1 2002-11-10 Joseph S. Myers <jsm@polyomino.org.uk>
3 * c-decl.c (grokdeclarator): Make error for duplicate type
4 qualifiers into a pedwarn, disabled for C99.
6 2002-11-10 Hans-Peter Nilsson <hp@bitrange.com>
8 * config/mmix/mmix.h (FUNCTION_ARG_CALLEE_COPIES): Define the same
9 as FUNCTION_ARG_PASS_BY_REFERENCE.
11 2002-11-09 John David Anglin <dave@hiauly1.hia.nrc.ca>
13 * pa.h (STARTING_FRAME_OFFSET): Change offset for TARGET_64BIT to 16.
15 * config.gcc (hppa*64*-*-linux*): Shorten lines in tm_file define.
16 (hppa*64*-*-hpux11*): Likewise. Use elfos.h with gas.
17 * pa.c (output_millicode_call): Use symbol difference rather than
18 $PIC_pcrel$0 when using HP assembler.
19 * pa64-hpux.h (TARGET_GAS): Define to 1 or 0 depending on whether or
20 not elfos.h (i.e., gas) is being used.
21 (ASM_FILE_START, STRING_ASM_OP, TEXT_SECTION_ASM_OP,
22 DATA_SECTION_ASM_OP, BSS_SECTION_ASM_OP, ASM_OUTPUT_ALIGNED_COMMON,
23 ASM_OUTPUT_ALIGNED_LOCAL, GLOBAL_ASM_OP, ASM_DECLARE_FUNCTION_NAME,
24 ASM_OUTPUT_EXTERNAL, ASM_OUTPUT_EXTERNAL_LIBCALL,
25 ASM_OUTPUT_INTERNAL_LABEL, ASM_GENERATE_INTERNAL_LABEL): Define when
27 (TARGET_ASM_GLOBALIZE_LABEL): Undefine when using elfos.h.
28 (DWARF2_ASM_LINE_DEBUG_INFO): Delete.
29 (ASM_FILE_START): Add standard .SPACE and .SUBSPA defines when not
31 (TEXT_SECTION_ASM_OP, READONLY_DATA_SECTION_ASM_OP, DATA_SECTION_ASM_OP,
32 BSS_SECTION_ASM_OP): New HP style defines when not using elfos.h.
33 (TARGET_ASM_NAMED_SECTION, MAKE_DECL_ONE_ONLY, ASM_WEAKEN_LABEL):
34 Don't define when not using elfos.h.
35 (ASM_DECLARE_RESULT): Don't define.
36 * doc/install.texi (hppa*-hp-hpux*): Remove statement that HP assembler
37 doesn't work on hppa64-hp-hpux11.
38 (hppa*-hp-hpux11): Update.
40 2002-11-09 Jason Thorpe <thorpej@wasabisystems.com>
42 * config/mips/netbsd.h (SUBTARGET_ASM_SPEC): Don't pass -KPIC
43 to the assembler if -mno-abicalls was specified.
45 2002-11-09 John David Anglin <dave@hiauly1.hia.nrc.ca>
47 * pa-linux.h (PREFERRED_DEBUGGING_TYPE, DWARF2_ASM_LINE_DEBUG_INFO,
48 ASM_OUTPUT_DEF): Delete.
50 2002-11-09 Neil Booth <neil@daikokuya.co.uk>
52 * c-opts.c (COMMAND_LINE_OPTIONS): Fix -Wimplicit.
54 2002-11-08 Dale Johannesen <dalej@apple.com>
56 * dbxout.c (dbxout_type): Fix stabs info for vector types.
58 2002-11-08 Neil Booth <neil@daikokuya.co.uk>
62 * cpptrad.c (scan_out_logical_line): A '#' from a macro doesn't
63 start a directive. In assembler, #NUM is not a line directive.
65 2002-11-08 Neil Booth <neil@daikokuya.co.uk>
67 * cppmain.c (cpp_preprocess_file): Loop to pop any -included
70 2002-11-08 Kazu Hirata <kazu@cs.umass.edu>
72 * config/h8300/h8300.md (two anonymous test insns): New.
74 Fri Nov 8 11:20:19 CET 2002 Jan Hubicka <jh@suse.cz>
76 * jump.c (mark_jump_label): Handle subregs of label_refs.
78 2002-11-07 David Mosberger <davidm@hpl.hp.com>
80 * config/ia64/crtend.asm: Include "auto-host.h".
81 [HAVE_INITFINI_ARRAY]: Invoke __do_global_ctors_aux via .init_array.
82 * config/ia64/crtbegin.asm: Similarly.
83 * config/ia64/t-ia64 (crtbegin.o): Include from current directory.
84 (crtend.o, crtbeginS.o, crtendS.o): Likewise.
86 * aclocal.m4 (gcc_AC_INITFINI_ARRAY): New.
87 * configure.in: Use it if --enable-initfini-array not specified.
88 * doc/install.texi (Configuration): Document --enable-initfini-array.
89 * configure, config.in: Rebuild.
91 2002-11-07 Jason Thorpe <thorpej@wasabisystems.com>
93 * config/arm/arm-protos.h (arm_get_frame_size)
94 (thumb_get_frame_size): New prototypes.
95 * config/arm/arm.c (arm_get_frame_size)
96 (thumb_get_frame_size): New functions.
97 (use_return_insn, arm_output_epilogue, arm_output_function_epilogue)
98 (arm_compute_initial_elimination_offset, arm_expand_prologue): Use
100 (thumb_expand_prologue, thumb_expand_epilogue): Use
101 thumb_get_frame_size.
102 * config/arm/arm.h (PREFERRED_STACK_BOUNDARY): Define.
103 (machine_function): Add frame_size member.
104 (THUMB_INITIAL_ELIMINATION_OFFSET): Use thumb_get_frame_size.
106 2002-11-07 Richard Earnshaw <rearnsha@arm.com>
108 * arm.c (bit_count): Make argument unsigned long. Return unsigned.
109 Adjust code to use portable unsigned bit manipulation.
110 (insn_flags, tune_flags): Change type to unsigned.
111 (struct processors): Make flags unsigned long.
112 (arm_override_options): Change type of count and current_bit_count
115 2002-11-07 Richard Earnshaw <rearnsha@arm.com>
117 * arm/elf.h (TYPE_OPERAND_FMT): Prefix type with %.
119 Thu Nov 7 15:50:18 2002 J"orn Rennecke <joern.rennecke@superh.com>
121 * sh.h (DWARF_FRAME_RETURN_COLUMN): Use DWARF_FRAME_REGNUM.
123 Thu Nov 7 11:18:01 CET 2002 Jan Hubicka <jh@suse.cz>
125 * reg-stack.c (compensate_edge): Fix sanity check.
127 2002-11-05 Geoffrey Keating <geoffk@apple.com>
129 * config.gcc: Don't create crtbegin, crtend on Darwin; do create
130 crt2.o. Rearrange t-darwin makefiles.
131 * crtstuff.c [OBJECT_FORMAT_MACHO]: Delete.
132 * unwind-dw2-fde-darwin.c: New.
133 * unwind-dw2-fde-glibc.c: Correct comment.
134 * unwind-dw2-fde.c (__register_frame_info_bases)
135 [DWARF2_OBJECT_END_PTR_EXTENSION]: Clear fde_end.
136 (classify_object_over_fdes): Use last_fde.
137 (add_fdes): Likewise.
138 (linear_search_fdes): Likewise.
139 * unwind-dw2-fde.h (struct object)
140 [DWARF2_OBJECT_END_PTR_EXTENSION]: Add fde_end field.
142 * config/darwin.h (STARTFILE_SPEC): Include crt2.o not crtbegin.o.
143 (ENDFILE_SPEC): No crtend.o.
144 * config/t-darwin: New.
145 * config/i386/t-darwin: Delete.
146 * config/darwin-crt2.c: New.
147 * config/rs6000/t-darwin: Delete contents duplicated in t-rs6000
150 2002-11-06 David Edelsohn <edelsohn@gnu.org>
153 * config/rs6000/rs6000.md (movdi_internal64): Discourage
156 2002-11-06 Janis Johnson <janis187@us.ibm.com>
158 * doc/contrib.texi: Merge in the list from the Java web pages.
160 2002-11-06 David O'Brien <obrien@FreeBSD.org>
162 * config/sparc/freebsd: Fix typo.
164 2002-11-06 John David Anglin <dave@hiauly1.hia.nrc.ca>
166 * pa64-hpux.h (LDD_SUFFIX, PARSE_LDD_OUTPUT): Define.
168 2002-11-06 Alexandre Oliva <aoliva@redhat.com>
170 * config/mips/mips.md (call_value_multiple_internal2): Use dla for
171 non-SImode addresses.
173 2002-11-05 Bob Wilson <bob.wilson@acm.org>
175 * config/xtensa/elf.h (LIB_SPEC): Add "-lhal".
177 2002-11-05 John David Anglin <dave2hiauly1.hia.nrc.ca>
179 * pa64-hpux.h (LIB_SPEC): Fix p and pg options.
180 (STARTFILE_SPEC): Remove p and pg options.
182 2002-11-05 Andrew Haley <aph@redhat.com>
184 * fold-const.c (fold): Don't transform (a0 op compound(a1,a2))
185 to (compound(a1,a0 op a2)) if a0 or a1 have side effects.
187 2002-11-05 Richard Sandiford <rsandifo@redhat.com>
189 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Move comment to...
190 * config/mips/mips.c (mips_cannot_change_mode_class): ...here.
192 2002-11-04 Dale Johannesen <dalej@apple.com>
194 * doloop.c (doloop_modify_runtime): Fix loop count computation
196 * loop.c (loop_invariant_p): Support calling from unroller.
198 2002-11-04 Ulrich Weigand <uweigand@de.ibm.com>
200 * config/s390/s390.c (s390_decompose_address): Use arg_pointer_rtx
203 2002-11-04 Aldy Hernandez <aldyh@redhat.com>
205 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): New.
207 * config/rs6000/rs6000.h (CLASS_CANNOT_CHANGE_MODE_P): Remove.
208 (CLASS_CANNOT_CHANGE_MODE): Remove.
209 (CANNOT_CHANGE_MODE_CLASS): New.
211 * config/alpha/alpha.h: Same.
213 * config/ia64/ia64.h: Same.
215 * config/mips/mips.h: Same.
217 * config/s390/s390.h: Same.
219 * config/sh/sh.h: Same.
221 * config/pa/pa64-regs.h: Same.
223 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Add prototype.
225 * config/sh/sh.c (sh_cannot_change_mode_class): New.
227 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Add
230 * config/mips/mips.c (mips_cannot_change_mode_class): New.
232 * doc/tm.texi (Register Classes): Remove
233 CLASS_CANNOT_CHANGE_MODE and CLASS_CANNOT_CHANGE_MODE_P.
234 Document CANNOT_CHANGE_MODE_CLASS.
236 * reload.c (push_reload): Use CANNOT_CHANGE_MODE_CLASS.
239 * simplify-rtx.c (simplify_subreg): Same.
241 * reload1.c (choose_reload_regs): Same.
243 * recog.c (register_operand): Same.
245 * regrename.c (mode_change_ok): Change to use new
246 CANNOT_CHANGE_MODE_CLASS infrastructure.
248 * regclass.c (cannot_change_mode_set_regs): New.
249 Declare subregs_of_mode.
250 (regclass): Use subregs_of_mode.
251 Remove references to reg_changes_mode.
252 (init_reg_sets_1): Remove class_can_change_mode and
253 reg_changes_mode code.
254 (invalid_mode_change_p): New.
255 (dump_regclass): Use invalid_mode_change_p instead of
256 class_can_change_mode.
258 (record_operand_costs): Do not set reg_changes_mode.
260 * local-alloc.c (struct qty): Remove changes_mode field.
261 (alloc_qty): Remove changes_mode initialization.
262 (update_qty_class): Remove set of changes_mode.
263 (find_free_reg): Use subregs_of_mode.
265 * global.c (find_reg): Use subregs_of_mode info.
267 * rtl.h (cannot_change_mode_set_regs): New prototype.
268 (invalid_mode_change_p): Same.
269 (REG_CANNOT_CHANGE_MODE_P): New macro.
271 * flow.c (mark_used_regs): Calculate subregs_of_mode. Remove
273 (life_analysis): Clear subregs_of_mode.
275 * combine.c (subst): Pass class to CLASS_CANNOT_CHANGE_MODE_P.
276 Remove use of CLASS_CANNOT_CHANGE_MODE.
277 (simplify_set): Same.
278 (gen_lowpart_for_combine): Calculate subregs_of_mode. Remove
281 * regs.h: Add extern for subregs_of_mode;
282 Include hard-reg-set and basic-block.
283 (REG_CHANGES_MODE): Delete.
285 2002-11-03 John David Anglin <dave@hiauly1.hia.nrc.ca>
287 * jump.c (never_reached_warning): Don't set contains_insn until the
288 first line note is seen.
290 2002-11-03 David Edelsohn <edelsohn@gnu.org>
292 * config/rs6000/rs6000.md (movti_string): Use string instructions.
294 2002-11-03 Roger Sayle <roger@eyesopen.com>
297 * c-typeck.c (c_expand_asm_operands): Defend against
298 error_mark_nodes in the output argument to avoid ICE.
300 2002-11-03 Eric Botcazou <ebotcazou@libertysurf.fr>
303 * genrecog.c (preds): Handle ADDRESSOF.
304 (validate_pattern): Mark it as an lvalue.
306 2002-11-02 David Edelsohn <edelsohn@gnu.org>
308 * config/rs6000/rs6000.c (rs6000_override_options): Use string
309 instructions when optimizing for size.
311 2002-11-02 Kazu Hirata <kazu@cs.umass.edu>
313 * config/h8300/h8300.h: Fix comment typos.
314 * config/h8300/h8300.md: Likewise.
315 * config/h8300/lib1funcs.asm: Likewise.
317 2002-11-02 Gerald Pfeifer <pfeifer@dbai.tuwien.ac.at>
321 *doc/install.texi (Installing GCC: Configuration): Clarify
322 the only supported ways to configure gcc.
324 2002-11-01 Kazu Hirata <kazu@cs.umass.edu>
326 * config/h8300/h8300.md (anonymous and:QI pattern): Use 'n'
327 instead of 'O' for the constraint for the second operand.
329 2002-11-01 Mark Mitchell <mark@codesourcery.com>
332 * toplev.c (rest_of_compilation): Do not refuse to output code for
333 an inline function in a local class.
335 2002-11-01 David O'Brien <obrien@FreeBSD.org>
337 * config/sparc/freebsd.h (CPP_CPU64_DEFAULT_SPEC): Define __arch64__.
338 (TRANSFER_FROM_TRAMPOLINE): Reformat.
341 2002-11-01 Kazu Hirata <kazu@cs.umass.edu>
343 * config/h8300/h8300.h (CAN_ELIMINATE): Simplify.
345 2002-11-01 Toshiyasu Morita <toshiyasu.morita@hsa.hitachi.com>
347 * config/h8300/h8300.h (OPTIMIZATION_OPTIONS): New.
349 2002-11-01 Steve Ellcey <sje@cup.hp.com>
351 * config/ia64/ia64.h (MASK_INLINE_DIV_LAT): Remove.
352 (MASK_INLINE_DIV_THR): Remove.
353 (TARGET_INLINE_DIV_LAT): Remove.
354 (TARGET_INLINE_DIV_THR): Remove.
355 (TARGET_INLINE_DIV): Remove.
356 (MASK_INLINE_FLOAT_DIV_LAT): New macro.
357 (MASK_INLINE_FLOAT_DIV_THR): New macro.
358 (MASK_INLINE_INT_DIV_LAT): New macro.
359 (MASK_INLINE_INT_DIV_THR): New macro.
360 (TARGET_INLINE_FLOAT_DIV_LAT): New macro.
361 (TARGET_INLINE_FLOAT_DIV_THR): New macro.
362 (TARGET_INLINE_INT_DIV_LAT): New macro.
363 (TARGET_INLINE_INT_DIV_THR): New macro.
364 (TARGET_INLINE_FLOAT_DIV): New macro.
365 (TARGET_INLINE_INT_DIV): New macro.
366 * config/ia64/ia64.md (divsi3): Change to use new macros.
370 (divsi3_internal): Ditto.
375 (divdi3_internal_lat): Ditto.
376 (divdi3_internal_thr): Ditto.
378 (divsf3_internal_lat): Ditto.
379 (divsf3_internal_thr): Ditto.
381 (divdf3_internal_lat): Ditto.
382 (divdf3_internal_thr): Ditto.
384 (divtf3_internal_lat): Ditto.
385 (divtf3_internal_thr): Ditto.
386 * config/ia64/ia64.c (ia64_override_options): Change
387 to check new macros for conflicts in settings.
388 * doc/invoke.texi (-minline-divide-min-latency): Remove.
389 (-minline-divide-max-throughput): Remove.
390 (-minline-float-divide-min-latency): New.
391 (-minline-float-divide-max-throughput): New.
392 (-minline-int-divide-min-latency): New.
393 (-minline-int-divide-max-throughput): New.
395 2002-11-01 Richard Earnshaw (rearnsha@arm.com)
398 * arm.c (use_return_insn): Don't use a return insn if there are
399 saved integer regs, but LR is not one of them.
401 Fri Nov 1 10:33:15 CET 2002 Jan Hubicka <jh@suse.cz>
403 * expr.c (emit_move_insn): Use SCALAR_FLOAT_MODE_P
404 * machmode.h (SCALAR_FLOAT_MODE_P): New macro.
406 2002-10-31 Nathanael Nerode <neroden@gcc.gnu.org>
409 * doc/md.texi: Document restriction on commutative operand
412 2002-10-31 Eric Christopher <echristo@redhat.com>
414 * explow.c (convert_memory_address): Use shallow_copy_rtx.
416 2002-10-31 Steve Ellcey <sje@cup.hp.com>
418 * expmed.c (store_bit_field): Check FUNCTION_ARG_REG_LITTLE_ENDIAN.
420 2002-10-31 Steve Ellcey <sje@cup.hp.com>
422 * config/ia64/hpux.h (MEMBER_TYPE_FORCES_BLK): Set for non-floats.
424 Thu Oct 31 Dale Johannesen <dalej@apple.com>
426 * config/rs6000/darwin.h: Correct formatting in previous.
428 Thu Oct 31 Dale Johannesen <dalej@apple.com>
430 * config/rs6000/darwin.h: Enable -falign-xxx options.
432 Thu Oct 31 18:08:00 CET 2002 Jan Hubicka <jh@suse.cz>
434 * i386.c (override_options): Set defaults for flag_omit_frame_pointer,
435 flag_asynchronous_unwind_tables, flag_pcc_struct_return.
436 * i386.c (optimization_options): Set flag_omit_frame_pointer,
437 flag_asynchronous_unwind_tables, flag_pcc_struct_return to 2.
438 Do not clear -momit-leaf-frame-pointer when profiling.
439 (ix86_frame_pointer_required): Frame pointer is allways required when
442 Thu Oct 31 16:09:44 CET 2002 Jan Hubicka <jh@suse.cz>
444 * i386.md (negdf2_ifs_rex64): Don't allow GPR operand.
446 Thu Oct 31 12:45:55 2002 J"orn Rennecke <joern.rennecke@superh.com>
448 * sh.h (binary_logical_operator): Declare.
449 * sh.c (binary_logical_operator): New function.
450 * sh.md (xordi3+1): New combiner splitter pattern.
452 2002-10-31 David O'Brien <obrien@FreeBSD.org>
454 * config/sparc/freebsd.h (TRANSFER_FROM_TRAMPOLINE): Define
455 __enable_execute_stack function.
457 2002-10-30 Aldy Hernandez <aldyh@redhat.com>
459 * c-common.c: Add GTY to vector_type_node_list.
461 2002-10-30 John David Anglin <dave@hiauly.hia.nrc.ca>
463 * pa-linux.h (ASM_OUTPUT_EXTERNAL_LIBCALL): Define.
464 * pa-protos.h (attr_length_millicode_call, attr_length_call,
465 pa_init_machine_status): Declare new global functions.
466 * pa.c (void copy_fp_args, length_fp_args, get_plabel): Declare and
467 implement new functions.
468 (attr_length_millicode_call, attr_length_call): Implement.
469 (total_code_bytes): Change type to long.
470 (pa_output_function_prologue): Compute total_code_bytes on TARGET_64BIT.
471 Reset counter if flag_function_sections.
472 (output_deferred_plabels): Set output alignment to 3 for TARGET_64BIT.
473 (output_cbranch): Move call to gen_label_rtx.
474 (output_millicode_call): Rewrite adding long TARGET_64BIT call, expose
475 delay slot in all variants, shorten pc-relative calls.
476 (output_call): Rewrite adding long TARGET_64BIT call, improved delay
477 slot usage and exposure, various new call variants, and shortened
478 sequences for some variants on TARGET_PA_20.
479 Miscellaneous format changes.
480 * pa.h (total_code_bytes): Change type to long.
481 (MASK_LONG_CALLS, TARGET_LONG_CALLS, TARGET_LONG_ABS_CALL,
482 TARGET_LONG_PIC_SDIFF_CALL, TARGET_LONG_PIC_PCREL_CALL): Define.
483 (TARGET_SWITCHES): Add "-mlong-calls" and "-mno-long-calls" options.
