target.h (globalize_decl_name): New.
[gcc.git] / gcc / ChangeLog
1 2007-01-24 Andreas Krebbel <krebbel1@de.ibm.com>
2
3 * unwind-dw2-fde.c (get_cie_encoding): Replaced _Unwind_Word with
4 _uleb128_t and _Unwind_SWord with _sleb128_t.
5 * unwind-dw2.c (extract_cie_info, execute_stack_op, execute_cfa_program,
6 uw_frame_state_for, uw_update_context_1): Likewise.
7 * unwind-c.c (parse_lsda_header, PERSONALITY_FUNCTION): Likewise.
8 * unwind-pe.h (read_uleb128, read_sleb128,
9 read_encoded_value_with_base): Likewise.
10 * unwind-generic.h: Define _sleb128_t and _uleb128_t types.
11
12 2007-01-24 Richard Guenther <rguenther@suse.de>
13
14 * builtins.c (expand_builtin_cexpi): Get the fndecl
15 for cexp in the correct way.
16
17 2007-01-24 Jan Hubicka <jh@suse.cz>
18
19 * tree-ssa-dce.c (eliminate_unnecesary_stmts): Remove dead LHS of calls.
20
21 2007-01-24 Andreas Krebbel <krebbel1@de.ibm.com>
22
23 * c-cppbuiltin.c (builtin_define_type_sizeof): New function.
24 (c_cpp_builtins): New builtin macros: __SIZEOF_INT__, __SIZEOF_LONG__,
25 __SIZEOF_LONG_LONG__, __SIZEOF_SHORT__, __SIZEOF_POINTER__,
26 __SIZEOF_FLOAT__, __SIZEOF_DOUBLE__, __SIZEOF_LONG_DOUBLE__,
27 __SIZEOF_SIZE_T__, __SIZEOF_WCHAR_T__, __SIZEOF_WINT_T__ and
28 __SIZEOF_PTRDIFF_T__.
29 * doc/cpp.texi: Documentation for the new builtin macros added.
30
31 2007-01-24 Uros Bizjak <ubizjak@gmail.com>
32
33 * config/i386/i386.md (tanxf2, tan<mode>2, atan<mode>2, log<mode>2,
34 log10<mode>2, log2<mode>2, expxf2, exp10xf2, exp2xf2): Use op2
35 instead of operands[2] to avoid access past the end of array.
36
37 2007-01-24 Richard Sandiford <richard@codesourcery.com>
38
39 * reload1.c (emit_reload_insns): Pass the reload register
40 for a non-spill output reload through forget_old_reloads_1.
41
42 2007-01-23 Joseph Myers <joseph@codesourcery.com>
43
44 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Define
45 _SOFT_DOUBLE if doubles use software floating-point.
46 * config/rs6000/libgcc-ppc-glibc.ver: Export additional long
47 double functions if _SOFT_DOUBLE, not _SOFT_FLOAT.
48 * config/rs6000/darwin-ldouble.c: Also compile functions for
49 hard-float without FPRs. Use fmsub function for all __NO_FPRS__
50 cases. Compile extra functions if _SOFT_DOUBLE, not _SOFT_FLOAT.
51 * config/rs6000/linuxspe.h (SUBSUBTARGET_OVERRIDE_OPTIONS): Remove
52 commented-out long double override.
53 (CPP_LONGDOUBLE_DEFAULT_SPEC): Likewise.
54 * config/rs6000/eabispe.h: Likewise.
55 * config/rs6000/rs6000.c (rs6000_override_options): Don't override
56 long double for non-SPE.
57 (rs6000_handle_option): Likewise.
58 (invalid_e500_subreg): Disallow more subregs involding DImode,
59 DFmode, TImode or TFmode.
60 (rs6000_legitimate_offset_address_p): Check TFmode offsets for
61 E500 double.
62 (legitimate_lo_sum_address_p): Also check for TFmode for E500
63 double.
64 (rs6000_legitimize_address): Also handle TFmode for E500 double.
65 (rs6000_legitimize_reload_address): Also handle TFmode for E500
66 double.
67 (rs6000_legitimate_address): Also check for TFmode for E500
68 double.
69 (rs6000_emit_move): Use DFmode subregs of TFmode for E500 double.
70 (spe_build_register_parallel): Handle TFmode and TCmode.
71 (rs6000_spe_function_arg): Handle TFmode and TCmode for E500
72 double.
73 (function_arg): Handle TFmode and TCmode for E500 double.
74 (rs6000_init_libfuncs): Initialize extra libfuncs for soft double
75 in general.
76 (print_operand): Handle TFmode and TImode for %y.
77 (rs6000_generate_compare): Handle TFmode comparisons for E500
78 double.
79 (spe_func_has_64bit_regs_p): Check for TFmode for E500 double.
80 (rs6000_function_value): Handle TFmode and TCmode for E500 double.
81 (rs6000_libcall_value): Handle TFmode and TCmode for E500 double.
82 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Check for
83 TFmode for E500 double.
84 * config/rs6000/rs6000.md (FP): Allow TF for E500 double.
85 (floatsidf2): Enable for E500 double.
86 (movtf_softfloat): Use rs6000_nonimmediate_operand.
87 (extenddftf2): Change to extenddftf2_fprs.
88 (extenddftf2): Call gen_spe_extenddftf2 or gen_extenddftf2_fprs
89 depending on TARGET_E500_DOUBLE.
90 (extendsftf2): Enable for E500 double.
91 (trunctfdf2): Enable for E500 double.
92 (trunctfsf2): Change to trunctfsf2_fprs.
93 (trunctfsf2): Call gen_spe_trunctfsf2 or gen_trunctfsf2_fprs
94 depending on TARGET_E500_DOUBLE.
95 (floatsitf2): Enable for E500 double.
96 (fix_trunctfsi2): Change to fix_trunctfsi2_fprs.
97 (fix_trunctfsi2): Call gen_spe_fix_trunctfsi2 or
98 gen_fix_trunctfsi2_fprs depending on TARGET_E500_DOUBLE.
99 (negtf2): Change to negtf2_internal.
100 (negtf2): New expander.
101 (abstf2): Enable for E500 double. Call gen_spe_abstf2_tst,
102 gen_spe_abstf2_cmp or gen_abstf2_internal depending on
103 TARGET_E500_DOUBLE and flag_unsafe_math_optimizations.
104 (movdi_internal32): Use rs6000_nonimmediate_operand.
105 (unnamed splitter): Likewise.
106 * config/rs6000/spe.md (CMPTFEQ_GPR, TSTTFEQ_GPR, CMPTFGT_GPR,
107 TSTTFGT_GPR, CMPTFLT_GPR, TSTTFLT_GPR): New unspecs.
108 (SPE64TF, DITI): New mode macros.
109 (frob_df_di): Change to frob_<SPE64:mode>_<DITI:mode>; allow more
110 modes.
111 (frob_tf_ti): New.
112 (frob_<mode>_di_2): New.
113 (frob_tf_di_8_2): New.
114 (frob_di_df): Change to frob_di_<mode>; allow more modes.
115 (frob_ti_tf): New.
116 (frob_di_df_2): Change to frob_<DITI:mode>_<SPE64:mode>_2; allow
117 more modes.
118 (frob_ti_<mode>_8_2): New.
119 (frob_ti_tf_2): New.
120 (mov_si<mode>_e500_subreg0, mov_si<mode>_e500_subreg0_2,
121 mov_si<mode>_e500_subreg4, mov_si<mode>_e500_subreg4_2): Allow
122 TFmode.
123 (mov_sitf_e500_subreg8, mov_sitf_e500_subreg8_2,
124 mov_sitf_e500_subreg12, mov_sitf_e500_subreg12_2): New.
125 (spe_trunctfdf2_internal1, spe_trunctfsf2, spe_extenddftf2,
126 spe_fix_trunctfsi2, spe_fix_trunctfsi2_internal,
127 spe_negtf2_internal, spe_abstf2_cmp, spe_abstf2_tst): New.
128 (cmptfeq_gpr, tsttfeq_gpr, cmptfgt_gpr, tsttfgt_gpr, cmptflt_gpr,
129 tsttflt_gp): New.
130
131 2007-01-23 Ian Lance Taylor <iant@google.com>
132
133 * Makefile.in (OBJS-common): Reformat, alphabetize, but put
134 insn-*.o first.
135 (OBJS-archive): Reformat, alphabetize.
136 (OBJS): Change out_object_file to OBJS-md.
137
138 2007-01-23 Uros Bizjak <ubizjak@gmail.com>
139
140 * config/i386/i386.md: Use REG_P, MEM_P, CONST_INT_P, LABEL_P,
141 JUMP_P and CALL_P predicates where applicable.
142 * config/i386/i386.c: Ditto.
143 * config/i386/i386.md: Ditto.
144 * config/i386/mmx.md: Ditto.
145 * config/i386/predicates.md: Ditto.
146
147 2007-01-22 Andreas Schwab <schwab@suse.de>
148
149 * config/m68k/m68k.h: Fix comment.
150
151 2007-01-22 Jan Hubicka <jh@suse.cz>
152
153 * passes.c (init_optimization_passes): Do not rerun
154 pass_early_warn_uninitialized.
155
156 2007-01-22 Richard Guenther <rguenther@suse.de>
157
158 PR tree-optimization/30038
159 * tree-ssa-math-opts.c (maybe_record_sincos): New static helper
160 function.
161 (execute_cse_sincos_1): Likewise.
162 (execute_cse_sincos): Likewise.
163 (gate_cse_sincos): Likewise.
164 (pass_cse_sincos): New pass CSEing sin() and cos() calls using
165 the cexpi() canonicalization of sincos().
166 * tree-pass.h (pass_cse_sincos): Declare.
167 * passes.c (init_optimization_passes): New pass pas_cse_sincos.
168
169 2007-01-21 Eric Botcazou <ebotcazou@libertysurf.fr>
170
171 PR rtl-optimization/29329
172 * combine.c (replaced_rhs_insn): Rename to i2mod.
173 (replaced_rhs_value): Rename to i2mod_new_rhs.
174 (i2mod_old_rhs): New global variable.
175 (combine_instructions): Adjust for above change. Save a copy of
176 the old RHS into i2mod_old_rhs when the contents of a REG_EQUAL
177 note are substituted in the second instruction.
178 (distribute_notes) <REG_DEAD>: Adjust for above change. Do not
179 ditch the note if it pertains to the second eliminated register
180 and this register is mentioned in i2mod_old_rhs.
181
182 Revert:
183 2006-09-12 Eric Botcazou <ebotcazou@libertysurf.fr>
184
185 * combine.c (distribute_notes) <REG_DEAD>: Do not consider SETs past
186 the insn to which the note was originally attached.
187
188 2007-01-21 Jan Hubicka <jh@suse.cz>
189
190 * ipa-inline.c (inlining_mode): Comment, move up.
191 (cgraph_decide_inlining_incrementally): Do not perform inlining
192 itself; fix handling of flattening of self recursive functions.
193 (cgraph_find_cycles): Remove.
194 (cgraph_flatten_node): Remove.
195 (cgraph_decide_inlining): Use incremental inliner to handle flattening.
196 (try_inline): New function.
197 (cgraph_early_inlining): Update call of
198 cgraph_decide_inlining_incrementally. Apply inlining here.
199 (apply_inline): Update call of cgraph_decide_inlining_incrementally.
200
201 2007-01-21 Dirk Mueller <dmueller@suse.de>
202
203 PR bootstrap/30511
204 * tree-vrp.c (check_array_bounds): do not warn
205 about ADDR_EXPR's of ARRAY_REF's which are immediately
206 used in binary expressions.
207
208 2007-01-21 Ira Rosen <irar@il.ibm.com>
209
210 * tree-vectorizer.h (struct _stmt_vec_info): Add new field
211 read_write_dep and macros for its access.
212 * tree-vectorizer.c (new_stmt_vec_info): Initialize the new field.
213 * tree-vect-analyze.c (vect_analyze_data_ref_dependence): Remove
214 argument, call vect_check_interleaving for every independent pair of
215 data-refs. Mark loads that access the same memory location as a store
216 in the loop.
217 (vect_check_dependences): Remove.
218 (vect_analyze_data_ref_dependences): Remove vect_check_dependences
219 call, fix the call to vect_analyze_data_ref_dependence.
220 (vect_analyze_data_ref_access): For statements that access the same
221 data-ref, check that they are not stores; for loads, check that there
222 is no store that access the same location.
223
224 2007-01-20 Roger Sayle <roger@eyesopen.com>
225 Joseph Myers <joseph@codesourcery.com>
226
227 * doc/invoke.texi (-fdump-rtl-combine): Fix under/overfull hbox.
228 (-fdump-rtl-stack): Likewise.
229 (-fno-signed-zeros): Use @minus{} for a minus sign. Correct typo.
230 (-mcheck-zero-division, -mcpu): Fix under/overfull hbox.
231 (-mpt-fixed): Use @minus{} for minus sign.
232 (Using Precompiled Headers): Fix under/overfull hbox.
233
234 2007-01-20 Jan Hubicka <jh@suse.cz>
235
236 * tree-flow.h (struct stmt_ann_d): Move references_memory to proper
237 place within annotation.
238
239 2007-01-20 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
240
241 * pa.c (output_move_double): Change array size of xoperands to 4.
242
243 2007-01-20 Jan Hubicka <jh@suse.cz>
244
245 * tree-tailcall.c (adjust_return_value): Do not use RESULT_DECL
246 as temporary.
247
248 2007-01-19 Ian Lance Taylor <iant@google.com>
249
250 * expmed.c (expand_divmod) [TRUNC_MOD_EXPR, TRUNC_DIV_EXPR]: Cast
251 constant to unsigned HOST_WIDE_INT before negating.
252
253 2007-01-19 Ian Lance Taylor <iant@google.com>
254
255 * tree-ssa-operands.h (struct vuse_vec_d): Change num_vuse field
256 to unsigned.
257 (VUSE_VECT_ELEMENT) [ENABLE_CHECKING]: Use unsigned comparison.
258 (VUSE_ELEMENT_PTR) [ENABLE_CHECKING]: Likewise.
259 (SET_VUSE_VECT_ELEMENT) [ENABLE_CHECKING]: Likewise.
260 (SET_VUSE_ELEMENT_VAR) [ENABLE_CHECKING]: Likewise.
261 (SET_VUSE_ELEMENT_PTR) [ENABLE_CHECKING]: Likewise.
262 (realloc_vdef, realloc_vuse): Change second parameter to
263 unsigned.
264 (ssa_operand_iterator_d): Change vuse_index and mayuse_index
265 fields to unsigned.
266 * tree-ssa-operands.c (realloc_vop): Change num_elem parameter to
267 unsigned. Change x and lim locals to unsigned.
268 (realloc_vdef, realloc_vuse): Change num_elem parameter to
269 unsigned.
270 (finalize_ssa_vuse_ops): Change old_i local to unsigned.
271 (copy_virtual_operands): Change i and n locals to unsigned.
272
273 2007-01-19 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
274
275 PR middle-end/29335
276 * builtins.c (fold_builtin_1): Handle builtin fdim.
277
278 2007-01-20 Jan Hubicka <jh@suse.cz>
279
280 * tree-ssa.c (init_tree_ssa): Do not call init_alias_heapvars.
281 * tree-ssa-structalias.c (compute_points_to_sets): Do call
282 init_alias_heapvars.
283 (init_alias_heapvars): Initialize only when not already initialized.
284 (delete_alias_heapvars): Set heapvar_for_stmt to NULL.
285
286 2007-01-19 Roger Sayle <roger@eyesopen.com>
287
288 * common.opt (fsigned-zeros): New command line option.
289 * flags.h (HONOR_SIGNED_ZEROS): Control via flag_signed_zeros instead
290 of flag_unsafe_math_optimizations.
291 * opts.c (set_fast_math_flags): The -ffast-math command line option
292 implies -fno-signed-zeros.
293 (fast_math_flags_set_p): Likewise.
294
295 * doc/invoke.texi: Document new -fno-signed-zeros option, and update
296 the documentation of -ffast-math appropriately. Wrap long lines.
297
298 2007-01-19 Steve Ellcey <sje@cup.hp.com>
299
300 * system.h (ASM_MAKE_LABEL_LINKONCE): Poison.
301 * varasm.c (globalize_decl): Remove ASM_MAKE_LABEL_LINKONCE ifdef.
302
303 2007-01-19 Tomas Bily <tbily@suse.cz>
304
305 * cgraphunit.c (cgraph_finalize_function): Updating of pid
306 * tree-profile.c:
307 (tree_init_ic_make_global_vars): New function
308 (tree_init_edge_profiler): call of tree_init_ic_make_global_vars
309 (tree_gen_ic_profiler): New function
310 (tree_gen_ic_func_profiler): New function
311 (tree_profiling): Added calling of tree_gen_ic_func_profiler
312 (tree_profile_hooks): Added hook for indirec/virtual calls
313 * value-prof.c (tree_find_values_to_profile): New case for
314 indirect calls
315 (tree_values_to_profile): Call for determining indirect/virtual
316 counters
317 (tree_indirect_call_to_profile): New function
318 (tree_ic_transform): New function
319 (tree_ic): New function
320 (find_func_by_pid): New function
321 (init_pid_map): New function
322 (tree_value_profile_transformations): Added check for
323 indirect/virtual call transformation
324 * value-prof.h (enum hist_type): New counter type for
325 indirect/virtual calls
326 (profile_hooks): Added new hook for profiling indirect/virtual
327 calls
328 * profile.c (instrument_values): New case for indirect/virtual
329 call added
330 * gcov-io.h (GCOV_LAST_VALUE_COUNTER): Changed to 6
331 (GCOV_COUNTER_V_INDIR): New counter type
332 (GCOV_COUNTER_NAMES): New name of counter "indirect" added
333 (GCOV_MERGE_FUNCTIONS): New merge function for indirect/virtual
334 call added
335 * cgraph.c: Definition of cgraph_max_pid
336 (cgraph_create_node): Default init of pid attribute
337 * cgraph.h: Declaration of cgraph_max_pid
338 (struct cgraph_node): Added pid attribute
339 * libgcov.c (__gcov_indirect_call_profiler): New function
340 (__gcov_one_value_profiler_body): New function
341 (__gcov_one_value_profiler): Body was moved to
342 __gcov_one_value_profiler_body and calls it
343
344 2007-01-19 Basile Starynkevitch <basile@starynkevitch.net>
345
346 * doc/gty.texi (Options): Document the mark_hook option to GTY.
347 * gengtype.c (write_types_data, write_func_for_structure,
348 write_types, ggc_wtd, pch_wtd): Add skip_hooks to
349 write_types_data, ggc_wtd, pch_wtd for processing mark_hook.
350 (walk_type, write_func_for_structure): Generate the mark_hook if
351 needed.
352
353 2007-01-19 Jan Hubicka <jh@suse.cz>
354
355 * ipa-inline.c (cgraph_decide_inlining_incrementally): Instead of
356 'early' argument take inlining mode argument specifying whether to
357 inline for size/speeed or all functions; add support for flattening;
358 improve dumpting.
359 (cgraph_early_inlining): Update call of decide_inlining_incrementally.
360
361 2007-01-19 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
362
363 PR c++/17947
364 * toplev.c (warn_deprecated_use): Use %qD instead of %qs to print
365 the name of the declared identifier.
366
367 2007-01-19 Dirk Mueller <dmueller@suse.de>
368
369 * config/i386.h (CONDITIONAL_REGISTER_USAGE): Store
370 result of PIC_OFFSET_TABLE_REGNUM in temporary variable to avoid
371 duplicate evaluation.
372
373 2007-01-19 Uros Bizjak <ubizjak@gmail.com>
374
375 * config/i386/i386.md (acos<mode>2): Rename from acossf2 and acosdf2.
376 Macroize expander using X87MODEF12 mode macro. Extend operand 1
377 to XFMode, use acosxf2 and truncate result to requested mode.
