99fcc33ed8c469c19c55dae9d82ae1f037a863a0
[gcc.git] / gcc / ChangeLog
1 2007-01-27 Steven Bosscher <steven@gcc.gnu.org>
2
3 * tracer.c (rest_of_handle_tracer): We already cleaned
4 up the CFG in tracer() so don't do it here again.
5 * cfgcleanup.c (rest_of_handle_jump2): Don't repeat
6 cleanup_cfg here, either. And don't call renumber_insns.
7
8 * cfgrtl.c (rtl_verify_flow_info_1): Don't verify that BB_END
9 and BB_HEAD are in the insn stream here. Instead make sure
10 that BB_INSN is valid on all insns. Also, do check here that
11 there are no pending branch predictions...
12 (rtl_verify_flow_info): ...instead of doing it here. Checks
13 for BB_END and BB_HEAD moved from rtl_verify_flow_info_1 to
14 here.
15
16 2007-01-26 Roger Sayle <roger@eyesopen.com>
17
18 * config/i386/i386.c (ix86_swap_binary_operands_p): New helper
19 function to simplify/factorize operand order canonicalization.
20 (ix86_fixup_binary_operands): Reorganize using the above function.
21 (ix86_binary_operator_ok): Likewise.
22
23 2007-01-27 Jakub Jelinek <jakub@redhat.com>
24
25 * genattrtab.c (struct attr_value_list, insn_code_values): Move to
26 file scope from optimize_attrs.
27 (simplify_test_exp): If insn_code_values is not NULL, use it to speed
28 up search.
29 (optimize_attrs): Clear insn_code_values after freeing it.
30
31 2007-01-26 Zdenek Dvorak <dvorakz@suse.cz>
32
33 * tree-ssa-address.c (create_mem_ref): Remove ", bsi" from
34 a parts.base assignment.
35
36 2007-01-26 Zdenek Dvorak <dvorakz@suse.cz>
37
38 * tree-data-ref.c (dump_subscript): Use dump_conflict_function.
39 (compute_subscript_distance, initialize_data_dependence_relation,
40 finalize_ddr_dependent, analyze_ziv_subscript,
41 analyze_siv_subscript_cst_affine,
42 compute_overlap_steps_for_affine_univar,
43 compute_overlap_steps_for_affine_1_2, analyze_subscript_affine_affine,
44 analyze_siv_subscript, analyze_miv_subscript,
45 analyze_overlapping_iterations, subscript_dependence_tester_1,
46 compute_self_dependence, free_dependence_relation): Work
47 with affine_fn instead of chrecs.
48 (dump_affine_function, dump_conflict_function, affine_function_equal_p,
49 common_affine_function, affine_function_base,
50 affine_function_constant_p, affine_fn_op, affine_fn_plus,
51 affine_fn_minus, affine_fn_free, conflict_fn_not_known,
52 conflict_fn_no_dependence, free_conflict_function, free_subscripts,
53 conflict_fn, affine_fn_cst, affine_fn_univar): New functions.
54 (all_chrecs_equal_p): Removed.
55 * tree-data-ref.h (affine_fn, conflict_function): New types.
56 (struct subscript): Change type of conflicting_iterations_in_a
57 and conflicting_iterations_in_b.
58
59 2007-01-26 Steve Ellcey <sje@cup.hp.com>
60
61 PR other/30182
62 * config/pa/pa.h (TARGET_HPUX_11): New.
63 * config/pa/pa-hpux11.h (TARGET_HPUX_11): New.
64 * config/pa/pa.c (pa_init_builtins): Use TARGET_HPUX_11.
65
66 2007-01-26 Daniel Berlin <dberlin@dberlin.org>
67 Richard Guenther <rguenther@suse.de>
68
69 * tree-ssa-structalias.c (solve_graph): Handle case
70 we merged the variable to another.
71
72 2007-01-25 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
73
74 * builtins.c (fold_builtin_1): Treat ccos and ccosh as 'even'
75 functions.
76
77 * fold-const.c (negate_mathfn_p): Treat casin, casinh, catan,
78 catanh, cproj, csin, csinh, ctan and ctanh as 'odd' functions.
79
80 2007-01-25 DJ Delorie <dj@redhat.com>
81
82 * config/m32c/m32c.c (m32c_cannot_change_mode_class): We don't
83 allow changes to modes which don't fit in those registers.
84
85 * reload1.c (choose_reload_regs): Check for invalid subregs before
86 computing their locations, not after.
87
88 2007-01-25 Geoffrey Keating <geoffk@apple.com>
89
90 PR 25127
91 * config/rs6000/rs6000.c (first_altivec_reg_to_save): On Darwin,
92 save Altivec registers in an eh_return function.
93 (compute_vrsave_mask): Likewise.
94 (rs6000_stack_info): Correct AIX/Darwin stack alignment computation
95 for saving Altivec registers.
96 (rs6000_emit_prologue): Don't allocate stack twice in
97 eh_return function. Correct expected value of altivec_save_offset
98 when using save_world. Describe save of R0 to stack when using
99 save_world. Describe stack pointer adjustment when using
100 save_world. Remove duplicated eh_return parameter register saving.
101 Update sp_offset variable after save_world.
102 * config/rs6000/t-darwin (LIB2FUNCS_STATIC_EXTRA): Remove
103 darwin-world.asm.
104 (LIB2FUNCS_EXTRA): Add darwin-world.asm.
105 * config/rs6000/darwin.h (SUBTARGET_OVERRIDE_OPTIONS): -m64
106 implies Altivec.
107
108 2007-01-25 Steve Ellcey <sje@cup.hp.com>
109
110 * config.gcc (ia64*-*-hpux*): Make posix threads the default.
111
112 2007-01-25 Steve Ellcey <sje@cup.hp.com>
113
114 PR other/30182
115 * config/pa/pa.c (pa_init_builtins): Set asm names for finite routines.
116 * config/ia64/ia64.c (ia64_init_builtins): Ditto.
117
118 2007-01-25 Richard Guenther <rguenther@suse.de>
119
120 * doc/invoke.texi (-Wcoverage-mismatch): Document.
121 * common.opt (-Wcoverage-mismatch): New warning option.
122 * coverage.c (get_coverage_counts): Ignore coverage mismatch
123 if -Wcoverage-mismatch is given.
124
125 2007-01-25 Razya Ladelsky <razya@il.ibm.com>
126
127 * ipa-cp.c (ipcp_insert_stage, ipcp_driver): Support for SSA.
128 (ipcp_driver): Change to static definition.
129 Add dumping of the ifunctions.
130 (constant_val_insert): Remove unused parameter. Support for SSA.
131 (ipcp_propagate_const): Support for SSA.
132 (ipcp_profile_bb_print): Print only analyzed nodes.
133 (ipcp_replace_map_create): Remove support for Fortran constant
134 for now.
135 * ipa-prop.c (ipa_method_modify_stmt,
136 ipa_callsite_compute_param): Support for SSA.
137 * ipa-prop.h (ipcp_driver): Remove declaration.
138 (IS_VALID_TREE_MAP_INDEX): Add define.
139
140 2007-01-24 Geoffrey Keating <geoffk@apple.com>
141
142 * unwind-dw2.c (execute_stack_op): Handle DW_OP_swap.
143
144 2007-01-24 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
145
146 PR middle-end/30447
147 * builtins.c (fold_builtin_cabs): Use MPFR to evaluate a
148 constant argument to cabs and do it without checking for
149 -funsafe-math-optimizations.
150
151 2007-01-24 Douglas Gregor <dgregor@osl.iu.edu>
152
153 * c-common.h (RID_FIRST_CXX0X): New.
154 (RID_LAST_CXX0X): New.
155 * c-opts.c (c_common_handle_option): -Wc++0x-compat is triggered
156 by -Wall.
157 * c.opt (Wc++0x-compat): New.
158 * doc/invoke.texi (-Wc++0x-compat): Document.
159
160 2007-01-24 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
161
162 * builtins.c (fold_builtin_carg): New.
163 (fold_builtin_1): Use it.
164
165 2007-01-24 Jan Hubicka <jh@suse.cz>
166
167 * ipa-inline.c (cgraph_decide_inlining): Initialize initial_insns.
168
169 * ipa-inline.c (initial_insns, max_insns): Delete.
170 (compute_max_insns): New function.
171 (cgraph_decide_inlining_of_small_function): Use it; take minimal amount
172 of insns as base for code growth.
173 (cgraph_decide_inlining): Make initial_insns local; do not compute
174 max_insns.
175 * params.def (PARAM_INLINE_UNIT_GROWTH): Set to 60.
176 * doc/invoke.texi (inline-unit-growth): Update docs.
177
178 2007-01-24 Jakub Jelinek <jakub@redhat.com>
179
180 * config/i386/i386.h (x86_cmpxchg16b): Remove const.
181 (TARGET_CMPXCHG16B): Define to x86_cmpxchg16b.
182 * config/i386/i386.c (x86_cmpxchg16b): Remove const.
183 (override_options): Add PTA_CX16 flag. Set x86_cmpxchg16b
184 for CPUs that have PTA_CX16 set.
185
186 PR middle-end/27416
187 * gimplify.c (omp_check_private): New function.
188 (gimplify_scan_omp_clauses): Use it for
189 firstprivate/lastprivate/reduction.
190
191 PR middle-end/30494
192 * gimplify.c (omp_add_variable): Don't call omp_notice_variable
193 on TYPE_SIZE_UNIT for GOVD_LOCAL VLAs.
194
195 PR middle-end/30421
196 * omp-low.c (lower_omp_for_lastprivate): Add dlist argument.
197 If lower_lastprivate_clauses emits some statements, append them
198 to dlist rather than body_p and to body_p append an initializer.
199 (lower_omp_for): Adjust caller.
200
201 2007-01-24 Steve Ellcey <sje@cup.hp.com>
202
203 * target.h (globalize_decl_name): New.
204 * target-def.h (TARGET_ASM_GLOBALIZE_DECL_NAME): New.
205 * output.h (default_globalize_decl_name): New.
206 * varasm.c (asm_output_bss): Use globalize_decl_name instead of
207 globalize_label.
208 (globalize_decl): Ditto.
209 (default_globalize_decl_name): New.
210 * config/ia64/ia64.c (ia64_globalize_decl_name): New.
211 (ia64_handle_version_id_attribute): New.
212 (TARGET_ASM_GLOBALIZE_DECL_NAME): New.
213 (ia64_asm_output_external): Use globalize_decl_name instead
214 of globalize_label.
215 * doc/extend.texi (version_id): New pragma.
216 * doc/tm.texi (ARGET_ASM_GLOBALIZE_DECL_NAME): New target hook.
217
218 2007-01-24 Andreas Krebbel <krebbel1@de.ibm.com>
219
220 * unwind-dw2-fde.c (get_cie_encoding): Replaced _Unwind_Word with
221 _uleb128_t and _Unwind_SWord with _sleb128_t.
222 * unwind-dw2.c (extract_cie_info, execute_stack_op, execute_cfa_program,
223 uw_frame_state_for, uw_update_context_1): Likewise.
224 * unwind-c.c (parse_lsda_header, PERSONALITY_FUNCTION): Likewise.
225 * unwind-pe.h (read_uleb128, read_sleb128,
226 read_encoded_value_with_base): Likewise.
227 * unwind-generic.h: Define _sleb128_t and _uleb128_t types.
228
229 2007-01-24 Richard Guenther <rguenther@suse.de>
230
231 * builtins.c (expand_builtin_cexpi): Get the fndecl
232 for cexp in the correct way.
233
234 2007-01-24 Jan Hubicka <jh@suse.cz>
235
236 * tree-ssa-dce.c (eliminate_unnecesary_stmts): Remove dead LHS of calls.
237
238 2007-01-24 Andreas Krebbel <krebbel1@de.ibm.com>
239
240 * c-cppbuiltin.c (builtin_define_type_sizeof): New function.
241 (c_cpp_builtins): New builtin macros: __SIZEOF_INT__, __SIZEOF_LONG__,
242 __SIZEOF_LONG_LONG__, __SIZEOF_SHORT__, __SIZEOF_POINTER__,
243 __SIZEOF_FLOAT__, __SIZEOF_DOUBLE__, __SIZEOF_LONG_DOUBLE__,
244 __SIZEOF_SIZE_T__, __SIZEOF_WCHAR_T__, __SIZEOF_WINT_T__ and
245 __SIZEOF_PTRDIFF_T__.
246 * doc/cpp.texi: Documentation for the new builtin macros added.
247
248 2007-01-24 Uros Bizjak <ubizjak@gmail.com>
249
250 * config/i386/i386.md (tanxf2, tan<mode>2, atan<mode>2, log<mode>2,
251 log10<mode>2, log2<mode>2, expxf2, exp10xf2, exp2xf2): Use op2
252 instead of operands[2] to avoid access past the end of array.
253
254 2007-01-24 Richard Sandiford <richard@codesourcery.com>
255
256 * reload1.c (emit_reload_insns): Pass the reload register
257 for a non-spill output reload through forget_old_reloads_1.
258
259 2007-01-23 Joseph Myers <joseph@codesourcery.com>
260
261 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Define
262 _SOFT_DOUBLE if doubles use software floating-point.
263 * config/rs6000/libgcc-ppc-glibc.ver: Export additional long
264 double functions if _SOFT_DOUBLE, not _SOFT_FLOAT.
265 * config/rs6000/darwin-ldouble.c: Also compile functions for
266 hard-float without FPRs. Use fmsub function for all __NO_FPRS__
267 cases. Compile extra functions if _SOFT_DOUBLE, not _SOFT_FLOAT.
268 * config/rs6000/linuxspe.h (SUBSUBTARGET_OVERRIDE_OPTIONS): Remove
269 commented-out long double override.
270 (CPP_LONGDOUBLE_DEFAULT_SPEC): Likewise.
271 * config/rs6000/eabispe.h: Likewise.
272 * config/rs6000/rs6000.c (rs6000_override_options): Don't override
273 long double for non-SPE.
274 (rs6000_handle_option): Likewise.
275 (invalid_e500_subreg): Disallow more subregs involding DImode,
276 DFmode, TImode or TFmode.
277 (rs6000_legitimate_offset_address_p): Check TFmode offsets for
278 E500 double.
279 (legitimate_lo_sum_address_p): Also check for TFmode for E500
280 double.
281 (rs6000_legitimize_address): Also handle TFmode for E500 double.
282 (rs6000_legitimize_reload_address): Also handle TFmode for E500
283 double.
284 (rs6000_legitimate_address): Also check for TFmode for E500
285 double.
286 (rs6000_emit_move): Use DFmode subregs of TFmode for E500 double.
287 (spe_build_register_parallel): Handle TFmode and TCmode.
288 (rs6000_spe_function_arg): Handle TFmode and TCmode for E500
289 double.
290 (function_arg): Handle TFmode and TCmode for E500 double.
291 (rs6000_init_libfuncs): Initialize extra libfuncs for soft double
292 in general.
293 (print_operand): Handle TFmode and TImode for %y.
294 (rs6000_generate_compare): Handle TFmode comparisons for E500
295 double.
296 (spe_func_has_64bit_regs_p): Check for TFmode for E500 double.
297 (rs6000_function_value): Handle TFmode and TCmode for E500 double.
298 (rs6000_libcall_value): Handle TFmode and TCmode for E500 double.
299 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Check for
300 TFmode for E500 double.
301 * config/rs6000/rs6000.md (FP): Allow TF for E500 double.
302 (floatsidf2): Enable for E500 double.
303 (movtf_softfloat): Use rs6000_nonimmediate_operand.
304 (extenddftf2): Change to extenddftf2_fprs.
305 (extenddftf2): Call gen_spe_extenddftf2 or gen_extenddftf2_fprs
306 depending on TARGET_E500_DOUBLE.
307 (extendsftf2): Enable for E500 double.
308 (trunctfdf2): Enable for E500 double.
309 (trunctfsf2): Change to trunctfsf2_fprs.
310 (trunctfsf2): Call gen_spe_trunctfsf2 or gen_trunctfsf2_fprs
311 depending on TARGET_E500_DOUBLE.
312 (floatsitf2): Enable for E500 double.
313 (fix_trunctfsi2): Change to fix_trunctfsi2_fprs.
314 (fix_trunctfsi2): Call gen_spe_fix_trunctfsi2 or
315 gen_fix_trunctfsi2_fprs depending on TARGET_E500_DOUBLE.
316 (negtf2): Change to negtf2_internal.
317 (negtf2): New expander.
318 (abstf2): Enable for E500 double. Call gen_spe_abstf2_tst,
319 gen_spe_abstf2_cmp or gen_abstf2_internal depending on
320 TARGET_E500_DOUBLE and flag_unsafe_math_optimizations.
321 (movdi_internal32): Use rs6000_nonimmediate_operand.
322 (unnamed splitter): Likewise.
323 * config/rs6000/spe.md (CMPTFEQ_GPR, TSTTFEQ_GPR, CMPTFGT_GPR,
324 TSTTFGT_GPR, CMPTFLT_GPR, TSTTFLT_GPR): New unspecs.
325 (SPE64TF, DITI): New mode macros.
326 (frob_df_di): Change to frob_<SPE64:mode>_<DITI:mode>; allow more
327 modes.
328 (frob_tf_ti): New.
329 (frob_<mode>_di_2): New.
330 (frob_tf_di_8_2): New.
331 (frob_di_df): Change to frob_di_<mode>; allow more modes.
332 (frob_ti_tf): New.
333 (frob_di_df_2): Change to frob_<DITI:mode>_<SPE64:mode>_2; allow
334 more modes.
335 (frob_ti_<mode>_8_2): New.
336 (frob_ti_tf_2): New.
337 (mov_si<mode>_e500_subreg0, mov_si<mode>_e500_subreg0_2,
338 mov_si<mode>_e500_subreg4, mov_si<mode>_e500_subreg4_2): Allow
339 TFmode.
340 (mov_sitf_e500_subreg8, mov_sitf_e500_subreg8_2,
341 mov_sitf_e500_subreg12, mov_sitf_e500_subreg12_2): New.
342 (spe_trunctfdf2_internal1, spe_trunctfsf2, spe_extenddftf2,
343 spe_fix_trunctfsi2, spe_fix_trunctfsi2_internal,
344 spe_negtf2_internal, spe_abstf2_cmp, spe_abstf2_tst): New.
345 (cmptfeq_gpr, tsttfeq_gpr, cmptfgt_gpr, tsttfgt_gpr, cmptflt_gpr,
346 tsttflt_gp): New.
347
348 2007-01-23 Ian Lance Taylor <iant@google.com>
349
350 * Makefile.in (OBJS-common): Reformat, alphabetize, but put
351 insn-*.o first.
352 (OBJS-archive): Reformat, alphabetize.
353 (OBJS): Change out_object_file to OBJS-md.
354
355 2007-01-23 Uros Bizjak <ubizjak@gmail.com>
356
357 * config/i386/i386.md: Use REG_P, MEM_P, CONST_INT_P, LABEL_P,
358 JUMP_P and CALL_P predicates where applicable.
359 * config/i386/i386.c: Ditto.
360 * config/i386/i386.md: Ditto.
361 * config/i386/mmx.md: Ditto.
362 * config/i386/predicates.md: Ditto.
363
364 2007-01-22 Andreas Schwab <schwab@suse.de>
365
366 * config/m68k/m68k.h: Fix comment.
367
368 2007-01-22 Jan Hubicka <jh@suse.cz>
369
370 * passes.c (init_optimization_passes): Do not rerun
371 pass_early_warn_uninitialized.
372
373 2007-01-22 Richard Guenther <rguenther@suse.de>
374
375 PR tree-optimization/30038
376 * tree-ssa-math-opts.c (maybe_record_sincos): New static helper
377 function.
378 (execute_cse_sincos_1): Likewise.
379 (execute_cse_sincos): Likewise.
380 (gate_cse_sincos): Likewise.
381 (pass_cse_sincos): New pass CSEing sin() and cos() calls using
382 the cexpi() canonicalization of sincos().
383 * tree-pass.h (pass_cse_sincos): Declare.
384 * passes.c (init_optimization_passes): New pass pas_cse_sincos.
385
386 2007-01-21 Eric Botcazou <ebotcazou@libertysurf.fr>
387
388 PR rtl-optimization/29329
389 * combine.c (replaced_rhs_insn): Rename to i2mod.
390 (replaced_rhs_value): Rename to i2mod_new_rhs.
391 (i2mod_old_rhs): New global variable.
392 (combine_instructions): Adjust for above change. Save a copy of
393 the old RHS into i2mod_old_rhs when the contents of a REG_EQUAL
394 note are substituted in the second instruction.
395 (distribute_notes) <REG_DEAD>: Adjust for above change. Do not
396 ditch the note if it pertains to the second eliminated register
397 and this register is mentioned in i2mod_old_rhs.
398
399 Revert:
400 2006-09-12 Eric Botcazou <ebotcazou@libertysurf.fr>
401
402 * combine.c (distribute_notes) <REG_DEAD>: Do not consider SETs past
403 the insn to which the note was originally attached.
404
405 2007-01-21 Jan Hubicka <jh@suse.cz>
406
407 * ipa-inline.c (inlining_mode): Comment, move up.
408 (cgraph_decide_inlining_incrementally): Do not perform inlining
409 itself; fix handling of flattening of self recursive functions.
410 (cgraph_find_cycles): Remove.