484 (EXTRA_CONSTRAINT, GO_IF_LEGITIMATE_ADDRESS,
485 LEGITIMIZE_RELOAD_ADDRESS): Don't use long floating point loads and
486 stores on TARGET_ELF32.
487 *pa.md (define_delay): Allow insns in delay on TARGET_PORTABLE_RUNTIME.
488 (unnamed patterns for mulsi3, divsi3, udivsi3, modsi3, umodsi3 and
489 canonicalize_funcptr_for_compare expanders): Calculate attribute length
490 attr_length_millicode_call().
491 (call_internal_symref, call_value_internal_symref): Clobber register 1.
492 Calculate attribute length using attr_length_call().
493 (call_internal_reg_64bit, call_value_internal_reg_64bit): Move gp load
495 (sibcall, sibcall_value): Rewrite.
496 (sibcall_internal_symref, sibcall_value_internal_symref): Clobber
497 register 1. Use attr_length_call().
498 (sibcall_internal_symref_64bit, sibcall_value_internal_symref_64bit):
500 (unamed pattern for canonicalize_funcptr_for_compare): Rewrite.
501 * som.h (MEMBER_TYPE_FORCES_BLK): Define.
502 * t-pa64 (TARGET_LIBGCC2_CFLAGS): Add "-mlong-calls".
503 * doc/invoke.texi (mlong-calls): Document.
505 2002-10-30 Roger Sayle <roger@eyesopen.com>
507 * fold-const.c (fold_binary_op_with_conditional_arg): Improve
508 handling of cases where one or both branches of the conditional
509 have void type, i.e. throw an exception or don't return.
510 (fold): Only apply (and undo) type conversion to the non-void
511 branches of a COND_EXPR.
513 2002-10-30 Mark Mitchell <mark@codesourcery.com>
516 * varasm.c (asm_output_aligned_bss): Do not call
519 2002-10-30 David Edelsohn <edelsohn@gnu.org>
520 Torbjorn Granlund <tege@swox.com>
522 * config/rs6000/rs6000.md (load_toc_v4_PIC_1): Use preferred form
524 (load_toc_v4_PIC_1b): Same.
526 2002-10-30 Kazu Hirata <kazu@cs.umass.edu>
528 * config/h8300/h8300.c (h8300_eightbit_constant_address_p):
529 Truncate the addresses for H8/300 using HImode.
531 2002-10-29 Hans-Peter Nilsson <hp@bitrange.com>
533 * toplev.c (rest_of_type_compilation): Return early in case of
535 (check_global_declarations): Don't call debug_hooks->global_decl
538 2002-10-28 Andreas Bauer <baueran@in.tum.de>
540 * doc/c-tree.texi (Tree overview): Fix typos.
542 2002-10-29 Phil Edwards <pme@gcc.gnu.org>
544 * Makefile.in (gnucompare*): Only record bad comparisons
545 if there really was a bad comparison.
547 Tue Oct 29 19:32:16 CET 2002 Jan Hubicka <jh@suse.cz>
549 * i386.h (CONST_DOUBLE_OK_FOR_LETTER_P): Remove 'H'
550 * i386.md (movsf*, movdf*): Use 'C' instead of 'H'
551 * md.texi (machine dependent constraints): Document 'C'
553 * simplify-rtx.c (simplify_subreg): Fix const_int->vector subregging.
555 * i386.c (ix86_expand_vector_move): Fix.
557 * i386.c (ix86_expand_builtin): Use sse2_maskmovdqu_rex64.
558 * i386.md (sse2_maskmovdqu_rex64): New pattern
561 * xmmintrin.h (_mm_stream_pi, _mm_stream_pd): Fix cast.
562 (ix86_init_mmx_sse_builtins): Fix type.
564 2002-10-29 Jason Thorpe <thorpej@wasabisystems.com>
566 * gthr-posix.h: Include <unistd.h> for feature tests.
567 (sched_get_priority_max, sched_get_priority_min)
568 (pthread_getschedparam, pthread_setschedparam): Only use
569 if _POSIX_THREAD_PRIORITY_SCHEDULING is defined.
570 (__gthread_objc_thread_set_priority): Don't treat all non-zero
571 returns from sched_get_priority_max and sched_get_priority_min
574 2002-10-29 Kazu Hirata <kazu@cs.umass.edu>
576 * config/h8300/h8300.h (TARGET_DEFAULT): Make it
579 2002-10-29 Kazu Hirata <kazu@cs.umass.edu>
581 * config/h8300/h8300.c (h8300_eightbit_constant_address_p): New.
582 (h8300_tiny_constant_address_p): Likewise.
583 * config/h8300/h8300.h (EIGHTBIT_CONSTANT_ADDRESS_P): Use
584 h8300_eightbit_constant_address_p.
585 (TINY_CONSTANT_ADDRESS_P): Use h8300_tiny_constant_address_p.
586 * config/h8300/h8300-protos.h: Add the prototypes for the two
589 2002-10-29 Kazu Hirata <kazu@cs.umass.edu>
591 * reload1.c (update_eliminables): Unconditionally check if
592 frame_pointer_needed has changed.
594 Tue Oct 29 15:37:39 CET 2002 Jan Hubicka <jh@suse.cz>
596 * toplev.c (rest_of_compilation): Reorganize way reg_scan is called
599 2002-10-29 Eric Botcazou <ebotcazou@libertysurf.fr>
601 * expr.c (expand_expr) [PLUS]: Don't use simplify_binary_operation;
602 check for zero operands explicitly.
604 2002-10-29 Richard Sandiford <rsandifo@redhat.com>
606 * config/mips/mips.md (extv, extzv, insv): Set size of referenced
607 memory after adjusting to BLKmode.
609 2002-10-29 Kazu Hirata <kazu@cs.umass.edu>
611 * config/h8300/h8300.h (MASK_*): New.
612 (TARGET_*): Use MASK_*.
614 2002-10-28 Jason Thorpe <thorpej@wasabisystems.com>
616 * config.gcc (*-*-netbsd*): Add NETBSD_ENABLE_PTHREADS to
617 tm_defines if pthreads are enabled.
618 * config/netbsd.h (LIB_SPEC): Only support the -pthread option
619 if NETBSD_ENABLE_PTHREADS is defined.
621 2002-10-28 Kazu Hirata <kazu@cs.umass.edu>
623 * ChangeLog.1: Fix typos.
624 * cse.c: Fix a comment typo.
625 * reload1.c: Likewise.
627 2002-10-27 Hans-Peter Nilsson <hp@bitrange.com>
629 * fixinc/inclhack.def (libc1_G_va_list): Correct test_text.
630 * fixinc/tests/base/_G_config.h: New file.
632 2002-10-27 Kazu Hirata <kazu@cs.umass.edu>
634 * combine.c: Fix comment formatting.
637 * regclass.c: Likewise.
638 * regmove.c: Likewise.
639 * regrename.c: Likewise.
640 * reg-stack.c: Likewise.
641 * reload1.c: Likewise.
642 * reload.c: Likewise.
643 * reload.h: Likewise.
644 * unroll.c: Likewise.
646 2002-10-27 Kazu Hirata <kazu@cs.umass.edu>
648 * reload1.c (reload): Fix a comment typo.
650 Sun Oct 27 10:15:24 CET 2002 Jan Hubicka <jh@suse.cz>
652 * linux64.h (DEFAULT_PCC_STRUCT_RETURN): Define.
654 2002-10-27 Zdenek Dvorak <rakdver@atrey.karlin.mff.cuni.cz>
656 * Makefile.in (dwarf2out.o): Add dependendcy on hashtab.h.
657 * dwarf2out.c: Include hashtab.h.
658 (is_main_source): New static variable.
659 (attr_checksum, die_checksum): Modified to handle die references.
660 (same_loc_p, same_dw_val_p, same_attr_p, same_die_p, same_die_p_wrap,
661 unmark_all_dies, htab_cu_hash, htab_cu_eq, htab_cu_del, check_duplicate_cu,
662 record_comdat_symbol_number): New static functions.
663 (output_comp_unit, compute_section_prefix, is_type_die, break_out_includes,
664 mark_dies, unmark_dies, dwarf2out_start_source_file): Modified.
665 * toplev.c (rest_of_decl_compilation): Call of dwarf2out_decl for type
668 2002-10-26 Kazu Hirata <kazu@cs.umass.edu>
670 * config/h8300/h8300.c (initial_offset): Change to
671 h8300_initial_elimination_offset.
672 * config/h8300/h8300.h (INITIAL_ELIMINATION_OFFSET): Use
673 h8300_initial_elimination_offset.
674 * config/h8300/h8300-protos.h: Update the prototype.
676 2002-10-26 Hans-Peter Nilsson <hp@bitrange.com>
678 * config/mmix/mmix.h (LIBCALL_VALUE): Use
679 MMIX_RETURN_VALUE_REGNUM, not MMIX_OUTGOING_RETURN_VALUE_REGNUM.
680 (FUNCTION_VALUE_REGNO_P): Similar, but move code to...
681 * config/mmix/mmix.c (mmix_function_value_regno_p): New.
682 * config/mmix/mmix-protos.h: Remove needless ifdefs on TREE_CODE
684 (mmix_function_value_regno_p): Declare.
686 * config/mmix/mmix.md ("fixuns_truncdfdi2"): Replace unsigned_fix,
687 invalid for floating point mode result, with fix.
689 2002-10-25 Mike Stump <mrs@apple.com>
691 Fixes gcc.dg/warn-1.c.
692 * c-typeck.c (warn_for_assignment): Don't print argument number, if zero.
694 Sat Oct 26 01:44:46 CEST 2002 Jan Hubicka <jh@suse.cz>
696 * toplev.c (dump_file_index): Add DFI_ce3.
697 (dump_file_info): Likewise.
698 (rest_of_compilation): Run first ifcvt pass before tracer.
700 2002-10-25 Steve Ellcey <sje@cup.hp.com>
702 * config/ia64/hpux.h (BITS_BIG_ENDIAN): Remove.
704 2002-10-25 Richard Henderson <rth@redhat.com>
706 * real.c (real_to_decimal): If the >1 tens reduction loop results
707 in a negative exponent, fall into the <1 pten computation.
709 2002-10-25 Zack Weinberg <zack@codesourcery.com>
712 * c-objc-common.c (inline_forbidden_p): Can not inline
713 functions containing structures or unions containing VLAs.
714 * tree-inline.c (walk_tree): For all class 't' nodes, walk
715 TYPE_SIZE and TYPE_SIZE_UNIT.
716 (copy_tree_r): Copy types if they are variably modified.
718 2002-10-25 Ulrich Weigand <uweigand@de.ibm.com>
720 * config/s390/s390.md: Remove old-style peepholes.
722 2002-10-25 Ulrich Weigand <uweigand@de.ibm.com>
724 * config/s390/s390.c (s390_decompose_address): Do not range check the
725 displacement if base or index is the argument pointer register.
727 2002-10-24 Hans-Peter Nilsson <hp@bitrange.com>
732 * fixinc/inclhack.def (libc1_G_va_list): New fix.
733 * fixinc/fixincl.x: Regenerate.
734 * config/i386/linux.h: Move MD_FALLBACK_FRAME_STATE_FOR inside
735 ifndef IN_LIBGCC2. Wrap it together with signal.h and
736 sys/ucontext.h inclusion in ifndef USE_GNULIBC_1.
737 * configure.in (gcc_AC_CHECK_DECLS): Check vasprintf too.
738 * config.in, configure: Regenerate.
740 2002-10-24 Igor Shevlyakov <igor@microunity.com>
742 * varasm.c (struct rtx_const): Array size 16 for V16QImode.
744 2002-10-24 Richard Henderson <rth@redhat.com>
746 * config/i386/i386.c (x86_output_mi_thunk): Fix x86_64 pic jump.
748 2002-10-24 Kazu Hirata <kazu@cs.umass.edu>
750 * config/h8300/h8300.c (initial_offset): Simplify by using
753 2002-10-24 Marek Michalkiewicz <marekm@amelek.gda.pl>
755 * doc/install.texi (avr): Update required binutils version.
757 2002-10-24 Theodore A. Roth <troth@openavr.org>
759 * doc/install.texi: Point avr users at more up-to-date information.
761 2002-10-24 Ulrich Weigand <uweigand@de.ibm.com>
763 * config/s390/s390.md (movdi, movsi, movhi, movqi): Add peepholes2
764 to pull operands out of the literal pool where possible.
766 2002-10-24 Denis Chertykov <denisc@overta.ru>
768 * config/avr/avr.c (init_cumulative_args): Test fntype for zero.
770 2002-10-24 Steve Ellcey <sje@cup.hp.com>
772 * expr.c (convert_move): If unsignedp is less then zero there
773 is no equivalent code.
775 2002-10-24 Zack Weinberg <zack@codesourcery.com>
777 * tree.def: Delete mention of nonexistent ARRAY_TYPE fields.
779 2002-10-24 Ulrich Weigand <uweigand@de.ibm.com>
781 * config/s390/s390.h: Rework comments; re-sort target macro definitions
782 according to the sequence they are defined in the manual.
783 (POINTER_BOUNDARY): Remove.
785 2002-10-24 Kazu Hirata <kazu@cs.umass.edu>
787 * config/h8300/h8300.c (round_frame_size): Replace 8 with
790 2002-10-24 Kazu Hirata <kazu@cs.umass.edu>
792 * config/h8300/h8300.h (EIGHTBIT_CONSTANT_ADDRESS_P): Make it
794 (TINY_CONSTANT_ADDRESS_P): Likewise.
796 2002-10-24 Richard Henderson <rth@redhat.com>
798 * config/ia64/ia64.c (TARGET_ASM_CAN_OUTPUT_MI_THUNK): True.
799 (ia64_output_mi_thunk): Rewrite to use rtl, and to handle the
802 2002-10-24 Richard Henderson <rth@redhat.com>
805 * reload.c (find_reloads_toplev): Mode of X is not important
806 when simplifying subregs of constants.
808 2002-10-24 Richard Sandiford <rsandifo@redhat.com>
810 * config.gcc (mips64vr-*-elf*, mips64vrel-*-elf*): Add
811 MIPS_MARCH_CONTROLS_SOFT_FLOAT=1 to $tm_defines.
812 * config/mips/mips.c (MIPS_MARCH_CONTROLS_SOFT_FLOAT): Default to 0.
813 (override_options): Base default setting of MASK_SOFT_FLOAT on -march
814 if MIPS_MARCH_CONTROLS_SOFT_FLOAT.
816 2002-10-24 Richard Sandiford <rsandifo@redhat.com>
818 * optabs.c (expand_binop): Don't reuse the shift target in the
819 middle of shift sequences.
821 2002-10-23 Ziemowit Laski <zlaski@apple.com>
823 * objc/objc-act.c (get_static_reference): Remove unneeded
824 TYPE_BINFO initialization.
825 (get_object-reference): Likewise.
826 (build_constructor): Tighten precondition check.
827 (finish_message_expr): Likewise.
829 2002-10-23 Jakub Jelinek <jakub@redhat.com>
831 * config/i386/i386.c (local_symbolic_operand): Move LABEL_REF test
834 2002-10-23 Steve Ellcey <sje@cup.hp.com>
836 * config/ia64/ia64.c (hfa_element_mode): Don't allow 128 bit floats
839 2002-10-23 Richard Henderson <rth@redhat.com>
841 * config/alpha/alpha.c (TARGET_ASM_CAN_OUTPUT_MI_THUNK): True.
842 (alpha_output_mi_thunk_osf): Handle vcall_offset.
844 2002-10-23 Zack Weinberg <zack@codesourcery.com>
846 * langhooks.h (struct lang_hooks_for_tree_inlining): Add
848 * langhooks-def.h: Default for tree_inlining.var_mod_type_p is
849 hook_tree_bool_false.
851 * tree.c (variably_modified_type_p): Moved here from
852 cp/tree.c. Use lang_hooks.tree_inlining.var_mod_type_p for
853 language-specific cases. Due to this, must weaken some 'if
854 and only if' checks to merely 'if'.
855 * tree.h: Prototype variably_modified_type_p.
857 * tree-inline.c (walk_tree): #undef WALK_SUBTREE_TAIL at end.
859 2002-10-23 Ulrich Weigand <uweigand@de.ibm.com>
861 * config/s390/linux.h (CC1_SPEC, CC1PLUS_SPEC): Remove.
862 * config/s390/s390.c (optimization_options): Disable -fcaller-saves.
864 * config/s390/s390-protos.h (fp_operand): Remove.
865 * config/s390/s390.c (fp_operand): Remove.
866 * config/s390/s390.md ("movdi"): Replace fp_operand by FP_REG_P.
867 ("*movdi_lhi", "*movdi_lli", "*movdi_larl"): Likewise.
868 ("movsi", "*movsi_lhi", "*movsi_lli"): Likewise.
869 (movdi_31, movdf_31 splitters): Likewise.
871 * config/s390/s390.h (IEEE_FLOAT): Remove.
872 (TARGET_FLOAT_FORMAT): Define in terms of TARGET_IEEE_FLOAT.
873 (INT_REGNO_P): Rename to ...
874 (GENERAL_REGNO_P): ... this.
875 (FLOAT_REGNO_P): Rename to ...
876 (FP_REGNO_P): ... this.
877 (ADDR_REGNO_P): New macro.
878 (GENERAL_REG_P, ADDR_REG_P, FP_REG_P, CC_REG_P): New macros.
879 (REGNO_OK_FOR_DATA_P, REGNO_OK_FOR_FP_P): Remove.
880 (DATA_REG_P, FP_REG_P, ADDRESS_REG_P): Likewise.
881 (HARD_REGNO_NREGS): Adapt to macro renaming.
882 (HARD_REGNO_MODE_OK): Likewise.
884 2002-10-23 David Edelsohn <edelsohn@gnu.org>
885 Geoff Keating <geoffk@apple.com>
887 * config/rs6000/rs6000.c (rs6000_register_move_cost): New function.
888 (rs6000_memory_move_cost): New function.
889 * config/rs6000/rs6000-protos.h: Declare them.
890 * config/rs6000/rs6000.h: Use them.
892 2002-10-23 Ulrich Weigand <uweigand@de.ibm.com>
894 * libgcc2.c (__udiv_w_sdiv): Use attribute ((always_inline)) when
895 inlining it into other libgcc2 routines.
896 (__udivmoddi4): Likewise.
898 2002-10-22 Nathanael Nerode <neroden@gcc.gnu.org>
900 * doc/sourcebuild.texi (Test Suites): Improve.
902 2002-10-22 Stan Shebs <shebs@apple.com>
904 * config/rs6000/rs6000.c (rs6000_output_mi_thunk): Add missing
907 2002-10-22 Jim Wilson <wilson@redhat.com>
909 * config/i386/i386.md (subdi3_1): Add call to ix86_binary_operator_ok.
911 Wed Oct 23 01:52:36 CEST 2002 Jan Hubicka <jh@suse.cz>
914 * xmmintrin.h: Add const to the argument of loads.
916 * i386.md (pushv2di): New pattern.
918 * xmmintrin.h (_MM_TRANSPOSE4_PS): New.
920 2002-10-22 Richard Henderson <rth@redhat.com>
922 * target.h (gcc_target.asm_out): Merge output_mi_thunk and
923 output_mi_vcall_thunk into a single hook. Add can_output_mi_thunk.
924 * target-def.h (TARGET_ASM_OUTPUT_MI_THUNK): Don't conditionalize.
925 (TARGET_ASM_OUTPUT_MI_VCALL_THUNK): Remove.
926 (TARGET_ASM_CAN_OUTPUT_MI_THUNK): New.
927 (TARGET_ASM_OUT): Update.
928 * hooks.c (hook_bool_tree_hwi_hwi_tree_false): New.
929 (hook_bool_tree_hwi_hwi_tree_true): New.
930 (default_can_output_mi_thunk_no_vcall): New.
931 * hooks.h: Declare them.
932 * system.h (ASM_OUTPUT_MI_THUNK): Poison.
934 * config/alpha/alpha.c (TARGET_ASM_CAN_OUTPUT_MI_THUNK): New.
935 (alpha_output_mi_thunk_osf): Add VCALL_OFFSET parameter.
936 * config/arm/arm.c, config/cris/cris.c, config/frv/frv.c,
937 config/i960/i960.c, config/ia64/ia64.c, config/m68k/m68k.c,
938 config/mmix/mmix.c, config/pa/pa.c, config/sparc/sparc.c,
939 config/stormy16/stormy16.c: Similarly.
941 * config/i386/i386.c (x86_output_mi_thunk): Merge vcall_offset code.
942 Handle 64-bit properly. Streamline.
943 (x86_output_mi_vcall_thunk): Remove.
944 (x86_this_parameter): Rename from ia32_this_parameter; handle 64-bit.
945 (x86_can_output_mi_thunk): New.
946 (TARGET_ASM_OUTPUT_MI_VCALL_THUNK): Remove.
947 (TARGET_ASM_CAN_OUTPUT_MI_THUNK): New.
948 (override_options): Don't zap targetm.asm_out.output_mi_vcall_thunk.
