1 2007-01-19 Basile Starynkevitch <basile@starynkevitch.net>
3 * doc/gty.texi (Options): Document the mark_hook option to
5 * gcc/gengtype.c (write_types_data, write_func_for_structure,
6 write_types, ggc_wtd, pch_wtd): Added skip_hooks to
7 write_types_data, ggc_wtd, pch_wtd for mark_hook.
8 (walk_type, write_func_for_structure): Generate the mark_hook if
11 2007-01-19 Jan Hubicka <jh@suse.cz>
13 * ipa-inline.c (cgraph_decide_inlining_incrementally): Instead of 'early' argument
14 take inlining mode argument specifying whether to inline for size/speeed or all
15 functions; add support for flattening; improve dumpting.
16 (cgraph_early_inlining): Update call of decide_inlining_incrementally.
18 2007-01-19 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
21 * toplev.c (warn_deprecated_use): Use %qD instead of %qs to print
22 the name of the declared identifier.
24 2007-01-19 Dirk Mueller <dmueller@suse.de>
26 * config/i386.h (CONDITIONAL_REGISTER_USAGE): Store
27 result of PIC_OFFSET_TABLE_REGNUM in temporary variable to avoid
30 2007-01-19 Uros Bizjak <ubizjak@gmail.com>
32 * config/i386/i386.md (acos<mode>2): Rename from acossf2 and acosdf2.
33 Macroize expander using X87MODEF12 mode macro. Extend operand 1
34 to XFMode, use acosxf2 and truncate result to requested mode.
35 Use SSE_FLOAT_MODE_P to disable patterns for SSE math.
36 (asin<mode>2): Similarly, with asin expanders.
37 (*fscalexf4_i387): Rename from *fscalexf4.
38 (expNcorexf3): New expander.
39 (expxf2, exp10xf2, exp2xf2): Use expNcorexf3 expander.
40 (exp<mode>2): Rename from expsf2 and expdf2. Macroize expander using
41 X87MODEF12 mode macro. Extend operand 1 to XFMode, use expxf2 and
42 truncate result to requested mode. Use SSE_FLOAT_MODE_P to disable
43 patterns for SSE math.
44 (exp10<mode>2): Similarly, with exp10 expanders.
45 (exp2<mode>2): Similarly, with exp2 expanders.
46 (expm1<mode>2): Similarly, with expm1 expanders.
47 (ldexp<mode>3): Similarly, with ldexp expanders.
48 (log<mode>2, log10<mode>2, log2<mode>2, log1p<mode>2, logb<mode>2):
49 Use gen_truncxf<mode>2_i387_noop to truncate result.
51 2007-01-19 Richard Sandiford <richard@codesourcery.com>
53 * config/m68k/m68k.h (PREFERRED_STACK_BOUNDARY): Define to 32
56 2007-01-19 Nathan Sidwell <nathan@codesourcery.com>
57 Richard Sandiford <richard@codesourcery.com>
59 * config/m68k/m68k.h (M68K_STATIC_CHAIN_REG_NAME): New macro.
60 (INITIALIZE_TRAMPOLINE): Use STATIC_CHAIN_REGNUM.
61 (__transfer_from_trampoline): Use M68K_STATIC_CHAIN_REG_NAME.
62 * config/m68k/m68kelf.h (STATIC_CHAIN_REGNUM): Override.
63 (M68K_STATIC_CHAIN_REG_NAME): Likewise.
64 * config/m68k/netbsd-elf.h (M68K_STATIC_CHAIN_REG_NAME): Likewise.
66 2007-01-19 Richard Sandiford <richard@codesourcery.com>
68 * config/m68k/m68k.md (adddi_dilshr32): Rename to...
69 (*adddi_dilshr32): ...this. Fix formatting. Remove commented-out
70 non-canonical pattern. Restrict to !TARGET_COLDFIRE.
71 (*adddi_dilshr32_cf): New pattern.
72 (adddi3, subdi3): Remove first alternatives.
74 2007-01-19 Richard Sandiford <richard@codesourcery.com>
76 * config/m68k/m68k.c (notice_update_cc): If an SFmode move is
77 implemented using move.l, do not use its cc result for floating-point
80 2007-01-19 Richard Sandiford <richard@codesourcery.com>
82 * config/m68k/m68k.h (EXTRA_CONSTRAINT): Stop the 'T' constraint
83 from accepting 's' constraints if flag_pic.
85 2007-01-19 Richard Sandiford <richard@codesourcery.com>
87 * config/m68k/m68k.md (bordered, bunordered, buneq, bunge, bungt)
88 (bunle, bunlt, bltgt, bordered_rev, bunordered_rev, buneq_rev)
89 (bunge_rev, bungt_rev, bunle_rev, bunlt_rev, bltgt_rev): Change
90 condition from TARGET_68881 to TARGET_HARD_FLOAT.
92 2007-01-19 Sandra Loosemore <sandra@codesourcery.com>
94 * longlong.h (count_leading_zeros, COUNT_LEADING_ZEROS_0): Add
95 ColdFire alternatives.
96 * config/m68k/m68k.h (CLZ_DEFINED_VALUE_AT_ZERO): New macro.
97 * config/m68k/m68k.md (clzsi2): Define for ColdFire
98 architectures that support ff1 instruction.
100 2007-01-19 Richard Sandiford <richard@codesourcery.com>
101 Julian Brown <julian@codesourcery.com>
103 * config/m68k/m68k.h (CONST_OK_FOR_LETTER_P): Add an 'R' case.
104 * config/m68k/m68k.md (*movsi_cfv4): Fold into...
105 (*movsi_cf): ...here. Remove unnecessary 'R' from 'Rg'.
108 2007-01-19 Richard Sandiford <richard@codesourcery.com>
110 * config/m68k/m68k-protos.h (valid_mov3q_const): Take a HOST_WIDE_INT
112 (output_move_const_into_data_reg, output_move_simode_const): Delete.
113 * config/m68k/m68k.c (const_method, const_int_cost): Take a
114 HOST_WIDE_INT instead of an rtx.
115 (m68k_rtx_costs): Update call accordingly.
116 (output_move_const_into_data_reg): Likewise. Fix formatting.
117 (valid_mov3q_const): Take a HOST_WIDE_INT instead of an rtx.
119 (output_move_simode_const): Update calls after above changes.
120 Rework to use automatic variables and predicates like MEM_P.
121 * config/m68k/m68k.md (pushexthisi_const): Update call to
124 2007-01-19 Dirk Mueller <dmueller@suse.de>
126 * tree-ssa-alias.c (perform_var_substitution): Fix typo
129 2007-01-19 Richard Guenther <rguenther@suse.de>
131 * builtins.c (expand_builtin_cexpi): Fall back to expanding
132 via cexp in case sincos is not available.
134 2007-01-19 Richard Guenther <rguenther@suse.de>
136 * doc/tm.texi (TARGET_HAS_SINCOS): Document new target macro.
137 * defaults.h (TARGET_HAS_SINCOS): Default to off.
138 * config/linux.h (TARGET_HAS_SINCOS): Set to on if we have glibc.
139 * config/alpha/linux.h (TARGET_HAS_SINCOS): Likewise.
140 * config/sparc/linux.h (TARGET_HAS_SINCOS): Likewise.
141 * config/sparc/linux64.h (TARGET_HAS_SINCOS): Likewise.
142 * config/rs6000/linux.h (TARGET_HAS_SINCOS): Likewise.
143 * config/rs6000/linux64.h (TARGET_HAS_SINCOS): Likewise.
145 2007-01-19 Uros Bizjak <ubizjak@gmail.com>
147 * config/i386/i386.md (*fpatanxf3_i387, fpatan_extend<mode>xf3_i387):
149 (atan2sf3_1, atan2df3_1, atan2xf3_1): Remove insn patterns.
150 (atan2xf3): Directly generate RTL pattern.
151 (atan2<mode>3): Rename from atan2sf3 and atan2df3 and macroize insn
152 patterns using X87MODEF12 mode macro. Use fpatan_extend<mode>xf3_i387
153 and truncate result to requested mode. Use SSE_FLOAT_MODE_P to
154 disable patterns for SSE math.
155 (atan<mode>2): Rename from atansf2 and atandf2 and macroize insn
156 patterns using X87MODEF12 mode macro. Use fpatan_extend<mode>xf3_i387
157 and truncate result to requested mode. Use SSE_FLOAT_MODE_P to
158 disable patterns for SSE math.
160 2007-01-19 Alexandre Oliva <aoliva@redhat.com>
162 * libgcc-std.ver: Fix typo in %inherit for GCC_4.3.0.
164 2007-01-18 Roger Sayle <roger@eyesopen.com>
166 * fold-const.c (fold_unary) <VIEW_CONVERT_EXPR>: Optimize away a
167 VIEW_CONVERT_EXPR to the same type as it's operand.
169 2007-01-18 David Edelsohn <edelsohn@gnu.org>
171 * config/rs6000/darwin-ldouble.c: Only build _SOFT_FLOAT if
172 configured for long double 128.
174 2007-01-18 Mike Stump <mrs@apple.com>
176 * config/rs6000/rs6000.c (rs6000_emit_vector_compare): Fix build
179 2007-01-18 Michael Meissner <michael.meissner@amd.com>
181 * i386.c (ix86_compute_frame_layout): Make fprintf's in #if 0 code
184 2007-01-18 Jan Hubicka <jh@suse.cz>
186 * tree-ssa-operands.c (vop_free_bucket_size): Never return value
187 greater than NUM_VOP_FREE_BUCKETS.
189 2007-01-18 Daniel Berlin <dberlin@dberlin.org>
191 * tree-ssa-structalias.c: Update comments.
192 (ptabitmap_obstack): Removed.
194 (oldpta_obstack): Ditto.
195 (stats): Add a few members.
196 (struct variable_info): Remove node, complex, address_taken, and
197 indirect_target members. Add oldsolution member.
198 (new_var_info): Do not initialize removed members.
199 (constraint_expr_type): Remove INCLUDES.
200 (constraint_graph): Add size, implicit_preds, rep,
201 indirect_cycles, eq_rep, label, direct_nodes, and complex members.
202 (FIRST_REF_NODE): New macro.
203 (LAST_REF_NODE): Ditto.
204 (FIRST_ADDR_NODE): Ditto.
205 (find): New function.
207 (dump_constraint): Do not handle INCLUDES.
208 (insert_into_complex): Do not insert duplicate constraints.
209 (condense_varmap_nodes): Renamed and rewritten into ...
210 (merge_node_constraints): This. Also fix bug in handling of
211 offseted copy constraints.
212 (clear_edges_for_node): No longer need to deal with preds at all,
213 or removing associated preds/succs.
214 (merge_graph_nodes): Deal with indirect_cycles.
215 Don't deal with predecessors.
216 (add_implicit_graph_edge): New function.
217 (add_pred_graph_edge): Ditto.
218 (add_graph_edge): Don't deal with predecessors.
219 (build_constraint_graph): Removed.
220 (build_pred_graph): New function.
221 (build_succ_graph): Ditto.
222 (struct scc_info): Removed in_component. Added roots, dfs, and
223 node_mapping. Remove visited_index, unification_queue.
224 (scc_visit): Deal with union-find we do now.
225 Deal with cycles with REF nodes.
226 (collapse_nodes): Renamed and rewritten to ...
228 (process_unification_queue): Removed.
229 (topo_visit): Cleanup
230 (do_da_constraint): Use find.
231 (do_sd_constraint): Ditto.
232 (do_ds_constraint): Ditto.
233 (do_complex_constraint): Ditto.
234 (init_scc_info): Update for removed and added members.
235 (find_and_collapse_graph_cycles): Renamed and rewritten into ...
236 (find_indirect_cycles): This.
237 (equivalence_class): New variable.
238 (label_visit): New function.
239 (perform_variable_substitution): Rewritten.
240 (free_var_substitution_info): New function.
241 (find_equivalent_node): Ditto.
242 (move_complex_constraints): Ditto.
243 (eliminate_indirect_cycles): Ditto.
244 (solve_graph): Only propagate changed bits.
245 Use indirect cycle elimination.
247 (tree_id_t): Rename to tree_vi_t, delete id member, add vi member.
248 (tree_id_eq): Renamed to ...
249 (tree_vi_eq): This. Update for member change
250 (insert_id_for_tree): Renamed and rewritten to ...
251 (insert_vi_for_tree): This.
252 (lookup_id_for_tree): Renamed and rewritten to ...
253 (lookup_vi_for_tree): This.
254 (get_id_for_tree): Renamed and rewritten to ...
255 (get_vi_for_tree): Ditto.
256 (get_constraint_exp_from_ssa_var): Update to use get_vi_for_tree.
257 (process_constraint): Don't handle INCLUDES.
258 Remove special ADDRESSOF case.
259 (find_func_aliases): Rewrite to use vi functions instead of id
261 (create_function_info_for): Ditto.
262 (create_variable_info_for): Ditto.
263 (intra_create_variable_infos): Ditto.
264 (merge_smts_into): Ditto.
265 (find_what_p_points_to): Ditto.
266 (init_base_vars): Ditto.
267 (init_alias_vars): Ditto.
268 (remove_preds_and_fake_succs): New function.
269 (dump_sa_points_to_info): Dump new stats.
270 (dump_solution_for_var): Use find.
271 (set_used_smts): Fix formatting.
272 (compute_points_to_sets): Updated for new functions.
273 (ipa_pta_execute): Ditto.
275 2007-01-18 Kazu Hirata <kazu@codesourcery.com>
276 Richard Sandiford <richard@codesourcery.com>
278 * doc/tm.texi (TARGET_FUNCTION_VALUE): Expand documentation of
280 * calls.c (expand_call): If the return value is a PARALLEL,
281 extract its first member.
282 * config/m68k/linux.h (FUNCTION_EXTRA_EPILOGUE): Remove.
283 * config/m68k/m68k.c (m68k_output_function_epilogue): Don't
284 use FUNCTION_EXTRA_EPILOGUE.
285 (m68k_function_value): Return a PARALLEL if the return value
286 is of a pointer type.
287 * config/m68k/netbsd-elf.h (current_function_returns_pointer)
288 (FUNCTION_EXTRA_EPILOGUE): Remove.
289 * config/m68k/m68k.md (D0_REG): New constant.
291 2007-01-18 Kazu Hirata <kazu@codesourcery.com>
293 * config/m68k/m68k.c (m68k_output_function_epilogue): Don't
294 output a NOP for empty epilogues.
296 2007-01-18 Richard Sandiford <richard@codesourcery.com>
298 * config/m68k/m68k.c (m68k_use_return_insn): Update comments
299 before function. Extend register save check to include all
300 registers, not just integer ones.
302 2007-01-18 Kazu Hirata <kazu@codesourcery.com>
304 * config/m68k/m68k-protos.h (use_return_insn): Rename to...
305 (m68k_use_return_insn): ...this.
306 * config/m68k/m68k.h (USE_RETURN_INSN): Delete.
307 * config/m68k/m68k.c (use_return_insn): Rename to...
308 (m68k_use_return_insn): ...this.
309 * config/m68k/m68k.md (return): Use m68k_use_return_insn instead
312 2007-01-18 Nathan Sidwell <nathan@codesourcery.com>
314 * config/m68k/fpgnulib.c (__truncdfsf2): Implement round to
315 nearest even, fix denormal rounding overflow.
317 2007-01-18 Richard Sandiford <richard@codesourcery.com>
319 * config/m68k/m68k.md (movsf_cf_hard): Use fsmove instead of
320 f%$smove and f%$move.
321 (movdf_cf_hard): Use fdmove for cases 0 and 3 and fmove for case 1.
322 (extendsfdf2_cf): Use fdmove instead of f%&move.
323 (truncdfsf2_cf): Use fsmove instead of f%$smove.
324 (add<mode>3_cf, sub<mode>3_cf): Use <FP:prec> instead of <FP:round>.
326 2007-01-18 Richard Sandiford <richard@codesourcery.com>
328 * config/m68k/m68k.md (movdf_cf_hard): Use output_move_double for
331 2007-01-18 Richard Sandiford <richard@codesourcery.com>
333 * real.h (real_format): Add a canonical_nan_lsbs_set field.
334 (coldfire_single_format): Declare.
335 (coldfire_double_format): Likewise.
336 * real.c (encode_ieee_single): Use canonical_nan_lsbs_set instead
337 of qnan_msb_set to determine the lower bits of a canonical
339 (encode_ieee_double): Likewise.
340 (encode_ieee_quad): Likewise.
341 (ieee_single_format): Initialize canonical_nan_lsbs_set.
342 (mips_single_format): Likewise.
343 (ieee_double_format): Likewise.
344 (mips_double_format): Likewise.
345 (ieee_extended_motorola_format): Likewise.
346 (ieee_extended_intel_96_format): Likewise.
