| blob | 5a6d033e06ad565c70ce2093a93d12c1fbf24f63 |
1 2007-01-28 Zdenek Dvorak <dvorakz@suse.cz>
3 * tree-ssa-loop-unswitch.c: Include tree-inline.h.
4 (tree_unswitch_single_loop): Pass eni_size_weights to
5 tree_num_loop_insns.
6 * tree-ssa-loop-manip.c: Include tree-inline.h.
7 (can_unroll_loop_p): Pass eni_size_weights to
8 tree_num_loop_insns.
9 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p):
10 Pass eni_size_weights to estimate_num_insns.
11 * tree.h (init_inline_once): Export.
12 * toplev.c (backend_init): Call init_inline_once.
13 * cgraphunit.c (cgraph_process_new_functions,
14 cgraph_analyze_function): Pass eni_inlining_weights to
15 estimate_num_insns.
16 * ipa-inline.c (compute_inline_parameters): Ditto.
17 * tree-ssa-loop-ivcanon.c (tree_num_loop_insns): Pass weights
18 to estimate_num_insns.
19 (try_unroll_loop_completely): Pass eni_size_weights to
20 tree_num_loop_insns.
21 * tree-eh.c (decide_copy_try_finally): Pass eni_size_weights
22 ot estimate_num_insns.
23 * tree-ssa-loop-prefetch.c: Include tree-inline.h.
24 (loop_prefetch_arrays): Pass eni_time_weights to tree_num_loop_insns.
25 * tree-inline.c (eni_inlining_weights, eni_size_weights,
26 eni_time_weights): New variables.
27 (init_inline_once): Initialize them.
28 (struct eni_data): Mew.
29 (estimate_num_insns_1, estimate_num_insns): Use weights.
30 * tree-inline.h (struct eni_weights_d): New.
31 (eni_inlining_weights, eni_size_weights, eni_time_weights): Declare.
32 (estimate_num_insns): Declaration changed.
33 * cfgloop.h (tree_num_loop_insns): Declaration changed.
34 * Makefile.in (tree-ssa-loop-unswitch.o, tree-ssa-loop-prefetch.o,
35 tree-ssa-loop-manip.o): Add TREE_INLINE_H dependency.
37 2007-01-28 Zdenek Dvorak <dvorakz@suse.cz>
39 * tree-data-ref.c (conflict_fn): Assert that the number of affine
40 relations in the conflict function is valid.
42 2007-01-27 Ian Lance Taylor <iant@google.com>
44 * common.opt: Add fstrict-overflow.
45 * opts.c (decode_options): Set flag_strict_overflow if -O2.
46 * flags.h (TYPE_OVERFLOW_WRAPS): Define.
47 (TYPE_OVERFLOW_UNDEFINED): Define.
48 (TYPE_OVERFLOW_TRAPS): Define. This replaces TYPE_TRAP_SIGNED.
49 Replace all uses.
50 * tree.h (TYPE_TRAP_SIGNED): Don't define.
51 * fold-const.c (negate_expr_p): Use TYPE_OVERFLOW_UNDEFINED.
52 (fold_negate_expr): Likewise.
53 (make_range): Likewise.
54 (extract_muldiv_1): Likewise.
55 (maybe_canonicalize_comparison): Likewise.
56 (fold_comparison): Likewise.
57 (fold_binary): Likewise.
58 (tree_expr_nonnegative_p): Likewise.
59 (tree_expr_nonzero_p): Likewise.
60 * tree-vrp.c (compare_values): Likewise.
61 (extract_range_from_binary_expr): Likewise.
62 (extract_range_from_unary_expr): Likewise.
63 * tree-ssa-loop-niter.c (infer_loop_bounds_from_signedness):
64 Likewise.
65 (nowrap_type_p): Likewise.
66 * tree-scalar-evolution.c (simple_iv): Likewise.
67 * fold-const.c (negate_expr_p): Use TYPE_OVERFLOW_WRAPS.
68 (build_range_check): Likewise.
69 (extract_muldiv_1): Likewise.
70 (fold_comparison): Likewise.
71 * tree-vrp.c (vrp_int_const_binop): Likewise.
72 (extract_range_from_unary_expr): Likewise.
73 * convert.c (convert_to_integer): Likewise.
74 * fold-const.c (fold_negate_expr): Use TYPE_OVERFLOW_TRAPS.
75 (fold_comparison): Likewise.
76 (fold_binary): Likewise.
77 * optabs.c (optab_for_tree_code): Likewise.
78 * tree-vectorizer.c (vect_is_simple_reduction): Likewise.
79 * simplify-rtx.c (simplify_const_relational_operation): Check
80 flag_strict_overflow and flag_trapv.
81 (simplify_const_relational_operation): Likewise.
82 * doc/invoke.texi (Option Summary): Mention -fstrict-overflow.
83 (Optimize Options): Add -fstrict-overflow to -O2 list. Document
84 -fstrict-overflow.
86 2007-01-27 Roger Sayle <roger@eyesopen.com>
88 * tree.c (tree_fold_gcd): Delete.
89 * tree.h (tree_fold_gcd): Remove prototype.
90 * tree-data-ref.c (tree_fold_divides_p): Don't use tree_fold_gcd to
91 test whether one constant integer is a multiple of another. Instead
92 call int_const_binop with TRUNC_MOD_EXPR and test for a zero result.
93 * fold-const.c (multiple_of_p): We've determined both TOP and
94 BOTTOM are integer constants so we can call int_const_binop directly
95 instead of the more generic const_binop.
97 2007-01-27 Roger Sayle <roger@eyesopen.com>
99 * fold-const.c (size_binop): In the fast-paths for X+0, 0+X, X-0 and
100 1*X check that the constant hasn't overflowed, to preserve the
101 TREE_OVERFLOW bit.
102 (round_up): Provide an efficient implementation when rouding-up an
103 INTEGER_CST to a power-of-two.
105 2007-01-28 Ralf Wildenhues <Ralf.Wildenhues@gmx.de>
107 * doc/sourcebuild.texi: Add comma for clarity.
108 * doc/extend.texi: Fix some typos.
109 * doc/passes.texi: Likewise.
110 * doc/cppinternals.texi: Likewise.
111 * doc/c-tree.texi: Likewise.
112 * doc/tree-ssa.texi: Likewise.
113 * doc/install.texi: Likewise.
115 2007-01-27 Jan Hubicka <jh@suse.cz>
117 * tree-sra.c (sra_walk_function): Don't rely on aliases being build.
118 (pass_sra): Do not require alias information.
119 * passes.c (init_optimization_passes): Add SRA
121 2007-01-27 Steven Bosscher <steven@gcc.gnu.org>
123 * tracer.c (rest_of_handle_tracer): We already cleaned
124 up the CFG in tracer() so don't do it here again.
125 * cfgcleanup.c (rest_of_handle_jump2): Don't repeat
126 cleanup_cfg here, either. And don't call renumber_insns.
128 * cfgrtl.c (rtl_verify_flow_info_1): Don't verify that BB_END
129 and BB_HEAD are in the insn stream here. Instead make sure
130 that BB_INSN is valid on all insns. Also, do check here that
131 there are no pending branch predictions...
132 (rtl_verify_flow_info): ...instead of doing it here. Checks
133 for BB_END and BB_HEAD moved from rtl_verify_flow_info_1 to
134 here.
136 2007-01-26 Roger Sayle <roger@eyesopen.com>
138 * config/i386/i386.c (ix86_swap_binary_operands_p): New helper
139 function to simplify/factorize operand order canonicalization.
140 (ix86_fixup_binary_operands): Reorganize using the above function.
141 (ix86_binary_operator_ok): Likewise.
143 2007-01-27 Jakub Jelinek <jakub@redhat.com>
145 * genattrtab.c (struct attr_value_list, insn_code_values): Move to
146 file scope from optimize_attrs.
147 (simplify_test_exp): If insn_code_values is not NULL, use it to speed
148 up search.
149 (optimize_attrs): Clear insn_code_values after freeing it.
151 2007-01-26 Zdenek Dvorak <dvorakz@suse.cz>
153 * tree-ssa-address.c (create_mem_ref): Remove ", bsi" from
154 a parts.base assignment.
156 2007-01-26 Zdenek Dvorak <dvorakz@suse.cz>
158 * tree-data-ref.c (dump_subscript): Use dump_conflict_function.
159 (compute_subscript_distance, initialize_data_dependence_relation,
160 finalize_ddr_dependent, analyze_ziv_subscript,
161 analyze_siv_subscript_cst_affine,
162 compute_overlap_steps_for_affine_univar,
163 compute_overlap_steps_for_affine_1_2, analyze_subscript_affine_affine,
164 analyze_siv_subscript, analyze_miv_subscript,
165 analyze_overlapping_iterations, subscript_dependence_tester_1,
166 compute_self_dependence, free_dependence_relation): Work
167 with affine_fn instead of chrecs.
168 (dump_affine_function, dump_conflict_function, affine_function_equal_p,
169 common_affine_function, affine_function_base,
170 affine_function_constant_p, affine_fn_op, affine_fn_plus,
171 affine_fn_minus, affine_fn_free, conflict_fn_not_known,
172 conflict_fn_no_dependence, free_conflict_function, free_subscripts,
173 conflict_fn, affine_fn_cst, affine_fn_univar): New functions.
174 (all_chrecs_equal_p): Removed.
175 * tree-data-ref.h (affine_fn, conflict_function): New types.
176 (struct subscript): Change type of conflicting_iterations_in_a
177 and conflicting_iterations_in_b.
179 2007-01-26 Steve Ellcey <sje@cup.hp.com>
181 PR other/30182
182 * config/pa/pa.h (TARGET_HPUX_11): New.
183 * config/pa/pa-hpux11.h (TARGET_HPUX_11): New.
184 * config/pa/pa.c (pa_init_builtins): Use TARGET_HPUX_11.
186 2007-01-26 Daniel Berlin <dberlin@dberlin.org>
187 Richard Guenther <rguenther@suse.de>
189 * tree-ssa-structalias.c (solve_graph): Handle case
190 we merged the variable to another.
192 2007-01-25 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
194 * builtins.c (fold_builtin_1): Treat ccos and ccosh as 'even'
195 functions.
197 * fold-const.c (negate_mathfn_p): Treat casin, casinh, catan,
198 catanh, cproj, csin, csinh, ctan and ctanh as 'odd' functions.
200 2007-01-25 DJ Delorie <dj@redhat.com>
202 * config/m32c/m32c.c (m32c_cannot_change_mode_class): We don't
203 allow changes to modes which don't fit in those registers.
205 * reload1.c (choose_reload_regs): Check for invalid subregs before
206 computing their locations, not after.
208 2007-01-25 Geoffrey Keating <geoffk@apple.com>
210 PR 25127
211 * config/rs6000/rs6000.c (first_altivec_reg_to_save): On Darwin,
212 save Altivec registers in an eh_return function.
213 (compute_vrsave_mask): Likewise.
214 (rs6000_stack_info): Correct AIX/Darwin stack alignment computation
215 for saving Altivec registers.
216 (rs6000_emit_prologue): Don't allocate stack twice in
217 eh_return function. Correct expected value of altivec_save_offset
218 when using save_world. Describe save of R0 to stack when using
219 save_world. Describe stack pointer adjustment when using
220 save_world. Remove duplicated eh_return parameter register saving.
221 Update sp_offset variable after save_world.
222 * config/rs6000/t-darwin (LIB2FUNCS_STATIC_EXTRA): Remove
223 darwin-world.asm.
224 (LIB2FUNCS_EXTRA): Add darwin-world.asm.
225 * config/rs6000/darwin.h (SUBTARGET_OVERRIDE_OPTIONS): -m64
226 implies Altivec.
228 2007-01-25 Steve Ellcey <sje@cup.hp.com>
230 * config.gcc (ia64*-*-hpux*): Make posix threads the default.
232 2007-01-25 Steve Ellcey <sje@cup.hp.com>
234 PR other/30182
235 * config/pa/pa.c (pa_init_builtins): Set asm names for finite routines.
236 * config/ia64/ia64.c (ia64_init_builtins): Ditto.
238 2007-01-25 Richard Guenther <rguenther@suse.de>
240 * doc/invoke.texi (-Wcoverage-mismatch): Document.
241 * common.opt (-Wcoverage-mismatch): New warning option.
242 * coverage.c (get_coverage_counts): Ignore coverage mismatch
243 if -Wcoverage-mismatch is given.
245 2007-01-25 Razya Ladelsky <razya@il.ibm.com>
247 * ipa-cp.c (ipcp_insert_stage, ipcp_driver): Support for SSA.
248 (ipcp_driver): Change to static definition.
249 Add dumping of the ifunctions.
250 (constant_val_insert): Remove unused parameter. Support for SSA.
251 (ipcp_propagate_const): Support for SSA.
252 (ipcp_profile_bb_print): Print only analyzed nodes.
253 (ipcp_replace_map_create): Remove support for Fortran constant
254 for now.
255 * ipa-prop.c (ipa_method_modify_stmt,
256 ipa_callsite_compute_param): Support for SSA.
257 * ipa-prop.h (ipcp_driver): Remove declaration.
258 (IS_VALID_TREE_MAP_INDEX): Add define.
260 2007-01-24 Geoffrey Keating <geoffk@apple.com>
262 * unwind-dw2.c (execute_stack_op): Handle DW_OP_swap.
264 2007-01-24 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
266 PR middle-end/30447
267 * builtins.c (fold_builtin_cabs): Use MPFR to evaluate a
268 constant argument to cabs and do it without checking for
269 -funsafe-math-optimizations.
271 2007-01-24 Douglas Gregor <dgregor@osl.iu.edu>
273 * c-common.h (RID_FIRST_CXX0X): New.
274 (RID_LAST_CXX0X): New.
275 * c-opts.c (c_common_handle_option): -Wc++0x-compat is triggered
276 by -Wall.
277 * c.opt (Wc++0x-compat): New.
278 * doc/invoke.texi (-Wc++0x-compat): Document.
280 2007-01-24 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
282 * builtins.c (fold_builtin_carg): New.
283 (fold_builtin_1): Use it.
285 2007-01-24 Jan Hubicka <jh@suse.cz>
287 * ipa-inline.c (cgraph_decide_inlining): Initialize initial_insns.
289 * ipa-inline.c (initial_insns, max_insns): Delete.
290 (compute_max_insns): New function.
291 (cgraph_decide_inlining_of_small_function): Use it; take minimal amount
292 of insns as base for code growth.
293 (cgraph_decide_inlining): Make initial_insns local; do not compute
294 max_insns.
295 * params.def (PARAM_INLINE_UNIT_GROWTH): Set to 60.
296 * doc/invoke.texi (inline-unit-growth): Update docs.
298 2007-01-24 Jakub Jelinek <jakub@redhat.com>
300 * config/i386/i386.h (x86_cmpxchg16b): Remove const.
301 (TARGET_CMPXCHG16B): Define to x86_cmpxchg16b.
302 * config/i386/i386.c (x86_cmpxchg16b): Remove const.
303 (override_options): Add PTA_CX16 flag. Set x86_cmpxchg16b
304 for CPUs that have PTA_CX16 set.
306 PR middle-end/27416
307 * gimplify.c (omp_check_private): New function.
308 (gimplify_scan_omp_clauses): Use it for
309 firstprivate/lastprivate/reduction.
311 PR middle-end/30494
312 * gimplify.c (omp_add_variable): Don't call omp_notice_variable
313 on TYPE_SIZE_UNIT for GOVD_LOCAL VLAs.
315 PR middle-end/30421
316 * omp-low.c (lower_omp_for_lastprivate): Add dlist argument.
317 If lower_lastprivate_clauses emits some statements, append them
318 to dlist rather than body_p and to body_p append an initializer.
319 (lower_omp_for): Adjust caller.
321 2007-01-24 Steve Ellcey <sje@cup.hp.com>
323 * target.h (globalize_decl_name): New.
324 * target-def.h (TARGET_ASM_GLOBALIZE_DECL_NAME): New.
325 * output.h (default_globalize_decl_name): New.
326 * varasm.c (asm_output_bss): Use globalize_decl_name instead of
327 globalize_label.
328 (globalize_decl): Ditto.
329 (default_globalize_decl_name): New.
330 * config/ia64/ia64.c (ia64_globalize_decl_name): New.
331 (ia64_handle_version_id_attribute): New.
332 (TARGET_ASM_GLOBALIZE_DECL_NAME): New.
333 (ia64_asm_output_external): Use globalize_decl_name instead
334 of globalize_label.
335 * doc/extend.texi (version_id): New pragma.
336 * doc/tm.texi (ARGET_ASM_GLOBALIZE_DECL_NAME): New target hook.
338 2007-01-24 Andreas Krebbel <krebbel1@de.ibm.com>
340 * unwind-dw2-fde.c (get_cie_encoding): Replaced _Unwind_Word with
341 _uleb128_t and _Unwind_SWord with _sleb128_t.
342 * unwind-dw2.c (extract_cie_info, execute_stack_op, execute_cfa_program,
343 uw_frame_state_for, uw_update_context_1): Likewise.
344 * unwind-c.c (parse_lsda_header, PERSONALITY_FUNCTION): Likewise.
345 * unwind-pe.h (read_uleb128, read_sleb128,
346 read_encoded_value_with_base): Likewise.
347 * unwind-generic.h: Define _sleb128_t and _uleb128_t types.
349 2007-01-24 Richard Guenther <rguenther@suse.de>
351 * builtins.c (expand_builtin_cexpi): Get the fndecl
352 for cexp in the correct way.
354 2007-01-24 Jan Hubicka <jh@suse.cz>
356 * tree-ssa-dce.c (eliminate_unnecesary_stmts): Remove dead LHS of calls.
358 2007-01-24 Andreas Krebbel <krebbel1@de.ibm.com>
360 * c-cppbuiltin.c (builtin_define_type_sizeof): New function.
361 (c_cpp_builtins): New builtin macros: __SIZEOF_INT__, __SIZEOF_LONG__,
362 __SIZEOF_LONG_LONG__, __SIZEOF_SHORT__, __SIZEOF_POINTER__,
363 __SIZEOF_FLOAT__, __SIZEOF_DOUBLE__, __SIZEOF_LONG_DOUBLE__,
364 __SIZEOF_SIZE_T__, __SIZEOF_WCHAR_T__, __SIZEOF_WINT_T__ and
365 __SIZEOF_PTRDIFF_T__.
366 * doc/cpp.texi: Documentation for the new builtin macros added.
368 2007-01-24 Uros Bizjak <ubizjak@gmail.com>
370 * config/i386/i386.md (tanxf2, tan<mode>2, atan<mode>2, log<mode>2,
371 log10<mode>2, log2<mode>2, expxf2, exp10xf2, exp2xf2): Use op2
372 instead of operands[2] to avoid access past the end of array.
374 2007-01-24 Richard Sandiford <richard@codesourcery.com>
376 * reload1.c (emit_reload_insns): Pass the reload register
377 for a non-spill output reload through forget_old_reloads_1.
379 2007-01-23 Joseph Myers <joseph@codesourcery.com>
381 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Define
382 _SOFT_DOUBLE if doubles use software floating-point.
383 * config/rs6000/libgcc-ppc-glibc.ver: Export additional long
384 double functions if _SOFT_DOUBLE, not _SOFT_FLOAT.
385 * config/rs6000/darwin-ldouble.c: Also compile functions for
386 hard-float without FPRs. Use fmsub function for all __NO_FPRS__
387 cases. Compile extra functions if _SOFT_DOUBLE, not _SOFT_FLOAT.
388 * config/rs6000/linuxspe.h (SUBSUBTARGET_OVERRIDE_OPTIONS): Remove
389 commented-out long double override.
390 (CPP_LONGDOUBLE_DEFAULT_SPEC): Likewise.
391 * config/rs6000/eabispe.h: Likewise.
392 * config/rs6000/rs6000.c (rs6000_override_options): Don't override
393 long double for non-SPE.
394 (rs6000_handle_option): Likewise.
395 (invalid_e500_subreg): Disallow more subregs involding DImode,
396 DFmode, TImode or TFmode.
397 (rs6000_legitimate_offset_address_p): Check TFmode offsets for
398 E500 double.
399 (legitimate_lo_sum_address_p): Also check for TFmode for E500
400 double.
401 (rs6000_legitimize_address): Also handle TFmode for E500 double.
402 (rs6000_legitimize_reload_address): Also handle TFmode for E500
403 double.
404 (rs6000_legitimate_address): Also check for TFmode for E500
405 double.
406 (rs6000_emit_move): Use DFmode subregs of TFmode for E500 double.
407 (spe_build_register_parallel): Handle TFmode and TCmode.
408 (rs6000_spe_function_arg): Handle TFmode and TCmode for E500
409 double.
410 (function_arg): Handle TFmode and TCmode for E500 double.
411 (rs6000_init_libfuncs): Initialize extra libfuncs for soft double
412 in general.
413 (print_operand): Handle TFmode and TImode for %y.
414 (rs6000_generate_compare): Handle TFmode comparisons for E500
415 double.
416 (spe_func_has_64bit_regs_p): Check for TFmode for E500 double.
417 (rs6000_function_value): Handle TFmode and TCmode for E500 double.
418 (rs6000_libcall_value): Handle TFmode and TCmode for E500 double.
419 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Check for
420 TFmode for E500 double.
421 * config/rs6000/rs6000.md (FP): Allow TF for E500 double.
422 (floatsidf2): Enable for E500 double.
423 (movtf_softfloat): Use rs6000_nonimmediate_operand.
424 (extenddftf2): Change to extenddftf2_fprs.
425 (extenddftf2): Call gen_spe_extenddftf2 or gen_extenddftf2_fprs
426 depending on TARGET_E500_DOUBLE.
