1 2007-01-19 Dirk Mueller <dmueller@suse.de>
3 * config/i386.h (CONDITIONAL_REGISTER_USAGE): Store
4 result of PIC_OFFSET_TABLE_REGNUM in temporary variable to avoid
7 2007-01-19 Uros Bizjak <ubizjak@gmail.com>
9 * config/i386/i386.md (acos<mode>2): Rename from acossf2 and acosdf2.
10 Macroize expander using X87MODEF12 mode macro. Extend operand 1
11 to XFMode, use acosxf2 and truncate result to requested mode.
12 Use SSE_FLOAT_MODE_P to disable patterns for SSE math.
13 (asin<mode>2): Similarly, with asin expanders.
14 (*fscalexf4_i387): Rename from *fscalexf4.
15 (expNcorexf3): New expander.
16 (expxf2, exp10xf2, exp2xf2): Use expNcorexf3 expander.
17 (exp<mode>2): Rename from expsf2 and expdf2. Macroize expander using
18 X87MODEF12 mode macro. Extend operand 1 to XFMode, use expxf2 and
19 truncate result to requested mode. Use SSE_FLOAT_MODE_P to disable
20 patterns for SSE math.
21 (exp10<mode>2): Similarly, with exp10 expanders.
22 (exp2<mode>2): Similarly, with exp2 expanders.
23 (expm1<mode>2): Similarly, with expm1 expanders.
24 (ldexp<mode>3): Similarly, with ldexp expanders.
25 (log<mode>2, log10<mode>2, log2<mode>2, log1p<mode>2, logb<mode>2):
26 Use gen_truncxf<mode>2_i387_noop to truncate result.
28 2007-01-19 Richard Sandiford <richard@codesourcery.com>
30 * config/m68k/m68k.h (PREFERRED_STACK_BOUNDARY): Define to 32
33 2007-01-19 Nathan Sidwell <nathan@codesourcery.com>
34 Richard Sandiford <richard@codesourcery.com>
36 * config/m68k/m68k.h (M68K_STATIC_CHAIN_REG_NAME): New macro.
37 (INITIALIZE_TRAMPOLINE): Use STATIC_CHAIN_REGNUM.
38 (__transfer_from_trampoline): Use M68K_STATIC_CHAIN_REG_NAME.
39 * config/m68k/m68kelf.h (STATIC_CHAIN_REGNUM): Override.
40 (M68K_STATIC_CHAIN_REG_NAME): Likewise.
41 * config/m68k/netbsd-elf.h (M68K_STATIC_CHAIN_REG_NAME): Likewise.
43 2007-01-19 Richard Sandiford <richard@codesourcery.com>
45 * config/m68k/m68k.md (adddi_dilshr32): Rename to...
46 (*adddi_dilshr32): ...this. Fix formatting. Remove commented-out
47 non-canonical pattern. Restrict to !TARGET_COLDFIRE.
48 (*adddi_dilshr32_cf): New pattern.
49 (adddi3, subdi3): Remove first alternatives.
51 2007-01-19 Richard Sandiford <richard@codesourcery.com>
53 * config/m68k/m68k.c (notice_update_cc): If an SFmode move is
54 implemented using move.l, do not use its cc result for floating-point
57 2007-01-19 Richard Sandiford <richard@codesourcery.com>
59 * config/m68k/m68k.h (EXTRA_CONSTRAINT): Stop the 'T' constraint
60 from accepting 's' constraints if flag_pic.
62 2007-01-19 Richard Sandiford <richard@codesourcery.com>
64 * config/m68k/m68k.md (bordered, bunordered, buneq, bunge, bungt)
65 (bunle, bunlt, bltgt, bordered_rev, bunordered_rev, buneq_rev)
66 (bunge_rev, bungt_rev, bunle_rev, bunlt_rev, bltgt_rev): Change
67 condition from TARGET_68881 to TARGET_HARD_FLOAT.
69 2007-01-19 Sandra Loosemore <sandra@codesourcery.com>
71 * longlong.h (count_leading_zeros, COUNT_LEADING_ZEROS_0): Add
72 ColdFire alternatives.
73 * config/m68k/m68k.h (CLZ_DEFINED_VALUE_AT_ZERO): New macro.
74 * config/m68k/m68k.md (clzsi2): Define for ColdFire
75 architectures that support ff1 instruction.
77 2007-01-19 Richard Sandiford <richard@codesourcery.com>
78 Julian Brown <julian@codesourcery.com>
80 * config/m68k/m68k.h (CONST_OK_FOR_LETTER_P): Add an 'R' case.
81 * config/m68k/m68k.md (*movsi_cfv4): Fold into...
82 (*movsi_cf): ...here. Remove unnecessary 'R' from 'Rg'.
85 2007-01-19 Richard Sandiford <richard@codesourcery.com>
87 * config/m68k/m68k-protos.h (valid_mov3q_const): Take a HOST_WIDE_INT
89 (output_move_const_into_data_reg, output_move_simode_const): Delete.
90 * config/m68k/m68k.c (const_method, const_int_cost): Take a
91 HOST_WIDE_INT instead of an rtx.
92 (m68k_rtx_costs): Update call accordingly.
93 (output_move_const_into_data_reg): Likewise. Fix formatting.
94 (valid_mov3q_const): Take a HOST_WIDE_INT instead of an rtx.
96 (output_move_simode_const): Update calls after above changes.
97 Rework to use automatic variables and predicates like MEM_P.
98 * config/m68k/m68k.md (pushexthisi_const): Update call to
101 2007-01-19 Dirk Mueller <dmueller@suse.de>
103 * tree-ssa-alias.c (perform_var_substitution): Fix typo
106 2007-01-19 Richard Guenther <rguenther@suse.de>
108 * builtins.c (expand_builtin_cexpi): Fall back to expanding
109 via cexp in case sincos is not available.
111 2007-01-19 Richard Guenther <rguenther@suse.de>
113 * doc/tm.texi (TARGET_HAS_SINCOS): Document new target macro.
114 * defaults.h (TARGET_HAS_SINCOS): Default to off.
115 * config/linux.h (TARGET_HAS_SINCOS): Set to on if we have glibc.
116 * config/alpha/linux.h (TARGET_HAS_SINCOS): Likewise.
117 * config/sparc/linux.h (TARGET_HAS_SINCOS): Likewise.
118 * config/sparc/linux64.h (TARGET_HAS_SINCOS): Likewise.
119 * config/rs6000/linux.h (TARGET_HAS_SINCOS): Likewise.
120 * config/rs6000/linux64.h (TARGET_HAS_SINCOS): Likewise.
122 2007-01-19 Uros Bizjak <ubizjak@gmail.com>
124 * config/i386/i386.md (*fpatanxf3_i387, fpatan_extend<mode>xf3_i387):
126 (atan2sf3_1, atan2df3_1, atan2xf3_1): Remove insn patterns.
127 (atan2xf3): Directly generate RTL pattern.
128 (atan2<mode>3): Rename from atan2sf3 and atan2df3 and macroize insn
129 patterns using X87MODEF12 mode macro. Use fpatan_extend<mode>xf3_i387
130 and truncate result to requested mode. Use SSE_FLOAT_MODE_P to
131 disable patterns for SSE math.
132 (atan<mode>2): Rename from atansf2 and atandf2 and macroize insn
133 patterns using X87MODEF12 mode macro. Use fpatan_extend<mode>xf3_i387
134 and truncate result to requested mode. Use SSE_FLOAT_MODE_P to
135 disable patterns for SSE math.
137 2007-01-19 Alexandre Oliva <aoliva@redhat.com>
139 * libgcc-std.ver: Fix typo in %inherit for GCC_4.3.0.
141 2007-01-18 Roger Sayle <roger@eyesopen.com>
143 * fold-const.c (fold_unary) <VIEW_CONVERT_EXPR>: Optimize away a
144 VIEW_CONVERT_EXPR to the same type as it's operand.
146 2007-01-18 David Edelsohn <edelsohn@gnu.org>
148 * config/rs6000/darwin-ldouble.c: Only build _SOFT_FLOAT if
149 configured for long double 128.
151 2007-01-18 Mike Stump <mrs@apple.com>
153 * config/rs6000/rs6000.c (rs6000_emit_vector_compare): Fix build
156 2007-01-18 Michael Meissner <michael.meissner@amd.com>
158 * i386.c (ix86_compute_frame_layout): Make fprintf's in #if 0 code
161 2007-01-18 Jan Hubicka <jh@suse.cz>
163 * tree-ssa-operands.c (vop_free_bucket_size): Never return value
164 greater than NUM_VOP_FREE_BUCKETS.
166 2007-01-18 Daniel Berlin <dberlin@dberlin.org>
168 * tree-ssa-structalias.c: Update comments.
169 (ptabitmap_obstack): Removed.
171 (oldpta_obstack): Ditto.
172 (stats): Add a few members.
173 (struct variable_info): Remove node, complex, address_taken, and
174 indirect_target members. Add oldsolution member.
175 (new_var_info): Do not initialize removed members.
176 (constraint_expr_type): Remove INCLUDES.
177 (constraint_graph): Add size, implicit_preds, rep,
178 indirect_cycles, eq_rep, label, direct_nodes, and complex members.
179 (FIRST_REF_NODE): New macro.
180 (LAST_REF_NODE): Ditto.
181 (FIRST_ADDR_NODE): Ditto.
182 (find): New function.
184 (dump_constraint): Do not handle INCLUDES.
185 (insert_into_complex): Do not insert duplicate constraints.
186 (condense_varmap_nodes): Renamed and rewritten into ...
187 (merge_node_constraints): This. Also fix bug in handling of
188 offseted copy constraints.
189 (clear_edges_for_node): No longer need to deal with preds at all,
190 or removing associated preds/succs.
191 (merge_graph_nodes): Deal with indirect_cycles.
192 Don't deal with predecessors.
193 (add_implicit_graph_edge): New function.
194 (add_pred_graph_edge): Ditto.
195 (add_graph_edge): Don't deal with predecessors.
196 (build_constraint_graph): Removed.
197 (build_pred_graph): New function.
198 (build_succ_graph): Ditto.
199 (struct scc_info): Removed in_component. Added roots, dfs, and
200 node_mapping. Remove visited_index, unification_queue.
201 (scc_visit): Deal with union-find we do now.
202 Deal with cycles with REF nodes.
203 (collapse_nodes): Renamed and rewritten to ...
205 (process_unification_queue): Removed.
206 (topo_visit): Cleanup
207 (do_da_constraint): Use find.
208 (do_sd_constraint): Ditto.
209 (do_ds_constraint): Ditto.
210 (do_complex_constraint): Ditto.
211 (init_scc_info): Update for removed and added members.
212 (find_and_collapse_graph_cycles): Renamed and rewritten into ...
213 (find_indirect_cycles): This.
214 (equivalence_class): New variable.
215 (label_visit): New function.
216 (perform_variable_substitution): Rewritten.
217 (free_var_substitution_info): New function.
218 (find_equivalent_node): Ditto.
219 (move_complex_constraints): Ditto.
220 (eliminate_indirect_cycles): Ditto.
221 (solve_graph): Only propagate changed bits.
222 Use indirect cycle elimination.
224 (tree_id_t): Rename to tree_vi_t, delete id member, add vi member.
225 (tree_id_eq): Renamed to ...
226 (tree_vi_eq): This. Update for member change
227 (insert_id_for_tree): Renamed and rewritten to ...
228 (insert_vi_for_tree): This.
229 (lookup_id_for_tree): Renamed and rewritten to ...
230 (lookup_vi_for_tree): This.
231 (get_id_for_tree): Renamed and rewritten to ...
232 (get_vi_for_tree): Ditto.
233 (get_constraint_exp_from_ssa_var): Update to use get_vi_for_tree.
234 (process_constraint): Don't handle INCLUDES.
235 Remove special ADDRESSOF case.
236 (find_func_aliases): Rewrite to use vi functions instead of id
238 (create_function_info_for): Ditto.
239 (create_variable_info_for): Ditto.
240 (intra_create_variable_infos): Ditto.
241 (merge_smts_into): Ditto.
242 (find_what_p_points_to): Ditto.
243 (init_base_vars): Ditto.
244 (init_alias_vars): Ditto.
245 (remove_preds_and_fake_succs): New function.
246 (dump_sa_points_to_info): Dump new stats.
247 (dump_solution_for_var): Use find.
248 (set_used_smts): Fix formatting.
249 (compute_points_to_sets): Updated for new functions.
250 (ipa_pta_execute): Ditto.
252 2007-01-18 Kazu Hirata <kazu@codesourcery.com>
253 Richard Sandiford <richard@codesourcery.com>
255 * doc/tm.texi (TARGET_FUNCTION_VALUE): Expand documentation of
257 * calls.c (expand_call): If the return value is a PARALLEL,
258 extract its first member.
259 * config/m68k/linux.h (FUNCTION_EXTRA_EPILOGUE): Remove.
260 * config/m68k/m68k.c (m68k_output_function_epilogue): Don't
261 use FUNCTION_EXTRA_EPILOGUE.
262 (m68k_function_value): Return a PARALLEL if the return value
263 is of a pointer type.
264 * config/m68k/netbsd-elf.h (current_function_returns_pointer)
265 (FUNCTION_EXTRA_EPILOGUE): Remove.
266 * config/m68k/m68k.md (D0_REG): New constant.
268 2007-01-18 Kazu Hirata <kazu@codesourcery.com>
270 * config/m68k/m68k.c (m68k_output_function_epilogue): Don't
271 output a NOP for empty epilogues.
273 2007-01-18 Richard Sandiford <richard@codesourcery.com>
275 * config/m68k/m68k.c (m68k_use_return_insn): Update comments
276 before function. Extend register save check to include all
277 registers, not just integer ones.
279 2007-01-18 Kazu Hirata <kazu@codesourcery.com>
281 * config/m68k/m68k-protos.h (use_return_insn): Rename to...
282 (m68k_use_return_insn): ...this.
283 * config/m68k/m68k.h (USE_RETURN_INSN): Delete.
284 * config/m68k/m68k.c (use_return_insn): Rename to...
285 (m68k_use_return_insn): ...this.
286 * config/m68k/m68k.md (return): Use m68k_use_return_insn instead
289 2007-01-18 Nathan Sidwell <nathan@codesourcery.com>
291 * config/m68k/fpgnulib.c (__truncdfsf2): Implement round to
292 nearest even, fix denormal rounding overflow.
294 2007-01-18 Richard Sandiford <richard@codesourcery.com>
296 * config/m68k/m68k.md (movsf_cf_hard): Use fsmove instead of
297 f%$smove and f%$move.
298 (movdf_cf_hard): Use fdmove for cases 0 and 3 and fmove for case 1.
299 (extendsfdf2_cf): Use fdmove instead of f%&move.
300 (truncdfsf2_cf): Use fsmove instead of f%$smove.
301 (add<mode>3_cf, sub<mode>3_cf): Use <FP:prec> instead of <FP:round>.
303 2007-01-18 Richard Sandiford <richard@codesourcery.com>
305 * config/m68k/m68k.md (movdf_cf_hard): Use output_move_double for
308 2007-01-18 Richard Sandiford <richard@codesourcery.com>
310 * real.h (real_format): Add a canonical_nan_lsbs_set field.
311 (coldfire_single_format): Declare.
312 (coldfire_double_format): Likewise.
313 * real.c (encode_ieee_single): Use canonical_nan_lsbs_set instead
314 of qnan_msb_set to determine the lower bits of a canonical
316 (encode_ieee_double): Likewise.
317 (encode_ieee_quad): Likewise.
318 (ieee_single_format): Initialize canonical_nan_lsbs_set.
319 (mips_single_format): Likewise.
320 (ieee_double_format): Likewise.
321 (mips_double_format): Likewise.
322 (ieee_extended_motorola_format): Likewise.
323 (ieee_extended_intel_96_format): Likewise.
324 (ieee_extended_intel_128_format): Likewise.
325 (ieee_extended_intel_96_round_53_format): Likewise.
326 (ibm_extended_format): Likewise.
327 (mips_extended_format): Likewise.
328 (ieee_quad_format): Likewise.
329 (mips_quad_format): Likewise.
330 (vax_f_format): Likewise.
331 (vax_d_format): Likewise.
332 (vax_g_format): Likewise.
333 (i370_single_format): Likewise.
334 (i370_double_format): Likewise.
335 (decimal_single_format): Likewise.
336 (decimal_double_format): Likewise.
337 (decimal_quad_format): Likewise.
338 (c4x_single_format): Likewise.
339 (c4x_extended_format): Likewise.
340 (real_internal_format): Likewise.
341 (coldfire_single_format): New real_format.
