| blob | b866ea0b49b1524c92175b3a5d3cbf197565c62a |
1 2007-01-22 Jan Hubicka <jh@suse.cz>
3 * passes.c (init_optimization_passes): Do not rerun
4 pass_early_warn_uninitialized.
6 2007-01-22 Richard Guenther <rguenther@suse.de>
8 PR tree-optimization/30038
9 * tree-ssa-math-opts.c (maybe_record_sincos): New static helper
10 function.
11 (execute_cse_sincos_1): Likewise.
12 (execute_cse_sincos): Likewise.
13 (gate_cse_sincos): Likewise.
14 (pass_cse_sincos): New pass CSEing sin() and cos() calls using
15 the cexpi() canonicalization of sincos().
16 * tree-pass.h (pass_cse_sincos): Declare.
17 * passes.c (init_optimization_passes): New pass pas_cse_sincos.
19 2007-01-21 Eric Botcazou <ebotcazou@libertysurf.fr>
21 PR rtl-optimization/29329
22 * combine.c (replaced_rhs_insn): Rename to i2mod.
23 (replaced_rhs_value): Rename to i2mod_new_rhs.
24 (i2mod_old_rhs): New global variable.
25 (combine_instructions): Adjust for above change. Save a copy of
26 the old RHS into i2mod_old_rhs when the contents of a REG_EQUAL
27 note are substituted in the second instruction.
28 (distribute_notes) <REG_DEAD>: Adjust for above change. Do not
29 ditch the note if it pertains to the second eliminated register
30 and this register is mentioned in i2mod_old_rhs.
32 Revert:
33 2006-09-12 Eric Botcazou <ebotcazou@libertysurf.fr>
35 * combine.c (distribute_notes) <REG_DEAD>: Do not consider SETs past
36 the insn to which the note was originally attached.
38 2007-01-21 Jan Hubicka <jh@suse.cz>
40 * ipa-inline.c (inlining_mode): Comment, move up.
41 (cgraph_decide_inlining_incrementally): Do not perform inlining itself;
42 fix handling of flattening of self recursive functions.
43 (cgraph_find_cycles): Remove.
44 (cgraph_flatten_node): Remove.
45 (cgraph_decide_inlining): Use incremental inliner to handle flattening.
46 (try_inline): New function.
47 (cgraph_early_inlining): Update call of
48 cgraph_decide_inlining_incrementally. Apply inlining here.
49 (apply_inline): Update call of cgraph_decide_inlining_incrementally.
51 2007-01-21 Dirk Mueller <dmueller@suse.de>
53 PR bootstrap/30511
54 * tree-vrp.c (check_array_bounds): do not warn
55 about ADDR_EXPR's of ARRAY_REF's which are immediately
56 used in binary expressions.
58 2007-01-21 Ira Rosen <irar@il.ibm.com>
60 * tree-vectorizer.h (struct _stmt_vec_info): Add new field
61 read_write_dep and macros for its access.
62 * tree-vectorizer.c (new_stmt_vec_info): Initialize the new field.
63 * tree-vect-analyze.c (vect_analyze_data_ref_dependence): Remove
64 argument, call vect_check_interleaving for every independent pair of
65 data-refs. Mark loads that access the same memory location as a store
66 in the loop.
67 (vect_check_dependences): Remove.
68 (vect_analyze_data_ref_dependences): Remove vect_check_dependences
69 call, fix the call to vect_analyze_data_ref_dependence.
70 (vect_analyze_data_ref_access): For statements that access the same
71 data-ref, check that they are not stores; for loads, check that there
72 is no store that access the same location.
74 2007-01-20 Roger Sayle <roger@eyesopen.com>
75 Joseph Myers <joseph@codesourcery.com>
77 * doc/invoke.texi (-fdump-rtl-combine): Fix under/overfull hbox.
78 (-fdump-rtl-stack): Likewise.
79 (-fno-signed-zeros): Use @minus{} for a minus sign. Correct typo.
80 (-mcheck-zero-division, -mcpu): Fix under/overfull hbox.
81 (-mpt-fixed): Use @minus{} for minus sign.
82 (Using Precompiled Headers): Fix under/overfull hbox.
84 2007-01-20 Jan Hubicka <jh@suse.cz>
86 * tree-flow.h (struct stmt_ann_d): Move references_memory to proper
87 place within annotation.
89 2007-01-20 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
91 * pa.c (output_move_double): Change array size of xoperands to 4.
93 2007-01-20 Jan Hubicka <jh@suse.cz>
95 * tree-tailcall.c (adjust_return_value): Do not use RESULT_DECL
96 as temporary.
98 2007-01-19 Ian Lance Taylor <iant@google.com>
100 * expmed.c (expand_divmod) [TRUNC_MOD_EXPR, TRUNC_DIV_EXPR]: Cast
101 constant to unsigned HOST_WIDE_INT before negating.
103 2007-01-19 Ian Lance Taylor <iant@google.com>
105 * tree-ssa-operands.h (struct vuse_vec_d): Change num_vuse field
106 to unsigned.
107 (VUSE_VECT_ELEMENT) [ENABLE_CHECKING]: Use unsigned comparison.
108 (VUSE_ELEMENT_PTR) [ENABLE_CHECKING]: Likewise.
109 (SET_VUSE_VECT_ELEMENT) [ENABLE_CHECKING]: Likewise.
110 (SET_VUSE_ELEMENT_VAR) [ENABLE_CHECKING]: Likewise.
111 (SET_VUSE_ELEMENT_PTR) [ENABLE_CHECKING]: Likewise.
112 (realloc_vdef, realloc_vuse): Change second parameter to
113 unsigned.
114 (ssa_operand_iterator_d): Change vuse_index and mayuse_index
115 fields to unsigned.
116 * tree-ssa-operands.c (realloc_vop): Change num_elem parameter to
117 unsigned. Change x and lim locals to unsigned.
118 (realloc_vdef, realloc_vuse): Change num_elem parameter to
119 unsigned.
120 (finalize_ssa_vuse_ops): Change old_i local to unsigned.
121 (copy_virtual_operands): Change i and n locals to unsigned.
123 2007-01-19 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
125 PR middle-end/29335
126 * builtins.c (fold_builtin_1): Handle builtin fdim.
128 2007-01-20 Jan Hubicka <jh@suse.cz>
130 * tree-ssa.c (init_tree_ssa): Do not call init_alias_heapvars.
131 * tree-ssa-structalias.c (compute_points_to_sets): Do call
132 init_alias_heapvars.
133 (init_alias_heapvars): Initialize only when not already initialized.
134 (delete_alias_heapvars): Set heapvar_for_stmt to NULL.
136 2007-01-19 Roger Sayle <roger@eyesopen.com>
138 * common.opt (fsigned-zeros): New command line option.
139 * flags.h (HONOR_SIGNED_ZEROS): Control via flag_signed_zeros instead
140 of flag_unsafe_math_optimizations.
141 * opts.c (set_fast_math_flags): The -ffast-math command line option
142 implies -fno-signed-zeros.
143 (fast_math_flags_set_p): Likewise.
145 * doc/invoke.texi: Document new -fno-signed-zeros option, and update
146 the documentation of -ffast-math appropriately. Wrap long lines.
148 2007-01-19 Steve Ellcey <sje@cup.hp.com>
150 * system.h (ASM_MAKE_LABEL_LINKONCE): Poison.
151 * varasm.c (globalize_decl): Remove ASM_MAKE_LABEL_LINKONCE ifdef.
153 2007-01-19 Tomas Bily <tbily@suse.cz>
155 * cgraphunit.c (cgraph_finalize_function): Updating of pid
156 * tree-profile.c:
157 (tree_init_ic_make_global_vars): New function
158 (tree_init_edge_profiler): call of tree_init_ic_make_global_vars
159 (tree_gen_ic_profiler): New function
160 (tree_gen_ic_func_profiler): New function
161 (tree_profiling): Added calling of tree_gen_ic_func_profiler
162 (tree_profile_hooks): Added hook for indirec/virtual calls
163 * value-prof.c (tree_find_values_to_profile): New case for
164 indirect calls
165 (tree_values_to_profile): Call for determining indirect/virtual
166 counters
167 (tree_indirect_call_to_profile): New function
168 (tree_ic_transform): New function
169 (tree_ic): New function
170 (find_func_by_pid): New function
171 (init_pid_map): New function
172 (tree_value_profile_transformations): Added check for
173 indirect/virtual call transformation
174 * value-prof.h (enum hist_type): New counter type for
175 indirect/virtual calls
176 (profile_hooks): Added new hook for profiling indirect/virtual
177 calls
178 * profile.c (instrument_values): New case for indirect/virtual
179 call added
180 * gcov-io.h (GCOV_LAST_VALUE_COUNTER): Changed to 6
181 (GCOV_COUNTER_V_INDIR): New counter type
182 (GCOV_COUNTER_NAMES): New name of counter "indirect" added
183 (GCOV_MERGE_FUNCTIONS): New merge function for indirect/virtual
184 call added
185 * cgraph.c: Definition of cgraph_max_pid
186 (cgraph_create_node): Default init of pid attribute
187 * cgraph.h: Declaration of cgraph_max_pid
188 (struct cgraph_node): Added pid attribute
189 * libgcov.c (__gcov_indirect_call_profiler): New function
190 (__gcov_one_value_profiler_body): New function
191 (__gcov_one_value_profiler): Body was moved to
192 __gcov_one_value_profiler_body and calls it
194 2007-01-19 Basile Starynkevitch <basile@starynkevitch.net>
196 * doc/gty.texi (Options): Document the mark_hook option to GTY.
197 * gengtype.c (write_types_data, write_func_for_structure,
198 write_types, ggc_wtd, pch_wtd): Add skip_hooks to
199 write_types_data, ggc_wtd, pch_wtd for processing mark_hook.
200 (walk_type, write_func_for_structure): Generate the mark_hook if
201 needed.
203 2007-01-19 Jan Hubicka <jh@suse.cz>
205 * ipa-inline.c (cgraph_decide_inlining_incrementally): Instead of
206 'early' argument take inlining mode argument specifying whether to
207 inline for size/speeed or all functions; add support for flattening;
208 improve dumpting.
209 (cgraph_early_inlining): Update call of decide_inlining_incrementally.
211 2007-01-19 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
213 PR c++/17947
214 * toplev.c (warn_deprecated_use): Use %qD instead of %qs to print
215 the name of the declared identifier.
217 2007-01-19 Dirk Mueller <dmueller@suse.de>
219 * config/i386.h (CONDITIONAL_REGISTER_USAGE): Store
220 result of PIC_OFFSET_TABLE_REGNUM in temporary variable to avoid
221 duplicate evaluation.
223 2007-01-19 Uros Bizjak <ubizjak@gmail.com>
225 * config/i386/i386.md (acos<mode>2): Rename from acossf2 and acosdf2.
226 Macroize expander using X87MODEF12 mode macro. Extend operand 1
227 to XFMode, use acosxf2 and truncate result to requested mode.
228 Use SSE_FLOAT_MODE_P to disable patterns for SSE math.
229 (asin<mode>2): Similarly, with asin expanders.
230 (*fscalexf4_i387): Rename from *fscalexf4.
231 (expNcorexf3): New expander.
232 (expxf2, exp10xf2, exp2xf2): Use expNcorexf3 expander.
233 (exp<mode>2): Rename from expsf2 and expdf2. Macroize expander using
234 X87MODEF12 mode macro. Extend operand 1 to XFMode, use expxf2 and
235 truncate result to requested mode. Use SSE_FLOAT_MODE_P to disable
236 patterns for SSE math.
237 (exp10<mode>2): Similarly, with exp10 expanders.
238 (exp2<mode>2): Similarly, with exp2 expanders.
239 (expm1<mode>2): Similarly, with expm1 expanders.
240 (ldexp<mode>3): Similarly, with ldexp expanders.
241 (log<mode>2, log10<mode>2, log2<mode>2, log1p<mode>2, logb<mode>2):
242 Use gen_truncxf<mode>2_i387_noop to truncate result.
244 2007-01-19 Richard Sandiford <richard@codesourcery.com>
246 * config/m68k/m68k.h (PREFERRED_STACK_BOUNDARY): Define to 32
247 for ColdFire targets.
249 2007-01-19 Nathan Sidwell <nathan@codesourcery.com>
250 Richard Sandiford <richard@codesourcery.com>
252 * config/m68k/m68k.h (M68K_STATIC_CHAIN_REG_NAME): New macro.
253 (INITIALIZE_TRAMPOLINE): Use STATIC_CHAIN_REGNUM.
254 (__transfer_from_trampoline): Use M68K_STATIC_CHAIN_REG_NAME.
255 * config/m68k/m68kelf.h (STATIC_CHAIN_REGNUM): Override.
256 (M68K_STATIC_CHAIN_REG_NAME): Likewise.
257 * config/m68k/netbsd-elf.h (M68K_STATIC_CHAIN_REG_NAME): Likewise.
259 2007-01-19 Richard Sandiford <richard@codesourcery.com>
261 * config/m68k/m68k.md (adddi_dilshr32): Rename to...
262 (*adddi_dilshr32): ...this. Fix formatting. Remove commented-out
263 non-canonical pattern. Restrict to !TARGET_COLDFIRE.
264 (*adddi_dilshr32_cf): New pattern.
265 (adddi3, subdi3): Remove first alternatives.
267 2007-01-19 Richard Sandiford <richard@codesourcery.com>
269 * config/m68k/m68k.c (notice_update_cc): If an SFmode move is
270 implemented using move.l, do not use its cc result for floating-point
271 comparisons.
273 2007-01-19 Richard Sandiford <richard@codesourcery.com>
275 * config/m68k/m68k.h (EXTRA_CONSTRAINT): Stop the 'T' constraint
276 from accepting 's' constraints if flag_pic.
278 2007-01-19 Richard Sandiford <richard@codesourcery.com>
280 * config/m68k/m68k.md (bordered, bunordered, buneq, bunge, bungt)
281 (bunle, bunlt, bltgt, bordered_rev, bunordered_rev, buneq_rev)
282 (bunge_rev, bungt_rev, bunle_rev, bunlt_rev, bltgt_rev): Change
283 condition from TARGET_68881 to TARGET_HARD_FLOAT.
285 2007-01-19 Sandra Loosemore <sandra@codesourcery.com>
287 * longlong.h (count_leading_zeros, COUNT_LEADING_ZEROS_0): Add
288 ColdFire alternatives.
289 * config/m68k/m68k.h (CLZ_DEFINED_VALUE_AT_ZERO): New macro.
290 * config/m68k/m68k.md (clzsi2): Define for ColdFire
291 architectures that support ff1 instruction.
293 2007-01-19 Richard Sandiford <richard@codesourcery.com>
294 Julian Brown <julian@codesourcery.com>
296 * config/m68k/m68k.h (CONST_OK_FOR_LETTER_P): Add an 'R' case.
297 * config/m68k/m68k.md (*movsi_cfv4): Fold into...
298 (*movsi_cf): ...here. Remove unnecessary 'R' from 'Rg'.
299 Add commentary.
301 2007-01-19 Richard Sandiford <richard@codesourcery.com>
303 * config/m68k/m68k-protos.h (valid_mov3q_const): Take a HOST_WIDE_INT
304 and return a bool.
305 (output_move_const_into_data_reg, output_move_simode_const): Delete.
306 * config/m68k/m68k.c (const_method, const_int_cost): Take a
307 HOST_WIDE_INT instead of an rtx.
308 (m68k_rtx_costs): Update call accordingly.
309 (output_move_const_into_data_reg): Likewise. Fix formatting.
310 (valid_mov3q_const): Take a HOST_WIDE_INT instead of an rtx.
311 Return a bool.
312 (output_move_simode_const): Update calls after above changes.
313 Rework to use automatic variables and predicates like MEM_P.
314 * config/m68k/m68k.md (pushexthisi_const): Update call to
315 valid_mov3q_const.
317 2007-01-19 Dirk Mueller <dmueller@suse.de>
319 * tree-ssa-alias.c (perform_var_substitution): Fix typo
320 in dump_flags test.
322 2007-01-19 Richard Guenther <rguenther@suse.de>
324 * builtins.c (expand_builtin_cexpi): Fall back to expanding
325 via cexp in case sincos is not available.
327 2007-01-19 Richard Guenther <rguenther@suse.de>
329 * doc/tm.texi (TARGET_HAS_SINCOS): Document new target macro.
330 * defaults.h (TARGET_HAS_SINCOS): Default to off.
331 * config/linux.h (TARGET_HAS_SINCOS): Set to on if we have glibc.
332 * config/alpha/linux.h (TARGET_HAS_SINCOS): Likewise.
333 * config/sparc/linux.h (TARGET_HAS_SINCOS): Likewise.
334 * config/sparc/linux64.h (TARGET_HAS_SINCOS): Likewise.
335 * config/rs6000/linux.h (TARGET_HAS_SINCOS): Likewise.
336 * config/rs6000/linux64.h (TARGET_HAS_SINCOS): Likewise.
338 2007-01-19 Uros Bizjak <ubizjak@gmail.com>
340 * config/i386/i386.md (*fpatanxf3_i387, fpatan_extend<mode>xf3_i387):
341 New insn patterns.
342 (atan2sf3_1, atan2df3_1, atan2xf3_1): Remove insn patterns.
343 (atan2xf3): Directly generate RTL pattern.
344 (atan2<mode>3): Rename from atan2sf3 and atan2df3 and macroize insn
345 patterns using X87MODEF12 mode macro. Use fpatan_extend<mode>xf3_i387
346 and truncate result to requested mode. Use SSE_FLOAT_MODE_P to
347 disable patterns for SSE math.
348 (atan<mode>2): Rename from atansf2 and atandf2 and macroize insn
349 patterns using X87MODEF12 mode macro. Use fpatan_extend<mode>xf3_i387
350 and truncate result to requested mode. Use SSE_FLOAT_MODE_P to
351 disable patterns for SSE math.
353 2007-01-19 Alexandre Oliva <aoliva@redhat.com>
355 * libgcc-std.ver: Fix typo in %inherit for GCC_4.3.0.
357 2007-01-18 Roger Sayle <roger@eyesopen.com>
359 * fold-const.c (fold_unary) <VIEW_CONVERT_EXPR>: Optimize away a
360 VIEW_CONVERT_EXPR to the same type as it's operand.
362 2007-01-18 David Edelsohn <edelsohn@gnu.org>
364 * config/rs6000/darwin-ldouble.c: Only build _SOFT_FLOAT if
365 configured for long double 128.
