1 Mon Sep 7 22:41:46 1998 Michael Meissner <meissner@cygnus.com>
3 * rs6000.c (rs6000_override_options): Fix name for ec603e, to add
5 * t-ppccomm (MULTILIB_MATCHES_FLOAT): Add support for -mcpu=xxx
6 for all targets that set -msoft-float.
8 Mon Sep 7 23:30:07 1998 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
10 * toplev.c (print_switch_values): Make static to match prototype.
12 Mon Sep 7 19:13:59 1998 Jeffrey A Law (law@cygnus.com)
14 * configure.in: If we are unable to find the "gnatbind" program,
15 then do not configure the ada subdir.
18 Sun Sep 6 14:03:58 PDT 1998 Jeff Law (law@cygnus.com)
20 * version.c: Bump for snapshot.
22 Sun Sep 6 13:28:07 PDT 1998 Jeff Law (law@cygnus.com)
24 * version.c: Bump for snapshot.
26 Sun Sep 6 08:54:14 1998 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
28 * Makefile.in (toplev.o): Depend on $(EXPR_H).
29 (insn-extract.o, insn-attrtab.o): Depend on toplev.h.
31 * gansidecl.h: Define ATTRIBUTE_NORETURN.
33 * genattrtab.c: Have insn-attrtab.c include toplev.h.
35 * genextract.c: Have insn-extract.c include toplev.h.
37 * rtl.h: Don't prototype `fatal_insn_not_found' and `fatal_insn'.
39 * toplev.c: Include expr.h.
40 (really_sorry, fancy_abort): Remove prototypes.
41 (set_target_switch): Add argument in prototype.
42 (vfatal): Mark prototype with ATTRIBUTE_NORETURN.
43 (v_really_sorry): Likewise.
44 (print_version, print_single_switch, print_switch_values): Make
45 static and add prototype arguments.
46 (decl_printable_name): Add prototype arguments.
47 (lang_expand_expr_t): New typedef.
48 (lang_expand_expr): Declare as a lang_expand_expr_t.
49 (incomplete_decl_finalize_hook): Add prototype argument.
50 (decl_name): Mark variable `verbosity' with ATTRIBUTE_UNUSED.
51 (botch): Likewise for variable `s'.
52 (rest_of_type_compilation): Mark variables `type' and `toplev'
53 with ATTRIBUTE_UNUSED if none of DBX_DEBUGGING_INFO,
54 XCOFF_DEBUGGING_INFO or SDB_DEBUGGING_INFO are defined.
55 (display_help): Make variable `i' an `unsigned long'.
56 (main): Remove unused parameter `envp'.
57 Cast assignment to `lang_expand_expr' to a `lang_expand_expr_t'.
58 Cast -1 when comparing it with a `size_t'.
60 * toplev.h (fatal, fatal_io_error, pfatal_with_name): Mark
61 prototype with ATTRIBUTE_NORETURN.
62 (fatal_insn_not_found, fatal_insn, really_sorry,
63 push_float_handler, pop_float_handler): Add prototypes.
64 (fancy_abort): Mark prototype with ATTRIBUTE_NORETURN.
65 (do_abort, botch): Add prototypes.
67 Sat Sep 6 12:05:18 1998 John Carr <jfc@mit.edu>
69 * final.c (final): If a label is reached only from a single jump,
70 call NOTICE_UPDATE_CC on the jump and its predecessor before
71 emitting the insn after the label.
73 * i386.h: Add AMD K6 support.
74 Change TARGET_* macros to use table lookup.
75 (INITIALIZE_TRAMPOLINE): Improve trampoline code.
76 (ADJUST_COST): Change definition to call function in i386.c.
77 (ISSUE_RATE): Define as 2 for anything newer than an 80486.
78 * i386.c: Add AMD K6 support.
79 Add constants for feature tests used by TARGET_* macros.
80 (split_di): If before reload, call gen_lowpart and gen_highpart.
81 (x86_adjust_cost): New function.
82 (put_jump_code): New function.
83 (print_operand): New codes 'D' and 'd'.
84 * i386.md: New insn types. New insn attribute "memory".
85 Redefine scheduling parameters to use new types and add AMD K6
86 support. Explicitly set type of most insns.
87 (move insns): K6 prefers movl $0,reg to xorl reg,reg. Pentium
88 Pro and K6 prefer movl $1,reg to incl reg.
89 (adddi3, subdi3): Set cc_status.
90 (DImode shift patterns): Change label counters from HOST_WIDE_INT
91 to int; x86 can't have more than 2^31 DImode shifts per file.
92 (setcc): Combine all setcc patterns. Allow writing memory.
93 Combine all jump patterns using match_operator.
94 (*bzero): Name pattern. Emit mutliple stos instructions when that
95 is faster than rep stos.
96 (xordi3, anddi3, iordi3): Simplify DImode logical patterns and
99 Sun Sep 6 11:17:20 1998 Dave Love <d.love@dl.ac.uk>
101 * config/m68k/x-next (BOOT_LDFLAGS): Define suitably for f771
104 Sat Sep 5 22:05:25 1998 Richard Henderson <rth@cygnus.com>
106 * alpha.c (alpha_ra_ever_killed): Inspect the topmost sequence,
107 not whatever we're generating now.
109 * alpha.c (set_frame_related_p, FRP): New.
110 (alpha_expand_prologue): Mark frame related insns.
111 (alpha_expand_epilogue): Likewise, but with a null FRP.
112 * alpha.h (INCOMING_RETURN_ADDR_RTX): New.
113 * alpha.md (exception_receiver): New.
114 * alpha/crtbegin.asm (.eh_frame): New beginning.
115 (__do_frame_setup, __do_frame_takedown): New.
116 * alpha/crtend.asm (.eh_frame): New ending.
117 * alpha/elf.h (DWARF2_DEBUGGING_INFO): Define.
118 (ASM_SPEC): Don't emit both dwarf2 and mdebug.
119 (ASM_FILE_START): Don't emit .file for dwarf2.
121 * rtl.h (enum reg_note): Add REG_FRAME_RELATED_EXPR.
122 * rtl.c (reg_note_name): Likewise.
123 * rtl.texi (REG_NOTES): Likewise.
124 * dwarf2out.c (dwarf2out_frame_debug): Use it. Recognize a store
127 Sat Sep 5 14:47:17 1998 Richard Henderson <rth@cygnus.com>
129 * i386.h (PREFERRED_RELOAD_CLASS): Standard fp constants load to TOS.
130 * i386.md (movsf, movdf, movxf): Validate memory address returned
131 from force_const_mem. Kill useless REG_EQUAL setting code.
133 Sat Sep 5 14:23:31 1998 Torbjorn Granlund <tege@matematik.su.se>
135 * m68k.md (zero_extendsidi2): Fix typo.
137 Sat Sep 5 13:40:24 1998 Krister Walfridsson <cato@df.lth.se>
139 * configure.in: Removed references to the removed file
140 * config/xm-netbsd.h. Use ${cpu_type}/xm-netbsd.h for
141 * arm*-*-netbsd* and ns32k-*-netbsd*.
142 * config/i386/xm-netbsd.h: Removed unnecessary file.
143 * config/m68k/xm-netbsd.h: Likewise.
144 * config/sparc/xm-netbsd.h: Likewise.
145 * config/mips/xm-netbsd.h: Likewise.
147 Sat Aug 29 13:32:58 1998 Mumit Khan <khan@xraylith.wisc.edu>
149 * i386/cygwin32.h (BIGGEST_ALIGNMENT): Define.
150 (PCC_BITFIELD_TYPE_MATTERS): Define to be 0.
152 * i386/cygwin32.h (ASM_OUTPUT_SECTION_NAME): Don't check for
153 for exact section attributions.
155 * i386/mingw32.h (CPP_PREDEFINES): Add __MSVCRT__ for msvc
157 * i386/crtdll.h (CPP_PREDEFINES): Define.
159 Sat Sep 5 03:23:05 1998 Jeffrey A Law (law@cygnus.com)
161 * m68k.md (5200 movqi): Do not allow byte sized memory references
163 * m68k.c (output_move_qimode): Do not use byte sized operations on
166 * Makefile.in (pexecute.o): Use pexecute.c from libiberty. Provide
167 explicit rules for building. Similarly for alloca, vfprintf,
168 choose-temp and mkstemp, getopt, getopt1, and obstack.
169 (INCLUDES): Add $(srcdir)/../include.
170 * pexecute.c, alloca.c, vfprintf.c, choose-temp.c, mkstemp.c: Delete.
171 * getopt.h, getopt.c getopt1.c, obstack.c, obstack.h: Likewise.
173 Fri Sep 4 11:57:50 1998 Tom Tromey <tromey@cygnus.com>
175 * gcc.c (do_spec_1): [case 'o'] Account for
176 lang_specific_extra_outfiles.
177 (main): Correctly clear all slots in outfiles for
178 lang_specific_extra_outfiles. Set input_file_number before
179 calling lang_specific_pre_link.
181 Fri Sep 4 10:37:07 1998 Jim Wilson <wilson@cygnus.com>
183 * loop.c (load_mems): Fix JUMP_LABEL field after for_each_rtx call.
185 Fri Sep 4 02:01:05 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
187 * config/sparc/sparc.c (output_double_int): In all V9 symbolic
189 (sparc_output_deferred_case_vectors): If no work to do, return.
190 Fix thinko in Sept 1 change.
192 1998-09-03 SL Baur <steve@altair.xemacs.org>
194 * Makefile.in: add semicolon in BISON definition for portability.
196 Thu Sep 3 13:34:41 1998 Toon Moene <toon@moene.indiv.nluug.nl>
198 * config/nextstep.c (handle_pragma): Correct name of third
201 Tue Sep 1 11:30:33 1998 Nick Clifton <nickc@cygnus.com>
203 * config/m32r/m32r.md: Change (reg:CC 17) to (reg:SI 17).
204 * config/m32r/m32r.h: Make register 17 be fixed.
205 * config/m32r/m32r.c: Use SImode for cc operations.
207 Thu Sep 3 18:17:34 1998 Benjamin Kosnik <bkoz@cygnus.com>
209 * invoke.texi (Warning Options): Add -Wnon-template-friend
212 Thu Sep 3 18:16:16 1998 Michael Meissner <meissner@cygnus.com>
214 * rs6000.c (rs6000_override_options): Add -mcpu={401,e603e}.
216 Thu Sep 3 18:05:16 1998 David Edelsohn <edelsohn@mhpcc.edu>
218 * rs6000.md (movsf): Disable explicit secondary-reload-like
219 functionality if TARGET_POWERPC64.
220 (movdf): Remove TARGET_POWERPC64 explicit secondary-reload-like
224 Thu Sep 3 11:41:40 1998 Robert Lipe <robertl@dgii.com>
226 * fixinc.sco: Borrow code to wrap 'bool' typedefs from tinfo.h
227 and term.h from fixinc.wrap.
229 Thu Sep 3 09:47:31 1998 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
231 * aclocal.m4 (GCC_HEADER_STRING): New macro to detect if it is
232 safe to include both string.h and strings.h together.
233 (GCC_NEED_DECLARATION): Test STRING_WITH_STRINGS when deciding
234 which headers to search for function declarations. Continue to
235 prefer string.h over strings.h when both are not acceptable.
237 * acconfig.h (STRING_WITH_STRINGS): Add stub.
239 * configure.in: Call GCC_HEADER_STRING.
241 * system.h: Test STRING_WITH_STRINGS when deciding which headers
242 to include. Continue to prefer string.h over strings.h when both
245 Wed Sep 2 23:56:29 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
247 * config/sparc/sparc.c (output_double_int): If V9 and MEDLOW, do
248 not assume top 32-bits of symbolic addresses are zero if
251 Thu Sep 3 00:23:21 1998 Richard Henderson <rth@cygnus.com>
253 * ginclude/va-alpha.h: Protect entire second portion of the
254 file against double inclusion.
256 Thu Sep 3 00:37:55 1998 Ovidiu Predescu <ovidiu@aracnet.com>
258 Added support for the Boehm's garbage collector.
259 * configure.in: Handle --enable-objc-gc.
260 * configure: Rebuilt.
261 * Makefile.in (CHECK_TARGETS): Add check-objc.
262 (check-objc): New rule.
263 * objc/Make-lang.in: Build a different Objective-C library that
264 runs with the Boehm's collector.
265 * objc/encoding.c (objc_round_acc_size_for_types): New function.
266 * objc/encoding.c: Correctly compute the size of compound types in
267 the presence of bitfields. Skip the variable name of the type if
268 any. Added support for long long.
269 * objc/encoding.h (_C_GCINVISIBLE): New specifier.
270 (_F_GCINVISIBLE): New mask.
271 * objc/gc.c: New file. Compute the type memory mask associated with
272 a class based on the runtime information.
273 * objc/misc.c: Added the hooks that use the Boehm's collector
274 allocation functions.
275 * objc/objc-act.c (build_class_template): Generate a new class
276 member (gc_object_type) to hold the class' type memory mask.
277 (build_shared_structure_initializer): Initialize the new member to
279 (encode_complete_bitfield): New function. Generate the new
281 (encode_field_decl): Generate the new encoding only for the GNU
283 * objc/objc-api.h (_C_LNG_LNG, _C_ULNG_LNG): New specifiers for the
285 (class_get_gc_object_type): New function to mark a pointer instance
286 variable as a weak pointer.
287 * objc/objc-features.texi: New file.
288 * objc/objc.h (gc_object_type): New class member.
289 * objc/objects.c (class_create_instance): Create a typed memory
290 object when compiled with Boehm's collector support.
291 * objc/sendmsg.c (__objc_init_install_dtable): Call
292 __objc_send_initialize instead of setting the initialize flag.
293 (__objc_send_initialize): Call __objc_generate_gc_type_description
294 to generate the class type memory mask. Rewrite the code that
295 sends the +initialize so that it is called only once (bug report
296 and fix from Ronald Pijnacker <Ronald.Pijnacker@best.ms.philips.com>).
297 * testsuite/objc: New testsuite for Objective-C type encoding.
298 * testsuite/lib/objc-torture.exp: New file.
299 * testsuite/lib/objc.exp: New file.
