1 2007-01-22 Andreas Schwab <schwab@suse.de>
3 * config/m68k/m68k.h: Fix comment.
5 2007-01-22 Jan Hubicka <jh@suse.cz>
7 * passes.c (init_optimization_passes): Do not rerun
8 pass_early_warn_uninitialized.
10 2007-01-22 Richard Guenther <rguenther@suse.de>
12 PR tree-optimization/30038
13 * tree-ssa-math-opts.c (maybe_record_sincos): New static helper
15 (execute_cse_sincos_1): Likewise.
16 (execute_cse_sincos): Likewise.
17 (gate_cse_sincos): Likewise.
18 (pass_cse_sincos): New pass CSEing sin() and cos() calls using
19 the cexpi() canonicalization of sincos().
20 * tree-pass.h (pass_cse_sincos): Declare.
21 * passes.c (init_optimization_passes): New pass pas_cse_sincos.
23 2007-01-21 Eric Botcazou <ebotcazou@libertysurf.fr>
25 PR rtl-optimization/29329
26 * combine.c (replaced_rhs_insn): Rename to i2mod.
27 (replaced_rhs_value): Rename to i2mod_new_rhs.
28 (i2mod_old_rhs): New global variable.
29 (combine_instructions): Adjust for above change. Save a copy of
30 the old RHS into i2mod_old_rhs when the contents of a REG_EQUAL
31 note are substituted in the second instruction.
32 (distribute_notes) <REG_DEAD>: Adjust for above change. Do not
33 ditch the note if it pertains to the second eliminated register
34 and this register is mentioned in i2mod_old_rhs.
37 2006-09-12 Eric Botcazou <ebotcazou@libertysurf.fr>
39 * combine.c (distribute_notes) <REG_DEAD>: Do not consider SETs past
40 the insn to which the note was originally attached.
42 2007-01-21 Jan Hubicka <jh@suse.cz>
44 * ipa-inline.c (inlining_mode): Comment, move up.
45 (cgraph_decide_inlining_incrementally): Do not perform inlining itself;
46 fix handling of flattening of self recursive functions.
47 (cgraph_find_cycles): Remove.
48 (cgraph_flatten_node): Remove.
49 (cgraph_decide_inlining): Use incremental inliner to handle flattening.
50 (try_inline): New function.
51 (cgraph_early_inlining): Update call of
52 cgraph_decide_inlining_incrementally. Apply inlining here.
53 (apply_inline): Update call of cgraph_decide_inlining_incrementally.
55 2007-01-21 Dirk Mueller <dmueller@suse.de>
58 * tree-vrp.c (check_array_bounds): do not warn
59 about ADDR_EXPR's of ARRAY_REF's which are immediately
60 used in binary expressions.
62 2007-01-21 Ira Rosen <irar@il.ibm.com>
64 * tree-vectorizer.h (struct _stmt_vec_info): Add new field
65 read_write_dep and macros for its access.
66 * tree-vectorizer.c (new_stmt_vec_info): Initialize the new field.
67 * tree-vect-analyze.c (vect_analyze_data_ref_dependence): Remove
68 argument, call vect_check_interleaving for every independent pair of
69 data-refs. Mark loads that access the same memory location as a store
71 (vect_check_dependences): Remove.
72 (vect_analyze_data_ref_dependences): Remove vect_check_dependences
73 call, fix the call to vect_analyze_data_ref_dependence.
74 (vect_analyze_data_ref_access): For statements that access the same
75 data-ref, check that they are not stores; for loads, check that there
76 is no store that access the same location.
78 2007-01-20 Roger Sayle <roger@eyesopen.com>
79 Joseph Myers <joseph@codesourcery.com>
81 * doc/invoke.texi (-fdump-rtl-combine): Fix under/overfull hbox.
82 (-fdump-rtl-stack): Likewise.
83 (-fno-signed-zeros): Use @minus{} for a minus sign. Correct typo.
84 (-mcheck-zero-division, -mcpu): Fix under/overfull hbox.
85 (-mpt-fixed): Use @minus{} for minus sign.
86 (Using Precompiled Headers): Fix under/overfull hbox.
88 2007-01-20 Jan Hubicka <jh@suse.cz>
90 * tree-flow.h (struct stmt_ann_d): Move references_memory to proper
91 place within annotation.
93 2007-01-20 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
95 * pa.c (output_move_double): Change array size of xoperands to 4.
97 2007-01-20 Jan Hubicka <jh@suse.cz>
99 * tree-tailcall.c (adjust_return_value): Do not use RESULT_DECL
102 2007-01-19 Ian Lance Taylor <iant@google.com>
104 * expmed.c (expand_divmod) [TRUNC_MOD_EXPR, TRUNC_DIV_EXPR]: Cast
105 constant to unsigned HOST_WIDE_INT before negating.
107 2007-01-19 Ian Lance Taylor <iant@google.com>
109 * tree-ssa-operands.h (struct vuse_vec_d): Change num_vuse field
111 (VUSE_VECT_ELEMENT) [ENABLE_CHECKING]: Use unsigned comparison.
112 (VUSE_ELEMENT_PTR) [ENABLE_CHECKING]: Likewise.
113 (SET_VUSE_VECT_ELEMENT) [ENABLE_CHECKING]: Likewise.
114 (SET_VUSE_ELEMENT_VAR) [ENABLE_CHECKING]: Likewise.
115 (SET_VUSE_ELEMENT_PTR) [ENABLE_CHECKING]: Likewise.
116 (realloc_vdef, realloc_vuse): Change second parameter to
118 (ssa_operand_iterator_d): Change vuse_index and mayuse_index
120 * tree-ssa-operands.c (realloc_vop): Change num_elem parameter to
121 unsigned. Change x and lim locals to unsigned.
122 (realloc_vdef, realloc_vuse): Change num_elem parameter to
124 (finalize_ssa_vuse_ops): Change old_i local to unsigned.
125 (copy_virtual_operands): Change i and n locals to unsigned.
127 2007-01-19 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
130 * builtins.c (fold_builtin_1): Handle builtin fdim.
132 2007-01-20 Jan Hubicka <jh@suse.cz>
134 * tree-ssa.c (init_tree_ssa): Do not call init_alias_heapvars.
135 * tree-ssa-structalias.c (compute_points_to_sets): Do call
137 (init_alias_heapvars): Initialize only when not already initialized.
138 (delete_alias_heapvars): Set heapvar_for_stmt to NULL.
140 2007-01-19 Roger Sayle <roger@eyesopen.com>
142 * common.opt (fsigned-zeros): New command line option.
143 * flags.h (HONOR_SIGNED_ZEROS): Control via flag_signed_zeros instead
144 of flag_unsafe_math_optimizations.
145 * opts.c (set_fast_math_flags): The -ffast-math command line option
146 implies -fno-signed-zeros.
147 (fast_math_flags_set_p): Likewise.
149 * doc/invoke.texi: Document new -fno-signed-zeros option, and update
150 the documentation of -ffast-math appropriately. Wrap long lines.
152 2007-01-19 Steve Ellcey <sje@cup.hp.com>
154 * system.h (ASM_MAKE_LABEL_LINKONCE): Poison.
155 * varasm.c (globalize_decl): Remove ASM_MAKE_LABEL_LINKONCE ifdef.
157 2007-01-19 Tomas Bily <tbily@suse.cz>
159 * cgraphunit.c (cgraph_finalize_function): Updating of pid
161 (tree_init_ic_make_global_vars): New function
162 (tree_init_edge_profiler): call of tree_init_ic_make_global_vars
163 (tree_gen_ic_profiler): New function
164 (tree_gen_ic_func_profiler): New function
165 (tree_profiling): Added calling of tree_gen_ic_func_profiler
166 (tree_profile_hooks): Added hook for indirec/virtual calls
167 * value-prof.c (tree_find_values_to_profile): New case for
169 (tree_values_to_profile): Call for determining indirect/virtual
171 (tree_indirect_call_to_profile): New function
172 (tree_ic_transform): New function
173 (tree_ic): New function
174 (find_func_by_pid): New function
175 (init_pid_map): New function
176 (tree_value_profile_transformations): Added check for
177 indirect/virtual call transformation
178 * value-prof.h (enum hist_type): New counter type for
179 indirect/virtual calls
180 (profile_hooks): Added new hook for profiling indirect/virtual
182 * profile.c (instrument_values): New case for indirect/virtual
184 * gcov-io.h (GCOV_LAST_VALUE_COUNTER): Changed to 6
185 (GCOV_COUNTER_V_INDIR): New counter type
186 (GCOV_COUNTER_NAMES): New name of counter "indirect" added
187 (GCOV_MERGE_FUNCTIONS): New merge function for indirect/virtual
189 * cgraph.c: Definition of cgraph_max_pid
190 (cgraph_create_node): Default init of pid attribute
191 * cgraph.h: Declaration of cgraph_max_pid
192 (struct cgraph_node): Added pid attribute
193 * libgcov.c (__gcov_indirect_call_profiler): New function
194 (__gcov_one_value_profiler_body): New function
195 (__gcov_one_value_profiler): Body was moved to
196 __gcov_one_value_profiler_body and calls it
198 2007-01-19 Basile Starynkevitch <basile@starynkevitch.net>
200 * doc/gty.texi (Options): Document the mark_hook option to GTY.
201 * gengtype.c (write_types_data, write_func_for_structure,
202 write_types, ggc_wtd, pch_wtd): Add skip_hooks to
203 write_types_data, ggc_wtd, pch_wtd for processing mark_hook.
204 (walk_type, write_func_for_structure): Generate the mark_hook if
207 2007-01-19 Jan Hubicka <jh@suse.cz>
209 * ipa-inline.c (cgraph_decide_inlining_incrementally): Instead of
210 'early' argument take inlining mode argument specifying whether to
211 inline for size/speeed or all functions; add support for flattening;
213 (cgraph_early_inlining): Update call of decide_inlining_incrementally.
215 2007-01-19 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
218 * toplev.c (warn_deprecated_use): Use %qD instead of %qs to print
219 the name of the declared identifier.
221 2007-01-19 Dirk Mueller <dmueller@suse.de>
223 * config/i386.h (CONDITIONAL_REGISTER_USAGE): Store
224 result of PIC_OFFSET_TABLE_REGNUM in temporary variable to avoid
225 duplicate evaluation.
227 2007-01-19 Uros Bizjak <ubizjak@gmail.com>
229 * config/i386/i386.md (acos<mode>2): Rename from acossf2 and acosdf2.
230 Macroize expander using X87MODEF12 mode macro. Extend operand 1
231 to XFMode, use acosxf2 and truncate result to requested mode.
232 Use SSE_FLOAT_MODE_P to disable patterns for SSE math.
233 (asin<mode>2): Similarly, with asin expanders.
234 (*fscalexf4_i387): Rename from *fscalexf4.
235 (expNcorexf3): New expander.
236 (expxf2, exp10xf2, exp2xf2): Use expNcorexf3 expander.
237 (exp<mode>2): Rename from expsf2 and expdf2. Macroize expander using
238 X87MODEF12 mode macro. Extend operand 1 to XFMode, use expxf2 and
239 truncate result to requested mode. Use SSE_FLOAT_MODE_P to disable
240 patterns for SSE math.
241 (exp10<mode>2): Similarly, with exp10 expanders.
242 (exp2<mode>2): Similarly, with exp2 expanders.
243 (expm1<mode>2): Similarly, with expm1 expanders.
244 (ldexp<mode>3): Similarly, with ldexp expanders.
245 (log<mode>2, log10<mode>2, log2<mode>2, log1p<mode>2, logb<mode>2):
246 Use gen_truncxf<mode>2_i387_noop to truncate result.
248 2007-01-19 Richard Sandiford <richard@codesourcery.com>
250 * config/m68k/m68k.h (PREFERRED_STACK_BOUNDARY): Define to 32
251 for ColdFire targets.
253 2007-01-19 Nathan Sidwell <nathan@codesourcery.com>
254 Richard Sandiford <richard@codesourcery.com>
256 * config/m68k/m68k.h (M68K_STATIC_CHAIN_REG_NAME): New macro.
257 (INITIALIZE_TRAMPOLINE): Use STATIC_CHAIN_REGNUM.
258 (__transfer_from_trampoline): Use M68K_STATIC_CHAIN_REG_NAME.
259 * config/m68k/m68kelf.h (STATIC_CHAIN_REGNUM): Override.
260 (M68K_STATIC_CHAIN_REG_NAME): Likewise.
261 * config/m68k/netbsd-elf.h (M68K_STATIC_CHAIN_REG_NAME): Likewise.
263 2007-01-19 Richard Sandiford <richard@codesourcery.com>
265 * config/m68k/m68k.md (adddi_dilshr32): Rename to...
266 (*adddi_dilshr32): ...this. Fix formatting. Remove commented-out
267 non-canonical pattern. Restrict to !TARGET_COLDFIRE.
268 (*adddi_dilshr32_cf): New pattern.
269 (adddi3, subdi3): Remove first alternatives.
271 2007-01-19 Richard Sandiford <richard@codesourcery.com>
273 * config/m68k/m68k.c (notice_update_cc): If an SFmode move is
274 implemented using move.l, do not use its cc result for floating-point
277 2007-01-19 Richard Sandiford <richard@codesourcery.com>
279 * config/m68k/m68k.h (EXTRA_CONSTRAINT): Stop the 'T' constraint
280 from accepting 's' constraints if flag_pic.
282 2007-01-19 Richard Sandiford <richard@codesourcery.com>
284 * config/m68k/m68k.md (bordered, bunordered, buneq, bunge, bungt)
285 (bunle, bunlt, bltgt, bordered_rev, bunordered_rev, buneq_rev)
286 (bunge_rev, bungt_rev, bunle_rev, bunlt_rev, bltgt_rev): Change
287 condition from TARGET_68881 to TARGET_HARD_FLOAT.
289 2007-01-19 Sandra Loosemore <sandra@codesourcery.com>
291 * longlong.h (count_leading_zeros, COUNT_LEADING_ZEROS_0): Add
292 ColdFire alternatives.
293 * config/m68k/m68k.h (CLZ_DEFINED_VALUE_AT_ZERO): New macro.
294 * config/m68k/m68k.md (clzsi2): Define for ColdFire
295 architectures that support ff1 instruction.
297 2007-01-19 Richard Sandiford <richard@codesourcery.com>
298 Julian Brown <julian@codesourcery.com>
300 * config/m68k/m68k.h (CONST_OK_FOR_LETTER_P): Add an 'R' case.
301 * config/m68k/m68k.md (*movsi_cfv4): Fold into...
302 (*movsi_cf): ...here. Remove unnecessary 'R' from 'Rg'.
305 2007-01-19 Richard Sandiford <richard@codesourcery.com>
307 * config/m68k/m68k-protos.h (valid_mov3q_const): Take a HOST_WIDE_INT
309 (output_move_const_into_data_reg, output_move_simode_const): Delete.
310 * config/m68k/m68k.c (const_method, const_int_cost): Take a
311 HOST_WIDE_INT instead of an rtx.
312 (m68k_rtx_costs): Update call accordingly.
313 (output_move_const_into_data_reg): Likewise. Fix formatting.
314 (valid_mov3q_const): Take a HOST_WIDE_INT instead of an rtx.
316 (output_move_simode_const): Update calls after above changes.
317 Rework to use automatic variables and predicates like MEM_P.
318 * config/m68k/m68k.md (pushexthisi_const): Update call to
321 2007-01-19 Dirk Mueller <dmueller@suse.de>
323 * tree-ssa-alias.c (perform_var_substitution): Fix typo
326 2007-01-19 Richard Guenther <rguenther@suse.de>
328 * builtins.c (expand_builtin_cexpi): Fall back to expanding
329 via cexp in case sincos is not available.
331 2007-01-19 Richard Guenther <rguenther@suse.de>
333 * doc/tm.texi (TARGET_HAS_SINCOS): Document new target macro.
334 * defaults.h (TARGET_HAS_SINCOS): Default to off.
335 * config/linux.h (TARGET_HAS_SINCOS): Set to on if we have glibc.
336 * config/alpha/linux.h (TARGET_HAS_SINCOS): Likewise.
337 * config/sparc/linux.h (TARGET_HAS_SINCOS): Likewise.
338 * config/sparc/linux64.h (TARGET_HAS_SINCOS): Likewise.
339 * config/rs6000/linux.h (TARGET_HAS_SINCOS): Likewise.
340 * config/rs6000/linux64.h (TARGET_HAS_SINCOS): Likewise.
342 2007-01-19 Uros Bizjak <ubizjak@gmail.com>
344 * config/i386/i386.md (*fpatanxf3_i387, fpatan_extend<mode>xf3_i387):
346 (atan2sf3_1, atan2df3_1, atan2xf3_1): Remove insn patterns.
347 (atan2xf3): Directly generate RTL pattern.
348 (atan2<mode>3): Rename from atan2sf3 and atan2df3 and macroize insn
349 patterns using X87MODEF12 mode macro. Use fpatan_extend<mode>xf3_i387
350 and truncate result to requested mode. Use SSE_FLOAT_MODE_P to
351 disable patterns for SSE math.
352 (atan<mode>2): Rename from atansf2 and atandf2 and macroize insn
353 patterns using X87MODEF12 mode macro. Use fpatan_extend<mode>xf3_i387
354 and truncate result to requested mode. Use SSE_FLOAT_MODE_P to
355 disable patterns for SSE math.
357 2007-01-19 Alexandre Oliva <aoliva@redhat.com>
359 * libgcc-std.ver: Fix typo in %inherit for GCC_4.3.0.
361 2007-01-18 Roger Sayle <roger@eyesopen.com>
363 * fold-const.c (fold_unary) <VIEW_CONVERT_EXPR>: Optimize away a
364 VIEW_CONVERT_EXPR to the same type as it's operand.
366 2007-01-18 David Edelsohn <edelsohn@gnu.org>
368 * config/rs6000/darwin-ldouble.c: Only build _SOFT_FLOAT if
369 configured for long double 128.
