1 2007-01-18 Kazu Hirata <kazu@codesourcery.com>
3 * config/m68k/m68k-protos.h (use_return_insn): Rename to...
4 (m68k_use_return_insn): ...this.
5 * config/m68k/m68k.h (USE_RETURN_INSN): Delete.
6 * config/m68k/m68k.c (use_return_insn): Rename to...
7 (m68k_use_return_insn): ...this.
8 * config/m68k/m68k.md (return): Use m68k_use_return_insn instead
11 2007-01-18 Nathan Sidwell <nathan@codesourcery.com>
13 * config/m68k/fpgnulib.c (__truncdfsf2): Implement round to
14 nearest even, fix denormal rounding overflow.
16 2007-01-18 Richard Sandiford <richard@codesourcery.com>
18 * config/m68k/m68k.md (movsf_cf_hard): Use fsmove instead of
20 (movdf_cf_hard): Use fdmove for cases 0 and 3 and fmove for case 1.
21 (extendsfdf2_cf): Use fdmove instead of f%&move.
22 (truncdfsf2_cf): Use fsmove instead of f%$smove.
23 (add<mode>3_cf, sub<mode>3_cf): Use <FP:prec> instead of <FP:round>.
25 2007-01-18 Richard Sandiford <richard@codesourcery.com>
27 * config/m68k/m68k.md (movdf_cf_hard): Use output_move_double for
30 2007-01-18 Richard Sandiford <richard@codesourcery.com>
32 * real.h (real_format): Add a canonical_nan_lsbs_set field.
33 (coldfire_single_format): Declare.
34 (coldfire_double_format): Likewise.
35 * real.c (encode_ieee_single): Use canonical_nan_lsbs_set instead
36 of qnan_msb_set to determine the lower bits of a canonical
38 (encode_ieee_double): Likewise.
39 (encode_ieee_quad): Likewise.
40 (ieee_single_format): Initialize canonical_nan_lsbs_set.
41 (mips_single_format): Likewise.
42 (ieee_double_format): Likewise.
43 (mips_double_format): Likewise.
44 (ieee_extended_motorola_format): Likewise.
45 (ieee_extended_intel_96_format): Likewise.
46 (ieee_extended_intel_128_format): Likewise.
47 (ieee_extended_intel_96_round_53_format): Likewise.
48 (ibm_extended_format): Likewise.
49 (mips_extended_format): Likewise.
50 (ieee_quad_format): Likewise.
51 (mips_quad_format): Likewise.
52 (vax_f_format): Likewise.
53 (vax_d_format): Likewise.
54 (vax_g_format): Likewise.
55 (i370_single_format): Likewise.
56 (i370_double_format): Likewise.
57 (decimal_single_format): Likewise.
58 (decimal_double_format): Likewise.
59 (decimal_quad_format): Likewise.
60 (c4x_single_format): Likewise.
61 (c4x_extended_format): Likewise.
62 (real_internal_format): Likewise.
63 (coldfire_single_format): New real_format.
64 (coldfire_double_format): Likewise.
65 * config/pdp11/pdp11.c (pdp11_f_format): Initialize
66 canonical_nan_lsbs_set.
67 (pdp11_d_format): Likewise.
68 * config/m68k/m68k.c (override_options): Override REAL_FORMAT_MODE
69 if TARGET_COLDFIRE_CPU.
71 2007-01-18 Richard Sandiford <richard@codesourcery.com>
73 * config/m68k/m68k-protos.h (m68k_output_pic_call): Delete.
74 (output_call, m68k_legitimize_call_address): Declare.
75 * config/m68k/m68k.h (EXTRA_CONSTRAINT): Remove unnecessary
76 parenthesees. Add support for a 'W' constraint.
77 (LEGITIMATE_PIC_OPERAND_P): Remove SYMBOL_REF_FLAG handling.
78 (PRINT_OPERAND_PUNCT_VALID_P): Remove comment about 'o'.
79 (m68k_symbolic_call, m68k_symbolic_jump): Declare.
80 * config/m68k/m68k.c (m68k_symbolic_call, m68k_symbolic_jump): New
82 (override_options): Initialize them. Do not set flag_no_function_cse
83 for TARGET_ID_SHARED_LIBRARY.
84 (m68k_output_pic_call): Delete.
85 (m68k_legitimize_call_address): New function.
86 (print_operand): Remove the %o prefix. Handle the %p prefix.
87 (output_call): New function.
88 (m68k_output_mi_thunk): Use m68k_symbolic_jump. Always load the
89 target address from the GOT if symbolic jumps are not allowed.
90 * config/m68k/m68k.md (call, general_operand): Do not set
91 SYMBOL_REF_FLAG. Use m68k_legitimize_call_address instead.
92 Merge separate flag_pic and !flag_pic define_insns into...
93 (*call, *call_value): ...these new patterns. Match the address
94 rather than the containing MEM and require it to be a call_operand.
95 Use output_call to generate the asm template.
96 * config/m68k/predicates.md (const_call_operand): New predicate.
97 (call_operand): Likewise.
99 2007-01-18 Nathan Sidwell <nathan@codesourcery.com>
101 * config/m68k/m68k.h (REGISTER_MOVE_COST): Simplify definition.
102 (STACK_GROWS_DOWNWARD): Define to 1.
103 (FUNCTION_VALUE, LIBCALL_VALUE, FUNCTION_VALUE_REGNO_P): Equivocate
104 comments, emphasizing that these values are only defaults.
105 * config/m68k/linux.h (LINK_SPEC): Fix formatting in #undef.
106 * config/m68k/m68k.c (const_method): Remove trailing whitespace.
108 2007-01-18 Richard Sandiford <richard@codesourcery.com>
110 * config/m68k/m68k.md (cmpsi): Remove outdated flag_pic handling.
112 2007-01-18 Kazu Hirata <kazu@codesourcery.com>
113 Richard Sandiford <richard@codesourcery.com>
115 * config/m68k/m68k.h (DATA_REGNO_P, ADDRESS_REGNO_P, INT_REGNO_P)
116 (FP_REGNO_P): New macros.
117 (REGNO_OK_FOR_INDEX_P, REGNO_OK_FOR_BASE_P, REGNO_OK_FOR_DATA_P)
118 (REGNO_OK_FOR_FP_P, REG_OK_FOR_INDEX_P, REG_OK_FOR_BASE_P): Use them.
119 (EH_RETURN_STACKADJ_RTX): Use A0_REG.
120 * config/m68k/m68k.c (m68k_regno_mode_ok): Use the new REGNO macros.
122 2007-01-18 Nathan Sidwell <nathan@codesourcery.com>
124 * config.gcc (m68k-*-aout*, m68k-*-coff*, m68020-*-elf*, m68k-*-elf*)
125 (m68k-*-uclinuxoldabi, m68k-*-uclinux*, m68k-*-rtems*): Add t-floatlib
127 * config/m68k/t-floatlib: New file, extracting common code from...
128 * config/m68k/t-m68kbare, config/m68k/t-m68kelf,
129 * config/m68k/t-uclinux: Here.
130 * config/m68k/fpgnulib.c: Do not compile extendeed precision
131 routines on ColdFire targets.
133 2007-01-18 Nathan Sidwell <nathan@codesourcery.com>
135 * config.gcc (m68k-*-aout*, m68k-*-coff*, m68020-*-elf*, m68k-*-elf*)
136 (m68010-*-netbsdelf*, m68k*-*-netbsdelf*, m68k*-*-openbsd*)
137 (m68k-*-uclinuxoldabi, m68k-*-uclinux*, m68k-*-linux*)
138 (m68k-*-rtems*): Use tm_file rather than m68k/m68k.h and
139 explicitly set MOTOROLA to 1.
140 * config/m68k/m68k.h (MOTOROLA): Simplify definition accordingly.
142 2007-01-18 Richard Sandiford <richard@codesourcery.com>
143 Nathan Sidwell <nathan@codesourcery.com>
145 * config/m68k/m68k.h (PCC_STATIC_STRUCT_RETURN): Delete.
146 (ASM_OUTPUT_REG_PUSH, ASM_OUTPUT_REG_POP): Add MOTOROLA cases.
147 * config/m68k/coff.h (REGISTER_PREFIX_MD): Delete.
148 * config/m68k/m68020-elf.h (LIB_SPEC): Delete.
149 * config/m68k/m68k-none.h (CC1_SPEC, CPP_SUBTARGET_SPEC): Delete.
150 * config/m68k/m68kelf.h (IMMEDIATE_PREFIX, REGISTER_PREFIX_MD)
151 (ASM_OUTPUT_REG_PUSH, ASM_OUTPUT_REG_POP): Delete.
152 (NO_DOLLAR_IN_LABEL, PCC_STATIC_STRUCT_RETURN): Don't undefine.
153 (BSS_ASM_OP, ASM_OUTPUT_SKIP, ASM_OUTPUT_ASCII): Delete.
154 * config/m68k/m68kemb.h (PCC_STATIC_STRUCT_RETURN): Don't undefine.
155 * config/m68k/linux.h (SIZE_TYPE, PTRDIFF_TYPE, WCHAR_TYPE)
156 (WCHAR_TYPE_SIZE, TARGET_OBJFMT_CPP_BUILTINS): Delete.
157 (TARGET_OS_CPP_BUILTINS): Don't define mc68000 and mc68020 here.
158 (DBX_REGISTER_NUMBER): Delete.
159 * config/m68k/netbsd-elf.h (IMMEDIATE_PREFIX): Delete.
160 (PCC_STATIC_STRUCT_RETURN): Don't undefine.
161 * config/m68k/openbsd.h (PCC_STATIC_STRUCT_RETURN): Define.
163 2007-01-18 Richard Sandiford <richard@codesourcery.com>
165 * config.gcc (m68k-*-uclinux*): Add flat.h to $tm_file.
166 * config/flat.h: New file.
167 * crtstuff.c (USE_PT_GNU_EH_FRAME): Don't define if
169 * config/m68k/m68k.h (ASM_PREFERRED_EH_DATA_FORMAT): Do not use
170 indirect references for -msep-data or -mid-shared-library.
171 Do not use PC-relative code addresses either.
172 * config/m68k/m68k.c (override_options): Restrict -fPIC error
174 * config/m68k/uclinux.h (STARTFILE_SPEC): Define. Use Scrt1.o
175 for shared libraries and crt1.o for executables. Use crti.o and
177 (ENDFILE_SPEC): Use crtend.o and crtn.o.
178 (LIB_SPEC): Suppress -Rlibc.gdb if -static-libc is given.
179 Do not add -elf2flt or -shared-lib-id options here.
180 (LINK_SPEC): Define. Pass -elf2flt if no -elf2flt option is given.
181 Pass -shared-lib-id if -mid-shared-library, taking the library
182 identifier from -mshared-library-id if given, otherwise
184 (EH_FRAME_IN_DATA_SECTION): Do not undefine.
185 (INIT_SECTION_ASM_OP, FINI_SECTION_ASM_OP): Likewise.
186 (TARGET_OS_CPP_BUILTINS): Define __GXX_MERGED_TYPEINFO_NAMES=0
187 and __GXX_TYPEINFO_EQUALITY_INLINE=0 if -mid-shared-library.
188 (DRIVER_SELF_SPECS): Map unadorned PIC options to -msep-data.
189 * config/m68k/t-uclinux (EXTRA_MULTILIB_PARTS): Add crtbegin.o
191 * config/m68k/lb1sf68.asm (PICCALL): Use an lea and pc-relative
192 jump sequence for ISA A and ISA A+.
195 2007-01-18 Richard Sandiford <richard@codesourcery.com>
197 * config.gcc (m68k-*-uclinux*): Base the port on the common
198 and m68k GNU/Linux files rather than on the generic ELF ones.
199 * config/m68k/uclinux.h (TARGET_VERSION): Override.
200 (TARGET_OS_CPP_BUILTINS): Use LINUX_TARGET_OS_CPP_BUILTINS.
202 2007-01-18 Julian Brown <julian@codesourcery.com>
203 Richard Sandiford <richard@codesourcery.com>
205 * config/m68k/m68k.h (LONG_DOUBLE_TYPE_SIZE): Make 64-bit on ColdFire.
206 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
207 * config/m68k/netbsd-elf.h (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Undefine
209 * config/m68k/uclinux-oldabi.h (LONG_DOUBLE_TYPE_SIZE): Redefine to
211 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
213 2007-01-18 Richard Sandiford <richard@codesourcery.com>
215 * doc/install.texi: Document m68k-uclinuxoldabi.
216 * config.gcc (m68k-*-uclinuxoldabi): New configuration.
217 * config/m68k/uclinux-oldabi.h: New file, copied from
218 config/m68k/uclinux.h.
220 2007-01-18 Kaz Kojima <kkojima@rr.iij4u.or.jp>
222 * config/m32r/m32r-protos.h (m32r_expand_epilogue): Declare it.
223 * config/m32r/m32r.c (m32r_setup_incoming_varargs): Use gen_frame_mem.
224 (m32r_compute_frame_size): Use unsigned for regno.
225 (m32r_reload_lr): Use gen_frame_mem.
227 (m32r_output_function_epilogue): Don't output the function epilogue
229 (m32r_expand_epilogue): New.
230 (direct_return): Return false if the function has the interrupt
232 (m32r_hard_regno_rename_ok): Remove code for the textual epilogue.
233 * config/m32r/m32r.md (epilogue): New expander.
234 (return_lr, return_rte): New insns.
235 (return): Make it expander.
236 (return_normal): New expander.
238 2007-01-18 Josh Conner <jconner@apple.com>
241 * config/rs6000/rs6000.c (rs6000_emit_vector_compare): Add
242 support for UNLE, UNLT, UNGE, and UNGT.
244 2007-01-18 Jan Hubicka <jh@suse.cz>
246 * tree-vrp.c (finalize_jump_threads): Do not call cleanup_cfg by hand.
247 * tree-tailcall (add_virtual_phis): Likewise.
248 (optimize_tail_call): Return TODOs.
249 (execute_tail_calls): Return TODOs.
250 * tree-ssa-ccp (execute_fold_all_builtins): Do cleanup_cfg via TODO.
251 * tree-cfgcleanup.c (cleanup_tree_cfg_loop): Return if something
253 * tree-ssa-forwprop.c (tree_ssa_forward_propagate_single_use_value):
254 Cleanup cfg using TODO.
255 * tree-flow.h (cleanup_tree_cfg_loop): Update prototype.
256 * passes.c (execute_function_todo): When cleanup did something, remove
258 * tree-cfg.c (pass_build_cfg): Add cleanup_cfg TODO.
259 (make_edges): Don't cleanup_cfg.
261 2007-01-18 Uros Bizjak <ubizjak@gmail.com>
263 * reg-stack.c (subst_stack_regs_pat) [UNSPEC_SINCOS_COS,
264 UNSPEC_XTRACT_FRACT]: Use generic code for instructions that
265 operate on the top of stack.