411 (cgraph_flatten_node): Remove.
412 (cgraph_decide_inlining): Use incremental inliner to handle flattening.
413 (try_inline): New function.
414 (cgraph_early_inlining): Update call of
415 cgraph_decide_inlining_incrementally. Apply inlining here.
416 (apply_inline): Update call of cgraph_decide_inlining_incrementally.
417
418 2007-01-21 Dirk Mueller <dmueller@suse.de>
419
420 PR bootstrap/30511
421 * tree-vrp.c (check_array_bounds): do not warn
422 about ADDR_EXPR's of ARRAY_REF's which are immediately
423 used in binary expressions.
424
425 2007-01-21 Ira Rosen <irar@il.ibm.com>
426
427 * tree-vectorizer.h (struct _stmt_vec_info): Add new field
428 read_write_dep and macros for its access.
429 * tree-vectorizer.c (new_stmt_vec_info): Initialize the new field.
430 * tree-vect-analyze.c (vect_analyze_data_ref_dependence): Remove
431 argument, call vect_check_interleaving for every independent pair of
432 data-refs. Mark loads that access the same memory location as a store
433 in the loop.
434 (vect_check_dependences): Remove.
435 (vect_analyze_data_ref_dependences): Remove vect_check_dependences
436 call, fix the call to vect_analyze_data_ref_dependence.
437 (vect_analyze_data_ref_access): For statements that access the same
438 data-ref, check that they are not stores; for loads, check that there
439 is no store that access the same location.
440
441 2007-01-20 Roger Sayle <roger@eyesopen.com>
442 Joseph Myers <joseph@codesourcery.com>
443
444 * doc/invoke.texi (-fdump-rtl-combine): Fix under/overfull hbox.
445 (-fdump-rtl-stack): Likewise.
446 (-fno-signed-zeros): Use @minus{} for a minus sign. Correct typo.
447 (-mcheck-zero-division, -mcpu): Fix under/overfull hbox.
448 (-mpt-fixed): Use @minus{} for minus sign.
449 (Using Precompiled Headers): Fix under/overfull hbox.
450
451 2007-01-20 Jan Hubicka <jh@suse.cz>
452
453 * tree-flow.h (struct stmt_ann_d): Move references_memory to proper
454 place within annotation.
455
456 2007-01-20 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
457
458 * pa.c (output_move_double): Change array size of xoperands to 4.
459
460 2007-01-20 Jan Hubicka <jh@suse.cz>
461
462 * tree-tailcall.c (adjust_return_value): Do not use RESULT_DECL
463 as temporary.
464
465 2007-01-19 Ian Lance Taylor <iant@google.com>
466
467 * expmed.c (expand_divmod) [TRUNC_MOD_EXPR, TRUNC_DIV_EXPR]: Cast
468 constant to unsigned HOST_WIDE_INT before negating.
469
470 2007-01-19 Ian Lance Taylor <iant@google.com>
471
472 * tree-ssa-operands.h (struct vuse_vec_d): Change num_vuse field
473 to unsigned.
474 (VUSE_VECT_ELEMENT) [ENABLE_CHECKING]: Use unsigned comparison.
475 (VUSE_ELEMENT_PTR) [ENABLE_CHECKING]: Likewise.
476 (SET_VUSE_VECT_ELEMENT) [ENABLE_CHECKING]: Likewise.
477 (SET_VUSE_ELEMENT_VAR) [ENABLE_CHECKING]: Likewise.
478 (SET_VUSE_ELEMENT_PTR) [ENABLE_CHECKING]: Likewise.
479 (realloc_vdef, realloc_vuse): Change second parameter to
480 unsigned.
481 (ssa_operand_iterator_d): Change vuse_index and mayuse_index
482 fields to unsigned.
483 * tree-ssa-operands.c (realloc_vop): Change num_elem parameter to
484 unsigned. Change x and lim locals to unsigned.
485 (realloc_vdef, realloc_vuse): Change num_elem parameter to
486 unsigned.
487 (finalize_ssa_vuse_ops): Change old_i local to unsigned.
488 (copy_virtual_operands): Change i and n locals to unsigned.
489
490 2007-01-19 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
491
492 PR middle-end/29335
493 * builtins.c (fold_builtin_1): Handle builtin fdim.
494
495 2007-01-20 Jan Hubicka <jh@suse.cz>
496
497 * tree-ssa.c (init_tree_ssa): Do not call init_alias_heapvars.
498 * tree-ssa-structalias.c (compute_points_to_sets): Do call
499 init_alias_heapvars.
500 (init_alias_heapvars): Initialize only when not already initialized.
501 (delete_alias_heapvars): Set heapvar_for_stmt to NULL.
502
503 2007-01-19 Roger Sayle <roger@eyesopen.com>
504
505 * common.opt (fsigned-zeros): New command line option.
506 * flags.h (HONOR_SIGNED_ZEROS): Control via flag_signed_zeros instead
507 of flag_unsafe_math_optimizations.
508 * opts.c (set_fast_math_flags): The -ffast-math command line option
509 implies -fno-signed-zeros.
510 (fast_math_flags_set_p): Likewise.
511
512 * doc/invoke.texi: Document new -fno-signed-zeros option, and update
513 the documentation of -ffast-math appropriately. Wrap long lines.
514
515 2007-01-19 Steve Ellcey <sje@cup.hp.com>
516
517 * system.h (ASM_MAKE_LABEL_LINKONCE): Poison.
518 * varasm.c (globalize_decl): Remove ASM_MAKE_LABEL_LINKONCE ifdef.
519
520 2007-01-19 Tomas Bily <tbily@suse.cz>
521
522 * cgraphunit.c (cgraph_finalize_function): Updating of pid
523 * tree-profile.c:
524 (tree_init_ic_make_global_vars): New function
525 (tree_init_edge_profiler): call of tree_init_ic_make_global_vars
526 (tree_gen_ic_profiler): New function
527 (tree_gen_ic_func_profiler): New function
528 (tree_profiling): Added calling of tree_gen_ic_func_profiler
529 (tree_profile_hooks): Added hook for indirec/virtual calls
530 * value-prof.c (tree_find_values_to_profile): New case for
531 indirect calls
532 (tree_values_to_profile): Call for determining indirect/virtual
533 counters
534 (tree_indirect_call_to_profile): New function
535 (tree_ic_transform): New function
536 (tree_ic): New function
537 (find_func_by_pid): New function
538 (init_pid_map): New function
539 (tree_value_profile_transformations): Added check for
540 indirect/virtual call transformation
541 * value-prof.h (enum hist_type): New counter type for
542 indirect/virtual calls
543 (profile_hooks): Added new hook for profiling indirect/virtual
544 calls
545 * profile.c (instrument_values): New case for indirect/virtual
546 call added
547 * gcov-io.h (GCOV_LAST_VALUE_COUNTER): Changed to 6
548 (GCOV_COUNTER_V_INDIR): New counter type
549 (GCOV_COUNTER_NAMES): New name of counter "indirect" added
550 (GCOV_MERGE_FUNCTIONS): New merge function for indirect/virtual
551 call added
552 * cgraph.c: Definition of cgraph_max_pid
553 (cgraph_create_node): Default init of pid attribute
554 * cgraph.h: Declaration of cgraph_max_pid
555 (struct cgraph_node): Added pid attribute
556 * libgcov.c (__gcov_indirect_call_profiler): New function
557 (__gcov_one_value_profiler_body): New function
558 (__gcov_one_value_profiler): Body was moved to
559 __gcov_one_value_profiler_body and calls it
560
561 2007-01-19 Basile Starynkevitch <basile@starynkevitch.net>
562
563 * doc/gty.texi (Options): Document the mark_hook option to GTY.
564 * gengtype.c (write_types_data, write_func_for_structure,
565 write_types, ggc_wtd, pch_wtd): Add skip_hooks to
566 write_types_data, ggc_wtd, pch_wtd for processing mark_hook.
567 (walk_type, write_func_for_structure): Generate the mark_hook if
568 needed.
569
570 2007-01-19 Jan Hubicka <jh@suse.cz>
571
572 * ipa-inline.c (cgraph_decide_inlining_incrementally): Instead of
573 'early' argument take inlining mode argument specifying whether to
574 inline for size/speeed or all functions; add support for flattening;
575 improve dumpting.
576 (cgraph_early_inlining): Update call of decide_inlining_incrementally.
577
578 2007-01-19 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
579
580 PR c++/17947
581 * toplev.c (warn_deprecated_use): Use %qD instead of %qs to print
582 the name of the declared identifier.
583
584 2007-01-19 Dirk Mueller <dmueller@suse.de>
585
586 * config/i386.h (CONDITIONAL_REGISTER_USAGE): Store
587 result of PIC_OFFSET_TABLE_REGNUM in temporary variable to avoid
588 duplicate evaluation.
589
590 2007-01-19 Uros Bizjak <ubizjak@gmail.com>
591
592 * config/i386/i386.md (acos<mode>2): Rename from acossf2 and acosdf2.
593 Macroize expander using X87MODEF12 mode macro. Extend operand 1
594 to XFMode, use acosxf2 and truncate result to requested mode.
595 Use SSE_FLOAT_MODE_P to disable patterns for SSE math.
596 (asin<mode>2): Similarly, with asin expanders.
597 (*fscalexf4_i387): Rename from *fscalexf4.
598 (expNcorexf3): New expander.
599 (expxf2, exp10xf2, exp2xf2): Use expNcorexf3 expander.
600 (exp<mode>2): Rename from expsf2 and expdf2. Macroize expander using
601 X87MODEF12 mode macro. Extend operand 1 to XFMode, use expxf2 and
602 truncate result to requested mode. Use SSE_FLOAT_MODE_P to disable
603 patterns for SSE math.
604 (exp10<mode>2): Similarly, with exp10 expanders.
605 (exp2<mode>2): Similarly, with exp2 expanders.
606 (expm1<mode>2): Similarly, with expm1 expanders.
607 (ldexp<mode>3): Similarly, with ldexp expanders.
608 (log<mode>2, log10<mode>2, log2<mode>2, log1p<mode>2, logb<mode>2):
609 Use gen_truncxf<mode>2_i387_noop to truncate result.
610
611 2007-01-19 Richard Sandiford <richard@codesourcery.com>
612
613 * config/m68k/m68k.h (PREFERRED_STACK_BOUNDARY): Define to 32
614 for ColdFire targets.
615
616 2007-01-19 Nathan Sidwell <nathan@codesourcery.com>
617 Richard Sandiford <richard@codesourcery.com>
618
619 * config/m68k/m68k.h (M68K_STATIC_CHAIN_REG_NAME): New macro.
620 (INITIALIZE_TRAMPOLINE): Use STATIC_CHAIN_REGNUM.
621 (__transfer_from_trampoline): Use M68K_STATIC_CHAIN_REG_NAME.
622 * config/m68k/m68kelf.h (STATIC_CHAIN_REGNUM): Override.
623 (M68K_STATIC_CHAIN_REG_NAME): Likewise.
624 * config/m68k/netbsd-elf.h (M68K_STATIC_CHAIN_REG_NAME): Likewise.
625
626 2007-01-19 Richard Sandiford <richard@codesourcery.com>
627
628 * config/m68k/m68k.md (adddi_dilshr32): Rename to...
629 (*adddi_dilshr32): ...this. Fix formatting. Remove commented-out
630 non-canonical pattern. Restrict to !TARGET_COLDFIRE.
631 (*adddi_dilshr32_cf): New pattern.
632 (adddi3, subdi3): Remove first alternatives.
633
634 2007-01-19 Richard Sandiford <richard@codesourcery.com>
635
636 * config/m68k/m68k.c (notice_update_cc): If an SFmode move is
637 implemented using move.l, do not use its cc result for floating-point
638 comparisons.
639
640 2007-01-19 Richard Sandiford <richard@codesourcery.com>
641
642 * config/m68k/m68k.h (EXTRA_CONSTRAINT): Stop the 'T' constraint
643 from accepting 's' constraints if flag_pic.
644
645 2007-01-19 Richard Sandiford <richard@codesourcery.com>
646
647 * config/m68k/m68k.md (bordered, bunordered, buneq, bunge, bungt)
648 (bunle, bunlt, bltgt, bordered_rev, bunordered_rev, buneq_rev)
649 (bunge_rev, bungt_rev, bunle_rev, bunlt_rev, bltgt_rev): Change
650 condition from TARGET_68881 to TARGET_HARD_FLOAT.
651
652 2007-01-19 Sandra Loosemore <sandra@codesourcery.com>
653
654 * longlong.h (count_leading_zeros, COUNT_LEADING_ZEROS_0): Add
655 ColdFire alternatives.
656 * config/m68k/m68k.h (CLZ_DEFINED_VALUE_AT_ZERO): New macro.
657 * config/m68k/m68k.md (clzsi2): Define for ColdFire
658 architectures that support ff1 instruction.
659
660 2007-01-19 Richard Sandiford <richard@codesourcery.com>
661 Julian Brown <julian@codesourcery.com>
662
663 * config/m68k/m68k.h (CONST_OK_FOR_LETTER_P): Add an 'R' case.
664 * config/m68k/m68k.md (*movsi_cfv4): Fold into...
665 (*movsi_cf): ...here. Remove unnecessary 'R' from 'Rg'.
666 Add commentary.
667
668 2007-01-19 Richard Sandiford <richard@codesourcery.com>
669
670 * config/m68k/m68k-protos.h (valid_mov3q_const): Take a HOST_WIDE_INT
671 and return a bool.
672 (output_move_const_into_data_reg, output_move_simode_const): Delete.
673 * config/m68k/m68k.c (const_method, const_int_cost): Take a
674 HOST_WIDE_INT instead of an rtx.
675 (m68k_rtx_costs): Update call accordingly.
676 (output_move_const_into_data_reg): Likewise. Fix formatting.
677 (valid_mov3q_const): Take a HOST_WIDE_INT instead of an rtx.
678 Return a bool.
679 (output_move_simode_const): Update calls after above changes.
680 Rework to use automatic variables and predicates like MEM_P.
681 * config/m68k/m68k.md (pushexthisi_const): Update call to
682 valid_mov3q_const.
683
684 2007-01-19 Dirk Mueller <dmueller@suse.de>
685
686 * tree-ssa-alias.c (perform_var_substitution): Fix typo
687 in dump_flags test.
688
689 2007-01-19 Richard Guenther <rguenther@suse.de>
690
691 * builtins.c (expand_builtin_cexpi): Fall back to expanding
692 via cexp in case sincos is not available.
693
694 2007-01-19 Richard Guenther <rguenther@suse.de>
695
696 * doc/tm.texi (TARGET_HAS_SINCOS): Document new target macro.
697 * defaults.h (TARGET_HAS_SINCOS): Default to off.
698 * config/linux.h (TARGET_HAS_SINCOS): Set to on if we have glibc.
699 * config/alpha/linux.h (TARGET_HAS_SINCOS): Likewise.
700 * config/sparc/linux.h (TARGET_HAS_SINCOS): Likewise.
701 * config/sparc/linux64.h (TARGET_HAS_SINCOS): Likewise.
702 * config/rs6000/linux.h (TARGET_HAS_SINCOS): Likewise.
703 * config/rs6000/linux64.h (TARGET_HAS_SINCOS): Likewise.
704
705 2007-01-19 Uros Bizjak <ubizjak@gmail.com>
706
707 * config/i386/i386.md (*fpatanxf3_i387, fpatan_extend<mode>xf3_i387):
708 New insn patterns.
709 (atan2sf3_1, atan2df3_1, atan2xf3_1): Remove insn patterns.
710 (atan2xf3): Directly generate RTL pattern.
711 (atan2<mode>3): Rename from atan2sf3 and atan2df3 and macroize insn
712 patterns using X87MODEF12 mode macro. Use fpatan_extend<mode>xf3_i387
713 and truncate result to requested mode. Use SSE_FLOAT_MODE_P to
714 disable patterns for SSE math.
715 (atan<mode>2): Rename from atansf2 and atandf2 and macroize insn
716 patterns using X87MODEF12 mode macro. Use fpatan_extend<mode>xf3_i387
717 and truncate result to requested mode. Use SSE_FLOAT_MODE_P to
718 disable patterns for SSE math.
719
720 2007-01-19 Alexandre Oliva <aoliva@redhat.com>
721
722 * libgcc-std.ver: Fix typo in %inherit for GCC_4.3.0.
723
724 2007-01-18 Roger Sayle <roger@eyesopen.com>
725
726 * fold-const.c (fold_unary) <VIEW_CONVERT_EXPR>: Optimize away a
727 VIEW_CONVERT_EXPR to the same type as it's operand.
728
729 2007-01-18 David Edelsohn <edelsohn@gnu.org>
730
731 * config/rs6000/darwin-ldouble.c: Only build _SOFT_FLOAT if
732 configured for long double 128.
733
734 2007-01-18 Mike Stump <mrs@apple.com>
735
736 * config/rs6000/rs6000.c (rs6000_emit_vector_compare): Fix build
737 error.
738
739 2007-01-18 Michael Meissner <michael.meissner@amd.com>
740
741 * i386.c (ix86_compute_frame_layout): Make fprintf's in #if 0 code
742 type correct.
743
744 2007-01-18 Jan Hubicka <jh@suse.cz>
745
746 * tree-ssa-operands.c (vop_free_bucket_size): Never return value
747 greater than NUM_VOP_FREE_BUCKETS.
748
749 2007-01-18 Daniel Berlin <dberlin@dberlin.org>
750
751 * tree-ssa-structalias.c: Update comments.
752 (ptabitmap_obstack): Removed.
753 (pta_obstack): New.
754 (oldpta_obstack): Ditto.
755 (stats): Add a few members.
756 (struct variable_info): Remove node, complex, address_taken, and
757 indirect_target members. Add oldsolution member.
758 (new_var_info): Do not initialize removed members.
759 (constraint_expr_type): Remove INCLUDES.
760 (constraint_graph): Add size, implicit_preds, rep,
761 indirect_cycles, eq_rep, label, direct_nodes, and complex members.
762 (FIRST_REF_NODE): New macro.
763 (LAST_REF_NODE): Ditto.
764 (FIRST_ADDR_NODE): Ditto.
765 (find): New function.
766 (unite): Ditto.
767 (dump_constraint): Do not handle INCLUDES.
768 (insert_into_complex): Do not insert duplicate constraints.
769 (condense_varmap_nodes): Renamed and rewritten into ...
770 (merge_node_constraints): This. Also fix bug in handling of
771 offseted copy constraints.
772 (clear_edges_for_node): No longer need to deal with preds at all,
773 or removing associated preds/succs.
774 (merge_graph_nodes): Deal with indirect_cycles.
775 Don't deal with predecessors.
776 (add_implicit_graph_edge): New function.
777 (add_pred_graph_edge): Ditto.
778 (add_graph_edge): Don't deal with predecessors.
779 (build_constraint_graph): Removed.
780 (build_pred_graph): New function.
781 (build_succ_graph): Ditto.
782 (struct scc_info): Removed in_component. Added roots, dfs, and
783 node_mapping. Remove visited_index, unification_queue.
784 (scc_visit): Deal with union-find we do now.
785 Deal with cycles with REF nodes.
786 (collapse_nodes): Renamed and rewritten to ...
787 (unify_nodes): This.
788 (process_unification_queue): Removed.
789 (topo_visit): Cleanup
790 (do_da_constraint): Use find.
791 (do_sd_constraint): Ditto.
792 (do_ds_constraint): Ditto.
793 (do_complex_constraint): Ditto.
794 (init_scc_info): Update for removed and added members.
795 (find_and_collapse_graph_cycles): Renamed and rewritten into ...
796 (find_indirect_cycles): This.
797 (equivalence_class): New variable.
798 (label_visit): New function.
799 (perform_variable_substitution): Rewritten.
800 (free_var_substitution_info): New function.
801 (find_equivalent_node): Ditto.
802 (move_complex_constraints): Ditto.
803 (eliminate_indirect_cycles): Ditto.
804 (solve_graph): Only propagate changed bits.
805 Use indirect cycle elimination.
806 Use find.
807 (tree_id_t): Rename to tree_vi_t, delete id member, add vi member.
808 (tree_id_eq): Renamed to ...
809 (tree_vi_eq): This. Update for member change
810 (insert_id_for_tree): Renamed and rewritten to ...
811 (insert_vi_for_tree): This.
812 (lookup_id_for_tree): Renamed and rewritten to ...
813 (lookup_vi_for_tree): This.
814 (get_id_for_tree): Renamed and rewritten to ...
815 (get_vi_for_tree): Ditto.
816 (get_constraint_exp_from_ssa_var): Update to use get_vi_for_tree.
817 (process_constraint): Don't handle INCLUDES.
818 Remove special ADDRESSOF case.
819 (find_func_aliases): Rewrite to use vi functions instead of id
820 ones.
821 (create_function_info_for): Ditto.
822 (create_variable_info_for): Ditto.
823 (intra_create_variable_infos): Ditto.
824 (merge_smts_into): Ditto.
825 (find_what_p_points_to): Ditto.
826 (init_base_vars): Ditto.
827 (init_alias_vars): Ditto.
828 (remove_preds_and_fake_succs): New function.
829 (dump_sa_points_to_info): Dump new stats.
830 (dump_solution_for_var): Use find.
831 (set_used_smts): Fix formatting.
