1 2007-01-29 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
3 * tree-optimize.c (update_inlined_to_pointers): Delete unused
6 2007-01-29 Janis Johnson <janis187@us.ibm.com>
8 * Makefile.in (USER_H): Remove decfloat.h.
9 * ginclude/decfloat.h: Delete, moving contents to ...
10 * ginclude/float.h: Add support for decimal floating point,
11 guarded by __STDC_WANT_DEC_FP__.
13 2007-01-29 Mike Stump <mrs@apple.com>
15 * doc/gccint.texi (Top): Rename Loop Representation to Loop
16 Analysis and Representation to resolve case insensitive conflict.
17 * doc/loop.texi (Loop Analysis and Representation): Likewise.
19 2007-01-28 Daniel Berlin <dberlin@dberlin.org>
21 * tree.h (struct tree_memory_tag): Add aliases member.
22 (MTAG_ALIASES): New macro.
23 * tree-ssa-alias.c (alias_bitmap_obstack): New variable.
24 (add_may_alias): Remove pointer-set. Update for may_aliases being
26 (mark_aliases_call_clobbered): Update for may_aliases being a
28 (compute_tag_properties): Ditto.
29 (create_partition_for): Ditto.
30 (compute_memory_partitions): Ditto.
31 (dump_may_aliases_for): Ditto.
32 (is_aliased_with): Ditto.
33 (add_may_alias_for_new_tag): Ditto.
34 (rewrite_alias_set_for): Rewrite for may_aliases being a bitmap.
35 (compute_is_aliased): New function.
36 (compute_may_aliases): Call compute_is_aliased).
37 (init_alias_info): Initialize alias_bitmap_obstack.
38 (union_alias_set_into): New function.
39 (compute_flow_sensitive_aliasing): Use union_aliases_into.
40 (have_common_aliases_p): Rewrite to take two bitmaps and use
42 (compute_flow_insensitive_aliasing): Stop using pointer-sets.
44 (finalize_ref_all_pointers): Update for add_may_alias changes.
45 (new_type_alias): Ditto.
46 * tree-flow-inline.h (may_aliases): Return a bitmap.
47 * tree-dfa.c (dump_variable): Check for MTAG_P'ness.
48 * tree-ssa.c (verify_flow_insensitive_alias_info): Update for
49 may_aliases being a bitmap.
50 * tree-flow.h (struct var_ann_d): Remove may_aliases member.
51 may_aliases now returns a bitmap.
52 * tree-ssa-structalias.c (merge_smts_into): Update for may_aliases
54 * tree-ssa-operands.c (add_virtual_operand): Update for
55 may_aliases being a bitmap.
57 2007-01-29 Daniel Berlin <dberlin@dberlin.org>
59 PR tree-optimization/30630
60 * tree-ssa-structalias.c (do_complex_constraint): Mark correct
63 2007-01-29 Simon Martin <simartin@users.sourceforge.net>
66 * gimplify.c (gimplify_target_expr): Make sure that the TARGET_EXPR is
67 expanded only once even if an error occurs.
69 2007-01-29 Ben Elliston <bje@au.ibm.com>
71 * gcov-io.h (__gcov_indirect_call_profiler): Declare.
72 (__gcov_average_profiler): Likewise.
73 (__gcov_ior_profiler): Likewise.
74 (__gcov_merge_ior): Likewise.
76 2007-01-28 Jan Hubicka <jh@suse.cz>
78 * builtins.c (expand_builtin_memset): Fix typo in my last patch.
79 * value-prof.c (stringop_block_profile): Likewise.
81 2007-01-28 Jan Hubicka <jh@suse.cz>
83 * expr.c (emit_block_move_via_movmem, emit_block_move_via_libcall): Add
84 variant handling histograms; add wrapper.
85 (clear_storage_via_libcall): Export.
86 (emit_block_move_hints): Break out from ...; add histograms.
87 (emit_block_move): ... this one.
88 (clear_storage_hints): Break out from ...; add histograms.
89 (clear_storage): ... this one.
90 (set_storage_via_memset): Handle histogram.
91 * expr.h (emit_block_move_via_libcall, emit_block_move_hints): Declare.
92 (clear_storage_hints, clear_storage_via_libcall): Declare.
93 (set_storage_via_setmem): Update prototype.
94 * doc/md.texi (movmem, setmem): Document new arguments.
96 * value-prof.c (dump_histogram_value, tree_find_values_to_profile): Add
98 (stringop_block_profile): New global function.
99 (tree_stringops_values_to_profile): Profile block size and alignment.
100 * value-prof.h (enum hist_type): add HIST_TYPE_AVERAGE and
102 (struct profile_hooks): Add gen_average_profiler and gen_ior_profiler.
103 (stringop_block_profile): Declare.
104 * builtins.c: Include value-prof.h.
105 (expand_builtin_memcpy, expand_builtin_memset): Pass block profile.
106 * gcov-ui.h (GCOV_COUNTER_NAMES): Add new counter.
107 (GCOV_COUNTER_AVERAGE, GCOV_COUNTER_IOR): New constants.
108 (GCOV_COUNTERS, GCOV_LAST_VALUE_COUNTER): Update.
109 * profile.c (instrument_values): Add new counters.
110 * cfgexpand.c (expand_gimple_basic_block): Propagate histograms to
112 * tree-profile.c (tree_average_profiler_fn, tree_ior_profiler_fn): New.
113 (tree_init_edge_profiler): Build new profilers.
114 (tree_gen_average_profiler, tree_gen_ior_profiler): New.
115 (pass_tree_profile): Add dump.
116 (tree_profile_hooks): Update.
117 * Makefile.in (LIBGCOV): Add new constants.
118 * libgcov.c (__gcov_merge_ior, __gcov_average_profiler,
119 __gcov_ior_profiler): New.
120 * i386.md (movmem/setmem expanders): Add new optional arguments.
122 2007-01-28 David Edelsohn <edelsohn@gnu.org>
124 * doc/md.texi (Standard Pattern Names): Document blockage pattern.
126 2007-01-28 Zdenek Dvorak <dvorakz@suse.cz>
128 * tree-ssa-loop-unswitch.c: Include tree-inline.h.
129 (tree_unswitch_single_loop): Pass eni_size_weights to
131 * tree-ssa-loop-manip.c: Include tree-inline.h.
132 (can_unroll_loop_p): Pass eni_size_weights to
134 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p):
135 Pass eni_size_weights to estimate_num_insns.
136 * tree.h (init_inline_once): Export.
137 * toplev.c (backend_init): Call init_inline_once.
138 * cgraphunit.c (cgraph_process_new_functions,
139 cgraph_analyze_function): Pass eni_inlining_weights to
141 * ipa-inline.c (compute_inline_parameters): Ditto.
142 * tree-ssa-loop-ivcanon.c (tree_num_loop_insns): Pass weights
143 to estimate_num_insns.
144 (try_unroll_loop_completely): Pass eni_size_weights to
146 * tree-eh.c (decide_copy_try_finally): Pass eni_size_weights
147 ot estimate_num_insns.
148 * tree-ssa-loop-prefetch.c: Include tree-inline.h.
149 (loop_prefetch_arrays): Pass eni_time_weights to tree_num_loop_insns.
150 * tree-inline.c (eni_inlining_weights, eni_size_weights,
151 eni_time_weights): New variables.
152 (init_inline_once): Initialize them.
153 (struct eni_data): Mew.
154 (estimate_num_insns_1, estimate_num_insns): Use weights.
155 * tree-inline.h (struct eni_weights_d): New.
156 (eni_inlining_weights, eni_size_weights, eni_time_weights): Declare.
157 (estimate_num_insns): Declaration changed.
158 * cfgloop.h (tree_num_loop_insns): Declaration changed.
159 * Makefile.in (tree-ssa-loop-unswitch.o, tree-ssa-loop-prefetch.o,
160 tree-ssa-loop-manip.o): Add TREE_INLINE_H dependency.
162 2007-01-28 Zdenek Dvorak <dvorakz@suse.cz>
164 * tree-data-ref.c (conflict_fn): Assert that the number of affine
165 relations in the conflict function is valid.
167 2007-01-27 Ian Lance Taylor <iant@google.com>
169 * common.opt: Add fstrict-overflow.
170 * opts.c (decode_options): Set flag_strict_overflow if -O2.
171 * flags.h (TYPE_OVERFLOW_WRAPS): Define.
172 (TYPE_OVERFLOW_UNDEFINED): Define.
173 (TYPE_OVERFLOW_TRAPS): Define. This replaces TYPE_TRAP_SIGNED.
175 * tree.h (TYPE_TRAP_SIGNED): Don't define.
176 * fold-const.c (negate_expr_p): Use TYPE_OVERFLOW_UNDEFINED.
177 (fold_negate_expr): Likewise.
178 (make_range): Likewise.
179 (extract_muldiv_1): Likewise.
180 (maybe_canonicalize_comparison): Likewise.
181 (fold_comparison): Likewise.
182 (fold_binary): Likewise.
183 (tree_expr_nonnegative_p): Likewise.
184 (tree_expr_nonzero_p): Likewise.
185 * tree-vrp.c (compare_values): Likewise.
186 (extract_range_from_binary_expr): Likewise.
187 (extract_range_from_unary_expr): Likewise.
188 * tree-ssa-loop-niter.c (infer_loop_bounds_from_signedness):
190 (nowrap_type_p): Likewise.
191 * tree-scalar-evolution.c (simple_iv): Likewise.
192 * fold-const.c (negate_expr_p): Use TYPE_OVERFLOW_WRAPS.
193 (build_range_check): Likewise.
194 (extract_muldiv_1): Likewise.
195 (fold_comparison): Likewise.
196 * tree-vrp.c (vrp_int_const_binop): Likewise.
197 (extract_range_from_unary_expr): Likewise.
198 * convert.c (convert_to_integer): Likewise.
199 * fold-const.c (fold_negate_expr): Use TYPE_OVERFLOW_TRAPS.
200 (fold_comparison): Likewise.
201 (fold_binary): Likewise.
202 * optabs.c (optab_for_tree_code): Likewise.
203 * tree-vectorizer.c (vect_is_simple_reduction): Likewise.
204 * simplify-rtx.c (simplify_const_relational_operation): Check
205 flag_strict_overflow and flag_trapv.
206 (simplify_const_relational_operation): Likewise.
207 * doc/invoke.texi (Option Summary): Mention -fstrict-overflow.
208 (Optimize Options): Add -fstrict-overflow to -O2 list. Document
211 2007-01-27 Roger Sayle <roger@eyesopen.com>
213 * tree.c (tree_fold_gcd): Delete.
214 * tree.h (tree_fold_gcd): Remove prototype.
215 * tree-data-ref.c (tree_fold_divides_p): Don't use tree_fold_gcd to
216 test whether one constant integer is a multiple of another. Instead
217 call int_const_binop with TRUNC_MOD_EXPR and test for a zero result.
218 * fold-const.c (multiple_of_p): We've determined both TOP and
219 BOTTOM are integer constants so we can call int_const_binop directly
220 instead of the more generic const_binop.
222 2007-01-27 Roger Sayle <roger@eyesopen.com>
224 * fold-const.c (size_binop): In the fast-paths for X+0, 0+X, X-0 and
225 1*X check that the constant hasn't overflowed, to preserve the
227 (round_up): Provide an efficient implementation when rouding-up an
228 INTEGER_CST to a power-of-two.
230 2007-01-28 Ralf Wildenhues <Ralf.Wildenhues@gmx.de>
232 * doc/sourcebuild.texi: Add comma for clarity.
233 * doc/extend.texi: Fix some typos.
234 * doc/passes.texi: Likewise.
235 * doc/cppinternals.texi: Likewise.
236 * doc/c-tree.texi: Likewise.
237 * doc/tree-ssa.texi: Likewise.
238 * doc/install.texi: Likewise.
240 2007-01-27 Jan Hubicka <jh@suse.cz>
242 * tree-sra.c (sra_walk_function): Don't rely on aliases being build.
243 (pass_sra): Do not require alias information.
244 * passes.c (init_optimization_passes): Add SRA
246 2007-01-27 Steven Bosscher <steven@gcc.gnu.org>
248 * tracer.c (rest_of_handle_tracer): We already cleaned
249 up the CFG in tracer() so don't do it here again.
250 * cfgcleanup.c (rest_of_handle_jump2): Don't repeat
251 cleanup_cfg here, either. And don't call renumber_insns.
253 * cfgrtl.c (rtl_verify_flow_info_1): Don't verify that BB_END
254 and BB_HEAD are in the insn stream here. Instead make sure
255 that BB_INSN is valid on all insns. Also, do check here that
256 there are no pending branch predictions...
257 (rtl_verify_flow_info): ...instead of doing it here. Checks
258 for BB_END and BB_HEAD moved from rtl_verify_flow_info_1 to
261 2007-01-26 Roger Sayle <roger@eyesopen.com>
263 * config/i386/i386.c (ix86_swap_binary_operands_p): New helper
264 function to simplify/factorize operand order canonicalization.
265 (ix86_fixup_binary_operands): Reorganize using the above function.
266 (ix86_binary_operator_ok): Likewise.
268 2007-01-27 Jakub Jelinek <jakub@redhat.com>
270 * genattrtab.c (struct attr_value_list, insn_code_values): Move to
271 file scope from optimize_attrs.
272 (simplify_test_exp): If insn_code_values is not NULL, use it to speed
274 (optimize_attrs): Clear insn_code_values after freeing it.
276 2007-01-26 Zdenek Dvorak <dvorakz@suse.cz>
278 * tree-ssa-address.c (create_mem_ref): Remove ", bsi" from
279 a parts.base assignment.
281 2007-01-26 Zdenek Dvorak <dvorakz@suse.cz>
283 * tree-data-ref.c (dump_subscript): Use dump_conflict_function.
284 (compute_subscript_distance, initialize_data_dependence_relation,
285 finalize_ddr_dependent, analyze_ziv_subscript,
286 analyze_siv_subscript_cst_affine,
287 compute_overlap_steps_for_affine_univar,
288 compute_overlap_steps_for_affine_1_2, analyze_subscript_affine_affine,
289 analyze_siv_subscript, analyze_miv_subscript,
290 analyze_overlapping_iterations, subscript_dependence_tester_1,
291 compute_self_dependence, free_dependence_relation): Work
292 with affine_fn instead of chrecs.
293 (dump_affine_function, dump_conflict_function, affine_function_equal_p,
294 common_affine_function, affine_function_base,
295 affine_function_constant_p, affine_fn_op, affine_fn_plus,
296 affine_fn_minus, affine_fn_free, conflict_fn_not_known,
297 conflict_fn_no_dependence, free_conflict_function, free_subscripts,
298 conflict_fn, affine_fn_cst, affine_fn_univar): New functions.
299 (all_chrecs_equal_p): Removed.
300 * tree-data-ref.h (affine_fn, conflict_function): New types.
301 (struct subscript): Change type of conflicting_iterations_in_a
302 and conflicting_iterations_in_b.
304 2007-01-26 Steve Ellcey <sje@cup.hp.com>
307 * config/pa/pa.h (TARGET_HPUX_11): New.
308 * config/pa/pa-hpux11.h (TARGET_HPUX_11): New.
309 * config/pa/pa.c (pa_init_builtins): Use TARGET_HPUX_11.
311 2007-01-26 Daniel Berlin <dberlin@dberlin.org>
312 Richard Guenther <rguenther@suse.de>
314 * tree-ssa-structalias.c (solve_graph): Handle case
315 we merged the variable to another.
317 2007-01-25 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
319 * builtins.c (fold_builtin_1): Treat ccos and ccosh as 'even'
322 * fold-const.c (negate_mathfn_p): Treat casin, casinh, catan,
323 catanh, cproj, csin, csinh, ctan and ctanh as 'odd' functions.
325 2007-01-25 DJ Delorie <dj@redhat.com>
327 * config/m32c/m32c.c (m32c_cannot_change_mode_class): We don't
328 allow changes to modes which don't fit in those registers.
330 * reload1.c (choose_reload_regs): Check for invalid subregs before
331 computing their locations, not after.
333 2007-01-25 Geoffrey Keating <geoffk@apple.com>
336 * config/rs6000/rs6000.c (first_altivec_reg_to_save): On Darwin,
337 save Altivec registers in an eh_return function.
338 (compute_vrsave_mask): Likewise.
339 (rs6000_stack_info): Correct AIX/Darwin stack alignment computation
340 for saving Altivec registers.
341 (rs6000_emit_prologue): Don't allocate stack twice in
342 eh_return function. Correct expected value of altivec_save_offset
343 when using save_world. Describe save of R0 to stack when using
344 save_world. Describe stack pointer adjustment when using
345 save_world. Remove duplicated eh_return parameter register saving.
346 Update sp_offset variable after save_world.
347 * config/rs6000/t-darwin (LIB2FUNCS_STATIC_EXTRA): Remove
349 (LIB2FUNCS_EXTRA): Add darwin-world.asm.
350 * config/rs6000/darwin.h (SUBTARGET_OVERRIDE_OPTIONS): -m64
353 2007-01-25 Steve Ellcey <sje@cup.hp.com>
355 * config.gcc (ia64*-*-hpux*): Make posix threads the default.
357 2007-01-25 Steve Ellcey <sje@cup.hp.com>
360 * config/pa/pa.c (pa_init_builtins): Set asm names for finite routines.
361 * config/ia64/ia64.c (ia64_init_builtins): Ditto.
363 2007-01-25 Richard Guenther <rguenther@suse.de>
365 * doc/invoke.texi (-Wcoverage-mismatch): Document.
366 * common.opt (-Wcoverage-mismatch): New warning option.
367 * coverage.c (get_coverage_counts): Ignore coverage mismatch
368 if -Wcoverage-mismatch is given.
370 2007-01-25 Razya Ladelsky <razya@il.ibm.com>
372 * ipa-cp.c (ipcp_insert_stage, ipcp_driver): Support for SSA.
373 (ipcp_driver): Change to static definition.
374 Add dumping of the ifunctions.
375 (constant_val_insert): Remove unused parameter. Support for SSA.
376 (ipcp_propagate_const): Support for SSA.
377 (ipcp_profile_bb_print): Print only analyzed nodes.
378 (ipcp_replace_map_create): Remove support for Fortran constant
380 * ipa-prop.c (ipa_method_modify_stmt,
381 ipa_callsite_compute_param): Support for SSA.
382 * ipa-prop.h (ipcp_driver): Remove declaration.
383 (IS_VALID_TREE_MAP_INDEX): Add define.
385 2007-01-24 Geoffrey Keating <geoffk@apple.com>
387 * unwind-dw2.c (execute_stack_op): Handle DW_OP_swap.
389 2007-01-24 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
392 * builtins.c (fold_builtin_cabs): Use MPFR to evaluate a
393 constant argument to cabs and do it without checking for
394 -funsafe-math-optimizations.
396 2007-01-24 Douglas Gregor <dgregor@osl.iu.edu>
398 * c-common.h (RID_FIRST_CXX0X): New.
399 (RID_LAST_CXX0X): New.
400 * c-opts.c (c_common_handle_option): -Wc++0x-compat is triggered
402 * c.opt (Wc++0x-compat): New.
403 * doc/invoke.texi (-Wc++0x-compat): Document.
405 2007-01-24 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
407 * builtins.c (fold_builtin_carg): New.
408 (fold_builtin_1): Use it.
410 2007-01-24 Jan Hubicka <jh@suse.cz>
412 * ipa-inline.c (cgraph_decide_inlining): Initialize initial_insns.
414 * ipa-inline.c (initial_insns, max_insns): Delete.
415 (compute_max_insns): New function.
416 (cgraph_decide_inlining_of_small_function): Use it; take minimal amount
417 of insns as base for code growth.
418 (cgraph_decide_inlining): Make initial_insns local; do not compute
420 * params.def (PARAM_INLINE_UNIT_GROWTH): Set to 60.
421 * doc/invoke.texi (inline-unit-growth): Update docs.
423 2007-01-24 Jakub Jelinek <jakub@redhat.com>
425 * config/i386/i386.h (x86_cmpxchg16b): Remove const.
426 (TARGET_CMPXCHG16B): Define to x86_cmpxchg16b.
427 * config/i386/i386.c (x86_cmpxchg16b): Remove const.
428 (override_options): Add PTA_CX16 flag. Set x86_cmpxchg16b
429 for CPUs that have PTA_CX16 set.
432 * gimplify.c (omp_check_private): New function.
433 (gimplify_scan_omp_clauses): Use it for
434 firstprivate/lastprivate/reduction.
437 * gimplify.c (omp_add_variable): Don't call omp_notice_variable
438 on TYPE_SIZE_UNIT for GOVD_LOCAL VLAs.
441 * omp-low.c (lower_omp_for_lastprivate): Add dlist argument.
442 If lower_lastprivate_clauses emits some statements, append them
443 to dlist rather than body_p and to body_p append an initializer.
444 (lower_omp_for): Adjust caller.
446 2007-01-24 Steve Ellcey <sje@cup.hp.com>
448 * target.h (globalize_decl_name): New.
449 * target-def.h (TARGET_ASM_GLOBALIZE_DECL_NAME): New.
450 * output.h (default_globalize_decl_name): New.
451 * varasm.c (asm_output_bss): Use globalize_decl_name instead of
453 (globalize_decl): Ditto.
454 (default_globalize_decl_name): New.
455 * config/ia64/ia64.c (ia64_globalize_decl_name): New.
456 (ia64_handle_version_id_attribute): New.
457 (TARGET_ASM_GLOBALIZE_DECL_NAME): New.
458 (ia64_asm_output_external): Use globalize_decl_name instead
460 * doc/extend.texi (version_id): New pragma.
461 * doc/tm.texi (ARGET_ASM_GLOBALIZE_DECL_NAME): New target hook.
