1 2007-01-19 Ian Lance Taylor <iant@google.com>
3 * expmed.c (expand_divmod) [TRUNC_MOD_EXPR, TRUNC_DIV_EXPR]: Cast
4 constant to unsigned HOST_WIDE_INT before negating.
6 2007-01-19 Ian Lance Taylor <iant@google.com>
8 * tree-ssa-operands.h (struct vuse_vec_d): Change num_vuse field
10 (VUSE_VECT_ELEMENT) [ENABLE_CHECKING]: Use unsigned comparison.
11 (VUSE_ELEMENT_PTR) [ENABLE_CHECKING]: Likewise.
12 (SET_VUSE_VECT_ELEMENT) [ENABLE_CHECKING]: Likewise.
13 (SET_VUSE_ELEMENT_VAR) [ENABLE_CHECKING]: Likewise.
14 (SET_VUSE_ELEMENT_PTR) [ENABLE_CHECKING]: Likewise.
15 (realloc_vdef, realloc_vuse): Change second parameter to
17 (ssa_operand_iterator_d): Change vuse_index and mayuse_index
19 * tree-ssa-operands.c (realloc_vop): Change num_elem parameter to
20 unsigned. Change x and lim locals to unsigned.
21 (realloc_vdef, realloc_vuse): Change num_elem parameter to
23 (finalize_ssa_vuse_ops): Change old_i local to unsigned.
24 (copy_virtual_operands): Change i and n locals to unsigned.
26 2007-01-19 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
29 * builtins.c (fold_builtin_1): Handle builtin fdim.
31 2007-01-20 Jan Hubicka <jh@suse.cz>
33 * tree-ssa.c (init_tree_ssa): Do not call init_alias_heapvars.
34 * tree-ssa-structalias.c (compute_points_to_sets): Do call
36 (init_alias_heapvars): Initialize only when not already initialized.
37 (delete_alias_heapvars): Set heapvar_for_stmt to NULL.
39 2007-01-19 Roger Sayle <roger@eyesopen.com>
41 * common.opt (fsigned-zeros): New command line option.
42 * flags.h (HONOR_SIGNED_ZEROS): Control via flag_signed_zeros instead
43 of flag_unsafe_math_optimizations.
44 * opts.c (set_fast_math_flags): The -ffast-math command line option
45 implies -fno-signed-zeros.
46 (fast_math_flags_set_p): Likewise.
48 * doc/invoke.texi: Document new -fno-signed-zeros option, and update
49 the documentation of -ffast-math appropriately. Wrap long lines.
51 2007-01-19 Steve Ellcey <sje@cup.hp.com>
53 * system.h (ASM_MAKE_LABEL_LINKONCE): Poison.
54 * varasm.c (globalize_decl): Remove ASM_MAKE_LABEL_LINKONCE ifdef.
56 2007-01-19 Tomas Bily <tbily@suse.cz>
58 * cgraphunit.c (cgraph_finalize_function): Updating of pid
60 (tree_init_ic_make_global_vars): New function
61 (tree_init_edge_profiler): call of tree_init_ic_make_global_vars
62 (tree_gen_ic_profiler): New function
63 (tree_gen_ic_func_profiler): New function
64 (tree_profiling): Added calling of tree_gen_ic_func_profiler
65 (tree_profile_hooks): Added hook for indirec/virtual calls
66 * value-prof.c (tree_find_values_to_profile): New case for
68 (tree_values_to_profile): Call for determining indirect/virtual
70 (tree_indirect_call_to_profile): New function
71 (tree_ic_transform): New function
72 (tree_ic): New function
73 (find_func_by_pid): New function
74 (init_pid_map): New function
75 (tree_value_profile_transformations): Added check for
76 indirect/virtual call transformation
77 * value-prof.h (enum hist_type): New counter type for
78 indirect/virtual calls
79 (profile_hooks): Added new hook for profiling indirect/virtual
81 * profile.c (instrument_values): New case for indirect/virtual
83 * gcov-io.h (GCOV_LAST_VALUE_COUNTER): Changed to 6
84 (GCOV_COUNTER_V_INDIR): New counter type
85 (GCOV_COUNTER_NAMES): New name of counter "indirect" added
86 (GCOV_MERGE_FUNCTIONS): New merge function for indirect/virtual
88 * cgraph.c: Definition of cgraph_max_pid
89 (cgraph_create_node): Default init of pid attribute
90 * cgraph.h: Declaration of cgraph_max_pid
91 (struct cgraph_node): Added pid attribute
92 * libgcov.c (__gcov_indirect_call_profiler): New function
93 (__gcov_one_value_profiler_body): New function
94 (__gcov_one_value_profiler): Body was moved to
95 __gcov_one_value_profiler_body and calls it
97 2007-01-19 Basile Starynkevitch <basile@starynkevitch.net>
99 * doc/gty.texi (Options): Document the mark_hook option to
101 * gcc/gengtype.c (write_types_data, write_func_for_structure,
102 write_types, ggc_wtd, pch_wtd): Added skip_hooks to
103 write_types_data, ggc_wtd, pch_wtd for mark_hook.
104 (walk_type, write_func_for_structure): Generate the mark_hook if
107 2007-01-19 Jan Hubicka <jh@suse.cz>
109 * ipa-inline.c (cgraph_decide_inlining_incrementally): Instead of 'early' argument
110 take inlining mode argument specifying whether to inline for size/speeed or all
111 functions; add support for flattening; improve dumpting.
112 (cgraph_early_inlining): Update call of decide_inlining_incrementally.
114 2007-01-19 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
117 * toplev.c (warn_deprecated_use): Use %qD instead of %qs to print
118 the name of the declared identifier.
120 2007-01-19 Dirk Mueller <dmueller@suse.de>
122 * config/i386.h (CONDITIONAL_REGISTER_USAGE): Store
123 result of PIC_OFFSET_TABLE_REGNUM in temporary variable to avoid
124 duplicate evaluation.
126 2007-01-19 Uros Bizjak <ubizjak@gmail.com>
128 * config/i386/i386.md (acos<mode>2): Rename from acossf2 and acosdf2.
129 Macroize expander using X87MODEF12 mode macro. Extend operand 1
130 to XFMode, use acosxf2 and truncate result to requested mode.
131 Use SSE_FLOAT_MODE_P to disable patterns for SSE math.
132 (asin<mode>2): Similarly, with asin expanders.
133 (*fscalexf4_i387): Rename from *fscalexf4.
134 (expNcorexf3): New expander.
135 (expxf2, exp10xf2, exp2xf2): Use expNcorexf3 expander.
136 (exp<mode>2): Rename from expsf2 and expdf2. Macroize expander using
137 X87MODEF12 mode macro. Extend operand 1 to XFMode, use expxf2 and
138 truncate result to requested mode. Use SSE_FLOAT_MODE_P to disable
139 patterns for SSE math.
140 (exp10<mode>2): Similarly, with exp10 expanders.
141 (exp2<mode>2): Similarly, with exp2 expanders.
142 (expm1<mode>2): Similarly, with expm1 expanders.
143 (ldexp<mode>3): Similarly, with ldexp expanders.
144 (log<mode>2, log10<mode>2, log2<mode>2, log1p<mode>2, logb<mode>2):
145 Use gen_truncxf<mode>2_i387_noop to truncate result.
147 2007-01-19 Richard Sandiford <richard@codesourcery.com>
149 * config/m68k/m68k.h (PREFERRED_STACK_BOUNDARY): Define to 32
150 for ColdFire targets.
152 2007-01-19 Nathan Sidwell <nathan@codesourcery.com>
153 Richard Sandiford <richard@codesourcery.com>
155 * config/m68k/m68k.h (M68K_STATIC_CHAIN_REG_NAME): New macro.
156 (INITIALIZE_TRAMPOLINE): Use STATIC_CHAIN_REGNUM.
157 (__transfer_from_trampoline): Use M68K_STATIC_CHAIN_REG_NAME.
158 * config/m68k/m68kelf.h (STATIC_CHAIN_REGNUM): Override.
159 (M68K_STATIC_CHAIN_REG_NAME): Likewise.
160 * config/m68k/netbsd-elf.h (M68K_STATIC_CHAIN_REG_NAME): Likewise.
162 2007-01-19 Richard Sandiford <richard@codesourcery.com>
164 * config/m68k/m68k.md (adddi_dilshr32): Rename to...
165 (*adddi_dilshr32): ...this. Fix formatting. Remove commented-out
166 non-canonical pattern. Restrict to !TARGET_COLDFIRE.
167 (*adddi_dilshr32_cf): New pattern.
168 (adddi3, subdi3): Remove first alternatives.
170 2007-01-19 Richard Sandiford <richard@codesourcery.com>
172 * config/m68k/m68k.c (notice_update_cc): If an SFmode move is
173 implemented using move.l, do not use its cc result for floating-point
176 2007-01-19 Richard Sandiford <richard@codesourcery.com>
178 * config/m68k/m68k.h (EXTRA_CONSTRAINT): Stop the 'T' constraint
179 from accepting 's' constraints if flag_pic.
181 2007-01-19 Richard Sandiford <richard@codesourcery.com>
183 * config/m68k/m68k.md (bordered, bunordered, buneq, bunge, bungt)
184 (bunle, bunlt, bltgt, bordered_rev, bunordered_rev, buneq_rev)
185 (bunge_rev, bungt_rev, bunle_rev, bunlt_rev, bltgt_rev): Change
186 condition from TARGET_68881 to TARGET_HARD_FLOAT.
188 2007-01-19 Sandra Loosemore <sandra@codesourcery.com>
190 * longlong.h (count_leading_zeros, COUNT_LEADING_ZEROS_0): Add
191 ColdFire alternatives.
192 * config/m68k/m68k.h (CLZ_DEFINED_VALUE_AT_ZERO): New macro.
193 * config/m68k/m68k.md (clzsi2): Define for ColdFire
194 architectures that support ff1 instruction.
196 2007-01-19 Richard Sandiford <richard@codesourcery.com>
197 Julian Brown <julian@codesourcery.com>
199 * config/m68k/m68k.h (CONST_OK_FOR_LETTER_P): Add an 'R' case.
200 * config/m68k/m68k.md (*movsi_cfv4): Fold into...
201 (*movsi_cf): ...here. Remove unnecessary 'R' from 'Rg'.
204 2007-01-19 Richard Sandiford <richard@codesourcery.com>
206 * config/m68k/m68k-protos.h (valid_mov3q_const): Take a HOST_WIDE_INT
208 (output_move_const_into_data_reg, output_move_simode_const): Delete.
209 * config/m68k/m68k.c (const_method, const_int_cost): Take a
210 HOST_WIDE_INT instead of an rtx.
211 (m68k_rtx_costs): Update call accordingly.
212 (output_move_const_into_data_reg): Likewise. Fix formatting.
213 (valid_mov3q_const): Take a HOST_WIDE_INT instead of an rtx.
215 (output_move_simode_const): Update calls after above changes.
216 Rework to use automatic variables and predicates like MEM_P.
217 * config/m68k/m68k.md (pushexthisi_const): Update call to
220 2007-01-19 Dirk Mueller <dmueller@suse.de>
222 * tree-ssa-alias.c (perform_var_substitution): Fix typo
225 2007-01-19 Richard Guenther <rguenther@suse.de>
227 * builtins.c (expand_builtin_cexpi): Fall back to expanding
228 via cexp in case sincos is not available.
230 2007-01-19 Richard Guenther <rguenther@suse.de>
232 * doc/tm.texi (TARGET_HAS_SINCOS): Document new target macro.
233 * defaults.h (TARGET_HAS_SINCOS): Default to off.
234 * config/linux.h (TARGET_HAS_SINCOS): Set to on if we have glibc.
235 * config/alpha/linux.h (TARGET_HAS_SINCOS): Likewise.
236 * config/sparc/linux.h (TARGET_HAS_SINCOS): Likewise.
237 * config/sparc/linux64.h (TARGET_HAS_SINCOS): Likewise.
238 * config/rs6000/linux.h (TARGET_HAS_SINCOS): Likewise.
239 * config/rs6000/linux64.h (TARGET_HAS_SINCOS): Likewise.
241 2007-01-19 Uros Bizjak <ubizjak@gmail.com>
243 * config/i386/i386.md (*fpatanxf3_i387, fpatan_extend<mode>xf3_i387):
245 (atan2sf3_1, atan2df3_1, atan2xf3_1): Remove insn patterns.
246 (atan2xf3): Directly generate RTL pattern.
247 (atan2<mode>3): Rename from atan2sf3 and atan2df3 and macroize insn
248 patterns using X87MODEF12 mode macro. Use fpatan_extend<mode>xf3_i387
249 and truncate result to requested mode. Use SSE_FLOAT_MODE_P to
250 disable patterns for SSE math.
251 (atan<mode>2): Rename from atansf2 and atandf2 and macroize insn
252 patterns using X87MODEF12 mode macro. Use fpatan_extend<mode>xf3_i387
253 and truncate result to requested mode. Use SSE_FLOAT_MODE_P to
254 disable patterns for SSE math.
256 2007-01-19 Alexandre Oliva <aoliva@redhat.com>
258 * libgcc-std.ver: Fix typo in %inherit for GCC_4.3.0.
260 2007-01-18 Roger Sayle <roger@eyesopen.com>
262 * fold-const.c (fold_unary) <VIEW_CONVERT_EXPR>: Optimize away a
263 VIEW_CONVERT_EXPR to the same type as it's operand.
265 2007-01-18 David Edelsohn <edelsohn@gnu.org>
267 * config/rs6000/darwin-ldouble.c: Only build _SOFT_FLOAT if
268 configured for long double 128.
270 2007-01-18 Mike Stump <mrs@apple.com>
272 * config/rs6000/rs6000.c (rs6000_emit_vector_compare): Fix build
275 2007-01-18 Michael Meissner <michael.meissner@amd.com>
277 * i386.c (ix86_compute_frame_layout): Make fprintf's in #if 0 code
280 2007-01-18 Jan Hubicka <jh@suse.cz>
282 * tree-ssa-operands.c (vop_free_bucket_size): Never return value
283 greater than NUM_VOP_FREE_BUCKETS.
285 2007-01-18 Daniel Berlin <dberlin@dberlin.org>
287 * tree-ssa-structalias.c: Update comments.
288 (ptabitmap_obstack): Removed.
290 (oldpta_obstack): Ditto.
291 (stats): Add a few members.
292 (struct variable_info): Remove node, complex, address_taken, and
293 indirect_target members. Add oldsolution member.
294 (new_var_info): Do not initialize removed members.
295 (constraint_expr_type): Remove INCLUDES.
296 (constraint_graph): Add size, implicit_preds, rep,
297 indirect_cycles, eq_rep, label, direct_nodes, and complex members.
298 (FIRST_REF_NODE): New macro.
299 (LAST_REF_NODE): Ditto.
300 (FIRST_ADDR_NODE): Ditto.
301 (find): New function.
303 (dump_constraint): Do not handle INCLUDES.
304 (insert_into_complex): Do not insert duplicate constraints.
305 (condense_varmap_nodes): Renamed and rewritten into ...
306 (merge_node_constraints): This. Also fix bug in handling of
307 offseted copy constraints.
308 (clear_edges_for_node): No longer need to deal with preds at all,
309 or removing associated preds/succs.
310 (merge_graph_nodes): Deal with indirect_cycles.
311 Don't deal with predecessors.
312 (add_implicit_graph_edge): New function.
313 (add_pred_graph_edge): Ditto.
314 (add_graph_edge): Don't deal with predecessors.
315 (build_constraint_graph): Removed.
316 (build_pred_graph): New function.
317 (build_succ_graph): Ditto.
318 (struct scc_info): Removed in_component. Added roots, dfs, and
319 node_mapping. Remove visited_index, unification_queue.
320 (scc_visit): Deal with union-find we do now.
321 Deal with cycles with REF nodes.
322 (collapse_nodes): Renamed and rewritten to ...
324 (process_unification_queue): Removed.
325 (topo_visit): Cleanup
326 (do_da_constraint): Use find.
327 (do_sd_constraint): Ditto.
328 (do_ds_constraint): Ditto.
329 (do_complex_constraint): Ditto.
330 (init_scc_info): Update for removed and added members.
331 (find_and_collapse_graph_cycles): Renamed and rewritten into ...
332 (find_indirect_cycles): This.
333 (equivalence_class): New variable.
334 (label_visit): New function.
335 (perform_variable_substitution): Rewritten.
336 (free_var_substitution_info): New function.
337 (find_equivalent_node): Ditto.
338 (move_complex_constraints): Ditto.
339 (eliminate_indirect_cycles): Ditto.
340 (solve_graph): Only propagate changed bits.
341 Use indirect cycle elimination.
343 (tree_id_t): Rename to tree_vi_t, delete id member, add vi member.
344 (tree_id_eq): Renamed to ...
345 (tree_vi_eq): This. Update for member change
346 (insert_id_for_tree): Renamed and rewritten to ...
347 (insert_vi_for_tree): This.
348 (lookup_id_for_tree): Renamed and rewritten to ...
349 (lookup_vi_for_tree): This.
350 (get_id_for_tree): Renamed and rewritten to ...
351 (get_vi_for_tree): Ditto.
352 (get_constraint_exp_from_ssa_var): Update to use get_vi_for_tree.
353 (process_constraint): Don't handle INCLUDES.
