1 Tue Sep 15 09:59:01 1998 Mark Mitchell <mark@markmitchell.com>
3 * integrate.c (copy_decl_list): Fix typo.
5 Tue Sep 15 04:18:52 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
7 * config/sparc/sparc.md (movdf_const_intreg_sp32): Fix length
10 Mon Sep 14 14:02:53 PDT 1998 Jeff Law (law@cygnus.com)
12 * version.c: Bump for snapshot.
14 Mon Sep 14 10:33:56 PDT 1998 Jeff Law (law@cygnus.com)
16 * version.c: Bump for snapshot.
18 Mon Sep 14 09:51:05 PDT 1998 Jeff Law (law@cygnus.com)
20 * version.c: Bump for snapshot.
22 Sun Sep 13 22:10:18 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
24 * invoke.texi (C Dialect Options): Put back missing @end itemize.
26 Mon Sep 14 02:33:46 1998 Alexandre Oliva <oliva@dcc.unicamp.br>
28 * configure.in: remove usage of `!' to negate the result of a
29 command; some common shells do not support it
31 Sun Sep 13 19:17:35 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
33 * configure.in: in sparc9-sol2 config, use 'if test' not
37 * config/sparc/sol2-sld-64.h (SPARC_DEFAULT_CMODEL): Change to
39 (CPP_CPU_SPEC): Do not define _LP64, header files do this.
40 (CPP_CPU_DEFAULT_SPEC): Likewise.
41 * config/sparc/sol2.h (INIT_SUBTARGET_OPTABS): Get the names right
44 * config/sparc/sparc.md (goto_handler_and_restore): Allow any mode
47 Sun Sep 13 09:11:59 1998 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
49 * acconfig.h (NEED_DECLARATION_STRSIGNAL): Provide a stub.
51 * collect2.c: Don't declare `sys_siglist' here.
52 (my_strsignal): Prototype and define new function. Use it in
53 place of `sys_siglist' hacks.
55 * mips_tfile.c: Likewise.
57 * configure.in (AC_CHECK_FUNCS): Check for strsignal.
58 (GCC_NEED_DECLARATIONS): Likewise.
60 * system.h (strsignal): Prototype it, if necessary.
61 (sys_siglist): Declare it, if necessary.
63 Sun Sep 13 04:37:28 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
65 * loop.c (move_movables): While removing insn sequences, preserve
66 the next pointer of the most recently deleted insn when we skip
69 Sat Sep 12 11:37:19 1998 Michael Meissner <meissner@cygnus.com>
71 * rs6000.h ({ASM,CPP}_CPU_SPEC): Add support for all machines
72 supported with -mcpu=xxx.
74 Fri Sep 11 23:55:54 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
76 * flow.c (mark_set_1): Recognize multi-register structure return
78 (mark_used_regs): Likewise.
79 (count_reg_sets_1): Likewise.
80 (count_reg_references): Likewise.
81 * rtlanal.c (note_stores): Likewise.
82 (reg_overlap_mentioned_p): Likewise.
83 * haifa-sched.c (check_live_1): Likewise.
84 (update_live_1): Likewise.
85 (sched_analyze_1): Likewise.
86 (sched_note_set): Likewise.
87 (birthing_insn_p): Likewise.
88 (attach_deaths): Likewise.
90 * config/sparc/sparc.md (movdf_const_intreg_sp64): Disable.
93 Fri Sep 11 22:57:55 1998 Eric Dumazet <dumazet@cosmosbay.com>
95 * config/i386/sco5.h (ASM_WEAKEN_LABEL): Defined as in svr4.h.
97 Thu Sep 10 22:02:04 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
99 * glimits.h (__LONG_MAX__): Recognize __sparcv9 too.
101 Thu Sep 10 21:19:10 1998 Jakub Jelinek <jj@sunsite.ms.mff.cuni.cz>
103 * configure.in: Add check for GAS subsection -1 support.
104 * acconfig.h (HAVE_GAS_SUBSECTION_ORDERING): Add.
105 * configure config.in: Rebuilt.
106 * config/sparc/sparc.h (CASE_VECTOR_MODE): For V9 flag_pic, use
107 SImode is subsection -1 works, else use DImode.
108 (ASM_OUTPUT_ADDR_VEC_START, ASM_OUTPUT_ADDR_VEC_END): Define if
110 * config/sparc/sparc.c (sparc_output_addr_vec,
111 sparc_output_addr_diff_vec): Use them if defined.
113 Thu Sep 10 10:46:01 1998 Mark Mitchell <mark@markmitchell.com>
115 * tree.h (DECL_ORIGIN): New macro.
116 * integrate.c (copy_and_set_decl_abstract_origin): New function.
117 (copy_decl_list): Use it.
118 (integrate_parm_decls): Likewise.
119 (integrate_decl_tree): Likewise.
120 * dwarf2out.c (decl_ultimate_origin): Simplify.
121 * dwarfout.c (decl_ultimate_origin): Likewise.
122 * c-decl.c (duplicate_decls): Use DECL_ORIGIN.
123 (pushdecl): Likewise.
125 Thu Sep 10 08:01:31 1998 Anthony Green <green@cygnus.com>
127 * config/rs6000/rs6000.c (output_epilog): Add Java support.
129 Thu Sep 10 14:48:59 1998 Martin von Löwis <loewis@informatik.hu-berlin.de>
131 * invoke.texi (C++ Dialect Options): Document -fhonor-std.
133 Thu Sep 10 01:38:05 1998 Jeffrey A Law (law@cygnus.com)
135 * reg-stack.c (straighten_stack): Do nothing if the virtual stack is
136 empty or has a single entry.
138 * toplev.c (rest_of_compilation): Open up the dump file for reg-stack
139 before calling reg_to_stack.
141 Thu Sep 10 00:03:34 1998 Richard Henderson <rth@cygnus.com>
143 * alpha.c (alphaev5_insn_pipe): Abort on default case.
144 (alphaev5_next_group): Swallow CLOBBERs and USEs.
146 * c-tree.h (warn_long_long): Declare it.
148 Wed Sep 9 23:31:36 1998 (Stephen L Moshier) <moshier@world.std.com>
150 * emit-rtl.c (gen_lowpart_common): Disable optimization of
151 initialized float-int union if the value is a NaN.
153 Wed Sep 9 23:00:48 1998 Nathan Sidwell <nathan@acm.org>
155 * c-lex.c (real_yylex): Don't warn about long long constants if
156 we're allowing long long
158 Wed Sep 9 21:58:41 1998 Bernd Schmidt <crux@pool.informatik.rwth-aachen.de>
160 * except.h (current_function_eh_stub_label): Declare.
161 (current_function_eh_old_stub_label): Declare.
162 * function.h (struct function): New members eh_stub_label and
164 * except.c (current_function_eh_stub_label): New variable.
165 (current_function_eh_old_stub_label): New variable.
166 (init_eh_for_function): Clear them.
167 (save_eh_status): Save them.
168 (restore_eh_status): Restore them.
169 (expand_builtin_eh_stub): Set current_function_eh_stub_label.
170 (expand_builtin_eh_stub_old): Set current_function_eh_old_stub_label.
171 * flow.c (find_basic_blocks_1): When handling a REG_LABEL note, don't
172 make an edge from the block that contains it to the block starting
173 with the label if this label is one of the eh stub labels.
174 If eh stub labels exist, show they are reachable from the last block
177 * reload1.c (reload): Break out several subroutines and make some
179 (calculate_needs_all_insns): New function, broken out of reload.
180 (calculate_needs): Likewise.
181 (find_reload_regs): Likewise.
182 (find_group): Likewise.
183 (find_tworeg_group): Likewise.
184 (something_needs_reloads): New global variable, formerly in reload.
185 (something_needs_elimination): Likewise.
186 (caller_save_spill_class): Likewise.
187 (caller_save_group_size): Likewise.
188 (max_needs): Likewise.
189 (group_size): Likewise.
190 (max_groups): Likewise.
191 (max_nongroups): Likewise.
192 (group_mode): Likewise.
193 (max_needs_insn): Likewise.
194 (max_groups_insn): Likewise.
195 (max_nongroups_insn): Likewise.
198 * print-rtl.c (print_rtx): For MEMs, print MEM_ALIAS_SET.
200 Wed Sep 9 13:14:41 1998 Richard Henderson <rth@cygnus.com>
202 * loop.c (load_mems): Copy rtx for output mem.
204 Wed Sep 9 15:16:58 1998 Gavin Romig-Koch <gavin@cygnus.com>
206 * mips/abi64.h (LONG_MAX_SPEC): Don't set LONG_MAX for
207 mips1 or mips2 either.
209 Wed Sep 9 12:31:35 1998 Jeffrey A Law (law@cygnus.com)
211 * pa.c (pa_reorg): New marking scheme for jumps inside switch
213 (pa_adjust_insn_length): Update to work with new marking scheme
214 for jumps inside switch tables.
215 * pa.md (switch_jump): Remove pattern.
216 (jump): Handle jumps inside jump tables.
218 * Makefile.in (profile.o): Depend on insn-config.h
220 Wed Sep 9 09:36:51 1998 Jim Wilson <wilson@cygnus.com>
222 * iris6.h (DWARF2_UNWIND_INFO): Undef.
224 Wed Sep 9 01:32:01 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
226 Add preliminary native sparcv9 Solaris support.
227 * configure.in: Recognize sparv9-*-solaris2*
229 * config.sub: Recognize sparcv9 just like sparc64.
230 * config/sparc/sol2-c1.asm config/sparc/sol2-ci.asm
231 config/sparc/sol2-cn.asm: Macroize so it can be shared between
232 32-bit and 64-bit Solaris systems.
233 * config/sparc/t-sol2: Assemble those with cpp.
234 * config/sparc/sparc.h (TARGET_CPU_sparcv9): New alias for v9.
235 (*TF*_LIBCALL): If ARCH64 use V9 names.
236 * config/sparc/{xm-sysv4-64,sol2-sld-64}.h: New files.
238 Wed Sep 9 01:07:30 1998 Jakub Jelinek <jj@sunsite.ms.mff.cuni.cz>
240 * config/sparc/sparc.h (TARGET_CM_MEDMID): Fix documentation.
241 (CASE_VECTOR_MODE): Set to SImode even if PTR64, when MEDLOW and
243 (ASM_OUTPUT_ADDR_{VEC,DIFF}_ELT): Check CASE_VECTOR_MODE not
245 * config/sparc/sparc.md (tablejump): Likewise, and sign extend op0
246 to Pmode if CASE_VECTOR_MODE is something else.
248 Wed Sep 9 00:10:31 1998 Jeffrey A Law (law@cygnus.com)
250 * prefix.c (update_path): Correctly handle cases where PATH is
251 a substring of the builtin prefix, but specifies a different
254 Tue Sep 8 23:46:04 1998 Hans-Peter Nilsson <hp@axis.se>
256 * expr.c: Corrected comment about what MOVE_RATIO does.
257 * config/alpha/alpha.h: Ditto.
258 * config/1750a/1750a.h: Ditto.
259 * config/clipper/clipper.h: Ditto.
260 * config/i386/i386.h: Ditto.
262 Tue Sep 8 22:56:12 1998 Jeffrey A Law (law@cygnus.com)
264 * configure.in (m68k-next-nextstep3*): Use collect2.
265 Similarly for x86 NeXT configurations.
266 * configure: Rebuilt.
268 Tue Sep 8 01:38:57 1998 Nathan Sidwell <nathan@acm.org>
270 * configure.in: Don't assume srcdir is .../gcc
271 * configure: Rebuilt.
273 Sat Sep 5 16:34:34 EDT 1998 John Wehle (john@feith.com)
275 * global.c: Update comments.
276 (global_alloc): Assign allocation-numbers
277 even for registers allocated by local_alloc in case
278 they are later spilled and retry_global_alloc is called.
279 (mark_reg_store, mark_reg_clobber,
280 mark_reg_conflicts, mark_reg_death): Always record a
281 conflict with a pseudo register even if it has been
282 assigned to a hard register.
283 (dump_conflicts): Don't list pseudo registers already assigned to
284 a hard register as needing to be allocated, but do list their
286 * local-alloc.c: Update comment.
288 Mon Sep 7 23:38:01 1998 Jeffrey A Law (law@cygnus.com)
290 * configure.in: Check for bogus GCC_EXEC_PREFIX and LIBRARY_PATH.
291 * configure: Rebuilt.
293 Mon Sep 7 22:41:46 1998 Michael Meissner <meissner@cygnus.com>
295 * rs6000.c (rs6000_override_options): Fix name for ec603e, to add
297 * t-ppccomm (MULTILIB_MATCHES_FLOAT): Add support for -mcpu=xxx
298 for all targets that set -msoft-float.
300 Mon Sep 7 23:30:07 1998 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
302 * toplev.c (print_switch_values): Make static to match prototype.
304 Mon Sep 7 19:13:59 1998 Jeffrey A Law (law@cygnus.com)
306 * configure.in: If we are unable to find the "gnatbind" program,
307 then do not configure the ada subdir.
308 * configure: Rebuilt.
310 Sun Sep 6 14:03:58 PDT 1998 Jeff Law (law@cygnus.com)
312 * version.c: Bump for snapshot.
314 Sun Sep 6 13:28:07 PDT 1998 Jeff Law (law@cygnus.com)
316 * version.c: Bump for snapshot.
318 Sun Sep 6 08:54:14 1998 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
320 * Makefile.in (toplev.o): Depend on $(EXPR_H).
321 (insn-extract.o, insn-attrtab.o): Depend on toplev.h.
323 * gansidecl.h: Define ATTRIBUTE_NORETURN.
325 * genattrtab.c: Have insn-attrtab.c include toplev.h.
327 * genextract.c: Have insn-extract.c include toplev.h.
329 * rtl.h: Don't prototype `fatal_insn_not_found' and `fatal_insn'.
331 * toplev.c: Include expr.h.
332 (really_sorry, fancy_abort): Remove prototypes.
333 (set_target_switch): Add argument in prototype.
334 (vfatal): Mark prototype with ATTRIBUTE_NORETURN.
335 (v_really_sorry): Likewise.
336 (print_version, print_single_switch, print_switch_values): Make
337 static and add prototype arguments.
338 (decl_printable_name): Add prototype arguments.
339 (lang_expand_expr_t): New typedef.
340 (lang_expand_expr): Declare as a lang_expand_expr_t.
341 (incomplete_decl_finalize_hook): Add prototype argument.
342 (decl_name): Mark variable `verbosity' with ATTRIBUTE_UNUSED.
343 (botch): Likewise for variable `s'.
344 (rest_of_type_compilation): Mark variables `type' and `toplev'
345 with ATTRIBUTE_UNUSED if none of DBX_DEBUGGING_INFO,
346 XCOFF_DEBUGGING_INFO or SDB_DEBUGGING_INFO are defined.
347 (display_help): Make variable `i' an `unsigned long'.
348 (main): Remove unused parameter `envp'.
349 Cast assignment to `lang_expand_expr' to a `lang_expand_expr_t'.
350 Cast -1 when comparing it with a `size_t'.
352 * toplev.h (fatal, fatal_io_error, pfatal_with_name): Mark
353 prototype with ATTRIBUTE_NORETURN.
354 (fatal_insn_not_found, fatal_insn, really_sorry,
355 push_float_handler, pop_float_handler): Add prototypes.
356 (fancy_abort): Mark prototype with ATTRIBUTE_NORETURN.
357 (do_abort, botch): Add prototypes.
359 Sat Sep 6 12:05:18 1998 John Carr <jfc@mit.edu>
361 * final.c (final): If a label is reached only from a single jump,
362 call NOTICE_UPDATE_CC on the jump and its predecessor before
363 emitting the insn after the label.
365 * i386.h: Add AMD K6 support.
366 Change TARGET_* macros to use table lookup.
367 (INITIALIZE_TRAMPOLINE): Improve trampoline code.
