1 Wed Sep 9 12:31:35 1998 Jeffrey A Law (law@cygnus.com)
3 * Makefile.in (profile.o): Depend on insn-config.h
5 Wed Sep 9 09:36:51 1998 Jim Wilson <wilson@cygnus.com>
7 * iris6.h (DWARF2_UNWIND_INFO): Undef.
9 Wed Sep 9 01:32:01 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
11 Add preliminary native sparcv9 Solaris support.
12 * configure.in: Recognize sparv9-*-solaris2*
14 * config.sub: Recognize sparcv9 just like sparc64.
15 * config/sparc/sol2-c1.asm config/sparc/sol2-ci.asm
16 config/sparc/sol2-cn.asm: Macroize so it can be shared between
17 32-bit and 64-bit Solaris systems.
18 * config/sparc/t-sol2: Assemble those with cpp.
19 * config/sparc/sparc.h (TARGET_CPU_sparcv9): New alias for v9.
20 (*TF*_LIBCALL): If ARCH64 use V9 names.
21 * config/sparc/{xm-sysv4-64,sol2-sld-64}.h: New files.
23 Wed Sep 9 01:07:30 1998 Jakub Jelinek <jj@sunsite.ms.mff.cuni.cz>
25 * config/sparc/sparc.h (TARGET_CM_MEDMID): Fix documentation.
26 (CASE_VECTOR_MODE): Set to SImode even if PTR64, when MEDLOW and
28 (ASM_OUTPUT_ADDR_{VEC,DIFF}_ELT): Check CASE_VECTOR_MODE not
30 * config/sparc/sparc.md (tablejump): Likewise, and sign extend op0
31 to Pmode if CASE_VECTOR_MODE is something else.
33 Wed Sep 9 00:10:31 1998 Jeffrey A Law (law@cygnus.com)
35 * prefix.c (update_path): Correctly handle cases where PATH is
36 a substring of the builtin prefix, but specifies a different
39 Tue Sep 8 23:46:04 1998 Hans-Peter Nilsson <hp@axis.se>
41 * expr.c: Corrected comment about what MOVE_RATIO does.
42 * config/alpha/alpha.h: Ditto.
43 * config/1750a/1750a.h: Ditto.
44 * config/clipper/clipper.h: Ditto.
45 * config/i386/i386.h: Ditto.
47 Tue Sep 8 22:56:12 1998 Jeffrey A Law (law@cygnus.com)
49 * configure.in (m68k-next-nextstep3*): Use collect2.
50 Similarly for x86 NeXT configurations.
53 Tue Sep 8 01:38:57 1998 Nathan Sidwell <nathan@acm.org>
55 * configure.in: Don't assume srcdir is .../gcc
58 Sat Sep 5 16:34:34 EDT 1998 John Wehle (john@feith.com)
60 * global.c: Update comments.
61 (global_alloc): Assign allocation-numbers
62 even for registers allocated by local_alloc in case
63 they are later spilled and retry_global_alloc is called.
64 (mark_reg_store, mark_reg_clobber,
65 mark_reg_conflicts, mark_reg_death): Always record a
66 conflict with a pseudo register even if it has been
67 assigned to a hard register.
68 (dump_conflicts): Don't list pseudo registers already assigned to
69 a hard register as needing to be allocated, but do list their
71 * local-alloc.c: Update comment.
73 Mon Sep 7 23:38:01 1998 Jeffrey A Law (law@cygnus.com)
75 * configure.in: Check for bogus GCC_EXEC_PREFIX and LIBRARY_PATH.
78 Mon Sep 7 22:41:46 1998 Michael Meissner <meissner@cygnus.com>
80 * rs6000.c (rs6000_override_options): Fix name for ec603e, to add
82 * t-ppccomm (MULTILIB_MATCHES_FLOAT): Add support for -mcpu=xxx
83 for all targets that set -msoft-float.
85 Mon Sep 7 23:30:07 1998 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
87 * toplev.c (print_switch_values): Make static to match prototype.
89 Mon Sep 7 19:13:59 1998 Jeffrey A Law (law@cygnus.com)
91 * configure.in: If we are unable to find the "gnatbind" program,
92 then do not configure the ada subdir.
95 Sun Sep 6 14:03:58 PDT 1998 Jeff Law (law@cygnus.com)
97 * version.c: Bump for snapshot.
99 Sun Sep 6 13:28:07 PDT 1998 Jeff Law (law@cygnus.com)
101 * version.c: Bump for snapshot.
103 Sun Sep 6 08:54:14 1998 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
105 * Makefile.in (toplev.o): Depend on $(EXPR_H).
106 (insn-extract.o, insn-attrtab.o): Depend on toplev.h.
108 * gansidecl.h: Define ATTRIBUTE_NORETURN.
110 * genattrtab.c: Have insn-attrtab.c include toplev.h.
112 * genextract.c: Have insn-extract.c include toplev.h.
114 * rtl.h: Don't prototype `fatal_insn_not_found' and `fatal_insn'.
116 * toplev.c: Include expr.h.
117 (really_sorry, fancy_abort): Remove prototypes.
118 (set_target_switch): Add argument in prototype.
119 (vfatal): Mark prototype with ATTRIBUTE_NORETURN.
120 (v_really_sorry): Likewise.
121 (print_version, print_single_switch, print_switch_values): Make
122 static and add prototype arguments.
123 (decl_printable_name): Add prototype arguments.
124 (lang_expand_expr_t): New typedef.
125 (lang_expand_expr): Declare as a lang_expand_expr_t.
126 (incomplete_decl_finalize_hook): Add prototype argument.
127 (decl_name): Mark variable `verbosity' with ATTRIBUTE_UNUSED.
128 (botch): Likewise for variable `s'.
129 (rest_of_type_compilation): Mark variables `type' and `toplev'
130 with ATTRIBUTE_UNUSED if none of DBX_DEBUGGING_INFO,
131 XCOFF_DEBUGGING_INFO or SDB_DEBUGGING_INFO are defined.
132 (display_help): Make variable `i' an `unsigned long'.
133 (main): Remove unused parameter `envp'.
134 Cast assignment to `lang_expand_expr' to a `lang_expand_expr_t'.
135 Cast -1 when comparing it with a `size_t'.
137 * toplev.h (fatal, fatal_io_error, pfatal_with_name): Mark
138 prototype with ATTRIBUTE_NORETURN.
139 (fatal_insn_not_found, fatal_insn, really_sorry,
140 push_float_handler, pop_float_handler): Add prototypes.
141 (fancy_abort): Mark prototype with ATTRIBUTE_NORETURN.
142 (do_abort, botch): Add prototypes.
144 Sat Sep 6 12:05:18 1998 John Carr <jfc@mit.edu>
146 * final.c (final): If a label is reached only from a single jump,
147 call NOTICE_UPDATE_CC on the jump and its predecessor before
148 emitting the insn after the label.
150 * i386.h: Add AMD K6 support.
151 Change TARGET_* macros to use table lookup.
152 (INITIALIZE_TRAMPOLINE): Improve trampoline code.
153 (ADJUST_COST): Change definition to call function in i386.c.
154 (ISSUE_RATE): Define as 2 for anything newer than an 80486.
155 * i386.c: Add AMD K6 support.
156 Add constants for feature tests used by TARGET_* macros.
157 (split_di): If before reload, call gen_lowpart and gen_highpart.
158 (x86_adjust_cost): New function.
159 (put_jump_code): New function.
160 (print_operand): New codes 'D' and 'd'.
161 * i386.md: New insn types. New insn attribute "memory".
162 Redefine scheduling parameters to use new types and add AMD K6
163 support. Explicitly set type of most insns.
164 (move insns): K6 prefers movl $0,reg to xorl reg,reg. Pentium
165 Pro and K6 prefer movl $1,reg to incl reg.
166 (adddi3, subdi3): Set cc_status.
167 (DImode shift patterns): Change label counters from HOST_WIDE_INT
168 to int; x86 can't have more than 2^31 DImode shifts per file.
169 (setcc): Combine all setcc patterns. Allow writing memory.
170 Combine all jump patterns using match_operator.
171 (*bzero): Name pattern. Emit mutliple stos instructions when that
172 is faster than rep stos.
173 (xordi3, anddi3, iordi3): Simplify DImode logical patterns and
176 Sun Sep 6 11:17:20 1998 Dave Love <d.love@dl.ac.uk>
178 * config/m68k/x-next (BOOT_LDFLAGS): Define suitably for f771
181 Sat Sep 5 22:05:25 1998 Richard Henderson <rth@cygnus.com>
183 * alpha.c (alpha_ra_ever_killed): Inspect the topmost sequence,
184 not whatever we're generating now.
186 * alpha.c (set_frame_related_p, FRP): New.
187 (alpha_expand_prologue): Mark frame related insns.
188 (alpha_expand_epilogue): Likewise, but with a null FRP.
189 * alpha.h (INCOMING_RETURN_ADDR_RTX): New.
190 * alpha.md (exception_receiver): New.
191 * alpha/crtbegin.asm (.eh_frame): New beginning.
192 (__do_frame_setup, __do_frame_takedown): New.
193 * alpha/crtend.asm (.eh_frame): New ending.
194 * alpha/elf.h (DWARF2_DEBUGGING_INFO): Define.
195 (ASM_SPEC): Don't emit both dwarf2 and mdebug.
196 (ASM_FILE_START): Don't emit .file for dwarf2.
198 * rtl.h (enum reg_note): Add REG_FRAME_RELATED_EXPR.
199 * rtl.c (reg_note_name): Likewise.
200 * rtl.texi (REG_NOTES): Likewise.
201 * dwarf2out.c (dwarf2out_frame_debug): Use it. Recognize a store
204 Sat Sep 5 14:47:17 1998 Richard Henderson <rth@cygnus.com>
206 * i386.h (PREFERRED_RELOAD_CLASS): Standard fp constants load to TOS.
207 * i386.md (movsf, movdf, movxf): Validate memory address returned
208 from force_const_mem. Kill useless REG_EQUAL setting code.
210 Sat Sep 5 14:23:31 1998 Torbjorn Granlund <tege@matematik.su.se>
212 * m68k.md (zero_extendsidi2): Fix typo.
214 Sat Sep 5 13:40:24 1998 Krister Walfridsson <cato@df.lth.se>
216 * configure.in: Removed references to the removed file
217 * config/xm-netbsd.h. Use ${cpu_type}/xm-netbsd.h for
218 * arm*-*-netbsd* and ns32k-*-netbsd*.
219 * config/i386/xm-netbsd.h: Removed unnecessary file.
220 * config/m68k/xm-netbsd.h: Likewise.
221 * config/sparc/xm-netbsd.h: Likewise.
222 * config/mips/xm-netbsd.h: Likewise.
224 Sat Aug 29 13:32:58 1998 Mumit Khan <khan@xraylith.wisc.edu>
226 * i386/cygwin32.h (BIGGEST_ALIGNMENT): Define.
227 (PCC_BITFIELD_TYPE_MATTERS): Define to be 0.
229 * i386/cygwin32.h (ASM_OUTPUT_SECTION_NAME): Don't check for
230 for exact section attributions.
232 * i386/mingw32.h (CPP_PREDEFINES): Add __MSVCRT__ for msvc
234 * i386/crtdll.h (CPP_PREDEFINES): Define.
236 Sat Sep 5 03:23:05 1998 Jeffrey A Law (law@cygnus.com)
238 * m68k.md (5200 movqi): Do not allow byte sized memory references
240 * m68k.c (output_move_qimode): Do not use byte sized operations on
243 * Makefile.in (pexecute.o): Use pexecute.c from libiberty. Provide
244 explicit rules for building. Similarly for alloca, vfprintf,
245 choose-temp and mkstemp, getopt, getopt1, and obstack.
246 (INCLUDES): Add $(srcdir)/../include.
247 * pexecute.c, alloca.c, vfprintf.c, choose-temp.c, mkstemp.c: Delete.
248 * getopt.h, getopt.c getopt1.c, obstack.c, obstack.h: Likewise.
250 Fri Sep 4 11:57:50 1998 Tom Tromey <tromey@cygnus.com>
252 * gcc.c (do_spec_1): [case 'o'] Account for
253 lang_specific_extra_outfiles.
254 (main): Correctly clear all slots in outfiles for
255 lang_specific_extra_outfiles. Set input_file_number before
256 calling lang_specific_pre_link.
258 Fri Sep 4 10:37:07 1998 Jim Wilson <wilson@cygnus.com>
260 * loop.c (load_mems): Fix JUMP_LABEL field after for_each_rtx call.
262 Fri Sep 4 02:01:05 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
264 * config/sparc/sparc.c (output_double_int): In all V9 symbolic
266 (sparc_output_deferred_case_vectors): If no work to do, return.
267 Fix thinko in Sept 1 change.
269 1998-09-03 SL Baur <steve@altair.xemacs.org>
271 * Makefile.in: add semicolon in BISON definition for portability.
273 Thu Sep 3 13:34:41 1998 Toon Moene <toon@moene.indiv.nluug.nl>
275 * config/nextstep.c (handle_pragma): Correct name of third
278 Tue Sep 1 11:30:33 1998 Nick Clifton <nickc@cygnus.com>
280 * config/m32r/m32r.md: Change (reg:CC 17) to (reg:SI 17).
281 * config/m32r/m32r.h: Make register 17 be fixed.
282 * config/m32r/m32r.c: Use SImode for cc operations.
284 Thu Sep 3 18:17:34 1998 Benjamin Kosnik <bkoz@cygnus.com>
286 * invoke.texi (Warning Options): Add -Wnon-template-friend
289 Thu Sep 3 18:16:16 1998 Michael Meissner <meissner@cygnus.com>
291 * rs6000.c (rs6000_override_options): Add -mcpu={401,e603e}.
293 Thu Sep 3 18:05:16 1998 David Edelsohn <edelsohn@mhpcc.edu>
295 * rs6000.md (movsf): Disable explicit secondary-reload-like
296 functionality if TARGET_POWERPC64.
297 (movdf): Remove TARGET_POWERPC64 explicit secondary-reload-like
301 Thu Sep 3 11:41:40 1998 Robert Lipe <robertl@dgii.com>
303 * fixinc.sco: Borrow code to wrap 'bool' typedefs from tinfo.h
304 and term.h from fixinc.wrap.
306 Thu Sep 3 09:47:31 1998 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
308 * aclocal.m4 (GCC_HEADER_STRING): New macro to detect if it is
309 safe to include both string.h and strings.h together.
310 (GCC_NEED_DECLARATION): Test STRING_WITH_STRINGS when deciding
311 which headers to search for function declarations. Continue to
312 prefer string.h over strings.h when both are not acceptable.
314 * acconfig.h (STRING_WITH_STRINGS): Add stub.