347 (ieee_extended_intel_128_format): Likewise.
348 (ieee_extended_intel_96_round_53_format): Likewise.
349 (ibm_extended_format): Likewise.
350 (mips_extended_format): Likewise.
351 (ieee_quad_format): Likewise.
352 (mips_quad_format): Likewise.
353 (vax_f_format): Likewise.
354 (vax_d_format): Likewise.
355 (vax_g_format): Likewise.
356 (i370_single_format): Likewise.
357 (i370_double_format): Likewise.
358 (decimal_single_format): Likewise.
359 (decimal_double_format): Likewise.
360 (decimal_quad_format): Likewise.
361 (c4x_single_format): Likewise.
362 (c4x_extended_format): Likewise.
363 (real_internal_format): Likewise.
364 (coldfire_single_format): New real_format.
365 (coldfire_double_format): Likewise.
366 * config/pdp11/pdp11.c (pdp11_f_format): Initialize
367 canonical_nan_lsbs_set.
368 (pdp11_d_format): Likewise.
369 * config/m68k/m68k.c (override_options): Override REAL_FORMAT_MODE
370 if TARGET_COLDFIRE_CPU.
372 2007-01-18 Richard Sandiford <richard@codesourcery.com>
374 * config/m68k/m68k-protos.h (m68k_output_pic_call): Delete.
375 (output_call, m68k_legitimize_call_address): Declare.
376 * config/m68k/m68k.h (EXTRA_CONSTRAINT): Remove unnecessary
377 parenthesees. Add support for a 'W' constraint.
378 (LEGITIMATE_PIC_OPERAND_P): Remove SYMBOL_REF_FLAG handling.
379 (PRINT_OPERAND_PUNCT_VALID_P): Remove comment about 'o'.
380 (m68k_symbolic_call, m68k_symbolic_jump): Declare.
381 * config/m68k/m68k.c (m68k_symbolic_call, m68k_symbolic_jump): New
383 (override_options): Initialize them. Do not set flag_no_function_cse
384 for TARGET_ID_SHARED_LIBRARY.
385 (m68k_output_pic_call): Delete.
386 (m68k_legitimize_call_address): New function.
387 (print_operand): Remove the %o prefix. Handle the %p prefix.
388 (output_call): New function.
389 (m68k_output_mi_thunk): Use m68k_symbolic_jump. Always load the
390 target address from the GOT if symbolic jumps are not allowed.
391 * config/m68k/m68k.md (call, general_operand): Do not set
392 SYMBOL_REF_FLAG. Use m68k_legitimize_call_address instead.
393 Merge separate flag_pic and !flag_pic define_insns into...
394 (*call, *call_value): ...these new patterns. Match the address
395 rather than the containing MEM and require it to be a call_operand.
396 Use output_call to generate the asm template.
397 * config/m68k/predicates.md (const_call_operand): New predicate.
398 (call_operand): Likewise.
400 2007-01-18 Nathan Sidwell <nathan@codesourcery.com>
402 * config/m68k/m68k.h (REGISTER_MOVE_COST): Simplify definition.
403 (STACK_GROWS_DOWNWARD): Define to 1.
404 (FUNCTION_VALUE, LIBCALL_VALUE, FUNCTION_VALUE_REGNO_P): Equivocate
405 comments, emphasizing that these values are only defaults.
406 * config/m68k/linux.h (LINK_SPEC): Fix formatting in #undef.
407 * config/m68k/m68k.c (const_method): Remove trailing whitespace.
409 2007-01-18 Richard Sandiford <richard@codesourcery.com>
411 * config/m68k/m68k.md (cmpsi): Remove outdated flag_pic handling.
413 2007-01-18 Kazu Hirata <kazu@codesourcery.com>
414 Richard Sandiford <richard@codesourcery.com>
416 * config/m68k/m68k.h (DATA_REGNO_P, ADDRESS_REGNO_P, INT_REGNO_P)
417 (FP_REGNO_P): New macros.
418 (REGNO_OK_FOR_INDEX_P, REGNO_OK_FOR_BASE_P, REGNO_OK_FOR_DATA_P)
419 (REGNO_OK_FOR_FP_P, REG_OK_FOR_INDEX_P, REG_OK_FOR_BASE_P): Use them.
420 (EH_RETURN_STACKADJ_RTX): Use A0_REG.
421 * config/m68k/m68k.c (m68k_regno_mode_ok): Use the new REGNO macros.
423 2007-01-18 Nathan Sidwell <nathan@codesourcery.com>
425 * config.gcc (m68k-*-aout*, m68k-*-coff*, m68020-*-elf*, m68k-*-elf*)
426 (m68k-*-uclinuxoldabi, m68k-*-uclinux*, m68k-*-rtems*): Add t-floatlib
428 * config/m68k/t-floatlib: New file, extracting common code from...
429 * config/m68k/t-m68kbare, config/m68k/t-m68kelf,
430 * config/m68k/t-uclinux: Here.
431 * config/m68k/fpgnulib.c: Do not compile extendeed precision
432 routines on ColdFire targets.
434 2007-01-18 Nathan Sidwell <nathan@codesourcery.com>
436 * config.gcc (m68k-*-aout*, m68k-*-coff*, m68020-*-elf*, m68k-*-elf*)
437 (m68010-*-netbsdelf*, m68k*-*-netbsdelf*, m68k*-*-openbsd*)
438 (m68k-*-uclinuxoldabi, m68k-*-uclinux*, m68k-*-linux*)
439 (m68k-*-rtems*): Use tm_file rather than m68k/m68k.h and
440 explicitly set MOTOROLA to 1.
441 * config/m68k/m68k.h (MOTOROLA): Simplify definition accordingly.
443 2007-01-18 Richard Sandiford <richard@codesourcery.com>
444 Nathan Sidwell <nathan@codesourcery.com>
446 * config/m68k/m68k.h (PCC_STATIC_STRUCT_RETURN): Delete.
447 (ASM_OUTPUT_REG_PUSH, ASM_OUTPUT_REG_POP): Add MOTOROLA cases.
448 * config/m68k/coff.h (REGISTER_PREFIX_MD): Delete.
449 * config/m68k/m68020-elf.h (LIB_SPEC): Delete.
450 * config/m68k/m68k-none.h (CC1_SPEC, CPP_SUBTARGET_SPEC): Delete.
451 * config/m68k/m68kelf.h (IMMEDIATE_PREFIX, REGISTER_PREFIX_MD)
452 (ASM_OUTPUT_REG_PUSH, ASM_OUTPUT_REG_POP): Delete.
453 (NO_DOLLAR_IN_LABEL, PCC_STATIC_STRUCT_RETURN): Don't undefine.
454 (BSS_ASM_OP, ASM_OUTPUT_SKIP, ASM_OUTPUT_ASCII): Delete.
455 * config/m68k/m68kemb.h (PCC_STATIC_STRUCT_RETURN): Don't undefine.
456 * config/m68k/linux.h (SIZE_TYPE, PTRDIFF_TYPE, WCHAR_TYPE)
457 (WCHAR_TYPE_SIZE, TARGET_OBJFMT_CPP_BUILTINS): Delete.
458 (TARGET_OS_CPP_BUILTINS): Don't define mc68000 and mc68020 here.
459 (DBX_REGISTER_NUMBER): Delete.
460 * config/m68k/netbsd-elf.h (IMMEDIATE_PREFIX): Delete.
461 (PCC_STATIC_STRUCT_RETURN): Don't undefine.
462 * config/m68k/openbsd.h (PCC_STATIC_STRUCT_RETURN): Define.
464 2007-01-18 Richard Sandiford <richard@codesourcery.com>
466 * config.gcc (m68k-*-uclinux*): Add flat.h to $tm_file.
467 * config/flat.h: New file.
468 * crtstuff.c (USE_PT_GNU_EH_FRAME): Don't define if
470 * config/m68k/m68k.h (ASM_PREFERRED_EH_DATA_FORMAT): Do not use
471 indirect references for -msep-data or -mid-shared-library.
472 Do not use PC-relative code addresses either.
473 * config/m68k/m68k.c (override_options): Restrict -fPIC error
475 * config/m68k/uclinux.h (STARTFILE_SPEC): Define. Use Scrt1.o
476 for shared libraries and crt1.o for executables. Use crti.o and
478 (ENDFILE_SPEC): Use crtend.o and crtn.o.
479 (LIB_SPEC): Suppress -Rlibc.gdb if -static-libc is given.
480 Do not add -elf2flt or -shared-lib-id options here.
481 (LINK_SPEC): Define. Pass -elf2flt if no -elf2flt option is given.
482 Pass -shared-lib-id if -mid-shared-library, taking the library
483 identifier from -mshared-library-id if given, otherwise
485 (EH_FRAME_IN_DATA_SECTION): Do not undefine.
486 (INIT_SECTION_ASM_OP, FINI_SECTION_ASM_OP): Likewise.
487 (TARGET_OS_CPP_BUILTINS): Define __GXX_MERGED_TYPEINFO_NAMES=0
488 and __GXX_TYPEINFO_EQUALITY_INLINE=0 if -mid-shared-library.
489 (DRIVER_SELF_SPECS): Map unadorned PIC options to -msep-data.
490 * config/m68k/t-uclinux (EXTRA_MULTILIB_PARTS): Add crtbegin.o
492 * config/m68k/lb1sf68.asm (PICCALL): Use an lea and pc-relative
493 jump sequence for ISA A and ISA A+.
496 2007-01-18 Richard Sandiford <richard@codesourcery.com>
498 * config.gcc (m68k-*-uclinux*): Base the port on the common
499 and m68k GNU/Linux files rather than on the generic ELF ones.
500 * config/m68k/uclinux.h (TARGET_VERSION): Override.
501 (TARGET_OS_CPP_BUILTINS): Use LINUX_TARGET_OS_CPP_BUILTINS.
503 2007-01-18 Julian Brown <julian@codesourcery.com>
504 Richard Sandiford <richard@codesourcery.com>
506 * config/m68k/m68k.h (LONG_DOUBLE_TYPE_SIZE): Make 64-bit on ColdFire.
507 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
508 * config/m68k/netbsd-elf.h (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Undefine
510 * config/m68k/uclinux-oldabi.h (LONG_DOUBLE_TYPE_SIZE): Redefine to
512 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
514 2007-01-18 Richard Sandiford <richard@codesourcery.com>
516 * doc/install.texi: Document m68k-uclinuxoldabi.
517 * config.gcc (m68k-*-uclinuxoldabi): New configuration.
518 * config/m68k/uclinux-oldabi.h: New file, copied from
519 config/m68k/uclinux.h.
521 2007-01-18 Kaz Kojima <kkojima@rr.iij4u.or.jp>
523 * config/m32r/m32r-protos.h (m32r_expand_epilogue): Declare it.
524 * config/m32r/m32r.c (m32r_setup_incoming_varargs): Use gen_frame_mem.
525 (m32r_compute_frame_size): Use unsigned for regno.
526 (m32r_reload_lr): Use gen_frame_mem.
528 (m32r_output_function_epilogue): Don't output the function epilogue
530 (m32r_expand_epilogue): New.
531 (direct_return): Return false if the function has the interrupt
533 (m32r_hard_regno_rename_ok): Remove code for the textual epilogue.
534 * config/m32r/m32r.md (epilogue): New expander.
535 (return_lr, return_rte): New insns.
536 (return): Make it expander.
537 (return_normal): New expander.
539 2007-01-18 Josh Conner <jconner@apple.com>
542 * config/rs6000/rs6000.c (rs6000_emit_vector_compare): Add
543 support for UNLE, UNLT, UNGE, and UNGT.
545 2007-01-18 Jan Hubicka <jh@suse.cz>
547 * tree-vrp.c (finalize_jump_threads): Do not call cleanup_cfg by hand.
548 * tree-tailcall (add_virtual_phis): Likewise.
549 (optimize_tail_call): Return TODOs.
550 (execute_tail_calls): Return TODOs.
551 * tree-ssa-ccp (execute_fold_all_builtins): Do cleanup_cfg via TODO.
552 * tree-cfgcleanup.c (cleanup_tree_cfg_loop): Return if something
554 * tree-ssa-forwprop.c (tree_ssa_forward_propagate_single_use_value):
555 Cleanup cfg using TODO.
556 * tree-flow.h (cleanup_tree_cfg_loop): Update prototype.
557 * passes.c (execute_function_todo): When cleanup did something, remove
559 * tree-cfg.c (pass_build_cfg): Add cleanup_cfg TODO.
560 (make_edges): Don't cleanup_cfg.
562 2007-01-18 Uros Bizjak <ubizjak@gmail.com>
564 * reg-stack.c (subst_stack_regs_pat) [UNSPEC_SINCOS_COS,
565 UNSPEC_XTRACT_FRACT]: Use generic code for instructions that
566 operate on the top of stack.
567 [UNSPEC_SINCOS_SIN, UNSPEC_XTRACT_EXP, UNSPEC_TAN]: Rewrite
568 register handling of instructions that output to the second
570 [UNSPEC_TAN_ONE, UNSPEC_TAN_TAN]: Remove.
571 (move_for_stack_reg): Special-case check for dead destination
572 stack slot for constant load of 1.0 inside UNSPEC_TAN.
574 * config/i386/i386.md (UNSPEC_TAN): New constant.
575 (UNSPEC_TAN_ONE, UNSPEC_TAN_TAN): Remove.
576 (fptanxf4_i387, fptan_extend<mode>xf4_i387): New patterns
577 to correctly model move of constant 1.0 to top stack slot.
578 (*tandf3_1, *tansf3_1, *tanxf3_1): Remove insn patterns.
579 (unnamed peephole2 pattern): Remove corresponding peephole2
580 pattern that optimizes tan insn and loading of constant 1.0.
581 (tanxf2): Use fptanxf4_i387.
582 (tan<mode>2): Rename from tansf2 and tandf2 and macroize insn
583 patterns using X87MODEF12 mode macro. Use fptan_extend<mode>xf4_i387
584 and truncate result to requested mode. Use SSE_FLOAT_MODE_P to
585 disable patterns for SSE math.
586 (sincos<mode>3): Use truncxf<mode>2_i387_noop for truncation.
587 (fyl2x_extend<mode>xf3_i387): Use X87MODEF12 for operand 1.
589 2007-01-18 Dirk Mueller <dmueller@suse.de>
590 Richard Guenther <rguenther@suse.de>
593 * doc/invoke.texi (Warray-bounds): Document -Warray-bounds.
594 * common.opt (Warray-bounds): Add new warning option.
595 * c-opts.c (c_common_handle_option): Define -Warray-bounds
597 * Makefile.in: make tree-vrp.o depend on toplev.h
598 * tree-vrp.c (vrp_finalize): Call check_array_refs if -Warray-bounds
600 (check_array_refs, check_array_bounds, check_array_ref): New.
602 2007-01-18 Jan Hubicka <jh@suse.cz>
604 * tree-ssa-ccp.c (ccp_finalize): Return if something changed.
605 (execute_ssa_ccp): Return flags conditionally.
606 * tree-ssa-propagate.c (substitue_and_fold): Return if something was
608 * tree-ssa-propagate.h (substitute_and_fold): Update prototype.
610 2007-01-18 Steven Bosscher <steven@gcc.gnu.org>
612 * cfgcleanup.c (cleanup_cfg): Detect cfglayout mode and set
613 the CLEANUP_CFGLAYOUT flag when in cfglayout mode.
615 * Makefile.c (GTFILES): Add cfglayout.h.
616 * gengtype.c (open_base_files): Likewise.
617 * cfglayout.c (cfg_layout_function_footer,
618 cfg_layout_function_header) Reindent to make gengtype happy.
619 * cfglayout.h (cfg_layout_function_footer,
620 cfg_layout_function_header): Add GTY(()) marker.
622 * ifcvt.c (noce_try_sign_mask): Make sure INSN_B is non-null.
624 2007-01-18 Ben Elliston <bje@au.ibm.com>
626 * genautomata.c (write_automata): Include xstrerror output in the
627 error message if writing the DFA description file fails.
629 2007-01-17 H.J. Lu <hongjiu.lu@intel.com>
631 * config/mips/mips-protos.h (mips_output_external): Make it
633 * config/mips/iris.h (TARGET_ASM_EXTERNAL_LIBCALL): Removed.
634 * config/mips/mips.c (irix_output_external_libcall): Likewise.
635 (extern_list): Likewise.
636 (extern_head): Likewise.
637 (TARGET_ASM_FILE_END): Likewise.
638 (mips_file_end): Likewise.
639 (mips_output_external): Rewritten.
641 2007-01-18 Ben Elliston <bje@au.ibm.com>
643 * genpreds.c (write_insn_preds_c): Only write out the function
644 body for regclass_for_constraint if we have register constraints.