427 (extendsftf2): Enable for E500 double.
428 (trunctfdf2): Enable for E500 double.
429 (trunctfsf2): Change to trunctfsf2_fprs.
430 (trunctfsf2): Call gen_spe_trunctfsf2 or gen_trunctfsf2_fprs
431 depending on TARGET_E500_DOUBLE.
432 (floatsitf2): Enable for E500 double.
433 (fix_trunctfsi2): Change to fix_trunctfsi2_fprs.
434 (fix_trunctfsi2): Call gen_spe_fix_trunctfsi2 or
435 gen_fix_trunctfsi2_fprs depending on TARGET_E500_DOUBLE.
436 (negtf2): Change to negtf2_internal.
437 (negtf2): New expander.
438 (abstf2): Enable for E500 double. Call gen_spe_abstf2_tst,
439 gen_spe_abstf2_cmp or gen_abstf2_internal depending on
440 TARGET_E500_DOUBLE and flag_unsafe_math_optimizations.
441 (movdi_internal32): Use rs6000_nonimmediate_operand.
442 (unnamed splitter): Likewise.
443 * config/rs6000/spe.md (CMPTFEQ_GPR, TSTTFEQ_GPR, CMPTFGT_GPR,
444 TSTTFGT_GPR, CMPTFLT_GPR, TSTTFLT_GPR): New unspecs.
445 (SPE64TF, DITI): New mode macros.
446 (frob_df_di): Change to frob_<SPE64:mode>_<DITI:mode>; allow more
447 modes.
448 (frob_tf_ti): New.
449 (frob_<mode>_di_2): New.
450 (frob_tf_di_8_2): New.
451 (frob_di_df): Change to frob_di_<mode>; allow more modes.
452 (frob_ti_tf): New.
453 (frob_di_df_2): Change to frob_<DITI:mode>_<SPE64:mode>_2; allow
454 more modes.
455 (frob_ti_<mode>_8_2): New.
456 (frob_ti_tf_2): New.
457 (mov_si<mode>_e500_subreg0, mov_si<mode>_e500_subreg0_2,
458 mov_si<mode>_e500_subreg4, mov_si<mode>_e500_subreg4_2): Allow
459 TFmode.
460 (mov_sitf_e500_subreg8, mov_sitf_e500_subreg8_2,
461 mov_sitf_e500_subreg12, mov_sitf_e500_subreg12_2): New.
462 (spe_trunctfdf2_internal1, spe_trunctfsf2, spe_extenddftf2,
463 spe_fix_trunctfsi2, spe_fix_trunctfsi2_internal,
464 spe_negtf2_internal, spe_abstf2_cmp, spe_abstf2_tst): New.
465 (cmptfeq_gpr, tsttfeq_gpr, cmptfgt_gpr, tsttfgt_gpr, cmptflt_gpr,
466 tsttflt_gp): New.
468 2007-01-23 Ian Lance Taylor <iant@google.com>
470 * Makefile.in (OBJS-common): Reformat, alphabetize, but put
471 insn-*.o first.
472 (OBJS-archive): Reformat, alphabetize.
473 (OBJS): Change out_object_file to OBJS-md.
475 2007-01-23 Uros Bizjak <ubizjak@gmail.com>
477 * config/i386/i386.md: Use REG_P, MEM_P, CONST_INT_P, LABEL_P,
478 JUMP_P and CALL_P predicates where applicable.
479 * config/i386/i386.c: Ditto.
480 * config/i386/i386.md: Ditto.
481 * config/i386/mmx.md: Ditto.
482 * config/i386/predicates.md: Ditto.
484 2007-01-22 Andreas Schwab <schwab@suse.de>
486 * config/m68k/m68k.h: Fix comment.
488 2007-01-22 Jan Hubicka <jh@suse.cz>
490 * passes.c (init_optimization_passes): Do not rerun
491 pass_early_warn_uninitialized.
493 2007-01-22 Richard Guenther <rguenther@suse.de>
495 PR tree-optimization/30038
496 * tree-ssa-math-opts.c (maybe_record_sincos): New static helper
497 function.
498 (execute_cse_sincos_1): Likewise.
499 (execute_cse_sincos): Likewise.
500 (gate_cse_sincos): Likewise.
501 (pass_cse_sincos): New pass CSEing sin() and cos() calls using
502 the cexpi() canonicalization of sincos().
503 * tree-pass.h (pass_cse_sincos): Declare.
504 * passes.c (init_optimization_passes): New pass pas_cse_sincos.
506 2007-01-21 Eric Botcazou <ebotcazou@libertysurf.fr>
508 PR rtl-optimization/29329
509 * combine.c (replaced_rhs_insn): Rename to i2mod.
510 (replaced_rhs_value): Rename to i2mod_new_rhs.
511 (i2mod_old_rhs): New global variable.
512 (combine_instructions): Adjust for above change. Save a copy of
513 the old RHS into i2mod_old_rhs when the contents of a REG_EQUAL
514 note are substituted in the second instruction.
515 (distribute_notes) <REG_DEAD>: Adjust for above change. Do not
516 ditch the note if it pertains to the second eliminated register
517 and this register is mentioned in i2mod_old_rhs.
519 Revert:
520 2006-09-12 Eric Botcazou <ebotcazou@libertysurf.fr>
522 * combine.c (distribute_notes) <REG_DEAD>: Do not consider SETs past
523 the insn to which the note was originally attached.
525 2007-01-21 Jan Hubicka <jh@suse.cz>
527 * ipa-inline.c (inlining_mode): Comment, move up.
528 (cgraph_decide_inlining_incrementally): Do not perform inlining
529 itself; fix handling of flattening of self recursive functions.
530 (cgraph_find_cycles): Remove.
531 (cgraph_flatten_node): Remove.
532 (cgraph_decide_inlining): Use incremental inliner to handle flattening.
533 (try_inline): New function.
534 (cgraph_early_inlining): Update call of
535 cgraph_decide_inlining_incrementally. Apply inlining here.
536 (apply_inline): Update call of cgraph_decide_inlining_incrementally.
538 2007-01-21 Dirk Mueller <dmueller@suse.de>
540 PR bootstrap/30511
541 * tree-vrp.c (check_array_bounds): do not warn
542 about ADDR_EXPR's of ARRAY_REF's which are immediately
543 used in binary expressions.
545 2007-01-21 Ira Rosen <irar@il.ibm.com>
547 * tree-vectorizer.h (struct _stmt_vec_info): Add new field
548 read_write_dep and macros for its access.
549 * tree-vectorizer.c (new_stmt_vec_info): Initialize the new field.
550 * tree-vect-analyze.c (vect_analyze_data_ref_dependence): Remove
551 argument, call vect_check_interleaving for every independent pair of
552 data-refs. Mark loads that access the same memory location as a store
553 in the loop.
554 (vect_check_dependences): Remove.
555 (vect_analyze_data_ref_dependences): Remove vect_check_dependences
556 call, fix the call to vect_analyze_data_ref_dependence.
557 (vect_analyze_data_ref_access): For statements that access the same
558 data-ref, check that they are not stores; for loads, check that there
559 is no store that access the same location.
561 2007-01-20 Roger Sayle <roger@eyesopen.com>
562 Joseph Myers <joseph@codesourcery.com>
564 * doc/invoke.texi (-fdump-rtl-combine): Fix under/overfull hbox.
565 (-fdump-rtl-stack): Likewise.
566 (-fno-signed-zeros): Use @minus{} for a minus sign. Correct typo.
567 (-mcheck-zero-division, -mcpu): Fix under/overfull hbox.
568 (-mpt-fixed): Use @minus{} for minus sign.
569 (Using Precompiled Headers): Fix under/overfull hbox.
571 2007-01-20 Jan Hubicka <jh@suse.cz>
573 * tree-flow.h (struct stmt_ann_d): Move references_memory to proper
574 place within annotation.
576 2007-01-20 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
578 * pa.c (output_move_double): Change array size of xoperands to 4.
580 2007-01-20 Jan Hubicka <jh@suse.cz>
582 * tree-tailcall.c (adjust_return_value): Do not use RESULT_DECL
583 as temporary.
585 2007-01-19 Ian Lance Taylor <iant@google.com>
587 * expmed.c (expand_divmod) [TRUNC_MOD_EXPR, TRUNC_DIV_EXPR]: Cast
588 constant to unsigned HOST_WIDE_INT before negating.
590 2007-01-19 Ian Lance Taylor <iant@google.com>
592 * tree-ssa-operands.h (struct vuse_vec_d): Change num_vuse field
593 to unsigned.
594 (VUSE_VECT_ELEMENT) [ENABLE_CHECKING]: Use unsigned comparison.
595 (VUSE_ELEMENT_PTR) [ENABLE_CHECKING]: Likewise.
596 (SET_VUSE_VECT_ELEMENT) [ENABLE_CHECKING]: Likewise.
597 (SET_VUSE_ELEMENT_VAR) [ENABLE_CHECKING]: Likewise.
598 (SET_VUSE_ELEMENT_PTR) [ENABLE_CHECKING]: Likewise.
599 (realloc_vdef, realloc_vuse): Change second parameter to
600 unsigned.
601 (ssa_operand_iterator_d): Change vuse_index and mayuse_index
602 fields to unsigned.
603 * tree-ssa-operands.c (realloc_vop): Change num_elem parameter to
604 unsigned. Change x and lim locals to unsigned.
605 (realloc_vdef, realloc_vuse): Change num_elem parameter to
606 unsigned.
607 (finalize_ssa_vuse_ops): Change old_i local to unsigned.
608 (copy_virtual_operands): Change i and n locals to unsigned.
610 2007-01-19 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
612 PR middle-end/29335
613 * builtins.c (fold_builtin_1): Handle builtin fdim.
615 2007-01-20 Jan Hubicka <jh@suse.cz>
617 * tree-ssa.c (init_tree_ssa): Do not call init_alias_heapvars.
618 * tree-ssa-structalias.c (compute_points_to_sets): Do call
619 init_alias_heapvars.
620 (init_alias_heapvars): Initialize only when not already initialized.
621 (delete_alias_heapvars): Set heapvar_for_stmt to NULL.
623 2007-01-19 Roger Sayle <roger@eyesopen.com>
625 * common.opt (fsigned-zeros): New command line option.
626 * flags.h (HONOR_SIGNED_ZEROS): Control via flag_signed_zeros instead
627 of flag_unsafe_math_optimizations.
628 * opts.c (set_fast_math_flags): The -ffast-math command line option
629 implies -fno-signed-zeros.
630 (fast_math_flags_set_p): Likewise.
632 * doc/invoke.texi: Document new -fno-signed-zeros option, and update
633 the documentation of -ffast-math appropriately. Wrap long lines.
635 2007-01-19 Steve Ellcey <sje@cup.hp.com>
637 * system.h (ASM_MAKE_LABEL_LINKONCE): Poison.
638 * varasm.c (globalize_decl): Remove ASM_MAKE_LABEL_LINKONCE ifdef.
640 2007-01-19 Tomas Bily <tbily@suse.cz>
642 * cgraphunit.c (cgraph_finalize_function): Updating of pid
643 * tree-profile.c:
644 (tree_init_ic_make_global_vars): New function
645 (tree_init_edge_profiler): call of tree_init_ic_make_global_vars
646 (tree_gen_ic_profiler): New function
647 (tree_gen_ic_func_profiler): New function
648 (tree_profiling): Added calling of tree_gen_ic_func_profiler
649 (tree_profile_hooks): Added hook for indirec/virtual calls
650 * value-prof.c (tree_find_values_to_profile): New case for
651 indirect calls
652 (tree_values_to_profile): Call for determining indirect/virtual
653 counters
654 (tree_indirect_call_to_profile): New function
655 (tree_ic_transform): New function
656 (tree_ic): New function
657 (find_func_by_pid): New function
658 (init_pid_map): New function
659 (tree_value_profile_transformations): Added check for
660 indirect/virtual call transformation
661 * value-prof.h (enum hist_type): New counter type for
662 indirect/virtual calls
663 (profile_hooks): Added new hook for profiling indirect/virtual
664 calls
665 * profile.c (instrument_values): New case for indirect/virtual
666 call added
667 * gcov-io.h (GCOV_LAST_VALUE_COUNTER): Changed to 6
668 (GCOV_COUNTER_V_INDIR): New counter type
669 (GCOV_COUNTER_NAMES): New name of counter "indirect" added
670 (GCOV_MERGE_FUNCTIONS): New merge function for indirect/virtual
671 call added
672 * cgraph.c: Definition of cgraph_max_pid
673 (cgraph_create_node): Default init of pid attribute
674 * cgraph.h: Declaration of cgraph_max_pid
675 (struct cgraph_node): Added pid attribute
676 * libgcov.c (__gcov_indirect_call_profiler): New function
677 (__gcov_one_value_profiler_body): New function
678 (__gcov_one_value_profiler): Body was moved to
679 __gcov_one_value_profiler_body and calls it
681 2007-01-19 Basile Starynkevitch <basile@starynkevitch.net>
683 * doc/gty.texi (Options): Document the mark_hook option to GTY.
684 * gengtype.c (write_types_data, write_func_for_structure,
685 write_types, ggc_wtd, pch_wtd): Add skip_hooks to
686 write_types_data, ggc_wtd, pch_wtd for processing mark_hook.
687 (walk_type, write_func_for_structure): Generate the mark_hook if
688 needed.
690 2007-01-19 Jan Hubicka <jh@suse.cz>
692 * ipa-inline.c (cgraph_decide_inlining_incrementally): Instead of
693 'early' argument take inlining mode argument specifying whether to
694 inline for size/speeed or all functions; add support for flattening;
695 improve dumpting.
696 (cgraph_early_inlining): Update call of decide_inlining_incrementally.
698 2007-01-19 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
700 PR c++/17947
701 * toplev.c (warn_deprecated_use): Use %qD instead of %qs to print
702 the name of the declared identifier.
704 2007-01-19 Dirk Mueller <dmueller@suse.de>
706 * config/i386.h (CONDITIONAL_REGISTER_USAGE): Store
707 result of PIC_OFFSET_TABLE_REGNUM in temporary variable to avoid
708 duplicate evaluation.
710 2007-01-19 Uros Bizjak <ubizjak@gmail.com>
712 * config/i386/i386.md (acos<mode>2): Rename from acossf2 and acosdf2.
713 Macroize expander using X87MODEF12 mode macro. Extend operand 1
714 to XFMode, use acosxf2 and truncate result to requested mode.
715 Use SSE_FLOAT_MODE_P to disable patterns for SSE math.
716 (asin<mode>2): Similarly, with asin expanders.
717 (*fscalexf4_i387): Rename from *fscalexf4.
718 (expNcorexf3): New expander.
719 (expxf2, exp10xf2, exp2xf2): Use expNcorexf3 expander.
720 (exp<mode>2): Rename from expsf2 and expdf2. Macroize expander using
721 X87MODEF12 mode macro. Extend operand 1 to XFMode, use expxf2 and
722 truncate result to requested mode. Use SSE_FLOAT_MODE_P to disable
723 patterns for SSE math.
724 (exp10<mode>2): Similarly, with exp10 expanders.
725 (exp2<mode>2): Similarly, with exp2 expanders.
726 (expm1<mode>2): Similarly, with expm1 expanders.
727 (ldexp<mode>3): Similarly, with ldexp expanders.
728 (log<mode>2, log10<mode>2, log2<mode>2, log1p<mode>2, logb<mode>2):
729 Use gen_truncxf<mode>2_i387_noop to truncate result.
731 2007-01-19 Richard Sandiford <richard@codesourcery.com>
733 * config/m68k/m68k.h (PREFERRED_STACK_BOUNDARY): Define to 32
734 for ColdFire targets.
736 2007-01-19 Nathan Sidwell <nathan@codesourcery.com>
737 Richard Sandiford <richard@codesourcery.com>
739 * config/m68k/m68k.h (M68K_STATIC_CHAIN_REG_NAME): New macro.
740 (INITIALIZE_TRAMPOLINE): Use STATIC_CHAIN_REGNUM.
741 (__transfer_from_trampoline): Use M68K_STATIC_CHAIN_REG_NAME.
742 * config/m68k/m68kelf.h (STATIC_CHAIN_REGNUM): Override.
743 (M68K_STATIC_CHAIN_REG_NAME): Likewise.
744 * config/m68k/netbsd-elf.h (M68K_STATIC_CHAIN_REG_NAME): Likewise.
746 2007-01-19 Richard Sandiford <richard@codesourcery.com>
748 * config/m68k/m68k.md (adddi_dilshr32): Rename to...
749 (*adddi_dilshr32): ...this. Fix formatting. Remove commented-out
750 non-canonical pattern. Restrict to !TARGET_COLDFIRE.
751 (*adddi_dilshr32_cf): New pattern.
752 (adddi3, subdi3): Remove first alternatives.
754 2007-01-19 Richard Sandiford <richard@codesourcery.com>
756 * config/m68k/m68k.c (notice_update_cc): If an SFmode move is
757 implemented using move.l, do not use its cc result for floating-point
758 comparisons.
760 2007-01-19 Richard Sandiford <richard@codesourcery.com>
762 * config/m68k/m68k.h (EXTRA_CONSTRAINT): Stop the 'T' constraint
763 from accepting 's' constraints if flag_pic.
765 2007-01-19 Richard Sandiford <richard@codesourcery.com>
767 * config/m68k/m68k.md (bordered, bunordered, buneq, bunge, bungt)
768 (bunle, bunlt, bltgt, bordered_rev, bunordered_rev, buneq_rev)
769 (bunge_rev, bungt_rev, bunle_rev, bunlt_rev, bltgt_rev): Change
770 condition from TARGET_68881 to TARGET_HARD_FLOAT.
772 2007-01-19 Sandra Loosemore <sandra@codesourcery.com>
774 * longlong.h (count_leading_zeros, COUNT_LEADING_ZEROS_0): Add
775 ColdFire alternatives.
776 * config/m68k/m68k.h (CLZ_DEFINED_VALUE_AT_ZERO): New macro.
777 * config/m68k/m68k.md (clzsi2): Define for ColdFire
778 architectures that support ff1 instruction.
780 2007-01-19 Richard Sandiford <richard@codesourcery.com>
781 Julian Brown <julian@codesourcery.com>
783 * config/m68k/m68k.h (CONST_OK_FOR_LETTER_P): Add an 'R' case.
784 * config/m68k/m68k.md (*movsi_cfv4): Fold into...
785 (*movsi_cf): ...here. Remove unnecessary 'R' from 'Rg'.
786 Add commentary.
788 2007-01-19 Richard Sandiford <richard@codesourcery.com>
790 * config/m68k/m68k-protos.h (valid_mov3q_const): Take a HOST_WIDE_INT
791 and return a bool.
792 (output_move_const_into_data_reg, output_move_simode_const): Delete.
793 * config/m68k/m68k.c (const_method, const_int_cost): Take a
794 HOST_WIDE_INT instead of an rtx.
795 (m68k_rtx_costs): Update call accordingly.
796 (output_move_const_into_data_reg): Likewise. Fix formatting.
797 (valid_mov3q_const): Take a HOST_WIDE_INT instead of an rtx.
798 Return a bool.
799 (output_move_simode_const): Update calls after above changes.
800 Rework to use automatic variables and predicates like MEM_P.
801 * config/m68k/m68k.md (pushexthisi_const): Update call to
802 valid_mov3q_const.
804 2007-01-19 Dirk Mueller <dmueller@suse.de>
806 * tree-ssa-alias.c (perform_var_substitution): Fix typo
807 in dump_flags test.
809 2007-01-19 Richard Guenther <rguenther@suse.de>
811 * builtins.c (expand_builtin_cexpi): Fall back to expanding
812 via cexp in case sincos is not available.
814 2007-01-19 Richard Guenther <rguenther@suse.de>
816 * doc/tm.texi (TARGET_HAS_SINCOS): Document new target macro.
817 * defaults.h (TARGET_HAS_SINCOS): Default to off.
818 * config/linux.h (TARGET_HAS_SINCOS): Set to on if we have glibc.
819 * config/alpha/linux.h (TARGET_HAS_SINCOS): Likewise.
820 * config/sparc/linux.h (TARGET_HAS_SINCOS): Likewise.
821 * config/sparc/linux64.h (TARGET_HAS_SINCOS): Likewise.
822 * config/rs6000/linux.h (TARGET_HAS_SINCOS): Likewise.
823 * config/rs6000/linux64.h (TARGET_HAS_SINCOS): Likewise.
825 2007-01-19 Uros Bizjak <ubizjak@gmail.com>
827 * config/i386/i386.md (*fpatanxf3_i387, fpatan_extend<mode>xf3_i387):
828 New insn patterns.
829 (atan2sf3_1, atan2df3_1, atan2xf3_1): Remove insn patterns.
830 (atan2xf3): Directly generate RTL pattern.
831 (atan2<mode>3): Rename from atan2sf3 and atan2df3 and macroize insn
832 patterns using X87MODEF12 mode macro. Use fpatan_extend<mode>xf3_i387
833 and truncate result to requested mode. Use SSE_FLOAT_MODE_P to
834 disable patterns for SSE math.
835 (atan<mode>2): Rename from atansf2 and atandf2 and macroize insn
836 patterns using X87MODEF12 mode macro. Use fpatan_extend<mode>xf3_i387
837 and truncate result to requested mode. Use SSE_FLOAT_MODE_P to
838 disable patterns for SSE math.
840 2007-01-19 Alexandre Oliva <aoliva@redhat.com>
842 * libgcc-std.ver: Fix typo in %inherit for GCC_4.3.0.
