1 2007-01-19 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
4 * toplev.c (warn_deprecated_use): Use %qD instead of %qs to print
5 the name of the declared identifier.
7 2007-01-19 Dirk Mueller <dmueller@suse.de>
9 * config/i386.h (CONDITIONAL_REGISTER_USAGE): Store
10 result of PIC_OFFSET_TABLE_REGNUM in temporary variable to avoid
13 2007-01-19 Uros Bizjak <ubizjak@gmail.com>
15 * config/i386/i386.md (acos<mode>2): Rename from acossf2 and acosdf2.
16 Macroize expander using X87MODEF12 mode macro. Extend operand 1
17 to XFMode, use acosxf2 and truncate result to requested mode.
18 Use SSE_FLOAT_MODE_P to disable patterns for SSE math.
19 (asin<mode>2): Similarly, with asin expanders.
20 (*fscalexf4_i387): Rename from *fscalexf4.
21 (expNcorexf3): New expander.
22 (expxf2, exp10xf2, exp2xf2): Use expNcorexf3 expander.
23 (exp<mode>2): Rename from expsf2 and expdf2. Macroize expander using
24 X87MODEF12 mode macro. Extend operand 1 to XFMode, use expxf2 and
25 truncate result to requested mode. Use SSE_FLOAT_MODE_P to disable
26 patterns for SSE math.
27 (exp10<mode>2): Similarly, with exp10 expanders.
28 (exp2<mode>2): Similarly, with exp2 expanders.
29 (expm1<mode>2): Similarly, with expm1 expanders.
30 (ldexp<mode>3): Similarly, with ldexp expanders.
31 (log<mode>2, log10<mode>2, log2<mode>2, log1p<mode>2, logb<mode>2):
32 Use gen_truncxf<mode>2_i387_noop to truncate result.
34 2007-01-19 Richard Sandiford <richard@codesourcery.com>
36 * config/m68k/m68k.h (PREFERRED_STACK_BOUNDARY): Define to 32
39 2007-01-19 Nathan Sidwell <nathan@codesourcery.com>
40 Richard Sandiford <richard@codesourcery.com>
42 * config/m68k/m68k.h (M68K_STATIC_CHAIN_REG_NAME): New macro.
43 (INITIALIZE_TRAMPOLINE): Use STATIC_CHAIN_REGNUM.
44 (__transfer_from_trampoline): Use M68K_STATIC_CHAIN_REG_NAME.
45 * config/m68k/m68kelf.h (STATIC_CHAIN_REGNUM): Override.
46 (M68K_STATIC_CHAIN_REG_NAME): Likewise.
47 * config/m68k/netbsd-elf.h (M68K_STATIC_CHAIN_REG_NAME): Likewise.
49 2007-01-19 Richard Sandiford <richard@codesourcery.com>
51 * config/m68k/m68k.md (adddi_dilshr32): Rename to...
52 (*adddi_dilshr32): ...this. Fix formatting. Remove commented-out
53 non-canonical pattern. Restrict to !TARGET_COLDFIRE.
54 (*adddi_dilshr32_cf): New pattern.
55 (adddi3, subdi3): Remove first alternatives.
57 2007-01-19 Richard Sandiford <richard@codesourcery.com>
59 * config/m68k/m68k.c (notice_update_cc): If an SFmode move is
60 implemented using move.l, do not use its cc result for floating-point
63 2007-01-19 Richard Sandiford <richard@codesourcery.com>
65 * config/m68k/m68k.h (EXTRA_CONSTRAINT): Stop the 'T' constraint
66 from accepting 's' constraints if flag_pic.
68 2007-01-19 Richard Sandiford <richard@codesourcery.com>
70 * config/m68k/m68k.md (bordered, bunordered, buneq, bunge, bungt)
71 (bunle, bunlt, bltgt, bordered_rev, bunordered_rev, buneq_rev)
72 (bunge_rev, bungt_rev, bunle_rev, bunlt_rev, bltgt_rev): Change
73 condition from TARGET_68881 to TARGET_HARD_FLOAT.
75 2007-01-19 Sandra Loosemore <sandra@codesourcery.com>
77 * longlong.h (count_leading_zeros, COUNT_LEADING_ZEROS_0): Add
78 ColdFire alternatives.
79 * config/m68k/m68k.h (CLZ_DEFINED_VALUE_AT_ZERO): New macro.
80 * config/m68k/m68k.md (clzsi2): Define for ColdFire
81 architectures that support ff1 instruction.
83 2007-01-19 Richard Sandiford <richard@codesourcery.com>
84 Julian Brown <julian@codesourcery.com>
86 * config/m68k/m68k.h (CONST_OK_FOR_LETTER_P): Add an 'R' case.
87 * config/m68k/m68k.md (*movsi_cfv4): Fold into...
88 (*movsi_cf): ...here. Remove unnecessary 'R' from 'Rg'.
91 2007-01-19 Richard Sandiford <richard@codesourcery.com>
93 * config/m68k/m68k-protos.h (valid_mov3q_const): Take a HOST_WIDE_INT
95 (output_move_const_into_data_reg, output_move_simode_const): Delete.
96 * config/m68k/m68k.c (const_method, const_int_cost): Take a
97 HOST_WIDE_INT instead of an rtx.
98 (m68k_rtx_costs): Update call accordingly.
99 (output_move_const_into_data_reg): Likewise. Fix formatting.
100 (valid_mov3q_const): Take a HOST_WIDE_INT instead of an rtx.
102 (output_move_simode_const): Update calls after above changes.
103 Rework to use automatic variables and predicates like MEM_P.
104 * config/m68k/m68k.md (pushexthisi_const): Update call to
107 2007-01-19 Dirk Mueller <dmueller@suse.de>
109 * tree-ssa-alias.c (perform_var_substitution): Fix typo
112 2007-01-19 Richard Guenther <rguenther@suse.de>
114 * builtins.c (expand_builtin_cexpi): Fall back to expanding
115 via cexp in case sincos is not available.
117 2007-01-19 Richard Guenther <rguenther@suse.de>
119 * doc/tm.texi (TARGET_HAS_SINCOS): Document new target macro.
120 * defaults.h (TARGET_HAS_SINCOS): Default to off.
121 * config/linux.h (TARGET_HAS_SINCOS): Set to on if we have glibc.
122 * config/alpha/linux.h (TARGET_HAS_SINCOS): Likewise.
123 * config/sparc/linux.h (TARGET_HAS_SINCOS): Likewise.
124 * config/sparc/linux64.h (TARGET_HAS_SINCOS): Likewise.
125 * config/rs6000/linux.h (TARGET_HAS_SINCOS): Likewise.
126 * config/rs6000/linux64.h (TARGET_HAS_SINCOS): Likewise.
128 2007-01-19 Uros Bizjak <ubizjak@gmail.com>
130 * config/i386/i386.md (*fpatanxf3_i387, fpatan_extend<mode>xf3_i387):
132 (atan2sf3_1, atan2df3_1, atan2xf3_1): Remove insn patterns.
133 (atan2xf3): Directly generate RTL pattern.
134 (atan2<mode>3): Rename from atan2sf3 and atan2df3 and macroize insn
135 patterns using X87MODEF12 mode macro. Use fpatan_extend<mode>xf3_i387
136 and truncate result to requested mode. Use SSE_FLOAT_MODE_P to
137 disable patterns for SSE math.
138 (atan<mode>2): Rename from atansf2 and atandf2 and macroize insn
139 patterns using X87MODEF12 mode macro. Use fpatan_extend<mode>xf3_i387
140 and truncate result to requested mode. Use SSE_FLOAT_MODE_P to
141 disable patterns for SSE math.
143 2007-01-19 Alexandre Oliva <aoliva@redhat.com>
145 * libgcc-std.ver: Fix typo in %inherit for GCC_4.3.0.
147 2007-01-18 Roger Sayle <roger@eyesopen.com>
149 * fold-const.c (fold_unary) <VIEW_CONVERT_EXPR>: Optimize away a
150 VIEW_CONVERT_EXPR to the same type as it's operand.
152 2007-01-18 David Edelsohn <edelsohn@gnu.org>
154 * config/rs6000/darwin-ldouble.c: Only build _SOFT_FLOAT if
155 configured for long double 128.
157 2007-01-18 Mike Stump <mrs@apple.com>
159 * config/rs6000/rs6000.c (rs6000_emit_vector_compare): Fix build
162 2007-01-18 Michael Meissner <michael.meissner@amd.com>
164 * i386.c (ix86_compute_frame_layout): Make fprintf's in #if 0 code
167 2007-01-18 Jan Hubicka <jh@suse.cz>
169 * tree-ssa-operands.c (vop_free_bucket_size): Never return value
170 greater than NUM_VOP_FREE_BUCKETS.
172 2007-01-18 Daniel Berlin <dberlin@dberlin.org>
174 * tree-ssa-structalias.c: Update comments.
175 (ptabitmap_obstack): Removed.
177 (oldpta_obstack): Ditto.
178 (stats): Add a few members.
179 (struct variable_info): Remove node, complex, address_taken, and
180 indirect_target members. Add oldsolution member.
181 (new_var_info): Do not initialize removed members.
182 (constraint_expr_type): Remove INCLUDES.
183 (constraint_graph): Add size, implicit_preds, rep,
184 indirect_cycles, eq_rep, label, direct_nodes, and complex members.
185 (FIRST_REF_NODE): New macro.
186 (LAST_REF_NODE): Ditto.
187 (FIRST_ADDR_NODE): Ditto.
188 (find): New function.
190 (dump_constraint): Do not handle INCLUDES.
191 (insert_into_complex): Do not insert duplicate constraints.
192 (condense_varmap_nodes): Renamed and rewritten into ...
193 (merge_node_constraints): This. Also fix bug in handling of
194 offseted copy constraints.
195 (clear_edges_for_node): No longer need to deal with preds at all,
196 or removing associated preds/succs.
197 (merge_graph_nodes): Deal with indirect_cycles.
198 Don't deal with predecessors.
199 (add_implicit_graph_edge): New function.
200 (add_pred_graph_edge): Ditto.
201 (add_graph_edge): Don't deal with predecessors.
202 (build_constraint_graph): Removed.
203 (build_pred_graph): New function.
204 (build_succ_graph): Ditto.
205 (struct scc_info): Removed in_component. Added roots, dfs, and
206 node_mapping. Remove visited_index, unification_queue.
207 (scc_visit): Deal with union-find we do now.
208 Deal with cycles with REF nodes.
209 (collapse_nodes): Renamed and rewritten to ...
211 (process_unification_queue): Removed.
212 (topo_visit): Cleanup
213 (do_da_constraint): Use find.
214 (do_sd_constraint): Ditto.
215 (do_ds_constraint): Ditto.
216 (do_complex_constraint): Ditto.
217 (init_scc_info): Update for removed and added members.
218 (find_and_collapse_graph_cycles): Renamed and rewritten into ...
219 (find_indirect_cycles): This.
220 (equivalence_class): New variable.
221 (label_visit): New function.
222 (perform_variable_substitution): Rewritten.
223 (free_var_substitution_info): New function.
224 (find_equivalent_node): Ditto.
225 (move_complex_constraints): Ditto.
226 (eliminate_indirect_cycles): Ditto.
227 (solve_graph): Only propagate changed bits.
228 Use indirect cycle elimination.
230 (tree_id_t): Rename to tree_vi_t, delete id member, add vi member.
231 (tree_id_eq): Renamed to ...
232 (tree_vi_eq): This. Update for member change
233 (insert_id_for_tree): Renamed and rewritten to ...
234 (insert_vi_for_tree): This.
235 (lookup_id_for_tree): Renamed and rewritten to ...
236 (lookup_vi_for_tree): This.
237 (get_id_for_tree): Renamed and rewritten to ...
238 (get_vi_for_tree): Ditto.
239 (get_constraint_exp_from_ssa_var): Update to use get_vi_for_tree.
240 (process_constraint): Don't handle INCLUDES.
241 Remove special ADDRESSOF case.
242 (find_func_aliases): Rewrite to use vi functions instead of id
244 (create_function_info_for): Ditto.
245 (create_variable_info_for): Ditto.
246 (intra_create_variable_infos): Ditto.
247 (merge_smts_into): Ditto.
248 (find_what_p_points_to): Ditto.
249 (init_base_vars): Ditto.
250 (init_alias_vars): Ditto.
251 (remove_preds_and_fake_succs): New function.
252 (dump_sa_points_to_info): Dump new stats.
253 (dump_solution_for_var): Use find.
254 (set_used_smts): Fix formatting.
255 (compute_points_to_sets): Updated for new functions.
256 (ipa_pta_execute): Ditto.
258 2007-01-18 Kazu Hirata <kazu@codesourcery.com>
259 Richard Sandiford <richard@codesourcery.com>
261 * doc/tm.texi (TARGET_FUNCTION_VALUE): Expand documentation of
263 * calls.c (expand_call): If the return value is a PARALLEL,
264 extract its first member.
265 * config/m68k/linux.h (FUNCTION_EXTRA_EPILOGUE): Remove.
266 * config/m68k/m68k.c (m68k_output_function_epilogue): Don't
267 use FUNCTION_EXTRA_EPILOGUE.
268 (m68k_function_value): Return a PARALLEL if the return value
269 is of a pointer type.
270 * config/m68k/netbsd-elf.h (current_function_returns_pointer)
271 (FUNCTION_EXTRA_EPILOGUE): Remove.
272 * config/m68k/m68k.md (D0_REG): New constant.
274 2007-01-18 Kazu Hirata <kazu@codesourcery.com>
276 * config/m68k/m68k.c (m68k_output_function_epilogue): Don't
277 output a NOP for empty epilogues.
279 2007-01-18 Richard Sandiford <richard@codesourcery.com>
281 * config/m68k/m68k.c (m68k_use_return_insn): Update comments
282 before function. Extend register save check to include all
283 registers, not just integer ones.
285 2007-01-18 Kazu Hirata <kazu@codesourcery.com>
287 * config/m68k/m68k-protos.h (use_return_insn): Rename to...
288 (m68k_use_return_insn): ...this.
289 * config/m68k/m68k.h (USE_RETURN_INSN): Delete.
290 * config/m68k/m68k.c (use_return_insn): Rename to...
291 (m68k_use_return_insn): ...this.
292 * config/m68k/m68k.md (return): Use m68k_use_return_insn instead
295 2007-01-18 Nathan Sidwell <nathan@codesourcery.com>
297 * config/m68k/fpgnulib.c (__truncdfsf2): Implement round to
298 nearest even, fix denormal rounding overflow.
300 2007-01-18 Richard Sandiford <richard@codesourcery.com>
302 * config/m68k/m68k.md (movsf_cf_hard): Use fsmove instead of
303 f%$smove and f%$move.
304 (movdf_cf_hard): Use fdmove for cases 0 and 3 and fmove for case 1.
305 (extendsfdf2_cf): Use fdmove instead of f%&move.
306 (truncdfsf2_cf): Use fsmove instead of f%$smove.
307 (add<mode>3_cf, sub<mode>3_cf): Use <FP:prec> instead of <FP:round>.
309 2007-01-18 Richard Sandiford <richard@codesourcery.com>
311 * config/m68k/m68k.md (movdf_cf_hard): Use output_move_double for
314 2007-01-18 Richard Sandiford <richard@codesourcery.com>
316 * real.h (real_format): Add a canonical_nan_lsbs_set field.
317 (coldfire_single_format): Declare.
318 (coldfire_double_format): Likewise.
319 * real.c (encode_ieee_single): Use canonical_nan_lsbs_set instead
320 of qnan_msb_set to determine the lower bits of a canonical
322 (encode_ieee_double): Likewise.
323 (encode_ieee_quad): Likewise.
324 (ieee_single_format): Initialize canonical_nan_lsbs_set.
325 (mips_single_format): Likewise.
326 (ieee_double_format): Likewise.
327 (mips_double_format): Likewise.
328 (ieee_extended_motorola_format): Likewise.
329 (ieee_extended_intel_96_format): Likewise.
330 (ieee_extended_intel_128_format): Likewise.
331 (ieee_extended_intel_96_round_53_format): Likewise.
332 (ibm_extended_format): Likewise.
333 (mips_extended_format): Likewise.
334 (ieee_quad_format): Likewise.
335 (mips_quad_format): Likewise.
336 (vax_f_format): Likewise.
337 (vax_d_format): Likewise.
338 (vax_g_format): Likewise.
339 (i370_single_format): Likewise.
340 (i370_double_format): Likewise.
341 (decimal_single_format): Likewise.
342 (decimal_double_format): Likewise.
343 (decimal_quad_format): Likewise.
344 (c4x_single_format): Likewise.
345 (c4x_extended_format): Likewise.
