1 Fri Sep 4 02:01:05 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
3 * config/sparc/sparc.c (output_double_int): In all V9 symbolic
5 (sparc_output_deferred_case_vectors): If no work to do, return.
6 Fix thinko in Sept 1 change.
8 1998-09-03 SL Baur <steve@altair.xemacs.org>
10 * Makefile.in: add semicolon in BISON definition for portability.
12 Thu Sep 3 13:34:41 1998 Toon Moene <toon@moene.indiv.nluug.nl>
14 * config/nextstep.c (handle_pragma): Correct name of third
17 Tue Sep 1 11:30:33 1998 Nick Clifton <nickc@cygnus.com>
19 * config/m32r/m32r.md: Change (reg:CC 17) to (reg:SI 17).
20 * config/m32r/m32r.h: Make register 17 be fixed.
21 * config/m32r/m32r.c: Use SImode for cc operations.
23 Thu Sep 3 18:17:34 1998 Benjamin Kosnik <bkoz@cygnus.com>
25 * invoke.texi (Warning Options): Add -Wnon-template-friend
28 Thu Sep 3 18:16:16 1998 Michael Meissner <meissner@cygnus.com>
30 * rs6000.c (rs6000_override_options): Add -mcpu={401,e603e}.
32 Thu Sep 3 18:05:16 1998 David Edelsohn <edelsohn@mhpcc.edu>
34 * rs6000.md (movsf): Disable explicit secondary-reload-like
35 functionality if TARGET_POWERPC64.
36 (movdf): Remove TARGET_POWERPC64 explicit secondary-reload-like
39 Thu Sep 3 11:41:40 1998 Robert Lipe <robertl@dgii.com>
41 * fixinc.sco: Borrow code to wrap 'bool' typedefs from tinfo.h
42 and term.h from fixinc.wrap.
44 Thu Sep 3 09:47:31 1998 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
46 * aclocal.m4 (GCC_HEADER_STRING): New macro to detect if it is
47 safe to include both string.h and strings.h together.
48 (GCC_NEED_DECLARATION): Test STRING_WITH_STRINGS when deciding
49 which headers to search for function declarations. Continue to
50 prefer string.h over strings.h when both are not acceptable.
52 * acconfig.h (STRING_WITH_STRINGS): Add stub.
54 * configure.in: Call GCC_HEADER_STRING.
56 * system.h: Test STRING_WITH_STRINGS when deciding which headers
57 to include. Continue to prefer string.h over strings.h when both
60 Wed Sep 2 23:56:29 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
62 * config/sparc/sparc.c (output_double_int): If V9 and MEDLOW, do
63 not assume top 32-bits of symbolic addresses are zero if
66 Thu Sep 3 00:23:21 1998 Richard Henderson <rth@cygnus.com>
68 * ginclude/va-alpha.h: Protect entire second portion of the
69 file against double inclusion.
71 Thu Sep 3 00:37:55 1998 Ovidiu Predescu <ovidiu@aracnet.com>
73 Added support for the Boehm's garbage collector.
74 * configure.in: Handle --enable-objc-gc.
76 * Makefile.in (CHECK_TARGETS): Add check-objc.
77 (check-objc): New rule.
78 * objc/Make-lang.in: Build a different Objective-C library that
79 runs with the Boehm's collector.
80 * objc/encoding.c (objc_round_acc_size_for_types): New function.
81 * objc/encoding.c: Correctly compute the size of compound types in
82 the presence of bitfields. Skip the variable name of the type if
83 any. Added support for long long.
84 * objc/encoding.h (_C_GCINVISIBLE): New specifier.
85 (_F_GCINVISIBLE): New mask.
86 * objc/gc.c: New file. Compute the type memory mask associated with
87 a class based on the runtime information.
88 * objc/misc.c: Added the hooks that use the Boehm's collector
90 * objc/objc-act.c (build_class_template): Generate a new class
91 member (gc_object_type) to hold the class' type memory mask.
92 (build_shared_structure_initializer): Initialize the new member to
94 (encode_complete_bitfield): New function. Generate the new
96 (encode_field_decl): Generate the new encoding only for the GNU
98 * objc/objc-api.h (_C_LNG_LNG, _C_ULNG_LNG): New specifiers for the
100 (class_get_gc_object_type): New function to mark a pointer instance
101 variable as a weak pointer.
102 * objc/objc-features.texi: New file.
103 * objc/objc.h (gc_object_type): New class member.
104 * objc/objects.c (class_create_instance): Create a typed memory
105 object when compiled with Boehm's collector support.
106 * objc/sendmsg.c (__objc_init_install_dtable): Call
107 __objc_send_initialize instead of setting the initialize flag.
108 (__objc_send_initialize): Call __objc_generate_gc_type_description
109 to generate the class type memory mask. Rewrite the code that
110 sends the +initialize so that it is called only once (bug report
111 and fix from Ronald Pijnacker <Ronald.Pijnacker@best.ms.philips.com>).
112 * testsuite/objc: New testsuite for Objective-C type encoding.
113 * testsuite/lib/objc-torture.exp: New file.
114 * testsuite/lib/objc.exp: New file.
116 Wed Sep 2 14:47:36 1998 Jim Wilson <wilson@cygnus.com>
118 * jump.c (jump_optimize): In if/then/else transformations, add
119 another call to modified_between_p for the jump insn.
121 Wed Sep 2 14:16:49 1998 Jeffrey A Law (law@cygnus.com)
123 * fix-header.c (symlink): Treat like readlink.
125 Wed Sep 2 19:30:06 1998 J"orn Rennecke <amylaar@cygnus.co.uk>
127 * dwarfout.c (fundamental_type_code): Encode 32 bit floats/doubles
130 Wed Sep 2 10:06:07 1998 Nick Clifton <nickc@cygnus.com>
132 * config/nextstep.h: Update HANDLE_PRAGMA macro.
133 * config/h8300/h8300.h: Update HANDLE_PRAGMA macro.
134 * config/i960/i960.h: Update HANDLE_PRAGMA macro.
136 * config/nextstep.c (handle_pragma): Take three arguments, as per
137 the new HANDLE_PRAGMA macro specification.
138 * config/h8300/h8300.c (handle_pragma): Take three arguments, as
139 per the new HANDLE_PRAGMA macro specification.
140 * config/i960/i960.c (process_pragma): Take three arguments, as
141 per the new HANDLE_PRAGMA macro specification.
143 Wed Sep 2 09:25:29 1998 Nick Clifton <nickc@cygnus.com>
145 * c-lex.c (check_newline): Call HANDLE_PRAGMA before
146 HANDLE_SYSV_PRAGMA if both are defined. Generate warning messages
147 if unknown pragmas are encountered.
148 (handle_sysv_pragma): Interpret return code from
149 handle_pragma_token (). Return success/failure indication rather
150 than next unprocessed character.
151 (pragma_getc): New function: retrieves characters from the
152 input stream. Defined when HANDLE_PRAGMA is enabled.
153 (pragma_ungetc): New function: replaces characters back into the
154 input stream. Defined when HANDLE_PRAGMA is enabled.
156 * c-pragma.c (handle_pragma_token): Return success/failure status
159 * c-pragma.h: Change prototype of handle_pragma_token().
161 * varasm.c: (handle_pragma_weak): Only create this function if
162 HANDLE_PRAGMA_WEAK is defined.
164 * c-common,c (decl_attributes): If defined call the expression
165 contained within the INSERT_ATTRIBUTES macro before adding
166 attributes to a decl.
168 * tm.texi (HANDLE_PRAGMA): Document the new verion of
169 HANDLE_PRAGMA, which takes three arguments.
170 (INSERT_ATTRIBUTES): Document this new macro.
172 * LANGUAGES: Document the new version of HANDLE_PRAGMA and the
173 new INSERT_ATTRIBUTES macro.
175 Wed Sep 2 02:03:23 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
177 * config/sparc/sparc.md (movdf): Only generate special RTL for
179 (move_label_di): Remove
180 (movdi_pic_label_ref, movdi_high_pic_label_ref,
181 movdi_lo_sum_pic_label_ref): New patterns for 64-bit label
183 * config/sparc/sparc.h (ASM_OUTPUT_ADDR_VEC_ELT,
184 ASM_OUTPUT_ADDR_DIFF_ELT): Don't do anything special for MEDLOW,
185 output an .xword for all 64-bit cases.
187 Tue Sep 1 15:55:17 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
189 * config/sparc/sparc.c (finalize_pic): Don't output arbitrary
190 alignment, use FUNCTION_BOUNDARY instead.
191 (sparc_output_deferred_case_vectors): Likewise.
193 Mon Aug 31 17:25:41 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
195 * config/sparc/sparc.md (movsf_const_intreg): Kill warning.
196 (movtf_insn_sp64, movtf_no_e_insn_sp64): Reorder alternatives.
198 Mon Aug 31 13:57:55 1998 Richard Henderson <rth@cygnus.com>
200 * alpha/va_list.h: New file.
201 * alpha/x-alpha (EXTRA_HEADERS): New. Add va_list.h.
203 Mon Aug 31 14:55:02 1998 Jeffrey A Law (law@cygnus.com)
205 * NEWS: Add SCO Openserver and Unixware 7 notes.
209 Mon Aug 31 15:42:18 1998 Dave Brolley <brolley@cygnus.com>
211 * varasm.c (compare_constant_1): Handle RANGE_EXPR.
212 (record_constant_1): Handle RANGE_EXPR.
214 Mon Aug 31 10:54:03 1998 Richard Henderson <rth@cygnus.com>
216 * print-rtl.c (print_rtx): NOTE_INSN_LIVE has an rtx not a bitmap.
217 * haifa-sched.c (sched_analyze): Handle NOTE_INSN_RANGE_START
218 and NOTE_INSN_RANGE_END specially.