844 2007-01-18 Roger Sayle <roger@eyesopen.com>
846 * fold-const.c (fold_unary) <VIEW_CONVERT_EXPR>: Optimize away a
847 VIEW_CONVERT_EXPR to the same type as it's operand.
849 2007-01-18 David Edelsohn <edelsohn@gnu.org>
851 * config/rs6000/darwin-ldouble.c: Only build _SOFT_FLOAT if
852 configured for long double 128.
854 2007-01-18 Mike Stump <mrs@apple.com>
856 * config/rs6000/rs6000.c (rs6000_emit_vector_compare): Fix build
857 error.
859 2007-01-18 Michael Meissner <michael.meissner@amd.com>
861 * i386.c (ix86_compute_frame_layout): Make fprintf's in #if 0 code
862 type correct.
864 2007-01-18 Jan Hubicka <jh@suse.cz>
866 * tree-ssa-operands.c (vop_free_bucket_size): Never return value
867 greater than NUM_VOP_FREE_BUCKETS.
869 2007-01-18 Daniel Berlin <dberlin@dberlin.org>
871 * tree-ssa-structalias.c: Update comments.
872 (ptabitmap_obstack): Removed.
873 (pta_obstack): New.
874 (oldpta_obstack): Ditto.
875 (stats): Add a few members.
876 (struct variable_info): Remove node, complex, address_taken, and
877 indirect_target members. Add oldsolution member.
878 (new_var_info): Do not initialize removed members.
879 (constraint_expr_type): Remove INCLUDES.
880 (constraint_graph): Add size, implicit_preds, rep,
881 indirect_cycles, eq_rep, label, direct_nodes, and complex members.
882 (FIRST_REF_NODE): New macro.
883 (LAST_REF_NODE): Ditto.
884 (FIRST_ADDR_NODE): Ditto.
885 (find): New function.
886 (unite): Ditto.
887 (dump_constraint): Do not handle INCLUDES.
888 (insert_into_complex): Do not insert duplicate constraints.
889 (condense_varmap_nodes): Renamed and rewritten into ...
890 (merge_node_constraints): This. Also fix bug in handling of
891 offseted copy constraints.
892 (clear_edges_for_node): No longer need to deal with preds at all,
893 or removing associated preds/succs.
894 (merge_graph_nodes): Deal with indirect_cycles.
895 Don't deal with predecessors.
896 (add_implicit_graph_edge): New function.
897 (add_pred_graph_edge): Ditto.
898 (add_graph_edge): Don't deal with predecessors.
899 (build_constraint_graph): Removed.
900 (build_pred_graph): New function.
901 (build_succ_graph): Ditto.
902 (struct scc_info): Removed in_component. Added roots, dfs, and
903 node_mapping. Remove visited_index, unification_queue.
904 (scc_visit): Deal with union-find we do now.
905 Deal with cycles with REF nodes.
906 (collapse_nodes): Renamed and rewritten to ...
907 (unify_nodes): This.
908 (process_unification_queue): Removed.
909 (topo_visit): Cleanup
910 (do_da_constraint): Use find.
911 (do_sd_constraint): Ditto.
912 (do_ds_constraint): Ditto.
913 (do_complex_constraint): Ditto.
914 (init_scc_info): Update for removed and added members.
915 (find_and_collapse_graph_cycles): Renamed and rewritten into ...
916 (find_indirect_cycles): This.
917 (equivalence_class): New variable.
918 (label_visit): New function.
919 (perform_variable_substitution): Rewritten.
920 (free_var_substitution_info): New function.
921 (find_equivalent_node): Ditto.
922 (move_complex_constraints): Ditto.
923 (eliminate_indirect_cycles): Ditto.
924 (solve_graph): Only propagate changed bits.
925 Use indirect cycle elimination.
926 Use find.
927 (tree_id_t): Rename to tree_vi_t, delete id member, add vi member.
928 (tree_id_eq): Renamed to ...
929 (tree_vi_eq): This. Update for member change
930 (insert_id_for_tree): Renamed and rewritten to ...
931 (insert_vi_for_tree): This.
932 (lookup_id_for_tree): Renamed and rewritten to ...
933 (lookup_vi_for_tree): This.
934 (get_id_for_tree): Renamed and rewritten to ...
935 (get_vi_for_tree): Ditto.
936 (get_constraint_exp_from_ssa_var): Update to use get_vi_for_tree.
937 (process_constraint): Don't handle INCLUDES.
938 Remove special ADDRESSOF case.
939 (find_func_aliases): Rewrite to use vi functions instead of id
940 ones.
941 (create_function_info_for): Ditto.
942 (create_variable_info_for): Ditto.
943 (intra_create_variable_infos): Ditto.
944 (merge_smts_into): Ditto.
945 (find_what_p_points_to): Ditto.
946 (init_base_vars): Ditto.
947 (init_alias_vars): Ditto.
948 (remove_preds_and_fake_succs): New function.
949 (dump_sa_points_to_info): Dump new stats.
950 (dump_solution_for_var): Use find.
951 (set_used_smts): Fix formatting.
952 (compute_points_to_sets): Updated for new functions.
953 (ipa_pta_execute): Ditto.
955 2007-01-18 Kazu Hirata <kazu@codesourcery.com>
956 Richard Sandiford <richard@codesourcery.com>
958 * doc/tm.texi (TARGET_FUNCTION_VALUE): Expand documentation of
959 parallels.
960 * calls.c (expand_call): If the return value is a PARALLEL,
961 extract its first member.
962 * config/m68k/linux.h (FUNCTION_EXTRA_EPILOGUE): Remove.
963 * config/m68k/m68k.c (m68k_output_function_epilogue): Don't
964 use FUNCTION_EXTRA_EPILOGUE.
965 (m68k_function_value): Return a PARALLEL if the return value
966 is of a pointer type.
967 * config/m68k/netbsd-elf.h (current_function_returns_pointer)
968 (FUNCTION_EXTRA_EPILOGUE): Remove.
969 * config/m68k/m68k.md (D0_REG): New constant.
971 2007-01-18 Kazu Hirata <kazu@codesourcery.com>
973 * config/m68k/m68k.c (m68k_output_function_epilogue): Don't
974 output a NOP for empty epilogues.
976 2007-01-18 Richard Sandiford <richard@codesourcery.com>
978 * config/m68k/m68k.c (m68k_use_return_insn): Update comments
979 before function. Extend register save check to include all
980 registers, not just integer ones.
982 2007-01-18 Kazu Hirata <kazu@codesourcery.com>
984 * config/m68k/m68k-protos.h (use_return_insn): Rename to...
985 (m68k_use_return_insn): ...this.
986 * config/m68k/m68k.h (USE_RETURN_INSN): Delete.
987 * config/m68k/m68k.c (use_return_insn): Rename to...
988 (m68k_use_return_insn): ...this.
989 * config/m68k/m68k.md (return): Use m68k_use_return_insn instead
990 of USE_RETURN_INSN.
992 2007-01-18 Nathan Sidwell <nathan@codesourcery.com>
994 * config/m68k/fpgnulib.c (__truncdfsf2): Implement round to
995 nearest even, fix denormal rounding overflow.
997 2007-01-18 Richard Sandiford <richard@codesourcery.com>
999 * config/m68k/m68k.md (movsf_cf_hard): Use fsmove instead of
1000 f%$smove and f%$move.
1001 (movdf_cf_hard): Use fdmove for cases 0 and 3 and fmove for case 1.
1002 (extendsfdf2_cf): Use fdmove instead of f%&move.
1003 (truncdfsf2_cf): Use fsmove instead of f%$smove.
1004 (add<mode>3_cf, sub<mode>3_cf): Use <FP:prec> instead of <FP:round>.
1006 2007-01-18 Richard Sandiford <richard@codesourcery.com>
1008 * config/m68k/m68k.md (movdf_cf_hard): Use output_move_double for
1009 GPR<-GPR moves.
1011 2007-01-18 Richard Sandiford <richard@codesourcery.com>
1013 * real.h (real_format): Add a canonical_nan_lsbs_set field.
1014 (coldfire_single_format): Declare.
1015 (coldfire_double_format): Likewise.
1016 * real.c (encode_ieee_single): Use canonical_nan_lsbs_set instead
1017 of qnan_msb_set to determine the lower bits of a canonical
1018 NaN significand.
1019 (encode_ieee_double): Likewise.
1020 (encode_ieee_quad): Likewise.
1021 (ieee_single_format): Initialize canonical_nan_lsbs_set.
1022 (mips_single_format): Likewise.
1023 (ieee_double_format): Likewise.
1024 (mips_double_format): Likewise.
1025 (ieee_extended_motorola_format): Likewise.
1026 (ieee_extended_intel_96_format): Likewise.
1027 (ieee_extended_intel_128_format): Likewise.
1028 (ieee_extended_intel_96_round_53_format): Likewise.
1029 (ibm_extended_format): Likewise.
1030 (mips_extended_format): Likewise.
1031 (ieee_quad_format): Likewise.
1032 (mips_quad_format): Likewise.
1033 (vax_f_format): Likewise.
1034 (vax_d_format): Likewise.
1035 (vax_g_format): Likewise.
1036 (i370_single_format): Likewise.
1037 (i370_double_format): Likewise.
1038 (decimal_single_format): Likewise.
1039 (decimal_double_format): Likewise.
1040 (decimal_quad_format): Likewise.
1041 (c4x_single_format): Likewise.
1042 (c4x_extended_format): Likewise.
1043 (real_internal_format): Likewise.
1044 (coldfire_single_format): New real_format.
1045 (coldfire_double_format): Likewise.
1046 * config/pdp11/pdp11.c (pdp11_f_format): Initialize
1047 canonical_nan_lsbs_set.
1048 (pdp11_d_format): Likewise.
1049 * config/m68k/m68k.c (override_options): Override REAL_FORMAT_MODE
1050 if TARGET_COLDFIRE_CPU.
1052 2007-01-18 Richard Sandiford <richard@codesourcery.com>
1054 * config/m68k/m68k-protos.h (m68k_output_pic_call): Delete.
1055 (output_call, m68k_legitimize_call_address): Declare.
1056 * config/m68k/m68k.h (EXTRA_CONSTRAINT): Remove unnecessary
1057 parenthesees. Add support for a 'W' constraint.
1058 (LEGITIMATE_PIC_OPERAND_P): Remove SYMBOL_REF_FLAG handling.
1059 (PRINT_OPERAND_PUNCT_VALID_P): Remove comment about 'o'.
1060 (m68k_symbolic_call, m68k_symbolic_jump): Declare.
1061 * config/m68k/m68k.c (m68k_symbolic_call, m68k_symbolic_jump): New
1062 variables.
1063 (override_options): Initialize them. Do not set flag_no_function_cse
1064 for TARGET_ID_SHARED_LIBRARY.
1065 (m68k_output_pic_call): Delete.
1066 (m68k_legitimize_call_address): New function.
1067 (print_operand): Remove the %o prefix. Handle the %p prefix.
1068 (output_call): New function.
1069 (m68k_output_mi_thunk): Use m68k_symbolic_jump. Always load the
1070 target address from the GOT if symbolic jumps are not allowed.
1071 * config/m68k/m68k.md (call, general_operand): Do not set
1072 SYMBOL_REF_FLAG. Use m68k_legitimize_call_address instead.
1073 Merge separate flag_pic and !flag_pic define_insns into...
1074 (*call, *call_value): ...these new patterns. Match the address
1075 rather than the containing MEM and require it to be a call_operand.
1076 Use output_call to generate the asm template.
1077 * config/m68k/predicates.md (const_call_operand): New predicate.
1078 (call_operand): Likewise.
1080 2007-01-18 Nathan Sidwell <nathan@codesourcery.com>
1082 * config/m68k/m68k.h (REGISTER_MOVE_COST): Simplify definition.
1083 (STACK_GROWS_DOWNWARD): Define to 1.
1084 (FUNCTION_VALUE, LIBCALL_VALUE, FUNCTION_VALUE_REGNO_P): Equivocate
1085 comments, emphasizing that these values are only defaults.
1086 * config/m68k/linux.h (LINK_SPEC): Fix formatting in #undef.
1087 * config/m68k/m68k.c (const_method): Remove trailing whitespace.
1089 2007-01-18 Richard Sandiford <richard@codesourcery.com>
1091 * config/m68k/m68k.md (cmpsi): Remove outdated flag_pic handling.
1093 2007-01-18 Kazu Hirata <kazu@codesourcery.com>
1094 Richard Sandiford <richard@codesourcery.com>
1096 * config/m68k/m68k.h (DATA_REGNO_P, ADDRESS_REGNO_P, INT_REGNO_P)
1097 (FP_REGNO_P): New macros.
1098 (REGNO_OK_FOR_INDEX_P, REGNO_OK_FOR_BASE_P, REGNO_OK_FOR_DATA_P)
1099 (REGNO_OK_FOR_FP_P, REG_OK_FOR_INDEX_P, REG_OK_FOR_BASE_P): Use them.
1100 (EH_RETURN_STACKADJ_RTX): Use A0_REG.
1101 * config/m68k/m68k.c (m68k_regno_mode_ok): Use the new REGNO macros.
1103 2007-01-18 Nathan Sidwell <nathan@codesourcery.com>
1105 * config.gcc (m68k-*-aout*, m68k-*-coff*, m68020-*-elf*, m68k-*-elf*)
1106 (m68k-*-uclinuxoldabi, m68k-*-uclinux*, m68k-*-rtems*): Add t-floatlib
1107 to $tmake_file.
1108 * config/m68k/t-floatlib: New file, extracting common code from...
1109 * config/m68k/t-m68kbare, config/m68k/t-m68kelf,
1110 * config/m68k/t-uclinux: Here.
1111 * config/m68k/fpgnulib.c: Do not compile extendeed precision
1112 routines on ColdFire targets.
1114 2007-01-18 Nathan Sidwell <nathan@codesourcery.com>
1116 * config.gcc (m68k-*-aout*, m68k-*-coff*, m68020-*-elf*, m68k-*-elf*)
1117 (m68010-*-netbsdelf*, m68k*-*-netbsdelf*, m68k*-*-openbsd*)
1118 (m68k-*-uclinuxoldabi, m68k-*-uclinux*, m68k-*-linux*)
1119 (m68k-*-rtems*): Use tm_file rather than m68k/m68k.h and
1120 explicitly set MOTOROLA to 1.
1121 * config/m68k/m68k.h (MOTOROLA): Simplify definition accordingly.
1123 2007-01-18 Richard Sandiford <richard@codesourcery.com>
1124 Nathan Sidwell <nathan@codesourcery.com>
1126 * config/m68k/m68k.h (PCC_STATIC_STRUCT_RETURN): Delete.
1127 (ASM_OUTPUT_REG_PUSH, ASM_OUTPUT_REG_POP): Add MOTOROLA cases.
1128 * config/m68k/coff.h (REGISTER_PREFIX_MD): Delete.
1129 * config/m68k/m68020-elf.h (LIB_SPEC): Delete.
1130 * config/m68k/m68k-none.h (CC1_SPEC, CPP_SUBTARGET_SPEC): Delete.
1131 * config/m68k/m68kelf.h (IMMEDIATE_PREFIX, REGISTER_PREFIX_MD)
1132 (ASM_OUTPUT_REG_PUSH, ASM_OUTPUT_REG_POP): Delete.
1133 (NO_DOLLAR_IN_LABEL, PCC_STATIC_STRUCT_RETURN): Don't undefine.
1134 (BSS_ASM_OP, ASM_OUTPUT_SKIP, ASM_OUTPUT_ASCII): Delete.
1135 * config/m68k/m68kemb.h (PCC_STATIC_STRUCT_RETURN): Don't undefine.
1136 * config/m68k/linux.h (SIZE_TYPE, PTRDIFF_TYPE, WCHAR_TYPE)
1137 (WCHAR_TYPE_SIZE, TARGET_OBJFMT_CPP_BUILTINS): Delete.
1138 (TARGET_OS_CPP_BUILTINS): Don't define mc68000 and mc68020 here.
1139 (DBX_REGISTER_NUMBER): Delete.
1140 * config/m68k/netbsd-elf.h (IMMEDIATE_PREFIX): Delete.
1141 (PCC_STATIC_STRUCT_RETURN): Don't undefine.
1142 * config/m68k/openbsd.h (PCC_STATIC_STRUCT_RETURN): Define.
1144 2007-01-18 Richard Sandiford <richard@codesourcery.com>
1146 * config.gcc (m68k-*-uclinux*): Add flat.h to $tm_file.
1147 * config/flat.h: New file.
1148 * crtstuff.c (USE_PT_GNU_EH_FRAME): Don't define if
1149 OBJECT_FORMAT_FLAT.
1150 * config/m68k/m68k.h (ASM_PREFERRED_EH_DATA_FORMAT): Do not use
1151 indirect references for -msep-data or -mid-shared-library.
1152 Do not use PC-relative code addresses either.
1153 * config/m68k/m68k.c (override_options): Restrict -fPIC error
1154 to -mpcrel.
1155 * config/m68k/uclinux.h (STARTFILE_SPEC): Define. Use Scrt1.o
1156 for shared libraries and crt1.o for executables. Use crti.o and
1157 crtbegin.o.
1158 (ENDFILE_SPEC): Use crtend.o and crtn.o.
1159 (LIB_SPEC): Suppress -Rlibc.gdb if -static-libc is given.
1160 Do not add -elf2flt or -shared-lib-id options here.
1161 (LINK_SPEC): Define. Pass -elf2flt if no -elf2flt option is given.
1162 Pass -shared-lib-id if -mid-shared-library, taking the library
1163 identifier from -mshared-library-id if given, otherwise
1164 defaulting to 0.
1165 (EH_FRAME_IN_DATA_SECTION): Do not undefine.
1166 (INIT_SECTION_ASM_OP, FINI_SECTION_ASM_OP): Likewise.
1167 (TARGET_OS_CPP_BUILTINS): Define __GXX_MERGED_TYPEINFO_NAMES=0
1168 and __GXX_TYPEINFO_EQUALITY_INLINE=0 if -mid-shared-library.
1169 (DRIVER_SELF_SPECS): Map unadorned PIC options to -msep-data.
1170 * config/m68k/t-uclinux (EXTRA_MULTILIB_PARTS): Add crtbegin.o
1171 and crtend.o.
1172 * config/m68k/lb1sf68.asm (PICCALL): Use an lea and pc-relative
1173 jump sequence for ISA A and ISA A+.
1174 (PICJUMP): Likewise.
1176 2007-01-18 Richard Sandiford <richard@codesourcery.com>
1178 * config.gcc (m68k-*-uclinux*): Base the port on the common
1179 and m68k GNU/Linux files rather than on the generic ELF ones.
1180 * config/m68k/uclinux.h (TARGET_VERSION): Override.
1181 (TARGET_OS_CPP_BUILTINS): Use LINUX_TARGET_OS_CPP_BUILTINS.
1183 2007-01-18 Julian Brown <julian@codesourcery.com>
1184 Richard Sandiford <richard@codesourcery.com>
1186 * config/m68k/m68k.h (LONG_DOUBLE_TYPE_SIZE): Make 64-bit on ColdFire.
1187 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
1188 * config/m68k/netbsd-elf.h (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Undefine
1189 before redefining.
1190 * config/m68k/uclinux-oldabi.h (LONG_DOUBLE_TYPE_SIZE): Redefine to
1191 80 unconditionally.
1192 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
1194 2007-01-18 Richard Sandiford <richard@codesourcery.com>
1196 * doc/install.texi: Document m68k-uclinuxoldabi.
1197 * config.gcc (m68k-*-uclinuxoldabi): New configuration.
1198 * config/m68k/uclinux-oldabi.h: New file, copied from
1199 config/m68k/uclinux.h.
1201 2007-01-18 Kaz Kojima <kkojima@rr.iij4u.or.jp>
1203 * config/m32r/m32r-protos.h (m32r_expand_epilogue): Declare it.
1204 * config/m32r/m32r.c (m32r_setup_incoming_varargs): Use gen_frame_mem.
1205 (m32r_compute_frame_size): Use unsigned for regno.
1206 (m32r_reload_lr): Use gen_frame_mem.
1207 (pop): New.
1208 (m32r_output_function_epilogue): Don't output the function epilogue
1209 textually here.
1210 (m32r_expand_epilogue): New.
1211 (direct_return): Return false if the function has the interrupt
1212 attribute.
1213 (m32r_hard_regno_rename_ok): Remove code for the textual epilogue.
1214 * config/m32r/m32r.md (epilogue): New expander.
1215 (return_lr, return_rte): New insns.
1216 (return): Make it expander.
1217 (return_normal): New expander.
1219 2007-01-18 Josh Conner <jconner@apple.com>
1221 PR target/30485
1222 * config/rs6000/rs6000.c (rs6000_emit_vector_compare): Add
1223 support for UNLE, UNLT, UNGE, and UNGT.
1225 2007-01-18 Jan Hubicka <jh@suse.cz>
1227 * tree-vrp.c (finalize_jump_threads): Do not call cleanup_cfg by hand.
1228 * tree-tailcall (add_virtual_phis): Likewise.
1229 (optimize_tail_call): Return TODOs.
1230 (execute_tail_calls): Return TODOs.
1231 * tree-ssa-ccp (execute_fold_all_builtins): Do cleanup_cfg via TODO.
1232 * tree-cfgcleanup.c (cleanup_tree_cfg_loop): Return if something
1233 changed.
1234 * tree-ssa-forwprop.c (tree_ssa_forward_propagate_single_use_value):
1235 Cleanup cfg using TODO.
1236 * tree-flow.h (cleanup_tree_cfg_loop): Update prototype.
1237 * passes.c (execute_function_todo): When cleanup did something, remove
1238 unused locals.
1239 * tree-cfg.c (pass_build_cfg): Add cleanup_cfg TODO.
1240 (make_edges): Don't cleanup_cfg.
