frv.md (reload_incc, [...]): Remove invalid patterns.
[gcc.git] / gcc / ChangeLog
1 2007-02-15 Alexandre Oliva <aoliva@redhat.com>
2
3 * config/frv/frv.md (reload_incc, reload_outcc, reload_incc_uns,
4 reload_outcc_uns, reload_incc_nz, reload_outcc_nz): Remove
5 invalid patterns.
6
7 2007-02-15 Alexandre Oliva <aoliva@redhat.com>
8
9 * tree-sra.c (instantiate_missing_elements): Canonicalize
10 bit-field types.
11 (sra_build_assignment): New.
12 (generate_copy_inout, generate_element_copy,
13 generate_element_zero, generate_one_element_init): Use it.
14
15 2007-02-15 Alexandre Oliva <aoliva@redhat.com>
16
17 * tree-sra.c (instantiate_missing_elements): Canonicalize
18 bit-field types.
19 (sra_build_assignment): New.
20 (generate_copy_inout, generate_element_copy,
21 generate_element_zero, generate_one_element_init): Use it.
22
23 2007-02-15 Alexandre Oliva <aoliva@redhat.com>
24
25 * dwarf2out.c (dwarf2out_finish): Accept namespaces as context of
26 limbo die nodes.
27
28 2007-02-14 Joseph Myers <joseph@codesourcery.com>
29
30 * emit-rtl.c (set_mem_attributes_minus_bitpos): Treat complex
31 types as aggregates not scalars.
32 * function.c (assign_stack_temp_for_type): Likewise.
33
34 2007-02-14 Roger Sayle <roger@eyesopen.com>
35 Zdenek Dvorak <dvorakz@suse.cz>
36
37 * tree-dump.c (dump_switch_p_1): Require exact match of the option
38 name.
39
40 2007-02-14 Zdenek Dvorak <dvorakz@suse.cz>
41
42 * passes.c (next_pass_1): Clear the next field of the copied
43 pass structure.
44
45 2007-02-14 Richard Henderson <rth@redhat.com>
46
47 * tree-sra.c (early_sra): New.
48 (decl_can_be_decomposed_p): Deny va_list if early_sra.
49 (tree_sra_early, pass_sra_early): New.
50 * tree-pass.h (pass_sra_early): Declare.
51 * passes.c (init_optimization_passes): Use it.
52
53 2007-02-14 Richard Guenther <rguenther@suse.de>
54
55 * flags.h (issue_strict_overflow_warning): Convert to a macro.
56
57 2007-02-14 Dorit Nuzman <dorit@il.ibm.com>
58
59 PR tree-optimization/30771
60 * tree-vect-analyze.c (vect_determine_vectorization_factor): Traverse
61 also phi nodes.
62 (vect_analyze_operations): Induction phis can now be marked as
63 used_in_loop.
64 (vect_mark_stmts_to_be_vectorized): No special treatment for phis.
65 Update documentation accordingly.
66
67 2007-02-14 Nick Clifton <nickc@redhat.com>
68
69 * builtin-types.def (DEF_FUNCTION_TYPE_x): Do not imply that at
70 most 3 arguments are supported.
71 (DEF_FUNCTION_TYPE_VAR_5): Fix typo in its description.
72
73 2007-02-13 Seongbae Park <seongbae.park@gmail.com>
74
75 * bitmap.c (bitmap_and, bitmap_and_compl, bitmap_xor):
76 Ensure dst->current is valid.
77
78 2007-02-13 Paul Brook <paul@codesourcery.com>
79
80 * config.gcc: Add arm*-*-uclinux-*eabi.
81 * config/arm/uclinux-elf.h (TARGET_OS_CPP_BUILTINS): Define.
82 (SUBTARGET_EXTRA_LINK_SPEC): Define.
83 (STARTFILE_SPEC, ENDFILE_SPEC): Remove broken -shared handling.
84 (LINK_GCC_C_SEQUENCE_SPEC): Undef.
85 (LINK_SPEC): Define.
86 (LIB_SPEC): Define.
87 * config/arm/arm.c (arm_override_options): Use r9 as EABI PIC
88 register.
89 * config/arm/uclinux-eabi.h: New file.
90 * config/arm/linux-eabi.h (WCHAR_TYPE): Remove.
91 * config/arm/linux-gas.h (WCHAR_TYPE): Use unsigned long on AAPCS
92 based targets.
93
94 2007-02-13 Ian Lance Taylor <iant@google.com>
95
96 * common.opt: Add Wstrict-overflow and Wstrict-overflow=.
97 * flags.h (warn_strict_overflow): Declare.
98 (enum warn_strict_overflow_code): Define.
99 (issue_strict_overflow_warning): New static inline function.
100 * opts.c (warn_strict_overflow): New variable.
101 (common_handle_option): Handle OPT_Wstrict_overflow and
102 OPT_Wstrict_overflow_.
103 * c-opts.c (c_common_handle_option): Set warn_strict_overflow for
104 OPT_Wall.
105 * fold-const.c: Include intl.h.
106 (fold_deferring_overflow_warnings): New static variable.
107 (fold_deferred_overflow_warning): New static variable.
108 (fold_deferred_overflow_code): New static variable.
109 (fold_defer_overflow_warnings): New function.
110 (fold_undefer_overflow_warnings): New function.
111 (fold_undefer_and_ignore_overflow_warnings): New function.
112 (fold_deferring_overflow_warnings_p): New function.
113 (fold_overflow_warning): New static function.
114 (make_range): Add strict_overflow_p parameter. Change all
115 callers.
116 (extract_muldiv, extract_muldiv_1): Likewise.
117 (fold_unary) [ABS_EXPR]: Check ABS_EXPR before calling
118 tree_expr_nonnegative_p.
119 (fold_negate_expr): Call fold_overflow_warning.
120 (fold_range_test): Likewise.
121 (fold_comparison): Likewise.
122 (fold_binary): Likewise. Call tree_expr_nonnegative_warnv_p
123 instead of tree_expr_nonnegative_p.
124 (tree_expr_nonnegative_warnv_p): Rename from
125 tree_expr_nonnegative_p, add strict_overflow_p parameter.
126 (tree_expr_nonnegative_p): New function.
127 (tree_expr_nonzero_warnv_p): Rename from tree_expr_nonzero_p, add
128 strict_overflow_p parameter.
129 (tree_expr_nonzero_p): New function.
130 * passes.c (verify_interpass_invariants): New static function.
131 (execute_one_pass): Call it.
132 * tree-ssa-loop-niter.c (expand_simple_operations): Ignore fold
133 warnings.
134 (number_of_iterations_exit, loop_niter_by_eval): Likewise.
135 (estimate_numbers_of_iterations): Likewise.
136 (scev_probably_wraps_p): Likewise.
137 * tree-ssa-ccp.c: Include "toplev.h".
138 (evaluate_stmt): Defer fold overflow warnings until we know we are
139 going to optimize.
140 (struct fold_stmt_r_data): Add stmt field.
141 (fold_stmt_r): Defer fold overflow warnings until we know we
142 optimized.
143 (fold_stmt): Initialize stmt field of fold_stmt_r_data.
144 (fold_stmt_inplace): Likewise.
145 * tree-cfgcleanup.c: Include "toplev.h" rather than "errors.h".
146 (cleanup_control_expr_graph): Defer fold overflow warnings until
147 we know we are going to optimize.
148 * tree-cfg.c (fold_cond_expr_cond): Likewise.
149 * tree-ssa-threadedge.c (simplify_control_stmt_condition):
150 Likewise.
151 * tree-vrp.c (vrp_expr_computes_nonnegative): Call
152 tree_expr_nonnegative_warnv_p instead of tree_expr_nonnegative_p.
153 * tree-ssa-loop-manip.c (create_iv): Likewise.
154 * c-typeck.c (build_conditional_expr): Likewise.
155 (build_binary_op): Likewise.
156 * tree-vrp.c (vrp_expr_computes_nonzero): Call
157 tree_expr_nonzero_warnv_p instead of tree_expr_nonzero_p.
158 (extract_range_from_unary_expr): Likewise.
159 * simplify-rtx.c (simplify_const_relational_operation): Warn when
160 assuming that signed overflow does not occur.
161 * c-common.c (pointer_int_sum): Ignore fold overflow warnings.
162 * tree.h (tree_expr_nonnegative_warnv_p): Declare.
163 (fold_defer_overflow_warnings): Declare.
164 (fold_undefer_overflow_warnings): Declare.
165 (fold_undefer_and_ignore_overflow_warnings): Declare.
166 (fold_deferring_overflow_warnings_p): Declare.
167 (tree_expr_nonzero_warnv_p): Declare.
168 * doc/invoke.texi (Option Summary): Add -Wstrict-overflow to list
169 of warning options.
170 (Warning Options): Document -Wstrict-overflow.
171 * Makefile.in (tree-ssa-threadedge.o): Depend on toplev.h.
172 (tree-ssa-ccp.o): Likewise.
173 (tree-cfgcleanup.o): Change errors.h dependency to toplev.h.
174 (fold-const.o): Depend on intl.h.
175
176 2007-02-13 Ian Lance Taylor <iant@google.com>
177
178 PR middle-end/30751
179 * lower-subreg.c (resolve_simple_move): Decompose subregs in
180 addresses.
181
182 2007-02-13 Stuart Hastings <stuart@apple.com>
183
184 * gcc/config/i386/i386.md (fixuns_truncdfhi2): Require SSE2.
185
186 2007-02-13 Richard Henderson <rth@redhat.com>
187
188 * config/alpha/alpha.c (alpha_stdarg_optimize_hook): Strip
189 handled_component_p before looking for the indirect_ref.
190
191 2007-02-13 Richard Henderson <rth@redhat.com>
192
193 * config/i386/i386.md (bswapsi_1): Rename from bswapsi2,
194 remove flags clobber.
195 (bswapsi2): New expander, emit code for !TARGET_BSWAP.
196 (bswaphi_lowpart): New.
197 (bswapdi2): Rename from bswapdi2_rex, remove flags clobber,
198 remove TARGET_BSWAP test. Delete expander of the same name.
199
200 * optabs.c (widen_bswap, expand_doubleword_bswap): New.
201 (expand_unop): Use them.
202
203 2007-02-13 Uros Bizjak <ubizjak@gmail.com>
204
205 * config/i386/i386.md (cmpdi_ccno_1_rex64, *cmpsi_ccno_1,
206 *cmphi_ccno_1, *cmpqi_ccno_1, *movsi_xor, *movstricthi_xor,
207 *movstrictqi_xor, *movdi_xor_rex64, *ashldi3_1_rex64,
208 *ashldi3_cmp_rex64, *ashldi3_cconly_rex64, ashlsi3, *ashlsi3_1_zext,
209 *ashlsi3_cmp, *ashlsi3_cconly, *ashlsi3_cmp_zext, *ashlhi3_1_lea,
210 *ashlhi3_1, *ashlhi3_cmp, *ashlhi3_cconly, *ashlqi3_1_lea,
211 *ashlqi3_1, *ashlqi3_cmp, *ashlqi3_cconly): Remove equivalent
212 assembler dialect choice from asm templates.
213
214 2007-02-12 Richard Henderson <rth@redhat.com>
215
216 * config/i386/i386.md (fixuns_trunc<SSEMODEF>si_1): New insn.
217 (fixuns_trunc<SSEMODEF>si2): Use it.
218 * config/i386/sse.md (vec_setv4sf_0): Export.
219 * config/i386/i386.c (ix86_build_const_vector): Export.
220 (ix86_split_convert_uns_si_sse): Rename from
221 ix86_expand_convert_uns_si_sse and rewrite as a splitter.
222 * config/i386/i386-protos.h: Update.
223
224 2007-02-13 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
225
226 PR c/29521
227 * c-typeck.c (c_finish_return): Improve warning message.
228
229 2007-02-12 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
230
231 * alias.c (find_symbolic_term): Delete unused function.
232
233 2007-02-12 Uros Bizjak <ubizjak@gmail.com>
234
235 * config/i386/i386.md (paritydi2, paritysi2): New expanders.
236 (paritydi2_cmp, paritydi2_cmp): New insn and split patterns.
237 (*parityhi2_cmp, *parityqi2_cmp): New insn patterns.
238
239 2007-02-12 Eric Botcazou <ebotcazou@adacore.com>
240
241 * tree.h (DECL_IGNORED_P): Document further effect for FUNCTION_DECL.
242 * cgraphunit.c (cgraph_expand_function): If DECL_IGNORED_P is set on
243 the function, temporarily point the debug interface to the null one.
244
245 2007-02-12 Eric Botcazou <ebotcazou@adacore.com>
246
247 * dwarf2out.c (round_up_to_align): New static function.
248 (field_byte_offset): Use it to round the offset.
249
250 2007-02-12 Richard Henderson <rth@redhat.com>
251
252 * config/alpha/alpha.md (bswapsi2, bswapdi2): New.
253 (inswl_const): Export.
254
255 2007-02-12 Richard Henderson <rth@redhat.com>
256
257 * calls.c (emit_library_call_value_1): If PROMOTE_MODE modifed the
258 result mode of the libcall, convert back to outmode.
259
260 2007-02-12 Roger Sayle <roger@eyesopen.com>
261
262 * config/i386/i386.md (*bswapdi2_rex): Renamed from bswapdi2.
263 (bswapdi2): New define_expand to implement 32-bit implementation.
264
265 2007-02-12 Nick Clifton <nickc@redhat.com>
266
267 * doc/invoke.texi (Overall Options): Document --help=.
268 * gcc.c (target_help_flag): Rename to print_subprocess_flag.
269 (cc1_options): Pass --help= on to cc1.
270 (display_help): Add description of --help=.
271 (process_command): Add code to handle --help=. Allow translated
272 --help and --target-help switches to be passed on to compiler
273 sub-process.
274 (main): Remove unused if statement.
275 * opts.c (columns): Remove.
276 (LEFT_COLUMN): Define.
277 (wrap_help): Add columns argument.
278 (print_filtered_help): Change parameters to be an include bitmask,
279 an exclude bitmask, an any bitmask and the column width. Move the
280 code to display the params list here. Add code to display the
281 status of options rather than their descriptions if the quiet flag
282 is not active.
283 (print_specific_help): Change parameters to be an include bitmask,
284 an exclude bitmask and an any bitmask. Move code to look up the
285 column width here. Decide upon the title for an options listing.
286 (common_handle_options): Add code to handle --help=. Adapt code
287 for --help and --target-help to use the revised form of the
288 print_specific_help function.
289 (print_help): Delete.
290 (print_param_help): Delete.
291 (print_switch): Delete.
292 * opts.h (cl_lang_count): Add prototype.
293 (CL_PARAMS, CL_WARNING, CL_OPTIMIZATION, CL_MIN_OPTION_CLASS,
294 CL_MAX_OPTION_CLASS): New defines.
295 * optc-gen.awk: Add construction of cl_lang_count.
296 * c.opt: Add Warning attribute to warning options and Optimization
297 attribute to optimization options.
298 * common.opt: Likewise.
299 Add --help=.
300 Add -fhelp and -ftarget-help as aliases for the transformed --help
301 and --target-help options.
302 * opt-functions.awk: Add code to handle Warning and Optimization
303 attributes.
304
305 2007-02-12 Richard Henderson <rth@redhat.com>
306
307 * config/alpha/constraints.md: New file.
308 * config/alpha/alpha.c: Include tm-constrs.h.
309 (alpha_const_ok_for_letter_p, alpha_const_double_ok_for_letter_p,
310 alpha_extra_constraint): Remove.
311 (alpha_emit_conditional_branch): Use satisfies_constraint_*.
312 * config/alpha/alpha-protos.h: Update.
313 * config/alpha/alpha.h (REG_CLASS_FROM_LETTER): Remove.
314 (CONST_OK_FOR_LETTER_P, CONST_DOUBLE_OK_FOR_LETTER_P): Remove.
315 (EXTRA_CONSTRAINT): Remove.
316 * config/alpha/alpha.md: Include constraints.md.
317 (adddi splitter): Use satisfies_constraint_*.
318 * config/alpha/predicates.md (add_operand): Likewise.
319 (sext_add_operand, addition_operation): Likewise.
320
321 2007-02-12 Dorit Nuzman <dorit@il.ibm.com>
322
323 PR tree-optimization/29145
324 * tree-data-ref.c (base_addr_differ_p): Make us more conservative
325 in our handling of restrict qualified pointers.
326
327 2007-02-12 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
328
329 PR middle-end/7651
330 * doc/invoke.texi (Wunused-value): Update description.
331 (Wextra): Delete item.
332 * opts.c (set_Wextra): Don't use the value of Wextra to set the
333 value of Wunused-value.
334 * c-typeck.c (c_process_expr_stmt): Don't check extra_warnings.
335 (c_finish_stmt_expr): Don't check extra_warnings.
336 (emit_side_effect_warnings): The caller is responsible to check
337 warn_unused_value.
338
339 2007-02-11 Roger Sayle <roger@eyesopen.com>
340 Matt Thomas <matt@3am-software.com>
341
342 * simplify-rtx.c (simplify_relational_operation_1): Correct typo.
343
344 2007-02-11 Roger Sayle <roger@eyesopen.com>
345
346 * simplify-rtx.c (simplify_relational_operation_1): Optimize
347 comparisons of POPCOUNT against zero.
348 (simplify_const_relational_operation): Likewise.
349
350 2007-02-11 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
351
352 * doc/invoke.texi (Wextra): Delete outdated paragraph.
353
354 2007-02-11 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
355
356 * dwarf2out.c (root_type): Delete unused function.
357
358 2007-02-11 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
359
360 * genattrtab.c (contained_in_p): Delete unused function.
361 (write_expr_attr_cache): Likewise.
362
363 2007-02-11 Jan Hubicka <jh@suse.cz>
364
365 * ipa-inline.c (cgraph_edge_badness): Add "else" missing in the
366 previous patch.
367
368 2007-02-11 Steven Bosscher <steven@gcc.gnu.org>
369
370 * fwprop.c (try_fwprop_subst): Use set_unique_reg_note
371 to add the REG_EQ* note.
372 * see.c (see_merge_one_use_extension): Likewise.
373 * local-alloc.c (update_equiv_regs): Likewise. Also don't
374 turn REG_EQUAL notes into REG_EQUIV notes if the target
375 register may have more than one set.
376 * function.c (assign_parm_setup_reg): Use set_unique_reg_note.
377 * gcse.c (try_replace_reg): Likewise.
378 * alias.c (init_alias_analysis): Use find_reg_equal_equiv_note.
379 * calls.c (fixup_tail_calls): Likewise. Abort if there is
380 more than one REG_EQUIV note.
381 * reload1.c (gen_reload): Use set_unique_reg_note.
382
383 2007-02-11 Uros Bizjak <ubizjak@gmail.com>
384
385 * config/i386/i386.c (TARGET_VECTORIZE_BUILTIN_CONVERSION): Define.
386 (ix86_builtin_conversion): New function.
387
388 2007-02-06 Mark Mitchell <mark@codesourcery.com>
389
390 PR target/29487
391 * tree.h (DECL_REPLACEABLE_P): New macro.
392 * except.c (set_nothrow_function_flags): Likewise.
393
394 2007-02-11 Tehila Meyzels <tehila@il.ibm.com>
395 Ira Rosen <irar@il.ibm.com>
396 Dorit Nuzman <dorit@il.ibm.com>
397
398 * doc/tm.texi (TARGET_VECTORIZE_BUILTIN_CONVERSION): New target hook.
399 * targhooks.c (default_builtin_vectorized_conversion): New.
400 * targhooks.h (default_builtin_vectorized_function): New declaration.
401 * target.h (struct vectorize): Add builtin_conversion field.
402 * tree-vectorizer.h (type_conversion_vec_info_type): New enum
403 stmt_vec_info_type value.
404 (vectorizable_conversion): New declaration.
405 * tree-vect-analyze.c (vect_analyze_operations): Add
406 vectorizable_conversion call.
407 * target-def.h (TARGET_VECTORIZE_BUILTIN_CONVERSION): New.
408 * tree-vect-transform.c (vectorizable_conversion): New function.
409 (vect_transform_stmt): Add case for type_conversion_vec_info_type.
410 * tree-vect-generic.c (expand_vector_operations_1): Consider correct
411 mode.
412 * config/rs6000/rs6000.c (rs6000_builtin_conversion): New.
413 (TARGET_VECTORIZE_BUILTIN_CONVERSION): Defined.
414 (rs6000_expand_builtin): Add handling a case of ALTIVEC_BUILTIN_VCFUX
415 or ALTIVEC_BUILTIN_VCFSX.
416
417 2007-02-10 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
418
419 PR target/30634
420 * pa.md (movdf): For 64-bit target, fail if operand 1 is a non-zero
421 CONST_DOUBLE and operand 0 is a hard register.
422 (movdi): For 64-bit target, remove code to force CONST_DOUBLE to
423 memory. Fail if operand 1 is a non-zero CONST_INT and operand 0
424 is a hard floating-point register.
425
426 2007-02-10 Richard Henderson <rth@redhat.com>
427 Jakub Jelinek <jakub@redhat.com>
428 Alexandre Oliva <aoliva@redhat.com>
429
430 * Makefile.in (libgcc-support, libgcc.mvars): Add emutls.c.
431 * builtin-types.def (BT_WORD): Make unsigned.
432 (BT_FN_VOID_PTR_WORD_WORD_PTR): New.
433 * builtins.def (BUILT_IN_EMUTLS_GET_ADDRESS): New.
434 (BUILT_IN_EMUTLS_REGISTER_COMMON): New.
435 * c-decl.c (grokdeclarator): Don't error if !have_tls.
436 * c-parser.c (c_parser_omp_threadprivate): Likewise.
437 * dwarf2out.c (loc_descriptor_from_tree_1): Don't do anything for
438 emulated tls.
439 * expr.c (emutls_var_address): New.
440 (expand_expr_real_1): Expand emulated tls.
441 (expand_expr_addr_expr_1): Likewise.
442 * libgcc-std.ver: Add __emutls_get_address, __emutls_register_common.
443 * output.h (emutls_finish): Declare.
444 * toplev.c (compile_file): Call it.
445 * tree-ssa-address.c (gen_addr_rtx): Check for const-ness of the
446 address before wrapping in CONST.
447 * varasm.c (emutls_htab, emutls_object_type): New.
448 (EMUTLS_VAR_PREFIX, EMUTLS_TMPL_PREFIX): New.
449 (get_emutls_object_name, get_emutls_object_type): New.
450 (get_emutls_init_templ_addr, emutls_decl): New.
451 (emutls_common_1, emutls_finish): New.
452 (assemble_variable): When emulating tls, swap decls; generate
453 constructor for the emutls objects.
454 (do_assemble_alias): When emulating tls, swap decl and target name.
455 (default_encode_section_info): Don't add SYMBOL_FLAG_TLS_SHIFT
456 for emulated tls.
457 * varpool.c (decide_is_variable_needed): Look at force_output.
458 Recurse for emulated tls.
459 (cgraph_varpool_remove_unreferenced_decls): Remove checks redundant
460 with decide_is_variable_needed.
461 * emutls.c: New file.
462 * config/sparc/sol2.h (ASM_DECLARE_OBJECT_NAME): Only emit
463 tls_object for real tls.
464
465 2007-02-10 Kaz Kojima <kkojima@gcc.gnu.org>
466
467 PR rtl-optimization/29599
468 * reload1.c (eliminate_regs_in_insn): Take the destination
469 mode into account when computing the offset.
470
471 2007-02-09 Stuart Hastings <stuart@apple.com>
472 Richard Henderson <rth@redhat.com>
473
474 * gcc/config/i386/i386.h (TARGET_KEEPS_VECTOR_ALIGNED_STACK): New.
475 * gcc/config/i386/darwin.h: (TARGET_KEEPS_VECTOR_ALIGNED_STACK): New.
476 * gcc/config/i386/i386.md (fixuns_trunc<mode>si2, fixuns_truncsfhi2,
477 fixuns_truncdfhi2): New.
478 (fix_truncsfdi_sse): Call ix86_expand_convert_sign_didf_sse.
479 (floatunsdidf2): Call ix86_expand_convert_uns_didf_sse.
480 (floatunssisf2): Add call to ix86_expand_convert_uns_sisf_sse.
481 (floatunssidf2): Allow nonimmediate source.
482 * gcc/config/i386/sse.md (movdi_to_sse): New.
483 (vec_concatv2di): Drop '*'.
484 * gcc/config/i386/i386-protos.h (ix86_expand_convert_uns_si_sse,
485 ix86_expand_convert_uns_didf_sse, ix86_expand_convert_uns_sidf_sse,
486 ix86_expand_convert_uns_sisf_sse, ix86_expand_convert_sign_didf_sse):
487 New.
488 * gcc/config/i386/i386.c (ix86_expand_convert_uns_si_sse,
489 ix86_expand_convert_uns_didf_sse, ix86_expand_convert_uns_sidf_sse,
490 ix86_expand_convert_uns_sisf_sse, ix86_expand_convert_sign_didf_sse,
491 ix86_build_const_vector, ix86_expand_vector_init_one_nonzero): New.
492 (ix86_build_signbit_mask): Fix decl of v, refactor to call
493 ix86_build_const_vector.
494 (x86_emit_floatuns): Rewrite.
495
496 2007-02-10 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
497
498 * genautomata.c (longest_path_length): Delete unused function.
499 (struct state): Delete unused longest_path_length.
500 (UNDEFINED_LONGEST_PATH_LENGTH): Delete unused macro.
501 (get_free_state): Delete unused.
502
503 2007-02-09 Jan Hubicka <jh@suse.cz>
504
505 * params.def (PARAM_INLINE_UNIT_GROWTH): Set to 30.
506 * doc/invoke.texi (inline-unit-growth): Update default value.
507
508 * Makefile.in (passes.o, ipa-inline.o): Add dependencies.
509 * cgraphbuild.c (build_cgraph_edges): Compute frequencies.
510 (rebuild_cgraph_edges): Likewise.
511 * cgraph.c (cgraph_set_call_stmt): Add new argument frequency.
512 (dump_cgraph_node): Dump frequencies.
513 (cgraph_clone_edge): Add frequency scales.
514 (cgraph_clone_node): Add freuqnecy.
515 * cgraph.h (cgraph_edge): Add freuqnecy argument.
516 (CGRAPH_FREQ_BASE, CGRAPH_FREQ_MAX): New constants.
517 (cgraph_create_edge, cgraph_clone_edge, cgraph_clone_node): Update.
518 * tree-pass.h (TODO_rebuild_frequencies): New constant.
519 * cgraphunit.c (verify_cgraph_node): Verify frequencies.
520 (cgraph_copy_node_for_versioning): Update call of cgraph_clone_edge.
521 (save_inline_function_body): Likewise.
522 * ipa-inline.c: inluce rtl.h
523 (cgraph_clone_inlined_nods): Update call of cgraph_clone_node.
524 (cgraph_edge_badness): Use frequencies.
525 (cgraph_decide_recursive_inlining): Update clonning.
526 (cgraph_decide_inlining_of_small_function): Dump frequency.
527 * predict.c (estimate_bb_frequencies): Export.
528 * predict.h (estimate_bb_frequencies): Declare.
529 * tree-inline.c (copy_bb): Watch overflows.
530 (expand_call_inline): Update call of cgraph_create_edge.
531 (optimize_inline_calls): Use TODO flags to update frequnecies.
532 * passes.h: Include predict.h
533 (init_optimization_passes): Move profile ahead.
534 (execute_function_todo): Handle TODO_rebuild_frequencies.
535
536 2007-02-09 Roger Sayle <roger@eyesopen.com>
537
538 * config/alpha/alpha.c (emit_insxl): Force the first operand of
539 the insbl or inswl pattern into a register.
540
541 2007-02-09 Roger Sayle <roger@eyesopen.com>
542
543 * config/ia64/ia64.md (bswapdi2): New define_insn.
544
545 2007-02-09 Richard Henderson <rth@redhat.com>
546
547 * config/i386/constraints.md (Ym): New constraint.
548 * config/i386/i386.md (movsi_1): Change Y2 to Yi constraints.
549 (movdi_1_rex64): Split sse and xmm general register moves from
550 memory move alternatives. Use conditional register constraints.
551 (movsf_1, movdf_integer): Likewise.
552 (zero_extendsidi2_32, zero_extendsidi2_rex64): Likewise.
553 (movdf_integer_rex64): New.
554 (pushsf_rex64): Fix output constraints.
555 * config/i386/sse.md (sse2_loadld): Split rm alternative, use Yi.
556 (sse2_stored): Likewise.
557 (sse2_storeq_rex64): New.
558 * config/i386/i386.c (x86_inter_unit_moves): Enable for not
559 amd and not generic.
560 (ix86_secondary_memory_needed): Don't bypass TARGET_INTER_UNIT_MOVES
561 for optimize_size. Remove SF/DFmode hack.
562
563 2007-02-09 Dwarakanath Rajagopal <dwarak.rajagopal@amd.com>
564
565 * config/i386/driver-i386.c: Turn on -mtune=native for AMDFAM10.
566 (bit_SSE4a): New.
567
568 2007-02-09 Nathan Sidwell <nathan@codesourcery.com>
569 Richard Sandiford <richard@codesourcery.com>
570
571 * config.gcc (m68010-*-netbsdelf*, m68k*-*-netbsdelf*)
572 (m68k*-*-openbsd*, m68k-*-linux*): Set default_cf_cpu.
573 (m68k-*-aout*, m68k-*-coff*, m68k-*-uclinux*, m68k-*-rtems*): Add
574 m68k/t-mlib to tmake_file.
575 (m68020-*-elf*, m68k-*-elf*): Likewise. Add t-m68kbare as well.
576 (m68k*-*-*): Use --with-arch to pick a default for --with-cpu.
577 (m680[012]0-*-*, m68k*-*-*): Add support for --with-arch.
578 Allow it to be cf or m68k. Set m68k_arch_family. If that
579 variable is not empty, add t-$m68k_arch_family to tmake_file.
580 Add t-mlibs to tmake_file.
581 * doc/install.texi: Document --with-arch=m68k and --with-arch=cf.
582 * config/m68k/t-cf: New file.
583 * config/m68k/t-m68k: Likewise.
584 * config/m68k/t-mlibs: Likewise.
585 * config/m68k/t-m68kbare (MULTILIB_OPTIONS, MULTILIB_DIRNAMES)
586 (MULTILIB_MATCHES, MULTILIB_EXCEPTIONS): Delete.
587 (M68K_MLIB_DIRNAMES, M68K_MLIB_OPTIONS): Define.
588 * config/m68k/t-m68kelf (MULTILIB_OPTIONS, MULTILIB_DIRNAMES)
589 (MULTILIB_MATCHES, MULTILIB_EXCEPTIONS, LIBGCC, INSTALL_LIBGCC):
590 Delete.
591 * config/m68k/t-openbsd (MULTILIB_OPTIONS, LIBGCC): Delete.
592 (INSTALL_LIBGCC): Delete.
593 (M68K_MLIB_DIRNAMES, M68K_MLIB_OPTIONS): Define.
594 * config/m68k/t-rtems (MULTILIB_OPTIONS, MULTILIB_DIRNAMES)
595 (MULTILIB_MATCHES, MULTILIB_EXCEPTIONS): Delete.
596 (M68K_MLIB_CPU): Define.
597 * config/m68k/t-uclinux (MULTILIB_OPTIONS, MULTILIB_DIRNAMES)
598 (MULTILIB_MATCHES, MULTILIB_EXCEPTIONS): Delete.
599 (M68K_MLIB_CPU, M68K_MLIB_OPTIONS, M68K_MLIB_DIRNAMES): Define.
600
601 2007-02-09 Zdenek Dvorak <dvorakz@suse.cz>
602 Richard Guenther <rguenther@suse.de>
603
604 PR middle-end/23361
605 * fold-const.c (fold_comparison): Handle obfuscated comparisons
606 against INT_MIN/INT_MAX.
607 * tree-ssa-loop-ivcanon.c (remove_empty_loop): Print to dump
608 file if a loop is removed.
609
610 2007-02-09 Joseph Myers <joseph@codesourcery.com>
611
612 * calls.c (store_one_arg): Pass correct alignment to
613 emit_push_insn for non-BLKmode values.
614 * expr.c (emit_push_insn): If STRICT_ALIGNMENT, copy to an
615 unaligned stack slot via a suitably aligned slot.
616
617 2007-02-08 DJ Delorie <dj@redhat.com>
618
619 * config/m32c/m32c.c (m32c_unpend_compare): Add default to silence
620 warnings.
621 (legal_subregs): Use unsigned char, make const.
622 (m32c_illegal_subreg_p): Use ARRAY_SIZE. Delete unused variables.
623
624 2007-02-08 Paul Brook <paul@codesourcery.com>
625
626 * config/arm/lib1funcs.asm (RETLDM): Pop directly into PC when no
627 special interworking needed.
628
629 2007-02-08 Harsha Jagasia <harsha.jagasia@amd.com>
630
631 * config/i386/xmmintrin.h: Make inclusion of emmintrin.h
632 conditional to __SSE2__.
633 (Entries below should have been added to first ChangeLog
634 entry for amdfam10 dated 2007-02-05)
635 * config/i386/emmintrin.h: Generate #error if __SSE2__ is not
636 defined.
637 * config/i386/pmmintrin.h: Generate #error if __SSE3__ is not
638 defined.
639 * config/i386/tmmintrin.h: Generate #error if __SSSE3__ is not
640 defined.
641
642 2007-02-08 DJ Delorie <dj@redhat.com>
643
644 * config/m32c/m32c-protos.h (m32c_illegal_subreg_p): New.
645 * config/m32c/m32c.c (legal_subregs): New.
646 (m32c_illegal_subreg_p): New.
647 * config/m32c/predicates.md (m32c_any_operand): Use it to reject
648 unsupported subregs of hard regs.
649
650 2007-02-08 Jan Hubicka <jh@suse.cz>
651
652 * tree-cfg.c (bsi_replace): Shortcut when replacing the statement with
653 the same one; always update histograms.
654
655 2007-02-08 Diego Novillo <dnovillo@redhat.com>
656
657 * passes.c (init_optimization_passes): Tidy comment.
658
659 2007-02-08 Roger Sayle <roger@eyesopen.com>
660
661 * simplify-rtx.c (simplify_unary_operation_1) <POPCOUNT>: We can
662 strip zero_extend, bswap and rotates from POCOUNT's argument.
663 <PARITY>: Likewise, we can strip not, bswap, sign_extend,
664 zero_extend and rotates from PARITY's argument.
665 <BSWAP>: A byte-swap followed by a byte-swap is an identity.
666 (simplify_const_unary_operation) <BSWAP>: Evaluate the byte-swap
667 of an integer constant at compile-time.
668
669 2007-02-08 Diego Novillo <dnovillo@redhat.com>
670
671 PR 30562
672 * tree-flow.h (struct var_ann_d): Remove field 'is_used'.
673 Update all users.
674 * tree-ssa-alias.c (compute_is_aliased): Remove. Update all
675 users.
676 (init_alias_info):
677 * tree-ssa-live.c (remove_unused_locals): Do not remove
678 TREE_ADDRESSABLE variables.
679 * tree-ssa-structalias.c (compute_points_to_sets): Tidy.
680 * tree-ssa-operands.c (add_virtual_operand): Remove argument
681 FOR_CLOBBER. Update all users.
682 If VAR has an associated alias set, add a virtual operand for
683 it if no alias is found to conflict with the memory reference.