346 (real_internal_format): Likewise.
347 (coldfire_single_format): New real_format.
348 (coldfire_double_format): Likewise.
349 * config/pdp11/pdp11.c (pdp11_f_format): Initialize
350 canonical_nan_lsbs_set.
351 (pdp11_d_format): Likewise.
352 * config/m68k/m68k.c (override_options): Override REAL_FORMAT_MODE
353 if TARGET_COLDFIRE_CPU.
355 2007-01-18 Richard Sandiford <richard@codesourcery.com>
357 * config/m68k/m68k-protos.h (m68k_output_pic_call): Delete.
358 (output_call, m68k_legitimize_call_address): Declare.
359 * config/m68k/m68k.h (EXTRA_CONSTRAINT): Remove unnecessary
360 parenthesees. Add support for a 'W' constraint.
361 (LEGITIMATE_PIC_OPERAND_P): Remove SYMBOL_REF_FLAG handling.
362 (PRINT_OPERAND_PUNCT_VALID_P): Remove comment about 'o'.
363 (m68k_symbolic_call, m68k_symbolic_jump): Declare.
364 * config/m68k/m68k.c (m68k_symbolic_call, m68k_symbolic_jump): New
366 (override_options): Initialize them. Do not set flag_no_function_cse
367 for TARGET_ID_SHARED_LIBRARY.
368 (m68k_output_pic_call): Delete.
369 (m68k_legitimize_call_address): New function.
370 (print_operand): Remove the %o prefix. Handle the %p prefix.
371 (output_call): New function.
372 (m68k_output_mi_thunk): Use m68k_symbolic_jump. Always load the
373 target address from the GOT if symbolic jumps are not allowed.
374 * config/m68k/m68k.md (call, general_operand): Do not set
375 SYMBOL_REF_FLAG. Use m68k_legitimize_call_address instead.
376 Merge separate flag_pic and !flag_pic define_insns into...
377 (*call, *call_value): ...these new patterns. Match the address
378 rather than the containing MEM and require it to be a call_operand.
379 Use output_call to generate the asm template.
380 * config/m68k/predicates.md (const_call_operand): New predicate.
381 (call_operand): Likewise.
383 2007-01-18 Nathan Sidwell <nathan@codesourcery.com>
385 * config/m68k/m68k.h (REGISTER_MOVE_COST): Simplify definition.
386 (STACK_GROWS_DOWNWARD): Define to 1.
387 (FUNCTION_VALUE, LIBCALL_VALUE, FUNCTION_VALUE_REGNO_P): Equivocate
388 comments, emphasizing that these values are only defaults.
389 * config/m68k/linux.h (LINK_SPEC): Fix formatting in #undef.
390 * config/m68k/m68k.c (const_method): Remove trailing whitespace.
392 2007-01-18 Richard Sandiford <richard@codesourcery.com>
394 * config/m68k/m68k.md (cmpsi): Remove outdated flag_pic handling.
396 2007-01-18 Kazu Hirata <kazu@codesourcery.com>
397 Richard Sandiford <richard@codesourcery.com>
399 * config/m68k/m68k.h (DATA_REGNO_P, ADDRESS_REGNO_P, INT_REGNO_P)
400 (FP_REGNO_P): New macros.
401 (REGNO_OK_FOR_INDEX_P, REGNO_OK_FOR_BASE_P, REGNO_OK_FOR_DATA_P)
402 (REGNO_OK_FOR_FP_P, REG_OK_FOR_INDEX_P, REG_OK_FOR_BASE_P): Use them.
403 (EH_RETURN_STACKADJ_RTX): Use A0_REG.
404 * config/m68k/m68k.c (m68k_regno_mode_ok): Use the new REGNO macros.
406 2007-01-18 Nathan Sidwell <nathan@codesourcery.com>
408 * config.gcc (m68k-*-aout*, m68k-*-coff*, m68020-*-elf*, m68k-*-elf*)
409 (m68k-*-uclinuxoldabi, m68k-*-uclinux*, m68k-*-rtems*): Add t-floatlib
411 * config/m68k/t-floatlib: New file, extracting common code from...
412 * config/m68k/t-m68kbare, config/m68k/t-m68kelf,
413 * config/m68k/t-uclinux: Here.
414 * config/m68k/fpgnulib.c: Do not compile extendeed precision
415 routines on ColdFire targets.
417 2007-01-18 Nathan Sidwell <nathan@codesourcery.com>
419 * config.gcc (m68k-*-aout*, m68k-*-coff*, m68020-*-elf*, m68k-*-elf*)
420 (m68010-*-netbsdelf*, m68k*-*-netbsdelf*, m68k*-*-openbsd*)
421 (m68k-*-uclinuxoldabi, m68k-*-uclinux*, m68k-*-linux*)
422 (m68k-*-rtems*): Use tm_file rather than m68k/m68k.h and
423 explicitly set MOTOROLA to 1.
424 * config/m68k/m68k.h (MOTOROLA): Simplify definition accordingly.
426 2007-01-18 Richard Sandiford <richard@codesourcery.com>
427 Nathan Sidwell <nathan@codesourcery.com>
429 * config/m68k/m68k.h (PCC_STATIC_STRUCT_RETURN): Delete.
430 (ASM_OUTPUT_REG_PUSH, ASM_OUTPUT_REG_POP): Add MOTOROLA cases.
431 * config/m68k/coff.h (REGISTER_PREFIX_MD): Delete.
432 * config/m68k/m68020-elf.h (LIB_SPEC): Delete.
433 * config/m68k/m68k-none.h (CC1_SPEC, CPP_SUBTARGET_SPEC): Delete.
434 * config/m68k/m68kelf.h (IMMEDIATE_PREFIX, REGISTER_PREFIX_MD)
435 (ASM_OUTPUT_REG_PUSH, ASM_OUTPUT_REG_POP): Delete.
436 (NO_DOLLAR_IN_LABEL, PCC_STATIC_STRUCT_RETURN): Don't undefine.
437 (BSS_ASM_OP, ASM_OUTPUT_SKIP, ASM_OUTPUT_ASCII): Delete.
438 * config/m68k/m68kemb.h (PCC_STATIC_STRUCT_RETURN): Don't undefine.
439 * config/m68k/linux.h (SIZE_TYPE, PTRDIFF_TYPE, WCHAR_TYPE)
440 (WCHAR_TYPE_SIZE, TARGET_OBJFMT_CPP_BUILTINS): Delete.
441 (TARGET_OS_CPP_BUILTINS): Don't define mc68000 and mc68020 here.
442 (DBX_REGISTER_NUMBER): Delete.
443 * config/m68k/netbsd-elf.h (IMMEDIATE_PREFIX): Delete.
444 (PCC_STATIC_STRUCT_RETURN): Don't undefine.
445 * config/m68k/openbsd.h (PCC_STATIC_STRUCT_RETURN): Define.
447 2007-01-18 Richard Sandiford <richard@codesourcery.com>
449 * config.gcc (m68k-*-uclinux*): Add flat.h to $tm_file.
450 * config/flat.h: New file.
451 * crtstuff.c (USE_PT_GNU_EH_FRAME): Don't define if
453 * config/m68k/m68k.h (ASM_PREFERRED_EH_DATA_FORMAT): Do not use
454 indirect references for -msep-data or -mid-shared-library.
455 Do not use PC-relative code addresses either.
456 * config/m68k/m68k.c (override_options): Restrict -fPIC error
458 * config/m68k/uclinux.h (STARTFILE_SPEC): Define. Use Scrt1.o
459 for shared libraries and crt1.o for executables. Use crti.o and
461 (ENDFILE_SPEC): Use crtend.o and crtn.o.
462 (LIB_SPEC): Suppress -Rlibc.gdb if -static-libc is given.
463 Do not add -elf2flt or -shared-lib-id options here.
464 (LINK_SPEC): Define. Pass -elf2flt if no -elf2flt option is given.
465 Pass -shared-lib-id if -mid-shared-library, taking the library
466 identifier from -mshared-library-id if given, otherwise
468 (EH_FRAME_IN_DATA_SECTION): Do not undefine.
469 (INIT_SECTION_ASM_OP, FINI_SECTION_ASM_OP): Likewise.
470 (TARGET_OS_CPP_BUILTINS): Define __GXX_MERGED_TYPEINFO_NAMES=0
471 and __GXX_TYPEINFO_EQUALITY_INLINE=0 if -mid-shared-library.
472 (DRIVER_SELF_SPECS): Map unadorned PIC options to -msep-data.
473 * config/m68k/t-uclinux (EXTRA_MULTILIB_PARTS): Add crtbegin.o
475 * config/m68k/lb1sf68.asm (PICCALL): Use an lea and pc-relative
476 jump sequence for ISA A and ISA A+.
479 2007-01-18 Richard Sandiford <richard@codesourcery.com>
481 * config.gcc (m68k-*-uclinux*): Base the port on the common
482 and m68k GNU/Linux files rather than on the generic ELF ones.
483 * config/m68k/uclinux.h (TARGET_VERSION): Override.
484 (TARGET_OS_CPP_BUILTINS): Use LINUX_TARGET_OS_CPP_BUILTINS.
486 2007-01-18 Julian Brown <julian@codesourcery.com>
487 Richard Sandiford <richard@codesourcery.com>
489 * config/m68k/m68k.h (LONG_DOUBLE_TYPE_SIZE): Make 64-bit on ColdFire.
490 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
491 * config/m68k/netbsd-elf.h (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Undefine
493 * config/m68k/uclinux-oldabi.h (LONG_DOUBLE_TYPE_SIZE): Redefine to
495 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
497 2007-01-18 Richard Sandiford <richard@codesourcery.com>
499 * doc/install.texi: Document m68k-uclinuxoldabi.
500 * config.gcc (m68k-*-uclinuxoldabi): New configuration.
501 * config/m68k/uclinux-oldabi.h: New file, copied from
502 config/m68k/uclinux.h.
504 2007-01-18 Kaz Kojima <kkojima@rr.iij4u.or.jp>
506 * config/m32r/m32r-protos.h (m32r_expand_epilogue): Declare it.
507 * config/m32r/m32r.c (m32r_setup_incoming_varargs): Use gen_frame_mem.
508 (m32r_compute_frame_size): Use unsigned for regno.
509 (m32r_reload_lr): Use gen_frame_mem.
511 (m32r_output_function_epilogue): Don't output the function epilogue
513 (m32r_expand_epilogue): New.
514 (direct_return): Return false if the function has the interrupt
516 (m32r_hard_regno_rename_ok): Remove code for the textual epilogue.
517 * config/m32r/m32r.md (epilogue): New expander.
518 (return_lr, return_rte): New insns.
519 (return): Make it expander.
520 (return_normal): New expander.
522 2007-01-18 Josh Conner <jconner@apple.com>
525 * config/rs6000/rs6000.c (rs6000_emit_vector_compare): Add
526 support for UNLE, UNLT, UNGE, and UNGT.
528 2007-01-18 Jan Hubicka <jh@suse.cz>
530 * tree-vrp.c (finalize_jump_threads): Do not call cleanup_cfg by hand.
531 * tree-tailcall (add_virtual_phis): Likewise.
532 (optimize_tail_call): Return TODOs.
533 (execute_tail_calls): Return TODOs.
534 * tree-ssa-ccp (execute_fold_all_builtins): Do cleanup_cfg via TODO.
535 * tree-cfgcleanup.c (cleanup_tree_cfg_loop): Return if something
537 * tree-ssa-forwprop.c (tree_ssa_forward_propagate_single_use_value):
538 Cleanup cfg using TODO.
539 * tree-flow.h (cleanup_tree_cfg_loop): Update prototype.
540 * passes.c (execute_function_todo): When cleanup did something, remove
542 * tree-cfg.c (pass_build_cfg): Add cleanup_cfg TODO.
543 (make_edges): Don't cleanup_cfg.
545 2007-01-18 Uros Bizjak <ubizjak@gmail.com>
547 * reg-stack.c (subst_stack_regs_pat) [UNSPEC_SINCOS_COS,
548 UNSPEC_XTRACT_FRACT]: Use generic code for instructions that
549 operate on the top of stack.
550 [UNSPEC_SINCOS_SIN, UNSPEC_XTRACT_EXP, UNSPEC_TAN]: Rewrite
551 register handling of instructions that output to the second
553 [UNSPEC_TAN_ONE, UNSPEC_TAN_TAN]: Remove.
554 (move_for_stack_reg): Special-case check for dead destination
555 stack slot for constant load of 1.0 inside UNSPEC_TAN.
557 * config/i386/i386.md (UNSPEC_TAN): New constant.
558 (UNSPEC_TAN_ONE, UNSPEC_TAN_TAN): Remove.
559 (fptanxf4_i387, fptan_extend<mode>xf4_i387): New patterns
560 to correctly model move of constant 1.0 to top stack slot.
561 (*tandf3_1, *tansf3_1, *tanxf3_1): Remove insn patterns.
562 (unnamed peephole2 pattern): Remove corresponding peephole2
563 pattern that optimizes tan insn and loading of constant 1.0.
564 (tanxf2): Use fptanxf4_i387.
565 (tan<mode>2): Rename from tansf2 and tandf2 and macroize insn
566 patterns using X87MODEF12 mode macro. Use fptan_extend<mode>xf4_i387
567 and truncate result to requested mode. Use SSE_FLOAT_MODE_P to
568 disable patterns for SSE math.
569 (sincos<mode>3): Use truncxf<mode>2_i387_noop for truncation.
570 (fyl2x_extend<mode>xf3_i387): Use X87MODEF12 for operand 1.
572 2007-01-18 Dirk Mueller <dmueller@suse.de>
573 Richard Guenther <rguenther@suse.de>
576 * doc/invoke.texi (Warray-bounds): Document -Warray-bounds.
577 * common.opt (Warray-bounds): Add new warning option.
578 * c-opts.c (c_common_handle_option): Define -Warray-bounds
580 * Makefile.in: make tree-vrp.o depend on toplev.h
581 * tree-vrp.c (vrp_finalize): Call check_array_refs if -Warray-bounds
583 (check_array_refs, check_array_bounds, check_array_ref): New.
585 2007-01-18 Jan Hubicka <jh@suse.cz>
587 * tree-ssa-ccp.c (ccp_finalize): Return if something changed.
588 (execute_ssa_ccp): Return flags conditionally.
589 * tree-ssa-propagate.c (substitue_and_fold): Return if something was
591 * tree-ssa-propagate.h (substitute_and_fold): Update prototype.
593 2007-01-18 Steven Bosscher <steven@gcc.gnu.org>
595 * cfgcleanup.c (cleanup_cfg): Detect cfglayout mode and set
596 the CLEANUP_CFGLAYOUT flag when in cfglayout mode.
598 * Makefile.c (GTFILES): Add cfglayout.h.
599 * gengtype.c (open_base_files): Likewise.
600 * cfglayout.c (cfg_layout_function_footer,
601 cfg_layout_function_header) Reindent to make gengtype happy.
602 * cfglayout.h (cfg_layout_function_footer,
603 cfg_layout_function_header): Add GTY(()) marker.
605 * ifcvt.c (noce_try_sign_mask): Make sure INSN_B is non-null.
607 2007-01-18 Ben Elliston <bje@au.ibm.com>
609 * genautomata.c (write_automata): Include xstrerror output in the
610 error message if writing the DFA description file fails.
612 2007-01-17 H.J. Lu <hongjiu.lu@intel.com>
614 * config/mips/mips-protos.h (mips_output_external): Make it
616 * config/mips/iris.h (TARGET_ASM_EXTERNAL_LIBCALL): Removed.
617 * config/mips/mips.c (irix_output_external_libcall): Likewise.
618 (extern_list): Likewise.
619 (extern_head): Likewise.
620 (TARGET_ASM_FILE_END): Likewise.
621 (mips_file_end): Likewise.
622 (mips_output_external): Rewritten.
624 2007-01-18 Ben Elliston <bje@au.ibm.com>
626 * genpreds.c (write_insn_preds_c): Only write out the function
627 body for regclass_for_constraint if we have register constraints.
629 2007-01-17 Tom Tromey <tromey@redhat.com>
631 * doc/sourcebuild.texi (libgcj Tests): Use sourceware.org.
632 * doc/install.texi (Testing): Use sourceware.org.
633 (Binaries): Likewise.
634 (Specific): Likewise.
635 * doc/contrib.texi (Contributors): Use sourceware.org.
637 2007-01-17 Anatoly Sokolov <aesok@post.ru>
639 * config/avr/avr.h (AVR_HAVE_LPMX): New macro.
640 (AVR_ENHANCED): Rename to ...
641 (AVR_HAVE_MUL): ... new.
642 (avr_enhanced_p): Rename to ...