219 (reemit_notes): Likewise.
221 Mon Aug 31 10:18:52 1998 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
223 * sparc.c (TMASK, UMASK): Use `(unsigned)1' not `1U'.
224 (ultrasparc_sched_init): Remove unneeded &.
226 Mon Aug 31 10:47:16 1998 Andreas Schwab <schwab@issan.informatik.uni-dortmund.de>
228 * config/m68k/m68k.h (TARGET_SWITCHES): Don't remove MASK_68040
229 for m68020-60, to prevent the use of fintrz.
231 Sun Aug 30 22:17:20 1998 Mark Mitchell <mark@markmitchell.com>
233 * configure.in: If the native compiler is GCC use $(WARN_CFLAGS)
235 * Makefile.in: Likewise.
236 * configure: Regenerated.
238 Sun Aug 30 22:15:41 1998 H.J. Lu (hjl@gnu.org)
240 * configure.in (gxx_include_dir): Changed to
241 '${prefix}/include/g++'-${libstdcxx_interface}.
242 * configure: Rebuilt.
244 Sun Aug 30 20:19:43 1998 Hans-Peter Nilsson <hp@axis.se>
246 * expr.c (expand_expr): Change ">" to ">=" making MOVE_RATIO use
248 * tm.texi (Costs): Say MOVE_RATIO is number of mem-mem move
249 *sequences* *below* which scalar moves will be used.
251 Sun Aug 30 17:18:43 1998 Jeffrey A Law (law@cygnus.com)
253 * collect2.c (mktemp): Delete unused declaration.
255 * config/xm-netbsd.h: Remove unnecessary file.
256 * config/*/xm-netbsd.h: Do not include the generic xm-netbsd.h
257 file anymore, it is not needed.
259 Sun Aug 30 16:05:45 1998 Mark Mitchell <mark@markmitchell.com>
261 * convert.c (convert_to_integer): Issue an error on conversions to
264 Sun Aug 30 16:47:20 1998 Martin von Lvwis <loewis@informatik.hu-berlin.de>
266 * Makefile.in: Add lang_tree_files and gencheck.h.
267 * configure.in: Generate them.
268 * gencheck.c: Include gencheck.h.
270 Sat Aug 29 21:38:24 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
272 * config/sparc/sparc.md (pic_lo_sum_di, pic_sethi_di): Rename to
273 movdi_lo_sum_pic and movdi_high_pic and make visible.
274 * config/sparc/sparc.c (legitimize_pic_address): For -fPIC,
275 emit these when Pmode is not SImode.
276 * config/sparc/linux64.h (SPARC_DEFAULT_CMODEL): Make CM_MEDLOW.
278 Sat Aug 29 14:59:32 1998 Mumit Khan <khan@xraylith.wisc.edu>
280 * i386/cygwin32.h (ASM_OUTPUT_SECTION_NAME): Don't emit
281 .linkonce directive after the first time.
283 Sat Aug 29 12:39:56 1998 Jeffrey A Law (law@cygnus.com)
285 * m68k.md (beq0_di): Generate correct (and more efficient) code when
286 the clobbered operand overlaps with an input.
287 (bne0_di): Similarly.
289 * Makefile.in (INSTALL): Remove "--no-header" argument.
291 * NEWS: Various updates.
293 Fri Aug 28 19:00:44 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
295 * config/sparc/sparc.c (arith_operand, const64_operand,
296 const64_high_operand, arith_double_4096_operand): Mark mode as
298 (create_simple_focus_bits): Remove unused arg highest_bit_set, all
300 (sparc_emit_set_const64): Remove unused variable i.
301 (sparc_splitdi_legitimate): Likewise for addr_part.
302 (ultra_code_from_mask): Likewise for mask.
303 (ultra_cmove_results_ready_p): Fixup entry modulo calc. and
304 reverse return values so it matches usage and comments.
305 (ultra_flush_pipeline): Likewise.
306 (ultra_fpmode_conflict_exists): Likewise, remove unused variable
307 this_type, and allow loads and stores of differing FP modes as
308 they do not create a conflict.
309 (ultra_find_type): Initialize fpmode to SFmode, fix
310 parenthesization thinkos in large conditional.
311 (ultrasparc_sched_init): Mark dump and sched_verbose as unused.
312 Init free_slot_mask after ultra_cur_hist is reset, not before.
313 (ultrasparc_rescan_pipeline_state): Remove unused variable ucode.
314 (ultrasparc_sched_reorder): Don't bzero current pipeline state,
315 use ultra_flush_pipeline instead, then re-init group pointer.
316 Fix statement with no effect. If no progress made in, and no
317 instructions scheduled at all, advance to new pipeline cycle else
318 we get into an endless loop.
319 (ultrasparc_adjust_cost): Remove previous arg.
320 * config/sparc/sparc.h (ADJUST_COST): Update to reflect that.
322 Fri Aug 28 13:52:35 1998 Jim Wilson <wilson@cygnus.com>
324 * sparc.md (DImode, DFmode, TFmode splits): Delete self_reference
325 code. Use reg_overlap_mentioned_p to detect when source and
327 (negtf2_notv9+1): Use DFmode instead of SFmode in last two operands.
329 1998-08-28 Brendan Kehoe <brendan@cygnus.com>
331 * loop.c (check_dbra_loop): Pass COMPARISON_VALUE, not
332 COMPARISON_VAL, into invariant_p.
334 Fri Aug 28 15:13:25 1998 J"orn Rennecke <amylaar@cygnus.co.uk>
336 * regmove.c (regclass_compatible_p): New function.
337 (regmove_optimize): Use it.
339 Use NREGS parameter insted of calling max_reg_num.
341 (fixup_match_1): Don't use code = MINUS when later tieing with
342 a hard register is likely.
344 Fri Aug 28 14:54:07 1998 J"orn Rennecke <amylaar@cygnus.co.uk>
346 * loop.c (check_dbra_loop): Fix calculation of FINAL_VALUE when
347 COMPARISON_VAL was normalized.
349 Thu Aug 27 20:10:46 1998 Jeffrey A Law (law@cygnus.com)
351 * loop.c (check_dbra_loop): The loop ending comparison value
352 must be an invariant or we can not reverse the loop.
354 * loop.c (scan_loop): Count down from max_reg_num - 1 to
355 FIRST_PSEUDO_REGISTER to avoid calling max_reg_num each iteration
357 (load_mems_and_recount_loop_regs_set): Likewise.
359 * i386.c (print_operand): Remove obsolete 'c' docs.
361 Wed Aug 26 17:13:37 1998 Tom Tromey <tromey@cygnus.com>
363 * gthr.h: Document __GTHREAD_MUTEX_INIT_FUNCTION.
364 * frame.c (init_object_mutex): New function.
365 (init_object_mutex_once): Likewise.
367 (__register_frame_info): Likewise.
368 (__register_frame_info_table): Likewise.
369 (__deregister_frame_info): Likewise.
371 Thu Aug 27 15:14:18 1998 Jeffrey A Law (law@cygnus.com)
373 * haifa-sched.c (sched_analyze_insn): Fix thinko in last change.
375 Thu Aug 27 16:34:51 1998 J"orn Rennecke <amylaar@cygnus.co.uk>
377 * loop.c (check_dbra_loop): Enable code for reversal
378 of some loops without a known constant loop end.
380 Wed Aug 26 18:38:15 1998 Richard Henderson <rth@cygnus.com>
382 * haifa-sched.c (last_clock_var): New.
383 (schedule_block): Initialize it.
384 (schedule_insn): Use it to fill insn modes with issue information.
386 * alpha.c (alpha_handle_trap_shadows): Remove do-nothing exit.
387 Tag trapb and next insn with TImode.
388 (alphaev5_insn_pipe, alphaev5_next_group, alphaev5_align_insns): New.
389 (alpha_reorg): Add conditional for alpha_handle_trap_shadows.
390 Invoke alphaev5_align_insns as appropriate.
391 * alpha.h (LABEL_ALIGN_AFTER_BARRIER): Was ALIGN_LABEL_AFTER_BARRIER.
392 (MD_SCHED_VARIABLE_ISSUE): New.
393 * alpha.md (attr type): Add multi.
394 (define_asm_attributes): New.
395 (prologue_stack_probe_loop, builtin_setjmp_receiver): Set type multi.
396 (arg_home): Likewise.
397 (fnop, unop, realign): New.
399 Wed Aug 26 15:55:41 1998 Jim Wilson <wilson@cygnus.com>
401 * iris5.h (PREFERRED_DEBUGGING_TYPE): Undef.
402 * iris5gas.h (PREFERRED_DEBUGGING_TYPE): Define.
404 * configure.in (powerpc-ibm-aix4.[12]*): Change from 4.[12].*.
405 (rs6000-ibm-aix4.[12]*): Likewise.
406 * configure: Regnerate.
408 Wed Aug 26 09:30:59 1998 Nick Clifton <nickc@cygnus.com>
410 * config/arm/thumb.c (thumb_exit): Do not move a4 into lr if it
411 already contains the return address.
413 Wed Aug 26 12:57:09 1998 Jeffrey A Law (law@cygnus.com)
415 * calls.c (expand_call): Use bitfield instructions to extract/deposit
416 word sized hunks when loading unaligned args into registers.
418 * haifa-sched.c (sched_analyze_insn): Only create scheduling
419 barriers for LOOP, EH and SETJMP notes on the loop_notes list.
421 * mn10300.h (RTX_COSTS): Handle UDIV and UMOD too.
423 Wed Aug 26 16:35:37 1998 J"orn Rennecke <amylaar@cygnus.co.uk>
425 * loop.c (check_dbra_loop): Add some code that would allow reversal
426 of some loops without a known constant loop end if it were enabled.
