1 2007-01-18 Nathan Sidwell <nathan@codesourcery.com>
3 * config/m68k/fpgnulib.c (__truncdfsf2): Implement round to
4 nearest even, fix denormal rounding overflow.
6 2007-01-18 Richard Sandiford <richard@codesourcery.com>
8 * config/m68k/m68k.md (movsf_cf_hard): Use fsmove instead of
10 (movdf_cf_hard): Use fdmove for cases 0 and 3 and fmove for case 1.
11 (extendsfdf2_cf): Use fdmove instead of f%&move.
12 (truncdfsf2_cf): Use fsmove instead of f%$smove.
13 (add<mode>3_cf, sub<mode>3_cf): Use <FP:prec> instead of <FP:round>.
15 2007-01-18 Richard Sandiford <richard@codesourcery.com>
17 * config/m68k/m68k.md (movdf_cf_hard): Use output_move_double for
20 2007-01-18 Richard Sandiford <richard@codesourcery.com>
22 * real.h (real_format): Add a canonical_nan_lsbs_set field.
23 (coldfire_single_format): Declare.
24 (coldfire_double_format): Likewise.
25 * real.c (encode_ieee_single): Use canonical_nan_lsbs_set instead
26 of qnan_msb_set to determine the lower bits of a canonical
28 (encode_ieee_double): Likewise.
29 (encode_ieee_quad): Likewise.
30 (ieee_single_format): Initialize canonical_nan_lsbs_set.
31 (mips_single_format): Likewise.
32 (ieee_double_format): Likewise.
33 (mips_double_format): Likewise.
34 (ieee_extended_motorola_format): Likewise.
35 (ieee_extended_intel_96_format): Likewise.
36 (ieee_extended_intel_128_format): Likewise.
37 (ieee_extended_intel_96_round_53_format): Likewise.
38 (ibm_extended_format): Likewise.
39 (mips_extended_format): Likewise.
40 (ieee_quad_format): Likewise.
41 (mips_quad_format): Likewise.
42 (vax_f_format): Likewise.
43 (vax_d_format): Likewise.
44 (vax_g_format): Likewise.
45 (i370_single_format): Likewise.
46 (i370_double_format): Likewise.
47 (decimal_single_format): Likewise.
48 (decimal_double_format): Likewise.
49 (decimal_quad_format): Likewise.
50 (c4x_single_format): Likewise.
51 (c4x_extended_format): Likewise.
52 (real_internal_format): Likewise.
53 (coldfire_single_format): New real_format.
54 (coldfire_double_format): Likewise.
55 * config/pdp11/pdp11.c (pdp11_f_format): Initialize
56 canonical_nan_lsbs_set.
57 (pdp11_d_format): Likewise.
58 * config/m68k/m68k.c (override_options): Override REAL_FORMAT_MODE
59 if TARGET_COLDFIRE_CPU.
61 2007-01-18 Richard Sandiford <richard@codesourcery.com>
63 * config/m68k/m68k-protos.h (m68k_output_pic_call): Delete.
64 (output_call, m68k_legitimize_call_address): Declare.
65 * config/m68k/m68k.h (EXTRA_CONSTRAINT): Remove unnecessary
66 parenthesees. Add support for a 'W' constraint.
67 (LEGITIMATE_PIC_OPERAND_P): Remove SYMBOL_REF_FLAG handling.
68 (PRINT_OPERAND_PUNCT_VALID_P): Remove comment about 'o'.
69 (m68k_symbolic_call, m68k_symbolic_jump): Declare.
70 * config/m68k/m68k.c (m68k_symbolic_call, m68k_symbolic_jump): New
72 (override_options): Initialize them. Do not set flag_no_function_cse
73 for TARGET_ID_SHARED_LIBRARY.
74 (m68k_output_pic_call): Delete.
75 (m68k_legitimize_call_address): New function.
76 (print_operand): Remove the %o prefix. Handle the %p prefix.
77 (output_call): New function.
78 (m68k_output_mi_thunk): Use m68k_symbolic_jump. Always load the
79 target address from the GOT if symbolic jumps are not allowed.
80 * config/m68k/m68k.md (call, general_operand): Do not set
81 SYMBOL_REF_FLAG. Use m68k_legitimize_call_address instead.
82 Merge separate flag_pic and !flag_pic define_insns into...
83 (*call, *call_value): ...these new patterns. Match the address
84 rather than the containing MEM and require it to be a call_operand.
85 Use output_call to generate the asm template.
86 * config/m68k/predicates.md (const_call_operand): New predicate.
87 (call_operand): Likewise.
89 2007-01-18 Nathan Sidwell <nathan@codesourcery.com>
91 * config/m68k/m68k.h (REGISTER_MOVE_COST): Simplify definition.
92 (STACK_GROWS_DOWNWARD): Define to 1.
93 (FUNCTION_VALUE, LIBCALL_VALUE, FUNCTION_VALUE_REGNO_P): Equivocate
94 comments, emphasizing that these values are only defaults.
95 * config/m68k/linux.h (LINK_SPEC): Fix formatting in #undef.
96 * config/m68k/m68k.c (const_method): Remove trailing whitespace.
98 2007-01-18 Richard Sandiford <richard@codesourcery.com>
100 * config/m68k/m68k.md (cmpsi): Remove outdated flag_pic handling.
102 2007-01-18 Kazu Hirata <kazu@codesourcery.com>
103 Richard Sandiford <richard@codesourcery.com>
105 * config/m68k/m68k.h (DATA_REGNO_P, ADDRESS_REGNO_P, INT_REGNO_P)
106 (FP_REGNO_P): New macros.
107 (REGNO_OK_FOR_INDEX_P, REGNO_OK_FOR_BASE_P, REGNO_OK_FOR_DATA_P)
108 (REGNO_OK_FOR_FP_P, REG_OK_FOR_INDEX_P, REG_OK_FOR_BASE_P): Use them.
109 (EH_RETURN_STACKADJ_RTX): Use A0_REG.
110 * config/m68k/m68k.c (m68k_regno_mode_ok): Use the new REGNO macros.
112 2007-01-18 Nathan Sidwell <nathan@codesourcery.com>
114 * config.gcc (m68k-*-aout*, m68k-*-coff*, m68020-*-elf*, m68k-*-elf*)
115 (m68k-*-uclinuxoldabi, m68k-*-uclinux*, m68k-*-rtems*): Add t-floatlib
117 * config/m68k/t-floatlib: New file, extracting common code from...
118 * config/m68k/t-m68kbare, config/m68k/t-m68kelf,
119 * config/m68k/t-uclinux: Here.
120 * config/m68k/fpgnulib.c: Do not compile extendeed precision
121 routines on ColdFire targets.
123 2007-01-18 Nathan Sidwell <nathan@codesourcery.com>
125 * config.gcc (m68k-*-aout*, m68k-*-coff*, m68020-*-elf*, m68k-*-elf*)
126 (m68010-*-netbsdelf*, m68k*-*-netbsdelf*, m68k*-*-openbsd*)
127 (m68k-*-uclinuxoldabi, m68k-*-uclinux*, m68k-*-linux*)
128 (m68k-*-rtems*): Use tm_file rather than m68k/m68k.h and
129 explicitly set MOTOROLA to 1.
130 * config/m68k/m68k.h (MOTOROLA): Simplify definition accordingly.
132 2007-01-18 Richard Sandiford <richard@codesourcery.com>
133 Nathan Sidwell <nathan@codesourcery.com>
135 * config/m68k/m68k.h (PCC_STATIC_STRUCT_RETURN): Delete.
136 (ASM_OUTPUT_REG_PUSH, ASM_OUTPUT_REG_POP): Add MOTOROLA cases.
137 * config/m68k/coff.h (REGISTER_PREFIX_MD): Delete.
138 * config/m68k/m68020-elf.h (LIB_SPEC): Delete.
139 * config/m68k/m68k-none.h (CC1_SPEC, CPP_SUBTARGET_SPEC): Delete.
140 * config/m68k/m68kelf.h (IMMEDIATE_PREFIX, REGISTER_PREFIX_MD)
141 (ASM_OUTPUT_REG_PUSH, ASM_OUTPUT_REG_POP): Delete.
142 (NO_DOLLAR_IN_LABEL, PCC_STATIC_STRUCT_RETURN): Don't undefine.
143 (BSS_ASM_OP, ASM_OUTPUT_SKIP, ASM_OUTPUT_ASCII): Delete.
144 * config/m68k/m68kemb.h (PCC_STATIC_STRUCT_RETURN): Don't undefine.
145 * config/m68k/linux.h (SIZE_TYPE, PTRDIFF_TYPE, WCHAR_TYPE)
146 (WCHAR_TYPE_SIZE, TARGET_OBJFMT_CPP_BUILTINS): Delete.
147 (TARGET_OS_CPP_BUILTINS): Don't define mc68000 and mc68020 here.
148 (DBX_REGISTER_NUMBER): Delete.
149 * config/m68k/netbsd-elf.h (IMMEDIATE_PREFIX): Delete.
150 (PCC_STATIC_STRUCT_RETURN): Don't undefine.
151 * config/m68k/openbsd.h (PCC_STATIC_STRUCT_RETURN): Define.
153 2007-01-18 Richard Sandiford <richard@codesourcery.com>
155 * config.gcc (m68k-*-uclinux*): Add flat.h to $tm_file.
156 * config/flat.h: New file.
157 * crtstuff.c (USE_PT_GNU_EH_FRAME): Don't define if
159 * config/m68k/m68k.h (ASM_PREFERRED_EH_DATA_FORMAT): Do not use
160 indirect references for -msep-data or -mid-shared-library.
161 Do not use PC-relative code addresses either.
162 * config/m68k/m68k.c (override_options): Restrict -fPIC error
164 * config/m68k/uclinux.h (STARTFILE_SPEC): Define. Use Scrt1.o
165 for shared libraries and crt1.o for executables. Use crti.o and
167 (ENDFILE_SPEC): Use crtend.o and crtn.o.
168 (LIB_SPEC): Suppress -Rlibc.gdb if -static-libc is given.
169 Do not add -elf2flt or -shared-lib-id options here.
170 (LINK_SPEC): Define. Pass -elf2flt if no -elf2flt option is given.
171 Pass -shared-lib-id if -mid-shared-library, taking the library
172 identifier from -mshared-library-id if given, otherwise
174 (EH_FRAME_IN_DATA_SECTION): Do not undefine.
175 (INIT_SECTION_ASM_OP, FINI_SECTION_ASM_OP): Likewise.
176 (TARGET_OS_CPP_BUILTINS): Define __GXX_MERGED_TYPEINFO_NAMES=0
177 and __GXX_TYPEINFO_EQUALITY_INLINE=0 if -mid-shared-library.
178 (DRIVER_SELF_SPECS): Map unadorned PIC options to -msep-data.
179 * config/m68k/t-uclinux (EXTRA_MULTILIB_PARTS): Add crtbegin.o
181 * config/m68k/lb1sf68.asm (PICCALL): Use an lea and pc-relative
182 jump sequence for ISA A and ISA A+.
185 2007-01-18 Richard Sandiford <richard@codesourcery.com>
187 * config.gcc (m68k-*-uclinux*): Base the port on the common
188 and m68k GNU/Linux files rather than on the generic ELF ones.
189 * config/m68k/uclinux.h (TARGET_VERSION): Override.
190 (TARGET_OS_CPP_BUILTINS): Use LINUX_TARGET_OS_CPP_BUILTINS.
192 2007-01-18 Julian Brown <julian@codesourcery.com>
193 Richard Sandiford <richard@codesourcery.com>
195 * config/m68k/m68k.h (LONG_DOUBLE_TYPE_SIZE): Make 64-bit on ColdFire.
196 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
197 * config/m68k/netbsd-elf.h (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Undefine
199 * config/m68k/uclinux-oldabi.h (LONG_DOUBLE_TYPE_SIZE): Redefine to
201 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
203 2007-01-18 Richard Sandiford <richard@codesourcery.com>
205 * doc/install.texi: Document m68k-uclinuxoldabi.
206 * config.gcc (m68k-*-uclinuxoldabi): New configuration.
207 * config/m68k/uclinux-oldabi.h: New file, copied from
208 config/m68k/uclinux.h.
210 2007-01-18 Kaz Kojima <kkojima@rr.iij4u.or.jp>
212 * config/m32r/m32r-protos.h (m32r_expand_epilogue): Declare it.
213 * config/m32r/m32r.c (m32r_setup_incoming_varargs): Use gen_frame_mem.
214 (m32r_compute_frame_size): Use unsigned for regno.
215 (m32r_reload_lr): Use gen_frame_mem.
217 (m32r_output_function_epilogue): Don't output the function epilogue
219 (m32r_expand_epilogue): New.
220 (direct_return): Return false if the function has the interrupt
222 (m32r_hard_regno_rename_ok): Remove code for the textual epilogue.
223 * config/m32r/m32r.md (epilogue): New expander.
224 (return_lr, return_rte): New insns.
225 (return): Make it expander.
226 (return_normal): New expander.
228 2007-01-18 Josh Conner <jconner@apple.com>
231 * config/rs6000/rs6000.c (rs6000_emit_vector_compare): Add
232 support for UNLE, UNLT, UNGE, and UNGT.
234 2007-01-18 Jan Hubicka <jh@suse.cz>
236 * tree-vrp.c (finalize_jump_threads): Do not call cleanup_cfg by hand.
237 * tree-tailcall (add_virtual_phis): Likewise.
238 (optimize_tail_call): Return TODOs.
239 (execute_tail_calls): Return TODOs.
240 * tree-ssa-ccp (execute_fold_all_builtins): Do cleanup_cfg via TODO.
241 * tree-cfgcleanup.c (cleanup_tree_cfg_loop): Return if something
243 * tree-ssa-forwprop.c (tree_ssa_forward_propagate_single_use_value):
244 Cleanup cfg using TODO.
245 * tree-flow.h (cleanup_tree_cfg_loop): Update prototype.
246 * passes.c (execute_function_todo): When cleanup did something, remove
248 * tree-cfg.c (pass_build_cfg): Add cleanup_cfg TODO.
249 (make_edges): Don't cleanup_cfg.
251 2007-01-18 Uros Bizjak <ubizjak@gmail.com>
253 * reg-stack.c (subst_stack_regs_pat) [UNSPEC_SINCOS_COS,
254 UNSPEC_XTRACT_FRACT]: Use generic code for instructions that
255 operate on the top of stack.
256 [UNSPEC_SINCOS_SIN, UNSPEC_XTRACT_EXP, UNSPEC_TAN]: Rewrite
257 register handling of instructions that output to the second
259 [UNSPEC_TAN_ONE, UNSPEC_TAN_TAN]: Remove.
260 (move_for_stack_reg): Special-case check for dead destination
261 stack slot for constant load of 1.0 inside UNSPEC_TAN.
