1 2007-01-19 Richard Sandiford <richard@codesourcery.com>
2 Julian Brown <julian@codesourcery.com>
4 * config/m68k/m68k.h (CONST_OK_FOR_LETTER_P): Add an 'R' case.
5 * config/m68k/m68k.md (*movsi_cfv4): Fold into...
6 (*movsi_cf): ...here. Remove unnecessary 'R' from 'Rg'.
9 2007-01-19 Richard Sandiford <richard@codesourcery.com>
11 * config/m68k/m68k-protos.h (valid_mov3q_const): Take a HOST_WIDE_INT
13 (output_move_const_into_data_reg, output_move_simode_const): Delete.
14 * config/m68k/m68k.c (const_method, const_int_cost): Take a
15 HOST_WIDE_INT instead of an rtx.
16 (m68k_rtx_costs): Update call accordingly.
17 (output_move_const_into_data_reg): Likewise. Fix formatting.
18 (valid_mov3q_const): Take a HOST_WIDE_INT instead of an rtx.
20 (output_move_simode_const): Update calls after above changes.
21 Rework to use automatic variables and predicates like MEM_P.
22 * config/m68k/m68k.md (pushexthisi_const): Update call to
25 2007-01-19 Dirk Mueller <dmueller@suse.de>
27 * tree-ssa-alias.c (perform_var_substitution): Fix typo
30 2007-01-19 Richard Guenther <rguenther@suse.de>
32 * builtins.c (expand_builtin_cexpi): Fall back to expanding
33 via cexp in case sincos is not available.
35 2007-01-19 Richard Guenther <rguenther@suse.de>
37 * doc/tm.texi (TARGET_HAS_SINCOS): Document new target macro.
38 * defaults.h (TARGET_HAS_SINCOS): Default to off.
39 * config/linux.h (TARGET_HAS_SINCOS): Set to on if we have glibc.
40 * config/alpha/linux.h (TARGET_HAS_SINCOS): Likewise.
41 * config/sparc/linux.h (TARGET_HAS_SINCOS): Likewise.
42 * config/sparc/linux64.h (TARGET_HAS_SINCOS): Likewise.
43 * config/rs6000/linux.h (TARGET_HAS_SINCOS): Likewise.
44 * config/rs6000/linux64.h (TARGET_HAS_SINCOS): Likewise.
46 2007-01-19 Uros Bizjak <ubizjak@gmail.com>
48 * config/i386/i386.md (*fpatanxf3_i387, fpatan_extend<mode>xf3_i387):
50 (atan2sf3_1, atan2df3_1, atan2xf3_1): Remove insn patterns.
51 (atan2xf3): Directly generate RTL pattern.
52 (atan2<mode>3): Rename from atan2sf3 and atan2df3 and macroize insn
53 patterns using X87MODEF12 mode macro. Use fpatan_extend<mode>xf3_i387
54 and truncate result to requested mode. Use SSE_FLOAT_MODE_P to
55 disable patterns for SSE math.
56 (atan<mode>2): Rename from atansf2 and atandf2 and macroize insn
57 patterns using X87MODEF12 mode macro. Use fpatan_extend<mode>xf3_i387
58 and truncate result to requested mode. Use SSE_FLOAT_MODE_P to
59 disable patterns for SSE math.
61 2007-01-19 Alexandre Oliva <aoliva@redhat.com>
63 * libgcc-std.ver: Fix typo in %inherit for GCC_4.3.0.
65 2007-01-18 Roger Sayle <roger@eyesopen.com>
67 * fold-const.c (fold_unary) <VIEW_CONVERT_EXPR>: Optimize away a
68 VIEW_CONVERT_EXPR to the same type as it's operand.
70 2007-01-18 David Edelsohn <edelsohn@gnu.org>
72 * config/rs6000/darwin-ldouble.c: Only build _SOFT_FLOAT if
73 configured for long double 128.
75 2007-01-18 Mike Stump <mrs@apple.com>
77 * config/rs6000/rs6000.c (rs6000_emit_vector_compare): Fix build
80 2007-01-18 Michael Meissner <michael.meissner@amd.com>
82 * i386.c (ix86_compute_frame_layout): Make fprintf's in #if 0 code
85 2007-01-18 Jan Hubicka <jh@suse.cz>
87 * tree-ssa-operands.c (vop_free_bucket_size): Never return value
88 greater than NUM_VOP_FREE_BUCKETS.
90 2007-01-18 Daniel Berlin <dberlin@dberlin.org>
92 * tree-ssa-structalias.c: Update comments.
93 (ptabitmap_obstack): Removed.
95 (oldpta_obstack): Ditto.
96 (stats): Add a few members.
97 (struct variable_info): Remove node, complex, address_taken, and
98 indirect_target members. Add oldsolution member.
99 (new_var_info): Do not initialize removed members.
100 (constraint_expr_type): Remove INCLUDES.
101 (constraint_graph): Add size, implicit_preds, rep,
102 indirect_cycles, eq_rep, label, direct_nodes, and complex members.
103 (FIRST_REF_NODE): New macro.
104 (LAST_REF_NODE): Ditto.
105 (FIRST_ADDR_NODE): Ditto.
106 (find): New function.
108 (dump_constraint): Do not handle INCLUDES.
109 (insert_into_complex): Do not insert duplicate constraints.
110 (condense_varmap_nodes): Renamed and rewritten into ...
111 (merge_node_constraints): This. Also fix bug in handling of
112 offseted copy constraints.
113 (clear_edges_for_node): No longer need to deal with preds at all,
114 or removing associated preds/succs.
115 (merge_graph_nodes): Deal with indirect_cycles.
116 Don't deal with predecessors.
117 (add_implicit_graph_edge): New function.
118 (add_pred_graph_edge): Ditto.
119 (add_graph_edge): Don't deal with predecessors.
120 (build_constraint_graph): Removed.
121 (build_pred_graph): New function.
122 (build_succ_graph): Ditto.
123 (struct scc_info): Removed in_component. Added roots, dfs, and
124 node_mapping. Remove visited_index, unification_queue.
125 (scc_visit): Deal with union-find we do now.
126 Deal with cycles with REF nodes.
127 (collapse_nodes): Renamed and rewritten to ...
129 (process_unification_queue): Removed.
130 (topo_visit): Cleanup
131 (do_da_constraint): Use find.
132 (do_sd_constraint): Ditto.
133 (do_ds_constraint): Ditto.
134 (do_complex_constraint): Ditto.
135 (init_scc_info): Update for removed and added members.
136 (find_and_collapse_graph_cycles): Renamed and rewritten into ...
137 (find_indirect_cycles): This.
138 (equivalence_class): New variable.
139 (label_visit): New function.
140 (perform_variable_substitution): Rewritten.
141 (free_var_substitution_info): New function.
142 (find_equivalent_node): Ditto.
143 (move_complex_constraints): Ditto.
144 (eliminate_indirect_cycles): Ditto.
145 (solve_graph): Only propagate changed bits.
146 Use indirect cycle elimination.
148 (tree_id_t): Rename to tree_vi_t, delete id member, add vi member.
149 (tree_id_eq): Renamed to ...
150 (tree_vi_eq): This. Update for member change
151 (insert_id_for_tree): Renamed and rewritten to ...
152 (insert_vi_for_tree): This.
153 (lookup_id_for_tree): Renamed and rewritten to ...
154 (lookup_vi_for_tree): This.
155 (get_id_for_tree): Renamed and rewritten to ...
156 (get_vi_for_tree): Ditto.
157 (get_constraint_exp_from_ssa_var): Update to use get_vi_for_tree.
158 (process_constraint): Don't handle INCLUDES.
159 Remove special ADDRESSOF case.
160 (find_func_aliases): Rewrite to use vi functions instead of id
162 (create_function_info_for): Ditto.
163 (create_variable_info_for): Ditto.
164 (intra_create_variable_infos): Ditto.
165 (merge_smts_into): Ditto.
166 (find_what_p_points_to): Ditto.
167 (init_base_vars): Ditto.
168 (init_alias_vars): Ditto.
169 (remove_preds_and_fake_succs): New function.
170 (dump_sa_points_to_info): Dump new stats.
171 (dump_solution_for_var): Use find.
172 (set_used_smts): Fix formatting.
173 (compute_points_to_sets): Updated for new functions.
174 (ipa_pta_execute): Ditto.
176 2007-01-18 Kazu Hirata <kazu@codesourcery.com>
177 Richard Sandiford <richard@codesourcery.com>
179 * doc/tm.texi (TARGET_FUNCTION_VALUE): Expand documentation of
181 * calls.c (expand_call): If the return value is a PARALLEL,
182 extract its first member.
183 * config/m68k/linux.h (FUNCTION_EXTRA_EPILOGUE): Remove.
184 * config/m68k/m68k.c (m68k_output_function_epilogue): Don't
185 use FUNCTION_EXTRA_EPILOGUE.
186 (m68k_function_value): Return a PARALLEL if the return value
187 is of a pointer type.
188 * config/m68k/netbsd-elf.h (current_function_returns_pointer)
189 (FUNCTION_EXTRA_EPILOGUE): Remove.
190 * config/m68k/m68k.md (D0_REG): New constant.
192 2007-01-18 Kazu Hirata <kazu@codesourcery.com>
194 * config/m68k/m68k.c (m68k_output_function_epilogue): Don't
195 output a NOP for empty epilogues.
197 2007-01-18 Richard Sandiford <richard@codesourcery.com>
199 * config/m68k/m68k.c (m68k_use_return_insn): Update comments
200 before function. Extend register save check to include all
201 registers, not just integer ones.
203 2007-01-18 Kazu Hirata <kazu@codesourcery.com>
205 * config/m68k/m68k-protos.h (use_return_insn): Rename to...
206 (m68k_use_return_insn): ...this.
207 * config/m68k/m68k.h (USE_RETURN_INSN): Delete.
208 * config/m68k/m68k.c (use_return_insn): Rename to...
209 (m68k_use_return_insn): ...this.
210 * config/m68k/m68k.md (return): Use m68k_use_return_insn instead
213 2007-01-18 Nathan Sidwell <nathan@codesourcery.com>
215 * config/m68k/fpgnulib.c (__truncdfsf2): Implement round to
216 nearest even, fix denormal rounding overflow.
218 2007-01-18 Richard Sandiford <richard@codesourcery.com>
220 * config/m68k/m68k.md (movsf_cf_hard): Use fsmove instead of
221 f%$smove and f%$move.
222 (movdf_cf_hard): Use fdmove for cases 0 and 3 and fmove for case 1.
223 (extendsfdf2_cf): Use fdmove instead of f%&move.
224 (truncdfsf2_cf): Use fsmove instead of f%$smove.
225 (add<mode>3_cf, sub<mode>3_cf): Use <FP:prec> instead of <FP:round>.
227 2007-01-18 Richard Sandiford <richard@codesourcery.com>
229 * config/m68k/m68k.md (movdf_cf_hard): Use output_move_double for
232 2007-01-18 Richard Sandiford <richard@codesourcery.com>
234 * real.h (real_format): Add a canonical_nan_lsbs_set field.
235 (coldfire_single_format): Declare.
236 (coldfire_double_format): Likewise.
237 * real.c (encode_ieee_single): Use canonical_nan_lsbs_set instead
238 of qnan_msb_set to determine the lower bits of a canonical
240 (encode_ieee_double): Likewise.
241 (encode_ieee_quad): Likewise.
242 (ieee_single_format): Initialize canonical_nan_lsbs_set.
243 (mips_single_format): Likewise.
244 (ieee_double_format): Likewise.
245 (mips_double_format): Likewise.
246 (ieee_extended_motorola_format): Likewise.
247 (ieee_extended_intel_96_format): Likewise.
248 (ieee_extended_intel_128_format): Likewise.
249 (ieee_extended_intel_96_round_53_format): Likewise.
250 (ibm_extended_format): Likewise.
251 (mips_extended_format): Likewise.
252 (ieee_quad_format): Likewise.
253 (mips_quad_format): Likewise.
254 (vax_f_format): Likewise.
255 (vax_d_format): Likewise.
256 (vax_g_format): Likewise.
257 (i370_single_format): Likewise.
258 (i370_double_format): Likewise.
259 (decimal_single_format): Likewise.
260 (decimal_double_format): Likewise.
261 (decimal_quad_format): Likewise.
262 (c4x_single_format): Likewise.
263 (c4x_extended_format): Likewise.
264 (real_internal_format): Likewise.
265 (coldfire_single_format): New real_format.
266 (coldfire_double_format): Likewise.
267 * config/pdp11/pdp11.c (pdp11_f_format): Initialize
268 canonical_nan_lsbs_set.
269 (pdp11_d_format): Likewise.
270 * config/m68k/m68k.c (override_options): Override REAL_FORMAT_MODE
271 if TARGET_COLDFIRE_CPU.
273 2007-01-18 Richard Sandiford <richard@codesourcery.com>
275 * config/m68k/m68k-protos.h (m68k_output_pic_call): Delete.
276 (output_call, m68k_legitimize_call_address): Declare.
277 * config/m68k/m68k.h (EXTRA_CONSTRAINT): Remove unnecessary
278 parenthesees. Add support for a 'W' constraint.
279 (LEGITIMATE_PIC_OPERAND_P): Remove SYMBOL_REF_FLAG handling.
280 (PRINT_OPERAND_PUNCT_VALID_P): Remove comment about 'o'.
281 (m68k_symbolic_call, m68k_symbolic_jump): Declare.
282 * config/m68k/m68k.c (m68k_symbolic_call, m68k_symbolic_jump): New
284 (override_options): Initialize them. Do not set flag_no_function_cse
285 for TARGET_ID_SHARED_LIBRARY.
286 (m68k_output_pic_call): Delete.
287 (m68k_legitimize_call_address): New function.
288 (print_operand): Remove the %o prefix. Handle the %p prefix.
289 (output_call): New function.
290 (m68k_output_mi_thunk): Use m68k_symbolic_jump. Always load the
291 target address from the GOT if symbolic jumps are not allowed.
292 * config/m68k/m68k.md (call, general_operand): Do not set
293 SYMBOL_REF_FLAG. Use m68k_legitimize_call_address instead.
294 Merge separate flag_pic and !flag_pic define_insns into...
295 (*call, *call_value): ...these new patterns. Match the address
296 rather than the containing MEM and require it to be a call_operand.
297 Use output_call to generate the asm template.
298 * config/m68k/predicates.md (const_call_operand): New predicate.
299 (call_operand): Likewise.
301 2007-01-18 Nathan Sidwell <nathan@codesourcery.com>
303 * config/m68k/m68k.h (REGISTER_MOVE_COST): Simplify definition.
304 (STACK_GROWS_DOWNWARD): Define to 1.
305 (FUNCTION_VALUE, LIBCALL_VALUE, FUNCTION_VALUE_REGNO_P): Equivocate
306 comments, emphasizing that these values are only defaults.
307 * config/m68k/linux.h (LINK_SPEC): Fix formatting in #undef.
308 * config/m68k/m68k.c (const_method): Remove trailing whitespace.
310 2007-01-18 Richard Sandiford <richard@codesourcery.com>
312 * config/m68k/m68k.md (cmpsi): Remove outdated flag_pic handling.
314 2007-01-18 Kazu Hirata <kazu@codesourcery.com>
315 Richard Sandiford <richard@codesourcery.com>
317 * config/m68k/m68k.h (DATA_REGNO_P, ADDRESS_REGNO_P, INT_REGNO_P)
318 (FP_REGNO_P): New macros.
319 (REGNO_OK_FOR_INDEX_P, REGNO_OK_FOR_BASE_P, REGNO_OK_FOR_DATA_P)
320 (REGNO_OK_FOR_FP_P, REG_OK_FOR_INDEX_P, REG_OK_FOR_BASE_P): Use them.
321 (EH_RETURN_STACKADJ_RTX): Use A0_REG.
322 * config/m68k/m68k.c (m68k_regno_mode_ok): Use the new REGNO macros.