646 2007-01-17 Tom Tromey <tromey@redhat.com>
648 * doc/sourcebuild.texi (libgcj Tests): Use sourceware.org.
649 * doc/install.texi (Testing): Use sourceware.org.
650 (Binaries): Likewise.
651 (Specific): Likewise.
652 * doc/contrib.texi (Contributors): Use sourceware.org.
654 2007-01-17 Anatoly Sokolov <aesok@post.ru>
656 * config/avr/avr.h (AVR_HAVE_LPMX): New macro.
657 (AVR_ENHANCED): Rename to ...
658 (AVR_HAVE_MUL): ... new.
659 (avr_enhanced_p): Rename to ...
660 (avr_have_mul_p): ... new.
661 (TARGET_CPU_CPP_BUILTINS): Use 'avr_have_mul_p' instead of
662 'avr_enhanced_p' for "__AVR_ENHANCED__". Define "__AVR_HAVE_MUL__".
663 * config/avr/avr.c (avr_enhanced_p): Rename to ...
664 (avr_have_mul_p): ... new.
665 (base_arch_s): Rename 'enhanced' to 'have_mul'.
666 (avr_override_options): Use 'avr_have_mul_p' and 'have_mul' instead of
667 'avr_enhanced_p' and 'enhanced'.
668 (ashlhi3_out, ashrhi3_out, lshrhi3_out, avr_rtx_costs): Use
669 AVR_HAVE_MUL instead of AVR_ENHANCED.
670 * avr.md (*tablejump_enh): Use AVR_HAVE_LPMX instead of AVR_ENHANCED.
671 (mulqi3, *mulqi3_enh, *mulqi3_call, mulqihi3, umulqihi3, mulhi3,
672 *mulhi3_enh, *mulhi3_call, mulsi3, *mulsi3_call): Use AVR_HAVE_MUL
673 instead of AVR_ENHANCED.
674 (*tablejump_enh): Use AVR_HAVE_LPMX instead of AVR_ENHANCED.
675 * libgcc.S: Use __AVR_HAVE_MUL__ instead of __AVR_ENHANCED__.
676 (__tablejump__): Use __AVR_HAVE_LPMX__ instead of __AVR_ENHANCED__.
678 2007-01-17 Ian Lance Taylor <iant@google.com>
680 * vec.h (VEC_reserve_exact): Define.
681 (vec_gc_p_reserve_exact): Declare.
682 (vec_gc_o_reserve_exact): Declare.
683 (vec_heap_p_reserve_exact): Declare.
684 (vec_heap_o_reserve_exact): Declare.
685 (VEC_OP (T,A,reserve_exact)): New static inline function, three
687 (VEC_OP (T,A,reserve)) [all versions]: Remove handling of
689 (VEC_OP (T,A,alloc)) [all versions]: Call ...reserve_exact.
690 (VEC_OP (T,A,copy)) [all versions]: Likewise.
691 (VEC_OP (T,a,safe_grow)) [all versions]: Likewise.
692 * vec.c (calculate_allocation): Add exact parameter. Change all
694 (vec_gc_o_reserve_1): New static function, from vec_gc_o_reserve.
695 (vec_gc_p_reserve, vec_gc_o_reserve): Call vec_gc_o_reserve_1.
696 (vec_gc_p_reserve_exact, vec_gc_o_reserve_exact): New functions.
697 (vec_heap_o_reserve_1): New static function, from vec_heap_o_reserve.
698 (vec_heap_p_reserve, vec_heap_o_reserve): Call vec_heap_o_reserve_1.
699 (vec_heap_p_reserve_exact): New function.
700 (vec_heap_o_reserve_exact): New function.
702 2007-01-17 Jan Hubicka <jh@suse.cz>
704 * ipa-type-escape.c (look_for_casts): Revamp using handled_component_p.
706 2007-01-17 Eric Christopher <echristo@apple.com>
708 * config.gcc: Support core2 processor.
710 2007-01-16 Jan Hubicka <jh@suse.cz>
712 * tree-ssanames.c (release_dead_ssa_names): Instead of ggc_freeing
713 the names, just unlink the chain so we don't crash on dangling pointers
716 2007-01-16 Jan Hubicka <jh@suse.cz>
718 * cgraph.h (cgraph_decide_inlining_incrementally): Kill.
719 * tree-pass.h: Reorder to make IPA passes appear toegher.
720 (pass_early_inline, pass_inline_parameters, pass_apply_inline): Declare.
721 * cgraphunit.c (cgraph_finalize_function): Do not compute inling
722 parameters, do not call early inliner.
723 * ipa-inline.c: Update comments. Include tree-flow.h
724 (cgraph_decide_inlining): Do not compute inlining parameters.
725 (cgraph_decide_inlining_incrementally): Return TODOs; assume to
726 be called with function context set up.
727 (pass_ipa_inline): Remove unreachable functions before pass.
728 (cgraph_early_inlining): Simplify assuming to be called from the
730 (pass_early_inline): New pass.
731 (cgraph_gate_ipa_early_inlining): New gate.
732 (pass_ipa_early_inline): Turn into simple wrapper.
733 (compute_inline_parameters): New function.
734 (gate_inline_passes): New gate.
735 (pass_inline_parameters): New pass.
736 (apply_inline): Move here from tree-optimize.c
737 (pass_apply_inline): New pass.
738 * ipa.c (cgraph_remove_unreachable_nodes): Verify cgraph after
740 * tree-inline.c (optimize_inline_calls): Return TODOs rather than
742 (tree_function_versioning): Do not allocate dummy struct function.
743 * tree-inline.h (optimize_inline_calls): Update prototype.
744 * tree-optimize.c (execute_fixup_cfg): Export.
745 (pass_fixup_cfg): Remove
746 (tree_rest_of_compilation): Do not apply inlines.
747 * tree-flow.h (execute_fixup_cfg): Declare.
748 * Makefile.in (gt-passes.c): New.
749 * passes.c: Include gt-passes.h
750 (init_optimization_passes): New passes.
751 (nnodes, order): New static vars.
752 (do_per_function_toporder): New function.
753 (execute_one_pass): Dump current pass here.
754 (execute_ipa_pass_list): Don't dump current pass here.
756 2007-01-16 Janis Johnson <janis187@us.ibm.com>
758 * config/dfp-bit.c (dfp_compare_op): Return separate value for NaN.
759 (DFP_NE, DFP_LE, DFP_GE): Return false for NaN.
761 2007-01-16 David Edelsohn <edelsohn@gnu.org>
763 * config/rs6000/darwin-ldouble.c: Build file for SOFT_FLOAT.
764 (strong_alias): Define.
765 (__gcc_qmul): Provide non-FMA for soft-float.
781 * config/rs6000/rs6000.c (rs6000_init_libfuncs): Initialize
782 soft-float functions.
783 * config/rs6000/libgcc-ppc-glibc.ver: Version soft-float symbols.
784 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Do not warn
785 about long double soft float.
787 2007-01-16 Dorit Nuzman <dorit@il.ibm.com>
788 Tehila Meyzels <tehila@il.ibm.com>
790 * tree-vectorizer.h (is_pattern_stmt_p): New.
791 * tree-vect-analyze.c (vect_determine_vectorization_factor): Fix
792 formatting (tabs instead of spaces). Cleanup and clarify setting
793 of STMT_VINFO_VECTYPE. Call is_pattern_stmt_p.
794 * tree-vect-transform.c (vect_get_vec_def_for_operand): Fix typo.
795 (vectorizable_type_demotion): Check that types are integral.
796 (vectorizable_type_promotion): Likewise.
797 (vectorizable_store): Fix typo. Eliminate new-line at end of
800 2007-01-16 Jan Hubicka <jh@suse.cz>
802 * tree-ssanames.c (release_dead_ssa_names): Remove invalidated
805 2007-01-15 Eric Christopher <echristo@apple.com>
807 * ifcvt.c: Include vec.h, vecprim.h.
808 (check_cond_move_block): New argument regs.
809 Reorganize. Add registers used to regs.
810 (cond_move_process_if_block): Use regs set above as
813 2007-01-15 Eric Christopher <echristo@apple.com>
815 * config/darwin.h: Update copyright.
816 (TARGET_OPTION_TRANSLATE_TABLE): Add umbrella.
817 (LINK_COMMAND_SPEC): Add -u.
818 (LINK_SPEC): Fix umbrella for above.
820 2007-01-15 Joseph S. Myers <joseph@codesourcery.com>
822 * config/soft-fp/op-common.h, config/soft-fp/op-4.h: Update from
825 2007-01-15 Tom Tromey <tromey@redhat.com>
827 * doc/sourcebuild.texi (libgcj Tests): Don't mention jacks.
828 * doc/install.texi (Testing): Don't mention jacks.
829 (Configuration): Document --enable-java-maintainer-mode. Move
830 --with-java-home to libgcj-specific section. Document
832 (Prerequisites): Mention --enable-java-maintainer-mode, ecj1.
834 2007-01-15 Jan Hubicka <jh@suse.cz>
836 * tree-ssa-dce.c (DCE_TODOs): New.
837 (propagate_necessity): Return if something changed.
838 (eliminate_unnecessary_stmts): Likewise.
839 (perform_tree_ssa_dce): Return TODO flags when needed.
840 (pass_dce, pass_dce_loop, pass_cd_dce): Remove TODO flags.
842 2007-01-15 Uros Bizjak <ubizjak@gmail.com>
844 * config/i386/i386.md (fyl2xxf3_i387): Rename from fyl2x_xf3.
845 (fyl2x_extend<mode>xf3_i387): New insn pattern.
846 (log<mode>2): Rename from logsf2 and logdf2 and macroize insn
847 patterns using X87MODEF12 mode macro. Extend operand 1
848 to XFmode. Use SSE_FLOAT_MODE_P to disable patterns for SSE math.
849 (log10<mode>2): Ditto.
850 (log2<mode>2): Ditto.
851 (log1p<mode>2): Ditto.
852 (logb<mode>2): Ditto.
853 (fyl2xp1xf3_i387): Rename from fyl2xp1_xf3.
854 (fyl2xp1_extend<mode>xf3_i387): New insn pattern.
855 (*fxtractxf3_i387): Rename from *fxtractxf3.
856 (fxtract_extend<mode>xf3_i387): New insn pattern.
857 (ilogbsi2): Use match_dup 3, not match_operand:XF 3.
858 * config/i386/i386.c (ix86_emit_i387_log1p): Use gen_fyl2xp1xf3_i387()
859 and gen_fyl2xxf3_i387().
861 2007-01-14 Zdenek Dvorak <dvorakz@suse.cz>
863 * loop-unswitch.c (unswitch_loop): Do not call fix_loop_placement.
864 * cfgloopmanip.c (fix_loop_placement): Made static. Use
865 get_loop_exit_edges. Changed return type to bool.
866 * cfgloop.h (fix_loop_placement): Declaration removed.
868 2007-01-14 Dorit Nuzman <dorit@il.ibm.com>
870 * param.h (MIN_VECT_LOOP_BOUND): New.
871 * params.def (MIN_VECT_LOOP_BOUND): New.
872 * tree-vectorizer.c (slpeel_tree_peel_loop_to_edge): Takes another
873 argument - minimum threshold for number of iterations.
874 * tree-vectorizer.h (slpeel_tree_peel_loop_to_edge): Add another
875 argument to declaration.
876 * tree-vect-analyze.c (vect_analyze_operations): Check value of
878 * tree-vect-transform.c (vect_do_peeling_for_loop_bound): Call
879 slpeel_tree_peel_loop_to_edge with additional argument.
880 (vect_do_peeling_for_alignment): Likewise.
881 * doc/invoke.texi (min-vect-loop-bound): Document new param option.
883 2007-01-14 Uros Bizjak <ubizjak@gmail.com>
886 * config/i386/i386.c (print_operand) ['z']: Output 'b' for
889 2007-01-14 Jan Hubicka <jh@suse.cz>
891 * tree-dfa.c (remove_referenced_var): New function.
892 * tree-ssa-live.c (remove_unused_locals): Walk referenced vars and
893 prune referenced vars list too.
894 * tree-flow.h (remove_referenced_var): Declare.
896 2007-01-14 Jan Hubicka <jh@suse.cz>
898 * tree-eh.c (add_stmt_to_eh_region_fn): Do not add call_exprs
900 (remove_stmt_from_eh_region_fn): Do not remove call_exprs.
901 (verify_eh_throw_stmt_node, verify_eh_throw_table_statements): Kill.
902 * except.h (verify_eh_throw_table_statements): Kill prototype.
903 * cfgexpand.c (expand_gimple_basic_block): Propagate Eh regions
904 into call exrepssions.
905 * tree-optimize.c (execute_free_cfg_annotatiosn): Do not call
907 * tree-cfg.c: Include pointer-set.h.
908 (verify_node_sharing): Work on pointer set.
909 (verify_eh_throw_stmt_node): New.
910 (verify_stmts): Use pointers sets, verify throw_stmt.
912 2007-01-13 Zdenek Dvorak <dvorakz@suse.cz>
914 * ipa-reference.c (analyze_function): Consider also addresses taken
917 2007-01-12 Roger Sayle <roger@eyesopen.com>
919 * c-typeck.c (null_pointer_constant_p): Replace use of
920 TREE_CONSTANT_OVERFLOW with TREE_OVERFLOW.
921 (build_c_cast): Likewise.
923 2007-01-12 Roger Sayle <roger@eyesopen.com>
925 * tree.h (force_fit_type_double): Remove unused final argument.
926 * c-common.c (constant_expression_warning): Replace use of
927 TREE_CONSTANT_OVERFLOW with TREE_OVERFLOW.
928 (convert_and_check): Likewise.
929 (shorten_compare): Update call to force_fit_type_double.
930 (c_common_truthvalue_conversion) <INTEGER_CST>: Use integer_zerop.
931 * convert.c (convert_to_pointer): Update call to
932 force_fit_type_double.
933 * fold-const.c (force_fit_type_double): Remove overflowed_const
935 (int_const_binop, fold_convert_const_int_from_int,
936 fold_convert_const_int_from_real, fold_div_compare,
937 fold_sign_changed_comparison, fold_unary, fold_negate_const,
938 fold_abs_const, fold_not_const): Remove the final argument from
939 calls to force_fit_type_double.
941 2007-01-12 Andrew Pinski <andrew_pinski@playstation.sony.com>
943 * configure.ac: Set insn to "nop" for spu-*-* also.
944 * configure: Regenerate.
946 2007-01-12 Olga Golovanevsky <olga@il.ibm.com>
948 * builtins.def : Add BUILT_IN_FREE.
950 2007-01-12 Jan Hubicka <jh@suse.cz>
952 PR tree-optimization/30443
953 * tree-inline.c (tree_function_versioning): Do not optimize when
954 cloning for inlining.
956 2007-01-12 Zdenek Dvorak <dvorakz@suse.cz>
958 * doc/loop.texi: Document recording of loop exits.
959 * cfgloopmanip.c (loopify, duplicate_loop): Use alloc_loop.
960 (update_single_exits_after_duplication,
961 update_single_exit_for_duplicated_loop,
962 update_single_exit_for_duplicated_loops): Removed.
963 (duplicate_loop_to_header_edge): Do not call
964 update_single_exits_after_duplication and
965 update_single_exit_for_duplicated_loops.
966 (loop_version): Do not update single_exit information.
967 (fix_loop_structure): Use record_loop_exits instead of
968 mark_single_exit_loops.
969 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update
970 the lists of loop exits.
971 * cfghooks.c (redirect_edge_and_branch, redirect_edge_and_branch_force,
972 split_edge, merge_blocks): Update the lists of loop exits.
973 * modulo-sched.c (sms_schedule): Pass LOOPS_HAVE_RECORDED_EXITS to
975 * loop-init.c (loop_optimizer_init): Call record_loop_exits instead
976 of mark_single_exit_loops.
977 (loop_optimizer_finalize): Call release_recorded_exits.
978 * tree-ssa-loop.c (tree_loop_optimizer_init): Pass
979 LOOPS_HAVE_RECORDED_EXITS to loop_optimizer_init.
980 * tree-vectorizer.c (slpeel_tree_duplicate_loop_to_edge_cfg): Do not
981 update single exit information.
982 * lambda-code.c (perfect_nestify): Ditto.
983 * cfgloop.c (flow_loop_free): Destroy the list of exits of the loop.
984 (mark_single_exit_loops): Removed.
985 (alloc_loop, loop_exit_hash, loop_exit_eq, loop_exit_free,
986 get_exit_descriptions, rescan_loop_exit, record_loop_exits,
987 dump_recorded_exit, dump_recorded_exits, release_recorded_exits): New
989 (get_loop_exit_edges, single_exit): Use recorded exit lists.
990 (add_bb_to_loop, remove_bb_from_loops): Update the lists of loop exits.
