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