428 Wed Aug 26 11:08:44 1998 Gavin Romig-Koch <gavin@cygnus.com>
430 * mips.md (lshrsi3_internal2+2): Fix type-o.
432 Wed Aug 26 10:53:03 1998 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
434 * system.h: Include stdarg.h/varargs.h, make sure they are ordered
435 correctly with regards to stdio.h.
437 * calls.c: Remove stdarg.h/varargs.h.
440 * combine.c: Likewise.
441 * cpperror.c: Likewise.
442 * cpplib.c: Likewise.
443 * cpplib.h: Likewise.
444 * doprint.c: Likewise.
445 * emit-rtl.c: Likewise.
447 * fix-header.c: Likewise.
449 * genattr.c: Likewise.
450 * genattrtab.c: Likewise.
451 * gencodes.c: Likewise.
452 * genconfig.c: Likewise.
453 * genemit.c: Likewise.
454 * genextract.c: Likewise.
455 * genflags.c: Likewise.
456 * genopinit.c: Likewise.
457 * genoutput.c: Likewise.
458 * genpeep.c: Likewise.
459 * genrecog.c: Likewise.
460 * mips-tfile.c: Likewise.
461 * prefix.c: Likewise.
462 * protoize.c: Likewise.
463 * regmove.c: Likewise.
464 * toplev.c: Likewise.
467 Wed Aug 26 05:09:27 1998 Jakub Jelinek <jj@sunsite.ms.mff.cuni.cz>
469 * config/sparc/sparc.c (sparc_override_options): If not
470 TARGET_FPU, turn off TARGET_VIS.
471 * config/sparc/sparc.h (TARGET_SWITCHES): Add no-vis.
472 (LEGITIMATE_CONSTANT_P): Allow SF/DF mode zero when TARGET_VIS.
473 * config/sparc/sparc.md (movsi_insn): Use fzeros not fzero.
474 (movdi_insn_sp64): Add VIS fzero alternative.
475 (clear_sf, clear_df): New VIS patterns.
476 (movsf, movdf expanders): Allow fp_zero_operand flat out when
478 (one_cmpldi2_sp64): Provide new fnot1 VIS alternative.
480 Tue Aug 25 10:57:41 1998 Mark Mitchell <mark@markmitchell.com>
482 * loop.c (n_times_set, n_times_used, may_not_optimize,
483 reg_single_usage): Convert to varrays. All uses changed.
484 (insert_loop_mem): Return a value.
485 (scan_loop): Tweak AVOID_CC_MODE_COPIES code.
486 (load_mems_and_recount_loop_regs_set): Likewise. Grow the arrays, if
489 Tue Aug 25 23:57:12 1998 Jeffrey A Law (law@cygnus.com)
492 * configure.in: Do not set thread_file to "irix" since no such
495 * sparc.md (float abs/neg splits): Check reload_completed before
496 calling alter_subreg.
498 Tue Aug 25 19:17:59 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
500 * config/sparc/sparc.c (sparc_absnegfloat_split_legitimate): New
502 * config/sparc/sparc.h: Declare it.
503 * config/sparc/sparc.md (float abs/neg splits): Use it.
504 (all other splits): Handle SUBREGs properly where necessary.
505 (unnamed (1<<x)-1 V8PLUS pattern): Disable for now.
507 Tue Aug 25 19:48:46 1998 Jeffrey A Law (law@cygnus.com)
509 * reorg.c (fill_simple_delay_slots): Do not abort if we encounter
510 an insn on the unfilled_slots_list that has no delay slots.
511 (fill_eager_delay_slots): Similarly.
513 Tue Aug 25 13:35:20 1998 Nick Clifton <nickc@cygnus.com>
515 * config/v850/v850.c (movsi_source_operand): Treat CONSTANT_P_RTX
516 as an ordinary operand.
518 Tue Aug 25 12:54:57 1998 Jason Merrill <jason@yorick.cygnus.com>
520 * tree.c (valid_machine_attribute): Don't apply attributes to both
523 Tue Aug 25 12:23:20 PDT 1998 Richard Henderson <rth@cygnus.com>
525 * reload.c (operands_match_p): Handle rtvecs.
527 * i386.c (legitimate_pic_address_disp_p): New.
528 (legitimate_address_p): Use it.
529 (legitimize_pic_address): Use unspecs to represent @GOT and @GOTOFF.
530 Handle constant pool symbols just like statics.
531 (emit_pic_move): Use Pmode not SImode for clarity.
532 (output_pic_addr_const) [SYMBOL_REF]: Remove @GOT and @GOTOFF hacks.
533 [UNSPEC]: New, handling what we killed above.
534 [PLUS]: Detect and abort on invalid symbol arithmetic.
535 * i386.h (CONSTANT_ADDRESS_P): Remove HIGH.
537 Tue Aug 25 12:02:23 1998 Mark Mitchell <mark@markmitchell.com>
539 * alias.c: Include output.h.
540 (DIFFERENT_ALIAS_SETS_P): Don't treat alias sets as
541 different if we're in a varargs function.
542 * Makefile.in (alias.o): Depend on output.h
544 Tue Aug 25 19:20:12 1998 J"orn Rennecke <amylaar@cygnus.co.uk>
546 * sh.h (GIV_SORT_CRITERION): Delete.
548 Tue Aug 25 13:19:46 1998 Dave Brolley <brolley@cygnus.com>
550 * regclass.c (regclass): Use xmalloc/free instead of alloca.
551 * stupid.c (stupid_life_analysis): Ditto.
552 * reload1.c (reload): Ditto.
554 Tue Aug 25 05:48:18 1998 Jakub Jelinek <jj@sunsite.ms.mff.cuni.cz>
556 * config/sparc/sparc.c (arith_4096_operand, arith_add_operand,
557 arith_double_4096_operand, arith_double_add_operand): New
559 * config/sparc/sparc.h (PREDICATE_CODES): Add them, declare them.
560 * config/sparc/sparc.md (adddi3, addsi3, subdi3, subsi3): Use
561 them to transform add/sub 4096 into add/sub -4096.
563 Mon Aug 24 23:31:03 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
565 * loop.c (scan_loop): Allocate some slop to handle pseudos
566 generated by move_movables.
567 (load_mems_and_recount_loop_regs_set): Honor AVOID_CC_MODE_COPIES
570 Mon Aug 24 19:45:40 1998 Jim Wilson <wilson@cygnus.com>
572 * tree.def (DECL_RESULT): Correct documentation.
574 Tue Aug 25 01:15:27 1998 J"orn Rennecke <amylaar@cygnus.co.uk>
576 * reload1.c (reload_reg_free_before_p): New argument EQUIV; Changed
577 all callers. Abort for RELOAD_FOR_INSN. RELOAD_FOR_OUTADDR_ADDR:
578 conflicts will all RELOAD_FOR_OUTPUT reloads.
580 * reload1.c (reload_cse_regs_1): When deleting a no-op move that
581 loads the function result, substitute with a USE.
583 Mon Aug 24 15:20:19 1998 David Edelsohn <edelsohn@mhpcc.edu>
585 * rs6000.h (GO_IF_LEGITIMATE_ADDRESS): Use TARGET_POWERPC64
586 when testing LEGITIMATE_INDEXED_ADDRESS_P DFmode and DImode.
587 (LEGITIMIZE_ADDRESS): Use TARGET_POWERPC64 for INDEXED fixup.
588 * rs6000.c (print_operand, case 'L'): Add UNITS_PER_WORD, not 4.
589 (print_operand, cases 'O' and 'T'): Fix typos in lossage strings.
590 * rs6000.md (fix_truncdfsi2_store): Remove %w from non-CONST_INT
592 (movdf_softfloat32, movdf_hardfloat64, movdf_softfloat64): Change
593 'o' to 'm' for GPR variant constraints.
595 Mon Aug 24 10:25:46 1998 Jeffrey A Law (law@cygnus.com)
597 * loop.c (scan_loop): Honor AVOID_CC_MODE_COPIES.
599 * h8300.h (STRIP_NAME_ENCODING): Fix typo.
601 * sparc.md (TFmode splits): Use reg_overlap_mentioned_p to detect
602 when the source and destination overlap.
604 * stmt.c (emit_case_nodes): Change rtx_function to rtx_fn to avoid
605 clash with global type.
607 Mon Aug 24 00:53:53 1998 Jason Merrill <jason@yorick.cygnus.com>
609 * fixinc.irix: Add curses.h handling from fixinc.wrap.
611 * c-common.c (combine_strings): Also set TREE_READONLY.
612 Change warn_write_strings to flag_const_strings.
613 * c-decl.c, c-tree.h: Likewise.
615 Sun Aug 23 18:39:11 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
617 * config/sparc/sparc.c (sparc_emit_set_const32): If outputting a
618 CONST_INT, not a symbolic reference, don't use a HIGH/LO_SUM
619 sequence, use SET/IOR instead so CSE can see it.
620 * config/sparc/sparc.md (movhi_const64_special,
621 movsi_const64_special): New patterns necessitated by that change.
622 (movhi_high): Remove.
623 (movhi_lo_sum): Change to match an IOR.
624 (movdf_insn_sp32): Test TARGET_V9 not TARGET_ARCH64.
625 (movdf_insn_v9only): New pattern for when V9 but not ARCH64.
626 (movdf_insn_sp64): Test both TARGET_V9 and TARGET_ARCH64.
627 (movdf splits): Allow when not V9 or when not ARCH64 and integer
628 registers are involved.
629 (snesi_zero_extend split): Remove reload_completed test.
630 (unnamed plus and minus zero_extend sidi splits): Add it.
632 Sun Aug 23 11:56:08 1998 Mark Mitchell <mark@markmitchell.com>
634 * extend.texi: Remove description of extension to explicit
635 instantiation that is now endorsed by standard C++.