1242 2007-01-18 Uros Bizjak <ubizjak@gmail.com>
1244 * reg-stack.c (subst_stack_regs_pat) [UNSPEC_SINCOS_COS,
1245 UNSPEC_XTRACT_FRACT]: Use generic code for instructions that
1246 operate on the top of stack.
1247 [UNSPEC_SINCOS_SIN, UNSPEC_XTRACT_EXP, UNSPEC_TAN]: Rewrite
1248 register handling of instructions that output to the second
1249 stack slot.
1250 [UNSPEC_TAN_ONE, UNSPEC_TAN_TAN]: Remove.
1251 (move_for_stack_reg): Special-case check for dead destination
1252 stack slot for constant load of 1.0 inside UNSPEC_TAN.
1254 * config/i386/i386.md (UNSPEC_TAN): New constant.
1255 (UNSPEC_TAN_ONE, UNSPEC_TAN_TAN): Remove.
1256 (fptanxf4_i387, fptan_extend<mode>xf4_i387): New patterns
1257 to correctly model move of constant 1.0 to top stack slot.
1258 (*tandf3_1, *tansf3_1, *tanxf3_1): Remove insn patterns.
1259 (unnamed peephole2 pattern): Remove corresponding peephole2
1260 pattern that optimizes tan insn and loading of constant 1.0.
1261 (tanxf2): Use fptanxf4_i387.
1262 (tan<mode>2): Rename from tansf2 and tandf2 and macroize insn
1263 patterns using X87MODEF12 mode macro. Use fptan_extend<mode>xf4_i387
1264 and truncate result to requested mode. Use SSE_FLOAT_MODE_P to
1265 disable patterns for SSE math.
1266 (sincos<mode>3): Use truncxf<mode>2_i387_noop for truncation.
1267 (fyl2x_extend<mode>xf3_i387): Use X87MODEF12 for operand 1.
1269 2007-01-18 Dirk Mueller <dmueller@suse.de>
1270 Richard Guenther <rguenther@suse.de>
1272 PR diagnostic/8268
1273 * doc/invoke.texi (Warray-bounds): Document -Warray-bounds.
1274 * common.opt (Warray-bounds): Add new warning option.
1275 * c-opts.c (c_common_handle_option): Define -Warray-bounds
1276 if -Wall is given.
1277 * Makefile.in: make tree-vrp.o depend on toplev.h
1278 * tree-vrp.c (vrp_finalize): Call check_array_refs if -Warray-bounds
1279 is enabled.
1280 (check_array_refs, check_array_bounds, check_array_ref): New.
1282 2007-01-18 Jan Hubicka <jh@suse.cz>
1284 * tree-ssa-ccp.c (ccp_finalize): Return if something changed.
1285 (execute_ssa_ccp): Return flags conditionally.
1286 * tree-ssa-propagate.c (substitue_and_fold): Return if something was
1287 changed.
1288 * tree-ssa-propagate.h (substitute_and_fold): Update prototype.
1290 2007-01-18 Steven Bosscher <steven@gcc.gnu.org>
1292 * cfgcleanup.c (cleanup_cfg): Detect cfglayout mode and set
1293 the CLEANUP_CFGLAYOUT flag when in cfglayout mode.
1295 * Makefile.c (GTFILES): Add cfglayout.h.
1296 * gengtype.c (open_base_files): Likewise.
1297 * cfglayout.c (cfg_layout_function_footer,
1298 cfg_layout_function_header) Reindent to make gengtype happy.
1299 * cfglayout.h (cfg_layout_function_footer,
1300 cfg_layout_function_header): Add GTY(()) marker.
1302 * ifcvt.c (noce_try_sign_mask): Make sure INSN_B is non-null.
1304 2007-01-18 Ben Elliston <bje@au.ibm.com>
1306 * genautomata.c (write_automata): Include xstrerror output in the
1307 error message if writing the DFA description file fails.
1309 2007-01-17 H.J. Lu <hongjiu.lu@intel.com>
1311 * config/mips/mips-protos.h (mips_output_external): Make it
1312 return void.
1313 * config/mips/iris.h (TARGET_ASM_EXTERNAL_LIBCALL): Removed.
1314 * config/mips/mips.c (irix_output_external_libcall): Likewise.
1315 (extern_list): Likewise.
1316 (extern_head): Likewise.
1317 (TARGET_ASM_FILE_END): Likewise.
1318 (mips_file_end): Likewise.
1319 (mips_output_external): Rewritten.
1321 2007-01-18 Ben Elliston <bje@au.ibm.com>
1323 * genpreds.c (write_insn_preds_c): Only write out the function
1324 body for regclass_for_constraint if we have register constraints.
1326 2007-01-17 Tom Tromey <tromey@redhat.com>
1328 * doc/sourcebuild.texi (libgcj Tests): Use sourceware.org.
1329 * doc/install.texi (Testing): Use sourceware.org.
1330 (Binaries): Likewise.
1331 (Specific): Likewise.
1332 * doc/contrib.texi (Contributors): Use sourceware.org.
1334 2007-01-17 Anatoly Sokolov <aesok@post.ru>
1336 * config/avr/avr.h (AVR_HAVE_LPMX): New macro.
1337 (AVR_ENHANCED): Rename to ...
1338 (AVR_HAVE_MUL): ... new.
1339 (avr_enhanced_p): Rename to ...
1340 (avr_have_mul_p): ... new.
1341 (TARGET_CPU_CPP_BUILTINS): Use 'avr_have_mul_p' instead of
1342 'avr_enhanced_p' for "__AVR_ENHANCED__". Define "__AVR_HAVE_MUL__".
1343 * config/avr/avr.c (avr_enhanced_p): Rename to ...
1344 (avr_have_mul_p): ... new.
1345 (base_arch_s): Rename 'enhanced' to 'have_mul'.
1346 (avr_override_options): Use 'avr_have_mul_p' and 'have_mul' instead of
1347 'avr_enhanced_p' and 'enhanced'.
1348 (ashlhi3_out, ashrhi3_out, lshrhi3_out, avr_rtx_costs): Use
1349 AVR_HAVE_MUL instead of AVR_ENHANCED.
1350 * avr.md (*tablejump_enh): Use AVR_HAVE_LPMX instead of AVR_ENHANCED.
1351 (mulqi3, *mulqi3_enh, *mulqi3_call, mulqihi3, umulqihi3, mulhi3,
1352 *mulhi3_enh, *mulhi3_call, mulsi3, *mulsi3_call): Use AVR_HAVE_MUL
1353 instead of AVR_ENHANCED.
1354 (*tablejump_enh): Use AVR_HAVE_LPMX instead of AVR_ENHANCED.
1355 * libgcc.S: Use __AVR_HAVE_MUL__ instead of __AVR_ENHANCED__.
1356 (__tablejump__): Use __AVR_HAVE_LPMX__ instead of __AVR_ENHANCED__.
1358 2007-01-17 Ian Lance Taylor <iant@google.com>
1360 * vec.h (VEC_reserve_exact): Define.
1361 (vec_gc_p_reserve_exact): Declare.
1362 (vec_gc_o_reserve_exact): Declare.
1363 (vec_heap_p_reserve_exact): Declare.
1364 (vec_heap_o_reserve_exact): Declare.
1365 (VEC_OP (T,A,reserve_exact)): New static inline function, three
1366 versions.
1367 (VEC_OP (T,A,reserve)) [all versions]: Remove handling of
1368 negative parameter.
1369 (VEC_OP (T,A,alloc)) [all versions]: Call ...reserve_exact.
1370 (VEC_OP (T,A,copy)) [all versions]: Likewise.
1371 (VEC_OP (T,a,safe_grow)) [all versions]: Likewise.
1372 * vec.c (calculate_allocation): Add exact parameter. Change all
1373 callers.
1374 (vec_gc_o_reserve_1): New static function, from vec_gc_o_reserve.
1375 (vec_gc_p_reserve, vec_gc_o_reserve): Call vec_gc_o_reserve_1.
1376 (vec_gc_p_reserve_exact, vec_gc_o_reserve_exact): New functions.
1377 (vec_heap_o_reserve_1): New static function, from vec_heap_o_reserve.
1378 (vec_heap_p_reserve, vec_heap_o_reserve): Call vec_heap_o_reserve_1.
1379 (vec_heap_p_reserve_exact): New function.
1380 (vec_heap_o_reserve_exact): New function.
1382 2007-01-17 Jan Hubicka <jh@suse.cz>
1384 * ipa-type-escape.c (look_for_casts): Revamp using handled_component_p.
1386 2007-01-17 Eric Christopher <echristo@apple.com>
1388 * config.gcc: Support core2 processor.
1390 2007-01-16 Jan Hubicka <jh@suse.cz>
1392 * tree-ssanames.c (release_dead_ssa_names): Instead of ggc_freeing
1393 the names, just unlink the chain so we don't crash on dangling pointers
1394 to dead SSA names.
1396 2007-01-16 Jan Hubicka <jh@suse.cz>
1398 * cgraph.h (cgraph_decide_inlining_incrementally): Kill.
1399 * tree-pass.h: Reorder to make IPA passes appear toegher.
1400 (pass_early_inline, pass_inline_parameters, pass_apply_inline): Declare.
1401 * cgraphunit.c (cgraph_finalize_function): Do not compute inling
1402 parameters, do not call early inliner.
1403 * ipa-inline.c: Update comments. Include tree-flow.h
1404 (cgraph_decide_inlining): Do not compute inlining parameters.
1405 (cgraph_decide_inlining_incrementally): Return TODOs; assume to
1406 be called with function context set up.
1407 (pass_ipa_inline): Remove unreachable functions before pass.
1408 (cgraph_early_inlining): Simplify assuming to be called from the
1409 PM as local pass.
1410 (pass_early_inline): New pass.
1411 (cgraph_gate_ipa_early_inlining): New gate.
1412 (pass_ipa_early_inline): Turn into simple wrapper.
1413 (compute_inline_parameters): New function.
1414 (gate_inline_passes): New gate.
1415 (pass_inline_parameters): New pass.
1416 (apply_inline): Move here from tree-optimize.c
1417 (pass_apply_inline): New pass.
1418 * ipa.c (cgraph_remove_unreachable_nodes): Verify cgraph after
1419 transforming.
1420 * tree-inline.c (optimize_inline_calls): Return TODOs rather than
1421 doing them by hand.
1422 (tree_function_versioning): Do not allocate dummy struct function.
1423 * tree-inline.h (optimize_inline_calls): Update prototype.
1424 * tree-optimize.c (execute_fixup_cfg): Export.
1425 (pass_fixup_cfg): Remove
1426 (tree_rest_of_compilation): Do not apply inlines.
1427 * tree-flow.h (execute_fixup_cfg): Declare.
1428 * Makefile.in (gt-passes.c): New.
1429 * passes.c: Include gt-passes.h
1430 (init_optimization_passes): New passes.
1431 (nnodes, order): New static vars.
1432 (do_per_function_toporder): New function.
1433 (execute_one_pass): Dump current pass here.
1434 (execute_ipa_pass_list): Don't dump current pass here.
1436 2007-01-16 Janis Johnson <janis187@us.ibm.com>
1438 * config/dfp-bit.c (dfp_compare_op): Return separate value for NaN.
1439 (DFP_NE, DFP_LE, DFP_GE): Return false for NaN.
1441 2007-01-16 David Edelsohn <edelsohn@gnu.org>
1443 * config/rs6000/darwin-ldouble.c: Build file for SOFT_FLOAT.
1444 (strong_alias): Define.
1445 (__gcc_qmul): Provide non-FMA for soft-float.
1446 (__gcc_qdiv): Same.
1447 (__gcc_qneg): New.
1448 (__gcc_qeq): New.
1449 (__gcc_qle): New.
1450 (__gcc_qge): New.
1451 (__gcc_qunord): New.
1452 (__gcc_stoq): New.
1453 (__gcc_dtoq): New.
1454 (__gcc_qtos): New.
1455 (__gcc_qtod): New.
1456 (__gcc_qtoi): New.
1457 (__gcc_qtou): New.
1458 (__gcc_itoq): New.
1459 (__gcc_utoq): New.
1460 (fmsub): New.
1461 * config/rs6000/rs6000.c (rs6000_init_libfuncs): Initialize
1462 soft-float functions.
1463 * config/rs6000/libgcc-ppc-glibc.ver: Version soft-float symbols.
1464 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Do not warn
1465 about long double soft float.
1467 2007-01-16 Dorit Nuzman <dorit@il.ibm.com>
1468 Tehila Meyzels <tehila@il.ibm.com>
1470 * tree-vectorizer.h (is_pattern_stmt_p): New.
1471 * tree-vect-analyze.c (vect_determine_vectorization_factor): Fix
1472 formatting (tabs instead of spaces). Cleanup and clarify setting
1473 of STMT_VINFO_VECTYPE. Call is_pattern_stmt_p.
1474 * tree-vect-transform.c (vect_get_vec_def_for_operand): Fix typo.
1475 (vectorizable_type_demotion): Check that types are integral.
1476 (vectorizable_type_promotion): Likewise.
1477 (vectorizable_store): Fix typo. Eliminate new-line at end of
1478 comments.
1480 2007-01-16 Jan Hubicka <jh@suse.cz>
1482 * tree-ssanames.c (release_dead_ssa_names): Remove invalidated
1483 cgraph edges too.
1485 2007-01-15 Eric Christopher <echristo@apple.com>
1487 * ifcvt.c: Include vec.h, vecprim.h.
1488 (check_cond_move_block): New argument regs.
1489 Reorganize. Add registers used to regs.
1490 (cond_move_process_if_block): Use regs set above as
1491 loop bounds.
1493 2007-01-15 Eric Christopher <echristo@apple.com>
1495 * config/darwin.h: Update copyright.
1496 (TARGET_OPTION_TRANSLATE_TABLE): Add umbrella.
1497 (LINK_COMMAND_SPEC): Add -u.
1498 (LINK_SPEC): Fix umbrella for above.
1500 2007-01-15 Joseph S. Myers <joseph@codesourcery.com>
1502 * config/soft-fp/op-common.h, config/soft-fp/op-4.h: Update from
1503 glibc CVS.
1505 2007-01-15 Tom Tromey <tromey@redhat.com>
1507 * doc/sourcebuild.texi (libgcj Tests): Don't mention jacks.
1508 * doc/install.texi (Testing): Don't mention jacks.
1509 (Configuration): Document --enable-java-maintainer-mode. Move
1510 --with-java-home to libgcj-specific section. Document
1511 --with-ecj-jar.
1512 (Prerequisites): Mention --enable-java-maintainer-mode, ecj1.
1514 2007-01-15 Jan Hubicka <jh@suse.cz>
1516 * tree-ssa-dce.c (DCE_TODOs): New.
1517 (propagate_necessity): Return if something changed.
1518 (eliminate_unnecessary_stmts): Likewise.
1519 (perform_tree_ssa_dce): Return TODO flags when needed.
1520 (pass_dce, pass_dce_loop, pass_cd_dce): Remove TODO flags.
1522 2007-01-15 Uros Bizjak <ubizjak@gmail.com>
1524 * config/i386/i386.md (fyl2xxf3_i387): Rename from fyl2x_xf3.
1525 (fyl2x_extend<mode>xf3_i387): New insn pattern.
1526 (log<mode>2): Rename from logsf2 and logdf2 and macroize insn
1527 patterns using X87MODEF12 mode macro. Extend operand 1
1528 to XFmode. Use SSE_FLOAT_MODE_P to disable patterns for SSE math.
1529 (log10<mode>2): Ditto.
1530 (log2<mode>2): Ditto.
1531 (log1p<mode>2): Ditto.
1532 (logb<mode>2): Ditto.
1533 (fyl2xp1xf3_i387): Rename from fyl2xp1_xf3.
1534 (fyl2xp1_extend<mode>xf3_i387): New insn pattern.
1535 (*fxtractxf3_i387): Rename from *fxtractxf3.
1536 (fxtract_extend<mode>xf3_i387): New insn pattern.
1537 (ilogbsi2): Use match_dup 3, not match_operand:XF 3.
1538 * config/i386/i386.c (ix86_emit_i387_log1p): Use gen_fyl2xp1xf3_i387()
1539 and gen_fyl2xxf3_i387().
1541 2007-01-14 Zdenek Dvorak <dvorakz@suse.cz>
1543 * loop-unswitch.c (unswitch_loop): Do not call fix_loop_placement.
1544 * cfgloopmanip.c (fix_loop_placement): Made static. Use
1545 get_loop_exit_edges. Changed return type to bool.
1546 * cfgloop.h (fix_loop_placement): Declaration removed.
1548 2007-01-14 Dorit Nuzman <dorit@il.ibm.com>
1550 * param.h (MIN_VECT_LOOP_BOUND): New.
1551 * params.def (MIN_VECT_LOOP_BOUND): New.
1552 * tree-vectorizer.c (slpeel_tree_peel_loop_to_edge): Takes another
1553 argument - minimum threshold for number of iterations.
1554 * tree-vectorizer.h (slpeel_tree_peel_loop_to_edge): Add another
1555 argument to declaration.
1556 * tree-vect-analyze.c (vect_analyze_operations): Check value of
1557 MIN_VECT_LOOP_BOUND.
1558 * tree-vect-transform.c (vect_do_peeling_for_loop_bound): Call
1559 slpeel_tree_peel_loop_to_edge with additional argument.
1560 (vect_do_peeling_for_alignment): Likewise.
1561 * doc/invoke.texi (min-vect-loop-bound): Document new param option.
1563 2007-01-14 Uros Bizjak <ubizjak@gmail.com>
1565 PR target/30413
1566 * config/i386/i386.c (print_operand) ['z']: Output 'b' for
1567 operands of size 1.
1569 2007-01-14 Jan Hubicka <jh@suse.cz>
1571 * tree-dfa.c (remove_referenced_var): New function.
1572 * tree-ssa-live.c (remove_unused_locals): Walk referenced vars and
1573 prune referenced vars list too.
1574 * tree-flow.h (remove_referenced_var): Declare.
1576 2007-01-14 Jan Hubicka <jh@suse.cz>
1578 * tree-eh.c (add_stmt_to_eh_region_fn): Do not add call_exprs
1579 separately.
1580 (remove_stmt_from_eh_region_fn): Do not remove call_exprs.
1581 (verify_eh_throw_stmt_node, verify_eh_throw_table_statements): Kill.
1582 * except.h (verify_eh_throw_table_statements): Kill prototype.
1583 * cfgexpand.c (expand_gimple_basic_block): Propagate Eh regions
1584 into call exrepssions.
1585 * tree-optimize.c (execute_free_cfg_annotatiosn): Do not call
1586 eh trhow verifier.
1587 * tree-cfg.c: Include pointer-set.h.
1588 (verify_node_sharing): Work on pointer set.
1589 (verify_eh_throw_stmt_node): New.
1590 (verify_stmts): Use pointers sets, verify throw_stmt.
1592 2007-01-13 Zdenek Dvorak <dvorakz@suse.cz>
1594 * ipa-reference.c (analyze_function): Consider also addresses taken
1595 in phi nodes.
1597 2007-01-12 Roger Sayle <roger@eyesopen.com>
1599 * c-typeck.c (null_pointer_constant_p): Replace use of
1600 TREE_CONSTANT_OVERFLOW with TREE_OVERFLOW.
1601 (build_c_cast): Likewise.
1603 2007-01-12 Roger Sayle <roger@eyesopen.com>
1605 * tree.h (force_fit_type_double): Remove unused final argument.
1606 * c-common.c (constant_expression_warning): Replace use of
1607 TREE_CONSTANT_OVERFLOW with TREE_OVERFLOW.
1608 (convert_and_check): Likewise.
1609 (shorten_compare): Update call to force_fit_type_double.
1610 (c_common_truthvalue_conversion) <INTEGER_CST>: Use integer_zerop.
1611 * convert.c (convert_to_pointer): Update call to
1612 force_fit_type_double.
1613 * fold-const.c (force_fit_type_double): Remove overflowed_const
1614 argument.
1615 (int_const_binop, fold_convert_const_int_from_int,
1616 fold_convert_const_int_from_real, fold_div_compare,
1617 fold_sign_changed_comparison, fold_unary, fold_negate_const,
1618 fold_abs_const, fold_not_const): Remove the final argument from
1619 calls to force_fit_type_double.
1621 2007-01-12 Andrew Pinski <andrew_pinski@playstation.sony.com>
1623 * configure.ac: Set insn to "nop" for spu-*-* also.
1624 * configure: Regenerate.
1626 2007-01-12 Olga Golovanevsky <olga@il.ibm.com>
1628 * builtins.def : Add BUILT_IN_FREE.
1630 2007-01-12 Jan Hubicka <jh@suse.cz>
1632 PR tree-optimization/30443
1633 * tree-inline.c (tree_function_versioning): Do not optimize when
1634 cloning for inlining.
1636 2007-01-12 Zdenek Dvorak <dvorakz@suse.cz>
1638 * doc/loop.texi: Document recording of loop exits.
1639 * cfgloopmanip.c (loopify, duplicate_loop): Use alloc_loop.
1640 (update_single_exits_after_duplication,
1641 update_single_exit_for_duplicated_loop,
1642 update_single_exit_for_duplicated_loops): Removed.
1643 (duplicate_loop_to_header_edge): Do not call
1644 update_single_exits_after_duplication and
1645 update_single_exit_for_duplicated_loops.
1646 (loop_version): Do not update single_exit information.
1647 (fix_loop_structure): Use record_loop_exits instead of
1648 mark_single_exit_loops.
1649 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update
1650 the lists of loop exits.
1651 * cfghooks.c (redirect_edge_and_branch, redirect_edge_and_branch_force,
1652 split_edge, merge_blocks): Update the lists of loop exits.