684
685 2007-02-07 Jan Hubicka <jh@suse.cz>
686 Robert Kidd <rkidd@crhc.uiuc.edu>
687
688 * value-prof.c (visit_hist, free_hist): Return 1 instead of 0.
689
690 2007-02-07 Ian Lance Taylor <iant@google.com>
691
692 * lower-subreg.c (simple_move): Reject PARTIAL_INT modes.
693
694 2007-02-07 Roger Sayle <roger@eyesopen.com>
695
696 * config/rs6000/rs6000.md (ctz<mode>2, ffs<mode>2, popcount<mode>2,
697 parity<mode>2, smulsi3_highpart, abstf2_internal, allocate_stack,
698 tablejumpdi, movsi_to_cr_one): Remove constraints from
699 define_expand's match_operands.
700
701 2007-02-07 Roger Sayle <roger@eyesopen.com>
702
703 * global.c (compute_regsets): Move declatation of "i" inside of
704 #ifdef ELIMINABLE_REGS to avoid unused variable bootstrap failure.
705
706 2007-02-07 Jakub Jelinek <jakub@redhat.com>
707
708 PR c++/30703
709 * gimplify.c (gimplify_scan_omp_clauses): Remove special casing
710 of INDIRECT_REF <RESULT_DECL>.
711
712 * config/i386/i386.c (override_options): Set PTA_SSSE3 for core2.
713
714 2007-02-06 J"orn Rennecke <joern.rennecke@arc.com>
715 Kaz Kojima <kkojima@gcc.gnu.org>
716
717 PR target/29746
718 * config/sh/sh.c (expand_cbranchdi4): Use scratch register
719 properly.
720 (sh_initialize_trampoline): Add parentheses to avoid a warning.
721
722 2007-02-06 Zdenek Dvorak <dvorakz@suse.cz>
723
724 * doc/loop.texi: Document possibility not to perform disambiguation
725 of loops with multiple latches.
726 * cfgloopmanip.c (alp_enum_p): Removed.
727 (add_loop): Handle subloops. Use get_loop_body_with_size.
728 (create_preheader): Do not allow ENTRY_BLOCK_PTR to be preheader.
729 * cfghooks.c (redirect_edge_and_branch_force): Set dominator for
730 the new forwarder block.
731 (make_forwarder_block): Only call new_bb_cbk if it is not NULL.
732 Handle the case latch is NULL.
733 * tree-ssa-dom.c (tree_ssa_dominator_optimize): Avoid cfg modifications
734 when marking loop exits.
735 * ifcvt.c (if_convert): Ditto. Mark loop exits even if cfg cannot
736 be modified.
737 * loop-init.c (loop_optimizer_init): Do not modify cfg. Call
738 disambiguate_loops_with_multiple_latches.
739 * tree-cfgcleanup.c (cleanup_tree_cfg_loop): Calculate dominators
740 before fix_loop_structure.
741 * cfgloop.c: Include pointer-set.h and output.h.
742 (canonicalize_loop_headers, HEADER_BLOCK, LATCH_EDGE,
743 update_latch_info, mfb_keep_just, mfb_keep_nonlatch): Removed.
744 (get_loop_latch_edges, find_subloop_latch_edge_by_profile,
745 find_subloop_latch_edge_by_ivs, find_subloop_latch_edge,
746 mfb_redirect_edges_in_set, form_subloop, merge_latch_edges,
747 disambiguate_multiple_latches, get_loop_body_with_size,
748 disambiguate_loops_with_multiple_latches): New functions.
749 (flow_loop_dump): Dump multiple latch edges.
750 (flow_loop_nodes_find): Handle loops with multiple latches.
751 (flow_loops_find): Ditto. Do not call canonicalize_loop_headers.
752 (glb_enum_p): Modified.
753 (get_loop_body): Use get_loop_body_with_size.
754 * cfgloop.h (LOOPS_HAVE_RECORDED_EXITS): New flag.
755 (AVOID_CFG_MODIFICATIONS): New constant.
756 (disambiguate_loops_with_multiple_latches, add_loop,
757 get_loop_body_with_size): Declare.
758 * Makefile.in (cfgloop.o): Add pointer-set.h and output.h.
759
760 2007-02-06 Seongbae Park <seongbae.park@gmail.com>
761
762 PR inline-asm/28686
763 * global.c (compute_regsets): New function.
764 (global_alloc): Refactored ELIMINABLE_REGSET
765 and NO_GLOBAL_ALLOC_REGS computation out.
766 (rest_of_handle_global_alloc): Call compute_regsets()
767 for non-optimizing case.
768
769 2007-02-06 Richard Henderson <rth@redhat.com>
770
771 * config/i386/constraints.md (Y2): Rename from Y.
772 (Yi): New constraint.
773 * config/i386/i386.md (movsi_1, movdi_2, pushdf_nointeger,
774 pushdf_integer, movdf_nointeger, movdf_integer, zero_extendsidi2_32,
775 zero_extendsidi2_rex64, truncxfdf2_mixed): Change Y constraints to Y2.
776 (extendsfdf2_mixed, extendsfdf2_sse, truncdfsf_fast_mixed,
777 truncdfsf_fast_sse, truncdfsf_mixed, fix_truncdfdi_sse,
778 fix_truncdfsi_sse, floatsidf2_mixed, floatsidf2_sse,
779 floatdidf2_mixed, floatdidf2_sse, absnegdf2_mixed,
780 absnegdf2_sse, sse_setccdf, fop_df_comm_mixed, fop_df_comm_sse,
781 fop_df_1_mixed, fop_df_1_sse): Change Y constraints to x.
782 * config/i386/mmx.md (mov<MMXMODEI>_internal_rex64,
783 mov<MMXMODEI>_internal, movv2sf_internal_rex64, movv2sf_internal,
784 vec_extractv2si_1): Change Y constraints to Y2.
785 * config/i386/sse.md (vec_setv4sf_0, vec_concatv2df, vec_dupv4si,
786 vec_dupv2di, sse2_concatv2si, vec_concatv4si_1, vec_concatv2di):
787 Change Y constraints to Y2.
788 (sse2_loadld): Change Y constraints to x.
789
790 2007-02-06 Roger Sayle <roger@eyesopen.com>
791
792 * config/rs6000/rs6000.md (popcount<mode>2): Rewrite.
793 (parity<mode>2): New define_expand using rs6000_emit_parity.
794 * config/rs6000/rs6000.c (rs6000_emit_popcount,
795 rs6000_emit_parity): New functions.
796 * config/rs6000/rs6000-protos.h (rs6000_emit_popcount,
797 rs6000_emit_parity): Prototype here.
798
799 2007-02-06 Ian Lance Taylor <iant@google.com>
800
801 * lower-subreg.c (simple_move_operand): Reject CONST.
802 (resolve_clobber): Call validate_change rather than directly
803 assigning to XEXP (pat, 0).
804
805 2006-02-06 Paolo Bonzini <bonzini@gnu.org>
806
807 * Makefile.in (tree-ssa-loop-ivopts.o): Add pointer-set.h dependency.
808 (tree-ssa-reassoc.o): Add pointer-set.h dependency.
809 (tree-cfg.o): Remove hashtab.h dependency.
810
811 * tree-ssa-loop-ivopts.c: Include pointer-set.h.
812 (struct ivopts_data): Change niters to pointer_map_t.
813 (struct nfe_cache_elt, nfe_hash, nfe_eq): Delete.
814 (niter_for_exit): Create pointer_map on demand. Change for
815 pointer_map API.
816 (tree_ssa_iv_optimize_init): Initialize data->niters to NULL.
817 (free_loop_data): Destroy data->niters if created and reset field.
818 (tree_ssa_iv_optimize_finalize): Don't delete data->niters here.
819 (tree_ssa_iv_optimize_loop): Check for presence of stale data.
820
821 * tree-ssa-reassoc.c: Include pointer-set.h.
822 (bb_rank): Change to long *.
823 (operand_rank): Change to pointer_map_t.
824 (find_operand_rank): Return long, -1 if not found. Declare as inline.
825 (insert_operand_rank): Accept long.
826 (operand_entry_hash, operand_entry_eq): Remove.
827 (get_rank): Return long. Adjust for changes above.
828 (init_reassoc): Change rank type to long. Adjust creation of bb_rank
829 and operand_rank.
830 (fini_reassoc): Delete operand_rank with pointer_map_destroy.
831
832 * tree-ssa-structalias.c (vi_for_tree): Change to pointer_map.
833 (struct tree_vi, tree_vi_t, tree_vi_hash, tree_vi_eq): Delete.
834 (insert_vi_for_tree): Rewrite for pointer_map API. Assert argument
835 is not NULL.
836 (lookup_vi_for_tree): Rewrite for pointer_map API. Return varinfo_t
837 directly since it cannot be NULL.
838 (get_vi_for_tree): Rewrite for pointer_map API.
839 (find_what_p_points_to): Adjust for change to lookup_vi_for_tree.
840 (init_alias_vars): Create vi_for_tree as pointer_map.
841 (delete_points_to_sets): Delete vi_for_tree using pointer_map_destroy.
842
843 * tree-cfg.c: Don't include hashtab.h.
844 (edge_to_cases): Declare as pointer_map.
845 (struct edge_to_cases_elt, edge_to_cases_hash, edge_to_cases_eq):
846 Delete.
847 (edge_to_cases_cleanup): Rewrite as pointer_map_traverse callback.
848 (start_recording_case_labels): Create edge_to_cases as pointer_map.
849 (end_recoding_case_labels): Cleanup edge_to_cases manually before
850 destroying it.
851 (record_switch_edge): Delete.
852 (get_cases_for_edge): Adjust for pointer_map API, inline
853 record_switch_edge (rewritten for new API), remove goto.
854
855 2006-02-06 Paolo Bonzini <bonzini@gnu.org>
856
857 * Makefile.in (tree-nested.o): Add pointer-set.h dependency.
858 * tree-nested.c: Include pointer-set.h.
859 (var_map_elt, var_map_eq, var_map_hash): Delete.
860 (struct nesting_info): Remove GTY marker. Change the two htab_t's
861 to pointer_map_t's.
862 (nesting_info_bitmap_obstack): New.
863 (lookup_field_for_decl): Adjust for pointer_map API.
864 (lookup_tramp_for_decl): Adjust for pointer_map API.
865 (get_nonlocal_debug_decl): Adjust for pointer_map API.
866 (get_local_debug_decl): Adjust for pointer_map API.
867 (convert_nl_goto_reference): Adjust for pointer_map API.
868 (convert_nl_goto_receiver): Adjust for pointer_map API.
869 (create_nesting_tree): Create outside GGC space. Create bitmap on
870 the new obstack. Create field_map and var_map as pointer_maps.
871 (free_nesting_tree): Adjust for changes to create_nesting_tree.
872 (root): Delete.
873 (lower_nested_functions): Move root here, no need to NULL it.
874 Initialize and release the obstack.
875
876 2007-02-06 Paolo Bonzini <bonzini@gnu.org>
877
878 * tree.c (tree_int_map_hash, tree_int_map_eq, tree_int_map_marked_p):
879 Remove prototypes and make them non-static.
880 (struct tree_int_map): Remove.
881 * tree.h (struct tree_int_map): Move here, turning TO into an
882 unsigned int.
883 (tree_int_map_hash, tree_int_map_eq, tree_int_map_marked_p): Declare.
884
885 * tree.h (TREE_COMPLEXITY): Remove.
886 (struct tree_exp): Remove complexity field.
887 * tree.c (build1_stat): Don't set it.
888
889 2007-02-06 Dorit Nuzman <dorit@il.ibm.com>
890 Victor Kaplansky <victork@il.ibm.com>
891
892 * tree-vectorizer.c (vect_is_simple_use): Support induction.
893 (vect_is_simple_reduction): Support reduction with induction as
894 one of the operands.
895 (vect_is_simple_iv_evolution): Fix formatting.
896 * tree-vect-analyze.c (vect_mark_stmts_to_be_vectorized): Fix
897 formatting. Don't mark induction phis for vectorization.
898 (vect_analyze_scalar_cycles): Analyze all inductions, then reductions.
899 * tree-vect-transform.c (get_initial_def_for_induction): New function.
900 (vect_get_vec_def_for_operand): Support induction.
901 (vect_get_vec_def_for_stmt_copy): Fix formatting and add check for
902 induction case.
903 (vectorizable_reduction): Support reduction with induction as one of
904 the operands.
905 (vectorizable_type_demotion): Use def-type of stmt argument rather
906 than dummy def-type.
907
908 * tree-ssa-loop.c (gate_scev_const_prop): Return the value of
909 flag_tree_scev_cprop.
910 * common.opt (tree-scev-cprop): New flag.
911
912 * tree-vect-transform.c (vect_create_destination_var): Use 'kind' in
913 call to vect_get_new_vect_var.
914
915 2007-02-06 Ira Rosen <irar@il.ibm.com>
916
917 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Check that
918 vectype is not NULL.
919 (vect_pattern_recog_1): Likewise.
920
921 2007-02-05 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
922
923 * fold-const.c (negate_expr_p): Handle CONJ_EXPR.
924 (fold_negate_expr): Likewise.
925
926 2007-02-05 Alexandre Oliva <aoliva@redhat.com>
927
928 PR debug/30189
929 * dwarf2out.c (modified_type_die): Follow DECL_ORIGINAL_TYPE
930 even if cv-qualification is the same.
931
932 2007-02-05 Geoffrey Keating <geoffk@apple.com>
933
934 * config/rs6000/darwin-tramp.asm (__trampoline_setup): Call
935 __enable_execute_stack on completion.
936
937 2007-02-05 Harsha Jagasia <harsha.jagasia@amd.com>
938
939 * config/i386/athlon.md (athlon_fldxf_k8, athlon_fld_k8,
940 athlon_fstxf_k8, athlon_fst_k8, athlon_fist, athlon_fmov,
941 athlon_fadd_load, athlon_fadd_load_k8, athlon_fadd, athlon_fmul,
942 athlon_fmul_load, athlon_fmul_load_k8, athlon_fsgn,
943 athlon_fdiv_load, athlon_fdiv_load_k8, athlon_fdiv_k8,
944 athlon_fpspc_load, athlon_fpspc, athlon_fcmov_load,
945 athlon_fcmov_load_k8, athlon_fcmov_k8, athlon_fcomi_load_k8,
946 athlon_fcomi, athlon_fcom_load_k8, athlon_fcom): Added amdfam10.
947
948 2007-02-05 Harsha Jagasia <harsha.jagasia@amd.com>
949
950 * config/i386/i386.md (x86_sahf_1, cmpfp_i_mixed, cmpfp_i_sse,
951 cmpfp_i_i387, cmpfp_iu_mixed, cmpfp_iu_sse, cmpfp_iu_387,
952 swapsi, swaphi_1, swapqi_1, swapdi_rex64, fix_truncsfdi_sse,
953 fix_truncdfdi_sse, fix_truncsfsi_sse, fix_truncdfsi_sse,
954 x86_fldcw_1, floatsisf2_mixed, floatsisf2_sse, floatdisf2_mixed,
955 floatdisf2_sse, floatsidf2_mixed, floatsidf2_sse,
956 floatdidf2_mixed, floatdidf2_sse, muldi3_1_rex64, mulsi3_1,
957 mulsi3_1_zext, mulhi3_1, mulqi3_1, umulqihi3_1, mulqihi3_insn,
958 umulditi3_insn, umulsidi3_insn, mulditi3_insn, mulsidi3_insn,
959 umuldi3_highpart_rex64, umulsi3_highpart_insn,
960 umulsi3_highpart_zext, smuldi3_highpart_rex64,
961 smulsi3_highpart_insn, smulsi3_highpart_zext, x86_64_shld,
962 x86_shld_1, x86_64_shrd, sqrtsf2_mixed, sqrtsf2_sse,
963 sqrtsf2_i387, sqrtdf2_mixed, sqrtdf2_sse, sqrtdf2_i387,
964 sqrtextendsfdf2_i387, sqrtxf2, sqrtextendsfxf2_i387,
965 sqrtextenddfxf2_i387): Added amdfam10_decode.
966
967 * config/i386/athlon.md (athlon_idirect_amdfam10,
968 athlon_ivector_amdfam10, athlon_idirect_load_amdfam10,
969 athlon_ivector_load_amdfam10, athlon_idirect_both_amdfam10,
970 athlon_ivector_both_amdfam10, athlon_idirect_store_amdfam10,
971 athlon_ivector_store_amdfam10): New define_insn_reservation.
972 (athlon_idirect_loadmov, athlon_idirect_movstore): Added
973 amdfam10.
974
975 2007-02-05 Harsha Jagasia <harsha.jagasia@amd.com>
976
977 * config/i386/athlon.md (athlon_call_amdfam10,
978 athlon_pop_amdfam10, athlon_lea_amdfam10): New
979 define_insn_reservation.
980 (athlon_branch, athlon_push, athlon_leave_k8, athlon_imul_k8,
981 athlon_imul_k8_DI, athlon_imul_mem_k8, athlon_imul_mem_k8_DI,
982 athlon_idiv, athlon_idiv_mem, athlon_str): Added amdfam10.
983
984 2007-02-05 Harsha Jagasia <harsha.jagasia@amd.com>
985
986 * config/i386/athlon.md (athlon_sseld_amdfam10,
987 athlon_mmxld_amdfam10, athlon_ssest_amdfam10,
988 athlon_mmxssest_short_amdfam10): New define_insn_reservation.
989
990 2007-02-05 Harsha Jagasia <harsha.jagasia@amd.com>
991
992 * config/i386/athlon.md (athlon_sseins_amdfam10): New
993 define_insn_reservation.
994 * config/i386/i386.md (sseins): Added sseins to define_attr type
995 and define_attr unit.
996 * config/i386/sse.md: Set type attribute to sseins for insertq
997 and insertqi.
998
999 2007-02-05 Harsha Jagasia <harsha.jagasia@amd.com>
1000
1001 * config/i386/athlon.md (sselog_load_amdfam10, sselog_amdfam10,
1002 ssecmpvector_load_amdfam10, ssecmpvector_amdfam10,
1003 ssecomi_load_amdfam10, ssecomi_amdfam10,
1004 sseaddvector_load_amdfam10, sseaddvector_amdfam10): New
1005 define_insn_reservation.
1006 (ssecmp_load_k8, ssecmp, sseadd_load_k8, seadd): Added amdfam10.
1007
1008 2007-02-05 Harsha Jagasia <harsha.jagasia@amd.com>
1009
1010 * config/i386/athlon.md (cvtss2sd_load_amdfam10,
1011 cvtss2sd_amdfam10, cvtps2pd_load_amdfam10, cvtps2pd_amdfam10,
1012 cvtsi2sd_load_amdfam10, cvtsi2ss_load_amdfam10,
1013 cvtsi2sd_amdfam10, cvtsi2ss_amdfam10, cvtsd2ss_load_amdfam10,
1014 cvtsd2ss_amdfam10, cvtpd2ps_load_amdfam10, cvtpd2ps_amdfam10,
1015 cvtsX2si_load_amdfam10, cvtsX2si_amdfam10): New
1016 define_insn_reservation.
1017
1018 * config/i386/sse.md (cvtsi2ss, cvtsi2ssq, cvtss2si,
1019 cvtss2siq, cvttss2si, cvttss2siq, cvtsi2sd, cvtsi2sdq,
1020 cvtsd2si, cvtsd2siq, cvttsd2si, cvttsd2siq,
1021 cvtpd2dq, cvttpd2dq, cvtsd2ss, cvtss2sd,
1022 cvtpd2ps, cvtps2pd): Added amdfam10_decode attribute.
1023
1024 2007-02-05 Harsha Jagasia <harsha.jagasia@amd.com>
1025
1026 * config/i386/athlon.md (athlon_ssedivvector_amdfam10,
1027 athlon_ssedivvector_load_amdfam10, athlon_ssemulvector_amdfam10,
1028 athlon_ssemulvector_load_amdfam10): New define_insn_reservation.
1029 (athlon_ssediv, athlon_ssediv_load_k8, athlon_ssemul,
1030 athlon_ssemul_load_k8): Added amdfam10.
1031
1032 2007-02-05 Harsha Jagasia <harsha.jagasia@amd.com>
1033
1034 * config/i386/i386.h (TARGET_SSE_UNALIGNED_MOVE_OPTIMAL): New macro.
1035 (x86_sse_unaligned_move_optimal): New variable.
1036
1037 * config/i386/i386.c (x86_sse_unaligned_move_optimal): Enable for
1038 m_AMDFAM10.
1039 (ix86_expand_vector_move_misalign): Add code to generate movupd/movups
1040 for unaligned vector SSE double/single precision loads for AMDFAM10.
1041
1042 2007-02-05 Harsha Jagasia <harsha.jagasia@amd.com>
1043
1044 * config/i386/i386.h (TARGET_AMDFAM10): New macro.
1045 (TARGET_CPU_CPP_BUILTINS): Add code for amdfam10.
1046 Define TARGET_CPU_DEFAULT_amdfam10.
1047 (TARGET_CPU_DEFAULT_NAMES): Add amdfam10.
1048 (processor_type): Add PROCESSOR_AMDFAM10.
1049
1050 * config/i386/i386.md: Add amdfam10 as a new cpu attribute to match
1051 processor_type in config/i386/i386.h.
1052 Enable imul peepholes for TARGET_AMDFAM10.
1053
1054 * config.gcc: Add support for --with-cpu option for amdfam10.
1055
1056 * config/i386/i386.c (amdfam10_cost): New variable.
1057 (m_AMDFAM10): New macro.
1058 (m_ATHLON_K8_AMDFAM10): New macro.
1059 (x86_use_leave, x86_push_memory, x86_movx, x86_unroll_strlen,
1060 x86_cmove, x86_3dnow_a, x86_deep_branch, x86_use_simode_fiop,
1061 x86_promote_QImode, x86_integer_DFmode_moves,
1062 x86_partial_reg_dependency, x86_memory_mismatch_stall,
1063 x86_accumulate_outgoing_args, x86_arch_always_fancy_math_387,
1064 x86_sse_partial_reg_dependency, x86_sse_typeless_stores,
1065 x86_use_ffreep, x86_use_incdec, x86_four_jump_limit,
1066 x86_schedule, x86_use_bt, x86_cmpxchg16b, x86_pad_returns):
1067 Enable/disable for amdfam10.
1068 (override_options): Add amdfam10_cost to processor_target_table.
1069 Set up PROCESSOR_AMDFAM10 for amdfam10 entry in
1070 processor_alias_table.
1071 (ix86_issue_rate): Add PROCESSOR_AMDFAM10.
1072 (ix86_adjust_cost): Add code for amdfam10.
1073
1074 2007-02-05 Harsha Jagasia <harsha.jagasia@amd.com>
1075
1076 * config/i386/i386.opt: Add new Advanced Bit Manipulation (-mabm)
1077 instruction set feature flag. Add new (-mpopcnt) flag for popcnt
1078 instruction. Add new SSE4A (-msse4a) instruction set feature flag.
1079 * config/i386/i386.h: Add builtin definition for SSE4A.
1080 * config/i386/i386.md: Add support for ABM instructions
1081 (popcnt and lzcnt).
1082 * config/i386/sse.md: Add support for SSE4A instructions
1083 (movntss, movntsd, extrq, insertq).
1084 * config/i386/i386.c: Add support for ABM and SSE4A builtins.
1085 Add -march=amdfam10 flag.
1086 * config/i386/ammintrin.h: Add support for SSE4A intrinsics.
1087 * doc/invoke.texi: Add documentation on flags for sse4a, abm, popcnt
1088 and amdfam10.
1089 * doc/extend.texi: Add documentation for SSE4A builtins.
1090
1091 2007-02-05 Bob Wilson <bob.wilson@acm.org>
1092
1093 * config/xtensa/xtensa.c (constantpool_mem_p): Skip over SUBREGs.
1094
1095 2007-02-05 Richard Guenther <rguenther@suse.de>
1096
1097 * tree-vectorizer.h (vectorizable_function): Add argument type
1098 argument, change return type.
1099 * tree-vect-patterns.c (vect_recog_pow_pattern): Adjust caller.
1100 * tree-vect-transform.c (vectorizable_function): Handle extra
1101 argument, return vectorized function decl.
1102 (build_vectorized_function_call): Remove.
1103 (vectorizable_call): Handle calls with result and argument types
1104 differing. Handle loop vectorization factor correctly.
1105 * targhooks.c (default_builtin_vectorized_function): Adjust for
1106 extra argument.
1107 * targhooks.h (default_builtin_vectorized_function): Likewise.
1108 * target.h (builtin_vectorized_function): Add argument type
1109 argument.
1110 * config/i386/i386.c (ix86_builtin_vectorized_function): Handle
1111 extra argument, allow vectorizing of lrintf.
1112 * doc/tm.texi (TARGET_VECTORIZE_BUILTIN_VECTORIZED_FUNCTION): Adjust
1113 documentation of target hook.
1114
1115 2007-02-05 Hans-Peter Nilsson <hp@axis.com>
1116
1117 PR target/30665
1118 * config/cris/cris.md ("*andsi_movu", "*andsi_clear", "*andhi_movu")
1119 ("*andhi_clear", andu (casesi+45)): For size-changed operand where
1120 memory is allowed, require !side_effects_p, not just !MEM_VOLATILE_P.
1121
1122 2007-02-05 Roger Sayle <roger@eyesopen.com>
1123
1124 * fold-const.c (fold_unary) <REAL_PART>: Test for availability of
1125 BUILT_IN_COS before simplifying REAL_PART(CEXPI)) to COS.
1126 <IMAG_PART>: Likewise, check for availability of BUILT_IN_SIN.
1127 * builtins.c (fold_builtin_sincos): Check for TARGET_C99_FUNCTIONS
1128 before canonicalizing sincos to cexpi.
1129 (fold_builtin_cexp): Likewise, for canonicalizing cexp to cexpi.
1130
1131 2007-02-05 Roger Sayle <roger@eyesopen.com>
1132
1133 * config/alpha/alpha.c (alpha_add_builtins): New Helper function.
1134 Set TREE_READONLY and TREE_NOTHROW directly, not via attributes.
1135 (alpha_init_builtins): Use alpha_add_builtins to process tables.
1136
1137 2007-02-05 Roger Sayle <roger@eyesopen.com>
1138
1139 * mips-tfile.c (initialize_init_file): Correct endianness test.
1140
1141 2007-02-05 Kazu Hirata <kazu@codesourcery.com>
1142
1143 * config/m68k/m68k.md (pushdi-1, pushdi, movsi+1): Don't use
1144 the 'y' constraint.
1145
1146 2007-02-05 Richard Sandiford <richard@codesourcery.com>
1147
1148 * dwarf2out.c (dwarf2out_frame_debug_expr): Record the register
1149 saves in a PARALLEL before the register assignments.
1150
1151 2007-02-05 Richard Sandiford <richard@codesourcery.com>
1152
1153 * doc/tm.texi (DWARF_ALT_FRAME_RETURN_COLUMN): Do not require
1154 DWARF_FRAME_RETURN_COLUMN to be a general register.
1155 * dwarf2out.c (init_return_column_size): New function, split from...
1156 (expand_builtin_init_dwarf_reg_sizes): ...here. Allow both
1157 DWARF_FRAME_RETURN_COLUMN and DWARF_ALT_FRAME_RETURN_COLUMN
1158 to be nongeneral registers.
1159 * config/m68k/m68k.h (DWARF_FRAME_REGNUM): Only map FP and
1160 integer registers.
1161 (DWARF_FRAME_REGISTERS, DWARF_FRAME_RETURN_COLUMN): Define.
1162 (DWARF_ALT_FRAME_RETURN_COLUMN): Define.
1163
1164 2007-02-04 Zdenek Dvorak <dvorakz@suse.cz>
1165
1166 * cfgcleanup.c (try_optimize_cfg): Avoid removing ENTRY_BLOCK_PTR.
1167
1168 2007-02-04 Zdenek Dvorak <dvorakz@suse.cz>
1169
1170 * cfgloopmanip.c (loop_delete_branch_edge): Removed.
1171 (remove_path): Use can_remove_branch_p and remove_branch instead
1172 of loop_delete_branch_edge.
1173 * tree-ssa-loop-manip.c (scale_dominated_blocks_in_loop): New function.
1174 (tree_transform_and_unroll_loop): Remove dead branches immediately.
1175 Update profile using scale_dominated_blocks_in_loop.
1176 * cfghooks.c (can_remove_branch_p, remove_branch): New functions.
1177 * cfghooks.h (struct cfg_hooks): Add can_remove_branch_p.
1178 (can_remove_branch_p, remove_branch): Declare.
1179 * tree-cfg.c (tree_can_remove_branch_p): New function.
1180 (tree_cfg_hooks): Add tree_can_remove_branch_p.
1181 * cfgrtl.c (rtl_can_remove_branch_p): New function.
1182 (rtl_cfg_hooks, cfg_layout_rtl_cfg_hook): Add rtl_can_remove_branch_p.
1183
1184 2007-02-05 Jan Hubicka <jh@suse.cz>
1185
1186 PR middle-end/30696
1187 * ipa-inline.c (cgraph_clone_inlined_nodes): When there are unanalyzed
1188 nodes in cgraph, don't remove offline copy of the function.
1189
1190 2007-02-04 Jan Hubicka <jh@suse.cz>
1191
1192 * tree-sra.c (sra_walk_expr): Add linebreaks. BITFIELD_REFs into
1193 vectors might cause maybe_lookup_element_for_expr to be called
1194 on non-sra-candidate.
1195
1196 2007-02-04 Kazu Hirata <kazu@codesourcery.com>
1197
1198 * config/bfin/bfin-modes.def, config/bfin/bfin.c,
1199 config/bfin/bfin.md, config/bfin/predicates.md: Follow
1200 spelling conventions.
1201
1202 2007-02-04 Richard Guenther <rguenther@suse.de>
1203
1204 PR middle-end/30636
1205 * fold-const.c (try_move_mult_to_index): Make sure to not
1206 overflow one dimension of a multi-dimensional array access.
1207
1208 2007-02-04 Jan Hubicka <jh@suse.cz>
1209
1210 * passes.c (init_optimization_passes): Reindent.
1211
1212 2007-02-04 Jan Hubicka <jh@suse.cz>
1213 Eric Botcazou <ebotcazou@adacore.com>
1214
1215 * tree-optimize.c (has_abnormal_outgoing_edge_p): Move to...
1216 (execute_fixup_cfg): Break out the abnormal goto code.
1217 * tree-inline.c (has_abnormal_outgoing_edge_p): ...here.
1218 (make_nonlocal_label_edges): Move here from execute_fixup_cfg.
1219 (optimize_inline_calls): Call make_nonlocal_label_edges.
1220
1221 2007-02-04 Jan Hubicka <jh@suse.cz>
1222
1223 * tree-ssa-copyrename.c (copy_rename_partition_coalesce): Return
1224 true when something was changed.
1225 (rename_ssa_copies): When something was changed, do
1226 TODO_remove_unused_locals.
1227 * tree-ssa-forwprop.c (tree_ssa_forward_propagate_single_use_value):
1228 add TODO_remove_unused_locals when instruction was removed.
1229
1230 2007-02-04 Jan Hubicka <jh@suse.cz>
1231
1232 * ipa-inline.c (try_inline): Improve debug output; work on already
1233 inline edges too.
1234 (cgraph_decide_inlining_incrementally): Indent; improve debug output;
1235 call try_inline for already inlined edges too when flattening;
1236 inline also functions that make callee growth but overall unit size
1237 reduce.
1238
1239 2007-02-04 Kazu Hirata <kazu@codesourcery.com>
1240
1241 * config/m32c/bitops.md, config/m32c/jump.md,
1242 config/m32c/m32c.c, config/m32c/m32c.h, config/m32r/m32r.c,
1243 config/m32r/m32r.h, config/m32r/m32r.md,
1244 config/m32r/predicates.md, config/m68hc11/larith.asm,
1245 config/m68hc11/m68hc11.c, config/m68hc11/m68hc11.h,
1246 config/m68k/m68k.h, config/mcore/mcore.md, config/mips/4k.md,
1247 config/mips/mips-protos.h, config/mips/mips.c,
1248 config/mips/mips.h, config/mips/mips.md, config/mips/mips16.S,
1249 config/mn10300/mn10300.h, config/mn10300/predicates.md,
1250 config/mt/mt.c, config/mt/mt.h, config/mt/mt.md: Follow
1251 spelling conventions.
1252
1253 * config/v850/v850.c, config/v850/v850.h, config/v850/v850.md:
1254 Follow spelling conventions.
1255
1256 2007-02-03 Douglas Gregor <doug.gregor@gmail.com>
1257
1258 * c-opts.c (c_common_post_options): If C++0x mode is enabled, don't
1259 warn about C++0x compatibility.
1260
1261 2007-02-04 Kazu Hirata <kazu@codesourcery.com>
1262
1263 * config/h8300/h8300.c, config/h8300/h8300.h,
1264 config/h8300/h8300.md: Follow spelling conventions.
1265
1266 2007-02-03 Uros Bizjak <ubizjak@gmail.com>
1267
1268 PR middle-end/30667
1269 * combine.c (try_combine): Do not substitute source operand
1270 with constants wider than 2 * HOST_BITS_PER_WIDE_INT.
1271
1272 2007-02-03 Jan Hubicka <jh@suse.cz>
1273
1274 PR gcov-profile/30650
1275 * value-prof.c (stringop_block_profile): Fix handling of size counter;
1276 do not divide by zero for never executed counters.
1277 (tree_find_values_to_profile): Fix counters.
1278 * gcov-ui.h (GCOV_COUNTER_AVERAGE, GCOV_COUNTER_IOR): Fix comments.
1279
1280 2007-02-03 Ian Lance Taylor <iant@google.com>
1281
1282 * lower-subreg.c (simple_move_operand): New static function,
1283 broken out of simple_move. Reject LABEL_REF, SYMBOL_REF, and HIGH
1284 operands.
1285 (simple_move): Call simple_move_operand.
1286 (find_decomposable_subregs): Add special handling of MEMs.
1287 (can_decompose_p): Rename from cannot_decompose_p. Reverse
1288 meaning of return value. If we see a hard register, test whether
1289 it can store a word_mode value. Change all callers.
1290
1291 2007-02-03 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
1292
1293 * pa.md (addvdi3, addvsi3, subvdi3, subvsi3, negvdi2, negvsi2): New
1294 ftrapv insns and expanders.
1295 (subdi3): Change define_expand operand 1 to arith11_operand, and
1296 operand 2 to reg_or_0_operand. Change constraints of 64-bit insn
1297 pattern to handle reg_or_0 operands. Revise 32-bit insn pattern to
1298 handle 11-bit constants and reg_or_0 operands in operands 1 and 2,
1299 respectively.
1300
1301 PR middle-end/30174
1302 * varasm.c (notice_global_symbol): Treat global objects as weak when
1303 flag_shlib is true.
1304
1305 2007-02-03 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
1306
1307 * emit-rtl.c (dconstpi): Delete.
1308 (dconstsqrt2): New.
1309 (init_emit_once): Delete dconstpi and init dconstsqrt2.
1310 * real.h (dconstpi): Delete.
1311 (dconstsqrt2): New.
1312 * builtins.c (fold_builtin_cabs): Use dconstsqrt2.
1313 (fold_builtin_hypot): Likewise.
1314
1315 2007-02-03 Tom Tromey <tromey@redhat.com>
1316
1317 PR driver/30246
1318 * gcc.c (cpp_unique_options): Any of -ggdb3, -gstabs3,
1319 -gcoff3, -gxcoff3, -gvms3 implies -dD.