991 (verify_loop_structure): Verify consistency of the exit lists.
992 (flow_loops_find): Use alloc_loop. Initialize exits hash.
993 (set_single_exit): Removed.
994 * cfgloop.h (struct loop_exit): New function.
995 (struct loop): single_exit_ field replaced by exits field.
996 (LOOPS_HAVE_MARKED_SINGLE_EXITS): Replaced by LOOPS_HAVE_RECORDED_EXITS.
997 (struct loops): Added exits hash.
998 (mark_single_exit_loops, set_single_exit): Declaration removed.
999 (release_recorded_exits, record_loop_exits, rescan_loop_exit): Declare.
1001 2007-01-12 Richard Sandiford <richard@codesourcery.com>
1003 * doc/invoke.texi: Avoid use of @headitem.
1005 2007-01-12 Richard Sandiford <richard@codesourcery.com>
1007 * cse.c (cse_insn): Move HAVE_CC0 code after declarations.
1009 2007-01-12 Richard Sandiford <richard@codesourcery.com>
1011 * doc/install.texi: Fix m68k-*-* anchor and add m68k-*-* to the
1014 2007-01-12 Nathan Sidwell <nathan@codesourcery.com>
1015 Richard Sandiford <richard@codesourcery.com>
1017 * doc/invoke.texi: Document -mno-bitfield, -mno-rtd and -mno-short.
1018 * config/m68k/m68k.opt: Resort options.
1019 (mbitfield, mrtd, mshort): Remove RejectNegative properties.
1021 2007-01-12 Nathan Sidwell <nathan@codesourcery.com>
1022 Richard Sandiford <richard@codesourcery.com>
1024 * doc/invoke.texi: Document the macros that are defined by
1025 m68k's -mtune and -mhard-float options.
1026 * config/m68k/m68k-protos.h (m68k_cpp_cpu_ident) Declare.
1027 (m68k_cpp_cpu_family): Likewise.
1028 * config/m68k/m68k.h (TARGET_CPU_CPP_BUILTINS): Add a full set
1029 of __ucfv*__ macros. Define __mcffpu__ if generating code for
1030 ColdFire FPUs. Define __mcf_cpu_* and __mcf_family_* macros.
1031 * config/m68k/m68k.c (m68k_cpp_cpu_ident): New function.
1032 (m68k_cpp_cpu_family): Likewise.
1034 2007-01-12 Richard Sandiford <richard@codesourcery.com>
1036 * config/m68k/m68k.h (TARGET_CPU_CPP_BUILTINS): Treat all mc68*
1037 macros besides mc68000 as tuning macros. Use a switch statement
1038 to set them and mcpu32.
1040 2007-01-12 Julian Brown <julian@codesourcery.com>
1042 * config/m68k/m68k.h: Use TARGET_68040 instead of TARGET_68040_ONLY.
1043 (TARGET_68040_ONLY): Rename to...
1044 (TARGET_68040): ...this.
1045 * config/m68k/m68k.c: Use TARGET_68040 instead of TARGET_68040_ONLY.
1046 * config/m68k/m68k.md: Likewise.
1048 2007-01-12 Julian Brown <julian@codesourcery.com>
1049 Nathan Sidwell <nathan@codesourcery.com>
1050 Richard Sandiford <richard@codesourcery.com>
1052 * config.gcc (m680[012]0-*-*, m68k*-*-*): Set m68k_cpu_ident to
1053 the -mcpu= argument associated with the --with-cpu setting.
1054 Define M68K_DEFAULT_TUNE to the default -mtune= option,
1055 if different from the one implied by the -mcpu setting.
1056 Accept --with-cpu=FOO if FOO is listed in m68k-devices.def,
1057 using mcpu=FOO as the default CPU option. Set target_cpu_default2.
1058 * doc/invoke.texi: Mention ColdFire in the introduction to the
1059 m68k options. Document the new -march, -mcpu, -mtune, -mdiv,
1060 -mno-div and -mhard-float options. Make -m68881 a synonym for
1061 -mhard-float. Document the previously-undocumented -m5206e,
1062 -m528x, -m5307 and -m5407 options. Tweak the existing option
1063 documentation for consistency.
1064 * doc/install.texi: Mention new --with-cpu arguments.
1065 * config/m68k/m68k.h (OPTION_DEFAULT_SPECS): Only use the
1066 default CPU if neither -mcpu nor -march are specified.
1067 (ASM_CPU_SPEC): Pass down -mcpu and -march options.
1068 (TARGET_CPU_CPP_BUILTINS): Set __mcfisa*__ macros from
1069 TARGET_ISA*. Set the legacy __mcf*__ cpu macros in the same way,
1070 using m68k_tune to decide between families that implement the
1071 same ISA. Use m68k_tune to set __mcfv4e__.
1072 (FL_BITFIELD, FL_68881, FL_COLDFIRE, FL_CF_HWDIV, FL_CF_MAC)
1073 (FL_CF_EMAC, FL_CF_EMAC_B, FL_CF_USP, FL_CF_FPU, FL_ISA_68000)
1074 (FL_ISA_68010, FL_ISA_68020, FL_ISA_68040, FL_ISA_A, FL_ISA_B)
1075 (FL_ISA_C, FL_ISA_MMU): New macros.
1076 (MASK_COLDFIRE): Delete.
1077 (TARGET_68010, TARGET_68020, TARGET_68040_ONLY, TARGET_COLDFIRE)
1078 (TARGET_ISAB): Redefine in terms of m68k_cpu_flags.
1079 (TARGET_68881, TARGET_COLDFIRE_FPU): Redefine in terms of m68k_fpu.
1080 (TARGET_HARD_FLOAT): Do not define here.
1081 (TARGET_ISAAPLUS, TARGET_ISAC): New macros.
1082 (TUNE_68000): New macro.
1083 (TUNE_68000_10): Redefine in terms of TUNE_68000 and TUNE_68010.
1084 (TUNE_68010, TUNE_68030, TUNE_68040, TUNE_68060, TUNE_CPU32)
1085 (TUNE_CFV2): Redefine in terms of m68k_tune.
1086 (uarch_type, target_device, fpu_type): New enums.
1087 (m68k_cpu, m68k_tune, m68k_fpu, m68k_cpu_flags): Declare.
1088 * config/m68k/m68k.c (TARGET_DEFAULT): Remove MASK_68881.
1089 (FL_FOR_isa_00, FL_FOR_isa_10, FL_FOR_isa_20, FL_FOR_isa_40)
1090 (FL_FOR_isa_cpu32, FL_FOR_isa_a, FL_FOR_isa_aplus, FL_FOR_isa_b)
1091 (FL_FOR_isa_c): New macros.
1092 (m68k_isa): New enum.
1093 (m68k_target_selection): New structure.
1094 (all_devices, all_isas, all_microarchs): New tables.
1095 (m68k_cpu_entry, m68k_arch_entry, m68k_tune_entry, m68k_cpu)
1096 (m68k_tune, m68k_fpu, m68k_cpu_flags): New variables.
1097 (MASK_ALL_CPU_BITS): Delete.
1098 (m68k_find_selection): New function.
1099 (m68k_handle_option): Handle -mcpu=, -march= and -mtune=.
1100 Map the legacy target options to a combination of the new ones.
1101 (override_options): Set m68k_cpu, m68k_tune, m68k_fpu and
1102 m68k_cpu_flags. Handle M68K_DEFAULT_TUNE. Use m68k_cpu_flags
1103 to derive default MASK_BITFIELD, MASK_CF_HWDIV and MASK_HARD_FLOAT
1105 * config/m68k/m68k.opt (m5200, m5206e, m528x, m5307, m5407, mcfv4e)
1106 (m68010, m68020, m68020-40, m68020-60, m68030, m68040): Remove Mask
1108 (m68881, msoft-float): Change mask from 68881 to HARD_FLOAT.
1109 (march=, mcpu=, mdiv, mhard-float, mtune=): New options.
1110 * config/m68k/m68k-devices.def: New file.
1112 2007-01-12 Richard Sandiford <richard@codesourcery.com>
1113 Nathan Sidwell <nathan@codesourcery.com>
1115 * config/m68k/m68k.h (ASM_CPU_SPEC, ASM_SPEC, EXTRA_SPECS)
1116 (SUBTARGET_EXTRA_SPECS): New macros.
1117 * config/m68k/linux.h (ASM_SPEC): Remove CPU flags;
1118 use %(asm_cpu_spec) instead.
1119 * config/m68k/m68k-none.h (ASM_SPEC): Likewise.
1120 * config/m68k/openbsd.h (ASM_SPEC): Likewise.
1121 * config/m68k/netbsd-elf.h (ASM_SPEC): Likewise.
1122 (EXTRA_SPECS): Rename to...
1123 (SUBTARGET_EXTRA_SPECS): ...this.
1125 2007-01-12 Nathan Sidwell <nathan@codesourcery.com>
1126 Richard Sandiford <richard@codesourcery.com>
1127 Julian Brown <julian@codesourcery.com>
1129 * config.gcc (m68k-*-aout*, m68k-*-coff*, m68020-*-elf*, m68k-*-elf*)
1130 (m68k-*-uclinux*, m68k-*-linux*, m68k-*-rtems*): Set default_m68k_cpu
1131 to the configuration's default CPU.
1132 (m68010-*-netbsdelf*, m68k*-*-netbsdelf*, m68k*-*-openbsd*): Likewise.
1133 Remove default masks.
1134 (m680[012]0-*-*): Set the default with_cpu to the first part of
1136 (m68k*-*-*): Set the default with_cpu to m$default_m68k_cpu.
1137 (m68k*-*-linux): Extend the --with-cpu handling to...
1138 (m680[012]0-*-*, m68k*-*-*): ...these configurations. Allow m68000
1139 and m68010. Don't set target_cpu_default2.
1140 * doc/install.texi: Document --with-cpu for m68k.
1141 * config/m68k/m68k.h (OPTION_DEFAULT_SPECS): Define.
1142 * config/m68k/m68k-none.h (TARGET_CPU_DEFAULT, M68K_CPU_m68k)
1143 (M68K_CPU_m68000, M68K_CPU_m68010, M68K_CPU_m68020, M68K_CPU_m68030)
1144 (M68K_CPU_m68040, M68K_CPU_m68302, M68K_CPU_m68332, TARGET_DEFAULT)
1145 (ASM_CPU_DEFAULT_SPEC, CC1_CPU_DEFAULT_SPEC): Delete.
1146 (ASM_SPEC): Remove use of %(asm_cpu_default).
1147 (EXTRA_SPECS, SUBTARGET_EXTRA_SPECS, MULTILIB_DEFAULTS): Delete.
1148 * config/m68k/linux.h (TARGET_DEFAULT): Delete.
1149 (CPP_SPEC): Merge definitions. Do not handle __HAVE_68881__ here.
1150 * config/m68k/netbsd-elf.h (TARGET_OS_CPP_BUILTINS): Define
1151 __HAVE_FPU__ if TARGET_HARD_FLOAT.
1152 (TARGET_DEFAULT): Delete.
1153 (EXTRA_SPECS): Delete cpp_cpu_default_spec, cpp_cpu_spec,
1154 cpp_fpu_spec, asm_default_spec and netbsd_cpp_spec.
1155 (CPP_CPU_SPEC): Delete.
1156 (TARGET_VERSION): Merge definitions, using TARGET_68010 to pick
1157 the appropriate string.
1158 (CPP_CPU_DEFAULT_SPEC, ASM_DEFAULT_SPEC, CPP_FPU_SPEC): Delete.
1159 (CPP_SPEC): Define to NETBSD_CPP_SPEC.
1160 (ASM_SPEC): Don't use %(asm_default_spec).
1161 * config/m68k/m68k.c (TARGET_DEFAULT_TARGET_FLAGS): Remove
1162 TARGET_DEFAULT and add MASK_68881.
1163 * config/m68k/m68k.md: Remove mention of TARGET_DEFAULT from comments.
1165 2007-01-12 Richard Sandiford <richard@codesourcery.com>
1167 * config.gcc (m68010-*-netbsdelf*): Add MASK_68010.
1168 (m68k*-*-netbsdelf*, m68k*-*-openbsd*, m68k*-linux*): Add
1169 MASK_68010 alongside MASK_68020.
1170 * doc/invoke.texi: Document -m68010.
1171 * config/m68k/m68k.opt (m68010): New.
1172 * config/m68k/m68k.h (TARGET_CPU_CPP_BUILTINS): Define mc68010
1174 (TUNE_68010): New macro.
1175 * config/m68k/m68k-none.h (M68K_CPU_m68k, M68K_CPU_m68010)
1176 (M68K_CPU_m68020, M68K_CPU_m68030, M68K_CPU_m68040)
1177 (M68K_CPU_m68332): Add MASK_68010.
1178 * config/m68k/linux.h (TARGET_DEFAULT): Add MASK_68010 to
1179 fallback definition.
1180 * config/m68k/netbsd-elf.h (CPP_CPU_SPEC): Remove now-redundant
1182 * config/m68k/m68k.c (MASK_ALL_CPU_BITS): Add MASK_68010.
1183 (m68k_handle_option): Handle OPT_m68010. Add MASK_68010
1184 to all entries that use MASK_68020.
1185 (output_move_simode_const, output_move_himode, output_move_qimode)
1186 (output_move_stricthi, output_move_strictqi): Use TARGET_68010
1187 instead of TARGET_68020 to select clr behavior. Remove comment
1188 about there being no TARGET_68010.
1189 * config/m68k/m68k.md: Likewise throughout.
1191 2007-01-12 Julian Brown <julian@codesourcery.com>
1193 * config/m68k/m68k.h (TARGET_ISAB): New macro.
1194 * config/m68k/m68k.c: Use TARGET_ISAB rather than TARGET_CFV4.
1195 * config/m68k/m68k.md: Likewise.
1197 2007-01-12 Julian Brown <julian@codesourcery.com>
1199 * config/m68k/m68k.h (LEGITIMATE_INDEX_P, LEGITIMIZE_ADDRESS): Use
1200 TARGET_COLDFIRE_FPU instead of TARGET_CFV4E.
1202 2007-01-12 Julian Brown <julian@codesourcery.com>
1204 * config/m68k/m68k.h (TUNE_68040_60): New macro.
1205 * config/m68k/m68k.c (standard_68881_constant_p): Use it.
1206 * config/m68k/m68k.md: Likewise.
1208 2007-01-12 Julian Brown <julian@codesourcery.com>
1209 Richard Sandiford <richard@codesourcery.com>
1211 * config/m68k/m68k.h (TARGET_CPU_CPP_BUILTINS): Use TUNE_68030
1212 instead of TARGET_68030, TUNE_68040 instead of TARGET_68040,
1213 TUNE_68060 instead of TARGET_68060 and TUNE_CPU32 instead of
1215 (TARGET_CPU32): Rename to...
1216 (TUNE_CPU32): ...this.
1217 (TUNE_68000_10, TUNE_68030, TUNE_68040, TUNE_68060)
1218 (TUNE_CFV2): New macros.
1219 * config/m68k/netbsd-elf.h (LONG_DOUBLE_TYPE_SIZE): Simplify;
1220 remove conditions that are implied by TARGET_68020.
1221 * config/m68k/m68k.c (m68k_output_function_prologue): Use TUNE_68040
1222 instead of TARGET_68040 and TUNE_CPU32 instead of TARGET_CPU32.
1223 (m68k_output_function_epilogue): Likewise.
1224 (m68k_rtx_costs): Likewise. Use TUNE_68060 instead of TARGET_68060
1225 and TUNE_CFV2 instead of TARGET_5200. Use TUNE_68000_10 instead of
1226 "!TARGET_68020 && !TARGET_COLDFIRE" to choose between 68000 and
1227 non-68000 timings. Refactor multiplication and division costs.
1228 (output_addsi3): Use TUNE_68040 instead of TARGET_68040 and
1229 TUNE_CPU32 instead of TARGET_CPU32.
1230 (standard_68881_constant_p): Use TUNE_68040 instead of TARGET_68040
1231 and TUNE_68060 instead of TARGET_68060.
1232 * config/m68k/m68k.md: Use TUNE_68040 instead of TARGET_68040,
1233 TUNE_68060 instead of TARGET_68060, and TUNE_CPU32 instead of
1235 (movsi_const0): Use TUNE_68000_10 rather than "!TARGET_68020
1236 && !TARGET_COLDFIRE" to choose between moveq and clr.
1237 Likewise in the unnamed movsf pattern.
1238 (ashlsi_17_24, lshrsi_17_24): Guard with TUNE_68000_10 rather than
1239 "!TARGET_68020 && !TARGET_COLDFIRE". Likewise the unnamed
1242 2007-01-12 Richard Sandiford <richard@codesourcery.com>
1244 * config/m68k/m68k.h (TARGET_CPU_CPP_BUILTINS): Increase amount
1245 of tabbing before backslashes.