354 Remove special ADDRESSOF case.
355 (find_func_aliases): Rewrite to use vi functions instead of id
357 (create_function_info_for): Ditto.
358 (create_variable_info_for): Ditto.
359 (intra_create_variable_infos): Ditto.
360 (merge_smts_into): Ditto.
361 (find_what_p_points_to): Ditto.
362 (init_base_vars): Ditto.
363 (init_alias_vars): Ditto.
364 (remove_preds_and_fake_succs): New function.
365 (dump_sa_points_to_info): Dump new stats.
366 (dump_solution_for_var): Use find.
367 (set_used_smts): Fix formatting.
368 (compute_points_to_sets): Updated for new functions.
369 (ipa_pta_execute): Ditto.
371 2007-01-18 Kazu Hirata <kazu@codesourcery.com>
372 Richard Sandiford <richard@codesourcery.com>
374 * doc/tm.texi (TARGET_FUNCTION_VALUE): Expand documentation of
376 * calls.c (expand_call): If the return value is a PARALLEL,
377 extract its first member.
378 * config/m68k/linux.h (FUNCTION_EXTRA_EPILOGUE): Remove.
379 * config/m68k/m68k.c (m68k_output_function_epilogue): Don't
380 use FUNCTION_EXTRA_EPILOGUE.
381 (m68k_function_value): Return a PARALLEL if the return value
382 is of a pointer type.
383 * config/m68k/netbsd-elf.h (current_function_returns_pointer)
384 (FUNCTION_EXTRA_EPILOGUE): Remove.
385 * config/m68k/m68k.md (D0_REG): New constant.
387 2007-01-18 Kazu Hirata <kazu@codesourcery.com>
389 * config/m68k/m68k.c (m68k_output_function_epilogue): Don't
390 output a NOP for empty epilogues.
392 2007-01-18 Richard Sandiford <richard@codesourcery.com>
394 * config/m68k/m68k.c (m68k_use_return_insn): Update comments
395 before function. Extend register save check to include all
396 registers, not just integer ones.
398 2007-01-18 Kazu Hirata <kazu@codesourcery.com>
400 * config/m68k/m68k-protos.h (use_return_insn): Rename to...
401 (m68k_use_return_insn): ...this.
402 * config/m68k/m68k.h (USE_RETURN_INSN): Delete.
403 * config/m68k/m68k.c (use_return_insn): Rename to...
404 (m68k_use_return_insn): ...this.
405 * config/m68k/m68k.md (return): Use m68k_use_return_insn instead
408 2007-01-18 Nathan Sidwell <nathan@codesourcery.com>
410 * config/m68k/fpgnulib.c (__truncdfsf2): Implement round to
411 nearest even, fix denormal rounding overflow.
413 2007-01-18 Richard Sandiford <richard@codesourcery.com>
415 * config/m68k/m68k.md (movsf_cf_hard): Use fsmove instead of
416 f%$smove and f%$move.
417 (movdf_cf_hard): Use fdmove for cases 0 and 3 and fmove for case 1.
418 (extendsfdf2_cf): Use fdmove instead of f%&move.
419 (truncdfsf2_cf): Use fsmove instead of f%$smove.
420 (add<mode>3_cf, sub<mode>3_cf): Use <FP:prec> instead of <FP:round>.
422 2007-01-18 Richard Sandiford <richard@codesourcery.com>
424 * config/m68k/m68k.md (movdf_cf_hard): Use output_move_double for
427 2007-01-18 Richard Sandiford <richard@codesourcery.com>
429 * real.h (real_format): Add a canonical_nan_lsbs_set field.
430 (coldfire_single_format): Declare.
431 (coldfire_double_format): Likewise.
432 * real.c (encode_ieee_single): Use canonical_nan_lsbs_set instead
433 of qnan_msb_set to determine the lower bits of a canonical
435 (encode_ieee_double): Likewise.
436 (encode_ieee_quad): Likewise.
437 (ieee_single_format): Initialize canonical_nan_lsbs_set.
438 (mips_single_format): Likewise.
439 (ieee_double_format): Likewise.
440 (mips_double_format): Likewise.
441 (ieee_extended_motorola_format): Likewise.
442 (ieee_extended_intel_96_format): Likewise.
443 (ieee_extended_intel_128_format): Likewise.
444 (ieee_extended_intel_96_round_53_format): Likewise.
445 (ibm_extended_format): Likewise.
446 (mips_extended_format): Likewise.
447 (ieee_quad_format): Likewise.
448 (mips_quad_format): Likewise.
449 (vax_f_format): Likewise.
450 (vax_d_format): Likewise.
451 (vax_g_format): Likewise.
452 (i370_single_format): Likewise.
453 (i370_double_format): Likewise.
454 (decimal_single_format): Likewise.
455 (decimal_double_format): Likewise.
456 (decimal_quad_format): Likewise.
457 (c4x_single_format): Likewise.
458 (c4x_extended_format): Likewise.
459 (real_internal_format): Likewise.
460 (coldfire_single_format): New real_format.
461 (coldfire_double_format): Likewise.
462 * config/pdp11/pdp11.c (pdp11_f_format): Initialize
463 canonical_nan_lsbs_set.
464 (pdp11_d_format): Likewise.
465 * config/m68k/m68k.c (override_options): Override REAL_FORMAT_MODE
466 if TARGET_COLDFIRE_CPU.
468 2007-01-18 Richard Sandiford <richard@codesourcery.com>
470 * config/m68k/m68k-protos.h (m68k_output_pic_call): Delete.
471 (output_call, m68k_legitimize_call_address): Declare.
472 * config/m68k/m68k.h (EXTRA_CONSTRAINT): Remove unnecessary
473 parenthesees. Add support for a 'W' constraint.
474 (LEGITIMATE_PIC_OPERAND_P): Remove SYMBOL_REF_FLAG handling.
475 (PRINT_OPERAND_PUNCT_VALID_P): Remove comment about 'o'.
476 (m68k_symbolic_call, m68k_symbolic_jump): Declare.
477 * config/m68k/m68k.c (m68k_symbolic_call, m68k_symbolic_jump): New
479 (override_options): Initialize them. Do not set flag_no_function_cse
480 for TARGET_ID_SHARED_LIBRARY.
481 (m68k_output_pic_call): Delete.
482 (m68k_legitimize_call_address): New function.
483 (print_operand): Remove the %o prefix. Handle the %p prefix.
484 (output_call): New function.
485 (m68k_output_mi_thunk): Use m68k_symbolic_jump. Always load the
486 target address from the GOT if symbolic jumps are not allowed.
487 * config/m68k/m68k.md (call, general_operand): Do not set
488 SYMBOL_REF_FLAG. Use m68k_legitimize_call_address instead.
489 Merge separate flag_pic and !flag_pic define_insns into...
490 (*call, *call_value): ...these new patterns. Match the address
491 rather than the containing MEM and require it to be a call_operand.
492 Use output_call to generate the asm template.
493 * config/m68k/predicates.md (const_call_operand): New predicate.
494 (call_operand): Likewise.
496 2007-01-18 Nathan Sidwell <nathan@codesourcery.com>
498 * config/m68k/m68k.h (REGISTER_MOVE_COST): Simplify definition.
499 (STACK_GROWS_DOWNWARD): Define to 1.
500 (FUNCTION_VALUE, LIBCALL_VALUE, FUNCTION_VALUE_REGNO_P): Equivocate
501 comments, emphasizing that these values are only defaults.
502 * config/m68k/linux.h (LINK_SPEC): Fix formatting in #undef.
503 * config/m68k/m68k.c (const_method): Remove trailing whitespace.
505 2007-01-18 Richard Sandiford <richard@codesourcery.com>
507 * config/m68k/m68k.md (cmpsi): Remove outdated flag_pic handling.
509 2007-01-18 Kazu Hirata <kazu@codesourcery.com>
510 Richard Sandiford <richard@codesourcery.com>
512 * config/m68k/m68k.h (DATA_REGNO_P, ADDRESS_REGNO_P, INT_REGNO_P)
513 (FP_REGNO_P): New macros.
514 (REGNO_OK_FOR_INDEX_P, REGNO_OK_FOR_BASE_P, REGNO_OK_FOR_DATA_P)
515 (REGNO_OK_FOR_FP_P, REG_OK_FOR_INDEX_P, REG_OK_FOR_BASE_P): Use them.
516 (EH_RETURN_STACKADJ_RTX): Use A0_REG.
517 * config/m68k/m68k.c (m68k_regno_mode_ok): Use the new REGNO macros.
519 2007-01-18 Nathan Sidwell <nathan@codesourcery.com>
521 * config.gcc (m68k-*-aout*, m68k-*-coff*, m68020-*-elf*, m68k-*-elf*)
522 (m68k-*-uclinuxoldabi, m68k-*-uclinux*, m68k-*-rtems*): Add t-floatlib
524 * config/m68k/t-floatlib: New file, extracting common code from...
525 * config/m68k/t-m68kbare, config/m68k/t-m68kelf,
526 * config/m68k/t-uclinux: Here.
527 * config/m68k/fpgnulib.c: Do not compile extendeed precision
528 routines on ColdFire targets.
530 2007-01-18 Nathan Sidwell <nathan@codesourcery.com>
532 * config.gcc (m68k-*-aout*, m68k-*-coff*, m68020-*-elf*, m68k-*-elf*)
533 (m68010-*-netbsdelf*, m68k*-*-netbsdelf*, m68k*-*-openbsd*)
534 (m68k-*-uclinuxoldabi, m68k-*-uclinux*, m68k-*-linux*)
535 (m68k-*-rtems*): Use tm_file rather than m68k/m68k.h and
536 explicitly set MOTOROLA to 1.
537 * config/m68k/m68k.h (MOTOROLA): Simplify definition accordingly.
539 2007-01-18 Richard Sandiford <richard@codesourcery.com>
540 Nathan Sidwell <nathan@codesourcery.com>
542 * config/m68k/m68k.h (PCC_STATIC_STRUCT_RETURN): Delete.
543 (ASM_OUTPUT_REG_PUSH, ASM_OUTPUT_REG_POP): Add MOTOROLA cases.
544 * config/m68k/coff.h (REGISTER_PREFIX_MD): Delete.
545 * config/m68k/m68020-elf.h (LIB_SPEC): Delete.
546 * config/m68k/m68k-none.h (CC1_SPEC, CPP_SUBTARGET_SPEC): Delete.
547 * config/m68k/m68kelf.h (IMMEDIATE_PREFIX, REGISTER_PREFIX_MD)
548 (ASM_OUTPUT_REG_PUSH, ASM_OUTPUT_REG_POP): Delete.
549 (NO_DOLLAR_IN_LABEL, PCC_STATIC_STRUCT_RETURN): Don't undefine.
550 (BSS_ASM_OP, ASM_OUTPUT_SKIP, ASM_OUTPUT_ASCII): Delete.
551 * config/m68k/m68kemb.h (PCC_STATIC_STRUCT_RETURN): Don't undefine.
552 * config/m68k/linux.h (SIZE_TYPE, PTRDIFF_TYPE, WCHAR_TYPE)
553 (WCHAR_TYPE_SIZE, TARGET_OBJFMT_CPP_BUILTINS): Delete.
554 (TARGET_OS_CPP_BUILTINS): Don't define mc68000 and mc68020 here.
555 (DBX_REGISTER_NUMBER): Delete.
556 * config/m68k/netbsd-elf.h (IMMEDIATE_PREFIX): Delete.
557 (PCC_STATIC_STRUCT_RETURN): Don't undefine.
558 * config/m68k/openbsd.h (PCC_STATIC_STRUCT_RETURN): Define.
560 2007-01-18 Richard Sandiford <richard@codesourcery.com>
562 * config.gcc (m68k-*-uclinux*): Add flat.h to $tm_file.
563 * config/flat.h: New file.
564 * crtstuff.c (USE_PT_GNU_EH_FRAME): Don't define if
566 * config/m68k/m68k.h (ASM_PREFERRED_EH_DATA_FORMAT): Do not use
567 indirect references for -msep-data or -mid-shared-library.
568 Do not use PC-relative code addresses either.
569 * config/m68k/m68k.c (override_options): Restrict -fPIC error
571 * config/m68k/uclinux.h (STARTFILE_SPEC): Define. Use Scrt1.o
572 for shared libraries and crt1.o for executables. Use crti.o and
574 (ENDFILE_SPEC): Use crtend.o and crtn.o.
575 (LIB_SPEC): Suppress -Rlibc.gdb if -static-libc is given.
576 Do not add -elf2flt or -shared-lib-id options here.
577 (LINK_SPEC): Define. Pass -elf2flt if no -elf2flt option is given.
578 Pass -shared-lib-id if -mid-shared-library, taking the library
579 identifier from -mshared-library-id if given, otherwise
581 (EH_FRAME_IN_DATA_SECTION): Do not undefine.
582 (INIT_SECTION_ASM_OP, FINI_SECTION_ASM_OP): Likewise.
583 (TARGET_OS_CPP_BUILTINS): Define __GXX_MERGED_TYPEINFO_NAMES=0
584 and __GXX_TYPEINFO_EQUALITY_INLINE=0 if -mid-shared-library.
585 (DRIVER_SELF_SPECS): Map unadorned PIC options to -msep-data.
586 * config/m68k/t-uclinux (EXTRA_MULTILIB_PARTS): Add crtbegin.o
588 * config/m68k/lb1sf68.asm (PICCALL): Use an lea and pc-relative
589 jump sequence for ISA A and ISA A+.
592 2007-01-18 Richard Sandiford <richard@codesourcery.com>
594 * config.gcc (m68k-*-uclinux*): Base the port on the common
595 and m68k GNU/Linux files rather than on the generic ELF ones.
596 * config/m68k/uclinux.h (TARGET_VERSION): Override.
597 (TARGET_OS_CPP_BUILTINS): Use LINUX_TARGET_OS_CPP_BUILTINS.
599 2007-01-18 Julian Brown <julian@codesourcery.com>
600 Richard Sandiford <richard@codesourcery.com>
602 * config/m68k/m68k.h (LONG_DOUBLE_TYPE_SIZE): Make 64-bit on ColdFire.
603 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
604 * config/m68k/netbsd-elf.h (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Undefine
606 * config/m68k/uclinux-oldabi.h (LONG_DOUBLE_TYPE_SIZE): Redefine to
608 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
610 2007-01-18 Richard Sandiford <richard@codesourcery.com>
612 * doc/install.texi: Document m68k-uclinuxoldabi.
613 * config.gcc (m68k-*-uclinuxoldabi): New configuration.
614 * config/m68k/uclinux-oldabi.h: New file, copied from
615 config/m68k/uclinux.h.
617 2007-01-18 Kaz Kojima <kkojima@rr.iij4u.or.jp>
619 * config/m32r/m32r-protos.h (m32r_expand_epilogue): Declare it.
620 * config/m32r/m32r.c (m32r_setup_incoming_varargs): Use gen_frame_mem.
621 (m32r_compute_frame_size): Use unsigned for regno.
622 (m32r_reload_lr): Use gen_frame_mem.
624 (m32r_output_function_epilogue): Don't output the function epilogue
626 (m32r_expand_epilogue): New.
627 (direct_return): Return false if the function has the interrupt
629 (m32r_hard_regno_rename_ok): Remove code for the textual epilogue.
630 * config/m32r/m32r.md (epilogue): New expander.
631 (return_lr, return_rte): New insns.
632 (return): Make it expander.
633 (return_normal): New expander.
635 2007-01-18 Josh Conner <jconner@apple.com>
638 * config/rs6000/rs6000.c (rs6000_emit_vector_compare): Add
639 support for UNLE, UNLT, UNGE, and UNGT.
641 2007-01-18 Jan Hubicka <jh@suse.cz>
643 * tree-vrp.c (finalize_jump_threads): Do not call cleanup_cfg by hand.
644 * tree-tailcall (add_virtual_phis): Likewise.
645 (optimize_tail_call): Return TODOs.
646 (execute_tail_calls): Return TODOs.
647 * tree-ssa-ccp (execute_fold_all_builtins): Do cleanup_cfg via TODO.
648 * tree-cfgcleanup.c (cleanup_tree_cfg_loop): Return if something
650 * tree-ssa-forwprop.c (tree_ssa_forward_propagate_single_use_value):
651 Cleanup cfg using TODO.
652 * tree-flow.h (cleanup_tree_cfg_loop): Update prototype.
653 * passes.c (execute_function_todo): When cleanup did something, remove
655 * tree-cfg.c (pass_build_cfg): Add cleanup_cfg TODO.
656 (make_edges): Don't cleanup_cfg.
658 2007-01-18 Uros Bizjak <ubizjak@gmail.com>
660 * reg-stack.c (subst_stack_regs_pat) [UNSPEC_SINCOS_COS,
661 UNSPEC_XTRACT_FRACT]: Use generic code for instructions that
662 operate on the top of stack.
663 [UNSPEC_SINCOS_SIN, UNSPEC_XTRACT_EXP, UNSPEC_TAN]: Rewrite
664 register handling of instructions that output to the second
666 [UNSPEC_TAN_ONE, UNSPEC_TAN_TAN]: Remove.
667 (move_for_stack_reg): Special-case check for dead destination
668 stack slot for constant load of 1.0 inside UNSPEC_TAN.
670 * config/i386/i386.md (UNSPEC_TAN): New constant.
671 (UNSPEC_TAN_ONE, UNSPEC_TAN_TAN): Remove.