832 (compute_points_to_sets): Updated for new functions.
833 (ipa_pta_execute): Ditto.
834
835 2007-01-18 Kazu Hirata <kazu@codesourcery.com>
836 Richard Sandiford <richard@codesourcery.com>
837
838 * doc/tm.texi (TARGET_FUNCTION_VALUE): Expand documentation of
839 parallels.
840 * calls.c (expand_call): If the return value is a PARALLEL,
841 extract its first member.
842 * config/m68k/linux.h (FUNCTION_EXTRA_EPILOGUE): Remove.
843 * config/m68k/m68k.c (m68k_output_function_epilogue): Don't
844 use FUNCTION_EXTRA_EPILOGUE.
845 (m68k_function_value): Return a PARALLEL if the return value
846 is of a pointer type.
847 * config/m68k/netbsd-elf.h (current_function_returns_pointer)
848 (FUNCTION_EXTRA_EPILOGUE): Remove.
849 * config/m68k/m68k.md (D0_REG): New constant.
850
851 2007-01-18 Kazu Hirata <kazu@codesourcery.com>
852
853 * config/m68k/m68k.c (m68k_output_function_epilogue): Don't
854 output a NOP for empty epilogues.
855
856 2007-01-18 Richard Sandiford <richard@codesourcery.com>
857
858 * config/m68k/m68k.c (m68k_use_return_insn): Update comments
859 before function. Extend register save check to include all
860 registers, not just integer ones.
861
862 2007-01-18 Kazu Hirata <kazu@codesourcery.com>
863
864 * config/m68k/m68k-protos.h (use_return_insn): Rename to...
865 (m68k_use_return_insn): ...this.
866 * config/m68k/m68k.h (USE_RETURN_INSN): Delete.
867 * config/m68k/m68k.c (use_return_insn): Rename to...
868 (m68k_use_return_insn): ...this.
869 * config/m68k/m68k.md (return): Use m68k_use_return_insn instead
870 of USE_RETURN_INSN.
871
872 2007-01-18 Nathan Sidwell <nathan@codesourcery.com>
873
874 * config/m68k/fpgnulib.c (__truncdfsf2): Implement round to
875 nearest even, fix denormal rounding overflow.
876
877 2007-01-18 Richard Sandiford <richard@codesourcery.com>
878
879 * config/m68k/m68k.md (movsf_cf_hard): Use fsmove instead of
880 f%$smove and f%$move.
881 (movdf_cf_hard): Use fdmove for cases 0 and 3 and fmove for case 1.
882 (extendsfdf2_cf): Use fdmove instead of f%&move.
883 (truncdfsf2_cf): Use fsmove instead of f%$smove.
884 (add<mode>3_cf, sub<mode>3_cf): Use <FP:prec> instead of <FP:round>.
885
886 2007-01-18 Richard Sandiford <richard@codesourcery.com>
887
888 * config/m68k/m68k.md (movdf_cf_hard): Use output_move_double for
889 GPR<-GPR moves.
890
891 2007-01-18 Richard Sandiford <richard@codesourcery.com>
892
893 * real.h (real_format): Add a canonical_nan_lsbs_set field.
894 (coldfire_single_format): Declare.
895 (coldfire_double_format): Likewise.
896 * real.c (encode_ieee_single): Use canonical_nan_lsbs_set instead
897 of qnan_msb_set to determine the lower bits of a canonical
898 NaN significand.
899 (encode_ieee_double): Likewise.
900 (encode_ieee_quad): Likewise.
901 (ieee_single_format): Initialize canonical_nan_lsbs_set.
902 (mips_single_format): Likewise.
903 (ieee_double_format): Likewise.
904 (mips_double_format): Likewise.
905 (ieee_extended_motorola_format): Likewise.
906 (ieee_extended_intel_96_format): Likewise.
907 (ieee_extended_intel_128_format): Likewise.
908 (ieee_extended_intel_96_round_53_format): Likewise.
909 (ibm_extended_format): Likewise.
910 (mips_extended_format): Likewise.
911 (ieee_quad_format): Likewise.
912 (mips_quad_format): Likewise.
913 (vax_f_format): Likewise.
914 (vax_d_format): Likewise.
915 (vax_g_format): Likewise.
916 (i370_single_format): Likewise.
917 (i370_double_format): Likewise.
918 (decimal_single_format): Likewise.
919 (decimal_double_format): Likewise.
920 (decimal_quad_format): Likewise.
921 (c4x_single_format): Likewise.
922 (c4x_extended_format): Likewise.
923 (real_internal_format): Likewise.
924 (coldfire_single_format): New real_format.
925 (coldfire_double_format): Likewise.
926 * config/pdp11/pdp11.c (pdp11_f_format): Initialize
927 canonical_nan_lsbs_set.
928 (pdp11_d_format): Likewise.
929 * config/m68k/m68k.c (override_options): Override REAL_FORMAT_MODE
930 if TARGET_COLDFIRE_CPU.
931
932 2007-01-18 Richard Sandiford <richard@codesourcery.com>
933
934 * config/m68k/m68k-protos.h (m68k_output_pic_call): Delete.
935 (output_call, m68k_legitimize_call_address): Declare.
936 * config/m68k/m68k.h (EXTRA_CONSTRAINT): Remove unnecessary
937 parenthesees. Add support for a 'W' constraint.
938 (LEGITIMATE_PIC_OPERAND_P): Remove SYMBOL_REF_FLAG handling.
939 (PRINT_OPERAND_PUNCT_VALID_P): Remove comment about 'o'.
940 (m68k_symbolic_call, m68k_symbolic_jump): Declare.
941 * config/m68k/m68k.c (m68k_symbolic_call, m68k_symbolic_jump): New
942 variables.
943 (override_options): Initialize them. Do not set flag_no_function_cse
944 for TARGET_ID_SHARED_LIBRARY.
945 (m68k_output_pic_call): Delete.
946 (m68k_legitimize_call_address): New function.
947 (print_operand): Remove the %o prefix. Handle the %p prefix.
948 (output_call): New function.
949 (m68k_output_mi_thunk): Use m68k_symbolic_jump. Always load the
950 target address from the GOT if symbolic jumps are not allowed.
951 * config/m68k/m68k.md (call, general_operand): Do not set
952 SYMBOL_REF_FLAG. Use m68k_legitimize_call_address instead.
953 Merge separate flag_pic and !flag_pic define_insns into...
954 (*call, *call_value): ...these new patterns. Match the address
955 rather than the containing MEM and require it to be a call_operand.
956 Use output_call to generate the asm template.
957 * config/m68k/predicates.md (const_call_operand): New predicate.
958 (call_operand): Likewise.
959
960 2007-01-18 Nathan Sidwell <nathan@codesourcery.com>
961
962 * config/m68k/m68k.h (REGISTER_MOVE_COST): Simplify definition.
963 (STACK_GROWS_DOWNWARD): Define to 1.
964 (FUNCTION_VALUE, LIBCALL_VALUE, FUNCTION_VALUE_REGNO_P): Equivocate
965 comments, emphasizing that these values are only defaults.
966 * config/m68k/linux.h (LINK_SPEC): Fix formatting in #undef.
967 * config/m68k/m68k.c (const_method): Remove trailing whitespace.
968
969 2007-01-18 Richard Sandiford <richard@codesourcery.com>
970
971 * config/m68k/m68k.md (cmpsi): Remove outdated flag_pic handling.
972
973 2007-01-18 Kazu Hirata <kazu@codesourcery.com>
974 Richard Sandiford <richard@codesourcery.com>
975
976 * config/m68k/m68k.h (DATA_REGNO_P, ADDRESS_REGNO_P, INT_REGNO_P)
977 (FP_REGNO_P): New macros.
978 (REGNO_OK_FOR_INDEX_P, REGNO_OK_FOR_BASE_P, REGNO_OK_FOR_DATA_P)
979 (REGNO_OK_FOR_FP_P, REG_OK_FOR_INDEX_P, REG_OK_FOR_BASE_P): Use them.
980 (EH_RETURN_STACKADJ_RTX): Use A0_REG.
981 * config/m68k/m68k.c (m68k_regno_mode_ok): Use the new REGNO macros.
982
983 2007-01-18 Nathan Sidwell <nathan@codesourcery.com>
984
985 * config.gcc (m68k-*-aout*, m68k-*-coff*, m68020-*-elf*, m68k-*-elf*)
986 (m68k-*-uclinuxoldabi, m68k-*-uclinux*, m68k-*-rtems*): Add t-floatlib
987 to $tmake_file.
988 * config/m68k/t-floatlib: New file, extracting common code from...
989 * config/m68k/t-m68kbare, config/m68k/t-m68kelf,
990 * config/m68k/t-uclinux: Here.
991 * config/m68k/fpgnulib.c: Do not compile extendeed precision
992 routines on ColdFire targets.
993
994 2007-01-18 Nathan Sidwell <nathan@codesourcery.com>
995
996 * config.gcc (m68k-*-aout*, m68k-*-coff*, m68020-*-elf*, m68k-*-elf*)
997 (m68010-*-netbsdelf*, m68k*-*-netbsdelf*, m68k*-*-openbsd*)
998 (m68k-*-uclinuxoldabi, m68k-*-uclinux*, m68k-*-linux*)
999 (m68k-*-rtems*): Use tm_file rather than m68k/m68k.h and
1000 explicitly set MOTOROLA to 1.
1001 * config/m68k/m68k.h (MOTOROLA): Simplify definition accordingly.
1002
1003 2007-01-18 Richard Sandiford <richard@codesourcery.com>
1004 Nathan Sidwell <nathan@codesourcery.com>
1005
1006 * config/m68k/m68k.h (PCC_STATIC_STRUCT_RETURN): Delete.
1007 (ASM_OUTPUT_REG_PUSH, ASM_OUTPUT_REG_POP): Add MOTOROLA cases.
1008 * config/m68k/coff.h (REGISTER_PREFIX_MD): Delete.
1009 * config/m68k/m68020-elf.h (LIB_SPEC): Delete.
1010 * config/m68k/m68k-none.h (CC1_SPEC, CPP_SUBTARGET_SPEC): Delete.
1011 * config/m68k/m68kelf.h (IMMEDIATE_PREFIX, REGISTER_PREFIX_MD)
1012 (ASM_OUTPUT_REG_PUSH, ASM_OUTPUT_REG_POP): Delete.
1013 (NO_DOLLAR_IN_LABEL, PCC_STATIC_STRUCT_RETURN): Don't undefine.
1014 (BSS_ASM_OP, ASM_OUTPUT_SKIP, ASM_OUTPUT_ASCII): Delete.
1015 * config/m68k/m68kemb.h (PCC_STATIC_STRUCT_RETURN): Don't undefine.
1016 * config/m68k/linux.h (SIZE_TYPE, PTRDIFF_TYPE, WCHAR_TYPE)
1017 (WCHAR_TYPE_SIZE, TARGET_OBJFMT_CPP_BUILTINS): Delete.
1018 (TARGET_OS_CPP_BUILTINS): Don't define mc68000 and mc68020 here.
1019 (DBX_REGISTER_NUMBER): Delete.
1020 * config/m68k/netbsd-elf.h (IMMEDIATE_PREFIX): Delete.
1021 (PCC_STATIC_STRUCT_RETURN): Don't undefine.
1022 * config/m68k/openbsd.h (PCC_STATIC_STRUCT_RETURN): Define.
1023
1024 2007-01-18 Richard Sandiford <richard@codesourcery.com>
1025
1026 * config.gcc (m68k-*-uclinux*): Add flat.h to $tm_file.
1027 * config/flat.h: New file.
1028 * crtstuff.c (USE_PT_GNU_EH_FRAME): Don't define if
1029 OBJECT_FORMAT_FLAT.
1030 * config/m68k/m68k.h (ASM_PREFERRED_EH_DATA_FORMAT): Do not use
1031 indirect references for -msep-data or -mid-shared-library.
1032 Do not use PC-relative code addresses either.
1033 * config/m68k/m68k.c (override_options): Restrict -fPIC error
1034 to -mpcrel.
1035 * config/m68k/uclinux.h (STARTFILE_SPEC): Define. Use Scrt1.o
1036 for shared libraries and crt1.o for executables. Use crti.o and
1037 crtbegin.o.
1038 (ENDFILE_SPEC): Use crtend.o and crtn.o.
1039 (LIB_SPEC): Suppress -Rlibc.gdb if -static-libc is given.
1040 Do not add -elf2flt or -shared-lib-id options here.
1041 (LINK_SPEC): Define. Pass -elf2flt if no -elf2flt option is given.
1042 Pass -shared-lib-id if -mid-shared-library, taking the library
1043 identifier from -mshared-library-id if given, otherwise
1044 defaulting to 0.
1045 (EH_FRAME_IN_DATA_SECTION): Do not undefine.
1046 (INIT_SECTION_ASM_OP, FINI_SECTION_ASM_OP): Likewise.
1047 (TARGET_OS_CPP_BUILTINS): Define __GXX_MERGED_TYPEINFO_NAMES=0
1048 and __GXX_TYPEINFO_EQUALITY_INLINE=0 if -mid-shared-library.
1049 (DRIVER_SELF_SPECS): Map unadorned PIC options to -msep-data.
1050 * config/m68k/t-uclinux (EXTRA_MULTILIB_PARTS): Add crtbegin.o
1051 and crtend.o.
1052 * config/m68k/lb1sf68.asm (PICCALL): Use an lea and pc-relative
1053 jump sequence for ISA A and ISA A+.
1054 (PICJUMP): Likewise.
1055
1056 2007-01-18 Richard Sandiford <richard@codesourcery.com>
1057
1058 * config.gcc (m68k-*-uclinux*): Base the port on the common
1059 and m68k GNU/Linux files rather than on the generic ELF ones.
1060 * config/m68k/uclinux.h (TARGET_VERSION): Override.
1061 (TARGET_OS_CPP_BUILTINS): Use LINUX_TARGET_OS_CPP_BUILTINS.
1062
1063 2007-01-18 Julian Brown <julian@codesourcery.com>
1064 Richard Sandiford <richard@codesourcery.com>
1065
1066 * config/m68k/m68k.h (LONG_DOUBLE_TYPE_SIZE): Make 64-bit on ColdFire.
1067 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
1068 * config/m68k/netbsd-elf.h (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Undefine
1069 before redefining.
1070 * config/m68k/uclinux-oldabi.h (LONG_DOUBLE_TYPE_SIZE): Redefine to
1071 80 unconditionally.
1072 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
1073
1074 2007-01-18 Richard Sandiford <richard@codesourcery.com>
1075
1076 * doc/install.texi: Document m68k-uclinuxoldabi.
1077 * config.gcc (m68k-*-uclinuxoldabi): New configuration.
1078 * config/m68k/uclinux-oldabi.h: New file, copied from
1079 config/m68k/uclinux.h.
1080
1081 2007-01-18 Kaz Kojima <kkojima@rr.iij4u.or.jp>
1082
1083 * config/m32r/m32r-protos.h (m32r_expand_epilogue): Declare it.
1084 * config/m32r/m32r.c (m32r_setup_incoming_varargs): Use gen_frame_mem.
1085 (m32r_compute_frame_size): Use unsigned for regno.
1086 (m32r_reload_lr): Use gen_frame_mem.
1087 (pop): New.
1088 (m32r_output_function_epilogue): Don't output the function epilogue
1089 textually here.
1090 (m32r_expand_epilogue): New.
1091 (direct_return): Return false if the function has the interrupt
1092 attribute.
1093 (m32r_hard_regno_rename_ok): Remove code for the textual epilogue.
1094 * config/m32r/m32r.md (epilogue): New expander.
1095 (return_lr, return_rte): New insns.
1096 (return): Make it expander.
1097 (return_normal): New expander.
1098
1099 2007-01-18 Josh Conner <jconner@apple.com>
1100
1101 PR target/30485
1102 * config/rs6000/rs6000.c (rs6000_emit_vector_compare): Add
1103 support for UNLE, UNLT, UNGE, and UNGT.
1104
1105 2007-01-18 Jan Hubicka <jh@suse.cz>
1106
1107 * tree-vrp.c (finalize_jump_threads): Do not call cleanup_cfg by hand.
1108 * tree-tailcall (add_virtual_phis): Likewise.
1109 (optimize_tail_call): Return TODOs.
1110 (execute_tail_calls): Return TODOs.
1111 * tree-ssa-ccp (execute_fold_all_builtins): Do cleanup_cfg via TODO.
1112 * tree-cfgcleanup.c (cleanup_tree_cfg_loop): Return if something
1113 changed.
1114 * tree-ssa-forwprop.c (tree_ssa_forward_propagate_single_use_value):
1115 Cleanup cfg using TODO.
1116 * tree-flow.h (cleanup_tree_cfg_loop): Update prototype.
1117 * passes.c (execute_function_todo): When cleanup did something, remove
1118 unused locals.
1119 * tree-cfg.c (pass_build_cfg): Add cleanup_cfg TODO.
1120 (make_edges): Don't cleanup_cfg.
1121
1122 2007-01-18 Uros Bizjak <ubizjak@gmail.com>
1123
1124 * reg-stack.c (subst_stack_regs_pat) [UNSPEC_SINCOS_COS,
1125 UNSPEC_XTRACT_FRACT]: Use generic code for instructions that
1126 operate on the top of stack.
1127 [UNSPEC_SINCOS_SIN, UNSPEC_XTRACT_EXP, UNSPEC_TAN]: Rewrite
1128 register handling of instructions that output to the second
1129 stack slot.
1130 [UNSPEC_TAN_ONE, UNSPEC_TAN_TAN]: Remove.
1131 (move_for_stack_reg): Special-case check for dead destination
1132 stack slot for constant load of 1.0 inside UNSPEC_TAN.
1133
1134 * config/i386/i386.md (UNSPEC_TAN): New constant.
1135 (UNSPEC_TAN_ONE, UNSPEC_TAN_TAN): Remove.
1136 (fptanxf4_i387, fptan_extend<mode>xf4_i387): New patterns
1137 to correctly model move of constant 1.0 to top stack slot.
1138 (*tandf3_1, *tansf3_1, *tanxf3_1): Remove insn patterns.
1139 (unnamed peephole2 pattern): Remove corresponding peephole2
1140 pattern that optimizes tan insn and loading of constant 1.0.
1141 (tanxf2): Use fptanxf4_i387.
1142 (tan<mode>2): Rename from tansf2 and tandf2 and macroize insn
1143 patterns using X87MODEF12 mode macro. Use fptan_extend<mode>xf4_i387
1144 and truncate result to requested mode. Use SSE_FLOAT_MODE_P to
1145 disable patterns for SSE math.
1146 (sincos<mode>3): Use truncxf<mode>2_i387_noop for truncation.
1147 (fyl2x_extend<mode>xf3_i387): Use X87MODEF12 for operand 1.
1148
1149 2007-01-18 Dirk Mueller <dmueller@suse.de>
1150 Richard Guenther <rguenther@suse.de>
1151
1152 PR diagnostic/8268
1153 * doc/invoke.texi (Warray-bounds): Document -Warray-bounds.
1154 * common.opt (Warray-bounds): Add new warning option.
1155 * c-opts.c (c_common_handle_option): Define -Warray-bounds
1156 if -Wall is given.
1157 * Makefile.in: make tree-vrp.o depend on toplev.h
1158 * tree-vrp.c (vrp_finalize): Call check_array_refs if -Warray-bounds
1159 is enabled.
1160 (check_array_refs, check_array_bounds, check_array_ref): New.
1161
1162 2007-01-18 Jan Hubicka <jh@suse.cz>
1163
1164 * tree-ssa-ccp.c (ccp_finalize): Return if something changed.
1165 (execute_ssa_ccp): Return flags conditionally.
1166 * tree-ssa-propagate.c (substitue_and_fold): Return if something was
1167 changed.
1168 * tree-ssa-propagate.h (substitute_and_fold): Update prototype.
1169
1170 2007-01-18 Steven Bosscher <steven@gcc.gnu.org>
1171
1172 * cfgcleanup.c (cleanup_cfg): Detect cfglayout mode and set
1173 the CLEANUP_CFGLAYOUT flag when in cfglayout mode.
1174
1175 * Makefile.c (GTFILES): Add cfglayout.h.
1176 * gengtype.c (open_base_files): Likewise.
1177 * cfglayout.c (cfg_layout_function_footer,
1178 cfg_layout_function_header) Reindent to make gengtype happy.
1179 * cfglayout.h (cfg_layout_function_footer,
1180 cfg_layout_function_header): Add GTY(()) marker.
1181
1182 * ifcvt.c (noce_try_sign_mask): Make sure INSN_B is non-null.
1183
1184 2007-01-18 Ben Elliston <bje@au.ibm.com>
1185
1186 * genautomata.c (write_automata): Include xstrerror output in the
1187 error message if writing the DFA description file fails.
1188
1189 2007-01-17 H.J. Lu <hongjiu.lu@intel.com>
1190
1191 * config/mips/mips-protos.h (mips_output_external): Make it
1192 return void.
1193 * config/mips/iris.h (TARGET_ASM_EXTERNAL_LIBCALL): Removed.
1194 * config/mips/mips.c (irix_output_external_libcall): Likewise.
1195 (extern_list): Likewise.
1196 (extern_head): Likewise.
1197 (TARGET_ASM_FILE_END): Likewise.
1198 (mips_file_end): Likewise.
