1 Wed Sep 9 15:16:58 1998 Gavin Romig-Koch <gavin@cygnus.com>
3 * mips/abi64.h (LONG_MAX_SPEC): Don't set LONG_MAX for
6 Wed Sep 9 12:31:35 1998 Jeffrey A Law (law@cygnus.com)
8 * Makefile.in (profile.o): Depend on insn-config.h
10 Wed Sep 9 09:36:51 1998 Jim Wilson <wilson@cygnus.com>
12 * iris6.h (DWARF2_UNWIND_INFO): Undef.
14 Wed Sep 9 01:32:01 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
16 Add preliminary native sparcv9 Solaris support.
17 * configure.in: Recognize sparv9-*-solaris2*
19 * config.sub: Recognize sparcv9 just like sparc64.
20 * config/sparc/sol2-c1.asm config/sparc/sol2-ci.asm
21 config/sparc/sol2-cn.asm: Macroize so it can be shared between
22 32-bit and 64-bit Solaris systems.
23 * config/sparc/t-sol2: Assemble those with cpp.
24 * config/sparc/sparc.h (TARGET_CPU_sparcv9): New alias for v9.
25 (*TF*_LIBCALL): If ARCH64 use V9 names.
26 * config/sparc/{xm-sysv4-64,sol2-sld-64}.h: New files.
28 Wed Sep 9 01:07:30 1998 Jakub Jelinek <jj@sunsite.ms.mff.cuni.cz>
30 * config/sparc/sparc.h (TARGET_CM_MEDMID): Fix documentation.
31 (CASE_VECTOR_MODE): Set to SImode even if PTR64, when MEDLOW and
33 (ASM_OUTPUT_ADDR_{VEC,DIFF}_ELT): Check CASE_VECTOR_MODE not
35 * config/sparc/sparc.md (tablejump): Likewise, and sign extend op0
36 to Pmode if CASE_VECTOR_MODE is something else.
38 Wed Sep 9 00:10:31 1998 Jeffrey A Law (law@cygnus.com)
40 * prefix.c (update_path): Correctly handle cases where PATH is
41 a substring of the builtin prefix, but specifies a different
44 Tue Sep 8 23:46:04 1998 Hans-Peter Nilsson <hp@axis.se>
46 * expr.c: Corrected comment about what MOVE_RATIO does.
47 * config/alpha/alpha.h: Ditto.
48 * config/1750a/1750a.h: Ditto.
49 * config/clipper/clipper.h: Ditto.
50 * config/i386/i386.h: Ditto.
52 Tue Sep 8 22:56:12 1998 Jeffrey A Law (law@cygnus.com)
54 * configure.in (m68k-next-nextstep3*): Use collect2.
55 Similarly for x86 NeXT configurations.
58 Tue Sep 8 01:38:57 1998 Nathan Sidwell <nathan@acm.org>
60 * configure.in: Don't assume srcdir is .../gcc
63 Sat Sep 5 16:34:34 EDT 1998 John Wehle (john@feith.com)
65 * global.c: Update comments.
66 (global_alloc): Assign allocation-numbers
67 even for registers allocated by local_alloc in case
68 they are later spilled and retry_global_alloc is called.
69 (mark_reg_store, mark_reg_clobber,
70 mark_reg_conflicts, mark_reg_death): Always record a
71 conflict with a pseudo register even if it has been
72 assigned to a hard register.
73 (dump_conflicts): Don't list pseudo registers already assigned to
74 a hard register as needing to be allocated, but do list their
76 * local-alloc.c: Update comment.
78 Mon Sep 7 23:38:01 1998 Jeffrey A Law (law@cygnus.com)
80 * configure.in: Check for bogus GCC_EXEC_PREFIX and LIBRARY_PATH.
83 Mon Sep 7 22:41:46 1998 Michael Meissner <meissner@cygnus.com>
85 * rs6000.c (rs6000_override_options): Fix name for ec603e, to add
87 * t-ppccomm (MULTILIB_MATCHES_FLOAT): Add support for -mcpu=xxx
88 for all targets that set -msoft-float.
90 Mon Sep 7 23:30:07 1998 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
92 * toplev.c (print_switch_values): Make static to match prototype.
94 Mon Sep 7 19:13:59 1998 Jeffrey A Law (law@cygnus.com)
96 * configure.in: If we are unable to find the "gnatbind" program,
97 then do not configure the ada subdir.
100 Sun Sep 6 14:03:58 PDT 1998 Jeff Law (law@cygnus.com)
102 * version.c: Bump for snapshot.
104 Sun Sep 6 13:28:07 PDT 1998 Jeff Law (law@cygnus.com)
106 * version.c: Bump for snapshot.
108 Sun Sep 6 08:54:14 1998 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
110 * Makefile.in (toplev.o): Depend on $(EXPR_H).
111 (insn-extract.o, insn-attrtab.o): Depend on toplev.h.
113 * gansidecl.h: Define ATTRIBUTE_NORETURN.
115 * genattrtab.c: Have insn-attrtab.c include toplev.h.
117 * genextract.c: Have insn-extract.c include toplev.h.
119 * rtl.h: Don't prototype `fatal_insn_not_found' and `fatal_insn'.
121 * toplev.c: Include expr.h.
122 (really_sorry, fancy_abort): Remove prototypes.
123 (set_target_switch): Add argument in prototype.
124 (vfatal): Mark prototype with ATTRIBUTE_NORETURN.
125 (v_really_sorry): Likewise.
126 (print_version, print_single_switch, print_switch_values): Make
127 static and add prototype arguments.
128 (decl_printable_name): Add prototype arguments.
129 (lang_expand_expr_t): New typedef.
130 (lang_expand_expr): Declare as a lang_expand_expr_t.
131 (incomplete_decl_finalize_hook): Add prototype argument.
132 (decl_name): Mark variable `verbosity' with ATTRIBUTE_UNUSED.
133 (botch): Likewise for variable `s'.
134 (rest_of_type_compilation): Mark variables `type' and `toplev'
135 with ATTRIBUTE_UNUSED if none of DBX_DEBUGGING_INFO,
136 XCOFF_DEBUGGING_INFO or SDB_DEBUGGING_INFO are defined.
137 (display_help): Make variable `i' an `unsigned long'.
138 (main): Remove unused parameter `envp'.
139 Cast assignment to `lang_expand_expr' to a `lang_expand_expr_t'.
140 Cast -1 when comparing it with a `size_t'.
142 * toplev.h (fatal, fatal_io_error, pfatal_with_name): Mark
143 prototype with ATTRIBUTE_NORETURN.
144 (fatal_insn_not_found, fatal_insn, really_sorry,
145 push_float_handler, pop_float_handler): Add prototypes.
146 (fancy_abort): Mark prototype with ATTRIBUTE_NORETURN.
147 (do_abort, botch): Add prototypes.
149 Sat Sep 6 12:05:18 1998 John Carr <jfc@mit.edu>
151 * final.c (final): If a label is reached only from a single jump,
152 call NOTICE_UPDATE_CC on the jump and its predecessor before
153 emitting the insn after the label.
155 * i386.h: Add AMD K6 support.
156 Change TARGET_* macros to use table lookup.
157 (INITIALIZE_TRAMPOLINE): Improve trampoline code.
158 (ADJUST_COST): Change definition to call function in i386.c.
159 (ISSUE_RATE): Define as 2 for anything newer than an 80486.
160 * i386.c: Add AMD K6 support.
161 Add constants for feature tests used by TARGET_* macros.
162 (split_di): If before reload, call gen_lowpart and gen_highpart.
163 (x86_adjust_cost): New function.
164 (put_jump_code): New function.
165 (print_operand): New codes 'D' and 'd'.
166 * i386.md: New insn types. New insn attribute "memory".
167 Redefine scheduling parameters to use new types and add AMD K6
168 support. Explicitly set type of most insns.
169 (move insns): K6 prefers movl $0,reg to xorl reg,reg. Pentium
170 Pro and K6 prefer movl $1,reg to incl reg.
171 (adddi3, subdi3): Set cc_status.
172 (DImode shift patterns): Change label counters from HOST_WIDE_INT
173 to int; x86 can't have more than 2^31 DImode shifts per file.
174 (setcc): Combine all setcc patterns. Allow writing memory.
175 Combine all jump patterns using match_operator.
176 (*bzero): Name pattern. Emit mutliple stos instructions when that
177 is faster than rep stos.
178 (xordi3, anddi3, iordi3): Simplify DImode logical patterns and
181 Sun Sep 6 11:17:20 1998 Dave Love <d.love@dl.ac.uk>
183 * config/m68k/x-next (BOOT_LDFLAGS): Define suitably for f771
186 Sat Sep 5 22:05:25 1998 Richard Henderson <rth@cygnus.com>
188 * alpha.c (alpha_ra_ever_killed): Inspect the topmost sequence,
189 not whatever we're generating now.
191 * alpha.c (set_frame_related_p, FRP): New.
192 (alpha_expand_prologue): Mark frame related insns.
193 (alpha_expand_epilogue): Likewise, but with a null FRP.
194 * alpha.h (INCOMING_RETURN_ADDR_RTX): New.
195 * alpha.md (exception_receiver): New.
196 * alpha/crtbegin.asm (.eh_frame): New beginning.
197 (__do_frame_setup, __do_frame_takedown): New.
198 * alpha/crtend.asm (.eh_frame): New ending.
199 * alpha/elf.h (DWARF2_DEBUGGING_INFO): Define.
200 (ASM_SPEC): Don't emit both dwarf2 and mdebug.
201 (ASM_FILE_START): Don't emit .file for dwarf2.
203 * rtl.h (enum reg_note): Add REG_FRAME_RELATED_EXPR.
204 * rtl.c (reg_note_name): Likewise.
205 * rtl.texi (REG_NOTES): Likewise.
206 * dwarf2out.c (dwarf2out_frame_debug): Use it. Recognize a store
209 Sat Sep 5 14:47:17 1998 Richard Henderson <rth@cygnus.com>
211 * i386.h (PREFERRED_RELOAD_CLASS): Standard fp constants load to TOS.
212 * i386.md (movsf, movdf, movxf): Validate memory address returned
213 from force_const_mem. Kill useless REG_EQUAL setting code.
215 Sat Sep 5 14:23:31 1998 Torbjorn Granlund <tege@matematik.su.se>
217 * m68k.md (zero_extendsidi2): Fix typo.
219 Sat Sep 5 13:40:24 1998 Krister Walfridsson <cato@df.lth.se>
221 * configure.in: Removed references to the removed file
222 * config/xm-netbsd.h. Use ${cpu_type}/xm-netbsd.h for
223 * arm*-*-netbsd* and ns32k-*-netbsd*.
224 * config/i386/xm-netbsd.h: Removed unnecessary file.
225 * config/m68k/xm-netbsd.h: Likewise.
226 * config/sparc/xm-netbsd.h: Likewise.
227 * config/mips/xm-netbsd.h: Likewise.
229 Sat Aug 29 13:32:58 1998 Mumit Khan <khan@xraylith.wisc.edu>
231 * i386/cygwin32.h (BIGGEST_ALIGNMENT): Define.
232 (PCC_BITFIELD_TYPE_MATTERS): Define to be 0.
234 * i386/cygwin32.h (ASM_OUTPUT_SECTION_NAME): Don't check for
235 for exact section attributions.
237 * i386/mingw32.h (CPP_PREDEFINES): Add __MSVCRT__ for msvc
239 * i386/crtdll.h (CPP_PREDEFINES): Define.
241 Sat Sep 5 03:23:05 1998 Jeffrey A Law (law@cygnus.com)
243 * m68k.md (5200 movqi): Do not allow byte sized memory references
245 * m68k.c (output_move_qimode): Do not use byte sized operations on
248 * Makefile.in (pexecute.o): Use pexecute.c from libiberty. Provide
249 explicit rules for building. Similarly for alloca, vfprintf,
250 choose-temp and mkstemp, getopt, getopt1, and obstack.
251 (INCLUDES): Add $(srcdir)/../include.
252 * pexecute.c, alloca.c, vfprintf.c, choose-temp.c, mkstemp.c: Delete.
253 * getopt.h, getopt.c getopt1.c, obstack.c, obstack.h: Likewise.
255 Fri Sep 4 11:57:50 1998 Tom Tromey <tromey@cygnus.com>
257 * gcc.c (do_spec_1): [case 'o'] Account for
258 lang_specific_extra_outfiles.
259 (main): Correctly clear all slots in outfiles for
260 lang_specific_extra_outfiles. Set input_file_number before
261 calling lang_specific_pre_link.
263 Fri Sep 4 10:37:07 1998 Jim Wilson <wilson@cygnus.com>
265 * loop.c (load_mems): Fix JUMP_LABEL field after for_each_rtx call.
267 Fri Sep 4 02:01:05 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
269 * config/sparc/sparc.c (output_double_int): In all V9 symbolic
271 (sparc_output_deferred_case_vectors): If no work to do, return.
272 Fix thinko in Sept 1 change.
274 1998-09-03 SL Baur <steve@altair.xemacs.org>
276 * Makefile.in: add semicolon in BISON definition for portability.
278 Thu Sep 3 13:34:41 1998 Toon Moene <toon@moene.indiv.nluug.nl>
280 * config/nextstep.c (handle_pragma): Correct name of third
283 Tue Sep 1 11:30:33 1998 Nick Clifton <nickc@cygnus.com>
285 * config/m32r/m32r.md: Change (reg:CC 17) to (reg:SI 17).
286 * config/m32r/m32r.h: Make register 17 be fixed.
287 * config/m32r/m32r.c: Use SImode for cc operations.
289 Thu Sep 3 18:17:34 1998 Benjamin Kosnik <bkoz@cygnus.com>
291 * invoke.texi (Warning Options): Add -Wnon-template-friend
294 Thu Sep 3 18:16:16 1998 Michael Meissner <meissner@cygnus.com>
296 * rs6000.c (rs6000_override_options): Add -mcpu={401,e603e}.
298 Thu Sep 3 18:05:16 1998 David Edelsohn <edelsohn@mhpcc.edu>
300 * rs6000.md (movsf): Disable explicit secondary-reload-like
301 functionality if TARGET_POWERPC64.
302 (movdf): Remove TARGET_POWERPC64 explicit secondary-reload-like
306 Thu Sep 3 11:41:40 1998 Robert Lipe <robertl@dgii.com>
308 * fixinc.sco: Borrow code to wrap 'bool' typedefs from tinfo.h
309 and term.h from fixinc.wrap.
311 Thu Sep 3 09:47:31 1998 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
313 * aclocal.m4 (GCC_HEADER_STRING): New macro to detect if it is
314 safe to include both string.h and strings.h together.
