old_rhs when the contents of a REG_EQUAL
825 note are substituted in the second instruction.
826 (distribute_notes) <REG_DEAD>: Adjust for above change. Do not
827 ditch the note if it pertains to the second eliminated register
828 and this register is mentioned in i2mod_old_rhs.
831 2006-09-12 Eric Botcazou <ebotcazou@libertysurf.fr>
833 * combine.c (distribute_notes) <REG_DEAD>: Do not consider SETs past
834 the insn to which the note was originally attached.
836 2007-01-21 Jan Hubicka <jh@suse.cz>
838 * ipa-inline.c (inlining_mode): Comment, move up.
839 (cgraph_decide_inlining_incrementally): Do not perform inlining
840 itself; fix handling of flattening of self recursive functions.
841 (cgraph_find_cycles): Remove.
842 (cgraph_flatten_node): Remove.
843 (cgraph_decide_inlining): Use incremental inliner to handle flattening.
844 (try_inline): New function.
845 (cgraph_early_inlining): Update call of
846 cgraph_decide_inlining_incrementally. Apply inlining here.
847 (apply_inline): Update call of cgraph_decide_inlining_incrementally.
849 2007-01-21 Dirk Mueller <dmueller@suse.de>
852 * tree-vrp.c (check_array_bounds): do not warn
853 about ADDR_EXPR's of ARRAY_REF's which are immediately
854 used in binary expressions.
856 2007-01-21 Ira Rosen <irar@il.ibm.com>
858 * tree-vectorizer.h (struct _stmt_vec_info): Add new field
859 read_write_dep and macros for its access.
860 * tree-vectorizer.c (new_stmt_vec_info): Initialize the new field.
861 * tree-vect-analyze.c (vect_analyze_data_ref_dependence): Remove
862 argument, call vect_check_interleaving for every independent pair of
863 data-refs. Mark loads that access the same memory location as a store
865 (vect_check_dependences): Remove.
866 (vect_analyze_data_ref_dependences): Remove vect_check_dependences
867 call, fix the call to vect_analyze_data_ref_dependence.
868 (vect_analyze_data_ref_access): For statements that access the same
869 data-ref, check that they are not stores; for loads, check that there
870 is no store that access the same location.
872 2007-01-20 Roger Sayle <roger@eyesopen.com>
873 Joseph Myers <joseph@codesourcery.com>
875 * doc/invoke.texi (-fdump-rtl-combine): Fix under/overfull hbox.
876 (-fdump-rtl-stack): Likewise.
877 (-fno-signed-zeros): Use @minus{} for a minus sign. Correct typo.
878 (-mcheck-zero-division, -mcpu): Fix under/overfull hbox.
879 (-mpt-fixed): Use @minus{} for minus sign.
880 (Using Precompiled Headers): Fix under/overfull hbox.
882 2007-01-20 Jan Hubicka <jh@suse.cz>
884 * tree-flow.h (struct stmt_ann_d): Move references_memory to proper
885 place within annotation.
887 2007-01-20 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
889 * pa.c (output_move_double): Change array size of xoperands to 4.
891 2007-01-20 Jan Hubicka <jh@suse.cz>
893 * tree-tailcall.c (adjust_return_value): Do not use RESULT_DECL
896 2007-01-19 Ian Lance Taylor <iant@google.com>
898 * expmed.c (expand_divmod) [TRUNC_MOD_EXPR, TRUNC_DIV_EXPR]: Cast
899 constant to unsigned HOST_WIDE_INT before negating.
901 2007-01-19 Ian Lance Taylor <iant@google.com>
903 * tree-ssa-operands.h (struct vuse_vec_d): Change num_vuse field
905 (VUSE_VECT_ELEMENT) [ENABLE_CHECKING]: Use unsigned comparison.
906 (VUSE_ELEMENT_PTR) [ENABLE_CHECKING]: Likewise.
907 (SET_VUSE_VECT_ELEMENT) [ENABLE_CHECKING]: Likewise.
908 (SET_VUSE_ELEMENT_VAR) [ENABLE_CHECKING]: Likewise.
909 (SET_VUSE_ELEMENT_PTR) [ENABLE_CHECKING]: Likewise.
910 (realloc_vdef, realloc_vuse): Change second parameter to
912 (ssa_operand_iterator_d): Change vuse_index and mayuse_index
914 * tree-ssa-operands.c (realloc_vop): Change num_elem parameter to
915 unsigned. Change x and lim locals to unsigned.
916 (realloc_vdef, realloc_vuse): Change num_elem parameter to
918 (finalize_ssa_vuse_ops): Change old_i local to unsigned.
919 (copy_virtual_operands): Change i and n locals to unsigned.
921 2007-01-19 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
924 * builtins.c (fold_builtin_1): Handle builtin fdim.
926 2007-01-20 Jan Hubicka <jh@suse.cz>
928 * tree-ssa.c (init_tree_ssa): Do not call init_alias_heapvars.
929 * tree-ssa-structalias.c (compute_points_to_sets): Do call
931 (init_alias_heapvars): Initialize only when not already initialized.
932 (delete_alias_heapvars): Set heapvar_for_stmt to NULL.
934 2007-01-19 Roger Sayle <roger@eyesopen.com>
936 * common.opt (fsigned-zeros): New command line option.
937 * flags.h (HONOR_SIGNED_ZEROS): Control via flag_signed_zeros instead
938 of flag_unsafe_math_optimizations.
939 * opts.c (set_fast_math_flags): The -ffast-math command line option
940 implies -fno-signed-zeros.
941 (fast_math_flags_set_p): Likewise.
943 * doc/invoke.texi: Document new -fno-signed-zeros option, and update
944 the documentation of -ffast-math appropriately. Wrap long lines.
946 2007-01-19 Steve Ellcey <sje@cup.hp.com>
948 * system.h (ASM_MAKE_LABEL_LINKONCE): Poison.
949 * varasm.c (globalize_decl): Remove ASM_MAKE_LABEL_LINKONCE ifdef.
951 2007-01-19 Tomas Bily <tbily@suse.cz>
953 * cgraphunit.c (cgraph_finalize_function): Updating of pid
955 (tree_init_ic_make_global_vars): New function
956 (tree_init_edge_profiler): call of tree_init_ic_make_global_vars
957 (tree_gen_ic_profiler): New function
958 (tree_gen_ic_func_profiler): New function
959 (tree_profiling): Added calling of tree_gen_ic_func_profiler
960 (tree_profile_hooks): Added hook for indirec/virtual calls
961 * value-prof.c (tree_find_values_to_profile): New case for
963 (tree_values_to_profile): Call for determining indirect/virtual
965 (tree_indirect_call_to_profile): New function
966 (tree_ic_transform): New function
967 (tree_ic): New function
968 (find_func_by_pid): New function
969 (init_pid_map): New function
970 (tree_value_profile_transformations): Added check for
971 indirect/virtual call transformation
972 * value-prof.h (enum hist_type): New counter type for
973 indirect/virtual calls
974 (profile_hooks): Added new hook for profiling indirect/virtual
976 * profile.c (instrument_values): New case for indirect/virtual
978 * gcov-io.h (GCOV_LAST_VALUE_COUNTER): Changed to 6
979 (GCOV_COUNTER_V_INDIR): New counter type
980 (GCOV_COUNTER_NAMES): New name of counter "indirect" added
981 (GCOV_MERGE_FUNCTIONS): New merge function for indirect/virtual
983 * cgraph.c: Definition of cgraph_max_pid
984 (cgraph_create_node): Default init of pid attribute
985 * cgraph.h: Declaration of cgraph_max_pid
986 (struct cgraph_node): Added pid attribute
987 * libgcov.c (__gcov_indirect_call_profiler): New function
988 (__gcov_one_value_profiler_body): New function
989 (__gcov_one_value_profiler): Body was moved to
990 __gcov_one_value_profiler_body and calls it
992 2007-01-19 Basile Starynkevitch <basile@starynkevitch.net>
994 * doc/gty.texi (Options): Document the mark_hook option to GTY.
995 * gengtype.c (write_types_data, write_func_for_structure,
996 write_types, ggc_wtd, pch_wtd): Add skip_hooks to
997 write_types_data, ggc_wtd, pch_wtd for processing mark_hook.
998 (walk_type, write_func_for_structure): Generate the mark_hook if
1001 2007-01-19 Jan Hubicka <jh@suse.cz>
1003 * ipa-inline.c (cgraph_decide_inlining_incrementally): Instead of
1004 'early' argument take inlining mode argument specifying whether to
1005 inline for size/speeed or all functions; add support for flattening;
1007 (cgraph_early_inlining): Update call of decide_inlining_incrementally.
1009 2007-01-19 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
1012 * toplev.c (warn_deprecated_use): Use %qD instead of %qs to print
1013 the name of the declared identifier.
1015 2007-01-19 Dirk Mueller <dmueller@suse.de>
1017 * config/i386.h (CONDITIONAL_REGISTER_USAGE): Store
1018 result of PIC_OFFSET_TABLE_REGNUM in temporary variable to avoid
1019 duplicate evaluation.
1021 2007-01-19 Uros Bizjak <ubizjak@gmail.com>
1023 * config/i386/i386.md (acos<mode>2): Rename from acossf2 and acosdf2.
1024 Macroize expander using X87MODEF12 mode macro. Extend operand 1
1025 to XFMode, use acosxf2 and truncate result to requested mode.
1026 Use SSE_FLOAT_MODE_P to disable patterns for SSE math.
1027 (asin<mode>2): Similarly, with asin expanders.
1028 (*fscalexf4_i387): Rename from *fscalexf4.
1029 (expNcorexf3): New expander.
1030 (expxf2, exp10xf2, exp2xf2): Use expNcorexf3 expander.
1031 (exp<mode>2): Rename from expsf2 and expdf2. Macroize expander using
1032 X87MODEF12 mode macro. Extend operand 1 to XFMode, use expxf2 and
1033 truncate result to requested mode. Use SSE_FLOAT_MODE_P to disable
1034 patterns for SSE math.
1035 (exp10<mode>2): Similarly, with exp10 expanders.
1036 (exp2<mode>2): Similarly, with exp2 expanders.
1037 (expm1<mode>2): Similarly, with expm1 expanders.
1038 (ldexp<mode>3): Similarly, with ldexp expanders.
1039 (log<mode>2, log10<mode>2, log2<mode>2, log1p<mode>2, logb<mode>2):
1040 Use gen_truncxf<mode>2_i387_noop to truncate result.
1042 2007-01-19 Richard Sandiford <richard@codesourcery.com>
1044 * config/m68k/m68k.h (PREFERRED_STACK_BOUNDARY): Define to 32
1045 for ColdFire targets.
1047 2007-01-19 Nathan Sidwell <nathan@codesourcery.com>
1048 Richard Sandiford <richard@codesourcery.com>
1050 * config/m68k/m68k.h (M68K_STATIC_CHAIN_REG_NAME): New macro.
1051 (INITIALIZE_TRAMPOLINE): Use STATIC_CHAIN_REGNUM.
1052 (__transfer_from_trampoline): Use M68K_STATIC_CHAIN_REG_NAME.
1053 * config/m68k/m68kelf.h (STATIC_CHAIN_REGNUM): Override.
1054 (M68K_STATIC_CHAIN_REG_NAME): Likewise.
1055 * config/m68k/netbsd-elf.h (M68K_STATIC_CHAIN_REG_NAME): Likewise.
1057 2007-01-19 Richard Sandiford <richard@codesourcery.com>
1059 * config/m68k/m68k.md (adddi_dilshr32): Rename to...
1060 (*adddi_dilshr32): ...this. Fix formatting. Remove commented-out
1061 non-canonical pattern. Restrict to !TARGET_COLDFIRE.
1062 (*adddi_dilshr32_cf): New pattern.
1063 (adddi3, subdi3): Remove first alternatives.
1065 2007-01-19 Richard Sandiford <richard@codesourcery.com>
1067 * config/m68k/m68k.c (notice_update_cc): If an SFmode move is
1068 implemented using move.l, do not use its cc result for floating-point
1071 2007-01-19 Richard Sandiford <richard@codesourcery.com>
1073 * config/m68k/m68k.h (EXTRA_CONSTRAINT): Stop the 'T' constraint
1074 from accepting 's' constraints if flag_pic.
1076 2007-01-19 Richard Sandiford <richard@codesourcery.com>
1078 * config/m68k/m68k.md (bordered, bunordered, buneq, bunge, bungt)
1079 (bunle, bunlt, bltgt, bordered_rev, bunordered_rev, buneq_rev)
1080 (bunge_rev, bungt_rev, bunle_rev, bunlt_rev, bltgt_rev): Change
1081 condition from TARGET_68881 to TARGET_HARD_FLOAT.
1083 2007-01-19 Sandra Loosemore <sandra@codesourcery.com>
1085 * longlong.h (count_leading_zeros, COUNT_LEADING_ZEROS_0): Add
1086 ColdFire alternatives.
1087 * config/m68k/m68k.h (CLZ_DEFINED_VALUE_AT_ZERO): New macro.
1088 * config/m68k/m68k.md (clzsi2): Define for ColdFire
1089 architectures that support ff1 instruction.
1091 2007-01-19 Richard Sandiford <richard@codesourcery.com>
1092 Julian Brown <julian@codesourcery.com>
1094 * config/m68k/m68k.h (CONST_OK_FOR_LETTER_P): Add an 'R' case.
1095 * config/m68k/m68k.md (*movsi_cfv4): Fold into...
1096 (*movsi_cf): ...here. Remove unnecessary 'R' from 'Rg'.
1099 2007-01-19 Richard Sandiford <richard@codesourcery.com>
1101 * config/m68k/m68k-protos.h (valid_mov3q_const): Take a HOST_WIDE_INT
1103 (output_move_const_into_data_reg, output_move_simode_const): Delete.
1104 * config/m68k/m68k.c (const_method, const_int_cost): Take a
1105 HOST_WIDE_INT instead of an rtx.
1106 (m68k_rtx_costs): Update call accordingly.
1107 (output_move_const_into_data_reg): Likewise. Fix formatting.
1108 (valid_mov3q_const): Take a HOST_WIDE_INT instead of an rtx.
1110 (output_move_simode_const): Update calls after above changes.
1111 Rework to use automatic variables and predicates like MEM_P.
1112 * config/m68k/m68k.md (pushexthisi_const): Update call to
1115 2007-01-19 Dirk Mueller <dmueller@suse.de>
1117 * tree-ssa-alias.c (perform_var_substitution): Fix typo
1120 2007-01-19 Richard Guenther <rguenther@suse.de>
1122 * builtins.c (expand_builtin_cexpi): Fall back to expanding
1123 via cexp in case sincos is not available.
1125 2007-01-19 Richard Guenther <rguenther@suse.de>
1127 * doc/tm.texi (TARGET_HAS_SINCOS): Document new target macro.
1128 * defaults.h (TARGET_HAS_SINCOS): Default to off.
1129 * config/linux.h (TARGET_HAS_SINCOS): Set to on if we have glibc.
1130 * config/alpha/linux.h (TARGET_HAS_SINCOS): Likewise.
1131 * config/sparc/linux.h (TARGET_HAS_SINCOS): Likewise.
1132 * config/sparc/linux64.h (TARGET_HAS_SINCOS): Likewise.
1133 * config/rs6000/linux.h (TARGET_HAS_SINCOS): Likewise.
1134 * config/rs6000/linux64.h (TARGET_HAS_SINCOS): Likewise.
1136 2007-01-19 Uros Bizjak <ubizjak@gmail.com>
1138 * config/i386/i386.md (*fpatanxf3_i387, fpatan_extend<mode>xf3_i387):
1140 (atan2sf3_1, atan2df3_1, atan2xf3_1): Remove insn patterns.
1141 (atan2xf3): Directly generate RTL pattern.
1142 (atan2<mode>3): Rename from atan2sf3 and atan2df3 and macroize insn
1143 patterns using X87MODEF12 mode macro. Use fpatan_extend<mode>xf3_i387
1144 and truncate result to requested mode. Use SSE_FLOAT_MODE_P to
1145 disable patterns for SSE math.
1146 (atan<mode>2): Rename from atansf2 and atandf2 and macroize insn
1147 patterns using X87MODEF12 mode macro. Use fpatan_extend<mode>xf3_i387
1148 and truncate result to requested mode. Use SSE_FLOAT_MODE_P to
1149 disable patterns for SSE math.
1151 2007-01-19 Alexandre Oliva <aoliva@redhat.com>
1153 * libgcc-std.ver: Fix typo in %inherit for GCC_4.3.0.
1155 2007-01-18 Roger Sayle <roger@eyesopen.com>
1157 * fold-const.c (fold_unary) <VIEW_CONVERT_EXPR>: Optimize away a
1158 VIEW_CONVERT_EXPR to the same type as it's operand.
1160 2007-01-18 David Edelsohn <edelsohn@gnu.org>
1162 * config/rs6000/darwin-ldouble.c: Only build _SOFT_FLOAT if
1163 configured for long double 128.
1165 2007-01-18 Mike Stump <mrs@apple.com>
1167 * config/rs6000/rs6000.c (rs6000_emit_vector_compare): Fix build
1170 2007-01-18 Michael Meissner <michael.meissner@amd.com>
1172 * i386.c (ix86_compute_frame_layout): Make fprintf's in #if 0 code
1175 2007-01-18 Jan Hubicka <jh@suse.cz>
1177 * tree-ssa-operands.c (vop_free_bucket_size): Never return value
1178 greater than NUM_VOP_FREE_BUCKETS.
1180 2007-01-18 Daniel Berlin <dberlin@dberlin.org>
1182 * tree-ssa-structalias.c: Update comments.
1183 (ptabitmap_obstack): Removed.
1185 (oldpta_obstack): Ditto.
1186 (stats): Add a few members.
1187 (struct variable_info): Remove node, complex, address_taken, and
1188 indirect_target members. Add oldsolution member.
1189 (new_var_info): Do not initialize removed members.
1190 (constraint_expr_type): Remove INCLUDES.