263 * config/i386/i386.md (UNSPEC_TAN): New constant.
264 (UNSPEC_TAN_ONE, UNSPEC_TAN_TAN): Remove.
265 (fptanxf4_i387, fptan_extend<mode>xf4_i387): New patterns
266 to correctly model move of constant 1.0 to top stack slot.
267 (*tandf3_1, *tansf3_1, *tanxf3_1): Remove insn patterns.
268 (unnamed peephole2 pattern): Remove corresponding peephole2
269 pattern that optimizes tan insn and loading of constant 1.0.
270 (tanxf2): Use fptanxf4_i387.
271 (tan<mode>2): Rename from tansf2 and tandf2 and macroize insn
272 patterns using X87MODEF12 mode macro. Use fptan_extend<mode>xf4_i387
273 and truncate result to requested mode. Use SSE_FLOAT_MODE_P to
274 disable patterns for SSE math.
275 (sincos<mode>3): Use truncxf<mode>2_i387_noop for truncation.
276 (fyl2x_extend<mode>xf3_i387): Use X87MODEF12 for operand 1.
278 2007-01-18 Dirk Mueller <dmueller@suse.de>
279 Richard Guenther <rguenther@suse.de>
282 * doc/invoke.texi (Warray-bounds): Document -Warray-bounds.
283 * common.opt (Warray-bounds): Add new warning option.
284 * c-opts.c (c_common_handle_option): Define -Warray-bounds
286 * Makefile.in: make tree-vrp.o depend on toplev.h
287 * tree-vrp.c (vrp_finalize): Call check_array_refs if -Warray-bounds
289 (check_array_refs, check_array_bounds, check_array_ref): New.
291 2007-01-18 Jan Hubicka <jh@suse.cz>
293 * tree-ssa-ccp.c (ccp_finalize): Return if something changed.
294 (execute_ssa_ccp): Return flags conditionally.
295 * tree-ssa-propagate.c (substitue_and_fold): Return if something was
297 * tree-ssa-propagate.h (substitute_and_fold): Update prototype.
299 2007-01-18 Steven Bosscher <steven@gcc.gnu.org>
301 * cfgcleanup.c (cleanup_cfg): Detect cfglayout mode and set
302 the CLEANUP_CFGLAYOUT flag when in cfglayout mode.
304 * Makefile.c (GTFILES): Add cfglayout.h.
305 * gengtype.c (open_base_files): Likewise.
306 * cfglayout.c (cfg_layout_function_footer,
307 cfg_layout_function_header) Reindent to make gengtype happy.
308 * cfglayout.h (cfg_layout_function_footer,
309 cfg_layout_function_header): Add GTY(()) marker.
311 * ifcvt.c (noce_try_sign_mask): Make sure INSN_B is non-null.
313 2007-01-18 Ben Elliston <bje@au.ibm.com>
315 * genautomata.c (write_automata): Include xstrerror output in the
316 error message if writing the DFA description file fails.
318 2007-01-17 H.J. Lu <hongjiu.lu@intel.com>
320 * config/mips/mips-protos.h (mips_output_external): Make it
322 * config/mips/iris.h (TARGET_ASM_EXTERNAL_LIBCALL): Removed.
323 * config/mips/mips.c (irix_output_external_libcall): Likewise.
324 (extern_list): Likewise.
325 (extern_head): Likewise.
326 (TARGET_ASM_FILE_END): Likewise.
327 (mips_file_end): Likewise.
328 (mips_output_external): Rewritten.
330 2007-01-18 Ben Elliston <bje@au.ibm.com>
332 * genpreds.c (write_insn_preds_c): Only write out the function
333 body for regclass_for_constraint if we have register constraints.
335 2007-01-17 Tom Tromey <tromey@redhat.com>
337 * doc/sourcebuild.texi (libgcj Tests): Use sourceware.org.
338 * doc/install.texi (Testing): Use sourceware.org.
339 (Binaries): Likewise.
340 (Specific): Likewise.
341 * doc/contrib.texi (Contributors): Use sourceware.org.
343 2007-01-17 Anatoly Sokolov <aesok@post.ru>
345 * config/avr/avr.h (AVR_HAVE_LPMX): New macro.
346 (AVR_ENHANCED): Rename to ...
347 (AVR_HAVE_MUL): ... new.
348 (avr_enhanced_p): Rename to ...
349 (avr_have_mul_p): ... new.
350 (TARGET_CPU_CPP_BUILTINS): Use 'avr_have_mul_p' instead of
351 'avr_enhanced_p' for "__AVR_ENHANCED__". Define "__AVR_HAVE_MUL__".
352 * config/avr/avr.c (avr_enhanced_p): Rename to ...
353 (avr_have_mul_p): ... new.
354 (base_arch_s): Rename 'enhanced' to 'have_mul'.
355 (avr_override_options): Use 'avr_have_mul_p' and 'have_mul' instead of
356 'avr_enhanced_p' and 'enhanced'.
357 (ashlhi3_out, ashrhi3_out, lshrhi3_out, avr_rtx_costs): Use
358 AVR_HAVE_MUL instead of AVR_ENHANCED.
359 * avr.md (*tablejump_enh): Use AVR_HAVE_LPMX instead of AVR_ENHANCED.
360 (mulqi3, *mulqi3_enh, *mulqi3_call, mulqihi3, umulqihi3, mulhi3,
361 *mulhi3_enh, *mulhi3_call, mulsi3, *mulsi3_call): Use AVR_HAVE_MUL
362 instead of AVR_ENHANCED.
363 (*tablejump_enh): Use AVR_HAVE_LPMX instead of AVR_ENHANCED.
364 * libgcc.S: Use __AVR_HAVE_MUL__ instead of __AVR_ENHANCED__.
365 (__tablejump__): Use __AVR_HAVE_LPMX__ instead of __AVR_ENHANCED__.
367 2007-01-17 Ian Lance Taylor <iant@google.com>
369 * vec.h (VEC_reserve_exact): Define.
370 (vec_gc_p_reserve_exact): Declare.
371 (vec_gc_o_reserve_exact): Declare.
372 (vec_heap_p_reserve_exact): Declare.
373 (vec_heap_o_reserve_exact): Declare.
374 (VEC_OP (T,A,reserve_exact)): New static inline function, three
376 (VEC_OP (T,A,reserve)) [all versions]: Remove handling of
378 (VEC_OP (T,A,alloc)) [all versions]: Call ...reserve_exact.
379 (VEC_OP (T,A,copy)) [all versions]: Likewise.
380 (VEC_OP (T,a,safe_grow)) [all versions]: Likewise.
381 * vec.c (calculate_allocation): Add exact parameter. Change all
383 (vec_gc_o_reserve_1): New static function, from vec_gc_o_reserve.
384 (vec_gc_p_reserve, vec_gc_o_reserve): Call vec_gc_o_reserve_1.
385 (vec_gc_p_reserve_exact, vec_gc_o_reserve_exact): New functions.
386 (vec_heap_o_reserve_1): New static function, from vec_heap_o_reserve.
387 (vec_heap_p_reserve, vec_heap_o_reserve): Call vec_heap_o_reserve_1.
388 (vec_heap_p_reserve_exact): New function.
389 (vec_heap_o_reserve_exact): New function.
391 2007-01-17 Jan Hubicka <jh@suse.cz>
393 * ipa-type-escape.c (look_for_casts): Revamp using handled_component_p.
395 2007-01-17 Eric Christopher <echristo@apple.com>
397 * config.gcc: Support core2 processor.
399 2007-01-16 Jan Hubicka <jh@suse.cz>
401 * tree-ssanames.c (release_dead_ssa_names): Instead of ggc_freeing
402 the names, just unlink the chain so we don't crash on dangling pointers
405 2007-01-16 Jan Hubicka <jh@suse.cz>
407 * cgraph.h (cgraph_decide_inlining_incrementally): Kill.
408 * tree-pass.h: Reorder to make IPA passes appear toegher.
409 (pass_early_inline, pass_inline_parameters, pass_apply_inline): Declare.
410 * cgraphunit.c (cgraph_finalize_function): Do not compute inling
411 parameters, do not call early inliner.
412 * ipa-inline.c: Update comments. Include tree-flow.h
413 (cgraph_decide_inlining): Do not compute inlining parameters.
414 (cgraph_decide_inlining_incrementally): Return TODOs; assume to
415 be called with function context set up.
416 (pass_ipa_inline): Remove unreachable functions before pass.
417 (cgraph_early_inlining): Simplify assuming to be called from the
419 (pass_early_inline): New pass.
420 (cgraph_gate_ipa_early_inlining): New gate.
421 (pass_ipa_early_inline): Turn into simple wrapper.
422 (compute_inline_parameters): New function.
423 (gate_inline_passes): New gate.
424 (pass_inline_parameters): New pass.
425 (apply_inline): Move here from tree-optimize.c
426 (pass_apply_inline): New pass.
427 * ipa.c (cgraph_remove_unreachable_nodes): Verify cgraph after
429 * tree-inline.c (optimize_inline_calls): Return TODOs rather than
431 (tree_function_versioning): Do not allocate dummy struct function.
432 * tree-inline.h (optimize_inline_calls): Update prototype.
433 * tree-optimize.c (execute_fixup_cfg): Export.
434 (pass_fixup_cfg): Remove
435 (tree_rest_of_compilation): Do not apply inlines.
436 * tree-flow.h (execute_fixup_cfg): Declare.
437 * Makefile.in (gt-passes.c): New.
438 * passes.c: Include gt-passes.h
439 (init_optimization_passes): New passes.
440 (nnodes, order): New static vars.
441 (do_per_function_toporder): New function.
442 (execute_one_pass): Dump current pass here.
443 (execute_ipa_pass_list): Don't dump current pass here.
445 2007-01-16 Janis Johnson <janis187@us.ibm.com>
447 * config/dfp-bit.c (dfp_compare_op): Return separate value for NaN.
448 (DFP_NE, DFP_LE, DFP_GE): Return false for NaN.
450 2007-01-16 David Edelsohn <edelsohn@gnu.org>
452 * config/rs6000/darwin-ldouble.c: Build file for SOFT_FLOAT.
453 (strong_alias): Define.
454 (__gcc_qmul): Provide non-FMA for soft-float.
470 * config/rs6000/rs6000.c (rs6000_init_libfuncs): Initialize
471 soft-float functions.
472 * config/rs6000/libgcc-ppc-glibc.ver: Version soft-float symbols.
473 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Do not warn
474 about long double soft float.
476 2007-01-16 Dorit Nuzman <dorit@il.ibm.com>
477 Tehila Meyzels <tehila@il.ibm.com>
479 * tree-vectorizer.h (is_pattern_stmt_p): New.
480 * tree-vect-analyze.c (vect_determine_vectorization_factor): Fix
481 formatting (tabs instead of spaces). Cleanup and clarify setting
482 of STMT_VINFO_VECTYPE. Call is_pattern_stmt_p.
483 * tree-vect-transform.c (vect_get_vec_def_for_operand): Fix typo.
484 (vectorizable_type_demotion): Check that types are integral.
485 (vectorizable_type_promotion): Likewise.
486 (vectorizable_store): Fix typo. Eliminate new-line at end of
489 2007-01-16 Jan Hubicka <jh@suse.cz>
491 * tree-ssanames.c (release_dead_ssa_names): Remove invalidated
494 2007-01-15 Eric Christopher <echristo@apple.com>
496 * ifcvt.c: Include vec.h, vecprim.h.
497 (check_cond_move_block): New argument regs.
498 Reorganize. Add registers used to regs.
499 (cond_move_process_if_block): Use regs set above as
502 2007-01-15 Eric Christopher <echristo@apple.com>
504 * config/darwin.h: Update copyright.
505 (TARGET_OPTION_TRANSLATE_TABLE): Add umbrella.
506 (LINK_COMMAND_SPEC): Add -u.
507 (LINK_SPEC): Fix umbrella for above.
509 2007-01-15 Joseph S. Myers <joseph@codesourcery.com>
511 * config/soft-fp/op-common.h, config/soft-fp/op-4.h: Update from
514 2007-01-15 Tom Tromey <tromey@redhat.com>
516 * doc/sourcebuild.texi (libgcj Tests): Don't mention jacks.
517 * doc/install.texi (Testing): Don't mention jacks.
518 (Configuration): Document --enable-java-maintainer-mode. Move
519 --with-java-home to libgcj-specific section. Document
521 (Prerequisites): Mention --enable-java-maintainer-mode, ecj1.
523 2007-01-15 Jan Hubicka <jh@suse.cz>
525 * tree-ssa-dce.c (DCE_TODOs): New.
526 (propagate_necessity): Return if something changed.
527 (eliminate_unnecessary_stmts): Likewise.
528 (perform_tree_ssa_dce): Return TODO flags when needed.
529 (pass_dce, pass_dce_loop, pass_cd_dce): Remove TODO flags.
531 2007-01-15 Uros Bizjak <ubizjak@gmail.com>
533 * config/i386/i386.md (fyl2xxf3_i387): Rename from fyl2x_xf3.
534 (fyl2x_extend<mode>xf3_i387): New insn pattern.
535 (log<mode>2): Rename from logsf2 and logdf2 and macroize insn
536 patterns using X87MODEF12 mode macro. Extend operand 1
537 to XFmode. Use SSE_FLOAT_MODE_P to disable patterns for SSE math.
538 (log10<mode>2): Ditto.
539 (log2<mode>2): Ditto.
540 (log1p<mode>2): Ditto.
541 (logb<mode>2): Ditto.
542 (fyl2xp1xf3_i387): Rename from fyl2xp1_xf3.
543 (fyl2xp1_extend<mode>xf3_i387): New insn pattern.
544 (*fxtractxf3_i387): Rename from *fxtractxf3.
545 (fxtract_extend<mode>xf3_i387): New insn pattern.
546 (ilogbsi2): Use match_dup 3, not match_operand:XF 3.
547 * config/i386/i386.c (ix86_emit_i387_log1p): Use gen_fyl2xp1xf3_i387()
548 and gen_fyl2xxf3_i387().
550 2007-01-14 Zdenek Dvorak <dvorakz@suse.cz>
552 * loop-unswitch.c (unswitch_loop): Do not call fix_loop_placement.
553 * cfgloopmanip.c (fix_loop_placement): Made static. Use
554 get_loop_exit_edges. Changed return type to bool.
555 * cfgloop.h (fix_loop_placement): Declaration removed.
557 2007-01-14 Dorit Nuzman <dorit@il.ibm.com>
559 * param.h (MIN_VECT_LOOP_BOUND): New.
560 * params.def (MIN_VECT_LOOP_BOUND): New.
561 * tree-vectorizer.c (slpeel_tree_peel_loop_to_edge): Takes another
562 argument - minimum threshold for number of iterations.
563 * tree-vectorizer.h (slpeel_tree_peel_loop_to_edge): Add another
564 argument to declaration.