371 2007-01-18 Mike Stump <mrs@apple.com>
373 * config/rs6000/rs6000.c (rs6000_emit_vector_compare): Fix build
376 2007-01-18 Michael Meissner <michael.meissner@amd.com>
378 * i386.c (ix86_compute_frame_layout): Make fprintf's in #if 0 code
381 2007-01-18 Jan Hubicka <jh@suse.cz>
383 * tree-ssa-operands.c (vop_free_bucket_size): Never return value
384 greater than NUM_VOP_FREE_BUCKETS.
386 2007-01-18 Daniel Berlin <dberlin@dberlin.org>
388 * tree-ssa-structalias.c: Update comments.
389 (ptabitmap_obstack): Removed.
391 (oldpta_obstack): Ditto.
392 (stats): Add a few members.
393 (struct variable_info): Remove node, complex, address_taken, and
394 indirect_target members. Add oldsolution member.
395 (new_var_info): Do not initialize removed members.
396 (constraint_expr_type): Remove INCLUDES.
397 (constraint_graph): Add size, implicit_preds, rep,
398 indirect_cycles, eq_rep, label, direct_nodes, and complex members.
399 (FIRST_REF_NODE): New macro.
400 (LAST_REF_NODE): Ditto.
401 (FIRST_ADDR_NODE): Ditto.
402 (find): New function.
404 (dump_constraint): Do not handle INCLUDES.
405 (insert_into_complex): Do not insert duplicate constraints.
406 (condense_varmap_nodes): Renamed and rewritten into ...
407 (merge_node_constraints): This. Also fix bug in handling of
408 offseted copy constraints.
409 (clear_edges_for_node): No longer need to deal with preds at all,
410 or removing associated preds/succs.
411 (merge_graph_nodes): Deal with indirect_cycles.
412 Don't deal with predecessors.
413 (add_implicit_graph_edge): New function.
414 (add_pred_graph_edge): Ditto.
415 (add_graph_edge): Don't deal with predecessors.
416 (build_constraint_graph): Removed.
417 (build_pred_graph): New function.
418 (build_succ_graph): Ditto.
419 (struct scc_info): Removed in_component. Added roots, dfs, and
420 node_mapping. Remove visited_index, unification_queue.
421 (scc_visit): Deal with union-find we do now.
422 Deal with cycles with REF nodes.
423 (collapse_nodes): Renamed and rewritten to ...
425 (process_unification_queue): Removed.
426 (topo_visit): Cleanup
427 (do_da_constraint): Use find.
428 (do_sd_constraint): Ditto.
429 (do_ds_constraint): Ditto.
430 (do_complex_constraint): Ditto.
431 (init_scc_info): Update for removed and added members.
432 (find_and_collapse_graph_cycles): Renamed and rewritten into ...
433 (find_indirect_cycles): This.
434 (equivalence_class): New variable.
435 (label_visit): New function.
436 (perform_variable_substitution): Rewritten.
437 (free_var_substitution_info): New function.
438 (find_equivalent_node): Ditto.
439 (move_complex_constraints): Ditto.
440 (eliminate_indirect_cycles): Ditto.
441 (solve_graph): Only propagate changed bits.
442 Use indirect cycle elimination.
444 (tree_id_t): Rename to tree_vi_t, delete id member, add vi member.
445 (tree_id_eq): Renamed to ...
446 (tree_vi_eq): This. Update for member change
447 (insert_id_for_tree): Renamed and rewritten to ...
448 (insert_vi_for_tree): This.
449 (lookup_id_for_tree): Renamed and rewritten to ...
450 (lookup_vi_for_tree): This.
451 (get_id_for_tree): Renamed and rewritten to ...
452 (get_vi_for_tree): Ditto.
453 (get_constraint_exp_from_ssa_var): Update to use get_vi_for_tree.
454 (process_constraint): Don't handle INCLUDES.
455 Remove special ADDRESSOF case.
456 (find_func_aliases): Rewrite to use vi functions instead of id
458 (create_function_info_for): Ditto.
459 (create_variable_info_for): Ditto.
460 (intra_create_variable_infos): Ditto.
461 (merge_smts_into): Ditto.
462 (find_what_p_points_to): Ditto.
463 (init_base_vars): Ditto.
464 (init_alias_vars): Ditto.
465 (remove_preds_and_fake_succs): New function.
466 (dump_sa_points_to_info): Dump new stats.
467 (dump_solution_for_var): Use find.
468 (set_used_smts): Fix formatting.
469 (compute_points_to_sets): Updated for new functions.
470 (ipa_pta_execute): Ditto.
472 2007-01-18 Kazu Hirata <kazu@codesourcery.com>
473 Richard Sandiford <richard@codesourcery.com>
475 * doc/tm.texi (TARGET_FUNCTION_VALUE): Expand documentation of
477 * calls.c (expand_call): If the return value is a PARALLEL,
478 extract its first member.
479 * config/m68k/linux.h (FUNCTION_EXTRA_EPILOGUE): Remove.
480 * config/m68k/m68k.c (m68k_output_function_epilogue): Don't
481 use FUNCTION_EXTRA_EPILOGUE.
482 (m68k_function_value): Return a PARALLEL if the return value
483 is of a pointer type.
484 * config/m68k/netbsd-elf.h (current_function_returns_pointer)
485 (FUNCTION_EXTRA_EPILOGUE): Remove.
486 * config/m68k/m68k.md (D0_REG): New constant.
488 2007-01-18 Kazu Hirata <kazu@codesourcery.com>
490 * config/m68k/m68k.c (m68k_output_function_epilogue): Don't
491 output a NOP for empty epilogues.
493 2007-01-18 Richard Sandiford <richard@codesourcery.com>
495 * config/m68k/m68k.c (m68k_use_return_insn): Update comments
496 before function. Extend register save check to include all
497 registers, not just integer ones.
499 2007-01-18 Kazu Hirata <kazu@codesourcery.com>
501 * config/m68k/m68k-protos.h (use_return_insn): Rename to...
502 (m68k_use_return_insn): ...this.
503 * config/m68k/m68k.h (USE_RETURN_INSN): Delete.
504 * config/m68k/m68k.c (use_return_insn): Rename to...
505 (m68k_use_return_insn): ...this.
506 * config/m68k/m68k.md (return): Use m68k_use_return_insn instead
509 2007-01-18 Nathan Sidwell <nathan@codesourcery.com>
511 * config/m68k/fpgnulib.c (__truncdfsf2): Implement round to
512 nearest even, fix denormal rounding overflow.
514 2007-01-18 Richard Sandiford <richard@codesourcery.com>
516 * config/m68k/m68k.md (movsf_cf_hard): Use fsmove instead of
517 f%$smove and f%$move.
518 (movdf_cf_hard): Use fdmove for cases 0 and 3 and fmove for case 1.
519 (extendsfdf2_cf): Use fdmove instead of f%&move.
520 (truncdfsf2_cf): Use fsmove instead of f%$smove.
521 (add<mode>3_cf, sub<mode>3_cf): Use <FP:prec> instead of <FP:round>.
523 2007-01-18 Richard Sandiford <richard@codesourcery.com>
525 * config/m68k/m68k.md (movdf_cf_hard): Use output_move_double for
528 2007-01-18 Richard Sandiford <richard@codesourcery.com>
530 * real.h (real_format): Add a canonical_nan_lsbs_set field.
531 (coldfire_single_format): Declare.
532 (coldfire_double_format): Likewise.
533 * real.c (encode_ieee_single): Use canonical_nan_lsbs_set instead
534 of qnan_msb_set to determine the lower bits of a canonical
536 (encode_ieee_double): Likewise.
537 (encode_ieee_quad): Likewise.
538 (ieee_single_format): Initialize canonical_nan_lsbs_set.
539 (mips_single_format): Likewise.
540 (ieee_double_format): Likewise.
541 (mips_double_format): Likewise.
542 (ieee_extended_motorola_format): Likewise.
543 (ieee_extended_intel_96_format): Likewise.
544 (ieee_extended_intel_128_format): Likewise.
545 (ieee_extended_intel_96_round_53_format): Likewise.
546 (ibm_extended_format): Likewise.
547 (mips_extended_format): Likewise.
548 (ieee_quad_format): Likewise.
549 (mips_quad_format): Likewise.
550 (vax_f_format): Likewise.
551 (vax_d_format): Likewise.
552 (vax_g_format): Likewise.
553 (i370_single_format): Likewise.
554 (i370_double_format): Likewise.
555 (decimal_single_format): Likewise.
556 (decimal_double_format): Likewise.
557 (decimal_quad_format): Likewise.
558 (c4x_single_format): Likewise.
559 (c4x_extended_format): Likewise.
560 (real_internal_format): Likewise.
561 (coldfire_single_format): New real_format.
562 (coldfire_double_format): Likewise.
563 * config/pdp11/pdp11.c (pdp11_f_format): Initialize
564 canonical_nan_lsbs_set.
565 (pdp11_d_format): Likewise.
566 * config/m68k/m68k.c (override_options): Override REAL_FORMAT_MODE
567 if TARGET_COLDFIRE_CPU.
569 2007-01-18 Richard Sandiford <richard@codesourcery.com>
571 * config/m68k/m68k-protos.h (m68k_output_pic_call): Delete.
572 (output_call, m68k_legitimize_call_address): Declare.
573 * config/m68k/m68k.h (EXTRA_CONSTRAINT): Remove unnecessary
574 parenthesees. Add support for a 'W' constraint.
575 (LEGITIMATE_PIC_OPERAND_P): Remove SYMBOL_REF_FLAG handling.
576 (PRINT_OPERAND_PUNCT_VALID_P): Remove comment about 'o'.
577 (m68k_symbolic_call, m68k_symbolic_jump): Declare.
578 * config/m68k/m68k.c (m68k_symbolic_call, m68k_symbolic_jump): New
580 (override_options): Initialize them. Do not set flag_no_function_cse
581 for TARGET_ID_SHARED_LIBRARY.
582 (m68k_output_pic_call): Delete.
583 (m68k_legitimize_call_address): New function.
584 (print_operand): Remove the %o prefix. Handle the %p prefix.
585 (output_call): New function.
586 (m68k_output_mi_thunk): Use m68k_symbolic_jump. Always load the
587 target address from the GOT if symbolic jumps are not allowed.
588 * config/m68k/m68k.md (call, general_operand): Do not set
589 SYMBOL_REF_FLAG. Use m68k_legitimize_call_address instead.
590 Merge separate flag_pic and !flag_pic define_insns into...
591 (*call, *call_value): ...these new patterns. Match the address
592 rather than the containing MEM and require it to be a call_operand.
593 Use output_call to generate the asm template.
594 * config/m68k/predicates.md (const_call_operand): New predicate.
595 (call_operand): Likewise.
597 2007-01-18 Nathan Sidwell <nathan@codesourcery.com>
599 * config/m68k/m68k.h (REGISTER_MOVE_COST): Simplify definition.
600 (STACK_GROWS_DOWNWARD): Define to 1.
601 (FUNCTION_VALUE, LIBCALL_VALUE, FUNCTION_VALUE_REGNO_P): Equivocate
602 comments, emphasizing that these values are only defaults.
603 * config/m68k/linux.h (LINK_SPEC): Fix formatting in #undef.
604 * config/m68k/m68k.c (const_method): Remove trailing whitespace.
606 2007-01-18 Richard Sandiford <richard@codesourcery.com>
608 * config/m68k/m68k.md (cmpsi): Remove outdated flag_pic handling.
610 2007-01-18 Kazu Hirata <kazu@codesourcery.com>
611 Richard Sandiford <richard@codesourcery.com>
613 * config/m68k/m68k.h (DATA_REGNO_P, ADDRESS_REGNO_P, INT_REGNO_P)
614 (FP_REGNO_P): New macros.
615 (REGNO_OK_FOR_INDEX_P, REGNO_OK_FOR_BASE_P, REGNO_OK_FOR_DATA_P)
616 (REGNO_OK_FOR_FP_P, REG_OK_FOR_INDEX_P, REG_OK_FOR_BASE_P): Use them.
617 (EH_RETURN_STACKADJ_RTX): Use A0_REG.
618 * config/m68k/m68k.c (m68k_regno_mode_ok): Use the new REGNO macros.
620 2007-01-18 Nathan Sidwell <nathan@codesourcery.com>
622 * config.gcc (m68k-*-aout*, m68k-*-coff*, m68020-*-elf*, m68k-*-elf*)
623 (m68k-*-uclinuxoldabi, m68k-*-uclinux*, m68k-*-rtems*): Add t-floatlib
625 * config/m68k/t-floatlib: New file, extracting common code from...
626 * config/m68k/t-m68kbare, config/m68k/t-m68kelf,
627 * config/m68k/t-uclinux: Here.
628 * config/m68k/fpgnulib.c: Do not compile extendeed precision
629 routines on ColdFire targets.
631 2007-01-18 Nathan Sidwell <nathan@codesourcery.com>
633 * config.gcc (m68k-*-aout*, m68k-*-coff*, m68020-*-elf*, m68k-*-elf*)
634 (m68010-*-netbsdelf*, m68k*-*-netbsdelf*, m68k*-*-openbsd*)
635 (m68k-*-uclinuxoldabi, m68k-*-uclinux*, m68k-*-linux*)
636 (m68k-*-rtems*): Use tm_file rather than m68k/m68k.h and
637 explicitly set MOTOROLA to 1.
638 * config/m68k/m68k.h (MOTOROLA): Simplify definition accordingly.
640 2007-01-18 Richard Sandiford <richard@codesourcery.com>
641 Nathan Sidwell <nathan@codesourcery.com>
643 * config/m68k/m68k.h (PCC_STATIC_STRUCT_RETURN): Delete.
644 (ASM_OUTPUT_REG_PUSH, ASM_OUTPUT_REG_POP): Add MOTOROLA cases.
645 * config/m68k/coff.h (REGISTER_PREFIX_MD): Delete.
646 * config/m68k/m68020-elf.h (LIB_SPEC): Delete.
647 * config/m68k/m68k-none.h (CC1_SPEC, CPP_SUBTARGET_SPEC): Delete.
648 * config/m68k/m68kelf.h (IMMEDIATE_PREFIX, REGISTER_PREFIX_MD)
649 (ASM_OUTPUT_REG_PUSH, ASM_OUTPUT_REG_POP): Delete.
650 (NO_DOLLAR_IN_LABEL, PCC_STATIC_STRUCT_RETURN): Don't undefine.
651 (BSS_ASM_OP, ASM_OUTPUT_SKIP, ASM_OUTPUT_ASCII): Delete.
652 * config/m68k/m68kemb.h (PCC_STATIC_STRUCT_RETURN): Don't undefine.
653 * config/m68k/linux.h (SIZE_TYPE, PTRDIFF_TYPE, WCHAR_TYPE)
654 (WCHAR_TYPE_SIZE, TARGET_OBJFMT_CPP_BUILTINS): Delete.
655 (TARGET_OS_CPP_BUILTINS): Don't define mc68000 and mc68020 here.
656 (DBX_REGISTER_NUMBER): Delete.
657 * config/m68k/netbsd-elf.h (IMMEDIATE_PREFIX): Delete.
658 (PCC_STATIC_STRUCT_RETURN): Don't undefine.
659 * config/m68k/openbsd.h (PCC_STATIC_STRUCT_RETURN): Define.
661 2007-01-18 Richard Sandiford <richard@codesourcery.com>
663 * config.gcc (m68k-*-uclinux*): Add flat.h to $tm_file.
664 * config/flat.h: New file.
665 * crtstuff.c (USE_PT_GNU_EH_FRAME): Don't define if
667 * config/m68k/m68k.h (ASM_PREFERRED_EH_DATA_FORMAT): Do not use
668 indirect references for -msep-data or -mid-shared-library.
669 Do not use PC-relative code addresses either.
670 * config/m68k/m68k.c (override_options): Restrict -fPIC error
672 * config/m68k/uclinux.h (STARTFILE_SPEC): Define. Use Scrt1.o
673 for shared libraries and crt1.o for executables. Use crti.o and
675 (ENDFILE_SPEC): Use crtend.o and crtn.o.
676 (LIB_SPEC): Suppress -Rlibc.gdb if -static-libc is given.
677 Do not add -elf2flt or -shared-lib-id options here.
678 (LINK_SPEC): Define. Pass -elf2flt if no -elf2flt option is given.
679 Pass -shared-lib-id if -mid-shared-library, taking the library
680 identifier from -mshared-library-id if given, otherwise
682 (EH_FRAME_IN_DATA_SECTION): Do not undefine.
683 (INIT_SECTION_ASM_OP, FINI_SECTION_ASM_OP): Likewise.
684 (TARGET_OS_CPP_BUILTINS): Define __GXX_MERGED_TYPEINFO_NAMES=0
685 and __GXX_TYPEINFO_EQUALITY_INLINE=0 if -mid-shared-library.
686 (DRIVER_SELF_SPECS): Map unadorned PIC options to -msep-data.
687 * config/m68k/t-uclinux (EXTRA_MULTILIB_PARTS): Add crtbegin.o
689 * config/m68k/lb1sf68.asm (PICCALL): Use an lea and pc-relative
690 jump sequence for ISA A and ISA A+.
693 2007-01-18 Richard Sandiford <richard@codesourcery.com>
695 * config.gcc (m68k-*-uclinux*): Base the port on the common
696 and m68k GNU/Linux files rather than on the generic ELF ones.
697 * config/m68k/uclinux.h (TARGET_VERSION): Override.
698 (TARGET_OS_CPP_BUILTINS): Use LINUX_TARGET_OS_CPP_BUILTINS.
700 2007-01-18 Julian Brown <julian@codesourcery.com>
701 Richard Sandiford <richard@codesourcery.com>
703 * config/m68k/m68k.h (LONG_DOUBLE_TYPE_SIZE): Make 64-bit on ColdFire.
