1 2007-02-14 Roger Sayle <roger@eyesopen.com>
2 Zdenek Dvorak <dvorakz@suse.cz>
4 * tree-dump.c (dump_switch_p_1): Require exact match of the option
7 2007-02-14 Zdenek Dvorak <dvorakz@suse.cz>
9 * passes.c (next_pass_1): Clear the next field of the copied
12 2007-02-14 Richard Henderson <rth@redhat.com>
14 * tree-sra.c (early_sra): New.
15 (decl_can_be_decomposed_p): Deny va_list if early_sra.
16 (tree_sra_early, pass_sra_early): New.
17 * tree-pass.h (pass_sra_early): Declare.
18 * passes.c (init_optimization_passes): Use it.
20 2007-02-14 Richard Guenther <rguenther@suse.de>
22 * flags.h (issue_strict_overflow_warning): Convert to a macro.
24 2007-02-14 Dorit Nuzman <dorit@il.ibm.com>
26 PR tree-optimization/30771
27 * tree-vect-analyze.c (vect_determine_vectorization_factor): Traverse
29 (vect_analyze_operations): Induction phis can now be marked as
31 (vect_mark_stmts_to_be_vectorized): No special treatment for phis.
32 Update documentation accordingly.
34 2007-02-14 Nick Clifton <nickc@redhat.com>
36 * builtin-types.def (DEF_FUNCTION_TYPE_x): Do not imply that at
37 most 3 arguments are supported.
38 (DEF_FUNCTION_TYPE_VAR_5): Fix typo in its description.
40 2007-02-13 Seongbae Park <seongbae.park@gmail.com>
42 * bitmap.c (bitmap_and, bitmap_and_compl, bitmap_xor):
43 Ensure dst->current is valid.
45 2007-02-13 Paul Brook <paul@codesourcery.com>
47 * config.gcc: Add arm*-*-uclinux-*eabi.
48 * config/arm/uclinux-elf.h (TARGET_OS_CPP_BUILTINS): Define.
49 (SUBTARGET_EXTRA_LINK_SPEC): Define.
50 (STARTFILE_SPEC, ENDFILE_SPEC): Remove broken -shared handling.
51 (LINK_GCC_C_SEQUENCE_SPEC): Undef.
54 * config/arm/arm.c (arm_override_options): Use r9 as EABI PIC
56 * config/arm/uclinux-eabi.h: New file.
57 * config/arm/linux-eabi.h (WCHAR_TYPE): Remove.
58 * config/arm/linux-gas.h (WCHAR_TYPE): Use unsigned long on AAPCS
61 2007-02-13 Ian Lance Taylor <iant@google.com>
63 * common.opt: Add Wstrict-overflow and Wstrict-overflow=.
64 * flags.h (warn_strict_overflow): Declare.
65 (enum warn_strict_overflow_code): Define.
66 (issue_strict_overflow_warning): New static inline function.
67 * opts.c (warn_strict_overflow): New variable.
68 (common_handle_option): Handle OPT_Wstrict_overflow and
69 OPT_Wstrict_overflow_.
70 * c-opts.c (c_common_handle_option): Set warn_strict_overflow for
72 * fold-const.c: Include intl.h.
73 (fold_deferring_overflow_warnings): New static variable.
74 (fold_deferred_overflow_warning): New static variable.
75 (fold_deferred_overflow_code): New static variable.
76 (fold_defer_overflow_warnings): New function.
77 (fold_undefer_overflow_warnings): New function.
78 (fold_undefer_and_ignore_overflow_warnings): New function.
79 (fold_deferring_overflow_warnings_p): New function.
80 (fold_overflow_warning): New static function.
81 (make_range): Add strict_overflow_p parameter. Change all
83 (extract_muldiv, extract_muldiv_1): Likewise.
84 (fold_unary) [ABS_EXPR]: Check ABS_EXPR before calling
85 tree_expr_nonnegative_p.
86 (fold_negate_expr): Call fold_overflow_warning.
87 (fold_range_test): Likewise.
88 (fold_comparison): Likewise.
89 (fold_binary): Likewise. Call tree_expr_nonnegative_warnv_p
90 instead of tree_expr_nonnegative_p.
91 (tree_expr_nonnegative_warnv_p): Rename from
92 tree_expr_nonnegative_p, add strict_overflow_p parameter.
93 (tree_expr_nonnegative_p): New function.
94 (tree_expr_nonzero_warnv_p): Rename from tree_expr_nonzero_p, add
95 strict_overflow_p parameter.
96 (tree_expr_nonzero_p): New function.
97 * passes.c (verify_interpass_invariants): New static function.
98 (execute_one_pass): Call it.
99 * tree-ssa-loop-niter.c (expand_simple_operations): Ignore fold
101 (number_of_iterations_exit, loop_niter_by_eval): Likewise.
102 (estimate_numbers_of_iterations): Likewise.
103 (scev_probably_wraps_p): Likewise.
104 * tree-ssa-ccp.c: Include "toplev.h".
105 (evaluate_stmt): Defer fold overflow warnings until we know we are
107 (struct fold_stmt_r_data): Add stmt field.
108 (fold_stmt_r): Defer fold overflow warnings until we know we
110 (fold_stmt): Initialize stmt field of fold_stmt_r_data.
111 (fold_stmt_inplace): Likewise.
112 * tree-cfgcleanup.c: Include "toplev.h" rather than "errors.h".
113 (cleanup_control_expr_graph): Defer fold overflow warnings until
114 we know we are going to optimize.
115 * tree-cfg.c (fold_cond_expr_cond): Likewise.
116 * tree-ssa-threadedge.c (simplify_control_stmt_condition):
118 * tree-vrp.c (vrp_expr_computes_nonnegative): Call
119 tree_expr_nonnegative_warnv_p instead of tree_expr_nonnegative_p.
120 * tree-ssa-loop-manip.c (create_iv): Likewise.
121 * c-typeck.c (build_conditional_expr): Likewise.
122 (build_binary_op): Likewise.
123 * tree-vrp.c (vrp_expr_computes_nonzero): Call
124 tree_expr_nonzero_warnv_p instead of tree_expr_nonzero_p.
125 (extract_range_from_unary_expr): Likewise.
126 * simplify-rtx.c (simplify_const_relational_operation): Warn when
127 assuming that signed overflow does not occur.
128 * c-common.c (pointer_int_sum): Ignore fold overflow warnings.
129 * tree.h (tree_expr_nonnegative_warnv_p): Declare.
130 (fold_defer_overflow_warnings): Declare.
131 (fold_undefer_overflow_warnings): Declare.
132 (fold_undefer_and_ignore_overflow_warnings): Declare.
133 (fold_deferring_overflow_warnings_p): Declare.
134 (tree_expr_nonzero_warnv_p): Declare.
135 * doc/invoke.texi (Option Summary): Add -Wstrict-overflow to list
137 (Warning Options): Document -Wstrict-overflow.
138 * Makefile.in (tree-ssa-threadedge.o): Depend on toplev.h.
139 (tree-ssa-ccp.o): Likewise.
140 (tree-cfgcleanup.o): Change errors.h dependency to toplev.h.
141 (fold-const.o): Depend on intl.h.
143 2007-02-13 Ian Lance Taylor <iant@google.com>
146 * lower-subreg.c (resolve_simple_move): Decompose subregs in
149 2007-02-13 Stuart Hastings <stuart@apple.com>
151 * gcc/config/i386/i386.md (fixuns_truncdfhi2): Require SSE2.
153 2007-02-13 Richard Henderson <rth@redhat.com>
155 * config/alpha/alpha.c (alpha_stdarg_optimize_hook): Strip
156 handled_component_p before looking for the indirect_ref.
158 2007-02-13 Richard Henderson <rth@redhat.com>
160 * config/i386/i386.md (bswapsi_1): Rename from bswapsi2,
161 remove flags clobber.
162 (bswapsi2): New expander, emit code for !TARGET_BSWAP.
163 (bswaphi_lowpart): New.
164 (bswapdi2): Rename from bswapdi2_rex, remove flags clobber,
165 remove TARGET_BSWAP test. Delete expander of the same name.
167 * optabs.c (widen_bswap, expand_doubleword_bswap): New.
168 (expand_unop): Use them.
170 2007-02-13 Uros Bizjak <ubizjak@gmail.com>
172 * config/i386/i386.md (cmpdi_ccno_1_rex64, *cmpsi_ccno_1,
173 *cmphi_ccno_1, *cmpqi_ccno_1, *movsi_xor, *movstricthi_xor,
174 *movstrictqi_xor, *movdi_xor_rex64, *ashldi3_1_rex64,
175 *ashldi3_cmp_rex64, *ashldi3_cconly_rex64, ashlsi3, *ashlsi3_1_zext,
176 *ashlsi3_cmp, *ashlsi3_cconly, *ashlsi3_cmp_zext, *ashlhi3_1_lea,
177 *ashlhi3_1, *ashlhi3_cmp, *ashlhi3_cconly, *ashlqi3_1_lea,
178 *ashlqi3_1, *ashlqi3_cmp, *ashlqi3_cconly): Remove equivalent
179 assembler dialect choice from asm templates.
181 2007-02-12 Richard Henderson <rth@redhat.com>
183 * config/i386/i386.md (fixuns_trunc<SSEMODEF>si_1): New insn.
184 (fixuns_trunc<SSEMODEF>si2): Use it.
185 * config/i386/sse.md (vec_setv4sf_0): Export.
186 * config/i386/i386.c (ix86_build_const_vector): Export.
187 (ix86_split_convert_uns_si_sse): Rename from
188 ix86_expand_convert_uns_si_sse and rewrite as a splitter.
189 * config/i386/i386-protos.h: Update.
191 2007-02-13 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
194 * c-typeck.c (c_finish_return): Improve warning message.
196 2007-02-12 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
198 * alias.c (find_symbolic_term): Delete unused function.
200 2007-02-12 Uros Bizjak <ubizjak@gmail.com>
202 * config/i386/i386.md (paritydi2, paritysi2): New expanders.
203 (paritydi2_cmp, paritydi2_cmp): New insn and split patterns.
204 (*parityhi2_cmp, *parityqi2_cmp): New insn patterns.
206 2007-02-12 Eric Botcazou <ebotcazou@adacore.com>
208 * tree.h (DECL_IGNORED_P): Document further effect for FUNCTION_DECL.
209 * cgraphunit.c (cgraph_expand_function): If DECL_IGNORED_P is set on
210 the function, temporarily point the debug interface to the null one.
212 2007-02-12 Eric Botcazou <ebotcazou@adacore.com>
214 * dwarf2out.c (round_up_to_align): New static function.
215 (field_byte_offset): Use it to round the offset.
217 2007-02-12 Richard Henderson <rth@redhat.com>
219 * config/alpha/alpha.md (bswapsi2, bswapdi2): New.
220 (inswl_const): Export.
222 2007-02-12 Richard Henderson <rth@redhat.com>
224 * calls.c (emit_library_call_value_1): If PROMOTE_MODE modifed the
225 result mode of the libcall, convert back to outmode.
227 2007-02-12 Roger Sayle <roger@eyesopen.com>
229 * config/i386/i386.md (*bswapdi2_rex): Renamed from bswapdi2.
230 (bswapdi2): New define_expand to implement 32-bit implementation.
232 2007-02-12 Nick Clifton <nickc@redhat.com>
234 * doc/invoke.texi (Overall Options): Document --help=.
235 * gcc.c (target_help_flag): Rename to print_subprocess_flag.
236 (cc1_options): Pass --help= on to cc1.
237 (display_help): Add description of --help=.
238 (process_command): Add code to handle --help=. Allow translated
239 --help and --target-help switches to be passed on to compiler
241 (main): Remove unused if statement.
242 * opts.c (columns): Remove.
243 (LEFT_COLUMN): Define.
244 (wrap_help): Add columns argument.
245 (print_filtered_help): Change parameters to be an include bitmask,
246 an exclude bitmask, an any bitmask and the column width. Move the
247 code to display the params list here. Add code to display the
248 status of options rather than their descriptions if the quiet flag
250 (print_specific_help): Change parameters to be an include bitmask,
251 an exclude bitmask and an any bitmask. Move code to look up the
252 column width here. Decide upon the title for an options listing.
253 (common_handle_options): Add code to handle --help=. Adapt code
254 for --help and --target-help to use the revised form of the
255 print_specific_help function.
256 (print_help): Delete.
257 (print_param_help): Delete.
258 (print_switch): Delete.
259 * opts.h (cl_lang_count): Add prototype.
260 (CL_PARAMS, CL_WARNING, CL_OPTIMIZATION, CL_MIN_OPTION_CLASS,
261 CL_MAX_OPTION_CLASS): New defines.
262 * optc-gen.awk: Add construction of cl_lang_count.
263 * c.opt: Add Warning attribute to warning options and Optimization
264 attribute to optimization options.
265 * common.opt: Likewise.
267 Add -fhelp and -ftarget-help as aliases for the transformed --help
268 and --target-help options.
269 * opt-functions.awk: Add code to handle Warning and Optimization
272 2007-02-12 Richard Henderson <rth@redhat.com>
274 * config/alpha/constraints.md: New file.
275 * config/alpha/alpha.c: Include tm-constrs.h.
276 (alpha_const_ok_for_letter_p, alpha_const_double_ok_for_letter_p,
277 alpha_extra_constraint): Remove.
278 (alpha_emit_conditional_branch): Use satisfies_constraint_*.
279 * config/alpha/alpha-protos.h: Update.
280 * config/alpha/alpha.h (REG_CLASS_FROM_LETTER): Remove.
281 (CONST_OK_FOR_LETTER_P, CONST_DOUBLE_OK_FOR_LETTER_P): Remove.
282 (EXTRA_CONSTRAINT): Remove.
283 * config/alpha/alpha.md: Include constraints.md.
284 (adddi splitter): Use satisfies_constraint_*.
285 * config/alpha/predicates.md (add_operand): Likewise.
286 (sext_add_operand, addition_operation): Likewise.
288 2007-02-12 Dorit Nuzman <dorit@il.ibm.com>
290 PR tree-optimization/29145
291 * tree-data-ref.c (base_addr_differ_p): Make us more conservative
292 in our handling of restrict qualified pointers.
294 2007-02-12 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
297 * doc/invoke.texi (Wunused-value): Update description.
298 (Wextra): Delete item.
299 * opts.c (set_Wextra): Don't use the value of Wextra to set the
300 value of Wunused-value.
301 * c-typeck.c (c_process_expr_stmt): Don't check extra_warnings.
302 (c_finish_stmt_expr): Don't check extra_warnings.
303 (emit_side_effect_warnings): The caller is responsible to check
306 2007-02-11 Roger Sayle <roger@eyesopen.com>
307 Matt Thomas <matt@3am-software.com>
309 * simplify-rtx.c (simplify_relational_operation_1): Correct typo.
311 2007-02-11 Roger Sayle <roger@eyesopen.com>
313 * simplify-rtx.c (simplify_relational_operation_1): Optimize
314 comparisons of POPCOUNT against zero.
315 (simplify_const_relational_operation): Likewise.
317 2007-02-11 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
319 * doc/invoke.texi (Wextra): Delete outdated paragraph.
321 2007-02-11 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
323 * dwarf2out.c (root_type): Delete unused function.
325 2007-02-11 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
327 * genattrtab.c (contained_in_p): Delete unused function.
328 (write_expr_attr_cache): Likewise.
330 2007-02-11 Jan Hubicka <jh@suse.cz>
332 * ipa-inline.c (cgraph_edge_badness): Add "else" missing in the
335 2007-02-11 Steven Bosscher <steven@gcc.gnu.org>
337 * fwprop.c (try_fwprop_subst): Use set_unique_reg_note
338 to add the REG_EQ* note.
339 * see.c (see_merge_one_use_extension): Likewise.
340 * local-alloc.c (update_equiv_regs): Likewise. Also don't
341 turn REG_EQUAL notes into REG_EQUIV notes if the target
342 register may have more than one set.
343 * function.c (assign_parm_setup_reg): Use set_unique_reg_note.
344 * gcse.c (try_replace_reg): Likewise.
345 * alias.c (init_alias_analysis): Use find_reg_equal_equiv_note.
346 * calls.c (fixup_tail_calls): Likewise. Abort if there is
347 more than one REG_EQUIV note.
348 * reload1.c (gen_reload): Use set_unique_reg_note.
350 2007-02-11 Uros Bizjak <ubizjak@gmail.com>
352 * config/i386/i386.c (TARGET_VECTORIZE_BUILTIN_CONVERSION): Define.
353 (ix86_builtin_conversion): New function.
355 2007-02-06 Mark Mitchell <mark@codesourcery.com>
358 * tree.h (DECL_REPLACEABLE_P): New macro.
359 * except.c (set_nothrow_function_flags): Likewise.
361 2007-02-11 Tehila Meyzels <tehila@il.ibm.com>
362 Ira Rosen <irar@il.ibm.com>
363 Dorit Nuzman <dorit@il.ibm.com>
365 * doc/tm.texi (TARGET_VECTORIZE_BUILTIN_CONVERSION): New target hook.
366 * targhooks.c (default_builtin_vectorized_conversion): New.
367 * targhooks.h (default_builtin_vectorized_function): New declaration.
368 * target.h (struct vectorize): Add builtin_conversion field.
369 * tree-vectorizer.h (type_conversion_vec_info_type): New enum
370 stmt_vec_info_type value.
371 (vectorizable_conversion): New declaration.
372 * tree-vect-analyze.c (vect_analyze_operations): Add
373 vectorizable_conversion call.
374 * target-def.h (TARGET_VECTORIZE_BUILTIN_CONVERSION): New.
375 * tree-vect-transform.c (vectorizable_conversion): New function.
376 (vect_transform_stmt): Add case for type_conversion_vec_info_type.
377 * tree-vect-generic.c (expand_vector_operations_1): Consider correct
379 * config/rs6000/rs6000.c (rs6000_builtin_conversion): New.
380 (TARGET_VECTORIZE_BUILTIN_CONVERSION): Defined.
381 (rs6000_expand_builtin): Add handling a case of ALTIVEC_BUILTIN_VCFUX
382 or ALTIVEC_BUILTIN_VCFSX.
384 2007-02-10 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
387 * pa.md (movdf): For 64-bit target, fail if operand 1 is a non-zero
388 CONST_DOUBLE and operand 0 is a hard register.
389 (movdi): For 64-bit target, remove code to force CONST_DOUBLE to
390 memory. Fail if operand 1 is a non-zero CONST_INT and operand 0
391 is a hard floating-point register.
393 2007-02-10 Richard Henderson <rth@redhat.com>
394 Jakub Jelinek <jakub@redhat.com>
395 Alexandre Oliva <aoliva@redhat.com>
397 * Makefile.in (libgcc-support, libgcc.mvars): Add emutls.c.
398 * builtin-types.def (BT_WORD): Make unsigned.
399 (BT_FN_VOID_PTR_WORD_WORD_PTR): New.
400 * builtins.def (BUILT_IN_EMUTLS_GET_ADDRESS): New.
401 (BUILT_IN_EMUTLS_REGISTER_COMMON): New.
402 * c-decl.c (grokdeclarator): Don't error if !have_tls.
403 * c-parser.c (c_parser_omp_threadprivate): Likewise.
404 * dwarf2out.c (loc_descriptor_from_tree_1): Don't do anything for
406 * expr.c (emutls_var_address): New.
407 (expand_expr_real_1): Expand emulated tls.
408 (expand_expr_addr_expr_1): Likewise.
409 * libgcc-std.ver: Add __emutls_get_address, __emutls_register_common.
410 * output.h (emutls_finish): Declare.
411 * toplev.c (compile_file): Call it.
412 * tree-ssa-address.c (gen_addr_rtx): Check for const-ness of the
413 address before wrapping in CONST.
414 * varasm.c (emutls_htab, emutls_object_type): New.
415 (EMUTLS_VAR_PREFIX, EMUTLS_TMPL_PREFIX): New.
416 (get_emutls_object_name, get_emutls_object_type): New.
417 (get_emutls_init_templ_addr, emutls_decl): New.
418 (emutls_common_1, emutls_finish): New.
419 (assemble_variable): When emulating tls, swap decls; generate
420 constructor for the emutls objects.
421 (do_assemble_alias): When emulating tls, swap decl and target name.
422 (default_encode_section_info): Don't add SYMBOL_FLAG_TLS_SHIFT
424 * varpool.c (decide_is_variable_needed): Look at force_output.
425 Recurse for emulated tls.
426 (cgraph_varpool_remove_unreferenced_decls): Remove checks redundant
427 with decide_is_variable_needed.
428 * emutls.c: New file.
429 * config/sparc/sol2.h (ASM_DECLARE_OBJECT_NAME): Only emit
430 tls_object for real tls.
432 2007-02-10 Kaz Kojima <kkojima@gcc.gnu.org>
434 PR rtl-optimization/29599
435 * reload1.c (eliminate_regs_in_insn): Take the destination
436 mode into account when computing the offset.
438 2007-02-09 Stuart Hastings <stuart@apple.com>
439 Richard Henderson <rth@redhat.com>
441 * gcc/config/i386/i386.h (TARGET_KEEPS_VECTOR_ALIGNED_STACK): New.
442 * gcc/config/i386/darwin.h: (TARGET_KEEPS_VECTOR_ALIGNED_STACK): New.
443 * gcc/config/i386/i386.md (fixuns_trunc<mode>si2, fixuns_truncsfhi2,
444 fixuns_truncdfhi2): New.
445 (fix_truncsfdi_sse): Call ix86_expand_convert_sign_didf_sse.
446 (floatunsdidf2): Call ix86_expand_convert_uns_didf_sse.
447 (floatunssisf2): Add call to ix86_expand_convert_uns_sisf_sse.
448 (floatunssidf2): Allow nonimmediate source.
449 * gcc/config/i386/sse.md (movdi_to_sse): New.
450 (vec_concatv2di): Drop '*'.
451 * gcc/config/i386/i386-protos.h (ix86_expand_convert_uns_si_sse,
452 ix86_expand_convert_uns_didf_sse, ix86_expand_convert_uns_sidf_sse,
453 ix86_expand_convert_uns_sisf_sse, ix86_expand_convert_sign_didf_sse):
455 * gcc/config/i386/i386.c (ix86_expand_convert_uns_si_sse,
456 ix86_expand_convert_uns_didf_sse, ix86_expand_convert_uns_sidf_sse,
457 ix86_expand_convert_uns_sisf_sse, ix86_expand_convert_sign_didf_sse,
458 ix86_build_const_vector, ix86_expand_vector_init_one_nonzero): New.
459 (ix86_build_signbit_mask): Fix decl of v, refactor to call
460 ix86_build_const_vector.
461 (x86_emit_floatuns): Rewrite.
463 2007-02-10 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
465 * genautomata.c (longest_path_length): Delete unused function.
466 (struct state): Delete unused longest_path_length.
467 (UNDEFINED_LONGEST_PATH_LENGTH): Delete unused macro.
468 (get_free_state): Delete unused.
470 2007-02-09 Jan Hubicka <jh@suse.cz>
472 * params.def (PARAM_INLINE_UNIT_GROWTH): Set to 30.
473 * doc/invoke.texi (inline-unit-growth): Update default value.
475 * Makefile.in (passes.o, ipa-inline.o): Add dependencies.
476 * cgraphbuild.c (build_cgraph_edges): Compute frequencies.
477 (rebuild_cgraph_edges): Likewise.
478 * cgraph.c (cgraph_set_call_stmt): Add new argument frequency.
479 (dump_cgraph_node): Dump frequencies.
480 (cgraph_clone_edge): Add frequency scales.
481 (cgraph_clone_node): Add freuqnecy.
482 * cgraph.h (cgraph_edge): Add freuqnecy argument.
483 (CGRAPH_FREQ_BASE, CGRAPH_FREQ_MAX): New constants.
484 (cgraph_create_edge, cgraph_clone_edge, cgraph_clone_node): Update.
485 * tree-pass.h (TODO_rebuild_frequencies): New constant.
486 * cgraphunit.c (verify_cgraph_node): Verify frequencies.
487 (cgraph_copy_node_for_versioning): Update call of cgraph_clone_edge.
488 (save_inline_function_body): Likewise.
489 * ipa-inline.c: inluce rtl.h
490 (cgraph_clone_inlined_nods): Update call of cgraph_clone_node.
491 (cgraph_edge_badness): Use frequencies.
492 (cgraph_decide_recursive_inlining): Update clonning.
493 (cgraph_decide_inlining_of_small_function): Dump frequency.
494 * predict.c (estimate_bb_frequencies): Export.
495 * predict.h (estimate_bb_frequencies): Declare.
496 * tree-inline.c (copy_bb): Watch overflows.
497 (expand_call_inline): Update call of cgraph_create_edge.
498 (optimize_inline_calls): Use TODO flags to update frequnecies.
499 * passes.h: Include predict.h
500 (init_optimization_passes): Move profile ahead.
501 (execute_function_todo): Handle TODO_rebuild_frequencies.
503 2007-02-09 Roger Sayle <roger@eyesopen.com>
505 * config/alpha/alpha.c (emit_insxl): Force the first operand of
506 the insbl or inswl pattern into a register.
508 2007-02-09 Roger Sayle <roger@eyesopen.com>
510 * config/ia64/ia64.md (bswapdi2): New define_insn.
512 2007-02-09 Richard Henderson <rth@redhat.com>
514 * config/i386/constraints.md (Ym): New constraint.
515 * config/i386/i386.md (movsi_1): Change Y2 to Yi constraints.
516 (movdi_1_rex64): Split sse and xmm general register moves from
517 memory move alternatives. Use conditional register constraints.
518 (movsf_1, movdf_integer): Likewise.
519 (zero_extendsidi2_32, zero_extendsidi2_rex64): Likewise.
520 (movdf_integer_rex64): New.
521 (pushsf_rex64): Fix output constraints.
522 * config/i386/sse.md (sse2_loadld): Split rm alternative, use Yi.
523 (sse2_stored): Likewise.
524 (sse2_storeq_rex64): New.
525 * config/i386/i386.c (x86_inter_unit_moves): Enable for not
527 (ix86_secondary_memory_needed): Don't bypass TARGET_INTER_UNIT_MOVES
528 for optimize_size. Remove SF/DFmode hack.
530 2007-02-09 Dwarakanath Rajagopal <dwarak.rajagopal@amd.com>
532 * config/i386/driver-i386.c: Turn on -mtune=native for AMDFAM10.
535 2007-02-09 Nathan Sidwell <nathan@codesourcery.com>
536 Richard Sandiford <richard@codesourcery.com>
538 * config.gcc (m68010-*-netbsdelf*, m68k*-*-netbsdelf*)
539 (m68k*-*-openbsd*, m68k-*-linux*): Set default_cf_cpu.
540 (m68k-*-aout*, m68k-*-coff*, m68k-*-uclinux*, m68k-*-rtems*): Add
541 m68k/t-mlib to tmake_file.
542 (m68020-*-elf*, m68k-*-elf*): Likewise. Add t-m68kbare as well.
543 (m68k*-*-*): Use --with-arch to pick a default for --with-cpu.
544 (m680[012]0-*-*, m68k*-*-*): Add support for --with-arch.
545 Allow it to be cf or m68k. Set m68k_arch_family. If that
546 variable is not empty, add t-$m68k_arch_family to tmake_file.
547 Add t-mlibs to tmake_file.
548 * doc/install.texi: Document --with-arch=m68k and --with-arch=cf.
549 * config/m68k/t-cf: New file.
550 * config/m68k/t-m68k: Likewise.
551 * config/m68k/t-mlibs: Likewise.
552 * config/m68k/t-m68kbare (MULTILIB_OPTIONS, MULTILIB_DIRNAMES)
553 (MULTILIB_MATCHES, MULTILIB_EXCEPTIONS): Delete.
554 (M68K_MLIB_DIRNAMES, M68K_MLIB_OPTIONS): Define.
555 * config/m68k/t-m68kelf (MULTILIB_OPTIONS, MULTILIB_DIRNAMES)
556 (MULTILIB_MATCHES, MULTILIB_EXCEPTIONS, LIBGCC, INSTALL_LIBGCC):
558 * config/m68k/t-openbsd (MULTILIB_OPTIONS, LIBGCC): Delete.
559 (INSTALL_LIBGCC): Delete.
560 (M68K_MLIB_DIRNAMES, M68K_MLIB_OPTIONS): Define.
561 * config/m68k/t-rtems (MULTILIB_OPTIONS, MULTILIB_DIRNAMES)
562 (MULTILIB_MATCHES, MULTILIB_EXCEPTIONS): Delete.
563 (M68K_MLIB_CPU): Define.
564 * config/m68k/t-uclinux (MULTILIB_OPTIONS, MULTILIB_DIRNAMES)
565 (MULTILIB_MATCHES, MULTILIB_EXCEPTIONS): Delete.
566 (M68K_MLIB_CPU, M68K_MLIB_OPTIONS, M68K_MLIB_DIRNAMES): Define.
568 2007-02-09 Zdenek Dvorak <dvorakz@suse.cz>
569 Richard Guenther <rguenther@suse.de>
572 * fold-const.c (fold_comparison): Handle obfuscated comparisons
573 against INT_MIN/INT_MAX.
574 * tree-ssa-loop-ivcanon.c (remove_empty_loop): Print to dump
575 file if a loop is removed.
577 2007-02-09 Joseph Myers <joseph@codesourcery.com>
579 * calls.c (store_one_arg): Pass correct alignment to
580 emit_push_insn for non-BLKmode values.
581 * expr.c (emit_push_insn): If STRICT_ALIGNMENT, copy to an
582 unaligned stack slot via a suitably aligned slot.
584 2007-02-08 DJ Delorie <dj@redhat.com>
586 * config/m32c/m32c.c (m32c_unpend_compare): Add default to silence
588 (legal_subregs): Use unsigned char, make const.
589 (m32c_illegal_subreg_p): Use ARRAY_SIZE. Delete unused variables.
591 2007-02-08 Paul Brook <paul@codesourcery.com>
593 * config/arm/lib1funcs.asm (RETLDM): Pop directly into PC when no
594 special interworking needed.
596 2007-02-08 Harsha Jagasia <harsha.jagasia@amd.com>
598 * config/i386/xmmintrin.h: Make inclusion of emmintrin.h
599 conditional to __SSE2__.
600 (Entries below should have been added to first ChangeLog
601 entry for amdfam10 dated 2007-02-05)
602 * config/i386/emmintrin.h: Generate #error if __SSE2__ is not
604 * config/i386/pmmintrin.h: Generate #error if __SSE3__ is not
606 * config/i386/tmmintrin.h: Generate #error if __SSSE3__ is not
609 2007-02-08 DJ Delorie <dj@redhat.com>
611 * config/m32c/m32c-protos.h (m32c_illegal_subreg_p): New.
612 * config/m32c/m32c.c (legal_subregs): New.
613 (m32c_illegal_subreg_p): New.
614 * config/m32c/predicates.md (m32c_any_operand): Use it to reject
615 unsupported subregs of hard regs.
617 2007-02-08 Jan Hubicka <jh@suse.cz>
619 * tree-cfg.c (bsi_replace): Shortcut when replacing the statement with
620 the same one; always update histograms.
622 2007-02-08 Diego Novillo <dnovillo@redhat.com>
624 * passes.c (init_optimization_passes): Tidy comment.
626 2007-02-08 Roger Sayle <roger@eyesopen.com>
628 * simplify-rtx.c (simplify_unary_operation_1) <POPCOUNT>: We can
629 strip zero_extend, bswap and rotates from POCOUNT's argument.
630 <PARITY>: Likewise, we can strip not, bswap, sign_extend,
631 zero_extend and rotates from PARITY's argument.
632 <BSWAP>: A byte-swap followed by a byte-swap is an identity.
633 (simplify_const_unary_operation) <BSWAP>: Evaluate the byte-swap
634 of an integer constant at compile-time.
636 2007-02-08 Diego Novillo <dnovillo@redhat.com>
639 * tree-flow.h (struct var_ann_d): Remove field 'is_used'.
641 * tree-ssa-alias.c (compute_is_aliased): Remove. Update all
644 * tree-ssa-live.c (remove_unused_locals): Do not remove
645 TREE_ADDRESSABLE variables.
646 * tree-ssa-structalias.c (compute_points_to_sets): Tidy.
647 * tree-ssa-operands.c (add_virtual_operand): Remove argument
648 FOR_CLOBBER. Update all users.
649 If VAR has an associated alias set, add a virtual operand for
650 it if no alias is found to conflict with the memory reference.
652 2007-02-07 Jan Hubicka <jh@suse.cz>
653 Robert Kidd <rkidd@crhc.uiuc.edu>
655 * value-prof.c (visit_hist, free_hist): Return 1 instead of 0.
657 2007-02-07 Ian Lance Taylor <iant@google.com>
659 * lower-subreg.c (simple_move): Reject PARTIAL_INT modes.
661 2007-02-07 Roger Sayle <roger@eyesopen.com>
663 * config/rs6000/rs6000.md (ctz<mode>2, ffs<mode>2, popcount<mode>2,
664 parity<mode>2, smulsi3_highpart, abstf2_internal, allocate_stack,
665 tablejumpdi, movsi_to_cr_one): Remove constraints from
666 define_expand's match_operands.
668 2007-02-07 Roger Sayle <roger@eyesopen.com>
670 * global.c (compute_regsets): Move declatation of "i" inside of
671 #ifdef ELIMINABLE_REGS to avoid unused variable bootstrap failure.