950 * config/rs6000/rs6000.c (rs6000_output_mi_thunk): Rename from
951 output_mi_thunk; make static; always use function_section.
952 (TARGET_ASM_OUTPUT_MI_THUNK): New.
953 (TARGET_ASM_CAN_OUTPUT_MI_THUNK): New.
954 (rs6000_ra_ever_killed): Test no_new_pseudos not
955 targetm.asm_out.output_mi_thunk in conjunction with thunks.
956 * config/rs6000/rs6000-protos.h: Update.
957 * config/rs6000/sysv4.h (TARGET_ASM_OUTPUT_MI_THUNK): Remove.
958 * config/rs6000/xcoff.h (ASM_DECLARE_FUNCTION_NAME): Don't call
959 xcoffout_declare_function when using rs6000_output_mi_thunk.
961 * config/s390/s390.c (s390_output_mi_thunk): Rename from
962 s390_output_mi_vcall_thunk.
963 (TARGET_ASM_OUTPUT_MI_THUNK): Remove.
964 (TARGET_ASM_CAN_OUTPUT_MI_THUNK): New.
966 * config/vax/vax.c (vax_output_mi_thunk): Static; add vcall_offset.
967 (TARGET_ASM_OUTPUT_MI_THUNK, TARGET_ASM_CAN_OUTPUT_MI_THUNK): New.
968 * config/vax/vax-protos.h: Update.
969 * config/vax/vax.h (ASM_OUTPUT_MI_THUNK): Remove.
971 Wed Oct 23 00:33:11 CEST 2002 Jan Hubicka <jh@suse,cz>
973 * i386.c (standard_sse_constant_p): Accept vector and integer zeros too.
974 * i386.h (EXTRA_CONSTRAINT): Recognize 'C'
975 * i386.md (movti_internal): Use 'C'
977 * xmmintrin.h (_mm_cmplt_epi*): New.
979 2002-10-22 Ulrich Weigand <uweigand@de.ibm.com>
981 * config/s390/s390.md ("*movdi_64"): Fix op_type attribute.
982 ("*movdf_64"): Likewise.
983 ("*lshrdi3_64"): Likewise.
984 ("blockage"): Add length attribute.
987 Tue Oct 22 23:51:34 CEST 2002 Jan Hubicka <jh@suse.cz>
990 (sse2_cvtsi2sd, sse2_pslrdq): Fix template.
991 (sse2_umulv2siv2di3): Fix predicate.
992 (sse2_psadbw, ashrv8hi3, ashrv4si3, lshrv8hi3 lshrv4si3,
993 lshrv2di3, ashlv8hi3, ashlv4si3, ashlv2di3): Likewise.
994 * xmmintrin.h (_mm_mul_epu16): Rename to...
995 (_mm_mul_epu32): This one.
996 (_mm_cvtsi32_si128, _mm_cvtsi128_si32): New.
998 (contains_128bit_aligned_vector_p): Undo accidental checkin.
1000 2002-10-22 Eric Christopher <echristo@redhat.com>
1002 * config/sparc/sparc.h: Add #error.
1004 2002-10-22 Ulrich Weigand <uweigand@de.ibm.com>
1006 * config.gcc [s390-*-linux]: Remove s390/t-linux from tmake_file.
1007 [s390x-*-linux*]: Likewise.
1008 * config/s390/t-linux: Remove.
1009 * config/s390/s390.h: Include fixdfdi.h when building libgcc2.
1011 Tue Oct 22 19:07:03 CEST 2002 Jan Hubicka <jh@suse.cz>
1013 * i386.c (builtin_description): Add IX86_BUILTIN_PUNPCKHQDQ128.
1014 (ix86_expand_builtin): Fix MASKMOVDQU expasion.
1015 * i386.h (ix86_builtins): Add IX86_BUILTIN_PUNPCKHQDQ128.
1016 * i386.md (mmx_punpck?dq): Simplify.
1017 (sse2_pubpcklqdq): Fix.
1018 (sse2_pubpckhqdq): New.
1019 * xmmintrin.h (_mm_unpackhi_epi32): New.
1021 * xmmintrin.h (_mm_cvt*, _mm_stream_pd): Fix prototypes.
1022 (_mm_shufflehi_epi16, _mm_shufflelo_epi16): Fix typo.
1024 2002-10-22 Nathan Sidwell <nathan@codesourcery.com>
1027 * fold_const.c (fold_binary_op_with_conditional_arg): Always
1028 build compound_expr if we used save_expr.
1030 2002-10-22 Alan Modra <amodra@bigpond.net.au>
1032 * output.h (SECTION_NOTYPE): Define.
1033 * varasm.c (default_section_type_flags_1): Set SECTION_NOTYPE for
1034 init array sections.
1035 (default_elf_asm_named_section): Mind SECTION_NOTYPE.
1036 * config/arm/arm.c (arm_elf_asm_named_section): Likewise. Also
1039 2002-10-21 Richard Henderson <rth@redhat.com>
1041 * real.c (sticky_rshift_significand): Return inexact, don't
1042 or it in immediately.
1043 (sub_significands): Accept incomming carry.
1044 (div_significands, rtd_divmod): Update for sub_significands change.
1045 (round_for_format): Update for sticky_rshift_significand change.
1046 (do_add): Don't involve the inexact bit in addition, do give the
1047 inexact bit as the subtraction carry-in.
1048 (encode_internal, decode_internal, real_internal_format): New.
1049 * real.h (real_internal_format): Declare.
1051 2002-10-21 Ulrich Weigand <uweigand@de.ibm.com>
1053 * libgcc2.c: Fix __udiv_w_sdiv breakage on platforms that
1054 don't define sdiv_qrnnd.
1056 2002-10-21 Kazu Hirata <kazu@cs.umass.edu>
1058 * config/h8300/h8300.h (EIGHTBIT_CONSTANT_ADDRESS_P): Simplify
1060 (TINY_CONSTANT_ADDRESS_P): Likewise.
1062 Tue Oct 22 00:04:20 CEST 2002 Jan Hubicka <jh@suse.cz>
1064 * i386.c (builtin_description): Add punpcklqdq and movdq2q
1065 (ix86_init_mmx_sse_builtins): Add v2di_ftype_void, di_ftype_v2di,
1066 v16qi_ftype_pchar, void_ftype_pchar_v16qi, v4si_ftype_pchar,
1067 void_ftype_pchar_v4si; Initialize __builtin_ia32_movdq2q,
1068 __builtin_ia32_loaddqa, __builtin_ia32_loaddqu, __builtin_ia32_loadd
1069 __builtin_ia32_storedqa, __builtin_ia32_storedqu, __builtin_ia32_stored
1070 __builtin_ia32_setzero128.
1071 (ix86_expand_builtin): Handle IX86_BUILTIN_CLRTI, IX86_BUILTIN_LOADDQA,
1072 IX86_BUILTIN_LOADDQU, IX86_BUILTIN_LOADD, IX86_BUILTIN_STOREDQA,
1073 IX86_BUILTIN_STOREDQU, IX86_BUILTIN_STORED, Ix86_BUILTIN_MOVQ.
1074 * i386.h (ix86_builtins): Add IX86_BUILTIN_LOADDQA, IX86_BUILTIN_LOADDQU,
1075 IX86_BUILTIN_STOREDQA, IX86_BUILTIN_STOREDQU, IX86_BUILTIN_LOADD,
1076 IX86_BUILTIN_STORED, IX86_BUILTIN_CLRTI, IX86_BUILTIN_MOVDQ2Q,
1077 IX86_BUILTIN_PUNPCKLQDQ128, Ix86_BUILTIN_MOVQ.
1078 * i386.md (sse2_punpcklqdq, sse2_movqsse2_loadd, sse2_stored,
1079 sse2_movq): New patterns.
1080 (sse2_movdqa, sse2_movdqu, sse2_movdq2q): Fix.
1081 * xmmintrin.h (_mm_load_si128, _mm_loadu_si128, _mm_loadl_epi64,
1082 _mm_store_si128, _mm_storeu_si128, _mm_storel_epi64,
1083 _mm_setzero_si128, _mm_set_epi64, _mm_set_epi32, _mm_set_epi16,
1084 _mm_set_epi8, _mm_set1_epi64, _mm_set1_epi32, _mm_set1_epi16,
1085 _mm_set1_epi8, _mm_setr_epi64, _mm_setr_epi32, _mm_setr_epi16,
1086 _mm_setr_epi8, _mm_unpacklo_epi64,_mm_set_moveq): New functions.
1087 (_mm_insert_epi16): Fix.
1089 2002-10-21 Dale Johannesen <dalej@apple.com>
1091 * config/rs6000/rs6000.c (rs6000_reverse_condition): Handle
1092 unsafe math reversals correctly for RTL generation.
1093 (output_cbranch): Replace rs6000_reverse_condition call
1094 by its former definition.
1096 2002-10-21 Jakub Jelinek <jakub@redhat.com>
1098 * config/i386/i386.c (x86_64_sign_extended_value): Add allow_rip
1099 argument. In CM_SMALL_PIC model consider SYMBOL_REFs binding locally or
1100 from constant pool or LABEL_REFs as sign extended if allow_rip.
1101 Change all +-1GB limits to +-16MB.
1102 (x86_64_general_operand, x86_64_szext_general_operand,
1103 x86_64_nonmemory_operand, x86_64_movabs_operand,
1104 x86_64_szext_nonmemory_operand, x86_64_immediate_operand,
1105 legitimate_address_p, ix86_expand_int_movcc): Update callers.
1106 (local_symbolic_operand): Don't allow offsets bigger than +-16MB
1107 in CM_SMALL_PIC model.
1108 (legitimate_pic_address_disp_p): Don't check offsets before
1109 calling local_symbolic_operand.
1110 (legitimize_pic_address): Force offsets bigger than +-16MB into
1112 * config/i386/i386.h (EXTRA_CONSTRAINT, CONST_COSTS): Likewise.
1113 * config/i386/i386-protos.h (x86_64_sign_extended_value): Update
1116 * configure.in: Test for @GOTNTPOFF and @INDNTPOFF on IA-32 too.
1117 Add x86-64 test. Set tls_first_minor to 14 on IA-32 and x86-64.
1118 * configure: Rebuilt.
1119 * config/i386/i386.c (x86_64_sign_extended_value): Don't allow TLS
1120 SYMBOL_REFs unless enclosed in UNSPEC. Handle UNSPEC_DTPOFF,
1121 UNSPEC_GOTNTPOFF and UNSPEC_NTPOFF.
1122 (legitimate_address_p): Allow foo@dtpoff(base) even on TARGET_64BIT
1124 (ix86_encode_section_info): Don't ever generate TLSGD or TLSLD for
1125 non-pic code if TARGET_64BIT.
1126 (legitimize_address): Generate 64-bit TLS sequences.
1127 (output_pic_addr_const): Support x86-64 TLS operators.
1128 (i386_output_dwarf_dtprel): Output 64-bit DTPOFF as .long f@DTPOFF, 0.
1129 (print_operand_address): Use %fs instead of %gs on TARGET_64BIT.
1130 Don't append (%rip) in 64-bit TLSGD and TLSLD sequences.
1131 (output_addr_const_extra): Support x86-64 TLS operators.
1132 (maybe_get_pool_constant): Handle TARGET_64BIT -fpic.
1133 (ix86_tls_get_addr): Use __tls_get_addr on TARGET_64BIT
1135 * config/i386/i386.md (*tls_global_dynamic_gnu): Renamed to...
1136 (*tls_global_dynamic_32_gnu): ..., add !TARGET_64BIT.
1137 (*tls_global_dynamic_sun): Renamed to...
1138 (*tls_global_dynamic_32_sun): ..., add !TARGET_64BIT.
1139 (tls_global_dynamic): Renamed to...
1140 (tls_global_dynamic_32): ... this.
1141 (tls_global_dynamic_64, *tls_global_dynamic_64): New.
1142 (*tls_local_dynamic_base_dynamic_gnu): Renamed to...
1143 (*tls_local_dynamic_base_dynamic_32_gnu): ..., add !TARGET_64BIT.
1144 (*tls_local_dynamic_base_dynamic_sun): Renamed to...
1145 (*tls_local_dynamic_base_dynamic_32_sun): ..., add !TARGET_64BIT.
1146 (tls_local_dynamic_base_dynamic): Renamed to...
1147 (tls_local_dynamic_base_dynamic_32): ... this.
1148 (tls_local_dynamic_base_dynamic_64,
1149 *tls_local_dynamic_base_dynamic_64): New.
1150 (*tls_local_dynamic_once): Renamed to...
1151 (*tls_local_dynamic_32_once): ... this.
1153 2002-10-21 Ulrich Weigand <uweigand@de.ibm.com>
1155 * libgcc2.c: Inline __udiv_w_sdiv when compiling __udivdi3,
1156 __divdi3, __umoddi3, or __moddi3.
1158 2002-10-21 Ulrich Weigand <uweigand@de.ibm.com>
1160 * c-opts.c (missing_arg): Use cl_options[opt_index].opt_code
1161 instead of just opt_index as switch expression.
1163 * calls.c (store_one_arg): Change type of 'excess_align'
1166 * profile.c (output_gcov_string): Change type of 'temp'
1169 2002-10-21 Ulrich Weigand <uweigand@de.ibm.com>
1171 * config/s390/fixdfdi.h (__fixunsdfdi, __fixdfdi): Add prototypes.
1172 (__fixunssfdi, __fixsfdi): Likewise.
1173 * config/s390/s390.c (s390_single_hi): Initialize 'value'.
1174 (s390_single_qi): Likewise.
1175 (s390_emit_epilogue): Initialize 'offset'. Remove signed vs.
1176 unsigned comparison warning.
1177 (s390_return_addr_rtx): New function.
1178 * config/s390/s390-protos.h (s390_return_addr_rtx): Declare it.
1179 * config/s390/s390.h (RETURN_ADDR_RTX): Use it.
1180 (HARD_REGNO_MODE_OK): Rewrite condition to silence warnings.
1182 2002-10-21 Ulrich Weigand <uweigand@de.ibm.com>
1184 * config/s390/s390.c (s390_output_mi_vcall_thunk): New function.
1185 (TARGET_ASM_OUTPUT_MI_VCALL_THUNK): Define target hook.
1186 (s390_output_mi_thunk): Remove.
1187 (TARGET_ASM_OUTPUT_MI_THUNK): Remove.
1189 2002-10-21 Kazu Hirata <kazu@cs.umass.edu>
1191 * config/h8300/h8300.h (N_REG_CLASSES): Parenthesize.
1193 2002-10-20 Richard Henderson <rth@redhat.com>
1195 * target.h (struct gcc_target): Line wrap.
1197 * config/alpha/alpha.c (alpha_output_mi_thunk_osf): Static.
1198 (TARGET_ASM_OUTPUT_MI_THUNK): Define here...
1199 * config/alpha/alpha.h: ... not here.
1200 * config/alpha/alpha-protos.h: Update.
1202 * config/arm/arm.c, config/arm/arm.h, config/arm/arm-protos.h
1203 config/cris/cris-protos.h, config/cris/cris.c, config/cris/cris.h,
1204 config/frv/frv-protos.h, config/frv/frv.c, config/frv/frv.h,
1205 config/i386/i386-protos.h, config/i386/i386.c, config/i386/openbsd.h,
1206 config/i386/unix.h, config/i960/i960-protos.h, config/i960/i960.c,
1207 config/i960/i960.h, config/ia64/ia64-protos.h, config/ia64/ia64.c,
1208 config/ia64/ia64.h, config/m68k/linux.h, config/m68k/m68k-protos.h,
1209 config/m68k/m68k.c, config/m68k/netbsd-elf.h, config/m68k/openbsd.h,
1210 config/mmix/mmix-protos.h, config/mmix/mmix.c, config/mmix/mmix.h,
1211 config/pa/pa-protos.h, config/pa/pa.c, config/pa/pa.h,
1212 config/s390/s390-protos.h, config/s390/s390.c, config/s390/s390.h,
1213 config/sparc/openbsd.h, config/sparc/sparc-protos.h,
1214 config/sparc/sparc.c, config/sparc/sparc.h,
1215 config/stormy16/stormy16-protos.h, config/stormy16/stormy16.c,
1216 config/stormy16/stormy16.h: Similarly.
1218 * config/m68k/m68k.c (m68k_output_mi_thunk): Replicate mnemonic
1219 selection logic from call patterns.
1221 2002-10-20 Mark Mitchell <mark@codesourcery.com>
1223 * config/m68k/m68k.c (m68k_output_mi_thunk): Fix typo.
1225 2002-10-20 Zdenek Dvorak <rakdver@atrey.karlin.mff.cuni.cz>
1228 * i386.c (ix86_init_mmx_sse_builtins, ix86_expand_builtin): Define and
1229 expand __builtin_ia32_pslldqi128 and __builtin_ia32_psrldqi128.
1230 * i386.h (IX86_BUILTIN_PSLLDQI128, IX86_BUILTIN_PSRLDQI128): New.
1231 * xmmintrin.h (_mm_srli_si128, _mm_slli_si128): New.
1233 2002-10-20 Roger Sayle <roger@eyesopen.com>
1236 * toplev.c (flag_unsafe_profile_arcs): Remove.
1237 (flag_bounded_pointers): Remove.
1238 (flag_bounds_check): Correct comments.
1239 (lang_independent_options): Remove -funsafe-profile-arcs and
1240 -fbounded-pointers. Correct -fbounds-check comments.
1242 * flags.h: Correct flag_schedule_interblock comments.
1243 (flag_bounded_pointers): Remove prototype.
1244 (flag_bounds_check): Correct comments.
1246 * c-opts.c (c_common_init_options): No need to mark
1247 flag_bounds_check as unspecified.
1248 (c_common_post_options): And no need to set it from
1249 flag_bounded_pointers if its still unspecified.
1251 * doc/invoke.texi: Fix some overfull hboxes in "make dvi".
1252 Document --version, -feliminate-dwarf-2-dups, -fno-sched-interblock,
1253 -fno-sched-spec, -fsched-spec-load, -fsched-spec-load-dangerous,
1254 -fsched-verbose=n, -fno-branch-count-reg and -fbounds-check.
1256 2002-10-19 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
1257 Mark Mitchell <mark@codesourcery.com>
1259 * alpha-protos.h (alpha_output_mi_thunk_osf): Update signature to
1261 * arm-protos.h, arm.c (arm_output_mi_thunk): Likewise.
1262 * cris-protos.h, cris.c (cris_asm_output_mi_thunk): Likewise.
1263 * frv-protos.h, frv.c (frv_asm_output_mi_thunk): Likewise.
1264 * i386-protos.h, i386.c (x86_output_mi_vcall_thunk,
1265 x86_output_mi_thunk): Likewise.
1266 * i960-protos.h, i960.c (i960_output_mi_thunk): Likewise.
1267 * ia64-protos.h, ia64.c (ia64_output_mi_thunk): Likewise.
1268 * m68k-protos.h, m68k.c (m68k_output_mi_thunk): Likewise.
1269 * mmix-protos.h, mmix.c (mmix_asm_output_mi_thunk): Likewise.
1270 * rs6000-protos.h, rs6000.c (output_mi_thunk): Likewise.
1271 * s390-protos.h, s390.c (s390_output_mi_thunk): Likewise.
1272 * stormy16-protos.h, stormy16.c (xstormy16_asm_output_mi_thunk):
1274 * vax-protos.h, vax.c (vax_output_mi_thunk): Likewise.
1276 * target.h (gcc_target): Update output_mi_thunk and
1277 output_mi_vcall_thunk to take a HOST_WIDE_INT delta and
1280 * config/alpha/alpha.c: Replace ASM_OUTPUT_MI_THUNK with
1281 TARGET_ASM_OUTPUT_MI_THUNK in comments.
1282 * config/alpha/vms.h (ASM_OUTPUT_MI_THUNK): Don't #undef it.
1283 (TARGET_ASM_OUTPUT_MI_THUNK): #undef it.
1284 * config/frv/frv.h (DEFAULT_VTABLE_THUNKS): Remove definition.
1285 * config/i386/i386-protos.h (x86_output_mi_vcall_thunk): Update
1287 * config/i386/i386.c (x86_output_mi_vcall_thunk): Likewise.
1288 * config/i386/openbsd.h: Replace ASM_OUTPUT_MI_THUNK with
1289 TARGET_ASM_OUTPUT_MI_THUNK in comments.
1290 * config/i960/i960.h (ASM_OUTPUT_MI_THUNK): Don't define.
1291 (TARGET_ASM_OUTPUT_MI_THUNK): Do define.
1292 * config/m68k/openbsd.h: Replace ASM_OUTPUT_MI_THUNK with
1293 TARGET_ASM_OUTPUT_MI_THUNK in comments.
1294 * config/rs6000/rs6000.c (rs6000_ra_ever_killed): Remove #ifdef
1295 ASM_OUTPUT_MI_THUNK and replace with check of targetm.
1297 * doc/tm.texi (TARGET_ASM_OUTPUT_MI_THUNK): Update signature.
1298 (TARGET_ASM_OUTPU_MI_VCALL_THUNK): Likewise.
1300 2002-10-19 Brad Lucier <lucier@math.purdue.edu>
1302 * real.c (do_add): Fix 0+0 sign corner case.
1303 (do_divide): Fix Inf/0 corner case.