704 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
705 * config/m68k/netbsd-elf.h (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Undefine
707 * config/m68k/uclinux-oldabi.h (LONG_DOUBLE_TYPE_SIZE): Redefine to
709 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
711 2007-01-18 Richard Sandiford <richard@codesourcery.com>
713 * doc/install.texi: Document m68k-uclinuxoldabi.
714 * config.gcc (m68k-*-uclinuxoldabi): New configuration.
715 * config/m68k/uclinux-oldabi.h: New file, copied from
716 config/m68k/uclinux.h.
718 2007-01-18 Kaz Kojima <kkojima@rr.iij4u.or.jp>
720 * config/m32r/m32r-protos.h (m32r_expand_epilogue): Declare it.
721 * config/m32r/m32r.c (m32r_setup_incoming_varargs): Use gen_frame_mem.
722 (m32r_compute_frame_size): Use unsigned for regno.
723 (m32r_reload_lr): Use gen_frame_mem.
725 (m32r_output_function_epilogue): Don't output the function epilogue
727 (m32r_expand_epilogue): New.
728 (direct_return): Return false if the function has the interrupt
730 (m32r_hard_regno_rename_ok): Remove code for the textual epilogue.
731 * config/m32r/m32r.md (epilogue): New expander.
732 (return_lr, return_rte): New insns.
733 (return): Make it expander.
734 (return_normal): New expander.
736 2007-01-18 Josh Conner <jconner@apple.com>
739 * config/rs6000/rs6000.c (rs6000_emit_vector_compare): Add
740 support for UNLE, UNLT, UNGE, and UNGT.
742 2007-01-18 Jan Hubicka <jh@suse.cz>
744 * tree-vrp.c (finalize_jump_threads): Do not call cleanup_cfg by hand.
745 * tree-tailcall (add_virtual_phis): Likewise.
746 (optimize_tail_call): Return TODOs.
747 (execute_tail_calls): Return TODOs.
748 * tree-ssa-ccp (execute_fold_all_builtins): Do cleanup_cfg via TODO.
749 * tree-cfgcleanup.c (cleanup_tree_cfg_loop): Return if something
751 * tree-ssa-forwprop.c (tree_ssa_forward_propagate_single_use_value):
752 Cleanup cfg using TODO.
753 * tree-flow.h (cleanup_tree_cfg_loop): Update prototype.
754 * passes.c (execute_function_todo): When cleanup did something, remove
756 * tree-cfg.c (pass_build_cfg): Add cleanup_cfg TODO.
757 (make_edges): Don't cleanup_cfg.
759 2007-01-18 Uros Bizjak <ubizjak@gmail.com>
761 * reg-stack.c (subst_stack_regs_pat) [UNSPEC_SINCOS_COS,
762 UNSPEC_XTRACT_FRACT]: Use generic code for instructions that
763 operate on the top of stack.
764 [UNSPEC_SINCOS_SIN, UNSPEC_XTRACT_EXP, UNSPEC_TAN]: Rewrite
765 register handling of instructions that output to the second
767 [UNSPEC_TAN_ONE, UNSPEC_TAN_TAN]: Remove.
768 (move_for_stack_reg): Special-case check for dead destination
769 stack slot for constant load of 1.0 inside UNSPEC_TAN.
771 * config/i386/i386.md (UNSPEC_TAN): New constant.
772 (UNSPEC_TAN_ONE, UNSPEC_TAN_TAN): Remove.
773 (fptanxf4_i387, fptan_extend<mode>xf4_i387): New patterns
774 to correctly model move of constant 1.0 to top stack slot.
775 (*tandf3_1, *tansf3_1, *tanxf3_1): Remove insn patterns.
776 (unnamed peephole2 pattern): Remove corresponding peephole2
777 pattern that optimizes tan insn and loading of constant 1.0.
778 (tanxf2): Use fptanxf4_i387.
779 (tan<mode>2): Rename from tansf2 and tandf2 and macroize insn
780 patterns using X87MODEF12 mode macro. Use fptan_extend<mode>xf4_i387
781 and truncate result to requested mode. Use SSE_FLOAT_MODE_P to
782 disable patterns for SSE math.
783 (sincos<mode>3): Use truncxf<mode>2_i387_noop for truncation.
784 (fyl2x_extend<mode>xf3_i387): Use X87MODEF12 for operand 1.
786 2007-01-18 Dirk Mueller <dmueller@suse.de>
787 Richard Guenther <rguenther@suse.de>
790 * doc/invoke.texi (Warray-bounds): Document -Warray-bounds.
791 * common.opt (Warray-bounds): Add new warning option.
792 * c-opts.c (c_common_handle_option): Define -Warray-bounds
794 * Makefile.in: make tree-vrp.o depend on toplev.h
795 * tree-vrp.c (vrp_finalize): Call check_array_refs if -Warray-bounds
797 (check_array_refs, check_array_bounds, check_array_ref): New.
799 2007-01-18 Jan Hubicka <jh@suse.cz>
801 * tree-ssa-ccp.c (ccp_finalize): Return if something changed.
802 (execute_ssa_ccp): Return flags conditionally.
803 * tree-ssa-propagate.c (substitue_and_fold): Return if something was
805 * tree-ssa-propagate.h (substitute_and_fold): Update prototype.
807 2007-01-18 Steven Bosscher <steven@gcc.gnu.org>
809 * cfgcleanup.c (cleanup_cfg): Detect cfglayout mode and set
810 the CLEANUP_CFGLAYOUT flag when in cfglayout mode.
812 * Makefile.c (GTFILES): Add cfglayout.h.
813 * gengtype.c (open_base_files): Likewise.
814 * cfglayout.c (cfg_layout_function_footer,
815 cfg_layout_function_header) Reindent to make gengtype happy.
816 * cfglayout.h (cfg_layout_function_footer,
817 cfg_layout_function_header): Add GTY(()) marker.
819 * ifcvt.c (noce_try_sign_mask): Make sure INSN_B is non-null.
821 2007-01-18 Ben Elliston <bje@au.ibm.com>
823 * genautomata.c (write_automata): Include xstrerror output in the
824 error message if writing the DFA description file fails.
826 2007-01-17 H.J. Lu <hongjiu.lu@intel.com>
828 * config/mips/mips-protos.h (mips_output_external): Make it
830 * config/mips/iris.h (TARGET_ASM_EXTERNAL_LIBCALL): Removed.
831 * config/mips/mips.c (irix_output_external_libcall): Likewise.
832 (extern_list): Likewise.
833 (extern_head): Likewise.
834 (TARGET_ASM_FILE_END): Likewise.
835 (mips_file_end): Likewise.
836 (mips_output_external): Rewritten.
838 2007-01-18 Ben Elliston <bje@au.ibm.com>
840 * genpreds.c (write_insn_preds_c): Only write out the function
841 body for regclass_for_constraint if we have register constraints.
843 2007-01-17 Tom Tromey <tromey@redhat.com>
845 * doc/sourcebuild.texi (libgcj Tests): Use sourceware.org.
846 * doc/install.texi (Testing): Use sourceware.org.
847 (Binaries): Likewise.
848 (Specific): Likewise.
849 * doc/contrib.texi (Contributors): Use sourceware.org.
851 2007-01-17 Anatoly Sokolov <aesok@post.ru>
853 * config/avr/avr.h (AVR_HAVE_LPMX): New macro.
854 (AVR_ENHANCED): Rename to ...
855 (AVR_HAVE_MUL): ... new.
856 (avr_enhanced_p): Rename to ...
857 (avr_have_mul_p): ... new.
858 (TARGET_CPU_CPP_BUILTINS): Use 'avr_have_mul_p' instead of
859 'avr_enhanced_p' for "__AVR_ENHANCED__". Define "__AVR_HAVE_MUL__".
860 * config/avr/avr.c (avr_enhanced_p): Rename to ...
861 (avr_have_mul_p): ... new.
862 (base_arch_s): Rename 'enhanced' to 'have_mul'.
863 (avr_override_options): Use 'avr_have_mul_p' and 'have_mul' instead of
864 'avr_enhanced_p' and 'enhanced'.
865 (ashlhi3_out, ashrhi3_out, lshrhi3_out, avr_rtx_costs): Use
866 AVR_HAVE_MUL instead of AVR_ENHANCED.
867 * avr.md (*tablejump_enh): Use AVR_HAVE_LPMX instead of AVR_ENHANCED.
868 (mulqi3, *mulqi3_enh, *mulqi3_call, mulqihi3, umulqihi3, mulhi3,
869 *mulhi3_enh, *mulhi3_call, mulsi3, *mulsi3_call): Use AVR_HAVE_MUL
870 instead of AVR_ENHANCED.
871 (*tablejump_enh): Use AVR_HAVE_LPMX instead of AVR_ENHANCED.
872 * libgcc.S: Use __AVR_HAVE_MUL__ instead of __AVR_ENHANCED__.
873 (__tablejump__): Use __AVR_HAVE_LPMX__ instead of __AVR_ENHANCED__.
875 2007-01-17 Ian Lance Taylor <iant@google.com>
877 * vec.h (VEC_reserve_exact): Define.
878 (vec_gc_p_reserve_exact): Declare.
879 (vec_gc_o_reserve_exact): Declare.
880 (vec_heap_p_reserve_exact): Declare.
881 (vec_heap_o_reserve_exact): Declare.
882 (VEC_OP (T,A,reserve_exact)): New static inline function, three
884 (VEC_OP (T,A,reserve)) [all versions]: Remove handling of
886 (VEC_OP (T,A,alloc)) [all versions]: Call ...reserve_exact.
887 (VEC_OP (T,A,copy)) [all versions]: Likewise.
888 (VEC_OP (T,a,safe_grow)) [all versions]: Likewise.
889 * vec.c (calculate_allocation): Add exact parameter. Change all
891 (vec_gc_o_reserve_1): New static function, from vec_gc_o_reserve.
892 (vec_gc_p_reserve, vec_gc_o_reserve): Call vec_gc_o_reserve_1.
893 (vec_gc_p_reserve_exact, vec_gc_o_reserve_exact): New functions.
894 (vec_heap_o_reserve_1): New static function, from vec_heap_o_reserve.
895 (vec_heap_p_reserve, vec_heap_o_reserve): Call vec_heap_o_reserve_1.
896 (vec_heap_p_reserve_exact): New function.
897 (vec_heap_o_reserve_exact): New function.
899 2007-01-17 Jan Hubicka <jh@suse.cz>
901 * ipa-type-escape.c (look_for_casts): Revamp using handled_component_p.
903 2007-01-17 Eric Christopher <echristo@apple.com>
905 * config.gcc: Support core2 processor.
907 2007-01-16 Jan Hubicka <jh@suse.cz>
909 * tree-ssanames.c (release_dead_ssa_names): Instead of ggc_freeing
910 the names, just unlink the chain so we don't crash on dangling pointers
913 2007-01-16 Jan Hubicka <jh@suse.cz>
915 * cgraph.h (cgraph_decide_inlining_incrementally): Kill.
916 * tree-pass.h: Reorder to make IPA passes appear toegher.
917 (pass_early_inline, pass_inline_parameters, pass_apply_inline): Declare.
918 * cgraphunit.c (cgraph_finalize_function): Do not compute inling
919 parameters, do not call early inliner.
920 * ipa-inline.c: Update comments. Include tree-flow.h
921 (cgraph_decide_inlining): Do not compute inlining parameters.
922 (cgraph_decide_inlining_incrementally): Return TODOs; assume to
923 be called with function context set up.
924 (pass_ipa_inline): Remove unreachable functions before pass.
925 (cgraph_early_inlining): Simplify assuming to be called from the
927 (pass_early_inline): New pass.
928 (cgraph_gate_ipa_early_inlining): New gate.
929 (pass_ipa_early_inline): Turn into simple wrapper.
930 (compute_inline_parameters): New function.
931 (gate_inline_passes): New gate.
932 (pass_inline_parameters): New pass.
933 (apply_inline): Move here from tree-optimize.c
934 (pass_apply_inline): New pass.
935 * ipa.c (cgraph_remove_unreachable_nodes): Verify cgraph after
937 * tree-inline.c (optimize_inline_calls): Return TODOs rather than
939 (tree_function_versioning): Do not allocate dummy struct function.
940 * tree-inline.h (optimize_inline_calls): Update prototype.
941 * tree-optimize.c (execute_fixup_cfg): Export.
942 (pass_fixup_cfg): Remove
943 (tree_rest_of_compilation): Do not apply inlines.
944 * tree-flow.h (execute_fixup_cfg): Declare.
945 * Makefile.in (gt-passes.c): New.
946 * passes.c: Include gt-passes.h
947 (init_optimization_passes): New passes.
948 (nnodes, order): New static vars.
949 (do_per_function_toporder): New function.
950 (execute_one_pass): Dump current pass here.
951 (execute_ipa_pass_list): Don't dump current pass here.
953 2007-01-16 Janis Johnson <janis187@us.ibm.com>
955 * config/dfp-bit.c (dfp_compare_op): Return separate value for NaN.
956 (DFP_NE, DFP_LE, DFP_GE): Return false for NaN.
958 2007-01-16 David Edelsohn <edelsohn@gnu.org>
960 * config/rs6000/darwin-ldouble.c: Build file for SOFT_FLOAT.
961 (strong_alias): Define.
962 (__gcc_qmul): Provide non-FMA for soft-float.
978 * config/rs6000/rs6000.c (rs6000_init_libfuncs): Initialize
979 soft-float functions.
980 * config/rs6000/libgcc-ppc-glibc.ver: Version soft-float symbols.
981 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Do not warn
982 about long double soft float.
984 2007-01-16 Dorit Nuzman <dorit@il.ibm.com>
985 Tehila Meyzels <tehila@il.ibm.com>
987 * tree-vectorizer.h (is_pattern_stmt_p): New.
988 * tree-vect-analyze.c (vect_determine_vectorization_factor): Fix
989 formatting (tabs instead of spaces). Cleanup and clarify setting
990 of STMT_VINFO_VECTYPE. Call is_pattern_stmt_p.
991 * tree-vect-transform.c (vect_get_vec_def_for_operand): Fix typo.
992 (vectorizable_type_demotion): Check that types are integral.
993 (vectorizable_type_promotion): Likewise.
994 (vectorizable_store): Fix typo. Eliminate new-line at end of
997 2007-01-16 Jan Hubicka <jh@suse.cz>
999 * tree-ssanames.c (release_dead_ssa_names): Remove invalidated
1002 2007-01-15 Eric Christopher <echristo@apple.com>
1004 * ifcvt.c: Include vec.h, vecprim.h.
1005 (check_cond_move_block): New argument regs.
1006 Reorganize. Add registers used to regs.
1007 (cond_move_process_if_block): Use regs set above as
1010 2007-01-15 Eric Christopher <echristo@apple.com>
1012 * config/darwin.h: Update copyright.
1013 (TARGET_OPTION_TRANSLATE_TABLE): Add umbrella.
1014 (LINK_COMMAND_SPEC): Add -u.
1015 (LINK_SPEC): Fix umbrella for above.
1017 2007-01-15 Joseph S. Myers <joseph@codesourcery.com>
1019 * config/soft-fp/op-common.h, config/soft-fp/op-4.h: Update from
1022 2007-01-15 Tom Tromey <tromey@redhat.com>
1024 * doc/sourcebuild.texi (libgcj Tests): Don't mention jacks.
1025 * doc/install.texi (Testing): Don't mention jacks.
1026 (Configuration): Document --enable-java-maintainer-mode. Move
1027 --with-java-home to libgcj-specific section. Document
1029 (Prerequisites): Mention --enable-java-maintainer-mode, ecj1.
1031 2007-01-15 Jan Hubicka <jh@suse.cz>
1033 * tree-ssa-dce.c (DCE_TODOs): New.
1034 (propagate_necessity): Return if something changed.
1035 (eliminate_unnecessary_stmts): Likewise.
1036 (perform_tree_ssa_dce): Return TODO flags when needed.
1037 (pass_dce, pass_dce_loop, pass_cd_dce): Remove TODO flags.
1039 2007-01-15 Uros Bizjak <ubizjak@gmail.com>
1041 * config/i386/i386.md (fyl2xxf3_i387): Rename from fyl2x_xf3.
1042 (fyl2x_extend<mode>xf3_i387): New insn pattern.
1043 (log<mode>2): Rename from logsf2 and logdf2 and macroize insn
1044 patterns using X87MODEF12 mode macro. Extend operand 1
1045 to XFmode. Use SSE_FLOAT_MODE_P to disable patterns for SSE math.
1046 (log10<mode>2): Ditto.
1047 (log2<mode>2): Ditto.
1048 (log1p<mode>2): Ditto.
1049 (logb<mode>2): Ditto.
1050 (fyl2xp1xf3_i387): Rename from fyl2xp1_xf3.
1051 (fyl2xp1_extend<mode>xf3_i387): New insn pattern.
1052 (*fxtractxf3_i387): Rename from *fxtractxf3.
1053 (fxtract_extend<mode>xf3_i387): New insn pattern.
1054 (ilogbsi2): Use match_dup 3, not match_operand:XF 3.
1055 * config/i386/i386.c (ix86_emit_i387_log1p): Use gen_fyl2xp1xf3_i387()
1056 and gen_fyl2xxf3_i387().
1058 2007-01-14 Zdenek Dvorak <dvorakz@suse.cz>
1060 * loop-unswitch.c (unswitch_loop): Do not call fix_loop_placement.
1061 * cfgloopmanip.c (fix_loop_placement): Made static. Use
1062 get_loop_exit_edges. Changed return type to bool.
1063 * cfgloop.h (fix_loop_placement): Declaration removed.
1065 2007-01-14 Dorit Nuzman <dorit@il.ibm.com>
1067 * param.h (MIN_VECT_LOOP_BOUND): New.
1068 * params.def (MIN_VECT_LOOP_BOUND): New.
1069 * tree-vectorizer.c (slpeel_tree_peel_loop_to_edge): Takes another
1070 argument - minimum threshold for number of iterations.