266 [UNSPEC_SINCOS_SIN, UNSPEC_XTRACT_EXP, UNSPEC_TAN]: Rewrite
267 register handling of instructions that output to the second
269 [UNSPEC_TAN_ONE, UNSPEC_TAN_TAN]: Remove.
270 (move_for_stack_reg): Special-case check for dead destination
271 stack slot for constant load of 1.0 inside UNSPEC_TAN.
273 * config/i386/i386.md (UNSPEC_TAN): New constant.
274 (UNSPEC_TAN_ONE, UNSPEC_TAN_TAN): Remove.
275 (fptanxf4_i387, fptan_extend<mode>xf4_i387): New patterns
276 to correctly model move of constant 1.0 to top stack slot.
277 (*tandf3_1, *tansf3_1, *tanxf3_1): Remove insn patterns.
278 (unnamed peephole2 pattern): Remove corresponding peephole2
279 pattern that optimizes tan insn and loading of constant 1.0.
280 (tanxf2): Use fptanxf4_i387.
281 (tan<mode>2): Rename from tansf2 and tandf2 and macroize insn
282 patterns using X87MODEF12 mode macro. Use fptan_extend<mode>xf4_i387
283 and truncate result to requested mode. Use SSE_FLOAT_MODE_P to
284 disable patterns for SSE math.
285 (sincos<mode>3): Use truncxf<mode>2_i387_noop for truncation.
286 (fyl2x_extend<mode>xf3_i387): Use X87MODEF12 for operand 1.
288 2007-01-18 Dirk Mueller <dmueller@suse.de>
289 Richard Guenther <rguenther@suse.de>
292 * doc/invoke.texi (Warray-bounds): Document -Warray-bounds.
293 * common.opt (Warray-bounds): Add new warning option.
294 * c-opts.c (c_common_handle_option): Define -Warray-bounds
296 * Makefile.in: make tree-vrp.o depend on toplev.h
297 * tree-vrp.c (vrp_finalize): Call check_array_refs if -Warray-bounds
299 (check_array_refs, check_array_bounds, check_array_ref): New.
301 2007-01-18 Jan Hubicka <jh@suse.cz>
303 * tree-ssa-ccp.c (ccp_finalize): Return if something changed.
304 (execute_ssa_ccp): Return flags conditionally.
305 * tree-ssa-propagate.c (substitue_and_fold): Return if something was
307 * tree-ssa-propagate.h (substitute_and_fold): Update prototype.
309 2007-01-18 Steven Bosscher <steven@gcc.gnu.org>
311 * cfgcleanup.c (cleanup_cfg): Detect cfglayout mode and set
312 the CLEANUP_CFGLAYOUT flag when in cfglayout mode.
314 * Makefile.c (GTFILES): Add cfglayout.h.
315 * gengtype.c (open_base_files): Likewise.
316 * cfglayout.c (cfg_layout_function_footer,
317 cfg_layout_function_header) Reindent to make gengtype happy.
318 * cfglayout.h (cfg_layout_function_footer,
319 cfg_layout_function_header): Add GTY(()) marker.
321 * ifcvt.c (noce_try_sign_mask): Make sure INSN_B is non-null.
323 2007-01-18 Ben Elliston <bje@au.ibm.com>
325 * genautomata.c (write_automata): Include xstrerror output in the
326 error message if writing the DFA description file fails.
328 2007-01-17 H.J. Lu <hongjiu.lu@intel.com>
330 * config/mips/mips-protos.h (mips_output_external): Make it
332 * config/mips/iris.h (TARGET_ASM_EXTERNAL_LIBCALL): Removed.
333 * config/mips/mips.c (irix_output_external_libcall): Likewise.
334 (extern_list): Likewise.
335 (extern_head): Likewise.
336 (TARGET_ASM_FILE_END): Likewise.
337 (mips_file_end): Likewise.
338 (mips_output_external): Rewritten.
340 2007-01-18 Ben Elliston <bje@au.ibm.com>
342 * genpreds.c (write_insn_preds_c): Only write out the function
343 body for regclass_for_constraint if we have register constraints.
345 2007-01-17 Tom Tromey <tromey@redhat.com>
347 * doc/sourcebuild.texi (libgcj Tests): Use sourceware.org.
348 * doc/install.texi (Testing): Use sourceware.org.
349 (Binaries): Likewise.
350 (Specific): Likewise.
351 * doc/contrib.texi (Contributors): Use sourceware.org.
353 2007-01-17 Anatoly Sokolov <aesok@post.ru>
355 * config/avr/avr.h (AVR_HAVE_LPMX): New macro.
356 (AVR_ENHANCED): Rename to ...
357 (AVR_HAVE_MUL): ... new.
358 (avr_enhanced_p): Rename to ...
359 (avr_have_mul_p): ... new.
360 (TARGET_CPU_CPP_BUILTINS): Use 'avr_have_mul_p' instead of
361 'avr_enhanced_p' for "__AVR_ENHANCED__". Define "__AVR_HAVE_MUL__".
362 * config/avr/avr.c (avr_enhanced_p): Rename to ...
363 (avr_have_mul_p): ... new.
364 (base_arch_s): Rename 'enhanced' to 'have_mul'.
365 (avr_override_options): Use 'avr_have_mul_p' and 'have_mul' instead of
366 'avr_enhanced_p' and 'enhanced'.
367 (ashlhi3_out, ashrhi3_out, lshrhi3_out, avr_rtx_costs): Use
368 AVR_HAVE_MUL instead of AVR_ENHANCED.
369 * avr.md (*tablejump_enh): Use AVR_HAVE_LPMX instead of AVR_ENHANCED.
370 (mulqi3, *mulqi3_enh, *mulqi3_call, mulqihi3, umulqihi3, mulhi3,
371 *mulhi3_enh, *mulhi3_call, mulsi3, *mulsi3_call): Use AVR_HAVE_MUL
372 instead of AVR_ENHANCED.
373 (*tablejump_enh): Use AVR_HAVE_LPMX instead of AVR_ENHANCED.
374 * libgcc.S: Use __AVR_HAVE_MUL__ instead of __AVR_ENHANCED__.
375 (__tablejump__): Use __AVR_HAVE_LPMX__ instead of __AVR_ENHANCED__.
377 2007-01-17 Ian Lance Taylor <iant@google.com>
379 * vec.h (VEC_reserve_exact): Define.
380 (vec_gc_p_reserve_exact): Declare.
381 (vec_gc_o_reserve_exact): Declare.
382 (vec_heap_p_reserve_exact): Declare.
383 (vec_heap_o_reserve_exact): Declare.
384 (VEC_OP (T,A,reserve_exact)): New static inline function, three
386 (VEC_OP (T,A,reserve)) [all versions]: Remove handling of
388 (VEC_OP (T,A,alloc)) [all versions]: Call ...reserve_exact.
389 (VEC_OP (T,A,copy)) [all versions]: Likewise.
390 (VEC_OP (T,a,safe_grow)) [all versions]: Likewise.
391 * vec.c (calculate_allocation): Add exact parameter. Change all
393 (vec_gc_o_reserve_1): New static function, from vec_gc_o_reserve.
394 (vec_gc_p_reserve, vec_gc_o_reserve): Call vec_gc_o_reserve_1.
395 (vec_gc_p_reserve_exact, vec_gc_o_reserve_exact): New functions.
396 (vec_heap_o_reserve_1): New static function, from vec_heap_o_reserve.
397 (vec_heap_p_reserve, vec_heap_o_reserve): Call vec_heap_o_reserve_1.
398 (vec_heap_p_reserve_exact): New function.
399 (vec_heap_o_reserve_exact): New function.
401 2007-01-17 Jan Hubicka <jh@suse.cz>
403 * ipa-type-escape.c (look_for_casts): Revamp using handled_component_p.
405 2007-01-17 Eric Christopher <echristo@apple.com>
407 * config.gcc: Support core2 processor.
409 2007-01-16 Jan Hubicka <jh@suse.cz>
411 * tree-ssanames.c (release_dead_ssa_names): Instead of ggc_freeing
412 the names, just unlink the chain so we don't crash on dangling pointers
415 2007-01-16 Jan Hubicka <jh@suse.cz>
417 * cgraph.h (cgraph_decide_inlining_incrementally): Kill.
418 * tree-pass.h: Reorder to make IPA passes appear toegher.
419 (pass_early_inline, pass_inline_parameters, pass_apply_inline): Declare.
420 * cgraphunit.c (cgraph_finalize_function): Do not compute inling
421 parameters, do not call early inliner.
422 * ipa-inline.c: Update comments. Include tree-flow.h
423 (cgraph_decide_inlining): Do not compute inlining parameters.
424 (cgraph_decide_inlining_incrementally): Return TODOs; assume to
425 be called with function context set up.
426 (pass_ipa_inline): Remove unreachable functions before pass.
427 (cgraph_early_inlining): Simplify assuming to be called from the
429 (pass_early_inline): New pass.
430 (cgraph_gate_ipa_early_inlining): New gate.
431 (pass_ipa_early_inline): Turn into simple wrapper.
432 (compute_inline_parameters): New function.
433 (gate_inline_passes): New gate.
434 (pass_inline_parameters): New pass.
435 (apply_inline): Move here from tree-optimize.c
436 (pass_apply_inline): New pass.
437 * ipa.c (cgraph_remove_unreachable_nodes): Verify cgraph after
439 * tree-inline.c (optimize_inline_calls): Return TODOs rather than
441 (tree_function_versioning): Do not allocate dummy struct function.
442 * tree-inline.h (optimize_inline_calls): Update prototype.
443 * tree-optimize.c (execute_fixup_cfg): Export.
444 (pass_fixup_cfg): Remove
445 (tree_rest_of_compilation): Do not apply inlines.
446 * tree-flow.h (execute_fixup_cfg): Declare.
447 * Makefile.in (gt-passes.c): New.
448 * passes.c: Include gt-passes.h
449 (init_optimization_passes): New passes.
450 (nnodes, order): New static vars.
451 (do_per_function_toporder): New function.
452 (execute_one_pass): Dump current pass here.
453 (execute_ipa_pass_list): Don't dump current pass here.
455 2007-01-16 Janis Johnson <janis187@us.ibm.com>
457 * config/dfp-bit.c (dfp_compare_op): Return separate value for NaN.
458 (DFP_NE, DFP_LE, DFP_GE): Return false for NaN.
460 2007-01-16 David Edelsohn <edelsohn@gnu.org>
462 * config/rs6000/darwin-ldouble.c: Build file for SOFT_FLOAT.
463 (strong_alias): Define.
464 (__gcc_qmul): Provide non-FMA for soft-float.
480 * config/rs6000/rs6000.c (rs6000_init_libfuncs): Initialize
481 soft-float functions.
482 * config/rs6000/libgcc-ppc-glibc.ver: Version soft-float symbols.
483 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Do not warn
484 about long double soft float.
486 2007-01-16 Dorit Nuzman <dorit@il.ibm.com>
487 Tehila Meyzels <tehila@il.ibm.com>
489 * tree-vectorizer.h (is_pattern_stmt_p): New.
490 * tree-vect-analyze.c (vect_determine_vectorization_factor): Fix
491 formatting (tabs instead of spaces). Cleanup and clarify setting
492 of STMT_VINFO_VECTYPE. Call is_pattern_stmt_p.
493 * tree-vect-transform.c (vect_get_vec_def_for_operand): Fix typo.
494 (vectorizable_type_demotion): Check that types are integral.
495 (vectorizable_type_promotion): Likewise.
496 (vectorizable_store): Fix typo. Eliminate new-line at end of
499 2007-01-16 Jan Hubicka <jh@suse.cz>
501 * tree-ssanames.c (release_dead_ssa_names): Remove invalidated
504 2007-01-15 Eric Christopher <echristo@apple.com>
506 * ifcvt.c: Include vec.h, vecprim.h.
507 (check_cond_move_block): New argument regs.
508 Reorganize. Add registers used to regs.
509 (cond_move_process_if_block): Use regs set above as
512 2007-01-15 Eric Christopher <echristo@apple.com>
514 * config/darwin.h: Update copyright.
515 (TARGET_OPTION_TRANSLATE_TABLE): Add umbrella.
516 (LINK_COMMAND_SPEC): Add -u.
517 (LINK_SPEC): Fix umbrella for above.
519 2007-01-15 Joseph S. Myers <joseph@codesourcery.com>
521 * config/soft-fp/op-common.h, config/soft-fp/op-4.h: Update from
524 2007-01-15 Tom Tromey <tromey@redhat.com>
526 * doc/sourcebuild.texi (libgcj Tests): Don't mention jacks.
527 * doc/install.texi (Testing): Don't mention jacks.
528 (Configuration): Document --enable-java-maintainer-mode. Move
529 --with-java-home to libgcj-specific section. Document
531 (Prerequisites): Mention --enable-java-maintainer-mode, ecj1.
533 2007-01-15 Jan Hubicka <jh@suse.cz>
535 * tree-ssa-dce.c (DCE_TODOs): New.
536 (propagate_necessity): Return if something changed.
537 (eliminate_unnecessary_stmts): Likewise.
538 (perform_tree_ssa_dce): Return TODO flags when needed.
539 (pass_dce, pass_dce_loop, pass_cd_dce): Remove TODO flags.
541 2007-01-15 Uros Bizjak <ubizjak@gmail.com>
543 * config/i386/i386.md (fyl2xxf3_i387): Rename from fyl2x_xf3.
544 (fyl2x_extend<mode>xf3_i387): New insn pattern.
545 (log<mode>2): Rename from logsf2 and logdf2 and macroize insn
546 patterns using X87MODEF12 mode macro. Extend operand 1
547 to XFmode. Use SSE_FLOAT_MODE_P to disable patterns for SSE math.
548 (log10<mode>2): Ditto.
549 (log2<mode>2): Ditto.
550 (log1p<mode>2): Ditto.
551 (logb<mode>2): Ditto.
552 (fyl2xp1xf3_i387): Rename from fyl2xp1_xf3.
553 (fyl2xp1_extend<mode>xf3_i387): New insn pattern.
554 (*fxtractxf3_i387): Rename from *fxtractxf3.
555 (fxtract_extend<mode>xf3_i387): New insn pattern.
556 (ilogbsi2): Use match_dup 3, not match_operand:XF 3.
557 * config/i386/i386.c (ix86_emit_i387_log1p): Use gen_fyl2xp1xf3_i387()
558 and gen_fyl2xxf3_i387().
560 2007-01-14 Zdenek Dvorak <dvorakz@suse.cz>
562 * loop-unswitch.c (unswitch_loop): Do not call fix_loop_placement.
563 * cfgloopmanip.c (fix_loop_placement): Made static. Use
564 get_loop_exit_edges. Changed return type to bool.
565 * cfgloop.h (fix_loop_placement): Declaration removed.
567 2007-01-14 Dorit Nuzman <dorit@il.ibm.com>
569 * param.h (MIN_VECT_LOOP_BOUND): New.