565 * tree-vect-analyze.c (vect_analyze_operations): Check value of
567 * tree-vect-transform.c (vect_do_peeling_for_loop_bound): Call
568 slpeel_tree_peel_loop_to_edge with additional argument.
569 (vect_do_peeling_for_alignment): Likewise.
570 * doc/invoke.texi (min-vect-loop-bound): Document new param option.
572 2007-01-14 Uros Bizjak <ubizjak@gmail.com>
575 * config/i386/i386.c (print_operand) ['z']: Output 'b' for
578 2007-01-14 Jan Hubicka <jh@suse.cz>
580 * tree-dfa.c (remove_referenced_var): New function.
581 * tree-ssa-live.c (remove_unused_locals): Walk referenced vars and
582 prune referenced vars list too.
583 * tree-flow.h (remove_referenced_var): Declare.
585 2007-01-14 Jan Hubicka <jh@suse.cz>
587 * tree-eh.c (add_stmt_to_eh_region_fn): Do not add call_exprs
589 (remove_stmt_from_eh_region_fn): Do not remove call_exprs.
590 (verify_eh_throw_stmt_node, verify_eh_throw_table_statements): Kill.
591 * except.h (verify_eh_throw_table_statements): Kill prototype.
592 * cfgexpand.c (expand_gimple_basic_block): Propagate Eh regions
593 into call exrepssions.
594 * tree-optimize.c (execute_free_cfg_annotatiosn): Do not call
596 * tree-cfg.c: Include pointer-set.h.
597 (verify_node_sharing): Work on pointer set.
598 (verify_eh_throw_stmt_node): New.
599 (verify_stmts): Use pointers sets, verify throw_stmt.
601 2007-01-13 Zdenek Dvorak <dvorakz@suse.cz>
603 * ipa-reference.c (analyze_function): Consider also addresses taken
606 2007-01-12 Roger Sayle <roger@eyesopen.com>
608 * c-typeck.c (null_pointer_constant_p): Replace use of
609 TREE_CONSTANT_OVERFLOW with TREE_OVERFLOW.
610 (build_c_cast): Likewise.
612 2007-01-12 Roger Sayle <roger@eyesopen.com>
614 * tree.h (force_fit_type_double): Remove unused final argument.
615 * c-common.c (constant_expression_warning): Replace use of
616 TREE_CONSTANT_OVERFLOW with TREE_OVERFLOW.
617 (convert_and_check): Likewise.
618 (shorten_compare): Update call to force_fit_type_double.
619 (c_common_truthvalue_conversion) <INTEGER_CST>: Use integer_zerop.
620 * convert.c (convert_to_pointer): Update call to
621 force_fit_type_double.
622 * fold-const.c (force_fit_type_double): Remove overflowed_const
624 (int_const_binop, fold_convert_const_int_from_int,
625 fold_convert_const_int_from_real, fold_div_compare,
626 fold_sign_changed_comparison, fold_unary, fold_negate_const,
627 fold_abs_const, fold_not_const): Remove the final argument from
628 calls to force_fit_type_double.
630 2007-01-12 Andrew Pinski <andrew_pinski@playstation.sony.com>
632 * configure.ac: Set insn to "nop" for spu-*-* also.
633 * configure: Regenerate.
635 2007-01-12 Olga Golovanevsky <olga@il.ibm.com>
637 * builtins.def : Add BUILT_IN_FREE.
639 2007-01-12 Jan Hubicka <jh@suse.cz>
641 PR tree-optimization/30443
642 * tree-inline.c (tree_function_versioning): Do not optimize when
643 cloning for inlining.
645 2007-01-12 Zdenek Dvorak <dvorakz@suse.cz>
647 * doc/loop.texi: Document recording of loop exits.
648 * cfgloopmanip.c (loopify, duplicate_loop): Use alloc_loop.
649 (update_single_exits_after_duplication,
650 update_single_exit_for_duplicated_loop,
651 update_single_exit_for_duplicated_loops): Removed.
652 (duplicate_loop_to_header_edge): Do not call
653 update_single_exits_after_duplication and
654 update_single_exit_for_duplicated_loops.
655 (loop_version): Do not update single_exit information.
656 (fix_loop_structure): Use record_loop_exits instead of
657 mark_single_exit_loops.
658 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update
659 the lists of loop exits.
660 * cfghooks.c (redirect_edge_and_branch, redirect_edge_and_branch_force,
661 split_edge, merge_blocks): Update the lists of loop exits.
662 * modulo-sched.c (sms_schedule): Pass LOOPS_HAVE_RECORDED_EXITS to
664 * loop-init.c (loop_optimizer_init): Call record_loop_exits instead
665 of mark_single_exit_loops.
666 (loop_optimizer_finalize): Call release_recorded_exits.
667 * tree-ssa-loop.c (tree_loop_optimizer_init): Pass
668 LOOPS_HAVE_RECORDED_EXITS to loop_optimizer_init.
669 * tree-vectorizer.c (slpeel_tree_duplicate_loop_to_edge_cfg): Do not
670 update single exit information.
671 * lambda-code.c (perfect_nestify): Ditto.
672 * cfgloop.c (flow_loop_free): Destroy the list of exits of the loop.
673 (mark_single_exit_loops): Removed.
674 (alloc_loop, loop_exit_hash, loop_exit_eq, loop_exit_free,
675 get_exit_descriptions, rescan_loop_exit, record_loop_exits,
676 dump_recorded_exit, dump_recorded_exits, release_recorded_exits): New
678 (get_loop_exit_edges, single_exit): Use recorded exit lists.
679 (add_bb_to_loop, remove_bb_from_loops): Update the lists of loop exits.
680 (verify_loop_structure): Verify consistency of the exit lists.
681 (flow_loops_find): Use alloc_loop. Initialize exits hash.
682 (set_single_exit): Removed.
683 * cfgloop.h (struct loop_exit): New function.
684 (struct loop): single_exit_ field replaced by exits field.
685 (LOOPS_HAVE_MARKED_SINGLE_EXITS): Replaced by LOOPS_HAVE_RECORDED_EXITS.
686 (struct loops): Added exits hash.
687 (mark_single_exit_loops, set_single_exit): Declaration removed.
688 (release_recorded_exits, record_loop_exits, rescan_loop_exit): Declare.
690 2007-01-12 Richard Sandiford <richard@codesourcery.com>
692 * doc/invoke.texi: Avoid use of @headitem.
694 2007-01-12 Richard Sandiford <richard@codesourcery.com>
696 * cse.c (cse_insn): Move HAVE_CC0 code after declarations.
698 2007-01-12 Richard Sandiford <richard@codesourcery.com>
700 * doc/install.texi: Fix m68k-*-* anchor and add m68k-*-* to the
703 2007-01-12 Nathan Sidwell <nathan@codesourcery.com>
704 Richard Sandiford <richard@codesourcery.com>
706 * doc/invoke.texi: Document -mno-bitfield, -mno-rtd and -mno-short.
707 * config/m68k/m68k.opt: Resort options.
708 (mbitfield, mrtd, mshort): Remove RejectNegative properties.
710 2007-01-12 Nathan Sidwell <nathan@codesourcery.com>
711 Richard Sandiford <richard@codesourcery.com>
713 * doc/invoke.texi: Document the macros that are defined by
714 m68k's -mtune and -mhard-float options.
715 * config/m68k/m68k-protos.h (m68k_cpp_cpu_ident) Declare.
716 (m68k_cpp_cpu_family): Likewise.
717 * config/m68k/m68k.h (TARGET_CPU_CPP_BUILTINS): Add a full set
718 of __ucfv*__ macros. Define __mcffpu__ if generating code for
719 ColdFire FPUs. Define __mcf_cpu_* and __mcf_family_* macros.
720 * config/m68k/m68k.c (m68k_cpp_cpu_ident): New function.
721 (m68k_cpp_cpu_family): Likewise.
723 2007-01-12 Richard Sandiford <richard@codesourcery.com>
725 * config/m68k/m68k.h (TARGET_CPU_CPP_BUILTINS): Treat all mc68*
726 macros besides mc68000 as tuning macros. Use a switch statement
727 to set them and mcpu32.
729 2007-01-12 Julian Brown <julian@codesourcery.com>
731 * config/m68k/m68k.h: Use TARGET_68040 instead of TARGET_68040_ONLY.
732 (TARGET_68040_ONLY): Rename to...
733 (TARGET_68040): ...this.
734 * config/m68k/m68k.c: Use TARGET_68040 instead of TARGET_68040_ONLY.
735 * config/m68k/m68k.md: Likewise.
737 2007-01-12 Julian Brown <julian@codesourcery.com>
738 Nathan Sidwell <nathan@codesourcery.com>
739 Richard Sandiford <richard@codesourcery.com>
741 * config.gcc (m680[012]0-*-*, m68k*-*-*): Set m68k_cpu_ident to
742 the -mcpu= argument associated with the --with-cpu setting.
743 Define M68K_DEFAULT_TUNE to the default -mtune= option,
744 if different from the one implied by the -mcpu setting.
745 Accept --with-cpu=FOO if FOO is listed in m68k-devices.def,
746 using mcpu=FOO as the default CPU option. Set target_cpu_default2.
747 * doc/invoke.texi: Mention ColdFire in the introduction to the
748 m68k options. Document the new -march, -mcpu, -mtune, -mdiv,
749 -mno-div and -mhard-float options. Make -m68881 a synonym for
750 -mhard-float. Document the previously-undocumented -m5206e,
751 -m528x, -m5307 and -m5407 options. Tweak the existing option
752 documentation for consistency.
753 * doc/install.texi: Mention new --with-cpu arguments.
754 * config/m68k/m68k.h (OPTION_DEFAULT_SPECS): Only use the
755 default CPU if neither -mcpu nor -march are specified.
756 (ASM_CPU_SPEC): Pass down -mcpu and -march options.
757 (TARGET_CPU_CPP_BUILTINS): Set __mcfisa*__ macros from
758 TARGET_ISA*. Set the legacy __mcf*__ cpu macros in the same way,
759 using m68k_tune to decide between families that implement the
760 same ISA. Use m68k_tune to set __mcfv4e__.
761 (FL_BITFIELD, FL_68881, FL_COLDFIRE, FL_CF_HWDIV, FL_CF_MAC)
762 (FL_CF_EMAC, FL_CF_EMAC_B, FL_CF_USP, FL_CF_FPU, FL_ISA_68000)
763 (FL_ISA_68010, FL_ISA_68020, FL_ISA_68040, FL_ISA_A, FL_ISA_B)
764 (FL_ISA_C, FL_ISA_MMU): New macros.
765 (MASK_COLDFIRE): Delete.
766 (TARGET_68010, TARGET_68020, TARGET_68040_ONLY, TARGET_COLDFIRE)
767 (TARGET_ISAB): Redefine in terms of m68k_cpu_flags.
768 (TARGET_68881, TARGET_COLDFIRE_FPU): Redefine in terms of m68k_fpu.
769 (TARGET_HARD_FLOAT): Do not define here.
770 (TARGET_ISAAPLUS, TARGET_ISAC): New macros.
771 (TUNE_68000): New macro.
772 (TUNE_68000_10): Redefine in terms of TUNE_68000 and TUNE_68010.
773 (TUNE_68010, TUNE_68030, TUNE_68040, TUNE_68060, TUNE_CPU32)
774 (TUNE_CFV2): Redefine in terms of m68k_tune.
775 (uarch_type, target_device, fpu_type): New enums.
776 (m68k_cpu, m68k_tune, m68k_fpu, m68k_cpu_flags): Declare.
777 * config/m68k/m68k.c (TARGET_DEFAULT): Remove MASK_68881.
778 (FL_FOR_isa_00, FL_FOR_isa_10, FL_FOR_isa_20, FL_FOR_isa_40)
779 (FL_FOR_isa_cpu32, FL_FOR_isa_a, FL_FOR_isa_aplus, FL_FOR_isa_b)
780 (FL_FOR_isa_c): New macros.
781 (m68k_isa): New enum.
782 (m68k_target_selection): New structure.
783 (all_devices, all_isas, all_microarchs): New tables.
784 (m68k_cpu_entry, m68k_arch_entry, m68k_tune_entry, m68k_cpu)
785 (m68k_tune, m68k_fpu, m68k_cpu_flags): New variables.
786 (MASK_ALL_CPU_BITS): Delete.
787 (m68k_find_selection): New function.
788 (m68k_handle_option): Handle -mcpu=, -march= and -mtune=.
789 Map the legacy target options to a combination of the new ones.
790 (override_options): Set m68k_cpu, m68k_tune, m68k_fpu and
791 m68k_cpu_flags. Handle M68K_DEFAULT_TUNE. Use m68k_cpu_flags
792 to derive default MASK_BITFIELD, MASK_CF_HWDIV and MASK_HARD_FLOAT
794 * config/m68k/m68k.opt (m5200, m5206e, m528x, m5307, m5407, mcfv4e)
795 (m68010, m68020, m68020-40, m68020-60, m68030, m68040): Remove Mask
797 (m68881, msoft-float): Change mask from 68881 to HARD_FLOAT.
798 (march=, mcpu=, mdiv, mhard-float, mtune=): New options.
799 * config/m68k/m68k-devices.def: New file.
801 2007-01-12 Richard Sandiford <richard@codesourcery.com>
802 Nathan Sidwell <nathan@codesourcery.com>
804 * config/m68k/m68k.h (ASM_CPU_SPEC, ASM_SPEC, EXTRA_SPECS)
805 (SUBTARGET_EXTRA_SPECS): New macros.
806 * config/m68k/linux.h (ASM_SPEC): Remove CPU flags;
807 use %(asm_cpu_spec) instead.
808 * config/m68k/m68k-none.h (ASM_SPEC): Likewise.
809 * config/m68k/openbsd.h (ASM_SPEC): Likewise.
810 * config/m68k/netbsd-elf.h (ASM_SPEC): Likewise.
811 (EXTRA_SPECS): Rename to...
812 (SUBTARGET_EXTRA_SPECS): ...this.
814 2007-01-12 Nathan Sidwell <nathan@codesourcery.com>
815 Richard Sandiford <richard@codesourcery.com>
816 Julian Brown <julian@codesourcery.com>
818 * config.gcc (m68k-*-aout*, m68k-*-coff*, m68020-*-elf*, m68k-*-elf*)
819 (m68k-*-uclinux*, m68k-*-linux*, m68k-*-rtems*): Set default_m68k_cpu
820 to the configuration's default CPU.
821 (m68010-*-netbsdelf*, m68k*-*-netbsdelf*, m68k*-*-openbsd*): Likewise.
822 Remove default masks.
823 (m680[012]0-*-*): Set the default with_cpu to the first part of
825 (m68k*-*-*): Set the default with_cpu to m$default_m68k_cpu.
826 (m68k*-*-linux): Extend the --with-cpu handling to...