324 2007-01-18 Nathan Sidwell <nathan@codesourcery.com>
326 * config.gcc (m68k-*-aout*, m68k-*-coff*, m68020-*-elf*, m68k-*-elf*)
327 (m68k-*-uclinuxoldabi, m68k-*-uclinux*, m68k-*-rtems*): Add t-floatlib
329 * config/m68k/t-floatlib: New file, extracting common code from...
330 * config/m68k/t-m68kbare, config/m68k/t-m68kelf,
331 * config/m68k/t-uclinux: Here.
332 * config/m68k/fpgnulib.c: Do not compile extendeed precision
333 routines on ColdFire targets.
335 2007-01-18 Nathan Sidwell <nathan@codesourcery.com>
337 * config.gcc (m68k-*-aout*, m68k-*-coff*, m68020-*-elf*, m68k-*-elf*)
338 (m68010-*-netbsdelf*, m68k*-*-netbsdelf*, m68k*-*-openbsd*)
339 (m68k-*-uclinuxoldabi, m68k-*-uclinux*, m68k-*-linux*)
340 (m68k-*-rtems*): Use tm_file rather than m68k/m68k.h and
341 explicitly set MOTOROLA to 1.
342 * config/m68k/m68k.h (MOTOROLA): Simplify definition accordingly.
344 2007-01-18 Richard Sandiford <richard@codesourcery.com>
345 Nathan Sidwell <nathan@codesourcery.com>
347 * config/m68k/m68k.h (PCC_STATIC_STRUCT_RETURN): Delete.
348 (ASM_OUTPUT_REG_PUSH, ASM_OUTPUT_REG_POP): Add MOTOROLA cases.
349 * config/m68k/coff.h (REGISTER_PREFIX_MD): Delete.
350 * config/m68k/m68020-elf.h (LIB_SPEC): Delete.
351 * config/m68k/m68k-none.h (CC1_SPEC, CPP_SUBTARGET_SPEC): Delete.
352 * config/m68k/m68kelf.h (IMMEDIATE_PREFIX, REGISTER_PREFIX_MD)
353 (ASM_OUTPUT_REG_PUSH, ASM_OUTPUT_REG_POP): Delete.
354 (NO_DOLLAR_IN_LABEL, PCC_STATIC_STRUCT_RETURN): Don't undefine.
355 (BSS_ASM_OP, ASM_OUTPUT_SKIP, ASM_OUTPUT_ASCII): Delete.
356 * config/m68k/m68kemb.h (PCC_STATIC_STRUCT_RETURN): Don't undefine.
357 * config/m68k/linux.h (SIZE_TYPE, PTRDIFF_TYPE, WCHAR_TYPE)
358 (WCHAR_TYPE_SIZE, TARGET_OBJFMT_CPP_BUILTINS): Delete.
359 (TARGET_OS_CPP_BUILTINS): Don't define mc68000 and mc68020 here.
360 (DBX_REGISTER_NUMBER): Delete.
361 * config/m68k/netbsd-elf.h (IMMEDIATE_PREFIX): Delete.
362 (PCC_STATIC_STRUCT_RETURN): Don't undefine.
363 * config/m68k/openbsd.h (PCC_STATIC_STRUCT_RETURN): Define.
365 2007-01-18 Richard Sandiford <richard@codesourcery.com>
367 * config.gcc (m68k-*-uclinux*): Add flat.h to $tm_file.
368 * config/flat.h: New file.
369 * crtstuff.c (USE_PT_GNU_EH_FRAME): Don't define if
371 * config/m68k/m68k.h (ASM_PREFERRED_EH_DATA_FORMAT): Do not use
372 indirect references for -msep-data or -mid-shared-library.
373 Do not use PC-relative code addresses either.
374 * config/m68k/m68k.c (override_options): Restrict -fPIC error
376 * config/m68k/uclinux.h (STARTFILE_SPEC): Define. Use Scrt1.o
377 for shared libraries and crt1.o for executables. Use crti.o and
379 (ENDFILE_SPEC): Use crtend.o and crtn.o.
380 (LIB_SPEC): Suppress -Rlibc.gdb if -static-libc is given.
381 Do not add -elf2flt or -shared-lib-id options here.
382 (LINK_SPEC): Define. Pass -elf2flt if no -elf2flt option is given.
383 Pass -shared-lib-id if -mid-shared-library, taking the library
384 identifier from -mshared-library-id if given, otherwise
386 (EH_FRAME_IN_DATA_SECTION): Do not undefine.
387 (INIT_SECTION_ASM_OP, FINI_SECTION_ASM_OP): Likewise.
388 (TARGET_OS_CPP_BUILTINS): Define __GXX_MERGED_TYPEINFO_NAMES=0
389 and __GXX_TYPEINFO_EQUALITY_INLINE=0 if -mid-shared-library.
390 (DRIVER_SELF_SPECS): Map unadorned PIC options to -msep-data.
391 * config/m68k/t-uclinux (EXTRA_MULTILIB_PARTS): Add crtbegin.o
393 * config/m68k/lb1sf68.asm (PICCALL): Use an lea and pc-relative
394 jump sequence for ISA A and ISA A+.
397 2007-01-18 Richard Sandiford <richard@codesourcery.com>
399 * config.gcc (m68k-*-uclinux*): Base the port on the common
400 and m68k GNU/Linux files rather than on the generic ELF ones.
401 * config/m68k/uclinux.h (TARGET_VERSION): Override.
402 (TARGET_OS_CPP_BUILTINS): Use LINUX_TARGET_OS_CPP_BUILTINS.
404 2007-01-18 Julian Brown <julian@codesourcery.com>
405 Richard Sandiford <richard@codesourcery.com>
407 * config/m68k/m68k.h (LONG_DOUBLE_TYPE_SIZE): Make 64-bit on ColdFire.
408 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
409 * config/m68k/netbsd-elf.h (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Undefine
411 * config/m68k/uclinux-oldabi.h (LONG_DOUBLE_TYPE_SIZE): Redefine to
413 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
415 2007-01-18 Richard Sandiford <richard@codesourcery.com>
417 * doc/install.texi: Document m68k-uclinuxoldabi.
418 * config.gcc (m68k-*-uclinuxoldabi): New configuration.
419 * config/m68k/uclinux-oldabi.h: New file, copied from
420 config/m68k/uclinux.h.
422 2007-01-18 Kaz Kojima <kkojima@rr.iij4u.or.jp>
424 * config/m32r/m32r-protos.h (m32r_expand_epilogue): Declare it.
425 * config/m32r/m32r.c (m32r_setup_incoming_varargs): Use gen_frame_mem.
426 (m32r_compute_frame_size): Use unsigned for regno.
427 (m32r_reload_lr): Use gen_frame_mem.
429 (m32r_output_function_epilogue): Don't output the function epilogue
431 (m32r_expand_epilogue): New.
432 (direct_return): Return false if the function has the interrupt
434 (m32r_hard_regno_rename_ok): Remove code for the textual epilogue.
435 * config/m32r/m32r.md (epilogue): New expander.
436 (return_lr, return_rte): New insns.
437 (return): Make it expander.
438 (return_normal): New expander.
440 2007-01-18 Josh Conner <jconner@apple.com>
443 * config/rs6000/rs6000.c (rs6000_emit_vector_compare): Add
444 support for UNLE, UNLT, UNGE, and UNGT.
446 2007-01-18 Jan Hubicka <jh@suse.cz>
448 * tree-vrp.c (finalize_jump_threads): Do not call cleanup_cfg by hand.
449 * tree-tailcall (add_virtual_phis): Likewise.
450 (optimize_tail_call): Return TODOs.
451 (execute_tail_calls): Return TODOs.
452 * tree-ssa-ccp (execute_fold_all_builtins): Do cleanup_cfg via TODO.
453 * tree-cfgcleanup.c (cleanup_tree_cfg_loop): Return if something
455 * tree-ssa-forwprop.c (tree_ssa_forward_propagate_single_use_value):
456 Cleanup cfg using TODO.
457 * tree-flow.h (cleanup_tree_cfg_loop): Update prototype.
458 * passes.c (execute_function_todo): When cleanup did something, remove
460 * tree-cfg.c (pass_build_cfg): Add cleanup_cfg TODO.
461 (make_edges): Don't cleanup_cfg.
463 2007-01-18 Uros Bizjak <ubizjak@gmail.com>
465 * reg-stack.c (subst_stack_regs_pat) [UNSPEC_SINCOS_COS,
466 UNSPEC_XTRACT_FRACT]: Use generic code for instructions that
467 operate on the top of stack.
468 [UNSPEC_SINCOS_SIN, UNSPEC_XTRACT_EXP, UNSPEC_TAN]: Rewrite
469 register handling of instructions that output to the second
471 [UNSPEC_TAN_ONE, UNSPEC_TAN_TAN]: Remove.
472 (move_for_stack_reg): Special-case check for dead destination
473 stack slot for constant load of 1.0 inside UNSPEC_TAN.
475 * config/i386/i386.md (UNSPEC_TAN): New constant.
476 (UNSPEC_TAN_ONE, UNSPEC_TAN_TAN): Remove.
477 (fptanxf4_i387, fptan_extend<mode>xf4_i387): New patterns
478 to correctly model move of constant 1.0 to top stack slot.
479 (*tandf3_1, *tansf3_1, *tanxf3_1): Remove insn patterns.
480 (unnamed peephole2 pattern): Remove corresponding peephole2
481 pattern that optimizes tan insn and loading of constant 1.0.
482 (tanxf2): Use fptanxf4_i387.
483 (tan<mode>2): Rename from tansf2 and tandf2 and macroize insn
484 patterns using X87MODEF12 mode macro. Use fptan_extend<mode>xf4_i387
485 and truncate result to requested mode. Use SSE_FLOAT_MODE_P to
486 disable patterns for SSE math.
487 (sincos<mode>3): Use truncxf<mode>2_i387_noop for truncation.
488 (fyl2x_extend<mode>xf3_i387): Use X87MODEF12 for operand 1.
490 2007-01-18 Dirk Mueller <dmueller@suse.de>
491 Richard Guenther <rguenther@suse.de>
494 * doc/invoke.texi (Warray-bounds): Document -Warray-bounds.
495 * common.opt (Warray-bounds): Add new warning option.
496 * c-opts.c (c_common_handle_option): Define -Warray-bounds
498 * Makefile.in: make tree-vrp.o depend on toplev.h
499 * tree-vrp.c (vrp_finalize): Call check_array_refs if -Warray-bounds
501 (check_array_refs, check_array_bounds, check_array_ref): New.
503 2007-01-18 Jan Hubicka <jh@suse.cz>
505 * tree-ssa-ccp.c (ccp_finalize): Return if something changed.
506 (execute_ssa_ccp): Return flags conditionally.
507 * tree-ssa-propagate.c (substitue_and_fold): Return if something was
509 * tree-ssa-propagate.h (substitute_and_fold): Update prototype.
511 2007-01-18 Steven Bosscher <steven@gcc.gnu.org>
513 * cfgcleanup.c (cleanup_cfg): Detect cfglayout mode and set
514 the CLEANUP_CFGLAYOUT flag when in cfglayout mode.
516 * Makefile.c (GTFILES): Add cfglayout.h.
517 * gengtype.c (open_base_files): Likewise.
518 * cfglayout.c (cfg_layout_function_footer,
519 cfg_layout_function_header) Reindent to make gengtype happy.
520 * cfglayout.h (cfg_layout_function_footer,
521 cfg_layout_function_header): Add GTY(()) marker.
523 * ifcvt.c (noce_try_sign_mask): Make sure INSN_B is non-null.
525 2007-01-18 Ben Elliston <bje@au.ibm.com>
527 * genautomata.c (write_automata): Include xstrerror output in the
528 error message if writing the DFA description file fails.
530 2007-01-17 H.J. Lu <hongjiu.lu@intel.com>
532 * config/mips/mips-protos.h (mips_output_external): Make it
534 * config/mips/iris.h (TARGET_ASM_EXTERNAL_LIBCALL): Removed.
535 * config/mips/mips.c (irix_output_external_libcall): Likewise.
536 (extern_list): Likewise.
537 (extern_head): Likewise.
538 (TARGET_ASM_FILE_END): Likewise.
539 (mips_file_end): Likewise.
540 (mips_output_external): Rewritten.
542 2007-01-18 Ben Elliston <bje@au.ibm.com>
544 * genpreds.c (write_insn_preds_c): Only write out the function
545 body for regclass_for_constraint if we have register constraints.
547 2007-01-17 Tom Tromey <tromey@redhat.com>
549 * doc/sourcebuild.texi (libgcj Tests): Use sourceware.org.
550 * doc/install.texi (Testing): Use sourceware.org.
551 (Binaries): Likewise.
552 (Specific): Likewise.
553 * doc/contrib.texi (Contributors): Use sourceware.org.
555 2007-01-17 Anatoly Sokolov <aesok@post.ru>
557 * config/avr/avr.h (AVR_HAVE_LPMX): New macro.
558 (AVR_ENHANCED): Rename to ...
559 (AVR_HAVE_MUL): ... new.
560 (avr_enhanced_p): Rename to ...
561 (avr_have_mul_p): ... new.
562 (TARGET_CPU_CPP_BUILTINS): Use 'avr_have_mul_p' instead of
563 'avr_enhanced_p' for "__AVR_ENHANCED__". Define "__AVR_HAVE_MUL__".
564 * config/avr/avr.c (avr_enhanced_p): Rename to ...
565 (avr_have_mul_p): ... new.
566 (base_arch_s): Rename 'enhanced' to 'have_mul'.
567 (avr_override_options): Use 'avr_have_mul_p' and 'have_mul' instead of
568 'avr_enhanced_p' and 'enhanced'.
569 (ashlhi3_out, ashrhi3_out, lshrhi3_out, avr_rtx_costs): Use
570 AVR_HAVE_MUL instead of AVR_ENHANCED.
571 * avr.md (*tablejump_enh): Use AVR_HAVE_LPMX instead of AVR_ENHANCED.
572 (mulqi3, *mulqi3_enh, *mulqi3_call, mulqihi3, umulqihi3, mulhi3,
573 *mulhi3_enh, *mulhi3_call, mulsi3, *mulsi3_call): Use AVR_HAVE_MUL
574 instead of AVR_ENHANCED.
575 (*tablejump_enh): Use AVR_HAVE_LPMX instead of AVR_ENHANCED.
576 * libgcc.S: Use __AVR_HAVE_MUL__ instead of __AVR_ENHANCED__.
577 (__tablejump__): Use __AVR_HAVE_LPMX__ instead of __AVR_ENHANCED__.
579 2007-01-17 Ian Lance Taylor <iant@google.com>
581 * vec.h (VEC_reserve_exact): Define.
582 (vec_gc_p_reserve_exact): Declare.
583 (vec_gc_o_reserve_exact): Declare.
584 (vec_heap_p_reserve_exact): Declare.
585 (vec_heap_o_reserve_exact): Declare.
586 (VEC_OP (T,A,reserve_exact)): New static inline function, three
588 (VEC_OP (T,A,reserve)) [all versions]: Remove handling of
590 (VEC_OP (T,A,alloc)) [all versions]: Call ...reserve_exact.
591 (VEC_OP (T,A,copy)) [all versions]: Likewise.
592 (VEC_OP (T,a,safe_grow)) [all versions]: Likewise.
593 * vec.c (calculate_allocation): Add exact parameter. Change all
595 (vec_gc_o_reserve_1): New static function, from vec_gc_o_reserve.
596 (vec_gc_p_reserve, vec_gc_o_reserve): Call vec_gc_o_reserve_1.
597 (vec_gc_p_reserve_exact, vec_gc_o_reserve_exact): New functions.
598 (vec_heap_o_reserve_1): New static function, from vec_heap_o_reserve.
599 (vec_heap_p_reserve, vec_heap_o_reserve): Call vec_heap_o_reserve_1.
600 (vec_heap_p_reserve_exact): New function.
601 (vec_heap_o_reserve_exact): New function.
603 2007-01-17 Jan Hubicka <jh@suse.cz>
605 * ipa-type-escape.c (look_for_casts): Revamp using handled_component_p.
607 2007-01-17 Eric Christopher <echristo@apple.com>
609 * config.gcc: Support core2 processor.