1320
1321 2007-02-03 Kazu Hirata <kazu@codesourcery.com>
1322
1323 * c-decl.c, config/avr/avr.c, config/avr/avr.h,
1324 config/m68k/m68k.c, config/m68k/netbsd-elf.h,
1325 config/mn10300/mn10300.c, config/pdp11/pdp11.h,
1326 config/rs6000/cell.md, config/rs6000/darwin.h,
1327 config/sh/sh.md, config/sh/sh4-300.md, config/spu/spu.c,
1328 config/spu/spu.md, cselib.c, expr.c, haifa-sched.c, hwint.h,
1329 jump.c, reload.c, sched-deps.c, sched-int.h, tree-inline.c,
1330 tree-profile.c, tree-ssa-live.h, tree-vrp.c: Fix comment
1331 typos. Follow spelling conventions.
1332 * doc/invoke.texi: Follow spelling conventions.
1333
1334 2007-02-03 Roger Sayle <roger@eyesopen.com>
1335
1336 * simplify-rtx.c (simplify_relational_operation_1): Implement some
1337 canonicalization transformations that attempt to simplify integer
1338 constant comparisons to become comparisons against zero.
1339
1340 2007-02-03 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
1341
1342 * builtins.c (fold_builtin_cabs): Fold cabs(x+xi) into
1343 fabs(x)*sqrt(2).
1344 * fold-const.c (fold_binary): Fix comment typos. Fold complex
1345 (x,0)-(0,y) into (x,-y). Likewise (0,y)-(x,0) into (-x,y).
1346
1347 2007-02-02 Mike Stump <mrs@apple.com>
1348
1349 * config/darwin9.h (DARWIN_LINKER_GENERATES_ISLANDS): Add.
1350 * config/rs6000/rs6000.c (DARWIN_GENERATE_ISLANDS): Add.
1351 (output_call): Use DARWIN_GENERATE_ISLANDS to decide when to
1352 generate a branch island.
1353
1354 2007-02-02 Bob Wilson <bob.wilson@acm.org>
1355
1356 * config/xtensa/xtensa.c (smalloffset_mem_p): Use BASE_REG_P.
1357 (xtensa_legitimate_address_p): New.
1358 (xtensa_legitimize_address): New.
1359 (xtensa_output_addr_const_extra): New.
1360 * config/xtensa/xtensa.h (REG_OK_STRICT_FLAG): Define.
1361 (BASE_REG_P): New.
1362 (REG_OK_FOR_BASE_P): Use BASE_REG_P.
1363 (GO_IF_LEGITIMATE_ADDRESS): Move code to xtensa_legitimate_address_p.
1364 (LEGITIMIZE_ADDRESS): Move code to xtensa_legitimize_address.
1365 (OUTPUT_ADDR_CONST_EXTRA): Move code to xtensa_output_addr_const_extra.
1366 * config/xtensa/xtensa-protos.h (xtensa_legitimate_address_p): New.
1367 (xtensa_legitimize_address): New.
1368 (xtensa_output_addr_const_extra): New.
1369
1370 2007-02-02 Steve Ellcey <sje@cup.hp.com>
1371
1372 * config/ia64/ia64.c (ia64_print_operand): Fix compare strings.
1373
1374 2007-02-02 Ian Lance Taylor <iant@google.com>
1375
1376 * expmed.c (expand_divmod): Add comment.
1377
1378 2007-02-02 Kazu Hirata <kazu@codesourcery.com>
1379
1380 * emit-rtl.c (renumber_insns): Remove.
1381 * flags.h: Remove the extern for flag_renumber_insns.
1382 * rtl.h: Remove the prototype for renumber_insns.
1383 * toplev.c (flag_renumber_insns): Remove.
1384
1385 2007-02-02 Hui-May Chang <hm.chang@apple.com>
1386
1387 Revert for x86 darwin:
1388 2005-06-19 Uros Bizjak <uros@kss-loka.si>
1389
1390 * config/i386/i386.c (ix86_function_arg_regno_p): Put back the
1391 code before the following patch under TARGET_MACHO.
1392 (ix86_function_value_regno_p): Likewise.
1393
1394 2007-02-02 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
1395
1396 * fold-const.c (negate_expr_p, fold_negate_expr): Handle
1397 COMPLEX_EXPR.
1398
1399 2007-02-02 Maxim Kuvyrkov <mkuvyrkov@ispras.ru>
1400
1401 PR target/29682
1402 * config/ia64/ia64.c (ia64_speculate_insn): Restrict to memory loads to
1403 general or fp registers. Add comments.
1404 * config/ia64/ia64.md (reg_pred_prefix): Add comment.
1405
1406 2007-02-02 Paolo Bonzini <bonzini@gnu.org>
1407
1408 * pointer-set.c (insert_aux): Only return insertion slot.
1409 (pointer_set_insert): Adjust.
1410 (pointer_set_traverse, struct pointer_map_t, pointer_map_create,
1411 pointer_map_destroy, pointer_map_insert, pointer_map_contains,
1412 pointer_map_traverse): New.
1413 * pointer-set.h (pointer_set_traverse, struct pointer_map_t,
1414 pointer_map_create, pointer_map_destroy, pointer_map_insert,
1415 pointer_map_contains, pointer_map_traverse): Declare.
1416
1417 2007-02-02 Jakub Jelinek <jakub@redhat.com>
1418
1419 PR middle-end/30473
1420 * builtins.c (fold_builtin_sprintf): Do not attempt to optimize
1421 sprintf (str, "%s"). Do not optimize sprintf (str, "nopercent", p++).
1422
1423 2007-02-02 Maxim Kuvyrkov <mkuvyrkov@ispras.ru>
1424
1425 * sched-int.h (ds_to_dk, dk_to_ds): Declare functions.
1426
1427 (struct _dep): New type.
1428 (dep_t): New typedef.
1429 (DEP_PRO, DEP_CON, DEP_KIND): New access macros.
1430 (DEP_STATUS): New access macro. The macro with the same name was
1431 renamed to DEP_LINK_STATUS.
1432 (dep_init): Declare function
1433
1434 (struct _dep_link): New type.
1435 (dep_link_t): New typedef.
1436 (DEP_LINK_NODE, DEP_LINK_NEXT, DEP_LINK_PREV_NEXTP): New access macros.
1437 (DEP_LINK_DEP, DEP_LINK_PRO, DEP_LINK_CON, DEP_LINK_KIND): New macros.
1438 (DEP_LINK_STATUS): New macro.
1439 (debug_dep_links): New debug function.
1440
1441 (struct _deps_list): New type.
1442 (deps_list_t): New typedef.
1443 (DEPS_LIST_FIRST): New access macro.
1444 (FOR_EACH_DEP_LINK): New cycle macro.
1445 (create_deps_list, free_deps_list, delete_deps_list): Declare
1446 functions.
1447 (deps_list_empty_p, debug_deps_list, add_back_dep_to_deps_list): Ditto.
1448 (find_link_by_pro_in_deps_list, find_link_by_con_in_deps_list): Ditto.
1449 (copy_deps_list_change_con): Ditto.
1450
1451 (move_dep_link): Declare function.
1452
1453 (struct _dep_node): New type.
1454 (dep_node_t): New typedef.
1455 (DEP_NODE_BACK, DEP_NODE_DEP, DEP_NODE_FORW): New access macros.
1456
1457 (struct haifa_insn_data.back_deps): New field to hold backward
1458 dependencies of the insn.
1459 (struct haifa_insn_data.depend): Rename to forw_deps. Change its type
1460 to deps_list_t.
1461 (struct haifa_insn_data.resolved_deps): Rename to resolved_back_deps.
1462 Change its type to deps_list_t.
1463 (INSN_BACK_DEPS): New access macro to use instead of LOG_LINKS.
1464 (INSN_DEPEND): Rename to INSN_FORW_DEPS.
1465 (RESOLVED_DEPS): Rename to INSN_RESOLVED_BACK_DEPS.
1466
1467 (INSN_COST): Move to haifa-sched.c. Use insn_cost () instead.
1468
1469 (DEP_STATUS): Rename to DEP_LINK_STATUS. Fix typo in the comment.
1470
1471 (add_forw_dep, delete_back_forw_dep, insn_cost): Update declaration and
1472 all callers.
1473 (dep_cost): Declare.
1474
1475 * sched-deps.c (CHECK): New macro to (en/dis)able sanity checks.
1476 (ds_to_dk, dk_to_ds): New functions.
1477
1478 (init_dep_1): New static function.
1479 (init_dep): New function.
1480 (copy_dep): New static function.
1481
1482 (dep_link_consistent_p, attach_dep_link, add_to_deps_list): New static
1483 functions.
1484 (detach_dep_link): New static function.
1485 (move_dep_link): New function.
1486
1487 (dep_links_consistent_p, dump_dep_links): New static functions.
1488 (debug_dep_links): New debugging function.
1489
1490 (deps_obstack, dl_obstack, dn_obstack): New static variables.
1491
1492 (alloc_deps_list, init_deps_list): New static functions.
1493 (create_deps_list): New function.
1494 (clear_deps_list): New static function.
1495 (free_deps_list, delete_deps_list, deps_list_empty_p): New functions.
1496 (deps_list_consistent_p, dump_deps_list): New static functions.
1497 (debug_deps_list): New function.
1498 (add_back_dep_to_deps_list, find_link_by_pro_in_deps_list): New
1499 functions.
1500 (find_link_by_con_in_deps_list, copy_deps_list_change_con): Ditto.
1501
1502 (maybe_add_or_update_back_dep_1, add_or_update_back_dep_1): Update to
1503 use new scheduler dependencies lists.
1504 (add_back_dep, delete_all_dependences, fixup_sched_groups): Ditto.
1505 (sched_analyze): Ditto. Initialize dependencies lists.
1506 (add_forw_dep, compute_forward_dependences): Update to use new
1507 scheduler dependencies lists.
1508
1509 (init_dependency_caches): Init deps_obstack.
1510 (free_dependency_caches): Free deps_obstack.
1511
1512 (adjust_add_sorted_back_dep, adjust_back_add_forw_dep): Update to use
1513 new scheduler dependencies lists.
1514 (delete_forw_dep, add_or_update_back_forw_dep): Ditto.
1515 (add_back_forw_dep, delete_back_forw_dep): Ditto.
1516
1517 * sched-rgn.c (set_spec_fed, find_conditional_protection, is_pfree):
1518 Update to use new scheduler dependencies lists.
1519 (is_conditionally_protected, is_prisky, add_branch_dependences): Ditto.
1520 (debug_dependencies): Ditto.
1521 (schedule_region): Update comments.
1522
1523 * sched-ebb.c (earliest_block_with_similiar_load): Update to use new
1524 scheduler dependencies lists.
1525 (schedule_ebb): Update comments.
1526
1527 * rtl.def (DEPS_LIST): Remove.
1528
1529 * lists.c (unused_deps_list): Remove.
1530 (free_list): Update assertions.
1531
1532 (alloc_DEPS_LIST, free_DEPS_LIST_list, free_DEPS_LIST_node): Remove.
1533 (remove_free_DEPS_LIST_elem, copy_DEPS_LIST_list): Ditto.
1534
1535 * rtl.h (free_DEPS_LIST_list, alloc_DEPS_LIST): Remove declarations.
1536 (remove_free_DEPS_LIST_elem, copy_DEPS_LIST_list): Ditto.
1537
1538 * haifa-sched.c (comments): Update.
1539 (insn_cost1): Remove. Inline the code into insn_cost ().
1540 (insn_cost): Update to use new scheduler dependencies lists. Move
1541 processing of the dependency cost to dep_cost ().
1542 (dep_cost): New function. Use it instead of insn_cost () when
1543 evaluating cost of the dependency. Use compatible interface to
1544 interact with the target.
1545 (priority): Update to use new scheduler dependencies lists.
1546 (rank_for_schedule): Ditto. Optimize heuristic that prefers the insn
1547 with greater number of insns that depend on the insn.
1548 (schedule_insn): Update to use new scheduler dependencies lists. Add
1549 code to free backward dependencies lists. Inline and optimize code
1550 from resolve_dep () - see PR28071.
1551 (ok_for_early_queue_removal): Update to use new scheduler dependencies
1552 lists. Update call to targetm.sched.is_costly_dependence hook.
1553
1554 (fix_inter_tick, try_ready, fix_tick_ready): Update to use new
1555 scheduler dependencies lists.
1556
1557 (resolve_dep): Remove. Move the logic to schedule_insn ().
1558 (init_h_i_d): Initialize dependencies lists.
1559
1560 (process_insn_depend_be_in_spec): Rename to
1561 process_insn_forw_deps_be_in_spec. Update to use new scheduler
1562 dependencies lists.
1563 (add_to_speculative_block, create_check_block_twin, fix_recovery_deps):
1564 Update to use new scheduler dependencies lists.
1565 (clear_priorities, calc_priorities, add_jump_dependencies): Ditto.
1566
1567 * ddg.c (create_ddg_dependence, create_ddg_dep_no_link): Update to use
1568 new scheduler dependencies lists.
1569 (build_intra_loop_deps): Ditto.
1570
1571 * target.h (struct _dep): Declare to use in
1572 gcc_target.sched.is_costly_dependence.
1573 (struct gcc_target.sched.adjust_cost): Fix typo.
1574 (struct gcc_target.sched.is_costly_dependence): Change signature to use
1575 single dep_t parameter instead of an equivalent triad.
1576 (struct gcc_target.sched.adjust_cost_2): Remove.
1577
1578 * target-def.h (TARGET_SCHED_ADJUST_COST_2): Remove.
1579
1580 * reg-notes.def (DEP_TRUE, DEP_OUTPUT, DEP_ANTI): Update comments.
1581
1582 * doc/tm.texi (TARGET_SCHED_IS_COSTLY_DEPENDENCE): Update
1583 documentation.
1584 (TARGET_SCHED_ADJUST_COST_2): Remove documentation.
1585
1586 * doc/rtl.texi (LOG_LINKS): Remove part about instruction scheduler.
1587 (REG_DEP_TRUE): Document.
1588
1589 * config/ia64/ia64.c (ia64_adjust_cost_2): Rename to ia64_adjust_cost.
1590 Change signature to correspond to the targetm.sched.adjust_cost hook.
1591 Update use in TARGET_SCHED_ADJUST_COST_2.
1592 (TARGET_SCHED_ADJUST_COST_2): Rename to TARGET_SCHED_ADJUST_COST.
1593 (ia64_dependencies_evaluation_hook, ia64_dfa_new_cycle): Update to use
1594 new scheduler dependencies lists.
1595 (ia64_gen_check): Ditto.
1596
1597 * config/mips/mips.c (vr4130_swap_insns_p): Update to use new scheduler
1598 dependencies lists.
1599
1600 * config/rs6000/rs6000.c (rs6000_is_costly_dependence): Change
1601 signature to correspond to the targetm.sched.is_costly_dependence hook.
1602 (is_costly_group): Update to use new scheduler dependencies lists.
1603
1604 * config/spu/spu.c (spu_sched_adjust_cost): Use insn_cost () function
1605 instead of INSN_COST () macro.
1606
1607 2007-02-01 Ian Lance Taylor <iant@google.com>
1608
1609 * lower-subreg.c (resolve_clobber): Handle a subreg of a concatn.
1610
1611 2007-02-01 Guy Martin <gmsoft@gentoo.org>
1612
1613 * pa.md (tp_load): Correct mfctl instruction syntax.
1614
1615 2007-02-01 Geoffrey Keating <geoffk@apple.com>
1616
1617 * config/rs6000/rs6000.c (rs6000_stack_info): Correct
1618 altivec_padding_size calculation on AIX. Improve comment, add
1619 assert to verify that it's right.
1620
1621 * config/rs6000/darwin.h (MD_UNWIND_SUPPORT): Don't define for
1622 64-bit.
1623
1624 2007-2-01 Seongbae Park <seongbae.park@gmail.com>
1625
1626 PR inline-asm/28686
1627 * global.c (global_alloc): Add mising initialization of
1628 ELIMINABLE_REGSET.
1629
1630 2007-02-01 Roger Sayle <roger@eyesopen.com>
1631
1632 * alias.c (init_alias_analysis): Correct whitespace.
1633 * bb-reorder.c (fix_edges_for_rarely_executed_code,
1634 partition_hot_cold_basic_blocks): Likewise.
1635 * builtins.c (expand_builtin_printf, expand_builtin_fprintf,
1636 expand_builtin_sprintf, fold_builtin_carg, fold_builtin_sprintf,
1637 maybe_emit_sprintf_chk_warning, fold_builtin_sprintf_chk,
1638 fold_builtin_snprintf_chk, fold_builtin_printf,
1639 fold_builtin_fprintf, do_mpfr_ckconv, do_mpfr_arg1, do_mpfr_arg2,
1640 do_mpfr_arg3, do_mpfr_sincos): Likewise.
1641 * cfgcleanup.c (cleanup_cfg): Likewise.
1642 * cfgexpand.c (tree_expand_cfg): Likewise.
1643 * fold-const.c (fold_binary) <RDIV_EXPR>: Likewise.
1644 * function.c (get_next_funcdef_no): Likewise.
1645 * gengtype.c (main): Likewise.
1646 * genmodes.c (main): Likewise.
1647 * gcse.c (bypass_conditional_jumps, print_ldst_list): Likewise.
1648 * haifa-sched.c (schedule_block, extend_h_i_d): Likewise.
1649 * ifcvt.c (noce_emit_move_insn): Likewise.
1650 * modulo-sched.c (generate_prolog_epilog, sms_schedule_by_order):
1651 Likewise.
1652 * stor-layout.c (get_best_mode): Likewise.
1653 * tree-ssa-loop-niter.c (get_val_for): Likewise.
1654 * tree-ssa-structalias.c (get_varinfo, get_varinfo_fc,
1655 scc_visit, do_ds_constraint, do_complex_constraint, label_visit,
1656 perform_var_substitution, solve_graph): Likewise.
1657 * tree-vrp.c (vrp_finalize): Likewise.
1658
1659 2007-02-01 Ian Lance Taylor <iant@google.com>
1660
1661 * lower-subreg.c (simplify_gen_subreg_concatn): If we ask for the
1662 high part of a paradoxical subreg, return a constant zero.
1663
1664 2007-02-01 Zdenek Dvorak <dvorakz@suse.cz>
1665
1666 * toplev.c (lang_dependent_init): Call init_set_costs.
1667 * loop-init.c (loop_optimizer_init): Do not call init_set_costs.
1668
1669 2007-02-01 Richard Guenther <rguenther@suse.de>
1670
1671 PR middle-end/30656
1672 * fold-const.c (fold_negate_expr): Allow negating a
1673 constant if overflow does not change.
1674
1675 2007-02-01 Ralf Wildenhues <Ralf.Wildenhues@gmx.de>
1676
1677 * doc/c-tree.texi (Expression trees): Improve markup.
1678 * doc/tm.texi (Register Classes, Addressing Modes)
1679 (Floating Point): Fix spacing after abbreviations. Fix some
1680 typos.
1681
1682 2007-02-01 Ben Elliston <bje@au.ibm.com>
1683
1684 * doc/invoke.texi: Replace "bugfix" with "bug fix" throughout.
1685 * doc/contrib.texi: Likewise.
1686 * doc/install.texi: Likewise.
1687
1688 2007-01-31 Richard Henderson <rth@redhat.com>
1689 Ian Lance Taylor <iant@google.com>
1690
1691 * lower-subreg.c: New file.
1692 * rtl.def (CONCATN): Define.
1693 * passes.c (init_optimization_passes): Add pass_lower_subreg and
1694 pass_lower_subreg2.
1695 * emit-rtl.c (update_reg_offset): New static function, broken out
1696 of gen_rtx_REG_offset.
1697 (gen_rtx_REG_offset): Call update_reg_offset.
1698 (gen_reg_rtx_offset): New function.
1699 * regclass.c: Revert patch of 2006-03-05, restoring
1700 reg_scan_update.
1701 (clear_reg_info_regno): New function.
1702 * dwarf2out.c (concatn_loc_descriptor): New static function.
1703 (loc_descriptor): Handle CONCATN.
1704 * common.opt (fsplit_wide_types): New option.
1705 * opts.c (decode_options): Set flag_split_wide_types when
1706 optimizing.
1707 * timevar.def (TV_LOWER_SUBREG): Define.
1708 * rtl.h (gen_reg_rtx_offset): Declare.
1709 (reg_scan_update): Declare.
1710 * regs.h (clear_reg_info_regno): Declare.
1711 * tree-pass.h (pass_lower_subreg): Declare.
1712 (pass_lower_subreg2): Declare.
1713 * doc/invoke.texi (Option Summary): List -fno-split-wide-types.
1714 (Optimize Options): Add -fsplit-wide-types to -O1 list. Document
1715 -fsplit-wide-types.
1716 * doc/rtl.texi (Regs and Memory): Document concat and concatn.
1717 * Makefile.in (OBJS-common): Add lower-subreg.o.
1718 (lower-subreg.o): New target.
1719
1720 2007-01-31 Kazu Hirata <kazu@codesourcery.com>
1721
1722 * config/sh/sh.h (HAVE_SECONDARY_RELOADS): Remove.
1723
1724 2007-01-31 Anatoly Sokolov <aesok@post.ru>
1725
1726 PR target/19087
1727 * config/avr/avr.c (DWARF2_ADDR_SIZE): Define.
1728
1729 2007-01-31 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
1730
1731 PR middle-end/29335
1732 * builtins.c (fold_builtin_sqrt): Use MPFR for constant args.
1733
1734 2007-01-31 Zdenek Dvorak <dvorakz@suse.cz>
1735
1736 * cfgloop.h: Include vec-prim.h.
1737 (enum li_flags): Remove LI_ONLY_OLD.
1738 (loop_iterator): Changed.
1739 (fel_next, fel_init): Iterate over loop tree.
1740 (FOR_EACH_LOOP_BREAK): New macro.
1741 * loop-unswitch.c (unswitch_loops): Do not pass LI_ONLY_OLD to
1742 FOR_EACH_LOOP.
1743 * tree-ssa-loop-unswitch.c (tree_ssa_unswitch_loops): Ditto.
1744 * modulo-sched.c (sms_schedule): Ditto.
1745 * tree-vectorizer.c (vectorize_loops): Ditto.
1746 * doc/loop.texi: Update information on loop numbering and behavior of
1747 FOR_EACH_LOOP wrto new loops.
1748 * tree-scalar-evolution.c (compute_overall_effect_of_inner_loop,
1749 add_to_evolution_1): Test nestedness of loops instead of comparing
1750 their numbers.
1751 * tree-chrec.c (chrec_fold_plus_poly_poly,
1752 chrec_fold_multiply_poly_poly, chrec_evaluate,
1753 hide_evolution_in_other_loops_than_loop, chrec_component_in_loop_num,
1754 reset_evolution_in_loop): Ditto.
1755 * Makefile.in (CFGLOOP_H): Add vecprim.h dependency.
1756
1757 2007-01-31 Dirk Mueller <dmueller@suse.de>
1758
1759 * c-common.c (warn_about_parentheses): Separate warning about
1760 un-parenthized sequence of comparison operators from the one
1761 which is supposed to warn about x <= y <= z.
1762
1763 2007-01-31 Uros Bizjak <ubizjak@gmail.com>
1764
1765 * optabs.h (enum optab_index): Add new OTI_isinf.
1766 (isinf_optab): Define corresponding macro.
1767 * optabs.c (init_optabs): Initialize isinf_optab.
1768 * genopinit.c (optabs): Implement isinf_optab using isinf?f2
1769 patterns.
1770 * builtins.c (mathfn_built_in): Handle BUILT_IN_ISINF{,F,L}.
1771 (expand_builtin_interclass_mathfn): Expand BUILT_IN_ISINF{,F,L}
1772 using isinf_optab.
1773 (expand_builtin): Expand BUILT_IN_ISINF{,F,L} using
1774 expand_builtin_interclass_mathfn.
1775 * reg_stack.c (subst_stack_regs_pat): Handle UNSPEC_FXAM.
1776 * config/i386/i386.md (UNSPEC_FXAM): New constant.
1777 (fxam<mode>2_i387): New insn pattern.
1778 (isinf<mode>2) New expander to implement isinf, isinff and isinfl
1779 built-in functions as x87 inline asm.
1780
1781 2007-01-31 Kazu Hirata <kazu@codesourcery.com>
1782
1783 * gcc/config/arm/unwind-arm.h (_sleb128_t, _uleb128_t): New.
1784
1785 2007-01-30 Eric Christopher <echristo@apple.com>
1786
1787 * config.gcc: Add geode.
1788
1789 2007-01-31 Kazu Hirata <kazu@codesourcery.com>
1790
1791 * cgraphunit.c, config/arm/arm.c, config/m68k/m68k.c,
1792 ipa-inline.c, tree-profile.c, tree-ssa-live.c,
1793 tree-ssa-math-opts.c, tree-ssanames.c, tree-vect-analyze.c,
1794 value-prof.c: Fix comment typos.
1795
1796 2007-01-30 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
1797
1798 PR c++/24745
1799 * doc/invoke.texi (Wpointer-arith): Document warning.
1800
1801 2007-01-30 Janis Johnson <janis187@us.ibm.com>
1802
1803 * doc/extend.texi (Decimal Floating Types): Remove decfloat.h from
1804 the list of discrepancies from the draft TR.
1805
1806 2007-01-30 Dirk Mueller <dmueller@suse.de>
1807
1808 PR c++/30601
1809 * doc/invoke.texi (-Wreturn-type): Update description to
1810 match new behavior.
1811
1812 2007-01-30 Richard Sandiford <richard@codesourcery.com>
1813
1814 * cfgrtl.c (try_redirect_by_replacing_jump): Check only_sets_cc0_p.
1815
1816 2007-01-30 Uros Bizjak <ubizjak@gmail.com>
1817
1818 * builtins.c (expand_builtin_int_interclass_roundingfn): New function
1819 to handle optabs that operate on floating point input argument and
1820 output to integer output.
1821 (expand_builtin_mathfn) [BUILT_IN_ILOGB]: Move from here ...
1822 (expand_builtin_interclass_mathfn) [BUILT_IN_ILOGB]: ... to here.
1823 (expand_builtin): Expand BUILT_IN_ILOGB{,F,L} using
1824 expand_builtin_interclass_mathfn ().
1825 * config/i386/i386.md (fxtractxf3_i387): Rename from *fxtractxf3_i387.
1826 (ilogbsi2): Remove.
1827 (ilogbxf2, ilogb<mode>2): New expanders to implement ilogb, ilogbf and
1828 ilogbl built-in functions as x87 intrinsics.
1829
1830 2007-01-30 Richard Guenther <rguenther@suse.de>
1831
1832 PR middle-end/27657
1833 * dwarf2out.c (reference_to_unused): Query varpool if the
1834 variable was output.
1835
1836 2007-01-30 Richard Guenther <rguenther@suse.de>
1837
1838 PR middle-end/30313
1839 * passes.c (execute_one_pass): Reset in_gimple_form to not
1840 confuse non-unit-at-a-time mode.
1841
1842 2007-01-29 Roger Sayle <roger@eyesopen.com>
1843 Richard Guenther <rguenther@suse.de>
1844
1845 * fold-const.c (round_up): Make HIGH an unsigned HOST_WIDE_INT to
1846 avoid undefined behaviour on overflow. Use force_fit_type_double
1847 to construct the constant with the specified TREE_OVERFLOW.
1848
1849 2007-01-29 Janis Johnson <janis187@us.ibm.com>
1850
1851 * config/dfp-bit.c: Add parameterized support for fp exceptions.
1852 * config/dfp-bit.h: Ditto.
1853
1854 2007-01-29 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
1855
1856 * c-decl.c (pop_scope): Replace warnings with call to
1857 warn_for_unused_label.
1858 * c-common.h (warn_for_unused_label): Declare.
1859 * c-common.c (warn_for_unused_label): Define.
1860
1861 2007-01-29 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
1862
1863 * tree-optimize.c (update_inlined_to_pointers): Delete unused
1864 function.
1865
1866 2007-01-29 Janis Johnson <janis187@us.ibm.com>
1867
1868 * Makefile.in (USER_H): Remove decfloat.h.
1869 * ginclude/decfloat.h: Delete, moving contents to ...
1870 * ginclude/float.h: Add support for decimal floating point,
1871 guarded by __STDC_WANT_DEC_FP__.
1872
1873 2007-01-29 Mike Stump <mrs@apple.com>
1874
1875 * doc/gccint.texi (Top): Rename Loop Representation to Loop
1876 Analysis and Representation to resolve case insensitive conflict.
1877 * doc/loop.texi (Loop Analysis and Representation): Likewise.
1878
1879 2007-01-28 Daniel Berlin <dberlin@dberlin.org>
1880
1881 * tree.h (struct tree_memory_tag): Add aliases member.
1882 (MTAG_ALIASES): New macro.
1883 * tree-ssa-alias.c (alias_bitmap_obstack): New variable.
1884 (add_may_alias): Remove pointer-set. Update for may_aliases being
1885 a bitmap.
1886 (mark_aliases_call_clobbered): Update for may_aliases being a
1887 bitmap.
1888 (compute_tag_properties): Ditto.
1889 (create_partition_for): Ditto.
1890 (compute_memory_partitions): Ditto.
1891 (dump_may_aliases_for): Ditto.
1892 (is_aliased_with): Ditto.
1893 (add_may_alias_for_new_tag): Ditto.
1894 (rewrite_alias_set_for): Rewrite for may_aliases being a bitmap.
1895 (compute_is_aliased): New function.
1896 (compute_may_aliases): Call compute_is_aliased).
1897 (init_alias_info): Initialize alias_bitmap_obstack.
1898 (union_alias_set_into): New function.
1899 (compute_flow_sensitive_aliasing): Use union_aliases_into.
1900 (have_common_aliases_p): Rewrite to take two bitmaps and use
1901 intersection.
1902 (compute_flow_insensitive_aliasing): Stop using pointer-sets.
1903 Update for bitmaps.
1904 (finalize_ref_all_pointers): Update for add_may_alias changes.
1905 (new_type_alias): Ditto.
1906 * tree-flow-inline.h (may_aliases): Return a bitmap.
1907 * tree-dfa.c (dump_variable): Check for MTAG_P'ness.
1908 * tree-ssa.c (verify_flow_insensitive_alias_info): Update for
1909 may_aliases being a bitmap.
1910 * tree-flow.h (struct var_ann_d): Remove may_aliases member.
1911 may_aliases now returns a bitmap.
1912 * tree-ssa-structalias.c (merge_smts_into): Update for may_aliases
1913 being a bitmap.
1914 * tree-ssa-operands.c (add_virtual_operand): Update for
1915 may_aliases being a bitmap.
1916
1917 2007-01-29 Daniel Berlin <dberlin@dberlin.org>
1918
1919 PR tree-optimization/30630
1920 * tree-ssa-structalias.c (do_complex_constraint): Mark correct
1921 variable as changed.
1922
1923 2007-01-29 Simon Martin <simartin@users.sourceforge.net>
1924
1925 PR c++/28266
1926 * gimplify.c (gimplify_target_expr): Make sure that the TARGET_EXPR is
1927 expanded only once even if an error occurs.
1928
1929 2007-01-29 Ben Elliston <bje@au.ibm.com>
1930
1931 * gcov-io.h (__gcov_indirect_call_profiler): Declare.
1932 (__gcov_average_profiler): Likewise.
1933 (__gcov_ior_profiler): Likewise.
1934 (__gcov_merge_ior): Likewise.
1935
1936 2007-01-28 Jan Hubicka <jh@suse.cz>
1937
1938 * builtins.c (expand_builtin_memset): Fix typo in my last patch.
1939 * value-prof.c (stringop_block_profile): Likewise.
1940
1941 2007-01-28 Jan Hubicka <jh@suse.cz>
1942
1943 * expr.c (emit_block_move_via_movmem, emit_block_move_via_libcall): Add
1944 variant handling histograms; add wrapper.
1945 (clear_storage_via_libcall): Export.
1946 (emit_block_move_hints): Break out from ...; add histograms.
1947 (emit_block_move): ... this one.
1948 (clear_storage_hints): Break out from ...; add histograms.
1949 (clear_storage): ... this one.
1950 (set_storage_via_memset): Handle histogram.
1951 * expr.h (emit_block_move_via_libcall, emit_block_move_hints): Declare.
1952 (clear_storage_hints, clear_storage_via_libcall): Declare.
1953 (set_storage_via_setmem): Update prototype.
1954 * doc/md.texi (movmem, setmem): Document new arguments.
1955
1956 * value-prof.c (dump_histogram_value, tree_find_values_to_profile): Add
1957 new histograms.
1958 (stringop_block_profile): New global function.
1959 (tree_stringops_values_to_profile): Profile block size and alignment.
1960 * value-prof.h (enum hist_type): add HIST_TYPE_AVERAGE and
1961 HIST_TYPE_IOR.
1962 (struct profile_hooks): Add gen_average_profiler and gen_ior_profiler.
1963 (stringop_block_profile): Declare.
1964 * builtins.c: Include value-prof.h.
1965 (expand_builtin_memcpy, expand_builtin_memset): Pass block profile.
1966 * gcov-ui.h (GCOV_COUNTER_NAMES): Add new counter.
1967 (GCOV_COUNTER_AVERAGE, GCOV_COUNTER_IOR): New constants.
1968 (GCOV_COUNTERS, GCOV_LAST_VALUE_COUNTER): Update.
1969 * profile.c (instrument_values): Add new counters.
1970 * cfgexpand.c (expand_gimple_basic_block): Propagate histograms to
1971 calls.
1972 * tree-profile.c (tree_average_profiler_fn, tree_ior_profiler_fn): New.
1973 (tree_init_edge_profiler): Build new profilers.
1974 (tree_gen_average_profiler, tree_gen_ior_profiler): New.
1975 (pass_tree_profile): Add dump.
1976 (tree_profile_hooks): Update.
1977 * Makefile.in (LIBGCOV): Add new constants.
1978 * libgcov.c (__gcov_merge_ior, __gcov_average_profiler,
1979 __gcov_ior_profiler): New.
1980 * i386.md (movmem/setmem expanders): Add new optional arguments.
1981
1982 2007-01-28 David Edelsohn <edelsohn@gnu.org>
1983
1984 * doc/md.texi (Standard Pattern Names): Document blockage pattern.
1985
1986 2007-01-28 Zdenek Dvorak <dvorakz@suse.cz>
1987
1988 * tree-ssa-loop-unswitch.c: Include tree-inline.h.
1989 (tree_unswitch_single_loop): Pass eni_size_weights to
1990 tree_num_loop_insns.
1991 * tree-ssa-loop-manip.c: Include tree-inline.h.
1992 (can_unroll_loop_p): Pass eni_size_weights to
1993 tree_num_loop_insns.
1994 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p):
1995 Pass eni_size_weights to estimate_num_insns.
1996 * tree.h (init_inline_once): Export.
1997 * toplev.c (backend_init): Call init_inline_once.
1998 * cgraphunit.c (cgraph_process_new_functions,
1999 cgraph_analyze_function): Pass eni_inlining_weights to
2000 estimate_num_insns.
2001 * ipa-inline.c (compute_inline_parameters): Ditto.
2002 * tree-ssa-loop-ivcanon.c (tree_num_loop_insns): Pass weights
2003 to estimate_num_insns.
2004 (try_unroll_loop_completely): Pass eni_size_weights to
2005 tree_num_loop_insns.
2006 * tree-eh.c (decide_copy_try_finally): Pass eni_size_weights
2007 ot estimate_num_insns.