1191 (constraint_graph): Add size, implicit_preds, rep,
1192 indirect_cycles, eq_rep, label, direct_nodes, and complex members.
1193 (FIRST_REF_NODE): New macro.
1194 (LAST_REF_NODE): Ditto.
1195 (FIRST_ADDR_NODE): Ditto.
1196 (find): New function.
1198 (dump_constraint): Do not handle INCLUDES.
1199 (insert_into_complex): Do not insert duplicate constraints.
1200 (condense_varmap_nodes): Renamed and rewritten into ...
1201 (merge_node_constraints): This. Also fix bug in handling of
1202 offseted copy constraints.
1203 (clear_edges_for_node): No longer need to deal with preds at all,
1204 or removing associated preds/succs.
1205 (merge_graph_nodes): Deal with indirect_cycles.
1206 Don't deal with predecessors.
1207 (add_implicit_graph_edge): New function.
1208 (add_pred_graph_edge): Ditto.
1209 (add_graph_edge): Don't deal with predecessors.
1210 (build_constraint_graph): Removed.
1211 (build_pred_graph): New function.
1212 (build_succ_graph): Ditto.
1213 (struct scc_info): Removed in_component. Added roots, dfs, and
1214 node_mapping. Remove visited_index, unification_queue.
1215 (scc_visit): Deal with union-find we do now.
1216 Deal with cycles with REF nodes.
1217 (collapse_nodes): Renamed and rewritten to ...
1218 (unify_nodes): This.
1219 (process_unification_queue): Removed.
1220 (topo_visit): Cleanup
1221 (do_da_constraint): Use find.
1222 (do_sd_constraint): Ditto.
1223 (do_ds_constraint): Ditto.
1224 (do_complex_constraint): Ditto.
1225 (init_scc_info): Update for removed and added members.
1226 (find_and_collapse_graph_cycles): Renamed and rewritten into ...
1227 (find_indirect_cycles): This.
1228 (equivalence_class): New variable.
1229 (label_visit): New function.
1230 (perform_variable_substitution): Rewritten.
1231 (free_var_substitution_info): New function.
1232 (find_equivalent_node): Ditto.
1233 (move_complex_constraints): Ditto.
1234 (eliminate_indirect_cycles): Ditto.
1235 (solve_graph): Only propagate changed bits.
1236 Use indirect cycle elimination.
1238 (tree_id_t): Rename to tree_vi_t, delete id member, add vi member.
1239 (tree_id_eq): Renamed to ...
1240 (tree_vi_eq): This. Update for member change
1241 (insert_id_for_tree): Renamed and rewritten to ...
1242 (insert_vi_for_tree): This.
1243 (lookup_id_for_tree): Renamed and rewritten to ...
1244 (lookup_vi_for_tree): This.
1245 (get_id_for_tree): Renamed and rewritten to ...
1246 (get_vi_for_tree): Ditto.
1247 (get_constraint_exp_from_ssa_var): Update to use get_vi_for_tree.
1248 (process_constraint): Don't handle INCLUDES.
1249 Remove special ADDRESSOF case.
1250 (find_func_aliases): Rewrite to use vi functions instead of id
1252 (create_function_info_for): Ditto.
1253 (create_variable_info_for): Ditto.
1254 (intra_create_variable_infos): Ditto.
1255 (merge_smts_into): Ditto.
1256 (find_what_p_points_to): Ditto.
1257 (init_base_vars): Ditto.
1258 (init_alias_vars): Ditto.
1259 (remove_preds_and_fake_succs): New function.
1260 (dump_sa_points_to_info): Dump new stats.
1261 (dump_solution_for_var): Use find.
1262 (set_used_smts): Fix formatting.
1263 (compute_points_to_sets): Updated for new functions.
1264 (ipa_pta_execute): Ditto.
1266 2007-01-18 Kazu Hirata <kazu@codesourcery.com>
1267 Richard Sandiford <richard@codesourcery.com>
1269 * doc/tm.texi (TARGET_FUNCTION_VALUE): Expand documentation of
1271 * calls.c (expand_call): If the return value is a PARALLEL,
1272 extract its first member.
1273 * config/m68k/linux.h (FUNCTION_EXTRA_EPILOGUE): Remove.
1274 * config/m68k/m68k.c (m68k_output_function_epilogue): Don't
1275 use FUNCTION_EXTRA_EPILOGUE.
1276 (m68k_function_value): Return a PARALLEL if the return value
1277 is of a pointer type.
1278 * config/m68k/netbsd-elf.h (current_function_returns_pointer)
1279 (FUNCTION_EXTRA_EPILOGUE): Remove.
1280 * config/m68k/m68k.md (D0_REG): New constant.
1282 2007-01-18 Kazu Hirata <kazu@codesourcery.com>
1284 * config/m68k/m68k.c (m68k_output_function_epilogue): Don't
1285 output a NOP for empty epilogues.
1287 2007-01-18 Richard Sandiford <richard@codesourcery.com>
1289 * config/m68k/m68k.c (m68k_use_return_insn): Update comments
1290 before function. Extend register save check to include all
1291 registers, not just integer ones.
1293 2007-01-18 Kazu Hirata <kazu@codesourcery.com>
1295 * config/m68k/m68k-protos.h (use_return_insn): Rename to...
1296 (m68k_use_return_insn): ...this.
1297 * config/m68k/m68k.h (USE_RETURN_INSN): Delete.
1298 * config/m68k/m68k.c (use_return_insn): Rename to...
1299 (m68k_use_return_insn): ...this.
1300 * config/m68k/m68k.md (return): Use m68k_use_return_insn instead
1303 2007-01-18 Nathan Sidwell <nathan@codesourcery.com>
1305 * config/m68k/fpgnulib.c (__truncdfsf2): Implement round to
1306 nearest even, fix denormal rounding overflow.
1308 2007-01-18 Richard Sandiford <richard@codesourcery.com>
1310 * config/m68k/m68k.md (movsf_cf_hard): Use fsmove instead of
1311 f%$smove and f%$move.
1312 (movdf_cf_hard): Use fdmove for cases 0 and 3 and fmove for case 1.
1313 (extendsfdf2_cf): Use fdmove instead of f%&move.
1314 (truncdfsf2_cf): Use fsmove instead of f%$smove.
1315 (add<mode>3_cf, sub<mode>3_cf): Use <FP:prec> instead of <FP:round>.
1317 2007-01-18 Richard Sandiford <richard@codesourcery.com>
1319 * config/m68k/m68k.md (movdf_cf_hard): Use output_move_double for
1322 2007-01-18 Richard Sandiford <richard@codesourcery.com>
1324 * real.h (real_format): Add a canonical_nan_lsbs_set field.
1325 (coldfire_single_format): Declare.
1326 (coldfire_double_format): Likewise.
1327 * real.c (encode_ieee_single): Use canonical_nan_lsbs_set instead
1328 of qnan_msb_set to determine the lower bits of a canonical
1330 (encode_ieee_double): Likewise.
1331 (encode_ieee_quad): Likewise.
1332 (ieee_single_format): Initialize canonical_nan_lsbs_set.
1333 (mips_single_format): Likewise.
1334 (ieee_double_format): Likewise.
1335 (mips_double_format): Likewise.
1336 (ieee_extended_motorola_format): Likewise.
1337 (ieee_extended_intel_96_format): Likewise.
1338 (ieee_extended_intel_128_format): Likewise.
1339 (ieee_extended_intel_96_round_53_format): Likewise.
1340 (ibm_extended_format): Likewise.
1341 (mips_extended_format): Likewise.
1342 (ieee_quad_format): Likewise.
1343 (mips_quad_format): Likewise.
1344 (vax_f_format): Likewise.
1345 (vax_d_format): Likewise.
1346 (vax_g_format): Likewise.
1347 (i370_single_format): Likewise.
1348 (i370_double_format): Likewise.
1349 (decimal_single_format): Likewise.
1350 (decimal_double_format): Likewise.
1351 (decimal_quad_format): Likewise.
1352 (c4x_single_format): Likewise.
1353 (c4x_extended_format): Likewise.
1354 (real_internal_format): Likewise.
1355 (coldfire_single_format): New real_format.
1356 (coldfire_double_format): Likewise.
1357 * config/pdp11/pdp11.c (pdp11_f_format): Initialize
1358 canonical_nan_lsbs_set.
1359 (pdp11_d_format): Likewise.
1360 * config/m68k/m68k.c (override_options): Override REAL_FORMAT_MODE
1361 if TARGET_COLDFIRE_CPU.
1363 2007-01-18 Richard Sandiford <richard@codesourcery.com>
1365 * config/m68k/m68k-protos.h (m68k_output_pic_call): Delete.
1366 (output_call, m68k_legitimize_call_address): Declare.
1367 * config/m68k/m68k.h (EXTRA_CONSTRAINT): Remove unnecessary
1368 parenthesees. Add support for a 'W' constraint.
1369 (LEGITIMATE_PIC_OPERAND_P): Remove SYMBOL_REF_FLAG handling.
1370 (PRINT_OPERAND_PUNCT_VALID_P): Remove comment about 'o'.
1371 (m68k_symbolic_call, m68k_symbolic_jump): Declare.
1372 * config/m68k/m68k.c (m68k_symbolic_call, m68k_symbolic_jump): New
1374 (override_options): Initialize them. Do not set flag_no_function_cse
1375 for TARGET_ID_SHARED_LIBRARY.
1376 (m68k_output_pic_call): Delete.
1377 (m68k_legitimize_call_address): New function.
1378 (print_operand): Remove the %o prefix. Handle the %p prefix.
1379 (output_call): New function.
1380 (m68k_output_mi_thunk): Use m68k_symbolic_jump. Always load the
1381 target address from the GOT if symbolic jumps are not allowed.
1382 * config/m68k/m68k.md (call, general_operand): Do not set
1383 SYMBOL_REF_FLAG. Use m68k_legitimize_call_address instead.
1384 Merge separate flag_pic and !flag_pic define_insns into...
1385 (*call, *call_value): ...these new patterns. Match the address
1386 rather than the containing MEM and require it to be a call_operand.
1387 Use output_call to generate the asm template.
1388 * config/m68k/predicates.md (const_call_operand): New predicate.
1389 (call_operand): Likewise.
1391 2007-01-18 Nathan Sidwell <nathan@codesourcery.com>
1393 * config/m68k/m68k.h (REGISTER_MOVE_COST): Simplify definition.
1394 (STACK_GROWS_DOWNWARD): Define to 1.
1395 (FUNCTION_VALUE, LIBCALL_VALUE, FUNCTION_VALUE_REGNO_P): Equivocate
1396 comments, emphasizing that these values are only defaults.
1397 * config/m68k/linux.h (LINK_SPEC): Fix formatting in #undef.
1398 * config/m68k/m68k.c (const_method): Remove trailing whitespace.
1400 2007-01-18 Richard Sandiford <richard@codesourcery.com>
1402 * config/m68k/m68k.md (cmpsi): Remove outdated flag_pic handling.
1404 2007-01-18 Kazu Hirata <kazu@codesourcery.com>
1405 Richard Sandiford <richard@codesourcery.com>
1407 * config/m68k/m68k.h (DATA_REGNO_P, ADDRESS_REGNO_P, INT_REGNO_P)
1408 (FP_REGNO_P): New macros.
1409 (REGNO_OK_FOR_INDEX_P, REGNO_OK_FOR_BASE_P, REGNO_OK_FOR_DATA_P)
1410 (REGNO_OK_FOR_FP_P, REG_OK_FOR_INDEX_P, REG_OK_FOR_BASE_P): Use them.
1411 (EH_RETURN_STACKADJ_RTX): Use A0_REG.
1412 * config/m68k/m68k.c (m68k_regno_mode_ok): Use the new REGNO macros.
1414 2007-01-18 Nathan Sidwell <nathan@codesourcery.com>
1416 * config.gcc (m68k-*-aout*, m68k-*-coff*, m68020-*-elf*, m68k-*-elf*)
1417 (m68k-*-uclinuxoldabi, m68k-*-uclinux*, m68k-*-rtems*): Add t-floatlib
1419 * config/m68k/t-floatlib: New file, extracting common code from...
1420 * config/m68k/t-m68kbare, config/m68k/t-m68kelf,
1421 * config/m68k/t-uclinux: Here.
1422 * config/m68k/fpgnulib.c: Do not compile extendeed precision
1423 routines on ColdFire targets.
1425 2007-01-18 Nathan Sidwell <nathan@codesourcery.com>
1427 * config.gcc (m68k-*-aout*, m68k-*-coff*, m68020-*-elf*, m68k-*-elf*)
1428 (m68010-*-netbsdelf*, m68k*-*-netbsdelf*, m68k*-*-openbsd*)
1429 (m68k-*-uclinuxoldabi, m68k-*-uclinux*, m68k-*-linux*)
1430 (m68k-*-rtems*): Use tm_file rather than m68k/m68k.h and
1431 explicitly set MOTOROLA to 1.
1432 * config/m68k/m68k.h (MOTOROLA): Simplify definition accordingly.
1434 2007-01-18 Richard Sandiford <richard@codesourcery.com>
1435 Nathan Sidwell <nathan@codesourcery.com>
1437 * config/m68k/m68k.h (PCC_STATIC_STRUCT_RETURN): Delete.
1438 (ASM_OUTPUT_REG_PUSH, ASM_OUTPUT_REG_POP): Add MOTOROLA cases.
1439 * config/m68k/coff.h (REGISTER_PREFIX_MD): Delete.
1440 * config/m68k/m68020-elf.h (LIB_SPEC): Delete.
1441 * config/m68k/m68k-none.h (CC1_SPEC, CPP_SUBTARGET_SPEC): Delete.
1442 * config/m68k/m68kelf.h (IMMEDIATE_PREFIX, REGISTER_PREFIX_MD)
1443 (ASM_OUTPUT_REG_PUSH, ASM_OUTPUT_REG_POP): Delete.
1444 (NO_DOLLAR_IN_LABEL, PCC_STATIC_STRUCT_RETURN): Don't undefine.
1445 (BSS_ASM_OP, ASM_OUTPUT_SKIP, ASM_OUTPUT_ASCII): Delete.
1446 * config/m68k/m68kemb.h (PCC_STATIC_STRUCT_RETURN): Don't undefine.
1447 * config/m68k/linux.h (SIZE_TYPE, PTRDIFF_TYPE, WCHAR_TYPE)
1448 (WCHAR_TYPE_SIZE, TARGET_OBJFMT_CPP_BUILTINS): Delete.
1449 (TARGET_OS_CPP_BUILTINS): Don't define mc68000 and mc68020 here.
1450 (DBX_REGISTER_NUMBER): Delete.
1451 * config/m68k/netbsd-elf.h (IMMEDIATE_PREFIX): Delete.
1452 (PCC_STATIC_STRUCT_RETURN): Don't undefine.
1453 * config/m68k/openbsd.h (PCC_STATIC_STRUCT_RETURN): Define.
1455 2007-01-18 Richard Sandiford <richard@codesourcery.com>
1457 * config.gcc (m68k-*-uclinux*): Add flat.h to $tm_file.
1458 * config/flat.h: New file.
1459 * crtstuff.c (USE_PT_GNU_EH_FRAME): Don't define if
1461 * config/m68k/m68k.h (ASM_PREFERRED_EH_DATA_FORMAT): Do not use
1462 indirect references for -msep-data or -mid-shared-library.
1463 Do not use PC-relative code addresses either.
1464 * config/m68k/m68k.c (override_options): Restrict -fPIC error
1466 * config/m68k/uclinux.h (STARTFILE_SPEC): Define. Use Scrt1.o
1467 for shared libraries and crt1.o for executables. Use crti.o and
1469 (ENDFILE_SPEC): Use crtend.o and crtn.o.
1470 (LIB_SPEC): Suppress -Rlibc.gdb if -static-libc is given.
1471 Do not add -elf2flt or -shared-lib-id options here.
1472 (LINK_SPEC): Define. Pass -elf2flt if no -elf2flt option is given.
1473 Pass -shared-lib-id if -mid-shared-library, taking the library
1474 identifier from -mshared-library-id if given, otherwise
1476 (EH_FRAME_IN_DATA_SECTION): Do not undefine.
1477 (INIT_SECTION_ASM_OP, FINI_SECTION_ASM_OP): Likewise.
1478 (TARGET_OS_CPP_BUILTINS): Define __GXX_MERGED_TYPEINFO_NAMES=0
1479 and __GXX_TYPEINFO_EQUALITY_INLINE=0 if -mid-shared-library.
1480 (DRIVER_SELF_SPECS): Map unadorned PIC options to -msep-data.
1481 * config/m68k/t-uclinux (EXTRA_MULTILIB_PARTS): Add crtbegin.o
1483 * config/m68k/lb1sf68.asm (PICCALL): Use an lea and pc-relative
1484 jump sequence for ISA A and ISA A+.
1485 (PICJUMP): Likewise.
1487 2007-01-18 Richard Sandiford <richard@codesourcery.com>
1489 * config.gcc (m68k-*-uclinux*): Base the port on the common
1490 and m68k GNU/Linux files rather than on the generic ELF ones.
1491 * config/m68k/uclinux.h (TARGET_VERSION): Override.
1492 (TARGET_OS_CPP_BUILTINS): Use LINUX_TARGET_OS_CPP_BUILTINS.
1494 2007-01-18 Julian Brown <julian@codesourcery.com>
1495 Richard Sandiford <richard@codesourcery.com>
1497 * config/m68k/m68k.h (LONG_DOUBLE_TYPE_SIZE): Make 64-bit on ColdFire.
1498 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
1499 * config/m68k/netbsd-elf.h (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Undefine
1501 * config/m68k/uclinux-oldabi.h (LONG_DOUBLE_TYPE_SIZE): Redefine to
1503 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
1505 2007-01-18 Richard Sandiford <richard@codesourcery.com>
1507 * doc/install.texi: Document m68k-uclinuxoldabi.
1508 * config.gcc (m68k-*-uclinuxoldabi): New configuration.
