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