643 (avr_have_mul_p): ... new.
644 (TARGET_CPU_CPP_BUILTINS): Use 'avr_have_mul_p' instead of
645 'avr_enhanced_p' for "__AVR_ENHANCED__". Define "__AVR_HAVE_MUL__".
646 * config/avr/avr.c (avr_enhanced_p): Rename to ...
647 (avr_have_mul_p): ... new.
648 (base_arch_s): Rename 'enhanced' to 'have_mul'.
649 (avr_override_options): Use 'avr_have_mul_p' and 'have_mul' instead of
650 'avr_enhanced_p' and 'enhanced'.
651 (ashlhi3_out, ashrhi3_out, lshrhi3_out, avr_rtx_costs): Use
652 AVR_HAVE_MUL instead of AVR_ENHANCED.
653 * avr.md (*tablejump_enh): Use AVR_HAVE_LPMX instead of AVR_ENHANCED.
654 (mulqi3, *mulqi3_enh, *mulqi3_call, mulqihi3, umulqihi3, mulhi3,
655 *mulhi3_enh, *mulhi3_call, mulsi3, *mulsi3_call): Use AVR_HAVE_MUL
656 instead of AVR_ENHANCED.
657 (*tablejump_enh): Use AVR_HAVE_LPMX instead of AVR_ENHANCED.
658 * libgcc.S: Use __AVR_HAVE_MUL__ instead of __AVR_ENHANCED__.
659 (__tablejump__): Use __AVR_HAVE_LPMX__ instead of __AVR_ENHANCED__.
661 2007-01-17 Ian Lance Taylor <iant@google.com>
663 * vec.h (VEC_reserve_exact): Define.
664 (vec_gc_p_reserve_exact): Declare.
665 (vec_gc_o_reserve_exact): Declare.
666 (vec_heap_p_reserve_exact): Declare.
667 (vec_heap_o_reserve_exact): Declare.
668 (VEC_OP (T,A,reserve_exact)): New static inline function, three
670 (VEC_OP (T,A,reserve)) [all versions]: Remove handling of
672 (VEC_OP (T,A,alloc)) [all versions]: Call ...reserve_exact.
673 (VEC_OP (T,A,copy)) [all versions]: Likewise.
674 (VEC_OP (T,a,safe_grow)) [all versions]: Likewise.
675 * vec.c (calculate_allocation): Add exact parameter. Change all
677 (vec_gc_o_reserve_1): New static function, from vec_gc_o_reserve.
678 (vec_gc_p_reserve, vec_gc_o_reserve): Call vec_gc_o_reserve_1.
679 (vec_gc_p_reserve_exact, vec_gc_o_reserve_exact): New functions.
680 (vec_heap_o_reserve_1): New static function, from vec_heap_o_reserve.
681 (vec_heap_p_reserve, vec_heap_o_reserve): Call vec_heap_o_reserve_1.
682 (vec_heap_p_reserve_exact): New function.
683 (vec_heap_o_reserve_exact): New function.
685 2007-01-17 Jan Hubicka <jh@suse.cz>
687 * ipa-type-escape.c (look_for_casts): Revamp using handled_component_p.
689 2007-01-17 Eric Christopher <echristo@apple.com>
691 * config.gcc: Support core2 processor.
693 2007-01-16 Jan Hubicka <jh@suse.cz>
695 * tree-ssanames.c (release_dead_ssa_names): Instead of ggc_freeing
696 the names, just unlink the chain so we don't crash on dangling pointers
699 2007-01-16 Jan Hubicka <jh@suse.cz>
701 * cgraph.h (cgraph_decide_inlining_incrementally): Kill.
702 * tree-pass.h: Reorder to make IPA passes appear toegher.
703 (pass_early_inline, pass_inline_parameters, pass_apply_inline): Declare.
704 * cgraphunit.c (cgraph_finalize_function): Do not compute inling
705 parameters, do not call early inliner.
706 * ipa-inline.c: Update comments. Include tree-flow.h
707 (cgraph_decide_inlining): Do not compute inlining parameters.
708 (cgraph_decide_inlining_incrementally): Return TODOs; assume to
709 be called with function context set up.
710 (pass_ipa_inline): Remove unreachable functions before pass.
711 (cgraph_early_inlining): Simplify assuming to be called from the
713 (pass_early_inline): New pass.
714 (cgraph_gate_ipa_early_inlining): New gate.
715 (pass_ipa_early_inline): Turn into simple wrapper.
716 (compute_inline_parameters): New function.
717 (gate_inline_passes): New gate.
718 (pass_inline_parameters): New pass.
719 (apply_inline): Move here from tree-optimize.c
720 (pass_apply_inline): New pass.
721 * ipa.c (cgraph_remove_unreachable_nodes): Verify cgraph after
723 * tree-inline.c (optimize_inline_calls): Return TODOs rather than
725 (tree_function_versioning): Do not allocate dummy struct function.
726 * tree-inline.h (optimize_inline_calls): Update prototype.
727 * tree-optimize.c (execute_fixup_cfg): Export.
728 (pass_fixup_cfg): Remove
729 (tree_rest_of_compilation): Do not apply inlines.
730 * tree-flow.h (execute_fixup_cfg): Declare.
731 * Makefile.in (gt-passes.c): New.
732 * passes.c: Include gt-passes.h
733 (init_optimization_passes): New passes.
734 (nnodes, order): New static vars.
735 (do_per_function_toporder): New function.
736 (execute_one_pass): Dump current pass here.
737 (execute_ipa_pass_list): Don't dump current pass here.
739 2007-01-16 Janis Johnson <janis187@us.ibm.com>
741 * config/dfp-bit.c (dfp_compare_op): Return separate value for NaN.
742 (DFP_NE, DFP_LE, DFP_GE): Return false for NaN.
744 2007-01-16 David Edelsohn <edelsohn@gnu.org>
746 * config/rs6000/darwin-ldouble.c: Build file for SOFT_FLOAT.
747 (strong_alias): Define.
748 (__gcc_qmul): Provide non-FMA for soft-float.
764 * config/rs6000/rs6000.c (rs6000_init_libfuncs): Initialize
765 soft-float functions.
766 * config/rs6000/libgcc-ppc-glibc.ver: Version soft-float symbols.
767 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Do not warn
768 about long double soft float.
770 2007-01-16 Dorit Nuzman <dorit@il.ibm.com>
771 Tehila Meyzels <tehila@il.ibm.com>
773 * tree-vectorizer.h (is_pattern_stmt_p): New.
774 * tree-vect-analyze.c (vect_determine_vectorization_factor): Fix
775 formatting (tabs instead of spaces). Cleanup and clarify setting
776 of STMT_VINFO_VECTYPE. Call is_pattern_stmt_p.
777 * tree-vect-transform.c (vect_get_vec_def_for_operand): Fix typo.
778 (vectorizable_type_demotion): Check that types are integral.
779 (vectorizable_type_promotion): Likewise.
780 (vectorizable_store): Fix typo. Eliminate new-line at end of
783 2007-01-16 Jan Hubicka <jh@suse.cz>
785 * tree-ssanames.c (release_dead_ssa_names): Remove invalidated
788 2007-01-15 Eric Christopher <echristo@apple.com>
790 * ifcvt.c: Include vec.h, vecprim.h.
791 (check_cond_move_block): New argument regs.
792 Reorganize. Add registers used to regs.
793 (cond_move_process_if_block): Use regs set above as
796 2007-01-15 Eric Christopher <echristo@apple.com>
798 * config/darwin.h: Update copyright.
799 (TARGET_OPTION_TRANSLATE_TABLE): Add umbrella.
800 (LINK_COMMAND_SPEC): Add -u.
801 (LINK_SPEC): Fix umbrella for above.
803 2007-01-15 Joseph S. Myers <joseph@codesourcery.com>
805 * config/soft-fp/op-common.h, config/soft-fp/op-4.h: Update from
808 2007-01-15 Tom Tromey <tromey@redhat.com>
810 * doc/sourcebuild.texi (libgcj Tests): Don't mention jacks.
811 * doc/install.texi (Testing): Don't mention jacks.
812 (Configuration): Document --enable-java-maintainer-mode. Move
813 --with-java-home to libgcj-specific section. Document
815 (Prerequisites): Mention --enable-java-maintainer-mode, ecj1.
817 2007-01-15 Jan Hubicka <jh@suse.cz>
819 * tree-ssa-dce.c (DCE_TODOs): New.
820 (propagate_necessity): Return if something changed.
821 (eliminate_unnecessary_stmts): Likewise.
822 (perform_tree_ssa_dce): Return TODO flags when needed.
823 (pass_dce, pass_dce_loop, pass_cd_dce): Remove TODO flags.
825 2007-01-15 Uros Bizjak <ubizjak@gmail.com>
827 * config/i386/i386.md (fyl2xxf3_i387): Rename from fyl2x_xf3.
828 (fyl2x_extend<mode>xf3_i387): New insn pattern.
829 (log<mode>2): Rename from logsf2 and logdf2 and macroize insn
830 patterns using X87MODEF12 mode macro. Extend operand 1
831 to XFmode. Use SSE_FLOAT_MODE_P to disable patterns for SSE math.
832 (log10<mode>2): Ditto.
833 (log2<mode>2): Ditto.
834 (log1p<mode>2): Ditto.
835 (logb<mode>2): Ditto.
836 (fyl2xp1xf3_i387): Rename from fyl2xp1_xf3.
837 (fyl2xp1_extend<mode>xf3_i387): New insn pattern.
838 (*fxtractxf3_i387): Rename from *fxtractxf3.
839 (fxtract_extend<mode>xf3_i387): New insn pattern.
840 (ilogbsi2): Use match_dup 3, not match_operand:XF 3.
841 * config/i386/i386.c (ix86_emit_i387_log1p): Use gen_fyl2xp1xf3_i387()
842 and gen_fyl2xxf3_i387().
844 2007-01-14 Zdenek Dvorak <dvorakz@suse.cz>
846 * loop-unswitch.c (unswitch_loop): Do not call fix_loop_placement.
847 * cfgloopmanip.c (fix_loop_placement): Made static. Use
848 get_loop_exit_edges. Changed return type to bool.
849 * cfgloop.h (fix_loop_placement): Declaration removed.
851 2007-01-14 Dorit Nuzman <dorit@il.ibm.com>
853 * param.h (MIN_VECT_LOOP_BOUND): New.
854 * params.def (MIN_VECT_LOOP_BOUND): New.
855 * tree-vectorizer.c (slpeel_tree_peel_loop_to_edge): Takes another
856 argument - minimum threshold for number of iterations.
857 * tree-vectorizer.h (slpeel_tree_peel_loop_to_edge): Add another
858 argument to declaration.
859 * tree-vect-analyze.c (vect_analyze_operations): Check value of
861 * tree-vect-transform.c (vect_do_peeling_for_loop_bound): Call
862 slpeel_tree_peel_loop_to_edge with additional argument.
863 (vect_do_peeling_for_alignment): Likewise.
864 * doc/invoke.texi (min-vect-loop-bound): Document new param option.
866 2007-01-14 Uros Bizjak <ubizjak@gmail.com>
869 * config/i386/i386.c (print_operand) ['z']: Output 'b' for
872 2007-01-14 Jan Hubicka <jh@suse.cz>
874 * tree-dfa.c (remove_referenced_var): New function.
875 * tree-ssa-live.c (remove_unused_locals): Walk referenced vars and
876 prune referenced vars list too.
877 * tree-flow.h (remove_referenced_var): Declare.
879 2007-01-14 Jan Hubicka <jh@suse.cz>
881 * tree-eh.c (add_stmt_to_eh_region_fn): Do not add call_exprs
883 (remove_stmt_from_eh_region_fn): Do not remove call_exprs.
884 (verify_eh_throw_stmt_node, verify_eh_throw_table_statements): Kill.
885 * except.h (verify_eh_throw_table_statements): Kill prototype.
886 * cfgexpand.c (expand_gimple_basic_block): Propagate Eh regions
887 into call exrepssions.
888 * tree-optimize.c (execute_free_cfg_annotatiosn): Do not call
890 * tree-cfg.c: Include pointer-set.h.
891 (verify_node_sharing): Work on pointer set.
892 (verify_eh_throw_stmt_node): New.
893 (verify_stmts): Use pointers sets, verify throw_stmt.
895 2007-01-13 Zdenek Dvorak <dvorakz@suse.cz>
897 * ipa-reference.c (analyze_function): Consider also addresses taken
900 2007-01-12 Roger Sayle <roger@eyesopen.com>
902 * c-typeck.c (null_pointer_constant_p): Replace use of
903 TREE_CONSTANT_OVERFLOW with TREE_OVERFLOW.
904 (build_c_cast): Likewise.
906 2007-01-12 Roger Sayle <roger@eyesopen.com>
908 * tree.h (force_fit_type_double): Remove unused final argument.
909 * c-common.c (constant_expression_warning): Replace use of
910 TREE_CONSTANT_OVERFLOW with TREE_OVERFLOW.
911 (convert_and_check): Likewise.
912 (shorten_compare): Update call to force_fit_type_double.
913 (c_common_truthvalue_conversion) <INTEGER_CST>: Use integer_zerop.
914 * convert.c (convert_to_pointer): Update call to
915 force_fit_type_double.
916 * fold-const.c (force_fit_type_double): Remove overflowed_const
918 (int_const_binop, fold_convert_const_int_from_int,
919 fold_convert_const_int_from_real, fold_div_compare,
920 fold_sign_changed_comparison, fold_unary, fold_negate_const,
921 fold_abs_const, fold_not_const): Remove the final argument from
922 calls to force_fit_type_double.
924 2007-01-12 Andrew Pinski <andrew_pinski@playstation.sony.com>
926 * configure.ac: Set insn to "nop" for spu-*-* also.
927 * configure: Regenerate.
929 2007-01-12 Olga Golovanevsky <olga@il.ibm.com>
931 * builtins.def : Add BUILT_IN_FREE.
933 2007-01-12 Jan Hubicka <jh@suse.cz>
935 PR tree-optimization/30443
936 * tree-inline.c (tree_function_versioning): Do not optimize when
937 cloning for inlining.
939 2007-01-12 Zdenek Dvorak <dvorakz@suse.cz>
941 * doc/loop.texi: Document recording of loop exits.
942 * cfgloopmanip.c (loopify, duplicate_loop): Use alloc_loop.
943 (update_single_exits_after_duplication,
944 update_single_exit_for_duplicated_loop,
945 update_single_exit_for_duplicated_loops): Removed.
946 (duplicate_loop_to_header_edge): Do not call
947 update_single_exits_after_duplication and
948 update_single_exit_for_duplicated_loops.
949 (loop_version): Do not update single_exit information.
950 (fix_loop_structure): Use record_loop_exits instead of
951 mark_single_exit_loops.
952 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update
953 the lists of loop exits.
954 * cfghooks.c (redirect_edge_and_branch, redirect_edge_and_branch_force,
955 split_edge, merge_blocks): Update the lists of loop exits.
956 * modulo-sched.c (sms_schedule): Pass LOOPS_HAVE_RECORDED_EXITS to
958 * loop-init.c (loop_optimizer_init): Call record_loop_exits instead
959 of mark_single_exit_loops.
960 (loop_optimizer_finalize): Call release_recorded_exits.
961 * tree-ssa-loop.c (tree_loop_optimizer_init): Pass
962 LOOPS_HAVE_RECORDED_EXITS to loop_optimizer_init.
963 * tree-vectorizer.c (slpeel_tree_duplicate_loop_to_edge_cfg): Do not
964 update single exit information.
965 * lambda-code.c (perfect_nestify): Ditto.
966 * cfgloop.c (flow_loop_free): Destroy the list of exits of the loop.
967 (mark_single_exit_loops): Removed.
968 (alloc_loop, loop_exit_hash, loop_exit_eq, loop_exit_free,
969 get_exit_descriptions, rescan_loop_exit, record_loop_exits,
970 dump_recorded_exit, dump_recorded_exits, release_recorded_exits): New
972 (get_loop_exit_edges, single_exit): Use recorded exit lists.
973 (add_bb_to_loop, remove_bb_from_loops): Update the lists of loop exits.
974 (verify_loop_structure): Verify consistency of the exit lists.
975 (flow_loops_find): Use alloc_loop. Initialize exits hash.
976 (set_single_exit): Removed.
977 * cfgloop.h (struct loop_exit): New function.
978 (struct loop): single_exit_ field replaced by exits field.
979 (LOOPS_HAVE_MARKED_SINGLE_EXITS): Replaced by LOOPS_HAVE_RECORDED_EXITS.
980 (struct loops): Added exits hash.
981 (mark_single_exit_loops, set_single_exit): Declaration removed.