316 * configure.in: Call GCC_HEADER_STRING.
318 * system.h: Test STRING_WITH_STRINGS when deciding which headers
319 to include. Continue to prefer string.h over strings.h when both
322 Wed Sep 2 23:56:29 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
324 * config/sparc/sparc.c (output_double_int): If V9 and MEDLOW, do
325 not assume top 32-bits of symbolic addresses are zero if
328 Thu Sep 3 00:23:21 1998 Richard Henderson <rth@cygnus.com>
330 * ginclude/va-alpha.h: Protect entire second portion of the
331 file against double inclusion.
333 Thu Sep 3 00:37:55 1998 Ovidiu Predescu <ovidiu@aracnet.com>
335 Added support for the Boehm's garbage collector.
336 * configure.in: Handle --enable-objc-gc.
337 * configure: Rebuilt.
338 * Makefile.in (CHECK_TARGETS): Add check-objc.
339 (check-objc): New rule.
340 * objc/Make-lang.in: Build a different Objective-C library that
341 runs with the Boehm's collector.
342 * objc/encoding.c (objc_round_acc_size_for_types): New function.
343 * objc/encoding.c: Correctly compute the size of compound types in
344 the presence of bitfields. Skip the variable name of the type if
345 any. Added support for long long.
346 * objc/encoding.h (_C_GCINVISIBLE): New specifier.
347 (_F_GCINVISIBLE): New mask.
348 * objc/gc.c: New file. Compute the type memory mask associated with
349 a class based on the runtime information.
350 * objc/misc.c: Added the hooks that use the Boehm's collector
351 allocation functions.
352 * objc/objc-act.c (build_class_template): Generate a new class
353 member (gc_object_type) to hold the class' type memory mask.
354 (build_shared_structure_initializer): Initialize the new member to
356 (encode_complete_bitfield): New function. Generate the new
358 (encode_field_decl): Generate the new encoding only for the GNU
360 * objc/objc-api.h (_C_LNG_LNG, _C_ULNG_LNG): New specifiers for the
362 (class_get_gc_object_type): New function to mark a pointer instance
363 variable as a weak pointer.
364 * objc/objc-features.texi: New file.
365 * objc/objc.h (gc_object_type): New class member.
366 * objc/objects.c (class_create_instance): Create a typed memory
367 object when compiled with Boehm's collector support.
368 * objc/sendmsg.c (__objc_init_install_dtable): Call
369 __objc_send_initialize instead of setting the initialize flag.
370 (__objc_send_initialize): Call __objc_generate_gc_type_description
371 to generate the class type memory mask. Rewrite the code that
372 sends the +initialize so that it is called only once (bug report
373 and fix from Ronald Pijnacker <Ronald.Pijnacker@best.ms.philips.com>).
374 * testsuite/objc: New testsuite for Objective-C type encoding.
375 * testsuite/lib/objc-torture.exp: New file.
376 * testsuite/lib/objc.exp: New file.
378 Wed Sep 2 14:47:36 1998 Jim Wilson <wilson@cygnus.com>
380 * jump.c (jump_optimize): In if/then/else transformations, add
381 another call to modified_between_p for the jump insn.
383 Wed Sep 2 14:16:49 1998 Jeffrey A Law (law@cygnus.com)
385 * fix-header.c (symlink): Treat like readlink.
387 Wed Sep 2 19:30:06 1998 J"orn Rennecke <amylaar@cygnus.co.uk>
389 * dwarfout.c (fundamental_type_code): Encode 32 bit floats/doubles
392 Wed Sep 2 10:06:07 1998 Nick Clifton <nickc@cygnus.com>
394 * config/nextstep.h: Update HANDLE_PRAGMA macro.
395 * config/h8300/h8300.h: Update HANDLE_PRAGMA macro.
396 * config/i960/i960.h: Update HANDLE_PRAGMA macro.
398 * config/nextstep.c (handle_pragma): Take three arguments, as per
399 the new HANDLE_PRAGMA macro specification.
400 * config/h8300/h8300.c (handle_pragma): Take three arguments, as
401 per the new HANDLE_PRAGMA macro specification.
402 * config/i960/i960.c (process_pragma): Take three arguments, as
403 per the new HANDLE_PRAGMA macro specification.
405 Wed Sep 2 09:25:29 1998 Nick Clifton <nickc@cygnus.com>
407 * c-lex.c (check_newline): Call HANDLE_PRAGMA before
408 HANDLE_SYSV_PRAGMA if both are defined. Generate warning messages
409 if unknown pragmas are encountered.
410 (handle_sysv_pragma): Interpret return code from
411 handle_pragma_token (). Return success/failure indication rather
412 than next unprocessed character.
413 (pragma_getc): New function: retrieves characters from the
414 input stream. Defined when HANDLE_PRAGMA is enabled.
415 (pragma_ungetc): New function: replaces characters back into the
416 input stream. Defined when HANDLE_PRAGMA is enabled.
418 * c-pragma.c (handle_pragma_token): Return success/failure status
421 * c-pragma.h: Change prototype of handle_pragma_token().
423 * varasm.c: (handle_pragma_weak): Only create this function if
424 HANDLE_PRAGMA_WEAK is defined.
426 * c-common,c (decl_attributes): If defined call the expression
427 contained within the INSERT_ATTRIBUTES macro before adding
428 attributes to a decl.
430 * tm.texi (HANDLE_PRAGMA): Document the new verion of
431 HANDLE_PRAGMA, which takes three arguments.
432 (INSERT_ATTRIBUTES): Document this new macro.
434 * LANGUAGES: Document the new version of HANDLE_PRAGMA and the
435 new INSERT_ATTRIBUTES macro.
437 Wed Sep 2 02:03:23 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
439 * config/sparc/sparc.md (movdf): Only generate special RTL for
441 (move_label_di): Remove
442 (movdi_pic_label_ref, movdi_high_pic_label_ref,
443 movdi_lo_sum_pic_label_ref): New patterns for 64-bit label
445 * config/sparc/sparc.h (ASM_OUTPUT_ADDR_VEC_ELT,
446 ASM_OUTPUT_ADDR_DIFF_ELT): Don't do anything special for MEDLOW,
447 output an .xword for all 64-bit cases.
449 Tue Sep 1 15:55:17 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
451 * config/sparc/sparc.c (finalize_pic): Don't output arbitrary
452 alignment, use FUNCTION_BOUNDARY instead.
453 (sparc_output_deferred_case_vectors): Likewise.
455 Mon Aug 31 17:25:41 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
457 * config/sparc/sparc.md (movsf_const_intreg): Kill warning.
458 (movtf_insn_sp64, movtf_no_e_insn_sp64): Reorder alternatives.
460 Mon Aug 31 13:57:55 1998 Richard Henderson <rth@cygnus.com>
462 * alpha/va_list.h: New file.
463 * alpha/x-alpha (EXTRA_HEADERS): New. Add va_list.h.
465 Mon Aug 31 14:55:02 1998 Jeffrey A Law (law@cygnus.com)
467 * NEWS: Add SCO Openserver and Unixware 7 notes.
471 Mon Aug 31 15:42:18 1998 Dave Brolley <brolley@cygnus.com>
473 * varasm.c (compare_constant_1): Handle RANGE_EXPR.
474 (record_constant_1): Handle RANGE_EXPR.
476 Mon Aug 31 10:54:03 1998 Richard Henderson <rth@cygnus.com>
478 * print-rtl.c (print_rtx): NOTE_INSN_LIVE has an rtx not a bitmap.
479 * haifa-sched.c (sched_analyze): Handle NOTE_INSN_RANGE_START
480 and NOTE_INSN_RANGE_END specially.
481 (reemit_notes): Likewise.
483 Mon Aug 31 10:18:52 1998 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
485 * sparc.c (TMASK, UMASK): Use `(unsigned)1' not `1U'.
486 (ultrasparc_sched_init): Remove unneeded &.
488 Mon Aug 31 10:47:16 1998 Andreas Schwab <schwab@issan.informatik.uni-dortmund.de>
490 * config/m68k/m68k.h (TARGET_SWITCHES): Don't remove MASK_68040
491 for m68020-60, to prevent the use of fintrz.
493 Sun Aug 30 22:17:20 1998 Mark Mitchell <mark@markmitchell.com>
495 * configure.in: If the native compiler is GCC use $(WARN_CFLAGS)
497 * Makefile.in: Likewise.
498 * configure: Regenerated.
500 Sun Aug 30 22:15:41 1998 H.J. Lu (hjl@gnu.org)
502 * configure.in (gxx_include_dir): Changed to
503 '${prefix}/include/g++'-${libstdcxx_interface}.
504 * configure: Rebuilt.
506 Sun Aug 30 20:19:43 1998 Hans-Peter Nilsson <hp@axis.se>
508 * expr.c (expand_expr): Change ">" to ">=" making MOVE_RATIO use
510 * tm.texi (Costs): Say MOVE_RATIO is number of mem-mem move
511 *sequences* *below* which scalar moves will be used.
513 Sun Aug 30 17:18:43 1998 Jeffrey A Law (law@cygnus.com)
515 * collect2.c (mktemp): Delete unused declaration.
517 * config/xm-netbsd.h: Remove unnecessary file.
518 * config/*/xm-netbsd.h: Do not include the generic xm-netbsd.h
519 file anymore, it is not needed.
521 Sun Aug 30 16:05:45 1998 Mark Mitchell <mark@markmitchell.com>
523 * convert.c (convert_to_integer): Issue an error on conversions to
526 Sun Aug 30 16:47:20 1998 Martin von Lvwis <loewis@informatik.hu-berlin.de>
528 * Makefile.in: Add lang_tree_files and gencheck.h.
529 * configure.in: Generate them.
530 * gencheck.c: Include gencheck.h.
532 Sat Aug 29 21:38:24 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
534 * config/sparc/sparc.md (pic_lo_sum_di, pic_sethi_di): Rename to
535 movdi_lo_sum_pic and movdi_high_pic and make visible.
536 * config/sparc/sparc.c (legitimize_pic_address): For -fPIC,
537 emit these when Pmode is not SImode.
538 * config/sparc/linux64.h (SPARC_DEFAULT_CMODEL): Make CM_MEDLOW.
540 Sat Aug 29 14:59:32 1998 Mumit Khan <khan@xraylith.wisc.edu>
542 * i386/cygwin32.h (ASM_OUTPUT_SECTION_NAME): Don't emit
543 .linkonce directive after the first time.
545 Sat Aug 29 12:39:56 1998 Jeffrey A Law (law@cygnus.com)
547 * m68k.md (beq0_di): Generate correct (and more efficient) code when
548 the clobbered operand overlaps with an input.
549 (bne0_di): Similarly.
551 * Makefile.in (INSTALL): Remove "--no-header" argument.
553 * NEWS: Various updates.
555 Fri Aug 28 19:00:44 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
557 * config/sparc/sparc.c (arith_operand, const64_operand,
558 const64_high_operand, arith_double_4096_operand): Mark mode as
560 (create_simple_focus_bits): Remove unused arg highest_bit_set, all
562 (sparc_emit_set_const64): Remove unused variable i.
563 (sparc_splitdi_legitimate): Likewise for addr_part.
564 (ultra_code_from_mask): Likewise for mask.
565 (ultra_cmove_results_ready_p): Fixup entry modulo calc. and
566 reverse return values so it matches usage and comments.
567 (ultra_flush_pipeline): Likewise.
568 (ultra_fpmode_conflict_exists): Likewise, remove unused variable
569 this_type, and allow loads and stores of differing FP modes as
570 they do not create a conflict.
571 (ultra_find_type): Initialize fpmode to SFmode, fix
572 parenthesization thinkos in large conditional.
573 (ultrasparc_sched_init): Mark dump and sched_verbose as unused.
574 Init free_slot_mask after ultra_cur_hist is reset, not before.
575 (ultrasparc_rescan_pipeline_state): Remove unused variable ucode.
576 (ultrasparc_sched_reorder): Don't bzero current pipeline state,
577 use ultra_flush_pipeline instead, then re-init group pointer.
578 Fix statement with no effect. If no progress made in, and no
579 instructions scheduled at all, advance to new pipeline cycle else
580 we get into an endless loop.
581 (ultrasparc_adjust_cost): Remove previous arg.
582 * config/sparc/sparc.h (ADJUST_COST): Update to reflect that.
584 Fri Aug 28 13:52:35 1998 Jim Wilson <wilson@cygnus.com>
586 * sparc.md (DImode, DFmode, TFmode splits): Delete self_reference
587 code. Use reg_overlap_mentioned_p to detect when source and
589 (negtf2_notv9+1): Use DFmode instead of SFmode in last two operands.
591 1998-08-28 Brendan Kehoe <brendan@cygnus.com>
593 * loop.c (check_dbra_loop): Pass COMPARISON_VALUE, not
594 COMPARISON_VAL, into invariant_p.
596 Fri Aug 28 15:13:25 1998 J"orn Rennecke <amylaar@cygnus.co.uk>
598 * regmove.c (regclass_compatible_p): New function.
599 (regmove_optimize): Use it.
601 Use NREGS parameter insted of calling max_reg_num.
603 (fixup_match_1): Don't use code = MINUS when later tieing with
604 a hard register is likely.
606 Fri Aug 28 14:54:07 1998 J"orn Rennecke <amylaar@cygnus.co.uk>
608 * loop.c (check_dbra_loop): Fix calculation of FINAL_VALUE when
609 COMPARISON_VAL was normalized.
611 Thu Aug 27 20:10:46 1998 Jeffrey A Law (law@cygnus.com)
613 * loop.c (check_dbra_loop): The loop ending comparison value
614 must be an invariant or we can not reverse the loop.
616 * loop.c (scan_loop): Count down from max_reg_num - 1 to
617 FIRST_PSEUDO_REGISTER to avoid calling max_reg_num each iteration
619 (load_mems_and_recount_loop_regs_set): Likewise.
621 * i386.c (print_operand): Remove obsolete 'c' docs.
623 Wed Aug 26 17:13:37 1998 Tom Tromey <tromey@cygnus.com>
625 * gthr.h: Document __GTHREAD_MUTEX_INIT_FUNCTION.
626 * frame.c (init_object_mutex): New function.
627 (init_object_mutex_once): Likewise.
629 (__register_frame_info): Likewise.
630 (__register_frame_info_table): Likewise.
631 (__deregister_frame_info): Likewise.
633 Thu Aug 27 15:14:18 1998 Jeffrey A Law (law@cygnus.com)
635 * haifa-sched.c (sched_analyze_insn): Fix thinko in last change.
637 Thu Aug 27 16:34:51 1998 J"orn Rennecke <amylaar@cygnus.co.uk>
639 * loop.c (check_dbra_loop): Enable code for reversal
640 of some loops without a known constant loop end.