368 (ADJUST_COST): Change definition to call function in i386.c.
369 (ISSUE_RATE): Define as 2 for anything newer than an 80486.
370 * i386.c: Add AMD K6 support.
371 Add constants for feature tests used by TARGET_* macros.
372 (split_di): If before reload, call gen_lowpart and gen_highpart.
373 (x86_adjust_cost): New function.
374 (put_jump_code): New function.
375 (print_operand): New codes 'D' and 'd'.
376 * i386.md: New insn types. New insn attribute "memory".
377 Redefine scheduling parameters to use new types and add AMD K6
378 support. Explicitly set type of most insns.
379 (move insns): K6 prefers movl $0,reg to xorl reg,reg. Pentium
380 Pro and K6 prefer movl $1,reg to incl reg.
381 (adddi3, subdi3): Set cc_status.
382 (DImode shift patterns): Change label counters from HOST_WIDE_INT
383 to int; x86 can't have more than 2^31 DImode shifts per file.
384 (setcc): Combine all setcc patterns. Allow writing memory.
385 Combine all jump patterns using match_operator.
386 (*bzero): Name pattern. Emit mutliple stos instructions when that
387 is faster than rep stos.
388 (xordi3, anddi3, iordi3): Simplify DImode logical patterns and
391 Sun Sep 6 11:17:20 1998 Dave Love <d.love@dl.ac.uk>
393 * config/m68k/x-next (BOOT_LDFLAGS): Define suitably for f771
396 Sat Sep 5 22:05:25 1998 Richard Henderson <rth@cygnus.com>
398 * alpha.c (alpha_ra_ever_killed): Inspect the topmost sequence,
399 not whatever we're generating now.
401 * alpha.c (set_frame_related_p, FRP): New.
402 (alpha_expand_prologue): Mark frame related insns.
403 (alpha_expand_epilogue): Likewise, but with a null FRP.
404 * alpha.h (INCOMING_RETURN_ADDR_RTX): New.
405 * alpha.md (exception_receiver): New.
406 * alpha/crtbegin.asm (.eh_frame): New beginning.
407 (__do_frame_setup, __do_frame_takedown): New.
408 * alpha/crtend.asm (.eh_frame): New ending.
409 * alpha/elf.h (DWARF2_DEBUGGING_INFO): Define.
410 (ASM_SPEC): Don't emit both dwarf2 and mdebug.
411 (ASM_FILE_START): Don't emit .file for dwarf2.
413 * rtl.h (enum reg_note): Add REG_FRAME_RELATED_EXPR.
414 * rtl.c (reg_note_name): Likewise.
415 * rtl.texi (REG_NOTES): Likewise.
416 * dwarf2out.c (dwarf2out_frame_debug): Use it. Recognize a store
419 Sat Sep 5 14:47:17 1998 Richard Henderson <rth@cygnus.com>
421 * i386.h (PREFERRED_RELOAD_CLASS): Standard fp constants load to TOS.
422 * i386.md (movsf, movdf, movxf): Validate memory address returned
423 from force_const_mem. Kill useless REG_EQUAL setting code.
425 Sat Sep 5 14:23:31 1998 Torbjorn Granlund <tege@matematik.su.se>
427 * m68k.md (zero_extendsidi2): Fix typo.
429 Sat Sep 5 13:40:24 1998 Krister Walfridsson <cato@df.lth.se>
431 * configure.in: Removed references to the removed file
432 * config/xm-netbsd.h. Use ${cpu_type}/xm-netbsd.h for
433 * arm*-*-netbsd* and ns32k-*-netbsd*.
434 * config/i386/xm-netbsd.h: Removed unnecessary file.
435 * config/m68k/xm-netbsd.h: Likewise.
436 * config/sparc/xm-netbsd.h: Likewise.
437 * config/mips/xm-netbsd.h: Likewise.
439 Sat Aug 29 13:32:58 1998 Mumit Khan <khan@xraylith.wisc.edu>
441 * i386/cygwin32.h (BIGGEST_ALIGNMENT): Define.
442 (PCC_BITFIELD_TYPE_MATTERS): Define to be 0.
444 * i386/cygwin32.h (ASM_OUTPUT_SECTION_NAME): Don't check for
445 for exact section attributions.
447 * i386/mingw32.h (CPP_PREDEFINES): Add __MSVCRT__ for msvc
449 * i386/crtdll.h (CPP_PREDEFINES): Define.
451 Sat Sep 5 03:23:05 1998 Jeffrey A Law (law@cygnus.com)
453 * m68k.md (5200 movqi): Do not allow byte sized memory references
455 * m68k.c (output_move_qimode): Do not use byte sized operations on
458 * Makefile.in (pexecute.o): Use pexecute.c from libiberty. Provide
459 explicit rules for building. Similarly for alloca, vfprintf,
460 choose-temp and mkstemp, getopt, getopt1, and obstack.
461 (INCLUDES): Add $(srcdir)/../include.
462 * pexecute.c, alloca.c, vfprintf.c, choose-temp.c, mkstemp.c: Delete.
463 * getopt.h, getopt.c getopt1.c, obstack.c, obstack.h: Likewise.
465 Fri Sep 4 11:57:50 1998 Tom Tromey <tromey@cygnus.com>
467 * gcc.c (do_spec_1): [case 'o'] Account for
468 lang_specific_extra_outfiles.
469 (main): Correctly clear all slots in outfiles for
470 lang_specific_extra_outfiles. Set input_file_number before
471 calling lang_specific_pre_link.
473 Fri Sep 4 10:37:07 1998 Jim Wilson <wilson@cygnus.com>
475 * loop.c (load_mems): Fix JUMP_LABEL field after for_each_rtx call.
477 Fri Sep 4 02:01:05 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
479 * config/sparc/sparc.c (output_double_int): In all V9 symbolic
481 (sparc_output_deferred_case_vectors): If no work to do, return.
482 Fix thinko in Sept 1 change.
484 1998-09-03 SL Baur <steve@altair.xemacs.org>
486 * Makefile.in: add semicolon in BISON definition for portability.
488 Thu Sep 3 13:34:41 1998 Toon Moene <toon@moene.indiv.nluug.nl>
490 * config/nextstep.c (handle_pragma): Correct name of third
493 Tue Sep 1 11:30:33 1998 Nick Clifton <nickc@cygnus.com>
495 * config/m32r/m32r.md: Change (reg:CC 17) to (reg:SI 17).
496 * config/m32r/m32r.h: Make register 17 be fixed.
497 * config/m32r/m32r.c: Use SImode for cc operations.
499 Thu Sep 3 18:17:34 1998 Benjamin Kosnik <bkoz@cygnus.com>
501 * invoke.texi (Warning Options): Add -Wnon-template-friend
504 Thu Sep 3 18:16:16 1998 Michael Meissner <meissner@cygnus.com>
506 * rs6000.c (rs6000_override_options): Add -mcpu={401,e603e}.
508 Thu Sep 3 18:05:16 1998 David Edelsohn <edelsohn@mhpcc.edu>
510 * rs6000.md (movsf): Disable explicit secondary-reload-like
511 functionality if TARGET_POWERPC64.
512 (movdf): Remove TARGET_POWERPC64 explicit secondary-reload-like
515 Thu Sep 3 11:41:40 1998 Robert Lipe <robertl@dgii.com>
517 * fixinc.sco: Borrow code to wrap 'bool' typedefs from tinfo.h
518 and term.h from fixinc.wrap.
520 Thu Sep 3 09:47:31 1998 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
522 * aclocal.m4 (GCC_HEADER_STRING): New macro to detect if it is
523 safe to include both string.h and strings.h together.
524 (GCC_NEED_DECLARATION): Test STRING_WITH_STRINGS when deciding
525 which headers to search for function declarations. Continue to
526 prefer string.h over strings.h when both are not acceptable.
528 * acconfig.h (STRING_WITH_STRINGS): Add stub.
530 * configure.in: Call GCC_HEADER_STRING.
532 * system.h: Test STRING_WITH_STRINGS when deciding which headers
533 to include. Continue to prefer string.h over strings.h when both
536 Wed Sep 2 23:56:29 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
538 * config/sparc/sparc.c (output_double_int): If V9 and MEDLOW, do
539 not assume top 32-bits of symbolic addresses are zero if
542 Thu Sep 3 00:23:21 1998 Richard Henderson <rth@cygnus.com>
544 * ginclude/va-alpha.h: Protect entire second portion of the
545 file against double inclusion.
547 Thu Sep 3 00:37:55 1998 Ovidiu Predescu <ovidiu@aracnet.com>
549 Added support for the Boehm's garbage collector.
550 * configure.in: Handle --enable-objc-gc.
551 * configure: Rebuilt.
552 * Makefile.in (CHECK_TARGETS): Add check-objc.
553 (check-objc): New rule.
554 * objc/Make-lang.in: Build a different Objective-C library that
555 runs with the Boehm's collector.
556 * objc/encoding.c (objc_round_acc_size_for_types): New function.
557 * objc/encoding.c: Correctly compute the size of compound types in
558 the presence of bitfields. Skip the variable name of the type if
559 any. Added support for long long.
560 * objc/encoding.h (_C_GCINVISIBLE): New specifier.
561 (_F_GCINVISIBLE): New mask.
562 * objc/gc.c: New file. Compute the type memory mask associated with
563 a class based on the runtime information.
564 * objc/misc.c: Added the hooks that use the Boehm's collector
565 allocation functions.
566 * objc/objc-act.c (build_class_template): Generate a new class
567 member (gc_object_type) to hold the class' type memory mask.
568 (build_shared_structure_initializer): Initialize the new member to
570 (encode_complete_bitfield): New function. Generate the new
572 (encode_field_decl): Generate the new encoding only for the GNU
574 * objc/objc-api.h (_C_LNG_LNG, _C_ULNG_LNG): New specifiers for the
576 (class_get_gc_object_type): New function to mark a pointer instance
577 variable as a weak pointer.
578 * objc/objc-features.texi: New file.
579 * objc/objc.h (gc_object_type): New class member.
580 * objc/objects.c (class_create_instance): Create a typed memory
581 object when compiled with Boehm's collector support.
582 * objc/sendmsg.c (__objc_init_install_dtable): Call
583 __objc_send_initialize instead of setting the initialize flag.
584 (__objc_send_initialize): Call __objc_generate_gc_type_description
585 to generate the class type memory mask. Rewrite the code that
586 sends the +initialize so that it is called only once (bug report
587 and fix from Ronald Pijnacker <Ronald.Pijnacker@best.ms.philips.com>).
588 * testsuite/objc: New testsuite for Objective-C type encoding.
589 * testsuite/lib/objc-torture.exp: New file.
590 * testsuite/lib/objc.exp: New file.
592 Wed Sep 2 14:47:36 1998 Jim Wilson <wilson@cygnus.com>
594 * jump.c (jump_optimize): In if/then/else transformations, add
595 another call to modified_between_p for the jump insn.
597 Wed Sep 2 14:16:49 1998 Jeffrey A Law (law@cygnus.com)
599 * fix-header.c (symlink): Treat like readlink.
601 Wed Sep 2 19:30:06 1998 J"orn Rennecke <amylaar@cygnus.co.uk>
603 * dwarfout.c (fundamental_type_code): Encode 32 bit floats/doubles
606 Wed Sep 2 10:06:07 1998 Nick Clifton <nickc@cygnus.com>
608 * config/nextstep.h: Update HANDLE_PRAGMA macro.
609 * config/h8300/h8300.h: Update HANDLE_PRAGMA macro.
610 * config/i960/i960.h: Update HANDLE_PRAGMA macro.
612 * config/nextstep.c (handle_pragma): Take three arguments, as per
613 the new HANDLE_PRAGMA macro specification.
614 * config/h8300/h8300.c (handle_pragma): Take three arguments, as
615 per the new HANDLE_PRAGMA macro specification.
616 * config/i960/i960.c (process_pragma): Take three arguments, as
617 per the new HANDLE_PRAGMA macro specification.
619 Wed Sep 2 09:25:29 1998 Nick Clifton <nickc@cygnus.com>
621 * c-lex.c (check_newline): Call HANDLE_PRAGMA before
622 HANDLE_SYSV_PRAGMA if both are defined. Generate warning messages
623 if unknown pragmas are encountered.
624 (handle_sysv_pragma): Interpret return code from
625 handle_pragma_token (). Return success/failure indication rather
626 than next unprocessed character.
627 (pragma_getc): New function: retrieves characters from the
628 input stream. Defined when HANDLE_PRAGMA is enabled.
629 (pragma_ungetc): New function: replaces characters back into the
630 input stream. Defined when HANDLE_PRAGMA is enabled.
632 * c-pragma.c (handle_pragma_token): Return success/failure status
635 * c-pragma.h: Change prototype of handle_pragma_token().
637 * varasm.c: (handle_pragma_weak): Only create this function if
638 HANDLE_PRAGMA_WEAK is defined.
640 * c-common,c (decl_attributes): If defined call the expression
641 contained within the INSERT_ATTRIBUTES macro before adding
642 attributes to a decl.
644 * tm.texi (HANDLE_PRAGMA): Document the new verion of
645 HANDLE_PRAGMA, which takes three arguments.
646 (INSERT_ATTRIBUTES): Document this new macro.
648 * LANGUAGES: Document the new version of HANDLE_PRAGMA and the
649 new INSERT_ATTRIBUTES macro.
651 Wed Sep 2 02:03:23 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
653 * config/sparc/sparc.md (movdf): Only generate special RTL for
655 (move_label_di): Remove
656 (movdi_pic_label_ref, movdi_high_pic_label_ref,
657 movdi_lo_sum_pic_label_ref): New patterns for 64-bit label
659 * config/sparc/sparc.h (ASM_OUTPUT_ADDR_VEC_ELT,
660 ASM_OUTPUT_ADDR_DIFF_ELT): Don't do anything special for MEDLOW,
661 output an .xword for all 64-bit cases.
663 Tue Sep 1 15:55:17 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
665 * config/sparc/sparc.c (finalize_pic): Don't output arbitrary
666 alignment, use FUNCTION_BOUNDARY instead.
667 (sparc_output_deferred_case_vectors): Likewise.
669 Mon Aug 31 17:25:41 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
671 * config/sparc/sparc.md (movsf_const_intreg): Kill warning.
672 (movtf_insn_sp64, movtf_no_e_insn_sp64): Reorder alternatives.
674 Mon Aug 31 13:57:55 1998 Richard Henderson <rth@cygnus.com>
676 * alpha/va_list.h: New file.
677 * alpha/x-alpha (EXTRA_HEADERS): New. Add va_list.h.
679 Mon Aug 31 14:55:02 1998 Jeffrey A Law (law@cygnus.com)
681 * NEWS: Add SCO Openserver and Unixware 7 notes.
685 Mon Aug 31 15:42:18 1998 Dave Brolley <brolley@cygnus.com>
687 * varasm.c (compare_constant_1): Handle RANGE_EXPR.
688 (record_constant_1): Handle RANGE_EXPR.
690 Mon Aug 31 10:54:03 1998 Richard Henderson <rth@cygnus.com>
692 * print-rtl.c (print_rtx): NOTE_INSN_LIVE has an rtx not a bitmap.
693 * haifa-sched.c (sched_analyze): Handle NOTE_INSN_RANGE_START
694 and NOTE_INSN_RANGE_END specially.
695 (reemit_notes): Likewise.
697 Mon Aug 31 10:18:52 1998 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
699 * sparc.c (TMASK, UMASK): Use `(unsigned)1' not `1U'.
700 (ultrasparc_sched_init): Remove unneeded &.
702 Mon Aug 31 10:47:16 1998 Andreas Schwab <schwab@issan.informatik.uni-dortmund.de>
704 * config/m68k/m68k.h (TARGET_SWITCHES): Don't remove MASK_68040
705 for m68020-60, to prevent the use of fintrz.