1071 * tree-vectorizer.h (slpeel_tree_peel_loop_to_edge): Add another
1072 argument to declaration.
1073 * tree-vect-analyze.c (vect_analyze_operations): Check value of
1074 MIN_VECT_LOOP_BOUND.
1075 * tree-vect-transform.c (vect_do_peeling_for_loop_bound): Call
1076 slpeel_tree_peel_loop_to_edge with additional argument.
1077 (vect_do_peeling_for_alignment): Likewise.
1078 * doc/invoke.texi (min-vect-loop-bound): Document new param option.
1080 2007-01-14 Uros Bizjak <ubizjak@gmail.com>
1083 * config/i386/i386.c (print_operand) ['z']: Output 'b' for
1086 2007-01-14 Jan Hubicka <jh@suse.cz>
1088 * tree-dfa.c (remove_referenced_var): New function.
1089 * tree-ssa-live.c (remove_unused_locals): Walk referenced vars and
1090 prune referenced vars list too.
1091 * tree-flow.h (remove_referenced_var): Declare.
1093 2007-01-14 Jan Hubicka <jh@suse.cz>
1095 * tree-eh.c (add_stmt_to_eh_region_fn): Do not add call_exprs
1097 (remove_stmt_from_eh_region_fn): Do not remove call_exprs.
1098 (verify_eh_throw_stmt_node, verify_eh_throw_table_statements): Kill.
1099 * except.h (verify_eh_throw_table_statements): Kill prototype.
1100 * cfgexpand.c (expand_gimple_basic_block): Propagate Eh regions
1101 into call exrepssions.
1102 * tree-optimize.c (execute_free_cfg_annotatiosn): Do not call
1104 * tree-cfg.c: Include pointer-set.h.
1105 (verify_node_sharing): Work on pointer set.
1106 (verify_eh_throw_stmt_node): New.
1107 (verify_stmts): Use pointers sets, verify throw_stmt.
1109 2007-01-13 Zdenek Dvorak <dvorakz@suse.cz>
1111 * ipa-reference.c (analyze_function): Consider also addresses taken
1114 2007-01-12 Roger Sayle <roger@eyesopen.com>
1116 * c-typeck.c (null_pointer_constant_p): Replace use of
1117 TREE_CONSTANT_OVERFLOW with TREE_OVERFLOW.
1118 (build_c_cast): Likewise.
1120 2007-01-12 Roger Sayle <roger@eyesopen.com>
1122 * tree.h (force_fit_type_double): Remove unused final argument.
1123 * c-common.c (constant_expression_warning): Replace use of
1124 TREE_CONSTANT_OVERFLOW with TREE_OVERFLOW.
1125 (convert_and_check): Likewise.
1126 (shorten_compare): Update call to force_fit_type_double.
1127 (c_common_truthvalue_conversion) <INTEGER_CST>: Use integer_zerop.
1128 * convert.c (convert_to_pointer): Update call to
1129 force_fit_type_double.
1130 * fold-const.c (force_fit_type_double): Remove overflowed_const
1132 (int_const_binop, fold_convert_const_int_from_int,
1133 fold_convert_const_int_from_real, fold_div_compare,
1134 fold_sign_changed_comparison, fold_unary, fold_negate_const,
1135 fold_abs_const, fold_not_const): Remove the final argument from
1136 calls to force_fit_type_double.
1138 2007-01-12 Andrew Pinski <andrew_pinski@playstation.sony.com>
1140 * configure.ac: Set insn to "nop" for spu-*-* also.
1141 * configure: Regenerate.
1143 2007-01-12 Olga Golovanevsky <olga@il.ibm.com>
1145 * builtins.def : Add BUILT_IN_FREE.
1147 2007-01-12 Jan Hubicka <jh@suse.cz>
1149 PR tree-optimization/30443
1150 * tree-inline.c (tree_function_versioning): Do not optimize when
1151 cloning for inlining.
1153 2007-01-12 Zdenek Dvorak <dvorakz@suse.cz>
1155 * doc/loop.texi: Document recording of loop exits.
1156 * cfgloopmanip.c (loopify, duplicate_loop): Use alloc_loop.
1157 (update_single_exits_after_duplication,
1158 update_single_exit_for_duplicated_loop,
1159 update_single_exit_for_duplicated_loops): Removed.
1160 (duplicate_loop_to_header_edge): Do not call
1161 update_single_exits_after_duplication and
1162 update_single_exit_for_duplicated_loops.
1163 (loop_version): Do not update single_exit information.
1164 (fix_loop_structure): Use record_loop_exits instead of
1165 mark_single_exit_loops.
1166 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update
1167 the lists of loop exits.
1168 * cfghooks.c (redirect_edge_and_branch, redirect_edge_and_branch_force,
1169 split_edge, merge_blocks): Update the lists of loop exits.
1170 * modulo-sched.c (sms_schedule): Pass LOOPS_HAVE_RECORDED_EXITS to
1171 loop_optimizer_init.
1172 * loop-init.c (loop_optimizer_init): Call record_loop_exits instead
1173 of mark_single_exit_loops.
1174 (loop_optimizer_finalize): Call release_recorded_exits.
1175 * tree-ssa-loop.c (tree_loop_optimizer_init): Pass
1176 LOOPS_HAVE_RECORDED_EXITS to loop_optimizer_init.
1177 * tree-vectorizer.c (slpeel_tree_duplicate_loop_to_edge_cfg): Do not
1178 update single exit information.
1179 * lambda-code.c (perfect_nestify): Ditto.
1180 * cfgloop.c (flow_loop_free): Destroy the list of exits of the loop.
1181 (mark_single_exit_loops): Removed.
1182 (alloc_loop, loop_exit_hash, loop_exit_eq, loop_exit_free,
1183 get_exit_descriptions, rescan_loop_exit, record_loop_exits,
1184 dump_recorded_exit, dump_recorded_exits, release_recorded_exits): New
1186 (get_loop_exit_edges, single_exit): Use recorded exit lists.
1187 (add_bb_to_loop, remove_bb_from_loops): Update the lists of loop exits.
1188 (verify_loop_structure): Verify consistency of the exit lists.
1189 (flow_loops_find): Use alloc_loop. Initialize exits hash.
1190 (set_single_exit): Removed.
1191 * cfgloop.h (struct loop_exit): New function.
1192 (struct loop): single_exit_ field replaced by exits field.
1193 (LOOPS_HAVE_MARKED_SINGLE_EXITS): Replaced by LOOPS_HAVE_RECORDED_EXITS.
1194 (struct loops): Added exits hash.
1195 (mark_single_exit_loops, set_single_exit): Declaration removed.
1196 (release_recorded_exits, record_loop_exits, rescan_loop_exit): Declare.
1198 2007-01-12 Richard Sandiford <richard@codesourcery.com>
1200 * doc/invoke.texi: Avoid use of @headitem.
1202 2007-01-12 Richard Sandiford <richard@codesourcery.com>
1204 * cse.c (cse_insn): Move HAVE_CC0 code after declarations.
1206 2007-01-12 Richard Sandiford <richard@codesourcery.com>
1208 * doc/install.texi: Fix m68k-*-* anchor and add m68k-*-* to the
1211 2007-01-12 Nathan Sidwell <nathan@codesourcery.com>
1212 Richard Sandiford <richard@codesourcery.com>
1214 * doc/invoke.texi: Document -mno-bitfield, -mno-rtd and -mno-short.
1215 * config/m68k/m68k.opt: Resort options.
1216 (mbitfield, mrtd, mshort): Remove RejectNegative properties.
1218 2007-01-12 Nathan Sidwell <nathan@codesourcery.com>
1219 Richard Sandiford <richard@codesourcery.com>
1221 * doc/invoke.texi: Document the macros that are defined by
1222 m68k's -mtune and -mhard-float options.
1223 * config/m68k/m68k-protos.h (m68k_cpp_cpu_ident) Declare.
1224 (m68k_cpp_cpu_family): Likewise.
1225 * config/m68k/m68k.h (TARGET_CPU_CPP_BUILTINS): Add a full set
1226 of __ucfv*__ macros. Define __mcffpu__ if generating code for
1227 ColdFire FPUs. Define __mcf_cpu_* and __mcf_family_* macros.
1228 * config/m68k/m68k.c (m68k_cpp_cpu_ident): New function.
1229 (m68k_cpp_cpu_family): Likewise.
1231 2007-01-12 Richard Sandiford <richard@codesourcery.com>
1233 * config/m68k/m68k.h (TARGET_CPU_CPP_BUILTINS): Treat all mc68*
1234 macros besides mc68000 as tuning macros. Use a switch statement
1235 to set them and mcpu32.
1237 2007-01-12 Julian Brown <julian@codesourcery.com>
1239 * config/m68k/m68k.h: Use TARGET_68040 instead of TARGET_68040_ONLY.
1240 (TARGET_68040_ONLY): Rename to...
1241 (TARGET_68040): ...this.
1242 * config/m68k/m68k.c: Use TARGET_68040 instead of TARGET_68040_ONLY.
1243 * config/m68k/m68k.md: Likewise.
1245 2007-01-12 Julian Brown <julian@codesourcery.com>
1246 Nathan Sidwell <nathan@codesourcery.com>
1247 Richard Sandiford <richard@codesourcery.com>
1249 * config.gcc (m680[012]0-*-*, m68k*-*-*): Set m68k_cpu_ident to
1250 the -mcpu= argument associated with the --with-cpu setting.
1251 Define M68K_DEFAULT_TUNE to the default -mtune= option,
1252 if different from the one implied by the -mcpu setting.
1253 Accept --with-cpu=FOO if FOO is listed in m68k-devices.def,
1254 using mcpu=FOO as the default CPU option. Set target_cpu_default2.
1255 * doc/invoke.texi: Mention ColdFire in the introduction to the
1256 m68k options. Document the new -march, -mcpu, -mtune, -mdiv,
1257 -mno-div and -mhard-float options. Make -m68881 a synonym for
1258 -mhard-float. Document the previously-undocumented -m5206e,
1259 -m528x, -m5307 and -m5407 options. Tweak the existing option
1260 documentation for consistency.
1261 * doc/install.texi: Mention new --with-cpu arguments.
1262 * config/m68k/m68k.h (OPTION_DEFAULT_SPECS): Only use the
1263 default CPU if neither -mcpu nor -march are specified.
1264 (ASM_CPU_SPEC): Pass down -mcpu and -march options.
1265 (TARGET_CPU_CPP_BUILTINS): Set __mcfisa*__ macros from
1266 TARGET_ISA*. Set the legacy __mcf*__ cpu macros in the same way,
1267 using m68k_tune to decide between families that implement the
1268 same ISA. Use m68k_tune to set __mcfv4e__.
1269 (FL_BITFIELD, FL_68881, FL_COLDFIRE, FL_CF_HWDIV, FL_CF_MAC)
1270 (FL_CF_EMAC, FL_CF_EMAC_B, FL_CF_USP, FL_CF_FPU, FL_ISA_68000)
1271 (FL_ISA_68010, FL_ISA_68020, FL_ISA_68040, FL_ISA_A, FL_ISA_B)
1272 (FL_ISA_C, FL_ISA_MMU): New macros.
1273 (MASK_COLDFIRE): Delete.
1274 (TARGET_68010, TARGET_68020, TARGET_68040_ONLY, TARGET_COLDFIRE)
1275 (TARGET_ISAB): Redefine in terms of m68k_cpu_flags.
1276 (TARGET_68881, TARGET_COLDFIRE_FPU): Redefine in terms of m68k_fpu.
1277 (TARGET_HARD_FLOAT): Do not define here.
1278 (TARGET_ISAAPLUS, TARGET_ISAC): New macros.
1279 (TUNE_68000): New macro.
1280 (TUNE_68000_10): Redefine in terms of TUNE_68000 and TUNE_68010.
1281 (TUNE_68010, TUNE_68030, TUNE_68040, TUNE_68060, TUNE_CPU32)
1282 (TUNE_CFV2): Redefine in terms of m68k_tune.
1283 (uarch_type, target_device, fpu_type): New enums.
1284 (m68k_cpu, m68k_tune, m68k_fpu, m68k_cpu_flags): Declare.
1285 * config/m68k/m68k.c (TARGET_DEFAULT): Remove MASK_68881.
1286 (FL_FOR_isa_00, FL_FOR_isa_10, FL_FOR_isa_20, FL_FOR_isa_40)
1287 (FL_FOR_isa_cpu32, FL_FOR_isa_a, FL_FOR_isa_aplus, FL_FOR_isa_b)
1288 (FL_FOR_isa_c): New macros.
1289 (m68k_isa): New enum.
1290 (m68k_target_selection): New structure.
1291 (all_devices, all_isas, all_microarchs): New tables.
1292 (m68k_cpu_entry, m68k_arch_entry, m68k_tune_entry, m68k_cpu)
1293 (m68k_tune, m68k_fpu, m68k_cpu_flags): New variables.
1294 (MASK_ALL_CPU_BITS): Delete.
1295 (m68k_find_selection): New function.
1296 (m68k_handle_option): Handle -mcpu=, -march= and -mtune=.
1297 Map the legacy target options to a combination of the new ones.
1298 (override_options): Set m68k_cpu, m68k_tune, m68k_fpu and
1299 m68k_cpu_flags. Handle M68K_DEFAULT_TUNE. Use m68k_cpu_flags
1300 to derive default MASK_BITFIELD, MASK_CF_HWDIV and MASK_HARD_FLOAT
1302 * config/m68k/m68k.opt (m5200, m5206e, m528x, m5307, m5407, mcfv4e)
1303 (m68010, m68020, m68020-40, m68020-60, m68030, m68040): Remove Mask
1305 (m68881, msoft-float): Change mask from 68881 to HARD_FLOAT.
1306 (march=, mcpu=, mdiv, mhard-float, mtune=): New options.
1307 * config/m68k/m68k-devices.def: New file.
1309 2007-01-12 Richard Sandiford <richard@codesourcery.com>
1310 Nathan Sidwell <nathan@codesourcery.com>
1312 * config/m68k/m68k.h (ASM_CPU_SPEC, ASM_SPEC, EXTRA_SPECS)
1313 (SUBTARGET_EXTRA_SPECS): New macros.
1314 * config/m68k/linux.h (ASM_SPEC): Remove CPU flags;
1315 use %(asm_cpu_spec) instead.
1316 * config/m68k/m68k-none.h (ASM_SPEC): Likewise.
1317 * config/m68k/openbsd.h (ASM_SPEC): Likewise.
1318 * config/m68k/netbsd-elf.h (ASM_SPEC): Likewise.
1319 (EXTRA_SPECS): Rename to...
1320 (SUBTARGET_EXTRA_SPECS): ...this.
1322 2007-01-12 Nathan Sidwell <nathan@codesourcery.com>
1323 Richard Sandiford <richard@codesourcery.com>
1324 Julian Brown <julian@codesourcery.com>
1326 * config.gcc (m68k-*-aout*, m68k-*-coff*, m68020-*-elf*, m68k-*-elf*)
1327 (m68k-*-uclinux*, m68k-*-linux*, m68k-*-rtems*): Set default_m68k_cpu
1328 to the configuration's default CPU.
1329 (m68010-*-netbsdelf*, m68k*-*-netbsdelf*, m68k*-*-openbsd*): Likewise.
1330 Remove default masks.
1331 (m680[012]0-*-*): Set the default with_cpu to the first part of
1333 (m68k*-*-*): Set the default with_cpu to m$default_m68k_cpu.
1334 (m68k*-*-linux): Extend the --with-cpu handling to...
1335 (m680[012]0-*-*, m68k*-*-*): ...these configurations. Allow m68000
1336 and m68010. Don't set target_cpu_default2.
1337 * doc/install.texi: Document --with-cpu for m68k.
1338 * config/m68k/m68k.h (OPTION_DEFAULT_SPECS): Define.
1339 * config/m68k/m68k-none.h (TARGET_CPU_DEFAULT, M68K_CPU_m68k)
1340 (M68K_CPU_m68000, M68K_CPU_m68010, M68K_CPU_m68020, M68K_CPU_m68030)
1341 (M68K_CPU_m68040, M68K_CPU_m68302, M68K_CPU_m68332, TARGET_DEFAULT)
1342 (ASM_CPU_DEFAULT_SPEC, CC1_CPU_DEFAULT_SPEC): Delete.
1343 (ASM_SPEC): Remove use of %(asm_cpu_default).
1344 (EXTRA_SPECS, SUBTARGET_EXTRA_SPECS, MULTILIB_DEFAULTS): Delete.
1345 * config/m68k/linux.h (TARGET_DEFAULT): Delete.
1346 (CPP_SPEC): Merge definitions. Do not handle __HAVE_68881__ here.
1347 * config/m68k/netbsd-elf.h (TARGET_OS_CPP_BUILTINS): Define
1348 __HAVE_FPU__ if TARGET_HARD_FLOAT.
1349 (TARGET_DEFAULT): Delete.
1350 (EXTRA_SPECS): Delete cpp_cpu_default_spec, cpp_cpu_spec,
1351 cpp_fpu_spec, asm_default_spec and netbsd_cpp_spec.
1352 (CPP_CPU_SPEC): Delete.
1353 (TARGET_VERSION): Merge definitions, using TARGET_68010 to pick
1354 the appropriate string.
1355 (CPP_CPU_DEFAULT_SPEC, ASM_DEFAULT_SPEC, CPP_FPU_SPEC): Delete.
1356 (CPP_SPEC): Define to NETBSD_CPP_SPEC.
1357 (ASM_SPEC): Don't use %(asm_default_spec).
1358 * config/m68k/m68k.c (TARGET_DEFAULT_TARGET_FLAGS): Remove
1359 TARGET_DEFAULT and add MASK_68881.
1360 * config/m68k/m68k.md: Remove mention of TARGET_DEFAULT from comments.
1362 2007-01-12 Richard Sandiford <richard@codesourcery.com>
1364 * config.gcc (m68010-*-netbsdelf*): Add MASK_68010.