642 Wed Aug 26 18:38:15 1998 Richard Henderson <rth@cygnus.com>
644 * haifa-sched.c (last_clock_var): New.
645 (schedule_block): Initialize it.
646 (schedule_insn): Use it to fill insn modes with issue information.
648 * alpha.c (alpha_handle_trap_shadows): Remove do-nothing exit.
649 Tag trapb and next insn with TImode.
650 (alphaev5_insn_pipe, alphaev5_next_group, alphaev5_align_insns): New.
651 (alpha_reorg): Add conditional for alpha_handle_trap_shadows.
652 Invoke alphaev5_align_insns as appropriate.
653 * alpha.h (LABEL_ALIGN_AFTER_BARRIER): Was ALIGN_LABEL_AFTER_BARRIER.
654 (MD_SCHED_VARIABLE_ISSUE): New.
655 * alpha.md (attr type): Add multi.
656 (define_asm_attributes): New.
657 (prologue_stack_probe_loop, builtin_setjmp_receiver): Set type multi.
658 (arg_home): Likewise.
659 (fnop, unop, realign): New.
661 Wed Aug 26 15:55:41 1998 Jim Wilson <wilson@cygnus.com>
663 * iris5.h (PREFERRED_DEBUGGING_TYPE): Undef.
664 * iris5gas.h (PREFERRED_DEBUGGING_TYPE): Define.
666 * configure.in (powerpc-ibm-aix4.[12]*): Change from 4.[12].*.
667 (rs6000-ibm-aix4.[12]*): Likewise.
668 * configure: Regnerate.
670 Wed Aug 26 09:30:59 1998 Nick Clifton <nickc@cygnus.com>
672 * config/arm/thumb.c (thumb_exit): Do not move a4 into lr if it
673 already contains the return address.
675 Wed Aug 26 12:57:09 1998 Jeffrey A Law (law@cygnus.com)
677 * calls.c (expand_call): Use bitfield instructions to extract/deposit
678 word sized hunks when loading unaligned args into registers.
680 * haifa-sched.c (sched_analyze_insn): Only create scheduling
681 barriers for LOOP, EH and SETJMP notes on the loop_notes list.
683 * mn10300.h (RTX_COSTS): Handle UDIV and UMOD too.
685 Wed Aug 26 16:35:37 1998 J"orn Rennecke <amylaar@cygnus.co.uk>
687 * loop.c (check_dbra_loop): Add some code that would allow reversal
688 of some loops without a known constant loop end if it were enabled.
690 Wed Aug 26 11:08:44 1998 Gavin Romig-Koch <gavin@cygnus.com>
692 * mips.md (lshrsi3_internal2+2): Fix type-o.
694 Wed Aug 26 10:53:03 1998 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
696 * system.h: Include stdarg.h/varargs.h, make sure they are ordered
697 correctly with regards to stdio.h.
699 * calls.c: Remove stdarg.h/varargs.h.
702 * combine.c: Likewise.
703 * cpperror.c: Likewise.
704 * cpplib.c: Likewise.
705 * cpplib.h: Likewise.
706 * doprint.c: Likewise.
707 * emit-rtl.c: Likewise.
709 * fix-header.c: Likewise.
711 * genattr.c: Likewise.
712 * genattrtab.c: Likewise.
713 * gencodes.c: Likewise.
714 * genconfig.c: Likewise.
715 * genemit.c: Likewise.
716 * genextract.c: Likewise.
717 * genflags.c: Likewise.
718 * genopinit.c: Likewise.
719 * genoutput.c: Likewise.
720 * genpeep.c: Likewise.
721 * genrecog.c: Likewise.
722 * mips-tfile.c: Likewise.
723 * prefix.c: Likewise.
724 * protoize.c: Likewise.
725 * regmove.c: Likewise.
726 * toplev.c: Likewise.
729 Wed Aug 26 05:09:27 1998 Jakub Jelinek <jj@sunsite.ms.mff.cuni.cz>
731 * config/sparc/sparc.c (sparc_override_options): If not
732 TARGET_FPU, turn off TARGET_VIS.
733 * config/sparc/sparc.h (TARGET_SWITCHES): Add no-vis.
734 (LEGITIMATE_CONSTANT_P): Allow SF/DF mode zero when TARGET_VIS.
735 * config/sparc/sparc.md (movsi_insn): Use fzeros not fzero.
736 (movdi_insn_sp64): Add VIS fzero alternative.
737 (clear_sf, clear_df): New VIS patterns.
738 (movsf, movdf expanders): Allow fp_zero_operand flat out when
740 (one_cmpldi2_sp64): Provide new fnot1 VIS alternative.
742 Tue Aug 25 10:57:41 1998 Mark Mitchell <mark@markmitchell.com>
744 * loop.c (n_times_set, n_times_used, may_not_optimize,
745 reg_single_usage): Convert to varrays. All uses changed.
746 (insert_loop_mem): Return a value.
747 (scan_loop): Tweak AVOID_CC_MODE_COPIES code.
748 (load_mems_and_recount_loop_regs_set): Likewise. Grow the arrays, if
751 Tue Aug 25 23:57:12 1998 Jeffrey A Law (law@cygnus.com)
754 * configure.in: Do not set thread_file to "irix" since no such
757 * sparc.md (float abs/neg splits): Check reload_completed before
758 calling alter_subreg.
760 Tue Aug 25 19:17:59 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
762 * config/sparc/sparc.c (sparc_absnegfloat_split_legitimate): New
764 * config/sparc/sparc.h: Declare it.
765 * config/sparc/sparc.md (float abs/neg splits): Use it.
766 (all other splits): Handle SUBREGs properly where necessary.
767 (unnamed (1<<x)-1 V8PLUS pattern): Disable for now.
769 Tue Aug 25 19:48:46 1998 Jeffrey A Law (law@cygnus.com)
771 * reorg.c (fill_simple_delay_slots): Do not abort if we encounter
772 an insn on the unfilled_slots_list that has no delay slots.
773 (fill_eager_delay_slots): Similarly.
775 Tue Aug 25 13:35:20 1998 Nick Clifton <nickc@cygnus.com>
777 * config/v850/v850.c (movsi_source_operand): Treat CONSTANT_P_RTX
778 as an ordinary operand.
780 Tue Aug 25 12:54:57 1998 Jason Merrill <jason@yorick.cygnus.com>
782 * tree.c (valid_machine_attribute): Don't apply attributes to both
785 Tue Aug 25 12:23:20 PDT 1998 Richard Henderson <rth@cygnus.com>
787 * reload.c (operands_match_p): Handle rtvecs.
789 * i386.c (legitimate_pic_address_disp_p): New.
790 (legitimate_address_p): Use it.
791 (legitimize_pic_address): Use unspecs to represent @GOT and @GOTOFF.
792 Handle constant pool symbols just like statics.
793 (emit_pic_move): Use Pmode not SImode for clarity.
794 (output_pic_addr_const) [SYMBOL_REF]: Remove @GOT and @GOTOFF hacks.
795 [UNSPEC]: New, handling what we killed above.
796 [PLUS]: Detect and abort on invalid symbol arithmetic.
797 * i386.h (CONSTANT_ADDRESS_P): Remove HIGH.
799 Tue Aug 25 12:02:23 1998 Mark Mitchell <mark@markmitchell.com>
801 * alias.c: Include output.h.
802 (DIFFERENT_ALIAS_SETS_P): Don't treat alias sets as
803 different if we're in a varargs function.
804 * Makefile.in (alias.o): Depend on output.h
806 Tue Aug 25 19:20:12 1998 J"orn Rennecke <amylaar@cygnus.co.uk>
808 * sh.h (GIV_SORT_CRITERION): Delete.
810 Tue Aug 25 13:19:46 1998 Dave Brolley <brolley@cygnus.com>
812 * regclass.c (regclass): Use xmalloc/free instead of alloca.
813 * stupid.c (stupid_life_analysis): Ditto.
814 * reload1.c (reload): Ditto.
816 Tue Aug 25 05:48:18 1998 Jakub Jelinek <jj@sunsite.ms.mff.cuni.cz>
818 * config/sparc/sparc.c (arith_4096_operand, arith_add_operand,
819 arith_double_4096_operand, arith_double_add_operand): New
821 * config/sparc/sparc.h (PREDICATE_CODES): Add them, declare them.
822 * config/sparc/sparc.md (adddi3, addsi3, subdi3, subsi3): Use
823 them to transform add/sub 4096 into add/sub -4096.
825 Mon Aug 24 23:31:03 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
827 * loop.c (scan_loop): Allocate some slop to handle pseudos
828 generated by move_movables.
829 (load_mems_and_recount_loop_regs_set): Honor AVOID_CC_MODE_COPIES
832 Mon Aug 24 19:45:40 1998 Jim Wilson <wilson@cygnus.com>
834 * tree.def (DECL_RESULT): Correct documentation.
836 Tue Aug 25 01:15:27 1998 J"orn Rennecke <amylaar@cygnus.co.uk>
838 * reload1.c (reload_reg_free_before_p): New argument EQUIV; Changed
839 all callers. Abort for RELOAD_FOR_INSN. RELOAD_FOR_OUTADDR_ADDR:
840 conflicts will all RELOAD_FOR_OUTPUT reloads.
842 * reload1.c (reload_cse_regs_1): When deleting a no-op move that
843 loads the function result, substitute with a USE.
845 Mon Aug 24 15:20:19 1998 David Edelsohn <edelsohn@mhpcc.edu>
847 * rs6000.h (GO_IF_LEGITIMATE_ADDRESS): Use TARGET_POWERPC64
848 when testing LEGITIMATE_INDEXED_ADDRESS_P DFmode and DImode.
849 (LEGITIMIZE_ADDRESS): Use TARGET_POWERPC64 for INDEXED fixup.
850 * rs6000.c (print_operand, case 'L'): Add UNITS_PER_WORD, not 4.
851 (print_operand, cases 'O' and 'T'): Fix typos in lossage strings.
852 * rs6000.md (fix_truncdfsi2_store): Remove %w from non-CONST_INT
854 (movdf_softfloat32, movdf_hardfloat64, movdf_softfloat64): Change
855 'o' to 'm' for GPR variant constraints.
857 Mon Aug 24 10:25:46 1998 Jeffrey A Law (law@cygnus.com)
859 * loop.c (scan_loop): Honor AVOID_CC_MODE_COPIES.
861 * h8300.h (STRIP_NAME_ENCODING): Fix typo.
863 * sparc.md (TFmode splits): Use reg_overlap_mentioned_p to detect
864 when the source and destination overlap.
866 * stmt.c (emit_case_nodes): Change rtx_function to rtx_fn to avoid
867 clash with global type.
869 Mon Aug 24 00:53:53 1998 Jason Merrill <jason@yorick.cygnus.com>
871 * fixinc.irix: Add curses.h handling from fixinc.wrap.
873 * c-common.c (combine_strings): Also set TREE_READONLY.
874 Change warn_write_strings to flag_const_strings.
875 * c-decl.c, c-tree.h: Likewise.
877 Sun Aug 23 18:39:11 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
879 * config/sparc/sparc.c (sparc_emit_set_const32): If outputting a
880 CONST_INT, not a symbolic reference, don't use a HIGH/LO_SUM
881 sequence, use SET/IOR instead so CSE can see it.
882 * config/sparc/sparc.md (movhi_const64_special,
883 movsi_const64_special): New patterns necessitated by that change.
884 (movhi_high): Remove.
885 (movhi_lo_sum): Change to match an IOR.
886 (movdf_insn_sp32): Test TARGET_V9 not TARGET_ARCH64.
887 (movdf_insn_v9only): New pattern for when V9 but not ARCH64.
888 (movdf_insn_sp64): Test both TARGET_V9 and TARGET_ARCH64.
889 (movdf splits): Allow when not V9 or when not ARCH64 and integer
890 registers are involved.
891 (snesi_zero_extend split): Remove reload_completed test.
892 (unnamed plus and minus zero_extend sidi splits): Add it.
894 Sun Aug 23 11:56:08 1998 Mark Mitchell <mark@markmitchell.com>
896 * extend.texi: Remove description of extension to explicit
897 instantiation that is now endorsed by standard C++.
899 Sun Aug 23 09:39:09 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
901 * config/arc/arc.c (arc_initialize_pic): Remove.
902 * config/arc/arc.h (INITIALIZE_PIC): Similarly, this routine does
903 nothing on any platform and is invoked by no-one, it does not even
904 appear in the documentation.
905 * config/sparc/sparc.h (INITIALIZE_PIC): Likewise.
906 * config/sparc/sparc.c (initialize_pic): Likewise.
907 (find_addr_reg): Remove this as well, no longer referenced after
910 Sun Aug 23 00:17:14 1998 Jeffrey A Law (law@cygnus.com)
912 * recog.c (validate_replace_rtx_group): New function.
913 * recog.h (validate_replace_rtx_group): Declare it.
914 * regmove.c (optimize_reg_copy_3): If any substitution fails, then undo
915 the entire group of substitutions.
917 Sat Aug 22 23:31:00 1998 Klaus-Georg Adams (Klaus-Georg.Adams@chemie.uni-karlsruhe.de)
919 * loop.c (load_mems): Fix initializers.
921 Fri Aug 21 23:07:46 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
923 * config/sparc/sparc.md (TFmode splits): Handle destination
924 registers being referenced in the address correctly.
926 * expmed.c (make_tree) [CONST_INT]: Sign extend even if
927 TREE_UNSIGNED, when bitsize of type's mode is larger than
928 HOST_BITS_PER_WIDE_INT.
930 Fri Aug 21 19:31:31 1998 Alexandre Petit-Bianco <apbianco@cygnus.com>
932 * tree.def (LABELED_BLOCK_EXPR, EXIT_BLOCK_EXPR): New tree nodes.
933 * tree.h (LABELED_BLOCK_LABEL, LABELED_BLOCK_BODY,
934 EXIT_BLOCK_LABELED_BLOCK, EXIT_BLOCK_RETURN, LOOP_EXPR_BODY): New
936 * expr.c (expand_expr): Handle LABELED_BLOCK_EXPR and
939 Thu Aug 20 19:43:44 1998 Jeffrey A Law (law@cygnus.com)
941 * h8300.c (h8300_encode_label): Use '&' for tiny data items.
942 * h8300.h (TINY_DATA_NAME_P): Likewise.
943 (STRIP_NAME_ENCODING): Handle '&'.
945 * mn10200.h (REG_OK_FOR_INDEX_P): Do not check the mode of the
946 register (it could be accessed via an outer SUBREG).
947 (REG_OK_FOR_BASE_P): Likewwise.