342 (coldfire_double_format): Likewise.
343 * config/pdp11/pdp11.c (pdp11_f_format): Initialize
344 canonical_nan_lsbs_set.
345 (pdp11_d_format): Likewise.
346 * config/m68k/m68k.c (override_options): Override REAL_FORMAT_MODE
347 if TARGET_COLDFIRE_CPU.
349 2007-01-18 Richard Sandiford <richard@codesourcery.com>
351 * config/m68k/m68k-protos.h (m68k_output_pic_call): Delete.
352 (output_call, m68k_legitimize_call_address): Declare.
353 * config/m68k/m68k.h (EXTRA_CONSTRAINT): Remove unnecessary
354 parenthesees. Add support for a 'W' constraint.
355 (LEGITIMATE_PIC_OPERAND_P): Remove SYMBOL_REF_FLAG handling.
356 (PRINT_OPERAND_PUNCT_VALID_P): Remove comment about 'o'.
357 (m68k_symbolic_call, m68k_symbolic_jump): Declare.
358 * config/m68k/m68k.c (m68k_symbolic_call, m68k_symbolic_jump): New
360 (override_options): Initialize them. Do not set flag_no_function_cse
361 for TARGET_ID_SHARED_LIBRARY.
362 (m68k_output_pic_call): Delete.
363 (m68k_legitimize_call_address): New function.
364 (print_operand): Remove the %o prefix. Handle the %p prefix.
365 (output_call): New function.
366 (m68k_output_mi_thunk): Use m68k_symbolic_jump. Always load the
367 target address from the GOT if symbolic jumps are not allowed.
368 * config/m68k/m68k.md (call, general_operand): Do not set
369 SYMBOL_REF_FLAG. Use m68k_legitimize_call_address instead.
370 Merge separate flag_pic and !flag_pic define_insns into...
371 (*call, *call_value): ...these new patterns. Match the address
372 rather than the containing MEM and require it to be a call_operand.
373 Use output_call to generate the asm template.
374 * config/m68k/predicates.md (const_call_operand): New predicate.
375 (call_operand): Likewise.
377 2007-01-18 Nathan Sidwell <nathan@codesourcery.com>
379 * config/m68k/m68k.h (REGISTER_MOVE_COST): Simplify definition.
380 (STACK_GROWS_DOWNWARD): Define to 1.
381 (FUNCTION_VALUE, LIBCALL_VALUE, FUNCTION_VALUE_REGNO_P): Equivocate
382 comments, emphasizing that these values are only defaults.
383 * config/m68k/linux.h (LINK_SPEC): Fix formatting in #undef.
384 * config/m68k/m68k.c (const_method): Remove trailing whitespace.
386 2007-01-18 Richard Sandiford <richard@codesourcery.com>
388 * config/m68k/m68k.md (cmpsi): Remove outdated flag_pic handling.
390 2007-01-18 Kazu Hirata <kazu@codesourcery.com>
391 Richard Sandiford <richard@codesourcery.com>
393 * config/m68k/m68k.h (DATA_REGNO_P, ADDRESS_REGNO_P, INT_REGNO_P)
394 (FP_REGNO_P): New macros.
395 (REGNO_OK_FOR_INDEX_P, REGNO_OK_FOR_BASE_P, REGNO_OK_FOR_DATA_P)
396 (REGNO_OK_FOR_FP_P, REG_OK_FOR_INDEX_P, REG_OK_FOR_BASE_P): Use them.
397 (EH_RETURN_STACKADJ_RTX): Use A0_REG.
398 * config/m68k/m68k.c (m68k_regno_mode_ok): Use the new REGNO macros.
400 2007-01-18 Nathan Sidwell <nathan@codesourcery.com>
402 * config.gcc (m68k-*-aout*, m68k-*-coff*, m68020-*-elf*, m68k-*-elf*)
403 (m68k-*-uclinuxoldabi, m68k-*-uclinux*, m68k-*-rtems*): Add t-floatlib
405 * config/m68k/t-floatlib: New file, extracting common code from...
406 * config/m68k/t-m68kbare, config/m68k/t-m68kelf,
407 * config/m68k/t-uclinux: Here.
408 * config/m68k/fpgnulib.c: Do not compile extendeed precision
409 routines on ColdFire targets.
411 2007-01-18 Nathan Sidwell <nathan@codesourcery.com>
413 * config.gcc (m68k-*-aout*, m68k-*-coff*, m68020-*-elf*, m68k-*-elf*)
414 (m68010-*-netbsdelf*, m68k*-*-netbsdelf*, m68k*-*-openbsd*)
415 (m68k-*-uclinuxoldabi, m68k-*-uclinux*, m68k-*-linux*)
416 (m68k-*-rtems*): Use tm_file rather than m68k/m68k.h and
417 explicitly set MOTOROLA to 1.
418 * config/m68k/m68k.h (MOTOROLA): Simplify definition accordingly.
420 2007-01-18 Richard Sandiford <richard@codesourcery.com>
421 Nathan Sidwell <nathan@codesourcery.com>
423 * config/m68k/m68k.h (PCC_STATIC_STRUCT_RETURN): Delete.
424 (ASM_OUTPUT_REG_PUSH, ASM_OUTPUT_REG_POP): Add MOTOROLA cases.
425 * config/m68k/coff.h (REGISTER_PREFIX_MD): Delete.
426 * config/m68k/m68020-elf.h (LIB_SPEC): Delete.
427 * config/m68k/m68k-none.h (CC1_SPEC, CPP_SUBTARGET_SPEC): Delete.
428 * config/m68k/m68kelf.h (IMMEDIATE_PREFIX, REGISTER_PREFIX_MD)
429 (ASM_OUTPUT_REG_PUSH, ASM_OUTPUT_REG_POP): Delete.
430 (NO_DOLLAR_IN_LABEL, PCC_STATIC_STRUCT_RETURN): Don't undefine.
431 (BSS_ASM_OP, ASM_OUTPUT_SKIP, ASM_OUTPUT_ASCII): Delete.
432 * config/m68k/m68kemb.h (PCC_STATIC_STRUCT_RETURN): Don't undefine.
433 * config/m68k/linux.h (SIZE_TYPE, PTRDIFF_TYPE, WCHAR_TYPE)
434 (WCHAR_TYPE_SIZE, TARGET_OBJFMT_CPP_BUILTINS): Delete.
435 (TARGET_OS_CPP_BUILTINS): Don't define mc68000 and mc68020 here.
436 (DBX_REGISTER_NUMBER): Delete.
437 * config/m68k/netbsd-elf.h (IMMEDIATE_PREFIX): Delete.
438 (PCC_STATIC_STRUCT_RETURN): Don't undefine.
439 * config/m68k/openbsd.h (PCC_STATIC_STRUCT_RETURN): Define.
441 2007-01-18 Richard Sandiford <richard@codesourcery.com>
443 * config.gcc (m68k-*-uclinux*): Add flat.h to $tm_file.
444 * config/flat.h: New file.
445 * crtstuff.c (USE_PT_GNU_EH_FRAME): Don't define if
447 * config/m68k/m68k.h (ASM_PREFERRED_EH_DATA_FORMAT): Do not use
448 indirect references for -msep-data or -mid-shared-library.
449 Do not use PC-relative code addresses either.
450 * config/m68k/m68k.c (override_options): Restrict -fPIC error
452 * config/m68k/uclinux.h (STARTFILE_SPEC): Define. Use Scrt1.o
453 for shared libraries and crt1.o for executables. Use crti.o and
455 (ENDFILE_SPEC): Use crtend.o and crtn.o.
456 (LIB_SPEC): Suppress -Rlibc.gdb if -static-libc is given.
457 Do not add -elf2flt or -shared-lib-id options here.
458 (LINK_SPEC): Define. Pass -elf2flt if no -elf2flt option is given.
459 Pass -shared-lib-id if -mid-shared-library, taking the library
460 identifier from -mshared-library-id if given, otherwise
462 (EH_FRAME_IN_DATA_SECTION): Do not undefine.
463 (INIT_SECTION_ASM_OP, FINI_SECTION_ASM_OP): Likewise.
464 (TARGET_OS_CPP_BUILTINS): Define __GXX_MERGED_TYPEINFO_NAMES=0
465 and __GXX_TYPEINFO_EQUALITY_INLINE=0 if -mid-shared-library.
466 (DRIVER_SELF_SPECS): Map unadorned PIC options to -msep-data.
467 * config/m68k/t-uclinux (EXTRA_MULTILIB_PARTS): Add crtbegin.o
469 * config/m68k/lb1sf68.asm (PICCALL): Use an lea and pc-relative
470 jump sequence for ISA A and ISA A+.
473 2007-01-18 Richard Sandiford <richard@codesourcery.com>
475 * config.gcc (m68k-*-uclinux*): Base the port on the common
476 and m68k GNU/Linux files rather than on the generic ELF ones.
477 * config/m68k/uclinux.h (TARGET_VERSION): Override.
478 (TARGET_OS_CPP_BUILTINS): Use LINUX_TARGET_OS_CPP_BUILTINS.
480 2007-01-18 Julian Brown <julian@codesourcery.com>
481 Richard Sandiford <richard@codesourcery.com>
483 * config/m68k/m68k.h (LONG_DOUBLE_TYPE_SIZE): Make 64-bit on ColdFire.
484 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
485 * config/m68k/netbsd-elf.h (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Undefine
487 * config/m68k/uclinux-oldabi.h (LONG_DOUBLE_TYPE_SIZE): Redefine to
489 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
491 2007-01-18 Richard Sandiford <richard@codesourcery.com>
493 * doc/install.texi: Document m68k-uclinuxoldabi.
494 * config.gcc (m68k-*-uclinuxoldabi): New configuration.
495 * config/m68k/uclinux-oldabi.h: New file, copied from
496 config/m68k/uclinux.h.
498 2007-01-18 Kaz Kojima <kkojima@rr.iij4u.or.jp>
500 * config/m32r/m32r-protos.h (m32r_expand_epilogue): Declare it.
501 * config/m32r/m32r.c (m32r_setup_incoming_varargs): Use gen_frame_mem.
502 (m32r_compute_frame_size): Use unsigned for regno.
503 (m32r_reload_lr): Use gen_frame_mem.
505 (m32r_output_function_epilogue): Don't output the function epilogue
507 (m32r_expand_epilogue): New.
508 (direct_return): Return false if the function has the interrupt
510 (m32r_hard_regno_rename_ok): Remove code for the textual epilogue.
511 * config/m32r/m32r.md (epilogue): New expander.
512 (return_lr, return_rte): New insns.
513 (return): Make it expander.
514 (return_normal): New expander.
516 2007-01-18 Josh Conner <jconner@apple.com>
519 * config/rs6000/rs6000.c (rs6000_emit_vector_compare): Add
520 support for UNLE, UNLT, UNGE, and UNGT.
522 2007-01-18 Jan Hubicka <jh@suse.cz>
524 * tree-vrp.c (finalize_jump_threads): Do not call cleanup_cfg by hand.
525 * tree-tailcall (add_virtual_phis): Likewise.
526 (optimize_tail_call): Return TODOs.
527 (execute_tail_calls): Return TODOs.
528 * tree-ssa-ccp (execute_fold_all_builtins): Do cleanup_cfg via TODO.
529 * tree-cfgcleanup.c (cleanup_tree_cfg_loop): Return if something
531 * tree-ssa-forwprop.c (tree_ssa_forward_propagate_single_use_value):
532 Cleanup cfg using TODO.
533 * tree-flow.h (cleanup_tree_cfg_loop): Update prototype.
534 * passes.c (execute_function_todo): When cleanup did something, remove
536 * tree-cfg.c (pass_build_cfg): Add cleanup_cfg TODO.
537 (make_edges): Don't cleanup_cfg.
539 2007-01-18 Uros Bizjak <ubizjak@gmail.com>
541 * reg-stack.c (subst_stack_regs_pat) [UNSPEC_SINCOS_COS,
542 UNSPEC_XTRACT_FRACT]: Use generic code for instructions that
543 operate on the top of stack.
544 [UNSPEC_SINCOS_SIN, UNSPEC_XTRACT_EXP, UNSPEC_TAN]: Rewrite
545 register handling of instructions that output to the second
547 [UNSPEC_TAN_ONE, UNSPEC_TAN_TAN]: Remove.
548 (move_for_stack_reg): Special-case check for dead destination
549 stack slot for constant load of 1.0 inside UNSPEC_TAN.
551 * config/i386/i386.md (UNSPEC_TAN): New constant.
552 (UNSPEC_TAN_ONE, UNSPEC_TAN_TAN): Remove.
553 (fptanxf4_i387, fptan_extend<mode>xf4_i387): New patterns
554 to correctly model move of constant 1.0 to top stack slot.
555 (*tandf3_1, *tansf3_1, *tanxf3_1): Remove insn patterns.
556 (unnamed peephole2 pattern): Remove corresponding peephole2
557 pattern that optimizes tan insn and loading of constant 1.0.
558 (tanxf2): Use fptanxf4_i387.
559 (tan<mode>2): Rename from tansf2 and tandf2 and macroize insn
560 patterns using X87MODEF12 mode macro. Use fptan_extend<mode>xf4_i387
561 and truncate result to requested mode. Use SSE_FLOAT_MODE_P to
562 disable patterns for SSE math.
563 (sincos<mode>3): Use truncxf<mode>2_i387_noop for truncation.
564 (fyl2x_extend<mode>xf3_i387): Use X87MODEF12 for operand 1.
566 2007-01-18 Dirk Mueller <dmueller@suse.de>
567 Richard Guenther <rguenther@suse.de>
570 * doc/invoke.texi (Warray-bounds): Document -Warray-bounds.
571 * common.opt (Warray-bounds): Add new warning option.
572 * c-opts.c (c_common_handle_option): Define -Warray-bounds
574 * Makefile.in: make tree-vrp.o depend on toplev.h
575 * tree-vrp.c (vrp_finalize): Call check_array_refs if -Warray-bounds
577 (check_array_refs, check_array_bounds, check_array_ref): New.
579 2007-01-18 Jan Hubicka <jh@suse.cz>
581 * tree-ssa-ccp.c (ccp_finalize): Return if something changed.
582 (execute_ssa_ccp): Return flags conditionally.
583 * tree-ssa-propagate.c (substitue_and_fold): Return if something was
585 * tree-ssa-propagate.h (substitute_and_fold): Update prototype.
587 2007-01-18 Steven Bosscher <steven@gcc.gnu.org>
589 * cfgcleanup.c (cleanup_cfg): Detect cfglayout mode and set
590 the CLEANUP_CFGLAYOUT flag when in cfglayout mode.
592 * Makefile.c (GTFILES): Add cfglayout.h.
593 * gengtype.c (open_base_files): Likewise.
594 * cfglayout.c (cfg_layout_function_footer,
595 cfg_layout_function_header) Reindent to make gengtype happy.
596 * cfglayout.h (cfg_layout_function_footer,
597 cfg_layout_function_header): Add GTY(()) marker.
599 * ifcvt.c (noce_try_sign_mask): Make sure INSN_B is non-null.
601 2007-01-18 Ben Elliston <bje@au.ibm.com>
603 * genautomata.c (write_automata): Include xstrerror output in the
604 error message if writing the DFA description file fails.
606 2007-01-17 H.J. Lu <hongjiu.lu@intel.com>
608 * config/mips/mips-protos.h (mips_output_external): Make it
610 * config/mips/iris.h (TARGET_ASM_EXTERNAL_LIBCALL): Removed.
611 * config/mips/mips.c (irix_output_external_libcall): Likewise.
612 (extern_list): Likewise.
613 (extern_head): Likewise.
614 (TARGET_ASM_FILE_END): Likewise.
615 (mips_file_end): Likewise.
616 (mips_output_external): Rewritten.
618 2007-01-18 Ben Elliston <bje@au.ibm.com>
620 * genpreds.c (write_insn_preds_c): Only write out the function
621 body for regclass_for_constraint if we have register constraints.
623 2007-01-17 Tom Tromey <tromey@redhat.com>
625 * doc/sourcebuild.texi (libgcj Tests): Use sourceware.org.
626 * doc/install.texi (Testing): Use sourceware.org.
627 (Binaries): Likewise.
628 (Specific): Likewise.
629 * doc/contrib.texi (Contributors): Use sourceware.org.
631 2007-01-17 Anatoly Sokolov <aesok@post.ru>
633 * config/avr/avr.h (AVR_HAVE_LPMX): New macro.
634 (AVR_ENHANCED): Rename to ...
635 (AVR_HAVE_MUL): ... new.
636 (avr_enhanced_p): Rename to ...
637 (avr_have_mul_p): ... new.