673 2007-02-07 Jakub Jelinek <jakub@redhat.com>
676 * gimplify.c (gimplify_scan_omp_clauses): Remove special casing
677 of INDIRECT_REF <RESULT_DECL>.
679 * config/i386/i386.c (override_options): Set PTA_SSSE3 for core2.
681 2007-02-06 J"orn Rennecke <joern.rennecke@arc.com>
682 Kaz Kojima <kkojima@gcc.gnu.org>
685 * config/sh/sh.c (expand_cbranchdi4): Use scratch register
687 (sh_initialize_trampoline): Add parentheses to avoid a warning.
689 2007-02-06 Zdenek Dvorak <dvorakz@suse.cz>
691 * doc/loop.texi: Document possibility not to perform disambiguation
692 of loops with multiple latches.
693 * cfgloopmanip.c (alp_enum_p): Removed.
694 (add_loop): Handle subloops. Use get_loop_body_with_size.
695 (create_preheader): Do not allow ENTRY_BLOCK_PTR to be preheader.
696 * cfghooks.c (redirect_edge_and_branch_force): Set dominator for
697 the new forwarder block.
698 (make_forwarder_block): Only call new_bb_cbk if it is not NULL.
699 Handle the case latch is NULL.
700 * tree-ssa-dom.c (tree_ssa_dominator_optimize): Avoid cfg modifications
701 when marking loop exits.
702 * ifcvt.c (if_convert): Ditto. Mark loop exits even if cfg cannot
704 * loop-init.c (loop_optimizer_init): Do not modify cfg. Call
705 disambiguate_loops_with_multiple_latches.
706 * tree-cfgcleanup.c (cleanup_tree_cfg_loop): Calculate dominators
707 before fix_loop_structure.
708 * cfgloop.c: Include pointer-set.h and output.h.
709 (canonicalize_loop_headers, HEADER_BLOCK, LATCH_EDGE,
710 update_latch_info, mfb_keep_just, mfb_keep_nonlatch): Removed.
711 (get_loop_latch_edges, find_subloop_latch_edge_by_profile,
712 find_subloop_latch_edge_by_ivs, find_subloop_latch_edge,
713 mfb_redirect_edges_in_set, form_subloop, merge_latch_edges,
714 disambiguate_multiple_latches, get_loop_body_with_size,
715 disambiguate_loops_with_multiple_latches): New functions.
716 (flow_loop_dump): Dump multiple latch edges.
717 (flow_loop_nodes_find): Handle loops with multiple latches.
718 (flow_loops_find): Ditto. Do not call canonicalize_loop_headers.
719 (glb_enum_p): Modified.
720 (get_loop_body): Use get_loop_body_with_size.
721 * cfgloop.h (LOOPS_HAVE_RECORDED_EXITS): New flag.
722 (AVOID_CFG_MODIFICATIONS): New constant.
723 (disambiguate_loops_with_multiple_latches, add_loop,
724 get_loop_body_with_size): Declare.
725 * Makefile.in (cfgloop.o): Add pointer-set.h and output.h.
727 2007-02-06 Seongbae Park <seongbae.park@gmail.com>
730 * global.c (compute_regsets): New function.
731 (global_alloc): Refactored ELIMINABLE_REGSET
732 and NO_GLOBAL_ALLOC_REGS computation out.
733 (rest_of_handle_global_alloc): Call compute_regsets()
734 for non-optimizing case.
736 2007-02-06 Richard Henderson <rth@redhat.com>
738 * config/i386/constraints.md (Y2): Rename from Y.
739 (Yi): New constraint.
740 * config/i386/i386.md (movsi_1, movdi_2, pushdf_nointeger,
741 pushdf_integer, movdf_nointeger, movdf_integer, zero_extendsidi2_32,
742 zero_extendsidi2_rex64, truncxfdf2_mixed): Change Y constraints to Y2.
743 (extendsfdf2_mixed, extendsfdf2_sse, truncdfsf_fast_mixed,
744 truncdfsf_fast_sse, truncdfsf_mixed, fix_truncdfdi_sse,
745 fix_truncdfsi_sse, floatsidf2_mixed, floatsidf2_sse,
746 floatdidf2_mixed, floatdidf2_sse, absnegdf2_mixed,
747 absnegdf2_sse, sse_setccdf, fop_df_comm_mixed, fop_df_comm_sse,
748 fop_df_1_mixed, fop_df_1_sse): Change Y constraints to x.
749 * config/i386/mmx.md (mov<MMXMODEI>_internal_rex64,
750 mov<MMXMODEI>_internal, movv2sf_internal_rex64, movv2sf_internal,
751 vec_extractv2si_1): Change Y constraints to Y2.
752 * config/i386/sse.md (vec_setv4sf_0, vec_concatv2df, vec_dupv4si,
753 vec_dupv2di, sse2_concatv2si, vec_concatv4si_1, vec_concatv2di):
754 Change Y constraints to Y2.
755 (sse2_loadld): Change Y constraints to x.
757 2007-02-06 Roger Sayle <roger@eyesopen.com>
759 * config/rs6000/rs6000.md (popcount<mode>2): Rewrite.
760 (parity<mode>2): New define_expand using rs6000_emit_parity.
761 * config/rs6000/rs6000.c (rs6000_emit_popcount,
762 rs6000_emit_parity): New functions.
763 * config/rs6000/rs6000-protos.h (rs6000_emit_popcount,
764 rs6000_emit_parity): Prototype here.
766 2007-02-06 Ian Lance Taylor <iant@google.com>
768 * lower-subreg.c (simple_move_operand): Reject CONST.
769 (resolve_clobber): Call validate_change rather than directly
770 assigning to XEXP (pat, 0).
772 2006-02-06 Paolo Bonzini <bonzini@gnu.org>
774 * Makefile.in (tree-ssa-loop-ivopts.o): Add pointer-set.h dependency.
775 (tree-ssa-reassoc.o): Add pointer-set.h dependency.
776 (tree-cfg.o): Remove hashtab.h dependency.
778 * tree-ssa-loop-ivopts.c: Include pointer-set.h.
779 (struct ivopts_data): Change niters to pointer_map_t.
780 (struct nfe_cache_elt, nfe_hash, nfe_eq): Delete.
781 (niter_for_exit): Create pointer_map on demand. Change for
783 (tree_ssa_iv_optimize_init): Initialize data->niters to NULL.
784 (free_loop_data): Destroy data->niters if created and reset field.
785 (tree_ssa_iv_optimize_finalize): Don't delete data->niters here.
786 (tree_ssa_iv_optimize_loop): Check for presence of stale data.
788 * tree-ssa-reassoc.c: Include pointer-set.h.
789 (bb_rank): Change to long *.
790 (operand_rank): Change to pointer_map_t.
791 (find_operand_rank): Return long, -1 if not found. Declare as inline.
792 (insert_operand_rank): Accept long.
793 (operand_entry_hash, operand_entry_eq): Remove.
794 (get_rank): Return long. Adjust for changes above.
795 (init_reassoc): Change rank type to long. Adjust creation of bb_rank
797 (fini_reassoc): Delete operand_rank with pointer_map_destroy.
799 * tree-ssa-structalias.c (vi_for_tree): Change to pointer_map.
800 (struct tree_vi, tree_vi_t, tree_vi_hash, tree_vi_eq): Delete.
801 (insert_vi_for_tree): Rewrite for pointer_map API. Assert argument
803 (lookup_vi_for_tree): Rewrite for pointer_map API. Return varinfo_t
804 directly since it cannot be NULL.
805 (get_vi_for_tree): Rewrite for pointer_map API.
806 (find_what_p_points_to): Adjust for change to lookup_vi_for_tree.
807 (init_alias_vars): Create vi_for_tree as pointer_map.
808 (delete_points_to_sets): Delete vi_for_tree using pointer_map_destroy.
810 * tree-cfg.c: Don't include hashtab.h.
811 (edge_to_cases): Declare as pointer_map.
812 (struct edge_to_cases_elt, edge_to_cases_hash, edge_to_cases_eq):
814 (edge_to_cases_cleanup): Rewrite as pointer_map_traverse callback.
815 (start_recording_case_labels): Create edge_to_cases as pointer_map.
816 (end_recoding_case_labels): Cleanup edge_to_cases manually before
818 (record_switch_edge): Delete.
819 (get_cases_for_edge): Adjust for pointer_map API, inline
820 record_switch_edge (rewritten for new API), remove goto.
822 2006-02-06 Paolo Bonzini <bonzini@gnu.org>
824 * Makefile.in (tree-nested.o): Add pointer-set.h dependency.
825 * tree-nested.c: Include pointer-set.h.
826 (var_map_elt, var_map_eq, var_map_hash): Delete.
827 (struct nesting_info): Remove GTY marker. Change the two htab_t's
829 (nesting_info_bitmap_obstack): New.
830 (lookup_field_for_decl): Adjust for pointer_map API.
831 (lookup_tramp_for_decl): Adjust for pointer_map API.
832 (get_nonlocal_debug_decl): Adjust for pointer_map API.
833 (get_local_debug_decl): Adjust for pointer_map API.
834 (convert_nl_goto_reference): Adjust for pointer_map API.
835 (convert_nl_goto_receiver): Adjust for pointer_map API.
836 (create_nesting_tree): Create outside GGC space. Create bitmap on
837 the new obstack. Create field_map and var_map as pointer_maps.
838 (free_nesting_tree): Adjust for changes to create_nesting_tree.
840 (lower_nested_functions): Move root here, no need to NULL it.
841 Initialize and release the obstack.
843 2007-02-06 Paolo Bonzini <bonzini@gnu.org>
845 * tree.c (tree_int_map_hash, tree_int_map_eq, tree_int_map_marked_p):
846 Remove prototypes and make them non-static.
847 (struct tree_int_map): Remove.
848 * tree.h (struct tree_int_map): Move here, turning TO into an
850 (tree_int_map_hash, tree_int_map_eq, tree_int_map_marked_p): Declare.
852 * tree.h (TREE_COMPLEXITY): Remove.
853 (struct tree_exp): Remove complexity field.
854 * tree.c (build1_stat): Don't set it.
856 2007-02-06 Dorit Nuzman <dorit@il.ibm.com>
857 Victor Kaplansky <victork@il.ibm.com>
859 * tree-vectorizer.c (vect_is_simple_use): Support induction.
860 (vect_is_simple_reduction): Support reduction with induction as
862 (vect_is_simple_iv_evolution): Fix formatting.
863 * tree-vect-analyze.c (vect_mark_stmts_to_be_vectorized): Fix
864 formatting. Don't mark induction phis for vectorization.
865 (vect_analyze_scalar_cycles): Analyze all inductions, then reductions.
866 * tree-vect-transform.c (get_initial_def_for_induction): New function.
867 (vect_get_vec_def_for_operand): Support induction.
868 (vect_get_vec_def_for_stmt_copy): Fix formatting and add check for
870 (vectorizable_reduction): Support reduction with induction as one of
872 (vectorizable_type_demotion): Use def-type of stmt argument rather
875 * tree-ssa-loop.c (gate_scev_const_prop): Return the value of
876 flag_tree_scev_cprop.
877 * common.opt (tree-scev-cprop): New flag.
879 * tree-vect-transform.c (vect_create_destination_var): Use 'kind' in
880 call to vect_get_new_vect_var.
882 2007-02-06 Ira Rosen <irar@il.ibm.com>
884 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Check that
886 (vect_pattern_recog_1): Likewise.
888 2007-02-05 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
890 * fold-const.c (negate_expr_p): Handle CONJ_EXPR.
891 (fold_negate_expr): Likewise.
893 2007-02-05 Alexandre Oliva <aoliva@redhat.com>
896 * dwarf2out.c (modified_type_die): Follow DECL_ORIGINAL_TYPE
897 even if cv-qualification is the same.
899 2007-02-05 Geoffrey Keating <geoffk@apple.com>
901 * config/rs6000/darwin-tramp.asm (__trampoline_setup): Call
902 __enable_execute_stack on completion.
904 2007-02-05 Harsha Jagasia <harsha.jagasia@amd.com>
906 * config/i386/athlon.md (athlon_fldxf_k8, athlon_fld_k8,
907 athlon_fstxf_k8, athlon_fst_k8, athlon_fist, athlon_fmov,
908 athlon_fadd_load, athlon_fadd_load_k8, athlon_fadd, athlon_fmul,
909 athlon_fmul_load, athlon_fmul_load_k8, athlon_fsgn,
910 athlon_fdiv_load, athlon_fdiv_load_k8, athlon_fdiv_k8,
911 athlon_fpspc_load, athlon_fpspc, athlon_fcmov_load,
912 athlon_fcmov_load_k8, athlon_fcmov_k8, athlon_fcomi_load_k8,
913 athlon_fcomi, athlon_fcom_load_k8, athlon_fcom): Added amdfam10.
915 2007-02-05 Harsha Jagasia <harsha.jagasia@amd.com>
917 * config/i386/i386.md (x86_sahf_1, cmpfp_i_mixed, cmpfp_i_sse,
918 cmpfp_i_i387, cmpfp_iu_mixed, cmpfp_iu_sse, cmpfp_iu_387,
919 swapsi, swaphi_1, swapqi_1, swapdi_rex64, fix_truncsfdi_sse,
920 fix_truncdfdi_sse, fix_truncsfsi_sse, fix_truncdfsi_sse,
921 x86_fldcw_1, floatsisf2_mixed, floatsisf2_sse, floatdisf2_mixed,
922 floatdisf2_sse, floatsidf2_mixed, floatsidf2_sse,
923 floatdidf2_mixed, floatdidf2_sse, muldi3_1_rex64, mulsi3_1,
924 mulsi3_1_zext, mulhi3_1, mulqi3_1, umulqihi3_1, mulqihi3_insn,
925 umulditi3_insn, umulsidi3_insn, mulditi3_insn, mulsidi3_insn,
926 umuldi3_highpart_rex64, umulsi3_highpart_insn,
927 umulsi3_highpart_zext, smuldi3_highpart_rex64,
928 smulsi3_highpart_insn, smulsi3_highpart_zext, x86_64_shld,
929 x86_shld_1, x86_64_shrd, sqrtsf2_mixed, sqrtsf2_sse,
930 sqrtsf2_i387, sqrtdf2_mixed, sqrtdf2_sse, sqrtdf2_i387,
931 sqrtextendsfdf2_i387, sqrtxf2, sqrtextendsfxf2_i387,
932 sqrtextenddfxf2_i387): Added amdfam10_decode.
934 * config/i386/athlon.md (athlon_idirect_amdfam10,
935 athlon_ivector_amdfam10, athlon_idirect_load_amdfam10,
936 athlon_ivector_load_amdfam10, athlon_idirect_both_amdfam10,
937 athlon_ivector_both_amdfam10, athlon_idirect_store_amdfam10,
938 athlon_ivector_store_amdfam10): New define_insn_reservation.
939 (athlon_idirect_loadmov, athlon_idirect_movstore): Added
942 2007-02-05 Harsha Jagasia <harsha.jagasia@amd.com>
944 * config/i386/athlon.md (athlon_call_amdfam10,
945 athlon_pop_amdfam10, athlon_lea_amdfam10): New
946 define_insn_reservation.
947 (athlon_branch, athlon_push, athlon_leave_k8, athlon_imul_k8,
948 athlon_imul_k8_DI, athlon_imul_mem_k8, athlon_imul_mem_k8_DI,
949 athlon_idiv, athlon_idiv_mem, athlon_str): Added amdfam10.
951 2007-02-05 Harsha Jagasia <harsha.jagasia@amd.com>
953 * config/i386/athlon.md (athlon_sseld_amdfam10,
954 athlon_mmxld_amdfam10, athlon_ssest_amdfam10,
955 athlon_mmxssest_short_amdfam10): New define_insn_reservation.
957 2007-02-05 Harsha Jagasia <harsha.jagasia@amd.com>
959 * config/i386/athlon.md (athlon_sseins_amdfam10): New
960 define_insn_reservation.
961 * config/i386/i386.md (sseins): Added sseins to define_attr type
962 and define_attr unit.
963 * config/i386/sse.md: Set type attribute to sseins for insertq
966 2007-02-05 Harsha Jagasia <harsha.jagasia@amd.com>
968 * config/i386/athlon.md (sselog_load_amdfam10, sselog_amdfam10,
969 ssecmpvector_load_amdfam10, ssecmpvector_amdfam10,
970 ssecomi_load_amdfam10, ssecomi_amdfam10,
971 sseaddvector_load_amdfam10, sseaddvector_amdfam10): New
972 define_insn_reservation.
973 (ssecmp_load_k8, ssecmp, sseadd_load_k8, seadd): Added amdfam10.
975 2007-02-05 Harsha Jagasia <harsha.jagasia@amd.com>
977 * config/i386/athlon.md (cvtss2sd_load_amdfam10,
978 cvtss2sd_amdfam10, cvtps2pd_load_amdfam10, cvtps2pd_amdfam10,
979 cvtsi2sd_load_amdfam10, cvtsi2ss_load_amdfam10,
980 cvtsi2sd_amdfam10, cvtsi2ss_amdfam10, cvtsd2ss_load_amdfam10,
981 cvtsd2ss_amdfam10, cvtpd2ps_load_amdfam10, cvtpd2ps_amdfam10,
982 cvtsX2si_load_amdfam10, cvtsX2si_amdfam10): New
983 define_insn_reservation.
985 * config/i386/sse.md (cvtsi2ss, cvtsi2ssq, cvtss2si,
986 cvtss2siq, cvttss2si, cvttss2siq, cvtsi2sd, cvtsi2sdq,
987 cvtsd2si, cvtsd2siq, cvttsd2si, cvttsd2siq,
988 cvtpd2dq, cvttpd2dq, cvtsd2ss, cvtss2sd,
989 cvtpd2ps, cvtps2pd): Added amdfam10_decode attribute.
991 2007-02-05 Harsha Jagasia <harsha.jagasia@amd.com>
993 * config/i386/athlon.md (athlon_ssedivvector_amdfam10,
994 athlon_ssedivvector_load_amdfam10, athlon_ssemulvector_amdfam10,
995 athlon_ssemulvector_load_amdfam10): New define_insn_reservation.
996 (athlon_ssediv, athlon_ssediv_load_k8, athlon_ssemul,
997 athlon_ssemul_load_k8): Added amdfam10.
999 2007-02-05 Harsha Jagasia <harsha.jagasia@amd.com>
1001 * config/i386/i386.h (TARGET_SSE_UNALIGNED_MOVE_OPTIMAL): New macro.
1002 (x86_sse_unaligned_move_optimal): New variable.
1004 * config/i386/i386.c (x86_sse_unaligned_move_optimal): Enable for
1006 (ix86_expand_vector_move_misalign): Add code to generate movupd/movups
1007 for unaligned vector SSE double/single precision loads for AMDFAM10.
1009 2007-02-05 Harsha Jagasia <harsha.jagasia@amd.com>
1011 * config/i386/i386.h (TARGET_AMDFAM10): New macro.
1012 (TARGET_CPU_CPP_BUILTINS): Add code for amdfam10.
1013 Define TARGET_CPU_DEFAULT_amdfam10.
1014 (TARGET_CPU_DEFAULT_NAMES): Add amdfam10.
1015 (processor_type): Add PROCESSOR_AMDFAM10.
1017 * config/i386/i386.md: Add amdfam10 as a new cpu attribute to match
1018 processor_type in config/i386/i386.h.
1019 Enable imul peepholes for TARGET_AMDFAM10.
1021 * config.gcc: Add support for --with-cpu option for amdfam10.
1023 * config/i386/i386.c (amdfam10_cost): New variable.
1024 (m_AMDFAM10): New macro.
1025 (m_ATHLON_K8_AMDFAM10): New macro.
1026 (x86_use_leave, x86_push_memory, x86_movx, x86_unroll_strlen,
1027 x86_cmove, x86_3dnow_a, x86_deep_branch, x86_use_simode_fiop,
1028 x86_promote_QImode, x86_integer_DFmode_moves,
1029 x86_partial_reg_dependency, x86_memory_mismatch_stall,
1030 x86_accumulate_outgoing_args, x86_arch_always_fancy_math_387,
1031 x86_sse_partial_reg_dependency, x86_sse_typeless_stores,
1032 x86_use_ffreep, x86_use_incdec, x86_four_jump_limit,
1033 x86_schedule, x86_use_bt, x86_cmpxchg16b, x86_pad_returns):
1034 Enable/disable for amdfam10.
1035 (override_options): Add amdfam10_cost to processor_target_table.
1036 Set up PROCESSOR_AMDFAM10 for amdfam10 entry in
1037 processor_alias_table.
1038 (ix86_issue_rate): Add PROCESSOR_AMDFAM10.
1039 (ix86_adjust_cost): Add code for amdfam10.
1041 2007-02-05 Harsha Jagasia <harsha.jagasia@amd.com>
1043 * config/i386/i386.opt: Add new Advanced Bit Manipulation (-mabm)
1044 instruction set feature flag. Add new (-mpopcnt) flag for popcnt
1045 instruction. Add new SSE4A (-msse4a) instruction set feature flag.
1046 * config/i386/i386.h: Add builtin definition for SSE4A.
1047 * config/i386/i386.md: Add support for ABM instructions
1049 * config/i386/sse.md: Add support for SSE4A instructions
1050 (movntss, movntsd, extrq, insertq).
1051 * config/i386/i386.c: Add support for ABM and SSE4A builtins.
1052 Add -march=amdfam10 flag.
1053 * config/i386/ammintrin.h: Add support for SSE4A intrinsics.
1054 * doc/invoke.texi: Add documentation on flags for sse4a, abm, popcnt
1056 * doc/extend.texi: Add documentation for SSE4A builtins.
1058 2007-02-05 Bob Wilson <bob.wilson@acm.org>
1060 * config/xtensa/xtensa.c (constantpool_mem_p): Skip over SUBREGs.
1062 2007-02-05 Richard Guenther <rguenther@suse.de>
1064 * tree-vectorizer.h (vectorizable_function): Add argument type
1065 argument, change return type.
1066 * tree-vect-patterns.c (vect_recog_pow_pattern): Adjust caller.
1067 * tree-vect-transform.c (vectorizable_function): Handle extra
1068 argument, return vectorized function decl.
1069 (build_vectorized_function_call): Remove.
1070 (vectorizable_call): Handle calls with result and argument types
1071 differing. Handle loop vectorization factor correctly.
1072 * targhooks.c (default_builtin_vectorized_function): Adjust for
1074 * targhooks.h (default_builtin_vectorized_function): Likewise.
1075 * target.h (builtin_vectorized_function): Add argument type
1077 * config/i386/i386.c (ix86_builtin_vectorized_function): Handle
1078 extra argument, allow vectorizing of lrintf.
1079 * doc/tm.texi (TARGET_VECTORIZE_BUILTIN_VECTORIZED_FUNCTION): Adjust
1080 documentation of target hook.
1082 2007-02-05 Hans-Peter Nilsson <hp@axis.com>
1085 * config/cris/cris.md ("*andsi_movu", "*andsi_clear", "*andhi_movu")
1086 ("*andhi_clear", andu (casesi+45)): For size-changed operand where
1087 memory is allowed, require !side_effects_p, not just !MEM_VOLATILE_P.
1089 2007-02-05 Roger Sayle <roger@eyesopen.com>
1091 * fold-const.c (fold_unary) <REAL_PART>: Test for availability of
1092 BUILT_IN_COS before simplifying REAL_PART(CEXPI)) to COS.
1093 <IMAG_PART>: Likewise, check for availability of BUILT_IN_SIN.
1094 * builtins.c (fold_builtin_sincos): Check for TARGET_C99_FUNCTIONS
1095 before canonicalizing sincos to cexpi.
1096 (fold_builtin_cexp): Likewise, for canonicalizing cexp to cexpi.
1098 2007-02-05 Roger Sayle <roger@eyesopen.com>
1100 * config/alpha/alpha.c (alpha_add_builtins): New Helper function.
1101 Set TREE_READONLY and TREE_NOTHROW directly, not via attributes.
1102 (alpha_init_builtins): Use alpha_add_builtins to process tables.
1104 2007-02-05 Roger Sayle <roger@eyesopen.com>
1106 * mips-tfile.c (initialize_init_file): Correct endianness test.
1108 2007-02-05 Kazu Hirata <kazu@codesourcery.com>
1110 * config/m68k/m68k.md (pushdi-1, pushdi, movsi+1): Don't use
1113 2007-02-05 Richard Sandiford <richard@codesourcery.com>
1115 * dwarf2out.c (dwarf2out_frame_debug_expr): Record the register
1116 saves in a PARALLEL before the register assignments.
1118 2007-02-05 Richard Sandiford <richard@codesourcery.com>
1120 * doc/tm.texi (DWARF_ALT_FRAME_RETURN_COLUMN): Do not require
1121 DWARF_FRAME_RETURN_COLUMN to be a general register.
1122 * dwarf2out.c (init_return_column_size): New function, split from...
1123 (expand_builtin_init_dwarf_reg_sizes): ...here. Allow both
1124 DWARF_FRAME_RETURN_COLUMN and DWARF_ALT_FRAME_RETURN_COLUMN
1125 to be nongeneral registers.
1126 * config/m68k/m68k.h (DWARF_FRAME_REGNUM): Only map FP and
1128 (DWARF_FRAME_REGISTERS, DWARF_FRAME_RETURN_COLUMN): Define.
1129 (DWARF_ALT_FRAME_RETURN_COLUMN): Define.
1131 2007-02-04 Zdenek Dvorak <dvorakz@suse.cz>
1133 * cfgcleanup.c (try_optimize_cfg): Avoid removing ENTRY_BLOCK_PTR.
1135 2007-02-04 Zdenek Dvorak <dvorakz@suse.cz>
1137 * cfgloopmanip.c (loop_delete_branch_edge): Removed.
1138 (remove_path): Use can_remove_branch_p and remove_branch instead
1139 of loop_delete_branch_edge.
1140 * tree-ssa-loop-manip.c (scale_dominated_blocks_in_loop): New function.
1141 (tree_transform_and_unroll_loop): Remove dead branches immediately.
1142 Update profile using scale_dominated_blocks_in_loop.
1143 * cfghooks.c (can_remove_branch_p, remove_branch): New functions.
1144 * cfghooks.h (struct cfg_hooks): Add can_remove_branch_p.
1145 (can_remove_branch_p, remove_branch): Declare.
1146 * tree-cfg.c (tree_can_remove_branch_p): New function.
1147 (tree_cfg_hooks): Add tree_can_remove_branch_p.
1148 * cfgrtl.c (rtl_can_remove_branch_p): New function.
1149 (rtl_cfg_hooks, cfg_layout_rtl_cfg_hook): Add rtl_can_remove_branch_p.
1151 2007-02-05 Jan Hubicka <jh@suse.cz>
1154 * ipa-inline.c (cgraph_clone_inlined_nodes): When there are unanalyzed
1155 nodes in cgraph, don't remove offline copy of the function.
1157 2007-02-04 Jan Hubicka <jh@suse.cz>
1159 * tree-sra.c (sra_walk_expr): Add linebreaks. BITFIELD_REFs into
1160 vectors might cause maybe_lookup_element_for_expr to be called
1161 on non-sra-candidate.
1163 2007-02-04 Kazu Hirata <kazu@codesourcery.com>
1165 * config/bfin/bfin-modes.def, config/bfin/bfin.c,
1166 config/bfin/bfin.md, config/bfin/predicates.md: Follow
1167 spelling conventions.
1169 2007-02-04 Richard Guenther <rguenther@suse.de>
1172 * fold-const.c (try_move_mult_to_index): Make sure to not
1173 overflow one dimension of a multi-dimensional array access.
1175 2007-02-04 Jan Hubicka <jh@suse.cz>
1177 * passes.c (init_optimization_passes): Reindent.
1179 2007-02-04 Jan Hubicka <jh@suse.cz>
1180 Eric Botcazou <ebotcazou@adacore.com>
1182 * tree-optimize.c (has_abnormal_outgoing_edge_p): Move to...
1183 (execute_fixup_cfg): Break out the abnormal goto code.
1184 * tree-inline.c (has_abnormal_outgoing_edge_p): ...here.
1185 (make_nonlocal_label_edges): Move here from execute_fixup_cfg.
1186 (optimize_inline_calls): Call make_nonlocal_label_edges.
1188 2007-02-04 Jan Hubicka <jh@suse.cz>
1190 * tree-ssa-copyrename.c (copy_rename_partition_coalesce): Return
1191 true when something was changed.
1192 (rename_ssa_copies): When something was changed, do
1193 TODO_remove_unused_locals.
1194 * tree-ssa-forwprop.c (tree_ssa_forward_propagate_single_use_value):
1195 add TODO_remove_unused_locals when instruction was removed.
1197 2007-02-04 Jan Hubicka <jh@suse.cz>
1199 * ipa-inline.c (try_inline): Improve debug output; work on already
1201 (cgraph_decide_inlining_incrementally): Indent; improve debug output;
1202 call try_inline for already inlined edges too when flattening;
1203 inline also functions that make callee growth but overall unit size
1206 2007-02-04 Kazu Hirata <kazu@codesourcery.com>
1208 * config/m32c/bitops.md, config/m32c/jump.md,
1209 config/m32c/m32c.c, config/m32c/m32c.h, config/m32r/m32r.c,
1210 config/m32r/m32r.h, config/m32r/m32r.md,
1211 config/m32r/predicates.md, config/m68hc11/larith.asm,
1212 config/m68hc11/m68hc11.c, config/m68hc11/m68hc11.h,
1213 config/m68k/m68k.h, config/mcore/mcore.md, config/mips/4k.md,
1214 config/mips/mips-protos.h, config/mips/mips.c,
1215 config/mips/mips.h, config/mips/mips.md, config/mips/mips16.S,
1216 config/mn10300/mn10300.h, config/mn10300/predicates.md,
1217 config/mt/mt.c, config/mt/mt.h, config/mt/mt.md: Follow
1218 spelling conventions.
1220 * config/v850/v850.c, config/v850/v850.h, config/v850/v850.md:
1221 Follow spelling conventions.
1223 2007-02-03 Douglas Gregor <doug.gregor@gmail.com>
1225 * c-opts.c (c_common_post_options): If C++0x mode is enabled, don't
1226 warn about C++0x compatibility.
1228 2007-02-04 Kazu Hirata <kazu@codesourcery.com>
1230 * config/h8300/h8300.c, config/h8300/h8300.h,
1231 config/h8300/h8300.md: Follow spelling conventions.
1233 2007-02-03 Uros Bizjak <ubizjak@gmail.com>
1236 * combine.c (try_combine): Do not substitute source operand
1237 with constants wider than 2 * HOST_BITS_PER_WIDE_INT.
1239 2007-02-03 Jan Hubicka <jh@suse.cz>
1241 PR gcov-profile/30650
1242 * value-prof.c (stringop_block_profile): Fix handling of size counter;
1243 do not divide by zero for never executed counters.
1244 (tree_find_values_to_profile): Fix counters.
1245 * gcov-ui.h (GCOV_COUNTER_AVERAGE, GCOV_COUNTER_IOR): Fix comments.
1247 2007-02-03 Ian Lance Taylor <iant@google.com>
1249 * lower-subreg.c (simple_move_operand): New static function,
1250 broken out of simple_move. Reject LABEL_REF, SYMBOL_REF, and HIGH
1252 (simple_move): Call simple_move_operand.
1253 (find_decomposable_subregs): Add special handling of MEMs.
1254 (can_decompose_p): Rename from cannot_decompose_p. Reverse
1255 meaning of return value. If we see a hard register, test whether
1256 it can store a word_mode value. Change all callers.
1258 2007-02-03 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
1260 * pa.md (addvdi3, addvsi3, subvdi3, subvsi3, negvdi2, negvsi2): New
1261 ftrapv insns and expanders.
1262 (subdi3): Change define_expand operand 1 to arith11_operand, and
1263 operand 2 to reg_or_0_operand. Change constraints of 64-bit insn
1264 pattern to handle reg_or_0 operands. Revise 32-bit insn pattern to
1265 handle 11-bit constants and reg_or_0 operands in operands 1 and 2,
1269 * varasm.c (notice_global_symbol): Treat global objects as weak when
1272 2007-02-03 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
1274 * emit-rtl.c (dconstpi): Delete.
1276 (init_emit_once): Delete dconstpi and init dconstsqrt2.
1277 * real.h (dconstpi): Delete.
1279 * builtins.c (fold_builtin_cabs): Use dconstsqrt2.
1280 (fold_builtin_hypot): Likewise.
1282 2007-02-03 Tom Tromey <tromey@redhat.com>
1285 * gcc.c (cpp_unique_options): Any of -ggdb3, -gstabs3,
1286 -gcoff3, -gxcoff3, -gvms3 implies -dD.
1288 2007-02-03 Kazu Hirata <kazu@codesourcery.com>
1290 * c-decl.c, config/avr/avr.c, config/avr/avr.h,
1291 config/m68k/m68k.c, config/m68k/netbsd-elf.h,
1292 config/mn10300/mn10300.c, config/pdp11/pdp11.h,
1293 config/rs6000/cell.md, config/rs6000/darwin.h,
1294 config/sh/sh.md, config/sh/sh4-300.md, config/spu/spu.c,
1295 config/spu/spu.md, cselib.c, expr.c, haifa-sched.c, hwint.h,
1296 jump.c, reload.c, sched-deps.c, sched-int.h, tree-inline.c,
1297 tree-profile.c, tree-ssa-live.h, tree-vrp.c: Fix comment
1298 typos. Follow spelling conventions.