1247 2007-01-11 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
1249 * pa-linux.h (ASM_OUTPUT_INTERNAL_LABEL): Undefine.
1250 * pa.h (ASM_OUTPUT_LABEL): Output colon when using GAS.
1251 (ASM_OUTPUT_INTERNAL_LABEL): Define.
1253 2007-01-11 Zdenek Dvorak <dvorakz@suse.cz>
1255 * tree-ssa-loop-ivopts.c (extract_cond_operands): Split from
1256 find_interesting_uses_cond.
1257 (find_interesting_uses_cond): Use extract_cond_operands.
1258 (rewrite_use_compare): Use extract_cond_operands and
1259 force_gimple_operand_bsi. Do not call update_stmt.
1260 (determine_use_iv_cost_condition): Use extract_cond_operands.
1261 Return cheaper of using original bound and changing the exit bound.
1263 2007-01-11 Zdenek Dvorak <dvorakz@suse.cz>
1265 PR tree-optimization/29516
1266 * tree-ssa-address.c (tree_mem_ref_addr, add_to_parts,
1267 most_expensive_mult_to_index, addr_to_parts,
1268 create_mem_ref, maybe_fold_tmr): Make the type of
1269 fields of TARGET_MEM_REF sizetype.
1270 (move_fixed_address_to_symbol, move_pointer_to_base):
1272 * tree.def (TARGET_MEM_REF): Add comment on types of
1275 2007-01-11 Joseph Myers <joseph@codesourcery.com>
1277 * c-common.c (vector_types_convertible_p): Treat opaque types as
1278 always convertible if they have the same size, but not otherwise.
1280 2007-01-11 Steven Bosscher <steven@gcc.gnu.org>
1282 * ifcvt.c (struct noce_if_info): Add comments to the fields.
1283 Remove the b_unconditional field.
1284 (noce_try_sign_mask): Do not look at b_unconditional.
1285 (noce_process_if_block): Do not use merge_if_blocks. Update
1286 the CFG here. Do not set b_unconditional.
1287 (cond_move_process_if_block): Likewise.
1288 (find_cond_trap): Likewise.
1289 (check_cond_move_block): Require simple jump insns at the end
1292 2007-01-11 Jan Hubicka <jh@suse.cz>
1294 PR tree-optimization/1046
1295 * tree-tailcall.c (suitable_for_tail_call_opt_p): Use TREE_ADDRESSABLE
1296 when alias info is not ready.
1297 (pass_tail_recursion): Do not require aliasing.
1298 * tree-ssa-copyrename.c (pass_rename_ssa_cop): Likewise.
1299 * tree-ssa-ccp.c (pass_ccp, pass_fold_builtins): Likewise.
1300 * tree-ssa-copy.c (pass_copy_prop): Likewise.
1301 * tree-ssa-forwprop.c (pass_forwprop): Likewise.
1302 * tree-ssa-dce.c (pass_dce, pass_dce_loop, pass_cd_dce): Likewise.
1303 * passes.c (init_optimization_passes): Execute rename_ssa_copies,
1304 ccp, forwprop, copy_prop, merge_phi, copy_prop, dce and tail recursion
1306 * tree-ssa-operands.c (add_virtual_operand, get_indirect_ref_operand):
1307 When aliasing is not build, mark statement as volatile.
1309 2007-01-11 Tom Tromey <tromey@redhat.com>
1311 PR preprocessor/15185, PR preprocessor/20989:
1312 * doc/cppopts.texi <-MT>: Update description of algorithm for
1313 computing default target.
1314 <-M, -MD>: Reword "basename" text.
1316 2007-01-11 Roger Sayle <roger@eyesopen.com>
1318 * builtins.c (expand_builtin_pow, expand_builtin_powi,
1319 fold_builtin_cabs, fold_builtin_sqrt, fold_builtin_trunc,
1320 fold_builtin_floor, fold_builtin_ceil, fold_builtin_round,
1321 fold_builtin_int_int_roundingfn, fold_builtin_bitop,
1322 fold_builtin_bswap, real_constp, fold_builtin_pow,
1323 fold_builtin_powi, fold_builtin_signbit, fold_builtin_copysign,
1324 do_mpfr_arg1, do_mpfr_arg2, do_mpfr_arg3, do_mpfr_sincos): Replace
1325 uses of the macro TREE_CONSTANT_OVERFLOW with TREE_OVERFLOW.
1326 * convert.c (convert_to_pointer): Likewise.
1327 * expr.c (highest_pow2_factor, expand_expr_real_1): Likewise.
1328 * fold-const.c (force_fit_type, fold_negate_expr, int_const_binop,
1329 const_binop, fold_convert_const_int_from_int,
1330 fold_convert_const_int_from_real,
1331 fold_convert_const_real_from_real, sign_bit_p,
1332 optimize_minmax_comparison, extract_muldiv_1, fold_div_compare,
1333 fold_sign_changed_comparison, fold_unary, fold_comparison,
1334 fold_binary, multiple_of_p, tree_Expr_non_zero_p,
1335 fold_negate_const, fold_abs_const, fold_not_const): Likewise.
1336 * print-tree.c (print_node_brief, print_node): Likewise.
1337 * stor-layout.c (place_field, layout_type): Likewise.
1338 * tree-chrec.c (keep_cast): Likewise.
1339 * tree.c (build_vector, build_real, build_real_from_int_cst,
1340 build_complex): Likewise.
1342 2007-01-11 Roger Sayle <roger@eyesopen.com>
1344 * tree.h (TREE_CONSTANT_OVERFLOW): Obsolete. For the time being,
1345 treat TREE_CONSTANT_OVERFLOW as a synonym of TREE_OVERFLOW.
1347 2007-01-11 Paolo Bonzini <bonzini@gnu.org>
1349 * configure.ac (strict1_warn): Rename to strict_warn.
1350 (WERROR, --enable-werror, symlink hacks, stage1_cflags,
1351 cc_set_by_configure, quoted_cc_set_by_configure,
1352 stage_prefix_set_by_configure, quoted_stage_prefix_set_by_configure,
1353 all_boot_languages, all_stagestuff): Remove.
1354 (target_list): Remove bootstrap targets.
1355 * Makefile.in (quickstrap): Unconditionally make a synonym of all.
1356 (BOOT_LANGUAGES, STAGE1_CFLAGS, STAGE1_CHECKING,
1357 REMAKEFLAGS, FLAGS_TO_PASS, PREPEND_DOTDOT_TO_RELATIVE_PATHS,
1358 SUBDIR_FLAGS_TO_PASS, WERROR_FLAGS, STRICT2_WARN, LANG_STAGESTUFF,
1359 VOL_FILES, POSTSTAGE1_FLAGS_TO_PASS, STAGE2_FLAGS_TO_PASS,
1360 STAGEPROFILE_FLAGS_TO_PASS, STAGEFEEDBACK_FLAGS_TO_PASS, stage1_build,
1361 stage1_copy, stage2_build, stage2_copy, stageprofile_build,
1362 stageprofile_copy, stage3_build, stage3_copy, stagefeedback_build,
1363 stagefeedback_copy, stage4_build, clean_s1, clean_sw, bootstrap,
1364 bootstrap-lean, bootstrap2, bootstrap2-lean, bootstrap3,
1365 bootstrap3-lean, bootstrap4, bootstrap4-lean, unstage1, unstage2,
1366 unstage3, unstage4, unstageprofile, unstagefeedback, restage, restage2,
1367 restage3, restage4, restageprofile, restagefeedback, bubbleestrap,
1368 cleanstrap, unstrap, restrap, *compare, *compare3, *compare4,
1369 *compare-lean, *compare3-lean, *compare4-lean, stage1-start, stage1,
1370 stage2-start, stage2, stage3-start, stage3, stage4-start, stage4,
1371 stageprofile-start, stageprofile, stagefeedback-start, stagefeedback,
1372 risky-stage1, risky-stage2, risky-stage3, risky-stage4): Remove.
1373 (ORDINARY_FLAGS_TO_PASS): Rename to FLAGS_TO_PASS.
1374 (STAGECOPYSTUFF, STAGEMOVESTUFF): Consolidate into MOSTLYCLEANFILES.
1375 (mostlyclean): Adjust.
1376 (clean, distclean): Don't mention bootstrap stuff.
1377 * configure: Regenerate.
1378 * ada/config-lang.in, cp/config-lang.in, forttran/config-lang.in,
1379 java/config-lang.in, objc/config-lang.in, objcp/config-lang.in,
1380 treelang/config-lang.in (stagestuff): Remove.
1381 * doc/sourcebuild.texi (stage1, stage2, stage3, stage4,
1382 stageprofile, stagefeedback, stagestuff): Remove mention.
1384 2007-01-11 Nick Clifton <nickc@redhat.com>
1386 * config/mcore/predicates.md (mcore_general_movesrc_operand):
1388 (mcore_general_movdst_operand): Do not accept CONST_INTs.
1389 (mcore_arith_K_S_operand): Run the test for the S constraint not
1390 the test for the M constraint.
1391 (mcore_addsub_operand): Do not accept integer values that are
1392 larger than 32 bits.
1393 * config/mcore/mcore.md: Remove unused constraints from split.
1394 (andsi3): Use HOST_WIDE_INT instead of int to hold an INTVAL.
1396 (allocate_stack): Likewise.
1397 * config/mcore/mcore.c (mcore_print_operand): Restrict output of P
1398 operands to 32 bits.
1399 (mcore_const_costs): Use HOST_WIDE_INT instead of int to hold an
1401 (mcore_and_cost, mcore_modify_comparison, const_ok_for_mcore,
1402 mcore_const_ok_for_inline, mcore_const_trick_uses_not,
1403 try_constant_tricks, mcore_num_ones, mcore_num_zeros,
1404 mcore_output_bclri, mcore_output_andn, output_inline_const,
1405 mcore_output_move, mcore_output_movedouble): Likewise.
1406 (mcore_output_cmov): Use CONST_OK_FOR_M and CONST_OK_FOR_N.
1407 (output_inline_const): Likewise.
1408 (output_inline_const): Fix format strings used in sprintf
1410 * config/mcore/mcore-protos.h: Update prototypes for changed
1411 functions in mcore.c.
1412 * config/mcore/mcore.h (CONST_OK_FOR_I): Cast values to
1413 HOST_WIDE_INT and not int.
1414 (CONST_OK_FOR_J, CONST_OK_FOR_K, CONST_OK_FOR_L, CONST_OK_FOR_M,
1415 CONST_OK_FOR_N): Likewise.
1416 (LEGITIMATE_CONSTANT_P): Also check CONSTANT_P.
1417 (GO_IF_LEGITIMATE_INDEX): Use HOST_WIDE_INT instead of int to hold
1420 2007-01-10 Jan Hubicka <jh@suse.cz>
1422 * tree-vrp.c (remove_range_assertions): Release defs.
1423 * tree-ssa-loop-ivopts.c (rmeove_statement): Likewise.
1424 * tree-ssa-dom.c (remove_stmt_or_phi): Likewise.
1426 2007-01-10 Paul Brook <paul@codesourcery.com>
1428 * config/arm/arm.c (arm_rtx_costs_1): Handle mutiply-subtract.
1429 * config/arm/arm.md (mulsi3subsi): New insn.
1431 2007-01-10 Zdenek Dvorak <dvorakz@suse.cz>
1433 * tree-ssa-loop-manip.c (tree_unroll_loop): Make it a wrapper over ...
1434 (tree_transform_and_unroll_loop): New.
1435 * tree-flow.h (transform_callback, tree_transform_and_unroll_loop):
1438 2007-01-10 Robert Kennedy <jimbob@google.com>
1440 * fold-const.c (fold_comparison): Fold comparisons like (x *
1441 1000 < 0) to (x < 0).
1443 2007-01-10 Ian Lance Taylor <iant@google.com>
1445 * tree-pretty-print.c (dump_generic_node): Print parentheses when
1446 operands have the same priority.
1448 2007-01-10 Tom Tromey <tromey@redhat.com>
1450 * fold-const.c (fold_truthop): Don't check can_use_bit_fields_p.
1451 (fold_binary): Likewise.
1452 * langhooks.c (lhd_can_use_bit_fields_p): Removed.
1453 * langhooks-def.h (lhd_can_use_bit_fields_p): Removed.
1454 (LANG_HOOKS_CAN_USE_BIT_FIELDS_P): Removed.
1455 (LANG_HOOKS_INITIALIZER): Remove LANG_HOOKS_CAN_USE_BIT_FIELDS_P.
1456 * langhooks.h (struct lang_hooks): Removed field
1457 'can_use_bit_fields_p'.
1459 2007-01-10 Ralf Corsépius <ralf.corsepius@rtems.org>
1461 * config/bfin/t-bfin, config/bfin/t-bfin-elf: Remove GCC_CFLAGS.
1463 2007-01-10 Razya Ladelsky <razya@il.ibm.com>
1465 * function.c (get_last_funcdef_no): New function.
1466 * function.h (get_last_funcdef_no): Declare.
1467 * tree-inline.c (initialize_cfun): Add initialization.
1468 (tree_function_versioning): Cleanup.
1470 2007-01-10 Jan Hubicka <jh@suse.cz>
1472 * tree-inline.c (setup_one_parameter): Do not propagate into abnormal
1475 2007-01-10 Sa Liu <saliu@de.ibm.com>
1476 Ben Elliston <bje@au.ibm.com>
1478 * spu.h (STACK_SAVE_AREA): Use VOIDmode for SAVE_FUNCTION, SImode
1479 for SAVE_NONLOCAL and Pmode for any other save level.
1480 * spu-protos.h (spu_restore_stack_block): Declare.
1481 * spu.md (save_stack_block): Remove.
1482 (restore_stack_block): Call spu_restore_stack_block.
1483 * spu.c (spu_restore_stack_block): New function.
1484 (spu_expand_epilogue): Remove old comment.
1486 2007-01-09 Zdenek Dvorak <dvorakz@suse.cz>
1488 PR tree-optimization/30322
1489 * tree-ssa-loop-ivopts.c (fold_affine_expr, iv_value): Removed.
1490 (cand_value_at): Return the value as aff_tree.
1491 (may_eliminate_iv): Convert the bound from aff_tree to tree.
1492 * tree-affine.c (aff_combination_add_cst, aff_combination_add_product,
1493 aff_combination_mult): New functions.
1494 (aff_combination_add): Use aff_combination_add_cst.
1495 (aff_combination_convert): Allow conversions to a wider type.
1496 (tree_to_aff_combination): Handle BIT_NOT_EXPR.
1497 * tree-affine.h (aff_combination_mult): Declare.
1499 2007-01-09 Carlos O'Donell <carlos@codesourcery.com>
1501 * doc/tm.texi: Update documentation to reflect reality of exec
1502 and start file search behaviours. Update copyright year.
1503 * doc/invoke.texi: Explain how GCC_EXEC_PREFIX is used to find
1504 header file directories.
1506 2007-01-09 Uros Bizjak <ubizjak@gmail.com>
1508 * config/i386/i386.md (*sinxf2): Rename to *sinxf2_i387.
1509 (*cosxf2): Rename to cosxf2_i387.
1510 (*sindf2, *sinsf2): Extend operand 1 to XFmode. Macroize patterns
1511 using X87MODEF12 mode macro. Rename patterns to
1512 *sin_extend<mode>xf2_i387. Use SSE_FLOAT_MODE_P to disable patterns
1514 (*cosdf2, *cossf2): Ditto.
1515 (sincosdf3, sincossf3): Ditto. Rewrite corresponding splitters
1516 to match extended input operands.
1517 (sincos<mode>3): New expander.
1518 (*sinextendsfdf2, *cosextendsfdf2, *sincosextendsfdf3): Remove
1519 insn patterns and corresponding splitters.
1521 2007-01-09 Kaz Kojima <kkojima@gcc.gnu.org>
1523 * config/sh/t-linux (TARGET_LIBGCC2_CFLAGS): Delete.
1524 (SHLIB_MAPFILES, SHLIB_LINK, SHLIB_INSTALL): Likewise.
1526 2007-01-09 Nicolas Pitre <nico@cam.org>
1529 * arm/ieee754-df.S (Lad_s): Also test the low word of X for zero.
1531 2007-01-08 Geoffrey Keating <geoffk@apple.com>
1533 * target.h (struct gcc_target): New field library_rtti_comdat.
1534 * target-def.h (TARGET_CXX_LIBRARY_RTTI_COMDAT): New.
1535 (TARGET_CXX): Add TARGET_CXX_LIBRARY_RTTI_COMDAT.
1536 * doc/tm.texi (C++ ABI): Document TARGET_CXX_LIBRARY_RTTI_COMDAT.
1537 * config/darwin.h (TARGET_CXX_LIBRARY_RTTI_COMDAT): Define.