982 (release_recorded_exits, record_loop_exits, rescan_loop_exit): Declare.
984 2007-01-12 Richard Sandiford <richard@codesourcery.com>
986 * doc/invoke.texi: Avoid use of @headitem.
988 2007-01-12 Richard Sandiford <richard@codesourcery.com>
990 * cse.c (cse_insn): Move HAVE_CC0 code after declarations.
992 2007-01-12 Richard Sandiford <richard@codesourcery.com>
994 * doc/install.texi: Fix m68k-*-* anchor and add m68k-*-* to the
997 2007-01-12 Nathan Sidwell <nathan@codesourcery.com>
998 Richard Sandiford <richard@codesourcery.com>
1000 * doc/invoke.texi: Document -mno-bitfield, -mno-rtd and -mno-short.
1001 * config/m68k/m68k.opt: Resort options.
1002 (mbitfield, mrtd, mshort): Remove RejectNegative properties.
1004 2007-01-12 Nathan Sidwell <nathan@codesourcery.com>
1005 Richard Sandiford <richard@codesourcery.com>
1007 * doc/invoke.texi: Document the macros that are defined by
1008 m68k's -mtune and -mhard-float options.
1009 * config/m68k/m68k-protos.h (m68k_cpp_cpu_ident) Declare.
1010 (m68k_cpp_cpu_family): Likewise.
1011 * config/m68k/m68k.h (TARGET_CPU_CPP_BUILTINS): Add a full set
1012 of __ucfv*__ macros. Define __mcffpu__ if generating code for
1013 ColdFire FPUs. Define __mcf_cpu_* and __mcf_family_* macros.
1014 * config/m68k/m68k.c (m68k_cpp_cpu_ident): New function.
1015 (m68k_cpp_cpu_family): Likewise.
1017 2007-01-12 Richard Sandiford <richard@codesourcery.com>
1019 * config/m68k/m68k.h (TARGET_CPU_CPP_BUILTINS): Treat all mc68*
1020 macros besides mc68000 as tuning macros. Use a switch statement
1021 to set them and mcpu32.
1023 2007-01-12 Julian Brown <julian@codesourcery.com>
1025 * config/m68k/m68k.h: Use TARGET_68040 instead of TARGET_68040_ONLY.
1026 (TARGET_68040_ONLY): Rename to...
1027 (TARGET_68040): ...this.
1028 * config/m68k/m68k.c: Use TARGET_68040 instead of TARGET_68040_ONLY.
1029 * config/m68k/m68k.md: Likewise.
1031 2007-01-12 Julian Brown <julian@codesourcery.com>
1032 Nathan Sidwell <nathan@codesourcery.com>
1033 Richard Sandiford <richard@codesourcery.com>
1035 * config.gcc (m680[012]0-*-*, m68k*-*-*): Set m68k_cpu_ident to
1036 the -mcpu= argument associated with the --with-cpu setting.
1037 Define M68K_DEFAULT_TUNE to the default -mtune= option,
1038 if different from the one implied by the -mcpu setting.
1039 Accept --with-cpu=FOO if FOO is listed in m68k-devices.def,
1040 using mcpu=FOO as the default CPU option. Set target_cpu_default2.
1041 * doc/invoke.texi: Mention ColdFire in the introduction to the
1042 m68k options. Document the new -march, -mcpu, -mtune, -mdiv,
1043 -mno-div and -mhard-float options. Make -m68881 a synonym for
1044 -mhard-float. Document the previously-undocumented -m5206e,
1045 -m528x, -m5307 and -m5407 options. Tweak the existing option
1046 documentation for consistency.
1047 * doc/install.texi: Mention new --with-cpu arguments.
1048 * config/m68k/m68k.h (OPTION_DEFAULT_SPECS): Only use the
1049 default CPU if neither -mcpu nor -march are specified.
1050 (ASM_CPU_SPEC): Pass down -mcpu and -march options.
1051 (TARGET_CPU_CPP_BUILTINS): Set __mcfisa*__ macros from
1052 TARGET_ISA*. Set the legacy __mcf*__ cpu macros in the same way,
1053 using m68k_tune to decide between families that implement the
1054 same ISA. Use m68k_tune to set __mcfv4e__.
1055 (FL_BITFIELD, FL_68881, FL_COLDFIRE, FL_CF_HWDIV, FL_CF_MAC)
1056 (FL_CF_EMAC, FL_CF_EMAC_B, FL_CF_USP, FL_CF_FPU, FL_ISA_68000)
1057 (FL_ISA_68010, FL_ISA_68020, FL_ISA_68040, FL_ISA_A, FL_ISA_B)
1058 (FL_ISA_C, FL_ISA_MMU): New macros.
1059 (MASK_COLDFIRE): Delete.
1060 (TARGET_68010, TARGET_68020, TARGET_68040_ONLY, TARGET_COLDFIRE)
1061 (TARGET_ISAB): Redefine in terms of m68k_cpu_flags.
1062 (TARGET_68881, TARGET_COLDFIRE_FPU): Redefine in terms of m68k_fpu.
1063 (TARGET_HARD_FLOAT): Do not define here.
1064 (TARGET_ISAAPLUS, TARGET_ISAC): New macros.
1065 (TUNE_68000): New macro.
1066 (TUNE_68000_10): Redefine in terms of TUNE_68000 and TUNE_68010.
1067 (TUNE_68010, TUNE_68030, TUNE_68040, TUNE_68060, TUNE_CPU32)
1068 (TUNE_CFV2): Redefine in terms of m68k_tune.
1069 (uarch_type, target_device, fpu_type): New enums.
1070 (m68k_cpu, m68k_tune, m68k_fpu, m68k_cpu_flags): Declare.
1071 * config/m68k/m68k.c (TARGET_DEFAULT): Remove MASK_68881.
1072 (FL_FOR_isa_00, FL_FOR_isa_10, FL_FOR_isa_20, FL_FOR_isa_40)
1073 (FL_FOR_isa_cpu32, FL_FOR_isa_a, FL_FOR_isa_aplus, FL_FOR_isa_b)
1074 (FL_FOR_isa_c): New macros.
1075 (m68k_isa): New enum.
1076 (m68k_target_selection): New structure.
1077 (all_devices, all_isas, all_microarchs): New tables.
1078 (m68k_cpu_entry, m68k_arch_entry, m68k_tune_entry, m68k_cpu)
1079 (m68k_tune, m68k_fpu, m68k_cpu_flags): New variables.
1080 (MASK_ALL_CPU_BITS): Delete.
1081 (m68k_find_selection): New function.
1082 (m68k_handle_option): Handle -mcpu=, -march= and -mtune=.
1083 Map the legacy target options to a combination of the new ones.
1084 (override_options): Set m68k_cpu, m68k_tune, m68k_fpu and
1085 m68k_cpu_flags. Handle M68K_DEFAULT_TUNE. Use m68k_cpu_flags
1086 to derive default MASK_BITFIELD, MASK_CF_HWDIV and MASK_HARD_FLOAT
1088 * config/m68k/m68k.opt (m5200, m5206e, m528x, m5307, m5407, mcfv4e)
1089 (m68010, m68020, m68020-40, m68020-60, m68030, m68040): Remove Mask
1091 (m68881, msoft-float): Change mask from 68881 to HARD_FLOAT.
1092 (march=, mcpu=, mdiv, mhard-float, mtune=): New options.
1093 * config/m68k/m68k-devices.def: New file.
1095 2007-01-12 Richard Sandiford <richard@codesourcery.com>
1096 Nathan Sidwell <nathan@codesourcery.com>
1098 * config/m68k/m68k.h (ASM_CPU_SPEC, ASM_SPEC, EXTRA_SPECS)
1099 (SUBTARGET_EXTRA_SPECS): New macros.
1100 * config/m68k/linux.h (ASM_SPEC): Remove CPU flags;
1101 use %(asm_cpu_spec) instead.
1102 * config/m68k/m68k-none.h (ASM_SPEC): Likewise.
1103 * config/m68k/openbsd.h (ASM_SPEC): Likewise.
1104 * config/m68k/netbsd-elf.h (ASM_SPEC): Likewise.
1105 (EXTRA_SPECS): Rename to...
1106 (SUBTARGET_EXTRA_SPECS): ...this.
1108 2007-01-12 Nathan Sidwell <nathan@codesourcery.com>
1109 Richard Sandiford <richard@codesourcery.com>
1110 Julian Brown <julian@codesourcery.com>
1112 * config.gcc (m68k-*-aout*, m68k-*-coff*, m68020-*-elf*, m68k-*-elf*)
1113 (m68k-*-uclinux*, m68k-*-linux*, m68k-*-rtems*): Set default_m68k_cpu
1114 to the configuration's default CPU.
1115 (m68010-*-netbsdelf*, m68k*-*-netbsdelf*, m68k*-*-openbsd*): Likewise.
1116 Remove default masks.
1117 (m680[012]0-*-*): Set the default with_cpu to the first part of
1119 (m68k*-*-*): Set the default with_cpu to m$default_m68k_cpu.
1120 (m68k*-*-linux): Extend the --with-cpu handling to...
1121 (m680[012]0-*-*, m68k*-*-*): ...these configurations. Allow m68000
1122 and m68010. Don't set target_cpu_default2.
1123 * doc/install.texi: Document --with-cpu for m68k.
1124 * config/m68k/m68k.h (OPTION_DEFAULT_SPECS): Define.
1125 * config/m68k/m68k-none.h (TARGET_CPU_DEFAULT, M68K_CPU_m68k)
1126 (M68K_CPU_m68000, M68K_CPU_m68010, M68K_CPU_m68020, M68K_CPU_m68030)
1127 (M68K_CPU_m68040, M68K_CPU_m68302, M68K_CPU_m68332, TARGET_DEFAULT)
1128 (ASM_CPU_DEFAULT_SPEC, CC1_CPU_DEFAULT_SPEC): Delete.
1129 (ASM_SPEC): Remove use of %(asm_cpu_default).
1130 (EXTRA_SPECS, SUBTARGET_EXTRA_SPECS, MULTILIB_DEFAULTS): Delete.
1131 * config/m68k/linux.h (TARGET_DEFAULT): Delete.
1132 (CPP_SPEC): Merge definitions. Do not handle __HAVE_68881__ here.
1133 * config/m68k/netbsd-elf.h (TARGET_OS_CPP_BUILTINS): Define
1134 __HAVE_FPU__ if TARGET_HARD_FLOAT.
1135 (TARGET_DEFAULT): Delete.
1136 (EXTRA_SPECS): Delete cpp_cpu_default_spec, cpp_cpu_spec,
1137 cpp_fpu_spec, asm_default_spec and netbsd_cpp_spec.
1138 (CPP_CPU_SPEC): Delete.
1139 (TARGET_VERSION): Merge definitions, using TARGET_68010 to pick
1140 the appropriate string.
1141 (CPP_CPU_DEFAULT_SPEC, ASM_DEFAULT_SPEC, CPP_FPU_SPEC): Delete.
1142 (CPP_SPEC): Define to NETBSD_CPP_SPEC.
1143 (ASM_SPEC): Don't use %(asm_default_spec).
1144 * config/m68k/m68k.c (TARGET_DEFAULT_TARGET_FLAGS): Remove
1145 TARGET_DEFAULT and add MASK_68881.
1146 * config/m68k/m68k.md: Remove mention of TARGET_DEFAULT from comments.
1148 2007-01-12 Richard Sandiford <richard@codesourcery.com>
1150 * config.gcc (m68010-*-netbsdelf*): Add MASK_68010.
1151 (m68k*-*-netbsdelf*, m68k*-*-openbsd*, m68k*-linux*): Add
1152 MASK_68010 alongside MASK_68020.
1153 * doc/invoke.texi: Document -m68010.
1154 * config/m68k/m68k.opt (m68010): New.
1155 * config/m68k/m68k.h (TARGET_CPU_CPP_BUILTINS): Define mc68010
1157 (TUNE_68010): New macro.
1158 * config/m68k/m68k-none.h (M68K_CPU_m68k, M68K_CPU_m68010)
1159 (M68K_CPU_m68020, M68K_CPU_m68030, M68K_CPU_m68040)
1160 (M68K_CPU_m68332): Add MASK_68010.
1161 * config/m68k/linux.h (TARGET_DEFAULT): Add MASK_68010 to
1162 fallback definition.
1163 * config/m68k/netbsd-elf.h (CPP_CPU_SPEC): Remove now-redundant
1165 * config/m68k/m68k.c (MASK_ALL_CPU_BITS): Add MASK_68010.
1166 (m68k_handle_option): Handle OPT_m68010. Add MASK_68010
1167 to all entries that use MASK_68020.
1168 (output_move_simode_const, output_move_himode, output_move_qimode)
1169 (output_move_stricthi, output_move_strictqi): Use TARGET_68010
1170 instead of TARGET_68020 to select clr behavior. Remove comment
1171 about there being no TARGET_68010.
1172 * config/m68k/m68k.md: Likewise throughout.
1174 2007-01-12 Julian Brown <julian@codesourcery.com>
1176 * config/m68k/m68k.h (TARGET_ISAB): New macro.
1177 * config/m68k/m68k.c: Use TARGET_ISAB rather than TARGET_CFV4.
1178 * config/m68k/m68k.md: Likewise.
1180 2007-01-12 Julian Brown <julian@codesourcery.com>
1182 * config/m68k/m68k.h (LEGITIMATE_INDEX_P, LEGITIMIZE_ADDRESS): Use
1183 TARGET_COLDFIRE_FPU instead of TARGET_CFV4E.
1185 2007-01-12 Julian Brown <julian@codesourcery.com>
1187 * config/m68k/m68k.h (TUNE_68040_60): New macro.
1188 * config/m68k/m68k.c (standard_68881_constant_p): Use it.
1189 * config/m68k/m68k.md: Likewise.
1191 2007-01-12 Julian Brown <julian@codesourcery.com>
1192 Richard Sandiford <richard@codesourcery.com>
1194 * config/m68k/m68k.h (TARGET_CPU_CPP_BUILTINS): Use TUNE_68030
1195 instead of TARGET_68030, TUNE_68040 instead of TARGET_68040,
1196 TUNE_68060 instead of TARGET_68060 and TUNE_CPU32 instead of
1198 (TARGET_CPU32): Rename to...
1199 (TUNE_CPU32): ...this.
1200 (TUNE_68000_10, TUNE_68030, TUNE_68040, TUNE_68060)
1201 (TUNE_CFV2): New macros.
1202 * config/m68k/netbsd-elf.h (LONG_DOUBLE_TYPE_SIZE): Simplify;
1203 remove conditions that are implied by TARGET_68020.
1204 * config/m68k/m68k.c (m68k_output_function_prologue): Use TUNE_68040
1205 instead of TARGET_68040 and TUNE_CPU32 instead of TARGET_CPU32.
1206 (m68k_output_function_epilogue): Likewise.
1207 (m68k_rtx_costs): Likewise. Use TUNE_68060 instead of TARGET_68060
1208 and TUNE_CFV2 instead of TARGET_5200. Use TUNE_68000_10 instead of
1209 "!TARGET_68020 && !TARGET_COLDFIRE" to choose between 68000 and
1210 non-68000 timings. Refactor multiplication and division costs.
1211 (output_addsi3): Use TUNE_68040 instead of TARGET_68040 and
1212 TUNE_CPU32 instead of TARGET_CPU32.
1213 (standard_68881_constant_p): Use TUNE_68040 instead of TARGET_68040
1214 and TUNE_68060 instead of TARGET_68060.
1215 * config/m68k/m68k.md: Use TUNE_68040 instead of TARGET_68040,
1216 TUNE_68060 instead of TARGET_68060, and TUNE_CPU32 instead of
1218 (movsi_const0): Use TUNE_68000_10 rather than "!TARGET_68020
1219 && !TARGET_COLDFIRE" to choose between moveq and clr.
1220 Likewise in the unnamed movsf pattern.
1221 (ashlsi_17_24, lshrsi_17_24): Guard with TUNE_68000_10 rather than
1222 "!TARGET_68020 && !TARGET_COLDFIRE". Likewise the unnamed
1225 2007-01-12 Richard Sandiford <richard@codesourcery.com>
1227 * config/m68k/m68k.h (TARGET_CPU_CPP_BUILTINS): Increase amount
1228 of tabbing before backslashes.
1230 2007-01-11 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
1232 * pa-linux.h (ASM_OUTPUT_INTERNAL_LABEL): Undefine.
1233 * pa.h (ASM_OUTPUT_LABEL): Output colon when using GAS.
1234 (ASM_OUTPUT_INTERNAL_LABEL): Define.