1299 * doc/invoke.texi: Follow spelling conventions.
1301 2007-02-03 Roger Sayle <roger@eyesopen.com>
1303 * simplify-rtx.c (simplify_relational_operation_1): Implement some
1304 canonicalization transformations that attempt to simplify integer
1305 constant comparisons to become comparisons against zero.
1307 2007-02-03 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
1309 * builtins.c (fold_builtin_cabs): Fold cabs(x+xi) into
1311 * fold-const.c (fold_binary): Fix comment typos. Fold complex
1312 (x,0)-(0,y) into (x,-y). Likewise (0,y)-(x,0) into (-x,y).
1314 2007-02-02 Mike Stump <mrs@apple.com>
1316 * config/darwin9.h (DARWIN_LINKER_GENERATES_ISLANDS): Add.
1317 * config/rs6000/rs6000.c (DARWIN_GENERATE_ISLANDS): Add.
1318 (output_call): Use DARWIN_GENERATE_ISLANDS to decide when to
1319 generate a branch island.
1321 2007-02-02 Bob Wilson <bob.wilson@acm.org>
1323 * config/xtensa/xtensa.c (smalloffset_mem_p): Use BASE_REG_P.
1324 (xtensa_legitimate_address_p): New.
1325 (xtensa_legitimize_address): New.
1326 (xtensa_output_addr_const_extra): New.
1327 * config/xtensa/xtensa.h (REG_OK_STRICT_FLAG): Define.
1329 (REG_OK_FOR_BASE_P): Use BASE_REG_P.
1330 (GO_IF_LEGITIMATE_ADDRESS): Move code to xtensa_legitimate_address_p.
1331 (LEGITIMIZE_ADDRESS): Move code to xtensa_legitimize_address.
1332 (OUTPUT_ADDR_CONST_EXTRA): Move code to xtensa_output_addr_const_extra.
1333 * config/xtensa/xtensa-protos.h (xtensa_legitimate_address_p): New.
1334 (xtensa_legitimize_address): New.
1335 (xtensa_output_addr_const_extra): New.
1337 2007-02-02 Steve Ellcey <sje@cup.hp.com>
1339 * config/ia64/ia64.c (ia64_print_operand): Fix compare strings.
1341 2007-02-02 Ian Lance Taylor <iant@google.com>
1343 * expmed.c (expand_divmod): Add comment.
1345 2007-02-02 Kazu Hirata <kazu@codesourcery.com>
1347 * emit-rtl.c (renumber_insns): Remove.
1348 * flags.h: Remove the extern for flag_renumber_insns.
1349 * rtl.h: Remove the prototype for renumber_insns.
1350 * toplev.c (flag_renumber_insns): Remove.
1352 2007-02-02 Hui-May Chang <hm.chang@apple.com>
1354 Revert for x86 darwin:
1355 2005-06-19 Uros Bizjak <uros@kss-loka.si>
1357 * config/i386/i386.c (ix86_function_arg_regno_p): Put back the
1358 code before the following patch under TARGET_MACHO.
1359 (ix86_function_value_regno_p): Likewise.
1361 2007-02-02 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
1363 * fold-const.c (negate_expr_p, fold_negate_expr): Handle
1366 2007-02-02 Maxim Kuvyrkov <mkuvyrkov@ispras.ru>
1369 * config/ia64/ia64.c (ia64_speculate_insn): Restrict to memory loads to
1370 general or fp registers. Add comments.
1371 * config/ia64/ia64.md (reg_pred_prefix): Add comment.
1373 2007-02-02 Paolo Bonzini <bonzini@gnu.org>
1375 * pointer-set.c (insert_aux): Only return insertion slot.
1376 (pointer_set_insert): Adjust.
1377 (pointer_set_traverse, struct pointer_map_t, pointer_map_create,
1378 pointer_map_destroy, pointer_map_insert, pointer_map_contains,
1379 pointer_map_traverse): New.
1380 * pointer-set.h (pointer_set_traverse, struct pointer_map_t,
1381 pointer_map_create, pointer_map_destroy, pointer_map_insert,
1382 pointer_map_contains, pointer_map_traverse): Declare.
1384 2007-02-02 Jakub Jelinek <jakub@redhat.com>
1387 * builtins.c (fold_builtin_sprintf): Do not attempt to optimize
1388 sprintf (str, "%s"). Do not optimize sprintf (str, "nopercent", p++).
1390 2007-02-02 Maxim Kuvyrkov <mkuvyrkov@ispras.ru>
1392 * sched-int.h (ds_to_dk, dk_to_ds): Declare functions.
1394 (struct _dep): New type.
1395 (dep_t): New typedef.
1396 (DEP_PRO, DEP_CON, DEP_KIND): New access macros.
1397 (DEP_STATUS): New access macro. The macro with the same name was
1398 renamed to DEP_LINK_STATUS.
1399 (dep_init): Declare function
1401 (struct _dep_link): New type.
1402 (dep_link_t): New typedef.
1403 (DEP_LINK_NODE, DEP_LINK_NEXT, DEP_LINK_PREV_NEXTP): New access macros.
1404 (DEP_LINK_DEP, DEP_LINK_PRO, DEP_LINK_CON, DEP_LINK_KIND): New macros.
1405 (DEP_LINK_STATUS): New macro.
1406 (debug_dep_links): New debug function.
1408 (struct _deps_list): New type.
1409 (deps_list_t): New typedef.
1410 (DEPS_LIST_FIRST): New access macro.
1411 (FOR_EACH_DEP_LINK): New cycle macro.
1412 (create_deps_list, free_deps_list, delete_deps_list): Declare
1414 (deps_list_empty_p, debug_deps_list, add_back_dep_to_deps_list): Ditto.
1415 (find_link_by_pro_in_deps_list, find_link_by_con_in_deps_list): Ditto.
1416 (copy_deps_list_change_con): Ditto.
1418 (move_dep_link): Declare function.
1420 (struct _dep_node): New type.
1421 (dep_node_t): New typedef.
1422 (DEP_NODE_BACK, DEP_NODE_DEP, DEP_NODE_FORW): New access macros.
1424 (struct haifa_insn_data.back_deps): New field to hold backward
1425 dependencies of the insn.
1426 (struct haifa_insn_data.depend): Rename to forw_deps. Change its type
1428 (struct haifa_insn_data.resolved_deps): Rename to resolved_back_deps.
1429 Change its type to deps_list_t.
1430 (INSN_BACK_DEPS): New access macro to use instead of LOG_LINKS.
1431 (INSN_DEPEND): Rename to INSN_FORW_DEPS.
1432 (RESOLVED_DEPS): Rename to INSN_RESOLVED_BACK_DEPS.
1434 (INSN_COST): Move to haifa-sched.c. Use insn_cost () instead.
1436 (DEP_STATUS): Rename to DEP_LINK_STATUS. Fix typo in the comment.
1438 (add_forw_dep, delete_back_forw_dep, insn_cost): Update declaration and
1440 (dep_cost): Declare.
1442 * sched-deps.c (CHECK): New macro to (en/dis)able sanity checks.
1443 (ds_to_dk, dk_to_ds): New functions.
1445 (init_dep_1): New static function.
1446 (init_dep): New function.
1447 (copy_dep): New static function.
1449 (dep_link_consistent_p, attach_dep_link, add_to_deps_list): New static
1451 (detach_dep_link): New static function.
1452 (move_dep_link): New function.
1454 (dep_links_consistent_p, dump_dep_links): New static functions.
1455 (debug_dep_links): New debugging function.
1457 (deps_obstack, dl_obstack, dn_obstack): New static variables.
1459 (alloc_deps_list, init_deps_list): New static functions.
1460 (create_deps_list): New function.
1461 (clear_deps_list): New static function.
1462 (free_deps_list, delete_deps_list, deps_list_empty_p): New functions.
1463 (deps_list_consistent_p, dump_deps_list): New static functions.
1464 (debug_deps_list): New function.
1465 (add_back_dep_to_deps_list, find_link_by_pro_in_deps_list): New
1467 (find_link_by_con_in_deps_list, copy_deps_list_change_con): Ditto.
1469 (maybe_add_or_update_back_dep_1, add_or_update_back_dep_1): Update to
1470 use new scheduler dependencies lists.
1471 (add_back_dep, delete_all_dependences, fixup_sched_groups): Ditto.
1472 (sched_analyze): Ditto. Initialize dependencies lists.
1473 (add_forw_dep, compute_forward_dependences): Update to use new
1474 scheduler dependencies lists.
1476 (init_dependency_caches): Init deps_obstack.
1477 (free_dependency_caches): Free deps_obstack.
1479 (adjust_add_sorted_back_dep, adjust_back_add_forw_dep): Update to use
1480 new scheduler dependencies lists.
1481 (delete_forw_dep, add_or_update_back_forw_dep): Ditto.
1482 (add_back_forw_dep, delete_back_forw_dep): Ditto.
1484 * sched-rgn.c (set_spec_fed, find_conditional_protection, is_pfree):
1485 Update to use new scheduler dependencies lists.
1486 (is_conditionally_protected, is_prisky, add_branch_dependences): Ditto.
1487 (debug_dependencies): Ditto.
1488 (schedule_region): Update comments.
1490 * sched-ebb.c (earliest_block_with_similiar_load): Update to use new
1491 scheduler dependencies lists.
1492 (schedule_ebb): Update comments.
1494 * rtl.def (DEPS_LIST): Remove.
1496 * lists.c (unused_deps_list): Remove.
1497 (free_list): Update assertions.
1499 (alloc_DEPS_LIST, free_DEPS_LIST_list, free_DEPS_LIST_node): Remove.
1500 (remove_free_DEPS_LIST_elem, copy_DEPS_LIST_list): Ditto.
1502 * rtl.h (free_DEPS_LIST_list, alloc_DEPS_LIST): Remove declarations.
1503 (remove_free_DEPS_LIST_elem, copy_DEPS_LIST_list): Ditto.
1505 * haifa-sched.c (comments): Update.
1506 (insn_cost1): Remove. Inline the code into insn_cost ().
1507 (insn_cost): Update to use new scheduler dependencies lists. Move
1508 processing of the dependency cost to dep_cost ().
1509 (dep_cost): New function. Use it instead of insn_cost () when
1510 evaluating cost of the dependency. Use compatible interface to
1511 interact with the target.
1512 (priority): Update to use new scheduler dependencies lists.
1513 (rank_for_schedule): Ditto. Optimize heuristic that prefers the insn
1514 with greater number of insns that depend on the insn.
1515 (schedule_insn): Update to use new scheduler dependencies lists. Add
1516 code to free backward dependencies lists. Inline and optimize code
1517 from resolve_dep () - see PR28071.
1518 (ok_for_early_queue_removal): Update to use new scheduler dependencies
1519 lists. Update call to targetm.sched.is_costly_dependence hook.
1521 (fix_inter_tick, try_ready, fix_tick_ready): Update to use new
1522 scheduler dependencies lists.
1524 (resolve_dep): Remove. Move the logic to schedule_insn ().
1525 (init_h_i_d): Initialize dependencies lists.
1527 (process_insn_depend_be_in_spec): Rename to
1528 process_insn_forw_deps_be_in_spec. Update to use new scheduler
1530 (add_to_speculative_block, create_check_block_twin, fix_recovery_deps):
1531 Update to use new scheduler dependencies lists.
1532 (clear_priorities, calc_priorities, add_jump_dependencies): Ditto.
1534 * ddg.c (create_ddg_dependence, create_ddg_dep_no_link): Update to use
1535 new scheduler dependencies lists.
1536 (build_intra_loop_deps): Ditto.
1538 * target.h (struct _dep): Declare to use in
1539 gcc_target.sched.is_costly_dependence.
1540 (struct gcc_target.sched.adjust_cost): Fix typo.
1541 (struct gcc_target.sched.is_costly_dependence): Change signature to use
1542 single dep_t parameter instead of an equivalent triad.
1543 (struct gcc_target.sched.adjust_cost_2): Remove.
1545 * target-def.h (TARGET_SCHED_ADJUST_COST_2): Remove.
1547 * reg-notes.def (DEP_TRUE, DEP_OUTPUT, DEP_ANTI): Update comments.
1549 * doc/tm.texi (TARGET_SCHED_IS_COSTLY_DEPENDENCE): Update
1551 (TARGET_SCHED_ADJUST_COST_2): Remove documentation.
1553 * doc/rtl.texi (LOG_LINKS): Remove part about instruction scheduler.
1554 (REG_DEP_TRUE): Document.
1556 * config/ia64/ia64.c (ia64_adjust_cost_2): Rename to ia64_adjust_cost.
1557 Change signature to correspond to the targetm.sched.adjust_cost hook.
1558 Update use in TARGET_SCHED_ADJUST_COST_2.
1559 (TARGET_SCHED_ADJUST_COST_2): Rename to TARGET_SCHED_ADJUST_COST.
1560 (ia64_dependencies_evaluation_hook, ia64_dfa_new_cycle): Update to use
1561 new scheduler dependencies lists.
1562 (ia64_gen_check): Ditto.
1564 * config/mips/mips.c (vr4130_swap_insns_p): Update to use new scheduler
1567 * config/rs6000/rs6000.c (rs6000_is_costly_dependence): Change
1568 signature to correspond to the targetm.sched.is_costly_dependence hook.
1569 (is_costly_group): Update to use new scheduler dependencies lists.
1571 * config/spu/spu.c (spu_sched_adjust_cost): Use insn_cost () function
1572 instead of INSN_COST () macro.
1574 2007-02-01 Ian Lance Taylor <iant@google.com>
1576 * lower-subreg.c (resolve_clobber): Handle a subreg of a concatn.
1578 2007-02-01 Guy Martin <gmsoft@gentoo.org>
1580 * pa.md (tp_load): Correct mfctl instruction syntax.
1582 2007-02-01 Geoffrey Keating <geoffk@apple.com>
1584 * config/rs6000/rs6000.c (rs6000_stack_info): Correct
1585 altivec_padding_size calculation on AIX. Improve comment, add
1586 assert to verify that it's right.
1588 * config/rs6000/darwin.h (MD_UNWIND_SUPPORT): Don't define for
1591 2007-2-01 Seongbae Park <seongbae.park@gmail.com>
1594 * global.c (global_alloc): Add mising initialization of
1597 2007-02-01 Roger Sayle <roger@eyesopen.com>
1599 * alias.c (init_alias_analysis): Correct whitespace.
1600 * bb-reorder.c (fix_edges_for_rarely_executed_code,
1601 partition_hot_cold_basic_blocks): Likewise.
1602 * builtins.c (expand_builtin_printf, expand_builtin_fprintf,
1603 expand_builtin_sprintf, fold_builtin_carg, fold_builtin_sprintf,
1604 maybe_emit_sprintf_chk_warning, fold_builtin_sprintf_chk,
1605 fold_builtin_snprintf_chk, fold_builtin_printf,
1606 fold_builtin_fprintf, do_mpfr_ckconv, do_mpfr_arg1, do_mpfr_arg2,
1607 do_mpfr_arg3, do_mpfr_sincos): Likewise.
1608 * cfgcleanup.c (cleanup_cfg): Likewise.
1609 * cfgexpand.c (tree_expand_cfg): Likewise.
1610 * fold-const.c (fold_binary) <RDIV_EXPR>: Likewise.
1611 * function.c (get_next_funcdef_no): Likewise.
1612 * gengtype.c (main): Likewise.
1613 * genmodes.c (main): Likewise.
1614 * gcse.c (bypass_conditional_jumps, print_ldst_list): Likewise.
1615 * haifa-sched.c (schedule_block, extend_h_i_d): Likewise.
1616 * ifcvt.c (noce_emit_move_insn): Likewise.
1617 * modulo-sched.c (generate_prolog_epilog, sms_schedule_by_order):
1619 * stor-layout.c (get_best_mode): Likewise.
1620 * tree-ssa-loop-niter.c (get_val_for): Likewise.
1621 * tree-ssa-structalias.c (get_varinfo, get_varinfo_fc,
1622 scc_visit, do_ds_constraint, do_complex_constraint, label_visit,
1623 perform_var_substitution, solve_graph): Likewise.
1624 * tree-vrp.c (vrp_finalize): Likewise.
1626 2007-02-01 Ian Lance Taylor <iant@google.com>
1628 * lower-subreg.c (simplify_gen_subreg_concatn): If we ask for the
1629 high part of a paradoxical subreg, return a constant zero.
1631 2007-02-01 Zdenek Dvorak <dvorakz@suse.cz>
1633 * toplev.c (lang_dependent_init): Call init_set_costs.
1634 * loop-init.c (loop_optimizer_init): Do not call init_set_costs.
1636 2007-02-01 Richard Guenther <rguenther@suse.de>
1639 * fold-const.c (fold_negate_expr): Allow negating a
1640 constant if overflow does not change.
1642 2007-02-01 Ralf Wildenhues <Ralf.Wildenhues@gmx.de>
1644 * doc/c-tree.texi (Expression trees): Improve markup.
1645 * doc/tm.texi (Register Classes, Addressing Modes)
1646 (Floating Point): Fix spacing after abbreviations. Fix some
1649 2007-02-01 Ben Elliston <bje@au.ibm.com>
1651 * doc/invoke.texi: Replace "bugfix" with "bug fix" throughout.
1652 * doc/contrib.texi: Likewise.
1653 * doc/install.texi: Likewise.
1655 2007-01-31 Richard Henderson <rth@redhat.com>
1656 Ian Lance Taylor <iant@google.com>
1658 * lower-subreg.c: New file.
1659 * rtl.def (CONCATN): Define.
1660 * passes.c (init_optimization_passes): Add pass_lower_subreg and
1662 * emit-rtl.c (update_reg_offset): New static function, broken out
1663 of gen_rtx_REG_offset.
1664 (gen_rtx_REG_offset): Call update_reg_offset.
1665 (gen_reg_rtx_offset): New function.
1666 * regclass.c: Revert patch of 2006-03-05, restoring
1668 (clear_reg_info_regno): New function.
1669 * dwarf2out.c (concatn_loc_descriptor): New static function.
1670 (loc_descriptor): Handle CONCATN.
1671 * common.opt (fsplit_wide_types): New option.
1672 * opts.c (decode_options): Set flag_split_wide_types when
1674 * timevar.def (TV_LOWER_SUBREG): Define.
1675 * rtl.h (gen_reg_rtx_offset): Declare.
1676 (reg_scan_update): Declare.
1677 * regs.h (clear_reg_info_regno): Declare.
1678 * tree-pass.h (pass_lower_subreg): Declare.
1679 (pass_lower_subreg2): Declare.
1680 * doc/invoke.texi (Option Summary): List -fno-split-wide-types.
1681 (Optimize Options): Add -fsplit-wide-types to -O1 list. Document
1683 * doc/rtl.texi (Regs and Memory): Document concat and concatn.
1684 * Makefile.in (OBJS-common): Add lower-subreg.o.
1685 (lower-subreg.o): New target.
1687 2007-01-31 Kazu Hirata <kazu@codesourcery.com>
1689 * config/sh/sh.h (HAVE_SECONDARY_RELOADS): Remove.
1691 2007-01-31 Anatoly Sokolov <aesok@post.ru>
1694 * config/avr/avr.c (DWARF2_ADDR_SIZE): Define.
1696 2007-01-31 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
1699 * builtins.c (fold_builtin_sqrt): Use MPFR for constant args.
1701 2007-01-31 Zdenek Dvorak <dvorakz@suse.cz>
1703 * cfgloop.h: Include vec-prim.h.
1704 (enum li_flags): Remove LI_ONLY_OLD.
1705 (loop_iterator): Changed.
1706 (fel_next, fel_init): Iterate over loop tree.
1707 (FOR_EACH_LOOP_BREAK): New macro.
1708 * loop-unswitch.c (unswitch_loops): Do not pass LI_ONLY_OLD to
1710 * tree-ssa-loop-unswitch.c (tree_ssa_unswitch_loops): Ditto.
1711 * modulo-sched.c (sms_schedule): Ditto.
1712 * tree-vectorizer.c (vectorize_loops): Ditto.
1713 * doc/loop.texi: Update information on loop numbering and behavior of
1714 FOR_EACH_LOOP wrto new loops.
1715 * tree-scalar-evolution.c (compute_overall_effect_of_inner_loop,
1716 add_to_evolution_1): Test nestedness of loops instead of comparing
1718 * tree-chrec.c (chrec_fold_plus_poly_poly,
1719 chrec_fold_multiply_poly_poly, chrec_evaluate,
1720 hide_evolution_in_other_loops_than_loop, chrec_component_in_loop_num,
1721 reset_evolution_in_loop): Ditto.
1722 * Makefile.in (CFGLOOP_H): Add vecprim.h dependency.
1724 2007-01-31 Dirk Mueller <dmueller@suse.de>
1726 * c-common.c (warn_about_parentheses): Separate warning about
1727 un-parenthized sequence of comparison operators from the one
1728 which is supposed to warn about x <= y <= z.
1730 2007-01-31 Uros Bizjak <ubizjak@gmail.com>
1732 * optabs.h (enum optab_index): Add new OTI_isinf.
1733 (isinf_optab): Define corresponding macro.
1734 * optabs.c (init_optabs): Initialize isinf_optab.
1735 * genopinit.c (optabs): Implement isinf_optab using isinf?f2
1737 * builtins.c (mathfn_built_in): Handle BUILT_IN_ISINF{,F,L}.
1738 (expand_builtin_interclass_mathfn): Expand BUILT_IN_ISINF{,F,L}
1740 (expand_builtin): Expand BUILT_IN_ISINF{,F,L} using
1741 expand_builtin_interclass_mathfn.
1742 * reg_stack.c (subst_stack_regs_pat): Handle UNSPEC_FXAM.
1743 * config/i386/i386.md (UNSPEC_FXAM): New constant.
1744 (fxam<mode>2_i387): New insn pattern.
1745 (isinf<mode>2) New expander to implement isinf, isinff and isinfl
1746 built-in functions as x87 inline asm.
1748 2007-01-31 Kazu Hirata <kazu@codesourcery.com>
1750 * gcc/config/arm/unwind-arm.h (_sleb128_t, _uleb128_t): New.
1752 2007-01-30 Eric Christopher <echristo@apple.com>
1754 * config.gcc: Add geode.
1756 2007-01-31 Kazu Hirata <kazu@codesourcery.com>
1758 * cgraphunit.c, config/arm/arm.c, config/m68k/m68k.c,
1759 ipa-inline.c, tree-profile.c, tree-ssa-live.c,
1760 tree-ssa-math-opts.c, tree-ssanames.c, tree-vect-analyze.c,
1761 value-prof.c: Fix comment typos.
1763 2007-01-30 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
1766 * doc/invoke.texi (Wpointer-arith): Document warning.
1768 2007-01-30 Janis Johnson <janis187@us.ibm.com>
1770 * doc/extend.texi (Decimal Floating Types): Remove decfloat.h from
1771 the list of discrepancies from the draft TR.
1773 2007-01-30 Dirk Mueller <dmueller@suse.de>
1776 * doc/invoke.texi (-Wreturn-type): Update description to
1779 2007-01-30 Richard Sandiford <richard@codesourcery.com>
1781 * cfgrtl.c (try_redirect_by_replacing_jump): Check only_sets_cc0_p.
1783 2007-01-30 Uros Bizjak <ubizjak@gmail.com>
1785 * builtins.c (expand_builtin_int_interclass_roundingfn): New function
1786 to handle optabs that operate on floating point input argument and
1787 output to integer output.
1788 (expand_builtin_mathfn) [BUILT_IN_ILOGB]: Move from here ...
1789 (expand_builtin_interclass_mathfn) [BUILT_IN_ILOGB]: ... to here.
1790 (expand_builtin): Expand BUILT_IN_ILOGB{,F,L} using
1791 expand_builtin_interclass_mathfn ().
1792 * config/i386/i386.md (fxtractxf3_i387): Rename from *fxtractxf3_i387.
1794 (ilogbxf2, ilogb<mode>2): New expanders to implement ilogb, ilogbf and
1795 ilogbl built-in functions as x87 intrinsics.
1797 2007-01-30 Richard Guenther <rguenther@suse.de>
1800 * dwarf2out.c (reference_to_unused): Query varpool if the
1801 variable was output.
1803 2007-01-30 Richard Guenther <rguenther@suse.de>
1806 * passes.c (execute_one_pass): Reset in_gimple_form to not
1807 confuse non-unit-at-a-time mode.
1809 2007-01-29 Roger Sayle <roger@eyesopen.com>
1810 Richard Guenther <rguenther@suse.de>
1812 * fold-const.c (round_up): Make HIGH an unsigned HOST_WIDE_INT to
1813 avoid undefined behaviour on overflow. Use force_fit_type_double
1814 to construct the constant with the specified TREE_OVERFLOW.
1816 2007-01-29 Janis Johnson <janis187@us.ibm.com>
1818 * config/dfp-bit.c: Add parameterized support for fp exceptions.
1819 * config/dfp-bit.h: Ditto.
1821 2007-01-29 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
1823 * c-decl.c (pop_scope): Replace warnings with call to
1824 warn_for_unused_label.
1825 * c-common.h (warn_for_unused_label): Declare.
1826 * c-common.c (warn_for_unused_label): Define.
1828 2007-01-29 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
1830 * tree-optimize.c (update_inlined_to_pointers): Delete unused
1833 2007-01-29 Janis Johnson <janis187@us.ibm.com>
1835 * Makefile.in (USER_H): Remove decfloat.h.
1836 * ginclude/decfloat.h: Delete, moving contents to ...
1837 * ginclude/float.h: Add support for decimal floating point,
1838 guarded by __STDC_WANT_DEC_FP__.
1840 2007-01-29 Mike Stump <mrs@apple.com>
1842 * doc/gccint.texi (Top): Rename Loop Representation to Loop
1843 Analysis and Representation to resolve case insensitive conflict.
1844 * doc/loop.texi (Loop Analysis and Representation): Likewise.
1846 2007-01-28 Daniel Berlin <dberlin@dberlin.org>
1848 * tree.h (struct tree_memory_tag): Add aliases member.
1849 (MTAG_ALIASES): New macro.
1850 * tree-ssa-alias.c (alias_bitmap_obstack): New variable.
1851 (add_may_alias): Remove pointer-set. Update for may_aliases being
1853 (mark_aliases_call_clobbered): Update for may_aliases being a
1855 (compute_tag_properties): Ditto.
1856 (create_partition_for): Ditto.
1857 (compute_memory_partitions): Ditto.
1858 (dump_may_aliases_for): Ditto.
1859 (is_aliased_with): Ditto.
1860 (add_may_alias_for_new_tag): Ditto.
1861 (rewrite_alias_set_for): Rewrite for may_aliases being a bitmap.
1862 (compute_is_aliased): New function.
1863 (compute_may_aliases): Call compute_is_aliased).
1864 (init_alias_info): Initialize alias_bitmap_obstack.
1865 (union_alias_set_into): New function.
1866 (compute_flow_sensitive_aliasing): Use union_aliases_into.
1867 (have_common_aliases_p): Rewrite to take two bitmaps and use
1869 (compute_flow_insensitive_aliasing): Stop using pointer-sets.
1871 (finalize_ref_all_pointers): Update for add_may_alias changes.
1872 (new_type_alias): Ditto.
1873 * tree-flow-inline.h (may_aliases): Return a bitmap.
1874 * tree-dfa.c (dump_variable): Check for MTAG_P'ness.
1875 * tree-ssa.c (verify_flow_insensitive_alias_info): Update for
1876 may_aliases being a bitmap.
1877 * tree-flow.h (struct var_ann_d): Remove may_aliases member.
1878 may_aliases now returns a bitmap.
1879 * tree-ssa-structalias.c (merge_smts_into): Update for may_aliases
1881 * tree-ssa-operands.c (add_virtual_operand): Update for
1882 may_aliases being a bitmap.
1884 2007-01-29 Daniel Berlin <dberlin@dberlin.org>
1886 PR tree-optimization/30630
1887 * tree-ssa-structalias.c (do_complex_constraint): Mark correct
1888 variable as changed.
1890 2007-01-29 Simon Martin <simartin@users.sourceforge.net>
1893 * gimplify.c (gimplify_target_expr): Make sure that the TARGET_EXPR is
1894 expanded only once even if an error occurs.
1896 2007-01-29 Ben Elliston <bje@au.ibm.com>
1898 * gcov-io.h (__gcov_indirect_call_profiler): Declare.
1899 (__gcov_average_profiler): Likewise.
1900 (__gcov_ior_profiler): Likewise.
1901 (__gcov_merge_ior): Likewise.
1903 2007-01-28 Jan Hubicka <jh@suse.cz>
1905 * builtins.c (expand_builtin_memset): Fix typo in my last patch.
1906 * value-prof.c (stringop_block_profile): Likewise.
1908 2007-01-28 Jan Hubicka <jh@suse.cz>
1910 * expr.c (emit_block_move_via_movmem, emit_block_move_via_libcall): Add
1911 variant handling histograms; add wrapper.
1912 (clear_storage_via_libcall): Export.
1913 (emit_block_move_hints): Break out from ...; add histograms.
1914 (emit_block_move): ... this one.
1915 (clear_storage_hints): Break out from ...; add histograms.
1916 (clear_storage): ... this one.
1917 (set_storage_via_memset): Handle histogram.
1918 * expr.h (emit_block_move_via_libcall, emit_block_move_hints): Declare.
1919 (clear_storage_hints, clear_storage_via_libcall): Declare.
1920 (set_storage_via_setmem): Update prototype.
1921 * doc/md.texi (movmem, setmem): Document new arguments.
1923 * value-prof.c (dump_histogram_value, tree_find_values_to_profile): Add
1925 (stringop_block_profile): New global function.
1926 (tree_stringops_values_to_profile): Profile block size and alignment.
1927 * value-prof.h (enum hist_type): add HIST_TYPE_AVERAGE and
1929 (struct profile_hooks): Add gen_average_profiler and gen_ior_profiler.
1930 (stringop_block_profile): Declare.
1931 * builtins.c: Include value-prof.h.
1932 (expand_builtin_memcpy, expand_builtin_memset): Pass block profile.
1933 * gcov-ui.h (GCOV_COUNTER_NAMES): Add new counter.
1934 (GCOV_COUNTER_AVERAGE, GCOV_COUNTER_IOR): New constants.
1935 (GCOV_COUNTERS, GCOV_LAST_VALUE_COUNTER): Update.
1936 * profile.c (instrument_values): Add new counters.
1937 * cfgexpand.c (expand_gimple_basic_block): Propagate histograms to
1939 * tree-profile.c (tree_average_profiler_fn, tree_ior_profiler_fn): New.
1940 (tree_init_edge_profiler): Build new profilers.
1941 (tree_gen_average_profiler, tree_gen_ior_profiler): New.
1942 (pass_tree_profile): Add dump.
1943 (tree_profile_hooks): Update.
1944 * Makefile.in (LIBGCOV): Add new constants.
1945 * libgcov.c (__gcov_merge_ior, __gcov_average_profiler,
1946 __gcov_ior_profiler): New.
1947 * i386.md (movmem/setmem expanders): Add new optional arguments.
1949 2007-01-28 David Edelsohn <edelsohn@gnu.org>
1951 * doc/md.texi (Standard Pattern Names): Document blockage pattern.
1953 2007-01-28 Zdenek Dvorak <dvorakz@suse.cz>
1955 * tree-ssa-loop-unswitch.c: Include tree-inline.h.
1956 (tree_unswitch_single_loop): Pass eni_size_weights to
1957 tree_num_loop_insns.
1958 * tree-ssa-loop-manip.c: Include tree-inline.h.
1959 (can_unroll_loop_p): Pass eni_size_weights to
1960 tree_num_loop_insns.
1961 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p):
1962 Pass eni_size_weights to estimate_num_insns.
1963 * tree.h (init_inline_once): Export.
1964 * toplev.c (backend_init): Call init_inline_once.
1965 * cgraphunit.c (cgraph_process_new_functions,
1966 cgraph_analyze_function): Pass eni_inlining_weights to
1968 * ipa-inline.c (compute_inline_parameters): Ditto.
1969 * tree-ssa-loop-ivcanon.c (tree_num_loop_insns): Pass weights
1970 to estimate_num_insns.
1971 (try_unroll_loop_completely): Pass eni_size_weights to
1972 tree_num_loop_insns.
1973 * tree-eh.c (decide_copy_try_finally): Pass eni_size_weights
1974 ot estimate_num_insns.
1975 * tree-ssa-loop-prefetch.c: Include tree-inline.h.
1976 (loop_prefetch_arrays): Pass eni_time_weights to tree_num_loop_insns.
1977 * tree-inline.c (eni_inlining_weights, eni_size_weights,
1978 eni_time_weights): New variables.
1979 (init_inline_once): Initialize them.
1980 (struct eni_data): Mew.
1981 (estimate_num_insns_1, estimate_num_insns): Use weights.
1982 * tree-inline.h (struct eni_weights_d): New.
1983 (eni_inlining_weights, eni_size_weights, eni_time_weights): Declare.
1984 (estimate_num_insns): Declaration changed.
1985 * cfgloop.h (tree_num_loop_insns): Declaration changed.
1986 * Makefile.in (tree-ssa-loop-unswitch.o, tree-ssa-loop-prefetch.o,
1987 tree-ssa-loop-manip.o): Add TREE_INLINE_H dependency.
1989 2007-01-28 Zdenek Dvorak <dvorakz@suse.cz>
1991 * tree-data-ref.c (conflict_fn): Assert that the number of affine
1992 relations in the conflict function is valid.