2008 * tree-ssa-loop-prefetch.c: Include tree-inline.h.
2009 (loop_prefetch_arrays): Pass eni_time_weights to tree_num_loop_insns.
2010 * tree-inline.c (eni_inlining_weights, eni_size_weights,
2011 eni_time_weights): New variables.
2012 (init_inline_once): Initialize them.
2013 (struct eni_data): Mew.
2014 (estimate_num_insns_1, estimate_num_insns): Use weights.
2015 * tree-inline.h (struct eni_weights_d): New.
2016 (eni_inlining_weights, eni_size_weights, eni_time_weights): Declare.
2017 (estimate_num_insns): Declaration changed.
2018 * cfgloop.h (tree_num_loop_insns): Declaration changed.
2019 * Makefile.in (tree-ssa-loop-unswitch.o, tree-ssa-loop-prefetch.o,
2020 tree-ssa-loop-manip.o): Add TREE_INLINE_H dependency.
2021
2022 2007-01-28 Zdenek Dvorak <dvorakz@suse.cz>
2023
2024 * tree-data-ref.c (conflict_fn): Assert that the number of affine
2025 relations in the conflict function is valid.
2026
2027 2007-01-27 Ian Lance Taylor <iant@google.com>
2028
2029 * common.opt: Add fstrict-overflow.
2030 * opts.c (decode_options): Set flag_strict_overflow if -O2.
2031 * flags.h (TYPE_OVERFLOW_WRAPS): Define.
2032 (TYPE_OVERFLOW_UNDEFINED): Define.
2033 (TYPE_OVERFLOW_TRAPS): Define. This replaces TYPE_TRAP_SIGNED.
2034 Replace all uses.
2035 * tree.h (TYPE_TRAP_SIGNED): Don't define.
2036 * fold-const.c (negate_expr_p): Use TYPE_OVERFLOW_UNDEFINED.
2037 (fold_negate_expr): Likewise.
2038 (make_range): Likewise.
2039 (extract_muldiv_1): Likewise.
2040 (maybe_canonicalize_comparison): Likewise.
2041 (fold_comparison): Likewise.
2042 (fold_binary): Likewise.
2043 (tree_expr_nonnegative_p): Likewise.
2044 (tree_expr_nonzero_p): Likewise.
2045 * tree-vrp.c (compare_values): Likewise.
2046 (extract_range_from_binary_expr): Likewise.
2047 (extract_range_from_unary_expr): Likewise.
2048 * tree-ssa-loop-niter.c (infer_loop_bounds_from_signedness):
2049 Likewise.
2050 (nowrap_type_p): Likewise.
2051 * tree-scalar-evolution.c (simple_iv): Likewise.
2052 * fold-const.c (negate_expr_p): Use TYPE_OVERFLOW_WRAPS.
2053 (build_range_check): Likewise.
2054 (extract_muldiv_1): Likewise.
2055 (fold_comparison): Likewise.
2056 * tree-vrp.c (vrp_int_const_binop): Likewise.
2057 (extract_range_from_unary_expr): Likewise.
2058 * convert.c (convert_to_integer): Likewise.
2059 * fold-const.c (fold_negate_expr): Use TYPE_OVERFLOW_TRAPS.
2060 (fold_comparison): Likewise.
2061 (fold_binary): Likewise.
2062 * optabs.c (optab_for_tree_code): Likewise.
2063 * tree-vectorizer.c (vect_is_simple_reduction): Likewise.
2064 * simplify-rtx.c (simplify_const_relational_operation): Check
2065 flag_strict_overflow and flag_trapv.
2066 (simplify_const_relational_operation): Likewise.
2067 * doc/invoke.texi (Option Summary): Mention -fstrict-overflow.
2068 (Optimize Options): Add -fstrict-overflow to -O2 list. Document
2069 -fstrict-overflow.
2070
2071 2007-01-27 Roger Sayle <roger@eyesopen.com>
2072
2073 * tree.c (tree_fold_gcd): Delete.
2074 * tree.h (tree_fold_gcd): Remove prototype.
2075 * tree-data-ref.c (tree_fold_divides_p): Don't use tree_fold_gcd to
2076 test whether one constant integer is a multiple of another. Instead
2077 call int_const_binop with TRUNC_MOD_EXPR and test for a zero result.
2078 * fold-const.c (multiple_of_p): We've determined both TOP and
2079 BOTTOM are integer constants so we can call int_const_binop directly
2080 instead of the more generic const_binop.
2081
2082 2007-01-27 Roger Sayle <roger@eyesopen.com>
2083
2084 * fold-const.c (size_binop): In the fast-paths for X+0, 0+X, X-0 and
2085 1*X check that the constant hasn't overflowed, to preserve the
2086 TREE_OVERFLOW bit.
2087 (round_up): Provide an efficient implementation when rouding-up an
2088 INTEGER_CST to a power-of-two.
2089
2090 2007-01-28 Ralf Wildenhues <Ralf.Wildenhues@gmx.de>
2091
2092 * doc/sourcebuild.texi: Add comma for clarity.
2093 * doc/extend.texi: Fix some typos.
2094 * doc/passes.texi: Likewise.
2095 * doc/cppinternals.texi: Likewise.
2096 * doc/c-tree.texi: Likewise.
2097 * doc/tree-ssa.texi: Likewise.
2098 * doc/install.texi: Likewise.
2099
2100 2007-01-27 Jan Hubicka <jh@suse.cz>
2101
2102 * tree-sra.c (sra_walk_function): Don't rely on aliases being build.
2103 (pass_sra): Do not require alias information.
2104 * passes.c (init_optimization_passes): Add SRA
2105
2106 2007-01-27 Steven Bosscher <steven@gcc.gnu.org>
2107
2108 * tracer.c (rest_of_handle_tracer): We already cleaned
2109 up the CFG in tracer() so don't do it here again.
2110 * cfgcleanup.c (rest_of_handle_jump2): Don't repeat
2111 cleanup_cfg here, either. And don't call renumber_insns.
2112
2113 * cfgrtl.c (rtl_verify_flow_info_1): Don't verify that BB_END
2114 and BB_HEAD are in the insn stream here. Instead make sure
2115 that BB_INSN is valid on all insns. Also, do check here that
2116 there are no pending branch predictions...
2117 (rtl_verify_flow_info): ...instead of doing it here. Checks
2118 for BB_END and BB_HEAD moved from rtl_verify_flow_info_1 to
2119 here.
2120
2121 2007-01-26 Roger Sayle <roger@eyesopen.com>
2122
2123 * config/i386/i386.c (ix86_swap_binary_operands_p): New helper
2124 function to simplify/factorize operand order canonicalization.
2125 (ix86_fixup_binary_operands): Reorganize using the above function.
2126 (ix86_binary_operator_ok): Likewise.
2127
2128 2007-01-27 Jakub Jelinek <jakub@redhat.com>
2129
2130 * genattrtab.c (struct attr_value_list, insn_code_values): Move to
2131 file scope from optimize_attrs.
2132 (simplify_test_exp): If insn_code_values is not NULL, use it to speed
2133 up search.
2134 (optimize_attrs): Clear insn_code_values after freeing it.
2135
2136 2007-01-26 Zdenek Dvorak <dvorakz@suse.cz>
2137
2138 * tree-ssa-address.c (create_mem_ref): Remove ", bsi" from
2139 a parts.base assignment.
2140
2141 2007-01-26 Zdenek Dvorak <dvorakz@suse.cz>
2142
2143 * tree-data-ref.c (dump_subscript): Use dump_conflict_function.
2144 (compute_subscript_distance, initialize_data_dependence_relation,
2145 finalize_ddr_dependent, analyze_ziv_subscript,
2146 analyze_siv_subscript_cst_affine,
2147 compute_overlap_steps_for_affine_univar,
2148 compute_overlap_steps_for_affine_1_2, analyze_subscript_affine_affine,
2149 analyze_siv_subscript, analyze_miv_subscript,
2150 analyze_overlapping_iterations, subscript_dependence_tester_1,
2151 compute_self_dependence, free_dependence_relation): Work
2152 with affine_fn instead of chrecs.
2153 (dump_affine_function, dump_conflict_function, affine_function_equal_p,
2154 common_affine_function, affine_function_base,
2155 affine_function_constant_p, affine_fn_op, affine_fn_plus,
2156 affine_fn_minus, affine_fn_free, conflict_fn_not_known,
2157 conflict_fn_no_dependence, free_conflict_function, free_subscripts,
2158 conflict_fn, affine_fn_cst, affine_fn_univar): New functions.
2159 (all_chrecs_equal_p): Removed.
2160 * tree-data-ref.h (affine_fn, conflict_function): New types.
2161 (struct subscript): Change type of conflicting_iterations_in_a
2162 and conflicting_iterations_in_b.
2163
2164 2007-01-26 Steve Ellcey <sje@cup.hp.com>
2165
2166 PR other/30182
2167 * config/pa/pa.h (TARGET_HPUX_11): New.
2168 * config/pa/pa-hpux11.h (TARGET_HPUX_11): New.
2169 * config/pa/pa.c (pa_init_builtins): Use TARGET_HPUX_11.
2170
2171 2007-01-26 Daniel Berlin <dberlin@dberlin.org>
2172 Richard Guenther <rguenther@suse.de>
2173
2174 * tree-ssa-structalias.c (solve_graph): Handle case
2175 we merged the variable to another.
2176
2177 2007-01-25 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
2178
2179 * builtins.c (fold_builtin_1): Treat ccos and ccosh as 'even'
2180 functions.
2181
2182 * fold-const.c (negate_mathfn_p): Treat casin, casinh, catan,
2183 catanh, cproj, csin, csinh, ctan and ctanh as 'odd' functions.
2184
2185 2007-01-25 DJ Delorie <dj@redhat.com>
2186
2187 * config/m32c/m32c.c (m32c_cannot_change_mode_class): We don't
2188 allow changes to modes which don't fit in those registers.
2189
2190 * reload1.c (choose_reload_regs): Check for invalid subregs before
2191 computing their locations, not after.
2192
2193 2007-01-25 Geoffrey Keating <geoffk@apple.com>
2194
2195 PR 25127
2196 * config/rs6000/rs6000.c (first_altivec_reg_to_save): On Darwin,
2197 save Altivec registers in an eh_return function.
2198 (compute_vrsave_mask): Likewise.
2199 (rs6000_stack_info): Correct AIX/Darwin stack alignment computation
2200 for saving Altivec registers.
2201 (rs6000_emit_prologue): Don't allocate stack twice in
2202 eh_return function. Correct expected value of altivec_save_offset
2203 when using save_world. Describe save of R0 to stack when using
2204 save_world. Describe stack pointer adjustment when using
2205 save_world. Remove duplicated eh_return parameter register saving.
2206 Update sp_offset variable after save_world.
2207 * config/rs6000/t-darwin (LIB2FUNCS_STATIC_EXTRA): Remove
2208 darwin-world.asm.
2209 (LIB2FUNCS_EXTRA): Add darwin-world.asm.
2210 * config/rs6000/darwin.h (SUBTARGET_OVERRIDE_OPTIONS): -m64
2211 implies Altivec.
2212
2213 2007-01-25 Steve Ellcey <sje@cup.hp.com>
2214
2215 * config.gcc (ia64*-*-hpux*): Make posix threads the default.
2216
2217 2007-01-25 Steve Ellcey <sje@cup.hp.com>
2218
2219 PR other/30182
2220 * config/pa/pa.c (pa_init_builtins): Set asm names for finite routines.
2221 * config/ia64/ia64.c (ia64_init_builtins): Ditto.
2222
2223 2007-01-25 Richard Guenther <rguenther@suse.de>
2224
2225 * doc/invoke.texi (-Wcoverage-mismatch): Document.
2226 * common.opt (-Wcoverage-mismatch): New warning option.
2227 * coverage.c (get_coverage_counts): Ignore coverage mismatch
2228 if -Wcoverage-mismatch is given.
2229
2230 2007-01-25 Razya Ladelsky <razya@il.ibm.com>
2231
2232 * ipa-cp.c (ipcp_insert_stage, ipcp_driver): Support for SSA.
2233 (ipcp_driver): Change to static definition.
2234 Add dumping of the ifunctions.
2235 (constant_val_insert): Remove unused parameter. Support for SSA.
2236 (ipcp_propagate_const): Support for SSA.
2237 (ipcp_profile_bb_print): Print only analyzed nodes.
2238 (ipcp_replace_map_create): Remove support for Fortran constant
2239 for now.
2240 * ipa-prop.c (ipa_method_modify_stmt,
2241 ipa_callsite_compute_param): Support for SSA.
2242 * ipa-prop.h (ipcp_driver): Remove declaration.
2243 (IS_VALID_TREE_MAP_INDEX): Add define.
2244
2245 2007-01-24 Geoffrey Keating <geoffk@apple.com>
2246
2247 * unwind-dw2.c (execute_stack_op): Handle DW_OP_swap.
2248
2249 2007-01-24 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
2250
2251 PR middle-end/30447
2252 * builtins.c (fold_builtin_cabs): Use MPFR to evaluate a
2253 constant argument to cabs and do it without checking for
2254 -funsafe-math-optimizations.
2255
2256 2007-01-24 Douglas Gregor <dgregor@osl.iu.edu>
2257
2258 * c-common.h (RID_FIRST_CXX0X): New.
2259 (RID_LAST_CXX0X): New.
2260 * c-opts.c (c_common_handle_option): -Wc++0x-compat is triggered
2261 by -Wall.
2262 * c.opt (Wc++0x-compat): New.
2263 * doc/invoke.texi (-Wc++0x-compat): Document.
2264
2265 2007-01-24 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
2266
2267 * builtins.c (fold_builtin_carg): New.
2268 (fold_builtin_1): Use it.
2269
2270 2007-01-24 Jan Hubicka <jh@suse.cz>
2271
2272 * ipa-inline.c (cgraph_decide_inlining): Initialize initial_insns.
2273
2274 * ipa-inline.c (initial_insns, max_insns): Delete.
2275 (compute_max_insns): New function.
2276 (cgraph_decide_inlining_of_small_function): Use it; take minimal amount
2277 of insns as base for code growth.
2278 (cgraph_decide_inlining): Make initial_insns local; do not compute
2279 max_insns.
2280 * params.def (PARAM_INLINE_UNIT_GROWTH): Set to 60.
2281 * doc/invoke.texi (inline-unit-growth): Update docs.
2282
2283 2007-01-24 Jakub Jelinek <jakub@redhat.com>
2284
2285 * config/i386/i386.h (x86_cmpxchg16b): Remove const.
2286 (TARGET_CMPXCHG16B): Define to x86_cmpxchg16b.
2287 * config/i386/i386.c (x86_cmpxchg16b): Remove const.
2288 (override_options): Add PTA_CX16 flag. Set x86_cmpxchg16b
2289 for CPUs that have PTA_CX16 set.
2290
2291 PR middle-end/27416
2292 * gimplify.c (omp_check_private): New function.
2293 (gimplify_scan_omp_clauses): Use it for
2294 firstprivate/lastprivate/reduction.
2295
2296 PR middle-end/30494
2297 * gimplify.c (omp_add_variable): Don't call omp_notice_variable
2298 on TYPE_SIZE_UNIT for GOVD_LOCAL VLAs.
2299
2300 PR middle-end/30421
2301 * omp-low.c (lower_omp_for_lastprivate): Add dlist argument.
2302 If lower_lastprivate_clauses emits some statements, append them
2303 to dlist rather than body_p and to body_p append an initializer.
2304 (lower_omp_for): Adjust caller.
2305
2306 2007-01-24 Steve Ellcey <sje@cup.hp.com>
2307
2308 * target.h (globalize_decl_name): New.
2309 * target-def.h (TARGET_ASM_GLOBALIZE_DECL_NAME): New.
2310 * output.h (default_globalize_decl_name): New.
2311 * varasm.c (asm_output_bss): Use globalize_decl_name instead of
2312 globalize_label.
2313 (globalize_decl): Ditto.
2314 (default_globalize_decl_name): New.
2315 * config/ia64/ia64.c (ia64_globalize_decl_name): New.
2316 (ia64_handle_version_id_attribute): New.
2317 (TARGET_ASM_GLOBALIZE_DECL_NAME): New.
2318 (ia64_asm_output_external): Use globalize_decl_name instead
2319 of globalize_label.
2320 * doc/extend.texi (version_id): New pragma.
2321 * doc/tm.texi (ARGET_ASM_GLOBALIZE_DECL_NAME): New target hook.
2322
2323 2007-01-24 Andreas Krebbel <krebbel1@de.ibm.com>
2324
2325 * unwind-dw2-fde.c (get_cie_encoding): Replaced _Unwind_Word with
2326 _uleb128_t and _Unwind_SWord with _sleb128_t.
2327 * unwind-dw2.c (extract_cie_info, execute_stack_op, execute_cfa_program,
2328 uw_frame_state_for, uw_update_context_1): Likewise.
2329 * unwind-c.c (parse_lsda_header, PERSONALITY_FUNCTION): Likewise.
2330 * unwind-pe.h (read_uleb128, read_sleb128,
2331 read_encoded_value_with_base): Likewise.
2332 * unwind-generic.h: Define _sleb128_t and _uleb128_t types.
2333
2334 2007-01-24 Richard Guenther <rguenther@suse.de>
2335
2336 * builtins.c (expand_builtin_cexpi): Get the fndecl
2337 for cexp in the correct way.
2338
2339 2007-01-24 Jan Hubicka <jh@suse.cz>
2340
2341 * tree-ssa-dce.c (eliminate_unnecesary_stmts): Remove dead LHS of calls.
2342
2343 2007-01-24 Andreas Krebbel <krebbel1@de.ibm.com>
2344
2345 * c-cppbuiltin.c (builtin_define_type_sizeof): New function.
2346 (c_cpp_builtins): New builtin macros: __SIZEOF_INT__, __SIZEOF_LONG__,
2347 __SIZEOF_LONG_LONG__, __SIZEOF_SHORT__, __SIZEOF_POINTER__,
2348 __SIZEOF_FLOAT__, __SIZEOF_DOUBLE__, __SIZEOF_LONG_DOUBLE__,
2349 __SIZEOF_SIZE_T__, __SIZEOF_WCHAR_T__, __SIZEOF_WINT_T__ and
2350 __SIZEOF_PTRDIFF_T__.
2351 * doc/cpp.texi: Documentation for the new builtin macros added.
2352
2353 2007-01-24 Uros Bizjak <ubizjak@gmail.com>
2354
2355 * config/i386/i386.md (tanxf2, tan<mode>2, atan<mode>2, log<mode>2,
2356 log10<mode>2, log2<mode>2, expxf2, exp10xf2, exp2xf2): Use op2
2357 instead of operands[2] to avoid access past the end of array.
2358
2359 2007-01-24 Richard Sandiford <richard@codesourcery.com>
2360
2361 * reload1.c (emit_reload_insns): Pass the reload register
2362 for a non-spill output reload through forget_old_reloads_1.
2363
2364 2007-01-23 Joseph Myers <joseph@codesourcery.com>
2365
2366 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Define
2367 _SOFT_DOUBLE if doubles use software floating-point.
2368 * config/rs6000/libgcc-ppc-glibc.ver: Export additional long
2369 double functions if _SOFT_DOUBLE, not _SOFT_FLOAT.
2370 * config/rs6000/darwin-ldouble.c: Also compile functions for
2371 hard-float without FPRs. Use fmsub function for all __NO_FPRS__
2372 cases. Compile extra functions if _SOFT_DOUBLE, not _SOFT_FLOAT.
2373 * config/rs6000/linuxspe.h (SUBSUBTARGET_OVERRIDE_OPTIONS): Remove
2374 commented-out long double override.
2375 (CPP_LONGDOUBLE_DEFAULT_SPEC): Likewise.
2376 * config/rs6000/eabispe.h: Likewise.
2377 * config/rs6000/rs6000.c (rs6000_override_options): Don't override
2378 long double for non-SPE.
2379 (rs6000_handle_option): Likewise.
2380 (invalid_e500_subreg): Disallow more subregs involding DImode,
2381 DFmode, TImode or TFmode.
2382 (rs6000_legitimate_offset_address_p): Check TFmode offsets for
2383 E500 double.
2384 (legitimate_lo_sum_address_p): Also check for TFmode for E500
2385 double.
2386 (rs6000_legitimize_address): Also handle TFmode for E500 double.
2387 (rs6000_legitimize_reload_address): Also handle TFmode for E500
2388 double.
2389 (rs6000_legitimate_address): Also check for TFmode for E500
2390 double.
2391 (rs6000_emit_move): Use DFmode subregs of TFmode for E500 double.
2392 (spe_build_register_parallel): Handle TFmode and TCmode.
2393 (rs6000_spe_function_arg): Handle TFmode and TCmode for E500
2394 double.
2395 (function_arg): Handle TFmode and TCmode for E500 double.
2396 (rs6000_init_libfuncs): Initialize extra libfuncs for soft double
2397 in general.
2398 (print_operand): Handle TFmode and TImode for %y.
2399 (rs6000_generate_compare): Handle TFmode comparisons for E500
2400 double.
2401 (spe_func_has_64bit_regs_p): Check for TFmode for E500 double.
2402 (rs6000_function_value): Handle TFmode and TCmode for E500 double.
2403 (rs6000_libcall_value): Handle TFmode and TCmode for E500 double.
2404 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Check for
2405 TFmode for E500 double.
2406 * config/rs6000/rs6000.md (FP): Allow TF for E500 double.
2407 (floatsidf2): Enable for E500 double.
2408 (movtf_softfloat): Use rs6000_nonimmediate_operand.
2409 (extenddftf2): Change to extenddftf2_fprs.
2410 (extenddftf2): Call gen_spe_extenddftf2 or gen_extenddftf2_fprs
2411 depending on TARGET_E500_DOUBLE.
2412 (extendsftf2): Enable for E500 double.
2413 (trunctfdf2): Enable for E500 double.
2414 (trunctfsf2): Change to trunctfsf2_fprs.
2415 (trunctfsf2): Call gen_spe_trunctfsf2 or gen_trunctfsf2_fprs
2416 depending on TARGET_E500_DOUBLE.
2417 (floatsitf2): Enable for E500 double.
2418 (fix_trunctfsi2): Change to fix_trunctfsi2_fprs.
2419 (fix_trunctfsi2): Call gen_spe_fix_trunctfsi2 or
2420 gen_fix_trunctfsi2_fprs depending on TARGET_E500_DOUBLE.
2421 (negtf2): Change to negtf2_internal.
2422 (negtf2): New expander.
2423 (abstf2): Enable for E500 double. Call gen_spe_abstf2_tst,
2424 gen_spe_abstf2_cmp or gen_abstf2_internal depending on
2425 TARGET_E500_DOUBLE and flag_unsafe_math_optimizations.
2426 (movdi_internal32): Use rs6000_nonimmediate_operand.
2427 (unnamed splitter): Likewise.
2428 * config/rs6000/spe.md (CMPTFEQ_GPR, TSTTFEQ_GPR, CMPTFGT_GPR,
2429 TSTTFGT_GPR, CMPTFLT_GPR, TSTTFLT_GPR): New unspecs.
2430 (SPE64TF, DITI): New mode macros.
2431 (frob_df_di): Change to frob_<SPE64:mode>_<DITI:mode>; allow more
2432 modes.
2433 (frob_tf_ti): New.
2434 (frob_<mode>_di_2): New.
2435 (frob_tf_di_8_2): New.
2436 (frob_di_df): Change to frob_di_<mode>; allow more modes.
2437 (frob_ti_tf): New.
2438 (frob_di_df_2): Change to frob_<DITI:mode>_<SPE64:mode>_2; allow
2439 more modes.
2440 (frob_ti_<mode>_8_2): New.
2441 (frob_ti_tf_2): New.
2442 (mov_si<mode>_e500_subreg0, mov_si<mode>_e500_subreg0_2,
2443 mov_si<mode>_e500_subreg4, mov_si<mode>_e500_subreg4_2): Allow
2444 TFmode.
2445 (mov_sitf_e500_subreg8, mov_sitf_e500_subreg8_2,
2446 mov_sitf_e500_subreg12, mov_sitf_e500_subreg12_2): New.
2447 (spe_trunctfdf2_internal1, spe_trunctfsf2, spe_extenddftf2,
2448 spe_fix_trunctfsi2, spe_fix_trunctfsi2_internal,
2449 spe_negtf2_internal, spe_abstf2_cmp, spe_abstf2_tst): New.
2450 (cmptfeq_gpr, tsttfeq_gpr, cmptfgt_gpr, tsttfgt_gpr, cmptflt_gpr,
2451 tsttflt_gp): New.
2452
2453 2007-01-23 Ian Lance Taylor <iant@google.com>
2454
2455 * Makefile.in (OBJS-common): Reformat, alphabetize, but put
2456 insn-*.o first.
2457 (OBJS-archive): Reformat, alphabetize.
2458 (OBJS): Change out_object_file to OBJS-md.
2459
2460 2007-01-23 Uros Bizjak <ubizjak@gmail.com>
2461
2462 * config/i386/i386.md: Use REG_P, MEM_P, CONST_INT_P, LABEL_P,
2463 JUMP_P and CALL_P predicates where applicable.
2464 * config/i386/i386.c: Ditto.
2465 * config/i386/i386.md: Ditto.
2466 * config/i386/mmx.md: Ditto.
2467 * config/i386/predicates.md: Ditto.
2468
2469 2007-01-22 Andreas Schwab <schwab@suse.de>
2470
2471 * config/m68k/m68k.h: Fix comment.
2472
2473 2007-01-22 Jan Hubicka <jh@suse.cz>
2474
2475 * passes.c (init_optimization_passes): Do not rerun
2476 pass_early_warn_uninitialized.
2477
2478 2007-01-22 Richard Guenther <rguenther@suse.de>
2479
2480 PR tree-optimization/30038
2481 * tree-ssa-math-opts.c (maybe_record_sincos): New static helper
2482 function.
2483 (execute_cse_sincos_1): Likewise.
2484 (execute_cse_sincos): Likewise.
2485 (gate_cse_sincos): Likewise.
2486 (pass_cse_sincos): New pass CSEing sin() and cos() calls using
2487 the cexpi() canonicalization of sincos().
2488 * tree-pass.h (pass_cse_sincos): Declare.
2489 * passes.c (init_optimization_passes): New pass pas_cse_sincos.
2490
2491 2007-01-21 Eric Botcazou <ebotcazou@libertysurf.fr>
2492
2493 PR rtl-optimization/29329
2494 * combine.c (replaced_rhs_insn): Rename to i2mod.
2495 (replaced_rhs_value): Rename to i2mod_new_rhs.
2496 (i2mod_old_rhs): New global variable.
2497 (combine_instructions): Adjust for above change. Save a copy of
2498 the old RHS into i2mod_old_rhs when the contents of a REG_EQUAL
2499 note are substituted in the second instruction.
2500 (distribute_notes) <REG_DEAD>: Adjust for above change. Do not
2501 ditch the note if it pertains to the second eliminated register
2502 and this register is mentioned in i2mod_old_rhs.
2503
2504 Revert:
2505 2006-09-12 Eric Botcazou <ebotcazou@libertysurf.fr>
2506
2507 * combine.c (distribute_notes) <REG_DEAD>: Do not consider SETs past
2508 the insn to which the note was originally attached.
2509
2510 2007-01-21 Jan Hubicka <jh@suse.cz>
2511
2512 * ipa-inline.c (inlining_mode): Comment, move up.
2513 (cgraph_decide_inlining_incrementally): Do not perform inlining
2514 itself; fix handling of flattening of self recursive functions.
2515 (cgraph_find_cycles): Remove.
2516 (cgraph_flatten_node): Remove.
2517 (cgraph_decide_inlining): Use incremental inliner to handle flattening.
2518 (try_inline): New function.
2519 (cgraph_early_inlining): Update call of
2520 cgraph_decide_inlining_incrementally. Apply inlining here.
2521 (apply_inline): Update call of cgraph_decide_inlining_incrementally.
2522
2523 2007-01-21 Dirk Mueller <dmueller@suse.de>
2524
2525 PR bootstrap/30511
2526 * tree-vrp.c (check_array_bounds): do not warn
2527 about ADDR_EXPR's of ARRAY_REF's which are immediately
2528 used in binary expressions.
2529
2530 2007-01-21 Ira Rosen <irar@il.ibm.com>
2531
2532 * tree-vectorizer.h (struct _stmt_vec_info): Add new field
2533 read_write_dep and macros for its access.
2534 * tree-vectorizer.c (new_stmt_vec_info): Initialize the new field.
2535 * tree-vect-analyze.c (vect_analyze_data_ref_dependence): Remove
2536 argument, call vect_check_interleaving for every independent pair of
2537 data-refs. Mark loads that access the same memory location as a store
2538 in the loop.
2539 (vect_check_dependences): Remove.
2540 (vect_analyze_data_ref_dependences): Remove vect_check_dependences
2541 call, fix the call to vect_analyze_data_ref_dependence.
2542 (vect_analyze_data_ref_access): For statements that access the same
2543 data-ref, check that they are not stores; for loads, check that there
2544 is no store that access the same location.
2545
2546 2007-01-20 Roger Sayle <roger@eyesopen.com>
2547 Joseph Myers <joseph@codesourcery.com>
2548
2549 * doc/invoke.texi (-fdump-rtl-combine): Fix under/overfull hbox.
2550 (-fdump-rtl-stack): Likewise.
2551 (-fno-signed-zeros): Use @minus{} for a minus sign. Correct typo.
2552 (-mcheck-zero-division, -mcpu): Fix under/overfull hbox.
2553 (-mpt-fixed): Use @minus{} for minus sign.
2554 (Using Precompiled Headers): Fix under/overfull hbox.
2555
2556 2007-01-20 Jan Hubicka <jh@suse.cz>
2557
2558 * tree-flow.h (struct stmt_ann_d): Move references_memory to proper
2559 place within annotation.
2560
2561 2007-01-20 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
2562
2563 * pa.c (output_move_double): Change array size of xoperands to 4.
2564
2565 2007-01-20 Jan Hubicka <jh@suse.cz>
2566
2567 * tree-tailcall.c (adjust_return_value): Do not use RESULT_DECL
2568 as temporary.
2569
2570 2007-01-19 Ian Lance Taylor <iant@google.com>
2571
2572 * expmed.c (expand_divmod) [TRUNC_MOD_EXPR, TRUNC_DIV_EXPR]: Cast
2573 constant to unsigned HOST_WIDE_INT before negating.
2574
2575 2007-01-19 Ian Lance Taylor <iant@google.com>
2576
2577 * tree-ssa-operands.h (struct vuse_vec_d): Change num_vuse field
2578 to unsigned.
2579 (VUSE_VECT_ELEMENT) [ENABLE_CHECKING]: Use unsigned comparison.
2580 (VUSE_ELEMENT_PTR) [ENABLE_CHECKING]: Likewise.
2581 (SET_VUSE_VECT_ELEMENT) [ENABLE_CHECKING]: Likewise.
2582 (SET_VUSE_ELEMENT_VAR) [ENABLE_CHECKING]: Likewise.
2583 (SET_VUSE_ELEMENT_PTR) [ENABLE_CHECKING]: Likewise.
2584 (realloc_vdef, realloc_vuse): Change second parameter to
2585 unsigned.
2586 (ssa_operand_iterator_d): Change vuse_index and mayuse_index
2587 fields to unsigned.
2588 * tree-ssa-operands.c (realloc_vop): Change num_elem parameter to
2589 unsigned. Change x and lim locals to unsigned.
2590 (realloc_vdef, realloc_vuse): Change num_elem parameter to
2591 unsigned.
2592 (finalize_ssa_vuse_ops): Change old_i local to unsigned.
2593 (copy_virtual_operands): Change i and n locals to unsigned.
2594
2595 2007-01-19 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
2596
2597 PR middle-end/29335
2598 * builtins.c (fold_builtin_1): Handle builtin fdim.
2599
2600 2007-01-20 Jan Hubicka <jh@suse.cz>
2601
2602 * tree-ssa.c (init_tree_ssa): Do not call init_alias_heapvars.
2603 * tree-ssa-structalias.c (compute_points_to_sets): Do call
2604 init_alias_heapvars.
2605 (init_alias_heapvars): Initialize only when not already initialized.
2606 (delete_alias_heapvars): Set heapvar_for_stmt to NULL.
2607
2608 2007-01-19 Roger Sayle <roger@eyesopen.com>
2609
2610 * common.opt (fsigned-zeros): New command line option.
2611 * flags.h (HONOR_SIGNED_ZEROS): Control via flag_signed_zeros instead
2612 of flag_unsafe_math_optimizations.
2613 * opts.c (set_fast_math_flags): The -ffast-math command line option
2614 implies -fno-signed-zeros.
2615 (fast_math_flags_set_p): Likewise.
2616
2617 * doc/invoke.texi: Document new -fno-signed-zeros option, and update
2618 the documentation of -ffast-math appropriately. Wrap long lines.
2619
2620 2007-01-19 Steve Ellcey <sje@cup.hp.com>
2621
2622 * system.h (ASM_MAKE_LABEL_LINKONCE): Poison.
2623 * varasm.c (globalize_decl): Remove ASM_MAKE_LABEL_LINKONCE ifdef.
2624
2625 2007-01-19 Tomas Bily <tbily@suse.cz>
2626
2627 * cgraphunit.c (cgraph_finalize_function): Updating of pid
2628 * tree-profile.c:
2629 (tree_init_ic_make_global_vars): New function
2630 (tree_init_edge_profiler): call of tree_init_ic_make_global_vars
2631 (tree_gen_ic_profiler): New function
2632 (tree_gen_ic_func_profiler): New function
2633 (tree_profiling): Added calling of tree_gen_ic_func_profiler
2634 (tree_profile_hooks): Added hook for indirec/virtual calls
2635 * value-prof.c (tree_find_values_to_profile): New case for
2636 indirect calls
2637 (tree_values_to_profile): Call for determining indirect/virtual
2638 counters
2639 (tree_indirect_call_to_profile): New function
2640 (tree_ic_transform): New function
2641 (tree_ic): New function
2642 (find_func_by_pid): New function
2643 (init_pid_map): New function
2644 (tree_value_profile_transformations): Added check for
2645 indirect/virtual call transformation
2646 * value-prof.h (enum hist_type): New counter type for
2647 indirect/virtual calls
2648 (profile_hooks): Added new hook for profiling indirect/virtual
2649 calls
2650 * profile.c (instrument_values): New case for indirect/virtual
2651 call added
2652 * gcov-io.h (GCOV_LAST_VALUE_COUNTER): Changed to 6
2653 (GCOV_COUNTER_V_INDIR): New counter type
2654 (GCOV_COUNTER_NAMES): New name of counter "indirect" added
2655 (GCOV_MERGE_FUNCTIONS): New merge function for indirect/virtual
2656 call added
2657 * cgraph.c: Definition of cgraph_max_pid
2658 (cgraph_create_node): Default init of pid attribute
2659 * cgraph.h: Declaration of cgraph_max_pid
2660 (struct cgraph_node): Added pid attribute
2661 * libgcov.c (__gcov_indirect_call_profiler): New function
2662 (__gcov_one_value_profiler_body): New function
2663 (__gcov_one_value_profiler): Body was moved to
2664 __gcov_one_value_profiler_body and calls it
2665
2666 2007-01-19 Basile Starynkevitch <basile@starynkevitch.net>
2667
2668 * doc/gty.texi (Options): Document the mark_hook option to GTY.