367 2007-01-18 Mike Stump <mrs@apple.com>
369 * config/rs6000/rs6000.c (rs6000_emit_vector_compare): Fix build
370 error.
372 2007-01-18 Michael Meissner <michael.meissner@amd.com>
374 * i386.c (ix86_compute_frame_layout): Make fprintf's in #if 0 code
375 type correct.
377 2007-01-18 Jan Hubicka <jh@suse.cz>
379 * tree-ssa-operands.c (vop_free_bucket_size): Never return value
380 greater than NUM_VOP_FREE_BUCKETS.
382 2007-01-18 Daniel Berlin <dberlin@dberlin.org>
384 * tree-ssa-structalias.c: Update comments.
385 (ptabitmap_obstack): Removed.
386 (pta_obstack): New.
387 (oldpta_obstack): Ditto.
388 (stats): Add a few members.
389 (struct variable_info): Remove node, complex, address_taken, and
390 indirect_target members. Add oldsolution member.
391 (new_var_info): Do not initialize removed members.
392 (constraint_expr_type): Remove INCLUDES.
393 (constraint_graph): Add size, implicit_preds, rep,
394 indirect_cycles, eq_rep, label, direct_nodes, and complex members.
395 (FIRST_REF_NODE): New macro.
396 (LAST_REF_NODE): Ditto.
397 (FIRST_ADDR_NODE): Ditto.
398 (find): New function.
399 (unite): Ditto.
400 (dump_constraint): Do not handle INCLUDES.
401 (insert_into_complex): Do not insert duplicate constraints.
402 (condense_varmap_nodes): Renamed and rewritten into ...
403 (merge_node_constraints): This. Also fix bug in handling of
404 offseted copy constraints.
405 (clear_edges_for_node): No longer need to deal with preds at all,
406 or removing associated preds/succs.
407 (merge_graph_nodes): Deal with indirect_cycles.
408 Don't deal with predecessors.
409 (add_implicit_graph_edge): New function.
410 (add_pred_graph_edge): Ditto.
411 (add_graph_edge): Don't deal with predecessors.
412 (build_constraint_graph): Removed.
413 (build_pred_graph): New function.
414 (build_succ_graph): Ditto.
415 (struct scc_info): Removed in_component. Added roots, dfs, and
416 node_mapping. Remove visited_index, unification_queue.
417 (scc_visit): Deal with union-find we do now.
418 Deal with cycles with REF nodes.
419 (collapse_nodes): Renamed and rewritten to ...
420 (unify_nodes): This.
421 (process_unification_queue): Removed.
422 (topo_visit): Cleanup
423 (do_da_constraint): Use find.
424 (do_sd_constraint): Ditto.
425 (do_ds_constraint): Ditto.
426 (do_complex_constraint): Ditto.
427 (init_scc_info): Update for removed and added members.
428 (find_and_collapse_graph_cycles): Renamed and rewritten into ...
429 (find_indirect_cycles): This.
430 (equivalence_class): New variable.
431 (label_visit): New function.
432 (perform_variable_substitution): Rewritten.
433 (free_var_substitution_info): New function.
434 (find_equivalent_node): Ditto.
435 (move_complex_constraints): Ditto.
436 (eliminate_indirect_cycles): Ditto.
437 (solve_graph): Only propagate changed bits.
438 Use indirect cycle elimination.
439 Use find.
440 (tree_id_t): Rename to tree_vi_t, delete id member, add vi member.
441 (tree_id_eq): Renamed to ...
442 (tree_vi_eq): This. Update for member change
443 (insert_id_for_tree): Renamed and rewritten to ...
444 (insert_vi_for_tree): This.
445 (lookup_id_for_tree): Renamed and rewritten to ...
446 (lookup_vi_for_tree): This.
447 (get_id_for_tree): Renamed and rewritten to ...
448 (get_vi_for_tree): Ditto.
449 (get_constraint_exp_from_ssa_var): Update to use get_vi_for_tree.
450 (process_constraint): Don't handle INCLUDES.
451 Remove special ADDRESSOF case.
452 (find_func_aliases): Rewrite to use vi functions instead of id
453 ones.
454 (create_function_info_for): Ditto.
455 (create_variable_info_for): Ditto.
456 (intra_create_variable_infos): Ditto.
457 (merge_smts_into): Ditto.
458 (find_what_p_points_to): Ditto.
459 (init_base_vars): Ditto.
460 (init_alias_vars): Ditto.
461 (remove_preds_and_fake_succs): New function.
462 (dump_sa_points_to_info): Dump new stats.
463 (dump_solution_for_var): Use find.
464 (set_used_smts): Fix formatting.
465 (compute_points_to_sets): Updated for new functions.
466 (ipa_pta_execute): Ditto.
468 2007-01-18 Kazu Hirata <kazu@codesourcery.com>
469 Richard Sandiford <richard@codesourcery.com>
471 * doc/tm.texi (TARGET_FUNCTION_VALUE): Expand documentation of
472 parallels.
473 * calls.c (expand_call): If the return value is a PARALLEL,
474 extract its first member.
475 * config/m68k/linux.h (FUNCTION_EXTRA_EPILOGUE): Remove.
476 * config/m68k/m68k.c (m68k_output_function_epilogue): Don't
477 use FUNCTION_EXTRA_EPILOGUE.
478 (m68k_function_value): Return a PARALLEL if the return value
479 is of a pointer type.
480 * config/m68k/netbsd-elf.h (current_function_returns_pointer)
481 (FUNCTION_EXTRA_EPILOGUE): Remove.
482 * config/m68k/m68k.md (D0_REG): New constant.
484 2007-01-18 Kazu Hirata <kazu@codesourcery.com>
486 * config/m68k/m68k.c (m68k_output_function_epilogue): Don't
487 output a NOP for empty epilogues.
489 2007-01-18 Richard Sandiford <richard@codesourcery.com>
491 * config/m68k/m68k.c (m68k_use_return_insn): Update comments
492 before function. Extend register save check to include all
493 registers, not just integer ones.
495 2007-01-18 Kazu Hirata <kazu@codesourcery.com>
497 * config/m68k/m68k-protos.h (use_return_insn): Rename to...
498 (m68k_use_return_insn): ...this.
499 * config/m68k/m68k.h (USE_RETURN_INSN): Delete.
500 * config/m68k/m68k.c (use_return_insn): Rename to...
501 (m68k_use_return_insn): ...this.
502 * config/m68k/m68k.md (return): Use m68k_use_return_insn instead
503 of USE_RETURN_INSN.
505 2007-01-18 Nathan Sidwell <nathan@codesourcery.com>
507 * config/m68k/fpgnulib.c (__truncdfsf2): Implement round to
508 nearest even, fix denormal rounding overflow.
510 2007-01-18 Richard Sandiford <richard@codesourcery.com>
512 * config/m68k/m68k.md (movsf_cf_hard): Use fsmove instead of
513 f%$smove and f%$move.
514 (movdf_cf_hard): Use fdmove for cases 0 and 3 and fmove for case 1.
515 (extendsfdf2_cf): Use fdmove instead of f%&move.
516 (truncdfsf2_cf): Use fsmove instead of f%$smove.
517 (add<mode>3_cf, sub<mode>3_cf): Use <FP:prec> instead of <FP:round>.
519 2007-01-18 Richard Sandiford <richard@codesourcery.com>
521 * config/m68k/m68k.md (movdf_cf_hard): Use output_move_double for
522 GPR<-GPR moves.
524 2007-01-18 Richard Sandiford <richard@codesourcery.com>
526 * real.h (real_format): Add a canonical_nan_lsbs_set field.
527 (coldfire_single_format): Declare.
528 (coldfire_double_format): Likewise.
529 * real.c (encode_ieee_single): Use canonical_nan_lsbs_set instead
530 of qnan_msb_set to determine the lower bits of a canonical
531 NaN significand.
532 (encode_ieee_double): Likewise.
533 (encode_ieee_quad): Likewise.
534 (ieee_single_format): Initialize canonical_nan_lsbs_set.
535 (mips_single_format): Likewise.
536 (ieee_double_format): Likewise.
537 (mips_double_format): Likewise.
538 (ieee_extended_motorola_format): Likewise.
539 (ieee_extended_intel_96_format): Likewise.
540 (ieee_extended_intel_128_format): Likewise.
541 (ieee_extended_intel_96_round_53_format): Likewise.
542 (ibm_extended_format): Likewise.
543 (mips_extended_format): Likewise.
544 (ieee_quad_format): Likewise.
545 (mips_quad_format): Likewise.
546 (vax_f_format): Likewise.
547 (vax_d_format): Likewise.
548 (vax_g_format): Likewise.
549 (i370_single_format): Likewise.
550 (i370_double_format): Likewise.
551 (decimal_single_format): Likewise.
552 (decimal_double_format): Likewise.
553 (decimal_quad_format): Likewise.
554 (c4x_single_format): Likewise.
555 (c4x_extended_format): Likewise.
556 (real_internal_format): Likewise.
557 (coldfire_single_format): New real_format.
558 (coldfire_double_format): Likewise.
559 * config/pdp11/pdp11.c (pdp11_f_format): Initialize
560 canonical_nan_lsbs_set.
561 (pdp11_d_format): Likewise.
562 * config/m68k/m68k.c (override_options): Override REAL_FORMAT_MODE
563 if TARGET_COLDFIRE_CPU.
565 2007-01-18 Richard Sandiford <richard@codesourcery.com>
567 * config/m68k/m68k-protos.h (m68k_output_pic_call): Delete.
568 (output_call, m68k_legitimize_call_address): Declare.
569 * config/m68k/m68k.h (EXTRA_CONSTRAINT): Remove unnecessary
570 parenthesees. Add support for a 'W' constraint.
571 (LEGITIMATE_PIC_OPERAND_P): Remove SYMBOL_REF_FLAG handling.
572 (PRINT_OPERAND_PUNCT_VALID_P): Remove comment about 'o'.
573 (m68k_symbolic_call, m68k_symbolic_jump): Declare.
574 * config/m68k/m68k.c (m68k_symbolic_call, m68k_symbolic_jump): New
575 variables.
576 (override_options): Initialize them. Do not set flag_no_function_cse
577 for TARGET_ID_SHARED_LIBRARY.
578 (m68k_output_pic_call): Delete.
579 (m68k_legitimize_call_address): New function.
580 (print_operand): Remove the %o prefix. Handle the %p prefix.
581 (output_call): New function.
582 (m68k_output_mi_thunk): Use m68k_symbolic_jump. Always load the
583 target address from the GOT if symbolic jumps are not allowed.
584 * config/m68k/m68k.md (call, general_operand): Do not set
585 SYMBOL_REF_FLAG. Use m68k_legitimize_call_address instead.
586 Merge separate flag_pic and !flag_pic define_insns into...
587 (*call, *call_value): ...these new patterns. Match the address
588 rather than the containing MEM and require it to be a call_operand.
589 Use output_call to generate the asm template.
590 * config/m68k/predicates.md (const_call_operand): New predicate.
591 (call_operand): Likewise.
593 2007-01-18 Nathan Sidwell <nathan@codesourcery.com>
595 * config/m68k/m68k.h (REGISTER_MOVE_COST): Simplify definition.
596 (STACK_GROWS_DOWNWARD): Define to 1.
597 (FUNCTION_VALUE, LIBCALL_VALUE, FUNCTION_VALUE_REGNO_P): Equivocate
598 comments, emphasizing that these values are only defaults.
599 * config/m68k/linux.h (LINK_SPEC): Fix formatting in #undef.
600 * config/m68k/m68k.c (const_method): Remove trailing whitespace.
602 2007-01-18 Richard Sandiford <richard@codesourcery.com>
604 * config/m68k/m68k.md (cmpsi): Remove outdated flag_pic handling.
606 2007-01-18 Kazu Hirata <kazu@codesourcery.com>
607 Richard Sandiford <richard@codesourcery.com>
609 * config/m68k/m68k.h (DATA_REGNO_P, ADDRESS_REGNO_P, INT_REGNO_P)
610 (FP_REGNO_P): New macros.
611 (REGNO_OK_FOR_INDEX_P, REGNO_OK_FOR_BASE_P, REGNO_OK_FOR_DATA_P)
612 (REGNO_OK_FOR_FP_P, REG_OK_FOR_INDEX_P, REG_OK_FOR_BASE_P): Use them.
613 (EH_RETURN_STACKADJ_RTX): Use A0_REG.
614 * config/m68k/m68k.c (m68k_regno_mode_ok): Use the new REGNO macros.
616 2007-01-18 Nathan Sidwell <nathan@codesourcery.com>
618 * config.gcc (m68k-*-aout*, m68k-*-coff*, m68020-*-elf*, m68k-*-elf*)
619 (m68k-*-uclinuxoldabi, m68k-*-uclinux*, m68k-*-rtems*): Add t-floatlib
620 to $tmake_file.
621 * config/m68k/t-floatlib: New file, extracting common code from...
622 * config/m68k/t-m68kbare, config/m68k/t-m68kelf,
623 * config/m68k/t-uclinux: Here.
624 * config/m68k/fpgnulib.c: Do not compile extendeed precision
625 routines on ColdFire targets.
627 2007-01-18 Nathan Sidwell <nathan@codesourcery.com>
629 * config.gcc (m68k-*-aout*, m68k-*-coff*, m68020-*-elf*, m68k-*-elf*)
630 (m68010-*-netbsdelf*, m68k*-*-netbsdelf*, m68k*-*-openbsd*)
631 (m68k-*-uclinuxoldabi, m68k-*-uclinux*, m68k-*-linux*)
632 (m68k-*-rtems*): Use tm_file rather than m68k/m68k.h and
633 explicitly set MOTOROLA to 1.
634 * config/m68k/m68k.h (MOTOROLA): Simplify definition accordingly.
636 2007-01-18 Richard Sandiford <richard@codesourcery.com>
637 Nathan Sidwell <nathan@codesourcery.com>
639 * config/m68k/m68k.h (PCC_STATIC_STRUCT_RETURN): Delete.
640 (ASM_OUTPUT_REG_PUSH, ASM_OUTPUT_REG_POP): Add MOTOROLA cases.
641 * config/m68k/coff.h (REGISTER_PREFIX_MD): Delete.
642 * config/m68k/m68020-elf.h (LIB_SPEC): Delete.
643 * config/m68k/m68k-none.h (CC1_SPEC, CPP_SUBTARGET_SPEC): Delete.
644 * config/m68k/m68kelf.h (IMMEDIATE_PREFIX, REGISTER_PREFIX_MD)
645 (ASM_OUTPUT_REG_PUSH, ASM_OUTPUT_REG_POP): Delete.
646 (NO_DOLLAR_IN_LABEL, PCC_STATIC_STRUCT_RETURN): Don't undefine.
647 (BSS_ASM_OP, ASM_OUTPUT_SKIP, ASM_OUTPUT_ASCII): Delete.
648 * config/m68k/m68kemb.h (PCC_STATIC_STRUCT_RETURN): Don't undefine.
649 * config/m68k/linux.h (SIZE_TYPE, PTRDIFF_TYPE, WCHAR_TYPE)
650 (WCHAR_TYPE_SIZE, TARGET_OBJFMT_CPP_BUILTINS): Delete.
651 (TARGET_OS_CPP_BUILTINS): Don't define mc68000 and mc68020 here.
652 (DBX_REGISTER_NUMBER): Delete.
653 * config/m68k/netbsd-elf.h (IMMEDIATE_PREFIX): Delete.
654 (PCC_STATIC_STRUCT_RETURN): Don't undefine.
655 * config/m68k/openbsd.h (PCC_STATIC_STRUCT_RETURN): Define.
657 2007-01-18 Richard Sandiford <richard@codesourcery.com>
659 * config.gcc (m68k-*-uclinux*): Add flat.h to $tm_file.
660 * config/flat.h: New file.
661 * crtstuff.c (USE_PT_GNU_EH_FRAME): Don't define if
662 OBJECT_FORMAT_FLAT.
663 * config/m68k/m68k.h (ASM_PREFERRED_EH_DATA_FORMAT): Do not use
664 indirect references for -msep-data or -mid-shared-library.
665 Do not use PC-relative code addresses either.
666 * config/m68k/m68k.c (override_options): Restrict -fPIC error
667 to -mpcrel.
668 * config/m68k/uclinux.h (STARTFILE_SPEC): Define. Use Scrt1.o
669 for shared libraries and crt1.o for executables. Use crti.o and
670 crtbegin.o.
671 (ENDFILE_SPEC): Use crtend.o and crtn.o.
672 (LIB_SPEC): Suppress -Rlibc.gdb if -static-libc is given.
673 Do not add -elf2flt or -shared-lib-id options here.
674 (LINK_SPEC): Define. Pass -elf2flt if no -elf2flt option is given.
675 Pass -shared-lib-id if -mid-shared-library, taking the library
676 identifier from -mshared-library-id if given, otherwise
677 defaulting to 0.
678 (EH_FRAME_IN_DATA_SECTION): Do not undefine.
679 (INIT_SECTION_ASM_OP, FINI_SECTION_ASM_OP): Likewise.
680 (TARGET_OS_CPP_BUILTINS): Define __GXX_MERGED_TYPEINFO_NAMES=0
681 and __GXX_TYPEINFO_EQUALITY_INLINE=0 if -mid-shared-library.
682 (DRIVER_SELF_SPECS): Map unadorned PIC options to -msep-data.
683 * config/m68k/t-uclinux (EXTRA_MULTILIB_PARTS): Add crtbegin.o
684 and crtend.o.
685 * config/m68k/lb1sf68.asm (PICCALL): Use an lea and pc-relative
686 jump sequence for ISA A and ISA A+.
687 (PICJUMP): Likewise.
689 2007-01-18 Richard Sandiford <richard@codesourcery.com>
691 * config.gcc (m68k-*-uclinux*): Base the port on the common
692 and m68k GNU/Linux files rather than on the generic ELF ones.
693 * config/m68k/uclinux.h (TARGET_VERSION): Override.
694 (TARGET_OS_CPP_BUILTINS): Use LINUX_TARGET_OS_CPP_BUILTINS.
696 2007-01-18 Julian Brown <julian@codesourcery.com>
697 Richard Sandiford <richard@codesourcery.com>
699 * config/m68k/m68k.h (LONG_DOUBLE_TYPE_SIZE): Make 64-bit on ColdFire.
700 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
701 * config/m68k/netbsd-elf.h (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Undefine
702 before redefining.