1653 * modulo-sched.c (sms_schedule): Pass LOOPS_HAVE_RECORDED_EXITS to
1654 loop_optimizer_init.
1655 * loop-init.c (loop_optimizer_init): Call record_loop_exits instead
1656 of mark_single_exit_loops.
1657 (loop_optimizer_finalize): Call release_recorded_exits.
1658 * tree-ssa-loop.c (tree_loop_optimizer_init): Pass
1659 LOOPS_HAVE_RECORDED_EXITS to loop_optimizer_init.
1660 * tree-vectorizer.c (slpeel_tree_duplicate_loop_to_edge_cfg): Do not
1661 update single exit information.
1662 * lambda-code.c (perfect_nestify): Ditto.
1663 * cfgloop.c (flow_loop_free): Destroy the list of exits of the loop.
1664 (mark_single_exit_loops): Removed.
1665 (alloc_loop, loop_exit_hash, loop_exit_eq, loop_exit_free,
1666 get_exit_descriptions, rescan_loop_exit, record_loop_exits,
1667 dump_recorded_exit, dump_recorded_exits, release_recorded_exits): New
1668 functions.
1669 (get_loop_exit_edges, single_exit): Use recorded exit lists.
1670 (add_bb_to_loop, remove_bb_from_loops): Update the lists of loop exits.
1671 (verify_loop_structure): Verify consistency of the exit lists.
1672 (flow_loops_find): Use alloc_loop. Initialize exits hash.
1673 (set_single_exit): Removed.
1674 * cfgloop.h (struct loop_exit): New function.
1675 (struct loop): single_exit_ field replaced by exits field.
1676 (LOOPS_HAVE_MARKED_SINGLE_EXITS): Replaced by LOOPS_HAVE_RECORDED_EXITS.
1677 (struct loops): Added exits hash.
1678 (mark_single_exit_loops, set_single_exit): Declaration removed.
1679 (release_recorded_exits, record_loop_exits, rescan_loop_exit): Declare.
1681 2007-01-12 Richard Sandiford <richard@codesourcery.com>
1683 * doc/invoke.texi: Avoid use of @headitem.
1685 2007-01-12 Richard Sandiford <richard@codesourcery.com>
1687 * cse.c (cse_insn): Move HAVE_CC0 code after declarations.
1689 2007-01-12 Richard Sandiford <richard@codesourcery.com>
1691 * doc/install.texi: Fix m68k-*-* anchor and add m68k-*-* to the
1692 list of targets.
1694 2007-01-12 Nathan Sidwell <nathan@codesourcery.com>
1695 Richard Sandiford <richard@codesourcery.com>
1697 * doc/invoke.texi: Document -mno-bitfield, -mno-rtd and -mno-short.
1698 * config/m68k/m68k.opt: Resort options.
1699 (mbitfield, mrtd, mshort): Remove RejectNegative properties.
1701 2007-01-12 Nathan Sidwell <nathan@codesourcery.com>
1702 Richard Sandiford <richard@codesourcery.com>
1704 * doc/invoke.texi: Document the macros that are defined by
1705 m68k's -mtune and -mhard-float options.
1706 * config/m68k/m68k-protos.h (m68k_cpp_cpu_ident) Declare.
1707 (m68k_cpp_cpu_family): Likewise.
1708 * config/m68k/m68k.h (TARGET_CPU_CPP_BUILTINS): Add a full set
1709 of __ucfv*__ macros. Define __mcffpu__ if generating code for
1710 ColdFire FPUs. Define __mcf_cpu_* and __mcf_family_* macros.
1711 * config/m68k/m68k.c (m68k_cpp_cpu_ident): New function.
1712 (m68k_cpp_cpu_family): Likewise.
1714 2007-01-12 Richard Sandiford <richard@codesourcery.com>
1716 * config/m68k/m68k.h (TARGET_CPU_CPP_BUILTINS): Treat all mc68*
1717 macros besides mc68000 as tuning macros. Use a switch statement
1718 to set them and mcpu32.
1720 2007-01-12 Julian Brown <julian@codesourcery.com>
1722 * config/m68k/m68k.h: Use TARGET_68040 instead of TARGET_68040_ONLY.
1723 (TARGET_68040_ONLY): Rename to...
1724 (TARGET_68040): ...this.
1725 * config/m68k/m68k.c: Use TARGET_68040 instead of TARGET_68040_ONLY.
1726 * config/m68k/m68k.md: Likewise.
1728 2007-01-12 Julian Brown <julian@codesourcery.com>
1729 Nathan Sidwell <nathan@codesourcery.com>
1730 Richard Sandiford <richard@codesourcery.com>
1732 * config.gcc (m680[012]0-*-*, m68k*-*-*): Set m68k_cpu_ident to
1733 the -mcpu= argument associated with the --with-cpu setting.
1734 Define M68K_DEFAULT_TUNE to the default -mtune= option,
1735 if different from the one implied by the -mcpu setting.
1736 Accept --with-cpu=FOO if FOO is listed in m68k-devices.def,
1737 using mcpu=FOO as the default CPU option. Set target_cpu_default2.
1738 * doc/invoke.texi: Mention ColdFire in the introduction to the
1739 m68k options. Document the new -march, -mcpu, -mtune, -mdiv,
1740 -mno-div and -mhard-float options. Make -m68881 a synonym for
1741 -mhard-float. Document the previously-undocumented -m5206e,
1742 -m528x, -m5307 and -m5407 options. Tweak the existing option
1743 documentation for consistency.
1744 * doc/install.texi: Mention new --with-cpu arguments.
1745 * config/m68k/m68k.h (OPTION_DEFAULT_SPECS): Only use the
1746 default CPU if neither -mcpu nor -march are specified.
1747 (ASM_CPU_SPEC): Pass down -mcpu and -march options.
1748 (TARGET_CPU_CPP_BUILTINS): Set __mcfisa*__ macros from
1749 TARGET_ISA*. Set the legacy __mcf*__ cpu macros in the same way,
1750 using m68k_tune to decide between families that implement the
1751 same ISA. Use m68k_tune to set __mcfv4e__.
1752 (FL_BITFIELD, FL_68881, FL_COLDFIRE, FL_CF_HWDIV, FL_CF_MAC)
1753 (FL_CF_EMAC, FL_CF_EMAC_B, FL_CF_USP, FL_CF_FPU, FL_ISA_68000)
1754 (FL_ISA_68010, FL_ISA_68020, FL_ISA_68040, FL_ISA_A, FL_ISA_B)
1755 (FL_ISA_C, FL_ISA_MMU): New macros.
1756 (MASK_COLDFIRE): Delete.
1757 (TARGET_68010, TARGET_68020, TARGET_68040_ONLY, TARGET_COLDFIRE)
1758 (TARGET_ISAB): Redefine in terms of m68k_cpu_flags.
1759 (TARGET_68881, TARGET_COLDFIRE_FPU): Redefine in terms of m68k_fpu.
1760 (TARGET_HARD_FLOAT): Do not define here.
1761 (TARGET_ISAAPLUS, TARGET_ISAC): New macros.
1762 (TUNE_68000): New macro.
1763 (TUNE_68000_10): Redefine in terms of TUNE_68000 and TUNE_68010.
1764 (TUNE_68010, TUNE_68030, TUNE_68040, TUNE_68060, TUNE_CPU32)
1765 (TUNE_CFV2): Redefine in terms of m68k_tune.
1766 (uarch_type, target_device, fpu_type): New enums.
1767 (m68k_cpu, m68k_tune, m68k_fpu, m68k_cpu_flags): Declare.
1768 * config/m68k/m68k.c (TARGET_DEFAULT): Remove MASK_68881.
1769 (FL_FOR_isa_00, FL_FOR_isa_10, FL_FOR_isa_20, FL_FOR_isa_40)
1770 (FL_FOR_isa_cpu32, FL_FOR_isa_a, FL_FOR_isa_aplus, FL_FOR_isa_b)
1771 (FL_FOR_isa_c): New macros.
1772 (m68k_isa): New enum.
1773 (m68k_target_selection): New structure.
1774 (all_devices, all_isas, all_microarchs): New tables.
1775 (m68k_cpu_entry, m68k_arch_entry, m68k_tune_entry, m68k_cpu)
1776 (m68k_tune, m68k_fpu, m68k_cpu_flags): New variables.
1777 (MASK_ALL_CPU_BITS): Delete.
1778 (m68k_find_selection): New function.
1779 (m68k_handle_option): Handle -mcpu=, -march= and -mtune=.
1780 Map the legacy target options to a combination of the new ones.
1781 (override_options): Set m68k_cpu, m68k_tune, m68k_fpu and
1782 m68k_cpu_flags. Handle M68K_DEFAULT_TUNE. Use m68k_cpu_flags
1783 to derive default MASK_BITFIELD, MASK_CF_HWDIV and MASK_HARD_FLOAT
1784 settings.
1785 * config/m68k/m68k.opt (m5200, m5206e, m528x, m5307, m5407, mcfv4e)
1786 (m68010, m68020, m68020-40, m68020-60, m68030, m68040): Remove Mask
1787 properties.
1788 (m68881, msoft-float): Change mask from 68881 to HARD_FLOAT.
1789 (march=, mcpu=, mdiv, mhard-float, mtune=): New options.
1790 * config/m68k/m68k-devices.def: New file.
1792 2007-01-12 Richard Sandiford <richard@codesourcery.com>
1793 Nathan Sidwell <nathan@codesourcery.com>
1795 * config/m68k/m68k.h (ASM_CPU_SPEC, ASM_SPEC, EXTRA_SPECS)
1796 (SUBTARGET_EXTRA_SPECS): New macros.
1797 * config/m68k/linux.h (ASM_SPEC): Remove CPU flags;
1798 use %(asm_cpu_spec) instead.
1799 * config/m68k/m68k-none.h (ASM_SPEC): Likewise.
1800 * config/m68k/openbsd.h (ASM_SPEC): Likewise.
1801 * config/m68k/netbsd-elf.h (ASM_SPEC): Likewise.
1802 (EXTRA_SPECS): Rename to...
1803 (SUBTARGET_EXTRA_SPECS): ...this.
1805 2007-01-12 Nathan Sidwell <nathan@codesourcery.com>
1806 Richard Sandiford <richard@codesourcery.com>
1807 Julian Brown <julian@codesourcery.com>
1809 * config.gcc (m68k-*-aout*, m68k-*-coff*, m68020-*-elf*, m68k-*-elf*)
1810 (m68k-*-uclinux*, m68k-*-linux*, m68k-*-rtems*): Set default_m68k_cpu
1811 to the configuration's default CPU.
1812 (m68010-*-netbsdelf*, m68k*-*-netbsdelf*, m68k*-*-openbsd*): Likewise.
1813 Remove default masks.
1814 (m680[012]0-*-*): Set the default with_cpu to the first part of
1815 the target name.
1816 (m68k*-*-*): Set the default with_cpu to m$default_m68k_cpu.
1817 (m68k*-*-linux): Extend the --with-cpu handling to...
1818 (m680[012]0-*-*, m68k*-*-*): ...these configurations. Allow m68000
1819 and m68010. Don't set target_cpu_default2.
1820 * doc/install.texi: Document --with-cpu for m68k.
1821 * config/m68k/m68k.h (OPTION_DEFAULT_SPECS): Define.
1822 * config/m68k/m68k-none.h (TARGET_CPU_DEFAULT, M68K_CPU_m68k)
1823 (M68K_CPU_m68000, M68K_CPU_m68010, M68K_CPU_m68020, M68K_CPU_m68030)
1824 (M68K_CPU_m68040, M68K_CPU_m68302, M68K_CPU_m68332, TARGET_DEFAULT)
1825 (ASM_CPU_DEFAULT_SPEC, CC1_CPU_DEFAULT_SPEC): Delete.
1826 (ASM_SPEC): Remove use of %(asm_cpu_default).
1827 (EXTRA_SPECS, SUBTARGET_EXTRA_SPECS, MULTILIB_DEFAULTS): Delete.
1828 * config/m68k/linux.h (TARGET_DEFAULT): Delete.
1829 (CPP_SPEC): Merge definitions. Do not handle __HAVE_68881__ here.
1830 * config/m68k/netbsd-elf.h (TARGET_OS_CPP_BUILTINS): Define
1831 __HAVE_FPU__ if TARGET_HARD_FLOAT.
1832 (TARGET_DEFAULT): Delete.
1833 (EXTRA_SPECS): Delete cpp_cpu_default_spec, cpp_cpu_spec,
1834 cpp_fpu_spec, asm_default_spec and netbsd_cpp_spec.
1835 (CPP_CPU_SPEC): Delete.
1836 (TARGET_VERSION): Merge definitions, using TARGET_68010 to pick
1837 the appropriate string.
1838 (CPP_CPU_DEFAULT_SPEC, ASM_DEFAULT_SPEC, CPP_FPU_SPEC): Delete.
1839 (CPP_SPEC): Define to NETBSD_CPP_SPEC.
1840 (ASM_SPEC): Don't use %(asm_default_spec).
1841 * config/m68k/m68k.c (TARGET_DEFAULT_TARGET_FLAGS): Remove
1842 TARGET_DEFAULT and add MASK_68881.
1843 * config/m68k/m68k.md: Remove mention of TARGET_DEFAULT from comments.
1845 2007-01-12 Richard Sandiford <richard@codesourcery.com>
1847 * config.gcc (m68010-*-netbsdelf*): Add MASK_68010.
1848 (m68k*-*-netbsdelf*, m68k*-*-openbsd*, m68k*-linux*): Add
1849 MASK_68010 alongside MASK_68020.
1850 * doc/invoke.texi: Document -m68010.
1851 * config/m68k/m68k.opt (m68010): New.
1852 * config/m68k/m68k.h (TARGET_CPU_CPP_BUILTINS): Define mc68010
1853 if TUNE_68010.
1854 (TUNE_68010): New macro.
1855 * config/m68k/m68k-none.h (M68K_CPU_m68k, M68K_CPU_m68010)
1856 (M68K_CPU_m68020, M68K_CPU_m68030, M68K_CPU_m68040)
1857 (M68K_CPU_m68332): Add MASK_68010.
1858 * config/m68k/linux.h (TARGET_DEFAULT): Add MASK_68010 to
1859 fallback definition.
1860 * config/m68k/netbsd-elf.h (CPP_CPU_SPEC): Remove now-redundant
1861 defines.
1862 * config/m68k/m68k.c (MASK_ALL_CPU_BITS): Add MASK_68010.
1863 (m68k_handle_option): Handle OPT_m68010. Add MASK_68010
1864 to all entries that use MASK_68020.
1865 (output_move_simode_const, output_move_himode, output_move_qimode)
1866 (output_move_stricthi, output_move_strictqi): Use TARGET_68010
1867 instead of TARGET_68020 to select clr behavior. Remove comment
1868 about there being no TARGET_68010.
1869 * config/m68k/m68k.md: Likewise throughout.
1871 2007-01-12 Julian Brown <julian@codesourcery.com>
1873 * config/m68k/m68k.h (TARGET_ISAB): New macro.
1874 * config/m68k/m68k.c: Use TARGET_ISAB rather than TARGET_CFV4.
1875 * config/m68k/m68k.md: Likewise.
1877 2007-01-12 Julian Brown <julian@codesourcery.com>
1879 * config/m68k/m68k.h (LEGITIMATE_INDEX_P, LEGITIMIZE_ADDRESS): Use
1880 TARGET_COLDFIRE_FPU instead of TARGET_CFV4E.
1882 2007-01-12 Julian Brown <julian@codesourcery.com>
1884 * config/m68k/m68k.h (TUNE_68040_60): New macro.
1885 * config/m68k/m68k.c (standard_68881_constant_p): Use it.
1886 * config/m68k/m68k.md: Likewise.
1888 2007-01-12 Julian Brown <julian@codesourcery.com>
1889 Richard Sandiford <richard@codesourcery.com>
1891 * config/m68k/m68k.h (TARGET_CPU_CPP_BUILTINS): Use TUNE_68030
1892 instead of TARGET_68030, TUNE_68040 instead of TARGET_68040,
1893 TUNE_68060 instead of TARGET_68060 and TUNE_CPU32 instead of
1894 TARGET_CPU32.
1895 (TARGET_CPU32): Rename to...
1896 (TUNE_CPU32): ...this.
1897 (TUNE_68000_10, TUNE_68030, TUNE_68040, TUNE_68060)
1898 (TUNE_CFV2): New macros.
1899 * config/m68k/netbsd-elf.h (LONG_DOUBLE_TYPE_SIZE): Simplify;
1900 remove conditions that are implied by TARGET_68020.
1901 * config/m68k/m68k.c (m68k_output_function_prologue): Use TUNE_68040
1902 instead of TARGET_68040 and TUNE_CPU32 instead of TARGET_CPU32.
1903 (m68k_output_function_epilogue): Likewise.
1904 (m68k_rtx_costs): Likewise. Use TUNE_68060 instead of TARGET_68060
1905 and TUNE_CFV2 instead of TARGET_5200. Use TUNE_68000_10 instead of
1906 "!TARGET_68020 && !TARGET_COLDFIRE" to choose between 68000 and
1907 non-68000 timings. Refactor multiplication and division costs.
1908 (output_addsi3): Use TUNE_68040 instead of TARGET_68040 and
1909 TUNE_CPU32 instead of TARGET_CPU32.
1910 (standard_68881_constant_p): Use TUNE_68040 instead of TARGET_68040
1911 and TUNE_68060 instead of TARGET_68060.
1912 * config/m68k/m68k.md: Use TUNE_68040 instead of TARGET_68040,
1913 TUNE_68060 instead of TARGET_68060, and TUNE_CPU32 instead of
1914 TARGET_CPU32.
1915 (movsi_const0): Use TUNE_68000_10 rather than "!TARGET_68020
1916 && !TARGET_COLDFIRE" to choose between moveq and clr.
1917 Likewise in the unnamed movsf pattern.
1918 (ashlsi_17_24, lshrsi_17_24): Guard with TUNE_68000_10 rather than
1919 "!TARGET_68020 && !TARGET_COLDFIRE". Likewise the unnamed
1920 ashiftrt pattern.
1922 2007-01-12 Richard Sandiford <richard@codesourcery.com>
1924 * config/m68k/m68k.h (TARGET_CPU_CPP_BUILTINS): Increase amount
1925 of tabbing before backslashes.
1927 2007-01-11 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
1929 * pa-linux.h (ASM_OUTPUT_INTERNAL_LABEL): Undefine.
1930 * pa.h (ASM_OUTPUT_LABEL): Output colon when using GAS.
1931 (ASM_OUTPUT_INTERNAL_LABEL): Define.
1933 2007-01-11 Zdenek Dvorak <dvorakz@suse.cz>
1935 * tree-ssa-loop-ivopts.c (extract_cond_operands): Split from
1936 find_interesting_uses_cond.
1937 (find_interesting_uses_cond): Use extract_cond_operands.
1938 (rewrite_use_compare): Use extract_cond_operands and
1939 force_gimple_operand_bsi. Do not call update_stmt.
1940 (determine_use_iv_cost_condition): Use extract_cond_operands.
1941 Return cheaper of using original bound and changing the exit bound.
1943 2007-01-11 Zdenek Dvorak <dvorakz@suse.cz>
1945 PR tree-optimization/29516
1946 * tree-ssa-address.c (tree_mem_ref_addr, add_to_parts,
1947 most_expensive_mult_to_index, addr_to_parts,
1948 create_mem_ref, maybe_fold_tmr): Make the type of
1949 fields of TARGET_MEM_REF sizetype.
1950 (move_fixed_address_to_symbol, move_pointer_to_base):
1951 New functions.
1952 * tree.def (TARGET_MEM_REF): Add comment on types of
1953 the operands.
1955 2007-01-11 Joseph Myers <joseph@codesourcery.com>
1957 * c-common.c (vector_types_convertible_p): Treat opaque types as
1958 always convertible if they have the same size, but not otherwise.
1960 2007-01-11 Steven Bosscher <steven@gcc.gnu.org>
1962 * ifcvt.c (struct noce_if_info): Add comments to the fields.
1963 Remove the b_unconditional field.
1964 (noce_try_sign_mask): Do not look at b_unconditional.
1965 (noce_process_if_block): Do not use merge_if_blocks. Update
1966 the CFG here. Do not set b_unconditional.
1967 (cond_move_process_if_block): Likewise.
1968 (find_cond_trap): Likewise.
1969 (check_cond_move_block): Require simple jump insns at the end
1970 of the basic block.
1972 2007-01-11 Jan Hubicka <jh@suse.cz>
1974 PR tree-optimization/1046
1975 * tree-tailcall.c (suitable_for_tail_call_opt_p): Use TREE_ADDRESSABLE
1976 when alias info is not ready.
1977 (pass_tail_recursion): Do not require aliasing.
1978 * tree-ssa-copyrename.c (pass_rename_ssa_cop): Likewise.
1979 * tree-ssa-ccp.c (pass_ccp, pass_fold_builtins): Likewise.
1980 * tree-ssa-copy.c (pass_copy_prop): Likewise.
1981 * tree-ssa-forwprop.c (pass_forwprop): Likewise.
1982 * tree-ssa-dce.c (pass_dce, pass_dce_loop, pass_cd_dce): Likewise.
1983 * passes.c (init_optimization_passes): Execute rename_ssa_copies,
1984 ccp, forwprop, copy_prop, merge_phi, copy_prop, dce and tail recursion
1985 before inlining.
1986 * tree-ssa-operands.c (add_virtual_operand, get_indirect_ref_operand):
1987 When aliasing is not build, mark statement as volatile.
1989 2007-01-11 Tom Tromey <tromey@redhat.com>
1991 PR preprocessor/15185, PR preprocessor/20989:
1992 * doc/cppopts.texi <-MT>: Update description of algorithm for
1993 computing default target.