315 (GCC_NEED_DECLARATION): Test STRING_WITH_STRINGS when deciding
316 which headers to search for function declarations. Continue to
317 prefer string.h over strings.h when both are not acceptable.
319 * acconfig.h (STRING_WITH_STRINGS): Add stub.
321 * configure.in: Call GCC_HEADER_STRING.
323 * system.h: Test STRING_WITH_STRINGS when deciding which headers
324 to include. Continue to prefer string.h over strings.h when both
327 Wed Sep 2 23:56:29 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
329 * config/sparc/sparc.c (output_double_int): If V9 and MEDLOW, do
330 not assume top 32-bits of symbolic addresses are zero if
333 Thu Sep 3 00:23:21 1998 Richard Henderson <rth@cygnus.com>
335 * ginclude/va-alpha.h: Protect entire second portion of the
336 file against double inclusion.
338 Thu Sep 3 00:37:55 1998 Ovidiu Predescu <ovidiu@aracnet.com>
340 Added support for the Boehm's garbage collector.
341 * configure.in: Handle --enable-objc-gc.
342 * configure: Rebuilt.
343 * Makefile.in (CHECK_TARGETS): Add check-objc.
344 (check-objc): New rule.
345 * objc/Make-lang.in: Build a different Objective-C library that
346 runs with the Boehm's collector.
347 * objc/encoding.c (objc_round_acc_size_for_types): New function.
348 * objc/encoding.c: Correctly compute the size of compound types in
349 the presence of bitfields. Skip the variable name of the type if
350 any. Added support for long long.
351 * objc/encoding.h (_C_GCINVISIBLE): New specifier.
352 (_F_GCINVISIBLE): New mask.
353 * objc/gc.c: New file. Compute the type memory mask associated with
354 a class based on the runtime information.
355 * objc/misc.c: Added the hooks that use the Boehm's collector
356 allocation functions.
357 * objc/objc-act.c (build_class_template): Generate a new class
358 member (gc_object_type) to hold the class' type memory mask.
359 (build_shared_structure_initializer): Initialize the new member to
361 (encode_complete_bitfield): New function. Generate the new
363 (encode_field_decl): Generate the new encoding only for the GNU
365 * objc/objc-api.h (_C_LNG_LNG, _C_ULNG_LNG): New specifiers for the
367 (class_get_gc_object_type): New function to mark a pointer instance
368 variable as a weak pointer.
369 * objc/objc-features.texi: New file.
370 * objc/objc.h (gc_object_type): New class member.
371 * objc/objects.c (class_create_instance): Create a typed memory
372 object when compiled with Boehm's collector support.
373 * objc/sendmsg.c (__objc_init_install_dtable): Call
374 __objc_send_initialize instead of setting the initialize flag.
375 (__objc_send_initialize): Call __objc_generate_gc_type_description
376 to generate the class type memory mask. Rewrite the code that
377 sends the +initialize so that it is called only once (bug report
378 and fix from Ronald Pijnacker <Ronald.Pijnacker@best.ms.philips.com>).
379 * testsuite/objc: New testsuite for Objective-C type encoding.
380 * testsuite/lib/objc-torture.exp: New file.
381 * testsuite/lib/objc.exp: New file.
383 Wed Sep 2 14:47:36 1998 Jim Wilson <wilson@cygnus.com>
385 * jump.c (jump_optimize): In if/then/else transformations, add
386 another call to modified_between_p for the jump insn.
388 Wed Sep 2 14:16:49 1998 Jeffrey A Law (law@cygnus.com)
390 * fix-header.c (symlink): Treat like readlink.
392 Wed Sep 2 19:30:06 1998 J"orn Rennecke <amylaar@cygnus.co.uk>
394 * dwarfout.c (fundamental_type_code): Encode 32 bit floats/doubles
397 Wed Sep 2 10:06:07 1998 Nick Clifton <nickc@cygnus.com>
399 * config/nextstep.h: Update HANDLE_PRAGMA macro.
400 * config/h8300/h8300.h: Update HANDLE_PRAGMA macro.
401 * config/i960/i960.h: Update HANDLE_PRAGMA macro.
403 * config/nextstep.c (handle_pragma): Take three arguments, as per
404 the new HANDLE_PRAGMA macro specification.
405 * config/h8300/h8300.c (handle_pragma): Take three arguments, as
406 per the new HANDLE_PRAGMA macro specification.
407 * config/i960/i960.c (process_pragma): Take three arguments, as
408 per the new HANDLE_PRAGMA macro specification.
410 Wed Sep 2 09:25:29 1998 Nick Clifton <nickc@cygnus.com>
412 * c-lex.c (check_newline): Call HANDLE_PRAGMA before
413 HANDLE_SYSV_PRAGMA if both are defined. Generate warning messages
414 if unknown pragmas are encountered.
415 (handle_sysv_pragma): Interpret return code from
416 handle_pragma_token (). Return success/failure indication rather
417 than next unprocessed character.
418 (pragma_getc): New function: retrieves characters from the
419 input stream. Defined when HANDLE_PRAGMA is enabled.
420 (pragma_ungetc): New function: replaces characters back into the
421 input stream. Defined when HANDLE_PRAGMA is enabled.
423 * c-pragma.c (handle_pragma_token): Return success/failure status
426 * c-pragma.h: Change prototype of handle_pragma_token().
428 * varasm.c: (handle_pragma_weak): Only create this function if
429 HANDLE_PRAGMA_WEAK is defined.
431 * c-common,c (decl_attributes): If defined call the expression
432 contained within the INSERT_ATTRIBUTES macro before adding
433 attributes to a decl.
435 * tm.texi (HANDLE_PRAGMA): Document the new verion of
436 HANDLE_PRAGMA, which takes three arguments.
437 (INSERT_ATTRIBUTES): Document this new macro.
439 * LANGUAGES: Document the new version of HANDLE_PRAGMA and the
440 new INSERT_ATTRIBUTES macro.
442 Wed Sep 2 02:03:23 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
444 * config/sparc/sparc.md (movdf): Only generate special RTL for
446 (move_label_di): Remove
447 (movdi_pic_label_ref, movdi_high_pic_label_ref,
448 movdi_lo_sum_pic_label_ref): New patterns for 64-bit label
450 * config/sparc/sparc.h (ASM_OUTPUT_ADDR_VEC_ELT,
451 ASM_OUTPUT_ADDR_DIFF_ELT): Don't do anything special for MEDLOW,
452 output an .xword for all 64-bit cases.
454 Tue Sep 1 15:55:17 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
456 * config/sparc/sparc.c (finalize_pic): Don't output arbitrary
457 alignment, use FUNCTION_BOUNDARY instead.
458 (sparc_output_deferred_case_vectors): Likewise.
460 Mon Aug 31 17:25:41 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
462 * config/sparc/sparc.md (movsf_const_intreg): Kill warning.
463 (movtf_insn_sp64, movtf_no_e_insn_sp64): Reorder alternatives.
465 Mon Aug 31 13:57:55 1998 Richard Henderson <rth@cygnus.com>
467 * alpha/va_list.h: New file.
468 * alpha/x-alpha (EXTRA_HEADERS): New. Add va_list.h.
470 Mon Aug 31 14:55:02 1998 Jeffrey A Law (law@cygnus.com)
472 * NEWS: Add SCO Openserver and Unixware 7 notes.
476 Mon Aug 31 15:42:18 1998 Dave Brolley <brolley@cygnus.com>
478 * varasm.c (compare_constant_1): Handle RANGE_EXPR.
479 (record_constant_1): Handle RANGE_EXPR.
481 Mon Aug 31 10:54:03 1998 Richard Henderson <rth@cygnus.com>
483 * print-rtl.c (print_rtx): NOTE_INSN_LIVE has an rtx not a bitmap.
484 * haifa-sched.c (sched_analyze): Handle NOTE_INSN_RANGE_START
485 and NOTE_INSN_RANGE_END specially.
486 (reemit_notes): Likewise.
488 Mon Aug 31 10:18:52 1998 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
490 * sparc.c (TMASK, UMASK): Use `(unsigned)1' not `1U'.
491 (ultrasparc_sched_init): Remove unneeded &.
493 Mon Aug 31 10:47:16 1998 Andreas Schwab <schwab@issan.informatik.uni-dortmund.de>
495 * config/m68k/m68k.h (TARGET_SWITCHES): Don't remove MASK_68040
496 for m68020-60, to prevent the use of fintrz.
498 Sun Aug 30 22:17:20 1998 Mark Mitchell <mark@markmitchell.com>
500 * configure.in: If the native compiler is GCC use $(WARN_CFLAGS)
502 * Makefile.in: Likewise.
503 * configure: Regenerated.
505 Sun Aug 30 22:15:41 1998 H.J. Lu (hjl@gnu.org)
507 * configure.in (gxx_include_dir): Changed to
508 '${prefix}/include/g++'-${libstdcxx_interface}.
509 * configure: Rebuilt.
511 Sun Aug 30 20:19:43 1998 Hans-Peter Nilsson <hp@axis.se>
513 * expr.c (expand_expr): Change ">" to ">=" making MOVE_RATIO use
515 * tm.texi (Costs): Say MOVE_RATIO is number of mem-mem move
516 *sequences* *below* which scalar moves will be used.
518 Sun Aug 30 17:18:43 1998 Jeffrey A Law (law@cygnus.com)
520 * collect2.c (mktemp): Delete unused declaration.
522 * config/xm-netbsd.h: Remove unnecessary file.
523 * config/*/xm-netbsd.h: Do not include the generic xm-netbsd.h
524 file anymore, it is not needed.
526 Sun Aug 30 16:05:45 1998 Mark Mitchell <mark@markmitchell.com>
528 * convert.c (convert_to_integer): Issue an error on conversions to
531 Sun Aug 30 16:47:20 1998 Martin von Lvwis <loewis@informatik.hu-berlin.de>
533 * Makefile.in: Add lang_tree_files and gencheck.h.
534 * configure.in: Generate them.
535 * gencheck.c: Include gencheck.h.
537 Sat Aug 29 21:38:24 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
539 * config/sparc/sparc.md (pic_lo_sum_di, pic_sethi_di): Rename to
540 movdi_lo_sum_pic and movdi_high_pic and make visible.
541 * config/sparc/sparc.c (legitimize_pic_address): For -fPIC,
542 emit these when Pmode is not SImode.
543 * config/sparc/linux64.h (SPARC_DEFAULT_CMODEL): Make CM_MEDLOW.
545 Sat Aug 29 14:59:32 1998 Mumit Khan <khan@xraylith.wisc.edu>
547 * i386/cygwin32.h (ASM_OUTPUT_SECTION_NAME): Don't emit
548 .linkonce directive after the first time.
550 Sat Aug 29 12:39:56 1998 Jeffrey A Law (law@cygnus.com)
552 * m68k.md (beq0_di): Generate correct (and more efficient) code when
553 the clobbered operand overlaps with an input.
554 (bne0_di): Similarly.
556 * Makefile.in (INSTALL): Remove "--no-header" argument.
558 * NEWS: Various updates.
560 Fri Aug 28 19:00:44 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
562 * config/sparc/sparc.c (arith_operand, const64_operand,
563 const64_high_operand, arith_double_4096_operand): Mark mode as
565 (create_simple_focus_bits): Remove unused arg highest_bit_set, all
567 (sparc_emit_set_const64): Remove unused variable i.
568 (sparc_splitdi_legitimate): Likewise for addr_part.
569 (ultra_code_from_mask): Likewise for mask.
570 (ultra_cmove_results_ready_p): Fixup entry modulo calc. and
571 reverse return values so it matches usage and comments.
572 (ultra_flush_pipeline): Likewise.
573 (ultra_fpmode_conflict_exists): Likewise, remove unused variable
574 this_type, and allow loads and stores of differing FP modes as
575 they do not create a conflict.
576 (ultra_find_type): Initialize fpmode to SFmode, fix
577 parenthesization thinkos in large conditional.
578 (ultrasparc_sched_init): Mark dump and sched_verbose as unused.
579 Init free_slot_mask after ultra_cur_hist is reset, not before.
580 (ultrasparc_rescan_pipeline_state): Remove unused variable ucode.
581 (ultrasparc_sched_reorder): Don't bzero current pipeline state,
582 use ultra_flush_pipeline instead, then re-init group pointer.
583 Fix statement with no effect. If no progress made in, and no
584 instructions scheduled at all, advance to new pipeline cycle else
585 we get into an endless loop.
586 (ultrasparc_adjust_cost): Remove previous arg.
587 * config/sparc/sparc.h (ADJUST_COST): Update to reflect that.
589 Fri Aug 28 13:52:35 1998 Jim Wilson <wilson@cygnus.com>
591 * sparc.md (DImode, DFmode, TFmode splits): Delete self_reference
592 code. Use reg_overlap_mentioned_p to detect when source and
594 (negtf2_notv9+1): Use DFmode instead of SFmode in last two operands.
596 1998-08-28 Brendan Kehoe <brendan@cygnus.com>
598 * loop.c (check_dbra_loop): Pass COMPARISON_VALUE, not
599 COMPARISON_VAL, into invariant_p.
601 Fri Aug 28 15:13:25 1998 J"orn Rennecke <amylaar@cygnus.co.uk>
603 * regmove.c (regclass_compatible_p): New function.
604 (regmove_optimize): Use it.
606 Use NREGS parameter insted of calling max_reg_num.
608 (fixup_match_1): Don't use code = MINUS when later tieing with
609 a hard register is likely.
611 Fri Aug 28 14:54:07 1998 J"orn Rennecke <amylaar@cygnus.co.uk>
613 * loop.c (check_dbra_loop): Fix calculation of FINAL_VALUE when
614 COMPARISON_VAL was normalized.
616 Thu Aug 27 20:10:46 1998 Jeffrey A Law (law@cygnus.com)
618 * loop.c (check_dbra_loop): The loop ending comparison value
619 must be an invariant or we can not reverse the loop.
621 * loop.c (scan_loop): Count down from max_reg_num - 1 to
622 FIRST_PSEUDO_REGISTER to avoid calling max_reg_num each iteration
624 (load_mems_and_recount_loop_regs_set): Likewise.
626 * i386.c (print_operand): Remove obsolete 'c' docs.
628 Wed Aug 26 17:13:37 1998 Tom Tromey <tromey@cygnus.com>
630 * gthr.h: Document __GTHREAD_MUTEX_INIT_FUNCTION.
631 * frame.c (init_object_mutex): New function.
632 (init_object_mutex_once): Likewise.
634 (__register_frame_info): Likewise.
635 (__register_frame_info_table): Likewise.
636 (__deregister_frame_info): Likewise.
638 Thu Aug 27 15:14:18 1998 Jeffrey A Law (law@cygnus.com)
640 * haifa-sched.c (sched_analyze_insn): Fix thinko in last change.