703 * config/m68k/uclinux-oldabi.h (LONG_DOUBLE_TYPE_SIZE): Redefine to
704 80 unconditionally.
705 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
707 2007-01-18 Richard Sandiford <richard@codesourcery.com>
709 * doc/install.texi: Document m68k-uclinuxoldabi.
710 * config.gcc (m68k-*-uclinuxoldabi): New configuration.
711 * config/m68k/uclinux-oldabi.h: New file, copied from
712 config/m68k/uclinux.h.
714 2007-01-18 Kaz Kojima <kkojima@rr.iij4u.or.jp>
716 * config/m32r/m32r-protos.h (m32r_expand_epilogue): Declare it.
717 * config/m32r/m32r.c (m32r_setup_incoming_varargs): Use gen_frame_mem.
718 (m32r_compute_frame_size): Use unsigned for regno.
719 (m32r_reload_lr): Use gen_frame_mem.
720 (pop): New.
721 (m32r_output_function_epilogue): Don't output the function epilogue
722 textually here.
723 (m32r_expand_epilogue): New.
724 (direct_return): Return false if the function has the interrupt
725 attribute.
726 (m32r_hard_regno_rename_ok): Remove code for the textual epilogue.
727 * config/m32r/m32r.md (epilogue): New expander.
728 (return_lr, return_rte): New insns.
729 (return): Make it expander.
730 (return_normal): New expander.
732 2007-01-18 Josh Conner <jconner@apple.com>
734 PR target/30485
735 * config/rs6000/rs6000.c (rs6000_emit_vector_compare): Add
736 support for UNLE, UNLT, UNGE, and UNGT.
738 2007-01-18 Jan Hubicka <jh@suse.cz>
740 * tree-vrp.c (finalize_jump_threads): Do not call cleanup_cfg by hand.
741 * tree-tailcall (add_virtual_phis): Likewise.
742 (optimize_tail_call): Return TODOs.
743 (execute_tail_calls): Return TODOs.
744 * tree-ssa-ccp (execute_fold_all_builtins): Do cleanup_cfg via TODO.
745 * tree-cfgcleanup.c (cleanup_tree_cfg_loop): Return if something
746 changed.
747 * tree-ssa-forwprop.c (tree_ssa_forward_propagate_single_use_value):
748 Cleanup cfg using TODO.
749 * tree-flow.h (cleanup_tree_cfg_loop): Update prototype.
750 * passes.c (execute_function_todo): When cleanup did something, remove
751 unused locals.
752 * tree-cfg.c (pass_build_cfg): Add cleanup_cfg TODO.
753 (make_edges): Don't cleanup_cfg.
755 2007-01-18 Uros Bizjak <ubizjak@gmail.com>
757 * reg-stack.c (subst_stack_regs_pat) [UNSPEC_SINCOS_COS,
758 UNSPEC_XTRACT_FRACT]: Use generic code for instructions that
759 operate on the top of stack.
760 [UNSPEC_SINCOS_SIN, UNSPEC_XTRACT_EXP, UNSPEC_TAN]: Rewrite
761 register handling of instructions that output to the second
762 stack slot.
763 [UNSPEC_TAN_ONE, UNSPEC_TAN_TAN]: Remove.
764 (move_for_stack_reg): Special-case check for dead destination
765 stack slot for constant load of 1.0 inside UNSPEC_TAN.
767 * config/i386/i386.md (UNSPEC_TAN): New constant.
768 (UNSPEC_TAN_ONE, UNSPEC_TAN_TAN): Remove.
769 (fptanxf4_i387, fptan_extend<mode>xf4_i387): New patterns
770 to correctly model move of constant 1.0 to top stack slot.
771 (*tandf3_1, *tansf3_1, *tanxf3_1): Remove insn patterns.
772 (unnamed peephole2 pattern): Remove corresponding peephole2
773 pattern that optimizes tan insn and loading of constant 1.0.
774 (tanxf2): Use fptanxf4_i387.
775 (tan<mode>2): Rename from tansf2 and tandf2 and macroize insn
776 patterns using X87MODEF12 mode macro. Use fptan_extend<mode>xf4_i387
777 and truncate result to requested mode. Use SSE_FLOAT_MODE_P to
778 disable patterns for SSE math.
779 (sincos<mode>3): Use truncxf<mode>2_i387_noop for truncation.
780 (fyl2x_extend<mode>xf3_i387): Use X87MODEF12 for operand 1.
782 2007-01-18 Dirk Mueller <dmueller@suse.de>
783 Richard Guenther <rguenther@suse.de>
785 PR diagnostic/8268
786 * doc/invoke.texi (Warray-bounds): Document -Warray-bounds.
787 * common.opt (Warray-bounds): Add new warning option.
788 * c-opts.c (c_common_handle_option): Define -Warray-bounds
789 if -Wall is given.
790 * Makefile.in: make tree-vrp.o depend on toplev.h
791 * tree-vrp.c (vrp_finalize): Call check_array_refs if -Warray-bounds
792 is enabled.
793 (check_array_refs, check_array_bounds, check_array_ref): New.
795 2007-01-18 Jan Hubicka <jh@suse.cz>
797 * tree-ssa-ccp.c (ccp_finalize): Return if something changed.
798 (execute_ssa_ccp): Return flags conditionally.
799 * tree-ssa-propagate.c (substitue_and_fold): Return if something was
800 changed.
801 * tree-ssa-propagate.h (substitute_and_fold): Update prototype.
803 2007-01-18 Steven Bosscher <steven@gcc.gnu.org>
805 * cfgcleanup.c (cleanup_cfg): Detect cfglayout mode and set
806 the CLEANUP_CFGLAYOUT flag when in cfglayout mode.
808 * Makefile.c (GTFILES): Add cfglayout.h.
809 * gengtype.c (open_base_files): Likewise.
810 * cfglayout.c (cfg_layout_function_footer,
811 cfg_layout_function_header) Reindent to make gengtype happy.
812 * cfglayout.h (cfg_layout_function_footer,
813 cfg_layout_function_header): Add GTY(()) marker.
815 * ifcvt.c (noce_try_sign_mask): Make sure INSN_B is non-null.
817 2007-01-18 Ben Elliston <bje@au.ibm.com>
819 * genautomata.c (write_automata): Include xstrerror output in the
820 error message if writing the DFA description file fails.
822 2007-01-17 H.J. Lu <hongjiu.lu@intel.com>
824 * config/mips/mips-protos.h (mips_output_external): Make it
825 return void.
826 * config/mips/iris.h (TARGET_ASM_EXTERNAL_LIBCALL): Removed.
827 * config/mips/mips.c (irix_output_external_libcall): Likewise.
828 (extern_list): Likewise.
829 (extern_head): Likewise.
830 (TARGET_ASM_FILE_END): Likewise.
831 (mips_file_end): Likewise.
832 (mips_output_external): Rewritten.
834 2007-01-18 Ben Elliston <bje@au.ibm.com>
836 * genpreds.c (write_insn_preds_c): Only write out the function
837 body for regclass_for_constraint if we have register constraints.
839 2007-01-17 Tom Tromey <tromey@redhat.com>
841 * doc/sourcebuild.texi (libgcj Tests): Use sourceware.org.
842 * doc/install.texi (Testing): Use sourceware.org.
843 (Binaries): Likewise.
844 (Specific): Likewise.
845 * doc/contrib.texi (Contributors): Use sourceware.org.
847 2007-01-17 Anatoly Sokolov <aesok@post.ru>
849 * config/avr/avr.h (AVR_HAVE_LPMX): New macro.
850 (AVR_ENHANCED): Rename to ...
851 (AVR_HAVE_MUL): ... new.
852 (avr_enhanced_p): Rename to ...
853 (avr_have_mul_p): ... new.
854 (TARGET_CPU_CPP_BUILTINS): Use 'avr_have_mul_p' instead of
855 'avr_enhanced_p' for "__AVR_ENHANCED__". Define "__AVR_HAVE_MUL__".
856 * config/avr/avr.c (avr_enhanced_p): Rename to ...
857 (avr_have_mul_p): ... new.
858 (base_arch_s): Rename 'enhanced' to 'have_mul'.
859 (avr_override_options): Use 'avr_have_mul_p' and 'have_mul' instead of
860 'avr_enhanced_p' and 'enhanced'.
861 (ashlhi3_out, ashrhi3_out, lshrhi3_out, avr_rtx_costs): Use
862 AVR_HAVE_MUL instead of AVR_ENHANCED.
863 * avr.md (*tablejump_enh): Use AVR_HAVE_LPMX instead of AVR_ENHANCED.
864 (mulqi3, *mulqi3_enh, *mulqi3_call, mulqihi3, umulqihi3, mulhi3,
865 *mulhi3_enh, *mulhi3_call, mulsi3, *mulsi3_call): Use AVR_HAVE_MUL
866 instead of AVR_ENHANCED.
867 (*tablejump_enh): Use AVR_HAVE_LPMX instead of AVR_ENHANCED.
868 * libgcc.S: Use __AVR_HAVE_MUL__ instead of __AVR_ENHANCED__.
869 (__tablejump__): Use __AVR_HAVE_LPMX__ instead of __AVR_ENHANCED__.
871 2007-01-17 Ian Lance Taylor <iant@google.com>
873 * vec.h (VEC_reserve_exact): Define.
874 (vec_gc_p_reserve_exact): Declare.
875 (vec_gc_o_reserve_exact): Declare.
876 (vec_heap_p_reserve_exact): Declare.
877 (vec_heap_o_reserve_exact): Declare.
878 (VEC_OP (T,A,reserve_exact)): New static inline function, three
879 versions.
880 (VEC_OP (T,A,reserve)) [all versions]: Remove handling of
881 negative parameter.
882 (VEC_OP (T,A,alloc)) [all versions]: Call ...reserve_exact.
883 (VEC_OP (T,A,copy)) [all versions]: Likewise.
884 (VEC_OP (T,a,safe_grow)) [all versions]: Likewise.
885 * vec.c (calculate_allocation): Add exact parameter. Change all
886 callers.
887 (vec_gc_o_reserve_1): New static function, from vec_gc_o_reserve.
888 (vec_gc_p_reserve, vec_gc_o_reserve): Call vec_gc_o_reserve_1.
889 (vec_gc_p_reserve_exact, vec_gc_o_reserve_exact): New functions.
890 (vec_heap_o_reserve_1): New static function, from vec_heap_o_reserve.
891 (vec_heap_p_reserve, vec_heap_o_reserve): Call vec_heap_o_reserve_1.
892 (vec_heap_p_reserve_exact): New function.
893 (vec_heap_o_reserve_exact): New function.
895 2007-01-17 Jan Hubicka <jh@suse.cz>
897 * ipa-type-escape.c (look_for_casts): Revamp using handled_component_p.
899 2007-01-17 Eric Christopher <echristo@apple.com>
901 * config.gcc: Support core2 processor.
903 2007-01-16 Jan Hubicka <jh@suse.cz>
905 * tree-ssanames.c (release_dead_ssa_names): Instead of ggc_freeing
906 the names, just unlink the chain so we don't crash on dangling pointers
907 to dead SSA names.
909 2007-01-16 Jan Hubicka <jh@suse.cz>
911 * cgraph.h (cgraph_decide_inlining_incrementally): Kill.
912 * tree-pass.h: Reorder to make IPA passes appear toegher.
913 (pass_early_inline, pass_inline_parameters, pass_apply_inline): Declare.
914 * cgraphunit.c (cgraph_finalize_function): Do not compute inling
915 parameters, do not call early inliner.
916 * ipa-inline.c: Update comments. Include tree-flow.h
917 (cgraph_decide_inlining): Do not compute inlining parameters.
918 (cgraph_decide_inlining_incrementally): Return TODOs; assume to
919 be called with function context set up.
920 (pass_ipa_inline): Remove unreachable functions before pass.
921 (cgraph_early_inlining): Simplify assuming to be called from the
922 PM as local pass.
923 (pass_early_inline): New pass.
924 (cgraph_gate_ipa_early_inlining): New gate.
925 (pass_ipa_early_inline): Turn into simple wrapper.
926 (compute_inline_parameters): New function.
927 (gate_inline_passes): New gate.
928 (pass_inline_parameters): New pass.
929 (apply_inline): Move here from tree-optimize.c
930 (pass_apply_inline): New pass.
931 * ipa.c (cgraph_remove_unreachable_nodes): Verify cgraph after
932 transforming.
933 * tree-inline.c (optimize_inline_calls): Return TODOs rather than
934 doing them by hand.
935 (tree_function_versioning): Do not allocate dummy struct function.
936 * tree-inline.h (optimize_inline_calls): Update prototype.
937 * tree-optimize.c (execute_fixup_cfg): Export.
938 (pass_fixup_cfg): Remove
939 (tree_rest_of_compilation): Do not apply inlines.
940 * tree-flow.h (execute_fixup_cfg): Declare.
941 * Makefile.in (gt-passes.c): New.
942 * passes.c: Include gt-passes.h
943 (init_optimization_passes): New passes.
944 (nnodes, order): New static vars.
945 (do_per_function_toporder): New function.
946 (execute_one_pass): Dump current pass here.
947 (execute_ipa_pass_list): Don't dump current pass here.
949 2007-01-16 Janis Johnson <janis187@us.ibm.com>
951 * config/dfp-bit.c (dfp_compare_op): Return separate value for NaN.
952 (DFP_NE, DFP_LE, DFP_GE): Return false for NaN.
954 2007-01-16 David Edelsohn <edelsohn@gnu.org>
956 * config/rs6000/darwin-ldouble.c: Build file for SOFT_FLOAT.
957 (strong_alias): Define.
958 (__gcc_qmul): Provide non-FMA for soft-float.
959 (__gcc_qdiv): Same.
960 (__gcc_qneg): New.
961 (__gcc_qeq): New.
962 (__gcc_qle): New.
963 (__gcc_qge): New.
964 (__gcc_qunord): New.
965 (__gcc_stoq): New.
966 (__gcc_dtoq): New.
967 (__gcc_qtos): New.
968 (__gcc_qtod): New.
969 (__gcc_qtoi): New.
970 (__gcc_qtou): New.
971 (__gcc_itoq): New.
972 (__gcc_utoq): New.
973 (fmsub): New.
974 * config/rs6000/rs6000.c (rs6000_init_libfuncs): Initialize
975 soft-float functions.
976 * config/rs6000/libgcc-ppc-glibc.ver: Version soft-float symbols.
977 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Do not warn
978 about long double soft float.
980 2007-01-16 Dorit Nuzman <dorit@il.ibm.com>
981 Tehila Meyzels <tehila@il.ibm.com>
983 * tree-vectorizer.h (is_pattern_stmt_p): New.
984 * tree-vect-analyze.c (vect_determine_vectorization_factor): Fix
985 formatting (tabs instead of spaces). Cleanup and clarify setting
986 of STMT_VINFO_VECTYPE. Call is_pattern_stmt_p.
987 * tree-vect-transform.c (vect_get_vec_def_for_operand): Fix typo.
988 (vectorizable_type_demotion): Check that types are integral.
989 (vectorizable_type_promotion): Likewise.
990 (vectorizable_store): Fix typo. Eliminate new-line at end of
991 comments.
993 2007-01-16 Jan Hubicka <jh@suse.cz>
995 * tree-ssanames.c (release_dead_ssa_names): Remove invalidated
996 cgraph edges too.
998 2007-01-15 Eric Christopher <echristo@apple.com>
1000 * ifcvt.c: Include vec.h, vecprim.h.
1001 (check_cond_move_block): New argument regs.
1002 Reorganize. Add registers used to regs.
1003 (cond_move_process_if_block): Use regs set above as
1004 loop bounds.
1006 2007-01-15 Eric Christopher <echristo@apple.com>
1008 * config/darwin.h: Update copyright.
1009 (TARGET_OPTION_TRANSLATE_TABLE): Add umbrella.
1010 (LINK_COMMAND_SPEC): Add -u.
1011 (LINK_SPEC): Fix umbrella for above.
1013 2007-01-15 Joseph S. Myers <joseph@codesourcery.com>
1015 * config/soft-fp/op-common.h, config/soft-fp/op-4.h: Update from
1016 glibc CVS.
1018 2007-01-15 Tom Tromey <tromey@redhat.com>
1020 * doc/sourcebuild.texi (libgcj Tests): Don't mention jacks.
1021 * doc/install.texi (Testing): Don't mention jacks.
1022 (Configuration): Document --enable-java-maintainer-mode. Move
1023 --with-java-home to libgcj-specific section. Document
1024 --with-ecj-jar.
1025 (Prerequisites): Mention --enable-java-maintainer-mode, ecj1.
1027 2007-01-15 Jan Hubicka <jh@suse.cz>
1029 * tree-ssa-dce.c (DCE_TODOs): New.
1030 (propagate_necessity): Return if something changed.
1031 (eliminate_unnecessary_stmts): Likewise.
1032 (perform_tree_ssa_dce): Return TODO flags when needed.
1033 (pass_dce, pass_dce_loop, pass_cd_dce): Remove TODO flags.
1035 2007-01-15 Uros Bizjak <ubizjak@gmail.com>
1037 * config/i386/i386.md (fyl2xxf3_i387): Rename from fyl2x_xf3.
1038 (fyl2x_extend<mode>xf3_i387): New insn pattern.
1039 (log<mode>2): Rename from logsf2 and logdf2 and macroize insn
1040 patterns using X87MODEF12 mode macro. Extend operand 1
1041 to XFmode. Use SSE_FLOAT_MODE_P to disable patterns for SSE math.
1042 (log10<mode>2): Ditto.
1043 (log2<mode>2): Ditto.
1044 (log1p<mode>2): Ditto.
1045 (logb<mode>2): Ditto.
1046 (fyl2xp1xf3_i387): Rename from fyl2xp1_xf3.
1047 (fyl2xp1_extend<mode>xf3_i387): New insn pattern.
1048 (*fxtractxf3_i387): Rename from *fxtractxf3.
1049 (fxtract_extend<mode>xf3_i387): New insn pattern.
1050 (ilogbsi2): Use match_dup 3, not match_operand:XF 3.
1051 * config/i386/i386.c (ix86_emit_i387_log1p): Use gen_fyl2xp1xf3_i387()
1052 and gen_fyl2xxf3_i387().
1054 2007-01-14 Zdenek Dvorak <dvorakz@suse.cz>
1056 * loop-unswitch.c (unswitch_loop): Do not call fix_loop_placement.
1057 * cfgloopmanip.c (fix_loop_placement): Made static. Use
1058 get_loop_exit_edges. Changed return type to bool.