1365 (m68k*-*-netbsdelf*, m68k*-*-openbsd*, m68k*-linux*): Add
1366 MASK_68010 alongside MASK_68020.
1367 * doc/invoke.texi: Document -m68010.
1368 * config/m68k/m68k.opt (m68010): New.
1369 * config/m68k/m68k.h (TARGET_CPU_CPP_BUILTINS): Define mc68010
1371 (TUNE_68010): New macro.
1372 * config/m68k/m68k-none.h (M68K_CPU_m68k, M68K_CPU_m68010)
1373 (M68K_CPU_m68020, M68K_CPU_m68030, M68K_CPU_m68040)
1374 (M68K_CPU_m68332): Add MASK_68010.
1375 * config/m68k/linux.h (TARGET_DEFAULT): Add MASK_68010 to
1376 fallback definition.
1377 * config/m68k/netbsd-elf.h (CPP_CPU_SPEC): Remove now-redundant
1379 * config/m68k/m68k.c (MASK_ALL_CPU_BITS): Add MASK_68010.
1380 (m68k_handle_option): Handle OPT_m68010. Add MASK_68010
1381 to all entries that use MASK_68020.
1382 (output_move_simode_const, output_move_himode, output_move_qimode)
1383 (output_move_stricthi, output_move_strictqi): Use TARGET_68010
1384 instead of TARGET_68020 to select clr behavior. Remove comment
1385 about there being no TARGET_68010.
1386 * config/m68k/m68k.md: Likewise throughout.
1388 2007-01-12 Julian Brown <julian@codesourcery.com>
1390 * config/m68k/m68k.h (TARGET_ISAB): New macro.
1391 * config/m68k/m68k.c: Use TARGET_ISAB rather than TARGET_CFV4.
1392 * config/m68k/m68k.md: Likewise.
1394 2007-01-12 Julian Brown <julian@codesourcery.com>
1396 * config/m68k/m68k.h (LEGITIMATE_INDEX_P, LEGITIMIZE_ADDRESS): Use
1397 TARGET_COLDFIRE_FPU instead of TARGET_CFV4E.
1399 2007-01-12 Julian Brown <julian@codesourcery.com>
1401 * config/m68k/m68k.h (TUNE_68040_60): New macro.
1402 * config/m68k/m68k.c (standard_68881_constant_p): Use it.
1403 * config/m68k/m68k.md: Likewise.
1405 2007-01-12 Julian Brown <julian@codesourcery.com>
1406 Richard Sandiford <richard@codesourcery.com>
1408 * config/m68k/m68k.h (TARGET_CPU_CPP_BUILTINS): Use TUNE_68030
1409 instead of TARGET_68030, TUNE_68040 instead of TARGET_68040,
1410 TUNE_68060 instead of TARGET_68060 and TUNE_CPU32 instead of
1412 (TARGET_CPU32): Rename to...
1413 (TUNE_CPU32): ...this.
1414 (TUNE_68000_10, TUNE_68030, TUNE_68040, TUNE_68060)
1415 (TUNE_CFV2): New macros.
1416 * config/m68k/netbsd-elf.h (LONG_DOUBLE_TYPE_SIZE): Simplify;
1417 remove conditions that are implied by TARGET_68020.
1418 * config/m68k/m68k.c (m68k_output_function_prologue): Use TUNE_68040
1419 instead of TARGET_68040 and TUNE_CPU32 instead of TARGET_CPU32.
1420 (m68k_output_function_epilogue): Likewise.
1421 (m68k_rtx_costs): Likewise. Use TUNE_68060 instead of TARGET_68060
1422 and TUNE_CFV2 instead of TARGET_5200. Use TUNE_68000_10 instead of
1423 "!TARGET_68020 && !TARGET_COLDFIRE" to choose between 68000 and
1424 non-68000 timings. Refactor multiplication and division costs.
1425 (output_addsi3): Use TUNE_68040 instead of TARGET_68040 and
1426 TUNE_CPU32 instead of TARGET_CPU32.
1427 (standard_68881_constant_p): Use TUNE_68040 instead of TARGET_68040
1428 and TUNE_68060 instead of TARGET_68060.
1429 * config/m68k/m68k.md: Use TUNE_68040 instead of TARGET_68040,
1430 TUNE_68060 instead of TARGET_68060, and TUNE_CPU32 instead of
1432 (movsi_const0): Use TUNE_68000_10 rather than "!TARGET_68020
1433 && !TARGET_COLDFIRE" to choose between moveq and clr.
1434 Likewise in the unnamed movsf pattern.
1435 (ashlsi_17_24, lshrsi_17_24): Guard with TUNE_68000_10 rather than
1436 "!TARGET_68020 && !TARGET_COLDFIRE". Likewise the unnamed
1439 2007-01-12 Richard Sandiford <richard@codesourcery.com>
1441 * config/m68k/m68k.h (TARGET_CPU_CPP_BUILTINS): Increase amount
1442 of tabbing before backslashes.
1444 2007-01-11 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
1446 * pa-linux.h (ASM_OUTPUT_INTERNAL_LABEL): Undefine.
1447 * pa.h (ASM_OUTPUT_LABEL): Output colon when using GAS.
1448 (ASM_OUTPUT_INTERNAL_LABEL): Define.
1450 2007-01-11 Zdenek Dvorak <dvorakz@suse.cz>
1452 * tree-ssa-loop-ivopts.c (extract_cond_operands): Split from
1453 find_interesting_uses_cond.
1454 (find_interesting_uses_cond): Use extract_cond_operands.
1455 (rewrite_use_compare): Use extract_cond_operands and
1456 force_gimple_operand_bsi. Do not call update_stmt.
1457 (determine_use_iv_cost_condition): Use extract_cond_operands.
1458 Return cheaper of using original bound and changing the exit bound.
1460 2007-01-11 Zdenek Dvorak <dvorakz@suse.cz>
1462 PR tree-optimization/29516
1463 * tree-ssa-address.c (tree_mem_ref_addr, add_to_parts,
1464 most_expensive_mult_to_index, addr_to_parts,
1465 create_mem_ref, maybe_fold_tmr): Make the type of
1466 fields of TARGET_MEM_REF sizetype.
1467 (move_fixed_address_to_symbol, move_pointer_to_base):
1469 * tree.def (TARGET_MEM_REF): Add comment on types of
1472 2007-01-11 Joseph Myers <joseph@codesourcery.com>
1474 * c-common.c (vector_types_convertible_p): Treat opaque types as
1475 always convertible if they have the same size, but not otherwise.
1477 2007-01-11 Steven Bosscher <steven@gcc.gnu.org>
1479 * ifcvt.c (struct noce_if_info): Add comments to the fields.
1480 Remove the b_unconditional field.
1481 (noce_try_sign_mask): Do not look at b_unconditional.
1482 (noce_process_if_block): Do not use merge_if_blocks. Update
1483 the CFG here. Do not set b_unconditional.
1484 (cond_move_process_if_block): Likewise.
1485 (find_cond_trap): Likewise.
1486 (check_cond_move_block): Require simple jump insns at the end
1489 2007-01-11 Jan Hubicka <jh@suse.cz>
1491 PR tree-optimization/1046
1492 * tree-tailcall.c (suitable_for_tail_call_opt_p): Use TREE_ADDRESSABLE
1493 when alias info is not ready.
1494 (pass_tail_recursion): Do not require aliasing.
1495 * tree-ssa-copyrename.c (pass_rename_ssa_cop): Likewise.
1496 * tree-ssa-ccp.c (pass_ccp, pass_fold_builtins): Likewise.
1497 * tree-ssa-copy.c (pass_copy_prop): Likewise.
1498 * tree-ssa-forwprop.c (pass_forwprop): Likewise.
1499 * tree-ssa-dce.c (pass_dce, pass_dce_loop, pass_cd_dce): Likewise.
1500 * passes.c (init_optimization_passes): Execute rename_ssa_copies,
1501 ccp, forwprop, copy_prop, merge_phi, copy_prop, dce and tail recursion
1503 * tree-ssa-operands.c (add_virtual_operand, get_indirect_ref_operand):
1504 When aliasing is not build, mark statement as volatile.
1506 2007-01-11 Tom Tromey <tromey@redhat.com>
1508 PR preprocessor/15185, PR preprocessor/20989:
1509 * doc/cppopts.texi <-MT>: Update description of algorithm for
1510 computing default target.
1511 <-M, -MD>: Reword "basename" text.
1513 2007-01-11 Roger Sayle <roger@eyesopen.com>
1515 * builtins.c (expand_builtin_pow, expand_builtin_powi,
1516 fold_builtin_cabs, fold_builtin_sqrt, fold_builtin_trunc,
1517 fold_builtin_floor, fold_builtin_ceil, fold_builtin_round,
1518 fold_builtin_int_int_roundingfn, fold_builtin_bitop,
1519 fold_builtin_bswap, real_constp, fold_builtin_pow,
1520 fold_builtin_powi, fold_builtin_signbit, fold_builtin_copysign,
1521 do_mpfr_arg1, do_mpfr_arg2, do_mpfr_arg3, do_mpfr_sincos): Replace
1522 uses of the macro TREE_CONSTANT_OVERFLOW with TREE_OVERFLOW.
1523 * convert.c (convert_to_pointer): Likewise.
1524 * expr.c (highest_pow2_factor, expand_expr_real_1): Likewise.
1525 * fold-const.c (force_fit_type, fold_negate_expr, int_const_binop,
1526 const_binop, fold_convert_const_int_from_int,
1527 fold_convert_const_int_from_real,
1528 fold_convert_const_real_from_real, sign_bit_p,
1529 optimize_minmax_comparison, extract_muldiv_1, fold_div_compare,
1530 fold_sign_changed_comparison, fold_unary, fold_comparison,
1531 fold_binary, multiple_of_p, tree_Expr_non_zero_p,
1532 fold_negate_const, fold_abs_const, fold_not_const): Likewise.
1533 * print-tree.c (print_node_brief, print_node): Likewise.
1534 * stor-layout.c (place_field, layout_type): Likewise.
1535 * tree-chrec.c (keep_cast): Likewise.
1536 * tree.c (build_vector, build_real, build_real_from_int_cst,
1537 build_complex): Likewise.
1539 2007-01-11 Roger Sayle <roger@eyesopen.com>
1541 * tree.h (TREE_CONSTANT_OVERFLOW): Obsolete. For the time being,
1542 treat TREE_CONSTANT_OVERFLOW as a synonym of TREE_OVERFLOW.
1544 2007-01-11 Paolo Bonzini <bonzini@gnu.org>
1546 * configure.ac (strict1_warn): Rename to strict_warn.
1547 (WERROR, --enable-werror, symlink hacks, stage1_cflags,
1548 cc_set_by_configure, quoted_cc_set_by_configure,
1549 stage_prefix_set_by_configure, quoted_stage_prefix_set_by_configure,
1550 all_boot_languages, all_stagestuff): Remove.
1551 (target_list): Remove bootstrap targets.
1552 * Makefile.in (quickstrap): Unconditionally make a synonym of all.
1553 (BOOT_LANGUAGES, STAGE1_CFLAGS, STAGE1_CHECKING,
1554 REMAKEFLAGS, FLAGS_TO_PASS, PREPEND_DOTDOT_TO_RELATIVE_PATHS,
1555 SUBDIR_FLAGS_TO_PASS, WERROR_FLAGS, STRICT2_WARN, LANG_STAGESTUFF,
1556 VOL_FILES, POSTSTAGE1_FLAGS_TO_PASS, STAGE2_FLAGS_TO_PASS,
1557 STAGEPROFILE_FLAGS_TO_PASS, STAGEFEEDBACK_FLAGS_TO_PASS, stage1_build,
1558 stage1_copy, stage2_build, stage2_copy, stageprofile_build,
1559 stageprofile_copy, stage3_build, stage3_copy, stagefeedback_build,
1560 stagefeedback_copy, stage4_build, clean_s1, clean_sw, bootstrap,
1561 bootstrap-lean, bootstrap2, bootstrap2-lean, bootstrap3,
1562 bootstrap3-lean, bootstrap4, bootstrap4-lean, unstage1, unstage2,
1563 unstage3, unstage4, unstageprofile, unstagefeedback, restage, restage2,
1564 restage3, restage4, restageprofile, restagefeedback, bubbleestrap,
1565 cleanstrap, unstrap, restrap, *compare, *compare3, *compare4,
1566 *compare-lean, *compare3-lean, *compare4-lean, stage1-start, stage1,
1567 stage2-start, stage2, stage3-start, stage3, stage4-start, stage4,
1568 stageprofile-start, stageprofile, stagefeedback-start, stagefeedback,
1569 risky-stage1, risky-stage2, risky-stage3, risky-stage4): Remove.
1570 (ORDINARY_FLAGS_TO_PASS): Rename to FLAGS_TO_PASS.
1571 (STAGECOPYSTUFF, STAGEMOVESTUFF): Consolidate into MOSTLYCLEANFILES.
1572 (mostlyclean): Adjust.
1573 (clean, distclean): Don't mention bootstrap stuff.
1574 * configure: Regenerate.
1575 * ada/config-lang.in, cp/config-lang.in, forttran/config-lang.in,
1576 java/config-lang.in, objc/config-lang.in, objcp/config-lang.in,
1577 treelang/config-lang.in (stagestuff): Remove.
1578 * doc/sourcebuild.texi (stage1, stage2, stage3, stage4,
1579 stageprofile, stagefeedback, stagestuff): Remove mention.
1581 2007-01-11 Nick Clifton <nickc@redhat.com>
1583 * config/mcore/predicates.md (mcore_general_movesrc_operand):
1585 (mcore_general_movdst_operand): Do not accept CONST_INTs.
1586 (mcore_arith_K_S_operand): Run the test for the S constraint not
1587 the test for the M constraint.
1588 (mcore_addsub_operand): Do not accept integer values that are
1589 larger than 32 bits.
1590 * config/mcore/mcore.md: Remove unused constraints from split.
1591 (andsi3): Use HOST_WIDE_INT instead of int to hold an INTVAL.
1593 (allocate_stack): Likewise.
1594 * config/mcore/mcore.c (mcore_print_operand): Restrict output of P
1595 operands to 32 bits.
1596 (mcore_const_costs): Use HOST_WIDE_INT instead of int to hold an
1598 (mcore_and_cost, mcore_modify_comparison, const_ok_for_mcore,
1599 mcore_const_ok_for_inline, mcore_const_trick_uses_not,
1600 try_constant_tricks, mcore_num_ones, mcore_num_zeros,
1601 mcore_output_bclri, mcore_output_andn, output_inline_const,
1602 mcore_output_move, mcore_output_movedouble): Likewise.
1603 (mcore_output_cmov): Use CONST_OK_FOR_M and CONST_OK_FOR_N.
1604 (output_inline_const): Likewise.
1605 (output_inline_const): Fix format strings used in sprintf
1607 * config/mcore/mcore-protos.h: Update prototypes for changed
1608 functions in mcore.c.
1609 * config/mcore/mcore.h (CONST_OK_FOR_I): Cast values to
1610 HOST_WIDE_INT and not int.
1611 (CONST_OK_FOR_J, CONST_OK_FOR_K, CONST_OK_FOR_L, CONST_OK_FOR_M,
1612 CONST_OK_FOR_N): Likewise.
1613 (LEGITIMATE_CONSTANT_P): Also check CONSTANT_P.
1614 (GO_IF_LEGITIMATE_INDEX): Use HOST_WIDE_INT instead of int to hold
1617 2007-01-10 Jan Hubicka <jh@suse.cz>
1619 * tree-vrp.c (remove_range_assertions): Release defs.
1620 * tree-ssa-loop-ivopts.c (rmeove_statement): Likewise.
1621 * tree-ssa-dom.c (remove_stmt_or_phi): Likewise.
1623 2007-01-10 Paul Brook <paul@codesourcery.com>
1625 * config/arm/arm.c (arm_rtx_costs_1): Handle mutiply-subtract.
1626 * config/arm/arm.md (mulsi3subsi): New insn.
1628 2007-01-10 Zdenek Dvorak <dvorakz@suse.cz>
1630 * tree-ssa-loop-manip.c (tree_unroll_loop): Make it a wrapper over ...
1631 (tree_transform_and_unroll_loop): New.
1632 * tree-flow.h (transform_callback, tree_transform_and_unroll_loop):
1635 2007-01-10 Robert Kennedy <jimbob@google.com>
1637 * fold-const.c (fold_comparison): Fold comparisons like (x *
1638 1000 < 0) to (x < 0).
1640 2007-01-10 Ian Lance Taylor <iant@google.com>
1642 * tree-pretty-print.c (dump_generic_node): Print parentheses when
1643 operands have the same priority.
1645 2007-01-10 Tom Tromey <tromey@redhat.com>
1647 * fold-const.c (fold_truthop): Don't check can_use_bit_fields_p.
1648 (fold_binary): Likewise.
1649 * langhooks.c (lhd_can_use_bit_fields_p): Removed.
1650 * langhooks-def.h (lhd_can_use_bit_fields_p): Removed.
1651 (LANG_HOOKS_CAN_USE_BIT_FIELDS_P): Removed.
1652 (LANG_HOOKS_INITIALIZER): Remove LANG_HOOKS_CAN_USE_BIT_FIELDS_P.
1653 * langhooks.h (struct lang_hooks): Removed field
1654 'can_use_bit_fields_p'.
1656 2007-01-10 Ralf Corsépius <ralf.corsepius@rtems.org>
1658 * config/bfin/t-bfin, config/bfin/t-bfin-elf: Remove GCC_CFLAGS.
1660 2007-01-10 Razya Ladelsky <razya@il.ibm.com>
1662 * function.c (get_last_funcdef_no): New function.
1663 * function.h (get_last_funcdef_no): Declare.
1664 * tree-inline.c (initialize_cfun): Add initialization.
1665 (tree_function_versioning): Cleanup.