378 Use SSE_FLOAT_MODE_P to disable patterns for SSE math.
379 (asin<mode>2): Similarly, with asin expanders.
380 (*fscalexf4_i387): Rename from *fscalexf4.
381 (expNcorexf3): New expander.
382 (expxf2, exp10xf2, exp2xf2): Use expNcorexf3 expander.
383 (exp<mode>2): Rename from expsf2 and expdf2. Macroize expander using
384 X87MODEF12 mode macro. Extend operand 1 to XFMode, use expxf2 and
385 truncate result to requested mode. Use SSE_FLOAT_MODE_P to disable
386 patterns for SSE math.
387 (exp10<mode>2): Similarly, with exp10 expanders.
388 (exp2<mode>2): Similarly, with exp2 expanders.
389 (expm1<mode>2): Similarly, with expm1 expanders.
390 (ldexp<mode>3): Similarly, with ldexp expanders.
391 (log<mode>2, log10<mode>2, log2<mode>2, log1p<mode>2, logb<mode>2):
392 Use gen_truncxf<mode>2_i387_noop to truncate result.
393
394 2007-01-19 Richard Sandiford <richard@codesourcery.com>
395
396 * config/m68k/m68k.h (PREFERRED_STACK_BOUNDARY): Define to 32
397 for ColdFire targets.
398
399 2007-01-19 Nathan Sidwell <nathan@codesourcery.com>
400 Richard Sandiford <richard@codesourcery.com>
401
402 * config/m68k/m68k.h (M68K_STATIC_CHAIN_REG_NAME): New macro.
403 (INITIALIZE_TRAMPOLINE): Use STATIC_CHAIN_REGNUM.
404 (__transfer_from_trampoline): Use M68K_STATIC_CHAIN_REG_NAME.
405 * config/m68k/m68kelf.h (STATIC_CHAIN_REGNUM): Override.
406 (M68K_STATIC_CHAIN_REG_NAME): Likewise.
407 * config/m68k/netbsd-elf.h (M68K_STATIC_CHAIN_REG_NAME): Likewise.
408
409 2007-01-19 Richard Sandiford <richard@codesourcery.com>
410
411 * config/m68k/m68k.md (adddi_dilshr32): Rename to...
412 (*adddi_dilshr32): ...this. Fix formatting. Remove commented-out
413 non-canonical pattern. Restrict to !TARGET_COLDFIRE.
414 (*adddi_dilshr32_cf): New pattern.
415 (adddi3, subdi3): Remove first alternatives.
416
417 2007-01-19 Richard Sandiford <richard@codesourcery.com>
418
419 * config/m68k/m68k.c (notice_update_cc): If an SFmode move is
420 implemented using move.l, do not use its cc result for floating-point
421 comparisons.
422
423 2007-01-19 Richard Sandiford <richard@codesourcery.com>
424
425 * config/m68k/m68k.h (EXTRA_CONSTRAINT): Stop the 'T' constraint
426 from accepting 's' constraints if flag_pic.
427
428 2007-01-19 Richard Sandiford <richard@codesourcery.com>
429
430 * config/m68k/m68k.md (bordered, bunordered, buneq, bunge, bungt)
431 (bunle, bunlt, bltgt, bordered_rev, bunordered_rev, buneq_rev)
432 (bunge_rev, bungt_rev, bunle_rev, bunlt_rev, bltgt_rev): Change
433 condition from TARGET_68881 to TARGET_HARD_FLOAT.
434
435 2007-01-19 Sandra Loosemore <sandra@codesourcery.com>
436
437 * longlong.h (count_leading_zeros, COUNT_LEADING_ZEROS_0): Add
438 ColdFire alternatives.
439 * config/m68k/m68k.h (CLZ_DEFINED_VALUE_AT_ZERO): New macro.
440 * config/m68k/m68k.md (clzsi2): Define for ColdFire
441 architectures that support ff1 instruction.
442
443 2007-01-19 Richard Sandiford <richard@codesourcery.com>
444 Julian Brown <julian@codesourcery.com>
445
446 * config/m68k/m68k.h (CONST_OK_FOR_LETTER_P): Add an 'R' case.
447 * config/m68k/m68k.md (*movsi_cfv4): Fold into...
448 (*movsi_cf): ...here. Remove unnecessary 'R' from 'Rg'.
449 Add commentary.
450
451 2007-01-19 Richard Sandiford <richard@codesourcery.com>
452
453 * config/m68k/m68k-protos.h (valid_mov3q_const): Take a HOST_WIDE_INT
454 and return a bool.
455 (output_move_const_into_data_reg, output_move_simode_const): Delete.
456 * config/m68k/m68k.c (const_method, const_int_cost): Take a
457 HOST_WIDE_INT instead of an rtx.
458 (m68k_rtx_costs): Update call accordingly.
459 (output_move_const_into_data_reg): Likewise. Fix formatting.
460 (valid_mov3q_const): Take a HOST_WIDE_INT instead of an rtx.
461 Return a bool.
462 (output_move_simode_const): Update calls after above changes.
463 Rework to use automatic variables and predicates like MEM_P.
464 * config/m68k/m68k.md (pushexthisi_const): Update call to
465 valid_mov3q_const.
466
467 2007-01-19 Dirk Mueller <dmueller@suse.de>
468
469 * tree-ssa-alias.c (perform_var_substitution): Fix typo
470 in dump_flags test.
471
472 2007-01-19 Richard Guenther <rguenther@suse.de>
473
474 * builtins.c (expand_builtin_cexpi): Fall back to expanding
475 via cexp in case sincos is not available.
476
477 2007-01-19 Richard Guenther <rguenther@suse.de>
478
479 * doc/tm.texi (TARGET_HAS_SINCOS): Document new target macro.
480 * defaults.h (TARGET_HAS_SINCOS): Default to off.
481 * config/linux.h (TARGET_HAS_SINCOS): Set to on if we have glibc.
482 * config/alpha/linux.h (TARGET_HAS_SINCOS): Likewise.
483 * config/sparc/linux.h (TARGET_HAS_SINCOS): Likewise.
484 * config/sparc/linux64.h (TARGET_HAS_SINCOS): Likewise.
485 * config/rs6000/linux.h (TARGET_HAS_SINCOS): Likewise.
486 * config/rs6000/linux64.h (TARGET_HAS_SINCOS): Likewise.
487
488 2007-01-19 Uros Bizjak <ubizjak@gmail.com>
489
490 * config/i386/i386.md (*fpatanxf3_i387, fpatan_extend<mode>xf3_i387):
491 New insn patterns.
492 (atan2sf3_1, atan2df3_1, atan2xf3_1): Remove insn patterns.
493 (atan2xf3): Directly generate RTL pattern.
494 (atan2<mode>3): Rename from atan2sf3 and atan2df3 and macroize insn
495 patterns using X87MODEF12 mode macro. Use fpatan_extend<mode>xf3_i387
496 and truncate result to requested mode. Use SSE_FLOAT_MODE_P to
497 disable patterns for SSE math.
498 (atan<mode>2): Rename from atansf2 and atandf2 and macroize insn
499 patterns using X87MODEF12 mode macro. Use fpatan_extend<mode>xf3_i387
500 and truncate result to requested mode. Use SSE_FLOAT_MODE_P to
501 disable patterns for SSE math.
502
503 2007-01-19 Alexandre Oliva <aoliva@redhat.com>
504
505 * libgcc-std.ver: Fix typo in %inherit for GCC_4.3.0.
506
507 2007-01-18 Roger Sayle <roger@eyesopen.com>
508
509 * fold-const.c (fold_unary) <VIEW_CONVERT_EXPR>: Optimize away a
510 VIEW_CONVERT_EXPR to the same type as it's operand.
511
512 2007-01-18 David Edelsohn <edelsohn@gnu.org>
513
514 * config/rs6000/darwin-ldouble.c: Only build _SOFT_FLOAT if
515 configured for long double 128.
516
517 2007-01-18 Mike Stump <mrs@apple.com>
518
519 * config/rs6000/rs6000.c (rs6000_emit_vector_compare): Fix build
520 error.
521
522 2007-01-18 Michael Meissner <michael.meissner@amd.com>
523
524 * i386.c (ix86_compute_frame_layout): Make fprintf's in #if 0 code
525 type correct.
526
527 2007-01-18 Jan Hubicka <jh@suse.cz>
528
529 * tree-ssa-operands.c (vop_free_bucket_size): Never return value
530 greater than NUM_VOP_FREE_BUCKETS.
531
532 2007-01-18 Daniel Berlin <dberlin@dberlin.org>
533
534 * tree-ssa-structalias.c: Update comments.
535 (ptabitmap_obstack): Removed.
536 (pta_obstack): New.
537 (oldpta_obstack): Ditto.
538 (stats): Add a few members.
539 (struct variable_info): Remove node, complex, address_taken, and
540 indirect_target members. Add oldsolution member.
541 (new_var_info): Do not initialize removed members.
542 (constraint_expr_type): Remove INCLUDES.
543 (constraint_graph): Add size, implicit_preds, rep,
544 indirect_cycles, eq_rep, label, direct_nodes, and complex members.
545 (FIRST_REF_NODE): New macro.
546 (LAST_REF_NODE): Ditto.
547 (FIRST_ADDR_NODE): Ditto.
548 (find): New function.
549 (unite): Ditto.
550 (dump_constraint): Do not handle INCLUDES.
551 (insert_into_complex): Do not insert duplicate constraints.
552 (condense_varmap_nodes): Renamed and rewritten into ...
553 (merge_node_constraints): This. Also fix bug in handling of
554 offseted copy constraints.
555 (clear_edges_for_node): No longer need to deal with preds at all,
556 or removing associated preds/succs.
557 (merge_graph_nodes): Deal with indirect_cycles.
558 Don't deal with predecessors.
559 (add_implicit_graph_edge): New function.
560 (add_pred_graph_edge): Ditto.
561 (add_graph_edge): Don't deal with predecessors.
562 (build_constraint_graph): Removed.
563 (build_pred_graph): New function.
564 (build_succ_graph): Ditto.
565 (struct scc_info): Removed in_component. Added roots, dfs, and
566 node_mapping. Remove visited_index, unification_queue.
567 (scc_visit): Deal with union-find we do now.
568 Deal with cycles with REF nodes.
569 (collapse_nodes): Renamed and rewritten to ...
570 (unify_nodes): This.
571 (process_unification_queue): Removed.
572 (topo_visit): Cleanup
573 (do_da_constraint): Use find.
574 (do_sd_constraint): Ditto.
575 (do_ds_constraint): Ditto.
576 (do_complex_constraint): Ditto.
577 (init_scc_info): Update for removed and added members.
578 (find_and_collapse_graph_cycles): Renamed and rewritten into ...
579 (find_indirect_cycles): This.
580 (equivalence_class): New variable.
581 (label_visit): New function.
582 (perform_variable_substitution): Rewritten.
583 (free_var_substitution_info): New function.
584 (find_equivalent_node): Ditto.
585 (move_complex_constraints): Ditto.
586 (eliminate_indirect_cycles): Ditto.
587 (solve_graph): Only propagate changed bits.
588 Use indirect cycle elimination.
589 Use find.
590 (tree_id_t): Rename to tree_vi_t, delete id member, add vi member.
591 (tree_id_eq): Renamed to ...
592 (tree_vi_eq): This. Update for member change
593 (insert_id_for_tree): Renamed and rewritten to ...
594 (insert_vi_for_tree): This.
595 (lookup_id_for_tree): Renamed and rewritten to ...
596 (lookup_vi_for_tree): This.
597 (get_id_for_tree): Renamed and rewritten to ...
598 (get_vi_for_tree): Ditto.
599 (get_constraint_exp_from_ssa_var): Update to use get_vi_for_tree.
600 (process_constraint): Don't handle INCLUDES.
601 Remove special ADDRESSOF case.
602 (find_func_aliases): Rewrite to use vi functions instead of id
603 ones.
604 (create_function_info_for): Ditto.
605 (create_variable_info_for): Ditto.
606 (intra_create_variable_infos): Ditto.
607 (merge_smts_into): Ditto.
608 (find_what_p_points_to): Ditto.
609 (init_base_vars): Ditto.
610 (init_alias_vars): Ditto.
611 (remove_preds_and_fake_succs): New function.
612 (dump_sa_points_to_info): Dump new stats.
613 (dump_solution_for_var): Use find.
614 (set_used_smts): Fix formatting.
615 (compute_points_to_sets): Updated for new functions.
616 (ipa_pta_execute): Ditto.
617
618 2007-01-18 Kazu Hirata <kazu@codesourcery.com>
619 Richard Sandiford <richard@codesourcery.com>
620
621 * doc/tm.texi (TARGET_FUNCTION_VALUE): Expand documentation of
622 parallels.
623 * calls.c (expand_call): If the return value is a PARALLEL,
624 extract its first member.
625 * config/m68k/linux.h (FUNCTION_EXTRA_EPILOGUE): Remove.
626 * config/m68k/m68k.c (m68k_output_function_epilogue): Don't
627 use FUNCTION_EXTRA_EPILOGUE.
628 (m68k_function_value): Return a PARALLEL if the return value
629 is of a pointer type.
630 * config/m68k/netbsd-elf.h (current_function_returns_pointer)
631 (FUNCTION_EXTRA_EPILOGUE): Remove.
632 * config/m68k/m68k.md (D0_REG): New constant.
633
634 2007-01-18 Kazu Hirata <kazu@codesourcery.com>
635
636 * config/m68k/m68k.c (m68k_output_function_epilogue): Don't
637 output a NOP for empty epilogues.
638
639 2007-01-18 Richard Sandiford <richard@codesourcery.com>
640
641 * config/m68k/m68k.c (m68k_use_return_insn): Update comments
642 before function. Extend register save check to include all
643 registers, not just integer ones.
644
645 2007-01-18 Kazu Hirata <kazu@codesourcery.com>
646
647 * config/m68k/m68k-protos.h (use_return_insn): Rename to...
648 (m68k_use_return_insn): ...this.
649 * config/m68k/m68k.h (USE_RETURN_INSN): Delete.
650 * config/m68k/m68k.c (use_return_insn): Rename to...
651 (m68k_use_return_insn): ...this.
652 * config/m68k/m68k.md (return): Use m68k_use_return_insn instead
653 of USE_RETURN_INSN.
654
655 2007-01-18 Nathan Sidwell <nathan@codesourcery.com>
656
657 * config/m68k/fpgnulib.c (__truncdfsf2): Implement round to
658 nearest even, fix denormal rounding overflow.
659
660 2007-01-18 Richard Sandiford <richard@codesourcery.com>
661
662 * config/m68k/m68k.md (movsf_cf_hard): Use fsmove instead of
663 f%$smove and f%$move.
664 (movdf_cf_hard): Use fdmove for cases 0 and 3 and fmove for case 1.
665 (extendsfdf2_cf): Use fdmove instead of f%&move.
666 (truncdfsf2_cf): Use fsmove instead of f%$smove.
667 (add<mode>3_cf, sub<mode>3_cf): Use <FP:prec> instead of <FP:round>.
668
669 2007-01-18 Richard Sandiford <richard@codesourcery.com>
670
671 * config/m68k/m68k.md (movdf_cf_hard): Use output_move_double for
672 GPR<-GPR moves.
673
674 2007-01-18 Richard Sandiford <richard@codesourcery.com>
675
676 * real.h (real_format): Add a canonical_nan_lsbs_set field.
677 (coldfire_single_format): Declare.
678 (coldfire_double_format): Likewise.
679 * real.c (encode_ieee_single): Use canonical_nan_lsbs_set instead
680 of qnan_msb_set to determine the lower bits of a canonical
681 NaN significand.
682 (encode_ieee_double): Likewise.
683 (encode_ieee_quad): Likewise.
684 (ieee_single_format): Initialize canonical_nan_lsbs_set.
685 (mips_single_format): Likewise.
686 (ieee_double_format): Likewise.
687 (mips_double_format): Likewise.
688 (ieee_extended_motorola_format): Likewise.
689 (ieee_extended_intel_96_format): Likewise.
690 (ieee_extended_intel_128_format): Likewise.
691 (ieee_extended_intel_96_round_53_format): Likewise.
692 (ibm_extended_format): Likewise.
693 (mips_extended_format): Likewise.
694 (ieee_quad_format): Likewise.
695 (mips_quad_format): Likewise.
696 (vax_f_format): Likewise.
697 (vax_d_format): Likewise.
698 (vax_g_format): Likewise.
699 (i370_single_format): Likewise.
700 (i370_double_format): Likewise.
701 (decimal_single_format): Likewise.
702 (decimal_double_format): Likewise.
703 (decimal_quad_format): Likewise.
704 (c4x_single_format): Likewise.
705 (c4x_extended_format): Likewise.
706 (real_internal_format): Likewise.
707 (coldfire_single_format): New real_format.
708 (coldfire_double_format): Likewise.
709 * config/pdp11/pdp11.c (pdp11_f_format): Initialize
710 canonical_nan_lsbs_set.
711 (pdp11_d_format): Likewise.
712 * config/m68k/m68k.c (override_options): Override REAL_FORMAT_MODE
713 if TARGET_COLDFIRE_CPU.
714
715 2007-01-18 Richard Sandiford <richard@codesourcery.com>
716
717 * config/m68k/m68k-protos.h (m68k_output_pic_call): Delete.
718 (output_call, m68k_legitimize_call_address): Declare.
719 * config/m68k/m68k.h (EXTRA_CONSTRAINT): Remove unnecessary
720 parenthesees. Add support for a 'W' constraint.
721 (LEGITIMATE_PIC_OPERAND_P): Remove SYMBOL_REF_FLAG handling.
722 (PRINT_OPERAND_PUNCT_VALID_P): Remove comment about 'o'.
723 (m68k_symbolic_call, m68k_symbolic_jump): Declare.
724 * config/m68k/m68k.c (m68k_symbolic_call, m68k_symbolic_jump): New
725 variables.
726 (override_options): Initialize them. Do not set flag_no_function_cse
727 for TARGET_ID_SHARED_LIBRARY.
728 (m68k_output_pic_call): Delete.
729 (m68k_legitimize_call_address): New function.
730 (print_operand): Remove the %o prefix. Handle the %p prefix.
731 (output_call): New function.
732 (m68k_output_mi_thunk): Use m68k_symbolic_jump. Always load the
733 target address from the GOT if symbolic jumps are not allowed.
734 * config/m68k/m68k.md (call, general_operand): Do not set
735 SYMBOL_REF_FLAG. Use m68k_legitimize_call_address instead.
736 Merge separate flag_pic and !flag_pic define_insns into...
737 (*call, *call_value): ...these new patterns. Match the address
738 rather than the containing MEM and require it to be a call_operand.
739 Use output_call to generate the asm template.
740 * config/m68k/predicates.md (const_call_operand): New predicate.
741 (call_operand): Likewise.
742
743 2007-01-18 Nathan Sidwell <nathan@codesourcery.com>
744
745 * config/m68k/m68k.h (REGISTER_MOVE_COST): Simplify definition.
746 (STACK_GROWS_DOWNWARD): Define to 1.
747 (FUNCTION_VALUE, LIBCALL_VALUE, FUNCTION_VALUE_REGNO_P): Equivocate
748 comments, emphasizing that these values are only defaults.
749 * config/m68k/linux.h (LINK_SPEC): Fix formatting in #undef.
750 * config/m68k/m68k.c (const_method): Remove trailing whitespace.
751
752 2007-01-18 Richard Sandiford <richard@codesourcery.com>
753
754 * config/m68k/m68k.md (cmpsi): Remove outdated flag_pic handling.
755
756 2007-01-18 Kazu Hirata <kazu@codesourcery.com>
757 Richard Sandiford <richard@codesourcery.com>
758
759 * config/m68k/m68k.h (DATA_REGNO_P, ADDRESS_REGNO_P, INT_REGNO_P)
760 (FP_REGNO_P): New macros.
761 (REGNO_OK_FOR_INDEX_P, REGNO_OK_FOR_BASE_P, REGNO_OK_FOR_DATA_P)
762 (REGNO_OK_FOR_FP_P, REG_OK_FOR_INDEX_P, REG_OK_FOR_BASE_P): Use them.