827 (m680[012]0-*-*, m68k*-*-*): ...these configurations. Allow m68000
828 and m68010. Don't set target_cpu_default2.
829 * doc/install.texi: Document --with-cpu for m68k.
830 * config/m68k/m68k.h (OPTION_DEFAULT_SPECS): Define.
831 * config/m68k/m68k-none.h (TARGET_CPU_DEFAULT, M68K_CPU_m68k)
832 (M68K_CPU_m68000, M68K_CPU_m68010, M68K_CPU_m68020, M68K_CPU_m68030)
833 (M68K_CPU_m68040, M68K_CPU_m68302, M68K_CPU_m68332, TARGET_DEFAULT)
834 (ASM_CPU_DEFAULT_SPEC, CC1_CPU_DEFAULT_SPEC): Delete.
835 (ASM_SPEC): Remove use of %(asm_cpu_default).
836 (EXTRA_SPECS, SUBTARGET_EXTRA_SPECS, MULTILIB_DEFAULTS): Delete.
837 * config/m68k/linux.h (TARGET_DEFAULT): Delete.
838 (CPP_SPEC): Merge definitions. Do not handle __HAVE_68881__ here.
839 * config/m68k/netbsd-elf.h (TARGET_OS_CPP_BUILTINS): Define
840 __HAVE_FPU__ if TARGET_HARD_FLOAT.
841 (TARGET_DEFAULT): Delete.
842 (EXTRA_SPECS): Delete cpp_cpu_default_spec, cpp_cpu_spec,
843 cpp_fpu_spec, asm_default_spec and netbsd_cpp_spec.
844 (CPP_CPU_SPEC): Delete.
845 (TARGET_VERSION): Merge definitions, using TARGET_68010 to pick
846 the appropriate string.
847 (CPP_CPU_DEFAULT_SPEC, ASM_DEFAULT_SPEC, CPP_FPU_SPEC): Delete.
848 (CPP_SPEC): Define to NETBSD_CPP_SPEC.
849 (ASM_SPEC): Don't use %(asm_default_spec).
850 * config/m68k/m68k.c (TARGET_DEFAULT_TARGET_FLAGS): Remove
851 TARGET_DEFAULT and add MASK_68881.
852 * config/m68k/m68k.md: Remove mention of TARGET_DEFAULT from comments.
854 2007-01-12 Richard Sandiford <richard@codesourcery.com>
856 * config.gcc (m68010-*-netbsdelf*): Add MASK_68010.
857 (m68k*-*-netbsdelf*, m68k*-*-openbsd*, m68k*-linux*): Add
858 MASK_68010 alongside MASK_68020.
859 * doc/invoke.texi: Document -m68010.
860 * config/m68k/m68k.opt (m68010): New.
861 * config/m68k/m68k.h (TARGET_CPU_CPP_BUILTINS): Define mc68010
863 (TUNE_68010): New macro.
864 * config/m68k/m68k-none.h (M68K_CPU_m68k, M68K_CPU_m68010)
865 (M68K_CPU_m68020, M68K_CPU_m68030, M68K_CPU_m68040)
866 (M68K_CPU_m68332): Add MASK_68010.
867 * config/m68k/linux.h (TARGET_DEFAULT): Add MASK_68010 to
869 * config/m68k/netbsd-elf.h (CPP_CPU_SPEC): Remove now-redundant
871 * config/m68k/m68k.c (MASK_ALL_CPU_BITS): Add MASK_68010.
872 (m68k_handle_option): Handle OPT_m68010. Add MASK_68010
873 to all entries that use MASK_68020.
874 (output_move_simode_const, output_move_himode, output_move_qimode)
875 (output_move_stricthi, output_move_strictqi): Use TARGET_68010
876 instead of TARGET_68020 to select clr behavior. Remove comment
877 about there being no TARGET_68010.
878 * config/m68k/m68k.md: Likewise throughout.
880 2007-01-12 Julian Brown <julian@codesourcery.com>
882 * config/m68k/m68k.h (TARGET_ISAB): New macro.
883 * config/m68k/m68k.c: Use TARGET_ISAB rather than TARGET_CFV4.
884 * config/m68k/m68k.md: Likewise.
886 2007-01-12 Julian Brown <julian@codesourcery.com>
888 * config/m68k/m68k.h (LEGITIMATE_INDEX_P, LEGITIMIZE_ADDRESS): Use
889 TARGET_COLDFIRE_FPU instead of TARGET_CFV4E.
891 2007-01-12 Julian Brown <julian@codesourcery.com>
893 * config/m68k/m68k.h (TUNE_68040_60): New macro.
894 * config/m68k/m68k.c (standard_68881_constant_p): Use it.
895 * config/m68k/m68k.md: Likewise.
897 2007-01-12 Julian Brown <julian@codesourcery.com>
898 Richard Sandiford <richard@codesourcery.com>
900 * config/m68k/m68k.h (TARGET_CPU_CPP_BUILTINS): Use TUNE_68030
901 instead of TARGET_68030, TUNE_68040 instead of TARGET_68040,
902 TUNE_68060 instead of TARGET_68060 and TUNE_CPU32 instead of
904 (TARGET_CPU32): Rename to...
905 (TUNE_CPU32): ...this.
906 (TUNE_68000_10, TUNE_68030, TUNE_68040, TUNE_68060)
907 (TUNE_CFV2): New macros.
908 * config/m68k/netbsd-elf.h (LONG_DOUBLE_TYPE_SIZE): Simplify;
909 remove conditions that are implied by TARGET_68020.
910 * config/m68k/m68k.c (m68k_output_function_prologue): Use TUNE_68040
911 instead of TARGET_68040 and TUNE_CPU32 instead of TARGET_CPU32.
912 (m68k_output_function_epilogue): Likewise.
913 (m68k_rtx_costs): Likewise. Use TUNE_68060 instead of TARGET_68060
914 and TUNE_CFV2 instead of TARGET_5200. Use TUNE_68000_10 instead of
915 "!TARGET_68020 && !TARGET_COLDFIRE" to choose between 68000 and
916 non-68000 timings. Refactor multiplication and division costs.
917 (output_addsi3): Use TUNE_68040 instead of TARGET_68040 and
918 TUNE_CPU32 instead of TARGET_CPU32.
919 (standard_68881_constant_p): Use TUNE_68040 instead of TARGET_68040
920 and TUNE_68060 instead of TARGET_68060.
921 * config/m68k/m68k.md: Use TUNE_68040 instead of TARGET_68040,
922 TUNE_68060 instead of TARGET_68060, and TUNE_CPU32 instead of
924 (movsi_const0): Use TUNE_68000_10 rather than "!TARGET_68020
925 && !TARGET_COLDFIRE" to choose between moveq and clr.
926 Likewise in the unnamed movsf pattern.
927 (ashlsi_17_24, lshrsi_17_24): Guard with TUNE_68000_10 rather than
928 "!TARGET_68020 && !TARGET_COLDFIRE". Likewise the unnamed
931 2007-01-12 Richard Sandiford <richard@codesourcery.com>
933 * config/m68k/m68k.h (TARGET_CPU_CPP_BUILTINS): Increase amount
934 of tabbing before backslashes.
936 2007-01-11 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
938 * pa-linux.h (ASM_OUTPUT_INTERNAL_LABEL): Undefine.
939 * pa.h (ASM_OUTPUT_LABEL): Output colon when using GAS.
940 (ASM_OUTPUT_INTERNAL_LABEL): Define.
942 2007-01-11 Zdenek Dvorak <dvorakz@suse.cz>
944 * tree-ssa-loop-ivopts.c (extract_cond_operands): Split from
945 find_interesting_uses_cond.
946 (find_interesting_uses_cond): Use extract_cond_operands.
947 (rewrite_use_compare): Use extract_cond_operands and
948 force_gimple_operand_bsi. Do not call update_stmt.
949 (determine_use_iv_cost_condition): Use extract_cond_operands.
950 Return cheaper of using original bound and changing the exit bound.
952 2007-01-11 Zdenek Dvorak <dvorakz@suse.cz>
954 PR tree-optimization/29516
955 * tree-ssa-address.c (tree_mem_ref_addr, add_to_parts,
956 most_expensive_mult_to_index, addr_to_parts,
957 create_mem_ref, maybe_fold_tmr): Make the type of
958 fields of TARGET_MEM_REF sizetype.
959 (move_fixed_address_to_symbol, move_pointer_to_base):
961 * tree.def (TARGET_MEM_REF): Add comment on types of
964 2007-01-11 Joseph Myers <joseph@codesourcery.com>
966 * c-common.c (vector_types_convertible_p): Treat opaque types as
967 always convertible if they have the same size, but not otherwise.
969 2007-01-11 Steven Bosscher <steven@gcc.gnu.org>
971 * ifcvt.c (struct noce_if_info): Add comments to the fields.
972 Remove the b_unconditional field.
973 (noce_try_sign_mask): Do not look at b_unconditional.
974 (noce_process_if_block): Do not use merge_if_blocks. Update
975 the CFG here. Do not set b_unconditional.
976 (cond_move_process_if_block): Likewise.
977 (find_cond_trap): Likewise.
978 (check_cond_move_block): Require simple jump insns at the end
981 2007-01-11 Jan Hubicka <jh@suse.cz>
983 PR tree-optimization/1046
984 * tree-tailcall.c (suitable_for_tail_call_opt_p): Use TREE_ADDRESSABLE
985 when alias info is not ready.
986 (pass_tail_recursion): Do not require aliasing.
987 * tree-ssa-copyrename.c (pass_rename_ssa_cop): Likewise.
988 * tree-ssa-ccp.c (pass_ccp, pass_fold_builtins): Likewise.
989 * tree-ssa-copy.c (pass_copy_prop): Likewise.
990 * tree-ssa-forwprop.c (pass_forwprop): Likewise.
991 * tree-ssa-dce.c (pass_dce, pass_dce_loop, pass_cd_dce): Likewise.
992 * passes.c (init_optimization_passes): Execute rename_ssa_copies,
993 ccp, forwprop, copy_prop, merge_phi, copy_prop, dce and tail recursion
995 * tree-ssa-operands.c (add_virtual_operand, get_indirect_ref_operand):
996 When aliasing is not build, mark statement as volatile.
998 2007-01-11 Tom Tromey <tromey@redhat.com>
1000 PR preprocessor/15185, PR preprocessor/20989:
1001 * doc/cppopts.texi <-MT>: Update description of algorithm for
1002 computing default target.
1003 <-M, -MD>: Reword "basename" text.
1005 2007-01-11 Roger Sayle <roger@eyesopen.com>
1007 * builtins.c (expand_builtin_pow, expand_builtin_powi,
1008 fold_builtin_cabs, fold_builtin_sqrt, fold_builtin_trunc,
1009 fold_builtin_floor, fold_builtin_ceil, fold_builtin_round,
1010 fold_builtin_int_int_roundingfn, fold_builtin_bitop,
1011 fold_builtin_bswap, real_constp, fold_builtin_pow,
1012 fold_builtin_powi, fold_builtin_signbit, fold_builtin_copysign,
1013 do_mpfr_arg1, do_mpfr_arg2, do_mpfr_arg3, do_mpfr_sincos): Replace
1014 uses of the macro TREE_CONSTANT_OVERFLOW with TREE_OVERFLOW.
1015 * convert.c (convert_to_pointer): Likewise.
1016 * expr.c (highest_pow2_factor, expand_expr_real_1): Likewise.
1017 * fold-const.c (force_fit_type, fold_negate_expr, int_const_binop,
1018 const_binop, fold_convert_const_int_from_int,
1019 fold_convert_const_int_from_real,
1020 fold_convert_const_real_from_real, sign_bit_p,
1021 optimize_minmax_comparison, extract_muldiv_1, fold_div_compare,
1022 fold_sign_changed_comparison, fold_unary, fold_comparison,
1023 fold_binary, multiple_of_p, tree_Expr_non_zero_p,
1024 fold_negate_const, fold_abs_const, fold_not_const): Likewise.
1025 * print-tree.c (print_node_brief, print_node): Likewise.
1026 * stor-layout.c (place_field, layout_type): Likewise.
1027 * tree-chrec.c (keep_cast): Likewise.
1028 * tree.c (build_vector, build_real, build_real_from_int_cst,
1029 build_complex): Likewise.
1031 2007-01-11 Roger Sayle <roger@eyesopen.com>
1033 * tree.h (TREE_CONSTANT_OVERFLOW): Obsolete. For the time being,
1034 treat TREE_CONSTANT_OVERFLOW as a synonym of TREE_OVERFLOW.
1036 2007-01-11 Paolo Bonzini <bonzini@gnu.org>
1038 * configure.ac (strict1_warn): Rename to strict_warn.
1039 (WERROR, --enable-werror, symlink hacks, stage1_cflags,
1040 cc_set_by_configure, quoted_cc_set_by_configure,
1041 stage_prefix_set_by_configure, quoted_stage_prefix_set_by_configure,
1042 all_boot_languages, all_stagestuff): Remove.
1043 (target_list): Remove bootstrap targets.
1044 * Makefile.in (quickstrap): Unconditionally make a synonym of all.
1045 (BOOT_LANGUAGES, STAGE1_CFLAGS, STAGE1_CHECKING,
1046 REMAKEFLAGS, FLAGS_TO_PASS, PREPEND_DOTDOT_TO_RELATIVE_PATHS,
1047 SUBDIR_FLAGS_TO_PASS, WERROR_FLAGS, STRICT2_WARN, LANG_STAGESTUFF,
1048 VOL_FILES, POSTSTAGE1_FLAGS_TO_PASS, STAGE2_FLAGS_TO_PASS,
1049 STAGEPROFILE_FLAGS_TO_PASS, STAGEFEEDBACK_FLAGS_TO_PASS, stage1_build,
1050 stage1_copy, stage2_build, stage2_copy, stageprofile_build,
1051 stageprofile_copy, stage3_build, stage3_copy, stagefeedback_build,
1052 stagefeedback_copy, stage4_build, clean_s1, clean_sw, bootstrap,
1053 bootstrap-lean, bootstrap2, bootstrap2-lean, bootstrap3,
1054 bootstrap3-lean, bootstrap4, bootstrap4-lean, unstage1, unstage2,
1055 unstage3, unstage4, unstageprofile, unstagefeedback, restage, restage2,
1056 restage3, restage4, restageprofile, restagefeedback, bubbleestrap,
1057 cleanstrap, unstrap, restrap, *compare, *compare3, *compare4,
1058 *compare-lean, *compare3-lean, *compare4-lean, stage1-start, stage1,
1059 stage2-start, stage2, stage3-start, stage3, stage4-start, stage4,
1060 stageprofile-start, stageprofile, stagefeedback-start, stagefeedback,
1061 risky-stage1, risky-stage2, risky-stage3, risky-stage4): Remove.