301 Wed Sep 2 14:47:36 1998 Jim Wilson <wilson@cygnus.com>
303 * jump.c (jump_optimize): In if/then/else transformations, add
304 another call to modified_between_p for the jump insn.
306 Wed Sep 2 14:16:49 1998 Jeffrey A Law (law@cygnus.com)
308 * fix-header.c (symlink): Treat like readlink.
310 Wed Sep 2 19:30:06 1998 J"orn Rennecke <amylaar@cygnus.co.uk>
312 * dwarfout.c (fundamental_type_code): Encode 32 bit floats/doubles
315 Wed Sep 2 10:06:07 1998 Nick Clifton <nickc@cygnus.com>
317 * config/nextstep.h: Update HANDLE_PRAGMA macro.
318 * config/h8300/h8300.h: Update HANDLE_PRAGMA macro.
319 * config/i960/i960.h: Update HANDLE_PRAGMA macro.
321 * config/nextstep.c (handle_pragma): Take three arguments, as per
322 the new HANDLE_PRAGMA macro specification.
323 * config/h8300/h8300.c (handle_pragma): Take three arguments, as
324 per the new HANDLE_PRAGMA macro specification.
325 * config/i960/i960.c (process_pragma): Take three arguments, as
326 per the new HANDLE_PRAGMA macro specification.
328 Wed Sep 2 09:25:29 1998 Nick Clifton <nickc@cygnus.com>
330 * c-lex.c (check_newline): Call HANDLE_PRAGMA before
331 HANDLE_SYSV_PRAGMA if both are defined. Generate warning messages
332 if unknown pragmas are encountered.
333 (handle_sysv_pragma): Interpret return code from
334 handle_pragma_token (). Return success/failure indication rather
335 than next unprocessed character.
336 (pragma_getc): New function: retrieves characters from the
337 input stream. Defined when HANDLE_PRAGMA is enabled.
338 (pragma_ungetc): New function: replaces characters back into the
339 input stream. Defined when HANDLE_PRAGMA is enabled.
341 * c-pragma.c (handle_pragma_token): Return success/failure status
344 * c-pragma.h: Change prototype of handle_pragma_token().
346 * varasm.c: (handle_pragma_weak): Only create this function if
347 HANDLE_PRAGMA_WEAK is defined.
349 * c-common,c (decl_attributes): If defined call the expression
350 contained within the INSERT_ATTRIBUTES macro before adding
351 attributes to a decl.
353 * tm.texi (HANDLE_PRAGMA): Document the new verion of
354 HANDLE_PRAGMA, which takes three arguments.
355 (INSERT_ATTRIBUTES): Document this new macro.
357 * LANGUAGES: Document the new version of HANDLE_PRAGMA and the
358 new INSERT_ATTRIBUTES macro.
360 Wed Sep 2 02:03:23 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
362 * config/sparc/sparc.md (movdf): Only generate special RTL for
364 (move_label_di): Remove
365 (movdi_pic_label_ref, movdi_high_pic_label_ref,
366 movdi_lo_sum_pic_label_ref): New patterns for 64-bit label
368 * config/sparc/sparc.h (ASM_OUTPUT_ADDR_VEC_ELT,
369 ASM_OUTPUT_ADDR_DIFF_ELT): Don't do anything special for MEDLOW,
370 output an .xword for all 64-bit cases.
372 Tue Sep 1 15:55:17 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
374 * config/sparc/sparc.c (finalize_pic): Don't output arbitrary
375 alignment, use FUNCTION_BOUNDARY instead.
376 (sparc_output_deferred_case_vectors): Likewise.
378 Mon Aug 31 17:25:41 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
380 * config/sparc/sparc.md (movsf_const_intreg): Kill warning.
381 (movtf_insn_sp64, movtf_no_e_insn_sp64): Reorder alternatives.
383 Mon Aug 31 13:57:55 1998 Richard Henderson <rth@cygnus.com>
385 * alpha/va_list.h: New file.
386 * alpha/x-alpha (EXTRA_HEADERS): New. Add va_list.h.
388 Mon Aug 31 14:55:02 1998 Jeffrey A Law (law@cygnus.com)
390 * NEWS: Add SCO Openserver and Unixware 7 notes.
394 Mon Aug 31 15:42:18 1998 Dave Brolley <brolley@cygnus.com>
396 * varasm.c (compare_constant_1): Handle RANGE_EXPR.
397 (record_constant_1): Handle RANGE_EXPR.
399 Mon Aug 31 10:54:03 1998 Richard Henderson <rth@cygnus.com>
401 * print-rtl.c (print_rtx): NOTE_INSN_LIVE has an rtx not a bitmap.
402 * haifa-sched.c (sched_analyze): Handle NOTE_INSN_RANGE_START
403 and NOTE_INSN_RANGE_END specially.
404 (reemit_notes): Likewise.
406 Mon Aug 31 10:18:52 1998 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
408 * sparc.c (TMASK, UMASK): Use `(unsigned)1' not `1U'.
409 (ultrasparc_sched_init): Remove unneeded &.
411 Mon Aug 31 10:47:16 1998 Andreas Schwab <schwab@issan.informatik.uni-dortmund.de>
413 * config/m68k/m68k.h (TARGET_SWITCHES): Don't remove MASK_68040
414 for m68020-60, to prevent the use of fintrz.
416 Sun Aug 30 22:17:20 1998 Mark Mitchell <mark@markmitchell.com>
418 * configure.in: If the native compiler is GCC use $(WARN_CFLAGS)
420 * Makefile.in: Likewise.
421 * configure: Regenerated.
423 Sun Aug 30 22:15:41 1998 H.J. Lu (hjl@gnu.org)
425 * configure.in (gxx_include_dir): Changed to
426 '${prefix}/include/g++'-${libstdcxx_interface}.
427 * configure: Rebuilt.
429 Sun Aug 30 20:19:43 1998 Hans-Peter Nilsson <hp@axis.se>
431 * expr.c (expand_expr): Change ">" to ">=" making MOVE_RATIO use
433 * tm.texi (Costs): Say MOVE_RATIO is number of mem-mem move
434 *sequences* *below* which scalar moves will be used.
436 Sun Aug 30 17:18:43 1998 Jeffrey A Law (law@cygnus.com)
438 * collect2.c (mktemp): Delete unused declaration.
440 * config/xm-netbsd.h: Remove unnecessary file.
441 * config/*/xm-netbsd.h: Do not include the generic xm-netbsd.h
442 file anymore, it is not needed.
444 Sun Aug 30 16:05:45 1998 Mark Mitchell <mark@markmitchell.com>
446 * convert.c (convert_to_integer): Issue an error on conversions to
449 Sun Aug 30 16:47:20 1998 Martin von Lvwis <loewis@informatik.hu-berlin.de>
451 * Makefile.in: Add lang_tree_files and gencheck.h.
452 * configure.in: Generate them.
453 * gencheck.c: Include gencheck.h.
455 Sat Aug 29 21:38:24 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
457 * config/sparc/sparc.md (pic_lo_sum_di, pic_sethi_di): Rename to
458 movdi_lo_sum_pic and movdi_high_pic and make visible.
459 * config/sparc/sparc.c (legitimize_pic_address): For -fPIC,
460 emit these when Pmode is not SImode.
461 * config/sparc/linux64.h (SPARC_DEFAULT_CMODEL): Make CM_MEDLOW.
463 Sat Aug 29 14:59:32 1998 Mumit Khan <khan@xraylith.wisc.edu>
465 * i386/cygwin32.h (ASM_OUTPUT_SECTION_NAME): Don't emit
466 .linkonce directive after the first time.
468 Sat Aug 29 12:39:56 1998 Jeffrey A Law (law@cygnus.com)
470 * m68k.md (beq0_di): Generate correct (and more efficient) code when
471 the clobbered operand overlaps with an input.
472 (bne0_di): Similarly.
474 * Makefile.in (INSTALL): Remove "--no-header" argument.
476 * NEWS: Various updates.
478 Fri Aug 28 19:00:44 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
480 * config/sparc/sparc.c (arith_operand, const64_operand,
481 const64_high_operand, arith_double_4096_operand): Mark mode as
483 (create_simple_focus_bits): Remove unused arg highest_bit_set, all
485 (sparc_emit_set_const64): Remove unused variable i.
486 (sparc_splitdi_legitimate): Likewise for addr_part.
487 (ultra_code_from_mask): Likewise for mask.
488 (ultra_cmove_results_ready_p): Fixup entry modulo calc. and
489 reverse return values so it matches usage and comments.
490 (ultra_flush_pipeline): Likewise.
491 (ultra_fpmode_conflict_exists): Likewise, remove unused variable
492 this_type, and allow loads and stores of differing FP modes as
493 they do not create a conflict.
494 (ultra_find_type): Initialize fpmode to SFmode, fix
495 parenthesization thinkos in large conditional.
496 (ultrasparc_sched_init): Mark dump and sched_verbose as unused.
497 Init free_slot_mask after ultra_cur_hist is reset, not before.
498 (ultrasparc_rescan_pipeline_state): Remove unused variable ucode.
499 (ultrasparc_sched_reorder): Don't bzero current pipeline state,
500 use ultra_flush_pipeline instead, then re-init group pointer.
501 Fix statement with no effect. If no progress made in, and no
502 instructions scheduled at all, advance to new pipeline cycle else
503 we get into an endless loop.
504 (ultrasparc_adjust_cost): Remove previous arg.
505 * config/sparc/sparc.h (ADJUST_COST): Update to reflect that.
507 Fri Aug 28 13:52:35 1998 Jim Wilson <wilson@cygnus.com>
509 * sparc.md (DImode, DFmode, TFmode splits): Delete self_reference
510 code. Use reg_overlap_mentioned_p to detect when source and
512 (negtf2_notv9+1): Use DFmode instead of SFmode in last two operands.
514 1998-08-28 Brendan Kehoe <brendan@cygnus.com>
516 * loop.c (check_dbra_loop): Pass COMPARISON_VALUE, not
517 COMPARISON_VAL, into invariant_p.
519 Fri Aug 28 15:13:25 1998 J"orn Rennecke <amylaar@cygnus.co.uk>
521 * regmove.c (regclass_compatible_p): New function.
522 (regmove_optimize): Use it.
524 Use NREGS parameter insted of calling max_reg_num.
526 (fixup_match_1): Don't use code = MINUS when later tieing with
527 a hard register is likely.
529 Fri Aug 28 14:54:07 1998 J"orn Rennecke <amylaar@cygnus.co.uk>
531 * loop.c (check_dbra_loop): Fix calculation of FINAL_VALUE when
532 COMPARISON_VAL was normalized.
534 Thu Aug 27 20:10:46 1998 Jeffrey A Law (law@cygnus.com)
536 * loop.c (check_dbra_loop): The loop ending comparison value
537 must be an invariant or we can not reverse the loop.
539 * loop.c (scan_loop): Count down from max_reg_num - 1 to
540 FIRST_PSEUDO_REGISTER to avoid calling max_reg_num each iteration
542 (load_mems_and_recount_loop_regs_set): Likewise.
544 * i386.c (print_operand): Remove obsolete 'c' docs.
546 Wed Aug 26 17:13:37 1998 Tom Tromey <tromey@cygnus.com>
548 * gthr.h: Document __GTHREAD_MUTEX_INIT_FUNCTION.
549 * frame.c (init_object_mutex): New function.
550 (init_object_mutex_once): Likewise.
552 (__register_frame_info): Likewise.
553 (__register_frame_info_table): Likewise.
554 (__deregister_frame_info): Likewise.
556 Thu Aug 27 15:14:18 1998 Jeffrey A Law (law@cygnus.com)
558 * haifa-sched.c (sched_analyze_insn): Fix thinko in last change.
560 Thu Aug 27 16:34:51 1998 J"orn Rennecke <amylaar@cygnus.co.uk>
562 * loop.c (check_dbra_loop): Enable code for reversal
563 of some loops without a known constant loop end.
565 Wed Aug 26 18:38:15 1998 Richard Henderson <rth@cygnus.com>
567 * haifa-sched.c (last_clock_var): New.
568 (schedule_block): Initialize it.
569 (schedule_insn): Use it to fill insn modes with issue information.
571 * alpha.c (alpha_handle_trap_shadows): Remove do-nothing exit.
572 Tag trapb and next insn with TImode.
573 (alphaev5_insn_pipe, alphaev5_next_group, alphaev5_align_insns): New.
574 (alpha_reorg): Add conditional for alpha_handle_trap_shadows.
575 Invoke alphaev5_align_insns as appropriate.
576 * alpha.h (LABEL_ALIGN_AFTER_BARRIER): Was ALIGN_LABEL_AFTER_BARRIER.
577 (MD_SCHED_VARIABLE_ISSUE): New.
578 * alpha.md (attr type): Add multi.
579 (define_asm_attributes): New.
580 (prologue_stack_probe_loop, builtin_setjmp_receiver): Set type multi.
581 (arg_home): Likewise.
582 (fnop, unop, realign): New.
584 Wed Aug 26 15:55:41 1998 Jim Wilson <wilson@cygnus.com>
586 * iris5.h (PREFERRED_DEBUGGING_TYPE): Undef.
587 * iris5gas.h (PREFERRED_DEBUGGING_TYPE): Define.
589 * configure.in (powerpc-ibm-aix4.[12]*): Change from 4.[12].*.
590 (rs6000-ibm-aix4.[12]*): Likewise.
591 * configure: Regnerate.
593 Wed Aug 26 09:30:59 1998 Nick Clifton <nickc@cygnus.com>
595 * config/arm/thumb.c (thumb_exit): Do not move a4 into lr if it
596 already contains the return address.
598 Wed Aug 26 12:57:09 1998 Jeffrey A Law (law@cygnus.com)
600 * calls.c (expand_call): Use bitfield instructions to extract/deposit
601 word sized hunks when loading unaligned args into registers.
603 * haifa-sched.c (sched_analyze_insn): Only create scheduling
604 barriers for LOOP, EH and SETJMP notes on the loop_notes list.
606 * mn10300.h (RTX_COSTS): Handle UDIV and UMOD too.
608 Wed Aug 26 16:35:37 1998 J"orn Rennecke <amylaar@cygnus.co.uk>
610 * loop.c (check_dbra_loop): Add some code that would allow reversal
611 of some loops without a known constant loop end if it were enabled.