1994 <-M, -MD>: Reword "basename" text.
1996 2007-01-11 Roger Sayle <roger@eyesopen.com>
1998 * builtins.c (expand_builtin_pow, expand_builtin_powi,
1999 fold_builtin_cabs, fold_builtin_sqrt, fold_builtin_trunc,
2000 fold_builtin_floor, fold_builtin_ceil, fold_builtin_round,
2001 fold_builtin_int_int_roundingfn, fold_builtin_bitop,
2002 fold_builtin_bswap, real_constp, fold_builtin_pow,
2003 fold_builtin_powi, fold_builtin_signbit, fold_builtin_copysign,
2004 do_mpfr_arg1, do_mpfr_arg2, do_mpfr_arg3, do_mpfr_sincos): Replace
2005 uses of the macro TREE_CONSTANT_OVERFLOW with TREE_OVERFLOW.
2006 * convert.c (convert_to_pointer): Likewise.
2007 * expr.c (highest_pow2_factor, expand_expr_real_1): Likewise.
2008 * fold-const.c (force_fit_type, fold_negate_expr, int_const_binop,
2009 const_binop, fold_convert_const_int_from_int,
2010 fold_convert_const_int_from_real,
2011 fold_convert_const_real_from_real, sign_bit_p,
2012 optimize_minmax_comparison, extract_muldiv_1, fold_div_compare,
2013 fold_sign_changed_comparison, fold_unary, fold_comparison,
2014 fold_binary, multiple_of_p, tree_Expr_non_zero_p,
2015 fold_negate_const, fold_abs_const, fold_not_const): Likewise.
2016 * print-tree.c (print_node_brief, print_node): Likewise.
2017 * stor-layout.c (place_field, layout_type): Likewise.
2018 * tree-chrec.c (keep_cast): Likewise.
2019 * tree.c (build_vector, build_real, build_real_from_int_cst,
2020 build_complex): Likewise.
2022 2007-01-11 Roger Sayle <roger@eyesopen.com>
2024 * tree.h (TREE_CONSTANT_OVERFLOW): Obsolete. For the time being,
2025 treat TREE_CONSTANT_OVERFLOW as a synonym of TREE_OVERFLOW.
2027 2007-01-11 Paolo Bonzini <bonzini@gnu.org>
2029 * configure.ac (strict1_warn): Rename to strict_warn.
2030 (WERROR, --enable-werror, symlink hacks, stage1_cflags,
2031 cc_set_by_configure, quoted_cc_set_by_configure,
2032 stage_prefix_set_by_configure, quoted_stage_prefix_set_by_configure,
2033 all_boot_languages, all_stagestuff): Remove.
2034 (target_list): Remove bootstrap targets.
2035 * Makefile.in (quickstrap): Unconditionally make a synonym of all.
2036 (BOOT_LANGUAGES, STAGE1_CFLAGS, STAGE1_CHECKING,
2037 REMAKEFLAGS, FLAGS_TO_PASS, PREPEND_DOTDOT_TO_RELATIVE_PATHS,
2038 SUBDIR_FLAGS_TO_PASS, WERROR_FLAGS, STRICT2_WARN, LANG_STAGESTUFF,
2039 VOL_FILES, POSTSTAGE1_FLAGS_TO_PASS, STAGE2_FLAGS_TO_PASS,
2040 STAGEPROFILE_FLAGS_TO_PASS, STAGEFEEDBACK_FLAGS_TO_PASS, stage1_build,
2041 stage1_copy, stage2_build, stage2_copy, stageprofile_build,
2042 stageprofile_copy, stage3_build, stage3_copy, stagefeedback_build,
2043 stagefeedback_copy, stage4_build, clean_s1, clean_sw, bootstrap,
2044 bootstrap-lean, bootstrap2, bootstrap2-lean, bootstrap3,
2045 bootstrap3-lean, bootstrap4, bootstrap4-lean, unstage1, unstage2,
2046 unstage3, unstage4, unstageprofile, unstagefeedback, restage, restage2,
2047 restage3, restage4, restageprofile, restagefeedback, bubbleestrap,
2048 cleanstrap, unstrap, restrap, *compare, *compare3, *compare4,
2049 *compare-lean, *compare3-lean, *compare4-lean, stage1-start, stage1,
2050 stage2-start, stage2, stage3-start, stage3, stage4-start, stage4,
2051 stageprofile-start, stageprofile, stagefeedback-start, stagefeedback,
2052 risky-stage1, risky-stage2, risky-stage3, risky-stage4): Remove.
2053 (ORDINARY_FLAGS_TO_PASS): Rename to FLAGS_TO_PASS.
2054 (STAGECOPYSTUFF, STAGEMOVESTUFF): Consolidate into MOSTLYCLEANFILES.
2055 (mostlyclean): Adjust.
2056 (clean, distclean): Don't mention bootstrap stuff.
2057 * configure: Regenerate.
2058 * ada/config-lang.in, cp/config-lang.in, forttran/config-lang.in,
2059 java/config-lang.in, objc/config-lang.in, objcp/config-lang.in,
2060 treelang/config-lang.in (stagestuff): Remove.
2061 * doc/sourcebuild.texi (stage1, stage2, stage3, stage4,
2062 stageprofile, stagefeedback, stagestuff): Remove mention.
2064 2007-01-11 Nick Clifton <nickc@redhat.com>
2066 * config/mcore/predicates.md (mcore_general_movesrc_operand):
2067 Accept CONSTs.
2068 (mcore_general_movdst_operand): Do not accept CONST_INTs.
2069 (mcore_arith_K_S_operand): Run the test for the S constraint not
2070 the test for the M constraint.
2071 (mcore_addsub_operand): Do not accept integer values that are
2072 larger than 32 bits.
2073 * config/mcore/mcore.md: Remove unused constraints from split.
2074 (andsi3): Use HOST_WIDE_INT instead of int to hold an INTVAL.
2075 (addsi3): Likewise.
2076 (allocate_stack): Likewise.
2077 * config/mcore/mcore.c (mcore_print_operand): Restrict output of P
2078 operands to 32 bits.
2079 (mcore_const_costs): Use HOST_WIDE_INT instead of int to hold an
2080 INTVAL.
2081 (mcore_and_cost, mcore_modify_comparison, const_ok_for_mcore,
2082 mcore_const_ok_for_inline, mcore_const_trick_uses_not,
2083 try_constant_tricks, mcore_num_ones, mcore_num_zeros,
2084 mcore_output_bclri, mcore_output_andn, output_inline_const,
2085 mcore_output_move, mcore_output_movedouble): Likewise.
2086 (mcore_output_cmov): Use CONST_OK_FOR_M and CONST_OK_FOR_N.
2087 (output_inline_const): Likewise.
2088 (output_inline_const): Fix format strings used in sprintf
2089 statements.
2090 * config/mcore/mcore-protos.h: Update prototypes for changed
2091 functions in mcore.c.
2092 * config/mcore/mcore.h (CONST_OK_FOR_I): Cast values to
2093 HOST_WIDE_INT and not int.
2094 (CONST_OK_FOR_J, CONST_OK_FOR_K, CONST_OK_FOR_L, CONST_OK_FOR_M,
2095 CONST_OK_FOR_N): Likewise.
2096 (LEGITIMATE_CONSTANT_P): Also check CONSTANT_P.
2097 (GO_IF_LEGITIMATE_INDEX): Use HOST_WIDE_INT instead of int to hold
2098 an INTVAL.
2100 2007-01-10 Jan Hubicka <jh@suse.cz>
2102 * tree-vrp.c (remove_range_assertions): Release defs.
2103 * tree-ssa-loop-ivopts.c (rmeove_statement): Likewise.
2104 * tree-ssa-dom.c (remove_stmt_or_phi): Likewise.
2106 2007-01-10 Paul Brook <paul@codesourcery.com>
2108 * config/arm/arm.c (arm_rtx_costs_1): Handle mutiply-subtract.
2109 * config/arm/arm.md (mulsi3subsi): New insn.
2111 2007-01-10 Zdenek Dvorak <dvorakz@suse.cz>
2113 * tree-ssa-loop-manip.c (tree_unroll_loop): Make it a wrapper over ...
2114 (tree_transform_and_unroll_loop): New.
2115 * tree-flow.h (transform_callback, tree_transform_and_unroll_loop):
2116 Declare.
2118 2007-01-10 Robert Kennedy <jimbob@google.com>
2120 * fold-const.c (fold_comparison): Fold comparisons like (x *
2121 1000 < 0) to (x < 0).
2123 2007-01-10 Ian Lance Taylor <iant@google.com>
2125 * tree-pretty-print.c (dump_generic_node): Print parentheses when
2126 operands have the same priority.
2128 2007-01-10 Tom Tromey <tromey@redhat.com>
2130 * fold-const.c (fold_truthop): Don't check can_use_bit_fields_p.
2131 (fold_binary): Likewise.
2132 * langhooks.c (lhd_can_use_bit_fields_p): Removed.
2133 * langhooks-def.h (lhd_can_use_bit_fields_p): Removed.
2134 (LANG_HOOKS_CAN_USE_BIT_FIELDS_P): Removed.
2135 (LANG_HOOKS_INITIALIZER): Remove LANG_HOOKS_CAN_USE_BIT_FIELDS_P.
2136 * langhooks.h (struct lang_hooks): Removed field
2137 'can_use_bit_fields_p'.
2139 2007-01-10 Ralf Corsépius <ralf.corsepius@rtems.org>
2141 * config/bfin/t-bfin, config/bfin/t-bfin-elf: Remove GCC_CFLAGS.
2143 2007-01-10 Razya Ladelsky <razya@il.ibm.com>
2145 * function.c (get_last_funcdef_no): New function.
2146 * function.h (get_last_funcdef_no): Declare.
2147 * tree-inline.c (initialize_cfun): Add initialization.
2148 (tree_function_versioning): Cleanup.
2150 2007-01-10 Jan Hubicka <jh@suse.cz>
2152 * tree-inline.c (setup_one_parameter): Do not propagate into abnormal
2153 PHIs.
2155 2007-01-10 Sa Liu <saliu@de.ibm.com>
2156 Ben Elliston <bje@au.ibm.com>
2158 * spu.h (STACK_SAVE_AREA): Use VOIDmode for SAVE_FUNCTION, SImode
2159 for SAVE_NONLOCAL and Pmode for any other save level.
2160 * spu-protos.h (spu_restore_stack_block): Declare.
2161 * spu.md (save_stack_block): Remove.
2162 (restore_stack_block): Call spu_restore_stack_block.
2163 * spu.c (spu_restore_stack_block): New function.
2164 (spu_expand_epilogue): Remove old comment.
2166 2007-01-09 Zdenek Dvorak <dvorakz@suse.cz>
2168 PR tree-optimization/30322
2169 * tree-ssa-loop-ivopts.c (fold_affine_expr, iv_value): Removed.
2170 (cand_value_at): Return the value as aff_tree.
2171 (may_eliminate_iv): Convert the bound from aff_tree to tree.
2172 * tree-affine.c (aff_combination_add_cst, aff_combination_add_product,
2173 aff_combination_mult): New functions.
2174 (aff_combination_add): Use aff_combination_add_cst.
2175 (aff_combination_convert): Allow conversions to a wider type.
2176 (tree_to_aff_combination): Handle BIT_NOT_EXPR.
2177 * tree-affine.h (aff_combination_mult): Declare.
2179 2007-01-09 Carlos O'Donell <carlos@codesourcery.com>
2181 * doc/tm.texi: Update documentation to reflect reality of exec
2182 and start file search behaviours. Update copyright year.
2183 * doc/invoke.texi: Explain how GCC_EXEC_PREFIX is used to find
2184 header file directories.
2186 2007-01-09 Uros Bizjak <ubizjak@gmail.com>
2188 * config/i386/i386.md (*sinxf2): Rename to *sinxf2_i387.
2189 (*cosxf2): Rename to cosxf2_i387.
2190 (*sindf2, *sinsf2): Extend operand 1 to XFmode. Macroize patterns
2191 using X87MODEF12 mode macro. Rename patterns to
2192 *sin_extend<mode>xf2_i387. Use SSE_FLOAT_MODE_P to disable patterns
2193 for SSE math.
2194 (*cosdf2, *cossf2): Ditto.
2195 (sincosdf3, sincossf3): Ditto. Rewrite corresponding splitters
2196 to match extended input operands.
2197 (sincos<mode>3): New expander.
2198 (*sinextendsfdf2, *cosextendsfdf2, *sincosextendsfdf3): Remove
2199 insn patterns and corresponding splitters.
2201 2007-01-09 Kaz Kojima <kkojima@gcc.gnu.org>
2203 * config/sh/t-linux (TARGET_LIBGCC2_CFLAGS): Delete.
2204 (SHLIB_MAPFILES, SHLIB_LINK, SHLIB_INSTALL): Likewise.
2206 2007-01-09 Nicolas Pitre <nico@cam.org>
2208 PR target/30173
2209 * arm/ieee754-df.S (Lad_s): Also test the low word of X for zero.
2211 2007-01-08 Geoffrey Keating <geoffk@apple.com>
2213 * target.h (struct gcc_target): New field library_rtti_comdat.
2214 * target-def.h (TARGET_CXX_LIBRARY_RTTI_COMDAT): New.
2215 (TARGET_CXX): Add TARGET_CXX_LIBRARY_RTTI_COMDAT.
2216 * doc/tm.texi (C++ ABI): Document TARGET_CXX_LIBRARY_RTTI_COMDAT.
2217 * config/darwin.h (TARGET_CXX_LIBRARY_RTTI_COMDAT): Define.
2219 2007-01-08 Geoffrey Keating <geoffk@apple.com>
2221 * doc/invoke.texi (Optimize Options): Correct description of -O0.
2223 2007-01-08 Richard Guenther <rguenther@suse.de>
2225 * tree.h (force_fit_type_double): Export.
2226 (force_fit_type): Remove.
2227 * fold-const.c (force_fit_type_double): New function.
2228 (force_fit_type): Remove.
2229 (int_const_binop): Use it.
2230 (fold_convert_const_int_from_int): Likewise.
2231 (fold_convert_const_int_from_real): Likewise.
2232 (fold_div_compare): Likewise.
2233 (fold_sign_changed_comparison): Likewise.
2234 (fold_unary): Likewise.
2235 (fold_negate_const): Likewise.
2236 (fold_abs_const): Likewise.
2237 (fold_not_const): Likewise.
2238 * c-common.c (shorten_compare): Use force_fit_type_double.
2239 * convert.c (convert_to_pointer): Likewise.
2241 2007-01-08 Richard Guenther <rguenther@suse.de>
2243 * tree.h (build_int_cst_wide_type): Export.
2244 * tree.c (build_int_cst_wide_type): New function.
2245 (build_int_cst_wide): Fix comment.
2246 * builtins.c (fold_builtin_object_size): Use build_int_cst
2247 to build -1 or 0 of the correct type. Use fit_double_type
2248 to check for overflow.
2249 * fold-const.c (optimize_bit_field_compare): Use build_int_cst_type
2250 to build the mask.
2251 (decode_field_reference): Likewise.
2252 (all_ones_mask_p): Likewise.
2253 (native_interpret_int): Use build_int_cst_wide_type.
2254 (fold_binary): Use build_int_cst_type to build an all-ones
2255 value.
2256 * stor-layout.c (set_sizetype): Use build_int_cst_wide_type.
2258 2007-01-08 Daniel Jacobowitz <dan@codesourcery.com>
2260 * config/pa/t-pa64 (libgcc_stub.a): Use $(T).
2262 2007-01-09 Ben Elliston <bje@au.ibm.com>
2264 * genautomata.c (STATS_OPTION): New option.
2265 (stats_flag): New flag.
2266 (gen_automata_option): Handle it.
2267 (initiate_automaton_gen): Ditto.
2268 (write_automata): Output statistics only if stats_flag is
2269 set. Likewise, output time statistics only if time_flag is set.
2270 * doc/md.texi (Processor pipeline description): Document new flag.
2272 2007-01-08 Richard Guenther <rguenther@suse.de>
2274 * builtins.c (fold_builtin_int_roundingfn): Use fit_double_type.
2275 * tree.c (build_int_cst_type): Likewise.
2276 (size_in_bytes): Don't call force_fit_type on the result.
2277 (int_fits_type_p): Use fit_double_type.
2278 * fold-const.c (fit_double_type): New function.
2279 (force_fit_type): Use it.
2280 * tree.h (fit_double_type): Export.
2282 2007-01-08 Jan Hubicka <jh@suse.cz>
2284 * tree-vectorizer.c (gate_increase_alignment): Fix return type.
2285 * ipa.c (function_and_variable_visibility): Fix return type.
2287 2007-01-08 Richard Guenther <rguenther@suse.de>
2289 * tree-ssa-ccp.c (maybe_fold_offset_to_array_ref): Use type
2290 of offset to build the index.
2291 * tree-pretty-print.c (dump_generic_node): Don't build negated
2292 const just for printing.
2293 * c-pretty-print.c (pp_c_integer_constant): Likewise.
2294 * builtins.c (fold_builtin_int_roundingfn): Check if result
2295 fits the type by using force_fit_type and comparing the result.
2296 * predict.c (predict_loops): Use compare_tree_int for comparison.
2297 * tree.c (build_int_cst): Fall back to integer_type_node for
2298 NULL_TREE type.
2299 (build_int_cst_wide): Assert type is non-null.
2301 2007-01-08 Roberto Costa <roberto.costa@st.com>
2303 * tree-vrp.c (extract_range_from_cond_expr): New.
2304 (extract_range_from_expr): Handle COND_EXPR nodes used as expressions.
2305 * tree-ssa-ccp.c (get_maxval_strlen): Handle COND_EXPR nodes used
2306 as expressions.
2307 (fold_stmt): Bug fix, avoid infinite recursion when folding COND_EXPRs.
2308 * tree-ssa-forwprop.c (simplify_cond, forward_propagate_into_cond,
2309 tree_ssa_forward_propagate_single_use_vars): Handle COND_EXPR nodes
2310 used as expressions.
2311 * tree-object-size.c (cond_expr_object_size): New.
2312 (collect_object_sizes_for): Handle COND_EXPR nodes used as expressions.
2314 2007-01-08 Jan Hubicka <jh@suse.cz>
2316 * tree-ssa-forwprop.c (forward_propagate_into_cond,
2317 tree_ssa_forward_propagate_single_use_va): Release defs of propagated
2318 statement.
2320 2007-01-08 Richard Guenther <rguenther@suse.de>
2322 PR tree-optimization/23603
2323 * tree-vrp.c (set_value_range_to_truthvalue): New function.
2324 (extract_range_from_binary): Fall back to truthvalue instead of
2325 varying for TRUTH_*_EXPR.
2326 (extract_range_from_comparison): Fall back to truthvalue instead of
2327 varying.
2328 (vrp_visit_phi_node): Don't adjust new range bounds to +INF/-INF
2329 if all visited PHI values were constant.
2331 2007-01-08 Jan Hubicka <jh@suse.cz>
2333 * cgraphunit.c (cgraph_process_new_functions): Reset reachable flag.
2334 (cgraph_analyze_function): break out from ...
2335 (cgraph_finalize_compilation_unit): ... here.
2336 (cgraph_expand_function): Remove forgoten commented out line.
2337 (cgraph_optimize): Analyze functions.
2339 2007-01-08 Jan Hubicka <jh@suse.cz>
2341 * tree-pas.h (TODO_remove_function): New flag.
2342 (TODO_update*): Renumber.
2343 (pass_ipa_increase_alignment,
2344 pass_ipa_function_and_variable_visibility): New passes.
2345 * cgraphunit.c (cgraph_increase_alignment): Move to tree-vectorizer.c
2346 (cgraph_function_and_variable_visibility): Move to ipa.c
2347 (cgraph_optimize): Don't call cgraph_function_and_variable_visibility,
2348 cgraph_increase_alignment.
2349 * ipa-inline.c (cgraph_decide_inlining): Don't push timevar.
2350 (cgraph_decide_inlining_incrementally): Push TV_INTEGRATION before
2351 calling tree-inline.
2352 (cgraph_early_inlining): Do not call cgraph_remove_unreachable_nodes.
2353 (pass_ipa_inline, pass_early_ipa_inlining): Set TODO_remove_functions
2354 * tree-vectorizer.c (increase_alignment): Move here from cgraphunit.c
2355 (gate_increase_alignment): New function.
2356 (pass_ipa_increase_alignment): New pass.
2357 * ipa.c: Inline tree-pass.h and timevar.h
2358 (function_and_variable_visibility): Move here from cgraphunit.c
2359 * tree-optimize.c (pass_early_local_passes): Add TODO_remove_functions.
2360 * passes.c (init_optimization_passes): Add the two new passes.
2361 (execute_todo): Handle cgraph_remove_functions.
2363 2007-01-08 Nick Clifton <nickc@redhat.com>
2365 * config/frv/predicates.md (reg_or_0_operand): Accept
2366 CONST_DOUBLEs.
2368 2007-01-08 Ralf Corsépius <ralf.corsepius@rtems.org>
2370 * config/bfin/rtems.h, config/bfin/t-rtems: New.
2371 * config.gcc: Add bfin*-rtems*.
2373 2007-01-08 Mark Shinwell <shinwell@codesourcery.com>
2375 * c.opt: Add -flax-vector-conversions.
2376 * c-typeck.c (convert_for_assignment): Pass flag to
2377 vector_types_convertible_p to allow emission of note.
2378 (digest_init): Likewise.
2379 * c-opts.c: Handle -flax-vector-conversions.
2380 * c-common.c (flag_lax_vector_conversions): New.
