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