570 * params.def (MIN_VECT_LOOP_BOUND): New.
571 * tree-vectorizer.c (slpeel_tree_peel_loop_to_edge): Takes another
572 argument - minimum threshold for number of iterations.
573 * tree-vectorizer.h (slpeel_tree_peel_loop_to_edge): Add another
574 argument to declaration.
575 * tree-vect-analyze.c (vect_analyze_operations): Check value of
577 * tree-vect-transform.c (vect_do_peeling_for_loop_bound): Call
578 slpeel_tree_peel_loop_to_edge with additional argument.
579 (vect_do_peeling_for_alignment): Likewise.
580 * doc/invoke.texi (min-vect-loop-bound): Document new param option.
582 2007-01-14 Uros Bizjak <ubizjak@gmail.com>
585 * config/i386/i386.c (print_operand) ['z']: Output 'b' for
588 2007-01-14 Jan Hubicka <jh@suse.cz>
590 * tree-dfa.c (remove_referenced_var): New function.
591 * tree-ssa-live.c (remove_unused_locals): Walk referenced vars and
592 prune referenced vars list too.
593 * tree-flow.h (remove_referenced_var): Declare.
595 2007-01-14 Jan Hubicka <jh@suse.cz>
597 * tree-eh.c (add_stmt_to_eh_region_fn): Do not add call_exprs
599 (remove_stmt_from_eh_region_fn): Do not remove call_exprs.
600 (verify_eh_throw_stmt_node, verify_eh_throw_table_statements): Kill.
601 * except.h (verify_eh_throw_table_statements): Kill prototype.
602 * cfgexpand.c (expand_gimple_basic_block): Propagate Eh regions
603 into call exrepssions.
604 * tree-optimize.c (execute_free_cfg_annotatiosn): Do not call
606 * tree-cfg.c: Include pointer-set.h.
607 (verify_node_sharing): Work on pointer set.
608 (verify_eh_throw_stmt_node): New.
609 (verify_stmts): Use pointers sets, verify throw_stmt.
611 2007-01-13 Zdenek Dvorak <dvorakz@suse.cz>
613 * ipa-reference.c (analyze_function): Consider also addresses taken
616 2007-01-12 Roger Sayle <roger@eyesopen.com>
618 * c-typeck.c (null_pointer_constant_p): Replace use of
619 TREE_CONSTANT_OVERFLOW with TREE_OVERFLOW.
620 (build_c_cast): Likewise.
622 2007-01-12 Roger Sayle <roger@eyesopen.com>
624 * tree.h (force_fit_type_double): Remove unused final argument.
625 * c-common.c (constant_expression_warning): Replace use of
626 TREE_CONSTANT_OVERFLOW with TREE_OVERFLOW.
627 (convert_and_check): Likewise.
628 (shorten_compare): Update call to force_fit_type_double.
629 (c_common_truthvalue_conversion) <INTEGER_CST>: Use integer_zerop.
630 * convert.c (convert_to_pointer): Update call to
631 force_fit_type_double.
632 * fold-const.c (force_fit_type_double): Remove overflowed_const
634 (int_const_binop, fold_convert_const_int_from_int,
635 fold_convert_const_int_from_real, fold_div_compare,
636 fold_sign_changed_comparison, fold_unary, fold_negate_const,
637 fold_abs_const, fold_not_const): Remove the final argument from
638 calls to force_fit_type_double.
640 2007-01-12 Andrew Pinski <andrew_pinski@playstation.sony.com>
642 * configure.ac: Set insn to "nop" for spu-*-* also.
643 * configure: Regenerate.
645 2007-01-12 Olga Golovanevsky <olga@il.ibm.com>
647 * builtins.def : Add BUILT_IN_FREE.
649 2007-01-12 Jan Hubicka <jh@suse.cz>
651 PR tree-optimization/30443
652 * tree-inline.c (tree_function_versioning): Do not optimize when
653 cloning for inlining.
655 2007-01-12 Zdenek Dvorak <dvorakz@suse.cz>
657 * doc/loop.texi: Document recording of loop exits.
658 * cfgloopmanip.c (loopify, duplicate_loop): Use alloc_loop.
659 (update_single_exits_after_duplication,
660 update_single_exit_for_duplicated_loop,
661 update_single_exit_for_duplicated_loops): Removed.
662 (duplicate_loop_to_header_edge): Do not call
663 update_single_exits_after_duplication and
664 update_single_exit_for_duplicated_loops.
665 (loop_version): Do not update single_exit information.
666 (fix_loop_structure): Use record_loop_exits instead of
667 mark_single_exit_loops.
668 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update
669 the lists of loop exits.
670 * cfghooks.c (redirect_edge_and_branch, redirect_edge_and_branch_force,
671 split_edge, merge_blocks): Update the lists of loop exits.
672 * modulo-sched.c (sms_schedule): Pass LOOPS_HAVE_RECORDED_EXITS to
674 * loop-init.c (loop_optimizer_init): Call record_loop_exits instead
675 of mark_single_exit_loops.
676 (loop_optimizer_finalize): Call release_recorded_exits.
677 * tree-ssa-loop.c (tree_loop_optimizer_init): Pass
678 LOOPS_HAVE_RECORDED_EXITS to loop_optimizer_init.
679 * tree-vectorizer.c (slpeel_tree_duplicate_loop_to_edge_cfg): Do not
680 update single exit information.
681 * lambda-code.c (perfect_nestify): Ditto.
682 * cfgloop.c (flow_loop_free): Destroy the list of exits of the loop.
683 (mark_single_exit_loops): Removed.
684 (alloc_loop, loop_exit_hash, loop_exit_eq, loop_exit_free,
685 get_exit_descriptions, rescan_loop_exit, record_loop_exits,
686 dump_recorded_exit, dump_recorded_exits, release_recorded_exits): New
688 (get_loop_exit_edges, single_exit): Use recorded exit lists.
689 (add_bb_to_loop, remove_bb_from_loops): Update the lists of loop exits.
690 (verify_loop_structure): Verify consistency of the exit lists.
691 (flow_loops_find): Use alloc_loop. Initialize exits hash.
692 (set_single_exit): Removed.
693 * cfgloop.h (struct loop_exit): New function.
694 (struct loop): single_exit_ field replaced by exits field.
695 (LOOPS_HAVE_MARKED_SINGLE_EXITS): Replaced by LOOPS_HAVE_RECORDED_EXITS.
696 (struct loops): Added exits hash.
697 (mark_single_exit_loops, set_single_exit): Declaration removed.
698 (release_recorded_exits, record_loop_exits, rescan_loop_exit): Declare.
700 2007-01-12 Richard Sandiford <richard@codesourcery.com>
702 * doc/invoke.texi: Avoid use of @headitem.
704 2007-01-12 Richard Sandiford <richard@codesourcery.com>
706 * cse.c (cse_insn): Move HAVE_CC0 code after declarations.
708 2007-01-12 Richard Sandiford <richard@codesourcery.com>
710 * doc/install.texi: Fix m68k-*-* anchor and add m68k-*-* to the
713 2007-01-12 Nathan Sidwell <nathan@codesourcery.com>
714 Richard Sandiford <richard@codesourcery.com>
716 * doc/invoke.texi: Document -mno-bitfield, -mno-rtd and -mno-short.
717 * config/m68k/m68k.opt: Resort options.
718 (mbitfield, mrtd, mshort): Remove RejectNegative properties.
720 2007-01-12 Nathan Sidwell <nathan@codesourcery.com>
721 Richard Sandiford <richard@codesourcery.com>
723 * doc/invoke.texi: Document the macros that are defined by
724 m68k's -mtune and -mhard-float options.
725 * config/m68k/m68k-protos.h (m68k_cpp_cpu_ident) Declare.
726 (m68k_cpp_cpu_family): Likewise.
727 * config/m68k/m68k.h (TARGET_CPU_CPP_BUILTINS): Add a full set
728 of __ucfv*__ macros. Define __mcffpu__ if generating code for
729 ColdFire FPUs. Define __mcf_cpu_* and __mcf_family_* macros.
730 * config/m68k/m68k.c (m68k_cpp_cpu_ident): New function.
731 (m68k_cpp_cpu_family): Likewise.
733 2007-01-12 Richard Sandiford <richard@codesourcery.com>
735 * config/m68k/m68k.h (TARGET_CPU_CPP_BUILTINS): Treat all mc68*
736 macros besides mc68000 as tuning macros. Use a switch statement
737 to set them and mcpu32.
739 2007-01-12 Julian Brown <julian@codesourcery.com>
741 * config/m68k/m68k.h: Use TARGET_68040 instead of TARGET_68040_ONLY.
742 (TARGET_68040_ONLY): Rename to...
743 (TARGET_68040): ...this.
744 * config/m68k/m68k.c: Use TARGET_68040 instead of TARGET_68040_ONLY.
745 * config/m68k/m68k.md: Likewise.
747 2007-01-12 Julian Brown <julian@codesourcery.com>
748 Nathan Sidwell <nathan@codesourcery.com>
749 Richard Sandiford <richard@codesourcery.com>
751 * config.gcc (m680[012]0-*-*, m68k*-*-*): Set m68k_cpu_ident to
752 the -mcpu= argument associated with the --with-cpu setting.
753 Define M68K_DEFAULT_TUNE to the default -mtune= option,
754 if different from the one implied by the -mcpu setting.
755 Accept --with-cpu=FOO if FOO is listed in m68k-devices.def,
756 using mcpu=FOO as the default CPU option. Set target_cpu_default2.
757 * doc/invoke.texi: Mention ColdFire in the introduction to the
758 m68k options. Document the new -march, -mcpu, -mtune, -mdiv,
759 -mno-div and -mhard-float options. Make -m68881 a synonym for
760 -mhard-float. Document the previously-undocumented -m5206e,
761 -m528x, -m5307 and -m5407 options. Tweak the existing option
762 documentation for consistency.
763 * doc/install.texi: Mention new --with-cpu arguments.
764 * config/m68k/m68k.h (OPTION_DEFAULT_SPECS): Only use the
765 default CPU if neither -mcpu nor -march are specified.
766 (ASM_CPU_SPEC): Pass down -mcpu and -march options.
767 (TARGET_CPU_CPP_BUILTINS): Set __mcfisa*__ macros from
768 TARGET_ISA*. Set the legacy __mcf*__ cpu macros in the same way,
769 using m68k_tune to decide between families that implement the
770 same ISA. Use m68k_tune to set __mcfv4e__.
771 (FL_BITFIELD, FL_68881, FL_COLDFIRE, FL_CF_HWDIV, FL_CF_MAC)
772 (FL_CF_EMAC, FL_CF_EMAC_B, FL_CF_USP, FL_CF_FPU, FL_ISA_68000)
773 (FL_ISA_68010, FL_ISA_68020, FL_ISA_68040, FL_ISA_A, FL_ISA_B)
774 (FL_ISA_C, FL_ISA_MMU): New macros.
775 (MASK_COLDFIRE): Delete.
776 (TARGET_68010, TARGET_68020, TARGET_68040_ONLY, TARGET_COLDFIRE)
777 (TARGET_ISAB): Redefine in terms of m68k_cpu_flags.
778 (TARGET_68881, TARGET_COLDFIRE_FPU): Redefine in terms of m68k_fpu.
779 (TARGET_HARD_FLOAT): Do not define here.
780 (TARGET_ISAAPLUS, TARGET_ISAC): New macros.
781 (TUNE_68000): New macro.
782 (TUNE_68000_10): Redefine in terms of TUNE_68000 and TUNE_68010.
783 (TUNE_68010, TUNE_68030, TUNE_68040, TUNE_68060, TUNE_CPU32)
784 (TUNE_CFV2): Redefine in terms of m68k_tune.
785 (uarch_type, target_device, fpu_type): New enums.
786 (m68k_cpu, m68k_tune, m68k_fpu, m68k_cpu_flags): Declare.
787 * config/m68k/m68k.c (TARGET_DEFAULT): Remove MASK_68881.
788 (FL_FOR_isa_00, FL_FOR_isa_10, FL_FOR_isa_20, FL_FOR_isa_40)
789 (FL_FOR_isa_cpu32, FL_FOR_isa_a, FL_FOR_isa_aplus, FL_FOR_isa_b)
790 (FL_FOR_isa_c): New macros.
791 (m68k_isa): New enum.
792 (m68k_target_selection): New structure.
793 (all_devices, all_isas, all_microarchs): New tables.
794 (m68k_cpu_entry, m68k_arch_entry, m68k_tune_entry, m68k_cpu)
795 (m68k_tune, m68k_fpu, m68k_cpu_flags): New variables.
796 (MASK_ALL_CPU_BITS): Delete.
797 (m68k_find_selection): New function.
798 (m68k_handle_option): Handle -mcpu=, -march= and -mtune=.
799 Map the legacy target options to a combination of the new ones.
800 (override_options): Set m68k_cpu, m68k_tune, m68k_fpu and
801 m68k_cpu_flags. Handle M68K_DEFAULT_TUNE. Use m68k_cpu_flags
802 to derive default MASK_BITFIELD, MASK_CF_HWDIV and MASK_HARD_FLOAT
804 * config/m68k/m68k.opt (m5200, m5206e, m528x, m5307, m5407, mcfv4e)
805 (m68010, m68020, m68020-40, m68020-60, m68030, m68040): Remove Mask
807 (m68881, msoft-float): Change mask from 68881 to HARD_FLOAT.
808 (march=, mcpu=, mdiv, mhard-float, mtune=): New options.
809 * config/m68k/m68k-devices.def: New file.
811 2007-01-12 Richard Sandiford <richard@codesourcery.com>
812 Nathan Sidwell <nathan@codesourcery.com>
814 * config/m68k/m68k.h (ASM_CPU_SPEC, ASM_SPEC, EXTRA_SPECS)
815 (SUBTARGET_EXTRA_SPECS): New macros.
816 * config/m68k/linux.h (ASM_SPEC): Remove CPU flags;
817 use %(asm_cpu_spec) instead.
818 * config/m68k/m68k-none.h (ASM_SPEC): Likewise.
819 * config/m68k/openbsd.h (ASM_SPEC): Likewise.
820 * config/m68k/netbsd-elf.h (ASM_SPEC): Likewise.
821 (EXTRA_SPECS): Rename to...
822 (SUBTARGET_EXTRA_SPECS): ...this.
824 2007-01-12 Nathan Sidwell <nathan@codesourcery.com>
825 Richard Sandiford <richard@codesourcery.com>
826 Julian Brown <julian@codesourcery.com>
828 * config.gcc (m68k-*-aout*, m68k-*-coff*, m68020-*-elf*, m68k-*-elf*)
829 (m68k-*-uclinux*, m68k-*-linux*, m68k-*-rtems*): Set default_m68k_cpu
830 to the configuration's default CPU.