642 Thu Aug 27 16:34:51 1998 J"orn Rennecke <amylaar@cygnus.co.uk>
644 * loop.c (check_dbra_loop): Enable code for reversal
645 of some loops without a known constant loop end.
647 Wed Aug 26 18:38:15 1998 Richard Henderson <rth@cygnus.com>
649 * haifa-sched.c (last_clock_var): New.
650 (schedule_block): Initialize it.
651 (schedule_insn): Use it to fill insn modes with issue information.
653 * alpha.c (alpha_handle_trap_shadows): Remove do-nothing exit.
654 Tag trapb and next insn with TImode.
655 (alphaev5_insn_pipe, alphaev5_next_group, alphaev5_align_insns): New.
656 (alpha_reorg): Add conditional for alpha_handle_trap_shadows.
657 Invoke alphaev5_align_insns as appropriate.
658 * alpha.h (LABEL_ALIGN_AFTER_BARRIER): Was ALIGN_LABEL_AFTER_BARRIER.
659 (MD_SCHED_VARIABLE_ISSUE): New.
660 * alpha.md (attr type): Add multi.
661 (define_asm_attributes): New.
662 (prologue_stack_probe_loop, builtin_setjmp_receiver): Set type multi.
663 (arg_home): Likewise.
664 (fnop, unop, realign): New.
666 Wed Aug 26 15:55:41 1998 Jim Wilson <wilson@cygnus.com>
668 * iris5.h (PREFERRED_DEBUGGING_TYPE): Undef.
669 * iris5gas.h (PREFERRED_DEBUGGING_TYPE): Define.
671 * configure.in (powerpc-ibm-aix4.[12]*): Change from 4.[12].*.
672 (rs6000-ibm-aix4.[12]*): Likewise.
673 * configure: Regnerate.
675 Wed Aug 26 09:30:59 1998 Nick Clifton <nickc@cygnus.com>
677 * config/arm/thumb.c (thumb_exit): Do not move a4 into lr if it
678 already contains the return address.
680 Wed Aug 26 12:57:09 1998 Jeffrey A Law (law@cygnus.com)
682 * calls.c (expand_call): Use bitfield instructions to extract/deposit
683 word sized hunks when loading unaligned args into registers.
685 * haifa-sched.c (sched_analyze_insn): Only create scheduling
686 barriers for LOOP, EH and SETJMP notes on the loop_notes list.
688 * mn10300.h (RTX_COSTS): Handle UDIV and UMOD too.
690 Wed Aug 26 16:35:37 1998 J"orn Rennecke <amylaar@cygnus.co.uk>
692 * loop.c (check_dbra_loop): Add some code that would allow reversal
693 of some loops without a known constant loop end if it were enabled.
695 Wed Aug 26 11:08:44 1998 Gavin Romig-Koch <gavin@cygnus.com>
697 * mips.md (lshrsi3_internal2+2): Fix type-o.
699 Wed Aug 26 10:53:03 1998 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
701 * system.h: Include stdarg.h/varargs.h, make sure they are ordered
702 correctly with regards to stdio.h.
704 * calls.c: Remove stdarg.h/varargs.h.
707 * combine.c: Likewise.
708 * cpperror.c: Likewise.
709 * cpplib.c: Likewise.
710 * cpplib.h: Likewise.
711 * doprint.c: Likewise.
712 * emit-rtl.c: Likewise.
714 * fix-header.c: Likewise.
716 * genattr.c: Likewise.
717 * genattrtab.c: Likewise.
718 * gencodes.c: Likewise.
719 * genconfig.c: Likewise.
720 * genemit.c: Likewise.
721 * genextract.c: Likewise.
722 * genflags.c: Likewise.
723 * genopinit.c: Likewise.
724 * genoutput.c: Likewise.
725 * genpeep.c: Likewise.
726 * genrecog.c: Likewise.
727 * mips-tfile.c: Likewise.
728 * prefix.c: Likewise.
729 * protoize.c: Likewise.
730 * regmove.c: Likewise.
731 * toplev.c: Likewise.
734 Wed Aug 26 05:09:27 1998 Jakub Jelinek <jj@sunsite.ms.mff.cuni.cz>
736 * config/sparc/sparc.c (sparc_override_options): If not
737 TARGET_FPU, turn off TARGET_VIS.
738 * config/sparc/sparc.h (TARGET_SWITCHES): Add no-vis.
739 (LEGITIMATE_CONSTANT_P): Allow SF/DF mode zero when TARGET_VIS.
740 * config/sparc/sparc.md (movsi_insn): Use fzeros not fzero.
741 (movdi_insn_sp64): Add VIS fzero alternative.
742 (clear_sf, clear_df): New VIS patterns.
743 (movsf, movdf expanders): Allow fp_zero_operand flat out when
745 (one_cmpldi2_sp64): Provide new fnot1 VIS alternative.
747 Tue Aug 25 10:57:41 1998 Mark Mitchell <mark@markmitchell.com>
749 * loop.c (n_times_set, n_times_used, may_not_optimize,
750 reg_single_usage): Convert to varrays. All uses changed.
751 (insert_loop_mem): Return a value.
752 (scan_loop): Tweak AVOID_CC_MODE_COPIES code.
753 (load_mems_and_recount_loop_regs_set): Likewise. Grow the arrays, if
756 Tue Aug 25 23:57:12 1998 Jeffrey A Law (law@cygnus.com)
759 * configure.in: Do not set thread_file to "irix" since no such
762 * sparc.md (float abs/neg splits): Check reload_completed before
763 calling alter_subreg.
765 Tue Aug 25 19:17:59 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
767 * config/sparc/sparc.c (sparc_absnegfloat_split_legitimate): New
769 * config/sparc/sparc.h: Declare it.
770 * config/sparc/sparc.md (float abs/neg splits): Use it.
771 (all other splits): Handle SUBREGs properly where necessary.
772 (unnamed (1<<x)-1 V8PLUS pattern): Disable for now.
774 Tue Aug 25 19:48:46 1998 Jeffrey A Law (law@cygnus.com)
776 * reorg.c (fill_simple_delay_slots): Do not abort if we encounter
777 an insn on the unfilled_slots_list that has no delay slots.
778 (fill_eager_delay_slots): Similarly.
780 Tue Aug 25 13:35:20 1998 Nick Clifton <nickc@cygnus.com>
782 * config/v850/v850.c (movsi_source_operand): Treat CONSTANT_P_RTX
783 as an ordinary operand.
785 Tue Aug 25 12:54:57 1998 Jason Merrill <jason@yorick.cygnus.com>
787 * tree.c (valid_machine_attribute): Don't apply attributes to both
790 Tue Aug 25 12:23:20 PDT 1998 Richard Henderson <rth@cygnus.com>
792 * reload.c (operands_match_p): Handle rtvecs.
794 * i386.c (legitimate_pic_address_disp_p): New.
795 (legitimate_address_p): Use it.
796 (legitimize_pic_address): Use unspecs to represent @GOT and @GOTOFF.
797 Handle constant pool symbols just like statics.
798 (emit_pic_move): Use Pmode not SImode for clarity.
799 (output_pic_addr_const) [SYMBOL_REF]: Remove @GOT and @GOTOFF hacks.
800 [UNSPEC]: New, handling what we killed above.
801 [PLUS]: Detect and abort on invalid symbol arithmetic.
802 * i386.h (CONSTANT_ADDRESS_P): Remove HIGH.
804 Tue Aug 25 12:02:23 1998 Mark Mitchell <mark@markmitchell.com>
806 * alias.c: Include output.h.
807 (DIFFERENT_ALIAS_SETS_P): Don't treat alias sets as
808 different if we're in a varargs function.
809 * Makefile.in (alias.o): Depend on output.h
811 Tue Aug 25 19:20:12 1998 J"orn Rennecke <amylaar@cygnus.co.uk>
813 * sh.h (GIV_SORT_CRITERION): Delete.
815 Tue Aug 25 13:19:46 1998 Dave Brolley <brolley@cygnus.com>
817 * regclass.c (regclass): Use xmalloc/free instead of alloca.
818 * stupid.c (stupid_life_analysis): Ditto.
819 * reload1.c (reload): Ditto.
821 Tue Aug 25 05:48:18 1998 Jakub Jelinek <jj@sunsite.ms.mff.cuni.cz>
823 * config/sparc/sparc.c (arith_4096_operand, arith_add_operand,
824 arith_double_4096_operand, arith_double_add_operand): New
826 * config/sparc/sparc.h (PREDICATE_CODES): Add them, declare them.
827 * config/sparc/sparc.md (adddi3, addsi3, subdi3, subsi3): Use
828 them to transform add/sub 4096 into add/sub -4096.
830 Mon Aug 24 23:31:03 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
832 * loop.c (scan_loop): Allocate some slop to handle pseudos
833 generated by move_movables.
834 (load_mems_and_recount_loop_regs_set): Honor AVOID_CC_MODE_COPIES
837 Mon Aug 24 19:45:40 1998 Jim Wilson <wilson@cygnus.com>
839 * tree.def (DECL_RESULT): Correct documentation.
841 Tue Aug 25 01:15:27 1998 J"orn Rennecke <amylaar@cygnus.co.uk>
843 * reload1.c (reload_reg_free_before_p): New argument EQUIV; Changed
844 all callers. Abort for RELOAD_FOR_INSN. RELOAD_FOR_OUTADDR_ADDR:
845 conflicts will all RELOAD_FOR_OUTPUT reloads.
847 * reload1.c (reload_cse_regs_1): When deleting a no-op move that
848 loads the function result, substitute with a USE.
850 Mon Aug 24 15:20:19 1998 David Edelsohn <edelsohn@mhpcc.edu>
852 * rs6000.h (GO_IF_LEGITIMATE_ADDRESS): Use TARGET_POWERPC64
853 when testing LEGITIMATE_INDEXED_ADDRESS_P DFmode and DImode.
854 (LEGITIMIZE_ADDRESS): Use TARGET_POWERPC64 for INDEXED fixup.
855 * rs6000.c (print_operand, case 'L'): Add UNITS_PER_WORD, not 4.
856 (print_operand, cases 'O' and 'T'): Fix typos in lossage strings.
857 * rs6000.md (fix_truncdfsi2_store): Remove %w from non-CONST_INT
859 (movdf_softfloat32, movdf_hardfloat64, movdf_softfloat64): Change
860 'o' to 'm' for GPR variant constraints.
862 Mon Aug 24 10:25:46 1998 Jeffrey A Law (law@cygnus.com)
864 * loop.c (scan_loop): Honor AVOID_CC_MODE_COPIES.
866 * h8300.h (STRIP_NAME_ENCODING): Fix typo.
868 * sparc.md (TFmode splits): Use reg_overlap_mentioned_p to detect
869 when the source and destination overlap.
871 * stmt.c (emit_case_nodes): Change rtx_function to rtx_fn to avoid
872 clash with global type.
874 Mon Aug 24 00:53:53 1998 Jason Merrill <jason@yorick.cygnus.com>
876 * fixinc.irix: Add curses.h handling from fixinc.wrap.
878 * c-common.c (combine_strings): Also set TREE_READONLY.
879 Change warn_write_strings to flag_const_strings.
880 * c-decl.c, c-tree.h: Likewise.
882 Sun Aug 23 18:39:11 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
884 * config/sparc/sparc.c (sparc_emit_set_const32): If outputting a
885 CONST_INT, not a symbolic reference, don't use a HIGH/LO_SUM
886 sequence, use SET/IOR instead so CSE can see it.
887 * config/sparc/sparc.md (movhi_const64_special,
888 movsi_const64_special): New patterns necessitated by that change.
889 (movhi_high): Remove.
890 (movhi_lo_sum): Change to match an IOR.
891 (movdf_insn_sp32): Test TARGET_V9 not TARGET_ARCH64.
892 (movdf_insn_v9only): New pattern for when V9 but not ARCH64.
893 (movdf_insn_sp64): Test both TARGET_V9 and TARGET_ARCH64.
894 (movdf splits): Allow when not V9 or when not ARCH64 and integer
895 registers are involved.
896 (snesi_zero_extend split): Remove reload_completed test.
897 (unnamed plus and minus zero_extend sidi splits): Add it.
899 Sun Aug 23 11:56:08 1998 Mark Mitchell <mark@markmitchell.com>
901 * extend.texi: Remove description of extension to explicit
902 instantiation that is now endorsed by standard C++.
904 Sun Aug 23 09:39:09 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
906 * config/arc/arc.c (arc_initialize_pic): Remove.
907 * config/arc/arc.h (INITIALIZE_PIC): Similarly, this routine does
908 nothing on any platform and is invoked by no-one, it does not even
909 appear in the documentation.
910 * config/sparc/sparc.h (INITIALIZE_PIC): Likewise.
911 * config/sparc/sparc.c (initialize_pic): Likewise.
912 (find_addr_reg): Remove this as well, no longer referenced after
915 Sun Aug 23 00:17:14 1998 Jeffrey A Law (law@cygnus.com)
917 * recog.c (validate_replace_rtx_group): New function.
918 * recog.h (validate_replace_rtx_group): Declare it.
919 * regmove.c (optimize_reg_copy_3): If any substitution fails, then undo
920 the entire group of substitutions.
922 Sat Aug 22 23:31:00 1998 Klaus-Georg Adams (Klaus-Georg.Adams@chemie.uni-karlsruhe.de)
924 * loop.c (load_mems): Fix initializers.
926 Fri Aug 21 23:07:46 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
928 * config/sparc/sparc.md (TFmode splits): Handle destination
929 registers being referenced in the address correctly.
931 * expmed.c (make_tree) [CONST_INT]: Sign extend even if
932 TREE_UNSIGNED, when bitsize of type's mode is larger than
933 HOST_BITS_PER_WIDE_INT.
935 Fri Aug 21 19:31:31 1998 Alexandre Petit-Bianco <apbianco@cygnus.com>
937 * tree.def (LABELED_BLOCK_EXPR, EXIT_BLOCK_EXPR): New tree nodes.
938 * tree.h (LABELED_BLOCK_LABEL, LABELED_BLOCK_BODY,
939 EXIT_BLOCK_LABELED_BLOCK, EXIT_BLOCK_RETURN, LOOP_EXPR_BODY): New
941 * expr.c (expand_expr): Handle LABELED_BLOCK_EXPR and
944 Thu Aug 20 19:43:44 1998 Jeffrey A Law (law@cygnus.com)
946 * h8300.c (h8300_encode_label): Use '&' for tiny data items.
947 * h8300.h (TINY_DATA_NAME_P): Likewise.
948 (STRIP_NAME_ENCODING): Handle '&'.
950 * mn10200.h (REG_OK_FOR_INDEX_P): Do not check the mode of the
951 register (it could be accessed via an outer SUBREG).