1059 * cfgloop.h (fix_loop_placement): Declaration removed.
1061 2007-01-14 Dorit Nuzman <dorit@il.ibm.com>
1063 * param.h (MIN_VECT_LOOP_BOUND): New.
1064 * params.def (MIN_VECT_LOOP_BOUND): New.
1065 * tree-vectorizer.c (slpeel_tree_peel_loop_to_edge): Takes another
1066 argument - minimum threshold for number of iterations.
1067 * tree-vectorizer.h (slpeel_tree_peel_loop_to_edge): Add another
1068 argument to declaration.
1069 * tree-vect-analyze.c (vect_analyze_operations): Check value of
1070 MIN_VECT_LOOP_BOUND.
1071 * tree-vect-transform.c (vect_do_peeling_for_loop_bound): Call
1072 slpeel_tree_peel_loop_to_edge with additional argument.
1073 (vect_do_peeling_for_alignment): Likewise.
1074 * doc/invoke.texi (min-vect-loop-bound): Document new param option.
1076 2007-01-14 Uros Bizjak <ubizjak@gmail.com>
1078 PR target/30413
1079 * config/i386/i386.c (print_operand) ['z']: Output 'b' for
1080 operands of size 1.
1082 2007-01-14 Jan Hubicka <jh@suse.cz>
1084 * tree-dfa.c (remove_referenced_var): New function.
1085 * tree-ssa-live.c (remove_unused_locals): Walk referenced vars and
1086 prune referenced vars list too.
1087 * tree-flow.h (remove_referenced_var): Declare.
1089 2007-01-14 Jan Hubicka <jh@suse.cz>
1091 * tree-eh.c (add_stmt_to_eh_region_fn): Do not add call_exprs
1092 separately.
1093 (remove_stmt_from_eh_region_fn): Do not remove call_exprs.
1094 (verify_eh_throw_stmt_node, verify_eh_throw_table_statements): Kill.
1095 * except.h (verify_eh_throw_table_statements): Kill prototype.
1096 * cfgexpand.c (expand_gimple_basic_block): Propagate Eh regions
1097 into call exrepssions.
1098 * tree-optimize.c (execute_free_cfg_annotatiosn): Do not call
1099 eh trhow verifier.
1100 * tree-cfg.c: Include pointer-set.h.
1101 (verify_node_sharing): Work on pointer set.
1102 (verify_eh_throw_stmt_node): New.
1103 (verify_stmts): Use pointers sets, verify throw_stmt.
1105 2007-01-13 Zdenek Dvorak <dvorakz@suse.cz>
1107 * ipa-reference.c (analyze_function): Consider also addresses taken
1108 in phi nodes.
1110 2007-01-12 Roger Sayle <roger@eyesopen.com>
1112 * c-typeck.c (null_pointer_constant_p): Replace use of
1113 TREE_CONSTANT_OVERFLOW with TREE_OVERFLOW.
1114 (build_c_cast): Likewise.
1116 2007-01-12 Roger Sayle <roger@eyesopen.com>
1118 * tree.h (force_fit_type_double): Remove unused final argument.
1119 * c-common.c (constant_expression_warning): Replace use of
1120 TREE_CONSTANT_OVERFLOW with TREE_OVERFLOW.
1121 (convert_and_check): Likewise.
1122 (shorten_compare): Update call to force_fit_type_double.
1123 (c_common_truthvalue_conversion) <INTEGER_CST>: Use integer_zerop.
1124 * convert.c (convert_to_pointer): Update call to
1125 force_fit_type_double.
1126 * fold-const.c (force_fit_type_double): Remove overflowed_const
1127 argument.
1128 (int_const_binop, fold_convert_const_int_from_int,
1129 fold_convert_const_int_from_real, fold_div_compare,
1130 fold_sign_changed_comparison, fold_unary, fold_negate_const,
1131 fold_abs_const, fold_not_const): Remove the final argument from
1132 calls to force_fit_type_double.
1134 2007-01-12 Andrew Pinski <andrew_pinski@playstation.sony.com>
1136 * configure.ac: Set insn to "nop" for spu-*-* also.
1137 * configure: Regenerate.
1139 2007-01-12 Olga Golovanevsky <olga@il.ibm.com>
1141 * builtins.def : Add BUILT_IN_FREE.
1143 2007-01-12 Jan Hubicka <jh@suse.cz>
1145 PR tree-optimization/30443
1146 * tree-inline.c (tree_function_versioning): Do not optimize when
1147 cloning for inlining.
1149 2007-01-12 Zdenek Dvorak <dvorakz@suse.cz>
1151 * doc/loop.texi: Document recording of loop exits.
1152 * cfgloopmanip.c (loopify, duplicate_loop): Use alloc_loop.
1153 (update_single_exits_after_duplication,
1154 update_single_exit_for_duplicated_loop,
1155 update_single_exit_for_duplicated_loops): Removed.
1156 (duplicate_loop_to_header_edge): Do not call
1157 update_single_exits_after_duplication and
1158 update_single_exit_for_duplicated_loops.
1159 (loop_version): Do not update single_exit information.
1160 (fix_loop_structure): Use record_loop_exits instead of
1161 mark_single_exit_loops.
1162 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update
1163 the lists of loop exits.
1164 * cfghooks.c (redirect_edge_and_branch, redirect_edge_and_branch_force,
1165 split_edge, merge_blocks): Update the lists of loop exits.
1166 * modulo-sched.c (sms_schedule): Pass LOOPS_HAVE_RECORDED_EXITS to
1167 loop_optimizer_init.
1168 * loop-init.c (loop_optimizer_init): Call record_loop_exits instead
1169 of mark_single_exit_loops.
1170 (loop_optimizer_finalize): Call release_recorded_exits.
1171 * tree-ssa-loop.c (tree_loop_optimizer_init): Pass
1172 LOOPS_HAVE_RECORDED_EXITS to loop_optimizer_init.
1173 * tree-vectorizer.c (slpeel_tree_duplicate_loop_to_edge_cfg): Do not
1174 update single exit information.
1175 * lambda-code.c (perfect_nestify): Ditto.
1176 * cfgloop.c (flow_loop_free): Destroy the list of exits of the loop.
1177 (mark_single_exit_loops): Removed.
1178 (alloc_loop, loop_exit_hash, loop_exit_eq, loop_exit_free,
1179 get_exit_descriptions, rescan_loop_exit, record_loop_exits,
1180 dump_recorded_exit, dump_recorded_exits, release_recorded_exits): New
1181 functions.
1182 (get_loop_exit_edges, single_exit): Use recorded exit lists.
1183 (add_bb_to_loop, remove_bb_from_loops): Update the lists of loop exits.
1184 (verify_loop_structure): Verify consistency of the exit lists.
1185 (flow_loops_find): Use alloc_loop. Initialize exits hash.
1186 (set_single_exit): Removed.
1187 * cfgloop.h (struct loop_exit): New function.
1188 (struct loop): single_exit_ field replaced by exits field.
1189 (LOOPS_HAVE_MARKED_SINGLE_EXITS): Replaced by LOOPS_HAVE_RECORDED_EXITS.
1190 (struct loops): Added exits hash.
1191 (mark_single_exit_loops, set_single_exit): Declaration removed.
1192 (release_recorded_exits, record_loop_exits, rescan_loop_exit): Declare.
1194 2007-01-12 Richard Sandiford <richard@codesourcery.com>
1196 * doc/invoke.texi: Avoid use of @headitem.
1198 2007-01-12 Richard Sandiford <richard@codesourcery.com>
1200 * cse.c (cse_insn): Move HAVE_CC0 code after declarations.
1202 2007-01-12 Richard Sandiford <richard@codesourcery.com>
1204 * doc/install.texi: Fix m68k-*-* anchor and add m68k-*-* to the
1205 list of targets.
1207 2007-01-12 Nathan Sidwell <nathan@codesourcery.com>
1208 Richard Sandiford <richard@codesourcery.com>
1210 * doc/invoke.texi: Document -mno-bitfield, -mno-rtd and -mno-short.
1211 * config/m68k/m68k.opt: Resort options.
1212 (mbitfield, mrtd, mshort): Remove RejectNegative properties.
1214 2007-01-12 Nathan Sidwell <nathan@codesourcery.com>
1215 Richard Sandiford <richard@codesourcery.com>
1217 * doc/invoke.texi: Document the macros that are defined by
1218 m68k's -mtune and -mhard-float options.
1219 * config/m68k/m68k-protos.h (m68k_cpp_cpu_ident) Declare.
1220 (m68k_cpp_cpu_family): Likewise.
1221 * config/m68k/m68k.h (TARGET_CPU_CPP_BUILTINS): Add a full set
1222 of __ucfv*__ macros. Define __mcffpu__ if generating code for
1223 ColdFire FPUs. Define __mcf_cpu_* and __mcf_family_* macros.
1224 * config/m68k/m68k.c (m68k_cpp_cpu_ident): New function.
1225 (m68k_cpp_cpu_family): Likewise.
1227 2007-01-12 Richard Sandiford <richard@codesourcery.com>
1229 * config/m68k/m68k.h (TARGET_CPU_CPP_BUILTINS): Treat all mc68*
1230 macros besides mc68000 as tuning macros. Use a switch statement
1231 to set them and mcpu32.
1233 2007-01-12 Julian Brown <julian@codesourcery.com>
1235 * config/m68k/m68k.h: Use TARGET_68040 instead of TARGET_68040_ONLY.
1236 (TARGET_68040_ONLY): Rename to...
1237 (TARGET_68040): ...this.
1238 * config/m68k/m68k.c: Use TARGET_68040 instead of TARGET_68040_ONLY.
1239 * config/m68k/m68k.md: Likewise.
1241 2007-01-12 Julian Brown <julian@codesourcery.com>
1242 Nathan Sidwell <nathan@codesourcery.com>
1243 Richard Sandiford <richard@codesourcery.com>
1245 * config.gcc (m680[012]0-*-*, m68k*-*-*): Set m68k_cpu_ident to
1246 the -mcpu= argument associated with the --with-cpu setting.
1247 Define M68K_DEFAULT_TUNE to the default -mtune= option,
1248 if different from the one implied by the -mcpu setting.
1249 Accept --with-cpu=FOO if FOO is listed in m68k-devices.def,
1250 using mcpu=FOO as the default CPU option. Set target_cpu_default2.
1251 * doc/invoke.texi: Mention ColdFire in the introduction to the
1252 m68k options. Document the new -march, -mcpu, -mtune, -mdiv,
1253 -mno-div and -mhard-float options. Make -m68881 a synonym for
1254 -mhard-float. Document the previously-undocumented -m5206e,
1255 -m528x, -m5307 and -m5407 options. Tweak the existing option
1256 documentation for consistency.
1257 * doc/install.texi: Mention new --with-cpu arguments.
1258 * config/m68k/m68k.h (OPTION_DEFAULT_SPECS): Only use the
1259 default CPU if neither -mcpu nor -march are specified.
1260 (ASM_CPU_SPEC): Pass down -mcpu and -march options.
1261 (TARGET_CPU_CPP_BUILTINS): Set __mcfisa*__ macros from
1262 TARGET_ISA*. Set the legacy __mcf*__ cpu macros in the same way,
1263 using m68k_tune to decide between families that implement the
1264 same ISA. Use m68k_tune to set __mcfv4e__.
1265 (FL_BITFIELD, FL_68881, FL_COLDFIRE, FL_CF_HWDIV, FL_CF_MAC)
1266 (FL_CF_EMAC, FL_CF_EMAC_B, FL_CF_USP, FL_CF_FPU, FL_ISA_68000)
1267 (FL_ISA_68010, FL_ISA_68020, FL_ISA_68040, FL_ISA_A, FL_ISA_B)
1268 (FL_ISA_C, FL_ISA_MMU): New macros.
1269 (MASK_COLDFIRE): Delete.
1270 (TARGET_68010, TARGET_68020, TARGET_68040_ONLY, TARGET_COLDFIRE)
1271 (TARGET_ISAB): Redefine in terms of m68k_cpu_flags.
1272 (TARGET_68881, TARGET_COLDFIRE_FPU): Redefine in terms of m68k_fpu.
1273 (TARGET_HARD_FLOAT): Do not define here.
1274 (TARGET_ISAAPLUS, TARGET_ISAC): New macros.
1275 (TUNE_68000): New macro.
1276 (TUNE_68000_10): Redefine in terms of TUNE_68000 and TUNE_68010.
1277 (TUNE_68010, TUNE_68030, TUNE_68040, TUNE_68060, TUNE_CPU32)
1278 (TUNE_CFV2): Redefine in terms of m68k_tune.
1279 (uarch_type, target_device, fpu_type): New enums.
1280 (m68k_cpu, m68k_tune, m68k_fpu, m68k_cpu_flags): Declare.
1281 * config/m68k/m68k.c (TARGET_DEFAULT): Remove MASK_68881.
1282 (FL_FOR_isa_00, FL_FOR_isa_10, FL_FOR_isa_20, FL_FOR_isa_40)
1283 (FL_FOR_isa_cpu32, FL_FOR_isa_a, FL_FOR_isa_aplus, FL_FOR_isa_b)
1284 (FL_FOR_isa_c): New macros.
1285 (m68k_isa): New enum.
1286 (m68k_target_selection): New structure.
1287 (all_devices, all_isas, all_microarchs): New tables.
1288 (m68k_cpu_entry, m68k_arch_entry, m68k_tune_entry, m68k_cpu)
1289 (m68k_tune, m68k_fpu, m68k_cpu_flags): New variables.
1290 (MASK_ALL_CPU_BITS): Delete.
1291 (m68k_find_selection): New function.
1292 (m68k_handle_option): Handle -mcpu=, -march= and -mtune=.
1293 Map the legacy target options to a combination of the new ones.
1294 (override_options): Set m68k_cpu, m68k_tune, m68k_fpu and
1295 m68k_cpu_flags. Handle M68K_DEFAULT_TUNE. Use m68k_cpu_flags
1296 to derive default MASK_BITFIELD, MASK_CF_HWDIV and MASK_HARD_FLOAT
1297 settings.
1298 * config/m68k/m68k.opt (m5200, m5206e, m528x, m5307, m5407, mcfv4e)
1299 (m68010, m68020, m68020-40, m68020-60, m68030, m68040): Remove Mask
1300 properties.
1301 (m68881, msoft-float): Change mask from 68881 to HARD_FLOAT.
1302 (march=, mcpu=, mdiv, mhard-float, mtune=): New options.
1303 * config/m68k/m68k-devices.def: New file.
1305 2007-01-12 Richard Sandiford <richard@codesourcery.com>
1306 Nathan Sidwell <nathan@codesourcery.com>
1308 * config/m68k/m68k.h (ASM_CPU_SPEC, ASM_SPEC, EXTRA_SPECS)
1309 (SUBTARGET_EXTRA_SPECS): New macros.
1310 * config/m68k/linux.h (ASM_SPEC): Remove CPU flags;
1311 use %(asm_cpu_spec) instead.
1312 * config/m68k/m68k-none.h (ASM_SPEC): Likewise.
1313 * config/m68k/openbsd.h (ASM_SPEC): Likewise.
1314 * config/m68k/netbsd-elf.h (ASM_SPEC): Likewise.
1315 (EXTRA_SPECS): Rename to...
1316 (SUBTARGET_EXTRA_SPECS): ...this.
1318 2007-01-12 Nathan Sidwell <nathan@codesourcery.com>
1319 Richard Sandiford <richard@codesourcery.com>
1320 Julian Brown <julian@codesourcery.com>
1322 * config.gcc (m68k-*-aout*, m68k-*-coff*, m68020-*-elf*, m68k-*-elf*)
1323 (m68k-*-uclinux*, m68k-*-linux*, m68k-*-rtems*): Set default_m68k_cpu
1324 to the configuration's default CPU.
1325 (m68010-*-netbsdelf*, m68k*-*-netbsdelf*, m68k*-*-openbsd*): Likewise.
1326 Remove default masks.
1327 (m680[012]0-*-*): Set the default with_cpu to the first part of
1328 the target name.
1329 (m68k*-*-*): Set the default with_cpu to m$default_m68k_cpu.
1330 (m68k*-*-linux): Extend the --with-cpu handling to...
1331 (m680[012]0-*-*, m68k*-*-*): ...these configurations. Allow m68000
1332 and m68010. Don't set target_cpu_default2.
1333 * doc/install.texi: Document --with-cpu for m68k.
1334 * config/m68k/m68k.h (OPTION_DEFAULT_SPECS): Define.
1335 * config/m68k/m68k-none.h (TARGET_CPU_DEFAULT, M68K_CPU_m68k)
1336 (M68K_CPU_m68000, M68K_CPU_m68010, M68K_CPU_m68020, M68K_CPU_m68030)
1337 (M68K_CPU_m68040, M68K_CPU_m68302, M68K_CPU_m68332, TARGET_DEFAULT)
1338 (ASM_CPU_DEFAULT_SPEC, CC1_CPU_DEFAULT_SPEC): Delete.
1339 (ASM_SPEC): Remove use of %(asm_cpu_default).
1340 (EXTRA_SPECS, SUBTARGET_EXTRA_SPECS, MULTILIB_DEFAULTS): Delete.
1341 * config/m68k/linux.h (TARGET_DEFAULT): Delete.
1342 (CPP_SPEC): Merge definitions. Do not handle __HAVE_68881__ here.
1343 * config/m68k/netbsd-elf.h (TARGET_OS_CPP_BUILTINS): Define
1344 __HAVE_FPU__ if TARGET_HARD_FLOAT.
1345 (TARGET_DEFAULT): Delete.
1346 (EXTRA_SPECS): Delete cpp_cpu_default_spec, cpp_cpu_spec,
1347 cpp_fpu_spec, asm_default_spec and netbsd_cpp_spec.
1348 (CPP_CPU_SPEC): Delete.
1349 (TARGET_VERSION): Merge definitions, using TARGET_68010 to pick
1350 the appropriate string.
1351 (CPP_CPU_DEFAULT_SPEC, ASM_DEFAULT_SPEC, CPP_FPU_SPEC): Delete.
1352 (CPP_SPEC): Define to NETBSD_CPP_SPEC.
1353 (ASM_SPEC): Don't use %(asm_default_spec).
1354 * config/m68k/m68k.c (TARGET_DEFAULT_TARGET_FLAGS): Remove
1355 TARGET_DEFAULT and add MASK_68881.
1356 * config/m68k/m68k.md: Remove mention of TARGET_DEFAULT from comments.