637 Sun Aug 23 09:39:09 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
639 * config/arc/arc.c (arc_initialize_pic): Remove.
640 * config/arc/arc.h (INITIALIZE_PIC): Similarly, this routine does
641 nothing on any platform and is invoked by no-one, it does not even
642 appear in the documentation.
643 * config/sparc/sparc.h (INITIALIZE_PIC): Likewise.
644 * config/sparc/sparc.c (initialize_pic): Likewise.
645 (find_addr_reg): Remove this as well, no longer referenced after
648 Sun Aug 23 00:17:14 1998 Jeffrey A Law (law@cygnus.com)
650 * recog.c (validate_replace_rtx_group): New function.
651 * recog.h (validate_replace_rtx_group): Declare it.
652 * regmove.c (optimize_reg_copy_3): If any substitution fails, then undo
653 the entire group of substitutions.
655 Sat Aug 22 23:31:00 1998 Klaus-Georg Adams (Klaus-Georg.Adams@chemie.uni-karlsruhe.de)
657 * loop.c (load_mems): Fix initializers.
659 Fri Aug 21 23:07:46 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
661 * config/sparc/sparc.md (TFmode splits): Handle destination
662 registers being referenced in the address correctly.
664 * expmed.c (make_tree) [CONST_INT]: Sign extend even if
665 TREE_UNSIGNED, when bitsize of type's mode is larger than
666 HOST_BITS_PER_WIDE_INT.
668 Fri Aug 21 19:31:31 1998 Alexandre Petit-Bianco <apbianco@cygnus.com>
670 * tree.def (LABELED_BLOCK_EXPR, EXIT_BLOCK_EXPR): New tree nodes.
671 * tree.h (LABELED_BLOCK_LABEL, LABELED_BLOCK_BODY,
672 EXIT_BLOCK_LABELED_BLOCK, EXIT_BLOCK_RETURN, LOOP_EXPR_BODY): New
674 * expr.c (expand_expr): Handle LABELED_BLOCK_EXPR and
677 Thu Aug 20 19:43:44 1998 Jeffrey A Law (law@cygnus.com)
679 * h8300.c (h8300_encode_label): Use '&' for tiny data items.
680 * h8300.h (TINY_DATA_NAME_P): Likewise.
681 (STRIP_NAME_ENCODING): Handle '&'.
683 * mn10200.h (REG_OK_FOR_INDEX_P): Do not check the mode of the
684 register (it could be accessed via an outer SUBREG).
685 (REG_OK_FOR_BASE_P): Likewwise.
686 (GO_IF_LEGITIMATE_ADDRESS): Consistently use REGNO_OK_FOR_BASE_P.
688 * remove.c (optimize_reg_copy_3): Abort instead of silently generating
691 * jump.c (rtx_renumbered_equal_p): Do not consider PLUS commutative.
693 Thu Aug 20 17:35:20 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
695 * config/sparc/sparc.md (movtf_insn_sp32): All memory operands
696 must be offsettable so the splits can be made.
698 Thu Aug 20 13:56:53 1998 Michael Meissner <meissner@cygnus.com>
700 * config/i386/winnt.c: Include system.h, not stdio.h to get
701 sys/param.h pulled in before rtl.h in case the system defines MIN
704 Thu Aug 20 13:44:20 1998 David Edelsohn <edelsohn@mhpcc.edu>
706 * rs6000.md (movqi, movhi): Add CONSTANT_P_RTX.
708 Thu Aug 20 13:15:11 1998 Dave Brolley <brolley@cygnus.com>
710 * stor-layout.c (layout_type): Compute TYPE_SIZE_UNIT correctly for
712 * cpplib.c (cpp_define): Handle macros with parameters.
714 Wed Aug 19 21:33:19 1998 David Edelsohn <edelsohn@mhpcc.edu>
716 * rs6000.c (rs6000_output_load_toc_table): Use ld for 64-bit.
717 (output_toc): Use single TOC slot or llong minimal-toc for DFmode
718 and DImode 64-bit. Use llong for minimal-toc SFmode and
719 SYMBOL_REF / LABEL_REF 64-bit.
720 (output_function_profiler): Use llong for profiler label and ld to
721 load 64-bit label address.
723 Wed Aug 19 17:52:27 1998 Nick Clifton (nickc@cygnus.com)
725 * config/arm/thumb.md (extendqisi2_insn): Cope with REG +
728 Wed Aug 19 14:13:31 PDT 1998 Jeff Law (law@cygnus.com)
730 * version.c: Bump for snapshot.
732 Wed Aug 19 13:10:30 PDT 1998 Jeff Law (law@cygnus.com)
734 * version.c: Bump for snapshot.
736 Wed Aug 19 13:06:47 1998 Jason Merrill <jason@yorick.cygnus.com>
738 * collect2.c (extract_init_priority): Use atoi instead of strtoul.
740 Wed Aug 19 13:51:35 1998 Hans-Peter Nilsson <hp@axis.se>
742 * tm.texi (Misc): Fix typo "teh".
744 * tm.texi (PIC): Fix typo "PPIC".
746 * tm.texi (Caller Saves): Say that DEFAULT_CALLER_SAVES has no
747 effect when -O2 and higher.
748 * invoke.texi (Optimize Options): Likewise for -fcaller-saves.
750 1998-08-19 Michael Hayes <michaelh@ongaonga.chch.cri.nz>
752 * regclass.c: Changed register set documentation to be consistent
755 * final.c (final_start_function) Removed redundant test for
758 Wed Aug 19 13:28:41 1998 Mark Mitchell <mark@markmitchell.com>
760 * rtl.h (rtx_function): New type.
761 (for_each_rtx): New function.
762 * rtlanal.c (for_each_rtx): Define it.
764 * recog.c (change_t): New type.
765 (change_objects, change_old_codes, change_locs, change_olds):
767 (changes): New variable.
768 (validate_change): Dynamically allocate room for more changes, if
769 necessary. Uses changes array instead of change_objects, etc.
770 (apply_change_group): Use changes array instead of
773 * loop.c (loop_mem_info): New type.
774 (loop_mems): New variable.
775 (loop_mems_idx): Likewise.
776 (looop_mems_allocated): Likewise.
777 (scan_loop): Remove nregs parameter.
778 (next_insn_in_loop): New function.
779 (load_mems_and_recount_loop_regs_set): Likewise.
780 (load_mems): Likewise.
781 (insert_loop_mem): Likewise.
782 (replace_loop_mem): Likewise.
783 (replace_label): Likewise.
784 (INSN_IN_RANGE_P): New macro.
785 (loop_optimize): Don't pass max_reg_num() to scan_loop.
786 (scan_loop): Remove nregs parameter, compute it after any new
787 registers are created by load_mems. Use INSN_IN_RANGE_P and
788 next_insn_in_loop rather than expanding them inline. Call
789 load_mems to load memory into pseudos, if appropriate.
790 (prescan_loop): Figure out whether or not there are jumps from the
791 loop to targets other than the label immediately following the
792 loop. Call insert_loop_mem to notice all the MEMs used in the
793 loop, if it could be safe to pull MEMs into REGs for the duration
795 (strength_reduce): Use next_insn_in_loop. Tweak comments.
797 Wed Aug 19 08:29:44 1998 Richard Earnshaw (rearnsha@arm.com)
799 * arm.c (arm_override_options): Remove lie about ignoring PIC flag.
801 Wed Aug 19 07:08:15 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
803 * config/sparc/sparc.c (finalize_pic): Check for the correct
804 nonlocal_goto_receiver UNSPEC number.
805 * config/sparc/sparc.md (nonlocal_goto_receiver): Add comment
806 making note of this dependency existing in sparc.c
807 (negtf2_notv9 split): Give NEG SFmode.
808 (negsf2): Fix insn output string.
810 Tue Aug 18 12:40:27 1998 Richard Henderson <rth@cygnus.com>
812 * c-common.c (decl_attributes): Issue an error if the argument
813 to alias is not a string.
815 Tue Aug 18 10:33:30 1998 Jeffrey A Law (law@cygnus.com)
817 * haifa-sched.c (sched_analyze): Put all JUMP_INSNs on the last
818 pending memory flush list.
820 * combine.c (can_combine_p): Allow combining insns with REG_RETVAL
822 (try_combine): Allow combining insns with REG_LIBCALL notes.
824 * expr.c (emit_block_move): Do not call memcpy as a libcall
825 instead build up a CALL_EXPR and call it like any other
827 (clear_storage): Similarly for memset.
829 * regmove.c (fixup_match_2): Do not call reg_overlap_mentioned_p
832 * Makefile.in (cplus-dem.o): Provide explicit rules for building
835 * regmove.c (optimize_reg_copy_1): Update REG_N_CALLS_CROSSED
836 and REG_LIVE_LENGTH as successful substitutions are made.
838 Tue Aug 18 07:15:27 1998 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
840 * config/sparc/sparc.c (ultra_find_type): Add empty semicolon
841 statement after end of loop label.
843 Tue Aug 18 07:13:27 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
845 * config/sparc/sparc.c (ultra_types_avail): New variable.
846 (ultra_build_types_avail): New function to record mask of insn
847 types in ready list at this cycle.
848 (ultrasparc_sched_reorder): Call it.
849 (ultra_find_type): Use it to quicken the search. Also simplif
850 dependency check, don't use rtx_equal_p because we know exactly
851 what we are looking for.
853 Tue Aug 18 03:20:53 1998 Richard Earnshaw (rearnsha@arm.com)
855 * arm.h (SECONDARY_INPUT_RELOAD_CLASS): Return NO_REGS if compiling
858 Mon Aug 17 21:26:38 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
860 * config/sparc/sparc.md (sltu, sgeu): Don't FAIL, call
862 (movsf_const_intreg, movsf_const_high, movsf_const_lo,
863 movdf_const_intreg and helper splits): New patterns to move float
864 constants into integer registers.