611 2007-01-16 Jan Hubicka <jh@suse.cz>
613 * tree-ssanames.c (release_dead_ssa_names): Instead of ggc_freeing
614 the names, just unlink the chain so we don't crash on dangling pointers
617 2007-01-16 Jan Hubicka <jh@suse.cz>
619 * cgraph.h (cgraph_decide_inlining_incrementally): Kill.
620 * tree-pass.h: Reorder to make IPA passes appear toegher.
621 (pass_early_inline, pass_inline_parameters, pass_apply_inline): Declare.
622 * cgraphunit.c (cgraph_finalize_function): Do not compute inling
623 parameters, do not call early inliner.
624 * ipa-inline.c: Update comments. Include tree-flow.h
625 (cgraph_decide_inlining): Do not compute inlining parameters.
626 (cgraph_decide_inlining_incrementally): Return TODOs; assume to
627 be called with function context set up.
628 (pass_ipa_inline): Remove unreachable functions before pass.
629 (cgraph_early_inlining): Simplify assuming to be called from the
631 (pass_early_inline): New pass.
632 (cgraph_gate_ipa_early_inlining): New gate.
633 (pass_ipa_early_inline): Turn into simple wrapper.
634 (compute_inline_parameters): New function.
635 (gate_inline_passes): New gate.
636 (pass_inline_parameters): New pass.
637 (apply_inline): Move here from tree-optimize.c
638 (pass_apply_inline): New pass.
639 * ipa.c (cgraph_remove_unreachable_nodes): Verify cgraph after
641 * tree-inline.c (optimize_inline_calls): Return TODOs rather than
643 (tree_function_versioning): Do not allocate dummy struct function.
644 * tree-inline.h (optimize_inline_calls): Update prototype.
645 * tree-optimize.c (execute_fixup_cfg): Export.
646 (pass_fixup_cfg): Remove
647 (tree_rest_of_compilation): Do not apply inlines.
648 * tree-flow.h (execute_fixup_cfg): Declare.
649 * Makefile.in (gt-passes.c): New.
650 * passes.c: Include gt-passes.h
651 (init_optimization_passes): New passes.
652 (nnodes, order): New static vars.
653 (do_per_function_toporder): New function.
654 (execute_one_pass): Dump current pass here.
655 (execute_ipa_pass_list): Don't dump current pass here.
657 2007-01-16 Janis Johnson <janis187@us.ibm.com>
659 * config/dfp-bit.c (dfp_compare_op): Return separate value for NaN.
660 (DFP_NE, DFP_LE, DFP_GE): Return false for NaN.
662 2007-01-16 David Edelsohn <edelsohn@gnu.org>
664 * config/rs6000/darwin-ldouble.c: Build file for SOFT_FLOAT.
665 (strong_alias): Define.
666 (__gcc_qmul): Provide non-FMA for soft-float.
682 * config/rs6000/rs6000.c (rs6000_init_libfuncs): Initialize
683 soft-float functions.
684 * config/rs6000/libgcc-ppc-glibc.ver: Version soft-float symbols.
685 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Do not warn
686 about long double soft float.
688 2007-01-16 Dorit Nuzman <dorit@il.ibm.com>
689 Tehila Meyzels <tehila@il.ibm.com>
691 * tree-vectorizer.h (is_pattern_stmt_p): New.
692 * tree-vect-analyze.c (vect_determine_vectorization_factor): Fix
693 formatting (tabs instead of spaces). Cleanup and clarify setting
694 of STMT_VINFO_VECTYPE. Call is_pattern_stmt_p.
695 * tree-vect-transform.c (vect_get_vec_def_for_operand): Fix typo.
696 (vectorizable_type_demotion): Check that types are integral.
697 (vectorizable_type_promotion): Likewise.
698 (vectorizable_store): Fix typo. Eliminate new-line at end of
701 2007-01-16 Jan Hubicka <jh@suse.cz>
703 * tree-ssanames.c (release_dead_ssa_names): Remove invalidated
706 2007-01-15 Eric Christopher <echristo@apple.com>
708 * ifcvt.c: Include vec.h, vecprim.h.
709 (check_cond_move_block): New argument regs.
710 Reorganize. Add registers used to regs.
711 (cond_move_process_if_block): Use regs set above as
714 2007-01-15 Eric Christopher <echristo@apple.com>
716 * config/darwin.h: Update copyright.
717 (TARGET_OPTION_TRANSLATE_TABLE): Add umbrella.
718 (LINK_COMMAND_SPEC): Add -u.
719 (LINK_SPEC): Fix umbrella for above.
721 2007-01-15 Joseph S. Myers <joseph@codesourcery.com>
723 * config/soft-fp/op-common.h, config/soft-fp/op-4.h: Update from
726 2007-01-15 Tom Tromey <tromey@redhat.com>
728 * doc/sourcebuild.texi (libgcj Tests): Don't mention jacks.
729 * doc/install.texi (Testing): Don't mention jacks.
730 (Configuration): Document --enable-java-maintainer-mode. Move
731 --with-java-home to libgcj-specific section. Document
733 (Prerequisites): Mention --enable-java-maintainer-mode, ecj1.
735 2007-01-15 Jan Hubicka <jh@suse.cz>
737 * tree-ssa-dce.c (DCE_TODOs): New.
738 (propagate_necessity): Return if something changed.
739 (eliminate_unnecessary_stmts): Likewise.
740 (perform_tree_ssa_dce): Return TODO flags when needed.
741 (pass_dce, pass_dce_loop, pass_cd_dce): Remove TODO flags.
743 2007-01-15 Uros Bizjak <ubizjak@gmail.com>
745 * config/i386/i386.md (fyl2xxf3_i387): Rename from fyl2x_xf3.
746 (fyl2x_extend<mode>xf3_i387): New insn pattern.
747 (log<mode>2): Rename from logsf2 and logdf2 and macroize insn
748 patterns using X87MODEF12 mode macro. Extend operand 1
749 to XFmode. Use SSE_FLOAT_MODE_P to disable patterns for SSE math.
750 (log10<mode>2): Ditto.
751 (log2<mode>2): Ditto.
752 (log1p<mode>2): Ditto.
753 (logb<mode>2): Ditto.
754 (fyl2xp1xf3_i387): Rename from fyl2xp1_xf3.
755 (fyl2xp1_extend<mode>xf3_i387): New insn pattern.
756 (*fxtractxf3_i387): Rename from *fxtractxf3.
757 (fxtract_extend<mode>xf3_i387): New insn pattern.
758 (ilogbsi2): Use match_dup 3, not match_operand:XF 3.
759 * config/i386/i386.c (ix86_emit_i387_log1p): Use gen_fyl2xp1xf3_i387()
760 and gen_fyl2xxf3_i387().
762 2007-01-14 Zdenek Dvorak <dvorakz@suse.cz>
764 * loop-unswitch.c (unswitch_loop): Do not call fix_loop_placement.
765 * cfgloopmanip.c (fix_loop_placement): Made static. Use
766 get_loop_exit_edges. Changed return type to bool.
767 * cfgloop.h (fix_loop_placement): Declaration removed.
769 2007-01-14 Dorit Nuzman <dorit@il.ibm.com>
771 * param.h (MIN_VECT_LOOP_BOUND): New.
772 * params.def (MIN_VECT_LOOP_BOUND): New.
773 * tree-vectorizer.c (slpeel_tree_peel_loop_to_edge): Takes another
774 argument - minimum threshold for number of iterations.
775 * tree-vectorizer.h (slpeel_tree_peel_loop_to_edge): Add another
776 argument to declaration.
777 * tree-vect-analyze.c (vect_analyze_operations): Check value of
779 * tree-vect-transform.c (vect_do_peeling_for_loop_bound): Call
780 slpeel_tree_peel_loop_to_edge with additional argument.
781 (vect_do_peeling_for_alignment): Likewise.
782 * doc/invoke.texi (min-vect-loop-bound): Document new param option.
784 2007-01-14 Uros Bizjak <ubizjak@gmail.com>
787 * config/i386/i386.c (print_operand) ['z']: Output 'b' for
790 2007-01-14 Jan Hubicka <jh@suse.cz>
792 * tree-dfa.c (remove_referenced_var): New function.
793 * tree-ssa-live.c (remove_unused_locals): Walk referenced vars and
794 prune referenced vars list too.
795 * tree-flow.h (remove_referenced_var): Declare.
797 2007-01-14 Jan Hubicka <jh@suse.cz>
799 * tree-eh.c (add_stmt_to_eh_region_fn): Do not add call_exprs
801 (remove_stmt_from_eh_region_fn): Do not remove call_exprs.
802 (verify_eh_throw_stmt_node, verify_eh_throw_table_statements): Kill.
803 * except.h (verify_eh_throw_table_statements): Kill prototype.
804 * cfgexpand.c (expand_gimple_basic_block): Propagate Eh regions
805 into call exrepssions.
806 * tree-optimize.c (execute_free_cfg_annotatiosn): Do not call
808 * tree-cfg.c: Include pointer-set.h.
809 (verify_node_sharing): Work on pointer set.
810 (verify_eh_throw_stmt_node): New.
811 (verify_stmts): Use pointers sets, verify throw_stmt.
813 2007-01-13 Zdenek Dvorak <dvorakz@suse.cz>
815 * ipa-reference.c (analyze_function): Consider also addresses taken
818 2007-01-12 Roger Sayle <roger@eyesopen.com>
820 * c-typeck.c (null_pointer_constant_p): Replace use of
821 TREE_CONSTANT_OVERFLOW with TREE_OVERFLOW.
822 (build_c_cast): Likewise.
824 2007-01-12 Roger Sayle <roger@eyesopen.com>
826 * tree.h (force_fit_type_double): Remove unused final argument.
827 * c-common.c (constant_expression_warning): Replace use of
828 TREE_CONSTANT_OVERFLOW with TREE_OVERFLOW.
829 (convert_and_check): Likewise.
830 (shorten_compare): Update call to force_fit_type_double.
831 (c_common_truthvalue_conversion) <INTEGER_CST>: Use integer_zerop.
832 * convert.c (convert_to_pointer): Update call to
833 force_fit_type_double.
834 * fold-const.c (force_fit_type_double): Remove overflowed_const
836 (int_const_binop, fold_convert_const_int_from_int,
837 fold_convert_const_int_from_real, fold_div_compare,
838 fold_sign_changed_comparison, fold_unary, fold_negate_const,
839 fold_abs_const, fold_not_const): Remove the final argument from
840 calls to force_fit_type_double.
842 2007-01-12 Andrew Pinski <andrew_pinski@playstation.sony.com>
844 * configure.ac: Set insn to "nop" for spu-*-* also.
845 * configure: Regenerate.
847 2007-01-12 Olga Golovanevsky <olga@il.ibm.com>
849 * builtins.def : Add BUILT_IN_FREE.
851 2007-01-12 Jan Hubicka <jh@suse.cz>
853 PR tree-optimization/30443
854 * tree-inline.c (tree_function_versioning): Do not optimize when
855 cloning for inlining.
857 2007-01-12 Zdenek Dvorak <dvorakz@suse.cz>
859 * doc/loop.texi: Document recording of loop exits.
860 * cfgloopmanip.c (loopify, duplicate_loop): Use alloc_loop.
861 (update_single_exits_after_duplication,
862 update_single_exit_for_duplicated_loop,
863 update_single_exit_for_duplicated_loops): Removed.
864 (duplicate_loop_to_header_edge): Do not call
865 update_single_exits_after_duplication and
866 update_single_exit_for_duplicated_loops.
867 (loop_version): Do not update single_exit information.
868 (fix_loop_structure): Use record_loop_exits instead of
869 mark_single_exit_loops.
870 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update
871 the lists of loop exits.
872 * cfghooks.c (redirect_edge_and_branch, redirect_edge_and_branch_force,
873 split_edge, merge_blocks): Update the lists of loop exits.
874 * modulo-sched.c (sms_schedule): Pass LOOPS_HAVE_RECORDED_EXITS to
876 * loop-init.c (loop_optimizer_init): Call record_loop_exits instead
877 of mark_single_exit_loops.
878 (loop_optimizer_finalize): Call release_recorded_exits.
879 * tree-ssa-loop.c (tree_loop_optimizer_init): Pass
880 LOOPS_HAVE_RECORDED_EXITS to loop_optimizer_init.
881 * tree-vectorizer.c (slpeel_tree_duplicate_loop_to_edge_cfg): Do not
882 update single exit information.
883 * lambda-code.c (perfect_nestify): Ditto.
884 * cfgloop.c (flow_loop_free): Destroy the list of exits of the loop.
885 (mark_single_exit_loops): Removed.
886 (alloc_loop, loop_exit_hash, loop_exit_eq, loop_exit_free,
887 get_exit_descriptions, rescan_loop_exit, record_loop_exits,
888 dump_recorded_exit, dump_recorded_exits, release_recorded_exits): New
890 (get_loop_exit_edges, single_exit): Use recorded exit lists.
891 (add_bb_to_loop, remove_bb_from_loops): Update the lists of loop exits.
892 (verify_loop_structure): Verify consistency of the exit lists.
893 (flow_loops_find): Use alloc_loop. Initialize exits hash.
894 (set_single_exit): Removed.
895 * cfgloop.h (struct loop_exit): New function.
896 (struct loop): single_exit_ field replaced by exits field.
897 (LOOPS_HAVE_MARKED_SINGLE_EXITS): Replaced by LOOPS_HAVE_RECORDED_EXITS.
898 (struct loops): Added exits hash.
899 (mark_single_exit_loops, set_single_exit): Declaration removed.
900 (release_recorded_exits, record_loop_exits, rescan_loop_exit): Declare.
902 2007-01-12 Richard Sandiford <richard@codesourcery.com>
904 * doc/invoke.texi: Avoid use of @headitem.
906 2007-01-12 Richard Sandiford <richard@codesourcery.com>
908 * cse.c (cse_insn): Move HAVE_CC0 code after declarations.
910 2007-01-12 Richard Sandiford <richard@codesourcery.com>
912 * doc/install.texi: Fix m68k-*-* anchor and add m68k-*-* to the
915 2007-01-12 Nathan Sidwell <nathan@codesourcery.com>
916 Richard Sandiford <richard@codesourcery.com>
918 * doc/invoke.texi: Document -mno-bitfield, -mno-rtd and -mno-short.
919 * config/m68k/m68k.opt: Resort options.
920 (mbitfield, mrtd, mshort): Remove RejectNegative properties.
922 2007-01-12 Nathan Sidwell <nathan@codesourcery.com>
923 Richard Sandiford <richard@codesourcery.com>
925 * doc/invoke.texi: Document the macros that are defined by
926 m68k's -mtune and -mhard-float options.
927 * config/m68k/m68k-protos.h (m68k_cpp_cpu_ident) Declare.
928 (m68k_cpp_cpu_family): Likewise.
929 * config/m68k/m68k.h (TARGET_CPU_CPP_BUILTINS): Add a full set
930 of __ucfv*__ macros. Define __mcffpu__ if generating code for
931 ColdFire FPUs. Define __mcf_cpu_* and __mcf_family_* macros.
932 * config/m68k/m68k.c (m68k_cpp_cpu_ident): New function.
933 (m68k_cpp_cpu_family): Likewise.
935 2007-01-12 Richard Sandiford <richard@codesourcery.com>
937 * config/m68k/m68k.h (TARGET_CPU_CPP_BUILTINS): Treat all mc68*
938 macros besides mc68000 as tuning macros. Use a switch statement
939 to set them and mcpu32.
941 2007-01-12 Julian Brown <julian@codesourcery.com>
943 * config/m68k/m68k.h: Use TARGET_68040 instead of TARGET_68040_ONLY.
944 (TARGET_68040_ONLY): Rename to...
945 (TARGET_68040): ...this.
946 * config/m68k/m68k.c: Use TARGET_68040 instead of TARGET_68040_ONLY.
947 * config/m68k/m68k.md: Likewise.
949 2007-01-12 Julian Brown <julian@codesourcery.com>
950 Nathan Sidwell <nathan@codesourcery.com>
951 Richard Sandiford <richard@codesourcery.com>
953 * config.gcc (m680[012]0-*-*, m68k*-*-*): Set m68k_cpu_ident to
954 the -mcpu= argument associated with the --with-cpu setting.