948 (GO_IF_LEGITIMATE_ADDRESS): Consistently use REGNO_OK_FOR_BASE_P.
950 * remove.c (optimize_reg_copy_3): Abort instead of silently generating
953 * jump.c (rtx_renumbered_equal_p): Do not consider PLUS commutative.
955 Thu Aug 20 17:35:20 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
957 * config/sparc/sparc.md (movtf_insn_sp32): All memory operands
958 must be offsettable so the splits can be made.
960 Thu Aug 20 13:56:53 1998 Michael Meissner <meissner@cygnus.com>
962 * config/i386/winnt.c: Include system.h, not stdio.h to get
963 sys/param.h pulled in before rtl.h in case the system defines MIN
966 Thu Aug 20 13:44:20 1998 David Edelsohn <edelsohn@mhpcc.edu>
968 * rs6000.md (movqi, movhi): Add CONSTANT_P_RTX.
970 Thu Aug 20 13:15:11 1998 Dave Brolley <brolley@cygnus.com>
972 * stor-layout.c (layout_type): Compute TYPE_SIZE_UNIT correctly for
974 * cpplib.c (cpp_define): Handle macros with parameters.
976 Wed Aug 19 21:33:19 1998 David Edelsohn <edelsohn@mhpcc.edu>
978 * rs6000.c (rs6000_output_load_toc_table): Use ld for 64-bit.
979 (output_toc): Use single TOC slot or llong minimal-toc for DFmode
980 and DImode 64-bit. Use llong for minimal-toc SFmode and
981 SYMBOL_REF / LABEL_REF 64-bit.
982 (output_function_profiler): Use llong for profiler label and ld to
983 load 64-bit label address.
985 Wed Aug 19 17:52:27 1998 Nick Clifton (nickc@cygnus.com)
987 * config/arm/thumb.md (extendqisi2_insn): Cope with REG +
990 Wed Aug 19 14:13:31 PDT 1998 Jeff Law (law@cygnus.com)
992 * version.c: Bump for snapshot.
994 Wed Aug 19 13:10:30 PDT 1998 Jeff Law (law@cygnus.com)
996 * version.c: Bump for snapshot.
998 Wed Aug 19 13:06:47 1998 Jason Merrill <jason@yorick.cygnus.com>
1000 * collect2.c (extract_init_priority): Use atoi instead of strtoul.
1002 Wed Aug 19 13:51:35 1998 Hans-Peter Nilsson <hp@axis.se>
1004 * tm.texi (Misc): Fix typo "teh".
1006 * tm.texi (PIC): Fix typo "PPIC".
1008 * tm.texi (Caller Saves): Say that DEFAULT_CALLER_SAVES has no
1009 effect when -O2 and higher.
1010 * invoke.texi (Optimize Options): Likewise for -fcaller-saves.
1012 1998-08-19 Michael Hayes <michaelh@ongaonga.chch.cri.nz>
1014 * regclass.c: Changed register set documentation to be consistent
1017 * final.c (final_start_function) Removed redundant test for
1020 Wed Aug 19 13:28:41 1998 Mark Mitchell <mark@markmitchell.com>
1022 * rtl.h (rtx_function): New type.
1023 (for_each_rtx): New function.
1024 * rtlanal.c (for_each_rtx): Define it.
1026 * recog.c (change_t): New type.
1027 (change_objects, change_old_codes, change_locs, change_olds):
1029 (changes): New variable.
1030 (validate_change): Dynamically allocate room for more changes, if
1031 necessary. Uses changes array instead of change_objects, etc.
1032 (apply_change_group): Use changes array instead of
1033 change_objects, etc.
1035 * loop.c (loop_mem_info): New type.
1036 (loop_mems): New variable.
1037 (loop_mems_idx): Likewise.
1038 (looop_mems_allocated): Likewise.
1039 (scan_loop): Remove nregs parameter.
1040 (next_insn_in_loop): New function.
1041 (load_mems_and_recount_loop_regs_set): Likewise.
1042 (load_mems): Likewise.
1043 (insert_loop_mem): Likewise.
1044 (replace_loop_mem): Likewise.
1045 (replace_label): Likewise.
1046 (INSN_IN_RANGE_P): New macro.
1047 (loop_optimize): Don't pass max_reg_num() to scan_loop.
1048 (scan_loop): Remove nregs parameter, compute it after any new
1049 registers are created by load_mems. Use INSN_IN_RANGE_P and
1050 next_insn_in_loop rather than expanding them inline. Call
1051 load_mems to load memory into pseudos, if appropriate.
1052 (prescan_loop): Figure out whether or not there are jumps from the
1053 loop to targets other than the label immediately following the
1054 loop. Call insert_loop_mem to notice all the MEMs used in the
1055 loop, if it could be safe to pull MEMs into REGs for the duration
1057 (strength_reduce): Use next_insn_in_loop. Tweak comments.
1059 Wed Aug 19 08:29:44 1998 Richard Earnshaw (rearnsha@arm.com)
1061 * arm.c (arm_override_options): Remove lie about ignoring PIC flag.
1063 Wed Aug 19 07:08:15 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
1065 * config/sparc/sparc.c (finalize_pic): Check for the correct
1066 nonlocal_goto_receiver UNSPEC number.
1067 * config/sparc/sparc.md (nonlocal_goto_receiver): Add comment
1068 making note of this dependency existing in sparc.c
1069 (negtf2_notv9 split): Give NEG SFmode.
1070 (negsf2): Fix insn output string.
1072 Tue Aug 18 12:40:27 1998 Richard Henderson <rth@cygnus.com>
1074 * c-common.c (decl_attributes): Issue an error if the argument
1075 to alias is not a string.
1077 Tue Aug 18 10:33:30 1998 Jeffrey A Law (law@cygnus.com)
1079 * haifa-sched.c (sched_analyze): Put all JUMP_INSNs on the last
1080 pending memory flush list.
1082 * combine.c (can_combine_p): Allow combining insns with REG_RETVAL
1084 (try_combine): Allow combining insns with REG_LIBCALL notes.
1086 * expr.c (emit_block_move): Do not call memcpy as a libcall
1087 instead build up a CALL_EXPR and call it like any other
1089 (clear_storage): Similarly for memset.
1091 * regmove.c (fixup_match_2): Do not call reg_overlap_mentioned_p
1094 * Makefile.in (cplus-dem.o): Provide explicit rules for building
1097 * regmove.c (optimize_reg_copy_1): Update REG_N_CALLS_CROSSED
1098 and REG_LIVE_LENGTH as successful substitutions are made.
1100 Tue Aug 18 07:15:27 1998 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
1102 * config/sparc/sparc.c (ultra_find_type): Add empty semicolon
1103 statement after end of loop label.
1105 Tue Aug 18 07:13:27 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
1107 * config/sparc/sparc.c (ultra_types_avail): New variable.
1108 (ultra_build_types_avail): New function to record mask of insn
1109 types in ready list at this cycle.
1110 (ultrasparc_sched_reorder): Call it.
1111 (ultra_find_type): Use it to quicken the search. Also simplif
1112 dependency check, don't use rtx_equal_p because we know exactly
1113 what we are looking for.
1115 Tue Aug 18 03:20:53 1998 Richard Earnshaw (rearnsha@arm.com)
1117 * arm.h (SECONDARY_INPUT_RELOAD_CLASS): Return NO_REGS if compiling
1118 for architecture v4.
1120 Mon Aug 17 21:26:38 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
1122 * config/sparc/sparc.md (sltu, sgeu): Don't FAIL, call
1124 (movsf_const_intreg, movsf_const_high, movsf_const_lo,
1125 movdf_const_intreg and helper splits): New patterns to move float
1126 constants into integer registers.
1127 (negtf2, negdf2, abstf2, absdf2): Rework using new patterns and
1130 Mon Aug 17 11:46:19 1998 Jeffrey A Law (law@cygnus.com)
1133 * tree.c (build_index_type): Copy TYPE_SIZE_UNIT from sizetype
1135 * c-decl.c (finish_enum): Copy TYPE_SIZ_UNIT from enumtype to tem.
1137 * rs6000.c (secondary_reload_class): For TARGET_ELF, indicate that
1138 a BASE_REGS register is needed as an intermediate when copying
1139 a symbolic value into any register class other than BASE_REGS.
1141 * expr.c (move_by_pieces): No longer static. Remove prototype.
1142 * rtl.h (move_by_pieces): Add extern prototype.
1143 * mips.c (expand_block_move): Handle aligned straight line copy by
1144 calling move_by_pieces.
1146 * expr.c (expand_expr): Allow assignments from TImode PARM_DECLs
1149 Mon Aug 17 10:28:52 1998 Mark Mitchell <mark@markmitchell.com>
1151 * stmt.c (expand_end_loop): Tidy. Allow unconditional
1152 jumps out of the loop to be treated as part of the exit test.
1154 Mon Aug 17 10:06:11 1998 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
1155 Jeff Law <law@cygnus.com>
1157 * Makefile.in (cplus-dep.o): Use cplus-dem.c from libiberty.
1158 * cplus-dem.c: Delete.
1160 * Makefile.in (fold-const.o): depend on $(RTL_H).
1162 * fold-const.c: Include rtl.h to get the prototype for
1163 `set_identifier_local_value'.
1165 * loop.c (express_from_1): Remove unused variable `tmp'.
1166 (combine_givs): cast the first argument of bzero to char *.
1168 * toplev.c (display_help): Remove unused variable `looking_for_start'.
1170 * c-decl.c (init_decl_processing): Remove unneeded &.
1172 * alpha.h (alpha_initialize_trampoline): Provide prototype.
1174 * except.c (set_exception_lang_code, set_exception_version_code):
1175 Change parameter from `short' to `int' to avoid using a gcc
1178 * except.h (set_exception_lang_code, set_exception_version_code):
1179 Likewise for prototypes.
1181 * flow.c (count_reg_references): Remove unused variables `regno'
1184 * gcse.c (hash_scan_insn): Declare parameter `in_libcall_block'.
1186 * prefix.c (translate_name): Cast the result of `alloca'.
1188 * varray.h (VARRAY_FREE): Reimplement as a `do-while(0)' statement.
1190 Mon Aug 17 09:23:42 1998 Andreas Schwab <schwab@issan.informatik.uni-dortmund.de>
1192 * config/m68k/m68k.c: Include "system.h" instead of <stdio.h>.
1194 (valid_dbcc_comparison_p): Mark mode argument as unused.
1195 (symbolic_operand): Likewise.
1196 (legitimize_pic_address): Likewise.
1197 (const_uint32_operand): Likewise.
1198 (const_sint32_operand): Likewise.
1199 * sched.c [!INSN_SCHEDULING]: Define only dummy function
1200 schedule_insns and comment out rest of file.
1202 * m68k.c (output_move_simode_const): Use subl to move a zero into an
1204 (output_move_[hq]imode): Likewise.
1206 Mon Aug 17 09:15:47 1998 Jeffrey A Law (law@cygnus.com)
1208 * toplev.c (main): Enable -fstrict-aliasing for -O2 and above.
1209 * invoke.texi: Corresponding changes.
1211 Mon Aug 17 02:03:55 1998 Richard Henderson <rth@cygnus.com>
1213 * regclass.c (allocate_reg_info): Respect MIN when clearing data.
1215 Sun Aug 16 17:37:06 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
1217 * config/sparc/sparc.c (ultra_code_from_mask,
1218 ultra_cmove_results_ready_p, ultra_fpmode_conflict_exists,
1219 ultra_find_type, ultra_schedule_insn, ultra_flush_pipeline,
1220 ultrasparc_sched_init, ultrasparc_variable_issue,
1221 ultra_rescan_pipeline_state, ultrasparc_sched_reorder): New
1222 functions to describe UltraSPARC pipeline exactly to Haifa.
1223 (ultrasparc_adjust_cost): Indicate IMUL type insns have zero cost,
1224 as there is nothing the scheduler can do about it. Indicate that
1225 REG_DEP_OUTPUT's collide. Fixup formatting.
1226 * config/sparc/sparc.h (RTX_COSTS): Fixup integer multiply and
1227 divide costs on Ultra for DImode.
1228 (MD_SCHED_INIT, MD_SCHED_REORDER, MD_SCHED_VARIABLE_ISSUE):
1230 * config/sparc/sparc.md (ieu_unnamed function unit): Rename to
1231 ieuN and add call_no_delay_slot to type list.
1232 (cti function unit): New unit for branches on UltraSPARC.
1233 (subx/addx insns): Set type to misc.
1234 (sidi zero/sign extension insns on arch64): Set type to shift.
1235 (sign_extendhidi2_insn): Set type to sload.
1237 Sun Aug 16 13:52:00 1998 David Edelsohn <edelsohn@mhpcc.edu>
1239 * rs6000.c (rs6000_stack_info): Use if == 0 for sizes.
1240 (output_epilog): Use if != 0 for offset.
1241 (rs6000_fatal_bad_address): Prepare for Intl.
1242 * rs6000.h (rs6000_fatal_bad_address): Declare.
1243 * rs6000.md (movsfcc, movdfcc): Use else if.
1244 (elf_high): Use {liu|lis}.
1245 (elf_low): Use {cal|la}. Remove %a template from old mnemonics.
1246 (movsi): Use rs6000_fatal_bad_address.
1248 Sun Aug 16 01:53:21 1998 Richard Henderson <rth@cygnus.com>
1250 * reload.c (find_equiv_reg): Reject equivalences separated
1251 by a volatile instruction.
1253 Sun Aug 16 00:21:44 1998 Franz Sirl <Franz.Sirl-kernel@lauterbach.com>
1255 * rs6000/linux.h (CPP_OS_DEFAULT_SPEC): Define.
1257 Sat Aug 15 20:51:35 1998 Richard Henderson <rth@cygnus.com>
1259 * alpha.md (movsicc): Fix mode mismatch.
1261 Sat Aug 15 20:22:33 1998 H.J. Lu (hjl@gnu.org)
1263 * config/alpha/alpha.h (ASM_OUTPUT_MI_THUNK): Handle aggregated
1265 * config/alpha/win-nt.h (ASM_OUTPUT_MI_THUNK): Likewise.
1267 Sat Aug 15 08:39:49 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
1269 * config/sparc/sparc.md (movsi_lo_sum_pic_label_reg): Remove
1270 write-only modifier from operand 1 constraint.
1272 Sat Aug 15 06:28:19 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
1274 * config/sparc/sparc.c (sparc_emit_set_const64_quick1): If
1275 emitting a XOR of -1 at the end, emit a NOT instead for combine's
1277 (sparc_emit_set_const64): Likewise, also when computing trailing
1278 bits do not negate low_bits and make fast_int an int.