1667 2007-01-10 Jan Hubicka <jh@suse.cz>
1669 * tree-inline.c (setup_one_parameter): Do not propagate into abnormal
1672 2007-01-10 Sa Liu <saliu@de.ibm.com>
1673 Ben Elliston <bje@au.ibm.com>
1675 * spu.h (STACK_SAVE_AREA): Use VOIDmode for SAVE_FUNCTION, SImode
1676 for SAVE_NONLOCAL and Pmode for any other save level.
1677 * spu-protos.h (spu_restore_stack_block): Declare.
1678 * spu.md (save_stack_block): Remove.
1679 (restore_stack_block): Call spu_restore_stack_block.
1680 * spu.c (spu_restore_stack_block): New function.
1681 (spu_expand_epilogue): Remove old comment.
1683 2007-01-09 Zdenek Dvorak <dvorakz@suse.cz>
1685 PR tree-optimization/30322
1686 * tree-ssa-loop-ivopts.c (fold_affine_expr, iv_value): Removed.
1687 (cand_value_at): Return the value as aff_tree.
1688 (may_eliminate_iv): Convert the bound from aff_tree to tree.
1689 * tree-affine.c (aff_combination_add_cst, aff_combination_add_product,
1690 aff_combination_mult): New functions.
1691 (aff_combination_add): Use aff_combination_add_cst.
1692 (aff_combination_convert): Allow conversions to a wider type.
1693 (tree_to_aff_combination): Handle BIT_NOT_EXPR.
1694 * tree-affine.h (aff_combination_mult): Declare.
1696 2007-01-09 Carlos O'Donell <carlos@codesourcery.com>
1698 * doc/tm.texi: Update documentation to reflect reality of exec
1699 and start file search behaviours. Update copyright year.
1700 * doc/invoke.texi: Explain how GCC_EXEC_PREFIX is used to find
1701 header file directories.
1703 2007-01-09 Uros Bizjak <ubizjak@gmail.com>
1705 * config/i386/i386.md (*sinxf2): Rename to *sinxf2_i387.
1706 (*cosxf2): Rename to cosxf2_i387.
1707 (*sindf2, *sinsf2): Extend operand 1 to XFmode. Macroize patterns
1708 using X87MODEF12 mode macro. Rename patterns to
1709 *sin_extend<mode>xf2_i387. Use SSE_FLOAT_MODE_P to disable patterns
1711 (*cosdf2, *cossf2): Ditto.
1712 (sincosdf3, sincossf3): Ditto. Rewrite corresponding splitters
1713 to match extended input operands.
1714 (sincos<mode>3): New expander.
1715 (*sinextendsfdf2, *cosextendsfdf2, *sincosextendsfdf3): Remove
1716 insn patterns and corresponding splitters.
1718 2007-01-09 Kaz Kojima <kkojima@gcc.gnu.org>
1720 * config/sh/t-linux (TARGET_LIBGCC2_CFLAGS): Delete.
1721 (SHLIB_MAPFILES, SHLIB_LINK, SHLIB_INSTALL): Likewise.
1723 2007-01-09 Nicolas Pitre <nico@cam.org>
1726 * arm/ieee754-df.S (Lad_s): Also test the low word of X for zero.
1728 2007-01-08 Geoffrey Keating <geoffk@apple.com>
1730 * target.h (struct gcc_target): New field library_rtti_comdat.
1731 * target-def.h (TARGET_CXX_LIBRARY_RTTI_COMDAT): New.
1732 (TARGET_CXX): Add TARGET_CXX_LIBRARY_RTTI_COMDAT.
1733 * doc/tm.texi (C++ ABI): Document TARGET_CXX_LIBRARY_RTTI_COMDAT.
1734 * config/darwin.h (TARGET_CXX_LIBRARY_RTTI_COMDAT): Define.
1736 2007-01-08 Geoffrey Keating <geoffk@apple.com>
1738 * doc/invoke.texi (Optimize Options): Correct description of -O0.
1740 2007-01-08 Richard Guenther <rguenther@suse.de>
1742 * tree.h (force_fit_type_double): Export.
1743 (force_fit_type): Remove.
1744 * fold-const.c (force_fit_type_double): New function.
1745 (force_fit_type): Remove.
1746 (int_const_binop): Use it.
1747 (fold_convert_const_int_from_int): Likewise.
1748 (fold_convert_const_int_from_real): Likewise.
1749 (fold_div_compare): Likewise.
1750 (fold_sign_changed_comparison): Likewise.
1751 (fold_unary): Likewise.
1752 (fold_negate_const): Likewise.
1753 (fold_abs_const): Likewise.
1754 (fold_not_const): Likewise.
1755 * c-common.c (shorten_compare): Use force_fit_type_double.
1756 * convert.c (convert_to_pointer): Likewise.
1758 2007-01-08 Richard Guenther <rguenther@suse.de>
1760 * tree.h (build_int_cst_wide_type): Export.
1761 * tree.c (build_int_cst_wide_type): New function.
1762 (build_int_cst_wide): Fix comment.
1763 * builtins.c (fold_builtin_object_size): Use build_int_cst
1764 to build -1 or 0 of the correct type. Use fit_double_type
1765 to check for overflow.
1766 * fold-const.c (optimize_bit_field_compare): Use build_int_cst_type
1768 (decode_field_reference): Likewise.
1769 (all_ones_mask_p): Likewise.
1770 (native_interpret_int): Use build_int_cst_wide_type.
1771 (fold_binary): Use build_int_cst_type to build an all-ones
1773 * stor-layout.c (set_sizetype): Use build_int_cst_wide_type.
1775 2007-01-08 Daniel Jacobowitz <dan@codesourcery.com>
1777 * config/pa/t-pa64 (libgcc_stub.a): Use $(T).
1779 2007-01-09 Ben Elliston <bje@au.ibm.com>
1781 * genautomata.c (STATS_OPTION): New option.
1782 (stats_flag): New flag.
1783 (gen_automata_option): Handle it.
1784 (initiate_automaton_gen): Ditto.
1785 (write_automata): Output statistics only if stats_flag is
1786 set. Likewise, output time statistics only if time_flag is set.
1787 * doc/md.texi (Processor pipeline description): Document new flag.
1789 2007-01-08 Richard Guenther <rguenther@suse.de>
1791 * builtins.c (fold_builtin_int_roundingfn): Use fit_double_type.
1792 * tree.c (build_int_cst_type): Likewise.
1793 (size_in_bytes): Don't call force_fit_type on the result.
1794 (int_fits_type_p): Use fit_double_type.
1795 * fold-const.c (fit_double_type): New function.
1796 (force_fit_type): Use it.
1797 * tree.h (fit_double_type): Export.
1799 2007-01-08 Jan Hubicka <jh@suse.cz>
1801 * tree-vectorizer.c (gate_increase_alignment): Fix return type.
1802 * ipa.c (function_and_variable_visibility): Fix return type.
1804 2007-01-08 Richard Guenther <rguenther@suse.de>
1806 * tree-ssa-ccp.c (maybe_fold_offset_to_array_ref): Use type
1807 of offset to build the index.
1808 * tree-pretty-print.c (dump_generic_node): Don't build negated
1809 const just for printing.
1810 * c-pretty-print.c (pp_c_integer_constant): Likewise.
1811 * builtins.c (fold_builtin_int_roundingfn): Check if result
1812 fits the type by using force_fit_type and comparing the result.
1813 * predict.c (predict_loops): Use compare_tree_int for comparison.
1814 * tree.c (build_int_cst): Fall back to integer_type_node for
1816 (build_int_cst_wide): Assert type is non-null.
1818 2007-01-08 Roberto Costa <roberto.costa@st.com>
1820 * tree-vrp.c (extract_range_from_cond_expr): New.
1821 (extract_range_from_expr): Handle COND_EXPR nodes used as expressions.
1822 * tree-ssa-ccp.c (get_maxval_strlen): Handle COND_EXPR nodes used
1824 (fold_stmt): Bug fix, avoid infinite recursion when folding COND_EXPRs.
1825 * tree-ssa-forwprop.c (simplify_cond, forward_propagate_into_cond,
1826 tree_ssa_forward_propagate_single_use_vars): Handle COND_EXPR nodes
1827 used as expressions.
1828 * tree-object-size.c (cond_expr_object_size): New.
1829 (collect_object_sizes_for): Handle COND_EXPR nodes used as expressions.
1831 2007-01-08 Jan Hubicka <jh@suse.cz>
1833 * tree-ssa-forwprop.c (forward_propagate_into_cond,
1834 tree_ssa_forward_propagate_single_use_va): Release defs of propagated
1837 2007-01-08 Richard Guenther <rguenther@suse.de>
1839 PR tree-optimization/23603
1840 * tree-vrp.c (set_value_range_to_truthvalue): New function.
1841 (extract_range_from_binary): Fall back to truthvalue instead of
1842 varying for TRUTH_*_EXPR.
1843 (extract_range_from_comparison): Fall back to truthvalue instead of
1845 (vrp_visit_phi_node): Don't adjust new range bounds to +INF/-INF
1846 if all visited PHI values were constant.
1848 2007-01-08 Jan Hubicka <jh@suse.cz>
1850 * cgraphunit.c (cgraph_process_new_functions): Reset reachable flag.
1851 (cgraph_analyze_function): break out from ...
1852 (cgraph_finalize_compilation_unit): ... here.
1853 (cgraph_expand_function): Remove forgoten commented out line.
1854 (cgraph_optimize): Analyze functions.
1856 2007-01-08 Jan Hubicka <jh@suse.cz>
1858 * tree-pas.h (TODO_remove_function): New flag.
1859 (TODO_update*): Renumber.
1860 (pass_ipa_increase_alignment,
1861 pass_ipa_function_and_variable_visibility): New passes.
1862 * cgraphunit.c (cgraph_increase_alignment): Move to tree-vectorizer.c
1863 (cgraph_function_and_variable_visibility): Move to ipa.c
1864 (cgraph_optimize): Don't call cgraph_function_and_variable_visibility,
1865 cgraph_increase_alignment.
1866 * ipa-inline.c (cgraph_decide_inlining): Don't push timevar.
1867 (cgraph_decide_inlining_incrementally): Push TV_INTEGRATION before
1868 calling tree-inline.
1869 (cgraph_early_inlining): Do not call cgraph_remove_unreachable_nodes.
1870 (pass_ipa_inline, pass_early_ipa_inlining): Set TODO_remove_functions
1871 * tree-vectorizer.c (increase_alignment): Move here from cgraphunit.c
1872 (gate_increase_alignment): New function.
1873 (pass_ipa_increase_alignment): New pass.
1874 * ipa.c: Inline tree-pass.h and timevar.h
1875 (function_and_variable_visibility): Move here from cgraphunit.c
1876 * tree-optimize.c (pass_early_local_passes): Add TODO_remove_functions.
1877 * passes.c (init_optimization_passes): Add the two new passes.
1878 (execute_todo): Handle cgraph_remove_functions.
1880 2007-01-08 Nick Clifton <nickc@redhat.com>
1882 * config/frv/predicates.md (reg_or_0_operand): Accept
1885 2007-01-08 Ralf Corsépius <ralf.corsepius@rtems.org>
1887 * config/bfin/rtems.h, config/bfin/t-rtems: New.
1888 * config.gcc: Add bfin*-rtems*.
1890 2007-01-08 Mark Shinwell <shinwell@codesourcery.com>
1892 * c.opt: Add -flax-vector-conversions.
1893 * c-typeck.c (convert_for_assignment): Pass flag to
1894 vector_types_convertible_p to allow emission of note.
1895 (digest_init): Likewise.
1896 * c-opts.c: Handle -flax-vector-conversions.
1897 * c-common.c (flag_lax_vector_conversions): New.
1898 (vector_types_convertible_p): Unless -flax-vector conversions
1899 has been passed, disallow conversions between vectors with
1900 differing numbers of subparts and/or element types. If such
1901 a conversion is disallowed, possibly emit a note on the first
1902 occasion only to inform the user of -flax-vector-conversions.
1903 The new last argument specifies this.
1904 * c-common.h (flag_lax_vector_conversions): New.
1905 (vector_types_convertible_p): Add extra argument.
1906 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use
1907 char_type_node for V*QI type vectors.
1908 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
1909 Update to satisfy new typechecking rules.
1910 * config/rs6000/altivec.h (vec_cmple): Use vec_cmpge, for both
1912 * doc/invoke.texi (C Dialect Options): Document
1913 -flax-vector-conversions.
1915 2007-01-08 Mark Shinwell <shinwell@codesourcery.com>
1917 PR tree-optimization/29877
1918 * tree-ssa-ter.c (is_replaceable_p): Deem assignments with
1919 a register variable on the RHS to not be replaceable.
1921 2007-01-08 Chen Liqin <liqin@sunnorth.com.cn>
1922 * config/score/t-score-elf (MULTILIB_OPTIONS): Change.
1923 * config/score/predicates.md (const_uimm5, sr0_operand, const_simm12,
1924 const_simm15, const_pow2, const_npow2): Added.
1925 * config/score/misc.md (insv, extv, extzv, movmemsi,
1926 move_lbu_a/b, mov_lhu_a/b etc): Added and fix some bug.
1927 * config/score/score.c (score_address_cost, score_select_cc_mode):
1929 Change CONST_OK_FOR_LETTER_P/EXTRA_CONSTRAINT define.
1930 Update score_rtx_costs for MACRO TARGET_RTX_COSTS.
1931 Update score_print_operand.
1932 * config/score/score.h (DATA_ALIGNMENT, SELECT_CC_MODE): Added.
1933 Adjust register allocate order and update some macro define.
1934 * config/score/score-mdaux.c (mdx_unaligned_load, mdx_unsigned_store,
1935 mdx_block_move_straight, mdx_block_move_loop_head,
1936 mdx_block_move_loop_body, mdx_block_move_loop_foot, mdx_block_move_loop,
1937 mdx_block_move): Added.
1938 (mdx_movsicc, mdp_select_add_imm, mdp_select, mds_zero_extract_andi,
1939 mdp_limm): Updated and fix some bug and typo.
1940 * config/score/score.md (movqi/hi/si, add/sub/zero/ext): Updated.
1941 (movsf, movdf, doloop_end): Added.
1943 2007-01-08 Kazu Hirata <kazu@codesourcery.com>
1945 * config/arm/arm.c, config/arm/arm.h, config/arm/arm.md,
1946 config/arm/thumb2.md: Fix comment typos.
1947 * doc/extend.texi: Fix a typo.
1949 2007-01-07 Eric Christopher <echristo@apple.com>
1951 * configure.ac: Check for __stack_chk_fail for darwin.
1952 * configure: Regenerate.
1954 2007-01-07 Richard Guenther <rguenther@suse.de>
1956 * tree-vrp.c (extract_range_from_assert): CSE calls to
1957 compare_values where possible.
1958 (extract_range_from_unary_expr): Likewise.
1960 2007-01-07 Anatoly Sokolov <aesok@post.ru>
1962 * config/avr/avr-protos.h (call_insn_operand): Delete prototype.
1963 * config/avr/avr.c (call_insn_operand): Delete function.
1964 * config/avr/avr.md (*pushqi, *pushhi, *pushsi, *pushsf): Use REG_SP
1965 instead of register number. Use predicates.
1966 * config/avr/predicates.md (const0_operand, reg_or_0_operand,
1967 call_insn_operand): Add.
1969 2007-01-06 Jan Hubicka <jh@suse.cz>
1971 * tree-pass.h (pass_build_cgraph_edges): Declare.
1972 * cgraphunit.c (record_refernece): Move to cgraphbuild.c
1973 (visited_nodes): Remove.
1974 (cgraph_create_edges): Move to cgraphbuild.c; rename to
1975 build_cgrpah_edges; make visited_nodes local.
1976 (cgraph_process_new_functions): DO not call initialize_inline_failed.
1977 (record_references_in_initializer): Move to cgraphbuild.c
1978 (initialize_inline_failed, rebuild_cgraph_edges,
1979 pass_rebuild_cgraph_edges): Move to cgraphbuild.c.
1980 (verify_cgraph_node): Make visited_nodes local.
1981 (cgraph_analyze_function): Do not call cgraph_create_edges and
1982 initialize_inline_failed.
1983 (cgraph_expand_function): Do not call cgraph_lower_function;
1984 assert that function is already lowered.
1985 * Makefile.in (cgraphbuild.o): New.
1986 * passes.c (init_optimization_passes): Add pass_build_cgraph_edges
1987 at the end of lowering passes.
1989 2007-01-06 Steven Bosscher <steven@gcc.gnu.org>
1991 * ifcvt.c (cond_move_convert_if_block): New function, code
1992 factored out from...
1993 (cond_move_process_if_block): ...here. Call the new function
1994 on the THEN and ELSE blocks.
1995 (merge_if_block): Do not copy global_live_at_end, merge_blocks
1996 already takes care of this.
1998 2007-01-05 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
2001 * tree.h (TREE_OVERFLOW_P): New.
2002 * c-typeck.c (parser_build_unary_op): Warn only if result
2003 overflowed and operands did not.
2004 (parser_build_binary_op): Likewise.
2005 (convert_for_assignment): Remove redundant overflow_warning.
2006 * c-common.c (overflow_warning): Don't check or set TREE_OVERFLOW.
2008 2007-01-05 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
2010 * c-typeck.c (store_init_value): Split over two lines to follow
2011 the GNU coding style.
2013 2007-01-05 Benjamin Kosnik <bkoz@redhat.com>
2015 * c-cppbuiltin.c (c_cpp_builtins): __GXX_EXPERIMENTAL_CPP0X__ to
2016 __GXX_EXPERIMENTAL_CXX0X__.
2017 * doc/cpp.texi: Same.
2019 2007-01-05 Richard Guenther <rguenther@suse.de>
2022 * tree.c (get_narrower): Do not construct COMPONENT_REFs
2023 with mismatched types. Instead explicitly build a
2024 conversion NOP_EXPR.
2026 2007-01-05 Ian Lance Taylor <iant@google.com>
2028 * c-common.c (decl_with_nonnull_addr_p): New function.