831 (m68010-*-netbsdelf*, m68k*-*-netbsdelf*, m68k*-*-openbsd*): Likewise.
832 Remove default masks.
833 (m680[012]0-*-*): Set the default with_cpu to the first part of
835 (m68k*-*-*): Set the default with_cpu to m$default_m68k_cpu.
836 (m68k*-*-linux): Extend the --with-cpu handling to...
837 (m680[012]0-*-*, m68k*-*-*): ...these configurations. Allow m68000
838 and m68010. Don't set target_cpu_default2.
839 * doc/install.texi: Document --with-cpu for m68k.
840 * config/m68k/m68k.h (OPTION_DEFAULT_SPECS): Define.
841 * config/m68k/m68k-none.h (TARGET_CPU_DEFAULT, M68K_CPU_m68k)
842 (M68K_CPU_m68000, M68K_CPU_m68010, M68K_CPU_m68020, M68K_CPU_m68030)
843 (M68K_CPU_m68040, M68K_CPU_m68302, M68K_CPU_m68332, TARGET_DEFAULT)
844 (ASM_CPU_DEFAULT_SPEC, CC1_CPU_DEFAULT_SPEC): Delete.
845 (ASM_SPEC): Remove use of %(asm_cpu_default).
846 (EXTRA_SPECS, SUBTARGET_EXTRA_SPECS, MULTILIB_DEFAULTS): Delete.
847 * config/m68k/linux.h (TARGET_DEFAULT): Delete.
848 (CPP_SPEC): Merge definitions. Do not handle __HAVE_68881__ here.
849 * config/m68k/netbsd-elf.h (TARGET_OS_CPP_BUILTINS): Define
850 __HAVE_FPU__ if TARGET_HARD_FLOAT.
851 (TARGET_DEFAULT): Delete.
852 (EXTRA_SPECS): Delete cpp_cpu_default_spec, cpp_cpu_spec,
853 cpp_fpu_spec, asm_default_spec and netbsd_cpp_spec.
854 (CPP_CPU_SPEC): Delete.
855 (TARGET_VERSION): Merge definitions, using TARGET_68010 to pick
856 the appropriate string.
857 (CPP_CPU_DEFAULT_SPEC, ASM_DEFAULT_SPEC, CPP_FPU_SPEC): Delete.
858 (CPP_SPEC): Define to NETBSD_CPP_SPEC.
859 (ASM_SPEC): Don't use %(asm_default_spec).
860 * config/m68k/m68k.c (TARGET_DEFAULT_TARGET_FLAGS): Remove
861 TARGET_DEFAULT and add MASK_68881.
862 * config/m68k/m68k.md: Remove mention of TARGET_DEFAULT from comments.
864 2007-01-12 Richard Sandiford <richard@codesourcery.com>
866 * config.gcc (m68010-*-netbsdelf*): Add MASK_68010.
867 (m68k*-*-netbsdelf*, m68k*-*-openbsd*, m68k*-linux*): Add
868 MASK_68010 alongside MASK_68020.
869 * doc/invoke.texi: Document -m68010.
870 * config/m68k/m68k.opt (m68010): New.
871 * config/m68k/m68k.h (TARGET_CPU_CPP_BUILTINS): Define mc68010
873 (TUNE_68010): New macro.
874 * config/m68k/m68k-none.h (M68K_CPU_m68k, M68K_CPU_m68010)
875 (M68K_CPU_m68020, M68K_CPU_m68030, M68K_CPU_m68040)
876 (M68K_CPU_m68332): Add MASK_68010.
877 * config/m68k/linux.h (TARGET_DEFAULT): Add MASK_68010 to
879 * config/m68k/netbsd-elf.h (CPP_CPU_SPEC): Remove now-redundant
881 * config/m68k/m68k.c (MASK_ALL_CPU_BITS): Add MASK_68010.
882 (m68k_handle_option): Handle OPT_m68010. Add MASK_68010
883 to all entries that use MASK_68020.
884 (output_move_simode_const, output_move_himode, output_move_qimode)
885 (output_move_stricthi, output_move_strictqi): Use TARGET_68010
886 instead of TARGET_68020 to select clr behavior. Remove comment
887 about there being no TARGET_68010.
888 * config/m68k/m68k.md: Likewise throughout.
890 2007-01-12 Julian Brown <julian@codesourcery.com>
892 * config/m68k/m68k.h (TARGET_ISAB): New macro.
893 * config/m68k/m68k.c: Use TARGET_ISAB rather than TARGET_CFV4.
894 * config/m68k/m68k.md: Likewise.
896 2007-01-12 Julian Brown <julian@codesourcery.com>
898 * config/m68k/m68k.h (LEGITIMATE_INDEX_P, LEGITIMIZE_ADDRESS): Use
899 TARGET_COLDFIRE_FPU instead of TARGET_CFV4E.
901 2007-01-12 Julian Brown <julian@codesourcery.com>
903 * config/m68k/m68k.h (TUNE_68040_60): New macro.
904 * config/m68k/m68k.c (standard_68881_constant_p): Use it.
905 * config/m68k/m68k.md: Likewise.
907 2007-01-12 Julian Brown <julian@codesourcery.com>
908 Richard Sandiford <richard@codesourcery.com>
910 * config/m68k/m68k.h (TARGET_CPU_CPP_BUILTINS): Use TUNE_68030
911 instead of TARGET_68030, TUNE_68040 instead of TARGET_68040,
912 TUNE_68060 instead of TARGET_68060 and TUNE_CPU32 instead of
914 (TARGET_CPU32): Rename to...
915 (TUNE_CPU32): ...this.
916 (TUNE_68000_10, TUNE_68030, TUNE_68040, TUNE_68060)
917 (TUNE_CFV2): New macros.
918 * config/m68k/netbsd-elf.h (LONG_DOUBLE_TYPE_SIZE): Simplify;
919 remove conditions that are implied by TARGET_68020.
920 * config/m68k/m68k.c (m68k_output_function_prologue): Use TUNE_68040
921 instead of TARGET_68040 and TUNE_CPU32 instead of TARGET_CPU32.
922 (m68k_output_function_epilogue): Likewise.
923 (m68k_rtx_costs): Likewise. Use TUNE_68060 instead of TARGET_68060
924 and TUNE_CFV2 instead of TARGET_5200. Use TUNE_68000_10 instead of
925 "!TARGET_68020 && !TARGET_COLDFIRE" to choose between 68000 and
926 non-68000 timings. Refactor multiplication and division costs.
927 (output_addsi3): Use TUNE_68040 instead of TARGET_68040 and
928 TUNE_CPU32 instead of TARGET_CPU32.
929 (standard_68881_constant_p): Use TUNE_68040 instead of TARGET_68040
930 and TUNE_68060 instead of TARGET_68060.
931 * config/m68k/m68k.md: Use TUNE_68040 instead of TARGET_68040,
932 TUNE_68060 instead of TARGET_68060, and TUNE_CPU32 instead of
934 (movsi_const0): Use TUNE_68000_10 rather than "!TARGET_68020
935 && !TARGET_COLDFIRE" to choose between moveq and clr.
936 Likewise in the unnamed movsf pattern.
937 (ashlsi_17_24, lshrsi_17_24): Guard with TUNE_68000_10 rather than
938 "!TARGET_68020 && !TARGET_COLDFIRE". Likewise the unnamed
941 2007-01-12 Richard Sandiford <richard@codesourcery.com>
943 * config/m68k/m68k.h (TARGET_CPU_CPP_BUILTINS): Increase amount
944 of tabbing before backslashes.
946 2007-01-11 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
948 * pa-linux.h (ASM_OUTPUT_INTERNAL_LABEL): Undefine.
949 * pa.h (ASM_OUTPUT_LABEL): Output colon when using GAS.
950 (ASM_OUTPUT_INTERNAL_LABEL): Define.
952 2007-01-11 Zdenek Dvorak <dvorakz@suse.cz>
954 * tree-ssa-loop-ivopts.c (extract_cond_operands): Split from
955 find_interesting_uses_cond.
956 (find_interesting_uses_cond): Use extract_cond_operands.
957 (rewrite_use_compare): Use extract_cond_operands and
958 force_gimple_operand_bsi. Do not call update_stmt.
959 (determine_use_iv_cost_condition): Use extract_cond_operands.
960 Return cheaper of using original bound and changing the exit bound.
962 2007-01-11 Zdenek Dvorak <dvorakz@suse.cz>
964 PR tree-optimization/29516
965 * tree-ssa-address.c (tree_mem_ref_addr, add_to_parts,
966 most_expensive_mult_to_index, addr_to_parts,
967 create_mem_ref, maybe_fold_tmr): Make the type of
968 fields of TARGET_MEM_REF sizetype.
969 (move_fixed_address_to_symbol, move_pointer_to_base):
971 * tree.def (TARGET_MEM_REF): Add comment on types of
974 2007-01-11 Joseph Myers <joseph@codesourcery.com>
976 * c-common.c (vector_types_convertible_p): Treat opaque types as
977 always convertible if they have the same size, but not otherwise.
979 2007-01-11 Steven Bosscher <steven@gcc.gnu.org>
981 * ifcvt.c (struct noce_if_info): Add comments to the fields.
982 Remove the b_unconditional field.
983 (noce_try_sign_mask): Do not look at b_unconditional.
984 (noce_process_if_block): Do not use merge_if_blocks. Update
985 the CFG here. Do not set b_unconditional.
986 (cond_move_process_if_block): Likewise.
987 (find_cond_trap): Likewise.
988 (check_cond_move_block): Require simple jump insns at the end
991 2007-01-11 Jan Hubicka <jh@suse.cz>
993 PR tree-optimization/1046
994 * tree-tailcall.c (suitable_for_tail_call_opt_p): Use TREE_ADDRESSABLE
995 when alias info is not ready.
996 (pass_tail_recursion): Do not require aliasing.
997 * tree-ssa-copyrename.c (pass_rename_ssa_cop): Likewise.
998 * tree-ssa-ccp.c (pass_ccp, pass_fold_builtins): Likewise.
999 * tree-ssa-copy.c (pass_copy_prop): Likewise.
1000 * tree-ssa-forwprop.c (pass_forwprop): Likewise.
1001 * tree-ssa-dce.c (pass_dce, pass_dce_loop, pass_cd_dce): Likewise.
1002 * passes.c (init_optimization_passes): Execute rename_ssa_copies,
1003 ccp, forwprop, copy_prop, merge_phi, copy_prop, dce and tail recursion
1005 * tree-ssa-operands.c (add_virtual_operand, get_indirect_ref_operand):
1006 When aliasing is not build, mark statement as volatile.
1008 2007-01-11 Tom Tromey <tromey@redhat.com>
1010 PR preprocessor/15185, PR preprocessor/20989:
1011 * doc/cppopts.texi <-MT>: Update description of algorithm for
1012 computing default target.
1013 <-M, -MD>: Reword "basename" text.
1015 2007-01-11 Roger Sayle <roger@eyesopen.com>
1017 * builtins.c (expand_builtin_pow, expand_builtin_powi,
1018 fold_builtin_cabs, fold_builtin_sqrt, fold_builtin_trunc,
1019 fold_builtin_floor, fold_builtin_ceil, fold_builtin_round,
1020 fold_builtin_int_int_roundingfn, fold_builtin_bitop,
1021 fold_builtin_bswap, real_constp, fold_builtin_pow,
1022 fold_builtin_powi, fold_builtin_signbit, fold_builtin_copysign,
1023 do_mpfr_arg1, do_mpfr_arg2, do_mpfr_arg3, do_mpfr_sincos): Replace
1024 uses of the macro TREE_CONSTANT_OVERFLOW with TREE_OVERFLOW.
1025 * convert.c (convert_to_pointer): Likewise.
1026 * expr.c (highest_pow2_factor, expand_expr_real_1): Likewise.
1027 * fold-const.c (force_fit_type, fold_negate_expr, int_const_binop,
1028 const_binop, fold_convert_const_int_from_int,
1029 fold_convert_const_int_from_real,
1030 fold_convert_const_real_from_real, sign_bit_p,
1031 optimize_minmax_comparison, extract_muldiv_1, fold_div_compare,
1032 fold_sign_changed_comparison, fold_unary, fold_comparison,
1033 fold_binary, multiple_of_p, tree_Expr_non_zero_p,
1034 fold_negate_const, fold_abs_const, fold_not_const): Likewise.
1035 * print-tree.c (print_node_brief, print_node): Likewise.
1036 * stor-layout.c (place_field, layout_type): Likewise.
1037 * tree-chrec.c (keep_cast): Likewise.
1038 * tree.c (build_vector, build_real, build_real_from_int_cst,
1039 build_complex): Likewise.
1041 2007-01-11 Roger Sayle <roger@eyesopen.com>
1043 * tree.h (TREE_CONSTANT_OVERFLOW): Obsolete. For the time being,
1044 treat TREE_CONSTANT_OVERFLOW as a synonym of TREE_OVERFLOW.
1046 2007-01-11 Paolo Bonzini <bonzini@gnu.org>
1048 * configure.ac (strict1_warn): Rename to strict_warn.
1049 (WERROR, --enable-werror, symlink hacks, stage1_cflags,
1050 cc_set_by_configure, quoted_cc_set_by_configure,
1051 stage_prefix_set_by_configure, quoted_stage_prefix_set_by_configure,
1052 all_boot_languages, all_stagestuff): Remove.
1053 (target_list): Remove bootstrap targets.
1054 * Makefile.in (quickstrap): Unconditionally make a synonym of all.
1055 (BOOT_LANGUAGES, STAGE1_CFLAGS, STAGE1_CHECKING,
1056 REMAKEFLAGS, FLAGS_TO_PASS, PREPEND_DOTDOT_TO_RELATIVE_PATHS,
1057 SUBDIR_FLAGS_TO_PASS, WERROR_FLAGS, STRICT2_WARN, LANG_STAGESTUFF,
1058 VOL_FILES, POSTSTAGE1_FLAGS_TO_PASS, STAGE2_FLAGS_TO_PASS,
1059 STAGEPROFILE_FLAGS_TO_PASS, STAGEFEEDBACK_FLAGS_TO_PASS, stage1_build,
1060 stage1_copy, stage2_build, stage2_copy, stageprofile_build,
1061 stageprofile_copy, stage3_build, stage3_copy, stagefeedback_build,
1062 stagefeedback_copy, stage4_build, clean_s1, clean_sw, bootstrap,
1063 bootstrap-lean, bootstrap2, bootstrap2-lean, bootstrap3,
1064 bootstrap3-lean, bootstrap4, bootstrap4-lean, unstage1, unstage2,
1065 unstage3, unstage4, unstageprofile, unstagefeedback, restage, restage2,
1066 restage3, restage4, restageprofile, restagefeedback, bubbleestrap,
1067 cleanstrap, unstrap, restrap, *compare, *compare3, *compare4,
1068 *compare-lean, *compare3-lean, *compare4-lean, stage1-start, stage1,
1069 stage2-start, stage2, stage3-start, stage3, stage4-start, stage4,
1070 stageprofile-start, stageprofile, stagefeedback-start, stagefeedback,
1071 risky-stage1, risky-stage2, risky-stage3, risky-stage4): Remove.