1280 Fri Aug 14 21:07:03 1998 Jeffrey A Law (law@cygnus.com)
1282 * loop.c (add_label_notes): Do not ignore references to labels
1283 before dispatch tables. Mirrors Apr 8 change to mark_jump_label.
1284 * gcse.c (add_label_notes): Similarly.
1286 * pa.h (ASM_OUTPUT_MI_THUNK): Strip name encoding.
1288 * m68k.md (adddi_dilshr32): One of the operands must be a register.
1289 (adddi_dishl32): Similarly.
1291 Fri Aug 14 14:12:59 1998 Jason Merrill <jason@yorick.cygnus.com>
1293 * i386.h (MODES_TIEABLE_P): Reorganize to shut up warnings.
1294 * alias.c (memrefs_conflict_p): Add braces to shut up warnings.
1295 * cse.c (cse_basic_block): Add parens to shut up warnings.
1297 Fri Aug 14 12:58:21 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
1299 * config/sparc/sparc.c (sparc_emit_set_const64_quick2,
1300 sparc_emit_set_const64_longway, const64_is_2insns,
1301 create_simple_focus_bits, sparc_emit_set_const64): Fix more bugs
1302 in 64-bit constant formation.
1303 * config/sparc/sparc.md (snesi_zero_extend split): Generate
1304 rtl for addx not subx.
1305 (define_insn movdi_const64_special): Make available even when
1306 HOST_BITS_PER_WIDE_INT is not 64.
1307 (movdi_lo_sum_sp64_cint, movdi_high_sp64_cint): Remove.
1308 (losum_di_medlow, sethm, setlo): Make op2 symbolic_operand.
1309 (cmp_siqi_trunc_set, cmp_diqi_trunc_set): Encapsulate both
1310 instances of operand 1 inside a QI subreg.
1311 (xordi3_sp64_dbl): Remove '%' constraint for op1.
1312 (one_cmpldi2_sp64): Fix output string.
1313 (one_cmplsi2_not_liveg0): Rewrite to remove unneeded extra
1315 (unnamed arch64 ashift DI): Truncate shift count if greater than
1318 Fri Aug 14 21:52:53 1998 J"orn Rennecke <amylaar@cygnus.co.uk>
1320 * expr.c (store_expr): Don't optimize away load-store pair
1321 when either source or destination have a side effect.
1323 Fri Aug 14 16:50:10 1998 John Carr <jfc@mit.edu>
1325 * genrecog.c (add_to_sequence): Fatal error if the modes of the
1326 operands of SET are incompatible.
1328 * alpha.md: Fix max and min patterns so modes of SET operands match.
1330 Fri Aug 14 12:22:55 1998 Ian Lance Taylor <ian@cygnus.com>
1332 * configure.in: Avoid [[ by using test and changequote.
1333 * configure: Rebuild.
1335 Fri Aug 14 01:22:31 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
1337 * rtl.def (CONSTANT_P_RTX): Fix typo in string name.
1339 * config/sparc/sparc.md (seqdi_special_trunc, snedi_special_trunc,
1340 seqsi_special_extend, snesi_special_extend, snesi_zero_extend and
1341 split, snedi_zero_trunc and split, seqsi_zero_extend and split,
1342 seqdi_zero_trunc and split, pic_lo_sum_di, pic_sethi_di,
1343 movdi_cc_sp64_trunc, movdi_cc_reg_sp64_trunc, addx_extend_sp32 and
1344 split, addx_extend_sp64, subx_extend_sp64, subx_extend and split):
1345 Fix mismatching modes in SET operands.
1346 (conditional move patterns): Fix formatting.
1347 (unnamed subx arch64 pattern): Remove duplicate insn.
1349 Fri Aug 14 00:34:34 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
1351 * config/sparc/sparc.c (const64_operand, const64_high_operand):
1352 Get it right when HOST_BITS_PER_WIDE_INT is not 64.
1353 (input_operand): Fixup test for what we accept for constant
1355 (sparc_emit_set_const32, sparc_emit_set_symbolic_const64): Give
1357 (safe_constDI): Remove.
1358 (sparc_emit_set_safe_HIGH64, gen_safe_SET64, gen_safe_OR64,
1359 gen_safe_XOR64): New functions.
1360 (sparc_emit_set_const64_quick1, sparc_emit_set_const64_quick2,
1361 sparc_emit_set_const64_longway, sparc_emit_set_const64): Use
1363 * config/sparc/sparc.md (define_insn xordi3_sp64_dbl): Only make
1364 available when HOST_BITS_PER_WIDE_INT is not 64.
1365 (define_insn movdi_sp64_dbl, movdi_const64_special): Likewise and
1366 move before movdi_insn_sp64 pattern.
1367 (define_insn movdi_lo_sum_sp64_dbl, movdi_high_sp64_dbl): Remove.
1368 (define_insn sethi_di_medlow, seth44, setm44, sethh): Use
1369 symbolic_operand as predicate for second operand.
1370 (DImode minus split on arch32, negsi2 expander, one_cmplsi2
1371 expander): Give set VOIDmode.
1373 Fri Aug 14 01:45:06 1998 Mumit Khan <khan@xraylith.wisc.edu>
1375 * i386/cygwin32 (DEFAULT_PCC_STRUCT_RETURN): Define.
1377 Fri Aug 14 01:40:21 1998 Geoffrey Keating <geoffk@ozemail.com.au>
1379 * rs6000/linux.h (LINK_SPEC): Pass -G args to the linker.
1381 Fri Aug 14 01:23:23 1998 Richard Earnshaw (rearnsha@arm.com)
1383 * arm/netbsd.h (TARGET_DEFAULT): Default includes software floating
1385 (CPP_FLOAT_DEFAULT_SPEC): Re-define accordingly.
1387 Fri Aug 14 01:19:08 1998 Robert Lipe <robertl@dgii.com>
1389 * install.texi: Various SCO OpenServer tweaks.
1391 Thu Aug 13 20:14:40 1998 Jim Wilson <wilson@cygnus.com>
1393 * reload1.c (eliminate_regs_in_insn): Handle another case when
1394 eliminating the frame pointer to the hard frame pointer. Add
1395 missing ep->to_rtx check to one existing case.
1397 * mips/mips.md (movhi_internal2+2): Fix typo mem:SI -> mem:HI.
1399 Thu Aug 13 17:08:11 1998 Jason Merrill <jason@yorick.cygnus.com>
1401 * tree.h: De-conditionalize init_priority code.
1403 * mips.h (NM_FLAGS): Change from -Bp to -Bn.
1404 * collect2.c (NM_FLAGS): Change from -p to -n.
1406 * configure.in: Turn on collect2 for mipstx39-elf.
1407 Handle use_collect2=no properly.
1409 * c-common.c: De-conditionalize init_priority code.
1410 * collect2.c (extract_init_priority, sort_ids): New fns.
1411 (main): Call sort_ids.
1412 Move sequence_number to file scope.
1414 * configure.in: Handle --enable-init-priority.
1415 * c-common.c (attrs): Add A_INIT_PRIORITY.
1416 (init_attributes, decl_attributes): Likewise.
1417 * tree.h (DEFAULT_INIT_PRIORITY, MAX_INIT_PRIORITY): New macros.
1418 * tree.c (get_file_function_name_long): Split out...
1419 (get_file_function_name): ...from here.
1421 Thu Aug 13 16:09:53 1998 Martin von Loewis <loewis@informatik.hu-berlin.de>
1423 * expr.c (safe_from_p): Change code to ERROR_MARK only when not
1426 Thu Aug 13 15:24:48 1998 Jason Merrill <jason@yorick.cygnus.com>
1428 * toplev.c (display_help): Add braces to shut up warnings.
1429 * tree.c (simple_cst_equal): Likewise.
1431 * fold-const.c (non_lvalue): Don't deal with null pointer
1433 (fold, case COMPOUND_EXPR): Wrap a constant 0 in a NOP_EXPR.
1435 * c-typeck.c (initializer_constant_valid_p): Allow conversion of 0
1436 of any size to a pointer.
1438 Thu Aug 13 12:53:13 1998 Jim Wilson <wilson@cygnus.com>
1440 * i386/winnt.c (i386_pe_asm_file_end): Check TREE_SYMBOL_REFERENCED.
1442 Wed Aug 12 17:25:18 1998 Jeffrey A Law (law@cygnus.com)
1444 * mn10300.c (REG_SAVE_BYTES): Only reserve space for registers
1445 which will be saved.
1446 * mn10300.md (prologue insn): Only save registers which need saving.
1447 (epilogue insn): Similarly.
1449 * mn10300.c, mn10300.h, mn10300.md: Remove "global zero register"
1452 Wed Aug 12 12:39:16 1998 Gavin Romig-Koch <gavin@cygnus.com>
1454 * mips/mips.h (ENCODE_SECTION_INFO): Set SYMBOL_REF_FLAG for
1455 VAR_DECL's in gp addressable sections.
1457 Tue Aug 11 23:02:31 1998 John Carr <jfc@mit.edu>
1459 * sparc.c: Change return <exp> to <exp>; return; in functions
1461 * sparc.md: Add empty semicolon statement after final label in
1464 Tue Aug 11 22:42:01 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
1466 * config/sparc/sparc.md (define_insn addx_extend): Rename to
1467 addx_extend_sp64, only allow when TARGET_ARCH64.
1468 (define_insn addx_extend_sp32 and split): Version that works when
1470 (define_insn subx_extend): Likewise.
1471 (define_split adddi3 and subdi3 with zero extension): Fixup and
1472 correct bugs when not TARGET_ARCH64.
1474 Tue Aug 11 16:04:34 1998 John Carr <jfc@mit.edu>
1476 * except.c (set_exception_lang_code, set_exception_version_code):
1477 Use prototype-style definition if __STDC__, to match declaration
1480 * genemit.c: Change FAIL and DONE macros not to use loops.
1482 Tue Aug 11 12:27:03 1998 Jim Wilson <wilson@cygnus.com>
1484 * dwarf2out.c (ASM_OUTPUT_DWARF_ADDR_CONST): Use
1485 ASM_OUTPUT_DWARF2_ADDR_CONST if defined.
1487 * mips/mips.md (reload_outsi): Use M16_REG_P when TARGET_MIPS16.
1489 Tue Aug 11 18:12:53 1998 Dave Love <d.love@dl.ac.uk>
1491 * README.g77: Update from Craig.
1493 Tue Aug 11 04:46:01 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
1495 * config/sparc/sparc.c (sparc_emit_set_const32): INTVAL is of
1497 (safe_constDI sparc_emit_set_const64_quick1,
1498 sparc_emit_set_const64_quick2, sparc_emit_set_const64_longway,
1499 analyze_64bit_constant, const64_is_2insns,
1500 create_simple_focus_bits): Fix some bugs when compiled on real
1502 (function_arg_record_value_3, function_arg_record_value_2,
1503 function_arg_record_value): Add fully prototyped forward decls.
1504 * config/sparc/sparc.md (define_insn cmpsi_insn_sp32): Rename back
1505 to cmpsi_insn and use on both 64 and 32 bit targets.
1506 (define_insn cmpsi_insn_sp64): Remove.
1507 (define_expand zero_extendsidi2): Allow for 32-bit target too.
1508 (define_insn zero_extendsidi2_insn): Rename to
1509 zero_extendsidi2_insn_sp64.
1510 (define_insn zero_extendsidi2_insn_sp32): New pattern and
1511 assosciated forced split for it.
1513 * config/sparc/sparc.c (const64_operand, const64_high_operand):
1515 * config/sparc/sparc.h: Declare them.
1516 (PREDICATE_CODES): Add them.
1517 * config/sparc/sparc.md (movdi_lo_sum_sp64_dbl,
1518 movdi_high_sp64_dbl, xordi3_sp64_dbl): Use them.
1520 Mon Aug 10 22:57:24 1998 John Carr <jfc@mit.edu>
1522 * config/sparc/sparc.md (define_insn jump): Output ba,pt not b,pt
1523 in v9 case as the latter makes the Solaris assembler crash.
1525 Mon Aug 10 22:39:09 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
1527 * config/sparc/sparc.c (input_operand): Do not accept a LO_SUM MEM
1528 for TFmode when !v9. We require offsettable memory addresses.
1529 * config/sparc/sparc.h (ALTER_HARD_SUBREG): Handle TFmode to
1530 DFmode register number conversions.
1531 * config/sparc/sparc.md (define_split DFmode moves): If register
1532 is a SUBREG do alter_subreg on it before using.
1533 (define_expand movtf): Fixup comment about alignment on v9.
1534 (define_split TFmode moves): Don't use gen_{high,low}part, create
1535 explicit SUBREGs instead.
1537 Mon Aug 10 19:02:55 1998 John Carr <jfc@mit.edu>
1539 * Makefile.in (mbchar.o): Depend on mbchar.c.
1541 Mon Aug 10 04:28:13 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
1542 Richard Henderson <rth@cygnus.com>
1544 Rewrite Sparc backend for better code generation and
1545 improved sparc64 support.
1546 * config/sparc/sp64-elf.h: Set JUMP_TABLES_IN_TEXT_SECTION to
1548 * config/sparc/sysv4.h: Likewise.
1549 * config/sparc/sparc.c (v8plus_regcmp_p, sparc_operand,
1550 move_operand, v8plus_regcmp_op, emit_move_sequence,
1551 singlemove_string, doublemove_string, mem_aligned_8,
1552 output_move_double, output_move_quad, output_fp_move_double,
1553 move_quad_direction, output_fp_move_quad, output_scc_insn):
1555 (small_int_or_double): New predicate.
1556 (gen_compare_reg): Remove TARGET_V8PLUS cmpdi_v8plus emission.
1557 (legitimize_pic_address): Emit movsi_{high,lo_sum}_pic instead of
1558 old pic_{sethi,lo_sum}_si patterns.
1559 (mem_min_alignment): New generic function to replace
1560 mem_aligned_8, which uses REGNO_POINTER_ALIGN information when
1561 available and can test for arbitrary alignments. All callers
1563 (save_regs, restore_regs, build_big_number,
1564 output_function_prologue, output_cbranch, output_return,
1565 sparc_flat_save_restore, sparc_flat_output_function_prologue,
1566 sparc_flat_output_function_epilogue): Prettify
1568 (output_function_epilogue): Likewise and add code to output
1569 deferred case vectors.
1570 (output_v9branch): Likewise, add new arg INSN and use it to tack
1571 on branch prediction settings. All callers changed.
1572 (print_operand): Likewise and output %l44 for LO_SUMs when
1574 (sparc_splitdi_legitimate): New function to make sure DImode
1575 splits can be run properly when !arch64.
1576 (sparc_initialize_trampoline, sparc64_initialize_trampoline):
1577 Reformat example code in comments.
1578 (set_extends): Remove UNSPEC/v8plus_clear_high case.