1199 (mips_output_external): Rewritten.
1200
1201 2007-01-18 Ben Elliston <bje@au.ibm.com>
1202
1203 * genpreds.c (write_insn_preds_c): Only write out the function
1204 body for regclass_for_constraint if we have register constraints.
1205
1206 2007-01-17 Tom Tromey <tromey@redhat.com>
1207
1208 * doc/sourcebuild.texi (libgcj Tests): Use sourceware.org.
1209 * doc/install.texi (Testing): Use sourceware.org.
1210 (Binaries): Likewise.
1211 (Specific): Likewise.
1212 * doc/contrib.texi (Contributors): Use sourceware.org.
1213
1214 2007-01-17 Anatoly Sokolov <aesok@post.ru>
1215
1216 * config/avr/avr.h (AVR_HAVE_LPMX): New macro.
1217 (AVR_ENHANCED): Rename to ...
1218 (AVR_HAVE_MUL): ... new.
1219 (avr_enhanced_p): Rename to ...
1220 (avr_have_mul_p): ... new.
1221 (TARGET_CPU_CPP_BUILTINS): Use 'avr_have_mul_p' instead of
1222 'avr_enhanced_p' for "__AVR_ENHANCED__". Define "__AVR_HAVE_MUL__".
1223 * config/avr/avr.c (avr_enhanced_p): Rename to ...
1224 (avr_have_mul_p): ... new.
1225 (base_arch_s): Rename 'enhanced' to 'have_mul'.
1226 (avr_override_options): Use 'avr_have_mul_p' and 'have_mul' instead of
1227 'avr_enhanced_p' and 'enhanced'.
1228 (ashlhi3_out, ashrhi3_out, lshrhi3_out, avr_rtx_costs): Use
1229 AVR_HAVE_MUL instead of AVR_ENHANCED.
1230 * avr.md (*tablejump_enh): Use AVR_HAVE_LPMX instead of AVR_ENHANCED.
1231 (mulqi3, *mulqi3_enh, *mulqi3_call, mulqihi3, umulqihi3, mulhi3,
1232 *mulhi3_enh, *mulhi3_call, mulsi3, *mulsi3_call): Use AVR_HAVE_MUL
1233 instead of AVR_ENHANCED.
1234 (*tablejump_enh): Use AVR_HAVE_LPMX instead of AVR_ENHANCED.
1235 * libgcc.S: Use __AVR_HAVE_MUL__ instead of __AVR_ENHANCED__.
1236 (__tablejump__): Use __AVR_HAVE_LPMX__ instead of __AVR_ENHANCED__.
1237
1238 2007-01-17 Ian Lance Taylor <iant@google.com>
1239
1240 * vec.h (VEC_reserve_exact): Define.
1241 (vec_gc_p_reserve_exact): Declare.
1242 (vec_gc_o_reserve_exact): Declare.
1243 (vec_heap_p_reserve_exact): Declare.
1244 (vec_heap_o_reserve_exact): Declare.
1245 (VEC_OP (T,A,reserve_exact)): New static inline function, three
1246 versions.
1247 (VEC_OP (T,A,reserve)) [all versions]: Remove handling of
1248 negative parameter.
1249 (VEC_OP (T,A,alloc)) [all versions]: Call ...reserve_exact.
1250 (VEC_OP (T,A,copy)) [all versions]: Likewise.
1251 (VEC_OP (T,a,safe_grow)) [all versions]: Likewise.
1252 * vec.c (calculate_allocation): Add exact parameter. Change all
1253 callers.
1254 (vec_gc_o_reserve_1): New static function, from vec_gc_o_reserve.
1255 (vec_gc_p_reserve, vec_gc_o_reserve): Call vec_gc_o_reserve_1.
1256 (vec_gc_p_reserve_exact, vec_gc_o_reserve_exact): New functions.
1257 (vec_heap_o_reserve_1): New static function, from vec_heap_o_reserve.
1258 (vec_heap_p_reserve, vec_heap_o_reserve): Call vec_heap_o_reserve_1.
1259 (vec_heap_p_reserve_exact): New function.
1260 (vec_heap_o_reserve_exact): New function.
1261
1262 2007-01-17 Jan Hubicka <jh@suse.cz>
1263
1264 * ipa-type-escape.c (look_for_casts): Revamp using handled_component_p.
1265
1266 2007-01-17 Eric Christopher <echristo@apple.com>
1267
1268 * config.gcc: Support core2 processor.
1269
1270 2007-01-16 Jan Hubicka <jh@suse.cz>
1271
1272 * tree-ssanames.c (release_dead_ssa_names): Instead of ggc_freeing
1273 the names, just unlink the chain so we don't crash on dangling pointers
1274 to dead SSA names.
1275
1276 2007-01-16 Jan Hubicka <jh@suse.cz>
1277
1278 * cgraph.h (cgraph_decide_inlining_incrementally): Kill.
1279 * tree-pass.h: Reorder to make IPA passes appear toegher.
1280 (pass_early_inline, pass_inline_parameters, pass_apply_inline): Declare.
1281 * cgraphunit.c (cgraph_finalize_function): Do not compute inling
1282 parameters, do not call early inliner.
1283 * ipa-inline.c: Update comments. Include tree-flow.h
1284 (cgraph_decide_inlining): Do not compute inlining parameters.
1285 (cgraph_decide_inlining_incrementally): Return TODOs; assume to
1286 be called with function context set up.
1287 (pass_ipa_inline): Remove unreachable functions before pass.
1288 (cgraph_early_inlining): Simplify assuming to be called from the
1289 PM as local pass.
1290 (pass_early_inline): New pass.
1291 (cgraph_gate_ipa_early_inlining): New gate.
1292 (pass_ipa_early_inline): Turn into simple wrapper.
1293 (compute_inline_parameters): New function.
1294 (gate_inline_passes): New gate.
1295 (pass_inline_parameters): New pass.
1296 (apply_inline): Move here from tree-optimize.c
1297 (pass_apply_inline): New pass.
1298 * ipa.c (cgraph_remove_unreachable_nodes): Verify cgraph after
1299 transforming.
1300 * tree-inline.c (optimize_inline_calls): Return TODOs rather than
1301 doing them by hand.
1302 (tree_function_versioning): Do not allocate dummy struct function.
1303 * tree-inline.h (optimize_inline_calls): Update prototype.
1304 * tree-optimize.c (execute_fixup_cfg): Export.
1305 (pass_fixup_cfg): Remove
1306 (tree_rest_of_compilation): Do not apply inlines.
1307 * tree-flow.h (execute_fixup_cfg): Declare.
1308 * Makefile.in (gt-passes.c): New.
1309 * passes.c: Include gt-passes.h
1310 (init_optimization_passes): New passes.
1311 (nnodes, order): New static vars.
1312 (do_per_function_toporder): New function.
1313 (execute_one_pass): Dump current pass here.
1314 (execute_ipa_pass_list): Don't dump current pass here.
1315
1316 2007-01-16 Janis Johnson <janis187@us.ibm.com>
1317
1318 * config/dfp-bit.c (dfp_compare_op): Return separate value for NaN.
1319 (DFP_NE, DFP_LE, DFP_GE): Return false for NaN.
1320
1321 2007-01-16 David Edelsohn <edelsohn@gnu.org>
1322
1323 * config/rs6000/darwin-ldouble.c: Build file for SOFT_FLOAT.
1324 (strong_alias): Define.
1325 (__gcc_qmul): Provide non-FMA for soft-float.
1326 (__gcc_qdiv): Same.
1327 (__gcc_qneg): New.
1328 (__gcc_qeq): New.
1329 (__gcc_qle): New.
1330 (__gcc_qge): New.
1331 (__gcc_qunord): New.
1332 (__gcc_stoq): New.
1333 (__gcc_dtoq): New.
1334 (__gcc_qtos): New.
1335 (__gcc_qtod): New.
1336 (__gcc_qtoi): New.
1337 (__gcc_qtou): New.
1338 (__gcc_itoq): New.
1339 (__gcc_utoq): New.
1340 (fmsub): New.
1341 * config/rs6000/rs6000.c (rs6000_init_libfuncs): Initialize
1342 soft-float functions.
1343 * config/rs6000/libgcc-ppc-glibc.ver: Version soft-float symbols.
1344 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Do not warn
1345 about long double soft float.
1346
1347 2007-01-16 Dorit Nuzman <dorit@il.ibm.com>
1348 Tehila Meyzels <tehila@il.ibm.com>
1349
1350 * tree-vectorizer.h (is_pattern_stmt_p): New.
1351 * tree-vect-analyze.c (vect_determine_vectorization_factor): Fix
1352 formatting (tabs instead of spaces). Cleanup and clarify setting
1353 of STMT_VINFO_VECTYPE. Call is_pattern_stmt_p.
1354 * tree-vect-transform.c (vect_get_vec_def_for_operand): Fix typo.
1355 (vectorizable_type_demotion): Check that types are integral.
1356 (vectorizable_type_promotion): Likewise.
1357 (vectorizable_store): Fix typo. Eliminate new-line at end of
1358 comments.
1359
1360 2007-01-16 Jan Hubicka <jh@suse.cz>
1361
1362 * tree-ssanames.c (release_dead_ssa_names): Remove invalidated
1363 cgraph edges too.
1364
1365 2007-01-15 Eric Christopher <echristo@apple.com>
1366
1367 * ifcvt.c: Include vec.h, vecprim.h.
1368 (check_cond_move_block): New argument regs.
1369 Reorganize. Add registers used to regs.
1370 (cond_move_process_if_block): Use regs set above as
1371 loop bounds.
1372
1373 2007-01-15 Eric Christopher <echristo@apple.com>
1374
1375 * config/darwin.h: Update copyright.
1376 (TARGET_OPTION_TRANSLATE_TABLE): Add umbrella.
1377 (LINK_COMMAND_SPEC): Add -u.
1378 (LINK_SPEC): Fix umbrella for above.
1379
1380 2007-01-15 Joseph S. Myers <joseph@codesourcery.com>
1381
1382 * config/soft-fp/op-common.h, config/soft-fp/op-4.h: Update from
1383 glibc CVS.
1384
1385 2007-01-15 Tom Tromey <tromey@redhat.com>
1386
1387 * doc/sourcebuild.texi (libgcj Tests): Don't mention jacks.
1388 * doc/install.texi (Testing): Don't mention jacks.
1389 (Configuration): Document --enable-java-maintainer-mode. Move
1390 --with-java-home to libgcj-specific section. Document
1391 --with-ecj-jar.
1392 (Prerequisites): Mention --enable-java-maintainer-mode, ecj1.
1393
1394 2007-01-15 Jan Hubicka <jh@suse.cz>
1395
1396 * tree-ssa-dce.c (DCE_TODOs): New.
1397 (propagate_necessity): Return if something changed.
1398 (eliminate_unnecessary_stmts): Likewise.
1399 (perform_tree_ssa_dce): Return TODO flags when needed.
1400 (pass_dce, pass_dce_loop, pass_cd_dce): Remove TODO flags.
1401
1402 2007-01-15 Uros Bizjak <ubizjak@gmail.com>
1403
1404 * config/i386/i386.md (fyl2xxf3_i387): Rename from fyl2x_xf3.
1405 (fyl2x_extend<mode>xf3_i387): New insn pattern.
1406 (log<mode>2): Rename from logsf2 and logdf2 and macroize insn
1407 patterns using X87MODEF12 mode macro. Extend operand 1
1408 to XFmode. Use SSE_FLOAT_MODE_P to disable patterns for SSE math.
1409 (log10<mode>2): Ditto.
1410 (log2<mode>2): Ditto.
1411 (log1p<mode>2): Ditto.
1412 (logb<mode>2): Ditto.
1413 (fyl2xp1xf3_i387): Rename from fyl2xp1_xf3.
1414 (fyl2xp1_extend<mode>xf3_i387): New insn pattern.
1415 (*fxtractxf3_i387): Rename from *fxtractxf3.
1416 (fxtract_extend<mode>xf3_i387): New insn pattern.
1417 (ilogbsi2): Use match_dup 3, not match_operand:XF 3.
1418 * config/i386/i386.c (ix86_emit_i387_log1p): Use gen_fyl2xp1xf3_i387()
1419 and gen_fyl2xxf3_i387().
1420
1421 2007-01-14 Zdenek Dvorak <dvorakz@suse.cz>
1422
1423 * loop-unswitch.c (unswitch_loop): Do not call fix_loop_placement.
1424 * cfgloopmanip.c (fix_loop_placement): Made static. Use
1425 get_loop_exit_edges. Changed return type to bool.
1426 * cfgloop.h (fix_loop_placement): Declaration removed.
1427
1428 2007-01-14 Dorit Nuzman <dorit@il.ibm.com>
1429
1430 * param.h (MIN_VECT_LOOP_BOUND): New.
1431 * params.def (MIN_VECT_LOOP_BOUND): New.
1432 * tree-vectorizer.c (slpeel_tree_peel_loop_to_edge): Takes another
1433 argument - minimum threshold for number of iterations.
1434 * tree-vectorizer.h (slpeel_tree_peel_loop_to_edge): Add another
1435 argument to declaration.
1436 * tree-vect-analyze.c (vect_analyze_operations): Check value of
1437 MIN_VECT_LOOP_BOUND.
1438 * tree-vect-transform.c (vect_do_peeling_for_loop_bound): Call
1439 slpeel_tree_peel_loop_to_edge with additional argument.
1440 (vect_do_peeling_for_alignment): Likewise.
1441 * doc/invoke.texi (min-vect-loop-bound): Document new param option.
1442
1443 2007-01-14 Uros Bizjak <ubizjak@gmail.com>
1444
1445 PR target/30413
1446 * config/i386/i386.c (print_operand) ['z']: Output 'b' for
1447 operands of size 1.
1448
1449 2007-01-14 Jan Hubicka <jh@suse.cz>
1450
1451 * tree-dfa.c (remove_referenced_var): New function.
1452 * tree-ssa-live.c (remove_unused_locals): Walk referenced vars and
1453 prune referenced vars list too.
1454 * tree-flow.h (remove_referenced_var): Declare.
1455
1456 2007-01-14 Jan Hubicka <jh@suse.cz>
1457
1458 * tree-eh.c (add_stmt_to_eh_region_fn): Do not add call_exprs
1459 separately.
1460 (remove_stmt_from_eh_region_fn): Do not remove call_exprs.
1461 (verify_eh_throw_stmt_node, verify_eh_throw_table_statements): Kill.
1462 * except.h (verify_eh_throw_table_statements): Kill prototype.
1463 * cfgexpand.c (expand_gimple_basic_block): Propagate Eh regions
1464 into call exrepssions.
1465 * tree-optimize.c (execute_free_cfg_annotatiosn): Do not call
1466 eh trhow verifier.
1467 * tree-cfg.c: Include pointer-set.h.
1468 (verify_node_sharing): Work on pointer set.
1469 (verify_eh_throw_stmt_node): New.
1470 (verify_stmts): Use pointers sets, verify throw_stmt.
1471
1472 2007-01-13 Zdenek Dvorak <dvorakz@suse.cz>
1473
1474 * ipa-reference.c (analyze_function): Consider also addresses taken
1475 in phi nodes.
1476
1477 2007-01-12 Roger Sayle <roger@eyesopen.com>
1478
1479 * c-typeck.c (null_pointer_constant_p): Replace use of
1480 TREE_CONSTANT_OVERFLOW with TREE_OVERFLOW.
1481 (build_c_cast): Likewise.
1482
1483 2007-01-12 Roger Sayle <roger@eyesopen.com>
1484
1485 * tree.h (force_fit_type_double): Remove unused final argument.
1486 * c-common.c (constant_expression_warning): Replace use of
1487 TREE_CONSTANT_OVERFLOW with TREE_OVERFLOW.
1488 (convert_and_check): Likewise.
1489 (shorten_compare): Update call to force_fit_type_double.
1490 (c_common_truthvalue_conversion) <INTEGER_CST>: Use integer_zerop.
1491 * convert.c (convert_to_pointer): Update call to
1492 force_fit_type_double.
1493 * fold-const.c (force_fit_type_double): Remove overflowed_const
1494 argument.
1495 (int_const_binop, fold_convert_const_int_from_int,
1496 fold_convert_const_int_from_real, fold_div_compare,
1497 fold_sign_changed_comparison, fold_unary, fold_negate_const,
1498 fold_abs_const, fold_not_const): Remove the final argument from
1499 calls to force_fit_type_double.
1500
1501 2007-01-12 Andrew Pinski <andrew_pinski@playstation.sony.com>
1502
1503 * configure.ac: Set insn to "nop" for spu-*-* also.
1504 * configure: Regenerate.
1505
1506 2007-01-12 Olga Golovanevsky <olga@il.ibm.com>
1507
1508 * builtins.def : Add BUILT_IN_FREE.
1509
1510 2007-01-12 Jan Hubicka <jh@suse.cz>
1511
1512 PR tree-optimization/30443
1513 * tree-inline.c (tree_function_versioning): Do not optimize when
1514 cloning for inlining.
1515
1516 2007-01-12 Zdenek Dvorak <dvorakz@suse.cz>
1517
1518 * doc/loop.texi: Document recording of loop exits.
1519 * cfgloopmanip.c (loopify, duplicate_loop): Use alloc_loop.
1520 (update_single_exits_after_duplication,
1521 update_single_exit_for_duplicated_loop,
1522 update_single_exit_for_duplicated_loops): Removed.
1523 (duplicate_loop_to_header_edge): Do not call
1524 update_single_exits_after_duplication and
1525 update_single_exit_for_duplicated_loops.
1526 (loop_version): Do not update single_exit information.
1527 (fix_loop_structure): Use record_loop_exits instead of
1528 mark_single_exit_loops.
1529 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update
1530 the lists of loop exits.
1531 * cfghooks.c (redirect_edge_and_branch, redirect_edge_and_branch_force,
1532 split_edge, merge_blocks): Update the lists of loop exits.
1533 * modulo-sched.c (sms_schedule): Pass LOOPS_HAVE_RECORDED_EXITS to
1534 loop_optimizer_init.
1535 * loop-init.c (loop_optimizer_init): Call record_loop_exits instead
1536 of mark_single_exit_loops.
1537 (loop_optimizer_finalize): Call release_recorded_exits.
1538 * tree-ssa-loop.c (tree_loop_optimizer_init): Pass
1539 LOOPS_HAVE_RECORDED_EXITS to loop_optimizer_init.
1540 * tree-vectorizer.c (slpeel_tree_duplicate_loop_to_edge_cfg): Do not
1541 update single exit information.
1542 * lambda-code.c (perfect_nestify): Ditto.
1543 * cfgloop.c (flow_loop_free): Destroy the list of exits of the loop.
1544 (mark_single_exit_loops): Removed.
1545 (alloc_loop, loop_exit_hash, loop_exit_eq, loop_exit_free,
1546 get_exit_descriptions, rescan_loop_exit, record_loop_exits,
1547 dump_recorded_exit, dump_recorded_exits, release_recorded_exits): New
1548 functions.
1549 (get_loop_exit_edges, single_exit): Use recorded exit lists.
1550 (add_bb_to_loop, remove_bb_from_loops): Update the lists of loop exits.
1551 (verify_loop_structure): Verify consistency of the exit lists.
1552 (flow_loops_find): Use alloc_loop. Initialize exits hash.
1553 (set_single_exit): Removed.
1554 * cfgloop.h (struct loop_exit): New function.
1555 (struct loop): single_exit_ field replaced by exits field.
1556 (LOOPS_HAVE_MARKED_SINGLE_EXITS): Replaced by LOOPS_HAVE_RECORDED_EXITS.
1557 (struct loops): Added exits hash.
1558 (mark_single_exit_loops, set_single_exit): Declaration removed.
1559 (release_recorded_exits, record_loop_exits, rescan_loop_exit): Declare.
1560
1561 2007-01-12 Richard Sandiford <richard@codesourcery.com>
1562
1563 * doc/invoke.texi: Avoid use of @headitem.
1564
1565 2007-01-12 Richard Sandiford <richard@codesourcery.com>
1566
1567 * cse.c (cse_insn): Move HAVE_CC0 code after declarations.
1568
1569 2007-01-12 Richard Sandiford <richard@codesourcery.com>
1570
1571 * doc/install.texi: Fix m68k-*-* anchor and add m68k-*-* to the
1572 list of targets.
1573
1574 2007-01-12 Nathan Sidwell <nathan@codesourcery.com>
1575 Richard Sandiford <richard@codesourcery.com>
1576
1577 * doc/invoke.texi: Document -mno-bitfield, -mno-rtd and -mno-short.
1578 * config/m68k/m68k.opt: Resort options.
1579 (mbitfield, mrtd, mshort): Remove RejectNegative properties.
1580
1581 2007-01-12 Nathan Sidwell <nathan@codesourcery.com>
1582 Richard Sandiford <richard@codesourcery.com>
1583
1584 * doc/invoke.texi: Document the macros that are defined by
1585 m68k's -mtune and -mhard-float options.
1586 * config/m68k/m68k-protos.h (m68k_cpp_cpu_ident) Declare.
1587 (m68k_cpp_cpu_family): Likewise.
1588 * config/m68k/m68k.h (TARGET_CPU_CPP_BUILTINS): Add a full set
1589 of __ucfv*__ macros. Define __mcffpu__ if generating code for
1590 ColdFire FPUs. Define __mcf_cpu_* and __mcf_family_* macros.
1591 * config/m68k/m68k.c (m68k_cpp_cpu_ident): New function.