1539 2007-01-08 Geoffrey Keating <geoffk@apple.com>
1541 * doc/invoke.texi (Optimize Options): Correct description of -O0.
1543 2007-01-08 Richard Guenther <rguenther@suse.de>
1545 * tree.h (force_fit_type_double): Export.
1546 (force_fit_type): Remove.
1547 * fold-const.c (force_fit_type_double): New function.
1548 (force_fit_type): Remove.
1549 (int_const_binop): Use it.
1550 (fold_convert_const_int_from_int): Likewise.
1551 (fold_convert_const_int_from_real): Likewise.
1552 (fold_div_compare): Likewise.
1553 (fold_sign_changed_comparison): Likewise.
1554 (fold_unary): Likewise.
1555 (fold_negate_const): Likewise.
1556 (fold_abs_const): Likewise.
1557 (fold_not_const): Likewise.
1558 * c-common.c (shorten_compare): Use force_fit_type_double.
1559 * convert.c (convert_to_pointer): Likewise.
1561 2007-01-08 Richard Guenther <rguenther@suse.de>
1563 * tree.h (build_int_cst_wide_type): Export.
1564 * tree.c (build_int_cst_wide_type): New function.
1565 (build_int_cst_wide): Fix comment.
1566 * builtins.c (fold_builtin_object_size): Use build_int_cst
1567 to build -1 or 0 of the correct type. Use fit_double_type
1568 to check for overflow.
1569 * fold-const.c (optimize_bit_field_compare): Use build_int_cst_type
1571 (decode_field_reference): Likewise.
1572 (all_ones_mask_p): Likewise.
1573 (native_interpret_int): Use build_int_cst_wide_type.
1574 (fold_binary): Use build_int_cst_type to build an all-ones
1576 * stor-layout.c (set_sizetype): Use build_int_cst_wide_type.
1578 2007-01-08 Daniel Jacobowitz <dan@codesourcery.com>
1580 * config/pa/t-pa64 (libgcc_stub.a): Use $(T).
1582 2007-01-09 Ben Elliston <bje@au.ibm.com>
1584 * genautomata.c (STATS_OPTION): New option.
1585 (stats_flag): New flag.
1586 (gen_automata_option): Handle it.
1587 (initiate_automaton_gen): Ditto.
1588 (write_automata): Output statistics only if stats_flag is
1589 set. Likewise, output time statistics only if time_flag is set.
1590 * doc/md.texi (Processor pipeline description): Document new flag.
1592 2007-01-08 Richard Guenther <rguenther@suse.de>
1594 * builtins.c (fold_builtin_int_roundingfn): Use fit_double_type.
1595 * tree.c (build_int_cst_type): Likewise.
1596 (size_in_bytes): Don't call force_fit_type on the result.
1597 (int_fits_type_p): Use fit_double_type.
1598 * fold-const.c (fit_double_type): New function.
1599 (force_fit_type): Use it.
1600 * tree.h (fit_double_type): Export.
1602 2007-01-08 Jan Hubicka <jh@suse.cz>
1604 * tree-vectorizer.c (gate_increase_alignment): Fix return type.
1605 * ipa.c (function_and_variable_visibility): Fix return type.
1607 2007-01-08 Richard Guenther <rguenther@suse.de>
1609 * tree-ssa-ccp.c (maybe_fold_offset_to_array_ref): Use type
1610 of offset to build the index.
1611 * tree-pretty-print.c (dump_generic_node): Don't build negated
1612 const just for printing.
1613 * c-pretty-print.c (pp_c_integer_constant): Likewise.
1614 * builtins.c (fold_builtin_int_roundingfn): Check if result
1615 fits the type by using force_fit_type and comparing the result.
1616 * predict.c (predict_loops): Use compare_tree_int for comparison.
1617 * tree.c (build_int_cst): Fall back to integer_type_node for
1619 (build_int_cst_wide): Assert type is non-null.
1621 2007-01-08 Roberto Costa <roberto.costa@st.com>
1623 * tree-vrp.c (extract_range_from_cond_expr): New.
1624 (extract_range_from_expr): Handle COND_EXPR nodes used as expressions.
1625 * tree-ssa-ccp.c (get_maxval_strlen): Handle COND_EXPR nodes used
1627 (fold_stmt): Bug fix, avoid infinite recursion when folding COND_EXPRs.
1628 * tree-ssa-forwprop.c (simplify_cond, forward_propagate_into_cond,
1629 tree_ssa_forward_propagate_single_use_vars): Handle COND_EXPR nodes
1630 used as expressions.
1631 * tree-object-size.c (cond_expr_object_size): New.
1632 (collect_object_sizes_for): Handle COND_EXPR nodes used as expressions.
1634 2007-01-08 Jan Hubicka <jh@suse.cz>
1636 * tree-ssa-forwprop.c (forward_propagate_into_cond,
1637 tree_ssa_forward_propagate_single_use_va): Release defs of propagated
1640 2007-01-08 Richard Guenther <rguenther@suse.de>
1642 PR tree-optimization/23603
1643 * tree-vrp.c (set_value_range_to_truthvalue): New function.
1644 (extract_range_from_binary): Fall back to truthvalue instead of
1645 varying for TRUTH_*_EXPR.
1646 (extract_range_from_comparison): Fall back to truthvalue instead of
1648 (vrp_visit_phi_node): Don't adjust new range bounds to +INF/-INF
1649 if all visited PHI values were constant.
1651 2007-01-08 Jan Hubicka <jh@suse.cz>
1653 * cgraphunit.c (cgraph_process_new_functions): Reset reachable flag.
1654 (cgraph_analyze_function): break out from ...
1655 (cgraph_finalize_compilation_unit): ... here.
1656 (cgraph_expand_function): Remove forgoten commented out line.
1657 (cgraph_optimize): Analyze functions.
1659 2007-01-08 Jan Hubicka <jh@suse.cz>
1661 * tree-pas.h (TODO_remove_function): New flag.
1662 (TODO_update*): Renumber.
1663 (pass_ipa_increase_alignment,
1664 pass_ipa_function_and_variable_visibility): New passes.
1665 * cgraphunit.c (cgraph_increase_alignment): Move to tree-vectorizer.c
1666 (cgraph_function_and_variable_visibility): Move to ipa.c
1667 (cgraph_optimize): Don't call cgraph_function_and_variable_visibility,
1668 cgraph_increase_alignment.
1669 * ipa-inline.c (cgraph_decide_inlining): Don't push timevar.
1670 (cgraph_decide_inlining_incrementally): Push TV_INTEGRATION before
1671 calling tree-inline.
1672 (cgraph_early_inlining): Do not call cgraph_remove_unreachable_nodes.
1673 (pass_ipa_inline, pass_early_ipa_inlining): Set TODO_remove_functions
1674 * tree-vectorizer.c (increase_alignment): Move here from cgraphunit.c
1675 (gate_increase_alignment): New function.
1676 (pass_ipa_increase_alignment): New pass.
1677 * ipa.c: Inline tree-pass.h and timevar.h
1678 (function_and_variable_visibility): Move here from cgraphunit.c
1679 * tree-optimize.c (pass_early_local_passes): Add TODO_remove_functions.
1680 * passes.c (init_optimization_passes): Add the two new passes.
1681 (execute_todo): Handle cgraph_remove_functions.
1683 2007-01-08 Nick Clifton <nickc@redhat.com>
1685 * config/frv/predicates.md (reg_or_0_operand): Accept
1688 2007-01-08 Ralf Corsépius <ralf.corsepius@rtems.org>
1690 * config/bfin/rtems.h, config/bfin/t-rtems: New.
1691 * config.gcc: Add bfin*-rtems*.
1693 2007-01-08 Mark Shinwell <shinwell@codesourcery.com>
1695 * c.opt: Add -flax-vector-conversions.
1696 * c-typeck.c (convert_for_assignment): Pass flag to
1697 vector_types_convertible_p to allow emission of note.
1698 (digest_init): Likewise.
1699 * c-opts.c: Handle -flax-vector-conversions.
1700 * c-common.c (flag_lax_vector_conversions): New.
1701 (vector_types_convertible_p): Unless -flax-vector conversions
1702 has been passed, disallow conversions between vectors with
1703 differing numbers of subparts and/or element types. If such
1704 a conversion is disallowed, possibly emit a note on the first
1705 occasion only to inform the user of -flax-vector-conversions.
1706 The new last argument specifies this.
1707 * c-common.h (flag_lax_vector_conversions): New.
1708 (vector_types_convertible_p): Add extra argument.
1709 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use
1710 char_type_node for V*QI type vectors.
1711 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
1712 Update to satisfy new typechecking rules.
1713 * config/rs6000/altivec.h (vec_cmple): Use vec_cmpge, for both
1715 * doc/invoke.texi (C Dialect Options): Document
1716 -flax-vector-conversions.
1718 2007-01-08 Mark Shinwell <shinwell@codesourcery.com>
1720 PR tree-optimization/29877
1721 * tree-ssa-ter.c (is_replaceable_p): Deem assignments with
1722 a register variable on the RHS to not be replaceable.
1724 2007-01-08 Chen Liqin <liqin@sunnorth.com.cn>
1725 * config/score/t-score-elf (MULTILIB_OPTIONS): Change.
1726 * config/score/predicates.md (const_uimm5, sr0_operand, const_simm12,
1727 const_simm15, const_pow2, const_npow2): Added.
1728 * config/score/misc.md (insv, extv, extzv, movmemsi,
1729 move_lbu_a/b, mov_lhu_a/b etc): Added and fix some bug.
1730 * config/score/score.c (score_address_cost, score_select_cc_mode):
1732 Change CONST_OK_FOR_LETTER_P/EXTRA_CONSTRAINT define.
1733 Update score_rtx_costs for MACRO TARGET_RTX_COSTS.
1734 Update score_print_operand.
1735 * config/score/score.h (DATA_ALIGNMENT, SELECT_CC_MODE): Added.
1736 Adjust register allocate order and update some macro define.
1737 * config/score/score-mdaux.c (mdx_unaligned_load, mdx_unsigned_store,
1738 mdx_block_move_straight, mdx_block_move_loop_head,
1739 mdx_block_move_loop_body, mdx_block_move_loop_foot, mdx_block_move_loop,
1740 mdx_block_move): Added.
1741 (mdx_movsicc, mdp_select_add_imm, mdp_select, mds_zero_extract_andi,
1742 mdp_limm): Updated and fix some bug and typo.
1743 * config/score/score.md (movqi/hi/si, add/sub/zero/ext): Updated.
1744 (movsf, movdf, doloop_end): Added.
1746 2007-01-08 Kazu Hirata <kazu@codesourcery.com>
1748 * config/arm/arm.c, config/arm/arm.h, config/arm/arm.md,
1749 config/arm/thumb2.md: Fix comment typos.
1750 * doc/extend.texi: Fix a typo.
1752 2007-01-07 Eric Christopher <echristo@apple.com>
1754 * configure.ac: Check for __stack_chk_fail for darwin.
1755 * configure: Regenerate.
1757 2007-01-07 Richard Guenther <rguenther@suse.de>
1759 * tree-vrp.c (extract_range_from_assert): CSE calls to
1760 compare_values where possible.
1761 (extract_range_from_unary_expr): Likewise.
1763 2007-01-07 Anatoly Sokolov <aesok@post.ru>
1765 * config/avr/avr-protos.h (call_insn_operand): Delete prototype.
1766 * config/avr/avr.c (call_insn_operand): Delete function.
1767 * config/avr/avr.md (*pushqi, *pushhi, *pushsi, *pushsf): Use REG_SP
1768 instead of register number. Use predicates.
1769 * config/avr/predicates.md (const0_operand, reg_or_0_operand,
1770 call_insn_operand): Add.
1772 2007-01-06 Jan Hubicka <jh@suse.cz>
1774 * tree-pass.h (pass_build_cgraph_edges): Declare.
1775 * cgraphunit.c (record_refernece): Move to cgraphbuild.c
1776 (visited_nodes): Remove.
1777 (cgraph_create_edges): Move to cgraphbuild.c; rename to
1778 build_cgrpah_edges; make visited_nodes local.
1779 (cgraph_process_new_functions): DO not call initialize_inline_failed.
1780 (record_references_in_initializer): Move to cgraphbuild.c
1781 (initialize_inline_failed, rebuild_cgraph_edges,
1782 pass_rebuild_cgraph_edges): Move to cgraphbuild.c.
1783 (verify_cgraph_node): Make visited_nodes local.
1784 (cgraph_analyze_function): Do not call cgraph_create_edges and
1785 initialize_inline_failed.
1786 (cgraph_expand_function): Do not call cgraph_lower_function;
1787 assert that function is already lowered.
1788 * Makefile.in (cgraphbuild.o): New.
1789 * passes.c (init_optimization_passes): Add pass_build_cgraph_edges
1790 at the end of lowering passes.
1792 2007-01-06 Steven Bosscher <steven@gcc.gnu.org>
1794 * ifcvt.c (cond_move_convert_if_block): New function, code
1795 factored out from...
1796 (cond_move_process_if_block): ...here. Call the new function
1797 on the THEN and ELSE blocks.
1798 (merge_if_block): Do not copy global_live_at_end, merge_blocks
1799 already takes care of this.
1801 2007-01-05 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
1804 * tree.h (TREE_OVERFLOW_P): New.
1805 * c-typeck.c (parser_build_unary_op): Warn only if result
1806 overflowed and operands did not.
1807 (parser_build_binary_op): Likewise.
1808 (convert_for_assignment): Remove redundant overflow_warning.
1809 * c-common.c (overflow_warning): Don't check or set TREE_OVERFLOW.
1811 2007-01-05 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
1813 * c-typeck.c (store_init_value): Split over two lines to follow
1814 the GNU coding style.
1816 2007-01-05 Benjamin Kosnik <bkoz@redhat.com>
1818 * c-cppbuiltin.c (c_cpp_builtins): __GXX_EXPERIMENTAL_CPP0X__ to
1819 __GXX_EXPERIMENTAL_CXX0X__.
1820 * doc/cpp.texi: Same.
1822 2007-01-05 Richard Guenther <rguenther@suse.de>
1825 * tree.c (get_narrower): Do not construct COMPONENT_REFs
1826 with mismatched types. Instead explicitly build a
1827 conversion NOP_EXPR.
1829 2007-01-05 Ian Lance Taylor <iant@google.com>
1831 * c-common.c (decl_with_nonnull_addr_p): New function.
1832 (c_common_truthvalue_conversion): Call it.
1833 * c-typeck.c (build_binary_op): Likewise.
1834 * c-common.h (decl_with_nonnull_addr_p): Declare.
1836 2007-01-05 Jakub Jelinek <jakub@redhat.com>
1839 * libgcc2.c (__divdc3): Compare c and d against 0.0 instead of
1842 2007-01-05 Joel Brobecker <brobecker@adacore.com>
1844 * doc/install.texi (Final install): Document the fact that
1845 the GNAT runtime should not be stripped.
1847 2007-01-04 Jan Hubicka <jh@suse.cz>
1849 * tree-inline.c (fold_marked_statements): Update operand caches
1850 and EH after folding
1852 2007-01-04 Ian Lance Taylor <iant@google.com>
1854 * c-common.c (check_function_nonnull): Whitespace fix.
1856 2007-01-04 Jan Hubicka <jh@suse.cz>
1858 * tree-optimize.c (execute_fixup_cfg): Correct previously mistakely
1859 comitted older version of patch.
1860 (pass_fixup_cfg): Add TODOs to verify flow and statements, dump
1861 function, celanup cfg and collect garbage.
1863 2007-01-04 Mike Stump <mrs@apple.com>
1865 * Makefile.in (mostlyclean): Don't remove libgcc anymore.
1868 2007-01-04 Eric Christopher <echristo@apple.com>
1870 * libgcc2.c (__bswapsi2): Use SItype.
1871 (__bswapdi2): Use DItype.
1872 * libgcc2.h: Update for above.
1874 2007-01-04 Paul Brook <paul@codesourcery.com>
1876 * config/arm/arm.md (arm_mulsi3, thumb_mulsi3, mulsi3_compare0,
1877 mulsi_compare0_scratch, mulsi3addsi, mulsi3addsi_compare0,
1878 mulsi3addsi_compare0_scratch, mulsidi3adddi, mulsidi3,
1879 umulsidi3, umulsidi3adddi, smulsi3_highpart,
1880 umulsi3_highpart): Make conditional on !arm_arch6.
1881 (arm_mulsi3_v6, thumb_mulsi3_v6, mulsi3_compare0_v6,
1882 mulsi_compare0_scratch_v6, mulsi3addsi_v6, mulsi3addsi_compare0_v6,
1883 mulsi3addsi_compare0_scratch_v6, mulsidi3adddi_v6, mulsidi3_v6,
1884 umulsidi3_v6, umulsidi3adddi_v6, smulsi3_highpart_v6,
1885 umulsi3_highpart_v6): New insns.