613 Wed Aug 26 11:08:44 1998 Gavin Romig-Koch <gavin@cygnus.com>
615 * mips.md (lshrsi3_internal2+2): Fix type-o.
617 Wed Aug 26 10:53:03 1998 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
619 * system.h: Include stdarg.h/varargs.h, make sure they are ordered
620 correctly with regards to stdio.h.
622 * calls.c: Remove stdarg.h/varargs.h.
625 * combine.c: Likewise.
626 * cpperror.c: Likewise.
627 * cpplib.c: Likewise.
628 * cpplib.h: Likewise.
629 * doprint.c: Likewise.
630 * emit-rtl.c: Likewise.
632 * fix-header.c: Likewise.
634 * genattr.c: Likewise.
635 * genattrtab.c: Likewise.
636 * gencodes.c: Likewise.
637 * genconfig.c: Likewise.
638 * genemit.c: Likewise.
639 * genextract.c: Likewise.
640 * genflags.c: Likewise.
641 * genopinit.c: Likewise.
642 * genoutput.c: Likewise.
643 * genpeep.c: Likewise.
644 * genrecog.c: Likewise.
645 * mips-tfile.c: Likewise.
646 * prefix.c: Likewise.
647 * protoize.c: Likewise.
648 * regmove.c: Likewise.
649 * toplev.c: Likewise.
652 Wed Aug 26 05:09:27 1998 Jakub Jelinek <jj@sunsite.ms.mff.cuni.cz>
654 * config/sparc/sparc.c (sparc_override_options): If not
655 TARGET_FPU, turn off TARGET_VIS.
656 * config/sparc/sparc.h (TARGET_SWITCHES): Add no-vis.
657 (LEGITIMATE_CONSTANT_P): Allow SF/DF mode zero when TARGET_VIS.
658 * config/sparc/sparc.md (movsi_insn): Use fzeros not fzero.
659 (movdi_insn_sp64): Add VIS fzero alternative.
660 (clear_sf, clear_df): New VIS patterns.
661 (movsf, movdf expanders): Allow fp_zero_operand flat out when
663 (one_cmpldi2_sp64): Provide new fnot1 VIS alternative.
665 Tue Aug 25 10:57:41 1998 Mark Mitchell <mark@markmitchell.com>
667 * loop.c (n_times_set, n_times_used, may_not_optimize,
668 reg_single_usage): Convert to varrays. All uses changed.
669 (insert_loop_mem): Return a value.
670 (scan_loop): Tweak AVOID_CC_MODE_COPIES code.
671 (load_mems_and_recount_loop_regs_set): Likewise. Grow the arrays, if
674 Tue Aug 25 23:57:12 1998 Jeffrey A Law (law@cygnus.com)
677 * configure.in: Do not set thread_file to "irix" since no such
680 * sparc.md (float abs/neg splits): Check reload_completed before
681 calling alter_subreg.
683 Tue Aug 25 19:17:59 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
685 * config/sparc/sparc.c (sparc_absnegfloat_split_legitimate): New
687 * config/sparc/sparc.h: Declare it.
688 * config/sparc/sparc.md (float abs/neg splits): Use it.
689 (all other splits): Handle SUBREGs properly where necessary.
690 (unnamed (1<<x)-1 V8PLUS pattern): Disable for now.
692 Tue Aug 25 19:48:46 1998 Jeffrey A Law (law@cygnus.com)
694 * reorg.c (fill_simple_delay_slots): Do not abort if we encounter
695 an insn on the unfilled_slots_list that has no delay slots.
696 (fill_eager_delay_slots): Similarly.
698 Tue Aug 25 13:35:20 1998 Nick Clifton <nickc@cygnus.com>
700 * config/v850/v850.c (movsi_source_operand): Treat CONSTANT_P_RTX
701 as an ordinary operand.
703 Tue Aug 25 12:54:57 1998 Jason Merrill <jason@yorick.cygnus.com>
705 * tree.c (valid_machine_attribute): Don't apply attributes to both
708 Tue Aug 25 12:23:20 PDT 1998 Richard Henderson <rth@cygnus.com>
710 * reload.c (operands_match_p): Handle rtvecs.
712 * i386.c (legitimate_pic_address_disp_p): New.
713 (legitimate_address_p): Use it.
714 (legitimize_pic_address): Use unspecs to represent @GOT and @GOTOFF.
715 Handle constant pool symbols just like statics.
716 (emit_pic_move): Use Pmode not SImode for clarity.
717 (output_pic_addr_const) [SYMBOL_REF]: Remove @GOT and @GOTOFF hacks.
718 [UNSPEC]: New, handling what we killed above.
719 [PLUS]: Detect and abort on invalid symbol arithmetic.
720 * i386.h (CONSTANT_ADDRESS_P): Remove HIGH.
722 Tue Aug 25 12:02:23 1998 Mark Mitchell <mark@markmitchell.com>
724 * alias.c: Include output.h.
725 (DIFFERENT_ALIAS_SETS_P): Don't treat alias sets as
726 different if we're in a varargs function.
727 * Makefile.in (alias.o): Depend on output.h
729 Tue Aug 25 19:20:12 1998 J"orn Rennecke <amylaar@cygnus.co.uk>
731 * sh.h (GIV_SORT_CRITERION): Delete.
733 Tue Aug 25 13:19:46 1998 Dave Brolley <brolley@cygnus.com>
735 * regclass.c (regclass): Use xmalloc/free instead of alloca.
736 * stupid.c (stupid_life_analysis): Ditto.
737 * reload1.c (reload): Ditto.
739 Tue Aug 25 05:48:18 1998 Jakub Jelinek <jj@sunsite.ms.mff.cuni.cz>
741 * config/sparc/sparc.c (arith_4096_operand, arith_add_operand,
742 arith_double_4096_operand, arith_double_add_operand): New
744 * config/sparc/sparc.h (PREDICATE_CODES): Add them, declare them.
745 * config/sparc/sparc.md (adddi3, addsi3, subdi3, subsi3): Use
746 them to transform add/sub 4096 into add/sub -4096.
748 Mon Aug 24 23:31:03 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
750 * loop.c (scan_loop): Allocate some slop to handle pseudos
751 generated by move_movables.
752 (load_mems_and_recount_loop_regs_set): Honor AVOID_CC_MODE_COPIES
755 Mon Aug 24 19:45:40 1998 Jim Wilson <wilson@cygnus.com>
757 * tree.def (DECL_RESULT): Correct documentation.
759 Tue Aug 25 01:15:27 1998 J"orn Rennecke <amylaar@cygnus.co.uk>
761 * reload1.c (reload_reg_free_before_p): New argument EQUIV; Changed
762 all callers. Abort for RELOAD_FOR_INSN. RELOAD_FOR_OUTADDR_ADDR:
763 conflicts will all RELOAD_FOR_OUTPUT reloads.
765 * reload1.c (reload_cse_regs_1): When deleting a no-op move that
766 loads the function result, substitute with a USE.
768 Mon Aug 24 15:20:19 1998 David Edelsohn <edelsohn@mhpcc.edu>
770 * rs6000.h (GO_IF_LEGITIMATE_ADDRESS): Use TARGET_POWERPC64
771 when testing LEGITIMATE_INDEXED_ADDRESS_P DFmode and DImode.
772 (LEGITIMIZE_ADDRESS): Use TARGET_POWERPC64 for INDEXED fixup.
773 * rs6000.c (print_operand, case 'L'): Add UNITS_PER_WORD, not 4.
774 (print_operand, cases 'O' and 'T'): Fix typos in lossage strings.
775 * rs6000.md (fix_truncdfsi2_store): Remove %w from non-CONST_INT
777 (movdf_softfloat32, movdf_hardfloat64, movdf_softfloat64): Change
778 'o' to 'm' for GPR variant constraints.
780 Mon Aug 24 10:25:46 1998 Jeffrey A Law (law@cygnus.com)
782 * loop.c (scan_loop): Honor AVOID_CC_MODE_COPIES.
784 * h8300.h (STRIP_NAME_ENCODING): Fix typo.
786 * sparc.md (TFmode splits): Use reg_overlap_mentioned_p to detect
787 when the source and destination overlap.
789 * stmt.c (emit_case_nodes): Change rtx_function to rtx_fn to avoid
790 clash with global type.
792 Mon Aug 24 00:53:53 1998 Jason Merrill <jason@yorick.cygnus.com>
794 * fixinc.irix: Add curses.h handling from fixinc.wrap.
796 * c-common.c (combine_strings): Also set TREE_READONLY.
797 Change warn_write_strings to flag_const_strings.
798 * c-decl.c, c-tree.h: Likewise.
800 Sun Aug 23 18:39:11 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
802 * config/sparc/sparc.c (sparc_emit_set_const32): If outputting a
803 CONST_INT, not a symbolic reference, don't use a HIGH/LO_SUM
804 sequence, use SET/IOR instead so CSE can see it.
805 * config/sparc/sparc.md (movhi_const64_special,
806 movsi_const64_special): New patterns necessitated by that change.
807 (movhi_high): Remove.
808 (movhi_lo_sum): Change to match an IOR.
809 (movdf_insn_sp32): Test TARGET_V9 not TARGET_ARCH64.
810 (movdf_insn_v9only): New pattern for when V9 but not ARCH64.
811 (movdf_insn_sp64): Test both TARGET_V9 and TARGET_ARCH64.
812 (movdf splits): Allow when not V9 or when not ARCH64 and integer
813 registers are involved.
814 (snesi_zero_extend split): Remove reload_completed test.
815 (unnamed plus and minus zero_extend sidi splits): Add it.
817 Sun Aug 23 11:56:08 1998 Mark Mitchell <mark@markmitchell.com>
819 * extend.texi: Remove description of extension to explicit
820 instantiation that is now endorsed by standard C++.
822 Sun Aug 23 09:39:09 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
824 * config/arc/arc.c (arc_initialize_pic): Remove.
825 * config/arc/arc.h (INITIALIZE_PIC): Similarly, this routine does
826 nothing on any platform and is invoked by no-one, it does not even
827 appear in the documentation.
828 * config/sparc/sparc.h (INITIALIZE_PIC): Likewise.
829 * config/sparc/sparc.c (initialize_pic): Likewise.
830 (find_addr_reg): Remove this as well, no longer referenced after
833 Sun Aug 23 00:17:14 1998 Jeffrey A Law (law@cygnus.com)
835 * recog.c (validate_replace_rtx_group): New function.
836 * recog.h (validate_replace_rtx_group): Declare it.
837 * regmove.c (optimize_reg_copy_3): If any substitution fails, then undo
838 the entire group of substitutions.
840 Sat Aug 22 23:31:00 1998 Klaus-Georg Adams (Klaus-Georg.Adams@chemie.uni-karlsruhe.de)
842 * loop.c (load_mems): Fix initializers.
844 Fri Aug 21 23:07:46 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
846 * config/sparc/sparc.md (TFmode splits): Handle destination
847 registers being referenced in the address correctly.
849 * expmed.c (make_tree) [CONST_INT]: Sign extend even if
850 TREE_UNSIGNED, when bitsize of type's mode is larger than
851 HOST_BITS_PER_WIDE_INT.
853 Fri Aug 21 19:31:31 1998 Alexandre Petit-Bianco <apbianco@cygnus.com>
855 * tree.def (LABELED_BLOCK_EXPR, EXIT_BLOCK_EXPR): New tree nodes.
856 * tree.h (LABELED_BLOCK_LABEL, LABELED_BLOCK_BODY,
857 EXIT_BLOCK_LABELED_BLOCK, EXIT_BLOCK_RETURN, LOOP_EXPR_BODY): New
859 * expr.c (expand_expr): Handle LABELED_BLOCK_EXPR and
862 Thu Aug 20 19:43:44 1998 Jeffrey A Law (law@cygnus.com)
864 * h8300.c (h8300_encode_label): Use '&' for tiny data items.
865 * h8300.h (TINY_DATA_NAME_P): Likewise.
866 (STRIP_NAME_ENCODING): Handle '&'.
868 * mn10200.h (REG_OK_FOR_INDEX_P): Do not check the mode of the
869 register (it could be accessed via an outer SUBREG).
870 (REG_OK_FOR_BASE_P): Likewwise.
871 (GO_IF_LEGITIMATE_ADDRESS): Consistently use REGNO_OK_FOR_BASE_P.
873 * remove.c (optimize_reg_copy_3): Abort instead of silently generating
876 * jump.c (rtx_renumbered_equal_p): Do not consider PLUS commutative.
878 Thu Aug 20 17:35:20 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
880 * config/sparc/sparc.md (movtf_insn_sp32): All memory operands
881 must be offsettable so the splits can be made.
883 Thu Aug 20 13:56:53 1998 Michael Meissner <meissner@cygnus.com>
885 * config/i386/winnt.c: Include system.h, not stdio.h to get
886 sys/param.h pulled in before rtl.h in case the system defines MIN
889 Thu Aug 20 13:44:20 1998 David Edelsohn <edelsohn@mhpcc.edu>
891 * rs6000.md (movqi, movhi): Add CONSTANT_P_RTX.
893 Thu Aug 20 13:15:11 1998 Dave Brolley <brolley@cygnus.com>
895 * stor-layout.c (layout_type): Compute TYPE_SIZE_UNIT correctly for
897 * cpplib.c (cpp_define): Handle macros with parameters.
899 Wed Aug 19 21:33:19 1998 David Edelsohn <edelsohn@mhpcc.edu>
901 * rs6000.c (rs6000_output_load_toc_table): Use ld for 64-bit.
902 (output_toc): Use single TOC slot or llong minimal-toc for DFmode
903 and DImode 64-bit. Use llong for minimal-toc SFmode and
904 SYMBOL_REF / LABEL_REF 64-bit.
905 (output_function_profiler): Use llong for profiler label and ld to
906 load 64-bit label address.
908 Wed Aug 19 17:52:27 1998 Nick Clifton (nickc@cygnus.com)
910 * config/arm/thumb.md (extendqisi2_insn): Cope with REG +
913 Wed Aug 19 14:13:31 PDT 1998 Jeff Law (law@cygnus.com)
915 * version.c: Bump for snapshot.
917 Wed Aug 19 13:10:30 PDT 1998 Jeff Law (law@cygnus.com)
919 * version.c: Bump for snapshot.