672 (fptanxf4_i387, fptan_extend<mode>xf4_i387): New patterns
673 to correctly model move of constant 1.0 to top stack slot.
674 (*tandf3_1, *tansf3_1, *tanxf3_1): Remove insn patterns.
675 (unnamed peephole2 pattern): Remove corresponding peephole2
676 pattern that optimizes tan insn and loading of constant 1.0.
677 (tanxf2): Use fptanxf4_i387.
678 (tan<mode>2): Rename from tansf2 and tandf2 and macroize insn
679 patterns using X87MODEF12 mode macro. Use fptan_extend<mode>xf4_i387
680 and truncate result to requested mode. Use SSE_FLOAT_MODE_P to
681 disable patterns for SSE math.
682 (sincos<mode>3): Use truncxf<mode>2_i387_noop for truncation.
683 (fyl2x_extend<mode>xf3_i387): Use X87MODEF12 for operand 1.
685 2007-01-18 Dirk Mueller <dmueller@suse.de>
686 Richard Guenther <rguenther@suse.de>
689 * doc/invoke.texi (Warray-bounds): Document -Warray-bounds.
690 * common.opt (Warray-bounds): Add new warning option.
691 * c-opts.c (c_common_handle_option): Define -Warray-bounds
693 * Makefile.in: make tree-vrp.o depend on toplev.h
694 * tree-vrp.c (vrp_finalize): Call check_array_refs if -Warray-bounds
696 (check_array_refs, check_array_bounds, check_array_ref): New.
698 2007-01-18 Jan Hubicka <jh@suse.cz>
700 * tree-ssa-ccp.c (ccp_finalize): Return if something changed.
701 (execute_ssa_ccp): Return flags conditionally.
702 * tree-ssa-propagate.c (substitue_and_fold): Return if something was
704 * tree-ssa-propagate.h (substitute_and_fold): Update prototype.
706 2007-01-18 Steven Bosscher <steven@gcc.gnu.org>
708 * cfgcleanup.c (cleanup_cfg): Detect cfglayout mode and set
709 the CLEANUP_CFGLAYOUT flag when in cfglayout mode.
711 * Makefile.c (GTFILES): Add cfglayout.h.
712 * gengtype.c (open_base_files): Likewise.
713 * cfglayout.c (cfg_layout_function_footer,
714 cfg_layout_function_header) Reindent to make gengtype happy.
715 * cfglayout.h (cfg_layout_function_footer,
716 cfg_layout_function_header): Add GTY(()) marker.
718 * ifcvt.c (noce_try_sign_mask): Make sure INSN_B is non-null.
720 2007-01-18 Ben Elliston <bje@au.ibm.com>
722 * genautomata.c (write_automata): Include xstrerror output in the
723 error message if writing the DFA description file fails.
725 2007-01-17 H.J. Lu <hongjiu.lu@intel.com>
727 * config/mips/mips-protos.h (mips_output_external): Make it
729 * config/mips/iris.h (TARGET_ASM_EXTERNAL_LIBCALL): Removed.
730 * config/mips/mips.c (irix_output_external_libcall): Likewise.
731 (extern_list): Likewise.
732 (extern_head): Likewise.
733 (TARGET_ASM_FILE_END): Likewise.
734 (mips_file_end): Likewise.
735 (mips_output_external): Rewritten.
737 2007-01-18 Ben Elliston <bje@au.ibm.com>
739 * genpreds.c (write_insn_preds_c): Only write out the function
740 body for regclass_for_constraint if we have register constraints.
742 2007-01-17 Tom Tromey <tromey@redhat.com>
744 * doc/sourcebuild.texi (libgcj Tests): Use sourceware.org.
745 * doc/install.texi (Testing): Use sourceware.org.
746 (Binaries): Likewise.
747 (Specific): Likewise.
748 * doc/contrib.texi (Contributors): Use sourceware.org.
750 2007-01-17 Anatoly Sokolov <aesok@post.ru>
752 * config/avr/avr.h (AVR_HAVE_LPMX): New macro.
753 (AVR_ENHANCED): Rename to ...
754 (AVR_HAVE_MUL): ... new.
755 (avr_enhanced_p): Rename to ...
756 (avr_have_mul_p): ... new.
757 (TARGET_CPU_CPP_BUILTINS): Use 'avr_have_mul_p' instead of
758 'avr_enhanced_p' for "__AVR_ENHANCED__". Define "__AVR_HAVE_MUL__".
759 * config/avr/avr.c (avr_enhanced_p): Rename to ...
760 (avr_have_mul_p): ... new.
761 (base_arch_s): Rename 'enhanced' to 'have_mul'.
762 (avr_override_options): Use 'avr_have_mul_p' and 'have_mul' instead of
763 'avr_enhanced_p' and 'enhanced'.
764 (ashlhi3_out, ashrhi3_out, lshrhi3_out, avr_rtx_costs): Use
765 AVR_HAVE_MUL instead of AVR_ENHANCED.
766 * avr.md (*tablejump_enh): Use AVR_HAVE_LPMX instead of AVR_ENHANCED.
767 (mulqi3, *mulqi3_enh, *mulqi3_call, mulqihi3, umulqihi3, mulhi3,
768 *mulhi3_enh, *mulhi3_call, mulsi3, *mulsi3_call): Use AVR_HAVE_MUL
769 instead of AVR_ENHANCED.
770 (*tablejump_enh): Use AVR_HAVE_LPMX instead of AVR_ENHANCED.
771 * libgcc.S: Use __AVR_HAVE_MUL__ instead of __AVR_ENHANCED__.
772 (__tablejump__): Use __AVR_HAVE_LPMX__ instead of __AVR_ENHANCED__.
774 2007-01-17 Ian Lance Taylor <iant@google.com>
776 * vec.h (VEC_reserve_exact): Define.
777 (vec_gc_p_reserve_exact): Declare.
778 (vec_gc_o_reserve_exact): Declare.
779 (vec_heap_p_reserve_exact): Declare.
780 (vec_heap_o_reserve_exact): Declare.
781 (VEC_OP (T,A,reserve_exact)): New static inline function, three
783 (VEC_OP (T,A,reserve)) [all versions]: Remove handling of
785 (VEC_OP (T,A,alloc)) [all versions]: Call ...reserve_exact.
786 (VEC_OP (T,A,copy)) [all versions]: Likewise.
787 (VEC_OP (T,a,safe_grow)) [all versions]: Likewise.
788 * vec.c (calculate_allocation): Add exact parameter. Change all
790 (vec_gc_o_reserve_1): New static function, from vec_gc_o_reserve.
791 (vec_gc_p_reserve, vec_gc_o_reserve): Call vec_gc_o_reserve_1.
792 (vec_gc_p_reserve_exact, vec_gc_o_reserve_exact): New functions.
793 (vec_heap_o_reserve_1): New static function, from vec_heap_o_reserve.
794 (vec_heap_p_reserve, vec_heap_o_reserve): Call vec_heap_o_reserve_1.
795 (vec_heap_p_reserve_exact): New function.
796 (vec_heap_o_reserve_exact): New function.
798 2007-01-17 Jan Hubicka <jh@suse.cz>
800 * ipa-type-escape.c (look_for_casts): Revamp using handled_component_p.
802 2007-01-17 Eric Christopher <echristo@apple.com>
804 * config.gcc: Support core2 processor.
806 2007-01-16 Jan Hubicka <jh@suse.cz>
808 * tree-ssanames.c (release_dead_ssa_names): Instead of ggc_freeing
809 the names, just unlink the chain so we don't crash on dangling pointers
812 2007-01-16 Jan Hubicka <jh@suse.cz>
814 * cgraph.h (cgraph_decide_inlining_incrementally): Kill.
815 * tree-pass.h: Reorder to make IPA passes appear toegher.
816 (pass_early_inline, pass_inline_parameters, pass_apply_inline): Declare.
817 * cgraphunit.c (cgraph_finalize_function): Do not compute inling
818 parameters, do not call early inliner.
819 * ipa-inline.c: Update comments. Include tree-flow.h
820 (cgraph_decide_inlining): Do not compute inlining parameters.
821 (cgraph_decide_inlining_incrementally): Return TODOs; assume to
822 be called with function context set up.
823 (pass_ipa_inline): Remove unreachable functions before pass.
824 (cgraph_early_inlining): Simplify assuming to be called from the
826 (pass_early_inline): New pass.
827 (cgraph_gate_ipa_early_inlining): New gate.
828 (pass_ipa_early_inline): Turn into simple wrapper.
829 (compute_inline_parameters): New function.
830 (gate_inline_passes): New gate.
831 (pass_inline_parameters): New pass.
832 (apply_inline): Move here from tree-optimize.c
833 (pass_apply_inline): New pass.
834 * ipa.c (cgraph_remove_unreachable_nodes): Verify cgraph after
836 * tree-inline.c (optimize_inline_calls): Return TODOs rather than
838 (tree_function_versioning): Do not allocate dummy struct function.
839 * tree-inline.h (optimize_inline_calls): Update prototype.
840 * tree-optimize.c (execute_fixup_cfg): Export.
841 (pass_fixup_cfg): Remove
842 (tree_rest_of_compilation): Do not apply inlines.
843 * tree-flow.h (execute_fixup_cfg): Declare.
844 * Makefile.in (gt-passes.c): New.
845 * passes.c: Include gt-passes.h
846 (init_optimization_passes): New passes.
847 (nnodes, order): New static vars.
848 (do_per_function_toporder): New function.
849 (execute_one_pass): Dump current pass here.
850 (execute_ipa_pass_list): Don't dump current pass here.
852 2007-01-16 Janis Johnson <janis187@us.ibm.com>
854 * config/dfp-bit.c (dfp_compare_op): Return separate value for NaN.
855 (DFP_NE, DFP_LE, DFP_GE): Return false for NaN.
857 2007-01-16 David Edelsohn <edelsohn@gnu.org>
859 * config/rs6000/darwin-ldouble.c: Build file for SOFT_FLOAT.
860 (strong_alias): Define.
861 (__gcc_qmul): Provide non-FMA for soft-float.
877 * config/rs6000/rs6000.c (rs6000_init_libfuncs): Initialize
878 soft-float functions.
879 * config/rs6000/libgcc-ppc-glibc.ver: Version soft-float symbols.
880 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Do not warn
881 about long double soft float.
883 2007-01-16 Dorit Nuzman <dorit@il.ibm.com>
884 Tehila Meyzels <tehila@il.ibm.com>
886 * tree-vectorizer.h (is_pattern_stmt_p): New.
887 * tree-vect-analyze.c (vect_determine_vectorization_factor): Fix
888 formatting (tabs instead of spaces). Cleanup and clarify setting
889 of STMT_VINFO_VECTYPE. Call is_pattern_stmt_p.
890 * tree-vect-transform.c (vect_get_vec_def_for_operand): Fix typo.
891 (vectorizable_type_demotion): Check that types are integral.
892 (vectorizable_type_promotion): Likewise.
893 (vectorizable_store): Fix typo. Eliminate new-line at end of
896 2007-01-16 Jan Hubicka <jh@suse.cz>
898 * tree-ssanames.c (release_dead_ssa_names): Remove invalidated
901 2007-01-15 Eric Christopher <echristo@apple.com>
903 * ifcvt.c: Include vec.h, vecprim.h.
904 (check_cond_move_block): New argument regs.
905 Reorganize. Add registers used to regs.
906 (cond_move_process_if_block): Use regs set above as
909 2007-01-15 Eric Christopher <echristo@apple.com>
911 * config/darwin.h: Update copyright.
912 (TARGET_OPTION_TRANSLATE_TABLE): Add umbrella.
913 (LINK_COMMAND_SPEC): Add -u.
914 (LINK_SPEC): Fix umbrella for above.
916 2007-01-15 Joseph S. Myers <joseph@codesourcery.com>
918 * config/soft-fp/op-common.h, config/soft-fp/op-4.h: Update from
921 2007-01-15 Tom Tromey <tromey@redhat.com>
923 * doc/sourcebuild.texi (libgcj Tests): Don't mention jacks.
924 * doc/install.texi (Testing): Don't mention jacks.
925 (Configuration): Document --enable-java-maintainer-mode. Move
926 --with-java-home to libgcj-specific section. Document
928 (Prerequisites): Mention --enable-java-maintainer-mode, ecj1.
930 2007-01-15 Jan Hubicka <jh@suse.cz>
932 * tree-ssa-dce.c (DCE_TODOs): New.
933 (propagate_necessity): Return if something changed.
934 (eliminate_unnecessary_stmts): Likewise.
935 (perform_tree_ssa_dce): Return TODO flags when needed.
936 (pass_dce, pass_dce_loop, pass_cd_dce): Remove TODO flags.
938 2007-01-15 Uros Bizjak <ubizjak@gmail.com>
940 * config/i386/i386.md (fyl2xxf3_i387): Rename from fyl2x_xf3.
941 (fyl2x_extend<mode>xf3_i387): New insn pattern.
942 (log<mode>2): Rename from logsf2 and logdf2 and macroize insn
943 patterns using X87MODEF12 mode macro. Extend operand 1
944 to XFmode. Use SSE_FLOAT_MODE_P to disable patterns for SSE math.
945 (log10<mode>2): Ditto.
946 (log2<mode>2): Ditto.
947 (log1p<mode>2): Ditto.
948 (logb<mode>2): Ditto.
949 (fyl2xp1xf3_i387): Rename from fyl2xp1_xf3.
950 (fyl2xp1_extend<mode>xf3_i387): New insn pattern.
951 (*fxtractxf3_i387): Rename from *fxtractxf3.
952 (fxtract_extend<mode>xf3_i387): New insn pattern.
953 (ilogbsi2): Use match_dup 3, not match_operand:XF 3.
954 * config/i386/i386.c (ix86_emit_i387_log1p): Use gen_fyl2xp1xf3_i387()
955 and gen_fyl2xxf3_i387().
957 2007-01-14 Zdenek Dvorak <dvorakz@suse.cz>
959 * loop-unswitch.c (unswitch_loop): Do not call fix_loop_placement.
960 * cfgloopmanip.c (fix_loop_placement): Made static. Use
961 get_loop_exit_edges. Changed return type to bool.
962 * cfgloop.h (fix_loop_placement): Declaration removed.
964 2007-01-14 Dorit Nuzman <dorit@il.ibm.com>
966 * param.h (MIN_VECT_LOOP_BOUND): New.
967 * params.def (MIN_VECT_LOOP_BOUND): New.
968 * tree-vectorizer.c (slpeel_tree_peel_loop_to_edge): Takes another
969 argument - minimum threshold for number of iterations.
970 * tree-vectorizer.h (slpeel_tree_peel_loop_to_edge): Add another
971 argument to declaration.
972 * tree-vect-analyze.c (vect_analyze_operations): Check value of
974 * tree-vect-transform.c (vect_do_peeling_for_loop_bound): Call
975 slpeel_tree_peel_loop_to_edge with additional argument.
976 (vect_do_peeling_for_alignment): Likewise.
977 * doc/invoke.texi (min-vect-loop-bound): Document new param option.
979 2007-01-14 Uros Bizjak <ubizjak@gmail.com>
982 * config/i386/i386.c (print_operand) ['z']: Output 'b' for
985 2007-01-14 Jan Hubicka <jh@suse.cz>
987 * tree-dfa.c (remove_referenced_var): New function.
988 * tree-ssa-live.c (remove_unused_locals): Walk referenced vars and
989 prune referenced vars list too.
990 * tree-flow.h (remove_referenced_var): Declare.
992 2007-01-14 Jan Hubicka <jh@suse.cz>
994 * tree-eh.c (add_stmt_to_eh_region_fn): Do not add call_exprs
996 (remove_stmt_from_eh_region_fn): Do not remove call_exprs.
997 (verify_eh_throw_stmt_node, verify_eh_throw_table_statements): Kill.
998 * except.h (verify_eh_throw_table_statements): Kill prototype.
999 * cfgexpand.c (expand_gimple_basic_block): Propagate Eh regions
1000 into call exrepssions.
1001 * tree-optimize.c (execute_free_cfg_annotatiosn): Do not call
1003 * tree-cfg.c: Include pointer-set.h.
1004 (verify_node_sharing): Work on pointer set.
1005 (verify_eh_throw_stmt_node): New.
1006 (verify_stmts): Use pointers sets, verify throw_stmt.
1008 2007-01-13 Zdenek Dvorak <dvorakz@suse.cz>
1010 * ipa-reference.c (analyze_function): Consider also addresses taken
1013 2007-01-12 Roger Sayle <roger@eyesopen.com>
1015 * c-typeck.c (null_pointer_constant_p): Replace use of
1016 TREE_CONSTANT_OVERFLOW with TREE_OVERFLOW.
1017 (build_c_cast): Likewise.
1019 2007-01-12 Roger Sayle <roger@eyesopen.com>
1021 * tree.h (force_fit_type_double): Remove unused final argument.
1022 * c-common.c (constant_expression_warning): Replace use of
1023 TREE_CONSTANT_OVERFLOW with TREE_OVERFLOW.
1024 (convert_and_check): Likewise.
1025 (shorten_compare): Update call to force_fit_type_double.
1026 (c_common_truthvalue_conversion) <INTEGER_CST>: Use integer_zerop.
1027 * convert.c (convert_to_pointer): Update call to
1028 force_fit_type_double.