2381 (vector_types_convertible_p): Unless -flax-vector conversions
2382 has been passed, disallow conversions between vectors with
2383 differing numbers of subparts and/or element types. If such
2384 a conversion is disallowed, possibly emit a note on the first
2385 occasion only to inform the user of -flax-vector-conversions.
2386 The new last argument specifies this.
2387 * c-common.h (flag_lax_vector_conversions): New.
2388 (vector_types_convertible_p): Add extra argument.
2389 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use
2390 char_type_node for V*QI type vectors.
2391 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
2392 Update to satisfy new typechecking rules.
2393 * config/rs6000/altivec.h (vec_cmple): Use vec_cmpge, for both
2394 C and C++ variants.
2395 * doc/invoke.texi (C Dialect Options): Document
2396 -flax-vector-conversions.
2398 2007-01-08 Mark Shinwell <shinwell@codesourcery.com>
2400 PR tree-optimization/29877
2401 * tree-ssa-ter.c (is_replaceable_p): Deem assignments with
2402 a register variable on the RHS to not be replaceable.
2404 2007-01-08 Chen Liqin <liqin@sunnorth.com.cn>
2405 * config/score/t-score-elf (MULTILIB_OPTIONS): Change.
2406 * config/score/predicates.md (const_uimm5, sr0_operand, const_simm12,
2407 const_simm15, const_pow2, const_npow2): Added.
2408 * config/score/misc.md (insv, extv, extzv, movmemsi,
2409 move_lbu_a/b, mov_lhu_a/b etc): Added and fix some bug.
2410 * config/score/score.c (score_address_cost, score_select_cc_mode):
2411 Added.
2412 Change CONST_OK_FOR_LETTER_P/EXTRA_CONSTRAINT define.
2413 Update score_rtx_costs for MACRO TARGET_RTX_COSTS.
2414 Update score_print_operand.
2415 * config/score/score.h (DATA_ALIGNMENT, SELECT_CC_MODE): Added.
2416 Adjust register allocate order and update some macro define.
2417 * config/score/score-mdaux.c (mdx_unaligned_load, mdx_unsigned_store,
2418 mdx_block_move_straight, mdx_block_move_loop_head,
2419 mdx_block_move_loop_body, mdx_block_move_loop_foot, mdx_block_move_loop,
2420 mdx_block_move): Added.
2421 (mdx_movsicc, mdp_select_add_imm, mdp_select, mds_zero_extract_andi,
2422 mdp_limm): Updated and fix some bug and typo.
2423 * config/score/score.md (movqi/hi/si, add/sub/zero/ext): Updated.
2424 (movsf, movdf, doloop_end): Added.
2426 2007-01-08 Kazu Hirata <kazu@codesourcery.com>
2428 * config/arm/arm.c, config/arm/arm.h, config/arm/arm.md,
2429 config/arm/thumb2.md: Fix comment typos.
2430 * doc/extend.texi: Fix a typo.
2432 2007-01-07 Eric Christopher <echristo@apple.com>
2434 * configure.ac: Check for __stack_chk_fail for darwin.
2435 * configure: Regenerate.
2437 2007-01-07 Richard Guenther <rguenther@suse.de>
2439 * tree-vrp.c (extract_range_from_assert): CSE calls to
2440 compare_values where possible.
2441 (extract_range_from_unary_expr): Likewise.
2443 2007-01-07 Anatoly Sokolov <aesok@post.ru>
2445 * config/avr/avr-protos.h (call_insn_operand): Delete prototype.
2446 * config/avr/avr.c (call_insn_operand): Delete function.
2447 * config/avr/avr.md (*pushqi, *pushhi, *pushsi, *pushsf): Use REG_SP
2448 instead of register number. Use predicates.
2449 * config/avr/predicates.md (const0_operand, reg_or_0_operand,
2450 call_insn_operand): Add.
2452 2007-01-06 Jan Hubicka <jh@suse.cz>
2454 * tree-pass.h (pass_build_cgraph_edges): Declare.
2455 * cgraphunit.c (record_refernece): Move to cgraphbuild.c
2456 (visited_nodes): Remove.
2457 (cgraph_create_edges): Move to cgraphbuild.c; rename to
2458 build_cgrpah_edges; make visited_nodes local.
2459 (cgraph_process_new_functions): DO not call initialize_inline_failed.
2460 (record_references_in_initializer): Move to cgraphbuild.c
2461 (initialize_inline_failed, rebuild_cgraph_edges,
2462 pass_rebuild_cgraph_edges): Move to cgraphbuild.c.
2463 (verify_cgraph_node): Make visited_nodes local.
2464 (cgraph_analyze_function): Do not call cgraph_create_edges and
2465 initialize_inline_failed.
2466 (cgraph_expand_function): Do not call cgraph_lower_function;
2467 assert that function is already lowered.
2468 * Makefile.in (cgraphbuild.o): New.
2469 * passes.c (init_optimization_passes): Add pass_build_cgraph_edges
2470 at the end of lowering passes.
2472 2007-01-06 Steven Bosscher <steven@gcc.gnu.org>
2474 * ifcvt.c (cond_move_convert_if_block): New function, code
2475 factored out from...
2476 (cond_move_process_if_block): ...here. Call the new function
2477 on the THEN and ELSE blocks.
2478 (merge_if_block): Do not copy global_live_at_end, merge_blocks
2479 already takes care of this.
2481 2007-01-05 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
2483 PR c/19978
2484 * tree.h (TREE_OVERFLOW_P): New.
2485 * c-typeck.c (parser_build_unary_op): Warn only if result
2486 overflowed and operands did not.
2487 (parser_build_binary_op): Likewise.
2488 (convert_for_assignment): Remove redundant overflow_warning.
2489 * c-common.c (overflow_warning): Don't check or set TREE_OVERFLOW.
2491 2007-01-05 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
2493 * c-typeck.c (store_init_value): Split over two lines to follow
2494 the GNU coding style.
2496 2007-01-05 Benjamin Kosnik <bkoz@redhat.com>
2498 * c-cppbuiltin.c (c_cpp_builtins): __GXX_EXPERIMENTAL_CPP0X__ to
2499 __GXX_EXPERIMENTAL_CXX0X__.
2500 * doc/cpp.texi: Same.
2502 2007-01-05 Richard Guenther <rguenther@suse.de>
2504 PR middle-end/27826
2505 * tree.c (get_narrower): Do not construct COMPONENT_REFs
2506 with mismatched types. Instead explicitly build a
2507 conversion NOP_EXPR.
2509 2007-01-05 Ian Lance Taylor <iant@google.com>
2511 * c-common.c (decl_with_nonnull_addr_p): New function.
2512 (c_common_truthvalue_conversion): Call it.
2513 * c-typeck.c (build_binary_op): Likewise.
2514 * c-common.h (decl_with_nonnull_addr_p): Declare.
2516 2007-01-05 Jakub Jelinek <jakub@redhat.com>
2518 PR c/30360
2519 * libgcc2.c (__divdc3): Compare c and d against 0.0 instead of
2520 denom against 0.0.
2522 2007-01-05 Joel Brobecker <brobecker@adacore.com>
2524 * doc/install.texi (Final install): Document the fact that
2525 the GNAT runtime should not be stripped.
2527 2007-01-04 Jan Hubicka <jh@suse.cz>
2529 * tree-inline.c (fold_marked_statements): Update operand caches
2530 and EH after folding
2532 2007-01-04 Ian Lance Taylor <iant@google.com>
2534 * c-common.c (check_function_nonnull): Whitespace fix.
2536 2007-01-04 Jan Hubicka <jh@suse.cz>
2538 * tree-optimize.c (execute_fixup_cfg): Correct previously mistakely
2539 comitted older version of patch.
2540 (pass_fixup_cfg): Add TODOs to verify flow and statements, dump
2541 function, celanup cfg and collect garbage.
2543 2007-01-04 Mike Stump <mrs@apple.com>
2545 * Makefile.in (mostlyclean): Don't remove libgcc anymore.
2546 (clean): Likewise.
2548 2007-01-04 Eric Christopher <echristo@apple.com>
2550 * libgcc2.c (__bswapsi2): Use SItype.
2551 (__bswapdi2): Use DItype.
2552 * libgcc2.h: Update for above.
2554 2007-01-04 Paul Brook <paul@codesourcery.com>
2556 * config/arm/arm.md (arm_mulsi3, thumb_mulsi3, mulsi3_compare0,
2557 mulsi_compare0_scratch, mulsi3addsi, mulsi3addsi_compare0,
2558 mulsi3addsi_compare0_scratch, mulsidi3adddi, mulsidi3,
2559 umulsidi3, umulsidi3adddi, smulsi3_highpart,
2560 umulsi3_highpart): Make conditional on !arm_arch6.
2561 (arm_mulsi3_v6, thumb_mulsi3_v6, mulsi3_compare0_v6,
2562 mulsi_compare0_scratch_v6, mulsi3addsi_v6, mulsi3addsi_compare0_v6,
2563 mulsi3addsi_compare0_scratch_v6, mulsidi3adddi_v6, mulsidi3_v6,
2564 umulsidi3_v6, umulsidi3adddi_v6, smulsi3_highpart_v6,
2565 umulsi3_highpart_v6): New insns.
2567 2007-01-04 Roger Sayle <roger@eyesopen.com>
2569 * fold-const.c (fold_convert): When casting an expression to void,
2570 fold_ignored_result may discover a GIMPLE_MODIFY_STMT which doesn't
2571 have a type. Instead of attempting to build a NOP_EXPR, return
2572 these "special" trees directly.
2574 2007-01-04 Joseph Myers <joseph@codesourcery.com>
2576 * config/rs6000/rs6000.c (rs6000_rtx_costs): Make adjustment for
2577 MULT inside MINUS as either argument. Use rs6000_cost->dmul -
2578 rs6000_cost->fp not 0 as adjustment for outer NEG.
2580 2007-01-04 Jan Hubicka <jh@suse.cz>
2582 * cgraph.c (cgraph_release_function_body): New function.
2583 (cgraph_remove_node): Use it.
2584 * cgraph.h (cgraph_release_function_body): Declare.
2585 * cgraphunit.c (cgraph_expand_function): Use it.
2586 * ipa.c (cgraph_remove_unreahchable_nodes): Use it.
2587 * tree-ssa.c (delete_tree_ssa): Allow to be called before aliasing
2588 is initialized and while compilation of other function is running.
2589 * tree-optimize.c (execute_free_cfg_annotations): Move code to clear
2590 statement CFG annotations from here to ...
2591 * tree-cfg.c (delete_tree_cfg_annotations): ... here.
2593 2007-01-04 Zdenek Dvorak <dvorakz@suse.cz>
2595 * cfgloop.h (enum li_flags): Make the constants powers of two.
2597 2007-01-04 Jan Hubicka <jh@suse.cz>
2599 * tree-inline.c (copy_bb): Insert new statements to statements_to_fold
2600 set.
2601 (fold_marked_statements): New function.
2602 (optimize_inline_calls, tree_function_versioning): Fold new statements.
2603 * tree-inline.h (copy_body_data): Add statements_to_fold.
2605 2007-01-03 Daniel Jacobowitz <dan@codesourcery.com>
2607 * config.gcc: Mention libgcc/config.host.
2608 * Makefile.in: Update comments mentioning libgcc.
2609 (LIBGCC, INSTALL_LIBGCC, GCC_PARTS, mklibgcc): Delete.
2610 (all.cross, start.encap, rest.encap, rest.cross): Update
2611 dependencies for libgcc move.
2612 (libgcc.mk, LIBGCC_DEPS, libgcov.a, libgcc.a, stmp-multilib)
2613 (clean-target, clean-target-libgcc): Delete.
2614 (srcdirify, GCC_EXTRA_PARTS): New macros.
2615 (libgcc-support, libgcc.mvars): New rules.
2616 (distclean): Remove mention of mklibgcc.
2617 (install): Don't reference INSTALL_LIBGCC.
2618 (install-common): Don't reference EXTRA_PARTS.
2619 (install-libgcc, install-multilib): Delete rules.
2620 * mklibgcc.in: Delete file.
2621 * doc/configfiles.texi: Don't mention mklibgcc.
2623 * config/i386/t-darwin (SHLIB_VERPFX): Delete (moved to libgcc).
2624 * config/i386/t-darwin64 (SHLIB_VERPFX): Likewise.
2625 * config/rs6000/t-darwin (SHLIB_VERPFX): Likewise.
2626 * config/rs6000/t-ppccomm (TARGET_LIBGCC2_CFLAGS, SHLIB_MAPFILES)
2627 (mklibgcc, ldblspecs): Likewise.
2629 * config/i386/t-nwld (libgcc.def, libc.def, libpcre.def)
2630 (posixpre.def): Use $(T).
2631 (SHLIB_EXT, SHLIB_NAME, SHLIB_SLIBDIR_QUAL, SHLIB_DEF, SHLIB_MAP)
2632 (SHLIB_SRC, SHLIB_INSTALL): Delete.
2633 (SHLIB_LINK): Make dummy.
2634 * config/t-slibgcc-darwin: Delete contents except for dummy SHLIB_LINK.
2636 * config/frv/t-linux (EXTRA_MULTILIB_PARTS): Clear.
2638 * config/alpha/t-crtfm: Use $(T) in rules for EXTRA_PARTS.
2639 * config/alpha/t-vms, config/alpha/t-vms64, config/fr30/t-fr30,
2640 config/i386/t-rtems-i386, config/ia64/t-ia64, config/rs6000/t-beos,
2641 config/rs6000/t-newas, config/sparc/t-elf: Likewise.
2643 * configure.ac (all_outputs): Remove mklibgcc.
2644 * configure: Regenerated.
2646 2007-01-03 Josh Conner <jconner@apple.com>
2648 PR middle-end/29683
2649 * calls.c (compute_argument_addresses): Set stack and stack_slot
2650 for partial args, too.
2651 (store_one_arg): Use locate.size.constant for the size when
2652 generating a save_area.
2654 2007-01-03 Robert Kennedy <jimbob@google.com>
2656 * tree-cfg.c (tree_merge_blocks): Release SSA_NAME phi results
2657 whose definitions are deleted due to basic block merging.
2659 2007-01-03 Paul Brook <paul@codesourcery.com>
2661 PR target/16634
2662 * config/arm/arm.c (output_return_instruction): Pop PC in interrupt
2663 functions.
2664 (use_return_insn): Return 0 for Thumb interrupt functions.
2665 (print_multi_reg): Add rfe argument for IRQ returns.
2666 (arm_output_epilogue): Pop interrupt return address directly into PC.
2667 (arm_expand_prologue): Only adjust IRQ return address in Arm mode.
2669 2007-01-03 Paul Brook <paul@codesourcery.com>
2671 Merge from sourcerygxx-4_1.
2672 * config/arm/thumb2.md: New file.
2673 * config/arm/elf.h (JUMP_TABLES_IN_TEXT_SECTION): Return True for
2674 Thumb-2.
2675 * config/arm/coff.h (JUMP_TABLES_IN_TEXT_SECTION): Ditto.
2676 * config/arm/aout.h (ASM_OUTPUT_ADDR_VEC_ELT): Add !Thumb-2 assertion.
2677 (ASM_OUTPUT_ADDR_DIFF_ELT): Output Thumb-2 jump tables.
2678 * config/arm/aof.h (ASM_OUTPUT_ADDR_DIFF_ELT): Output Thumb-2 jump
2679 tables.
2680 (ASM_OUTPUT_ADDR_VEC_ELT): Add !Thumb-2 assertion.
2681 * config/arm/ieee754-df.S: Use macros for Thumb-2/Unified asm
2682 comptibility.
2683 * config/arm/ieee754-sf.S: Ditto.
2684 * config/arm/arm.c (thumb_base_register_rtx_p): Rename...
2685 (thumb1_base_register_rtx_p): ... to this.
2686 (thumb_index_register_rtx_p): Rename...
2687 (thumb1_index_register_rtx_p): ... to this.
2688 (thumb_output_function_prologue): Rename...
2689 (thumb1_output_function_prologue): ... to this.
2690 (thumb_legitimate_address_p): Rename...
2691 (thumb1_legitimate_address_p): ... to this.
2692 (thumb_rtx_costs): Rename...
2693 (thumb1_rtx_costs): ... to this.
2694 (thumb_compute_save_reg_mask): Rename...
2695 (thumb1_compute_save_reg_mask): ... to this.
2696 (thumb_final_prescan_insn): Rename...
2697 (thumb1_final_prescan_insn): ... to this.
2698 (thumb_expand_epilogue): Rename...
2699 (thumb1_expand_epilogue): ... to this.
2700 (arm_unwind_emit_stm): Rename...
2701 (arm_unwind_emit_sequence): ... to this.
2702 (thumb2_legitimate_index_p, thumb2_legitimate_address_p,
2703 thumb1_compute_save_reg_mask, arm_dwarf_handle_frame_unspec,
2704 thumb2_index_mul_operand, output_move_vfp, arm_shift_nmem,
2705 arm_save_coproc_regs, thumb_set_frame_pointer, arm_print_condition,
2706 thumb2_final_prescan_insn, thumb2_asm_output_opcode, arm_output_shift,
2707 thumb2_output_casesi): New functions.
2708 (TARGET_DWARF_HANDLE_FRAME_UNSPEC): Define.
2709 (FL_THUMB2, FL_NOTM, FL_DIV, FL_FOR_ARCH6T2, FL_FOR_ARCH7,
2710 FL_FOR_ARCH7A, FL_FOR_ARCH7R, FL_FOR_ARCH7M, ARM_LSL_NAME,
2711 THUMB2_WORK_REGS): Define.
2712 (arm_arch_notm, arm_arch_thumb2, arm_arch_hwdiv, arm_condexec_count,
2713 arm_condexec_mask, arm_condexec_masklen)): New variables.
2714 (all_architectures): Add armv6t2, armv7, armv7a, armv7r and armv7m.
2715 (arm_override_options): Check new CPU capabilities.
2716 Set new architecture flag variables.
2717 (arm_isr_value): Handle v7m interrupt functions.
2718 (user_return_insn): Return 0 for v7m interrupt functions. Handle
2719 Thumb-2.
2720 (const_ok_for_arm): Handle Thumb-2 constants.
2721 (arm_gen_constant): Ditto. Use movw when available.
2722 (arm_function_ok_for_sibcall): Return false for v7m interrupt
2723 functions.
2724 (legitimize_pic_address, arm_call_tls_get_addr): Handle Thumb-2.
2725 (thumb_find_work_register, arm_load_pic_register,
2726 legitimize_tls_address, arm_address_cost, load_multiple_sequence,
2727 emit_ldm_seq, emit_stm_seq, arm_select_cc_mode, get_jump_table_size,
2728 print_multi_reg, output_mov_long_double_fpa_from_arm,
2729 output_mov_long_double_arm_from_fpa, output_mov_double_fpa_from_arm,
2730 output_mov_double_fpa_from_arm, output_move_double,
2731 arm_compute_save_reg_mask, arm_compute_save_reg0_reg12_mask,
2732 output_return_instruction, arm_output_function_prologue,
2733 arm_output_epilogue, arm_get_frame_offsets, arm_regno_class,
2734 arm_output_mi_thunk, thumb_set_return_address): Ditto.
2735 (arm_expand_prologue): Handle Thumb-2. Use arm_save_coproc_regs.
2736 (arm_coproc_mem_operand): Allow POST_INC/PRE_DEC.
2737 (arithmetic_instr, shift_op): Use arm_shift_nmem.
2738 (arm_print_operand): Use arm_print_condition. Handle '(', ')', '.',
2739 '!' and 'L'.
2740 (arm_final_prescan_insn): Use extract_constrain_insn_cached.
2741 (thumb_expand_prologue): Use thumb_set_frame_pointer.
2742 (arm_file_start): Output directive for unified syntax.
2743 (arm_unwind_emit_set): Handle stack alignment instruction.
2744 * config/arm/lib1funcs.asm: Remove default for __ARM_ARCH__.
2745 Add v6t2, v7, v7a, v7r and v7m.
2746 (RETLDM): Add Thumb-2 code.
2747 (do_it, shift1, do_push, do_pop, COND, THUMB_SYNTAX): New macros.
2748 * config/arm/arm.h (TARGET_CPU_CPP_BUILTINS): Define __thumb2__.
2749 (TARGET_THUMB1, TARGET_32BIT, TARGET_THUMB2, TARGET_DSP_MULTIPLY,
2750 TARGET_INT_SIMD, TARGET_UNIFIED_ASM, ARM_FT_STACKALIGN, IS_STACKALIGN,
2751 THUMB2_TRAMPOLINE_TEMPLATE, TRAMPOLINE_ADJUST_ADDRESS,
2752 ASM_OUTPUT_OPCODE, THUMB2_GO_IF_LEGITIMATE_ADDRESS,
2753 THUMB2_LEGITIMIZE_ADDRESS, CASE_VECTOR_PC_RELATIVE,
2754 CASE_VECTOR_SHORTEN_MODE, ADDR_VEC_ALIGN, ASM_OUTPUT_CASE_END,
2755 ADJUST_INSN_LENGTH): Define.
2756 (TARGET_REALLY_IWMMXT, TARGET_IWMMXT_ABI, CONDITIONAL_REGISTER_USAGE,
2757 STATIC_CHAIN_REGNUM, HARD_REGNO_NREGS, INDEX_REG_CLASS,
2758 BASE_REG_CLASS, MODE_BASE_REG_CLASS, SMALL_REGISTER_CLASSES,
2759 PREFERRED_RELOAD_CLASS, SECONDARY_OUTPUT_RELOAD_CLASS,
2760 SECONDARY_INPUT_RELOAD_CLASS, LIBCALL_VALUE, FUNCTION_VALUE_REGNO_P,
2761 TRAMPOLINE_SIZE, INITIALIZE_TRAMPOLINE, HAVE_PRE_INCREMENT,
2762 HAVE_POST_DECREMENT, HAVE_PRE_DECREMENT, HAVE_PRE_MODIFY_DISP,
2763 HAVE_POST_MODIFY_DISP, HAVE_PRE_MODIFY_REG, HAVE_POST_MODIFY_REG,
2764 REGNO_MODE_OK_FOR_BASE_P, LEGITIMATE_CONSTANT_P,
2765 REG_MODE_OK_FOR_BASE_P, REG_OK_FOR_INDEX_P, GO_IF_LEGITIMATE_ADDRESS,
2766 LEGITIMIZE_ADDRESS, THUMB2_LEGITIMIZE_ADDRESS,
2767 GO_IF_MODE_DEPENDENT_ADDRESS, MEMORY_MOVE_COST, BRANCH_COST,
2768 ASM_APP_OFF, ASM_OUTPUT_CASE_LABEL, ARM_DECLARE_FUNCTION_NAME,
2769 FINAL_PRESCAN_INSN, PRINT_OPERAND_PUNCT_VALID_P,
2770 PRINT_OPERAND_ADDRESS): Adjust for Thumb-2.
2771 (arm_arch_notm, arm_arch_thumb2, arm_arch_hwdiv): New declarations.