921 Wed Aug 19 13:06:47 1998 Jason Merrill <jason@yorick.cygnus.com>
923 * collect2.c (extract_init_priority): Use atoi instead of strtoul.
925 Wed Aug 19 13:51:35 1998 Hans-Peter Nilsson <hp@axis.se>
927 * tm.texi (Misc): Fix typo "teh".
929 * tm.texi (PIC): Fix typo "PPIC".
931 * tm.texi (Caller Saves): Say that DEFAULT_CALLER_SAVES has no
932 effect when -O2 and higher.
933 * invoke.texi (Optimize Options): Likewise for -fcaller-saves.
935 1998-08-19 Michael Hayes <michaelh@ongaonga.chch.cri.nz>
937 * regclass.c: Changed register set documentation to be consistent
940 * final.c (final_start_function) Removed redundant test for
943 Wed Aug 19 13:28:41 1998 Mark Mitchell <mark@markmitchell.com>
945 * rtl.h (rtx_function): New type.
946 (for_each_rtx): New function.
947 * rtlanal.c (for_each_rtx): Define it.
949 * recog.c (change_t): New type.
950 (change_objects, change_old_codes, change_locs, change_olds):
952 (changes): New variable.
953 (validate_change): Dynamically allocate room for more changes, if
954 necessary. Uses changes array instead of change_objects, etc.
955 (apply_change_group): Use changes array instead of
958 * loop.c (loop_mem_info): New type.
959 (loop_mems): New variable.
960 (loop_mems_idx): Likewise.
961 (looop_mems_allocated): Likewise.
962 (scan_loop): Remove nregs parameter.
963 (next_insn_in_loop): New function.
964 (load_mems_and_recount_loop_regs_set): Likewise.
965 (load_mems): Likewise.
966 (insert_loop_mem): Likewise.
967 (replace_loop_mem): Likewise.
968 (replace_label): Likewise.
969 (INSN_IN_RANGE_P): New macro.
970 (loop_optimize): Don't pass max_reg_num() to scan_loop.
971 (scan_loop): Remove nregs parameter, compute it after any new
972 registers are created by load_mems. Use INSN_IN_RANGE_P and
973 next_insn_in_loop rather than expanding them inline. Call
974 load_mems to load memory into pseudos, if appropriate.
975 (prescan_loop): Figure out whether or not there are jumps from the
976 loop to targets other than the label immediately following the
977 loop. Call insert_loop_mem to notice all the MEMs used in the
978 loop, if it could be safe to pull MEMs into REGs for the duration
980 (strength_reduce): Use next_insn_in_loop. Tweak comments.
982 Wed Aug 19 08:29:44 1998 Richard Earnshaw (rearnsha@arm.com)
984 * arm.c (arm_override_options): Remove lie about ignoring PIC flag.
986 Wed Aug 19 07:08:15 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
988 * config/sparc/sparc.c (finalize_pic): Check for the correct
989 nonlocal_goto_receiver UNSPEC number.
990 * config/sparc/sparc.md (nonlocal_goto_receiver): Add comment
991 making note of this dependency existing in sparc.c
992 (negtf2_notv9 split): Give NEG SFmode.
993 (negsf2): Fix insn output string.
995 Tue Aug 18 12:40:27 1998 Richard Henderson <rth@cygnus.com>
997 * c-common.c (decl_attributes): Issue an error if the argument
998 to alias is not a string.
1000 Tue Aug 18 10:33:30 1998 Jeffrey A Law (law@cygnus.com)
1002 * haifa-sched.c (sched_analyze): Put all JUMP_INSNs on the last
1003 pending memory flush list.
1005 * combine.c (can_combine_p): Allow combining insns with REG_RETVAL
1007 (try_combine): Allow combining insns with REG_LIBCALL notes.
1009 * expr.c (emit_block_move): Do not call memcpy as a libcall
1010 instead build up a CALL_EXPR and call it like any other
1012 (clear_storage): Similarly for memset.
1014 * regmove.c (fixup_match_2): Do not call reg_overlap_mentioned_p
1017 * Makefile.in (cplus-dem.o): Provide explicit rules for building
1020 * regmove.c (optimize_reg_copy_1): Update REG_N_CALLS_CROSSED
1021 and REG_LIVE_LENGTH as successful substitutions are made.
1023 Tue Aug 18 07:15:27 1998 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
1025 * config/sparc/sparc.c (ultra_find_type): Add empty semicolon
1026 statement after end of loop label.
1028 Tue Aug 18 07:13:27 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
1030 * config/sparc/sparc.c (ultra_types_avail): New variable.
1031 (ultra_build_types_avail): New function to record mask of insn
1032 types in ready list at this cycle.
1033 (ultrasparc_sched_reorder): Call it.
1034 (ultra_find_type): Use it to quicken the search. Also simplif
1035 dependency check, don't use rtx_equal_p because we know exactly
1036 what we are looking for.
1038 Tue Aug 18 03:20:53 1998 Richard Earnshaw (rearnsha@arm.com)
1040 * arm.h (SECONDARY_INPUT_RELOAD_CLASS): Return NO_REGS if compiling
1041 for architecture v4.
1043 Mon Aug 17 21:26:38 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
1045 * config/sparc/sparc.md (sltu, sgeu): Don't FAIL, call
1047 (movsf_const_intreg, movsf_const_high, movsf_const_lo,
1048 movdf_const_intreg and helper splits): New patterns to move float
1049 constants into integer registers.
1050 (negtf2, negdf2, abstf2, absdf2): Rework using new patterns and
1053 Mon Aug 17 11:46:19 1998 Jeffrey A Law (law@cygnus.com)
1056 * tree.c (build_index_type): Copy TYPE_SIZE_UNIT from sizetype
1058 * c-decl.c (finish_enum): Copy TYPE_SIZ_UNIT from enumtype to tem.
1060 * rs6000.c (secondary_reload_class): For TARGET_ELF, indicate that
1061 a BASE_REGS register is needed as an intermediate when copying
1062 a symbolic value into any register class other than BASE_REGS.
1064 * expr.c (move_by_pieces): No longer static. Remove prototype.
1065 * rtl.h (move_by_pieces): Add extern prototype.
1066 * mips.c (expand_block_move): Handle aligned straight line copy by
1067 calling move_by_pieces.
1069 * expr.c (expand_expr): Allow assignments from TImode PARM_DECLs
1072 Mon Aug 17 10:28:52 1998 Mark Mitchell <mark@markmitchell.com>
1074 * stmt.c (expand_end_loop): Tidy. Allow unconditional
1075 jumps out of the loop to be treated as part of the exit test.
1077 Mon Aug 17 10:06:11 1998 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
1078 Jeff Law <law@cygnus.com>
1080 * Makefile.in (cplus-dep.o): Use cplus-dem.c from libiberty.
1081 * cplus-dem.c: Delete.
1083 * Makefile.in (fold-const.o): depend on $(RTL_H).
1085 * fold-const.c: Include rtl.h to get the prototype for
1086 `set_identifier_local_value'.
1088 * loop.c (express_from_1): Remove unused variable `tmp'.
1089 (combine_givs): cast the first argument of bzero to char *.
1091 * toplev.c (display_help): Remove unused variable `looking_for_start'.
1093 * c-decl.c (init_decl_processing): Remove unneeded &.
1095 * alpha.h (alpha_initialize_trampoline): Provide prototype.
1097 * except.c (set_exception_lang_code, set_exception_version_code):
1098 Change parameter from `short' to `int' to avoid using a gcc
1101 * except.h (set_exception_lang_code, set_exception_version_code):
1102 Likewise for prototypes.
1104 * flow.c (count_reg_references): Remove unused variables `regno'
1107 * gcse.c (hash_scan_insn): Declare parameter `in_libcall_block'.
1109 * prefix.c (translate_name): Cast the result of `alloca'.
1111 * varray.h (VARRAY_FREE): Reimplement as a `do-while(0)' statement.
1113 Mon Aug 17 09:23:42 1998 Andreas Schwab <schwab@issan.informatik.uni-dortmund.de>
1115 * config/m68k/m68k.c: Include "system.h" instead of <stdio.h>.
1117 (valid_dbcc_comparison_p): Mark mode argument as unused.
1118 (symbolic_operand): Likewise.
1119 (legitimize_pic_address): Likewise.
1120 (const_uint32_operand): Likewise.
1121 (const_sint32_operand): Likewise.
1122 * sched.c [!INSN_SCHEDULING]: Define only dummy function
1123 schedule_insns and comment out rest of file.
1125 * m68k.c (output_move_simode_const): Use subl to move a zero into an
1127 (output_move_[hq]imode): Likewise.
1129 Mon Aug 17 09:15:47 1998 Jeffrey A Law (law@cygnus.com)
1131 * toplev.c (main): Enable -fstrict-aliasing for -O2 and above.
1132 * invoke.texi: Corresponding changes.
1134 Mon Aug 17 02:03:55 1998 Richard Henderson <rth@cygnus.com>
1136 * regclass.c (allocate_reg_info): Respect MIN when clearing data.
1138 Sun Aug 16 17:37:06 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
1140 * config/sparc/sparc.c (ultra_code_from_mask,
1141 ultra_cmove_results_ready_p, ultra_fpmode_conflict_exists,
1142 ultra_find_type, ultra_schedule_insn, ultra_flush_pipeline,
1143 ultrasparc_sched_init, ultrasparc_variable_issue,
1144 ultra_rescan_pipeline_state, ultrasparc_sched_reorder): New
1145 functions to describe UltraSPARC pipeline exactly to Haifa.
1146 (ultrasparc_adjust_cost): Indicate IMUL type insns have zero cost,
1147 as there is nothing the scheduler can do about it. Indicate that
1148 REG_DEP_OUTPUT's collide. Fixup formatting.
1149 * config/sparc/sparc.h (RTX_COSTS): Fixup integer multiply and
1150 divide costs on Ultra for DImode.
1151 (MD_SCHED_INIT, MD_SCHED_REORDER, MD_SCHED_VARIABLE_ISSUE):
1153 * config/sparc/sparc.md (ieu_unnamed function unit): Rename to
1154 ieuN and add call_no_delay_slot to type list.
1155 (cti function unit): New unit for branches on UltraSPARC.
1156 (subx/addx insns): Set type to misc.
1157 (sidi zero/sign extension insns on arch64): Set type to shift.
1158 (sign_extendhidi2_insn): Set type to sload.
1160 Sun Aug 16 13:52:00 1998 David Edelsohn <edelsohn@mhpcc.edu>
1162 * rs6000.c (rs6000_stack_info): Use if == 0 for sizes.
1163 (output_epilog): Use if != 0 for offset.
1164 (rs6000_fatal_bad_address): Prepare for Intl.
1165 * rs6000.h (rs6000_fatal_bad_address): Declare.
1166 * rs6000.md (movsfcc, movdfcc): Use else if.
1167 (elf_high): Use {liu|lis}.
1168 (elf_low): Use {cal|la}. Remove %a template from old mnemonics.
1169 (movsi): Use rs6000_fatal_bad_address.
1171 Sun Aug 16 01:53:21 1998 Richard Henderson <rth@cygnus.com>
1173 * reload.c (find_equiv_reg): Reject equivalences separated
1174 by a volatile instruction.
1176 Sun Aug 16 00:21:44 1998 Franz Sirl <Franz.Sirl-kernel@lauterbach.com>
1178 * rs6000/linux.h (CPP_OS_DEFAULT_SPEC): Define.
1180 Sat Aug 15 20:51:35 1998 Richard Henderson <rth@cygnus.com>
1182 * alpha.md (movsicc): Fix mode mismatch.
1184 Sat Aug 15 20:22:33 1998 H.J. Lu (hjl@gnu.org)
1186 * config/alpha/alpha.h (ASM_OUTPUT_MI_THUNK): Handle aggregated
1188 * config/alpha/win-nt.h (ASM_OUTPUT_MI_THUNK): Likewise.
1190 Sat Aug 15 08:39:49 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
1192 * config/sparc/sparc.md (movsi_lo_sum_pic_label_reg): Remove
1193 write-only modifier from operand 1 constraint.
1195 Sat Aug 15 06:28:19 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
1197 * config/sparc/sparc.c (sparc_emit_set_const64_quick1): If
1198 emitting a XOR of -1 at the end, emit a NOT instead for combine's
1200 (sparc_emit_set_const64): Likewise, also when computing trailing
1201 bits do not negate low_bits and make fast_int an int.
1203 Fri Aug 14 21:07:03 1998 Jeffrey A Law (law@cygnus.com)
1205 * loop.c (add_label_notes): Do not ignore references to labels
1206 before dispatch tables. Mirrors Apr 8 change to mark_jump_label.
1207 * gcse.c (add_label_notes): Similarly.
1209 * pa.h (ASM_OUTPUT_MI_THUNK): Strip name encoding.
1211 * m68k.md (adddi_dilshr32): One of the operands must be a register.
1212 (adddi_dishl32): Similarly.
1214 Fri Aug 14 14:12:59 1998 Jason Merrill <jason@yorick.cygnus.com>
1216 * i386.h (MODES_TIEABLE_P): Reorganize to shut up warnings.
1217 * alias.c (memrefs_conflict_p): Add braces to shut up warnings.
1218 * cse.c (cse_basic_block): Add parens to shut up warnings.
1220 Fri Aug 14 12:58:21 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
1222 * config/sparc/sparc.c (sparc_emit_set_const64_quick2,
1223 sparc_emit_set_const64_longway, const64_is_2insns,
1224 create_simple_focus_bits, sparc_emit_set_const64): Fix more bugs
1225 in 64-bit constant formation.
1226 * config/sparc/sparc.md (snesi_zero_extend split): Generate
1227 rtl for addx not subx.
1228 (define_insn movdi_const64_special): Make available even when
1229 HOST_BITS_PER_WIDE_INT is not 64.
1230 (movdi_lo_sum_sp64_cint, movdi_high_sp64_cint): Remove.
1231 (losum_di_medlow, sethm, setlo): Make op2 symbolic_operand.
1232 (cmp_siqi_trunc_set, cmp_diqi_trunc_set): Encapsulate both
1233 instances of operand 1 inside a QI subreg.
1234 (xordi3_sp64_dbl): Remove '%' constraint for op1.