1305 Sun Oct 20 00:31:31 CEST 2002 Jan Hubicka <jh@suse.cz>
1307 * i386.c (classify_argument): Pass MMX arguments in memory
1308 (ix86_expand_builtin): Expand proper address mode for cflush.
1309 * i386.md (movdqa): Fix typo.
1310 (sse2_cflush): Accept DImode addresses.
1312 * xmmintrin.h (_mm_sqrt_sd): Accept two arguments.
1313 (_mm_max_sd): Fix pasto.
1314 (_mm_storeh_pd, _mm_storel_pd): Fix.
1316 * i386.c (bdesc_comi): Fix to match specification.
1317 (ix86_expand_sse_comi): Emit the comparison properly.
1318 * i386.md (sse_comi, sse2_comi, sse_ucomi, sse2_ucomi):
1319 Do not use comparison operator.
1320 (vnmaskcmp): Fix template.
1322 * xmmintrin.h (_mm_cvtps_pi16): Fix.
1324 2002-10-19 Sebastian Pop <s.pop@laposte.net>
1326 * dependence.c : Removed.
1327 * Makefile.in : Remove dependence.o.
1329 Sat Oct 19 10:46:52 CEST 2002 Jan Hubicka <jh@suse.cz>
1331 * mmintrin.h (__m64): typedef it to v2si.
1332 (_mm_cvtsi32_si64, _mm_cvtsi32_si64_mm_sll_pi16,
1333 _mm_sll_pi32, _mm_sll_pi64, _mm_slli_pi64, _mm_sra_pi16,
1334 _mm_sra_pi32, _mm_srl_pi16, _mm_srl_pi32, _mm_srl_pi64,
1335 _mm_srli_pi64, _mm_and_si64, _mm_andnot_si64,
1336 _mm_or_si64, _mm_xor_si64): Add neccesary casts.
1337 * xmmintrin.h (_mm_setzero_si64): Likewise.
1339 * i386.h (ALIGN_MODE_128): Update comment; add missing modes
1340 (SSE_REG_MODE_P, MMX_REG_MODE_P): New macros.
1343 Patch by Shawn Wagner
1344 * mmintrin.h: Replace pi64 by si64.
1346 2002-10-18 David Edelsohn <edelsohn@gnu.org>
1348 * rs6000.md (movdf_hardfloat32): Order alternatives consistently.
1349 Use length of 4 not *.
1350 (movdf_hardfloat64): Same. Support DFmode moves to/from CTR/LR.
1351 (movdf_softfloat64): Likewise.
1352 (movdi_internal32): Use length of 4 not *.
1353 (movti_power): Same.
1354 (ctrsi, ctrdi): Same.
1356 2002-10-18 Zack Weinberg <zack@codesourcery.com>
1358 * c-decl.c (start_decl): Point users of the old initialized-
1359 typedef extension at __typeof__.
1361 2002-10-18 Richard Henderson <rth@redhat.com>
1363 * real.c (cmp_significand_0, rtd_divmod, ten_to_mptwo): New.
1364 (real_to_decimal): Re-implement using the logic from the
1365 gcc 3.2 etoasc. Comment heavily.
1366 (div_significands): Simplify loop startup and comparison logic.
1368 2002-10-18 Mark Mitchell <mark@codesourcery.com>
1370 * target-def.h (TARGET_ASM_OUTPUT_MI_THUNK): Default to NULL.
1371 (TARGET_ASM_OUTPUT_MI_VCALL_THUNK): Likewise.
1372 (TARGET_ASM_OUT): Add them.
1373 * target.h (asm_out): Add output_mi_thunk and
1374 output_mi_vcall_thunk.
1375 * config/alpha/alpha.h (ASM_OUTPUT_MI_THUNK): Rename to ...
1376 (TARGET_ASM_OUTPUT_MI_THUNK): ... this.
1377 * config/arm/arm-protos.h (arm_output_mi_thunk): Declare.
1378 * config/arm/arm.c (arm_output_mi_thunk): Define.
1379 * config/arm/arm.h (ASM_OUTPUT_MI_THUNK): Rename to ...
1380 (TARGET_ASM_OUTPUT_MI_THUNK): ... this.
1381 * config/cris/cris.h (ASM_OUTPUT_MI_THUNK): Rename to ...
1382 (TARGET_ASM_OUTPUT_MI_THUNK): ... this.
1383 * config/frv/frv.h (ASM_OUTPUT_MI_THUNK): Rename to ...
1384 (TARGET_ASM_OUTPUT_MI_THUNK): ... this.
1385 * config/i386/i386-protos.h (x86_output_mi_thunk): Adjust
1387 (x86_output_mi_vcall_thunk): Declare.
1388 * config/i386/i386.c (override_options): Clear
1389 output_mi_vcall_thunk in 64-bit mode.
1390 (ix86_fntype_regparm): New function.
1391 (ix86_return_pops_args): Use it.
1392 (ia32_this_parameter): New function.
1393 (x86_output_mi_vcall_thunk): New function.
1394 (x86_output_mi_thunk): Use it
1395 * config/i386/unix.h (TARGET_ASM_OUTPUT_MI_THUNK): Adjust.
1396 (TARGET_ASM_OUTPUT_MI_VCALL_THUNK): Define.
1397 * config/i960/i960-protos.h (i960_output_mi_thunk): Declare.
1398 * config/i960/i960.c (i960_output_mi_thunk): New function.
1399 * config/i960/i960.h (ASM_OUTPUT_MI_THUNK): Adjust.
1400 * config/ia64/ia64-protos.h (ia64_output_mi_thunk): Declare.
1401 * config/ia64/ia64.c (ia64_output_mi_thunk): Define.
1402 * config/ia64/ia64.h (ASM_OUTPUT_MI_THUNK): Rename to ...
1403 (TARGET_ASM_OUTPUT_MI_THUNK): ... this.
1404 * config/m68k/m68k-protos.h (m68k_output_mi_thunk): New function.
1405 * config/m68k/linux.h (ASM_OUTPUT_MI_THUNK): Rename to ...
1406 (TARGET_ASM_OUTPUT_MI_THUNK): ... this.
1407 * config/m68k/netbsd-elf.h (ASM_OUTPUT_MI_THUNK): Rename to ...
1408 (TARGET_ASM_OUTPUT_MI_THUNK): ... this.
1409 * config/mmix/mmix.h (ASM_OUTPUT_MI_THUNK): Rename to ...
1410 (TARGET_ASM_OUTPUT_MI_THUNK): ... this.
1411 * config/pa/pa.h (ASM_OUTPUT_MI_THUNK): Rename to ...
1412 (TARGET_ASM_OUTPUT_MI_THUNK): ... this.
1413 * config/rs6000/sysv4.h (ASM_OUTPUT_MI_THUNK): Rename to ...
1414 (TARGET_ASM_OUTPUT_MI_THUNK): ... this.
1415 * config/s390/s390-protos.h (s390_output_mi_thunk): Declare.
1416 * config/s390/s390.c (s390_output_mi_thunk): Define.
1417 * config/s390/s390.h (ASM_OUTPUT_MI_THUNK): Rename to ...
1418 (TARGET_ASM_OUTPUT_MI_THUNK): ... this.
1419 * config/sparc/sparc.h (ASM_OUTPUT_MI_THUNK): Rename to ...
1420 (TARGET_ASM_OUTPUT_MI_THUNK): ... this.
1421 * config/stormy16/stormy16.h (ASM_OUTPUT_MI_THUNK): Rename to ...
1422 (TARGET_ASM_OUTPUT_MI_THUNK): ... this.
1423 * config/vax/vax-protos.h (vax_output_mi_thunk): Declare.
1424 * config/vax/vax.c (vax_output_mi_thunk): Define.
1425 * config/vax/vax.h (ASM_OUTPUT_MI_THUNK): Rename to ...
1426 (TARGET_ASM_OUTPUT_MI_THUNK): ... this.
1427 * doc/tm.texi: Adjust documentation.
1429 2002-10-18 Jason Thorpe <thorpej@wasabisystems.com>
1431 * config/netbsd.h (NETBSD_ENABLE_EXECUTE_STACK): Define
1432 __enable_execute_stack function.
1433 * config/alpha/netbsd.h (TRANSFER_FROM_TRAMPOLINE): Define
1434 as NETBSD_ENABLE_EXECUTE_STACK.
1435 * config/i386/netbsd-elf.h (TRANSFER_FROM_TRAMPOLINE): Ditto.
1436 * config/i386/netbsd.h (TRANSFER_FROM_TRAMPOLINE): Ditto.
1437 * config/i386/netbsd64.h (TRANSFER_FROM_TRAMPOLINE): Ditto.
1438 * config/sparc/netbsd-elf.h (TRANSFER_FROM_TRAMPOLINE): Ditto.
1439 * config/sparc/netbsd.h (TRANSFER_FROM_TRAMPOLINE): Ditto.
1441 2002-10-18 Jason Thorpe <thorpej@wasabisystems.com>
1443 * config/i386/i386.c (x86_initialize_trampoline): Emit a call
1444 to __enable_execute_stack with the address of the trampoline
1445 if TRANSFER_FROM_TRAMPOLINE is defined.
1446 * config/i386/i386.h (TARGET_64BIT): Expand to a compile-time
1447 constant if building libgcc2.
1449 2002-10-17 Roger Sayle <roger@eyesopen.com>
1451 * doc/c-tree.texi: Update description of COND_EXPR tree nodes.
1453 2002-10-17 Geoffrey Keating <geoffk@apple.com>
1455 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Allow arbitrary modes
1457 * config/rs6000/rs6000.md (movcc_internal1): Support CCmode moves
1459 (movsf_hardfloat): Support SFmode moves to/from CTR/LR/MQ.
1460 (movsf_softfloat): Likewise.
1462 2002-10-17 Janis Johnson <janis187@us.ibm.com>
1464 * Makefile.in (site.exp): Add ALT_CXX_UNDER_TEST and COMPAT_OPTIONS.
1466 2002-10-17 Jason Thorpe <thorpej@wasabisystems.com>
1468 * config/alpha/alpha.c (alpha_initialize_trampoline): Use
1469 tramp, not addr, to pass the trampoline address to
1470 __enable_execute_stack.
1472 Thu Oct 17 18:40:47 CEST 2002 Jan Hubicka <jh@suse.cz>
1474 * mmintrin.h: Guard by __MMX__
1475 * xmmintrin.h: Guard by __SSE__
1478 * xmmintrin.h (_MM_SHUFFLE2): New macro.
1479 (_mm_load*_?d): New functions.
1480 (_mm_set*_?d): New functions.
1481 (_mm_store*_?d): New functions.
1483 Wed Oct 16 15:01:29 CEST 2002 Jan Hubicka <jh@suse.cz>
1485 Really commit patch announced at Oct 14
1487 * predict.c (can_predict_insn_p): New function.
1488 (estimate_probability): Avoid unnecesary work.
1489 (process_note_prediction): Likewise.
1490 * toplev.c (rest_of_compilation): Account early branch prediction pass
1494 Found by Ian Ollmann
1495 * xmmintrin.h (_mm_shuffle_pd): Fix typo.
1496 (_mm_load?_pd): Likewise.
1497 (_mm_store?_pd): Likewise.
1500 * i386.c (builtin_description):Drop cmpg[te]s[sd].
1501 * xmmintrin.h (__mm_cmpg[te]_s[sd]): Rewrite using
1502 swapped alternative.
1505 * reload1.c (reload_inner_reg_of_subreg): New argument output;
1506 (push_reload): Update call.
1508 2002-10-17 Richard Sandiford <rsandifo@redhat.com>
1510 * config.gcc (mips*-*-*): Add OBJECT_FORMAT_ELF to $tm_defines
1511 if using mips/elf.h or mips/elf64.h.
1512 * config/mips/elf.h (OBJECT_FORMAT_ELF): Remove.
1513 * config/mips/elf64.h (OBJECT_FORMAT_ELF): Remove.
1515 2002-10-16 Aldy Hernandez <aldyh@redhat.com>
1517 * config/rs6000/rs6000.c (function_arg): Set inner mode of V1DI to
1520 2002-10-16 Ulrich Weigand <uweigand@de.ibm.com>
1522 * config/s390/linux.h (ASM_DOUBLE, _ASM_OUTPUT_LONG): Remove.
1523 (LPREFIX): Likewise.
1524 (ASM_COMMENT_START, LOCAL_LABEL_PREFIX, ASM_FORMAT_PRIVATE_NAME,
1525 ASM_OUTPUT_ADDR_VEC_ELT, ASM_OUTPUT_ADDR_DIFF_ELT,
1526 ASM_OUTPUT_ALIGN, ASM_OUTPUT_SKIP, ASM_OUTPUT_ALIGNED_BSS,
1527 TEXT_SECTION_ASM_OP, DATA_SECTION_ASM_OP, BSS_SECTION_ASM_OP,
1528 GLOBAL_ASM_OP, ASM_OUTPUT_MI_THUNK): Move to s390.h.
1530 * config/s390/s390.h (ASM_COMMENT_START, LOCAL_LABEL_PREFIX,
1531 ASM_FORMAT_PRIVATE_NAME, ASM_OUTPUT_ALIGN, ASM_OUTPUT_SKIP,
1532 ASM_OUTPUT_ALIGNED_BSS, TEXT_SECTION_ASM_OP, DATA_SECTION_ASM_OP,
1533 BSS_SECTION_ASM_OP): Move from linux.h.
1534 (ASM_OUTPUT_ADDR_VEC_ELT, ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
1535 Also, use ASM_GENERATE_INTERNAL_LABEL instead of LPREFIX.
1537 * config/s390/s390.c (s390_function_profiler): Use
1538 ASM_GENERATE_INTERNAL_LABEL instead of LPREFIX.
1540 2002-10-15 Richard Henderson <rth@redhat.com>
1542 * real.c (real_to_decimal): Accept BUF_SIZE and CROP_TRAILING_ZEROS
1543 as arguments. Bound DIGITS by the available buffer size.
1544 (real_to_hexadecimal): Likewise.
1545 * real.h (real_to_decimal, real_to_hexadecimal): Update prototypes.
1546 (REAL_VALUE_TO_DECIMAL): Remove.
1547 * c-common.c, c-pretty-print.c, print-rtl.c, print-tree.c,
1548 sched-vis.c, config/arc/arc.c, config/c4x/c4x.c, config/fr30/fr30.c,
1549 config/i370/i370.h, config/i386/i386.c, config/i960/i960.c,
1550 config/ip2k/ip2k.c, config/m32r/m32r.c, config/m68hc11/m68hc11.c,
1551 config/m68k/hp320.h, config/m68k/m68k.h, config/m68k/sun2o4.h,
1552 config/m68k/sun3.h, config/mips/mips.c, config/ns32k/ns32k.c,
1553 config/pdp11/pdp11.h, config/vax/vax.h: Update all callers to
1554 use real_to_decimal directly, and with the proper arguments.
1555 * doc/tm.texi (REAL_VALUE_TO_DECIMAL): Remove.
1557 2002-10-15 Jim Wilson <wilson@redhat.com>
1559 * reload1.c (merge_assigned_reloads): After converting overlapping
1560 reloads to RELOAD_OTHER, abort if there are now conflicting reloads.
1562 * config/i386/i386.md (adddi3_1): Add call to ix86_binary_operator_ok.
1564 Tue Oct 15 22:08:35 CEST 2002 Jan Hubicka <jh@suse.cz>
1566 * expr.c (do_tablejump): Fix typo in my previous commit.
1568 2002-10-15 Richard Sandiford <rsandifo@redhat.com>
1570 * config/mips/vr.h (DRIVER_SELF_SPECS): Change %<mgp32 to %{<mgp32}.
1572 2002-10-15 Ulrich Weigand <uweigand@de.ibm.com>
1574 * config/s390/s390.c (s390_split_branches): Add return
1575 value. Add parameters TEMP_REG and TEMP_USED. Use unspec 104.
1577 (find_base_register_in_addr): New function.
1578 (find_base_register_ref): New function.
1579 (replace_base_register_ref): New function.
1581 (struct constant_pool): Add members pool_insn, insns, and anchor.
1582 Remove member last_insn.
1583 (s390_start_pool): Initialize them.
1584 (s390_end_pool): Emit pool placeholder insn.
1585 (s390_add_pool_insn): New function.
1586 (s390_find_pool): Use insns bitmap instead of addresses.
1587 (s390_dump_pool): Replace placeholder insn. Emit anchor.
1588 Replace unspec 104 by local-pool-relative references.
1589 (s390_output_constant_pool): Output anchor label if required.
1590 (s390_output_symbolic_const): Handle unspec 104 and 105.
1591 (s390_add_pool): Remove, replace by ...
1592 (s390_add_constant, s390_find_constant): ... these new functions.
1593 (s390_add_anchor): New function.
1595 (s390_chunkify_pool): Delete, replace by ...
1596 (s390_chunkify_start, s390_chunkify_finish,
1597 s390_chunkify_cancel): ... these new functions.
1598 (s390_optimize_prolog): Add parameter TEMP_REGNO.
1599 Recompute register live data for special registers.
1600 (s390_fixup_clobbered_return_reg): New function.
1601 (s390_machine_dependent_reorg): Rewrite to use new
1602 s390_chunkify_... routines.
1604 config/s390/s390.md ("reload_base"): Rename to ...
1605 ("reload_base_31"): ... this.
1606 ("reload_base_64"): New insn.
1607 ("reload_base2"): Remove.
1608 ("reload_anchor"): New insn.
1611 s390.c (s390_pool_overflow): Remove.
1612 s390.h (s390_pool_overflow): Likewise.
1613 s390.md ("cjump", "icjump", "doloop_si"): Remove s390_pool_overflow.
1615 Tue Oct 15 16:51:04 2002 J"orn Rennecke <joern.rennecke@superh.com>
1617 * sh.md (movv8qi_i+2): Don't split if source is -1.
1619 2002-10-15 Janis Johnson <janis187@us.ibm.com>
1621 * doc/install.texi: Formatting changes for conformance to HTML 4.01.
1623 2002-10-15 Ulrich Weigand <uweigand@de.ibm.com>
1626 * loop.c (loop_regs_scan): Mark registers used for function
1627 argument passing as MAY_NOT_OPTIMIZE.
1629 Mon Oct 14 20:33:12 CEST 2002 Jan Hubicka <jh@suse.cz>
1631 * i386.md (movv2di_internal): New pattern.
1632 (movv2df_internal, movv8hi_internal, movv16qi_internal): Fix predicate.
1633 (movv2di): New expander.
1634 * i386.c (ix86_preferred_reload_class): Return NO_REGS for vector operands.
1636 * i386.c (ix86_expand_timode_binop_builtin): Delete.
1637 (builtin_description): Add SSE1 logicals; rename SSE2 logicals.
1638 (ix86_init_mmx_sse_builtins): Kill SSE1 logicals.
1639 (ix86_expand_builtin): Likewise.
1640 * i386.h (sse_andti4_df_1, sse_andti3_df_2, sse_andti3_sf_1, sse_andti3_sf_2,
1642 sse_andnti4_df_1, sse_andti3_df_2, sse_andti3_sf_1, sse_andti3_sf_2,
1644 sse_orti4_df_1, sse_orti3_df_2, sse_orti3_sf_1, sse_orti3_sf_2,
1646 sse_xorti4_df_1, sse_xorti3_df_2, sse_xorti3_sf_1, sse_xorti3_sf_2,
1648 (sse_andv4sf3, sse_andnv4sf3, sse_orv2df3, sse_xorv2df3, sse_andv2df3,
1649 sse_andnv2df3, sse_orv2df3, sse_xorv2df3): New expanders.
1650 (*sse_andv4sf3, *sse_andnv2df3, *sse_orv4sf3, *sse_xorv4sf3, *sse_andv2df3,
1651 *sse_andnv2df3, *sse_orv2df3, *sse_xorv2df3): New patterns.
1652 (*sse_andsf3, *sse_andndf3, *sse_ordf3, *sse_xordf3, *sse_anddf3,
1653 *sse_andndf3, *sse_orv2df3, *sse_xorv2df3): New patterns.
1655 * xmmintrin.h (__m128i): Define as __v2di.
1658 (expand_expr): Use DECL_RTL_SET_P.
1660 2002-10-14 Roger Sayle <roger@eyesopen.com>
1662 * combine.c (simplify_set): Treat MODE_CC registers like cc0.
1664 2002-10-14 Roger Sayle <roger@eyesopen.com>
1665 Zack Weinberg <zack@codesourcery.com>
1667 * config/i386/i386.c (k6_cost): Correct typo.
1669 2002-10-14 Mark Mitchell <mark@codesourcery.com>
1671 PR optimization/6631
1672 * alias.c (objects_must_conflict_p): Check honor_readonly when
1673 examining TYPE_READONLY.
1674 * function.c (assign_stack_temp_for_type): Likewise.
1676 2002-10-14 Falk Hueffner <falk.hueffner@student.uni-tuebingen.de>
1678 * config/alpha/alpha.md (extendsidi2_nofix, extendsidi2_fix):
1679 Swap zero extension arguments.
1680 (umaxhi3): Fix instruction class.
1682 (prefetch): Fix prefetch instructions.
1684 (pkwb): Fix output constraint.