1072 (ORDINARY_FLAGS_TO_PASS): Rename to FLAGS_TO_PASS.
1073 (STAGECOPYSTUFF, STAGEMOVESTUFF): Consolidate into MOSTLYCLEANFILES.
1074 (mostlyclean): Adjust.
1075 (clean, distclean): Don't mention bootstrap stuff.
1076 * configure: Regenerate.
1077 * ada/config-lang.in, cp/config-lang.in, forttran/config-lang.in,
1078 java/config-lang.in, objc/config-lang.in, objcp/config-lang.in,
1079 treelang/config-lang.in (stagestuff): Remove.
1080 * doc/sourcebuild.texi (stage1, stage2, stage3, stage4,
1081 stageprofile, stagefeedback, stagestuff): Remove mention.
1083 2007-01-11 Nick Clifton <nickc@redhat.com>
1085 * config/mcore/predicates.md (mcore_general_movesrc_operand):
1087 (mcore_general_movdst_operand): Do not accept CONST_INTs.
1088 (mcore_arith_K_S_operand): Run the test for the S constraint not
1089 the test for the M constraint.
1090 (mcore_addsub_operand): Do not accept integer values that are
1091 larger than 32 bits.
1092 * config/mcore/mcore.md: Remove unused constraints from split.
1093 (andsi3): Use HOST_WIDE_INT instead of int to hold an INTVAL.
1095 (allocate_stack): Likewise.
1096 * config/mcore/mcore.c (mcore_print_operand): Restrict output of P
1097 operands to 32 bits.
1098 (mcore_const_costs): Use HOST_WIDE_INT instead of int to hold an
1100 (mcore_and_cost, mcore_modify_comparison, const_ok_for_mcore,
1101 mcore_const_ok_for_inline, mcore_const_trick_uses_not,
1102 try_constant_tricks, mcore_num_ones, mcore_num_zeros,
1103 mcore_output_bclri, mcore_output_andn, output_inline_const,
1104 mcore_output_move, mcore_output_movedouble): Likewise.
1105 (mcore_output_cmov): Use CONST_OK_FOR_M and CONST_OK_FOR_N.
1106 (output_inline_const): Likewise.
1107 (output_inline_const): Fix format strings used in sprintf
1109 * config/mcore/mcore-protos.h: Update prototypes for changed
1110 functions in mcore.c.
1111 * config/mcore/mcore.h (CONST_OK_FOR_I): Cast values to
1112 HOST_WIDE_INT and not int.
1113 (CONST_OK_FOR_J, CONST_OK_FOR_K, CONST_OK_FOR_L, CONST_OK_FOR_M,
1114 CONST_OK_FOR_N): Likewise.
1115 (LEGITIMATE_CONSTANT_P): Also check CONSTANT_P.
1116 (GO_IF_LEGITIMATE_INDEX): Use HOST_WIDE_INT instead of int to hold
1119 2007-01-10 Jan Hubicka <jh@suse.cz>
1121 * tree-vrp.c (remove_range_assertions): Release defs.
1122 * tree-ssa-loop-ivopts.c (rmeove_statement): Likewise.
1123 * tree-ssa-dom.c (remove_stmt_or_phi): Likewise.
1125 2007-01-10 Paul Brook <paul@codesourcery.com>
1127 * config/arm/arm.c (arm_rtx_costs_1): Handle mutiply-subtract.
1128 * config/arm/arm.md (mulsi3subsi): New insn.
1130 2007-01-10 Zdenek Dvorak <dvorakz@suse.cz>
1132 * tree-ssa-loop-manip.c (tree_unroll_loop): Make it a wrapper over ...
1133 (tree_transform_and_unroll_loop): New.
1134 * tree-flow.h (transform_callback, tree_transform_and_unroll_loop):
1137 2007-01-10 Robert Kennedy <jimbob@google.com>
1139 * fold-const.c (fold_comparison): Fold comparisons like (x *
1140 1000 < 0) to (x < 0).
1142 2007-01-10 Ian Lance Taylor <iant@google.com>
1144 * tree-pretty-print.c (dump_generic_node): Print parentheses when
1145 operands have the same priority.
1147 2007-01-10 Tom Tromey <tromey@redhat.com>
1149 * fold-const.c (fold_truthop): Don't check can_use_bit_fields_p.
1150 (fold_binary): Likewise.
1151 * langhooks.c (lhd_can_use_bit_fields_p): Removed.
1152 * langhooks-def.h (lhd_can_use_bit_fields_p): Removed.
1153 (LANG_HOOKS_CAN_USE_BIT_FIELDS_P): Removed.
1154 (LANG_HOOKS_INITIALIZER): Remove LANG_HOOKS_CAN_USE_BIT_FIELDS_P.
1155 * langhooks.h (struct lang_hooks): Removed field
1156 'can_use_bit_fields_p'.
1158 2007-01-10 Ralf Corsépius <ralf.corsepius@rtems.org>
1160 * config/bfin/t-bfin, config/bfin/t-bfin-elf: Remove GCC_CFLAGS.
1162 2007-01-10 Razya Ladelsky <razya@il.ibm.com>
1164 * function.c (get_last_funcdef_no): New function.
1165 * function.h (get_last_funcdef_no): Declare.
1166 * tree-inline.c (initialize_cfun): Add initialization.
1167 (tree_function_versioning): Cleanup.
1169 2007-01-10 Jan Hubicka <jh@suse.cz>
1171 * tree-inline.c (setup_one_parameter): Do not propagate into abnormal
1174 2007-01-10 Sa Liu <saliu@de.ibm.com>
1175 Ben Elliston <bje@au.ibm.com>
1177 * spu.h (STACK_SAVE_AREA): Use VOIDmode for SAVE_FUNCTION, SImode
1178 for SAVE_NONLOCAL and Pmode for any other save level.
1179 * spu-protos.h (spu_restore_stack_block): Declare.
1180 * spu.md (save_stack_block): Remove.
1181 (restore_stack_block): Call spu_restore_stack_block.
1182 * spu.c (spu_restore_stack_block): New function.
1183 (spu_expand_epilogue): Remove old comment.
1185 2007-01-09 Zdenek Dvorak <dvorakz@suse.cz>
1187 PR tree-optimization/30322
1188 * tree-ssa-loop-ivopts.c (fold_affine_expr, iv_value): Removed.
1189 (cand_value_at): Return the value as aff_tree.
1190 (may_eliminate_iv): Convert the bound from aff_tree to tree.
1191 * tree-affine.c (aff_combination_add_cst, aff_combination_add_product,
1192 aff_combination_mult): New functions.
1193 (aff_combination_add): Use aff_combination_add_cst.
1194 (aff_combination_convert): Allow conversions to a wider type.
1195 (tree_to_aff_combination): Handle BIT_NOT_EXPR.
1196 * tree-affine.h (aff_combination_mult): Declare.
1198 2007-01-09 Carlos O'Donell <carlos@codesourcery.com>
1200 * doc/tm.texi: Update documentation to reflect reality of exec
1201 and start file search behaviours. Update copyright year.
1202 * doc/invoke.texi: Explain how GCC_EXEC_PREFIX is used to find
1203 header file directories.
1205 2007-01-09 Uros Bizjak <ubizjak@gmail.com>
1207 * config/i386/i386.md (*sinxf2): Rename to *sinxf2_i387.
1208 (*cosxf2): Rename to cosxf2_i387.
1209 (*sindf2, *sinsf2): Extend operand 1 to XFmode. Macroize patterns
1210 using X87MODEF12 mode macro. Rename patterns to
1211 *sin_extend<mode>xf2_i387. Use SSE_FLOAT_MODE_P to disable patterns
1213 (*cosdf2, *cossf2): Ditto.
1214 (sincosdf3, sincossf3): Ditto. Rewrite corresponding splitters
1215 to match extended input operands.
1216 (sincos<mode>3): New expander.
1217 (*sinextendsfdf2, *cosextendsfdf2, *sincosextendsfdf3): Remove
1218 insn patterns and corresponding splitters.
1220 2007-01-09 Kaz Kojima <kkojima@gcc.gnu.org>
1222 * config/sh/t-linux (TARGET_LIBGCC2_CFLAGS): Delete.
1223 (SHLIB_MAPFILES, SHLIB_LINK, SHLIB_INSTALL): Likewise.
1225 2007-01-09 Nicolas Pitre <nico@cam.org>
1228 * arm/ieee754-df.S (Lad_s): Also test the low word of X for zero.
1230 2007-01-08 Geoffrey Keating <geoffk@apple.com>
1232 * target.h (struct gcc_target): New field library_rtti_comdat.
1233 * target-def.h (TARGET_CXX_LIBRARY_RTTI_COMDAT): New.
1234 (TARGET_CXX): Add TARGET_CXX_LIBRARY_RTTI_COMDAT.
1235 * doc/tm.texi (C++ ABI): Document TARGET_CXX_LIBRARY_RTTI_COMDAT.
1236 * config/darwin.h (TARGET_CXX_LIBRARY_RTTI_COMDAT): Define.
1238 2007-01-08 Geoffrey Keating <geoffk@apple.com>
1240 * doc/invoke.texi (Optimize Options): Correct description of -O0.
1242 2007-01-08 Richard Guenther <rguenther@suse.de>
1244 * tree.h (force_fit_type_double): Export.
1245 (force_fit_type): Remove.
1246 * fold-const.c (force_fit_type_double): New function.
1247 (force_fit_type): Remove.
1248 (int_const_binop): Use it.
1249 (fold_convert_const_int_from_int): Likewise.
1250 (fold_convert_const_int_from_real): Likewise.
1251 (fold_div_compare): Likewise.
1252 (fold_sign_changed_comparison): Likewise.
1253 (fold_unary): Likewise.
1254 (fold_negate_const): Likewise.
1255 (fold_abs_const): Likewise.
1256 (fold_not_const): Likewise.
1257 * c-common.c (shorten_compare): Use force_fit_type_double.
1258 * convert.c (convert_to_pointer): Likewise.
1260 2007-01-08 Richard Guenther <rguenther@suse.de>
1262 * tree.h (build_int_cst_wide_type): Export.
1263 * tree.c (build_int_cst_wide_type): New function.
1264 (build_int_cst_wide): Fix comment.
1265 * builtins.c (fold_builtin_object_size): Use build_int_cst
1266 to build -1 or 0 of the correct type. Use fit_double_type
1267 to check for overflow.
1268 * fold-const.c (optimize_bit_field_compare): Use build_int_cst_type
1270 (decode_field_reference): Likewise.
1271 (all_ones_mask_p): Likewise.
1272 (native_interpret_int): Use build_int_cst_wide_type.
1273 (fold_binary): Use build_int_cst_type to build an all-ones
1275 * stor-layout.c (set_sizetype): Use build_int_cst_wide_type.
1277 2007-01-08 Daniel Jacobowitz <dan@codesourcery.com>
1279 * config/pa/t-pa64 (libgcc_stub.a): Use $(T).
1281 2007-01-09 Ben Elliston <bje@au.ibm.com>
1283 * genautomata.c (STATS_OPTION): New option.
1284 (stats_flag): New flag.
1285 (gen_automata_option): Handle it.
1286 (initiate_automaton_gen): Ditto.
1287 (write_automata): Output statistics only if stats_flag is
1288 set. Likewise, output time statistics only if time_flag is set.
1289 * doc/md.texi (Processor pipeline description): Document new flag.
1291 2007-01-08 Richard Guenther <rguenther@suse.de>
1293 * builtins.c (fold_builtin_int_roundingfn): Use fit_double_type.
1294 * tree.c (build_int_cst_type): Likewise.
1295 (size_in_bytes): Don't call force_fit_type on the result.
1296 (int_fits_type_p): Use fit_double_type.
1297 * fold-const.c (fit_double_type): New function.
1298 (force_fit_type): Use it.
1299 * tree.h (fit_double_type): Export.
1301 2007-01-08 Jan Hubicka <jh@suse.cz>
1303 * tree-vectorizer.c (gate_increase_alignment): Fix return type.
1304 * ipa.c (function_and_variable_visibility): Fix return type.
1306 2007-01-08 Richard Guenther <rguenther@suse.de>
1308 * tree-ssa-ccp.c (maybe_fold_offset_to_array_ref): Use type
1309 of offset to build the index.
1310 * tree-pretty-print.c (dump_generic_node): Don't build negated
1311 const just for printing.
1312 * c-pretty-print.c (pp_c_integer_constant): Likewise.
1313 * builtins.c (fold_builtin_int_roundingfn): Check if result
1314 fits the type by using force_fit_type and comparing the result.
1315 * predict.c (predict_loops): Use compare_tree_int for comparison.
1316 * tree.c (build_int_cst): Fall back to integer_type_node for
1318 (build_int_cst_wide): Assert type is non-null.
1320 2007-01-08 Roberto Costa <roberto.costa@st.com>
1322 * tree-vrp.c (extract_range_from_cond_expr): New.
1323 (extract_range_from_expr): Handle COND_EXPR nodes used as expressions.
1324 * tree-ssa-ccp.c (get_maxval_strlen): Handle COND_EXPR nodes used
1326 (fold_stmt): Bug fix, avoid infinite recursion when folding COND_EXPRs.
1327 * tree-ssa-forwprop.c (simplify_cond, forward_propagate_into_cond,
1328 tree_ssa_forward_propagate_single_use_vars): Handle COND_EXPR nodes
1329 used as expressions.
1330 * tree-object-size.c (cond_expr_object_size): New.
1331 (collect_object_sizes_for): Handle COND_EXPR nodes used as expressions.
1333 2007-01-08 Jan Hubicka <jh@suse.cz>
1335 * tree-ssa-forwprop.c (forward_propagate_into_cond,
1336 tree_ssa_forward_propagate_single_use_va): Release defs of propagated
1339 2007-01-08 Richard Guenther <rguenther@suse.de>
1341 PR tree-optimization/23603
1342 * tree-vrp.c (set_value_range_to_truthvalue): New function.
1343 (extract_range_from_binary): Fall back to truthvalue instead of
1344 varying for TRUTH_*_EXPR.
1345 (extract_range_from_comparison): Fall back to truthvalue instead of
1347 (vrp_visit_phi_node): Don't adjust new range bounds to +INF/-INF
1348 if all visited PHI values were constant.
1350 2007-01-08 Jan Hubicka <jh@suse.cz>
1352 * cgraphunit.c (cgraph_process_new_functions): Reset reachable flag.
1353 (cgraph_analyze_function): break out from ...
1354 (cgraph_finalize_compilation_unit): ... here.
1355 (cgraph_expand_function): Remove forgoten commented out line.
1356 (cgraph_optimize): Analyze functions.