1994 2007-01-27 Ian Lance Taylor <iant@google.com>
1996 * common.opt: Add fstrict-overflow.
1997 * opts.c (decode_options): Set flag_strict_overflow if -O2.
1998 * flags.h (TYPE_OVERFLOW_WRAPS): Define.
1999 (TYPE_OVERFLOW_UNDEFINED): Define.
2000 (TYPE_OVERFLOW_TRAPS): Define. This replaces TYPE_TRAP_SIGNED.
2002 * tree.h (TYPE_TRAP_SIGNED): Don't define.
2003 * fold-const.c (negate_expr_p): Use TYPE_OVERFLOW_UNDEFINED.
2004 (fold_negate_expr): Likewise.
2005 (make_range): Likewise.
2006 (extract_muldiv_1): Likewise.
2007 (maybe_canonicalize_comparison): Likewise.
2008 (fold_comparison): Likewise.
2009 (fold_binary): Likewise.
2010 (tree_expr_nonnegative_p): Likewise.
2011 (tree_expr_nonzero_p): Likewise.
2012 * tree-vrp.c (compare_values): Likewise.
2013 (extract_range_from_binary_expr): Likewise.
2014 (extract_range_from_unary_expr): Likewise.
2015 * tree-ssa-loop-niter.c (infer_loop_bounds_from_signedness):
2017 (nowrap_type_p): Likewise.
2018 * tree-scalar-evolution.c (simple_iv): Likewise.
2019 * fold-const.c (negate_expr_p): Use TYPE_OVERFLOW_WRAPS.
2020 (build_range_check): Likewise.
2021 (extract_muldiv_1): Likewise.
2022 (fold_comparison): Likewise.
2023 * tree-vrp.c (vrp_int_const_binop): Likewise.
2024 (extract_range_from_unary_expr): Likewise.
2025 * convert.c (convert_to_integer): Likewise.
2026 * fold-const.c (fold_negate_expr): Use TYPE_OVERFLOW_TRAPS.
2027 (fold_comparison): Likewise.
2028 (fold_binary): Likewise.
2029 * optabs.c (optab_for_tree_code): Likewise.
2030 * tree-vectorizer.c (vect_is_simple_reduction): Likewise.
2031 * simplify-rtx.c (simplify_const_relational_operation): Check
2032 flag_strict_overflow and flag_trapv.
2033 (simplify_const_relational_operation): Likewise.
2034 * doc/invoke.texi (Option Summary): Mention -fstrict-overflow.
2035 (Optimize Options): Add -fstrict-overflow to -O2 list. Document
2038 2007-01-27 Roger Sayle <roger@eyesopen.com>
2040 * tree.c (tree_fold_gcd): Delete.
2041 * tree.h (tree_fold_gcd): Remove prototype.
2042 * tree-data-ref.c (tree_fold_divides_p): Don't use tree_fold_gcd to
2043 test whether one constant integer is a multiple of another. Instead
2044 call int_const_binop with TRUNC_MOD_EXPR and test for a zero result.
2045 * fold-const.c (multiple_of_p): We've determined both TOP and
2046 BOTTOM are integer constants so we can call int_const_binop directly
2047 instead of the more generic const_binop.
2049 2007-01-27 Roger Sayle <roger@eyesopen.com>
2051 * fold-const.c (size_binop): In the fast-paths for X+0, 0+X, X-0 and
2052 1*X check that the constant hasn't overflowed, to preserve the
2054 (round_up): Provide an efficient implementation when rouding-up an
2055 INTEGER_CST to a power-of-two.
2057 2007-01-28 Ralf Wildenhues <Ralf.Wildenhues@gmx.de>
2059 * doc/sourcebuild.texi: Add comma for clarity.
2060 * doc/extend.texi: Fix some typos.
2061 * doc/passes.texi: Likewise.
2062 * doc/cppinternals.texi: Likewise.
2063 * doc/c-tree.texi: Likewise.
2064 * doc/tree-ssa.texi: Likewise.
2065 * doc/install.texi: Likewise.
2067 2007-01-27 Jan Hubicka <jh@suse.cz>
2069 * tree-sra.c (sra_walk_function): Don't rely on aliases being build.
2070 (pass_sra): Do not require alias information.
2071 * passes.c (init_optimization_passes): Add SRA
2073 2007-01-27 Steven Bosscher <steven@gcc.gnu.org>
2075 * tracer.c (rest_of_handle_tracer): We already cleaned
2076 up the CFG in tracer() so don't do it here again.
2077 * cfgcleanup.c (rest_of_handle_jump2): Don't repeat
2078 cleanup_cfg here, either. And don't call renumber_insns.
2080 * cfgrtl.c (rtl_verify_flow_info_1): Don't verify that BB_END
2081 and BB_HEAD are in the insn stream here. Instead make sure
2082 that BB_INSN is valid on all insns. Also, do check here that
2083 there are no pending branch predictions...
2084 (rtl_verify_flow_info): ...instead of doing it here. Checks
2085 for BB_END and BB_HEAD moved from rtl_verify_flow_info_1 to
2088 2007-01-26 Roger Sayle <roger@eyesopen.com>
2090 * config/i386/i386.c (ix86_swap_binary_operands_p): New helper
2091 function to simplify/factorize operand order canonicalization.
2092 (ix86_fixup_binary_operands): Reorganize using the above function.
2093 (ix86_binary_operator_ok): Likewise.
2095 2007-01-27 Jakub Jelinek <jakub@redhat.com>
2097 * genattrtab.c (struct attr_value_list, insn_code_values): Move to
2098 file scope from optimize_attrs.
2099 (simplify_test_exp): If insn_code_values is not NULL, use it to speed
2101 (optimize_attrs): Clear insn_code_values after freeing it.
2103 2007-01-26 Zdenek Dvorak <dvorakz@suse.cz>
2105 * tree-ssa-address.c (create_mem_ref): Remove ", bsi" from
2106 a parts.base assignment.
2108 2007-01-26 Zdenek Dvorak <dvorakz@suse.cz>
2110 * tree-data-ref.c (dump_subscript): Use dump_conflict_function.
2111 (compute_subscript_distance, initialize_data_dependence_relation,
2112 finalize_ddr_dependent, analyze_ziv_subscript,
2113 analyze_siv_subscript_cst_affine,
2114 compute_overlap_steps_for_affine_univar,
2115 compute_overlap_steps_for_affine_1_2, analyze_subscript_affine_affine,
2116 analyze_siv_subscript, analyze_miv_subscript,
2117 analyze_overlapping_iterations, subscript_dependence_tester_1,
2118 compute_self_dependence, free_dependence_relation): Work
2119 with affine_fn instead of chrecs.
2120 (dump_affine_function, dump_conflict_function, affine_function_equal_p,
2121 common_affine_function, affine_function_base,
2122 affine_function_constant_p, affine_fn_op, affine_fn_plus,
2123 affine_fn_minus, affine_fn_free, conflict_fn_not_known,
2124 conflict_fn_no_dependence, free_conflict_function, free_subscripts,
2125 conflict_fn, affine_fn_cst, affine_fn_univar): New functions.
2126 (all_chrecs_equal_p): Removed.
2127 * tree-data-ref.h (affine_fn, conflict_function): New types.
2128 (struct subscript): Change type of conflicting_iterations_in_a
2129 and conflicting_iterations_in_b.
2131 2007-01-26 Steve Ellcey <sje@cup.hp.com>
2134 * config/pa/pa.h (TARGET_HPUX_11): New.
2135 * config/pa/pa-hpux11.h (TARGET_HPUX_11): New.
2136 * config/pa/pa.c (pa_init_builtins): Use TARGET_HPUX_11.
2138 2007-01-26 Daniel Berlin <dberlin@dberlin.org>
2139 Richard Guenther <rguenther@suse.de>
2141 * tree-ssa-structalias.c (solve_graph): Handle case
2142 we merged the variable to another.
2144 2007-01-25 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
2146 * builtins.c (fold_builtin_1): Treat ccos and ccosh as 'even'
2149 * fold-const.c (negate_mathfn_p): Treat casin, casinh, catan,
2150 catanh, cproj, csin, csinh, ctan and ctanh as 'odd' functions.
2152 2007-01-25 DJ Delorie <dj@redhat.com>
2154 * config/m32c/m32c.c (m32c_cannot_change_mode_class): We don't
2155 allow changes to modes which don't fit in those registers.
2157 * reload1.c (choose_reload_regs): Check for invalid subregs before
2158 computing their locations, not after.
2160 2007-01-25 Geoffrey Keating <geoffk@apple.com>
2163 * config/rs6000/rs6000.c (first_altivec_reg_to_save): On Darwin,
2164 save Altivec registers in an eh_return function.
2165 (compute_vrsave_mask): Likewise.
2166 (rs6000_stack_info): Correct AIX/Darwin stack alignment computation
2167 for saving Altivec registers.
2168 (rs6000_emit_prologue): Don't allocate stack twice in
2169 eh_return function. Correct expected value of altivec_save_offset
2170 when using save_world. Describe save of R0 to stack when using
2171 save_world. Describe stack pointer adjustment when using
2172 save_world. Remove duplicated eh_return parameter register saving.
2173 Update sp_offset variable after save_world.
2174 * config/rs6000/t-darwin (LIB2FUNCS_STATIC_EXTRA): Remove
2176 (LIB2FUNCS_EXTRA): Add darwin-world.asm.
2177 * config/rs6000/darwin.h (SUBTARGET_OVERRIDE_OPTIONS): -m64
2180 2007-01-25 Steve Ellcey <sje@cup.hp.com>
2182 * config.gcc (ia64*-*-hpux*): Make posix threads the default.
2184 2007-01-25 Steve Ellcey <sje@cup.hp.com>
2187 * config/pa/pa.c (pa_init_builtins): Set asm names for finite routines.
2188 * config/ia64/ia64.c (ia64_init_builtins): Ditto.
2190 2007-01-25 Richard Guenther <rguenther@suse.de>
2192 * doc/invoke.texi (-Wcoverage-mismatch): Document.
2193 * common.opt (-Wcoverage-mismatch): New warning option.
2194 * coverage.c (get_coverage_counts): Ignore coverage mismatch
2195 if -Wcoverage-mismatch is given.
2197 2007-01-25 Razya Ladelsky <razya@il.ibm.com>
2199 * ipa-cp.c (ipcp_insert_stage, ipcp_driver): Support for SSA.
2200 (ipcp_driver): Change to static definition.
2201 Add dumping of the ifunctions.
2202 (constant_val_insert): Remove unused parameter. Support for SSA.
2203 (ipcp_propagate_const): Support for SSA.
2204 (ipcp_profile_bb_print): Print only analyzed nodes.
2205 (ipcp_replace_map_create): Remove support for Fortran constant
2207 * ipa-prop.c (ipa_method_modify_stmt,
2208 ipa_callsite_compute_param): Support for SSA.
2209 * ipa-prop.h (ipcp_driver): Remove declaration.
2210 (IS_VALID_TREE_MAP_INDEX): Add define.
2212 2007-01-24 Geoffrey Keating <geoffk@apple.com>
2214 * unwind-dw2.c (execute_stack_op): Handle DW_OP_swap.
2216 2007-01-24 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
2219 * builtins.c (fold_builtin_cabs): Use MPFR to evaluate a
2220 constant argument to cabs and do it without checking for
2221 -funsafe-math-optimizations.
2223 2007-01-24 Douglas Gregor <dgregor@osl.iu.edu>
2225 * c-common.h (RID_FIRST_CXX0X): New.
2226 (RID_LAST_CXX0X): New.
2227 * c-opts.c (c_common_handle_option): -Wc++0x-compat is triggered
2229 * c.opt (Wc++0x-compat): New.
2230 * doc/invoke.texi (-Wc++0x-compat): Document.
2232 2007-01-24 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
2234 * builtins.c (fold_builtin_carg): New.
2235 (fold_builtin_1): Use it.
2237 2007-01-24 Jan Hubicka <jh@suse.cz>
2239 * ipa-inline.c (cgraph_decide_inlining): Initialize initial_insns.
2241 * ipa-inline.c (initial_insns, max_insns): Delete.
2242 (compute_max_insns): New function.
2243 (cgraph_decide_inlining_of_small_function): Use it; take minimal amount
2244 of insns as base for code growth.
2245 (cgraph_decide_inlining): Make initial_insns local; do not compute
2247 * params.def (PARAM_INLINE_UNIT_GROWTH): Set to 60.
2248 * doc/invoke.texi (inline-unit-growth): Update docs.
2250 2007-01-24 Jakub Jelinek <jakub@redhat.com>
2252 * config/i386/i386.h (x86_cmpxchg16b): Remove const.
2253 (TARGET_CMPXCHG16B): Define to x86_cmpxchg16b.
2254 * config/i386/i386.c (x86_cmpxchg16b): Remove const.
2255 (override_options): Add PTA_CX16 flag. Set x86_cmpxchg16b
2256 for CPUs that have PTA_CX16 set.
2259 * gimplify.c (omp_check_private): New function.
2260 (gimplify_scan_omp_clauses): Use it for
2261 firstprivate/lastprivate/reduction.
2264 * gimplify.c (omp_add_variable): Don't call omp_notice_variable
2265 on TYPE_SIZE_UNIT for GOVD_LOCAL VLAs.
2268 * omp-low.c (lower_omp_for_lastprivate): Add dlist argument.
2269 If lower_lastprivate_clauses emits some statements, append them
2270 to dlist rather than body_p and to body_p append an initializer.
2271 (lower_omp_for): Adjust caller.
2273 2007-01-24 Steve Ellcey <sje@cup.hp.com>
2275 * target.h (globalize_decl_name): New.
2276 * target-def.h (TARGET_ASM_GLOBALIZE_DECL_NAME): New.
2277 * output.h (default_globalize_decl_name): New.
2278 * varasm.c (asm_output_bss): Use globalize_decl_name instead of
2280 (globalize_decl): Ditto.
2281 (default_globalize_decl_name): New.
2282 * config/ia64/ia64.c (ia64_globalize_decl_name): New.
2283 (ia64_handle_version_id_attribute): New.
2284 (TARGET_ASM_GLOBALIZE_DECL_NAME): New.
2285 (ia64_asm_output_external): Use globalize_decl_name instead
2287 * doc/extend.texi (version_id): New pragma.
2288 * doc/tm.texi (ARGET_ASM_GLOBALIZE_DECL_NAME): New target hook.
2290 2007-01-24 Andreas Krebbel <krebbel1@de.ibm.com>
2292 * unwind-dw2-fde.c (get_cie_encoding): Replaced _Unwind_Word with
2293 _uleb128_t and _Unwind_SWord with _sleb128_t.
2294 * unwind-dw2.c (extract_cie_info, execute_stack_op, execute_cfa_program,
2295 uw_frame_state_for, uw_update_context_1): Likewise.
2296 * unwind-c.c (parse_lsda_header, PERSONALITY_FUNCTION): Likewise.
2297 * unwind-pe.h (read_uleb128, read_sleb128,
2298 read_encoded_value_with_base): Likewise.
2299 * unwind-generic.h: Define _sleb128_t and _uleb128_t types.
2301 2007-01-24 Richard Guenther <rguenther@suse.de>
2303 * builtins.c (expand_builtin_cexpi): Get the fndecl
2304 for cexp in the correct way.
2306 2007-01-24 Jan Hubicka <jh@suse.cz>
2308 * tree-ssa-dce.c (eliminate_unnecesary_stmts): Remove dead LHS of calls.
2310 2007-01-24 Andreas Krebbel <krebbel1@de.ibm.com>
2312 * c-cppbuiltin.c (builtin_define_type_sizeof): New function.
2313 (c_cpp_builtins): New builtin macros: __SIZEOF_INT__, __SIZEOF_LONG__,
2314 __SIZEOF_LONG_LONG__, __SIZEOF_SHORT__, __SIZEOF_POINTER__,
2315 __SIZEOF_FLOAT__, __SIZEOF_DOUBLE__, __SIZEOF_LONG_DOUBLE__,
2316 __SIZEOF_SIZE_T__, __SIZEOF_WCHAR_T__, __SIZEOF_WINT_T__ and
2317 __SIZEOF_PTRDIFF_T__.
2318 * doc/cpp.texi: Documentation for the new builtin macros added.
2320 2007-01-24 Uros Bizjak <ubizjak@gmail.com>
2322 * config/i386/i386.md (tanxf2, tan<mode>2, atan<mode>2, log<mode>2,
2323 log10<mode>2, log2<mode>2, expxf2, exp10xf2, exp2xf2): Use op2
2324 instead of operands[2] to avoid access past the end of array.
2326 2007-01-24 Richard Sandiford <richard@codesourcery.com>
2328 * reload1.c (emit_reload_insns): Pass the reload register
2329 for a non-spill output reload through forget_old_reloads_1.
2331 2007-01-23 Joseph Myers <joseph@codesourcery.com>
2333 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Define
2334 _SOFT_DOUBLE if doubles use software floating-point.
2335 * config/rs6000/libgcc-ppc-glibc.ver: Export additional long
2336 double functions if _SOFT_DOUBLE, not _SOFT_FLOAT.
2337 * config/rs6000/darwin-ldouble.c: Also compile functions for
2338 hard-float without FPRs. Use fmsub function for all __NO_FPRS__
2339 cases. Compile extra functions if _SOFT_DOUBLE, not _SOFT_FLOAT.
2340 * config/rs6000/linuxspe.h (SUBSUBTARGET_OVERRIDE_OPTIONS): Remove
2341 commented-out long double override.
2342 (CPP_LONGDOUBLE_DEFAULT_SPEC): Likewise.
2343 * config/rs6000/eabispe.h: Likewise.
2344 * config/rs6000/rs6000.c (rs6000_override_options): Don't override
2345 long double for non-SPE.
2346 (rs6000_handle_option): Likewise.
2347 (invalid_e500_subreg): Disallow more subregs involding DImode,
2348 DFmode, TImode or TFmode.
2349 (rs6000_legitimate_offset_address_p): Check TFmode offsets for
2351 (legitimate_lo_sum_address_p): Also check for TFmode for E500
2353 (rs6000_legitimize_address): Also handle TFmode for E500 double.
2354 (rs6000_legitimize_reload_address): Also handle TFmode for E500
2356 (rs6000_legitimate_address): Also check for TFmode for E500
2358 (rs6000_emit_move): Use DFmode subregs of TFmode for E500 double.
2359 (spe_build_register_parallel): Handle TFmode and TCmode.
2360 (rs6000_spe_function_arg): Handle TFmode and TCmode for E500
2362 (function_arg): Handle TFmode and TCmode for E500 double.
2363 (rs6000_init_libfuncs): Initialize extra libfuncs for soft double
2365 (print_operand): Handle TFmode and TImode for %y.
2366 (rs6000_generate_compare): Handle TFmode comparisons for E500
2368 (spe_func_has_64bit_regs_p): Check for TFmode for E500 double.
2369 (rs6000_function_value): Handle TFmode and TCmode for E500 double.
2370 (rs6000_libcall_value): Handle TFmode and TCmode for E500 double.
2371 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Check for
2372 TFmode for E500 double.
2373 * config/rs6000/rs6000.md (FP): Allow TF for E500 double.
2374 (floatsidf2): Enable for E500 double.
2375 (movtf_softfloat): Use rs6000_nonimmediate_operand.
2376 (extenddftf2): Change to extenddftf2_fprs.
2377 (extenddftf2): Call gen_spe_extenddftf2 or gen_extenddftf2_fprs
2378 depending on TARGET_E500_DOUBLE.
2379 (extendsftf2): Enable for E500 double.
2380 (trunctfdf2): Enable for E500 double.
2381 (trunctfsf2): Change to trunctfsf2_fprs.
2382 (trunctfsf2): Call gen_spe_trunctfsf2 or gen_trunctfsf2_fprs
2383 depending on TARGET_E500_DOUBLE.
2384 (floatsitf2): Enable for E500 double.
2385 (fix_trunctfsi2): Change to fix_trunctfsi2_fprs.
2386 (fix_trunctfsi2): Call gen_spe_fix_trunctfsi2 or
2387 gen_fix_trunctfsi2_fprs depending on TARGET_E500_DOUBLE.
2388 (negtf2): Change to negtf2_internal.
2389 (negtf2): New expander.
2390 (abstf2): Enable for E500 double. Call gen_spe_abstf2_tst,
2391 gen_spe_abstf2_cmp or gen_abstf2_internal depending on
2392 TARGET_E500_DOUBLE and flag_unsafe_math_optimizations.
2393 (movdi_internal32): Use rs6000_nonimmediate_operand.
2394 (unnamed splitter): Likewise.
2395 * config/rs6000/spe.md (CMPTFEQ_GPR, TSTTFEQ_GPR, CMPTFGT_GPR,
2396 TSTTFGT_GPR, CMPTFLT_GPR, TSTTFLT_GPR): New unspecs.
2397 (SPE64TF, DITI): New mode macros.
2398 (frob_df_di): Change to frob_<SPE64:mode>_<DITI:mode>; allow more
2401 (frob_<mode>_di_2): New.
2402 (frob_tf_di_8_2): New.
2403 (frob_di_df): Change to frob_di_<mode>; allow more modes.
2405 (frob_di_df_2): Change to frob_<DITI:mode>_<SPE64:mode>_2; allow
2407 (frob_ti_<mode>_8_2): New.
2408 (frob_ti_tf_2): New.
2409 (mov_si<mode>_e500_subreg0, mov_si<mode>_e500_subreg0_2,
2410 mov_si<mode>_e500_subreg4, mov_si<mode>_e500_subreg4_2): Allow
2412 (mov_sitf_e500_subreg8, mov_sitf_e500_subreg8_2,
2413 mov_sitf_e500_subreg12, mov_sitf_e500_subreg12_2): New.
2414 (spe_trunctfdf2_internal1, spe_trunctfsf2, spe_extenddftf2,
2415 spe_fix_trunctfsi2, spe_fix_trunctfsi2_internal,
2416 spe_negtf2_internal, spe_abstf2_cmp, spe_abstf2_tst): New.
2417 (cmptfeq_gpr, tsttfeq_gpr, cmptfgt_gpr, tsttfgt_gpr, cmptflt_gpr,
2420 2007-01-23 Ian Lance Taylor <iant@google.com>
2422 * Makefile.in (OBJS-common): Reformat, alphabetize, but put
2424 (OBJS-archive): Reformat, alphabetize.
2425 (OBJS): Change out_object_file to OBJS-md.
2427 2007-01-23 Uros Bizjak <ubizjak@gmail.com>
2429 * config/i386/i386.md: Use REG_P, MEM_P, CONST_INT_P, LABEL_P,
2430 JUMP_P and CALL_P predicates where applicable.
2431 * config/i386/i386.c: Ditto.
2432 * config/i386/i386.md: Ditto.
2433 * config/i386/mmx.md: Ditto.
2434 * config/i386/predicates.md: Ditto.
2436 2007-01-22 Andreas Schwab <schwab@suse.de>
2438 * config/m68k/m68k.h: Fix comment.
2440 2007-01-22 Jan Hubicka <jh@suse.cz>
2442 * passes.c (init_optimization_passes): Do not rerun
2443 pass_early_warn_uninitialized.
2445 2007-01-22 Richard Guenther <rguenther@suse.de>
2447 PR tree-optimization/30038
2448 * tree-ssa-math-opts.c (maybe_record_sincos): New static helper
2450 (execute_cse_sincos_1): Likewise.
2451 (execute_cse_sincos): Likewise.
2452 (gate_cse_sincos): Likewise.
2453 (pass_cse_sincos): New pass CSEing sin() and cos() calls using
2454 the cexpi() canonicalization of sincos().
2455 * tree-pass.h (pass_cse_sincos): Declare.
2456 * passes.c (init_optimization_passes): New pass pas_cse_sincos.
2458 2007-01-21 Eric Botcazou <ebotcazou@libertysurf.fr>
2460 PR rtl-optimization/29329
2461 * combine.c (replaced_rhs_insn): Rename to i2mod.
2462 (replaced_rhs_value): Rename to i2mod_new_rhs.
2463 (i2mod_old_rhs): New global variable.
2464 (combine_instructions): Adjust for above change. Save a copy of
2465 the old RHS into i2mod_old_rhs when the contents of a REG_EQUAL
2466 note are substituted in the second instruction.
2467 (distribute_notes) <REG_DEAD>: Adjust for above change. Do not
2468 ditch the note if it pertains to the second eliminated register
2469 and this register is mentioned in i2mod_old_rhs.
2472 2006-09-12 Eric Botcazou <ebotcazou@libertysurf.fr>
2474 * combine.c (distribute_notes) <REG_DEAD>: Do not consider SETs past
2475 the insn to which the note was originally attached.
2477 2007-01-21 Jan Hubicka <jh@suse.cz>
2479 * ipa-inline.c (inlining_mode): Comment, move up.
2480 (cgraph_decide_inlining_incrementally): Do not perform inlining
2481 itself; fix handling of flattening of self recursive functions.
2482 (cgraph_find_cycles): Remove.
2483 (cgraph_flatten_node): Remove.
2484 (cgraph_decide_inlining): Use incremental inliner to handle flattening.
2485 (try_inline): New function.
2486 (cgraph_early_inlining): Update call of
2487 cgraph_decide_inlining_incrementally. Apply inlining here.
2488 (apply_inline): Update call of cgraph_decide_inlining_incrementally.
2490 2007-01-21 Dirk Mueller <dmueller@suse.de>
2493 * tree-vrp.c (check_array_bounds): do not warn
2494 about ADDR_EXPR's of ARRAY_REF's which are immediately
2495 used in binary expressions.
2497 2007-01-21 Ira Rosen <irar@il.ibm.com>
2499 * tree-vectorizer.h (struct _stmt_vec_info): Add new field
2500 read_write_dep and macros for its access.
2501 * tree-vectorizer.c (new_stmt_vec_info): Initialize the new field.
2502 * tree-vect-analyze.c (vect_analyze_data_ref_dependence): Remove
2503 argument, call vect_check_interleaving for every independent pair of
2504 data-refs. Mark loads that access the same memory location as a store
2506 (vect_check_dependences): Remove.
2507 (vect_analyze_data_ref_dependences): Remove vect_check_dependences
2508 call, fix the call to vect_analyze_data_ref_dependence.
2509 (vect_analyze_data_ref_access): For statements that access the same
2510 data-ref, check that they are not stores; for loads, check that there
2511 is no store that access the same location.
2513 2007-01-20 Roger Sayle <roger@eyesopen.com>
2514 Joseph Myers <joseph@codesourcery.com>
2516 * doc/invoke.texi (-fdump-rtl-combine): Fix under/overfull hbox.
2517 (-fdump-rtl-stack): Likewise.
2518 (-fno-signed-zeros): Use @minus{} for a minus sign. Correct typo.
2519 (-mcheck-zero-division, -mcpu): Fix under/overfull hbox.
2520 (-mpt-fixed): Use @minus{} for minus sign.
2521 (Using Precompiled Headers): Fix under/overfull hbox.
2523 2007-01-20 Jan Hubicka <jh@suse.cz>
2525 * tree-flow.h (struct stmt_ann_d): Move references_memory to proper
2526 place within annotation.
2528 2007-01-20 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
2530 * pa.c (output_move_double): Change array size of xoperands to 4.
2532 2007-01-20 Jan Hubicka <jh@suse.cz>
2534 * tree-tailcall.c (adjust_return_value): Do not use RESULT_DECL
2537 2007-01-19 Ian Lance Taylor <iant@google.com>
2539 * expmed.c (expand_divmod) [TRUNC_MOD_EXPR, TRUNC_DIV_EXPR]: Cast
2540 constant to unsigned HOST_WIDE_INT before negating.
2542 2007-01-19 Ian Lance Taylor <iant@google.com>
2544 * tree-ssa-operands.h (struct vuse_vec_d): Change num_vuse field
2546 (VUSE_VECT_ELEMENT) [ENABLE_CHECKING]: Use unsigned comparison.
2547 (VUSE_ELEMENT_PTR) [ENABLE_CHECKING]: Likewise.
2548 (SET_VUSE_VECT_ELEMENT) [ENABLE_CHECKING]: Likewise.
2549 (SET_VUSE_ELEMENT_VAR) [ENABLE_CHECKING]: Likewise.
2550 (SET_VUSE_ELEMENT_PTR) [ENABLE_CHECKING]: Likewise.
2551 (realloc_vdef, realloc_vuse): Change second parameter to
2553 (ssa_operand_iterator_d): Change vuse_index and mayuse_index
2555 * tree-ssa-operands.c (realloc_vop): Change num_elem parameter to
2556 unsigned. Change x and lim locals to unsigned.
2557 (realloc_vdef, realloc_vuse): Change num_elem parameter to
2559 (finalize_ssa_vuse_ops): Change old_i local to unsigned.
2560 (copy_virtual_operands): Change i and n locals to unsigned.
2562 2007-01-19 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
2565 * builtins.c (fold_builtin_1): Handle builtin fdim.
2567 2007-01-20 Jan Hubicka <jh@suse.cz>
2569 * tree-ssa.c (init_tree_ssa): Do not call init_alias_heapvars.
2570 * tree-ssa-structalias.c (compute_points_to_sets): Do call
2571 init_alias_heapvars.
2572 (init_alias_heapvars): Initialize only when not already initialized.
2573 (delete_alias_heapvars): Set heapvar_for_stmt to NULL.
2575 2007-01-19 Roger Sayle <roger@eyesopen.com>
2577 * common.opt (fsigned-zeros): New command line option.
2578 * flags.h (HONOR_SIGNED_ZEROS): Control via flag_signed_zeros instead
2579 of flag_unsafe_math_optimizations.
2580 * opts.c (set_fast_math_flags): The -ffast-math command line option
2581 implies -fno-signed-zeros.
2582 (fast_math_flags_set_p): Likewise.
2584 * doc/invoke.texi: Document new -fno-signed-zeros option, and update
2585 the documentation of -ffast-math appropriately. Wrap long lines.
2587 2007-01-19 Steve Ellcey <sje@cup.hp.com>
2589 * system.h (ASM_MAKE_LABEL_LINKONCE): Poison.
2590 * varasm.c (globalize_decl): Remove ASM_MAKE_LABEL_LINKONCE ifdef.
2592 2007-01-19 Tomas Bily <tbily@suse.cz>
2594 * cgraphunit.c (cgraph_finalize_function): Updating of pid
2596 (tree_init_ic_make_global_vars): New function
2597 (tree_init_edge_profiler): call of tree_init_ic_make_global_vars
2598 (tree_gen_ic_profiler): New function
2599 (tree_gen_ic_func_profiler): New function
2600 (tree_profiling): Added calling of tree_gen_ic_func_profiler
2601 (tree_profile_hooks): Added hook for indirec/virtual calls
2602 * value-prof.c (tree_find_values_to_profile): New case for
2604 (tree_values_to_profile): Call for determining indirect/virtual
2606 (tree_indirect_call_to_profile): New function
2607 (tree_ic_transform): New function
2608 (tree_ic): New function
2609 (find_func_by_pid): New function
2610 (init_pid_map): New function
2611 (tree_value_profile_transformations): Added check for
2612 indirect/virtual call transformation
2613 * value-prof.h (enum hist_type): New counter type for
2614 indirect/virtual calls
2615 (profile_hooks): Added new hook for profiling indirect/virtual
2617 * profile.c (instrument_values): New case for indirect/virtual
2619 * gcov-io.h (GCOV_LAST_VALUE_COUNTER): Changed to 6
2620 (GCOV_COUNTER_V_INDIR): New counter type
2621 (GCOV_COUNTER_NAMES): New name of counter "indirect" added
2622 (GCOV_MERGE_FUNCTIONS): New merge function for indirect/virtual
2624 * cgraph.c: Definition of cgraph_max_pid
2625 (cgraph_create_node): Default init of pid attribute
2626 * cgraph.h: Declaration of cgraph_max_pid
2627 (struct cgraph_node): Added pid attribute
2628 * libgcov.c (__gcov_indirect_call_profiler): New function
2629 (__gcov_one_value_profiler_body): New function
2630 (__gcov_one_value_profiler): Body was moved to
2631 __gcov_one_value_profiler_body and calls it
2633 2007-01-19 Basile Starynkevitch <basile@starynkevitch.net>
2635 * doc/gty.texi (Options): Document the mark_hook option to GTY.
2636 * gengtype.c (write_types_data, write_func_for_structure,
2637 write_types, ggc_wtd, pch_wtd): Add skip_hooks to
2638 write_types_data, ggc_wtd, pch_wtd for processing mark_hook.
2639 (walk_type, write_func_for_structure): Generate the mark_hook if
2642 2007-01-19 Jan Hubicka <jh@suse.cz>
2644 * ipa-inline.c (cgraph_decide_inlining_incrementally): Instead of
2645 'early' argument take inlining mode argument specifying whether to
2646 inline for size/speeed or all functions; add support for flattening;
2648 (cgraph_early_inlining): Update call of decide_inlining_incrementally.
2650 2007-01-19 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
2653 * toplev.c (warn_deprecated_use): Use %qD instead of %qs to print
2654 the name of the declared identifier.
2656 2007-01-19 Dirk Mueller <dmueller@suse.de>
2658 * config/i386.h (CONDITIONAL_REGISTER_USAGE): Store
2659 result of PIC_OFFSET_TABLE_REGNUM in temporary variable to avoid
2660 duplicate evaluation.