1686 2002-10-14 Alexandre Oliva <aoliva@redhat.com>
1688 * config/mips/mips.c (print_operand): Increase buffer size for
1691 2002-10-14 Richard Henderson <rth@redhat.com>
1694 * gcse.c (adjust_libcall_notes): Revert last change.
1695 * simplify-rtx.c (simplify_replace_rtx): Handle LO_SUM.
1697 2002-10-14 Andrew Haley <aph@redhat.com>
1699 * tree-inline.c (remap_block): All local class initialization
1700 flags go in the outermost scope.
1701 (expand_call_inline): Call java_inlining_map_static_initializers.
1702 (expand_call_inline): Call java_inlining_merge_static_initializers.
1703 * java/lang.c (merge_init_test_initialization): New.
1704 (java_inlining_merge_static_initializers): New.
1705 (inline_init_test_initialization): New.
1706 (java_inlining_map_static_initializers): New.
1708 * tree-inline.c (expand_call_inline): Convert retvar to expected
1711 2002-10-14 Graham Stott <graham.stott@btinternet.com>
1713 * stmt.c (decl_conflicts_with_clobbers_p): Add REG_P check.
1715 2002-10-14 Aldy Hernandez <aldyh@redhat.com>
1717 * stmt.c: Fix typo in comment.
1719 Mon Oct 14 11:35:49 2002 J"orn Rennecke <joern.rennecke@superh.com>
1721 * c-common.c (c_common_type_for_mode): Add V2HImode case.
1722 * tree.c (build_common_tree_nodes_2): Initialize
1723 unsigned_V2HI_type_node and V2HI_type_node.
1724 * tree.h (enum tree_index): Add TI_UV2HI_TYPE and TI_V2HI_TYPE.
1725 (unsigned_V2HI_type_node, V2HI_type_node): Define.
1727 2002-10-14 Jakub Jelinek <jakub@redhat.com>
1729 * config/i386/i386.h (ASM_OUTPUT_REG_PUSH, ASM_OUTPUT_REG_POP):
1730 Handle TARGET_64BIT.
1732 2002-10-14 Richard Sandiford <rsandifo@redhat.com>
1734 * config/mips/vr.h (DRIVER_SELF_SPECS): Define.
1735 * config/mips/t-vr (MULTILIB_OPTIONS): Remove mlong32.
1736 (MULTILIB_DIRNAMES): Remove long32.
1737 (MULTILIB_EXCEPTIONS): Don't build -mabi=32 -mgp32 multilibs.
1738 (MULTILIB_REDUNDANT_DIRS): Remove.
1740 2002-10-14 Richard Sandiford <rsandifo@redhat.com>
1742 * doc/tm.texi (DRIVER_SELF_SPECS): Document.
1743 * gcc.c (driver_self_specs): New variable.
1744 (do_self_spec): New function.
1745 (main): Use it to process driver_self_specs.
1747 2002-10-13 Richard Henderson <rth@redhat.com>
1749 * rtl.c (shallow_copy_rtx): Use memcpy for the entire node.
1751 2002-10-12 Neil Booth <neil@daikokuya.co.uk>
1753 PR preprocessor/7862
1754 PR preprocessor/8190
1755 * gcc.c (cpp_unique_options): Don't delete .d files.
1756 Remove stray whitespace.
1758 2002-10-12 Naohiko Shimizu <pshimizu@fa2.so-net.ne.jp>
1760 * pdp11.h (ASM_OUTPUT_SKIP): Add preceding 0 for octal constant.
1761 (ASM_OUTPUT_COMMON, ASM_OUTPUT_LOCAL): Likewise.
1762 * pdp11.c (pdp11_output_function_prologue): 0%o -> %#o.
1763 (pdp11_output_function_epilogue, output_ascii): Likewise.
1764 (output_addr_const_pdp11): Likewise.
1765 * pdp11.md (movdi): Use offsetable memory for floating store.
1766 (lshrsi3, negsi2): Delete irrelevant comment.
1768 2002-10-11 Geoffrey Keating <geoffk@apple.com>
1770 * cse.c (mention_regs): Set SUBREG_TICKED to the register number,
1771 not the address of the REG.
1772 (struct cse_reg_info): Make subreg_ticked unsigned.
1774 2002-10-11 Janis Johnson <janis187@us.ibm.com>
1776 * doc/compat.texi: Add info about C++ libraries.
1778 2002-10-11 Richard Henderson <rth@redhat.com>
1781 * gcse.c (adjust_libcall_notes): Also adjust notes for INSN.
1783 2002-10-11 John David Anglin <dave@hiauly1.hia.nrc.ca>
1785 * cfganal.c (dfs_enumerate_from): Use PARAMS.
1786 * genautomata.c (output_insn_code_cases): Likewise.
1787 * real.c (real_format): Likewise.
1788 * tree.c (tree_size): Revise expressions using TREE_CODE_LENGTH to
1789 ensure value is promoted before doing subtraction.
1791 Fri Oct 11 22:22:38 CEST 2002 Jan Hubicka <jh@suse.cz>
1793 * calls.c (expand_call): Simplify noreturn call.
1796 * cfgbuild.c (make_edges): Create edge cache when we do have
1798 * expr.c (do_tablejump): Note size of maximal jumptable.
1799 * function.c (prepare_function_start): Zero out size.
1800 * function.h (function): Add max_jumptable_ents.
1802 * cfgcleanup.c (insn_match_p): Verify sibcall flag for calls to.
1804 Fri Oct 11 12:34:33 2002 J"orn Rennecke <joern.rennecke@superh.com>
1806 * sh.md (movv8qi_i+2): For V8QI destinations, generate V4HI
1807 register for mperm_w operation.
1809 Fri Oct 11 10:56:17 2002 J"orn Rennecke <joern.rennecke@superh.com>
1811 * emit-rtl.c (gen_lowpart_common): When asked to make a vector from
1812 an integer, use simplify_gen_subreg.
1814 2002-10-10 Aldy Hernandez <aldyh@redhat.com>
1816 * extend.texi (Vector Extensions): Remove comment about single
1819 2002-10-10 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
1821 * fold-const.c (size_htab_hash): Use htab_hash_pointer.
1822 * function.c (insns_for_mem_hash): Likewise.
1823 * varasm.c (STRHASH): Likewise.
1825 2002-10-10 Stuart Hastings <stuart@apple.com>
1827 * cse.c (struct cse_reg_info): Add subreg_ticked.
1828 (SUBREG_TICKED): New.
1829 (get_cse_reg_info): Initialize SUBREG_TICKED.
1830 (mention_regs): Use it.
1831 (invalidate): Set SUBREG_TICKED.
1832 (invalidate_for_call): Likewise.
1833 (addr_affects_sp_p): Likewise.
1835 2002-10-10 Jakub Jelinek <jakub@redhat.com>
1837 * config/i386/i386.md (tls_local_dynamic_base): Put pic reg
1838 into proper operand.
1840 2002-10-10 Denis Chertykov <denisc@overta.ru>
1842 * config/ip2k/ip2k.c (function_epilogue): Optimize stack
1844 * config/ip2k/libgcc.S: Combine routines used by function
1847 2002-10-10 Jim Wilson <wilson@redhat.com>
1849 * cse.c (fold_rtx): Don't perform associative optimization for DIV and
1852 2002-10-10 David Edelsohn <edelsohn@gnu.org>
1854 * config/rs6000/aix52.h: New file.
1855 * config/rs6000/t-aix52: New File.
1856 * config.gcc (rs6000-ibm-aix5.1.*): New entry.
1857 (rs6000-ibm-aix[56789].*): Default to AIX 5.2.
1859 Thu Oct 10 19:37:54 CEST 2002 Jan Hubicka <jh@suse.cz>
1862 * invoke.texi (-msse-math): Kill
1863 (-msse): Add note to mfpmath=sse.
1865 Thu Oct 10 17:08:30 CEST 2002 Jan Hubicka <jh@suse.cz>
1868 * i386.c (ix86_expand_vector_move): Do not generate const0->mem moves.
1870 2002-10-10 Neil Booth <neil@daikokuya.co.uk>
1872 PR preprocessor/8179
1873 * gcc.c (cpp_options): Add {ansi}, move %{m*} to same location
1875 (default_compilers): Pass debug options when preprocessing
1878 2002-10-06 Richard Henderson <rth@redhat.com>
1880 * toplev.c (rest_of_compilation): Revert opt/2960 change.
1882 2002-10-09 Zack Weinberg <zack@codesourcery.com>
1885 * c-decl.c (start_decl): Unconditionally issue error for
1886 'typedef foo = bar'.
1887 (finish_decl): Remove special case for TYPE_DECL with initializer.
1889 * doc/extend.texi: Delete "Naming Types" section. Change all
1890 cross-references to that section to refer to "Typeof" instead.
1891 Add the useful safe-max()-macro example from "Naming Types" to
1892 "Typeof", rewritten using that extension. Add some compatibility
1895 2002-10-09 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
1897 * loop.c: Revert 2002-08-15 change.
1898 (LOOP_REGNO_NREGS): Ensure type is int.
1900 2002-10-09 David Edelsohn <edelsohn@gnu.org>
1902 * config/rs6000/rs6000.md (extenddftf2): Change to define_insn
1903 which copies first FPR and clears second.
1904 (extendsftf2): Same.
1905 (floatditf2): Fix typo.
1907 (fix_trunctfdi2): Same.
1908 (fix_trunctfsi2): Same.
1910 2002-10-09 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
1912 * conflict.c (arc_hash): Change return type to hashval_t.
1913 * cselib.c (get_value_hash): Likewise.
1914 * genautomata.c (automaton_decl_hash, insn_decl_hash, decl_hash,
1915 state_hash, automata_list_hash): Likewise.
1916 * read-rtl.c (def_hash): Likewise.
1917 * tree.c (type_hash_hash): Likewise.
1919 2002-10-08 Aldy Hernandez <aldyh@redhat.com>
1921 * config/rs6000/rs6000.c (rs6000_ra_ever_killed): Call
1922 prologue_epilogue_contains instead of using REG_MAYBE_DEAD notes.
1924 Wed Oct 9 15:54:49 2002 J"orn Rennecke <joern.rennecke@superh.com>
1926 * sh.md (ffssi2): Fix emitted code.
1928 2002-10-09 Ulrich Weigand <uweigand@de.ibm.com>
1930 * cse.c (insn_live_p): Pass insn pattern, not full insn
1933 2002-10-09 Neil Booth <neil@daikokuya.co.uk>
1935 * cppmacro.c (paste_tokens): Only allow / to paste with =.
1937 2002-10-09 David Edelsohn <edelsohn@gnu.org>
1939 * config/rs6000/rs6000.md (movdf splitter): Use gen_int_mode on
1941 (movtf_internal): Reference correct displacement for second value
1943 (movtf splitter): Correct generation of constants in 64-bit mode.
1945 2002-10-09 Alan Modra <amodra@bigpond.net.au>
1947 * libgcc2.c (__floatdisf): Properly cure double rounding.
1949 2002-10-09 Gabriel Dos Reis <gdr@integrable-solutions.net>
1951 * c-common.c (cb_register_builtins): Define __WCHAR_MAX__.
1952 * doc/cpp.texi (Common Predefined Macros): Document.
1954 2002-10-09 Gabriel Dos Reis <gdr@integrable-solutions.net>
1957 * doc/invoke.texi (Option Summary): List
1958 -Wmissing-declarations as a C only option.
1960 2002-10-08 Jakub Jelinek <jakub@redhat.com>
1962 * config/sparc/t-linux64 (MULTILIB_OPTIONS): Remove
1963 mno-app-regs|mcmodel=medany.
1964 (MULTILIB_DIRNAMES, MULTILIB_OSDIRNAMES): Remove alt.
1965 (MULTILIB_EXCEPTIONS, MULTILIB_EXCLUSIONS, MULTILIB_MATCHES): Remove.
1966 (CRTSTUFF_T_CFLAGS): Define.
1968 2002-10-08 Roger Sayle <roger@eyesopen.com>
1971 * simplify-rtx.c (avoid_constant_pool_reference): Allow constant
1972 pool references that are constructed using LO_SUM.
1974 2002-10-08 Nathan Sidwell <nathan@codesourcery.com>
1976 * c-opts.c (c_common_decode_option): Add warn_strict_aliasing to
1978 * c-typeck.c (build_c_cast): Use warn_strict_aliasing, tweak
1980 * flags.h (warn_strict_aliasing): Declare.
1981 * toplev.c (warn_strict_aliasing): Define.
1982 (lang_independent_options): Add it.
1983 * doc/invoke.texi (-Wstrict-aliasing): Document it.
1985 2002-10-08 Zack Weinberg <zack@codesourcery.com>
1987 * system.h (GCCBUGURL): Delete.
1988 * version.c (bug_report_url): New. Add commentary about
1989 modifying both these strings in modified distributions.
1990 * version.h: Declare bug_report_url.
1992 * diagnostic.c, gcc.c, gcov.c: Globally replace GCCBUGURL with
1995 2002-10-08 Nick Clifton <nickc@redhat.com>
1997 * config/rs6000/spe.h (__ev_set_acc_u64): Use __ev_create_u64 to
1998 convert uint64_t into __ev64_opaque__.
1999 (__ev_set_acc_s64): Likewise, but using signed types.
2001 2002-10-08 Ulrich Weigand <uweigand@de.ibm.com>
2003 * config/s390/s390.md ("*doloop_si_long"): Add missing operand.
2004 ("*doloop_di_long"): Likewise.
2006 Tue Oct 8 16:50:10 CEST 2002 Jan Hubicka <jh@suse.cz>
2008 * print-rtl.c (print_rtx): Increase buffer size for real numbers.
2010 2002-10-08 Richard Sandiford <rsandifo@redhat.com>
2012 * config/mips/mips.md (define_attr cpu): Add r4111.
2014 2002-10-08 Anthony Green <green@redhat.com>
2016 * bitmap.c (bitmap_equal_p): Clear all bitmap_head fields.
2018 2002-10-08 Michael Hayes <m.hayes@elec.canterbury.ac.nz>
2020 * config/c4x/c4x.c (c4x_print_operand): Enlarge buffer
2021 for REAL_VALUE_TO_DECIMAL output.
2023 2002-10-07 David Edelsohn <edelsohn@gnu.org>
2025 * config/rs6000/rs6000.c (rs6000_override_options): Set
2026 real_format_for_mode for IBM extended format, if enabled.
2027 (easy_fp_constant): Add TFmode.
2028 (rs6000_legitimize_address): Add TFmode.
2029 (rs6000_legitimate_address): Same.
2030 (function_arg_advance): TFmode uses two FPRs.
2031 (rs6000_emit_prologue): Fix warning.
2032 (rs6000_output_function_epilogue): Add TFmode.
2033 (output_toc): Add TFmode.
2034 * rs6000.h (SLOW_UNALIGNED_ACCESS): Add TFmode.
2035 (LEGITIMATE_OFFSET_ADDRESS_P): Add TFmode.
2036 * rs6000.md (movtf splitter): Load TFmode constant.
2038 2002-10-07 Dale Johannesen <dalej@apple.com>
2040 * rtl.h: Add NOTE_PRECONDITIONED.
2042 * loop.c: Set loop_info->preconditioned from it.
2043 * doloop.c: Permit doloop treatment when loop_info->preconditoned.
2045 2002-10-07 Richard Henderson <rth@redhat.com>
2047 * config/i960/i960.c (i960_setup_incoming_varargs): Create a
2048 new rtx for comparing the argument pointer against zero.
2049 (i960_va_start): Similarly.
2051 2002-10-07 Richard Henderson <rth@redhat.com>
2053 * config/i960/i960.md (*): Use TFmode, not XFmode.
2054 * config/i960/i960.c (*): Likewise.
2055 (i960_arg_size_and_align): Remove XFmode alignment hack.
2056 (i960_round_align): Merge code from ROUND_TYPE_ALIGN.
2057 * config/i960/i960.h (LONG_DOUBLE_TYPE_SIZE): Use 128, not 96.
2058 (MAX_LONG_DOUBLE_TYPE_SIZE): Likewise.
2059 (DATA_ALIGNMENT, ROUND_TYPE_SIZE): Remove.
2061 2002-10-07 Richard Henderson <rth@redhat.com>
2063 * config/fp-bit.c (EXTENDED_FLOAT_STUBS): Flush out all XF/TFmode
2064 entry points; use void return value and argument list.
2066 2002-10-05 Naohiko Shimizu <nshimizu@keyaki.cc.u-tokai.ac.jp>
2068 * t-pdp11: Add MULTILIB support for msoft-float.
2069 * pdp11.h (LEGITIMATE_CONSTANT_P): Fix soft-float case.
2071 * t-pdp11: Add LIB2FUNCS_EXTRA.
2072 * pdp11.c (pdp11_output_function_prologue): Restrict offset to 16bit,
2073 add preceding 0 to the octal constant, rename 'fp' to 'r5', rename
2074 'fldd' to 'ldd', rename 'fstd' to 'std'.
2075 (pdp11_output_function_epilogue): Likewise.
2076 (output_move_quad): Make the comment gas compatible.
2077 (output_ascii): Add preceding 0 to the octal constant.
2078 (print_operand_address): Add pre_modify, post_modify.
2079 (output_addr_const_pdp11): Add preceding 0 to the octal constant.
2080 * pdp11.h (GO_IF_LEGITIMATE_ADDRESS) : Add 'movb' pre_modify case
2081 with the indication of Paul Koning.
2082 (PRINT_OPERAND): Fix floating constant.
2083 * pdp11.md (movdi): Restrict matching pattern.
2084 (movqi): Generalize the matching pattern.
2085 (movdf): Restrict matching pattern.
2086 (zero_extendqihi2): Change constant representation.
2087 (floatsidf2): Fix wrong operands.
2088 (addqi3): Fix wrong instruction name.
2089 (subqi3): Fix wrong instruction name.
2090 (andsi3, andhi3, andqi3): Simplify and fix to use 'bic'.
2091 (xorsi3): Fix wrong insn.
2092 (one_cmplqi2): Add two operand pattern.
2095 (call): Add register indirect case.
2096 (mod): Fix wrong subreg.
2098 2002-09-25 Eric Botcazou <ebotcazou@libertysurf.fr>
2099 Volker Reichelt <reichelt@igpm.rwth-aachen.de>
2102 * expr.c (expand_expr) [PLUS]: Simplify after the operands
2103 have been expanded in EXPAND_NORMAL mode.
2105 2002-10-06 Richard Henderson <rth@redhat.com>
2107 * config/rs6000/rs6000.md (load_toc_v4_PIC_2): Fix base constraint.
2109 2002-10-06 Richard Henderson <rth@redhat.com>
2111 PR optimization/2960
2112 * toplev.c (rest_of_compilation): Don't copy_loop_headers if
2115 2002-10-06 Alexandre Oliva <aoliva@redhat.com>
2117 * config/mips/mips.h (SIZE_TYPE, PTRDIFF_TYPE): Override
2118 previously definitions.
2120 2002-10-06 Frank Ch. Eigler <fche@redhat.com>
2122 * cppinit.c (init_standard_includes, parse_option): Use strncmp.
2123 * c-opts.c (find_opt): Similarly.
2125 2002-10-05 Jakub Jelinek <jakub@redhat.com>
2127 * gcc.c (set_multilib_dir): Don't access *end.
2128 Use memcpy instead of strncpy. Don't write beyond malloced buffer.
2129 (print_multilib_info): Don't show paths starting with ".:".
2130 * genmultilib: Add new option, "yes" if multilibs are enabled.
2131 Update comments. If multilibs not enabled, print .:${osdirout}
2132 for each directory. If multilibs are enabled, always print
2133 ${dirout}:${osdirout}, even if the two are the same.
2134 * Makefile.in (s-mlib): Pass @enable_multilib@ to genmultilib.
2135 Pass all MULTILIB_* variables to genmultilib even if
2136 --disable-multilib but MULTILIB_OSDIRNAMES is not empty.
2138 2002-10-04 Bruce Korb <bkorb@gnu.org>
2140 * fixinc/inclhack.def(hpux11_abs): use format fix
2141 * fixinc/fixincl.x: regenerate
2142 * fixinc/tests/base/stdlib.h: accommodate new fix test
2144 Sat Oct 5 19:42:45 CEST 2002 Jan Hubicka <jh@suse.cz>
2146 * c-common.c (cb_register_builtins): Use really_no_inline.
2148 2002-10-04 David Edelsohn <edelsohn@gnu.org>
2150 * unroll.c (copy_loop_body): Remove REG_EQUAL note attached to
2151 copied instruction if the note is not loop invariant.
2153 2002-10-04 Loren J. Rittle <ljrittle@acm.org>
2155 * gcc/ginclude/stddef.h: Support the FreeBSD 5 typedef system.
2157 2002-10-04 Steve Ellcey <sje@cup.hp.com>
2159 * doc/invoke.texi (HPPA): Add -mlinker-opt, -mgnu-ld,
2160 and -mhp-ld options to list of options. Add -mgnu-ld
2161 and -mhp-ld option descriptions.
2163 2002-10-04 Steve Ellcey <sje@cup.hp.com>
2165 * fixinc/inclhack.def (hpux11_abs): New.