1592 (m68k_cpp_cpu_family): Likewise.
1593
1594 2007-01-12 Richard Sandiford <richard@codesourcery.com>
1595
1596 * config/m68k/m68k.h (TARGET_CPU_CPP_BUILTINS): Treat all mc68*
1597 macros besides mc68000 as tuning macros. Use a switch statement
1598 to set them and mcpu32.
1599
1600 2007-01-12 Julian Brown <julian@codesourcery.com>
1601
1602 * config/m68k/m68k.h: Use TARGET_68040 instead of TARGET_68040_ONLY.
1603 (TARGET_68040_ONLY): Rename to...
1604 (TARGET_68040): ...this.
1605 * config/m68k/m68k.c: Use TARGET_68040 instead of TARGET_68040_ONLY.
1606 * config/m68k/m68k.md: Likewise.
1607
1608 2007-01-12 Julian Brown <julian@codesourcery.com>
1609 Nathan Sidwell <nathan@codesourcery.com>
1610 Richard Sandiford <richard@codesourcery.com>
1611
1612 * config.gcc (m680[012]0-*-*, m68k*-*-*): Set m68k_cpu_ident to
1613 the -mcpu= argument associated with the --with-cpu setting.
1614 Define M68K_DEFAULT_TUNE to the default -mtune= option,
1615 if different from the one implied by the -mcpu setting.
1616 Accept --with-cpu=FOO if FOO is listed in m68k-devices.def,
1617 using mcpu=FOO as the default CPU option. Set target_cpu_default2.
1618 * doc/invoke.texi: Mention ColdFire in the introduction to the
1619 m68k options. Document the new -march, -mcpu, -mtune, -mdiv,
1620 -mno-div and -mhard-float options. Make -m68881 a synonym for
1621 -mhard-float. Document the previously-undocumented -m5206e,
1622 -m528x, -m5307 and -m5407 options. Tweak the existing option
1623 documentation for consistency.
1624 * doc/install.texi: Mention new --with-cpu arguments.
1625 * config/m68k/m68k.h (OPTION_DEFAULT_SPECS): Only use the
1626 default CPU if neither -mcpu nor -march are specified.
1627 (ASM_CPU_SPEC): Pass down -mcpu and -march options.
1628 (TARGET_CPU_CPP_BUILTINS): Set __mcfisa*__ macros from
1629 TARGET_ISA*. Set the legacy __mcf*__ cpu macros in the same way,
1630 using m68k_tune to decide between families that implement the
1631 same ISA. Use m68k_tune to set __mcfv4e__.
1632 (FL_BITFIELD, FL_68881, FL_COLDFIRE, FL_CF_HWDIV, FL_CF_MAC)
1633 (FL_CF_EMAC, FL_CF_EMAC_B, FL_CF_USP, FL_CF_FPU, FL_ISA_68000)
1634 (FL_ISA_68010, FL_ISA_68020, FL_ISA_68040, FL_ISA_A, FL_ISA_B)
1635 (FL_ISA_C, FL_ISA_MMU): New macros.
1636 (MASK_COLDFIRE): Delete.
1637 (TARGET_68010, TARGET_68020, TARGET_68040_ONLY, TARGET_COLDFIRE)
1638 (TARGET_ISAB): Redefine in terms of m68k_cpu_flags.
1639 (TARGET_68881, TARGET_COLDFIRE_FPU): Redefine in terms of m68k_fpu.
1640 (TARGET_HARD_FLOAT): Do not define here.
1641 (TARGET_ISAAPLUS, TARGET_ISAC): New macros.
1642 (TUNE_68000): New macro.
1643 (TUNE_68000_10): Redefine in terms of TUNE_68000 and TUNE_68010.
1644 (TUNE_68010, TUNE_68030, TUNE_68040, TUNE_68060, TUNE_CPU32)
1645 (TUNE_CFV2): Redefine in terms of m68k_tune.
1646 (uarch_type, target_device, fpu_type): New enums.
1647 (m68k_cpu, m68k_tune, m68k_fpu, m68k_cpu_flags): Declare.
1648 * config/m68k/m68k.c (TARGET_DEFAULT): Remove MASK_68881.
1649 (FL_FOR_isa_00, FL_FOR_isa_10, FL_FOR_isa_20, FL_FOR_isa_40)
1650 (FL_FOR_isa_cpu32, FL_FOR_isa_a, FL_FOR_isa_aplus, FL_FOR_isa_b)
1651 (FL_FOR_isa_c): New macros.
1652 (m68k_isa): New enum.
1653 (m68k_target_selection): New structure.
1654 (all_devices, all_isas, all_microarchs): New tables.
1655 (m68k_cpu_entry, m68k_arch_entry, m68k_tune_entry, m68k_cpu)
1656 (m68k_tune, m68k_fpu, m68k_cpu_flags): New variables.
1657 (MASK_ALL_CPU_BITS): Delete.
1658 (m68k_find_selection): New function.
1659 (m68k_handle_option): Handle -mcpu=, -march= and -mtune=.
1660 Map the legacy target options to a combination of the new ones.
1661 (override_options): Set m68k_cpu, m68k_tune, m68k_fpu and
1662 m68k_cpu_flags. Handle M68K_DEFAULT_TUNE. Use m68k_cpu_flags
1663 to derive default MASK_BITFIELD, MASK_CF_HWDIV and MASK_HARD_FLOAT
1664 settings.
1665 * config/m68k/m68k.opt (m5200, m5206e, m528x, m5307, m5407, mcfv4e)
1666 (m68010, m68020, m68020-40, m68020-60, m68030, m68040): Remove Mask
1667 properties.
1668 (m68881, msoft-float): Change mask from 68881 to HARD_FLOAT.
1669 (march=, mcpu=, mdiv, mhard-float, mtune=): New options.
1670 * config/m68k/m68k-devices.def: New file.
1671
1672 2007-01-12 Richard Sandiford <richard@codesourcery.com>
1673 Nathan Sidwell <nathan@codesourcery.com>
1674
1675 * config/m68k/m68k.h (ASM_CPU_SPEC, ASM_SPEC, EXTRA_SPECS)
1676 (SUBTARGET_EXTRA_SPECS): New macros.
1677 * config/m68k/linux.h (ASM_SPEC): Remove CPU flags;
1678 use %(asm_cpu_spec) instead.
1679 * config/m68k/m68k-none.h (ASM_SPEC): Likewise.
1680 * config/m68k/openbsd.h (ASM_SPEC): Likewise.
1681 * config/m68k/netbsd-elf.h (ASM_SPEC): Likewise.
1682 (EXTRA_SPECS): Rename to...
1683 (SUBTARGET_EXTRA_SPECS): ...this.
1684
1685 2007-01-12 Nathan Sidwell <nathan@codesourcery.com>
1686 Richard Sandiford <richard@codesourcery.com>
1687 Julian Brown <julian@codesourcery.com>
1688
1689 * config.gcc (m68k-*-aout*, m68k-*-coff*, m68020-*-elf*, m68k-*-elf*)
1690 (m68k-*-uclinux*, m68k-*-linux*, m68k-*-rtems*): Set default_m68k_cpu
1691 to the configuration's default CPU.
1692 (m68010-*-netbsdelf*, m68k*-*-netbsdelf*, m68k*-*-openbsd*): Likewise.
1693 Remove default masks.
1694 (m680[012]0-*-*): Set the default with_cpu to the first part of
1695 the target name.
1696 (m68k*-*-*): Set the default with_cpu to m$default_m68k_cpu.
1697 (m68k*-*-linux): Extend the --with-cpu handling to...
1698 (m680[012]0-*-*, m68k*-*-*): ...these configurations. Allow m68000
1699 and m68010. Don't set target_cpu_default2.
1700 * doc/install.texi: Document --with-cpu for m68k.
1701 * config/m68k/m68k.h (OPTION_DEFAULT_SPECS): Define.
1702 * config/m68k/m68k-none.h (TARGET_CPU_DEFAULT, M68K_CPU_m68k)
1703 (M68K_CPU_m68000, M68K_CPU_m68010, M68K_CPU_m68020, M68K_CPU_m68030)
1704 (M68K_CPU_m68040, M68K_CPU_m68302, M68K_CPU_m68332, TARGET_DEFAULT)
1705 (ASM_CPU_DEFAULT_SPEC, CC1_CPU_DEFAULT_SPEC): Delete.
1706 (ASM_SPEC): Remove use of %(asm_cpu_default).
1707 (EXTRA_SPECS, SUBTARGET_EXTRA_SPECS, MULTILIB_DEFAULTS): Delete.
1708 * config/m68k/linux.h (TARGET_DEFAULT): Delete.
1709 (CPP_SPEC): Merge definitions. Do not handle __HAVE_68881__ here.
1710 * config/m68k/netbsd-elf.h (TARGET_OS_CPP_BUILTINS): Define
1711 __HAVE_FPU__ if TARGET_HARD_FLOAT.
1712 (TARGET_DEFAULT): Delete.
1713 (EXTRA_SPECS): Delete cpp_cpu_default_spec, cpp_cpu_spec,
1714 cpp_fpu_spec, asm_default_spec and netbsd_cpp_spec.
1715 (CPP_CPU_SPEC): Delete.
1716 (TARGET_VERSION): Merge definitions, using TARGET_68010 to pick
1717 the appropriate string.
1718 (CPP_CPU_DEFAULT_SPEC, ASM_DEFAULT_SPEC, CPP_FPU_SPEC): Delete.
1719 (CPP_SPEC): Define to NETBSD_CPP_SPEC.
1720 (ASM_SPEC): Don't use %(asm_default_spec).
1721 * config/m68k/m68k.c (TARGET_DEFAULT_TARGET_FLAGS): Remove
1722 TARGET_DEFAULT and add MASK_68881.
1723 * config/m68k/m68k.md: Remove mention of TARGET_DEFAULT from comments.
1724
1725 2007-01-12 Richard Sandiford <richard@codesourcery.com>
1726
1727 * config.gcc (m68010-*-netbsdelf*): Add MASK_68010.
1728 (m68k*-*-netbsdelf*, m68k*-*-openbsd*, m68k*-linux*): Add
1729 MASK_68010 alongside MASK_68020.
1730 * doc/invoke.texi: Document -m68010.
1731 * config/m68k/m68k.opt (m68010): New.
1732 * config/m68k/m68k.h (TARGET_CPU_CPP_BUILTINS): Define mc68010
1733 if TUNE_68010.
1734 (TUNE_68010): New macro.
1735 * config/m68k/m68k-none.h (M68K_CPU_m68k, M68K_CPU_m68010)
1736 (M68K_CPU_m68020, M68K_CPU_m68030, M68K_CPU_m68040)
1737 (M68K_CPU_m68332): Add MASK_68010.
1738 * config/m68k/linux.h (TARGET_DEFAULT): Add MASK_68010 to
1739 fallback definition.
1740 * config/m68k/netbsd-elf.h (CPP_CPU_SPEC): Remove now-redundant
1741 defines.
1742 * config/m68k/m68k.c (MASK_ALL_CPU_BITS): Add MASK_68010.
1743 (m68k_handle_option): Handle OPT_m68010. Add MASK_68010
1744 to all entries that use MASK_68020.
1745 (output_move_simode_const, output_move_himode, output_move_qimode)
1746 (output_move_stricthi, output_move_strictqi): Use TARGET_68010
1747 instead of TARGET_68020 to select clr behavior. Remove comment
1748 about there being no TARGET_68010.
1749 * config/m68k/m68k.md: Likewise throughout.
1750
1751 2007-01-12 Julian Brown <julian@codesourcery.com>
1752
1753 * config/m68k/m68k.h (TARGET_ISAB): New macro.
1754 * config/m68k/m68k.c: Use TARGET_ISAB rather than TARGET_CFV4.
1755 * config/m68k/m68k.md: Likewise.
1756
1757 2007-01-12 Julian Brown <julian@codesourcery.com>
1758
1759 * config/m68k/m68k.h (LEGITIMATE_INDEX_P, LEGITIMIZE_ADDRESS): Use
1760 TARGET_COLDFIRE_FPU instead of TARGET_CFV4E.
1761
1762 2007-01-12 Julian Brown <julian@codesourcery.com>
1763
1764 * config/m68k/m68k.h (TUNE_68040_60): New macro.
1765 * config/m68k/m68k.c (standard_68881_constant_p): Use it.
1766 * config/m68k/m68k.md: Likewise.
1767
1768 2007-01-12 Julian Brown <julian@codesourcery.com>
1769 Richard Sandiford <richard@codesourcery.com>
1770
1771 * config/m68k/m68k.h (TARGET_CPU_CPP_BUILTINS): Use TUNE_68030
1772 instead of TARGET_68030, TUNE_68040 instead of TARGET_68040,
1773 TUNE_68060 instead of TARGET_68060 and TUNE_CPU32 instead of
1774 TARGET_CPU32.
1775 (TARGET_CPU32): Rename to...
1776 (TUNE_CPU32): ...this.
1777 (TUNE_68000_10, TUNE_68030, TUNE_68040, TUNE_68060)
1778 (TUNE_CFV2): New macros.
1779 * config/m68k/netbsd-elf.h (LONG_DOUBLE_TYPE_SIZE): Simplify;
1780 remove conditions that are implied by TARGET_68020.
1781 * config/m68k/m68k.c (m68k_output_function_prologue): Use TUNE_68040
1782 instead of TARGET_68040 and TUNE_CPU32 instead of TARGET_CPU32.
1783 (m68k_output_function_epilogue): Likewise.
1784 (m68k_rtx_costs): Likewise. Use TUNE_68060 instead of TARGET_68060
1785 and TUNE_CFV2 instead of TARGET_5200. Use TUNE_68000_10 instead of
1786 "!TARGET_68020 && !TARGET_COLDFIRE" to choose between 68000 and
1787 non-68000 timings. Refactor multiplication and division costs.
1788 (output_addsi3): Use TUNE_68040 instead of TARGET_68040 and
1789 TUNE_CPU32 instead of TARGET_CPU32.
1790 (standard_68881_constant_p): Use TUNE_68040 instead of TARGET_68040
1791 and TUNE_68060 instead of TARGET_68060.
1792 * config/m68k/m68k.md: Use TUNE_68040 instead of TARGET_68040,
1793 TUNE_68060 instead of TARGET_68060, and TUNE_CPU32 instead of
1794 TARGET_CPU32.
1795 (movsi_const0): Use TUNE_68000_10 rather than "!TARGET_68020
1796 && !TARGET_COLDFIRE" to choose between moveq and clr.
1797 Likewise in the unnamed movsf pattern.
1798 (ashlsi_17_24, lshrsi_17_24): Guard with TUNE_68000_10 rather than
1799 "!TARGET_68020 && !TARGET_COLDFIRE". Likewise the unnamed
1800 ashiftrt pattern.
1801
1802 2007-01-12 Richard Sandiford <richard@codesourcery.com>
1803
1804 * config/m68k/m68k.h (TARGET_CPU_CPP_BUILTINS): Increase amount
1805 of tabbing before backslashes.
1806
1807 2007-01-11 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
1808
1809 * pa-linux.h (ASM_OUTPUT_INTERNAL_LABEL): Undefine.
1810 * pa.h (ASM_OUTPUT_LABEL): Output colon when using GAS.
1811 (ASM_OUTPUT_INTERNAL_LABEL): Define.
1812
1813 2007-01-11 Zdenek Dvorak <dvorakz@suse.cz>
1814
1815 * tree-ssa-loop-ivopts.c (extract_cond_operands): Split from
1816 find_interesting_uses_cond.
1817 (find_interesting_uses_cond): Use extract_cond_operands.
1818 (rewrite_use_compare): Use extract_cond_operands and
1819 force_gimple_operand_bsi. Do not call update_stmt.
1820 (determine_use_iv_cost_condition): Use extract_cond_operands.
1821 Return cheaper of using original bound and changing the exit bound.
1822
1823 2007-01-11 Zdenek Dvorak <dvorakz@suse.cz>
1824
1825 PR tree-optimization/29516
1826 * tree-ssa-address.c (tree_mem_ref_addr, add_to_parts,
1827 most_expensive_mult_to_index, addr_to_parts,
1828 create_mem_ref, maybe_fold_tmr): Make the type of
1829 fields of TARGET_MEM_REF sizetype.
1830 (move_fixed_address_to_symbol, move_pointer_to_base):
1831 New functions.
1832 * tree.def (TARGET_MEM_REF): Add comment on types of
1833 the operands.
1834
1835 2007-01-11 Joseph Myers <joseph@codesourcery.com>
1836
1837 * c-common.c (vector_types_convertible_p): Treat opaque types as
1838 always convertible if they have the same size, but not otherwise.
1839
1840 2007-01-11 Steven Bosscher <steven@gcc.gnu.org>
1841
1842 * ifcvt.c (struct noce_if_info): Add comments to the fields.
1843 Remove the b_unconditional field.
1844 (noce_try_sign_mask): Do not look at b_unconditional.
1845 (noce_process_if_block): Do not use merge_if_blocks. Update
1846 the CFG here. Do not set b_unconditional.
1847 (cond_move_process_if_block): Likewise.
1848 (find_cond_trap): Likewise.
1849 (check_cond_move_block): Require simple jump insns at the end
1850 of the basic block.
1851
1852 2007-01-11 Jan Hubicka <jh@suse.cz>
1853
1854 PR tree-optimization/1046
1855 * tree-tailcall.c (suitable_for_tail_call_opt_p): Use TREE_ADDRESSABLE
1856 when alias info is not ready.
1857 (pass_tail_recursion): Do not require aliasing.
1858 * tree-ssa-copyrename.c (pass_rename_ssa_cop): Likewise.
1859 * tree-ssa-ccp.c (pass_ccp, pass_fold_builtins): Likewise.
1860 * tree-ssa-copy.c (pass_copy_prop): Likewise.
1861 * tree-ssa-forwprop.c (pass_forwprop): Likewise.
1862 * tree-ssa-dce.c (pass_dce, pass_dce_loop, pass_cd_dce): Likewise.
1863 * passes.c (init_optimization_passes): Execute rename_ssa_copies,
1864 ccp, forwprop, copy_prop, merge_phi, copy_prop, dce and tail recursion
1865 before inlining.
1866 * tree-ssa-operands.c (add_virtual_operand, get_indirect_ref_operand):
1867 When aliasing is not build, mark statement as volatile.
1868
1869 2007-01-11 Tom Tromey <tromey@redhat.com>
1870
1871 PR preprocessor/15185, PR preprocessor/20989:
1872 * doc/cppopts.texi <-MT>: Update description of algorithm for
1873 computing default target.
1874 <-M, -MD>: Reword "basename" text.
1875
1876 2007-01-11 Roger Sayle <roger@eyesopen.com>
1877
1878 * builtins.c (expand_builtin_pow, expand_builtin_powi,
1879 fold_builtin_cabs, fold_builtin_sqrt, fold_builtin_trunc,
1880 fold_builtin_floor, fold_builtin_ceil, fold_builtin_round,
1881 fold_builtin_int_int_roundingfn, fold_builtin_bitop,
1882 fold_builtin_bswap, real_constp, fold_builtin_pow,
1883 fold_builtin_powi, fold_builtin_signbit, fold_builtin_copysign,
1884 do_mpfr_arg1, do_mpfr_arg2, do_mpfr_arg3, do_mpfr_sincos): Replace
1885 uses of the macro TREE_CONSTANT_OVERFLOW with TREE_OVERFLOW.
1886 * convert.c (convert_to_pointer): Likewise.
1887 * expr.c (highest_pow2_factor, expand_expr_real_1): Likewise.
1888 * fold-const.c (force_fit_type, fold_negate_expr, int_const_binop,
1889 const_binop, fold_convert_const_int_from_int,
1890 fold_convert_const_int_from_real,
1891 fold_convert_const_real_from_real, sign_bit_p,
1892 optimize_minmax_comparison, extract_muldiv_1, fold_div_compare,
1893 fold_sign_changed_comparison, fold_unary, fold_comparison,
1894 fold_binary, multiple_of_p, tree_Expr_non_zero_p,
1895 fold_negate_const, fold_abs_const, fold_not_const): Likewise.
1896 * print-tree.c (print_node_brief, print_node): Likewise.
1897 * stor-layout.c (place_field, layout_type): Likewise.
1898 * tree-chrec.c (keep_cast): Likewise.
1899 * tree.c (build_vector, build_real, build_real_from_int_cst,
1900 build_complex): Likewise.
1901
1902 2007-01-11 Roger Sayle <roger@eyesopen.com>
1903
1904 * tree.h (TREE_CONSTANT_OVERFLOW): Obsolete. For the time being,
1905 treat TREE_CONSTANT_OVERFLOW as a synonym of TREE_OVERFLOW.
1906
1907 2007-01-11 Paolo Bonzini <bonzini@gnu.org>
1908
1909 * configure.ac (strict1_warn): Rename to strict_warn.
1910 (WERROR, --enable-werror, symlink hacks, stage1_cflags,
1911 cc_set_by_configure, quoted_cc_set_by_configure,
1912 stage_prefix_set_by_configure, quoted_stage_prefix_set_by_configure,
1913 all_boot_languages, all_stagestuff): Remove.