955 Define M68K_DEFAULT_TUNE to the default -mtune= option,
956 if different from the one implied by the -mcpu setting.
957 Accept --with-cpu=FOO if FOO is listed in m68k-devices.def,
958 using mcpu=FOO as the default CPU option. Set target_cpu_default2.
959 * doc/invoke.texi: Mention ColdFire in the introduction to the
960 m68k options. Document the new -march, -mcpu, -mtune, -mdiv,
961 -mno-div and -mhard-float options. Make -m68881 a synonym for
962 -mhard-float. Document the previously-undocumented -m5206e,
963 -m528x, -m5307 and -m5407 options. Tweak the existing option
964 documentation for consistency.
965 * doc/install.texi: Mention new --with-cpu arguments.
966 * config/m68k/m68k.h (OPTION_DEFAULT_SPECS): Only use the
967 default CPU if neither -mcpu nor -march are specified.
968 (ASM_CPU_SPEC): Pass down -mcpu and -march options.
969 (TARGET_CPU_CPP_BUILTINS): Set __mcfisa*__ macros from
970 TARGET_ISA*. Set the legacy __mcf*__ cpu macros in the same way,
971 using m68k_tune to decide between families that implement the
972 same ISA. Use m68k_tune to set __mcfv4e__.
973 (FL_BITFIELD, FL_68881, FL_COLDFIRE, FL_CF_HWDIV, FL_CF_MAC)
974 (FL_CF_EMAC, FL_CF_EMAC_B, FL_CF_USP, FL_CF_FPU, FL_ISA_68000)
975 (FL_ISA_68010, FL_ISA_68020, FL_ISA_68040, FL_ISA_A, FL_ISA_B)
976 (FL_ISA_C, FL_ISA_MMU): New macros.
977 (MASK_COLDFIRE): Delete.
978 (TARGET_68010, TARGET_68020, TARGET_68040_ONLY, TARGET_COLDFIRE)
979 (TARGET_ISAB): Redefine in terms of m68k_cpu_flags.
980 (TARGET_68881, TARGET_COLDFIRE_FPU): Redefine in terms of m68k_fpu.
981 (TARGET_HARD_FLOAT): Do not define here.
982 (TARGET_ISAAPLUS, TARGET_ISAC): New macros.
983 (TUNE_68000): New macro.
984 (TUNE_68000_10): Redefine in terms of TUNE_68000 and TUNE_68010.
985 (TUNE_68010, TUNE_68030, TUNE_68040, TUNE_68060, TUNE_CPU32)
986 (TUNE_CFV2): Redefine in terms of m68k_tune.
987 (uarch_type, target_device, fpu_type): New enums.
988 (m68k_cpu, m68k_tune, m68k_fpu, m68k_cpu_flags): Declare.
989 * config/m68k/m68k.c (TARGET_DEFAULT): Remove MASK_68881.
990 (FL_FOR_isa_00, FL_FOR_isa_10, FL_FOR_isa_20, FL_FOR_isa_40)
991 (FL_FOR_isa_cpu32, FL_FOR_isa_a, FL_FOR_isa_aplus, FL_FOR_isa_b)
992 (FL_FOR_isa_c): New macros.
993 (m68k_isa): New enum.
994 (m68k_target_selection): New structure.
995 (all_devices, all_isas, all_microarchs): New tables.
996 (m68k_cpu_entry, m68k_arch_entry, m68k_tune_entry, m68k_cpu)
997 (m68k_tune, m68k_fpu, m68k_cpu_flags): New variables.
998 (MASK_ALL_CPU_BITS): Delete.
999 (m68k_find_selection): New function.
1000 (m68k_handle_option): Handle -mcpu=, -march= and -mtune=.
1001 Map the legacy target options to a combination of the new ones.
1002 (override_options): Set m68k_cpu, m68k_tune, m68k_fpu and
1003 m68k_cpu_flags. Handle M68K_DEFAULT_TUNE. Use m68k_cpu_flags
1004 to derive default MASK_BITFIELD, MASK_CF_HWDIV and MASK_HARD_FLOAT
1006 * config/m68k/m68k.opt (m5200, m5206e, m528x, m5307, m5407, mcfv4e)
1007 (m68010, m68020, m68020-40, m68020-60, m68030, m68040): Remove Mask
1009 (m68881, msoft-float): Change mask from 68881 to HARD_FLOAT.
1010 (march=, mcpu=, mdiv, mhard-float, mtune=): New options.
1011 * config/m68k/m68k-devices.def: New file.
1013 2007-01-12 Richard Sandiford <richard@codesourcery.com>
1014 Nathan Sidwell <nathan@codesourcery.com>
1016 * config/m68k/m68k.h (ASM_CPU_SPEC, ASM_SPEC, EXTRA_SPECS)
1017 (SUBTARGET_EXTRA_SPECS): New macros.
1018 * config/m68k/linux.h (ASM_SPEC): Remove CPU flags;
1019 use %(asm_cpu_spec) instead.
1020 * config/m68k/m68k-none.h (ASM_SPEC): Likewise.
1021 * config/m68k/openbsd.h (ASM_SPEC): Likewise.
1022 * config/m68k/netbsd-elf.h (ASM_SPEC): Likewise.
1023 (EXTRA_SPECS): Rename to...
1024 (SUBTARGET_EXTRA_SPECS): ...this.
1026 2007-01-12 Nathan Sidwell <nathan@codesourcery.com>
1027 Richard Sandiford <richard@codesourcery.com>
1028 Julian Brown <julian@codesourcery.com>
1030 * config.gcc (m68k-*-aout*, m68k-*-coff*, m68020-*-elf*, m68k-*-elf*)
1031 (m68k-*-uclinux*, m68k-*-linux*, m68k-*-rtems*): Set default_m68k_cpu
1032 to the configuration's default CPU.
1033 (m68010-*-netbsdelf*, m68k*-*-netbsdelf*, m68k*-*-openbsd*): Likewise.
1034 Remove default masks.
1035 (m680[012]0-*-*): Set the default with_cpu to the first part of
1037 (m68k*-*-*): Set the default with_cpu to m$default_m68k_cpu.
1038 (m68k*-*-linux): Extend the --with-cpu handling to...
1039 (m680[012]0-*-*, m68k*-*-*): ...these configurations. Allow m68000
1040 and m68010. Don't set target_cpu_default2.
1041 * doc/install.texi: Document --with-cpu for m68k.
1042 * config/m68k/m68k.h (OPTION_DEFAULT_SPECS): Define.
1043 * config/m68k/m68k-none.h (TARGET_CPU_DEFAULT, M68K_CPU_m68k)
1044 (M68K_CPU_m68000, M68K_CPU_m68010, M68K_CPU_m68020, M68K_CPU_m68030)
1045 (M68K_CPU_m68040, M68K_CPU_m68302, M68K_CPU_m68332, TARGET_DEFAULT)
1046 (ASM_CPU_DEFAULT_SPEC, CC1_CPU_DEFAULT_SPEC): Delete.
1047 (ASM_SPEC): Remove use of %(asm_cpu_default).
1048 (EXTRA_SPECS, SUBTARGET_EXTRA_SPECS, MULTILIB_DEFAULTS): Delete.
1049 * config/m68k/linux.h (TARGET_DEFAULT): Delete.
1050 (CPP_SPEC): Merge definitions. Do not handle __HAVE_68881__ here.
1051 * config/m68k/netbsd-elf.h (TARGET_OS_CPP_BUILTINS): Define
1052 __HAVE_FPU__ if TARGET_HARD_FLOAT.
1053 (TARGET_DEFAULT): Delete.
1054 (EXTRA_SPECS): Delete cpp_cpu_default_spec, cpp_cpu_spec,
1055 cpp_fpu_spec, asm_default_spec and netbsd_cpp_spec.
1056 (CPP_CPU_SPEC): Delete.
1057 (TARGET_VERSION): Merge definitions, using TARGET_68010 to pick
1058 the appropriate string.
1059 (CPP_CPU_DEFAULT_SPEC, ASM_DEFAULT_SPEC, CPP_FPU_SPEC): Delete.
1060 (CPP_SPEC): Define to NETBSD_CPP_SPEC.
1061 (ASM_SPEC): Don't use %(asm_default_spec).
1062 * config/m68k/m68k.c (TARGET_DEFAULT_TARGET_FLAGS): Remove
1063 TARGET_DEFAULT and add MASK_68881.
1064 * config/m68k/m68k.md: Remove mention of TARGET_DEFAULT from comments.
1066 2007-01-12 Richard Sandiford <richard@codesourcery.com>
1068 * config.gcc (m68010-*-netbsdelf*): Add MASK_68010.
1069 (m68k*-*-netbsdelf*, m68k*-*-openbsd*, m68k*-linux*): Add
1070 MASK_68010 alongside MASK_68020.
1071 * doc/invoke.texi: Document -m68010.
1072 * config/m68k/m68k.opt (m68010): New.
1073 * config/m68k/m68k.h (TARGET_CPU_CPP_BUILTINS): Define mc68010
1075 (TUNE_68010): New macro.
1076 * config/m68k/m68k-none.h (M68K_CPU_m68k, M68K_CPU_m68010)
1077 (M68K_CPU_m68020, M68K_CPU_m68030, M68K_CPU_m68040)
1078 (M68K_CPU_m68332): Add MASK_68010.
1079 * config/m68k/linux.h (TARGET_DEFAULT): Add MASK_68010 to
1080 fallback definition.
1081 * config/m68k/netbsd-elf.h (CPP_CPU_SPEC): Remove now-redundant
1083 * config/m68k/m68k.c (MASK_ALL_CPU_BITS): Add MASK_68010.
1084 (m68k_handle_option): Handle OPT_m68010. Add MASK_68010
1085 to all entries that use MASK_68020.
1086 (output_move_simode_const, output_move_himode, output_move_qimode)
1087 (output_move_stricthi, output_move_strictqi): Use TARGET_68010
1088 instead of TARGET_68020 to select clr behavior. Remove comment
1089 about there being no TARGET_68010.
1090 * config/m68k/m68k.md: Likewise throughout.
1092 2007-01-12 Julian Brown <julian@codesourcery.com>
1094 * config/m68k/m68k.h (TARGET_ISAB): New macro.
1095 * config/m68k/m68k.c: Use TARGET_ISAB rather than TARGET_CFV4.
1096 * config/m68k/m68k.md: Likewise.
1098 2007-01-12 Julian Brown <julian@codesourcery.com>
1100 * config/m68k/m68k.h (LEGITIMATE_INDEX_P, LEGITIMIZE_ADDRESS): Use
1101 TARGET_COLDFIRE_FPU instead of TARGET_CFV4E.
1103 2007-01-12 Julian Brown <julian@codesourcery.com>
1105 * config/m68k/m68k.h (TUNE_68040_60): New macro.
1106 * config/m68k/m68k.c (standard_68881_constant_p): Use it.
1107 * config/m68k/m68k.md: Likewise.
1109 2007-01-12 Julian Brown <julian@codesourcery.com>
1110 Richard Sandiford <richard@codesourcery.com>
1112 * config/m68k/m68k.h (TARGET_CPU_CPP_BUILTINS): Use TUNE_68030
1113 instead of TARGET_68030, TUNE_68040 instead of TARGET_68040,
1114 TUNE_68060 instead of TARGET_68060 and TUNE_CPU32 instead of
1116 (TARGET_CPU32): Rename to...
1117 (TUNE_CPU32): ...this.
1118 (TUNE_68000_10, TUNE_68030, TUNE_68040, TUNE_68060)
1119 (TUNE_CFV2): New macros.
1120 * config/m68k/netbsd-elf.h (LONG_DOUBLE_TYPE_SIZE): Simplify;
1121 remove conditions that are implied by TARGET_68020.
1122 * config/m68k/m68k.c (m68k_output_function_prologue): Use TUNE_68040
1123 instead of TARGET_68040 and TUNE_CPU32 instead of TARGET_CPU32.
1124 (m68k_output_function_epilogue): Likewise.
1125 (m68k_rtx_costs): Likewise. Use TUNE_68060 instead of TARGET_68060
1126 and TUNE_CFV2 instead of TARGET_5200. Use TUNE_68000_10 instead of
1127 "!TARGET_68020 && !TARGET_COLDFIRE" to choose between 68000 and
1128 non-68000 timings. Refactor multiplication and division costs.
1129 (output_addsi3): Use TUNE_68040 instead of TARGET_68040 and
1130 TUNE_CPU32 instead of TARGET_CPU32.
1131 (standard_68881_constant_p): Use TUNE_68040 instead of TARGET_68040
1132 and TUNE_68060 instead of TARGET_68060.
1133 * config/m68k/m68k.md: Use TUNE_68040 instead of TARGET_68040,
1134 TUNE_68060 instead of TARGET_68060, and TUNE_CPU32 instead of
1136 (movsi_const0): Use TUNE_68000_10 rather than "!TARGET_68020
1137 && !TARGET_COLDFIRE" to choose between moveq and clr.
1138 Likewise in the unnamed movsf pattern.
1139 (ashlsi_17_24, lshrsi_17_24): Guard with TUNE_68000_10 rather than
1140 "!TARGET_68020 && !TARGET_COLDFIRE". Likewise the unnamed
1143 2007-01-12 Richard Sandiford <richard@codesourcery.com>
1145 * config/m68k/m68k.h (TARGET_CPU_CPP_BUILTINS): Increase amount
1146 of tabbing before backslashes.
1148 2007-01-11 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
1150 * pa-linux.h (ASM_OUTPUT_INTERNAL_LABEL): Undefine.
1151 * pa.h (ASM_OUTPUT_LABEL): Output colon when using GAS.
1152 (ASM_OUTPUT_INTERNAL_LABEL): Define.
1154 2007-01-11 Zdenek Dvorak <dvorakz@suse.cz>
1156 * tree-ssa-loop-ivopts.c (extract_cond_operands): Split from
1157 find_interesting_uses_cond.
1158 (find_interesting_uses_cond): Use extract_cond_operands.
1159 (rewrite_use_compare): Use extract_cond_operands and
1160 force_gimple_operand_bsi. Do not call update_stmt.
1161 (determine_use_iv_cost_condition): Use extract_cond_operands.
1162 Return cheaper of using original bound and changing the exit bound.
1164 2007-01-11 Zdenek Dvorak <dvorakz@suse.cz>
1166 PR tree-optimization/29516
1167 * tree-ssa-address.c (tree_mem_ref_addr, add_to_parts,
1168 most_expensive_mult_to_index, addr_to_parts,
1169 create_mem_ref, maybe_fold_tmr): Make the type of
1170 fields of TARGET_MEM_REF sizetype.
1171 (move_fixed_address_to_symbol, move_pointer_to_base):
1173 * tree.def (TARGET_MEM_REF): Add comment on types of
1176 2007-01-11 Joseph Myers <joseph@codesourcery.com>
1178 * c-common.c (vector_types_convertible_p): Treat opaque types as
1179 always convertible if they have the same size, but not otherwise.
1181 2007-01-11 Steven Bosscher <steven@gcc.gnu.org>
1183 * ifcvt.c (struct noce_if_info): Add comments to the fields.
1184 Remove the b_unconditional field.
1185 (noce_try_sign_mask): Do not look at b_unconditional.
1186 (noce_process_if_block): Do not use merge_if_blocks. Update
1187 the CFG here. Do not set b_unconditional.
1188 (cond_move_process_if_block): Likewise.
1189 (find_cond_trap): Likewise.
1190 (check_cond_move_block): Require simple jump insns at the end
1193 2007-01-11 Jan Hubicka <jh@suse.cz>
1195 PR tree-optimization/1046
1196 * tree-tailcall.c (suitable_for_tail_call_opt_p): Use TREE_ADDRESSABLE
1197 when alias info is not ready.
1198 (pass_tail_recursion): Do not require aliasing.
1199 * tree-ssa-copyrename.c (pass_rename_ssa_cop): Likewise.
1200 * tree-ssa-ccp.c (pass_ccp, pass_fold_builtins): Likewise.