1029 * fold-const.c (force_fit_type_double): Remove overflowed_const
1031 (int_const_binop, fold_convert_const_int_from_int,
1032 fold_convert_const_int_from_real, fold_div_compare,
1033 fold_sign_changed_comparison, fold_unary, fold_negate_const,
1034 fold_abs_const, fold_not_const): Remove the final argument from
1035 calls to force_fit_type_double.
1037 2007-01-12 Andrew Pinski <andrew_pinski@playstation.sony.com>
1039 * configure.ac: Set insn to "nop" for spu-*-* also.
1040 * configure: Regenerate.
1042 2007-01-12 Olga Golovanevsky <olga@il.ibm.com>
1044 * builtins.def : Add BUILT_IN_FREE.
1046 2007-01-12 Jan Hubicka <jh@suse.cz>
1048 PR tree-optimization/30443
1049 * tree-inline.c (tree_function_versioning): Do not optimize when
1050 cloning for inlining.
1052 2007-01-12 Zdenek Dvorak <dvorakz@suse.cz>
1054 * doc/loop.texi: Document recording of loop exits.
1055 * cfgloopmanip.c (loopify, duplicate_loop): Use alloc_loop.
1056 (update_single_exits_after_duplication,
1057 update_single_exit_for_duplicated_loop,
1058 update_single_exit_for_duplicated_loops): Removed.
1059 (duplicate_loop_to_header_edge): Do not call
1060 update_single_exits_after_duplication and
1061 update_single_exit_for_duplicated_loops.
1062 (loop_version): Do not update single_exit information.
1063 (fix_loop_structure): Use record_loop_exits instead of
1064 mark_single_exit_loops.
1065 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update
1066 the lists of loop exits.
1067 * cfghooks.c (redirect_edge_and_branch, redirect_edge_and_branch_force,
1068 split_edge, merge_blocks): Update the lists of loop exits.
1069 * modulo-sched.c (sms_schedule): Pass LOOPS_HAVE_RECORDED_EXITS to
1070 loop_optimizer_init.
1071 * loop-init.c (loop_optimizer_init): Call record_loop_exits instead
1072 of mark_single_exit_loops.
1073 (loop_optimizer_finalize): Call release_recorded_exits.
1074 * tree-ssa-loop.c (tree_loop_optimizer_init): Pass
1075 LOOPS_HAVE_RECORDED_EXITS to loop_optimizer_init.
1076 * tree-vectorizer.c (slpeel_tree_duplicate_loop_to_edge_cfg): Do not
1077 update single exit information.
1078 * lambda-code.c (perfect_nestify): Ditto.
1079 * cfgloop.c (flow_loop_free): Destroy the list of exits of the loop.
1080 (mark_single_exit_loops): Removed.
1081 (alloc_loop, loop_exit_hash, loop_exit_eq, loop_exit_free,
1082 get_exit_descriptions, rescan_loop_exit, record_loop_exits,
1083 dump_recorded_exit, dump_recorded_exits, release_recorded_exits): New
1085 (get_loop_exit_edges, single_exit): Use recorded exit lists.
1086 (add_bb_to_loop, remove_bb_from_loops): Update the lists of loop exits.
1087 (verify_loop_structure): Verify consistency of the exit lists.
1088 (flow_loops_find): Use alloc_loop. Initialize exits hash.
1089 (set_single_exit): Removed.
1090 * cfgloop.h (struct loop_exit): New function.
1091 (struct loop): single_exit_ field replaced by exits field.
1092 (LOOPS_HAVE_MARKED_SINGLE_EXITS): Replaced by LOOPS_HAVE_RECORDED_EXITS.
1093 (struct loops): Added exits hash.
1094 (mark_single_exit_loops, set_single_exit): Declaration removed.
1095 (release_recorded_exits, record_loop_exits, rescan_loop_exit): Declare.
1097 2007-01-12 Richard Sandiford <richard@codesourcery.com>
1099 * doc/invoke.texi: Avoid use of @headitem.
1101 2007-01-12 Richard Sandiford <richard@codesourcery.com>
1103 * cse.c (cse_insn): Move HAVE_CC0 code after declarations.
1105 2007-01-12 Richard Sandiford <richard@codesourcery.com>
1107 * doc/install.texi: Fix m68k-*-* anchor and add m68k-*-* to the
1110 2007-01-12 Nathan Sidwell <nathan@codesourcery.com>
1111 Richard Sandiford <richard@codesourcery.com>
1113 * doc/invoke.texi: Document -mno-bitfield, -mno-rtd and -mno-short.
1114 * config/m68k/m68k.opt: Resort options.
1115 (mbitfield, mrtd, mshort): Remove RejectNegative properties.
1117 2007-01-12 Nathan Sidwell <nathan@codesourcery.com>
1118 Richard Sandiford <richard@codesourcery.com>
1120 * doc/invoke.texi: Document the macros that are defined by
1121 m68k's -mtune and -mhard-float options.
1122 * config/m68k/m68k-protos.h (m68k_cpp_cpu_ident) Declare.
1123 (m68k_cpp_cpu_family): Likewise.
1124 * config/m68k/m68k.h (TARGET_CPU_CPP_BUILTINS): Add a full set
1125 of __ucfv*__ macros. Define __mcffpu__ if generating code for
1126 ColdFire FPUs. Define __mcf_cpu_* and __mcf_family_* macros.
1127 * config/m68k/m68k.c (m68k_cpp_cpu_ident): New function.
1128 (m68k_cpp_cpu_family): Likewise.
1130 2007-01-12 Richard Sandiford <richard@codesourcery.com>
1132 * config/m68k/m68k.h (TARGET_CPU_CPP_BUILTINS): Treat all mc68*
1133 macros besides mc68000 as tuning macros. Use a switch statement
1134 to set them and mcpu32.
1136 2007-01-12 Julian Brown <julian@codesourcery.com>
1138 * config/m68k/m68k.h: Use TARGET_68040 instead of TARGET_68040_ONLY.
1139 (TARGET_68040_ONLY): Rename to...
1140 (TARGET_68040): ...this.
1141 * config/m68k/m68k.c: Use TARGET_68040 instead of TARGET_68040_ONLY.
1142 * config/m68k/m68k.md: Likewise.
1144 2007-01-12 Julian Brown <julian@codesourcery.com>
1145 Nathan Sidwell <nathan@codesourcery.com>
1146 Richard Sandiford <richard@codesourcery.com>
1148 * config.gcc (m680[012]0-*-*, m68k*-*-*): Set m68k_cpu_ident to
1149 the -mcpu= argument associated with the --with-cpu setting.
1150 Define M68K_DEFAULT_TUNE to the default -mtune= option,
1151 if different from the one implied by the -mcpu setting.
1152 Accept --with-cpu=FOO if FOO is listed in m68k-devices.def,
1153 using mcpu=FOO as the default CPU option. Set target_cpu_default2.
1154 * doc/invoke.texi: Mention ColdFire in the introduction to the
1155 m68k options. Document the new -march, -mcpu, -mtune, -mdiv,
1156 -mno-div and -mhard-float options. Make -m68881 a synonym for
1157 -mhard-float. Document the previously-undocumented -m5206e,
1158 -m528x, -m5307 and -m5407 options. Tweak the existing option
1159 documentation for consistency.
1160 * doc/install.texi: Mention new --with-cpu arguments.
1161 * config/m68k/m68k.h (OPTION_DEFAULT_SPECS): Only use the
1162 default CPU if neither -mcpu nor -march are specified.
1163 (ASM_CPU_SPEC): Pass down -mcpu and -march options.
1164 (TARGET_CPU_CPP_BUILTINS): Set __mcfisa*__ macros from
1165 TARGET_ISA*. Set the legacy __mcf*__ cpu macros in the same way,
1166 using m68k_tune to decide between families that implement the
1167 same ISA. Use m68k_tune to set __mcfv4e__.
1168 (FL_BITFIELD, FL_68881, FL_COLDFIRE, FL_CF_HWDIV, FL_CF_MAC)
1169 (FL_CF_EMAC, FL_CF_EMAC_B, FL_CF_USP, FL_CF_FPU, FL_ISA_68000)
1170 (FL_ISA_68010, FL_ISA_68020, FL_ISA_68040, FL_ISA_A, FL_ISA_B)
1171 (FL_ISA_C, FL_ISA_MMU): New macros.
1172 (MASK_COLDFIRE): Delete.
1173 (TARGET_68010, TARGET_68020, TARGET_68040_ONLY, TARGET_COLDFIRE)
1174 (TARGET_ISAB): Redefine in terms of m68k_cpu_flags.
1175 (TARGET_68881, TARGET_COLDFIRE_FPU): Redefine in terms of m68k_fpu.
1176 (TARGET_HARD_FLOAT): Do not define here.
1177 (TARGET_ISAAPLUS, TARGET_ISAC): New macros.
1178 (TUNE_68000): New macro.
1179 (TUNE_68000_10): Redefine in terms of TUNE_68000 and TUNE_68010.
1180 (TUNE_68010, TUNE_68030, TUNE_68040, TUNE_68060, TUNE_CPU32)
1181 (TUNE_CFV2): Redefine in terms of m68k_tune.
1182 (uarch_type, target_device, fpu_type): New enums.
1183 (m68k_cpu, m68k_tune, m68k_fpu, m68k_cpu_flags): Declare.
1184 * config/m68k/m68k.c (TARGET_DEFAULT): Remove MASK_68881.
1185 (FL_FOR_isa_00, FL_FOR_isa_10, FL_FOR_isa_20, FL_FOR_isa_40)
1186 (FL_FOR_isa_cpu32, FL_FOR_isa_a, FL_FOR_isa_aplus, FL_FOR_isa_b)
1187 (FL_FOR_isa_c): New macros.
1188 (m68k_isa): New enum.
1189 (m68k_target_selection): New structure.
1190 (all_devices, all_isas, all_microarchs): New tables.
1191 (m68k_cpu_entry, m68k_arch_entry, m68k_tune_entry, m68k_cpu)
1192 (m68k_tune, m68k_fpu, m68k_cpu_flags): New variables.
1193 (MASK_ALL_CPU_BITS): Delete.
1194 (m68k_find_selection): New function.
1195 (m68k_handle_option): Handle -mcpu=, -march= and -mtune=.
1196 Map the legacy target options to a combination of the new ones.
1197 (override_options): Set m68k_cpu, m68k_tune, m68k_fpu and
1198 m68k_cpu_flags. Handle M68K_DEFAULT_TUNE. Use m68k_cpu_flags
1199 to derive default MASK_BITFIELD, MASK_CF_HWDIV and MASK_HARD_FLOAT
1201 * config/m68k/m68k.opt (m5200, m5206e, m528x, m5307, m5407, mcfv4e)
1202 (m68010, m68020, m68020-40, m68020-60, m68030, m68040): Remove Mask
1204 (m68881, msoft-float): Change mask from 68881 to HARD_FLOAT.
1205 (march=, mcpu=, mdiv, mhard-float, mtune=): New options.
1206 * config/m68k/m68k-devices.def: New file.
1208 2007-01-12 Richard Sandiford <richard@codesourcery.com>
1209 Nathan Sidwell <nathan@codesourcery.com>
1211 * config/m68k/m68k.h (ASM_CPU_SPEC, ASM_SPEC, EXTRA_SPECS)
1212 (SUBTARGET_EXTRA_SPECS): New macros.
1213 * config/m68k/linux.h (ASM_SPEC): Remove CPU flags;
1214 use %(asm_cpu_spec) instead.
1215 * config/m68k/m68k-none.h (ASM_SPEC): Likewise.
1216 * config/m68k/openbsd.h (ASM_SPEC): Likewise.
1217 * config/m68k/netbsd-elf.h (ASM_SPEC): Likewise.
1218 (EXTRA_SPECS): Rename to...
1219 (SUBTARGET_EXTRA_SPECS): ...this.
1221 2007-01-12 Nathan Sidwell <nathan@codesourcery.com>
1222 Richard Sandiford <richard@codesourcery.com>
1223 Julian Brown <julian@codesourcery.com>
1225 * config.gcc (m68k-*-aout*, m68k-*-coff*, m68020-*-elf*, m68k-*-elf*)
1226 (m68k-*-uclinux*, m68k-*-linux*, m68k-*-rtems*): Set default_m68k_cpu
1227 to the configuration's default CPU.
1228 (m68010-*-netbsdelf*, m68k*-*-netbsdelf*, m68k*-*-openbsd*): Likewise.
1229 Remove default masks.
1230 (m680[012]0-*-*): Set the default with_cpu to the first part of
1232 (m68k*-*-*): Set the default with_cpu to m$default_m68k_cpu.
1233 (m68k*-*-linux): Extend the --with-cpu handling to...
1234 (m680[012]0-*-*, m68k*-*-*): ...these configurations. Allow m68000
1235 and m68010. Don't set target_cpu_default2.
1236 * doc/install.texi: Document --with-cpu for m68k.
1237 * config/m68k/m68k.h (OPTION_DEFAULT_SPECS): Define.
1238 * config/m68k/m68k-none.h (TARGET_CPU_DEFAULT, M68K_CPU_m68k)
1239 (M68K_CPU_m68000, M68K_CPU_m68010, M68K_CPU_m68020, M68K_CPU_m68030)
1240 (M68K_CPU_m68040, M68K_CPU_m68302, M68K_CPU_m68332, TARGET_DEFAULT)
1241 (ASM_CPU_DEFAULT_SPEC, CC1_CPU_DEFAULT_SPEC): Delete.
1242 (ASM_SPEC): Remove use of %(asm_cpu_default).
1243 (EXTRA_SPECS, SUBTARGET_EXTRA_SPECS, MULTILIB_DEFAULTS): Delete.
1244 * config/m68k/linux.h (TARGET_DEFAULT): Delete.
1245 (CPP_SPEC): Merge definitions. Do not handle __HAVE_68881__ here.
1246 * config/m68k/netbsd-elf.h (TARGET_OS_CPP_BUILTINS): Define
1247 __HAVE_FPU__ if TARGET_HARD_FLOAT.
1248 (TARGET_DEFAULT): Delete.
1249 (EXTRA_SPECS): Delete cpp_cpu_default_spec, cpp_cpu_spec,
1250 cpp_fpu_spec, asm_default_spec and netbsd_cpp_spec.
1251 (CPP_CPU_SPEC): Delete.
1252 (TARGET_VERSION): Merge definitions, using TARGET_68010 to pick
1253 the appropriate string.
1254 (CPP_CPU_DEFAULT_SPEC, ASM_DEFAULT_SPEC, CPP_FPU_SPEC): Delete.
1255 (CPP_SPEC): Define to NETBSD_CPP_SPEC.
1256 (ASM_SPEC): Don't use %(asm_default_spec).
1257 * config/m68k/m68k.c (TARGET_DEFAULT_TARGET_FLAGS): Remove
1258 TARGET_DEFAULT and add MASK_68881.
1259 * config/m68k/m68k.md: Remove mention of TARGET_DEFAULT from comments.
1261 2007-01-12 Richard Sandiford <richard@codesourcery.com>
1263 * config.gcc (m68010-*-netbsdelf*): Add MASK_68010.
1264 (m68k*-*-netbsdelf*, m68k*-*-openbsd*, m68k*-linux*): Add
1265 MASK_68010 alongside MASK_68020.
1266 * doc/invoke.texi: Document -m68010.
1267 * config/m68k/m68k.opt (m68010): New.
1268 * config/m68k/m68k.h (TARGET_CPU_CPP_BUILTINS): Define mc68010
1270 (TUNE_68010): New macro.
1271 * config/m68k/m68k-none.h (M68K_CPU_m68k, M68K_CPU_m68010)
1272 (M68K_CPU_m68020, M68K_CPU_m68030, M68K_CPU_m68040)
1273 (M68K_CPU_m68332): Add MASK_68010.
1274 * config/m68k/linux.h (TARGET_DEFAULT): Add MASK_68010 to
1275 fallback definition.
1276 * config/m68k/netbsd-elf.h (CPP_CPU_SPEC): Remove now-redundant
1278 * config/m68k/m68k.c (MASK_ALL_CPU_BITS): Add MASK_68010.
1279 (m68k_handle_option): Handle OPT_m68010. Add MASK_68010
1280 to all entries that use MASK_68020.
1281 (output_move_simode_const, output_move_himode, output_move_qimode)
1282 (output_move_stricthi, output_move_strictqi): Use TARGET_68010
1283 instead of TARGET_68020 to select clr behavior. Remove comment
1284 about there being no TARGET_68010.
1285 * config/m68k/m68k.md: Likewise throughout.
1287 2007-01-12 Julian Brown <julian@codesourcery.com>
1289 * config/m68k/m68k.h (TARGET_ISAB): New macro.
1290 * config/m68k/m68k.c: Use TARGET_ISAB rather than TARGET_CFV4.
1291 * config/m68k/m68k.md: Likewise.
1293 2007-01-12 Julian Brown <julian@codesourcery.com>
1295 * config/m68k/m68k.h (LEGITIMATE_INDEX_P, LEGITIMIZE_ADDRESS): Use
1296 TARGET_COLDFIRE_FPU instead of TARGET_CFV4E.
1298 2007-01-12 Julian Brown <julian@codesourcery.com>
1300 * config/m68k/m68k.h (TUNE_68040_60): New macro.
1301 * config/m68k/m68k.c (standard_68881_constant_p): Use it.
1302 * config/m68k/m68k.md: Likewise.