865 (negtf2, negdf2, abstf2, absdf2): Rework using new patterns and
868 Mon Aug 17 11:46:19 1998 Jeffrey A Law (law@cygnus.com)
871 * tree.c (build_index_type): Copy TYPE_SIZE_UNIT from sizetype
873 * c-decl.c (finish_enum): Copy TYPE_SIZ_UNIT from enumtype to tem.
875 * rs6000.c (secondary_reload_class): For TARGET_ELF, indicate that
876 a BASE_REGS register is needed as an intermediate when copying
877 a symbolic value into any register class other than BASE_REGS.
879 * expr.c (move_by_pieces): No longer static. Remove prototype.
880 * rtl.h (move_by_pieces): Add extern prototype.
881 * mips.c (expand_block_move): Handle aligned straight line copy by
882 calling move_by_pieces.
884 * expr.c (expand_expr): Allow assignments from TImode PARM_DECLs
887 Mon Aug 17 10:28:52 1998 Mark Mitchell <mark@markmitchell.com>
889 * stmt.c (expand_end_loop): Tidy. Allow unconditional
890 jumps out of the loop to be treated as part of the exit test.
892 Mon Aug 17 10:06:11 1998 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
893 Jeff Law <law@cygnus.com>
895 * Makefile.in (cplus-dep.o): Use cplus-dem.c from libiberty.
896 * cplus-dem.c: Delete.
898 * Makefile.in (fold-const.o): depend on $(RTL_H).
900 * fold-const.c: Include rtl.h to get the prototype for
901 `set_identifier_local_value'.
903 * loop.c (express_from_1): Remove unused variable `tmp'.
904 (combine_givs): cast the first argument of bzero to char *.
906 * toplev.c (display_help): Remove unused variable `looking_for_start'.
908 * c-decl.c (init_decl_processing): Remove unneeded &.
910 * alpha.h (alpha_initialize_trampoline): Provide prototype.
912 * except.c (set_exception_lang_code, set_exception_version_code):
913 Change parameter from `short' to `int' to avoid using a gcc
916 * except.h (set_exception_lang_code, set_exception_version_code):
917 Likewise for prototypes.
919 * flow.c (count_reg_references): Remove unused variables `regno'
922 * gcse.c (hash_scan_insn): Declare parameter `in_libcall_block'.
924 * prefix.c (translate_name): Cast the result of `alloca'.
926 * varray.h (VARRAY_FREE): Reimplement as a `do-while(0)' statement.
928 Mon Aug 17 09:23:42 1998 Andreas Schwab <schwab@issan.informatik.uni-dortmund.de>
930 * config/m68k/m68k.c: Include "system.h" instead of <stdio.h>.
932 (valid_dbcc_comparison_p): Mark mode argument as unused.
933 (symbolic_operand): Likewise.
934 (legitimize_pic_address): Likewise.
935 (const_uint32_operand): Likewise.
936 (const_sint32_operand): Likewise.
937 * sched.c [!INSN_SCHEDULING]: Define only dummy function
938 schedule_insns and comment out rest of file.
940 * m68k.c (output_move_simode_const): Use subl to move a zero into an
942 (output_move_[hq]imode): Likewise.
944 Mon Aug 17 09:15:47 1998 Jeffrey A Law (law@cygnus.com)
946 * toplev.c (main): Enable -fstrict-aliasing for -O2 and above.
947 * invoke.texi: Corresponding changes.
949 Mon Aug 17 02:03:55 1998 Richard Henderson <rth@cygnus.com>
951 * regclass.c (allocate_reg_info): Respect MIN when clearing data.
953 Sun Aug 16 17:37:06 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
955 * config/sparc/sparc.c (ultra_code_from_mask,
956 ultra_cmove_results_ready_p, ultra_fpmode_conflict_exists,
957 ultra_find_type, ultra_schedule_insn, ultra_flush_pipeline,
958 ultrasparc_sched_init, ultrasparc_variable_issue,
959 ultra_rescan_pipeline_state, ultrasparc_sched_reorder): New
960 functions to describe UltraSPARC pipeline exactly to Haifa.
961 (ultrasparc_adjust_cost): Indicate IMUL type insns have zero cost,
962 as there is nothing the scheduler can do about it. Indicate that
963 REG_DEP_OUTPUT's collide. Fixup formatting.
964 * config/sparc/sparc.h (RTX_COSTS): Fixup integer multiply and
965 divide costs on Ultra for DImode.
966 (MD_SCHED_INIT, MD_SCHED_REORDER, MD_SCHED_VARIABLE_ISSUE):
968 * config/sparc/sparc.md (ieu_unnamed function unit): Rename to
969 ieuN and add call_no_delay_slot to type list.
970 (cti function unit): New unit for branches on UltraSPARC.
971 (subx/addx insns): Set type to misc.
972 (sidi zero/sign extension insns on arch64): Set type to shift.
973 (sign_extendhidi2_insn): Set type to sload.
975 Sun Aug 16 13:52:00 1998 David Edelsohn <edelsohn@mhpcc.edu>
977 * rs6000.c (rs6000_stack_info): Use if == 0 for sizes.
978 (output_epilog): Use if != 0 for offset.
979 (rs6000_fatal_bad_address): Prepare for Intl.
980 * rs6000.h (rs6000_fatal_bad_address): Declare.
981 * rs6000.md (movsfcc, movdfcc): Use else if.
982 (elf_high): Use {liu|lis}.
983 (elf_low): Use {cal|la}. Remove %a template from old mnemonics.
984 (movsi): Use rs6000_fatal_bad_address.
986 Sun Aug 16 01:53:21 1998 Richard Henderson <rth@cygnus.com>
988 * reload.c (find_equiv_reg): Reject equivalences separated
989 by a volatile instruction.
991 Sun Aug 16 00:21:44 1998 Franz Sirl <Franz.Sirl-kernel@lauterbach.com>
993 * rs6000/linux.h (CPP_OS_DEFAULT_SPEC): Define.
995 Sat Aug 15 20:51:35 1998 Richard Henderson <rth@cygnus.com>
997 * alpha.md (movsicc): Fix mode mismatch.
999 Sat Aug 15 20:22:33 1998 H.J. Lu (hjl@gnu.org)
1001 * config/alpha/alpha.h (ASM_OUTPUT_MI_THUNK): Handle aggregated
1003 * config/alpha/win-nt.h (ASM_OUTPUT_MI_THUNK): Likewise.
1005 Sat Aug 15 08:39:49 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
1007 * config/sparc/sparc.md (movsi_lo_sum_pic_label_reg): Remove
1008 write-only modifier from operand 1 constraint.
1010 Sat Aug 15 06:28:19 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
1012 * config/sparc/sparc.c (sparc_emit_set_const64_quick1): If
1013 emitting a XOR of -1 at the end, emit a NOT instead for combine's
1015 (sparc_emit_set_const64): Likewise, also when computing trailing
1016 bits do not negate low_bits and make fast_int an int.
1018 Fri Aug 14 21:07:03 1998 Jeffrey A Law (law@cygnus.com)
1020 * loop.c (add_label_notes): Do not ignore references to labels
1021 before dispatch tables. Mirrors Apr 8 change to mark_jump_label.
1022 * gcse.c (add_label_notes): Similarly.
1024 * pa.h (ASM_OUTPUT_MI_THUNK): Strip name encoding.
1026 * m68k.md (adddi_dilshr32): One of the operands must be a register.
1027 (adddi_dishl32): Similarly.
1029 Fri Aug 14 14:12:59 1998 Jason Merrill <jason@yorick.cygnus.com>
1031 * i386.h (MODES_TIEABLE_P): Reorganize to shut up warnings.
1032 * alias.c (memrefs_conflict_p): Add braces to shut up warnings.
1033 * cse.c (cse_basic_block): Add parens to shut up warnings.
1035 Fri Aug 14 12:58:21 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
1037 * config/sparc/sparc.c (sparc_emit_set_const64_quick2,
1038 sparc_emit_set_const64_longway, const64_is_2insns,
1039 create_simple_focus_bits, sparc_emit_set_const64): Fix more bugs
1040 in 64-bit constant formation.
1041 * config/sparc/sparc.md (snesi_zero_extend split): Generate
1042 rtl for addx not subx.
1043 (define_insn movdi_const64_special): Make available even when
1044 HOST_BITS_PER_WIDE_INT is not 64.
1045 (movdi_lo_sum_sp64_cint, movdi_high_sp64_cint): Remove.
1046 (losum_di_medlow, sethm, setlo): Make op2 symbolic_operand.
1047 (cmp_siqi_trunc_set, cmp_diqi_trunc_set): Encapsulate both
1048 instances of operand 1 inside a QI subreg.
1049 (xordi3_sp64_dbl): Remove '%' constraint for op1.
1050 (one_cmpldi2_sp64): Fix output string.
1051 (one_cmplsi2_not_liveg0): Rewrite to remove unneeded extra
1053 (unnamed arch64 ashift DI): Truncate shift count if greater than
1056 Fri Aug 14 21:52:53 1998 J"orn Rennecke <amylaar@cygnus.co.uk>
1058 * expr.c (store_expr): Don't optimize away load-store pair
1059 when either source or destination have a side effect.
1061 Fri Aug 14 16:50:10 1998 John Carr <jfc@mit.edu>
1063 * genrecog.c (add_to_sequence): Fatal error if the modes of the
1064 operands of SET are incompatible.
1066 * alpha.md: Fix max and min patterns so modes of SET operands match.