463 2007-01-24 Andreas Krebbel <krebbel1@de.ibm.com>
465 * unwind-dw2-fde.c (get_cie_encoding): Replaced _Unwind_Word with
466 _uleb128_t and _Unwind_SWord with _sleb128_t.
467 * unwind-dw2.c (extract_cie_info, execute_stack_op, execute_cfa_program,
468 uw_frame_state_for, uw_update_context_1): Likewise.
469 * unwind-c.c (parse_lsda_header, PERSONALITY_FUNCTION): Likewise.
470 * unwind-pe.h (read_uleb128, read_sleb128,
471 read_encoded_value_with_base): Likewise.
472 * unwind-generic.h: Define _sleb128_t and _uleb128_t types.
474 2007-01-24 Richard Guenther <rguenther@suse.de>
476 * builtins.c (expand_builtin_cexpi): Get the fndecl
477 for cexp in the correct way.
479 2007-01-24 Jan Hubicka <jh@suse.cz>
481 * tree-ssa-dce.c (eliminate_unnecesary_stmts): Remove dead LHS of calls.
483 2007-01-24 Andreas Krebbel <krebbel1@de.ibm.com>
485 * c-cppbuiltin.c (builtin_define_type_sizeof): New function.
486 (c_cpp_builtins): New builtin macros: __SIZEOF_INT__, __SIZEOF_LONG__,
487 __SIZEOF_LONG_LONG__, __SIZEOF_SHORT__, __SIZEOF_POINTER__,
488 __SIZEOF_FLOAT__, __SIZEOF_DOUBLE__, __SIZEOF_LONG_DOUBLE__,
489 __SIZEOF_SIZE_T__, __SIZEOF_WCHAR_T__, __SIZEOF_WINT_T__ and
490 __SIZEOF_PTRDIFF_T__.
491 * doc/cpp.texi: Documentation for the new builtin macros added.
493 2007-01-24 Uros Bizjak <ubizjak@gmail.com>
495 * config/i386/i386.md (tanxf2, tan<mode>2, atan<mode>2, log<mode>2,
496 log10<mode>2, log2<mode>2, expxf2, exp10xf2, exp2xf2): Use op2
497 instead of operands[2] to avoid access past the end of array.
499 2007-01-24 Richard Sandiford <richard@codesourcery.com>
501 * reload1.c (emit_reload_insns): Pass the reload register
502 for a non-spill output reload through forget_old_reloads_1.
504 2007-01-23 Joseph Myers <joseph@codesourcery.com>
506 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Define
507 _SOFT_DOUBLE if doubles use software floating-point.
508 * config/rs6000/libgcc-ppc-glibc.ver: Export additional long
509 double functions if _SOFT_DOUBLE, not _SOFT_FLOAT.
510 * config/rs6000/darwin-ldouble.c: Also compile functions for
511 hard-float without FPRs. Use fmsub function for all __NO_FPRS__
512 cases. Compile extra functions if _SOFT_DOUBLE, not _SOFT_FLOAT.
513 * config/rs6000/linuxspe.h (SUBSUBTARGET_OVERRIDE_OPTIONS): Remove
514 commented-out long double override.
515 (CPP_LONGDOUBLE_DEFAULT_SPEC): Likewise.
516 * config/rs6000/eabispe.h: Likewise.
517 * config/rs6000/rs6000.c (rs6000_override_options): Don't override
518 long double for non-SPE.
519 (rs6000_handle_option): Likewise.
520 (invalid_e500_subreg): Disallow more subregs involding DImode,
521 DFmode, TImode or TFmode.
522 (rs6000_legitimate_offset_address_p): Check TFmode offsets for
524 (legitimate_lo_sum_address_p): Also check for TFmode for E500
526 (rs6000_legitimize_address): Also handle TFmode for E500 double.
527 (rs6000_legitimize_reload_address): Also handle TFmode for E500
529 (rs6000_legitimate_address): Also check for TFmode for E500
531 (rs6000_emit_move): Use DFmode subregs of TFmode for E500 double.
532 (spe_build_register_parallel): Handle TFmode and TCmode.
533 (rs6000_spe_function_arg): Handle TFmode and TCmode for E500
535 (function_arg): Handle TFmode and TCmode for E500 double.
536 (rs6000_init_libfuncs): Initialize extra libfuncs for soft double
538 (print_operand): Handle TFmode and TImode for %y.
539 (rs6000_generate_compare): Handle TFmode comparisons for E500
541 (spe_func_has_64bit_regs_p): Check for TFmode for E500 double.
542 (rs6000_function_value): Handle TFmode and TCmode for E500 double.
543 (rs6000_libcall_value): Handle TFmode and TCmode for E500 double.
544 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Check for
545 TFmode for E500 double.
546 * config/rs6000/rs6000.md (FP): Allow TF for E500 double.
547 (floatsidf2): Enable for E500 double.
548 (movtf_softfloat): Use rs6000_nonimmediate_operand.
549 (extenddftf2): Change to extenddftf2_fprs.
550 (extenddftf2): Call gen_spe_extenddftf2 or gen_extenddftf2_fprs
551 depending on TARGET_E500_DOUBLE.
552 (extendsftf2): Enable for E500 double.
553 (trunctfdf2): Enable for E500 double.
554 (trunctfsf2): Change to trunctfsf2_fprs.
555 (trunctfsf2): Call gen_spe_trunctfsf2 or gen_trunctfsf2_fprs
556 depending on TARGET_E500_DOUBLE.
557 (floatsitf2): Enable for E500 double.
558 (fix_trunctfsi2): Change to fix_trunctfsi2_fprs.
559 (fix_trunctfsi2): Call gen_spe_fix_trunctfsi2 or
560 gen_fix_trunctfsi2_fprs depending on TARGET_E500_DOUBLE.
561 (negtf2): Change to negtf2_internal.
562 (negtf2): New expander.
563 (abstf2): Enable for E500 double. Call gen_spe_abstf2_tst,
564 gen_spe_abstf2_cmp or gen_abstf2_internal depending on
565 TARGET_E500_DOUBLE and flag_unsafe_math_optimizations.
566 (movdi_internal32): Use rs6000_nonimmediate_operand.
567 (unnamed splitter): Likewise.
568 * config/rs6000/spe.md (CMPTFEQ_GPR, TSTTFEQ_GPR, CMPTFGT_GPR,
569 TSTTFGT_GPR, CMPTFLT_GPR, TSTTFLT_GPR): New unspecs.
570 (SPE64TF, DITI): New mode macros.
571 (frob_df_di): Change to frob_<SPE64:mode>_<DITI:mode>; allow more
574 (frob_<mode>_di_2): New.
575 (frob_tf_di_8_2): New.
576 (frob_di_df): Change to frob_di_<mode>; allow more modes.
578 (frob_di_df_2): Change to frob_<DITI:mode>_<SPE64:mode>_2; allow
580 (frob_ti_<mode>_8_2): New.
582 (mov_si<mode>_e500_subreg0, mov_si<mode>_e500_subreg0_2,
583 mov_si<mode>_e500_subreg4, mov_si<mode>_e500_subreg4_2): Allow
585 (mov_sitf_e500_subreg8, mov_sitf_e500_subreg8_2,
586 mov_sitf_e500_subreg12, mov_sitf_e500_subreg12_2): New.
587 (spe_trunctfdf2_internal1, spe_trunctfsf2, spe_extenddftf2,
588 spe_fix_trunctfsi2, spe_fix_trunctfsi2_internal,
589 spe_negtf2_internal, spe_abstf2_cmp, spe_abstf2_tst): New.
590 (cmptfeq_gpr, tsttfeq_gpr, cmptfgt_gpr, tsttfgt_gpr, cmptflt_gpr,
593 2007-01-23 Ian Lance Taylor <iant@google.com>
595 * Makefile.in (OBJS-common): Reformat, alphabetize, but put
597 (OBJS-archive): Reformat, alphabetize.
598 (OBJS): Change out_object_file to OBJS-md.
600 2007-01-23 Uros Bizjak <ubizjak@gmail.com>
602 * config/i386/i386.md: Use REG_P, MEM_P, CONST_INT_P, LABEL_P,
603 JUMP_P and CALL_P predicates where applicable.
604 * config/i386/i386.c: Ditto.
605 * config/i386/i386.md: Ditto.
606 * config/i386/mmx.md: Ditto.
607 * config/i386/predicates.md: Ditto.
609 2007-01-22 Andreas Schwab <schwab@suse.de>
611 * config/m68k/m68k.h: Fix comment.
613 2007-01-22 Jan Hubicka <jh@suse.cz>
615 * passes.c (init_optimization_passes): Do not rerun
616 pass_early_warn_uninitialized.
618 2007-01-22 Richard Guenther <rguenther@suse.de>
620 PR tree-optimization/30038
621 * tree-ssa-math-opts.c (maybe_record_sincos): New static helper
623 (execute_cse_sincos_1): Likewise.
624 (execute_cse_sincos): Likewise.
625 (gate_cse_sincos): Likewise.
626 (pass_cse_sincos): New pass CSEing sin() and cos() calls using
627 the cexpi() canonicalization of sincos().
628 * tree-pass.h (pass_cse_sincos): Declare.
629 * passes.c (init_optimization_passes): New pass pas_cse_sincos.
631 2007-01-21 Eric Botcazou <ebotcazou@libertysurf.fr>
633 PR rtl-optimization/29329
634 * combine.c (replaced_rhs_insn): Rename to i2mod.
635 (replaced_rhs_value): Rename to i2mod_new_rhs.
636 (i2mod_old_rhs): New global variable.
637 (combine_instructions): Adjust for above change. Save a copy of
638 the old RHS into i2mod_old_rhs when the contents of a REG_EQUAL
639 note are substituted in the second instruction.
640 (distribute_notes) <REG_DEAD>: Adjust for above change. Do not
641 ditch the note if it pertains to the second eliminated register
642 and this register is mentioned in i2mod_old_rhs.
645 2006-09-12 Eric Botcazou <ebotcazou@libertysurf.fr>
647 * combine.c (distribute_notes) <REG_DEAD>: Do not consider SETs past
648 the insn to which the note was originally attached.
650 2007-01-21 Jan Hubicka <jh@suse.cz>
652 * ipa-inline.c (inlining_mode): Comment, move up.
653 (cgraph_decide_inlining_incrementally): Do not perform inlining
654 itself; fix handling of flattening of self recursive functions.
655 (cgraph_find_cycles): Remove.
656 (cgraph_flatten_node): Remove.
657 (cgraph_decide_inlining): Use incremental inliner to handle flattening.
658 (try_inline): New function.
659 (cgraph_early_inlining): Update call of
660 cgraph_decide_inlining_incrementally. Apply inlining here.
661 (apply_inline): Update call of cgraph_decide_inlining_incrementally.
663 2007-01-21 Dirk Mueller <dmueller@suse.de>
666 * tree-vrp.c (check_array_bounds): do not warn
667 about ADDR_EXPR's of ARRAY_REF's which are immediately
668 used in binary expressions.
670 2007-01-21 Ira Rosen <irar@il.ibm.com>
672 * tree-vectorizer.h (struct _stmt_vec_info): Add new field
673 read_write_dep and macros for its access.
674 * tree-vectorizer.c (new_stmt_vec_info): Initialize the new field.
675 * tree-vect-analyze.c (vect_analyze_data_ref_dependence): Remove
676 argument, call vect_check_interleaving for every independent pair of
677 data-refs. Mark loads that access the same memory location as a store
679 (vect_check_dependences): Remove.
680 (vect_analyze_data_ref_dependences): Remove vect_check_dependences
681 call, fix the call to vect_analyze_data_ref_dependence.
682 (vect_analyze_data_ref_access): For statements that access the same
683 data-ref, check that they are not stores; for loads, check that there
684 is no store that access the same location.
686 2007-01-20 Roger Sayle <roger@eyesopen.com>
687 Joseph Myers <joseph@codesourcery.com>
689 * doc/invoke.texi (-fdump-rtl-combine): Fix under/overfull hbox.
690 (-fdump-rtl-stack): Likewise.
691 (-fno-signed-zeros): Use @minus{} for a minus sign. Correct typo.
692 (-mcheck-zero-division, -mcpu): Fix under/overfull hbox.
693 (-mpt-fixed): Use @minus{} for minus sign.
694 (Using Precompiled Headers): Fix under/overfull hbox.
696 2007-01-20 Jan Hubicka <jh@suse.cz>
698 * tree-flow.h (struct stmt_ann_d): Move references_memory to proper
699 place within annotation.
701 2007-01-20 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
703 * pa.c (output_move_double): Change array size of xoperands to 4.
705 2007-01-20 Jan Hubicka <jh@suse.cz>
707 * tree-tailcall.c (adjust_return_value): Do not use RESULT_DECL
710 2007-01-19 Ian Lance Taylor <iant@google.com>
712 * expmed.c (expand_divmod) [TRUNC_MOD_EXPR, TRUNC_DIV_EXPR]: Cast
713 constant to unsigned HOST_WIDE_INT before negating.
715 2007-01-19 Ian Lance Taylor <iant@google.com>
717 * tree-ssa-operands.h (struct vuse_vec_d): Change num_vuse field
719 (VUSE_VECT_ELEMENT) [ENABLE_CHECKING]: Use unsigned comparison.
720 (VUSE_ELEMENT_PTR) [ENABLE_CHECKING]: Likewise.
721 (SET_VUSE_VECT_ELEMENT) [ENABLE_CHECKING]: Likewise.
722 (SET_VUSE_ELEMENT_VAR) [ENABLE_CHECKING]: Likewise.
723 (SET_VUSE_ELEMENT_PTR) [ENABLE_CHECKING]: Likewise.
724 (realloc_vdef, realloc_vuse): Change second parameter to
726 (ssa_operand_iterator_d): Change vuse_index and mayuse_index
728 * tree-ssa-operands.c (realloc_vop): Change num_elem parameter to
729 unsigned. Change x and lim locals to unsigned.
730 (realloc_vdef, realloc_vuse): Change num_elem parameter to
732 (finalize_ssa_vuse_ops): Change old_i local to unsigned.
733 (copy_virtual_operands): Change i and n locals to unsigned.
735 2007-01-19 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
738 * builtins.c (fold_builtin_1): Handle builtin fdim.
740 2007-01-20 Jan Hubicka <jh@suse.cz>
742 * tree-ssa.c (init_tree_ssa): Do not call init_alias_heapvars.
743 * tree-ssa-structalias.c (compute_points_to_sets): Do call
745 (init_alias_heapvars): Initialize only when not already initialized.
746 (delete_alias_heapvars): Set heapvar_for_stmt to NULL.
748 2007-01-19 Roger Sayle <roger@eyesopen.com>
750 * common.opt (fsigned-zeros): New command line option.
751 * flags.h (HONOR_SIGNED_ZEROS): Control via flag_signed_zeros instead
752 of flag_unsafe_math_optimizations.
753 * opts.c (set_fast_math_flags): The -ffast-math command line option
754 implies -fno-signed-zeros.
755 (fast_math_flags_set_p): Likewise.
757 * doc/invoke.texi: Document new -fno-signed-zeros option, and update
758 the documentation of -ffast-math appropriately. Wrap long lines.
760 2007-01-19 Steve Ellcey <sje@cup.hp.com>
762 * system.h (ASM_MAKE_LABEL_LINKONCE): Poison.
763 * varasm.c (globalize_decl): Remove ASM_MAKE_LABEL_LINKONCE ifdef.
765 2007-01-19 Tomas Bily <tbily@suse.cz>
767 * cgraphunit.c (cgraph_finalize_function): Updating of pid
769 (tree_init_ic_make_global_vars): New function
770 (tree_init_edge_profiler): call of tree_init_ic_make_global_vars
771 (tree_gen_ic_profiler): New function
772 (tree_gen_ic_func_profiler): New function
773 (tree_profiling): Added calling of tree_gen_ic_func_profiler
774 (tree_profile_hooks): Added hook for indirec/virtual calls
775 * value-prof.c (tree_find_values_to_profile): New case for
777 (tree_values_to_profile): Call for determining indirect/virtual
779 (tree_indirect_call_to_profile): New function
780 (tree_ic_transform): New function
781 (tree_ic): New function
782 (find_func_by_pid): New function
783 (init_pid_map): New function
784 (tree_value_profile_transformations): Added check for
785 indirect/virtual call transformation
786 * value-prof.h (enum hist_type): New counter type for
787 indirect/virtual calls
788 (profile_hooks): Added new hook for profiling indirect/virtual
790 * profile.c (instrument_values): New case for indirect/virtual
792 * gcov-io.h (GCOV_LAST_VALUE_COUNTER): Changed to 6
793 (GCOV_COUNTER_V_INDIR): New counter type
794 (GCOV_COUNTER_NAMES): New name of counter "indirect" added
795 (GCOV_MERGE_FUNCTIONS): New merge function for indirect/virtual
797 * cgraph.c: Definition of cgraph_max_pid
798 (cgraph_create_node): Default init of pid attribute
799 * cgraph.h: Declaration of cgraph_max_pid
800 (struct cgraph_node): Added pid attribute
801 * libgcov.c (__gcov_indirect_call_profiler): New function
802 (__gcov_one_value_profiler_body): New function
803 (__gcov_one_value_profiler): Body was moved to
804 __gcov_one_value_profiler_body and calls it
806 2007-01-19 Basile Starynkevitch <basile@starynkevitch.net>
808 * doc/gty.texi (Options): Document the mark_hook option to GTY.
809 * gengtype.c (write_types_data, write_func_for_structure,
810 write_types, ggc_wtd, pch_wtd): Add skip_hooks to
811 write_types_data, ggc_wtd, pch_wtd for processing mark_hook.
812 (walk_type, write_func_for_structure): Generate the mark_hook if
815 2007-01-19 Jan Hubicka <jh@suse.cz>
817 * ipa-inline.c (cgraph_decide_inlining_incrementally): Instead of
818 'early' argument take inlining mode argument specifying whether to
819 inline for size/speeed or all functions; add support for flattening;
821 (cgraph_early_inlining): Update call of decide_inlining_incrementally.
823 2007-01-19 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
826 * toplev.c (warn_deprecated_use): Use %qD instead of %qs to print
827 the name of the declared identifier.
829 2007-01-19 Dirk Mueller <dmueller@suse.de>
831 * config/i386.h (CONDITIONAL_REGISTER_USAGE): Store
832 result of PIC_OFFSET_TABLE_REGNUM in temporary variable to avoid
833 duplicate evaluation.
835 2007-01-19 Uros Bizjak <ubizjak@gmail.com>
837 * config/i386/i386.md (acos<mode>2): Rename from acossf2 and acosdf2.
838 Macroize expander using X87MODEF12 mode macro. Extend operand 1
839 to XFMode, use acosxf2 and truncate result to requested mode.
840 Use SSE_FLOAT_MODE_P to disable patterns for SSE math.
841 (asin<mode>2): Similarly, with asin expanders.
842 (*fscalexf4_i387): Rename from *fscalexf4.
843 (expNcorexf3): New expander.
844 (expxf2, exp10xf2, exp2xf2): Use expNcorexf3 expander.
845 (exp<mode>2): Rename from expsf2 and expdf2. Macroize expander using
846 X87MODEF12 mode macro. Extend operand 1 to XFMode, use expxf2 and
847 truncate result to requested mode. Use SSE_FLOAT_MODE_P to disable
848 patterns for SSE math.
849 (exp10<mode>2): Similarly, with exp10 expanders.
850 (exp2<mode>2): Similarly, with exp2 expanders.
851 (expm1<mode>2): Similarly, with expm1 expanders.
852 (ldexp<mode>3): Similarly, with ldexp expanders.
853 (log<mode>2, log10<mode>2, log2<mode>2, log1p<mode>2, logb<mode>2):
854 Use gen_truncxf<mode>2_i387_noop to truncate result.
856 2007-01-19 Richard Sandiford <richard@codesourcery.com>
858 * config/m68k/m68k.h (PREFERRED_STACK_BOUNDARY): Define to 32
859 for ColdFire targets.
861 2007-01-19 Nathan Sidwell <nathan@codesourcery.com>
862 Richard Sandiford <richard@codesourcery.com>
864 * config/m68k/m68k.h (M68K_STATIC_CHAIN_REG_NAME): New macro.
865 (INITIALIZE_TRAMPOLINE): Use STATIC_CHAIN_REGNUM.
866 (__transfer_from_trampoline): Use M68K_STATIC_CHAIN_REG_NAME.
867 * config/m68k/m68kelf.h (STATIC_CHAIN_REGNUM): Override.
868 (M68K_STATIC_CHAIN_REG_NAME): Likewise.
869 * config/m68k/netbsd-elf.h (M68K_STATIC_CHAIN_REG_NAME): Likewise.
871 2007-01-19 Richard Sandiford <richard@codesourcery.com>
873 * config/m68k/m68k.md (adddi_dilshr32): Rename to...
874 (*adddi_dilshr32): ...this. Fix formatting. Remove commented-out
875 non-canonical pattern. Restrict to !TARGET_COLDFIRE.
876 (*adddi_dilshr32_cf): New pattern.
877 (adddi3, subdi3): Remove first alternatives.
879 2007-01-19 Richard Sandiford <richard@codesourcery.com>
881 * config/m68k/m68k.c (notice_update_cc): If an SFmode move is
882 implemented using move.l, do not use its cc result for floating-point
885 2007-01-19 Richard Sandiford <richard@codesourcery.com>
887 * config/m68k/m68k.h (EXTRA_CONSTRAINT): Stop the 'T' constraint
888 from accepting 's' constraints if flag_pic.
890 2007-01-19 Richard Sandiford <richard@codesourcery.com>
892 * config/m68k/m68k.md (bordered, bunordered, buneq, bunge, bungt)
893 (bunle, bunlt, bltgt, bordered_rev, bunordered_rev, buneq_rev)
894 (bunge_rev, bungt_rev, bunle_rev, bunlt_rev, bltgt_rev): Change
895 condition from TARGET_68881 to TARGET_HARD_FLOAT.