1304 2007-01-12 Julian Brown <julian@codesourcery.com>
1305 Richard Sandiford <richard@codesourcery.com>
1307 * config/m68k/m68k.h (TARGET_CPU_CPP_BUILTINS): Use TUNE_68030
1308 instead of TARGET_68030, TUNE_68040 instead of TARGET_68040,
1309 TUNE_68060 instead of TARGET_68060 and TUNE_CPU32 instead of
1311 (TARGET_CPU32): Rename to...
1312 (TUNE_CPU32): ...this.
1313 (TUNE_68000_10, TUNE_68030, TUNE_68040, TUNE_68060)
1314 (TUNE_CFV2): New macros.
1315 * config/m68k/netbsd-elf.h (LONG_DOUBLE_TYPE_SIZE): Simplify;
1316 remove conditions that are implied by TARGET_68020.
1317 * config/m68k/m68k.c (m68k_output_function_prologue): Use TUNE_68040
1318 instead of TARGET_68040 and TUNE_CPU32 instead of TARGET_CPU32.
1319 (m68k_output_function_epilogue): Likewise.
1320 (m68k_rtx_costs): Likewise. Use TUNE_68060 instead of TARGET_68060
1321 and TUNE_CFV2 instead of TARGET_5200. Use TUNE_68000_10 instead of
1322 "!TARGET_68020 && !TARGET_COLDFIRE" to choose between 68000 and
1323 non-68000 timings. Refactor multiplication and division costs.
1324 (output_addsi3): Use TUNE_68040 instead of TARGET_68040 and
1325 TUNE_CPU32 instead of TARGET_CPU32.
1326 (standard_68881_constant_p): Use TUNE_68040 instead of TARGET_68040
1327 and TUNE_68060 instead of TARGET_68060.
1328 * config/m68k/m68k.md: Use TUNE_68040 instead of TARGET_68040,
1329 TUNE_68060 instead of TARGET_68060, and TUNE_CPU32 instead of
1331 (movsi_const0): Use TUNE_68000_10 rather than "!TARGET_68020
1332 && !TARGET_COLDFIRE" to choose between moveq and clr.
1333 Likewise in the unnamed movsf pattern.
1334 (ashlsi_17_24, lshrsi_17_24): Guard with TUNE_68000_10 rather than
1335 "!TARGET_68020 && !TARGET_COLDFIRE". Likewise the unnamed
1338 2007-01-12 Richard Sandiford <richard@codesourcery.com>
1340 * config/m68k/m68k.h (TARGET_CPU_CPP_BUILTINS): Increase amount
1341 of tabbing before backslashes.
1343 2007-01-11 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
1345 * pa-linux.h (ASM_OUTPUT_INTERNAL_LABEL): Undefine.
1346 * pa.h (ASM_OUTPUT_LABEL): Output colon when using GAS.
1347 (ASM_OUTPUT_INTERNAL_LABEL): Define.
1349 2007-01-11 Zdenek Dvorak <dvorakz@suse.cz>
1351 * tree-ssa-loop-ivopts.c (extract_cond_operands): Split from
1352 find_interesting_uses_cond.
1353 (find_interesting_uses_cond): Use extract_cond_operands.
1354 (rewrite_use_compare): Use extract_cond_operands and
1355 force_gimple_operand_bsi. Do not call update_stmt.
1356 (determine_use_iv_cost_condition): Use extract_cond_operands.
1357 Return cheaper of using original bound and changing the exit bound.
1359 2007-01-11 Zdenek Dvorak <dvorakz@suse.cz>
1361 PR tree-optimization/29516
1362 * tree-ssa-address.c (tree_mem_ref_addr, add_to_parts,
1363 most_expensive_mult_to_index, addr_to_parts,
1364 create_mem_ref, maybe_fold_tmr): Make the type of
1365 fields of TARGET_MEM_REF sizetype.
1366 (move_fixed_address_to_symbol, move_pointer_to_base):
1368 * tree.def (TARGET_MEM_REF): Add comment on types of
1371 2007-01-11 Joseph Myers <joseph@codesourcery.com>
1373 * c-common.c (vector_types_convertible_p): Treat opaque types as
1374 always convertible if they have the same size, but not otherwise.
1376 2007-01-11 Steven Bosscher <steven@gcc.gnu.org>
1378 * ifcvt.c (struct noce_if_info): Add comments to the fields.
1379 Remove the b_unconditional field.
1380 (noce_try_sign_mask): Do not look at b_unconditional.
1381 (noce_process_if_block): Do not use merge_if_blocks. Update
1382 the CFG here. Do not set b_unconditional.
1383 (cond_move_process_if_block): Likewise.
1384 (find_cond_trap): Likewise.
1385 (check_cond_move_block): Require simple jump insns at the end
1388 2007-01-11 Jan Hubicka <jh@suse.cz>
1390 PR tree-optimization/1046
1391 * tree-tailcall.c (suitable_for_tail_call_opt_p): Use TREE_ADDRESSABLE
1392 when alias info is not ready.
1393 (pass_tail_recursion): Do not require aliasing.
1394 * tree-ssa-copyrename.c (pass_rename_ssa_cop): Likewise.
1395 * tree-ssa-ccp.c (pass_ccp, pass_fold_builtins): Likewise.
1396 * tree-ssa-copy.c (pass_copy_prop): Likewise.
1397 * tree-ssa-forwprop.c (pass_forwprop): Likewise.
1398 * tree-ssa-dce.c (pass_dce, pass_dce_loop, pass_cd_dce): Likewise.
1399 * passes.c (init_optimization_passes): Execute rename_ssa_copies,
1400 ccp, forwprop, copy_prop, merge_phi, copy_prop, dce and tail recursion
1402 * tree-ssa-operands.c (add_virtual_operand, get_indirect_ref_operand):
1403 When aliasing is not build, mark statement as volatile.
1405 2007-01-11 Tom Tromey <tromey@redhat.com>
1407 PR preprocessor/15185, PR preprocessor/20989:
1408 * doc/cppopts.texi <-MT>: Update description of algorithm for
1409 computing default target.
1410 <-M, -MD>: Reword "basename" text.
1412 2007-01-11 Roger Sayle <roger@eyesopen.com>
1414 * builtins.c (expand_builtin_pow, expand_builtin_powi,
1415 fold_builtin_cabs, fold_builtin_sqrt, fold_builtin_trunc,
1416 fold_builtin_floor, fold_builtin_ceil, fold_builtin_round,
1417 fold_builtin_int_int_roundingfn, fold_builtin_bitop,
1418 fold_builtin_bswap, real_constp, fold_builtin_pow,
1419 fold_builtin_powi, fold_builtin_signbit, fold_builtin_copysign,
1420 do_mpfr_arg1, do_mpfr_arg2, do_mpfr_arg3, do_mpfr_sincos): Replace
1421 uses of the macro TREE_CONSTANT_OVERFLOW with TREE_OVERFLOW.
1422 * convert.c (convert_to_pointer): Likewise.
1423 * expr.c (highest_pow2_factor, expand_expr_real_1): Likewise.
1424 * fold-const.c (force_fit_type, fold_negate_expr, int_const_binop,
1425 const_binop, fold_convert_const_int_from_int,
1426 fold_convert_const_int_from_real,
1427 fold_convert_const_real_from_real, sign_bit_p,
1428 optimize_minmax_comparison, extract_muldiv_1, fold_div_compare,
1429 fold_sign_changed_comparison, fold_unary, fold_comparison,
1430 fold_binary, multiple_of_p, tree_Expr_non_zero_p,
1431 fold_negate_const, fold_abs_const, fold_not_const): Likewise.
1432 * print-tree.c (print_node_brief, print_node): Likewise.
1433 * stor-layout.c (place_field, layout_type): Likewise.
1434 * tree-chrec.c (keep_cast): Likewise.
1435 * tree.c (build_vector, build_real, build_real_from_int_cst,
1436 build_complex): Likewise.
1438 2007-01-11 Roger Sayle <roger@eyesopen.com>
1440 * tree.h (TREE_CONSTANT_OVERFLOW): Obsolete. For the time being,
1441 treat TREE_CONSTANT_OVERFLOW as a synonym of TREE_OVERFLOW.
1443 2007-01-11 Paolo Bonzini <bonzini@gnu.org>
1445 * configure.ac (strict1_warn): Rename to strict_warn.
1446 (WERROR, --enable-werror, symlink hacks, stage1_cflags,
1447 cc_set_by_configure, quoted_cc_set_by_configure,
1448 stage_prefix_set_by_configure, quoted_stage_prefix_set_by_configure,
1449 all_boot_languages, all_stagestuff): Remove.
1450 (target_list): Remove bootstrap targets.
1451 * Makefile.in (quickstrap): Unconditionally make a synonym of all.
1452 (BOOT_LANGUAGES, STAGE1_CFLAGS, STAGE1_CHECKING,
1453 REMAKEFLAGS, FLAGS_TO_PASS, PREPEND_DOTDOT_TO_RELATIVE_PATHS,
1454 SUBDIR_FLAGS_TO_PASS, WERROR_FLAGS, STRICT2_WARN, LANG_STAGESTUFF,
1455 VOL_FILES, POSTSTAGE1_FLAGS_TO_PASS, STAGE2_FLAGS_TO_PASS,
1456 STAGEPROFILE_FLAGS_TO_PASS, STAGEFEEDBACK_FLAGS_TO_PASS, stage1_build,
1457 stage1_copy, stage2_build, stage2_copy, stageprofile_build,
1458 stageprofile_copy, stage3_build, stage3_copy, stagefeedback_build,
1459 stagefeedback_copy, stage4_build, clean_s1, clean_sw, bootstrap,
1460 bootstrap-lean, bootstrap2, bootstrap2-lean, bootstrap3,
1461 bootstrap3-lean, bootstrap4, bootstrap4-lean, unstage1, unstage2,
1462 unstage3, unstage4, unstageprofile, unstagefeedback, restage, restage2,
1463 restage3, restage4, restageprofile, restagefeedback, bubbleestrap,
1464 cleanstrap, unstrap, restrap, *compare, *compare3, *compare4,
1465 *compare-lean, *compare3-lean, *compare4-lean, stage1-start, stage1,
1466 stage2-start, stage2, stage3-start, stage3, stage4-start, stage4,
1467 stageprofile-start, stageprofile, stagefeedback-start, stagefeedback,
1468 risky-stage1, risky-stage2, risky-stage3, risky-stage4): Remove.
1469 (ORDINARY_FLAGS_TO_PASS): Rename to FLAGS_TO_PASS.
1470 (STAGECOPYSTUFF, STAGEMOVESTUFF): Consolidate into MOSTLYCLEANFILES.
1471 (mostlyclean): Adjust.
1472 (clean, distclean): Don't mention bootstrap stuff.
1473 * configure: Regenerate.
1474 * ada/config-lang.in, cp/config-lang.in, forttran/config-lang.in,
1475 java/config-lang.in, objc/config-lang.in, objcp/config-lang.in,
1476 treelang/config-lang.in (stagestuff): Remove.
1477 * doc/sourcebuild.texi (stage1, stage2, stage3, stage4,
1478 stageprofile, stagefeedback, stagestuff): Remove mention.
1480 2007-01-11 Nick Clifton <nickc@redhat.com>
1482 * config/mcore/predicates.md (mcore_general_movesrc_operand):
1484 (mcore_general_movdst_operand): Do not accept CONST_INTs.
1485 (mcore_arith_K_S_operand): Run the test for the S constraint not
1486 the test for the M constraint.
1487 (mcore_addsub_operand): Do not accept integer values that are
1488 larger than 32 bits.
1489 * config/mcore/mcore.md: Remove unused constraints from split.
1490 (andsi3): Use HOST_WIDE_INT instead of int to hold an INTVAL.
1492 (allocate_stack): Likewise.
1493 * config/mcore/mcore.c (mcore_print_operand): Restrict output of P
1494 operands to 32 bits.
1495 (mcore_const_costs): Use HOST_WIDE_INT instead of int to hold an
1497 (mcore_and_cost, mcore_modify_comparison, const_ok_for_mcore,
1498 mcore_const_ok_for_inline, mcore_const_trick_uses_not,
1499 try_constant_tricks, mcore_num_ones, mcore_num_zeros,
1500 mcore_output_bclri, mcore_output_andn, output_inline_const,
1501 mcore_output_move, mcore_output_movedouble): Likewise.
1502 (mcore_output_cmov): Use CONST_OK_FOR_M and CONST_OK_FOR_N.
1503 (output_inline_const): Likewise.
1504 (output_inline_const): Fix format strings used in sprintf
1506 * config/mcore/mcore-protos.h: Update prototypes for changed
1507 functions in mcore.c.
1508 * config/mcore/mcore.h (CONST_OK_FOR_I): Cast values to
1509 HOST_WIDE_INT and not int.
1510 (CONST_OK_FOR_J, CONST_OK_FOR_K, CONST_OK_FOR_L, CONST_OK_FOR_M,
1511 CONST_OK_FOR_N): Likewise.
1512 (LEGITIMATE_CONSTANT_P): Also check CONSTANT_P.
1513 (GO_IF_LEGITIMATE_INDEX): Use HOST_WIDE_INT instead of int to hold
1516 2007-01-10 Jan Hubicka <jh@suse.cz>
1518 * tree-vrp.c (remove_range_assertions): Release defs.
1519 * tree-ssa-loop-ivopts.c (rmeove_statement): Likewise.
1520 * tree-ssa-dom.c (remove_stmt_or_phi): Likewise.
1522 2007-01-10 Paul Brook <paul@codesourcery.com>
1524 * config/arm/arm.c (arm_rtx_costs_1): Handle mutiply-subtract.
1525 * config/arm/arm.md (mulsi3subsi): New insn.
1527 2007-01-10 Zdenek Dvorak <dvorakz@suse.cz>
1529 * tree-ssa-loop-manip.c (tree_unroll_loop): Make it a wrapper over ...
1530 (tree_transform_and_unroll_loop): New.
1531 * tree-flow.h (transform_callback, tree_transform_and_unroll_loop):
1534 2007-01-10 Robert Kennedy <jimbob@google.com>
1536 * fold-const.c (fold_comparison): Fold comparisons like (x *
1537 1000 < 0) to (x < 0).
1539 2007-01-10 Ian Lance Taylor <iant@google.com>
1541 * tree-pretty-print.c (dump_generic_node): Print parentheses when
1542 operands have the same priority.
1544 2007-01-10 Tom Tromey <tromey@redhat.com>
1546 * fold-const.c (fold_truthop): Don't check can_use_bit_fields_p.
1547 (fold_binary): Likewise.
1548 * langhooks.c (lhd_can_use_bit_fields_p): Removed.
1549 * langhooks-def.h (lhd_can_use_bit_fields_p): Removed.
1550 (LANG_HOOKS_CAN_USE_BIT_FIELDS_P): Removed.
1551 (LANG_HOOKS_INITIALIZER): Remove LANG_HOOKS_CAN_USE_BIT_FIELDS_P.
1552 * langhooks.h (struct lang_hooks): Removed field
1553 'can_use_bit_fields_p'.
1555 2007-01-10 Ralf Corsépius <ralf.corsepius@rtems.org>
1557 * config/bfin/t-bfin, config/bfin/t-bfin-elf: Remove GCC_CFLAGS.
1559 2007-01-10 Razya Ladelsky <razya@il.ibm.com>
1561 * function.c (get_last_funcdef_no): New function.
1562 * function.h (get_last_funcdef_no): Declare.
1563 * tree-inline.c (initialize_cfun): Add initialization.
1564 (tree_function_versioning): Cleanup.
1566 2007-01-10 Jan Hubicka <jh@suse.cz>
1568 * tree-inline.c (setup_one_parameter): Do not propagate into abnormal
1571 2007-01-10 Sa Liu <saliu@de.ibm.com>
1572 Ben Elliston <bje@au.ibm.com>
1574 * spu.h (STACK_SAVE_AREA): Use VOIDmode for SAVE_FUNCTION, SImode
1575 for SAVE_NONLOCAL and Pmode for any other save level.
1576 * spu-protos.h (spu_restore_stack_block): Declare.
1577 * spu.md (save_stack_block): Remove.
1578 (restore_stack_block): Call spu_restore_stack_block.
1579 * spu.c (spu_restore_stack_block): New function.
1580 (spu_expand_epilogue): Remove old comment.
1582 2007-01-09 Zdenek Dvorak <dvorakz@suse.cz>
1584 PR tree-optimization/30322
1585 * tree-ssa-loop-ivopts.c (fold_affine_expr, iv_value): Removed.
1586 (cand_value_at): Return the value as aff_tree.
1587 (may_eliminate_iv): Convert the bound from aff_tree to tree.
1588 * tree-affine.c (aff_combination_add_cst, aff_combination_add_product,
1589 aff_combination_mult): New functions.
1590 (aff_combination_add): Use aff_combination_add_cst.
1591 (aff_combination_convert): Allow conversions to a wider type.
1592 (tree_to_aff_combination): Handle BIT_NOT_EXPR.
1593 * tree-affine.h (aff_combination_mult): Declare.
1595 2007-01-09 Carlos O'Donell <carlos@codesourcery.com>
1597 * doc/tm.texi: Update documentation to reflect reality of exec
1598 and start file search behaviours. Update copyright year.
1599 * doc/invoke.texi: Explain how GCC_EXEC_PREFIX is used to find
1600 header file directories.
