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