1068 Fri Aug 14 12:22:55 1998 Ian Lance Taylor <ian@cygnus.com>
1070 * configure.in: Avoid [[ by using test and changequote.
1071 * configure: Rebuild.
1073 Fri Aug 14 01:22:31 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
1075 * rtl.def (CONSTANT_P_RTX): Fix typo in string name.
1077 * config/sparc/sparc.md (seqdi_special_trunc, snedi_special_trunc,
1078 seqsi_special_extend, snesi_special_extend, snesi_zero_extend and
1079 split, snedi_zero_trunc and split, seqsi_zero_extend and split,
1080 seqdi_zero_trunc and split, pic_lo_sum_di, pic_sethi_di,
1081 movdi_cc_sp64_trunc, movdi_cc_reg_sp64_trunc, addx_extend_sp32 and
1082 split, addx_extend_sp64, subx_extend_sp64, subx_extend and split):
1083 Fix mismatching modes in SET operands.
1084 (conditional move patterns): Fix formatting.
1085 (unnamed subx arch64 pattern): Remove duplicate insn.
1087 Fri Aug 14 00:34:34 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
1089 * config/sparc/sparc.c (const64_operand, const64_high_operand):
1090 Get it right when HOST_BITS_PER_WIDE_INT is not 64.
1091 (input_operand): Fixup test for what we accept for constant
1093 (sparc_emit_set_const32, sparc_emit_set_symbolic_const64): Give
1095 (safe_constDI): Remove.
1096 (sparc_emit_set_safe_HIGH64, gen_safe_SET64, gen_safe_OR64,
1097 gen_safe_XOR64): New functions.
1098 (sparc_emit_set_const64_quick1, sparc_emit_set_const64_quick2,
1099 sparc_emit_set_const64_longway, sparc_emit_set_const64): Use
1101 * config/sparc/sparc.md (define_insn xordi3_sp64_dbl): Only make
1102 available when HOST_BITS_PER_WIDE_INT is not 64.
1103 (define_insn movdi_sp64_dbl, movdi_const64_special): Likewise and
1104 move before movdi_insn_sp64 pattern.
1105 (define_insn movdi_lo_sum_sp64_dbl, movdi_high_sp64_dbl): Remove.
1106 (define_insn sethi_di_medlow, seth44, setm44, sethh): Use
1107 symbolic_operand as predicate for second operand.
1108 (DImode minus split on arch32, negsi2 expander, one_cmplsi2
1109 expander): Give set VOIDmode.
1111 Fri Aug 14 01:45:06 1998 Mumit Khan <khan@xraylith.wisc.edu>
1113 * i386/cygwin32 (DEFAULT_PCC_STRUCT_RETURN): Define.
1115 Fri Aug 14 01:40:21 1998 Geoffrey Keating <geoffk@ozemail.com.au>
1117 * rs6000/linux.h (LINK_SPEC): Pass -G args to the linker.
1119 Fri Aug 14 01:23:23 1998 Richard Earnshaw (rearnsha@arm.com)
1121 * arm/netbsd.h (TARGET_DEFAULT): Default includes software floating
1123 (CPP_FLOAT_DEFAULT_SPEC): Re-define accordingly.
1125 Fri Aug 14 01:19:08 1998 Robert Lipe <robertl@dgii.com>
1127 * install.texi: Various SCO OpenServer tweaks.
1129 Thu Aug 13 20:14:40 1998 Jim Wilson <wilson@cygnus.com>
1131 * reload1.c (eliminate_regs_in_insn): Handle another case when
1132 eliminating the frame pointer to the hard frame pointer. Add
1133 missing ep->to_rtx check to one existing case.
1135 * mips/mips.md (movhi_internal2+2): Fix typo mem:SI -> mem:HI.
1137 Thu Aug 13 17:08:11 1998 Jason Merrill <jason@yorick.cygnus.com>
1139 * tree.h: De-conditionalize init_priority code.
1141 * mips.h (NM_FLAGS): Change from -Bp to -Bn.
1142 * collect2.c (NM_FLAGS): Change from -p to -n.
1144 * configure.in: Turn on collect2 for mipstx39-elf.
1145 Handle use_collect2=no properly.
1147 * c-common.c: De-conditionalize init_priority code.
1148 * collect2.c (extract_init_priority, sort_ids): New fns.
1149 (main): Call sort_ids.
1150 Move sequence_number to file scope.
1152 * configure.in: Handle --enable-init-priority.
1153 * c-common.c (attrs): Add A_INIT_PRIORITY.
1154 (init_attributes, decl_attributes): Likewise.
1155 * tree.h (DEFAULT_INIT_PRIORITY, MAX_INIT_PRIORITY): New macros.
1156 * tree.c (get_file_function_name_long): Split out...
1157 (get_file_function_name): ...from here.
1159 Thu Aug 13 16:09:53 1998 Martin von Loewis <loewis@informatik.hu-berlin.de>
1161 * expr.c (safe_from_p): Change code to ERROR_MARK only when not
1164 Thu Aug 13 15:24:48 1998 Jason Merrill <jason@yorick.cygnus.com>
1166 * toplev.c (display_help): Add braces to shut up warnings.
1167 * tree.c (simple_cst_equal): Likewise.
1169 * fold-const.c (non_lvalue): Don't deal with null pointer
1171 (fold, case COMPOUND_EXPR): Wrap a constant 0 in a NOP_EXPR.
1173 * c-typeck.c (initializer_constant_valid_p): Allow conversion of 0
1174 of any size to a pointer.
1176 Thu Aug 13 12:53:13 1998 Jim Wilson <wilson@cygnus.com>
1178 * i386/winnt.c (i386_pe_asm_file_end): Check TREE_SYMBOL_REFERENCED.
1180 Wed Aug 12 17:25:18 1998 Jeffrey A Law (law@cygnus.com)
1182 * mn10300.c (REG_SAVE_BYTES): Only reserve space for registers
1183 which will be saved.
1184 * mn10300.md (prologue insn): Only save registers which need saving.
1185 (epilogue insn): Similarly.
1187 * mn10300.c, mn10300.h, mn10300.md: Remove "global zero register"
1190 Wed Aug 12 12:39:16 1998 Gavin Romig-Koch <gavin@cygnus.com>
1192 * mips/mips.h (ENCODE_SECTION_INFO): Set SYMBOL_REF_FLAG for
1193 VAR_DECL's in gp addressable sections.
1195 Tue Aug 11 23:02:31 1998 John Carr <jfc@mit.edu>
1197 * sparc.c: Change return <exp> to <exp>; return; in functions
1199 * sparc.md: Add empty semicolon statement after final label in
1202 Tue Aug 11 22:42:01 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
1204 * config/sparc/sparc.md (define_insn addx_extend): Rename to
1205 addx_extend_sp64, only allow when TARGET_ARCH64.
1206 (define_insn addx_extend_sp32 and split): Version that works when
1208 (define_insn subx_extend): Likewise.
1209 (define_split adddi3 and subdi3 with zero extension): Fixup and
1210 correct bugs when not TARGET_ARCH64.
1212 Tue Aug 11 16:04:34 1998 John Carr <jfc@mit.edu>
1214 * except.c (set_exception_lang_code, set_exception_version_code):
1215 Use prototype-style definition if __STDC__, to match declaration
1218 * genemit.c: Change FAIL and DONE macros not to use loops.
1220 Tue Aug 11 12:27:03 1998 Jim Wilson <wilson@cygnus.com>
1222 * dwarf2out.c (ASM_OUTPUT_DWARF_ADDR_CONST): Use
1223 ASM_OUTPUT_DWARF2_ADDR_CONST if defined.
1225 * mips/mips.md (reload_outsi): Use M16_REG_P when TARGET_MIPS16.
1227 Tue Aug 11 18:12:53 1998 Dave Love <d.love@dl.ac.uk>
1229 * README.g77: Update from Craig.
1231 Tue Aug 11 04:46:01 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
1233 * config/sparc/sparc.c (sparc_emit_set_const32): INTVAL is of
1235 (safe_constDI sparc_emit_set_const64_quick1,
1236 sparc_emit_set_const64_quick2, sparc_emit_set_const64_longway,
1237 analyze_64bit_constant, const64_is_2insns,
1238 create_simple_focus_bits): Fix some bugs when compiled on real
1240 (function_arg_record_value_3, function_arg_record_value_2,
1241 function_arg_record_value): Add fully prototyped forward decls.
1242 * config/sparc/sparc.md (define_insn cmpsi_insn_sp32): Rename back
1243 to cmpsi_insn and use on both 64 and 32 bit targets.
1244 (define_insn cmpsi_insn_sp64): Remove.
1245 (define_expand zero_extendsidi2): Allow for 32-bit target too.
1246 (define_insn zero_extendsidi2_insn): Rename to
1247 zero_extendsidi2_insn_sp64.
1248 (define_insn zero_extendsidi2_insn_sp32): New pattern and
1249 assosciated forced split for it.
1251 * config/sparc/sparc.c (const64_operand, const64_high_operand):
1253 * config/sparc/sparc.h: Declare them.
1254 (PREDICATE_CODES): Add them.
1255 * config/sparc/sparc.md (movdi_lo_sum_sp64_dbl,
1256 movdi_high_sp64_dbl, xordi3_sp64_dbl): Use them.
1258 Mon Aug 10 22:57:24 1998 John Carr <jfc@mit.edu>
1260 * config/sparc/sparc.md (define_insn jump): Output ba,pt not b,pt
1261 in v9 case as the latter makes the Solaris assembler crash.
1263 Mon Aug 10 22:39:09 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
1265 * config/sparc/sparc.c (input_operand): Do not accept a LO_SUM MEM
1266 for TFmode when !v9. We require offsettable memory addresses.
