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