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