1579 (sparc_addr_diff_list, sparc_addr_list): New statics to keep track
1580 of deferred case vectors we need to output.
1581 (sparc_defer_case_vector): Record a case vector.
1582 (sparc_output_addr_vec, sparc_output_addr_diff_vec,
1583 sparc_output_deferred_case_vectors): New functions to output them.
1584 (sparc_emit_set_const32): New function to form 32-bit constants in
1585 registers when that requires more than one instruction.
1586 (safe_constDI, sparc_emit_set_const64_quick1,
1587 sparc_emit_set_const64_quick2, sparc_emit_set_const64_longway,
1588 analyze_64bit_constant, const64_is_2insns,
1589 create_simple_focus_bits, sparc_emit_set_const64): New functions
1590 which do the same for 64-bit constants when arch64.
1591 (sparc_emit_set_symbolic_const64): New function to emit address
1592 loading for all code models on v9.
1593 * config/sparc/sparc.h (CONDITIONAL_REGISTER_USAGE): Do not make
1594 %g1 fixed when arch64, unfix %g0 when TARGET_LIVE_G0.
1595 (ALTER_HARD_SUBREG): Fix thinko, return REGNO + 1 not 1.
1596 (SECONDARY_INPUT_RELOAD_CLASS, SECONDARY_OUTPUT_RELOAD_CLASS): Fix
1597 inaccuracies in comments, add symbolic and text_segment operands
1598 when TARGET_CM_MEDANY and TARGET_CM_EMBMEDANY respectively. Use
1599 GENERAL_REGS in these cases as a temp REG is needed to load these
1600 addresses into a register properly.
1601 (EXTRA_CONSTRAINT): Document more accurately, remove Q case as it
1603 (GO_IF_LEGITIMATE_ADDRESS): Allow TFmode for LO_SUM on v9 since fp
1604 quads are guarenteed to have 16-byte alignment.
1605 (LEGITIMIZE_ADDRESS): For SYMBOL_REF, CONST, and LABEL_REF use
1606 copy_to_suggested_reg instead of explicit LO_SUM and HIGH.
1607 (ASM_OUTPUT_ADDR_VEC, ASM_OUTPUT_ADDR_DIFF_VEC): New macros for
1608 deferred case vector implementation.
1609 (ASM_OUTPUT_ADDR_VEC_ELT): Use fputc to output newline.
1610 (ASM_OUTPUT_ADDR_DIFF_ELT): Parenthesize LABEL in macro calls.
1611 Generate "internal label - label" instead of "label - 1b".
1612 (PRINT_OPERAND_ADDRESS): For LO_SUM use %l44 on TARGET_CM_MEDMID.
1613 (PREDICATE_CODES): Remove sparc_operand, move_operand,
1614 v8plus_regcmp_op. Add small_int_or_double, input_operand, and
1616 (doublemove_string, output_block_move, output_fp_move_double,
1617 output_fp_move_quad, output_move_double, output_move_quad,
1618 output_scc_insn, singlemove_string, mem_aligned_8, move_operand,
1619 sparc_operand, v8plus_regcmp_op, v8plus_regcmp_p): Remove externs.
1620 (sparc_emit_set_const32, sparc_emit_set_const64,
1621 sparc_emit_set_symbolic_const64, input_operand, zero_operand,
1622 mem_min_alignment, small_int_or_double): Add externs.
1623 * config/sparc/sparc.md: Document the many uses of UNSPEC and
1624 UNSPEC_VOLATILE in this backend.
1625 (define_function_unit ieu): Rename to ieu_unnamed. Add move and
1626 unary to types which execute in it.
1627 (define_function_unit ieu_shift): Rename to ieu0.
1628 (define_function_unit ieu1): New, executes compare, call, and
1629 uncond_branch type insns.
1630 (define_function_units for type fdivs, fdivd, fsqrt): These
1631 execute in the fpu multiply unit not the adder on UltraSparc.
1632 (define_expand cmpdi): Disallow TARGET_V8PLUS.
1633 (define_insn cmpsi_insn): Rename to cmpsi_insn_sp32.
1634 (define_insn cmpsi_insn_sp64): New, same as sp32 variant except it
1635 allows the arith_double_operand predicate and rHI constraint when
1637 (define_insn cmpdi_sp64, cmpsf_fpe, cmpdf_fpe, cmptf_fpe,
1638 cmpsf_fp, cmpdf_fp, cmptf_fp, sltu_insn, neg_sltu_insn,
1639 neg_sltu_minux_x, neg_sltu_plus_x, sgeu_insn, neg_sgeu_insn,
1640 sltu_plus_x, sltu_plus_x, sltu_plus_x_plus_y, x_minus_sltu,
1641 sgeu_plus_x, x_minus_sgeu, movqi_cc_sp64, movhi_cc_sp64,
1642 movsi_cc_sp64, movdi_cc_sp64, movsf_cc_sp64, movdf_cc_sp64,
1643 movtf_cc_sp64, movqi_cc_reg_sp64, movhi_cc_reg_sp64,
1644 movsi_cc_reg_sp64, movdi_cc_reg_sp64, movsf_cc_reg_sp64,
1645 movdf_cc_reg_sp64, movtf_cc_reg_sp64, zero_extendhisi2_insn,
1646 cmp_siqi_trunc, cmp_siqi_trunc_set, sign_extendhisi2_insn,
1647 sign_extendqihi2_insn, sign_extendqisi2_insn,
1648 sign_extendqidi2_insn, sign_extendhidi2_insn,
1649 extendsfdf2, extendsftf2, extenddftf2, truncdfsf2, trunctfsf2,
1650 trunctfdf2, floatsisf2, floatsidf2, floatsitf2, floatdisf2,
1651 floatdidf2, floatditf2, fix_truncsfsi2, fix_truncdfsi2,
1652 fix_trunctfsi2, fix_truncsfdi2, fix_truncdfdi2, fix_trunctfdi2,
1653 adddi3_sp64, addsi3, cmp_ccx_plus, cmp_cc_plus_set, subdi_sp64,
1654 subsi3, cmp_minus_ccx, cmp_minus_ccx_set, mulsi3, muldi3,
1655 muldi3_v8plus, cmp_mul_set, mulsidi3, mulsidi3_v8plus,
1656 const_mulsidi3_v8plus, mulsidi3_sp32, const_mulsidi3,
1657 smulsi3_highpart_v8plus, unnamed subreg mult,
1658 const_smulsi3_highpart_v8plus, smulsi3_highpart_sp32,
1659 const_smulsi3_highpart, umulsidi3_v8plus, umulsidi3_sp32,
1660 const_umulsidi3, const_umulsidi3_v8plus, umulsi3_highpart_v8plus,
1661 const_umulsi3_highpart_v8plus, umulsi3_highpart_sp32,
1662 const_umulsi3_highpart, divsi3, divdi3, cmp_sdiv_cc_set, udivsi3,
1663 udivdi3, cmp_udiv_cc_set, smacsi, smacdi, umacdi, anddi3_sp64,
1664 andsi3, and_not_di_sp64, and_not_si, iordi3_sp64, iorsi3,
1665 or_not_di_sp64, or_not_si, xordi3_sp64, xorsi3, xor_not_di_sp64,
1666 xor_not_si, cmp_cc_arith_op, cmp_ccx_arith_op,
1667 cmp_cc_arith_op_set, cmp_ccx_arith_op_set, cmp_ccx_xor_not,
1668 cmp_cc_xor_not_set, cmp_ccx_xor_not_set, cmp_cc_arith_op_not,
1669 cmp_ccx_arith_op_not, cmp_cc_arith_op_not_set,
1670 cmp_ccx_arith_op_not_set, negdi2_sp64, cmp_cc_neg, cmp_ccx_neg,
1671 cmp_cc_set_neg, cmp_ccx_set_neg, one_cmpldi2_sp64, cmp_cc_not,
1672 cmp_ccx_not, cmp_cc_set_not, cmp_ccx_set_not, addtf3, adddf3,
1673 addsf3, subtf3, subdf3, subsf3, multf3, muldf3, mulsf3,
1674 muldf3_extend, multf3_extend, divtf3, divdf3, divsf3, negtf2,
1675 negdf2, negsf2, abstf2, absdf2, abssf2, sqrttf2, sqrtdf2, sqrtsf2,
1676 ashlsi3, ashldi3, unnamed DI ashift, cmp_cc_ashift_1,
1677 cmp_cc_set_ashift_1, ashrsi3, ashrdi3, unnamed DI ashiftrt,
1678 ashrdi3_v8plus, lshrsi3, lshrdi3, unnamed DI lshiftrt,
1679 lshrdi3_v8plus, tablejump_sp32, tablejump_sp64, call_address_sp32,
1680 call_symbolic_sp32, call_address_sp64, call_symbolic_sp64,
1681 call_address_struct_value_sp32, call_symbolic_struct_value_sp32,
1682 call_address_untyped_struct_value_sp32,
1683 call_symbolic_untyped_struct_value_sp32, call_value_address_sp32,
1684 call_value_symbolic_sp32, call_value_address_sp64,
1685 call_value_symbolic_sp64, branch_sp32, branch_sp64,
1686 flush_register_windows, goto_handler_and_restore,
1687 goto_handler_and_restore_v9, goto_handler_and_restore_v9_sp64,
1688 flush, all ldd/std peepholes, return_qi, return_hi, return_si,
1689 return_addsi, return_di, return_adddi, return_sf, all call+jump
1690 peepholes, trap, unnamed trap insns): Prettify output strings.
1691 (define_insn anddi3_sp32, and_not_di_sp32, iordi3_sp32,
1692 or_not_di_sp32, xordi3_sp32, xor_not_di_sp32, one_cmpldi2):
1693 Likewise and force + implement splits for integer cases.
1694 (define_insn return_sf_no_fpu): Likewise and allow to match when
1695 no-fpu because of our subreg SFmode splits.
1696 (define_insn zero_extendqihi2, zero_extendqisi2_insn,
1697 zero_extendqidi2_insn, zero_extendhidi2_insn,
1698 zero_extendsidi2_insn, sign_extendsidi2_insn): Likewise and use
1699 input_operand for second operand.
1700 (cmp_minus_cc, cmp_minus_cc_set): Likewise and use
1701 reg_or_0_operand for operand 2 so new splits can use it.
1702 (cmp_zero_extendqisi2, cmp_zero_extendqisi2_set, cmp_cc_plus,
1703 cmp_cc_xor_not): Likewise and don't forget to check TARGET_LIVE_G0
1705 (cmp_zero_extract, cmp_zero_extract_sp64): Likewise and allow
1706 CONST_DOUBLEs for operand 2.
1707 (define_insn move_label_di): Likewise and label distance
1708 optimization because it no longer works with new deferred case
1709 vector scheme. To be revisited.
1710 (define_insn x_minus_y_minus_sltu, x_minus_sltu_plus_y): Likewise
1711 and allow reg_or_0_operand and J constraint for second operand.
1712 (define_insn jump): Set branch predict taken on V9.
1713 (define_insn tablejump): Emit LABEL_REF + PLUS memory address for
1714 new deferred case vector scheme.
1715 (define_insn pic_tablejump_32, pic_tablejump_64): Remove.
1716 (define_insn negdi2_sp32): Force + implement splits.
1717 (define_insn negsi2, one_cmplsi2): Rename to negsi2_not_liveg0 and
1718 one_cmplsi2_not_liveg0 respectively, and create expander of original
1719 names which emit special rtl for TARGET_LIVE_G0.
1720 (define_insn cmpdi_v8plus, scc_si, scc_di): Remove.
1721 (define_insn seq, sne, slt, sge, sle, sltu, sgeu): Don't do
1722 gen_compare_reg, FAIL instead.
1723 (define_insn sgtu, sleu): Likewise and check gen_s*() return
1724 values when trying to reverse condition codes, if they FAIL then
1726 (define_insn snesi_zero, neg_snesi_zero, snesi_zero_extend,
1727 snedi_zero, neg_snedi_zero, snedi_zero_trunc, seqsi_zero,
1728 neg_seqsi_zero, seqsi_zero_extend, seqdi_zero, neg_seqdi_zero,
1729 seqdi_zero_trunc, x_plus_i_ne_0, x_minus_i_ne_0, x_plus_i_eq_0,
1730 x_minus_i_eq_0): Add new splits to perform these multi-insn cases,
1731 set output string to # to indicate they are mandatory splits.
1732 (define_insn pic_lo_sum_si, pic_sethi_si, pic_lo_sum_di,
1733 pic_sethi_di, move_pic_label_si): Remove.
1734 (define_insn movsi_low_sum, movsi_high, movsi_lo_sum_pic,
1735 movsi_high_pic, movsi_pic_label_reg): New patterns to take their
1737 (define_expand movsi_pic_label_ref, define_insn
1738 movsi_high_pic_label_ref, movsi_lo_sum_pic_label_ref): New
1739 expander and insns to handle PIC label references and deferred
1741 (define_insn get_pc_via_rdpc): Comment out as it is no longer
1743 (define_expand movqi, movhi, movsi, movdi, movsf, movdf, movtf):
1744 Rewrite to not use emit_move_sequence, make use of new constant
1745 formation code, and new splits for all multi-insn cases.
1746 (define_insn movqi_insn): Remove sethi case, it can never happen.
1747 Use reg_or_zero_operand instead of const0_rtx explicit test,
1748 use input_operand instead of move_operand for source, and use
1749 general_operand now for dest.
1750 (define_insn movhi_insn): Similar but leave sethi case.
1751 (define_insn lo_sum_qi, store_qi, store_hi): Remove.
1752 (define_insn sethi_hi lo_sum_hi): Rename to movhi_high and
1753 movhi_lo_sum respectively, prettify output string.
1754 (define_insn movsi_zero_liveg0): New pattern to put zero into a
1755 register when needed on TARGET_LIVE_G0.
1756 (define_insn movsi_insn): Use general_operand and input_operand
1757 for dest and src respectively. Simplify applicability test.
1758 Prettify output strings, and add clr alternative for J
1760 (define_insn movdi_sp32_v9, movdi_sp32, define_splits for
1761 deprecated std and reg-reg DI moves): Remove and...
1762 (define_insn movdi_insn_sp32, movdi_insn_sp64): Replace with new
1763 implementation which uses forced splits for all non-single insn
1765 (define_split DI move cases on !arch64): New splits to handle all
1766 situations of 64-bit double register DImode on 32bit, and
1767 unaligned registers and memory addresses for all subtargets.