2166 (stdio_va_list): change __va_list__ to __gnuc_va_list.
2167 * fixinc/fixincl.x: Rebuild.
2169 2002-10-04 Roger Sayle <roger@eyesopen.com>
2171 * config/i386/i386.h (processor_costs): Add new fields fadd,
2172 fmul, fdiv, fabs, fchs and fsqrt to costs structure.
2173 (RTX_COSTS): Use these fields to determine the RTX costs
2174 of floating point addition/subtraction, multiplication,
2175 division, fabs, negation and square root respectively.
2176 * config/i386/i386.c (size_cost): Provide instruction sizes
2177 for these new fields.
2178 (i386_cost, i486_cost, pentium_cost, pentiumpro_cost,
2179 k6_cost, athlon_cost, pentium4_cost): Provide typical cycle
2180 counts for these new fields for all x86 processor variants.
2182 2002-10-04 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
2184 * mips.c (mips_const_double_ok): Delete unused variable.
2186 * gengtype.c (rtx_next): Change type to int.
2188 2002-10-04 Andreas Jaeger <aj@suse.de>
2190 * config/i386/t-linux64 (MULTILIB_OSDIRNAMES): Fix value.
2192 2002-10-04 Richard Henderson <rth@redhat.com>
2194 * real.h (SIGNIFICAND_BITS): Add one more word.
2195 (CONST_DOUBLE_FORMAT): Accomodate 6 words.
2196 * real.c (times_pten): New.
2197 (real_to_decimal, real_from_string): Use it.
2198 (sticky_rshift_significand): Use & to find modulus.
2199 (rshift_significand, lshift_significand): Likewise.
2200 (do_divide): Apply sticky bit after normalization.
2201 (real_to_decimal, real_to_hexadecimal): Fix sign of Inf and NaN.
2203 2002-10-03 Andreas Jaeger <aj@suse.de>
2205 * gengtype.c (adjust_field_rtx_def): Cast variables of type size_t
2206 to unsigned long, adjust printf format string.
2207 (output_mangled_typename): Likewise.
2209 2002-10-03 Jason Thorpe <thorpej@wasabisystems.com>
2211 * config/vax/vax.c (vax_output_function_prologue): Use asm_fprintf.
2212 * config/vax/vax.h (VAX_FUNCTION_PROFILER_NAME): New.
2213 (FUNCTION_PROFILER): Rewrite to use ASM_GENERATE_INTERNAL_LABEL,
2214 assemble_name, asm_fprintf, and VAX_FUNCTION_PROFILER_NAME.
2215 (ASM_OUTPUT_MI_THUNK): Use asm_fprintf instead of REGISTER_PREFIX.
2216 (PRINT_OPERAND_PUNCT_VALID_P): Fix comment.
2217 * config/vax/elf.h (FUNCTION_PROFILER): Remove.
2218 (VAX_FUNCTION_PROFILER_NAME): Redefine as "__mcount".
2220 2002-10-03 Mark Mitchell <mark@codesourcery.com>
2222 * doc/invoke.texi (-Wabi): Document mangling bug.
2224 2002-10-04 Alan Modra <amodra@bigpond.net.au>
2226 * config/rs6000/rs6000.c (rs6000_output_function_epilogue): Use a
2227 name for the tbtab label that depends on the function asm name.
2228 Don't output tbtab label unless optional_tbtab.
2229 (output_mi_thunk): Formatting.
2231 2002-10-03 Richard Henderson <rth@redhat.com>
2233 * config/m68k/m68k.h (OVERRIDE_OPTIONS): Move additional code ...
2234 * config/m68k/m68k.c (override_options): ... here.
2235 * config/m68k/m68kelf.h (OVERRIDE_OPTIONS): Remove.
2236 * config/m68k/m68kv4.h (OVERRIDE_OPTIONS): Remove.
2237 * config/m68k/linux.h (SUBTARGET_OVERRIDE_OPTIONS): Remove.
2238 * config/m68k/netbsd-elf.h (SUBTARGET_OVERRIDE_OPTIONS): Remove.
2240 2002-10-03 Richard Henderson <rth@redhat.com>
2242 * real.h (struct real_value): Use ENUM_BITFIELD.
2244 2002-10-03 Richard Henderson <rth@redhat.com>
2246 * config/i960/i960.md (call, call_value): Use emit_call_insn.
2248 2002-10-03 Steve Ellcey <sje@cup.hp.com>
2250 * config/pa/pa64-hpux.h (INIT_ENVIRONMENT): New.
2252 2002-10-03 Steve Ellcey <sje@cup.hp.com>
2254 * config.gcc (hppa*64*-*-hpux11*): Check gnu_ld.
2255 * config/pa/pa.h (MASK_GNU_LD): New.
2256 (TARGET_GNU_LD): New.
2257 * config/pa/pa64-hpux.h (LINK_SPEC): Set based
2258 on gnu-ld and MASK_GNU_LD.
2259 (SUBTARGET_SWITCHES): New gnu-ld & hp-ld flags.
2261 Thu Oct 3 23:35:51 CEST 2002 Jan Hubicka <jh@suse.cz>
2263 * i386.c (athlon_cost): Fix the move costs.
2265 Thu Oct 3 23:20:58 CEST 2002 Jan Hubicka <jh@suse.cz>
2267 * final.c (final): Use symbol name as function name for profiling.
2268 * profile.c (get_exec_counts): Likewise.
2269 (branch_prob): Likewise.
2271 2002-10-03 Jakub Jelinek <jakub@redhat.com>
2273 * longlong.h (__udiv_qrnnd): Remove PARAMS from prototype.
2275 2002-10-03 Jakub Jelinek <jakub@redhat.com>
2277 * gcc.c (print_multi_os_directory): New variable.
2278 (option_map): Support --print-multi-os-directory.
2279 (struct prefix_list): Add os_multilib field.
2280 (multilib_os_dir): New variable.
2281 (static_specs): Add multilib_options.
2282 (find_a_file): Add multilib argument. Search in GCC or OS multilib
2283 subdirs if non-zero.
2284 (read_specs, execute): Update callers.
2285 (find_file): Likewise. Don't prefix name with multilib_dir, instead
2286 pass 1 as multilib option.
2287 (display_help): Include --print-multi-os-directory.
2288 (add_prefix): Add os_multilib argument. Initialize pl->os_multilib.
2289 (process_command): Update callers. Handle --print-multi-os-directory.
2290 (do_spec_1) ['D']: Use multilib_os_directory if pl->os_multilib is
2292 (main): Update find_a_file and add_prefix callers.
2293 Handle print_multi_os_directory.
2294 (struct mdswitchstr): New.
2295 (mdswitches, n_mdswitches): New variables.
2296 (used_arg): Add MULTILIB_DEFAULT switches too if they are not
2297 present on the command line nor their mutually incompatible
2299 (default_arg): Optimize.
2300 (set_multilib_dir): Compute multilib_os_dir. Initialize mdswitches
2302 (print_multilib_info): Only print GCC multilib dir name, not OS
2304 * genmultilib: Add osdirnames parameter. Output multilib_options
2305 variable. If osdirnames is specified, output dirnames as
2307 * mklibgcc.in: Use MULTILIB_OSDIRNAMES, --print-multi-directory
2308 and --print-multi-os-directory instead of SHLIB_SLIBDIR_SUFFIXES
2309 to compute libgcc_s soname and install path.
2310 * Makefile.in (libgcc.mk): Pass MULTILIB_OSDIRNAMES instead of
2311 SHLIB_SLIBDIR_SUFFIXES to mklibgcc.
2312 (s_mlib): Pass MULTILIB_OSDIRNAMES or nothing as last genmultilib
2315 * config/sparc/t-linux64 (MULTILIB_OSDIRNAMES): Set.
2316 (SHLIB_SLIBDIR_SUFFIXES): Remove.
2317 * config/sparc/linux64.h (STARTFILE_SPEC32, STARTFILE_SPEC64,
2318 ENDFILE_SPEC32, ENDFILE_SPEC64, ENDFILE_COMMON): Remove.
2319 (STARTFILE_SPEC, ENDFILE_SPEC): Don't distinguish between -m32
2321 * config/sparc/t-sol2-64 (MULTILIB_OSDIRNAMES): Set.
2322 (SHLIB_SLIBDIR_SUFFIXES): Remove.
2323 * config/sparc/sol2-bi.h (STARTFILE_ARCH64_SPEC): Remove.
2324 (STARTFILE_ARCH_SPEC): Remove.
2325 * config/i386/t-linux64 (MULTILIB_OSDIRNAMES): Set.
2326 (SHLIB_SLIBDIR_SUFFIXES): Remove.
2327 * config/i386/linux64.h (STARTFILE_PREFIX_SPEC): Remove.
2328 * config/mips/t-iris6 (MULTILIB_OSDIRNAMES): Set.
2329 (SHLIB_SLIBDIR_SUFFIXES): Remove.
2331 Thu Oct 3 21:42:20 CEST 2002 Jan Hubicka <jh@suse.cz>
2333 * predict.c (choose_function_section): Avoid choice for linkonce functions.
2335 Thu Oct 3 15:15:00 CEST 2002 Jan Hubicka <jh@suse.cz>
2337 * i386.md (lea to mul peep2): Fix condition.
2339 2002-10-02 John David Anglin <dave@hiauly1.hia.nrc.ca>
2341 * pa-linux.h (FUNCTION_OK_FOR_SIBCALL): Delete macro.
2342 * pa32-linux.h (FUNCTION_OK_FOR_SIBCALL): Define.
2344 2002-10-02 Richard Henderson <rth@redhat.com>
2347 * config/i386/i386.c (ix86_register_move_cost): Increase cost
2348 for secondary_memory_needed pairs.
2350 2002-10-02 Nathanael Nerode <neroden@gcc.gnu.org>
2352 * doc/vms.texi: Blow away false include file section.
2354 2002-10-02 Roger Sayle <roger@eyesopen.com>
2356 PR optimization/6627
2357 * toplev.c (force_align_functions_log): New global variable.
2358 * flags.h (force_align_functions_log): Add extern prototype.
2359 * varasm.c (assemble_start_function): Use it to force minimum
2361 * config/i386/i386.h (FUNCTION_BOUNDARY): Set the correct
2362 minimum function alignment to one byte.
2363 (TARGET_PTRMEMFUNC_VBIT_LOCATION): Store the virtual bit in
2364 the least significant bit of vtable member function pointers.
2365 * tree.h (enum ptrmemfunc_vbit_where_t): Move definition to
2366 here from cp/cp-tree.h.
2368 Wed Oct 2 17:01:36 CEST 2002 Jan Hubicka <jh@suse.cz>
2370 * i386.c (print_operand_address): Use RIP addressing for offsetted
2373 2002-09-30 David S. Miller <davem@redhat.com>
2376 * config/sparc/sparc.md (movdi_insn_sp32_v9): Accept 'e' regs.
2377 (movdi reg/reg split): Match only on sparc32, and v9 when int regs.
2379 2002-10-01 Roger Sayle <roger@eyesopen.com>
2381 * unroll.c (loop_iterations): Revert 2002-09-08 change.
2383 2002-10-01 Richard Henderson <rth@redhat.com>
2385 * real.c (real_to_decimal): Crop trailing zeros for DIGITS < 0.
2386 (real_to_hexadecimal): Likewise.
2387 * print-rtl.c (print_rtx): If we are linked with real.c, don't
2388 dump the XWINT fields of a floating point CONST_DOUBLE.
2390 2002-10-01 Jason Thorpe <thorpej@wasabisystems.com>
2392 * config/vax/elf.h (FUNCTION_PROFILER): Fix __mcount call.
2394 2002-10-01 Richard Henderson <rth@redhat.com>
2396 * calls.c (precompute_register_parameters): Force non-legitimate
2397 constants into pseudos.
2399 2002-10-01 Nick Clifton <nickc@redhat.com>
2401 * config/rs6000/spe.md (spe_evrlwi): Add missing third operand
2402 to assembler template.
2404 2002-10-01 Richard Henderson <rth@redhat.com>
2406 * dwarf2out.c (loc_descriptor_from_tree): Relax requirement
2407 for TLS debug info to !DECL_EXTERNAL.
2409 2002-10-01 Matt Thomas <matt@3am-software.com>
2410 Jason Thorpe <thorpej@wasabisystems.com>
2412 * config.gcc (vax-*-netbsdelf*): Enable configuration.
2413 * config/elfos.h (PCC_BITFIELD_TYPE_MATTERS): Define only
2414 if not already defined.
2415 * config/vax/elf.h: New file.
2416 * config/vax/netbsd-elf.h: New file.
2417 * config/vax/vax.c: Include "debug.h".
2418 (vax_output_function_prologue): Add dwarf2 support. Use
2419 MAIN_NAME_P when checking for VMS_TARGET stack adjust.
2420 * config/vax/vax.h (CONST_OK_FOR_LETTER_P): Add cases for
2421 'J' [0..63], 'K' [-128..127], 'L' [-32768..32767],
2422 'M' [0..255], 'N' [0..65535], and, 'O' [-63..-1].
2423 (VAX_ISTREAM_SYNC): Remove.
2424 (INITIALIZE_TRAMPOLINE): Use gen_sync_istream.
2425 (JUMP_TABLES_IN_TEXT_SECTION): Define.
2426 (ASM_OUTPUT_REG_POP): Use reg_names for the stack pointer.
2427 (ASM_OUTPUT_ADDR_VEC_ELT): Use ASM_GENERATE_INTERNAL_LABEL
2429 (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
2430 (PRINT_OPERAND_PUNCT_VALID_P): Accept '|'.
2431 (PRINT_OPERAND): Output REGISTER_PREFIX for '|'.
2432 (INCOMING_RETURN_ADDR_RTX): Define.
2433 * config/vax/vax.md (VUNSPEC_BLOCKAGE)
2434 (VUNSPEC_SYNC_ISTREAM): Define.
2435 (blockage): Use VUNSPEC_BLOCKAGE.
2436 (sync_istream): New insn.
2438 2002-10-01 Richard Henderson <rth@redhat.com>
2440 * config/vax/vax.md (call_pop, *call_pop, call_value_pop)
2441 (*call_value_pop, call, call_value): Add dwarf2 EH support.
2444 2002-10-01 Nathan Sidwell <nathan@codesourcery.com>
2447 * c-typeck.c (build_c_cast): Warn about type punning which breaks
2448 type based aliasing.
2450 2002-10-01 Mark Mitchell <mark@codesourcery.com>
2452 * stor-layout.c (update_alignment_for_field): New function.
2453 (place_union_field): Use it.
2454 (place_field): Likewise.
2456 2002-10-01 Nathan Sidwell <nathan@codesourcery.com>
2459 * gcc.c (cc1_options): Add space on -auxbase-strip.
2461 2002-10-01 Jim Wilson <wilson@redhat.com>
2463 * config/v850/v850.h (EPILOGUE_USES): Define.
2465 2002-09-30 Andrew Haley <aph@redhat.com>
2467 * flow.c (insn_dead_p): When using non-call-exceptions, don't
2468 eliminate insns that may trap.
2469 * cse.c (insn_live_p): Likewise.
2471 2002-10-01 Richard Sandiford <rsandifo@redhat.com>
2473 * config/mips/mips.h (PROCESSOR_R4121): Rename to PROCESSOR_R4120.
2474 (TARGET_MIPS4121): Rename to TARGET_MIPS4120.
2475 * config/mips/mips.c (mips_cpu_info): Rename vr4121 to vr4120.
2476 * config/mips/mips.md: Apply same renaming here.
2478 2002-10-01 Richard Sandiford <rsandifo@redhat.com>
2480 * config/mips/mips.c (PROCESSOR_R4320, TARGET_MIPS4320): Remove.
2481 (GENERATE_MULT3_SI): Remove use of TARGET_MIPS4320.
2482 * config/mips/mips.c (mips_cpu_info): Remove vr4320 entry.
2483 * config/mips/mips.md (define_attr cpu): Remove r4320.
2484 Remove vr4320 scheduler and uses of TARGET_MIPS4320.
2486 2002-10-01 Richard Sandiford <rsandifo@redhat.com>
2488 * config/mips/mips.c (mips16_strings): New variable.
2489 (mips_output_function_epilogue): Clear the SYMBOL_REF_FLAG of every
2490 symbol in mips16_strings. Free the list.
2491 (mips_encode_section_info): Keep track of local strings.
2493 2002-10-01 Richard Sandiford <rsandifo@redhat.com>
2495 * config/mips/mips.md (bunge, bltgt, bungt): New define_expands.
2496 (sordered_df, sordered_sf): Remove.
2497 * config/mips/mips.c (get_float_compare_codes): New fn.
2498 (gen_int_relational, gen_conditional_move): Use it.
2500 2002-10-01 Richard Sandiford <rsandifo@redhat.com>
2502 * config/mips/mips-protos.h (mips_emit_fcc_reload): Declare.
2503 * config/mips/mips.h (PREDICATE_CODES): Add fcc_register_operand.
2504 * config/mips/mips.c (fcc_register_operand): New function.
2505 (mips_emit_fcc_reload): New function, extracted from reload_incc.
2506 (override_options): Allow TFmode values in float registers
2508 * cnfig/mips/mips.md (reload_incc): Change destination prediate
2509 to fcc_register_operand. Remove misleading source constraint.
2510 Use mips_emit_fcc_reload.
2511 (reload_outcc): Duplicate reload_incc.
2513 2002-09-30 Ulrich Weigand <uweigand@de.ibm.com>
2515 * longlong.h: Partially synchronize with GMP-4.1 version:
2516 Use i370 definitions also for s390.
2517 Add generic definition of umul_ppmm in terms of smul_ppmm.
2518 [s390] (umul_ppmm): Remove.
2519 [s390] (smul_ppmm): Fix incorrect assembler constraints.
2520 [s390] (smul_ppmm, sdiv_qrnnd): Rename __xx to __x.
2522 2002-09-30 Bob Wilson <bob.wilson@acm.org>
2524 * config/xtensa/xtensa.h (REG_CLASS_NAMES, REG_CLASS_CONTENTS):
2525 Add new RL_REGS register class.
2526 (PREFERRED_RELOAD_CLASS, PREFERRED_OUTPUT_RELOAD_CLASS):
2527 Call xtensa_preferred_reload_class for both input and output reloads.
2528 * config/xtensa/xtensa.c (xtensa_regno_to_class): Use new RL_REGS class.
2529 (xtensa_preferred_reload_class): Handle output reloads; use RL_REGS
2530 instead of either AR_REGS or GR_REGS classes.
2531 (xtensa_secondary_reload_class): Use new RL_REGS class.
2532 * config/xtensa/xtensa-protos.h (xtensa_preferred_reload_class): Update.
2534 2002-09-30 John David Anglin <dave@hiauly1.hia.nrc.ca>
2536 * pa.c (hppa_encode_label): Don't drop '*' from function labels.
2537 (pa_strip_name_encoding): Strip '@' and '*', in that order.
2538 * pa.h (ASM_OUTPUT_LABELREF): Output user_label_prefix except when
2539 there is a '*' prefix in NAME.
2541 Mon Sep 30 21:33:23 CEST 2002 Jan Hubicka <jh@suse.cz>
2543 * reload.c (push_reload): Handle subregs and secondary memory.
2544 * reload1.c (gen_reload): Likewise.
2546 * jump.c (reg_or_subregno): New function.
2547 * rtl.h (reg_or_subregno): Declare
2548 * unroll.c (find_splittable_givs): Handle subregs.
2550 2002-09-30 Mark Mitchell <mark@codesourcery.com>
2552 * store-layout.c (finish_record_layout): Add free_p parameter.
2553 (layout_type): Pass it.
2554 * tree.h (finish_record_layout): Update prototype.
2556 Mon Sep 30 14:57:18 CEST 2002 Jan Hubicka <jh@suse.cz>
2558 * i386.h (TARGET_CPP_CPU_BUILTINS): Define __SSE_MATH__.
2560 * gcse.c (cprop_jump): Check that the register has not
2562 (cprop_jump): Likewise.
2564 2002-09-30 Richard Earnshaw <rearnsha@arm.com>
2566 * arm.h (BASE_REG_CLASS): Always return LO_REGS for Thumb.
2567 (MODE_BASE_REG_CLASS, case Thumb): Only return BASE_REGS if we know
2568 that we have a SImode access, and only then if reload hasn't completed;
2569 for all other cases, use LO_REGS.
2571 2002-09-29 Richard Henderson <rth@redhat.com>
2573 * real.c (real_from_string): Apply sign last. Tidy exponent handling.
2575 2002-09-29 Richard Henderson <rth@redhat.com>
2578 * combine.c (force_to_mode): Handle FLOAT_MODE destinations
2581 2002-09-29 David Edelsohn <edelsohn@gnu.org>
2583 * real.h (ibm_extended_format): Declare.
2584 * real.c (encode_ibm_extended, decode_ibm_extended): New
2587 2002-09-29 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
2589 * darwin-protos.h (darwin_asm_output_dwarf_delta): Prototype.
2591 * ia64.c (ia64_hpux_asm_file_end): Const-ify.
2593 2002-09-29 John David Anglin <dave@hiauly1.hia.nrc.ca>
2595 * expmed.c (extract_bit_field): Fix bit-field extraction from SUBREGs.