2029 (c_common_truthvalue_conversion): Call it.
2030 * c-typeck.c (build_binary_op): Likewise.
2031 * c-common.h (decl_with_nonnull_addr_p): Declare.
2033 2007-01-05 Jakub Jelinek <jakub@redhat.com>
2036 * libgcc2.c (__divdc3): Compare c and d against 0.0 instead of
2039 2007-01-05 Joel Brobecker <brobecker@adacore.com>
2041 * doc/install.texi (Final install): Document the fact that
2042 the GNAT runtime should not be stripped.
2044 2007-01-04 Jan Hubicka <jh@suse.cz>
2046 * tree-inline.c (fold_marked_statements): Update operand caches
2047 and EH after folding
2049 2007-01-04 Ian Lance Taylor <iant@google.com>
2051 * c-common.c (check_function_nonnull): Whitespace fix.
2053 2007-01-04 Jan Hubicka <jh@suse.cz>
2055 * tree-optimize.c (execute_fixup_cfg): Correct previously mistakely
2056 comitted older version of patch.
2057 (pass_fixup_cfg): Add TODOs to verify flow and statements, dump
2058 function, celanup cfg and collect garbage.
2060 2007-01-04 Mike Stump <mrs@apple.com>
2062 * Makefile.in (mostlyclean): Don't remove libgcc anymore.
2065 2007-01-04 Eric Christopher <echristo@apple.com>
2067 * libgcc2.c (__bswapsi2): Use SItype.
2068 (__bswapdi2): Use DItype.
2069 * libgcc2.h: Update for above.
2071 2007-01-04 Paul Brook <paul@codesourcery.com>
2073 * config/arm/arm.md (arm_mulsi3, thumb_mulsi3, mulsi3_compare0,
2074 mulsi_compare0_scratch, mulsi3addsi, mulsi3addsi_compare0,
2075 mulsi3addsi_compare0_scratch, mulsidi3adddi, mulsidi3,
2076 umulsidi3, umulsidi3adddi, smulsi3_highpart,
2077 umulsi3_highpart): Make conditional on !arm_arch6.
2078 (arm_mulsi3_v6, thumb_mulsi3_v6, mulsi3_compare0_v6,
2079 mulsi_compare0_scratch_v6, mulsi3addsi_v6, mulsi3addsi_compare0_v6,
2080 mulsi3addsi_compare0_scratch_v6, mulsidi3adddi_v6, mulsidi3_v6,
2081 umulsidi3_v6, umulsidi3adddi_v6, smulsi3_highpart_v6,
2082 umulsi3_highpart_v6): New insns.
2084 2007-01-04 Roger Sayle <roger@eyesopen.com>
2086 * fold-const.c (fold_convert): When casting an expression to void,
2087 fold_ignored_result may discover a GIMPLE_MODIFY_STMT which doesn't
2088 have a type. Instead of attempting to build a NOP_EXPR, return
2089 these "special" trees directly.
2091 2007-01-04 Joseph Myers <joseph@codesourcery.com>
2093 * config/rs6000/rs6000.c (rs6000_rtx_costs): Make adjustment for
2094 MULT inside MINUS as either argument. Use rs6000_cost->dmul -
2095 rs6000_cost->fp not 0 as adjustment for outer NEG.
2097 2007-01-04 Jan Hubicka <jh@suse.cz>
2099 * cgraph.c (cgraph_release_function_body): New function.
2100 (cgraph_remove_node): Use it.
2101 * cgraph.h (cgraph_release_function_body): Declare.
2102 * cgraphunit.c (cgraph_expand_function): Use it.
2103 * ipa.c (cgraph_remove_unreahchable_nodes): Use it.
2104 * tree-ssa.c (delete_tree_ssa): Allow to be called before aliasing
2105 is initialized and while compilation of other function is running.
2106 * tree-optimize.c (execute_free_cfg_annotations): Move code to clear
2107 statement CFG annotations from here to ...
2108 * tree-cfg.c (delete_tree_cfg_annotations): ... here.
2110 2007-01-04 Zdenek Dvorak <dvorakz@suse.cz>
2112 * cfgloop.h (enum li_flags): Make the constants powers of two.
2114 2007-01-04 Jan Hubicka <jh@suse.cz>
2116 * tree-inline.c (copy_bb): Insert new statements to statements_to_fold
2118 (fold_marked_statements): New function.
2119 (optimize_inline_calls, tree_function_versioning): Fold new statements.
2120 * tree-inline.h (copy_body_data): Add statements_to_fold.
2122 2007-01-03 Daniel Jacobowitz <dan@codesourcery.com>
2124 * config.gcc: Mention libgcc/config.host.
2125 * Makefile.in: Update comments mentioning libgcc.
2126 (LIBGCC, INSTALL_LIBGCC, GCC_PARTS, mklibgcc): Delete.
2127 (all.cross, start.encap, rest.encap, rest.cross): Update
2128 dependencies for libgcc move.
2129 (libgcc.mk, LIBGCC_DEPS, libgcov.a, libgcc.a, stmp-multilib)
2130 (clean-target, clean-target-libgcc): Delete.
2131 (srcdirify, GCC_EXTRA_PARTS): New macros.
2132 (libgcc-support, libgcc.mvars): New rules.
2133 (distclean): Remove mention of mklibgcc.
2134 (install): Don't reference INSTALL_LIBGCC.
2135 (install-common): Don't reference EXTRA_PARTS.
2136 (install-libgcc, install-multilib): Delete rules.
2137 * mklibgcc.in: Delete file.
2138 * doc/configfiles.texi: Don't mention mklibgcc.
2140 * config/i386/t-darwin (SHLIB_VERPFX): Delete (moved to libgcc).
2141 * config/i386/t-darwin64 (SHLIB_VERPFX): Likewise.
2142 * config/rs6000/t-darwin (SHLIB_VERPFX): Likewise.
2143 * config/rs6000/t-ppccomm (TARGET_LIBGCC2_CFLAGS, SHLIB_MAPFILES)
2144 (mklibgcc, ldblspecs): Likewise.
2146 * config/i386/t-nwld (libgcc.def, libc.def, libpcre.def)
2147 (posixpre.def): Use $(T).
2148 (SHLIB_EXT, SHLIB_NAME, SHLIB_SLIBDIR_QUAL, SHLIB_DEF, SHLIB_MAP)
2149 (SHLIB_SRC, SHLIB_INSTALL): Delete.
2150 (SHLIB_LINK): Make dummy.
2151 * config/t-slibgcc-darwin: Delete contents except for dummy SHLIB_LINK.
2153 * config/frv/t-linux (EXTRA_MULTILIB_PARTS): Clear.
2155 * config/alpha/t-crtfm: Use $(T) in rules for EXTRA_PARTS.
2156 * config/alpha/t-vms, config/alpha/t-vms64, config/fr30/t-fr30,
2157 config/i386/t-rtems-i386, config/ia64/t-ia64, config/rs6000/t-beos,
2158 config/rs6000/t-newas, config/sparc/t-elf: Likewise.
2160 * configure.ac (all_outputs): Remove mklibgcc.
2161 * configure: Regenerated.
2163 2007-01-03 Josh Conner <jconner@apple.com>
2166 * calls.c (compute_argument_addresses): Set stack and stack_slot
2167 for partial args, too.
2168 (store_one_arg): Use locate.size.constant for the size when
2169 generating a save_area.
2171 2007-01-03 Robert Kennedy <jimbob@google.com>
2173 * tree-cfg.c (tree_merge_blocks): Release SSA_NAME phi results
2174 whose definitions are deleted due to basic block merging.
2176 2007-01-03 Paul Brook <paul@codesourcery.com>
2179 * config/arm/arm.c (output_return_instruction): Pop PC in interrupt
2181 (use_return_insn): Return 0 for Thumb interrupt functions.
2182 (print_multi_reg): Add rfe argument for IRQ returns.
2183 (arm_output_epilogue): Pop interrupt return address directly into PC.
2184 (arm_expand_prologue): Only adjust IRQ return address in Arm mode.
2186 2007-01-03 Paul Brook <paul@codesourcery.com>
2188 Merge from sourcerygxx-4_1.
2189 * config/arm/thumb2.md: New file.
2190 * config/arm/elf.h (JUMP_TABLES_IN_TEXT_SECTION): Return True for
2192 * config/arm/coff.h (JUMP_TABLES_IN_TEXT_SECTION): Ditto.
2193 * config/arm/aout.h (ASM_OUTPUT_ADDR_VEC_ELT): Add !Thumb-2 assertion.
2194 (ASM_OUTPUT_ADDR_DIFF_ELT): Output Thumb-2 jump tables.
2195 * config/arm/aof.h (ASM_OUTPUT_ADDR_DIFF_ELT): Output Thumb-2 jump
2197 (ASM_OUTPUT_ADDR_VEC_ELT): Add !Thumb-2 assertion.
2198 * config/arm/ieee754-df.S: Use macros for Thumb-2/Unified asm
2200 * config/arm/ieee754-sf.S: Ditto.
2201 * config/arm/arm.c (thumb_base_register_rtx_p): Rename...
2202 (thumb1_base_register_rtx_p): ... to this.
2203 (thumb_index_register_rtx_p): Rename...
2204 (thumb1_index_register_rtx_p): ... to this.
2205 (thumb_output_function_prologue): Rename...
2206 (thumb1_output_function_prologue): ... to this.
2207 (thumb_legitimate_address_p): Rename...
2208 (thumb1_legitimate_address_p): ... to this.
2209 (thumb_rtx_costs): Rename...
2210 (thumb1_rtx_costs): ... to this.
2211 (thumb_compute_save_reg_mask): Rename...
2212 (thumb1_compute_save_reg_mask): ... to this.
2213 (thumb_final_prescan_insn): Rename...
2214 (thumb1_final_prescan_insn): ... to this.
2215 (thumb_expand_epilogue): Rename...
2216 (thumb1_expand_epilogue): ... to this.
2217 (arm_unwind_emit_stm): Rename...
2218 (arm_unwind_emit_sequence): ... to this.
2219 (thumb2_legitimate_index_p, thumb2_legitimate_address_p,
2220 thumb1_compute_save_reg_mask, arm_dwarf_handle_frame_unspec,
2221 thumb2_index_mul_operand, output_move_vfp, arm_shift_nmem,
2222 arm_save_coproc_regs, thumb_set_frame_pointer, arm_print_condition,
2223 thumb2_final_prescan_insn, thumb2_asm_output_opcode, arm_output_shift,
2224 thumb2_output_casesi): New functions.
2225 (TARGET_DWARF_HANDLE_FRAME_UNSPEC): Define.
2226 (FL_THUMB2, FL_NOTM, FL_DIV, FL_FOR_ARCH6T2, FL_FOR_ARCH7,
2227 FL_FOR_ARCH7A, FL_FOR_ARCH7R, FL_FOR_ARCH7M, ARM_LSL_NAME,
2228 THUMB2_WORK_REGS): Define.
2229 (arm_arch_notm, arm_arch_thumb2, arm_arch_hwdiv, arm_condexec_count,
2230 arm_condexec_mask, arm_condexec_masklen)): New variables.
2231 (all_architectures): Add armv6t2, armv7, armv7a, armv7r and armv7m.
2232 (arm_override_options): Check new CPU capabilities.
2233 Set new architecture flag variables.
2234 (arm_isr_value): Handle v7m interrupt functions.
2235 (user_return_insn): Return 0 for v7m interrupt functions. Handle
2237 (const_ok_for_arm): Handle Thumb-2 constants.
2238 (arm_gen_constant): Ditto. Use movw when available.
2239 (arm_function_ok_for_sibcall): Return false for v7m interrupt
2241 (legitimize_pic_address, arm_call_tls_get_addr): Handle Thumb-2.
2242 (thumb_find_work_register, arm_load_pic_register,
2243 legitimize_tls_address, arm_address_cost, load_multiple_sequence,
2244 emit_ldm_seq, emit_stm_seq, arm_select_cc_mode, get_jump_table_size,
2245 print_multi_reg, output_mov_long_double_fpa_from_arm,
2246 output_mov_long_double_arm_from_fpa, output_mov_double_fpa_from_arm,
2247 output_mov_double_fpa_from_arm, output_move_double,
2248 arm_compute_save_reg_mask, arm_compute_save_reg0_reg12_mask,
2249 output_return_instruction, arm_output_function_prologue,
2250 arm_output_epilogue, arm_get_frame_offsets, arm_regno_class,
2251 arm_output_mi_thunk, thumb_set_return_address): Ditto.
2252 (arm_expand_prologue): Handle Thumb-2. Use arm_save_coproc_regs.
2253 (arm_coproc_mem_operand): Allow POST_INC/PRE_DEC.
2254 (arithmetic_instr, shift_op): Use arm_shift_nmem.
2255 (arm_print_operand): Use arm_print_condition. Handle '(', ')', '.',
2257 (arm_final_prescan_insn): Use extract_constrain_insn_cached.
2258 (thumb_expand_prologue): Use thumb_set_frame_pointer.
2259 (arm_file_start): Output directive for unified syntax.
2260 (arm_unwind_emit_set): Handle stack alignment instruction.
2261 * config/arm/lib1funcs.asm: Remove default for __ARM_ARCH__.
2262 Add v6t2, v7, v7a, v7r and v7m.
2263 (RETLDM): Add Thumb-2 code.
2264 (do_it, shift1, do_push, do_pop, COND, THUMB_SYNTAX): New macros.
2265 * config/arm/arm.h (TARGET_CPU_CPP_BUILTINS): Define __thumb2__.
2266 (TARGET_THUMB1, TARGET_32BIT, TARGET_THUMB2, TARGET_DSP_MULTIPLY,
2267 TARGET_INT_SIMD, TARGET_UNIFIED_ASM, ARM_FT_STACKALIGN, IS_STACKALIGN,
2268 THUMB2_TRAMPOLINE_TEMPLATE, TRAMPOLINE_ADJUST_ADDRESS,
2269 ASM_OUTPUT_OPCODE, THUMB2_GO_IF_LEGITIMATE_ADDRESS,
2270 THUMB2_LEGITIMIZE_ADDRESS, CASE_VECTOR_PC_RELATIVE,
2271 CASE_VECTOR_SHORTEN_MODE, ADDR_VEC_ALIGN, ASM_OUTPUT_CASE_END,
2272 ADJUST_INSN_LENGTH): Define.
2273 (TARGET_REALLY_IWMMXT, TARGET_IWMMXT_ABI, CONDITIONAL_REGISTER_USAGE,
2274 STATIC_CHAIN_REGNUM, HARD_REGNO_NREGS, INDEX_REG_CLASS,
2275 BASE_REG_CLASS, MODE_BASE_REG_CLASS, SMALL_REGISTER_CLASSES,
2276 PREFERRED_RELOAD_CLASS, SECONDARY_OUTPUT_RELOAD_CLASS,
2277 SECONDARY_INPUT_RELOAD_CLASS, LIBCALL_VALUE, FUNCTION_VALUE_REGNO_P,
2278 TRAMPOLINE_SIZE, INITIALIZE_TRAMPOLINE, HAVE_PRE_INCREMENT,
2279 HAVE_POST_DECREMENT, HAVE_PRE_DECREMENT, HAVE_PRE_MODIFY_DISP,
2280 HAVE_POST_MODIFY_DISP, HAVE_PRE_MODIFY_REG, HAVE_POST_MODIFY_REG,
2281 REGNO_MODE_OK_FOR_BASE_P, LEGITIMATE_CONSTANT_P,
2282 REG_MODE_OK_FOR_BASE_P, REG_OK_FOR_INDEX_P, GO_IF_LEGITIMATE_ADDRESS,
2283 LEGITIMIZE_ADDRESS, THUMB2_LEGITIMIZE_ADDRESS,
2284 GO_IF_MODE_DEPENDENT_ADDRESS, MEMORY_MOVE_COST, BRANCH_COST,
2285 ASM_APP_OFF, ASM_OUTPUT_CASE_LABEL, ARM_DECLARE_FUNCTION_NAME,
2286 FINAL_PRESCAN_INSN, PRINT_OPERAND_PUNCT_VALID_P,
2287 PRINT_OPERAND_ADDRESS): Adjust for Thumb-2.
2288 (arm_arch_notm, arm_arch_thumb2, arm_arch_hwdiv): New declarations.
2289 * config/arm/arm-cores.def: Add arm1156t2-s, cortex-a8, cortex-r4 and
2291 * config/arm/arm-tune.md: Regenerate.
2292 * config/arm/arm-protos.h: Update prototypes.
2293 * config/arm/vfp.md: Enable patterns for Thumb-2.
2294 (arm_movsi_vfp): Add movw alternative. Use output_move_vfp.
2295 (arm_movdi_vfp, movsf_vfp, movdf_vfp): Use output_move_vfp.
2296 (thumb2_movsi_vfp, thumb2_movdi_vfp, thumb2_movsf_vfp,
2297 thumb2_movdf_vfp, thumb2_movsfcc_vfp, thumb2_movdfcc_vfp): New.
2298 * config/arm/libunwind.S: Add Thumb-2 code.
2299 * config/arm/constraints.md: Update include Thumb-2.
2300 * config/arm/ieee754-sf.S: Add Thumb-2/Unified asm support.
2301 * config/arm/ieee754-df.S: Ditto.
2302 * config/arm/bpabi.S: Ditto.
2303 * config/arm/t-arm (MD_INCLUDES): Add thumb2.md.
2304 * config/arm/predicates.md (low_register_operand,
2305 low_reg_or_int_operand, thumb_16bit_operator): New.
2306 (thumb_cmp_operand, thumb_cmpneg_operand): Rename...
2307 (thumb1_cmp_operand, thumb1_cmpneg_operand): ... to this.
2308 * config/arm/t-arm-elf: Add armv7 multilib.
2309 * config/arm/arm.md: Update patterns for Thumb-2 and Unified asm.
2311 (UNSPEC_STACK_ALIGN, ce_count): New.