1201 * tree-ssa-copy.c (pass_copy_prop): Likewise.
1202 * tree-ssa-forwprop.c (pass_forwprop): Likewise.
1203 * tree-ssa-dce.c (pass_dce, pass_dce_loop, pass_cd_dce): Likewise.
1204 * passes.c (init_optimization_passes): Execute rename_ssa_copies,
1205 ccp, forwprop, copy_prop, merge_phi, copy_prop, dce and tail recursion
1207 * tree-ssa-operands.c (add_virtual_operand, get_indirect_ref_operand):
1208 When aliasing is not build, mark statement as volatile.
1210 2007-01-11 Tom Tromey <tromey@redhat.com>
1212 PR preprocessor/15185, PR preprocessor/20989:
1213 * doc/cppopts.texi <-MT>: Update description of algorithm for
1214 computing default target.
1215 <-M, -MD>: Reword "basename" text.
1217 2007-01-11 Roger Sayle <roger@eyesopen.com>
1219 * builtins.c (expand_builtin_pow, expand_builtin_powi,
1220 fold_builtin_cabs, fold_builtin_sqrt, fold_builtin_trunc,
1221 fold_builtin_floor, fold_builtin_ceil, fold_builtin_round,
1222 fold_builtin_int_int_roundingfn, fold_builtin_bitop,
1223 fold_builtin_bswap, real_constp, fold_builtin_pow,
1224 fold_builtin_powi, fold_builtin_signbit, fold_builtin_copysign,
1225 do_mpfr_arg1, do_mpfr_arg2, do_mpfr_arg3, do_mpfr_sincos): Replace
1226 uses of the macro TREE_CONSTANT_OVERFLOW with TREE_OVERFLOW.
1227 * convert.c (convert_to_pointer): Likewise.
1228 * expr.c (highest_pow2_factor, expand_expr_real_1): Likewise.
1229 * fold-const.c (force_fit_type, fold_negate_expr, int_const_binop,
1230 const_binop, fold_convert_const_int_from_int,
1231 fold_convert_const_int_from_real,
1232 fold_convert_const_real_from_real, sign_bit_p,
1233 optimize_minmax_comparison, extract_muldiv_1, fold_div_compare,
1234 fold_sign_changed_comparison, fold_unary, fold_comparison,
1235 fold_binary, multiple_of_p, tree_Expr_non_zero_p,
1236 fold_negate_const, fold_abs_const, fold_not_const): Likewise.
1237 * print-tree.c (print_node_brief, print_node): Likewise.
1238 * stor-layout.c (place_field, layout_type): Likewise.
1239 * tree-chrec.c (keep_cast): Likewise.
1240 * tree.c (build_vector, build_real, build_real_from_int_cst,
1241 build_complex): Likewise.
1243 2007-01-11 Roger Sayle <roger@eyesopen.com>
1245 * tree.h (TREE_CONSTANT_OVERFLOW): Obsolete. For the time being,
1246 treat TREE_CONSTANT_OVERFLOW as a synonym of TREE_OVERFLOW.
1248 2007-01-11 Paolo Bonzini <bonzini@gnu.org>
1250 * configure.ac (strict1_warn): Rename to strict_warn.
1251 (WERROR, --enable-werror, symlink hacks, stage1_cflags,
1252 cc_set_by_configure, quoted_cc_set_by_configure,
1253 stage_prefix_set_by_configure, quoted_stage_prefix_set_by_configure,
1254 all_boot_languages, all_stagestuff): Remove.
1255 (target_list): Remove bootstrap targets.
1256 * Makefile.in (quickstrap): Unconditionally make a synonym of all.
1257 (BOOT_LANGUAGES, STAGE1_CFLAGS, STAGE1_CHECKING,
1258 REMAKEFLAGS, FLAGS_TO_PASS, PREPEND_DOTDOT_TO_RELATIVE_PATHS,
1259 SUBDIR_FLAGS_TO_PASS, WERROR_FLAGS, STRICT2_WARN, LANG_STAGESTUFF,
1260 VOL_FILES, POSTSTAGE1_FLAGS_TO_PASS, STAGE2_FLAGS_TO_PASS,
1261 STAGEPROFILE_FLAGS_TO_PASS, STAGEFEEDBACK_FLAGS_TO_PASS, stage1_build,
1262 stage1_copy, stage2_build, stage2_copy, stageprofile_build,
1263 stageprofile_copy, stage3_build, stage3_copy, stagefeedback_build,
1264 stagefeedback_copy, stage4_build, clean_s1, clean_sw, bootstrap,
1265 bootstrap-lean, bootstrap2, bootstrap2-lean, bootstrap3,
1266 bootstrap3-lean, bootstrap4, bootstrap4-lean, unstage1, unstage2,
1267 unstage3, unstage4, unstageprofile, unstagefeedback, restage, restage2,
1268 restage3, restage4, restageprofile, restagefeedback, bubbleestrap,
1269 cleanstrap, unstrap, restrap, *compare, *compare3, *compare4,
1270 *compare-lean, *compare3-lean, *compare4-lean, stage1-start, stage1,
1271 stage2-start, stage2, stage3-start, stage3, stage4-start, stage4,
1272 stageprofile-start, stageprofile, stagefeedback-start, stagefeedback,
1273 risky-stage1, risky-stage2, risky-stage3, risky-stage4): Remove.
1274 (ORDINARY_FLAGS_TO_PASS): Rename to FLAGS_TO_PASS.
1275 (STAGECOPYSTUFF, STAGEMOVESTUFF): Consolidate into MOSTLYCLEANFILES.
1276 (mostlyclean): Adjust.
1277 (clean, distclean): Don't mention bootstrap stuff.
1278 * configure: Regenerate.
1279 * ada/config-lang.in, cp/config-lang.in, forttran/config-lang.in,
1280 java/config-lang.in, objc/config-lang.in, objcp/config-lang.in,
1281 treelang/config-lang.in (stagestuff): Remove.
1282 * doc/sourcebuild.texi (stage1, stage2, stage3, stage4,
1283 stageprofile, stagefeedback, stagestuff): Remove mention.
1285 2007-01-11 Nick Clifton <nickc@redhat.com>
1287 * config/mcore/predicates.md (mcore_general_movesrc_operand):
1289 (mcore_general_movdst_operand): Do not accept CONST_INTs.
1290 (mcore_arith_K_S_operand): Run the test for the S constraint not
1291 the test for the M constraint.
1292 (mcore_addsub_operand): Do not accept integer values that are
1293 larger than 32 bits.
1294 * config/mcore/mcore.md: Remove unused constraints from split.
1295 (andsi3): Use HOST_WIDE_INT instead of int to hold an INTVAL.
1297 (allocate_stack): Likewise.
1298 * config/mcore/mcore.c (mcore_print_operand): Restrict output of P
1299 operands to 32 bits.
1300 (mcore_const_costs): Use HOST_WIDE_INT instead of int to hold an
1302 (mcore_and_cost, mcore_modify_comparison, const_ok_for_mcore,
1303 mcore_const_ok_for_inline, mcore_const_trick_uses_not,
1304 try_constant_tricks, mcore_num_ones, mcore_num_zeros,
1305 mcore_output_bclri, mcore_output_andn, output_inline_const,
1306 mcore_output_move, mcore_output_movedouble): Likewise.
1307 (mcore_output_cmov): Use CONST_OK_FOR_M and CONST_OK_FOR_N.
1308 (output_inline_const): Likewise.
1309 (output_inline_const): Fix format strings used in sprintf
1311 * config/mcore/mcore-protos.h: Update prototypes for changed
1312 functions in mcore.c.
1313 * config/mcore/mcore.h (CONST_OK_FOR_I): Cast values to
1314 HOST_WIDE_INT and not int.
1315 (CONST_OK_FOR_J, CONST_OK_FOR_K, CONST_OK_FOR_L, CONST_OK_FOR_M,
1316 CONST_OK_FOR_N): Likewise.
1317 (LEGITIMATE_CONSTANT_P): Also check CONSTANT_P.
1318 (GO_IF_LEGITIMATE_INDEX): Use HOST_WIDE_INT instead of int to hold
1321 2007-01-10 Jan Hubicka <jh@suse.cz>
1323 * tree-vrp.c (remove_range_assertions): Release defs.
1324 * tree-ssa-loop-ivopts.c (rmeove_statement): Likewise.
1325 * tree-ssa-dom.c (remove_stmt_or_phi): Likewise.
1327 2007-01-10 Paul Brook <paul@codesourcery.com>
1329 * config/arm/arm.c (arm_rtx_costs_1): Handle mutiply-subtract.
1330 * config/arm/arm.md (mulsi3subsi): New insn.
1332 2007-01-10 Zdenek Dvorak <dvorakz@suse.cz>
1334 * tree-ssa-loop-manip.c (tree_unroll_loop): Make it a wrapper over ...
1335 (tree_transform_and_unroll_loop): New.
1336 * tree-flow.h (transform_callback, tree_transform_and_unroll_loop):
1339 2007-01-10 Robert Kennedy <jimbob@google.com>
1341 * fold-const.c (fold_comparison): Fold comparisons like (x *
1342 1000 < 0) to (x < 0).
1344 2007-01-10 Ian Lance Taylor <iant@google.com>
1346 * tree-pretty-print.c (dump_generic_node): Print parentheses when
1347 operands have the same priority.
1349 2007-01-10 Tom Tromey <tromey@redhat.com>
1351 * fold-const.c (fold_truthop): Don't check can_use_bit_fields_p.
1352 (fold_binary): Likewise.
1353 * langhooks.c (lhd_can_use_bit_fields_p): Removed.
1354 * langhooks-def.h (lhd_can_use_bit_fields_p): Removed.
1355 (LANG_HOOKS_CAN_USE_BIT_FIELDS_P): Removed.
1356 (LANG_HOOKS_INITIALIZER): Remove LANG_HOOKS_CAN_USE_BIT_FIELDS_P.
1357 * langhooks.h (struct lang_hooks): Removed field
1358 'can_use_bit_fields_p'.
1360 2007-01-10 Ralf Corsépius <ralf.corsepius@rtems.org>
1362 * config/bfin/t-bfin, config/bfin/t-bfin-elf: Remove GCC_CFLAGS.
1364 2007-01-10 Razya Ladelsky <razya@il.ibm.com>
1366 * function.c (get_last_funcdef_no): New function.
1367 * function.h (get_last_funcdef_no): Declare.
1368 * tree-inline.c (initialize_cfun): Add initialization.
1369 (tree_function_versioning): Cleanup.
1371 2007-01-10 Jan Hubicka <jh@suse.cz>
1373 * tree-inline.c (setup_one_parameter): Do not propagate into abnormal
1376 2007-01-10 Sa Liu <saliu@de.ibm.com>
1377 Ben Elliston <bje@au.ibm.com>
1379 * spu.h (STACK_SAVE_AREA): Use VOIDmode for SAVE_FUNCTION, SImode
1380 for SAVE_NONLOCAL and Pmode for any other save level.
1381 * spu-protos.h (spu_restore_stack_block): Declare.
1382 * spu.md (save_stack_block): Remove.
1383 (restore_stack_block): Call spu_restore_stack_block.
1384 * spu.c (spu_restore_stack_block): New function.
1385 (spu_expand_epilogue): Remove old comment.
1387 2007-01-09 Zdenek Dvorak <dvorakz@suse.cz>
1389 PR tree-optimization/30322
1390 * tree-ssa-loop-ivopts.c (fold_affine_expr, iv_value): Removed.
1391 (cand_value_at): Return the value as aff_tree.
1392 (may_eliminate_iv): Convert the bound from aff_tree to tree.
1393 * tree-affine.c (aff_combination_add_cst, aff_combination_add_product,
1394 aff_combination_mult): New functions.
1395 (aff_combination_add): Use aff_combination_add_cst.
1396 (aff_combination_convert): Allow conversions to a wider type.
1397 (tree_to_aff_combination): Handle BIT_NOT_EXPR.
1398 * tree-affine.h (aff_combination_mult): Declare.
1400 2007-01-09 Carlos O'Donell <carlos@codesourcery.com>
1402 * doc/tm.texi: Update documentation to reflect reality of exec
1403 and start file search behaviours. Update copyright year.
1404 * doc/invoke.texi: Explain how GCC_EXEC_PREFIX is used to find
1405 header file directories.
1407 2007-01-09 Uros Bizjak <ubizjak@gmail.com>
1409 * config/i386/i386.md (*sinxf2): Rename to *sinxf2_i387.
1410 (*cosxf2): Rename to cosxf2_i387.
1411 (*sindf2, *sinsf2): Extend operand 1 to XFmode. Macroize patterns
1412 using X87MODEF12 mode macro. Rename patterns to
1413 *sin_extend<mode>xf2_i387. Use SSE_FLOAT_MODE_P to disable patterns
1415 (*cosdf2, *cossf2): Ditto.
1416 (sincosdf3, sincossf3): Ditto. Rewrite corresponding splitters
1417 to match extended input operands.
1418 (sincos<mode>3): New expander.
1419 (*sinextendsfdf2, *cosextendsfdf2, *sincosextendsfdf3): Remove
1420 insn patterns and corresponding splitters.
1422 2007-01-09 Kaz Kojima <kkojima@gcc.gnu.org>
1424 * config/sh/t-linux (TARGET_LIBGCC2_CFLAGS): Delete.
1425 (SHLIB_MAPFILES, SHLIB_LINK, SHLIB_INSTALL): Likewise.
1427 2007-01-09 Nicolas Pitre <nico@cam.org>
1430 * arm/ieee754-df.S (Lad_s): Also test the low word of X for zero.
1432 2007-01-08 Geoffrey Keating <geoffk@apple.com>
1434 * target.h (struct gcc_target): New field library_rtti_comdat.
1435 * target-def.h (TARGET_CXX_LIBRARY_RTTI_COMDAT): New.
1436 (TARGET_CXX): Add TARGET_CXX_LIBRARY_RTTI_COMDAT.
1437 * doc/tm.texi (C++ ABI): Document TARGET_CXX_LIBRARY_RTTI_COMDAT.
1438 * config/darwin.h (TARGET_CXX_LIBRARY_RTTI_COMDAT): Define.
1440 2007-01-08 Geoffrey Keating <geoffk@apple.com>
1442 * doc/invoke.texi (Optimize Options): Correct description of -O0.
1444 2007-01-08 Richard Guenther <rguenther@suse.de>
1446 * tree.h (force_fit_type_double): Export.
1447 (force_fit_type): Remove.
1448 * fold-const.c (force_fit_type_double): New function.
1449 (force_fit_type): Remove.
1450 (int_const_binop): Use it.
1451 (fold_convert_const_int_from_int): Likewise.
1452 (fold_convert_const_int_from_real): Likewise.
1453 (fold_div_compare): Likewise.
1454 (fold_sign_changed_comparison): Likewise.
1455 (fold_unary): Likewise.
1456 (fold_negate_const): Likewise.
1457 (fold_abs_const): Likewise.
1458 (fold_not_const): Likewise.
1459 * c-common.c (shorten_compare): Use force_fit_type_double.
1460 * convert.c (convert_to_pointer): Likewise.
1462 2007-01-08 Richard Guenther <rguenther@suse.de>
1464 * tree.h (build_int_cst_wide_type): Export.
1465 * tree.c (build_int_cst_wide_type): New function.
1466 (build_int_cst_wide): Fix comment.
1467 * builtins.c (fold_builtin_object_size): Use build_int_cst
1468 to build -1 or 0 of the correct type. Use fit_double_type
1469 to check for overflow.
1470 * fold-const.c (optimize_bit_field_compare): Use build_int_cst_type
1472 (decode_field_reference): Likewise.
1473 (all_ones_mask_p): Likewise.
1474 (native_interpret_int): Use build_int_cst_wide_type.
1475 (fold_binary): Use build_int_cst_type to build an all-ones
1477 * stor-layout.c (set_sizetype): Use build_int_cst_wide_type.
1479 2007-01-08 Daniel Jacobowitz <dan@codesourcery.com>
1481 * config/pa/t-pa64 (libgcc_stub.a): Use $(T).