1358 2007-01-12 Richard Sandiford <richard@codesourcery.com>
1360 * config.gcc (m68010-*-netbsdelf*): Add MASK_68010.
1361 (m68k*-*-netbsdelf*, m68k*-*-openbsd*, m68k*-linux*): Add
1362 MASK_68010 alongside MASK_68020.
1363 * doc/invoke.texi: Document -m68010.
1364 * config/m68k/m68k.opt (m68010): New.
1365 * config/m68k/m68k.h (TARGET_CPU_CPP_BUILTINS): Define mc68010
1366 if TUNE_68010.
1367 (TUNE_68010): New macro.
1368 * config/m68k/m68k-none.h (M68K_CPU_m68k, M68K_CPU_m68010)
1369 (M68K_CPU_m68020, M68K_CPU_m68030, M68K_CPU_m68040)
1370 (M68K_CPU_m68332): Add MASK_68010.
1371 * config/m68k/linux.h (TARGET_DEFAULT): Add MASK_68010 to
1372 fallback definition.
1373 * config/m68k/netbsd-elf.h (CPP_CPU_SPEC): Remove now-redundant
1374 defines.
1375 * config/m68k/m68k.c (MASK_ALL_CPU_BITS): Add MASK_68010.
1376 (m68k_handle_option): Handle OPT_m68010. Add MASK_68010
1377 to all entries that use MASK_68020.
1378 (output_move_simode_const, output_move_himode, output_move_qimode)
1379 (output_move_stricthi, output_move_strictqi): Use TARGET_68010
1380 instead of TARGET_68020 to select clr behavior. Remove comment
1381 about there being no TARGET_68010.
1382 * config/m68k/m68k.md: Likewise throughout.
1384 2007-01-12 Julian Brown <julian@codesourcery.com>
1386 * config/m68k/m68k.h (TARGET_ISAB): New macro.
1387 * config/m68k/m68k.c: Use TARGET_ISAB rather than TARGET_CFV4.
1388 * config/m68k/m68k.md: Likewise.
1390 2007-01-12 Julian Brown <julian@codesourcery.com>
1392 * config/m68k/m68k.h (LEGITIMATE_INDEX_P, LEGITIMIZE_ADDRESS): Use
1393 TARGET_COLDFIRE_FPU instead of TARGET_CFV4E.
1395 2007-01-12 Julian Brown <julian@codesourcery.com>
1397 * config/m68k/m68k.h (TUNE_68040_60): New macro.
1398 * config/m68k/m68k.c (standard_68881_constant_p): Use it.
1399 * config/m68k/m68k.md: Likewise.
1401 2007-01-12 Julian Brown <julian@codesourcery.com>
1402 Richard Sandiford <richard@codesourcery.com>
1404 * config/m68k/m68k.h (TARGET_CPU_CPP_BUILTINS): Use TUNE_68030
1405 instead of TARGET_68030, TUNE_68040 instead of TARGET_68040,
1406 TUNE_68060 instead of TARGET_68060 and TUNE_CPU32 instead of
1407 TARGET_CPU32.
1408 (TARGET_CPU32): Rename to...
1409 (TUNE_CPU32): ...this.
1410 (TUNE_68000_10, TUNE_68030, TUNE_68040, TUNE_68060)
1411 (TUNE_CFV2): New macros.
1412 * config/m68k/netbsd-elf.h (LONG_DOUBLE_TYPE_SIZE): Simplify;
1413 remove conditions that are implied by TARGET_68020.
1414 * config/m68k/m68k.c (m68k_output_function_prologue): Use TUNE_68040
1415 instead of TARGET_68040 and TUNE_CPU32 instead of TARGET_CPU32.
1416 (m68k_output_function_epilogue): Likewise.
1417 (m68k_rtx_costs): Likewise. Use TUNE_68060 instead of TARGET_68060
1418 and TUNE_CFV2 instead of TARGET_5200. Use TUNE_68000_10 instead of
1419 "!TARGET_68020 && !TARGET_COLDFIRE" to choose between 68000 and
1420 non-68000 timings. Refactor multiplication and division costs.
1421 (output_addsi3): Use TUNE_68040 instead of TARGET_68040 and
1422 TUNE_CPU32 instead of TARGET_CPU32.
1423 (standard_68881_constant_p): Use TUNE_68040 instead of TARGET_68040
1424 and TUNE_68060 instead of TARGET_68060.
1425 * config/m68k/m68k.md: Use TUNE_68040 instead of TARGET_68040,
1426 TUNE_68060 instead of TARGET_68060, and TUNE_CPU32 instead of
1427 TARGET_CPU32.
1428 (movsi_const0): Use TUNE_68000_10 rather than "!TARGET_68020
1429 && !TARGET_COLDFIRE" to choose between moveq and clr.
1430 Likewise in the unnamed movsf pattern.
1431 (ashlsi_17_24, lshrsi_17_24): Guard with TUNE_68000_10 rather than
1432 "!TARGET_68020 && !TARGET_COLDFIRE". Likewise the unnamed
1433 ashiftrt pattern.
1435 2007-01-12 Richard Sandiford <richard@codesourcery.com>
1437 * config/m68k/m68k.h (TARGET_CPU_CPP_BUILTINS): Increase amount
1438 of tabbing before backslashes.
1440 2007-01-11 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
1442 * pa-linux.h (ASM_OUTPUT_INTERNAL_LABEL): Undefine.
1443 * pa.h (ASM_OUTPUT_LABEL): Output colon when using GAS.
1444 (ASM_OUTPUT_INTERNAL_LABEL): Define.
1446 2007-01-11 Zdenek Dvorak <dvorakz@suse.cz>
1448 * tree-ssa-loop-ivopts.c (extract_cond_operands): Split from
1449 find_interesting_uses_cond.
1450 (find_interesting_uses_cond): Use extract_cond_operands.
1451 (rewrite_use_compare): Use extract_cond_operands and
1452 force_gimple_operand_bsi. Do not call update_stmt.
1453 (determine_use_iv_cost_condition): Use extract_cond_operands.
1454 Return cheaper of using original bound and changing the exit bound.
1456 2007-01-11 Zdenek Dvorak <dvorakz@suse.cz>
1458 PR tree-optimization/29516
1459 * tree-ssa-address.c (tree_mem_ref_addr, add_to_parts,
1460 most_expensive_mult_to_index, addr_to_parts,
1461 create_mem_ref, maybe_fold_tmr): Make the type of
1462 fields of TARGET_MEM_REF sizetype.
1463 (move_fixed_address_to_symbol, move_pointer_to_base):
1464 New functions.
1465 * tree.def (TARGET_MEM_REF): Add comment on types of
1466 the operands.
1468 2007-01-11 Joseph Myers <joseph@codesourcery.com>
1470 * c-common.c (vector_types_convertible_p): Treat opaque types as
1471 always convertible if they have the same size, but not otherwise.
1473 2007-01-11 Steven Bosscher <steven@gcc.gnu.org>
1475 * ifcvt.c (struct noce_if_info): Add comments to the fields.
1476 Remove the b_unconditional field.
1477 (noce_try_sign_mask): Do not look at b_unconditional.
1478 (noce_process_if_block): Do not use merge_if_blocks. Update
1479 the CFG here. Do not set b_unconditional.
1480 (cond_move_process_if_block): Likewise.
1481 (find_cond_trap): Likewise.
1482 (check_cond_move_block): Require simple jump insns at the end
1483 of the basic block.
1485 2007-01-11 Jan Hubicka <jh@suse.cz>
1487 PR tree-optimization/1046
1488 * tree-tailcall.c (suitable_for_tail_call_opt_p): Use TREE_ADDRESSABLE
1489 when alias info is not ready.
1490 (pass_tail_recursion): Do not require aliasing.
1491 * tree-ssa-copyrename.c (pass_rename_ssa_cop): Likewise.
1492 * tree-ssa-ccp.c (pass_ccp, pass_fold_builtins): Likewise.
1493 * tree-ssa-copy.c (pass_copy_prop): Likewise.
1494 * tree-ssa-forwprop.c (pass_forwprop): Likewise.
1495 * tree-ssa-dce.c (pass_dce, pass_dce_loop, pass_cd_dce): Likewise.
1496 * passes.c (init_optimization_passes): Execute rename_ssa_copies,
1497 ccp, forwprop, copy_prop, merge_phi, copy_prop, dce and tail recursion
1498 before inlining.
1499 * tree-ssa-operands.c (add_virtual_operand, get_indirect_ref_operand):
1500 When aliasing is not build, mark statement as volatile.
1502 2007-01-11 Tom Tromey <tromey@redhat.com>
1504 PR preprocessor/15185, PR preprocessor/20989:
1505 * doc/cppopts.texi <-MT>: Update description of algorithm for
1506 computing default target.
1507 <-M, -MD>: Reword "basename" text.
1509 2007-01-11 Roger Sayle <roger@eyesopen.com>
1511 * builtins.c (expand_builtin_pow, expand_builtin_powi,
1512 fold_builtin_cabs, fold_builtin_sqrt, fold_builtin_trunc,
1513 fold_builtin_floor, fold_builtin_ceil, fold_builtin_round,
1514 fold_builtin_int_int_roundingfn, fold_builtin_bitop,
1515 fold_builtin_bswap, real_constp, fold_builtin_pow,
1516 fold_builtin_powi, fold_builtin_signbit, fold_builtin_copysign,
1517 do_mpfr_arg1, do_mpfr_arg2, do_mpfr_arg3, do_mpfr_sincos): Replace
1518 uses of the macro TREE_CONSTANT_OVERFLOW with TREE_OVERFLOW.
1519 * convert.c (convert_to_pointer): Likewise.
1520 * expr.c (highest_pow2_factor, expand_expr_real_1): Likewise.
1521 * fold-const.c (force_fit_type, fold_negate_expr, int_const_binop,
1522 const_binop, fold_convert_const_int_from_int,
1523 fold_convert_const_int_from_real,
1524 fold_convert_const_real_from_real, sign_bit_p,
1525 optimize_minmax_comparison, extract_muldiv_1, fold_div_compare,
1526 fold_sign_changed_comparison, fold_unary, fold_comparison,
1527 fold_binary, multiple_of_p, tree_Expr_non_zero_p,
1528 fold_negate_const, fold_abs_const, fold_not_const): Likewise.
1529 * print-tree.c (print_node_brief, print_node): Likewise.
1530 * stor-layout.c (place_field, layout_type): Likewise.
1531 * tree-chrec.c (keep_cast): Likewise.
1532 * tree.c (build_vector, build_real, build_real_from_int_cst,
1533 build_complex): Likewise.
1535 2007-01-11 Roger Sayle <roger@eyesopen.com>
1537 * tree.h (TREE_CONSTANT_OVERFLOW): Obsolete. For the time being,
1538 treat TREE_CONSTANT_OVERFLOW as a synonym of TREE_OVERFLOW.
1540 2007-01-11 Paolo Bonzini <bonzini@gnu.org>
1542 * configure.ac (strict1_warn): Rename to strict_warn.
1543 (WERROR, --enable-werror, symlink hacks, stage1_cflags,
1544 cc_set_by_configure, quoted_cc_set_by_configure,
1545 stage_prefix_set_by_configure, quoted_stage_prefix_set_by_configure,
1546 all_boot_languages, all_stagestuff): Remove.
1547 (target_list): Remove bootstrap targets.
1548 * Makefile.in (quickstrap): Unconditionally make a synonym of all.
1549 (BOOT_LANGUAGES, STAGE1_CFLAGS, STAGE1_CHECKING,
1550 REMAKEFLAGS, FLAGS_TO_PASS, PREPEND_DOTDOT_TO_RELATIVE_PATHS,
1551 SUBDIR_FLAGS_TO_PASS, WERROR_FLAGS, STRICT2_WARN, LANG_STAGESTUFF,
1552 VOL_FILES, POSTSTAGE1_FLAGS_TO_PASS, STAGE2_FLAGS_TO_PASS,
1553 STAGEPROFILE_FLAGS_TO_PASS, STAGEFEEDBACK_FLAGS_TO_PASS, stage1_build,
1554 stage1_copy, stage2_build, stage2_copy, stageprofile_build,
1555 stageprofile_copy, stage3_build, stage3_copy, stagefeedback_build,
1556 stagefeedback_copy, stage4_build, clean_s1, clean_sw, bootstrap,
1557 bootstrap-lean, bootstrap2, bootstrap2-lean, bootstrap3,
1558 bootstrap3-lean, bootstrap4, bootstrap4-lean, unstage1, unstage2,
1559 unstage3, unstage4, unstageprofile, unstagefeedback, restage, restage2,
1560 restage3, restage4, restageprofile, restagefeedback, bubbleestrap,
1561 cleanstrap, unstrap, restrap, *compare, *compare3, *compare4,
1562 *compare-lean, *compare3-lean, *compare4-lean, stage1-start, stage1,
1563 stage2-start, stage2, stage3-start, stage3, stage4-start, stage4,
1564 stageprofile-start, stageprofile, stagefeedback-start, stagefeedback,
1565 risky-stage1, risky-stage2, risky-stage3, risky-stage4): Remove.
1566 (ORDINARY_FLAGS_TO_PASS): Rename to FLAGS_TO_PASS.
1567 (STAGECOPYSTUFF, STAGEMOVESTUFF): Consolidate into MOSTLYCLEANFILES.
1568 (mostlyclean): Adjust.
1569 (clean, distclean): Don't mention bootstrap stuff.
1570 * configure: Regenerate.
1571 * ada/config-lang.in, cp/config-lang.in, forttran/config-lang.in,
1572 java/config-lang.in, objc/config-lang.in, objcp/config-lang.in,
1573 treelang/config-lang.in (stagestuff): Remove.
1574 * doc/sourcebuild.texi (stage1, stage2, stage3, stage4,
1575 stageprofile, stagefeedback, stagestuff): Remove mention.
1577 2007-01-11 Nick Clifton <nickc@redhat.com>
1579 * config/mcore/predicates.md (mcore_general_movesrc_operand):
1580 Accept CONSTs.
1581 (mcore_general_movdst_operand): Do not accept CONST_INTs.
1582 (mcore_arith_K_S_operand): Run the test for the S constraint not
1583 the test for the M constraint.
1584 (mcore_addsub_operand): Do not accept integer values that are
1585 larger than 32 bits.
1586 * config/mcore/mcore.md: Remove unused constraints from split.
1587 (andsi3): Use HOST_WIDE_INT instead of int to hold an INTVAL.
1588 (addsi3): Likewise.
1589 (allocate_stack): Likewise.
1590 * config/mcore/mcore.c (mcore_print_operand): Restrict output of P
1591 operands to 32 bits.
1592 (mcore_const_costs): Use HOST_WIDE_INT instead of int to hold an
1593 INTVAL.
1594 (mcore_and_cost, mcore_modify_comparison, const_ok_for_mcore,
1595 mcore_const_ok_for_inline, mcore_const_trick_uses_not,
1596 try_constant_tricks, mcore_num_ones, mcore_num_zeros,
1597 mcore_output_bclri, mcore_output_andn, output_inline_const,
1598 mcore_output_move, mcore_output_movedouble): Likewise.
1599 (mcore_output_cmov): Use CONST_OK_FOR_M and CONST_OK_FOR_N.
1600 (output_inline_const): Likewise.
1601 (output_inline_const): Fix format strings used in sprintf
1602 statements.
1603 * config/mcore/mcore-protos.h: Update prototypes for changed
1604 functions in mcore.c.
1605 * config/mcore/mcore.h (CONST_OK_FOR_I): Cast values to
1606 HOST_WIDE_INT and not int.
1607 (CONST_OK_FOR_J, CONST_OK_FOR_K, CONST_OK_FOR_L, CONST_OK_FOR_M,
1608 CONST_OK_FOR_N): Likewise.
1609 (LEGITIMATE_CONSTANT_P): Also check CONSTANT_P.
1610 (GO_IF_LEGITIMATE_INDEX): Use HOST_WIDE_INT instead of int to hold
1611 an INTVAL.
1613 2007-01-10 Jan Hubicka <jh@suse.cz>
1615 * tree-vrp.c (remove_range_assertions): Release defs.
1616 * tree-ssa-loop-ivopts.c (rmeove_statement): Likewise.
1617 * tree-ssa-dom.c (remove_stmt_or_phi): Likewise.
1619 2007-01-10 Paul Brook <paul@codesourcery.com>
1621 * config/arm/arm.c (arm_rtx_costs_1): Handle mutiply-subtract.
1622 * config/arm/arm.md (mulsi3subsi): New insn.
1624 2007-01-10 Zdenek Dvorak <dvorakz@suse.cz>
1626 * tree-ssa-loop-manip.c (tree_unroll_loop): Make it a wrapper over ...
1627 (tree_transform_and_unroll_loop): New.
1628 * tree-flow.h (transform_callback, tree_transform_and_unroll_loop):
1629 Declare.
1631 2007-01-10 Robert Kennedy <jimbob@google.com>
1633 * fold-const.c (fold_comparison): Fold comparisons like (x *
1634 1000 < 0) to (x < 0).
1636 2007-01-10 Ian Lance Taylor <iant@google.com>
1638 * tree-pretty-print.c (dump_generic_node): Print parentheses when
1639 operands have the same priority.
1641 2007-01-10 Tom Tromey <tromey@redhat.com>
1643 * fold-const.c (fold_truthop): Don't check can_use_bit_fields_p.
1644 (fold_binary): Likewise.
1645 * langhooks.c (lhd_can_use_bit_fields_p): Removed.
1646 * langhooks-def.h (lhd_can_use_bit_fields_p): Removed.
1647 (LANG_HOOKS_CAN_USE_BIT_FIELDS_P): Removed.
1648 (LANG_HOOKS_INITIALIZER): Remove LANG_HOOKS_CAN_USE_BIT_FIELDS_P.
1649 * langhooks.h (struct lang_hooks): Removed field
1650 'can_use_bit_fields_p'.
1652 2007-01-10 Ralf Corsépius <ralf.corsepius@rtems.org>
1654 * config/bfin/t-bfin, config/bfin/t-bfin-elf: Remove GCC_CFLAGS.
1656 2007-01-10 Razya Ladelsky <razya@il.ibm.com>
1658 * function.c (get_last_funcdef_no): New function.
1659 * function.h (get_last_funcdef_no): Declare.
1660 * tree-inline.c (initialize_cfun): Add initialization.
1661 (tree_function_versioning): Cleanup.