707 Sun Aug 30 22:17:20 1998 Mark Mitchell <mark@markmitchell.com>
709 * configure.in: If the native compiler is GCC use $(WARN_CFLAGS)
711 * Makefile.in: Likewise.
712 * configure: Regenerated.
714 Sun Aug 30 22:15:41 1998 H.J. Lu (hjl@gnu.org)
716 * configure.in (gxx_include_dir): Changed to
717 '${prefix}/include/g++'-${libstdcxx_interface}.
718 * configure: Rebuilt.
720 Sun Aug 30 20:19:43 1998 Hans-Peter Nilsson <hp@axis.se>
722 * expr.c (expand_expr): Change ">" to ">=" making MOVE_RATIO use
724 * tm.texi (Costs): Say MOVE_RATIO is number of mem-mem move
725 *sequences* *below* which scalar moves will be used.
727 Sun Aug 30 17:18:43 1998 Jeffrey A Law (law@cygnus.com)
729 * collect2.c (mktemp): Delete unused declaration.
731 * config/xm-netbsd.h: Remove unnecessary file.
732 * config/*/xm-netbsd.h: Do not include the generic xm-netbsd.h
733 file anymore, it is not needed.
735 Sun Aug 30 16:05:45 1998 Mark Mitchell <mark@markmitchell.com>
737 * convert.c (convert_to_integer): Issue an error on conversions to
740 Sun Aug 30 16:47:20 1998 Martin von Lvwis <loewis@informatik.hu-berlin.de>
742 * Makefile.in: Add lang_tree_files and gencheck.h.
743 * configure.in: Generate them.
744 * gencheck.c: Include gencheck.h.
746 Sat Aug 29 21:38:24 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
748 * config/sparc/sparc.md (pic_lo_sum_di, pic_sethi_di): Rename to
749 movdi_lo_sum_pic and movdi_high_pic and make visible.
750 * config/sparc/sparc.c (legitimize_pic_address): For -fPIC,
751 emit these when Pmode is not SImode.
752 * config/sparc/linux64.h (SPARC_DEFAULT_CMODEL): Make CM_MEDLOW.
754 Sat Aug 29 14:59:32 1998 Mumit Khan <khan@xraylith.wisc.edu>
756 * i386/cygwin32.h (ASM_OUTPUT_SECTION_NAME): Don't emit
757 .linkonce directive after the first time.
759 Sat Aug 29 12:39:56 1998 Jeffrey A Law (law@cygnus.com)
761 * m68k.md (beq0_di): Generate correct (and more efficient) code when
762 the clobbered operand overlaps with an input.
763 (bne0_di): Similarly.
765 * Makefile.in (INSTALL): Remove "--no-header" argument.
767 * NEWS: Various updates.
769 Fri Aug 28 19:00:44 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
771 * config/sparc/sparc.c (arith_operand, const64_operand,
772 const64_high_operand, arith_double_4096_operand): Mark mode as
774 (create_simple_focus_bits): Remove unused arg highest_bit_set, all
776 (sparc_emit_set_const64): Remove unused variable i.
777 (sparc_splitdi_legitimate): Likewise for addr_part.
778 (ultra_code_from_mask): Likewise for mask.
779 (ultra_cmove_results_ready_p): Fixup entry modulo calc. and
780 reverse return values so it matches usage and comments.
781 (ultra_flush_pipeline): Likewise.
782 (ultra_fpmode_conflict_exists): Likewise, remove unused variable
783 this_type, and allow loads and stores of differing FP modes as
784 they do not create a conflict.
785 (ultra_find_type): Initialize fpmode to SFmode, fix
786 parenthesization thinkos in large conditional.
787 (ultrasparc_sched_init): Mark dump and sched_verbose as unused.
788 Init free_slot_mask after ultra_cur_hist is reset, not before.
789 (ultrasparc_rescan_pipeline_state): Remove unused variable ucode.
790 (ultrasparc_sched_reorder): Don't bzero current pipeline state,
791 use ultra_flush_pipeline instead, then re-init group pointer.
792 Fix statement with no effect. If no progress made in, and no
793 instructions scheduled at all, advance to new pipeline cycle else
794 we get into an endless loop.
795 (ultrasparc_adjust_cost): Remove previous arg.
796 * config/sparc/sparc.h (ADJUST_COST): Update to reflect that.
798 Fri Aug 28 13:52:35 1998 Jim Wilson <wilson@cygnus.com>
800 * sparc.md (DImode, DFmode, TFmode splits): Delete self_reference
801 code. Use reg_overlap_mentioned_p to detect when source and
803 (negtf2_notv9+1): Use DFmode instead of SFmode in last two operands.
805 1998-08-28 Brendan Kehoe <brendan@cygnus.com>
807 * loop.c (check_dbra_loop): Pass COMPARISON_VALUE, not
808 COMPARISON_VAL, into invariant_p.
810 Fri Aug 28 15:13:25 1998 J"orn Rennecke <amylaar@cygnus.co.uk>
812 * regmove.c (regclass_compatible_p): New function.
813 (regmove_optimize): Use it.
815 Use NREGS parameter insted of calling max_reg_num.
817 (fixup_match_1): Don't use code = MINUS when later tieing with
818 a hard register is likely.
820 Fri Aug 28 14:54:07 1998 J"orn Rennecke <amylaar@cygnus.co.uk>
822 * loop.c (check_dbra_loop): Fix calculation of FINAL_VALUE when
823 COMPARISON_VAL was normalized.
825 Thu Aug 27 20:10:46 1998 Jeffrey A Law (law@cygnus.com)
827 * loop.c (check_dbra_loop): The loop ending comparison value
828 must be an invariant or we can not reverse the loop.
830 * loop.c (scan_loop): Count down from max_reg_num - 1 to
831 FIRST_PSEUDO_REGISTER to avoid calling max_reg_num each iteration
833 (load_mems_and_recount_loop_regs_set): Likewise.
835 * i386.c (print_operand): Remove obsolete 'c' docs.
837 Wed Aug 26 17:13:37 1998 Tom Tromey <tromey@cygnus.com>
839 * gthr.h: Document __GTHREAD_MUTEX_INIT_FUNCTION.
840 * frame.c (init_object_mutex): New function.
841 (init_object_mutex_once): Likewise.
843 (__register_frame_info): Likewise.
844 (__register_frame_info_table): Likewise.
845 (__deregister_frame_info): Likewise.
847 Thu Aug 27 15:14:18 1998 Jeffrey A Law (law@cygnus.com)
849 * haifa-sched.c (sched_analyze_insn): Fix thinko in last change.
851 Thu Aug 27 16:34:51 1998 J"orn Rennecke <amylaar@cygnus.co.uk>
853 * loop.c (check_dbra_loop): Enable code for reversal
854 of some loops without a known constant loop end.
856 Wed Aug 26 18:38:15 1998 Richard Henderson <rth@cygnus.com>
858 * haifa-sched.c (last_clock_var): New.
859 (schedule_block): Initialize it.
860 (schedule_insn): Use it to fill insn modes with issue information.
862 * alpha.c (alpha_handle_trap_shadows): Remove do-nothing exit.
863 Tag trapb and next insn with TImode.
864 (alphaev5_insn_pipe, alphaev5_next_group, alphaev5_align_insns): New.
865 (alpha_reorg): Add conditional for alpha_handle_trap_shadows.
866 Invoke alphaev5_align_insns as appropriate.
867 * alpha.h (LABEL_ALIGN_AFTER_BARRIER): Was ALIGN_LABEL_AFTER_BARRIER.
868 (MD_SCHED_VARIABLE_ISSUE): New.
869 * alpha.md (attr type): Add multi.
870 (define_asm_attributes): New.
871 (prologue_stack_probe_loop, builtin_setjmp_receiver): Set type multi.
872 (arg_home): Likewise.
873 (fnop, unop, realign): New.
875 Wed Aug 26 15:55:41 1998 Jim Wilson <wilson@cygnus.com>
877 * iris5.h (PREFERRED_DEBUGGING_TYPE): Undef.
878 * iris5gas.h (PREFERRED_DEBUGGING_TYPE): Define.
880 * configure.in (powerpc-ibm-aix4.[12]*): Change from 4.[12].*.
881 (rs6000-ibm-aix4.[12]*): Likewise.
882 * configure: Regnerate.
884 Wed Aug 26 09:30:59 1998 Nick Clifton <nickc@cygnus.com>
886 * config/arm/thumb.c (thumb_exit): Do not move a4 into lr if it
887 already contains the return address.
889 Wed Aug 26 12:57:09 1998 Jeffrey A Law (law@cygnus.com)
891 * calls.c (expand_call): Use bitfield instructions to extract/deposit
892 word sized hunks when loading unaligned args into registers.
894 * haifa-sched.c (sched_analyze_insn): Only create scheduling
895 barriers for LOOP, EH and SETJMP notes on the loop_notes list.
897 * mn10300.h (RTX_COSTS): Handle UDIV and UMOD too.
899 Wed Aug 26 16:35:37 1998 J"orn Rennecke <amylaar@cygnus.co.uk>
901 * loop.c (check_dbra_loop): Add some code that would allow reversal
902 of some loops without a known constant loop end if it were enabled.
904 Wed Aug 26 11:08:44 1998 Gavin Romig-Koch <gavin@cygnus.com>
906 * mips.md (lshrsi3_internal2+2): Fix type-o.
908 Wed Aug 26 10:53:03 1998 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
910 * system.h: Include stdarg.h/varargs.h, make sure they are ordered
911 correctly with regards to stdio.h.
913 * calls.c: Remove stdarg.h/varargs.h.
916 * combine.c: Likewise.
917 * cpperror.c: Likewise.
918 * cpplib.c: Likewise.
919 * cpplib.h: Likewise.
920 * doprint.c: Likewise.
921 * emit-rtl.c: Likewise.
923 * fix-header.c: Likewise.
925 * genattr.c: Likewise.
926 * genattrtab.c: Likewise.
927 * gencodes.c: Likewise.
928 * genconfig.c: Likewise.
929 * genemit.c: Likewise.
930 * genextract.c: Likewise.
931 * genflags.c: Likewise.
932 * genopinit.c: Likewise.
933 * genoutput.c: Likewise.
934 * genpeep.c: Likewise.
935 * genrecog.c: Likewise.
936 * mips-tfile.c: Likewise.
937 * prefix.c: Likewise.
938 * protoize.c: Likewise.
939 * regmove.c: Likewise.
940 * toplev.c: Likewise.
943 Wed Aug 26 05:09:27 1998 Jakub Jelinek <jj@sunsite.ms.mff.cuni.cz>
945 * config/sparc/sparc.c (sparc_override_options): If not
946 TARGET_FPU, turn off TARGET_VIS.
947 * config/sparc/sparc.h (TARGET_SWITCHES): Add no-vis.
948 (LEGITIMATE_CONSTANT_P): Allow SF/DF mode zero when TARGET_VIS.
949 * config/sparc/sparc.md (movsi_insn): Use fzeros not fzero.
950 (movdi_insn_sp64): Add VIS fzero alternative.
951 (clear_sf, clear_df): New VIS patterns.
952 (movsf, movdf expanders): Allow fp_zero_operand flat out when
954 (one_cmpldi2_sp64): Provide new fnot1 VIS alternative.
956 Tue Aug 25 10:57:41 1998 Mark Mitchell <mark@markmitchell.com>
958 * loop.c (n_times_set, n_times_used, may_not_optimize,
959 reg_single_usage): Convert to varrays. All uses changed.
960 (insert_loop_mem): Return a value.
961 (scan_loop): Tweak AVOID_CC_MODE_COPIES code.
962 (load_mems_and_recount_loop_regs_set): Likewise. Grow the arrays, if
965 Tue Aug 25 23:57:12 1998 Jeffrey A Law (law@cygnus.com)
968 * configure.in: Do not set thread_file to "irix" since no such
971 * sparc.md (float abs/neg splits): Check reload_completed before
972 calling alter_subreg.
974 Tue Aug 25 19:17:59 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
976 * config/sparc/sparc.c (sparc_absnegfloat_split_legitimate): New
978 * config/sparc/sparc.h: Declare it.
979 * config/sparc/sparc.md (float abs/neg splits): Use it.
980 (all other splits): Handle SUBREGs properly where necessary.
981 (unnamed (1<<x)-1 V8PLUS pattern): Disable for now.
983 Tue Aug 25 19:48:46 1998 Jeffrey A Law (law@cygnus.com)
985 * reorg.c (fill_simple_delay_slots): Do not abort if we encounter
986 an insn on the unfilled_slots_list that has no delay slots.
987 (fill_eager_delay_slots): Similarly.
989 Tue Aug 25 13:35:20 1998 Nick Clifton <nickc@cygnus.com>
991 * config/v850/v850.c (movsi_source_operand): Treat CONSTANT_P_RTX
992 as an ordinary operand.
994 Tue Aug 25 12:54:57 1998 Jason Merrill <jason@yorick.cygnus.com>
996 * tree.c (valid_machine_attribute): Don't apply attributes to both
999 Tue Aug 25 12:23:20 PDT 1998 Richard Henderson <rth@cygnus.com>
1001 * reload.c (operands_match_p): Handle rtvecs.
1003 * i386.c (legitimate_pic_address_disp_p): New.
1004 (legitimate_address_p): Use it.
1005 (legitimize_pic_address): Use unspecs to represent @GOT and @GOTOFF.
1006 Handle constant pool symbols just like statics.
1007 (emit_pic_move): Use Pmode not SImode for clarity.
1008 (output_pic_addr_const) [SYMBOL_REF]: Remove @GOT and @GOTOFF hacks.
1009 [UNSPEC]: New, handling what we killed above.
1010 [PLUS]: Detect and abort on invalid symbol arithmetic.
1011 * i386.h (CONSTANT_ADDRESS_P): Remove HIGH.
1013 Tue Aug 25 12:02:23 1998 Mark Mitchell <mark@markmitchell.com>
1015 * alias.c: Include output.h.
1016 (DIFFERENT_ALIAS_SETS_P): Don't treat alias sets as
1017 different if we're in a varargs function.
1018 * Makefile.in (alias.o): Depend on output.h
1020 Tue Aug 25 19:20:12 1998 J"orn Rennecke <amylaar@cygnus.co.uk>
1022 * sh.h (GIV_SORT_CRITERION): Delete.
1024 Tue Aug 25 13:19:46 1998 Dave Brolley <brolley@cygnus.com>
1026 * regclass.c (regclass): Use xmalloc/free instead of alloca.
1027 * stupid.c (stupid_life_analysis): Ditto.
1028 * reload1.c (reload): Ditto.
1030 Tue Aug 25 05:48:18 1998 Jakub Jelinek <jj@sunsite.ms.mff.cuni.cz>
1032 * config/sparc/sparc.c (arith_4096_operand, arith_add_operand,
1033 arith_double_4096_operand, arith_double_add_operand): New
1035 * config/sparc/sparc.h (PREDICATE_CODES): Add them, declare them.
1036 * config/sparc/sparc.md (adddi3, addsi3, subdi3, subsi3): Use
1037 them to transform add/sub 4096 into add/sub -4096.
1039 Mon Aug 24 23:31:03 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
1041 * loop.c (scan_loop): Allocate some slop to handle pseudos
1042 generated by move_movables.
1043 (load_mems_and_recount_loop_regs_set): Honor AVOID_CC_MODE_COPIES
1046 Mon Aug 24 19:45:40 1998 Jim Wilson <wilson@cygnus.com>
1048 * tree.def (DECL_RESULT): Correct documentation.
1050 Tue Aug 25 01:15:27 1998 J"orn Rennecke <amylaar@cygnus.co.uk>
1052 * reload1.c (reload_reg_free_before_p): New argument EQUIV; Changed
1053 all callers. Abort for RELOAD_FOR_INSN. RELOAD_FOR_OUTADDR_ADDR:
1054 conflicts will all RELOAD_FOR_OUTPUT reloads.