2669 * gengtype.c (write_types_data, write_func_for_structure,
2670 write_types, ggc_wtd, pch_wtd): Add skip_hooks to
2671 write_types_data, ggc_wtd, pch_wtd for processing mark_hook.
2672 (walk_type, write_func_for_structure): Generate the mark_hook if
2673 needed.
2674
2675 2007-01-19 Jan Hubicka <jh@suse.cz>
2676
2677 * ipa-inline.c (cgraph_decide_inlining_incrementally): Instead of
2678 'early' argument take inlining mode argument specifying whether to
2679 inline for size/speeed or all functions; add support for flattening;
2680 improve dumpting.
2681 (cgraph_early_inlining): Update call of decide_inlining_incrementally.
2682
2683 2007-01-19 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
2684
2685 PR c++/17947
2686 * toplev.c (warn_deprecated_use): Use %qD instead of %qs to print
2687 the name of the declared identifier.
2688
2689 2007-01-19 Dirk Mueller <dmueller@suse.de>
2690
2691 * config/i386.h (CONDITIONAL_REGISTER_USAGE): Store
2692 result of PIC_OFFSET_TABLE_REGNUM in temporary variable to avoid
2693 duplicate evaluation.
2694
2695 2007-01-19 Uros Bizjak <ubizjak@gmail.com>
2696
2697 * config/i386/i386.md (acos<mode>2): Rename from acossf2 and acosdf2.
2698 Macroize expander using X87MODEF12 mode macro. Extend operand 1
2699 to XFMode, use acosxf2 and truncate result to requested mode.
2700 Use SSE_FLOAT_MODE_P to disable patterns for SSE math.
2701 (asin<mode>2): Similarly, with asin expanders.
2702 (*fscalexf4_i387): Rename from *fscalexf4.
2703 (expNcorexf3): New expander.
2704 (expxf2, exp10xf2, exp2xf2): Use expNcorexf3 expander.
2705 (exp<mode>2): Rename from expsf2 and expdf2. Macroize expander using
2706 X87MODEF12 mode macro. Extend operand 1 to XFMode, use expxf2 and
2707 truncate result to requested mode. Use SSE_FLOAT_MODE_P to disable
2708 patterns for SSE math.
2709 (exp10<mode>2): Similarly, with exp10 expanders.
2710 (exp2<mode>2): Similarly, with exp2 expanders.
2711 (expm1<mode>2): Similarly, with expm1 expanders.
2712 (ldexp<mode>3): Similarly, with ldexp expanders.
2713 (log<mode>2, log10<mode>2, log2<mode>2, log1p<mode>2, logb<mode>2):
2714 Use gen_truncxf<mode>2_i387_noop to truncate result.
2715
2716 2007-01-19 Richard Sandiford <richard@codesourcery.com>
2717
2718 * config/m68k/m68k.h (PREFERRED_STACK_BOUNDARY): Define to 32
2719 for ColdFire targets.
2720
2721 2007-01-19 Nathan Sidwell <nathan@codesourcery.com>
2722 Richard Sandiford <richard@codesourcery.com>
2723
2724 * config/m68k/m68k.h (M68K_STATIC_CHAIN_REG_NAME): New macro.
2725 (INITIALIZE_TRAMPOLINE): Use STATIC_CHAIN_REGNUM.
2726 (__transfer_from_trampoline): Use M68K_STATIC_CHAIN_REG_NAME.
2727 * config/m68k/m68kelf.h (STATIC_CHAIN_REGNUM): Override.
2728 (M68K_STATIC_CHAIN_REG_NAME): Likewise.
2729 * config/m68k/netbsd-elf.h (M68K_STATIC_CHAIN_REG_NAME): Likewise.
2730
2731 2007-01-19 Richard Sandiford <richard@codesourcery.com>
2732
2733 * config/m68k/m68k.md (adddi_dilshr32): Rename to...
2734 (*adddi_dilshr32): ...this. Fix formatting. Remove commented-out
2735 non-canonical pattern. Restrict to !TARGET_COLDFIRE.
2736 (*adddi_dilshr32_cf): New pattern.
2737 (adddi3, subdi3): Remove first alternatives.
2738
2739 2007-01-19 Richard Sandiford <richard@codesourcery.com>
2740
2741 * config/m68k/m68k.c (notice_update_cc): If an SFmode move is
2742 implemented using move.l, do not use its cc result for floating-point
2743 comparisons.
2744
2745 2007-01-19 Richard Sandiford <richard@codesourcery.com>
2746
2747 * config/m68k/m68k.h (EXTRA_CONSTRAINT): Stop the 'T' constraint
2748 from accepting 's' constraints if flag_pic.
2749
2750 2007-01-19 Richard Sandiford <richard@codesourcery.com>
2751
2752 * config/m68k/m68k.md (bordered, bunordered, buneq, bunge, bungt)
2753 (bunle, bunlt, bltgt, bordered_rev, bunordered_rev, buneq_rev)
2754 (bunge_rev, bungt_rev, bunle_rev, bunlt_rev, bltgt_rev): Change
2755 condition from TARGET_68881 to TARGET_HARD_FLOAT.
2756
2757 2007-01-19 Sandra Loosemore <sandra@codesourcery.com>
2758
2759 * longlong.h (count_leading_zeros, COUNT_LEADING_ZEROS_0): Add
2760 ColdFire alternatives.
2761 * config/m68k/m68k.h (CLZ_DEFINED_VALUE_AT_ZERO): New macro.
2762 * config/m68k/m68k.md (clzsi2): Define for ColdFire
2763 architectures that support ff1 instruction.
2764
2765 2007-01-19 Richard Sandiford <richard@codesourcery.com>
2766 Julian Brown <julian@codesourcery.com>
2767
2768 * config/m68k/m68k.h (CONST_OK_FOR_LETTER_P): Add an 'R' case.
2769 * config/m68k/m68k.md (*movsi_cfv4): Fold into...
2770 (*movsi_cf): ...here. Remove unnecessary 'R' from 'Rg'.
2771 Add commentary.
2772
2773 2007-01-19 Richard Sandiford <richard@codesourcery.com>
2774
2775 * config/m68k/m68k-protos.h (valid_mov3q_const): Take a HOST_WIDE_INT
2776 and return a bool.
2777 (output_move_const_into_data_reg, output_move_simode_const): Delete.
2778 * config/m68k/m68k.c (const_method, const_int_cost): Take a
2779 HOST_WIDE_INT instead of an rtx.
2780 (m68k_rtx_costs): Update call accordingly.
2781 (output_move_const_into_data_reg): Likewise. Fix formatting.
2782 (valid_mov3q_const): Take a HOST_WIDE_INT instead of an rtx.
2783 Return a bool.
2784 (output_move_simode_const): Update calls after above changes.
2785 Rework to use automatic variables and predicates like MEM_P.
2786 * config/m68k/m68k.md (pushexthisi_const): Update call to
2787 valid_mov3q_const.
2788
2789 2007-01-19 Dirk Mueller <dmueller@suse.de>
2790
2791 * tree-ssa-alias.c (perform_var_substitution): Fix typo
2792 in dump_flags test.
2793
2794 2007-01-19 Richard Guenther <rguenther@suse.de>
2795
2796 * builtins.c (expand_builtin_cexpi): Fall back to expanding
2797 via cexp in case sincos is not available.
2798
2799 2007-01-19 Richard Guenther <rguenther@suse.de>
2800
2801 * doc/tm.texi (TARGET_HAS_SINCOS): Document new target macro.
2802 * defaults.h (TARGET_HAS_SINCOS): Default to off.
2803 * config/linux.h (TARGET_HAS_SINCOS): Set to on if we have glibc.
2804 * config/alpha/linux.h (TARGET_HAS_SINCOS): Likewise.
2805 * config/sparc/linux.h (TARGET_HAS_SINCOS): Likewise.
2806 * config/sparc/linux64.h (TARGET_HAS_SINCOS): Likewise.
2807 * config/rs6000/linux.h (TARGET_HAS_SINCOS): Likewise.
2808 * config/rs6000/linux64.h (TARGET_HAS_SINCOS): Likewise.
2809
2810 2007-01-19 Uros Bizjak <ubizjak@gmail.com>
2811
2812 * config/i386/i386.md (*fpatanxf3_i387, fpatan_extend<mode>xf3_i387):
2813 New insn patterns.
2814 (atan2sf3_1, atan2df3_1, atan2xf3_1): Remove insn patterns.
2815 (atan2xf3): Directly generate RTL pattern.
2816 (atan2<mode>3): Rename from atan2sf3 and atan2df3 and macroize insn
2817 patterns using X87MODEF12 mode macro. Use fpatan_extend<mode>xf3_i387
2818 and truncate result to requested mode. Use SSE_FLOAT_MODE_P to
2819 disable patterns for SSE math.
2820 (atan<mode>2): Rename from atansf2 and atandf2 and macroize insn
2821 patterns using X87MODEF12 mode macro. Use fpatan_extend<mode>xf3_i387
2822 and truncate result to requested mode. Use SSE_FLOAT_MODE_P to
2823 disable patterns for SSE math.
2824
2825 2007-01-19 Alexandre Oliva <aoliva@redhat.com>
2826
2827 * libgcc-std.ver: Fix typo in %inherit for GCC_4.3.0.
2828
2829 2007-01-18 Roger Sayle <roger@eyesopen.com>
2830
2831 * fold-const.c (fold_unary) <VIEW_CONVERT_EXPR>: Optimize away a
2832 VIEW_CONVERT_EXPR to the same type as it's operand.
2833
2834 2007-01-18 David Edelsohn <edelsohn@gnu.org>
2835
2836 * config/rs6000/darwin-ldouble.c: Only build _SOFT_FLOAT if
2837 configured for long double 128.
2838
2839 2007-01-18 Mike Stump <mrs@apple.com>
2840
2841 * config/rs6000/rs6000.c (rs6000_emit_vector_compare): Fix build
2842 error.
2843
2844 2007-01-18 Michael Meissner <michael.meissner@amd.com>
2845
2846 * i386.c (ix86_compute_frame_layout): Make fprintf's in #if 0 code
2847 type correct.
2848
2849 2007-01-18 Jan Hubicka <jh@suse.cz>
2850
2851 * tree-ssa-operands.c (vop_free_bucket_size): Never return value
2852 greater than NUM_VOP_FREE_BUCKETS.
2853
2854 2007-01-18 Daniel Berlin <dberlin@dberlin.org>
2855
2856 * tree-ssa-structalias.c: Update comments.
2857 (ptabitmap_obstack): Removed.
2858 (pta_obstack): New.
2859 (oldpta_obstack): Ditto.
2860 (stats): Add a few members.
2861 (struct variable_info): Remove node, complex, address_taken, and
2862 indirect_target members. Add oldsolution member.
2863 (new_var_info): Do not initialize removed members.
2864 (constraint_expr_type): Remove INCLUDES.
2865 (constraint_graph): Add size, implicit_preds, rep,
2866 indirect_cycles, eq_rep, label, direct_nodes, and complex members.
2867 (FIRST_REF_NODE): New macro.
2868 (LAST_REF_NODE): Ditto.
2869 (FIRST_ADDR_NODE): Ditto.
2870 (find): New function.
2871 (unite): Ditto.
2872 (dump_constraint): Do not handle INCLUDES.
2873 (insert_into_complex): Do not insert duplicate constraints.
2874 (condense_varmap_nodes): Renamed and rewritten into ...
2875 (merge_node_constraints): This. Also fix bug in handling of
2876 offseted copy constraints.
2877 (clear_edges_for_node): No longer need to deal with preds at all,
2878 or removing associated preds/succs.
2879 (merge_graph_nodes): Deal with indirect_cycles.
2880 Don't deal with predecessors.
2881 (add_implicit_graph_edge): New function.
2882 (add_pred_graph_edge): Ditto.
2883 (add_graph_edge): Don't deal with predecessors.
2884 (build_constraint_graph): Removed.
2885 (build_pred_graph): New function.
2886 (build_succ_graph): Ditto.
2887 (struct scc_info): Removed in_component. Added roots, dfs, and
2888 node_mapping. Remove visited_index, unification_queue.
2889 (scc_visit): Deal with union-find we do now.
2890 Deal with cycles with REF nodes.
2891 (collapse_nodes): Renamed and rewritten to ...
2892 (unify_nodes): This.
2893 (process_unification_queue): Removed.
2894 (topo_visit): Cleanup
2895 (do_da_constraint): Use find.
2896 (do_sd_constraint): Ditto.
2897 (do_ds_constraint): Ditto.
2898 (do_complex_constraint): Ditto.
2899 (init_scc_info): Update for removed and added members.
2900 (find_and_collapse_graph_cycles): Renamed and rewritten into ...
2901 (find_indirect_cycles): This.
2902 (equivalence_class): New variable.
2903 (label_visit): New function.
2904 (perform_variable_substitution): Rewritten.
2905 (free_var_substitution_info): New function.
2906 (find_equivalent_node): Ditto.
2907 (move_complex_constraints): Ditto.
2908 (eliminate_indirect_cycles): Ditto.
2909 (solve_graph): Only propagate changed bits.
2910 Use indirect cycle elimination.
2911 Use find.
2912 (tree_id_t): Rename to tree_vi_t, delete id member, add vi member.
2913 (tree_id_eq): Renamed to ...
2914 (tree_vi_eq): This. Update for member change
2915 (insert_id_for_tree): Renamed and rewritten to ...
2916 (insert_vi_for_tree): This.
2917 (lookup_id_for_tree): Renamed and rewritten to ...
2918 (lookup_vi_for_tree): This.
2919 (get_id_for_tree): Renamed and rewritten to ...
2920 (get_vi_for_tree): Ditto.
2921 (get_constraint_exp_from_ssa_var): Update to use get_vi_for_tree.
2922 (process_constraint): Don't handle INCLUDES.
2923 Remove special ADDRESSOF case.
2924 (find_func_aliases): Rewrite to use vi functions instead of id
2925 ones.
2926 (create_function_info_for): Ditto.
2927 (create_variable_info_for): Ditto.
2928 (intra_create_variable_infos): Ditto.
2929 (merge_smts_into): Ditto.
2930 (find_what_p_points_to): Ditto.
2931 (init_base_vars): Ditto.
2932 (init_alias_vars): Ditto.
2933 (remove_preds_and_fake_succs): New function.
2934 (dump_sa_points_to_info): Dump new stats.
2935 (dump_solution_for_var): Use find.
2936 (set_used_smts): Fix formatting.
2937 (compute_points_to_sets): Updated for new functions.
2938 (ipa_pta_execute): Ditto.
2939
2940 2007-01-18 Kazu Hirata <kazu@codesourcery.com>
2941 Richard Sandiford <richard@codesourcery.com>
2942
2943 * doc/tm.texi (TARGET_FUNCTION_VALUE): Expand documentation of
2944 parallels.
2945 * calls.c (expand_call): If the return value is a PARALLEL,
2946 extract its first member.
2947 * config/m68k/linux.h (FUNCTION_EXTRA_EPILOGUE): Remove.
2948 * config/m68k/m68k.c (m68k_output_function_epilogue): Don't
2949 use FUNCTION_EXTRA_EPILOGUE.
2950 (m68k_function_value): Return a PARALLEL if the return value
2951 is of a pointer type.
2952 * config/m68k/netbsd-elf.h (current_function_returns_pointer)
2953 (FUNCTION_EXTRA_EPILOGUE): Remove.
2954 * config/m68k/m68k.md (D0_REG): New constant.
2955
2956 2007-01-18 Kazu Hirata <kazu@codesourcery.com>
2957
2958 * config/m68k/m68k.c (m68k_output_function_epilogue): Don't
2959 output a NOP for empty epilogues.
2960
2961 2007-01-18 Richard Sandiford <richard@codesourcery.com>
2962
2963 * config/m68k/m68k.c (m68k_use_return_insn): Update comments
2964 before function. Extend register save check to include all
2965 registers, not just integer ones.
2966
2967 2007-01-18 Kazu Hirata <kazu@codesourcery.com>
2968
2969 * config/m68k/m68k-protos.h (use_return_insn): Rename to...
2970 (m68k_use_return_insn): ...this.
2971 * config/m68k/m68k.h (USE_RETURN_INSN): Delete.
2972 * config/m68k/m68k.c (use_return_insn): Rename to...
2973 (m68k_use_return_insn): ...this.
2974 * config/m68k/m68k.md (return): Use m68k_use_return_insn instead
2975 of USE_RETURN_INSN.
2976
2977 2007-01-18 Nathan Sidwell <nathan@codesourcery.com>
2978
2979 * config/m68k/fpgnulib.c (__truncdfsf2): Implement round to
2980 nearest even, fix denormal rounding overflow.
2981
2982 2007-01-18 Richard Sandiford <richard@codesourcery.com>
2983
2984 * config/m68k/m68k.md (movsf_cf_hard): Use fsmove instead of
2985 f%$smove and f%$move.
2986 (movdf_cf_hard): Use fdmove for cases 0 and 3 and fmove for case 1.
2987 (extendsfdf2_cf): Use fdmove instead of f%&move.
2988 (truncdfsf2_cf): Use fsmove instead of f%$smove.
2989 (add<mode>3_cf, sub<mode>3_cf): Use <FP:prec> instead of <FP:round>.
2990
2991 2007-01-18 Richard Sandiford <richard@codesourcery.com>
2992
2993 * config/m68k/m68k.md (movdf_cf_hard): Use output_move_double for
2994 GPR<-GPR moves.
2995
2996 2007-01-18 Richard Sandiford <richard@codesourcery.com>
2997
2998 * real.h (real_format): Add a canonical_nan_lsbs_set field.
2999 (coldfire_single_format): Declare.
3000 (coldfire_double_format): Likewise.
3001 * real.c (encode_ieee_single): Use canonical_nan_lsbs_set instead
3002 of qnan_msb_set to determine the lower bits of a canonical
3003 NaN significand.
3004 (encode_ieee_double): Likewise.
3005 (encode_ieee_quad): Likewise.
3006 (ieee_single_format): Initialize canonical_nan_lsbs_set.
3007 (mips_single_format): Likewise.
3008 (ieee_double_format): Likewise.
3009 (mips_double_format): Likewise.
3010 (ieee_extended_motorola_format): Likewise.
3011 (ieee_extended_intel_96_format): Likewise.
3012 (ieee_extended_intel_128_format): Likewise.
3013 (ieee_extended_intel_96_round_53_format): Likewise.
3014 (ibm_extended_format): Likewise.
3015 (mips_extended_format): Likewise.
3016 (ieee_quad_format): Likewise.
3017 (mips_quad_format): Likewise.
3018 (vax_f_format): Likewise.
3019 (vax_d_format): Likewise.
3020 (vax_g_format): Likewise.
3021 (i370_single_format): Likewise.
3022 (i370_double_format): Likewise.
3023 (decimal_single_format): Likewise.
3024 (decimal_double_format): Likewise.
3025 (decimal_quad_format): Likewise.
3026 (c4x_single_format): Likewise.
3027 (c4x_extended_format): Likewise.
3028 (real_internal_format): Likewise.
3029 (coldfire_single_format): New real_format.
3030 (coldfire_double_format): Likewise.
3031 * config/pdp11/pdp11.c (pdp11_f_format): Initialize
3032 canonical_nan_lsbs_set.
3033 (pdp11_d_format): Likewise.
3034 * config/m68k/m68k.c (override_options): Override REAL_FORMAT_MODE
3035 if TARGET_COLDFIRE_CPU.
3036
3037 2007-01-18 Richard Sandiford <richard@codesourcery.com>
3038
3039 * config/m68k/m68k-protos.h (m68k_output_pic_call): Delete.
3040 (output_call, m68k_legitimize_call_address): Declare.
3041 * config/m68k/m68k.h (EXTRA_CONSTRAINT): Remove unnecessary
3042 parenthesees. Add support for a 'W' constraint.
3043 (LEGITIMATE_PIC_OPERAND_P): Remove SYMBOL_REF_FLAG handling.
3044 (PRINT_OPERAND_PUNCT_VALID_P): Remove comment about 'o'.
3045 (m68k_symbolic_call, m68k_symbolic_jump): Declare.
3046 * config/m68k/m68k.c (m68k_symbolic_call, m68k_symbolic_jump): New
3047 variables.
3048 (override_options): Initialize them. Do not set flag_no_function_cse
3049 for TARGET_ID_SHARED_LIBRARY.
3050 (m68k_output_pic_call): Delete.
3051 (m68k_legitimize_call_address): New function.
3052 (print_operand): Remove the %o prefix. Handle the %p prefix.
3053 (output_call): New function.
3054 (m68k_output_mi_thunk): Use m68k_symbolic_jump. Always load the
3055 target address from the GOT if symbolic jumps are not allowed.
3056 * config/m68k/m68k.md (call, general_operand): Do not set
3057 SYMBOL_REF_FLAG. Use m68k_legitimize_call_address instead.
3058 Merge separate flag_pic and !flag_pic define_insns into...
3059 (*call, *call_value): ...these new patterns. Match the address
3060 rather than the containing MEM and require it to be a call_operand.
3061 Use output_call to generate the asm template.
3062 * config/m68k/predicates.md (const_call_operand): New predicate.
3063 (call_operand): Likewise.
3064
3065 2007-01-18 Nathan Sidwell <nathan@codesourcery.com>
3066
3067 * config/m68k/m68k.h (REGISTER_MOVE_COST): Simplify definition.
3068 (STACK_GROWS_DOWNWARD): Define to 1.
3069 (FUNCTION_VALUE, LIBCALL_VALUE, FUNCTION_VALUE_REGNO_P): Equivocate
3070 comments, emphasizing that these values are only defaults.
3071 * config/m68k/linux.h (LINK_SPEC): Fix formatting in #undef.
3072 * config/m68k/m68k.c (const_method): Remove trailing whitespace.
3073
3074 2007-01-18 Richard Sandiford <richard@codesourcery.com>
3075
3076 * config/m68k/m68k.md (cmpsi): Remove outdated flag_pic handling.
3077
3078 2007-01-18 Kazu Hirata <kazu@codesourcery.com>
3079 Richard Sandiford <richard@codesourcery.com>
3080
3081 * config/m68k/m68k.h (DATA_REGNO_P, ADDRESS_REGNO_P, INT_REGNO_P)
3082 (FP_REGNO_P): New macros.
3083 (REGNO_OK_FOR_INDEX_P, REGNO_OK_FOR_BASE_P, REGNO_OK_FOR_DATA_P)
3084 (REGNO_OK_FOR_FP_P, REG_OK_FOR_INDEX_P, REG_OK_FOR_BASE_P): Use them.
3085 (EH_RETURN_STACKADJ_RTX): Use A0_REG.
3086 * config/m68k/m68k.c (m68k_regno_mode_ok): Use the new REGNO macros.
3087
3088 2007-01-18 Nathan Sidwell <nathan@codesourcery.com>
3089
3090 * config.gcc (m68k-*-aout*, m68k-*-coff*, m68020-*-elf*, m68k-*-elf*)
3091 (m68k-*-uclinuxoldabi, m68k-*-uclinux*, m68k-*-rtems*): Add t-floatlib
3092 to $tmake_file.
3093 * config/m68k/t-floatlib: New file, extracting common code from...
3094 * config/m68k/t-m68kbare, config/m68k/t-m68kelf,
3095 * config/m68k/t-uclinux: Here.
3096 * config/m68k/fpgnulib.c: Do not compile extendeed precision
3097 routines on ColdFire targets.
3098
3099 2007-01-18 Nathan Sidwell <nathan@codesourcery.com>
3100
3101 * config.gcc (m68k-*-aout*, m68k-*-coff*, m68020-*-elf*, m68k-*-elf*)
3102 (m68010-*-netbsdelf*, m68k*-*-netbsdelf*, m68k*-*-openbsd*)
3103 (m68k-*-uclinuxoldabi, m68k-*-uclinux*, m68k-*-linux*)
3104 (m68k-*-rtems*): Use tm_file rather than m68k/m68k.h and
3105 explicitly set MOTOROLA to 1.
3106 * config/m68k/m68k.h (MOTOROLA): Simplify definition accordingly.
3107
3108 2007-01-18 Richard Sandiford <richard@codesourcery.com>
3109 Nathan Sidwell <nathan@codesourcery.com>
3110
3111 * config/m68k/m68k.h (PCC_STATIC_STRUCT_RETURN): Delete.
3112 (ASM_OUTPUT_REG_PUSH, ASM_OUTPUT_REG_POP): Add MOTOROLA cases.
3113 * config/m68k/coff.h (REGISTER_PREFIX_MD): Delete.
3114 * config/m68k/m68020-elf.h (LIB_SPEC): Delete.
3115 * config/m68k/m68k-none.h (CC1_SPEC, CPP_SUBTARGET_SPEC): Delete.
3116 * config/m68k/m68kelf.h (IMMEDIATE_PREFIX, REGISTER_PREFIX_MD)
3117 (ASM_OUTPUT_REG_PUSH, ASM_OUTPUT_REG_POP): Delete.
3118 (NO_DOLLAR_IN_LABEL, PCC_STATIC_STRUCT_RETURN): Don't undefine.
3119 (BSS_ASM_OP, ASM_OUTPUT_SKIP, ASM_OUTPUT_ASCII): Delete.
3120 * config/m68k/m68kemb.h (PCC_STATIC_STRUCT_RETURN): Don't undefine.
3121 * config/m68k/linux.h (SIZE_TYPE, PTRDIFF_TYPE, WCHAR_TYPE)
3122 (WCHAR_TYPE_SIZE, TARGET_OBJFMT_CPP_BUILTINS): Delete.
3123 (TARGET_OS_CPP_BUILTINS): Don't define mc68000 and mc68020 here.
3124 (DBX_REGISTER_NUMBER): Delete.
3125 * config/m68k/netbsd-elf.h (IMMEDIATE_PREFIX): Delete.
3126 (PCC_STATIC_STRUCT_RETURN): Don't undefine.
3127 * config/m68k/openbsd.h (PCC_STATIC_STRUCT_RETURN): Define.
3128
3129 2007-01-18 Richard Sandiford <richard@codesourcery.com>
3130
3131 * config.gcc (m68k-*-uclinux*): Add flat.h to $tm_file.
3132 * config/flat.h: New file.
3133 * crtstuff.c (USE_PT_GNU_EH_FRAME): Don't define if
3134 OBJECT_FORMAT_FLAT.
3135 * config/m68k/m68k.h (ASM_PREFERRED_EH_DATA_FORMAT): Do not use
3136 indirect references for -msep-data or -mid-shared-library.
3137 Do not use PC-relative code addresses either.
3138 * config/m68k/m68k.c (override_options): Restrict -fPIC error
3139 to -mpcrel.
3140 * config/m68k/uclinux.h (STARTFILE_SPEC): Define. Use Scrt1.o
3141 for shared libraries and crt1.o for executables. Use crti.o and
3142 crtbegin.o.
3143 (ENDFILE_SPEC): Use crtend.o and crtn.o.
3144 (LIB_SPEC): Suppress -Rlibc.gdb if -static-libc is given.
3145 Do not add -elf2flt or -shared-lib-id options here.
3146 (LINK_SPEC): Define. Pass -elf2flt if no -elf2flt option is given.
3147 Pass -shared-lib-id if -mid-shared-library, taking the library
3148 identifier from -mshared-library-id if given, otherwise
3149 defaulting to 0.
3150 (EH_FRAME_IN_DATA_SECTION): Do not undefine.
3151 (INIT_SECTION_ASM_OP, FINI_SECTION_ASM_OP): Likewise.
3152 (TARGET_OS_CPP_BUILTINS): Define __GXX_MERGED_TYPEINFO_NAMES=0
3153 and __GXX_TYPEINFO_EQUALITY_INLINE=0 if -mid-shared-library.
3154 (DRIVER_SELF_SPECS): Map unadorned PIC options to -msep-data.
3155 * config/m68k/t-uclinux (EXTRA_MULTILIB_PARTS): Add crtbegin.o
3156 and crtend.o.
3157 * config/m68k/lb1sf68.asm (PICCALL): Use an lea and pc-relative
3158 jump sequence for ISA A and ISA A+.
3159 (PICJUMP): Likewise.
3160
3161 2007-01-18 Richard Sandiford <richard@codesourcery.com>
3162
3163 * config.gcc (m68k-*-uclinux*): Base the port on the common
3164 and m68k GNU/Linux files rather than on the generic ELF ones.
3165 * config/m68k/uclinux.h (TARGET_VERSION): Override.
3166 (TARGET_OS_CPP_BUILTINS): Use LINUX_TARGET_OS_CPP_BUILTINS.
3167
3168 2007-01-18 Julian Brown <julian@codesourcery.com>
3169 Richard Sandiford <richard@codesourcery.com>
3170
3171 * config/m68k/m68k.h (LONG_DOUBLE_TYPE_SIZE): Make 64-bit on ColdFire.
3172 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
3173 * config/m68k/netbsd-elf.h (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Undefine
3174 before redefining.
3175 * config/m68k/uclinux-oldabi.h (LONG_DOUBLE_TYPE_SIZE): Redefine to
3176 80 unconditionally.
3177 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
3178
3179 2007-01-18 Richard Sandiford <richard@codesourcery.com>
3180
3181 * doc/install.texi: Document m68k-uclinuxoldabi.
3182 * config.gcc (m68k-*-uclinuxoldabi): New configuration.
3183 * config/m68k/uclinux-oldabi.h: New file, copied from
3184 config/m68k/uclinux.h.
3185
3186 2007-01-18 Kaz Kojima <kkojima@rr.iij4u.or.jp>
3187
3188 * config/m32r/m32r-protos.h (m32r_expand_epilogue): Declare it.
3189 * config/m32r/m32r.c (m32r_setup_incoming_varargs): Use gen_frame_mem.
3190 (m32r_compute_frame_size): Use unsigned for regno.
3191 (m32r_reload_lr): Use gen_frame_mem.
3192 (pop): New.
3193 (m32r_output_function_epilogue): Don't output the function epilogue
3194 textually here.
3195 (m32r_expand_epilogue): New.
3196 (direct_return): Return false if the function has the interrupt
3197 attribute.
3198 (m32r_hard_regno_rename_ok): Remove code for the textual epilogue.
3199 * config/m32r/m32r.md (epilogue): New expander.
3200 (return_lr, return_rte): New insns.
3201 (return): Make it expander.
3202 (return_normal): New expander.
3203
3204 2007-01-18 Josh Conner <jconner@apple.com>
3205
3206 PR target/30485
3207 * config/rs6000/rs6000.c (rs6000_emit_vector_compare): Add
3208 support for UNLE, UNLT, UNGE, and UNGT.
3209
3210 2007-01-18 Jan Hubicka <jh@suse.cz>
3211
3212 * tree-vrp.c (finalize_jump_threads): Do not call cleanup_cfg by hand.
3213 * tree-tailcall (add_virtual_phis): Likewise.
3214 (optimize_tail_call): Return TODOs.
3215 (execute_tail_calls): Return TODOs.
3216 * tree-ssa-ccp (execute_fold_all_builtins): Do cleanup_cfg via TODO.
3217 * tree-cfgcleanup.c (cleanup_tree_cfg_loop): Return if something
3218 changed.
3219 * tree-ssa-forwprop.c (tree_ssa_forward_propagate_single_use_value):
3220 Cleanup cfg using TODO.
3221 * tree-flow.h (cleanup_tree_cfg_loop): Update prototype.
3222 * passes.c (execute_function_todo): When cleanup did something, remove
3223 unused locals.
3224 * tree-cfg.c (pass_build_cfg): Add cleanup_cfg TODO.
3225 (make_edges): Don't cleanup_cfg.
3226
3227 2007-01-18 Uros Bizjak <ubizjak@gmail.com>
3228
3229 * reg-stack.c (subst_stack_regs_pat) [UNSPEC_SINCOS_COS,
3230 UNSPEC_XTRACT_FRACT]: Use generic code for instructions that
3231 operate on the top of stack.
3232 [UNSPEC_SINCOS_SIN, UNSPEC_XTRACT_EXP, UNSPEC_TAN]: Rewrite
3233 register handling of instructions that output to the second
3234 stack slot.
3235 [UNSPEC_TAN_ONE, UNSPEC_TAN_TAN]: Remove.
3236 (move_for_stack_reg): Special-case check for dead destination
3237 stack slot for constant load of 1.0 inside UNSPEC_TAN.
3238
3239 * config/i386/i386.md (UNSPEC_TAN): New constant.
3240 (UNSPEC_TAN_ONE, UNSPEC_TAN_TAN): Remove.
3241 (fptanxf4_i387, fptan_extend<mode>xf4_i387): New patterns
3242 to correctly model move of constant 1.0 to top stack slot.
3243 (*tandf3_1, *tansf3_1, *tanxf3_1): Remove insn patterns.
3244 (unnamed peephole2 pattern): Remove corresponding peephole2
3245 pattern that optimizes tan insn and loading of constant 1.0.
3246 (tanxf2): Use fptanxf4_i387.
3247 (tan<mode>2): Rename from tansf2 and tandf2 and macroize insn
3248 patterns using X87MODEF12 mode macro. Use fptan_extend<mode>xf4_i387
3249 and truncate result to requested mode. Use SSE_FLOAT_MODE_P to
3250 disable patterns for SSE math.
3251 (sincos<mode>3): Use truncxf<mode>2_i387_noop for truncation.
3252 (fyl2x_extend<mode>xf3_i387): Use X87MODEF12 for operand 1.
3253
3254 2007-01-18 Dirk Mueller <dmueller@suse.de>
3255 Richard Guenther <rguenther@suse.de>
3256
3257 PR diagnostic/8268
3258 * doc/invoke.texi (Warray-bounds): Document -Warray-bounds.
3259 * common.opt (Warray-bounds): Add new warning option.
3260 * c-opts.c (c_common_handle_option): Define -Warray-bounds
3261 if -Wall is given.
3262 * Makefile.in: make tree-vrp.o depend on toplev.h
3263 * tree-vrp.c (vrp_finalize): Call check_array_refs if -Warray-bounds
3264 is enabled.
3265 (check_array_refs, check_array_bounds, check_array_ref): New.
3266
3267 2007-01-18 Jan Hubicka <jh@suse.cz>
3268
3269 * tree-ssa-ccp.c (ccp_finalize): Return if something changed.
3270 (execute_ssa_ccp): Return flags conditionally.
3271 * tree-ssa-propagate.c (substitue_and_fold): Return if something was
3272 changed.
3273 * tree-ssa-propagate.h (substitute_and_fold): Update prototype.
3274
3275 2007-01-18 Steven Bosscher <steven@gcc.gnu.org>
3276
3277 * cfgcleanup.c (cleanup_cfg): Detect cfglayout mode and set
3278 the CLEANUP_CFGLAYOUT flag when in cfglayout mode.
3279
3280 * Makefile.c (GTFILES): Add cfglayout.h.
3281 * gengtype.c (open_base_files): Likewise.
3282 * cfglayout.c (cfg_layout_function_footer,
3283 cfg_layout_function_header) Reindent to make gengtype happy.
3284 * cfglayout.h (cfg_layout_function_footer,
3285 cfg_layout_function_header): Add GTY(()) marker.
3286
3287 * ifcvt.c (noce_try_sign_mask): Make sure INSN_B is non-null.
3288
3289 2007-01-18 Ben Elliston <bje@au.ibm.com>
3290
3291 * genautomata.c (write_automata): Include xstrerror output in the
3292 error message if writing the DFA description file fails.