1235 (one_cmpldi2_sp64): Fix output string.
1236 (one_cmplsi2_not_liveg0): Rewrite to remove unneeded extra
1238 (unnamed arch64 ashift DI): Truncate shift count if greater than
1241 Fri Aug 14 21:52:53 1998 J"orn Rennecke <amylaar@cygnus.co.uk>
1243 * expr.c (store_expr): Don't optimize away load-store pair
1244 when either source or destination have a side effect.
1246 Fri Aug 14 16:50:10 1998 John Carr <jfc@mit.edu>
1248 * genrecog.c (add_to_sequence): Fatal error if the modes of the
1249 operands of SET are incompatible.
1251 * alpha.md: Fix max and min patterns so modes of SET operands match.
1253 Fri Aug 14 12:22:55 1998 Ian Lance Taylor <ian@cygnus.com>
1255 * configure.in: Avoid [[ by using test and changequote.
1256 * configure: Rebuild.
1258 Fri Aug 14 01:22:31 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
1260 * rtl.def (CONSTANT_P_RTX): Fix typo in string name.
1262 * config/sparc/sparc.md (seqdi_special_trunc, snedi_special_trunc,
1263 seqsi_special_extend, snesi_special_extend, snesi_zero_extend and
1264 split, snedi_zero_trunc and split, seqsi_zero_extend and split,
1265 seqdi_zero_trunc and split, pic_lo_sum_di, pic_sethi_di,
1266 movdi_cc_sp64_trunc, movdi_cc_reg_sp64_trunc, addx_extend_sp32 and
1267 split, addx_extend_sp64, subx_extend_sp64, subx_extend and split):
1268 Fix mismatching modes in SET operands.
1269 (conditional move patterns): Fix formatting.
1270 (unnamed subx arch64 pattern): Remove duplicate insn.
1272 Fri Aug 14 00:34:34 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
1274 * config/sparc/sparc.c (const64_operand, const64_high_operand):
1275 Get it right when HOST_BITS_PER_WIDE_INT is not 64.
1276 (input_operand): Fixup test for what we accept for constant
1278 (sparc_emit_set_const32, sparc_emit_set_symbolic_const64): Give
1280 (safe_constDI): Remove.
1281 (sparc_emit_set_safe_HIGH64, gen_safe_SET64, gen_safe_OR64,
1282 gen_safe_XOR64): New functions.
1283 (sparc_emit_set_const64_quick1, sparc_emit_set_const64_quick2,
1284 sparc_emit_set_const64_longway, sparc_emit_set_const64): Use
1286 * config/sparc/sparc.md (define_insn xordi3_sp64_dbl): Only make
1287 available when HOST_BITS_PER_WIDE_INT is not 64.
1288 (define_insn movdi_sp64_dbl, movdi_const64_special): Likewise and
1289 move before movdi_insn_sp64 pattern.
1290 (define_insn movdi_lo_sum_sp64_dbl, movdi_high_sp64_dbl): Remove.
1291 (define_insn sethi_di_medlow, seth44, setm44, sethh): Use
1292 symbolic_operand as predicate for second operand.
1293 (DImode minus split on arch32, negsi2 expander, one_cmplsi2
1294 expander): Give set VOIDmode.
1296 Fri Aug 14 01:45:06 1998 Mumit Khan <khan@xraylith.wisc.edu>
1298 * i386/cygwin32 (DEFAULT_PCC_STRUCT_RETURN): Define.
1300 Fri Aug 14 01:40:21 1998 Geoffrey Keating <geoffk@ozemail.com.au>
1302 * rs6000/linux.h (LINK_SPEC): Pass -G args to the linker.
1304 Fri Aug 14 01:23:23 1998 Richard Earnshaw (rearnsha@arm.com)
1306 * arm/netbsd.h (TARGET_DEFAULT): Default includes software floating
1308 (CPP_FLOAT_DEFAULT_SPEC): Re-define accordingly.
1310 Fri Aug 14 01:19:08 1998 Robert Lipe <robertl@dgii.com>
1312 * install.texi: Various SCO OpenServer tweaks.
1314 Thu Aug 13 20:14:40 1998 Jim Wilson <wilson@cygnus.com>
1316 * reload1.c (eliminate_regs_in_insn): Handle another case when
1317 eliminating the frame pointer to the hard frame pointer. Add
1318 missing ep->to_rtx check to one existing case.
1320 * mips/mips.md (movhi_internal2+2): Fix typo mem:SI -> mem:HI.
1322 Thu Aug 13 17:08:11 1998 Jason Merrill <jason@yorick.cygnus.com>
1324 * tree.h: De-conditionalize init_priority code.
1326 * mips.h (NM_FLAGS): Change from -Bp to -Bn.
1327 * collect2.c (NM_FLAGS): Change from -p to -n.
1329 * configure.in: Turn on collect2 for mipstx39-elf.
1330 Handle use_collect2=no properly.
1332 * c-common.c: De-conditionalize init_priority code.
1333 * collect2.c (extract_init_priority, sort_ids): New fns.
1334 (main): Call sort_ids.
1335 Move sequence_number to file scope.
1337 * configure.in: Handle --enable-init-priority.
1338 * c-common.c (attrs): Add A_INIT_PRIORITY.
1339 (init_attributes, decl_attributes): Likewise.
1340 * tree.h (DEFAULT_INIT_PRIORITY, MAX_INIT_PRIORITY): New macros.
1341 * tree.c (get_file_function_name_long): Split out...
1342 (get_file_function_name): ...from here.
1344 Thu Aug 13 16:09:53 1998 Martin von Loewis <loewis@informatik.hu-berlin.de>
1346 * expr.c (safe_from_p): Change code to ERROR_MARK only when not
1349 Thu Aug 13 15:24:48 1998 Jason Merrill <jason@yorick.cygnus.com>
1351 * toplev.c (display_help): Add braces to shut up warnings.
1352 * tree.c (simple_cst_equal): Likewise.
1354 * fold-const.c (non_lvalue): Don't deal with null pointer
1356 (fold, case COMPOUND_EXPR): Wrap a constant 0 in a NOP_EXPR.
1358 * c-typeck.c (initializer_constant_valid_p): Allow conversion of 0
1359 of any size to a pointer.
1361 Thu Aug 13 12:53:13 1998 Jim Wilson <wilson@cygnus.com>
1363 * i386/winnt.c (i386_pe_asm_file_end): Check TREE_SYMBOL_REFERENCED.
1365 Wed Aug 12 17:25:18 1998 Jeffrey A Law (law@cygnus.com)
1367 * mn10300.c (REG_SAVE_BYTES): Only reserve space for registers
1368 which will be saved.
1369 * mn10300.md (prologue insn): Only save registers which need saving.
1370 (epilogue insn): Similarly.
1372 * mn10300.c, mn10300.h, mn10300.md: Remove "global zero register"
1375 Wed Aug 12 12:39:16 1998 Gavin Romig-Koch <gavin@cygnus.com>
1377 * mips/mips.h (ENCODE_SECTION_INFO): Set SYMBOL_REF_FLAG for
1378 VAR_DECL's in gp addressable sections.
1380 Tue Aug 11 23:02:31 1998 John Carr <jfc@mit.edu>
1382 * sparc.c: Change return <exp> to <exp>; return; in functions
1384 * sparc.md: Add empty semicolon statement after final label in
1387 Tue Aug 11 22:42:01 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
1389 * config/sparc/sparc.md (define_insn addx_extend): Rename to
1390 addx_extend_sp64, only allow when TARGET_ARCH64.
1391 (define_insn addx_extend_sp32 and split): Version that works when
1393 (define_insn subx_extend): Likewise.
1394 (define_split adddi3 and subdi3 with zero extension): Fixup and
1395 correct bugs when not TARGET_ARCH64.
1397 Tue Aug 11 16:04:34 1998 John Carr <jfc@mit.edu>
1399 * except.c (set_exception_lang_code, set_exception_version_code):
1400 Use prototype-style definition if __STDC__, to match declaration
1403 * genemit.c: Change FAIL and DONE macros not to use loops.
1405 Tue Aug 11 12:27:03 1998 Jim Wilson <wilson@cygnus.com>
1407 * dwarf2out.c (ASM_OUTPUT_DWARF_ADDR_CONST): Use
1408 ASM_OUTPUT_DWARF2_ADDR_CONST if defined.
1410 * mips/mips.md (reload_outsi): Use M16_REG_P when TARGET_MIPS16.
1412 Tue Aug 11 18:12:53 1998 Dave Love <d.love@dl.ac.uk>
1414 * README.g77: Update from Craig.
1416 Tue Aug 11 04:46:01 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
1418 * config/sparc/sparc.c (sparc_emit_set_const32): INTVAL is of
1420 (safe_constDI sparc_emit_set_const64_quick1,
1421 sparc_emit_set_const64_quick2, sparc_emit_set_const64_longway,
1422 analyze_64bit_constant, const64_is_2insns,
1423 create_simple_focus_bits): Fix some bugs when compiled on real
1425 (function_arg_record_value_3, function_arg_record_value_2,
1426 function_arg_record_value): Add fully prototyped forward decls.
1427 * config/sparc/sparc.md (define_insn cmpsi_insn_sp32): Rename back
1428 to cmpsi_insn and use on both 64 and 32 bit targets.
1429 (define_insn cmpsi_insn_sp64): Remove.
1430 (define_expand zero_extendsidi2): Allow for 32-bit target too.
1431 (define_insn zero_extendsidi2_insn): Rename to
1432 zero_extendsidi2_insn_sp64.
1433 (define_insn zero_extendsidi2_insn_sp32): New pattern and
1434 assosciated forced split for it.
1436 * config/sparc/sparc.c (const64_operand, const64_high_operand):
1438 * config/sparc/sparc.h: Declare them.
1439 (PREDICATE_CODES): Add them.
1440 * config/sparc/sparc.md (movdi_lo_sum_sp64_dbl,
1441 movdi_high_sp64_dbl, xordi3_sp64_dbl): Use them.
1443 Mon Aug 10 22:57:24 1998 John Carr <jfc@mit.edu>
1445 * config/sparc/sparc.md (define_insn jump): Output ba,pt not b,pt
1446 in v9 case as the latter makes the Solaris assembler crash.
1448 Mon Aug 10 22:39:09 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
1450 * config/sparc/sparc.c (input_operand): Do not accept a LO_SUM MEM
1451 for TFmode when !v9. We require offsettable memory addresses.
1452 * config/sparc/sparc.h (ALTER_HARD_SUBREG): Handle TFmode to
1453 DFmode register number conversions.
1454 * config/sparc/sparc.md (define_split DFmode moves): If register
1455 is a SUBREG do alter_subreg on it before using.
1456 (define_expand movtf): Fixup comment about alignment on v9.
1457 (define_split TFmode moves): Don't use gen_{high,low}part, create
1458 explicit SUBREGs instead.
1460 Mon Aug 10 19:02:55 1998 John Carr <jfc@mit.edu>
1462 * Makefile.in (mbchar.o): Depend on mbchar.c.
1464 Mon Aug 10 04:28:13 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
1465 Richard Henderson <rth@cygnus.com>
1467 Rewrite Sparc backend for better code generation and
1468 improved sparc64 support.
1469 * config/sparc/sp64-elf.h: Set JUMP_TABLES_IN_TEXT_SECTION to
1471 * config/sparc/sysv4.h: Likewise.
1472 * config/sparc/sparc.c (v8plus_regcmp_p, sparc_operand,
1473 move_operand, v8plus_regcmp_op, emit_move_sequence,
1474 singlemove_string, doublemove_string, mem_aligned_8,
1475 output_move_double, output_move_quad, output_fp_move_double,
1476 move_quad_direction, output_fp_move_quad, output_scc_insn):
1478 (small_int_or_double): New predicate.
1479 (gen_compare_reg): Remove TARGET_V8PLUS cmpdi_v8plus emission.
1480 (legitimize_pic_address): Emit movsi_{high,lo_sum}_pic instead of
1481 old pic_{sethi,lo_sum}_si patterns.
1482 (mem_min_alignment): New generic function to replace
1483 mem_aligned_8, which uses REGNO_POINTER_ALIGN information when
1484 available and can test for arbitrary alignments. All callers
1486 (save_regs, restore_regs, build_big_number,
1487 output_function_prologue, output_cbranch, output_return,
1488 sparc_flat_save_restore, sparc_flat_output_function_prologue,
1489 sparc_flat_output_function_epilogue): Prettify
1491 (output_function_epilogue): Likewise and add code to output
1492 deferred case vectors.
1493 (output_v9branch): Likewise, add new arg INSN and use it to tack
1494 on branch prediction settings. All callers changed.
1495 (print_operand): Likewise and output %l44 for LO_SUMs when
1497 (sparc_splitdi_legitimate): New function to make sure DImode
1498 splits can be run properly when !arch64.
1499 (sparc_initialize_trampoline, sparc64_initialize_trampoline):
1500 Reformat example code in comments.
1501 (set_extends): Remove UNSPEC/v8plus_clear_high case.
1502 (sparc_addr_diff_list, sparc_addr_list): New statics to keep track
1503 of deferred case vectors we need to output.
1504 (sparc_defer_case_vector): Record a case vector.
1505 (sparc_output_addr_vec, sparc_output_addr_diff_vec,
1506 sparc_output_deferred_case_vectors): New functions to output them.
1507 (sparc_emit_set_const32): New function to form 32-bit constants in
1508 registers when that requires more than one instruction.
1509 (safe_constDI, sparc_emit_set_const64_quick1,
1510 sparc_emit_set_const64_quick2, sparc_emit_set_const64_longway,
1511 analyze_64bit_constant, const64_is_2insns,
1512 create_simple_focus_bits, sparc_emit_set_const64): New functions
1513 which do the same for 64-bit constants when arch64.
1514 (sparc_emit_set_symbolic_const64): New function to emit address
1515 loading for all code models on v9.
1516 * config/sparc/sparc.h (CONDITIONAL_REGISTER_USAGE): Do not make
1517 %g1 fixed when arch64, unfix %g0 when TARGET_LIVE_G0.
1518 (ALTER_HARD_SUBREG): Fix thinko, return REGNO + 1 not 1.