897 2007-01-19 Sandra Loosemore <sandra@codesourcery.com>
899 * longlong.h (count_leading_zeros, COUNT_LEADING_ZEROS_0): Add
900 ColdFire alternatives.
901 * config/m68k/m68k.h (CLZ_DEFINED_VALUE_AT_ZERO): New macro.
902 * config/m68k/m68k.md (clzsi2): Define for ColdFire
903 architectures that support ff1 instruction.
905 2007-01-19 Richard Sandiford <richard@codesourcery.com>
906 Julian Brown <julian@codesourcery.com>
908 * config/m68k/m68k.h (CONST_OK_FOR_LETTER_P): Add an 'R' case.
909 * config/m68k/m68k.md (*movsi_cfv4): Fold into...
910 (*movsi_cf): ...here. Remove unnecessary 'R' from 'Rg'.
913 2007-01-19 Richard Sandiford <richard@codesourcery.com>
915 * config/m68k/m68k-protos.h (valid_mov3q_const): Take a HOST_WIDE_INT
917 (output_move_const_into_data_reg, output_move_simode_const): Delete.
918 * config/m68k/m68k.c (const_method, const_int_cost): Take a
919 HOST_WIDE_INT instead of an rtx.
920 (m68k_rtx_costs): Update call accordingly.
921 (output_move_const_into_data_reg): Likewise. Fix formatting.
922 (valid_mov3q_const): Take a HOST_WIDE_INT instead of an rtx.
924 (output_move_simode_const): Update calls after above changes.
925 Rework to use automatic variables and predicates like MEM_P.
926 * config/m68k/m68k.md (pushexthisi_const): Update call to
929 2007-01-19 Dirk Mueller <dmueller@suse.de>
931 * tree-ssa-alias.c (perform_var_substitution): Fix typo
934 2007-01-19 Richard Guenther <rguenther@suse.de>
936 * builtins.c (expand_builtin_cexpi): Fall back to expanding
937 via cexp in case sincos is not available.
939 2007-01-19 Richard Guenther <rguenther@suse.de>
941 * doc/tm.texi (TARGET_HAS_SINCOS): Document new target macro.
942 * defaults.h (TARGET_HAS_SINCOS): Default to off.
943 * config/linux.h (TARGET_HAS_SINCOS): Set to on if we have glibc.
944 * config/alpha/linux.h (TARGET_HAS_SINCOS): Likewise.
945 * config/sparc/linux.h (TARGET_HAS_SINCOS): Likewise.
946 * config/sparc/linux64.h (TARGET_HAS_SINCOS): Likewise.
947 * config/rs6000/linux.h (TARGET_HAS_SINCOS): Likewise.
948 * config/rs6000/linux64.h (TARGET_HAS_SINCOS): Likewise.
950 2007-01-19 Uros Bizjak <ubizjak@gmail.com>
952 * config/i386/i386.md (*fpatanxf3_i387, fpatan_extend<mode>xf3_i387):
954 (atan2sf3_1, atan2df3_1, atan2xf3_1): Remove insn patterns.
955 (atan2xf3): Directly generate RTL pattern.
956 (atan2<mode>3): Rename from atan2sf3 and atan2df3 and macroize insn
957 patterns using X87MODEF12 mode macro. Use fpatan_extend<mode>xf3_i387
958 and truncate result to requested mode. Use SSE_FLOAT_MODE_P to
959 disable patterns for SSE math.
960 (atan<mode>2): Rename from atansf2 and atandf2 and macroize insn
961 patterns using X87MODEF12 mode macro. Use fpatan_extend<mode>xf3_i387
962 and truncate result to requested mode. Use SSE_FLOAT_MODE_P to
963 disable patterns for SSE math.
965 2007-01-19 Alexandre Oliva <aoliva@redhat.com>
967 * libgcc-std.ver: Fix typo in %inherit for GCC_4.3.0.
969 2007-01-18 Roger Sayle <roger@eyesopen.com>
971 * fold-const.c (fold_unary) <VIEW_CONVERT_EXPR>: Optimize away a
972 VIEW_CONVERT_EXPR to the same type as it's operand.
974 2007-01-18 David Edelsohn <edelsohn@gnu.org>
976 * config/rs6000/darwin-ldouble.c: Only build _SOFT_FLOAT if
977 configured for long double 128.
979 2007-01-18 Mike Stump <mrs@apple.com>
981 * config/rs6000/rs6000.c (rs6000_emit_vector_compare): Fix build
984 2007-01-18 Michael Meissner <michael.meissner@amd.com>
986 * i386.c (ix86_compute_frame_layout): Make fprintf's in #if 0 code
989 2007-01-18 Jan Hubicka <jh@suse.cz>
991 * tree-ssa-operands.c (vop_free_bucket_size): Never return value
992 greater than NUM_VOP_FREE_BUCKETS.
994 2007-01-18 Daniel Berlin <dberlin@dberlin.org>
996 * tree-ssa-structalias.c: Update comments.
997 (ptabitmap_obstack): Removed.
999 (oldpta_obstack): Ditto.
1000 (stats): Add a few members.
1001 (struct variable_info): Remove node, complex, address_taken, and
1002 indirect_target members. Add oldsolution member.
1003 (new_var_info): Do not initialize removed members.
1004 (constraint_expr_type): Remove INCLUDES.
1005 (constraint_graph): Add size, implicit_preds, rep,
1006 indirect_cycles, eq_rep, label, direct_nodes, and complex members.
1007 (FIRST_REF_NODE): New macro.
1008 (LAST_REF_NODE): Ditto.
1009 (FIRST_ADDR_NODE): Ditto.
1010 (find): New function.
1012 (dump_constraint): Do not handle INCLUDES.
1013 (insert_into_complex): Do not insert duplicate constraints.
1014 (condense_varmap_nodes): Renamed and rewritten into ...
1015 (merge_node_constraints): This. Also fix bug in handling of
1016 offseted copy constraints.
1017 (clear_edges_for_node): No longer need to deal with preds at all,
1018 or removing associated preds/succs.
1019 (merge_graph_nodes): Deal with indirect_cycles.
1020 Don't deal with predecessors.
1021 (add_implicit_graph_edge): New function.
1022 (add_pred_graph_edge): Ditto.
1023 (add_graph_edge): Don't deal with predecessors.
1024 (build_constraint_graph): Removed.
1025 (build_pred_graph): New function.
1026 (build_succ_graph): Ditto.
1027 (struct scc_info): Removed in_component. Added roots, dfs, and
1028 node_mapping. Remove visited_index, unification_queue.
1029 (scc_visit): Deal with union-find we do now.
1030 Deal with cycles with REF nodes.
1031 (collapse_nodes): Renamed and rewritten to ...
1032 (unify_nodes): This.
1033 (process_unification_queue): Removed.
1034 (topo_visit): Cleanup
1035 (do_da_constraint): Use find.
1036 (do_sd_constraint): Ditto.
1037 (do_ds_constraint): Ditto.
1038 (do_complex_constraint): Ditto.
1039 (init_scc_info): Update for removed and added members.
1040 (find_and_collapse_graph_cycles): Renamed and rewritten into ...
1041 (find_indirect_cycles): This.
1042 (equivalence_class): New variable.
1043 (label_visit): New function.
1044 (perform_variable_substitution): Rewritten.
1045 (free_var_substitution_info): New function.
1046 (find_equivalent_node): Ditto.
1047 (move_complex_constraints): Ditto.
1048 (eliminate_indirect_cycles): Ditto.
1049 (solve_graph): Only propagate changed bits.
1050 Use indirect cycle elimination.
1052 (tree_id_t): Rename to tree_vi_t, delete id member, add vi member.
1053 (tree_id_eq): Renamed to ...
1054 (tree_vi_eq): This. Update for member change
1055 (insert_id_for_tree): Renamed and rewritten to ...
1056 (insert_vi_for_tree): This.
1057 (lookup_id_for_tree): Renamed and rewritten to ...
1058 (lookup_vi_for_tree): This.
1059 (get_id_for_tree): Renamed and rewritten to ...
1060 (get_vi_for_tree): Ditto.
1061 (get_constraint_exp_from_ssa_var): Update to use get_vi_for_tree.
1062 (process_constraint): Don't handle INCLUDES.
1063 Remove special ADDRESSOF case.
1064 (find_func_aliases): Rewrite to use vi functions instead of id
1066 (create_function_info_for): Ditto.
1067 (create_variable_info_for): Ditto.
1068 (intra_create_variable_infos): Ditto.
1069 (merge_smts_into): Ditto.
1070 (find_what_p_points_to): Ditto.
1071 (init_base_vars): Ditto.
1072 (init_alias_vars): Ditto.
1073 (remove_preds_and_fake_succs): New function.
1074 (dump_sa_points_to_info): Dump new stats.
1075 (dump_solution_for_var): Use find.
1076 (set_used_smts): Fix formatting.
1077 (compute_points_to_sets): Updated for new functions.
1078 (ipa_pta_execute): Ditto.
1080 2007-01-18 Kazu Hirata <kazu@codesourcery.com>
1081 Richard Sandiford <richard@codesourcery.com>
1083 * doc/tm.texi (TARGET_FUNCTION_VALUE): Expand documentation of
1085 * calls.c (expand_call): If the return value is a PARALLEL,
1086 extract its first member.
1087 * config/m68k/linux.h (FUNCTION_EXTRA_EPILOGUE): Remove.
1088 * config/m68k/m68k.c (m68k_output_function_epilogue): Don't
1089 use FUNCTION_EXTRA_EPILOGUE.
1090 (m68k_function_value): Return a PARALLEL if the return value
1091 is of a pointer type.
1092 * config/m68k/netbsd-elf.h (current_function_returns_pointer)
1093 (FUNCTION_EXTRA_EPILOGUE): Remove.
1094 * config/m68k/m68k.md (D0_REG): New constant.
1096 2007-01-18 Kazu Hirata <kazu@codesourcery.com>
1098 * config/m68k/m68k.c (m68k_output_function_epilogue): Don't
1099 output a NOP for empty epilogues.
1101 2007-01-18 Richard Sandiford <richard@codesourcery.com>
1103 * config/m68k/m68k.c (m68k_use_return_insn): Update comments
1104 before function. Extend register save check to include all
1105 registers, not just integer ones.
1107 2007-01-18 Kazu Hirata <kazu@codesourcery.com>
1109 * config/m68k/m68k-protos.h (use_return_insn): Rename to...
1110 (m68k_use_return_insn): ...this.
1111 * config/m68k/m68k.h (USE_RETURN_INSN): Delete.
1112 * config/m68k/m68k.c (use_return_insn): Rename to...
1113 (m68k_use_return_insn): ...this.
1114 * config/m68k/m68k.md (return): Use m68k_use_return_insn instead
1117 2007-01-18 Nathan Sidwell <nathan@codesourcery.com>
1119 * config/m68k/fpgnulib.c (__truncdfsf2): Implement round to
1120 nearest even, fix denormal rounding overflow.
1122 2007-01-18 Richard Sandiford <richard@codesourcery.com>
1124 * config/m68k/m68k.md (movsf_cf_hard): Use fsmove instead of
1125 f%$smove and f%$move.
1126 (movdf_cf_hard): Use fdmove for cases 0 and 3 and fmove for case 1.
1127 (extendsfdf2_cf): Use fdmove instead of f%&move.
1128 (truncdfsf2_cf): Use fsmove instead of f%$smove.
1129 (add<mode>3_cf, sub<mode>3_cf): Use <FP:prec> instead of <FP:round>.
1131 2007-01-18 Richard Sandiford <richard@codesourcery.com>
1133 * config/m68k/m68k.md (movdf_cf_hard): Use output_move_double for
1136 2007-01-18 Richard Sandiford <richard@codesourcery.com>
1138 * real.h (real_format): Add a canonical_nan_lsbs_set field.
1139 (coldfire_single_format): Declare.
1140 (coldfire_double_format): Likewise.
1141 * real.c (encode_ieee_single): Use canonical_nan_lsbs_set instead
1142 of qnan_msb_set to determine the lower bits of a canonical
1144 (encode_ieee_double): Likewise.
1145 (encode_ieee_quad): Likewise.
1146 (ieee_single_format): Initialize canonical_nan_lsbs_set.
1147 (mips_single_format): Likewise.
1148 (ieee_double_format): Likewise.
1149 (mips_double_format): Likewise.
1150 (ieee_extended_motorola_format): Likewise.
1151 (ieee_extended_intel_96_format): Likewise.
1152 (ieee_extended_intel_128_format): Likewise.
1153 (ieee_extended_intel_96_round_53_format): Likewise.
1154 (ibm_extended_format): Likewise.
1155 (mips_extended_format): Likewise.
1156 (ieee_quad_format): Likewise.
1157 (mips_quad_format): Likewise.
1158 (vax_f_format): Likewise.
1159 (vax_d_format): Likewise.
1160 (vax_g_format): Likewise.
1161 (i370_single_format): Likewise.
1162 (i370_double_format): Likewise.
1163 (decimal_single_format): Likewise.
1164 (decimal_double_format): Likewise.
1165 (decimal_quad_format): Likewise.
1166 (c4x_single_format): Likewise.
1167 (c4x_extended_format): Likewise.
1168 (real_internal_format): Likewise.
1169 (coldfire_single_format): New real_format.
1170 (coldfire_double_format): Likewise.
1171 * config/pdp11/pdp11.c (pdp11_f_format): Initialize
1172 canonical_nan_lsbs_set.
1173 (pdp11_d_format): Likewise.
1174 * config/m68k/m68k.c (override_options): Override REAL_FORMAT_MODE
1175 if TARGET_COLDFIRE_CPU.
1177 2007-01-18 Richard Sandiford <richard@codesourcery.com>
1179 * config/m68k/m68k-protos.h (m68k_output_pic_call): Delete.
1180 (output_call, m68k_legitimize_call_address): Declare.
1181 * config/m68k/m68k.h (EXTRA_CONSTRAINT): Remove unnecessary
1182 parenthesees. Add support for a 'W' constraint.
1183 (LEGITIMATE_PIC_OPERAND_P): Remove SYMBOL_REF_FLAG handling.
1184 (PRINT_OPERAND_PUNCT_VALID_P): Remove comment about 'o'.
1185 (m68k_symbolic_call, m68k_symbolic_jump): Declare.
1186 * config/m68k/m68k.c (m68k_symbolic_call, m68k_symbolic_jump): New
1188 (override_options): Initialize them. Do not set flag_no_function_cse
1189 for TARGET_ID_SHARED_LIBRARY.
1190 (m68k_output_pic_call): Delete.
1191 (m68k_legitimize_call_address): New function.
1192 (print_operand): Remove the %o prefix. Handle the %p prefix.
1193 (output_call): New function.
1194 (m68k_output_mi_thunk): Use m68k_symbolic_jump. Always load the
1195 target address from the GOT if symbolic jumps are not allowed.
1196 * config/m68k/m68k.md (call, general_operand): Do not set
1197 SYMBOL_REF_FLAG. Use m68k_legitimize_call_address instead.
1198 Merge separate flag_pic and !flag_pic define_insns into...
1199 (*call, *call_value): ...these new patterns. Match the address
1200 rather than the containing MEM and require it to be a call_operand.
1201 Use output_call to generate the asm template.
1202 * config/m68k/predicates.md (const_call_operand): New predicate.
1203 (call_operand): Likewise.
1205 2007-01-18 Nathan Sidwell <nathan@codesourcery.com>
1207 * config/m68k/m68k.h (REGISTER_MOVE_COST): Simplify definition.
1208 (STACK_GROWS_DOWNWARD): Define to 1.
1209 (FUNCTION_VALUE, LIBCALL_VALUE, FUNCTION_VALUE_REGNO_P): Equivocate
1210 comments, emphasizing that these values are only defaults.
1211 * config/m68k/linux.h (LINK_SPEC): Fix formatting in #undef.
1212 * config/m68k/m68k.c (const_method): Remove trailing whitespace.
1214 2007-01-18 Richard Sandiford <richard@codesourcery.com>
1216 * config/m68k/m68k.md (cmpsi): Remove outdated flag_pic handling.
1218 2007-01-18 Kazu Hirata <kazu@codesourcery.com>
1219 Richard Sandiford <richard@codesourcery.com>
1221 * config/m68k/m68k.h (DATA_REGNO_P, ADDRESS_REGNO_P, INT_REGNO_P)
1222 (FP_REGNO_P): New macros.
1223 (REGNO_OK_FOR_INDEX_P, REGNO_OK_FOR_BASE_P, REGNO_OK_FOR_DATA_P)
1224 (REGNO_OK_FOR_FP_P, REG_OK_FOR_INDEX_P, REG_OK_FOR_BASE_P): Use them.
1225 (EH_RETURN_STACKADJ_RTX): Use A0_REG.
1226 * config/m68k/m68k.c (m68k_regno_mode_ok): Use the new REGNO macros.
1228 2007-01-18 Nathan Sidwell <nathan@codesourcery.com>
1230 * config.gcc (m68k-*-aout*, m68k-*-coff*, m68020-*-elf*, m68k-*-elf*)
1231 (m68k-*-uclinuxoldabi, m68k-*-uclinux*, m68k-*-rtems*): Add t-floatlib
1233 * config/m68k/t-floatlib: New file, extracting common code from...
1234 * config/m68k/t-m68kbare, config/m68k/t-m68kelf,
1235 * config/m68k/t-uclinux: Here.
1236 * config/m68k/fpgnulib.c: Do not compile extendeed precision
1237 routines on ColdFire targets.
1239 2007-01-18 Nathan Sidwell <nathan@codesourcery.com>
1241 * config.gcc (m68k-*-aout*, m68k-*-coff*, m68020-*-elf*, m68k-*-elf*)
1242 (m68010-*-netbsdelf*, m68k*-*-netbsdelf*, m68k*-*-openbsd*)
1243 (m68k-*-uclinuxoldabi, m68k-*-uclinux*, m68k-*-linux*)
1244 (m68k-*-rtems*): Use tm_file rather than m68k/m68k.h and
1245 explicitly set MOTOROLA to 1.
1246 * config/m68k/m68k.h (MOTOROLA): Simplify definition accordingly.
1248 2007-01-18 Richard Sandiford <richard@codesourcery.com>
1249 Nathan Sidwell <nathan@codesourcery.com>
1251 * config/m68k/m68k.h (PCC_STATIC_STRUCT_RETURN): Delete.
1252 (ASM_OUTPUT_REG_PUSH, ASM_OUTPUT_REG_POP): Add MOTOROLA cases.
1253 * config/m68k/coff.h (REGISTER_PREFIX_MD): Delete.
1254 * config/m68k/m68020-elf.h (LIB_SPEC): Delete.
1255 * config/m68k/m68k-none.h (CC1_SPEC, CPP_SUBTARGET_SPEC): Delete.
1256 * config/m68k/m68kelf.h (IMMEDIATE_PREFIX, REGISTER_PREFIX_MD)
1257 (ASM_OUTPUT_REG_PUSH, ASM_OUTPUT_REG_POP): Delete.
1258 (NO_DOLLAR_IN_LABEL, PCC_STATIC_STRUCT_RETURN): Don't undefine.
1259 (BSS_ASM_OP, ASM_OUTPUT_SKIP, ASM_OUTPUT_ASCII): Delete.
1260 * config/m68k/m68kemb.h (PCC_STATIC_STRUCT_RETURN): Don't undefine.
1261 * config/m68k/linux.h (SIZE_TYPE, PTRDIFF_TYPE, WCHAR_TYPE)
1262 (WCHAR_TYPE_SIZE, TARGET_OBJFMT_CPP_BUILTINS): Delete.
1263 (TARGET_OS_CPP_BUILTINS): Don't define mc68000 and mc68020 here.
1264 (DBX_REGISTER_NUMBER): Delete.
1265 * config/m68k/netbsd-elf.h (IMMEDIATE_PREFIX): Delete.
1266 (PCC_STATIC_STRUCT_RETURN): Don't undefine.
1267 * config/m68k/openbsd.h (PCC_STATIC_STRUCT_RETURN): Define.
1269 2007-01-18 Richard Sandiford <richard@codesourcery.com>
1271 * config.gcc (m68k-*-uclinux*): Add flat.h to $tm_file.
1272 * config/flat.h: New file.
1273 * crtstuff.c (USE_PT_GNU_EH_FRAME): Don't define if
1275 * config/m68k/m68k.h (ASM_PREFERRED_EH_DATA_FORMAT): Do not use
1276 indirect references for -msep-data or -mid-shared-library.
1277 Do not use PC-relative code addresses either.
1278 * config/m68k/m68k.c (override_options): Restrict -fPIC error
1280 * config/m68k/uclinux.h (STARTFILE_SPEC): Define. Use Scrt1.o
1281 for shared libraries and crt1.o for executables. Use crti.o and
1283 (ENDFILE_SPEC): Use crtend.o and crtn.o.
1284 (LIB_SPEC): Suppress -Rlibc.gdb if -static-libc is given.
1285 Do not add -elf2flt or -shared-lib-id options here.
1286 (LINK_SPEC): Define. Pass -elf2flt if no -elf2flt option is given.
1287 Pass -shared-lib-id if -mid-shared-library, taking the library
1288 identifier from -mshared-library-id if given, otherwise
1290 (EH_FRAME_IN_DATA_SECTION): Do not undefine.
1291 (INIT_SECTION_ASM_OP, FINI_SECTION_ASM_OP): Likewise.
1292 (TARGET_OS_CPP_BUILTINS): Define __GXX_MERGED_TYPEINFO_NAMES=0
1293 and __GXX_TYPEINFO_EQUALITY_INLINE=0 if -mid-shared-library.