2597 2002-09-29 Kazu Hirata <kazu@cs.umass.edu>
2599 * builtins.def: Fix comment formatting.
2600 * c-common.def: Likewise.
2601 * cfgcleanup.c: Likewise.
2602 * combine.c: Likewise.
2603 * gengtype.c: Likewise.
2604 * params.def: Likewise.
2605 * predict.def: Likewise.
2606 * rtl.def: Likewise.
2607 * stab.def: Likewise.
2608 * stor-layout.c: Likewise.
2609 * tree.def: Likewise.
2610 * config/darwin.c: Likewise.
2611 * config/darwin.h: Likewise.
2612 * config/dbxcoff.h: Likewise.
2613 * config/elfos.h: Likewise.
2614 * config/fp-bit.c: Likewise.
2615 * config/freebsd-spec.h: Likewise.
2616 * config/interix.h: Likewise.
2617 * config/libgloss.h: Likewise.
2618 * config/linux-aout.h: Likewise.
2619 * config/linux.h: Likewise.
2620 * config/lynx-ng.h: Likewise.
2621 * config/lynx.h: Likewise.
2622 * config/netbsd-aout.h: Likewise.
2623 * config/netbsd.h: Likewise.
2624 * config/netware.h: Likewise.
2625 * config/psos.h: Likewise.
2626 * config/ptx4.h: Likewise.
2628 2002-09-28 Kazu Hirata <kazu@cs.umass.edu>
2630 * ChangeLog.4: Fix typos.
2631 * ChangeLog.6: Likewise.
2632 * FSFChangeLog.10: Likewise.
2633 * genattrtab.c: Fix comment typos.
2634 * haifa-sched.c: Likewise.
2637 * config/arm/arm.c: Likewise.
2638 * config/arm/crti.asm: Likewise.
2639 * config/arm/crtn.asm: Likewise.
2640 * config/frv/frv.c: Likewise.
2641 * config/frv/frv.md: Likewise.
2642 * config/h8300/h8300.md: Likewise.
2643 * config/i386/rtemself.h: Likewise.
2644 * config/ia64/unwind-ia64.c: Likewise.
2645 * config/ip2k/ip2k.h: Likewise.
2646 * config/m88k/m88k.c: Likewise.
2647 * config/m88k/m88k.md: Likewise.
2648 * config/mips/sr71k.md: Likewise.
2649 * config/mmix/mmix.c: Likewise.
2650 * config/rs6000/rs6000.c: Likewise.
2651 * config/sh/sh.md: Likewise.
2653 2002-09-26 Theodore A. Roth <troth@verinet.com>
2655 * config/avr/avr.c: Eliminate use of _PC_ in pc relative insns.
2656 * config/avr/avr.md: Ditto.
2658 2002-09-27 Alexander N. Kabaev <ak03@gte.com>
2660 PR preprocessor/8055
2661 * cppmacro.c (stringify_arg): Do not overflow the buffer
2662 with the terminating NUL when the argument to be stringified
2665 2002-09-27 Richard Henderson <rth@redhat.com>
2667 * unroll.c (simplify_cmp_and_jump_insns): New.
2668 (unroll_loop): Use it. Use simplify_gen_foo+force_operand
2669 instead of expand_simple_foo.
2671 2002-09-27 Richard Henderson <rth@redhat.com>
2673 PR optimization/7520
2674 * cfganal.c (flow_active_insn_p): New.
2675 (forwarder_block_p): Use it.
2677 2002-09-27 Richard Henderson <rth@redhat.com>
2679 * emit-rtl.c (active_insn_p): Revert last change.
2681 2002-09-27 Jakub Jelinek <jakub@redhat.com>
2683 * doc/extend.texi (tls_model): Document.
2684 * varasm.c (decl_tls_model): New.
2685 * c-common.c (handle_tls_model_attribute): New.
2686 (c_common_attribute_table): Add tls_model.
2687 * config/alpha/alpha.c (alpha_encode_section_info): Use
2689 * flags.h (enum tls_model, flag_tls_default): Move...
2690 * tree.h (enum tls_model, flag_tls_default): ...here.
2691 (decl_tls_model): New prototype.
2692 * config/ia64/ia64.c (ia64_encode_section_info): Likewise.
2693 * config/i386/i386.c (ix86_encode_section_info): Likewise.
2694 * config/i386/i386.md (tls_global_dynamic, tls_local_dynamic_base):
2697 2002-09-27 Kazu Hirata <kazu@cs.umass.edu>
2699 * LANGUAGES: Follow spelling conventions.
2700 * rtl.def: Likewise.
2701 * sbitmap.c: Likewise.
2702 * sched-int.h: Likewise.
2703 * sched-rgn.c: Likewise.
2704 * sibcall.c: Likewise.
2705 * simplify-rtx.c: Likewise.
2707 * stab.def: Likewise.
2709 * stor-layout.c: Likewise.
2710 * target.h: Likewise.
2711 * timevar.c: Likewise.
2712 * toplev.c: Likewise.
2713 * tree-dump.c: Likewise.
2714 * tree-inline.c: Likewise.
2716 * tree.def: Likewise.
2718 * unroll.c: Likewise.
2719 * varasm.c: Likewise.
2720 * vmsdbgout.c: Likewise.
2721 * treelang/treelang.texi: Likewise.
2722 * treelang/treetree.c: Likewise.
2724 2002-09-27 Kazu Hirata <kazu@cs.umass.edu>
2726 * config/h8300/h8300.c (compute_saved_regs): Use a macro
2727 instead of a hard register number.
2728 (get_shift_alg): Use an enumerated type instead of numbers.
2729 (h8300_shift_needs_scratch_p): Likewise.
2731 2002-09-26 David S. Miller <davem@redhat.com>
2733 PR optimization/7335
2734 * calls.c (emit_library_call_value_1): Passing args by reference
2735 converts a CONST function into a PURE one.
2737 2002-09-26 David Edelsohn <edelsohn@gnu.org>
2739 * dbxout.c (FORCE_TEXT): Switch to current_function_decl, not
2741 * xcoffout.h (DBX_STATIC_BLOCK_START): Remove explicit change to
2743 * config/rs6000/rs6000.c (rs6000_override_options): Allow
2744 function-sections and data-sections functionality on AIX.
2746 2002-09-26 David Edelsohn <edelsohn@gnu.org>
2747 Dale Johannesen <dalej@apple.com>
2749 * config/rs6000/rs6000.c (rs6000_emit_move): Insert zero-extend
2750 in RTL for sub-word loads from memory.
2752 2002-09-26 Richard Henderson <rth@redhat.com>
2755 * sched-deps.c (sched_analyze_insn): Make clobber insns depend
2758 2002-09-26 Richard Henderson <rth@redhat.com>
2760 * emit-rtl.c (const_double_htab_eq): Remove unused variable.
2762 2002-09-26 Chris Lattner <sabre@nondot.org>
2764 * ssa.c (rename_insn_1): Handle RENAME_NO_RTX correctly when
2765 handling undefined values.
2767 2002-09-26 Richard Henderson <rth@redhat.com>
2770 * emit-rtl.c (active_insn_p): Consider a clobber of the
2771 function return value to be active even after reload.
2773 2002-09-27 Alan Modra <amodra@bigpond.net.au>
2775 * doloop.c (doloop_modify_runtime <biv skips initial incr>): Adjust
2776 by absolute loop increment, not loop increment.
2778 2002-09-26 Kazu Hirata <kazu@cs.umass.edu>
2780 * c-common.h: Follow spelling conventions.
2781 * cpplex.c: Likewise.
2782 * cpplib.h: Likewise.
2783 * gthr-dce.h: Likewise.
2784 * gthr-posix.h: Likewise.
2785 * optabs.c: Likewise.
2786 * output.h: Likewise.
2787 * profile.c: Likewise.
2788 * protoize.c: Likewise.
2789 * ra-rewrite.c: Likewise.
2791 * recog.c: Likewise.
2792 * reg-stack.c: Likewise.
2793 * regclass.c: Likewise.
2794 * regmove.c: Likewise.
2795 * reload.c: Likewise.
2796 * reload.h: Likewise.
2797 * reload1.c: Likewise.
2798 * reorg.c: Likewise.
2799 * resource.c: Likewise.
2801 * rtlanal.c: Likewise.
2803 2002-09-26 Steve Ellcey <sje@cup.hp.com>
2805 * config/ia64/ia64.c (ia64_expand_load_address): Ensure correct mode
2808 2002-09-24 Eric Christopher <echristo@redhat.com>
2810 * config/mips/elf.h: Add HANDLE_SYSV_PRAGMA.
2811 * config/mips/elf64.h: Ditto.
2813 2002-09-24 Eric Christopher <echristo@redhat.com>
2815 * except.c (expand_builtin_extract_return_address): Handle case
2816 where Pmode != ptr_mode.
2818 2002-09-26 Steve Ellcey <sje@cup.hp.com>
2820 * config/ia64/hpux.h (ASM_OUTPUT_EXTERNAL_LIBCALL): New
2822 2002-09-26 Steve Ellcey <sje@cup.hp.com>
2824 * config/ia64/hpux.h (TARGET_DEFAULT): Include TARGET_ILP32.
2826 2002-09-26 Igor Shevlyakov <igor@microunity.com>
2828 * combine.c (simplify_set): Don't call to force_to_mode if size
2829 of integer type is larger than HOST_BITS_PER_WIDE_INT.
2831 2002-09-26 Janis Johnson <janis187@us.ibm.com>
2833 * Makefile.in (qmtest-g++): Fix file path.
2835 2002-09-26 Ulrich Weigand <uweigand@de.ibm.com>
2837 * expr.c (expand_expr) [MINUS_EXPR]: Convert A - const to
2838 A + (-const) on RTX level, even for unsigned types.
2840 2002-09-26 Ulrich Weigand <uweigand@de.ibm.com>
2842 * reload.c (dup_replacements): New function.
2843 (find_reloads): Use it to duplicate replacements at the top level
2844 of match_dup operands.
2846 2002-09-26 Miles Bader <miles@gnu.org>
2848 * v850.md ("length"): Change default value to 4.
2850 2002-09-26 Kazu Hirata <kazu@cs.umass.edu>
2852 * ChangeLog.1: Follow spelling conventions.
2853 * ChangeLog.4: Likewise.
2854 * ChangeLog.6: Likewise.
2855 * FSFChangeLog.11: Likewise.
2856 * doc/cpp.texi: Likewise.
2857 * doc/invoke.texi: Likewise.
2858 * doc/tm.texi: Likewise.
2860 2002-09-26 Nick Clifton <nickc@redhat.com>
2862 * config.gcc: Add x prefix to v850e case for handling
2865 2002-09-25 David S. Miller <davem@redhat.com>
2868 * config/sparc/sparc.c (set_extends): SImode ASHIFT does not
2871 2002-09-25 Richard Henderson <rth@redhat.com>
2873 * emit-rtl.c (const_double_htab_eq): Distinguish integer and
2874 fp CONST_DOUBLE; use real_identical.
2876 2002-09-25 Mark Mitchell <mark@codesourcery.com>
2878 * doc/invoke.texi: Add more -Wabi examples.
2880 2002-09-25 Richard Sandiford <rsandifo@redhat.com>
2882 * config/mips/mips.h (TARGET_MIPS4100): Add missing bracket.
2884 2002-09-24 Denis Chertykov <denisc@overta.ru>
2886 * config/ip2k/ip2k.c (function_epilogue): Fix wrong numbers in
2887 cases of optimizing "add sp,w" to "inc sp".
2889 2002-09-24 Adam Nemet <anemet@lnxw.com>
2891 * config/arm/arm.c (thumb_unexpanded_epilogue): Don't generate
2892 epilogue for naked functions.
2894 2002-09-24 Adam Nemet <anemet@lnxw.com>
2895 Nick Clifton <nickc@redhat.com>
2897 * config/arm/arm.h (THUMB_FUNCTION_PROFILER): Remove.
2898 (FUNCTION_PROFILER): Only invoke THUMB_FUNCTION_PROFILER if it
2901 2002-09-24 Ulrich Weigand <uweigand@de.ibm.com>
2903 * config/s390/s390.c (preferred_la_operand_p): New function.
2904 * config/s390/s390-protos.h (preferred_la_operand_p): Declare it.
2905 * config/s390/s390.md ("addaddr_esame", "*la_ccclobber"): Replace by ...
2906 ("*la_64_cc", "*la_31_cc", splitters): ... these.
2907 ("*la_31"): Deactivate for TARGET_64BIT.
2908 ("*la_31_and", "*la_31_and_cc"): New.
2910 2002-09-24 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
2912 * real.h (real_value): Make `exp' explicitly signed.
2914 2002-09-24 Kazu Hirata <kazu@cs.umass.edu>
2916 * config/elfos.h: Follow spelling conventions.
2917 * config/alpha/alpha.h: Likewise.
2918 * config/arc/arc.h: Likewise.
2919 * config/arm/arm.md: Likewise.
2920 * config/avr/avr.h: Likewise.
2921 * config/cris/cris.md: Likewise.
2922 * config/d30v/d30v.h: Likewise.
2923 * config/frv/frv.c: Likewise.
2924 * config/frv/frv.h: Likewise.
2925 * config/h8300/h8300.c: Likewise.
2926 * config/h8300/h8300.h: Likewise.
2927 * config/h8300/h8300.md: Likewise.
2928 * config/i386/cygwin.h: Likewise.
2929 * config/i386/i386.h: Likewise.
2930 * config/i386/sysv3.h: Likewise.
2931 * config/i960/i960.h: Likewise.
2932 * config/ia64/ia64.h: Likewise.
2933 * config/ia64/ia64.md: Likewise.
2934 * config/ip2k/ip2k.h: Likewise.
2935 * config/m32r/m32r.h: Likewise.
2936 * config/m68k/m68k.h: Likewise.
2937 * config/m88k/m88k.h: Likewise.
2938 * config/mcore/mcore.c: Likewise.
2939 * config/mcore/mcore.h: Likewise.
2940 * config/mcore/mcore.md: Likewise.
2941 * config/mips/mips.h: Likewise.
2942 * config/mmix/mmix.h: Likewise.
2943 * config/mmix/mmix.md: Likewise.
2944 * config/ns32k/netbsd.h: Likewise.
2945 * config/ns32k/ns32k.h: Likewise.
2946 * config/ns32k/ns32k.md: Likewise.
2947 * config/pa/pa.h: Likewise.
2948 * config/romp/romp.h: Likewise.
2949 * config/rs6000/rs6000.h: Likewise.
2950 * config/rs6000/rs6000.md: Likewise.
2951 * config/sparc/sparc.h: Likewise.
2952 * config/stormy16/stormy-abi: Likewise.
2953 * config/stormy16/stormy16.h: Likewise.
2954 * config/vax/vax.h: Likewise.
2956 2002-09-23 Zack Weinberg <zack@codesourcery.com>
2958 * version.c (version_string): Now const char[].
2959 * version.h: Update to match.
2961 2002-09-23 Richard Henderson <rth@redhat.com>
2963 * config/i386/i386.h (MASK_ACCUMULATE_OUTGOING_ARGS_SET, MASK_MMX_SET,
2964 MASK_SSE_SET, MASK_SSE2_SET, MASK_3DNOW_SET, MASK_3DNOW_A_SET): Kill.
2965 (TARGET_SWITCHES): Don't reference them.
2966 * config/i386/i386.c (override_options): Use target_flags_explicit
2967 to examine bits set by the user.
2969 2002-09-23 Dale Johannesen <dalej@apple.com>
2971 * dbxout.c (dbxout_parms): Set current_sym_code for params
2972 passed on stack by invisible reference.
2974 2002-09-23 Richard Earnshaw <rearnsha@arm.com>
2976 * arm/unknown-elf.h (ASM_OUTPUT_ALIGNED_DECL_LOCAL): Always allocate
2977 at least one byte of space.
2979 2002-09-23 Mark Mitchell <mark@codesourcery.com>
2981 * c-common.h (flag_abi_version): Fix typo in comment.
2982 * doc/invoke.texi (flag_abi_version): Document default value.
2984 2002-09-23 Hans-Peter Nilsson <hp@axis.com>
2986 * doc/extend.texi (Extended Asm): Clarify that overlap between
2987 asm-declared register variables used in an asm and the asm clobber
2988 list is not allowed.
2989 * stmt.c (decl_conflicts_with_clobbers_p): New function.
2990 (expand_asm_operands): Keep track of clobbered registers. Call
2991 decl_conflicts_with_clobbers_p for each input and output operand.
2992 If no conflicts found before, also do conflict sanity check when
2995 2002-09-23 Richard Henderson <rth@redhat.com>
2997 * c-common.c (cpp_define_data_format): Remove.
2998 (cb_register_builtins): Don't define __WCHAR_BIT__, __SHRT_BIT__,
2999 __INT_BIT__, __LONG_BIT__, __LONG_LONG_BIT__, __FLOAT_BIT__,
3000 __DOUBLE_BIT__, __LONG_DOUBLE_BIT__.
3001 * doc/cpp.texi: Don't document them either.
3002 (__SCHAR_MAX__, __SHRT_MAX__, __INT_MAX__, __LONG_MAX__,
3003 __LONG_LONG_MAX__): Document.
3004 (__TARGET_FLOAT_FORMAT__): Remove.
3006 2002-09-23 Richard Henderson <rth@redhat.com>
3008 * real.c (do_multiply): Normalize U before addition.
3010 2002-09-23 Mark Mitchell <mark@codesourcery.com>
3012 * c-common.c (flag_abi_version): New variable.
3013 * c-common.h (flag_abi_version): Declare it.
3014 * c-opts.c (missing_arg): Add -fabi-version.
3015 (c_common_decode_option): Process -fabi-version.
3016 * doc/invoke.texi (-fabi-version): Document it.
3017 (-Wabi): Add information about bit-fields in unions.
3019 2002-09-22 Jason Thorpe <thorpej@wasabisystems.com>
3021 * config/mips/netbsd.h (SUBTARGET_ASM_SPEC): Always pass -KPIC
3022 unless -fno-pic or -fno-PIC is specified.
3024 2002-09-22 John David Anglin <dave@hiauly1.hia.nrc.ca>
3026 * c-common.c (preprocessing_trad_p): Define.
3027 * pa-hiux.h, pa-hpux.h, pa-hpux7.h (CPP_PREDEFINES): Delete.
3028 (TARGET_OS_CPP_BUILTINS, SUBTARGET_SWITCHES): Define.
3029 * pa-hpux10.h (TARGET_OS_CPP_BUILTINS, CPP_SPEC): Define.
3030 * pa-hpux11.h (TARGET_OS_CPP_BUILTINS): Define.
3031 * pa-linux.h (CPP_PREDEFINES): Delete.
3032 (TARGET_OS_CPP_BUILTINS, CPP_SPEC): Define.
3033 * pa32-linux.h, pa64-linux.h (CPP_SPEC): Delete.
3034 * pa-osf.h, pa-pro-end.h, rtems.h (CPP_PREDEFINES): Delete.
3035 (TARGET_OS_CPP_BUILTINS): Define.
3036 * pa.h (MASK_SIO, TARGET_SIO, TARGET_PA_10): Define.
3037 (TARGET_SWITCHES): Reformat. Use N_() macro. Add SUBTARGET_SWITCHES.
3038 (SUBTARGET_SWITCHES): Provide default definition.
3039 (TARGET_OPTIONS): Reformat. Use N_() macro.
3040 (CPP_PA10_SPEC, CPP_PA11_SPEC, CPP_PA20_SPEC, CPP_64BIT_SPEC,
3041 CPP_CPU_DEFAULT_SPEC, CPP_64BIT_DEFAULT_SPEC, SUBTARGET_EXTRA_SPECS,
3042 EXTRA_SPECS, CPP_SPEC, CPLUSPLUS_CPP_SPEC, CPP_PREDEFINES): Delete.
3043 (TARGET_CPU_CPP_BUILTINS): Define.
3044 (TARGET_OS_CPP_BUILTINS): Define for BSD-like systems.
3045 * doc/invoke.texi (msio, mwsio): Document new hppa options.
3046 * doc/tm.texi (TARGET_CPU_CPP_BUILTINS): Document macro
3047 preprocessing_trad_p().
3049 2002-09-22 Jason Thorpe <thorpej@wasabisystems.com>
3051 * doc/install.texi: Document behavior of --with-headers and
3052 --with-libs when arguments are omitted.
3054 2002-09-22 Kazu Hirata <kazu@cs.umass.edu>
3056 * dbxout.c: Follow spelling conventions.
3057 * defaults.h: Likewise.
3059 * diagnostic.h: Likewise.
3060 * doloop.c: Likewise.
3061 * dwarf2out.c: Likewise.
3062 * dwarfout.c: Likewise.
3063 * emit-rtl.c: Likewise.
3064 * except.c: Likewise.
3065 * explow.c: Likewise.
3066 * expmed.c: Likewise.
3069 * flags.h: Likewise.
3071 * fold-const.c: Likewise.
3072 * function.c: Likewise.
3073 * function.h: Likewise.
3075 * gcov-io.h: Likewise.
3078 * genattrtab.c: Likewise.
3079 * genconfig.c: Likewise.
3080 * genrecog.c: Likewise.
3081 * ggc-page.c: Likewise.