1519 (SECONDARY_INPUT_RELOAD_CLASS, SECONDARY_OUTPUT_RELOAD_CLASS): Fix
1520 inaccuracies in comments, add symbolic and text_segment operands
1521 when TARGET_CM_MEDANY and TARGET_CM_EMBMEDANY respectively. Use
1522 GENERAL_REGS in these cases as a temp REG is needed to load these
1523 addresses into a register properly.
1524 (EXTRA_CONSTRAINT): Document more accurately, remove Q case as it
1526 (GO_IF_LEGITIMATE_ADDRESS): Allow TFmode for LO_SUM on v9 since fp
1527 quads are guarenteed to have 16-byte alignment.
1528 (LEGITIMIZE_ADDRESS): For SYMBOL_REF, CONST, and LABEL_REF use
1529 copy_to_suggested_reg instead of explicit LO_SUM and HIGH.
1530 (ASM_OUTPUT_ADDR_VEC, ASM_OUTPUT_ADDR_DIFF_VEC): New macros for
1531 deferred case vector implementation.
1532 (ASM_OUTPUT_ADDR_VEC_ELT): Use fputc to output newline.
1533 (ASM_OUTPUT_ADDR_DIFF_ELT): Parenthesize LABEL in macro calls.
1534 Generate "internal label - label" instead of "label - 1b".
1535 (PRINT_OPERAND_ADDRESS): For LO_SUM use %l44 on TARGET_CM_MEDMID.
1536 (PREDICATE_CODES): Remove sparc_operand, move_operand,
1537 v8plus_regcmp_op. Add small_int_or_double, input_operand, and
1539 (doublemove_string, output_block_move, output_fp_move_double,
1540 output_fp_move_quad, output_move_double, output_move_quad,
1541 output_scc_insn, singlemove_string, mem_aligned_8, move_operand,
1542 sparc_operand, v8plus_regcmp_op, v8plus_regcmp_p): Remove externs.
1543 (sparc_emit_set_const32, sparc_emit_set_const64,
1544 sparc_emit_set_symbolic_const64, input_operand, zero_operand,
1545 mem_min_alignment, small_int_or_double): Add externs.
1546 * config/sparc/sparc.md: Document the many uses of UNSPEC and
1547 UNSPEC_VOLATILE in this backend.
1548 (define_function_unit ieu): Rename to ieu_unnamed. Add move and
1549 unary to types which execute in it.
1550 (define_function_unit ieu_shift): Rename to ieu0.
1551 (define_function_unit ieu1): New, executes compare, call, and
1552 uncond_branch type insns.
1553 (define_function_units for type fdivs, fdivd, fsqrt): These
1554 execute in the fpu multiply unit not the adder on UltraSparc.
1555 (define_expand cmpdi): Disallow TARGET_V8PLUS.
1556 (define_insn cmpsi_insn): Rename to cmpsi_insn_sp32.
1557 (define_insn cmpsi_insn_sp64): New, same as sp32 variant except it
1558 allows the arith_double_operand predicate and rHI constraint when
1560 (define_insn cmpdi_sp64, cmpsf_fpe, cmpdf_fpe, cmptf_fpe,
1561 cmpsf_fp, cmpdf_fp, cmptf_fp, sltu_insn, neg_sltu_insn,
1562 neg_sltu_minux_x, neg_sltu_plus_x, sgeu_insn, neg_sgeu_insn,
1563 sltu_plus_x, sltu_plus_x, sltu_plus_x_plus_y, x_minus_sltu,
1564 sgeu_plus_x, x_minus_sgeu, movqi_cc_sp64, movhi_cc_sp64,
1565 movsi_cc_sp64, movdi_cc_sp64, movsf_cc_sp64, movdf_cc_sp64,
1566 movtf_cc_sp64, movqi_cc_reg_sp64, movhi_cc_reg_sp64,
1567 movsi_cc_reg_sp64, movdi_cc_reg_sp64, movsf_cc_reg_sp64,
1568 movdf_cc_reg_sp64, movtf_cc_reg_sp64, zero_extendhisi2_insn,
1569 cmp_siqi_trunc, cmp_siqi_trunc_set, sign_extendhisi2_insn,
1570 sign_extendqihi2_insn, sign_extendqisi2_insn,
1571 sign_extendqidi2_insn, sign_extendhidi2_insn,
1572 extendsfdf2, extendsftf2, extenddftf2, truncdfsf2, trunctfsf2,
1573 trunctfdf2, floatsisf2, floatsidf2, floatsitf2, floatdisf2,
1574 floatdidf2, floatditf2, fix_truncsfsi2, fix_truncdfsi2,
1575 fix_trunctfsi2, fix_truncsfdi2, fix_truncdfdi2, fix_trunctfdi2,
1576 adddi3_sp64, addsi3, cmp_ccx_plus, cmp_cc_plus_set, subdi_sp64,
1577 subsi3, cmp_minus_ccx, cmp_minus_ccx_set, mulsi3, muldi3,
1578 muldi3_v8plus, cmp_mul_set, mulsidi3, mulsidi3_v8plus,
1579 const_mulsidi3_v8plus, mulsidi3_sp32, const_mulsidi3,
1580 smulsi3_highpart_v8plus, unnamed subreg mult,
1581 const_smulsi3_highpart_v8plus, smulsi3_highpart_sp32,
1582 const_smulsi3_highpart, umulsidi3_v8plus, umulsidi3_sp32,
1583 const_umulsidi3, const_umulsidi3_v8plus, umulsi3_highpart_v8plus,
1584 const_umulsi3_highpart_v8plus, umulsi3_highpart_sp32,
1585 const_umulsi3_highpart, divsi3, divdi3, cmp_sdiv_cc_set, udivsi3,
1586 udivdi3, cmp_udiv_cc_set, smacsi, smacdi, umacdi, anddi3_sp64,
1587 andsi3, and_not_di_sp64, and_not_si, iordi3_sp64, iorsi3,
1588 or_not_di_sp64, or_not_si, xordi3_sp64, xorsi3, xor_not_di_sp64,
1589 xor_not_si, cmp_cc_arith_op, cmp_ccx_arith_op,
1590 cmp_cc_arith_op_set, cmp_ccx_arith_op_set, cmp_ccx_xor_not,
1591 cmp_cc_xor_not_set, cmp_ccx_xor_not_set, cmp_cc_arith_op_not,
1592 cmp_ccx_arith_op_not, cmp_cc_arith_op_not_set,
1593 cmp_ccx_arith_op_not_set, negdi2_sp64, cmp_cc_neg, cmp_ccx_neg,
1594 cmp_cc_set_neg, cmp_ccx_set_neg, one_cmpldi2_sp64, cmp_cc_not,
1595 cmp_ccx_not, cmp_cc_set_not, cmp_ccx_set_not, addtf3, adddf3,
1596 addsf3, subtf3, subdf3, subsf3, multf3, muldf3, mulsf3,
1597 muldf3_extend, multf3_extend, divtf3, divdf3, divsf3, negtf2,
1598 negdf2, negsf2, abstf2, absdf2, abssf2, sqrttf2, sqrtdf2, sqrtsf2,
1599 ashlsi3, ashldi3, unnamed DI ashift, cmp_cc_ashift_1,
1600 cmp_cc_set_ashift_1, ashrsi3, ashrdi3, unnamed DI ashiftrt,
1601 ashrdi3_v8plus, lshrsi3, lshrdi3, unnamed DI lshiftrt,
1602 lshrdi3_v8plus, tablejump_sp32, tablejump_sp64, call_address_sp32,
1603 call_symbolic_sp32, call_address_sp64, call_symbolic_sp64,
1604 call_address_struct_value_sp32, call_symbolic_struct_value_sp32,
1605 call_address_untyped_struct_value_sp32,
1606 call_symbolic_untyped_struct_value_sp32, call_value_address_sp32,
1607 call_value_symbolic_sp32, call_value_address_sp64,
1608 call_value_symbolic_sp64, branch_sp32, branch_sp64,
1609 flush_register_windows, goto_handler_and_restore,
1610 goto_handler_and_restore_v9, goto_handler_and_restore_v9_sp64,
1611 flush, all ldd/std peepholes, return_qi, return_hi, return_si,
1612 return_addsi, return_di, return_adddi, return_sf, all call+jump
1613 peepholes, trap, unnamed trap insns): Prettify output strings.
1614 (define_insn anddi3_sp32, and_not_di_sp32, iordi3_sp32,
1615 or_not_di_sp32, xordi3_sp32, xor_not_di_sp32, one_cmpldi2):
1616 Likewise and force + implement splits for integer cases.
1617 (define_insn return_sf_no_fpu): Likewise and allow to match when
1618 no-fpu because of our subreg SFmode splits.
1619 (define_insn zero_extendqihi2, zero_extendqisi2_insn,
1620 zero_extendqidi2_insn, zero_extendhidi2_insn,
1621 zero_extendsidi2_insn, sign_extendsidi2_insn): Likewise and use
1622 input_operand for second operand.
1623 (cmp_minus_cc, cmp_minus_cc_set): Likewise and use
1624 reg_or_0_operand for operand 2 so new splits can use it.
1625 (cmp_zero_extendqisi2, cmp_zero_extendqisi2_set, cmp_cc_plus,
1626 cmp_cc_xor_not): Likewise and don't forget to check TARGET_LIVE_G0
1628 (cmp_zero_extract, cmp_zero_extract_sp64): Likewise and allow
1629 CONST_DOUBLEs for operand 2.
1630 (define_insn move_label_di): Likewise and label distance
1631 optimization because it no longer works with new deferred case
1632 vector scheme. To be revisited.
1633 (define_insn x_minus_y_minus_sltu, x_minus_sltu_plus_y): Likewise
1634 and allow reg_or_0_operand and J constraint for second operand.
1635 (define_insn jump): Set branch predict taken on V9.
1636 (define_insn tablejump): Emit LABEL_REF + PLUS memory address for
1637 new deferred case vector scheme.
1638 (define_insn pic_tablejump_32, pic_tablejump_64): Remove.
1639 (define_insn negdi2_sp32): Force + implement splits.
1640 (define_insn negsi2, one_cmplsi2): Rename to negsi2_not_liveg0 and
1641 one_cmplsi2_not_liveg0 respectively, and create expander of original
1642 names which emit special rtl for TARGET_LIVE_G0.
1643 (define_insn cmpdi_v8plus, scc_si, scc_di): Remove.
1644 (define_insn seq, sne, slt, sge, sle, sltu, sgeu): Don't do
1645 gen_compare_reg, FAIL instead.
1646 (define_insn sgtu, sleu): Likewise and check gen_s*() return
1647 values when trying to reverse condition codes, if they FAIL then
1649 (define_insn snesi_zero, neg_snesi_zero, snesi_zero_extend,
1650 snedi_zero, neg_snedi_zero, snedi_zero_trunc, seqsi_zero,
1651 neg_seqsi_zero, seqsi_zero_extend, seqdi_zero, neg_seqdi_zero,
1652 seqdi_zero_trunc, x_plus_i_ne_0, x_minus_i_ne_0, x_plus_i_eq_0,
1653 x_minus_i_eq_0): Add new splits to perform these multi-insn cases,
1654 set output string to # to indicate they are mandatory splits.
1655 (define_insn pic_lo_sum_si, pic_sethi_si, pic_lo_sum_di,
1656 pic_sethi_di, move_pic_label_si): Remove.
1657 (define_insn movsi_low_sum, movsi_high, movsi_lo_sum_pic,
1658 movsi_high_pic, movsi_pic_label_reg): New patterns to take their
1660 (define_expand movsi_pic_label_ref, define_insn
1661 movsi_high_pic_label_ref, movsi_lo_sum_pic_label_ref): New
1662 expander and insns to handle PIC label references and deferred
1664 (define_insn get_pc_via_rdpc): Comment out as it is no longer
1666 (define_expand movqi, movhi, movsi, movdi, movsf, movdf, movtf):
1667 Rewrite to not use emit_move_sequence, make use of new constant
1668 formation code, and new splits for all multi-insn cases.
1669 (define_insn movqi_insn): Remove sethi case, it can never happen.
1670 Use reg_or_zero_operand instead of const0_rtx explicit test,
1671 use input_operand instead of move_operand for source, and use
1672 general_operand now for dest.
1673 (define_insn movhi_insn): Similar but leave sethi case.
1674 (define_insn lo_sum_qi, store_qi, store_hi): Remove.
1675 (define_insn sethi_hi lo_sum_hi): Rename to movhi_high and
1676 movhi_lo_sum respectively, prettify output string.
1677 (define_insn movsi_zero_liveg0): New pattern to put zero into a
1678 register when needed on TARGET_LIVE_G0.
1679 (define_insn movsi_insn): Use general_operand and input_operand
1680 for dest and src respectively. Simplify applicability test.
1681 Prettify output strings, and add clr alternative for J
1683 (define_insn movdi_sp32_v9, movdi_sp32, define_splits for
1684 deprecated std and reg-reg DI moves): Remove and...
1685 (define_insn movdi_insn_sp32, movdi_insn_sp64): Replace with new
1686 implementation which uses forced splits for all non-single insn
1688 (define_split DI move cases on !arch64): New splits to handle all
1689 situations of 64-bit double register DImode on 32bit, and
1690 unaligned registers and memory addresses for all subtargets.
1691 (define_insn movsf_const_insn, movdf_const_insn, store_sf):
1693 (define_insn movsf_insn, movsf_no_f_insn): Use general_operand and
1694 input_operand for dest and src respectively, prettify output
1696 (define_insn movdf_insn, movdf_no_e_insn, store_df,
1697 movtf_const_insn, movtf_insn, movtf_no_e_insn, store_tf): Remove
1699 (define_insn movdf_insn_sp32, movdf_no_e_insn_sp32,
1700 movdf_insn_sp64, movdf_no_e_insn_sp64, movtf_insn,
1701 movtf_no_e_insn_sp32, movtf_insn_hq_sp64, movtf_insn_sp64,
1702 movtf_no_e_insn_sp64) Replace with new
1703 implementation which uses forced splits for all non-single insn
1705 (define_split DF move cases): New splits in similar vein to DI
1707 (define_insn sethi_di_medlow, sethi_di_medium_pic,
1708 sethi_di_embmedany_data, sethi_di_embmedany_text, sethi_di_sp64,
1709 movdi_sp64_insn): Remove old v9 code model and constant loading
1711 (define_insn pic_lo_sum_di, pic_sethi_di,
1712 sethi_di_medlow_embmedany_pic, sethi_di_medlow, losum_di_medlow,
1713 seth44, setm44, setl44, sethh, setlm, sethm, setlo,
1714 embmedany_sethi, embmedany_losum, embmedany_brsum,
1715 embmedany_textuhi, embmedany_texthi, embmedany_textulo,
1716 embmedany_textlo, movdi_lo_sum_sp64_cint, movdi_lo_sum_sp64_dbl,
1717 movdi_high_sp64_cint, movdi_high_sp64_dbl): Replace with new
1718 scheme, using unspecs, secondary reloads, and one to one sparc
1719 insn to rtl insn mapping for better scheduling and code gen.