1602 2007-01-09 Uros Bizjak <ubizjak@gmail.com>
1604 * config/i386/i386.md (*sinxf2): Rename to *sinxf2_i387.
1605 (*cosxf2): Rename to cosxf2_i387.
1606 (*sindf2, *sinsf2): Extend operand 1 to XFmode. Macroize patterns
1607 using X87MODEF12 mode macro. Rename patterns to
1608 *sin_extend<mode>xf2_i387. Use SSE_FLOAT_MODE_P to disable patterns
1610 (*cosdf2, *cossf2): Ditto.
1611 (sincosdf3, sincossf3): Ditto. Rewrite corresponding splitters
1612 to match extended input operands.
1613 (sincos<mode>3): New expander.
1614 (*sinextendsfdf2, *cosextendsfdf2, *sincosextendsfdf3): Remove
1615 insn patterns and corresponding splitters.
1617 2007-01-09 Kaz Kojima <kkojima@gcc.gnu.org>
1619 * config/sh/t-linux (TARGET_LIBGCC2_CFLAGS): Delete.
1620 (SHLIB_MAPFILES, SHLIB_LINK, SHLIB_INSTALL): Likewise.
1622 2007-01-09 Nicolas Pitre <nico@cam.org>
1625 * arm/ieee754-df.S (Lad_s): Also test the low word of X for zero.
1627 2007-01-08 Geoffrey Keating <geoffk@apple.com>
1629 * target.h (struct gcc_target): New field library_rtti_comdat.
1630 * target-def.h (TARGET_CXX_LIBRARY_RTTI_COMDAT): New.
1631 (TARGET_CXX): Add TARGET_CXX_LIBRARY_RTTI_COMDAT.
1632 * doc/tm.texi (C++ ABI): Document TARGET_CXX_LIBRARY_RTTI_COMDAT.
1633 * config/darwin.h (TARGET_CXX_LIBRARY_RTTI_COMDAT): Define.
1635 2007-01-08 Geoffrey Keating <geoffk@apple.com>
1637 * doc/invoke.texi (Optimize Options): Correct description of -O0.
1639 2007-01-08 Richard Guenther <rguenther@suse.de>
1641 * tree.h (force_fit_type_double): Export.
1642 (force_fit_type): Remove.
1643 * fold-const.c (force_fit_type_double): New function.
1644 (force_fit_type): Remove.
1645 (int_const_binop): Use it.
1646 (fold_convert_const_int_from_int): Likewise.
1647 (fold_convert_const_int_from_real): Likewise.
1648 (fold_div_compare): Likewise.
1649 (fold_sign_changed_comparison): Likewise.
1650 (fold_unary): Likewise.
1651 (fold_negate_const): Likewise.
1652 (fold_abs_const): Likewise.
1653 (fold_not_const): Likewise.
1654 * c-common.c (shorten_compare): Use force_fit_type_double.
1655 * convert.c (convert_to_pointer): Likewise.
1657 2007-01-08 Richard Guenther <rguenther@suse.de>
1659 * tree.h (build_int_cst_wide_type): Export.
1660 * tree.c (build_int_cst_wide_type): New function.
1661 (build_int_cst_wide): Fix comment.
1662 * builtins.c (fold_builtin_object_size): Use build_int_cst
1663 to build -1 or 0 of the correct type. Use fit_double_type
1664 to check for overflow.
1665 * fold-const.c (optimize_bit_field_compare): Use build_int_cst_type
1667 (decode_field_reference): Likewise.
1668 (all_ones_mask_p): Likewise.
1669 (native_interpret_int): Use build_int_cst_wide_type.
1670 (fold_binary): Use build_int_cst_type to build an all-ones
1672 * stor-layout.c (set_sizetype): Use build_int_cst_wide_type.
1674 2007-01-08 Daniel Jacobowitz <dan@codesourcery.com>
1676 * config/pa/t-pa64 (libgcc_stub.a): Use $(T).
1678 2007-01-09 Ben Elliston <bje@au.ibm.com>
1680 * genautomata.c (STATS_OPTION): New option.
1681 (stats_flag): New flag.
1682 (gen_automata_option): Handle it.
1683 (initiate_automaton_gen): Ditto.
1684 (write_automata): Output statistics only if stats_flag is
1685 set. Likewise, output time statistics only if time_flag is set.
1686 * doc/md.texi (Processor pipeline description): Document new flag.
1688 2007-01-08 Richard Guenther <rguenther@suse.de>
1690 * builtins.c (fold_builtin_int_roundingfn): Use fit_double_type.
1691 * tree.c (build_int_cst_type): Likewise.
1692 (size_in_bytes): Don't call force_fit_type on the result.
1693 (int_fits_type_p): Use fit_double_type.
1694 * fold-const.c (fit_double_type): New function.
1695 (force_fit_type): Use it.
1696 * tree.h (fit_double_type): Export.
1698 2007-01-08 Jan Hubicka <jh@suse.cz>
1700 * tree-vectorizer.c (gate_increase_alignment): Fix return type.
1701 * ipa.c (function_and_variable_visibility): Fix return type.
1703 2007-01-08 Richard Guenther <rguenther@suse.de>
1705 * tree-ssa-ccp.c (maybe_fold_offset_to_array_ref): Use type
1706 of offset to build the index.
1707 * tree-pretty-print.c (dump_generic_node): Don't build negated
1708 const just for printing.
1709 * c-pretty-print.c (pp_c_integer_constant): Likewise.
1710 * builtins.c (fold_builtin_int_roundingfn): Check if result
1711 fits the type by using force_fit_type and comparing the result.
1712 * predict.c (predict_loops): Use compare_tree_int for comparison.
1713 * tree.c (build_int_cst): Fall back to integer_type_node for
1715 (build_int_cst_wide): Assert type is non-null.
1717 2007-01-08 Roberto Costa <roberto.costa@st.com>
1719 * tree-vrp.c (extract_range_from_cond_expr): New.
1720 (extract_range_from_expr): Handle COND_EXPR nodes used as expressions.
1721 * tree-ssa-ccp.c (get_maxval_strlen): Handle COND_EXPR nodes used
1723 (fold_stmt): Bug fix, avoid infinite recursion when folding COND_EXPRs.
1724 * tree-ssa-forwprop.c (simplify_cond, forward_propagate_into_cond,
1725 tree_ssa_forward_propagate_single_use_vars): Handle COND_EXPR nodes
1726 used as expressions.
1727 * tree-object-size.c (cond_expr_object_size): New.
1728 (collect_object_sizes_for): Handle COND_EXPR nodes used as expressions.
1730 2007-01-08 Jan Hubicka <jh@suse.cz>
1732 * tree-ssa-forwprop.c (forward_propagate_into_cond,
1733 tree_ssa_forward_propagate_single_use_va): Release defs of propagated
1736 2007-01-08 Richard Guenther <rguenther@suse.de>
1738 PR tree-optimization/23603
1739 * tree-vrp.c (set_value_range_to_truthvalue): New function.
1740 (extract_range_from_binary): Fall back to truthvalue instead of
1741 varying for TRUTH_*_EXPR.
1742 (extract_range_from_comparison): Fall back to truthvalue instead of
1744 (vrp_visit_phi_node): Don't adjust new range bounds to +INF/-INF
1745 if all visited PHI values were constant.
1747 2007-01-08 Jan Hubicka <jh@suse.cz>
1749 * cgraphunit.c (cgraph_process_new_functions): Reset reachable flag.
1750 (cgraph_analyze_function): break out from ...
1751 (cgraph_finalize_compilation_unit): ... here.
1752 (cgraph_expand_function): Remove forgoten commented out line.
1753 (cgraph_optimize): Analyze functions.
1755 2007-01-08 Jan Hubicka <jh@suse.cz>
1757 * tree-pas.h (TODO_remove_function): New flag.
1758 (TODO_update*): Renumber.
1759 (pass_ipa_increase_alignment,
1760 pass_ipa_function_and_variable_visibility): New passes.
1761 * cgraphunit.c (cgraph_increase_alignment): Move to tree-vectorizer.c
1762 (cgraph_function_and_variable_visibility): Move to ipa.c
1763 (cgraph_optimize): Don't call cgraph_function_and_variable_visibility,
1764 cgraph_increase_alignment.
1765 * ipa-inline.c (cgraph_decide_inlining): Don't push timevar.
1766 (cgraph_decide_inlining_incrementally): Push TV_INTEGRATION before
1767 calling tree-inline.
1768 (cgraph_early_inlining): Do not call cgraph_remove_unreachable_nodes.
1769 (pass_ipa_inline, pass_early_ipa_inlining): Set TODO_remove_functions
1770 * tree-vectorizer.c (increase_alignment): Move here from cgraphunit.c
1771 (gate_increase_alignment): New function.
1772 (pass_ipa_increase_alignment): New pass.
1773 * ipa.c: Inline tree-pass.h and timevar.h
1774 (function_and_variable_visibility): Move here from cgraphunit.c
1775 * tree-optimize.c (pass_early_local_passes): Add TODO_remove_functions.
1776 * passes.c (init_optimization_passes): Add the two new passes.
1777 (execute_todo): Handle cgraph_remove_functions.
1779 2007-01-08 Nick Clifton <nickc@redhat.com>
1781 * config/frv/predicates.md (reg_or_0_operand): Accept
1784 2007-01-08 Ralf Corsépius <ralf.corsepius@rtems.org>
1786 * config/bfin/rtems.h, config/bfin/t-rtems: New.
1787 * config.gcc: Add bfin*-rtems*.
1789 2007-01-08 Mark Shinwell <shinwell@codesourcery.com>
1791 * c.opt: Add -flax-vector-conversions.
1792 * c-typeck.c (convert_for_assignment): Pass flag to
1793 vector_types_convertible_p to allow emission of note.
1794 (digest_init): Likewise.
1795 * c-opts.c: Handle -flax-vector-conversions.
1796 * c-common.c (flag_lax_vector_conversions): New.
1797 (vector_types_convertible_p): Unless -flax-vector conversions
1798 has been passed, disallow conversions between vectors with
1799 differing numbers of subparts and/or element types. If such
1800 a conversion is disallowed, possibly emit a note on the first
1801 occasion only to inform the user of -flax-vector-conversions.
1802 The new last argument specifies this.
1803 * c-common.h (flag_lax_vector_conversions): New.
1804 (vector_types_convertible_p): Add extra argument.
1805 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use
1806 char_type_node for V*QI type vectors.
1807 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
1808 Update to satisfy new typechecking rules.
1809 * config/rs6000/altivec.h (vec_cmple): Use vec_cmpge, for both
1811 * doc/invoke.texi (C Dialect Options): Document
1812 -flax-vector-conversions.
1814 2007-01-08 Mark Shinwell <shinwell@codesourcery.com>
1816 PR tree-optimization/29877
1817 * tree-ssa-ter.c (is_replaceable_p): Deem assignments with
1818 a register variable on the RHS to not be replaceable.
1820 2007-01-08 Chen Liqin <liqin@sunnorth.com.cn>
1821 * config/score/t-score-elf (MULTILIB_OPTIONS): Change.
1822 * config/score/predicates.md (const_uimm5, sr0_operand, const_simm12,
1823 const_simm15, const_pow2, const_npow2): Added.
1824 * config/score/misc.md (insv, extv, extzv, movmemsi,
1825 move_lbu_a/b, mov_lhu_a/b etc): Added and fix some bug.
1826 * config/score/score.c (score_address_cost, score_select_cc_mode):
1828 Change CONST_OK_FOR_LETTER_P/EXTRA_CONSTRAINT define.
1829 Update score_rtx_costs for MACRO TARGET_RTX_COSTS.
1830 Update score_print_operand.
1831 * config/score/score.h (DATA_ALIGNMENT, SELECT_CC_MODE): Added.
1832 Adjust register allocate order and update some macro define.
1833 * config/score/score-mdaux.c (mdx_unaligned_load, mdx_unsigned_store,
1834 mdx_block_move_straight, mdx_block_move_loop_head,
1835 mdx_block_move_loop_body, mdx_block_move_loop_foot, mdx_block_move_loop,
1836 mdx_block_move): Added.
1837 (mdx_movsicc, mdp_select_add_imm, mdp_select, mds_zero_extract_andi,
1838 mdp_limm): Updated and fix some bug and typo.
1839 * config/score/score.md (movqi/hi/si, add/sub/zero/ext): Updated.
1840 (movsf, movdf, doloop_end): Added.
1842 2007-01-08 Kazu Hirata <kazu@codesourcery.com>
1844 * config/arm/arm.c, config/arm/arm.h, config/arm/arm.md,
1845 config/arm/thumb2.md: Fix comment typos.
1846 * doc/extend.texi: Fix a typo.
1848 2007-01-07 Eric Christopher <echristo@apple.com>
1850 * configure.ac: Check for __stack_chk_fail for darwin.
1851 * configure: Regenerate.
1853 2007-01-07 Richard Guenther <rguenther@suse.de>
1855 * tree-vrp.c (extract_range_from_assert): CSE calls to
1856 compare_values where possible.
1857 (extract_range_from_unary_expr): Likewise.
1859 2007-01-07 Anatoly Sokolov <aesok@post.ru>
1861 * config/avr/avr-protos.h (call_insn_operand): Delete prototype.
1862 * config/avr/avr.c (call_insn_operand): Delete function.
1863 * config/avr/avr.md (*pushqi, *pushhi, *pushsi, *pushsf): Use REG_SP
1864 instead of register number. Use predicates.
1865 * config/avr/predicates.md (const0_operand, reg_or_0_operand,
1866 call_insn_operand): Add.
1868 2007-01-06 Jan Hubicka <jh@suse.cz>
1870 * tree-pass.h (pass_build_cgraph_edges): Declare.
1871 * cgraphunit.c (record_refernece): Move to cgraphbuild.c
1872 (visited_nodes): Remove.
1873 (cgraph_create_edges): Move to cgraphbuild.c; rename to
1874 build_cgrpah_edges; make visited_nodes local.
1875 (cgraph_process_new_functions): DO not call initialize_inline_failed.
1876 (record_references_in_initializer): Move to cgraphbuild.c
1877 (initialize_inline_failed, rebuild_cgraph_edges,
1878 pass_rebuild_cgraph_edges): Move to cgraphbuild.c.
1879 (verify_cgraph_node): Make visited_nodes local.
1880 (cgraph_analyze_function): Do not call cgraph_create_edges and
1881 initialize_inline_failed.
1882 (cgraph_expand_function): Do not call cgraph_lower_function;
1883 assert that function is already lowered.
1884 * Makefile.in (cgraphbuild.o): New.
1885 * passes.c (init_optimization_passes): Add pass_build_cgraph_edges
1886 at the end of lowering passes.
1888 2007-01-06 Steven Bosscher <steven@gcc.gnu.org>
1890 * ifcvt.c (cond_move_convert_if_block): New function, code
1891 factored out from...
1892 (cond_move_process_if_block): ...here. Call the new function
1893 on the THEN and ELSE blocks.
1894 (merge_if_block): Do not copy global_live_at_end, merge_blocks
1895 already takes care of this.
1897 2007-01-05 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
1900 * tree.h (TREE_OVERFLOW_P): New.
1901 * c-typeck.c (parser_build_unary_op): Warn only if result
1902 overflowed and operands did not.
1903 (parser_build_binary_op): Likewise.
1904 (convert_for_assignment): Remove redundant overflow_warning.
1905 * c-common.c (overflow_warning): Don't check or set TREE_OVERFLOW.
1907 2007-01-05 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
1909 * c-typeck.c (store_init_value): Split over two lines to follow
1910 the GNU coding style.
1912 2007-01-05 Benjamin Kosnik <bkoz@redhat.com>
1914 * c-cppbuiltin.c (c_cpp_builtins): __GXX_EXPERIMENTAL_CPP0X__ to
1915 __GXX_EXPERIMENTAL_CXX0X__.
1916 * doc/cpp.texi: Same.
1918 2007-01-05 Richard Guenther <rguenther@suse.de>
1921 * tree.c (get_narrower): Do not construct COMPONENT_REFs
1922 with mismatched types. Instead explicitly build a
1923 conversion NOP_EXPR.
1925 2007-01-05 Ian Lance Taylor <iant@google.com>
1927 * c-common.c (decl_with_nonnull_addr_p): New function.
1928 (c_common_truthvalue_conversion): Call it.
1929 * c-typeck.c (build_binary_op): Likewise.
1930 * c-common.h (decl_with_nonnull_addr_p): Declare.
1932 2007-01-05 Jakub Jelinek <jakub@redhat.com>
1935 * libgcc2.c (__divdc3): Compare c and d against 0.0 instead of
1938 2007-01-05 Joel Brobecker <brobecker@adacore.com>
1940 * doc/install.texi (Final install): Document the fact that
1941 the GNAT runtime should not be stripped.
1943 2007-01-04 Jan Hubicka <jh@suse.cz>
1945 * tree-inline.c (fold_marked_statements): Update operand caches
1946 and EH after folding
1948 2007-01-04 Ian Lance Taylor <iant@google.com>
1950 * c-common.c (check_function_nonnull): Whitespace fix.
1952 2007-01-04 Jan Hubicka <jh@suse.cz>
1954 * tree-optimize.c (execute_fixup_cfg): Correct previously mistakely
1955 comitted older version of patch.
1956 (pass_fixup_cfg): Add TODOs to verify flow and statements, dump
1957 function, celanup cfg and collect garbage.