1294 (DRIVER_SELF_SPECS): Map unadorned PIC options to -msep-data.
1295 * config/m68k/t-uclinux (EXTRA_MULTILIB_PARTS): Add crtbegin.o
1297 * config/m68k/lb1sf68.asm (PICCALL): Use an lea and pc-relative
1298 jump sequence for ISA A and ISA A+.
1299 (PICJUMP): Likewise.
1301 2007-01-18 Richard Sandiford <richard@codesourcery.com>
1303 * config.gcc (m68k-*-uclinux*): Base the port on the common
1304 and m68k GNU/Linux files rather than on the generic ELF ones.
1305 * config/m68k/uclinux.h (TARGET_VERSION): Override.
1306 (TARGET_OS_CPP_BUILTINS): Use LINUX_TARGET_OS_CPP_BUILTINS.
1308 2007-01-18 Julian Brown <julian@codesourcery.com>
1309 Richard Sandiford <richard@codesourcery.com>
1311 * config/m68k/m68k.h (LONG_DOUBLE_TYPE_SIZE): Make 64-bit on ColdFire.
1312 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
1313 * config/m68k/netbsd-elf.h (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Undefine
1315 * config/m68k/uclinux-oldabi.h (LONG_DOUBLE_TYPE_SIZE): Redefine to
1317 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
1319 2007-01-18 Richard Sandiford <richard@codesourcery.com>
1321 * doc/install.texi: Document m68k-uclinuxoldabi.
1322 * config.gcc (m68k-*-uclinuxoldabi): New configuration.
1323 * config/m68k/uclinux-oldabi.h: New file, copied from
1324 config/m68k/uclinux.h.
1326 2007-01-18 Kaz Kojima <kkojima@rr.iij4u.or.jp>
1328 * config/m32r/m32r-protos.h (m32r_expand_epilogue): Declare it.
1329 * config/m32r/m32r.c (m32r_setup_incoming_varargs): Use gen_frame_mem.
1330 (m32r_compute_frame_size): Use unsigned for regno.
1331 (m32r_reload_lr): Use gen_frame_mem.
1333 (m32r_output_function_epilogue): Don't output the function epilogue
1335 (m32r_expand_epilogue): New.
1336 (direct_return): Return false if the function has the interrupt
1338 (m32r_hard_regno_rename_ok): Remove code for the textual epilogue.
1339 * config/m32r/m32r.md (epilogue): New expander.
1340 (return_lr, return_rte): New insns.
1341 (return): Make it expander.
1342 (return_normal): New expander.
1344 2007-01-18 Josh Conner <jconner@apple.com>
1347 * config/rs6000/rs6000.c (rs6000_emit_vector_compare): Add
1348 support for UNLE, UNLT, UNGE, and UNGT.
1350 2007-01-18 Jan Hubicka <jh@suse.cz>
1352 * tree-vrp.c (finalize_jump_threads): Do not call cleanup_cfg by hand.
1353 * tree-tailcall (add_virtual_phis): Likewise.
1354 (optimize_tail_call): Return TODOs.
1355 (execute_tail_calls): Return TODOs.
1356 * tree-ssa-ccp (execute_fold_all_builtins): Do cleanup_cfg via TODO.
1357 * tree-cfgcleanup.c (cleanup_tree_cfg_loop): Return if something
1359 * tree-ssa-forwprop.c (tree_ssa_forward_propagate_single_use_value):
1360 Cleanup cfg using TODO.
1361 * tree-flow.h (cleanup_tree_cfg_loop): Update prototype.
1362 * passes.c (execute_function_todo): When cleanup did something, remove
1364 * tree-cfg.c (pass_build_cfg): Add cleanup_cfg TODO.
1365 (make_edges): Don't cleanup_cfg.
1367 2007-01-18 Uros Bizjak <ubizjak@gmail.com>
1369 * reg-stack.c (subst_stack_regs_pat) [UNSPEC_SINCOS_COS,
1370 UNSPEC_XTRACT_FRACT]: Use generic code for instructions that
1371 operate on the top of stack.
1372 [UNSPEC_SINCOS_SIN, UNSPEC_XTRACT_EXP, UNSPEC_TAN]: Rewrite
1373 register handling of instructions that output to the second
1375 [UNSPEC_TAN_ONE, UNSPEC_TAN_TAN]: Remove.
1376 (move_for_stack_reg): Special-case check for dead destination
1377 stack slot for constant load of 1.0 inside UNSPEC_TAN.
1379 * config/i386/i386.md (UNSPEC_TAN): New constant.
1380 (UNSPEC_TAN_ONE, UNSPEC_TAN_TAN): Remove.
1381 (fptanxf4_i387, fptan_extend<mode>xf4_i387): New patterns
1382 to correctly model move of constant 1.0 to top stack slot.
1383 (*tandf3_1, *tansf3_1, *tanxf3_1): Remove insn patterns.
1384 (unnamed peephole2 pattern): Remove corresponding peephole2
1385 pattern that optimizes tan insn and loading of constant 1.0.
1386 (tanxf2): Use fptanxf4_i387.
1387 (tan<mode>2): Rename from tansf2 and tandf2 and macroize insn
1388 patterns using X87MODEF12 mode macro. Use fptan_extend<mode>xf4_i387
1389 and truncate result to requested mode. Use SSE_FLOAT_MODE_P to
1390 disable patterns for SSE math.
1391 (sincos<mode>3): Use truncxf<mode>2_i387_noop for truncation.
1392 (fyl2x_extend<mode>xf3_i387): Use X87MODEF12 for operand 1.
1394 2007-01-18 Dirk Mueller <dmueller@suse.de>
1395 Richard Guenther <rguenther@suse.de>
1398 * doc/invoke.texi (Warray-bounds): Document -Warray-bounds.
1399 * common.opt (Warray-bounds): Add new warning option.
1400 * c-opts.c (c_common_handle_option): Define -Warray-bounds
1402 * Makefile.in: make tree-vrp.o depend on toplev.h
1403 * tree-vrp.c (vrp_finalize): Call check_array_refs if -Warray-bounds
1405 (check_array_refs, check_array_bounds, check_array_ref): New.
1407 2007-01-18 Jan Hubicka <jh@suse.cz>
1409 * tree-ssa-ccp.c (ccp_finalize): Return if something changed.
1410 (execute_ssa_ccp): Return flags conditionally.
1411 * tree-ssa-propagate.c (substitue_and_fold): Return if something was
1413 * tree-ssa-propagate.h (substitute_and_fold): Update prototype.
1415 2007-01-18 Steven Bosscher <steven@gcc.gnu.org>
1417 * cfgcleanup.c (cleanup_cfg): Detect cfglayout mode and set
1418 the CLEANUP_CFGLAYOUT flag when in cfglayout mode.
1420 * Makefile.c (GTFILES): Add cfglayout.h.
1421 * gengtype.c (open_base_files): Likewise.
1422 * cfglayout.c (cfg_layout_function_footer,
1423 cfg_layout_function_header) Reindent to make gengtype happy.
1424 * cfglayout.h (cfg_layout_function_footer,
1425 cfg_layout_function_header): Add GTY(()) marker.
1427 * ifcvt.c (noce_try_sign_mask): Make sure INSN_B is non-null.
1429 2007-01-18 Ben Elliston <bje@au.ibm.com>
1431 * genautomata.c (write_automata): Include xstrerror output in the
1432 error message if writing the DFA description file fails.
1434 2007-01-17 H.J. Lu <hongjiu.lu@intel.com>
1436 * config/mips/mips-protos.h (mips_output_external): Make it
1438 * config/mips/iris.h (TARGET_ASM_EXTERNAL_LIBCALL): Removed.
1439 * config/mips/mips.c (irix_output_external_libcall): Likewise.
1440 (extern_list): Likewise.
1441 (extern_head): Likewise.
1442 (TARGET_ASM_FILE_END): Likewise.
1443 (mips_file_end): Likewise.
1444 (mips_output_external): Rewritten.
1446 2007-01-18 Ben Elliston <bje@au.ibm.com>
1448 * genpreds.c (write_insn_preds_c): Only write out the function
1449 body for regclass_for_constraint if we have register constraints.
1451 2007-01-17 Tom Tromey <tromey@redhat.com>
1453 * doc/sourcebuild.texi (libgcj Tests): Use sourceware.org.
1454 * doc/install.texi (Testing): Use sourceware.org.
1455 (Binaries): Likewise.
1456 (Specific): Likewise.
1457 * doc/contrib.texi (Contributors): Use sourceware.org.
1459 2007-01-17 Anatoly Sokolov <aesok@post.ru>
1461 * config/avr/avr.h (AVR_HAVE_LPMX): New macro.
1462 (AVR_ENHANCED): Rename to ...
1463 (AVR_HAVE_MUL): ... new.
1464 (avr_enhanced_p): Rename to ...
1465 (avr_have_mul_p): ... new.
1466 (TARGET_CPU_CPP_BUILTINS): Use 'avr_have_mul_p' instead of
1467 'avr_enhanced_p' for "__AVR_ENHANCED__". Define "__AVR_HAVE_MUL__".
1468 * config/avr/avr.c (avr_enhanced_p): Rename to ...
1469 (avr_have_mul_p): ... new.
1470 (base_arch_s): Rename 'enhanced' to 'have_mul'.
1471 (avr_override_options): Use 'avr_have_mul_p' and 'have_mul' instead of
1472 'avr_enhanced_p' and 'enhanced'.
1473 (ashlhi3_out, ashrhi3_out, lshrhi3_out, avr_rtx_costs): Use
1474 AVR_HAVE_MUL instead of AVR_ENHANCED.
1475 * avr.md (*tablejump_enh): Use AVR_HAVE_LPMX instead of AVR_ENHANCED.
1476 (mulqi3, *mulqi3_enh, *mulqi3_call, mulqihi3, umulqihi3, mulhi3,
1477 *mulhi3_enh, *mulhi3_call, mulsi3, *mulsi3_call): Use AVR_HAVE_MUL
1478 instead of AVR_ENHANCED.
1479 (*tablejump_enh): Use AVR_HAVE_LPMX instead of AVR_ENHANCED.
1480 * libgcc.S: Use __AVR_HAVE_MUL__ instead of __AVR_ENHANCED__.
1481 (__tablejump__): Use __AVR_HAVE_LPMX__ instead of __AVR_ENHANCED__.
1483 2007-01-17 Ian Lance Taylor <iant@google.com>
1485 * vec.h (VEC_reserve_exact): Define.
1486 (vec_gc_p_reserve_exact): Declare.
1487 (vec_gc_o_reserve_exact): Declare.
1488 (vec_heap_p_reserve_exact): Declare.
1489 (vec_heap_o_reserve_exact): Declare.
1490 (VEC_OP (T,A,reserve_exact)): New static inline function, three
1492 (VEC_OP (T,A,reserve)) [all versions]: Remove handling of
1494 (VEC_OP (T,A,alloc)) [all versions]: Call ...reserve_exact.
1495 (VEC_OP (T,A,copy)) [all versions]: Likewise.
1496 (VEC_OP (T,a,safe_grow)) [all versions]: Likewise.
1497 * vec.c (calculate_allocation): Add exact parameter. Change all
1499 (vec_gc_o_reserve_1): New static function, from vec_gc_o_reserve.
1500 (vec_gc_p_reserve, vec_gc_o_reserve): Call vec_gc_o_reserve_1.
1501 (vec_gc_p_reserve_exact, vec_gc_o_reserve_exact): New functions.
1502 (vec_heap_o_reserve_1): New static function, from vec_heap_o_reserve.
1503 (vec_heap_p_reserve, vec_heap_o_reserve): Call vec_heap_o_reserve_1.
1504 (vec_heap_p_reserve_exact): New function.
1505 (vec_heap_o_reserve_exact): New function.
1507 2007-01-17 Jan Hubicka <jh@suse.cz>
1509 * ipa-type-escape.c (look_for_casts): Revamp using handled_component_p.
1511 2007-01-17 Eric Christopher <echristo@apple.com>
1513 * config.gcc: Support core2 processor.
1515 2007-01-16 Jan Hubicka <jh@suse.cz>
1517 * tree-ssanames.c (release_dead_ssa_names): Instead of ggc_freeing
1518 the names, just unlink the chain so we don't crash on dangling pointers
1521 2007-01-16 Jan Hubicka <jh@suse.cz>
1523 * cgraph.h (cgraph_decide_inlining_incrementally): Kill.
1524 * tree-pass.h: Reorder to make IPA passes appear toegher.
1525 (pass_early_inline, pass_inline_parameters, pass_apply_inline): Declare.
1526 * cgraphunit.c (cgraph_finalize_function): Do not compute inling
1527 parameters, do not call early inliner.
1528 * ipa-inline.c: Update comments. Include tree-flow.h
1529 (cgraph_decide_inlining): Do not compute inlining parameters.
1530 (cgraph_decide_inlining_incrementally): Return TODOs; assume to
1531 be called with function context set up.
1532 (pass_ipa_inline): Remove unreachable functions before pass.
1533 (cgraph_early_inlining): Simplify assuming to be called from the
1535 (pass_early_inline): New pass.
1536 (cgraph_gate_ipa_early_inlining): New gate.
1537 (pass_ipa_early_inline): Turn into simple wrapper.
1538 (compute_inline_parameters): New function.
1539 (gate_inline_passes): New gate.
1540 (pass_inline_parameters): New pass.
1541 (apply_inline): Move here from tree-optimize.c
1542 (pass_apply_inline): New pass.
1543 * ipa.c (cgraph_remove_unreachable_nodes): Verify cgraph after
1545 * tree-inline.c (optimize_inline_calls): Return TODOs rather than
1547 (tree_function_versioning): Do not allocate dummy struct function.
1548 * tree-inline.h (optimize_inline_calls): Update prototype.
1549 * tree-optimize.c (execute_fixup_cfg): Export.
1550 (pass_fixup_cfg): Remove
1551 (tree_rest_of_compilation): Do not apply inlines.
1552 * tree-flow.h (execute_fixup_cfg): Declare.
1553 * Makefile.in (gt-passes.c): New.
1554 * passes.c: Include gt-passes.h
1555 (init_optimization_passes): New passes.
1556 (nnodes, order): New static vars.
1557 (do_per_function_toporder): New function.
1558 (execute_one_pass): Dump current pass here.
1559 (execute_ipa_pass_list): Don't dump current pass here.
1561 2007-01-16 Janis Johnson <janis187@us.ibm.com>
1563 * config/dfp-bit.c (dfp_compare_op): Return separate value for NaN.
1564 (DFP_NE, DFP_LE, DFP_GE): Return false for NaN.
1566 2007-01-16 David Edelsohn <edelsohn@gnu.org>
1568 * config/rs6000/darwin-ldouble.c: Build file for SOFT_FLOAT.
1569 (strong_alias): Define.
1570 (__gcc_qmul): Provide non-FMA for soft-float.
1576 (__gcc_qunord): New.
1586 * config/rs6000/rs6000.c (rs6000_init_libfuncs): Initialize
1587 soft-float functions.
1588 * config/rs6000/libgcc-ppc-glibc.ver: Version soft-float symbols.
1589 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Do not warn
1590 about long double soft float.
1592 2007-01-16 Dorit Nuzman <dorit@il.ibm.com>
1593 Tehila Meyzels <tehila@il.ibm.com>
1595 * tree-vectorizer.h (is_pattern_stmt_p): New.
1596 * tree-vect-analyze.c (vect_determine_vectorization_factor): Fix
1597 formatting (tabs instead of spaces). Cleanup and clarify setting
1598 of STMT_VINFO_VECTYPE. Call is_pattern_stmt_p.
1599 * tree-vect-transform.c (vect_get_vec_def_for_operand): Fix typo.
1600 (vectorizable_type_demotion): Check that types are integral.
1601 (vectorizable_type_promotion): Likewise.
1602 (vectorizable_store): Fix typo. Eliminate new-line at end of
1605 2007-01-16 Jan Hubicka <jh@suse.cz>
1607 * tree-ssanames.c (release_dead_ssa_names): Remove invalidated
1610 2007-01-15 Eric Christopher <echristo@apple.com>
1612 * ifcvt.c: Include vec.h, vecprim.h.
1613 (check_cond_move_block): New argument regs.
1614 Reorganize. Add registers used to regs.
1615 (cond_move_process_if_block): Use regs set above as
1618 2007-01-15 Eric Christopher <echristo@apple.com>
1620 * config/darwin.h: Update copyright.
1621 (TARGET_OPTION_TRANSLATE_TABLE): Add umbrella.
1622 (LINK_COMMAND_SPEC): Add -u.
1623 (LINK_SPEC): Fix umbrella for above.
1625 2007-01-15 Joseph S. Myers <joseph@codesourcery.com>
1627 * config/soft-fp/op-common.h, config/soft-fp/op-4.h: Update from
1630 2007-01-15 Tom Tromey <tromey@redhat.com>
1632 * doc/sourcebuild.texi (libgcj Tests): Don't mention jacks.
1633 * doc/install.texi (Testing): Don't mention jacks.
1634 (Configuration): Document --enable-java-maintainer-mode. Move
1635 --with-java-home to libgcj-specific section. Document
1637 (Prerequisites): Mention --enable-java-maintainer-mode, ecj1.
1639 2007-01-15 Jan Hubicka <jh@suse.cz>
1641 * tree-ssa-dce.c (DCE_TODOs): New.
1642 (propagate_necessity): Return if something changed.
1643 (eliminate_unnecessary_stmts): Likewise.
1644 (perform_tree_ssa_dce): Return TODO flags when needed.
1645 (pass_dce, pass_dce_loop, pass_cd_dce): Remove TODO flags.
1647 2007-01-15 Uros Bizjak <ubizjak@gmail.com>
1649 * config/i386/i386.md (fyl2xxf3_i387): Rename from fyl2x_xf3.
1650 (fyl2x_extend<mode>xf3_i387): New insn pattern.
1651 (log<mode>2): Rename from logsf2 and logdf2 and macroize insn
1652 patterns using X87MODEF12 mode macro. Extend operand 1
1653 to XFmode. Use SSE_FLOAT_MODE_P to disable patterns for SSE math.
1654 (log10<mode>2): Ditto.
1655 (log2<mode>2): Ditto.
1656 (log1p<mode>2): Ditto.
1657 (logb<mode>2): Ditto.
1658 (fyl2xp1xf3_i387): Rename from fyl2xp1_xf3.
1659 (fyl2xp1_extend<mode>xf3_i387): New insn pattern.
1660 (*fxtractxf3_i387): Rename from *fxtractxf3.
1661 (fxtract_extend<mode>xf3_i387): New insn pattern.
1662 (ilogbsi2): Use match_dup 3, not match_operand:XF 3.
1663 * config/i386/i386.c (ix86_emit_i387_log1p): Use gen_fyl2xp1xf3_i387()
1664 and gen_fyl2xxf3_i387().
1666 2007-01-14 Zdenek Dvorak <dvorakz@suse.cz>
1668 * loop-unswitch.c (unswitch_loop): Do not call fix_loop_placement.
1669 * cfgloopmanip.c (fix_loop_placement): Made static. Use
1670 get_loop_exit_edges. Changed return type to bool.
1671 * cfgloop.h (fix_loop_placement): Declaration removed.
1673 2007-01-14 Dorit Nuzman <dorit@il.ibm.com>
1675 * param.h (MIN_VECT_LOOP_BOUND): New.
1676 * params.def (MIN_VECT_LOOP_BOUND): New.
1677 * tree-vectorizer.c (slpeel_tree_peel_loop_to_edge): Takes another
1678 argument - minimum threshold for number of iterations.
1679 * tree-vectorizer.h (slpeel_tree_peel_loop_to_edge): Add another
1680 argument to declaration.
1681 * tree-vect-analyze.c (vect_analyze_operations): Check value of
1682 MIN_VECT_LOOP_BOUND.
1683 * tree-vect-transform.c (vect_do_peeling_for_loop_bound): Call
1684 slpeel_tree_peel_loop_to_edge with additional argument.
1685 (vect_do_peeling_for_alignment): Likewise.
1686 * doc/invoke.texi (min-vect-loop-bound): Document new param option.
1688 2007-01-14 Uros Bizjak <ubizjak@gmail.com>
1691 * config/i386/i386.c (print_operand) ['z']: Output 'b' for
1694 2007-01-14 Jan Hubicka <jh@suse.cz>
1696 * tree-dfa.c (remove_referenced_var): New function.
1697 * tree-ssa-live.c (remove_unused_locals): Walk referenced vars and
1698 prune referenced vars list too.
1699 * tree-flow.h (remove_referenced_var): Declare.
1701 2007-01-14 Jan Hubicka <jh@suse.cz>
1703 * tree-eh.c (add_stmt_to_eh_region_fn): Do not add call_exprs
1705 (remove_stmt_from_eh_region_fn): Do not remove call_exprs.
1706 (verify_eh_throw_stmt_node, verify_eh_throw_table_statements): Kill.
1707 * except.h (verify_eh_throw_table_statements): Kill prototype.
1708 * cfgexpand.c (expand_gimple_basic_block): Propagate Eh regions
1709 into call exrepssions.
1710 * tree-optimize.c (execute_free_cfg_annotatiosn): Do not call
1712 * tree-cfg.c: Include pointer-set.h.
1713 (verify_node_sharing): Work on pointer set.
1714 (verify_eh_throw_stmt_node): New.
1715 (verify_stmts): Use pointers sets, verify throw_stmt.
1717 2007-01-13 Zdenek Dvorak <dvorakz@suse.cz>
1719 * ipa-reference.c (analyze_function): Consider also addresses taken
1722 2007-01-12 Roger Sayle <roger@eyesopen.com>
1724 * c-typeck.c (null_pointer_constant_p): Replace use of
1725 TREE_CONSTANT_OVERFLOW with TREE_OVERFLOW.
1726 (build_c_cast): Likewise.
