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