1267 * config/sparc/sparc.h (ALTER_HARD_SUBREG): Handle TFmode to
1268 DFmode register number conversions.
1269 * config/sparc/sparc.md (define_split DFmode moves): If register
1270 is a SUBREG do alter_subreg on it before using.
1271 (define_expand movtf): Fixup comment about alignment on v9.
1272 (define_split TFmode moves): Don't use gen_{high,low}part, create
1273 explicit SUBREGs instead.
1275 Mon Aug 10 19:02:55 1998 John Carr <jfc@mit.edu>
1277 * Makefile.in (mbchar.o): Depend on mbchar.c.
1279 Mon Aug 10 04:28:13 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
1280 Richard Henderson <rth@cygnus.com>
1282 Rewrite Sparc backend for better code generation and
1283 improved sparc64 support.
1284 * config/sparc/sp64-elf.h: Set JUMP_TABLES_IN_TEXT_SECTION to
1286 * config/sparc/sysv4.h: Likewise.
1287 * config/sparc/sparc.c (v8plus_regcmp_p, sparc_operand,
1288 move_operand, v8plus_regcmp_op, emit_move_sequence,
1289 singlemove_string, doublemove_string, mem_aligned_8,
1290 output_move_double, output_move_quad, output_fp_move_double,
1291 move_quad_direction, output_fp_move_quad, output_scc_insn):
1293 (small_int_or_double): New predicate.
1294 (gen_compare_reg): Remove TARGET_V8PLUS cmpdi_v8plus emission.
1295 (legitimize_pic_address): Emit movsi_{high,lo_sum}_pic instead of
1296 old pic_{sethi,lo_sum}_si patterns.
1297 (mem_min_alignment): New generic function to replace
1298 mem_aligned_8, which uses REGNO_POINTER_ALIGN information when
1299 available and can test for arbitrary alignments. All callers
1301 (save_regs, restore_regs, build_big_number,
1302 output_function_prologue, output_cbranch, output_return,
1303 sparc_flat_save_restore, sparc_flat_output_function_prologue,
1304 sparc_flat_output_function_epilogue): Prettify
1306 (output_function_epilogue): Likewise and add code to output
1307 deferred case vectors.
1308 (output_v9branch): Likewise, add new arg INSN and use it to tack
1309 on branch prediction settings. All callers changed.
1310 (print_operand): Likewise and output %l44 for LO_SUMs when
1312 (sparc_splitdi_legitimate): New function to make sure DImode
1313 splits can be run properly when !arch64.
1314 (sparc_initialize_trampoline, sparc64_initialize_trampoline):
1315 Reformat example code in comments.
1316 (set_extends): Remove UNSPEC/v8plus_clear_high case.
1317 (sparc_addr_diff_list, sparc_addr_list): New statics to keep track
1318 of deferred case vectors we need to output.
1319 (sparc_defer_case_vector): Record a case vector.
1320 (sparc_output_addr_vec, sparc_output_addr_diff_vec,
1321 sparc_output_deferred_case_vectors): New functions to output them.
1322 (sparc_emit_set_const32): New function to form 32-bit constants in
1323 registers when that requires more than one instruction.
1324 (safe_constDI, sparc_emit_set_const64_quick1,
1325 sparc_emit_set_const64_quick2, sparc_emit_set_const64_longway,
1326 analyze_64bit_constant, const64_is_2insns,
1327 create_simple_focus_bits, sparc_emit_set_const64): New functions
1328 which do the same for 64-bit constants when arch64.
1329 (sparc_emit_set_symbolic_const64): New function to emit address
1330 loading for all code models on v9.
1331 * config/sparc/sparc.h (CONDITIONAL_REGISTER_USAGE): Do not make
1332 %g1 fixed when arch64, unfix %g0 when TARGET_LIVE_G0.
1333 (ALTER_HARD_SUBREG): Fix thinko, return REGNO + 1 not 1.
1334 (SECONDARY_INPUT_RELOAD_CLASS, SECONDARY_OUTPUT_RELOAD_CLASS): Fix
1335 inaccuracies in comments, add symbolic and text_segment operands
1336 when TARGET_CM_MEDANY and TARGET_CM_EMBMEDANY respectively. Use
1337 GENERAL_REGS in these cases as a temp REG is needed to load these
1338 addresses into a register properly.
1339 (EXTRA_CONSTRAINT): Document more accurately, remove Q case as it
1341 (GO_IF_LEGITIMATE_ADDRESS): Allow TFmode for LO_SUM on v9 since fp
1342 quads are guarenteed to have 16-byte alignment.
1343 (LEGITIMIZE_ADDRESS): For SYMBOL_REF, CONST, and LABEL_REF use
1344 copy_to_suggested_reg instead of explicit LO_SUM and HIGH.
1345 (ASM_OUTPUT_ADDR_VEC, ASM_OUTPUT_ADDR_DIFF_VEC): New macros for
1346 deferred case vector implementation.
1347 (ASM_OUTPUT_ADDR_VEC_ELT): Use fputc to output newline.
1348 (ASM_OUTPUT_ADDR_DIFF_ELT): Parenthesize LABEL in macro calls.
1349 Generate "internal label - label" instead of "label - 1b".
1350 (PRINT_OPERAND_ADDRESS): For LO_SUM use %l44 on TARGET_CM_MEDMID.
1351 (PREDICATE_CODES): Remove sparc_operand, move_operand,
1352 v8plus_regcmp_op. Add small_int_or_double, input_operand, and
1354 (doublemove_string, output_block_move, output_fp_move_double,
1355 output_fp_move_quad, output_move_double, output_move_quad,
1356 output_scc_insn, singlemove_string, mem_aligned_8, move_operand,
1357 sparc_operand, v8plus_regcmp_op, v8plus_regcmp_p): Remove externs.
1358 (sparc_emit_set_const32, sparc_emit_set_const64,
1359 sparc_emit_set_symbolic_const64, input_operand, zero_operand,
1360 mem_min_alignment, small_int_or_double): Add externs.
1361 * config/sparc/sparc.md: Document the many uses of UNSPEC and
1362 UNSPEC_VOLATILE in this backend.
1363 (define_function_unit ieu): Rename to ieu_unnamed. Add move and
1364 unary to types which execute in it.
1365 (define_function_unit ieu_shift): Rename to ieu0.
1366 (define_function_unit ieu1): New, executes compare, call, and
1367 uncond_branch type insns.
1368 (define_function_units for type fdivs, fdivd, fsqrt): These
1369 execute in the fpu multiply unit not the adder on UltraSparc.
1370 (define_expand cmpdi): Disallow TARGET_V8PLUS.
1371 (define_insn cmpsi_insn): Rename to cmpsi_insn_sp32.
1372 (define_insn cmpsi_insn_sp64): New, same as sp32 variant except it
1373 allows the arith_double_operand predicate and rHI constraint when
1375 (define_insn cmpdi_sp64, cmpsf_fpe, cmpdf_fpe, cmptf_fpe,
1376 cmpsf_fp, cmpdf_fp, cmptf_fp, sltu_insn, neg_sltu_insn,
1377 neg_sltu_minux_x, neg_sltu_plus_x, sgeu_insn, neg_sgeu_insn,
1378 sltu_plus_x, sltu_plus_x, sltu_plus_x_plus_y, x_minus_sltu,
1379 sgeu_plus_x, x_minus_sgeu, movqi_cc_sp64, movhi_cc_sp64,
1380 movsi_cc_sp64, movdi_cc_sp64, movsf_cc_sp64, movdf_cc_sp64,
1381 movtf_cc_sp64, movqi_cc_reg_sp64, movhi_cc_reg_sp64,
1382 movsi_cc_reg_sp64, movdi_cc_reg_sp64, movsf_cc_reg_sp64,
1383 movdf_cc_reg_sp64, movtf_cc_reg_sp64, zero_extendhisi2_insn,
1384 cmp_siqi_trunc, cmp_siqi_trunc_set, sign_extendhisi2_insn,
1385 sign_extendqihi2_insn, sign_extendqisi2_insn,
1386 sign_extendqidi2_insn, sign_extendhidi2_insn,
1387 extendsfdf2, extendsftf2, extenddftf2, truncdfsf2, trunctfsf2,
1388 trunctfdf2, floatsisf2, floatsidf2, floatsitf2, floatdisf2,
1389 floatdidf2, floatditf2, fix_truncsfsi2, fix_truncdfsi2,
1390 fix_trunctfsi2, fix_truncsfdi2, fix_truncdfdi2, fix_trunctfdi2,
1391 adddi3_sp64, addsi3, cmp_ccx_plus, cmp_cc_plus_set, subdi_sp64,
1392 subsi3, cmp_minus_ccx, cmp_minus_ccx_set, mulsi3, muldi3,
1393 muldi3_v8plus, cmp_mul_set, mulsidi3, mulsidi3_v8plus,
1394 const_mulsidi3_v8plus, mulsidi3_sp32, const_mulsidi3,
1395 smulsi3_highpart_v8plus, unnamed subreg mult,
1396 const_smulsi3_highpart_v8plus, smulsi3_highpart_sp32,
1397 const_smulsi3_highpart, umulsidi3_v8plus, umulsidi3_sp32,
1398 const_umulsidi3, const_umulsidi3_v8plus, umulsi3_highpart_v8plus,
1399 const_umulsi3_highpart_v8plus, umulsi3_highpart_sp32,
1400 const_umulsi3_highpart, divsi3, divdi3, cmp_sdiv_cc_set, udivsi3,
1401 udivdi3, cmp_udiv_cc_set, smacsi, smacdi, umacdi, anddi3_sp64,
1402 andsi3, and_not_di_sp64, and_not_si, iordi3_sp64, iorsi3,
1403 or_not_di_sp64, or_not_si, xordi3_sp64, xorsi3, xor_not_di_sp64,
1404 xor_not_si, cmp_cc_arith_op, cmp_ccx_arith_op,
1405 cmp_cc_arith_op_set, cmp_ccx_arith_op_set, cmp_ccx_xor_not,
1406 cmp_cc_xor_not_set, cmp_ccx_xor_not_set, cmp_cc_arith_op_not,
1407 cmp_ccx_arith_op_not, cmp_cc_arith_op_not_set,
1408 cmp_ccx_arith_op_not_set, negdi2_sp64, cmp_cc_neg, cmp_ccx_neg,
1409 cmp_cc_set_neg, cmp_ccx_set_neg, one_cmpldi2_sp64, cmp_cc_not,
1410 cmp_ccx_not, cmp_cc_set_not, cmp_ccx_set_not, addtf3, adddf3,
1411 addsf3, subtf3, subdf3, subsf3, multf3, muldf3, mulsf3,
1412 muldf3_extend, multf3_extend, divtf3, divdf3, divsf3, negtf2,
1413 negdf2, negsf2, abstf2, absdf2, abssf2, sqrttf2, sqrtdf2, sqrtsf2,
1414 ashlsi3, ashldi3, unnamed DI ashift, cmp_cc_ashift_1,
1415 cmp_cc_set_ashift_1, ashrsi3, ashrdi3, unnamed DI ashiftrt,
1416 ashrdi3_v8plus, lshrsi3, lshrdi3, unnamed DI lshiftrt,
1417 lshrdi3_v8plus, tablejump_sp32, tablejump_sp64, call_address_sp32,
1418 call_symbolic_sp32, call_address_sp64, call_symbolic_sp64,
1419 call_address_struct_value_sp32, call_symbolic_struct_value_sp32,
1420 call_address_untyped_struct_value_sp32,
1421 call_symbolic_untyped_struct_value_sp32, call_value_address_sp32,
1422 call_value_symbolic_sp32, call_value_address_sp64,
1423 call_value_symbolic_sp64, branch_sp32, branch_sp64,
1424 flush_register_windows, goto_handler_and_restore,
1425 goto_handler_and_restore_v9, goto_handler_and_restore_v9_sp64,
1426 flush, all ldd/std peepholes, return_qi, return_hi, return_si,
1427 return_addsi, return_di, return_adddi, return_sf, all call+jump
1428 peepholes, trap, unnamed trap insns): Prettify output strings.