1358 2007-01-08 Jan Hubicka <jh@suse.cz>
1360 * tree-pas.h (TODO_remove_function): New flag.
1361 (TODO_update*): Renumber.
1362 (pass_ipa_increase_alignment,
1363 pass_ipa_function_and_variable_visibility): New passes.
1364 * cgraphunit.c (cgraph_increase_alignment): Move to tree-vectorizer.c
1365 (cgraph_function_and_variable_visibility): Move to ipa.c
1366 (cgraph_optimize): Don't call cgraph_function_and_variable_visibility,
1367 cgraph_increase_alignment.
1368 * ipa-inline.c (cgraph_decide_inlining): Don't push timevar.
1369 (cgraph_decide_inlining_incrementally): Push TV_INTEGRATION before
1370 calling tree-inline.
1371 (cgraph_early_inlining): Do not call cgraph_remove_unreachable_nodes.
1372 (pass_ipa_inline, pass_early_ipa_inlining): Set TODO_remove_functions
1373 * tree-vectorizer.c (increase_alignment): Move here from cgraphunit.c
1374 (gate_increase_alignment): New function.
1375 (pass_ipa_increase_alignment): New pass.
1376 * ipa.c: Inline tree-pass.h and timevar.h
1377 (function_and_variable_visibility): Move here from cgraphunit.c
1378 * tree-optimize.c (pass_early_local_passes): Add TODO_remove_functions.
1379 * passes.c (init_optimization_passes): Add the two new passes.
1380 (execute_todo): Handle cgraph_remove_functions.
1382 2007-01-08 Nick Clifton <nickc@redhat.com>
1384 * config/frv/predicates.md (reg_or_0_operand): Accept
1387 2007-01-08 Ralf Corsépius <ralf.corsepius@rtems.org>
1389 * config/bfin/rtems.h, config/bfin/t-rtems: New.
1390 * config.gcc: Add bfin*-rtems*.
1392 2007-01-08 Mark Shinwell <shinwell@codesourcery.com>
1394 * c.opt: Add -flax-vector-conversions.
1395 * c-typeck.c (convert_for_assignment): Pass flag to
1396 vector_types_convertible_p to allow emission of note.
1397 (digest_init): Likewise.
1398 * c-opts.c: Handle -flax-vector-conversions.
1399 * c-common.c (flag_lax_vector_conversions): New.
1400 (vector_types_convertible_p): Unless -flax-vector conversions
1401 has been passed, disallow conversions between vectors with
1402 differing numbers of subparts and/or element types. If such
1403 a conversion is disallowed, possibly emit a note on the first
1404 occasion only to inform the user of -flax-vector-conversions.
1405 The new last argument specifies this.
1406 * c-common.h (flag_lax_vector_conversions): New.
1407 (vector_types_convertible_p): Add extra argument.
1408 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use
1409 char_type_node for V*QI type vectors.
1410 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
1411 Update to satisfy new typechecking rules.
1412 * config/rs6000/altivec.h (vec_cmple): Use vec_cmpge, for both
1414 * doc/invoke.texi (C Dialect Options): Document
1415 -flax-vector-conversions.
1417 2007-01-08 Mark Shinwell <shinwell@codesourcery.com>
1419 PR tree-optimization/29877
1420 * tree-ssa-ter.c (is_replaceable_p): Deem assignments with
1421 a register variable on the RHS to not be replaceable.
1423 2007-01-08 Chen Liqin <liqin@sunnorth.com.cn>
1424 * config/score/t-score-elf (MULTILIB_OPTIONS): Change.
1425 * config/score/predicates.md (const_uimm5, sr0_operand, const_simm12,
1426 const_simm15, const_pow2, const_npow2): Added.
1427 * config/score/misc.md (insv, extv, extzv, movmemsi,
1428 move_lbu_a/b, mov_lhu_a/b etc): Added and fix some bug.
1429 * config/score/score.c (score_address_cost, score_select_cc_mode):
1431 Change CONST_OK_FOR_LETTER_P/EXTRA_CONSTRAINT define.
1432 Update score_rtx_costs for MACRO TARGET_RTX_COSTS.
1433 Update score_print_operand.
1434 * config/score/score.h (DATA_ALIGNMENT, SELECT_CC_MODE): Added.
1435 Adjust register allocate order and update some macro define.
1436 * config/score/score-mdaux.c (mdx_unaligned_load, mdx_unsigned_store,
1437 mdx_block_move_straight, mdx_block_move_loop_head,
1438 mdx_block_move_loop_body, mdx_block_move_loop_foot, mdx_block_move_loop,
1439 mdx_block_move): Added.
1440 (mdx_movsicc, mdp_select_add_imm, mdp_select, mds_zero_extract_andi,
1441 mdp_limm): Updated and fix some bug and typo.
1442 * config/score/score.md (movqi/hi/si, add/sub/zero/ext): Updated.
1443 (movsf, movdf, doloop_end): Added.
1445 2007-01-08 Kazu Hirata <kazu@codesourcery.com>
1447 * config/arm/arm.c, config/arm/arm.h, config/arm/arm.md,
1448 config/arm/thumb2.md: Fix comment typos.
1449 * doc/extend.texi: Fix a typo.
1451 2007-01-07 Eric Christopher <echristo@apple.com>
1453 * configure.ac: Check for __stack_chk_fail for darwin.
1454 * configure: Regenerate.
1456 2007-01-07 Richard Guenther <rguenther@suse.de>
1458 * tree-vrp.c (extract_range_from_assert): CSE calls to
1459 compare_values where possible.
1460 (extract_range_from_unary_expr): Likewise.
1462 2007-01-07 Anatoly Sokolov <aesok@post.ru>
1464 * config/avr/avr-protos.h (call_insn_operand): Delete prototype.
1465 * config/avr/avr.c (call_insn_operand): Delete function.
1466 * config/avr/avr.md (*pushqi, *pushhi, *pushsi, *pushsf): Use REG_SP
1467 instead of register number. Use predicates.
1468 * config/avr/predicates.md (const0_operand, reg_or_0_operand,
1469 call_insn_operand): Add.
1471 2007-01-06 Jan Hubicka <jh@suse.cz>
1473 * tree-pass.h (pass_build_cgraph_edges): Declare.
1474 * cgraphunit.c (record_refernece): Move to cgraphbuild.c
1475 (visited_nodes): Remove.
1476 (cgraph_create_edges): Move to cgraphbuild.c; rename to
1477 build_cgrpah_edges; make visited_nodes local.
1478 (cgraph_process_new_functions): DO not call initialize_inline_failed.
1479 (record_references_in_initializer): Move to cgraphbuild.c
1480 (initialize_inline_failed, rebuild_cgraph_edges,
1481 pass_rebuild_cgraph_edges): Move to cgraphbuild.c.
1482 (verify_cgraph_node): Make visited_nodes local.
1483 (cgraph_analyze_function): Do not call cgraph_create_edges and
1484 initialize_inline_failed.
1485 (cgraph_expand_function): Do not call cgraph_lower_function;
1486 assert that function is already lowered.
1487 * Makefile.in (cgraphbuild.o): New.
1488 * passes.c (init_optimization_passes): Add pass_build_cgraph_edges
1489 at the end of lowering passes.
1491 2007-01-06 Steven Bosscher <steven@gcc.gnu.org>
1493 * ifcvt.c (cond_move_convert_if_block): New function, code
1494 factored out from...
1495 (cond_move_process_if_block): ...here. Call the new function
1496 on the THEN and ELSE blocks.
1497 (merge_if_block): Do not copy global_live_at_end, merge_blocks
1498 already takes care of this.
1500 2007-01-05 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
1503 * tree.h (TREE_OVERFLOW_P): New.
1504 * c-typeck.c (parser_build_unary_op): Warn only if result
1505 overflowed and operands did not.
1506 (parser_build_binary_op): Likewise.
1507 (convert_for_assignment): Remove redundant overflow_warning.
1508 * c-common.c (overflow_warning): Don't check or set TREE_OVERFLOW.
1510 2007-01-05 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
1512 * c-typeck.c (store_init_value): Split over two lines to follow
1513 the GNU coding style.
1515 2007-01-05 Benjamin Kosnik <bkoz@redhat.com>
1517 * c-cppbuiltin.c (c_cpp_builtins): __GXX_EXPERIMENTAL_CPP0X__ to
1518 __GXX_EXPERIMENTAL_CXX0X__.
1519 * doc/cpp.texi: Same.
1521 2007-01-05 Richard Guenther <rguenther@suse.de>
1524 * tree.c (get_narrower): Do not construct COMPONENT_REFs
1525 with mismatched types. Instead explicitly build a
1526 conversion NOP_EXPR.
1528 2007-01-05 Ian Lance Taylor <iant@google.com>
1530 * c-common.c (decl_with_nonnull_addr_p): New function.
1531 (c_common_truthvalue_conversion): Call it.
1532 * c-typeck.c (build_binary_op): Likewise.
1533 * c-common.h (decl_with_nonnull_addr_p): Declare.
1535 2007-01-05 Jakub Jelinek <jakub@redhat.com>
1538 * libgcc2.c (__divdc3): Compare c and d against 0.0 instead of
1541 2007-01-05 Joel Brobecker <brobecker@adacore.com>
1543 * doc/install.texi (Final install): Document the fact that
1544 the GNAT runtime should not be stripped.
1546 2007-01-04 Jan Hubicka <jh@suse.cz>
1548 * tree-inline.c (fold_marked_statements): Update operand caches
1549 and EH after folding
1551 2007-01-04 Ian Lance Taylor <iant@google.com>
1553 * c-common.c (check_function_nonnull): Whitespace fix.
1555 2007-01-04 Jan Hubicka <jh@suse.cz>
1557 * tree-optimize.c (execute_fixup_cfg): Correct previously mistakely
1558 comitted older version of patch.
1559 (pass_fixup_cfg): Add TODOs to verify flow and statements, dump
1560 function, celanup cfg and collect garbage.
1562 2007-01-04 Mike Stump <mrs@apple.com>
1564 * Makefile.in (mostlyclean): Don't remove libgcc anymore.
1567 2007-01-04 Eric Christopher <echristo@apple.com>
1569 * libgcc2.c (__bswapsi2): Use SItype.
1570 (__bswapdi2): Use DItype.
1571 * libgcc2.h: Update for above.
1573 2007-01-04 Paul Brook <paul@codesourcery.com>
1575 * config/arm/arm.md (arm_mulsi3, thumb_mulsi3, mulsi3_compare0,
1576 mulsi_compare0_scratch, mulsi3addsi, mulsi3addsi_compare0,
1577 mulsi3addsi_compare0_scratch, mulsidi3adddi, mulsidi3,
1578 umulsidi3, umulsidi3adddi, smulsi3_highpart,
1579 umulsi3_highpart): Make conditional on !arm_arch6.
1580 (arm_mulsi3_v6, thumb_mulsi3_v6, mulsi3_compare0_v6,
1581 mulsi_compare0_scratch_v6, mulsi3addsi_v6, mulsi3addsi_compare0_v6,
1582 mulsi3addsi_compare0_scratch_v6, mulsidi3adddi_v6, mulsidi3_v6,
1583 umulsidi3_v6, umulsidi3adddi_v6, smulsi3_highpart_v6,
1584 umulsi3_highpart_v6): New insns.
1586 2007-01-04 Roger Sayle <roger@eyesopen.com>
1588 * fold-const.c (fold_convert): When casting an expression to void,
1589 fold_ignored_result may discover a GIMPLE_MODIFY_STMT which doesn't
1590 have a type. Instead of attempting to build a NOP_EXPR, return
1591 these "special" trees directly.
1593 2007-01-04 Joseph Myers <joseph@codesourcery.com>
1595 * config/rs6000/rs6000.c (rs6000_rtx_costs): Make adjustment for
1596 MULT inside MINUS as either argument. Use rs6000_cost->dmul -
1597 rs6000_cost->fp not 0 as adjustment for outer NEG.
1599 2007-01-04 Jan Hubicka <jh@suse.cz>
1601 * cgraph.c (cgraph_release_function_body): New function.
1602 (cgraph_remove_node): Use it.
1603 * cgraph.h (cgraph_release_function_body): Declare.
1604 * cgraphunit.c (cgraph_expand_function): Use it.
1605 * ipa.c (cgraph_remove_unreahchable_nodes): Use it.
1606 * tree-ssa.c (delete_tree_ssa): Allow to be called before aliasing
1607 is initialized and while compilation of other function is running.
1608 * tree-optimize.c (execute_free_cfg_annotations): Move code to clear
1609 statement CFG annotations from here to ...
1610 * tree-cfg.c (delete_tree_cfg_annotations): ... here.
1612 2007-01-04 Zdenek Dvorak <dvorakz@suse.cz>
1614 * cfgloop.h (enum li_flags): Make the constants powers of two.
1616 2007-01-04 Jan Hubicka <jh@suse.cz>
1618 * tree-inline.c (copy_bb): Insert new statements to statements_to_fold
1620 (fold_marked_statements): New function.
1621 (optimize_inline_calls, tree_function_versioning): Fold new statements.
1622 * tree-inline.h (copy_body_data): Add statements_to_fold.
1624 2007-01-03 Daniel Jacobowitz <dan@codesourcery.com>
1626 * config.gcc: Mention libgcc/config.host.
1627 * Makefile.in: Update comments mentioning libgcc.
1628 (LIBGCC, INSTALL_LIBGCC, GCC_PARTS, mklibgcc): Delete.
1629 (all.cross, start.encap, rest.encap, rest.cross): Update
1630 dependencies for libgcc move.
1631 (libgcc.mk, LIBGCC_DEPS, libgcov.a, libgcc.a, stmp-multilib)
1632 (clean-target, clean-target-libgcc): Delete.
1633 (srcdirify, GCC_EXTRA_PARTS): New macros.
1634 (libgcc-support, libgcc.mvars): New rules.
1635 (distclean): Remove mention of mklibgcc.
1636 (install): Don't reference INSTALL_LIBGCC.
1637 (install-common): Don't reference EXTRA_PARTS.
1638 (install-libgcc, install-multilib): Delete rules.
1639 * mklibgcc.in: Delete file.
1640 * doc/configfiles.texi: Don't mention mklibgcc.
1642 * config/i386/t-darwin (SHLIB_VERPFX): Delete (moved to libgcc).
1643 * config/i386/t-darwin64 (SHLIB_VERPFX): Likewise.
1644 * config/rs6000/t-darwin (SHLIB_VERPFX): Likewise.
1645 * config/rs6000/t-ppccomm (TARGET_LIBGCC2_CFLAGS, SHLIB_MAPFILES)
1646 (mklibgcc, ldblspecs): Likewise.
1648 * config/i386/t-nwld (libgcc.def, libc.def, libpcre.def)
1649 (posixpre.def): Use $(T).
1650 (SHLIB_EXT, SHLIB_NAME, SHLIB_SLIBDIR_QUAL, SHLIB_DEF, SHLIB_MAP)
1651 (SHLIB_SRC, SHLIB_INSTALL): Delete.