3293
3294 2007-01-17 H.J. Lu <hongjiu.lu@intel.com>
3295
3296 * config/mips/mips-protos.h (mips_output_external): Make it
3297 return void.
3298 * config/mips/iris.h (TARGET_ASM_EXTERNAL_LIBCALL): Removed.
3299 * config/mips/mips.c (irix_output_external_libcall): Likewise.
3300 (extern_list): Likewise.
3301 (extern_head): Likewise.
3302 (TARGET_ASM_FILE_END): Likewise.
3303 (mips_file_end): Likewise.
3304 (mips_output_external): Rewritten.
3305
3306 2007-01-18 Ben Elliston <bje@au.ibm.com>
3307
3308 * genpreds.c (write_insn_preds_c): Only write out the function
3309 body for regclass_for_constraint if we have register constraints.
3310
3311 2007-01-17 Tom Tromey <tromey@redhat.com>
3312
3313 * doc/sourcebuild.texi (libgcj Tests): Use sourceware.org.
3314 * doc/install.texi (Testing): Use sourceware.org.
3315 (Binaries): Likewise.
3316 (Specific): Likewise.
3317 * doc/contrib.texi (Contributors): Use sourceware.org.
3318
3319 2007-01-17 Anatoly Sokolov <aesok@post.ru>
3320
3321 * config/avr/avr.h (AVR_HAVE_LPMX): New macro.
3322 (AVR_ENHANCED): Rename to ...
3323 (AVR_HAVE_MUL): ... new.
3324 (avr_enhanced_p): Rename to ...
3325 (avr_have_mul_p): ... new.
3326 (TARGET_CPU_CPP_BUILTINS): Use 'avr_have_mul_p' instead of
3327 'avr_enhanced_p' for "__AVR_ENHANCED__". Define "__AVR_HAVE_MUL__".
3328 * config/avr/avr.c (avr_enhanced_p): Rename to ...
3329 (avr_have_mul_p): ... new.
3330 (base_arch_s): Rename 'enhanced' to 'have_mul'.
3331 (avr_override_options): Use 'avr_have_mul_p' and 'have_mul' instead of
3332 'avr_enhanced_p' and 'enhanced'.
3333 (ashlhi3_out, ashrhi3_out, lshrhi3_out, avr_rtx_costs): Use
3334 AVR_HAVE_MUL instead of AVR_ENHANCED.
3335 * avr.md (*tablejump_enh): Use AVR_HAVE_LPMX instead of AVR_ENHANCED.
3336 (mulqi3, *mulqi3_enh, *mulqi3_call, mulqihi3, umulqihi3, mulhi3,
3337 *mulhi3_enh, *mulhi3_call, mulsi3, *mulsi3_call): Use AVR_HAVE_MUL
3338 instead of AVR_ENHANCED.
3339 (*tablejump_enh): Use AVR_HAVE_LPMX instead of AVR_ENHANCED.
3340 * libgcc.S: Use __AVR_HAVE_MUL__ instead of __AVR_ENHANCED__.
3341 (__tablejump__): Use __AVR_HAVE_LPMX__ instead of __AVR_ENHANCED__.
3342
3343 2007-01-17 Ian Lance Taylor <iant@google.com>
3344
3345 * vec.h (VEC_reserve_exact): Define.
3346 (vec_gc_p_reserve_exact): Declare.
3347 (vec_gc_o_reserve_exact): Declare.
3348 (vec_heap_p_reserve_exact): Declare.
3349 (vec_heap_o_reserve_exact): Declare.
3350 (VEC_OP (T,A,reserve_exact)): New static inline function, three
3351 versions.
3352 (VEC_OP (T,A,reserve)) [all versions]: Remove handling of
3353 negative parameter.
3354 (VEC_OP (T,A,alloc)) [all versions]: Call ...reserve_exact.
3355 (VEC_OP (T,A,copy)) [all versions]: Likewise.
3356 (VEC_OP (T,a,safe_grow)) [all versions]: Likewise.
3357 * vec.c (calculate_allocation): Add exact parameter. Change all
3358 callers.
3359 (vec_gc_o_reserve_1): New static function, from vec_gc_o_reserve.
3360 (vec_gc_p_reserve, vec_gc_o_reserve): Call vec_gc_o_reserve_1.
3361 (vec_gc_p_reserve_exact, vec_gc_o_reserve_exact): New functions.
3362 (vec_heap_o_reserve_1): New static function, from vec_heap_o_reserve.
3363 (vec_heap_p_reserve, vec_heap_o_reserve): Call vec_heap_o_reserve_1.
3364 (vec_heap_p_reserve_exact): New function.
3365 (vec_heap_o_reserve_exact): New function.
3366
3367 2007-01-17 Jan Hubicka <jh@suse.cz>
3368
3369 * ipa-type-escape.c (look_for_casts): Revamp using handled_component_p.
3370
3371 2007-01-17 Eric Christopher <echristo@apple.com>
3372
3373 * config.gcc: Support core2 processor.
3374
3375 2007-01-16 Jan Hubicka <jh@suse.cz>
3376
3377 * tree-ssanames.c (release_dead_ssa_names): Instead of ggc_freeing
3378 the names, just unlink the chain so we don't crash on dangling pointers
3379 to dead SSA names.
3380
3381 2007-01-16 Jan Hubicka <jh@suse.cz>
3382
3383 * cgraph.h (cgraph_decide_inlining_incrementally): Kill.
3384 * tree-pass.h: Reorder to make IPA passes appear toegher.
3385 (pass_early_inline, pass_inline_parameters, pass_apply_inline): Declare.
3386 * cgraphunit.c (cgraph_finalize_function): Do not compute inling
3387 parameters, do not call early inliner.
3388 * ipa-inline.c: Update comments. Include tree-flow.h
3389 (cgraph_decide_inlining): Do not compute inlining parameters.
3390 (cgraph_decide_inlining_incrementally): Return TODOs; assume to
3391 be called with function context set up.
3392 (pass_ipa_inline): Remove unreachable functions before pass.
3393 (cgraph_early_inlining): Simplify assuming to be called from the
3394 PM as local pass.
3395 (pass_early_inline): New pass.
3396 (cgraph_gate_ipa_early_inlining): New gate.
3397 (pass_ipa_early_inline): Turn into simple wrapper.
3398 (compute_inline_parameters): New function.
3399 (gate_inline_passes): New gate.
3400 (pass_inline_parameters): New pass.
3401 (apply_inline): Move here from tree-optimize.c
3402 (pass_apply_inline): New pass.
3403 * ipa.c (cgraph_remove_unreachable_nodes): Verify cgraph after
3404 transforming.
3405 * tree-inline.c (optimize_inline_calls): Return TODOs rather than
3406 doing them by hand.
3407 (tree_function_versioning): Do not allocate dummy struct function.
3408 * tree-inline.h (optimize_inline_calls): Update prototype.
3409 * tree-optimize.c (execute_fixup_cfg): Export.
3410 (pass_fixup_cfg): Remove
3411 (tree_rest_of_compilation): Do not apply inlines.
3412 * tree-flow.h (execute_fixup_cfg): Declare.
3413 * Makefile.in (gt-passes.c): New.
3414 * passes.c: Include gt-passes.h
3415 (init_optimization_passes): New passes.
3416 (nnodes, order): New static vars.
3417 (do_per_function_toporder): New function.
3418 (execute_one_pass): Dump current pass here.
3419 (execute_ipa_pass_list): Don't dump current pass here.
3420
3421 2007-01-16 Janis Johnson <janis187@us.ibm.com>
3422
3423 * config/dfp-bit.c (dfp_compare_op): Return separate value for NaN.
3424 (DFP_NE, DFP_LE, DFP_GE): Return false for NaN.
3425
3426 2007-01-16 David Edelsohn <edelsohn@gnu.org>
3427
3428 * config/rs6000/darwin-ldouble.c: Build file for SOFT_FLOAT.
3429 (strong_alias): Define.
3430 (__gcc_qmul): Provide non-FMA for soft-float.
3431 (__gcc_qdiv): Same.
3432 (__gcc_qneg): New.
3433 (__gcc_qeq): New.
3434 (__gcc_qle): New.
3435 (__gcc_qge): New.
3436 (__gcc_qunord): New.
3437 (__gcc_stoq): New.
3438 (__gcc_dtoq): New.
3439 (__gcc_qtos): New.
3440 (__gcc_qtod): New.
3441 (__gcc_qtoi): New.
3442 (__gcc_qtou): New.
3443 (__gcc_itoq): New.
3444 (__gcc_utoq): New.
3445 (fmsub): New.
3446 * config/rs6000/rs6000.c (rs6000_init_libfuncs): Initialize
3447 soft-float functions.
3448 * config/rs6000/libgcc-ppc-glibc.ver: Version soft-float symbols.
3449 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Do not warn
3450 about long double soft float.
3451
3452 2007-01-16 Dorit Nuzman <dorit@il.ibm.com>
3453 Tehila Meyzels <tehila@il.ibm.com>
3454
3455 * tree-vectorizer.h (is_pattern_stmt_p): New.
3456 * tree-vect-analyze.c (vect_determine_vectorization_factor): Fix
3457 formatting (tabs instead of spaces). Cleanup and clarify setting
3458 of STMT_VINFO_VECTYPE. Call is_pattern_stmt_p.
3459 * tree-vect-transform.c (vect_get_vec_def_for_operand): Fix typo.
3460 (vectorizable_type_demotion): Check that types are integral.
3461 (vectorizable_type_promotion): Likewise.
3462 (vectorizable_store): Fix typo. Eliminate new-line at end of
3463 comments.
3464
3465 2007-01-16 Jan Hubicka <jh@suse.cz>
3466
3467 * tree-ssanames.c (release_dead_ssa_names): Remove invalidated
3468 cgraph edges too.
3469
3470 2007-01-15 Eric Christopher <echristo@apple.com>
3471
3472 * ifcvt.c: Include vec.h, vecprim.h.
3473 (check_cond_move_block): New argument regs.
3474 Reorganize. Add registers used to regs.
3475 (cond_move_process_if_block): Use regs set above as
3476 loop bounds.
3477
3478 2007-01-15 Eric Christopher <echristo@apple.com>
3479
3480 * config/darwin.h: Update copyright.
3481 (TARGET_OPTION_TRANSLATE_TABLE): Add umbrella.
3482 (LINK_COMMAND_SPEC): Add -u.
3483 (LINK_SPEC): Fix umbrella for above.
3484
3485 2007-01-15 Joseph S. Myers <joseph@codesourcery.com>
3486
3487 * config/soft-fp/op-common.h, config/soft-fp/op-4.h: Update from
3488 glibc CVS.
3489
3490 2007-01-15 Tom Tromey <tromey@redhat.com>
3491
3492 * doc/sourcebuild.texi (libgcj Tests): Don't mention jacks.
3493 * doc/install.texi (Testing): Don't mention jacks.
3494 (Configuration): Document --enable-java-maintainer-mode. Move
3495 --with-java-home to libgcj-specific section. Document
3496 --with-ecj-jar.
3497 (Prerequisites): Mention --enable-java-maintainer-mode, ecj1.
3498
3499 2007-01-15 Jan Hubicka <jh@suse.cz>
3500
3501 * tree-ssa-dce.c (DCE_TODOs): New.
3502 (propagate_necessity): Return if something changed.
3503 (eliminate_unnecessary_stmts): Likewise.
3504 (perform_tree_ssa_dce): Return TODO flags when needed.
3505 (pass_dce, pass_dce_loop, pass_cd_dce): Remove TODO flags.
3506
3507 2007-01-15 Uros Bizjak <ubizjak@gmail.com>
3508
3509 * config/i386/i386.md (fyl2xxf3_i387): Rename from fyl2x_xf3.
3510 (fyl2x_extend<mode>xf3_i387): New insn pattern.
3511 (log<mode>2): Rename from logsf2 and logdf2 and macroize insn
3512 patterns using X87MODEF12 mode macro. Extend operand 1
3513 to XFmode. Use SSE_FLOAT_MODE_P to disable patterns for SSE math.
3514 (log10<mode>2): Ditto.
3515 (log2<mode>2): Ditto.
3516 (log1p<mode>2): Ditto.
3517 (logb<mode>2): Ditto.
3518 (fyl2xp1xf3_i387): Rename from fyl2xp1_xf3.
3519 (fyl2xp1_extend<mode>xf3_i387): New insn pattern.
3520 (*fxtractxf3_i387): Rename from *fxtractxf3.
3521 (fxtract_extend<mode>xf3_i387): New insn pattern.
3522 (ilogbsi2): Use match_dup 3, not match_operand:XF 3.
3523 * config/i386/i386.c (ix86_emit_i387_log1p): Use gen_fyl2xp1xf3_i387()
3524 and gen_fyl2xxf3_i387().
3525
3526 2007-01-14 Zdenek Dvorak <dvorakz@suse.cz>
3527
3528 * loop-unswitch.c (unswitch_loop): Do not call fix_loop_placement.
3529 * cfgloopmanip.c (fix_loop_placement): Made static. Use
3530 get_loop_exit_edges. Changed return type to bool.
3531 * cfgloop.h (fix_loop_placement): Declaration removed.
3532
3533 2007-01-14 Dorit Nuzman <dorit@il.ibm.com>
3534
3535 * param.h (MIN_VECT_LOOP_BOUND): New.
3536 * params.def (MIN_VECT_LOOP_BOUND): New.
3537 * tree-vectorizer.c (slpeel_tree_peel_loop_to_edge): Takes another
3538 argument - minimum threshold for number of iterations.
3539 * tree-vectorizer.h (slpeel_tree_peel_loop_to_edge): Add another
3540 argument to declaration.
3541 * tree-vect-analyze.c (vect_analyze_operations): Check value of
3542 MIN_VECT_LOOP_BOUND.
3543 * tree-vect-transform.c (vect_do_peeling_for_loop_bound): Call
3544 slpeel_tree_peel_loop_to_edge with additional argument.
3545 (vect_do_peeling_for_alignment): Likewise.
3546 * doc/invoke.texi (min-vect-loop-bound): Document new param option.
3547
3548 2007-01-14 Uros Bizjak <ubizjak@gmail.com>
3549
3550 PR target/30413
3551 * config/i386/i386.c (print_operand) ['z']: Output 'b' for
3552 operands of size 1.
3553
3554 2007-01-14 Jan Hubicka <jh@suse.cz>
3555
3556 * tree-dfa.c (remove_referenced_var): New function.
3557 * tree-ssa-live.c (remove_unused_locals): Walk referenced vars and
3558 prune referenced vars list too.
3559 * tree-flow.h (remove_referenced_var): Declare.
3560
3561 2007-01-14 Jan Hubicka <jh@suse.cz>
3562
3563 * tree-eh.c (add_stmt_to_eh_region_fn): Do not add call_exprs
3564 separately.
3565 (remove_stmt_from_eh_region_fn): Do not remove call_exprs.
3566 (verify_eh_throw_stmt_node, verify_eh_throw_table_statements): Kill.
3567 * except.h (verify_eh_throw_table_statements): Kill prototype.
3568 * cfgexpand.c (expand_gimple_basic_block): Propagate Eh regions
3569 into call exrepssions.
3570 * tree-optimize.c (execute_free_cfg_annotatiosn): Do not call
3571 eh trhow verifier.
3572 * tree-cfg.c: Include pointer-set.h.
3573 (verify_node_sharing): Work on pointer set.
3574 (verify_eh_throw_stmt_node): New.
3575 (verify_stmts): Use pointers sets, verify throw_stmt.
3576
3577 2007-01-13 Zdenek Dvorak <dvorakz@suse.cz>
3578
3579 * ipa-reference.c (analyze_function): Consider also addresses taken
3580 in phi nodes.
3581
3582 2007-01-12 Roger Sayle <roger@eyesopen.com>
3583
3584 * c-typeck.c (null_pointer_constant_p): Replace use of
3585 TREE_CONSTANT_OVERFLOW with TREE_OVERFLOW.
3586 (build_c_cast): Likewise.
3587
3588 2007-01-12 Roger Sayle <roger@eyesopen.com>
3589
3590 * tree.h (force_fit_type_double): Remove unused final argument.
3591 * c-common.c (constant_expression_warning): Replace use of
3592 TREE_CONSTANT_OVERFLOW with TREE_OVERFLOW.
3593 (convert_and_check): Likewise.
3594 (shorten_compare): Update call to force_fit_type_double.
3595 (c_common_truthvalue_conversion) <INTEGER_CST>: Use integer_zerop.
3596 * convert.c (convert_to_pointer): Update call to
3597 force_fit_type_double.
3598 * fold-const.c (force_fit_type_double): Remove overflowed_const
3599 argument.
3600 (int_const_binop, fold_convert_const_int_from_int,
3601 fold_convert_const_int_from_real, fold_div_compare,
3602 fold_sign_changed_comparison, fold_unary, fold_negate_const,
3603 fold_abs_const, fold_not_const): Remove the final argument from
3604 calls to force_fit_type_double.
3605
3606 2007-01-12 Andrew Pinski <andrew_pinski@playstation.sony.com>
3607
3608 * configure.ac: Set insn to "nop" for spu-*-* also.
3609 * configure: Regenerate.
3610
3611 2007-01-12 Olga Golovanevsky <olga@il.ibm.com>
3612
3613 * builtins.def : Add BUILT_IN_FREE.
3614
3615 2007-01-12 Jan Hubicka <jh@suse.cz>
3616
3617 PR tree-optimization/30443
3618 * tree-inline.c (tree_function_versioning): Do not optimize when
3619 cloning for inlining.
3620
3621 2007-01-12 Zdenek Dvorak <dvorakz@suse.cz>
3622
3623 * doc/loop.texi: Document recording of loop exits.
3624 * cfgloopmanip.c (loopify, duplicate_loop): Use alloc_loop.
3625 (update_single_exits_after_duplication,
3626 update_single_exit_for_duplicated_loop,
3627 update_single_exit_for_duplicated_loops): Removed.
3628 (duplicate_loop_to_header_edge): Do not call
3629 update_single_exits_after_duplication and
3630 update_single_exit_for_duplicated_loops.
3631 (loop_version): Do not update single_exit information.
3632 (fix_loop_structure): Use record_loop_exits instead of
3633 mark_single_exit_loops.
3634 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update
3635 the lists of loop exits.
3636 * cfghooks.c (redirect_edge_and_branch, redirect_edge_and_branch_force,
3637 split_edge, merge_blocks): Update the lists of loop exits.
3638 * modulo-sched.c (sms_schedule): Pass LOOPS_HAVE_RECORDED_EXITS to
3639 loop_optimizer_init.
3640 * loop-init.c (loop_optimizer_init): Call record_loop_exits instead
3641 of mark_single_exit_loops.
3642 (loop_optimizer_finalize): Call release_recorded_exits.
3643 * tree-ssa-loop.c (tree_loop_optimizer_init): Pass
3644 LOOPS_HAVE_RECORDED_EXITS to loop_optimizer_init.
3645 * tree-vectorizer.c (slpeel_tree_duplicate_loop_to_edge_cfg): Do not
3646 update single exit information.
3647 * lambda-code.c (perfect_nestify): Ditto.
3648 * cfgloop.c (flow_loop_free): Destroy the list of exits of the loop.
3649 (mark_single_exit_loops): Removed.
3650 (alloc_loop, loop_exit_hash, loop_exit_eq, loop_exit_free,
3651 get_exit_descriptions, rescan_loop_exit, record_loop_exits,
3652 dump_recorded_exit, dump_recorded_exits, release_recorded_exits): New
3653 functions.
3654 (get_loop_exit_edges, single_exit): Use recorded exit lists.
3655 (add_bb_to_loop, remove_bb_from_loops): Update the lists of loop exits.
3656 (verify_loop_structure): Verify consistency of the exit lists.
3657 (flow_loops_find): Use alloc_loop. Initialize exits hash.
3658 (set_single_exit): Removed.
3659 * cfgloop.h (struct loop_exit): New function.
3660 (struct loop): single_exit_ field replaced by exits field.
3661 (LOOPS_HAVE_MARKED_SINGLE_EXITS): Replaced by LOOPS_HAVE_RECORDED_EXITS.
3662 (struct loops): Added exits hash.
3663 (mark_single_exit_loops, set_single_exit): Declaration removed.
3664 (release_recorded_exits, record_loop_exits, rescan_loop_exit): Declare.
3665
3666 2007-01-12 Richard Sandiford <richard@codesourcery.com>
3667
3668 * doc/invoke.texi: Avoid use of @headitem.
3669
3670 2007-01-12 Richard Sandiford <richard@codesourcery.com>
3671
3672 * cse.c (cse_insn): Move HAVE_CC0 code after declarations.
3673
3674 2007-01-12 Richard Sandiford <richard@codesourcery.com>
3675
3676 * doc/install.texi: Fix m68k-*-* anchor and add m68k-*-* to the
3677 list of targets.
3678
3679 2007-01-12 Nathan Sidwell <nathan@codesourcery.com>
3680 Richard Sandiford <richard@codesourcery.com>
3681
3682 * doc/invoke.texi: Document -mno-bitfield, -mno-rtd and -mno-short.
3683 * config/m68k/m68k.opt: Resort options.
3684 (mbitfield, mrtd, mshort): Remove RejectNegative properties.
3685
3686 2007-01-12 Nathan Sidwell <nathan@codesourcery.com>
3687 Richard Sandiford <richard@codesourcery.com>
3688
3689 * doc/invoke.texi: Document the macros that are defined by
3690 m68k's -mtune and -mhard-float options.
3691 * config/m68k/m68k-protos.h (m68k_cpp_cpu_ident) Declare.
3692 (m68k_cpp_cpu_family): Likewise.
3693 * config/m68k/m68k.h (TARGET_CPU_CPP_BUILTINS): Add a full set
3694 of __ucfv*__ macros. Define __mcffpu__ if generating code for
3695 ColdFire FPUs. Define __mcf_cpu_* and __mcf_family_* macros.
3696 * config/m68k/m68k.c (m68k_cpp_cpu_ident): New function.
3697 (m68k_cpp_cpu_family): Likewise.
3698
3699 2007-01-12 Richard Sandiford <richard@codesourcery.com>
3700
3701 * config/m68k/m68k.h (TARGET_CPU_CPP_BUILTINS): Treat all mc68*
3702 macros besides mc68000 as tuning macros. Use a switch statement
3703 to set them and mcpu32.
3704
3705 2007-01-12 Julian Brown <julian@codesourcery.com>
3706
3707 * config/m68k/m68k.h: Use TARGET_68040 instead of TARGET_68040_ONLY.
3708 (TARGET_68040_ONLY): Rename to...
3709 (TARGET_68040): ...this.
3710 * config/m68k/m68k.c: Use TARGET_68040 instead of TARGET_68040_ONLY.
3711 * config/m68k/m68k.md: Likewise.
3712
3713 2007-01-12 Julian Brown <julian@codesourcery.com>
3714 Nathan Sidwell <nathan@codesourcery.com>
3715 Richard Sandiford <richard@codesourcery.com>
3716
3717 * config.gcc (m680[012]0-*-*, m68k*-*-*): Set m68k_cpu_ident to
3718 the -mcpu= argument associated with the --with-cpu setting.
3719 Define M68K_DEFAULT_TUNE to the default -mtune= option,
3720 if different from the one implied by the -mcpu setting.
3721 Accept --with-cpu=FOO if FOO is listed in m68k-devices.def,
3722 using mcpu=FOO as the default CPU option. Set target_cpu_default2.
3723 * doc/invoke.texi: Mention ColdFire in the introduction to the
3724 m68k options. Document the new -march, -mcpu, -mtune, -mdiv,
3725 -mno-div and -mhard-float options. Make -m68881 a synonym for
3726 -mhard-float. Document the previously-undocumented -m5206e,
3727 -m528x, -m5307 and -m5407 options. Tweak the existing option
3728 documentation for consistency.
3729 * doc/install.texi: Mention new --with-cpu arguments.
3730 * config/m68k/m68k.h (OPTION_DEFAULT_SPECS): Only use the
3731 default CPU if neither -mcpu nor -march are specified.
3732 (ASM_CPU_SPEC): Pass down -mcpu and -march options.
3733 (TARGET_CPU_CPP_BUILTINS): Set __mcfisa*__ macros from
3734 TARGET_ISA*. Set the legacy __mcf*__ cpu macros in the same way,
3735 using m68k_tune to decide between families that implement the
3736 same ISA. Use m68k_tune to set __mcfv4e__.
3737 (FL_BITFIELD, FL_68881, FL_COLDFIRE, FL_CF_HWDIV, FL_CF_MAC)
3738 (FL_CF_EMAC, FL_CF_EMAC_B, FL_CF_USP, FL_CF_FPU, FL_ISA_68000)
3739 (FL_ISA_68010, FL_ISA_68020, FL_ISA_68040, FL_ISA_A, FL_ISA_B)
3740 (FL_ISA_C, FL_ISA_MMU): New macros.
3741 (MASK_COLDFIRE): Delete.
3742 (TARGET_68010, TARGET_68020, TARGET_68040_ONLY, TARGET_COLDFIRE)
3743 (TARGET_ISAB): Redefine in terms of m68k_cpu_flags.
3744 (TARGET_68881, TARGET_COLDFIRE_FPU): Redefine in terms of m68k_fpu.
3745 (TARGET_HARD_FLOAT): Do not define here.
3746 (TARGET_ISAAPLUS, TARGET_ISAC): New macros.
3747 (TUNE_68000): New macro.
3748 (TUNE_68000_10): Redefine in terms of TUNE_68000 and TUNE_68010.
3749 (TUNE_68010, TUNE_68030, TUNE_68040, TUNE_68060, TUNE_CPU32)
3750 (TUNE_CFV2): Redefine in terms of m68k_tune.
3751 (uarch_type, target_device, fpu_type): New enums.
3752 (m68k_cpu, m68k_tune, m68k_fpu, m68k_cpu_flags): Declare.
3753 * config/m68k/m68k.c (TARGET_DEFAULT): Remove MASK_68881.
3754 (FL_FOR_isa_00, FL_FOR_isa_10, FL_FOR_isa_20, FL_FOR_isa_40)
3755 (FL_FOR_isa_cpu32, FL_FOR_isa_a, FL_FOR_isa_aplus, FL_FOR_isa_b)
3756 (FL_FOR_isa_c): New macros.
3757 (m68k_isa): New enum.
3758 (m68k_target_selection): New structure.
3759 (all_devices, all_isas, all_microarchs): New tables.
3760 (m68k_cpu_entry, m68k_arch_entry, m68k_tune_entry, m68k_cpu)
3761 (m68k_tune, m68k_fpu, m68k_cpu_flags): New variables.
3762 (MASK_ALL_CPU_BITS): Delete.
3763 (m68k_find_selection): New function.
3764 (m68k_handle_option): Handle -mcpu=, -march= and -mtune=.
3765 Map the legacy target options to a combination of the new ones.
3766 (override_options): Set m68k_cpu, m68k_tune, m68k_fpu and
3767 m68k_cpu_flags. Handle M68K_DEFAULT_TUNE. Use m68k_cpu_flags
3768 to derive default MASK_BITFIELD, MASK_CF_HWDIV and MASK_HARD_FLOAT
3769 settings.
3770 * config/m68k/m68k.opt (m5200, m5206e, m528x, m5307, m5407, mcfv4e)
3771 (m68010, m68020, m68020-40, m68020-60, m68030, m68040): Remove Mask
3772 properties.
3773 (m68881, msoft-float): Change mask from 68881 to HARD_FLOAT.
3774 (march=, mcpu=, mdiv, mhard-float, mtune=): New options.
3775 * config/m68k/m68k-devices.def: New file.
3776
3777 2007-01-12 Richard Sandiford <richard@codesourcery.com>
3778 Nathan Sidwell <nathan@codesourcery.com>
3779
3780 * config/m68k/m68k.h (ASM_CPU_SPEC, ASM_SPEC, EXTRA_SPECS)
3781 (SUBTARGET_EXTRA_SPECS): New macros.
3782 * config/m68k/linux.h (ASM_SPEC): Remove CPU flags;
3783 use %(asm_cpu_spec) instead.
3784 * config/m68k/m68k-none.h (ASM_SPEC): Likewise.
3785 * config/m68k/openbsd.h (ASM_SPEC): Likewise.
3786 * config/m68k/netbsd-elf.h (ASM_SPEC): Likewise.
3787 (EXTRA_SPECS): Rename to...
3788 (SUBTARGET_EXTRA_SPECS): ...this.
3789
3790 2007-01-12 Nathan Sidwell <nathan@codesourcery.com>
3791 Richard Sandiford <richard@codesourcery.com>
3792 Julian Brown <julian@codesourcery.com>
3793
3794 * config.gcc (m68k-*-aout*, m68k-*-coff*, m68020-*-elf*, m68k-*-elf*)
3795 (m68k-*-uclinux*, m68k-*-linux*, m68k-*-rtems*): Set default_m68k_cpu
3796 to the configuration's default CPU.
3797 (m68010-*-netbsdelf*, m68k*-*-netbsdelf*, m68k*-*-openbsd*): Likewise.
3798 Remove default masks.
3799 (m680[012]0-*-*): Set the default with_cpu to the first part of
3800 the target name.
3801 (m68k*-*-*): Set the default with_cpu to m$default_m68k_cpu.
3802 (m68k*-*-linux): Extend the --with-cpu handling to...
3803 (m680[012]0-*-*, m68k*-*-*): ...these configurations. Allow m68000
3804 and m68010. Don't set target_cpu_default2.
3805 * doc/install.texi: Document --with-cpu for m68k.
3806 * config/m68k/m68k.h (OPTION_DEFAULT_SPECS): Define.
3807 * config/m68k/m68k-none.h (TARGET_CPU_DEFAULT, M68K_CPU_m68k)
3808 (M68K_CPU_m68000, M68K_CPU_m68010, M68K_CPU_m68020, M68K_CPU_m68030)
3809 (M68K_CPU_m68040, M68K_CPU_m68302, M68K_CPU_m68332, TARGET_DEFAULT)
3810 (ASM_CPU_DEFAULT_SPEC, CC1_CPU_DEFAULT_SPEC): Delete.
3811 (ASM_SPEC): Remove use of %(asm_cpu_default).
3812 (EXTRA_SPECS, SUBTARGET_EXTRA_SPECS, MULTILIB_DEFAULTS): Delete.
3813 * config/m68k/linux.h (TARGET_DEFAULT): Delete.
3814 (CPP_SPEC): Merge definitions. Do not handle __HAVE_68881__ here.
3815 * config/m68k/netbsd-elf.h (TARGET_OS_CPP_BUILTINS): Define
3816 __HAVE_FPU__ if TARGET_HARD_FLOAT.
3817 (TARGET_DEFAULT): Delete.
3818 (EXTRA_SPECS): Delete cpp_cpu_default_spec, cpp_cpu_spec,
3819 cpp_fpu_spec, asm_default_spec and netbsd_cpp_spec.
3820 (CPP_CPU_SPEC): Delete.
3821 (TARGET_VERSION): Merge definitions, using TARGET_68010 to pick
3822 the appropriate string.
3823 (CPP_CPU_DEFAULT_SPEC, ASM_DEFAULT_SPEC, CPP_FPU_SPEC): Delete.
3824 (CPP_SPEC): Define to NETBSD_CPP_SPEC.
3825 (ASM_SPEC): Don't use %(asm_default_spec).
3826 * config/m68k/m68k.c (TARGET_DEFAULT_TARGET_FLAGS): Remove
3827 TARGET_DEFAULT and add MASK_68881.
3828 * config/m68k/m68k.md: Remove mention of TARGET_DEFAULT from comments.
3829
3830 2007-01-12 Richard Sandiford <richard@codesourcery.com>
3831
3832 * config.gcc (m68010-*-netbsdelf*): Add MASK_68010.
3833 (m68k*-*-netbsdelf*, m68k*-*-openbsd*, m68k*-linux*): Add
3834 MASK_68010 alongside MASK_68020.
3835 * doc/invoke.texi: Document -m68010.
3836 * config/m68k/m68k.opt (m68010): New.
3837 * config/m68k/m68k.h (TARGET_CPU_CPP_BUILTINS): Define mc68010
3838 if TUNE_68010.
3839 (TUNE_68010): New macro.
3840 * config/m68k/m68k-none.h (M68K_CPU_m68k, M68K_CPU_m68010)
3841 (M68K_CPU_m68020, M68K_CPU_m68030, M68K_CPU_m68040)
3842 (M68K_CPU_m68332): Add MASK_68010.
3843 * config/m68k/linux.h (TARGET_DEFAULT): Add MASK_68010 to
3844 fallback definition.
3845 * config/m68k/netbsd-elf.h (CPP_CPU_SPEC): Remove now-redundant
3846 defines.
3847 * config/m68k/m68k.c (MASK_ALL_CPU_BITS): Add MASK_68010.
3848 (m68k_handle_option): Handle OPT_m68010. Add MASK_68010
3849 to all entries that use MASK_68020.
3850 (output_move_simode_const, output_move_himode, output_move_qimode)
3851 (output_move_stricthi, output_move_strictqi): Use TARGET_68010
3852 instead of TARGET_68020 to select clr behavior. Remove comment
3853 about there being no TARGET_68010.
3854 * config/m68k/m68k.md: Likewise throughout.
3855
3856 2007-01-12 Julian Brown <julian@codesourcery.com>
3857
3858 * config/m68k/m68k.h (TARGET_ISAB): New macro.
3859 * config/m68k/m68k.c: Use TARGET_ISAB rather than TARGET_CFV4.
3860 * config/m68k/m68k.md: Likewise.
3861
3862 2007-01-12 Julian Brown <julian@codesourcery.com>
3863
3864 * config/m68k/m68k.h (LEGITIMATE_INDEX_P, LEGITIMIZE_ADDRESS): Use
3865 TARGET_COLDFIRE_FPU instead of TARGET_CFV4E.
3866
3867 2007-01-12 Julian Brown <julian@codesourcery.com>
3868
3869 * config/m68k/m68k.h (TUNE_68040_60): New macro.
3870 * config/m68k/m68k.c (standard_68881_constant_p): Use it.
3871 * config/m68k/m68k.md: Likewise.
3872
3873 2007-01-12 Julian Brown <julian@codesourcery.com>
3874 Richard Sandiford <richard@codesourcery.com>
3875
3876 * config/m68k/m68k.h (TARGET_CPU_CPP_BUILTINS): Use TUNE_68030
3877 instead of TARGET_68030, TUNE_68040 instead of TARGET_68040,
3878 TUNE_68060 instead of TARGET_68060 and TUNE_CPU32 instead of
3879 TARGET_CPU32.
3880 (TARGET_CPU32): Rename to...
3881 (TUNE_CPU32): ...this.
3882 (TUNE_68000_10, TUNE_68030, TUNE_68040, TUNE_68060)
3883 (TUNE_CFV2): New macros.
3884 * config/m68k/netbsd-elf.h (LONG_DOUBLE_TYPE_SIZE): Simplify;
3885 remove conditions that are implied by TARGET_68020.
3886 * config/m68k/m68k.c (m68k_output_function_prologue): Use TUNE_68040
3887 instead of TARGET_68040 and TUNE_CPU32 instead of TARGET_CPU32.
3888 (m68k_output_function_epilogue): Likewise.
3889 (m68k_rtx_costs): Likewise. Use TUNE_68060 instead of TARGET_68060
3890 and TUNE_CFV2 instead of TARGET_5200. Use TUNE_68000_10 instead of
3891 "!TARGET_68020 && !TARGET_COLDFIRE" to choose between 68000 and
3892 non-68000 timings. Refactor multiplication and division costs.