1914 (target_list): Remove bootstrap targets.
1915 * Makefile.in (quickstrap): Unconditionally make a synonym of all.
1916 (BOOT_LANGUAGES, STAGE1_CFLAGS, STAGE1_CHECKING,
1917 REMAKEFLAGS, FLAGS_TO_PASS, PREPEND_DOTDOT_TO_RELATIVE_PATHS,
1918 SUBDIR_FLAGS_TO_PASS, WERROR_FLAGS, STRICT2_WARN, LANG_STAGESTUFF,
1919 VOL_FILES, POSTSTAGE1_FLAGS_TO_PASS, STAGE2_FLAGS_TO_PASS,
1920 STAGEPROFILE_FLAGS_TO_PASS, STAGEFEEDBACK_FLAGS_TO_PASS, stage1_build,
1921 stage1_copy, stage2_build, stage2_copy, stageprofile_build,
1922 stageprofile_copy, stage3_build, stage3_copy, stagefeedback_build,
1923 stagefeedback_copy, stage4_build, clean_s1, clean_sw, bootstrap,
1924 bootstrap-lean, bootstrap2, bootstrap2-lean, bootstrap3,
1925 bootstrap3-lean, bootstrap4, bootstrap4-lean, unstage1, unstage2,
1926 unstage3, unstage4, unstageprofile, unstagefeedback, restage, restage2,
1927 restage3, restage4, restageprofile, restagefeedback, bubbleestrap,
1928 cleanstrap, unstrap, restrap, *compare, *compare3, *compare4,
1929 *compare-lean, *compare3-lean, *compare4-lean, stage1-start, stage1,
1930 stage2-start, stage2, stage3-start, stage3, stage4-start, stage4,
1931 stageprofile-start, stageprofile, stagefeedback-start, stagefeedback,
1932 risky-stage1, risky-stage2, risky-stage3, risky-stage4): Remove.
1933 (ORDINARY_FLAGS_TO_PASS): Rename to FLAGS_TO_PASS.
1934 (STAGECOPYSTUFF, STAGEMOVESTUFF): Consolidate into MOSTLYCLEANFILES.
1935 (mostlyclean): Adjust.
1936 (clean, distclean): Don't mention bootstrap stuff.
1937 * configure: Regenerate.
1938 * ada/config-lang.in, cp/config-lang.in, forttran/config-lang.in,
1939 java/config-lang.in, objc/config-lang.in, objcp/config-lang.in,
1940 treelang/config-lang.in (stagestuff): Remove.
1941 * doc/sourcebuild.texi (stage1, stage2, stage3, stage4,
1942 stageprofile, stagefeedback, stagestuff): Remove mention.
1943
1944 2007-01-11 Nick Clifton <nickc@redhat.com>
1945
1946 * config/mcore/predicates.md (mcore_general_movesrc_operand):
1947 Accept CONSTs.
1948 (mcore_general_movdst_operand): Do not accept CONST_INTs.
1949 (mcore_arith_K_S_operand): Run the test for the S constraint not
1950 the test for the M constraint.
1951 (mcore_addsub_operand): Do not accept integer values that are
1952 larger than 32 bits.
1953 * config/mcore/mcore.md: Remove unused constraints from split.
1954 (andsi3): Use HOST_WIDE_INT instead of int to hold an INTVAL.
1955 (addsi3): Likewise.
1956 (allocate_stack): Likewise.
1957 * config/mcore/mcore.c (mcore_print_operand): Restrict output of P
1958 operands to 32 bits.
1959 (mcore_const_costs): Use HOST_WIDE_INT instead of int to hold an
1960 INTVAL.
1961 (mcore_and_cost, mcore_modify_comparison, const_ok_for_mcore,
1962 mcore_const_ok_for_inline, mcore_const_trick_uses_not,
1963 try_constant_tricks, mcore_num_ones, mcore_num_zeros,
1964 mcore_output_bclri, mcore_output_andn, output_inline_const,
1965 mcore_output_move, mcore_output_movedouble): Likewise.
1966 (mcore_output_cmov): Use CONST_OK_FOR_M and CONST_OK_FOR_N.
1967 (output_inline_const): Likewise.
1968 (output_inline_const): Fix format strings used in sprintf
1969 statements.
1970 * config/mcore/mcore-protos.h: Update prototypes for changed
1971 functions in mcore.c.
1972 * config/mcore/mcore.h (CONST_OK_FOR_I): Cast values to
1973 HOST_WIDE_INT and not int.
1974 (CONST_OK_FOR_J, CONST_OK_FOR_K, CONST_OK_FOR_L, CONST_OK_FOR_M,
1975 CONST_OK_FOR_N): Likewise.
1976 (LEGITIMATE_CONSTANT_P): Also check CONSTANT_P.
1977 (GO_IF_LEGITIMATE_INDEX): Use HOST_WIDE_INT instead of int to hold
1978 an INTVAL.
1979
1980 2007-01-10 Jan Hubicka <jh@suse.cz>
1981
1982 * tree-vrp.c (remove_range_assertions): Release defs.
1983 * tree-ssa-loop-ivopts.c (rmeove_statement): Likewise.
1984 * tree-ssa-dom.c (remove_stmt_or_phi): Likewise.
1985
1986 2007-01-10 Paul Brook <paul@codesourcery.com>
1987
1988 * config/arm/arm.c (arm_rtx_costs_1): Handle mutiply-subtract.
1989 * config/arm/arm.md (mulsi3subsi): New insn.
1990
1991 2007-01-10 Zdenek Dvorak <dvorakz@suse.cz>
1992
1993 * tree-ssa-loop-manip.c (tree_unroll_loop): Make it a wrapper over ...
1994 (tree_transform_and_unroll_loop): New.
1995 * tree-flow.h (transform_callback, tree_transform_and_unroll_loop):
1996 Declare.
1997
1998 2007-01-10 Robert Kennedy <jimbob@google.com>
1999
2000 * fold-const.c (fold_comparison): Fold comparisons like (x *
2001 1000 < 0) to (x < 0).
2002
2003 2007-01-10 Ian Lance Taylor <iant@google.com>
2004
2005 * tree-pretty-print.c (dump_generic_node): Print parentheses when
2006 operands have the same priority.
2007
2008 2007-01-10 Tom Tromey <tromey@redhat.com>
2009
2010 * fold-const.c (fold_truthop): Don't check can_use_bit_fields_p.
2011 (fold_binary): Likewise.
2012 * langhooks.c (lhd_can_use_bit_fields_p): Removed.
2013 * langhooks-def.h (lhd_can_use_bit_fields_p): Removed.
2014 (LANG_HOOKS_CAN_USE_BIT_FIELDS_P): Removed.
2015 (LANG_HOOKS_INITIALIZER): Remove LANG_HOOKS_CAN_USE_BIT_FIELDS_P.
2016 * langhooks.h (struct lang_hooks): Removed field
2017 'can_use_bit_fields_p'.
2018
2019 2007-01-10 Ralf Corsépius <ralf.corsepius@rtems.org>
2020
2021 * config/bfin/t-bfin, config/bfin/t-bfin-elf: Remove GCC_CFLAGS.
2022
2023 2007-01-10 Razya Ladelsky <razya@il.ibm.com>
2024
2025 * function.c (get_last_funcdef_no): New function.
2026 * function.h (get_last_funcdef_no): Declare.
2027 * tree-inline.c (initialize_cfun): Add initialization.
2028 (tree_function_versioning): Cleanup.
2029
2030 2007-01-10 Jan Hubicka <jh@suse.cz>
2031
2032 * tree-inline.c (setup_one_parameter): Do not propagate into abnormal
2033 PHIs.
2034
2035 2007-01-10 Sa Liu <saliu@de.ibm.com>
2036 Ben Elliston <bje@au.ibm.com>
2037
2038 * spu.h (STACK_SAVE_AREA): Use VOIDmode for SAVE_FUNCTION, SImode
2039 for SAVE_NONLOCAL and Pmode for any other save level.
2040 * spu-protos.h (spu_restore_stack_block): Declare.
2041 * spu.md (save_stack_block): Remove.
2042 (restore_stack_block): Call spu_restore_stack_block.
2043 * spu.c (spu_restore_stack_block): New function.
2044 (spu_expand_epilogue): Remove old comment.
2045
2046 2007-01-09 Zdenek Dvorak <dvorakz@suse.cz>
2047
2048 PR tree-optimization/30322
2049 * tree-ssa-loop-ivopts.c (fold_affine_expr, iv_value): Removed.
2050 (cand_value_at): Return the value as aff_tree.
2051 (may_eliminate_iv): Convert the bound from aff_tree to tree.
2052 * tree-affine.c (aff_combination_add_cst, aff_combination_add_product,
2053 aff_combination_mult): New functions.
2054 (aff_combination_add): Use aff_combination_add_cst.
2055 (aff_combination_convert): Allow conversions to a wider type.
2056 (tree_to_aff_combination): Handle BIT_NOT_EXPR.
2057 * tree-affine.h (aff_combination_mult): Declare.
2058
2059 2007-01-09 Carlos O'Donell <carlos@codesourcery.com>
2060
2061 * doc/tm.texi: Update documentation to reflect reality of exec
2062 and start file search behaviours. Update copyright year.
2063 * doc/invoke.texi: Explain how GCC_EXEC_PREFIX is used to find
2064 header file directories.
2065
2066 2007-01-09 Uros Bizjak <ubizjak@gmail.com>
2067
2068 * config/i386/i386.md (*sinxf2): Rename to *sinxf2_i387.
2069 (*cosxf2): Rename to cosxf2_i387.
2070 (*sindf2, *sinsf2): Extend operand 1 to XFmode. Macroize patterns
2071 using X87MODEF12 mode macro. Rename patterns to
2072 *sin_extend<mode>xf2_i387. Use SSE_FLOAT_MODE_P to disable patterns
2073 for SSE math.
2074 (*cosdf2, *cossf2): Ditto.
2075 (sincosdf3, sincossf3): Ditto. Rewrite corresponding splitters
2076 to match extended input operands.
2077 (sincos<mode>3): New expander.
2078 (*sinextendsfdf2, *cosextendsfdf2, *sincosextendsfdf3): Remove
2079 insn patterns and corresponding splitters.
2080
2081 2007-01-09 Kaz Kojima <kkojima@gcc.gnu.org>
2082
2083 * config/sh/t-linux (TARGET_LIBGCC2_CFLAGS): Delete.
2084 (SHLIB_MAPFILES, SHLIB_LINK, SHLIB_INSTALL): Likewise.
2085
2086 2007-01-09 Nicolas Pitre <nico@cam.org>
2087
2088 PR target/30173
2089 * arm/ieee754-df.S (Lad_s): Also test the low word of X for zero.
2090
2091 2007-01-08 Geoffrey Keating <geoffk@apple.com>
2092
2093 * target.h (struct gcc_target): New field library_rtti_comdat.
2094 * target-def.h (TARGET_CXX_LIBRARY_RTTI_COMDAT): New.
2095 (TARGET_CXX): Add TARGET_CXX_LIBRARY_RTTI_COMDAT.
2096 * doc/tm.texi (C++ ABI): Document TARGET_CXX_LIBRARY_RTTI_COMDAT.
2097 * config/darwin.h (TARGET_CXX_LIBRARY_RTTI_COMDAT): Define.
2098
2099 2007-01-08 Geoffrey Keating <geoffk@apple.com>
2100
2101 * doc/invoke.texi (Optimize Options): Correct description of -O0.
2102
2103 2007-01-08 Richard Guenther <rguenther@suse.de>
2104
2105 * tree.h (force_fit_type_double): Export.
2106 (force_fit_type): Remove.
2107 * fold-const.c (force_fit_type_double): New function.
2108 (force_fit_type): Remove.
2109 (int_const_binop): Use it.
2110 (fold_convert_const_int_from_int): Likewise.
2111 (fold_convert_const_int_from_real): Likewise.
2112 (fold_div_compare): Likewise.
2113 (fold_sign_changed_comparison): Likewise.
2114 (fold_unary): Likewise.
2115 (fold_negate_const): Likewise.
2116 (fold_abs_const): Likewise.
2117 (fold_not_const): Likewise.
2118 * c-common.c (shorten_compare): Use force_fit_type_double.
2119 * convert.c (convert_to_pointer): Likewise.
2120
2121 2007-01-08 Richard Guenther <rguenther@suse.de>
2122
2123 * tree.h (build_int_cst_wide_type): Export.
2124 * tree.c (build_int_cst_wide_type): New function.
2125 (build_int_cst_wide): Fix comment.
2126 * builtins.c (fold_builtin_object_size): Use build_int_cst
2127 to build -1 or 0 of the correct type. Use fit_double_type
2128 to check for overflow.
2129 * fold-const.c (optimize_bit_field_compare): Use build_int_cst_type
2130 to build the mask.
2131 (decode_field_reference): Likewise.
2132 (all_ones_mask_p): Likewise.
2133 (native_interpret_int): Use build_int_cst_wide_type.
2134 (fold_binary): Use build_int_cst_type to build an all-ones
2135 value.
2136 * stor-layout.c (set_sizetype): Use build_int_cst_wide_type.
2137
2138 2007-01-08 Daniel Jacobowitz <dan@codesourcery.com>
2139
2140 * config/pa/t-pa64 (libgcc_stub.a): Use $(T).
2141
2142 2007-01-09 Ben Elliston <bje@au.ibm.com>
2143
2144 * genautomata.c (STATS_OPTION): New option.
2145 (stats_flag): New flag.
2146 (gen_automata_option): Handle it.
2147 (initiate_automaton_gen): Ditto.
2148 (write_automata): Output statistics only if stats_flag is
2149 set. Likewise, output time statistics only if time_flag is set.
2150 * doc/md.texi (Processor pipeline description): Document new flag.
2151
2152 2007-01-08 Richard Guenther <rguenther@suse.de>
2153
2154 * builtins.c (fold_builtin_int_roundingfn): Use fit_double_type.
2155 * tree.c (build_int_cst_type): Likewise.
2156 (size_in_bytes): Don't call force_fit_type on the result.
2157 (int_fits_type_p): Use fit_double_type.
2158 * fold-const.c (fit_double_type): New function.
2159 (force_fit_type): Use it.
2160 * tree.h (fit_double_type): Export.
2161
2162 2007-01-08 Jan Hubicka <jh@suse.cz>
2163
2164 * tree-vectorizer.c (gate_increase_alignment): Fix return type.
2165 * ipa.c (function_and_variable_visibility): Fix return type.
2166
2167 2007-01-08 Richard Guenther <rguenther@suse.de>
2168
2169 * tree-ssa-ccp.c (maybe_fold_offset_to_array_ref): Use type
2170 of offset to build the index.
2171 * tree-pretty-print.c (dump_generic_node): Don't build negated
2172 const just for printing.
2173 * c-pretty-print.c (pp_c_integer_constant): Likewise.
2174 * builtins.c (fold_builtin_int_roundingfn): Check if result
2175 fits the type by using force_fit_type and comparing the result.
2176 * predict.c (predict_loops): Use compare_tree_int for comparison.
2177 * tree.c (build_int_cst): Fall back to integer_type_node for
2178 NULL_TREE type.
2179 (build_int_cst_wide): Assert type is non-null.
2180
2181 2007-01-08 Roberto Costa <roberto.costa@st.com>
2182
2183 * tree-vrp.c (extract_range_from_cond_expr): New.
2184 (extract_range_from_expr): Handle COND_EXPR nodes used as expressions.
2185 * tree-ssa-ccp.c (get_maxval_strlen): Handle COND_EXPR nodes used
2186 as expressions.
2187 (fold_stmt): Bug fix, avoid infinite recursion when folding COND_EXPRs.
2188 * tree-ssa-forwprop.c (simplify_cond, forward_propagate_into_cond,
2189 tree_ssa_forward_propagate_single_use_vars): Handle COND_EXPR nodes
2190 used as expressions.
2191 * tree-object-size.c (cond_expr_object_size): New.
2192 (collect_object_sizes_for): Handle COND_EXPR nodes used as expressions.
2193
2194 2007-01-08 Jan Hubicka <jh@suse.cz>
2195
2196 * tree-ssa-forwprop.c (forward_propagate_into_cond,
2197 tree_ssa_forward_propagate_single_use_va): Release defs of propagated
2198 statement.
2199
2200 2007-01-08 Richard Guenther <rguenther@suse.de>
2201
2202 PR tree-optimization/23603
2203 * tree-vrp.c (set_value_range_to_truthvalue): New function.
2204 (extract_range_from_binary): Fall back to truthvalue instead of
2205 varying for TRUTH_*_EXPR.
2206 (extract_range_from_comparison): Fall back to truthvalue instead of
2207 varying.
2208 (vrp_visit_phi_node): Don't adjust new range bounds to +INF/-INF
2209 if all visited PHI values were constant.
2210
2211 2007-01-08 Jan Hubicka <jh@suse.cz>
2212
2213 * cgraphunit.c (cgraph_process_new_functions): Reset reachable flag.
2214 (cgraph_analyze_function): break out from ...
2215 (cgraph_finalize_compilation_unit): ... here.
2216 (cgraph_expand_function): Remove forgoten commented out line.
2217 (cgraph_optimize): Analyze functions.
2218
2219 2007-01-08 Jan Hubicka <jh@suse.cz>
2220
2221 * tree-pas.h (TODO_remove_function): New flag.
2222 (TODO_update*): Renumber.
2223 (pass_ipa_increase_alignment,
2224 pass_ipa_function_and_variable_visibility): New passes.
2225 * cgraphunit.c (cgraph_increase_alignment): Move to tree-vectorizer.c
2226 (cgraph_function_and_variable_visibility): Move to ipa.c
2227 (cgraph_optimize): Don't call cgraph_function_and_variable_visibility,
2228 cgraph_increase_alignment.
2229 * ipa-inline.c (cgraph_decide_inlining): Don't push timevar.
2230 (cgraph_decide_inlining_incrementally): Push TV_INTEGRATION before
2231 calling tree-inline.
2232 (cgraph_early_inlining): Do not call cgraph_remove_unreachable_nodes.
2233 (pass_ipa_inline, pass_early_ipa_inlining): Set TODO_remove_functions
2234 * tree-vectorizer.c (increase_alignment): Move here from cgraphunit.c
2235 (gate_increase_alignment): New function.
2236 (pass_ipa_increase_alignment): New pass.
2237 * ipa.c: Inline tree-pass.h and timevar.h
2238 (function_and_variable_visibility): Move here from cgraphunit.c
2239 * tree-optimize.c (pass_early_local_passes): Add TODO_remove_functions.
2240 * passes.c (init_optimization_passes): Add the two new passes.
2241 (execute_todo): Handle cgraph_remove_functions.
2242
2243 2007-01-08 Nick Clifton <nickc@redhat.com>
2244
2245 * config/frv/predicates.md (reg_or_0_operand): Accept
2246 CONST_DOUBLEs.
2247
2248 2007-01-08 Ralf Corsépius <ralf.corsepius@rtems.org>
2249
2250 * config/bfin/rtems.h, config/bfin/t-rtems: New.
2251 * config.gcc: Add bfin*-rtems*.
2252
2253 2007-01-08 Mark Shinwell <shinwell@codesourcery.com>
2254
2255 * c.opt: Add -flax-vector-conversions.
2256 * c-typeck.c (convert_for_assignment): Pass flag to
2257 vector_types_convertible_p to allow emission of note.
2258 (digest_init): Likewise.
2259 * c-opts.c: Handle -flax-vector-conversions.
2260 * c-common.c (flag_lax_vector_conversions): New.
2261 (vector_types_convertible_p): Unless -flax-vector conversions
2262 has been passed, disallow conversions between vectors with
2263 differing numbers of subparts and/or element types. If such
2264 a conversion is disallowed, possibly emit a note on the first
2265 occasion only to inform the user of -flax-vector-conversions.
2266 The new last argument specifies this.
2267 * c-common.h (flag_lax_vector_conversions): New.
2268 (vector_types_convertible_p): Add extra argument.
2269 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use
2270 char_type_node for V*QI type vectors.
2271 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
2272 Update to satisfy new typechecking rules.
2273 * config/rs6000/altivec.h (vec_cmple): Use vec_cmpge, for both
2274 C and C++ variants.
2275 * doc/invoke.texi (C Dialect Options): Document
2276 -flax-vector-conversions.
2277
2278 2007-01-08 Mark Shinwell <shinwell@codesourcery.com>
2279
2280 PR tree-optimization/29877
2281 * tree-ssa-ter.c (is_replaceable_p): Deem assignments with
2282 a register variable on the RHS to not be replaceable.
2283
2284 2007-01-08 Chen Liqin <liqin@sunnorth.com.cn>
2285 * config/score/t-score-elf (MULTILIB_OPTIONS): Change.
2286 * config/score/predicates.md (const_uimm5, sr0_operand, const_simm12,
2287 const_simm15, const_pow2, const_npow2): Added.
2288 * config/score/misc.md (insv, extv, extzv, movmemsi,
2289 move_lbu_a/b, mov_lhu_a/b etc): Added and fix some bug.
2290 * config/score/score.c (score_address_cost, score_select_cc_mode):
2291 Added.
2292 Change CONST_OK_FOR_LETTER_P/EXTRA_CONSTRAINT define.
2293 Update score_rtx_costs for MACRO TARGET_RTX_COSTS.
2294 Update score_print_operand.
2295 * config/score/score.h (DATA_ALIGNMENT, SELECT_CC_MODE): Added.