1652 (SHLIB_LINK): Make dummy.
1653 * config/t-slibgcc-darwin: Delete contents except for dummy SHLIB_LINK.
1655 * config/frv/t-linux (EXTRA_MULTILIB_PARTS): Clear.
1657 * config/alpha/t-crtfm: Use $(T) in rules for EXTRA_PARTS.
1658 * config/alpha/t-vms, config/alpha/t-vms64, config/fr30/t-fr30,
1659 config/i386/t-rtems-i386, config/ia64/t-ia64, config/rs6000/t-beos,
1660 config/rs6000/t-newas, config/sparc/t-elf: Likewise.
1662 * configure.ac (all_outputs): Remove mklibgcc.
1663 * configure: Regenerated.
1665 2007-01-03 Josh Conner <jconner@apple.com>
1668 * calls.c (compute_argument_addresses): Set stack and stack_slot
1669 for partial args, too.
1670 (store_one_arg): Use locate.size.constant for the size when
1671 generating a save_area.
1673 2007-01-03 Robert Kennedy <jimbob@google.com>
1675 * tree-cfg.c (tree_merge_blocks): Release SSA_NAME phi results
1676 whose definitions are deleted due to basic block merging.
1678 2007-01-03 Paul Brook <paul@codesourcery.com>
1681 * config/arm/arm.c (output_return_instruction): Pop PC in interrupt
1683 (use_return_insn): Return 0 for Thumb interrupt functions.
1684 (print_multi_reg): Add rfe argument for IRQ returns.
1685 (arm_output_epilogue): Pop interrupt return address directly into PC.
1686 (arm_expand_prologue): Only adjust IRQ return address in Arm mode.
1688 2007-01-03 Paul Brook <paul@codesourcery.com>
1690 Merge from sourcerygxx-4_1.
1691 * config/arm/thumb2.md: New file.
1692 * config/arm/elf.h (JUMP_TABLES_IN_TEXT_SECTION): Return True for
1694 * config/arm/coff.h (JUMP_TABLES_IN_TEXT_SECTION): Ditto.
1695 * config/arm/aout.h (ASM_OUTPUT_ADDR_VEC_ELT): Add !Thumb-2 assertion.
1696 (ASM_OUTPUT_ADDR_DIFF_ELT): Output Thumb-2 jump tables.
1697 * config/arm/aof.h (ASM_OUTPUT_ADDR_DIFF_ELT): Output Thumb-2 jump
1699 (ASM_OUTPUT_ADDR_VEC_ELT): Add !Thumb-2 assertion.
1700 * config/arm/ieee754-df.S: Use macros for Thumb-2/Unified asm
1702 * config/arm/ieee754-sf.S: Ditto.
1703 * config/arm/arm.c (thumb_base_register_rtx_p): Rename...
1704 (thumb1_base_register_rtx_p): ... to this.
1705 (thumb_index_register_rtx_p): Rename...
1706 (thumb1_index_register_rtx_p): ... to this.
1707 (thumb_output_function_prologue): Rename...
1708 (thumb1_output_function_prologue): ... to this.
1709 (thumb_legitimate_address_p): Rename...
1710 (thumb1_legitimate_address_p): ... to this.
1711 (thumb_rtx_costs): Rename...
1712 (thumb1_rtx_costs): ... to this.
1713 (thumb_compute_save_reg_mask): Rename...
1714 (thumb1_compute_save_reg_mask): ... to this.
1715 (thumb_final_prescan_insn): Rename...
1716 (thumb1_final_prescan_insn): ... to this.
1717 (thumb_expand_epilogue): Rename...
1718 (thumb1_expand_epilogue): ... to this.
1719 (arm_unwind_emit_stm): Rename...
1720 (arm_unwind_emit_sequence): ... to this.
1721 (thumb2_legitimate_index_p, thumb2_legitimate_address_p,
1722 thumb1_compute_save_reg_mask, arm_dwarf_handle_frame_unspec,
1723 thumb2_index_mul_operand, output_move_vfp, arm_shift_nmem,
1724 arm_save_coproc_regs, thumb_set_frame_pointer, arm_print_condition,
1725 thumb2_final_prescan_insn, thumb2_asm_output_opcode, arm_output_shift,
1726 thumb2_output_casesi): New functions.
1727 (TARGET_DWARF_HANDLE_FRAME_UNSPEC): Define.
1728 (FL_THUMB2, FL_NOTM, FL_DIV, FL_FOR_ARCH6T2, FL_FOR_ARCH7,
1729 FL_FOR_ARCH7A, FL_FOR_ARCH7R, FL_FOR_ARCH7M, ARM_LSL_NAME,
1730 THUMB2_WORK_REGS): Define.
1731 (arm_arch_notm, arm_arch_thumb2, arm_arch_hwdiv, arm_condexec_count,
1732 arm_condexec_mask, arm_condexec_masklen)): New variables.
1733 (all_architectures): Add armv6t2, armv7, armv7a, armv7r and armv7m.
1734 (arm_override_options): Check new CPU capabilities.
1735 Set new architecture flag variables.
1736 (arm_isr_value): Handle v7m interrupt functions.
1737 (user_return_insn): Return 0 for v7m interrupt functions. Handle
1739 (const_ok_for_arm): Handle Thumb-2 constants.
1740 (arm_gen_constant): Ditto. Use movw when available.
1741 (arm_function_ok_for_sibcall): Return false for v7m interrupt
1743 (legitimize_pic_address, arm_call_tls_get_addr): Handle Thumb-2.
1744 (thumb_find_work_register, arm_load_pic_register,
1745 legitimize_tls_address, arm_address_cost, load_multiple_sequence,
1746 emit_ldm_seq, emit_stm_seq, arm_select_cc_mode, get_jump_table_size,
1747 print_multi_reg, output_mov_long_double_fpa_from_arm,
1748 output_mov_long_double_arm_from_fpa, output_mov_double_fpa_from_arm,
1749 output_mov_double_fpa_from_arm, output_move_double,
1750 arm_compute_save_reg_mask, arm_compute_save_reg0_reg12_mask,
1751 output_return_instruction, arm_output_function_prologue,
1752 arm_output_epilogue, arm_get_frame_offsets, arm_regno_class,
1753 arm_output_mi_thunk, thumb_set_return_address): Ditto.
1754 (arm_expand_prologue): Handle Thumb-2. Use arm_save_coproc_regs.
1755 (arm_coproc_mem_operand): Allow POST_INC/PRE_DEC.
1756 (arithmetic_instr, shift_op): Use arm_shift_nmem.
1757 (arm_print_operand): Use arm_print_condition. Handle '(', ')', '.',
1759 (arm_final_prescan_insn): Use extract_constrain_insn_cached.
1760 (thumb_expand_prologue): Use thumb_set_frame_pointer.
1761 (arm_file_start): Output directive for unified syntax.
1762 (arm_unwind_emit_set): Handle stack alignment instruction.
1763 * config/arm/lib1funcs.asm: Remove default for __ARM_ARCH__.
1764 Add v6t2, v7, v7a, v7r and v7m.
1765 (RETLDM): Add Thumb-2 code.
1766 (do_it, shift1, do_push, do_pop, COND, THUMB_SYNTAX): New macros.
1767 * config/arm/arm.h (TARGET_CPU_CPP_BUILTINS): Define __thumb2__.
1768 (TARGET_THUMB1, TARGET_32BIT, TARGET_THUMB2, TARGET_DSP_MULTIPLY,
1769 TARGET_INT_SIMD, TARGET_UNIFIED_ASM, ARM_FT_STACKALIGN, IS_STACKALIGN,
1770 THUMB2_TRAMPOLINE_TEMPLATE, TRAMPOLINE_ADJUST_ADDRESS,
1771 ASM_OUTPUT_OPCODE, THUMB2_GO_IF_LEGITIMATE_ADDRESS,
1772 THUMB2_LEGITIMIZE_ADDRESS, CASE_VECTOR_PC_RELATIVE,
1773 CASE_VECTOR_SHORTEN_MODE, ADDR_VEC_ALIGN, ASM_OUTPUT_CASE_END,
1774 ADJUST_INSN_LENGTH): Define.
1775 (TARGET_REALLY_IWMMXT, TARGET_IWMMXT_ABI, CONDITIONAL_REGISTER_USAGE,
1776 STATIC_CHAIN_REGNUM, HARD_REGNO_NREGS, INDEX_REG_CLASS,
1777 BASE_REG_CLASS, MODE_BASE_REG_CLASS, SMALL_REGISTER_CLASSES,
1778 PREFERRED_RELOAD_CLASS, SECONDARY_OUTPUT_RELOAD_CLASS,
1779 SECONDARY_INPUT_RELOAD_CLASS, LIBCALL_VALUE, FUNCTION_VALUE_REGNO_P,
1780 TRAMPOLINE_SIZE, INITIALIZE_TRAMPOLINE, HAVE_PRE_INCREMENT,
1781 HAVE_POST_DECREMENT, HAVE_PRE_DECREMENT, HAVE_PRE_MODIFY_DISP,
1782 HAVE_POST_MODIFY_DISP, HAVE_PRE_MODIFY_REG, HAVE_POST_MODIFY_REG,
1783 REGNO_MODE_OK_FOR_BASE_P, LEGITIMATE_CONSTANT_P,
1784 REG_MODE_OK_FOR_BASE_P, REG_OK_FOR_INDEX_P, GO_IF_LEGITIMATE_ADDRESS,
1785 LEGITIMIZE_ADDRESS, THUMB2_LEGITIMIZE_ADDRESS,
1786 GO_IF_MODE_DEPENDENT_ADDRESS, MEMORY_MOVE_COST, BRANCH_COST,
1787 ASM_APP_OFF, ASM_OUTPUT_CASE_LABEL, ARM_DECLARE_FUNCTION_NAME,
1788 FINAL_PRESCAN_INSN, PRINT_OPERAND_PUNCT_VALID_P,
1789 PRINT_OPERAND_ADDRESS): Adjust for Thumb-2.
1790 (arm_arch_notm, arm_arch_thumb2, arm_arch_hwdiv): New declarations.
1791 * config/arm/arm-cores.def: Add arm1156t2-s, cortex-a8, cortex-r4 and
1793 * config/arm/arm-tune.md: Regenerate.
1794 * config/arm/arm-protos.h: Update prototypes.
1795 * config/arm/vfp.md: Enable patterns for Thumb-2.
1796 (arm_movsi_vfp): Add movw alternative. Use output_move_vfp.
1797 (arm_movdi_vfp, movsf_vfp, movdf_vfp): Use output_move_vfp.
1798 (thumb2_movsi_vfp, thumb2_movdi_vfp, thumb2_movsf_vfp,
1799 thumb2_movdf_vfp, thumb2_movsfcc_vfp, thumb2_movdfcc_vfp): New.
1800 * config/arm/libunwind.S: Add Thumb-2 code.
1801 * config/arm/constraints.md: Update include Thumb-2.
1802 * config/arm/ieee754-sf.S: Add Thumb-2/Unified asm support.
1803 * config/arm/ieee754-df.S: Ditto.
1804 * config/arm/bpabi.S: Ditto.
1805 * config/arm/t-arm (MD_INCLUDES): Add thumb2.md.
1806 * config/arm/predicates.md (low_register_operand,
1807 low_reg_or_int_operand, thumb_16bit_operator): New.
1808 (thumb_cmp_operand, thumb_cmpneg_operand): Rename...
1809 (thumb1_cmp_operand, thumb1_cmpneg_operand): ... to this.
1810 * config/arm/t-arm-elf: Add armv7 multilib.
1811 * config/arm/arm.md: Update patterns for Thumb-2 and Unified asm.
1813 (UNSPEC_STACK_ALIGN, ce_count): New.
1814 (arm_incscc, arm_decscc, arm_umaxsi3, arm_uminsi3,
1815 arm_zero_extendsidi2, arm_zero_extendqidi2): New
1817 * config/arm/fpa.md: Update patterns for Thumb-2 and Unified asm.
1818 (thumb2_movsf_fpa, thumb2_movdf_fpa, thumb2_movxf_fpa,
1819 thumb2_movsfcc_fpa, thumb2_movdfcc_fpa): New insns.
1820 * config/arm/cirrus.md: Update patterns for Thumb-2 and Unified asm.
1821 (cirrus_thumb2_movdi, cirrus_thumb2_movsi_insn,
1822 thumb2_cirrus_movsf_hard_insn, thumb2_cirrus_movdf_hard_insn): New
1824 * doc/extend.texi: Document ARMv7-M interrupt functions.
1825 * doc/invoke.texi: Document Thumb-2 new cores+architectures.
1827 2007-01-03 Jakub Jelinek <jakub@redhat.com>
1829 * unwind-dw2.c (SIGNAL_FRAME_BIT, EXTENDED_CONTEXT_BIT): Define.
1830 (struct _Unwind_Context): Rename args_size to flags, remove
1831 signal_frame field, add a new args_size field and version field.
1832 (_Unwind_IsSignalFrame, _Unwind_SetSignalFrame,
1833 _Unwind_IsExtendedContext): New inline functions.
1834 (_Unwind_GetGR, _Unwind_SetGR, _Unwind_GetGRPtr, _Unwind_SetGRPtr):
1835 Assume by_value array is only present if _Unwind_IsExtendedContext.
1836 (_Unwind_GetIPInfo, execute_cfa_program, uw_frame_state_for): Use
1837 _Unwind_IsSignalFrame.
1838 (__frame_state_for): Initialize context.flags to EXTENDED_CONTEXT_BIT.
1839 (uw_update_context_1): Use _Unwind_SetSignalFrame.
1840 (uw_init_context_1): Initialize context->flags to
1841 EXTENDED_CONTEXT_BIT.
1842 * config/rs6000/linux-unwind.h (frob_update_context): Use
1843 _Unwind_SetSignalFrame.
1845 2007-01-03 Andrew Pinski <andrew_pinski@playstation.sony.com>
1848 * gimplify.c (gimplify_modify_expr_complex_part): Move below
1849 tree_to_gimple_tuple. Call tree_to_gimple_tuple when we need
1852 2007-01-03 Kazu Hirata <kazu@codesourcery.com>
1854 * config/i386/i386.h (NON_STACK_REG_P, REGNO_OK_FOR_SIREG_P,
1855 REGNO_OK_FOR_DIREG_P, REWRITE_ADDRESS, ASM_OPERAND_LETTER,
1856 RET, AT_SP): Remove.
1857 * config/i386/i386.md (*sse_prologue_save_insn): Use return
1860 * alias.c (init_alias_analysis): Use VEC_safe_grow_cleared.
1861 * cfgbuild.c (find_basic_blocks): Likewise.
1862 * cfgrtl.c (rtl_create_basic_block): Likewise.
1863 * function.c (temp_slots_at_level): Likewise.