1663 2007-01-10 Jan Hubicka <jh@suse.cz>
1665 * tree-inline.c (setup_one_parameter): Do not propagate into abnormal
1666 PHIs.
1668 2007-01-10 Sa Liu <saliu@de.ibm.com>
1669 Ben Elliston <bje@au.ibm.com>
1671 * spu.h (STACK_SAVE_AREA): Use VOIDmode for SAVE_FUNCTION, SImode
1672 for SAVE_NONLOCAL and Pmode for any other save level.
1673 * spu-protos.h (spu_restore_stack_block): Declare.
1674 * spu.md (save_stack_block): Remove.
1675 (restore_stack_block): Call spu_restore_stack_block.
1676 * spu.c (spu_restore_stack_block): New function.
1677 (spu_expand_epilogue): Remove old comment.
1679 2007-01-09 Zdenek Dvorak <dvorakz@suse.cz>
1681 PR tree-optimization/30322
1682 * tree-ssa-loop-ivopts.c (fold_affine_expr, iv_value): Removed.
1683 (cand_value_at): Return the value as aff_tree.
1684 (may_eliminate_iv): Convert the bound from aff_tree to tree.
1685 * tree-affine.c (aff_combination_add_cst, aff_combination_add_product,
1686 aff_combination_mult): New functions.
1687 (aff_combination_add): Use aff_combination_add_cst.
1688 (aff_combination_convert): Allow conversions to a wider type.
1689 (tree_to_aff_combination): Handle BIT_NOT_EXPR.
1690 * tree-affine.h (aff_combination_mult): Declare.
1692 2007-01-09 Carlos O'Donell <carlos@codesourcery.com>
1694 * doc/tm.texi: Update documentation to reflect reality of exec
1695 and start file search behaviours. Update copyright year.
1696 * doc/invoke.texi: Explain how GCC_EXEC_PREFIX is used to find
1697 header file directories.
1699 2007-01-09 Uros Bizjak <ubizjak@gmail.com>
1701 * config/i386/i386.md (*sinxf2): Rename to *sinxf2_i387.
1702 (*cosxf2): Rename to cosxf2_i387.
1703 (*sindf2, *sinsf2): Extend operand 1 to XFmode. Macroize patterns
1704 using X87MODEF12 mode macro. Rename patterns to
1705 *sin_extend<mode>xf2_i387. Use SSE_FLOAT_MODE_P to disable patterns
1706 for SSE math.
1707 (*cosdf2, *cossf2): Ditto.
1708 (sincosdf3, sincossf3): Ditto. Rewrite corresponding splitters
1709 to match extended input operands.
1710 (sincos<mode>3): New expander.
1711 (*sinextendsfdf2, *cosextendsfdf2, *sincosextendsfdf3): Remove
1712 insn patterns and corresponding splitters.
1714 2007-01-09 Kaz Kojima <kkojima@gcc.gnu.org>
1716 * config/sh/t-linux (TARGET_LIBGCC2_CFLAGS): Delete.
1717 (SHLIB_MAPFILES, SHLIB_LINK, SHLIB_INSTALL): Likewise.
1719 2007-01-09 Nicolas Pitre <nico@cam.org>
1721 PR target/30173
1722 * arm/ieee754-df.S (Lad_s): Also test the low word of X for zero.
1724 2007-01-08 Geoffrey Keating <geoffk@apple.com>
1726 * target.h (struct gcc_target): New field library_rtti_comdat.
1727 * target-def.h (TARGET_CXX_LIBRARY_RTTI_COMDAT): New.
1728 (TARGET_CXX): Add TARGET_CXX_LIBRARY_RTTI_COMDAT.
1729 * doc/tm.texi (C++ ABI): Document TARGET_CXX_LIBRARY_RTTI_COMDAT.
1730 * config/darwin.h (TARGET_CXX_LIBRARY_RTTI_COMDAT): Define.
1732 2007-01-08 Geoffrey Keating <geoffk@apple.com>
1734 * doc/invoke.texi (Optimize Options): Correct description of -O0.
1736 2007-01-08 Richard Guenther <rguenther@suse.de>
1738 * tree.h (force_fit_type_double): Export.
1739 (force_fit_type): Remove.
1740 * fold-const.c (force_fit_type_double): New function.
1741 (force_fit_type): Remove.
1742 (int_const_binop): Use it.
1743 (fold_convert_const_int_from_int): Likewise.
1744 (fold_convert_const_int_from_real): Likewise.
1745 (fold_div_compare): Likewise.
1746 (fold_sign_changed_comparison): Likewise.
1747 (fold_unary): Likewise.
1748 (fold_negate_const): Likewise.
1749 (fold_abs_const): Likewise.
1750 (fold_not_const): Likewise.
1751 * c-common.c (shorten_compare): Use force_fit_type_double.
1752 * convert.c (convert_to_pointer): Likewise.
1754 2007-01-08 Richard Guenther <rguenther@suse.de>
1756 * tree.h (build_int_cst_wide_type): Export.
1757 * tree.c (build_int_cst_wide_type): New function.
1758 (build_int_cst_wide): Fix comment.
1759 * builtins.c (fold_builtin_object_size): Use build_int_cst
1760 to build -1 or 0 of the correct type. Use fit_double_type
1761 to check for overflow.
1762 * fold-const.c (optimize_bit_field_compare): Use build_int_cst_type
1763 to build the mask.
1764 (decode_field_reference): Likewise.
1765 (all_ones_mask_p): Likewise.
1766 (native_interpret_int): Use build_int_cst_wide_type.
1767 (fold_binary): Use build_int_cst_type to build an all-ones
1768 value.
1769 * stor-layout.c (set_sizetype): Use build_int_cst_wide_type.
1771 2007-01-08 Daniel Jacobowitz <dan@codesourcery.com>
1773 * config/pa/t-pa64 (libgcc_stub.a): Use $(T).
1775 2007-01-09 Ben Elliston <bje@au.ibm.com>
1777 * genautomata.c (STATS_OPTION): New option.
1778 (stats_flag): New flag.
1779 (gen_automata_option): Handle it.
1780 (initiate_automaton_gen): Ditto.
1781 (write_automata): Output statistics only if stats_flag is
1782 set. Likewise, output time statistics only if time_flag is set.
1783 * doc/md.texi (Processor pipeline description): Document new flag.
1785 2007-01-08 Richard Guenther <rguenther@suse.de>
1787 * builtins.c (fold_builtin_int_roundingfn): Use fit_double_type.
1788 * tree.c (build_int_cst_type): Likewise.
1789 (size_in_bytes): Don't call force_fit_type on the result.
1790 (int_fits_type_p): Use fit_double_type.
1791 * fold-const.c (fit_double_type): New function.
1792 (force_fit_type): Use it.
1793 * tree.h (fit_double_type): Export.
1795 2007-01-08 Jan Hubicka <jh@suse.cz>
1797 * tree-vectorizer.c (gate_increase_alignment): Fix return type.
1798 * ipa.c (function_and_variable_visibility): Fix return type.
1800 2007-01-08 Richard Guenther <rguenther@suse.de>
1802 * tree-ssa-ccp.c (maybe_fold_offset_to_array_ref): Use type
1803 of offset to build the index.
1804 * tree-pretty-print.c (dump_generic_node): Don't build negated
1805 const just for printing.
1806 * c-pretty-print.c (pp_c_integer_constant): Likewise.
1807 * builtins.c (fold_builtin_int_roundingfn): Check if result
1808 fits the type by using force_fit_type and comparing the result.
1809 * predict.c (predict_loops): Use compare_tree_int for comparison.
1810 * tree.c (build_int_cst): Fall back to integer_type_node for
1811 NULL_TREE type.
1812 (build_int_cst_wide): Assert type is non-null.
1814 2007-01-08 Roberto Costa <roberto.costa@st.com>
1816 * tree-vrp.c (extract_range_from_cond_expr): New.
1817 (extract_range_from_expr): Handle COND_EXPR nodes used as expressions.
1818 * tree-ssa-ccp.c (get_maxval_strlen): Handle COND_EXPR nodes used
1819 as expressions.
1820 (fold_stmt): Bug fix, avoid infinite recursion when folding COND_EXPRs.
1821 * tree-ssa-forwprop.c (simplify_cond, forward_propagate_into_cond,
1822 tree_ssa_forward_propagate_single_use_vars): Handle COND_EXPR nodes
1823 used as expressions.
1824 * tree-object-size.c (cond_expr_object_size): New.
1825 (collect_object_sizes_for): Handle COND_EXPR nodes used as expressions.
1827 2007-01-08 Jan Hubicka <jh@suse.cz>
1829 * tree-ssa-forwprop.c (forward_propagate_into_cond,
1830 tree_ssa_forward_propagate_single_use_va): Release defs of propagated
1831 statement.
1833 2007-01-08 Richard Guenther <rguenther@suse.de>
1835 PR tree-optimization/23603
1836 * tree-vrp.c (set_value_range_to_truthvalue): New function.
1837 (extract_range_from_binary): Fall back to truthvalue instead of
1838 varying for TRUTH_*_EXPR.
1839 (extract_range_from_comparison): Fall back to truthvalue instead of
1840 varying.
1841 (vrp_visit_phi_node): Don't adjust new range bounds to +INF/-INF
1842 if all visited PHI values were constant.
1844 2007-01-08 Jan Hubicka <jh@suse.cz>
1846 * cgraphunit.c (cgraph_process_new_functions): Reset reachable flag.
1847 (cgraph_analyze_function): break out from ...
1848 (cgraph_finalize_compilation_unit): ... here.
1849 (cgraph_expand_function): Remove forgoten commented out line.
1850 (cgraph_optimize): Analyze functions.
1852 2007-01-08 Jan Hubicka <jh@suse.cz>
1854 * tree-pas.h (TODO_remove_function): New flag.
1855 (TODO_update*): Renumber.
1856 (pass_ipa_increase_alignment,
1857 pass_ipa_function_and_variable_visibility): New passes.
1858 * cgraphunit.c (cgraph_increase_alignment): Move to tree-vectorizer.c
1859 (cgraph_function_and_variable_visibility): Move to ipa.c
1860 (cgraph_optimize): Don't call cgraph_function_and_variable_visibility,
1861 cgraph_increase_alignment.
1862 * ipa-inline.c (cgraph_decide_inlining): Don't push timevar.
1863 (cgraph_decide_inlining_incrementally): Push TV_INTEGRATION before
1864 calling tree-inline.
1865 (cgraph_early_inlining): Do not call cgraph_remove_unreachable_nodes.
1866 (pass_ipa_inline, pass_early_ipa_inlining): Set TODO_remove_functions
1867 * tree-vectorizer.c (increase_alignment): Move here from cgraphunit.c
1868 (gate_increase_alignment): New function.
1869 (pass_ipa_increase_alignment): New pass.
1870 * ipa.c: Inline tree-pass.h and timevar.h
1871 (function_and_variable_visibility): Move here from cgraphunit.c
1872 * tree-optimize.c (pass_early_local_passes): Add TODO_remove_functions.
1873 * passes.c (init_optimization_passes): Add the two new passes.
1874 (execute_todo): Handle cgraph_remove_functions.
1876 2007-01-08 Nick Clifton <nickc@redhat.com>
1878 * config/frv/predicates.md (reg_or_0_operand): Accept
1879 CONST_DOUBLEs.
1881 2007-01-08 Ralf Corsépius <ralf.corsepius@rtems.org>
1883 * config/bfin/rtems.h, config/bfin/t-rtems: New.
1884 * config.gcc: Add bfin*-rtems*.
1886 2007-01-08 Mark Shinwell <shinwell@codesourcery.com>
1888 * c.opt: Add -flax-vector-conversions.
1889 * c-typeck.c (convert_for_assignment): Pass flag to
1890 vector_types_convertible_p to allow emission of note.
1891 (digest_init): Likewise.
1892 * c-opts.c: Handle -flax-vector-conversions.
1893 * c-common.c (flag_lax_vector_conversions): New.
1894 (vector_types_convertible_p): Unless -flax-vector conversions
1895 has been passed, disallow conversions between vectors with
1896 differing numbers of subparts and/or element types. If such
1897 a conversion is disallowed, possibly emit a note on the first
1898 occasion only to inform the user of -flax-vector-conversions.
1899 The new last argument specifies this.
1900 * c-common.h (flag_lax_vector_conversions): New.
1901 (vector_types_convertible_p): Add extra argument.
1902 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use
1903 char_type_node for V*QI type vectors.
1904 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
1905 Update to satisfy new typechecking rules.
1906 * config/rs6000/altivec.h (vec_cmple): Use vec_cmpge, for both
1907 C and C++ variants.
1908 * doc/invoke.texi (C Dialect Options): Document
1909 -flax-vector-conversions.
1911 2007-01-08 Mark Shinwell <shinwell@codesourcery.com>
1913 PR tree-optimization/29877
1914 * tree-ssa-ter.c (is_replaceable_p): Deem assignments with
1915 a register variable on the RHS to not be replaceable.
1917 2007-01-08 Chen Liqin <liqin@sunnorth.com.cn>
1918 * config/score/t-score-elf (MULTILIB_OPTIONS): Change.
1919 * config/score/predicates.md (const_uimm5, sr0_operand, const_simm12,
1920 const_simm15, const_pow2, const_npow2): Added.
1921 * config/score/misc.md (insv, extv, extzv, movmemsi,
1922 move_lbu_a/b, mov_lhu_a/b etc): Added and fix some bug.
1923 * config/score/score.c (score_address_cost, score_select_cc_mode):
1924 Added.
1925 Change CONST_OK_FOR_LETTER_P/EXTRA_CONSTRAINT define.
1926 Update score_rtx_costs for MACRO TARGET_RTX_COSTS.
1927 Update score_print_operand.
1928 * config/score/score.h (DATA_ALIGNMENT, SELECT_CC_MODE): Added.
1929 Adjust register allocate order and update some macro define.
1930 * config/score/score-mdaux.c (mdx_unaligned_load, mdx_unsigned_store,
1931 mdx_block_move_straight, mdx_block_move_loop_head,
1932 mdx_block_move_loop_body, mdx_block_move_loop_foot, mdx_block_move_loop,
1933 mdx_block_move): Added.
1934 (mdx_movsicc, mdp_select_add_imm, mdp_select, mds_zero_extract_andi,
1935 mdp_limm): Updated and fix some bug and typo.
1936 * config/score/score.md (movqi/hi/si, add/sub/zero/ext): Updated.
1937 (movsf, movdf, doloop_end): Added.
1939 2007-01-08 Kazu Hirata <kazu@codesourcery.com>
1941 * config/arm/arm.c, config/arm/arm.h, config/arm/arm.md,
1942 config/arm/thumb2.md: Fix comment typos.
1943 * doc/extend.texi: Fix a typo.
1945 2007-01-07 Eric Christopher <echristo@apple.com>
1947 * configure.ac: Check for __stack_chk_fail for darwin.
1948 * configure: Regenerate.
1950 2007-01-07 Richard Guenther <rguenther@suse.de>
1952 * tree-vrp.c (extract_range_from_assert): CSE calls to
1953 compare_values where possible.
1954 (extract_range_from_unary_expr): Likewise.
1956 2007-01-07 Anatoly Sokolov <aesok@post.ru>
1958 * config/avr/avr-protos.h (call_insn_operand): Delete prototype.
1959 * config/avr/avr.c (call_insn_operand): Delete function.
1960 * config/avr/avr.md (*pushqi, *pushhi, *pushsi, *pushsf): Use REG_SP
1961 instead of register number. Use predicates.
1962 * config/avr/predicates.md (const0_operand, reg_or_0_operand,
1963 call_insn_operand): Add.
1965 2007-01-06 Jan Hubicka <jh@suse.cz>
1967 * tree-pass.h (pass_build_cgraph_edges): Declare.
1968 * cgraphunit.c (record_refernece): Move to cgraphbuild.c
1969 (visited_nodes): Remove.
1970 (cgraph_create_edges): Move to cgraphbuild.c; rename to
1971 build_cgrpah_edges; make visited_nodes local.
1972 (cgraph_process_new_functions): DO not call initialize_inline_failed.
1973 (record_references_in_initializer): Move to cgraphbuild.c
1974 (initialize_inline_failed, rebuild_cgraph_edges,
1975 pass_rebuild_cgraph_edges): Move to cgraphbuild.c.
1976 (verify_cgraph_node): Make visited_nodes local.
1977 (cgraph_analyze_function): Do not call cgraph_create_edges and
1978 initialize_inline_failed.
1979 (cgraph_expand_function): Do not call cgraph_lower_function;
1980 assert that function is already lowered.
1981 * Makefile.in (cgraphbuild.o): New.
1982 * passes.c (init_optimization_passes): Add pass_build_cgraph_edges
1983 at the end of lowering passes.
1985 2007-01-06 Steven Bosscher <steven@gcc.gnu.org>
1987 * ifcvt.c (cond_move_convert_if_block): New function, code
1988 factored out from...
1989 (cond_move_process_if_block): ...here. Call the new function
1990 on the THEN and ELSE blocks.
1991 (merge_if_block): Do not copy global_live_at_end, merge_blocks
1992 already takes care of this.
1994 2007-01-05 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
1996 PR c/19978
1997 * tree.h (TREE_OVERFLOW_P): New.
1998 * c-typeck.c (parser_build_unary_op): Warn only if result
1999 overflowed and operands did not.
2000 (parser_build_binary_op): Likewise.
2001 (convert_for_assignment): Remove redundant overflow_warning.
2002 * c-common.c (overflow_warning): Don't check or set TREE_OVERFLOW.
2004 2007-01-05 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
2006 * c-typeck.c (store_init_value): Split over two lines to follow
2007 the GNU coding style.
2009 2007-01-05 Benjamin Kosnik <bkoz@redhat.com>
2011 * c-cppbuiltin.c (c_cpp_builtins): __GXX_EXPERIMENTAL_CPP0X__ to
2012 __GXX_EXPERIMENTAL_CXX0X__.
2013 * doc/cpp.texi: Same.
2015 2007-01-05 Richard Guenther <rguenther@suse.de>
2017 PR middle-end/27826
2018 * tree.c (get_narrower): Do not construct COMPONENT_REFs
2019 with mismatched types. Instead explicitly build a
2020 conversion NOP_EXPR.
2022 2007-01-05 Ian Lance Taylor <iant@google.com>
2024 * c-common.c (decl_with_nonnull_addr_p): New function.
2025 (c_common_truthvalue_conversion): Call it.
2026 * c-typeck.c (build_binary_op): Likewise.