638 (TARGET_CPU_CPP_BUILTINS): Use 'avr_have_mul_p' instead of
639 'avr_enhanced_p' for "__AVR_ENHANCED__". Define "__AVR_HAVE_MUL__".
640 * config/avr/avr.c (avr_enhanced_p): Rename to ...
641 (avr_have_mul_p): ... new.
642 (base_arch_s): Rename 'enhanced' to 'have_mul'.
643 (avr_override_options): Use 'avr_have_mul_p' and 'have_mul' instead of
644 'avr_enhanced_p' and 'enhanced'.
645 (ashlhi3_out, ashrhi3_out, lshrhi3_out, avr_rtx_costs): Use
646 AVR_HAVE_MUL instead of AVR_ENHANCED.
647 * avr.md (*tablejump_enh): Use AVR_HAVE_LPMX instead of AVR_ENHANCED.
648 (mulqi3, *mulqi3_enh, *mulqi3_call, mulqihi3, umulqihi3, mulhi3,
649 *mulhi3_enh, *mulhi3_call, mulsi3, *mulsi3_call): Use AVR_HAVE_MUL
650 instead of AVR_ENHANCED.
651 (*tablejump_enh): Use AVR_HAVE_LPMX instead of AVR_ENHANCED.
652 * libgcc.S: Use __AVR_HAVE_MUL__ instead of __AVR_ENHANCED__.
653 (__tablejump__): Use __AVR_HAVE_LPMX__ instead of __AVR_ENHANCED__.
655 2007-01-17 Ian Lance Taylor <iant@google.com>
657 * vec.h (VEC_reserve_exact): Define.
658 (vec_gc_p_reserve_exact): Declare.
659 (vec_gc_o_reserve_exact): Declare.
660 (vec_heap_p_reserve_exact): Declare.
661 (vec_heap_o_reserve_exact): Declare.
662 (VEC_OP (T,A,reserve_exact)): New static inline function, three
664 (VEC_OP (T,A,reserve)) [all versions]: Remove handling of
666 (VEC_OP (T,A,alloc)) [all versions]: Call ...reserve_exact.
667 (VEC_OP (T,A,copy)) [all versions]: Likewise.
668 (VEC_OP (T,a,safe_grow)) [all versions]: Likewise.
669 * vec.c (calculate_allocation): Add exact parameter. Change all
671 (vec_gc_o_reserve_1): New static function, from vec_gc_o_reserve.
672 (vec_gc_p_reserve, vec_gc_o_reserve): Call vec_gc_o_reserve_1.
673 (vec_gc_p_reserve_exact, vec_gc_o_reserve_exact): New functions.
674 (vec_heap_o_reserve_1): New static function, from vec_heap_o_reserve.
675 (vec_heap_p_reserve, vec_heap_o_reserve): Call vec_heap_o_reserve_1.
676 (vec_heap_p_reserve_exact): New function.
677 (vec_heap_o_reserve_exact): New function.
679 2007-01-17 Jan Hubicka <jh@suse.cz>
681 * ipa-type-escape.c (look_for_casts): Revamp using handled_component_p.
683 2007-01-17 Eric Christopher <echristo@apple.com>
685 * config.gcc: Support core2 processor.
687 2007-01-16 Jan Hubicka <jh@suse.cz>
689 * tree-ssanames.c (release_dead_ssa_names): Instead of ggc_freeing
690 the names, just unlink the chain so we don't crash on dangling pointers
693 2007-01-16 Jan Hubicka <jh@suse.cz>
695 * cgraph.h (cgraph_decide_inlining_incrementally): Kill.
696 * tree-pass.h: Reorder to make IPA passes appear toegher.
697 (pass_early_inline, pass_inline_parameters, pass_apply_inline): Declare.
698 * cgraphunit.c (cgraph_finalize_function): Do not compute inling
699 parameters, do not call early inliner.
700 * ipa-inline.c: Update comments. Include tree-flow.h
701 (cgraph_decide_inlining): Do not compute inlining parameters.
702 (cgraph_decide_inlining_incrementally): Return TODOs; assume to
703 be called with function context set up.
704 (pass_ipa_inline): Remove unreachable functions before pass.
705 (cgraph_early_inlining): Simplify assuming to be called from the
707 (pass_early_inline): New pass.
708 (cgraph_gate_ipa_early_inlining): New gate.
709 (pass_ipa_early_inline): Turn into simple wrapper.
710 (compute_inline_parameters): New function.
711 (gate_inline_passes): New gate.
712 (pass_inline_parameters): New pass.
713 (apply_inline): Move here from tree-optimize.c
714 (pass_apply_inline): New pass.
715 * ipa.c (cgraph_remove_unreachable_nodes): Verify cgraph after
717 * tree-inline.c (optimize_inline_calls): Return TODOs rather than
719 (tree_function_versioning): Do not allocate dummy struct function.
720 * tree-inline.h (optimize_inline_calls): Update prototype.
721 * tree-optimize.c (execute_fixup_cfg): Export.
722 (pass_fixup_cfg): Remove
723 (tree_rest_of_compilation): Do not apply inlines.
724 * tree-flow.h (execute_fixup_cfg): Declare.
725 * Makefile.in (gt-passes.c): New.
726 * passes.c: Include gt-passes.h
727 (init_optimization_passes): New passes.
728 (nnodes, order): New static vars.
729 (do_per_function_toporder): New function.
730 (execute_one_pass): Dump current pass here.
731 (execute_ipa_pass_list): Don't dump current pass here.
733 2007-01-16 Janis Johnson <janis187@us.ibm.com>
735 * config/dfp-bit.c (dfp_compare_op): Return separate value for NaN.
736 (DFP_NE, DFP_LE, DFP_GE): Return false for NaN.
738 2007-01-16 David Edelsohn <edelsohn@gnu.org>
740 * config/rs6000/darwin-ldouble.c: Build file for SOFT_FLOAT.
741 (strong_alias): Define.
742 (__gcc_qmul): Provide non-FMA for soft-float.
758 * config/rs6000/rs6000.c (rs6000_init_libfuncs): Initialize
759 soft-float functions.
760 * config/rs6000/libgcc-ppc-glibc.ver: Version soft-float symbols.
761 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Do not warn
762 about long double soft float.
764 2007-01-16 Dorit Nuzman <dorit@il.ibm.com>
765 Tehila Meyzels <tehila@il.ibm.com>
767 * tree-vectorizer.h (is_pattern_stmt_p): New.
768 * tree-vect-analyze.c (vect_determine_vectorization_factor): Fix
769 formatting (tabs instead of spaces). Cleanup and clarify setting
770 of STMT_VINFO_VECTYPE. Call is_pattern_stmt_p.
771 * tree-vect-transform.c (vect_get_vec_def_for_operand): Fix typo.
772 (vectorizable_type_demotion): Check that types are integral.
773 (vectorizable_type_promotion): Likewise.
774 (vectorizable_store): Fix typo. Eliminate new-line at end of
777 2007-01-16 Jan Hubicka <jh@suse.cz>
779 * tree-ssanames.c (release_dead_ssa_names): Remove invalidated
782 2007-01-15 Eric Christopher <echristo@apple.com>
784 * ifcvt.c: Include vec.h, vecprim.h.
785 (check_cond_move_block): New argument regs.
786 Reorganize. Add registers used to regs.
787 (cond_move_process_if_block): Use regs set above as
790 2007-01-15 Eric Christopher <echristo@apple.com>
792 * config/darwin.h: Update copyright.
793 (TARGET_OPTION_TRANSLATE_TABLE): Add umbrella.
794 (LINK_COMMAND_SPEC): Add -u.
795 (LINK_SPEC): Fix umbrella for above.
797 2007-01-15 Joseph S. Myers <joseph@codesourcery.com>
799 * config/soft-fp/op-common.h, config/soft-fp/op-4.h: Update from
802 2007-01-15 Tom Tromey <tromey@redhat.com>
804 * doc/sourcebuild.texi (libgcj Tests): Don't mention jacks.
805 * doc/install.texi (Testing): Don't mention jacks.
806 (Configuration): Document --enable-java-maintainer-mode. Move
807 --with-java-home to libgcj-specific section. Document
809 (Prerequisites): Mention --enable-java-maintainer-mode, ecj1.
811 2007-01-15 Jan Hubicka <jh@suse.cz>
813 * tree-ssa-dce.c (DCE_TODOs): New.
814 (propagate_necessity): Return if something changed.
815 (eliminate_unnecessary_stmts): Likewise.
816 (perform_tree_ssa_dce): Return TODO flags when needed.
817 (pass_dce, pass_dce_loop, pass_cd_dce): Remove TODO flags.
819 2007-01-15 Uros Bizjak <ubizjak@gmail.com>
821 * config/i386/i386.md (fyl2xxf3_i387): Rename from fyl2x_xf3.
822 (fyl2x_extend<mode>xf3_i387): New insn pattern.
823 (log<mode>2): Rename from logsf2 and logdf2 and macroize insn
824 patterns using X87MODEF12 mode macro. Extend operand 1
825 to XFmode. Use SSE_FLOAT_MODE_P to disable patterns for SSE math.
826 (log10<mode>2): Ditto.
827 (log2<mode>2): Ditto.
828 (log1p<mode>2): Ditto.
829 (logb<mode>2): Ditto.
830 (fyl2xp1xf3_i387): Rename from fyl2xp1_xf3.
831 (fyl2xp1_extend<mode>xf3_i387): New insn pattern.
832 (*fxtractxf3_i387): Rename from *fxtractxf3.
833 (fxtract_extend<mode>xf3_i387): New insn pattern.
834 (ilogbsi2): Use match_dup 3, not match_operand:XF 3.
835 * config/i386/i386.c (ix86_emit_i387_log1p): Use gen_fyl2xp1xf3_i387()
836 and gen_fyl2xxf3_i387().
838 2007-01-14 Zdenek Dvorak <dvorakz@suse.cz>
840 * loop-unswitch.c (unswitch_loop): Do not call fix_loop_placement.
841 * cfgloopmanip.c (fix_loop_placement): Made static. Use
842 get_loop_exit_edges. Changed return type to bool.
843 * cfgloop.h (fix_loop_placement): Declaration removed.
845 2007-01-14 Dorit Nuzman <dorit@il.ibm.com>
847 * param.h (MIN_VECT_LOOP_BOUND): New.
848 * params.def (MIN_VECT_LOOP_BOUND): New.
849 * tree-vectorizer.c (slpeel_tree_peel_loop_to_edge): Takes another
850 argument - minimum threshold for number of iterations.
851 * tree-vectorizer.h (slpeel_tree_peel_loop_to_edge): Add another
852 argument to declaration.
853 * tree-vect-analyze.c (vect_analyze_operations): Check value of
855 * tree-vect-transform.c (vect_do_peeling_for_loop_bound): Call
856 slpeel_tree_peel_loop_to_edge with additional argument.
857 (vect_do_peeling_for_alignment): Likewise.
858 * doc/invoke.texi (min-vect-loop-bound): Document new param option.
860 2007-01-14 Uros Bizjak <ubizjak@gmail.com>
863 * config/i386/i386.c (print_operand) ['z']: Output 'b' for
866 2007-01-14 Jan Hubicka <jh@suse.cz>
868 * tree-dfa.c (remove_referenced_var): New function.
869 * tree-ssa-live.c (remove_unused_locals): Walk referenced vars and
870 prune referenced vars list too.
871 * tree-flow.h (remove_referenced_var): Declare.
873 2007-01-14 Jan Hubicka <jh@suse.cz>
875 * tree-eh.c (add_stmt_to_eh_region_fn): Do not add call_exprs
877 (remove_stmt_from_eh_region_fn): Do not remove call_exprs.
878 (verify_eh_throw_stmt_node, verify_eh_throw_table_statements): Kill.
879 * except.h (verify_eh_throw_table_statements): Kill prototype.
880 * cfgexpand.c (expand_gimple_basic_block): Propagate Eh regions
881 into call exrepssions.
882 * tree-optimize.c (execute_free_cfg_annotatiosn): Do not call
884 * tree-cfg.c: Include pointer-set.h.
885 (verify_node_sharing): Work on pointer set.
886 (verify_eh_throw_stmt_node): New.
887 (verify_stmts): Use pointers sets, verify throw_stmt.
889 2007-01-13 Zdenek Dvorak <dvorakz@suse.cz>
891 * ipa-reference.c (analyze_function): Consider also addresses taken
894 2007-01-12 Roger Sayle <roger@eyesopen.com>
896 * c-typeck.c (null_pointer_constant_p): Replace use of
897 TREE_CONSTANT_OVERFLOW with TREE_OVERFLOW.
898 (build_c_cast): Likewise.
900 2007-01-12 Roger Sayle <roger@eyesopen.com>
902 * tree.h (force_fit_type_double): Remove unused final argument.
903 * c-common.c (constant_expression_warning): Replace use of
904 TREE_CONSTANT_OVERFLOW with TREE_OVERFLOW.
905 (convert_and_check): Likewise.
906 (shorten_compare): Update call to force_fit_type_double.
907 (c_common_truthvalue_conversion) <INTEGER_CST>: Use integer_zerop.
908 * convert.c (convert_to_pointer): Update call to
909 force_fit_type_double.
910 * fold-const.c (force_fit_type_double): Remove overflowed_const
912 (int_const_binop, fold_convert_const_int_from_int,
913 fold_convert_const_int_from_real, fold_div_compare,
914 fold_sign_changed_comparison, fold_unary, fold_negate_const,
915 fold_abs_const, fold_not_const): Remove the final argument from
916 calls to force_fit_type_double.
918 2007-01-12 Andrew Pinski <andrew_pinski@playstation.sony.com>
920 * configure.ac: Set insn to "nop" for spu-*-* also.
921 * configure: Regenerate.
923 2007-01-12 Olga Golovanevsky <olga@il.ibm.com>
925 * builtins.def : Add BUILT_IN_FREE.
927 2007-01-12 Jan Hubicka <jh@suse.cz>
929 PR tree-optimization/30443
930 * tree-inline.c (tree_function_versioning): Do not optimize when
931 cloning for inlining.
933 2007-01-12 Zdenek Dvorak <dvorakz@suse.cz>
935 * doc/loop.texi: Document recording of loop exits.
936 * cfgloopmanip.c (loopify, duplicate_loop): Use alloc_loop.
937 (update_single_exits_after_duplication,
938 update_single_exit_for_duplicated_loop,
939 update_single_exit_for_duplicated_loops): Removed.
940 (duplicate_loop_to_header_edge): Do not call
941 update_single_exits_after_duplication and
942 update_single_exit_for_duplicated_loops.
943 (loop_version): Do not update single_exit information.
944 (fix_loop_structure): Use record_loop_exits instead of
945 mark_single_exit_loops.
946 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update
947 the lists of loop exits.
948 * cfghooks.c (redirect_edge_and_branch, redirect_edge_and_branch_force,
949 split_edge, merge_blocks): Update the lists of loop exits.
950 * modulo-sched.c (sms_schedule): Pass LOOPS_HAVE_RECORDED_EXITS to
952 * loop-init.c (loop_optimizer_init): Call record_loop_exits instead
953 of mark_single_exit_loops.
954 (loop_optimizer_finalize): Call release_recorded_exits.
955 * tree-ssa-loop.c (tree_loop_optimizer_init): Pass
956 LOOPS_HAVE_RECORDED_EXITS to loop_optimizer_init.
957 * tree-vectorizer.c (slpeel_tree_duplicate_loop_to_edge_cfg): Do not
958 update single exit information.
959 * lambda-code.c (perfect_nestify): Ditto.
960 * cfgloop.c (flow_loop_free): Destroy the list of exits of the loop.
961 (mark_single_exit_loops): Removed.
962 (alloc_loop, loop_exit_hash, loop_exit_eq, loop_exit_free,
963 get_exit_descriptions, rescan_loop_exit, record_loop_exits,
964 dump_recorded_exit, dump_recorded_exits, release_recorded_exits): New
966 (get_loop_exit_edges, single_exit): Use recorded exit lists.
967 (add_bb_to_loop, remove_bb_from_loops): Update the lists of loop exits.
968 (verify_loop_structure): Verify consistency of the exit lists.
969 (flow_loops_find): Use alloc_loop. Initialize exits hash.
970 (set_single_exit): Removed.
971 * cfgloop.h (struct loop_exit): New function.
972 (struct loop): single_exit_ field replaced by exits field.
973 (LOOPS_HAVE_MARKED_SINGLE_EXITS): Replaced by LOOPS_HAVE_RECORDED_EXITS.
974 (struct loops): Added exits hash.
975 (mark_single_exit_loops, set_single_exit): Declaration removed.
976 (release_recorded_exits, record_loop_exits, rescan_loop_exit): Declare.