2312 (arm_incscc, arm_decscc, arm_umaxsi3, arm_uminsi3,
2313 arm_zero_extendsidi2, arm_zero_extendqidi2): New
2315 * config/arm/fpa.md: Update patterns for Thumb-2 and Unified asm.
2316 (thumb2_movsf_fpa, thumb2_movdf_fpa, thumb2_movxf_fpa,
2317 thumb2_movsfcc_fpa, thumb2_movdfcc_fpa): New insns.
2318 * config/arm/cirrus.md: Update patterns for Thumb-2 and Unified asm.
2319 (cirrus_thumb2_movdi, cirrus_thumb2_movsi_insn,
2320 thumb2_cirrus_movsf_hard_insn, thumb2_cirrus_movdf_hard_insn): New
2322 * doc/extend.texi: Document ARMv7-M interrupt functions.
2323 * doc/invoke.texi: Document Thumb-2 new cores+architectures.
2325 2007-01-03 Jakub Jelinek <jakub@redhat.com>
2327 * unwind-dw2.c (SIGNAL_FRAME_BIT, EXTENDED_CONTEXT_BIT): Define.
2328 (struct _Unwind_Context): Rename args_size to flags, remove
2329 signal_frame field, add a new args_size field and version field.
2330 (_Unwind_IsSignalFrame, _Unwind_SetSignalFrame,
2331 _Unwind_IsExtendedContext): New inline functions.
2332 (_Unwind_GetGR, _Unwind_SetGR, _Unwind_GetGRPtr, _Unwind_SetGRPtr):
2333 Assume by_value array is only present if _Unwind_IsExtendedContext.
2334 (_Unwind_GetIPInfo, execute_cfa_program, uw_frame_state_for): Use
2335 _Unwind_IsSignalFrame.
2336 (__frame_state_for): Initialize context.flags to EXTENDED_CONTEXT_BIT.
2337 (uw_update_context_1): Use _Unwind_SetSignalFrame.
2338 (uw_init_context_1): Initialize context->flags to
2339 EXTENDED_CONTEXT_BIT.
2340 * config/rs6000/linux-unwind.h (frob_update_context): Use
2341 _Unwind_SetSignalFrame.
2343 2007-01-03 Andrew Pinski <andrew_pinski@playstation.sony.com>
2346 * gimplify.c (gimplify_modify_expr_complex_part): Move below
2347 tree_to_gimple_tuple. Call tree_to_gimple_tuple when we need
2350 2007-01-03 Kazu Hirata <kazu@codesourcery.com>
2352 * config/i386/i386.h (NON_STACK_REG_P, REGNO_OK_FOR_SIREG_P,
2353 REGNO_OK_FOR_DIREG_P, REWRITE_ADDRESS, ASM_OPERAND_LETTER,
2354 RET, AT_SP): Remove.
2355 * config/i386/i386.md (*sse_prologue_save_insn): Use return
2358 * alias.c (init_alias_analysis): Use VEC_safe_grow_cleared.
2359 * cfgbuild.c (find_basic_blocks): Likewise.
2360 * cfgrtl.c (rtl_create_basic_block): Likewise.
2361 * function.c (temp_slots_at_level): Likewise.
2362 * reg-stack.c (stack_regs_mentioned): Likewise.
2363 * regclass.c (allocate_reg_info): Likewise.
2364 * tree-cfg.c (init_empty_tree_cfg, build_tree_cfg, create_bb,
2365 set_bb_for_stmt, move_block_to_fn): Likewise.
2366 * tree-complex.c (tree_lower_complex): Likewise.
2367 * vec.h (VEC_safe_grow_cleared): New.
2369 * cgraphunit.c, tree-ssa-alias.c: Fix comment typos.
2371 2007-01-03 Zdenek Dvorak <dvorakz@suse.cz>
2373 * loop-unswitch.c (unswitch_loop): Pass probabilities to loopify.
2374 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Pass probabilities
2376 * cfgloopmanip.c (scale_loop_frequencies): Export.
2377 (loopify): Scale the frequencies by prescribed coefficients.
2378 (set_zero_probability): New function.
2379 (duplicate_loop_to_header_edge): Improve updating of frequencies.
2380 (lv_adjust_loop_entry_edge, loop_version): Set probabilities
2381 and frequencies according to arguments.
2382 * tree-ssa-loop-manip.c (tree_unroll_loop): Set probabilities
2384 * cfg.c (scale_bbs_frequencies_int): Allow scaling the frequencies up.
2385 * modulo-sched.c (sms_schedule): Set probabilities for entering
2386 versioned loop correctly.
2387 * tree-vect-transform.c (vect_transform_loop): Ditto.
2388 * cfgloop.h (loopify, loop_version): Declaration changed.
2389 (scale_loop_frequencies): Declared.
2391 2007-01-02 Jan Hubicka <jh@suse.cz>
2393 * cgraph.c: Include tree-flow.h
2394 (cgraph_add_new-function): Handle IPA_SSA mode; execute
2396 * cgraph.h (enum cgraph_state): Add CGRAPH_STATE_IPA_SSA.
2397 * tree-pass.h (pass_all_early_optimizations): Declare.
2398 * cgraphunit.c (cgraph_process_new_functions): Add IPA_SSA; execute
2400 (cgraph_analyze_function): Do early_local_passes.
2401 * tree-mudflap.c (mf_decl_cache_locals, mf_build_check_statement_for):
2402 Do not add referenced vars.
2403 * tree-optimize.c (gate_all_optimizations): Do not execute when not in
2405 (gate_all_early_local_passes): New gate.
2406 (pass_early_local_passes): Use new gate.
2407 (execute_early_local_optimizations): New functions.
2408 (gate_all_early_optimizations): New gate.
2409 (pass_all_early_optimizations): New pass.
2410 (execute_free_datastructures): Free SSA only when initialized.
2411 (gate_init_datastructures): Init only when optimizing.
2412 (tree_lowering_passes): Do early local passes when called late.
2413 * tree-profile.c (do_tree_profiling): Don't profile functions added
2415 (do_early_tree_profiling, pass_early_tree_profile): Kill.
2416 * tree-cfg.c (update_modified_stmts): Do not update when operands are
2418 * passes.c (init_optimizations_passes): Reorder so we go into SSA
2419 during early_local_passes.
2420 * Makefile.in (cgraph.o): Add dependency on tree-flow.h.
2423 2007-01-02 Carlos O'Donell <carlos@codesourcery.com>
2425 * Makefile.in: Update copyright year.
2427 2007-01-02 Carlos O'Donell <carlos@codesourcery.com>
2429 * Makefile.in: Export GCC_EXEC_PREFIX before calling $(RUNTEST)
2430 in $(lang_checks) and check-consistency targets.
2432 2007-01-02 Jan Hubicka <jh@suse.cz>
2434 * tree-mudflap.c (mf_decl_cache_locals, mf_build_check_statement_for):
2435 Do not add referenced vars.
2436 * tree-cfg.c (update_modified_stmts): Do not update when SSA operands
2438 * passes.c (init_optimization_passes): Put mudflap_2 after
2439 free_datastructures.
2441 2007-01-02 Jan Hubicka <jh@suse.cz>
2443 * tree-optimize (execute_fixup_cfg): Set after_inlining flag.
2444 Set NOTHROW flag on call statements proved to be nothrow.
2445 Update statement of local calls so new pure/const functions are
2446 updated. Update_ssa when in ssa form. Mark PHI nodes of nonlocal
2448 (tree_rest_of_compilation): Register hooks and initialize bitmap
2449 early. Do not set after_inlining flag.
2451 2007-01-02 Steve Ellcey <sje@cup.hp.com>
2453 * sbitmap.c (HOST_BITS_PER_LONG_LONG): Change to
2454 HOST_BITS_PER_LONGLONG
2456 2007-01-02 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
2459 * c-typeck.c (store_init_value): Don't emit pedantic overflow
2460 warning for non-static initializers.
2462 2007-01-02 Steven Bosscher <steven@gcc.gnu.org>
2464 * config/alpha/alpha.md, arm/arm.c, darwin.c, frv/frv.md,
2465 m32r/m32r.c, m32r/m32r.c, mn10300/mn10300.md, pa/pa.c,
2466 rs6000/rs6000.c, s390/s390.md, sh/sh.md, sparc/sparc.c:
2467 Always use set_unique_reg_note to add REG_EQUAL notes.
2469 2007-01-02 Kazu Hirata <kazu@codesourcery.com>
2472 2007-01-02 Kazu Hirata <kazu@codesourcery.com>
2474 * alias.c (init_alias_analysis): Use VEC_safe_grow_cleared.
2475 * cfgbuild.c (find_basic_blocks): Likewise.
2476 * cfgrtl.c (rtl_create_basic_block): Likewise.
2477 * function.c (temp_slots_at_level): Likewise.
2478 * reg-stack.c (stack_regs_mentioned): Likewise.
2479 * regclass.c (allocate_reg_info): Likewise.
2480 * tree-cfg.c (init_empty_tree_cfg, build_tree_cfg, create_bb,
2481 set_bb_for_stmt, move_block_to_fn): Likewise.
2482 * tree-complex.c (tree_lower_complex): Likewise.
2483 * vec.h (VEC_safe_grow_cleared): New.
2485 2007-01-02 Ian Lance Taylor <iant@google.com>
2487 * c-common.c (c_common_truthvalue_conversion): When warning about
2488 using an assignment as a truth value, set TREE_NO_WARNING.
2490 2007-01-02 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
2493 * c.opt (Wold-style-declaration): New.
2494 * doc/invoke.texi (C-only Warning Options): New.
2495 (Wold-style-declaration): Document it.
2496 (Wextra): Enabled by -Wextra.
2497 * c-opts.c (c_common_post_options): Enabled by -Wextra.
2498 * c-decl.c (declspecs_add_scspec): Replace -Wextra with
2499 -Wold-style-declaration.
2501 2007-01-02 Kazu Hirata <kazu@codesourcery.com>
2503 * alias.c (init_alias_analysis): Use VEC_safe_grow_cleared.
2504 * cfgbuild.c (find_basic_blocks): Likewise.
2505 * cfgrtl.c (rtl_create_basic_block): Likewise.
2506 * function.c (temp_slots_at_level): Likewise.
2507 * reg-stack.c (stack_regs_mentioned): Likewise.
2508 * regclass.c (allocate_reg_info): Likewise.
2509 * tree-cfg.c (init_empty_tree_cfg, build_tree_cfg, create_bb,
2510 set_bb_for_stmt, move_block_to_fn): Likewise.
2511 * tree-complex.c (tree_lower_complex): Likewise.
2512 * vec.h (VEC_safe_grow_cleared): New.
2514 2007-01-02 Douglas Gregor <doug.gregor@gmail.com>
2516 * c-common.c(c_common_nodes_and_builtins): Since variants of
2517 void_type_node get built before it is given a name, we need to
2518 give those variants the name, too.
2519 (complete_array_type): We need to work with the canonical main
2520 type of the array, from which we will build the qualified version.
2521 * params.def (PARAM_VERIFY_CANONICAL_TYPES): New.
2522 * print-tree.c (print_node): Display canonical type information
2524 * stor-layout.c (layout_type): When we don't know the
2525 alignment of a type for which we're building an array, we end up
2526 guessing wrong, so make the type require structural equality.
2527 * tree.c (make_node_stat): When we build a new type, it is its
2529 (build_type_attribute_qual_variant): When building an attribute
2530 variant, its canonical type is the non-attribute variant. However,
2531 if the attributes are target-dependent and they differ, we need to
2532 use structural equality checks for this type.
2533 (build_qualified_type): A qualified type is not equivalent to its
2534 unqualified variant; set the canonical type appropriately.
2535 (build_distinct_type_copy): When building a distinct type from
2536 another type, the new type is its own canonical type.
2537 (build_variant_type_copy): When building a new type variant, we
2538 assume that it is equivalent to the original type.
2539 (build_pointer_type_for_mode): When building a pointer type, also
2540 build a canonical type pointer.
2541 (build_reference_type_for_mode): When building a reference type,
2542 also build a canonical type reference.
2543 (build_index_type): When we can't hash an index type (e.g.,
2544 because its maximum value is negative), the index type requires
2545 structural equality tests.
2546 (build_array_type): Build the canonical form of an array type.
2547 (build_function_type): Function types require structural equality,
2548 because they contain default arguments, attributes, etc.
2549 (build_method_type_directly): Ditto for method types.
2550 (build_offset_type): Build the canonical offset type.
2551 (build_complex_type): Build the canonical vector type.
2552 (make_vector_type): Build the canonical vector type.
2553 * tree.h (TYPE_CANONICAL): New.
2554 (TYPE_STRUCTURAL_EQUALITY_P): New.
2555 (SET_TYPE_STRUCTURAL_EQUALITY): New.
2556 (struct tree_type): Added "canonical" field.
2557 * params.h (VERIFY_CANONICAL_TYPES): New.
2558 * doc/c-tree.texi (TYPE_CANONICAL): Document.
2559 (TYPE_STRUCTURAL_EQUALITY_P): Document.
2560 (SET_TYPE_STRUCTURAL_EQUALITY): Document.
2561 * doc/invoke.texi (verify-canonical-types): Document --param
2562 parameter for verifying canonical types.
2564 2007-01-02 Joseph Myers <joseph@codesourcery.com>
2566 * config.gcc (powerpc-*-eabispe*, powerpc-*-eabisimaltivec*,
2567 powerpc-*-eabisim*, powerpc-*-eabialtivec*, powerpc-*-eabi*,
2568 powerpc-*-rtems*, powerpc-wrs-vxworks, powerpc-wrs-vxworksae,
2569 powerpcle-*-eabisim*, powerpcle-*-eabi*): Add rs6000/e500.h to
2571 * config/rs6000/e500.h: New.
2572 * config/rs6000/eabi.h (TARGET_SPE_ABI, TARGET_SPE, TARGET_E500,
2573 TARGET_ISEL, TARGET_FPRS, TARGET_E500_SINGLE, TARGET_E500_DOUBLE):
2575 * config/rs6000/linuxspe.h (TARGET_SPE_ABI, TARGET_SPE,
2576 TARGET_E500, TARGET_ISEL, TARGET_FPRS, TARGET_E500_SINGLE,
2577 TARGET_E500_DOUBLE): Remove.
2578 * config/rs6000/vxworks.h (TARGET_SPE_ABI, TARGET_SPE,
2579 TARGET_E500, TARGET_ISEL, TARGET_FPRS): Remove.
2580 * config/rs6000/rs6000.h (CHECK_E500_OPTIONS): Define.
2581 * config/rs6000/rs6000.c (rs6000_override_options): Use
2584 2007-01-02 Joseph Myers <joseph@codesourcery.com>
2586 * config/rs6000/rs6000.c (print_operand): Check (TARGET_SPE ||
2587 TARGET_E500_DOUBLE), not TARGET_E500, for %y.
2588 (rs6000_generate_compare, rs6000_emit_sCOND, output_cbranch,
2589 rs6000_emit_cmove): Don't check TARGET_E500.
2590 * config/rs6000/rs6000.md (bunordered, bordered, sunordered,
2591 sordered): Don't check TARGET_E500.
2593 2007-01-01 Eric Christopher <echristo@apple.com>
2595 * config/mips/mips.c (mips_regno_mode_ok_for_base_p): Use
2596 HARD_REGISTER_NUM_P.
2598 2007-01-01 Roger Sayle <roger@eyesopen.com>
2600 * fold-const.c (fold_binary) <EQ_EXPR>: Fold "(X^C1) eq/ne C2" into
2601 "X eq/ne (C1^C2)". Fold "(X^Z) eq/ne (Y^Z)" as "X eq/ne Y" when Z
2602 has no side-effects. Fold "(X^C1) eq/ne (Y^C2)" as "(X^(C1^C2))
2605 2007-01-01 Mike Stump <mrs@apple.com>
2607 * configure.ac: Remove support for building with Apple's gcc-3.1.
2609 2007-01-02 Joseph Myers <joseph@codesourcery.com>
2612 * caller-save.c (add_stored_regs): Only handle SUBREGs if inner
2613 REG is a hard register. Do not modify REG before calling
2615 * rtlanal.c (subreg_get_info): Don't assert size of XMODE is a
2616 multiple of the size of YMODE for certain lowpart cases.
2618 2007-01-01 Andrew Pinski <pinskia@gmail.com>
2621 * gimplify (voidify_wrapper_expr): Update for
2624 2007-01-01 Andreas Schwab <schwab@suse.de>
2627 * config/ia64/ia64.c (ia64_compute_frame_size): Account space for
2628 save of BR0 in extra_spill_size instead of spill_size.
2629 (ia64_expand_prologue): Save BR0 outside of the gr/br/fr spill
2631 (ia64_expand_epilogue): Restore BR0 from its new location.
2633 2007-01-01 Andrew Pinski <pinskia@gmail.com>
2635 * gimplify.c (gimplify_init_constructor <case VECTOR_TYPE>):
2636 Use a temporary variable if the left hand side is not a gimple
2639 2007-01-01 Andrew Pinski <pinskia@gmail.com>
2641 * gimplify.c (gimplify_return_expr): Make the temporary variable
2642 for the return expression, a gimple register variable.
2644 2007-01-01 Jan Hubicka <jh@suse.cz>
2646 * emit-rtl.c (emit_copy_of_insn_after): Do not call copy_insn_1 for
2649 2007-01-01 Mike Stump <mrs@apple.com>
2651 * configure.ac (HAVE_GAS_LITERAL16): Add autoconf check for
2653 * config/darwin.c (machopic_select_rtx_section): Use
2655 (darwin_mergeable_constant_section): Likewise.
2656 * configure: Regenerate.
2657 * config.in: Regenerate.
2659 2007-01-01 Jan Hubicka <jh@suse.cz>
2660 Andrew Pinski <pinskia@gmail.com>
2662 * cgraphunit.c (cgraph_optimize): Call cgraph_add_new_functions
2663 before starting IPA passes.