1236 2007-01-11 Zdenek Dvorak <dvorakz@suse.cz>
1238 * tree-ssa-loop-ivopts.c (extract_cond_operands): Split from
1239 find_interesting_uses_cond.
1240 (find_interesting_uses_cond): Use extract_cond_operands.
1241 (rewrite_use_compare): Use extract_cond_operands and
1242 force_gimple_operand_bsi. Do not call update_stmt.
1243 (determine_use_iv_cost_condition): Use extract_cond_operands.
1244 Return cheaper of using original bound and changing the exit bound.
1246 2007-01-11 Zdenek Dvorak <dvorakz@suse.cz>
1248 PR tree-optimization/29516
1249 * tree-ssa-address.c (tree_mem_ref_addr, add_to_parts,
1250 most_expensive_mult_to_index, addr_to_parts,
1251 create_mem_ref, maybe_fold_tmr): Make the type of
1252 fields of TARGET_MEM_REF sizetype.
1253 (move_fixed_address_to_symbol, move_pointer_to_base):
1255 * tree.def (TARGET_MEM_REF): Add comment on types of
1258 2007-01-11 Joseph Myers <joseph@codesourcery.com>
1260 * c-common.c (vector_types_convertible_p): Treat opaque types as
1261 always convertible if they have the same size, but not otherwise.
1263 2007-01-11 Steven Bosscher <steven@gcc.gnu.org>
1265 * ifcvt.c (struct noce_if_info): Add comments to the fields.
1266 Remove the b_unconditional field.
1267 (noce_try_sign_mask): Do not look at b_unconditional.
1268 (noce_process_if_block): Do not use merge_if_blocks. Update
1269 the CFG here. Do not set b_unconditional.
1270 (cond_move_process_if_block): Likewise.
1271 (find_cond_trap): Likewise.
1272 (check_cond_move_block): Require simple jump insns at the end
1275 2007-01-11 Jan Hubicka <jh@suse.cz>
1277 PR tree-optimization/1046
1278 * tree-tailcall.c (suitable_for_tail_call_opt_p): Use TREE_ADDRESSABLE
1279 when alias info is not ready.
1280 (pass_tail_recursion): Do not require aliasing.
1281 * tree-ssa-copyrename.c (pass_rename_ssa_cop): Likewise.
1282 * tree-ssa-ccp.c (pass_ccp, pass_fold_builtins): Likewise.
1283 * tree-ssa-copy.c (pass_copy_prop): Likewise.
1284 * tree-ssa-forwprop.c (pass_forwprop): Likewise.
1285 * tree-ssa-dce.c (pass_dce, pass_dce_loop, pass_cd_dce): Likewise.
1286 * passes.c (init_optimization_passes): Execute rename_ssa_copies,
1287 ccp, forwprop, copy_prop, merge_phi, copy_prop, dce and tail recursion
1289 * tree-ssa-operands.c (add_virtual_operand, get_indirect_ref_operand):
1290 When aliasing is not build, mark statement as volatile.
1292 2007-01-11 Tom Tromey <tromey@redhat.com>
1294 PR preprocessor/15185, PR preprocessor/20989:
1295 * doc/cppopts.texi <-MT>: Update description of algorithm for
1296 computing default target.
1297 <-M, -MD>: Reword "basename" text.
1299 2007-01-11 Roger Sayle <roger@eyesopen.com>
1301 * builtins.c (expand_builtin_pow, expand_builtin_powi,
1302 fold_builtin_cabs, fold_builtin_sqrt, fold_builtin_trunc,
1303 fold_builtin_floor, fold_builtin_ceil, fold_builtin_round,
1304 fold_builtin_int_int_roundingfn, fold_builtin_bitop,
1305 fold_builtin_bswap, real_constp, fold_builtin_pow,
1306 fold_builtin_powi, fold_builtin_signbit, fold_builtin_copysign,
1307 do_mpfr_arg1, do_mpfr_arg2, do_mpfr_arg3, do_mpfr_sincos): Replace
1308 uses of the macro TREE_CONSTANT_OVERFLOW with TREE_OVERFLOW.
1309 * convert.c (convert_to_pointer): Likewise.
1310 * expr.c (highest_pow2_factor, expand_expr_real_1): Likewise.
1311 * fold-const.c (force_fit_type, fold_negate_expr, int_const_binop,
1312 const_binop, fold_convert_const_int_from_int,
1313 fold_convert_const_int_from_real,
1314 fold_convert_const_real_from_real, sign_bit_p,
1315 optimize_minmax_comparison, extract_muldiv_1, fold_div_compare,
1316 fold_sign_changed_comparison, fold_unary, fold_comparison,
1317 fold_binary, multiple_of_p, tree_Expr_non_zero_p,
1318 fold_negate_const, fold_abs_const, fold_not_const): Likewise.
1319 * print-tree.c (print_node_brief, print_node): Likewise.
1320 * stor-layout.c (place_field, layout_type): Likewise.
1321 * tree-chrec.c (keep_cast): Likewise.
1322 * tree.c (build_vector, build_real, build_real_from_int_cst,
1323 build_complex): Likewise.
1325 2007-01-11 Roger Sayle <roger@eyesopen.com>
1327 * tree.h (TREE_CONSTANT_OVERFLOW): Obsolete. For the time being,
1328 treat TREE_CONSTANT_OVERFLOW as a synonym of TREE_OVERFLOW.
1330 2007-01-11 Paolo Bonzini <bonzini@gnu.org>
1332 * configure.ac (strict1_warn): Rename to strict_warn.
1333 (WERROR, --enable-werror, symlink hacks, stage1_cflags,
1334 cc_set_by_configure, quoted_cc_set_by_configure,
1335 stage_prefix_set_by_configure, quoted_stage_prefix_set_by_configure,
1336 all_boot_languages, all_stagestuff): Remove.
1337 (target_list): Remove bootstrap targets.
1338 * Makefile.in (quickstrap): Unconditionally make a synonym of all.
1339 (BOOT_LANGUAGES, STAGE1_CFLAGS, STAGE1_CHECKING,
1340 REMAKEFLAGS, FLAGS_TO_PASS, PREPEND_DOTDOT_TO_RELATIVE_PATHS,
1341 SUBDIR_FLAGS_TO_PASS, WERROR_FLAGS, STRICT2_WARN, LANG_STAGESTUFF,
1342 VOL_FILES, POSTSTAGE1_FLAGS_TO_PASS, STAGE2_FLAGS_TO_PASS,
1343 STAGEPROFILE_FLAGS_TO_PASS, STAGEFEEDBACK_FLAGS_TO_PASS, stage1_build,
1344 stage1_copy, stage2_build, stage2_copy, stageprofile_build,
1345 stageprofile_copy, stage3_build, stage3_copy, stagefeedback_build,
1346 stagefeedback_copy, stage4_build, clean_s1, clean_sw, bootstrap,
1347 bootstrap-lean, bootstrap2, bootstrap2-lean, bootstrap3,
1348 bootstrap3-lean, bootstrap4, bootstrap4-lean, unstage1, unstage2,
1349 unstage3, unstage4, unstageprofile, unstagefeedback, restage, restage2,
1350 restage3, restage4, restageprofile, restagefeedback, bubbleestrap,
1351 cleanstrap, unstrap, restrap, *compare, *compare3, *compare4,
1352 *compare-lean, *compare3-lean, *compare4-lean, stage1-start, stage1,
1353 stage2-start, stage2, stage3-start, stage3, stage4-start, stage4,
1354 stageprofile-start, stageprofile, stagefeedback-start, stagefeedback,
1355 risky-stage1, risky-stage2, risky-stage3, risky-stage4): Remove.
1356 (ORDINARY_FLAGS_TO_PASS): Rename to FLAGS_TO_PASS.
1357 (STAGECOPYSTUFF, STAGEMOVESTUFF): Consolidate into MOSTLYCLEANFILES.
1358 (mostlyclean): Adjust.
1359 (clean, distclean): Don't mention bootstrap stuff.
1360 * configure: Regenerate.
1361 * ada/config-lang.in, cp/config-lang.in, forttran/config-lang.in,
1362 java/config-lang.in, objc/config-lang.in, objcp/config-lang.in,
1363 treelang/config-lang.in (stagestuff): Remove.
1364 * doc/sourcebuild.texi (stage1, stage2, stage3, stage4,
1365 stageprofile, stagefeedback, stagestuff): Remove mention.
1367 2007-01-11 Nick Clifton <nickc@redhat.com>
1369 * config/mcore/predicates.md (mcore_general_movesrc_operand):
1371 (mcore_general_movdst_operand): Do not accept CONST_INTs.
1372 (mcore_arith_K_S_operand): Run the test for the S constraint not
1373 the test for the M constraint.
1374 (mcore_addsub_operand): Do not accept integer values that are
1375 larger than 32 bits.
1376 * config/mcore/mcore.md: Remove unused constraints from split.
1377 (andsi3): Use HOST_WIDE_INT instead of int to hold an INTVAL.
1379 (allocate_stack): Likewise.
1380 * config/mcore/mcore.c (mcore_print_operand): Restrict output of P
1381 operands to 32 bits.
1382 (mcore_const_costs): Use HOST_WIDE_INT instead of int to hold an
1384 (mcore_and_cost, mcore_modify_comparison, const_ok_for_mcore,
1385 mcore_const_ok_for_inline, mcore_const_trick_uses_not,
1386 try_constant_tricks, mcore_num_ones, mcore_num_zeros,
1387 mcore_output_bclri, mcore_output_andn, output_inline_const,
1388 mcore_output_move, mcore_output_movedouble): Likewise.
1389 (mcore_output_cmov): Use CONST_OK_FOR_M and CONST_OK_FOR_N.
1390 (output_inline_const): Likewise.
1391 (output_inline_const): Fix format strings used in sprintf
1393 * config/mcore/mcore-protos.h: Update prototypes for changed
1394 functions in mcore.c.
1395 * config/mcore/mcore.h (CONST_OK_FOR_I): Cast values to
1396 HOST_WIDE_INT and not int.
1397 (CONST_OK_FOR_J, CONST_OK_FOR_K, CONST_OK_FOR_L, CONST_OK_FOR_M,
1398 CONST_OK_FOR_N): Likewise.
1399 (LEGITIMATE_CONSTANT_P): Also check CONSTANT_P.
1400 (GO_IF_LEGITIMATE_INDEX): Use HOST_WIDE_INT instead of int to hold
1403 2007-01-10 Jan Hubicka <jh@suse.cz>
1405 * tree-vrp.c (remove_range_assertions): Release defs.
1406 * tree-ssa-loop-ivopts.c (rmeove_statement): Likewise.
1407 * tree-ssa-dom.c (remove_stmt_or_phi): Likewise.
1409 2007-01-10 Paul Brook <paul@codesourcery.com>
1411 * config/arm/arm.c (arm_rtx_costs_1): Handle mutiply-subtract.
1412 * config/arm/arm.md (mulsi3subsi): New insn.
1414 2007-01-10 Zdenek Dvorak <dvorakz@suse.cz>
1416 * tree-ssa-loop-manip.c (tree_unroll_loop): Make it a wrapper over ...
1417 (tree_transform_and_unroll_loop): New.
1418 * tree-flow.h (transform_callback, tree_transform_and_unroll_loop):
1421 2007-01-10 Robert Kennedy <jimbob@google.com>
1423 * fold-const.c (fold_comparison): Fold comparisons like (x *
1424 1000 < 0) to (x < 0).
1426 2007-01-10 Ian Lance Taylor <iant@google.com>
1428 * tree-pretty-print.c (dump_generic_node): Print parentheses when
1429 operands have the same priority.
1431 2007-01-10 Tom Tromey <tromey@redhat.com>
1433 * fold-const.c (fold_truthop): Don't check can_use_bit_fields_p.
1434 (fold_binary): Likewise.
1435 * langhooks.c (lhd_can_use_bit_fields_p): Removed.
1436 * langhooks-def.h (lhd_can_use_bit_fields_p): Removed.
1437 (LANG_HOOKS_CAN_USE_BIT_FIELDS_P): Removed.
1438 (LANG_HOOKS_INITIALIZER): Remove LANG_HOOKS_CAN_USE_BIT_FIELDS_P.
1439 * langhooks.h (struct lang_hooks): Removed field
1440 'can_use_bit_fields_p'.
1442 2007-01-10 Ralf Corsépius <ralf.corsepius@rtems.org>
1444 * config/bfin/t-bfin, config/bfin/t-bfin-elf: Remove GCC_CFLAGS.
1446 2007-01-10 Razya Ladelsky <razya@il.ibm.com>
1448 * function.c (get_last_funcdef_no): New function.
1449 * function.h (get_last_funcdef_no): Declare.
1450 * tree-inline.c (initialize_cfun): Add initialization.
1451 (tree_function_versioning): Cleanup.
1453 2007-01-10 Jan Hubicka <jh@suse.cz>
1455 * tree-inline.c (setup_one_parameter): Do not propagate into abnormal
1458 2007-01-10 Sa Liu <saliu@de.ibm.com>
1459 Ben Elliston <bje@au.ibm.com>
1461 * spu.h (STACK_SAVE_AREA): Use VOIDmode for SAVE_FUNCTION, SImode
1462 for SAVE_NONLOCAL and Pmode for any other save level.
1463 * spu-protos.h (spu_restore_stack_block): Declare.
1464 * spu.md (save_stack_block): Remove.
1465 (restore_stack_block): Call spu_restore_stack_block.
1466 * spu.c (spu_restore_stack_block): New function.
1467 (spu_expand_epilogue): Remove old comment.
1469 2007-01-09 Zdenek Dvorak <dvorakz@suse.cz>
1471 PR tree-optimization/30322
1472 * tree-ssa-loop-ivopts.c (fold_affine_expr, iv_value): Removed.
1473 (cand_value_at): Return the value as aff_tree.
1474 (may_eliminate_iv): Convert the bound from aff_tree to tree.
1475 * tree-affine.c (aff_combination_add_cst, aff_combination_add_product,
1476 aff_combination_mult): New functions.
1477 (aff_combination_add): Use aff_combination_add_cst.
1478 (aff_combination_convert): Allow conversions to a wider type.
1479 (tree_to_aff_combination): Handle BIT_NOT_EXPR.
1480 * tree-affine.h (aff_combination_mult): Declare.
1482 2007-01-09 Carlos O'Donell <carlos@codesourcery.com>
1484 * doc/tm.texi: Update documentation to reflect reality of exec
1485 and start file search behaviours. Update copyright year.
1486 * doc/invoke.texi: Explain how GCC_EXEC_PREFIX is used to find
1487 header file directories.
1489 2007-01-09 Uros Bizjak <ubizjak@gmail.com>
1491 * config/i386/i386.md (*sinxf2): Rename to *sinxf2_i387.
1492 (*cosxf2): Rename to cosxf2_i387.
1493 (*sindf2, *sinsf2): Extend operand 1 to XFmode. Macroize patterns
1494 using X87MODEF12 mode macro. Rename patterns to
1495 *sin_extend<mode>xf2_i387. Use SSE_FLOAT_MODE_P to disable patterns
1497 (*cosdf2, *cossf2): Ditto.
1498 (sincosdf3, sincossf3): Ditto. Rewrite corresponding splitters
1499 to match extended input operands.
1500 (sincos<mode>3): New expander.
1501 (*sinextendsfdf2, *cosextendsfdf2, *sincosextendsfdf3): Remove
1502 insn patterns and corresponding splitters.
1504 2007-01-09 Kaz Kojima <kkojima@gcc.gnu.org>
1506 * config/sh/t-linux (TARGET_LIBGCC2_CFLAGS): Delete.
1507 (SHLIB_MAPFILES, SHLIB_LINK, SHLIB_INSTALL): Likewise.
1509 2007-01-09 Nicolas Pitre <nico@cam.org>
1512 * arm/ieee754-df.S (Lad_s): Also test the low word of X for zero.
1514 2007-01-08 Geoffrey Keating <geoffk@apple.com>
1516 * target.h (struct gcc_target): New field library_rtti_comdat.
1517 * target-def.h (TARGET_CXX_LIBRARY_RTTI_COMDAT): New.
1518 (TARGET_CXX): Add TARGET_CXX_LIBRARY_RTTI_COMDAT.
1519 * doc/tm.texi (C++ ABI): Document TARGET_CXX_LIBRARY_RTTI_COMDAT.
1520 * config/darwin.h (TARGET_CXX_LIBRARY_RTTI_COMDAT): Define.
1522 2007-01-08 Geoffrey Keating <geoffk@apple.com>
1524 * doc/invoke.texi (Optimize Options): Correct description of -O0.
1526 2007-01-08 Richard Guenther <rguenther@suse.de>
1528 * tree.h (force_fit_type_double): Export.