978 2007-01-12 Richard Sandiford <richard@codesourcery.com>
980 * doc/invoke.texi: Avoid use of @headitem.
982 2007-01-12 Richard Sandiford <richard@codesourcery.com>
984 * cse.c (cse_insn): Move HAVE_CC0 code after declarations.
986 2007-01-12 Richard Sandiford <richard@codesourcery.com>
988 * doc/install.texi: Fix m68k-*-* anchor and add m68k-*-* to the
991 2007-01-12 Nathan Sidwell <nathan@codesourcery.com>
992 Richard Sandiford <richard@codesourcery.com>
994 * doc/invoke.texi: Document -mno-bitfield, -mno-rtd and -mno-short.
995 * config/m68k/m68k.opt: Resort options.
996 (mbitfield, mrtd, mshort): Remove RejectNegative properties.
998 2007-01-12 Nathan Sidwell <nathan@codesourcery.com>
999 Richard Sandiford <richard@codesourcery.com>
1001 * doc/invoke.texi: Document the macros that are defined by
1002 m68k's -mtune and -mhard-float options.
1003 * config/m68k/m68k-protos.h (m68k_cpp_cpu_ident) Declare.
1004 (m68k_cpp_cpu_family): Likewise.
1005 * config/m68k/m68k.h (TARGET_CPU_CPP_BUILTINS): Add a full set
1006 of __ucfv*__ macros. Define __mcffpu__ if generating code for
1007 ColdFire FPUs. Define __mcf_cpu_* and __mcf_family_* macros.
1008 * config/m68k/m68k.c (m68k_cpp_cpu_ident): New function.
1009 (m68k_cpp_cpu_family): Likewise.
1011 2007-01-12 Richard Sandiford <richard@codesourcery.com>
1013 * config/m68k/m68k.h (TARGET_CPU_CPP_BUILTINS): Treat all mc68*
1014 macros besides mc68000 as tuning macros. Use a switch statement
1015 to set them and mcpu32.
1017 2007-01-12 Julian Brown <julian@codesourcery.com>
1019 * config/m68k/m68k.h: Use TARGET_68040 instead of TARGET_68040_ONLY.
1020 (TARGET_68040_ONLY): Rename to...
1021 (TARGET_68040): ...this.
1022 * config/m68k/m68k.c: Use TARGET_68040 instead of TARGET_68040_ONLY.
1023 * config/m68k/m68k.md: Likewise.
1025 2007-01-12 Julian Brown <julian@codesourcery.com>
1026 Nathan Sidwell <nathan@codesourcery.com>
1027 Richard Sandiford <richard@codesourcery.com>
1029 * config.gcc (m680[012]0-*-*, m68k*-*-*): Set m68k_cpu_ident to
1030 the -mcpu= argument associated with the --with-cpu setting.
1031 Define M68K_DEFAULT_TUNE to the default -mtune= option,
1032 if different from the one implied by the -mcpu setting.
1033 Accept --with-cpu=FOO if FOO is listed in m68k-devices.def,
1034 using mcpu=FOO as the default CPU option. Set target_cpu_default2.
1035 * doc/invoke.texi: Mention ColdFire in the introduction to the
1036 m68k options. Document the new -march, -mcpu, -mtune, -mdiv,
1037 -mno-div and -mhard-float options. Make -m68881 a synonym for
1038 -mhard-float. Document the previously-undocumented -m5206e,
1039 -m528x, -m5307 and -m5407 options. Tweak the existing option
1040 documentation for consistency.
1041 * doc/install.texi: Mention new --with-cpu arguments.
1042 * config/m68k/m68k.h (OPTION_DEFAULT_SPECS): Only use the
1043 default CPU if neither -mcpu nor -march are specified.
1044 (ASM_CPU_SPEC): Pass down -mcpu and -march options.
1045 (TARGET_CPU_CPP_BUILTINS): Set __mcfisa*__ macros from
1046 TARGET_ISA*. Set the legacy __mcf*__ cpu macros in the same way,
1047 using m68k_tune to decide between families that implement the
1048 same ISA. Use m68k_tune to set __mcfv4e__.
1049 (FL_BITFIELD, FL_68881, FL_COLDFIRE, FL_CF_HWDIV, FL_CF_MAC)
1050 (FL_CF_EMAC, FL_CF_EMAC_B, FL_CF_USP, FL_CF_FPU, FL_ISA_68000)
1051 (FL_ISA_68010, FL_ISA_68020, FL_ISA_68040, FL_ISA_A, FL_ISA_B)
1052 (FL_ISA_C, FL_ISA_MMU): New macros.
1053 (MASK_COLDFIRE): Delete.
1054 (TARGET_68010, TARGET_68020, TARGET_68040_ONLY, TARGET_COLDFIRE)
1055 (TARGET_ISAB): Redefine in terms of m68k_cpu_flags.
1056 (TARGET_68881, TARGET_COLDFIRE_FPU): Redefine in terms of m68k_fpu.
1057 (TARGET_HARD_FLOAT): Do not define here.
1058 (TARGET_ISAAPLUS, TARGET_ISAC): New macros.
1059 (TUNE_68000): New macro.
1060 (TUNE_68000_10): Redefine in terms of TUNE_68000 and TUNE_68010.
1061 (TUNE_68010, TUNE_68030, TUNE_68040, TUNE_68060, TUNE_CPU32)
1062 (TUNE_CFV2): Redefine in terms of m68k_tune.
1063 (uarch_type, target_device, fpu_type): New enums.
1064 (m68k_cpu, m68k_tune, m68k_fpu, m68k_cpu_flags): Declare.
1065 * config/m68k/m68k.c (TARGET_DEFAULT): Remove MASK_68881.
1066 (FL_FOR_isa_00, FL_FOR_isa_10, FL_FOR_isa_20, FL_FOR_isa_40)
1067 (FL_FOR_isa_cpu32, FL_FOR_isa_a, FL_FOR_isa_aplus, FL_FOR_isa_b)
1068 (FL_FOR_isa_c): New macros.
1069 (m68k_isa): New enum.
1070 (m68k_target_selection): New structure.
1071 (all_devices, all_isas, all_microarchs): New tables.
1072 (m68k_cpu_entry, m68k_arch_entry, m68k_tune_entry, m68k_cpu)
1073 (m68k_tune, m68k_fpu, m68k_cpu_flags): New variables.
1074 (MASK_ALL_CPU_BITS): Delete.
1075 (m68k_find_selection): New function.
1076 (m68k_handle_option): Handle -mcpu=, -march= and -mtune=.
1077 Map the legacy target options to a combination of the new ones.
1078 (override_options): Set m68k_cpu, m68k_tune, m68k_fpu and
1079 m68k_cpu_flags. Handle M68K_DEFAULT_TUNE. Use m68k_cpu_flags
1080 to derive default MASK_BITFIELD, MASK_CF_HWDIV and MASK_HARD_FLOAT
1082 * config/m68k/m68k.opt (m5200, m5206e, m528x, m5307, m5407, mcfv4e)
1083 (m68010, m68020, m68020-40, m68020-60, m68030, m68040): Remove Mask
1085 (m68881, msoft-float): Change mask from 68881 to HARD_FLOAT.
1086 (march=, mcpu=, mdiv, mhard-float, mtune=): New options.
1087 * config/m68k/m68k-devices.def: New file.
1089 2007-01-12 Richard Sandiford <richard@codesourcery.com>
1090 Nathan Sidwell <nathan@codesourcery.com>
1092 * config/m68k/m68k.h (ASM_CPU_SPEC, ASM_SPEC, EXTRA_SPECS)
1093 (SUBTARGET_EXTRA_SPECS): New macros.
1094 * config/m68k/linux.h (ASM_SPEC): Remove CPU flags;
1095 use %(asm_cpu_spec) instead.
1096 * config/m68k/m68k-none.h (ASM_SPEC): Likewise.
1097 * config/m68k/openbsd.h (ASM_SPEC): Likewise.
1098 * config/m68k/netbsd-elf.h (ASM_SPEC): Likewise.
1099 (EXTRA_SPECS): Rename to...
1100 (SUBTARGET_EXTRA_SPECS): ...this.
1102 2007-01-12 Nathan Sidwell <nathan@codesourcery.com>
1103 Richard Sandiford <richard@codesourcery.com>
1104 Julian Brown <julian@codesourcery.com>
1106 * config.gcc (m68k-*-aout*, m68k-*-coff*, m68020-*-elf*, m68k-*-elf*)
1107 (m68k-*-uclinux*, m68k-*-linux*, m68k-*-rtems*): Set default_m68k_cpu
1108 to the configuration's default CPU.
1109 (m68010-*-netbsdelf*, m68k*-*-netbsdelf*, m68k*-*-openbsd*): Likewise.
1110 Remove default masks.
1111 (m680[012]0-*-*): Set the default with_cpu to the first part of
1113 (m68k*-*-*): Set the default with_cpu to m$default_m68k_cpu.
1114 (m68k*-*-linux): Extend the --with-cpu handling to...
1115 (m680[012]0-*-*, m68k*-*-*): ...these configurations. Allow m68000
1116 and m68010. Don't set target_cpu_default2.
1117 * doc/install.texi: Document --with-cpu for m68k.
1118 * config/m68k/m68k.h (OPTION_DEFAULT_SPECS): Define.
1119 * config/m68k/m68k-none.h (TARGET_CPU_DEFAULT, M68K_CPU_m68k)
1120 (M68K_CPU_m68000, M68K_CPU_m68010, M68K_CPU_m68020, M68K_CPU_m68030)
1121 (M68K_CPU_m68040, M68K_CPU_m68302, M68K_CPU_m68332, TARGET_DEFAULT)
1122 (ASM_CPU_DEFAULT_SPEC, CC1_CPU_DEFAULT_SPEC): Delete.
1123 (ASM_SPEC): Remove use of %(asm_cpu_default).
1124 (EXTRA_SPECS, SUBTARGET_EXTRA_SPECS, MULTILIB_DEFAULTS): Delete.
1125 * config/m68k/linux.h (TARGET_DEFAULT): Delete.
1126 (CPP_SPEC): Merge definitions. Do not handle __HAVE_68881__ here.
1127 * config/m68k/netbsd-elf.h (TARGET_OS_CPP_BUILTINS): Define
1128 __HAVE_FPU__ if TARGET_HARD_FLOAT.
1129 (TARGET_DEFAULT): Delete.
1130 (EXTRA_SPECS): Delete cpp_cpu_default_spec, cpp_cpu_spec,
1131 cpp_fpu_spec, asm_default_spec and netbsd_cpp_spec.
1132 (CPP_CPU_SPEC): Delete.
1133 (TARGET_VERSION): Merge definitions, using TARGET_68010 to pick
1134 the appropriate string.
1135 (CPP_CPU_DEFAULT_SPEC, ASM_DEFAULT_SPEC, CPP_FPU_SPEC): Delete.
1136 (CPP_SPEC): Define to NETBSD_CPP_SPEC.
1137 (ASM_SPEC): Don't use %(asm_default_spec).
1138 * config/m68k/m68k.c (TARGET_DEFAULT_TARGET_FLAGS): Remove
1139 TARGET_DEFAULT and add MASK_68881.
1140 * config/m68k/m68k.md: Remove mention of TARGET_DEFAULT from comments.
1142 2007-01-12 Richard Sandiford <richard@codesourcery.com>
1144 * config.gcc (m68010-*-netbsdelf*): Add MASK_68010.
1145 (m68k*-*-netbsdelf*, m68k*-*-openbsd*, m68k*-linux*): Add
1146 MASK_68010 alongside MASK_68020.
1147 * doc/invoke.texi: Document -m68010.
1148 * config/m68k/m68k.opt (m68010): New.
1149 * config/m68k/m68k.h (TARGET_CPU_CPP_BUILTINS): Define mc68010
1151 (TUNE_68010): New macro.
1152 * config/m68k/m68k-none.h (M68K_CPU_m68k, M68K_CPU_m68010)
1153 (M68K_CPU_m68020, M68K_CPU_m68030, M68K_CPU_m68040)
1154 (M68K_CPU_m68332): Add MASK_68010.
1155 * config/m68k/linux.h (TARGET_DEFAULT): Add MASK_68010 to
1156 fallback definition.
1157 * config/m68k/netbsd-elf.h (CPP_CPU_SPEC): Remove now-redundant
1159 * config/m68k/m68k.c (MASK_ALL_CPU_BITS): Add MASK_68010.
1160 (m68k_handle_option): Handle OPT_m68010. Add MASK_68010
1161 to all entries that use MASK_68020.
1162 (output_move_simode_const, output_move_himode, output_move_qimode)
1163 (output_move_stricthi, output_move_strictqi): Use TARGET_68010
1164 instead of TARGET_68020 to select clr behavior. Remove comment
1165 about there being no TARGET_68010.
1166 * config/m68k/m68k.md: Likewise throughout.
1168 2007-01-12 Julian Brown <julian@codesourcery.com>
1170 * config/m68k/m68k.h (TARGET_ISAB): New macro.
1171 * config/m68k/m68k.c: Use TARGET_ISAB rather than TARGET_CFV4.
1172 * config/m68k/m68k.md: Likewise.
1174 2007-01-12 Julian Brown <julian@codesourcery.com>
1176 * config/m68k/m68k.h (LEGITIMATE_INDEX_P, LEGITIMIZE_ADDRESS): Use
1177 TARGET_COLDFIRE_FPU instead of TARGET_CFV4E.
1179 2007-01-12 Julian Brown <julian@codesourcery.com>
1181 * config/m68k/m68k.h (TUNE_68040_60): New macro.
1182 * config/m68k/m68k.c (standard_68881_constant_p): Use it.
1183 * config/m68k/m68k.md: Likewise.
1185 2007-01-12 Julian Brown <julian@codesourcery.com>
1186 Richard Sandiford <richard@codesourcery.com>
1188 * config/m68k/m68k.h (TARGET_CPU_CPP_BUILTINS): Use TUNE_68030
1189 instead of TARGET_68030, TUNE_68040 instead of TARGET_68040,
1190 TUNE_68060 instead of TARGET_68060 and TUNE_CPU32 instead of
1192 (TARGET_CPU32): Rename to...
1193 (TUNE_CPU32): ...this.
1194 (TUNE_68000_10, TUNE_68030, TUNE_68040, TUNE_68060)
1195 (TUNE_CFV2): New macros.
1196 * config/m68k/netbsd-elf.h (LONG_DOUBLE_TYPE_SIZE): Simplify;
1197 remove conditions that are implied by TARGET_68020.
1198 * config/m68k/m68k.c (m68k_output_function_prologue): Use TUNE_68040
1199 instead of TARGET_68040 and TUNE_CPU32 instead of TARGET_CPU32.
1200 (m68k_output_function_epilogue): Likewise.
1201 (m68k_rtx_costs): Likewise. Use TUNE_68060 instead of TARGET_68060
1202 and TUNE_CFV2 instead of TARGET_5200. Use TUNE_68000_10 instead of
1203 "!TARGET_68020 && !TARGET_COLDFIRE" to choose between 68000 and
1204 non-68000 timings. Refactor multiplication and division costs.
1205 (output_addsi3): Use TUNE_68040 instead of TARGET_68040 and
1206 TUNE_CPU32 instead of TARGET_CPU32.
1207 (standard_68881_constant_p): Use TUNE_68040 instead of TARGET_68040
1208 and TUNE_68060 instead of TARGET_68060.
1209 * config/m68k/m68k.md: Use TUNE_68040 instead of TARGET_68040,
1210 TUNE_68060 instead of TARGET_68060, and TUNE_CPU32 instead of
1212 (movsi_const0): Use TUNE_68000_10 rather than "!TARGET_68020
1213 && !TARGET_COLDFIRE" to choose between moveq and clr.
1214 Likewise in the unnamed movsf pattern.
1215 (ashlsi_17_24, lshrsi_17_24): Guard with TUNE_68000_10 rather than
1216 "!TARGET_68020 && !TARGET_COLDFIRE". Likewise the unnamed
1219 2007-01-12 Richard Sandiford <richard@codesourcery.com>
1221 * config/m68k/m68k.h (TARGET_CPU_CPP_BUILTINS): Increase amount
1222 of tabbing before backslashes.
1224 2007-01-11 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
1226 * pa-linux.h (ASM_OUTPUT_INTERNAL_LABEL): Undefine.
1227 * pa.h (ASM_OUTPUT_LABEL): Output colon when using GAS.
1228 (ASM_OUTPUT_INTERNAL_LABEL): Define.
1230 2007-01-11 Zdenek Dvorak <dvorakz@suse.cz>
1232 * tree-ssa-loop-ivopts.c (extract_cond_operands): Split from
1233 find_interesting_uses_cond.