763 (EH_RETURN_STACKADJ_RTX): Use A0_REG.
764 * config/m68k/m68k.c (m68k_regno_mode_ok): Use the new REGNO macros.
765
766 2007-01-18 Nathan Sidwell <nathan@codesourcery.com>
767
768 * config.gcc (m68k-*-aout*, m68k-*-coff*, m68020-*-elf*, m68k-*-elf*)
769 (m68k-*-uclinuxoldabi, m68k-*-uclinux*, m68k-*-rtems*): Add t-floatlib
770 to $tmake_file.
771 * config/m68k/t-floatlib: New file, extracting common code from...
772 * config/m68k/t-m68kbare, config/m68k/t-m68kelf,
773 * config/m68k/t-uclinux: Here.
774 * config/m68k/fpgnulib.c: Do not compile extendeed precision
775 routines on ColdFire targets.
776
777 2007-01-18 Nathan Sidwell <nathan@codesourcery.com>
778
779 * config.gcc (m68k-*-aout*, m68k-*-coff*, m68020-*-elf*, m68k-*-elf*)
780 (m68010-*-netbsdelf*, m68k*-*-netbsdelf*, m68k*-*-openbsd*)
781 (m68k-*-uclinuxoldabi, m68k-*-uclinux*, m68k-*-linux*)
782 (m68k-*-rtems*): Use tm_file rather than m68k/m68k.h and
783 explicitly set MOTOROLA to 1.
784 * config/m68k/m68k.h (MOTOROLA): Simplify definition accordingly.
785
786 2007-01-18 Richard Sandiford <richard@codesourcery.com>
787 Nathan Sidwell <nathan@codesourcery.com>
788
789 * config/m68k/m68k.h (PCC_STATIC_STRUCT_RETURN): Delete.
790 (ASM_OUTPUT_REG_PUSH, ASM_OUTPUT_REG_POP): Add MOTOROLA cases.
791 * config/m68k/coff.h (REGISTER_PREFIX_MD): Delete.
792 * config/m68k/m68020-elf.h (LIB_SPEC): Delete.
793 * config/m68k/m68k-none.h (CC1_SPEC, CPP_SUBTARGET_SPEC): Delete.
794 * config/m68k/m68kelf.h (IMMEDIATE_PREFIX, REGISTER_PREFIX_MD)
795 (ASM_OUTPUT_REG_PUSH, ASM_OUTPUT_REG_POP): Delete.
796 (NO_DOLLAR_IN_LABEL, PCC_STATIC_STRUCT_RETURN): Don't undefine.
797 (BSS_ASM_OP, ASM_OUTPUT_SKIP, ASM_OUTPUT_ASCII): Delete.
798 * config/m68k/m68kemb.h (PCC_STATIC_STRUCT_RETURN): Don't undefine.
799 * config/m68k/linux.h (SIZE_TYPE, PTRDIFF_TYPE, WCHAR_TYPE)
800 (WCHAR_TYPE_SIZE, TARGET_OBJFMT_CPP_BUILTINS): Delete.
801 (TARGET_OS_CPP_BUILTINS): Don't define mc68000 and mc68020 here.
802 (DBX_REGISTER_NUMBER): Delete.
803 * config/m68k/netbsd-elf.h (IMMEDIATE_PREFIX): Delete.
804 (PCC_STATIC_STRUCT_RETURN): Don't undefine.
805 * config/m68k/openbsd.h (PCC_STATIC_STRUCT_RETURN): Define.
806
807 2007-01-18 Richard Sandiford <richard@codesourcery.com>
808
809 * config.gcc (m68k-*-uclinux*): Add flat.h to $tm_file.
810 * config/flat.h: New file.
811 * crtstuff.c (USE_PT_GNU_EH_FRAME): Don't define if
812 OBJECT_FORMAT_FLAT.
813 * config/m68k/m68k.h (ASM_PREFERRED_EH_DATA_FORMAT): Do not use
814 indirect references for -msep-data or -mid-shared-library.
815 Do not use PC-relative code addresses either.
816 * config/m68k/m68k.c (override_options): Restrict -fPIC error
817 to -mpcrel.
818 * config/m68k/uclinux.h (STARTFILE_SPEC): Define. Use Scrt1.o
819 for shared libraries and crt1.o for executables. Use crti.o and
820 crtbegin.o.
821 (ENDFILE_SPEC): Use crtend.o and crtn.o.
822 (LIB_SPEC): Suppress -Rlibc.gdb if -static-libc is given.
823 Do not add -elf2flt or -shared-lib-id options here.
824 (LINK_SPEC): Define. Pass -elf2flt if no -elf2flt option is given.
825 Pass -shared-lib-id if -mid-shared-library, taking the library
826 identifier from -mshared-library-id if given, otherwise
827 defaulting to 0.
828 (EH_FRAME_IN_DATA_SECTION): Do not undefine.
829 (INIT_SECTION_ASM_OP, FINI_SECTION_ASM_OP): Likewise.
830 (TARGET_OS_CPP_BUILTINS): Define __GXX_MERGED_TYPEINFO_NAMES=0
831 and __GXX_TYPEINFO_EQUALITY_INLINE=0 if -mid-shared-library.
832 (DRIVER_SELF_SPECS): Map unadorned PIC options to -msep-data.
833 * config/m68k/t-uclinux (EXTRA_MULTILIB_PARTS): Add crtbegin.o
834 and crtend.o.
835 * config/m68k/lb1sf68.asm (PICCALL): Use an lea and pc-relative
836 jump sequence for ISA A and ISA A+.
837 (PICJUMP): Likewise.
838
839 2007-01-18 Richard Sandiford <richard@codesourcery.com>
840
841 * config.gcc (m68k-*-uclinux*): Base the port on the common
842 and m68k GNU/Linux files rather than on the generic ELF ones.
843 * config/m68k/uclinux.h (TARGET_VERSION): Override.
844 (TARGET_OS_CPP_BUILTINS): Use LINUX_TARGET_OS_CPP_BUILTINS.
845
846 2007-01-18 Julian Brown <julian@codesourcery.com>
847 Richard Sandiford <richard@codesourcery.com>
848
849 * config/m68k/m68k.h (LONG_DOUBLE_TYPE_SIZE): Make 64-bit on ColdFire.
850 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
851 * config/m68k/netbsd-elf.h (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Undefine
852 before redefining.
853 * config/m68k/uclinux-oldabi.h (LONG_DOUBLE_TYPE_SIZE): Redefine to
854 80 unconditionally.
855 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
856
857 2007-01-18 Richard Sandiford <richard@codesourcery.com>
858
859 * doc/install.texi: Document m68k-uclinuxoldabi.
860 * config.gcc (m68k-*-uclinuxoldabi): New configuration.
861 * config/m68k/uclinux-oldabi.h: New file, copied from
862 config/m68k/uclinux.h.
863
864 2007-01-18 Kaz Kojima <kkojima@rr.iij4u.or.jp>
865
866 * config/m32r/m32r-protos.h (m32r_expand_epilogue): Declare it.
867 * config/m32r/m32r.c (m32r_setup_incoming_varargs): Use gen_frame_mem.
868 (m32r_compute_frame_size): Use unsigned for regno.
869 (m32r_reload_lr): Use gen_frame_mem.
870 (pop): New.
871 (m32r_output_function_epilogue): Don't output the function epilogue
872 textually here.
873 (m32r_expand_epilogue): New.
874 (direct_return): Return false if the function has the interrupt
875 attribute.
876 (m32r_hard_regno_rename_ok): Remove code for the textual epilogue.
877 * config/m32r/m32r.md (epilogue): New expander.
878 (return_lr, return_rte): New insns.
879 (return): Make it expander.
880 (return_normal): New expander.
881
882 2007-01-18 Josh Conner <jconner@apple.com>
883
884 PR target/30485
885 * config/rs6000/rs6000.c (rs6000_emit_vector_compare): Add
886 support for UNLE, UNLT, UNGE, and UNGT.
887
888 2007-01-18 Jan Hubicka <jh@suse.cz>
889
890 * tree-vrp.c (finalize_jump_threads): Do not call cleanup_cfg by hand.
891 * tree-tailcall (add_virtual_phis): Likewise.
892 (optimize_tail_call): Return TODOs.
893 (execute_tail_calls): Return TODOs.
894 * tree-ssa-ccp (execute_fold_all_builtins): Do cleanup_cfg via TODO.
895 * tree-cfgcleanup.c (cleanup_tree_cfg_loop): Return if something
896 changed.
897 * tree-ssa-forwprop.c (tree_ssa_forward_propagate_single_use_value):
898 Cleanup cfg using TODO.
899 * tree-flow.h (cleanup_tree_cfg_loop): Update prototype.
900 * passes.c (execute_function_todo): When cleanup did something, remove
901 unused locals.
902 * tree-cfg.c (pass_build_cfg): Add cleanup_cfg TODO.
903 (make_edges): Don't cleanup_cfg.
904
905 2007-01-18 Uros Bizjak <ubizjak@gmail.com>
906
907 * reg-stack.c (subst_stack_regs_pat) [UNSPEC_SINCOS_COS,
908 UNSPEC_XTRACT_FRACT]: Use generic code for instructions that
909 operate on the top of stack.
910 [UNSPEC_SINCOS_SIN, UNSPEC_XTRACT_EXP, UNSPEC_TAN]: Rewrite
911 register handling of instructions that output to the second
912 stack slot.
913 [UNSPEC_TAN_ONE, UNSPEC_TAN_TAN]: Remove.
914 (move_for_stack_reg): Special-case check for dead destination
915 stack slot for constant load of 1.0 inside UNSPEC_TAN.
916
917 * config/i386/i386.md (UNSPEC_TAN): New constant.
918 (UNSPEC_TAN_ONE, UNSPEC_TAN_TAN): Remove.
919 (fptanxf4_i387, fptan_extend<mode>xf4_i387): New patterns
920 to correctly model move of constant 1.0 to top stack slot.
921 (*tandf3_1, *tansf3_1, *tanxf3_1): Remove insn patterns.
922 (unnamed peephole2 pattern): Remove corresponding peephole2
923 pattern that optimizes tan insn and loading of constant 1.0.
924 (tanxf2): Use fptanxf4_i387.
925 (tan<mode>2): Rename from tansf2 and tandf2 and macroize insn
926 patterns using X87MODEF12 mode macro. Use fptan_extend<mode>xf4_i387
927 and truncate result to requested mode. Use SSE_FLOAT_MODE_P to
928 disable patterns for SSE math.
929 (sincos<mode>3): Use truncxf<mode>2_i387_noop for truncation.
930 (fyl2x_extend<mode>xf3_i387): Use X87MODEF12 for operand 1.
931
932 2007-01-18 Dirk Mueller <dmueller@suse.de>
933 Richard Guenther <rguenther@suse.de>
934
935 PR diagnostic/8268
936 * doc/invoke.texi (Warray-bounds): Document -Warray-bounds.
937 * common.opt (Warray-bounds): Add new warning option.
938 * c-opts.c (c_common_handle_option): Define -Warray-bounds
939 if -Wall is given.
940 * Makefile.in: make tree-vrp.o depend on toplev.h
941 * tree-vrp.c (vrp_finalize): Call check_array_refs if -Warray-bounds
942 is enabled.
943 (check_array_refs, check_array_bounds, check_array_ref): New.
944
945 2007-01-18 Jan Hubicka <jh@suse.cz>
946
947 * tree-ssa-ccp.c (ccp_finalize): Return if something changed.
948 (execute_ssa_ccp): Return flags conditionally.
949 * tree-ssa-propagate.c (substitue_and_fold): Return if something was
950 changed.
951 * tree-ssa-propagate.h (substitute_and_fold): Update prototype.
952
953 2007-01-18 Steven Bosscher <steven@gcc.gnu.org>
954
955 * cfgcleanup.c (cleanup_cfg): Detect cfglayout mode and set
956 the CLEANUP_CFGLAYOUT flag when in cfglayout mode.
957
958 * Makefile.c (GTFILES): Add cfglayout.h.
959 * gengtype.c (open_base_files): Likewise.
960 * cfglayout.c (cfg_layout_function_footer,
961 cfg_layout_function_header) Reindent to make gengtype happy.
962 * cfglayout.h (cfg_layout_function_footer,
963 cfg_layout_function_header): Add GTY(()) marker.
964
965 * ifcvt.c (noce_try_sign_mask): Make sure INSN_B is non-null.
966
967 2007-01-18 Ben Elliston <bje@au.ibm.com>
968
969 * genautomata.c (write_automata): Include xstrerror output in the
970 error message if writing the DFA description file fails.
971
972 2007-01-17 H.J. Lu <hongjiu.lu@intel.com>
973
974 * config/mips/mips-protos.h (mips_output_external): Make it
975 return void.
976 * config/mips/iris.h (TARGET_ASM_EXTERNAL_LIBCALL): Removed.
977 * config/mips/mips.c (irix_output_external_libcall): Likewise.
978 (extern_list): Likewise.
979 (extern_head): Likewise.
980 (TARGET_ASM_FILE_END): Likewise.
981 (mips_file_end): Likewise.
982 (mips_output_external): Rewritten.
983
984 2007-01-18 Ben Elliston <bje@au.ibm.com>
985
986 * genpreds.c (write_insn_preds_c): Only write out the function
987 body for regclass_for_constraint if we have register constraints.
988
989 2007-01-17 Tom Tromey <tromey@redhat.com>
990
991 * doc/sourcebuild.texi (libgcj Tests): Use sourceware.org.
992 * doc/install.texi (Testing): Use sourceware.org.
993 (Binaries): Likewise.
994 (Specific): Likewise.
995 * doc/contrib.texi (Contributors): Use sourceware.org.
996
997 2007-01-17 Anatoly Sokolov <aesok@post.ru>
998
999 * config/avr/avr.h (AVR_HAVE_LPMX): New macro.
1000 (AVR_ENHANCED): Rename to ...
1001 (AVR_HAVE_MUL): ... new.
1002 (avr_enhanced_p): Rename to ...
1003 (avr_have_mul_p): ... new.
1004 (TARGET_CPU_CPP_BUILTINS): Use 'avr_have_mul_p' instead of
1005 'avr_enhanced_p' for "__AVR_ENHANCED__". Define "__AVR_HAVE_MUL__".
1006 * config/avr/avr.c (avr_enhanced_p): Rename to ...
1007 (avr_have_mul_p): ... new.
1008 (base_arch_s): Rename 'enhanced' to 'have_mul'.
1009 (avr_override_options): Use 'avr_have_mul_p' and 'have_mul' instead of
1010 'avr_enhanced_p' and 'enhanced'.
1011 (ashlhi3_out, ashrhi3_out, lshrhi3_out, avr_rtx_costs): Use
1012 AVR_HAVE_MUL instead of AVR_ENHANCED.
1013 * avr.md (*tablejump_enh): Use AVR_HAVE_LPMX instead of AVR_ENHANCED.
1014 (mulqi3, *mulqi3_enh, *mulqi3_call, mulqihi3, umulqihi3, mulhi3,
1015 *mulhi3_enh, *mulhi3_call, mulsi3, *mulsi3_call): Use AVR_HAVE_MUL
1016 instead of AVR_ENHANCED.
1017 (*tablejump_enh): Use AVR_HAVE_LPMX instead of AVR_ENHANCED.
1018 * libgcc.S: Use __AVR_HAVE_MUL__ instead of __AVR_ENHANCED__.
1019 (__tablejump__): Use __AVR_HAVE_LPMX__ instead of __AVR_ENHANCED__.
1020
1021 2007-01-17 Ian Lance Taylor <iant@google.com>
1022
1023 * vec.h (VEC_reserve_exact): Define.
1024 (vec_gc_p_reserve_exact): Declare.
1025 (vec_gc_o_reserve_exact): Declare.
1026 (vec_heap_p_reserve_exact): Declare.
1027 (vec_heap_o_reserve_exact): Declare.
1028 (VEC_OP (T,A,reserve_exact)): New static inline function, three
1029 versions.
1030 (VEC_OP (T,A,reserve)) [all versions]: Remove handling of
1031 negative parameter.
1032 (VEC_OP (T,A,alloc)) [all versions]: Call ...reserve_exact.
1033 (VEC_OP (T,A,copy)) [all versions]: Likewise.
1034 (VEC_OP (T,a,safe_grow)) [all versions]: Likewise.
1035 * vec.c (calculate_allocation): Add exact parameter. Change all
1036 callers.
1037 (vec_gc_o_reserve_1): New static function, from vec_gc_o_reserve.
1038 (vec_gc_p_reserve, vec_gc_o_reserve): Call vec_gc_o_reserve_1.
1039 (vec_gc_p_reserve_exact, vec_gc_o_reserve_exact): New functions.
1040 (vec_heap_o_reserve_1): New static function, from vec_heap_o_reserve.
1041 (vec_heap_p_reserve, vec_heap_o_reserve): Call vec_heap_o_reserve_1.
1042 (vec_heap_p_reserve_exact): New function.
1043 (vec_heap_o_reserve_exact): New function.
1044
1045 2007-01-17 Jan Hubicka <jh@suse.cz>
1046
1047 * ipa-type-escape.c (look_for_casts): Revamp using handled_component_p.
1048
1049 2007-01-17 Eric Christopher <echristo@apple.com>
1050
1051 * config.gcc: Support core2 processor.
1052
1053 2007-01-16 Jan Hubicka <jh@suse.cz>
1054
1055 * tree-ssanames.c (release_dead_ssa_names): Instead of ggc_freeing
1056 the names, just unlink the chain so we don't crash on dangling pointers
1057 to dead SSA names.
1058
1059 2007-01-16 Jan Hubicka <jh@suse.cz>
1060
1061 * cgraph.h (cgraph_decide_inlining_incrementally): Kill.
1062 * tree-pass.h: Reorder to make IPA passes appear toegher.
1063 (pass_early_inline, pass_inline_parameters, pass_apply_inline): Declare.
1064 * cgraphunit.c (cgraph_finalize_function): Do not compute inling
1065 parameters, do not call early inliner.
1066 * ipa-inline.c: Update comments. Include tree-flow.h
1067 (cgraph_decide_inlining): Do not compute inlining parameters.
1068 (cgraph_decide_inlining_incrementally): Return TODOs; assume to
1069 be called with function context set up.
1070 (pass_ipa_inline): Remove unreachable functions before pass.
1071 (cgraph_early_inlining): Simplify assuming to be called from the
1072 PM as local pass.
1073 (pass_early_inline): New pass.
1074 (cgraph_gate_ipa_early_inlining): New gate.
1075 (pass_ipa_early_inline): Turn into simple wrapper.
1076 (compute_inline_parameters): New function.
1077 (gate_inline_passes): New gate.
1078 (pass_inline_parameters): New pass.
1079 (apply_inline): Move here from tree-optimize.c
1080 (pass_apply_inline): New pass.
1081 * ipa.c (cgraph_remove_unreachable_nodes): Verify cgraph after
1082 transforming.
1083 * tree-inline.c (optimize_inline_calls): Return TODOs rather than
1084 doing them by hand.
1085 (tree_function_versioning): Do not allocate dummy struct function.
1086 * tree-inline.h (optimize_inline_calls): Update prototype.
1087 * tree-optimize.c (execute_fixup_cfg): Export.
1088 (pass_fixup_cfg): Remove
1089 (tree_rest_of_compilation): Do not apply inlines.
1090 * tree-flow.h (execute_fixup_cfg): Declare.
1091 * Makefile.in (gt-passes.c): New.
1092 * passes.c: Include gt-passes.h
1093 (init_optimization_passes): New passes.
1094 (nnodes, order): New static vars.
1095 (do_per_function_toporder): New function.
1096 (execute_one_pass): Dump current pass here.
1097 (execute_ipa_pass_list): Don't dump current pass here.