952 (REG_OK_FOR_BASE_P): Likewwise.
953 (GO_IF_LEGITIMATE_ADDRESS): Consistently use REGNO_OK_FOR_BASE_P.
955 * remove.c (optimize_reg_copy_3): Abort instead of silently generating
958 * jump.c (rtx_renumbered_equal_p): Do not consider PLUS commutative.
960 Thu Aug 20 17:35:20 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
962 * config/sparc/sparc.md (movtf_insn_sp32): All memory operands
963 must be offsettable so the splits can be made.
965 Thu Aug 20 13:56:53 1998 Michael Meissner <meissner@cygnus.com>
967 * config/i386/winnt.c: Include system.h, not stdio.h to get
968 sys/param.h pulled in before rtl.h in case the system defines MIN
971 Thu Aug 20 13:44:20 1998 David Edelsohn <edelsohn@mhpcc.edu>
973 * rs6000.md (movqi, movhi): Add CONSTANT_P_RTX.
975 Thu Aug 20 13:15:11 1998 Dave Brolley <brolley@cygnus.com>
977 * stor-layout.c (layout_type): Compute TYPE_SIZE_UNIT correctly for
979 * cpplib.c (cpp_define): Handle macros with parameters.
981 Wed Aug 19 21:33:19 1998 David Edelsohn <edelsohn@mhpcc.edu>
983 * rs6000.c (rs6000_output_load_toc_table): Use ld for 64-bit.
984 (output_toc): Use single TOC slot or llong minimal-toc for DFmode
985 and DImode 64-bit. Use llong for minimal-toc SFmode and
986 SYMBOL_REF / LABEL_REF 64-bit.
987 (output_function_profiler): Use llong for profiler label and ld to
988 load 64-bit label address.
990 Wed Aug 19 17:52:27 1998 Nick Clifton (nickc@cygnus.com)
992 * config/arm/thumb.md (extendqisi2_insn): Cope with REG +
995 Wed Aug 19 14:13:31 PDT 1998 Jeff Law (law@cygnus.com)
997 * version.c: Bump for snapshot.
999 Wed Aug 19 13:10:30 PDT 1998 Jeff Law (law@cygnus.com)
1001 * version.c: Bump for snapshot.
1003 Wed Aug 19 13:06:47 1998 Jason Merrill <jason@yorick.cygnus.com>
1005 * collect2.c (extract_init_priority): Use atoi instead of strtoul.
1007 Wed Aug 19 13:51:35 1998 Hans-Peter Nilsson <hp@axis.se>
1009 * tm.texi (Misc): Fix typo "teh".
1011 * tm.texi (PIC): Fix typo "PPIC".
1013 * tm.texi (Caller Saves): Say that DEFAULT_CALLER_SAVES has no
1014 effect when -O2 and higher.
1015 * invoke.texi (Optimize Options): Likewise for -fcaller-saves.
1017 1998-08-19 Michael Hayes <michaelh@ongaonga.chch.cri.nz>
1019 * regclass.c: Changed register set documentation to be consistent
1022 * final.c (final_start_function) Removed redundant test for
1025 Wed Aug 19 13:28:41 1998 Mark Mitchell <mark@markmitchell.com>
1027 * rtl.h (rtx_function): New type.
1028 (for_each_rtx): New function.
1029 * rtlanal.c (for_each_rtx): Define it.
1031 * recog.c (change_t): New type.
1032 (change_objects, change_old_codes, change_locs, change_olds):
1034 (changes): New variable.
1035 (validate_change): Dynamically allocate room for more changes, if
1036 necessary. Uses changes array instead of change_objects, etc.
1037 (apply_change_group): Use changes array instead of
1038 change_objects, etc.
1040 * loop.c (loop_mem_info): New type.
1041 (loop_mems): New variable.
1042 (loop_mems_idx): Likewise.
1043 (looop_mems_allocated): Likewise.
1044 (scan_loop): Remove nregs parameter.
1045 (next_insn_in_loop): New function.
1046 (load_mems_and_recount_loop_regs_set): Likewise.
1047 (load_mems): Likewise.
1048 (insert_loop_mem): Likewise.
1049 (replace_loop_mem): Likewise.
1050 (replace_label): Likewise.
1051 (INSN_IN_RANGE_P): New macro.
1052 (loop_optimize): Don't pass max_reg_num() to scan_loop.
1053 (scan_loop): Remove nregs parameter, compute it after any new
1054 registers are created by load_mems. Use INSN_IN_RANGE_P and
1055 next_insn_in_loop rather than expanding them inline. Call
1056 load_mems to load memory into pseudos, if appropriate.
1057 (prescan_loop): Figure out whether or not there are jumps from the
1058 loop to targets other than the label immediately following the
1059 loop. Call insert_loop_mem to notice all the MEMs used in the
1060 loop, if it could be safe to pull MEMs into REGs for the duration
1062 (strength_reduce): Use next_insn_in_loop. Tweak comments.
1064 Wed Aug 19 08:29:44 1998 Richard Earnshaw (rearnsha@arm.com)
1066 * arm.c (arm_override_options): Remove lie about ignoring PIC flag.
1068 Wed Aug 19 07:08:15 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
1070 * config/sparc/sparc.c (finalize_pic): Check for the correct
1071 nonlocal_goto_receiver UNSPEC number.
1072 * config/sparc/sparc.md (nonlocal_goto_receiver): Add comment
1073 making note of this dependency existing in sparc.c
1074 (negtf2_notv9 split): Give NEG SFmode.
1075 (negsf2): Fix insn output string.
1077 Tue Aug 18 12:40:27 1998 Richard Henderson <rth@cygnus.com>
1079 * c-common.c (decl_attributes): Issue an error if the argument
1080 to alias is not a string.
1082 Tue Aug 18 10:33:30 1998 Jeffrey A Law (law@cygnus.com)
1084 * haifa-sched.c (sched_analyze): Put all JUMP_INSNs on the last
1085 pending memory flush list.
1087 * combine.c (can_combine_p): Allow combining insns with REG_RETVAL
1089 (try_combine): Allow combining insns with REG_LIBCALL notes.
1091 * expr.c (emit_block_move): Do not call memcpy as a libcall
1092 instead build up a CALL_EXPR and call it like any other
1094 (clear_storage): Similarly for memset.
1096 * regmove.c (fixup_match_2): Do not call reg_overlap_mentioned_p
1099 * Makefile.in (cplus-dem.o): Provide explicit rules for building
1102 * regmove.c (optimize_reg_copy_1): Update REG_N_CALLS_CROSSED
1103 and REG_LIVE_LENGTH as successful substitutions are made.
1105 Tue Aug 18 07:15:27 1998 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
1107 * config/sparc/sparc.c (ultra_find_type): Add empty semicolon
1108 statement after end of loop label.
1110 Tue Aug 18 07:13:27 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
1112 * config/sparc/sparc.c (ultra_types_avail): New variable.
1113 (ultra_build_types_avail): New function to record mask of insn
1114 types in ready list at this cycle.
1115 (ultrasparc_sched_reorder): Call it.
1116 (ultra_find_type): Use it to quicken the search. Also simplif
1117 dependency check, don't use rtx_equal_p because we know exactly
1118 what we are looking for.
1120 Tue Aug 18 03:20:53 1998 Richard Earnshaw (rearnsha@arm.com)
1122 * arm.h (SECONDARY_INPUT_RELOAD_CLASS): Return NO_REGS if compiling
1123 for architecture v4.
1125 Mon Aug 17 21:26:38 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
1127 * config/sparc/sparc.md (sltu, sgeu): Don't FAIL, call
1129 (movsf_const_intreg, movsf_const_high, movsf_const_lo,
1130 movdf_const_intreg and helper splits): New patterns to move float
1131 constants into integer registers.
1132 (negtf2, negdf2, abstf2, absdf2): Rework using new patterns and
1135 Mon Aug 17 11:46:19 1998 Jeffrey A Law (law@cygnus.com)
1138 * tree.c (build_index_type): Copy TYPE_SIZE_UNIT from sizetype
1140 * c-decl.c (finish_enum): Copy TYPE_SIZ_UNIT from enumtype to tem.
1142 * rs6000.c (secondary_reload_class): For TARGET_ELF, indicate that
1143 a BASE_REGS register is needed as an intermediate when copying
1144 a symbolic value into any register class other than BASE_REGS.
1146 * expr.c (move_by_pieces): No longer static. Remove prototype.
1147 * rtl.h (move_by_pieces): Add extern prototype.
1148 * mips.c (expand_block_move): Handle aligned straight line copy by
1149 calling move_by_pieces.
1151 * expr.c (expand_expr): Allow assignments from TImode PARM_DECLs
1154 Mon Aug 17 10:28:52 1998 Mark Mitchell <mark@markmitchell.com>
1156 * stmt.c (expand_end_loop): Tidy. Allow unconditional
1157 jumps out of the loop to be treated as part of the exit test.
1159 Mon Aug 17 10:06:11 1998 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
1160 Jeff Law <law@cygnus.com>
1162 * Makefile.in (cplus-dep.o): Use cplus-dem.c from libiberty.
1163 * cplus-dem.c: Delete.
1165 * Makefile.in (fold-const.o): depend on $(RTL_H).
1167 * fold-const.c: Include rtl.h to get the prototype for
1168 `set_identifier_local_value'.
1170 * loop.c (express_from_1): Remove unused variable `tmp'.
1171 (combine_givs): cast the first argument of bzero to char *.
1173 * toplev.c (display_help): Remove unused variable `looking_for_start'.
1175 * c-decl.c (init_decl_processing): Remove unneeded &.
1177 * alpha.h (alpha_initialize_trampoline): Provide prototype.
1179 * except.c (set_exception_lang_code, set_exception_version_code):
1180 Change parameter from `short' to `int' to avoid using a gcc
1183 * except.h (set_exception_lang_code, set_exception_version_code):
1184 Likewise for prototypes.
1186 * flow.c (count_reg_references): Remove unused variables `regno'
1189 * gcse.c (hash_scan_insn): Declare parameter `in_libcall_block'.
1191 * prefix.c (translate_name): Cast the result of `alloca'.
1193 * varray.h (VARRAY_FREE): Reimplement as a `do-while(0)' statement.
1195 Mon Aug 17 09:23:42 1998 Andreas Schwab <schwab@issan.informatik.uni-dortmund.de>
1197 * config/m68k/m68k.c: Include "system.h" instead of <stdio.h>.
1199 (valid_dbcc_comparison_p): Mark mode argument as unused.
1200 (symbolic_operand): Likewise.
1201 (legitimize_pic_address): Likewise.
1202 (const_uint32_operand): Likewise.
1203 (const_sint32_operand): Likewise.
1204 * sched.c [!INSN_SCHEDULING]: Define only dummy function
1205 schedule_insns and comment out rest of file.
1207 * m68k.c (output_move_simode_const): Use subl to move a zero into an
1209 (output_move_[hq]imode): Likewise.
1211 Mon Aug 17 09:15:47 1998 Jeffrey A Law (law@cygnus.com)
1213 * toplev.c (main): Enable -fstrict-aliasing for -O2 and above.
1214 * invoke.texi: Corresponding changes.
1216 Mon Aug 17 02:03:55 1998 Richard Henderson <rth@cygnus.com>
1218 * regclass.c (allocate_reg_info): Respect MIN when clearing data.
1220 Sun Aug 16 17:37:06 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
1222 * config/sparc/sparc.c (ultra_code_from_mask,
1223 ultra_cmove_results_ready_p, ultra_fpmode_conflict_exists,
1224 ultra_find_type, ultra_schedule_insn, ultra_flush_pipeline,
1225 ultrasparc_sched_init, ultrasparc_variable_issue,
1226 ultra_rescan_pipeline_state, ultrasparc_sched_reorder): New
1227 functions to describe UltraSPARC pipeline exactly to Haifa.
1228 (ultrasparc_adjust_cost): Indicate IMUL type insns have zero cost,
1229 as there is nothing the scheduler can do about it. Indicate that
1230 REG_DEP_OUTPUT's collide. Fixup formatting.
1231 * config/sparc/sparc.h (RTX_COSTS): Fixup integer multiply and
1232 divide costs on Ultra for DImode.
1233 (MD_SCHED_INIT, MD_SCHED_REORDER, MD_SCHED_VARIABLE_ISSUE):
1235 * config/sparc/sparc.md (ieu_unnamed function unit): Rename to
1236 ieuN and add call_no_delay_slot to type list.
1237 (cti function unit): New unit for branches on UltraSPARC.
1238 (subx/addx insns): Set type to misc.
1239 (sidi zero/sign extension insns on arch64): Set type to shift.
1240 (sign_extendhidi2_insn): Set type to sload.
1242 Sun Aug 16 13:52:00 1998 David Edelsohn <edelsohn@mhpcc.edu>
1244 * rs6000.c (rs6000_stack_info): Use if == 0 for sizes.
1245 (output_epilog): Use if != 0 for offset.
1246 (rs6000_fatal_bad_address): Prepare for Intl.
1247 * rs6000.h (rs6000_fatal_bad_address): Declare.
1248 * rs6000.md (movsfcc, movdfcc): Use else if.
1249 (elf_high): Use {liu|lis}.
1250 (elf_low): Use {cal|la}. Remove %a template from old mnemonics.
1251 (movsi): Use rs6000_fatal_bad_address.
1253 Sun Aug 16 01:53:21 1998 Richard Henderson <rth@cygnus.com>
1255 * reload.c (find_equiv_reg): Reject equivalences separated
1256 by a volatile instruction.
1258 Sun Aug 16 00:21:44 1998 Franz Sirl <Franz.Sirl-kernel@lauterbach.com>
1260 * rs6000/linux.h (CPP_OS_DEFAULT_SPEC): Define.
1262 Sat Aug 15 20:51:35 1998 Richard Henderson <rth@cygnus.com>
1264 * alpha.md (movsicc): Fix mode mismatch.
1266 Sat Aug 15 20:22:33 1998 H.J. Lu (hjl@gnu.org)
1268 * config/alpha/alpha.h (ASM_OUTPUT_MI_THUNK): Handle aggregated
1270 * config/alpha/win-nt.h (ASM_OUTPUT_MI_THUNK): Likewise.
1272 Sat Aug 15 08:39:49 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
1274 * config/sparc/sparc.md (movsi_lo_sum_pic_label_reg): Remove
1275 write-only modifier from operand 1 constraint.
1277 Sat Aug 15 06:28:19 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
1279 * config/sparc/sparc.c (sparc_emit_set_const64_quick1): If
1280 emitting a XOR of -1 at the end, emit a NOT instead for combine's
1282 (sparc_emit_set_const64): Likewise, also when computing trailing
1283 bits do not negate low_bits and make fast_int an int.