1509 * config/m68k/uclinux-oldabi.h: New file, copied from
1510 config/m68k/uclinux.h.
1512 2007-01-18 Kaz Kojima <kkojima@rr.iij4u.or.jp>
1514 * config/m32r/m32r-protos.h (m32r_expand_epilogue): Declare it.
1515 * config/m32r/m32r.c (m32r_setup_incoming_varargs): Use gen_frame_mem.
1516 (m32r_compute_frame_size): Use unsigned for regno.
1517 (m32r_reload_lr): Use gen_frame_mem.
1519 (m32r_output_function_epilogue): Don't output the function epilogue
1521 (m32r_expand_epilogue): New.
1522 (direct_return): Return false if the function has the interrupt
1524 (m32r_hard_regno_rename_ok): Remove code for the textual epilogue.
1525 * config/m32r/m32r.md (epilogue): New expander.
1526 (return_lr, return_rte): New insns.
1527 (return): Make it expander.
1528 (return_normal): New expander.
1530 2007-01-18 Josh Conner <jconner@apple.com>
1533 * config/rs6000/rs6000.c (rs6000_emit_vector_compare): Add
1534 support for UNLE, UNLT, UNGE, and UNGT.
1536 2007-01-18 Jan Hubicka <jh@suse.cz>
1538 * tree-vrp.c (finalize_jump_threads): Do not call cleanup_cfg by hand.
1539 * tree-tailcall (add_virtual_phis): Likewise.
1540 (optimize_tail_call): Return TODOs.
1541 (execute_tail_calls): Return TODOs.
1542 * tree-ssa-ccp (execute_fold_all_builtins): Do cleanup_cfg via TODO.
1543 * tree-cfgcleanup.c (cleanup_tree_cfg_loop): Return if something
1545 * tree-ssa-forwprop.c (tree_ssa_forward_propagate_single_use_value):
1546 Cleanup cfg using TODO.
1547 * tree-flow.h (cleanup_tree_cfg_loop): Update prototype.
1548 * passes.c (execute_function_todo): When cleanup did something, remove
1550 * tree-cfg.c (pass_build_cfg): Add cleanup_cfg TODO.
1551 (make_edges): Don't cleanup_cfg.
1553 2007-01-18 Uros Bizjak <ubizjak@gmail.com>
1555 * reg-stack.c (subst_stack_regs_pat) [UNSPEC_SINCOS_COS,
1556 UNSPEC_XTRACT_FRACT]: Use generic code for instructions that
1557 operate on the top of stack.
1558 [UNSPEC_SINCOS_SIN, UNSPEC_XTRACT_EXP, UNSPEC_TAN]: Rewrite
1559 register handling of instructions that output to the second
1561 [UNSPEC_TAN_ONE, UNSPEC_TAN_TAN]: Remove.
1562 (move_for_stack_reg): Special-case check for dead destination
1563 stack slot for constant load of 1.0 inside UNSPEC_TAN.
1565 * config/i386/i386.md (UNSPEC_TAN): New constant.
1566 (UNSPEC_TAN_ONE, UNSPEC_TAN_TAN): Remove.
1567 (fptanxf4_i387, fptan_extend<mode>xf4_i387): New patterns
1568 to correctly model move of constant 1.0 to top stack slot.
1569 (*tandf3_1, *tansf3_1, *tanxf3_1): Remove insn patterns.
1570 (unnamed peephole2 pattern): Remove corresponding peephole2
1571 pattern that optimizes tan insn and loading of constant 1.0.
1572 (tanxf2): Use fptanxf4_i387.
1573 (tan<mode>2): Rename from tansf2 and tandf2 and macroize insn
1574 patterns using X87MODEF12 mode macro. Use fptan_extend<mode>xf4_i387
1575 and truncate result to requested mode. Use SSE_FLOAT_MODE_P to
1576 disable patterns for SSE math.
1577 (sincos<mode>3): Use truncxf<mode>2_i387_noop for truncation.
1578 (fyl2x_extend<mode>xf3_i387): Use X87MODEF12 for operand 1.
1580 2007-01-18 Dirk Mueller <dmueller@suse.de>
1581 Richard Guenther <rguenther@suse.de>
1584 * doc/invoke.texi (Warray-bounds): Document -Warray-bounds.
1585 * common.opt (Warray-bounds): Add new warning option.
1586 * c-opts.c (c_common_handle_option): Define -Warray-bounds
1588 * Makefile.in: make tree-vrp.o depend on toplev.h
1589 * tree-vrp.c (vrp_finalize): Call check_array_refs if -Warray-bounds
1591 (check_array_refs, check_array_bounds, check_array_ref): New.
1593 2007-01-18 Jan Hubicka <jh@suse.cz>
1595 * tree-ssa-ccp.c (ccp_finalize): Return if something changed.
1596 (execute_ssa_ccp): Return flags conditionally.
1597 * tree-ssa-propagate.c (substitue_and_fold): Return if something was
1599 * tree-ssa-propagate.h (substitute_and_fold): Update prototype.
1601 2007-01-18 Steven Bosscher <steven@gcc.gnu.org>
1603 * cfgcleanup.c (cleanup_cfg): Detect cfglayout mode and set
1604 the CLEANUP_CFGLAYOUT flag when in cfglayout mode.
1606 * Makefile.c (GTFILES): Add cfglayout.h.
1607 * gengtype.c (open_base_files): Likewise.
1608 * cfglayout.c (cfg_layout_function_footer,
1609 cfg_layout_function_header) Reindent to make gengtype happy.
1610 * cfglayout.h (cfg_layout_function_footer,
1611 cfg_layout_function_header): Add GTY(()) marker.
1613 * ifcvt.c (noce_try_sign_mask): Make sure INSN_B is non-null.
1615 2007-01-18 Ben Elliston <bje@au.ibm.com>
1617 * genautomata.c (write_automata): Include xstrerror output in the
1618 error message if writing the DFA description file fails.
1620 2007-01-17 H.J. Lu <hongjiu.lu@intel.com>
1622 * config/mips/mips-protos.h (mips_output_external): Make it
1624 * config/mips/iris.h (TARGET_ASM_EXTERNAL_LIBCALL): Removed.
1625 * config/mips/mips.c (irix_output_external_libcall): Likewise.
1626 (extern_list): Likewise.
1627 (extern_head): Likewise.
1628 (TARGET_ASM_FILE_END): Likewise.
1629 (mips_file_end): Likewise.
1630 (mips_output_external): Rewritten.
1632 2007-01-18 Ben Elliston <bje@au.ibm.com>
1634 * genpreds.c (write_insn_preds_c): Only write out the function
1635 body for regclass_for_constraint if we have register constraints.
1637 2007-01-17 Tom Tromey <tromey@redhat.com>
1639 * doc/sourcebuild.texi (libgcj Tests): Use sourceware.org.
1640 * doc/install.texi (Testing): Use sourceware.org.
1641 (Binaries): Likewise.
1642 (Specific): Likewise.
1643 * doc/contrib.texi (Contributors): Use sourceware.org.
1645 2007-01-17 Anatoly Sokolov <aesok@post.ru>
1647 * config/avr/avr.h (AVR_HAVE_LPMX): New macro.
1648 (AVR_ENHANCED): Rename to ...
1649 (AVR_HAVE_MUL): ... new.
1650 (avr_enhanced_p): Rename to ...
1651 (avr_have_mul_p): ... new.
1652 (TARGET_CPU_CPP_BUILTINS): Use 'avr_have_mul_p' instead of
1653 'avr_enhanced_p' for "__AVR_ENHANCED__". Define "__AVR_HAVE_MUL__".
1654 * config/avr/avr.c (avr_enhanced_p): Rename to ...
1655 (avr_have_mul_p): ... new.
1656 (base_arch_s): Rename 'enhanced' to 'have_mul'.
1657 (avr_override_options): Use 'avr_have_mul_p' and 'have_mul' instead of
1658 'avr_enhanced_p' and 'enhanced'.
1659 (ashlhi3_out, ashrhi3_out, lshrhi3_out, avr_rtx_costs): Use
1660 AVR_HAVE_MUL instead of AVR_ENHANCED.
1661 * avr.md (*tablejump_enh): Use AVR_HAVE_LPMX instead of AVR_ENHANCED.
1662 (mulqi3, *mulqi3_enh, *mulqi3_call, mulqihi3, umulqihi3, mulhi3,
1663 *mulhi3_enh, *mulhi3_call, mulsi3, *mulsi3_call): Use AVR_HAVE_MUL
1664 instead of AVR_ENHANCED.
1665 (*tablejump_enh): Use AVR_HAVE_LPMX instead of AVR_ENHANCED.
1666 * libgcc.S: Use __AVR_HAVE_MUL__ instead of __AVR_ENHANCED__.
1667 (__tablejump__): Use __AVR_HAVE_LPMX__ instead of __AVR_ENHANCED__.
1669 2007-01-17 Ian Lance Taylor <iant@google.com>
1671 * vec.h (VEC_reserve_exact): Define.
1672 (vec_gc_p_reserve_exact): Declare.
1673 (vec_gc_o_reserve_exact): Declare.
1674 (vec_heap_p_reserve_exact): Declare.
1675 (vec_heap_o_reserve_exact): Declare.
1676 (VEC_OP (T,A,reserve_exact)): New static inline function, three
1678 (VEC_OP (T,A,reserve)) [all versions]: Remove handling of
1680 (VEC_OP (T,A,alloc)) [all versions]: Call ...reserve_exact.
1681 (VEC_OP (T,A,copy)) [all versions]: Likewise.
1682 (VEC_OP (T,a,safe_grow)) [all versions]: Likewise.
1683 * vec.c (calculate_allocation): Add exact parameter. Change all
1685 (vec_gc_o_reserve_1): New static function, from vec_gc_o_reserve.
1686 (vec_gc_p_reserve, vec_gc_o_reserve): Call vec_gc_o_reserve_1.
1687 (vec_gc_p_reserve_exact, vec_gc_o_reserve_exact): New functions.
1688 (vec_heap_o_reserve_1): New static function, from vec_heap_o_reserve.
1689 (vec_heap_p_reserve, vec_heap_o_reserve): Call vec_heap_o_reserve_1.
1690 (vec_heap_p_reserve_exact): New function.
1691 (vec_heap_o_reserve_exact): New function.
1693 2007-01-17 Jan Hubicka <jh@suse.cz>
1695 * ipa-type-escape.c (look_for_casts): Revamp using handled_component_p.
1697 2007-01-17 Eric Christopher <echristo@apple.com>
1699 * config.gcc: Support core2 processor.
1701 2007-01-16 Jan Hubicka <jh@suse.cz>
1703 * tree-ssanames.c (release_dead_ssa_names): Instead of ggc_freeing
1704 the names, just unlink the chain so we don't crash on dangling pointers
1707 2007-01-16 Jan Hubicka <jh@suse.cz>
1709 * cgraph.h (cgraph_decide_inlining_incrementally): Kill.
1710 * tree-pass.h: Reorder to make IPA passes appear toegher.
1711 (pass_early_inline, pass_inline_parameters, pass_apply_inline): Declare.
1712 * cgraphunit.c (cgraph_finalize_function): Do not compute inling
1713 parameters, do not call early inliner.
1714 * ipa-inline.c: Update comments. Include tree-flow.h
1715 (cgraph_decide_inlining): Do not compute inlining parameters.
1716 (cgraph_decide_inlining_incrementally): Return TODOs; assume to
1717 be called with function context set up.
1718 (pass_ipa_inline): Remove unreachable functions before pass.
1719 (cgraph_early_inlining): Simplify assuming to be called from the
1721 (pass_early_inline): New pass.
1722 (cgraph_gate_ipa_early_inlining): New gate.
1723 (pass_ipa_early_inline): Turn into simple wrapper.
1724 (compute_inline_parameters): New function.
1725 (gate_inline_passes): New gate.
1726 (pass_inline_parameters): New pass.
1727 (apply_inline): Move here from tree-optimize.c
1728 (pass_apply_inline): New pass.
1729 * ipa.c (cgraph_remove_unreachable_nodes): Verify cgraph after
1731 * tree-inline.c (optimize_inline_calls): Return TODOs rather than
1733 (tree_function_versioning): Do not allocate dummy struct function.
1734 * tree-inline.h (optimize_inline_calls): Update prototype.
1735 * tree-optimize.c (execute_fixup_cfg): Export.
1736 (pass_fixup_cfg): Remove
1737 (tree_rest_of_compilation): Do not apply inlines.
1738 * tree-flow.h (execute_fixup_cfg): Declare.
1739 * Makefile.in (gt-passes.c): New.
1740 * passes.c: Include gt-passes.h
1741 (init_optimization_passes): New passes.
1742 (nnodes, order): New static vars.
1743 (do_per_function_toporder): New function.
1744 (execute_one_pass): Dump current pass here.
1745 (execute_ipa_pass_list): Don't dump current pass here.
1747 2007-01-16 Janis Johnson <janis187@us.ibm.com>
1749 * config/dfp-bit.c (dfp_compare_op): Return separate value for NaN.
1750 (DFP_NE, DFP_LE, DFP_GE): Return false for NaN.
1752 2007-01-16 David Edelsohn <edelsohn@gnu.org>
1754 * config/rs6000/darwin-ldouble.c: Build file for SOFT_FLOAT.
1755 (strong_alias): Define.
1756 (__gcc_qmul): Provide non-FMA for soft-float.
1762 (__gcc_qunord): New.
1772 * config/rs6000/rs6000.c (rs6000_init_libfuncs): Initialize
1773 soft-float functions.
1774 * config/rs6000/libgcc-ppc-glibc.ver: Version soft-float symbols.
1775 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Do not warn
1776 about long double soft float.
1778 2007-01-16 Dorit Nuzman <dorit@il.ibm.com>
1779 Tehila Meyzels <tehila@il.ibm.com>
1781 * tree-vectorizer.h (is_pattern_stmt_p): New.
1782 * tree-vect-analyze.c (vect_determine_vectorization_factor): Fix
1783 formatting (tabs instead of spaces). Cleanup and clarify setting
1784 of STMT_VINFO_VECTYPE. Call is_pattern_stmt_p.
1785 * tree-vect-transform.c (vect_get_vec_def_for_operand): Fix typo.
1786 (vectorizable_type_demotion): Check that types are integral.
1787 (vectorizable_type_promotion): Likewise.
1788 (vectorizable_store): Fix typo. Eliminate new-line at end of
1791 2007-01-16 Jan Hubicka <jh@suse.cz>
1793 * tree-ssanames.c (release_dead_ssa_names): Remove invalidated
1796 2007-01-15 Eric Christopher <echristo@apple.com>
1798 * ifcvt.c: Include vec.h, vecprim.h.
1799 (check_cond_move_block): New argument regs.
1800 Reorganize. Add registers used to regs.
1801 (cond_move_process_if_block): Use regs set above as
1804 2007-01-15 Eric Christopher <echristo@apple.com>
1806 * config/darwin.h: Update copyright.
1807 (TARGET_OPTION_TRANSLATE_TABLE): Add umbrella.
1808 (LINK_COMMAND_SPEC): Add -u.
1809 (LINK_SPEC): Fix umbrella for above.
1811 2007-01-15 Joseph S. Myers <joseph@codesourcery.com>
1813 * config/soft-fp/op-common.h, config/soft-fp/op-4.h: Update from
1816 2007-01-15 Tom Tromey <tromey@redhat.com>
1818 * doc/sourcebuild.texi (libgcj Tests): Don't mention jacks.
1819 * doc/install.texi (Testing): Don't mention jacks.
1820 (Configuration): Document --enable-java-maintainer-mode. Move
1821 --with-java-home to libgcj-specific section. Document
1823 (Prerequisites): Mention --enable-java-maintainer-mode, ecj1.
1825 2007-01-15 Jan Hubicka <jh@suse.cz>
1827 * tree-ssa-dce.c (DCE_TODOs): New.
1828 (propagate_necessity): Return if something changed.
1829 (eliminate_unnecessary_stmts): Likewise.
1830 (perform_tree_ssa_dce): Return TODO flags when needed.
1831 (pass_dce, pass_dce_loop, pass_cd_dce): Remove TODO flags.
1833 2007-01-15 Uros Bizjak <ubizjak@gmail.com>
1835 * config/i386/i386.md (fyl2xxf3_i387): Rename from fyl2x_xf3.
1836 (fyl2x_extend<mode>xf3_i387): New insn pattern.
1837 (log<mode>2): Rename from logsf2 and logdf2 and macroize insn
1838 patterns using X87MODEF12 mode macro. Extend operand 1
1839 to XFmode. Use SSE_FLOAT_MODE_P to disable patterns for SSE math.
1840 (log10<mode>2): Ditto.
1841 (log2<mode>2): Ditto.
1842 (log1p<mode>2): Ditto.
1843 (logb<mode>2): Ditto.
1844 (fyl2xp1xf3_i387): Rename from fyl2xp1_xf3.
1845 (fyl2xp1_extend<mode>xf3_i387): New insn pattern.
1846 (*fxtractxf3_i387): Rename from *fxtractxf3.
1847 (fxtract_extend<mode>xf3_i387): New insn pattern.
1848 (ilogbsi2): Use match_dup 3, not match_operand:XF 3.
1849 * config/i386/i386.c (ix86_emit_i387_log1p): Use gen_fyl2xp1xf3_i387()
1850 and gen_fyl2xxf3_i387().
1852 2007-01-14 Zdenek Dvorak <dvorakz@suse.cz>
1854 * loop-unswitch.c (unswitch_loop): Do not call fix_loop_placement.
1855 * cfgloopmanip.c (fix_loop_placement): Made static. Use
1856 get_loop_exit_edges. Changed return type to bool.
1857 * cfgloop.h (fix_loop_placement): Declaration removed.
1859 2007-01-14 Dorit Nuzman <dorit@il.ibm.com>
1861 * param.h (MIN_VECT_LOOP_BOUND): New.
1862 * params.def (MIN_VECT_LOOP_BOUND): New.