3893 (output_addsi3): Use TUNE_68040 instead of TARGET_68040 and
3894 TUNE_CPU32 instead of TARGET_CPU32.
3895 (standard_68881_constant_p): Use TUNE_68040 instead of TARGET_68040
3896 and TUNE_68060 instead of TARGET_68060.
3897 * config/m68k/m68k.md: Use TUNE_68040 instead of TARGET_68040,
3898 TUNE_68060 instead of TARGET_68060, and TUNE_CPU32 instead of
3899 TARGET_CPU32.
3900 (movsi_const0): Use TUNE_68000_10 rather than "!TARGET_68020
3901 && !TARGET_COLDFIRE" to choose between moveq and clr.
3902 Likewise in the unnamed movsf pattern.
3903 (ashlsi_17_24, lshrsi_17_24): Guard with TUNE_68000_10 rather than
3904 "!TARGET_68020 && !TARGET_COLDFIRE". Likewise the unnamed
3905 ashiftrt pattern.
3906
3907 2007-01-12 Richard Sandiford <richard@codesourcery.com>
3908
3909 * config/m68k/m68k.h (TARGET_CPU_CPP_BUILTINS): Increase amount
3910 of tabbing before backslashes.
3911
3912 2007-01-11 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
3913
3914 * pa-linux.h (ASM_OUTPUT_INTERNAL_LABEL): Undefine.
3915 * pa.h (ASM_OUTPUT_LABEL): Output colon when using GAS.
3916 (ASM_OUTPUT_INTERNAL_LABEL): Define.
3917
3918 2007-01-11 Zdenek Dvorak <dvorakz@suse.cz>
3919
3920 * tree-ssa-loop-ivopts.c (extract_cond_operands): Split from
3921 find_interesting_uses_cond.
3922 (find_interesting_uses_cond): Use extract_cond_operands.
3923 (rewrite_use_compare): Use extract_cond_operands and
3924 force_gimple_operand_bsi. Do not call update_stmt.
3925 (determine_use_iv_cost_condition): Use extract_cond_operands.
3926 Return cheaper of using original bound and changing the exit bound.
3927
3928 2007-01-11 Zdenek Dvorak <dvorakz@suse.cz>
3929
3930 PR tree-optimization/29516
3931 * tree-ssa-address.c (tree_mem_ref_addr, add_to_parts,
3932 most_expensive_mult_to_index, addr_to_parts,
3933 create_mem_ref, maybe_fold_tmr): Make the type of
3934 fields of TARGET_MEM_REF sizetype.
3935 (move_fixed_address_to_symbol, move_pointer_to_base):
3936 New functions.
3937 * tree.def (TARGET_MEM_REF): Add comment on types of
3938 the operands.
3939
3940 2007-01-11 Joseph Myers <joseph@codesourcery.com>
3941
3942 * c-common.c (vector_types_convertible_p): Treat opaque types as
3943 always convertible if they have the same size, but not otherwise.
3944
3945 2007-01-11 Steven Bosscher <steven@gcc.gnu.org>
3946
3947 * ifcvt.c (struct noce_if_info): Add comments to the fields.
3948 Remove the b_unconditional field.
3949 (noce_try_sign_mask): Do not look at b_unconditional.
3950 (noce_process_if_block): Do not use merge_if_blocks. Update
3951 the CFG here. Do not set b_unconditional.
3952 (cond_move_process_if_block): Likewise.
3953 (find_cond_trap): Likewise.
3954 (check_cond_move_block): Require simple jump insns at the end
3955 of the basic block.
3956
3957 2007-01-11 Jan Hubicka <jh@suse.cz>
3958
3959 PR tree-optimization/1046
3960 * tree-tailcall.c (suitable_for_tail_call_opt_p): Use TREE_ADDRESSABLE
3961 when alias info is not ready.
3962 (pass_tail_recursion): Do not require aliasing.
3963 * tree-ssa-copyrename.c (pass_rename_ssa_cop): Likewise.
3964 * tree-ssa-ccp.c (pass_ccp, pass_fold_builtins): Likewise.
3965 * tree-ssa-copy.c (pass_copy_prop): Likewise.
3966 * tree-ssa-forwprop.c (pass_forwprop): Likewise.
3967 * tree-ssa-dce.c (pass_dce, pass_dce_loop, pass_cd_dce): Likewise.
3968 * passes.c (init_optimization_passes): Execute rename_ssa_copies,
3969 ccp, forwprop, copy_prop, merge_phi, copy_prop, dce and tail recursion
3970 before inlining.
3971 * tree-ssa-operands.c (add_virtual_operand, get_indirect_ref_operand):
3972 When aliasing is not build, mark statement as volatile.
3973
3974 2007-01-11 Tom Tromey <tromey@redhat.com>
3975
3976 PR preprocessor/15185, PR preprocessor/20989:
3977 * doc/cppopts.texi <-MT>: Update description of algorithm for
3978 computing default target.
3979 <-M, -MD>: Reword "basename" text.
3980
3981 2007-01-11 Roger Sayle <roger@eyesopen.com>
3982
3983 * builtins.c (expand_builtin_pow, expand_builtin_powi,
3984 fold_builtin_cabs, fold_builtin_sqrt, fold_builtin_trunc,
3985 fold_builtin_floor, fold_builtin_ceil, fold_builtin_round,
3986 fold_builtin_int_int_roundingfn, fold_builtin_bitop,
3987 fold_builtin_bswap, real_constp, fold_builtin_pow,
3988 fold_builtin_powi, fold_builtin_signbit, fold_builtin_copysign,
3989 do_mpfr_arg1, do_mpfr_arg2, do_mpfr_arg3, do_mpfr_sincos): Replace
3990 uses of the macro TREE_CONSTANT_OVERFLOW with TREE_OVERFLOW.
3991 * convert.c (convert_to_pointer): Likewise.
3992 * expr.c (highest_pow2_factor, expand_expr_real_1): Likewise.
3993 * fold-const.c (force_fit_type, fold_negate_expr, int_const_binop,
3994 const_binop, fold_convert_const_int_from_int,
3995 fold_convert_const_int_from_real,
3996 fold_convert_const_real_from_real, sign_bit_p,
3997 optimize_minmax_comparison, extract_muldiv_1, fold_div_compare,
3998 fold_sign_changed_comparison, fold_unary, fold_comparison,
3999 fold_binary, multiple_of_p, tree_Expr_non_zero_p,
4000 fold_negate_const, fold_abs_const, fold_not_const): Likewise.
4001 * print-tree.c (print_node_brief, print_node): Likewise.
4002 * stor-layout.c (place_field, layout_type): Likewise.
4003 * tree-chrec.c (keep_cast): Likewise.
4004 * tree.c (build_vector, build_real, build_real_from_int_cst,
4005 build_complex): Likewise.
4006
4007 2007-01-11 Roger Sayle <roger@eyesopen.com>
4008
4009 * tree.h (TREE_CONSTANT_OVERFLOW): Obsolete. For the time being,
4010 treat TREE_CONSTANT_OVERFLOW as a synonym of TREE_OVERFLOW.
4011
4012 2007-01-11 Paolo Bonzini <bonzini@gnu.org>
4013
4014 * configure.ac (strict1_warn): Rename to strict_warn.
4015 (WERROR, --enable-werror, symlink hacks, stage1_cflags,
4016 cc_set_by_configure, quoted_cc_set_by_configure,
4017 stage_prefix_set_by_configure, quoted_stage_prefix_set_by_configure,
4018 all_boot_languages, all_stagestuff): Remove.
4019 (target_list): Remove bootstrap targets.
4020 * Makefile.in (quickstrap): Unconditionally make a synonym of all.
4021 (BOOT_LANGUAGES, STAGE1_CFLAGS, STAGE1_CHECKING,
4022 REMAKEFLAGS, FLAGS_TO_PASS, PREPEND_DOTDOT_TO_RELATIVE_PATHS,
4023 SUBDIR_FLAGS_TO_PASS, WERROR_FLAGS, STRICT2_WARN, LANG_STAGESTUFF,
4024 VOL_FILES, POSTSTAGE1_FLAGS_TO_PASS, STAGE2_FLAGS_TO_PASS,
4025 STAGEPROFILE_FLAGS_TO_PASS, STAGEFEEDBACK_FLAGS_TO_PASS, stage1_build,
4026 stage1_copy, stage2_build, stage2_copy, stageprofile_build,
4027 stageprofile_copy, stage3_build, stage3_copy, stagefeedback_build,
4028 stagefeedback_copy, stage4_build, clean_s1, clean_sw, bootstrap,
4029 bootstrap-lean, bootstrap2, bootstrap2-lean, bootstrap3,
4030 bootstrap3-lean, bootstrap4, bootstrap4-lean, unstage1, unstage2,
4031 unstage3, unstage4, unstageprofile, unstagefeedback, restage, restage2,
4032 restage3, restage4, restageprofile, restagefeedback, bubbleestrap,
4033 cleanstrap, unstrap, restrap, *compare, *compare3, *compare4,
4034 *compare-lean, *compare3-lean, *compare4-lean, stage1-start, stage1,
4035 stage2-start, stage2, stage3-start, stage3, stage4-start, stage4,
4036 stageprofile-start, stageprofile, stagefeedback-start, stagefeedback,
4037 risky-stage1, risky-stage2, risky-stage3, risky-stage4): Remove.
4038 (ORDINARY_FLAGS_TO_PASS): Rename to FLAGS_TO_PASS.
4039 (STAGECOPYSTUFF, STAGEMOVESTUFF): Consolidate into MOSTLYCLEANFILES.
4040 (mostlyclean): Adjust.
4041 (clean, distclean): Don't mention bootstrap stuff.
4042 * configure: Regenerate.
4043 * ada/config-lang.in, cp/config-lang.in, forttran/config-lang.in,
4044 java/config-lang.in, objc/config-lang.in, objcp/config-lang.in,
4045 treelang/config-lang.in (stagestuff): Remove.
4046 * doc/sourcebuild.texi (stage1, stage2, stage3, stage4,
4047 stageprofile, stagefeedback, stagestuff): Remove mention.
4048
4049 2007-01-11 Nick Clifton <nickc@redhat.com>
4050
4051 * config/mcore/predicates.md (mcore_general_movesrc_operand):
4052 Accept CONSTs.
4053 (mcore_general_movdst_operand): Do not accept CONST_INTs.
4054 (mcore_arith_K_S_operand): Run the test for the S constraint not
4055 the test for the M constraint.
4056 (mcore_addsub_operand): Do not accept integer values that are
4057 larger than 32 bits.
4058 * config/mcore/mcore.md: Remove unused constraints from split.
4059 (andsi3): Use HOST_WIDE_INT instead of int to hold an INTVAL.
4060 (addsi3): Likewise.
4061 (allocate_stack): Likewise.
4062 * config/mcore/mcore.c (mcore_print_operand): Restrict output of P
4063 operands to 32 bits.
4064 (mcore_const_costs): Use HOST_WIDE_INT instead of int to hold an
4065 INTVAL.
4066 (mcore_and_cost, mcore_modify_comparison, const_ok_for_mcore,
4067 mcore_const_ok_for_inline, mcore_const_trick_uses_not,
4068 try_constant_tricks, mcore_num_ones, mcore_num_zeros,
4069 mcore_output_bclri, mcore_output_andn, output_inline_const,
4070 mcore_output_move, mcore_output_movedouble): Likewise.
4071 (mcore_output_cmov): Use CONST_OK_FOR_M and CONST_OK_FOR_N.
4072 (output_inline_const): Likewise.
4073 (output_inline_const): Fix format strings used in sprintf
4074 statements.
4075 * config/mcore/mcore-protos.h: Update prototypes for changed
4076 functions in mcore.c.
4077 * config/mcore/mcore.h (CONST_OK_FOR_I): Cast values to
4078 HOST_WIDE_INT and not int.
4079 (CONST_OK_FOR_J, CONST_OK_FOR_K, CONST_OK_FOR_L, CONST_OK_FOR_M,
4080 CONST_OK_FOR_N): Likewise.
4081 (LEGITIMATE_CONSTANT_P): Also check CONSTANT_P.
4082 (GO_IF_LEGITIMATE_INDEX): Use HOST_WIDE_INT instead of int to hold
4083 an INTVAL.
4084
4085 2007-01-10 Jan Hubicka <jh@suse.cz>
4086
4087 * tree-vrp.c (remove_range_assertions): Release defs.
4088 * tree-ssa-loop-ivopts.c (rmeove_statement): Likewise.
4089 * tree-ssa-dom.c (remove_stmt_or_phi): Likewise.
4090
4091 2007-01-10 Paul Brook <paul@codesourcery.com>
4092
4093 * config/arm/arm.c (arm_rtx_costs_1): Handle mutiply-subtract.
4094 * config/arm/arm.md (mulsi3subsi): New insn.
4095
4096 2007-01-10 Zdenek Dvorak <dvorakz@suse.cz>
4097
4098 * tree-ssa-loop-manip.c (tree_unroll_loop): Make it a wrapper over ...
4099 (tree_transform_and_unroll_loop): New.
4100 * tree-flow.h (transform_callback, tree_transform_and_unroll_loop):
4101 Declare.
4102
4103 2007-01-10 Robert Kennedy <jimbob@google.com>
4104
4105 * fold-const.c (fold_comparison): Fold comparisons like (x *
4106 1000 < 0) to (x < 0).
4107
4108 2007-01-10 Ian Lance Taylor <iant@google.com>
4109
4110 * tree-pretty-print.c (dump_generic_node): Print parentheses when
4111 operands have the same priority.
4112
4113 2007-01-10 Tom Tromey <tromey@redhat.com>
4114
4115 * fold-const.c (fold_truthop): Don't check can_use_bit_fields_p.
4116 (fold_binary): Likewise.
4117 * langhooks.c (lhd_can_use_bit_fields_p): Removed.
4118 * langhooks-def.h (lhd_can_use_bit_fields_p): Removed.
4119 (LANG_HOOKS_CAN_USE_BIT_FIELDS_P): Removed.
4120 (LANG_HOOKS_INITIALIZER): Remove LANG_HOOKS_CAN_USE_BIT_FIELDS_P.
4121 * langhooks.h (struct lang_hooks): Removed field
4122 'can_use_bit_fields_p'.
4123
4124 2007-01-10 Ralf Corsépius <ralf.corsepius@rtems.org>
4125
4126 * config/bfin/t-bfin, config/bfin/t-bfin-elf: Remove GCC_CFLAGS.
4127
4128 2007-01-10 Razya Ladelsky <razya@il.ibm.com>
4129
4130 * function.c (get_last_funcdef_no): New function.
4131 * function.h (get_last_funcdef_no): Declare.
4132 * tree-inline.c (initialize_cfun): Add initialization.
4133 (tree_function_versioning): Cleanup.
4134
4135 2007-01-10 Jan Hubicka <jh@suse.cz>
4136
4137 * tree-inline.c (setup_one_parameter): Do not propagate into abnormal
4138 PHIs.
4139
4140 2007-01-10 Sa Liu <saliu@de.ibm.com>
4141 Ben Elliston <bje@au.ibm.com>
4142
4143 * spu.h (STACK_SAVE_AREA): Use VOIDmode for SAVE_FUNCTION, SImode
4144 for SAVE_NONLOCAL and Pmode for any other save level.
4145 * spu-protos.h (spu_restore_stack_block): Declare.
4146 * spu.md (save_stack_block): Remove.
4147 (restore_stack_block): Call spu_restore_stack_block.
4148 * spu.c (spu_restore_stack_block): New function.
4149 (spu_expand_epilogue): Remove old comment.
4150
4151 2007-01-09 Zdenek Dvorak <dvorakz@suse.cz>
4152
4153 PR tree-optimization/30322
4154 * tree-ssa-loop-ivopts.c (fold_affine_expr, iv_value): Removed.
4155 (cand_value_at): Return the value as aff_tree.
4156 (may_eliminate_iv): Convert the bound from aff_tree to tree.
4157 * tree-affine.c (aff_combination_add_cst, aff_combination_add_product,
4158 aff_combination_mult): New functions.
4159 (aff_combination_add): Use aff_combination_add_cst.
4160 (aff_combination_convert): Allow conversions to a wider type.
4161 (tree_to_aff_combination): Handle BIT_NOT_EXPR.
4162 * tree-affine.h (aff_combination_mult): Declare.
4163
4164 2007-01-09 Carlos O'Donell <carlos@codesourcery.com>
4165
4166 * doc/tm.texi: Update documentation to reflect reality of exec
4167 and start file search behaviours. Update copyright year.
4168 * doc/invoke.texi: Explain how GCC_EXEC_PREFIX is used to find
4169 header file directories.
4170
4171 2007-01-09 Uros Bizjak <ubizjak@gmail.com>
4172
4173 * config/i386/i386.md (*sinxf2): Rename to *sinxf2_i387.
4174 (*cosxf2): Rename to cosxf2_i387.
4175 (*sindf2, *sinsf2): Extend operand 1 to XFmode. Macroize patterns
4176 using X87MODEF12 mode macro. Rename patterns to
4177 *sin_extend<mode>xf2_i387. Use SSE_FLOAT_MODE_P to disable patterns
4178 for SSE math.
4179 (*cosdf2, *cossf2): Ditto.
4180 (sincosdf3, sincossf3): Ditto. Rewrite corresponding splitters
4181 to match extended input operands.
4182 (sincos<mode>3): New expander.
4183 (*sinextendsfdf2, *cosextendsfdf2, *sincosextendsfdf3): Remove
4184 insn patterns and corresponding splitters.
4185
4186 2007-01-09 Kaz Kojima <kkojima@gcc.gnu.org>
4187
4188 * config/sh/t-linux (TARGET_LIBGCC2_CFLAGS): Delete.
4189 (SHLIB_MAPFILES, SHLIB_LINK, SHLIB_INSTALL): Likewise.
4190
4191 2007-01-09 Nicolas Pitre <nico@cam.org>
4192
4193 PR target/30173
4194 * arm/ieee754-df.S (Lad_s): Also test the low word of X for zero.
4195
4196 2007-01-08 Geoffrey Keating <geoffk@apple.com>
4197
4198 * target.h (struct gcc_target): New field library_rtti_comdat.
4199 * target-def.h (TARGET_CXX_LIBRARY_RTTI_COMDAT): New.
4200 (TARGET_CXX): Add TARGET_CXX_LIBRARY_RTTI_COMDAT.
4201 * doc/tm.texi (C++ ABI): Document TARGET_CXX_LIBRARY_RTTI_COMDAT.
4202 * config/darwin.h (TARGET_CXX_LIBRARY_RTTI_COMDAT): Define.
4203
4204 2007-01-08 Geoffrey Keating <geoffk@apple.com>
4205
4206 * doc/invoke.texi (Optimize Options): Correct description of -O0.
4207
4208 2007-01-08 Richard Guenther <rguenther@suse.de>
4209
4210 * tree.h (force_fit_type_double): Export.
4211 (force_fit_type): Remove.
4212 * fold-const.c (force_fit_type_double): New function.
4213 (force_fit_type): Remove.
4214 (int_const_binop): Use it.
4215 (fold_convert_const_int_from_int): Likewise.
4216 (fold_convert_const_int_from_real): Likewise.
4217 (fold_div_compare): Likewise.
4218 (fold_sign_changed_comparison): Likewise.
4219 (fold_unary): Likewise.
4220 (fold_negate_const): Likewise.
4221 (fold_abs_const): Likewise.
4222 (fold_not_const): Likewise.
4223 * c-common.c (shorten_compare): Use force_fit_type_double.
4224 * convert.c (convert_to_pointer): Likewise.
4225
4226 2007-01-08 Richard Guenther <rguenther@suse.de>
4227
4228 * tree.h (build_int_cst_wide_type): Export.
4229 * tree.c (build_int_cst_wide_type): New function.
4230 (build_int_cst_wide): Fix comment.
4231 * builtins.c (fold_builtin_object_size): Use build_int_cst
4232 to build -1 or 0 of the correct type. Use fit_double_type
4233 to check for overflow.
4234 * fold-const.c (optimize_bit_field_compare): Use build_int_cst_type
4235 to build the mask.
4236 (decode_field_reference): Likewise.
4237 (all_ones_mask_p): Likewise.
4238 (native_interpret_int): Use build_int_cst_wide_type.
4239 (fold_binary): Use build_int_cst_type to build an all-ones
4240 value.
4241 * stor-layout.c (set_sizetype): Use build_int_cst_wide_type.
4242
4243 2007-01-08 Daniel Jacobowitz <dan@codesourcery.com>
4244
4245 * config/pa/t-pa64 (libgcc_stub.a): Use $(T).
4246
4247 2007-01-09 Ben Elliston <bje@au.ibm.com>
4248
4249 * genautomata.c (STATS_OPTION): New option.
4250 (stats_flag): New flag.
4251 (gen_automata_option): Handle it.
4252 (initiate_automaton_gen): Ditto.
4253 (write_automata): Output statistics only if stats_flag is
4254 set. Likewise, output time statistics only if time_flag is set.
4255 * doc/md.texi (Processor pipeline description): Document new flag.
4256
4257 2007-01-08 Richard Guenther <rguenther@suse.de>
4258
4259 * builtins.c (fold_builtin_int_roundingfn): Use fit_double_type.
4260 * tree.c (build_int_cst_type): Likewise.
4261 (size_in_bytes): Don't call force_fit_type on the result.
4262 (int_fits_type_p): Use fit_double_type.
4263 * fold-const.c (fit_double_type): New function.
4264 (force_fit_type): Use it.
4265 * tree.h (fit_double_type): Export.
4266
4267 2007-01-08 Jan Hubicka <jh@suse.cz>
4268
4269 * tree-vectorizer.c (gate_increase_alignment): Fix return type.
4270 * ipa.c (function_and_variable_visibility): Fix return type.
4271
4272 2007-01-08 Richard Guenther <rguenther@suse.de>
4273
4274 * tree-ssa-ccp.c (maybe_fold_offset_to_array_ref): Use type
4275 of offset to build the index.
4276 * tree-pretty-print.c (dump_generic_node): Don't build negated
4277 const just for printing.
4278 * c-pretty-print.c (pp_c_integer_constant): Likewise.
4279 * builtins.c (fold_builtin_int_roundingfn): Check if result
4280 fits the type by using force_fit_type and comparing the result.
4281 * predict.c (predict_loops): Use compare_tree_int for comparison.
4282 * tree.c (build_int_cst): Fall back to integer_type_node for
4283 NULL_TREE type.
4284 (build_int_cst_wide): Assert type is non-null.
4285
4286 2007-01-08 Roberto Costa <roberto.costa@st.com>
4287
4288 * tree-vrp.c (extract_range_from_cond_expr): New.
4289 (extract_range_from_expr): Handle COND_EXPR nodes used as expressions.
4290 * tree-ssa-ccp.c (get_maxval_strlen): Handle COND_EXPR nodes used
4291 as expressions.
4292 (fold_stmt): Bug fix, avoid infinite recursion when folding COND_EXPRs.
4293 * tree-ssa-forwprop.c (simplify_cond, forward_propagate_into_cond,
4294 tree_ssa_forward_propagate_single_use_vars): Handle COND_EXPR nodes
4295 used as expressions.
4296 * tree-object-size.c (cond_expr_object_size): New.
4297 (collect_object_sizes_for): Handle COND_EXPR nodes used as expressions.
4298
4299 2007-01-08 Jan Hubicka <jh@suse.cz>
4300
4301 * tree-ssa-forwprop.c (forward_propagate_into_cond,
4302 tree_ssa_forward_propagate_single_use_va): Release defs of propagated
4303 statement.
4304
4305 2007-01-08 Richard Guenther <rguenther@suse.de>
4306
4307 PR tree-optimization/23603
4308 * tree-vrp.c (set_value_range_to_truthvalue): New function.
4309 (extract_range_from_binary): Fall back to truthvalue instead of
4310 varying for TRUTH_*_EXPR.
4311 (extract_range_from_comparison): Fall back to truthvalue instead of
4312 varying.
4313 (vrp_visit_phi_node): Don't adjust new range bounds to +INF/-INF
4314 if all visited PHI values were constant.
4315
4316 2007-01-08 Jan Hubicka <jh@suse.cz>
4317
4318 * cgraphunit.c (cgraph_process_new_functions): Reset reachable flag.
4319 (cgraph_analyze_function): break out from ...
4320 (cgraph_finalize_compilation_unit): ... here.
4321 (cgraph_expand_function): Remove forgoten commented out line.
4322 (cgraph_optimize): Analyze functions.
4323
4324 2007-01-08 Jan Hubicka <jh@suse.cz>
4325
4326 * tree-pas.h (TODO_remove_function): New flag.
4327 (TODO_update*): Renumber.
4328 (pass_ipa_increase_alignment,
4329 pass_ipa_function_and_variable_visibility): New passes.
4330 * cgraphunit.c (cgraph_increase_alignment): Move to tree-vectorizer.c
4331 (cgraph_function_and_variable_visibility): Move to ipa.c
4332 (cgraph_optimize): Don't call cgraph_function_and_variable_visibility,
4333 cgraph_increase_alignment.
4334 * ipa-inline.c (cgraph_decide_inlining): Don't push timevar.
4335 (cgraph_decide_inlining_incrementally): Push TV_INTEGRATION before
4336 calling tree-inline.
4337 (cgraph_early_inlining): Do not call cgraph_remove_unreachable_nodes.
4338 (pass_ipa_inline, pass_early_ipa_inlining): Set TODO_remove_functions
4339 * tree-vectorizer.c (increase_alignment): Move here from cgraphunit.c
4340 (gate_increase_alignment): New function.
4341 (pass_ipa_increase_alignment): New pass.
4342 * ipa.c: Inline tree-pass.h and timevar.h
4343 (function_and_variable_visibility): Move here from cgraphunit.c
4344 * tree-optimize.c (pass_early_local_passes): Add TODO_remove_functions.
4345 * passes.c (init_optimization_passes): Add the two new passes.
4346 (execute_todo): Handle cgraph_remove_functions.
4347
4348 2007-01-08 Nick Clifton <nickc@redhat.com>
4349
4350 * config/frv/predicates.md (reg_or_0_operand): Accept
4351 CONST_DOUBLEs.
4352
4353 2007-01-08 Ralf Corsépius <ralf.corsepius@rtems.org>
4354
4355 * config/bfin/rtems.h, config/bfin/t-rtems: New.
4356 * config.gcc: Add bfin*-rtems*.
4357
4358 2007-01-08 Mark Shinwell <shinwell@codesourcery.com>
4359
4360 * c.opt: Add -flax-vector-conversions.
4361 * c-typeck.c (convert_for_assignment): Pass flag to
4362 vector_types_convertible_p to allow emission of note.
4363 (digest_init): Likewise.
4364 * c-opts.c: Handle -flax-vector-conversions.
4365 * c-common.c (flag_lax_vector_conversions): New.
4366 (vector_types_convertible_p): Unless -flax-vector conversions
4367 has been passed, disallow conversions between vectors with
4368 differing numbers of subparts and/or element types. If such
4369 a conversion is disallowed, possibly emit a note on the first
4370 occasion only to inform the user of -flax-vector-conversions.
4371 The new last argument specifies this.
4372 * c-common.h (flag_lax_vector_conversions): New.
4373 (vector_types_convertible_p): Add extra argument.
4374 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use
4375 char_type_node for V*QI type vectors.
4376 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
4377 Update to satisfy new typechecking rules.
4378 * config/rs6000/altivec.h (vec_cmple): Use vec_cmpge, for both
4379 C and C++ variants.
4380 * doc/invoke.texi (C Dialect Options): Document
4381 -flax-vector-conversions.
4382
4383 2007-01-08 Mark Shinwell <shinwell@codesourcery.com>
4384
4385 PR tree-optimization/29877
4386 * tree-ssa-ter.c (is_replaceable_p): Deem assignments with
4387 a register variable on the RHS to not be replaceable.
4388
4389 2007-01-08 Chen Liqin <liqin@sunnorth.com.cn>
4390 * config/score/t-score-elf (MULTILIB_OPTIONS): Change.
4391 * config/score/predicates.md (const_uimm5, sr0_operand, const_simm12,
4392 const_simm15, const_pow2, const_npow2): Added.
4393 * config/score/misc.md (insv, extv, extzv, movmemsi,
4394 move_lbu_a/b, mov_lhu_a/b etc): Added and fix some bug.
4395 * config/score/score.c (score_address_cost, score_select_cc_mode):
4396 Added.
4397 Change CONST_OK_FOR_LETTER_P/EXTRA_CONSTRAINT define.
4398 Update score_rtx_costs for MACRO TARGET_RTX_COSTS.
4399 Update score_print_operand.
4400 * config/score/score.h (DATA_ALIGNMENT, SELECT_CC_MODE): Added.
4401 Adjust register allocate order and update some macro define.
4402 * config/score/score-mdaux.c (mdx_unaligned_load, mdx_unsigned_store,
4403 mdx_block_move_straight, mdx_block_move_loop_head,
4404 mdx_block_move_loop_body, mdx_block_move_loop_foot, mdx_block_move_loop,
4405 mdx_block_move): Added.
4406 (mdx_movsicc, mdp_select_add_imm, mdp_select, mds_zero_extract_andi,
4407 mdp_limm): Updated and fix some bug and typo.
4408 * config/score/score.md (movqi/hi/si, add/sub/zero/ext): Updated.
4409 (movsf, movdf, doloop_end): Added.
4410
4411 2007-01-08 Kazu Hirata <kazu@codesourcery.com>
4412
4413 * config/arm/arm.c, config/arm/arm.h, config/arm/arm.md,
4414 config/arm/thumb2.md: Fix comment typos.
4415 * doc/extend.texi: Fix a typo.
4416
4417 2007-01-07 Eric Christopher <echristo@apple.com>
4418
4419 * configure.ac: Check for __stack_chk_fail for darwin.
4420 * configure: Regenerate.
4421
4422 2007-01-07 Richard Guenther <rguenther@suse.de>
4423
4424 * tree-vrp.c (extract_range_from_assert): CSE calls to
4425 compare_values where possible.
4426 (extract_range_from_unary_expr): Likewise.
4427
4428 2007-01-07 Anatoly Sokolov <aesok@post.ru>
4429
4430 * config/avr/avr-protos.h (call_insn_operand): Delete prototype.
4431 * config/avr/avr.c (call_insn_operand): Delete function.
4432 * config/avr/avr.md (*pushqi, *pushhi, *pushsi, *pushsf): Use REG_SP
4433 instead of register number. Use predicates.
4434 * config/avr/predicates.md (const0_operand, reg_or_0_operand,
4435 call_insn_operand): Add.
4436
4437 2007-01-06 Jan Hubicka <jh@suse.cz>
4438
4439 * tree-pass.h (pass_build_cgraph_edges): Declare.
4440 * cgraphunit.c (record_refernece): Move to cgraphbuild.c
4441 (visited_nodes): Remove.
4442 (cgraph_create_edges): Move to cgraphbuild.c; rename to
4443 build_cgrpah_edges; make visited_nodes local.
4444 (cgraph_process_new_functions): DO not call initialize_inline_failed.
4445 (record_references_in_initializer): Move to cgraphbuild.c
4446 (initialize_inline_failed, rebuild_cgraph_edges,
4447 pass_rebuild_cgraph_edges): Move to cgraphbuild.c.
4448 (verify_cgraph_node): Make visited_nodes local.
4449 (cgraph_analyze_function): Do not call cgraph_create_edges and
4450 initialize_inline_failed.
4451 (cgraph_expand_function): Do not call cgraph_lower_function;
4452 assert that function is already lowered.
4453 * Makefile.in (cgraphbuild.o): New.
4454 * passes.c (init_optimization_passes): Add pass_build_cgraph_edges
4455 at the end of lowering passes.
4456
4457 2007-01-06 Steven Bosscher <steven@gcc.gnu.org>
4458
4459 * ifcvt.c (cond_move_convert_if_block): New function, code
4460 factored out from...
4461 (cond_move_process_if_block): ...here. Call the new function
4462 on the THEN and ELSE blocks.
4463 (merge_if_block): Do not copy global_live_at_end, merge_blocks
4464 already takes care of this.
4465
4466 2007-01-05 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
4467
4468 PR c/19978
4469 * tree.h (TREE_OVERFLOW_P): New.
4470 * c-typeck.c (parser_build_unary_op): Warn only if result
4471 overflowed and operands did not.
4472 (parser_build_binary_op): Likewise.
4473 (convert_for_assignment): Remove redundant overflow_warning.
4474 * c-common.c (overflow_warning): Don't check or set TREE_OVERFLOW.
4475
4476 2007-01-05 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
4477
4478 * c-typeck.c (store_init_value): Split over two lines to follow
4479 the GNU coding style.
4480
4481 2007-01-05 Benjamin Kosnik <bkoz@redhat.com>
4482
4483 * c-cppbuiltin.c (c_cpp_builtins): __GXX_EXPERIMENTAL_CPP0X__ to
4484 __GXX_EXPERIMENTAL_CXX0X__.
4485 * doc/cpp.texi: Same.
4486
4487 2007-01-05 Richard Guenther <rguenther@suse.de>
4488
4489 PR middle-end/27826
4490 * tree.c (get_narrower): Do not construct COMPONENT_REFs
4491 with mismatched types. Instead explicitly build a
4492 conversion NOP_EXPR.
4493
4494 2007-01-05 Ian Lance Taylor <iant@google.com>
4495
4496 * c-common.c (decl_with_nonnull_addr_p): New function.
4497 (c_common_truthvalue_conversion): Call it.
4498 * c-typeck.c (build_binary_op): Likewise.
4499 * c-common.h (decl_with_nonnull_addr_p): Declare.
4500
4501 2007-01-05 Jakub Jelinek <jakub@redhat.com>
4502
4503 PR c/30360
4504 * libgcc2.c (__divdc3): Compare c and d against 0.0 instead of
4505 denom against 0.0.
4506
4507 2007-01-05 Joel Brobecker <brobecker@adacore.com>
4508
4509 * doc/install.texi (Final install): Document the fact that
4510 the GNAT runtime should not be stripped.
4511
4512 2007-01-04 Jan Hubicka <jh@suse.cz>
4513
4514 * tree-inline.c (fold_marked_statements): Update operand caches
4515 and EH after folding
4516
4517 2007-01-04 Ian Lance Taylor <iant@google.com>
4518
4519 * c-common.c (check_function_nonnull): Whitespace fix.
4520
4521 2007-01-04 Jan Hubicka <jh@suse.cz>
4522
4523 * tree-optimize.c (execute_fixup_cfg): Correct previously mistakely
4524 comitted older version of patch.
4525 (pass_fixup_cfg): Add TODOs to verify flow and statements, dump
4526 function, celanup cfg and collect garbage.
4527
4528 2007-01-04 Mike Stump <mrs@apple.com>
4529
4530 * Makefile.in (mostlyclean): Don't remove libgcc anymore.
4531 (clean): Likewise.
4532
4533 2007-01-04 Eric Christopher <echristo@apple.com>
4534
4535 * libgcc2.c (__bswapsi2): Use SItype.
4536 (__bswapdi2): Use DItype.
4537 * libgcc2.h: Update for above.
4538
4539 2007-01-04 Paul Brook <paul@codesourcery.com>
4540
4541 * config/arm/arm.md (arm_mulsi3, thumb_mulsi3, mulsi3_compare0,
4542 mulsi_compare0_scratch, mulsi3addsi, mulsi3addsi_compare0,
4543 mulsi3addsi_compare0_scratch, mulsidi3adddi, mulsidi3,
4544 umulsidi3, umulsidi3adddi, smulsi3_highpart,
4545 umulsi3_highpart): Make conditional on !arm_arch6.
