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