1720 (define_expand reload_indi, reload_outdi): Reload helpers for
1721 MEDANY and EMBMEDANY symbol address loading cases which require a
1723 (define_expand movsicc): Remove v8plus_regcmp cases.
1724 (define_insn movdi_cc_sp64_trunc, movdi_cc_reg_sp64_trunc,
1725 cmp_zero_extendqidi2, cmp_zero_extendqidi2_set, cmp_qidi_trunc,
1726 cmp_diqi_trunc_set): New patterns used by some of the new scc
1728 (define_insn xordi3_sp64_dbl): New pattern used for constant
1729 formation when crossing from 32-bit targets.
1730 (define_insn movsi_cc_reg_v8plus, v8plus_clear_high, and helper
1732 (define_insn addx, subx): Make visible and prettify.
1733 (define_insn adddi3_insn_sp32): Likewise and force split.
1734 (define_insn addx_extend, subx_extend, unnamed): New patterns for
1735 64bit scc split usage.
1736 (define_insn unnamed plusDI zero_extend, unnamed minusDI
1737 zero_extend, subdi3): Force and implement splits.
1739 * final.c (final_scan_insn): Don't output labels if target
1740 specifies ASM_OUTPUT_ADDR_{DIFF}_VEC. Do these macro operations
1743 * reorg.c (dbr_schedule): When taking on BR_PRED notes at the end,
1744 don't forget to walk inside SEQUENCESs too as these are what the
1745 delay slot scheduler will create.
1747 Mon Aug 10 01:21:01 1998 Richard Henderson <rth@cygnus.com>
1749 * alpha.md (extxl+1,+2): New patterns to work around
1752 Sat Aug 8 19:20:22 1998 Gary Thomas (gdt@linuxppc.org)
1754 * rs6000.c (rs6000_allocate_stack_space) Fix typo which
1755 caused bad assembly code to be generated.
1757 Sat Aug 8 18:53:28 1998 Jeffrey A Law (law@cygnus.com)
1759 * netbsd.h: Fix typo.
1761 Mon Aug 3 00:06:42 1998 Robert Lipe <robertl@dgii.com>
1763 * config.sub: Fix typo.
1765 Sun Aug 2 22:39:08 1998 Hans-Peter Nilsson <hp@axis.se>
1767 * invoke.texi (Environment Variables): Typo: Change "ascpects"
1769 (Running Protoize): Typo: Change "ther" into "other".
1771 Sun Aug 2 00:42:50 1998 Jeffrey A Law (law@cygnus.com)
1773 * i386/netbsd.h: Undo previous change to DWARF2_UNWIND_INFO.
1774 * m68k/netbsd.h: Likewise.
1775 * ns32k/netbsd.h: Likewise.
1776 * sparc/netbsd.h: Likewise.
1778 Sat Aug 1 17:59:30 1998 Richard Henderson <rth@cygnus.com>
1780 * ginclude/va-alpha.h (va_list): Use a typedef, not a define.
1781 * ginclude/va-clipper.h (va_list): Likewise.
1783 Fri Jul 31 20:22:02 1998 Michael Meissner <meissner@cygnus.com>
1785 * rs6000.c (rs6000_override_options): If big endian and -Os, use
1786 load/store multiple instructions unless user overrides.
1788 Fri Jul 31 17:08:59 1998 Jeffrey A Law (law@cygnus.com)
1790 * ns32k/netbsd.h: Fix typo.
1792 Fri Jul 31 10:23:55 1998 Doug Evans <devans@canuck.cygnus.com>
1794 * m32r/m32r.h (ASM_OUTPUT_SOURCE_LINE): Always output line number
1795 labels with .debugsym if no parallel insns.
1797 Thu Jul 30 19:15:53 1998 Richard Henderson <rth@cygnus.com>
1799 * alpha.md (fp cmp): Replicate patterns for ALPHA_TP_INSN.
1800 (fcmov): Remove ALPHA_TP_INSN patterns -- fcmov doesn't trap.
1802 Thu Jul 30 19:50:15 1998 David Edelsohn <edelsohn@mhpcc.edu>
1804 * rs6000/x-aix43 (AR_FOR_TARGET_FLAGS): Delete.
1805 (AR_FOR_TARGET): Define.
1807 Thu Jul 30 12:29:12 1998 Mark Mitchell <mark@markmitchell.com>
1809 * dyn-string.h: New file.
1810 * dyn-string.c: Likewise.
1811 * Makefile.in (OBJS): Add dyn-string.o.
1812 (dwarf2out.o): Add dyn-string.h dependency.
1813 (dyn-string.o): List dependencies.
1814 * dwarf2out.c: Include dyn-string.h.
1815 (ASM_NAME_TO_STRING): Use dyn_string_append, rather than strcpy.
1816 (addr_const_to_string): Take a dyn_string_t, not a char* as a
1817 prototype. Use dyn_string_append rather than strcat, throughout.
1818 (addr_to_string): Use dyn_string_t.
1820 Thu Jul 30 13:08:07 1998 Ken Raeburn <raeburn@cygnus.com>
1822 Function entry/exit profiling instrumentation:
1823 * expr.h (profile_function_entry_libfunc,
1824 profile_function_exit_libfunc): Declare new variables.
1825 * optabs.c: Define them here.
1826 (init_optabs): Initialize them.
1827 * tree.h (struct tree_decl): New flag
1828 no_instrument_function_entry_exit.
1829 (DECL_NO_INSTRUMENT_FUNCTION_ENTRY_EXIT): New accessor macro.
1830 * c-decl.c (duplicate_decls): Merge it.
1831 * c-common.c (enum attrs): New value A_NO_INSTRUMENT_FUNCTION.
1832 (init_attributes): Use it for "no_instrument_function".
1833 (decl_attributes): Handle it, for functions that have not yet been
1834 compiled. Set decl flag.
1835 * flags.h (flag_instrument_function_entry_exit): Declare new
1837 * toplev.c (flag_instrument_function_entry_exit): Define it here.
1838 (f_options): New option "instrument-functions".
1839 * function.h (struct function): New field instrument_entry_exit.
1840 * function.c (current_function_instrument_entry_exit): New
1842 (push_function_context_to, pop_function_context_from): Save and
1844 (expand_function_start): Set current_ variable, maybe emit return
1845 label and entry profile call.
1846 (expand_function_end): Maybe emit exit profile call.
1848 Thu Jul 30 00:58:34 1998 Jeffrey A Law (law@cygnus.com)
1850 * i386.md (movqi): When optimizing a load of (const_int 1) into a
1851 NON_QI_REG_P, pretend the register is SImode.
1853 Wed Jul 29 23:49:23 1998 Todd Vierling <tv@netbsd.org>
1855 * configure.in: Use xm-netbsd.h as the NetBSD xm file (not xm-siglist).
1856 Accept arm32 as arm, m68k4k as m68k, mipsle as mips-dec, and any
1857 manufacturer id for ns32k.
1858 * configure: Regenerated.
1859 * config/netbsd.h: When using ASM_WEAKEN_LABEL, make it global too.
1860 * config/t-netbsd: Don't compile libgcc1-test as the fns are in libc.
1861 * config/i386/netbsd.h: Undefine DWARF2_UNWIND_INFO, not define as 0.
1862 * config/m68k/netbsd.h: Same.
1863 * config/ns32k/netbsd.h: Same.
1864 * config/sparc/netbsd.h: Same.
1866 Wed Jul 29 22:39:21 1998 Jeffrey A Law (law@cygnus.com)
1868 * unroll.c (unroll_loop): Do not abort for an UNROLL_MODULO
1869 or UNROLL_COMPLETELY loop that starts with a jump to its
1872 Wed Jul 29 22:18:14 1998 David Edelsohn <edelsohn@mhpcc.edu>
1874 * rs6000/rs6000.md (absdi2 define_split): Swap operands of MINUS.
1875 * rs6000/rs6000.c (mask64_operand): Use HOST_BITS_PER_WIDE_INT.
1876 (print_operand, case 'B'): Don't fall through.
1877 (print_operand, case 'S'): Correct mask begin/end computation.
1878 Use HOST_BITS_PER_WIDE_INT.
1879 * rs6000/rs6000.h (CPP_PREDEFINES): Define _LONG_LONG.
1880 (CONDITIONAL_REGISTER_USAGE): GPR13 fixed if TARGET_64BIT.
1881 * rs6000/aix41.h (CPP_PREDEFINES): Same.
1882 * rs6000/aix43.h (CPP_PREDEFINES): Same.
1884 Wed Jul 29 11:47:10 1998 Nick Clifton <nickc@cygnus.com>
1886 * config/arm/thumb.md (extendqisi2_insn): Remove earlyclobber
1887 constraint from second alternative.
1889 Tue Jul 28 23:29:04 1998 Jason Merrill <jason@yorick.cygnus.com>
1891 * configure.in: Fix --without/--disable cases for local-prefix,
1892 gxx-include-dir and checking.
1894 Tue Jul 28 22:01:23 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
1896 * configure.in (enable_haifa): Set by default for sparc64 too.
1899 Tue Jul 28 23:29:04 1998 Jason Merrill <jason@yorick.cygnus.com>
1901 * i386/cygwin32.h (VALID_MACHINE_TYPE_ATTRIBUTE): New macro.
1902 * i386/winnt.c (associated_type): New fn.
1903 (i386_pe_valid_type_attribute_p): New fn.
1904 (i386_pe_check_vtable_importexport): Remove.
1905 (i386_pe_dllexport_p): Use associated_type.
1906 (i386_pe_dllimport_p): Likewise.
1908 From Antonio M. O. Neto <anmendes@cruzeironet.com.br>:
1909 * i386.c (i386_valid_type_attribute_p): Also accept
1910 attributes for METHOD_TYPEs.
1912 Tue Jul 28 23:17:39 1998 Peter Gerwinski <peter@gerwinski.de>
1914 * tree.c (build_range_type): Copy TYPE_SIZE_UNIT.
1916 Tue Jul 28 22:31:12 1998 Craig Burley <burley@gnu.org>
1918 * gcc.c: Fix commentary describing %g, %u, %U, and %O.
1920 * gcc.c (do_spec_1): Fix handling of %g%O and %U%O to prevent
1921 them from generating a new base name for each occurence of
1924 1998-07-28 Vladimir N. Makarov <vmakarov@cygnus.com>
1926 * cse.c (cse_insn): Enable subsitution inside libcall only for REG,
1928 * rtlanal.c (replace_rtx): Prohibit replaces in CONST_DOUBLE.
1932 * cplus-dem.c (type_kind_t): New type.
1933 (demangle_template_value_parm): Add type_kind_t parameter. Rely
1934 on this paramter, rather than demangling the type again.
1935 (demangle_integral_value): Pass tk_integral.
1936 (demangle_template_: Pass the value returned from do_type.
1937 (do_type): Return a type_kind_t. Pass tk_integral to
1938 demangle_template_value_parm for array bounds.
1939 (demangle_fund_type): Likewise.
1941 Mon Jul 27 00:54:41 1998 Jason Merrill <jason@yorick.cygnus.com>
1943 * tree.c (simple_cst_equal, case CONSTRUCTOR): OK if the elts are
1946 Mon Jul 27 22:18:36 1998 Jeffrey A Law (law@cygnus.com)
1948 * pa.c (move_operand): Accept CONSTANT_P_RTX.
1950 Mon Jul 27 17:18:52 1998 Dave Brolley <brolley@cygnus.com>
1952 * stor-layout.c (layout_type): Handle arrays of bits, for Chill.
1954 * expr.c (get_inner_reference): Handle zero-based, unsigned, array
1957 Mon Jul 27 14:51:33 1998 Jeffrey A Law (law@cygnus.com)
1959 * mn10300.h (DEBUGGER_AUTO_OFFSET): Define.
1960 (DEBUGGER_ARG_OFFSET): Likewise.
1962 * mn10300.md (movsf): Remove last change. Not needed.
1964 Mon Jul 27 14:22:36 1998 Dave Brolley <brolley@cygnus.com>
1966 * c-lex.c (yylex): Fix boundary conditions in character literal and
1967 string literal loops.
1969 Mon Jul 27 11:43:54 1998 Stan Cox <scox@cygnus.com>
1971 * longlong.h (count_leading_zeros): Sparclite scan instruction was
1972 being invoked incorrectly.
1974 * i386.c (ix86_prologue): Added SUBTARGET_PROLOGUE invocation.
1975 * i386/cygwin32.h (STARTFILE_SPEC, LIB_SPEC, SUBTARGET_PROLOGUE):
1977 * i386/win32.h: New file. Hybrid mingw32.h/cygwin32.h configuration.
1978 * configure.in: Added i[34567]86-*-win32
1979 * config.sub: Ditto.
1980 * configure: Rebuilt.
1982 Sun Jul 26 01:11:12 1998 H.J. Lu (hjl@gnu.org)
1984 * i386.h (CONST_DOUBLE_OK_FOR_LETTER_P): Return 0 when eliminating
1985 the frame pointer and compiling PIC code and reload has not completed.
1987 * i386.c (output_to_reg): Add code to emulate non-popping DImode
1990 Sun Jul 26 01:01:32 1998 Jeffrey A Law (law@cygnus.com)
1992 * regmove.c (regmove_optimize): Fix typo initializing regmove_bb_head.
1994 Sat Jul 25 23:29:23 1998 Gerald Pfeifer <pfeifer@dbai.tuwien.ac.at>
1996 * Makefile.in (install-info): Only try to update the info
1997 directory file if it exists in the first place.