1728 2007-01-12 Roger Sayle <roger@eyesopen.com>
1730 * tree.h (force_fit_type_double): Remove unused final argument.
1731 * c-common.c (constant_expression_warning): Replace use of
1732 TREE_CONSTANT_OVERFLOW with TREE_OVERFLOW.
1733 (convert_and_check): Likewise.
1734 (shorten_compare): Update call to force_fit_type_double.
1735 (c_common_truthvalue_conversion) <INTEGER_CST>: Use integer_zerop.
1736 * convert.c (convert_to_pointer): Update call to
1737 force_fit_type_double.
1738 * fold-const.c (force_fit_type_double): Remove overflowed_const
1740 (int_const_binop, fold_convert_const_int_from_int,
1741 fold_convert_const_int_from_real, fold_div_compare,
1742 fold_sign_changed_comparison, fold_unary, fold_negate_const,
1743 fold_abs_const, fold_not_const): Remove the final argument from
1744 calls to force_fit_type_double.
1746 2007-01-12 Andrew Pinski <andrew_pinski@playstation.sony.com>
1748 * configure.ac: Set insn to "nop" for spu-*-* also.
1749 * configure: Regenerate.
1751 2007-01-12 Olga Golovanevsky <olga@il.ibm.com>
1753 * builtins.def : Add BUILT_IN_FREE.
1755 2007-01-12 Jan Hubicka <jh@suse.cz>
1757 PR tree-optimization/30443
1758 * tree-inline.c (tree_function_versioning): Do not optimize when
1759 cloning for inlining.
1761 2007-01-12 Zdenek Dvorak <dvorakz@suse.cz>
1763 * doc/loop.texi: Document recording of loop exits.
1764 * cfgloopmanip.c (loopify, duplicate_loop): Use alloc_loop.
1765 (update_single_exits_after_duplication,
1766 update_single_exit_for_duplicated_loop,
1767 update_single_exit_for_duplicated_loops): Removed.
1768 (duplicate_loop_to_header_edge): Do not call
1769 update_single_exits_after_duplication and
1770 update_single_exit_for_duplicated_loops.
1771 (loop_version): Do not update single_exit information.
1772 (fix_loop_structure): Use record_loop_exits instead of
1773 mark_single_exit_loops.
1774 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update
1775 the lists of loop exits.
1776 * cfghooks.c (redirect_edge_and_branch, redirect_edge_and_branch_force,
1777 split_edge, merge_blocks): Update the lists of loop exits.
1778 * modulo-sched.c (sms_schedule): Pass LOOPS_HAVE_RECORDED_EXITS to
1779 loop_optimizer_init.
1780 * loop-init.c (loop_optimizer_init): Call record_loop_exits instead
1781 of mark_single_exit_loops.
1782 (loop_optimizer_finalize): Call release_recorded_exits.
1783 * tree-ssa-loop.c (tree_loop_optimizer_init): Pass
1784 LOOPS_HAVE_RECORDED_EXITS to loop_optimizer_init.
1785 * tree-vectorizer.c (slpeel_tree_duplicate_loop_to_edge_cfg): Do not
1786 update single exit information.
1787 * lambda-code.c (perfect_nestify): Ditto.
1788 * cfgloop.c (flow_loop_free): Destroy the list of exits of the loop.
1789 (mark_single_exit_loops): Removed.
1790 (alloc_loop, loop_exit_hash, loop_exit_eq, loop_exit_free,
1791 get_exit_descriptions, rescan_loop_exit, record_loop_exits,
1792 dump_recorded_exit, dump_recorded_exits, release_recorded_exits): New
1794 (get_loop_exit_edges, single_exit): Use recorded exit lists.
1795 (add_bb_to_loop, remove_bb_from_loops): Update the lists of loop exits.
1796 (verify_loop_structure): Verify consistency of the exit lists.
1797 (flow_loops_find): Use alloc_loop. Initialize exits hash.
1798 (set_single_exit): Removed.
1799 * cfgloop.h (struct loop_exit): New function.
1800 (struct loop): single_exit_ field replaced by exits field.
1801 (LOOPS_HAVE_MARKED_SINGLE_EXITS): Replaced by LOOPS_HAVE_RECORDED_EXITS.
1802 (struct loops): Added exits hash.
1803 (mark_single_exit_loops, set_single_exit): Declaration removed.
1804 (release_recorded_exits, record_loop_exits, rescan_loop_exit): Declare.
1806 2007-01-12 Richard Sandiford <richard@codesourcery.com>
1808 * doc/invoke.texi: Avoid use of @headitem.
1810 2007-01-12 Richard Sandiford <richard@codesourcery.com>
1812 * cse.c (cse_insn): Move HAVE_CC0 code after declarations.
1814 2007-01-12 Richard Sandiford <richard@codesourcery.com>
1816 * doc/install.texi: Fix m68k-*-* anchor and add m68k-*-* to the
1819 2007-01-12 Nathan Sidwell <nathan@codesourcery.com>
1820 Richard Sandiford <richard@codesourcery.com>
1822 * doc/invoke.texi: Document -mno-bitfield, -mno-rtd and -mno-short.
1823 * config/m68k/m68k.opt: Resort options.
1824 (mbitfield, mrtd, mshort): Remove RejectNegative properties.
1826 2007-01-12 Nathan Sidwell <nathan@codesourcery.com>
1827 Richard Sandiford <richard@codesourcery.com>
1829 * doc/invoke.texi: Document the macros that are defined by
1830 m68k's -mtune and -mhard-float options.
1831 * config/m68k/m68k-protos.h (m68k_cpp_cpu_ident) Declare.
1832 (m68k_cpp_cpu_family): Likewise.
1833 * config/m68k/m68k.h (TARGET_CPU_CPP_BUILTINS): Add a full set
1834 of __ucfv*__ macros. Define __mcffpu__ if generating code for
1835 ColdFire FPUs. Define __mcf_cpu_* and __mcf_family_* macros.
1836 * config/m68k/m68k.c (m68k_cpp_cpu_ident): New function.
1837 (m68k_cpp_cpu_family): Likewise.
1839 2007-01-12 Richard Sandiford <richard@codesourcery.com>
1841 * config/m68k/m68k.h (TARGET_CPU_CPP_BUILTINS): Treat all mc68*
1842 macros besides mc68000 as tuning macros. Use a switch statement
1843 to set them and mcpu32.
1845 2007-01-12 Julian Brown <julian@codesourcery.com>
1847 * config/m68k/m68k.h: Use TARGET_68040 instead of TARGET_68040_ONLY.
1848 (TARGET_68040_ONLY): Rename to...
1849 (TARGET_68040): ...this.
1850 * config/m68k/m68k.c: Use TARGET_68040 instead of TARGET_68040_ONLY.
1851 * config/m68k/m68k.md: Likewise.
1853 2007-01-12 Julian Brown <julian@codesourcery.com>
1854 Nathan Sidwell <nathan@codesourcery.com>
1855 Richard Sandiford <richard@codesourcery.com>
1857 * config.gcc (m680[012]0-*-*, m68k*-*-*): Set m68k_cpu_ident to
1858 the -mcpu= argument associated with the --with-cpu setting.
1859 Define M68K_DEFAULT_TUNE to the default -mtune= option,
1860 if different from the one implied by the -mcpu setting.
1861 Accept --with-cpu=FOO if FOO is listed in m68k-devices.def,
1862 using mcpu=FOO as the default CPU option. Set target_cpu_default2.
1863 * doc/invoke.texi: Mention ColdFire in the introduction to the
1864 m68k options. Document the new -march, -mcpu, -mtune, -mdiv,
1865 -mno-div and -mhard-float options. Make -m68881 a synonym for
1866 -mhard-float. Document the previously-undocumented -m5206e,
1867 -m528x, -m5307 and -m5407 options. Tweak the existing option
1868 documentation for consistency.
1869 * doc/install.texi: Mention new --with-cpu arguments.
1870 * config/m68k/m68k.h (OPTION_DEFAULT_SPECS): Only use the
1871 default CPU if neither -mcpu nor -march are specified.
1872 (ASM_CPU_SPEC): Pass down -mcpu and -march options.
1873 (TARGET_CPU_CPP_BUILTINS): Set __mcfisa*__ macros from
1874 TARGET_ISA*. Set the legacy __mcf*__ cpu macros in the same way,
1875 using m68k_tune to decide between families that implement the
1876 same ISA. Use m68k_tune to set __mcfv4e__.
1877 (FL_BITFIELD, FL_68881, FL_COLDFIRE, FL_CF_HWDIV, FL_CF_MAC)
1878 (FL_CF_EMAC, FL_CF_EMAC_B, FL_CF_USP, FL_CF_FPU, FL_ISA_68000)
1879 (FL_ISA_68010, FL_ISA_68020, FL_ISA_68040, FL_ISA_A, FL_ISA_B)
1880 (FL_ISA_C, FL_ISA_MMU): New macros.
1881 (MASK_COLDFIRE): Delete.
1882 (TARGET_68010, TARGET_68020, TARGET_68040_ONLY, TARGET_COLDFIRE)
1883 (TARGET_ISAB): Redefine in terms of m68k_cpu_flags.
1884 (TARGET_68881, TARGET_COLDFIRE_FPU): Redefine in terms of m68k_fpu.
1885 (TARGET_HARD_FLOAT): Do not define here.
1886 (TARGET_ISAAPLUS, TARGET_ISAC): New macros.
1887 (TUNE_68000): New macro.
1888 (TUNE_68000_10): Redefine in terms of TUNE_68000 and TUNE_68010.
1889 (TUNE_68010, TUNE_68030, TUNE_68040, TUNE_68060, TUNE_CPU32)
1890 (TUNE_CFV2): Redefine in terms of m68k_tune.
1891 (uarch_type, target_device, fpu_type): New enums.
1892 (m68k_cpu, m68k_tune, m68k_fpu, m68k_cpu_flags): Declare.
1893 * config/m68k/m68k.c (TARGET_DEFAULT): Remove MASK_68881.
1894 (FL_FOR_isa_00, FL_FOR_isa_10, FL_FOR_isa_20, FL_FOR_isa_40)
1895 (FL_FOR_isa_cpu32, FL_FOR_isa_a, FL_FOR_isa_aplus, FL_FOR_isa_b)
1896 (FL_FOR_isa_c): New macros.
1897 (m68k_isa): New enum.
1898 (m68k_target_selection): New structure.
1899 (all_devices, all_isas, all_microarchs): New tables.
1900 (m68k_cpu_entry, m68k_arch_entry, m68k_tune_entry, m68k_cpu)
1901 (m68k_tune, m68k_fpu, m68k_cpu_flags): New variables.
1902 (MASK_ALL_CPU_BITS): Delete.
1903 (m68k_find_selection): New function.
1904 (m68k_handle_option): Handle -mcpu=, -march= and -mtune=.
1905 Map the legacy target options to a combination of the new ones.
1906 (override_options): Set m68k_cpu, m68k_tune, m68k_fpu and
1907 m68k_cpu_flags. Handle M68K_DEFAULT_TUNE. Use m68k_cpu_flags
1908 to derive default MASK_BITFIELD, MASK_CF_HWDIV and MASK_HARD_FLOAT
1910 * config/m68k/m68k.opt (m5200, m5206e, m528x, m5307, m5407, mcfv4e)
1911 (m68010, m68020, m68020-40, m68020-60, m68030, m68040): Remove Mask
1913 (m68881, msoft-float): Change mask from 68881 to HARD_FLOAT.
1914 (march=, mcpu=, mdiv, mhard-float, mtune=): New options.
1915 * config/m68k/m68k-devices.def: New file.
1917 2007-01-12 Richard Sandiford <richard@codesourcery.com>
1918 Nathan Sidwell <nathan@codesourcery.com>
1920 * config/m68k/m68k.h (ASM_CPU_SPEC, ASM_SPEC, EXTRA_SPECS)
1921 (SUBTARGET_EXTRA_SPECS): New macros.
1922 * config/m68k/linux.h (ASM_SPEC): Remove CPU flags;
1923 use %(asm_cpu_spec) instead.
1924 * config/m68k/m68k-none.h (ASM_SPEC): Likewise.
1925 * config/m68k/openbsd.h (ASM_SPEC): Likewise.
1926 * config/m68k/netbsd-elf.h (ASM_SPEC): Likewise.
1927 (EXTRA_SPECS): Rename to...
1928 (SUBTARGET_EXTRA_SPECS): ...this.
1930 2007-01-12 Nathan Sidwell <nathan@codesourcery.com>
1931 Richard Sandiford <richard@codesourcery.com>
1932 Julian Brown <julian@codesourcery.com>
1934 * config.gcc (m68k-*-aout*, m68k-*-coff*, m68020-*-elf*, m68k-*-elf*)
1935 (m68k-*-uclinux*, m68k-*-linux*, m68k-*-rtems*): Set default_m68k_cpu
1936 to the configuration's default CPU.
1937 (m68010-*-netbsdelf*, m68k*-*-netbsdelf*, m68k*-*-openbsd*): Likewise.
1938 Remove default masks.
1939 (m680[012]0-*-*): Set the default with_cpu to the first part of
1941 (m68k*-*-*): Set the default with_cpu to m$default_m68k_cpu.
1942 (m68k*-*-linux): Extend the --with-cpu handling to...
1943 (m680[012]0-*-*, m68k*-*-*): ...these configurations. Allow m68000
1944 and m68010. Don't set target_cpu_default2.
1945 * doc/install.texi: Document --with-cpu for m68k.
1946 * config/m68k/m68k.h (OPTION_DEFAULT_SPECS): Define.
1947 * config/m68k/m68k-none.h (TARGET_CPU_DEFAULT, M68K_CPU_m68k)
1948 (M68K_CPU_m68000, M68K_CPU_m68010, M68K_CPU_m68020, M68K_CPU_m68030)
1949 (M68K_CPU_m68040, M68K_CPU_m68302, M68K_CPU_m68332, TARGET_DEFAULT)
1950 (ASM_CPU_DEFAULT_SPEC, CC1_CPU_DEFAULT_SPEC): Delete.
1951 (ASM_SPEC): Remove use of %(asm_cpu_default).
1952 (EXTRA_SPECS, SUBTARGET_EXTRA_SPECS, MULTILIB_DEFAULTS): Delete.
1953 * config/m68k/linux.h (TARGET_DEFAULT): Delete.
1954 (CPP_SPEC): Merge definitions. Do not handle __HAVE_68881__ here.
1955 * config/m68k/netbsd-elf.h (TARGET_OS_CPP_BUILTINS): Define
1956 __HAVE_FPU__ if TARGET_HARD_FLOAT.
1957 (TARGET_DEFAULT): Delete.
1958 (EXTRA_SPECS): Delete cpp_cpu_default_spec, cpp_cpu_spec,
1959 cpp_fpu_spec, asm_default_spec and netbsd_cpp_spec.
1960 (CPP_CPU_SPEC): Delete.
1961 (TARGET_VERSION): Merge definitions, using TARGET_68010 to pick
1962 the appropriate string.
1963 (CPP_CPU_DEFAULT_SPEC, ASM_DEFAULT_SPEC, CPP_FPU_SPEC): Delete.
1964 (CPP_SPEC): Define to NETBSD_CPP_SPEC.
1965 (ASM_SPEC): Don't use %(asm_default_spec).
1966 * config/m68k/m68k.c (TARGET_DEFAULT_TARGET_FLAGS): Remove
1967 TARGET_DEFAULT and add MASK_68881.
1968 * config/m68k/m68k.md: Remove mention of TARGET_DEFAULT from comments.
1970 2007-01-12 Richard Sandiford <richard@codesourcery.com>
1972 * config.gcc (m68010-*-netbsdelf*): Add MASK_68010.
1973 (m68k*-*-netbsdelf*, m68k*-*-openbsd*, m68k*-linux*): Add
1974 MASK_68010 alongside MASK_68020.
1975 * doc/invoke.texi: Document -m68010.
1976 * config/m68k/m68k.opt (m68010): New.
1977 * config/m68k/m68k.h (TARGET_CPU_CPP_BUILTINS): Define mc68010
1979 (TUNE_68010): New macro.
1980 * config/m68k/m68k-none.h (M68K_CPU_m68k, M68K_CPU_m68010)
1981 (M68K_CPU_m68020, M68K_CPU_m68030, M68K_CPU_m68040)
1982 (M68K_CPU_m68332): Add MASK_68010.
1983 * config/m68k/linux.h (TARGET_DEFAULT): Add MASK_68010 to
1984 fallback definition.
1985 * config/m68k/netbsd-elf.h (CPP_CPU_SPEC): Remove now-redundant
1987 * config/m68k/m68k.c (MASK_ALL_CPU_BITS): Add MASK_68010.
1988 (m68k_handle_option): Handle OPT_m68010. Add MASK_68010
1989 to all entries that use MASK_68020.
1990 (output_move_simode_const, output_move_himode, output_move_qimode)
1991 (output_move_stricthi, output_move_strictqi): Use TARGET_68010
1992 instead of TARGET_68020 to select clr behavior. Remove comment
1993 about there being no TARGET_68010.
1994 * config/m68k/m68k.md: Likewise throughout.
1996 2007-01-12 Julian Brown <julian@codesourcery.com>
1998 * config/m68k/m68k.h (TARGET_ISAB): New macro.
1999 * config/m68k/m68k.c: Use TARGET_ISAB rather than TARGET_CFV4.
2000 * config/m68k/m68k.md: Likewise.
2002 2007-01-12 Julian Brown <julian@codesourcery.com>
2004 * config/m68k/m68k.h (LEGITIMATE_INDEX_P, LEGITIMIZE_ADDRESS): Use
2005 TARGET_COLDFIRE_FPU instead of TARGET_CFV4E.
2007 2007-01-12 Julian Brown <julian@codesourcery.com>
2009 * config/m68k/m68k.h (TUNE_68040_60): New macro.
2010 * config/m68k/m68k.c (standard_68881_constant_p): Use it.
2011 * config/m68k/m68k.md: Likewise.
2013 2007-01-12 Julian Brown <julian@codesourcery.com>
2014 Richard Sandiford <richard@codesourcery.com>
2016 * config/m68k/m68k.h (TARGET_CPU_CPP_BUILTINS): Use TUNE_68030
2017 instead of TARGET_68030, TUNE_68040 instead of TARGET_68040,
2018 TUNE_68060 instead of TARGET_68060 and TUNE_CPU32 instead of
2020 (TARGET_CPU32): Rename to...
2021 (TUNE_CPU32): ...this.
2022 (TUNE_68000_10, TUNE_68030, TUNE_68040, TUNE_68060)
2023 (TUNE_CFV2): New macros.
2024 * config/m68k/netbsd-elf.h (LONG_DOUBLE_TYPE_SIZE): Simplify;
2025 remove conditions that are implied by TARGET_68020.
2026 * config/m68k/m68k.c (m68k_output_function_prologue): Use TUNE_68040
2027 instead of TARGET_68040 and TUNE_CPU32 instead of TARGET_CPU32.
2028 (m68k_output_function_epilogue): Likewise.
2029 (m68k_rtx_costs): Likewise. Use TUNE_68060 instead of TARGET_68060
2030 and TUNE_CFV2 instead of TARGET_5200. Use TUNE_68000_10 instead of
2031 "!TARGET_68020 && !TARGET_COLDFIRE" to choose between 68000 and
2032 non-68000 timings. Refactor multiplication and division costs.
2033 (output_addsi3): Use TUNE_68040 instead of TARGET_68040 and
2034 TUNE_CPU32 instead of TARGET_CPU32.
2035 (standard_68881_constant_p): Use TUNE_68040 instead of TARGET_68040
2036 and TUNE_68060 instead of TARGET_68060.
2037 * config/m68k/m68k.md: Use TUNE_68040 instead of TARGET_68040,
2038 TUNE_68060 instead of TARGET_68060, and TUNE_CPU32 instead of
2040 (movsi_const0): Use TUNE_68000_10 rather than "!TARGET_68020
2041 && !TARGET_COLDFIRE" to choose between moveq and clr.
2042 Likewise in the unnamed movsf pattern.
2043 (ashlsi_17_24, lshrsi_17_24): Guard with TUNE_68000_10 rather than
2044 "!TARGET_68020 && !TARGET_COLDFIRE". Likewise the unnamed
2047 2007-01-12 Richard Sandiford <richard@codesourcery.com>
2049 * config/m68k/m68k.h (TARGET_CPU_CPP_BUILTINS): Increase amount
2050 of tabbing before backslashes.
2052 2007-01-11 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
2054 * pa-linux.h (ASM_OUTPUT_INTERNAL_LABEL): Undefine.
2055 * pa.h (ASM_OUTPUT_LABEL): Output colon when using GAS.
2056 (ASM_OUTPUT_INTERNAL_LABEL): Define.
2058 2007-01-11 Zdenek Dvorak <dvorakz@suse.cz>
2060 * tree-ssa-loop-ivopts.c (extract_cond_operands): Split from
2061 find_interesting_uses_cond.
2062 (find_interesting_uses_cond): Use extract_cond_operands.
2063 (rewrite_use_compare): Use extract_cond_operands and
2064 force_gimple_operand_bsi. Do not call update_stmt.
2065 (determine_use_iv_cost_condition): Use extract_cond_operands.
2066 Return cheaper of using original bound and changing the exit bound.
2068 2007-01-11 Zdenek Dvorak <dvorakz@suse.cz>
2070 PR tree-optimization/29516
2071 * tree-ssa-address.c (tree_mem_ref_addr, add_to_parts,
2072 most_expensive_mult_to_index, addr_to_parts,
2073 create_mem_ref, maybe_fold_tmr): Make the type of
2074 fields of TARGET_MEM_REF sizetype.