1529 (force_fit_type): Remove.
1530 * fold-const.c (force_fit_type_double): New function.
1531 (force_fit_type): Remove.
1532 (int_const_binop): Use it.
1533 (fold_convert_const_int_from_int): Likewise.
1534 (fold_convert_const_int_from_real): Likewise.
1535 (fold_div_compare): Likewise.
1536 (fold_sign_changed_comparison): Likewise.
1537 (fold_unary): Likewise.
1538 (fold_negate_const): Likewise.
1539 (fold_abs_const): Likewise.
1540 (fold_not_const): Likewise.
1541 * c-common.c (shorten_compare): Use force_fit_type_double.
1542 * convert.c (convert_to_pointer): Likewise.
1544 2007-01-08 Richard Guenther <rguenther@suse.de>
1546 * tree.h (build_int_cst_wide_type): Export.
1547 * tree.c (build_int_cst_wide_type): New function.
1548 (build_int_cst_wide): Fix comment.
1549 * builtins.c (fold_builtin_object_size): Use build_int_cst
1550 to build -1 or 0 of the correct type. Use fit_double_type
1551 to check for overflow.
1552 * fold-const.c (optimize_bit_field_compare): Use build_int_cst_type
1554 (decode_field_reference): Likewise.
1555 (all_ones_mask_p): Likewise.
1556 (native_interpret_int): Use build_int_cst_wide_type.
1557 (fold_binary): Use build_int_cst_type to build an all-ones
1559 * stor-layout.c (set_sizetype): Use build_int_cst_wide_type.
1561 2007-01-08 Daniel Jacobowitz <dan@codesourcery.com>
1563 * config/pa/t-pa64 (libgcc_stub.a): Use $(T).
1565 2007-01-09 Ben Elliston <bje@au.ibm.com>
1567 * genautomata.c (STATS_OPTION): New option.
1568 (stats_flag): New flag.
1569 (gen_automata_option): Handle it.
1570 (initiate_automaton_gen): Ditto.
1571 (write_automata): Output statistics only if stats_flag is
1572 set. Likewise, output time statistics only if time_flag is set.
1573 * doc/md.texi (Processor pipeline description): Document new flag.
1575 2007-01-08 Richard Guenther <rguenther@suse.de>
1577 * builtins.c (fold_builtin_int_roundingfn): Use fit_double_type.
1578 * tree.c (build_int_cst_type): Likewise.
1579 (size_in_bytes): Don't call force_fit_type on the result.
1580 (int_fits_type_p): Use fit_double_type.
1581 * fold-const.c (fit_double_type): New function.
1582 (force_fit_type): Use it.
1583 * tree.h (fit_double_type): Export.
1585 2007-01-08 Jan Hubicka <jh@suse.cz>
1587 * tree-vectorizer.c (gate_increase_alignment): Fix return type.
1588 * ipa.c (function_and_variable_visibility): Fix return type.
1590 2007-01-08 Richard Guenther <rguenther@suse.de>
1592 * tree-ssa-ccp.c (maybe_fold_offset_to_array_ref): Use type
1593 of offset to build the index.
1594 * tree-pretty-print.c (dump_generic_node): Don't build negated
1595 const just for printing.
1596 * c-pretty-print.c (pp_c_integer_constant): Likewise.
1597 * builtins.c (fold_builtin_int_roundingfn): Check if result
1598 fits the type by using force_fit_type and comparing the result.
1599 * predict.c (predict_loops): Use compare_tree_int for comparison.
1600 * tree.c (build_int_cst): Fall back to integer_type_node for
1602 (build_int_cst_wide): Assert type is non-null.
1604 2007-01-08 Roberto Costa <roberto.costa@st.com>
1606 * tree-vrp.c (extract_range_from_cond_expr): New.
1607 (extract_range_from_expr): Handle COND_EXPR nodes used as expressions.
1608 * tree-ssa-ccp.c (get_maxval_strlen): Handle COND_EXPR nodes used
1610 (fold_stmt): Bug fix, avoid infinite recursion when folding COND_EXPRs.
1611 * tree-ssa-forwprop.c (simplify_cond, forward_propagate_into_cond,
1612 tree_ssa_forward_propagate_single_use_vars): Handle COND_EXPR nodes
1613 used as expressions.
1614 * tree-object-size.c (cond_expr_object_size): New.
1615 (collect_object_sizes_for): Handle COND_EXPR nodes used as expressions.
1617 2007-01-08 Jan Hubicka <jh@suse.cz>
1619 * tree-ssa-forwprop.c (forward_propagate_into_cond,
1620 tree_ssa_forward_propagate_single_use_va): Release defs of propagated
1623 2007-01-08 Richard Guenther <rguenther@suse.de>
1625 PR tree-optimization/23603
1626 * tree-vrp.c (set_value_range_to_truthvalue): New function.
1627 (extract_range_from_binary): Fall back to truthvalue instead of
1628 varying for TRUTH_*_EXPR.
1629 (extract_range_from_comparison): Fall back to truthvalue instead of
1631 (vrp_visit_phi_node): Don't adjust new range bounds to +INF/-INF
1632 if all visited PHI values were constant.
1634 2007-01-08 Jan Hubicka <jh@suse.cz>
1636 * cgraphunit.c (cgraph_process_new_functions): Reset reachable flag.
1637 (cgraph_analyze_function): break out from ...
1638 (cgraph_finalize_compilation_unit): ... here.
1639 (cgraph_expand_function): Remove forgoten commented out line.
1640 (cgraph_optimize): Analyze functions.
1642 2007-01-08 Jan Hubicka <jh@suse.cz>
1644 * tree-pas.h (TODO_remove_function): New flag.
1645 (TODO_update*): Renumber.
1646 (pass_ipa_increase_alignment,
1647 pass_ipa_function_and_variable_visibility): New passes.
1648 * cgraphunit.c (cgraph_increase_alignment): Move to tree-vectorizer.c
1649 (cgraph_function_and_variable_visibility): Move to ipa.c
1650 (cgraph_optimize): Don't call cgraph_function_and_variable_visibility,
1651 cgraph_increase_alignment.
1652 * ipa-inline.c (cgraph_decide_inlining): Don't push timevar.
1653 (cgraph_decide_inlining_incrementally): Push TV_INTEGRATION before
1654 calling tree-inline.
1655 (cgraph_early_inlining): Do not call cgraph_remove_unreachable_nodes.
1656 (pass_ipa_inline, pass_early_ipa_inlining): Set TODO_remove_functions
1657 * tree-vectorizer.c (increase_alignment): Move here from cgraphunit.c
1658 (gate_increase_alignment): New function.
1659 (pass_ipa_increase_alignment): New pass.
1660 * ipa.c: Inline tree-pass.h and timevar.h
1661 (function_and_variable_visibility): Move here from cgraphunit.c
1662 * tree-optimize.c (pass_early_local_passes): Add TODO_remove_functions.
1663 * passes.c (init_optimization_passes): Add the two new passes.
1664 (execute_todo): Handle cgraph_remove_functions.
1666 2007-01-08 Nick Clifton <nickc@redhat.com>
1668 * config/frv/predicates.md (reg_or_0_operand): Accept
1671 2007-01-08 Ralf Corsépius <ralf.corsepius@rtems.org>
1673 * config/bfin/rtems.h, config/bfin/t-rtems: New.
1674 * config.gcc: Add bfin*-rtems*.
1676 2007-01-08 Mark Shinwell <shinwell@codesourcery.com>
1678 * c.opt: Add -flax-vector-conversions.
1679 * c-typeck.c (convert_for_assignment): Pass flag to
1680 vector_types_convertible_p to allow emission of note.
1681 (digest_init): Likewise.
1682 * c-opts.c: Handle -flax-vector-conversions.
1683 * c-common.c (flag_lax_vector_conversions): New.
1684 (vector_types_convertible_p): Unless -flax-vector conversions
1685 has been passed, disallow conversions between vectors with
1686 differing numbers of subparts and/or element types. If such
1687 a conversion is disallowed, possibly emit a note on the first
1688 occasion only to inform the user of -flax-vector-conversions.
1689 The new last argument specifies this.
1690 * c-common.h (flag_lax_vector_conversions): New.
1691 (vector_types_convertible_p): Add extra argument.
1692 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use
1693 char_type_node for V*QI type vectors.
1694 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
1695 Update to satisfy new typechecking rules.
1696 * config/rs6000/altivec.h (vec_cmple): Use vec_cmpge, for both
1698 * doc/invoke.texi (C Dialect Options): Document
1699 -flax-vector-conversions.
1701 2007-01-08 Mark Shinwell <shinwell@codesourcery.com>
1703 PR tree-optimization/29877
1704 * tree-ssa-ter.c (is_replaceable_p): Deem assignments with
1705 a register variable on the RHS to not be replaceable.
1707 2007-01-08 Chen Liqin <liqin@sunnorth.com.cn>
1708 * config/score/t-score-elf (MULTILIB_OPTIONS): Change.
1709 * config/score/predicates.md (const_uimm5, sr0_operand, const_simm12,
1710 const_simm15, const_pow2, const_npow2): Added.
1711 * config/score/misc.md (insv, extv, extzv, movmemsi,
1712 move_lbu_a/b, mov_lhu_a/b etc): Added and fix some bug.
1713 * config/score/score.c (score_address_cost, score_select_cc_mode):
1715 Change CONST_OK_FOR_LETTER_P/EXTRA_CONSTRAINT define.
1716 Update score_rtx_costs for MACRO TARGET_RTX_COSTS.
1717 Update score_print_operand.
1718 * config/score/score.h (DATA_ALIGNMENT, SELECT_CC_MODE): Added.
1719 Adjust register allocate order and update some macro define.
1720 * config/score/score-mdaux.c (mdx_unaligned_load, mdx_unsigned_store,
1721 mdx_block_move_straight, mdx_block_move_loop_head,
1722 mdx_block_move_loop_body, mdx_block_move_loop_foot, mdx_block_move_loop,
1723 mdx_block_move): Added.
1724 (mdx_movsicc, mdp_select_add_imm, mdp_select, mds_zero_extract_andi,
1725 mdp_limm): Updated and fix some bug and typo.
1726 * config/score/score.md (movqi/hi/si, add/sub/zero/ext): Updated.
1727 (movsf, movdf, doloop_end): Added.
1729 2007-01-08 Kazu Hirata <kazu@codesourcery.com>
1731 * config/arm/arm.c, config/arm/arm.h, config/arm/arm.md,
1732 config/arm/thumb2.md: Fix comment typos.
1733 * doc/extend.texi: Fix a typo.
1735 2007-01-07 Eric Christopher <echristo@apple.com>
1737 * configure.ac: Check for __stack_chk_fail for darwin.
1738 * configure: Regenerate.
1740 2007-01-07 Richard Guenther <rguenther@suse.de>
1742 * tree-vrp.c (extract_range_from_assert): CSE calls to
1743 compare_values where possible.
1744 (extract_range_from_unary_expr): Likewise.
1746 2007-01-07 Anatoly Sokolov <aesok@post.ru>
1748 * config/avr/avr-protos.h (call_insn_operand): Delete prototype.
1749 * config/avr/avr.c (call_insn_operand): Delete function.
1750 * config/avr/avr.md (*pushqi, *pushhi, *pushsi, *pushsf): Use REG_SP
1751 instead of register number. Use predicates.
1752 * config/avr/predicates.md (const0_operand, reg_or_0_operand,
1753 call_insn_operand): Add.
1755 2007-01-06 Jan Hubicka <jh@suse.cz>
1757 * tree-pass.h (pass_build_cgraph_edges): Declare.
1758 * cgraphunit.c (record_refernece): Move to cgraphbuild.c
1759 (visited_nodes): Remove.
1760 (cgraph_create_edges): Move to cgraphbuild.c; rename to
1761 build_cgrpah_edges; make visited_nodes local.
1762 (cgraph_process_new_functions): DO not call initialize_inline_failed.
1763 (record_references_in_initializer): Move to cgraphbuild.c
1764 (initialize_inline_failed, rebuild_cgraph_edges,
1765 pass_rebuild_cgraph_edges): Move to cgraphbuild.c.
1766 (verify_cgraph_node): Make visited_nodes local.
1767 (cgraph_analyze_function): Do not call cgraph_create_edges and
1768 initialize_inline_failed.
1769 (cgraph_expand_function): Do not call cgraph_lower_function;
1770 assert that function is already lowered.
1771 * Makefile.in (cgraphbuild.o): New.
1772 * passes.c (init_optimization_passes): Add pass_build_cgraph_edges
1773 at the end of lowering passes.
1775 2007-01-06 Steven Bosscher <steven@gcc.gnu.org>
1777 * ifcvt.c (cond_move_convert_if_block): New function, code
1778 factored out from...
1779 (cond_move_process_if_block): ...here. Call the new function
1780 on the THEN and ELSE blocks.
1781 (merge_if_block): Do not copy global_live_at_end, merge_blocks
1782 already takes care of this.
1784 2007-01-05 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
1787 * tree.h (TREE_OVERFLOW_P): New.
1788 * c-typeck.c (parser_build_unary_op): Warn only if result
1789 overflowed and operands did not.
1790 (parser_build_binary_op): Likewise.
1791 (convert_for_assignment): Remove redundant overflow_warning.
1792 * c-common.c (overflow_warning): Don't check or set TREE_OVERFLOW.
1794 2007-01-05 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
1796 * c-typeck.c (store_init_value): Split over two lines to follow
1797 the GNU coding style.
1799 2007-01-05 Benjamin Kosnik <bkoz@redhat.com>
1801 * c-cppbuiltin.c (c_cpp_builtins): __GXX_EXPERIMENTAL_CPP0X__ to
1802 __GXX_EXPERIMENTAL_CXX0X__.
1803 * doc/cpp.texi: Same.
1805 2007-01-05 Richard Guenther <rguenther@suse.de>
1808 * tree.c (get_narrower): Do not construct COMPONENT_REFs
1809 with mismatched types. Instead explicitly build a
1810 conversion NOP_EXPR.
1812 2007-01-05 Ian Lance Taylor <iant@google.com>
1814 * c-common.c (decl_with_nonnull_addr_p): New function.
1815 (c_common_truthvalue_conversion): Call it.
1816 * c-typeck.c (build_binary_op): Likewise.
1817 * c-common.h (decl_with_nonnull_addr_p): Declare.
1819 2007-01-05 Jakub Jelinek <jakub@redhat.com>
1822 * libgcc2.c (__divdc3): Compare c and d against 0.0 instead of
1825 2007-01-05 Joel Brobecker <brobecker@adacore.com>
1827 * doc/install.texi (Final install): Document the fact that
1828 the GNAT runtime should not be stripped.
1830 2007-01-04 Jan Hubicka <jh@suse.cz>
1832 * tree-inline.c (fold_marked_statements): Update operand caches
1833 and EH after folding
1835 2007-01-04 Ian Lance Taylor <iant@google.com>
1837 * c-common.c (check_function_nonnull): Whitespace fix.
1839 2007-01-04 Jan Hubicka <jh@suse.cz>
1841 * tree-optimize.c (execute_fixup_cfg): Correct previously mistakely
1842 comitted older version of patch.
1843 (pass_fixup_cfg): Add TODOs to verify flow and statements, dump
1844 function, celanup cfg and collect garbage.
1846 2007-01-04 Mike Stump <mrs@apple.com>
1848 * Makefile.in (mostlyclean): Don't remove libgcc anymore.
1851 2007-01-04 Eric Christopher <echristo@apple.com>
1853 * libgcc2.c (__bswapsi2): Use SItype.
1854 (__bswapdi2): Use DItype.
1855 * libgcc2.h: Update for above.
1857 2007-01-04 Paul Brook <paul@codesourcery.com>
1859 * config/arm/arm.md (arm_mulsi3, thumb_mulsi3, mulsi3_compare0,
1860 mulsi_compare0_scratch, mulsi3addsi, mulsi3addsi_compare0,
1861 mulsi3addsi_compare0_scratch, mulsidi3adddi, mulsidi3,
1862 umulsidi3, umulsidi3adddi, smulsi3_highpart,
1863 umulsi3_highpart): Make conditional on !arm_arch6.
1864 (arm_mulsi3_v6, thumb_mulsi3_v6, mulsi3_compare0_v6,
1865 mulsi_compare0_scratch_v6, mulsi3addsi_v6, mulsi3addsi_compare0_v6,
1866 mulsi3addsi_compare0_scratch_v6, mulsidi3adddi_v6, mulsidi3_v6,
1867 umulsidi3_v6, umulsidi3adddi_v6, smulsi3_highpart_v6,
1868 umulsi3_highpart_v6): New insns.