1863 * tree-vectorizer.c (slpeel_tree_peel_loop_to_edge): Takes another
1864 argument - minimum threshold for number of iterations.
1865 * tree-vectorizer.h (slpeel_tree_peel_loop_to_edge): Add another
1866 argument to declaration.
1867 * tree-vect-analyze.c (vect_analyze_operations): Check value of
1868 MIN_VECT_LOOP_BOUND.
1869 * tree-vect-transform.c (vect_do_peeling_for_loop_bound): Call
1870 slpeel_tree_peel_loop_to_edge with additional argument.
1871 (vect_do_peeling_for_alignment): Likewise.
1872 * doc/invoke.texi (min-vect-loop-bound): Document new param option.
1874 2007-01-14 Uros Bizjak <ubizjak@gmail.com>
1877 * config/i386/i386.c (print_operand) ['z']: Output 'b' for
1880 2007-01-14 Jan Hubicka <jh@suse.cz>
1882 * tree-dfa.c (remove_referenced_var): New function.
1883 * tree-ssa-live.c (remove_unused_locals): Walk referenced vars and
1884 prune referenced vars list too.
1885 * tree-flow.h (remove_referenced_var): Declare.
1887 2007-01-14 Jan Hubicka <jh@suse.cz>
1889 * tree-eh.c (add_stmt_to_eh_region_fn): Do not add call_exprs
1891 (remove_stmt_from_eh_region_fn): Do not remove call_exprs.
1892 (verify_eh_throw_stmt_node, verify_eh_throw_table_statements): Kill.
1893 * except.h (verify_eh_throw_table_statements): Kill prototype.
1894 * cfgexpand.c (expand_gimple_basic_block): Propagate Eh regions
1895 into call exrepssions.
1896 * tree-optimize.c (execute_free_cfg_annotatiosn): Do not call
1898 * tree-cfg.c: Include pointer-set.h.
1899 (verify_node_sharing): Work on pointer set.
1900 (verify_eh_throw_stmt_node): New.
1901 (verify_stmts): Use pointers sets, verify throw_stmt.
1903 2007-01-13 Zdenek Dvorak <dvorakz@suse.cz>
1905 * ipa-reference.c (analyze_function): Consider also addresses taken
1908 2007-01-12 Roger Sayle <roger@eyesopen.com>
1910 * c-typeck.c (null_pointer_constant_p): Replace use of
1911 TREE_CONSTANT_OVERFLOW with TREE_OVERFLOW.
1912 (build_c_cast): Likewise.
1914 2007-01-12 Roger Sayle <roger@eyesopen.com>
1916 * tree.h (force_fit_type_double): Remove unused final argument.
1917 * c-common.c (constant_expression_warning): Replace use of
1918 TREE_CONSTANT_OVERFLOW with TREE_OVERFLOW.
1919 (convert_and_check): Likewise.
1920 (shorten_compare): Update call to force_fit_type_double.
1921 (c_common_truthvalue_conversion) <INTEGER_CST>: Use integer_zerop.
1922 * convert.c (convert_to_pointer): Update call to
1923 force_fit_type_double.
1924 * fold-const.c (force_fit_type_double): Remove overflowed_const
1926 (int_const_binop, fold_convert_const_int_from_int,
1927 fold_convert_const_int_from_real, fold_div_compare,
1928 fold_sign_changed_comparison, fold_unary, fold_negate_const,
1929 fold_abs_const, fold_not_const): Remove the final argument from
1930 calls to force_fit_type_double.
1932 2007-01-12 Andrew Pinski <andrew_pinski@playstation.sony.com>
1934 * configure.ac: Set insn to "nop" for spu-*-* also.
1935 * configure: Regenerate.
1937 2007-01-12 Olga Golovanevsky <olga@il.ibm.com>
1939 * builtins.def : Add BUILT_IN_FREE.
1941 2007-01-12 Jan Hubicka <jh@suse.cz>
1943 PR tree-optimization/30443
1944 * tree-inline.c (tree_function_versioning): Do not optimize when
1945 cloning for inlining.
1947 2007-01-12 Zdenek Dvorak <dvorakz@suse.cz>
1949 * doc/loop.texi: Document recording of loop exits.
1950 * cfgloopmanip.c (loopify, duplicate_loop): Use alloc_loop.
1951 (update_single_exits_after_duplication,
1952 update_single_exit_for_duplicated_loop,
1953 update_single_exit_for_duplicated_loops): Removed.
1954 (duplicate_loop_to_header_edge): Do not call
1955 update_single_exits_after_duplication and
1956 update_single_exit_for_duplicated_loops.
1957 (loop_version): Do not update single_exit information.
1958 (fix_loop_structure): Use record_loop_exits instead of
1959 mark_single_exit_loops.
1960 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update
1961 the lists of loop exits.
1962 * cfghooks.c (redirect_edge_and_branch, redirect_edge_and_branch_force,
1963 split_edge, merge_blocks): Update the lists of loop exits.
1964 * modulo-sched.c (sms_schedule): Pass LOOPS_HAVE_RECORDED_EXITS to
1965 loop_optimizer_init.
1966 * loop-init.c (loop_optimizer_init): Call record_loop_exits instead
1967 of mark_single_exit_loops.
1968 (loop_optimizer_finalize): Call release_recorded_exits.
1969 * tree-ssa-loop.c (tree_loop_optimizer_init): Pass
1970 LOOPS_HAVE_RECORDED_EXITS to loop_optimizer_init.
1971 * tree-vectorizer.c (slpeel_tree_duplicate_loop_to_edge_cfg): Do not
1972 update single exit information.
1973 * lambda-code.c (perfect_nestify): Ditto.
1974 * cfgloop.c (flow_loop_free): Destroy the list of exits of the loop.
1975 (mark_single_exit_loops): Removed.
1976 (alloc_loop, loop_exit_hash, loop_exit_eq, loop_exit_free,
1977 get_exit_descriptions, rescan_loop_exit, record_loop_exits,
1978 dump_recorded_exit, dump_recorded_exits, release_recorded_exits): New
1980 (get_loop_exit_edges, single_exit): Use recorded exit lists.
1981 (add_bb_to_loop, remove_bb_from_loops): Update the lists of loop exits.
1982 (verify_loop_structure): Verify consistency of the exit lists.
1983 (flow_loops_find): Use alloc_loop. Initialize exits hash.
1984 (set_single_exit): Removed.
1985 * cfgloop.h (struct loop_exit): New function.
1986 (struct loop): single_exit_ field replaced by exits field.
1987 (LOOPS_HAVE_MARKED_SINGLE_EXITS): Replaced by LOOPS_HAVE_RECORDED_EXITS.
1988 (struct loops): Added exits hash.
1989 (mark_single_exit_loops, set_single_exit): Declaration removed.
1990 (release_recorded_exits, record_loop_exits, rescan_loop_exit): Declare.
1992 2007-01-12 Richard Sandiford <richard@codesourcery.com>
1994 * doc/invoke.texi: Avoid use of @headitem.
1996 2007-01-12 Richard Sandiford <richard@codesourcery.com>
1998 * cse.c (cse_insn): Move HAVE_CC0 code after declarations.
2000 2007-01-12 Richard Sandiford <richard@codesourcery.com>
2002 * doc/install.texi: Fix m68k-*-* anchor and add m68k-*-* to the
2005 2007-01-12 Nathan Sidwell <nathan@codesourcery.com>
2006 Richard Sandiford <richard@codesourcery.com>
2008 * doc/invoke.texi: Document -mno-bitfield, -mno-rtd and -mno-short.
2009 * config/m68k/m68k.opt: Resort options.
2010 (mbitfield, mrtd, mshort): Remove RejectNegative properties.
2012 2007-01-12 Nathan Sidwell <nathan@codesourcery.com>
2013 Richard Sandiford <richard@codesourcery.com>
2015 * doc/invoke.texi: Document the macros that are defined by
2016 m68k's -mtune and -mhard-float options.
2017 * config/m68k/m68k-protos.h (m68k_cpp_cpu_ident) Declare.
2018 (m68k_cpp_cpu_family): Likewise.
2019 * config/m68k/m68k.h (TARGET_CPU_CPP_BUILTINS): Add a full set
2020 of __ucfv*__ macros. Define __mcffpu__ if generating code for
2021 ColdFire FPUs. Define __mcf_cpu_* and __mcf_family_* macros.
2022 * config/m68k/m68k.c (m68k_cpp_cpu_ident): New function.
2023 (m68k_cpp_cpu_family): Likewise.
2025 2007-01-12 Richard Sandiford <richard@codesourcery.com>
2027 * config/m68k/m68k.h (TARGET_CPU_CPP_BUILTINS): Treat all mc68*
2028 macros besides mc68000 as tuning macros. Use a switch statement
2029 to set them and mcpu32.
2031 2007-01-12 Julian Brown <julian@codesourcery.com>
2033 * config/m68k/m68k.h: Use TARGET_68040 instead of TARGET_68040_ONLY.
2034 (TARGET_68040_ONLY): Rename to...
2035 (TARGET_68040): ...this.
2036 * config/m68k/m68k.c: Use TARGET_68040 instead of TARGET_68040_ONLY.
2037 * config/m68k/m68k.md: Likewise.
2039 2007-01-12 Julian Brown <julian@codesourcery.com>
2040 Nathan Sidwell <nathan@codesourcery.com>
2041 Richard Sandiford <richard@codesourcery.com>
2043 * config.gcc (m680[012]0-*-*, m68k*-*-*): Set m68k_cpu_ident to
2044 the -mcpu= argument associated with the --with-cpu setting.
2045 Define M68K_DEFAULT_TUNE to the default -mtune= option,
2046 if different from the one implied by the -mcpu setting.
2047 Accept --with-cpu=FOO if FOO is listed in m68k-devices.def,
2048 using mcpu=FOO as the default CPU option. Set target_cpu_default2.
2049 * doc/invoke.texi: Mention ColdFire in the introduction to the
2050 m68k options. Document the new -march, -mcpu, -mtune, -mdiv,
2051 -mno-div and -mhard-float options. Make -m68881 a synonym for
2052 -mhard-float. Document the previously-undocumented -m5206e,
2053 -m528x, -m5307 and -m5407 options. Tweak the existing option
2054 documentation for consistency.
2055 * doc/install.texi: Mention new --with-cpu arguments.
2056 * config/m68k/m68k.h (OPTION_DEFAULT_SPECS): Only use the
2057 default CPU if neither -mcpu nor -march are specified.
2058 (ASM_CPU_SPEC): Pass down -mcpu and -march options.
2059 (TARGET_CPU_CPP_BUILTINS): Set __mcfisa*__ macros from
2060 TARGET_ISA*. Set the legacy __mcf*__ cpu macros in the same way,
2061 using m68k_tune to decide between families that implement the
2062 same ISA. Use m68k_tune to set __mcfv4e__.
2063 (FL_BITFIELD, FL_68881, FL_COLDFIRE, FL_CF_HWDIV, FL_CF_MAC)
2064 (FL_CF_EMAC, FL_CF_EMAC_B, FL_CF_USP, FL_CF_FPU, FL_ISA_68000)
2065 (FL_ISA_68010, FL_ISA_68020, FL_ISA_68040, FL_ISA_A, FL_ISA_B)
2066 (FL_ISA_C, FL_ISA_MMU): New macros.
2067 (MASK_COLDFIRE): Delete.
2068 (TARGET_68010, TARGET_68020, TARGET_68040_ONLY, TARGET_COLDFIRE)
2069 (TARGET_ISAB): Redefine in terms of m68k_cpu_flags.
2070 (TARGET_68881, TARGET_COLDFIRE_FPU): Redefine in terms of m68k_fpu.
2071 (TARGET_HARD_FLOAT): Do not define here.
2072 (TARGET_ISAAPLUS, TARGET_ISAC): New macros.
2073 (TUNE_68000): New macro.
2074 (TUNE_68000_10): Redefine in terms of TUNE_68000 and TUNE_68010.
2075 (TUNE_68010, TUNE_68030, TUNE_68040, TUNE_68060, TUNE_CPU32)
2076 (TUNE_CFV2): Redefine in terms of m68k_tune.
2077 (uarch_type, target_device, fpu_type): New enums.
2078 (m68k_cpu, m68k_tune, m68k_fpu, m68k_cpu_flags): Declare.
2079 * config/m68k/m68k.c (TARGET_DEFAULT): Remove MASK_68881.
2080 (FL_FOR_isa_00, FL_FOR_isa_10, FL_FOR_isa_20, FL_FOR_isa_40)
2081 (FL_FOR_isa_cpu32, FL_FOR_isa_a, FL_FOR_isa_aplus, FL_FOR_isa_b)
2082 (FL_FOR_isa_c): New macros.
2083 (m68k_isa): New enum.
2084 (m68k_target_selection): New structure.
2085 (all_devices, all_isas, all_microarchs): New tables.
2086 (m68k_cpu_entry, m68k_arch_entry, m68k_tune_entry, m68k_cpu)
2087 (m68k_tune, m68k_fpu, m68k_cpu_flags): New variables.
2088 (MASK_ALL_CPU_BITS): Delete.
2089 (m68k_find_selection): New function.
2090 (m68k_handle_option): Handle -mcpu=, -march= and -mtune=.
2091 Map the legacy target options to a combination of the new ones.
2092 (override_options): Set m68k_cpu, m68k_tune, m68k_fpu and
2093 m68k_cpu_flags. Handle M68K_DEFAULT_TUNE. Use m68k_cpu_flags
2094 to derive default MASK_BITFIELD, MASK_CF_HWDIV and MASK_HARD_FLOAT
2096 * config/m68k/m68k.opt (m5200, m5206e, m528x, m5307, m5407, mcfv4e)
2097 (m68010, m68020, m68020-40, m68020-60, m68030, m68040): Remove Mask
2099 (m68881, msoft-float): Change mask from 68881 to HARD_FLOAT.
2100 (march=, mcpu=, mdiv, mhard-float, mtune=): New options.
2101 * config/m68k/m68k-devices.def: New file.
2103 2007-01-12 Richard Sandiford <richard@codesourcery.com>
2104 Nathan Sidwell <nathan@codesourcery.com>
2106 * config/m68k/m68k.h (ASM_CPU_SPEC, ASM_SPEC, EXTRA_SPECS)
2107 (SUBTARGET_EXTRA_SPECS): New macros.
2108 * config/m68k/linux.h (ASM_SPEC): Remove CPU flags;
2109 use %(asm_cpu_spec) instead.
2110 * config/m68k/m68k-none.h (ASM_SPEC): Likewise.
2111 * config/m68k/openbsd.h (ASM_SPEC): Likewise.
2112 * config/m68k/netbsd-elf.h (ASM_SPEC): Likewise.
2113 (EXTRA_SPECS): Rename to...
2114 (SUBTARGET_EXTRA_SPECS): ...this.
2116 2007-01-12 Nathan Sidwell <nathan@codesourcery.com>
2117 Richard Sandiford <richard@codesourcery.com>
2118 Julian Brown <julian@codesourcery.com>
2120 * config.gcc (m68k-*-aout*, m68k-*-coff*, m68020-*-elf*, m68k-*-elf*)
2121 (m68k-*-uclinux*, m68k-*-linux*, m68k-*-rtems*): Set default_m68k_cpu
2122 to the configuration's default CPU.
2123 (m68010-*-netbsdelf*, m68k*-*-netbsdelf*, m68k*-*-openbsd*): Likewise.
2124 Remove default masks.
2125 (m680[012]0-*-*): Set the default with_cpu to the first part of
2127 (m68k*-*-*): Set the default with_cpu to m$default_m68k_cpu.
2128 (m68k*-*-linux): Extend the --with-cpu handling to...
2129 (m680[012]0-*-*, m68k*-*-*): ...these configurations. Allow m68000
2130 and m68010. Don't set target_cpu_default2.
2131 * doc/install.texi: Document --with-cpu for m68k.
2132 * config/m68k/m68k.h (OPTION_DEFAULT_SPECS): Define.
2133 * config/m68k/m68k-none.h (TARGET_CPU_DEFAULT, M68K_CPU_m68k)
2134 (M68K_CPU_m68000, M68K_CPU_m68010, M68K_CPU_m68020, M68K_CPU_m68030)
2135 (M68K_CPU_m68040, M68K_CPU_m68302, M68K_CPU_m68332, TARGET_DEFAULT)
2136 (ASM_CPU_DEFAULT_SPEC, CC1_CPU_DEFAULT_SPEC): Delete.
2137 (ASM_SPEC): Remove use of %(asm_cpu_default).
2138 (EXTRA_SPECS, SUBTARGET_EXTRA_SPECS, MULTILIB_DEFAULTS): Delete.
2139 * config/m68k/linux.h (TARGET_DEFAULT): Delete.
2140 (CPP_SPEC): Merge definitions. Do not handle __HAVE_68881__ here.
2141 * config/m68k/netbsd-elf.h (TARGET_OS_CPP_BUILTINS): Define
2142 __HAVE_FPU__ if TARGET_HARD_FLOAT.
2143 (TARGET_DEFAULT): Delete.
2144 (EXTRA_SPECS): Delete cpp_cpu_default_spec, cpp_cpu_spec,
2145 cpp_fpu_spec, asm_default_spec and netbsd_cpp_spec.
2146 (CPP_CPU_SPEC): Delete.
2147 (TARGET_VERSION): Merge definitions, using TARGET_68010 to pick
2148 the appropriate string.
2149 (CPP_CPU_DEFAULT_SPEC, ASM_DEFAULT_SPEC, CPP_FPU_SPEC): Delete.
2150 (CPP_SPEC): Define to NETBSD_CPP_SPEC.
2151 (ASM_SPEC): Don't use %(asm_default_spec).
2152 * config/m68k/m68k.c (TARGET_DEFAULT_TARGET_FLAGS): Remove
2153 TARGET_DEFAULT and add MASK_68881.
2154 * config/m68k/m68k.md: Remove mention of TARGET_DEFAULT from comments.