2296 Adjust register allocate order and update some macro define.
2297 * config/score/score-mdaux.c (mdx_unaligned_load, mdx_unsigned_store,
2298 mdx_block_move_straight, mdx_block_move_loop_head,
2299 mdx_block_move_loop_body, mdx_block_move_loop_foot, mdx_block_move_loop,
2300 mdx_block_move): Added.
2301 (mdx_movsicc, mdp_select_add_imm, mdp_select, mds_zero_extract_andi,
2302 mdp_limm): Updated and fix some bug and typo.
2303 * config/score/score.md (movqi/hi/si, add/sub/zero/ext): Updated.
2304 (movsf, movdf, doloop_end): Added.
2305
2306 2007-01-08 Kazu Hirata <kazu@codesourcery.com>
2307
2308 * config/arm/arm.c, config/arm/arm.h, config/arm/arm.md,
2309 config/arm/thumb2.md: Fix comment typos.
2310 * doc/extend.texi: Fix a typo.
2311
2312 2007-01-07 Eric Christopher <echristo@apple.com>
2313
2314 * configure.ac: Check for __stack_chk_fail for darwin.
2315 * configure: Regenerate.
2316
2317 2007-01-07 Richard Guenther <rguenther@suse.de>
2318
2319 * tree-vrp.c (extract_range_from_assert): CSE calls to
2320 compare_values where possible.
2321 (extract_range_from_unary_expr): Likewise.
2322
2323 2007-01-07 Anatoly Sokolov <aesok@post.ru>
2324
2325 * config/avr/avr-protos.h (call_insn_operand): Delete prototype.
2326 * config/avr/avr.c (call_insn_operand): Delete function.
2327 * config/avr/avr.md (*pushqi, *pushhi, *pushsi, *pushsf): Use REG_SP
2328 instead of register number. Use predicates.
2329 * config/avr/predicates.md (const0_operand, reg_or_0_operand,
2330 call_insn_operand): Add.
2331
2332 2007-01-06 Jan Hubicka <jh@suse.cz>
2333
2334 * tree-pass.h (pass_build_cgraph_edges): Declare.
2335 * cgraphunit.c (record_refernece): Move to cgraphbuild.c
2336 (visited_nodes): Remove.
2337 (cgraph_create_edges): Move to cgraphbuild.c; rename to
2338 build_cgrpah_edges; make visited_nodes local.
2339 (cgraph_process_new_functions): DO not call initialize_inline_failed.
2340 (record_references_in_initializer): Move to cgraphbuild.c
2341 (initialize_inline_failed, rebuild_cgraph_edges,
2342 pass_rebuild_cgraph_edges): Move to cgraphbuild.c.
2343 (verify_cgraph_node): Make visited_nodes local.
2344 (cgraph_analyze_function): Do not call cgraph_create_edges and
2345 initialize_inline_failed.
2346 (cgraph_expand_function): Do not call cgraph_lower_function;
2347 assert that function is already lowered.
2348 * Makefile.in (cgraphbuild.o): New.
2349 * passes.c (init_optimization_passes): Add pass_build_cgraph_edges
2350 at the end of lowering passes.
2351
2352 2007-01-06 Steven Bosscher <steven@gcc.gnu.org>
2353
2354 * ifcvt.c (cond_move_convert_if_block): New function, code
2355 factored out from...
2356 (cond_move_process_if_block): ...here. Call the new function
2357 on the THEN and ELSE blocks.
2358 (merge_if_block): Do not copy global_live_at_end, merge_blocks
2359 already takes care of this.
2360
2361 2007-01-05 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
2362
2363 PR c/19978
2364 * tree.h (TREE_OVERFLOW_P): New.
2365 * c-typeck.c (parser_build_unary_op): Warn only if result
2366 overflowed and operands did not.
2367 (parser_build_binary_op): Likewise.
2368 (convert_for_assignment): Remove redundant overflow_warning.
2369 * c-common.c (overflow_warning): Don't check or set TREE_OVERFLOW.
2370
2371 2007-01-05 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
2372
2373 * c-typeck.c (store_init_value): Split over two lines to follow
2374 the GNU coding style.
2375
2376 2007-01-05 Benjamin Kosnik <bkoz@redhat.com>
2377
2378 * c-cppbuiltin.c (c_cpp_builtins): __GXX_EXPERIMENTAL_CPP0X__ to
2379 __GXX_EXPERIMENTAL_CXX0X__.
2380 * doc/cpp.texi: Same.
2381
2382 2007-01-05 Richard Guenther <rguenther@suse.de>
2383
2384 PR middle-end/27826
2385 * tree.c (get_narrower): Do not construct COMPONENT_REFs
2386 with mismatched types. Instead explicitly build a
2387 conversion NOP_EXPR.
2388
2389 2007-01-05 Ian Lance Taylor <iant@google.com>
2390
2391 * c-common.c (decl_with_nonnull_addr_p): New function.
2392 (c_common_truthvalue_conversion): Call it.
2393 * c-typeck.c (build_binary_op): Likewise.
2394 * c-common.h (decl_with_nonnull_addr_p): Declare.
2395
2396 2007-01-05 Jakub Jelinek <jakub@redhat.com>
2397
2398 PR c/30360
2399 * libgcc2.c (__divdc3): Compare c and d against 0.0 instead of
2400 denom against 0.0.
2401
2402 2007-01-05 Joel Brobecker <brobecker@adacore.com>
2403
2404 * doc/install.texi (Final install): Document the fact that
2405 the GNAT runtime should not be stripped.
2406
2407 2007-01-04 Jan Hubicka <jh@suse.cz>
2408
2409 * tree-inline.c (fold_marked_statements): Update operand caches
2410 and EH after folding
2411
2412 2007-01-04 Ian Lance Taylor <iant@google.com>
2413
2414 * c-common.c (check_function_nonnull): Whitespace fix.
2415
2416 2007-01-04 Jan Hubicka <jh@suse.cz>
2417
2418 * tree-optimize.c (execute_fixup_cfg): Correct previously mistakely
2419 comitted older version of patch.
2420 (pass_fixup_cfg): Add TODOs to verify flow and statements, dump
2421 function, celanup cfg and collect garbage.
2422
2423 2007-01-04 Mike Stump <mrs@apple.com>
2424
2425 * Makefile.in (mostlyclean): Don't remove libgcc anymore.
2426 (clean): Likewise.
2427
2428 2007-01-04 Eric Christopher <echristo@apple.com>
2429
2430 * libgcc2.c (__bswapsi2): Use SItype.
2431 (__bswapdi2): Use DItype.
2432 * libgcc2.h: Update for above.
2433
2434 2007-01-04 Paul Brook <paul@codesourcery.com>
2435
2436 * config/arm/arm.md (arm_mulsi3, thumb_mulsi3, mulsi3_compare0,
2437 mulsi_compare0_scratch, mulsi3addsi, mulsi3addsi_compare0,
2438 mulsi3addsi_compare0_scratch, mulsidi3adddi, mulsidi3,
2439 umulsidi3, umulsidi3adddi, smulsi3_highpart,
2440 umulsi3_highpart): Make conditional on !arm_arch6.
2441 (arm_mulsi3_v6, thumb_mulsi3_v6, mulsi3_compare0_v6,
2442 mulsi_compare0_scratch_v6, mulsi3addsi_v6, mulsi3addsi_compare0_v6,
2443 mulsi3addsi_compare0_scratch_v6, mulsidi3adddi_v6, mulsidi3_v6,
2444 umulsidi3_v6, umulsidi3adddi_v6, smulsi3_highpart_v6,
2445 umulsi3_highpart_v6): New insns.
2446
2447 2007-01-04 Roger Sayle <roger@eyesopen.com>
2448
2449 * fold-const.c (fold_convert): When casting an expression to void,
2450 fold_ignored_result may discover a GIMPLE_MODIFY_STMT which doesn't
2451 have a type. Instead of attempting to build a NOP_EXPR, return
2452 these "special" trees directly.
2453
2454 2007-01-04 Joseph Myers <joseph@codesourcery.com>
2455
2456 * config/rs6000/rs6000.c (rs6000_rtx_costs): Make adjustment for
2457 MULT inside MINUS as either argument. Use rs6000_cost->dmul -
2458 rs6000_cost->fp not 0 as adjustment for outer NEG.
2459
2460 2007-01-04 Jan Hubicka <jh@suse.cz>
2461
2462 * cgraph.c (cgraph_release_function_body): New function.
2463 (cgraph_remove_node): Use it.
2464 * cgraph.h (cgraph_release_function_body): Declare.
2465 * cgraphunit.c (cgraph_expand_function): Use it.
2466 * ipa.c (cgraph_remove_unreahchable_nodes): Use it.
2467 * tree-ssa.c (delete_tree_ssa): Allow to be called before aliasing
2468 is initialized and while compilation of other function is running.
2469 * tree-optimize.c (execute_free_cfg_annotations): Move code to clear
2470 statement CFG annotations from here to ...
2471 * tree-cfg.c (delete_tree_cfg_annotations): ... here.
2472
2473 2007-01-04 Zdenek Dvorak <dvorakz@suse.cz>
2474
2475 * cfgloop.h (enum li_flags): Make the constants powers of two.
2476
2477 2007-01-04 Jan Hubicka <jh@suse.cz>
2478
2479 * tree-inline.c (copy_bb): Insert new statements to statements_to_fold
2480 set.
2481 (fold_marked_statements): New function.
2482 (optimize_inline_calls, tree_function_versioning): Fold new statements.
2483 * tree-inline.h (copy_body_data): Add statements_to_fold.
2484
2485 2007-01-03 Daniel Jacobowitz <dan@codesourcery.com>
2486
2487 * config.gcc: Mention libgcc/config.host.
2488 * Makefile.in: Update comments mentioning libgcc.
2489 (LIBGCC, INSTALL_LIBGCC, GCC_PARTS, mklibgcc): Delete.
2490 (all.cross, start.encap, rest.encap, rest.cross): Update
2491 dependencies for libgcc move.
2492 (libgcc.mk, LIBGCC_DEPS, libgcov.a, libgcc.a, stmp-multilib)
2493 (clean-target, clean-target-libgcc): Delete.
2494 (srcdirify, GCC_EXTRA_PARTS): New macros.
2495 (libgcc-support, libgcc.mvars): New rules.
2496 (distclean): Remove mention of mklibgcc.
2497 (install): Don't reference INSTALL_LIBGCC.
2498 (install-common): Don't reference EXTRA_PARTS.
2499 (install-libgcc, install-multilib): Delete rules.
2500 * mklibgcc.in: Delete file.
2501 * doc/configfiles.texi: Don't mention mklibgcc.
2502
2503 * config/i386/t-darwin (SHLIB_VERPFX): Delete (moved to libgcc).
2504 * config/i386/t-darwin64 (SHLIB_VERPFX): Likewise.
2505 * config/rs6000/t-darwin (SHLIB_VERPFX): Likewise.
2506 * config/rs6000/t-ppccomm (TARGET_LIBGCC2_CFLAGS, SHLIB_MAPFILES)
2507 (mklibgcc, ldblspecs): Likewise.
2508
2509 * config/i386/t-nwld (libgcc.def, libc.def, libpcre.def)
2510 (posixpre.def): Use $(T).
2511 (SHLIB_EXT, SHLIB_NAME, SHLIB_SLIBDIR_QUAL, SHLIB_DEF, SHLIB_MAP)
2512 (SHLIB_SRC, SHLIB_INSTALL): Delete.
2513 (SHLIB_LINK): Make dummy.
2514 * config/t-slibgcc-darwin: Delete contents except for dummy SHLIB_LINK.
2515
2516 * config/frv/t-linux (EXTRA_MULTILIB_PARTS): Clear.
2517
2518 * config/alpha/t-crtfm: Use $(T) in rules for EXTRA_PARTS.
2519 * config/alpha/t-vms, config/alpha/t-vms64, config/fr30/t-fr30,
2520 config/i386/t-rtems-i386, config/ia64/t-ia64, config/rs6000/t-beos,
2521 config/rs6000/t-newas, config/sparc/t-elf: Likewise.
2522
2523 * configure.ac (all_outputs): Remove mklibgcc.
2524 * configure: Regenerated.
2525
2526 2007-01-03 Josh Conner <jconner@apple.com>
2527
2528 PR middle-end/29683
2529 * calls.c (compute_argument_addresses): Set stack and stack_slot
2530 for partial args, too.
2531 (store_one_arg): Use locate.size.constant for the size when
2532 generating a save_area.
2533
2534 2007-01-03 Robert Kennedy <jimbob@google.com>
2535
2536 * tree-cfg.c (tree_merge_blocks): Release SSA_NAME phi results
2537 whose definitions are deleted due to basic block merging.
2538
2539 2007-01-03 Paul Brook <paul@codesourcery.com>
2540
2541 PR target/16634
2542 * config/arm/arm.c (output_return_instruction): Pop PC in interrupt
2543 functions.
2544 (use_return_insn): Return 0 for Thumb interrupt functions.
2545 (print_multi_reg): Add rfe argument for IRQ returns.
2546 (arm_output_epilogue): Pop interrupt return address directly into PC.
2547 (arm_expand_prologue): Only adjust IRQ return address in Arm mode.
2548
2549 2007-01-03 Paul Brook <paul@codesourcery.com>
2550
2551 Merge from sourcerygxx-4_1.
2552 * config/arm/thumb2.md: New file.
2553 * config/arm/elf.h (JUMP_TABLES_IN_TEXT_SECTION): Return True for
2554 Thumb-2.
2555 * config/arm/coff.h (JUMP_TABLES_IN_TEXT_SECTION): Ditto.
2556 * config/arm/aout.h (ASM_OUTPUT_ADDR_VEC_ELT): Add !Thumb-2 assertion.
2557 (ASM_OUTPUT_ADDR_DIFF_ELT): Output Thumb-2 jump tables.
2558 * config/arm/aof.h (ASM_OUTPUT_ADDR_DIFF_ELT): Output Thumb-2 jump
2559 tables.
2560 (ASM_OUTPUT_ADDR_VEC_ELT): Add !Thumb-2 assertion.
2561 * config/arm/ieee754-df.S: Use macros for Thumb-2/Unified asm
2562 comptibility.
2563 * config/arm/ieee754-sf.S: Ditto.
2564 * config/arm/arm.c (thumb_base_register_rtx_p): Rename...
2565 (thumb1_base_register_rtx_p): ... to this.
2566 (thumb_index_register_rtx_p): Rename...
2567 (thumb1_index_register_rtx_p): ... to this.
2568 (thumb_output_function_prologue): Rename...
2569 (thumb1_output_function_prologue): ... to this.
2570 (thumb_legitimate_address_p): Rename...
2571 (thumb1_legitimate_address_p): ... to this.
2572 (thumb_rtx_costs): Rename...
2573 (thumb1_rtx_costs): ... to this.
2574 (thumb_compute_save_reg_mask): Rename...
2575 (thumb1_compute_save_reg_mask): ... to this.
2576 (thumb_final_prescan_insn): Rename...
2577 (thumb1_final_prescan_insn): ... to this.
2578 (thumb_expand_epilogue): Rename...
2579 (thumb1_expand_epilogue): ... to this.
2580 (arm_unwind_emit_stm): Rename...
2581 (arm_unwind_emit_sequence): ... to this.
2582 (thumb2_legitimate_index_p, thumb2_legitimate_address_p,
2583 thumb1_compute_save_reg_mask, arm_dwarf_handle_frame_unspec,
2584 thumb2_index_mul_operand, output_move_vfp, arm_shift_nmem,
2585 arm_save_coproc_regs, thumb_set_frame_pointer, arm_print_condition,
2586 thumb2_final_prescan_insn, thumb2_asm_output_opcode, arm_output_shift,
2587 thumb2_output_casesi): New functions.
2588 (TARGET_DWARF_HANDLE_FRAME_UNSPEC): Define.
2589 (FL_THUMB2, FL_NOTM, FL_DIV, FL_FOR_ARCH6T2, FL_FOR_ARCH7,
2590 FL_FOR_ARCH7A, FL_FOR_ARCH7R, FL_FOR_ARCH7M, ARM_LSL_NAME,
2591 THUMB2_WORK_REGS): Define.
2592 (arm_arch_notm, arm_arch_thumb2, arm_arch_hwdiv, arm_condexec_count,
2593 arm_condexec_mask, arm_condexec_masklen)): New variables.
2594 (all_architectures): Add armv6t2, armv7, armv7a, armv7r and armv7m.
2595 (arm_override_options): Check new CPU capabilities.
2596 Set new architecture flag variables.
2597 (arm_isr_value): Handle v7m interrupt functions.
2598 (user_return_insn): Return 0 for v7m interrupt functions. Handle
2599 Thumb-2.
2600 (const_ok_for_arm): Handle Thumb-2 constants.
2601 (arm_gen_constant): Ditto. Use movw when available.
2602 (arm_function_ok_for_sibcall): Return false for v7m interrupt
2603 functions.
2604 (legitimize_pic_address, arm_call_tls_get_addr): Handle Thumb-2.
2605 (thumb_find_work_register, arm_load_pic_register,
2606 legitimize_tls_address, arm_address_cost, load_multiple_sequence,
2607 emit_ldm_seq, emit_stm_seq, arm_select_cc_mode, get_jump_table_size,
2608 print_multi_reg, output_mov_long_double_fpa_from_arm,
2609 output_mov_long_double_arm_from_fpa, output_mov_double_fpa_from_arm,
2610 output_mov_double_fpa_from_arm, output_move_double,
2611 arm_compute_save_reg_mask, arm_compute_save_reg0_reg12_mask,
2612 output_return_instruction, arm_output_function_prologue,
2613 arm_output_epilogue, arm_get_frame_offsets, arm_regno_class,
2614 arm_output_mi_thunk, thumb_set_return_address): Ditto.
2615 (arm_expand_prologue): Handle Thumb-2. Use arm_save_coproc_regs.
2616 (arm_coproc_mem_operand): Allow POST_INC/PRE_DEC.
2617 (arithmetic_instr, shift_op): Use arm_shift_nmem.
2618 (arm_print_operand): Use arm_print_condition. Handle '(', ')', '.',
2619 '!' and 'L'.
2620 (arm_final_prescan_insn): Use extract_constrain_insn_cached.
2621 (thumb_expand_prologue): Use thumb_set_frame_pointer.
2622 (arm_file_start): Output directive for unified syntax.
2623 (arm_unwind_emit_set): Handle stack alignment instruction.
2624 * config/arm/lib1funcs.asm: Remove default for __ARM_ARCH__.
2625 Add v6t2, v7, v7a, v7r and v7m.
2626 (RETLDM): Add Thumb-2 code.
2627 (do_it, shift1, do_push, do_pop, COND, THUMB_SYNTAX): New macros.
2628 * config/arm/arm.h (TARGET_CPU_CPP_BUILTINS): Define __thumb2__.
2629 (TARGET_THUMB1, TARGET_32BIT, TARGET_THUMB2, TARGET_DSP_MULTIPLY,
2630 TARGET_INT_SIMD, TARGET_UNIFIED_ASM, ARM_FT_STACKALIGN, IS_STACKALIGN,
2631 THUMB2_TRAMPOLINE_TEMPLATE, TRAMPOLINE_ADJUST_ADDRESS,
2632 ASM_OUTPUT_OPCODE, THUMB2_GO_IF_LEGITIMATE_ADDRESS,
2633 THUMB2_LEGITIMIZE_ADDRESS, CASE_VECTOR_PC_RELATIVE,
2634 CASE_VECTOR_SHORTEN_MODE, ADDR_VEC_ALIGN, ASM_OUTPUT_CASE_END,
2635 ADJUST_INSN_LENGTH): Define.
2636 (TARGET_REALLY_IWMMXT, TARGET_IWMMXT_ABI, CONDITIONAL_REGISTER_USAGE,
2637 STATIC_CHAIN_REGNUM, HARD_REGNO_NREGS, INDEX_REG_CLASS,
2638 BASE_REG_CLASS, MODE_BASE_REG_CLASS, SMALL_REGISTER_CLASSES,
2639 PREFERRED_RELOAD_CLASS, SECONDARY_OUTPUT_RELOAD_CLASS,
2640 SECONDARY_INPUT_RELOAD_CLASS, LIBCALL_VALUE, FUNCTION_VALUE_REGNO_P,
2641 TRAMPOLINE_SIZE, INITIALIZE_TRAMPOLINE, HAVE_PRE_INCREMENT,
2642 HAVE_POST_DECREMENT, HAVE_PRE_DECREMENT, HAVE_PRE_MODIFY_DISP,
2643 HAVE_POST_MODIFY_DISP, HAVE_PRE_MODIFY_REG, HAVE_POST_MODIFY_REG,
2644 REGNO_MODE_OK_FOR_BASE_P, LEGITIMATE_CONSTANT_P,
2645 REG_MODE_OK_FOR_BASE_P, REG_OK_FOR_INDEX_P, GO_IF_LEGITIMATE_ADDRESS,
2646 LEGITIMIZE_ADDRESS, THUMB2_LEGITIMIZE_ADDRESS,
2647 GO_IF_MODE_DEPENDENT_ADDRESS, MEMORY_MOVE_COST, BRANCH_COST,
2648 ASM_APP_OFF, ASM_OUTPUT_CASE_LABEL, ARM_DECLARE_FUNCTION_NAME,
2649 FINAL_PRESCAN_INSN, PRINT_OPERAND_PUNCT_VALID_P,
2650 PRINT_OPERAND_ADDRESS): Adjust for Thumb-2.
2651 (arm_arch_notm, arm_arch_thumb2, arm_arch_hwdiv): New declarations.