1062 (ORDINARY_FLAGS_TO_PASS): Rename to FLAGS_TO_PASS.
1063 (STAGECOPYSTUFF, STAGEMOVESTUFF): Consolidate into MOSTLYCLEANFILES.
1064 (mostlyclean): Adjust.
1065 (clean, distclean): Don't mention bootstrap stuff.
1066 * configure: Regenerate.
1067 * ada/config-lang.in, cp/config-lang.in, forttran/config-lang.in,
1068 java/config-lang.in, objc/config-lang.in, objcp/config-lang.in,
1069 treelang/config-lang.in (stagestuff): Remove.
1070 * doc/sourcebuild.texi (stage1, stage2, stage3, stage4,
1071 stageprofile, stagefeedback, stagestuff): Remove mention.
1073 2007-01-11 Nick Clifton <nickc@redhat.com>
1075 * config/mcore/predicates.md (mcore_general_movesrc_operand):
1077 (mcore_general_movdst_operand): Do not accept CONST_INTs.
1078 (mcore_arith_K_S_operand): Run the test for the S constraint not
1079 the test for the M constraint.
1080 (mcore_addsub_operand): Do not accept integer values that are
1081 larger than 32 bits.
1082 * config/mcore/mcore.md: Remove unused constraints from split.
1083 (andsi3): Use HOST_WIDE_INT instead of int to hold an INTVAL.
1085 (allocate_stack): Likewise.
1086 * config/mcore/mcore.c (mcore_print_operand): Restrict output of P
1087 operands to 32 bits.
1088 (mcore_const_costs): Use HOST_WIDE_INT instead of int to hold an
1090 (mcore_and_cost, mcore_modify_comparison, const_ok_for_mcore,
1091 mcore_const_ok_for_inline, mcore_const_trick_uses_not,
1092 try_constant_tricks, mcore_num_ones, mcore_num_zeros,
1093 mcore_output_bclri, mcore_output_andn, output_inline_const,
1094 mcore_output_move, mcore_output_movedouble): Likewise.
1095 (mcore_output_cmov): Use CONST_OK_FOR_M and CONST_OK_FOR_N.
1096 (output_inline_const): Likewise.
1097 (output_inline_const): Fix format strings used in sprintf
1099 * config/mcore/mcore-protos.h: Update prototypes for changed
1100 functions in mcore.c.
1101 * config/mcore/mcore.h (CONST_OK_FOR_I): Cast values to
1102 HOST_WIDE_INT and not int.
1103 (CONST_OK_FOR_J, CONST_OK_FOR_K, CONST_OK_FOR_L, CONST_OK_FOR_M,
1104 CONST_OK_FOR_N): Likewise.
1105 (LEGITIMATE_CONSTANT_P): Also check CONSTANT_P.
1106 (GO_IF_LEGITIMATE_INDEX): Use HOST_WIDE_INT instead of int to hold
1109 2007-01-10 Jan Hubicka <jh@suse.cz>
1111 * tree-vrp.c (remove_range_assertions): Release defs.
1112 * tree-ssa-loop-ivopts.c (rmeove_statement): Likewise.
1113 * tree-ssa-dom.c (remove_stmt_or_phi): Likewise.
1115 2007-01-10 Paul Brook <paul@codesourcery.com>
1117 * config/arm/arm.c (arm_rtx_costs_1): Handle mutiply-subtract.
1118 * config/arm/arm.md (mulsi3subsi): New insn.
1120 2007-01-10 Zdenek Dvorak <dvorakz@suse.cz>
1122 * tree-ssa-loop-manip.c (tree_unroll_loop): Make it a wrapper over ...
1123 (tree_transform_and_unroll_loop): New.
1124 * tree-flow.h (transform_callback, tree_transform_and_unroll_loop):
1127 2007-01-10 Robert Kennedy <jimbob@google.com>
1129 * fold-const.c (fold_comparison): Fold comparisons like (x *
1130 1000 < 0) to (x < 0).
1132 2007-01-10 Ian Lance Taylor <iant@google.com>
1134 * tree-pretty-print.c (dump_generic_node): Print parentheses when
1135 operands have the same priority.
1137 2007-01-10 Tom Tromey <tromey@redhat.com>
1139 * fold-const.c (fold_truthop): Don't check can_use_bit_fields_p.
1140 (fold_binary): Likewise.
1141 * langhooks.c (lhd_can_use_bit_fields_p): Removed.
1142 * langhooks-def.h (lhd_can_use_bit_fields_p): Removed.
1143 (LANG_HOOKS_CAN_USE_BIT_FIELDS_P): Removed.
1144 (LANG_HOOKS_INITIALIZER): Remove LANG_HOOKS_CAN_USE_BIT_FIELDS_P.
1145 * langhooks.h (struct lang_hooks): Removed field
1146 'can_use_bit_fields_p'.
1148 2007-01-10 Ralf Corsépius <ralf.corsepius@rtems.org>
1150 * config/bfin/t-bfin, config/bfin/t-bfin-elf: Remove GCC_CFLAGS.
1152 2007-01-10 Razya Ladelsky <razya@il.ibm.com>
1154 * function.c (get_last_funcdef_no): New function.
1155 * function.h (get_last_funcdef_no): Declare.
1156 * tree-inline.c (initialize_cfun): Add initialization.
1157 (tree_function_versioning): Cleanup.
1159 2007-01-10 Jan Hubicka <jh@suse.cz>
1161 * tree-inline.c (setup_one_parameter): Do not propagate into abnormal
1164 2007-01-10 Sa Liu <saliu@de.ibm.com>
1165 Ben Elliston <bje@au.ibm.com>
1167 * spu.h (STACK_SAVE_AREA): Use VOIDmode for SAVE_FUNCTION, SImode
1168 for SAVE_NONLOCAL and Pmode for any other save level.
1169 * spu-protos.h (spu_restore_stack_block): Declare.
1170 * spu.md (save_stack_block): Remove.
1171 (restore_stack_block): Call spu_restore_stack_block.
1172 * spu.c (spu_restore_stack_block): New function.
1173 (spu_expand_epilogue): Remove old comment.
1175 2007-01-09 Zdenek Dvorak <dvorakz@suse.cz>
1177 PR tree-optimization/30322
1178 * tree-ssa-loop-ivopts.c (fold_affine_expr, iv_value): Removed.
1179 (cand_value_at): Return the value as aff_tree.
1180 (may_eliminate_iv): Convert the bound from aff_tree to tree.
1181 * tree-affine.c (aff_combination_add_cst, aff_combination_add_product,
1182 aff_combination_mult): New functions.
1183 (aff_combination_add): Use aff_combination_add_cst.
1184 (aff_combination_convert): Allow conversions to a wider type.
1185 (tree_to_aff_combination): Handle BIT_NOT_EXPR.
1186 * tree-affine.h (aff_combination_mult): Declare.
1188 2007-01-09 Carlos O'Donell <carlos@codesourcery.com>
1190 * doc/tm.texi: Update documentation to reflect reality of exec
1191 and start file search behaviours. Update copyright year.
1192 * doc/invoke.texi: Explain how GCC_EXEC_PREFIX is used to find
1193 header file directories.
1195 2007-01-09 Uros Bizjak <ubizjak@gmail.com>
1197 * config/i386/i386.md (*sinxf2): Rename to *sinxf2_i387.
1198 (*cosxf2): Rename to cosxf2_i387.
1199 (*sindf2, *sinsf2): Extend operand 1 to XFmode. Macroize patterns
1200 using X87MODEF12 mode macro. Rename patterns to
1201 *sin_extend<mode>xf2_i387. Use SSE_FLOAT_MODE_P to disable patterns
1203 (*cosdf2, *cossf2): Ditto.
1204 (sincosdf3, sincossf3): Ditto. Rewrite corresponding splitters
1205 to match extended input operands.
1206 (sincos<mode>3): New expander.
1207 (*sinextendsfdf2, *cosextendsfdf2, *sincosextendsfdf3): Remove
1208 insn patterns and corresponding splitters.
1210 2007-01-09 Kaz Kojima <kkojima@gcc.gnu.org>
1212 * config/sh/t-linux (TARGET_LIBGCC2_CFLAGS): Delete.
1213 (SHLIB_MAPFILES, SHLIB_LINK, SHLIB_INSTALL): Likewise.
1215 2007-01-09 Nicolas Pitre <nico@cam.org>
1218 * arm/ieee754-df.S (Lad_s): Also test the low word of X for zero.
1220 2007-01-08 Geoffrey Keating <geoffk@apple.com>
1222 * target.h (struct gcc_target): New field library_rtti_comdat.
1223 * target-def.h (TARGET_CXX_LIBRARY_RTTI_COMDAT): New.
1224 (TARGET_CXX): Add TARGET_CXX_LIBRARY_RTTI_COMDAT.
1225 * doc/tm.texi (C++ ABI): Document TARGET_CXX_LIBRARY_RTTI_COMDAT.
1226 * config/darwin.h (TARGET_CXX_LIBRARY_RTTI_COMDAT): Define.
1228 2007-01-08 Geoffrey Keating <geoffk@apple.com>
1230 * doc/invoke.texi (Optimize Options): Correct description of -O0.
1232 2007-01-08 Richard Guenther <rguenther@suse.de>
1234 * tree.h (force_fit_type_double): Export.
1235 (force_fit_type): Remove.
1236 * fold-const.c (force_fit_type_double): New function.
1237 (force_fit_type): Remove.
1238 (int_const_binop): Use it.
1239 (fold_convert_const_int_from_int): Likewise.
1240 (fold_convert_const_int_from_real): Likewise.
1241 (fold_div_compare): Likewise.
1242 (fold_sign_changed_comparison): Likewise.
1243 (fold_unary): Likewise.
1244 (fold_negate_const): Likewise.
1245 (fold_abs_const): Likewise.
1246 (fold_not_const): Likewise.
1247 * c-common.c (shorten_compare): Use force_fit_type_double.
1248 * convert.c (convert_to_pointer): Likewise.
1250 2007-01-08 Richard Guenther <rguenther@suse.de>
1252 * tree.h (build_int_cst_wide_type): Export.
1253 * tree.c (build_int_cst_wide_type): New function.
1254 (build_int_cst_wide): Fix comment.
1255 * builtins.c (fold_builtin_object_size): Use build_int_cst
1256 to build -1 or 0 of the correct type. Use fit_double_type
1257 to check for overflow.
1258 * fold-const.c (optimize_bit_field_compare): Use build_int_cst_type
1260 (decode_field_reference): Likewise.
1261 (all_ones_mask_p): Likewise.
1262 (native_interpret_int): Use build_int_cst_wide_type.
1263 (fold_binary): Use build_int_cst_type to build an all-ones
1265 * stor-layout.c (set_sizetype): Use build_int_cst_wide_type.
1267 2007-01-08 Daniel Jacobowitz <dan@codesourcery.com>
1269 * config/pa/t-pa64 (libgcc_stub.a): Use $(T).
1271 2007-01-09 Ben Elliston <bje@au.ibm.com>
1273 * genautomata.c (STATS_OPTION): New option.
1274 (stats_flag): New flag.
1275 (gen_automata_option): Handle it.
1276 (initiate_automaton_gen): Ditto.
1277 (write_automata): Output statistics only if stats_flag is
1278 set. Likewise, output time statistics only if time_flag is set.
1279 * doc/md.texi (Processor pipeline description): Document new flag.
1281 2007-01-08 Richard Guenther <rguenther@suse.de>
1283 * builtins.c (fold_builtin_int_roundingfn): Use fit_double_type.
1284 * tree.c (build_int_cst_type): Likewise.
1285 (size_in_bytes): Don't call force_fit_type on the result.
1286 (int_fits_type_p): Use fit_double_type.
1287 * fold-const.c (fit_double_type): New function.
1288 (force_fit_type): Use it.
1289 * tree.h (fit_double_type): Export.
1291 2007-01-08 Jan Hubicka <jh@suse.cz>
1293 * tree-vectorizer.c (gate_increase_alignment): Fix return type.
1294 * ipa.c (function_and_variable_visibility): Fix return type.
1296 2007-01-08 Richard Guenther <rguenther@suse.de>
1298 * tree-ssa-ccp.c (maybe_fold_offset_to_array_ref): Use type
1299 of offset to build the index.
1300 * tree-pretty-print.c (dump_generic_node): Don't build negated
1301 const just for printing.
1302 * c-pretty-print.c (pp_c_integer_constant): Likewise.
1303 * builtins.c (fold_builtin_int_roundingfn): Check if result
1304 fits the type by using force_fit_type and comparing the result.
1305 * predict.c (predict_loops): Use compare_tree_int for comparison.
1306 * tree.c (build_int_cst): Fall back to integer_type_node for
1308 (build_int_cst_wide): Assert type is non-null.
1310 2007-01-08 Roberto Costa <roberto.costa@st.com>
1312 * tree-vrp.c (extract_range_from_cond_expr): New.
1313 (extract_range_from_expr): Handle COND_EXPR nodes used as expressions.
1314 * tree-ssa-ccp.c (get_maxval_strlen): Handle COND_EXPR nodes used
1316 (fold_stmt): Bug fix, avoid infinite recursion when folding COND_EXPRs.
1317 * tree-ssa-forwprop.c (simplify_cond, forward_propagate_into_cond,
1318 tree_ssa_forward_propagate_single_use_vars): Handle COND_EXPR nodes
1319 used as expressions.
1320 * tree-object-size.c (cond_expr_object_size): New.
1321 (collect_object_sizes_for): Handle COND_EXPR nodes used as expressions.
1323 2007-01-08 Jan Hubicka <jh@suse.cz>
1325 * tree-ssa-forwprop.c (forward_propagate_into_cond,
1326 tree_ssa_forward_propagate_single_use_va): Release defs of propagated
1329 2007-01-08 Richard Guenther <rguenther@suse.de>
1331 PR tree-optimization/23603
1332 * tree-vrp.c (set_value_range_to_truthvalue): New function.
1333 (extract_range_from_binary): Fall back to truthvalue instead of
1334 varying for TRUTH_*_EXPR.
1335 (extract_range_from_comparison): Fall back to truthvalue instead of
1337 (vrp_visit_phi_node): Don't adjust new range bounds to +INF/-INF
1338 if all visited PHI values were constant.
1340 2007-01-08 Jan Hubicka <jh@suse.cz>
1342 * cgraphunit.c (cgraph_process_new_functions): Reset reachable flag.
1343 (cgraph_analyze_function): break out from ...
1344 (cgraph_finalize_compilation_unit): ... here.
1345 (cgraph_expand_function): Remove forgoten commented out line.
1346 (cgraph_optimize): Analyze functions.
1348 2007-01-08 Jan Hubicka <jh@suse.cz>
1350 * tree-pas.h (TODO_remove_function): New flag.
1351 (TODO_update*): Renumber.