1285 Fri Aug 14 21:07:03 1998 Jeffrey A Law (law@cygnus.com)
1287 * loop.c (add_label_notes): Do not ignore references to labels
1288 before dispatch tables. Mirrors Apr 8 change to mark_jump_label.
1289 * gcse.c (add_label_notes): Similarly.
1291 * pa.h (ASM_OUTPUT_MI_THUNK): Strip name encoding.
1293 * m68k.md (adddi_dilshr32): One of the operands must be a register.
1294 (adddi_dishl32): Similarly.
1296 Fri Aug 14 14:12:59 1998 Jason Merrill <jason@yorick.cygnus.com>
1298 * i386.h (MODES_TIEABLE_P): Reorganize to shut up warnings.
1299 * alias.c (memrefs_conflict_p): Add braces to shut up warnings.
1300 * cse.c (cse_basic_block): Add parens to shut up warnings.
1302 Fri Aug 14 12:58:21 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
1304 * config/sparc/sparc.c (sparc_emit_set_const64_quick2,
1305 sparc_emit_set_const64_longway, const64_is_2insns,
1306 create_simple_focus_bits, sparc_emit_set_const64): Fix more bugs
1307 in 64-bit constant formation.
1308 * config/sparc/sparc.md (snesi_zero_extend split): Generate
1309 rtl for addx not subx.
1310 (define_insn movdi_const64_special): Make available even when
1311 HOST_BITS_PER_WIDE_INT is not 64.
1312 (movdi_lo_sum_sp64_cint, movdi_high_sp64_cint): Remove.
1313 (losum_di_medlow, sethm, setlo): Make op2 symbolic_operand.
1314 (cmp_siqi_trunc_set, cmp_diqi_trunc_set): Encapsulate both
1315 instances of operand 1 inside a QI subreg.
1316 (xordi3_sp64_dbl): Remove '%' constraint for op1.
1317 (one_cmpldi2_sp64): Fix output string.
1318 (one_cmplsi2_not_liveg0): Rewrite to remove unneeded extra
1320 (unnamed arch64 ashift DI): Truncate shift count if greater than
1323 Fri Aug 14 21:52:53 1998 J"orn Rennecke <amylaar@cygnus.co.uk>
1325 * expr.c (store_expr): Don't optimize away load-store pair
1326 when either source or destination have a side effect.
1328 Fri Aug 14 16:50:10 1998 John Carr <jfc@mit.edu>
1330 * genrecog.c (add_to_sequence): Fatal error if the modes of the
1331 operands of SET are incompatible.
1333 * alpha.md: Fix max and min patterns so modes of SET operands match.
1335 Fri Aug 14 12:22:55 1998 Ian Lance Taylor <ian@cygnus.com>
1337 * configure.in: Avoid [[ by using test and changequote.
1338 * configure: Rebuild.
1340 Fri Aug 14 01:22:31 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
1342 * rtl.def (CONSTANT_P_RTX): Fix typo in string name.
1344 * config/sparc/sparc.md (seqdi_special_trunc, snedi_special_trunc,
1345 seqsi_special_extend, snesi_special_extend, snesi_zero_extend and
1346 split, snedi_zero_trunc and split, seqsi_zero_extend and split,
1347 seqdi_zero_trunc and split, pic_lo_sum_di, pic_sethi_di,
1348 movdi_cc_sp64_trunc, movdi_cc_reg_sp64_trunc, addx_extend_sp32 and
1349 split, addx_extend_sp64, subx_extend_sp64, subx_extend and split):
1350 Fix mismatching modes in SET operands.
1351 (conditional move patterns): Fix formatting.
1352 (unnamed subx arch64 pattern): Remove duplicate insn.
1354 Fri Aug 14 00:34:34 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
1356 * config/sparc/sparc.c (const64_operand, const64_high_operand):
1357 Get it right when HOST_BITS_PER_WIDE_INT is not 64.
1358 (input_operand): Fixup test for what we accept for constant
1360 (sparc_emit_set_const32, sparc_emit_set_symbolic_const64): Give
1362 (safe_constDI): Remove.
1363 (sparc_emit_set_safe_HIGH64, gen_safe_SET64, gen_safe_OR64,
1364 gen_safe_XOR64): New functions.
1365 (sparc_emit_set_const64_quick1, sparc_emit_set_const64_quick2,
1366 sparc_emit_set_const64_longway, sparc_emit_set_const64): Use
1368 * config/sparc/sparc.md (define_insn xordi3_sp64_dbl): Only make
1369 available when HOST_BITS_PER_WIDE_INT is not 64.
1370 (define_insn movdi_sp64_dbl, movdi_const64_special): Likewise and
1371 move before movdi_insn_sp64 pattern.
1372 (define_insn movdi_lo_sum_sp64_dbl, movdi_high_sp64_dbl): Remove.
1373 (define_insn sethi_di_medlow, seth44, setm44, sethh): Use
1374 symbolic_operand as predicate for second operand.
1375 (DImode minus split on arch32, negsi2 expander, one_cmplsi2
1376 expander): Give set VOIDmode.
1378 Fri Aug 14 01:45:06 1998 Mumit Khan <khan@xraylith.wisc.edu>
1380 * i386/cygwin32 (DEFAULT_PCC_STRUCT_RETURN): Define.
1382 Fri Aug 14 01:40:21 1998 Geoffrey Keating <geoffk@ozemail.com.au>
1384 * rs6000/linux.h (LINK_SPEC): Pass -G args to the linker.
1386 Fri Aug 14 01:23:23 1998 Richard Earnshaw (rearnsha@arm.com)
1388 * arm/netbsd.h (TARGET_DEFAULT): Default includes software floating
1390 (CPP_FLOAT_DEFAULT_SPEC): Re-define accordingly.
1392 Fri Aug 14 01:19:08 1998 Robert Lipe <robertl@dgii.com>
1394 * install.texi: Various SCO OpenServer tweaks.
1396 Thu Aug 13 20:14:40 1998 Jim Wilson <wilson@cygnus.com>
1398 * reload1.c (eliminate_regs_in_insn): Handle another case when
1399 eliminating the frame pointer to the hard frame pointer. Add
1400 missing ep->to_rtx check to one existing case.
1402 * mips/mips.md (movhi_internal2+2): Fix typo mem:SI -> mem:HI.
1404 Thu Aug 13 17:08:11 1998 Jason Merrill <jason@yorick.cygnus.com>
1406 * tree.h: De-conditionalize init_priority code.
1408 * mips.h (NM_FLAGS): Change from -Bp to -Bn.
1409 * collect2.c (NM_FLAGS): Change from -p to -n.
1411 * configure.in: Turn on collect2 for mipstx39-elf.
1412 Handle use_collect2=no properly.
1414 * c-common.c: De-conditionalize init_priority code.
1415 * collect2.c (extract_init_priority, sort_ids): New fns.
1416 (main): Call sort_ids.
1417 Move sequence_number to file scope.
1419 * configure.in: Handle --enable-init-priority.
1420 * c-common.c (attrs): Add A_INIT_PRIORITY.
1421 (init_attributes, decl_attributes): Likewise.
1422 * tree.h (DEFAULT_INIT_PRIORITY, MAX_INIT_PRIORITY): New macros.
1423 * tree.c (get_file_function_name_long): Split out...
1424 (get_file_function_name): ...from here.
1426 Thu Aug 13 16:09:53 1998 Martin von Loewis <loewis@informatik.hu-berlin.de>
1428 * expr.c (safe_from_p): Change code to ERROR_MARK only when not
1431 Thu Aug 13 15:24:48 1998 Jason Merrill <jason@yorick.cygnus.com>
1433 * toplev.c (display_help): Add braces to shut up warnings.
1434 * tree.c (simple_cst_equal): Likewise.
1436 * fold-const.c (non_lvalue): Don't deal with null pointer
1438 (fold, case COMPOUND_EXPR): Wrap a constant 0 in a NOP_EXPR.
1440 * c-typeck.c (initializer_constant_valid_p): Allow conversion of 0
1441 of any size to a pointer.
1443 Thu Aug 13 12:53:13 1998 Jim Wilson <wilson@cygnus.com>
1445 * i386/winnt.c (i386_pe_asm_file_end): Check TREE_SYMBOL_REFERENCED.
1447 Wed Aug 12 17:25:18 1998 Jeffrey A Law (law@cygnus.com)
1449 * mn10300.c (REG_SAVE_BYTES): Only reserve space for registers
1450 which will be saved.
1451 * mn10300.md (prologue insn): Only save registers which need saving.
1452 (epilogue insn): Similarly.
1454 * mn10300.c, mn10300.h, mn10300.md: Remove "global zero register"
1457 Wed Aug 12 12:39:16 1998 Gavin Romig-Koch <gavin@cygnus.com>
1459 * mips/mips.h (ENCODE_SECTION_INFO): Set SYMBOL_REF_FLAG for
1460 VAR_DECL's in gp addressable sections.
1462 Tue Aug 11 23:02:31 1998 John Carr <jfc@mit.edu>
1464 * sparc.c: Change return <exp> to <exp>; return; in functions
1466 * sparc.md: Add empty semicolon statement after final label in
1469 Tue Aug 11 22:42:01 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
1471 * config/sparc/sparc.md (define_insn addx_extend): Rename to
1472 addx_extend_sp64, only allow when TARGET_ARCH64.
1473 (define_insn addx_extend_sp32 and split): Version that works when
1475 (define_insn subx_extend): Likewise.
1476 (define_split adddi3 and subdi3 with zero extension): Fixup and
1477 correct bugs when not TARGET_ARCH64.
1479 Tue Aug 11 16:04:34 1998 John Carr <jfc@mit.edu>
1481 * except.c (set_exception_lang_code, set_exception_version_code):
1482 Use prototype-style definition if __STDC__, to match declaration
1485 * genemit.c: Change FAIL and DONE macros not to use loops.
1487 Tue Aug 11 12:27:03 1998 Jim Wilson <wilson@cygnus.com>
1489 * dwarf2out.c (ASM_OUTPUT_DWARF_ADDR_CONST): Use
1490 ASM_OUTPUT_DWARF2_ADDR_CONST if defined.
1492 * mips/mips.md (reload_outsi): Use M16_REG_P when TARGET_MIPS16.
1494 Tue Aug 11 18:12:53 1998 Dave Love <d.love@dl.ac.uk>
1496 * README.g77: Update from Craig.
1498 Tue Aug 11 04:46:01 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
1500 * config/sparc/sparc.c (sparc_emit_set_const32): INTVAL is of
1502 (safe_constDI sparc_emit_set_const64_quick1,
1503 sparc_emit_set_const64_quick2, sparc_emit_set_const64_longway,
1504 analyze_64bit_constant, const64_is_2insns,
1505 create_simple_focus_bits): Fix some bugs when compiled on real
1507 (function_arg_record_value_3, function_arg_record_value_2,
1508 function_arg_record_value): Add fully prototyped forward decls.
1509 * config/sparc/sparc.md (define_insn cmpsi_insn_sp32): Rename back
1510 to cmpsi_insn and use on both 64 and 32 bit targets.
1511 (define_insn cmpsi_insn_sp64): Remove.
1512 (define_expand zero_extendsidi2): Allow for 32-bit target too.
1513 (define_insn zero_extendsidi2_insn): Rename to
1514 zero_extendsidi2_insn_sp64.
1515 (define_insn zero_extendsidi2_insn_sp32): New pattern and
1516 assosciated forced split for it.
1518 * config/sparc/sparc.c (const64_operand, const64_high_operand):
1520 * config/sparc/sparc.h: Declare them.
1521 (PREDICATE_CODES): Add them.
1522 * config/sparc/sparc.md (movdi_lo_sum_sp64_dbl,
1523 movdi_high_sp64_dbl, xordi3_sp64_dbl): Use them.
1525 Mon Aug 10 22:57:24 1998 John Carr <jfc@mit.edu>
1527 * config/sparc/sparc.md (define_insn jump): Output ba,pt not b,pt
1528 in v9 case as the latter makes the Solaris assembler crash.
1530 Mon Aug 10 22:39:09 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
1532 * config/sparc/sparc.c (input_operand): Do not accept a LO_SUM MEM
1533 for TFmode when !v9. We require offsettable memory addresses.
1534 * config/sparc/sparc.h (ALTER_HARD_SUBREG): Handle TFmode to
1535 DFmode register number conversions.
1536 * config/sparc/sparc.md (define_split DFmode moves): If register
1537 is a SUBREG do alter_subreg on it before using.
1538 (define_expand movtf): Fixup comment about alignment on v9.
1539 (define_split TFmode moves): Don't use gen_{high,low}part, create
1540 explicit SUBREGs instead.
1542 Mon Aug 10 19:02:55 1998 John Carr <jfc@mit.edu>
1544 * Makefile.in (mbchar.o): Depend on mbchar.c.
1546 Mon Aug 10 04:28:13 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
1547 Richard Henderson <rth@cygnus.com>
1549 Rewrite Sparc backend for better code generation and
1550 improved sparc64 support.
1551 * config/sparc/sp64-elf.h: Set JUMP_TABLES_IN_TEXT_SECTION to
1553 * config/sparc/sysv4.h: Likewise.
1554 * config/sparc/sparc.c (v8plus_regcmp_p, sparc_operand,
1555 move_operand, v8plus_regcmp_op, emit_move_sequence,
1556 singlemove_string, doublemove_string, mem_aligned_8,
1557 output_move_double, output_move_quad, output_fp_move_double,
1558 move_quad_direction, output_fp_move_quad, output_scc_insn):
1560 (small_int_or_double): New predicate.
1561 (gen_compare_reg): Remove TARGET_V8PLUS cmpdi_v8plus emission.
1562 (legitimize_pic_address): Emit movsi_{high,lo_sum}_pic instead of
1563 old pic_{sethi,lo_sum}_si patterns.
1564 (mem_min_alignment): New generic function to replace
1565 mem_aligned_8, which uses REGNO_POINTER_ALIGN information when
1566 available and can test for arbitrary alignments. All callers
1568 (save_regs, restore_regs, build_big_number,
1569 output_function_prologue, output_cbranch, output_return,
1570 sparc_flat_save_restore, sparc_flat_output_function_prologue,
1571 sparc_flat_output_function_epilogue): Prettify
1573 (output_function_epilogue): Likewise and add code to output
1574 deferred case vectors.
1575 (output_v9branch): Likewise, add new arg INSN and use it to tack
1576 on branch prediction settings. All callers changed.
1577 (print_operand): Likewise and output %l44 for LO_SUMs when
1579 (sparc_splitdi_legitimate): New function to make sure DImode
1580 splits can be run properly when !arch64.