1768 (define_insn movsf_const_insn, movdf_const_insn, store_sf):
1770 (define_insn movsf_insn, movsf_no_f_insn): Use general_operand and
1771 input_operand for dest and src respectively, prettify output
1773 (define_insn movdf_insn, movdf_no_e_insn, store_df,
1774 movtf_const_insn, movtf_insn, movtf_no_e_insn, store_tf): Remove
1776 (define_insn movdf_insn_sp32, movdf_no_e_insn_sp32,
1777 movdf_insn_sp64, movdf_no_e_insn_sp64, movtf_insn,
1778 movtf_no_e_insn_sp32, movtf_insn_hq_sp64, movtf_insn_sp64,
1779 movtf_no_e_insn_sp64) Replace with new
1780 implementation which uses forced splits for all non-single insn
1782 (define_split DF move cases): New splits in similar vein to DI
1784 (define_insn sethi_di_medlow, sethi_di_medium_pic,
1785 sethi_di_embmedany_data, sethi_di_embmedany_text, sethi_di_sp64,
1786 movdi_sp64_insn): Remove old v9 code model and constant loading
1788 (define_insn pic_lo_sum_di, pic_sethi_di,
1789 sethi_di_medlow_embmedany_pic, sethi_di_medlow, losum_di_medlow,
1790 seth44, setm44, setl44, sethh, setlm, sethm, setlo,
1791 embmedany_sethi, embmedany_losum, embmedany_brsum,
1792 embmedany_textuhi, embmedany_texthi, embmedany_textulo,
1793 embmedany_textlo, movdi_lo_sum_sp64_cint, movdi_lo_sum_sp64_dbl,
1794 movdi_high_sp64_cint, movdi_high_sp64_dbl): Replace with new
1795 scheme, using unspecs, secondary reloads, and one to one sparc
1796 insn to rtl insn mapping for better scheduling and code gen.
1797 (define_expand reload_indi, reload_outdi): Reload helpers for
1798 MEDANY and EMBMEDANY symbol address loading cases which require a
1800 (define_expand movsicc): Remove v8plus_regcmp cases.
1801 (define_insn movdi_cc_sp64_trunc, movdi_cc_reg_sp64_trunc,
1802 cmp_zero_extendqidi2, cmp_zero_extendqidi2_set, cmp_qidi_trunc,
1803 cmp_diqi_trunc_set): New patterns used by some of the new scc
1805 (define_insn xordi3_sp64_dbl): New pattern used for constant
1806 formation when crossing from 32-bit targets.
1807 (define_insn movsi_cc_reg_v8plus, v8plus_clear_high, and helper
1809 (define_insn addx, subx): Make visible and prettify.
1810 (define_insn adddi3_insn_sp32): Likewise and force split.
1811 (define_insn addx_extend, subx_extend, unnamed): New patterns for
1812 64bit scc split usage.
1813 (define_insn unnamed plusDI zero_extend, unnamed minusDI
1814 zero_extend, subdi3): Force and implement splits.
1816 * final.c (final_scan_insn): Don't output labels if target
1817 specifies ASM_OUTPUT_ADDR_{DIFF}_VEC. Do these macro operations
1820 * reorg.c (dbr_schedule): When taking on BR_PRED notes at the end,
1821 don't forget to walk inside SEQUENCESs too as these are what the
1822 delay slot scheduler will create.
1824 Mon Aug 10 01:21:01 1998 Richard Henderson <rth@cygnus.com>
1826 * alpha.md (extxl+1,+2): New patterns to work around
1829 Sat Aug 8 19:20:22 1998 Gary Thomas (gdt@linuxppc.org)
1831 * rs6000.c (rs6000_allocate_stack_space) Fix typo which
1832 caused bad assembly code to be generated.
1834 Sat Aug 8 18:53:28 1998 Jeffrey A Law (law@cygnus.com)
1836 * netbsd.h: Fix typo.
1838 Mon Aug 3 00:06:42 1998 Robert Lipe <robertl@dgii.com>
1840 * config.sub: Fix typo.
1842 Sun Aug 2 22:39:08 1998 Hans-Peter Nilsson <hp@axis.se>
1844 * invoke.texi (Environment Variables): Typo: Change "ascpects"
1846 (Running Protoize): Typo: Change "ther" into "other".
1848 Sun Aug 2 00:42:50 1998 Jeffrey A Law (law@cygnus.com)
1850 * i386/netbsd.h: Undo previous change to DWARF2_UNWIND_INFO.
1851 * m68k/netbsd.h: Likewise.
1852 * ns32k/netbsd.h: Likewise.
1853 * sparc/netbsd.h: Likewise.
1855 Sat Aug 1 17:59:30 1998 Richard Henderson <rth@cygnus.com>
1857 * ginclude/va-alpha.h (va_list): Use a typedef, not a define.
1858 * ginclude/va-clipper.h (va_list): Likewise.
1860 Fri Jul 31 20:22:02 1998 Michael Meissner <meissner@cygnus.com>
1862 * rs6000.c (rs6000_override_options): If big endian and -Os, use
1863 load/store multiple instructions unless user overrides.
1865 Fri Jul 31 17:08:59 1998 Jeffrey A Law (law@cygnus.com)
1867 * ns32k/netbsd.h: Fix typo.
1869 Fri Jul 31 10:23:55 1998 Doug Evans <devans@canuck.cygnus.com>
1871 * m32r/m32r.h (ASM_OUTPUT_SOURCE_LINE): Always output line number
1872 labels with .debugsym if no parallel insns.
1874 Thu Jul 30 19:15:53 1998 Richard Henderson <rth@cygnus.com>
1876 * alpha.md (fp cmp): Replicate patterns for ALPHA_TP_INSN.
1877 (fcmov): Remove ALPHA_TP_INSN patterns -- fcmov doesn't trap.
1879 Thu Jul 30 19:50:15 1998 David Edelsohn <edelsohn@mhpcc.edu>
1881 * rs6000/x-aix43 (AR_FOR_TARGET_FLAGS): Delete.
1882 (AR_FOR_TARGET): Define.
1884 Thu Jul 30 12:29:12 1998 Mark Mitchell <mark@markmitchell.com>
1886 * dyn-string.h: New file.
1887 * dyn-string.c: Likewise.
1888 * Makefile.in (OBJS): Add dyn-string.o.
1889 (dwarf2out.o): Add dyn-string.h dependency.
1890 (dyn-string.o): List dependencies.
1891 * dwarf2out.c: Include dyn-string.h.
1892 (ASM_NAME_TO_STRING): Use dyn_string_append, rather than strcpy.
1893 (addr_const_to_string): Take a dyn_string_t, not a char* as a
1894 prototype. Use dyn_string_append rather than strcat, throughout.
1895 (addr_to_string): Use dyn_string_t.
1897 Thu Jul 30 13:08:07 1998 Ken Raeburn <raeburn@cygnus.com>
1899 Function entry/exit profiling instrumentation:
1900 * expr.h (profile_function_entry_libfunc,
1901 profile_function_exit_libfunc): Declare new variables.
1902 * optabs.c: Define them here.
1903 (init_optabs): Initialize them.
1904 * tree.h (struct tree_decl): New flag
1905 no_instrument_function_entry_exit.
1906 (DECL_NO_INSTRUMENT_FUNCTION_ENTRY_EXIT): New accessor macro.
1907 * c-decl.c (duplicate_decls): Merge it.
1908 * c-common.c (enum attrs): New value A_NO_INSTRUMENT_FUNCTION.
1909 (init_attributes): Use it for "no_instrument_function".
1910 (decl_attributes): Handle it, for functions that have not yet been
1911 compiled. Set decl flag.
1912 * flags.h (flag_instrument_function_entry_exit): Declare new
1914 * toplev.c (flag_instrument_function_entry_exit): Define it here.
1915 (f_options): New option "instrument-functions".
1916 * function.h (struct function): New field instrument_entry_exit.
1917 * function.c (current_function_instrument_entry_exit): New
1919 (push_function_context_to, pop_function_context_from): Save and
1921 (expand_function_start): Set current_ variable, maybe emit return
1922 label and entry profile call.
1923 (expand_function_end): Maybe emit exit profile call.
1925 Thu Jul 30 00:58:34 1998 Jeffrey A Law (law@cygnus.com)
1927 * i386.md (movqi): When optimizing a load of (const_int 1) into a
1928 NON_QI_REG_P, pretend the register is SImode.
1930 Wed Jul 29 23:49:23 1998 Todd Vierling <tv@netbsd.org>
1932 * configure.in: Use xm-netbsd.h as the NetBSD xm file (not xm-siglist).
1933 Accept arm32 as arm, m68k4k as m68k, mipsle as mips-dec, and any
1934 manufacturer id for ns32k.
1935 * configure: Regenerated.
1936 * config/netbsd.h: When using ASM_WEAKEN_LABEL, make it global too.
1937 * config/t-netbsd: Don't compile libgcc1-test as the fns are in libc.
1938 * config/i386/netbsd.h: Undefine DWARF2_UNWIND_INFO, not define as 0.
1939 * config/m68k/netbsd.h: Same.
1940 * config/ns32k/netbsd.h: Same.
1941 * config/sparc/netbsd.h: Same.
1943 Wed Jul 29 22:39:21 1998 Jeffrey A Law (law@cygnus.com)
1945 * unroll.c (unroll_loop): Do not abort for an UNROLL_MODULO
1946 or UNROLL_COMPLETELY loop that starts with a jump to its
1949 Wed Jul 29 22:18:14 1998 David Edelsohn <edelsohn@mhpcc.edu>
1951 * rs6000/rs6000.md (absdi2 define_split): Swap operands of MINUS.
1952 * rs6000/rs6000.c (mask64_operand): Use HOST_BITS_PER_WIDE_INT.
1953 (print_operand, case 'B'): Don't fall through.
1954 (print_operand, case 'S'): Correct mask begin/end computation.
1955 Use HOST_BITS_PER_WIDE_INT.
1956 * rs6000/rs6000.h (CPP_PREDEFINES): Define _LONG_LONG.
1957 (CONDITIONAL_REGISTER_USAGE): GPR13 fixed if TARGET_64BIT.
1958 * rs6000/aix41.h (CPP_PREDEFINES): Same.
1959 * rs6000/aix43.h (CPP_PREDEFINES): Same.
1961 Wed Jul 29 11:47:10 1998 Nick Clifton <nickc@cygnus.com>
1963 * config/arm/thumb.md (extendqisi2_insn): Remove earlyclobber
1964 constraint from second alternative.
1966 Tue Jul 28 23:29:04 1998 Jason Merrill <jason@yorick.cygnus.com>
1968 * configure.in: Fix --without/--disable cases for local-prefix,
1969 gxx-include-dir and checking.
1971 Tue Jul 28 22:01:23 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
1973 * configure.in (enable_haifa): Set by default for sparc64 too.
1976 Tue Jul 28 23:29:04 1998 Jason Merrill <jason@yorick.cygnus.com>
1978 * i386/cygwin32.h (VALID_MACHINE_TYPE_ATTRIBUTE): New macro.
1979 * i386/winnt.c (associated_type): New fn.
1980 (i386_pe_valid_type_attribute_p): New fn.
1981 (i386_pe_check_vtable_importexport): Remove.
1982 (i386_pe_dllexport_p): Use associated_type.
1983 (i386_pe_dllimport_p): Likewise.
1985 From Antonio M. O. Neto <anmendes@cruzeironet.com.br>:
1986 * i386.c (i386_valid_type_attribute_p): Also accept
1987 attributes for METHOD_TYPEs.
1989 Tue Jul 28 23:17:39 1998 Peter Gerwinski <peter@gerwinski.de>
1991 * tree.c (build_range_type): Copy TYPE_SIZE_UNIT.
1993 Tue Jul 28 22:31:12 1998 Craig Burley <burley@gnu.org>
1995 * gcc.c: Fix commentary describing %g, %u, %U, and %O.
1997 * gcc.c (do_spec_1): Fix handling of %g%O and %U%O to prevent
1998 them from generating a new base name for each occurence of
2001 1998-07-28 Vladimir N. Makarov <vmakarov@cygnus.com>
2003 * cse.c (cse_insn): Enable subsitution inside libcall only for REG,
2005 * rtlanal.c (replace_rtx): Prohibit replaces in CONST_DOUBLE.
2009 * cplus-dem.c (type_kind_t): New type.
2010 (demangle_template_value_parm): Add type_kind_t parameter. Rely
2011 on this paramter, rather than demangling the type again.
2012 (demangle_integral_value): Pass tk_integral.
2013 (demangle_template_: Pass the value returned from do_type.
2014 (do_type): Return a type_kind_t. Pass tk_integral to
2015 demangle_template_value_parm for array bounds.
2016 (demangle_fund_type): Likewise.
2018 Mon Jul 27 00:54:41 1998 Jason Merrill <jason@yorick.cygnus.com>
2020 * tree.c (simple_cst_equal, case CONSTRUCTOR): OK if the elts are
2023 Mon Jul 27 22:18:36 1998 Jeffrey A Law (law@cygnus.com)
2025 * pa.c (move_operand): Accept CONSTANT_P_RTX.
2027 Mon Jul 27 17:18:52 1998 Dave Brolley <brolley@cygnus.com>
2029 * stor-layout.c (layout_type): Handle arrays of bits, for Chill.
2031 * expr.c (get_inner_reference): Handle zero-based, unsigned, array
2034 Mon Jul 27 14:51:33 1998 Jeffrey A Law (law@cygnus.com)
2036 * mn10300.h (DEBUGGER_AUTO_OFFSET): Define.
2037 (DEBUGGER_ARG_OFFSET): Likewise.
2039 * mn10300.md (movsf): Remove last change. Not needed.
2041 Mon Jul 27 14:22:36 1998 Dave Brolley <brolley@cygnus.com>
2043 * c-lex.c (yylex): Fix boundary conditions in character literal and
2044 string literal loops.
2046 Mon Jul 27 11:43:54 1998 Stan Cox <scox@cygnus.com>
2048 * longlong.h (count_leading_zeros): Sparclite scan instruction was
2049 being invoked incorrectly.
2051 * i386.c (ix86_prologue): Added SUBTARGET_PROLOGUE invocation.
2052 * i386/cygwin32.h (STARTFILE_SPEC, LIB_SPEC, SUBTARGET_PROLOGUE):
2054 * i386/win32.h: New file. Hybrid mingw32.h/cygwin32.h configuration.
2055 * configure.in: Added i[34567]86-*-win32
2056 * config.sub: Ditto.
2057 * configure: Rebuilt.
2059 Sun Jul 26 01:11:12 1998 H.J. Lu (hjl@gnu.org)
2061 * i386.h (CONST_DOUBLE_OK_FOR_LETTER_P): Return 0 when eliminating
2062 the frame pointer and compiling PIC code and reload has not completed.