1098
1099 2007-01-16 Janis Johnson <janis187@us.ibm.com>
1100
1101 * config/dfp-bit.c (dfp_compare_op): Return separate value for NaN.
1102 (DFP_NE, DFP_LE, DFP_GE): Return false for NaN.
1103
1104 2007-01-16 David Edelsohn <edelsohn@gnu.org>
1105
1106 * config/rs6000/darwin-ldouble.c: Build file for SOFT_FLOAT.
1107 (strong_alias): Define.
1108 (__gcc_qmul): Provide non-FMA for soft-float.
1109 (__gcc_qdiv): Same.
1110 (__gcc_qneg): New.
1111 (__gcc_qeq): New.
1112 (__gcc_qle): New.
1113 (__gcc_qge): New.
1114 (__gcc_qunord): New.
1115 (__gcc_stoq): New.
1116 (__gcc_dtoq): New.
1117 (__gcc_qtos): New.
1118 (__gcc_qtod): New.
1119 (__gcc_qtoi): New.
1120 (__gcc_qtou): New.
1121 (__gcc_itoq): New.
1122 (__gcc_utoq): New.
1123 (fmsub): New.
1124 * config/rs6000/rs6000.c (rs6000_init_libfuncs): Initialize
1125 soft-float functions.
1126 * config/rs6000/libgcc-ppc-glibc.ver: Version soft-float symbols.
1127 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Do not warn
1128 about long double soft float.
1129
1130 2007-01-16 Dorit Nuzman <dorit@il.ibm.com>
1131 Tehila Meyzels <tehila@il.ibm.com>
1132
1133 * tree-vectorizer.h (is_pattern_stmt_p): New.
1134 * tree-vect-analyze.c (vect_determine_vectorization_factor): Fix
1135 formatting (tabs instead of spaces). Cleanup and clarify setting
1136 of STMT_VINFO_VECTYPE. Call is_pattern_stmt_p.
1137 * tree-vect-transform.c (vect_get_vec_def_for_operand): Fix typo.
1138 (vectorizable_type_demotion): Check that types are integral.
1139 (vectorizable_type_promotion): Likewise.
1140 (vectorizable_store): Fix typo. Eliminate new-line at end of
1141 comments.
1142
1143 2007-01-16 Jan Hubicka <jh@suse.cz>
1144
1145 * tree-ssanames.c (release_dead_ssa_names): Remove invalidated
1146 cgraph edges too.
1147
1148 2007-01-15 Eric Christopher <echristo@apple.com>
1149
1150 * ifcvt.c: Include vec.h, vecprim.h.
1151 (check_cond_move_block): New argument regs.
1152 Reorganize. Add registers used to regs.
1153 (cond_move_process_if_block): Use regs set above as
1154 loop bounds.
1155
1156 2007-01-15 Eric Christopher <echristo@apple.com>
1157
1158 * config/darwin.h: Update copyright.
1159 (TARGET_OPTION_TRANSLATE_TABLE): Add umbrella.
1160 (LINK_COMMAND_SPEC): Add -u.
1161 (LINK_SPEC): Fix umbrella for above.
1162
1163 2007-01-15 Joseph S. Myers <joseph@codesourcery.com>
1164
1165 * config/soft-fp/op-common.h, config/soft-fp/op-4.h: Update from
1166 glibc CVS.
1167
1168 2007-01-15 Tom Tromey <tromey@redhat.com>
1169
1170 * doc/sourcebuild.texi (libgcj Tests): Don't mention jacks.
1171 * doc/install.texi (Testing): Don't mention jacks.
1172 (Configuration): Document --enable-java-maintainer-mode. Move
1173 --with-java-home to libgcj-specific section. Document
1174 --with-ecj-jar.
1175 (Prerequisites): Mention --enable-java-maintainer-mode, ecj1.
1176
1177 2007-01-15 Jan Hubicka <jh@suse.cz>
1178
1179 * tree-ssa-dce.c (DCE_TODOs): New.
1180 (propagate_necessity): Return if something changed.
1181 (eliminate_unnecessary_stmts): Likewise.
1182 (perform_tree_ssa_dce): Return TODO flags when needed.
1183 (pass_dce, pass_dce_loop, pass_cd_dce): Remove TODO flags.
1184
1185 2007-01-15 Uros Bizjak <ubizjak@gmail.com>
1186
1187 * config/i386/i386.md (fyl2xxf3_i387): Rename from fyl2x_xf3.
1188 (fyl2x_extend<mode>xf3_i387): New insn pattern.
1189 (log<mode>2): Rename from logsf2 and logdf2 and macroize insn
1190 patterns using X87MODEF12 mode macro. Extend operand 1
1191 to XFmode. Use SSE_FLOAT_MODE_P to disable patterns for SSE math.
1192 (log10<mode>2): Ditto.
1193 (log2<mode>2): Ditto.
1194 (log1p<mode>2): Ditto.
1195 (logb<mode>2): Ditto.
1196 (fyl2xp1xf3_i387): Rename from fyl2xp1_xf3.
1197 (fyl2xp1_extend<mode>xf3_i387): New insn pattern.
1198 (*fxtractxf3_i387): Rename from *fxtractxf3.
1199 (fxtract_extend<mode>xf3_i387): New insn pattern.
1200 (ilogbsi2): Use match_dup 3, not match_operand:XF 3.
1201 * config/i386/i386.c (ix86_emit_i387_log1p): Use gen_fyl2xp1xf3_i387()
1202 and gen_fyl2xxf3_i387().
1203
1204 2007-01-14 Zdenek Dvorak <dvorakz@suse.cz>
1205
1206 * loop-unswitch.c (unswitch_loop): Do not call fix_loop_placement.
1207 * cfgloopmanip.c (fix_loop_placement): Made static. Use
1208 get_loop_exit_edges. Changed return type to bool.
1209 * cfgloop.h (fix_loop_placement): Declaration removed.
1210
1211 2007-01-14 Dorit Nuzman <dorit@il.ibm.com>
1212
1213 * param.h (MIN_VECT_LOOP_BOUND): New.
1214 * params.def (MIN_VECT_LOOP_BOUND): New.
1215 * tree-vectorizer.c (slpeel_tree_peel_loop_to_edge): Takes another
1216 argument - minimum threshold for number of iterations.
1217 * tree-vectorizer.h (slpeel_tree_peel_loop_to_edge): Add another
1218 argument to declaration.
1219 * tree-vect-analyze.c (vect_analyze_operations): Check value of
1220 MIN_VECT_LOOP_BOUND.
1221 * tree-vect-transform.c (vect_do_peeling_for_loop_bound): Call
1222 slpeel_tree_peel_loop_to_edge with additional argument.
1223 (vect_do_peeling_for_alignment): Likewise.
1224 * doc/invoke.texi (min-vect-loop-bound): Document new param option.
1225
1226 2007-01-14 Uros Bizjak <ubizjak@gmail.com>
1227
1228 PR target/30413
1229 * config/i386/i386.c (print_operand) ['z']: Output 'b' for
1230 operands of size 1.
1231
1232 2007-01-14 Jan Hubicka <jh@suse.cz>
1233
1234 * tree-dfa.c (remove_referenced_var): New function.
1235 * tree-ssa-live.c (remove_unused_locals): Walk referenced vars and
1236 prune referenced vars list too.
1237 * tree-flow.h (remove_referenced_var): Declare.
1238
1239 2007-01-14 Jan Hubicka <jh@suse.cz>
1240
1241 * tree-eh.c (add_stmt_to_eh_region_fn): Do not add call_exprs
1242 separately.
1243 (remove_stmt_from_eh_region_fn): Do not remove call_exprs.
1244 (verify_eh_throw_stmt_node, verify_eh_throw_table_statements): Kill.
1245 * except.h (verify_eh_throw_table_statements): Kill prototype.
1246 * cfgexpand.c (expand_gimple_basic_block): Propagate Eh regions
1247 into call exrepssions.
1248 * tree-optimize.c (execute_free_cfg_annotatiosn): Do not call
1249 eh trhow verifier.
1250 * tree-cfg.c: Include pointer-set.h.
1251 (verify_node_sharing): Work on pointer set.
1252 (verify_eh_throw_stmt_node): New.
1253 (verify_stmts): Use pointers sets, verify throw_stmt.
1254
1255 2007-01-13 Zdenek Dvorak <dvorakz@suse.cz>
1256
1257 * ipa-reference.c (analyze_function): Consider also addresses taken
1258 in phi nodes.
1259
1260 2007-01-12 Roger Sayle <roger@eyesopen.com>
1261
1262 * c-typeck.c (null_pointer_constant_p): Replace use of
1263 TREE_CONSTANT_OVERFLOW with TREE_OVERFLOW.
1264 (build_c_cast): Likewise.
1265
1266 2007-01-12 Roger Sayle <roger@eyesopen.com>
1267
1268 * tree.h (force_fit_type_double): Remove unused final argument.
1269 * c-common.c (constant_expression_warning): Replace use of
1270 TREE_CONSTANT_OVERFLOW with TREE_OVERFLOW.
1271 (convert_and_check): Likewise.
1272 (shorten_compare): Update call to force_fit_type_double.
1273 (c_common_truthvalue_conversion) <INTEGER_CST>: Use integer_zerop.
1274 * convert.c (convert_to_pointer): Update call to
1275 force_fit_type_double.
1276 * fold-const.c (force_fit_type_double): Remove overflowed_const
1277 argument.
1278 (int_const_binop, fold_convert_const_int_from_int,
1279 fold_convert_const_int_from_real, fold_div_compare,
1280 fold_sign_changed_comparison, fold_unary, fold_negate_const,
1281 fold_abs_const, fold_not_const): Remove the final argument from
1282 calls to force_fit_type_double.
1283
1284 2007-01-12 Andrew Pinski <andrew_pinski@playstation.sony.com>
1285
1286 * configure.ac: Set insn to "nop" for spu-*-* also.
1287 * configure: Regenerate.
1288
1289 2007-01-12 Olga Golovanevsky <olga@il.ibm.com>
1290
1291 * builtins.def : Add BUILT_IN_FREE.
1292
1293 2007-01-12 Jan Hubicka <jh@suse.cz>
1294
1295 PR tree-optimization/30443
1296 * tree-inline.c (tree_function_versioning): Do not optimize when
1297 cloning for inlining.
1298
1299 2007-01-12 Zdenek Dvorak <dvorakz@suse.cz>
1300
1301 * doc/loop.texi: Document recording of loop exits.
1302 * cfgloopmanip.c (loopify, duplicate_loop): Use alloc_loop.
1303 (update_single_exits_after_duplication,
1304 update_single_exit_for_duplicated_loop,
1305 update_single_exit_for_duplicated_loops): Removed.
1306 (duplicate_loop_to_header_edge): Do not call
1307 update_single_exits_after_duplication and
1308 update_single_exit_for_duplicated_loops.
1309 (loop_version): Do not update single_exit information.
1310 (fix_loop_structure): Use record_loop_exits instead of
1311 mark_single_exit_loops.
1312 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update
1313 the lists of loop exits.
1314 * cfghooks.c (redirect_edge_and_branch, redirect_edge_and_branch_force,
1315 split_edge, merge_blocks): Update the lists of loop exits.
1316 * modulo-sched.c (sms_schedule): Pass LOOPS_HAVE_RECORDED_EXITS to
1317 loop_optimizer_init.
1318 * loop-init.c (loop_optimizer_init): Call record_loop_exits instead
1319 of mark_single_exit_loops.
1320 (loop_optimizer_finalize): Call release_recorded_exits.
1321 * tree-ssa-loop.c (tree_loop_optimizer_init): Pass
1322 LOOPS_HAVE_RECORDED_EXITS to loop_optimizer_init.
1323 * tree-vectorizer.c (slpeel_tree_duplicate_loop_to_edge_cfg): Do not
1324 update single exit information.
1325 * lambda-code.c (perfect_nestify): Ditto.
1326 * cfgloop.c (flow_loop_free): Destroy the list of exits of the loop.
1327 (mark_single_exit_loops): Removed.
1328 (alloc_loop, loop_exit_hash, loop_exit_eq, loop_exit_free,
1329 get_exit_descriptions, rescan_loop_exit, record_loop_exits,
1330 dump_recorded_exit, dump_recorded_exits, release_recorded_exits): New
1331 functions.
1332 (get_loop_exit_edges, single_exit): Use recorded exit lists.
1333 (add_bb_to_loop, remove_bb_from_loops): Update the lists of loop exits.
1334 (verify_loop_structure): Verify consistency of the exit lists.
1335 (flow_loops_find): Use alloc_loop. Initialize exits hash.
1336 (set_single_exit): Removed.
1337 * cfgloop.h (struct loop_exit): New function.
1338 (struct loop): single_exit_ field replaced by exits field.
1339 (LOOPS_HAVE_MARKED_SINGLE_EXITS): Replaced by LOOPS_HAVE_RECORDED_EXITS.
1340 (struct loops): Added exits hash.
1341 (mark_single_exit_loops, set_single_exit): Declaration removed.
1342 (release_recorded_exits, record_loop_exits, rescan_loop_exit): Declare.
1343
1344 2007-01-12 Richard Sandiford <richard@codesourcery.com>
1345
1346 * doc/invoke.texi: Avoid use of @headitem.
1347
1348 2007-01-12 Richard Sandiford <richard@codesourcery.com>
1349
1350 * cse.c (cse_insn): Move HAVE_CC0 code after declarations.
1351
1352 2007-01-12 Richard Sandiford <richard@codesourcery.com>
1353
1354 * doc/install.texi: Fix m68k-*-* anchor and add m68k-*-* to the
1355 list of targets.
1356
1357 2007-01-12 Nathan Sidwell <nathan@codesourcery.com>
1358 Richard Sandiford <richard@codesourcery.com>
1359
1360 * doc/invoke.texi: Document -mno-bitfield, -mno-rtd and -mno-short.
1361 * config/m68k/m68k.opt: Resort options.
1362 (mbitfield, mrtd, mshort): Remove RejectNegative properties.
1363
1364 2007-01-12 Nathan Sidwell <nathan@codesourcery.com>
1365 Richard Sandiford <richard@codesourcery.com>
1366
1367 * doc/invoke.texi: Document the macros that are defined by
1368 m68k's -mtune and -mhard-float options.
1369 * config/m68k/m68k-protos.h (m68k_cpp_cpu_ident) Declare.
1370 (m68k_cpp_cpu_family): Likewise.
1371 * config/m68k/m68k.h (TARGET_CPU_CPP_BUILTINS): Add a full set
1372 of __ucfv*__ macros. Define __mcffpu__ if generating code for
1373 ColdFire FPUs. Define __mcf_cpu_* and __mcf_family_* macros.
1374 * config/m68k/m68k.c (m68k_cpp_cpu_ident): New function.
1375 (m68k_cpp_cpu_family): Likewise.
1376
1377 2007-01-12 Richard Sandiford <richard@codesourcery.com>
1378
1379 * config/m68k/m68k.h (TARGET_CPU_CPP_BUILTINS): Treat all mc68*
1380 macros besides mc68000 as tuning macros. Use a switch statement
1381 to set them and mcpu32.
1382
1383 2007-01-12 Julian Brown <julian@codesourcery.com>
1384
1385 * config/m68k/m68k.h: Use TARGET_68040 instead of TARGET_68040_ONLY.
1386 (TARGET_68040_ONLY): Rename to...
1387 (TARGET_68040): ...this.
1388 * config/m68k/m68k.c: Use TARGET_68040 instead of TARGET_68040_ONLY.
1389 * config/m68k/m68k.md: Likewise.
1390
1391 2007-01-12 Julian Brown <julian@codesourcery.com>
1392 Nathan Sidwell <nathan@codesourcery.com>
1393 Richard Sandiford <richard@codesourcery.com>
1394
1395 * config.gcc (m680[012]0-*-*, m68k*-*-*): Set m68k_cpu_ident to
1396 the -mcpu= argument associated with the --with-cpu setting.
1397 Define M68K_DEFAULT_TUNE to the default -mtune= option,
1398 if different from the one implied by the -mcpu setting.
1399 Accept --with-cpu=FOO if FOO is listed in m68k-devices.def,
1400 using mcpu=FOO as the default CPU option. Set target_cpu_default2.
1401 * doc/invoke.texi: Mention ColdFire in the introduction to the
1402 m68k options. Document the new -march, -mcpu, -mtune, -mdiv,
1403 -mno-div and -mhard-float options. Make -m68881 a synonym for
1404 -mhard-float. Document the previously-undocumented -m5206e,
1405 -m528x, -m5307 and -m5407 options. Tweak the existing option
1406 documentation for consistency.
1407 * doc/install.texi: Mention new --with-cpu arguments.
1408 * config/m68k/m68k.h (OPTION_DEFAULT_SPECS): Only use the
1409 default CPU if neither -mcpu nor -march are specified.
1410 (ASM_CPU_SPEC): Pass down -mcpu and -march options.
1411 (TARGET_CPU_CPP_BUILTINS): Set __mcfisa*__ macros from
1412 TARGET_ISA*. Set the legacy __mcf*__ cpu macros in the same way,
1413 using m68k_tune to decide between families that implement the
1414 same ISA. Use m68k_tune to set __mcfv4e__.
1415 (FL_BITFIELD, FL_68881, FL_COLDFIRE, FL_CF_HWDIV, FL_CF_MAC)
1416 (FL_CF_EMAC, FL_CF_EMAC_B, FL_CF_USP, FL_CF_FPU, FL_ISA_68000)
1417 (FL_ISA_68010, FL_ISA_68020, FL_ISA_68040, FL_ISA_A, FL_ISA_B)
1418 (FL_ISA_C, FL_ISA_MMU): New macros.
1419 (MASK_COLDFIRE): Delete.
1420 (TARGET_68010, TARGET_68020, TARGET_68040_ONLY, TARGET_COLDFIRE)
1421 (TARGET_ISAB): Redefine in terms of m68k_cpu_flags.
1422 (TARGET_68881, TARGET_COLDFIRE_FPU): Redefine in terms of m68k_fpu.
1423 (TARGET_HARD_FLOAT): Do not define here.
1424 (TARGET_ISAAPLUS, TARGET_ISAC): New macros.
1425 (TUNE_68000): New macro.
1426 (TUNE_68000_10): Redefine in terms of TUNE_68000 and TUNE_68010.
1427 (TUNE_68010, TUNE_68030, TUNE_68040, TUNE_68060, TUNE_CPU32)
1428 (TUNE_CFV2): Redefine in terms of m68k_tune.
1429 (uarch_type, target_device, fpu_type): New enums.
1430 (m68k_cpu, m68k_tune, m68k_fpu, m68k_cpu_flags): Declare.
1431 * config/m68k/m68k.c (TARGET_DEFAULT): Remove MASK_68881.
1432 (FL_FOR_isa_00, FL_FOR_isa_10, FL_FOR_isa_20, FL_FOR_isa_40)
1433 (FL_FOR_isa_cpu32, FL_FOR_isa_a, FL_FOR_isa_aplus, FL_FOR_isa_b)
1434 (FL_FOR_isa_c): New macros.
1435 (m68k_isa): New enum.
1436 (m68k_target_selection): New structure.
1437 (all_devices, all_isas, all_microarchs): New tables.
1438 (m68k_cpu_entry, m68k_arch_entry, m68k_tune_entry, m68k_cpu)
1439 (m68k_tune, m68k_fpu, m68k_cpu_flags): New variables.
1440 (MASK_ALL_CPU_BITS): Delete.
1441 (m68k_find_selection): New function.
1442 (m68k_handle_option): Handle -mcpu=, -march= and -mtune=.
1443 Map the legacy target options to a combination of the new ones.
1444 (override_options): Set m68k_cpu, m68k_tune, m68k_fpu and
1445 m68k_cpu_flags. Handle M68K_DEFAULT_TUNE. Use m68k_cpu_flags
1446 to derive default MASK_BITFIELD, MASK_CF_HWDIV and MASK_HARD_FLOAT
1447 settings.