1870 2007-01-04 Roger Sayle <roger@eyesopen.com>
1872 * fold-const.c (fold_convert): When casting an expression to void,
1873 fold_ignored_result may discover a GIMPLE_MODIFY_STMT which doesn't
1874 have a type. Instead of attempting to build a NOP_EXPR, return
1875 these "special" trees directly.
1877 2007-01-04 Joseph Myers <joseph@codesourcery.com>
1879 * config/rs6000/rs6000.c (rs6000_rtx_costs): Make adjustment for
1880 MULT inside MINUS as either argument. Use rs6000_cost->dmul -
1881 rs6000_cost->fp not 0 as adjustment for outer NEG.
1883 2007-01-04 Jan Hubicka <jh@suse.cz>
1885 * cgraph.c (cgraph_release_function_body): New function.
1886 (cgraph_remove_node): Use it.
1887 * cgraph.h (cgraph_release_function_body): Declare.
1888 * cgraphunit.c (cgraph_expand_function): Use it.
1889 * ipa.c (cgraph_remove_unreahchable_nodes): Use it.
1890 * tree-ssa.c (delete_tree_ssa): Allow to be called before aliasing
1891 is initialized and while compilation of other function is running.
1892 * tree-optimize.c (execute_free_cfg_annotations): Move code to clear
1893 statement CFG annotations from here to ...
1894 * tree-cfg.c (delete_tree_cfg_annotations): ... here.
1896 2007-01-04 Zdenek Dvorak <dvorakz@suse.cz>
1898 * cfgloop.h (enum li_flags): Make the constants powers of two.
1900 2007-01-04 Jan Hubicka <jh@suse.cz>
1902 * tree-inline.c (copy_bb): Insert new statements to statements_to_fold
1904 (fold_marked_statements): New function.
1905 (optimize_inline_calls, tree_function_versioning): Fold new statements.
1906 * tree-inline.h (copy_body_data): Add statements_to_fold.
1908 2007-01-03 Daniel Jacobowitz <dan@codesourcery.com>
1910 * config.gcc: Mention libgcc/config.host.
1911 * Makefile.in: Update comments mentioning libgcc.
1912 (LIBGCC, INSTALL_LIBGCC, GCC_PARTS, mklibgcc): Delete.
1913 (all.cross, start.encap, rest.encap, rest.cross): Update
1914 dependencies for libgcc move.
1915 (libgcc.mk, LIBGCC_DEPS, libgcov.a, libgcc.a, stmp-multilib)
1916 (clean-target, clean-target-libgcc): Delete.
1917 (srcdirify, GCC_EXTRA_PARTS): New macros.
1918 (libgcc-support, libgcc.mvars): New rules.
1919 (distclean): Remove mention of mklibgcc.
1920 (install): Don't reference INSTALL_LIBGCC.
1921 (install-common): Don't reference EXTRA_PARTS.
1922 (install-libgcc, install-multilib): Delete rules.
1923 * mklibgcc.in: Delete file.
1924 * doc/configfiles.texi: Don't mention mklibgcc.
1926 * config/i386/t-darwin (SHLIB_VERPFX): Delete (moved to libgcc).
1927 * config/i386/t-darwin64 (SHLIB_VERPFX): Likewise.
1928 * config/rs6000/t-darwin (SHLIB_VERPFX): Likewise.
1929 * config/rs6000/t-ppccomm (TARGET_LIBGCC2_CFLAGS, SHLIB_MAPFILES)
1930 (mklibgcc, ldblspecs): Likewise.
1932 * config/i386/t-nwld (libgcc.def, libc.def, libpcre.def)
1933 (posixpre.def): Use $(T).
1934 (SHLIB_EXT, SHLIB_NAME, SHLIB_SLIBDIR_QUAL, SHLIB_DEF, SHLIB_MAP)
1935 (SHLIB_SRC, SHLIB_INSTALL): Delete.
1936 (SHLIB_LINK): Make dummy.
1937 * config/t-slibgcc-darwin: Delete contents except for dummy SHLIB_LINK.
1939 * config/frv/t-linux (EXTRA_MULTILIB_PARTS): Clear.
1941 * config/alpha/t-crtfm: Use $(T) in rules for EXTRA_PARTS.
1942 * config/alpha/t-vms, config/alpha/t-vms64, config/fr30/t-fr30,
1943 config/i386/t-rtems-i386, config/ia64/t-ia64, config/rs6000/t-beos,
1944 config/rs6000/t-newas, config/sparc/t-elf: Likewise.
1946 * configure.ac (all_outputs): Remove mklibgcc.
1947 * configure: Regenerated.
1949 2007-01-03 Josh Conner <jconner@apple.com>
1952 * calls.c (compute_argument_addresses): Set stack and stack_slot
1953 for partial args, too.
1954 (store_one_arg): Use locate.size.constant for the size when
1955 generating a save_area.
1957 2007-01-03 Robert Kennedy <jimbob@google.com>
1959 * tree-cfg.c (tree_merge_blocks): Release SSA_NAME phi results
1960 whose definitions are deleted due to basic block merging.
1962 2007-01-03 Paul Brook <paul@codesourcery.com>
1965 * config/arm/arm.c (output_return_instruction): Pop PC in interrupt
1967 (use_return_insn): Return 0 for Thumb interrupt functions.
1968 (print_multi_reg): Add rfe argument for IRQ returns.
1969 (arm_output_epilogue): Pop interrupt return address directly into PC.
1970 (arm_expand_prologue): Only adjust IRQ return address in Arm mode.
1972 2007-01-03 Paul Brook <paul@codesourcery.com>
1974 Merge from sourcerygxx-4_1.
1975 * config/arm/thumb2.md: New file.
1976 * config/arm/elf.h (JUMP_TABLES_IN_TEXT_SECTION): Return True for
1978 * config/arm/coff.h (JUMP_TABLES_IN_TEXT_SECTION): Ditto.
1979 * config/arm/aout.h (ASM_OUTPUT_ADDR_VEC_ELT): Add !Thumb-2 assertion.
1980 (ASM_OUTPUT_ADDR_DIFF_ELT): Output Thumb-2 jump tables.
1981 * config/arm/aof.h (ASM_OUTPUT_ADDR_DIFF_ELT): Output Thumb-2 jump
1983 (ASM_OUTPUT_ADDR_VEC_ELT): Add !Thumb-2 assertion.
1984 * config/arm/ieee754-df.S: Use macros for Thumb-2/Unified asm
1986 * config/arm/ieee754-sf.S: Ditto.
1987 * config/arm/arm.c (thumb_base_register_rtx_p): Rename...
1988 (thumb1_base_register_rtx_p): ... to this.
1989 (thumb_index_register_rtx_p): Rename...
1990 (thumb1_index_register_rtx_p): ... to this.
1991 (thumb_output_function_prologue): Rename...
1992 (thumb1_output_function_prologue): ... to this.
1993 (thumb_legitimate_address_p): Rename...
1994 (thumb1_legitimate_address_p): ... to this.
1995 (thumb_rtx_costs): Rename...
1996 (thumb1_rtx_costs): ... to this.
1997 (thumb_compute_save_reg_mask): Rename...
1998 (thumb1_compute_save_reg_mask): ... to this.
1999 (thumb_final_prescan_insn): Rename...
2000 (thumb1_final_prescan_insn): ... to this.
2001 (thumb_expand_epilogue): Rename...
2002 (thumb1_expand_epilogue): ... to this.
2003 (arm_unwind_emit_stm): Rename...
2004 (arm_unwind_emit_sequence): ... to this.
2005 (thumb2_legitimate_index_p, thumb2_legitimate_address_p,
2006 thumb1_compute_save_reg_mask, arm_dwarf_handle_frame_unspec,
2007 thumb2_index_mul_operand, output_move_vfp, arm_shift_nmem,
2008 arm_save_coproc_regs, thumb_set_frame_pointer, arm_print_condition,
2009 thumb2_final_prescan_insn, thumb2_asm_output_opcode, arm_output_shift,
2010 thumb2_output_casesi): New functions.
2011 (TARGET_DWARF_HANDLE_FRAME_UNSPEC): Define.
2012 (FL_THUMB2, FL_NOTM, FL_DIV, FL_FOR_ARCH6T2, FL_FOR_ARCH7,
2013 FL_FOR_ARCH7A, FL_FOR_ARCH7R, FL_FOR_ARCH7M, ARM_LSL_NAME,
2014 THUMB2_WORK_REGS): Define.
2015 (arm_arch_notm, arm_arch_thumb2, arm_arch_hwdiv, arm_condexec_count,
2016 arm_condexec_mask, arm_condexec_masklen)): New variables.
2017 (all_architectures): Add armv6t2, armv7, armv7a, armv7r and armv7m.
2018 (arm_override_options): Check new CPU capabilities.
2019 Set new architecture flag variables.
2020 (arm_isr_value): Handle v7m interrupt functions.
2021 (user_return_insn): Return 0 for v7m interrupt functions. Handle
2023 (const_ok_for_arm): Handle Thumb-2 constants.
2024 (arm_gen_constant): Ditto. Use movw when available.
2025 (arm_function_ok_for_sibcall): Return false for v7m interrupt
2027 (legitimize_pic_address, arm_call_tls_get_addr): Handle Thumb-2.
2028 (thumb_find_work_register, arm_load_pic_register,
2029 legitimize_tls_address, arm_address_cost, load_multiple_sequence,
2030 emit_ldm_seq, emit_stm_seq, arm_select_cc_mode, get_jump_table_size,
2031 print_multi_reg, output_mov_long_double_fpa_from_arm,
2032 output_mov_long_double_arm_from_fpa, output_mov_double_fpa_from_arm,
2033 output_mov_double_fpa_from_arm, output_move_double,
2034 arm_compute_save_reg_mask, arm_compute_save_reg0_reg12_mask,
2035 output_return_instruction, arm_output_function_prologue,
2036 arm_output_epilogue, arm_get_frame_offsets, arm_regno_class,
2037 arm_output_mi_thunk, thumb_set_return_address): Ditto.
2038 (arm_expand_prologue): Handle Thumb-2. Use arm_save_coproc_regs.
2039 (arm_coproc_mem_operand): Allow POST_INC/PRE_DEC.
2040 (arithmetic_instr, shift_op): Use arm_shift_nmem.
2041 (arm_print_operand): Use arm_print_condition. Handle '(', ')', '.',
2043 (arm_final_prescan_insn): Use extract_constrain_insn_cached.
2044 (thumb_expand_prologue): Use thumb_set_frame_pointer.
2045 (arm_file_start): Output directive for unified syntax.
2046 (arm_unwind_emit_set): Handle stack alignment instruction.
2047 * config/arm/lib1funcs.asm: Remove default for __ARM_ARCH__.
2048 Add v6t2, v7, v7a, v7r and v7m.
2049 (RETLDM): Add Thumb-2 code.
2050 (do_it, shift1, do_push, do_pop, COND, THUMB_SYNTAX): New macros.
2051 * config/arm/arm.h (TARGET_CPU_CPP_BUILTINS): Define __thumb2__.
2052 (TARGET_THUMB1, TARGET_32BIT, TARGET_THUMB2, TARGET_DSP_MULTIPLY,
2053 TARGET_INT_SIMD, TARGET_UNIFIED_ASM, ARM_FT_STACKALIGN, IS_STACKALIGN,
2054 THUMB2_TRAMPOLINE_TEMPLATE, TRAMPOLINE_ADJUST_ADDRESS,
2055 ASM_OUTPUT_OPCODE, THUMB2_GO_IF_LEGITIMATE_ADDRESS,
2056 THUMB2_LEGITIMIZE_ADDRESS, CASE_VECTOR_PC_RELATIVE,
2057 CASE_VECTOR_SHORTEN_MODE, ADDR_VEC_ALIGN, ASM_OUTPUT_CASE_END,
2058 ADJUST_INSN_LENGTH): Define.
2059 (TARGET_REALLY_IWMMXT, TARGET_IWMMXT_ABI, CONDITIONAL_REGISTER_USAGE,
2060 STATIC_CHAIN_REGNUM, HARD_REGNO_NREGS, INDEX_REG_CLASS,
2061 BASE_REG_CLASS, MODE_BASE_REG_CLASS, SMALL_REGISTER_CLASSES,
2062 PREFERRED_RELOAD_CLASS, SECONDARY_OUTPUT_RELOAD_CLASS,
2063 SECONDARY_INPUT_RELOAD_CLASS, LIBCALL_VALUE, FUNCTION_VALUE_REGNO_P,
2064 TRAMPOLINE_SIZE, INITIALIZE_TRAMPOLINE, HAVE_PRE_INCREMENT,
2065 HAVE_POST_DECREMENT, HAVE_PRE_DECREMENT, HAVE_PRE_MODIFY_DISP,
2066 HAVE_POST_MODIFY_DISP, HAVE_PRE_MODIFY_REG, HAVE_POST_MODIFY_REG,
2067 REGNO_MODE_OK_FOR_BASE_P, LEGITIMATE_CONSTANT_P,
2068 REG_MODE_OK_FOR_BASE_P, REG_OK_FOR_INDEX_P, GO_IF_LEGITIMATE_ADDRESS,
2069 LEGITIMIZE_ADDRESS, THUMB2_LEGITIMIZE_ADDRESS,
2070 GO_IF_MODE_DEPENDENT_ADDRESS, MEMORY_MOVE_COST, BRANCH_COST,
2071 ASM_APP_OFF, ASM_OUTPUT_CASE_LABEL, ARM_DECLARE_FUNCTION_NAME,
2072 FINAL_PRESCAN_INSN, PRINT_OPERAND_PUNCT_VALID_P,
2073 PRINT_OPERAND_ADDRESS): Adjust for Thumb-2.
2074 (arm_arch_notm, arm_arch_thumb2, arm_arch_hwdiv): New declarations.
2075 * config/arm/arm-cores.def: Add arm1156t2-s, cortex-a8, cortex-r4 and
2077 * config/arm/arm-tune.md: Regenerate.
2078 * config/arm/arm-protos.h: Update prototypes.
2079 * config/arm/vfp.md: Enable patterns for Thumb-2.
2080 (arm_movsi_vfp): Add movw alternative. Use output_move_vfp.
2081 (arm_movdi_vfp, movsf_vfp, movdf_vfp): Use output_move_vfp.
2082 (thumb2_movsi_vfp, thumb2_movdi_vfp, thumb2_movsf_vfp,
2083 thumb2_movdf_vfp, thumb2_movsfcc_vfp, thumb2_movdfcc_vfp): New.
2084 * config/arm/libunwind.S: Add Thumb-2 code.
2085 * config/arm/constraints.md: Update include Thumb-2.
2086 * config/arm/ieee754-sf.S: Add Thumb-2/Unified asm support.
2087 * config/arm/ieee754-df.S: Ditto.
2088 * config/arm/bpabi.S: Ditto.
2089 * config/arm/t-arm (MD_INCLUDES): Add thumb2.md.
2090 * config/arm/predicates.md (low_register_operand,
2091 low_reg_or_int_operand, thumb_16bit_operator): New.
2092 (thumb_cmp_operand, thumb_cmpneg_operand): Rename...
2093 (thumb1_cmp_operand, thumb1_cmpneg_operand): ... to this.
2094 * config/arm/t-arm-elf: Add armv7 multilib.
2095 * config/arm/arm.md: Update patterns for Thumb-2 and Unified asm.
2097 (UNSPEC_STACK_ALIGN, ce_count): New.
2098 (arm_incscc, arm_decscc, arm_umaxsi3, arm_uminsi3,
2099 arm_zero_extendsidi2, arm_zero_extendqidi2): New
2101 * config/arm/fpa.md: Update patterns for Thumb-2 and Unified asm.
2102 (thumb2_movsf_fpa, thumb2_movdf_fpa, thumb2_movxf_fpa,
2103 thumb2_movsfcc_fpa, thumb2_movdfcc_fpa): New insns.
2104 * config/arm/cirrus.md: Update patterns for Thumb-2 and Unified asm.
2105 (cirrus_thumb2_movdi, cirrus_thumb2_movsi_insn,
2106 thumb2_cirrus_movsf_hard_insn, thumb2_cirrus_movdf_hard_insn): New
2108 * doc/extend.texi: Document ARMv7-M interrupt functions.
2109 * doc/invoke.texi: Document Thumb-2 new cores+architectures.
2111 2007-01-03 Jakub Jelinek <jakub@redhat.com>
2113 * unwind-dw2.c (SIGNAL_FRAME_BIT, EXTENDED_CONTEXT_BIT): Define.
2114 (struct _Unwind_Context): Rename args_size to flags, remove
2115 signal_frame field, add a new args_size field and version field.