2075 (move_fixed_address_to_symbol, move_pointer_to_base):
2077 * tree.def (TARGET_MEM_REF): Add comment on types of
2080 2007-01-11 Joseph Myers <joseph@codesourcery.com>
2082 * c-common.c (vector_types_convertible_p): Treat opaque types as
2083 always convertible if they have the same size, but not otherwise.
2085 2007-01-11 Steven Bosscher <steven@gcc.gnu.org>
2087 * ifcvt.c (struct noce_if_info): Add comments to the fields.
2088 Remove the b_unconditional field.
2089 (noce_try_sign_mask): Do not look at b_unconditional.
2090 (noce_process_if_block): Do not use merge_if_blocks. Update
2091 the CFG here. Do not set b_unconditional.
2092 (cond_move_process_if_block): Likewise.
2093 (find_cond_trap): Likewise.
2094 (check_cond_move_block): Require simple jump insns at the end
2097 2007-01-11 Jan Hubicka <jh@suse.cz>
2099 PR tree-optimization/1046
2100 * tree-tailcall.c (suitable_for_tail_call_opt_p): Use TREE_ADDRESSABLE
2101 when alias info is not ready.
2102 (pass_tail_recursion): Do not require aliasing.
2103 * tree-ssa-copyrename.c (pass_rename_ssa_cop): Likewise.
2104 * tree-ssa-ccp.c (pass_ccp, pass_fold_builtins): Likewise.
2105 * tree-ssa-copy.c (pass_copy_prop): Likewise.
2106 * tree-ssa-forwprop.c (pass_forwprop): Likewise.
2107 * tree-ssa-dce.c (pass_dce, pass_dce_loop, pass_cd_dce): Likewise.
2108 * passes.c (init_optimization_passes): Execute rename_ssa_copies,
2109 ccp, forwprop, copy_prop, merge_phi, copy_prop, dce and tail recursion
2111 * tree-ssa-operands.c (add_virtual_operand, get_indirect_ref_operand):
2112 When aliasing is not build, mark statement as volatile.
2114 2007-01-11 Tom Tromey <tromey@redhat.com>
2116 PR preprocessor/15185, PR preprocessor/20989:
2117 * doc/cppopts.texi <-MT>: Update description of algorithm for
2118 computing default target.
2119 <-M, -MD>: Reword "basename" text.
2121 2007-01-11 Roger Sayle <roger@eyesopen.com>
2123 * builtins.c (expand_builtin_pow, expand_builtin_powi,
2124 fold_builtin_cabs, fold_builtin_sqrt, fold_builtin_trunc,
2125 fold_builtin_floor, fold_builtin_ceil, fold_builtin_round,
2126 fold_builtin_int_int_roundingfn, fold_builtin_bitop,
2127 fold_builtin_bswap, real_constp, fold_builtin_pow,
2128 fold_builtin_powi, fold_builtin_signbit, fold_builtin_copysign,
2129 do_mpfr_arg1, do_mpfr_arg2, do_mpfr_arg3, do_mpfr_sincos): Replace
2130 uses of the macro TREE_CONSTANT_OVERFLOW with TREE_OVERFLOW.
2131 * convert.c (convert_to_pointer): Likewise.
2132 * expr.c (highest_pow2_factor, expand_expr_real_1): Likewise.
2133 * fold-const.c (force_fit_type, fold_negate_expr, int_const_binop,
2134 const_binop, fold_convert_const_int_from_int,
2135 fold_convert_const_int_from_real,
2136 fold_convert_const_real_from_real, sign_bit_p,
2137 optimize_minmax_comparison, extract_muldiv_1, fold_div_compare,
2138 fold_sign_changed_comparison, fold_unary, fold_comparison,
2139 fold_binary, multiple_of_p, tree_Expr_non_zero_p,
2140 fold_negate_const, fold_abs_const, fold_not_const): Likewise.
2141 * print-tree.c (print_node_brief, print_node): Likewise.
2142 * stor-layout.c (place_field, layout_type): Likewise.
2143 * tree-chrec.c (keep_cast): Likewise.
2144 * tree.c (build_vector, build_real, build_real_from_int_cst,
2145 build_complex): Likewise.
2147 2007-01-11 Roger Sayle <roger@eyesopen.com>
2149 * tree.h (TREE_CONSTANT_OVERFLOW): Obsolete. For the time being,
2150 treat TREE_CONSTANT_OVERFLOW as a synonym of TREE_OVERFLOW.
2152 2007-01-11 Paolo Bonzini <bonzini@gnu.org>
2154 * configure.ac (strict1_warn): Rename to strict_warn.
2155 (WERROR, --enable-werror, symlink hacks, stage1_cflags,
2156 cc_set_by_configure, quoted_cc_set_by_configure,
2157 stage_prefix_set_by_configure, quoted_stage_prefix_set_by_configure,
2158 all_boot_languages, all_stagestuff): Remove.
2159 (target_list): Remove bootstrap targets.
2160 * Makefile.in (quickstrap): Unconditionally make a synonym of all.
2161 (BOOT_LANGUAGES, STAGE1_CFLAGS, STAGE1_CHECKING,
2162 REMAKEFLAGS, FLAGS_TO_PASS, PREPEND_DOTDOT_TO_RELATIVE_PATHS,
2163 SUBDIR_FLAGS_TO_PASS, WERROR_FLAGS, STRICT2_WARN, LANG_STAGESTUFF,
2164 VOL_FILES, POSTSTAGE1_FLAGS_TO_PASS, STAGE2_FLAGS_TO_PASS,
2165 STAGEPROFILE_FLAGS_TO_PASS, STAGEFEEDBACK_FLAGS_TO_PASS, stage1_build,
2166 stage1_copy, stage2_build, stage2_copy, stageprofile_build,
2167 stageprofile_copy, stage3_build, stage3_copy, stagefeedback_build,
2168 stagefeedback_copy, stage4_build, clean_s1, clean_sw, bootstrap,
2169 bootstrap-lean, bootstrap2, bootstrap2-lean, bootstrap3,
2170 bootstrap3-lean, bootstrap4, bootstrap4-lean, unstage1, unstage2,
2171 unstage3, unstage4, unstageprofile, unstagefeedback, restage, restage2,
2172 restage3, restage4, restageprofile, restagefeedback, bubbleestrap,
2173 cleanstrap, unstrap, restrap, *compare, *compare3, *compare4,
2174 *compare-lean, *compare3-lean, *compare4-lean, stage1-start, stage1,
2175 stage2-start, stage2, stage3-start, stage3, stage4-start, stage4,
2176 stageprofile-start, stageprofile, stagefeedback-start, stagefeedback,
2177 risky-stage1, risky-stage2, risky-stage3, risky-stage4): Remove.
2178 (ORDINARY_FLAGS_TO_PASS): Rename to FLAGS_TO_PASS.
2179 (STAGECOPYSTUFF, STAGEMOVESTUFF): Consolidate into MOSTLYCLEANFILES.
2180 (mostlyclean): Adjust.
2181 (clean, distclean): Don't mention bootstrap stuff.
2182 * configure: Regenerate.
2183 * ada/config-lang.in, cp/config-lang.in, forttran/config-lang.in,
2184 java/config-lang.in, objc/config-lang.in, objcp/config-lang.in,
2185 treelang/config-lang.in (stagestuff): Remove.
2186 * doc/sourcebuild.texi (stage1, stage2, stage3, stage4,
2187 stageprofile, stagefeedback, stagestuff): Remove mention.
2189 2007-01-11 Nick Clifton <nickc@redhat.com>
2191 * config/mcore/predicates.md (mcore_general_movesrc_operand):
2193 (mcore_general_movdst_operand): Do not accept CONST_INTs.
2194 (mcore_arith_K_S_operand): Run the test for the S constraint not
2195 the test for the M constraint.
2196 (mcore_addsub_operand): Do not accept integer values that are
2197 larger than 32 bits.
2198 * config/mcore/mcore.md: Remove unused constraints from split.
2199 (andsi3): Use HOST_WIDE_INT instead of int to hold an INTVAL.
2201 (allocate_stack): Likewise.
2202 * config/mcore/mcore.c (mcore_print_operand): Restrict output of P
2203 operands to 32 bits.
2204 (mcore_const_costs): Use HOST_WIDE_INT instead of int to hold an
2206 (mcore_and_cost, mcore_modify_comparison, const_ok_for_mcore,
2207 mcore_const_ok_for_inline, mcore_const_trick_uses_not,
2208 try_constant_tricks, mcore_num_ones, mcore_num_zeros,
2209 mcore_output_bclri, mcore_output_andn, output_inline_const,
2210 mcore_output_move, mcore_output_movedouble): Likewise.
2211 (mcore_output_cmov): Use CONST_OK_FOR_M and CONST_OK_FOR_N.
2212 (output_inline_const): Likewise.
2213 (output_inline_const): Fix format strings used in sprintf
2215 * config/mcore/mcore-protos.h: Update prototypes for changed
2216 functions in mcore.c.
2217 * config/mcore/mcore.h (CONST_OK_FOR_I): Cast values to
2218 HOST_WIDE_INT and not int.
2219 (CONST_OK_FOR_J, CONST_OK_FOR_K, CONST_OK_FOR_L, CONST_OK_FOR_M,
2220 CONST_OK_FOR_N): Likewise.
2221 (LEGITIMATE_CONSTANT_P): Also check CONSTANT_P.
2222 (GO_IF_LEGITIMATE_INDEX): Use HOST_WIDE_INT instead of int to hold
2225 2007-01-10 Jan Hubicka <jh@suse.cz>
2227 * tree-vrp.c (remove_range_assertions): Release defs.
2228 * tree-ssa-loop-ivopts.c (rmeove_statement): Likewise.
2229 * tree-ssa-dom.c (remove_stmt_or_phi): Likewise.
2231 2007-01-10 Paul Brook <paul@codesourcery.com>
2233 * config/arm/arm.c (arm_rtx_costs_1): Handle mutiply-subtract.
2234 * config/arm/arm.md (mulsi3subsi): New insn.
2236 2007-01-10 Zdenek Dvorak <dvorakz@suse.cz>
2238 * tree-ssa-loop-manip.c (tree_unroll_loop): Make it a wrapper over ...
2239 (tree_transform_and_unroll_loop): New.
2240 * tree-flow.h (transform_callback, tree_transform_and_unroll_loop):
2243 2007-01-10 Robert Kennedy <jimbob@google.com>
2245 * fold-const.c (fold_comparison): Fold comparisons like (x *
2246 1000 < 0) to (x < 0).
2248 2007-01-10 Ian Lance Taylor <iant@google.com>
2250 * tree-pretty-print.c (dump_generic_node): Print parentheses when
2251 operands have the same priority.
2253 2007-01-10 Tom Tromey <tromey@redhat.com>
2255 * fold-const.c (fold_truthop): Don't check can_use_bit_fields_p.
2256 (fold_binary): Likewise.
2257 * langhooks.c (lhd_can_use_bit_fields_p): Removed.
2258 * langhooks-def.h (lhd_can_use_bit_fields_p): Removed.
2259 (LANG_HOOKS_CAN_USE_BIT_FIELDS_P): Removed.
2260 (LANG_HOOKS_INITIALIZER): Remove LANG_HOOKS_CAN_USE_BIT_FIELDS_P.
2261 * langhooks.h (struct lang_hooks): Removed field
2262 'can_use_bit_fields_p'.
2264 2007-01-10 Ralf Corsépius <ralf.corsepius@rtems.org>
2266 * config/bfin/t-bfin, config/bfin/t-bfin-elf: Remove GCC_CFLAGS.
2268 2007-01-10 Razya Ladelsky <razya@il.ibm.com>
2270 * function.c (get_last_funcdef_no): New function.
2271 * function.h (get_last_funcdef_no): Declare.
2272 * tree-inline.c (initialize_cfun): Add initialization.
2273 (tree_function_versioning): Cleanup.
2275 2007-01-10 Jan Hubicka <jh@suse.cz>
2277 * tree-inline.c (setup_one_parameter): Do not propagate into abnormal
2280 2007-01-10 Sa Liu <saliu@de.ibm.com>
2281 Ben Elliston <bje@au.ibm.com>
2283 * spu.h (STACK_SAVE_AREA): Use VOIDmode for SAVE_FUNCTION, SImode
2284 for SAVE_NONLOCAL and Pmode for any other save level.
2285 * spu-protos.h (spu_restore_stack_block): Declare.
2286 * spu.md (save_stack_block): Remove.
2287 (restore_stack_block): Call spu_restore_stack_block.
2288 * spu.c (spu_restore_stack_block): New function.
2289 (spu_expand_epilogue): Remove old comment.
2291 2007-01-09 Zdenek Dvorak <dvorakz@suse.cz>
2293 PR tree-optimization/30322
2294 * tree-ssa-loop-ivopts.c (fold_affine_expr, iv_value): Removed.
2295 (cand_value_at): Return the value as aff_tree.
2296 (may_eliminate_iv): Convert the bound from aff_tree to tree.
2297 * tree-affine.c (aff_combination_add_cst, aff_combination_add_product,
2298 aff_combination_mult): New functions.
2299 (aff_combination_add): Use aff_combination_add_cst.
2300 (aff_combination_convert): Allow conversions to a wider type.
2301 (tree_to_aff_combination): Handle BIT_NOT_EXPR.
2302 * tree-affine.h (aff_combination_mult): Declare.
2304 2007-01-09 Carlos O'Donell <carlos@codesourcery.com>
2306 * doc/tm.texi: Update documentation to reflect reality of exec
2307 and start file search behaviours. Update copyright year.
2308 * doc/invoke.texi: Explain how GCC_EXEC_PREFIX is used to find
2309 header file directories.
2311 2007-01-09 Uros Bizjak <ubizjak@gmail.com>
2313 * config/i386/i386.md (*sinxf2): Rename to *sinxf2_i387.
2314 (*cosxf2): Rename to cosxf2_i387.
2315 (*sindf2, *sinsf2): Extend operand 1 to XFmode. Macroize patterns
2316 using X87MODEF12 mode macro. Rename patterns to
2317 *sin_extend<mode>xf2_i387. Use SSE_FLOAT_MODE_P to disable patterns
2319 (*cosdf2, *cossf2): Ditto.
2320 (sincosdf3, sincossf3): Ditto. Rewrite corresponding splitters
2321 to match extended input operands.
2322 (sincos<mode>3): New expander.
2323 (*sinextendsfdf2, *cosextendsfdf2, *sincosextendsfdf3): Remove
2324 insn patterns and corresponding splitters.
2326 2007-01-09 Kaz Kojima <kkojima@gcc.gnu.org>
2328 * config/sh/t-linux (TARGET_LIBGCC2_CFLAGS): Delete.
2329 (SHLIB_MAPFILES, SHLIB_LINK, SHLIB_INSTALL): Likewise.
2331 2007-01-09 Nicolas Pitre <nico@cam.org>
2334 * arm/ieee754-df.S (Lad_s): Also test the low word of X for zero.
2336 2007-01-08 Geoffrey Keating <geoffk@apple.com>
2338 * target.h (struct gcc_target): New field library_rtti_comdat.
2339 * target-def.h (TARGET_CXX_LIBRARY_RTTI_COMDAT): New.
2340 (TARGET_CXX): Add TARGET_CXX_LIBRARY_RTTI_COMDAT.
2341 * doc/tm.texi (C++ ABI): Document TARGET_CXX_LIBRARY_RTTI_COMDAT.
2342 * config/darwin.h (TARGET_CXX_LIBRARY_RTTI_COMDAT): Define.
2344 2007-01-08 Geoffrey Keating <geoffk@apple.com>
2346 * doc/invoke.texi (Optimize Options): Correct description of -O0.
2348 2007-01-08 Richard Guenther <rguenther@suse.de>
2350 * tree.h (force_fit_type_double): Export.
2351 (force_fit_type): Remove.
2352 * fold-const.c (force_fit_type_double): New function.
2353 (force_fit_type): Remove.
2354 (int_const_binop): Use it.
2355 (fold_convert_const_int_from_int): Likewise.
2356 (fold_convert_const_int_from_real): Likewise.
2357 (fold_div_compare): Likewise.
2358 (fold_sign_changed_comparison): Likewise.
2359 (fold_unary): Likewise.
2360 (fold_negate_const): Likewise.
2361 (fold_abs_const): Likewise.
2362 (fold_not_const): Likewise.
2363 * c-common.c (shorten_compare): Use force_fit_type_double.
2364 * convert.c (convert_to_pointer): Likewise.
2366 2007-01-08 Richard Guenther <rguenther@suse.de>
2368 * tree.h (build_int_cst_wide_type): Export.
2369 * tree.c (build_int_cst_wide_type): New function.
2370 (build_int_cst_wide): Fix comment.
2371 * builtins.c (fold_builtin_object_size): Use build_int_cst
2372 to build -1 or 0 of the correct type. Use fit_double_type
2373 to check for overflow.
2374 * fold-const.c (optimize_bit_field_compare): Use build_int_cst_type
2376 (decode_field_reference): Likewise.
2377 (all_ones_mask_p): Likewise.
2378 (native_interpret_int): Use build_int_cst_wide_type.
2379 (fold_binary): Use build_int_cst_type to build an all-ones
2381 * stor-layout.c (set_sizetype): Use build_int_cst_wide_type.
2383 2007-01-08 Daniel Jacobowitz <dan@codesourcery.com>
2385 * config/pa/t-pa64 (libgcc_stub.a): Use $(T).
2387 2007-01-09 Ben Elliston <bje@au.ibm.com>
2389 * genautomata.c (STATS_OPTION): New option.
2390 (stats_flag): New flag.
2391 (gen_automata_option): Handle it.
2392 (initiate_automaton_gen): Ditto.
2393 (write_automata): Output statistics only if stats_flag is
2394 set. Likewise, output time statistics only if time_flag is set.
2395 * doc/md.texi (Processor pipeline description): Document new flag.
2397 2007-01-08 Richard Guenther <rguenther@suse.de>
2399 * builtins.c (fold_builtin_int_roundingfn): Use fit_double_type.
2400 * tree.c (build_int_cst_type): Likewise.
2401 (size_in_bytes): Don't call force_fit_type on the result.
2402 (int_fits_type_p): Use fit_double_type.
2403 * fold-const.c (fit_double_type): New function.
2404 (force_fit_type): Use it.
2405 * tree.h (fit_double_type): Export.
2407 2007-01-08 Jan Hubicka <jh@suse.cz>
2409 * tree-vectorizer.c (gate_increase_alignment): Fix return type.
2410 * ipa.c (function_and_variable_visibility): Fix return type.
2412 2007-01-08 Richard Guenther <rguenther@suse.de>
2414 * tree-ssa-ccp.c (maybe_fold_offset_to_array_ref): Use type
2415 of offset to build the index.
2416 * tree-pretty-print.c (dump_generic_node): Don't build negated
2417 const just for printing.
2418 * c-pretty-print.c (pp_c_integer_constant): Likewise.
2419 * builtins.c (fold_builtin_int_roundingfn): Check if result
2420 fits the type by using force_fit_type and comparing the result.
2421 * predict.c (predict_loops): Use compare_tree_int for comparison.
2422 * tree.c (build_int_cst): Fall back to integer_type_node for
2424 (build_int_cst_wide): Assert type is non-null.
2426 2007-01-08 Roberto Costa <roberto.costa@st.com>
2428 * tree-vrp.c (extract_range_from_cond_expr): New.
2429 (extract_range_from_expr): Handle COND_EXPR nodes used as expressions.
2430 * tree-ssa-ccp.c (get_maxval_strlen): Handle COND_EXPR nodes used
2432 (fold_stmt): Bug fix, avoid infinite recursion when folding COND_EXPRs.
2433 * tree-ssa-forwprop.c (simplify_cond, forward_propagate_into_cond,
2434 tree_ssa_forward_propagate_single_use_vars): Handle COND_EXPR nodes
2435 used as expressions.
2436 * tree-object-size.c (cond_expr_object_size): New.
2437 (collect_object_sizes_for): Handle COND_EXPR nodes used as expressions.
2439 2007-01-08 Jan Hubicka <jh@suse.cz>
2441 * tree-ssa-forwprop.c (forward_propagate_into_cond,
2442 tree_ssa_forward_propagate_single_use_va): Release defs of propagated
2445 2007-01-08 Richard Guenther <rguenther@suse.de>
2447 PR tree-optimization/23603
2448 * tree-vrp.c (set_value_range_to_truthvalue): New function.
2449 (extract_range_from_binary): Fall back to truthvalue instead of
2450 varying for TRUTH_*_EXPR.
2451 (extract_range_from_comparison): Fall back to truthvalue instead of
2453 (vrp_visit_phi_node): Don't adjust new range bounds to +INF/-INF
2454 if all visited PHI values were constant.
2456 2007-01-08 Jan Hubicka <jh@suse.cz>
2458 * cgraphunit.c (cgraph_process_new_functions): Reset reachable flag.
2459 (cgraph_analyze_function): break out from ...
2460 (cgraph_finalize_compilation_unit): ... here.
2461 (cgraph_expand_function): Remove forgoten commented out line.
2462 (cgraph_optimize): Analyze functions.
2464 2007-01-08 Jan Hubicka <jh@suse.cz>
2466 * tree-pas.h (TODO_remove_function): New flag.
2467 (TODO_update*): Renumber.
2468 (pass_ipa_increase_alignment,
2469 pass_ipa_function_and_variable_visibility): New passes.
2470 * cgraphunit.c (cgraph_increase_alignment): Move to tree-vectorizer.c
2471 (cgraph_function_and_variable_visibility): Move to ipa.c
2472 (cgraph_optimize): Don't call cgraph_function_and_variable_visibility,
2473 cgraph_increase_alignment.
2474 * ipa-inline.c (cgraph_decide_inlining): Don't push timevar.
2475 (cgraph_decide_inlining_incrementally): Push TV_INTEGRATION before
2476 calling tree-inline.
2477 (cgraph_early_inlining): Do not call cgraph_remove_unreachable_nodes.