1448 * config/m68k/m68k.opt (m5200, m5206e, m528x, m5307, m5407, mcfv4e)
1449 (m68010, m68020, m68020-40, m68020-60, m68030, m68040): Remove Mask
1450 properties.
1451 (m68881, msoft-float): Change mask from 68881 to HARD_FLOAT.
1452 (march=, mcpu=, mdiv, mhard-float, mtune=): New options.
1453 * config/m68k/m68k-devices.def: New file.
1454
1455 2007-01-12 Richard Sandiford <richard@codesourcery.com>
1456 Nathan Sidwell <nathan@codesourcery.com>
1457
1458 * config/m68k/m68k.h (ASM_CPU_SPEC, ASM_SPEC, EXTRA_SPECS)
1459 (SUBTARGET_EXTRA_SPECS): New macros.
1460 * config/m68k/linux.h (ASM_SPEC): Remove CPU flags;
1461 use %(asm_cpu_spec) instead.
1462 * config/m68k/m68k-none.h (ASM_SPEC): Likewise.
1463 * config/m68k/openbsd.h (ASM_SPEC): Likewise.
1464 * config/m68k/netbsd-elf.h (ASM_SPEC): Likewise.
1465 (EXTRA_SPECS): Rename to...
1466 (SUBTARGET_EXTRA_SPECS): ...this.
1467
1468 2007-01-12 Nathan Sidwell <nathan@codesourcery.com>
1469 Richard Sandiford <richard@codesourcery.com>
1470 Julian Brown <julian@codesourcery.com>
1471
1472 * config.gcc (m68k-*-aout*, m68k-*-coff*, m68020-*-elf*, m68k-*-elf*)
1473 (m68k-*-uclinux*, m68k-*-linux*, m68k-*-rtems*): Set default_m68k_cpu
1474 to the configuration's default CPU.
1475 (m68010-*-netbsdelf*, m68k*-*-netbsdelf*, m68k*-*-openbsd*): Likewise.
1476 Remove default masks.
1477 (m680[012]0-*-*): Set the default with_cpu to the first part of
1478 the target name.
1479 (m68k*-*-*): Set the default with_cpu to m$default_m68k_cpu.
1480 (m68k*-*-linux): Extend the --with-cpu handling to...
1481 (m680[012]0-*-*, m68k*-*-*): ...these configurations. Allow m68000
1482 and m68010. Don't set target_cpu_default2.
1483 * doc/install.texi: Document --with-cpu for m68k.
1484 * config/m68k/m68k.h (OPTION_DEFAULT_SPECS): Define.
1485 * config/m68k/m68k-none.h (TARGET_CPU_DEFAULT, M68K_CPU_m68k)
1486 (M68K_CPU_m68000, M68K_CPU_m68010, M68K_CPU_m68020, M68K_CPU_m68030)
1487 (M68K_CPU_m68040, M68K_CPU_m68302, M68K_CPU_m68332, TARGET_DEFAULT)
1488 (ASM_CPU_DEFAULT_SPEC, CC1_CPU_DEFAULT_SPEC): Delete.
1489 (ASM_SPEC): Remove use of %(asm_cpu_default).
1490 (EXTRA_SPECS, SUBTARGET_EXTRA_SPECS, MULTILIB_DEFAULTS): Delete.
1491 * config/m68k/linux.h (TARGET_DEFAULT): Delete.
1492 (CPP_SPEC): Merge definitions. Do not handle __HAVE_68881__ here.
1493 * config/m68k/netbsd-elf.h (TARGET_OS_CPP_BUILTINS): Define
1494 __HAVE_FPU__ if TARGET_HARD_FLOAT.
1495 (TARGET_DEFAULT): Delete.
1496 (EXTRA_SPECS): Delete cpp_cpu_default_spec, cpp_cpu_spec,
1497 cpp_fpu_spec, asm_default_spec and netbsd_cpp_spec.
1498 (CPP_CPU_SPEC): Delete.
1499 (TARGET_VERSION): Merge definitions, using TARGET_68010 to pick
1500 the appropriate string.
1501 (CPP_CPU_DEFAULT_SPEC, ASM_DEFAULT_SPEC, CPP_FPU_SPEC): Delete.
1502 (CPP_SPEC): Define to NETBSD_CPP_SPEC.
1503 (ASM_SPEC): Don't use %(asm_default_spec).
1504 * config/m68k/m68k.c (TARGET_DEFAULT_TARGET_FLAGS): Remove
1505 TARGET_DEFAULT and add MASK_68881.
1506 * config/m68k/m68k.md: Remove mention of TARGET_DEFAULT from comments.
1507
1508 2007-01-12 Richard Sandiford <richard@codesourcery.com>
1509
1510 * config.gcc (m68010-*-netbsdelf*): Add MASK_68010.
1511 (m68k*-*-netbsdelf*, m68k*-*-openbsd*, m68k*-linux*): Add
1512 MASK_68010 alongside MASK_68020.
1513 * doc/invoke.texi: Document -m68010.
1514 * config/m68k/m68k.opt (m68010): New.
1515 * config/m68k/m68k.h (TARGET_CPU_CPP_BUILTINS): Define mc68010
1516 if TUNE_68010.
1517 (TUNE_68010): New macro.
1518 * config/m68k/m68k-none.h (M68K_CPU_m68k, M68K_CPU_m68010)
1519 (M68K_CPU_m68020, M68K_CPU_m68030, M68K_CPU_m68040)
1520 (M68K_CPU_m68332): Add MASK_68010.
1521 * config/m68k/linux.h (TARGET_DEFAULT): Add MASK_68010 to
1522 fallback definition.
1523 * config/m68k/netbsd-elf.h (CPP_CPU_SPEC): Remove now-redundant
1524 defines.
1525 * config/m68k/m68k.c (MASK_ALL_CPU_BITS): Add MASK_68010.
1526 (m68k_handle_option): Handle OPT_m68010. Add MASK_68010
1527 to all entries that use MASK_68020.
1528 (output_move_simode_const, output_move_himode, output_move_qimode)
1529 (output_move_stricthi, output_move_strictqi): Use TARGET_68010
1530 instead of TARGET_68020 to select clr behavior. Remove comment
1531 about there being no TARGET_68010.
1532 * config/m68k/m68k.md: Likewise throughout.
1533
1534 2007-01-12 Julian Brown <julian@codesourcery.com>
1535
1536 * config/m68k/m68k.h (TARGET_ISAB): New macro.
1537 * config/m68k/m68k.c: Use TARGET_ISAB rather than TARGET_CFV4.
1538 * config/m68k/m68k.md: Likewise.
1539
1540 2007-01-12 Julian Brown <julian@codesourcery.com>
1541
1542 * config/m68k/m68k.h (LEGITIMATE_INDEX_P, LEGITIMIZE_ADDRESS): Use
1543 TARGET_COLDFIRE_FPU instead of TARGET_CFV4E.
1544
1545 2007-01-12 Julian Brown <julian@codesourcery.com>
1546
1547 * config/m68k/m68k.h (TUNE_68040_60): New macro.
1548 * config/m68k/m68k.c (standard_68881_constant_p): Use it.
1549 * config/m68k/m68k.md: Likewise.
1550
1551 2007-01-12 Julian Brown <julian@codesourcery.com>
1552 Richard Sandiford <richard@codesourcery.com>
1553
1554 * config/m68k/m68k.h (TARGET_CPU_CPP_BUILTINS): Use TUNE_68030
1555 instead of TARGET_68030, TUNE_68040 instead of TARGET_68040,
1556 TUNE_68060 instead of TARGET_68060 and TUNE_CPU32 instead of
1557 TARGET_CPU32.
1558 (TARGET_CPU32): Rename to...
1559 (TUNE_CPU32): ...this.
1560 (TUNE_68000_10, TUNE_68030, TUNE_68040, TUNE_68060)
1561 (TUNE_CFV2): New macros.
1562 * config/m68k/netbsd-elf.h (LONG_DOUBLE_TYPE_SIZE): Simplify;
1563 remove conditions that are implied by TARGET_68020.
1564 * config/m68k/m68k.c (m68k_output_function_prologue): Use TUNE_68040
1565 instead of TARGET_68040 and TUNE_CPU32 instead of TARGET_CPU32.
1566 (m68k_output_function_epilogue): Likewise.
1567 (m68k_rtx_costs): Likewise. Use TUNE_68060 instead of TARGET_68060
1568 and TUNE_CFV2 instead of TARGET_5200. Use TUNE_68000_10 instead of
1569 "!TARGET_68020 && !TARGET_COLDFIRE" to choose between 68000 and
1570 non-68000 timings. Refactor multiplication and division costs.
1571 (output_addsi3): Use TUNE_68040 instead of TARGET_68040 and
1572 TUNE_CPU32 instead of TARGET_CPU32.
1573 (standard_68881_constant_p): Use TUNE_68040 instead of TARGET_68040
1574 and TUNE_68060 instead of TARGET_68060.
1575 * config/m68k/m68k.md: Use TUNE_68040 instead of TARGET_68040,
1576 TUNE_68060 instead of TARGET_68060, and TUNE_CPU32 instead of
1577 TARGET_CPU32.
1578 (movsi_const0): Use TUNE_68000_10 rather than "!TARGET_68020
1579 && !TARGET_COLDFIRE" to choose between moveq and clr.
1580 Likewise in the unnamed movsf pattern.
1581 (ashlsi_17_24, lshrsi_17_24): Guard with TUNE_68000_10 rather than
1582 "!TARGET_68020 && !TARGET_COLDFIRE". Likewise the unnamed
1583 ashiftrt pattern.
1584
1585 2007-01-12 Richard Sandiford <richard@codesourcery.com>
1586
1587 * config/m68k/m68k.h (TARGET_CPU_CPP_BUILTINS): Increase amount
1588 of tabbing before backslashes.
1589
1590 2007-01-11 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
1591
1592 * pa-linux.h (ASM_OUTPUT_INTERNAL_LABEL): Undefine.
1593 * pa.h (ASM_OUTPUT_LABEL): Output colon when using GAS.
1594 (ASM_OUTPUT_INTERNAL_LABEL): Define.
1595
1596 2007-01-11 Zdenek Dvorak <dvorakz@suse.cz>
1597
1598 * tree-ssa-loop-ivopts.c (extract_cond_operands): Split from
1599 find_interesting_uses_cond.
1600 (find_interesting_uses_cond): Use extract_cond_operands.
1601 (rewrite_use_compare): Use extract_cond_operands and
1602 force_gimple_operand_bsi. Do not call update_stmt.
1603 (determine_use_iv_cost_condition): Use extract_cond_operands.
1604 Return cheaper of using original bound and changing the exit bound.
1605
1606 2007-01-11 Zdenek Dvorak <dvorakz@suse.cz>
1607
1608 PR tree-optimization/29516
1609 * tree-ssa-address.c (tree_mem_ref_addr, add_to_parts,
1610 most_expensive_mult_to_index, addr_to_parts,
1611 create_mem_ref, maybe_fold_tmr): Make the type of
1612 fields of TARGET_MEM_REF sizetype.
1613 (move_fixed_address_to_symbol, move_pointer_to_base):
1614 New functions.
1615 * tree.def (TARGET_MEM_REF): Add comment on types of
1616 the operands.
1617
1618 2007-01-11 Joseph Myers <joseph@codesourcery.com>
1619
1620 * c-common.c (vector_types_convertible_p): Treat opaque types as
1621 always convertible if they have the same size, but not otherwise.
1622
1623 2007-01-11 Steven Bosscher <steven@gcc.gnu.org>
1624
1625 * ifcvt.c (struct noce_if_info): Add comments to the fields.
1626 Remove the b_unconditional field.
1627 (noce_try_sign_mask): Do not look at b_unconditional.
1628 (noce_process_if_block): Do not use merge_if_blocks. Update
1629 the CFG here. Do not set b_unconditional.
1630 (cond_move_process_if_block): Likewise.
1631 (find_cond_trap): Likewise.
1632 (check_cond_move_block): Require simple jump insns at the end
1633 of the basic block.
1634
1635 2007-01-11 Jan Hubicka <jh@suse.cz>
1636
1637 PR tree-optimization/1046
1638 * tree-tailcall.c (suitable_for_tail_call_opt_p): Use TREE_ADDRESSABLE
1639 when alias info is not ready.
1640 (pass_tail_recursion): Do not require aliasing.
1641 * tree-ssa-copyrename.c (pass_rename_ssa_cop): Likewise.
1642 * tree-ssa-ccp.c (pass_ccp, pass_fold_builtins): Likewise.
1643 * tree-ssa-copy.c (pass_copy_prop): Likewise.
1644 * tree-ssa-forwprop.c (pass_forwprop): Likewise.
1645 * tree-ssa-dce.c (pass_dce, pass_dce_loop, pass_cd_dce): Likewise.
1646 * passes.c (init_optimization_passes): Execute rename_ssa_copies,
1647 ccp, forwprop, copy_prop, merge_phi, copy_prop, dce and tail recursion
1648 before inlining.
1649 * tree-ssa-operands.c (add_virtual_operand, get_indirect_ref_operand):
1650 When aliasing is not build, mark statement as volatile.
1651
1652 2007-01-11 Tom Tromey <tromey@redhat.com>
1653
1654 PR preprocessor/15185, PR preprocessor/20989:
1655 * doc/cppopts.texi <-MT>: Update description of algorithm for
1656 computing default target.
1657 <-M, -MD>: Reword "basename" text.
1658
1659 2007-01-11 Roger Sayle <roger@eyesopen.com>
1660
1661 * builtins.c (expand_builtin_pow, expand_builtin_powi,
1662 fold_builtin_cabs, fold_builtin_sqrt, fold_builtin_trunc,
1663 fold_builtin_floor, fold_builtin_ceil, fold_builtin_round,
1664 fold_builtin_int_int_roundingfn, fold_builtin_bitop,
1665 fold_builtin_bswap, real_constp, fold_builtin_pow,
1666 fold_builtin_powi, fold_builtin_signbit, fold_builtin_copysign,
1667 do_mpfr_arg1, do_mpfr_arg2, do_mpfr_arg3, do_mpfr_sincos): Replace
1668 uses of the macro TREE_CONSTANT_OVERFLOW with TREE_OVERFLOW.
1669 * convert.c (convert_to_pointer): Likewise.
1670 * expr.c (highest_pow2_factor, expand_expr_real_1): Likewise.
1671 * fold-const.c (force_fit_type, fold_negate_expr, int_const_binop,
1672 const_binop, fold_convert_const_int_from_int,
1673 fold_convert_const_int_from_real,
1674 fold_convert_const_real_from_real, sign_bit_p,
1675 optimize_minmax_comparison, extract_muldiv_1, fold_div_compare,
1676 fold_sign_changed_comparison, fold_unary, fold_comparison,
1677 fold_binary, multiple_of_p, tree_Expr_non_zero_p,
1678 fold_negate_const, fold_abs_const, fold_not_const): Likewise.
1679 * print-tree.c (print_node_brief, print_node): Likewise.
1680 * stor-layout.c (place_field, layout_type): Likewise.
1681 * tree-chrec.c (keep_cast): Likewise.
1682 * tree.c (build_vector, build_real, build_real_from_int_cst,
1683 build_complex): Likewise.
1684
1685 2007-01-11 Roger Sayle <roger@eyesopen.com>
1686
1687 * tree.h (TREE_CONSTANT_OVERFLOW): Obsolete. For the time being,
1688 treat TREE_CONSTANT_OVERFLOW as a synonym of TREE_OVERFLOW.
1689
1690 2007-01-11 Paolo Bonzini <bonzini@gnu.org>
1691
1692 * configure.ac (strict1_warn): Rename to strict_warn.
1693 (WERROR, --enable-werror, symlink hacks, stage1_cflags,
1694 cc_set_by_configure, quoted_cc_set_by_configure,
1695 stage_prefix_set_by_configure, quoted_stage_prefix_set_by_configure,
1696 all_boot_languages, all_stagestuff): Remove.
1697 (target_list): Remove bootstrap targets.
1698 * Makefile.in (quickstrap): Unconditionally make a synonym of all.
1699 (BOOT_LANGUAGES, STAGE1_CFLAGS, STAGE1_CHECKING,
1700 REMAKEFLAGS, FLAGS_TO_PASS, PREPEND_DOTDOT_TO_RELATIVE_PATHS,
1701 SUBDIR_FLAGS_TO_PASS, WERROR_FLAGS, STRICT2_WARN, LANG_STAGESTUFF,
1702 VOL_FILES, POSTSTAGE1_FLAGS_TO_PASS, STAGE2_FLAGS_TO_PASS,
1703 STAGEPROFILE_FLAGS_TO_PASS, STAGEFEEDBACK_FLAGS_TO_PASS, stage1_build,
1704 stage1_copy, stage2_build, stage2_copy, stageprofile_build,
1705 stageprofile_copy, stage3_build, stage3_copy, stagefeedback_build,
1706 stagefeedback_copy, stage4_build, clean_s1, clean_sw, bootstrap,
1707 bootstrap-lean, bootstrap2, bootstrap2-lean, bootstrap3,
1708 bootstrap3-lean, bootstrap4, bootstrap4-lean, unstage1, unstage2,
1709 unstage3, unstage4, unstageprofile, unstagefeedback, restage, restage2,
1710 restage3, restage4, restageprofile, restagefeedback, bubbleestrap,
1711 cleanstrap, unstrap, restrap, *compare, *compare3, *compare4,
1712 *compare-lean, *compare3-lean, *compare4-lean, stage1-start, stage1,
1713 stage2-start, stage2, stage3-start, stage3, stage4-start, stage4,
1714 stageprofile-start, stageprofile, stagefeedback-start, stagefeedback,
1715 risky-stage1, risky-stage2, risky-stage3, risky-stage4): Remove.
1716 (ORDINARY_FLAGS_TO_PASS): Rename to FLAGS_TO_PASS.
1717 (STAGECOPYSTUFF, STAGEMOVESTUFF): Consolidate into MOSTLYCLEANFILES.
1718 (mostlyclean): Adjust.
1719 (clean, distclean): Don't mention bootstrap stuff.
1720 * configure: Regenerate.
1721 * ada/config-lang.in, cp/config-lang.in, forttran/config-lang.in,
1722 java/config-lang.in, objc/config-lang.in, objcp/config-lang.in,
1723 treelang/config-lang.in (stagestuff): Remove.
1724 * doc/sourcebuild.texi (stage1, stage2, stage3, stage4,
1725 stageprofile, stagefeedback, stagestuff): Remove mention.
1726
1727 2007-01-11 Nick Clifton <nickc@redhat.com>
1728
1729 * config/mcore/predicates.md (mcore_general_movesrc_operand):
1730 Accept CONSTs.
1731 (mcore_general_movdst_operand): Do not accept CONST_INTs.
1732 (mcore_arith_K_S_operand): Run the test for the S constraint not
1733 the test for the M constraint.
1734 (mcore_addsub_operand): Do not accept integer values that are
1735 larger than 32 bits.
1736 * config/mcore/mcore.md: Remove unused constraints from split.
1737 (andsi3): Use HOST_WIDE_INT instead of int to hold an INTVAL.
1738 (addsi3): Likewise.
1739 (allocate_stack): Likewise.
1740 * config/mcore/mcore.c (mcore_print_operand): Restrict output of P
1741 operands to 32 bits.
1742 (mcore_const_costs): Use HOST_WIDE_INT instead of int to hold an
1743 INTVAL.
1744 (mcore_and_cost, mcore_modify_comparison, const_ok_for_mcore,
1745 mcore_const_ok_for_inline, mcore_const_trick_uses_not,
1746 try_constant_tricks, mcore_num_ones, mcore_num_zeros,
1747 mcore_output_bclri, mcore_output_andn, output_inline_const,
1748 mcore_output_move, mcore_output_movedouble): Likewise.
1749 (mcore_output_cmov): Use CONST_OK_FOR_M and CONST_OK_FOR_N.
1750 (output_inline_const): Likewise.
1751 (output_inline_const): Fix format strings used in sprintf
1752 statements.