1056 * reload1.c (reload_cse_regs_1): When deleting a no-op move that
1057 loads the function result, substitute with a USE.
1059 Mon Aug 24 15:20:19 1998 David Edelsohn <edelsohn@mhpcc.edu>
1061 * rs6000.h (GO_IF_LEGITIMATE_ADDRESS): Use TARGET_POWERPC64
1062 when testing LEGITIMATE_INDEXED_ADDRESS_P DFmode and DImode.
1063 (LEGITIMIZE_ADDRESS): Use TARGET_POWERPC64 for INDEXED fixup.
1064 * rs6000.c (print_operand, case 'L'): Add UNITS_PER_WORD, not 4.
1065 (print_operand, cases 'O' and 'T'): Fix typos in lossage strings.
1066 * rs6000.md (fix_truncdfsi2_store): Remove %w from non-CONST_INT
1068 (movdf_softfloat32, movdf_hardfloat64, movdf_softfloat64): Change
1069 'o' to 'm' for GPR variant constraints.
1071 Mon Aug 24 10:25:46 1998 Jeffrey A Law (law@cygnus.com)
1073 * loop.c (scan_loop): Honor AVOID_CC_MODE_COPIES.
1075 * h8300.h (STRIP_NAME_ENCODING): Fix typo.
1077 * sparc.md (TFmode splits): Use reg_overlap_mentioned_p to detect
1078 when the source and destination overlap.
1080 * stmt.c (emit_case_nodes): Change rtx_function to rtx_fn to avoid
1081 clash with global type.
1083 Mon Aug 24 00:53:53 1998 Jason Merrill <jason@yorick.cygnus.com>
1085 * fixinc.irix: Add curses.h handling from fixinc.wrap.
1087 * c-common.c (combine_strings): Also set TREE_READONLY.
1088 Change warn_write_strings to flag_const_strings.
1089 * c-decl.c, c-tree.h: Likewise.
1091 Sun Aug 23 18:39:11 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
1093 * config/sparc/sparc.c (sparc_emit_set_const32): If outputting a
1094 CONST_INT, not a symbolic reference, don't use a HIGH/LO_SUM
1095 sequence, use SET/IOR instead so CSE can see it.
1096 * config/sparc/sparc.md (movhi_const64_special,
1097 movsi_const64_special): New patterns necessitated by that change.
1098 (movhi_high): Remove.
1099 (movhi_lo_sum): Change to match an IOR.
1100 (movdf_insn_sp32): Test TARGET_V9 not TARGET_ARCH64.
1101 (movdf_insn_v9only): New pattern for when V9 but not ARCH64.
1102 (movdf_insn_sp64): Test both TARGET_V9 and TARGET_ARCH64.
1103 (movdf splits): Allow when not V9 or when not ARCH64 and integer
1104 registers are involved.
1105 (snesi_zero_extend split): Remove reload_completed test.
1106 (unnamed plus and minus zero_extend sidi splits): Add it.
1108 Sun Aug 23 11:56:08 1998 Mark Mitchell <mark@markmitchell.com>
1110 * extend.texi: Remove description of extension to explicit
1111 instantiation that is now endorsed by standard C++.
1113 Sun Aug 23 09:39:09 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
1115 * config/arc/arc.c (arc_initialize_pic): Remove.
1116 * config/arc/arc.h (INITIALIZE_PIC): Similarly, this routine does
1117 nothing on any platform and is invoked by no-one, it does not even
1118 appear in the documentation.
1119 * config/sparc/sparc.h (INITIALIZE_PIC): Likewise.
1120 * config/sparc/sparc.c (initialize_pic): Likewise.
1121 (find_addr_reg): Remove this as well, no longer referenced after
1124 Sun Aug 23 00:17:14 1998 Jeffrey A Law (law@cygnus.com)
1126 * recog.c (validate_replace_rtx_group): New function.
1127 * recog.h (validate_replace_rtx_group): Declare it.
1128 * regmove.c (optimize_reg_copy_3): If any substitution fails, then undo
1129 the entire group of substitutions.
1131 Sat Aug 22 23:31:00 1998 Klaus-Georg Adams (Klaus-Georg.Adams@chemie.uni-karlsruhe.de)
1133 * loop.c (load_mems): Fix initializers.
1135 Fri Aug 21 23:07:46 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
1137 * config/sparc/sparc.md (TFmode splits): Handle destination
1138 registers being referenced in the address correctly.
1140 * expmed.c (make_tree) [CONST_INT]: Sign extend even if
1141 TREE_UNSIGNED, when bitsize of type's mode is larger than
1142 HOST_BITS_PER_WIDE_INT.
1144 Fri Aug 21 19:31:31 1998 Alexandre Petit-Bianco <apbianco@cygnus.com>
1146 * tree.def (LABELED_BLOCK_EXPR, EXIT_BLOCK_EXPR): New tree nodes.
1147 * tree.h (LABELED_BLOCK_LABEL, LABELED_BLOCK_BODY,
1148 EXIT_BLOCK_LABELED_BLOCK, EXIT_BLOCK_RETURN, LOOP_EXPR_BODY): New
1150 * expr.c (expand_expr): Handle LABELED_BLOCK_EXPR and
1153 Thu Aug 20 19:43:44 1998 Jeffrey A Law (law@cygnus.com)
1155 * h8300.c (h8300_encode_label): Use '&' for tiny data items.
1156 * h8300.h (TINY_DATA_NAME_P): Likewise.
1157 (STRIP_NAME_ENCODING): Handle '&'.
1159 * mn10200.h (REG_OK_FOR_INDEX_P): Do not check the mode of the
1160 register (it could be accessed via an outer SUBREG).
1161 (REG_OK_FOR_BASE_P): Likewwise.
1162 (GO_IF_LEGITIMATE_ADDRESS): Consistently use REGNO_OK_FOR_BASE_P.
1164 * remove.c (optimize_reg_copy_3): Abort instead of silently generating
1167 * jump.c (rtx_renumbered_equal_p): Do not consider PLUS commutative.
1169 Thu Aug 20 17:35:20 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
1171 * config/sparc/sparc.md (movtf_insn_sp32): All memory operands
1172 must be offsettable so the splits can be made.
1174 Thu Aug 20 13:56:53 1998 Michael Meissner <meissner@cygnus.com>
1176 * config/i386/winnt.c: Include system.h, not stdio.h to get
1177 sys/param.h pulled in before rtl.h in case the system defines MIN
1180 Thu Aug 20 13:44:20 1998 David Edelsohn <edelsohn@mhpcc.edu>
1182 * rs6000.md (movqi, movhi): Add CONSTANT_P_RTX.
1184 Thu Aug 20 13:15:11 1998 Dave Brolley <brolley@cygnus.com>
1186 * stor-layout.c (layout_type): Compute TYPE_SIZE_UNIT correctly for
1188 * cpplib.c (cpp_define): Handle macros with parameters.
1190 Wed Aug 19 21:33:19 1998 David Edelsohn <edelsohn@mhpcc.edu>
1192 * rs6000.c (rs6000_output_load_toc_table): Use ld for 64-bit.
1193 (output_toc): Use single TOC slot or llong minimal-toc for DFmode
1194 and DImode 64-bit. Use llong for minimal-toc SFmode and
1195 SYMBOL_REF / LABEL_REF 64-bit.
1196 (output_function_profiler): Use llong for profiler label and ld to
1197 load 64-bit label address.
1199 Wed Aug 19 17:52:27 1998 Nick Clifton (nickc@cygnus.com)
1201 * config/arm/thumb.md (extendqisi2_insn): Cope with REG +
1204 Wed Aug 19 14:13:31 PDT 1998 Jeff Law (law@cygnus.com)
1206 * version.c: Bump for snapshot.
1208 Wed Aug 19 13:10:30 PDT 1998 Jeff Law (law@cygnus.com)
1210 * version.c: Bump for snapshot.
1212 Wed Aug 19 13:06:47 1998 Jason Merrill <jason@yorick.cygnus.com>
1214 * collect2.c (extract_init_priority): Use atoi instead of strtoul.
1216 Wed Aug 19 13:51:35 1998 Hans-Peter Nilsson <hp@axis.se>
1218 * tm.texi (Misc): Fix typo "teh".
1220 * tm.texi (PIC): Fix typo "PPIC".
1222 * tm.texi (Caller Saves): Say that DEFAULT_CALLER_SAVES has no
1223 effect when -O2 and higher.
1224 * invoke.texi (Optimize Options): Likewise for -fcaller-saves.
1226 1998-08-19 Michael Hayes <michaelh@ongaonga.chch.cri.nz>
1228 * regclass.c: Changed register set documentation to be consistent
1231 * final.c (final_start_function) Removed redundant test for
1234 Wed Aug 19 13:28:41 1998 Mark Mitchell <mark@markmitchell.com>
1236 * rtl.h (rtx_function): New type.
1237 (for_each_rtx): New function.
1238 * rtlanal.c (for_each_rtx): Define it.
1240 * recog.c (change_t): New type.
1241 (change_objects, change_old_codes, change_locs, change_olds):
1243 (changes): New variable.
1244 (validate_change): Dynamically allocate room for more changes, if
1245 necessary. Uses changes array instead of change_objects, etc.
1246 (apply_change_group): Use changes array instead of
1247 change_objects, etc.
1249 * loop.c (loop_mem_info): New type.
1250 (loop_mems): New variable.
1251 (loop_mems_idx): Likewise.
1252 (looop_mems_allocated): Likewise.
1253 (scan_loop): Remove nregs parameter.
1254 (next_insn_in_loop): New function.
1255 (load_mems_and_recount_loop_regs_set): Likewise.
1256 (load_mems): Likewise.
1257 (insert_loop_mem): Likewise.
1258 (replace_loop_mem): Likewise.
1259 (replace_label): Likewise.
1260 (INSN_IN_RANGE_P): New macro.
1261 (loop_optimize): Don't pass max_reg_num() to scan_loop.
1262 (scan_loop): Remove nregs parameter, compute it after any new
1263 registers are created by load_mems. Use INSN_IN_RANGE_P and
1264 next_insn_in_loop rather than expanding them inline. Call
1265 load_mems to load memory into pseudos, if appropriate.
1266 (prescan_loop): Figure out whether or not there are jumps from the
1267 loop to targets other than the label immediately following the
1268 loop. Call insert_loop_mem to notice all the MEMs used in the
1269 loop, if it could be safe to pull MEMs into REGs for the duration
1271 (strength_reduce): Use next_insn_in_loop. Tweak comments.
1273 Wed Aug 19 08:29:44 1998 Richard Earnshaw (rearnsha@arm.com)
1275 * arm.c (arm_override_options): Remove lie about ignoring PIC flag.
1277 Wed Aug 19 07:08:15 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
1279 * config/sparc/sparc.c (finalize_pic): Check for the correct
1280 nonlocal_goto_receiver UNSPEC number.
1281 * config/sparc/sparc.md (nonlocal_goto_receiver): Add comment
1282 making note of this dependency existing in sparc.c
1283 (negtf2_notv9 split): Give NEG SFmode.
1284 (negsf2): Fix insn output string.
1286 Tue Aug 18 12:40:27 1998 Richard Henderson <rth@cygnus.com>
1288 * c-common.c (decl_attributes): Issue an error if the argument
1289 to alias is not a string.
1291 Tue Aug 18 10:33:30 1998 Jeffrey A Law (law@cygnus.com)
1293 * haifa-sched.c (sched_analyze): Put all JUMP_INSNs on the last
1294 pending memory flush list.
1296 * combine.c (can_combine_p): Allow combining insns with REG_RETVAL
1298 (try_combine): Allow combining insns with REG_LIBCALL notes.
1300 * expr.c (emit_block_move): Do not call memcpy as a libcall
1301 instead build up a CALL_EXPR and call it like any other
1303 (clear_storage): Similarly for memset.
1305 * regmove.c (fixup_match_2): Do not call reg_overlap_mentioned_p
1308 * Makefile.in (cplus-dem.o): Provide explicit rules for building
1311 * regmove.c (optimize_reg_copy_1): Update REG_N_CALLS_CROSSED
1312 and REG_LIVE_LENGTH as successful substitutions are made.
1314 Tue Aug 18 07:15:27 1998 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
1316 * config/sparc/sparc.c (ultra_find_type): Add empty semicolon
1317 statement after end of loop label.
1319 Tue Aug 18 07:13:27 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
1321 * config/sparc/sparc.c (ultra_types_avail): New variable.
1322 (ultra_build_types_avail): New function to record mask of insn
1323 types in ready list at this cycle.
1324 (ultrasparc_sched_reorder): Call it.
1325 (ultra_find_type): Use it to quicken the search. Also simplif
1326 dependency check, don't use rtx_equal_p because we know exactly
1327 what we are looking for.
1329 Tue Aug 18 03:20:53 1998 Richard Earnshaw (rearnsha@arm.com)
1331 * arm.h (SECONDARY_INPUT_RELOAD_CLASS): Return NO_REGS if compiling
1332 for architecture v4.
1334 Mon Aug 17 21:26:38 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
1336 * config/sparc/sparc.md (sltu, sgeu): Don't FAIL, call
1338 (movsf_const_intreg, movsf_const_high, movsf_const_lo,
1339 movdf_const_intreg and helper splits): New patterns to move float
1340 constants into integer registers.
1341 (negtf2, negdf2, abstf2, absdf2): Rework using new patterns and
1344 Mon Aug 17 11:46:19 1998 Jeffrey A Law (law@cygnus.com)
1347 * tree.c (build_index_type): Copy TYPE_SIZE_UNIT from sizetype
1349 * c-decl.c (finish_enum): Copy TYPE_SIZ_UNIT from enumtype to tem.
1351 * rs6000.c (secondary_reload_class): For TARGET_ELF, indicate that
1352 a BASE_REGS register is needed as an intermediate when copying
1353 a symbolic value into any register class other than BASE_REGS.
1355 * expr.c (move_by_pieces): No longer static. Remove prototype.
1356 * rtl.h (move_by_pieces): Add extern prototype.
1357 * mips.c (expand_block_move): Handle aligned straight line copy by
1358 calling move_by_pieces.
1360 * expr.c (expand_expr): Allow assignments from TImode PARM_DECLs
1363 Mon Aug 17 10:28:52 1998 Mark Mitchell <mark@markmitchell.com>
1365 * stmt.c (expand_end_loop): Tidy. Allow unconditional
1366 jumps out of the loop to be treated as part of the exit test.
1368 Mon Aug 17 10:06:11 1998 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
1369 Jeff Law <law@cygnus.com>
1371 * Makefile.in (cplus-dep.o): Use cplus-dem.c from libiberty.
1372 * cplus-dem.c: Delete.
1374 * Makefile.in (fold-const.o): depend on $(RTL_H).
1376 * fold-const.c: Include rtl.h to get the prototype for
1377 `set_identifier_local_value'.
1379 * loop.c (express_from_1): Remove unused variable `tmp'.
1380 (combine_givs): cast the first argument of bzero to char *.
1382 * toplev.c (display_help): Remove unused variable `looking_for_start'.
1384 * c-decl.c (init_decl_processing): Remove unneeded &.
1386 * alpha.h (alpha_initialize_trampoline): Provide prototype.
1388 * except.c (set_exception_lang_code, set_exception_version_code):
1389 Change parameter from `short' to `int' to avoid using a gcc
1392 * except.h (set_exception_lang_code, set_exception_version_code):
1393 Likewise for prototypes.
1395 * flow.c (count_reg_references): Remove unused variables `regno'
1398 * gcse.c (hash_scan_insn): Declare parameter `in_libcall_block'.
1400 * prefix.c (translate_name): Cast the result of `alloca'.
1402 * varray.h (VARRAY_FREE): Reimplement as a `do-while(0)' statement.