2662 2007-01-19 Uros Bizjak <ubizjak@gmail.com>
2664 * config/i386/i386.md (acos<mode>2): Rename from acossf2 and acosdf2.
2665 Macroize expander using X87MODEF12 mode macro. Extend operand 1
2666 to XFMode, use acosxf2 and truncate result to requested mode.
2667 Use SSE_FLOAT_MODE_P to disable patterns for SSE math.
2668 (asin<mode>2): Similarly, with asin expanders.
2669 (*fscalexf4_i387): Rename from *fscalexf4.
2670 (expNcorexf3): New expander.
2671 (expxf2, exp10xf2, exp2xf2): Use expNcorexf3 expander.
2672 (exp<mode>2): Rename from expsf2 and expdf2. Macroize expander using
2673 X87MODEF12 mode macro. Extend operand 1 to XFMode, use expxf2 and
2674 truncate result to requested mode. Use SSE_FLOAT_MODE_P to disable
2675 patterns for SSE math.
2676 (exp10<mode>2): Similarly, with exp10 expanders.
2677 (exp2<mode>2): Similarly, with exp2 expanders.
2678 (expm1<mode>2): Similarly, with expm1 expanders.
2679 (ldexp<mode>3): Similarly, with ldexp expanders.
2680 (log<mode>2, log10<mode>2, log2<mode>2, log1p<mode>2, logb<mode>2):
2681 Use gen_truncxf<mode>2_i387_noop to truncate result.
2683 2007-01-19 Richard Sandiford <richard@codesourcery.com>
2685 * config/m68k/m68k.h (PREFERRED_STACK_BOUNDARY): Define to 32
2686 for ColdFire targets.
2688 2007-01-19 Nathan Sidwell <nathan@codesourcery.com>
2689 Richard Sandiford <richard@codesourcery.com>
2691 * config/m68k/m68k.h (M68K_STATIC_CHAIN_REG_NAME): New macro.
2692 (INITIALIZE_TRAMPOLINE): Use STATIC_CHAIN_REGNUM.
2693 (__transfer_from_trampoline): Use M68K_STATIC_CHAIN_REG_NAME.
2694 * config/m68k/m68kelf.h (STATIC_CHAIN_REGNUM): Override.
2695 (M68K_STATIC_CHAIN_REG_NAME): Likewise.
2696 * config/m68k/netbsd-elf.h (M68K_STATIC_CHAIN_REG_NAME): Likewise.
2698 2007-01-19 Richard Sandiford <richard@codesourcery.com>
2700 * config/m68k/m68k.md (adddi_dilshr32): Rename to...
2701 (*adddi_dilshr32): ...this. Fix formatting. Remove commented-out
2702 non-canonical pattern. Restrict to !TARGET_COLDFIRE.
2703 (*adddi_dilshr32_cf): New pattern.
2704 (adddi3, subdi3): Remove first alternatives.
2706 2007-01-19 Richard Sandiford <richard@codesourcery.com>
2708 * config/m68k/m68k.c (notice_update_cc): If an SFmode move is
2709 implemented using move.l, do not use its cc result for floating-point
2712 2007-01-19 Richard Sandiford <richard@codesourcery.com>
2714 * config/m68k/m68k.h (EXTRA_CONSTRAINT): Stop the 'T' constraint
2715 from accepting 's' constraints if flag_pic.
2717 2007-01-19 Richard Sandiford <richard@codesourcery.com>
2719 * config/m68k/m68k.md (bordered, bunordered, buneq, bunge, bungt)
2720 (bunle, bunlt, bltgt, bordered_rev, bunordered_rev, buneq_rev)
2721 (bunge_rev, bungt_rev, bunle_rev, bunlt_rev, bltgt_rev): Change
2722 condition from TARGET_68881 to TARGET_HARD_FLOAT.
2724 2007-01-19 Sandra Loosemore <sandra@codesourcery.com>
2726 * longlong.h (count_leading_zeros, COUNT_LEADING_ZEROS_0): Add
2727 ColdFire alternatives.
2728 * config/m68k/m68k.h (CLZ_DEFINED_VALUE_AT_ZERO): New macro.
2729 * config/m68k/m68k.md (clzsi2): Define for ColdFire
2730 architectures that support ff1 instruction.
2732 2007-01-19 Richard Sandiford <richard@codesourcery.com>
2733 Julian Brown <julian@codesourcery.com>
2735 * config/m68k/m68k.h (CONST_OK_FOR_LETTER_P): Add an 'R' case.
2736 * config/m68k/m68k.md (*movsi_cfv4): Fold into...
2737 (*movsi_cf): ...here. Remove unnecessary 'R' from 'Rg'.
2740 2007-01-19 Richard Sandiford <richard@codesourcery.com>
2742 * config/m68k/m68k-protos.h (valid_mov3q_const): Take a HOST_WIDE_INT
2744 (output_move_const_into_data_reg, output_move_simode_const): Delete.
2745 * config/m68k/m68k.c (const_method, const_int_cost): Take a
2746 HOST_WIDE_INT instead of an rtx.
2747 (m68k_rtx_costs): Update call accordingly.
2748 (output_move_const_into_data_reg): Likewise. Fix formatting.
2749 (valid_mov3q_const): Take a HOST_WIDE_INT instead of an rtx.
2751 (output_move_simode_const): Update calls after above changes.
2752 Rework to use automatic variables and predicates like MEM_P.
2753 * config/m68k/m68k.md (pushexthisi_const): Update call to
2756 2007-01-19 Dirk Mueller <dmueller@suse.de>
2758 * tree-ssa-alias.c (perform_var_substitution): Fix typo
2761 2007-01-19 Richard Guenther <rguenther@suse.de>
2763 * builtins.c (expand_builtin_cexpi): Fall back to expanding
2764 via cexp in case sincos is not available.
2766 2007-01-19 Richard Guenther <rguenther@suse.de>
2768 * doc/tm.texi (TARGET_HAS_SINCOS): Document new target macro.
2769 * defaults.h (TARGET_HAS_SINCOS): Default to off.
2770 * config/linux.h (TARGET_HAS_SINCOS): Set to on if we have glibc.
2771 * config/alpha/linux.h (TARGET_HAS_SINCOS): Likewise.
2772 * config/sparc/linux.h (TARGET_HAS_SINCOS): Likewise.
2773 * config/sparc/linux64.h (TARGET_HAS_SINCOS): Likewise.
2774 * config/rs6000/linux.h (TARGET_HAS_SINCOS): Likewise.
2775 * config/rs6000/linux64.h (TARGET_HAS_SINCOS): Likewise.
2777 2007-01-19 Uros Bizjak <ubizjak@gmail.com>
2779 * config/i386/i386.md (*fpatanxf3_i387, fpatan_extend<mode>xf3_i387):
2781 (atan2sf3_1, atan2df3_1, atan2xf3_1): Remove insn patterns.
2782 (atan2xf3): Directly generate RTL pattern.
2783 (atan2<mode>3): Rename from atan2sf3 and atan2df3 and macroize insn
2784 patterns using X87MODEF12 mode macro. Use fpatan_extend<mode>xf3_i387
2785 and truncate result to requested mode. Use SSE_FLOAT_MODE_P to
2786 disable patterns for SSE math.
2787 (atan<mode>2): Rename from atansf2 and atandf2 and macroize insn
2788 patterns using X87MODEF12 mode macro. Use fpatan_extend<mode>xf3_i387
2789 and truncate result to requested mode. Use SSE_FLOAT_MODE_P to
2790 disable patterns for SSE math.
2792 2007-01-19 Alexandre Oliva <aoliva@redhat.com>
2794 * libgcc-std.ver: Fix typo in %inherit for GCC_4.3.0.
2796 2007-01-18 Roger Sayle <roger@eyesopen.com>
2798 * fold-const.c (fold_unary) <VIEW_CONVERT_EXPR>: Optimize away a
2799 VIEW_CONVERT_EXPR to the same type as it's operand.
2801 2007-01-18 David Edelsohn <edelsohn@gnu.org>
2803 * config/rs6000/darwin-ldouble.c: Only build _SOFT_FLOAT if
2804 configured for long double 128.
2806 2007-01-18 Mike Stump <mrs@apple.com>
2808 * config/rs6000/rs6000.c (rs6000_emit_vector_compare): Fix build
2811 2007-01-18 Michael Meissner <michael.meissner@amd.com>
2813 * i386.c (ix86_compute_frame_layout): Make fprintf's in #if 0 code
2816 2007-01-18 Jan Hubicka <jh@suse.cz>
2818 * tree-ssa-operands.c (vop_free_bucket_size): Never return value
2819 greater than NUM_VOP_FREE_BUCKETS.
2821 2007-01-18 Daniel Berlin <dberlin@dberlin.org>
2823 * tree-ssa-structalias.c: Update comments.
2824 (ptabitmap_obstack): Removed.
2826 (oldpta_obstack): Ditto.
2827 (stats): Add a few members.
2828 (struct variable_info): Remove node, complex, address_taken, and
2829 indirect_target members. Add oldsolution member.
2830 (new_var_info): Do not initialize removed members.
2831 (constraint_expr_type): Remove INCLUDES.
2832 (constraint_graph): Add size, implicit_preds, rep,
2833 indirect_cycles, eq_rep, label, direct_nodes, and complex members.
2834 (FIRST_REF_NODE): New macro.
2835 (LAST_REF_NODE): Ditto.
2836 (FIRST_ADDR_NODE): Ditto.
2837 (find): New function.
2839 (dump_constraint): Do not handle INCLUDES.
2840 (insert_into_complex): Do not insert duplicate constraints.
2841 (condense_varmap_nodes): Renamed and rewritten into ...
2842 (merge_node_constraints): This. Also fix bug in handling of
2843 offseted copy constraints.
2844 (clear_edges_for_node): No longer need to deal with preds at all,
2845 or removing associated preds/succs.
2846 (merge_graph_nodes): Deal with indirect_cycles.
2847 Don't deal with predecessors.
2848 (add_implicit_graph_edge): New function.
2849 (add_pred_graph_edge): Ditto.
2850 (add_graph_edge): Don't deal with predecessors.
2851 (build_constraint_graph): Removed.
2852 (build_pred_graph): New function.
2853 (build_succ_graph): Ditto.
2854 (struct scc_info): Removed in_component. Added roots, dfs, and
2855 node_mapping. Remove visited_index, unification_queue.
2856 (scc_visit): Deal with union-find we do now.
2857 Deal with cycles with REF nodes.
2858 (collapse_nodes): Renamed and rewritten to ...
2859 (unify_nodes): This.
2860 (process_unification_queue): Removed.
2861 (topo_visit): Cleanup
2862 (do_da_constraint): Use find.
2863 (do_sd_constraint): Ditto.
2864 (do_ds_constraint): Ditto.
2865 (do_complex_constraint): Ditto.
2866 (init_scc_info): Update for removed and added members.
2867 (find_and_collapse_graph_cycles): Renamed and rewritten into ...
2868 (find_indirect_cycles): This.
2869 (equivalence_class): New variable.
2870 (label_visit): New function.
2871 (perform_variable_substitution): Rewritten.
2872 (free_var_substitution_info): New function.
2873 (find_equivalent_node): Ditto.
2874 (move_complex_constraints): Ditto.
2875 (eliminate_indirect_cycles): Ditto.
2876 (solve_graph): Only propagate changed bits.
2877 Use indirect cycle elimination.
2879 (tree_id_t): Rename to tree_vi_t, delete id member, add vi member.
2880 (tree_id_eq): Renamed to ...
2881 (tree_vi_eq): This. Update for member change
2882 (insert_id_for_tree): Renamed and rewritten to ...
2883 (insert_vi_for_tree): This.
2884 (lookup_id_for_tree): Renamed and rewritten to ...
2885 (lookup_vi_for_tree): This.
2886 (get_id_for_tree): Renamed and rewritten to ...
2887 (get_vi_for_tree): Ditto.
2888 (get_constraint_exp_from_ssa_var): Update to use get_vi_for_tree.
2889 (process_constraint): Don't handle INCLUDES.
2890 Remove special ADDRESSOF case.
2891 (find_func_aliases): Rewrite to use vi functions instead of id
2893 (create_function_info_for): Ditto.
2894 (create_variable_info_for): Ditto.
2895 (intra_create_variable_infos): Ditto.
2896 (merge_smts_into): Ditto.
2897 (find_what_p_points_to): Ditto.
2898 (init_base_vars): Ditto.
2899 (init_alias_vars): Ditto.
2900 (remove_preds_and_fake_succs): New function.
2901 (dump_sa_points_to_info): Dump new stats.
2902 (dump_solution_for_var): Use find.
2903 (set_used_smts): Fix formatting.
2904 (compute_points_to_sets): Updated for new functions.
2905 (ipa_pta_execute): Ditto.
2907 2007-01-18 Kazu Hirata <kazu@codesourcery.com>
2908 Richard Sandiford <richard@codesourcery.com>
2910 * doc/tm.texi (TARGET_FUNCTION_VALUE): Expand documentation of
2912 * calls.c (expand_call): If the return value is a PARALLEL,
2913 extract its first member.
2914 * config/m68k/linux.h (FUNCTION_EXTRA_EPILOGUE): Remove.
2915 * config/m68k/m68k.c (m68k_output_function_epilogue): Don't
2916 use FUNCTION_EXTRA_EPILOGUE.
2917 (m68k_function_value): Return a PARALLEL if the return value
2918 is of a pointer type.
2919 * config/m68k/netbsd-elf.h (current_function_returns_pointer)
2920 (FUNCTION_EXTRA_EPILOGUE): Remove.
2921 * config/m68k/m68k.md (D0_REG): New constant.
2923 2007-01-18 Kazu Hirata <kazu@codesourcery.com>
2925 * config/m68k/m68k.c (m68k_output_function_epilogue): Don't
2926 output a NOP for empty epilogues.
2928 2007-01-18 Richard Sandiford <richard@codesourcery.com>
2930 * config/m68k/m68k.c (m68k_use_return_insn): Update comments
2931 before function. Extend register save check to include all
2932 registers, not just integer ones.
2934 2007-01-18 Kazu Hirata <kazu@codesourcery.com>
2936 * config/m68k/m68k-protos.h (use_return_insn): Rename to...
2937 (m68k_use_return_insn): ...this.
2938 * config/m68k/m68k.h (USE_RETURN_INSN): Delete.
2939 * config/m68k/m68k.c (use_return_insn): Rename to...
2940 (m68k_use_return_insn): ...this.
2941 * config/m68k/m68k.md (return): Use m68k_use_return_insn instead
2944 2007-01-18 Nathan Sidwell <nathan@codesourcery.com>
2946 * config/m68k/fpgnulib.c (__truncdfsf2): Implement round to
2947 nearest even, fix denormal rounding overflow.
2949 2007-01-18 Richard Sandiford <richard@codesourcery.com>
2951 * config/m68k/m68k.md (movsf_cf_hard): Use fsmove instead of
2952 f%$smove and f%$move.
2953 (movdf_cf_hard): Use fdmove for cases 0 and 3 and fmove for case 1.
2954 (extendsfdf2_cf): Use fdmove instead of f%&move.
2955 (truncdfsf2_cf): Use fsmove instead of f%$smove.
2956 (add<mode>3_cf, sub<mode>3_cf): Use <FP:prec> instead of <FP:round>.
2958 2007-01-18 Richard Sandiford <richard@codesourcery.com>
2960 * config/m68k/m68k.md (movdf_cf_hard): Use output_move_double for
2963 2007-01-18 Richard Sandiford <richard@codesourcery.com>
2965 * real.h (real_format): Add a canonical_nan_lsbs_set field.
2966 (coldfire_single_format): Declare.
2967 (coldfire_double_format): Likewise.
2968 * real.c (encode_ieee_single): Use canonical_nan_lsbs_set instead
2969 of qnan_msb_set to determine the lower bits of a canonical
2971 (encode_ieee_double): Likewise.
2972 (encode_ieee_quad): Likewise.
2973 (ieee_single_format): Initialize canonical_nan_lsbs_set.
2974 (mips_single_format): Likewise.
2975 (ieee_double_format): Likewise.
2976 (mips_double_format): Likewise.
2977 (ieee_extended_motorola_format): Likewise.
2978 (ieee_extended_intel_96_format): Likewise.
2979 (ieee_extended_intel_128_format): Likewise.
2980 (ieee_extended_intel_96_round_53_format): Likewise.
2981 (ibm_extended_format): Likewise.
2982 (mips_extended_format): Likewise.
2983 (ieee_quad_format): Likewise.
2984 (mips_quad_format): Likewise.
2985 (vax_f_format): Likewise.
2986 (vax_d_format): Likewise.
2987 (vax_g_format): Likewise.
2988 (i370_single_format): Likewise.
2989 (i370_double_format): Likewise.
2990 (decimal_single_format): Likewise.
2991 (decimal_double_format): Likewise.
2992 (decimal_quad_format): Likewise.
2993 (c4x_single_format): Likewise.
2994 (c4x_extended_format): Likewise.
2995 (real_internal_format): Likewise.
2996 (coldfire_single_format): New real_format.
2997 (coldfire_double_format): Likewise.
2998 * config/pdp11/pdp11.c (pdp11_f_format): Initialize
2999 canonical_nan_lsbs_set.
3000 (pdp11_d_format): Likewise.
3001 * config/m68k/m68k.c (override_options): Override REAL_FORMAT_MODE
3002 if TARGET_COLDFIRE_CPU.
3004 2007-01-18 Richard Sandiford <richard@codesourcery.com>
3006 * config/m68k/m68k-protos.h (m68k_output_pic_call): Delete.
3007 (output_call, m68k_legitimize_call_address): Declare.
3008 * config/m68k/m68k.h (EXTRA_CONSTRAINT): Remove unnecessary
3009 parenthesees. Add support for a 'W' constraint.
3010 (LEGITIMATE_PIC_OPERAND_P): Remove SYMBOL_REF_FLAG handling.
3011 (PRINT_OPERAND_PUNCT_VALID_P): Remove comment about 'o'.
3012 (m68k_symbolic_call, m68k_symbolic_jump): Declare.
3013 * config/m68k/m68k.c (m68k_symbolic_call, m68k_symbolic_jump): New
3015 (override_options): Initialize them. Do not set flag_no_function_cse
3016 for TARGET_ID_SHARED_LIBRARY.
3017 (m68k_output_pic_call): Delete.
3018 (m68k_legitimize_call_address): New function.
3019 (print_operand): Remove the %o prefix. Handle the %p prefix.
3020 (output_call): New function.
3021 (m68k_output_mi_thunk): Use m68k_symbolic_jump. Always load the
3022 target address from the GOT if symbolic jumps are not allowed.
3023 * config/m68k/m68k.md (call, general_operand): Do not set
3024 SYMBOL_REF_FLAG. Use m68k_legitimize_call_address instead.
3025 Merge separate flag_pic and !flag_pic define_insns into...
3026 (*call, *call_value): ...these new patterns. Match the address
3027 rather than the containing MEM and require it to be a call_operand.
3028 Use output_call to generate the asm template.
3029 * config/m68k/predicates.md (const_call_operand): New predicate.
3030 (call_operand): Likewise.
3032 2007-01-18 Nathan Sidwell <nathan@codesourcery.com>
3034 * config/m68k/m68k.h (REGISTER_MOVE_COST): Simplify definition.
3035 (STACK_GROWS_DOWNWARD): Define to 1.
3036 (FUNCTION_VALUE, LIBCALL_VALUE, FUNCTION_VALUE_REGNO_P): Equivocate
3037 comments, emphasizing that these values are only defaults.
3038 * config/m68k/linux.h (LINK_SPEC): Fix formatting in #undef.
3039 * config/m68k/m68k.c (const_method): Remove trailing whitespace.
3041 2007-01-18 Richard Sandiford <richard@codesourcery.com>
3043 * config/m68k/m68k.md (cmpsi): Remove outdated flag_pic handling.
3045 2007-01-18 Kazu Hirata <kazu@codesourcery.com>
3046 Richard Sandiford <richard@codesourcery.com>
3048 * config/m68k/m68k.h (DATA_REGNO_P, ADDRESS_REGNO_P, INT_REGNO_P)
3049 (FP_REGNO_P): New macros.
3050 (REGNO_OK_FOR_INDEX_P, REGNO_OK_FOR_BASE_P, REGNO_OK_FOR_DATA_P)
3051 (REGNO_OK_FOR_FP_P, REG_OK_FOR_INDEX_P, REG_OK_FOR_BASE_P): Use them.
3052 (EH_RETURN_STACKADJ_RTX): Use A0_REG.
3053 * config/m68k/m68k.c (m68k_regno_mode_ok): Use the new REGNO macros.
3055 2007-01-18 Nathan Sidwell <nathan@codesourcery.com>
3057 * config.gcc (m68k-*-aout*, m68k-*-coff*, m68020-*-elf*, m68k-*-elf*)
3058 (m68k-*-uclinuxoldabi, m68k-*-uclinux*, m68k-*-rtems*): Add t-floatlib
3060 * config/m68k/t-floatlib: New file, extracting common code from...
3061 * config/m68k/t-m68kbare, config/m68k/t-m68kelf,
3062 * config/m68k/t-uclinux: Here.
3063 * config/m68k/fpgnulib.c: Do not compile extendeed precision
3064 routines on ColdFire targets.
3066 2007-01-18 Nathan Sidwell <nathan@codesourcery.com>
3068 * config.gcc (m68k-*-aout*, m68k-*-coff*, m68020-*-elf*, m68k-*-elf*)
3069 (m68010-*-netbsdelf*, m68k*-*-netbsdelf*, m68k*-*-openbsd*)
3070 (m68k-*-uclinuxoldabi, m68k-*-uclinux*, m68k-*-linux*)
3071 (m68k-*-rtems*): Use tm_file rather than m68k/m68k.h and
3072 explicitly set MOTOROLA to 1.
3073 * config/m68k/m68k.h (MOTOROLA): Simplify definition accordingly.
3075 2007-01-18 Richard Sandiford <richard@codesourcery.com>
3076 Nathan Sidwell <nathan@codesourcery.com>
3078 * config/m68k/m68k.h (PCC_STATIC_STRUCT_RETURN): Delete.
3079 (ASM_OUTPUT_REG_PUSH, ASM_OUTPUT_REG_POP): Add MOTOROLA cases.
3080 * config/m68k/coff.h (REGISTER_PREFIX_MD): Delete.
3081 * config/m68k/m68020-elf.h (LIB_SPEC): Delete.
3082 * config/m68k/m68k-none.h (CC1_SPEC, CPP_SUBTARGET_SPEC): Delete.
3083 * config/m68k/m68kelf.h (IMMEDIATE_PREFIX, REGISTER_PREFIX_MD)
3084 (ASM_OUTPUT_REG_PUSH, ASM_OUTPUT_REG_POP): Delete.
3085 (NO_DOLLAR_IN_LABEL, PCC_STATIC_STRUCT_RETURN): Don't undefine.
3086 (BSS_ASM_OP, ASM_OUTPUT_SKIP, ASM_OUTPUT_ASCII): Delete.
3087 * config/m68k/m68kemb.h (PCC_STATIC_STRUCT_RETURN): Don't undefine.
3088 * config/m68k/linux.h (SIZE_TYPE, PTRDIFF_TYPE, WCHAR_TYPE)
3089 (WCHAR_TYPE_SIZE, TARGET_OBJFMT_CPP_BUILTINS): Delete.
3090 (TARGET_OS_CPP_BUILTINS): Don't define mc68000 and mc68020 here.
3091 (DBX_REGISTER_NUMBER): Delete.
3092 * config/m68k/netbsd-elf.h (IMMEDIATE_PREFIX): Delete.
3093 (PCC_STATIC_STRUCT_RETURN): Don't undefine.
3094 * config/m68k/openbsd.h (PCC_STATIC_STRUCT_RETURN): Define.
3096 2007-01-18 Richard Sandiford <richard@codesourcery.com>
3098 * config.gcc (m68k-*-uclinux*): Add flat.h to $tm_file.
3099 * config/flat.h: New file.
3100 * crtstuff.c (USE_PT_GNU_EH_FRAME): Don't define if
3102 * config/m68k/m68k.h (ASM_PREFERRED_EH_DATA_FORMAT): Do not use
3103 indirect references for -msep-data or -mid-shared-library.
3104 Do not use PC-relative code addresses either.
3105 * config/m68k/m68k.c (override_options): Restrict -fPIC error
3107 * config/m68k/uclinux.h (STARTFILE_SPEC): Define. Use Scrt1.o
3108 for shared libraries and crt1.o for executables. Use crti.o and
3110 (ENDFILE_SPEC): Use crtend.o and crtn.o.
3111 (LIB_SPEC): Suppress -Rlibc.gdb if -static-libc is given.
3112 Do not add -elf2flt or -shared-lib-id options here.
3113 (LINK_SPEC): Define. Pass -elf2flt if no -elf2flt option is given.
3114 Pass -shared-lib-id if -mid-shared-library, taking the library
3115 identifier from -mshared-library-id if given, otherwise
3117 (EH_FRAME_IN_DATA_SECTION): Do not undefine.
3118 (INIT_SECTION_ASM_OP, FINI_SECTION_ASM_OP): Likewise.
3119 (TARGET_OS_CPP_BUILTINS): Define __GXX_MERGED_TYPEINFO_NAMES=0
3120 and __GXX_TYPEINFO_EQUALITY_INLINE=0 if -mid-shared-library.
3121 (DRIVER_SELF_SPECS): Map unadorned PIC options to -msep-data.
3122 * config/m68k/t-uclinux (EXTRA_MULTILIB_PARTS): Add crtbegin.o
3124 * config/m68k/lb1sf68.asm (PICCALL): Use an lea and pc-relative
3125 jump sequence for ISA A and ISA A+.
3126 (PICJUMP): Likewise.
3128 2007-01-18 Richard Sandiford <richard@codesourcery.com>
3130 * config.gcc (m68k-*-uclinux*): Base the port on the common
3131 and m68k GNU/Linux files rather than on the generic ELF ones.
3132 * config/m68k/uclinux.h (TARGET_VERSION): Override.
3133 (TARGET_OS_CPP_BUILTINS): Use LINUX_TARGET_OS_CPP_BUILTINS.
3135 2007-01-18 Julian Brown <julian@codesourcery.com>
3136 Richard Sandiford <richard@codesourcery.com>
3138 * config/m68k/m68k.h (LONG_DOUBLE_TYPE_SIZE): Make 64-bit on ColdFire.
3139 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
3140 * config/m68k/netbsd-elf.h (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Undefine
3142 * config/m68k/uclinux-oldabi.h (LONG_DOUBLE_TYPE_SIZE): Redefine to
3144 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
3146 2007-01-18 Richard Sandiford <richard@codesourcery.com>
3148 * doc/install.texi: Document m68k-uclinuxoldabi.
3149 * config.gcc (m68k-*-uclinuxoldabi): New configuration.
3150 * config/m68k/uclinux-oldabi.h: New file, copied from
3151 config/m68k/uclinux.h.
3153 2007-01-18 Kaz Kojima <kkojima@rr.iij4u.or.jp>
3155 * config/m32r/m32r-protos.h (m32r_expand_epilogue): Declare it.
3156 * config/m32r/m32r.c (m32r_setup_incoming_varargs): Use gen_frame_mem.
3157 (m32r_compute_frame_size): Use unsigned for regno.
3158 (m32r_reload_lr): Use gen_frame_mem.
3160 (m32r_output_function_epilogue): Don't output the function epilogue
3162 (m32r_expand_epilogue): New.
3163 (direct_return): Return false if the function has the interrupt
3165 (m32r_hard_regno_rename_ok): Remove code for the textual epilogue.
3166 * config/m32r/m32r.md (epilogue): New expander.
3167 (return_lr, return_rte): New insns.
3168 (return): Make it expander.
3169 (return_normal): New expander.
3171 2007-01-18 Josh Conner <jconner@apple.com>
3174 * config/rs6000/rs6000.c (rs6000_emit_vector_compare): Add
3175 support for UNLE, UNLT, UNGE, and UNGT.
3177 2007-01-18 Jan Hubicka <jh@suse.cz>
3179 * tree-vrp.c (finalize_jump_threads): Do not call cleanup_cfg by hand.
3180 * tree-tailcall (add_virtual_phis): Likewise.
3181 (optimize_tail_call): Return TODOs.
3182 (execute_tail_calls): Return TODOs.
3183 * tree-ssa-ccp (execute_fold_all_builtins): Do cleanup_cfg via TODO.
3184 * tree-cfgcleanup.c (cleanup_tree_cfg_loop): Return if something
3186 * tree-ssa-forwprop.c (tree_ssa_forward_propagate_single_use_value):
3187 Cleanup cfg using TODO.
3188 * tree-flow.h (cleanup_tree_cfg_loop): Update prototype.
3189 * passes.c (execute_function_todo): When cleanup did something, remove
3191 * tree-cfg.c (pass_build_cfg): Add cleanup_cfg TODO.
3192 (make_edges): Don't cleanup_cfg.
3194 2007-01-18 Uros Bizjak <ubizjak@gmail.com>
3196 * reg-stack.c (subst_stack_regs_pat) [UNSPEC_SINCOS_COS,
3197 UNSPEC_XTRACT_FRACT]: Use generic code for instructions that
3198 operate on the top of stack.
3199 [UNSPEC_SINCOS_SIN, UNSPEC_XTRACT_EXP, UNSPEC_TAN]: Rewrite
3200 register handling of instructions that output to the second
3202 [UNSPEC_TAN_ONE, UNSPEC_TAN_TAN]: Remove.
3203 (move_for_stack_reg): Special-case check for dead destination
3204 stack slot for constant load of 1.0 inside UNSPEC_TAN.
3206 * config/i386/i386.md (UNSPEC_TAN): New constant.
3207 (UNSPEC_TAN_ONE, UNSPEC_TAN_TAN): Remove.
3208 (fptanxf4_i387, fptan_extend<mode>xf4_i387): New patterns
3209 to correctly model move of constant 1.0 to top stack slot.
3210 (*tandf3_1, *tansf3_1, *tanxf3_1): Remove insn patterns.
3211 (unnamed peephole2 pattern): Remove corresponding peephole2
3212 pattern that optimizes tan insn and loading of constant 1.0.
3213 (tanxf2): Use fptanxf4_i387.
3214 (tan<mode>2): Rename from tansf2 and tandf2 and macroize insn
3215 patterns using X87MODEF12 mode macro. Use fptan_extend<mode>xf4_i387
3216 and truncate result to requested mode. Use SSE_FLOAT_MODE_P to
3217 disable patterns for SSE math.
3218 (sincos<mode>3): Use truncxf<mode>2_i387_noop for truncation.
3219 (fyl2x_extend<mode>xf3_i387): Use X87MODEF12 for operand 1.
3221 2007-01-18 Dirk Mueller <dmueller@suse.de>
3222 Richard Guenther <rguenther@suse.de>
3225 * doc/invoke.texi (Warray-bounds): Document -Warray-bounds.
3226 * common.opt (Warray-bounds): Add new warning option.
3227 * c-opts.c (c_common_handle_option): Define -Warray-bounds
3229 * Makefile.in: make tree-vrp.o depend on toplev.h
3230 * tree-vrp.c (vrp_finalize): Call check_array_refs if -Warray-bounds
3232 (check_array_refs, check_array_bounds, check_array_ref): New.
3234 2007-01-18 Jan Hubicka <jh@suse.cz>
3236 * tree-ssa-ccp.c (ccp_finalize): Return if something changed.
3237 (execute_ssa_ccp): Return flags conditionally.
3238 * tree-ssa-propagate.c (substitue_and_fold): Return if something was
3240 * tree-ssa-propagate.h (substitute_and_fold): Update prototype.
3242 2007-01-18 Steven Bosscher <steven@gcc.gnu.org>
3244 * cfgcleanup.c (cleanup_cfg): Detect cfglayout mode and set
3245 the CLEANUP_CFGLAYOUT flag when in cfglayout mode.
3247 * Makefile.c (GTFILES): Add cfglayout.h.
3248 * gengtype.c (open_base_files): Likewise.
3249 * cfglayout.c (cfg_layout_function_footer,
3250 cfg_layout_function_header) Reindent to make gengtype happy.
3251 * cfglayout.h (cfg_layout_function_footer,
3252 cfg_layout_function_header): Add GTY(()) marker.
3254 * ifcvt.c (noce_try_sign_mask): Make sure INSN_B is non-null.
3256 2007-01-18 Ben Elliston <bje@au.ibm.com>
3258 * genautomata.c (write_automata): Include xstrerror output in the
3259 error message if writing the DFA description file fails.
3261 2007-01-17 H.J. Lu <hongjiu.lu@intel.com>
3263 * config/mips/mips-protos.h (mips_output_external): Make it
3265 * config/mips/iris.h (TARGET_ASM_EXTERNAL_LIBCALL): Removed.
3266 * config/mips/mips.c (irix_output_external_libcall): Likewise.
3267 (extern_list): Likewise.
3268 (extern_head): Likewise.
3269 (TARGET_ASM_FILE_END): Likewise.
3270 (mips_file_end): Likewise.
3271 (mips_output_external): Rewritten.
3273 2007-01-18 Ben Elliston <bje@au.ibm.com>
3275 * genpreds.c (write_insn_preds_c): Only write out the function
3276 body for regclass_for_constraint if we have register constraints.
3278 2007-01-17 Tom Tromey <tromey@redhat.com>
3280 * doc/sourcebuild.texi (libgcj Tests): Use sourceware.org.
3281 * doc/install.texi (Testing): Use sourceware.org.