1959 2007-01-04 Mike Stump <mrs@apple.com>
1961 * Makefile.in (mostlyclean): Don't remove libgcc anymore.
1964 2007-01-04 Eric Christopher <echristo@apple.com>
1966 * libgcc2.c (__bswapsi2): Use SItype.
1967 (__bswapdi2): Use DItype.
1968 * libgcc2.h: Update for above.
1970 2007-01-04 Paul Brook <paul@codesourcery.com>
1972 * config/arm/arm.md (arm_mulsi3, thumb_mulsi3, mulsi3_compare0,
1973 mulsi_compare0_scratch, mulsi3addsi, mulsi3addsi_compare0,
1974 mulsi3addsi_compare0_scratch, mulsidi3adddi, mulsidi3,
1975 umulsidi3, umulsidi3adddi, smulsi3_highpart,
1976 umulsi3_highpart): Make conditional on !arm_arch6.
1977 (arm_mulsi3_v6, thumb_mulsi3_v6, mulsi3_compare0_v6,
1978 mulsi_compare0_scratch_v6, mulsi3addsi_v6, mulsi3addsi_compare0_v6,
1979 mulsi3addsi_compare0_scratch_v6, mulsidi3adddi_v6, mulsidi3_v6,
1980 umulsidi3_v6, umulsidi3adddi_v6, smulsi3_highpart_v6,
1981 umulsi3_highpart_v6): New insns.
1983 2007-01-04 Roger Sayle <roger@eyesopen.com>
1985 * fold-const.c (fold_convert): When casting an expression to void,
1986 fold_ignored_result may discover a GIMPLE_MODIFY_STMT which doesn't
1987 have a type. Instead of attempting to build a NOP_EXPR, return
1988 these "special" trees directly.
1990 2007-01-04 Joseph Myers <joseph@codesourcery.com>
1992 * config/rs6000/rs6000.c (rs6000_rtx_costs): Make adjustment for
1993 MULT inside MINUS as either argument. Use rs6000_cost->dmul -
1994 rs6000_cost->fp not 0 as adjustment for outer NEG.
1996 2007-01-04 Jan Hubicka <jh@suse.cz>
1998 * cgraph.c (cgraph_release_function_body): New function.
1999 (cgraph_remove_node): Use it.
2000 * cgraph.h (cgraph_release_function_body): Declare.
2001 * cgraphunit.c (cgraph_expand_function): Use it.
2002 * ipa.c (cgraph_remove_unreahchable_nodes): Use it.
2003 * tree-ssa.c (delete_tree_ssa): Allow to be called before aliasing
2004 is initialized and while compilation of other function is running.
2005 * tree-optimize.c (execute_free_cfg_annotations): Move code to clear
2006 statement CFG annotations from here to ...
2007 * tree-cfg.c (delete_tree_cfg_annotations): ... here.
2009 2007-01-04 Zdenek Dvorak <dvorakz@suse.cz>
2011 * cfgloop.h (enum li_flags): Make the constants powers of two.
2013 2007-01-04 Jan Hubicka <jh@suse.cz>
2015 * tree-inline.c (copy_bb): Insert new statements to statements_to_fold
2017 (fold_marked_statements): New function.
2018 (optimize_inline_calls, tree_function_versioning): Fold new statements.
2019 * tree-inline.h (copy_body_data): Add statements_to_fold.
2021 2007-01-03 Daniel Jacobowitz <dan@codesourcery.com>
2023 * config.gcc: Mention libgcc/config.host.
2024 * Makefile.in: Update comments mentioning libgcc.
2025 (LIBGCC, INSTALL_LIBGCC, GCC_PARTS, mklibgcc): Delete.
2026 (all.cross, start.encap, rest.encap, rest.cross): Update
2027 dependencies for libgcc move.
2028 (libgcc.mk, LIBGCC_DEPS, libgcov.a, libgcc.a, stmp-multilib)
2029 (clean-target, clean-target-libgcc): Delete.
2030 (srcdirify, GCC_EXTRA_PARTS): New macros.
2031 (libgcc-support, libgcc.mvars): New rules.
2032 (distclean): Remove mention of mklibgcc.
2033 (install): Don't reference INSTALL_LIBGCC.
2034 (install-common): Don't reference EXTRA_PARTS.
2035 (install-libgcc, install-multilib): Delete rules.
2036 * mklibgcc.in: Delete file.
2037 * doc/configfiles.texi: Don't mention mklibgcc.
2039 * config/i386/t-darwin (SHLIB_VERPFX): Delete (moved to libgcc).
2040 * config/i386/t-darwin64 (SHLIB_VERPFX): Likewise.
2041 * config/rs6000/t-darwin (SHLIB_VERPFX): Likewise.
2042 * config/rs6000/t-ppccomm (TARGET_LIBGCC2_CFLAGS, SHLIB_MAPFILES)
2043 (mklibgcc, ldblspecs): Likewise.
2045 * config/i386/t-nwld (libgcc.def, libc.def, libpcre.def)
2046 (posixpre.def): Use $(T).
2047 (SHLIB_EXT, SHLIB_NAME, SHLIB_SLIBDIR_QUAL, SHLIB_DEF, SHLIB_MAP)
2048 (SHLIB_SRC, SHLIB_INSTALL): Delete.
2049 (SHLIB_LINK): Make dummy.
2050 * config/t-slibgcc-darwin: Delete contents except for dummy SHLIB_LINK.
2052 * config/frv/t-linux (EXTRA_MULTILIB_PARTS): Clear.
2054 * config/alpha/t-crtfm: Use $(T) in rules for EXTRA_PARTS.
2055 * config/alpha/t-vms, config/alpha/t-vms64, config/fr30/t-fr30,
2056 config/i386/t-rtems-i386, config/ia64/t-ia64, config/rs6000/t-beos,
2057 config/rs6000/t-newas, config/sparc/t-elf: Likewise.
2059 * configure.ac (all_outputs): Remove mklibgcc.
2060 * configure: Regenerated.
2062 2007-01-03 Josh Conner <jconner@apple.com>
2065 * calls.c (compute_argument_addresses): Set stack and stack_slot
2066 for partial args, too.
2067 (store_one_arg): Use locate.size.constant for the size when
2068 generating a save_area.
2070 2007-01-03 Robert Kennedy <jimbob@google.com>
2072 * tree-cfg.c (tree_merge_blocks): Release SSA_NAME phi results
2073 whose definitions are deleted due to basic block merging.
2075 2007-01-03 Paul Brook <paul@codesourcery.com>
2078 * config/arm/arm.c (output_return_instruction): Pop PC in interrupt
2080 (use_return_insn): Return 0 for Thumb interrupt functions.
2081 (print_multi_reg): Add rfe argument for IRQ returns.
2082 (arm_output_epilogue): Pop interrupt return address directly into PC.
2083 (arm_expand_prologue): Only adjust IRQ return address in Arm mode.
2085 2007-01-03 Paul Brook <paul@codesourcery.com>
2087 Merge from sourcerygxx-4_1.
2088 * config/arm/thumb2.md: New file.
2089 * config/arm/elf.h (JUMP_TABLES_IN_TEXT_SECTION): Return True for
2091 * config/arm/coff.h (JUMP_TABLES_IN_TEXT_SECTION): Ditto.
2092 * config/arm/aout.h (ASM_OUTPUT_ADDR_VEC_ELT): Add !Thumb-2 assertion.
2093 (ASM_OUTPUT_ADDR_DIFF_ELT): Output Thumb-2 jump tables.
2094 * config/arm/aof.h (ASM_OUTPUT_ADDR_DIFF_ELT): Output Thumb-2 jump
2096 (ASM_OUTPUT_ADDR_VEC_ELT): Add !Thumb-2 assertion.
2097 * config/arm/ieee754-df.S: Use macros for Thumb-2/Unified asm
2099 * config/arm/ieee754-sf.S: Ditto.
2100 * config/arm/arm.c (thumb_base_register_rtx_p): Rename...
2101 (thumb1_base_register_rtx_p): ... to this.
2102 (thumb_index_register_rtx_p): Rename...
2103 (thumb1_index_register_rtx_p): ... to this.
2104 (thumb_output_function_prologue): Rename...
2105 (thumb1_output_function_prologue): ... to this.
2106 (thumb_legitimate_address_p): Rename...
2107 (thumb1_legitimate_address_p): ... to this.
2108 (thumb_rtx_costs): Rename...
2109 (thumb1_rtx_costs): ... to this.
2110 (thumb_compute_save_reg_mask): Rename...
2111 (thumb1_compute_save_reg_mask): ... to this.
2112 (thumb_final_prescan_insn): Rename...
2113 (thumb1_final_prescan_insn): ... to this.
2114 (thumb_expand_epilogue): Rename...
2115 (thumb1_expand_epilogue): ... to this.
2116 (arm_unwind_emit_stm): Rename...
2117 (arm_unwind_emit_sequence): ... to this.
2118 (thumb2_legitimate_index_p, thumb2_legitimate_address_p,
2119 thumb1_compute_save_reg_mask, arm_dwarf_handle_frame_unspec,
2120 thumb2_index_mul_operand, output_move_vfp, arm_shift_nmem,
2121 arm_save_coproc_regs, thumb_set_frame_pointer, arm_print_condition,
2122 thumb2_final_prescan_insn, thumb2_asm_output_opcode, arm_output_shift,
2123 thumb2_output_casesi): New functions.
2124 (TARGET_DWARF_HANDLE_FRAME_UNSPEC): Define.
2125 (FL_THUMB2, FL_NOTM, FL_DIV, FL_FOR_ARCH6T2, FL_FOR_ARCH7,
2126 FL_FOR_ARCH7A, FL_FOR_ARCH7R, FL_FOR_ARCH7M, ARM_LSL_NAME,
2127 THUMB2_WORK_REGS): Define.
2128 (arm_arch_notm, arm_arch_thumb2, arm_arch_hwdiv, arm_condexec_count,
2129 arm_condexec_mask, arm_condexec_masklen)): New variables.
2130 (all_architectures): Add armv6t2, armv7, armv7a, armv7r and armv7m.
2131 (arm_override_options): Check new CPU capabilities.
2132 Set new architecture flag variables.
2133 (arm_isr_value): Handle v7m interrupt functions.
2134 (user_return_insn): Return 0 for v7m interrupt functions. Handle
2136 (const_ok_for_arm): Handle Thumb-2 constants.
2137 (arm_gen_constant): Ditto. Use movw when available.
2138 (arm_function_ok_for_sibcall): Return false for v7m interrupt
2140 (legitimize_pic_address, arm_call_tls_get_addr): Handle Thumb-2.
2141 (thumb_find_work_register, arm_load_pic_register,
2142 legitimize_tls_address, arm_address_cost, load_multiple_sequence,
2143 emit_ldm_seq, emit_stm_seq, arm_select_cc_mode, get_jump_table_size,
2144 print_multi_reg, output_mov_long_double_fpa_from_arm,
2145 output_mov_long_double_arm_from_fpa, output_mov_double_fpa_from_arm,
2146 output_mov_double_fpa_from_arm, output_move_double,
2147 arm_compute_save_reg_mask, arm_compute_save_reg0_reg12_mask,
2148 output_return_instruction, arm_output_function_prologue,
2149 arm_output_epilogue, arm_get_frame_offsets, arm_regno_class,
2150 arm_output_mi_thunk, thumb_set_return_address): Ditto.
2151 (arm_expand_prologue): Handle Thumb-2. Use arm_save_coproc_regs.
2152 (arm_coproc_mem_operand): Allow POST_INC/PRE_DEC.
2153 (arithmetic_instr, shift_op): Use arm_shift_nmem.
2154 (arm_print_operand): Use arm_print_condition. Handle '(', ')', '.',
2156 (arm_final_prescan_insn): Use extract_constrain_insn_cached.
2157 (thumb_expand_prologue): Use thumb_set_frame_pointer.
2158 (arm_file_start): Output directive for unified syntax.
2159 (arm_unwind_emit_set): Handle stack alignment instruction.
2160 * config/arm/lib1funcs.asm: Remove default for __ARM_ARCH__.
2161 Add v6t2, v7, v7a, v7r and v7m.
2162 (RETLDM): Add Thumb-2 code.
2163 (do_it, shift1, do_push, do_pop, COND, THUMB_SYNTAX): New macros.
2164 * config/arm/arm.h (TARGET_CPU_CPP_BUILTINS): Define __thumb2__.
2165 (TARGET_THUMB1, TARGET_32BIT, TARGET_THUMB2, TARGET_DSP_MULTIPLY,
2166 TARGET_INT_SIMD, TARGET_UNIFIED_ASM, ARM_FT_STACKALIGN, IS_STACKALIGN,
2167 THUMB2_TRAMPOLINE_TEMPLATE, TRAMPOLINE_ADJUST_ADDRESS,
2168 ASM_OUTPUT_OPCODE, THUMB2_GO_IF_LEGITIMATE_ADDRESS,
2169 THUMB2_LEGITIMIZE_ADDRESS, CASE_VECTOR_PC_RELATIVE,
2170 CASE_VECTOR_SHORTEN_MODE, ADDR_VEC_ALIGN, ASM_OUTPUT_CASE_END,
2171 ADJUST_INSN_LENGTH): Define.
2172 (TARGET_REALLY_IWMMXT, TARGET_IWMMXT_ABI, CONDITIONAL_REGISTER_USAGE,
2173 STATIC_CHAIN_REGNUM, HARD_REGNO_NREGS, INDEX_REG_CLASS,
2174 BASE_REG_CLASS, MODE_BASE_REG_CLASS, SMALL_REGISTER_CLASSES,
2175 PREFERRED_RELOAD_CLASS, SECONDARY_OUTPUT_RELOAD_CLASS,
2176 SECONDARY_INPUT_RELOAD_CLASS, LIBCALL_VALUE, FUNCTION_VALUE_REGNO_P,
2177 TRAMPOLINE_SIZE, INITIALIZE_TRAMPOLINE, HAVE_PRE_INCREMENT,
2178 HAVE_POST_DECREMENT, HAVE_PRE_DECREMENT, HAVE_PRE_MODIFY_DISP,
2179 HAVE_POST_MODIFY_DISP, HAVE_PRE_MODIFY_REG, HAVE_POST_MODIFY_REG,
2180 REGNO_MODE_OK_FOR_BASE_P, LEGITIMATE_CONSTANT_P,
2181 REG_MODE_OK_FOR_BASE_P, REG_OK_FOR_INDEX_P, GO_IF_LEGITIMATE_ADDRESS,
2182 LEGITIMIZE_ADDRESS, THUMB2_LEGITIMIZE_ADDRESS,
2183 GO_IF_MODE_DEPENDENT_ADDRESS, MEMORY_MOVE_COST, BRANCH_COST,
2184 ASM_APP_OFF, ASM_OUTPUT_CASE_LABEL, ARM_DECLARE_FUNCTION_NAME,
2185 FINAL_PRESCAN_INSN, PRINT_OPERAND_PUNCT_VALID_P,
2186 PRINT_OPERAND_ADDRESS): Adjust for Thumb-2.
2187 (arm_arch_notm, arm_arch_thumb2, arm_arch_hwdiv): New declarations.
2188 * config/arm/arm-cores.def: Add arm1156t2-s, cortex-a8, cortex-r4 and
2190 * config/arm/arm-tune.md: Regenerate.
2191 * config/arm/arm-protos.h: Update prototypes.
2192 * config/arm/vfp.md: Enable patterns for Thumb-2.
2193 (arm_movsi_vfp): Add movw alternative. Use output_move_vfp.
2194 (arm_movdi_vfp, movsf_vfp, movdf_vfp): Use output_move_vfp.
2195 (thumb2_movsi_vfp, thumb2_movdi_vfp, thumb2_movsf_vfp,
2196 thumb2_movdf_vfp, thumb2_movsfcc_vfp, thumb2_movdfcc_vfp): New.
2197 * config/arm/libunwind.S: Add Thumb-2 code.
2198 * config/arm/constraints.md: Update include Thumb-2.
2199 * config/arm/ieee754-sf.S: Add Thumb-2/Unified asm support.
2200 * config/arm/ieee754-df.S: Ditto.
2201 * config/arm/bpabi.S: Ditto.
2202 * config/arm/t-arm (MD_INCLUDES): Add thumb2.md.
2203 * config/arm/predicates.md (low_register_operand,
2204 low_reg_or_int_operand, thumb_16bit_operator): New.
2205 (thumb_cmp_operand, thumb_cmpneg_operand): Rename...
2206 (thumb1_cmp_operand, thumb1_cmpneg_operand): ... to this.
2207 * config/arm/t-arm-elf: Add armv7 multilib.
2208 * config/arm/arm.md: Update patterns for Thumb-2 and Unified asm.
2210 (UNSPEC_STACK_ALIGN, ce_count): New.
2211 (arm_incscc, arm_decscc, arm_umaxsi3, arm_uminsi3,
2212 arm_zero_extendsidi2, arm_zero_extendqidi2): New
2214 * config/arm/fpa.md: Update patterns for Thumb-2 and Unified asm.
2215 (thumb2_movsf_fpa, thumb2_movdf_fpa, thumb2_movxf_fpa,
2216 thumb2_movsfcc_fpa, thumb2_movdfcc_fpa): New insns.
2217 * config/arm/cirrus.md: Update patterns for Thumb-2 and Unified asm.