1404 Mon Aug 17 09:23:42 1998 Andreas Schwab <schwab@issan.informatik.uni-dortmund.de>
1406 * config/m68k/m68k.c: Include "system.h" instead of <stdio.h>.
1408 (valid_dbcc_comparison_p): Mark mode argument as unused.
1409 (symbolic_operand): Likewise.
1410 (legitimize_pic_address): Likewise.
1411 (const_uint32_operand): Likewise.
1412 (const_sint32_operand): Likewise.
1413 * sched.c [!INSN_SCHEDULING]: Define only dummy function
1414 schedule_insns and comment out rest of file.
1416 * m68k.c (output_move_simode_const): Use subl to move a zero into an
1418 (output_move_[hq]imode): Likewise.
1420 Mon Aug 17 09:15:47 1998 Jeffrey A Law (law@cygnus.com)
1422 * toplev.c (main): Enable -fstrict-aliasing for -O2 and above.
1423 * invoke.texi: Corresponding changes.
1425 Mon Aug 17 02:03:55 1998 Richard Henderson <rth@cygnus.com>
1427 * regclass.c (allocate_reg_info): Respect MIN when clearing data.
1429 Sun Aug 16 17:37:06 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
1431 * config/sparc/sparc.c (ultra_code_from_mask,
1432 ultra_cmove_results_ready_p, ultra_fpmode_conflict_exists,
1433 ultra_find_type, ultra_schedule_insn, ultra_flush_pipeline,
1434 ultrasparc_sched_init, ultrasparc_variable_issue,
1435 ultra_rescan_pipeline_state, ultrasparc_sched_reorder): New
1436 functions to describe UltraSPARC pipeline exactly to Haifa.
1437 (ultrasparc_adjust_cost): Indicate IMUL type insns have zero cost,
1438 as there is nothing the scheduler can do about it. Indicate that
1439 REG_DEP_OUTPUT's collide. Fixup formatting.
1440 * config/sparc/sparc.h (RTX_COSTS): Fixup integer multiply and
1441 divide costs on Ultra for DImode.
1442 (MD_SCHED_INIT, MD_SCHED_REORDER, MD_SCHED_VARIABLE_ISSUE):
1444 * config/sparc/sparc.md (ieu_unnamed function unit): Rename to
1445 ieuN and add call_no_delay_slot to type list.
1446 (cti function unit): New unit for branches on UltraSPARC.
1447 (subx/addx insns): Set type to misc.
1448 (sidi zero/sign extension insns on arch64): Set type to shift.
1449 (sign_extendhidi2_insn): Set type to sload.
1451 Sun Aug 16 13:52:00 1998 David Edelsohn <edelsohn@mhpcc.edu>
1453 * rs6000.c (rs6000_stack_info): Use if == 0 for sizes.
1454 (output_epilog): Use if != 0 for offset.
1455 (rs6000_fatal_bad_address): Prepare for Intl.
1456 * rs6000.h (rs6000_fatal_bad_address): Declare.
1457 * rs6000.md (movsfcc, movdfcc): Use else if.
1458 (elf_high): Use {liu|lis}.
1459 (elf_low): Use {cal|la}. Remove %a template from old mnemonics.
1460 (movsi): Use rs6000_fatal_bad_address.
1462 Sun Aug 16 01:53:21 1998 Richard Henderson <rth@cygnus.com>
1464 * reload.c (find_equiv_reg): Reject equivalences separated
1465 by a volatile instruction.
1467 Sun Aug 16 00:21:44 1998 Franz Sirl <Franz.Sirl-kernel@lauterbach.com>
1469 * rs6000/linux.h (CPP_OS_DEFAULT_SPEC): Define.
1471 Sat Aug 15 20:51:35 1998 Richard Henderson <rth@cygnus.com>
1473 * alpha.md (movsicc): Fix mode mismatch.
1475 Sat Aug 15 20:22:33 1998 H.J. Lu (hjl@gnu.org)
1477 * config/alpha/alpha.h (ASM_OUTPUT_MI_THUNK): Handle aggregated
1479 * config/alpha/win-nt.h (ASM_OUTPUT_MI_THUNK): Likewise.
1481 Sat Aug 15 08:39:49 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
1483 * config/sparc/sparc.md (movsi_lo_sum_pic_label_reg): Remove
1484 write-only modifier from operand 1 constraint.
1486 Sat Aug 15 06:28:19 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
1488 * config/sparc/sparc.c (sparc_emit_set_const64_quick1): If
1489 emitting a XOR of -1 at the end, emit a NOT instead for combine's
1491 (sparc_emit_set_const64): Likewise, also when computing trailing
1492 bits do not negate low_bits and make fast_int an int.
1494 Fri Aug 14 21:07:03 1998 Jeffrey A Law (law@cygnus.com)
1496 * loop.c (add_label_notes): Do not ignore references to labels
1497 before dispatch tables. Mirrors Apr 8 change to mark_jump_label.
1498 * gcse.c (add_label_notes): Similarly.
1500 * pa.h (ASM_OUTPUT_MI_THUNK): Strip name encoding.
1502 * m68k.md (adddi_dilshr32): One of the operands must be a register.
1503 (adddi_dishl32): Similarly.
1505 Fri Aug 14 14:12:59 1998 Jason Merrill <jason@yorick.cygnus.com>
1507 * i386.h (MODES_TIEABLE_P): Reorganize to shut up warnings.
1508 * alias.c (memrefs_conflict_p): Add braces to shut up warnings.
1509 * cse.c (cse_basic_block): Add parens to shut up warnings.
1511 Fri Aug 14 12:58:21 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
1513 * config/sparc/sparc.c (sparc_emit_set_const64_quick2,
1514 sparc_emit_set_const64_longway, const64_is_2insns,
1515 create_simple_focus_bits, sparc_emit_set_const64): Fix more bugs
1516 in 64-bit constant formation.
1517 * config/sparc/sparc.md (snesi_zero_extend split): Generate
1518 rtl for addx not subx.
1519 (define_insn movdi_const64_special): Make available even when
1520 HOST_BITS_PER_WIDE_INT is not 64.
1521 (movdi_lo_sum_sp64_cint, movdi_high_sp64_cint): Remove.
1522 (losum_di_medlow, sethm, setlo): Make op2 symbolic_operand.
1523 (cmp_siqi_trunc_set, cmp_diqi_trunc_set): Encapsulate both
1524 instances of operand 1 inside a QI subreg.
1525 (xordi3_sp64_dbl): Remove '%' constraint for op1.
1526 (one_cmpldi2_sp64): Fix output string.
1527 (one_cmplsi2_not_liveg0): Rewrite to remove unneeded extra
1529 (unnamed arch64 ashift DI): Truncate shift count if greater than
1532 Fri Aug 14 21:52:53 1998 J"orn Rennecke <amylaar@cygnus.co.uk>
1534 * expr.c (store_expr): Don't optimize away load-store pair
1535 when either source or destination have a side effect.
1537 Fri Aug 14 16:50:10 1998 John Carr <jfc@mit.edu>
1539 * genrecog.c (add_to_sequence): Fatal error if the modes of the
1540 operands of SET are incompatible.
1542 * alpha.md: Fix max and min patterns so modes of SET operands match.
1544 Fri Aug 14 12:22:55 1998 Ian Lance Taylor <ian@cygnus.com>
1546 * configure.in: Avoid [[ by using test and changequote.
1547 * configure: Rebuild.
1549 Fri Aug 14 01:22:31 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
1551 * rtl.def (CONSTANT_P_RTX): Fix typo in string name.
1553 * config/sparc/sparc.md (seqdi_special_trunc, snedi_special_trunc,
1554 seqsi_special_extend, snesi_special_extend, snesi_zero_extend and
1555 split, snedi_zero_trunc and split, seqsi_zero_extend and split,
1556 seqdi_zero_trunc and split, pic_lo_sum_di, pic_sethi_di,
1557 movdi_cc_sp64_trunc, movdi_cc_reg_sp64_trunc, addx_extend_sp32 and
1558 split, addx_extend_sp64, subx_extend_sp64, subx_extend and split):
1559 Fix mismatching modes in SET operands.
1560 (conditional move patterns): Fix formatting.
1561 (unnamed subx arch64 pattern): Remove duplicate insn.
1563 Fri Aug 14 00:34:34 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
1565 * config/sparc/sparc.c (const64_operand, const64_high_operand):
1566 Get it right when HOST_BITS_PER_WIDE_INT is not 64.
1567 (input_operand): Fixup test for what we accept for constant
1569 (sparc_emit_set_const32, sparc_emit_set_symbolic_const64): Give
1571 (safe_constDI): Remove.
1572 (sparc_emit_set_safe_HIGH64, gen_safe_SET64, gen_safe_OR64,
1573 gen_safe_XOR64): New functions.
1574 (sparc_emit_set_const64_quick1, sparc_emit_set_const64_quick2,
1575 sparc_emit_set_const64_longway, sparc_emit_set_const64): Use
1577 * config/sparc/sparc.md (define_insn xordi3_sp64_dbl): Only make
1578 available when HOST_BITS_PER_WIDE_INT is not 64.
1579 (define_insn movdi_sp64_dbl, movdi_const64_special): Likewise and
1580 move before movdi_insn_sp64 pattern.
1581 (define_insn movdi_lo_sum_sp64_dbl, movdi_high_sp64_dbl): Remove.
1582 (define_insn sethi_di_medlow, seth44, setm44, sethh): Use
1583 symbolic_operand as predicate for second operand.
1584 (DImode minus split on arch32, negsi2 expander, one_cmplsi2
1585 expander): Give set VOIDmode.
1587 Fri Aug 14 01:45:06 1998 Mumit Khan <khan@xraylith.wisc.edu>
1589 * i386/cygwin32 (DEFAULT_PCC_STRUCT_RETURN): Define.
1591 Fri Aug 14 01:40:21 1998 Geoffrey Keating <geoffk@ozemail.com.au>
1593 * rs6000/linux.h (LINK_SPEC): Pass -G args to the linker.
1595 Fri Aug 14 01:23:23 1998 Richard Earnshaw (rearnsha@arm.com)
1597 * arm/netbsd.h (TARGET_DEFAULT): Default includes software floating
1599 (CPP_FLOAT_DEFAULT_SPEC): Re-define accordingly.
1601 Fri Aug 14 01:19:08 1998 Robert Lipe <robertl@dgii.com>
1603 * install.texi: Various SCO OpenServer tweaks.
1605 Thu Aug 13 20:14:40 1998 Jim Wilson <wilson@cygnus.com>
1607 * reload1.c (eliminate_regs_in_insn): Handle another case when
1608 eliminating the frame pointer to the hard frame pointer. Add
1609 missing ep->to_rtx check to one existing case.
1611 * mips/mips.md (movhi_internal2+2): Fix typo mem:SI -> mem:HI.
1613 Thu Aug 13 17:08:11 1998 Jason Merrill <jason@yorick.cygnus.com>
1615 * tree.h: De-conditionalize init_priority code.
1617 * mips.h (NM_FLAGS): Change from -Bp to -Bn.
1618 * collect2.c (NM_FLAGS): Change from -p to -n.
1620 * configure.in: Turn on collect2 for mipstx39-elf.
1621 Handle use_collect2=no properly.
1623 * c-common.c: De-conditionalize init_priority code.
1624 * collect2.c (extract_init_priority, sort_ids): New fns.
1625 (main): Call sort_ids.
1626 Move sequence_number to file scope.
1628 * configure.in: Handle --enable-init-priority.
1629 * c-common.c (attrs): Add A_INIT_PRIORITY.
1630 (init_attributes, decl_attributes): Likewise.
1631 * tree.h (DEFAULT_INIT_PRIORITY, MAX_INIT_PRIORITY): New macros.
1632 * tree.c (get_file_function_name_long): Split out...
1633 (get_file_function_name): ...from here.
1635 Thu Aug 13 16:09:53 1998 Martin von Loewis <loewis@informatik.hu-berlin.de>
1637 * expr.c (safe_from_p): Change code to ERROR_MARK only when not
1640 Thu Aug 13 15:24:48 1998 Jason Merrill <jason@yorick.cygnus.com>
1642 * toplev.c (display_help): Add braces to shut up warnings.
1643 * tree.c (simple_cst_equal): Likewise.
1645 * fold-const.c (non_lvalue): Don't deal with null pointer
1647 (fold, case COMPOUND_EXPR): Wrap a constant 0 in a NOP_EXPR.
1649 * c-typeck.c (initializer_constant_valid_p): Allow conversion of 0
1650 of any size to a pointer.
1652 Thu Aug 13 12:53:13 1998 Jim Wilson <wilson@cygnus.com>
1654 * i386/winnt.c (i386_pe_asm_file_end): Check TREE_SYMBOL_REFERENCED.
1656 Wed Aug 12 17:25:18 1998 Jeffrey A Law (law@cygnus.com)
1658 * mn10300.c (REG_SAVE_BYTES): Only reserve space for registers
1659 which will be saved.
1660 * mn10300.md (prologue insn): Only save registers which need saving.
1661 (epilogue insn): Similarly.
1663 * mn10300.c, mn10300.h, mn10300.md: Remove "global zero register"
1666 Wed Aug 12 12:39:16 1998 Gavin Romig-Koch <gavin@cygnus.com>
1668 * mips/mips.h (ENCODE_SECTION_INFO): Set SYMBOL_REF_FLAG for
1669 VAR_DECL's in gp addressable sections.
1671 Tue Aug 11 23:02:31 1998 John Carr <jfc@mit.edu>
1673 * sparc.c: Change return <exp> to <exp>; return; in functions
1675 * sparc.md: Add empty semicolon statement after final label in
1678 Tue Aug 11 22:42:01 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
1680 * config/sparc/sparc.md (define_insn addx_extend): Rename to
1681 addx_extend_sp64, only allow when TARGET_ARCH64.
1682 (define_insn addx_extend_sp32 and split): Version that works when
1684 (define_insn subx_extend): Likewise.
1685 (define_split adddi3 and subdi3 with zero extension): Fixup and
1686 correct bugs when not TARGET_ARCH64.
1688 Tue Aug 11 16:04:34 1998 John Carr <jfc@mit.edu>
1690 * except.c (set_exception_lang_code, set_exception_version_code):
1691 Use prototype-style definition if __STDC__, to match declaration
1694 * genemit.c: Change FAIL and DONE macros not to use loops.
1696 Tue Aug 11 12:27:03 1998 Jim Wilson <wilson@cygnus.com>
1698 * dwarf2out.c (ASM_OUTPUT_DWARF_ADDR_CONST): Use
1699 ASM_OUTPUT_DWARF2_ADDR_CONST if defined.
1701 * mips/mips.md (reload_outsi): Use M16_REG_P when TARGET_MIPS16.
1703 Tue Aug 11 18:12:53 1998 Dave Love <d.love@dl.ac.uk>
1705 * README.g77: Update from Craig.
1707 Tue Aug 11 04:46:01 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
1709 * config/sparc/sparc.c (sparc_emit_set_const32): INTVAL is of
1711 (safe_constDI sparc_emit_set_const64_quick1,
1712 sparc_emit_set_const64_quick2, sparc_emit_set_const64_longway,
1713 analyze_64bit_constant, const64_is_2insns,
1714 create_simple_focus_bits): Fix some bugs when compiled on real
1716 (function_arg_record_value_3, function_arg_record_value_2,
1717 function_arg_record_value): Add fully prototyped forward decls.
1718 * config/sparc/sparc.md (define_insn cmpsi_insn_sp32): Rename back
1719 to cmpsi_insn and use on both 64 and 32 bit targets.
1720 (define_insn cmpsi_insn_sp64): Remove.
1721 (define_expand zero_extendsidi2): Allow for 32-bit target too.
1722 (define_insn zero_extendsidi2_insn): Rename to
1723 zero_extendsidi2_insn_sp64.