2027 * c-common.h (decl_with_nonnull_addr_p): Declare.
2029 2007-01-05 Jakub Jelinek <jakub@redhat.com>
2031 PR c/30360
2032 * libgcc2.c (__divdc3): Compare c and d against 0.0 instead of
2033 denom against 0.0.
2035 2007-01-05 Joel Brobecker <brobecker@adacore.com>
2037 * doc/install.texi (Final install): Document the fact that
2038 the GNAT runtime should not be stripped.
2040 2007-01-04 Jan Hubicka <jh@suse.cz>
2042 * tree-inline.c (fold_marked_statements): Update operand caches
2043 and EH after folding
2045 2007-01-04 Ian Lance Taylor <iant@google.com>
2047 * c-common.c (check_function_nonnull): Whitespace fix.
2049 2007-01-04 Jan Hubicka <jh@suse.cz>
2051 * tree-optimize.c (execute_fixup_cfg): Correct previously mistakely
2052 comitted older version of patch.
2053 (pass_fixup_cfg): Add TODOs to verify flow and statements, dump
2054 function, celanup cfg and collect garbage.
2056 2007-01-04 Mike Stump <mrs@apple.com>
2058 * Makefile.in (mostlyclean): Don't remove libgcc anymore.
2059 (clean): Likewise.
2061 2007-01-04 Eric Christopher <echristo@apple.com>
2063 * libgcc2.c (__bswapsi2): Use SItype.
2064 (__bswapdi2): Use DItype.
2065 * libgcc2.h: Update for above.
2067 2007-01-04 Paul Brook <paul@codesourcery.com>
2069 * config/arm/arm.md (arm_mulsi3, thumb_mulsi3, mulsi3_compare0,
2070 mulsi_compare0_scratch, mulsi3addsi, mulsi3addsi_compare0,
2071 mulsi3addsi_compare0_scratch, mulsidi3adddi, mulsidi3,
2072 umulsidi3, umulsidi3adddi, smulsi3_highpart,
2073 umulsi3_highpart): Make conditional on !arm_arch6.
2074 (arm_mulsi3_v6, thumb_mulsi3_v6, mulsi3_compare0_v6,
2075 mulsi_compare0_scratch_v6, mulsi3addsi_v6, mulsi3addsi_compare0_v6,
2076 mulsi3addsi_compare0_scratch_v6, mulsidi3adddi_v6, mulsidi3_v6,
2077 umulsidi3_v6, umulsidi3adddi_v6, smulsi3_highpart_v6,
2078 umulsi3_highpart_v6): New insns.
2080 2007-01-04 Roger Sayle <roger@eyesopen.com>
2082 * fold-const.c (fold_convert): When casting an expression to void,
2083 fold_ignored_result may discover a GIMPLE_MODIFY_STMT which doesn't
2084 have a type. Instead of attempting to build a NOP_EXPR, return
2085 these "special" trees directly.
2087 2007-01-04 Joseph Myers <joseph@codesourcery.com>
2089 * config/rs6000/rs6000.c (rs6000_rtx_costs): Make adjustment for
2090 MULT inside MINUS as either argument. Use rs6000_cost->dmul -
2091 rs6000_cost->fp not 0 as adjustment for outer NEG.
2093 2007-01-04 Jan Hubicka <jh@suse.cz>
2095 * cgraph.c (cgraph_release_function_body): New function.
2096 (cgraph_remove_node): Use it.
2097 * cgraph.h (cgraph_release_function_body): Declare.
2098 * cgraphunit.c (cgraph_expand_function): Use it.
2099 * ipa.c (cgraph_remove_unreahchable_nodes): Use it.
2100 * tree-ssa.c (delete_tree_ssa): Allow to be called before aliasing
2101 is initialized and while compilation of other function is running.
2102 * tree-optimize.c (execute_free_cfg_annotations): Move code to clear
2103 statement CFG annotations from here to ...
2104 * tree-cfg.c (delete_tree_cfg_annotations): ... here.
2106 2007-01-04 Zdenek Dvorak <dvorakz@suse.cz>
2108 * cfgloop.h (enum li_flags): Make the constants powers of two.
2110 2007-01-04 Jan Hubicka <jh@suse.cz>
2112 * tree-inline.c (copy_bb): Insert new statements to statements_to_fold
2113 set.
2114 (fold_marked_statements): New function.
2115 (optimize_inline_calls, tree_function_versioning): Fold new statements.
2116 * tree-inline.h (copy_body_data): Add statements_to_fold.
2118 2007-01-03 Daniel Jacobowitz <dan@codesourcery.com>
2120 * config.gcc: Mention libgcc/config.host.
2121 * Makefile.in: Update comments mentioning libgcc.
2122 (LIBGCC, INSTALL_LIBGCC, GCC_PARTS, mklibgcc): Delete.
2123 (all.cross, start.encap, rest.encap, rest.cross): Update
2124 dependencies for libgcc move.
2125 (libgcc.mk, LIBGCC_DEPS, libgcov.a, libgcc.a, stmp-multilib)
2126 (clean-target, clean-target-libgcc): Delete.
2127 (srcdirify, GCC_EXTRA_PARTS): New macros.
2128 (libgcc-support, libgcc.mvars): New rules.
2129 (distclean): Remove mention of mklibgcc.
2130 (install): Don't reference INSTALL_LIBGCC.
2131 (install-common): Don't reference EXTRA_PARTS.
2132 (install-libgcc, install-multilib): Delete rules.
2133 * mklibgcc.in: Delete file.
2134 * doc/configfiles.texi: Don't mention mklibgcc.
2136 * config/i386/t-darwin (SHLIB_VERPFX): Delete (moved to libgcc).
2137 * config/i386/t-darwin64 (SHLIB_VERPFX): Likewise.
2138 * config/rs6000/t-darwin (SHLIB_VERPFX): Likewise.
2139 * config/rs6000/t-ppccomm (TARGET_LIBGCC2_CFLAGS, SHLIB_MAPFILES)
2140 (mklibgcc, ldblspecs): Likewise.
2142 * config/i386/t-nwld (libgcc.def, libc.def, libpcre.def)
2143 (posixpre.def): Use $(T).
2144 (SHLIB_EXT, SHLIB_NAME, SHLIB_SLIBDIR_QUAL, SHLIB_DEF, SHLIB_MAP)
2145 (SHLIB_SRC, SHLIB_INSTALL): Delete.
2146 (SHLIB_LINK): Make dummy.
2147 * config/t-slibgcc-darwin: Delete contents except for dummy SHLIB_LINK.
2149 * config/frv/t-linux (EXTRA_MULTILIB_PARTS): Clear.
2151 * config/alpha/t-crtfm: Use $(T) in rules for EXTRA_PARTS.
2152 * config/alpha/t-vms, config/alpha/t-vms64, config/fr30/t-fr30,
2153 config/i386/t-rtems-i386, config/ia64/t-ia64, config/rs6000/t-beos,
2154 config/rs6000/t-newas, config/sparc/t-elf: Likewise.
2156 * configure.ac (all_outputs): Remove mklibgcc.
2157 * configure: Regenerated.
2159 2007-01-03 Josh Conner <jconner@apple.com>
2161 PR middle-end/29683
2162 * calls.c (compute_argument_addresses): Set stack and stack_slot
2163 for partial args, too.
2164 (store_one_arg): Use locate.size.constant for the size when
2165 generating a save_area.
2167 2007-01-03 Robert Kennedy <jimbob@google.com>
2169 * tree-cfg.c (tree_merge_blocks): Release SSA_NAME phi results
2170 whose definitions are deleted due to basic block merging.
2172 2007-01-03 Paul Brook <paul@codesourcery.com>
2174 PR target/16634
2175 * config/arm/arm.c (output_return_instruction): Pop PC in interrupt
2176 functions.
2177 (use_return_insn): Return 0 for Thumb interrupt functions.
2178 (print_multi_reg): Add rfe argument for IRQ returns.
2179 (arm_output_epilogue): Pop interrupt return address directly into PC.
2180 (arm_expand_prologue): Only adjust IRQ return address in Arm mode.
2182 2007-01-03 Paul Brook <paul@codesourcery.com>
2184 Merge from sourcerygxx-4_1.
2185 * config/arm/thumb2.md: New file.
2186 * config/arm/elf.h (JUMP_TABLES_IN_TEXT_SECTION): Return True for
2187 Thumb-2.
2188 * config/arm/coff.h (JUMP_TABLES_IN_TEXT_SECTION): Ditto.
2189 * config/arm/aout.h (ASM_OUTPUT_ADDR_VEC_ELT): Add !Thumb-2 assertion.
2190 (ASM_OUTPUT_ADDR_DIFF_ELT): Output Thumb-2 jump tables.
2191 * config/arm/aof.h (ASM_OUTPUT_ADDR_DIFF_ELT): Output Thumb-2 jump
2192 tables.
2193 (ASM_OUTPUT_ADDR_VEC_ELT): Add !Thumb-2 assertion.
2194 * config/arm/ieee754-df.S: Use macros for Thumb-2/Unified asm
2195 comptibility.
2196 * config/arm/ieee754-sf.S: Ditto.
2197 * config/arm/arm.c (thumb_base_register_rtx_p): Rename...
2198 (thumb1_base_register_rtx_p): ... to this.
2199 (thumb_index_register_rtx_p): Rename...
2200 (thumb1_index_register_rtx_p): ... to this.
2201 (thumb_output_function_prologue): Rename...
2202 (thumb1_output_function_prologue): ... to this.
2203 (thumb_legitimate_address_p): Rename...
2204 (thumb1_legitimate_address_p): ... to this.
2205 (thumb_rtx_costs): Rename...
2206 (thumb1_rtx_costs): ... to this.
2207 (thumb_compute_save_reg_mask): Rename...
2208 (thumb1_compute_save_reg_mask): ... to this.
2209 (thumb_final_prescan_insn): Rename...
2210 (thumb1_final_prescan_insn): ... to this.
2211 (thumb_expand_epilogue): Rename...
2212 (thumb1_expand_epilogue): ... to this.
2213 (arm_unwind_emit_stm): Rename...
2214 (arm_unwind_emit_sequence): ... to this.
2215 (thumb2_legitimate_index_p, thumb2_legitimate_address_p,
2216 thumb1_compute_save_reg_mask, arm_dwarf_handle_frame_unspec,
2217 thumb2_index_mul_operand, output_move_vfp, arm_shift_nmem,
2218 arm_save_coproc_regs, thumb_set_frame_pointer, arm_print_condition,
2219 thumb2_final_prescan_insn, thumb2_asm_output_opcode, arm_output_shift,
2220 thumb2_output_casesi): New functions.
2221 (TARGET_DWARF_HANDLE_FRAME_UNSPEC): Define.
2222 (FL_THUMB2, FL_NOTM, FL_DIV, FL_FOR_ARCH6T2, FL_FOR_ARCH7,
2223 FL_FOR_ARCH7A, FL_FOR_ARCH7R, FL_FOR_ARCH7M, ARM_LSL_NAME,
2224 THUMB2_WORK_REGS): Define.
2225 (arm_arch_notm, arm_arch_thumb2, arm_arch_hwdiv, arm_condexec_count,
2226 arm_condexec_mask, arm_condexec_masklen)): New variables.
2227 (all_architectures): Add armv6t2, armv7, armv7a, armv7r and armv7m.
2228 (arm_override_options): Check new CPU capabilities.
2229 Set new architecture flag variables.
2230 (arm_isr_value): Handle v7m interrupt functions.
2231 (user_return_insn): Return 0 for v7m interrupt functions. Handle
2232 Thumb-2.
2233 (const_ok_for_arm): Handle Thumb-2 constants.
2234 (arm_gen_constant): Ditto. Use movw when available.
2235 (arm_function_ok_for_sibcall): Return false for v7m interrupt
2236 functions.
2237 (legitimize_pic_address, arm_call_tls_get_addr): Handle Thumb-2.
2238 (thumb_find_work_register, arm_load_pic_register,
2239 legitimize_tls_address, arm_address_cost, load_multiple_sequence,
2240 emit_ldm_seq, emit_stm_seq, arm_select_cc_mode, get_jump_table_size,
2241 print_multi_reg, output_mov_long_double_fpa_from_arm,
2242 output_mov_long_double_arm_from_fpa, output_mov_double_fpa_from_arm,
2243 output_mov_double_fpa_from_arm, output_move_double,
2244 arm_compute_save_reg_mask, arm_compute_save_reg0_reg12_mask,
2245 output_return_instruction, arm_output_function_prologue,
2246 arm_output_epilogue, arm_get_frame_offsets, arm_regno_class,
2247 arm_output_mi_thunk, thumb_set_return_address): Ditto.
2248 (arm_expand_prologue): Handle Thumb-2. Use arm_save_coproc_regs.
2249 (arm_coproc_mem_operand): Allow POST_INC/PRE_DEC.
2250 (arithmetic_instr, shift_op): Use arm_shift_nmem.
2251 (arm_print_operand): Use arm_print_condition. Handle '(', ')', '.',
2252 '!' and 'L'.
2253 (arm_final_prescan_insn): Use extract_constrain_insn_cached.
2254 (thumb_expand_prologue): Use thumb_set_frame_pointer.
2255 (arm_file_start): Output directive for unified syntax.
2256 (arm_unwind_emit_set): Handle stack alignment instruction.
2257 * config/arm/lib1funcs.asm: Remove default for __ARM_ARCH__.
2258 Add v6t2, v7, v7a, v7r and v7m.
2259 (RETLDM): Add Thumb-2 code.
2260 (do_it, shift1, do_push, do_pop, COND, THUMB_SYNTAX): New macros.
2261 * config/arm/arm.h (TARGET_CPU_CPP_BUILTINS): Define __thumb2__.
2262 (TARGET_THUMB1, TARGET_32BIT, TARGET_THUMB2, TARGET_DSP_MULTIPLY,
2263 TARGET_INT_SIMD, TARGET_UNIFIED_ASM, ARM_FT_STACKALIGN, IS_STACKALIGN,
2264 THUMB2_TRAMPOLINE_TEMPLATE, TRAMPOLINE_ADJUST_ADDRESS,
2265 ASM_OUTPUT_OPCODE, THUMB2_GO_IF_LEGITIMATE_ADDRESS,
2266 THUMB2_LEGITIMIZE_ADDRESS, CASE_VECTOR_PC_RELATIVE,
2267 CASE_VECTOR_SHORTEN_MODE, ADDR_VEC_ALIGN, ASM_OUTPUT_CASE_END,
2268 ADJUST_INSN_LENGTH): Define.
2269 (TARGET_REALLY_IWMMXT, TARGET_IWMMXT_ABI, CONDITIONAL_REGISTER_USAGE,
2270 STATIC_CHAIN_REGNUM, HARD_REGNO_NREGS, INDEX_REG_CLASS,
2271 BASE_REG_CLASS, MODE_BASE_REG_CLASS, SMALL_REGISTER_CLASSES,
2272 PREFERRED_RELOAD_CLASS, SECONDARY_OUTPUT_RELOAD_CLASS,
2273 SECONDARY_INPUT_RELOAD_CLASS, LIBCALL_VALUE, FUNCTION_VALUE_REGNO_P,
2274 TRAMPOLINE_SIZE, INITIALIZE_TRAMPOLINE, HAVE_PRE_INCREMENT,
2275 HAVE_POST_DECREMENT, HAVE_PRE_DECREMENT, HAVE_PRE_MODIFY_DISP,
2276 HAVE_POST_MODIFY_DISP, HAVE_PRE_MODIFY_REG, HAVE_POST_MODIFY_REG,
2277 REGNO_MODE_OK_FOR_BASE_P, LEGITIMATE_CONSTANT_P,
2278 REG_MODE_OK_FOR_BASE_P, REG_OK_FOR_INDEX_P, GO_IF_LEGITIMATE_ADDRESS,
2279 LEGITIMIZE_ADDRESS, THUMB2_LEGITIMIZE_ADDRESS,
2280 GO_IF_MODE_DEPENDENT_ADDRESS, MEMORY_MOVE_COST, BRANCH_COST,
2281 ASM_APP_OFF, ASM_OUTPUT_CASE_LABEL, ARM_DECLARE_FUNCTION_NAME,
2282 FINAL_PRESCAN_INSN, PRINT_OPERAND_PUNCT_VALID_P,
2283 PRINT_OPERAND_ADDRESS): Adjust for Thumb-2.
2284 (arm_arch_notm, arm_arch_thumb2, arm_arch_hwdiv): New declarations.
2285 * config/arm/arm-cores.def: Add arm1156t2-s, cortex-a8, cortex-r4 and
2286 cortex-m3.
2287 * config/arm/arm-tune.md: Regenerate.
2288 * config/arm/arm-protos.h: Update prototypes.
2289 * config/arm/vfp.md: Enable patterns for Thumb-2.
2290 (arm_movsi_vfp): Add movw alternative. Use output_move_vfp.
2291 (arm_movdi_vfp, movsf_vfp, movdf_vfp): Use output_move_vfp.
2292 (thumb2_movsi_vfp, thumb2_movdi_vfp, thumb2_movsf_vfp,
2293 thumb2_movdf_vfp, thumb2_movsfcc_vfp, thumb2_movdfcc_vfp): New.
2294 * config/arm/libunwind.S: Add Thumb-2 code.
2295 * config/arm/constraints.md: Update include Thumb-2.
2296 * config/arm/ieee754-sf.S: Add Thumb-2/Unified asm support.
2297 * config/arm/ieee754-df.S: Ditto.
2298 * config/arm/bpabi.S: Ditto.
2299 * config/arm/t-arm (MD_INCLUDES): Add thumb2.md.
2300 * config/arm/predicates.md (low_register_operand,
2301 low_reg_or_int_operand, thumb_16bit_operator): New.
2302 (thumb_cmp_operand, thumb_cmpneg_operand): Rename...
2303 (thumb1_cmp_operand, thumb1_cmpneg_operand): ... to this.
2304 * config/arm/t-arm-elf: Add armv7 multilib.
2305 * config/arm/arm.md: Update patterns for Thumb-2 and Unified asm.
2306 Include thumb2.md.
2307 (UNSPEC_STACK_ALIGN, ce_count): New.
2308 (arm_incscc, arm_decscc, arm_umaxsi3, arm_uminsi3,
2309 arm_zero_extendsidi2, arm_zero_extendqidi2): New
2310 insns/expanders.
2311 * config/arm/fpa.md: Update patterns for Thumb-2 and Unified asm.