2116 (_Unwind_IsSignalFrame, _Unwind_SetSignalFrame,
2117 _Unwind_IsExtendedContext): New inline functions.
2118 (_Unwind_GetGR, _Unwind_SetGR, _Unwind_GetGRPtr, _Unwind_SetGRPtr):
2119 Assume by_value array is only present if _Unwind_IsExtendedContext.
2120 (_Unwind_GetIPInfo, execute_cfa_program, uw_frame_state_for): Use
2121 _Unwind_IsSignalFrame.
2122 (__frame_state_for): Initialize context.flags to EXTENDED_CONTEXT_BIT.
2123 (uw_update_context_1): Use _Unwind_SetSignalFrame.
2124 (uw_init_context_1): Initialize context->flags to
2125 EXTENDED_CONTEXT_BIT.
2126 * config/rs6000/linux-unwind.h (frob_update_context): Use
2127 _Unwind_SetSignalFrame.
2129 2007-01-03 Andrew Pinski <andrew_pinski@playstation.sony.com>
2132 * gimplify.c (gimplify_modify_expr_complex_part): Move below
2133 tree_to_gimple_tuple. Call tree_to_gimple_tuple when we need
2136 2007-01-03 Kazu Hirata <kazu@codesourcery.com>
2138 * config/i386/i386.h (NON_STACK_REG_P, REGNO_OK_FOR_SIREG_P,
2139 REGNO_OK_FOR_DIREG_P, REWRITE_ADDRESS, ASM_OPERAND_LETTER,
2140 RET, AT_SP): Remove.
2141 * config/i386/i386.md (*sse_prologue_save_insn): Use return
2144 * alias.c (init_alias_analysis): Use VEC_safe_grow_cleared.
2145 * cfgbuild.c (find_basic_blocks): Likewise.
2146 * cfgrtl.c (rtl_create_basic_block): Likewise.
2147 * function.c (temp_slots_at_level): Likewise.
2148 * reg-stack.c (stack_regs_mentioned): Likewise.
2149 * regclass.c (allocate_reg_info): Likewise.
2150 * tree-cfg.c (init_empty_tree_cfg, build_tree_cfg, create_bb,
2151 set_bb_for_stmt, move_block_to_fn): Likewise.
2152 * tree-complex.c (tree_lower_complex): Likewise.
2153 * vec.h (VEC_safe_grow_cleared): New.
2155 * cgraphunit.c, tree-ssa-alias.c: Fix comment typos.
2157 2007-01-03 Zdenek Dvorak <dvorakz@suse.cz>
2159 * loop-unswitch.c (unswitch_loop): Pass probabilities to loopify.
2160 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Pass probabilities
2162 * cfgloopmanip.c (scale_loop_frequencies): Export.
2163 (loopify): Scale the frequencies by prescribed coefficients.
2164 (set_zero_probability): New function.
2165 (duplicate_loop_to_header_edge): Improve updating of frequencies.
2166 (lv_adjust_loop_entry_edge, loop_version): Set probabilities
2167 and frequencies according to arguments.
2168 * tree-ssa-loop-manip.c (tree_unroll_loop): Set probabilities
2170 * cfg.c (scale_bbs_frequencies_int): Allow scaling the frequencies up.
2171 * modulo-sched.c (sms_schedule): Set probabilities for entering
2172 versioned loop correctly.
2173 * tree-vect-transform.c (vect_transform_loop): Ditto.
2174 * cfgloop.h (loopify, loop_version): Declaration changed.
2175 (scale_loop_frequencies): Declared.
2177 2007-01-02 Jan Hubicka <jh@suse.cz>
2179 * cgraph.c: Include tree-flow.h
2180 (cgraph_add_new-function): Handle IPA_SSA mode; execute
2182 * cgraph.h (enum cgraph_state): Add CGRAPH_STATE_IPA_SSA.
2183 * tree-pass.h (pass_all_early_optimizations): Declare.
2184 * cgraphunit.c (cgraph_process_new_functions): Add IPA_SSA; execute
2186 (cgraph_analyze_function): Do early_local_passes.
2187 * tree-mudflap.c (mf_decl_cache_locals, mf_build_check_statement_for):
2188 Do not add referenced vars.
2189 * tree-optimize.c (gate_all_optimizations): Do not execute when not in
2191 (gate_all_early_local_passes): New gate.
2192 (pass_early_local_passes): Use new gate.
2193 (execute_early_local_optimizations): New functions.
2194 (gate_all_early_optimizations): New gate.
2195 (pass_all_early_optimizations): New pass.
2196 (execute_free_datastructures): Free SSA only when initialized.
2197 (gate_init_datastructures): Init only when optimizing.
2198 (tree_lowering_passes): Do early local passes when called late.
2199 * tree-profile.c (do_tree_profiling): Don't profile functions added
2201 (do_early_tree_profiling, pass_early_tree_profile): Kill.
2202 * tree-cfg.c (update_modified_stmts): Do not update when operands are
2204 * passes.c (init_optimizations_passes): Reorder so we go into SSA
2205 during early_local_passes.
2206 * Makefile.in (cgraph.o): Add dependency on tree-flow.h.
2209 2007-01-02 Carlos O'Donell <carlos@codesourcery.com>
2211 * Makefile.in: Update copyright year.
2213 2007-01-02 Carlos O'Donell <carlos@codesourcery.com>
2215 * Makefile.in: Export GCC_EXEC_PREFIX before calling $(RUNTEST)
2216 in $(lang_checks) and check-consistency targets.
2218 2007-01-02 Jan Hubicka <jh@suse.cz>
2220 * tree-mudflap.c (mf_decl_cache_locals, mf_build_check_statement_for):
2221 Do not add referenced vars.
2222 * tree-cfg.c (update_modified_stmts): Do not update when SSA operands
2224 * passes.c (init_optimization_passes): Put mudflap_2 after
2225 free_datastructures.
2227 2007-01-02 Jan Hubicka <jh@suse.cz>
2229 * tree-optimize (execute_fixup_cfg): Set after_inlining flag.
2230 Set NOTHROW flag on call statements proved to be nothrow.
2231 Update statement of local calls so new pure/const functions are
2232 updated. Update_ssa when in ssa form. Mark PHI nodes of nonlocal
2234 (tree_rest_of_compilation): Register hooks and initialize bitmap
2235 early. Do not set after_inlining flag.
2237 2007-01-02 Steve Ellcey <sje@cup.hp.com>
2239 * sbitmap.c (HOST_BITS_PER_LONG_LONG): Change to
2240 HOST_BITS_PER_LONGLONG
2242 2007-01-02 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
2245 * c-typeck.c (store_init_value): Don't emit pedantic overflow
2246 warning for non-static initializers.
2248 2007-01-02 Steven Bosscher <steven@gcc.gnu.org>
2250 * config/alpha/alpha.md, arm/arm.c, darwin.c, frv/frv.md,
2251 m32r/m32r.c, m32r/m32r.c, mn10300/mn10300.md, pa/pa.c,
2252 rs6000/rs6000.c, s390/s390.md, sh/sh.md, sparc/sparc.c:
2253 Always use set_unique_reg_note to add REG_EQUAL notes.
2255 2007-01-02 Kazu Hirata <kazu@codesourcery.com>
2258 2007-01-02 Kazu Hirata <kazu@codesourcery.com>
2260 * alias.c (init_alias_analysis): Use VEC_safe_grow_cleared.
2261 * cfgbuild.c (find_basic_blocks): Likewise.
2262 * cfgrtl.c (rtl_create_basic_block): Likewise.
2263 * function.c (temp_slots_at_level): Likewise.
2264 * reg-stack.c (stack_regs_mentioned): Likewise.
2265 * regclass.c (allocate_reg_info): Likewise.
2266 * tree-cfg.c (init_empty_tree_cfg, build_tree_cfg, create_bb,
2267 set_bb_for_stmt, move_block_to_fn): Likewise.
2268 * tree-complex.c (tree_lower_complex): Likewise.
2269 * vec.h (VEC_safe_grow_cleared): New.
2271 2007-01-02 Ian Lance Taylor <iant@google.com>
2273 * c-common.c (c_common_truthvalue_conversion): When warning about
2274 using an assignment as a truth value, set TREE_NO_WARNING.
2276 2007-01-02 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
2279 * c.opt (Wold-style-declaration): New.
2280 * doc/invoke.texi (C-only Warning Options): New.
2281 (Wold-style-declaration): Document it.
2282 (Wextra): Enabled by -Wextra.
2283 * c-opts.c (c_common_post_options): Enabled by -Wextra.
2284 * c-decl.c (declspecs_add_scspec): Replace -Wextra with
2285 -Wold-style-declaration.
2287 2007-01-02 Kazu Hirata <kazu@codesourcery.com>
2289 * alias.c (init_alias_analysis): Use VEC_safe_grow_cleared.
2290 * cfgbuild.c (find_basic_blocks): Likewise.
2291 * cfgrtl.c (rtl_create_basic_block): Likewise.
2292 * function.c (temp_slots_at_level): Likewise.
2293 * reg-stack.c (stack_regs_mentioned): Likewise.
2294 * regclass.c (allocate_reg_info): Likewise.
2295 * tree-cfg.c (init_empty_tree_cfg, build_tree_cfg, create_bb,
2296 set_bb_for_stmt, move_block_to_fn): Likewise.
2297 * tree-complex.c (tree_lower_complex): Likewise.
2298 * vec.h (VEC_safe_grow_cleared): New.
2300 2007-01-02 Douglas Gregor <doug.gregor@gmail.com>
2302 * c-common.c(c_common_nodes_and_builtins): Since variants of
2303 void_type_node get built before it is given a name, we need to
2304 give those variants the name, too.
2305 (complete_array_type): We need to work with the canonical main
2306 type of the array, from which we will build the qualified version.
2307 * params.def (PARAM_VERIFY_CANONICAL_TYPES): New.
2308 * print-tree.c (print_node): Display canonical type information
2310 * stor-layout.c (layout_type): When we don't know the
2311 alignment of a type for which we're building an array, we end up
2312 guessing wrong, so make the type require structural equality.
2313 * tree.c (make_node_stat): When we build a new type, it is its
2315 (build_type_attribute_qual_variant): When building an attribute
2316 variant, its canonical type is the non-attribute variant. However,
2317 if the attributes are target-dependent and they differ, we need to
2318 use structural equality checks for this type.
2319 (build_qualified_type): A qualified type is not equivalent to its
2320 unqualified variant; set the canonical type appropriately.
2321 (build_distinct_type_copy): When building a distinct type from
2322 another type, the new type is its own canonical type.
2323 (build_variant_type_copy): When building a new type variant, we
2324 assume that it is equivalent to the original type.
2325 (build_pointer_type_for_mode): When building a pointer type, also
2326 build a canonical type pointer.
2327 (build_reference_type_for_mode): When building a reference type,
2328 also build a canonical type reference.
2329 (build_index_type): When we can't hash an index type (e.g.,
2330 because its maximum value is negative), the index type requires
2331 structural equality tests.
2332 (build_array_type): Build the canonical form of an array type.
2333 (build_function_type): Function types require structural equality,
2334 because they contain default arguments, attributes, etc.
2335 (build_method_type_directly): Ditto for method types.
2336 (build_offset_type): Build the canonical offset type.
2337 (build_complex_type): Build the canonical vector type.
2338 (make_vector_type): Build the canonical vector type.
2339 * tree.h (TYPE_CANONICAL): New.
2340 (TYPE_STRUCTURAL_EQUALITY_P): New.
2341 (SET_TYPE_STRUCTURAL_EQUALITY): New.
2342 (struct tree_type): Added "canonical" field.
2343 * params.h (VERIFY_CANONICAL_TYPES): New.
2344 * doc/c-tree.texi (TYPE_CANONICAL): Document.
2345 (TYPE_STRUCTURAL_EQUALITY_P): Document.
2346 (SET_TYPE_STRUCTURAL_EQUALITY): Document.
2347 * doc/invoke.texi (verify-canonical-types): Document --param
2348 parameter for verifying canonical types.
2350 2007-01-02 Joseph Myers <joseph@codesourcery.com>
2352 * config.gcc (powerpc-*-eabispe*, powerpc-*-eabisimaltivec*,
2353 powerpc-*-eabisim*, powerpc-*-eabialtivec*, powerpc-*-eabi*,
2354 powerpc-*-rtems*, powerpc-wrs-vxworks, powerpc-wrs-vxworksae,
2355 powerpcle-*-eabisim*, powerpcle-*-eabi*): Add rs6000/e500.h to
2357 * config/rs6000/e500.h: New.
2358 * config/rs6000/eabi.h (TARGET_SPE_ABI, TARGET_SPE, TARGET_E500,
2359 TARGET_ISEL, TARGET_FPRS, TARGET_E500_SINGLE, TARGET_E500_DOUBLE):
2361 * config/rs6000/linuxspe.h (TARGET_SPE_ABI, TARGET_SPE,
2362 TARGET_E500, TARGET_ISEL, TARGET_FPRS, TARGET_E500_SINGLE,
2363 TARGET_E500_DOUBLE): Remove.
2364 * config/rs6000/vxworks.h (TARGET_SPE_ABI, TARGET_SPE,
2365 TARGET_E500, TARGET_ISEL, TARGET_FPRS): Remove.
2366 * config/rs6000/rs6000.h (CHECK_E500_OPTIONS): Define.
2367 * config/rs6000/rs6000.c (rs6000_override_options): Use
2370 2007-01-02 Joseph Myers <joseph@codesourcery.com>
2372 * config/rs6000/rs6000.c (print_operand): Check (TARGET_SPE ||
2373 TARGET_E500_DOUBLE), not TARGET_E500, for %y.
2374 (rs6000_generate_compare, rs6000_emit_sCOND, output_cbranch,
2375 rs6000_emit_cmove): Don't check TARGET_E500.
2376 * config/rs6000/rs6000.md (bunordered, bordered, sunordered,
2377 sordered): Don't check TARGET_E500.
2379 2007-01-01 Eric Christopher <echristo@apple.com>
2381 * config/mips/mips.c (mips_regno_mode_ok_for_base_p): Use
2382 HARD_REGISTER_NUM_P.
2384 2007-01-01 Roger Sayle <roger@eyesopen.com>
2386 * fold-const.c (fold_binary) <EQ_EXPR>: Fold "(X^C1) eq/ne C2" into
2387 "X eq/ne (C1^C2)". Fold "(X^Z) eq/ne (Y^Z)" as "X eq/ne Y" when Z
2388 has no side-effects. Fold "(X^C1) eq/ne (Y^C2)" as "(X^(C1^C2))
2391 2007-01-01 Mike Stump <mrs@apple.com>
2393 * configure.ac: Remove support for building with Apple's gcc-3.1.
2395 2007-01-02 Joseph Myers <joseph@codesourcery.com>
2398 * caller-save.c (add_stored_regs): Only handle SUBREGs if inner
2399 REG is a hard register. Do not modify REG before calling
2401 * rtlanal.c (subreg_get_info): Don't assert size of XMODE is a
2402 multiple of the size of YMODE for certain lowpart cases.
2404 2007-01-01 Andrew Pinski <pinskia@gmail.com>
2407 * gimplify (voidify_wrapper_expr): Update for
2410 2007-01-01 Andreas Schwab <schwab@suse.de>
2413 * config/ia64/ia64.c (ia64_compute_frame_size): Account space for
2414 save of BR0 in extra_spill_size instead of spill_size.
2415 (ia64_expand_prologue): Save BR0 outside of the gr/br/fr spill
2417 (ia64_expand_epilogue): Restore BR0 from its new location.
2419 2007-01-01 Andrew Pinski <pinskia@gmail.com>
2421 * gimplify.c (gimplify_init_constructor <case VECTOR_TYPE>):
2422 Use a temporary variable if the left hand side is not a gimple
2425 2007-01-01 Andrew Pinski <pinskia@gmail.com>
2427 * gimplify.c (gimplify_return_expr): Make the temporary variable
2428 for the return expression, a gimple register variable.
2430 2007-01-01 Jan Hubicka <jh@suse.cz>
2432 * emit-rtl.c (emit_copy_of_insn_after): Do not call copy_insn_1 for
2435 2007-01-01 Mike Stump <mrs@apple.com>
2437 * configure.ac (HAVE_GAS_LITERAL16): Add autoconf check for
2439 * config/darwin.c (machopic_select_rtx_section): Use
2441 (darwin_mergeable_constant_section): Likewise.
2442 * configure: Regenerate.
2443 * config.in: Regenerate.
2445 2007-01-01 Jan Hubicka <jh@suse.cz>
2446 Andrew Pinski <pinskia@gmail.com>
2448 * cgraphunit.c (cgraph_optimize): Call cgraph_add_new_functions
2449 before starting IPA passes.