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