1483 2007-01-09 Ben Elliston <bje@au.ibm.com>
1485 * genautomata.c (STATS_OPTION): New option.
1486 (stats_flag): New flag.
1487 (gen_automata_option): Handle it.
1488 (initiate_automaton_gen): Ditto.
1489 (write_automata): Output statistics only if stats_flag is
1490 set. Likewise, output time statistics only if time_flag is set.
1491 * doc/md.texi (Processor pipeline description): Document new flag.
1493 2007-01-08 Richard Guenther <rguenther@suse.de>
1495 * builtins.c (fold_builtin_int_roundingfn): Use fit_double_type.
1496 * tree.c (build_int_cst_type): Likewise.
1497 (size_in_bytes): Don't call force_fit_type on the result.
1498 (int_fits_type_p): Use fit_double_type.
1499 * fold-const.c (fit_double_type): New function.
1500 (force_fit_type): Use it.
1501 * tree.h (fit_double_type): Export.
1503 2007-01-08 Jan Hubicka <jh@suse.cz>
1505 * tree-vectorizer.c (gate_increase_alignment): Fix return type.
1506 * ipa.c (function_and_variable_visibility): Fix return type.
1508 2007-01-08 Richard Guenther <rguenther@suse.de>
1510 * tree-ssa-ccp.c (maybe_fold_offset_to_array_ref): Use type
1511 of offset to build the index.
1512 * tree-pretty-print.c (dump_generic_node): Don't build negated
1513 const just for printing.
1514 * c-pretty-print.c (pp_c_integer_constant): Likewise.
1515 * builtins.c (fold_builtin_int_roundingfn): Check if result
1516 fits the type by using force_fit_type and comparing the result.
1517 * predict.c (predict_loops): Use compare_tree_int for comparison.
1518 * tree.c (build_int_cst): Fall back to integer_type_node for
1520 (build_int_cst_wide): Assert type is non-null.
1522 2007-01-08 Roberto Costa <roberto.costa@st.com>
1524 * tree-vrp.c (extract_range_from_cond_expr): New.
1525 (extract_range_from_expr): Handle COND_EXPR nodes used as expressions.
1526 * tree-ssa-ccp.c (get_maxval_strlen): Handle COND_EXPR nodes used
1528 (fold_stmt): Bug fix, avoid infinite recursion when folding COND_EXPRs.
1529 * tree-ssa-forwprop.c (simplify_cond, forward_propagate_into_cond,
1530 tree_ssa_forward_propagate_single_use_vars): Handle COND_EXPR nodes
1531 used as expressions.
1532 * tree-object-size.c (cond_expr_object_size): New.
1533 (collect_object_sizes_for): Handle COND_EXPR nodes used as expressions.
1535 2007-01-08 Jan Hubicka <jh@suse.cz>
1537 * tree-ssa-forwprop.c (forward_propagate_into_cond,
1538 tree_ssa_forward_propagate_single_use_va): Release defs of propagated
1541 2007-01-08 Richard Guenther <rguenther@suse.de>
1543 PR tree-optimization/23603
1544 * tree-vrp.c (set_value_range_to_truthvalue): New function.
1545 (extract_range_from_binary): Fall back to truthvalue instead of
1546 varying for TRUTH_*_EXPR.
1547 (extract_range_from_comparison): Fall back to truthvalue instead of
1549 (vrp_visit_phi_node): Don't adjust new range bounds to +INF/-INF
1550 if all visited PHI values were constant.
1552 2007-01-08 Jan Hubicka <jh@suse.cz>
1554 * cgraphunit.c (cgraph_process_new_functions): Reset reachable flag.
1555 (cgraph_analyze_function): break out from ...
1556 (cgraph_finalize_compilation_unit): ... here.
1557 (cgraph_expand_function): Remove forgoten commented out line.
1558 (cgraph_optimize): Analyze functions.
1560 2007-01-08 Jan Hubicka <jh@suse.cz>
1562 * tree-pas.h (TODO_remove_function): New flag.
1563 (TODO_update*): Renumber.
1564 (pass_ipa_increase_alignment,
1565 pass_ipa_function_and_variable_visibility): New passes.
1566 * cgraphunit.c (cgraph_increase_alignment): Move to tree-vectorizer.c
1567 (cgraph_function_and_variable_visibility): Move to ipa.c
1568 (cgraph_optimize): Don't call cgraph_function_and_variable_visibility,
1569 cgraph_increase_alignment.
1570 * ipa-inline.c (cgraph_decide_inlining): Don't push timevar.
1571 (cgraph_decide_inlining_incrementally): Push TV_INTEGRATION before
1572 calling tree-inline.
1573 (cgraph_early_inlining): Do not call cgraph_remove_unreachable_nodes.
1574 (pass_ipa_inline, pass_early_ipa_inlining): Set TODO_remove_functions
1575 * tree-vectorizer.c (increase_alignment): Move here from cgraphunit.c
1576 (gate_increase_alignment): New function.
1577 (pass_ipa_increase_alignment): New pass.
1578 * ipa.c: Inline tree-pass.h and timevar.h
1579 (function_and_variable_visibility): Move here from cgraphunit.c
1580 * tree-optimize.c (pass_early_local_passes): Add TODO_remove_functions.
1581 * passes.c (init_optimization_passes): Add the two new passes.
1582 (execute_todo): Handle cgraph_remove_functions.
1584 2007-01-08 Nick Clifton <nickc@redhat.com>
1586 * config/frv/predicates.md (reg_or_0_operand): Accept
1589 2007-01-08 Ralf Corsépius <ralf.corsepius@rtems.org>
1591 * config/bfin/rtems.h, config/bfin/t-rtems: New.
1592 * config.gcc: Add bfin*-rtems*.
1594 2007-01-08 Mark Shinwell <shinwell@codesourcery.com>
1596 * c.opt: Add -flax-vector-conversions.
1597 * c-typeck.c (convert_for_assignment): Pass flag to
1598 vector_types_convertible_p to allow emission of note.
1599 (digest_init): Likewise.
1600 * c-opts.c: Handle -flax-vector-conversions.
1601 * c-common.c (flag_lax_vector_conversions): New.
1602 (vector_types_convertible_p): Unless -flax-vector conversions
1603 has been passed, disallow conversions between vectors with
1604 differing numbers of subparts and/or element types. If such
1605 a conversion is disallowed, possibly emit a note on the first
1606 occasion only to inform the user of -flax-vector-conversions.
1607 The new last argument specifies this.
1608 * c-common.h (flag_lax_vector_conversions): New.
1609 (vector_types_convertible_p): Add extra argument.
1610 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use
1611 char_type_node for V*QI type vectors.
1612 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
1613 Update to satisfy new typechecking rules.
1614 * config/rs6000/altivec.h (vec_cmple): Use vec_cmpge, for both
1616 * doc/invoke.texi (C Dialect Options): Document
1617 -flax-vector-conversions.
1619 2007-01-08 Mark Shinwell <shinwell@codesourcery.com>
1621 PR tree-optimization/29877
1622 * tree-ssa-ter.c (is_replaceable_p): Deem assignments with
1623 a register variable on the RHS to not be replaceable.
1625 2007-01-08 Chen Liqin <liqin@sunnorth.com.cn>
1626 * config/score/t-score-elf (MULTILIB_OPTIONS): Change.
1627 * config/score/predicates.md (const_uimm5, sr0_operand, const_simm12,
1628 const_simm15, const_pow2, const_npow2): Added.
1629 * config/score/misc.md (insv, extv, extzv, movmemsi,
1630 move_lbu_a/b, mov_lhu_a/b etc): Added and fix some bug.
1631 * config/score/score.c (score_address_cost, score_select_cc_mode):
1633 Change CONST_OK_FOR_LETTER_P/EXTRA_CONSTRAINT define.
1634 Update score_rtx_costs for MACRO TARGET_RTX_COSTS.
1635 Update score_print_operand.
1636 * config/score/score.h (DATA_ALIGNMENT, SELECT_CC_MODE): Added.
1637 Adjust register allocate order and update some macro define.
1638 * config/score/score-mdaux.c (mdx_unaligned_load, mdx_unsigned_store,
1639 mdx_block_move_straight, mdx_block_move_loop_head,
1640 mdx_block_move_loop_body, mdx_block_move_loop_foot, mdx_block_move_loop,
1641 mdx_block_move): Added.
1642 (mdx_movsicc, mdp_select_add_imm, mdp_select, mds_zero_extract_andi,
1643 mdp_limm): Updated and fix some bug and typo.
1644 * config/score/score.md (movqi/hi/si, add/sub/zero/ext): Updated.
1645 (movsf, movdf, doloop_end): Added.
1647 2007-01-08 Kazu Hirata <kazu@codesourcery.com>
1649 * config/arm/arm.c, config/arm/arm.h, config/arm/arm.md,
1650 config/arm/thumb2.md: Fix comment typos.
1651 * doc/extend.texi: Fix a typo.
1653 2007-01-07 Eric Christopher <echristo@apple.com>
1655 * configure.ac: Check for __stack_chk_fail for darwin.
1656 * configure: Regenerate.
1658 2007-01-07 Richard Guenther <rguenther@suse.de>
1660 * tree-vrp.c (extract_range_from_assert): CSE calls to
1661 compare_values where possible.
1662 (extract_range_from_unary_expr): Likewise.
1664 2007-01-07 Anatoly Sokolov <aesok@post.ru>
1666 * config/avr/avr-protos.h (call_insn_operand): Delete prototype.
1667 * config/avr/avr.c (call_insn_operand): Delete function.
1668 * config/avr/avr.md (*pushqi, *pushhi, *pushsi, *pushsf): Use REG_SP
1669 instead of register number. Use predicates.
1670 * config/avr/predicates.md (const0_operand, reg_or_0_operand,
1671 call_insn_operand): Add.
1673 2007-01-06 Jan Hubicka <jh@suse.cz>
1675 * tree-pass.h (pass_build_cgraph_edges): Declare.
1676 * cgraphunit.c (record_refernece): Move to cgraphbuild.c
1677 (visited_nodes): Remove.
1678 (cgraph_create_edges): Move to cgraphbuild.c; rename to
1679 build_cgrpah_edges; make visited_nodes local.
1680 (cgraph_process_new_functions): DO not call initialize_inline_failed.
1681 (record_references_in_initializer): Move to cgraphbuild.c
1682 (initialize_inline_failed, rebuild_cgraph_edges,
1683 pass_rebuild_cgraph_edges): Move to cgraphbuild.c.
1684 (verify_cgraph_node): Make visited_nodes local.
1685 (cgraph_analyze_function): Do not call cgraph_create_edges and
1686 initialize_inline_failed.
1687 (cgraph_expand_function): Do not call cgraph_lower_function;
1688 assert that function is already lowered.
1689 * Makefile.in (cgraphbuild.o): New.
1690 * passes.c (init_optimization_passes): Add pass_build_cgraph_edges
1691 at the end of lowering passes.
1693 2007-01-06 Steven Bosscher <steven@gcc.gnu.org>
1695 * ifcvt.c (cond_move_convert_if_block): New function, code
1696 factored out from...
1697 (cond_move_process_if_block): ...here. Call the new function
1698 on the THEN and ELSE blocks.
1699 (merge_if_block): Do not copy global_live_at_end, merge_blocks
1700 already takes care of this.
1702 2007-01-05 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
1705 * tree.h (TREE_OVERFLOW_P): New.
1706 * c-typeck.c (parser_build_unary_op): Warn only if result
1707 overflowed and operands did not.
1708 (parser_build_binary_op): Likewise.
1709 (convert_for_assignment): Remove redundant overflow_warning.
1710 * c-common.c (overflow_warning): Don't check or set TREE_OVERFLOW.
1712 2007-01-05 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
1714 * c-typeck.c (store_init_value): Split over two lines to follow
1715 the GNU coding style.
1717 2007-01-05 Benjamin Kosnik <bkoz@redhat.com>
1719 * c-cppbuiltin.c (c_cpp_builtins): __GXX_EXPERIMENTAL_CPP0X__ to
1720 __GXX_EXPERIMENTAL_CXX0X__.
1721 * doc/cpp.texi: Same.
1723 2007-01-05 Richard Guenther <rguenther@suse.de>
1726 * tree.c (get_narrower): Do not construct COMPONENT_REFs
1727 with mismatched types. Instead explicitly build a
1728 conversion NOP_EXPR.
1730 2007-01-05 Ian Lance Taylor <iant@google.com>
1732 * c-common.c (decl_with_nonnull_addr_p): New function.
1733 (c_common_truthvalue_conversion): Call it.
1734 * c-typeck.c (build_binary_op): Likewise.
1735 * c-common.h (decl_with_nonnull_addr_p): Declare.
1737 2007-01-05 Jakub Jelinek <jakub@redhat.com>
1740 * libgcc2.c (__divdc3): Compare c and d against 0.0 instead of
1743 2007-01-05 Joel Brobecker <brobecker@adacore.com>
1745 * doc/install.texi (Final install): Document the fact that
1746 the GNAT runtime should not be stripped.
1748 2007-01-04 Jan Hubicka <jh@suse.cz>
1750 * tree-inline.c (fold_marked_statements): Update operand caches
1751 and EH after folding
1753 2007-01-04 Ian Lance Taylor <iant@google.com>
1755 * c-common.c (check_function_nonnull): Whitespace fix.
1757 2007-01-04 Jan Hubicka <jh@suse.cz>
1759 * tree-optimize.c (execute_fixup_cfg): Correct previously mistakely
1760 comitted older version of patch.
1761 (pass_fixup_cfg): Add TODOs to verify flow and statements, dump
1762 function, celanup cfg and collect garbage.
1764 2007-01-04 Mike Stump <mrs@apple.com>
1766 * Makefile.in (mostlyclean): Don't remove libgcc anymore.
1769 2007-01-04 Eric Christopher <echristo@apple.com>
1771 * libgcc2.c (__bswapsi2): Use SItype.
1772 (__bswapdi2): Use DItype.
1773 * libgcc2.h: Update for above.
1775 2007-01-04 Paul Brook <paul@codesourcery.com>
1777 * config/arm/arm.md (arm_mulsi3, thumb_mulsi3, mulsi3_compare0,
1778 mulsi_compare0_scratch, mulsi3addsi, mulsi3addsi_compare0,
1779 mulsi3addsi_compare0_scratch, mulsidi3adddi, mulsidi3,
1780 umulsidi3, umulsidi3adddi, smulsi3_highpart,
1781 umulsi3_highpart): Make conditional on !arm_arch6.
1782 (arm_mulsi3_v6, thumb_mulsi3_v6, mulsi3_compare0_v6,
1783 mulsi_compare0_scratch_v6, mulsi3addsi_v6, mulsi3addsi_compare0_v6,
1784 mulsi3addsi_compare0_scratch_v6, mulsidi3adddi_v6, mulsidi3_v6,
1785 umulsidi3_v6, umulsidi3adddi_v6, smulsi3_highpart_v6,
1786 umulsi3_highpart_v6): New insns.
1788 2007-01-04 Roger Sayle <roger@eyesopen.com>
1790 * fold-const.c (fold_convert): When casting an expression to void,
1791 fold_ignored_result may discover a GIMPLE_MODIFY_STMT which doesn't
1792 have a type. Instead of attempting to build a NOP_EXPR, return
1793 these "special" trees directly.
1795 2007-01-04 Joseph Myers <joseph@codesourcery.com>
1797 * config/rs6000/rs6000.c (rs6000_rtx_costs): Make adjustment for
1798 MULT inside MINUS as either argument. Use rs6000_cost->dmul -
1799 rs6000_cost->fp not 0 as adjustment for outer NEG.
1801 2007-01-04 Jan Hubicka <jh@suse.cz>
1803 * cgraph.c (cgraph_release_function_body): New function.
1804 (cgraph_remove_node): Use it.
1805 * cgraph.h (cgraph_release_function_body): Declare.
1806 * cgraphunit.c (cgraph_expand_function): Use it.
1807 * ipa.c (cgraph_remove_unreahchable_nodes): Use it.
1808 * tree-ssa.c (delete_tree_ssa): Allow to be called before aliasing
1809 is initialized and while compilation of other function is running.