2478 (pass_ipa_inline, pass_early_ipa_inlining): Set TODO_remove_functions
2479 * tree-vectorizer.c (increase_alignment): Move here from cgraphunit.c
2480 (gate_increase_alignment): New function.
2481 (pass_ipa_increase_alignment): New pass.
2482 * ipa.c: Inline tree-pass.h and timevar.h
2483 (function_and_variable_visibility): Move here from cgraphunit.c
2484 * tree-optimize.c (pass_early_local_passes): Add TODO_remove_functions.
2485 * passes.c (init_optimization_passes): Add the two new passes.
2486 (execute_todo): Handle cgraph_remove_functions.
2488 2007-01-08 Nick Clifton <nickc@redhat.com>
2490 * config/frv/predicates.md (reg_or_0_operand): Accept
2493 2007-01-08 Ralf Corsépius <ralf.corsepius@rtems.org>
2495 * config/bfin/rtems.h, config/bfin/t-rtems: New.
2496 * config.gcc: Add bfin*-rtems*.
2498 2007-01-08 Mark Shinwell <shinwell@codesourcery.com>
2500 * c.opt: Add -flax-vector-conversions.
2501 * c-typeck.c (convert_for_assignment): Pass flag to
2502 vector_types_convertible_p to allow emission of note.
2503 (digest_init): Likewise.
2504 * c-opts.c: Handle -flax-vector-conversions.
2505 * c-common.c (flag_lax_vector_conversions): New.
2506 (vector_types_convertible_p): Unless -flax-vector conversions
2507 has been passed, disallow conversions between vectors with
2508 differing numbers of subparts and/or element types. If such
2509 a conversion is disallowed, possibly emit a note on the first
2510 occasion only to inform the user of -flax-vector-conversions.
2511 The new last argument specifies this.
2512 * c-common.h (flag_lax_vector_conversions): New.
2513 (vector_types_convertible_p): Add extra argument.
2514 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use
2515 char_type_node for V*QI type vectors.
2516 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
2517 Update to satisfy new typechecking rules.
2518 * config/rs6000/altivec.h (vec_cmple): Use vec_cmpge, for both
2520 * doc/invoke.texi (C Dialect Options): Document
2521 -flax-vector-conversions.
2523 2007-01-08 Mark Shinwell <shinwell@codesourcery.com>
2525 PR tree-optimization/29877
2526 * tree-ssa-ter.c (is_replaceable_p): Deem assignments with
2527 a register variable on the RHS to not be replaceable.
2529 2007-01-08 Chen Liqin <liqin@sunnorth.com.cn>
2530 * config/score/t-score-elf (MULTILIB_OPTIONS): Change.
2531 * config/score/predicates.md (const_uimm5, sr0_operand, const_simm12,
2532 const_simm15, const_pow2, const_npow2): Added.
2533 * config/score/misc.md (insv, extv, extzv, movmemsi,
2534 move_lbu_a/b, mov_lhu_a/b etc): Added and fix some bug.
2535 * config/score/score.c (score_address_cost, score_select_cc_mode):
2537 Change CONST_OK_FOR_LETTER_P/EXTRA_CONSTRAINT define.
2538 Update score_rtx_costs for MACRO TARGET_RTX_COSTS.
2539 Update score_print_operand.
2540 * config/score/score.h (DATA_ALIGNMENT, SELECT_CC_MODE): Added.
2541 Adjust register allocate order and update some macro define.
2542 * config/score/score-mdaux.c (mdx_unaligned_load, mdx_unsigned_store,
2543 mdx_block_move_straight, mdx_block_move_loop_head,
2544 mdx_block_move_loop_body, mdx_block_move_loop_foot, mdx_block_move_loop,
2545 mdx_block_move): Added.
2546 (mdx_movsicc, mdp_select_add_imm, mdp_select, mds_zero_extract_andi,
2547 mdp_limm): Updated and fix some bug and typo.
2548 * config/score/score.md (movqi/hi/si, add/sub/zero/ext): Updated.
2549 (movsf, movdf, doloop_end): Added.
2551 2007-01-08 Kazu Hirata <kazu@codesourcery.com>
2553 * config/arm/arm.c, config/arm/arm.h, config/arm/arm.md,
2554 config/arm/thumb2.md: Fix comment typos.
2555 * doc/extend.texi: Fix a typo.
2557 2007-01-07 Eric Christopher <echristo@apple.com>
2559 * configure.ac: Check for __stack_chk_fail for darwin.
2560 * configure: Regenerate.
2562 2007-01-07 Richard Guenther <rguenther@suse.de>
2564 * tree-vrp.c (extract_range_from_assert): CSE calls to
2565 compare_values where possible.
2566 (extract_range_from_unary_expr): Likewise.
2568 2007-01-07 Anatoly Sokolov <aesok@post.ru>
2570 * config/avr/avr-protos.h (call_insn_operand): Delete prototype.
2571 * config/avr/avr.c (call_insn_operand): Delete function.
2572 * config/avr/avr.md (*pushqi, *pushhi, *pushsi, *pushsf): Use REG_SP
2573 instead of register number. Use predicates.
2574 * config/avr/predicates.md (const0_operand, reg_or_0_operand,
2575 call_insn_operand): Add.
2577 2007-01-06 Jan Hubicka <jh@suse.cz>
2579 * tree-pass.h (pass_build_cgraph_edges): Declare.
2580 * cgraphunit.c (record_refernece): Move to cgraphbuild.c
2581 (visited_nodes): Remove.
2582 (cgraph_create_edges): Move to cgraphbuild.c; rename to
2583 build_cgrpah_edges; make visited_nodes local.
2584 (cgraph_process_new_functions): DO not call initialize_inline_failed.
2585 (record_references_in_initializer): Move to cgraphbuild.c
2586 (initialize_inline_failed, rebuild_cgraph_edges,
2587 pass_rebuild_cgraph_edges): Move to cgraphbuild.c.
2588 (verify_cgraph_node): Make visited_nodes local.
2589 (cgraph_analyze_function): Do not call cgraph_create_edges and
2590 initialize_inline_failed.
2591 (cgraph_expand_function): Do not call cgraph_lower_function;
2592 assert that function is already lowered.
2593 * Makefile.in (cgraphbuild.o): New.
2594 * passes.c (init_optimization_passes): Add pass_build_cgraph_edges
2595 at the end of lowering passes.
2597 2007-01-06 Steven Bosscher <steven@gcc.gnu.org>
2599 * ifcvt.c (cond_move_convert_if_block): New function, code
2600 factored out from...
2601 (cond_move_process_if_block): ...here. Call the new function
2602 on the THEN and ELSE blocks.
2603 (merge_if_block): Do not copy global_live_at_end, merge_blocks
2604 already takes care of this.
2606 2007-01-05 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
2609 * tree.h (TREE_OVERFLOW_P): New.
2610 * c-typeck.c (parser_build_unary_op): Warn only if result
2611 overflowed and operands did not.
2612 (parser_build_binary_op): Likewise.
2613 (convert_for_assignment): Remove redundant overflow_warning.
2614 * c-common.c (overflow_warning): Don't check or set TREE_OVERFLOW.
2616 2007-01-05 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
2618 * c-typeck.c (store_init_value): Split over two lines to follow
2619 the GNU coding style.
2621 2007-01-05 Benjamin Kosnik <bkoz@redhat.com>
2623 * c-cppbuiltin.c (c_cpp_builtins): __GXX_EXPERIMENTAL_CPP0X__ to
2624 __GXX_EXPERIMENTAL_CXX0X__.
2625 * doc/cpp.texi: Same.
2627 2007-01-05 Richard Guenther <rguenther@suse.de>
2630 * tree.c (get_narrower): Do not construct COMPONENT_REFs
2631 with mismatched types. Instead explicitly build a
2632 conversion NOP_EXPR.
2634 2007-01-05 Ian Lance Taylor <iant@google.com>
2636 * c-common.c (decl_with_nonnull_addr_p): New function.
2637 (c_common_truthvalue_conversion): Call it.
2638 * c-typeck.c (build_binary_op): Likewise.
2639 * c-common.h (decl_with_nonnull_addr_p): Declare.
2641 2007-01-05 Jakub Jelinek <jakub@redhat.com>
2644 * libgcc2.c (__divdc3): Compare c and d against 0.0 instead of
2647 2007-01-05 Joel Brobecker <brobecker@adacore.com>
2649 * doc/install.texi (Final install): Document the fact that
2650 the GNAT runtime should not be stripped.
2652 2007-01-04 Jan Hubicka <jh@suse.cz>
2654 * tree-inline.c (fold_marked_statements): Update operand caches
2655 and EH after folding
2657 2007-01-04 Ian Lance Taylor <iant@google.com>
2659 * c-common.c (check_function_nonnull): Whitespace fix.
2661 2007-01-04 Jan Hubicka <jh@suse.cz>
2663 * tree-optimize.c (execute_fixup_cfg): Correct previously mistakely
2664 comitted older version of patch.
2665 (pass_fixup_cfg): Add TODOs to verify flow and statements, dump
2666 function, celanup cfg and collect garbage.
2668 2007-01-04 Mike Stump <mrs@apple.com>
2670 * Makefile.in (mostlyclean): Don't remove libgcc anymore.
2673 2007-01-04 Eric Christopher <echristo@apple.com>
2675 * libgcc2.c (__bswapsi2): Use SItype.
2676 (__bswapdi2): Use DItype.
2677 * libgcc2.h: Update for above.
2679 2007-01-04 Paul Brook <paul@codesourcery.com>
2681 * config/arm/arm.md (arm_mulsi3, thumb_mulsi3, mulsi3_compare0,
2682 mulsi_compare0_scratch, mulsi3addsi, mulsi3addsi_compare0,
2683 mulsi3addsi_compare0_scratch, mulsidi3adddi, mulsidi3,
2684 umulsidi3, umulsidi3adddi, smulsi3_highpart,
2685 umulsi3_highpart): Make conditional on !arm_arch6.
2686 (arm_mulsi3_v6, thumb_mulsi3_v6, mulsi3_compare0_v6,
2687 mulsi_compare0_scratch_v6, mulsi3addsi_v6, mulsi3addsi_compare0_v6,
2688 mulsi3addsi_compare0_scratch_v6, mulsidi3adddi_v6, mulsidi3_v6,
2689 umulsidi3_v6, umulsidi3adddi_v6, smulsi3_highpart_v6,
2690 umulsi3_highpart_v6): New insns.
2692 2007-01-04 Roger Sayle <roger@eyesopen.com>
2694 * fold-const.c (fold_convert): When casting an expression to void,
2695 fold_ignored_result may discover a GIMPLE_MODIFY_STMT which doesn't
2696 have a type. Instead of attempting to build a NOP_EXPR, return
2697 these "special" trees directly.
2699 2007-01-04 Joseph Myers <joseph@codesourcery.com>
2701 * config/rs6000/rs6000.c (rs6000_rtx_costs): Make adjustment for
2702 MULT inside MINUS as either argument. Use rs6000_cost->dmul -
2703 rs6000_cost->fp not 0 as adjustment for outer NEG.
2705 2007-01-04 Jan Hubicka <jh@suse.cz>
2707 * cgraph.c (cgraph_release_function_body): New function.
2708 (cgraph_remove_node): Use it.
2709 * cgraph.h (cgraph_release_function_body): Declare.
2710 * cgraphunit.c (cgraph_expand_function): Use it.
2711 * ipa.c (cgraph_remove_unreahchable_nodes): Use it.
2712 * tree-ssa.c (delete_tree_ssa): Allow to be called before aliasing
2713 is initialized and while compilation of other function is running.
2714 * tree-optimize.c (execute_free_cfg_annotations): Move code to clear
2715 statement CFG annotations from here to ...
2716 * tree-cfg.c (delete_tree_cfg_annotations): ... here.
2718 2007-01-04 Zdenek Dvorak <dvorakz@suse.cz>
2720 * cfgloop.h (enum li_flags): Make the constants powers of two.
2722 2007-01-04 Jan Hubicka <jh@suse.cz>
2724 * tree-inline.c (copy_bb): Insert new statements to statements_to_fold
2726 (fold_marked_statements): New function.
2727 (optimize_inline_calls, tree_function_versioning): Fold new statements.
2728 * tree-inline.h (copy_body_data): Add statements_to_fold.
2730 2007-01-03 Daniel Jacobowitz <dan@codesourcery.com>
2732 * config.gcc: Mention libgcc/config.host.
2733 * Makefile.in: Update comments mentioning libgcc.
2734 (LIBGCC, INSTALL_LIBGCC, GCC_PARTS, mklibgcc): Delete.
2735 (all.cross, start.encap, rest.encap, rest.cross): Update
2736 dependencies for libgcc move.
2737 (libgcc.mk, LIBGCC_DEPS, libgcov.a, libgcc.a, stmp-multilib)
2738 (clean-target, clean-target-libgcc): Delete.
2739 (srcdirify, GCC_EXTRA_PARTS): New macros.
2740 (libgcc-support, libgcc.mvars): New rules.
2741 (distclean): Remove mention of mklibgcc.
2742 (install): Don't reference INSTALL_LIBGCC.
2743 (install-common): Don't reference EXTRA_PARTS.
2744 (install-libgcc, install-multilib): Delete rules.
2745 * mklibgcc.in: Delete file.
2746 * doc/configfiles.texi: Don't mention mklibgcc.
2748 * config/i386/t-darwin (SHLIB_VERPFX): Delete (moved to libgcc).
2749 * config/i386/t-darwin64 (SHLIB_VERPFX): Likewise.
2750 * config/rs6000/t-darwin (SHLIB_VERPFX): Likewise.
2751 * config/rs6000/t-ppccomm (TARGET_LIBGCC2_CFLAGS, SHLIB_MAPFILES)
2752 (mklibgcc, ldblspecs): Likewise.
2754 * config/i386/t-nwld (libgcc.def, libc.def, libpcre.def)
2755 (posixpre.def): Use $(T).
2756 (SHLIB_EXT, SHLIB_NAME, SHLIB_SLIBDIR_QUAL, SHLIB_DEF, SHLIB_MAP)
2757 (SHLIB_SRC, SHLIB_INSTALL): Delete.
2758 (SHLIB_LINK): Make dummy.
2759 * config/t-slibgcc-darwin: Delete contents except for dummy SHLIB_LINK.
2761 * config/frv/t-linux (EXTRA_MULTILIB_PARTS): Clear.
2763 * config/alpha/t-crtfm: Use $(T) in rules for EXTRA_PARTS.
2764 * config/alpha/t-vms, config/alpha/t-vms64, config/fr30/t-fr30,
2765 config/i386/t-rtems-i386, config/ia64/t-ia64, config/rs6000/t-beos,
2766 config/rs6000/t-newas, config/sparc/t-elf: Likewise.
2768 * configure.ac (all_outputs): Remove mklibgcc.
2769 * configure: Regenerated.
2771 2007-01-03 Josh Conner <jconner@apple.com>
2774 * calls.c (compute_argument_addresses): Set stack and stack_slot
2775 for partial args, too.
2776 (store_one_arg): Use locate.size.constant for the size when
2777 generating a save_area.
2779 2007-01-03 Robert Kennedy <jimbob@google.com>
2781 * tree-cfg.c (tree_merge_blocks): Release SSA_NAME phi results
2782 whose definitions are deleted due to basic block merging.
2784 2007-01-03 Paul Brook <paul@codesourcery.com>
2787 * config/arm/arm.c (output_return_instruction): Pop PC in interrupt
2789 (use_return_insn): Return 0 for Thumb interrupt functions.
2790 (print_multi_reg): Add rfe argument for IRQ returns.
2791 (arm_output_epilogue): Pop interrupt return address directly into PC.
2792 (arm_expand_prologue): Only adjust IRQ return address in Arm mode.
2794 2007-01-03 Paul Brook <paul@codesourcery.com>
2796 Merge from sourcerygxx-4_1.
2797 * config/arm/thumb2.md: New file.
2798 * config/arm/elf.h (JUMP_TABLES_IN_TEXT_SECTION): Return True for
2800 * config/arm/coff.h (JUMP_TABLES_IN_TEXT_SECTION): Ditto.
2801 * config/arm/aout.h (ASM_OUTPUT_ADDR_VEC_ELT): Add !Thumb-2 assertion.
2802 (ASM_OUTPUT_ADDR_DIFF_ELT): Output Thumb-2 jump tables.
2803 * config/arm/aof.h (ASM_OUTPUT_ADDR_DIFF_ELT): Output Thumb-2 jump
2805 (ASM_OUTPUT_ADDR_VEC_ELT): Add !Thumb-2 assertion.
2806 * config/arm/ieee754-df.S: Use macros for Thumb-2/Unified asm
2808 * config/arm/ieee754-sf.S: Ditto.
2809 * config/arm/arm.c (thumb_base_register_rtx_p): Rename...
2810 (thumb1_base_register_rtx_p): ... to this.
2811 (thumb_index_register_rtx_p): Rename...
2812 (thumb1_index_register_rtx_p): ... to this.
2813 (thumb_output_function_prologue): Rename...
2814 (thumb1_output_function_prologue): ... to this.
2815 (thumb_legitimate_address_p): Rename...
2816 (thumb1_legitimate_address_p): ... to this.
2817 (thumb_rtx_costs): Rename...
2818 (thumb1_rtx_costs): ... to this.
2819 (thumb_compute_save_reg_mask): Rename...
2820 (thumb1_compute_save_reg_mask): ... to this.
2821 (thumb_final_prescan_insn): Rename...
2822 (thumb1_final_prescan_insn): ... to this.
2823 (thumb_expand_epilogue): Rename...
2824 (thumb1_expand_epilogue): ... to this.
2825 (arm_unwind_emit_stm): Rename...
2826 (arm_unwind_emit_sequence): ... to this.
2827 (thumb2_legitimate_index_p, thumb2_legitimate_address_p,
2828 thumb1_compute_save_reg_mask, arm_dwarf_handle_frame_unspec,
2829 thumb2_index_mul_operand, output_move_vfp, arm_shift_nmem,
2830 arm_save_coproc_regs, thumb_set_frame_pointer, arm_print_condition,
2831 thumb2_final_prescan_insn, thumb2_asm_output_opcode, arm_output_shift,
2832 thumb2_output_casesi): New functions.
2833 (TARGET_DWARF_HANDLE_FRAME_UNSPEC): Define.
2834 (FL_THUMB2, FL_NOTM, FL_DIV, FL_FOR_ARCH6T2, FL_FOR_ARCH7,
2835 FL_FOR_ARCH7A, FL_FOR_ARCH7R, FL_FOR_ARCH7M, ARM_LSL_NAME,
2836 THUMB2_WORK_REGS): Define.
2837 (arm_arch_notm, arm_arch_thumb2, arm_arch_hwdiv, arm_condexec_count,
2838 arm_condexec_mask, arm_condexec_masklen)): New variables.
2839 (all_architectures): Add armv6t2, armv7, armv7a, armv7r and armv7m.
2840 (arm_override_options): Check new CPU capabilities.
2841 Set new architecture flag variables.
2842 (arm_isr_value): Handle v7m interrupt functions.
2843 (user_return_insn): Return 0 for v7m interrupt functions. Handle
2845 (const_ok_for_arm): Handle Thumb-2 constants.
2846 (arm_gen_constant): Ditto. Use movw when available.
2847 (arm_function_ok_for_sibcall): Return false for v7m interrupt
2849 (legitimize_pic_address, arm_call_tls_get_addr): Handle Thumb-2.
2850 (thumb_find_work_register, arm_load_pic_register,
2851 legitimize_tls_address, arm_address_cost, load_multiple_sequence,
2852 emit_ldm_seq, emit_stm_seq, arm_select_cc_mode, get_jump_table_size,
2853 print_multi_reg, output_mov_long_double_fpa_from_arm,
2854 output_mov_long_double_arm_from_fpa, output_mov_double_fpa_from_arm,
2855 output_mov_double_fpa_from_arm, output_move_double,
2856 arm_compute_save_reg_mask, arm_compute_save_reg0_reg12_mask,
2857 output_return_instruction, arm_output_function_prologue,
2858 arm_output_epilogue, arm_get_frame_offsets, arm_regno_class,
2859 arm_output_mi_thunk, thumb_set_return_address): Ditto.
2860 (arm_expand_prologue): Handle Thumb-2. Use arm_save_coproc_regs.
2861 (arm_coproc_mem_operand): Allow POST_INC/PRE_DEC.
2862 (arithmetic_instr, shift_op): Use arm_shift_nmem.
2863 (arm_print_operand): Use arm_print_condition. Handle '(', ')', '.',
2865 (arm_final_prescan_insn): Use extract_constrain_insn_cached.
2866 (thumb_expand_prologue): Use thumb_set_frame_pointer.
2867 (arm_file_start): Output directive for unified syntax.
2868 (arm_unwind_emit_set): Handle stack alignment instruction.
2869 * config/arm/lib1funcs.asm: Remove default for __ARM_ARCH__.
2870 Add v6t2, v7, v7a, v7r and v7m.
2871 (RETLDM): Add Thumb-2 code.
2872 (do_it, shift1, do_push, do_pop, COND, THUMB_SYNTAX): New macros.
2873 * config/arm/arm.h (TARGET_CPU_CPP_BUILTINS): Define __thumb2__.