1887 2007-01-04 Roger Sayle <roger@eyesopen.com>
1889 * fold-const.c (fold_convert): When casting an expression to void,
1890 fold_ignored_result may discover a GIMPLE_MODIFY_STMT which doesn't
1891 have a type. Instead of attempting to build a NOP_EXPR, return
1892 these "special" trees directly.
1894 2007-01-04 Joseph Myers <joseph@codesourcery.com>
1896 * config/rs6000/rs6000.c (rs6000_rtx_costs): Make adjustment for
1897 MULT inside MINUS as either argument. Use rs6000_cost->dmul -
1898 rs6000_cost->fp not 0 as adjustment for outer NEG.
1900 2007-01-04 Jan Hubicka <jh@suse.cz>
1902 * cgraph.c (cgraph_release_function_body): New function.
1903 (cgraph_remove_node): Use it.
1904 * cgraph.h (cgraph_release_function_body): Declare.
1905 * cgraphunit.c (cgraph_expand_function): Use it.
1906 * ipa.c (cgraph_remove_unreahchable_nodes): Use it.
1907 * tree-ssa.c (delete_tree_ssa): Allow to be called before aliasing
1908 is initialized and while compilation of other function is running.
1909 * tree-optimize.c (execute_free_cfg_annotations): Move code to clear
1910 statement CFG annotations from here to ...
1911 * tree-cfg.c (delete_tree_cfg_annotations): ... here.
1913 2007-01-04 Zdenek Dvorak <dvorakz@suse.cz>
1915 * cfgloop.h (enum li_flags): Make the constants powers of two.
1917 2007-01-04 Jan Hubicka <jh@suse.cz>
1919 * tree-inline.c (copy_bb): Insert new statements to statements_to_fold
1921 (fold_marked_statements): New function.
1922 (optimize_inline_calls, tree_function_versioning): Fold new statements.
1923 * tree-inline.h (copy_body_data): Add statements_to_fold.
1925 2007-01-03 Daniel Jacobowitz <dan@codesourcery.com>
1927 * config.gcc: Mention libgcc/config.host.
1928 * Makefile.in: Update comments mentioning libgcc.
1929 (LIBGCC, INSTALL_LIBGCC, GCC_PARTS, mklibgcc): Delete.
1930 (all.cross, start.encap, rest.encap, rest.cross): Update
1931 dependencies for libgcc move.
1932 (libgcc.mk, LIBGCC_DEPS, libgcov.a, libgcc.a, stmp-multilib)
1933 (clean-target, clean-target-libgcc): Delete.
1934 (srcdirify, GCC_EXTRA_PARTS): New macros.
1935 (libgcc-support, libgcc.mvars): New rules.
1936 (distclean): Remove mention of mklibgcc.
1937 (install): Don't reference INSTALL_LIBGCC.
1938 (install-common): Don't reference EXTRA_PARTS.
1939 (install-libgcc, install-multilib): Delete rules.
1940 * mklibgcc.in: Delete file.
1941 * doc/configfiles.texi: Don't mention mklibgcc.
1943 * config/i386/t-darwin (SHLIB_VERPFX): Delete (moved to libgcc).
1944 * config/i386/t-darwin64 (SHLIB_VERPFX): Likewise.
1945 * config/rs6000/t-darwin (SHLIB_VERPFX): Likewise.
1946 * config/rs6000/t-ppccomm (TARGET_LIBGCC2_CFLAGS, SHLIB_MAPFILES)
1947 (mklibgcc, ldblspecs): Likewise.
1949 * config/i386/t-nwld (libgcc.def, libc.def, libpcre.def)
1950 (posixpre.def): Use $(T).
1951 (SHLIB_EXT, SHLIB_NAME, SHLIB_SLIBDIR_QUAL, SHLIB_DEF, SHLIB_MAP)
1952 (SHLIB_SRC, SHLIB_INSTALL): Delete.
1953 (SHLIB_LINK): Make dummy.
1954 * config/t-slibgcc-darwin: Delete contents except for dummy SHLIB_LINK.
1956 * config/frv/t-linux (EXTRA_MULTILIB_PARTS): Clear.
1958 * config/alpha/t-crtfm: Use $(T) in rules for EXTRA_PARTS.
1959 * config/alpha/t-vms, config/alpha/t-vms64, config/fr30/t-fr30,
1960 config/i386/t-rtems-i386, config/ia64/t-ia64, config/rs6000/t-beos,
1961 config/rs6000/t-newas, config/sparc/t-elf: Likewise.
1963 * configure.ac (all_outputs): Remove mklibgcc.
1964 * configure: Regenerated.
1966 2007-01-03 Josh Conner <jconner@apple.com>
1969 * calls.c (compute_argument_addresses): Set stack and stack_slot
1970 for partial args, too.
1971 (store_one_arg): Use locate.size.constant for the size when
1972 generating a save_area.
1974 2007-01-03 Robert Kennedy <jimbob@google.com>
1976 * tree-cfg.c (tree_merge_blocks): Release SSA_NAME phi results
1977 whose definitions are deleted due to basic block merging.
1979 2007-01-03 Paul Brook <paul@codesourcery.com>
1982 * config/arm/arm.c (output_return_instruction): Pop PC in interrupt
1984 (use_return_insn): Return 0 for Thumb interrupt functions.
1985 (print_multi_reg): Add rfe argument for IRQ returns.
1986 (arm_output_epilogue): Pop interrupt return address directly into PC.
1987 (arm_expand_prologue): Only adjust IRQ return address in Arm mode.
1989 2007-01-03 Paul Brook <paul@codesourcery.com>
1991 Merge from sourcerygxx-4_1.
1992 * config/arm/thumb2.md: New file.
1993 * config/arm/elf.h (JUMP_TABLES_IN_TEXT_SECTION): Return True for
1995 * config/arm/coff.h (JUMP_TABLES_IN_TEXT_SECTION): Ditto.
1996 * config/arm/aout.h (ASM_OUTPUT_ADDR_VEC_ELT): Add !Thumb-2 assertion.
1997 (ASM_OUTPUT_ADDR_DIFF_ELT): Output Thumb-2 jump tables.
1998 * config/arm/aof.h (ASM_OUTPUT_ADDR_DIFF_ELT): Output Thumb-2 jump
2000 (ASM_OUTPUT_ADDR_VEC_ELT): Add !Thumb-2 assertion.
2001 * config/arm/ieee754-df.S: Use macros for Thumb-2/Unified asm
2003 * config/arm/ieee754-sf.S: Ditto.
2004 * config/arm/arm.c (thumb_base_register_rtx_p): Rename...
2005 (thumb1_base_register_rtx_p): ... to this.
2006 (thumb_index_register_rtx_p): Rename...
2007 (thumb1_index_register_rtx_p): ... to this.
2008 (thumb_output_function_prologue): Rename...
2009 (thumb1_output_function_prologue): ... to this.
2010 (thumb_legitimate_address_p): Rename...
2011 (thumb1_legitimate_address_p): ... to this.
2012 (thumb_rtx_costs): Rename...
2013 (thumb1_rtx_costs): ... to this.
2014 (thumb_compute_save_reg_mask): Rename...
2015 (thumb1_compute_save_reg_mask): ... to this.
2016 (thumb_final_prescan_insn): Rename...
2017 (thumb1_final_prescan_insn): ... to this.
2018 (thumb_expand_epilogue): Rename...
2019 (thumb1_expand_epilogue): ... to this.
2020 (arm_unwind_emit_stm): Rename...
2021 (arm_unwind_emit_sequence): ... to this.
2022 (thumb2_legitimate_index_p, thumb2_legitimate_address_p,
2023 thumb1_compute_save_reg_mask, arm_dwarf_handle_frame_unspec,
2024 thumb2_index_mul_operand, output_move_vfp, arm_shift_nmem,
2025 arm_save_coproc_regs, thumb_set_frame_pointer, arm_print_condition,
2026 thumb2_final_prescan_insn, thumb2_asm_output_opcode, arm_output_shift,
2027 thumb2_output_casesi): New functions.
2028 (TARGET_DWARF_HANDLE_FRAME_UNSPEC): Define.
2029 (FL_THUMB2, FL_NOTM, FL_DIV, FL_FOR_ARCH6T2, FL_FOR_ARCH7,
2030 FL_FOR_ARCH7A, FL_FOR_ARCH7R, FL_FOR_ARCH7M, ARM_LSL_NAME,
2031 THUMB2_WORK_REGS): Define.
2032 (arm_arch_notm, arm_arch_thumb2, arm_arch_hwdiv, arm_condexec_count,
2033 arm_condexec_mask, arm_condexec_masklen)): New variables.
2034 (all_architectures): Add armv6t2, armv7, armv7a, armv7r and armv7m.
2035 (arm_override_options): Check new CPU capabilities.
2036 Set new architecture flag variables.
2037 (arm_isr_value): Handle v7m interrupt functions.
2038 (user_return_insn): Return 0 for v7m interrupt functions. Handle
2040 (const_ok_for_arm): Handle Thumb-2 constants.
2041 (arm_gen_constant): Ditto. Use movw when available.
2042 (arm_function_ok_for_sibcall): Return false for v7m interrupt
2044 (legitimize_pic_address, arm_call_tls_get_addr): Handle Thumb-2.
2045 (thumb_find_work_register, arm_load_pic_register,
2046 legitimize_tls_address, arm_address_cost, load_multiple_sequence,
2047 emit_ldm_seq, emit_stm_seq, arm_select_cc_mode, get_jump_table_size,
2048 print_multi_reg, output_mov_long_double_fpa_from_arm,
2049 output_mov_long_double_arm_from_fpa, output_mov_double_fpa_from_arm,
2050 output_mov_double_fpa_from_arm, output_move_double,
2051 arm_compute_save_reg_mask, arm_compute_save_reg0_reg12_mask,
2052 output_return_instruction, arm_output_function_prologue,
2053 arm_output_epilogue, arm_get_frame_offsets, arm_regno_class,
2054 arm_output_mi_thunk, thumb_set_return_address): Ditto.
2055 (arm_expand_prologue): Handle Thumb-2. Use arm_save_coproc_regs.
2056 (arm_coproc_mem_operand): Allow POST_INC/PRE_DEC.
2057 (arithmetic_instr, shift_op): Use arm_shift_nmem.
2058 (arm_print_operand): Use arm_print_condition. Handle '(', ')', '.',
2060 (arm_final_prescan_insn): Use extract_constrain_insn_cached.
2061 (thumb_expand_prologue): Use thumb_set_frame_pointer.
2062 (arm_file_start): Output directive for unified syntax.
2063 (arm_unwind_emit_set): Handle stack alignment instruction.
2064 * config/arm/lib1funcs.asm: Remove default for __ARM_ARCH__.
2065 Add v6t2, v7, v7a, v7r and v7m.
2066 (RETLDM): Add Thumb-2 code.
2067 (do_it, shift1, do_push, do_pop, COND, THUMB_SYNTAX): New macros.
2068 * config/arm/arm.h (TARGET_CPU_CPP_BUILTINS): Define __thumb2__.
2069 (TARGET_THUMB1, TARGET_32BIT, TARGET_THUMB2, TARGET_DSP_MULTIPLY,
2070 TARGET_INT_SIMD, TARGET_UNIFIED_ASM, ARM_FT_STACKALIGN, IS_STACKALIGN,
2071 THUMB2_TRAMPOLINE_TEMPLATE, TRAMPOLINE_ADJUST_ADDRESS,
2072 ASM_OUTPUT_OPCODE, THUMB2_GO_IF_LEGITIMATE_ADDRESS,
2073 THUMB2_LEGITIMIZE_ADDRESS, CASE_VECTOR_PC_RELATIVE,
2074 CASE_VECTOR_SHORTEN_MODE, ADDR_VEC_ALIGN, ASM_OUTPUT_CASE_END,
2075 ADJUST_INSN_LENGTH): Define.
2076 (TARGET_REALLY_IWMMXT, TARGET_IWMMXT_ABI, CONDITIONAL_REGISTER_USAGE,
2077 STATIC_CHAIN_REGNUM, HARD_REGNO_NREGS, INDEX_REG_CLASS,
2078 BASE_REG_CLASS, MODE_BASE_REG_CLASS, SMALL_REGISTER_CLASSES,
2079 PREFERRED_RELOAD_CLASS, SECONDARY_OUTPUT_RELOAD_CLASS,
2080 SECONDARY_INPUT_RELOAD_CLASS, LIBCALL_VALUE, FUNCTION_VALUE_REGNO_P,
2081 TRAMPOLINE_SIZE, INITIALIZE_TRAMPOLINE, HAVE_PRE_INCREMENT,
2082 HAVE_POST_DECREMENT, HAVE_PRE_DECREMENT, HAVE_PRE_MODIFY_DISP,
2083 HAVE_POST_MODIFY_DISP, HAVE_PRE_MODIFY_REG, HAVE_POST_MODIFY_REG,
2084 REGNO_MODE_OK_FOR_BASE_P, LEGITIMATE_CONSTANT_P,
2085 REG_MODE_OK_FOR_BASE_P, REG_OK_FOR_INDEX_P, GO_IF_LEGITIMATE_ADDRESS,
2086 LEGITIMIZE_ADDRESS, THUMB2_LEGITIMIZE_ADDRESS,
2087 GO_IF_MODE_DEPENDENT_ADDRESS, MEMORY_MOVE_COST, BRANCH_COST,
2088 ASM_APP_OFF, ASM_OUTPUT_CASE_LABEL, ARM_DECLARE_FUNCTION_NAME,
2089 FINAL_PRESCAN_INSN, PRINT_OPERAND_PUNCT_VALID_P,
2090 PRINT_OPERAND_ADDRESS): Adjust for Thumb-2.
2091 (arm_arch_notm, arm_arch_thumb2, arm_arch_hwdiv): New declarations.
2092 * config/arm/arm-cores.def: Add arm1156t2-s, cortex-a8, cortex-r4 and
2094 * config/arm/arm-tune.md: Regenerate.
2095 * config/arm/arm-protos.h: Update prototypes.
2096 * config/arm/vfp.md: Enable patterns for Thumb-2.
2097 (arm_movsi_vfp): Add movw alternative. Use output_move_vfp.
2098 (arm_movdi_vfp, movsf_vfp, movdf_vfp): Use output_move_vfp.
2099 (thumb2_movsi_vfp, thumb2_movdi_vfp, thumb2_movsf_vfp,
2100 thumb2_movdf_vfp, thumb2_movsfcc_vfp, thumb2_movdfcc_vfp): New.
2101 * config/arm/libunwind.S: Add Thumb-2 code.
2102 * config/arm/constraints.md: Update include Thumb-2.
2103 * config/arm/ieee754-sf.S: Add Thumb-2/Unified asm support.
2104 * config/arm/ieee754-df.S: Ditto.
2105 * config/arm/bpabi.S: Ditto.
2106 * config/arm/t-arm (MD_INCLUDES): Add thumb2.md.
2107 * config/arm/predicates.md (low_register_operand,
2108 low_reg_or_int_operand, thumb_16bit_operator): New.
2109 (thumb_cmp_operand, thumb_cmpneg_operand): Rename...
2110 (thumb1_cmp_operand, thumb1_cmpneg_operand): ... to this.
2111 * config/arm/t-arm-elf: Add armv7 multilib.
2112 * config/arm/arm.md: Update patterns for Thumb-2 and Unified asm.
2114 (UNSPEC_STACK_ALIGN, ce_count): New.
2115 (arm_incscc, arm_decscc, arm_umaxsi3, arm_uminsi3,
2116 arm_zero_extendsidi2, arm_zero_extendqidi2): New
2118 * config/arm/fpa.md: Update patterns for Thumb-2 and Unified asm.
2119 (thumb2_movsf_fpa, thumb2_movdf_fpa, thumb2_movxf_fpa,
2120 thumb2_movsfcc_fpa, thumb2_movdfcc_fpa): New insns.
2121 * config/arm/cirrus.md: Update patterns for Thumb-2 and Unified asm.
2122 (cirrus_thumb2_movdi, cirrus_thumb2_movsi_insn,
2123 thumb2_cirrus_movsf_hard_insn, thumb2_cirrus_movdf_hard_insn): New
2125 * doc/extend.texi: Document ARMv7-M interrupt functions.
2126 * doc/invoke.texi: Document Thumb-2 new cores+architectures.
2128 2007-01-03 Jakub Jelinek <jakub@redhat.com>
2130 * unwind-dw2.c (SIGNAL_FRAME_BIT, EXTENDED_CONTEXT_BIT): Define.
2131 (struct _Unwind_Context): Rename args_size to flags, remove
2132 signal_frame field, add a new args_size field and version field.