1234 (find_interesting_uses_cond): Use extract_cond_operands.
1235 (rewrite_use_compare): Use extract_cond_operands and
1236 force_gimple_operand_bsi. Do not call update_stmt.
1237 (determine_use_iv_cost_condition): Use extract_cond_operands.
1238 Return cheaper of using original bound and changing the exit bound.
1240 2007-01-11 Zdenek Dvorak <dvorakz@suse.cz>
1242 PR tree-optimization/29516
1243 * tree-ssa-address.c (tree_mem_ref_addr, add_to_parts,
1244 most_expensive_mult_to_index, addr_to_parts,
1245 create_mem_ref, maybe_fold_tmr): Make the type of
1246 fields of TARGET_MEM_REF sizetype.
1247 (move_fixed_address_to_symbol, move_pointer_to_base):
1249 * tree.def (TARGET_MEM_REF): Add comment on types of
1252 2007-01-11 Joseph Myers <joseph@codesourcery.com>
1254 * c-common.c (vector_types_convertible_p): Treat opaque types as
1255 always convertible if they have the same size, but not otherwise.
1257 2007-01-11 Steven Bosscher <steven@gcc.gnu.org>
1259 * ifcvt.c (struct noce_if_info): Add comments to the fields.
1260 Remove the b_unconditional field.
1261 (noce_try_sign_mask): Do not look at b_unconditional.
1262 (noce_process_if_block): Do not use merge_if_blocks. Update
1263 the CFG here. Do not set b_unconditional.
1264 (cond_move_process_if_block): Likewise.
1265 (find_cond_trap): Likewise.
1266 (check_cond_move_block): Require simple jump insns at the end
1269 2007-01-11 Jan Hubicka <jh@suse.cz>
1271 PR tree-optimization/1046
1272 * tree-tailcall.c (suitable_for_tail_call_opt_p): Use TREE_ADDRESSABLE
1273 when alias info is not ready.
1274 (pass_tail_recursion): Do not require aliasing.
1275 * tree-ssa-copyrename.c (pass_rename_ssa_cop): Likewise.
1276 * tree-ssa-ccp.c (pass_ccp, pass_fold_builtins): Likewise.
1277 * tree-ssa-copy.c (pass_copy_prop): Likewise.
1278 * tree-ssa-forwprop.c (pass_forwprop): Likewise.
1279 * tree-ssa-dce.c (pass_dce, pass_dce_loop, pass_cd_dce): Likewise.
1280 * passes.c (init_optimization_passes): Execute rename_ssa_copies,
1281 ccp, forwprop, copy_prop, merge_phi, copy_prop, dce and tail recursion
1283 * tree-ssa-operands.c (add_virtual_operand, get_indirect_ref_operand):
1284 When aliasing is not build, mark statement as volatile.
1286 2007-01-11 Tom Tromey <tromey@redhat.com>
1288 PR preprocessor/15185, PR preprocessor/20989:
1289 * doc/cppopts.texi <-MT>: Update description of algorithm for
1290 computing default target.
1291 <-M, -MD>: Reword "basename" text.
1293 2007-01-11 Roger Sayle <roger@eyesopen.com>
1295 * builtins.c (expand_builtin_pow, expand_builtin_powi,
1296 fold_builtin_cabs, fold_builtin_sqrt, fold_builtin_trunc,
1297 fold_builtin_floor, fold_builtin_ceil, fold_builtin_round,
1298 fold_builtin_int_int_roundingfn, fold_builtin_bitop,
1299 fold_builtin_bswap, real_constp, fold_builtin_pow,
1300 fold_builtin_powi, fold_builtin_signbit, fold_builtin_copysign,
1301 do_mpfr_arg1, do_mpfr_arg2, do_mpfr_arg3, do_mpfr_sincos): Replace
1302 uses of the macro TREE_CONSTANT_OVERFLOW with TREE_OVERFLOW.
1303 * convert.c (convert_to_pointer): Likewise.
1304 * expr.c (highest_pow2_factor, expand_expr_real_1): Likewise.
1305 * fold-const.c (force_fit_type, fold_negate_expr, int_const_binop,
1306 const_binop, fold_convert_const_int_from_int,
1307 fold_convert_const_int_from_real,
1308 fold_convert_const_real_from_real, sign_bit_p,
1309 optimize_minmax_comparison, extract_muldiv_1, fold_div_compare,
1310 fold_sign_changed_comparison, fold_unary, fold_comparison,
1311 fold_binary, multiple_of_p, tree_Expr_non_zero_p,
1312 fold_negate_const, fold_abs_const, fold_not_const): Likewise.
1313 * print-tree.c (print_node_brief, print_node): Likewise.
1314 * stor-layout.c (place_field, layout_type): Likewise.
1315 * tree-chrec.c (keep_cast): Likewise.
1316 * tree.c (build_vector, build_real, build_real_from_int_cst,
1317 build_complex): Likewise.
1319 2007-01-11 Roger Sayle <roger@eyesopen.com>
1321 * tree.h (TREE_CONSTANT_OVERFLOW): Obsolete. For the time being,
1322 treat TREE_CONSTANT_OVERFLOW as a synonym of TREE_OVERFLOW.
1324 2007-01-11 Paolo Bonzini <bonzini@gnu.org>
1326 * configure.ac (strict1_warn): Rename to strict_warn.
1327 (WERROR, --enable-werror, symlink hacks, stage1_cflags,
1328 cc_set_by_configure, quoted_cc_set_by_configure,
1329 stage_prefix_set_by_configure, quoted_stage_prefix_set_by_configure,
1330 all_boot_languages, all_stagestuff): Remove.
1331 (target_list): Remove bootstrap targets.
1332 * Makefile.in (quickstrap): Unconditionally make a synonym of all.
1333 (BOOT_LANGUAGES, STAGE1_CFLAGS, STAGE1_CHECKING,
1334 REMAKEFLAGS, FLAGS_TO_PASS, PREPEND_DOTDOT_TO_RELATIVE_PATHS,
1335 SUBDIR_FLAGS_TO_PASS, WERROR_FLAGS, STRICT2_WARN, LANG_STAGESTUFF,
1336 VOL_FILES, POSTSTAGE1_FLAGS_TO_PASS, STAGE2_FLAGS_TO_PASS,
1337 STAGEPROFILE_FLAGS_TO_PASS, STAGEFEEDBACK_FLAGS_TO_PASS, stage1_build,
1338 stage1_copy, stage2_build, stage2_copy, stageprofile_build,
1339 stageprofile_copy, stage3_build, stage3_copy, stagefeedback_build,
1340 stagefeedback_copy, stage4_build, clean_s1, clean_sw, bootstrap,
1341 bootstrap-lean, bootstrap2, bootstrap2-lean, bootstrap3,
1342 bootstrap3-lean, bootstrap4, bootstrap4-lean, unstage1, unstage2,
1343 unstage3, unstage4, unstageprofile, unstagefeedback, restage, restage2,
1344 restage3, restage4, restageprofile, restagefeedback, bubbleestrap,
1345 cleanstrap, unstrap, restrap, *compare, *compare3, *compare4,
1346 *compare-lean, *compare3-lean, *compare4-lean, stage1-start, stage1,
1347 stage2-start, stage2, stage3-start, stage3, stage4-start, stage4,
1348 stageprofile-start, stageprofile, stagefeedback-start, stagefeedback,
1349 risky-stage1, risky-stage2, risky-stage3, risky-stage4): Remove.
1350 (ORDINARY_FLAGS_TO_PASS): Rename to FLAGS_TO_PASS.
1351 (STAGECOPYSTUFF, STAGEMOVESTUFF): Consolidate into MOSTLYCLEANFILES.
1352 (mostlyclean): Adjust.
1353 (clean, distclean): Don't mention bootstrap stuff.
1354 * configure: Regenerate.
1355 * ada/config-lang.in, cp/config-lang.in, forttran/config-lang.in,
1356 java/config-lang.in, objc/config-lang.in, objcp/config-lang.in,
1357 treelang/config-lang.in (stagestuff): Remove.
1358 * doc/sourcebuild.texi (stage1, stage2, stage3, stage4,
1359 stageprofile, stagefeedback, stagestuff): Remove mention.
1361 2007-01-11 Nick Clifton <nickc@redhat.com>
1363 * config/mcore/predicates.md (mcore_general_movesrc_operand):
1365 (mcore_general_movdst_operand): Do not accept CONST_INTs.
1366 (mcore_arith_K_S_operand): Run the test for the S constraint not
1367 the test for the M constraint.
1368 (mcore_addsub_operand): Do not accept integer values that are
1369 larger than 32 bits.
1370 * config/mcore/mcore.md: Remove unused constraints from split.
1371 (andsi3): Use HOST_WIDE_INT instead of int to hold an INTVAL.
1373 (allocate_stack): Likewise.
1374 * config/mcore/mcore.c (mcore_print_operand): Restrict output of P
1375 operands to 32 bits.
1376 (mcore_const_costs): Use HOST_WIDE_INT instead of int to hold an
1378 (mcore_and_cost, mcore_modify_comparison, const_ok_for_mcore,
1379 mcore_const_ok_for_inline, mcore_const_trick_uses_not,
1380 try_constant_tricks, mcore_num_ones, mcore_num_zeros,
1381 mcore_output_bclri, mcore_output_andn, output_inline_const,
1382 mcore_output_move, mcore_output_movedouble): Likewise.
1383 (mcore_output_cmov): Use CONST_OK_FOR_M and CONST_OK_FOR_N.
1384 (output_inline_const): Likewise.
1385 (output_inline_const): Fix format strings used in sprintf
1387 * config/mcore/mcore-protos.h: Update prototypes for changed
1388 functions in mcore.c.
1389 * config/mcore/mcore.h (CONST_OK_FOR_I): Cast values to
1390 HOST_WIDE_INT and not int.
1391 (CONST_OK_FOR_J, CONST_OK_FOR_K, CONST_OK_FOR_L, CONST_OK_FOR_M,
1392 CONST_OK_FOR_N): Likewise.
1393 (LEGITIMATE_CONSTANT_P): Also check CONSTANT_P.
1394 (GO_IF_LEGITIMATE_INDEX): Use HOST_WIDE_INT instead of int to hold
1397 2007-01-10 Jan Hubicka <jh@suse.cz>
1399 * tree-vrp.c (remove_range_assertions): Release defs.
1400 * tree-ssa-loop-ivopts.c (rmeove_statement): Likewise.
1401 * tree-ssa-dom.c (remove_stmt_or_phi): Likewise.
1403 2007-01-10 Paul Brook <paul@codesourcery.com>
1405 * config/arm/arm.c (arm_rtx_costs_1): Handle mutiply-subtract.
1406 * config/arm/arm.md (mulsi3subsi): New insn.
1408 2007-01-10 Zdenek Dvorak <dvorakz@suse.cz>
1410 * tree-ssa-loop-manip.c (tree_unroll_loop): Make it a wrapper over ...
1411 (tree_transform_and_unroll_loop): New.
1412 * tree-flow.h (transform_callback, tree_transform_and_unroll_loop):
1415 2007-01-10 Robert Kennedy <jimbob@google.com>
1417 * fold-const.c (fold_comparison): Fold comparisons like (x *
1418 1000 < 0) to (x < 0).
1420 2007-01-10 Ian Lance Taylor <iant@google.com>
1422 * tree-pretty-print.c (dump_generic_node): Print parentheses when
1423 operands have the same priority.
1425 2007-01-10 Tom Tromey <tromey@redhat.com>
1427 * fold-const.c (fold_truthop): Don't check can_use_bit_fields_p.
1428 (fold_binary): Likewise.
1429 * langhooks.c (lhd_can_use_bit_fields_p): Removed.
1430 * langhooks-def.h (lhd_can_use_bit_fields_p): Removed.
1431 (LANG_HOOKS_CAN_USE_BIT_FIELDS_P): Removed.
1432 (LANG_HOOKS_INITIALIZER): Remove LANG_HOOKS_CAN_USE_BIT_FIELDS_P.
1433 * langhooks.h (struct lang_hooks): Removed field
1434 'can_use_bit_fields_p'.
1436 2007-01-10 Ralf Corsépius <ralf.corsepius@rtems.org>
1438 * config/bfin/t-bfin, config/bfin/t-bfin-elf: Remove GCC_CFLAGS.
1440 2007-01-10 Razya Ladelsky <razya@il.ibm.com>
1442 * function.c (get_last_funcdef_no): New function.
1443 * function.h (get_last_funcdef_no): Declare.
1444 * tree-inline.c (initialize_cfun): Add initialization.
1445 (tree_function_versioning): Cleanup.
1447 2007-01-10 Jan Hubicka <jh@suse.cz>
1449 * tree-inline.c (setup_one_parameter): Do not propagate into abnormal
1452 2007-01-10 Sa Liu <saliu@de.ibm.com>
1453 Ben Elliston <bje@au.ibm.com>
1455 * spu.h (STACK_SAVE_AREA): Use VOIDmode for SAVE_FUNCTION, SImode
1456 for SAVE_NONLOCAL and Pmode for any other save level.
1457 * spu-protos.h (spu_restore_stack_block): Declare.
1458 * spu.md (save_stack_block): Remove.
1459 (restore_stack_block): Call spu_restore_stack_block.
1460 * spu.c (spu_restore_stack_block): New function.
1461 (spu_expand_epilogue): Remove old comment.
1463 2007-01-09 Zdenek Dvorak <dvorakz@suse.cz>
1465 PR tree-optimization/30322
1466 * tree-ssa-loop-ivopts.c (fold_affine_expr, iv_value): Removed.
1467 (cand_value_at): Return the value as aff_tree.
1468 (may_eliminate_iv): Convert the bound from aff_tree to tree.
1469 * tree-affine.c (aff_combination_add_cst, aff_combination_add_product,
1470 aff_combination_mult): New functions.
1471 (aff_combination_add): Use aff_combination_add_cst.
1472 (aff_combination_convert): Allow conversions to a wider type.
1473 (tree_to_aff_combination): Handle BIT_NOT_EXPR.
1474 * tree-affine.h (aff_combination_mult): Declare.
1476 2007-01-09 Carlos O'Donell <carlos@codesourcery.com>
1478 * doc/tm.texi: Update documentation to reflect reality of exec
1479 and start file search behaviours. Update copyright year.
1480 * doc/invoke.texi: Explain how GCC_EXEC_PREFIX is used to find
1481 header file directories.
1483 2007-01-09 Uros Bizjak <ubizjak@gmail.com>
1485 * config/i386/i386.md (*sinxf2): Rename to *sinxf2_i387.
1486 (*cosxf2): Rename to cosxf2_i387.
1487 (*sindf2, *sinsf2): Extend operand 1 to XFmode. Macroize patterns
1488 using X87MODEF12 mode macro. Rename patterns to
1489 *sin_extend<mode>xf2_i387. Use SSE_FLOAT_MODE_P to disable patterns
1491 (*cosdf2, *cossf2): Ditto.
1492 (sincosdf3, sincossf3): Ditto. Rewrite corresponding splitters
1493 to match extended input operands.
1494 (sincos<mode>3): New expander.
1495 (*sinextendsfdf2, *cosextendsfdf2, *sincosextendsfdf3): Remove
1496 insn patterns and corresponding splitters.
1498 2007-01-09 Kaz Kojima <kkojima@gcc.gnu.org>
1500 * config/sh/t-linux (TARGET_LIBGCC2_CFLAGS): Delete.
1501 (SHLIB_MAPFILES, SHLIB_LINK, SHLIB_INSTALL): Likewise.
1503 2007-01-09 Nicolas Pitre <nico@cam.org>
1506 * arm/ieee754-df.S (Lad_s): Also test the low word of X for zero.
1508 2007-01-08 Geoffrey Keating <geoffk@apple.com>
1510 * target.h (struct gcc_target): New field library_rtti_comdat.
1511 * target-def.h (TARGET_CXX_LIBRARY_RTTI_COMDAT): New.
1512 (TARGET_CXX): Add TARGET_CXX_LIBRARY_RTTI_COMDAT.
1513 * doc/tm.texi (C++ ABI): Document TARGET_CXX_LIBRARY_RTTI_COMDAT.
1514 * config/darwin.h (TARGET_CXX_LIBRARY_RTTI_COMDAT): Define.
1516 2007-01-08 Geoffrey Keating <geoffk@apple.com>
1518 * doc/invoke.texi (Optimize Options): Correct description of -O0.
1520 2007-01-08 Richard Guenther <rguenther@suse.de>
1522 * tree.h (force_fit_type_double): Export.
1523 (force_fit_type): Remove.