2218 (cirrus_thumb2_movdi, cirrus_thumb2_movsi_insn,
2219 thumb2_cirrus_movsf_hard_insn, thumb2_cirrus_movdf_hard_insn): New
2221 * doc/extend.texi: Document ARMv7-M interrupt functions.
2222 * doc/invoke.texi: Document Thumb-2 new cores+architectures.
2224 2007-01-03 Jakub Jelinek <jakub@redhat.com>
2226 * unwind-dw2.c (SIGNAL_FRAME_BIT, EXTENDED_CONTEXT_BIT): Define.
2227 (struct _Unwind_Context): Rename args_size to flags, remove
2228 signal_frame field, add a new args_size field and version field.
2229 (_Unwind_IsSignalFrame, _Unwind_SetSignalFrame,
2230 _Unwind_IsExtendedContext): New inline functions.
2231 (_Unwind_GetGR, _Unwind_SetGR, _Unwind_GetGRPtr, _Unwind_SetGRPtr):
2232 Assume by_value array is only present if _Unwind_IsExtendedContext.
2233 (_Unwind_GetIPInfo, execute_cfa_program, uw_frame_state_for): Use
2234 _Unwind_IsSignalFrame.
2235 (__frame_state_for): Initialize context.flags to EXTENDED_CONTEXT_BIT.
2236 (uw_update_context_1): Use _Unwind_SetSignalFrame.
2237 (uw_init_context_1): Initialize context->flags to
2238 EXTENDED_CONTEXT_BIT.
2239 * config/rs6000/linux-unwind.h (frob_update_context): Use
2240 _Unwind_SetSignalFrame.
2242 2007-01-03 Andrew Pinski <andrew_pinski@playstation.sony.com>
2245 * gimplify.c (gimplify_modify_expr_complex_part): Move below
2246 tree_to_gimple_tuple. Call tree_to_gimple_tuple when we need
2249 2007-01-03 Kazu Hirata <kazu@codesourcery.com>
2251 * config/i386/i386.h (NON_STACK_REG_P, REGNO_OK_FOR_SIREG_P,
2252 REGNO_OK_FOR_DIREG_P, REWRITE_ADDRESS, ASM_OPERAND_LETTER,
2253 RET, AT_SP): Remove.
2254 * config/i386/i386.md (*sse_prologue_save_insn): Use return
2257 * alias.c (init_alias_analysis): Use VEC_safe_grow_cleared.
2258 * cfgbuild.c (find_basic_blocks): Likewise.
2259 * cfgrtl.c (rtl_create_basic_block): Likewise.
2260 * function.c (temp_slots_at_level): Likewise.
2261 * reg-stack.c (stack_regs_mentioned): Likewise.
2262 * regclass.c (allocate_reg_info): Likewise.
2263 * tree-cfg.c (init_empty_tree_cfg, build_tree_cfg, create_bb,
2264 set_bb_for_stmt, move_block_to_fn): Likewise.
2265 * tree-complex.c (tree_lower_complex): Likewise.
2266 * vec.h (VEC_safe_grow_cleared): New.
2268 * cgraphunit.c, tree-ssa-alias.c: Fix comment typos.
2270 2007-01-03 Zdenek Dvorak <dvorakz@suse.cz>
2272 * loop-unswitch.c (unswitch_loop): Pass probabilities to loopify.
2273 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Pass probabilities
2275 * cfgloopmanip.c (scale_loop_frequencies): Export.
2276 (loopify): Scale the frequencies by prescribed coefficients.
2277 (set_zero_probability): New function.
2278 (duplicate_loop_to_header_edge): Improve updating of frequencies.
2279 (lv_adjust_loop_entry_edge, loop_version): Set probabilities
2280 and frequencies according to arguments.
2281 * tree-ssa-loop-manip.c (tree_unroll_loop): Set probabilities
2283 * cfg.c (scale_bbs_frequencies_int): Allow scaling the frequencies up.
2284 * modulo-sched.c (sms_schedule): Set probabilities for entering
2285 versioned loop correctly.
2286 * tree-vect-transform.c (vect_transform_loop): Ditto.
2287 * cfgloop.h (loopify, loop_version): Declaration changed.
2288 (scale_loop_frequencies): Declared.
2290 2007-01-02 Jan Hubicka <jh@suse.cz>
2292 * cgraph.c: Include tree-flow.h
2293 (cgraph_add_new-function): Handle IPA_SSA mode; execute
2295 * cgraph.h (enum cgraph_state): Add CGRAPH_STATE_IPA_SSA.
2296 * tree-pass.h (pass_all_early_optimizations): Declare.
2297 * cgraphunit.c (cgraph_process_new_functions): Add IPA_SSA; execute
2299 (cgraph_analyze_function): Do early_local_passes.
2300 * tree-mudflap.c (mf_decl_cache_locals, mf_build_check_statement_for):
2301 Do not add referenced vars.
2302 * tree-optimize.c (gate_all_optimizations): Do not execute when not in
2304 (gate_all_early_local_passes): New gate.
2305 (pass_early_local_passes): Use new gate.
2306 (execute_early_local_optimizations): New functions.
2307 (gate_all_early_optimizations): New gate.
2308 (pass_all_early_optimizations): New pass.
2309 (execute_free_datastructures): Free SSA only when initialized.
2310 (gate_init_datastructures): Init only when optimizing.
2311 (tree_lowering_passes): Do early local passes when called late.
2312 * tree-profile.c (do_tree_profiling): Don't profile functions added
2314 (do_early_tree_profiling, pass_early_tree_profile): Kill.
2315 * tree-cfg.c (update_modified_stmts): Do not update when operands are
2317 * passes.c (init_optimizations_passes): Reorder so we go into SSA
2318 during early_local_passes.
2319 * Makefile.in (cgraph.o): Add dependency on tree-flow.h.
2322 2007-01-02 Carlos O'Donell <carlos@codesourcery.com>
2324 * Makefile.in: Update copyright year.
2326 2007-01-02 Carlos O'Donell <carlos@codesourcery.com>
2328 * Makefile.in: Export GCC_EXEC_PREFIX before calling $(RUNTEST)
2329 in $(lang_checks) and check-consistency targets.
2331 2007-01-02 Jan Hubicka <jh@suse.cz>
2333 * tree-mudflap.c (mf_decl_cache_locals, mf_build_check_statement_for):
2334 Do not add referenced vars.
2335 * tree-cfg.c (update_modified_stmts): Do not update when SSA operands
2337 * passes.c (init_optimization_passes): Put mudflap_2 after
2338 free_datastructures.
2340 2007-01-02 Jan Hubicka <jh@suse.cz>
2342 * tree-optimize (execute_fixup_cfg): Set after_inlining flag.
2343 Set NOTHROW flag on call statements proved to be nothrow.
2344 Update statement of local calls so new pure/const functions are
2345 updated. Update_ssa when in ssa form. Mark PHI nodes of nonlocal
2347 (tree_rest_of_compilation): Register hooks and initialize bitmap
2348 early. Do not set after_inlining flag.
2350 2007-01-02 Steve Ellcey <sje@cup.hp.com>
2352 * sbitmap.c (HOST_BITS_PER_LONG_LONG): Change to
2353 HOST_BITS_PER_LONGLONG
2355 2007-01-02 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
2358 * c-typeck.c (store_init_value): Don't emit pedantic overflow
2359 warning for non-static initializers.
2361 2007-01-02 Steven Bosscher <steven@gcc.gnu.org>
2363 * config/alpha/alpha.md, arm/arm.c, darwin.c, frv/frv.md,
2364 m32r/m32r.c, m32r/m32r.c, mn10300/mn10300.md, pa/pa.c,
2365 rs6000/rs6000.c, s390/s390.md, sh/sh.md, sparc/sparc.c:
2366 Always use set_unique_reg_note to add REG_EQUAL notes.
2368 2007-01-02 Kazu Hirata <kazu@codesourcery.com>
2371 2007-01-02 Kazu Hirata <kazu@codesourcery.com>
2373 * alias.c (init_alias_analysis): Use VEC_safe_grow_cleared.
2374 * cfgbuild.c (find_basic_blocks): Likewise.
2375 * cfgrtl.c (rtl_create_basic_block): Likewise.
2376 * function.c (temp_slots_at_level): Likewise.
2377 * reg-stack.c (stack_regs_mentioned): Likewise.
2378 * regclass.c (allocate_reg_info): Likewise.
2379 * tree-cfg.c (init_empty_tree_cfg, build_tree_cfg, create_bb,
2380 set_bb_for_stmt, move_block_to_fn): Likewise.
2381 * tree-complex.c (tree_lower_complex): Likewise.
2382 * vec.h (VEC_safe_grow_cleared): New.
2384 2007-01-02 Ian Lance Taylor <iant@google.com>
2386 * c-common.c (c_common_truthvalue_conversion): When warning about
2387 using an assignment as a truth value, set TREE_NO_WARNING.
2389 2007-01-02 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
2392 * c.opt (Wold-style-declaration): New.
2393 * doc/invoke.texi (C-only Warning Options): New.
2394 (Wold-style-declaration): Document it.
2395 (Wextra): Enabled by -Wextra.
2396 * c-opts.c (c_common_post_options): Enabled by -Wextra.
2397 * c-decl.c (declspecs_add_scspec): Replace -Wextra with
2398 -Wold-style-declaration.
2400 2007-01-02 Kazu Hirata <kazu@codesourcery.com>
2402 * alias.c (init_alias_analysis): Use VEC_safe_grow_cleared.
2403 * cfgbuild.c (find_basic_blocks): Likewise.
2404 * cfgrtl.c (rtl_create_basic_block): Likewise.
2405 * function.c (temp_slots_at_level): Likewise.
2406 * reg-stack.c (stack_regs_mentioned): Likewise.
2407 * regclass.c (allocate_reg_info): Likewise.
2408 * tree-cfg.c (init_empty_tree_cfg, build_tree_cfg, create_bb,
2409 set_bb_for_stmt, move_block_to_fn): Likewise.
2410 * tree-complex.c (tree_lower_complex): Likewise.
2411 * vec.h (VEC_safe_grow_cleared): New.
2413 2007-01-02 Douglas Gregor <doug.gregor@gmail.com>
2415 * c-common.c(c_common_nodes_and_builtins): Since variants of
2416 void_type_node get built before it is given a name, we need to
2417 give those variants the name, too.
2418 (complete_array_type): We need to work with the canonical main
2419 type of the array, from which we will build the qualified version.
2420 * params.def (PARAM_VERIFY_CANONICAL_TYPES): New.
2421 * print-tree.c (print_node): Display canonical type information
2423 * stor-layout.c (layout_type): When we don't know the
2424 alignment of a type for which we're building an array, we end up
2425 guessing wrong, so make the type require structural equality.
2426 * tree.c (make_node_stat): When we build a new type, it is its
2428 (build_type_attribute_qual_variant): When building an attribute
2429 variant, its canonical type is the non-attribute variant. However,
2430 if the attributes are target-dependent and they differ, we need to
2431 use structural equality checks for this type.
2432 (build_qualified_type): A qualified type is not equivalent to its
2433 unqualified variant; set the canonical type appropriately.
2434 (build_distinct_type_copy): When building a distinct type from
2435 another type, the new type is its own canonical type.
2436 (build_variant_type_copy): When building a new type variant, we
2437 assume that it is equivalent to the original type.
2438 (build_pointer_type_for_mode): When building a pointer type, also
2439 build a canonical type pointer.
2440 (build_reference_type_for_mode): When building a reference type,
2441 also build a canonical type reference.
2442 (build_index_type): When we can't hash an index type (e.g.,
2443 because its maximum value is negative), the index type requires
2444 structural equality tests.
2445 (build_array_type): Build the canonical form of an array type.
2446 (build_function_type): Function types require structural equality,
2447 because they contain default arguments, attributes, etc.
2448 (build_method_type_directly): Ditto for method types.
2449 (build_offset_type): Build the canonical offset type.
2450 (build_complex_type): Build the canonical vector type.
2451 (make_vector_type): Build the canonical vector type.
2452 * tree.h (TYPE_CANONICAL): New.
2453 (TYPE_STRUCTURAL_EQUALITY_P): New.
2454 (SET_TYPE_STRUCTURAL_EQUALITY): New.
2455 (struct tree_type): Added "canonical" field.
2456 * params.h (VERIFY_CANONICAL_TYPES): New.
2457 * doc/c-tree.texi (TYPE_CANONICAL): Document.
2458 (TYPE_STRUCTURAL_EQUALITY_P): Document.
2459 (SET_TYPE_STRUCTURAL_EQUALITY): Document.
2460 * doc/invoke.texi (verify-canonical-types): Document --param
2461 parameter for verifying canonical types.
2463 2007-01-02 Joseph Myers <joseph@codesourcery.com>
2465 * config.gcc (powerpc-*-eabispe*, powerpc-*-eabisimaltivec*,
2466 powerpc-*-eabisim*, powerpc-*-eabialtivec*, powerpc-*-eabi*,
2467 powerpc-*-rtems*, powerpc-wrs-vxworks, powerpc-wrs-vxworksae,
2468 powerpcle-*-eabisim*, powerpcle-*-eabi*): Add rs6000/e500.h to
2470 * config/rs6000/e500.h: New.
2471 * config/rs6000/eabi.h (TARGET_SPE_ABI, TARGET_SPE, TARGET_E500,
2472 TARGET_ISEL, TARGET_FPRS, TARGET_E500_SINGLE, TARGET_E500_DOUBLE):
2474 * config/rs6000/linuxspe.h (TARGET_SPE_ABI, TARGET_SPE,
2475 TARGET_E500, TARGET_ISEL, TARGET_FPRS, TARGET_E500_SINGLE,
2476 TARGET_E500_DOUBLE): Remove.
2477 * config/rs6000/vxworks.h (TARGET_SPE_ABI, TARGET_SPE,
2478 TARGET_E500, TARGET_ISEL, TARGET_FPRS): Remove.
2479 * config/rs6000/rs6000.h (CHECK_E500_OPTIONS): Define.
2480 * config/rs6000/rs6000.c (rs6000_override_options): Use
2483 2007-01-02 Joseph Myers <joseph@codesourcery.com>
2485 * config/rs6000/rs6000.c (print_operand): Check (TARGET_SPE ||
2486 TARGET_E500_DOUBLE), not TARGET_E500, for %y.
2487 (rs6000_generate_compare, rs6000_emit_sCOND, output_cbranch,
2488 rs6000_emit_cmove): Don't check TARGET_E500.
2489 * config/rs6000/rs6000.md (bunordered, bordered, sunordered,
2490 sordered): Don't check TARGET_E500.
2492 2007-01-01 Eric Christopher <echristo@apple.com>
2494 * config/mips/mips.c (mips_regno_mode_ok_for_base_p): Use
2495 HARD_REGISTER_NUM_P.
2497 2007-01-01 Roger Sayle <roger@eyesopen.com>
2499 * fold-const.c (fold_binary) <EQ_EXPR>: Fold "(X^C1) eq/ne C2" into
2500 "X eq/ne (C1^C2)". Fold "(X^Z) eq/ne (Y^Z)" as "X eq/ne Y" when Z
2501 has no side-effects. Fold "(X^C1) eq/ne (Y^C2)" as "(X^(C1^C2))
2504 2007-01-01 Mike Stump <mrs@apple.com>
2506 * configure.ac: Remove support for building with Apple's gcc-3.1.
2508 2007-01-02 Joseph Myers <joseph@codesourcery.com>
2511 * caller-save.c (add_stored_regs): Only handle SUBREGs if inner
2512 REG is a hard register. Do not modify REG before calling
2514 * rtlanal.c (subreg_get_info): Don't assert size of XMODE is a
2515 multiple of the size of YMODE for certain lowpart cases.
2517 2007-01-01 Andrew Pinski <pinskia@gmail.com>
2520 * gimplify (voidify_wrapper_expr): Update for
2523 2007-01-01 Andreas Schwab <schwab@suse.de>
2526 * config/ia64/ia64.c (ia64_compute_frame_size): Account space for
2527 save of BR0 in extra_spill_size instead of spill_size.
2528 (ia64_expand_prologue): Save BR0 outside of the gr/br/fr spill
2530 (ia64_expand_epilogue): Restore BR0 from its new location.
2532 2007-01-01 Andrew Pinski <pinskia@gmail.com>
2534 * gimplify.c (gimplify_init_constructor <case VECTOR_TYPE>):
2535 Use a temporary variable if the left hand side is not a gimple
2538 2007-01-01 Andrew Pinski <pinskia@gmail.com>
2540 * gimplify.c (gimplify_return_expr): Make the temporary variable
2541 for the return expression, a gimple register variable.
2543 2007-01-01 Jan Hubicka <jh@suse.cz>
2545 * emit-rtl.c (emit_copy_of_insn_after): Do not call copy_insn_1 for
2548 2007-01-01 Mike Stump <mrs@apple.com>
2550 * configure.ac (HAVE_GAS_LITERAL16): Add autoconf check for
2552 * config/darwin.c (machopic_select_rtx_section): Use
2554 (darwin_mergeable_constant_section): Likewise.
2555 * configure: Regenerate.
2556 * config.in: Regenerate.
2558 2007-01-01 Jan Hubicka <jh@suse.cz>
2559 Andrew Pinski <pinskia@gmail.com>
2561 * cgraphunit.c (cgraph_optimize): Call cgraph_add_new_functions
2562 before starting IPA passes.