2312 (thumb2_movsf_fpa, thumb2_movdf_fpa, thumb2_movxf_fpa,
2313 thumb2_movsfcc_fpa, thumb2_movdfcc_fpa): New insns.
2314 * config/arm/cirrus.md: Update patterns for Thumb-2 and Unified asm.
2315 (cirrus_thumb2_movdi, cirrus_thumb2_movsi_insn,
2316 thumb2_cirrus_movsf_hard_insn, thumb2_cirrus_movdf_hard_insn): New
2317 insns.
2318 * doc/extend.texi: Document ARMv7-M interrupt functions.
2319 * doc/invoke.texi: Document Thumb-2 new cores+architectures.
2321 2007-01-03 Jakub Jelinek <jakub@redhat.com>
2323 * unwind-dw2.c (SIGNAL_FRAME_BIT, EXTENDED_CONTEXT_BIT): Define.
2324 (struct _Unwind_Context): Rename args_size to flags, remove
2325 signal_frame field, add a new args_size field and version field.
2326 (_Unwind_IsSignalFrame, _Unwind_SetSignalFrame,
2327 _Unwind_IsExtendedContext): New inline functions.
2328 (_Unwind_GetGR, _Unwind_SetGR, _Unwind_GetGRPtr, _Unwind_SetGRPtr):
2329 Assume by_value array is only present if _Unwind_IsExtendedContext.
2330 (_Unwind_GetIPInfo, execute_cfa_program, uw_frame_state_for): Use
2331 _Unwind_IsSignalFrame.
2332 (__frame_state_for): Initialize context.flags to EXTENDED_CONTEXT_BIT.
2333 (uw_update_context_1): Use _Unwind_SetSignalFrame.
2334 (uw_init_context_1): Initialize context->flags to
2335 EXTENDED_CONTEXT_BIT.
2336 * config/rs6000/linux-unwind.h (frob_update_context): Use
2337 _Unwind_SetSignalFrame.
2339 2007-01-03 Andrew Pinski <andrew_pinski@playstation.sony.com>
2341 PR middle-end/30353
2342 * gimplify.c (gimplify_modify_expr_complex_part): Move below
2343 tree_to_gimple_tuple. Call tree_to_gimple_tuple when we need
2344 the value.
2346 2007-01-03 Kazu Hirata <kazu@codesourcery.com>
2348 * config/i386/i386.h (NON_STACK_REG_P, REGNO_OK_FOR_SIREG_P,
2349 REGNO_OK_FOR_DIREG_P, REWRITE_ADDRESS, ASM_OPERAND_LETTER,
2350 RET, AT_SP): Remove.
2351 * config/i386/i386.md (*sse_prologue_save_insn): Use return
2352 instead of RET.
2354 * alias.c (init_alias_analysis): Use VEC_safe_grow_cleared.
2355 * cfgbuild.c (find_basic_blocks): Likewise.
2356 * cfgrtl.c (rtl_create_basic_block): Likewise.
2357 * function.c (temp_slots_at_level): Likewise.
2358 * reg-stack.c (stack_regs_mentioned): Likewise.
2359 * regclass.c (allocate_reg_info): Likewise.
2360 * tree-cfg.c (init_empty_tree_cfg, build_tree_cfg, create_bb,
2361 set_bb_for_stmt, move_block_to_fn): Likewise.
2362 * tree-complex.c (tree_lower_complex): Likewise.
2363 * vec.h (VEC_safe_grow_cleared): New.
2365 * cgraphunit.c, tree-ssa-alias.c: Fix comment typos.
2367 2007-01-03 Zdenek Dvorak <dvorakz@suse.cz>
2369 * loop-unswitch.c (unswitch_loop): Pass probabilities to loopify.
2370 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Pass probabilities
2371 to loop_version.
2372 * cfgloopmanip.c (scale_loop_frequencies): Export.
2373 (loopify): Scale the frequencies by prescribed coefficients.
2374 (set_zero_probability): New function.
2375 (duplicate_loop_to_header_edge): Improve updating of frequencies.
2376 (lv_adjust_loop_entry_edge, loop_version): Set probabilities
2377 and frequencies according to arguments.
2378 * tree-ssa-loop-manip.c (tree_unroll_loop): Set probabilities
2379 correctly.
2380 * cfg.c (scale_bbs_frequencies_int): Allow scaling the frequencies up.
2381 * modulo-sched.c (sms_schedule): Set probabilities for entering
2382 versioned loop correctly.
2383 * tree-vect-transform.c (vect_transform_loop): Ditto.
2384 * cfgloop.h (loopify, loop_version): Declaration changed.
2385 (scale_loop_frequencies): Declared.
2387 2007-01-02 Jan Hubicka <jh@suse.cz>
2389 * cgraph.c: Include tree-flow.h
2390 (cgraph_add_new-function): Handle IPA_SSA mode; execute
2391 early_local_passes.
2392 * cgraph.h (enum cgraph_state): Add CGRAPH_STATE_IPA_SSA.
2393 * tree-pass.h (pass_all_early_optimizations): Declare.
2394 * cgraphunit.c (cgraph_process_new_functions): Add IPA_SSA; execute
2395 early_local_passes.
2396 (cgraph_analyze_function): Do early_local_passes.
2397 * tree-mudflap.c (mf_decl_cache_locals, mf_build_check_statement_for):
2398 Do not add referenced vars.
2399 * tree-optimize.c (gate_all_optimizations): Do not execute when not in
2400 SSA form.
2401 (gate_all_early_local_passes): New gate.
2402 (pass_early_local_passes): Use new gate.
2403 (execute_early_local_optimizations): New functions.
2404 (gate_all_early_optimizations): New gate.
2405 (pass_all_early_optimizations): New pass.
2406 (execute_free_datastructures): Free SSA only when initialized.
2407 (gate_init_datastructures): Init only when optimizing.
2408 (tree_lowering_passes): Do early local passes when called late.
2409 * tree-profile.c (do_tree_profiling): Don't profile functions added
2410 late.
2411 (do_early_tree_profiling, pass_early_tree_profile): Kill.
2412 * tree-cfg.c (update_modified_stmts): Do not update when operands are
2413 not active.
2414 * passes.c (init_optimizations_passes): Reorder so we go into SSA
2415 during early_local_passes.
2416 * Makefile.in (cgraph.o): Add dependency on tree-flow.h.
2419 2007-01-02 Carlos O'Donell <carlos@codesourcery.com>
2421 * Makefile.in: Update copyright year.
2423 2007-01-02 Carlos O'Donell <carlos@codesourcery.com>
2425 * Makefile.in: Export GCC_EXEC_PREFIX before calling $(RUNTEST)
2426 in $(lang_checks) and check-consistency targets.
2428 2007-01-02 Jan Hubicka <jh@suse.cz>
2430 * tree-mudflap.c (mf_decl_cache_locals, mf_build_check_statement_for):
2431 Do not add referenced vars.
2432 * tree-cfg.c (update_modified_stmts): Do not update when SSA operands
2433 are not active.
2434 * passes.c (init_optimization_passes): Put mudflap_2 after
2435 free_datastructures.
2437 2007-01-02 Jan Hubicka <jh@suse.cz>
2439 * tree-optimize (execute_fixup_cfg): Set after_inlining flag.
2440 Set NOTHROW flag on call statements proved to be nothrow.
2441 Update statement of local calls so new pure/const functions are
2442 updated. Update_ssa when in ssa form. Mark PHI nodes of nonlocal
2443 goto receivers.
2444 (tree_rest_of_compilation): Register hooks and initialize bitmap
2445 early. Do not set after_inlining flag.
2447 2007-01-02 Steve Ellcey <sje@cup.hp.com>
2449 * sbitmap.c (HOST_BITS_PER_LONG_LONG): Change to
2450 HOST_BITS_PER_LONGLONG
2452 2007-01-02 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
2454 PR c/19977
2455 * c-typeck.c (store_init_value): Don't emit pedantic overflow
2456 warning for non-static initializers.
2458 2007-01-02 Steven Bosscher <steven@gcc.gnu.org>
2460 * config/alpha/alpha.md, arm/arm.c, darwin.c, frv/frv.md,
2461 m32r/m32r.c, m32r/m32r.c, mn10300/mn10300.md, pa/pa.c,
2462 rs6000/rs6000.c, s390/s390.md, sh/sh.md, sparc/sparc.c:
2463 Always use set_unique_reg_note to add REG_EQUAL notes.
2465 2007-01-02 Kazu Hirata <kazu@codesourcery.com>
2467 Revert:
2468 2007-01-02 Kazu Hirata <kazu@codesourcery.com>
2470 * alias.c (init_alias_analysis): Use VEC_safe_grow_cleared.
2471 * cfgbuild.c (find_basic_blocks): Likewise.
2472 * cfgrtl.c (rtl_create_basic_block): Likewise.
2473 * function.c (temp_slots_at_level): Likewise.
2474 * reg-stack.c (stack_regs_mentioned): Likewise.
2475 * regclass.c (allocate_reg_info): Likewise.
2476 * tree-cfg.c (init_empty_tree_cfg, build_tree_cfg, create_bb,
2477 set_bb_for_stmt, move_block_to_fn): Likewise.
2478 * tree-complex.c (tree_lower_complex): Likewise.
2479 * vec.h (VEC_safe_grow_cleared): New.
2481 2007-01-02 Ian Lance Taylor <iant@google.com>
2483 * c-common.c (c_common_truthvalue_conversion): When warning about
2484 using an assignment as a truth value, set TREE_NO_WARNING.
2486 2007-01-02 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
2488 PR middle-end/7651
2489 * c.opt (Wold-style-declaration): New.
2490 * doc/invoke.texi (C-only Warning Options): New.
2491 (Wold-style-declaration): Document it.
2492 (Wextra): Enabled by -Wextra.
2493 * c-opts.c (c_common_post_options): Enabled by -Wextra.
2494 * c-decl.c (declspecs_add_scspec): Replace -Wextra with
2495 -Wold-style-declaration.
2497 2007-01-02 Kazu Hirata <kazu@codesourcery.com>
2499 * alias.c (init_alias_analysis): Use VEC_safe_grow_cleared.
2500 * cfgbuild.c (find_basic_blocks): Likewise.
2501 * cfgrtl.c (rtl_create_basic_block): Likewise.
2502 * function.c (temp_slots_at_level): Likewise.
2503 * reg-stack.c (stack_regs_mentioned): Likewise.
2504 * regclass.c (allocate_reg_info): Likewise.
2505 * tree-cfg.c (init_empty_tree_cfg, build_tree_cfg, create_bb,
2506 set_bb_for_stmt, move_block_to_fn): Likewise.
2507 * tree-complex.c (tree_lower_complex): Likewise.
2508 * vec.h (VEC_safe_grow_cleared): New.
2510 2007-01-02 Douglas Gregor <doug.gregor@gmail.com>
2512 * c-common.c(c_common_nodes_and_builtins): Since variants of
2513 void_type_node get built before it is given a name, we need to
2514 give those variants the name, too.
2515 (complete_array_type): We need to work with the canonical main
2516 type of the array, from which we will build the qualified version.
2517 * params.def (PARAM_VERIFY_CANONICAL_TYPES): New.
2518 * print-tree.c (print_node): Display canonical type information
2519 for each type.
2520 * stor-layout.c (layout_type): When we don't know the
2521 alignment of a type for which we're building an array, we end up
2522 guessing wrong, so make the type require structural equality.
2523 * tree.c (make_node_stat): When we build a new type, it is its
2524 own canonical type.
2525 (build_type_attribute_qual_variant): When building an attribute
2526 variant, its canonical type is the non-attribute variant. However,
2527 if the attributes are target-dependent and they differ, we need to
2528 use structural equality checks for this type.
2529 (build_qualified_type): A qualified type is not equivalent to its
2530 unqualified variant; set the canonical type appropriately.
2531 (build_distinct_type_copy): When building a distinct type from
2532 another type, the new type is its own canonical type.
2533 (build_variant_type_copy): When building a new type variant, we
2534 assume that it is equivalent to the original type.
2535 (build_pointer_type_for_mode): When building a pointer type, also
2536 build a canonical type pointer.
2537 (build_reference_type_for_mode): When building a reference type,
2538 also build a canonical type reference.
2539 (build_index_type): When we can't hash an index type (e.g.,
2540 because its maximum value is negative), the index type requires
2541 structural equality tests.
2542 (build_array_type): Build the canonical form of an array type.
2543 (build_function_type): Function types require structural equality,
2544 because they contain default arguments, attributes, etc.
2545 (build_method_type_directly): Ditto for method types.
2546 (build_offset_type): Build the canonical offset type.
2547 (build_complex_type): Build the canonical vector type.
2548 (make_vector_type): Build the canonical vector type.
2549 * tree.h (TYPE_CANONICAL): New.
2550 (TYPE_STRUCTURAL_EQUALITY_P): New.
2551 (SET_TYPE_STRUCTURAL_EQUALITY): New.
2552 (struct tree_type): Added "canonical" field.
2553 * params.h (VERIFY_CANONICAL_TYPES): New.
2554 * doc/c-tree.texi (TYPE_CANONICAL): Document.
2555 (TYPE_STRUCTURAL_EQUALITY_P): Document.
2556 (SET_TYPE_STRUCTURAL_EQUALITY): Document.
2557 * doc/invoke.texi (verify-canonical-types): Document --param
2558 parameter for verifying canonical types.
2560 2007-01-02 Joseph Myers <joseph@codesourcery.com>
2562 * config.gcc (powerpc-*-eabispe*, powerpc-*-eabisimaltivec*,
2563 powerpc-*-eabisim*, powerpc-*-eabialtivec*, powerpc-*-eabi*,
2564 powerpc-*-rtems*, powerpc-wrs-vxworks, powerpc-wrs-vxworksae,
2565 powerpcle-*-eabisim*, powerpcle-*-eabi*): Add rs6000/e500.h to
2566 tm_file.
2567 * config/rs6000/e500.h: New.
2568 * config/rs6000/eabi.h (TARGET_SPE_ABI, TARGET_SPE, TARGET_E500,
2569 TARGET_ISEL, TARGET_FPRS, TARGET_E500_SINGLE, TARGET_E500_DOUBLE):
2570 Remove.
2571 * config/rs6000/linuxspe.h (TARGET_SPE_ABI, TARGET_SPE,
2572 TARGET_E500, TARGET_ISEL, TARGET_FPRS, TARGET_E500_SINGLE,
2573 TARGET_E500_DOUBLE): Remove.
2574 * config/rs6000/vxworks.h (TARGET_SPE_ABI, TARGET_SPE,
2575 TARGET_E500, TARGET_ISEL, TARGET_FPRS): Remove.
2576 * config/rs6000/rs6000.h (CHECK_E500_OPTIONS): Define.
2577 * config/rs6000/rs6000.c (rs6000_override_options): Use
2578 CHECK_E500_OPTIONS.
2580 2007-01-02 Joseph Myers <joseph@codesourcery.com>
2582 * config/rs6000/rs6000.c (print_operand): Check (TARGET_SPE ||
2583 TARGET_E500_DOUBLE), not TARGET_E500, for %y.
2584 (rs6000_generate_compare, rs6000_emit_sCOND, output_cbranch,
2585 rs6000_emit_cmove): Don't check TARGET_E500.
2586 * config/rs6000/rs6000.md (bunordered, bordered, sunordered,
2587 sordered): Don't check TARGET_E500.
2589 2007-01-01 Eric Christopher <echristo@apple.com>
2591 * config/mips/mips.c (mips_regno_mode_ok_for_base_p): Use
2592 HARD_REGISTER_NUM_P.
2594 2007-01-01 Roger Sayle <roger@eyesopen.com>
2596 * fold-const.c (fold_binary) <EQ_EXPR>: Fold "(X^C1) eq/ne C2" into
2597 "X eq/ne (C1^C2)". Fold "(X^Z) eq/ne (Y^Z)" as "X eq/ne Y" when Z
2598 has no side-effects. Fold "(X^C1) eq/ne (Y^C2)" as "(X^(C1^C2))
2599 eq/ne Y".
2601 2007-01-01 Mike Stump <mrs@apple.com>
2603 * configure.ac: Remove support for building with Apple's gcc-3.1.
2605 2007-01-02 Joseph Myers <joseph@codesourcery.com>
2607 PR middle-end/30311
2608 * caller-save.c (add_stored_regs): Only handle SUBREGs if inner
2609 REG is a hard register. Do not modify REG before calling
2610 subreg_nregs.
2611 * rtlanal.c (subreg_get_info): Don't assert size of XMODE is a
2612 multiple of the size of YMODE for certain lowpart cases.
2614 2007-01-01 Andrew Pinski <pinskia@gmail.com>
2616 PR middle-end/30253
2617 * gimplify (voidify_wrapper_expr): Update for
2618 GIMPLE_MODIFY_STMT.
2620 2007-01-01 Andreas Schwab <schwab@suse.de>
2622 PR target/29166
2623 * config/ia64/ia64.c (ia64_compute_frame_size): Account space for
2624 save of BR0 in extra_spill_size instead of spill_size.
2625 (ia64_expand_prologue): Save BR0 outside of the gr/br/fr spill
2626 area.
2627 (ia64_expand_epilogue): Restore BR0 from its new location.
2629 2007-01-01 Andrew Pinski <pinskia@gmail.com>
2631 * gimplify.c (gimplify_init_constructor <case VECTOR_TYPE>):
2632 Use a temporary variable if the left hand side is not a gimple
2633 register.
2635 2007-01-01 Andrew Pinski <pinskia@gmail.com>
2637 * gimplify.c (gimplify_return_expr): Make the temporary variable
2638 for the return expression, a gimple register variable.
2640 2007-01-01 Jan Hubicka <jh@suse.cz>
2642 * emit-rtl.c (emit_copy_of_insn_after): Do not call copy_insn_1 for
2643 INSN_LIST.
2645 2007-01-01 Mike Stump <mrs@apple.com>
2647 * configure.ac (HAVE_GAS_LITERAL16): Add autoconf check for
2648 .literal16.
2649 * config/darwin.c (machopic_select_rtx_section): Use
2650 HAVE_GAS_LITERAL16.
2651 (darwin_mergeable_constant_section): Likewise.
2652 * configure: Regenerate.
2653 * config.in: Regenerate.
2655 2007-01-01 Jan Hubicka <jh@suse.cz>
2656 Andrew Pinski <pinskia@gmail.com>
2658 * cgraphunit.c (cgraph_optimize): Call cgraph_add_new_functions
2659 before starting IPA passes.