1999 Fri Jul 24 18:58:37 1998 Klaus Espenlaub <kespenla@student.informatik.uni-ulm.de>
2001 * rs6000.h (ASM_OUTPUT_CONSTRUCTOR, ASM_OUTPUT_DESTRUCTOR): Delete.
2003 Fri Jul 24 14:20:26 1998 Jeffrey A Law (law@cygnus.com)
2005 * mn10300.md (movqi, movhi, movsi, movsf): Correctly handle
2006 CONST_DOUBLE source.
2008 Fri Jul 24 11:17:04 1998 Nick Clifton <nickc@cygnus.com>
2010 * config/arm/thumb.c (thumb_print_operand): Decode %_ in asm
2011 strings as the insertion of USER_LABEL_PREFIX.
2012 * config/arm/thumb.h (PRINT_OPERAND_PUNCT_VALID_P): Accept _ as a
2014 * config/arm/thumb.md: Use %_ as a prefix to gcc library function
2017 Thu Jul 23 18:53:20 1998 Jim Wilson <wilson@cygnus.com>
2019 * dbxout.c (dbxout_range_type): Only call dbxout_type_index for
2020 already defined type.
2022 Thu Jul 23 13:49:41 1998 Jeffrey A Law (law@cygnus.com)
2024 * expr.c (check_max_integer_computation_mode): Allow conversions
2025 of constant integers to MAX_INTEGER_COMPUTATION_MODE.
2026 (expand_expr): Likewise.
2028 Thu Jul 23 11:12:06 1998 Alexandre Petit-Bianco <apbianco@cygnus.com>
2030 * expr.c (expand_expr): Expand RETURN_EXPR.
2032 Thu Jul 23 11:00:29 1998 Jim Wilson <wilson@cygnus.com>
2034 * dwarf2out.c (dwarf2out_finish): Call stripattributes on TEXT_SECTION.
2036 Wed Jul 22 19:10:00 1998 Catherine Moore <clm@cygnus.com>
2038 * dwarf2out.c (output_aranges): Call stripattributes
2039 for TEXT_SECTION references.
2040 (output_line_info): Likewise.
2042 Wed Jul 22 14:08:54 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
2044 * profile.c (branch_prob): Call allocate_reg_info after outputting
2045 profile rtl in instrument_arcs.
2047 Wed Jul 22 12:47:49 1998 Jim Wilson <wilson@cygnus.com>
2049 * fixinc.irix (math.h): Install wrapper instead of copying.
2051 Wed Jul 22 12:37:14 1998 Alexandre Petit-Bianco <apbianco@cygnus.com>
2053 * tree.def (EXPR_WITH_FILE_LOCATION): Defined as an 'e' expression
2054 so WFL are expanded correctly when contained in a COMPOUND_EXPR.
2055 * tree.h (EXPR_WFL_EMIT_LINE_NOTE): Change macro not to use
2056 lang_flag_0. Added documentation in the flag table.
2058 Tue Jul 21 23:28:35 1998 Klaus Kaempf <kkaempf@rmi.de>
2060 * cccp.c (do_include): Fix vax c style include handling.
2062 Tue Jul 21 13:28:19 1998 Jason Merrill <jason@yorick.cygnus.com>
2064 * cplus-dem.c (do_type): Use demangle_template_value_parm for arrays.
2066 Sun Jul 12 01:27:05 1998 Jason Merrill <jason@yorick.cygnus.com>
2068 * fold-const.c (non_lvalue): Don't deal with null pointer
2070 (fold, case COMPOUND_EXPR): Wrap a constant 0 in a NOP_EXPR.
2072 Tue Jul 21 15:49:31 1998 David Edelsohn <edelsohn@mhpcc.edu>
2074 * rs6000.h (PREDICATE_CODES): Add CONSTANT_P_RTX.
2075 * rs6000.md (movsi, movdi): Add CONSTANT_P_RTX.
2076 * rs6000.c (short_cint_operand): Add CONSTANT_P_RTX.
2077 (u_short_cint_operand): Same.
2078 (reg_or_cint_operand): Same.
2079 (logical_operand): Same.
2080 (input_operand): Same.
2081 (reg_or_short_operand): Use u_short_cint_operand.
2083 Tue Jul 21 08:56:42 1998 Richard Henderson <rth@cygnus.com>
2085 * alpha.md (fix_truncdfsi2, fix_truncsfsi2): Remove the define_expands,
2086 but keep the insns and splits. Adjust so when the ultimate destination
2087 is memory, use cvtql.
2089 Tue Jul 21 08:55:09 1998 Richard Henderson <rth@cygnus.com>
2091 * flow.c (regno_uninitialized): Fixed regs are never uninitialized.
2093 Tue Jul 21 00:31:01 1998 Jeffrey A Law (law@cygnus.com)
2095 * gcc.c (do_spec): Call "error" not "warning".
2097 * configure.in: Fix minor problems with gas feature detection code.
2098 * configure: Rebuilt.
2100 * gcc.c (do_spec): Issue a warning for '%[]' usage.
2103 * gcc.c: Delete %[spec] support.
2104 (do_spec_1, case '('): Likewise.
2105 (do_spec_1, case '['): Call error.
2107 Mon Jul 20 22:34:17 1998 Richard Henderson <rth@cygnus.com>
2109 * alpha.h (CPP_SPEC): Tidy. Hook to cpp_cpu and cpp_subtarget.
2110 (CPP_SUBTARGET_SPEC): Default to empty string.
2111 (CPP_AM_*, CPP_IM_*, CPP_CPU_*, CPP_CPU_SPEC): New.
2112 (EXTRA_SPECS, SUBTARGET_EXTRA_SPECS): New.
2113 * alpha/elf.h (LD_SPEC): Use %(elf_dynamic_linker).
2114 * alpha/linux-elf.h (SUBTARGET_EXTRA_SPECS): New.
2116 * alpha/linux.h (CPP_PREDEFINES): Tidy.
2117 * alpha/netbsd-elf.h (SUBTARGET_EXTRA_SPECS): New.
2118 * alpha/netbsd.h (CPP_PREDEFINES): Tidy.
2119 * alpha/osf.h (CPP_PREDEFINES): Remove bits subsumed by CPP_CPU_SPEC.
2120 * alpha/win-nt.h (CPP_PREDEFINES): Likewise.
2121 * alpha/vsf.h (CPP_PREDEFINES): Likewise.
2122 (CPP_SUBTARGET_SPEC): New. Do this instead of overriding CPP_SPEC.
2123 * alpha/vxworks.h: Likewise.
2125 Mon Jul 20 22:51:57 1998 Ken Raeburn <raeburn@cygnus.com>
2127 * mips.md (reload_outsi): Added missing REGNO call.
2128 (smulsi3_highpart, umulsi3_highpart): Provide prototype for
2130 (mul_acc_di, mul_acc_64bit_di): Don't use match_op_dup, use
2131 another match_operator and compare the codes.
2133 * mips.h (MASK_DEBUG_E, MASK_DEBUG_I): Set to zero.
2135 * MIPS multiply pattern fixes:
2136 * mips.h (enum reg_class, REG_CLASS_NAMES, REG_CLASS_CONTENTS):
2137 Add union classes for HI, LO, or HILO plus general registers.
2138 (GENERATE_MADD): Deleted.
2139 * mips.md (mulsi3_mult3): Don't disparage output-LO alternative.
2140 Add TARGET_MAD to condition.
2141 (mulsi3): Test HAVE_mulsi3_mult3, not specific flags.
2142 (mul_acc_si): Expand GENERATE_MADD here; it's the only use. Use
2143 "*d" for accumulator, to give preference to LO initially but not
2146 Mon Jul 20 16:16:38 1998 Dave Brolley <brolley@cygnus.com>
2148 * configure.in (enable_c_mbchar): New configure option.
2149 (extra_cpp_objs): Always available now.
2151 * cexp.y (mbchar.h): #include it.
2152 (yylex): Handle Multibyte characters in character literals.
2154 * cccp.c (mbchar.h): #include it.
2155 (main): Set character set based on LANG environment variable.
2156 (rescan): Handle multibyte characters in comments.
2157 (skip_if_group): See above.
2158 (validate_else): See above.
2159 (skip_to_end_of_comment): See above.
2160 (macarg1): See above.
2161 (discard_comments): See above.
2162 (rescan): Handle multibyte characters in string and character literals.
2163 (collect_expansion): See above.
2164 (skip_quoted_string): See above.
2165 (macroexpand): See above.
2166 (macarg1): See above.
2167 (discard_comments): See above.
2168 (change_newlines): See above.
2170 * c-lex.c (mbchar.h): #include it.
2171 (GET_ENVIRONMENT): New macro.
2172 (init_lex): Set character set based on LANG environment variable.
2173 (yylex): Handle multibyte characters in character literals.
2174 (yylex): Handle multibyte characters in string literals.
2176 * Makefile.in (mbchar.o): New target.
2177 (cccp$(exeext)): @extra_cpp_objs@ is always available.
2178 (cppmain$(exeext)): @extra_cpp_objs@ is always available.
2180 * mbchar.[ch]: New files for multibyte character handling.
2182 Mon Jul 20 01:11:11 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
2184 * jump.c (jump_optimize): When simplifying noop moves and
2185 PUSH_ROUNDING, fix thinko so we use same criterion for identifying
2186 the PUSHes to rewrite in second loop as we did in the first.
2188 Sun Jul 19 08:23:53 1998 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
2190 * cplus-dem.c (demangle_nested_args): Make function definition
2191 static to match the prototype.
2193 Fri Jul 17 14:58:44 1998 Richard Henderson <rth@cygnus.com>
2195 * alloca.c: Respect USE_C_ALLOCA.
2196 * gencheck.c (xmalloc): Ignore __GNUC__ for definition.
2197 * gengenrtl.c (xmalloc): Likewise.
2199 Fri Jul 17 14:18:14 1998 Richard Henderson <rth@cygnus.com>
2201 * loop.h (struct induction): Add no_const_addval.
2202 * loop.c (the_movables, reg_address_cost): New variables.
2203 (init_loop): Init reg_address_cost.
2204 (loop_optimize): Call end_alias_analysis.
2205 (scan_loop): Init the_movables.
2206 (record_giv): Init induction->no_const_addval.
2207 (basic_induction_var) [PLUS]: Use rtx_equal_p instead of ==.
2208 [REG]: Rearrange loop search test to catch more cases.
2209 (general_induction_var): Return success not benefit; take an extra
2210 argument for that. Change all callers.
2211 (simplify_giv_expr) [PLUS]: Always combine invariants. Use sge_plus.
2212 [MULT]: Use rtx_equal_p instead of ==. Combine simple invariants.
2213 [default]: Search the_movables for additional combinations.
2214 (sge_plus_constant, sge_plus): New functions.
2215 (express_from_1): New function.
2216 (express_from): Always define. Rewrite using express_from_1.
2217 (combine_givs_p): Handle more cases. Ignore address cost.
2218 (cmp_combine_givs_stats): New function.
2219 (combine_givs_used_once, combine_givs_benefit_from): New functions.
2220 (combine_givs): Rewrite to do best-fit combination.
2222 * fold-const.c (operand_equal_p): Handle RTL_EXPR.
2223 (fold): Do a complete (A*C)+(B*C) association check.
2225 Fri Jul 17 11:21:55 1998 Jim Wilson <wilson@cygnus.com>
2227 * function.c (fixup_var_refs_insns): Handle CLOBBER of a CONCAT.
2229 Fri Jul 17 11:48:55 1998 Jeffrey A Law (law@cygnus.com)
2231 * mn10300.c (MODES_TIEABLE_P): Fix typo.
2233 Fri Jul 17 03:26:12 1998 Rihcard Earnshaw (rearnsha@arm.com)
2235 * tree.c (valid_machine_attribute): Only create a new type variant if
2236 there is a decl to use it.
2238 Thu Jul 16 14:48:04 1998 Nick Clifton <nickc@cygnus.com>
2240 * gcc.c (do_spec_1): Cope with %g/%u/%U options which do not have
2243 Fri Jul 17 03:24:40 1998 Hans-Peter Nilsson <hp@axis.se>
2245 * extend.texi (Explicit Reg Vars): Typo: change "may deleted" into "may
2248 Thu Jul 16 14:48:47 1998 Jeffrey A Law (law@cygnus.com)
2250 * mn10300.c (count_tst_insns): New arg oreg_countp. Callers changed.
2251 Simplify tests for clearing an address register.
2252 (expand_prologue): Corresponding changes.
2254 * mn10300.md (movXX patterns): Make sure the destination is an
2255 ADDRESS_REG when substituting "zero_areg" for (const_int 0).
2256 (logical patterns): Split into expanders + patterns
2257 (zero and sign extension patterns): Similarly.
2258 (shift patterns): Similarly.
2260 Thu Jul 16 01:17:44 1998 Richard Henderson <rth@cygnus.com>
2262 * loop.c (emit_iv_add_mult): Scan the entire insn list generated
2263 for the sequence, recording base values.
2265 Wed Jul 15 10:49:55 1998 Richard Henderson <rth@cygnus.com>
2267 * i386.h (CPP_CPU_SPEC): Remove -Asystem(unix).
2269 Tue Jul 14 14:15:30 1998 Nick Clifton <nickc@cygnus.com>
2271 * gcc.c: Remove ANSI-C ism from --help code.
2273 * toplev.c: Support --help with USE_CPPLIB.
2275 Tue Jul 14 14:46:08 1998 Jeffrey A Law (law@cygnus.com)
2277 * configure.in: Rework gas feature code to work with symlink based
2280 * extend.texi: Clarify some issues related to local variables
2281 assigned to explicit registers.
2283 * mn10300.md (mulsi): Turn into expander + pattern.
2285 * mn10300.md (movsi, movsf, movdi, movdf): Remove "x" from I -> a
2288 Tue Jul 14 07:41:59 1998 Richard Earnshaw (rearnsha@arm.com)
2290 * arm/tcoff.h (USER_LABEL_PREFIX): Make it empty to match coff.h.
2292 Tue Jul 14 03:02:44 1998 Jeffrey A Law (law@cygnus.com)
2294 * version.c: Bump again to distinguish mainline tree from the
2297 See ChangeLog.0 for earlier changes.
2300 add-log-time-format: current-time-string