2156 2007-01-12 Richard Sandiford <richard@codesourcery.com>
2158 * config.gcc (m68010-*-netbsdelf*): Add MASK_68010.
2159 (m68k*-*-netbsdelf*, m68k*-*-openbsd*, m68k*-linux*): Add
2160 MASK_68010 alongside MASK_68020.
2161 * doc/invoke.texi: Document -m68010.
2162 * config/m68k/m68k.opt (m68010): New.
2163 * config/m68k/m68k.h (TARGET_CPU_CPP_BUILTINS): Define mc68010
2165 (TUNE_68010): New macro.
2166 * config/m68k/m68k-none.h (M68K_CPU_m68k, M68K_CPU_m68010)
2167 (M68K_CPU_m68020, M68K_CPU_m68030, M68K_CPU_m68040)
2168 (M68K_CPU_m68332): Add MASK_68010.
2169 * config/m68k/linux.h (TARGET_DEFAULT): Add MASK_68010 to
2170 fallback definition.
2171 * config/m68k/netbsd-elf.h (CPP_CPU_SPEC): Remove now-redundant
2173 * config/m68k/m68k.c (MASK_ALL_CPU_BITS): Add MASK_68010.
2174 (m68k_handle_option): Handle OPT_m68010. Add MASK_68010
2175 to all entries that use MASK_68020.
2176 (output_move_simode_const, output_move_himode, output_move_qimode)
2177 (output_move_stricthi, output_move_strictqi): Use TARGET_68010
2178 instead of TARGET_68020 to select clr behavior. Remove comment
2179 about there being no TARGET_68010.
2180 * config/m68k/m68k.md: Likewise throughout.
2182 2007-01-12 Julian Brown <julian@codesourcery.com>
2184 * config/m68k/m68k.h (TARGET_ISAB): New macro.
2185 * config/m68k/m68k.c: Use TARGET_ISAB rather than TARGET_CFV4.
2186 * config/m68k/m68k.md: Likewise.
2188 2007-01-12 Julian Brown <julian@codesourcery.com>
2190 * config/m68k/m68k.h (LEGITIMATE_INDEX_P, LEGITIMIZE_ADDRESS): Use
2191 TARGET_COLDFIRE_FPU instead of TARGET_CFV4E.
2193 2007-01-12 Julian Brown <julian@codesourcery.com>
2195 * config/m68k/m68k.h (TUNE_68040_60): New macro.
2196 * config/m68k/m68k.c (standard_68881_constant_p): Use it.
2197 * config/m68k/m68k.md: Likewise.
2199 2007-01-12 Julian Brown <julian@codesourcery.com>
2200 Richard Sandiford <richard@codesourcery.com>
2202 * config/m68k/m68k.h (TARGET_CPU_CPP_BUILTINS): Use TUNE_68030
2203 instead of TARGET_68030, TUNE_68040 instead of TARGET_68040,
2204 TUNE_68060 instead of TARGET_68060 and TUNE_CPU32 instead of
2206 (TARGET_CPU32): Rename to...
2207 (TUNE_CPU32): ...this.
2208 (TUNE_68000_10, TUNE_68030, TUNE_68040, TUNE_68060)
2209 (TUNE_CFV2): New macros.
2210 * config/m68k/netbsd-elf.h (LONG_DOUBLE_TYPE_SIZE): Simplify;
2211 remove conditions that are implied by TARGET_68020.
2212 * config/m68k/m68k.c (m68k_output_function_prologue): Use TUNE_68040
2213 instead of TARGET_68040 and TUNE_CPU32 instead of TARGET_CPU32.
2214 (m68k_output_function_epilogue): Likewise.
2215 (m68k_rtx_costs): Likewise. Use TUNE_68060 instead of TARGET_68060
2216 and TUNE_CFV2 instead of TARGET_5200. Use TUNE_68000_10 instead of
2217 "!TARGET_68020 && !TARGET_COLDFIRE" to choose between 68000 and
2218 non-68000 timings. Refactor multiplication and division costs.
2219 (output_addsi3): Use TUNE_68040 instead of TARGET_68040 and
2220 TUNE_CPU32 instead of TARGET_CPU32.
2221 (standard_68881_constant_p): Use TUNE_68040 instead of TARGET_68040
2222 and TUNE_68060 instead of TARGET_68060.
2223 * config/m68k/m68k.md: Use TUNE_68040 instead of TARGET_68040,
2224 TUNE_68060 instead of TARGET_68060, and TUNE_CPU32 instead of
2226 (movsi_const0): Use TUNE_68000_10 rather than "!TARGET_68020
2227 && !TARGET_COLDFIRE" to choose between moveq and clr.
2228 Likewise in the unnamed movsf pattern.
2229 (ashlsi_17_24, lshrsi_17_24): Guard with TUNE_68000_10 rather than
2230 "!TARGET_68020 && !TARGET_COLDFIRE". Likewise the unnamed
2233 2007-01-12 Richard Sandiford <richard@codesourcery.com>
2235 * config/m68k/m68k.h (TARGET_CPU_CPP_BUILTINS): Increase amount
2236 of tabbing before backslashes.
2238 2007-01-11 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
2240 * pa-linux.h (ASM_OUTPUT_INTERNAL_LABEL): Undefine.
2241 * pa.h (ASM_OUTPUT_LABEL): Output colon when using GAS.
2242 (ASM_OUTPUT_INTERNAL_LABEL): Define.
2244 2007-01-11 Zdenek Dvorak <dvorakz@suse.cz>
2246 * tree-ssa-loop-ivopts.c (extract_cond_operands): Split from
2247 find_interesting_uses_cond.
2248 (find_interesting_uses_cond): Use extract_cond_operands.
2249 (rewrite_use_compare): Use extract_cond_operands and
2250 force_gimple_operand_bsi. Do not call update_stmt.
2251 (determine_use_iv_cost_condition): Use extract_cond_operands.
2252 Return cheaper of using original bound and changing the exit bound.
2254 2007-01-11 Zdenek Dvorak <dvorakz@suse.cz>
2256 PR tree-optimization/29516
2257 * tree-ssa-address.c (tree_mem_ref_addr, add_to_parts,
2258 most_expensive_mult_to_index, addr_to_parts,
2259 create_mem_ref, maybe_fold_tmr): Make the type of
2260 fields of TARGET_MEM_REF sizetype.
2261 (move_fixed_address_to_symbol, move_pointer_to_base):
2263 * tree.def (TARGET_MEM_REF): Add comment on types of
2266 2007-01-11 Joseph Myers <joseph@codesourcery.com>
2268 * c-common.c (vector_types_convertible_p): Treat opaque types as
2269 always convertible if they have the same size, but not otherwise.
2271 2007-01-11 Steven Bosscher <steven@gcc.gnu.org>
2273 * ifcvt.c (struct noce_if_info): Add comments to the fields.
2274 Remove the b_unconditional field.
2275 (noce_try_sign_mask): Do not look at b_unconditional.
2276 (noce_process_if_block): Do not use merge_if_blocks. Update
2277 the CFG here. Do not set b_unconditional.
2278 (cond_move_process_if_block): Likewise.
2279 (find_cond_trap): Likewise.
2280 (check_cond_move_block): Require simple jump insns at the end
2283 2007-01-11 Jan Hubicka <jh@suse.cz>
2285 PR tree-optimization/1046
2286 * tree-tailcall.c (suitable_for_tail_call_opt_p): Use TREE_ADDRESSABLE
2287 when alias info is not ready.
2288 (pass_tail_recursion): Do not require aliasing.
2289 * tree-ssa-copyrename.c (pass_rename_ssa_cop): Likewise.
2290 * tree-ssa-ccp.c (pass_ccp, pass_fold_builtins): Likewise.
2291 * tree-ssa-copy.c (pass_copy_prop): Likewise.
2292 * tree-ssa-forwprop.c (pass_forwprop): Likewise.
2293 * tree-ssa-dce.c (pass_dce, pass_dce_loop, pass_cd_dce): Likewise.
2294 * passes.c (init_optimization_passes): Execute rename_ssa_copies,
2295 ccp, forwprop, copy_prop, merge_phi, copy_prop, dce and tail recursion
2297 * tree-ssa-operands.c (add_virtual_operand, get_indirect_ref_operand):
2298 When aliasing is not build, mark statement as volatile.
2300 2007-01-11 Tom Tromey <tromey@redhat.com>
2302 PR preprocessor/15185, PR preprocessor/20989:
2303 * doc/cppopts.texi <-MT>: Update description of algorithm for
2304 computing default target.
2305 <-M, -MD>: Reword "basename" text.
2307 2007-01-11 Roger Sayle <roger@eyesopen.com>
2309 * builtins.c (expand_builtin_pow, expand_builtin_powi,
2310 fold_builtin_cabs, fold_builtin_sqrt, fold_builtin_trunc,
2311 fold_builtin_floor, fold_builtin_ceil, fold_builtin_round,
2312 fold_builtin_int_int_roundingfn, fold_builtin_bitop,
2313 fold_builtin_bswap, real_constp, fold_builtin_pow,
2314 fold_builtin_powi, fold_builtin_signbit, fold_builtin_copysign,
2315 do_mpfr_arg1, do_mpfr_arg2, do_mpfr_arg3, do_mpfr_sincos): Replace
2316 uses of the macro TREE_CONSTANT_OVERFLOW with TREE_OVERFLOW.
2317 * convert.c (convert_to_pointer): Likewise.
2318 * expr.c (highest_pow2_factor, expand_expr_real_1): Likewise.
2319 * fold-const.c (force_fit_type, fold_negate_expr, int_const_binop,
2320 const_binop, fold_convert_const_int_from_int,
2321 fold_convert_const_int_from_real,
2322 fold_convert_const_real_from_real, sign_bit_p,
2323 optimize_minmax_comparison, extract_muldiv_1, fold_div_compare,
2324 fold_sign_changed_comparison, fold_unary, fold_comparison,
2325 fold_binary, multiple_of_p, tree_Expr_non_zero_p,
2326 fold_negate_const, fold_abs_const, fold_not_const): Likewise.
2327 * print-tree.c (print_node_brief, print_node): Likewise.
2328 * stor-layout.c (place_field, layout_type): Likewise.
2329 * tree-chrec.c (keep_cast): Likewise.
2330 * tree.c (build_vector, build_real, build_real_from_int_cst,
2331 build_complex): Likewise.
2333 2007-01-11 Roger Sayle <roger@eyesopen.com>
2335 * tree.h (TREE_CONSTANT_OVERFLOW): Obsolete. For the time being,
2336 treat TREE_CONSTANT_OVERFLOW as a synonym of TREE_OVERFLOW.
2338 2007-01-11 Paolo Bonzini <bonzini@gnu.org>
2340 * configure.ac (strict1_warn): Rename to strict_warn.
2341 (WERROR, --enable-werror, symlink hacks, stage1_cflags,
2342 cc_set_by_configure, quoted_cc_set_by_configure,
2343 stage_prefix_set_by_configure, quoted_stage_prefix_set_by_configure,
2344 all_boot_languages, all_stagestuff): Remove.
2345 (target_list): Remove bootstrap targets.
2346 * Makefile.in (quickstrap): Unconditionally make a synonym of all.
2347 (BOOT_LANGUAGES, STAGE1_CFLAGS, STAGE1_CHECKING,
2348 REMAKEFLAGS, FLAGS_TO_PASS, PREPEND_DOTDOT_TO_RELATIVE_PATHS,
2349 SUBDIR_FLAGS_TO_PASS, WERROR_FLAGS, STRICT2_WARN, LANG_STAGESTUFF,
2350 VOL_FILES, POSTSTAGE1_FLAGS_TO_PASS, STAGE2_FLAGS_TO_PASS,
2351 STAGEPROFILE_FLAGS_TO_PASS, STAGEFEEDBACK_FLAGS_TO_PASS, stage1_build,
2352 stage1_copy, stage2_build, stage2_copy, stageprofile_build,
2353 stageprofile_copy, stage3_build, stage3_copy, stagefeedback_build,
2354 stagefeedback_copy, stage4_build, clean_s1, clean_sw, bootstrap,
2355 bootstrap-lean, bootstrap2, bootstrap2-lean, bootstrap3,
2356 bootstrap3-lean, bootstrap4, bootstrap4-lean, unstage1, unstage2,
2357 unstage3, unstage4, unstageprofile, unstagefeedback, restage, restage2,
2358 restage3, restage4, restageprofile, restagefeedback, bubbleestrap,
2359 cleanstrap, unstrap, restrap, *compare, *compare3, *compare4,
2360 *compare-lean, *compare3-lean, *compare4-lean, stage1-start, stage1,
2361 stage2-start, stage2, stage3-start, stage3, stage4-start, stage4,
2362 stageprofile-start, stageprofile, stagefeedback-start, stagefeedback,
2363 risky-stage1, risky-stage2, risky-stage3, risky-stage4): Remove.
2364 (ORDINARY_FLAGS_TO_PASS): Rename to FLAGS_TO_PASS.
2365 (STAGECOPYSTUFF, STAGEMOVESTUFF): Consolidate into MOSTLYCLEANFILES.
2366 (mostlyclean): Adjust.
2367 (clean, distclean): Don't mention bootstrap stuff.
2368 * configure: Regenerate.
2369 * ada/config-lang.in, cp/config-lang.in, forttran/config-lang.in,
2370 java/config-lang.in, objc/config-lang.in, objcp/config-lang.in,
2371 treelang/config-lang.in (stagestuff): Remove.
2372 * doc/sourcebuild.texi (stage1, stage2, stage3, stage4,
2373 stageprofile, stagefeedback, stagestuff): Remove mention.
2375 2007-01-11 Nick Clifton <nickc@redhat.com>
2377 * config/mcore/predicates.md (mcore_general_movesrc_operand):
2379 (mcore_general_movdst_operand): Do not accept CONST_INTs.
2380 (mcore_arith_K_S_operand): Run the test for the S constraint not
2381 the test for the M constraint.
2382 (mcore_addsub_operand): Do not accept integer values that are
2383 larger than 32 bits.
2384 * config/mcore/mcore.md: Remove unused constraints from split.
2385 (andsi3): Use HOST_WIDE_INT instead of int to hold an INTVAL.
2387 (allocate_stack): Likewise.
2388 * config/mcore/mcore.c (mcore_print_operand): Restrict output of P
2389 operands to 32 bits.
2390 (mcore_const_costs): Use HOST_WIDE_INT instead of int to hold an
2392 (mcore_and_cost, mcore_modify_comparison, const_ok_for_mcore,
2393 mcore_const_ok_for_inline, mcore_const_trick_uses_not,
2394 try_constant_tricks, mcore_num_ones, mcore_num_zeros,
2395 mcore_output_bclri, mcore_output_andn, output_inline_const,
2396 mcore_output_move, mcore_output_movedouble): Likewise.
2397 (mcore_output_cmov): Use CONST_OK_FOR_M and CONST_OK_FOR_N.
2398 (output_inline_const): Likewise.
2399 (output_inline_const): Fix format strings used in sprintf
2401 * config/mcore/mcore-protos.h: Update prototypes for changed
2402 functions in mcore.c.
2403 * config/mcore/mcore.h (CONST_OK_FOR_I): Cast values to
2404 HOST_WIDE_INT and not int.
2405 (CONST_OK_FOR_J, CONST_OK_FOR_K, CONST_OK_FOR_L, CONST_OK_FOR_M,
2406 CONST_OK_FOR_N): Likewise.
2407 (LEGITIMATE_CONSTANT_P): Also check CONSTANT_P.
2408 (GO_IF_LEGITIMATE_INDEX): Use HOST_WIDE_INT instead of int to hold
2411 2007-01-10 Jan Hubicka <jh@suse.cz>
2413 * tree-vrp.c (remove_range_assertions): Release defs.
2414 * tree-ssa-loop-ivopts.c (rmeove_statement): Likewise.
2415 * tree-ssa-dom.c (remove_stmt_or_phi): Likewise.
2417 2007-01-10 Paul Brook <paul@codesourcery.com>
2419 * config/arm/arm.c (arm_rtx_costs_1): Handle mutiply-subtract.
2420 * config/arm/arm.md (mulsi3subsi): New insn.
2422 2007-01-10 Zdenek Dvorak <dvorakz@suse.cz>
2424 * tree-ssa-loop-manip.c (tree_unroll_loop): Make it a wrapper over ...
2425 (tree_transform_and_unroll_loop): New.
2426 * tree-flow.h (transform_callback, tree_transform_and_unroll_loop):
2429 2007-01-10 Robert Kennedy <jimbob@google.com>
2431 * fold-const.c (fold_comparison): Fold comparisons like (x *
2432 1000 < 0) to (x < 0).
2434 2007-01-10 Ian Lance Taylor <iant@google.com>
2436 * tree-pretty-print.c (dump_generic_node): Print parentheses when
2437 operands have the same priority.
2439 2007-01-10 Tom Tromey <tromey@redhat.com>
2441 * fold-const.c (fold_truthop): Don't check can_use_bit_fields_p.
2442 (fold_binary): Likewise.
2443 * langhooks.c (lhd_can_use_bit_fields_p): Removed.
2444 * langhooks-def.h (lhd_can_use_bit_fields_p): Removed.
2445 (LANG_HOOKS_CAN_USE_BIT_FIELDS_P): Removed.
2446 (LANG_HOOKS_INITIALIZER): Remove LANG_HOOKS_CAN_USE_BIT_FIELDS_P.
2447 * langhooks.h (struct lang_hooks): Removed field
2448 'can_use_bit_fields_p'.
2450 2007-01-10 Ralf Corsépius <ralf.corsepius@rtems.org>
2452 * config/bfin/t-bfin, config/bfin/t-bfin-elf: Remove GCC_CFLAGS.
2454 2007-01-10 Razya Ladelsky <razya@il.ibm.com>
2456 * function.c (get_last_funcdef_no): New function.
2457 * function.h (get_last_funcdef_no): Declare.
2458 * tree-inline.c (initialize_cfun): Add initialization.
2459 (tree_function_versioning): Cleanup.