1524 * fold-const.c (force_fit_type_double): New function.
1525 (force_fit_type): Remove.
1526 (int_const_binop): Use it.
1527 (fold_convert_const_int_from_int): Likewise.
1528 (fold_convert_const_int_from_real): Likewise.
1529 (fold_div_compare): Likewise.
1530 (fold_sign_changed_comparison): Likewise.
1531 (fold_unary): Likewise.
1532 (fold_negate_const): Likewise.
1533 (fold_abs_const): Likewise.
1534 (fold_not_const): Likewise.
1535 * c-common.c (shorten_compare): Use force_fit_type_double.
1536 * convert.c (convert_to_pointer): Likewise.
1538 2007-01-08 Richard Guenther <rguenther@suse.de>
1540 * tree.h (build_int_cst_wide_type): Export.
1541 * tree.c (build_int_cst_wide_type): New function.
1542 (build_int_cst_wide): Fix comment.
1543 * builtins.c (fold_builtin_object_size): Use build_int_cst
1544 to build -1 or 0 of the correct type. Use fit_double_type
1545 to check for overflow.
1546 * fold-const.c (optimize_bit_field_compare): Use build_int_cst_type
1548 (decode_field_reference): Likewise.
1549 (all_ones_mask_p): Likewise.
1550 (native_interpret_int): Use build_int_cst_wide_type.
1551 (fold_binary): Use build_int_cst_type to build an all-ones
1553 * stor-layout.c (set_sizetype): Use build_int_cst_wide_type.
1555 2007-01-08 Daniel Jacobowitz <dan@codesourcery.com>
1557 * config/pa/t-pa64 (libgcc_stub.a): Use $(T).
1559 2007-01-09 Ben Elliston <bje@au.ibm.com>
1561 * genautomata.c (STATS_OPTION): New option.
1562 (stats_flag): New flag.
1563 (gen_automata_option): Handle it.
1564 (initiate_automaton_gen): Ditto.
1565 (write_automata): Output statistics only if stats_flag is
1566 set. Likewise, output time statistics only if time_flag is set.
1567 * doc/md.texi (Processor pipeline description): Document new flag.
1569 2007-01-08 Richard Guenther <rguenther@suse.de>
1571 * builtins.c (fold_builtin_int_roundingfn): Use fit_double_type.
1572 * tree.c (build_int_cst_type): Likewise.
1573 (size_in_bytes): Don't call force_fit_type on the result.
1574 (int_fits_type_p): Use fit_double_type.
1575 * fold-const.c (fit_double_type): New function.
1576 (force_fit_type): Use it.
1577 * tree.h (fit_double_type): Export.
1579 2007-01-08 Jan Hubicka <jh@suse.cz>
1581 * tree-vectorizer.c (gate_increase_alignment): Fix return type.
1582 * ipa.c (function_and_variable_visibility): Fix return type.
1584 2007-01-08 Richard Guenther <rguenther@suse.de>
1586 * tree-ssa-ccp.c (maybe_fold_offset_to_array_ref): Use type
1587 of offset to build the index.
1588 * tree-pretty-print.c (dump_generic_node): Don't build negated
1589 const just for printing.
1590 * c-pretty-print.c (pp_c_integer_constant): Likewise.
1591 * builtins.c (fold_builtin_int_roundingfn): Check if result
1592 fits the type by using force_fit_type and comparing the result.
1593 * predict.c (predict_loops): Use compare_tree_int for comparison.
1594 * tree.c (build_int_cst): Fall back to integer_type_node for
1596 (build_int_cst_wide): Assert type is non-null.
1598 2007-01-08 Roberto Costa <roberto.costa@st.com>
1600 * tree-vrp.c (extract_range_from_cond_expr): New.
1601 (extract_range_from_expr): Handle COND_EXPR nodes used as expressions.
1602 * tree-ssa-ccp.c (get_maxval_strlen): Handle COND_EXPR nodes used
1604 (fold_stmt): Bug fix, avoid infinite recursion when folding COND_EXPRs.
1605 * tree-ssa-forwprop.c (simplify_cond, forward_propagate_into_cond,
1606 tree_ssa_forward_propagate_single_use_vars): Handle COND_EXPR nodes
1607 used as expressions.
1608 * tree-object-size.c (cond_expr_object_size): New.
1609 (collect_object_sizes_for): Handle COND_EXPR nodes used as expressions.
1611 2007-01-08 Jan Hubicka <jh@suse.cz>
1613 * tree-ssa-forwprop.c (forward_propagate_into_cond,
1614 tree_ssa_forward_propagate_single_use_va): Release defs of propagated
1617 2007-01-08 Richard Guenther <rguenther@suse.de>
1619 PR tree-optimization/23603
1620 * tree-vrp.c (set_value_range_to_truthvalue): New function.
1621 (extract_range_from_binary): Fall back to truthvalue instead of
1622 varying for TRUTH_*_EXPR.
1623 (extract_range_from_comparison): Fall back to truthvalue instead of
1625 (vrp_visit_phi_node): Don't adjust new range bounds to +INF/-INF
1626 if all visited PHI values were constant.
1628 2007-01-08 Jan Hubicka <jh@suse.cz>
1630 * cgraphunit.c (cgraph_process_new_functions): Reset reachable flag.
1631 (cgraph_analyze_function): break out from ...
1632 (cgraph_finalize_compilation_unit): ... here.
1633 (cgraph_expand_function): Remove forgoten commented out line.
1634 (cgraph_optimize): Analyze functions.
1636 2007-01-08 Jan Hubicka <jh@suse.cz>
1638 * tree-pas.h (TODO_remove_function): New flag.
1639 (TODO_update*): Renumber.
1640 (pass_ipa_increase_alignment,
1641 pass_ipa_function_and_variable_visibility): New passes.
1642 * cgraphunit.c (cgraph_increase_alignment): Move to tree-vectorizer.c
1643 (cgraph_function_and_variable_visibility): Move to ipa.c
1644 (cgraph_optimize): Don't call cgraph_function_and_variable_visibility,
1645 cgraph_increase_alignment.
1646 * ipa-inline.c (cgraph_decide_inlining): Don't push timevar.
1647 (cgraph_decide_inlining_incrementally): Push TV_INTEGRATION before
1648 calling tree-inline.
1649 (cgraph_early_inlining): Do not call cgraph_remove_unreachable_nodes.
1650 (pass_ipa_inline, pass_early_ipa_inlining): Set TODO_remove_functions
1651 * tree-vectorizer.c (increase_alignment): Move here from cgraphunit.c
1652 (gate_increase_alignment): New function.
1653 (pass_ipa_increase_alignment): New pass.
1654 * ipa.c: Inline tree-pass.h and timevar.h
1655 (function_and_variable_visibility): Move here from cgraphunit.c
1656 * tree-optimize.c (pass_early_local_passes): Add TODO_remove_functions.
1657 * passes.c (init_optimization_passes): Add the two new passes.
1658 (execute_todo): Handle cgraph_remove_functions.
1660 2007-01-08 Nick Clifton <nickc@redhat.com>
1662 * config/frv/predicates.md (reg_or_0_operand): Accept
1665 2007-01-08 Ralf Corsépius <ralf.corsepius@rtems.org>
1667 * config/bfin/rtems.h, config/bfin/t-rtems: New.
1668 * config.gcc: Add bfin*-rtems*.
1670 2007-01-08 Mark Shinwell <shinwell@codesourcery.com>
1672 * c.opt: Add -flax-vector-conversions.
1673 * c-typeck.c (convert_for_assignment): Pass flag to
1674 vector_types_convertible_p to allow emission of note.
1675 (digest_init): Likewise.
1676 * c-opts.c: Handle -flax-vector-conversions.
1677 * c-common.c (flag_lax_vector_conversions): New.
1678 (vector_types_convertible_p): Unless -flax-vector conversions
1679 has been passed, disallow conversions between vectors with
1680 differing numbers of subparts and/or element types. If such
1681 a conversion is disallowed, possibly emit a note on the first
1682 occasion only to inform the user of -flax-vector-conversions.
1683 The new last argument specifies this.
1684 * c-common.h (flag_lax_vector_conversions): New.
1685 (vector_types_convertible_p): Add extra argument.
1686 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use
1687 char_type_node for V*QI type vectors.
1688 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
1689 Update to satisfy new typechecking rules.
1690 * config/rs6000/altivec.h (vec_cmple): Use vec_cmpge, for both
1692 * doc/invoke.texi (C Dialect Options): Document
1693 -flax-vector-conversions.
1695 2007-01-08 Mark Shinwell <shinwell@codesourcery.com>
1697 PR tree-optimization/29877
1698 * tree-ssa-ter.c (is_replaceable_p): Deem assignments with
1699 a register variable on the RHS to not be replaceable.
1701 2007-01-08 Chen Liqin <liqin@sunnorth.com.cn>
1702 * config/score/t-score-elf (MULTILIB_OPTIONS): Change.
1703 * config/score/predicates.md (const_uimm5, sr0_operand, const_simm12,
1704 const_simm15, const_pow2, const_npow2): Added.
1705 * config/score/misc.md (insv, extv, extzv, movmemsi,
1706 move_lbu_a/b, mov_lhu_a/b etc): Added and fix some bug.
1707 * config/score/score.c (score_address_cost, score_select_cc_mode):
1709 Change CONST_OK_FOR_LETTER_P/EXTRA_CONSTRAINT define.
1710 Update score_rtx_costs for MACRO TARGET_RTX_COSTS.
1711 Update score_print_operand.
1712 * config/score/score.h (DATA_ALIGNMENT, SELECT_CC_MODE): Added.
1713 Adjust register allocate order and update some macro define.
1714 * config/score/score-mdaux.c (mdx_unaligned_load, mdx_unsigned_store,
1715 mdx_block_move_straight, mdx_block_move_loop_head,
1716 mdx_block_move_loop_body, mdx_block_move_loop_foot, mdx_block_move_loop,
1717 mdx_block_move): Added.
1718 (mdx_movsicc, mdp_select_add_imm, mdp_select, mds_zero_extract_andi,
1719 mdp_limm): Updated and fix some bug and typo.
1720 * config/score/score.md (movqi/hi/si, add/sub/zero/ext): Updated.
1721 (movsf, movdf, doloop_end): Added.
1723 2007-01-08 Kazu Hirata <kazu@codesourcery.com>
1725 * config/arm/arm.c, config/arm/arm.h, config/arm/arm.md,
1726 config/arm/thumb2.md: Fix comment typos.
1727 * doc/extend.texi: Fix a typo.
1729 2007-01-07 Eric Christopher <echristo@apple.com>
1731 * configure.ac: Check for __stack_chk_fail for darwin.
1732 * configure: Regenerate.
1734 2007-01-07 Richard Guenther <rguenther@suse.de>
1736 * tree-vrp.c (extract_range_from_assert): CSE calls to
1737 compare_values where possible.
1738 (extract_range_from_unary_expr): Likewise.
1740 2007-01-07 Anatoly Sokolov <aesok@post.ru>
1742 * config/avr/avr-protos.h (call_insn_operand): Delete prototype.
1743 * config/avr/avr.c (call_insn_operand): Delete function.
1744 * config/avr/avr.md (*pushqi, *pushhi, *pushsi, *pushsf): Use REG_SP
1745 instead of register number. Use predicates.
1746 * config/avr/predicates.md (const0_operand, reg_or_0_operand,
1747 call_insn_operand): Add.
1749 2007-01-06 Jan Hubicka <jh@suse.cz>
1751 * tree-pass.h (pass_build_cgraph_edges): Declare.
1752 * cgraphunit.c (record_refernece): Move to cgraphbuild.c
1753 (visited_nodes): Remove.
1754 (cgraph_create_edges): Move to cgraphbuild.c; rename to
1755 build_cgrpah_edges; make visited_nodes local.
1756 (cgraph_process_new_functions): DO not call initialize_inline_failed.
1757 (record_references_in_initializer): Move to cgraphbuild.c
1758 (initialize_inline_failed, rebuild_cgraph_edges,
1759 pass_rebuild_cgraph_edges): Move to cgraphbuild.c.
1760 (verify_cgraph_node): Make visited_nodes local.
1761 (cgraph_analyze_function): Do not call cgraph_create_edges and
1762 initialize_inline_failed.
1763 (cgraph_expand_function): Do not call cgraph_lower_function;
1764 assert that function is already lowered.
1765 * Makefile.in (cgraphbuild.o): New.
1766 * passes.c (init_optimization_passes): Add pass_build_cgraph_edges
1767 at the end of lowering passes.
1769 2007-01-06 Steven Bosscher <steven@gcc.gnu.org>
1771 * ifcvt.c (cond_move_convert_if_block): New function, code
1772 factored out from...
1773 (cond_move_process_if_block): ...here. Call the new function
1774 on the THEN and ELSE blocks.
1775 (merge_if_block): Do not copy global_live_at_end, merge_blocks
1776 already takes care of this.
1778 2007-01-05 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
1781 * tree.h (TREE_OVERFLOW_P): New.
1782 * c-typeck.c (parser_build_unary_op): Warn only if result
1783 overflowed and operands did not.
1784 (parser_build_binary_op): Likewise.
1785 (convert_for_assignment): Remove redundant overflow_warning.
1786 * c-common.c (overflow_warning): Don't check or set TREE_OVERFLOW.
1788 2007-01-05 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
1790 * c-typeck.c (store_init_value): Split over two lines to follow
1791 the GNU coding style.
1793 2007-01-05 Benjamin Kosnik <bkoz@redhat.com>
1795 * c-cppbuiltin.c (c_cpp_builtins): __GXX_EXPERIMENTAL_CPP0X__ to
1796 __GXX_EXPERIMENTAL_CXX0X__.
1797 * doc/cpp.texi: Same.
1799 2007-01-05 Richard Guenther <rguenther@suse.de>
1802 * tree.c (get_narrower): Do not construct COMPONENT_REFs
1803 with mismatched types. Instead explicitly build a
1804 conversion NOP_EXPR.
1806 2007-01-05 Ian Lance Taylor <iant@google.com>
1808 * c-common.c (decl_with_nonnull_addr_p): New function.
1809 (c_common_truthvalue_conversion): Call it.
1810 * c-typeck.c (build_binary_op): Likewise.
1811 * c-common.h (decl_with_nonnull_addr_p): Declare.
1813 2007-01-05 Jakub Jelinek <jakub@redhat.com>
1816 * libgcc2.c (__divdc3): Compare c and d against 0.0 instead of
1819 2007-01-05 Joel Brobecker <brobecker@adacore.com>
1821 * doc/install.texi (Final install): Document the fact that
1822 the GNAT runtime should not be stripped.
1824 2007-01-04 Jan Hubicka <jh@suse.cz>
1826 * tree-inline.c (fold_marked_statements): Update operand caches
1827 and EH after folding
1829 2007-01-04 Ian Lance Taylor <iant@google.com>
1831 * c-common.c (check_function_nonnull): Whitespace fix.
1833 2007-01-04 Jan Hubicka <jh@suse.cz>
1835 * tree-optimize.c (execute_fixup_cfg): Correct previously mistakely
1836 comitted older version of patch.
1837 (pass_fixup_cfg): Add TODOs to verify flow and statements, dump
1838 function, celanup cfg and collect garbage.
1840 2007-01-04 Mike Stump <mrs@apple.com>
1842 * Makefile.in (mostlyclean): Don't remove libgcc anymore.
1845 2007-01-04 Eric Christopher <echristo@apple.com>
1847 * libgcc2.c (__bswapsi2): Use SItype.
1848 (__bswapdi2): Use DItype.
1849 * libgcc2.h: Update for above.
1851 2007-01-04 Paul Brook <paul@codesourcery.com>
1853 * config/arm/arm.md (arm_mulsi3, thumb_mulsi3, mulsi3_compare0,
1854 mulsi_compare0_scratch, mulsi3addsi, mulsi3addsi_compare0,
1855 mulsi3addsi_compare0_scratch, mulsidi3adddi, mulsidi3,
1856 umulsidi3, umulsidi3adddi, smulsi3_highpart,
1857 umulsi3_highpart): Make conditional on !arm_arch6.
1858 (arm_mulsi3_v6, thumb_mulsi3_v6, mulsi3_compare0_v6,
1859 mulsi_compare0_scratch_v6, mulsi3addsi_v6, mulsi3addsi_compare0_v6,
1860 mulsi3addsi_compare0_scratch_v6, mulsidi3adddi_v6, mulsidi3_v6,
1861 umulsidi3_v6, umulsidi3adddi_v6, smulsi3_highpart_v6,
1862 umulsi3_highpart_v6): New insns.