3282 (Binaries): Likewise.
3283 (Specific): Likewise.
3284 * doc/contrib.texi (Contributors): Use sourceware.org.
3286 2007-01-17 Anatoly Sokolov <aesok@post.ru>
3288 * config/avr/avr.h (AVR_HAVE_LPMX): New macro.
3289 (AVR_ENHANCED): Rename to ...
3290 (AVR_HAVE_MUL): ... new.
3291 (avr_enhanced_p): Rename to ...
3292 (avr_have_mul_p): ... new.
3293 (TARGET_CPU_CPP_BUILTINS): Use 'avr_have_mul_p' instead of
3294 'avr_enhanced_p' for "__AVR_ENHANCED__". Define "__AVR_HAVE_MUL__".
3295 * config/avr/avr.c (avr_enhanced_p): Rename to ...
3296 (avr_have_mul_p): ... new.
3297 (base_arch_s): Rename 'enhanced' to 'have_mul'.
3298 (avr_override_options): Use 'avr_have_mul_p' and 'have_mul' instead of
3299 'avr_enhanced_p' and 'enhanced'.
3300 (ashlhi3_out, ashrhi3_out, lshrhi3_out, avr_rtx_costs): Use
3301 AVR_HAVE_MUL instead of AVR_ENHANCED.
3302 * avr.md (*tablejump_enh): Use AVR_HAVE_LPMX instead of AVR_ENHANCED.
3303 (mulqi3, *mulqi3_enh, *mulqi3_call, mulqihi3, umulqihi3, mulhi3,
3304 *mulhi3_enh, *mulhi3_call, mulsi3, *mulsi3_call): Use AVR_HAVE_MUL
3305 instead of AVR_ENHANCED.
3306 (*tablejump_enh): Use AVR_HAVE_LPMX instead of AVR_ENHANCED.
3307 * libgcc.S: Use __AVR_HAVE_MUL__ instead of __AVR_ENHANCED__.
3308 (__tablejump__): Use __AVR_HAVE_LPMX__ instead of __AVR_ENHANCED__.
3310 2007-01-17 Ian Lance Taylor <iant@google.com>
3312 * vec.h (VEC_reserve_exact): Define.
3313 (vec_gc_p_reserve_exact): Declare.
3314 (vec_gc_o_reserve_exact): Declare.
3315 (vec_heap_p_reserve_exact): Declare.
3316 (vec_heap_o_reserve_exact): Declare.
3317 (VEC_OP (T,A,reserve_exact)): New static inline function, three
3319 (VEC_OP (T,A,reserve)) [all versions]: Remove handling of
3321 (VEC_OP (T,A,alloc)) [all versions]: Call ...reserve_exact.
3322 (VEC_OP (T,A,copy)) [all versions]: Likewise.
3323 (VEC_OP (T,a,safe_grow)) [all versions]: Likewise.
3324 * vec.c (calculate_allocation): Add exact parameter. Change all
3326 (vec_gc_o_reserve_1): New static function, from vec_gc_o_reserve.
3327 (vec_gc_p_reserve, vec_gc_o_reserve): Call vec_gc_o_reserve_1.
3328 (vec_gc_p_reserve_exact, vec_gc_o_reserve_exact): New functions.
3329 (vec_heap_o_reserve_1): New static function, from vec_heap_o_reserve.
3330 (vec_heap_p_reserve, vec_heap_o_reserve): Call vec_heap_o_reserve_1.
3331 (vec_heap_p_reserve_exact): New function.
3332 (vec_heap_o_reserve_exact): New function.
3334 2007-01-17 Jan Hubicka <jh@suse.cz>
3336 * ipa-type-escape.c (look_for_casts): Revamp using handled_component_p.
3338 2007-01-17 Eric Christopher <echristo@apple.com>
3340 * config.gcc: Support core2 processor.
3342 2007-01-16 Jan Hubicka <jh@suse.cz>
3344 * tree-ssanames.c (release_dead_ssa_names): Instead of ggc_freeing
3345 the names, just unlink the chain so we don't crash on dangling pointers
3348 2007-01-16 Jan Hubicka <jh@suse.cz>
3350 * cgraph.h (cgraph_decide_inlining_incrementally): Kill.
3351 * tree-pass.h: Reorder to make IPA passes appear toegher.
3352 (pass_early_inline, pass_inline_parameters, pass_apply_inline): Declare.
3353 * cgraphunit.c (cgraph_finalize_function): Do not compute inling
3354 parameters, do not call early inliner.
3355 * ipa-inline.c: Update comments. Include tree-flow.h
3356 (cgraph_decide_inlining): Do not compute inlining parameters.
3357 (cgraph_decide_inlining_incrementally): Return TODOs; assume to
3358 be called with function context set up.
3359 (pass_ipa_inline): Remove unreachable functions before pass.
3360 (cgraph_early_inlining): Simplify assuming to be called from the
3362 (pass_early_inline): New pass.
3363 (cgraph_gate_ipa_early_inlining): New gate.
3364 (pass_ipa_early_inline): Turn into simple wrapper.
3365 (compute_inline_parameters): New function.
3366 (gate_inline_passes): New gate.
3367 (pass_inline_parameters): New pass.
3368 (apply_inline): Move here from tree-optimize.c
3369 (pass_apply_inline): New pass.
3370 * ipa.c (cgraph_remove_unreachable_nodes): Verify cgraph after
3372 * tree-inline.c (optimize_inline_calls): Return TODOs rather than
3374 (tree_function_versioning): Do not allocate dummy struct function.
3375 * tree-inline.h (optimize_inline_calls): Update prototype.
3376 * tree-optimize.c (execute_fixup_cfg): Export.
3377 (pass_fixup_cfg): Remove
3378 (tree_rest_of_compilation): Do not apply inlines.
3379 * tree-flow.h (execute_fixup_cfg): Declare.
3380 * Makefile.in (gt-passes.c): New.
3381 * passes.c: Include gt-passes.h
3382 (init_optimization_passes): New passes.
3383 (nnodes, order): New static vars.
3384 (do_per_function_toporder): New function.
3385 (execute_one_pass): Dump current pass here.
3386 (execute_ipa_pass_list): Don't dump current pass here.
3388 2007-01-16 Janis Johnson <janis187@us.ibm.com>
3390 * config/dfp-bit.c (dfp_compare_op): Return separate value for NaN.
3391 (DFP_NE, DFP_LE, DFP_GE): Return false for NaN.
3393 2007-01-16 David Edelsohn <edelsohn@gnu.org>
3395 * config/rs6000/darwin-ldouble.c: Build file for SOFT_FLOAT.
3396 (strong_alias): Define.
3397 (__gcc_qmul): Provide non-FMA for soft-float.
3403 (__gcc_qunord): New.
3413 * config/rs6000/rs6000.c (rs6000_init_libfuncs): Initialize
3414 soft-float functions.
3415 * config/rs6000/libgcc-ppc-glibc.ver: Version soft-float symbols.
3416 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Do not warn
3417 about long double soft float.
3419 2007-01-16 Dorit Nuzman <dorit@il.ibm.com>
3420 Tehila Meyzels <tehila@il.ibm.com>
3422 * tree-vectorizer.h (is_pattern_stmt_p): New.
3423 * tree-vect-analyze.c (vect_determine_vectorization_factor): Fix
3424 formatting (tabs instead of spaces). Cleanup and clarify setting
3425 of STMT_VINFO_VECTYPE. Call is_pattern_stmt_p.
3426 * tree-vect-transform.c (vect_get_vec_def_for_operand): Fix typo.
3427 (vectorizable_type_demotion): Check that types are integral.
3428 (vectorizable_type_promotion): Likewise.
3429 (vectorizable_store): Fix typo. Eliminate new-line at end of
3432 2007-01-16 Jan Hubicka <jh@suse.cz>
3434 * tree-ssanames.c (release_dead_ssa_names): Remove invalidated
3437 2007-01-15 Eric Christopher <echristo@apple.com>
3439 * ifcvt.c: Include vec.h, vecprim.h.
3440 (check_cond_move_block): New argument regs.
3441 Reorganize. Add registers used to regs.
3442 (cond_move_process_if_block): Use regs set above as
3445 2007-01-15 Eric Christopher <echristo@apple.com>
3447 * config/darwin.h: Update copyright.
3448 (TARGET_OPTION_TRANSLATE_TABLE): Add umbrella.
3449 (LINK_COMMAND_SPEC): Add -u.
3450 (LINK_SPEC): Fix umbrella for above.
3452 2007-01-15 Joseph S. Myers <joseph@codesourcery.com>
3454 * config/soft-fp/op-common.h, config/soft-fp/op-4.h: Update from
3457 2007-01-15 Tom Tromey <tromey@redhat.com>
3459 * doc/sourcebuild.texi (libgcj Tests): Don't mention jacks.
3460 * doc/install.texi (Testing): Don't mention jacks.
3461 (Configuration): Document --enable-java-maintainer-mode. Move
3462 --with-java-home to libgcj-specific section. Document
3464 (Prerequisites): Mention --enable-java-maintainer-mode, ecj1.
3466 2007-01-15 Jan Hubicka <jh@suse.cz>
3468 * tree-ssa-dce.c (DCE_TODOs): New.
3469 (propagate_necessity): Return if something changed.
3470 (eliminate_unnecessary_stmts): Likewise.
3471 (perform_tree_ssa_dce): Return TODO flags when needed.
3472 (pass_dce, pass_dce_loop, pass_cd_dce): Remove TODO flags.
3474 2007-01-15 Uros Bizjak <ubizjak@gmail.com>
3476 * config/i386/i386.md (fyl2xxf3_i387): Rename from fyl2x_xf3.
3477 (fyl2x_extend<mode>xf3_i387): New insn pattern.
3478 (log<mode>2): Rename from logsf2 and logdf2 and macroize insn
3479 patterns using X87MODEF12 mode macro. Extend operand 1
3480 to XFmode. Use SSE_FLOAT_MODE_P to disable patterns for SSE math.
3481 (log10<mode>2): Ditto.
3482 (log2<mode>2): Ditto.
3483 (log1p<mode>2): Ditto.
3484 (logb<mode>2): Ditto.
3485 (fyl2xp1xf3_i387): Rename from fyl2xp1_xf3.
3486 (fyl2xp1_extend<mode>xf3_i387): New insn pattern.
3487 (*fxtractxf3_i387): Rename from *fxtractxf3.
3488 (fxtract_extend<mode>xf3_i387): New insn pattern.
3489 (ilogbsi2): Use match_dup 3, not match_operand:XF 3.
3490 * config/i386/i386.c (ix86_emit_i387_log1p): Use gen_fyl2xp1xf3_i387()
3491 and gen_fyl2xxf3_i387().
3493 2007-01-14 Zdenek Dvorak <dvorakz@suse.cz>
3495 * loop-unswitch.c (unswitch_loop): Do not call fix_loop_placement.
3496 * cfgloopmanip.c (fix_loop_placement): Made static. Use
3497 get_loop_exit_edges. Changed return type to bool.
3498 * cfgloop.h (fix_loop_placement): Declaration removed.
3500 2007-01-14 Dorit Nuzman <dorit@il.ibm.com>
3502 * param.h (MIN_VECT_LOOP_BOUND): New.
3503 * params.def (MIN_VECT_LOOP_BOUND): New.
3504 * tree-vectorizer.c (slpeel_tree_peel_loop_to_edge): Takes another
3505 argument - minimum threshold for number of iterations.
3506 * tree-vectorizer.h (slpeel_tree_peel_loop_to_edge): Add another
3507 argument to declaration.
3508 * tree-vect-analyze.c (vect_analyze_operations): Check value of
3509 MIN_VECT_LOOP_BOUND.
3510 * tree-vect-transform.c (vect_do_peeling_for_loop_bound): Call
3511 slpeel_tree_peel_loop_to_edge with additional argument.
3512 (vect_do_peeling_for_alignment): Likewise.
3513 * doc/invoke.texi (min-vect-loop-bound): Document new param option.
3515 2007-01-14 Uros Bizjak <ubizjak@gmail.com>
3518 * config/i386/i386.c (print_operand) ['z']: Output 'b' for
3521 2007-01-14 Jan Hubicka <jh@suse.cz>
3523 * tree-dfa.c (remove_referenced_var): New function.
3524 * tree-ssa-live.c (remove_unused_locals): Walk referenced vars and
3525 prune referenced vars list too.
3526 * tree-flow.h (remove_referenced_var): Declare.
3528 2007-01-14 Jan Hubicka <jh@suse.cz>
3530 * tree-eh.c (add_stmt_to_eh_region_fn): Do not add call_exprs
3532 (remove_stmt_from_eh_region_fn): Do not remove call_exprs.
3533 (verify_eh_throw_stmt_node, verify_eh_throw_table_statements): Kill.
3534 * except.h (verify_eh_throw_table_statements): Kill prototype.
3535 * cfgexpand.c (expand_gimple_basic_block): Propagate Eh regions
3536 into call exrepssions.
3537 * tree-optimize.c (execute_free_cfg_annotatiosn): Do not call
3539 * tree-cfg.c: Include pointer-set.h.
3540 (verify_node_sharing): Work on pointer set.
3541 (verify_eh_throw_stmt_node): New.
3542 (verify_stmts): Use pointers sets, verify throw_stmt.
3544 2007-01-13 Zdenek Dvorak <dvorakz@suse.cz>
3546 * ipa-reference.c (analyze_function): Consider also addresses taken
3549 2007-01-12 Roger Sayle <roger@eyesopen.com>
3551 * c-typeck.c (null_pointer_constant_p): Replace use of
3552 TREE_CONSTANT_OVERFLOW with TREE_OVERFLOW.
3553 (build_c_cast): Likewise.
3555 2007-01-12 Roger Sayle <roger@eyesopen.com>
3557 * tree.h (force_fit_type_double): Remove unused final argument.
3558 * c-common.c (constant_expression_warning): Replace use of
3559 TREE_CONSTANT_OVERFLOW with TREE_OVERFLOW.
3560 (convert_and_check): Likewise.
3561 (shorten_compare): Update call to force_fit_type_double.
3562 (c_common_truthvalue_conversion) <INTEGER_CST>: Use integer_zerop.
3563 * convert.c (convert_to_pointer): Update call to
3564 force_fit_type_double.
3565 * fold-const.c (force_fit_type_double): Remove overflowed_const
3567 (int_const_binop, fold_convert_const_int_from_int,
3568 fold_convert_const_int_from_real, fold_div_compare,
3569 fold_sign_changed_comparison, fold_unary, fold_negate_const,
3570 fold_abs_const, fold_not_const): Remove the final argument from
3571 calls to force_fit_type_double.
3573 2007-01-12 Andrew Pinski <andrew_pinski@playstation.sony.com>
3575 * configure.ac: Set insn to "nop" for spu-*-* also.
3576 * configure: Regenerate.
3578 2007-01-12 Olga Golovanevsky <olga@il.ibm.com>
3580 * builtins.def : Add BUILT_IN_FREE.
3582 2007-01-12 Jan Hubicka <jh@suse.cz>
3584 PR tree-optimization/30443
3585 * tree-inline.c (tree_function_versioning): Do not optimize when
3586 cloning for inlining.
3588 2007-01-12 Zdenek Dvorak <dvorakz@suse.cz>
3590 * doc/loop.texi: Document recording of loop exits.
3591 * cfgloopmanip.c (loopify, duplicate_loop): Use alloc_loop.
3592 (update_single_exits_after_duplication,
3593 update_single_exit_for_duplicated_loop,
3594 update_single_exit_for_duplicated_loops): Removed.
3595 (duplicate_loop_to_header_edge): Do not call
3596 update_single_exits_after_duplication and
3597 update_single_exit_for_duplicated_loops.
3598 (loop_version): Do not update single_exit information.
3599 (fix_loop_structure): Use record_loop_exits instead of
3600 mark_single_exit_loops.
3601 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update
3602 the lists of loop exits.
3603 * cfghooks.c (redirect_edge_and_branch, redirect_edge_and_branch_force,
3604 split_edge, merge_blocks): Update the lists of loop exits.
3605 * modulo-sched.c (sms_schedule): Pass LOOPS_HAVE_RECORDED_EXITS to
3606 loop_optimizer_init.
3607 * loop-init.c (loop_optimizer_init): Call record_loop_exits instead
3608 of mark_single_exit_loops.
3609 (loop_optimizer_finalize): Call release_recorded_exits.
3610 * tree-ssa-loop.c (tree_loop_optimizer_init): Pass
3611 LOOPS_HAVE_RECORDED_EXITS to loop_optimizer_init.
3612 * tree-vectorizer.c (slpeel_tree_duplicate_loop_to_edge_cfg): Do not
3613 update single exit information.
3614 * lambda-code.c (perfect_nestify): Ditto.
3615 * cfgloop.c (flow_loop_free): Destroy the list of exits of the loop.
3616 (mark_single_exit_loops): Removed.
3617 (alloc_loop, loop_exit_hash, loop_exit_eq, loop_exit_free,
3618 get_exit_descriptions, rescan_loop_exit, record_loop_exits,
3619 dump_recorded_exit, dump_recorded_exits, release_recorded_exits): New
3621 (get_loop_exit_edges, single_exit): Use recorded exit lists.
3622 (add_bb_to_loop, remove_bb_from_loops): Update the lists of loop exits.
3623 (verify_loop_structure): Verify consistency of the exit lists.
3624 (flow_loops_find): Use alloc_loop. Initialize exits hash.
3625 (set_single_exit): Removed.
3626 * cfgloop.h (struct loop_exit): New function.
3627 (struct loop): single_exit_ field replaced by exits field.
3628 (LOOPS_HAVE_MARKED_SINGLE_EXITS): Replaced by LOOPS_HAVE_RECORDED_EXITS.
3629 (struct loops): Added exits hash.
3630 (mark_single_exit_loops, set_single_exit): Declaration removed.
3631 (release_recorded_exits, record_loop_exits, rescan_loop_exit): Declare.
3633 2007-01-12 Richard Sandiford <richard@codesourcery.com>
3635 * doc/invoke.texi: Avoid use of @headitem.
3637 2007-01-12 Richard Sandiford <richard@codesourcery.com>
3639 * cse.c (cse_insn): Move HAVE_CC0 code after declarations.
3641 2007-01-12 Richard Sandiford <richard@codesourcery.com>
3643 * doc/install.texi: Fix m68k-*-* anchor and add m68k-*-* to the
3646 2007-01-12 Nathan Sidwell <nathan@codesourcery.com>
3647 Richard Sandiford <richard@codesourcery.com>
3649 * doc/invoke.texi: Document -mno-bitfield, -mno-rtd and -mno-short.
3650 * config/m68k/m68k.opt: Resort options.
3651 (mbitfield, mrtd, mshort): Remove RejectNegative properties.
3653 2007-01-12 Nathan Sidwell <nathan@codesourcery.com>
3654 Richard Sandiford <richard@codesourcery.com>
3656 * doc/invoke.texi: Document the macros that are defined by
3657 m68k's -mtune and -mhard-float options.
3658 * config/m68k/m68k-protos.h (m68k_cpp_cpu_ident) Declare.
3659 (m68k_cpp_cpu_family): Likewise.
3660 * config/m68k/m68k.h (TARGET_CPU_CPP_BUILTINS): Add a full set
3661 of __ucfv*__ macros. Define __mcffpu__ if generating code for
3662 ColdFire FPUs. Define __mcf_cpu_* and __mcf_family_* macros.
3663 * config/m68k/m68k.c (m68k_cpp_cpu_ident): New function.
3664 (m68k_cpp_cpu_family): Likewise.
3666 2007-01-12 Richard Sandiford <richard@codesourcery.com>
3668 * config/m68k/m68k.h (TARGET_CPU_CPP_BUILTINS): Treat all mc68*
3669 macros besides mc68000 as tuning macros. Use a switch statement
3670 to set them and mcpu32.
3672 2007-01-12 Julian Brown <julian@codesourcery.com>
3674 * config/m68k/m68k.h: Use TARGET_68040 instead of TARGET_68040_ONLY.
3675 (TARGET_68040_ONLY): Rename to...
3676 (TARGET_68040): ...this.
3677 * config/m68k/m68k.c: Use TARGET_68040 instead of TARGET_68040_ONLY.
3678 * config/m68k/m68k.md: Likewise.
3680 2007-01-12 Julian Brown <julian@codesourcery.com>
3681 Nathan Sidwell <nathan@codesourcery.com>
3682 Richard Sandiford <richard@codesourcery.com>
3684 * config.gcc (m680[012]0-*-*, m68k*-*-*): Set m68k_cpu_ident to
3685 the -mcpu= argument associated with the --with-cpu setting.
3686 Define M68K_DEFAULT_TUNE to the default -mtune= option,
3687 if different from the one implied by the -mcpu setting.
3688 Accept --with-cpu=FOO if FOO is listed in m68k-devices.def,
3689 using mcpu=FOO as the default CPU option. Set target_cpu_default2.
3690 * doc/invoke.texi: Mention ColdFire in the introduction to the
3691 m68k options. Document the new -march, -mcpu, -mtune, -mdiv,
3692 -mno-div and -mhard-float options. Make -m68881 a synonym for
3693 -mhard-float. Document the previously-undocumented -m5206e,
3694 -m528x, -m5307 and -m5407 options. Tweak the existing option
3695 documentation for consistency.
3696 * doc/install.texi: Mention new --with-cpu arguments.
3697 * config/m68k/m68k.h (OPTION_DEFAULT_SPECS): Only use the
3698 default CPU if neither -mcpu nor -march are specified.
3699 (ASM_CPU_SPEC): Pass down -mcpu and -march options.
3700 (TARGET_CPU_CPP_BUILTINS): Set __mcfisa*__ macros from
3701 TARGET_ISA*. Set the legacy __mcf*__ cpu macros in the same way,
3702 using m68k_tune to decide between families that implement the
3703 same ISA. Use m68k_tune to set __mcfv4e__.
3704 (FL_BITFIELD, FL_68881, FL_COLDFIRE, FL_CF_HWDIV, FL_CF_MAC)
3705 (FL_CF_EMAC, FL_CF_EMAC_B, FL_CF_USP, FL_CF_FPU, FL_ISA_68000)
3706 (FL_ISA_68010, FL_ISA_68020, FL_ISA_68040, FL_ISA_A, FL_ISA_B)
3707 (FL_ISA_C, FL_ISA_MMU): New macros.
3708 (MASK_COLDFIRE): Delete.
3709 (TARGET_68010, TARGET_68020, TARGET_68040_ONLY, TARGET_COLDFIRE)
3710 (TARGET_ISAB): Redefine in terms of m68k_cpu_flags.
3711 (TARGET_68881, TARGET_COLDFIRE_FPU): Redefine in terms of m68k_fpu.
3712 (TARGET_HARD_FLOAT): Do not define here.
3713 (TARGET_ISAAPLUS, TARGET_ISAC): New macros.
3714 (TUNE_68000): New macro.
3715 (TUNE_68000_10): Redefine in terms of TUNE_68000 and TUNE_68010.
3716 (TUNE_68010, TUNE_68030, TUNE_68040, TUNE_68060, TUNE_CPU32)
3717 (TUNE_CFV2): Redefine in terms of m68k_tune.
3718 (uarch_type, target_device, fpu_type): New enums.
3719 (m68k_cpu, m68k_tune, m68k_fpu, m68k_cpu_flags): Declare.
3720 * config/m68k/m68k.c (TARGET_DEFAULT): Remove MASK_68881.
3721 (FL_FOR_isa_00, FL_FOR_isa_10, FL_FOR_isa_20, FL_FOR_isa_40)
3722 (FL_FOR_isa_cpu32, FL_FOR_isa_a, FL_FOR_isa_aplus, FL_FOR_isa_b)
3723 (FL_FOR_isa_c): New macros.
3724 (m68k_isa): New enum.
3725 (m68k_target_selection): New structure.
3726 (all_devices, all_isas, all_microarchs): New tables.
3727 (m68k_cpu_entry, m68k_arch_entry, m68k_tune_entry, m68k_cpu)
3728 (m68k_tune, m68k_fpu, m68k_cpu_flags): New variables.
3729 (MASK_ALL_CPU_BITS): Delete.
3730 (m68k_find_selection): New function.
3731 (m68k_handle_option): Handle -mcpu=, -march= and -mtune=.
3732 Map the legacy target options to a combination of the new ones.
3733 (override_options): Set m68k_cpu, m68k_tune, m68k_fpu and
3734 m68k_cpu_flags. Handle M68K_DEFAULT_TUNE. Use m68k_cpu_flags
3735 to derive default MASK_BITFIELD, MASK_CF_HWDIV and MASK_HARD_FLOAT
3737 * config/m68k/m68k.opt (m5200, m5206e, m528x, m5307, m5407, mcfv4e)
3738 (m68010, m68020, m68020-40, m68020-60, m68030, m68040): Remove Mask
3740 (m68881, msoft-float): Change mask from 68881 to HARD_FLOAT.
3741 (march=, mcpu=, mdiv, mhard-float, mtune=): New options.
3742 * config/m68k/m68k-devices.def: New file.
3744 2007-01-12 Richard Sandiford <richard@codesourcery.com>
3745 Nathan Sidwell <nathan@codesourcery.com>
3747 * config/m68k/m68k.h (ASM_CPU_SPEC, ASM_SPEC, EXTRA_SPECS)
3748 (SUBTARGET_EXTRA_SPECS): New macros.
3749 * config/m68k/linux.h (ASM_SPEC): Remove CPU flags;
3750 use %(asm_cpu_spec) instead.
3751 * config/m68k/m68k-none.h (ASM_SPEC): Likewise.
3752 * config/m68k/openbsd.h (ASM_SPEC): Likewise.
3753 * config/m68k/netbsd-elf.h (ASM_SPEC): Likewise.
3754 (EXTRA_SPECS): Rename to...
3755 (SUBTARGET_EXTRA_SPECS): ...this.
3757 2007-01-12 Nathan Sidwell <nathan@codesourcery.com>
3758 Richard Sandiford <richard@codesourcery.com>
3759 Julian Brown <julian@codesourcery.com>
3761 * config.gcc (m68k-*-aout*, m68k-*-coff*, m68020-*-elf*, m68k-*-elf*)
3762 (m68k-*-uclinux*, m68k-*-linux*, m68k-*-rtems*): Set default_m68k_cpu
3763 to the configuration's default CPU.
3764 (m68010-*-netbsdelf*, m68k*-*-netbsdelf*, m68k*-*-openbsd*): Likewise.
3765 Remove default masks.
3766 (m680[012]0-*-*): Set the default with_cpu to the first part of
3768 (m68k*-*-*): Set the default with_cpu to m$default_m68k_cpu.
3769 (m68k*-*-linux): Extend the --with-cpu handling to...
3770 (m680[012]0-*-*, m68k*-*-*): ...these configurations. Allow m68000
3771 and m68010. Don't set target_cpu_default2.
3772 * doc/install.texi: Document --with-cpu for m68k.
3773 * config/m68k/m68k.h (OPTION_DEFAULT_SPECS): Define.
3774 * config/m68k/m68k-none.h (TARGET_CPU_DEFAULT, M68K_CPU_m68k)
3775 (M68K_CPU_m68000, M68K_CPU_m68010, M68K_CPU_m68020, M68K_CPU_m68030)
3776 (M68K_CPU_m68040, M68K_CPU_m68302, M68K_CPU_m68332, TARGET_DEFAULT)
3777 (ASM_CPU_DEFAULT_SPEC, CC1_CPU_DEFAULT_SPEC): Delete.
3778 (ASM_SPEC): Remove use of %(asm_cpu_default).
3779 (EXTRA_SPECS, SUBTARGET_EXTRA_SPECS, MULTILIB_DEFAULTS): Delete.
3780 * config/m68k/linux.h (TARGET_DEFAULT): Delete.
3781 (CPP_SPEC): Merge definitions. Do not handle __HAVE_68881__ here.
3782 * config/m68k/netbsd-elf.h (TARGET_OS_CPP_BUILTINS): Define
3783 __HAVE_FPU__ if TARGET_HARD_FLOAT.
3784 (TARGET_DEFAULT): Delete.
3785 (EXTRA_SPECS): Delete cpp_cpu_default_spec, cpp_cpu_spec,
3786 cpp_fpu_spec, asm_default_spec and netbsd_cpp_spec.
3787 (CPP_CPU_SPEC): Delete.
3788 (TARGET_VERSION): Merge definitions, using TARGET_68010 to pick
3789 the appropriate string.
3790 (CPP_CPU_DEFAULT_SPEC, ASM_DEFAULT_SPEC, CPP_FPU_SPEC): Delete.
3791 (CPP_SPEC): Define to NETBSD_CPP_SPEC.
3792 (ASM_SPEC): Don't use %(asm_default_spec).
3793 * config/m68k/m68k.c (TARGET_DEFAULT_TARGET_FLAGS): Remove
3794 TARGET_DEFAULT and add MASK_68881.
3795 * config/m68k/m68k.md: Remove mention of TARGET_DEFAULT from comments.
3797 2007-01-12 Richard Sandiford <richard@codesourcery.com>
3799 * config.gcc (m68010-*-netbsdelf*): Add MASK_68010.
3800 (m68k*-*-netbsdelf*, m68k*-*-openbsd*, m68k*-linux*): Add
3801 MASK_68010 alongside MASK_68020.
3802 * doc/invoke.texi: Document -m68010.
3803 * config/m68k/m68k.opt (m68010): New.
3804 * config/m68k/m68k.h (TARGET_CPU_CPP_BUILTINS): Define mc68010
3806 (TUNE_68010): New macro.
3807 * config/m68k/m68k-none.h (M68K_CPU_m68k, M68K_CPU_m68010)
3808 (M68K_CPU_m68020, M68K_CPU_m68030, M68K_CPU_m68040)
3809 (M68K_CPU_m68332): Add MASK_68010.
3810 * config/m68k/linux.h (TARGET_DEFAULT): Add MASK_68010 to
3811 fallback definition.
3812 * config/m68k/netbsd-elf.h (CPP_CPU_SPEC): Remove now-redundant
3814 * config/m68k/m68k.c (MASK_ALL_CPU_BITS): Add MASK_68010.
3815 (m68k_handle_option): Handle OPT_m68010. Add MASK_68010
3816 to all entries that use MASK_68020.
3817 (output_move_simode_const, output_move_himode, output_move_qimode)
3818 (output_move_stricthi, output_move_strictqi): Use TARGET_68010
3819 instead of TARGET_68020 to select clr behavior. Remove comment
3820 about there being no TARGET_68010.
3821 * config/m68k/m68k.md: Likewise throughout.
3823 2007-01-12 Julian Brown <julian@codesourcery.com>
3825 * config/m68k/m68k.h (TARGET_ISAB): New macro.
3826 * config/m68k/m68k.c: Use TARGET_ISAB rather than TARGET_CFV4.
3827 * config/m68k/m68k.md: Likewise.
3829 2007-01-12 Julian Brown <julian@codesourcery.com>
3831 * config/m68k/m68k.h (LEGITIMATE_INDEX_P, LEGITIMIZE_ADDRESS): Use
3832 TARGET_COLDFIRE_FPU instead of TARGET_CFV4E.
3834 2007-01-12 Julian Brown <julian@codesourcery.com>
3836 * config/m68k/m68k.h (TUNE_68040_60): New macro.
3837 * config/m68k/m68k.c (standard_68881_constant_p): Use it.
3838 * config/m68k/m68k.md: Likewise.
3840 2007-01-12 Julian Brown <julian@codesourcery.com>
3841 Richard Sandiford <richard@codesourcery.com>
3843 * config/m68k/m68k.h (TARGET_CPU_CPP_BUILTINS): Use TUNE_68030
3844 instead of TARGET_68030, TUNE_68040 instead of TARGET_68040,
3845 TUNE_68060 instead of TARGET_68060 and TUNE_CPU32 instead of
3847 (TARGET_CPU32): Rename to...
3848 (TUNE_CPU32): ...this.
3849 (TUNE_68000_10, TUNE_68030, TUNE_68040, TUNE_68060)
3850 (TUNE_CFV2): New macros.
3851 * config/m68k/netbsd-elf.h (LONG_DOUBLE_TYPE_SIZE): Simplify;
3852 remove conditions that are implied by TARGET_68020.
3853 * config/m68k/m68k.c (m68k_output_function_prologue): Use TUNE_68040
3854 instead of TARGET_68040 and TUNE_CPU32 instead of TARGET_CPU32.
3855 (m68k_output_function_epilogue): Likewise.
3856 (m68k_rtx_costs): Likewise. Use TUNE_68060 instead of TARGET_68060
3857 and TUNE_CFV2 instead of TARGET_5200. Use TUNE_68000_10 instead of
3858 "!TARGET_68020 && !TARGET_COLDFIRE" to choose between 68000 and
3859 non-68000 timings. Refactor multiplication and division costs.
3860 (output_addsi3): Use TUNE_68040 instead of TARGET_68040 and
3861 TUNE_CPU32 instead of TARGET_CPU32.
3862 (standard_68881_constant_p): Use TUNE_68040 instead of TARGET_68040
3863 and TUNE_68060 instead of TARGET_68060.
3864 * config/m68k/m68k.md: Use TUNE_68040 instead of TARGET_68040,
3865 TUNE_68060 instead of TARGET_68060, and TUNE_CPU32 instead of
3867 (movsi_const0): Use TUNE_68000_10 rather than "!TARGET_68020
3868 && !TARGET_COLDFIRE" to choose between moveq and clr.