2461 2007-01-10 Jan Hubicka <jh@suse.cz>
2463 * tree-inline.c (setup_one_parameter): Do not propagate into abnormal
2466 2007-01-10 Sa Liu <saliu@de.ibm.com>
2467 Ben Elliston <bje@au.ibm.com>
2469 * spu.h (STACK_SAVE_AREA): Use VOIDmode for SAVE_FUNCTION, SImode
2470 for SAVE_NONLOCAL and Pmode for any other save level.
2471 * spu-protos.h (spu_restore_stack_block): Declare.
2472 * spu.md (save_stack_block): Remove.
2473 (restore_stack_block): Call spu_restore_stack_block.
2474 * spu.c (spu_restore_stack_block): New function.
2475 (spu_expand_epilogue): Remove old comment.
2477 2007-01-09 Zdenek Dvorak <dvorakz@suse.cz>
2479 PR tree-optimization/30322
2480 * tree-ssa-loop-ivopts.c (fold_affine_expr, iv_value): Removed.
2481 (cand_value_at): Return the value as aff_tree.
2482 (may_eliminate_iv): Convert the bound from aff_tree to tree.
2483 * tree-affine.c (aff_combination_add_cst, aff_combination_add_product,
2484 aff_combination_mult): New functions.
2485 (aff_combination_add): Use aff_combination_add_cst.
2486 (aff_combination_convert): Allow conversions to a wider type.
2487 (tree_to_aff_combination): Handle BIT_NOT_EXPR.
2488 * tree-affine.h (aff_combination_mult): Declare.
2490 2007-01-09 Carlos O'Donell <carlos@codesourcery.com>
2492 * doc/tm.texi: Update documentation to reflect reality of exec
2493 and start file search behaviours. Update copyright year.
2494 * doc/invoke.texi: Explain how GCC_EXEC_PREFIX is used to find
2495 header file directories.
2497 2007-01-09 Uros Bizjak <ubizjak@gmail.com>
2499 * config/i386/i386.md (*sinxf2): Rename to *sinxf2_i387.
2500 (*cosxf2): Rename to cosxf2_i387.
2501 (*sindf2, *sinsf2): Extend operand 1 to XFmode. Macroize patterns
2502 using X87MODEF12 mode macro. Rename patterns to
2503 *sin_extend<mode>xf2_i387. Use SSE_FLOAT_MODE_P to disable patterns
2505 (*cosdf2, *cossf2): Ditto.
2506 (sincosdf3, sincossf3): Ditto. Rewrite corresponding splitters
2507 to match extended input operands.
2508 (sincos<mode>3): New expander.
2509 (*sinextendsfdf2, *cosextendsfdf2, *sincosextendsfdf3): Remove
2510 insn patterns and corresponding splitters.
2512 2007-01-09 Kaz Kojima <kkojima@gcc.gnu.org>
2514 * config/sh/t-linux (TARGET_LIBGCC2_CFLAGS): Delete.
2515 (SHLIB_MAPFILES, SHLIB_LINK, SHLIB_INSTALL): Likewise.
2517 2007-01-09 Nicolas Pitre <nico@cam.org>
2520 * arm/ieee754-df.S (Lad_s): Also test the low word of X for zero.
2522 2007-01-08 Geoffrey Keating <geoffk@apple.com>
2524 * target.h (struct gcc_target): New field library_rtti_comdat.
2525 * target-def.h (TARGET_CXX_LIBRARY_RTTI_COMDAT): New.
2526 (TARGET_CXX): Add TARGET_CXX_LIBRARY_RTTI_COMDAT.
2527 * doc/tm.texi (C++ ABI): Document TARGET_CXX_LIBRARY_RTTI_COMDAT.
2528 * config/darwin.h (TARGET_CXX_LIBRARY_RTTI_COMDAT): Define.
2530 2007-01-08 Geoffrey Keating <geoffk@apple.com>
2532 * doc/invoke.texi (Optimize Options): Correct description of -O0.
2534 2007-01-08 Richard Guenther <rguenther@suse.de>
2536 * tree.h (force_fit_type_double): Export.
2537 (force_fit_type): Remove.
2538 * fold-const.c (force_fit_type_double): New function.
2539 (force_fit_type): Remove.
2540 (int_const_binop): Use it.
2541 (fold_convert_const_int_from_int): Likewise.
2542 (fold_convert_const_int_from_real): Likewise.
2543 (fold_div_compare): Likewise.
2544 (fold_sign_changed_comparison): Likewise.
2545 (fold_unary): Likewise.
2546 (fold_negate_const): Likewise.
2547 (fold_abs_const): Likewise.
2548 (fold_not_const): Likewise.
2549 * c-common.c (shorten_compare): Use force_fit_type_double.
2550 * convert.c (convert_to_pointer): Likewise.
2552 2007-01-08 Richard Guenther <rguenther@suse.de>
2554 * tree.h (build_int_cst_wide_type): Export.
2555 * tree.c (build_int_cst_wide_type): New function.
2556 (build_int_cst_wide): Fix comment.
2557 * builtins.c (fold_builtin_object_size): Use build_int_cst
2558 to build -1 or 0 of the correct type. Use fit_double_type
2559 to check for overflow.
2560 * fold-const.c (optimize_bit_field_compare): Use build_int_cst_type
2562 (decode_field_reference): Likewise.
2563 (all_ones_mask_p): Likewise.
2564 (native_interpret_int): Use build_int_cst_wide_type.
2565 (fold_binary): Use build_int_cst_type to build an all-ones
2567 * stor-layout.c (set_sizetype): Use build_int_cst_wide_type.
2569 2007-01-08 Daniel Jacobowitz <dan@codesourcery.com>
2571 * config/pa/t-pa64 (libgcc_stub.a): Use $(T).
2573 2007-01-09 Ben Elliston <bje@au.ibm.com>
2575 * genautomata.c (STATS_OPTION): New option.
2576 (stats_flag): New flag.
2577 (gen_automata_option): Handle it.
2578 (initiate_automaton_gen): Ditto.
2579 (write_automata): Output statistics only if stats_flag is
2580 set. Likewise, output time statistics only if time_flag is set.
2581 * doc/md.texi (Processor pipeline description): Document new flag.
2583 2007-01-08 Richard Guenther <rguenther@suse.de>
2585 * builtins.c (fold_builtin_int_roundingfn): Use fit_double_type.
2586 * tree.c (build_int_cst_type): Likewise.
2587 (size_in_bytes): Don't call force_fit_type on the result.
2588 (int_fits_type_p): Use fit_double_type.
2589 * fold-const.c (fit_double_type): New function.
2590 (force_fit_type): Use it.
2591 * tree.h (fit_double_type): Export.
2593 2007-01-08 Jan Hubicka <jh@suse.cz>
2595 * tree-vectorizer.c (gate_increase_alignment): Fix return type.
2596 * ipa.c (function_and_variable_visibility): Fix return type.
2598 2007-01-08 Richard Guenther <rguenther@suse.de>
2600 * tree-ssa-ccp.c (maybe_fold_offset_to_array_ref): Use type
2601 of offset to build the index.
2602 * tree-pretty-print.c (dump_generic_node): Don't build negated
2603 const just for printing.
2604 * c-pretty-print.c (pp_c_integer_constant): Likewise.
2605 * builtins.c (fold_builtin_int_roundingfn): Check if result
2606 fits the type by using force_fit_type and comparing the result.
2607 * predict.c (predict_loops): Use compare_tree_int for comparison.
2608 * tree.c (build_int_cst): Fall back to integer_type_node for
2610 (build_int_cst_wide): Assert type is non-null.
2612 2007-01-08 Roberto Costa <roberto.costa@st.com>
2614 * tree-vrp.c (extract_range_from_cond_expr): New.
2615 (extract_range_from_expr): Handle COND_EXPR nodes used as expressions.
2616 * tree-ssa-ccp.c (get_maxval_strlen): Handle COND_EXPR nodes used
2618 (fold_stmt): Bug fix, avoid infinite recursion when folding COND_EXPRs.
2619 * tree-ssa-forwprop.c (simplify_cond, forward_propagate_into_cond,
2620 tree_ssa_forward_propagate_single_use_vars): Handle COND_EXPR nodes
2621 used as expressions.
2622 * tree-object-size.c (cond_expr_object_size): New.
2623 (collect_object_sizes_for): Handle COND_EXPR nodes used as expressions.
2625 2007-01-08 Jan Hubicka <jh@suse.cz>
2627 * tree-ssa-forwprop.c (forward_propagate_into_cond,
2628 tree_ssa_forward_propagate_single_use_va): Release defs of propagated
2631 2007-01-08 Richard Guenther <rguenther@suse.de>
2633 PR tree-optimization/23603
2634 * tree-vrp.c (set_value_range_to_truthvalue): New function.
2635 (extract_range_from_binary): Fall back to truthvalue instead of
2636 varying for TRUTH_*_EXPR.
2637 (extract_range_from_comparison): Fall back to truthvalue instead of
2639 (vrp_visit_phi_node): Don't adjust new range bounds to +INF/-INF
2640 if all visited PHI values were constant.
2642 2007-01-08 Jan Hubicka <jh@suse.cz>
2644 * cgraphunit.c (cgraph_process_new_functions): Reset reachable flag.
2645 (cgraph_analyze_function): break out from ...
2646 (cgraph_finalize_compilation_unit): ... here.
2647 (cgraph_expand_function): Remove forgoten commented out line.
2648 (cgraph_optimize): Analyze functions.
2650 2007-01-08 Jan Hubicka <jh@suse.cz>
2652 * tree-pas.h (TODO_remove_function): New flag.
2653 (TODO_update*): Renumber.
2654 (pass_ipa_increase_alignment,
2655 pass_ipa_function_and_variable_visibility): New passes.
2656 * cgraphunit.c (cgraph_increase_alignment): Move to tree-vectorizer.c
2657 (cgraph_function_and_variable_visibility): Move to ipa.c
2658 (cgraph_optimize): Don't call cgraph_function_and_variable_visibility,
2659 cgraph_increase_alignment.
2660 * ipa-inline.c (cgraph_decide_inlining): Don't push timevar.
2661 (cgraph_decide_inlining_incrementally): Push TV_INTEGRATION before
2662 calling tree-inline.
2663 (cgraph_early_inlining): Do not call cgraph_remove_unreachable_nodes.
2664 (pass_ipa_inline, pass_early_ipa_inlining): Set TODO_remove_functions
2665 * tree-vectorizer.c (increase_alignment): Move here from cgraphunit.c
2666 (gate_increase_alignment): New function.
2667 (pass_ipa_increase_alignment): New pass.
2668 * ipa.c: Inline tree-pass.h and timevar.h
2669 (function_and_variable_visibility): Move here from cgraphunit.c
2670 * tree-optimize.c (pass_early_local_passes): Add TODO_remove_functions.
2671 * passes.c (init_optimization_passes): Add the two new passes.
2672 (execute_todo): Handle cgraph_remove_functions.
2674 2007-01-08 Nick Clifton <nickc@redhat.com>
2676 * config/frv/predicates.md (reg_or_0_operand): Accept
2679 2007-01-08 Ralf Corsépius <ralf.corsepius@rtems.org>
2681 * config/bfin/rtems.h, config/bfin/t-rtems: New.
2682 * config.gcc: Add bfin*-rtems*.
2684 2007-01-08 Mark Shinwell <shinwell@codesourcery.com>
2686 * c.opt: Add -flax-vector-conversions.
2687 * c-typeck.c (convert_for_assignment): Pass flag to
2688 vector_types_convertible_p to allow emission of note.
2689 (digest_init): Likewise.
2690 * c-opts.c: Handle -flax-vector-conversions.
2691 * c-common.c (flag_lax_vector_conversions): New.
2692 (vector_types_convertible_p): Unless -flax-vector conversions
2693 has been passed, disallow conversions between vectors with
2694 differing numbers of subparts and/or element types. If such
2695 a conversion is disallowed, possibly emit a note on the first
2696 occasion only to inform the user of -flax-vector-conversions.
2697 The new last argument specifies this.
2698 * c-common.h (flag_lax_vector_conversions): New.
2699 (vector_types_convertible_p): Add extra argument.
2700 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use
2701 char_type_node for V*QI type vectors.
2702 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
2703 Update to satisfy new typechecking rules.
2704 * config/rs6000/altivec.h (vec_cmple): Use vec_cmpge, for both
2706 * doc/invoke.texi (C Dialect Options): Document
2707 -flax-vector-conversions.
2709 2007-01-08 Mark Shinwell <shinwell@codesourcery.com>
2711 PR tree-optimization/29877
2712 * tree-ssa-ter.c (is_replaceable_p): Deem assignments with
2713 a register variable on the RHS to not be replaceable.
2715 2007-01-08 Chen Liqin <liqin@sunnorth.com.cn>
2716 * config/score/t-score-elf (MULTILIB_OPTIONS): Change.
2717 * config/score/predicates.md (const_uimm5, sr0_operand, const_simm12,
2718 const_simm15, const_pow2, const_npow2): Added.
2719 * config/score/misc.md (insv, extv, extzv, movmemsi,
2720 move_lbu_a/b, mov_lhu_a/b etc): Added and fix some bug.
2721 * config/score/score.c (score_address_cost, score_select_cc_mode):
2723 Change CONST_OK_FOR_LETTER_P/EXTRA_CONSTRAINT define.
2724 Update score_rtx_costs for MACRO TARGET_RTX_COSTS.
2725 Update score_print_operand.
2726 * config/score/score.h (DATA_ALIGNMENT, SELECT_CC_MODE): Added.
2727 Adjust register allocate order and update some macro define.
2728 * config/score/score-mdaux.c (mdx_unaligned_load, mdx_unsigned_store,
2729 mdx_block_move_straight, mdx_block_move_loop_head,
2730 mdx_block_move_loop_body, mdx_block_move_loop_foot, mdx_block_move_loop,
2731 mdx_block_move): Added.
2732 (mdx_movsicc, mdp_select_add_imm, mdp_select, mds_zero_extract_andi,
2733 mdp_limm): Updated and fix some bug and typo.
2734 * config/score/score.md (movqi/hi/si, add/sub/zero/ext): Updated.
2735 (movsf, movdf, doloop_end): Added.
2737 2007-01-08 Kazu Hirata <kazu@codesourcery.com>
2739 * config/arm/arm.c, config/arm/arm.h, config/arm/arm.md,
2740 config/arm/thumb2.md: Fix comment typos.
2741 * doc/extend.texi: Fix a typo.
2743 2007-01-07 Eric Christopher <echristo@apple.com>
2745 * configure.ac: Check for __stack_chk_fail for darwin.
2746 * configure: Regenerate.
2748 2007-01-07 Richard Guenther <rguenther@suse.de>
2750 * tree-vrp.c (extract_range_from_assert): CSE calls to
2751 compare_values where possible.
2752 (extract_range_from_unary_expr): Likewise.
2754 2007-01-07 Anatoly Sokolov <aesok@post.ru>
2756 * config/avr/avr-protos.h (call_insn_operand): Delete prototype.
2757 * config/avr/avr.c (call_insn_operand): Delete function.
2758 * config/avr/avr.md (*pushqi, *pushhi, *pushsi, *pushsf): Use REG_SP
2759 instead of register number. Use predicates.
2760 * config/avr/predicates.md (const0_operand, reg_or_0_operand,
2761 call_insn_operand): Add.
2763 2007-01-06 Jan Hubicka <jh@suse.cz>
2765 * tree-pass.h (pass_build_cgraph_edges): Declare.
2766 * cgraphunit.c (record_refernece): Move to cgraphbuild.c
2767 (visited_nodes): Remove.
2768 (cgraph_create_edges): Move to cgraphbuild.c; rename to
2769 build_cgrpah_edges; make visited_nodes local.
2770 (cgraph_process_new_functions): DO not call initialize_inline_failed.
2771 (record_references_in_initializer): Move to cgraphbuild.c
2772 (initialize_inline_failed, rebuild_cgraph_edges,
2773 pass_rebuild_cgraph_edges): Move to cgraphbuild.c.
2774 (verify_cgraph_node): Make visited_nodes local.
2775 (cgraph_analyze_function): Do not call cgraph_create_edges and
2776 initialize_inline_failed.
2777 (cgraph_expand_function): Do not call cgraph_lower_function;
2778 assert that function is already lowered.
2779 * Makefile.in (cgraphbuild.o): New.
2780 * passes.c (init_optimization_passes): Add pass_build_cgraph_edges
2781 at the end of lowering passes.
2783 2007-01-06 Steven Bosscher <steven@gcc.gnu.org>
2785 * ifcvt.c (cond_move_convert_if_block): New function, code
2786 factored out from...
2787 (cond_move_process_if_block): ...here. Call the new function
2788 on the THEN and ELSE blocks.
2789 (merge_if_block): Do not copy global_live_at_end, merge_blocks
2790 already takes care of this.
2792 2007-01-05 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
2795 * tree.h (TREE_OVERFLOW_P): New.
2796 * c-typeck.c (parser_build_unary_op): Warn only if result
2797 overflowed and operands did not.
2798 (parser_build_binary_op): Likewise.
2799 (convert_for_assignment): Remove redundant overflow_warning.
2800 * c-common.c (overflow_warning): Don't check or set TREE_OVERFLOW.
2802 2007-01-05 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
2804 * c-typeck.c (store_init_value): Split over two lines to follow
2805 the GNU coding style.
2807 2007-01-05 Benjamin Kosnik <bkoz@redhat.com>
2809 * c-cppbuiltin.c (c_cpp_builtins): __GXX_EXPERIMENTAL_CPP0X__ to
2810 __GXX_EXPERIMENTAL_CXX0X__.
2811 * doc/cpp.texi: Same.
2813 2007-01-05 Richard Guenther <rguenther@suse.de>
2816 * tree.c (get_narrower): Do not construct COMPONENT_REFs
2817 with mismatched types. Instead explicitly build a
2818 conversion NOP_EXPR.