1864 2007-01-04 Roger Sayle <roger@eyesopen.com>
1866 * fold-const.c (fold_convert): When casting an expression to void,
1867 fold_ignored_result may discover a GIMPLE_MODIFY_STMT which doesn't
1868 have a type. Instead of attempting to build a NOP_EXPR, return
1869 these "special" trees directly.
1871 2007-01-04 Joseph Myers <joseph@codesourcery.com>
1873 * config/rs6000/rs6000.c (rs6000_rtx_costs): Make adjustment for
1874 MULT inside MINUS as either argument. Use rs6000_cost->dmul -
1875 rs6000_cost->fp not 0 as adjustment for outer NEG.
1877 2007-01-04 Jan Hubicka <jh@suse.cz>
1879 * cgraph.c (cgraph_release_function_body): New function.
1880 (cgraph_remove_node): Use it.
1881 * cgraph.h (cgraph_release_function_body): Declare.
1882 * cgraphunit.c (cgraph_expand_function): Use it.
1883 * ipa.c (cgraph_remove_unreahchable_nodes): Use it.
1884 * tree-ssa.c (delete_tree_ssa): Allow to be called before aliasing
1885 is initialized and while compilation of other function is running.
1886 * tree-optimize.c (execute_free_cfg_annotations): Move code to clear
1887 statement CFG annotations from here to ...
1888 * tree-cfg.c (delete_tree_cfg_annotations): ... here.
1890 2007-01-04 Zdenek Dvorak <dvorakz@suse.cz>
1892 * cfgloop.h (enum li_flags): Make the constants powers of two.
1894 2007-01-04 Jan Hubicka <jh@suse.cz>
1896 * tree-inline.c (copy_bb): Insert new statements to statements_to_fold
1898 (fold_marked_statements): New function.
1899 (optimize_inline_calls, tree_function_versioning): Fold new statements.
1900 * tree-inline.h (copy_body_data): Add statements_to_fold.
1902 2007-01-03 Daniel Jacobowitz <dan@codesourcery.com>
1904 * config.gcc: Mention libgcc/config.host.
1905 * Makefile.in: Update comments mentioning libgcc.
1906 (LIBGCC, INSTALL_LIBGCC, GCC_PARTS, mklibgcc): Delete.
1907 (all.cross, start.encap, rest.encap, rest.cross): Update
1908 dependencies for libgcc move.
1909 (libgcc.mk, LIBGCC_DEPS, libgcov.a, libgcc.a, stmp-multilib)
1910 (clean-target, clean-target-libgcc): Delete.
1911 (srcdirify, GCC_EXTRA_PARTS): New macros.
1912 (libgcc-support, libgcc.mvars): New rules.
1913 (distclean): Remove mention of mklibgcc.
1914 (install): Don't reference INSTALL_LIBGCC.
1915 (install-common): Don't reference EXTRA_PARTS.
1916 (install-libgcc, install-multilib): Delete rules.
1917 * mklibgcc.in: Delete file.
1918 * doc/configfiles.texi: Don't mention mklibgcc.
1920 * config/i386/t-darwin (SHLIB_VERPFX): Delete (moved to libgcc).
1921 * config/i386/t-darwin64 (SHLIB_VERPFX): Likewise.
1922 * config/rs6000/t-darwin (SHLIB_VERPFX): Likewise.
1923 * config/rs6000/t-ppccomm (TARGET_LIBGCC2_CFLAGS, SHLIB_MAPFILES)
1924 (mklibgcc, ldblspecs): Likewise.
1926 * config/i386/t-nwld (libgcc.def, libc.def, libpcre.def)
1927 (posixpre.def): Use $(T).
1928 (SHLIB_EXT, SHLIB_NAME, SHLIB_SLIBDIR_QUAL, SHLIB_DEF, SHLIB_MAP)
1929 (SHLIB_SRC, SHLIB_INSTALL): Delete.
1930 (SHLIB_LINK): Make dummy.
1931 * config/t-slibgcc-darwin: Delete contents except for dummy SHLIB_LINK.
1933 * config/frv/t-linux (EXTRA_MULTILIB_PARTS): Clear.
1935 * config/alpha/t-crtfm: Use $(T) in rules for EXTRA_PARTS.
1936 * config/alpha/t-vms, config/alpha/t-vms64, config/fr30/t-fr30,
1937 config/i386/t-rtems-i386, config/ia64/t-ia64, config/rs6000/t-beos,
1938 config/rs6000/t-newas, config/sparc/t-elf: Likewise.
1940 * configure.ac (all_outputs): Remove mklibgcc.
1941 * configure: Regenerated.
1943 2007-01-03 Josh Conner <jconner@apple.com>
1946 * calls.c (compute_argument_addresses): Set stack and stack_slot
1947 for partial args, too.
1948 (store_one_arg): Use locate.size.constant for the size when
1949 generating a save_area.
1951 2007-01-03 Robert Kennedy <jimbob@google.com>
1953 * tree-cfg.c (tree_merge_blocks): Release SSA_NAME phi results
1954 whose definitions are deleted due to basic block merging.
1956 2007-01-03 Paul Brook <paul@codesourcery.com>
1959 * config/arm/arm.c (output_return_instruction): Pop PC in interrupt
1961 (use_return_insn): Return 0 for Thumb interrupt functions.
1962 (print_multi_reg): Add rfe argument for IRQ returns.
1963 (arm_output_epilogue): Pop interrupt return address directly into PC.
1964 (arm_expand_prologue): Only adjust IRQ return address in Arm mode.
1966 2007-01-03 Paul Brook <paul@codesourcery.com>
1968 Merge from sourcerygxx-4_1.
1969 * config/arm/thumb2.md: New file.
1970 * config/arm/elf.h (JUMP_TABLES_IN_TEXT_SECTION): Return True for
1972 * config/arm/coff.h (JUMP_TABLES_IN_TEXT_SECTION): Ditto.
1973 * config/arm/aout.h (ASM_OUTPUT_ADDR_VEC_ELT): Add !Thumb-2 assertion.
1974 (ASM_OUTPUT_ADDR_DIFF_ELT): Output Thumb-2 jump tables.
1975 * config/arm/aof.h (ASM_OUTPUT_ADDR_DIFF_ELT): Output Thumb-2 jump
1977 (ASM_OUTPUT_ADDR_VEC_ELT): Add !Thumb-2 assertion.
1978 * config/arm/ieee754-df.S: Use macros for Thumb-2/Unified asm
1980 * config/arm/ieee754-sf.S: Ditto.
1981 * config/arm/arm.c (thumb_base_register_rtx_p): Rename...
1982 (thumb1_base_register_rtx_p): ... to this.
1983 (thumb_index_register_rtx_p): Rename...
1984 (thumb1_index_register_rtx_p): ... to this.
1985 (thumb_output_function_prologue): Rename...
1986 (thumb1_output_function_prologue): ... to this.
1987 (thumb_legitimate_address_p): Rename...
1988 (thumb1_legitimate_address_p): ... to this.
1989 (thumb_rtx_costs): Rename...
1990 (thumb1_rtx_costs): ... to this.
1991 (thumb_compute_save_reg_mask): Rename...
1992 (thumb1_compute_save_reg_mask): ... to this.
1993 (thumb_final_prescan_insn): Rename...
1994 (thumb1_final_prescan_insn): ... to this.
1995 (thumb_expand_epilogue): Rename...
1996 (thumb1_expand_epilogue): ... to this.
1997 (arm_unwind_emit_stm): Rename...
1998 (arm_unwind_emit_sequence): ... to this.
1999 (thumb2_legitimate_index_p, thumb2_legitimate_address_p,
2000 thumb1_compute_save_reg_mask, arm_dwarf_handle_frame_unspec,
2001 thumb2_index_mul_operand, output_move_vfp, arm_shift_nmem,
2002 arm_save_coproc_regs, thumb_set_frame_pointer, arm_print_condition,
2003 thumb2_final_prescan_insn, thumb2_asm_output_opcode, arm_output_shift,
2004 thumb2_output_casesi): New functions.
2005 (TARGET_DWARF_HANDLE_FRAME_UNSPEC): Define.
2006 (FL_THUMB2, FL_NOTM, FL_DIV, FL_FOR_ARCH6T2, FL_FOR_ARCH7,
2007 FL_FOR_ARCH7A, FL_FOR_ARCH7R, FL_FOR_ARCH7M, ARM_LSL_NAME,
2008 THUMB2_WORK_REGS): Define.
2009 (arm_arch_notm, arm_arch_thumb2, arm_arch_hwdiv, arm_condexec_count,
2010 arm_condexec_mask, arm_condexec_masklen)): New variables.
2011 (all_architectures): Add armv6t2, armv7, armv7a, armv7r and armv7m.
2012 (arm_override_options): Check new CPU capabilities.
2013 Set new architecture flag variables.
2014 (arm_isr_value): Handle v7m interrupt functions.
2015 (user_return_insn): Return 0 for v7m interrupt functions. Handle
2017 (const_ok_for_arm): Handle Thumb-2 constants.
2018 (arm_gen_constant): Ditto. Use movw when available.
2019 (arm_function_ok_for_sibcall): Return false for v7m interrupt
2021 (legitimize_pic_address, arm_call_tls_get_addr): Handle Thumb-2.
2022 (thumb_find_work_register, arm_load_pic_register,
2023 legitimize_tls_address, arm_address_cost, load_multiple_sequence,
2024 emit_ldm_seq, emit_stm_seq, arm_select_cc_mode, get_jump_table_size,
2025 print_multi_reg, output_mov_long_double_fpa_from_arm,
2026 output_mov_long_double_arm_from_fpa, output_mov_double_fpa_from_arm,
2027 output_mov_double_fpa_from_arm, output_move_double,
2028 arm_compute_save_reg_mask, arm_compute_save_reg0_reg12_mask,
2029 output_return_instruction, arm_output_function_prologue,
2030 arm_output_epilogue, arm_get_frame_offsets, arm_regno_class,
2031 arm_output_mi_thunk, thumb_set_return_address): Ditto.
2032 (arm_expand_prologue): Handle Thumb-2. Use arm_save_coproc_regs.
2033 (arm_coproc_mem_operand): Allow POST_INC/PRE_DEC.
2034 (arithmetic_instr, shift_op): Use arm_shift_nmem.
2035 (arm_print_operand): Use arm_print_condition. Handle '(', ')', '.',
2037 (arm_final_prescan_insn): Use extract_constrain_insn_cached.
2038 (thumb_expand_prologue): Use thumb_set_frame_pointer.
2039 (arm_file_start): Output directive for unified syntax.
2040 (arm_unwind_emit_set): Handle stack alignment instruction.
2041 * config/arm/lib1funcs.asm: Remove default for __ARM_ARCH__.
2042 Add v6t2, v7, v7a, v7r and v7m.
2043 (RETLDM): Add Thumb-2 code.
2044 (do_it, shift1, do_push, do_pop, COND, THUMB_SYNTAX): New macros.
2045 * config/arm/arm.h (TARGET_CPU_CPP_BUILTINS): Define __thumb2__.
2046 (TARGET_THUMB1, TARGET_32BIT, TARGET_THUMB2, TARGET_DSP_MULTIPLY,
2047 TARGET_INT_SIMD, TARGET_UNIFIED_ASM, ARM_FT_STACKALIGN, IS_STACKALIGN,
2048 THUMB2_TRAMPOLINE_TEMPLATE, TRAMPOLINE_ADJUST_ADDRESS,
2049 ASM_OUTPUT_OPCODE, THUMB2_GO_IF_LEGITIMATE_ADDRESS,
2050 THUMB2_LEGITIMIZE_ADDRESS, CASE_VECTOR_PC_RELATIVE,
2051 CASE_VECTOR_SHORTEN_MODE, ADDR_VEC_ALIGN, ASM_OUTPUT_CASE_END,
2052 ADJUST_INSN_LENGTH): Define.
2053 (TARGET_REALLY_IWMMXT, TARGET_IWMMXT_ABI, CONDITIONAL_REGISTER_USAGE,
2054 STATIC_CHAIN_REGNUM, HARD_REGNO_NREGS, INDEX_REG_CLASS,
2055 BASE_REG_CLASS, MODE_BASE_REG_CLASS, SMALL_REGISTER_CLASSES,
2056 PREFERRED_RELOAD_CLASS, SECONDARY_OUTPUT_RELOAD_CLASS,
2057 SECONDARY_INPUT_RELOAD_CLASS, LIBCALL_VALUE, FUNCTION_VALUE_REGNO_P,
2058 TRAMPOLINE_SIZE, INITIALIZE_TRAMPOLINE, HAVE_PRE_INCREMENT,
2059 HAVE_POST_DECREMENT, HAVE_PRE_DECREMENT, HAVE_PRE_MODIFY_DISP,
2060 HAVE_POST_MODIFY_DISP, HAVE_PRE_MODIFY_REG, HAVE_POST_MODIFY_REG,
2061 REGNO_MODE_OK_FOR_BASE_P, LEGITIMATE_CONSTANT_P,
2062 REG_MODE_OK_FOR_BASE_P, REG_OK_FOR_INDEX_P, GO_IF_LEGITIMATE_ADDRESS,
2063 LEGITIMIZE_ADDRESS, THUMB2_LEGITIMIZE_ADDRESS,
2064 GO_IF_MODE_DEPENDENT_ADDRESS, MEMORY_MOVE_COST, BRANCH_COST,
2065 ASM_APP_OFF, ASM_OUTPUT_CASE_LABEL, ARM_DECLARE_FUNCTION_NAME,
2066 FINAL_PRESCAN_INSN, PRINT_OPERAND_PUNCT_VALID_P,
2067 PRINT_OPERAND_ADDRESS): Adjust for Thumb-2.
2068 (arm_arch_notm, arm_arch_thumb2, arm_arch_hwdiv): New declarations.
2069 * config/arm/arm-cores.def: Add arm1156t2-s, cortex-a8, cortex-r4 and
2071 * config/arm/arm-tune.md: Regenerate.
2072 * config/arm/arm-protos.h: Update prototypes.
2073 * config/arm/vfp.md: Enable patterns for Thumb-2.
2074 (arm_movsi_vfp): Add movw alternative. Use output_move_vfp.
2075 (arm_movdi_vfp, movsf_vfp, movdf_vfp): Use output_move_vfp.
2076 (thumb2_movsi_vfp, thumb2_movdi_vfp, thumb2_movsf_vfp,
2077 thumb2_movdf_vfp, thumb2_movsfcc_vfp, thumb2_movdfcc_vfp): New.
2078 * config/arm/libunwind.S: Add Thumb-2 code.
2079 * config/arm/constraints.md: Update include Thumb-2.
2080 * config/arm/ieee754-sf.S: Add Thumb-2/Unified asm support.
2081 * config/arm/ieee754-df.S: Ditto.
2082 * config/arm/bpabi.S: Ditto.
2083 * config/arm/t-arm (MD_INCLUDES): Add thumb2.md.
2084 * config/arm/predicates.md (low_register_operand,
2085 low_reg_or_int_operand, thumb_16bit_operator): New.
2086 (thumb_cmp_operand, thumb_cmpneg_operand): Rename...
2087 (thumb1_cmp_operand, thumb1_cmpneg_operand): ... to this.
2088 * config/arm/t-arm-elf: Add armv7 multilib.
2089 * config/arm/arm.md: Update patterns for Thumb-2 and Unified asm.
2091 (UNSPEC_STACK_ALIGN, ce_count): New.
2092 (arm_incscc, arm_decscc, arm_umaxsi3, arm_uminsi3,
2093 arm_zero_extendsidi2, arm_zero_extendqidi2): New
2095 * config/arm/fpa.md: Update patterns for Thumb-2 and Unified asm.
2096 (thumb2_movsf_fpa, thumb2_movdf_fpa, thumb2_movxf_fpa,
2097 thumb2_movsfcc_fpa, thumb2_movdfcc_fpa): New insns.
2098 * config/arm/cirrus.md: Update patterns for Thumb-2 and Unified asm.
2099 (cirrus_thumb2_movdi, cirrus_thumb2_movsi_insn,
2100 thumb2_cirrus_movsf_hard_insn, thumb2_cirrus_movdf_hard_insn): New
2102 * doc/extend.texi: Document ARMv7-M interrupt functions.
2103 * doc/invoke.texi: Document Thumb-2 new cores+architectures.
2105 2007-01-03 Jakub Jelinek <jakub@redhat.com>
2107 * unwind-dw2.c (SIGNAL_FRAME_BIT, EXTENDED_CONTEXT_BIT): Define.
2108 (struct _Unwind_Context): Rename args_size to flags, remove
2109 signal_frame field, add a new args_size field and version field.
2110 (_Unwind_IsSignalFrame, _Unwind_SetSignalFrame,
2111 _Unwind_IsExtendedContext): New inline functions.