1724 (define_insn zero_extendsidi2_insn_sp32): New pattern and
1725 assosciated forced split for it.
1727 * config/sparc/sparc.c (const64_operand, const64_high_operand):
1729 * config/sparc/sparc.h: Declare them.
1730 (PREDICATE_CODES): Add them.
1731 * config/sparc/sparc.md (movdi_lo_sum_sp64_dbl,
1732 movdi_high_sp64_dbl, xordi3_sp64_dbl): Use them.
1734 Mon Aug 10 22:57:24 1998 John Carr <jfc@mit.edu>
1736 * config/sparc/sparc.md (define_insn jump): Output ba,pt not b,pt
1737 in v9 case as the latter makes the Solaris assembler crash.
1739 Mon Aug 10 22:39:09 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
1741 * config/sparc/sparc.c (input_operand): Do not accept a LO_SUM MEM
1742 for TFmode when !v9. We require offsettable memory addresses.
1743 * config/sparc/sparc.h (ALTER_HARD_SUBREG): Handle TFmode to
1744 DFmode register number conversions.
1745 * config/sparc/sparc.md (define_split DFmode moves): If register
1746 is a SUBREG do alter_subreg on it before using.
1747 (define_expand movtf): Fixup comment about alignment on v9.
1748 (define_split TFmode moves): Don't use gen_{high,low}part, create
1749 explicit SUBREGs instead.
1751 Mon Aug 10 19:02:55 1998 John Carr <jfc@mit.edu>
1753 * Makefile.in (mbchar.o): Depend on mbchar.c.
1755 Mon Aug 10 04:28:13 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
1756 Richard Henderson <rth@cygnus.com>
1758 Rewrite Sparc backend for better code generation and
1759 improved sparc64 support.
1760 * config/sparc/sp64-elf.h: Set JUMP_TABLES_IN_TEXT_SECTION to
1762 * config/sparc/sysv4.h: Likewise.
1763 * config/sparc/sparc.c (v8plus_regcmp_p, sparc_operand,
1764 move_operand, v8plus_regcmp_op, emit_move_sequence,
1765 singlemove_string, doublemove_string, mem_aligned_8,
1766 output_move_double, output_move_quad, output_fp_move_double,
1767 move_quad_direction, output_fp_move_quad, output_scc_insn):
1769 (small_int_or_double): New predicate.
1770 (gen_compare_reg): Remove TARGET_V8PLUS cmpdi_v8plus emission.
1771 (legitimize_pic_address): Emit movsi_{high,lo_sum}_pic instead of
1772 old pic_{sethi,lo_sum}_si patterns.
1773 (mem_min_alignment): New generic function to replace
1774 mem_aligned_8, which uses REGNO_POINTER_ALIGN information when
1775 available and can test for arbitrary alignments. All callers
1777 (save_regs, restore_regs, build_big_number,
1778 output_function_prologue, output_cbranch, output_return,
1779 sparc_flat_save_restore, sparc_flat_output_function_prologue,
1780 sparc_flat_output_function_epilogue): Prettify
1782 (output_function_epilogue): Likewise and add code to output
1783 deferred case vectors.
1784 (output_v9branch): Likewise, add new arg INSN and use it to tack
1785 on branch prediction settings. All callers changed.
1786 (print_operand): Likewise and output %l44 for LO_SUMs when
1788 (sparc_splitdi_legitimate): New function to make sure DImode
1789 splits can be run properly when !arch64.
1790 (sparc_initialize_trampoline, sparc64_initialize_trampoline):
1791 Reformat example code in comments.
1792 (set_extends): Remove UNSPEC/v8plus_clear_high case.
1793 (sparc_addr_diff_list, sparc_addr_list): New statics to keep track
1794 of deferred case vectors we need to output.
1795 (sparc_defer_case_vector): Record a case vector.
1796 (sparc_output_addr_vec, sparc_output_addr_diff_vec,
1797 sparc_output_deferred_case_vectors): New functions to output them.
1798 (sparc_emit_set_const32): New function to form 32-bit constants in
1799 registers when that requires more than one instruction.
1800 (safe_constDI, sparc_emit_set_const64_quick1,
1801 sparc_emit_set_const64_quick2, sparc_emit_set_const64_longway,
1802 analyze_64bit_constant, const64_is_2insns,
1803 create_simple_focus_bits, sparc_emit_set_const64): New functions
1804 which do the same for 64-bit constants when arch64.
1805 (sparc_emit_set_symbolic_const64): New function to emit address
1806 loading for all code models on v9.
1807 * config/sparc/sparc.h (CONDITIONAL_REGISTER_USAGE): Do not make
1808 %g1 fixed when arch64, unfix %g0 when TARGET_LIVE_G0.
1809 (ALTER_HARD_SUBREG): Fix thinko, return REGNO + 1 not 1.
1810 (SECONDARY_INPUT_RELOAD_CLASS, SECONDARY_OUTPUT_RELOAD_CLASS): Fix
1811 inaccuracies in comments, add symbolic and text_segment operands
1812 when TARGET_CM_MEDANY and TARGET_CM_EMBMEDANY respectively. Use
1813 GENERAL_REGS in these cases as a temp REG is needed to load these
1814 addresses into a register properly.
1815 (EXTRA_CONSTRAINT): Document more accurately, remove Q case as it
1817 (GO_IF_LEGITIMATE_ADDRESS): Allow TFmode for LO_SUM on v9 since fp
1818 quads are guarenteed to have 16-byte alignment.
1819 (LEGITIMIZE_ADDRESS): For SYMBOL_REF, CONST, and LABEL_REF use
1820 copy_to_suggested_reg instead of explicit LO_SUM and HIGH.
1821 (ASM_OUTPUT_ADDR_VEC, ASM_OUTPUT_ADDR_DIFF_VEC): New macros for
1822 deferred case vector implementation.
1823 (ASM_OUTPUT_ADDR_VEC_ELT): Use fputc to output newline.
1824 (ASM_OUTPUT_ADDR_DIFF_ELT): Parenthesize LABEL in macro calls.
1825 Generate "internal label - label" instead of "label - 1b".
1826 (PRINT_OPERAND_ADDRESS): For LO_SUM use %l44 on TARGET_CM_MEDMID.
1827 (PREDICATE_CODES): Remove sparc_operand, move_operand,
1828 v8plus_regcmp_op. Add small_int_or_double, input_operand, and
1830 (doublemove_string, output_block_move, output_fp_move_double,
1831 output_fp_move_quad, output_move_double, output_move_quad,
1832 output_scc_insn, singlemove_string, mem_aligned_8, move_operand,
1833 sparc_operand, v8plus_regcmp_op, v8plus_regcmp_p): Remove externs.
1834 (sparc_emit_set_const32, sparc_emit_set_const64,
1835 sparc_emit_set_symbolic_const64, input_operand, zero_operand,
1836 mem_min_alignment, small_int_or_double): Add externs.
1837 * config/sparc/sparc.md: Document the many uses of UNSPEC and
1838 UNSPEC_VOLATILE in this backend.
1839 (define_function_unit ieu): Rename to ieu_unnamed. Add move and
1840 unary to types which execute in it.
1841 (define_function_unit ieu_shift): Rename to ieu0.
1842 (define_function_unit ieu1): New, executes compare, call, and
1843 uncond_branch type insns.
1844 (define_function_units for type fdivs, fdivd, fsqrt): These
1845 execute in the fpu multiply unit not the adder on UltraSparc.
1846 (define_expand cmpdi): Disallow TARGET_V8PLUS.
1847 (define_insn cmpsi_insn): Rename to cmpsi_insn_sp32.
1848 (define_insn cmpsi_insn_sp64): New, same as sp32 variant except it
1849 allows the arith_double_operand predicate and rHI constraint when
1851 (define_insn cmpdi_sp64, cmpsf_fpe, cmpdf_fpe, cmptf_fpe,
1852 cmpsf_fp, cmpdf_fp, cmptf_fp, sltu_insn, neg_sltu_insn,
1853 neg_sltu_minux_x, neg_sltu_plus_x, sgeu_insn, neg_sgeu_insn,
1854 sltu_plus_x, sltu_plus_x, sltu_plus_x_plus_y, x_minus_sltu,
1855 sgeu_plus_x, x_minus_sgeu, movqi_cc_sp64, movhi_cc_sp64,
1856 movsi_cc_sp64, movdi_cc_sp64, movsf_cc_sp64, movdf_cc_sp64,
1857 movtf_cc_sp64, movqi_cc_reg_sp64, movhi_cc_reg_sp64,
1858 movsi_cc_reg_sp64, movdi_cc_reg_sp64, movsf_cc_reg_sp64,
1859 movdf_cc_reg_sp64, movtf_cc_reg_sp64, zero_extendhisi2_insn,
1860 cmp_siqi_trunc, cmp_siqi_trunc_set, sign_extendhisi2_insn,
1861 sign_extendqihi2_insn, sign_extendqisi2_insn,
1862 sign_extendqidi2_insn, sign_extendhidi2_insn,
1863 extendsfdf2, extendsftf2, extenddftf2, truncdfsf2, trunctfsf2,
1864 trunctfdf2, floatsisf2, floatsidf2, floatsitf2, floatdisf2,
1865 floatdidf2, floatditf2, fix_truncsfsi2, fix_truncdfsi2,
1866 fix_trunctfsi2, fix_truncsfdi2, fix_truncdfdi2, fix_trunctfdi2,
1867 adddi3_sp64, addsi3, cmp_ccx_plus, cmp_cc_plus_set, subdi_sp64,
1868 subsi3, cmp_minus_ccx, cmp_minus_ccx_set, mulsi3, muldi3,
1869 muldi3_v8plus, cmp_mul_set, mulsidi3, mulsidi3_v8plus,
1870 const_mulsidi3_v8plus, mulsidi3_sp32, const_mulsidi3,
1871 smulsi3_highpart_v8plus, unnamed subreg mult,
1872 const_smulsi3_highpart_v8plus, smulsi3_highpart_sp32,
1873 const_smulsi3_highpart, umulsidi3_v8plus, umulsidi3_sp32,
1874 const_umulsidi3, const_umulsidi3_v8plus, umulsi3_highpart_v8plus,
1875 const_umulsi3_highpart_v8plus, umulsi3_highpart_sp32,
1876 const_umulsi3_highpart, divsi3, divdi3, cmp_sdiv_cc_set, udivsi3,
1877 udivdi3, cmp_udiv_cc_set, smacsi, smacdi, umacdi, anddi3_sp64,
1878 andsi3, and_not_di_sp64, and_not_si, iordi3_sp64, iorsi3,
1879 or_not_di_sp64, or_not_si, xordi3_sp64, xorsi3, xor_not_di_sp64,
1880 xor_not_si, cmp_cc_arith_op, cmp_ccx_arith_op,
1881 cmp_cc_arith_op_set, cmp_ccx_arith_op_set, cmp_ccx_xor_not,
1882 cmp_cc_xor_not_set, cmp_ccx_xor_not_set, cmp_cc_arith_op_not,
1883 cmp_ccx_arith_op_not, cmp_cc_arith_op_not_set,
1884 cmp_ccx_arith_op_not_set, negdi2_sp64, cmp_cc_neg, cmp_ccx_neg,
1885 cmp_cc_set_neg, cmp_ccx_set_neg, one_cmpldi2_sp64, cmp_cc_not,
1886 cmp_ccx_not, cmp_cc_set_not, cmp_ccx_set_not, addtf3, adddf3,
1887 addsf3, subtf3, subdf3, subsf3, multf3, muldf3, mulsf3,
1888 muldf3_extend, multf3_extend, divtf3, divdf3, divsf3, negtf2,
1889 negdf2, negsf2, abstf2, absdf2, abssf2, sqrttf2, sqrtdf2, sqrtsf2,
1890 ashlsi3, ashldi3, unnamed DI ashift, cmp_cc_ashift_1,
1891 cmp_cc_set_ashift_1, ashrsi3, ashrdi3, unnamed DI ashiftrt,
1892 ashrdi3_v8plus, lshrsi3, lshrdi3, unnamed DI lshiftrt,
1893 lshrdi3_v8plus, tablejump_sp32, tablejump_sp64, call_address_sp32,
1894 call_symbolic_sp32, call_address_sp64, call_symbolic_sp64,
1895 call_address_struct_value_sp32, call_symbolic_struct_value_sp32,
1896 call_address_untyped_struct_value_sp32,
1897 call_symbolic_untyped_struct_value_sp32, call_value_address_sp32,
1898 call_value_symbolic_sp32, call_value_address_sp64,
1899 call_value_symbolic_sp64, branch_sp32, branch_sp64,
1900 flush_register_windows, goto_handler_and_restore,
1901 goto_handler_and_restore_v9, goto_handler_and_restore_v9_sp64,
1902 flush, all ldd/std peepholes, return_qi, return_hi, return_si,
1903 return_addsi, return_di, return_adddi, return_sf, all call+jump
1904 peepholes, trap, unnamed trap insns): Prettify output strings.
1905 (define_insn anddi3_sp32, and_not_di_sp32, iordi3_sp32,
1906 or_not_di_sp32, xordi3_sp32, xor_not_di_sp32, one_cmpldi2):
1907 Likewise and force + implement splits for integer cases.
1908 (define_insn return_sf_no_fpu): Likewise and allow to match when
1909 no-fpu because of our subreg SFmode splits.
1910 (define_insn zero_extendqihi2, zero_extendqisi2_insn,
1911 zero_extendqidi2_insn, zero_extendhidi2_insn,
1912 zero_extendsidi2_insn, sign_extendsidi2_insn): Likewise and use
1913 input_operand for second operand.
1914 (cmp_minus_cc, cmp_minus_cc_set): Likewise and use
1915 reg_or_0_operand for operand 2 so new splits can use it.
1916 (cmp_zero_extendqisi2, cmp_zero_extendqisi2_set, cmp_cc_plus,
1917 cmp_cc_xor_not): Likewise and don't forget to check TARGET_LIVE_G0
1919 (cmp_zero_extract, cmp_zero_extract_sp64): Likewise and allow
1920 CONST_DOUBLEs for operand 2.
1921 (define_insn move_label_di): Likewise and label distance
1922 optimization because it no longer works with new deferred case
1923 vector scheme. To be revisited.
1924 (define_insn x_minus_y_minus_sltu, x_minus_sltu_plus_y): Likewise
1925 and allow reg_or_0_operand and J constraint for second operand.
1926 (define_insn jump): Set branch predict taken on V9.
1927 (define_insn tablejump): Emit LABEL_REF + PLUS memory address for
1928 new deferred case vector scheme.
1929 (define_insn pic_tablejump_32, pic_tablejump_64): Remove.
1930 (define_insn negdi2_sp32): Force + implement splits.
1931 (define_insn negsi2, one_cmplsi2): Rename to negsi2_not_liveg0 and
1932 one_cmplsi2_not_liveg0 respectively, and create expander of original
1933 names which emit special rtl for TARGET_LIVE_G0.
1934 (define_insn cmpdi_v8plus, scc_si, scc_di): Remove.
1935 (define_insn seq, sne, slt, sge, sle, sltu, sgeu): Don't do
1936 gen_compare_reg, FAIL instead.
1937 (define_insn sgtu, sleu): Likewise and check gen_s*() return
1938 values when trying to reverse condition codes, if they FAIL then
1940 (define_insn snesi_zero, neg_snesi_zero, snesi_zero_extend,
1941 snedi_zero, neg_snedi_zero, snedi_zero_trunc, seqsi_zero,
1942 neg_seqsi_zero, seqsi_zero_extend, seqdi_zero, neg_seqdi_zero,
1943 seqdi_zero_trunc, x_plus_i_ne_0, x_minus_i_ne_0, x_plus_i_eq_0,
1944 x_minus_i_eq_0): Add new splits to perform these multi-insn cases,
1945 set output string to # to indicate they are mandatory splits.