3083 * global.c: Likewise.
3084 * gthr-win32.h: Likewise.
3085 * integrate.c: Likewise.
3087 * langhooks.c: Likewise.
3088 * langhooks.h: Likewise.
3089 * line-map.h: Likewise.
3090 * local-alloc.c: Likewise.
3091 * longlong.h: Likewise.
3095 Tue Aug 27 22:26:35 CEST 2002 Jan Hubicka <jh@suse.cz>
3097 * i386.h (BIGGEST_FIELD_ALIGNMENT): Set proper default for x86_64.
3099 Tue Aug 27 20:07:01 CEST 2002 Jan Hubicka <jh@suse.cz>
3101 * i386.c (overwrite_options): Set -mpreferred-stack-boundary to 128
3102 for -Os/TARGET_64BIT too.
3104 2002-09-21 Kazu Hirata <kazu@cs.umass.edu>
3106 * ChangeLog: Follow spelling conventions.
3107 * ChangeLog.0: Likewise.
3108 * ChangeLog.1: Likewise.
3109 * ChangeLog.2: Likewise.
3110 * ChangeLog.3: Likewise.
3111 * ChangeLog.4: Likewise.
3112 * ChangeLog.5: Likewise.
3113 * ChangeLog.6: Likewise.
3114 * FSFChangeLog.10: Likewise.
3115 * FSFChangeLog.11: Likewise.
3116 * alias.c: Likewise.
3117 * basic-block.h: Likewise.
3118 * c-aux-info.c: Likewise.
3119 * c-common.c: Likewise.
3120 * c-common.h: Likewise.
3121 * c-decl.c: Likewise.
3122 * c-format.c: Likewise.
3123 * c-semantics.c: Likewise.
3124 * c-typeck.c: Likewise.
3125 * calls.c: Likewise.
3126 * cfganal.c: Likewise.
3127 * cfgloop.c: Likewise.
3128 * collect2.c: Likewise.
3129 * combine.c: Likewise.
3130 * conflict.c: Likewise.
3131 * cppexp.c: Likewise.
3132 * cppfiles.c: Likewise.
3133 * cpphash.h: Likewise.
3134 * cppinit.c: Likewise.
3135 * cpplex.c: Likewise.
3136 * cpplib.c: Likewise.
3137 * cpplib.h: Likewise.
3138 * cppmacro.c: Likewise.
3141 2002-09-21 Richard Earnshaw <rearnsha@arm.com>
3143 * netbsd-aout.h (NETBSD_LINK_SPEC_AOUT): New, takes old definition of
3145 (LINK_SPEC): Define to NETBSD_LINK_SPEC_AOUT.
3146 * arm/netbsd.h (SUBTARGET_EXTRA_SEPCS): Add NETBSD_LINK_SPEC_AOUT.
3147 (LINK_SPEC): Rework to use NETBSD_LINK_SPEC_AOUT).
3149 2002-09-21 Richard Earnshaw <rearnsha@arm.com>
3152 * cse.c (fold_rtx): Calculate old_cost before we fold each
3155 2002-09-21 Richard Henderson <rth@redhat.com>
3157 * c-common.c (cpp_define_data_format): Remove __GCC_LITTLE_ENDIAN__,
3158 __GCC_BIG_ENDIAN__, __TARGET_BITS_ORDER__, __TARGET_BYTES_ORDER__,
3159 __TARGET_INT_WORDS_ORDER__, __TARGET_FLOAT_WORDS_ORDER__,
3160 __TARGET_USES_VAX_F_FLOAT__, __TARGET_USES_VAX_D_FLOAT__,
3161 __TARGET_USES_VAX_G_FLOAT__, __TARGET_USES_VAX_H_FLOAT__.
3162 * doc/cpp.texi: Don't document them.
3164 2002-09-21 Richard Henderson <rth@redhat.com>
3166 * c-common.c (builtin_define_float_constants): Use real_format
3167 to get the floating-point parameters.
3169 2002-09-21 Richard Henderson <rth@redhat.com>
3171 * real.c (struct real_format): Move to real.h.
3172 (real_format_for_mode): Rename from fmt_for_mode; update all users;
3173 initialize with ieee defaults.
3174 (real_to_target_fmt, real_from_target_fmt): New.
3175 (ieee_single_format, ieee_double_format, ieee_extended_motorola_format,
3176 ieee_extended_intel_96_format, ieee_extended_intel_128_format,
3177 ieee_quad_format, i370_single_format, i370_double_format,
3178 c4x_single_format, c4x_extended_format): Rename from s/_format//.
3179 (ieee_quad_format): Fix emin.
3180 (format_for_size, init_real_once): Remove.
3181 * real.h (struct real_format): Move from real.c.
3182 (real_format_for_mode): Declare.
3183 (real_to_target_fmt, real_from_target_fmt): Declare.
3184 (ieee_single_format, ieee_double_format, ieee_extended_motorola_format,
3185 ieee_extended_intel_96_format, ieee_extended_intel_128_format,
3186 ieee_quad_format, vax_f_format, vax_d_format, vax_g_format,
3187 i370_single_format, i370_double_format, c4x_single_format,
3188 c4x_extended_format): Declare.
3189 * toplev.c (do_compile): Don't call init_real_once.
3191 * defaults.h (INTEL_EXTENDED_IEEE_FORMAT): Remove.
3192 * doc/tm.texi (INTEL_EXTENDED_IEEE_FORMAT): Remove.
3194 * config/alpha/alpha.h (TARGET_FLOAT_FORMAT): Define.
3195 * config/alpha/osf5.h (LONG_DOUBLE_TYPE_SIZE): 64, if vax mode.
3196 * config/alpha/alpha.c (override_options): Set real_format_for_mode
3197 for VAX, if enabled.
3199 * config/c4x/c4x.c (c4x_override_options): Set real_format_for_mode
3202 * config/i370/i370.h (OVERRIDE_OPTIONS): New.
3203 * config/i370/i370.c (override_options): New.
3204 * config/i370/i370-protos.h: Update.
3206 * config/i386/i386.c (override_options): Set real_format_for_mode
3207 for Intel 80-bit extended.
3208 * config/i386/i386.h (INTEL_EXTENDED_IEEE_FORMAT): Remove.
3210 * config/i960/i960.h (LONG_DOUBLE_TYPE_SIZE): Mind -mlong-double-64.
3211 (OVERRIDE_OPTIONS): Move code...
3212 * config/i960/i960.c (i960_initialize): ... here. Set
3213 real_format_for_mode for Intel 80-bit extended.
3215 * config/ia64/ia64.c (ia64_override_options): Set real_format_for_mode
3216 for Intel 80-bit extended, if enabled.
3218 * config/m68k/m68k.c (override_options): Set real_format_for_mode
3219 for Motorola 96-bit extended.
3221 * config/vax/vax.h (OVERRIDE_OPTIONS): New.
3222 * config/vax/vax.c (override_options): New.
3223 * config/vax/vax-protos.h: Update.
3225 2002-09-21 Alan Modra <amodra@bigpond.net.au>
3227 * config/rs6000/rs6000.md (builtin_setjmp_receiver): Add
3230 * config/rs6000/rs6000.md (floatdisf2_internal2): Combine
3231 insns. Supply missing clobber of scratch reg.
3233 2002-09-20 Kazu Hirata <kazu@cs.umass.edu>
3235 * config/m32r/m32r.c: Follow spelling conventions.
3236 * config/m32r/m32r.h: Likewise.
3237 * config/m32r/m32r.md: Likewise.
3238 * config/m68k/m68k.c: Likewise.
3239 * config/m88k/m88k.c: Likewise.
3240 * config/mcore/mcore.c: Likewise.
3241 * config/mips/mips.c: Likewise.
3242 * config/mips/mips.h: Likewise.
3243 * config/mmix/mmix.c: Likewise.
3244 * config/mn10200/mn10200.c: Likewise.
3245 * config/ns32k/ns32k.h: Likewise.
3246 * config/pa/pa.c: Likewise.
3247 * config/pa/pa64-linux.h: Likewise.
3248 * config/pdp11/pdp11.h: Likewise.
3249 * config/romp/romp.c: Likewise.
3250 * config/romp/romp.h: Likewise.
3251 * config/rs6000/eabi.asm: Likewise.
3252 * config/rs6000/linux64.h: Likewise.
3253 * config/rs6000/rs6000.c: Likewise.
3254 * config/rs6000/rs6000.h: Likewise.
3255 * config/rs6000/rs6000.md: Likewise.
3256 * config/rs6000/sysv4.h: Likewise.
3257 * config/rs6000/xcoff.h: Likewise.
3259 2002-09-20 Jim Wilson <wilson@redhat.com>
3261 * config/v850/v850/lib1funcs.asm (__muldi3): Change r5 to r28.
3263 2002-09-20 Jakub Jelinek <jakub@redhat.com>
3265 * config/i386/i386.md (UNSPEC_GOTNTPOFF, UNSPEC_INDNTPOFF): New.
3266 * config/i386/i386.c (legitimate_pic_address_disp_p): Handle
3267 UNSPEC_GOTNTPOFF and UNSPEC_INDNTPOFF like UNSPEC_GOTTPOFF.
3268 (legitimate_address_p): Likewise.
3269 (legitimize_address): Use @gotntpoff and @indntpoff.
3270 (output_pic_addr_const): Handle UNSPEC_GOTNTPOFF and UNSPEC_INDNTPOFF.
3271 (output_addr_const_extra): Likewise.
3273 2002-09-20 Jim Wilson <wilson@redhat.com>
3275 * combine.c (try_combine): When split an instruction pair, where the
3276 first has a sign_extend src, verify that the src and dest modes match.
3278 2002-09-20 Richard Henderson <rth@redhat.com>
3280 * config/mips/mips.c (dfhigh, dflow, sfhigh, sflow): Remove.
3281 (override_options): Do not initialize them.
3282 (mips_const_double_ok): Allow no fp constants except zero,
3283 and not even that for mips16.
3284 (const_float_1_operand): Use dconst1.
3285 * config/mips/mips.md (movsf, movsf_internal1, movsf_internal2,
3286 movdf, movdf_internal1, movdf_internal1a, movdf_internal2):
3287 Don't allow arbitrary constants; fix predicates and C constraint.
3289 2002-09-20 Neil Booth <neil@daikokuya.co.uk>
3291 * cppmacro.c: Don't warn about function-like macros without
3292 '(' during pre-expansion.
3294 2002-09-20 Jim Wilson <wilson@redhat.com>
3296 * config/v850/v850.c (current_function_anonymous_args): Delete.
3297 (expand_prologue): Use current_function_args_info.anonymous_args.
3298 (expand_epilogue): Delete use of current_function_anonymous_args.
3299 * config/v850/v850.h (struct cum_arg): Add anonymous_args field.
3300 (INIT_CUMULATIVE_ARGS): Clear anonymous_args field.
3301 (current_function_anonymous_args): Delete extern declaration.
3302 (SETUP_INCOMING_VARARGS): Set anonymous_args field.
3304 2002-09-20 Geoffrey Keating <geoffk@apple.com>
3306 * config/rs6000/rs6000.c (rs6000_emit_prologue): Update for change
3307 to load_macho_picbase.
3308 * config/rs6000/rs6000.md: Document Darwin-specific unspec IDs.
3309 (load_macho_picbase): Take the symbol to use as a parameter.
3310 (macho_correct_pic): New insn.
3311 (builtin_setjmp_reciever): On Darwin, restore the PIC register.
3313 * config/rs6000/rs6000.h (ELIMINABLE_REGS): Use
3314 RS6000_PIC_OFFSET_TABLE_REGNUM rather than hardcoding 30.
3315 (CAN_ELIMINATE): Likewise.
3316 (INITIAL_ELIMINATION_OFFSET): Likewise.
3317 (TOC_REGISTER): Likewise.
3319 2002-09-20 Richard Henderson <rth@redhat.com>
3321 * real.c (real_hash): New.
3322 * real.h: Declare it.
3323 * cse.c (canon_hash): Use it.
3324 * cselib.c (hash_rtx): Likewise.
3325 * emit-rtl.c (const_double_htab_hash): Likewise.
3326 * rtl.h (CONST_DOUBLE_REAL_VALUE): New.
3327 * varasm.c (struct rtx_const): Reduce vector size; separate
3328 integer and fp vectors.
3330 (const_hash_1): Rename from const_hash. Use real_hash. Do not
3331 take modulus MAX_HASH_TABLE.
3332 (const_hash): New. Do take modulus MAX_HASH_TABLE.
3333 (output_constant_def): Do not take modulus MAX_HASH_TABLE.
3334 (SYMHASH): Don't use HASHBITS.
3335 (decode_rtx_const): Copy only active bits from REAL_VALUE_TYPE.
3336 Fix CONST_VECTOR thinko wrt fp vectors. Fix kind comparison.
3337 (simplify_subtraction): Fix kind comparison.
3338 (const_hash_rtx): Return unsigned int. Don't use HASHBITS.
3339 Use a union to pun integer array.
3340 * config/rs6000/rs6000.c (rs6000_hash_constant): Use real_hash;
3341 only hash two words of integral CONST_DOUBLE.
3343 2002-09-20 Steve Ellcey <sje@cup.hp.com>
3345 * config/ia64/hpux.h (STARTFILE_SPEC): Modify.
3346 (STARTFILE_PREFIX_SPEC): New.
3347 (LINK_SPEC): Modify.
3351 2002-09-20 Jakub Jelinek <jakub@redhat.com>
3353 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
3354 UNSPEC_NTPOFF and UNSPEC_DTPOFF to be offsetted by constant.
3356 2002-09-20 Jeroen Dobbelaere <jeroen.dobbelaere@acunia.com>
3358 * config/arm/arm.md (sign_extract_onebit, not_signextract_onebit):
3359 Add clobber of the condition code register.
3361 2002-09-20 Richard Henderson <rth@redhat.com>
3363 * real.c (do_fix_trunc): Static.
3364 (encode_ieee_single, encode_ieee_double, encode_ieee_extended,
3365 encode_ieee_quad, encode_vax_f, encode_vax_d, encode_vax_g,
3366 encode_i370_single, encode_i370_double, encode_c4x_single,
3367 encode_c4x_extended): Add default abort case.
3369 2002-09-20 Richard Henderson <rth@redhat.com>
3371 * real.h (enum real_value_class, SIGNIFICAND_BITS, EXP_BITS,
3372 MAX_EXP, SIGSZ, SIG_MSB, struct real_value): Move from real.c.
3373 (struct realvaluetype): Remove.
3374 (REAL_VALUE_TYPE): Use struct real_value.
3375 (REAL_VALUE_TYPE_SIZE): Use SIGNIFICAND_BITS.
3376 (test_real_width): New.
3377 * real.c: Global replace struct real_value with REAL_VALUE_TYPE.
3378 (real_arithmetic): Avoid hoops for REAL_VALUE_TYPE parameters.
3379 (real_compare, real_exponent, real_ldexp, real_isinf, real_isnan,
3380 real_isneg, real_isnegzero, real_identical, exact_real_inverse,
3381 real_to_integer, real_to_integer2, real_to_decimal,
3382 real_to_hexadecimal, real_from_string, real_from_integer,
3383 real_inf, real_nan, real_2expN, real_convert, real_to_target,
3384 real_from_target): Likewise.
3385 * tree.h (struct tree_real_cst): Use real_value not realvaluetype.
3386 * gengtype-yacc.y (bitfieldopt): Accept an ID as well.
3388 2002-09-20 Richard Henderson <rth@redhat.com>
3390 * real.h (UNKNOWN_FLOAT_FORMAT, IEEE_FLOAT_FORMAT, VAX_FLOAT_FORMAT,
3391 IBM_FLOAT_FORMAT, C4X_FLOAT_FORMAT, TARGET_FLOAT_FORMAT): Move ...
3392 * defaults.h: ... here.
3393 * config/arm/arm.h, config/avr/avr.h, config/d30v/d30v.h,
3394 config/fr30/fr30.h, config/frv/frv.h, config/ia64/ia64.h,
3395 config/ip2k/ip2k.h, config/mips/mips.h, config/stormy16/stormy16.h,
3396 config/xtensa/xtensa.h (TARGET_FLOAT_FORMAT): Remove.
3398 2002-09-20 Hans-Peter Nilsson <hp@bitrange.com>
3400 * config/mmix/mmix.md ("negdf2"): Rewrite.
3401 ("*expanded_negdf2"): New.
3403 2002-09-19 Jim Wilson <wilson@redhat.com>
3405 * combine.c (simplify_set): When optimizing a subreg src with a
3406 cc0 dest, use GET_MODE (src) for mask instead of inner_mode.
3408 2002-09-19 Dale Johannesen <dalej@apple.com>
3409 * combine.c (make_extraction): Don't create
3412 2002-09-19 Ulrich Weigand <uweigand@de.ibm.com>
3414 * config/s390/s390.c (addr_generation_dependency_p): Handle SUBREG
3415 and STRICT_LOW_PART within SET_DEST.
3416 * config/s390/s390.md ("*extractqi", "*extracthi"): New insns with
3417 splitters, replacing pre-reload splitters.
3418 ("*zero_extendhisi2_31", "*zero_extendqisi2_31",
3419 "*zero_extendqihi2_31"): New insns.
3420 ("*zero_extendqihi2_64"): Do not clobber CC.
3422 2002-09-19 Steve Ellcey <sje@cup.hp.com>
3424 * config/ia64/hpux.h (CTORS_SECTION_ASM_OP): New.
3425 (DTORS_SECTION_ASM_OP): Ditto.
3426 (READONLY_DATA_SECTION_ASM_OP): Moved.
3427 (DATA_SECTION_ASM_OP): New.
3428 (SDATA_SECTION_ASM_OP): New.
3429 (BSS_SECTION_ASM_OP): New.
3430 (SBSS_SECTION_ASM_OP): New.
3431 (TEXT_SECTION_ASM_OP): New.
3433 2002-09-19 Kazu Hirata <kazu@cs.umass.edu>
3435 * config/fp-bit.c: Follow spelling conventions.
3436 * config/d30v/d30v.c: Likewise.
3437 * config/d30v/d30v.h: Likewise.
3438 * config/fr30/fr30.c: Likewise.
3439 * config/fr30/fr30.h: Likewise.
3440 * config/fr30/fr30.md: Likewise.
3441 * config/frv/frv.c: Likewise.
3442 * config/frv/frv.h: Likewise.
3443 * config/h8300/h8300.c: Likewise.
3444 * config/h8300/lib1funcs.asm: Likewise.
3445 * config/i370/i370.c: Likewise.
3446 * config/i386/i386.h: Likewise.
3447 * config/i386/i386.md: Likewise.
3448 * config/i386/pentium.md: Likewise.
3449 * config/i386/winnt.c: Likewise.
3450 * config/i960/i960.c: Likewise.
3451 * config/ia64/ia64.h: Likewise.
3452 * config/ip2k/ip2k.c: Likewise.
3453 * config/ip2k/ip2k.h: Likewise.
3454 * config/ip2k/ip2k.md: Likewise.
3455 * config/ip2k/libgcc.S: Likewise.
3457 2002-09-19 Stephen Clarke <stephen.clarke@superh.com>
3459 * config/sh/sh.h (UNSPEC_GOTOFF_P): Define.
3460 (GOTOFF_P): Extend to allow gotoff plus constant.
3462 2002-09-18 Richard Henderson <rth@redhat.com>
3464 * ifcvt.c (noce_process_if_block): Correctly detect X modified
3465 with INSN_B before COND_EARLIEST. Don't check A and B for
3466 modification in condition range. Reorder INSN_B for A==B properly.
3467 (if_convert): Iterate until no matches for a block.
3469 2002-09-18 Richard Henderson <rth@redhat.com>
3471 * calls.c (store_one_arg): Rename default_align to parm_align;
3472 always adjust parm_align for downward padding.
3474 2002-09-18 Richard Henderson <rth@redhat.com>
3476 * toplev.c (backend_init): Move init_real_once invocation ...
3477 (do_compile): ... here.
3479 2002-09-18 Richard Henderson <rth@redhat.com>
3481 * sibcall.c (optimize_sibling_and_tail_recursive_call): Also remove
3482 RTX_UNCHANGING_P markers for successful tail-recursive replacement.
3484 2002-09-18 Richard Henderson <rth@redhat.com>
3486 * real.c (round_for_format): Collect sticky as unsigned long, not bool.
3488 2002-09-19 Alan Modra <amodra@bigpond.net.au>
3490 * config/rs6000/rs6000.md: (floatdisf2): Rename to
3491 floatdisf2_internal1.
3492 (floatdisf2): New define_expand.
3493 (floatdisf2_internal2): Likewise.
3495 2002-09-18 Richard Henderson <rth@redhat.com>
3497 * real.c (sticky_rshift_significand): Collect sticky as
3498 unsigned long, not bool.
3500 2002-09-18 Ulrich Weigand <uweigand@de.ibm.com>
3502 * config/s390/s390.c (s390_address_cost): New function.
3503 config/s390/s390-protos.h (s390_address_cost): Add prototype.
3504 config/s390/s390.h (ADDRESS_COST): Call s390_address_cost.
3505 (RTX_COST): Use COSTS_N_INSNS.
OSDN Git Service