1581 (sparc_initialize_trampoline, sparc64_initialize_trampoline):
1582 Reformat example code in comments.
1583 (set_extends): Remove UNSPEC/v8plus_clear_high case.
1584 (sparc_addr_diff_list, sparc_addr_list): New statics to keep track
1585 of deferred case vectors we need to output.
1586 (sparc_defer_case_vector): Record a case vector.
1587 (sparc_output_addr_vec, sparc_output_addr_diff_vec,
1588 sparc_output_deferred_case_vectors): New functions to output them.
1589 (sparc_emit_set_const32): New function to form 32-bit constants in
1590 registers when that requires more than one instruction.
1591 (safe_constDI, sparc_emit_set_const64_quick1,
1592 sparc_emit_set_const64_quick2, sparc_emit_set_const64_longway,
1593 analyze_64bit_constant, const64_is_2insns,
1594 create_simple_focus_bits, sparc_emit_set_const64): New functions
1595 which do the same for 64-bit constants when arch64.
1596 (sparc_emit_set_symbolic_const64): New function to emit address
1597 loading for all code models on v9.
1598 * config/sparc/sparc.h (CONDITIONAL_REGISTER_USAGE): Do not make
1599 %g1 fixed when arch64, unfix %g0 when TARGET_LIVE_G0.
1600 (ALTER_HARD_SUBREG): Fix thinko, return REGNO + 1 not 1.
1601 (SECONDARY_INPUT_RELOAD_CLASS, SECONDARY_OUTPUT_RELOAD_CLASS): Fix
1602 inaccuracies in comments, add symbolic and text_segment operands
1603 when TARGET_CM_MEDANY and TARGET_CM_EMBMEDANY respectively. Use
1604 GENERAL_REGS in these cases as a temp REG is needed to load these
1605 addresses into a register properly.
1606 (EXTRA_CONSTRAINT): Document more accurately, remove Q case as it
1608 (GO_IF_LEGITIMATE_ADDRESS): Allow TFmode for LO_SUM on v9 since fp
1609 quads are guarenteed to have 16-byte alignment.
1610 (LEGITIMIZE_ADDRESS): For SYMBOL_REF, CONST, and LABEL_REF use
1611 copy_to_suggested_reg instead of explicit LO_SUM and HIGH.
1612 (ASM_OUTPUT_ADDR_VEC, ASM_OUTPUT_ADDR_DIFF_VEC): New macros for
1613 deferred case vector implementation.
1614 (ASM_OUTPUT_ADDR_VEC_ELT): Use fputc to output newline.
1615 (ASM_OUTPUT_ADDR_DIFF_ELT): Parenthesize LABEL in macro calls.
1616 Generate "internal label - label" instead of "label - 1b".
1617 (PRINT_OPERAND_ADDRESS): For LO_SUM use %l44 on TARGET_CM_MEDMID.
1618 (PREDICATE_CODES): Remove sparc_operand, move_operand,
1619 v8plus_regcmp_op. Add small_int_or_double, input_operand, and
1621 (doublemove_string, output_block_move, output_fp_move_double,
1622 output_fp_move_quad, output_move_double, output_move_quad,
1623 output_scc_insn, singlemove_string, mem_aligned_8, move_operand,
1624 sparc_operand, v8plus_regcmp_op, v8plus_regcmp_p): Remove externs.
1625 (sparc_emit_set_const32, sparc_emit_set_const64,
1626 sparc_emit_set_symbolic_const64, input_operand, zero_operand,
1627 mem_min_alignment, small_int_or_double): Add externs.
1628 * config/sparc/sparc.md: Document the many uses of UNSPEC and
1629 UNSPEC_VOLATILE in this backend.
1630 (define_function_unit ieu): Rename to ieu_unnamed. Add move and
1631 unary to types which execute in it.
1632 (define_function_unit ieu_shift): Rename to ieu0.
1633 (define_function_unit ieu1): New, executes compare, call, and
1634 uncond_branch type insns.
1635 (define_function_units for type fdivs, fdivd, fsqrt): These
1636 execute in the fpu multiply unit not the adder on UltraSparc.
1637 (define_expand cmpdi): Disallow TARGET_V8PLUS.
1638 (define_insn cmpsi_insn): Rename to cmpsi_insn_sp32.
1639 (define_insn cmpsi_insn_sp64): New, same as sp32 variant except it
1640 allows the arith_double_operand predicate and rHI constraint when
1642 (define_insn cmpdi_sp64, cmpsf_fpe, cmpdf_fpe, cmptf_fpe,
1643 cmpsf_fp, cmpdf_fp, cmptf_fp, sltu_insn, neg_sltu_insn,
1644 neg_sltu_minux_x, neg_sltu_plus_x, sgeu_insn, neg_sgeu_insn,
1645 sltu_plus_x, sltu_plus_x, sltu_plus_x_plus_y, x_minus_sltu,
1646 sgeu_plus_x, x_minus_sgeu, movqi_cc_sp64, movhi_cc_sp64,
1647 movsi_cc_sp64, movdi_cc_sp64, movsf_cc_sp64, movdf_cc_sp64,
1648 movtf_cc_sp64, movqi_cc_reg_sp64, movhi_cc_reg_sp64,
1649 movsi_cc_reg_sp64, movdi_cc_reg_sp64, movsf_cc_reg_sp64,
1650 movdf_cc_reg_sp64, movtf_cc_reg_sp64, zero_extendhisi2_insn,
1651 cmp_siqi_trunc, cmp_siqi_trunc_set, sign_extendhisi2_insn,
1652 sign_extendqihi2_insn, sign_extendqisi2_insn,
1653 sign_extendqidi2_insn, sign_extendhidi2_insn,
1654 extendsfdf2, extendsftf2, extenddftf2, truncdfsf2, trunctfsf2,
1655 trunctfdf2, floatsisf2, floatsidf2, floatsitf2, floatdisf2,
1656 floatdidf2, floatditf2, fix_truncsfsi2, fix_truncdfsi2,
1657 fix_trunctfsi2, fix_truncsfdi2, fix_truncdfdi2, fix_trunctfdi2,
1658 adddi3_sp64, addsi3, cmp_ccx_plus, cmp_cc_plus_set, subdi_sp64,
1659 subsi3, cmp_minus_ccx, cmp_minus_ccx_set, mulsi3, muldi3,
1660 muldi3_v8plus, cmp_mul_set, mulsidi3, mulsidi3_v8plus,
1661 const_mulsidi3_v8plus, mulsidi3_sp32, const_mulsidi3,
1662 smulsi3_highpart_v8plus, unnamed subreg mult,
1663 const_smulsi3_highpart_v8plus, smulsi3_highpart_sp32,
1664 const_smulsi3_highpart, umulsidi3_v8plus, umulsidi3_sp32,
1665 const_umulsidi3, const_umulsidi3_v8plus, umulsi3_highpart_v8plus,
1666 const_umulsi3_highpart_v8plus, umulsi3_highpart_sp32,
1667 const_umulsi3_highpart, divsi3, divdi3, cmp_sdiv_cc_set, udivsi3,
1668 udivdi3, cmp_udiv_cc_set, smacsi, smacdi, umacdi, anddi3_sp64,
1669 andsi3, and_not_di_sp64, and_not_si, iordi3_sp64, iorsi3,
1670 or_not_di_sp64, or_not_si, xordi3_sp64, xorsi3, xor_not_di_sp64,
1671 xor_not_si, cmp_cc_arith_op, cmp_ccx_arith_op,
1672 cmp_cc_arith_op_set, cmp_ccx_arith_op_set, cmp_ccx_xor_not,
1673 cmp_cc_xor_not_set, cmp_ccx_xor_not_set, cmp_cc_arith_op_not,
1674 cmp_ccx_arith_op_not, cmp_cc_arith_op_not_set,
1675 cmp_ccx_arith_op_not_set, negdi2_sp64, cmp_cc_neg, cmp_ccx_neg,
1676 cmp_cc_set_neg, cmp_ccx_set_neg, one_cmpldi2_sp64, cmp_cc_not,
1677 cmp_ccx_not, cmp_cc_set_not, cmp_ccx_set_not, addtf3, adddf3,
1678 addsf3, subtf3, subdf3, subsf3, multf3, muldf3, mulsf3,
1679 muldf3_extend, multf3_extend, divtf3, divdf3, divsf3, negtf2,
1680 negdf2, negsf2, abstf2, absdf2, abssf2, sqrttf2, sqrtdf2, sqrtsf2,
1681 ashlsi3, ashldi3, unnamed DI ashift, cmp_cc_ashift_1,
1682 cmp_cc_set_ashift_1, ashrsi3, ashrdi3, unnamed DI ashiftrt,
1683 ashrdi3_v8plus, lshrsi3, lshrdi3, unnamed DI lshiftrt,
1684 lshrdi3_v8plus, tablejump_sp32, tablejump_sp64, call_address_sp32,
1685 call_symbolic_sp32, call_address_sp64, call_symbolic_sp64,
1686 call_address_struct_value_sp32, call_symbolic_struct_value_sp32,
1687 call_address_untyped_struct_value_sp32,
1688 call_symbolic_untyped_struct_value_sp32, call_value_address_sp32,
1689 call_value_symbolic_sp32, call_value_address_sp64,
1690 call_value_symbolic_sp64, branch_sp32, branch_sp64,
1691 flush_register_windows, goto_handler_and_restore,
1692 goto_handler_and_restore_v9, goto_handler_and_restore_v9_sp64,
1693 flush, all ldd/std peepholes, return_qi, return_hi, return_si,
1694 return_addsi, return_di, return_adddi, return_sf, all call+jump
1695 peepholes, trap, unnamed trap insns): Prettify output strings.
1696 (define_insn anddi3_sp32, and_not_di_sp32, iordi3_sp32,
1697 or_not_di_sp32, xordi3_sp32, xor_not_di_sp32, one_cmpldi2):
1698 Likewise and force + implement splits for integer cases.
1699 (define_insn return_sf_no_fpu): Likewise and allow to match when
1700 no-fpu because of our subreg SFmode splits.
1701 (define_insn zero_extendqihi2, zero_extendqisi2_insn,
1702 zero_extendqidi2_insn, zero_extendhidi2_insn,
1703 zero_extendsidi2_insn, sign_extendsidi2_insn): Likewise and use
1704 input_operand for second operand.
1705 (cmp_minus_cc, cmp_minus_cc_set): Likewise and use
1706 reg_or_0_operand for operand 2 so new splits can use it.
1707 (cmp_zero_extendqisi2, cmp_zero_extendqisi2_set, cmp_cc_plus,
1708 cmp_cc_xor_not): Likewise and don't forget to check TARGET_LIVE_G0
1710 (cmp_zero_extract, cmp_zero_extract_sp64): Likewise and allow
1711 CONST_DOUBLEs for operand 2.
1712 (define_insn move_label_di): Likewise and label distance
1713 optimization because it no longer works with new deferred case
1714 vector scheme. To be revisited.
1715 (define_insn x_minus_y_minus_sltu, x_minus_sltu_plus_y): Likewise
1716 and allow reg_or_0_operand and J constraint for second operand.
1717 (define_insn jump): Set branch predict taken on V9.
1718 (define_insn tablejump): Emit LABEL_REF + PLUS memory address for
1719 new deferred case vector scheme.
1720 (define_insn pic_tablejump_32, pic_tablejump_64): Remove.
1721 (define_insn negdi2_sp32): Force + implement splits.
1722 (define_insn negsi2, one_cmplsi2): Rename to negsi2_not_liveg0 and
1723 one_cmplsi2_not_liveg0 respectively, and create expander of original
1724 names which emit special rtl for TARGET_LIVE_G0.
1725 (define_insn cmpdi_v8plus, scc_si, scc_di): Remove.
1726 (define_insn seq, sne, slt, sge, sle, sltu, sgeu): Don't do
1727 gen_compare_reg, FAIL instead.
1728 (define_insn sgtu, sleu): Likewise and check gen_s*() return
1729 values when trying to reverse condition codes, if they FAIL then
1731 (define_insn snesi_zero, neg_snesi_zero, snesi_zero_extend,
1732 snedi_zero, neg_snedi_zero, snedi_zero_trunc, seqsi_zero,
1733 neg_seqsi_zero, seqsi_zero_extend, seqdi_zero, neg_seqdi_zero,
1734 seqdi_zero_trunc, x_plus_i_ne_0, x_minus_i_ne_0, x_plus_i_eq_0,
1735 x_minus_i_eq_0): Add new splits to perform these multi-insn cases,
1736 set output string to # to indicate they are mandatory splits.
1737 (define_insn pic_lo_sum_si, pic_sethi_si, pic_lo_sum_di,
1738 pic_sethi_di, move_pic_label_si): Remove.
1739 (define_insn movsi_low_sum, movsi_high, movsi_lo_sum_pic,
1740 movsi_high_pic, movsi_pic_label_reg): New patterns to take their
1742 (define_expand movsi_pic_label_ref, define_insn
1743 movsi_high_pic_label_ref, movsi_lo_sum_pic_label_ref): New
1744 expander and insns to handle PIC label references and deferred
1746 (define_insn get_pc_via_rdpc): Comment out as it is no longer
1748 (define_expand movqi, movhi, movsi, movdi, movsf, movdf, movtf):
1749 Rewrite to not use emit_move_sequence, make use of new constant
1750 formation code, and new splits for all multi-insn cases.
1751 (define_insn movqi_insn): Remove sethi case, it can never happen.
1752 Use reg_or_zero_operand instead of const0_rtx explicit test,
1753 use input_operand instead of move_operand for source, and use
1754 general_operand now for dest.
1755 (define_insn movhi_insn): Similar but leave sethi case.
1756 (define_insn lo_sum_qi, store_qi, store_hi): Remove.
1757 (define_insn sethi_hi lo_sum_hi): Rename to movhi_high and
1758 movhi_lo_sum respectively, prettify output string.
1759 (define_insn movsi_zero_liveg0): New pattern to put zero into a
1760 register when needed on TARGET_LIVE_G0.
1761 (define_insn movsi_insn): Use general_operand and input_operand
1762 for dest and src respectively. Simplify applicability test.
1763 Prettify output strings, and add clr alternative for J
1765 (define_insn movdi_sp32_v9, movdi_sp32, define_splits for
1766 deprecated std and reg-reg DI moves): Remove and...
1767 (define_insn movdi_insn_sp32, movdi_insn_sp64): Replace with new
1768 implementation which uses forced splits for all non-single insn
1770 (define_split DI move cases on !arch64): New splits to handle all
1771 situations of 64-bit double register DImode on 32bit, and
1772 unaligned registers and memory addresses for all subtargets.