1946 (define_insn pic_lo_sum_si, pic_sethi_si, pic_lo_sum_di,
1947 pic_sethi_di, move_pic_label_si): Remove.
1948 (define_insn movsi_low_sum, movsi_high, movsi_lo_sum_pic,
1949 movsi_high_pic, movsi_pic_label_reg): New patterns to take their
1951 (define_expand movsi_pic_label_ref, define_insn
1952 movsi_high_pic_label_ref, movsi_lo_sum_pic_label_ref): New
1953 expander and insns to handle PIC label references and deferred
1955 (define_insn get_pc_via_rdpc): Comment out as it is no longer
1957 (define_expand movqi, movhi, movsi, movdi, movsf, movdf, movtf):
1958 Rewrite to not use emit_move_sequence, make use of new constant
1959 formation code, and new splits for all multi-insn cases.
1960 (define_insn movqi_insn): Remove sethi case, it can never happen.
1961 Use reg_or_zero_operand instead of const0_rtx explicit test,
1962 use input_operand instead of move_operand for source, and use
1963 general_operand now for dest.
1964 (define_insn movhi_insn): Similar but leave sethi case.
1965 (define_insn lo_sum_qi, store_qi, store_hi): Remove.
1966 (define_insn sethi_hi lo_sum_hi): Rename to movhi_high and
1967 movhi_lo_sum respectively, prettify output string.
1968 (define_insn movsi_zero_liveg0): New pattern to put zero into a
1969 register when needed on TARGET_LIVE_G0.
1970 (define_insn movsi_insn): Use general_operand and input_operand
1971 for dest and src respectively. Simplify applicability test.
1972 Prettify output strings, and add clr alternative for J
1974 (define_insn movdi_sp32_v9, movdi_sp32, define_splits for
1975 deprecated std and reg-reg DI moves): Remove and...
1976 (define_insn movdi_insn_sp32, movdi_insn_sp64): Replace with new
1977 implementation which uses forced splits for all non-single insn
1979 (define_split DI move cases on !arch64): New splits to handle all
1980 situations of 64-bit double register DImode on 32bit, and
1981 unaligned registers and memory addresses for all subtargets.
1982 (define_insn movsf_const_insn, movdf_const_insn, store_sf):
1984 (define_insn movsf_insn, movsf_no_f_insn): Use general_operand and
1985 input_operand for dest and src respectively, prettify output
1987 (define_insn movdf_insn, movdf_no_e_insn, store_df,
1988 movtf_const_insn, movtf_insn, movtf_no_e_insn, store_tf): Remove
1990 (define_insn movdf_insn_sp32, movdf_no_e_insn_sp32,
1991 movdf_insn_sp64, movdf_no_e_insn_sp64, movtf_insn,
1992 movtf_no_e_insn_sp32, movtf_insn_hq_sp64, movtf_insn_sp64,
1993 movtf_no_e_insn_sp64) Replace with new
1994 implementation which uses forced splits for all non-single insn
1996 (define_split DF move cases): New splits in similar vein to DI
1998 (define_insn sethi_di_medlow, sethi_di_medium_pic,
1999 sethi_di_embmedany_data, sethi_di_embmedany_text, sethi_di_sp64,
2000 movdi_sp64_insn): Remove old v9 code model and constant loading
2002 (define_insn pic_lo_sum_di, pic_sethi_di,
2003 sethi_di_medlow_embmedany_pic, sethi_di_medlow, losum_di_medlow,
2004 seth44, setm44, setl44, sethh, setlm, sethm, setlo,
2005 embmedany_sethi, embmedany_losum, embmedany_brsum,
2006 embmedany_textuhi, embmedany_texthi, embmedany_textulo,
2007 embmedany_textlo, movdi_lo_sum_sp64_cint, movdi_lo_sum_sp64_dbl,
2008 movdi_high_sp64_cint, movdi_high_sp64_dbl): Replace with new
2009 scheme, using unspecs, secondary reloads, and one to one sparc
2010 insn to rtl insn mapping for better scheduling and code gen.
2011 (define_expand reload_indi, reload_outdi): Reload helpers for
2012 MEDANY and EMBMEDANY symbol address loading cases which require a
2014 (define_expand movsicc): Remove v8plus_regcmp cases.
2015 (define_insn movdi_cc_sp64_trunc, movdi_cc_reg_sp64_trunc,
2016 cmp_zero_extendqidi2, cmp_zero_extendqidi2_set, cmp_qidi_trunc,
2017 cmp_diqi_trunc_set): New patterns used by some of the new scc
2019 (define_insn xordi3_sp64_dbl): New pattern used for constant
2020 formation when crossing from 32-bit targets.
2021 (define_insn movsi_cc_reg_v8plus, v8plus_clear_high, and helper
2023 (define_insn addx, subx): Make visible and prettify.
2024 (define_insn adddi3_insn_sp32): Likewise and force split.
2025 (define_insn addx_extend, subx_extend, unnamed): New patterns for
2026 64bit scc split usage.
2027 (define_insn unnamed plusDI zero_extend, unnamed minusDI
2028 zero_extend, subdi3): Force and implement splits.
2030 * final.c (final_scan_insn): Don't output labels if target
2031 specifies ASM_OUTPUT_ADDR_{DIFF}_VEC. Do these macro operations
2034 * reorg.c (dbr_schedule): When taking on BR_PRED notes at the end,
2035 don't forget to walk inside SEQUENCESs too as these are what the
2036 delay slot scheduler will create.
2038 Mon Aug 10 01:21:01 1998 Richard Henderson <rth@cygnus.com>
2040 * alpha.md (extxl+1,+2): New patterns to work around
2043 Sat Aug 8 19:20:22 1998 Gary Thomas (gdt@linuxppc.org)
2045 * rs6000.c (rs6000_allocate_stack_space) Fix typo which
2046 caused bad assembly code to be generated.
2048 Sat Aug 8 18:53:28 1998 Jeffrey A Law (law@cygnus.com)
2050 * netbsd.h: Fix typo.
2052 Mon Aug 3 00:06:42 1998 Robert Lipe <robertl@dgii.com>
2054 * config.sub: Fix typo.
2056 Sun Aug 2 22:39:08 1998 Hans-Peter Nilsson <hp@axis.se>
2058 * invoke.texi (Environment Variables): Typo: Change "ascpects"
2060 (Running Protoize): Typo: Change "ther" into "other".
2062 Sun Aug 2 00:42:50 1998 Jeffrey A Law (law@cygnus.com)
2064 * i386/netbsd.h: Undo previous change to DWARF2_UNWIND_INFO.
2065 * m68k/netbsd.h: Likewise.
2066 * ns32k/netbsd.h: Likewise.
2067 * sparc/netbsd.h: Likewise.
2069 Sat Aug 1 17:59:30 1998 Richard Henderson <rth@cygnus.com>
2071 * ginclude/va-alpha.h (va_list): Use a typedef, not a define.
2072 * ginclude/va-clipper.h (va_list): Likewise.
2074 Fri Jul 31 20:22:02 1998 Michael Meissner <meissner@cygnus.com>
2076 * rs6000.c (rs6000_override_options): If big endian and -Os, use
2077 load/store multiple instructions unless user overrides.
2079 Fri Jul 31 17:08:59 1998 Jeffrey A Law (law@cygnus.com)
2081 * ns32k/netbsd.h: Fix typo.
2083 Fri Jul 31 10:23:55 1998 Doug Evans <devans@canuck.cygnus.com>
2085 * m32r/m32r.h (ASM_OUTPUT_SOURCE_LINE): Always output line number
2086 labels with .debugsym if no parallel insns.
2088 Thu Jul 30 19:15:53 1998 Richard Henderson <rth@cygnus.com>
2090 * alpha.md (fp cmp): Replicate patterns for ALPHA_TP_INSN.
2091 (fcmov): Remove ALPHA_TP_INSN patterns -- fcmov doesn't trap.
2093 Thu Jul 30 19:50:15 1998 David Edelsohn <edelsohn@mhpcc.edu>
2095 * rs6000/x-aix43 (AR_FOR_TARGET_FLAGS): Delete.
2096 (AR_FOR_TARGET): Define.
2098 Thu Jul 30 12:29:12 1998 Mark Mitchell <mark@markmitchell.com>
2100 * dyn-string.h: New file.
2101 * dyn-string.c: Likewise.
2102 * Makefile.in (OBJS): Add dyn-string.o.
2103 (dwarf2out.o): Add dyn-string.h dependency.
2104 (dyn-string.o): List dependencies.
2105 * dwarf2out.c: Include dyn-string.h.
2106 (ASM_NAME_TO_STRING): Use dyn_string_append, rather than strcpy.
2107 (addr_const_to_string): Take a dyn_string_t, not a char* as a
2108 prototype. Use dyn_string_append rather than strcat, throughout.
2109 (addr_to_string): Use dyn_string_t.
2111 Thu Jul 30 13:08:07 1998 Ken Raeburn <raeburn@cygnus.com>
2113 Function entry/exit profiling instrumentation:
2114 * expr.h (profile_function_entry_libfunc,
2115 profile_function_exit_libfunc): Declare new variables.
2116 * optabs.c: Define them here.
2117 (init_optabs): Initialize them.
2118 * tree.h (struct tree_decl): New flag
2119 no_instrument_function_entry_exit.
2120 (DECL_NO_INSTRUMENT_FUNCTION_ENTRY_EXIT): New accessor macro.
2121 * c-decl.c (duplicate_decls): Merge it.
2122 * c-common.c (enum attrs): New value A_NO_INSTRUMENT_FUNCTION.
2123 (init_attributes): Use it for "no_instrument_function".
2124 (decl_attributes): Handle it, for functions that have not yet been
2125 compiled. Set decl flag.
2126 * flags.h (flag_instrument_function_entry_exit): Declare new
2128 * toplev.c (flag_instrument_function_entry_exit): Define it here.
2129 (f_options): New option "instrument-functions".
2130 * function.h (struct function): New field instrument_entry_exit.
2131 * function.c (current_function_instrument_entry_exit): New
2133 (push_function_context_to, pop_function_context_from): Save and
2135 (expand_function_start): Set current_ variable, maybe emit return
2136 label and entry profile call.
2137 (expand_function_end): Maybe emit exit profile call.
2139 Thu Jul 30 00:58:34 1998 Jeffrey A Law (law@cygnus.com)
2141 * i386.md (movqi): When optimizing a load of (const_int 1) into a
2142 NON_QI_REG_P, pretend the register is SImode.
2144 Wed Jul 29 23:49:23 1998 Todd Vierling <tv@netbsd.org>
2146 * configure.in: Use xm-netbsd.h as the NetBSD xm file (not xm-siglist).
2147 Accept arm32 as arm, m68k4k as m68k, mipsle as mips-dec, and any
2148 manufacturer id for ns32k.
2149 * configure: Regenerated.
2150 * config/netbsd.h: When using ASM_WEAKEN_LABEL, make it global too.
2151 * config/t-netbsd: Don't compile libgcc1-test as the fns are in libc.
2152 * config/i386/netbsd.h: Undefine DWARF2_UNWIND_INFO, not define as 0.
2153 * config/m68k/netbsd.h: Same.
2154 * config/ns32k/netbsd.h: Same.
2155 * config/sparc/netbsd.h: Same.
2157 Wed Jul 29 22:39:21 1998 Jeffrey A Law (law@cygnus.com)
2159 * unroll.c (unroll_loop): Do not abort for an UNROLL_MODULO
2160 or UNROLL_COMPLETELY loop that starts with a jump to its
2163 Wed Jul 29 22:18:14 1998 David Edelsohn <edelsohn@mhpcc.edu>
2165 * rs6000/rs6000.md (absdi2 define_split): Swap operands of MINUS.
2166 * rs6000/rs6000.c (mask64_operand): Use HOST_BITS_PER_WIDE_INT.
2167 (print_operand, case 'B'): Don't fall through.
2168 (print_operand, case 'S'): Correct mask begin/end computation.
2169 Use HOST_BITS_PER_WIDE_INT.
2170 * rs6000/rs6000.h (CPP_PREDEFINES): Define _LONG_LONG.
2171 (CONDITIONAL_REGISTER_USAGE): GPR13 fixed if TARGET_64BIT.
2172 * rs6000/aix41.h (CPP_PREDEFINES): Same.
2173 * rs6000/aix43.h (CPP_PREDEFINES): Same.
2175 Wed Jul 29 11:47:10 1998 Nick Clifton <nickc@cygnus.com>
2177 * config/arm/thumb.md (extendqisi2_insn): Remove earlyclobber
2178 constraint from second alternative.
2180 Tue Jul 28 23:29:04 1998 Jason Merrill <jason@yorick.cygnus.com>
2182 * configure.in: Fix --without/--disable cases for local-prefix,
2183 gxx-include-dir and checking.
2185 Tue Jul 28 22:01:23 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
2187 * configure.in (enable_haifa): Set by default for sparc64 too.
2190 Tue Jul 28 23:29:04 1998 Jason Merrill <jason@yorick.cygnus.com>
2192 * i386/cygwin32.h (VALID_MACHINE_TYPE_ATTRIBUTE): New macro.
2193 * i386/winnt.c (associated_type): New fn.
2194 (i386_pe_valid_type_attribute_p): New fn.
2195 (i386_pe_check_vtable_importexport): Remove.
2196 (i386_pe_dllexport_p): Use associated_type.
2197 (i386_pe_dllimport_p): Likewise.
2199 From Antonio M. O. Neto <anmendes@cruzeironet.com.br>:
2200 * i386.c (i386_valid_type_attribute_p): Also accept
2201 attributes for METHOD_TYPEs.
2203 Tue Jul 28 23:17:39 1998 Peter Gerwinski <peter@gerwinski.de>
2205 * tree.c (build_range_type): Copy TYPE_SIZE_UNIT.
2207 Tue Jul 28 22:31:12 1998 Craig Burley <burley@gnu.org>
2209 * gcc.c: Fix commentary describing %g, %u, %U, and %O.
2211 * gcc.c (do_spec_1): Fix handling of %g%O and %U%O to prevent
2212 them from generating a new base name for each occurence of
2215 1998-07-28 Vladimir N. Makarov <vmakarov@cygnus.com>
2217 * cse.c (cse_insn): Enable subsitution inside libcall only for REG,
2219 * rtlanal.c (replace_rtx): Prohibit replaces in CONST_DOUBLE.
2223 * cplus-dem.c (type_kind_t): New type.
2224 (demangle_template_value_parm): Add type_kind_t parameter. Rely
2225 on this paramter, rather than demangling the type again.
2226 (demangle_integral_value): Pass tk_integral.
2227 (demangle_template_: Pass the value returned from do_type.
2228 (do_type): Return a type_kind_t. Pass tk_integral to
2229 demangle_template_value_parm for array bounds.
2230 (demangle_fund_type): Likewise.
2232 Mon Jul 27 00:54:41 1998 Jason Merrill <jason@yorick.cygnus.com>
2234 * tree.c (simple_cst_equal, case CONSTRUCTOR): OK if the elts are
2237 Mon Jul 27 22:18:36 1998 Jeffrey A Law (law@cygnus.com)
2239 * pa.c (move_operand): Accept CONSTANT_P_RTX.
2241 Mon Jul 27 17:18:52 1998 Dave Brolley <brolley@cygnus.com>
2243 * stor-layout.c (layout_type): Handle arrays of bits, for Chill.
2245 * expr.c (get_inner_reference): Handle zero-based, unsigned, array
2248 Mon Jul 27 14:51:33 1998 Jeffrey A Law (law@cygnus.com)
2250 * mn10300.h (DEBUGGER_AUTO_OFFSET): Define.
2251 (DEBUGGER_ARG_OFFSET): Likewise.
2253 * mn10300.md (movsf): Remove last change. Not needed.