2133 (_Unwind_IsSignalFrame, _Unwind_SetSignalFrame,
2134 _Unwind_IsExtendedContext): New inline functions.
2135 (_Unwind_GetGR, _Unwind_SetGR, _Unwind_GetGRPtr, _Unwind_SetGRPtr):
2136 Assume by_value array is only present if _Unwind_IsExtendedContext.
2137 (_Unwind_GetIPInfo, execute_cfa_program, uw_frame_state_for): Use
2138 _Unwind_IsSignalFrame.
2139 (__frame_state_for): Initialize context.flags to EXTENDED_CONTEXT_BIT.
2140 (uw_update_context_1): Use _Unwind_SetSignalFrame.
2141 (uw_init_context_1): Initialize context->flags to
2142 EXTENDED_CONTEXT_BIT.
2143 * config/rs6000/linux-unwind.h (frob_update_context): Use
2144 _Unwind_SetSignalFrame.
2146 2007-01-03 Andrew Pinski <andrew_pinski@playstation.sony.com>
2149 * gimplify.c (gimplify_modify_expr_complex_part): Move below
2150 tree_to_gimple_tuple. Call tree_to_gimple_tuple when we need
2153 2007-01-03 Kazu Hirata <kazu@codesourcery.com>
2155 * config/i386/i386.h (NON_STACK_REG_P, REGNO_OK_FOR_SIREG_P,
2156 REGNO_OK_FOR_DIREG_P, REWRITE_ADDRESS, ASM_OPERAND_LETTER,
2157 RET, AT_SP): Remove.
2158 * config/i386/i386.md (*sse_prologue_save_insn): Use return
2161 * alias.c (init_alias_analysis): Use VEC_safe_grow_cleared.
2162 * cfgbuild.c (find_basic_blocks): Likewise.
2163 * cfgrtl.c (rtl_create_basic_block): Likewise.
2164 * function.c (temp_slots_at_level): Likewise.
2165 * reg-stack.c (stack_regs_mentioned): Likewise.
2166 * regclass.c (allocate_reg_info): Likewise.
2167 * tree-cfg.c (init_empty_tree_cfg, build_tree_cfg, create_bb,
2168 set_bb_for_stmt, move_block_to_fn): Likewise.
2169 * tree-complex.c (tree_lower_complex): Likewise.
2170 * vec.h (VEC_safe_grow_cleared): New.
2172 * cgraphunit.c, tree-ssa-alias.c: Fix comment typos.
2174 2007-01-03 Zdenek Dvorak <dvorakz@suse.cz>
2176 * loop-unswitch.c (unswitch_loop): Pass probabilities to loopify.
2177 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Pass probabilities
2179 * cfgloopmanip.c (scale_loop_frequencies): Export.
2180 (loopify): Scale the frequencies by prescribed coefficients.
2181 (set_zero_probability): New function.
2182 (duplicate_loop_to_header_edge): Improve updating of frequencies.
2183 (lv_adjust_loop_entry_edge, loop_version): Set probabilities
2184 and frequencies according to arguments.
2185 * tree-ssa-loop-manip.c (tree_unroll_loop): Set probabilities
2187 * cfg.c (scale_bbs_frequencies_int): Allow scaling the frequencies up.
2188 * modulo-sched.c (sms_schedule): Set probabilities for entering
2189 versioned loop correctly.
2190 * tree-vect-transform.c (vect_transform_loop): Ditto.
2191 * cfgloop.h (loopify, loop_version): Declaration changed.
2192 (scale_loop_frequencies): Declared.
2194 2007-01-02 Jan Hubicka <jh@suse.cz>
2196 * cgraph.c: Include tree-flow.h
2197 (cgraph_add_new-function): Handle IPA_SSA mode; execute
2199 * cgraph.h (enum cgraph_state): Add CGRAPH_STATE_IPA_SSA.
2200 * tree-pass.h (pass_all_early_optimizations): Declare.
2201 * cgraphunit.c (cgraph_process_new_functions): Add IPA_SSA; execute
2203 (cgraph_analyze_function): Do early_local_passes.
2204 * tree-mudflap.c (mf_decl_cache_locals, mf_build_check_statement_for):
2205 Do not add referenced vars.
2206 * tree-optimize.c (gate_all_optimizations): Do not execute when not in
2208 (gate_all_early_local_passes): New gate.
2209 (pass_early_local_passes): Use new gate.
2210 (execute_early_local_optimizations): New functions.
2211 (gate_all_early_optimizations): New gate.
2212 (pass_all_early_optimizations): New pass.
2213 (execute_free_datastructures): Free SSA only when initialized.
2214 (gate_init_datastructures): Init only when optimizing.
2215 (tree_lowering_passes): Do early local passes when called late.
2216 * tree-profile.c (do_tree_profiling): Don't profile functions added
2218 (do_early_tree_profiling, pass_early_tree_profile): Kill.
2219 * tree-cfg.c (update_modified_stmts): Do not update when operands are
2221 * passes.c (init_optimizations_passes): Reorder so we go into SSA
2222 during early_local_passes.
2223 * Makefile.in (cgraph.o): Add dependency on tree-flow.h.
2226 2007-01-02 Carlos O'Donell <carlos@codesourcery.com>
2228 * Makefile.in: Update copyright year.
2230 2007-01-02 Carlos O'Donell <carlos@codesourcery.com>
2232 * Makefile.in: Export GCC_EXEC_PREFIX before calling $(RUNTEST)
2233 in $(lang_checks) and check-consistency targets.
2235 2007-01-02 Jan Hubicka <jh@suse.cz>
2237 * tree-mudflap.c (mf_decl_cache_locals, mf_build_check_statement_for):
2238 Do not add referenced vars.
2239 * tree-cfg.c (update_modified_stmts): Do not update when SSA operands
2241 * passes.c (init_optimization_passes): Put mudflap_2 after
2242 free_datastructures.
2244 2007-01-02 Jan Hubicka <jh@suse.cz>
2246 * tree-optimize (execute_fixup_cfg): Set after_inlining flag.
2247 Set NOTHROW flag on call statements proved to be nothrow.
2248 Update statement of local calls so new pure/const functions are
2249 updated. Update_ssa when in ssa form. Mark PHI nodes of nonlocal
2251 (tree_rest_of_compilation): Register hooks and initialize bitmap
2252 early. Do not set after_inlining flag.
2254 2007-01-02 Steve Ellcey <sje@cup.hp.com>
2256 * sbitmap.c (HOST_BITS_PER_LONG_LONG): Change to
2257 HOST_BITS_PER_LONGLONG
2259 2007-01-02 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
2262 * c-typeck.c (store_init_value): Don't emit pedantic overflow
2263 warning for non-static initializers.
2265 2007-01-02 Steven Bosscher <steven@gcc.gnu.org>
2267 * config/alpha/alpha.md, arm/arm.c, darwin.c, frv/frv.md,
2268 m32r/m32r.c, m32r/m32r.c, mn10300/mn10300.md, pa/pa.c,
2269 rs6000/rs6000.c, s390/s390.md, sh/sh.md, sparc/sparc.c:
2270 Always use set_unique_reg_note to add REG_EQUAL notes.
2272 2007-01-02 Kazu Hirata <kazu@codesourcery.com>
2275 2007-01-02 Kazu Hirata <kazu@codesourcery.com>
2277 * alias.c (init_alias_analysis): Use VEC_safe_grow_cleared.
2278 * cfgbuild.c (find_basic_blocks): Likewise.
2279 * cfgrtl.c (rtl_create_basic_block): Likewise.
2280 * function.c (temp_slots_at_level): Likewise.
2281 * reg-stack.c (stack_regs_mentioned): Likewise.
2282 * regclass.c (allocate_reg_info): Likewise.
2283 * tree-cfg.c (init_empty_tree_cfg, build_tree_cfg, create_bb,
2284 set_bb_for_stmt, move_block_to_fn): Likewise.
2285 * tree-complex.c (tree_lower_complex): Likewise.
2286 * vec.h (VEC_safe_grow_cleared): New.
2288 2007-01-02 Ian Lance Taylor <iant@google.com>
2290 * c-common.c (c_common_truthvalue_conversion): When warning about
2291 using an assignment as a truth value, set TREE_NO_WARNING.
2293 2007-01-02 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
2296 * c.opt (Wold-style-declaration): New.
2297 * doc/invoke.texi (C-only Warning Options): New.
2298 (Wold-style-declaration): Document it.
2299 (Wextra): Enabled by -Wextra.
2300 * c-opts.c (c_common_post_options): Enabled by -Wextra.
2301 * c-decl.c (declspecs_add_scspec): Replace -Wextra with
2302 -Wold-style-declaration.
2304 2007-01-02 Kazu Hirata <kazu@codesourcery.com>
2306 * alias.c (init_alias_analysis): Use VEC_safe_grow_cleared.
2307 * cfgbuild.c (find_basic_blocks): Likewise.
2308 * cfgrtl.c (rtl_create_basic_block): Likewise.
2309 * function.c (temp_slots_at_level): Likewise.
2310 * reg-stack.c (stack_regs_mentioned): Likewise.
2311 * regclass.c (allocate_reg_info): Likewise.
2312 * tree-cfg.c (init_empty_tree_cfg, build_tree_cfg, create_bb,
2313 set_bb_for_stmt, move_block_to_fn): Likewise.
2314 * tree-complex.c (tree_lower_complex): Likewise.
2315 * vec.h (VEC_safe_grow_cleared): New.
2317 2007-01-02 Douglas Gregor <doug.gregor@gmail.com>
2319 * c-common.c(c_common_nodes_and_builtins): Since variants of
2320 void_type_node get built before it is given a name, we need to
2321 give those variants the name, too.
2322 (complete_array_type): We need to work with the canonical main
2323 type of the array, from which we will build the qualified version.
2324 * params.def (PARAM_VERIFY_CANONICAL_TYPES): New.
2325 * print-tree.c (print_node): Display canonical type information
2327 * stor-layout.c (layout_type): When we don't know the
2328 alignment of a type for which we're building an array, we end up
2329 guessing wrong, so make the type require structural equality.
2330 * tree.c (make_node_stat): When we build a new type, it is its
2332 (build_type_attribute_qual_variant): When building an attribute
2333 variant, its canonical type is the non-attribute variant. However,
2334 if the attributes are target-dependent and they differ, we need to
2335 use structural equality checks for this type.
2336 (build_qualified_type): A qualified type is not equivalent to its
2337 unqualified variant; set the canonical type appropriately.
2338 (build_distinct_type_copy): When building a distinct type from
2339 another type, the new type is its own canonical type.
2340 (build_variant_type_copy): When building a new type variant, we
2341 assume that it is equivalent to the original type.
2342 (build_pointer_type_for_mode): When building a pointer type, also
2343 build a canonical type pointer.
2344 (build_reference_type_for_mode): When building a reference type,
2345 also build a canonical type reference.
2346 (build_index_type): When we can't hash an index type (e.g.,
2347 because its maximum value is negative), the index type requires
2348 structural equality tests.
2349 (build_array_type): Build the canonical form of an array type.
2350 (build_function_type): Function types require structural equality,
2351 because they contain default arguments, attributes, etc.
2352 (build_method_type_directly): Ditto for method types.
2353 (build_offset_type): Build the canonical offset type.
2354 (build_complex_type): Build the canonical vector type.
2355 (make_vector_type): Build the canonical vector type.
2356 * tree.h (TYPE_CANONICAL): New.
2357 (TYPE_STRUCTURAL_EQUALITY_P): New.
2358 (SET_TYPE_STRUCTURAL_EQUALITY): New.
2359 (struct tree_type): Added "canonical" field.
2360 * params.h (VERIFY_CANONICAL_TYPES): New.
2361 * doc/c-tree.texi (TYPE_CANONICAL): Document.
2362 (TYPE_STRUCTURAL_EQUALITY_P): Document.
2363 (SET_TYPE_STRUCTURAL_EQUALITY): Document.
2364 * doc/invoke.texi (verify-canonical-types): Document --param
2365 parameter for verifying canonical types.
2367 2007-01-02 Joseph Myers <joseph@codesourcery.com>
2369 * config.gcc (powerpc-*-eabispe*, powerpc-*-eabisimaltivec*,
2370 powerpc-*-eabisim*, powerpc-*-eabialtivec*, powerpc-*-eabi*,
2371 powerpc-*-rtems*, powerpc-wrs-vxworks, powerpc-wrs-vxworksae,
2372 powerpcle-*-eabisim*, powerpcle-*-eabi*): Add rs6000/e500.h to
2374 * config/rs6000/e500.h: New.
2375 * config/rs6000/eabi.h (TARGET_SPE_ABI, TARGET_SPE, TARGET_E500,
2376 TARGET_ISEL, TARGET_FPRS, TARGET_E500_SINGLE, TARGET_E500_DOUBLE):
2378 * config/rs6000/linuxspe.h (TARGET_SPE_ABI, TARGET_SPE,
2379 TARGET_E500, TARGET_ISEL, TARGET_FPRS, TARGET_E500_SINGLE,
2380 TARGET_E500_DOUBLE): Remove.
2381 * config/rs6000/vxworks.h (TARGET_SPE_ABI, TARGET_SPE,
2382 TARGET_E500, TARGET_ISEL, TARGET_FPRS): Remove.
2383 * config/rs6000/rs6000.h (CHECK_E500_OPTIONS): Define.
2384 * config/rs6000/rs6000.c (rs6000_override_options): Use
2387 2007-01-02 Joseph Myers <joseph@codesourcery.com>
2389 * config/rs6000/rs6000.c (print_operand): Check (TARGET_SPE ||
2390 TARGET_E500_DOUBLE), not TARGET_E500, for %y.
2391 (rs6000_generate_compare, rs6000_emit_sCOND, output_cbranch,
2392 rs6000_emit_cmove): Don't check TARGET_E500.
2393 * config/rs6000/rs6000.md (bunordered, bordered, sunordered,
2394 sordered): Don't check TARGET_E500.
2396 2007-01-01 Eric Christopher <echristo@apple.com>
2398 * config/mips/mips.c (mips_regno_mode_ok_for_base_p): Use
2399 HARD_REGISTER_NUM_P.
2401 2007-01-01 Roger Sayle <roger@eyesopen.com>
2403 * fold-const.c (fold_binary) <EQ_EXPR>: Fold "(X^C1) eq/ne C2" into
2404 "X eq/ne (C1^C2)". Fold "(X^Z) eq/ne (Y^Z)" as "X eq/ne Y" when Z
2405 has no side-effects. Fold "(X^C1) eq/ne (Y^C2)" as "(X^(C1^C2))
2408 2007-01-01 Mike Stump <mrs@apple.com>
2410 * configure.ac: Remove support for building with Apple's gcc-3.1.
2412 2007-01-02 Joseph Myers <joseph@codesourcery.com>
2415 * caller-save.c (add_stored_regs): Only handle SUBREGs if inner
2416 REG is a hard register. Do not modify REG before calling
2418 * rtlanal.c (subreg_get_info): Don't assert size of XMODE is a
2419 multiple of the size of YMODE for certain lowpart cases.
2421 2007-01-01 Andrew Pinski <pinskia@gmail.com>
2424 * gimplify (voidify_wrapper_expr): Update for
2427 2007-01-01 Andreas Schwab <schwab@suse.de>
2430 * config/ia64/ia64.c (ia64_compute_frame_size): Account space for
2431 save of BR0 in extra_spill_size instead of spill_size.
2432 (ia64_expand_prologue): Save BR0 outside of the gr/br/fr spill
2434 (ia64_expand_epilogue): Restore BR0 from its new location.
2436 2007-01-01 Andrew Pinski <pinskia@gmail.com>
2438 * gimplify.c (gimplify_init_constructor <case VECTOR_TYPE>):
2439 Use a temporary variable if the left hand side is not a gimple
2442 2007-01-01 Andrew Pinski <pinskia@gmail.com>
2444 * gimplify.c (gimplify_return_expr): Make the temporary variable
2445 for the return expression, a gimple register variable.
2447 2007-01-01 Jan Hubicka <jh@suse.cz>
2449 * emit-rtl.c (emit_copy_of_insn_after): Do not call copy_insn_1 for
2452 2007-01-01 Mike Stump <mrs@apple.com>
2454 * configure.ac (HAVE_GAS_LITERAL16): Add autoconf check for
2456 * config/darwin.c (machopic_select_rtx_section): Use
2458 (darwin_mergeable_constant_section): Likewise.
2459 * configure: Regenerate.
2460 * config.in: Regenerate.
2462 2007-01-01 Jan Hubicka <jh@suse.cz>
2463 Andrew Pinski <pinskia@gmail.com>
2465 * cgraphunit.c (cgraph_optimize): Call cgraph_add_new_functions
2466 before starting IPA passes.