1773 (define_insn movsf_const_insn, movdf_const_insn, store_sf):
1775 (define_insn movsf_insn, movsf_no_f_insn): Use general_operand and
1776 input_operand for dest and src respectively, prettify output
1778 (define_insn movdf_insn, movdf_no_e_insn, store_df,
1779 movtf_const_insn, movtf_insn, movtf_no_e_insn, store_tf): Remove
1781 (define_insn movdf_insn_sp32, movdf_no_e_insn_sp32,
1782 movdf_insn_sp64, movdf_no_e_insn_sp64, movtf_insn,
1783 movtf_no_e_insn_sp32, movtf_insn_hq_sp64, movtf_insn_sp64,
1784 movtf_no_e_insn_sp64) Replace with new
1785 implementation which uses forced splits for all non-single insn
1787 (define_split DF move cases): New splits in similar vein to DI
1789 (define_insn sethi_di_medlow, sethi_di_medium_pic,
1790 sethi_di_embmedany_data, sethi_di_embmedany_text, sethi_di_sp64,
1791 movdi_sp64_insn): Remove old v9 code model and constant loading
1793 (define_insn pic_lo_sum_di, pic_sethi_di,
1794 sethi_di_medlow_embmedany_pic, sethi_di_medlow, losum_di_medlow,
1795 seth44, setm44, setl44, sethh, setlm, sethm, setlo,
1796 embmedany_sethi, embmedany_losum, embmedany_brsum,
1797 embmedany_textuhi, embmedany_texthi, embmedany_textulo,
1798 embmedany_textlo, movdi_lo_sum_sp64_cint, movdi_lo_sum_sp64_dbl,
1799 movdi_high_sp64_cint, movdi_high_sp64_dbl): Replace with new
1800 scheme, using unspecs, secondary reloads, and one to one sparc
1801 insn to rtl insn mapping for better scheduling and code gen.
1802 (define_expand reload_indi, reload_outdi): Reload helpers for
1803 MEDANY and EMBMEDANY symbol address loading cases which require a
1805 (define_expand movsicc): Remove v8plus_regcmp cases.
1806 (define_insn movdi_cc_sp64_trunc, movdi_cc_reg_sp64_trunc,
1807 cmp_zero_extendqidi2, cmp_zero_extendqidi2_set, cmp_qidi_trunc,
1808 cmp_diqi_trunc_set): New patterns used by some of the new scc
1810 (define_insn xordi3_sp64_dbl): New pattern used for constant
1811 formation when crossing from 32-bit targets.
1812 (define_insn movsi_cc_reg_v8plus, v8plus_clear_high, and helper
1814 (define_insn addx, subx): Make visible and prettify.
1815 (define_insn adddi3_insn_sp32): Likewise and force split.
1816 (define_insn addx_extend, subx_extend, unnamed): New patterns for
1817 64bit scc split usage.
1818 (define_insn unnamed plusDI zero_extend, unnamed minusDI
1819 zero_extend, subdi3): Force and implement splits.
1821 * final.c (final_scan_insn): Don't output labels if target
1822 specifies ASM_OUTPUT_ADDR_{DIFF}_VEC. Do these macro operations
1825 * reorg.c (dbr_schedule): When taking on BR_PRED notes at the end,
1826 don't forget to walk inside SEQUENCESs too as these are what the
1827 delay slot scheduler will create.
1829 Mon Aug 10 01:21:01 1998 Richard Henderson <rth@cygnus.com>
1831 * alpha.md (extxl+1,+2): New patterns to work around
1834 Sat Aug 8 19:20:22 1998 Gary Thomas (gdt@linuxppc.org)
1836 * rs6000.c (rs6000_allocate_stack_space) Fix typo which
1837 caused bad assembly code to be generated.
1839 Sat Aug 8 18:53:28 1998 Jeffrey A Law (law@cygnus.com)
1841 * netbsd.h: Fix typo.
1843 Mon Aug 3 00:06:42 1998 Robert Lipe <robertl@dgii.com>
1845 * config.sub: Fix typo.
1847 Sun Aug 2 22:39:08 1998 Hans-Peter Nilsson <hp@axis.se>
1849 * invoke.texi (Environment Variables): Typo: Change "ascpects"
1851 (Running Protoize): Typo: Change "ther" into "other".
1853 Sun Aug 2 00:42:50 1998 Jeffrey A Law (law@cygnus.com)
1855 * i386/netbsd.h: Undo previous change to DWARF2_UNWIND_INFO.
1856 * m68k/netbsd.h: Likewise.
1857 * ns32k/netbsd.h: Likewise.
1858 * sparc/netbsd.h: Likewise.
1860 Sat Aug 1 17:59:30 1998 Richard Henderson <rth@cygnus.com>
1862 * ginclude/va-alpha.h (va_list): Use a typedef, not a define.
1863 * ginclude/va-clipper.h (va_list): Likewise.
1865 Fri Jul 31 20:22:02 1998 Michael Meissner <meissner@cygnus.com>
1867 * rs6000.c (rs6000_override_options): If big endian and -Os, use
1868 load/store multiple instructions unless user overrides.
1870 Fri Jul 31 17:08:59 1998 Jeffrey A Law (law@cygnus.com)
1872 * ns32k/netbsd.h: Fix typo.
1874 Fri Jul 31 10:23:55 1998 Doug Evans <devans@canuck.cygnus.com>
1876 * m32r/m32r.h (ASM_OUTPUT_SOURCE_LINE): Always output line number
1877 labels with .debugsym if no parallel insns.
1879 Thu Jul 30 19:15:53 1998 Richard Henderson <rth@cygnus.com>
1881 * alpha.md (fp cmp): Replicate patterns for ALPHA_TP_INSN.
1882 (fcmov): Remove ALPHA_TP_INSN patterns -- fcmov doesn't trap.
1884 Thu Jul 30 19:50:15 1998 David Edelsohn <edelsohn@mhpcc.edu>
1886 * rs6000/x-aix43 (AR_FOR_TARGET_FLAGS): Delete.
1887 (AR_FOR_TARGET): Define.
1889 Thu Jul 30 12:29:12 1998 Mark Mitchell <mark@markmitchell.com>
1891 * dyn-string.h: New file.
1892 * dyn-string.c: Likewise.
1893 * Makefile.in (OBJS): Add dyn-string.o.
1894 (dwarf2out.o): Add dyn-string.h dependency.
1895 (dyn-string.o): List dependencies.
1896 * dwarf2out.c: Include dyn-string.h.
1897 (ASM_NAME_TO_STRING): Use dyn_string_append, rather than strcpy.
1898 (addr_const_to_string): Take a dyn_string_t, not a char* as a
1899 prototype. Use dyn_string_append rather than strcat, throughout.
1900 (addr_to_string): Use dyn_string_t.
1902 Thu Jul 30 13:08:07 1998 Ken Raeburn <raeburn@cygnus.com>
1904 Function entry/exit profiling instrumentation:
1905 * expr.h (profile_function_entry_libfunc,
1906 profile_function_exit_libfunc): Declare new variables.
1907 * optabs.c: Define them here.
1908 (init_optabs): Initialize them.
1909 * tree.h (struct tree_decl): New flag
1910 no_instrument_function_entry_exit.
1911 (DECL_NO_INSTRUMENT_FUNCTION_ENTRY_EXIT): New accessor macro.
1912 * c-decl.c (duplicate_decls): Merge it.
1913 * c-common.c (enum attrs): New value A_NO_INSTRUMENT_FUNCTION.
1914 (init_attributes): Use it for "no_instrument_function".
1915 (decl_attributes): Handle it, for functions that have not yet been
1916 compiled. Set decl flag.
1917 * flags.h (flag_instrument_function_entry_exit): Declare new
1919 * toplev.c (flag_instrument_function_entry_exit): Define it here.
1920 (f_options): New option "instrument-functions".
1921 * function.h (struct function): New field instrument_entry_exit.
1922 * function.c (current_function_instrument_entry_exit): New
1924 (push_function_context_to, pop_function_context_from): Save and
1926 (expand_function_start): Set current_ variable, maybe emit return
1927 label and entry profile call.
1928 (expand_function_end): Maybe emit exit profile call.
1930 Thu Jul 30 00:58:34 1998 Jeffrey A Law (law@cygnus.com)
1932 * i386.md (movqi): When optimizing a load of (const_int 1) into a
1933 NON_QI_REG_P, pretend the register is SImode.
1935 Wed Jul 29 23:49:23 1998 Todd Vierling <tv@netbsd.org>
1937 * configure.in: Use xm-netbsd.h as the NetBSD xm file (not xm-siglist).
1938 Accept arm32 as arm, m68k4k as m68k, mipsle as mips-dec, and any
1939 manufacturer id for ns32k.
1940 * configure: Regenerated.
1941 * config/netbsd.h: When using ASM_WEAKEN_LABEL, make it global too.
1942 * config/t-netbsd: Don't compile libgcc1-test as the fns are in libc.
1943 * config/i386/netbsd.h: Undefine DWARF2_UNWIND_INFO, not define as 0.
1944 * config/m68k/netbsd.h: Same.
1945 * config/ns32k/netbsd.h: Same.
1946 * config/sparc/netbsd.h: Same.
1948 Wed Jul 29 22:39:21 1998 Jeffrey A Law (law@cygnus.com)
1950 * unroll.c (unroll_loop): Do not abort for an UNROLL_MODULO
1951 or UNROLL_COMPLETELY loop that starts with a jump to its
1954 Wed Jul 29 22:18:14 1998 David Edelsohn <edelsohn@mhpcc.edu>
1956 * rs6000/rs6000.md (absdi2 define_split): Swap operands of MINUS.
1957 * rs6000/rs6000.c (mask64_operand): Use HOST_BITS_PER_WIDE_INT.
1958 (print_operand, case 'B'): Don't fall through.
1959 (print_operand, case 'S'): Correct mask begin/end computation.
1960 Use HOST_BITS_PER_WIDE_INT.
1961 * rs6000/rs6000.h (CPP_PREDEFINES): Define _LONG_LONG.
1962 (CONDITIONAL_REGISTER_USAGE): GPR13 fixed if TARGET_64BIT.
1963 * rs6000/aix41.h (CPP_PREDEFINES): Same.
1964 * rs6000/aix43.h (CPP_PREDEFINES): Same.
1966 Wed Jul 29 11:47:10 1998 Nick Clifton <nickc@cygnus.com>
1968 * config/arm/thumb.md (extendqisi2_insn): Remove earlyclobber
1969 constraint from second alternative.
1971 Tue Jul 28 23:29:04 1998 Jason Merrill <jason@yorick.cygnus.com>
1973 * configure.in: Fix --without/--disable cases for local-prefix,
1974 gxx-include-dir and checking.
1976 Tue Jul 28 22:01:23 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
1978 * configure.in (enable_haifa): Set by default for sparc64 too.
1981 Tue Jul 28 23:29:04 1998 Jason Merrill <jason@yorick.cygnus.com>
1983 * i386/cygwin32.h (VALID_MACHINE_TYPE_ATTRIBUTE): New macro.
1984 * i386/winnt.c (associated_type): New fn.
1985 (i386_pe_valid_type_attribute_p): New fn.
1986 (i386_pe_check_vtable_importexport): Remove.
1987 (i386_pe_dllexport_p): Use associated_type.
1988 (i386_pe_dllimport_p): Likewise.
1990 From Antonio M. O. Neto <anmendes@cruzeironet.com.br>:
1991 * i386.c (i386_valid_type_attribute_p): Also accept
1992 attributes for METHOD_TYPEs.
1994 Tue Jul 28 23:17:39 1998 Peter Gerwinski <peter@gerwinski.de>
1996 * tree.c (build_range_type): Copy TYPE_SIZE_UNIT.
1998 Tue Jul 28 22:31:12 1998 Craig Burley <burley@gnu.org>
2000 * gcc.c: Fix commentary describing %g, %u, %U, and %O.
2002 * gcc.c (do_spec_1): Fix handling of %g%O and %U%O to prevent
2003 them from generating a new base name for each occurence of
2006 1998-07-28 Vladimir N. Makarov <vmakarov@cygnus.com>
2008 * cse.c (cse_insn): Enable subsitution inside libcall only for REG,
2010 * rtlanal.c (replace_rtx): Prohibit replaces in CONST_DOUBLE.
2014 * cplus-dem.c (type_kind_t): New type.
2015 (demangle_template_value_parm): Add type_kind_t parameter. Rely
2016 on this paramter, rather than demangling the type again.
2017 (demangle_integral_value): Pass tk_integral.
2018 (demangle_template_: Pass the value returned from do_type.
2019 (do_type): Return a type_kind_t. Pass tk_integral to
2020 demangle_template_value_parm for array bounds.
2021 (demangle_fund_type): Likewise.
2023 Mon Jul 27 00:54:41 1998 Jason Merrill <jason@yorick.cygnus.com>
2025 * tree.c (simple_cst_equal, case CONSTRUCTOR): OK if the elts are
2028 Mon Jul 27 22:18:36 1998 Jeffrey A Law (law@cygnus.com)
2030 * pa.c (move_operand): Accept CONSTANT_P_RTX.
2032 Mon Jul 27 17:18:52 1998 Dave Brolley <brolley@cygnus.com>
2034 * stor-layout.c (layout_type): Handle arrays of bits, for Chill.
2036 * expr.c (get_inner_reference): Handle zero-based, unsigned, array
2039 Mon Jul 27 14:51:33 1998 Jeffrey A Law (law@cygnus.com)
2041 * mn10300.h (DEBUGGER_AUTO_OFFSET): Define.
2042 (DEBUGGER_ARG_OFFSET): Likewise.
2044 * mn10300.md (movsf): Remove last change. Not needed.
2046 Mon Jul 27 14:22:36 1998 Dave Brolley <brolley@cygnus.com>
2048 * c-lex.c (yylex): Fix boundary conditions in character literal and
2049 string literal loops.
2051 Mon Jul 27 11:43:54 1998 Stan Cox <scox@cygnus.com>
2053 * longlong.h (count_leading_zeros): Sparclite scan instruction was
2054 being invoked incorrectly.
2056 * i386.c (ix86_prologue): Added SUBTARGET_PROLOGUE invocation.
2057 * i386/cygwin32.h (STARTFILE_SPEC, LIB_SPEC, SUBTARGET_PROLOGUE):
2059 * i386/win32.h: New file. Hybrid mingw32.h/cygwin32.h configuration.
2060 * configure.in: Added i[34567]86-*-win32
2061 * config.sub: Ditto.
2062 * configure: Rebuilt.
2064 Sun Jul 26 01:11:12 1998 H.J. Lu (hjl@gnu.org)
2066 * i386.h (CONST_DOUBLE_OK_FOR_LETTER_P): Return 0 when eliminating
2067 the frame pointer and compiling PIC code and reload has not completed.