1352 (pass_ipa_increase_alignment,
1353 pass_ipa_function_and_variable_visibility): New passes.
1354 * cgraphunit.c (cgraph_increase_alignment): Move to tree-vectorizer.c
1355 (cgraph_function_and_variable_visibility): Move to ipa.c
1356 (cgraph_optimize): Don't call cgraph_function_and_variable_visibility,
1357 cgraph_increase_alignment.
1358 * ipa-inline.c (cgraph_decide_inlining): Don't push timevar.
1359 (cgraph_decide_inlining_incrementally): Push TV_INTEGRATION before
1360 calling tree-inline.
1361 (cgraph_early_inlining): Do not call cgraph_remove_unreachable_nodes.
1362 (pass_ipa_inline, pass_early_ipa_inlining): Set TODO_remove_functions
1363 * tree-vectorizer.c (increase_alignment): Move here from cgraphunit.c
1364 (gate_increase_alignment): New function.
1365 (pass_ipa_increase_alignment): New pass.
1366 * ipa.c: Inline tree-pass.h and timevar.h
1367 (function_and_variable_visibility): Move here from cgraphunit.c
1368 * tree-optimize.c (pass_early_local_passes): Add TODO_remove_functions.
1369 * passes.c (init_optimization_passes): Add the two new passes.
1370 (execute_todo): Handle cgraph_remove_functions.
1372 2007-01-08 Nick Clifton <nickc@redhat.com>
1374 * config/frv/predicates.md (reg_or_0_operand): Accept
1377 2007-01-08 Ralf Corsépius <ralf.corsepius@rtems.org>
1379 * config/bfin/rtems.h, config/bfin/t-rtems: New.
1380 * config.gcc: Add bfin*-rtems*.
1382 2007-01-08 Mark Shinwell <shinwell@codesourcery.com>
1384 * c.opt: Add -flax-vector-conversions.
1385 * c-typeck.c (convert_for_assignment): Pass flag to
1386 vector_types_convertible_p to allow emission of note.
1387 (digest_init): Likewise.
1388 * c-opts.c: Handle -flax-vector-conversions.
1389 * c-common.c (flag_lax_vector_conversions): New.
1390 (vector_types_convertible_p): Unless -flax-vector conversions
1391 has been passed, disallow conversions between vectors with
1392 differing numbers of subparts and/or element types. If such
1393 a conversion is disallowed, possibly emit a note on the first
1394 occasion only to inform the user of -flax-vector-conversions.
1395 The new last argument specifies this.
1396 * c-common.h (flag_lax_vector_conversions): New.
1397 (vector_types_convertible_p): Add extra argument.
1398 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use
1399 char_type_node for V*QI type vectors.
1400 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
1401 Update to satisfy new typechecking rules.
1402 * config/rs6000/altivec.h (vec_cmple): Use vec_cmpge, for both
1404 * doc/invoke.texi (C Dialect Options): Document
1405 -flax-vector-conversions.
1407 2007-01-08 Mark Shinwell <shinwell@codesourcery.com>
1409 PR tree-optimization/29877
1410 * tree-ssa-ter.c (is_replaceable_p): Deem assignments with
1411 a register variable on the RHS to not be replaceable.
1413 2007-01-08 Chen Liqin <liqin@sunnorth.com.cn>
1414 * config/score/t-score-elf (MULTILIB_OPTIONS): Change.
1415 * config/score/predicates.md (const_uimm5, sr0_operand, const_simm12,
1416 const_simm15, const_pow2, const_npow2): Added.
1417 * config/score/misc.md (insv, extv, extzv, movmemsi,
1418 move_lbu_a/b, mov_lhu_a/b etc): Added and fix some bug.
1419 * config/score/score.c (score_address_cost, score_select_cc_mode):
1421 Change CONST_OK_FOR_LETTER_P/EXTRA_CONSTRAINT define.
1422 Update score_rtx_costs for MACRO TARGET_RTX_COSTS.
1423 Update score_print_operand.
1424 * config/score/score.h (DATA_ALIGNMENT, SELECT_CC_MODE): Added.
1425 Adjust register allocate order and update some macro define.
1426 * config/score/score-mdaux.c (mdx_unaligned_load, mdx_unsigned_store,
1427 mdx_block_move_straight, mdx_block_move_loop_head,
1428 mdx_block_move_loop_body, mdx_block_move_loop_foot, mdx_block_move_loop,
1429 mdx_block_move): Added.
1430 (mdx_movsicc, mdp_select_add_imm, mdp_select, mds_zero_extract_andi,
1431 mdp_limm): Updated and fix some bug and typo.
1432 * config/score/score.md (movqi/hi/si, add/sub/zero/ext): Updated.
1433 (movsf, movdf, doloop_end): Added.
1435 2007-01-08 Kazu Hirata <kazu@codesourcery.com>
1437 * config/arm/arm.c, config/arm/arm.h, config/arm/arm.md,
1438 config/arm/thumb2.md: Fix comment typos.
1439 * doc/extend.texi: Fix a typo.
1441 2007-01-07 Eric Christopher <echristo@apple.com>
1443 * configure.ac: Check for __stack_chk_fail for darwin.
1444 * configure: Regenerate.
1446 2007-01-07 Richard Guenther <rguenther@suse.de>
1448 * tree-vrp.c (extract_range_from_assert): CSE calls to
1449 compare_values where possible.
1450 (extract_range_from_unary_expr): Likewise.
1452 2007-01-07 Anatoly Sokolov <aesok@post.ru>
1454 * config/avr/avr-protos.h (call_insn_operand): Delete prototype.
1455 * config/avr/avr.c (call_insn_operand): Delete function.
1456 * config/avr/avr.md (*pushqi, *pushhi, *pushsi, *pushsf): Use REG_SP
1457 instead of register number. Use predicates.
1458 * config/avr/predicates.md (const0_operand, reg_or_0_operand,
1459 call_insn_operand): Add.
1461 2007-01-06 Jan Hubicka <jh@suse.cz>
1463 * tree-pass.h (pass_build_cgraph_edges): Declare.
1464 * cgraphunit.c (record_refernece): Move to cgraphbuild.c
1465 (visited_nodes): Remove.
1466 (cgraph_create_edges): Move to cgraphbuild.c; rename to
1467 build_cgrpah_edges; make visited_nodes local.
1468 (cgraph_process_new_functions): DO not call initialize_inline_failed.
1469 (record_references_in_initializer): Move to cgraphbuild.c
1470 (initialize_inline_failed, rebuild_cgraph_edges,
1471 pass_rebuild_cgraph_edges): Move to cgraphbuild.c.
1472 (verify_cgraph_node): Make visited_nodes local.
1473 (cgraph_analyze_function): Do not call cgraph_create_edges and
1474 initialize_inline_failed.
1475 (cgraph_expand_function): Do not call cgraph_lower_function;
1476 assert that function is already lowered.
1477 * Makefile.in (cgraphbuild.o): New.
1478 * passes.c (init_optimization_passes): Add pass_build_cgraph_edges
1479 at the end of lowering passes.
1481 2007-01-06 Steven Bosscher <steven@gcc.gnu.org>
1483 * ifcvt.c (cond_move_convert_if_block): New function, code
1484 factored out from...
1485 (cond_move_process_if_block): ...here. Call the new function
1486 on the THEN and ELSE blocks.
1487 (merge_if_block): Do not copy global_live_at_end, merge_blocks
1488 already takes care of this.
1490 2007-01-05 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
1493 * tree.h (TREE_OVERFLOW_P): New.
1494 * c-typeck.c (parser_build_unary_op): Warn only if result
1495 overflowed and operands did not.
1496 (parser_build_binary_op): Likewise.
1497 (convert_for_assignment): Remove redundant overflow_warning.
1498 * c-common.c (overflow_warning): Don't check or set TREE_OVERFLOW.
1500 2007-01-05 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
1502 * c-typeck.c (store_init_value): Split over two lines to follow
1503 the GNU coding style.
1505 2007-01-05 Benjamin Kosnik <bkoz@redhat.com>
1507 * c-cppbuiltin.c (c_cpp_builtins): __GXX_EXPERIMENTAL_CPP0X__ to
1508 __GXX_EXPERIMENTAL_CXX0X__.
1509 * doc/cpp.texi: Same.
1511 2007-01-05 Richard Guenther <rguenther@suse.de>
1514 * tree.c (get_narrower): Do not construct COMPONENT_REFs
1515 with mismatched types. Instead explicitly build a
1516 conversion NOP_EXPR.
1518 2007-01-05 Ian Lance Taylor <iant@google.com>
1520 * c-common.c (decl_with_nonnull_addr_p): New function.
1521 (c_common_truthvalue_conversion): Call it.
1522 * c-typeck.c (build_binary_op): Likewise.
1523 * c-common.h (decl_with_nonnull_addr_p): Declare.
1525 2007-01-05 Jakub Jelinek <jakub@redhat.com>
1528 * libgcc2.c (__divdc3): Compare c and d against 0.0 instead of
1531 2007-01-05 Joel Brobecker <brobecker@adacore.com>
1533 * doc/install.texi (Final install): Document the fact that
1534 the GNAT runtime should not be stripped.
1536 2007-01-04 Jan Hubicka <jh@suse.cz>
1538 * tree-inline.c (fold_marked_statements): Update operand caches
1539 and EH after folding
1541 2007-01-04 Ian Lance Taylor <iant@google.com>
1543 * c-common.c (check_function_nonnull): Whitespace fix.
1545 2007-01-04 Jan Hubicka <jh@suse.cz>
1547 * tree-optimize.c (execute_fixup_cfg): Correct previously mistakely
1548 comitted older version of patch.
1549 (pass_fixup_cfg): Add TODOs to verify flow and statements, dump
1550 function, celanup cfg and collect garbage.
1552 2007-01-04 Mike Stump <mrs@apple.com>
1554 * Makefile.in (mostlyclean): Don't remove libgcc anymore.
1557 2007-01-04 Eric Christopher <echristo@apple.com>
1559 * libgcc2.c (__bswapsi2): Use SItype.
1560 (__bswapdi2): Use DItype.
1561 * libgcc2.h: Update for above.
1563 2007-01-04 Paul Brook <paul@codesourcery.com>
1565 * config/arm/arm.md (arm_mulsi3, thumb_mulsi3, mulsi3_compare0,
1566 mulsi_compare0_scratch, mulsi3addsi, mulsi3addsi_compare0,
1567 mulsi3addsi_compare0_scratch, mulsidi3adddi, mulsidi3,
1568 umulsidi3, umulsidi3adddi, smulsi3_highpart,
1569 umulsi3_highpart): Make conditional on !arm_arch6.
1570 (arm_mulsi3_v6, thumb_mulsi3_v6, mulsi3_compare0_v6,
1571 mulsi_compare0_scratch_v6, mulsi3addsi_v6, mulsi3addsi_compare0_v6,
1572 mulsi3addsi_compare0_scratch_v6, mulsidi3adddi_v6, mulsidi3_v6,
1573 umulsidi3_v6, umulsidi3adddi_v6, smulsi3_highpart_v6,
1574 umulsi3_highpart_v6): New insns.
1576 2007-01-04 Roger Sayle <roger@eyesopen.com>
1578 * fold-const.c (fold_convert): When casting an expression to void,
1579 fold_ignored_result may discover a GIMPLE_MODIFY_STMT which doesn't
1580 have a type. Instead of attempting to build a NOP_EXPR, return
1581 these "special" trees directly.
1583 2007-01-04 Joseph Myers <joseph@codesourcery.com>
1585 * config/rs6000/rs6000.c (rs6000_rtx_costs): Make adjustment for
1586 MULT inside MINUS as either argument. Use rs6000_cost->dmul -
1587 rs6000_cost->fp not 0 as adjustment for outer NEG.
1589 2007-01-04 Jan Hubicka <jh@suse.cz>
1591 * cgraph.c (cgraph_release_function_body): New function.
1592 (cgraph_remove_node): Use it.
1593 * cgraph.h (cgraph_release_function_body): Declare.
1594 * cgraphunit.c (cgraph_expand_function): Use it.
1595 * ipa.c (cgraph_remove_unreahchable_nodes): Use it.
1596 * tree-ssa.c (delete_tree_ssa): Allow to be called before aliasing
1597 is initialized and while compilation of other function is running.
1598 * tree-optimize.c (execute_free_cfg_annotations): Move code to clear
1599 statement CFG annotations from here to ...
1600 * tree-cfg.c (delete_tree_cfg_annotations): ... here.
1602 2007-01-04 Zdenek Dvorak <dvorakz@suse.cz>
1604 * cfgloop.h (enum li_flags): Make the constants powers of two.
1606 2007-01-04 Jan Hubicka <jh@suse.cz>
1608 * tree-inline.c (copy_bb): Insert new statements to statements_to_fold
1610 (fold_marked_statements): New function.
1611 (optimize_inline_calls, tree_function_versioning): Fold new statements.
1612 * tree-inline.h (copy_body_data): Add statements_to_fold.
1614 2007-01-03 Daniel Jacobowitz <dan@codesourcery.com>
1616 * config.gcc: Mention libgcc/config.host.
1617 * Makefile.in: Update comments mentioning libgcc.
1618 (LIBGCC, INSTALL_LIBGCC, GCC_PARTS, mklibgcc): Delete.
1619 (all.cross, start.encap, rest.encap, rest.cross): Update
1620 dependencies for libgcc move.
1621 (libgcc.mk, LIBGCC_DEPS, libgcov.a, libgcc.a, stmp-multilib)
1622 (clean-target, clean-target-libgcc): Delete.
1623 (srcdirify, GCC_EXTRA_PARTS): New macros.
1624 (libgcc-support, libgcc.mvars): New rules.
1625 (distclean): Remove mention of mklibgcc.
1626 (install): Don't reference INSTALL_LIBGCC.
1627 (install-common): Don't reference EXTRA_PARTS.
1628 (install-libgcc, install-multilib): Delete rules.
1629 * mklibgcc.in: Delete file.
1630 * doc/configfiles.texi: Don't mention mklibgcc.
1632 * config/i386/t-darwin (SHLIB_VERPFX): Delete (moved to libgcc).
1633 * config/i386/t-darwin64 (SHLIB_VERPFX): Likewise.
1634 * config/rs6000/t-darwin (SHLIB_VERPFX): Likewise.
1635 * config/rs6000/t-ppccomm (TARGET_LIBGCC2_CFLAGS, SHLIB_MAPFILES)
1636 (mklibgcc, ldblspecs): Likewise.
1638 * config/i386/t-nwld (libgcc.def, libc.def, libpcre.def)
1639 (posixpre.def): Use $(T).
1640 (SHLIB_EXT, SHLIB_NAME, SHLIB_SLIBDIR_QUAL, SHLIB_DEF, SHLIB_MAP)
1641 (SHLIB_SRC, SHLIB_INSTALL): Delete.
1642 (SHLIB_LINK): Make dummy.