3869 Likewise in the unnamed movsf pattern.
3870 (ashlsi_17_24, lshrsi_17_24): Guard with TUNE_68000_10 rather than
3871 "!TARGET_68020 && !TARGET_COLDFIRE". Likewise the unnamed
3874 2007-01-12 Richard Sandiford <richard@codesourcery.com>
3876 * config/m68k/m68k.h (TARGET_CPU_CPP_BUILTINS): Increase amount
3877 of tabbing before backslashes.
3879 2007-01-11 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
3881 * pa-linux.h (ASM_OUTPUT_INTERNAL_LABEL): Undefine.
3882 * pa.h (ASM_OUTPUT_LABEL): Output colon when using GAS.
3883 (ASM_OUTPUT_INTERNAL_LABEL): Define.
3885 2007-01-11 Zdenek Dvorak <dvorakz@suse.cz>
3887 * tree-ssa-loop-ivopts.c (extract_cond_operands): Split from
3888 find_interesting_uses_cond.
3889 (find_interesting_uses_cond): Use extract_cond_operands.
3890 (rewrite_use_compare): Use extract_cond_operands and
3891 force_gimple_operand_bsi. Do not call update_stmt.
3892 (determine_use_iv_cost_condition): Use extract_cond_operands.
3893 Return cheaper of using original bound and changing the exit bound.
3895 2007-01-11 Zdenek Dvorak <dvorakz@suse.cz>
3897 PR tree-optimization/29516
3898 * tree-ssa-address.c (tree_mem_ref_addr, add_to_parts,
3899 most_expensive_mult_to_index, addr_to_parts,
3900 create_mem_ref, maybe_fold_tmr): Make the type of
3901 fields of TARGET_MEM_REF sizetype.
3902 (move_fixed_address_to_symbol, move_pointer_to_base):
3904 * tree.def (TARGET_MEM_REF): Add comment on types of
3907 2007-01-11 Joseph Myers <joseph@codesourcery.com>
3909 * c-common.c (vector_types_convertible_p): Treat opaque types as
3910 always convertible if they have the same size, but not otherwise.
3912 2007-01-11 Steven Bosscher <steven@gcc.gnu.org>
3914 * ifcvt.c (struct noce_if_info): Add comments to the fields.
3915 Remove the b_unconditional field.
3916 (noce_try_sign_mask): Do not look at b_unconditional.
3917 (noce_process_if_block): Do not use merge_if_blocks. Update
3918 the CFG here. Do not set b_unconditional.
3919 (cond_move_process_if_block): Likewise.
3920 (find_cond_trap): Likewise.
3921 (check_cond_move_block): Require simple jump insns at the end
3924 2007-01-11 Jan Hubicka <jh@suse.cz>
3926 PR tree-optimization/1046
3927 * tree-tailcall.c (suitable_for_tail_call_opt_p): Use TREE_ADDRESSABLE
3928 when alias info is not ready.
3929 (pass_tail_recursion): Do not require aliasing.
3930 * tree-ssa-copyrename.c (pass_rename_ssa_cop): Likewise.
3931 * tree-ssa-ccp.c (pass_ccp, pass_fold_builtins): Likewise.
3932 * tree-ssa-copy.c (pass_copy_prop): Likewise.
3933 * tree-ssa-forwprop.c (pass_forwprop): Likewise.
3934 * tree-ssa-dce.c (pass_dce, pass_dce_loop, pass_cd_dce): Likewise.
3935 * passes.c (init_optimization_passes): Execute rename_ssa_copies,
3936 ccp, forwprop, copy_prop, merge_phi, copy_prop, dce and tail recursion
3938 * tree-ssa-operands.c (add_virtual_operand, get_indirect_ref_operand):
3939 When aliasing is not build, mark statement as volatile.
3941 2007-01-11 Tom Tromey <tromey@redhat.com>
3943 PR preprocessor/15185, PR preprocessor/20989:
3944 * doc/cppopts.texi <-MT>: Update description of algorithm for
3945 computing default target.
3946 <-M, -MD>: Reword "basename" text.
3948 2007-01-11 Roger Sayle <roger@eyesopen.com>
3950 * builtins.c (expand_builtin_pow, expand_builtin_powi,
3951 fold_builtin_cabs, fold_builtin_sqrt, fold_builtin_trunc,
3952 fold_builtin_floor, fold_builtin_ceil, fold_builtin_round,
3953 fold_builtin_int_int_roundingfn, fold_builtin_bitop,
3954 fold_builtin_bswap, real_constp, fold_builtin_pow,
3955 fold_builtin_powi, fold_builtin_signbit, fold_builtin_copysign,
3956 do_mpfr_arg1, do_mpfr_arg2, do_mpfr_arg3, do_mpfr_sincos): Replace
3957 uses of the macro TREE_CONSTANT_OVERFLOW with TREE_OVERFLOW.
3958 * convert.c (convert_to_pointer): Likewise.
3959 * expr.c (highest_pow2_factor, expand_expr_real_1): Likewise.
3960 * fold-const.c (force_fit_type, fold_negate_expr, int_const_binop,
3961 const_binop, fold_convert_const_int_from_int,
3962 fold_convert_const_int_from_real,
3963 fold_convert_const_real_from_real, sign_bit_p,
3964 optimize_minmax_comparison, extract_muldiv_1, fold_div_compare,
3965 fold_sign_changed_comparison, fold_unary, fold_comparison,
3966 fold_binary, multiple_of_p, tree_Expr_non_zero_p,
3967 fold_negate_const, fold_abs_const, fold_not_const): Likewise.
3968 * print-tree.c (print_node_brief, print_node): Likewise.
3969 * stor-layout.c (place_field, layout_type): Likewise.
3970 * tree-chrec.c (keep_cast): Likewise.
3971 * tree.c (build_vector, build_real, build_real_from_int_cst,
3972 build_complex): Likewise.
3974 2007-01-11 Roger Sayle <roger@eyesopen.com>
3976 * tree.h (TREE_CONSTANT_OVERFLOW): Obsolete. For the time being,
3977 treat TREE_CONSTANT_OVERFLOW as a synonym of TREE_OVERFLOW.
3979 2007-01-11 Paolo Bonzini <bonzini@gnu.org>
3981 * configure.ac (strict1_warn): Rename to strict_warn.
3982 (WERROR, --enable-werror, symlink hacks, stage1_cflags,
3983 cc_set_by_configure, quoted_cc_set_by_configure,
3984 stage_prefix_set_by_configure, quoted_stage_prefix_set_by_configure,
3985 all_boot_languages, all_stagestuff): Remove.
3986 (target_list): Remove bootstrap targets.
3987 * Makefile.in (quickstrap): Unconditionally make a synonym of all.
3988 (BOOT_LANGUAGES, STAGE1_CFLAGS, STAGE1_CHECKING,
3989 REMAKEFLAGS, FLAGS_TO_PASS, PREPEND_DOTDOT_TO_RELATIVE_PATHS,
3990 SUBDIR_FLAGS_TO_PASS, WERROR_FLAGS, STRICT2_WARN, LANG_STAGESTUFF,
3991 VOL_FILES, POSTSTAGE1_FLAGS_TO_PASS, STAGE2_FLAGS_TO_PASS,
3992 STAGEPROFILE_FLAGS_TO_PASS, STAGEFEEDBACK_FLAGS_TO_PASS, stage1_build,
3993 stage1_copy, stage2_build, stage2_copy, stageprofile_build,
3994 stageprofile_copy, stage3_build, stage3_copy, stagefeedback_build,
3995 stagefeedback_copy, stage4_build, clean_s1, clean_sw, bootstrap,
3996 bootstrap-lean, bootstrap2, bootstrap2-lean, bootstrap3,
3997 bootstrap3-lean, bootstrap4, bootstrap4-lean, unstage1, unstage2,
3998 unstage3, unstage4, unstageprofile, unstagefeedback, restage, restage2,
3999 restage3, restage4, restageprofile, restagefeedback, bubbleestrap,
4000 cleanstrap, unstrap, restrap, *compare, *compare3, *compare4,
4001 *compare-lean, *compare3-lean, *compare4-lean, stage1-start, stage1,
4002 stage2-start, stage2, stage3-start, stage3, stage4-start, stage4,
4003 stageprofile-start, stageprofile, stagefeedback-start, stagefeedback,
4004 risky-stage1, risky-stage2, risky-stage3, risky-stage4): Remove.
4005 (ORDINARY_FLAGS_TO_PASS): Rename to FLAGS_TO_PASS.
4006 (STAGECOPYSTUFF, STAGEMOVESTUFF): Consolidate into MOSTLYCLEANFILES.
4007 (mostlyclean): Adjust.
4008 (clean, distclean): Don't mention bootstrap stuff.
4009 * configure: Regenerate.
4010 * ada/config-lang.in, cp/config-lang.in, forttran/config-lang.in,
4011 java/config-lang.in, objc/config-lang.in, objcp/config-lang.in,
4012 treelang/config-lang.in (stagestuff): Remove.
4013 * doc/sourcebuild.texi (stage1, stage2, stage3, stage4,
4014 stageprofile, stagefeedback, stagestuff): Remove mention.
4016 2007-01-11 Nick Clifton <nickc@redhat.com>
4018 * config/mcore/predicates.md (mcore_general_movesrc_operand):
4020 (mcore_general_movdst_operand): Do not accept CONST_INTs.
4021 (mcore_arith_K_S_operand): Run the test for the S constraint not
4022 the test for the M constraint.
4023 (mcore_addsub_operand): Do not accept integer values that are
4024 larger than 32 bits.
4025 * config/mcore/mcore.md: Remove unused constraints from split.
4026 (andsi3): Use HOST_WIDE_INT instead of int to hold an INTVAL.
4028 (allocate_stack): Likewise.
4029 * config/mcore/mcore.c (mcore_print_operand): Restrict output of P
4030 operands to 32 bits.
4031 (mcore_const_costs): Use HOST_WIDE_INT instead of int to hold an
4033 (mcore_and_cost, mcore_modify_comparison, const_ok_for_mcore,
4034 mcore_const_ok_for_inline, mcore_const_trick_uses_not,
4035 try_constant_tricks, mcore_num_ones, mcore_num_zeros,
4036 mcore_output_bclri, mcore_output_andn, output_inline_const,
4037 mcore_output_move, mcore_output_movedouble): Likewise.
4038 (mcore_output_cmov): Use CONST_OK_FOR_M and CONST_OK_FOR_N.
4039 (output_inline_const): Likewise.
4040 (output_inline_const): Fix format strings used in sprintf
4042 * config/mcore/mcore-protos.h: Update prototypes for changed
4043 functions in mcore.c.
4044 * config/mcore/mcore.h (CONST_OK_FOR_I): Cast values to
4045 HOST_WIDE_INT and not int.
4046 (CONST_OK_FOR_J, CONST_OK_FOR_K, CONST_OK_FOR_L, CONST_OK_FOR_M,
4047 CONST_OK_FOR_N): Likewise.
4048 (LEGITIMATE_CONSTANT_P): Also check CONSTANT_P.
4049 (GO_IF_LEGITIMATE_INDEX): Use HOST_WIDE_INT instead of int to hold
4052 2007-01-10 Jan Hubicka <jh@suse.cz>
4054 * tree-vrp.c (remove_range_assertions): Release defs.
4055 * tree-ssa-loop-ivopts.c (rmeove_statement): Likewise.
4056 * tree-ssa-dom.c (remove_stmt_or_phi): Likewise.
4058 2007-01-10 Paul Brook <paul@codesourcery.com>
4060 * config/arm/arm.c (arm_rtx_costs_1): Handle mutiply-subtract.
4061 * config/arm/arm.md (mulsi3subsi): New insn.
4063 2007-01-10 Zdenek Dvorak <dvorakz@suse.cz>
4065 * tree-ssa-loop-manip.c (tree_unroll_loop): Make it a wrapper over ...
4066 (tree_transform_and_unroll_loop): New.
4067 * tree-flow.h (transform_callback, tree_transform_and_unroll_loop):
4070 2007-01-10 Robert Kennedy <jimbob@google.com>
4072 * fold-const.c (fold_comparison): Fold comparisons like (x *
4073 1000 < 0) to (x < 0).
4075 2007-01-10 Ian Lance Taylor <iant@google.com>
4077 * tree-pretty-print.c (dump_generic_node): Print parentheses when
4078 operands have the same priority.
4080 2007-01-10 Tom Tromey <tromey@redhat.com>
4082 * fold-const.c (fold_truthop): Don't check can_use_bit_fields_p.
4083 (fold_binary): Likewise.
4084 * langhooks.c (lhd_can_use_bit_fields_p): Removed.
4085 * langhooks-def.h (lhd_can_use_bit_fields_p): Removed.
4086 (LANG_HOOKS_CAN_USE_BIT_FIELDS_P): Removed.
4087 (LANG_HOOKS_INITIALIZER): Remove LANG_HOOKS_CAN_USE_BIT_FIELDS_P.
4088 * langhooks.h (struct lang_hooks): Removed field
4089 'can_use_bit_fields_p'.
4091 2007-01-10 Ralf Corsépius <ralf.corsepius@rtems.org>
4093 * config/bfin/t-bfin, config/bfin/t-bfin-elf: Remove GCC_CFLAGS.
4095 2007-01-10 Razya Ladelsky <razya@il.ibm.com>
4097 * function.c (get_last_funcdef_no): New function.
4098 * function.h (get_last_funcdef_no): Declare.
4099 * tree-inline.c (initialize_cfun): Add initialization.
4100 (tree_function_versioning): Cleanup.
4102 2007-01-10 Jan Hubicka <jh@suse.cz>
4104 * tree-inline.c (setup_one_parameter): Do not propagate into abnormal
4107 2007-01-10 Sa Liu <saliu@de.ibm.com>
4108 Ben Elliston <bje@au.ibm.com>
4110 * spu.h (STACK_SAVE_AREA): Use VOIDmode for SAVE_FUNCTION, SImode
4111 for SAVE_NONLOCAL and Pmode for any other save level.
4112 * spu-protos.h (spu_restore_stack_block): Declare.
4113 * spu.md (save_stack_block): Remove.
4114 (restore_stack_block): Call spu_restore_stack_block.
4115 * spu.c (spu_restore_stack_block): New function.
4116 (spu_expand_epilogue): Remove old comment.
4118 2007-01-09 Zdenek Dvorak <dvorakz@suse.cz>
4120 PR tree-optimization/30322
4121 * tree-ssa-loop-ivopts.c (fold_affine_expr, iv_value): Removed.
4122 (cand_value_at): Return the value as aff_tree.
4123 (may_eliminate_iv): Convert the bound from aff_tree to tree.
4124 * tree-affine.c (aff_combination_add_cst, aff_combination_add_product,
4125 aff_combination_mult): New functions.
4126 (aff_combination_add): Use aff_combination_add_cst.
4127 (aff_combination_convert): Allow conversions to a wider type.
4128 (tree_to_aff_combination): Handle BIT_NOT_EXPR.
4129 * tree-affine.h (aff_combination_mult): Declare.
4131 2007-01-09 Carlos O'Donell <carlos@codesourcery.com>
4133 * doc/tm.texi: Update documentation to reflect reality of exec
4134 and start file search behaviours. Update copyright year.
4135 * doc/invoke.texi: Explain how GCC_EXEC_PREFIX is used to find
4136 header file directories.
4138 2007-01-09 Uros Bizjak <ubizjak@gmail.com>
4140 * config/i386/i386.md (*sinxf2): Rename to *sinxf2_i387.
4141 (*cosxf2): Rename to cosxf2_i387.
4142 (*sindf2, *sinsf2): Extend operand 1 to XFmode. Macroize patterns
4143 using X87MODEF12 mode macro. Rename patterns to
4144 *sin_extend<mode>xf2_i387. Use SSE_FLOAT_MODE_P to disable patterns
4146 (*cosdf2, *cossf2): Ditto.
4147 (sincosdf3, sincossf3): Ditto. Rewrite corresponding splitters
4148 to match extended input operands.
4149 (sincos<mode>3): New expander.
4150 (*sinextendsfdf2, *cosextendsfdf2, *sincosextendsfdf3): Remove
4151 insn patterns and corresponding splitters.
4153 2007-01-09 Kaz Kojima <kkojima@gcc.gnu.org>
4155 * config/sh/t-linux (TARGET_LIBGCC2_CFLAGS): Delete.
4156 (SHLIB_MAPFILES, SHLIB_LINK, SHLIB_INSTALL): Likewise.
4158 2007-01-09 Nicolas Pitre <nico@cam.org>
4161 * arm/ieee754-df.S (Lad_s): Also test the low word of X for zero.
4163 2007-01-08 Geoffrey Keating <geoffk@apple.com>
4165 * target.h (struct gcc_target): New field library_rtti_comdat.
4166 * target-def.h (TARGET_CXX_LIBRARY_RTTI_COMDAT): New.
4167 (TARGET_CXX): Add TARGET_CXX_LIBRARY_RTTI_COMDAT.
4168 * doc/tm.texi (C++ ABI): Document TARGET_CXX_LIBRARY_RTTI_COMDAT.
4169 * config/darwin.h (TARGET_CXX_LIBRARY_RTTI_COMDAT): Define.
4171 2007-01-08 Geoffrey Keating <geoffk@apple.com>
4173 * doc/invoke.texi (Optimize Options): Correct description of -O0.
4175 2007-01-08 Richard Guenther <rguenther@suse.de>
4177 * tree.h (force_fit_type_double): Export.
4178 (force_fit_type): Remove.
4179 * fold-const.c (force_fit_type_double): New function.
4180 (force_fit_type): Remove.
4181 (int_const_binop): Use it.
4182 (fold_convert_const_int_from_int): Likewise.
4183 (fold_convert_const_int_from_real): Likewise.
4184 (fold_div_compare): Likewise.
4185 (fold_sign_changed_comparison): Likewise.
4186 (fold_unary): Likewise.
4187 (fold_negate_const): Likewise.
4188 (fold_abs_const): Likewise.
4189 (fold_not_const): Likewise.
4190 * c-common.c (shorten_compare): Use force_fit_type_double.
4191 * convert.c (convert_to_pointer): Likewise.
4193 2007-01-08 Richard Guenther <rguenther@suse.de>
4195 * tree.h (build_int_cst_wide_type): Export.
4196 * tree.c (build_int_cst_wide_type): New function.
4197 (build_int_cst_wide): Fix comment.
4198 * builtins.c (fold_builtin_object_size): Use build_int_cst
4199 to build -1 or 0 of the correct type. Use fit_double_type
4200 to check for overflow.
4201 * fold-const.c (optimize_bit_field_compare): Use build_int_cst_type
4203 (decode_field_reference): Likewise.
4204 (all_ones_mask_p): Likewise.
4205 (native_interpret_int): Use build_int_cst_wide_type.
4206 (fold_binary): Use build_int_cst_type to build an all-ones
4208 * stor-layout.c (set_sizetype): Use build_int_cst_wide_type.
4210 2007-01-08 Daniel Jacobowitz <dan@codesourcery.com>
4212 * config/pa/t-pa64 (libgcc_stub.a): Use $(T).
4214 2007-01-09 Ben Elliston <bje@au.ibm.com>
4216 * genautomata.c (STATS_OPTION): New option.
4217 (stats_flag): New flag.
4218 (gen_automata_option): Handle it.
4219 (initiate_automaton_gen): Ditto.
4220 (write_automata): Output statistics only if stats_flag is
4221 set. Likewise, output time statistics only if time_flag is set.
4222 * doc/md.texi (Processor pipeline description): Document new flag.
4224 2007-01-08 Richard Guenther <rguenther@suse.de>
4226 * builtins.c (fold_builtin_int_roundingfn): Use fit_double_type.
4227 * tree.c (build_int_cst_type): Likewise.
4228 (size_in_bytes): Don't call force_fit_type on the result.
4229 (int_fits_type_p): Use fit_double_type.
4230 * fold-const.c (fit_double_type): New function.
4231 (force_fit_type): Use it.
4232 * tree.h (fit_double_type): Export.
4234 2007-01-08 Jan Hubicka <jh@suse.cz>
4236 * tree-vectorizer.c (gate_increase_alignment): Fix return type.
4237 * ipa.c (function_and_variable_visibility): Fix return type.
4239 2007-01-08 Richard Guenther <rguenther@suse.de>
4241 * tree-ssa-ccp.c (maybe_fold_offset_to_array_ref): Use type
4242 of offset to build the index.
4243 * tree-pretty-print.c (dump_generic_node): Don't build negated
4244 const just for printing.
4245 * c-pretty-print.c (pp_c_integer_constant): Likewise.
4246 * builtins.c (fold_builtin_int_roundingfn): Check if result
4247 fits the type by using force_fit_type and comparing the result.
4248 * predict.c (predict_loops): Use compare_tree_int for comparison.
4249 * tree.c (build_int_cst): Fall back to integer_type_node for
4251 (build_int_cst_wide): Assert type is non-null.
4253 2007-01-08 Roberto Costa <roberto.costa@st.com>
4255 * tree-vrp.c (extract_range_from_cond_expr): New.
4256 (extract_range_from_expr): Handle COND_EXPR nodes used as expressions.
4257 * tree-ssa-ccp.c (get_maxval_strlen): Handle COND_EXPR nodes used
4259 (fold_stmt): Bug fix, avoid infinite recursion when folding COND_EXPRs.
4260 * tree-ssa-forwprop.c (simplify_cond, forward_propagate_into_cond,
4261 tree_ssa_forward_propagate_single_use_vars): Handle COND_EXPR nodes
4262 used as expressions.
4263 * tree-object-size.c (cond_expr_object_size): New.
4264 (collect_object_sizes_for): Handle COND_EXPR nodes used as expressions.
4266 2007-01-08 Jan Hubicka <jh@suse.cz>
4268 * tree-ssa-forwprop.c (forward_propagate_into_cond,
4269 tree_ssa_forward_propagate_single_use_va): Release defs of propagated
4272 2007-01-08 Richard Guenther <rguenther@suse.de>
4274 PR tree-optimization/23603
4275 * tree-vrp.c (set_value_range_to_truthvalue): New function.
4276 (extract_range_from_binary): Fall back to truthvalue instead of
4277 varying for TRUTH_*_EXPR.
4278 (extract_range_from_comparison): Fall back to truthvalue instead of
4280 (vrp_visit_phi_node): Don't adjust new range bounds to +INF/-INF
4281 if all visited PHI values were constant.
4283 2007-01-08 Jan Hubicka <jh@suse.cz>
4285 * cgraphunit.c (cgraph_process_new_functions): Reset reachable flag.
4286 (cgraph_analyze_function): break out from ...
4287 (cgraph_finalize_compilation_unit): ... here.
4288 (cgraph_expand_function): Remove forgoten commented out line.
4289 (cgraph_optimize): Analyze functions.
4291 2007-01-08 Jan Hubicka <jh@suse.cz>
4293 * tree-pas.h (TODO_remove_function): New flag.
4294 (TODO_update*): Renumber.
4295 (pass_ipa_increase_alignment,
4296 pass_ipa_function_and_variable_visibility): New passes.
4297 * cgraphunit.c (cgraph_increase_alignment): Move to tree-vectorizer.c
4298 (cgraph_function_and_variable_visibility): Move to ipa.c
4299 (cgraph_optimize): Don't call cgraph_function_and_variable_visibility,
4300 cgraph_increase_alignment.
4301 * ipa-inline.c (cgraph_decide_inlining): Don't push timevar.
4302 (cgraph_decide_inlining_incrementally): Push TV_INTEGRATION before
4303 calling tree-inline.
4304 (cgraph_early_inlining): Do not call cgraph_remove_unreachable_nodes.
4305 (pass_ipa_inline, pass_early_ipa_inlining): Set TODO_remove_functions
4306 * tree-vectorizer.c (increase_alignment): Move here from cgraphunit.c
4307 (gate_increase_alignment): New function.
4308 (pass_ipa_increase_alignment): New pass.
4309 * ipa.c: Inline tree-pass.h and timevar.h
4310 (function_and_variable_visibility): Move here from cgraphunit.c
4311 * tree-optimize.c (pass_early_local_passes): Add TODO_remove_functions.
4312 * passes.c (init_optimization_passes): Add the two new passes.
4313 (execute_todo): Handle cgraph_remove_functions.
4315 2007-01-08 Nick Clifton <nickc@redhat.com>
4317 * config/frv/predicates.md (reg_or_0_operand): Accept
4320 2007-01-08 Ralf Corsépius <ralf.corsepius@rtems.org>
4322 * config/bfin/rtems.h, config/bfin/t-rtems: New.
4323 * config.gcc: Add bfin*-rtems*.
4325 2007-01-08 Mark Shinwell <shinwell@codesourcery.com>
4327 * c.opt: Add -flax-vector-conversions.
4328 * c-typeck.c (convert_for_assignment): Pass flag to
4329 vector_types_convertible_p to allow emission of note.
4330 (digest_init): Likewise.
4331 * c-opts.c: Handle -flax-vector-conversions.
4332 * c-common.c (flag_lax_vector_conversions): New.
4333 (vector_types_convertible_p): Unless -flax-vector conversions
4334 has been passed, disallow conversions between vectors with
4335 differing numbers of subparts and/or element types. If such
4336 a conversion is disallowed, possibly emit a note on the first
4337 occasion only to inform the user of -flax-vector-conversions.
4338 The new last argument specifies this.
4339 * c-common.h (flag_lax_vector_conversions): New.
4340 (vector_types_convertible_p): Add extra argument.
4341 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use
4342 char_type_node for V*QI type vectors.
4343 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
4344 Update to satisfy new typechecking rules.
4345 * config/rs6000/altivec.h (vec_cmple): Use vec_cmpge, for both
4347 * doc/invoke.texi (C Dialect Options): Document
4348 -flax-vector-conversions.
4350 2007-01-08 Mark Shinwell <shinwell@codesourcery.com>
4352 PR tree-optimization/29877
4353 * tree-ssa-ter.c (is_replaceable_p): Deem assignments with
4354 a register variable on the RHS to not be replaceable.
4356 2007-01-08 Chen Liqin <liqin@sunnorth.com.cn>
4357 * config/score/t-score-elf (MULTILIB_OPTIONS): Change.
4358 * config/score/predicates.md (const_uimm5, sr0_operand, const_simm12,
4359 const_simm15, const_pow2, const_npow2): Added.
4360 * config/score/misc.md (insv, extv, extzv, movmemsi,
4361 move_lbu_a/b, mov_lhu_a/b etc): Added and fix some bug.
4362 * config/score/score.c (score_address_cost, score_select_cc_mode):
4364 Change CONST_OK_FOR_LETTER_P/EXTRA_CONSTRAINT define.
4365 Update score_rtx_costs for MACRO TARGET_RTX_COSTS.
4366 Update score_print_operand.
4367 * config/score/score.h (DATA_ALIGNMENT, SELECT_CC_MODE): Added.
4368 Adjust register allocate order and update some macro define.
4369 * config/score/score-mdaux.c (mdx_unaligned_load, mdx_unsigned_store,
4370 mdx_block_move_straight, mdx_block_move_loop_head,
4371 mdx_block_move_loop_body, mdx_block_move_loop_foot, mdx_block_move_loop,
4372 mdx_block_move): Added.
4373 (mdx_movsicc, mdp_select_add_imm, mdp_select, mds_zero_extract_andi,
4374 mdp_limm): Updated and fix some bug and typo.
4375 * config/score/score.md (movqi/hi/si, add/sub/zero/ext): Updated.
4376 (movsf, movdf, doloop_end): Added.
4378 2007-01-08 Kazu Hirata <kazu@codesourcery.com>
4380 * config/arm/arm.c, config/arm/arm.h, config/arm/arm.md,
4381 config/arm/thumb2.md: Fix comment typos.
4382 * doc/extend.texi: Fix a typo.
4384 2007-01-07 Eric Christopher <echristo@apple.com>
4386 * configure.ac: Check for __stack_chk_fail for darwin.
4387 * configure: Regenerate.
4389 2007-01-07 Richard Guenther <rguenther@suse.de>
4391 * tree-vrp.c (extract_range_from_assert): CSE calls to
4392 compare_values where possible.
4393 (extract_range_from_unary_expr): Likewise.
4395 2007-01-07 Anatoly Sokolov <aesok@post.ru>
4397 * config/avr/avr-protos.h (call_insn_operand): Delete prototype.
4398 * config/avr/avr.c (call_insn_operand): Delete function.
4399 * config/avr/avr.md (*pushqi, *pushhi, *pushsi, *pushsf): Use REG_SP
4400 instead of register number. Use predicates.
4401 * config/avr/predicates.md (const0_operand, reg_or_0_operand,
4402 call_insn_operand): Add.
4404 2007-01-06 Jan Hubicka <jh@suse.cz>
4406 * tree-pass.h (pass_build_cgraph_edges): Declare.
4407 * cgraphunit.c (record_refernece): Move to cgraphbuild.c
4408 (visited_nodes): Remove.
4409 (cgraph_create_edges): Move to cgraphbuild.c; rename to
4410 build_cgrpah_edges; make visited_nodes local.
4411 (cgraph_process_new_functions): DO not call initialize_inline_failed.
4412 (record_references_in_initializer): Move to cgraphbuild.c
4413 (initialize_inline_failed, rebuild_cgraph_edges,
4414 pass_rebuild_cgraph_edges): Move to cgraphbuild.c.
4415 (verify_cgraph_node): Make visited_nodes local.
4416 (cgraph_analyze_function): Do not call cgraph_create_edges and
4417 initialize_inline_failed.
4418 (cgraph_expand_function): Do not call cgraph_lower_function;
4419 assert that function is already lowered.
4420 * Makefile.in (cgraphbuild.o): New.
4421 * passes.c (init_optimization_passes): Add pass_build_cgraph_edges
4422 at the end of lowering passes.
4424 2007-01-06 Steven Bosscher <steven@gcc.gnu.org>
4426 * ifcvt.c (cond_move_convert_if_block): New function, code
4427 factored out from...
4428 (cond_move_process_if_block): ...here. Call the new function
4429 on the THEN and ELSE blocks.
4430 (merge_if_block): Do not copy global_live_at_end, merge_blocks
4431 already takes care of this.
4433 2007-01-05 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
4436 * tree.h (TREE_OVERFLOW_P): New.
4437 * c-typeck.c (parser_build_unary_op): Warn only if result
4438 overflowed and operands did not.
4439 (parser_build_binary_op): Likewise.
4440 (convert_for_assignment): Remove redundant overflow_warning.
4441 * c-common.c (overflow_warning): Don't check or set TREE_OVERFLOW.
4443 2007-01-05 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
4445 * c-typeck.c (store_init_value): Split over two lines to follow
4446 the GNU coding style.
4448 2007-01-05 Benjamin Kosnik <bkoz@redhat.com>
4450 * c-cppbuiltin.c (c_cpp_builtins): __GXX_EXPERIMENTAL_CPP0X__ to
4451 __GXX_EXPERIMENTAL_CXX0X__.
4452 * doc/cpp.texi: Same.
4454 2007-01-05 Richard Guenther <rguenther@suse.de>
4457 * tree.c (get_narrower): Do not construct COMPONENT_REFs
4458 with mismatched types. Instead explicitly build a
4459 conversion NOP_EXPR.
4461 2007-01-05 Ian Lance Taylor <iant@google.com>
4463 * c-common.c (decl_with_nonnull_addr_p): New function.
4464 (c_common_truthvalue_conversion): Call it.
4465 * c-typeck.c (build_binary_op): Likewise.
4466 * c-common.h (decl_with_nonnull_addr_p): Declare.
4468 2007-01-05 Jakub Jelinek <jakub@redhat.com>
4471 * libgcc2.c (__divdc3): Compare c and d against 0.0 instead of
4474 2007-01-05 Joel Brobecker <brobecker@adacore.com>
4476 * doc/install.texi (Final install): Document the fact that
4477 the GNAT runtime should not be stripped.
4479 2007-01-04 Jan Hubicka <jh@suse.cz>
4481 * tree-inline.c (fold_marked_statements): Update operand caches
4482 and EH after folding
4484 2007-01-04 Ian Lance Taylor <iant@google.com>
4486 * c-common.c (check_function_nonnull): Whitespace fix.
4488 2007-01-04 Jan Hubicka <jh@suse.cz>
4490 * tree-optimize.c (execute_fixup_cfg): Correct previously mistakely
4491 comitted older version of patch.
4492 (pass_fixup_cfg): Add TODOs to verify flow and statements, dump
4493 function, celanup cfg and collect garbage.
4495 2007-01-04 Mike Stump <mrs@apple.com>
4497 * Makefile.in (mostlyclean): Don't remove libgcc anymore.
4500 2007-01-04 Eric Christopher <echristo@apple.com>
4502 * libgcc2.c (__bswapsi2): Use SItype.
4503 (__bswapdi2): Use DItype.
4504 * libgcc2.h: Update for above.
4506 2007-01-04 Paul Brook <paul@codesourcery.com>
4508 * config/arm/arm.md (arm_mulsi3, thumb_mulsi3, mulsi3_compare0,
4509 mulsi_compare0_scratch, mulsi3addsi, mulsi3addsi_compare0,
4510 mulsi3addsi_compare0_scratch, mulsidi3adddi, mulsidi3,
4511 umulsidi3, umulsidi3adddi, smulsi3_highpart,
4512 umulsi3_highpart): Make conditional on !arm_arch6.