2064 * i386.c (output_to_reg): Add code to emulate non-popping DImode
2067 Sun Jul 26 01:01:32 1998 Jeffrey A Law (law@cygnus.com)
2069 * regmove.c (regmove_optimize): Fix typo initializing regmove_bb_head.
2071 Sat Jul 25 23:29:23 1998 Gerald Pfeifer <pfeifer@dbai.tuwien.ac.at>
2073 * Makefile.in (install-info): Only try to update the info
2074 directory file if it exists in the first place.
2076 Fri Jul 24 18:58:37 1998 Klaus Espenlaub <kespenla@student.informatik.uni-ulm.de>
2078 * rs6000.h (ASM_OUTPUT_CONSTRUCTOR, ASM_OUTPUT_DESTRUCTOR): Delete.
2080 Fri Jul 24 14:20:26 1998 Jeffrey A Law (law@cygnus.com)
2082 * mn10300.md (movqi, movhi, movsi, movsf): Correctly handle
2083 CONST_DOUBLE source.
2085 Fri Jul 24 11:17:04 1998 Nick Clifton <nickc@cygnus.com>
2087 * config/arm/thumb.c (thumb_print_operand): Decode %_ in asm
2088 strings as the insertion of USER_LABEL_PREFIX.
2089 * config/arm/thumb.h (PRINT_OPERAND_PUNCT_VALID_P): Accept _ as a
2091 * config/arm/thumb.md: Use %_ as a prefix to gcc library function
2094 Thu Jul 23 18:53:20 1998 Jim Wilson <wilson@cygnus.com>
2096 * dbxout.c (dbxout_range_type): Only call dbxout_type_index for
2097 already defined type.
2099 Thu Jul 23 13:49:41 1998 Jeffrey A Law (law@cygnus.com)
2101 * expr.c (check_max_integer_computation_mode): Allow conversions
2102 of constant integers to MAX_INTEGER_COMPUTATION_MODE.
2103 (expand_expr): Likewise.
2105 Thu Jul 23 11:12:06 1998 Alexandre Petit-Bianco <apbianco@cygnus.com>
2107 * expr.c (expand_expr): Expand RETURN_EXPR.
2109 Thu Jul 23 11:00:29 1998 Jim Wilson <wilson@cygnus.com>
2111 * dwarf2out.c (dwarf2out_finish): Call stripattributes on TEXT_SECTION.
2113 Wed Jul 22 19:10:00 1998 Catherine Moore <clm@cygnus.com>
2115 * dwarf2out.c (output_aranges): Call stripattributes
2116 for TEXT_SECTION references.
2117 (output_line_info): Likewise.
2119 Wed Jul 22 14:08:54 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
2121 * profile.c (branch_prob): Call allocate_reg_info after outputting
2122 profile rtl in instrument_arcs.
2124 Wed Jul 22 12:47:49 1998 Jim Wilson <wilson@cygnus.com>
2126 * fixinc.irix (math.h): Install wrapper instead of copying.
2128 Wed Jul 22 12:37:14 1998 Alexandre Petit-Bianco <apbianco@cygnus.com>
2130 * tree.def (EXPR_WITH_FILE_LOCATION): Defined as an 'e' expression
2131 so WFL are expanded correctly when contained in a COMPOUND_EXPR.
2132 * tree.h (EXPR_WFL_EMIT_LINE_NOTE): Change macro not to use
2133 lang_flag_0. Added documentation in the flag table.
2135 Tue Jul 21 23:28:35 1998 Klaus Kaempf <kkaempf@rmi.de>
2137 * cccp.c (do_include): Fix vax c style include handling.
2139 Tue Jul 21 13:28:19 1998 Jason Merrill <jason@yorick.cygnus.com>
2141 * cplus-dem.c (do_type): Use demangle_template_value_parm for arrays.
2143 Sun Jul 12 01:27:05 1998 Jason Merrill <jason@yorick.cygnus.com>
2145 * fold-const.c (non_lvalue): Don't deal with null pointer
2147 (fold, case COMPOUND_EXPR): Wrap a constant 0 in a NOP_EXPR.
2149 Tue Jul 21 15:49:31 1998 David Edelsohn <edelsohn@mhpcc.edu>
2151 * rs6000.h (PREDICATE_CODES): Add CONSTANT_P_RTX.
2152 * rs6000.md (movsi, movdi): Add CONSTANT_P_RTX.
2153 * rs6000.c (short_cint_operand): Add CONSTANT_P_RTX.
2154 (u_short_cint_operand): Same.
2155 (reg_or_cint_operand): Same.
2156 (logical_operand): Same.
2157 (input_operand): Same.
2158 (reg_or_short_operand): Use u_short_cint_operand.
2160 Tue Jul 21 08:56:42 1998 Richard Henderson <rth@cygnus.com>
2162 * alpha.md (fix_truncdfsi2, fix_truncsfsi2): Remove the define_expands,
2163 but keep the insns and splits. Adjust so when the ultimate destination
2164 is memory, use cvtql.
2166 Tue Jul 21 08:55:09 1998 Richard Henderson <rth@cygnus.com>
2168 * flow.c (regno_uninitialized): Fixed regs are never uninitialized.
2170 Tue Jul 21 00:31:01 1998 Jeffrey A Law (law@cygnus.com)
2172 * gcc.c (do_spec): Call "error" not "warning".
2174 * configure.in: Fix minor problems with gas feature detection code.
2175 * configure: Rebuilt.
2177 * gcc.c (do_spec): Issue a warning for '%[]' usage.
2180 * gcc.c: Delete %[spec] support.
2181 (do_spec_1, case '('): Likewise.
2182 (do_spec_1, case '['): Call error.
2184 Mon Jul 20 22:34:17 1998 Richard Henderson <rth@cygnus.com>
2186 * alpha.h (CPP_SPEC): Tidy. Hook to cpp_cpu and cpp_subtarget.
2187 (CPP_SUBTARGET_SPEC): Default to empty string.
2188 (CPP_AM_*, CPP_IM_*, CPP_CPU_*, CPP_CPU_SPEC): New.
2189 (EXTRA_SPECS, SUBTARGET_EXTRA_SPECS): New.
2190 * alpha/elf.h (LD_SPEC): Use %(elf_dynamic_linker).
2191 * alpha/linux-elf.h (SUBTARGET_EXTRA_SPECS): New.
2193 * alpha/linux.h (CPP_PREDEFINES): Tidy.
2194 * alpha/netbsd-elf.h (SUBTARGET_EXTRA_SPECS): New.
2195 * alpha/netbsd.h (CPP_PREDEFINES): Tidy.
2196 * alpha/osf.h (CPP_PREDEFINES): Remove bits subsumed by CPP_CPU_SPEC.
2197 * alpha/win-nt.h (CPP_PREDEFINES): Likewise.
2198 * alpha/vsf.h (CPP_PREDEFINES): Likewise.
2199 (CPP_SUBTARGET_SPEC): New. Do this instead of overriding CPP_SPEC.
2200 * alpha/vxworks.h: Likewise.
2202 Mon Jul 20 22:51:57 1998 Ken Raeburn <raeburn@cygnus.com>
2204 * mips.md (reload_outsi): Added missing REGNO call.
2205 (smulsi3_highpart, umulsi3_highpart): Provide prototype for
2207 (mul_acc_di, mul_acc_64bit_di): Don't use match_op_dup, use
2208 another match_operator and compare the codes.
2210 * mips.h (MASK_DEBUG_E, MASK_DEBUG_I): Set to zero.
2212 * MIPS multiply pattern fixes:
2213 * mips.h (enum reg_class, REG_CLASS_NAMES, REG_CLASS_CONTENTS):
2214 Add union classes for HI, LO, or HILO plus general registers.
2215 (GENERATE_MADD): Deleted.
2216 * mips.md (mulsi3_mult3): Don't disparage output-LO alternative.
2217 Add TARGET_MAD to condition.
2218 (mulsi3): Test HAVE_mulsi3_mult3, not specific flags.
2219 (mul_acc_si): Expand GENERATE_MADD here; it's the only use. Use
2220 "*d" for accumulator, to give preference to LO initially but not
2223 Mon Jul 20 16:16:38 1998 Dave Brolley <brolley@cygnus.com>
2225 * configure.in (enable_c_mbchar): New configure option.
2226 (extra_cpp_objs): Always available now.
2228 * cexp.y (mbchar.h): #include it.
2229 (yylex): Handle Multibyte characters in character literals.
2231 * cccp.c (mbchar.h): #include it.
2232 (main): Set character set based on LANG environment variable.
2233 (rescan): Handle multibyte characters in comments.
2234 (skip_if_group): See above.
2235 (validate_else): See above.
2236 (skip_to_end_of_comment): See above.
2237 (macarg1): See above.
2238 (discard_comments): See above.
2239 (rescan): Handle multibyte characters in string and character literals.
2240 (collect_expansion): See above.
2241 (skip_quoted_string): See above.
2242 (macroexpand): See above.
2243 (macarg1): See above.
2244 (discard_comments): See above.
2245 (change_newlines): See above.
2247 * c-lex.c (mbchar.h): #include it.
2248 (GET_ENVIRONMENT): New macro.
2249 (init_lex): Set character set based on LANG environment variable.
2250 (yylex): Handle multibyte characters in character literals.
2251 (yylex): Handle multibyte characters in string literals.
2253 * Makefile.in (mbchar.o): New target.
2254 (cccp$(exeext)): @extra_cpp_objs@ is always available.
2255 (cppmain$(exeext)): @extra_cpp_objs@ is always available.
2257 * mbchar.[ch]: New files for multibyte character handling.
2259 Mon Jul 20 01:11:11 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
2261 * jump.c (jump_optimize): When simplifying noop moves and
2262 PUSH_ROUNDING, fix thinko so we use same criterion for identifying
2263 the PUSHes to rewrite in second loop as we did in the first.
2265 Sun Jul 19 08:23:53 1998 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
2267 * cplus-dem.c (demangle_nested_args): Make function definition
2268 static to match the prototype.
2270 Fri Jul 17 14:58:44 1998 Richard Henderson <rth@cygnus.com>
2272 * alloca.c: Respect USE_C_ALLOCA.
2273 * gencheck.c (xmalloc): Ignore __GNUC__ for definition.
2274 * gengenrtl.c (xmalloc): Likewise.
2276 Fri Jul 17 14:18:14 1998 Richard Henderson <rth@cygnus.com>
2278 * loop.h (struct induction): Add no_const_addval.
2279 * loop.c (the_movables, reg_address_cost): New variables.
2280 (init_loop): Init reg_address_cost.
2281 (loop_optimize): Call end_alias_analysis.
2282 (scan_loop): Init the_movables.
2283 (record_giv): Init induction->no_const_addval.
2284 (basic_induction_var) [PLUS]: Use rtx_equal_p instead of ==.
2285 [REG]: Rearrange loop search test to catch more cases.
2286 (general_induction_var): Return success not benefit; take an extra
2287 argument for that. Change all callers.
2288 (simplify_giv_expr) [PLUS]: Always combine invariants. Use sge_plus.
2289 [MULT]: Use rtx_equal_p instead of ==. Combine simple invariants.
2290 [default]: Search the_movables for additional combinations.
2291 (sge_plus_constant, sge_plus): New functions.
2292 (express_from_1): New function.
2293 (express_from): Always define. Rewrite using express_from_1.
2294 (combine_givs_p): Handle more cases. Ignore address cost.
2295 (cmp_combine_givs_stats): New function.
2296 (combine_givs_used_once, combine_givs_benefit_from): New functions.
2297 (combine_givs): Rewrite to do best-fit combination.
2299 * fold-const.c (operand_equal_p): Handle RTL_EXPR.
2300 (fold): Do a complete (A*C)+(B*C) association check.
2302 Fri Jul 17 11:21:55 1998 Jim Wilson <wilson@cygnus.com>
2304 * function.c (fixup_var_refs_insns): Handle CLOBBER of a CONCAT.
2306 Fri Jul 17 11:48:55 1998 Jeffrey A Law (law@cygnus.com)
2308 * mn10300.c (MODES_TIEABLE_P): Fix typo.
2310 Fri Jul 17 03:26:12 1998 Rihcard Earnshaw (rearnsha@arm.com)
2312 * tree.c (valid_machine_attribute): Only create a new type variant if
2313 there is a decl to use it.
2315 Thu Jul 16 14:48:04 1998 Nick Clifton <nickc@cygnus.com>
2317 * gcc.c (do_spec_1): Cope with %g/%u/%U options which do not have
2320 Fri Jul 17 03:24:40 1998 Hans-Peter Nilsson <hp@axis.se>
2322 * extend.texi (Explicit Reg Vars): Typo: change "may deleted" into "may
2325 Thu Jul 16 14:48:47 1998 Jeffrey A Law (law@cygnus.com)
2327 * mn10300.c (count_tst_insns): New arg oreg_countp. Callers changed.
2328 Simplify tests for clearing an address register.
2329 (expand_prologue): Corresponding changes.
2331 * mn10300.md (movXX patterns): Make sure the destination is an
2332 ADDRESS_REG when substituting "zero_areg" for (const_int 0).
2333 (logical patterns): Split into expanders + patterns
2334 (zero and sign extension patterns): Similarly.
2335 (shift patterns): Similarly.
2337 Thu Jul 16 01:17:44 1998 Richard Henderson <rth@cygnus.com>
2339 * loop.c (emit_iv_add_mult): Scan the entire insn list generated
2340 for the sequence, recording base values.
2342 Wed Jul 15 10:49:55 1998 Richard Henderson <rth@cygnus.com>
2344 * i386.h (CPP_CPU_SPEC): Remove -Asystem(unix).
2346 Tue Jul 14 14:15:30 1998 Nick Clifton <nickc@cygnus.com>
2348 * gcc.c: Remove ANSI-C ism from --help code.
2350 * toplev.c: Support --help with USE_CPPLIB.
2352 Tue Jul 14 14:46:08 1998 Jeffrey A Law (law@cygnus.com)
2354 * configure.in: Rework gas feature code to work with symlink based
2357 * extend.texi: Clarify some issues related to local variables
2358 assigned to explicit registers.
2360 * mn10300.md (mulsi): Turn into expander + pattern.
2362 * mn10300.md (movsi, movsf, movdi, movdf): Remove "x" from I -> a
2365 Tue Jul 14 07:41:59 1998 Richard Earnshaw (rearnsha@arm.com)
2367 * arm/tcoff.h (USER_LABEL_PREFIX): Make it empty to match coff.h.
2369 Tue Jul 14 03:02:44 1998 Jeffrey A Law (law@cygnus.com)
2371 * version.c: Bump again to distinguish mainline tree from the
2374 See ChangeLog.0 for earlier changes.
2377 add-log-time-format: current-time-string