2820 2007-01-05 Ian Lance Taylor <iant@google.com>
2822 * c-common.c (decl_with_nonnull_addr_p): New function.
2823 (c_common_truthvalue_conversion): Call it.
2824 * c-typeck.c (build_binary_op): Likewise.
2825 * c-common.h (decl_with_nonnull_addr_p): Declare.
2827 2007-01-05 Jakub Jelinek <jakub@redhat.com>
2830 * libgcc2.c (__divdc3): Compare c and d against 0.0 instead of
2833 2007-01-05 Joel Brobecker <brobecker@adacore.com>
2835 * doc/install.texi (Final install): Document the fact that
2836 the GNAT runtime should not be stripped.
2838 2007-01-04 Jan Hubicka <jh@suse.cz>
2840 * tree-inline.c (fold_marked_statements): Update operand caches
2841 and EH after folding
2843 2007-01-04 Ian Lance Taylor <iant@google.com>
2845 * c-common.c (check_function_nonnull): Whitespace fix.
2847 2007-01-04 Jan Hubicka <jh@suse.cz>
2849 * tree-optimize.c (execute_fixup_cfg): Correct previously mistakely
2850 comitted older version of patch.
2851 (pass_fixup_cfg): Add TODOs to verify flow and statements, dump
2852 function, celanup cfg and collect garbage.
2854 2007-01-04 Mike Stump <mrs@apple.com>
2856 * Makefile.in (mostlyclean): Don't remove libgcc anymore.
2859 2007-01-04 Eric Christopher <echristo@apple.com>
2861 * libgcc2.c (__bswapsi2): Use SItype.
2862 (__bswapdi2): Use DItype.
2863 * libgcc2.h: Update for above.
2865 2007-01-04 Paul Brook <paul@codesourcery.com>
2867 * config/arm/arm.md (arm_mulsi3, thumb_mulsi3, mulsi3_compare0,
2868 mulsi_compare0_scratch, mulsi3addsi, mulsi3addsi_compare0,
2869 mulsi3addsi_compare0_scratch, mulsidi3adddi, mulsidi3,
2870 umulsidi3, umulsidi3adddi, smulsi3_highpart,
2871 umulsi3_highpart): Make conditional on !arm_arch6.
2872 (arm_mulsi3_v6, thumb_mulsi3_v6, mulsi3_compare0_v6,
2873 mulsi_compare0_scratch_v6, mulsi3addsi_v6, mulsi3addsi_compare0_v6,
2874 mulsi3addsi_compare0_scratch_v6, mulsidi3adddi_v6, mulsidi3_v6,
2875 umulsidi3_v6, umulsidi3adddi_v6, smulsi3_highpart_v6,
2876 umulsi3_highpart_v6): New insns.
2878 2007-01-04 Roger Sayle <roger@eyesopen.com>
2880 * fold-const.c (fold_convert): When casting an expression to void,
2881 fold_ignored_result may discover a GIMPLE_MODIFY_STMT which doesn't
2882 have a type. Instead of attempting to build a NOP_EXPR, return
2883 these "special" trees directly.
2885 2007-01-04 Joseph Myers <joseph@codesourcery.com>
2887 * config/rs6000/rs6000.c (rs6000_rtx_costs): Make adjustment for
2888 MULT inside MINUS as either argument. Use rs6000_cost->dmul -
2889 rs6000_cost->fp not 0 as adjustment for outer NEG.
2891 2007-01-04 Jan Hubicka <jh@suse.cz>
2893 * cgraph.c (cgraph_release_function_body): New function.
2894 (cgraph_remove_node): Use it.
2895 * cgraph.h (cgraph_release_function_body): Declare.
2896 * cgraphunit.c (cgraph_expand_function): Use it.
2897 * ipa.c (cgraph_remove_unreahchable_nodes): Use it.
2898 * tree-ssa.c (delete_tree_ssa): Allow to be called before aliasing
2899 is initialized and while compilation of other function is running.
2900 * tree-optimize.c (execute_free_cfg_annotations): Move code to clear
2901 statement CFG annotations from here to ...
2902 * tree-cfg.c (delete_tree_cfg_annotations): ... here.
2904 2007-01-04 Zdenek Dvorak <dvorakz@suse.cz>
2906 * cfgloop.h (enum li_flags): Make the constants powers of two.
2908 2007-01-04 Jan Hubicka <jh@suse.cz>
2910 * tree-inline.c (copy_bb): Insert new statements to statements_to_fold
2912 (fold_marked_statements): New function.
2913 (optimize_inline_calls, tree_function_versioning): Fold new statements.
2914 * tree-inline.h (copy_body_data): Add statements_to_fold.
2916 2007-01-03 Daniel Jacobowitz <dan@codesourcery.com>
2918 * config.gcc: Mention libgcc/config.host.
2919 * Makefile.in: Update comments mentioning libgcc.
2920 (LIBGCC, INSTALL_LIBGCC, GCC_PARTS, mklibgcc): Delete.
2921 (all.cross, start.encap, rest.encap, rest.cross): Update
2922 dependencies for libgcc move.
2923 (libgcc.mk, LIBGCC_DEPS, libgcov.a, libgcc.a, stmp-multilib)
2924 (clean-target, clean-target-libgcc): Delete.
2925 (srcdirify, GCC_EXTRA_PARTS): New macros.
2926 (libgcc-support, libgcc.mvars): New rules.
2927 (distclean): Remove mention of mklibgcc.
2928 (install): Don't reference INSTALL_LIBGCC.
2929 (install-common): Don't reference EXTRA_PARTS.
2930 (install-libgcc, install-multilib): Delete rules.
2931 * mklibgcc.in: Delete file.
2932 * doc/configfiles.texi: Don't mention mklibgcc.
2934 * config/i386/t-darwin (SHLIB_VERPFX): Delete (moved to libgcc).
2935 * config/i386/t-darwin64 (SHLIB_VERPFX): Likewise.
2936 * config/rs6000/t-darwin (SHLIB_VERPFX): Likewise.
2937 * config/rs6000/t-ppccomm (TARGET_LIBGCC2_CFLAGS, SHLIB_MAPFILES)
2938 (mklibgcc, ldblspecs): Likewise.
2940 * config/i386/t-nwld (libgcc.def, libc.def, libpcre.def)
2941 (posixpre.def): Use $(T).
2942 (SHLIB_EXT, SHLIB_NAME, SHLIB_SLIBDIR_QUAL, SHLIB_DEF, SHLIB_MAP)
2943 (SHLIB_SRC, SHLIB_INSTALL): Delete.
2944 (SHLIB_LINK): Make dummy.
2945 * config/t-slibgcc-darwin: Delete contents except for dummy SHLIB_LINK.
2947 * config/frv/t-linux (EXTRA_MULTILIB_PARTS): Clear.
2949 * config/alpha/t-crtfm: Use $(T) in rules for EXTRA_PARTS.
2950 * config/alpha/t-vms, config/alpha/t-vms64, config/fr30/t-fr30,
2951 config/i386/t-rtems-i386, config/ia64/t-ia64, config/rs6000/t-beos,
2952 config/rs6000/t-newas, config/sparc/t-elf: Likewise.
2954 * configure.ac (all_outputs): Remove mklibgcc.
2955 * configure: Regenerated.
2957 2007-01-03 Josh Conner <jconner@apple.com>
2960 * calls.c (compute_argument_addresses): Set stack and stack_slot
2961 for partial args, too.
2962 (store_one_arg): Use locate.size.constant for the size when
2963 generating a save_area.
2965 2007-01-03 Robert Kennedy <jimbob@google.com>
2967 * tree-cfg.c (tree_merge_blocks): Release SSA_NAME phi results
2968 whose definitions are deleted due to basic block merging.
2970 2007-01-03 Paul Brook <paul@codesourcery.com>
2973 * config/arm/arm.c (output_return_instruction): Pop PC in interrupt
2975 (use_return_insn): Return 0 for Thumb interrupt functions.
2976 (print_multi_reg): Add rfe argument for IRQ returns.
2977 (arm_output_epilogue): Pop interrupt return address directly into PC.
2978 (arm_expand_prologue): Only adjust IRQ return address in Arm mode.
2980 2007-01-03 Paul Brook <paul@codesourcery.com>
2982 Merge from sourcerygxx-4_1.
2983 * config/arm/thumb2.md: New file.
2984 * config/arm/elf.h (JUMP_TABLES_IN_TEXT_SECTION): Return True for
2986 * config/arm/coff.h (JUMP_TABLES_IN_TEXT_SECTION): Ditto.
2987 * config/arm/aout.h (ASM_OUTPUT_ADDR_VEC_ELT): Add !Thumb-2 assertion.
2988 (ASM_OUTPUT_ADDR_DIFF_ELT): Output Thumb-2 jump tables.
2989 * config/arm/aof.h (ASM_OUTPUT_ADDR_DIFF_ELT): Output Thumb-2 jump
2991 (ASM_OUTPUT_ADDR_VEC_ELT): Add !Thumb-2 assertion.
2992 * config/arm/ieee754-df.S: Use macros for Thumb-2/Unified asm
2994 * config/arm/ieee754-sf.S: Ditto.
2995 * config/arm/arm.c (thumb_base_register_rtx_p): Rename...
2996 (thumb1_base_register_rtx_p): ... to this.
2997 (thumb_index_register_rtx_p): Rename...
2998 (thumb1_index_register_rtx_p): ... to this.
2999 (thumb_output_function_prologue): Rename...
3000 (thumb1_output_function_prologue): ... to this.
3001 (thumb_legitimate_address_p): Rename...
3002 (thumb1_legitimate_address_p): ... to this.
3003 (thumb_rtx_costs): Rename...
3004 (thumb1_rtx_costs): ... to this.
3005 (thumb_compute_save_reg_mask): Rename...
3006 (thumb1_compute_save_reg_mask): ... to this.
3007 (thumb_final_prescan_insn): Rename...
3008 (thumb1_final_prescan_insn): ... to this.
3009 (thumb_expand_epilogue): Rename...
3010 (thumb1_expand_epilogue): ... to this.
3011 (arm_unwind_emit_stm): Rename...
3012 (arm_unwind_emit_sequence): ... to this.
3013 (thumb2_legitimate_index_p, thumb2_legitimate_address_p,
3014 thumb1_compute_save_reg_mask, arm_dwarf_handle_frame_unspec,
3015 thumb2_index_mul_operand, output_move_vfp, arm_shift_nmem,
3016 arm_save_coproc_regs, thumb_set_frame_pointer, arm_print_condition,
3017 thumb2_final_prescan_insn, thumb2_asm_output_opcode, arm_output_shift,
3018 thumb2_output_casesi): New functions.
3019 (TARGET_DWARF_HANDLE_FRAME_UNSPEC): Define.
3020 (FL_THUMB2, FL_NOTM, FL_DIV, FL_FOR_ARCH6T2, FL_FOR_ARCH7,
3021 FL_FOR_ARCH7A, FL_FOR_ARCH7R, FL_FOR_ARCH7M, ARM_LSL_NAME,
3022 THUMB2_WORK_REGS): Define.
3023 (arm_arch_notm, arm_arch_thumb2, arm_arch_hwdiv, arm_condexec_count,
3024 arm_condexec_mask, arm_condexec_masklen)): New variables.
3025 (all_architectures): Add armv6t2, armv7, armv7a, armv7r and armv7m.
3026 (arm_override_options): Check new CPU capabilities.
3027 Set new architecture flag variables.
3028 (arm_isr_value): Handle v7m interrupt functions.
3029 (user_return_insn): Return 0 for v7m interrupt functions. Handle
3031 (const_ok_for_arm): Handle Thumb-2 constants.
3032 (arm_gen_constant): Ditto. Use movw when available.
3033 (arm_function_ok_for_sibcall): Return false for v7m interrupt
3035 (legitimize_pic_address, arm_call_tls_get_addr): Handle Thumb-2.
3036 (thumb_find_work_register, arm_load_pic_register,
3037 legitimize_tls_address, arm_address_cost, load_multiple_sequence,
3038 emit_ldm_seq, emit_stm_seq, arm_select_cc_mode, get_jump_table_size,
3039 print_multi_reg, output_mov_long_double_fpa_from_arm,
3040 output_mov_long_double_arm_from_fpa, output_mov_double_fpa_from_arm,
3041 output_mov_double_fpa_from_arm, output_move_double,
3042 arm_compute_save_reg_mask, arm_compute_save_reg0_reg12_mask,
3043 output_return_instruction, arm_output_function_prologue,
3044 arm_output_epilogue, arm_get_frame_offsets, arm_regno_class,
3045 arm_output_mi_thunk, thumb_set_return_address): Ditto.
3046 (arm_expand_prologue): Handle Thumb-2. Use arm_save_coproc_regs.
3047 (arm_coproc_mem_operand): Allow POST_INC/PRE_DEC.
3048 (arithmetic_instr, shift_op): Use arm_shift_nmem.
3049 (arm_print_operand): Use arm_print_condition. Handle '(', ')', '.',
3051 (arm_final_prescan_insn): Use extract_constrain_insn_cached.
3052 (thumb_expand_prologue): Use thumb_set_frame_pointer.
3053 (arm_file_start): Output directive for unified syntax.
3054 (arm_unwind_emit_set): Handle stack alignment instruction.
3055 * config/arm/lib1funcs.asm: Remove default for __ARM_ARCH__.
3056 Add v6t2, v7, v7a, v7r and v7m.
3057 (RETLDM): Add Thumb-2 code.
3058 (do_it, shift1, do_push, do_pop, COND, THUMB_SYNTAX): New macros.
3059 * config/arm/arm.h (TARGET_CPU_CPP_BUILTINS): Define __thumb2__.
3060 (TARGET_THUMB1, TARGET_32BIT, TARGET_THUMB2, TARGET_DSP_MULTIPLY,
3061 TARGET_INT_SIMD, TARGET_UNIFIED_ASM, ARM_FT_STACKALIGN, IS_STACKALIGN,
3062 THUMB2_TRAMPOLINE_TEMPLATE, TRAMPOLINE_ADJUST_ADDRESS,
3063 ASM_OUTPUT_OPCODE, THUMB2_GO_IF_LEGITIMATE_ADDRESS,
3064 THUMB2_LEGITIMIZE_ADDRESS, CASE_VECTOR_PC_RELATIVE,
3065 CASE_VECTOR_SHORTEN_MODE, ADDR_VEC_ALIGN, ASM_OUTPUT_CASE_END,
3066 ADJUST_INSN_LENGTH): Define.
3067 (TARGET_REALLY_IWMMXT, TARGET_IWMMXT_ABI, CONDITIONAL_REGISTER_USAGE,
3068 STATIC_CHAIN_REGNUM, HARD_REGNO_NREGS, INDEX_REG_CLASS,
3069 BASE_REG_CLASS, MODE_BASE_REG_CLASS, SMALL_REGISTER_CLASSES,
3070 PREFERRED_RELOAD_CLASS, SECONDARY_OUTPUT_RELOAD_CLASS,
3071 SECONDARY_INPUT_RELOAD_CLASS, LIBCALL_VALUE, FUNCTION_VALUE_REGNO_P,
3072 TRAMPOLINE_SIZE, INITIALIZE_TRAMPOLINE, HAVE_PRE_INCREMENT,
3073 HAVE_POST_DECREMENT, HAVE_PRE_DECREMENT, HAVE_PRE_MODIFY_DISP,
3074 HAVE_POST_MODIFY_DISP, HAVE_PRE_MODIFY_REG, HAVE_POST_MODIFY_REG,
3075 REGNO_MODE_OK_FOR_BASE_P, LEGITIMATE_CONSTANT_P,
3076 REG_MODE_OK_FOR_BASE_P, REG_OK_FOR_INDEX_P, GO_IF_LEGITIMATE_ADDRESS,
3077 LEGITIMIZE_ADDRESS, THUMB2_LEGITIMIZE_ADDRESS,
3078 GO_IF_MODE_DEPENDENT_ADDRESS, MEMORY_MOVE_COST, BRANCH_COST,
3079 ASM_APP_OFF, ASM_OUTPUT_CASE_LABEL, ARM_DECLARE_FUNCTION_NAME,
3080 FINAL_PRESCAN_INSN, PRINT_OPERAND_PUNCT_VALID_P,
3081 PRINT_OPERAND_ADDRESS): Adjust for Thumb-2.
3082 (arm_arch_notm, arm_arch_thumb2, arm_arch_hwdiv): New declarations.
3083 * config/arm/arm-cores.def: Add arm1156t2-s, cortex-a8, cortex-r4 and
3085 * config/arm/arm-tune.md: Regenerate.
3086 * config/arm/arm-protos.h: Update prototypes.
3087 * config/arm/vfp.md: Enable patterns for Thumb-2.
3088 (arm_movsi_vfp): Add movw alternative. Use output_move_vfp.
3089 (arm_movdi_vfp, movsf_vfp, movdf_vfp): Use output_move_vfp.
3090 (thumb2_movsi_vfp, thumb2_movdi_vfp, thumb2_movsf_vfp,
3091 thumb2_movdf_vfp, thumb2_movsfcc_vfp, thumb2_movdfcc_vfp): New.
3092 * config/arm/libunwind.S: Add Thumb-2 code.
3093 * config/arm/constraints.md: Update include Thumb-2.
3094 * config/arm/ieee754-sf.S: Add Thumb-2/Unified asm support.
3095 * config/arm/ieee754-df.S: Ditto.
3096 * config/arm/bpabi.S: Ditto.
3097 * config/arm/t-arm (MD_INCLUDES): Add thumb2.md.
3098 * config/arm/predicates.md (low_register_operand,
3099 low_reg_or_int_operand, thumb_16bit_operator): New.
3100 (thumb_cmp_operand, thumb_cmpneg_operand): Rename...
3101 (thumb1_cmp_operand, thumb1_cmpneg_operand): ... to this.
3102 * config/arm/t-arm-elf: Add armv7 multilib.
3103 * config/arm/arm.md: Update patterns for Thumb-2 and Unified asm.
3105 (UNSPEC_STACK_ALIGN, ce_count): New.