2255 Mon Jul 27 14:22:36 1998 Dave Brolley <brolley@cygnus.com>
2257 * c-lex.c (yylex): Fix boundary conditions in character literal and
2258 string literal loops.
2260 Mon Jul 27 11:43:54 1998 Stan Cox <scox@cygnus.com>
2262 * longlong.h (count_leading_zeros): Sparclite scan instruction was
2263 being invoked incorrectly.
2265 * i386.c (ix86_prologue): Added SUBTARGET_PROLOGUE invocation.
2266 * i386/cygwin32.h (STARTFILE_SPEC, LIB_SPEC, SUBTARGET_PROLOGUE):
2268 * i386/win32.h: New file. Hybrid mingw32.h/cygwin32.h configuration.
2269 * configure.in: Added i[34567]86-*-win32
2270 * config.sub: Ditto.
2271 * configure: Rebuilt.
2273 Sun Jul 26 01:11:12 1998 H.J. Lu (hjl@gnu.org)
2275 * i386.h (CONST_DOUBLE_OK_FOR_LETTER_P): Return 0 when eliminating
2276 the frame pointer and compiling PIC code and reload has not completed.
2278 * i386.c (output_to_reg): Add code to emulate non-popping DImode
2281 Sun Jul 26 01:01:32 1998 Jeffrey A Law (law@cygnus.com)
2283 * regmove.c (regmove_optimize): Fix typo initializing regmove_bb_head.
2285 Sat Jul 25 23:29:23 1998 Gerald Pfeifer <pfeifer@dbai.tuwien.ac.at>
2287 * Makefile.in (install-info): Only try to update the info
2288 directory file if it exists in the first place.
2290 Fri Jul 24 18:58:37 1998 Klaus Espenlaub <kespenla@student.informatik.uni-ulm.de>
2292 * rs6000.h (ASM_OUTPUT_CONSTRUCTOR, ASM_OUTPUT_DESTRUCTOR): Delete.
2294 Fri Jul 24 14:20:26 1998 Jeffrey A Law (law@cygnus.com)
2296 * mn10300.md (movqi, movhi, movsi, movsf): Correctly handle
2297 CONST_DOUBLE source.
2299 Fri Jul 24 11:17:04 1998 Nick Clifton <nickc@cygnus.com>
2301 * config/arm/thumb.c (thumb_print_operand): Decode %_ in asm
2302 strings as the insertion of USER_LABEL_PREFIX.
2303 * config/arm/thumb.h (PRINT_OPERAND_PUNCT_VALID_P): Accept _ as a
2305 * config/arm/thumb.md: Use %_ as a prefix to gcc library function
2308 Thu Jul 23 18:53:20 1998 Jim Wilson <wilson@cygnus.com>
2310 * dbxout.c (dbxout_range_type): Only call dbxout_type_index for
2311 already defined type.
2313 Thu Jul 23 13:49:41 1998 Jeffrey A Law (law@cygnus.com)
2315 * expr.c (check_max_integer_computation_mode): Allow conversions
2316 of constant integers to MAX_INTEGER_COMPUTATION_MODE.
2317 (expand_expr): Likewise.
2319 Thu Jul 23 11:12:06 1998 Alexandre Petit-Bianco <apbianco@cygnus.com>
2321 * expr.c (expand_expr): Expand RETURN_EXPR.
2323 Thu Jul 23 11:00:29 1998 Jim Wilson <wilson@cygnus.com>
2325 * dwarf2out.c (dwarf2out_finish): Call stripattributes on TEXT_SECTION.
2327 Wed Jul 22 19:10:00 1998 Catherine Moore <clm@cygnus.com>
2329 * dwarf2out.c (output_aranges): Call stripattributes
2330 for TEXT_SECTION references.
2331 (output_line_info): Likewise.
2333 Wed Jul 22 14:08:54 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
2335 * profile.c (branch_prob): Call allocate_reg_info after outputting
2336 profile rtl in instrument_arcs.
2338 Wed Jul 22 12:47:49 1998 Jim Wilson <wilson@cygnus.com>
2340 * fixinc.irix (math.h): Install wrapper instead of copying.
2342 Wed Jul 22 12:37:14 1998 Alexandre Petit-Bianco <apbianco@cygnus.com>
2344 * tree.def (EXPR_WITH_FILE_LOCATION): Defined as an 'e' expression
2345 so WFL are expanded correctly when contained in a COMPOUND_EXPR.
2346 * tree.h (EXPR_WFL_EMIT_LINE_NOTE): Change macro not to use
2347 lang_flag_0. Added documentation in the flag table.
2349 Tue Jul 21 23:28:35 1998 Klaus Kaempf <kkaempf@rmi.de>
2351 * cccp.c (do_include): Fix vax c style include handling.
2353 Tue Jul 21 13:28:19 1998 Jason Merrill <jason@yorick.cygnus.com>
2355 * cplus-dem.c (do_type): Use demangle_template_value_parm for arrays.
2357 Sun Jul 12 01:27:05 1998 Jason Merrill <jason@yorick.cygnus.com>
2359 * fold-const.c (non_lvalue): Don't deal with null pointer
2361 (fold, case COMPOUND_EXPR): Wrap a constant 0 in a NOP_EXPR.
2363 Tue Jul 21 15:49:31 1998 David Edelsohn <edelsohn@mhpcc.edu>
2365 * rs6000.h (PREDICATE_CODES): Add CONSTANT_P_RTX.
2366 * rs6000.md (movsi, movdi): Add CONSTANT_P_RTX.
2367 * rs6000.c (short_cint_operand): Add CONSTANT_P_RTX.
2368 (u_short_cint_operand): Same.
2369 (reg_or_cint_operand): Same.
2370 (logical_operand): Same.
2371 (input_operand): Same.
2372 (reg_or_short_operand): Use u_short_cint_operand.
2374 Tue Jul 21 08:56:42 1998 Richard Henderson <rth@cygnus.com>
2376 * alpha.md (fix_truncdfsi2, fix_truncsfsi2): Remove the define_expands,
2377 but keep the insns and splits. Adjust so when the ultimate destination
2378 is memory, use cvtql.
2380 Tue Jul 21 08:55:09 1998 Richard Henderson <rth@cygnus.com>
2382 * flow.c (regno_uninitialized): Fixed regs are never uninitialized.
2384 Tue Jul 21 00:31:01 1998 Jeffrey A Law (law@cygnus.com)
2386 * gcc.c (do_spec): Call "error" not "warning".
2388 * configure.in: Fix minor problems with gas feature detection code.
2389 * configure: Rebuilt.
2391 * gcc.c (do_spec): Issue a warning for '%[]' usage.
2394 * gcc.c: Delete %[spec] support.
2395 (do_spec_1, case '('): Likewise.
2396 (do_spec_1, case '['): Call error.
2398 Mon Jul 20 22:34:17 1998 Richard Henderson <rth@cygnus.com>
2400 * alpha.h (CPP_SPEC): Tidy. Hook to cpp_cpu and cpp_subtarget.
2401 (CPP_SUBTARGET_SPEC): Default to empty string.
2402 (CPP_AM_*, CPP_IM_*, CPP_CPU_*, CPP_CPU_SPEC): New.
2403 (EXTRA_SPECS, SUBTARGET_EXTRA_SPECS): New.
2404 * alpha/elf.h (LD_SPEC): Use %(elf_dynamic_linker).
2405 * alpha/linux-elf.h (SUBTARGET_EXTRA_SPECS): New.
2407 * alpha/linux.h (CPP_PREDEFINES): Tidy.
2408 * alpha/netbsd-elf.h (SUBTARGET_EXTRA_SPECS): New.
2409 * alpha/netbsd.h (CPP_PREDEFINES): Tidy.
2410 * alpha/osf.h (CPP_PREDEFINES): Remove bits subsumed by CPP_CPU_SPEC.
2411 * alpha/win-nt.h (CPP_PREDEFINES): Likewise.
2412 * alpha/vsf.h (CPP_PREDEFINES): Likewise.
2413 (CPP_SUBTARGET_SPEC): New. Do this instead of overriding CPP_SPEC.
2414 * alpha/vxworks.h: Likewise.
2416 Mon Jul 20 22:51:57 1998 Ken Raeburn <raeburn@cygnus.com>
2418 * mips.md (reload_outsi): Added missing REGNO call.
2419 (smulsi3_highpart, umulsi3_highpart): Provide prototype for
2421 (mul_acc_di, mul_acc_64bit_di): Don't use match_op_dup, use
2422 another match_operator and compare the codes.
2424 * mips.h (MASK_DEBUG_E, MASK_DEBUG_I): Set to zero.
2426 * MIPS multiply pattern fixes:
2427 * mips.h (enum reg_class, REG_CLASS_NAMES, REG_CLASS_CONTENTS):
2428 Add union classes for HI, LO, or HILO plus general registers.
2429 (GENERATE_MADD): Deleted.
2430 * mips.md (mulsi3_mult3): Don't disparage output-LO alternative.
2431 Add TARGET_MAD to condition.
2432 (mulsi3): Test HAVE_mulsi3_mult3, not specific flags.
2433 (mul_acc_si): Expand GENERATE_MADD here; it's the only use. Use
2434 "*d" for accumulator, to give preference to LO initially but not
2437 Mon Jul 20 16:16:38 1998 Dave Brolley <brolley@cygnus.com>
2439 * configure.in (enable_c_mbchar): New configure option.
2440 (extra_cpp_objs): Always available now.
2442 * cexp.y (mbchar.h): #include it.
2443 (yylex): Handle Multibyte characters in character literals.
2445 * cccp.c (mbchar.h): #include it.
2446 (main): Set character set based on LANG environment variable.
2447 (rescan): Handle multibyte characters in comments.
2448 (skip_if_group): See above.
2449 (validate_else): See above.
2450 (skip_to_end_of_comment): See above.
2451 (macarg1): See above.
2452 (discard_comments): See above.
2453 (rescan): Handle multibyte characters in string and character literals.
2454 (collect_expansion): See above.
2455 (skip_quoted_string): See above.
2456 (macroexpand): See above.
2457 (macarg1): See above.
2458 (discard_comments): See above.
2459 (change_newlines): See above.
2461 * c-lex.c (mbchar.h): #include it.
2462 (GET_ENVIRONMENT): New macro.
2463 (init_lex): Set character set based on LANG environment variable.
2464 (yylex): Handle multibyte characters in character literals.
2465 (yylex): Handle multibyte characters in string literals.
2467 * Makefile.in (mbchar.o): New target.
2468 (cccp$(exeext)): @extra_cpp_objs@ is always available.
2469 (cppmain$(exeext)): @extra_cpp_objs@ is always available.
2471 * mbchar.[ch]: New files for multibyte character handling.
2473 Mon Jul 20 01:11:11 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
2475 * jump.c (jump_optimize): When simplifying noop moves and
2476 PUSH_ROUNDING, fix thinko so we use same criterion for identifying
2477 the PUSHes to rewrite in second loop as we did in the first.
2479 Sun Jul 19 08:23:53 1998 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
2481 * cplus-dem.c (demangle_nested_args): Make function definition
2482 static to match the prototype.
2484 Fri Jul 17 14:58:44 1998 Richard Henderson <rth@cygnus.com>
2486 * alloca.c: Respect USE_C_ALLOCA.
2487 * gencheck.c (xmalloc): Ignore __GNUC__ for definition.
2488 * gengenrtl.c (xmalloc): Likewise.
2490 Fri Jul 17 14:18:14 1998 Richard Henderson <rth@cygnus.com>
2492 * loop.h (struct induction): Add no_const_addval.
2493 * loop.c (the_movables, reg_address_cost): New variables.
2494 (init_loop): Init reg_address_cost.
2495 (loop_optimize): Call end_alias_analysis.
2496 (scan_loop): Init the_movables.
2497 (record_giv): Init induction->no_const_addval.
2498 (basic_induction_var) [PLUS]: Use rtx_equal_p instead of ==.
2499 [REG]: Rearrange loop search test to catch more cases.
2500 (general_induction_var): Return success not benefit; take an extra
2501 argument for that. Change all callers.
2502 (simplify_giv_expr) [PLUS]: Always combine invariants. Use sge_plus.
2503 [MULT]: Use rtx_equal_p instead of ==. Combine simple invariants.
2504 [default]: Search the_movables for additional combinations.
2505 (sge_plus_constant, sge_plus): New functions.
2506 (express_from_1): New function.
2507 (express_from): Always define. Rewrite using express_from_1.
2508 (combine_givs_p): Handle more cases. Ignore address cost.
2509 (cmp_combine_givs_stats): New function.
2510 (combine_givs_used_once, combine_givs_benefit_from): New functions.
2511 (combine_givs): Rewrite to do best-fit combination.
2513 * fold-const.c (operand_equal_p): Handle RTL_EXPR.
2514 (fold): Do a complete (A*C)+(B*C) association check.
2516 Fri Jul 17 11:21:55 1998 Jim Wilson <wilson@cygnus.com>
2518 * function.c (fixup_var_refs_insns): Handle CLOBBER of a CONCAT.
2520 Fri Jul 17 11:48:55 1998 Jeffrey A Law (law@cygnus.com)
2522 * mn10300.c (MODES_TIEABLE_P): Fix typo.
2524 Fri Jul 17 03:26:12 1998 Rihcard Earnshaw (rearnsha@arm.com)
2526 * tree.c (valid_machine_attribute): Only create a new type variant if
2527 there is a decl to use it.
2529 Thu Jul 16 14:48:04 1998 Nick Clifton <nickc@cygnus.com>
2531 * gcc.c (do_spec_1): Cope with %g/%u/%U options which do not have
2534 Fri Jul 17 03:24:40 1998 Hans-Peter Nilsson <hp@axis.se>
2536 * extend.texi (Explicit Reg Vars): Typo: change "may deleted" into "may
2539 Thu Jul 16 14:48:47 1998 Jeffrey A Law (law@cygnus.com)
2541 * mn10300.c (count_tst_insns): New arg oreg_countp. Callers changed.
2542 Simplify tests for clearing an address register.
2543 (expand_prologue): Corresponding changes.
2545 * mn10300.md (movXX patterns): Make sure the destination is an
2546 ADDRESS_REG when substituting "zero_areg" for (const_int 0).
2547 (logical patterns): Split into expanders + patterns
2548 (zero and sign extension patterns): Similarly.
2549 (shift patterns): Similarly.
2551 Thu Jul 16 01:17:44 1998 Richard Henderson <rth@cygnus.com>
2553 * loop.c (emit_iv_add_mult): Scan the entire insn list generated
2554 for the sequence, recording base values.
2556 Wed Jul 15 10:49:55 1998 Richard Henderson <rth@cygnus.com>
2558 * i386.h (CPP_CPU_SPEC): Remove -Asystem(unix).
2560 Tue Jul 14 14:15:30 1998 Nick Clifton <nickc@cygnus.com>
2562 * gcc.c: Remove ANSI-C ism from --help code.
2564 * toplev.c: Support --help with USE_CPPLIB.
2566 Tue Jul 14 14:46:08 1998 Jeffrey A Law (law@cygnus.com)
2568 * configure.in: Rework gas feature code to work with symlink based
2571 * extend.texi: Clarify some issues related to local variables
2572 assigned to explicit registers.
2574 * mn10300.md (mulsi): Turn into expander + pattern.
2576 * mn10300.md (movsi, movsf, movdi, movdf): Remove "x" from I -> a
2579 Tue Jul 14 07:41:59 1998 Richard Earnshaw (rearnsha@arm.com)
2581 * arm/tcoff.h (USER_LABEL_PREFIX): Make it empty to match coff.h.
2583 Tue Jul 14 03:02:44 1998 Jeffrey A Law (law@cygnus.com)
2585 * version.c: Bump again to distinguish mainline tree from the
2588 See ChangeLog.0 for earlier changes.
2591 add-log-time-format: current-time-string