2652 * config/arm/arm-cores.def: Add arm1156t2-s, cortex-a8, cortex-r4 and
2653 cortex-m3.
2654 * config/arm/arm-tune.md: Regenerate.
2655 * config/arm/arm-protos.h: Update prototypes.
2656 * config/arm/vfp.md: Enable patterns for Thumb-2.
2657 (arm_movsi_vfp): Add movw alternative. Use output_move_vfp.
2658 (arm_movdi_vfp, movsf_vfp, movdf_vfp): Use output_move_vfp.
2659 (thumb2_movsi_vfp, thumb2_movdi_vfp, thumb2_movsf_vfp,
2660 thumb2_movdf_vfp, thumb2_movsfcc_vfp, thumb2_movdfcc_vfp): New.
2661 * config/arm/libunwind.S: Add Thumb-2 code.
2662 * config/arm/constraints.md: Update include Thumb-2.
2663 * config/arm/ieee754-sf.S: Add Thumb-2/Unified asm support.
2664 * config/arm/ieee754-df.S: Ditto.
2665 * config/arm/bpabi.S: Ditto.
2666 * config/arm/t-arm (MD_INCLUDES): Add thumb2.md.
2667 * config/arm/predicates.md (low_register_operand,
2668 low_reg_or_int_operand, thumb_16bit_operator): New.
2669 (thumb_cmp_operand, thumb_cmpneg_operand): Rename...
2670 (thumb1_cmp_operand, thumb1_cmpneg_operand): ... to this.
2671 * config/arm/t-arm-elf: Add armv7 multilib.
2672 * config/arm/arm.md: Update patterns for Thumb-2 and Unified asm.
2673 Include thumb2.md.
2674 (UNSPEC_STACK_ALIGN, ce_count): New.
2675 (arm_incscc, arm_decscc, arm_umaxsi3, arm_uminsi3,
2676 arm_zero_extendsidi2, arm_zero_extendqidi2): New
2677 insns/expanders.
2678 * config/arm/fpa.md: Update patterns for Thumb-2 and Unified asm.
2679 (thumb2_movsf_fpa, thumb2_movdf_fpa, thumb2_movxf_fpa,
2680 thumb2_movsfcc_fpa, thumb2_movdfcc_fpa): New insns.
2681 * config/arm/cirrus.md: Update patterns for Thumb-2 and Unified asm.
2682 (cirrus_thumb2_movdi, cirrus_thumb2_movsi_insn,
2683 thumb2_cirrus_movsf_hard_insn, thumb2_cirrus_movdf_hard_insn): New
2684 insns.
2685 * doc/extend.texi: Document ARMv7-M interrupt functions.
2686 * doc/invoke.texi: Document Thumb-2 new cores+architectures.
2687
2688 2007-01-03 Jakub Jelinek <jakub@redhat.com>
2689
2690 * unwind-dw2.c (SIGNAL_FRAME_BIT, EXTENDED_CONTEXT_BIT): Define.
2691 (struct _Unwind_Context): Rename args_size to flags, remove
2692 signal_frame field, add a new args_size field and version field.
2693 (_Unwind_IsSignalFrame, _Unwind_SetSignalFrame,
2694 _Unwind_IsExtendedContext): New inline functions.
2695 (_Unwind_GetGR, _Unwind_SetGR, _Unwind_GetGRPtr, _Unwind_SetGRPtr):
2696 Assume by_value array is only present if _Unwind_IsExtendedContext.
2697 (_Unwind_GetIPInfo, execute_cfa_program, uw_frame_state_for): Use
2698 _Unwind_IsSignalFrame.
2699 (__frame_state_for): Initialize context.flags to EXTENDED_CONTEXT_BIT.
2700 (uw_update_context_1): Use _Unwind_SetSignalFrame.
2701 (uw_init_context_1): Initialize context->flags to
2702 EXTENDED_CONTEXT_BIT.
2703 * config/rs6000/linux-unwind.h (frob_update_context): Use
2704 _Unwind_SetSignalFrame.
2705
2706 2007-01-03 Andrew Pinski <andrew_pinski@playstation.sony.com>
2707
2708 PR middle-end/30353
2709 * gimplify.c (gimplify_modify_expr_complex_part): Move below
2710 tree_to_gimple_tuple. Call tree_to_gimple_tuple when we need
2711 the value.
2712
2713 2007-01-03 Kazu Hirata <kazu@codesourcery.com>
2714
2715 * config/i386/i386.h (NON_STACK_REG_P, REGNO_OK_FOR_SIREG_P,
2716 REGNO_OK_FOR_DIREG_P, REWRITE_ADDRESS, ASM_OPERAND_LETTER,
2717 RET, AT_SP): Remove.
2718 * config/i386/i386.md (*sse_prologue_save_insn): Use return
2719 instead of RET.
2720
2721 * alias.c (init_alias_analysis): Use VEC_safe_grow_cleared.
2722 * cfgbuild.c (find_basic_blocks): Likewise.
2723 * cfgrtl.c (rtl_create_basic_block): Likewise.
2724 * function.c (temp_slots_at_level): Likewise.
2725 * reg-stack.c (stack_regs_mentioned): Likewise.
2726 * regclass.c (allocate_reg_info): Likewise.
2727 * tree-cfg.c (init_empty_tree_cfg, build_tree_cfg, create_bb,
2728 set_bb_for_stmt, move_block_to_fn): Likewise.
2729 * tree-complex.c (tree_lower_complex): Likewise.
2730 * vec.h (VEC_safe_grow_cleared): New.
2731
2732 * cgraphunit.c, tree-ssa-alias.c: Fix comment typos.
2733
2734 2007-01-03 Zdenek Dvorak <dvorakz@suse.cz>
2735
2736 * loop-unswitch.c (unswitch_loop): Pass probabilities to loopify.
2737 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Pass probabilities
2738 to loop_version.
2739 * cfgloopmanip.c (scale_loop_frequencies): Export.
2740 (loopify): Scale the frequencies by prescribed coefficients.
2741 (set_zero_probability): New function.
2742 (duplicate_loop_to_header_edge): Improve updating of frequencies.
2743 (lv_adjust_loop_entry_edge, loop_version): Set probabilities
2744 and frequencies according to arguments.
2745 * tree-ssa-loop-manip.c (tree_unroll_loop): Set probabilities
2746 correctly.
2747 * cfg.c (scale_bbs_frequencies_int): Allow scaling the frequencies up.
2748 * modulo-sched.c (sms_schedule): Set probabilities for entering
2749 versioned loop correctly.
2750 * tree-vect-transform.c (vect_transform_loop): Ditto.
2751 * cfgloop.h (loopify, loop_version): Declaration changed.
2752 (scale_loop_frequencies): Declared.
2753
2754 2007-01-02 Jan Hubicka <jh@suse.cz>
2755
2756 * cgraph.c: Include tree-flow.h
2757 (cgraph_add_new-function): Handle IPA_SSA mode; execute
2758 early_local_passes.
2759 * cgraph.h (enum cgraph_state): Add CGRAPH_STATE_IPA_SSA.
2760 * tree-pass.h (pass_all_early_optimizations): Declare.
2761 * cgraphunit.c (cgraph_process_new_functions): Add IPA_SSA; execute
2762 early_local_passes.
2763 (cgraph_analyze_function): Do early_local_passes.
2764 * tree-mudflap.c (mf_decl_cache_locals, mf_build_check_statement_for):
2765 Do not add referenced vars.
2766 * tree-optimize.c (gate_all_optimizations): Do not execute when not in
2767 SSA form.
2768 (gate_all_early_local_passes): New gate.
2769 (pass_early_local_passes): Use new gate.
2770 (execute_early_local_optimizations): New functions.
2771 (gate_all_early_optimizations): New gate.
2772 (pass_all_early_optimizations): New pass.
2773 (execute_free_datastructures): Free SSA only when initialized.
2774 (gate_init_datastructures): Init only when optimizing.
2775 (tree_lowering_passes): Do early local passes when called late.
2776 * tree-profile.c (do_tree_profiling): Don't profile functions added
2777 late.
2778 (do_early_tree_profiling, pass_early_tree_profile): Kill.
2779 * tree-cfg.c (update_modified_stmts): Do not update when operands are
2780 not active.
2781 * passes.c (init_optimizations_passes): Reorder so we go into SSA
2782 during early_local_passes.
2783 * Makefile.in (cgraph.o): Add dependency on tree-flow.h.
2784
2785
2786 2007-01-02 Carlos O'Donell <carlos@codesourcery.com>
2787
2788 * Makefile.in: Update copyright year.
2789
2790 2007-01-02 Carlos O'Donell <carlos@codesourcery.com>
2791
2792 * Makefile.in: Export GCC_EXEC_PREFIX before calling $(RUNTEST)
2793 in $(lang_checks) and check-consistency targets.
2794
2795 2007-01-02 Jan Hubicka <jh@suse.cz>
2796
2797 * tree-mudflap.c (mf_decl_cache_locals, mf_build_check_statement_for):
2798 Do not add referenced vars.
2799 * tree-cfg.c (update_modified_stmts): Do not update when SSA operands
2800 are not active.
2801 * passes.c (init_optimization_passes): Put mudflap_2 after
2802 free_datastructures.
2803
2804 2007-01-02 Jan Hubicka <jh@suse.cz>
2805
2806 * tree-optimize (execute_fixup_cfg): Set after_inlining flag.
2807 Set NOTHROW flag on call statements proved to be nothrow.
2808 Update statement of local calls so new pure/const functions are
2809 updated. Update_ssa when in ssa form. Mark PHI nodes of nonlocal
2810 goto receivers.
2811 (tree_rest_of_compilation): Register hooks and initialize bitmap
2812 early. Do not set after_inlining flag.
2813
2814 2007-01-02 Steve Ellcey <sje@cup.hp.com>
2815
2816 * sbitmap.c (HOST_BITS_PER_LONG_LONG): Change to
2817 HOST_BITS_PER_LONGLONG
2818
2819 2007-01-02 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
2820
2821 PR c/19977
2822 * c-typeck.c (store_init_value): Don't emit pedantic overflow
2823 warning for non-static initializers.
2824
2825 2007-01-02 Steven Bosscher <steven@gcc.gnu.org>
2826
2827 * config/alpha/alpha.md, arm/arm.c, darwin.c, frv/frv.md,
2828 m32r/m32r.c, m32r/m32r.c, mn10300/mn10300.md, pa/pa.c,
2829 rs6000/rs6000.c, s390/s390.md, sh/sh.md, sparc/sparc.c:
2830 Always use set_unique_reg_note to add REG_EQUAL notes.
2831
2832 2007-01-02 Kazu Hirata <kazu@codesourcery.com>
2833
2834 Revert:
2835 2007-01-02 Kazu Hirata <kazu@codesourcery.com>
2836
2837 * alias.c (init_alias_analysis): Use VEC_safe_grow_cleared.
2838 * cfgbuild.c (find_basic_blocks): Likewise.
2839 * cfgrtl.c (rtl_create_basic_block): Likewise.
2840 * function.c (temp_slots_at_level): Likewise.
2841 * reg-stack.c (stack_regs_mentioned): Likewise.
2842 * regclass.c (allocate_reg_info): Likewise.
2843 * tree-cfg.c (init_empty_tree_cfg, build_tree_cfg, create_bb,
2844 set_bb_for_stmt, move_block_to_fn): Likewise.
2845 * tree-complex.c (tree_lower_complex): Likewise.
2846 * vec.h (VEC_safe_grow_cleared): New.
2847
2848 2007-01-02 Ian Lance Taylor <iant@google.com>
2849
2850 * c-common.c (c_common_truthvalue_conversion): When warning about
2851 using an assignment as a truth value, set TREE_NO_WARNING.
2852
2853 2007-01-02 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
2854
2855 PR middle-end/7651
2856 * c.opt (Wold-style-declaration): New.
2857 * doc/invoke.texi (C-only Warning Options): New.
2858 (Wold-style-declaration): Document it.
2859 (Wextra): Enabled by -Wextra.
2860 * c-opts.c (c_common_post_options): Enabled by -Wextra.
2861 * c-decl.c (declspecs_add_scspec): Replace -Wextra with
2862 -Wold-style-declaration.
2863
2864 2007-01-02 Kazu Hirata <kazu@codesourcery.com>
2865
2866 * alias.c (init_alias_analysis): Use VEC_safe_grow_cleared.
2867 * cfgbuild.c (find_basic_blocks): Likewise.
2868 * cfgrtl.c (rtl_create_basic_block): Likewise.
2869 * function.c (temp_slots_at_level): Likewise.
2870 * reg-stack.c (stack_regs_mentioned): Likewise.
2871 * regclass.c (allocate_reg_info): Likewise.
2872 * tree-cfg.c (init_empty_tree_cfg, build_tree_cfg, create_bb,
2873 set_bb_for_stmt, move_block_to_fn): Likewise.
2874 * tree-complex.c (tree_lower_complex): Likewise.
2875 * vec.h (VEC_safe_grow_cleared): New.
2876
2877 2007-01-02 Douglas Gregor <doug.gregor@gmail.com>
2878
2879 * c-common.c(c_common_nodes_and_builtins): Since variants of
2880 void_type_node get built before it is given a name, we need to
2881 give those variants the name, too.
2882 (complete_array_type): We need to work with the canonical main
2883 type of the array, from which we will build the qualified version.
2884 * params.def (PARAM_VERIFY_CANONICAL_TYPES): New.
2885 * print-tree.c (print_node): Display canonical type information
2886 for each type.
2887 * stor-layout.c (layout_type): When we don't know the
2888 alignment of a type for which we're building an array, we end up
2889 guessing wrong, so make the type require structural equality.
2890 * tree.c (make_node_stat): When we build a new type, it is its
2891 own canonical type.
2892 (build_type_attribute_qual_variant): When building an attribute
2893 variant, its canonical type is the non-attribute variant. However,
2894 if the attributes are target-dependent and they differ, we need to
2895 use structural equality checks for this type.
2896 (build_qualified_type): A qualified type is not equivalent to its
2897 unqualified variant; set the canonical type appropriately.
2898 (build_distinct_type_copy): When building a distinct type from
2899 another type, the new type is its own canonical type.
2900 (build_variant_type_copy): When building a new type variant, we
2901 assume that it is equivalent to the original type.
2902 (build_pointer_type_for_mode): When building a pointer type, also
2903 build a canonical type pointer.
2904 (build_reference_type_for_mode): When building a reference type,
2905 also build a canonical type reference.
2906 (build_index_type): When we can't hash an index type (e.g.,
2907 because its maximum value is negative), the index type requires
2908 structural equality tests.
2909 (build_array_type): Build the canonical form of an array type.
2910 (build_function_type): Function types require structural equality,
2911 because they contain default arguments, attributes, etc.
2912 (build_method_type_directly): Ditto for method types.
2913 (build_offset_type): Build the canonical offset type.
2914 (build_complex_type): Build the canonical vector type.
2915 (make_vector_type): Build the canonical vector type.
2916 * tree.h (TYPE_CANONICAL): New.
2917 (TYPE_STRUCTURAL_EQUALITY_P): New.
2918 (SET_TYPE_STRUCTURAL_EQUALITY): New.
2919 (struct tree_type): Added "canonical" field.
2920 * params.h (VERIFY_CANONICAL_TYPES): New.
2921 * doc/c-tree.texi (TYPE_CANONICAL): Document.
2922 (TYPE_STRUCTURAL_EQUALITY_P): Document.
2923 (SET_TYPE_STRUCTURAL_EQUALITY): Document.
2924 * doc/invoke.texi (verify-canonical-types): Document --param
2925 parameter for verifying canonical types.
2926
2927 2007-01-02 Joseph Myers <joseph@codesourcery.com>
2928
2929 * config.gcc (powerpc-*-eabispe*, powerpc-*-eabisimaltivec*,
2930 powerpc-*-eabisim*, powerpc-*-eabialtivec*, powerpc-*-eabi*,
2931 powerpc-*-rtems*, powerpc-wrs-vxworks, powerpc-wrs-vxworksae,
2932 powerpcle-*-eabisim*, powerpcle-*-eabi*): Add rs6000/e500.h to
2933 tm_file.
2934 * config/rs6000/e500.h: New.
2935 * config/rs6000/eabi.h (TARGET_SPE_ABI, TARGET_SPE, TARGET_E500,
2936 TARGET_ISEL, TARGET_FPRS, TARGET_E500_SINGLE, TARGET_E500_DOUBLE):
2937 Remove.
2938 * config/rs6000/linuxspe.h (TARGET_SPE_ABI, TARGET_SPE,
2939 TARGET_E500, TARGET_ISEL, TARGET_FPRS, TARGET_E500_SINGLE,
2940 TARGET_E500_DOUBLE): Remove.
2941 * config/rs6000/vxworks.h (TARGET_SPE_ABI, TARGET_SPE,
2942 TARGET_E500, TARGET_ISEL, TARGET_FPRS): Remove.
2943 * config/rs6000/rs6000.h (CHECK_E500_OPTIONS): Define.
2944 * config/rs6000/rs6000.c (rs6000_override_options): Use
2945 CHECK_E500_OPTIONS.
2946
2947 2007-01-02 Joseph Myers <joseph@codesourcery.com>
2948
2949 * config/rs6000/rs6000.c (print_operand): Check (TARGET_SPE ||
2950 TARGET_E500_DOUBLE), not TARGET_E500, for %y.
2951 (rs6000_generate_compare, rs6000_emit_sCOND, output_cbranch,
2952 rs6000_emit_cmove): Don't check TARGET_E500.
2953 * config/rs6000/rs6000.md (bunordered, bordered, sunordered,
2954 sordered): Don't check TARGET_E500.
2955
2956 2007-01-01 Eric Christopher <echristo@apple.com>
2957
2958 * config/mips/mips.c (mips_regno_mode_ok_for_base_p): Use
2959 HARD_REGISTER_NUM_P.
2960
2961 2007-01-01 Roger Sayle <roger@eyesopen.com>
2962
2963 * fold-const.c (fold_binary) <EQ_EXPR>: Fold "(X^C1) eq/ne C2" into
2964 "X eq/ne (C1^C2)". Fold "(X^Z) eq/ne (Y^Z)" as "X eq/ne Y" when Z
2965 has no side-effects. Fold "(X^C1) eq/ne (Y^C2)" as "(X^(C1^C2))
2966 eq/ne Y".
2967
2968 2007-01-01 Mike Stump <mrs@apple.com>
2969
2970 * configure.ac: Remove support for building with Apple's gcc-3.1.
2971
2972 2007-01-02 Joseph Myers <joseph@codesourcery.com>
2973
2974 PR middle-end/30311
2975 * caller-save.c (add_stored_regs): Only handle SUBREGs if inner
2976 REG is a hard register. Do not modify REG before calling
2977 subreg_nregs.
2978 * rtlanal.c (subreg_get_info): Don't assert size of XMODE is a
2979 multiple of the size of YMODE for certain lowpart cases.
2980
2981 2007-01-01 Andrew Pinski <pinskia@gmail.com>
2982
2983 PR middle-end/30253
2984 * gimplify (voidify_wrapper_expr): Update for
2985 GIMPLE_MODIFY_STMT.
2986
2987 2007-01-01 Andreas Schwab <schwab@suse.de>
2988
2989 PR target/29166
2990 * config/ia64/ia64.c (ia64_compute_frame_size): Account space for
2991 save of BR0 in extra_spill_size instead of spill_size.
2992 (ia64_expand_prologue): Save BR0 outside of the gr/br/fr spill
2993 area.
2994 (ia64_expand_epilogue): Restore BR0 from its new location.
2995
2996 2007-01-01 Andrew Pinski <pinskia@gmail.com>
2997
2998 * gimplify.c (gimplify_init_constructor <case VECTOR_TYPE>):
2999 Use a temporary variable if the left hand side is not a gimple
3000 register.
3001
3002 2007-01-01 Andrew Pinski <pinskia@gmail.com>
3003
3004 * gimplify.c (gimplify_return_expr): Make the temporary variable
3005 for the return expression, a gimple register variable.
3006
3007 2007-01-01 Jan Hubicka <jh@suse.cz>
3008
3009 * emit-rtl.c (emit_copy_of_insn_after): Do not call copy_insn_1 for
3010 INSN_LIST.
3011
3012 2007-01-01 Mike Stump <mrs@apple.com>
3013
3014 * configure.ac (HAVE_GAS_LITERAL16): Add autoconf check for
3015 .literal16.
3016 * config/darwin.c (machopic_select_rtx_section): Use
3017 HAVE_GAS_LITERAL16.
3018 (darwin_mergeable_constant_section): Likewise.
3019 * configure: Regenerate.
3020 * config.in: Regenerate.
3021
3022 2007-01-01 Jan Hubicka <jh@suse.cz>
3023 Andrew Pinski <pinskia@gmail.com>
3024
3025 * cgraphunit.c (cgraph_optimize): Call cgraph_add_new_functions
3026 before starting IPA passes.