4513 (arm_mulsi3_v6, thumb_mulsi3_v6, mulsi3_compare0_v6,
4514 mulsi_compare0_scratch_v6, mulsi3addsi_v6, mulsi3addsi_compare0_v6,
4515 mulsi3addsi_compare0_scratch_v6, mulsidi3adddi_v6, mulsidi3_v6,
4516 umulsidi3_v6, umulsidi3adddi_v6, smulsi3_highpart_v6,
4517 umulsi3_highpart_v6): New insns.
4519 2007-01-04 Roger Sayle <roger@eyesopen.com>
4521 * fold-const.c (fold_convert): When casting an expression to void,
4522 fold_ignored_result may discover a GIMPLE_MODIFY_STMT which doesn't
4523 have a type. Instead of attempting to build a NOP_EXPR, return
4524 these "special" trees directly.
4526 2007-01-04 Joseph Myers <joseph@codesourcery.com>
4528 * config/rs6000/rs6000.c (rs6000_rtx_costs): Make adjustment for
4529 MULT inside MINUS as either argument. Use rs6000_cost->dmul -
4530 rs6000_cost->fp not 0 as adjustment for outer NEG.
4532 2007-01-04 Jan Hubicka <jh@suse.cz>
4534 * cgraph.c (cgraph_release_function_body): New function.
4535 (cgraph_remove_node): Use it.
4536 * cgraph.h (cgraph_release_function_body): Declare.
4537 * cgraphunit.c (cgraph_expand_function): Use it.
4538 * ipa.c (cgraph_remove_unreahchable_nodes): Use it.
4539 * tree-ssa.c (delete_tree_ssa): Allow to be called before aliasing
4540 is initialized and while compilation of other function is running.
4541 * tree-optimize.c (execute_free_cfg_annotations): Move code to clear
4542 statement CFG annotations from here to ...
4543 * tree-cfg.c (delete_tree_cfg_annotations): ... here.
4545 2007-01-04 Zdenek Dvorak <dvorakz@suse.cz>
4547 * cfgloop.h (enum li_flags): Make the constants powers of two.
4549 2007-01-04 Jan Hubicka <jh@suse.cz>
4551 * tree-inline.c (copy_bb): Insert new statements to statements_to_fold
4553 (fold_marked_statements): New function.
4554 (optimize_inline_calls, tree_function_versioning): Fold new statements.
4555 * tree-inline.h (copy_body_data): Add statements_to_fold.
4557 2007-01-03 Daniel Jacobowitz <dan@codesourcery.com>
4559 * config.gcc: Mention libgcc/config.host.
4560 * Makefile.in: Update comments mentioning libgcc.
4561 (LIBGCC, INSTALL_LIBGCC, GCC_PARTS, mklibgcc): Delete.
4562 (all.cross, start.encap, rest.encap, rest.cross): Update
4563 dependencies for libgcc move.
4564 (libgcc.mk, LIBGCC_DEPS, libgcov.a, libgcc.a, stmp-multilib)
4565 (clean-target, clean-target-libgcc): Delete.
4566 (srcdirify, GCC_EXTRA_PARTS): New macros.
4567 (libgcc-support, libgcc.mvars): New rules.
4568 (distclean): Remove mention of mklibgcc.
4569 (install): Don't reference INSTALL_LIBGCC.
4570 (install-common): Don't reference EXTRA_PARTS.
4571 (install-libgcc, install-multilib): Delete rules.
4572 * mklibgcc.in: Delete file.
4573 * doc/configfiles.texi: Don't mention mklibgcc.
4575 * config/i386/t-darwin (SHLIB_VERPFX): Delete (moved to libgcc).
4576 * config/i386/t-darwin64 (SHLIB_VERPFX): Likewise.
4577 * config/rs6000/t-darwin (SHLIB_VERPFX): Likewise.
4578 * config/rs6000/t-ppccomm (TARGET_LIBGCC2_CFLAGS, SHLIB_MAPFILES)
4579 (mklibgcc, ldblspecs): Likewise.
4581 * config/i386/t-nwld (libgcc.def, libc.def, libpcre.def)
4582 (posixpre.def): Use $(T).
4583 (SHLIB_EXT, SHLIB_NAME, SHLIB_SLIBDIR_QUAL, SHLIB_DEF, SHLIB_MAP)
4584 (SHLIB_SRC, SHLIB_INSTALL): Delete.
4585 (SHLIB_LINK): Make dummy.
4586 * config/t-slibgcc-darwin: Delete contents except for dummy SHLIB_LINK.
4588 * config/frv/t-linux (EXTRA_MULTILIB_PARTS): Clear.
4590 * config/alpha/t-crtfm: Use $(T) in rules for EXTRA_PARTS.
4591 * config/alpha/t-vms, config/alpha/t-vms64, config/fr30/t-fr30,
4592 config/i386/t-rtems-i386, config/ia64/t-ia64, config/rs6000/t-beos,
4593 config/rs6000/t-newas, config/sparc/t-elf: Likewise.
4595 * configure.ac (all_outputs): Remove mklibgcc.
4596 * configure: Regenerated.
4598 2007-01-03 Josh Conner <jconner@apple.com>
4601 * calls.c (compute_argument_addresses): Set stack and stack_slot
4602 for partial args, too.
4603 (store_one_arg): Use locate.size.constant for the size when
4604 generating a save_area.
4606 2007-01-03 Robert Kennedy <jimbob@google.com>
4608 * tree-cfg.c (tree_merge_blocks): Release SSA_NAME phi results
4609 whose definitions are deleted due to basic block merging.
4611 2007-01-03 Paul Brook <paul@codesourcery.com>
4614 * config/arm/arm.c (output_return_instruction): Pop PC in interrupt
4616 (use_return_insn): Return 0 for Thumb interrupt functions.
4617 (print_multi_reg): Add rfe argument for IRQ returns.
4618 (arm_output_epilogue): Pop interrupt return address directly into PC.
4619 (arm_expand_prologue): Only adjust IRQ return address in Arm mode.
4621 2007-01-03 Paul Brook <paul@codesourcery.com>
4623 Merge from sourcerygxx-4_1.
4624 * config/arm/thumb2.md: New file.
4625 * config/arm/elf.h (JUMP_TABLES_IN_TEXT_SECTION): Return True for
4627 * config/arm/coff.h (JUMP_TABLES_IN_TEXT_SECTION): Ditto.
4628 * config/arm/aout.h (ASM_OUTPUT_ADDR_VEC_ELT): Add !Thumb-2 assertion.
4629 (ASM_OUTPUT_ADDR_DIFF_ELT): Output Thumb-2 jump tables.
4630 * config/arm/aof.h (ASM_OUTPUT_ADDR_DIFF_ELT): Output Thumb-2 jump
4632 (ASM_OUTPUT_ADDR_VEC_ELT): Add !Thumb-2 assertion.
4633 * config/arm/ieee754-df.S: Use macros for Thumb-2/Unified asm
4635 * config/arm/ieee754-sf.S: Ditto.
4636 * config/arm/arm.c (thumb_base_register_rtx_p): Rename...
4637 (thumb1_base_register_rtx_p): ... to this.
4638 (thumb_index_register_rtx_p): Rename...
4639 (thumb1_index_register_rtx_p): ... to this.
4640 (thumb_output_function_prologue): Rename...
4641 (thumb1_output_function_prologue): ... to this.
4642 (thumb_legitimate_address_p): Rename...
4643 (thumb1_legitimate_address_p): ... to this.
4644 (thumb_rtx_costs): Rename...
4645 (thumb1_rtx_costs): ... to this.
4646 (thumb_compute_save_reg_mask): Rename...
4647 (thumb1_compute_save_reg_mask): ... to this.
4648 (thumb_final_prescan_insn): Rename...
4649 (thumb1_final_prescan_insn): ... to this.
4650 (thumb_expand_epilogue): Rename...
4651 (thumb1_expand_epilogue): ... to this.
4652 (arm_unwind_emit_stm): Rename...
4653 (arm_unwind_emit_sequence): ... to this.
4654 (thumb2_legitimate_index_p, thumb2_legitimate_address_p,
4655 thumb1_compute_save_reg_mask, arm_dwarf_handle_frame_unspec,
4656 thumb2_index_mul_operand, output_move_vfp, arm_shift_nmem,
4657 arm_save_coproc_regs, thumb_set_frame_pointer, arm_print_condition,
4658 thumb2_final_prescan_insn, thumb2_asm_output_opcode, arm_output_shift,
4659 thumb2_output_casesi): New functions.
4660 (TARGET_DWARF_HANDLE_FRAME_UNSPEC): Define.
4661 (FL_THUMB2, FL_NOTM, FL_DIV, FL_FOR_ARCH6T2, FL_FOR_ARCH7,
4662 FL_FOR_ARCH7A, FL_FOR_ARCH7R, FL_FOR_ARCH7M, ARM_LSL_NAME,
4663 THUMB2_WORK_REGS): Define.
4664 (arm_arch_notm, arm_arch_thumb2, arm_arch_hwdiv, arm_condexec_count,
4665 arm_condexec_mask, arm_condexec_masklen)): New variables.
4666 (all_architectures): Add armv6t2, armv7, armv7a, armv7r and armv7m.
4667 (arm_override_options): Check new CPU capabilities.
4668 Set new architecture flag variables.
4669 (arm_isr_value): Handle v7m interrupt functions.
4670 (user_return_insn): Return 0 for v7m interrupt functions. Handle
4672 (const_ok_for_arm): Handle Thumb-2 constants.
4673 (arm_gen_constant): Ditto. Use movw when available.
4674 (arm_function_ok_for_sibcall): Return false for v7m interrupt
4676 (legitimize_pic_address, arm_call_tls_get_addr): Handle Thumb-2.
4677 (thumb_find_work_register, arm_load_pic_register,
4678 legitimize_tls_address, arm_address_cost, load_multiple_sequence,
4679 emit_ldm_seq, emit_stm_seq, arm_select_cc_mode, get_jump_table_size,
4680 print_multi_reg, output_mov_long_double_fpa_from_arm,
4681 output_mov_long_double_arm_from_fpa, output_mov_double_fpa_from_arm,
4682 output_mov_double_fpa_from_arm, output_move_double,
4683 arm_compute_save_reg_mask, arm_compute_save_reg0_reg12_mask,
4684 output_return_instruction, arm_output_function_prologue,
4685 arm_output_epilogue, arm_get_frame_offsets, arm_regno_class,
4686 arm_output_mi_thunk, thumb_set_return_address): Ditto.
4687 (arm_expand_prologue): Handle Thumb-2. Use arm_save_coproc_regs.
4688 (arm_coproc_mem_operand): Allow POST_INC/PRE_DEC.
4689 (arithmetic_instr, shift_op): Use arm_shift_nmem.
4690 (arm_print_operand): Use arm_print_condition. Handle '(', ')', '.',
4692 (arm_final_prescan_insn): Use extract_constrain_insn_cached.
4693 (thumb_expand_prologue): Use thumb_set_frame_pointer.
4694 (arm_file_start): Output directive for unified syntax.
4695 (arm_unwind_emit_set): Handle stack alignment instruction.
4696 * config/arm/lib1funcs.asm: Remove default for __ARM_ARCH__.
4697 Add v6t2, v7, v7a, v7r and v7m.
4698 (RETLDM): Add Thumb-2 code.
4699 (do_it, shift1, do_push, do_pop, COND, THUMB_SYNTAX): New macros.
4700 * config/arm/arm.h (TARGET_CPU_CPP_BUILTINS): Define __thumb2__.
4701 (TARGET_THUMB1, TARGET_32BIT, TARGET_THUMB2, TARGET_DSP_MULTIPLY,
4702 TARGET_INT_SIMD, TARGET_UNIFIED_ASM, ARM_FT_STACKALIGN, IS_STACKALIGN,
4703 THUMB2_TRAMPOLINE_TEMPLATE, TRAMPOLINE_ADJUST_ADDRESS,
4704 ASM_OUTPUT_OPCODE, THUMB2_GO_IF_LEGITIMATE_ADDRESS,
4705 THUMB2_LEGITIMIZE_ADDRESS, CASE_VECTOR_PC_RELATIVE,
4706 CASE_VECTOR_SHORTEN_MODE, ADDR_VEC_ALIGN, ASM_OUTPUT_CASE_END,
4707 ADJUST_INSN_LENGTH): Define.
4708 (TARGET_REALLY_IWMMXT, TARGET_IWMMXT_ABI, CONDITIONAL_REGISTER_USAGE,
4709 STATIC_CHAIN_REGNUM, HARD_REGNO_NREGS, INDEX_REG_CLASS,
4710 BASE_REG_CLASS, MODE_BASE_REG_CLASS, SMALL_REGISTER_CLASSES,
4711 PREFERRED_RELOAD_CLASS, SECONDARY_OUTPUT_RELOAD_CLASS,
4712 SECONDARY_INPUT_RELOAD_CLASS, LIBCALL_VALUE, FUNCTION_VALUE_REGNO_P,
4713 TRAMPOLINE_SIZE, INITIALIZE_TRAMPOLINE, HAVE_PRE_INCREMENT,
4714 HAVE_POST_DECREMENT, HAVE_PRE_DECREMENT, HAVE_PRE_MODIFY_DISP,
4715 HAVE_POST_MODIFY_DISP, HAVE_PRE_MODIFY_REG, HAVE_POST_MODIFY_REG,
4716 REGNO_MODE_OK_FOR_BASE_P, LEGITIMATE_CONSTANT_P,
4717 REG_MODE_OK_FOR_BASE_P, REG_OK_FOR_INDEX_P, GO_IF_LEGITIMATE_ADDRESS,
4718 LEGITIMIZE_ADDRESS, THUMB2_LEGITIMIZE_ADDRESS,
4719 GO_IF_MODE_DEPENDENT_ADDRESS, MEMORY_MOVE_COST, BRANCH_COST,
4720 ASM_APP_OFF, ASM_OUTPUT_CASE_LABEL, ARM_DECLARE_FUNCTION_NAME,
4721 FINAL_PRESCAN_INSN, PRINT_OPERAND_PUNCT_VALID_P,
4722 PRINT_OPERAND_ADDRESS): Adjust for Thumb-2.
4723 (arm_arch_notm, arm_arch_thumb2, arm_arch_hwdiv): New declarations.
4724 * config/arm/arm-cores.def: Add arm1156t2-s, cortex-a8, cortex-r4 and
4726 * config/arm/arm-tune.md: Regenerate.
4727 * config/arm/arm-protos.h: Update prototypes.
4728 * config/arm/vfp.md: Enable patterns for Thumb-2.
4729 (arm_movsi_vfp): Add movw alternative. Use output_move_vfp.
4730 (arm_movdi_vfp, movsf_vfp, movdf_vfp): Use output_move_vfp.
4731 (thumb2_movsi_vfp, thumb2_movdi_vfp, thumb2_movsf_vfp,
4732 thumb2_movdf_vfp, thumb2_movsfcc_vfp, thumb2_movdfcc_vfp): New.
4733 * config/arm/libunwind.S: Add Thumb-2 code.
4734 * config/arm/constraints.md: Update include Thumb-2.
4735 * config/arm/ieee754-sf.S: Add Thumb-2/Unified asm support.
4736 * config/arm/ieee754-df.S: Ditto.
4737 * config/arm/bpabi.S: Ditto.
4738 * config/arm/t-arm (MD_INCLUDES): Add thumb2.md.
4739 * config/arm/predicates.md (low_register_operand,
4740 low_reg_or_int_operand, thumb_16bit_operator): New.
4741 (thumb_cmp_operand, thumb_cmpneg_operand): Rename...
4742 (thumb1_cmp_operand, thumb1_cmpneg_operand): ... to this.
4743 * config/arm/t-arm-elf: Add armv7 multilib.
4744 * config/arm/arm.md: Update patterns for Thumb-2 and Unified asm.
4746 (UNSPEC_STACK_ALIGN, ce_count): New.
4747 (arm_incscc, arm_decscc, arm_umaxsi3, arm_uminsi3,
4748 arm_zero_extendsidi2, arm_zero_extendqidi2): New
4750 * config/arm/fpa.md: Update patterns for Thumb-2 and Unified asm.
4751 (thumb2_movsf_fpa, thumb2_movdf_fpa, thumb2_movxf_fpa,
4752 thumb2_movsfcc_fpa, thumb2_movdfcc_fpa): New insns.
4753 * config/arm/cirrus.md: Update patterns for Thumb-2 and Unified asm.
4754 (cirrus_thumb2_movdi, cirrus_thumb2_movsi_insn,
4755 thumb2_cirrus_movsf_hard_insn, thumb2_cirrus_movdf_hard_insn): New
4757 * doc/extend.texi: Document ARMv7-M interrupt functions.
4758 * doc/invoke.texi: Document Thumb-2 new cores+architectures.
4760 2007-01-03 Jakub Jelinek <jakub@redhat.com>
4762 * unwind-dw2.c (SIGNAL_FRAME_BIT, EXTENDED_CONTEXT_BIT): Define.
4763 (struct _Unwind_Context): Rename args_size to flags, remove
4764 signal_frame field, add a new args_size field and version field.
4765 (_Unwind_IsSignalFrame, _Unwind_SetSignalFrame,
4766 _Unwind_IsExtendedContext): New inline functions.
4767 (_Unwind_GetGR, _Unwind_SetGR, _Unwind_GetGRPtr, _Unwind_SetGRPtr):
4768 Assume by_value array is only present if _Unwind_IsExtendedContext.
4769 (_Unwind_GetIPInfo, execute_cfa_program, uw_frame_state_for): Use
4770 _Unwind_IsSignalFrame.
4771 (__frame_state_for): Initialize context.flags to EXTENDED_CONTEXT_BIT.
4772 (uw_update_context_1): Use _Unwind_SetSignalFrame.
4773 (uw_init_context_1): Initialize context->flags to
4774 EXTENDED_CONTEXT_BIT.
4775 * config/rs6000/linux-unwind.h (frob_update_context): Use
4776 _Unwind_SetSignalFrame.
4778 2007-01-03 Andrew Pinski <andrew_pinski@playstation.sony.com>
4781 * gimplify.c (gimplify_modify_expr_complex_part): Move below
4782 tree_to_gimple_tuple. Call tree_to_gimple_tuple when we need
4785 2007-01-03 Kazu Hirata <kazu@codesourcery.com>
4787 * config/i386/i386.h (NON_STACK_REG_P, REGNO_OK_FOR_SIREG_P,
4788 REGNO_OK_FOR_DIREG_P, REWRITE_ADDRESS, ASM_OPERAND_LETTER,
4789 RET, AT_SP): Remove.
4790 * config/i386/i386.md (*sse_prologue_save_insn): Use return
4793 * alias.c (init_alias_analysis): Use VEC_safe_grow_cleared.
4794 * cfgbuild.c (find_basic_blocks): Likewise.
4795 * cfgrtl.c (rtl_create_basic_block): Likewise.
4796 * function.c (temp_slots_at_level): Likewise.
4797 * reg-stack.c (stack_regs_mentioned): Likewise.
4798 * regclass.c (allocate_reg_info): Likewise.
4799 * tree-cfg.c (init_empty_tree_cfg, build_tree_cfg, create_bb,
4800 set_bb_for_stmt, move_block_to_fn): Likewise.
4801 * tree-complex.c (tree_lower_complex): Likewise.
4802 * vec.h (VEC_safe_grow_cleared): New.
4804 * cgraphunit.c, tree-ssa-alias.c: Fix comment typos.
4806 2007-01-03 Zdenek Dvorak <dvorakz@suse.cz>
4808 * loop-unswitch.c (unswitch_loop): Pass probabilities to loopify.
4809 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Pass probabilities
4811 * cfgloopmanip.c (scale_loop_frequencies): Export.
4812 (loopify): Scale the frequencies by prescribed coefficients.
4813 (set_zero_probability): New function.
4814 (duplicate_loop_to_header_edge): Improve updating of frequencies.
4815 (lv_adjust_loop_entry_edge, loop_version): Set probabilities
4816 and frequencies according to arguments.
4817 * tree-ssa-loop-manip.c (tree_unroll_loop): Set probabilities
4819 * cfg.c (scale_bbs_frequencies_int): Allow scaling the frequencies up.
4820 * modulo-sched.c (sms_schedule): Set probabilities for entering
4821 versioned loop correctly.
4822 * tree-vect-transform.c (vect_transform_loop): Ditto.
4823 * cfgloop.h (loopify, loop_version): Declaration changed.
4824 (scale_loop_frequencies): Declared.
4826 2007-01-02 Jan Hubicka <jh@suse.cz>
4828 * cgraph.c: Include tree-flow.h
4829 (cgraph_add_new-function): Handle IPA_SSA mode; execute
4831 * cgraph.h (enum cgraph_state): Add CGRAPH_STATE_IPA_SSA.
4832 * tree-pass.h (pass_all_early_optimizations): Declare.
4833 * cgraphunit.c (cgraph_process_new_functions): Add IPA_SSA; execute
4835 (cgraph_analyze_function): Do early_local_passes.
4836 * tree-mudflap.c (mf_decl_cache_locals, mf_build_check_statement_for):
4837 Do not add referenced vars.
4838 * tree-optimize.c (gate_all_optimizations): Do not execute when not in
4840 (gate_all_early_local_passes): New gate.
4841 (pass_early_local_passes): Use new gate.
4842 (execute_early_local_optimizations): New functions.
4843 (gate_all_early_optimizations): New gate.
4844 (pass_all_early_optimizations): New pass.
4845 (execute_free_datastructures): Free SSA only when initialized.
4846 (gate_init_datastructures): Init only when optimizing.
4847 (tree_lowering_passes): Do early local passes when called late.
4848 * tree-profile.c (do_tree_profiling): Don't profile functions added
4850 (do_early_tree_profiling, pass_early_tree_profile): Kill.
4851 * tree-cfg.c (update_modified_stmts): Do not update when operands are
4853 * passes.c (init_optimizations_passes): Reorder so we go into SSA
4854 during early_local_passes.
4855 * Makefile.in (cgraph.o): Add dependency on tree-flow.h.
4858 2007-01-02 Carlos O'Donell <carlos@codesourcery.com>
4860 * Makefile.in: Update copyright year.
4862 2007-01-02 Carlos O'Donell <carlos@codesourcery.com>
4864 * Makefile.in: Export GCC_EXEC_PREFIX before calling $(RUNTEST)
4865 in $(lang_checks) and check-consistency targets.
4867 2007-01-02 Jan Hubicka <jh@suse.cz>
4869 * tree-mudflap.c (mf_decl_cache_locals, mf_build_check_statement_for):
4870 Do not add referenced vars.
4871 * tree-cfg.c (update_modified_stmts): Do not update when SSA operands
4873 * passes.c (init_optimization_passes): Put mudflap_2 after
4874 free_datastructures.
4876 2007-01-02 Jan Hubicka <jh@suse.cz>
4878 * tree-optimize (execute_fixup_cfg): Set after_inlining flag.
4879 Set NOTHROW flag on call statements proved to be nothrow.
4880 Update statement of local calls so new pure/const functions are
4881 updated. Update_ssa when in ssa form. Mark PHI nodes of nonlocal
4883 (tree_rest_of_compilation): Register hooks and initialize bitmap
4884 early. Do not set after_inlining flag.
4886 2007-01-02 Steve Ellcey <sje@cup.hp.com>
4888 * sbitmap.c (HOST_BITS_PER_LONG_LONG): Change to
4889 HOST_BITS_PER_LONGLONG
4891 2007-01-02 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
4894 * c-typeck.c (store_init_value): Don't emit pedantic overflow
4895 warning for non-static initializers.
4897 2007-01-02 Steven Bosscher <steven@gcc.gnu.org>
4899 * config/alpha/alpha.md, arm/arm.c, darwin.c, frv/frv.md,
4900 m32r/m32r.c, m32r/m32r.c, mn10300/mn10300.md, pa/pa.c,
4901 rs6000/rs6000.c, s390/s390.md, sh/sh.md, sparc/sparc.c:
4902 Always use set_unique_reg_note to add REG_EQUAL notes.
4904 2007-01-02 Kazu Hirata <kazu@codesourcery.com>
4907 2007-01-02 Kazu Hirata <kazu@codesourcery.com>
4909 * alias.c (init_alias_analysis): Use VEC_safe_grow_cleared.
4910 * cfgbuild.c (find_basic_blocks): Likewise.
4911 * cfgrtl.c (rtl_create_basic_block): Likewise.
4912 * function.c (temp_slots_at_level): Likewise.
4913 * reg-stack.c (stack_regs_mentioned): Likewise.
4914 * regclass.c (allocate_reg_info): Likewise.
4915 * tree-cfg.c (init_empty_tree_cfg, build_tree_cfg, create_bb,
4916 set_bb_for_stmt, move_block_to_fn): Likewise.
4917 * tree-complex.c (tree_lower_complex): Likewise.
4918 * vec.h (VEC_safe_grow_cleared): New.
4920 2007-01-02 Ian Lance Taylor <iant@google.com>
4922 * c-common.c (c_common_truthvalue_conversion): When warning about
4923 using an assignment as a truth value, set TREE_NO_WARNING.
4925 2007-01-02 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
4928 * c.opt (Wold-style-declaration): New.
4929 * doc/invoke.texi (C-only Warning Options): New.
4930 (Wold-style-declaration): Document it.
4931 (Wextra): Enabled by -Wextra.
4932 * c-opts.c (c_common_post_options): Enabled by -Wextra.
4933 * c-decl.c (declspecs_add_scspec): Replace -Wextra with
4934 -Wold-style-declaration.
4936 2007-01-02 Kazu Hirata <kazu@codesourcery.com>
4938 * alias.c (init_alias_analysis): Use VEC_safe_grow_cleared.
4939 * cfgbuild.c (find_basic_blocks): Likewise.
4940 * cfgrtl.c (rtl_create_basic_block): Likewise.
4941 * function.c (temp_slots_at_level): Likewise.
4942 * reg-stack.c (stack_regs_mentioned): Likewise.
4943 * regclass.c (allocate_reg_info): Likewise.
4944 * tree-cfg.c (init_empty_tree_cfg, build_tree_cfg, create_bb,
4945 set_bb_for_stmt, move_block_to_fn): Likewise.
4946 * tree-complex.c (tree_lower_complex): Likewise.
4947 * vec.h (VEC_safe_grow_cleared): New.
4949 2007-01-02 Douglas Gregor <doug.gregor@gmail.com>
4951 * c-common.c (c_common_nodes_and_builtins): Since variants of
4952 void_type_node get built before it is given a name, we need to
4953 give those variants the name, too.
4954 (complete_array_type): We need to work with the canonical main
4955 type of the array, from which we will build the qualified version.
4956 * params.def (PARAM_VERIFY_CANONICAL_TYPES): New.
4957 * print-tree.c (print_node): Display canonical type information
4959 * stor-layout.c (layout_type): When we don't know the
4960 alignment of a type for which we're building an array, we end up
4961 guessing wrong, so make the type require structural equality.
4962 * tree.c (make_node_stat): When we build a new type, it is its
4964 (build_type_attribute_qual_variant): When building an attribute
4965 variant, its canonical type is the non-attribute variant. However,
4966 if the attributes are target-dependent and they differ, we need to
4967 use structural equality checks for this type.
4968 (build_qualified_type): A qualified type is not equivalent to its
4969 unqualified variant; set the canonical type appropriately.
4970 (build_distinct_type_copy): When building a distinct type from
4971 another type, the new type is its own canonical type.
4972 (build_variant_type_copy): When building a new type variant, we
4973 assume that it is equivalent to the original type.
4974 (build_pointer_type_for_mode): When building a pointer type, also
4975 build a canonical type pointer.
4976 (build_reference_type_for_mode): When building a reference type,
4977 also build a canonical type reference.
4978 (build_index_type): When we can't hash an index type (e.g.,
4979 because its maximum value is negative), the index type requires
4980 structural equality tests.
4981 (build_array_type): Build the canonical form of an array type.
4982 (build_function_type): Function types require structural equality,
4983 because they contain default arguments, attributes, etc.
4984 (build_method_type_directly): Ditto for method types.
4985 (build_offset_type): Build the canonical offset type.
4986 (build_complex_type): Build the canonical vector type.
4987 (make_vector_type): Build the canonical vector type.
4988 * tree.h (TYPE_CANONICAL): New.
4989 (TYPE_STRUCTURAL_EQUALITY_P): New.
4990 (SET_TYPE_STRUCTURAL_EQUALITY): New.
4991 (struct tree_type): Added "canonical" field.
4992 * params.h (VERIFY_CANONICAL_TYPES): New.
4993 * doc/c-tree.texi (TYPE_CANONICAL): Document.
4994 (TYPE_STRUCTURAL_EQUALITY_P): Document.
4995 (SET_TYPE_STRUCTURAL_EQUALITY): Document.
4996 * doc/invoke.texi (verify-canonical-types): Document --param
4997 parameter for verifying canonical types.
4999 2007-01-02 Joseph Myers <joseph@codesourcery.com>
5001 * config.gcc (powerpc-*-eabispe*, powerpc-*-eabisimaltivec*,
5002 powerpc-*-eabisim*, powerpc-*-eabialtivec*, powerpc-*-eabi*,
5003 powerpc-*-rtems*, powerpc-wrs-vxworks, powerpc-wrs-vxworksae,
5004 powerpcle-*-eabisim*, powerpcle-*-eabi*): Add rs6000/e500.h to
5006 * config/rs6000/e500.h: New.
5007 * config/rs6000/eabi.h (TARGET_SPE_ABI, TARGET_SPE, TARGET_E500,
5008 TARGET_ISEL, TARGET_FPRS, TARGET_E500_SINGLE, TARGET_E500_DOUBLE):
5010 * config/rs6000/linuxspe.h (TARGET_SPE_ABI, TARGET_SPE,
5011 TARGET_E500, TARGET_ISEL, TARGET_FPRS, TARGET_E500_SINGLE,
5012 TARGET_E500_DOUBLE): Remove.
5013 * config/rs6000/vxworks.h (TARGET_SPE_ABI, TARGET_SPE,
5014 TARGET_E500, TARGET_ISEL, TARGET_FPRS): Remove.
5015 * config/rs6000/rs6000.h (CHECK_E500_OPTIONS): Define.
5016 * config/rs6000/rs6000.c (rs6000_override_options): Use
5019 2007-01-02 Joseph Myers <joseph@codesourcery.com>
5021 * config/rs6000/rs6000.c (print_operand): Check (TARGET_SPE ||
5022 TARGET_E500_DOUBLE), not TARGET_E500, for %y.
5023 (rs6000_generate_compare, rs6000_emit_sCOND, output_cbranch,
5024 rs6000_emit_cmove): Don't check TARGET_E500.
5025 * config/rs6000/rs6000.md (bunordered, bordered, sunordered,
5026 sordered): Don't check TARGET_E500.
5028 2007-01-01 Eric Christopher <echristo@apple.com>
5030 * config/mips/mips.c (mips_regno_mode_ok_for_base_p): Use
5031 HARD_REGISTER_NUM_P.
5033 2007-01-01 Roger Sayle <roger@eyesopen.com>
5035 * fold-const.c (fold_binary) <EQ_EXPR>: Fold "(X^C1) eq/ne C2" into
5036 "X eq/ne (C1^C2)". Fold "(X^Z) eq/ne (Y^Z)" as "X eq/ne Y" when Z
5037 has no side-effects. Fold "(X^C1) eq/ne (Y^C2)" as "(X^(C1^C2))
5040 2007-01-01 Mike Stump <mrs@apple.com>
5042 * configure.ac: Remove support for building with Apple's gcc-3.1.
5044 2007-01-02 Joseph Myers <joseph@codesourcery.com>
5047 * caller-save.c (add_stored_regs): Only handle SUBREGs if inner
5048 REG is a hard register. Do not modify REG before calling
5050 * rtlanal.c (subreg_get_info): Don't assert size of XMODE is a
5051 multiple of the size of YMODE for certain lowpart cases.
5053 2007-01-01 Andrew Pinski <pinskia@gmail.com>
5056 * gimplify (voidify_wrapper_expr): Update for
5059 2007-01-01 Andreas Schwab <schwab@suse.de>
5062 * config/ia64/ia64.c (ia64_compute_frame_size): Account space for
5063 save of BR0 in extra_spill_size instead of spill_size.
5064 (ia64_expand_prologue): Save BR0 outside of the gr/br/fr spill
5066 (ia64_expand_epilogue): Restore BR0 from its new location.
5068 2007-01-01 Andrew Pinski <pinskia@gmail.com>
5070 * gimplify.c (gimplify_init_constructor <case VECTOR_TYPE>):
5071 Use a temporary variable if the left hand side is not a gimple
5074 2007-01-01 Andrew Pinski <pinskia@gmail.com>
5076 * gimplify.c (gimplify_return_expr): Make the temporary variable
5077 for the return expression, a gimple register variable.
5079 2007-01-01 Jan Hubicka <jh@suse.cz>
5081 * emit-rtl.c (emit_copy_of_insn_after): Do not call copy_insn_1 for
5084 2007-01-01 Mike Stump <mrs@apple.com>
5086 * configure.ac (HAVE_GAS_LITERAL16): Add autoconf check for
5088 * config/darwin.c (machopic_select_rtx_section): Use
5090 (darwin_mergeable_constant_section): Likewise.
5091 * configure: Regenerate.
5092 * config.in: Regenerate.
5094 2007-01-01 Jan Hubicka <jh@suse.cz>
5095 Andrew Pinski <pinskia@gmail.com>
5097 * cgraphunit.c (cgraph_optimize): Call cgraph_add_new_functions
5098 before starting IPA passes.