1810 * tree-optimize.c (execute_free_cfg_annotations): Move code to clear
1811 statement CFG annotations from here to ...
1812 * tree-cfg.c (delete_tree_cfg_annotations): ... here.
1814 2007-01-04 Zdenek Dvorak <dvorakz@suse.cz>
1816 * cfgloop.h (enum li_flags): Make the constants powers of two.
1818 2007-01-04 Jan Hubicka <jh@suse.cz>
1820 * tree-inline.c (copy_bb): Insert new statements to statements_to_fold
1822 (fold_marked_statements): New function.
1823 (optimize_inline_calls, tree_function_versioning): Fold new statements.
1824 * tree-inline.h (copy_body_data): Add statements_to_fold.
1826 2007-01-03 Daniel Jacobowitz <dan@codesourcery.com>
1828 * config.gcc: Mention libgcc/config.host.
1829 * Makefile.in: Update comments mentioning libgcc.
1830 (LIBGCC, INSTALL_LIBGCC, GCC_PARTS, mklibgcc): Delete.
1831 (all.cross, start.encap, rest.encap, rest.cross): Update
1832 dependencies for libgcc move.
1833 (libgcc.mk, LIBGCC_DEPS, libgcov.a, libgcc.a, stmp-multilib)
1834 (clean-target, clean-target-libgcc): Delete.
1835 (srcdirify, GCC_EXTRA_PARTS): New macros.
1836 (libgcc-support, libgcc.mvars): New rules.
1837 (distclean): Remove mention of mklibgcc.
1838 (install): Don't reference INSTALL_LIBGCC.
1839 (install-common): Don't reference EXTRA_PARTS.
1840 (install-libgcc, install-multilib): Delete rules.
1841 * mklibgcc.in: Delete file.
1842 * doc/configfiles.texi: Don't mention mklibgcc.
1844 * config/i386/t-darwin (SHLIB_VERPFX): Delete (moved to libgcc).
1845 * config/i386/t-darwin64 (SHLIB_VERPFX): Likewise.
1846 * config/rs6000/t-darwin (SHLIB_VERPFX): Likewise.
1847 * config/rs6000/t-ppccomm (TARGET_LIBGCC2_CFLAGS, SHLIB_MAPFILES)
1848 (mklibgcc, ldblspecs): Likewise.
1850 * config/i386/t-nwld (libgcc.def, libc.def, libpcre.def)
1851 (posixpre.def): Use $(T).
1852 (SHLIB_EXT, SHLIB_NAME, SHLIB_SLIBDIR_QUAL, SHLIB_DEF, SHLIB_MAP)
1853 (SHLIB_SRC, SHLIB_INSTALL): Delete.
1854 (SHLIB_LINK): Make dummy.
1855 * config/t-slibgcc-darwin: Delete contents except for dummy SHLIB_LINK.
1857 * config/frv/t-linux (EXTRA_MULTILIB_PARTS): Clear.
1859 * config/alpha/t-crtfm: Use $(T) in rules for EXTRA_PARTS.
1860 * config/alpha/t-vms, config/alpha/t-vms64, config/fr30/t-fr30,
1861 config/i386/t-rtems-i386, config/ia64/t-ia64, config/rs6000/t-beos,
1862 config/rs6000/t-newas, config/sparc/t-elf: Likewise.
1864 * configure.ac (all_outputs): Remove mklibgcc.
1865 * configure: Regenerated.
1867 2007-01-03 Josh Conner <jconner@apple.com>
1870 * calls.c (compute_argument_addresses): Set stack and stack_slot
1871 for partial args, too.
1872 (store_one_arg): Use locate.size.constant for the size when
1873 generating a save_area.
1875 2007-01-03 Robert Kennedy <jimbob@google.com>
1877 * tree-cfg.c (tree_merge_blocks): Release SSA_NAME phi results
1878 whose definitions are deleted due to basic block merging.
1880 2007-01-03 Paul Brook <paul@codesourcery.com>
1883 * config/arm/arm.c (output_return_instruction): Pop PC in interrupt
1885 (use_return_insn): Return 0 for Thumb interrupt functions.
1886 (print_multi_reg): Add rfe argument for IRQ returns.
1887 (arm_output_epilogue): Pop interrupt return address directly into PC.
1888 (arm_expand_prologue): Only adjust IRQ return address in Arm mode.
1890 2007-01-03 Paul Brook <paul@codesourcery.com>
1892 Merge from sourcerygxx-4_1.
1893 * config/arm/thumb2.md: New file.
1894 * config/arm/elf.h (JUMP_TABLES_IN_TEXT_SECTION): Return True for
1896 * config/arm/coff.h (JUMP_TABLES_IN_TEXT_SECTION): Ditto.
1897 * config/arm/aout.h (ASM_OUTPUT_ADDR_VEC_ELT): Add !Thumb-2 assertion.
1898 (ASM_OUTPUT_ADDR_DIFF_ELT): Output Thumb-2 jump tables.
1899 * config/arm/aof.h (ASM_OUTPUT_ADDR_DIFF_ELT): Output Thumb-2 jump
1901 (ASM_OUTPUT_ADDR_VEC_ELT): Add !Thumb-2 assertion.
1902 * config/arm/ieee754-df.S: Use macros for Thumb-2/Unified asm
1904 * config/arm/ieee754-sf.S: Ditto.
1905 * config/arm/arm.c (thumb_base_register_rtx_p): Rename...
1906 (thumb1_base_register_rtx_p): ... to this.
1907 (thumb_index_register_rtx_p): Rename...
1908 (thumb1_index_register_rtx_p): ... to this.
1909 (thumb_output_function_prologue): Rename...
1910 (thumb1_output_function_prologue): ... to this.
1911 (thumb_legitimate_address_p): Rename...
1912 (thumb1_legitimate_address_p): ... to this.
1913 (thumb_rtx_costs): Rename...
1914 (thumb1_rtx_costs): ... to this.
1915 (thumb_compute_save_reg_mask): Rename...
1916 (thumb1_compute_save_reg_mask): ... to this.
1917 (thumb_final_prescan_insn): Rename...
1918 (thumb1_final_prescan_insn): ... to this.
1919 (thumb_expand_epilogue): Rename...
1920 (thumb1_expand_epilogue): ... to this.
1921 (arm_unwind_emit_stm): Rename...
1922 (arm_unwind_emit_sequence): ... to this.
1923 (thumb2_legitimate_index_p, thumb2_legitimate_address_p,
1924 thumb1_compute_save_reg_mask, arm_dwarf_handle_frame_unspec,
1925 thumb2_index_mul_operand, output_move_vfp, arm_shift_nmem,
1926 arm_save_coproc_regs, thumb_set_frame_pointer, arm_print_condition,
1927 thumb2_final_prescan_insn, thumb2_asm_output_opcode, arm_output_shift,
1928 thumb2_output_casesi): New functions.
1929 (TARGET_DWARF_HANDLE_FRAME_UNSPEC): Define.
1930 (FL_THUMB2, FL_NOTM, FL_DIV, FL_FOR_ARCH6T2, FL_FOR_ARCH7,
1931 FL_FOR_ARCH7A, FL_FOR_ARCH7R, FL_FOR_ARCH7M, ARM_LSL_NAME,
1932 THUMB2_WORK_REGS): Define.
1933 (arm_arch_notm, arm_arch_thumb2, arm_arch_hwdiv, arm_condexec_count,
1934 arm_condexec_mask, arm_condexec_masklen)): New variables.
1935 (all_architectures): Add armv6t2, armv7, armv7a, armv7r and armv7m.
1936 (arm_override_options): Check new CPU capabilities.
1937 Set new architecture flag variables.
1938 (arm_isr_value): Handle v7m interrupt functions.
1939 (user_return_insn): Return 0 for v7m interrupt functions. Handle
1941 (const_ok_for_arm): Handle Thumb-2 constants.
1942 (arm_gen_constant): Ditto. Use movw when available.
1943 (arm_function_ok_for_sibcall): Return false for v7m interrupt
1945 (legitimize_pic_address, arm_call_tls_get_addr): Handle Thumb-2.
1946 (thumb_find_work_register, arm_load_pic_register,
1947 legitimize_tls_address, arm_address_cost, load_multiple_sequence,
1948 emit_ldm_seq, emit_stm_seq, arm_select_cc_mode, get_jump_table_size,
1949 print_multi_reg, output_mov_long_double_fpa_from_arm,
1950 output_mov_long_double_arm_from_fpa, output_mov_double_fpa_from_arm,
1951 output_mov_double_fpa_from_arm, output_move_double,
1952 arm_compute_save_reg_mask, arm_compute_save_reg0_reg12_mask,
1953 output_return_instruction, arm_output_function_prologue,
1954 arm_output_epilogue, arm_get_frame_offsets, arm_regno_class,
1955 arm_output_mi_thunk, thumb_set_return_address): Ditto.
1956 (arm_expand_prologue): Handle Thumb-2. Use arm_save_coproc_regs.
1957 (arm_coproc_mem_operand): Allow POST_INC/PRE_DEC.
1958 (arithmetic_instr, shift_op): Use arm_shift_nmem.
1959 (arm_print_operand): Use arm_print_condition. Handle '(', ')', '.',
1961 (arm_final_prescan_insn): Use extract_constrain_insn_cached.
1962 (thumb_expand_prologue): Use thumb_set_frame_pointer.
1963 (arm_file_start): Output directive for unified syntax.
1964 (arm_unwind_emit_set): Handle stack alignment instruction.
1965 * config/arm/lib1funcs.asm: Remove default for __ARM_ARCH__.
1966 Add v6t2, v7, v7a, v7r and v7m.
1967 (RETLDM): Add Thumb-2 code.
1968 (do_it, shift1, do_push, do_pop, COND, THUMB_SYNTAX): New macros.
1969 * config/arm/arm.h (TARGET_CPU_CPP_BUILTINS): Define __thumb2__.
1970 (TARGET_THUMB1, TARGET_32BIT, TARGET_THUMB2, TARGET_DSP_MULTIPLY,
1971 TARGET_INT_SIMD, TARGET_UNIFIED_ASM, ARM_FT_STACKALIGN, IS_STACKALIGN,
1972 THUMB2_TRAMPOLINE_TEMPLATE, TRAMPOLINE_ADJUST_ADDRESS,
1973 ASM_OUTPUT_OPCODE, THUMB2_GO_IF_LEGITIMATE_ADDRESS,
1974 THUMB2_LEGITIMIZE_ADDRESS, CASE_VECTOR_PC_RELATIVE,
1975 CASE_VECTOR_SHORTEN_MODE, ADDR_VEC_ALIGN, ASM_OUTPUT_CASE_END,
1976 ADJUST_INSN_LENGTH): Define.
1977 (TARGET_REALLY_IWMMXT, TARGET_IWMMXT_ABI, CONDITIONAL_REGISTER_USAGE,
1978 STATIC_CHAIN_REGNUM, HARD_REGNO_NREGS, INDEX_REG_CLASS,
1979 BASE_REG_CLASS, MODE_BASE_REG_CLASS, SMALL_REGISTER_CLASSES,
1980 PREFERRED_RELOAD_CLASS, SECONDARY_OUTPUT_RELOAD_CLASS,
1981 SECONDARY_INPUT_RELOAD_CLASS, LIBCALL_VALUE, FUNCTION_VALUE_REGNO_P,
1982 TRAMPOLINE_SIZE, INITIALIZE_TRAMPOLINE, HAVE_PRE_INCREMENT,
1983 HAVE_POST_DECREMENT, HAVE_PRE_DECREMENT, HAVE_PRE_MODIFY_DISP,
1984 HAVE_POST_MODIFY_DISP, HAVE_PRE_MODIFY_REG, HAVE_POST_MODIFY_REG,
1985 REGNO_MODE_OK_FOR_BASE_P, LEGITIMATE_CONSTANT_P,
1986 REG_MODE_OK_FOR_BASE_P, REG_OK_FOR_INDEX_P, GO_IF_LEGITIMATE_ADDRESS,
1987 LEGITIMIZE_ADDRESS, THUMB2_LEGITIMIZE_ADDRESS,
1988 GO_IF_MODE_DEPENDENT_ADDRESS, MEMORY_MOVE_COST, BRANCH_COST,
1989 ASM_APP_OFF, ASM_OUTPUT_CASE_LABEL, ARM_DECLARE_FUNCTION_NAME,
1990 FINAL_PRESCAN_INSN, PRINT_OPERAND_PUNCT_VALID_P,
1991 PRINT_OPERAND_ADDRESS): Adjust for Thumb-2.
1992 (arm_arch_notm, arm_arch_thumb2, arm_arch_hwdiv): New declarations.
1993 * config/arm/arm-cores.def: Add arm1156t2-s, cortex-a8, cortex-r4 and
1995 * config/arm/arm-tune.md: Regenerate.
1996 * config/arm/arm-protos.h: Update prototypes.
1997 * config/arm/vfp.md: Enable patterns for Thumb-2.
1998 (arm_movsi_vfp): Add movw alternative. Use output_move_vfp.
1999 (arm_movdi_vfp, movsf_vfp, movdf_vfp): Use output_move_vfp.
2000 (thumb2_movsi_vfp, thumb2_movdi_vfp, thumb2_movsf_vfp,
2001 thumb2_movdf_vfp, thumb2_movsfcc_vfp, thumb2_movdfcc_vfp): New.
2002 * config/arm/libunwind.S: Add Thumb-2 code.
2003 * config/arm/constraints.md: Update include Thumb-2.
2004 * config/arm/ieee754-sf.S: Add Thumb-2/Unified asm support.
2005 * config/arm/ieee754-df.S: Ditto.
2006 * config/arm/bpabi.S: Ditto.
2007 * config/arm/t-arm (MD_INCLUDES): Add thumb2.md.
2008 * config/arm/predicates.md (low_register_operand,
2009 low_reg_or_int_operand, thumb_16bit_operator): New.
2010 (thumb_cmp_operand, thumb_cmpneg_operand): Rename...
2011 (thumb1_cmp_operand, thumb1_cmpneg_operand): ... to this.
2012 * config/arm/t-arm-elf: Add armv7 multilib.
2013 * config/arm/arm.md: Update patterns for Thumb-2 and Unified asm.
2015 (UNSPEC_STACK_ALIGN, ce_count): New.
2016 (arm_incscc, arm_decscc, arm_umaxsi3, arm_uminsi3,
2017 arm_zero_extendsidi2, arm_zero_extendqidi2): New
2019 * config/arm/fpa.md: Update patterns for Thumb-2 and Unified asm.
2020 (thumb2_movsf_fpa, thumb2_movdf_fpa, thumb2_movxf_fpa,
2021 thumb2_movsfcc_fpa, thumb2_movdfcc_fpa): New insns.
2022 * config/arm/cirrus.md: Update patterns for Thumb-2 and Unified asm.
2023 (cirrus_thumb2_movdi, cirrus_thumb2_movsi_insn,
2024 thumb2_cirrus_movsf_hard_insn, thumb2_cirrus_movdf_hard_insn): New
2026 * doc/extend.texi: Document ARMv7-M interrupt functions.
2027 * doc/invoke.texi: Document Thumb-2 new cores+architectures.
2029 2007-01-03 Jakub Jelinek <jakub@redhat.com>
2031 * unwind-dw2.c (SIGNAL_FRAME_BIT, EXTENDED_CONTEXT_BIT): Define.
2032 (struct _Unwind_Context): Rename args_size to flags, remove
2033 signal_frame field, add a new args_size field and version field.
2034 (_Unwind_IsSignalFrame, _Unwind_SetSignalFrame,
2035 _Unwind_IsExtendedContext): New inline functions.
2036 (_Unwind_GetGR, _Unwind_SetGR, _Unwind_GetGRPtr, _Unwind_SetGRPtr):
2037 Assume by_value array is only present if _Unwind_IsExtendedContext.
2038 (_Unwind_GetIPInfo, execute_cfa_program, uw_frame_state_for): Use
2039 _Unwind_IsSignalFrame.
2040 (__frame_state_for): Initialize context.flags to EXTENDED_CONTEXT_BIT.
2041 (uw_update_context_1): Use _Unwind_SetSignalFrame.