2772 * config/arm/arm-cores.def: Add arm1156t2-s, cortex-a8, cortex-r4 and
2773 cortex-m3.
2774 * config/arm/arm-tune.md: Regenerate.
2775 * config/arm/arm-protos.h: Update prototypes.
2776 * config/arm/vfp.md: Enable patterns for Thumb-2.
2777 (arm_movsi_vfp): Add movw alternative. Use output_move_vfp.
2778 (arm_movdi_vfp, movsf_vfp, movdf_vfp): Use output_move_vfp.
2779 (thumb2_movsi_vfp, thumb2_movdi_vfp, thumb2_movsf_vfp,
2780 thumb2_movdf_vfp, thumb2_movsfcc_vfp, thumb2_movdfcc_vfp): New.
2781 * config/arm/libunwind.S: Add Thumb-2 code.
2782 * config/arm/constraints.md: Update include Thumb-2.
2783 * config/arm/ieee754-sf.S: Add Thumb-2/Unified asm support.
2784 * config/arm/ieee754-df.S: Ditto.
2785 * config/arm/bpabi.S: Ditto.
2786 * config/arm/t-arm (MD_INCLUDES): Add thumb2.md.
2787 * config/arm/predicates.md (low_register_operand,
2788 low_reg_or_int_operand, thumb_16bit_operator): New.
2789 (thumb_cmp_operand, thumb_cmpneg_operand): Rename...
2790 (thumb1_cmp_operand, thumb1_cmpneg_operand): ... to this.
2791 * config/arm/t-arm-elf: Add armv7 multilib.
2792 * config/arm/arm.md: Update patterns for Thumb-2 and Unified asm.
2793 Include thumb2.md.
2794 (UNSPEC_STACK_ALIGN, ce_count): New.
2795 (arm_incscc, arm_decscc, arm_umaxsi3, arm_uminsi3,
2796 arm_zero_extendsidi2, arm_zero_extendqidi2): New
2797 insns/expanders.
2798 * config/arm/fpa.md: Update patterns for Thumb-2 and Unified asm.
2799 (thumb2_movsf_fpa, thumb2_movdf_fpa, thumb2_movxf_fpa,
2800 thumb2_movsfcc_fpa, thumb2_movdfcc_fpa): New insns.
2801 * config/arm/cirrus.md: Update patterns for Thumb-2 and Unified asm.
2802 (cirrus_thumb2_movdi, cirrus_thumb2_movsi_insn,
2803 thumb2_cirrus_movsf_hard_insn, thumb2_cirrus_movdf_hard_insn): New
2804 insns.
2805 * doc/extend.texi: Document ARMv7-M interrupt functions.
2806 * doc/invoke.texi: Document Thumb-2 new cores+architectures.
2808 2007-01-03 Jakub Jelinek <jakub@redhat.com>
2810 * unwind-dw2.c (SIGNAL_FRAME_BIT, EXTENDED_CONTEXT_BIT): Define.
2811 (struct _Unwind_Context): Rename args_size to flags, remove
2812 signal_frame field, add a new args_size field and version field.
2813 (_Unwind_IsSignalFrame, _Unwind_SetSignalFrame,
2814 _Unwind_IsExtendedContext): New inline functions.
2815 (_Unwind_GetGR, _Unwind_SetGR, _Unwind_GetGRPtr, _Unwind_SetGRPtr):
2816 Assume by_value array is only present if _Unwind_IsExtendedContext.
2817 (_Unwind_GetIPInfo, execute_cfa_program, uw_frame_state_for): Use
2818 _Unwind_IsSignalFrame.
2819 (__frame_state_for): Initialize context.flags to EXTENDED_CONTEXT_BIT.
2820 (uw_update_context_1): Use _Unwind_SetSignalFrame.
2821 (uw_init_context_1): Initialize context->flags to
2822 EXTENDED_CONTEXT_BIT.
2823 * config/rs6000/linux-unwind.h (frob_update_context): Use
2824 _Unwind_SetSignalFrame.
2826 2007-01-03 Andrew Pinski <andrew_pinski@playstation.sony.com>
2828 PR middle-end/30353
2829 * gimplify.c (gimplify_modify_expr_complex_part): Move below
2830 tree_to_gimple_tuple. Call tree_to_gimple_tuple when we need
2831 the value.
2833 2007-01-03 Kazu Hirata <kazu@codesourcery.com>
2835 * config/i386/i386.h (NON_STACK_REG_P, REGNO_OK_FOR_SIREG_P,
2836 REGNO_OK_FOR_DIREG_P, REWRITE_ADDRESS, ASM_OPERAND_LETTER,
2837 RET, AT_SP): Remove.
2838 * config/i386/i386.md (*sse_prologue_save_insn): Use return
2839 instead of RET.
2841 * alias.c (init_alias_analysis): Use VEC_safe_grow_cleared.
2842 * cfgbuild.c (find_basic_blocks): Likewise.
2843 * cfgrtl.c (rtl_create_basic_block): Likewise.
2844 * function.c (temp_slots_at_level): Likewise.
2845 * reg-stack.c (stack_regs_mentioned): Likewise.
2846 * regclass.c (allocate_reg_info): Likewise.
2847 * tree-cfg.c (init_empty_tree_cfg, build_tree_cfg, create_bb,
2848 set_bb_for_stmt, move_block_to_fn): Likewise.
2849 * tree-complex.c (tree_lower_complex): Likewise.
2850 * vec.h (VEC_safe_grow_cleared): New.
2852 * cgraphunit.c, tree-ssa-alias.c: Fix comment typos.
2854 2007-01-03 Zdenek Dvorak <dvorakz@suse.cz>
2856 * loop-unswitch.c (unswitch_loop): Pass probabilities to loopify.
2857 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Pass probabilities
2858 to loop_version.
2859 * cfgloopmanip.c (scale_loop_frequencies): Export.
2860 (loopify): Scale the frequencies by prescribed coefficients.
2861 (set_zero_probability): New function.
2862 (duplicate_loop_to_header_edge): Improve updating of frequencies.
2863 (lv_adjust_loop_entry_edge, loop_version): Set probabilities
2864 and frequencies according to arguments.
2865 * tree-ssa-loop-manip.c (tree_unroll_loop): Set probabilities
2866 correctly.
2867 * cfg.c (scale_bbs_frequencies_int): Allow scaling the frequencies up.
2868 * modulo-sched.c (sms_schedule): Set probabilities for entering
2869 versioned loop correctly.
2870 * tree-vect-transform.c (vect_transform_loop): Ditto.
2871 * cfgloop.h (loopify, loop_version): Declaration changed.
2872 (scale_loop_frequencies): Declared.
2874 2007-01-02 Jan Hubicka <jh@suse.cz>
2876 * cgraph.c: Include tree-flow.h
2877 (cgraph_add_new-function): Handle IPA_SSA mode; execute
2878 early_local_passes.
2879 * cgraph.h (enum cgraph_state): Add CGRAPH_STATE_IPA_SSA.
2880 * tree-pass.h (pass_all_early_optimizations): Declare.
2881 * cgraphunit.c (cgraph_process_new_functions): Add IPA_SSA; execute
2882 early_local_passes.
2883 (cgraph_analyze_function): Do early_local_passes.
2884 * tree-mudflap.c (mf_decl_cache_locals, mf_build_check_statement_for):
2885 Do not add referenced vars.
2886 * tree-optimize.c (gate_all_optimizations): Do not execute when not in
2887 SSA form.
2888 (gate_all_early_local_passes): New gate.
2889 (pass_early_local_passes): Use new gate.
2890 (execute_early_local_optimizations): New functions.
2891 (gate_all_early_optimizations): New gate.
2892 (pass_all_early_optimizations): New pass.
2893 (execute_free_datastructures): Free SSA only when initialized.
2894 (gate_init_datastructures): Init only when optimizing.
2895 (tree_lowering_passes): Do early local passes when called late.
2896 * tree-profile.c (do_tree_profiling): Don't profile functions added
2897 late.
2898 (do_early_tree_profiling, pass_early_tree_profile): Kill.
2899 * tree-cfg.c (update_modified_stmts): Do not update when operands are
2900 not active.
2901 * passes.c (init_optimizations_passes): Reorder so we go into SSA
2902 during early_local_passes.
2903 * Makefile.in (cgraph.o): Add dependency on tree-flow.h.
2906 2007-01-02 Carlos O'Donell <carlos@codesourcery.com>
2908 * Makefile.in: Update copyright year.
2910 2007-01-02 Carlos O'Donell <carlos@codesourcery.com>
2912 * Makefile.in: Export GCC_EXEC_PREFIX before calling $(RUNTEST)
2913 in $(lang_checks) and check-consistency targets.
2915 2007-01-02 Jan Hubicka <jh@suse.cz>
2917 * tree-mudflap.c (mf_decl_cache_locals, mf_build_check_statement_for):
2918 Do not add referenced vars.
2919 * tree-cfg.c (update_modified_stmts): Do not update when SSA operands
2920 are not active.
2921 * passes.c (init_optimization_passes): Put mudflap_2 after
2922 free_datastructures.
2924 2007-01-02 Jan Hubicka <jh@suse.cz>
2926 * tree-optimize (execute_fixup_cfg): Set after_inlining flag.
2927 Set NOTHROW flag on call statements proved to be nothrow.
2928 Update statement of local calls so new pure/const functions are
2929 updated. Update_ssa when in ssa form. Mark PHI nodes of nonlocal
2930 goto receivers.
2931 (tree_rest_of_compilation): Register hooks and initialize bitmap
2932 early. Do not set after_inlining flag.
2934 2007-01-02 Steve Ellcey <sje@cup.hp.com>
2936 * sbitmap.c (HOST_BITS_PER_LONG_LONG): Change to
2937 HOST_BITS_PER_LONGLONG
2939 2007-01-02 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
2941 PR c/19977
2942 * c-typeck.c (store_init_value): Don't emit pedantic overflow
2943 warning for non-static initializers.
2945 2007-01-02 Steven Bosscher <steven@gcc.gnu.org>
2947 * config/alpha/alpha.md, arm/arm.c, darwin.c, frv/frv.md,
2948 m32r/m32r.c, m32r/m32r.c, mn10300/mn10300.md, pa/pa.c,
2949 rs6000/rs6000.c, s390/s390.md, sh/sh.md, sparc/sparc.c:
2950 Always use set_unique_reg_note to add REG_EQUAL notes.
2952 2007-01-02 Kazu Hirata <kazu@codesourcery.com>
2954 Revert:
2955 2007-01-02 Kazu Hirata <kazu@codesourcery.com>
2957 * alias.c (init_alias_analysis): Use VEC_safe_grow_cleared.
2958 * cfgbuild.c (find_basic_blocks): Likewise.
2959 * cfgrtl.c (rtl_create_basic_block): Likewise.
2960 * function.c (temp_slots_at_level): Likewise.
2961 * reg-stack.c (stack_regs_mentioned): Likewise.
2962 * regclass.c (allocate_reg_info): Likewise.
2963 * tree-cfg.c (init_empty_tree_cfg, build_tree_cfg, create_bb,
2964 set_bb_for_stmt, move_block_to_fn): Likewise.
2965 * tree-complex.c (tree_lower_complex): Likewise.
2966 * vec.h (VEC_safe_grow_cleared): New.
2968 2007-01-02 Ian Lance Taylor <iant@google.com>
2970 * c-common.c (c_common_truthvalue_conversion): When warning about
2971 using an assignment as a truth value, set TREE_NO_WARNING.
2973 2007-01-02 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
2975 PR middle-end/7651
2976 * c.opt (Wold-style-declaration): New.
2977 * doc/invoke.texi (C-only Warning Options): New.
2978 (Wold-style-declaration): Document it.
2979 (Wextra): Enabled by -Wextra.
2980 * c-opts.c (c_common_post_options): Enabled by -Wextra.
2981 * c-decl.c (declspecs_add_scspec): Replace -Wextra with
2982 -Wold-style-declaration.
2984 2007-01-02 Kazu Hirata <kazu@codesourcery.com>
2986 * alias.c (init_alias_analysis): Use VEC_safe_grow_cleared.
2987 * cfgbuild.c (find_basic_blocks): Likewise.
2988 * cfgrtl.c (rtl_create_basic_block): Likewise.
2989 * function.c (temp_slots_at_level): Likewise.
2990 * reg-stack.c (stack_regs_mentioned): Likewise.
2991 * regclass.c (allocate_reg_info): Likewise.
2992 * tree-cfg.c (init_empty_tree_cfg, build_tree_cfg, create_bb,
2993 set_bb_for_stmt, move_block_to_fn): Likewise.
2994 * tree-complex.c (tree_lower_complex): Likewise.
2995 * vec.h (VEC_safe_grow_cleared): New.
2997 2007-01-02 Douglas Gregor <doug.gregor@gmail.com>
2999 * c-common.c(c_common_nodes_and_builtins): Since variants of
3000 void_type_node get built before it is given a name, we need to
3001 give those variants the name, too.
3002 (complete_array_type): We need to work with the canonical main
3003 type of the array, from which we will build the qualified version.
3004 * params.def (PARAM_VERIFY_CANONICAL_TYPES): New.
3005 * print-tree.c (print_node): Display canonical type information
3006 for each type.
3007 * stor-layout.c (layout_type): When we don't know the
3008 alignment of a type for which we're building an array, we end up
3009 guessing wrong, so make the type require structural equality.
3010 * tree.c (make_node_stat): When we build a new type, it is its
3011 own canonical type.
3012 (build_type_attribute_qual_variant): When building an attribute
3013 variant, its canonical type is the non-attribute variant. However,
3014 if the attributes are target-dependent and they differ, we need to
3015 use structural equality checks for this type.
3016 (build_qualified_type): A qualified type is not equivalent to its
3017 unqualified variant; set the canonical type appropriately.
3018 (build_distinct_type_copy): When building a distinct type from
3019 another type, the new type is its own canonical type.
3020 (build_variant_type_copy): When building a new type variant, we
3021 assume that it is equivalent to the original type.
3022 (build_pointer_type_for_mode): When building a pointer type, also
3023 build a canonical type pointer.
3024 (build_reference_type_for_mode): When building a reference type,
3025 also build a canonical type reference.
3026 (build_index_type): When we can't hash an index type (e.g.,
3027 because its maximum value is negative), the index type requires
3028 structural equality tests.
3029 (build_array_type): Build the canonical form of an array type.
3030 (build_function_type): Function types require structural equality,
3031 because they contain default arguments, attributes, etc.
3032 (build_method_type_directly): Ditto for method types.
3033 (build_offset_type): Build the canonical offset type.
3034 (build_complex_type): Build the canonical vector type.
3035 (make_vector_type): Build the canonical vector type.
3036 * tree.h (TYPE_CANONICAL): New.
3037 (TYPE_STRUCTURAL_EQUALITY_P): New.
3038 (SET_TYPE_STRUCTURAL_EQUALITY): New.
3039 (struct tree_type): Added "canonical" field.
3040 * params.h (VERIFY_CANONICAL_TYPES): New.
3041 * doc/c-tree.texi (TYPE_CANONICAL): Document.
3042 (TYPE_STRUCTURAL_EQUALITY_P): Document.
3043 (SET_TYPE_STRUCTURAL_EQUALITY): Document.
3044 * doc/invoke.texi (verify-canonical-types): Document --param
3045 parameter for verifying canonical types.
3047 2007-01-02 Joseph Myers <joseph@codesourcery.com>
3049 * config.gcc (powerpc-*-eabispe*, powerpc-*-eabisimaltivec*,
3050 powerpc-*-eabisim*, powerpc-*-eabialtivec*, powerpc-*-eabi*,
3051 powerpc-*-rtems*, powerpc-wrs-vxworks, powerpc-wrs-vxworksae,
3052 powerpcle-*-eabisim*, powerpcle-*-eabi*): Add rs6000/e500.h to
3053 tm_file.
3054 * config/rs6000/e500.h: New.
3055 * config/rs6000/eabi.h (TARGET_SPE_ABI, TARGET_SPE, TARGET_E500,
3056 TARGET_ISEL, TARGET_FPRS, TARGET_E500_SINGLE, TARGET_E500_DOUBLE):
3057 Remove.
3058 * config/rs6000/linuxspe.h (TARGET_SPE_ABI, TARGET_SPE,
3059 TARGET_E500, TARGET_ISEL, TARGET_FPRS, TARGET_E500_SINGLE,
3060 TARGET_E500_DOUBLE): Remove.
3061 * config/rs6000/vxworks.h (TARGET_SPE_ABI, TARGET_SPE,
3062 TARGET_E500, TARGET_ISEL, TARGET_FPRS): Remove.
3063 * config/rs6000/rs6000.h (CHECK_E500_OPTIONS): Define.
3064 * config/rs6000/rs6000.c (rs6000_override_options): Use
3065 CHECK_E500_OPTIONS.
3067 2007-01-02 Joseph Myers <joseph@codesourcery.com>
3069 * config/rs6000/rs6000.c (print_operand): Check (TARGET_SPE ||
3070 TARGET_E500_DOUBLE), not TARGET_E500, for %y.
3071 (rs6000_generate_compare, rs6000_emit_sCOND, output_cbranch,
3072 rs6000_emit_cmove): Don't check TARGET_E500.
3073 * config/rs6000/rs6000.md (bunordered, bordered, sunordered,
3074 sordered): Don't check TARGET_E500.
3076 2007-01-01 Eric Christopher <echristo@apple.com>
3078 * config/mips/mips.c (mips_regno_mode_ok_for_base_p): Use
3079 HARD_REGISTER_NUM_P.
3081 2007-01-01 Roger Sayle <roger@eyesopen.com>
3083 * fold-const.c (fold_binary) <EQ_EXPR>: Fold "(X^C1) eq/ne C2" into
3084 "X eq/ne (C1^C2)". Fold "(X^Z) eq/ne (Y^Z)" as "X eq/ne Y" when Z
3085 has no side-effects. Fold "(X^C1) eq/ne (Y^C2)" as "(X^(C1^C2))
3086 eq/ne Y".
3088 2007-01-01 Mike Stump <mrs@apple.com>
3090 * configure.ac: Remove support for building with Apple's gcc-3.1.
3092 2007-01-02 Joseph Myers <joseph@codesourcery.com>
3094 PR middle-end/30311
3095 * caller-save.c (add_stored_regs): Only handle SUBREGs if inner
3096 REG is a hard register. Do not modify REG before calling
3097 subreg_nregs.
3098 * rtlanal.c (subreg_get_info): Don't assert size of XMODE is a
3099 multiple of the size of YMODE for certain lowpart cases.
3101 2007-01-01 Andrew Pinski <pinskia@gmail.com>
3103 PR middle-end/30253
3104 * gimplify (voidify_wrapper_expr): Update for
3105 GIMPLE_MODIFY_STMT.
3107 2007-01-01 Andreas Schwab <schwab@suse.de>
3109 PR target/29166
3110 * config/ia64/ia64.c (ia64_compute_frame_size): Account space for
3111 save of BR0 in extra_spill_size instead of spill_size.
3112 (ia64_expand_prologue): Save BR0 outside of the gr/br/fr spill
3113 area.
3114 (ia64_expand_epilogue): Restore BR0 from its new location.
3116 2007-01-01 Andrew Pinski <pinskia@gmail.com>
3118 * gimplify.c (gimplify_init_constructor <case VECTOR_TYPE>):
3119 Use a temporary variable if the left hand side is not a gimple
3120 register.
3122 2007-01-01 Andrew Pinski <pinskia@gmail.com>
3124 * gimplify.c (gimplify_return_expr): Make the temporary variable
3125 for the return expression, a gimple register variable.
3127 2007-01-01 Jan Hubicka <jh@suse.cz>
3129 * emit-rtl.c (emit_copy_of_insn_after): Do not call copy_insn_1 for
3130 INSN_LIST.
3132 2007-01-01 Mike Stump <mrs@apple.com>
3134 * configure.ac (HAVE_GAS_LITERAL16): Add autoconf check for
3135 .literal16.
3136 * config/darwin.c (machopic_select_rtx_section): Use
3137 HAVE_GAS_LITERAL16.
3138 (darwin_mergeable_constant_section): Likewise.
3139 * configure: Regenerate.
3140 * config.in: Regenerate.
3142 2007-01-01 Jan Hubicka <jh@suse.cz>
3143 Andrew Pinski <pinskia@gmail.com>
3145 * cgraphunit.c (cgraph_optimize): Call cgraph_add_new_functions
3146 before starting IPA passes.