2874 (TARGET_THUMB1, TARGET_32BIT, TARGET_THUMB2, TARGET_DSP_MULTIPLY,
2875 TARGET_INT_SIMD, TARGET_UNIFIED_ASM, ARM_FT_STACKALIGN, IS_STACKALIGN,
2876 THUMB2_TRAMPOLINE_TEMPLATE, TRAMPOLINE_ADJUST_ADDRESS,
2877 ASM_OUTPUT_OPCODE, THUMB2_GO_IF_LEGITIMATE_ADDRESS,
2878 THUMB2_LEGITIMIZE_ADDRESS, CASE_VECTOR_PC_RELATIVE,
2879 CASE_VECTOR_SHORTEN_MODE, ADDR_VEC_ALIGN, ASM_OUTPUT_CASE_END,
2880 ADJUST_INSN_LENGTH): Define.
2881 (TARGET_REALLY_IWMMXT, TARGET_IWMMXT_ABI, CONDITIONAL_REGISTER_USAGE,
2882 STATIC_CHAIN_REGNUM, HARD_REGNO_NREGS, INDEX_REG_CLASS,
2883 BASE_REG_CLASS, MODE_BASE_REG_CLASS, SMALL_REGISTER_CLASSES,
2884 PREFERRED_RELOAD_CLASS, SECONDARY_OUTPUT_RELOAD_CLASS,
2885 SECONDARY_INPUT_RELOAD_CLASS, LIBCALL_VALUE, FUNCTION_VALUE_REGNO_P,
2886 TRAMPOLINE_SIZE, INITIALIZE_TRAMPOLINE, HAVE_PRE_INCREMENT,
2887 HAVE_POST_DECREMENT, HAVE_PRE_DECREMENT, HAVE_PRE_MODIFY_DISP,
2888 HAVE_POST_MODIFY_DISP, HAVE_PRE_MODIFY_REG, HAVE_POST_MODIFY_REG,
2889 REGNO_MODE_OK_FOR_BASE_P, LEGITIMATE_CONSTANT_P,
2890 REG_MODE_OK_FOR_BASE_P, REG_OK_FOR_INDEX_P, GO_IF_LEGITIMATE_ADDRESS,
2891 LEGITIMIZE_ADDRESS, THUMB2_LEGITIMIZE_ADDRESS,
2892 GO_IF_MODE_DEPENDENT_ADDRESS, MEMORY_MOVE_COST, BRANCH_COST,
2893 ASM_APP_OFF, ASM_OUTPUT_CASE_LABEL, ARM_DECLARE_FUNCTION_NAME,
2894 FINAL_PRESCAN_INSN, PRINT_OPERAND_PUNCT_VALID_P,
2895 PRINT_OPERAND_ADDRESS): Adjust for Thumb-2.
2896 (arm_arch_notm, arm_arch_thumb2, arm_arch_hwdiv): New declarations.
2897 * config/arm/arm-cores.def: Add arm1156t2-s, cortex-a8, cortex-r4 and
2899 * config/arm/arm-tune.md: Regenerate.
2900 * config/arm/arm-protos.h: Update prototypes.
2901 * config/arm/vfp.md: Enable patterns for Thumb-2.
2902 (arm_movsi_vfp): Add movw alternative. Use output_move_vfp.
2903 (arm_movdi_vfp, movsf_vfp, movdf_vfp): Use output_move_vfp.
2904 (thumb2_movsi_vfp, thumb2_movdi_vfp, thumb2_movsf_vfp,
2905 thumb2_movdf_vfp, thumb2_movsfcc_vfp, thumb2_movdfcc_vfp): New.
2906 * config/arm/libunwind.S: Add Thumb-2 code.
2907 * config/arm/constraints.md: Update include Thumb-2.
2908 * config/arm/ieee754-sf.S: Add Thumb-2/Unified asm support.
2909 * config/arm/ieee754-df.S: Ditto.
2910 * config/arm/bpabi.S: Ditto.
2911 * config/arm/t-arm (MD_INCLUDES): Add thumb2.md.
2912 * config/arm/predicates.md (low_register_operand,
2913 low_reg_or_int_operand, thumb_16bit_operator): New.
2914 (thumb_cmp_operand, thumb_cmpneg_operand): Rename...
2915 (thumb1_cmp_operand, thumb1_cmpneg_operand): ... to this.
2916 * config/arm/t-arm-elf: Add armv7 multilib.
2917 * config/arm/arm.md: Update patterns for Thumb-2 and Unified asm.
2919 (UNSPEC_STACK_ALIGN, ce_count): New.
2920 (arm_incscc, arm_decscc, arm_umaxsi3, arm_uminsi3,
2921 arm_zero_extendsidi2, arm_zero_extendqidi2): New
2923 * config/arm/fpa.md: Update patterns for Thumb-2 and Unified asm.
2924 (thumb2_movsf_fpa, thumb2_movdf_fpa, thumb2_movxf_fpa,
2925 thumb2_movsfcc_fpa, thumb2_movdfcc_fpa): New insns.
2926 * config/arm/cirrus.md: Update patterns for Thumb-2 and Unified asm.
2927 (cirrus_thumb2_movdi, cirrus_thumb2_movsi_insn,
2928 thumb2_cirrus_movsf_hard_insn, thumb2_cirrus_movdf_hard_insn): New
2930 * doc/extend.texi: Document ARMv7-M interrupt functions.
2931 * doc/invoke.texi: Document Thumb-2 new cores+architectures.
2933 2007-01-03 Jakub Jelinek <jakub@redhat.com>
2935 * unwind-dw2.c (SIGNAL_FRAME_BIT, EXTENDED_CONTEXT_BIT): Define.
2936 (struct _Unwind_Context): Rename args_size to flags, remove
2937 signal_frame field, add a new args_size field and version field.
2938 (_Unwind_IsSignalFrame, _Unwind_SetSignalFrame,
2939 _Unwind_IsExtendedContext): New inline functions.
2940 (_Unwind_GetGR, _Unwind_SetGR, _Unwind_GetGRPtr, _Unwind_SetGRPtr):
2941 Assume by_value array is only present if _Unwind_IsExtendedContext.
2942 (_Unwind_GetIPInfo, execute_cfa_program, uw_frame_state_for): Use
2943 _Unwind_IsSignalFrame.
2944 (__frame_state_for): Initialize context.flags to EXTENDED_CONTEXT_BIT.
2945 (uw_update_context_1): Use _Unwind_SetSignalFrame.
2946 (uw_init_context_1): Initialize context->flags to
2947 EXTENDED_CONTEXT_BIT.
2948 * config/rs6000/linux-unwind.h (frob_update_context): Use
2949 _Unwind_SetSignalFrame.
2951 2007-01-03 Andrew Pinski <andrew_pinski@playstation.sony.com>
2954 * gimplify.c (gimplify_modify_expr_complex_part): Move below
2955 tree_to_gimple_tuple. Call tree_to_gimple_tuple when we need
2958 2007-01-03 Kazu Hirata <kazu@codesourcery.com>
2960 * config/i386/i386.h (NON_STACK_REG_P, REGNO_OK_FOR_SIREG_P,
2961 REGNO_OK_FOR_DIREG_P, REWRITE_ADDRESS, ASM_OPERAND_LETTER,
2962 RET, AT_SP): Remove.
2963 * config/i386/i386.md (*sse_prologue_save_insn): Use return
2966 * alias.c (init_alias_analysis): Use VEC_safe_grow_cleared.
2967 * cfgbuild.c (find_basic_blocks): Likewise.
2968 * cfgrtl.c (rtl_create_basic_block): Likewise.
2969 * function.c (temp_slots_at_level): Likewise.
2970 * reg-stack.c (stack_regs_mentioned): Likewise.
2971 * regclass.c (allocate_reg_info): Likewise.
2972 * tree-cfg.c (init_empty_tree_cfg, build_tree_cfg, create_bb,
2973 set_bb_for_stmt, move_block_to_fn): Likewise.
2974 * tree-complex.c (tree_lower_complex): Likewise.
2975 * vec.h (VEC_safe_grow_cleared): New.
2977 * cgraphunit.c, tree-ssa-alias.c: Fix comment typos.
2979 2007-01-03 Zdenek Dvorak <dvorakz@suse.cz>
2981 * loop-unswitch.c (unswitch_loop): Pass probabilities to loopify.
2982 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Pass probabilities
2984 * cfgloopmanip.c (scale_loop_frequencies): Export.
2985 (loopify): Scale the frequencies by prescribed coefficients.
2986 (set_zero_probability): New function.
2987 (duplicate_loop_to_header_edge): Improve updating of frequencies.
2988 (lv_adjust_loop_entry_edge, loop_version): Set probabilities
2989 and frequencies according to arguments.
2990 * tree-ssa-loop-manip.c (tree_unroll_loop): Set probabilities
2992 * cfg.c (scale_bbs_frequencies_int): Allow scaling the frequencies up.
2993 * modulo-sched.c (sms_schedule): Set probabilities for entering
2994 versioned loop correctly.
2995 * tree-vect-transform.c (vect_transform_loop): Ditto.
2996 * cfgloop.h (loopify, loop_version): Declaration changed.
2997 (scale_loop_frequencies): Declared.
2999 2007-01-02 Jan Hubicka <jh@suse.cz>
3001 * cgraph.c: Include tree-flow.h
3002 (cgraph_add_new-function): Handle IPA_SSA mode; execute
3004 * cgraph.h (enum cgraph_state): Add CGRAPH_STATE_IPA_SSA.
3005 * tree-pass.h (pass_all_early_optimizations): Declare.
3006 * cgraphunit.c (cgraph_process_new_functions): Add IPA_SSA; execute
3008 (cgraph_analyze_function): Do early_local_passes.
3009 * tree-mudflap.c (mf_decl_cache_locals, mf_build_check_statement_for):
3010 Do not add referenced vars.
3011 * tree-optimize.c (gate_all_optimizations): Do not execute when not in
3013 (gate_all_early_local_passes): New gate.
3014 (pass_early_local_passes): Use new gate.
3015 (execute_early_local_optimizations): New functions.
3016 (gate_all_early_optimizations): New gate.
3017 (pass_all_early_optimizations): New pass.
3018 (execute_free_datastructures): Free SSA only when initialized.
3019 (gate_init_datastructures): Init only when optimizing.
3020 (tree_lowering_passes): Do early local passes when called late.
3021 * tree-profile.c (do_tree_profiling): Don't profile functions added
3023 (do_early_tree_profiling, pass_early_tree_profile): Kill.
3024 * tree-cfg.c (update_modified_stmts): Do not update when operands are
3026 * passes.c (init_optimizations_passes): Reorder so we go into SSA
3027 during early_local_passes.
3028 * Makefile.in (cgraph.o): Add dependency on tree-flow.h.
3031 2007-01-02 Carlos O'Donell <carlos@codesourcery.com>
3033 * Makefile.in: Update copyright year.
3035 2007-01-02 Carlos O'Donell <carlos@codesourcery.com>
3037 * Makefile.in: Export GCC_EXEC_PREFIX before calling $(RUNTEST)
3038 in $(lang_checks) and check-consistency targets.
3040 2007-01-02 Jan Hubicka <jh@suse.cz>
3042 * tree-mudflap.c (mf_decl_cache_locals, mf_build_check_statement_for):
3043 Do not add referenced vars.
3044 * tree-cfg.c (update_modified_stmts): Do not update when SSA operands
3046 * passes.c (init_optimization_passes): Put mudflap_2 after
3047 free_datastructures.
3049 2007-01-02 Jan Hubicka <jh@suse.cz>
3051 * tree-optimize (execute_fixup_cfg): Set after_inlining flag.
3052 Set NOTHROW flag on call statements proved to be nothrow.
3053 Update statement of local calls so new pure/const functions are
3054 updated. Update_ssa when in ssa form. Mark PHI nodes of nonlocal
3056 (tree_rest_of_compilation): Register hooks and initialize bitmap
3057 early. Do not set after_inlining flag.
3059 2007-01-02 Steve Ellcey <sje@cup.hp.com>
3061 * sbitmap.c (HOST_BITS_PER_LONG_LONG): Change to
3062 HOST_BITS_PER_LONGLONG
3064 2007-01-02 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
3067 * c-typeck.c (store_init_value): Don't emit pedantic overflow
3068 warning for non-static initializers.
3070 2007-01-02 Steven Bosscher <steven@gcc.gnu.org>
3072 * config/alpha/alpha.md, arm/arm.c, darwin.c, frv/frv.md,
3073 m32r/m32r.c, m32r/m32r.c, mn10300/mn10300.md, pa/pa.c,
3074 rs6000/rs6000.c, s390/s390.md, sh/sh.md, sparc/sparc.c:
3075 Always use set_unique_reg_note to add REG_EQUAL notes.
3077 2007-01-02 Kazu Hirata <kazu@codesourcery.com>
3080 2007-01-02 Kazu Hirata <kazu@codesourcery.com>
3082 * alias.c (init_alias_analysis): Use VEC_safe_grow_cleared.
3083 * cfgbuild.c (find_basic_blocks): Likewise.
3084 * cfgrtl.c (rtl_create_basic_block): Likewise.
3085 * function.c (temp_slots_at_level): Likewise.
3086 * reg-stack.c (stack_regs_mentioned): Likewise.
3087 * regclass.c (allocate_reg_info): Likewise.
3088 * tree-cfg.c (init_empty_tree_cfg, build_tree_cfg, create_bb,
3089 set_bb_for_stmt, move_block_to_fn): Likewise.
3090 * tree-complex.c (tree_lower_complex): Likewise.
3091 * vec.h (VEC_safe_grow_cleared): New.
3093 2007-01-02 Ian Lance Taylor <iant@google.com>
3095 * c-common.c (c_common_truthvalue_conversion): When warning about
3096 using an assignment as a truth value, set TREE_NO_WARNING.
3098 2007-01-02 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
3101 * c.opt (Wold-style-declaration): New.
3102 * doc/invoke.texi (C-only Warning Options): New.
3103 (Wold-style-declaration): Document it.
3104 (Wextra): Enabled by -Wextra.
3105 * c-opts.c (c_common_post_options): Enabled by -Wextra.
3106 * c-decl.c (declspecs_add_scspec): Replace -Wextra with
3107 -Wold-style-declaration.
3109 2007-01-02 Kazu Hirata <kazu@codesourcery.com>
3111 * alias.c (init_alias_analysis): Use VEC_safe_grow_cleared.
3112 * cfgbuild.c (find_basic_blocks): Likewise.
3113 * cfgrtl.c (rtl_create_basic_block): Likewise.
3114 * function.c (temp_slots_at_level): Likewise.
3115 * reg-stack.c (stack_regs_mentioned): Likewise.
3116 * regclass.c (allocate_reg_info): Likewise.
3117 * tree-cfg.c (init_empty_tree_cfg, build_tree_cfg, create_bb,
3118 set_bb_for_stmt, move_block_to_fn): Likewise.
3119 * tree-complex.c (tree_lower_complex): Likewise.
3120 * vec.h (VEC_safe_grow_cleared): New.
3122 2007-01-02 Douglas Gregor <doug.gregor@gmail.com>
3124 * c-common.c(c_common_nodes_and_builtins): Since variants of
3125 void_type_node get built before it is given a name, we need to
3126 give those variants the name, too.
3127 (complete_array_type): We need to work with the canonical main
3128 type of the array, from which we will build the qualified version.
3129 * params.def (PARAM_VERIFY_CANONICAL_TYPES): New.
3130 * print-tree.c (print_node): Display canonical type information
3132 * stor-layout.c (layout_type): When we don't know the
3133 alignment of a type for which we're building an array, we end up
3134 guessing wrong, so make the type require structural equality.
3135 * tree.c (make_node_stat): When we build a new type, it is its
3137 (build_type_attribute_qual_variant): When building an attribute
3138 variant, its canonical type is the non-attribute variant. However,
3139 if the attributes are target-dependent and they differ, we need to
3140 use structural equality checks for this type.
3141 (build_qualified_type): A qualified type is not equivalent to its
3142 unqualified variant; set the canonical type appropriately.
3143 (build_distinct_type_copy): When building a distinct type from
3144 another type, the new type is its own canonical type.
3145 (build_variant_type_copy): When building a new type variant, we
3146 assume that it is equivalent to the original type.
3147 (build_pointer_type_for_mode): When building a pointer type, also
3148 build a canonical type pointer.
3149 (build_reference_type_for_mode): When building a reference type,
3150 also build a canonical type reference.
3151 (build_index_type): When we can't hash an index type (e.g.,
3152 because its maximum value is negative), the index type requires
3153 structural equality tests.
3154 (build_array_type): Build the canonical form of an array type.
3155 (build_function_type): Function types require structural equality,
3156 because they contain default arguments, attributes, etc.
3157 (build_method_type_directly): Ditto for method types.
3158 (build_offset_type): Build the canonical offset type.
3159 (build_complex_type): Build the canonical vector type.
3160 (make_vector_type): Build the canonical vector type.
3161 * tree.h (TYPE_CANONICAL): New.
3162 (TYPE_STRUCTURAL_EQUALITY_P): New.
3163 (SET_TYPE_STRUCTURAL_EQUALITY): New.
3164 (struct tree_type): Added "canonical" field.
3165 * params.h (VERIFY_CANONICAL_TYPES): New.
3166 * doc/c-tree.texi (TYPE_CANONICAL): Document.
3167 (TYPE_STRUCTURAL_EQUALITY_P): Document.
3168 (SET_TYPE_STRUCTURAL_EQUALITY): Document.
3169 * doc/invoke.texi (verify-canonical-types): Document --param
3170 parameter for verifying canonical types.
3172 2007-01-02 Joseph Myers <joseph@codesourcery.com>
3174 * config.gcc (powerpc-*-eabispe*, powerpc-*-eabisimaltivec*,
3175 powerpc-*-eabisim*, powerpc-*-eabialtivec*, powerpc-*-eabi*,
3176 powerpc-*-rtems*, powerpc-wrs-vxworks, powerpc-wrs-vxworksae,
3177 powerpcle-*-eabisim*, powerpcle-*-eabi*): Add rs6000/e500.h to
3179 * config/rs6000/e500.h: New.
3180 * config/rs6000/eabi.h (TARGET_SPE_ABI, TARGET_SPE, TARGET_E500,
3181 TARGET_ISEL, TARGET_FPRS, TARGET_E500_SINGLE, TARGET_E500_DOUBLE):
3183 * config/rs6000/linuxspe.h (TARGET_SPE_ABI, TARGET_SPE,
3184 TARGET_E500, TARGET_ISEL, TARGET_FPRS, TARGET_E500_SINGLE,
3185 TARGET_E500_DOUBLE): Remove.
3186 * config/rs6000/vxworks.h (TARGET_SPE_ABI, TARGET_SPE,
3187 TARGET_E500, TARGET_ISEL, TARGET_FPRS): Remove.
3188 * config/rs6000/rs6000.h (CHECK_E500_OPTIONS): Define.
3189 * config/rs6000/rs6000.c (rs6000_override_options): Use
3192 2007-01-02 Joseph Myers <joseph@codesourcery.com>
3194 * config/rs6000/rs6000.c (print_operand): Check (TARGET_SPE ||
3195 TARGET_E500_DOUBLE), not TARGET_E500, for %y.
3196 (rs6000_generate_compare, rs6000_emit_sCOND, output_cbranch,
3197 rs6000_emit_cmove): Don't check TARGET_E500.
3198 * config/rs6000/rs6000.md (bunordered, bordered, sunordered,
3199 sordered): Don't check TARGET_E500.
3201 2007-01-01 Eric Christopher <echristo@apple.com>
3203 * config/mips/mips.c (mips_regno_mode_ok_for_base_p): Use
3204 HARD_REGISTER_NUM_P.
3206 2007-01-01 Roger Sayle <roger@eyesopen.com>
3208 * fold-const.c (fold_binary) <EQ_EXPR>: Fold "(X^C1) eq/ne C2" into
3209 "X eq/ne (C1^C2)". Fold "(X^Z) eq/ne (Y^Z)" as "X eq/ne Y" when Z
3210 has no side-effects. Fold "(X^C1) eq/ne (Y^C2)" as "(X^(C1^C2))
3213 2007-01-01 Mike Stump <mrs@apple.com>
3215 * configure.ac: Remove support for building with Apple's gcc-3.1.
3217 2007-01-02 Joseph Myers <joseph@codesourcery.com>
3220 * caller-save.c (add_stored_regs): Only handle SUBREGs if inner
3221 REG is a hard register. Do not modify REG before calling
3223 * rtlanal.c (subreg_get_info): Don't assert size of XMODE is a
3224 multiple of the size of YMODE for certain lowpart cases.
3226 2007-01-01 Andrew Pinski <pinskia@gmail.com>
3229 * gimplify (voidify_wrapper_expr): Update for
3232 2007-01-01 Andreas Schwab <schwab@suse.de>
3235 * config/ia64/ia64.c (ia64_compute_frame_size): Account space for
3236 save of BR0 in extra_spill_size instead of spill_size.
3237 (ia64_expand_prologue): Save BR0 outside of the gr/br/fr spill
3239 (ia64_expand_epilogue): Restore BR0 from its new location.
3241 2007-01-01 Andrew Pinski <pinskia@gmail.com>
3243 * gimplify.c (gimplify_init_constructor <case VECTOR_TYPE>):
3244 Use a temporary variable if the left hand side is not a gimple
3247 2007-01-01 Andrew Pinski <pinskia@gmail.com>
3249 * gimplify.c (gimplify_return_expr): Make the temporary variable
3250 for the return expression, a gimple register variable.
3252 2007-01-01 Jan Hubicka <jh@suse.cz>
3254 * emit-rtl.c (emit_copy_of_insn_after): Do not call copy_insn_1 for
3257 2007-01-01 Mike Stump <mrs@apple.com>
3259 * configure.ac (HAVE_GAS_LITERAL16): Add autoconf check for
3261 * config/darwin.c (machopic_select_rtx_section): Use
3263 (darwin_mergeable_constant_section): Likewise.
3264 * configure: Regenerate.
3265 * config.in: Regenerate.
3267 2007-01-01 Jan Hubicka <jh@suse.cz>
3268 Andrew Pinski <pinskia@gmail.com>
3270 * cgraphunit.c (cgraph_optimize): Call cgraph_add_new_functions
3271 before starting IPA passes.