3106 (arm_incscc, arm_decscc, arm_umaxsi3, arm_uminsi3,
3107 arm_zero_extendsidi2, arm_zero_extendqidi2): New
3109 * config/arm/fpa.md: Update patterns for Thumb-2 and Unified asm.
3110 (thumb2_movsf_fpa, thumb2_movdf_fpa, thumb2_movxf_fpa,
3111 thumb2_movsfcc_fpa, thumb2_movdfcc_fpa): New insns.
3112 * config/arm/cirrus.md: Update patterns for Thumb-2 and Unified asm.
3113 (cirrus_thumb2_movdi, cirrus_thumb2_movsi_insn,
3114 thumb2_cirrus_movsf_hard_insn, thumb2_cirrus_movdf_hard_insn): New
3116 * doc/extend.texi: Document ARMv7-M interrupt functions.
3117 * doc/invoke.texi: Document Thumb-2 new cores+architectures.
3119 2007-01-03 Jakub Jelinek <jakub@redhat.com>
3121 * unwind-dw2.c (SIGNAL_FRAME_BIT, EXTENDED_CONTEXT_BIT): Define.
3122 (struct _Unwind_Context): Rename args_size to flags, remove
3123 signal_frame field, add a new args_size field and version field.
3124 (_Unwind_IsSignalFrame, _Unwind_SetSignalFrame,
3125 _Unwind_IsExtendedContext): New inline functions.
3126 (_Unwind_GetGR, _Unwind_SetGR, _Unwind_GetGRPtr, _Unwind_SetGRPtr):
3127 Assume by_value array is only present if _Unwind_IsExtendedContext.
3128 (_Unwind_GetIPInfo, execute_cfa_program, uw_frame_state_for): Use
3129 _Unwind_IsSignalFrame.
3130 (__frame_state_for): Initialize context.flags to EXTENDED_CONTEXT_BIT.
3131 (uw_update_context_1): Use _Unwind_SetSignalFrame.
3132 (uw_init_context_1): Initialize context->flags to
3133 EXTENDED_CONTEXT_BIT.
3134 * config/rs6000/linux-unwind.h (frob_update_context): Use
3135 _Unwind_SetSignalFrame.
3137 2007-01-03 Andrew Pinski <andrew_pinski@playstation.sony.com>
3140 * gimplify.c (gimplify_modify_expr_complex_part): Move below
3141 tree_to_gimple_tuple. Call tree_to_gimple_tuple when we need
3144 2007-01-03 Kazu Hirata <kazu@codesourcery.com>
3146 * config/i386/i386.h (NON_STACK_REG_P, REGNO_OK_FOR_SIREG_P,
3147 REGNO_OK_FOR_DIREG_P, REWRITE_ADDRESS, ASM_OPERAND_LETTER,
3148 RET, AT_SP): Remove.
3149 * config/i386/i386.md (*sse_prologue_save_insn): Use return
3152 * alias.c (init_alias_analysis): Use VEC_safe_grow_cleared.
3153 * cfgbuild.c (find_basic_blocks): Likewise.
3154 * cfgrtl.c (rtl_create_basic_block): Likewise.
3155 * function.c (temp_slots_at_level): Likewise.
3156 * reg-stack.c (stack_regs_mentioned): Likewise.
3157 * regclass.c (allocate_reg_info): Likewise.
3158 * tree-cfg.c (init_empty_tree_cfg, build_tree_cfg, create_bb,
3159 set_bb_for_stmt, move_block_to_fn): Likewise.
3160 * tree-complex.c (tree_lower_complex): Likewise.
3161 * vec.h (VEC_safe_grow_cleared): New.
3163 * cgraphunit.c, tree-ssa-alias.c: Fix comment typos.
3165 2007-01-03 Zdenek Dvorak <dvorakz@suse.cz>
3167 * loop-unswitch.c (unswitch_loop): Pass probabilities to loopify.
3168 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Pass probabilities
3170 * cfgloopmanip.c (scale_loop_frequencies): Export.
3171 (loopify): Scale the frequencies by prescribed coefficients.
3172 (set_zero_probability): New function.
3173 (duplicate_loop_to_header_edge): Improve updating of frequencies.
3174 (lv_adjust_loop_entry_edge, loop_version): Set probabilities
3175 and frequencies according to arguments.
3176 * tree-ssa-loop-manip.c (tree_unroll_loop): Set probabilities
3178 * cfg.c (scale_bbs_frequencies_int): Allow scaling the frequencies up.
3179 * modulo-sched.c (sms_schedule): Set probabilities for entering
3180 versioned loop correctly.
3181 * tree-vect-transform.c (vect_transform_loop): Ditto.
3182 * cfgloop.h (loopify, loop_version): Declaration changed.
3183 (scale_loop_frequencies): Declared.
3185 2007-01-02 Jan Hubicka <jh@suse.cz>
3187 * cgraph.c: Include tree-flow.h
3188 (cgraph_add_new-function): Handle IPA_SSA mode; execute
3190 * cgraph.h (enum cgraph_state): Add CGRAPH_STATE_IPA_SSA.
3191 * tree-pass.h (pass_all_early_optimizations): Declare.
3192 * cgraphunit.c (cgraph_process_new_functions): Add IPA_SSA; execute
3194 (cgraph_analyze_function): Do early_local_passes.
3195 * tree-mudflap.c (mf_decl_cache_locals, mf_build_check_statement_for):
3196 Do not add referenced vars.
3197 * tree-optimize.c (gate_all_optimizations): Do not execute when not in
3199 (gate_all_early_local_passes): New gate.
3200 (pass_early_local_passes): Use new gate.
3201 (execute_early_local_optimizations): New functions.
3202 (gate_all_early_optimizations): New gate.
3203 (pass_all_early_optimizations): New pass.
3204 (execute_free_datastructures): Free SSA only when initialized.
3205 (gate_init_datastructures): Init only when optimizing.
3206 (tree_lowering_passes): Do early local passes when called late.
3207 * tree-profile.c (do_tree_profiling): Don't profile functions added
3209 (do_early_tree_profiling, pass_early_tree_profile): Kill.
3210 * tree-cfg.c (update_modified_stmts): Do not update when operands are
3212 * passes.c (init_optimizations_passes): Reorder so we go into SSA
3213 during early_local_passes.
3214 * Makefile.in (cgraph.o): Add dependency on tree-flow.h.
3217 2007-01-02 Carlos O'Donell <carlos@codesourcery.com>
3219 * Makefile.in: Update copyright year.
3221 2007-01-02 Carlos O'Donell <carlos@codesourcery.com>
3223 * Makefile.in: Export GCC_EXEC_PREFIX before calling $(RUNTEST)
3224 in $(lang_checks) and check-consistency targets.
3226 2007-01-02 Jan Hubicka <jh@suse.cz>
3228 * tree-mudflap.c (mf_decl_cache_locals, mf_build_check_statement_for):
3229 Do not add referenced vars.
3230 * tree-cfg.c (update_modified_stmts): Do not update when SSA operands
3232 * passes.c (init_optimization_passes): Put mudflap_2 after
3233 free_datastructures.
3235 2007-01-02 Jan Hubicka <jh@suse.cz>
3237 * tree-optimize (execute_fixup_cfg): Set after_inlining flag.
3238 Set NOTHROW flag on call statements proved to be nothrow.
3239 Update statement of local calls so new pure/const functions are
3240 updated. Update_ssa when in ssa form. Mark PHI nodes of nonlocal
3242 (tree_rest_of_compilation): Register hooks and initialize bitmap
3243 early. Do not set after_inlining flag.
3245 2007-01-02 Steve Ellcey <sje@cup.hp.com>
3247 * sbitmap.c (HOST_BITS_PER_LONG_LONG): Change to
3248 HOST_BITS_PER_LONGLONG
3250 2007-01-02 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
3253 * c-typeck.c (store_init_value): Don't emit pedantic overflow
3254 warning for non-static initializers.
3256 2007-01-02 Steven Bosscher <steven@gcc.gnu.org>
3258 * config/alpha/alpha.md, arm/arm.c, darwin.c, frv/frv.md,
3259 m32r/m32r.c, m32r/m32r.c, mn10300/mn10300.md, pa/pa.c,
3260 rs6000/rs6000.c, s390/s390.md, sh/sh.md, sparc/sparc.c:
3261 Always use set_unique_reg_note to add REG_EQUAL notes.
3263 2007-01-02 Kazu Hirata <kazu@codesourcery.com>
3266 2007-01-02 Kazu Hirata <kazu@codesourcery.com>
3268 * alias.c (init_alias_analysis): Use VEC_safe_grow_cleared.
3269 * cfgbuild.c (find_basic_blocks): Likewise.
3270 * cfgrtl.c (rtl_create_basic_block): Likewise.
3271 * function.c (temp_slots_at_level): Likewise.
3272 * reg-stack.c (stack_regs_mentioned): Likewise.
3273 * regclass.c (allocate_reg_info): Likewise.
3274 * tree-cfg.c (init_empty_tree_cfg, build_tree_cfg, create_bb,
3275 set_bb_for_stmt, move_block_to_fn): Likewise.
3276 * tree-complex.c (tree_lower_complex): Likewise.
3277 * vec.h (VEC_safe_grow_cleared): New.
3279 2007-01-02 Ian Lance Taylor <iant@google.com>
3281 * c-common.c (c_common_truthvalue_conversion): When warning about
3282 using an assignment as a truth value, set TREE_NO_WARNING.
3284 2007-01-02 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
3287 * c.opt (Wold-style-declaration): New.
3288 * doc/invoke.texi (C-only Warning Options): New.
3289 (Wold-style-declaration): Document it.
3290 (Wextra): Enabled by -Wextra.
3291 * c-opts.c (c_common_post_options): Enabled by -Wextra.
3292 * c-decl.c (declspecs_add_scspec): Replace -Wextra with
3293 -Wold-style-declaration.
3295 2007-01-02 Kazu Hirata <kazu@codesourcery.com>
3297 * alias.c (init_alias_analysis): Use VEC_safe_grow_cleared.
3298 * cfgbuild.c (find_basic_blocks): Likewise.
3299 * cfgrtl.c (rtl_create_basic_block): Likewise.
3300 * function.c (temp_slots_at_level): Likewise.
3301 * reg-stack.c (stack_regs_mentioned): Likewise.
3302 * regclass.c (allocate_reg_info): Likewise.
3303 * tree-cfg.c (init_empty_tree_cfg, build_tree_cfg, create_bb,
3304 set_bb_for_stmt, move_block_to_fn): Likewise.
3305 * tree-complex.c (tree_lower_complex): Likewise.
3306 * vec.h (VEC_safe_grow_cleared): New.
3308 2007-01-02 Douglas Gregor <doug.gregor@gmail.com>
3310 * c-common.c (c_common_nodes_and_builtins): Since variants of
3311 void_type_node get built before it is given a name, we need to
3312 give those variants the name, too.
3313 (complete_array_type): We need to work with the canonical main
3314 type of the array, from which we will build the qualified version.
3315 * params.def (PARAM_VERIFY_CANONICAL_TYPES): New.
3316 * print-tree.c (print_node): Display canonical type information
3318 * stor-layout.c (layout_type): When we don't know the
3319 alignment of a type for which we're building an array, we end up
3320 guessing wrong, so make the type require structural equality.
3321 * tree.c (make_node_stat): When we build a new type, it is its
3323 (build_type_attribute_qual_variant): When building an attribute
3324 variant, its canonical type is the non-attribute variant. However,
3325 if the attributes are target-dependent and they differ, we need to
3326 use structural equality checks for this type.
3327 (build_qualified_type): A qualified type is not equivalent to its
3328 unqualified variant; set the canonical type appropriately.
3329 (build_distinct_type_copy): When building a distinct type from
3330 another type, the new type is its own canonical type.
3331 (build_variant_type_copy): When building a new type variant, we
3332 assume that it is equivalent to the original type.
3333 (build_pointer_type_for_mode): When building a pointer type, also
3334 build a canonical type pointer.
3335 (build_reference_type_for_mode): When building a reference type,
3336 also build a canonical type reference.
3337 (build_index_type): When we can't hash an index type (e.g.,
3338 because its maximum value is negative), the index type requires
3339 structural equality tests.
3340 (build_array_type): Build the canonical form of an array type.
3341 (build_function_type): Function types require structural equality,
3342 because they contain default arguments, attributes, etc.
3343 (build_method_type_directly): Ditto for method types.
3344 (build_offset_type): Build the canonical offset type.
3345 (build_complex_type): Build the canonical vector type.
3346 (make_vector_type): Build the canonical vector type.
3347 * tree.h (TYPE_CANONICAL): New.
3348 (TYPE_STRUCTURAL_EQUALITY_P): New.
3349 (SET_TYPE_STRUCTURAL_EQUALITY): New.
3350 (struct tree_type): Added "canonical" field.
3351 * params.h (VERIFY_CANONICAL_TYPES): New.
3352 * doc/c-tree.texi (TYPE_CANONICAL): Document.
3353 (TYPE_STRUCTURAL_EQUALITY_P): Document.
3354 (SET_TYPE_STRUCTURAL_EQUALITY): Document.
3355 * doc/invoke.texi (verify-canonical-types): Document --param
3356 parameter for verifying canonical types.
3358 2007-01-02 Joseph Myers <joseph@codesourcery.com>
3360 * config.gcc (powerpc-*-eabispe*, powerpc-*-eabisimaltivec*,
3361 powerpc-*-eabisim*, powerpc-*-eabialtivec*, powerpc-*-eabi*,
3362 powerpc-*-rtems*, powerpc-wrs-vxworks, powerpc-wrs-vxworksae,
3363 powerpcle-*-eabisim*, powerpcle-*-eabi*): Add rs6000/e500.h to
3365 * config/rs6000/e500.h: New.
3366 * config/rs6000/eabi.h (TARGET_SPE_ABI, TARGET_SPE, TARGET_E500,
3367 TARGET_ISEL, TARGET_FPRS, TARGET_E500_SINGLE, TARGET_E500_DOUBLE):
3369 * config/rs6000/linuxspe.h (TARGET_SPE_ABI, TARGET_SPE,
3370 TARGET_E500, TARGET_ISEL, TARGET_FPRS, TARGET_E500_SINGLE,
3371 TARGET_E500_DOUBLE): Remove.
3372 * config/rs6000/vxworks.h (TARGET_SPE_ABI, TARGET_SPE,
3373 TARGET_E500, TARGET_ISEL, TARGET_FPRS): Remove.
3374 * config/rs6000/rs6000.h (CHECK_E500_OPTIONS): Define.
3375 * config/rs6000/rs6000.c (rs6000_override_options): Use
3378 2007-01-02 Joseph Myers <joseph@codesourcery.com>
3380 * config/rs6000/rs6000.c (print_operand): Check (TARGET_SPE ||
3381 TARGET_E500_DOUBLE), not TARGET_E500, for %y.
3382 (rs6000_generate_compare, rs6000_emit_sCOND, output_cbranch,
3383 rs6000_emit_cmove): Don't check TARGET_E500.
3384 * config/rs6000/rs6000.md (bunordered, bordered, sunordered,
3385 sordered): Don't check TARGET_E500.
3387 2007-01-01 Eric Christopher <echristo@apple.com>
3389 * config/mips/mips.c (mips_regno_mode_ok_for_base_p): Use
3390 HARD_REGISTER_NUM_P.
3392 2007-01-01 Roger Sayle <roger@eyesopen.com>
3394 * fold-const.c (fold_binary) <EQ_EXPR>: Fold "(X^C1) eq/ne C2" into
3395 "X eq/ne (C1^C2)". Fold "(X^Z) eq/ne (Y^Z)" as "X eq/ne Y" when Z
3396 has no side-effects. Fold "(X^C1) eq/ne (Y^C2)" as "(X^(C1^C2))
3399 2007-01-01 Mike Stump <mrs@apple.com>
3401 * configure.ac: Remove support for building with Apple's gcc-3.1.
3403 2007-01-02 Joseph Myers <joseph@codesourcery.com>
3406 * caller-save.c (add_stored_regs): Only handle SUBREGs if inner
3407 REG is a hard register. Do not modify REG before calling
3409 * rtlanal.c (subreg_get_info): Don't assert size of XMODE is a
3410 multiple of the size of YMODE for certain lowpart cases.
3412 2007-01-01 Andrew Pinski <pinskia@gmail.com>
3415 * gimplify (voidify_wrapper_expr): Update for
3418 2007-01-01 Andreas Schwab <schwab@suse.de>
3421 * config/ia64/ia64.c (ia64_compute_frame_size): Account space for
3422 save of BR0 in extra_spill_size instead of spill_size.
3423 (ia64_expand_prologue): Save BR0 outside of the gr/br/fr spill
3425 (ia64_expand_epilogue): Restore BR0 from its new location.
3427 2007-01-01 Andrew Pinski <pinskia@gmail.com>
3429 * gimplify.c (gimplify_init_constructor <case VECTOR_TYPE>):
3430 Use a temporary variable if the left hand side is not a gimple
3433 2007-01-01 Andrew Pinski <pinskia@gmail.com>
3435 * gimplify.c (gimplify_return_expr): Make the temporary variable
3436 for the return expression, a gimple register variable.
3438 2007-01-01 Jan Hubicka <jh@suse.cz>
3440 * emit-rtl.c (emit_copy_of_insn_after): Do not call copy_insn_1 for
3443 2007-01-01 Mike Stump <mrs@apple.com>
3445 * configure.ac (HAVE_GAS_LITERAL16): Add autoconf check for
3447 * config/darwin.c (machopic_select_rtx_section): Use
3449 (darwin_mergeable_constant_section): Likewise.
3450 * configure: Regenerate.
3451 * config.in: Regenerate.
3453 2007-01-01 Jan Hubicka <jh@suse.cz>
3454 Andrew Pinski <pinskia@gmail.com>
3456 * cgraphunit.c (cgraph_optimize): Call cgraph_add_new_functions
3457 before starting IPA passes.