1753 * config/mcore/mcore-protos.h: Update prototypes for changed
1754 functions in mcore.c.
1755 * config/mcore/mcore.h (CONST_OK_FOR_I): Cast values to
1756 HOST_WIDE_INT and not int.
1757 (CONST_OK_FOR_J, CONST_OK_FOR_K, CONST_OK_FOR_L, CONST_OK_FOR_M,
1758 CONST_OK_FOR_N): Likewise.
1759 (LEGITIMATE_CONSTANT_P): Also check CONSTANT_P.
1760 (GO_IF_LEGITIMATE_INDEX): Use HOST_WIDE_INT instead of int to hold
1761 an INTVAL.
1762
1763 2007-01-10 Jan Hubicka <jh@suse.cz>
1764
1765 * tree-vrp.c (remove_range_assertions): Release defs.
1766 * tree-ssa-loop-ivopts.c (rmeove_statement): Likewise.
1767 * tree-ssa-dom.c (remove_stmt_or_phi): Likewise.
1768
1769 2007-01-10 Paul Brook <paul@codesourcery.com>
1770
1771 * config/arm/arm.c (arm_rtx_costs_1): Handle mutiply-subtract.
1772 * config/arm/arm.md (mulsi3subsi): New insn.
1773
1774 2007-01-10 Zdenek Dvorak <dvorakz@suse.cz>
1775
1776 * tree-ssa-loop-manip.c (tree_unroll_loop): Make it a wrapper over ...
1777 (tree_transform_and_unroll_loop): New.
1778 * tree-flow.h (transform_callback, tree_transform_and_unroll_loop):
1779 Declare.
1780
1781 2007-01-10 Robert Kennedy <jimbob@google.com>
1782
1783 * fold-const.c (fold_comparison): Fold comparisons like (x *
1784 1000 < 0) to (x < 0).
1785
1786 2007-01-10 Ian Lance Taylor <iant@google.com>
1787
1788 * tree-pretty-print.c (dump_generic_node): Print parentheses when
1789 operands have the same priority.
1790
1791 2007-01-10 Tom Tromey <tromey@redhat.com>
1792
1793 * fold-const.c (fold_truthop): Don't check can_use_bit_fields_p.
1794 (fold_binary): Likewise.
1795 * langhooks.c (lhd_can_use_bit_fields_p): Removed.
1796 * langhooks-def.h (lhd_can_use_bit_fields_p): Removed.
1797 (LANG_HOOKS_CAN_USE_BIT_FIELDS_P): Removed.
1798 (LANG_HOOKS_INITIALIZER): Remove LANG_HOOKS_CAN_USE_BIT_FIELDS_P.
1799 * langhooks.h (struct lang_hooks): Removed field
1800 'can_use_bit_fields_p'.
1801
1802 2007-01-10 Ralf Corsépius <ralf.corsepius@rtems.org>
1803
1804 * config/bfin/t-bfin, config/bfin/t-bfin-elf: Remove GCC_CFLAGS.
1805
1806 2007-01-10 Razya Ladelsky <razya@il.ibm.com>
1807
1808 * function.c (get_last_funcdef_no): New function.
1809 * function.h (get_last_funcdef_no): Declare.
1810 * tree-inline.c (initialize_cfun): Add initialization.
1811 (tree_function_versioning): Cleanup.
1812
1813 2007-01-10 Jan Hubicka <jh@suse.cz>
1814
1815 * tree-inline.c (setup_one_parameter): Do not propagate into abnormal
1816 PHIs.
1817
1818 2007-01-10 Sa Liu <saliu@de.ibm.com>
1819 Ben Elliston <bje@au.ibm.com>
1820
1821 * spu.h (STACK_SAVE_AREA): Use VOIDmode for SAVE_FUNCTION, SImode
1822 for SAVE_NONLOCAL and Pmode for any other save level.
1823 * spu-protos.h (spu_restore_stack_block): Declare.
1824 * spu.md (save_stack_block): Remove.
1825 (restore_stack_block): Call spu_restore_stack_block.
1826 * spu.c (spu_restore_stack_block): New function.
1827 (spu_expand_epilogue): Remove old comment.
1828
1829 2007-01-09 Zdenek Dvorak <dvorakz@suse.cz>
1830
1831 PR tree-optimization/30322
1832 * tree-ssa-loop-ivopts.c (fold_affine_expr, iv_value): Removed.
1833 (cand_value_at): Return the value as aff_tree.
1834 (may_eliminate_iv): Convert the bound from aff_tree to tree.
1835 * tree-affine.c (aff_combination_add_cst, aff_combination_add_product,
1836 aff_combination_mult): New functions.
1837 (aff_combination_add): Use aff_combination_add_cst.
1838 (aff_combination_convert): Allow conversions to a wider type.
1839 (tree_to_aff_combination): Handle BIT_NOT_EXPR.
1840 * tree-affine.h (aff_combination_mult): Declare.
1841
1842 2007-01-09 Carlos O'Donell <carlos@codesourcery.com>
1843
1844 * doc/tm.texi: Update documentation to reflect reality of exec
1845 and start file search behaviours. Update copyright year.
1846 * doc/invoke.texi: Explain how GCC_EXEC_PREFIX is used to find
1847 header file directories.
1848
1849 2007-01-09 Uros Bizjak <ubizjak@gmail.com>
1850
1851 * config/i386/i386.md (*sinxf2): Rename to *sinxf2_i387.
1852 (*cosxf2): Rename to cosxf2_i387.
1853 (*sindf2, *sinsf2): Extend operand 1 to XFmode. Macroize patterns
1854 using X87MODEF12 mode macro. Rename patterns to
1855 *sin_extend<mode>xf2_i387. Use SSE_FLOAT_MODE_P to disable patterns
1856 for SSE math.
1857 (*cosdf2, *cossf2): Ditto.
1858 (sincosdf3, sincossf3): Ditto. Rewrite corresponding splitters
1859 to match extended input operands.
1860 (sincos<mode>3): New expander.
1861 (*sinextendsfdf2, *cosextendsfdf2, *sincosextendsfdf3): Remove
1862 insn patterns and corresponding splitters.
1863
1864 2007-01-09 Kaz Kojima <kkojima@gcc.gnu.org>
1865
1866 * config/sh/t-linux (TARGET_LIBGCC2_CFLAGS): Delete.
1867 (SHLIB_MAPFILES, SHLIB_LINK, SHLIB_INSTALL): Likewise.
1868
1869 2007-01-09 Nicolas Pitre <nico@cam.org>
1870
1871 PR target/30173
1872 * arm/ieee754-df.S (Lad_s): Also test the low word of X for zero.
1873
1874 2007-01-08 Geoffrey Keating <geoffk@apple.com>
1875
1876 * target.h (struct gcc_target): New field library_rtti_comdat.
1877 * target-def.h (TARGET_CXX_LIBRARY_RTTI_COMDAT): New.
1878 (TARGET_CXX): Add TARGET_CXX_LIBRARY_RTTI_COMDAT.
1879 * doc/tm.texi (C++ ABI): Document TARGET_CXX_LIBRARY_RTTI_COMDAT.
1880 * config/darwin.h (TARGET_CXX_LIBRARY_RTTI_COMDAT): Define.
1881
1882 2007-01-08 Geoffrey Keating <geoffk@apple.com>
1883
1884 * doc/invoke.texi (Optimize Options): Correct description of -O0.
1885
1886 2007-01-08 Richard Guenther <rguenther@suse.de>
1887
1888 * tree.h (force_fit_type_double): Export.
1889 (force_fit_type): Remove.
1890 * fold-const.c (force_fit_type_double): New function.
1891 (force_fit_type): Remove.
1892 (int_const_binop): Use it.
1893 (fold_convert_const_int_from_int): Likewise.
1894 (fold_convert_const_int_from_real): Likewise.
1895 (fold_div_compare): Likewise.
1896 (fold_sign_changed_comparison): Likewise.
1897 (fold_unary): Likewise.
1898 (fold_negate_const): Likewise.
1899 (fold_abs_const): Likewise.
1900 (fold_not_const): Likewise.
1901 * c-common.c (shorten_compare): Use force_fit_type_double.
1902 * convert.c (convert_to_pointer): Likewise.
1903
1904 2007-01-08 Richard Guenther <rguenther@suse.de>
1905
1906 * tree.h (build_int_cst_wide_type): Export.
1907 * tree.c (build_int_cst_wide_type): New function.
1908 (build_int_cst_wide): Fix comment.
1909 * builtins.c (fold_builtin_object_size): Use build_int_cst
1910 to build -1 or 0 of the correct type. Use fit_double_type
1911 to check for overflow.
1912 * fold-const.c (optimize_bit_field_compare): Use build_int_cst_type
1913 to build the mask.
1914 (decode_field_reference): Likewise.
1915 (all_ones_mask_p): Likewise.
1916 (native_interpret_int): Use build_int_cst_wide_type.
1917 (fold_binary): Use build_int_cst_type to build an all-ones
1918 value.
1919 * stor-layout.c (set_sizetype): Use build_int_cst_wide_type.
1920
1921 2007-01-08 Daniel Jacobowitz <dan@codesourcery.com>
1922
1923 * config/pa/t-pa64 (libgcc_stub.a): Use $(T).
1924
1925 2007-01-09 Ben Elliston <bje@au.ibm.com>
1926
1927 * genautomata.c (STATS_OPTION): New option.
1928 (stats_flag): New flag.
1929 (gen_automata_option): Handle it.
1930 (initiate_automaton_gen): Ditto.
1931 (write_automata): Output statistics only if stats_flag is
1932 set. Likewise, output time statistics only if time_flag is set.
1933 * doc/md.texi (Processor pipeline description): Document new flag.
1934
1935 2007-01-08 Richard Guenther <rguenther@suse.de>
1936
1937 * builtins.c (fold_builtin_int_roundingfn): Use fit_double_type.
1938 * tree.c (build_int_cst_type): Likewise.
1939 (size_in_bytes): Don't call force_fit_type on the result.
1940 (int_fits_type_p): Use fit_double_type.
1941 * fold-const.c (fit_double_type): New function.
1942 (force_fit_type): Use it.
1943 * tree.h (fit_double_type): Export.
1944
1945 2007-01-08 Jan Hubicka <jh@suse.cz>
1946
1947 * tree-vectorizer.c (gate_increase_alignment): Fix return type.
1948 * ipa.c (function_and_variable_visibility): Fix return type.
1949
1950 2007-01-08 Richard Guenther <rguenther@suse.de>
1951
1952 * tree-ssa-ccp.c (maybe_fold_offset_to_array_ref): Use type
1953 of offset to build the index.
1954 * tree-pretty-print.c (dump_generic_node): Don't build negated
1955 const just for printing.
1956 * c-pretty-print.c (pp_c_integer_constant): Likewise.
1957 * builtins.c (fold_builtin_int_roundingfn): Check if result
1958 fits the type by using force_fit_type and comparing the result.
1959 * predict.c (predict_loops): Use compare_tree_int for comparison.
1960 * tree.c (build_int_cst): Fall back to integer_type_node for
1961 NULL_TREE type.
1962 (build_int_cst_wide): Assert type is non-null.
1963
1964 2007-01-08 Roberto Costa <roberto.costa@st.com>
1965
1966 * tree-vrp.c (extract_range_from_cond_expr): New.
1967 (extract_range_from_expr): Handle COND_EXPR nodes used as expressions.
1968 * tree-ssa-ccp.c (get_maxval_strlen): Handle COND_EXPR nodes used
1969 as expressions.
1970 (fold_stmt): Bug fix, avoid infinite recursion when folding COND_EXPRs.
1971 * tree-ssa-forwprop.c (simplify_cond, forward_propagate_into_cond,
1972 tree_ssa_forward_propagate_single_use_vars): Handle COND_EXPR nodes
1973 used as expressions.
1974 * tree-object-size.c (cond_expr_object_size): New.
1975 (collect_object_sizes_for): Handle COND_EXPR nodes used as expressions.
1976
1977 2007-01-08 Jan Hubicka <jh@suse.cz>
1978
1979 * tree-ssa-forwprop.c (forward_propagate_into_cond,
1980 tree_ssa_forward_propagate_single_use_va): Release defs of propagated
1981 statement.
1982
1983 2007-01-08 Richard Guenther <rguenther@suse.de>
1984
1985 PR tree-optimization/23603
1986 * tree-vrp.c (set_value_range_to_truthvalue): New function.
1987 (extract_range_from_binary): Fall back to truthvalue instead of
1988 varying for TRUTH_*_EXPR.
1989 (extract_range_from_comparison): Fall back to truthvalue instead of
1990 varying.
1991 (vrp_visit_phi_node): Don't adjust new range bounds to +INF/-INF
1992 if all visited PHI values were constant.
1993
1994 2007-01-08 Jan Hubicka <jh@suse.cz>
1995
1996 * cgraphunit.c (cgraph_process_new_functions): Reset reachable flag.
1997 (cgraph_analyze_function): break out from ...
1998 (cgraph_finalize_compilation_unit): ... here.
1999 (cgraph_expand_function): Remove forgoten commented out line.
2000 (cgraph_optimize): Analyze functions.
2001
2002 2007-01-08 Jan Hubicka <jh@suse.cz>
2003
2004 * tree-pas.h (TODO_remove_function): New flag.
2005 (TODO_update*): Renumber.
2006 (pass_ipa_increase_alignment,
2007 pass_ipa_function_and_variable_visibility): New passes.
2008 * cgraphunit.c (cgraph_increase_alignment): Move to tree-vectorizer.c
2009 (cgraph_function_and_variable_visibility): Move to ipa.c
2010 (cgraph_optimize): Don't call cgraph_function_and_variable_visibility,
2011 cgraph_increase_alignment.
2012 * ipa-inline.c (cgraph_decide_inlining): Don't push timevar.
2013 (cgraph_decide_inlining_incrementally): Push TV_INTEGRATION before
2014 calling tree-inline.
2015 (cgraph_early_inlining): Do not call cgraph_remove_unreachable_nodes.
2016 (pass_ipa_inline, pass_early_ipa_inlining): Set TODO_remove_functions
2017 * tree-vectorizer.c (increase_alignment): Move here from cgraphunit.c
2018 (gate_increase_alignment): New function.
2019 (pass_ipa_increase_alignment): New pass.
2020 * ipa.c: Inline tree-pass.h and timevar.h
2021 (function_and_variable_visibility): Move here from cgraphunit.c
2022 * tree-optimize.c (pass_early_local_passes): Add TODO_remove_functions.
2023 * passes.c (init_optimization_passes): Add the two new passes.
2024 (execute_todo): Handle cgraph_remove_functions.
2025
2026 2007-01-08 Nick Clifton <nickc@redhat.com>
2027
2028 * config/frv/predicates.md (reg_or_0_operand): Accept
2029 CONST_DOUBLEs.
2030
2031 2007-01-08 Ralf Corsépius <ralf.corsepius@rtems.org>
2032
2033 * config/bfin/rtems.h, config/bfin/t-rtems: New.
2034 * config.gcc: Add bfin*-rtems*.
2035
2036 2007-01-08 Mark Shinwell <shinwell@codesourcery.com>
2037
2038 * c.opt: Add -flax-vector-conversions.
2039 * c-typeck.c (convert_for_assignment): Pass flag to
2040 vector_types_convertible_p to allow emission of note.
2041 (digest_init): Likewise.
2042 * c-opts.c: Handle -flax-vector-conversions.
2043 * c-common.c (flag_lax_vector_conversions): New.
2044 (vector_types_convertible_p): Unless -flax-vector conversions
2045 has been passed, disallow conversions between vectors with
2046 differing numbers of subparts and/or element types. If such
2047 a conversion is disallowed, possibly emit a note on the first
2048 occasion only to inform the user of -flax-vector-conversions.
2049 The new last argument specifies this.
2050 * c-common.h (flag_lax_vector_conversions): New.
2051 (vector_types_convertible_p): Add extra argument.
2052 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use
2053 char_type_node for V*QI type vectors.
2054 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
2055 Update to satisfy new typechecking rules.
2056 * config/rs6000/altivec.h (vec_cmple): Use vec_cmpge, for both
2057 C and C++ variants.
2058 * doc/invoke.texi (C Dialect Options): Document
2059 -flax-vector-conversions.
2060
2061 2007-01-08 Mark Shinwell <shinwell@codesourcery.com>
2062
2063 PR tree-optimization/29877
2064 * tree-ssa-ter.c (is_replaceable_p): Deem assignments with
2065 a register variable on the RHS to not be replaceable.
2066
2067 2007-01-08 Chen Liqin <liqin@sunnorth.com.cn>
2068 * config/score/t-score-elf (MULTILIB_OPTIONS): Change.
2069 * config/score/predicates.md (const_uimm5, sr0_operand, const_simm12,
2070 const_simm15, const_pow2, const_npow2): Added.
2071 * config/score/misc.md (insv, extv, extzv, movmemsi,
2072 move_lbu_a/b, mov_lhu_a/b etc): Added and fix some bug.
2073 * config/score/score.c (score_address_cost, score_select_cc_mode):
2074 Added.
2075 Change CONST_OK_FOR_LETTER_P/EXTRA_CONSTRAINT define.
2076 Update score_rtx_costs for MACRO TARGET_RTX_COSTS.
2077 Update score_print_operand.
2078 * config/score/score.h (DATA_ALIGNMENT, SELECT_CC_MODE): Added.
2079 Adjust register allocate order and update some macro define.
2080 * config/score/score-mdaux.c (mdx_unaligned_load, mdx_unsigned_store,
2081 mdx_block_move_straight, mdx_block_move_loop_head,
2082 mdx_block_move_loop_body, mdx_block_move_loop_foot, mdx_block_move_loop,
2083 mdx_block_move): Added.
2084 (mdx_movsicc, mdp_select_add_imm, mdp_select, mds_zero_extract_andi,
2085 mdp_limm): Updated and fix some bug and typo.
2086 * config/score/score.md (movqi/hi/si, add/sub/zero/ext): Updated.
2087 (movsf, movdf, doloop_end): Added.
2088
2089 2007-01-08 Kazu Hirata <kazu@codesourcery.com>
2090
2091 * config/arm/arm.c, config/arm/arm.h, config/arm/arm.md,
2092 config/arm/thumb2.md: Fix comment typos.
2093 * doc/extend.texi: Fix a typo.
2094
2095 2007-01-07 Eric Christopher <echristo@apple.com>
2096
2097 * configure.ac: Check for __stack_chk_fail for darwin.
2098 * configure: Regenerate.
2099
2100 2007-01-07 Richard Guenther <rguenther@suse.de>
2101
2102 * tree-vrp.c (extract_range_from_assert): CSE calls to
2103 compare_values where possible.
2104 (extract_range_from_unary_expr): Likewise.
2105
2106 2007-01-07 Anatoly Sokolov <aesok@post.ru>
2107
2108 * config/avr/avr-protos.h (call_insn_operand): Delete prototype.
2109 * config/avr/avr.c (call_insn_operand): Delete function.
2110 * config/avr/avr.md (*pushqi, *pushhi, *pushsi, *pushsf): Use REG_SP
2111 instead of register number. Use predicates.
2112 * config/avr/predicates.md (const0_operand, reg_or_0_operand,
2113 call_insn_operand): Add.
2114
2115 2007-01-06 Jan Hubicka <jh@suse.cz>
2116
2117 * tree-pass.h (pass_build_cgraph_edges): Declare.
2118 * cgraphunit.c (record_refernece): Move to cgraphbuild.c
2119 (visited_nodes): Remove.
2120 (cgraph_create_edges): Move to cgraphbuild.c; rename to
2121 build_cgrpah_edges; make visited_nodes local.
2122 (cgraph_process_new_functions): DO not call initialize_inline_failed.
2123 (record_references_in_initializer): Move to cgraphbuild.c
2124 (initialize_inline_failed, rebuild_cgraph_edges,
2125 pass_rebuild_cgraph_edges): Move to cgraphbuild.c.
2126 (verify_cgraph_node): Make visited_nodes local.