1864 * reg-stack.c (stack_regs_mentioned): Likewise.
1865 * regclass.c (allocate_reg_info): Likewise.
1866 * tree-cfg.c (init_empty_tree_cfg, build_tree_cfg, create_bb,
1867 set_bb_for_stmt, move_block_to_fn): Likewise.
1868 * tree-complex.c (tree_lower_complex): Likewise.
1869 * vec.h (VEC_safe_grow_cleared): New.
1871 * cgraphunit.c, tree-ssa-alias.c: Fix comment typos.
1873 2007-01-03 Zdenek Dvorak <dvorakz@suse.cz>
1875 * loop-unswitch.c (unswitch_loop): Pass probabilities to loopify.
1876 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Pass probabilities
1878 * cfgloopmanip.c (scale_loop_frequencies): Export.
1879 (loopify): Scale the frequencies by prescribed coefficients.
1880 (set_zero_probability): New function.
1881 (duplicate_loop_to_header_edge): Improve updating of frequencies.
1882 (lv_adjust_loop_entry_edge, loop_version): Set probabilities
1883 and frequencies according to arguments.
1884 * tree-ssa-loop-manip.c (tree_unroll_loop): Set probabilities
1886 * cfg.c (scale_bbs_frequencies_int): Allow scaling the frequencies up.
1887 * modulo-sched.c (sms_schedule): Set probabilities for entering
1888 versioned loop correctly.
1889 * tree-vect-transform.c (vect_transform_loop): Ditto.
1890 * cfgloop.h (loopify, loop_version): Declaration changed.
1891 (scale_loop_frequencies): Declared.
1893 2007-01-02 Jan Hubicka <jh@suse.cz>
1895 * cgraph.c: Include tree-flow.h
1896 (cgraph_add_new-function): Handle IPA_SSA mode; execute
1898 * cgraph.h (enum cgraph_state): Add CGRAPH_STATE_IPA_SSA.
1899 * tree-pass.h (pass_all_early_optimizations): Declare.
1900 * cgraphunit.c (cgraph_process_new_functions): Add IPA_SSA; execute
1902 (cgraph_analyze_function): Do early_local_passes.
1903 * tree-mudflap.c (mf_decl_cache_locals, mf_build_check_statement_for):
1904 Do not add referenced vars.
1905 * tree-optimize.c (gate_all_optimizations): Do not execute when not in
1907 (gate_all_early_local_passes): New gate.
1908 (pass_early_local_passes): Use new gate.
1909 (execute_early_local_optimizations): New functions.
1910 (gate_all_early_optimizations): New gate.
1911 (pass_all_early_optimizations): New pass.
1912 (execute_free_datastructures): Free SSA only when initialized.
1913 (gate_init_datastructures): Init only when optimizing.
1914 (tree_lowering_passes): Do early local passes when called late.
1915 * tree-profile.c (do_tree_profiling): Don't profile functions added
1917 (do_early_tree_profiling, pass_early_tree_profile): Kill.
1918 * tree-cfg.c (update_modified_stmts): Do not update when operands are
1920 * passes.c (init_optimizations_passes): Reorder so we go into SSA
1921 during early_local_passes.
1922 * Makefile.in (cgraph.o): Add dependency on tree-flow.h.
1925 2007-01-02 Carlos O'Donell <carlos@codesourcery.com>
1927 * Makefile.in: Update copyright year.
1929 2007-01-02 Carlos O'Donell <carlos@codesourcery.com>
1931 * Makefile.in: Export GCC_EXEC_PREFIX before calling $(RUNTEST)
1932 in $(lang_checks) and check-consistency targets.
1934 2007-01-02 Jan Hubicka <jh@suse.cz>
1936 * tree-mudflap.c (mf_decl_cache_locals, mf_build_check_statement_for):
1937 Do not add referenced vars.
1938 * tree-cfg.c (update_modified_stmts): Do not update when SSA operands
1940 * passes.c (init_optimization_passes): Put mudflap_2 after
1941 free_datastructures.
1943 2007-01-02 Jan Hubicka <jh@suse.cz>
1945 * tree-optimize (execute_fixup_cfg): Set after_inlining flag.
1946 Set NOTHROW flag on call statements proved to be nothrow.
1947 Update statement of local calls so new pure/const functions are
1948 updated. Update_ssa when in ssa form. Mark PHI nodes of nonlocal
1950 (tree_rest_of_compilation): Register hooks and initialize bitmap
1951 early. Do not set after_inlining flag.
1953 2007-01-02 Steve Ellcey <sje@cup.hp.com>
1955 * sbitmap.c (HOST_BITS_PER_LONG_LONG): Change to
1956 HOST_BITS_PER_LONGLONG
1958 2007-01-02 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
1961 * c-typeck.c (store_init_value): Don't emit pedantic overflow
1962 warning for non-static initializers.
1964 2007-01-02 Steven Bosscher <steven@gcc.gnu.org>
1966 * config/alpha/alpha.md, arm/arm.c, darwin.c, frv/frv.md,
1967 m32r/m32r.c, m32r/m32r.c, mn10300/mn10300.md, pa/pa.c,
1968 rs6000/rs6000.c, s390/s390.md, sh/sh.md, sparc/sparc.c:
1969 Always use set_unique_reg_note to add REG_EQUAL notes.
1971 2007-01-02 Kazu Hirata <kazu@codesourcery.com>
1974 2007-01-02 Kazu Hirata <kazu@codesourcery.com>
1976 * alias.c (init_alias_analysis): Use VEC_safe_grow_cleared.
1977 * cfgbuild.c (find_basic_blocks): Likewise.
1978 * cfgrtl.c (rtl_create_basic_block): Likewise.
1979 * function.c (temp_slots_at_level): Likewise.
1980 * reg-stack.c (stack_regs_mentioned): Likewise.
1981 * regclass.c (allocate_reg_info): Likewise.
1982 * tree-cfg.c (init_empty_tree_cfg, build_tree_cfg, create_bb,
1983 set_bb_for_stmt, move_block_to_fn): Likewise.
1984 * tree-complex.c (tree_lower_complex): Likewise.
1985 * vec.h (VEC_safe_grow_cleared): New.
1987 2007-01-02 Ian Lance Taylor <iant@google.com>
1989 * c-common.c (c_common_truthvalue_conversion): When warning about
1990 using an assignment as a truth value, set TREE_NO_WARNING.
1992 2007-01-02 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
1995 * c.opt (Wold-style-declaration): New.
1996 * doc/invoke.texi (C-only Warning Options): New.
1997 (Wold-style-declaration): Document it.
1998 (Wextra): Enabled by -Wextra.
1999 * c-opts.c (c_common_post_options): Enabled by -Wextra.
2000 * c-decl.c (declspecs_add_scspec): Replace -Wextra with
2001 -Wold-style-declaration.
2003 2007-01-02 Kazu Hirata <kazu@codesourcery.com>
2005 * alias.c (init_alias_analysis): Use VEC_safe_grow_cleared.
2006 * cfgbuild.c (find_basic_blocks): Likewise.
2007 * cfgrtl.c (rtl_create_basic_block): Likewise.
2008 * function.c (temp_slots_at_level): Likewise.
2009 * reg-stack.c (stack_regs_mentioned): Likewise.
2010 * regclass.c (allocate_reg_info): Likewise.
2011 * tree-cfg.c (init_empty_tree_cfg, build_tree_cfg, create_bb,
2012 set_bb_for_stmt, move_block_to_fn): Likewise.
2013 * tree-complex.c (tree_lower_complex): Likewise.
2014 * vec.h (VEC_safe_grow_cleared): New.
2016 2007-01-02 Douglas Gregor <doug.gregor@gmail.com>
2018 * c-common.c(c_common_nodes_and_builtins): Since variants of
2019 void_type_node get built before it is given a name, we need to
2020 give those variants the name, too.
2021 (complete_array_type): We need to work with the canonical main
2022 type of the array, from which we will build the qualified version.
2023 * params.def (PARAM_VERIFY_CANONICAL_TYPES): New.
2024 * print-tree.c (print_node): Display canonical type information
2026 * stor-layout.c (layout_type): When we don't know the
2027 alignment of a type for which we're building an array, we end up
2028 guessing wrong, so make the type require structural equality.
2029 * tree.c (make_node_stat): When we build a new type, it is its
2031 (build_type_attribute_qual_variant): When building an attribute
2032 variant, its canonical type is the non-attribute variant. However,
2033 if the attributes are target-dependent and they differ, we need to
2034 use structural equality checks for this type.
2035 (build_qualified_type): A qualified type is not equivalent to its
2036 unqualified variant; set the canonical type appropriately.
2037 (build_distinct_type_copy): When building a distinct type from
2038 another type, the new type is its own canonical type.
2039 (build_variant_type_copy): When building a new type variant, we
2040 assume that it is equivalent to the original type.
2041 (build_pointer_type_for_mode): When building a pointer type, also
2042 build a canonical type pointer.
2043 (build_reference_type_for_mode): When building a reference type,
2044 also build a canonical type reference.
2045 (build_index_type): When we can't hash an index type (e.g.,
2046 because its maximum value is negative), the index type requires
2047 structural equality tests.
2048 (build_array_type): Build the canonical form of an array type.
2049 (build_function_type): Function types require structural equality,
2050 because they contain default arguments, attributes, etc.
2051 (build_method_type_directly): Ditto for method types.
2052 (build_offset_type): Build the canonical offset type.
2053 (build_complex_type): Build the canonical vector type.
2054 (make_vector_type): Build the canonical vector type.
2055 * tree.h (TYPE_CANONICAL): New.
2056 (TYPE_STRUCTURAL_EQUALITY_P): New.
2057 (SET_TYPE_STRUCTURAL_EQUALITY): New.
2058 (struct tree_type): Added "canonical" field.
2059 * params.h (VERIFY_CANONICAL_TYPES): New.
2060 * doc/c-tree.texi (TYPE_CANONICAL): Document.
2061 (TYPE_STRUCTURAL_EQUALITY_P): Document.
2062 (SET_TYPE_STRUCTURAL_EQUALITY): Document.
2063 * doc/invoke.texi (verify-canonical-types): Document --param
2064 parameter for verifying canonical types.
2066 2007-01-02 Joseph Myers <joseph@codesourcery.com>
2068 * config.gcc (powerpc-*-eabispe*, powerpc-*-eabisimaltivec*,
2069 powerpc-*-eabisim*, powerpc-*-eabialtivec*, powerpc-*-eabi*,
2070 powerpc-*-rtems*, powerpc-wrs-vxworks, powerpc-wrs-vxworksae,
2071 powerpcle-*-eabisim*, powerpcle-*-eabi*): Add rs6000/e500.h to
2073 * config/rs6000/e500.h: New.
2074 * config/rs6000/eabi.h (TARGET_SPE_ABI, TARGET_SPE, TARGET_E500,
2075 TARGET_ISEL, TARGET_FPRS, TARGET_E500_SINGLE, TARGET_E500_DOUBLE):
2077 * config/rs6000/linuxspe.h (TARGET_SPE_ABI, TARGET_SPE,
2078 TARGET_E500, TARGET_ISEL, TARGET_FPRS, TARGET_E500_SINGLE,
2079 TARGET_E500_DOUBLE): Remove.
2080 * config/rs6000/vxworks.h (TARGET_SPE_ABI, TARGET_SPE,
2081 TARGET_E500, TARGET_ISEL, TARGET_FPRS): Remove.
2082 * config/rs6000/rs6000.h (CHECK_E500_OPTIONS): Define.
2083 * config/rs6000/rs6000.c (rs6000_override_options): Use
2086 2007-01-02 Joseph Myers <joseph@codesourcery.com>
2088 * config/rs6000/rs6000.c (print_operand): Check (TARGET_SPE ||
2089 TARGET_E500_DOUBLE), not TARGET_E500, for %y.
2090 (rs6000_generate_compare, rs6000_emit_sCOND, output_cbranch,
2091 rs6000_emit_cmove): Don't check TARGET_E500.
2092 * config/rs6000/rs6000.md (bunordered, bordered, sunordered,
2093 sordered): Don't check TARGET_E500.
2095 2007-01-01 Eric Christopher <echristo@apple.com>
2097 * config/mips/mips.c (mips_regno_mode_ok_for_base_p): Use
2098 HARD_REGISTER_NUM_P.
2100 2007-01-01 Roger Sayle <roger@eyesopen.com>
2102 * fold-const.c (fold_binary) <EQ_EXPR>: Fold "(X^C1) eq/ne C2" into
2103 "X eq/ne (C1^C2)". Fold "(X^Z) eq/ne (Y^Z)" as "X eq/ne Y" when Z
2104 has no side-effects. Fold "(X^C1) eq/ne (Y^C2)" as "(X^(C1^C2))
2107 2007-01-01 Mike Stump <mrs@apple.com>
2109 * configure.ac: Remove support for building with Apple's gcc-3.1.
2111 2007-01-02 Joseph Myers <joseph@codesourcery.com>
2114 * caller-save.c (add_stored_regs): Only handle SUBREGs if inner
2115 REG is a hard register. Do not modify REG before calling
2117 * rtlanal.c (subreg_get_info): Don't assert size of XMODE is a
2118 multiple of the size of YMODE for certain lowpart cases.
2120 2007-01-01 Andrew Pinski <pinskia@gmail.com>
2123 * gimplify (voidify_wrapper_expr): Update for
2126 2007-01-01 Andreas Schwab <schwab@suse.de>
2129 * config/ia64/ia64.c (ia64_compute_frame_size): Account space for
2130 save of BR0 in extra_spill_size instead of spill_size.
2131 (ia64_expand_prologue): Save BR0 outside of the gr/br/fr spill
2133 (ia64_expand_epilogue): Restore BR0 from its new location.
2135 2007-01-01 Andrew Pinski <pinskia@gmail.com>
2137 * gimplify.c (gimplify_init_constructor <case VECTOR_TYPE>):
2138 Use a temporary variable if the left hand side is not a gimple
2141 2007-01-01 Andrew Pinski <pinskia@gmail.com>
2143 * gimplify.c (gimplify_return_expr): Make the temporary variable
2144 for the return expression, a gimple register variable.
2146 2007-01-01 Jan Hubicka <jh@suse.cz>
2148 * emit-rtl.c (emit_copy_of_insn_after): Do not call copy_insn_1 for
2151 2007-01-01 Mike Stump <mrs@apple.com>
2153 * configure.ac (HAVE_GAS_LITERAL16): Add autoconf check for
2155 * config/darwin.c (machopic_select_rtx_section): Use
2157 (darwin_mergeable_constant_section): Likewise.
2158 * configure: Regenerate.
2159 * config.in: Regenerate.
2161 2007-01-01 Jan Hubicka <jh@suse.cz>
2162 Andrew Pinski <pinskia@gmail.com>
2164 * cgraphunit.c (cgraph_optimize): Call cgraph_add_new_functions
2165 before starting IPA passes.