2069 * i386.c (output_to_reg): Add code to emulate non-popping DImode
2072 Sun Jul 26 01:01:32 1998 Jeffrey A Law (law@cygnus.com)
2074 * regmove.c (regmove_optimize): Fix typo initializing regmove_bb_head.
2076 Sat Jul 25 23:29:23 1998 Gerald Pfeifer <pfeifer@dbai.tuwien.ac.at>
2078 * Makefile.in (install-info): Only try to update the info
2079 directory file if it exists in the first place.
2081 Fri Jul 24 18:58:37 1998 Klaus Espenlaub <kespenla@student.informatik.uni-ulm.de>
2083 * rs6000.h (ASM_OUTPUT_CONSTRUCTOR, ASM_OUTPUT_DESTRUCTOR): Delete.
2085 Fri Jul 24 14:20:26 1998 Jeffrey A Law (law@cygnus.com)
2087 * mn10300.md (movqi, movhi, movsi, movsf): Correctly handle
2088 CONST_DOUBLE source.
2090 Fri Jul 24 11:17:04 1998 Nick Clifton <nickc@cygnus.com>
2092 * config/arm/thumb.c (thumb_print_operand): Decode %_ in asm
2093 strings as the insertion of USER_LABEL_PREFIX.
2094 * config/arm/thumb.h (PRINT_OPERAND_PUNCT_VALID_P): Accept _ as a
2096 * config/arm/thumb.md: Use %_ as a prefix to gcc library function
2099 Thu Jul 23 18:53:20 1998 Jim Wilson <wilson@cygnus.com>
2101 * dbxout.c (dbxout_range_type): Only call dbxout_type_index for
2102 already defined type.
2104 Thu Jul 23 13:49:41 1998 Jeffrey A Law (law@cygnus.com)
2106 * expr.c (check_max_integer_computation_mode): Allow conversions
2107 of constant integers to MAX_INTEGER_COMPUTATION_MODE.
2108 (expand_expr): Likewise.
2110 Thu Jul 23 11:12:06 1998 Alexandre Petit-Bianco <apbianco@cygnus.com>
2112 * expr.c (expand_expr): Expand RETURN_EXPR.
2114 Thu Jul 23 11:00:29 1998 Jim Wilson <wilson@cygnus.com>
2116 * dwarf2out.c (dwarf2out_finish): Call stripattributes on TEXT_SECTION.
2118 Wed Jul 22 19:10:00 1998 Catherine Moore <clm@cygnus.com>
2120 * dwarf2out.c (output_aranges): Call stripattributes
2121 for TEXT_SECTION references.
2122 (output_line_info): Likewise.
2124 Wed Jul 22 14:08:54 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
2126 * profile.c (branch_prob): Call allocate_reg_info after outputting
2127 profile rtl in instrument_arcs.
2129 Wed Jul 22 12:47:49 1998 Jim Wilson <wilson@cygnus.com>
2131 * fixinc.irix (math.h): Install wrapper instead of copying.
2133 Wed Jul 22 12:37:14 1998 Alexandre Petit-Bianco <apbianco@cygnus.com>
2135 * tree.def (EXPR_WITH_FILE_LOCATION): Defined as an 'e' expression
2136 so WFL are expanded correctly when contained in a COMPOUND_EXPR.
2137 * tree.h (EXPR_WFL_EMIT_LINE_NOTE): Change macro not to use
2138 lang_flag_0. Added documentation in the flag table.
2140 Tue Jul 21 23:28:35 1998 Klaus Kaempf <kkaempf@rmi.de>
2142 * cccp.c (do_include): Fix vax c style include handling.
2144 Tue Jul 21 13:28:19 1998 Jason Merrill <jason@yorick.cygnus.com>
2146 * cplus-dem.c (do_type): Use demangle_template_value_parm for arrays.
2148 Sun Jul 12 01:27:05 1998 Jason Merrill <jason@yorick.cygnus.com>
2150 * fold-const.c (non_lvalue): Don't deal with null pointer
2152 (fold, case COMPOUND_EXPR): Wrap a constant 0 in a NOP_EXPR.
2154 Tue Jul 21 15:49:31 1998 David Edelsohn <edelsohn@mhpcc.edu>
2156 * rs6000.h (PREDICATE_CODES): Add CONSTANT_P_RTX.
2157 * rs6000.md (movsi, movdi): Add CONSTANT_P_RTX.
2158 * rs6000.c (short_cint_operand): Add CONSTANT_P_RTX.
2159 (u_short_cint_operand): Same.
2160 (reg_or_cint_operand): Same.
2161 (logical_operand): Same.
2162 (input_operand): Same.
2163 (reg_or_short_operand): Use u_short_cint_operand.
2165 Tue Jul 21 08:56:42 1998 Richard Henderson <rth@cygnus.com>
2167 * alpha.md (fix_truncdfsi2, fix_truncsfsi2): Remove the define_expands,
2168 but keep the insns and splits. Adjust so when the ultimate destination
2169 is memory, use cvtql.
2171 Tue Jul 21 08:55:09 1998 Richard Henderson <rth@cygnus.com>
2173 * flow.c (regno_uninitialized): Fixed regs are never uninitialized.
2175 Tue Jul 21 00:31:01 1998 Jeffrey A Law (law@cygnus.com)
2177 * gcc.c (do_spec): Call "error" not "warning".
2179 * configure.in: Fix minor problems with gas feature detection code.
2180 * configure: Rebuilt.
2182 * gcc.c (do_spec): Issue a warning for '%[]' usage.
2185 * gcc.c: Delete %[spec] support.
2186 (do_spec_1, case '('): Likewise.
2187 (do_spec_1, case '['): Call error.
2189 Mon Jul 20 22:34:17 1998 Richard Henderson <rth@cygnus.com>
2191 * alpha.h (CPP_SPEC): Tidy. Hook to cpp_cpu and cpp_subtarget.
2192 (CPP_SUBTARGET_SPEC): Default to empty string.
2193 (CPP_AM_*, CPP_IM_*, CPP_CPU_*, CPP_CPU_SPEC): New.
2194 (EXTRA_SPECS, SUBTARGET_EXTRA_SPECS): New.
2195 * alpha/elf.h (LD_SPEC): Use %(elf_dynamic_linker).
2196 * alpha/linux-elf.h (SUBTARGET_EXTRA_SPECS): New.
2198 * alpha/linux.h (CPP_PREDEFINES): Tidy.
2199 * alpha/netbsd-elf.h (SUBTARGET_EXTRA_SPECS): New.
2200 * alpha/netbsd.h (CPP_PREDEFINES): Tidy.
2201 * alpha/osf.h (CPP_PREDEFINES): Remove bits subsumed by CPP_CPU_SPEC.
2202 * alpha/win-nt.h (CPP_PREDEFINES): Likewise.
2203 * alpha/vsf.h (CPP_PREDEFINES): Likewise.
2204 (CPP_SUBTARGET_SPEC): New. Do this instead of overriding CPP_SPEC.
2205 * alpha/vxworks.h: Likewise.
2207 Mon Jul 20 22:51:57 1998 Ken Raeburn <raeburn@cygnus.com>
2209 * mips.md (reload_outsi): Added missing REGNO call.
2210 (smulsi3_highpart, umulsi3_highpart): Provide prototype for
2212 (mul_acc_di, mul_acc_64bit_di): Don't use match_op_dup, use
2213 another match_operator and compare the codes.
2215 * mips.h (MASK_DEBUG_E, MASK_DEBUG_I): Set to zero.
2217 * MIPS multiply pattern fixes:
2218 * mips.h (enum reg_class, REG_CLASS_NAMES, REG_CLASS_CONTENTS):
2219 Add union classes for HI, LO, or HILO plus general registers.
2220 (GENERATE_MADD): Deleted.
2221 * mips.md (mulsi3_mult3): Don't disparage output-LO alternative.
2222 Add TARGET_MAD to condition.
2223 (mulsi3): Test HAVE_mulsi3_mult3, not specific flags.
2224 (mul_acc_si): Expand GENERATE_MADD here; it's the only use. Use
2225 "*d" for accumulator, to give preference to LO initially but not
2228 Mon Jul 20 16:16:38 1998 Dave Brolley <brolley@cygnus.com>
2230 * configure.in (enable_c_mbchar): New configure option.
2231 (extra_cpp_objs): Always available now.
2233 * cexp.y (mbchar.h): #include it.
2234 (yylex): Handle Multibyte characters in character literals.
2236 * cccp.c (mbchar.h): #include it.
2237 (main): Set character set based on LANG environment variable.
2238 (rescan): Handle multibyte characters in comments.
2239 (skip_if_group): See above.
2240 (validate_else): See above.
2241 (skip_to_end_of_comment): See above.
2242 (macarg1): See above.
2243 (discard_comments): See above.
2244 (rescan): Handle multibyte characters in string and character literals.
2245 (collect_expansion): See above.
2246 (skip_quoted_string): See above.
2247 (macroexpand): See above.
2248 (macarg1): See above.
2249 (discard_comments): See above.
2250 (change_newlines): See above.
2252 * c-lex.c (mbchar.h): #include it.
2253 (GET_ENVIRONMENT): New macro.
2254 (init_lex): Set character set based on LANG environment variable.
2255 (yylex): Handle multibyte characters in character literals.
2256 (yylex): Handle multibyte characters in string literals.
2258 * Makefile.in (mbchar.o): New target.
2259 (cccp$(exeext)): @extra_cpp_objs@ is always available.
2260 (cppmain$(exeext)): @extra_cpp_objs@ is always available.
2262 * mbchar.[ch]: New files for multibyte character handling.
2264 Mon Jul 20 01:11:11 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
2266 * jump.c (jump_optimize): When simplifying noop moves and
2267 PUSH_ROUNDING, fix thinko so we use same criterion for identifying
2268 the PUSHes to rewrite in second loop as we did in the first.
2270 Sun Jul 19 08:23:53 1998 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
2272 * cplus-dem.c (demangle_nested_args): Make function definition
2273 static to match the prototype.
2275 Fri Jul 17 14:58:44 1998 Richard Henderson <rth@cygnus.com>
2277 * alloca.c: Respect USE_C_ALLOCA.
2278 * gencheck.c (xmalloc): Ignore __GNUC__ for definition.
2279 * gengenrtl.c (xmalloc): Likewise.
2281 Fri Jul 17 14:18:14 1998 Richard Henderson <rth@cygnus.com>
2283 * loop.h (struct induction): Add no_const_addval.
2284 * loop.c (the_movables, reg_address_cost): New variables.
2285 (init_loop): Init reg_address_cost.
2286 (loop_optimize): Call end_alias_analysis.
2287 (scan_loop): Init the_movables.
2288 (record_giv): Init induction->no_const_addval.
2289 (basic_induction_var) [PLUS]: Use rtx_equal_p instead of ==.
2290 [REG]: Rearrange loop search test to catch more cases.
2291 (general_induction_var): Return success not benefit; take an extra
2292 argument for that. Change all callers.
2293 (simplify_giv_expr) [PLUS]: Always combine invariants. Use sge_plus.
2294 [MULT]: Use rtx_equal_p instead of ==. Combine simple invariants.
2295 [default]: Search the_movables for additional combinations.
2296 (sge_plus_constant, sge_plus): New functions.
2297 (express_from_1): New function.
2298 (express_from): Always define. Rewrite using express_from_1.
2299 (combine_givs_p): Handle more cases. Ignore address cost.
2300 (cmp_combine_givs_stats): New function.
2301 (combine_givs_used_once, combine_givs_benefit_from): New functions.
2302 (combine_givs): Rewrite to do best-fit combination.
2304 * fold-const.c (operand_equal_p): Handle RTL_EXPR.
2305 (fold): Do a complete (A*C)+(B*C) association check.
2307 Fri Jul 17 11:21:55 1998 Jim Wilson <wilson@cygnus.com>
2309 * function.c (fixup_var_refs_insns): Handle CLOBBER of a CONCAT.
2311 Fri Jul 17 11:48:55 1998 Jeffrey A Law (law@cygnus.com)
2313 * mn10300.c (MODES_TIEABLE_P): Fix typo.
2315 Fri Jul 17 03:26:12 1998 Rihcard Earnshaw (rearnsha@arm.com)
2317 * tree.c (valid_machine_attribute): Only create a new type variant if
2318 there is a decl to use it.
2320 Thu Jul 16 14:48:04 1998 Nick Clifton <nickc@cygnus.com>
2322 * gcc.c (do_spec_1): Cope with %g/%u/%U options which do not have
2325 Fri Jul 17 03:24:40 1998 Hans-Peter Nilsson <hp@axis.se>
2327 * extend.texi (Explicit Reg Vars): Typo: change "may deleted" into "may
2330 Thu Jul 16 14:48:47 1998 Jeffrey A Law (law@cygnus.com)
2332 * mn10300.c (count_tst_insns): New arg oreg_countp. Callers changed.
2333 Simplify tests for clearing an address register.
2334 (expand_prologue): Corresponding changes.
2336 * mn10300.md (movXX patterns): Make sure the destination is an
2337 ADDRESS_REG when substituting "zero_areg" for (const_int 0).
2338 (logical patterns): Split into expanders + patterns
2339 (zero and sign extension patterns): Similarly.
2340 (shift patterns): Similarly.
2342 Thu Jul 16 01:17:44 1998 Richard Henderson <rth@cygnus.com>
2344 * loop.c (emit_iv_add_mult): Scan the entire insn list generated
2345 for the sequence, recording base values.
2347 Wed Jul 15 10:49:55 1998 Richard Henderson <rth@cygnus.com>
2349 * i386.h (CPP_CPU_SPEC): Remove -Asystem(unix).
2351 Tue Jul 14 14:15:30 1998 Nick Clifton <nickc@cygnus.com>
2353 * gcc.c: Remove ANSI-C ism from --help code.
2355 * toplev.c: Support --help with USE_CPPLIB.
2357 Tue Jul 14 14:46:08 1998 Jeffrey A Law (law@cygnus.com)
2359 * configure.in: Rework gas feature code to work with symlink based
2362 * extend.texi: Clarify some issues related to local variables
2363 assigned to explicit registers.
2365 * mn10300.md (mulsi): Turn into expander + pattern.
2367 * mn10300.md (movsi, movsf, movdi, movdf): Remove "x" from I -> a
2370 Tue Jul 14 07:41:59 1998 Richard Earnshaw (rearnsha@arm.com)
2372 * arm/tcoff.h (USER_LABEL_PREFIX): Make it empty to match coff.h.
2374 Tue Jul 14 03:02:44 1998 Jeffrey A Law (law@cygnus.com)
2376 * version.c: Bump again to distinguish mainline tree from the
2379 See ChangeLog.0 for earlier changes.
2382 add-log-time-format: current-time-string
projects / platform / upstream / gcc.git / blob