2112 (_Unwind_GetGR, _Unwind_SetGR, _Unwind_GetGRPtr, _Unwind_SetGRPtr):
2113 Assume by_value array is only present if _Unwind_IsExtendedContext.
2114 (_Unwind_GetIPInfo, execute_cfa_program, uw_frame_state_for): Use
2115 _Unwind_IsSignalFrame.
2116 (__frame_state_for): Initialize context.flags to EXTENDED_CONTEXT_BIT.
2117 (uw_update_context_1): Use _Unwind_SetSignalFrame.
2118 (uw_init_context_1): Initialize context->flags to
2119 EXTENDED_CONTEXT_BIT.
2120 * config/rs6000/linux-unwind.h (frob_update_context): Use
2121 _Unwind_SetSignalFrame.
2123 2007-01-03 Andrew Pinski <andrew_pinski@playstation.sony.com>
2126 * gimplify.c (gimplify_modify_expr_complex_part): Move below
2127 tree_to_gimple_tuple. Call tree_to_gimple_tuple when we need
2130 2007-01-03 Kazu Hirata <kazu@codesourcery.com>
2132 * config/i386/i386.h (NON_STACK_REG_P, REGNO_OK_FOR_SIREG_P,
2133 REGNO_OK_FOR_DIREG_P, REWRITE_ADDRESS, ASM_OPERAND_LETTER,
2134 RET, AT_SP): Remove.
2135 * config/i386/i386.md (*sse_prologue_save_insn): Use return
2138 * alias.c (init_alias_analysis): Use VEC_safe_grow_cleared.
2139 * cfgbuild.c (find_basic_blocks): Likewise.
2140 * cfgrtl.c (rtl_create_basic_block): Likewise.
2141 * function.c (temp_slots_at_level): Likewise.
2142 * reg-stack.c (stack_regs_mentioned): Likewise.
2143 * regclass.c (allocate_reg_info): Likewise.
2144 * tree-cfg.c (init_empty_tree_cfg, build_tree_cfg, create_bb,
2145 set_bb_for_stmt, move_block_to_fn): Likewise.
2146 * tree-complex.c (tree_lower_complex): Likewise.
2147 * vec.h (VEC_safe_grow_cleared): New.
2149 * cgraphunit.c, tree-ssa-alias.c: Fix comment typos.
2151 2007-01-03 Zdenek Dvorak <dvorakz@suse.cz>
2153 * loop-unswitch.c (unswitch_loop): Pass probabilities to loopify.
2154 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Pass probabilities
2156 * cfgloopmanip.c (scale_loop_frequencies): Export.
2157 (loopify): Scale the frequencies by prescribed coefficients.
2158 (set_zero_probability): New function.
2159 (duplicate_loop_to_header_edge): Improve updating of frequencies.
2160 (lv_adjust_loop_entry_edge, loop_version): Set probabilities
2161 and frequencies according to arguments.
2162 * tree-ssa-loop-manip.c (tree_unroll_loop): Set probabilities
2164 * cfg.c (scale_bbs_frequencies_int): Allow scaling the frequencies up.
2165 * modulo-sched.c (sms_schedule): Set probabilities for entering
2166 versioned loop correctly.
2167 * tree-vect-transform.c (vect_transform_loop): Ditto.
2168 * cfgloop.h (loopify, loop_version): Declaration changed.
2169 (scale_loop_frequencies): Declared.
2171 2007-01-02 Jan Hubicka <jh@suse.cz>
2173 * cgraph.c: Include tree-flow.h
2174 (cgraph_add_new-function): Handle IPA_SSA mode; execute
2176 * cgraph.h (enum cgraph_state): Add CGRAPH_STATE_IPA_SSA.
2177 * tree-pass.h (pass_all_early_optimizations): Declare.
2178 * cgraphunit.c (cgraph_process_new_functions): Add IPA_SSA; execute
2180 (cgraph_analyze_function): Do early_local_passes.
2181 * tree-mudflap.c (mf_decl_cache_locals, mf_build_check_statement_for):
2182 Do not add referenced vars.
2183 * tree-optimize.c (gate_all_optimizations): Do not execute when not in
2185 (gate_all_early_local_passes): New gate.
2186 (pass_early_local_passes): Use new gate.
2187 (execute_early_local_optimizations): New functions.
2188 (gate_all_early_optimizations): New gate.
2189 (pass_all_early_optimizations): New pass.
2190 (execute_free_datastructures): Free SSA only when initialized.
2191 (gate_init_datastructures): Init only when optimizing.
2192 (tree_lowering_passes): Do early local passes when called late.
2193 * tree-profile.c (do_tree_profiling): Don't profile functions added
2195 (do_early_tree_profiling, pass_early_tree_profile): Kill.
2196 * tree-cfg.c (update_modified_stmts): Do not update when operands are
2198 * passes.c (init_optimizations_passes): Reorder so we go into SSA
2199 during early_local_passes.
2200 * Makefile.in (cgraph.o): Add dependency on tree-flow.h.
2203 2007-01-02 Carlos O'Donell <carlos@codesourcery.com>
2205 * Makefile.in: Update copyright year.
2207 2007-01-02 Carlos O'Donell <carlos@codesourcery.com>
2209 * Makefile.in: Export GCC_EXEC_PREFIX before calling $(RUNTEST)
2210 in $(lang_checks) and check-consistency targets.
2212 2007-01-02 Jan Hubicka <jh@suse.cz>
2214 * tree-mudflap.c (mf_decl_cache_locals, mf_build_check_statement_for):
2215 Do not add referenced vars.
2216 * tree-cfg.c (update_modified_stmts): Do not update when SSA operands
2218 * passes.c (init_optimization_passes): Put mudflap_2 after
2219 free_datastructures.
2221 2007-01-02 Jan Hubicka <jh@suse.cz>
2223 * tree-optimize (execute_fixup_cfg): Set after_inlining flag.
2224 Set NOTHROW flag on call statements proved to be nothrow.
2225 Update statement of local calls so new pure/const functions are
2226 updated. Update_ssa when in ssa form. Mark PHI nodes of nonlocal
2228 (tree_rest_of_compilation): Register hooks and initialize bitmap
2229 early. Do not set after_inlining flag.
2231 2007-01-02 Steve Ellcey <sje@cup.hp.com>
2233 * sbitmap.c (HOST_BITS_PER_LONG_LONG): Change to
2234 HOST_BITS_PER_LONGLONG
2236 2007-01-02 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
2239 * c-typeck.c (store_init_value): Don't emit pedantic overflow
2240 warning for non-static initializers.
2242 2007-01-02 Steven Bosscher <steven@gcc.gnu.org>
2244 * config/alpha/alpha.md, arm/arm.c, darwin.c, frv/frv.md,
2245 m32r/m32r.c, m32r/m32r.c, mn10300/mn10300.md, pa/pa.c,
2246 rs6000/rs6000.c, s390/s390.md, sh/sh.md, sparc/sparc.c:
2247 Always use set_unique_reg_note to add REG_EQUAL notes.
2249 2007-01-02 Kazu Hirata <kazu@codesourcery.com>
2252 2007-01-02 Kazu Hirata <kazu@codesourcery.com>
2254 * alias.c (init_alias_analysis): Use VEC_safe_grow_cleared.
2255 * cfgbuild.c (find_basic_blocks): Likewise.
2256 * cfgrtl.c (rtl_create_basic_block): Likewise.
2257 * function.c (temp_slots_at_level): Likewise.
2258 * reg-stack.c (stack_regs_mentioned): Likewise.
2259 * regclass.c (allocate_reg_info): Likewise.
2260 * tree-cfg.c (init_empty_tree_cfg, build_tree_cfg, create_bb,
2261 set_bb_for_stmt, move_block_to_fn): Likewise.
2262 * tree-complex.c (tree_lower_complex): Likewise.
2263 * vec.h (VEC_safe_grow_cleared): New.
2265 2007-01-02 Ian Lance Taylor <iant@google.com>
2267 * c-common.c (c_common_truthvalue_conversion): When warning about
2268 using an assignment as a truth value, set TREE_NO_WARNING.
2270 2007-01-02 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
2273 * c.opt (Wold-style-declaration): New.
2274 * doc/invoke.texi (C-only Warning Options): New.
2275 (Wold-style-declaration): Document it.
2276 (Wextra): Enabled by -Wextra.
2277 * c-opts.c (c_common_post_options): Enabled by -Wextra.
2278 * c-decl.c (declspecs_add_scspec): Replace -Wextra with
2279 -Wold-style-declaration.
2281 2007-01-02 Kazu Hirata <kazu@codesourcery.com>
2283 * alias.c (init_alias_analysis): Use VEC_safe_grow_cleared.
2284 * cfgbuild.c (find_basic_blocks): Likewise.
2285 * cfgrtl.c (rtl_create_basic_block): Likewise.
2286 * function.c (temp_slots_at_level): Likewise.
2287 * reg-stack.c (stack_regs_mentioned): Likewise.
2288 * regclass.c (allocate_reg_info): Likewise.
2289 * tree-cfg.c (init_empty_tree_cfg, build_tree_cfg, create_bb,
2290 set_bb_for_stmt, move_block_to_fn): Likewise.
2291 * tree-complex.c (tree_lower_complex): Likewise.
2292 * vec.h (VEC_safe_grow_cleared): New.
2294 2007-01-02 Douglas Gregor <doug.gregor@gmail.com>
2296 * c-common.c(c_common_nodes_and_builtins): Since variants of
2297 void_type_node get built before it is given a name, we need to
2298 give those variants the name, too.
2299 (complete_array_type): We need to work with the canonical main
2300 type of the array, from which we will build the qualified version.
2301 * params.def (PARAM_VERIFY_CANONICAL_TYPES): New.
2302 * print-tree.c (print_node): Display canonical type information
2304 * stor-layout.c (layout_type): When we don't know the
2305 alignment of a type for which we're building an array, we end up
2306 guessing wrong, so make the type require structural equality.
2307 * tree.c (make_node_stat): When we build a new type, it is its
2309 (build_type_attribute_qual_variant): When building an attribute
2310 variant, its canonical type is the non-attribute variant. However,
2311 if the attributes are target-dependent and they differ, we need to
2312 use structural equality checks for this type.
2313 (build_qualified_type): A qualified type is not equivalent to its
2314 unqualified variant; set the canonical type appropriately.
2315 (build_distinct_type_copy): When building a distinct type from
2316 another type, the new type is its own canonical type.
2317 (build_variant_type_copy): When building a new type variant, we
2318 assume that it is equivalent to the original type.
2319 (build_pointer_type_for_mode): When building a pointer type, also
2320 build a canonical type pointer.
2321 (build_reference_type_for_mode): When building a reference type,
2322 also build a canonical type reference.
2323 (build_index_type): When we can't hash an index type (e.g.,
2324 because its maximum value is negative), the index type requires
2325 structural equality tests.
2326 (build_array_type): Build the canonical form of an array type.
2327 (build_function_type): Function types require structural equality,
2328 because they contain default arguments, attributes, etc.
2329 (build_method_type_directly): Ditto for method types.
2330 (build_offset_type): Build the canonical offset type.
2331 (build_complex_type): Build the canonical vector type.
2332 (make_vector_type): Build the canonical vector type.
2333 * tree.h (TYPE_CANONICAL): New.
2334 (TYPE_STRUCTURAL_EQUALITY_P): New.
2335 (SET_TYPE_STRUCTURAL_EQUALITY): New.
2336 (struct tree_type): Added "canonical" field.
2337 * params.h (VERIFY_CANONICAL_TYPES): New.
2338 * doc/c-tree.texi (TYPE_CANONICAL): Document.
2339 (TYPE_STRUCTURAL_EQUALITY_P): Document.
2340 (SET_TYPE_STRUCTURAL_EQUALITY): Document.
2341 * doc/invoke.texi (verify-canonical-types): Document --param
2342 parameter for verifying canonical types.
2344 2007-01-02 Joseph Myers <joseph@codesourcery.com>
2346 * config.gcc (powerpc-*-eabispe*, powerpc-*-eabisimaltivec*,
2347 powerpc-*-eabisim*, powerpc-*-eabialtivec*, powerpc-*-eabi*,
2348 powerpc-*-rtems*, powerpc-wrs-vxworks, powerpc-wrs-vxworksae,
2349 powerpcle-*-eabisim*, powerpcle-*-eabi*): Add rs6000/e500.h to
2351 * config/rs6000/e500.h: New.
2352 * config/rs6000/eabi.h (TARGET_SPE_ABI, TARGET_SPE, TARGET_E500,
2353 TARGET_ISEL, TARGET_FPRS, TARGET_E500_SINGLE, TARGET_E500_DOUBLE):
2355 * config/rs6000/linuxspe.h (TARGET_SPE_ABI, TARGET_SPE,
2356 TARGET_E500, TARGET_ISEL, TARGET_FPRS, TARGET_E500_SINGLE,
2357 TARGET_E500_DOUBLE): Remove.
2358 * config/rs6000/vxworks.h (TARGET_SPE_ABI, TARGET_SPE,
2359 TARGET_E500, TARGET_ISEL, TARGET_FPRS): Remove.
2360 * config/rs6000/rs6000.h (CHECK_E500_OPTIONS): Define.
2361 * config/rs6000/rs6000.c (rs6000_override_options): Use
2364 2007-01-02 Joseph Myers <joseph@codesourcery.com>
2366 * config/rs6000/rs6000.c (print_operand): Check (TARGET_SPE ||
2367 TARGET_E500_DOUBLE), not TARGET_E500, for %y.
2368 (rs6000_generate_compare, rs6000_emit_sCOND, output_cbranch,
2369 rs6000_emit_cmove): Don't check TARGET_E500.
2370 * config/rs6000/rs6000.md (bunordered, bordered, sunordered,
2371 sordered): Don't check TARGET_E500.
2373 2007-01-01 Eric Christopher <echristo@apple.com>
2375 * config/mips/mips.c (mips_regno_mode_ok_for_base_p): Use
2376 HARD_REGISTER_NUM_P.
2378 2007-01-01 Roger Sayle <roger@eyesopen.com>
2380 * fold-const.c (fold_binary) <EQ_EXPR>: Fold "(X^C1) eq/ne C2" into
2381 "X eq/ne (C1^C2)". Fold "(X^Z) eq/ne (Y^Z)" as "X eq/ne Y" when Z
2382 has no side-effects. Fold "(X^C1) eq/ne (Y^C2)" as "(X^(C1^C2))
2385 2007-01-01 Mike Stump <mrs@apple.com>
2387 * configure.ac: Remove support for building with Apple's gcc-3.1.
2389 2007-01-02 Joseph Myers <joseph@codesourcery.com>
2392 * caller-save.c (add_stored_regs): Only handle SUBREGs if inner
2393 REG is a hard register. Do not modify REG before calling
2395 * rtlanal.c (subreg_get_info): Don't assert size of XMODE is a
2396 multiple of the size of YMODE for certain lowpart cases.
2398 2007-01-01 Andrew Pinski <pinskia@gmail.com>
2401 * gimplify (voidify_wrapper_expr): Update for
2404 2007-01-01 Andreas Schwab <schwab@suse.de>
2407 * config/ia64/ia64.c (ia64_compute_frame_size): Account space for
2408 save of BR0 in extra_spill_size instead of spill_size.
2409 (ia64_expand_prologue): Save BR0 outside of the gr/br/fr spill
2411 (ia64_expand_epilogue): Restore BR0 from its new location.
2413 2007-01-01 Andrew Pinski <pinskia@gmail.com>
2415 * gimplify.c (gimplify_init_constructor <case VECTOR_TYPE>):
2416 Use a temporary variable if the left hand side is not a gimple
2419 2007-01-01 Andrew Pinski <pinskia@gmail.com>
2421 * gimplify.c (gimplify_return_expr): Make the temporary variable
2422 for the return expression, a gimple register variable.
2424 2007-01-01 Jan Hubicka <jh@suse.cz>
2426 * emit-rtl.c (emit_copy_of_insn_after): Do not call copy_insn_1 for
2429 2007-01-01 Mike Stump <mrs@apple.com>
2431 * configure.ac (HAVE_GAS_LITERAL16): Add autoconf check for
2433 * config/darwin.c (machopic_select_rtx_section): Use
2435 (darwin_mergeable_constant_section): Likewise.
2436 * configure: Regenerate.
2437 * config.in: Regenerate.
2439 2007-01-01 Jan Hubicka <jh@suse.cz>
2440 Andrew Pinski <pinskia@gmail.com>
2442 * cgraphunit.c (cgraph_optimize): Call cgraph_add_new_functions
2443 before starting IPA passes.