1643 * config/t-slibgcc-darwin: Delete contents except for dummy SHLIB_LINK.
1645 * config/frv/t-linux (EXTRA_MULTILIB_PARTS): Clear.
1647 * config/alpha/t-crtfm: Use $(T) in rules for EXTRA_PARTS.
1648 * config/alpha/t-vms, config/alpha/t-vms64, config/fr30/t-fr30,
1649 config/i386/t-rtems-i386, config/ia64/t-ia64, config/rs6000/t-beos,
1650 config/rs6000/t-newas, config/sparc/t-elf: Likewise.
1652 * configure.ac (all_outputs): Remove mklibgcc.
1653 * configure: Regenerated.
1655 2007-01-03 Josh Conner <jconner@apple.com>
1658 * calls.c (compute_argument_addresses): Set stack and stack_slot
1659 for partial args, too.
1660 (store_one_arg): Use locate.size.constant for the size when
1661 generating a save_area.
1663 2007-01-03 Robert Kennedy <jimbob@google.com>
1665 * tree-cfg.c (tree_merge_blocks): Release SSA_NAME phi results
1666 whose definitions are deleted due to basic block merging.
1668 2007-01-03 Paul Brook <paul@codesourcery.com>
1671 * config/arm/arm.c (output_return_instruction): Pop PC in interrupt
1673 (use_return_insn): Return 0 for Thumb interrupt functions.
1674 (print_multi_reg): Add rfe argument for IRQ returns.
1675 (arm_output_epilogue): Pop interrupt return address directly into PC.
1676 (arm_expand_prologue): Only adjust IRQ return address in Arm mode.
1678 2007-01-03 Paul Brook <paul@codesourcery.com>
1680 Merge from sourcerygxx-4_1.
1681 * config/arm/thumb2.md: New file.
1682 * config/arm/elf.h (JUMP_TABLES_IN_TEXT_SECTION): Return True for
1684 * config/arm/coff.h (JUMP_TABLES_IN_TEXT_SECTION): Ditto.
1685 * config/arm/aout.h (ASM_OUTPUT_ADDR_VEC_ELT): Add !Thumb-2 assertion.
1686 (ASM_OUTPUT_ADDR_DIFF_ELT): Output Thumb-2 jump tables.
1687 * config/arm/aof.h (ASM_OUTPUT_ADDR_DIFF_ELT): Output Thumb-2 jump
1689 (ASM_OUTPUT_ADDR_VEC_ELT): Add !Thumb-2 assertion.
1690 * config/arm/ieee754-df.S: Use macros for Thumb-2/Unified asm
1692 * config/arm/ieee754-sf.S: Ditto.
1693 * config/arm/arm.c (thumb_base_register_rtx_p): Rename...
1694 (thumb1_base_register_rtx_p): ... to this.
1695 (thumb_index_register_rtx_p): Rename...
1696 (thumb1_index_register_rtx_p): ... to this.
1697 (thumb_output_function_prologue): Rename...
1698 (thumb1_output_function_prologue): ... to this.
1699 (thumb_legitimate_address_p): Rename...
1700 (thumb1_legitimate_address_p): ... to this.
1701 (thumb_rtx_costs): Rename...
1702 (thumb1_rtx_costs): ... to this.
1703 (thumb_compute_save_reg_mask): Rename...
1704 (thumb1_compute_save_reg_mask): ... to this.
1705 (thumb_final_prescan_insn): Rename...
1706 (thumb1_final_prescan_insn): ... to this.
1707 (thumb_expand_epilogue): Rename...
1708 (thumb1_expand_epilogue): ... to this.
1709 (arm_unwind_emit_stm): Rename...
1710 (arm_unwind_emit_sequence): ... to this.
1711 (thumb2_legitimate_index_p, thumb2_legitimate_address_p,
1712 thumb1_compute_save_reg_mask, arm_dwarf_handle_frame_unspec,
1713 thumb2_index_mul_operand, output_move_vfp, arm_shift_nmem,
1714 arm_save_coproc_regs, thumb_set_frame_pointer, arm_print_condition,
1715 thumb2_final_prescan_insn, thumb2_asm_output_opcode, arm_output_shift,
1716 thumb2_output_casesi): New functions.
1717 (TARGET_DWARF_HANDLE_FRAME_UNSPEC): Define.
1718 (FL_THUMB2, FL_NOTM, FL_DIV, FL_FOR_ARCH6T2, FL_FOR_ARCH7,
1719 FL_FOR_ARCH7A, FL_FOR_ARCH7R, FL_FOR_ARCH7M, ARM_LSL_NAME,
1720 THUMB2_WORK_REGS): Define.
1721 (arm_arch_notm, arm_arch_thumb2, arm_arch_hwdiv, arm_condexec_count,
1722 arm_condexec_mask, arm_condexec_masklen)): New variables.
1723 (all_architectures): Add armv6t2, armv7, armv7a, armv7r and armv7m.
1724 (arm_override_options): Check new CPU capabilities.
1725 Set new architecture flag variables.
1726 (arm_isr_value): Handle v7m interrupt functions.
1727 (user_return_insn): Return 0 for v7m interrupt functions. Handle
1729 (const_ok_for_arm): Handle Thumb-2 constants.
1730 (arm_gen_constant): Ditto. Use movw when available.
1731 (arm_function_ok_for_sibcall): Return false for v7m interrupt
1733 (legitimize_pic_address, arm_call_tls_get_addr): Handle Thumb-2.
1734 (thumb_find_work_register, arm_load_pic_register,
1735 legitimize_tls_address, arm_address_cost, load_multiple_sequence,
1736 emit_ldm_seq, emit_stm_seq, arm_select_cc_mode, get_jump_table_size,
1737 print_multi_reg, output_mov_long_double_fpa_from_arm,
1738 output_mov_long_double_arm_from_fpa, output_mov_double_fpa_from_arm,
1739 output_mov_double_fpa_from_arm, output_move_double,
1740 arm_compute_save_reg_mask, arm_compute_save_reg0_reg12_mask,
1741 output_return_instruction, arm_output_function_prologue,
1742 arm_output_epilogue, arm_get_frame_offsets, arm_regno_class,
1743 arm_output_mi_thunk, thumb_set_return_address): Ditto.
1744 (arm_expand_prologue): Handle Thumb-2. Use arm_save_coproc_regs.
1745 (arm_coproc_mem_operand): Allow POST_INC/PRE_DEC.
1746 (arithmetic_instr, shift_op): Use arm_shift_nmem.
1747 (arm_print_operand): Use arm_print_condition. Handle '(', ')', '.',
1749 (arm_final_prescan_insn): Use extract_constrain_insn_cached.
1750 (thumb_expand_prologue): Use thumb_set_frame_pointer.
1751 (arm_file_start): Output directive for unified syntax.
1752 (arm_unwind_emit_set): Handle stack alignment instruction.
1753 * config/arm/lib1funcs.asm: Remove default for __ARM_ARCH__.
1754 Add v6t2, v7, v7a, v7r and v7m.
1755 (RETLDM): Add Thumb-2 code.
1756 (do_it, shift1, do_push, do_pop, COND, THUMB_SYNTAX): New macros.
1757 * config/arm/arm.h (TARGET_CPU_CPP_BUILTINS): Define __thumb2__.
1758 (TARGET_THUMB1, TARGET_32BIT, TARGET_THUMB2, TARGET_DSP_MULTIPLY,
1759 TARGET_INT_SIMD, TARGET_UNIFIED_ASM, ARM_FT_STACKALIGN, IS_STACKALIGN,
1760 THUMB2_TRAMPOLINE_TEMPLATE, TRAMPOLINE_ADJUST_ADDRESS,
1761 ASM_OUTPUT_OPCODE, THUMB2_GO_IF_LEGITIMATE_ADDRESS,
1762 THUMB2_LEGITIMIZE_ADDRESS, CASE_VECTOR_PC_RELATIVE,
1763 CASE_VECTOR_SHORTEN_MODE, ADDR_VEC_ALIGN, ASM_OUTPUT_CASE_END,
1764 ADJUST_INSN_LENGTH): Define.
1765 (TARGET_REALLY_IWMMXT, TARGET_IWMMXT_ABI, CONDITIONAL_REGISTER_USAGE,
1766 STATIC_CHAIN_REGNUM, HARD_REGNO_NREGS, INDEX_REG_CLASS,
1767 BASE_REG_CLASS, MODE_BASE_REG_CLASS, SMALL_REGISTER_CLASSES,
1768 PREFERRED_RELOAD_CLASS, SECONDARY_OUTPUT_RELOAD_CLASS,
1769 SECONDARY_INPUT_RELOAD_CLASS, LIBCALL_VALUE, FUNCTION_VALUE_REGNO_P,
1770 TRAMPOLINE_SIZE, INITIALIZE_TRAMPOLINE, HAVE_PRE_INCREMENT,
1771 HAVE_POST_DECREMENT, HAVE_PRE_DECREMENT, HAVE_PRE_MODIFY_DISP,
1772 HAVE_POST_MODIFY_DISP, HAVE_PRE_MODIFY_REG, HAVE_POST_MODIFY_REG,
1773 REGNO_MODE_OK_FOR_BASE_P, LEGITIMATE_CONSTANT_P,
1774 REG_MODE_OK_FOR_BASE_P, REG_OK_FOR_INDEX_P, GO_IF_LEGITIMATE_ADDRESS,
1775 LEGITIMIZE_ADDRESS, THUMB2_LEGITIMIZE_ADDRESS,
1776 GO_IF_MODE_DEPENDENT_ADDRESS, MEMORY_MOVE_COST, BRANCH_COST,
1777 ASM_APP_OFF, ASM_OUTPUT_CASE_LABEL, ARM_DECLARE_FUNCTION_NAME,
1778 FINAL_PRESCAN_INSN, PRINT_OPERAND_PUNCT_VALID_P,
1779 PRINT_OPERAND_ADDRESS): Adjust for Thumb-2.
1780 (arm_arch_notm, arm_arch_thumb2, arm_arch_hwdiv): New declarations.
1781 * config/arm/arm-cores.def: Add arm1156t2-s, cortex-a8, cortex-r4 and
1783 * config/arm/arm-tune.md: Regenerate.
1784 * config/arm/arm-protos.h: Update prototypes.
1785 * config/arm/vfp.md: Enable patterns for Thumb-2.
1786 (arm_movsi_vfp): Add movw alternative. Use output_move_vfp.
1787 (arm_movdi_vfp, movsf_vfp, movdf_vfp): Use output_move_vfp.
1788 (thumb2_movsi_vfp, thumb2_movdi_vfp, thumb2_movsf_vfp,
1789 thumb2_movdf_vfp, thumb2_movsfcc_vfp, thumb2_movdfcc_vfp): New.
1790 * config/arm/libunwind.S: Add Thumb-2 code.
1791 * config/arm/constraints.md: Update include Thumb-2.
1792 * config/arm/ieee754-sf.S: Add Thumb-2/Unified asm support.
1793 * config/arm/ieee754-df.S: Ditto.
1794 * config/arm/bpabi.S: Ditto.
1795 * config/arm/t-arm (MD_INCLUDES): Add thumb2.md.
1796 * config/arm/predicates.md (low_register_operand,
1797 low_reg_or_int_operand, thumb_16bit_operator): New.
1798 (thumb_cmp_operand, thumb_cmpneg_operand): Rename...
1799 (thumb1_cmp_operand, thumb1_cmpneg_operand): ... to this.
1800 * config/arm/t-arm-elf: Add armv7 multilib.
1801 * config/arm/arm.md: Update patterns for Thumb-2 and Unified asm.
1803 (UNSPEC_STACK_ALIGN, ce_count): New.
1804 (arm_incscc, arm_decscc, arm_umaxsi3, arm_uminsi3,
1805 arm_zero_extendsidi2, arm_zero_extendqidi2): New
1807 * config/arm/fpa.md: Update patterns for Thumb-2 and Unified asm.
1808 (thumb2_movsf_fpa, thumb2_movdf_fpa, thumb2_movxf_fpa,
1809 thumb2_movsfcc_fpa, thumb2_movdfcc_fpa): New insns.
1810 * config/arm/cirrus.md: Update patterns for Thumb-2 and Unified asm.
1811 (cirrus_thumb2_movdi, cirrus_thumb2_movsi_insn,
1812 thumb2_cirrus_movsf_hard_insn, thumb2_cirrus_movdf_hard_insn): New
1814 * doc/extend.texi: Document ARMv7-M interrupt functions.
1815 * doc/invoke.texi: Document Thumb-2 new cores+architectures.
1817 2007-01-03 Jakub Jelinek <jakub@redhat.com>
1819 * unwind-dw2.c (SIGNAL_FRAME_BIT, EXTENDED_CONTEXT_BIT): Define.
1820 (struct _Unwind_Context): Rename args_size to flags, remove
1821 signal_frame field, add a new args_size field and version field.
1822 (_Unwind_IsSignalFrame, _Unwind_SetSignalFrame,
1823 _Unwind_IsExtendedContext): New inline functions.
1824 (_Unwind_GetGR, _Unwind_SetGR, _Unwind_GetGRPtr, _Unwind_SetGRPtr):
1825 Assume by_value array is only present if _Unwind_IsExtendedContext.
1826 (_Unwind_GetIPInfo, execute_cfa_program, uw_frame_state_for): Use
1827 _Unwind_IsSignalFrame.
1828 (__frame_state_for): Initialize context.flags to EXTENDED_CONTEXT_BIT.
1829 (uw_update_context_1): Use _Unwind_SetSignalFrame.
1830 (uw_init_context_1): Initialize context->flags to
1831 EXTENDED_CONTEXT_BIT.
1832 * config/rs6000/linux-unwind.h (frob_update_context): Use
1833 _Unwind_SetSignalFrame.
1835 2007-01-03 Andrew Pinski <andrew_pinski@playstation.sony.com>
1838 * gimplify.c (gimplify_modify_expr_complex_part): Move below
1839 tree_to_gimple_tuple. Call tree_to_gimple_tuple when we need
1842 2007-01-03 Kazu Hirata <kazu@codesourcery.com>
1844 * config/i386/i386.h (NON_STACK_REG_P, REGNO_OK_FOR_SIREG_P,
1845 REGNO_OK_FOR_DIREG_P, REWRITE_ADDRESS, ASM_OPERAND_LETTER,
1846 RET, AT_SP): Remove.
1847 * config/i386/i386.md (*sse_prologue_save_insn): Use return
1850 * alias.c (init_alias_analysis): Use VEC_safe_grow_cleared.
1851 * cfgbuild.c (find_basic_blocks): Likewise.
1852 * cfgrtl.c (rtl_create_basic_block): Likewise.
1853 * function.c (temp_slots_at_level): Likewise.
1854 * reg-stack.c (stack_regs_mentioned): Likewise.