4546 (arm_mulsi3_v6, thumb_mulsi3_v6, mulsi3_compare0_v6,
4547 mulsi_compare0_scratch_v6, mulsi3addsi_v6, mulsi3addsi_compare0_v6,
4548 mulsi3addsi_compare0_scratch_v6, mulsidi3adddi_v6, mulsidi3_v6,
4549 umulsidi3_v6, umulsidi3adddi_v6, smulsi3_highpart_v6,
4550 umulsi3_highpart_v6): New insns.
4551
4552 2007-01-04 Roger Sayle <roger@eyesopen.com>
4553
4554 * fold-const.c (fold_convert): When casting an expression to void,
4555 fold_ignored_result may discover a GIMPLE_MODIFY_STMT which doesn't
4556 have a type. Instead of attempting to build a NOP_EXPR, return
4557 these "special" trees directly.
4558
4559 2007-01-04 Joseph Myers <joseph@codesourcery.com>
4560
4561 * config/rs6000/rs6000.c (rs6000_rtx_costs): Make adjustment for
4562 MULT inside MINUS as either argument. Use rs6000_cost->dmul -
4563 rs6000_cost->fp not 0 as adjustment for outer NEG.
4564
4565 2007-01-04 Jan Hubicka <jh@suse.cz>
4566
4567 * cgraph.c (cgraph_release_function_body): New function.
4568 (cgraph_remove_node): Use it.
4569 * cgraph.h (cgraph_release_function_body): Declare.
4570 * cgraphunit.c (cgraph_expand_function): Use it.
4571 * ipa.c (cgraph_remove_unreahchable_nodes): Use it.
4572 * tree-ssa.c (delete_tree_ssa): Allow to be called before aliasing
4573 is initialized and while compilation of other function is running.
4574 * tree-optimize.c (execute_free_cfg_annotations): Move code to clear
4575 statement CFG annotations from here to ...
4576 * tree-cfg.c (delete_tree_cfg_annotations): ... here.
4577
4578 2007-01-04 Zdenek Dvorak <dvorakz@suse.cz>
4579
4580 * cfgloop.h (enum li_flags): Make the constants powers of two.
4581
4582 2007-01-04 Jan Hubicka <jh@suse.cz>
4583
4584 * tree-inline.c (copy_bb): Insert new statements to statements_to_fold
4585 set.
4586 (fold_marked_statements): New function.
4587 (optimize_inline_calls, tree_function_versioning): Fold new statements.
4588 * tree-inline.h (copy_body_data): Add statements_to_fold.
4589
4590 2007-01-03 Daniel Jacobowitz <dan@codesourcery.com>
4591
4592 * config.gcc: Mention libgcc/config.host.
4593 * Makefile.in: Update comments mentioning libgcc.
4594 (LIBGCC, INSTALL_LIBGCC, GCC_PARTS, mklibgcc): Delete.
4595 (all.cross, start.encap, rest.encap, rest.cross): Update
4596 dependencies for libgcc move.
4597 (libgcc.mk, LIBGCC_DEPS, libgcov.a, libgcc.a, stmp-multilib)
4598 (clean-target, clean-target-libgcc): Delete.
4599 (srcdirify, GCC_EXTRA_PARTS): New macros.
4600 (libgcc-support, libgcc.mvars): New rules.
4601 (distclean): Remove mention of mklibgcc.
4602 (install): Don't reference INSTALL_LIBGCC.
4603 (install-common): Don't reference EXTRA_PARTS.
4604 (install-libgcc, install-multilib): Delete rules.
4605 * mklibgcc.in: Delete file.
4606 * doc/configfiles.texi: Don't mention mklibgcc.
4607
4608 * config/i386/t-darwin (SHLIB_VERPFX): Delete (moved to libgcc).
4609 * config/i386/t-darwin64 (SHLIB_VERPFX): Likewise.
4610 * config/rs6000/t-darwin (SHLIB_VERPFX): Likewise.
4611 * config/rs6000/t-ppccomm (TARGET_LIBGCC2_CFLAGS, SHLIB_MAPFILES)
4612 (mklibgcc, ldblspecs): Likewise.
4613
4614 * config/i386/t-nwld (libgcc.def, libc.def, libpcre.def)
4615 (posixpre.def): Use $(T).
4616 (SHLIB_EXT, SHLIB_NAME, SHLIB_SLIBDIR_QUAL, SHLIB_DEF, SHLIB_MAP)
4617 (SHLIB_SRC, SHLIB_INSTALL): Delete.
4618 (SHLIB_LINK): Make dummy.
4619 * config/t-slibgcc-darwin: Delete contents except for dummy SHLIB_LINK.
4620
4621 * config/frv/t-linux (EXTRA_MULTILIB_PARTS): Clear.
4622
4623 * config/alpha/t-crtfm: Use $(T) in rules for EXTRA_PARTS.
4624 * config/alpha/t-vms, config/alpha/t-vms64, config/fr30/t-fr30,
4625 config/i386/t-rtems-i386, config/ia64/t-ia64, config/rs6000/t-beos,
4626 config/rs6000/t-newas, config/sparc/t-elf: Likewise.
4627
4628 * configure.ac (all_outputs): Remove mklibgcc.
4629 * configure: Regenerated.
4630
4631 2007-01-03 Josh Conner <jconner@apple.com>
4632
4633 PR middle-end/29683
4634 * calls.c (compute_argument_addresses): Set stack and stack_slot
4635 for partial args, too.
4636 (store_one_arg): Use locate.size.constant for the size when
4637 generating a save_area.
4638
4639 2007-01-03 Robert Kennedy <jimbob@google.com>
4640
4641 * tree-cfg.c (tree_merge_blocks): Release SSA_NAME phi results
4642 whose definitions are deleted due to basic block merging.
4643
4644 2007-01-03 Paul Brook <paul@codesourcery.com>
4645
4646 PR target/16634
4647 * config/arm/arm.c (output_return_instruction): Pop PC in interrupt
4648 functions.
4649 (use_return_insn): Return 0 for Thumb interrupt functions.
4650 (print_multi_reg): Add rfe argument for IRQ returns.
4651 (arm_output_epilogue): Pop interrupt return address directly into PC.
4652 (arm_expand_prologue): Only adjust IRQ return address in Arm mode.
4653
4654 2007-01-03 Paul Brook <paul@codesourcery.com>
4655
4656 Merge from sourcerygxx-4_1.
4657 * config/arm/thumb2.md: New file.
4658 * config/arm/elf.h (JUMP_TABLES_IN_TEXT_SECTION): Return True for
4659 Thumb-2.
4660 * config/arm/coff.h (JUMP_TABLES_IN_TEXT_SECTION): Ditto.
4661 * config/arm/aout.h (ASM_OUTPUT_ADDR_VEC_ELT): Add !Thumb-2 assertion.
4662 (ASM_OUTPUT_ADDR_DIFF_ELT): Output Thumb-2 jump tables.
4663 * config/arm/aof.h (ASM_OUTPUT_ADDR_DIFF_ELT): Output Thumb-2 jump
4664 tables.
4665 (ASM_OUTPUT_ADDR_VEC_ELT): Add !Thumb-2 assertion.
4666 * config/arm/ieee754-df.S: Use macros for Thumb-2/Unified asm
4667 comptibility.
4668 * config/arm/ieee754-sf.S: Ditto.
4669 * config/arm/arm.c (thumb_base_register_rtx_p): Rename...
4670 (thumb1_base_register_rtx_p): ... to this.
4671 (thumb_index_register_rtx_p): Rename...
4672 (thumb1_index_register_rtx_p): ... to this.
4673 (thumb_output_function_prologue): Rename...
4674 (thumb1_output_function_prologue): ... to this.
4675 (thumb_legitimate_address_p): Rename...
4676 (thumb1_legitimate_address_p): ... to this.
4677 (thumb_rtx_costs): Rename...
4678 (thumb1_rtx_costs): ... to this.
4679 (thumb_compute_save_reg_mask): Rename...
4680 (thumb1_compute_save_reg_mask): ... to this.
4681 (thumb_final_prescan_insn): Rename...
4682 (thumb1_final_prescan_insn): ... to this.
4683 (thumb_expand_epilogue): Rename...
4684 (thumb1_expand_epilogue): ... to this.
4685 (arm_unwind_emit_stm): Rename...
4686 (arm_unwind_emit_sequence): ... to this.
4687 (thumb2_legitimate_index_p, thumb2_legitimate_address_p,
4688 thumb1_compute_save_reg_mask, arm_dwarf_handle_frame_unspec,
4689 thumb2_index_mul_operand, output_move_vfp, arm_shift_nmem,
4690 arm_save_coproc_regs, thumb_set_frame_pointer, arm_print_condition,
4691 thumb2_final_prescan_insn, thumb2_asm_output_opcode, arm_output_shift,
4692 thumb2_output_casesi): New functions.
4693 (TARGET_DWARF_HANDLE_FRAME_UNSPEC): Define.
4694 (FL_THUMB2, FL_NOTM, FL_DIV, FL_FOR_ARCH6T2, FL_FOR_ARCH7,
4695 FL_FOR_ARCH7A, FL_FOR_ARCH7R, FL_FOR_ARCH7M, ARM_LSL_NAME,
4696 THUMB2_WORK_REGS): Define.
4697 (arm_arch_notm, arm_arch_thumb2, arm_arch_hwdiv, arm_condexec_count,
4698 arm_condexec_mask, arm_condexec_masklen)): New variables.
4699 (all_architectures): Add armv6t2, armv7, armv7a, armv7r and armv7m.
4700 (arm_override_options): Check new CPU capabilities.
4701 Set new architecture flag variables.
4702 (arm_isr_value): Handle v7m interrupt functions.
4703 (user_return_insn): Return 0 for v7m interrupt functions. Handle
4704 Thumb-2.
4705 (const_ok_for_arm): Handle Thumb-2 constants.
4706 (arm_gen_constant): Ditto. Use movw when available.
4707 (arm_function_ok_for_sibcall): Return false for v7m interrupt
4708 functions.
4709 (legitimize_pic_address, arm_call_tls_get_addr): Handle Thumb-2.
4710 (thumb_find_work_register, arm_load_pic_register,
4711 legitimize_tls_address, arm_address_cost, load_multiple_sequence,
4712 emit_ldm_seq, emit_stm_seq, arm_select_cc_mode, get_jump_table_size,
4713 print_multi_reg, output_mov_long_double_fpa_from_arm,
4714 output_mov_long_double_arm_from_fpa, output_mov_double_fpa_from_arm,
4715 output_mov_double_fpa_from_arm, output_move_double,
4716 arm_compute_save_reg_mask, arm_compute_save_reg0_reg12_mask,
4717 output_return_instruction, arm_output_function_prologue,
4718 arm_output_epilogue, arm_get_frame_offsets, arm_regno_class,
4719 arm_output_mi_thunk, thumb_set_return_address): Ditto.
4720 (arm_expand_prologue): Handle Thumb-2. Use arm_save_coproc_regs.
4721 (arm_coproc_mem_operand): Allow POST_INC/PRE_DEC.
4722 (arithmetic_instr, shift_op): Use arm_shift_nmem.
4723 (arm_print_operand): Use arm_print_condition. Handle '(', ')', '.',
4724 '!' and 'L'.
4725 (arm_final_prescan_insn): Use extract_constrain_insn_cached.
4726 (thumb_expand_prologue): Use thumb_set_frame_pointer.
4727 (arm_file_start): Output directive for unified syntax.
4728 (arm_unwind_emit_set): Handle stack alignment instruction.
4729 * config/arm/lib1funcs.asm: Remove default for __ARM_ARCH__.
4730 Add v6t2, v7, v7a, v7r and v7m.
4731 (RETLDM): Add Thumb-2 code.
4732 (do_it, shift1, do_push, do_pop, COND, THUMB_SYNTAX): New macros.
4733 * config/arm/arm.h (TARGET_CPU_CPP_BUILTINS): Define __thumb2__.
4734 (TARGET_THUMB1, TARGET_32BIT, TARGET_THUMB2, TARGET_DSP_MULTIPLY,
4735 TARGET_INT_SIMD, TARGET_UNIFIED_ASM, ARM_FT_STACKALIGN, IS_STACKALIGN,
4736 THUMB2_TRAMPOLINE_TEMPLATE, TRAMPOLINE_ADJUST_ADDRESS,
4737 ASM_OUTPUT_OPCODE, THUMB2_GO_IF_LEGITIMATE_ADDRESS,
4738 THUMB2_LEGITIMIZE_ADDRESS, CASE_VECTOR_PC_RELATIVE,
4739 CASE_VECTOR_SHORTEN_MODE, ADDR_VEC_ALIGN, ASM_OUTPUT_CASE_END,
4740 ADJUST_INSN_LENGTH): Define.
4741 (TARGET_REALLY_IWMMXT, TARGET_IWMMXT_ABI, CONDITIONAL_REGISTER_USAGE,
4742 STATIC_CHAIN_REGNUM, HARD_REGNO_NREGS, INDEX_REG_CLASS,
4743 BASE_REG_CLASS, MODE_BASE_REG_CLASS, SMALL_REGISTER_CLASSES,
4744 PREFERRED_RELOAD_CLASS, SECONDARY_OUTPUT_RELOAD_CLASS,
4745 SECONDARY_INPUT_RELOAD_CLASS, LIBCALL_VALUE, FUNCTION_VALUE_REGNO_P,
4746 TRAMPOLINE_SIZE, INITIALIZE_TRAMPOLINE, HAVE_PRE_INCREMENT,
4747 HAVE_POST_DECREMENT, HAVE_PRE_DECREMENT, HAVE_PRE_MODIFY_DISP,
4748 HAVE_POST_MODIFY_DISP, HAVE_PRE_MODIFY_REG, HAVE_POST_MODIFY_REG,
4749 REGNO_MODE_OK_FOR_BASE_P, LEGITIMATE_CONSTANT_P,
4750 REG_MODE_OK_FOR_BASE_P, REG_OK_FOR_INDEX_P, GO_IF_LEGITIMATE_ADDRESS,
4751 LEGITIMIZE_ADDRESS, THUMB2_LEGITIMIZE_ADDRESS,
4752 GO_IF_MODE_DEPENDENT_ADDRESS, MEMORY_MOVE_COST, BRANCH_COST,
4753 ASM_APP_OFF, ASM_OUTPUT_CASE_LABEL, ARM_DECLARE_FUNCTION_NAME,
4754 FINAL_PRESCAN_INSN, PRINT_OPERAND_PUNCT_VALID_P,
4755 PRINT_OPERAND_ADDRESS): Adjust for Thumb-2.
4756 (arm_arch_notm, arm_arch_thumb2, arm_arch_hwdiv): New declarations.
4757 * config/arm/arm-cores.def: Add arm1156t2-s, cortex-a8, cortex-r4 and
4758 cortex-m3.
4759 * config/arm/arm-tune.md: Regenerate.
4760 * config/arm/arm-protos.h: Update prototypes.
4761 * config/arm/vfp.md: Enable patterns for Thumb-2.
4762 (arm_movsi_vfp): Add movw alternative. Use output_move_vfp.
4763 (arm_movdi_vfp, movsf_vfp, movdf_vfp): Use output_move_vfp.
4764 (thumb2_movsi_vfp, thumb2_movdi_vfp, thumb2_movsf_vfp,
4765 thumb2_movdf_vfp, thumb2_movsfcc_vfp, thumb2_movdfcc_vfp): New.
4766 * config/arm/libunwind.S: Add Thumb-2 code.
4767 * config/arm/constraints.md: Update include Thumb-2.
4768 * config/arm/ieee754-sf.S: Add Thumb-2/Unified asm support.
4769 * config/arm/ieee754-df.S: Ditto.
4770 * config/arm/bpabi.S: Ditto.
4771 * config/arm/t-arm (MD_INCLUDES): Add thumb2.md.
4772 * config/arm/predicates.md (low_register_operand,
4773 low_reg_or_int_operand, thumb_16bit_operator): New.
4774 (thumb_cmp_operand, thumb_cmpneg_operand): Rename...
4775 (thumb1_cmp_operand, thumb1_cmpneg_operand): ... to this.
4776 * config/arm/t-arm-elf: Add armv7 multilib.
4777 * config/arm/arm.md: Update patterns for Thumb-2 and Unified asm.
4778 Include thumb2.md.
4779 (UNSPEC_STACK_ALIGN, ce_count): New.
4780 (arm_incscc, arm_decscc, arm_umaxsi3, arm_uminsi3,
4781 arm_zero_extendsidi2, arm_zero_extendqidi2): New
4782 insns/expanders.
4783 * config/arm/fpa.md: Update patterns for Thumb-2 and Unified asm.
4784 (thumb2_movsf_fpa, thumb2_movdf_fpa, thumb2_movxf_fpa,
4785 thumb2_movsfcc_fpa, thumb2_movdfcc_fpa): New insns.
4786 * config/arm/cirrus.md: Update patterns for Thumb-2 and Unified asm.
4787 (cirrus_thumb2_movdi, cirrus_thumb2_movsi_insn,
4788 thumb2_cirrus_movsf_hard_insn, thumb2_cirrus_movdf_hard_insn): New
4789 insns.
4790 * doc/extend.texi: Document ARMv7-M interrupt functions.
4791 * doc/invoke.texi: Document Thumb-2 new cores+architectures.
4792
4793 2007-01-03 Jakub Jelinek <jakub@redhat.com>
4794
4795 * unwind-dw2.c (SIGNAL_FRAME_BIT, EXTENDED_CONTEXT_BIT): Define.
4796 (struct _Unwind_Context): Rename args_size to flags, remove
4797 signal_frame field, add a new args_size field and version field.
4798 (_Unwind_IsSignalFrame, _Unwind_SetSignalFrame,
4799 _Unwind_IsExtendedContext): New inline functions.
4800 (_Unwind_GetGR, _Unwind_SetGR, _Unwind_GetGRPtr, _Unwind_SetGRPtr):
4801 Assume by_value array is only present if _Unwind_IsExtendedContext.
4802 (_Unwind_GetIPInfo, execute_cfa_program, uw_frame_state_for): Use
4803 _Unwind_IsSignalFrame.
4804 (__frame_state_for): Initialize context.flags to EXTENDED_CONTEXT_BIT.
4805 (uw_update_context_1): Use _Unwind_SetSignalFrame.
4806 (uw_init_context_1): Initialize context->flags to
4807 EXTENDED_CONTEXT_BIT.
4808 * config/rs6000/linux-unwind.h (frob_update_context): Use
4809 _Unwind_SetSignalFrame.
4810
4811 2007-01-03 Andrew Pinski <andrew_pinski@playstation.sony.com>
4812
4813 PR middle-end/30353
4814 * gimplify.c (gimplify_modify_expr_complex_part): Move below
4815 tree_to_gimple_tuple. Call tree_to_gimple_tuple when we need
4816 the value.
4817
4818 2007-01-03 Kazu Hirata <kazu@codesourcery.com>
4819
4820 * config/i386/i386.h (NON_STACK_REG_P, REGNO_OK_FOR_SIREG_P,
4821 REGNO_OK_FOR_DIREG_P, REWRITE_ADDRESS, ASM_OPERAND_LETTER,
4822 RET, AT_SP): Remove.
4823 * config/i386/i386.md (*sse_prologue_save_insn): Use return
4824 instead of RET.
4825
4826 * alias.c (init_alias_analysis): Use VEC_safe_grow_cleared.
4827 * cfgbuild.c (find_basic_blocks): Likewise.
4828 * cfgrtl.c (rtl_create_basic_block): Likewise.
4829 * function.c (temp_slots_at_level): Likewise.
4830 * reg-stack.c (stack_regs_mentioned): Likewise.
4831 * regclass.c (allocate_reg_info): Likewise.
4832 * tree-cfg.c (init_empty_tree_cfg, build_tree_cfg, create_bb,
4833 set_bb_for_stmt, move_block_to_fn): Likewise.
4834 * tree-complex.c (tree_lower_complex): Likewise.
4835 * vec.h (VEC_safe_grow_cleared): New.
4836
4837 * cgraphunit.c, tree-ssa-alias.c: Fix comment typos.
4838
4839 2007-01-03 Zdenek Dvorak <dvorakz@suse.cz>
4840
4841 * loop-unswitch.c (unswitch_loop): Pass probabilities to loopify.
4842 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Pass probabilities
4843 to loop_version.
4844 * cfgloopmanip.c (scale_loop_frequencies): Export.
4845 (loopify): Scale the frequencies by prescribed coefficients.
4846 (set_zero_probability): New function.
4847 (duplicate_loop_to_header_edge): Improve updating of frequencies.
4848 (lv_adjust_loop_entry_edge, loop_version): Set probabilities
4849 and frequencies according to arguments.
4850 * tree-ssa-loop-manip.c (tree_unroll_loop): Set probabilities
4851 correctly.
4852 * cfg.c (scale_bbs_frequencies_int): Allow scaling the frequencies up.
4853 * modulo-sched.c (sms_schedule): Set probabilities for entering
4854 versioned loop correctly.
4855 * tree-vect-transform.c (vect_transform_loop): Ditto.
4856 * cfgloop.h (loopify, loop_version): Declaration changed.
4857 (scale_loop_frequencies): Declared.
4858
4859 2007-01-02 Jan Hubicka <jh@suse.cz>
4860
4861 * cgraph.c: Include tree-flow.h
4862 (cgraph_add_new-function): Handle IPA_SSA mode; execute
4863 early_local_passes.
4864 * cgraph.h (enum cgraph_state): Add CGRAPH_STATE_IPA_SSA.
4865 * tree-pass.h (pass_all_early_optimizations): Declare.
4866 * cgraphunit.c (cgraph_process_new_functions): Add IPA_SSA; execute
4867 early_local_passes.
4868 (cgraph_analyze_function): Do early_local_passes.
4869 * tree-mudflap.c (mf_decl_cache_locals, mf_build_check_statement_for):
4870 Do not add referenced vars.
4871 * tree-optimize.c (gate_all_optimizations): Do not execute when not in
4872 SSA form.
4873 (gate_all_early_local_passes): New gate.
4874 (pass_early_local_passes): Use new gate.
4875 (execute_early_local_optimizations): New functions.
4876 (gate_all_early_optimizations): New gate.
4877 (pass_all_early_optimizations): New pass.
4878 (execute_free_datastructures): Free SSA only when initialized.
4879 (gate_init_datastructures): Init only when optimizing.
4880 (tree_lowering_passes): Do early local passes when called late.
4881 * tree-profile.c (do_tree_profiling): Don't profile functions added
4882 late.
4883 (do_early_tree_profiling, pass_early_tree_profile): Kill.
4884 * tree-cfg.c (update_modified_stmts): Do not update when operands are
4885 not active.
4886 * passes.c (init_optimizations_passes): Reorder so we go into SSA
4887 during early_local_passes.
4888 * Makefile.in (cgraph.o): Add dependency on tree-flow.h.
4889
4890
4891 2007-01-02 Carlos O'Donell <carlos@codesourcery.com>
4892
4893 * Makefile.in: Update copyright year.
4894
4895 2007-01-02 Carlos O'Donell <carlos@codesourcery.com>
4896
4897 * Makefile.in: Export GCC_EXEC_PREFIX before calling $(RUNTEST)
4898 in $(lang_checks) and check-consistency targets.
4899
4900 2007-01-02 Jan Hubicka <jh@suse.cz>
4901
4902 * tree-mudflap.c (mf_decl_cache_locals, mf_build_check_statement_for):
4903 Do not add referenced vars.
4904 * tree-cfg.c (update_modified_stmts): Do not update when SSA operands
4905 are not active.
4906 * passes.c (init_optimization_passes): Put mudflap_2 after
4907 free_datastructures.
4908
4909 2007-01-02 Jan Hubicka <jh@suse.cz>
4910
4911 * tree-optimize (execute_fixup_cfg): Set after_inlining flag.
4912 Set NOTHROW flag on call statements proved to be nothrow.
4913 Update statement of local calls so new pure/const functions are
4914 updated. Update_ssa when in ssa form. Mark PHI nodes of nonlocal
4915 goto receivers.
4916 (tree_rest_of_compilation): Register hooks and initialize bitmap
4917 early. Do not set after_inlining flag.
4918
4919 2007-01-02 Steve Ellcey <sje@cup.hp.com>
4920
4921 * sbitmap.c (HOST_BITS_PER_LONG_LONG): Change to
4922 HOST_BITS_PER_LONGLONG
4923
4924 2007-01-02 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
4925
4926 PR c/19977
4927 * c-typeck.c (store_init_value): Don't emit pedantic overflow
4928 warning for non-static initializers.
4929
4930 2007-01-02 Steven Bosscher <steven@gcc.gnu.org>
4931
4932 * config/alpha/alpha.md, arm/arm.c, darwin.c, frv/frv.md,
4933 m32r/m32r.c, m32r/m32r.c, mn10300/mn10300.md, pa/pa.c,
4934 rs6000/rs6000.c, s390/s390.md, sh/sh.md, sparc/sparc.c:
4935 Always use set_unique_reg_note to add REG_EQUAL notes.
4936
4937 2007-01-02 Kazu Hirata <kazu@codesourcery.com>
4938
4939 Revert:
4940 2007-01-02 Kazu Hirata <kazu@codesourcery.com>
4941
4942 * alias.c (init_alias_analysis): Use VEC_safe_grow_cleared.
4943 * cfgbuild.c (find_basic_blocks): Likewise.
4944 * cfgrtl.c (rtl_create_basic_block): Likewise.
4945 * function.c (temp_slots_at_level): Likewise.
4946 * reg-stack.c (stack_regs_mentioned): Likewise.
4947 * regclass.c (allocate_reg_info): Likewise.
4948 * tree-cfg.c (init_empty_tree_cfg, build_tree_cfg, create_bb,
4949 set_bb_for_stmt, move_block_to_fn): Likewise.
4950 * tree-complex.c (tree_lower_complex): Likewise.
4951 * vec.h (VEC_safe_grow_cleared): New.
4952
4953 2007-01-02 Ian Lance Taylor <iant@google.com>
4954
4955 * c-common.c (c_common_truthvalue_conversion): When warning about
4956 using an assignment as a truth value, set TREE_NO_WARNING.
4957
4958 2007-01-02 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
4959
4960 PR middle-end/7651
4961 * c.opt (Wold-style-declaration): New.
4962 * doc/invoke.texi (C-only Warning Options): New.
4963 (Wold-style-declaration): Document it.
4964 (Wextra): Enabled by -Wextra.
4965 * c-opts.c (c_common_post_options): Enabled by -Wextra.
4966 * c-decl.c (declspecs_add_scspec): Replace -Wextra with
4967 -Wold-style-declaration.
4968
4969 2007-01-02 Kazu Hirata <kazu@codesourcery.com>
4970
4971 * alias.c (init_alias_analysis): Use VEC_safe_grow_cleared.
4972 * cfgbuild.c (find_basic_blocks): Likewise.
4973 * cfgrtl.c (rtl_create_basic_block): Likewise.
4974 * function.c (temp_slots_at_level): Likewise.
4975 * reg-stack.c (stack_regs_mentioned): Likewise.
4976 * regclass.c (allocate_reg_info): Likewise.
4977 * tree-cfg.c (init_empty_tree_cfg, build_tree_cfg, create_bb,
4978 set_bb_for_stmt, move_block_to_fn): Likewise.
4979 * tree-complex.c (tree_lower_complex): Likewise.
4980 * vec.h (VEC_safe_grow_cleared): New.
4981
4982 2007-01-02 Douglas Gregor <doug.gregor@gmail.com>
4983
4984 * c-common.c (c_common_nodes_and_builtins): Since variants of
4985 void_type_node get built before it is given a name, we need to
4986 give those variants the name, too.
4987 (complete_array_type): We need to work with the canonical main
4988 type of the array, from which we will build the qualified version.
4989 * params.def (PARAM_VERIFY_CANONICAL_TYPES): New.
4990 * print-tree.c (print_node): Display canonical type information
4991 for each type.
4992 * stor-layout.c (layout_type): When we don't know the
4993 alignment of a type for which we're building an array, we end up
4994 guessing wrong, so make the type require structural equality.
4995 * tree.c (make_node_stat): When we build a new type, it is its
4996 own canonical type.
4997 (build_type_attribute_qual_variant): When building an attribute
4998 variant, its canonical type is the non-attribute variant. However,
4999 if the attributes are target-dependent and they differ, we need to
5000 use structural equality checks for this type.
5001 (build_qualified_type): A qualified type is not equivalent to its
5002 unqualified variant; set the canonical type appropriately.
5003 (build_distinct_type_copy): When building a distinct type from
5004 another type, the new type is its own canonical type.
5005 (build_variant_type_copy): When building a new type variant, we
5006 assume that it is equivalent to the original type.
5007 (build_pointer_type_for_mode): When building a pointer type, also
5008 build a canonical type pointer.
5009 (build_reference_type_for_mode): When building a reference type,
5010 also build a canonical type reference.
5011 (build_index_type): When we can't hash an index type (e.g.,
5012 because its maximum value is negative), the index type requires
5013 structural equality tests.
5014 (build_array_type): Build the canonical form of an array type.
5015 (build_function_type): Function types require structural equality,
5016 because they contain default arguments, attributes, etc.
5017 (build_method_type_directly): Ditto for method types.
5018 (build_offset_type): Build the canonical offset type.
5019 (build_complex_type): Build the canonical vector type.
5020 (make_vector_type): Build the canonical vector type.
5021 * tree.h (TYPE_CANONICAL): New.
5022 (TYPE_STRUCTURAL_EQUALITY_P): New.
5023 (SET_TYPE_STRUCTURAL_EQUALITY): New.
5024 (struct tree_type): Added "canonical" field.
5025 * params.h (VERIFY_CANONICAL_TYPES): New.
5026 * doc/c-tree.texi (TYPE_CANONICAL): Document.
5027 (TYPE_STRUCTURAL_EQUALITY_P): Document.
5028 (SET_TYPE_STRUCTURAL_EQUALITY): Document.
5029 * doc/invoke.texi (verify-canonical-types): Document --param
5030 parameter for verifying canonical types.
5031
5032 2007-01-02 Joseph Myers <joseph@codesourcery.com>
5033
5034 * config.gcc (powerpc-*-eabispe*, powerpc-*-eabisimaltivec*,
5035 powerpc-*-eabisim*, powerpc-*-eabialtivec*, powerpc-*-eabi*,
5036 powerpc-*-rtems*, powerpc-wrs-vxworks, powerpc-wrs-vxworksae,
5037 powerpcle-*-eabisim*, powerpcle-*-eabi*): Add rs6000/e500.h to
5038 tm_file.
5039 * config/rs6000/e500.h: New.
5040 * config/rs6000/eabi.h (TARGET_SPE_ABI, TARGET_SPE, TARGET_E500,
5041 TARGET_ISEL, TARGET_FPRS, TARGET_E500_SINGLE, TARGET_E500_DOUBLE):
5042 Remove.
5043 * config/rs6000/linuxspe.h (TARGET_SPE_ABI, TARGET_SPE,
5044 TARGET_E500, TARGET_ISEL, TARGET_FPRS, TARGET_E500_SINGLE,
5045 TARGET_E500_DOUBLE): Remove.
5046 * config/rs6000/vxworks.h (TARGET_SPE_ABI, TARGET_SPE,
5047 TARGET_E500, TARGET_ISEL, TARGET_FPRS): Remove.
5048 * config/rs6000/rs6000.h (CHECK_E500_OPTIONS): Define.
5049 * config/rs6000/rs6000.c (rs6000_override_options): Use
5050 CHECK_E500_OPTIONS.
5051
5052 2007-01-02 Joseph Myers <joseph@codesourcery.com>
5053
5054 * config/rs6000/rs6000.c (print_operand): Check (TARGET_SPE ||
5055 TARGET_E500_DOUBLE), not TARGET_E500, for %y.
5056 (rs6000_generate_compare, rs6000_emit_sCOND, output_cbranch,
5057 rs6000_emit_cmove): Don't check TARGET_E500.
5058 * config/rs6000/rs6000.md (bunordered, bordered, sunordered,
5059 sordered): Don't check TARGET_E500.
5060
5061 2007-01-01 Eric Christopher <echristo@apple.com>
5062
5063 * config/mips/mips.c (mips_regno_mode_ok_for_base_p): Use
5064 HARD_REGISTER_NUM_P.
5065
5066 2007-01-01 Roger Sayle <roger@eyesopen.com>
5067
5068 * fold-const.c (fold_binary) <EQ_EXPR>: Fold "(X^C1) eq/ne C2" into
5069 "X eq/ne (C1^C2)". Fold "(X^Z) eq/ne (Y^Z)" as "X eq/ne Y" when Z
5070 has no side-effects. Fold "(X^C1) eq/ne (Y^C2)" as "(X^(C1^C2))
5071 eq/ne Y".
5072
5073 2007-01-01 Mike Stump <mrs@apple.com>
5074
5075 * configure.ac: Remove support for building with Apple's gcc-3.1.
5076
5077 2007-01-02 Joseph Myers <joseph@codesourcery.com>
5078
5079 PR middle-end/30311
5080 * caller-save.c (add_stored_regs): Only handle SUBREGs if inner
5081 REG is a hard register. Do not modify REG before calling
5082 subreg_nregs.
5083 * rtlanal.c (subreg_get_info): Don't assert size of XMODE is a
5084 multiple of the size of YMODE for certain lowpart cases.
5085
5086 2007-01-01 Andrew Pinski <pinskia@gmail.com>
5087
5088 PR middle-end/30253
5089 * gimplify (voidify_wrapper_expr): Update for
5090 GIMPLE_MODIFY_STMT.
5091
5092 2007-01-01 Andreas Schwab <schwab@suse.de>
5093
5094 PR target/29166
5095 * config/ia64/ia64.c (ia64_compute_frame_size): Account space for
5096 save of BR0 in extra_spill_size instead of spill_size.
5097 (ia64_expand_prologue): Save BR0 outside of the gr/br/fr spill
5098 area.
5099 (ia64_expand_epilogue): Restore BR0 from its new location.
5100
5101 2007-01-01 Andrew Pinski <pinskia@gmail.com>
5102
5103 * gimplify.c (gimplify_init_constructor <case VECTOR_TYPE>):
5104 Use a temporary variable if the left hand side is not a gimple
5105 register.
5106
5107 2007-01-01 Andrew Pinski <pinskia@gmail.com>
5108
5109 * gimplify.c (gimplify_return_expr): Make the temporary variable
5110 for the return expression, a gimple register variable.
5111
5112 2007-01-01 Jan Hubicka <jh@suse.cz>
5113
5114 * emit-rtl.c (emit_copy_of_insn_after): Do not call copy_insn_1 for
5115 INSN_LIST.
5116
5117 2007-01-01 Mike Stump <mrs@apple.com>
5118
5119 * configure.ac (HAVE_GAS_LITERAL16): Add autoconf check for
5120 .literal16.
5121 * config/darwin.c (machopic_select_rtx_section): Use
5122 HAVE_GAS_LITERAL16.
5123 (darwin_mergeable_constant_section): Likewise.
5124 * configure: Regenerate.
5125 * config.in: Regenerate.
5126
5127 2007-01-01 Jan Hubicka <jh@suse.cz>
5128 Andrew Pinski <pinskia@gmail.com>
5129
5130 * cgraphunit.c (cgraph_optimize): Call cgraph_add_new_functions
5131 before starting IPA passes.