2127 (cgraph_analyze_function): Do not call cgraph_create_edges and
2128 initialize_inline_failed.
2129 (cgraph_expand_function): Do not call cgraph_lower_function;
2130 assert that function is already lowered.
2131 * Makefile.in (cgraphbuild.o): New.
2132 * passes.c (init_optimization_passes): Add pass_build_cgraph_edges
2133 at the end of lowering passes.
2134
2135 2007-01-06 Steven Bosscher <steven@gcc.gnu.org>
2136
2137 * ifcvt.c (cond_move_convert_if_block): New function, code
2138 factored out from...
2139 (cond_move_process_if_block): ...here. Call the new function
2140 on the THEN and ELSE blocks.
2141 (merge_if_block): Do not copy global_live_at_end, merge_blocks
2142 already takes care of this.
2143
2144 2007-01-05 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
2145
2146 PR c/19978
2147 * tree.h (TREE_OVERFLOW_P): New.
2148 * c-typeck.c (parser_build_unary_op): Warn only if result
2149 overflowed and operands did not.
2150 (parser_build_binary_op): Likewise.
2151 (convert_for_assignment): Remove redundant overflow_warning.
2152 * c-common.c (overflow_warning): Don't check or set TREE_OVERFLOW.
2153
2154 2007-01-05 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
2155
2156 * c-typeck.c (store_init_value): Split over two lines to follow
2157 the GNU coding style.
2158
2159 2007-01-05 Benjamin Kosnik <bkoz@redhat.com>
2160
2161 * c-cppbuiltin.c (c_cpp_builtins): __GXX_EXPERIMENTAL_CPP0X__ to
2162 __GXX_EXPERIMENTAL_CXX0X__.
2163 * doc/cpp.texi: Same.
2164
2165 2007-01-05 Richard Guenther <rguenther@suse.de>
2166
2167 PR middle-end/27826
2168 * tree.c (get_narrower): Do not construct COMPONENT_REFs
2169 with mismatched types. Instead explicitly build a
2170 conversion NOP_EXPR.
2171
2172 2007-01-05 Ian Lance Taylor <iant@google.com>
2173
2174 * c-common.c (decl_with_nonnull_addr_p): New function.
2175 (c_common_truthvalue_conversion): Call it.
2176 * c-typeck.c (build_binary_op): Likewise.
2177 * c-common.h (decl_with_nonnull_addr_p): Declare.
2178
2179 2007-01-05 Jakub Jelinek <jakub@redhat.com>
2180
2181 PR c/30360
2182 * libgcc2.c (__divdc3): Compare c and d against 0.0 instead of
2183 denom against 0.0.
2184
2185 2007-01-05 Joel Brobecker <brobecker@adacore.com>
2186
2187 * doc/install.texi (Final install): Document the fact that
2188 the GNAT runtime should not be stripped.
2189
2190 2007-01-04 Jan Hubicka <jh@suse.cz>
2191
2192 * tree-inline.c (fold_marked_statements): Update operand caches
2193 and EH after folding
2194
2195 2007-01-04 Ian Lance Taylor <iant@google.com>
2196
2197 * c-common.c (check_function_nonnull): Whitespace fix.
2198
2199 2007-01-04 Jan Hubicka <jh@suse.cz>
2200
2201 * tree-optimize.c (execute_fixup_cfg): Correct previously mistakely
2202 comitted older version of patch.
2203 (pass_fixup_cfg): Add TODOs to verify flow and statements, dump
2204 function, celanup cfg and collect garbage.
2205
2206 2007-01-04 Mike Stump <mrs@apple.com>
2207
2208 * Makefile.in (mostlyclean): Don't remove libgcc anymore.
2209 (clean): Likewise.
2210
2211 2007-01-04 Eric Christopher <echristo@apple.com>
2212
2213 * libgcc2.c (__bswapsi2): Use SItype.
2214 (__bswapdi2): Use DItype.
2215 * libgcc2.h: Update for above.
2216
2217 2007-01-04 Paul Brook <paul@codesourcery.com>
2218
2219 * config/arm/arm.md (arm_mulsi3, thumb_mulsi3, mulsi3_compare0,
2220 mulsi_compare0_scratch, mulsi3addsi, mulsi3addsi_compare0,
2221 mulsi3addsi_compare0_scratch, mulsidi3adddi, mulsidi3,
2222 umulsidi3, umulsidi3adddi, smulsi3_highpart,
2223 umulsi3_highpart): Make conditional on !arm_arch6.
2224 (arm_mulsi3_v6, thumb_mulsi3_v6, mulsi3_compare0_v6,
2225 mulsi_compare0_scratch_v6, mulsi3addsi_v6, mulsi3addsi_compare0_v6,
2226 mulsi3addsi_compare0_scratch_v6, mulsidi3adddi_v6, mulsidi3_v6,
2227 umulsidi3_v6, umulsidi3adddi_v6, smulsi3_highpart_v6,
2228 umulsi3_highpart_v6): New insns.
2229
2230 2007-01-04 Roger Sayle <roger@eyesopen.com>
2231
2232 * fold-const.c (fold_convert): When casting an expression to void,
2233 fold_ignored_result may discover a GIMPLE_MODIFY_STMT which doesn't
2234 have a type. Instead of attempting to build a NOP_EXPR, return
2235 these "special" trees directly.
2236
2237 2007-01-04 Joseph Myers <joseph@codesourcery.com>
2238
2239 * config/rs6000/rs6000.c (rs6000_rtx_costs): Make adjustment for
2240 MULT inside MINUS as either argument. Use rs6000_cost->dmul -
2241 rs6000_cost->fp not 0 as adjustment for outer NEG.
2242
2243 2007-01-04 Jan Hubicka <jh@suse.cz>
2244
2245 * cgraph.c (cgraph_release_function_body): New function.
2246 (cgraph_remove_node): Use it.
2247 * cgraph.h (cgraph_release_function_body): Declare.
2248 * cgraphunit.c (cgraph_expand_function): Use it.
2249 * ipa.c (cgraph_remove_unreahchable_nodes): Use it.
2250 * tree-ssa.c (delete_tree_ssa): Allow to be called before aliasing
2251 is initialized and while compilation of other function is running.
2252 * tree-optimize.c (execute_free_cfg_annotations): Move code to clear
2253 statement CFG annotations from here to ...
2254 * tree-cfg.c (delete_tree_cfg_annotations): ... here.
2255
2256 2007-01-04 Zdenek Dvorak <dvorakz@suse.cz>
2257
2258 * cfgloop.h (enum li_flags): Make the constants powers of two.
2259
2260 2007-01-04 Jan Hubicka <jh@suse.cz>
2261
2262 * tree-inline.c (copy_bb): Insert new statements to statements_to_fold
2263 set.
2264 (fold_marked_statements): New function.
2265 (optimize_inline_calls, tree_function_versioning): Fold new statements.
2266 * tree-inline.h (copy_body_data): Add statements_to_fold.
2267
2268 2007-01-03 Daniel Jacobowitz <dan@codesourcery.com>
2269
2270 * config.gcc: Mention libgcc/config.host.
2271 * Makefile.in: Update comments mentioning libgcc.
2272 (LIBGCC, INSTALL_LIBGCC, GCC_PARTS, mklibgcc): Delete.
2273 (all.cross, start.encap, rest.encap, rest.cross): Update
2274 dependencies for libgcc move.
2275 (libgcc.mk, LIBGCC_DEPS, libgcov.a, libgcc.a, stmp-multilib)
2276 (clean-target, clean-target-libgcc): Delete.
2277 (srcdirify, GCC_EXTRA_PARTS): New macros.
2278 (libgcc-support, libgcc.mvars): New rules.
2279 (distclean): Remove mention of mklibgcc.
2280 (install): Don't reference INSTALL_LIBGCC.
2281 (install-common): Don't reference EXTRA_PARTS.
2282 (install-libgcc, install-multilib): Delete rules.
2283 * mklibgcc.in: Delete file.
2284 * doc/configfiles.texi: Don't mention mklibgcc.
2285
2286 * config/i386/t-darwin (SHLIB_VERPFX): Delete (moved to libgcc).
2287 * config/i386/t-darwin64 (SHLIB_VERPFX): Likewise.
2288 * config/rs6000/t-darwin (SHLIB_VERPFX): Likewise.
2289 * config/rs6000/t-ppccomm (TARGET_LIBGCC2_CFLAGS, SHLIB_MAPFILES)
2290 (mklibgcc, ldblspecs): Likewise.
2291
2292 * config/i386/t-nwld (libgcc.def, libc.def, libpcre.def)
2293 (posixpre.def): Use $(T).
2294 (SHLIB_EXT, SHLIB_NAME, SHLIB_SLIBDIR_QUAL, SHLIB_DEF, SHLIB_MAP)
2295 (SHLIB_SRC, SHLIB_INSTALL): Delete.
2296 (SHLIB_LINK): Make dummy.
2297 * config/t-slibgcc-darwin: Delete contents except for dummy SHLIB_LINK.
2298
2299 * config/frv/t-linux (EXTRA_MULTILIB_PARTS): Clear.
2300
2301 * config/alpha/t-crtfm: Use $(T) in rules for EXTRA_PARTS.
2302 * config/alpha/t-vms, config/alpha/t-vms64, config/fr30/t-fr30,
2303 config/i386/t-rtems-i386, config/ia64/t-ia64, config/rs6000/t-beos,
2304 config/rs6000/t-newas, config/sparc/t-elf: Likewise.
2305
2306 * configure.ac (all_outputs): Remove mklibgcc.
2307 * configure: Regenerated.
2308
2309 2007-01-03 Josh Conner <jconner@apple.com>
2310
2311 PR middle-end/29683
2312 * calls.c (compute_argument_addresses): Set stack and stack_slot
2313 for partial args, too.
2314 (store_one_arg): Use locate.size.constant for the size when
2315 generating a save_area.
2316
2317 2007-01-03 Robert Kennedy <jimbob@google.com>
2318
2319 * tree-cfg.c (tree_merge_blocks): Release SSA_NAME phi results
2320 whose definitions are deleted due to basic block merging.
2321
2322 2007-01-03 Paul Brook <paul@codesourcery.com>
2323
2324 PR target/16634
2325 * config/arm/arm.c (output_return_instruction): Pop PC in interrupt
2326 functions.
2327 (use_return_insn): Return 0 for Thumb interrupt functions.
2328 (print_multi_reg): Add rfe argument for IRQ returns.
2329 (arm_output_epilogue): Pop interrupt return address directly into PC.
2330 (arm_expand_prologue): Only adjust IRQ return address in Arm mode.
2331
2332 2007-01-03 Paul Brook <paul@codesourcery.com>
2333
2334 Merge from sourcerygxx-4_1.
2335 * config/arm/thumb2.md: New file.
2336 * config/arm/elf.h (JUMP_TABLES_IN_TEXT_SECTION): Return True for
2337 Thumb-2.
2338 * config/arm/coff.h (JUMP_TABLES_IN_TEXT_SECTION): Ditto.
2339 * config/arm/aout.h (ASM_OUTPUT_ADDR_VEC_ELT): Add !Thumb-2 assertion.
2340 (ASM_OUTPUT_ADDR_DIFF_ELT): Output Thumb-2 jump tables.
2341 * config/arm/aof.h (ASM_OUTPUT_ADDR_DIFF_ELT): Output Thumb-2 jump
2342 tables.
2343 (ASM_OUTPUT_ADDR_VEC_ELT): Add !Thumb-2 assertion.
2344 * config/arm/ieee754-df.S: Use macros for Thumb-2/Unified asm
2345 comptibility.
2346 * config/arm/ieee754-sf.S: Ditto.
2347 * config/arm/arm.c (thumb_base_register_rtx_p): Rename...
2348 (thumb1_base_register_rtx_p): ... to this.
2349 (thumb_index_register_rtx_p): Rename...
2350 (thumb1_index_register_rtx_p): ... to this.
2351 (thumb_output_function_prologue): Rename...
2352 (thumb1_output_function_prologue): ... to this.
2353 (thumb_legitimate_address_p): Rename...
2354 (thumb1_legitimate_address_p): ... to this.
2355 (thumb_rtx_costs): Rename...
2356 (thumb1_rtx_costs): ... to this.
2357 (thumb_compute_save_reg_mask): Rename...
2358 (thumb1_compute_save_reg_mask): ... to this.
2359 (thumb_final_prescan_insn): Rename...
2360 (thumb1_final_prescan_insn): ... to this.
2361 (thumb_expand_epilogue): Rename...
2362 (thumb1_expand_epilogue): ... to this.
2363 (arm_unwind_emit_stm): Rename...
2364 (arm_unwind_emit_sequence): ... to this.
2365 (thumb2_legitimate_index_p, thumb2_legitimate_address_p,
2366 thumb1_compute_save_reg_mask, arm_dwarf_handle_frame_unspec,
2367 thumb2_index_mul_operand, output_move_vfp, arm_shift_nmem,
2368 arm_save_coproc_regs, thumb_set_frame_pointer, arm_print_condition,
2369 thumb2_final_prescan_insn, thumb2_asm_output_opcode, arm_output_shift,
2370 thumb2_output_casesi): New functions.
2371 (TARGET_DWARF_HANDLE_FRAME_UNSPEC): Define.
2372 (FL_THUMB2, FL_NOTM, FL_DIV, FL_FOR_ARCH6T2, FL_FOR_ARCH7,
2373 FL_FOR_ARCH7A, FL_FOR_ARCH7R, FL_FOR_ARCH7M, ARM_LSL_NAME,
2374 THUMB2_WORK_REGS): Define.
2375 (arm_arch_notm, arm_arch_thumb2, arm_arch_hwdiv, arm_condexec_count,
2376 arm_condexec_mask, arm_condexec_masklen)): New variables.
2377 (all_architectures): Add armv6t2, armv7, armv7a, armv7r and armv7m.
2378 (arm_override_options): Check new CPU capabilities.
2379 Set new architecture flag variables.
2380 (arm_isr_value): Handle v7m interrupt functions.
2381 (user_return_insn): Return 0 for v7m interrupt functions. Handle
2382 Thumb-2.
2383 (const_ok_for_arm): Handle Thumb-2 constants.
2384 (arm_gen_constant): Ditto. Use movw when available.
2385 (arm_function_ok_for_sibcall): Return false for v7m interrupt
2386 functions.
2387 (legitimize_pic_address, arm_call_tls_get_addr): Handle Thumb-2.
2388 (thumb_find_work_register, arm_load_pic_register,
2389 legitimize_tls_address, arm_address_cost, load_multiple_sequence,
2390 emit_ldm_seq, emit_stm_seq, arm_select_cc_mode, get_jump_table_size,
2391 print_multi_reg, output_mov_long_double_fpa_from_arm,
2392 output_mov_long_double_arm_from_fpa, output_mov_double_fpa_from_arm,
2393 output_mov_double_fpa_from_arm, output_move_double,
2394 arm_compute_save_reg_mask, arm_compute_save_reg0_reg12_mask,
2395 output_return_instruction, arm_output_function_prologue,
2396 arm_output_epilogue, arm_get_frame_offsets, arm_regno_class,
2397 arm_output_mi_thunk, thumb_set_return_address): Ditto.
2398 (arm_expand_prologue): Handle Thumb-2. Use arm_save_coproc_regs.
2399 (arm_coproc_mem_operand): Allow POST_INC/PRE_DEC.
2400 (arithmetic_instr, shift_op): Use arm_shift_nmem.
2401 (arm_print_operand): Use arm_print_condition. Handle '(', ')', '.',
2402 '!' and 'L'.
2403 (arm_final_prescan_insn): Use extract_constrain_insn_cached.
2404 (thumb_expand_prologue): Use thumb_set_frame_pointer.
2405 (arm_file_start): Output directive for unified syntax.
2406 (arm_unwind_emit_set): Handle stack alignment instruction.
2407 * config/arm/lib1funcs.asm: Remove default for __ARM_ARCH__.
2408 Add v6t2, v7, v7a, v7r and v7m.
2409 (RETLDM): Add Thumb-2 code.
2410 (do_it, shift1, do_push, do_pop, COND, THUMB_SYNTAX): New macros.
2411 * config/arm/arm.h (TARGET_CPU_CPP_BUILTINS): Define __thumb2__.
2412 (TARGET_THUMB1, TARGET_32BIT, TARGET_THUMB2, TARGET_DSP_MULTIPLY,
2413 TARGET_INT_SIMD, TARGET_UNIFIED_ASM, ARM_FT_STACKALIGN, IS_STACKALIGN,
2414 THUMB2_TRAMPOLINE_TEMPLATE, TRAMPOLINE_ADJUST_ADDRESS,
2415 ASM_OUTPUT_OPCODE, THUMB2_GO_IF_LEGITIMATE_ADDRESS,
2416 THUMB2_LEGITIMIZE_ADDRESS, CASE_VECTOR_PC_RELATIVE,
2417 CASE_VECTOR_SHORTEN_MODE, ADDR_VEC_ALIGN, ASM_OUTPUT_CASE_END,
2418 ADJUST_INSN_LENGTH): Define.
2419 (TARGET_REALLY_IWMMXT, TARGET_IWMMXT_ABI, CONDITIONAL_REGISTER_USAGE,
2420 STATIC_CHAIN_REGNUM, HARD_REGNO_NREGS, INDEX_REG_CLASS,
2421 BASE_REG_CLASS, MODE_BASE_REG_CLASS, SMALL_REGISTER_CLASSES,
2422 PREFERRED_RELOAD_CLASS, SECONDARY_OUTPUT_RELOAD_CLASS,
2423 SECONDARY_INPUT_RELOAD_CLASS, LIBCALL_VALUE, FUNCTION_VALUE_REGNO_P,
2424 TRAMPOLINE_SIZE, INITIALIZE_TRAMPOLINE, HAVE_PRE_INCREMENT,
2425 HAVE_POST_DECREMENT, HAVE_PRE_DECREMENT, HAVE_PRE_MODIFY_DISP,
2426 HAVE_POST_MODIFY_DISP, HAVE_PRE_MODIFY_REG, HAVE_POST_MODIFY_REG,
2427 REGNO_MODE_OK_FOR_BASE_P, LEGITIMATE_CONSTANT_P,
2428 REG_MODE_OK_FOR_BASE_P, REG_OK_FOR_INDEX_P, GO_IF_LEGITIMATE_ADDRESS,
2429 LEGITIMIZE_ADDRESS, THUMB2_LEGITIMIZE_ADDRESS,
2430 GO_IF_MODE_DEPENDENT_ADDRESS, MEMORY_MOVE_COST, BRANCH_COST,
2431 ASM_APP_OFF, ASM_OUTPUT_CASE_LABEL, ARM_DECLARE_FUNCTION_NAME,
2432 FINAL_PRESCAN_INSN, PRINT_OPERAND_PUNCT_VALID_P,
2433 PRINT_OPERAND_ADDRESS): Adjust for Thumb-2.
2434 (arm_arch_notm, arm_arch_thumb2, arm_arch_hwdiv): New declarations.
2435 * config/arm/arm-cores.def: Add arm1156t2-s, cortex-a8, cortex-r4 and
2436 cortex-m3.
2437 * config/arm/arm-tune.md: Regenerate.
2438 * config/arm/arm-protos.h: Update prototypes.
2439 * config/arm/vfp.md: Enable patterns for Thumb-2.
2440 (arm_movsi_vfp): Add movw alternative. Use output_move_vfp.
2441 (arm_movdi_vfp, movsf_vfp, movdf_vfp): Use output_move_vfp.
2442 (thumb2_movsi_vfp, thumb2_movdi_vfp, thumb2_movsf_vfp,
2443 thumb2_movdf_vfp, thumb2_movsfcc_vfp, thumb2_movdfcc_vfp): New.
2444 * config/arm/libunwind.S: Add Thumb-2 code.
2445 * config/arm/constraints.md: Update include Thumb-2.
2446 * config/arm/ieee754-sf.S: Add Thumb-2/Unified asm support.