1855 * regclass.c (allocate_reg_info): Likewise.
1856 * tree-cfg.c (init_empty_tree_cfg, build_tree_cfg, create_bb,
1857 set_bb_for_stmt, move_block_to_fn): Likewise.
1858 * tree-complex.c (tree_lower_complex): Likewise.
1859 * vec.h (VEC_safe_grow_cleared): New.
1861 * cgraphunit.c, tree-ssa-alias.c: Fix comment typos.
1863 2007-01-03 Zdenek Dvorak <dvorakz@suse.cz>
1865 * loop-unswitch.c (unswitch_loop): Pass probabilities to loopify.
1866 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Pass probabilities
1868 * cfgloopmanip.c (scale_loop_frequencies): Export.
1869 (loopify): Scale the frequencies by prescribed coefficients.
1870 (set_zero_probability): New function.
1871 (duplicate_loop_to_header_edge): Improve updating of frequencies.
1872 (lv_adjust_loop_entry_edge, loop_version): Set probabilities
1873 and frequencies according to arguments.
1874 * tree-ssa-loop-manip.c (tree_unroll_loop): Set probabilities
1876 * cfg.c (scale_bbs_frequencies_int): Allow scaling the frequencies up.
1877 * modulo-sched.c (sms_schedule): Set probabilities for entering
1878 versioned loop correctly.
1879 * tree-vect-transform.c (vect_transform_loop): Ditto.
1880 * cfgloop.h (loopify, loop_version): Declaration changed.
1881 (scale_loop_frequencies): Declared.
1883 2007-01-02 Jan Hubicka <jh@suse.cz>
1885 * cgraph.c: Include tree-flow.h
1886 (cgraph_add_new-function): Handle IPA_SSA mode; execute
1888 * cgraph.h (enum cgraph_state): Add CGRAPH_STATE_IPA_SSA.
1889 * tree-pass.h (pass_all_early_optimizations): Declare.
1890 * cgraphunit.c (cgraph_process_new_functions): Add IPA_SSA; execute
1892 (cgraph_analyze_function): Do early_local_passes.
1893 * tree-mudflap.c (mf_decl_cache_locals, mf_build_check_statement_for):
1894 Do not add referenced vars.
1895 * tree-optimize.c (gate_all_optimizations): Do not execute when not in
1897 (gate_all_early_local_passes): New gate.
1898 (pass_early_local_passes): Use new gate.
1899 (execute_early_local_optimizations): New functions.
1900 (gate_all_early_optimizations): New gate.
1901 (pass_all_early_optimizations): New pass.
1902 (execute_free_datastructures): Free SSA only when initialized.
1903 (gate_init_datastructures): Init only when optimizing.
1904 (tree_lowering_passes): Do early local passes when called late.
1905 * tree-profile.c (do_tree_profiling): Don't profile functions added
1907 (do_early_tree_profiling, pass_early_tree_profile): Kill.
1908 * tree-cfg.c (update_modified_stmts): Do not update when operands are
1910 * passes.c (init_optimizations_passes): Reorder so we go into SSA
1911 during early_local_passes.
1912 * Makefile.in (cgraph.o): Add dependency on tree-flow.h.
1915 2007-01-02 Carlos O'Donell <carlos@codesourcery.com>
1917 * Makefile.in: Update copyright year.
1919 2007-01-02 Carlos O'Donell <carlos@codesourcery.com>
1921 * Makefile.in: Export GCC_EXEC_PREFIX before calling $(RUNTEST)
1922 in $(lang_checks) and check-consistency targets.
1924 2007-01-02 Jan Hubicka <jh@suse.cz>
1926 * tree-mudflap.c (mf_decl_cache_locals, mf_build_check_statement_for):
1927 Do not add referenced vars.
1928 * tree-cfg.c (update_modified_stmts): Do not update when SSA operands
1930 * passes.c (init_optimization_passes): Put mudflap_2 after
1931 free_datastructures.
1933 2007-01-02 Jan Hubicka <jh@suse.cz>
1935 * tree-optimize (execute_fixup_cfg): Set after_inlining flag.
1936 Set NOTHROW flag on call statements proved to be nothrow.
1937 Update statement of local calls so new pure/const functions are
1938 updated. Update_ssa when in ssa form. Mark PHI nodes of nonlocal
1940 (tree_rest_of_compilation): Register hooks and initialize bitmap
1941 early. Do not set after_inlining flag.
1943 2007-01-02 Steve Ellcey <sje@cup.hp.com>
1945 * sbitmap.c (HOST_BITS_PER_LONG_LONG): Change to
1946 HOST_BITS_PER_LONGLONG
1948 2007-01-02 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
1951 * c-typeck.c (store_init_value): Don't emit pedantic overflow
1952 warning for non-static initializers.
1954 2007-01-02 Steven Bosscher <steven@gcc.gnu.org>
1956 * config/alpha/alpha.md, arm/arm.c, darwin.c, frv/frv.md,
1957 m32r/m32r.c, m32r/m32r.c, mn10300/mn10300.md, pa/pa.c,
1958 rs6000/rs6000.c, s390/s390.md, sh/sh.md, sparc/sparc.c:
1959 Always use set_unique_reg_note to add REG_EQUAL notes.
1961 2007-01-02 Kazu Hirata <kazu@codesourcery.com>
1964 2007-01-02 Kazu Hirata <kazu@codesourcery.com>
1966 * alias.c (init_alias_analysis): Use VEC_safe_grow_cleared.
1967 * cfgbuild.c (find_basic_blocks): Likewise.
1968 * cfgrtl.c (rtl_create_basic_block): Likewise.
1969 * function.c (temp_slots_at_level): Likewise.
1970 * reg-stack.c (stack_regs_mentioned): Likewise.
1971 * regclass.c (allocate_reg_info): Likewise.
1972 * tree-cfg.c (init_empty_tree_cfg, build_tree_cfg, create_bb,
1973 set_bb_for_stmt, move_block_to_fn): Likewise.
1974 * tree-complex.c (tree_lower_complex): Likewise.
1975 * vec.h (VEC_safe_grow_cleared): New.
1977 2007-01-02 Ian Lance Taylor <iant@google.com>
1979 * c-common.c (c_common_truthvalue_conversion): When warning about
1980 using an assignment as a truth value, set TREE_NO_WARNING.
1982 2007-01-02 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
1985 * c.opt (Wold-style-declaration): New.
1986 * doc/invoke.texi (C-only Warning Options): New.
1987 (Wold-style-declaration): Document it.
1988 (Wextra): Enabled by -Wextra.
1989 * c-opts.c (c_common_post_options): Enabled by -Wextra.
1990 * c-decl.c (declspecs_add_scspec): Replace -Wextra with
1991 -Wold-style-declaration.
1993 2007-01-02 Kazu Hirata <kazu@codesourcery.com>
1995 * alias.c (init_alias_analysis): Use VEC_safe_grow_cleared.
1996 * cfgbuild.c (find_basic_blocks): Likewise.
1997 * cfgrtl.c (rtl_create_basic_block): Likewise.
1998 * function.c (temp_slots_at_level): Likewise.
1999 * reg-stack.c (stack_regs_mentioned): Likewise.
2000 * regclass.c (allocate_reg_info): Likewise.
2001 * tree-cfg.c (init_empty_tree_cfg, build_tree_cfg, create_bb,
2002 set_bb_for_stmt, move_block_to_fn): Likewise.
2003 * tree-complex.c (tree_lower_complex): Likewise.
2004 * vec.h (VEC_safe_grow_cleared): New.
2006 2007-01-02 Douglas Gregor <doug.gregor@gmail.com>
2008 * c-common.c(c_common_nodes_and_builtins): Since variants of
2009 void_type_node get built before it is given a name, we need to
2010 give those variants the name, too.
2011 (complete_array_type): We need to work with the canonical main
2012 type of the array, from which we will build the qualified version.
2013 * params.def (PARAM_VERIFY_CANONICAL_TYPES): New.
2014 * print-tree.c (print_node): Display canonical type information
2016 * stor-layout.c (layout_type): When we don't know the
2017 alignment of a type for which we're building an array, we end up
2018 guessing wrong, so make the type require structural equality.
2019 * tree.c (make_node_stat): When we build a new type, it is its
2021 (build_type_attribute_qual_variant): When building an attribute
2022 variant, its canonical type is the non-attribute variant. However,
2023 if the attributes are target-dependent and they differ, we need to
2024 use structural equality checks for this type.
2025 (build_qualified_type): A qualified type is not equivalent to its
2026 unqualified variant; set the canonical type appropriately.
2027 (build_distinct_type_copy): When building a distinct type from
2028 another type, the new type is its own canonical type.
2029 (build_variant_type_copy): When building a new type variant, we
2030 assume that it is equivalent to the original type.
2031 (build_pointer_type_for_mode): When building a pointer type, also
2032 build a canonical type pointer.
2033 (build_reference_type_for_mode): When building a reference type,
2034 also build a canonical type reference.
2035 (build_index_type): When we can't hash an index type (e.g.,
2036 because its maximum value is negative), the index type requires
2037 structural equality tests.
2038 (build_array_type): Build the canonical form of an array type.
2039 (build_function_type): Function types require structural equality,
2040 because they contain default arguments, attributes, etc.
2041 (build_method_type_directly): Ditto for method types.
2042 (build_offset_type): Build the canonical offset type.
2043 (build_complex_type): Build the canonical vector type.
2044 (make_vector_type): Build the canonical vector type.
2045 * tree.h (TYPE_CANONICAL): New.
2046 (TYPE_STRUCTURAL_EQUALITY_P): New.
2047 (SET_TYPE_STRUCTURAL_EQUALITY): New.
2048 (struct tree_type): Added "canonical" field.
2049 * params.h (VERIFY_CANONICAL_TYPES): New.
2050 * doc/c-tree.texi (TYPE_CANONICAL): Document.
2051 (TYPE_STRUCTURAL_EQUALITY_P): Document.
2052 (SET_TYPE_STRUCTURAL_EQUALITY): Document.
2053 * doc/invoke.texi (verify-canonical-types): Document --param
2054 parameter for verifying canonical types.
2056 2007-01-02 Joseph Myers <joseph@codesourcery.com>
2058 * config.gcc (powerpc-*-eabispe*, powerpc-*-eabisimaltivec*,
2059 powerpc-*-eabisim*, powerpc-*-eabialtivec*, powerpc-*-eabi*,
2060 powerpc-*-rtems*, powerpc-wrs-vxworks, powerpc-wrs-vxworksae,
2061 powerpcle-*-eabisim*, powerpcle-*-eabi*): Add rs6000/e500.h to
2063 * config/rs6000/e500.h: New.
2064 * config/rs6000/eabi.h (TARGET_SPE_ABI, TARGET_SPE, TARGET_E500,
2065 TARGET_ISEL, TARGET_FPRS, TARGET_E500_SINGLE, TARGET_E500_DOUBLE):
2067 * config/rs6000/linuxspe.h (TARGET_SPE_ABI, TARGET_SPE,
2068 TARGET_E500, TARGET_ISEL, TARGET_FPRS, TARGET_E500_SINGLE,
2069 TARGET_E500_DOUBLE): Remove.
2070 * config/rs6000/vxworks.h (TARGET_SPE_ABI, TARGET_SPE,
2071 TARGET_E500, TARGET_ISEL, TARGET_FPRS): Remove.
2072 * config/rs6000/rs6000.h (CHECK_E500_OPTIONS): Define.
2073 * config/rs6000/rs6000.c (rs6000_override_options): Use
2076 2007-01-02 Joseph Myers <joseph@codesourcery.com>
2078 * config/rs6000/rs6000.c (print_operand): Check (TARGET_SPE ||
2079 TARGET_E500_DOUBLE), not TARGET_E500, for %y.
2080 (rs6000_generate_compare, rs6000_emit_sCOND, output_cbranch,
2081 rs6000_emit_cmove): Don't check TARGET_E500.
2082 * config/rs6000/rs6000.md (bunordered, bordered, sunordered,
2083 sordered): Don't check TARGET_E500.
2085 2007-01-01 Eric Christopher <echristo@apple.com>
2087 * config/mips/mips.c (mips_regno_mode_ok_for_base_p): Use
2088 HARD_REGISTER_NUM_P.
2090 2007-01-01 Roger Sayle <roger@eyesopen.com>
2092 * fold-const.c (fold_binary) <EQ_EXPR>: Fold "(X^C1) eq/ne C2" into
2093 "X eq/ne (C1^C2)". Fold "(X^Z) eq/ne (Y^Z)" as "X eq/ne Y" when Z
2094 has no side-effects. Fold "(X^C1) eq/ne (Y^C2)" as "(X^(C1^C2))
2097 2007-01-01 Mike Stump <mrs@apple.com>
2099 * configure.ac: Remove support for building with Apple's gcc-3.1.
2101 2007-01-02 Joseph Myers <joseph@codesourcery.com>
2104 * caller-save.c (add_stored_regs): Only handle SUBREGs if inner
2105 REG is a hard register. Do not modify REG before calling
2107 * rtlanal.c (subreg_get_info): Don't assert size of XMODE is a
2108 multiple of the size of YMODE for certain lowpart cases.
2110 2007-01-01 Andrew Pinski <pinskia@gmail.com>
2113 * gimplify (voidify_wrapper_expr): Update for
2116 2007-01-01 Andreas Schwab <schwab@suse.de>
2119 * config/ia64/ia64.c (ia64_compute_frame_size): Account space for
2120 save of BR0 in extra_spill_size instead of spill_size.
2121 (ia64_expand_prologue): Save BR0 outside of the gr/br/fr spill
2123 (ia64_expand_epilogue): Restore BR0 from its new location.
2125 2007-01-01 Andrew Pinski <pinskia@gmail.com>
2127 * gimplify.c (gimplify_init_constructor <case VECTOR_TYPE>):
2128 Use a temporary variable if the left hand side is not a gimple
2131 2007-01-01 Andrew Pinski <pinskia@gmail.com>
2133 * gimplify.c (gimplify_return_expr): Make the temporary variable
2134 for the return expression, a gimple register variable.
2136 2007-01-01 Jan Hubicka <jh@suse.cz>
2138 * emit-rtl.c (emit_copy_of_insn_after): Do not call copy_insn_1 for
2141 2007-01-01 Mike Stump <mrs@apple.com>
2143 * configure.ac (HAVE_GAS_LITERAL16): Add autoconf check for
2145 * config/darwin.c (machopic_select_rtx_section): Use
2147 (darwin_mergeable_constant_section): Likewise.
2148 * configure: Regenerate.
2149 * config.in: Regenerate.
2151 2007-01-01 Jan Hubicka <jh@suse.cz>
2152 Andrew Pinski <pinskia@gmail.com>
2154 * cgraphunit.c (cgraph_optimize): Call cgraph_add_new_functions
2155 before starting IPA passes.
OSDN Git Service