2042 (uw_init_context_1): Initialize context->flags to
2043 EXTENDED_CONTEXT_BIT.
2044 * config/rs6000/linux-unwind.h (frob_update_context): Use
2045 _Unwind_SetSignalFrame.
2047 2007-01-03 Andrew Pinski <andrew_pinski@playstation.sony.com>
2050 * gimplify.c (gimplify_modify_expr_complex_part): Move below
2051 tree_to_gimple_tuple. Call tree_to_gimple_tuple when we need
2054 2007-01-03 Kazu Hirata <kazu@codesourcery.com>
2056 * config/i386/i386.h (NON_STACK_REG_P, REGNO_OK_FOR_SIREG_P,
2057 REGNO_OK_FOR_DIREG_P, REWRITE_ADDRESS, ASM_OPERAND_LETTER,
2058 RET, AT_SP): Remove.
2059 * config/i386/i386.md (*sse_prologue_save_insn): Use return
2062 * alias.c (init_alias_analysis): Use VEC_safe_grow_cleared.
2063 * cfgbuild.c (find_basic_blocks): Likewise.
2064 * cfgrtl.c (rtl_create_basic_block): Likewise.
2065 * function.c (temp_slots_at_level): Likewise.
2066 * reg-stack.c (stack_regs_mentioned): Likewise.
2067 * regclass.c (allocate_reg_info): Likewise.
2068 * tree-cfg.c (init_empty_tree_cfg, build_tree_cfg, create_bb,
2069 set_bb_for_stmt, move_block_to_fn): Likewise.
2070 * tree-complex.c (tree_lower_complex): Likewise.
2071 * vec.h (VEC_safe_grow_cleared): New.
2073 * cgraphunit.c, tree-ssa-alias.c: Fix comment typos.
2075 2007-01-03 Zdenek Dvorak <dvorakz@suse.cz>
2077 * loop-unswitch.c (unswitch_loop): Pass probabilities to loopify.
2078 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Pass probabilities
2080 * cfgloopmanip.c (scale_loop_frequencies): Export.
2081 (loopify): Scale the frequencies by prescribed coefficients.
2082 (set_zero_probability): New function.
2083 (duplicate_loop_to_header_edge): Improve updating of frequencies.
2084 (lv_adjust_loop_entry_edge, loop_version): Set probabilities
2085 and frequencies according to arguments.
2086 * tree-ssa-loop-manip.c (tree_unroll_loop): Set probabilities
2088 * cfg.c (scale_bbs_frequencies_int): Allow scaling the frequencies up.
2089 * modulo-sched.c (sms_schedule): Set probabilities for entering
2090 versioned loop correctly.
2091 * tree-vect-transform.c (vect_transform_loop): Ditto.
2092 * cfgloop.h (loopify, loop_version): Declaration changed.
2093 (scale_loop_frequencies): Declared.
2095 2007-01-02 Jan Hubicka <jh@suse.cz>
2097 * cgraph.c: Include tree-flow.h
2098 (cgraph_add_new-function): Handle IPA_SSA mode; execute
2100 * cgraph.h (enum cgraph_state): Add CGRAPH_STATE_IPA_SSA.
2101 * tree-pass.h (pass_all_early_optimizations): Declare.
2102 * cgraphunit.c (cgraph_process_new_functions): Add IPA_SSA; execute
2104 (cgraph_analyze_function): Do early_local_passes.
2105 * tree-mudflap.c (mf_decl_cache_locals, mf_build_check_statement_for):
2106 Do not add referenced vars.
2107 * tree-optimize.c (gate_all_optimizations): Do not execute when not in
2109 (gate_all_early_local_passes): New gate.
2110 (pass_early_local_passes): Use new gate.
2111 (execute_early_local_optimizations): New functions.
2112 (gate_all_early_optimizations): New gate.
2113 (pass_all_early_optimizations): New pass.
2114 (execute_free_datastructures): Free SSA only when initialized.
2115 (gate_init_datastructures): Init only when optimizing.
2116 (tree_lowering_passes): Do early local passes when called late.
2117 * tree-profile.c (do_tree_profiling): Don't profile functions added
2119 (do_early_tree_profiling, pass_early_tree_profile): Kill.
2120 * tree-cfg.c (update_modified_stmts): Do not update when operands are
2122 * passes.c (init_optimizations_passes): Reorder so we go into SSA
2123 during early_local_passes.
2124 * Makefile.in (cgraph.o): Add dependency on tree-flow.h.
2127 2007-01-02 Carlos O'Donell <carlos@codesourcery.com>
2129 * Makefile.in: Update copyright year.
2131 2007-01-02 Carlos O'Donell <carlos@codesourcery.com>
2133 * Makefile.in: Export GCC_EXEC_PREFIX before calling $(RUNTEST)
2134 in $(lang_checks) and check-consistency targets.
2136 2007-01-02 Jan Hubicka <jh@suse.cz>
2138 * tree-mudflap.c (mf_decl_cache_locals, mf_build_check_statement_for):
2139 Do not add referenced vars.
2140 * tree-cfg.c (update_modified_stmts): Do not update when SSA operands
2142 * passes.c (init_optimization_passes): Put mudflap_2 after
2143 free_datastructures.
2145 2007-01-02 Jan Hubicka <jh@suse.cz>
2147 * tree-optimize (execute_fixup_cfg): Set after_inlining flag.
2148 Set NOTHROW flag on call statements proved to be nothrow.
2149 Update statement of local calls so new pure/const functions are
2150 updated. Update_ssa when in ssa form. Mark PHI nodes of nonlocal
2152 (tree_rest_of_compilation): Register hooks and initialize bitmap
2153 early. Do not set after_inlining flag.
2155 2007-01-02 Steve Ellcey <sje@cup.hp.com>
2157 * sbitmap.c (HOST_BITS_PER_LONG_LONG): Change to
2158 HOST_BITS_PER_LONGLONG
2160 2007-01-02 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
2163 * c-typeck.c (store_init_value): Don't emit pedantic overflow
2164 warning for non-static initializers.
2166 2007-01-02 Steven Bosscher <steven@gcc.gnu.org>
2168 * config/alpha/alpha.md, arm/arm.c, darwin.c, frv/frv.md,
2169 m32r/m32r.c, m32r/m32r.c, mn10300/mn10300.md, pa/pa.c,
2170 rs6000/rs6000.c, s390/s390.md, sh/sh.md, sparc/sparc.c:
2171 Always use set_unique_reg_note to add REG_EQUAL notes.
2173 2007-01-02 Kazu Hirata <kazu@codesourcery.com>
2176 2007-01-02 Kazu Hirata <kazu@codesourcery.com>
2178 * alias.c (init_alias_analysis): Use VEC_safe_grow_cleared.
2179 * cfgbuild.c (find_basic_blocks): Likewise.
2180 * cfgrtl.c (rtl_create_basic_block): Likewise.
2181 * function.c (temp_slots_at_level): Likewise.
2182 * reg-stack.c (stack_regs_mentioned): Likewise.
2183 * regclass.c (allocate_reg_info): Likewise.
2184 * tree-cfg.c (init_empty_tree_cfg, build_tree_cfg, create_bb,
2185 set_bb_for_stmt, move_block_to_fn): Likewise.
2186 * tree-complex.c (tree_lower_complex): Likewise.
2187 * vec.h (VEC_safe_grow_cleared): New.
2189 2007-01-02 Ian Lance Taylor <iant@google.com>
2191 * c-common.c (c_common_truthvalue_conversion): When warning about
2192 using an assignment as a truth value, set TREE_NO_WARNING.
2194 2007-01-02 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
2197 * c.opt (Wold-style-declaration): New.
2198 * doc/invoke.texi (C-only Warning Options): New.
2199 (Wold-style-declaration): Document it.
2200 (Wextra): Enabled by -Wextra.
2201 * c-opts.c (c_common_post_options): Enabled by -Wextra.
2202 * c-decl.c (declspecs_add_scspec): Replace -Wextra with
2203 -Wold-style-declaration.
2205 2007-01-02 Kazu Hirata <kazu@codesourcery.com>
2207 * alias.c (init_alias_analysis): Use VEC_safe_grow_cleared.
2208 * cfgbuild.c (find_basic_blocks): Likewise.
2209 * cfgrtl.c (rtl_create_basic_block): Likewise.
2210 * function.c (temp_slots_at_level): Likewise.
2211 * reg-stack.c (stack_regs_mentioned): Likewise.
2212 * regclass.c (allocate_reg_info): Likewise.
2213 * tree-cfg.c (init_empty_tree_cfg, build_tree_cfg, create_bb,
2214 set_bb_for_stmt, move_block_to_fn): Likewise.
2215 * tree-complex.c (tree_lower_complex): Likewise.
2216 * vec.h (VEC_safe_grow_cleared): New.
2218 2007-01-02 Douglas Gregor <doug.gregor@gmail.com>
2220 * c-common.c(c_common_nodes_and_builtins): Since variants of
2221 void_type_node get built before it is given a name, we need to
2222 give those variants the name, too.
2223 (complete_array_type): We need to work with the canonical main
2224 type of the array, from which we will build the qualified version.
2225 * params.def (PARAM_VERIFY_CANONICAL_TYPES): New.
2226 * print-tree.c (print_node): Display canonical type information
2228 * stor-layout.c (layout_type): When we don't know the
2229 alignment of a type for which we're building an array, we end up
2230 guessing wrong, so make the type require structural equality.
2231 * tree.c (make_node_stat): When we build a new type, it is its
2233 (build_type_attribute_qual_variant): When building an attribute
2234 variant, its canonical type is the non-attribute variant. However,
2235 if the attributes are target-dependent and they differ, we need to
2236 use structural equality checks for this type.
2237 (build_qualified_type): A qualified type is not equivalent to its
2238 unqualified variant; set the canonical type appropriately.
2239 (build_distinct_type_copy): When building a distinct type from
2240 another type, the new type is its own canonical type.
2241 (build_variant_type_copy): When building a new type variant, we
2242 assume that it is equivalent to the original type.
2243 (build_pointer_type_for_mode): When building a pointer type, also
2244 build a canonical type pointer.
2245 (build_reference_type_for_mode): When building a reference type,
2246 also build a canonical type reference.
2247 (build_index_type): When we can't hash an index type (e.g.,
2248 because its maximum value is negative), the index type requires
2249 structural equality tests.
2250 (build_array_type): Build the canonical form of an array type.
2251 (build_function_type): Function types require structural equality,
2252 because they contain default arguments, attributes, etc.
2253 (build_method_type_directly): Ditto for method types.
2254 (build_offset_type): Build the canonical offset type.
2255 (build_complex_type): Build the canonical vector type.
2256 (make_vector_type): Build the canonical vector type.
2257 * tree.h (TYPE_CANONICAL): New.
2258 (TYPE_STRUCTURAL_EQUALITY_P): New.
2259 (SET_TYPE_STRUCTURAL_EQUALITY): New.
2260 (struct tree_type): Added "canonical" field.
2261 * params.h (VERIFY_CANONICAL_TYPES): New.
2262 * doc/c-tree.texi (TYPE_CANONICAL): Document.
2263 (TYPE_STRUCTURAL_EQUALITY_P): Document.
2264 (SET_TYPE_STRUCTURAL_EQUALITY): Document.
2265 * doc/invoke.texi (verify-canonical-types): Document --param
2266 parameter for verifying canonical types.
2268 2007-01-02 Joseph Myers <joseph@codesourcery.com>
2270 * config.gcc (powerpc-*-eabispe*, powerpc-*-eabisimaltivec*,
2271 powerpc-*-eabisim*, powerpc-*-eabialtivec*, powerpc-*-eabi*,
2272 powerpc-*-rtems*, powerpc-wrs-vxworks, powerpc-wrs-vxworksae,
2273 powerpcle-*-eabisim*, powerpcle-*-eabi*): Add rs6000/e500.h to
2275 * config/rs6000/e500.h: New.
2276 * config/rs6000/eabi.h (TARGET_SPE_ABI, TARGET_SPE, TARGET_E500,
2277 TARGET_ISEL, TARGET_FPRS, TARGET_E500_SINGLE, TARGET_E500_DOUBLE):
2279 * config/rs6000/linuxspe.h (TARGET_SPE_ABI, TARGET_SPE,
2280 TARGET_E500, TARGET_ISEL, TARGET_FPRS, TARGET_E500_SINGLE,
2281 TARGET_E500_DOUBLE): Remove.
2282 * config/rs6000/vxworks.h (TARGET_SPE_ABI, TARGET_SPE,
2283 TARGET_E500, TARGET_ISEL, TARGET_FPRS): Remove.
2284 * config/rs6000/rs6000.h (CHECK_E500_OPTIONS): Define.
2285 * config/rs6000/rs6000.c (rs6000_override_options): Use
2288 2007-01-02 Joseph Myers <joseph@codesourcery.com>
2290 * config/rs6000/rs6000.c (print_operand): Check (TARGET_SPE ||
2291 TARGET_E500_DOUBLE), not TARGET_E500, for %y.
2292 (rs6000_generate_compare, rs6000_emit_sCOND, output_cbranch,
2293 rs6000_emit_cmove): Don't check TARGET_E500.
2294 * config/rs6000/rs6000.md (bunordered, bordered, sunordered,
2295 sordered): Don't check TARGET_E500.
2297 2007-01-01 Eric Christopher <echristo@apple.com>
2299 * config/mips/mips.c (mips_regno_mode_ok_for_base_p): Use
2300 HARD_REGISTER_NUM_P.
2302 2007-01-01 Roger Sayle <roger@eyesopen.com>
2304 * fold-const.c (fold_binary) <EQ_EXPR>: Fold "(X^C1) eq/ne C2" into
2305 "X eq/ne (C1^C2)". Fold "(X^Z) eq/ne (Y^Z)" as "X eq/ne Y" when Z
2306 has no side-effects. Fold "(X^C1) eq/ne (Y^C2)" as "(X^(C1^C2))
2309 2007-01-01 Mike Stump <mrs@apple.com>
2311 * configure.ac: Remove support for building with Apple's gcc-3.1.
2313 2007-01-02 Joseph Myers <joseph@codesourcery.com>
2316 * caller-save.c (add_stored_regs): Only handle SUBREGs if inner
2317 REG is a hard register. Do not modify REG before calling
2319 * rtlanal.c (subreg_get_info): Don't assert size of XMODE is a
2320 multiple of the size of YMODE for certain lowpart cases.
2322 2007-01-01 Andrew Pinski <pinskia@gmail.com>
2325 * gimplify (voidify_wrapper_expr): Update for
2328 2007-01-01 Andreas Schwab <schwab@suse.de>
2331 * config/ia64/ia64.c (ia64_compute_frame_size): Account space for
2332 save of BR0 in extra_spill_size instead of spill_size.
2333 (ia64_expand_prologue): Save BR0 outside of the gr/br/fr spill
2335 (ia64_expand_epilogue): Restore BR0 from its new location.
2337 2007-01-01 Andrew Pinski <pinskia@gmail.com>
2339 * gimplify.c (gimplify_init_constructor <case VECTOR_TYPE>):
2340 Use a temporary variable if the left hand side is not a gimple
2343 2007-01-01 Andrew Pinski <pinskia@gmail.com>
2345 * gimplify.c (gimplify_return_expr): Make the temporary variable
2346 for the return expression, a gimple register variable.
2348 2007-01-01 Jan Hubicka <jh@suse.cz>
2350 * emit-rtl.c (emit_copy_of_insn_after): Do not call copy_insn_1 for
2353 2007-01-01 Mike Stump <mrs@apple.com>
2355 * configure.ac (HAVE_GAS_LITERAL16): Add autoconf check for
2357 * config/darwin.c (machopic_select_rtx_section): Use
2359 (darwin_mergeable_constant_section): Likewise.
2360 * configure: Regenerate.
2361 * config.in: Regenerate.
2363 2007-01-01 Jan Hubicka <jh@suse.cz>
2364 Andrew Pinski <pinskia@gmail.com>
2366 * cgraphunit.c (cgraph_optimize): Call cgraph_add_new_functions
2367 before starting IPA passes.