2447 * config/arm/ieee754-df.S: Ditto.
2448 * config/arm/bpabi.S: Ditto.
2449 * config/arm/t-arm (MD_INCLUDES): Add thumb2.md.
2450 * config/arm/predicates.md (low_register_operand,
2451 low_reg_or_int_operand, thumb_16bit_operator): New.
2452 (thumb_cmp_operand, thumb_cmpneg_operand): Rename...
2453 (thumb1_cmp_operand, thumb1_cmpneg_operand): ... to this.
2454 * config/arm/t-arm-elf: Add armv7 multilib.
2455 * config/arm/arm.md: Update patterns for Thumb-2 and Unified asm.
2456 Include thumb2.md.
2457 (UNSPEC_STACK_ALIGN, ce_count): New.
2458 (arm_incscc, arm_decscc, arm_umaxsi3, arm_uminsi3,
2459 arm_zero_extendsidi2, arm_zero_extendqidi2): New
2460 insns/expanders.
2461 * config/arm/fpa.md: Update patterns for Thumb-2 and Unified asm.
2462 (thumb2_movsf_fpa, thumb2_movdf_fpa, thumb2_movxf_fpa,
2463 thumb2_movsfcc_fpa, thumb2_movdfcc_fpa): New insns.
2464 * config/arm/cirrus.md: Update patterns for Thumb-2 and Unified asm.
2465 (cirrus_thumb2_movdi, cirrus_thumb2_movsi_insn,
2466 thumb2_cirrus_movsf_hard_insn, thumb2_cirrus_movdf_hard_insn): New
2467 insns.
2468 * doc/extend.texi: Document ARMv7-M interrupt functions.
2469 * doc/invoke.texi: Document Thumb-2 new cores+architectures.
2470
2471 2007-01-03 Jakub Jelinek <jakub@redhat.com>
2472
2473 * unwind-dw2.c (SIGNAL_FRAME_BIT, EXTENDED_CONTEXT_BIT): Define.
2474 (struct _Unwind_Context): Rename args_size to flags, remove
2475 signal_frame field, add a new args_size field and version field.
2476 (_Unwind_IsSignalFrame, _Unwind_SetSignalFrame,
2477 _Unwind_IsExtendedContext): New inline functions.
2478 (_Unwind_GetGR, _Unwind_SetGR, _Unwind_GetGRPtr, _Unwind_SetGRPtr):
2479 Assume by_value array is only present if _Unwind_IsExtendedContext.
2480 (_Unwind_GetIPInfo, execute_cfa_program, uw_frame_state_for): Use
2481 _Unwind_IsSignalFrame.
2482 (__frame_state_for): Initialize context.flags to EXTENDED_CONTEXT_BIT.
2483 (uw_update_context_1): Use _Unwind_SetSignalFrame.
2484 (uw_init_context_1): Initialize context->flags to
2485 EXTENDED_CONTEXT_BIT.
2486 * config/rs6000/linux-unwind.h (frob_update_context): Use
2487 _Unwind_SetSignalFrame.
2488
2489 2007-01-03 Andrew Pinski <andrew_pinski@playstation.sony.com>
2490
2491 PR middle-end/30353
2492 * gimplify.c (gimplify_modify_expr_complex_part): Move below
2493 tree_to_gimple_tuple. Call tree_to_gimple_tuple when we need
2494 the value.
2495
2496 2007-01-03 Kazu Hirata <kazu@codesourcery.com>
2497
2498 * config/i386/i386.h (NON_STACK_REG_P, REGNO_OK_FOR_SIREG_P,
2499 REGNO_OK_FOR_DIREG_P, REWRITE_ADDRESS, ASM_OPERAND_LETTER,
2500 RET, AT_SP): Remove.
2501 * config/i386/i386.md (*sse_prologue_save_insn): Use return
2502 instead of RET.
2503
2504 * alias.c (init_alias_analysis): Use VEC_safe_grow_cleared.
2505 * cfgbuild.c (find_basic_blocks): Likewise.
2506 * cfgrtl.c (rtl_create_basic_block): Likewise.
2507 * function.c (temp_slots_at_level): Likewise.
2508 * reg-stack.c (stack_regs_mentioned): Likewise.
2509 * regclass.c (allocate_reg_info): Likewise.
2510 * tree-cfg.c (init_empty_tree_cfg, build_tree_cfg, create_bb,
2511 set_bb_for_stmt, move_block_to_fn): Likewise.
2512 * tree-complex.c (tree_lower_complex): Likewise.
2513 * vec.h (VEC_safe_grow_cleared): New.
2514
2515 * cgraphunit.c, tree-ssa-alias.c: Fix comment typos.
2516
2517 2007-01-03 Zdenek Dvorak <dvorakz@suse.cz>
2518
2519 * loop-unswitch.c (unswitch_loop): Pass probabilities to loopify.
2520 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Pass probabilities
2521 to loop_version.
2522 * cfgloopmanip.c (scale_loop_frequencies): Export.
2523 (loopify): Scale the frequencies by prescribed coefficients.
2524 (set_zero_probability): New function.
2525 (duplicate_loop_to_header_edge): Improve updating of frequencies.
2526 (lv_adjust_loop_entry_edge, loop_version): Set probabilities
2527 and frequencies according to arguments.
2528 * tree-ssa-loop-manip.c (tree_unroll_loop): Set probabilities
2529 correctly.
2530 * cfg.c (scale_bbs_frequencies_int): Allow scaling the frequencies up.
2531 * modulo-sched.c (sms_schedule): Set probabilities for entering
2532 versioned loop correctly.
2533 * tree-vect-transform.c (vect_transform_loop): Ditto.
2534 * cfgloop.h (loopify, loop_version): Declaration changed.
2535 (scale_loop_frequencies): Declared.
2536
2537 2007-01-02 Jan Hubicka <jh@suse.cz>
2538
2539 * cgraph.c: Include tree-flow.h
2540 (cgraph_add_new-function): Handle IPA_SSA mode; execute
2541 early_local_passes.
2542 * cgraph.h (enum cgraph_state): Add CGRAPH_STATE_IPA_SSA.
2543 * tree-pass.h (pass_all_early_optimizations): Declare.
2544 * cgraphunit.c (cgraph_process_new_functions): Add IPA_SSA; execute
2545 early_local_passes.
2546 (cgraph_analyze_function): Do early_local_passes.
2547 * tree-mudflap.c (mf_decl_cache_locals, mf_build_check_statement_for):
2548 Do not add referenced vars.
2549 * tree-optimize.c (gate_all_optimizations): Do not execute when not in
2550 SSA form.
2551 (gate_all_early_local_passes): New gate.
2552 (pass_early_local_passes): Use new gate.
2553 (execute_early_local_optimizations): New functions.
2554 (gate_all_early_optimizations): New gate.
2555 (pass_all_early_optimizations): New pass.
2556 (execute_free_datastructures): Free SSA only when initialized.
2557 (gate_init_datastructures): Init only when optimizing.
2558 (tree_lowering_passes): Do early local passes when called late.
2559 * tree-profile.c (do_tree_profiling): Don't profile functions added
2560 late.
2561 (do_early_tree_profiling, pass_early_tree_profile): Kill.
2562 * tree-cfg.c (update_modified_stmts): Do not update when operands are
2563 not active.
2564 * passes.c (init_optimizations_passes): Reorder so we go into SSA
2565 during early_local_passes.
2566 * Makefile.in (cgraph.o): Add dependency on tree-flow.h.
2567
2568
2569 2007-01-02 Carlos O'Donell <carlos@codesourcery.com>
2570
2571 * Makefile.in: Update copyright year.
2572
2573 2007-01-02 Carlos O'Donell <carlos@codesourcery.com>
2574
2575 * Makefile.in: Export GCC_EXEC_PREFIX before calling $(RUNTEST)
2576 in $(lang_checks) and check-consistency targets.
2577
2578 2007-01-02 Jan Hubicka <jh@suse.cz>
2579
2580 * tree-mudflap.c (mf_decl_cache_locals, mf_build_check_statement_for):
2581 Do not add referenced vars.
2582 * tree-cfg.c (update_modified_stmts): Do not update when SSA operands
2583 are not active.
2584 * passes.c (init_optimization_passes): Put mudflap_2 after
2585 free_datastructures.
2586
2587 2007-01-02 Jan Hubicka <jh@suse.cz>
2588
2589 * tree-optimize (execute_fixup_cfg): Set after_inlining flag.
2590 Set NOTHROW flag on call statements proved to be nothrow.
2591 Update statement of local calls so new pure/const functions are
2592 updated. Update_ssa when in ssa form. Mark PHI nodes of nonlocal
2593 goto receivers.
2594 (tree_rest_of_compilation): Register hooks and initialize bitmap
2595 early. Do not set after_inlining flag.
2596
2597 2007-01-02 Steve Ellcey <sje@cup.hp.com>
2598
2599 * sbitmap.c (HOST_BITS_PER_LONG_LONG): Change to
2600 HOST_BITS_PER_LONGLONG
2601
2602 2007-01-02 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
2603
2604 PR c/19977
2605 * c-typeck.c (store_init_value): Don't emit pedantic overflow
2606 warning for non-static initializers.
2607
2608 2007-01-02 Steven Bosscher <steven@gcc.gnu.org>
2609
2610 * config/alpha/alpha.md, arm/arm.c, darwin.c, frv/frv.md,
2611 m32r/m32r.c, m32r/m32r.c, mn10300/mn10300.md, pa/pa.c,
2612 rs6000/rs6000.c, s390/s390.md, sh/sh.md, sparc/sparc.c:
2613 Always use set_unique_reg_note to add REG_EQUAL notes.
2614
2615 2007-01-02 Kazu Hirata <kazu@codesourcery.com>
2616
2617 Revert:
2618 2007-01-02 Kazu Hirata <kazu@codesourcery.com>
2619
2620 * alias.c (init_alias_analysis): Use VEC_safe_grow_cleared.
2621 * cfgbuild.c (find_basic_blocks): Likewise.
2622 * cfgrtl.c (rtl_create_basic_block): Likewise.
2623 * function.c (temp_slots_at_level): Likewise.
2624 * reg-stack.c (stack_regs_mentioned): Likewise.
2625 * regclass.c (allocate_reg_info): Likewise.
2626 * tree-cfg.c (init_empty_tree_cfg, build_tree_cfg, create_bb,
2627 set_bb_for_stmt, move_block_to_fn): Likewise.
2628 * tree-complex.c (tree_lower_complex): Likewise.
2629 * vec.h (VEC_safe_grow_cleared): New.
2630
2631 2007-01-02 Ian Lance Taylor <iant@google.com>
2632
2633 * c-common.c (c_common_truthvalue_conversion): When warning about
2634 using an assignment as a truth value, set TREE_NO_WARNING.
2635
2636 2007-01-02 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
2637
2638 PR middle-end/7651
2639 * c.opt (Wold-style-declaration): New.
2640 * doc/invoke.texi (C-only Warning Options): New.
2641 (Wold-style-declaration): Document it.
2642 (Wextra): Enabled by -Wextra.
2643 * c-opts.c (c_common_post_options): Enabled by -Wextra.
2644 * c-decl.c (declspecs_add_scspec): Replace -Wextra with
2645 -Wold-style-declaration.
2646
2647 2007-01-02 Kazu Hirata <kazu@codesourcery.com>
2648
2649 * alias.c (init_alias_analysis): Use VEC_safe_grow_cleared.
2650 * cfgbuild.c (find_basic_blocks): Likewise.
2651 * cfgrtl.c (rtl_create_basic_block): Likewise.
2652 * function.c (temp_slots_at_level): Likewise.
2653 * reg-stack.c (stack_regs_mentioned): Likewise.
2654 * regclass.c (allocate_reg_info): Likewise.
2655 * tree-cfg.c (init_empty_tree_cfg, build_tree_cfg, create_bb,
2656 set_bb_for_stmt, move_block_to_fn): Likewise.
2657 * tree-complex.c (tree_lower_complex): Likewise.
2658 * vec.h (VEC_safe_grow_cleared): New.
2659
2660 2007-01-02 Douglas Gregor <doug.gregor@gmail.com>
2661
2662 * c-common.c(c_common_nodes_and_builtins): Since variants of
2663 void_type_node get built before it is given a name, we need to
2664 give those variants the name, too.
2665 (complete_array_type): We need to work with the canonical main
2666 type of the array, from which we will build the qualified version.
2667 * params.def (PARAM_VERIFY_CANONICAL_TYPES): New.
2668 * print-tree.c (print_node): Display canonical type information
2669 for each type.
2670 * stor-layout.c (layout_type): When we don't know the
2671 alignment of a type for which we're building an array, we end up
2672 guessing wrong, so make the type require structural equality.
2673 * tree.c (make_node_stat): When we build a new type, it is its
2674 own canonical type.
2675 (build_type_attribute_qual_variant): When building an attribute
2676 variant, its canonical type is the non-attribute variant. However,
2677 if the attributes are target-dependent and they differ, we need to
2678 use structural equality checks for this type.
2679 (build_qualified_type): A qualified type is not equivalent to its
2680 unqualified variant; set the canonical type appropriately.
2681 (build_distinct_type_copy): When building a distinct type from
2682 another type, the new type is its own canonical type.
2683 (build_variant_type_copy): When building a new type variant, we
2684 assume that it is equivalent to the original type.
2685 (build_pointer_type_for_mode): When building a pointer type, also
2686 build a canonical type pointer.
2687 (build_reference_type_for_mode): When building a reference type,
2688 also build a canonical type reference.
2689 (build_index_type): When we can't hash an index type (e.g.,
2690 because its maximum value is negative), the index type requires
2691 structural equality tests.
2692 (build_array_type): Build the canonical form of an array type.
2693 (build_function_type): Function types require structural equality,
2694 because they contain default arguments, attributes, etc.
2695 (build_method_type_directly): Ditto for method types.
2696 (build_offset_type): Build the canonical offset type.
2697 (build_complex_type): Build the canonical vector type.
2698 (make_vector_type): Build the canonical vector type.
2699 * tree.h (TYPE_CANONICAL): New.
2700 (TYPE_STRUCTURAL_EQUALITY_P): New.
2701 (SET_TYPE_STRUCTURAL_EQUALITY): New.
2702 (struct tree_type): Added "canonical" field.
2703 * params.h (VERIFY_CANONICAL_TYPES): New.
2704 * doc/c-tree.texi (TYPE_CANONICAL): Document.
2705 (TYPE_STRUCTURAL_EQUALITY_P): Document.
2706 (SET_TYPE_STRUCTURAL_EQUALITY): Document.
2707 * doc/invoke.texi (verify-canonical-types): Document --param
2708 parameter for verifying canonical types.
2709
2710 2007-01-02 Joseph Myers <joseph@codesourcery.com>
2711
2712 * config.gcc (powerpc-*-eabispe*, powerpc-*-eabisimaltivec*,
2713 powerpc-*-eabisim*, powerpc-*-eabialtivec*, powerpc-*-eabi*,
2714 powerpc-*-rtems*, powerpc-wrs-vxworks, powerpc-wrs-vxworksae,
2715 powerpcle-*-eabisim*, powerpcle-*-eabi*): Add rs6000/e500.h to
2716 tm_file.
2717 * config/rs6000/e500.h: New.
2718 * config/rs6000/eabi.h (TARGET_SPE_ABI, TARGET_SPE, TARGET_E500,
2719 TARGET_ISEL, TARGET_FPRS, TARGET_E500_SINGLE, TARGET_E500_DOUBLE):
2720 Remove.
2721 * config/rs6000/linuxspe.h (TARGET_SPE_ABI, TARGET_SPE,
2722 TARGET_E500, TARGET_ISEL, TARGET_FPRS, TARGET_E500_SINGLE,
2723 TARGET_E500_DOUBLE): Remove.
2724 * config/rs6000/vxworks.h (TARGET_SPE_ABI, TARGET_SPE,
2725 TARGET_E500, TARGET_ISEL, TARGET_FPRS): Remove.
2726 * config/rs6000/rs6000.h (CHECK_E500_OPTIONS): Define.
2727 * config/rs6000/rs6000.c (rs6000_override_options): Use
2728 CHECK_E500_OPTIONS.
2729
2730 2007-01-02 Joseph Myers <joseph@codesourcery.com>
2731
2732 * config/rs6000/rs6000.c (print_operand): Check (TARGET_SPE ||
2733 TARGET_E500_DOUBLE), not TARGET_E500, for %y.
2734 (rs6000_generate_compare, rs6000_emit_sCOND, output_cbranch,
2735 rs6000_emit_cmove): Don't check TARGET_E500.
2736 * config/rs6000/rs6000.md (bunordered, bordered, sunordered,
2737 sordered): Don't check TARGET_E500.
2738
2739 2007-01-01 Eric Christopher <echristo@apple.com>
2740
2741 * config/mips/mips.c (mips_regno_mode_ok_for_base_p): Use
2742 HARD_REGISTER_NUM_P.
2743
2744 2007-01-01 Roger Sayle <roger@eyesopen.com>
2745
2746 * fold-const.c (fold_binary) <EQ_EXPR>: Fold "(X^C1) eq/ne C2" into
2747 "X eq/ne (C1^C2)". Fold "(X^Z) eq/ne (Y^Z)" as "X eq/ne Y" when Z
2748 has no side-effects. Fold "(X^C1) eq/ne (Y^C2)" as "(X^(C1^C2))
2749 eq/ne Y".
2750
2751 2007-01-01 Mike Stump <mrs@apple.com>
2752
2753 * configure.ac: Remove support for building with Apple's gcc-3.1.
2754
2755 2007-01-02 Joseph Myers <joseph@codesourcery.com>
2756
2757 PR middle-end/30311
2758 * caller-save.c (add_stored_regs): Only handle SUBREGs if inner
2759 REG is a hard register. Do not modify REG before calling
2760 subreg_nregs.
2761 * rtlanal.c (subreg_get_info): Don't assert size of XMODE is a
2762 multiple of the size of YMODE for certain lowpart cases.
2763
2764 2007-01-01 Andrew Pinski <pinskia@gmail.com>
2765
2766 PR middle-end/30253
2767 * gimplify (voidify_wrapper_expr): Update for
2768 GIMPLE_MODIFY_STMT.
2769
2770 2007-01-01 Andreas Schwab <schwab@suse.de>
2771
2772 PR target/29166
2773 * config/ia64/ia64.c (ia64_compute_frame_size): Account space for
2774 save of BR0 in extra_spill_size instead of spill_size.
2775 (ia64_expand_prologue): Save BR0 outside of the gr/br/fr spill
2776 area.
2777 (ia64_expand_epilogue): Restore BR0 from its new location.
2778
2779 2007-01-01 Andrew Pinski <pinskia@gmail.com>
2780
2781 * gimplify.c (gimplify_init_constructor <case VECTOR_TYPE>):
2782 Use a temporary variable if the left hand side is not a gimple
2783 register.
2784
2785 2007-01-01 Andrew Pinski <pinskia@gmail.com>
2786
2787 * gimplify.c (gimplify_return_expr): Make the temporary variable
2788 for the return expression, a gimple register variable.
2789
2790 2007-01-01 Jan Hubicka <jh@suse.cz>
2791
2792 * emit-rtl.c (emit_copy_of_insn_after): Do not call copy_insn_1 for
2793 INSN_LIST.
2794
2795 2007-01-01 Mike Stump <mrs@apple.com>
2796
2797 * configure.ac (HAVE_GAS_LITERAL16): Add autoconf check for
2798 .literal16.
2799 * config/darwin.c (machopic_select_rtx_section): Use
2800 HAVE_GAS_LITERAL16.
2801 (darwin_mergeable_constant_section): Likewise.
2802 * configure: Regenerate.
2803 * config.in: Regenerate.
2804
2805 2007-01-01 Jan Hubicka <jh@suse.cz>
2806 Andrew Pinski <pinskia@gmail.com>
2807
2808 * cgraphunit.c (cgraph_optimize): Call cgraph_add_new_functions
2809 before starting IPA passes.