1429 (define_insn anddi3_sp32, and_not_di_sp32, iordi3_sp32,
1430 or_not_di_sp32, xordi3_sp32, xor_not_di_sp32, one_cmpldi2):
1431 Likewise and force + implement splits for integer cases.
1432 (define_insn return_sf_no_fpu): Likewise and allow to match when
1433 no-fpu because of our subreg SFmode splits.
1434 (define_insn zero_extendqihi2, zero_extendqisi2_insn,
1435 zero_extendqidi2_insn, zero_extendhidi2_insn,
1436 zero_extendsidi2_insn, sign_extendsidi2_insn): Likewise and use
1437 input_operand for second operand.
1438 (cmp_minus_cc, cmp_minus_cc_set): Likewise and use
1439 reg_or_0_operand for operand 2 so new splits can use it.
1440 (cmp_zero_extendqisi2, cmp_zero_extendqisi2_set, cmp_cc_plus,
1441 cmp_cc_xor_not): Likewise and don't forget to check TARGET_LIVE_G0
1443 (cmp_zero_extract, cmp_zero_extract_sp64): Likewise and allow
1444 CONST_DOUBLEs for operand 2.
1445 (define_insn move_label_di): Likewise and label distance
1446 optimization because it no longer works with new deferred case
1447 vector scheme. To be revisited.
1448 (define_insn x_minus_y_minus_sltu, x_minus_sltu_plus_y): Likewise
1449 and allow reg_or_0_operand and J constraint for second operand.
1450 (define_insn jump): Set branch predict taken on V9.
1451 (define_insn tablejump): Emit LABEL_REF + PLUS memory address for
1452 new deferred case vector scheme.
1453 (define_insn pic_tablejump_32, pic_tablejump_64): Remove.
1454 (define_insn negdi2_sp32): Force + implement splits.
1455 (define_insn negsi2, one_cmplsi2): Rename to negsi2_not_liveg0 and
1456 one_cmplsi2_not_liveg0 respectively, and create expander of original
1457 names which emit special rtl for TARGET_LIVE_G0.
1458 (define_insn cmpdi_v8plus, scc_si, scc_di): Remove.
1459 (define_insn seq, sne, slt, sge, sle, sltu, sgeu): Don't do
1460 gen_compare_reg, FAIL instead.
1461 (define_insn sgtu, sleu): Likewise and check gen_s*() return
1462 values when trying to reverse condition codes, if they FAIL then
1464 (define_insn snesi_zero, neg_snesi_zero, snesi_zero_extend,
1465 snedi_zero, neg_snedi_zero, snedi_zero_trunc, seqsi_zero,
1466 neg_seqsi_zero, seqsi_zero_extend, seqdi_zero, neg_seqdi_zero,
1467 seqdi_zero_trunc, x_plus_i_ne_0, x_minus_i_ne_0, x_plus_i_eq_0,
1468 x_minus_i_eq_0): Add new splits to perform these multi-insn cases,
1469 set output string to # to indicate they are mandatory splits.
1470 (define_insn pic_lo_sum_si, pic_sethi_si, pic_lo_sum_di,
1471 pic_sethi_di, move_pic_label_si): Remove.
1472 (define_insn movsi_low_sum, movsi_high, movsi_lo_sum_pic,
1473 movsi_high_pic, movsi_pic_label_reg): New patterns to take their
1475 (define_expand movsi_pic_label_ref, define_insn
1476 movsi_high_pic_label_ref, movsi_lo_sum_pic_label_ref): New
1477 expander and insns to handle PIC label references and deferred
1479 (define_insn get_pc_via_rdpc): Comment out as it is no longer
1481 (define_expand movqi, movhi, movsi, movdi, movsf, movdf, movtf):
1482 Rewrite to not use emit_move_sequence, make use of new constant
1483 formation code, and new splits for all multi-insn cases.
1484 (define_insn movqi_insn): Remove sethi case, it can never happen.
1485 Use reg_or_zero_operand instead of const0_rtx explicit test,
1486 use input_operand instead of move_operand for source, and use
1487 general_operand now for dest.
1488 (define_insn movhi_insn): Similar but leave sethi case.
1489 (define_insn lo_sum_qi, store_qi, store_hi): Remove.
1490 (define_insn sethi_hi lo_sum_hi): Rename to movhi_high and
1491 movhi_lo_sum respectively, prettify output string.
1492 (define_insn movsi_zero_liveg0): New pattern to put zero into a
1493 register when needed on TARGET_LIVE_G0.
1494 (define_insn movsi_insn): Use general_operand and input_operand
1495 for dest and src respectively. Simplify applicability test.
1496 Prettify output strings, and add clr alternative for J
1498 (define_insn movdi_sp32_v9, movdi_sp32, define_splits for
1499 deprecated std and reg-reg DI moves): Remove and...
1500 (define_insn movdi_insn_sp32, movdi_insn_sp64): Replace with new
1501 implementation which uses forced splits for all non-single insn
1503 (define_split DI move cases on !arch64): New splits to handle all
1504 situations of 64-bit double register DImode on 32bit, and
1505 unaligned registers and memory addresses for all subtargets.
1506 (define_insn movsf_const_insn, movdf_const_insn, store_sf):
1508 (define_insn movsf_insn, movsf_no_f_insn): Use general_operand and
1509 input_operand for dest and src respectively, prettify output
1511 (define_insn movdf_insn, movdf_no_e_insn, store_df,
1512 movtf_const_insn, movtf_insn, movtf_no_e_insn, store_tf): Remove
1514 (define_insn movdf_insn_sp32, movdf_no_e_insn_sp32,
1515 movdf_insn_sp64, movdf_no_e_insn_sp64, movtf_insn,
1516 movtf_no_e_insn_sp32, movtf_insn_hq_sp64, movtf_insn_sp64,
1517 movtf_no_e_insn_sp64) Replace with new
1518 implementation which uses forced splits for all non-single insn
1520 (define_split DF move cases): New splits in similar vein to DI
1522 (define_insn sethi_di_medlow, sethi_di_medium_pic,
1523 sethi_di_embmedany_data, sethi_di_embmedany_text, sethi_di_sp64,
1524 movdi_sp64_insn): Remove old v9 code model and constant loading
1526 (define_insn pic_lo_sum_di, pic_sethi_di,
1527 sethi_di_medlow_embmedany_pic, sethi_di_medlow, losum_di_medlow,
1528 seth44, setm44, setl44, sethh, setlm, sethm, setlo,
1529 embmedany_sethi, embmedany_losum, embmedany_brsum,
1530 embmedany_textuhi, embmedany_texthi, embmedany_textulo,
1531 embmedany_textlo, movdi_lo_sum_sp64_cint, movdi_lo_sum_sp64_dbl,
1532 movdi_high_sp64_cint, movdi_high_sp64_dbl): Replace with new
1533 scheme, using unspecs, secondary reloads, and one to one sparc
1534 insn to rtl insn mapping for better scheduling and code gen.
1535 (define_expand reload_indi, reload_outdi): Reload helpers for
1536 MEDANY and EMBMEDANY symbol address loading cases which require a
1538 (define_expand movsicc): Remove v8plus_regcmp cases.
1539 (define_insn movdi_cc_sp64_trunc, movdi_cc_reg_sp64_trunc,
1540 cmp_zero_extendqidi2, cmp_zero_extendqidi2_set, cmp_qidi_trunc,
1541 cmp_diqi_trunc_set): New patterns used by some of the new scc
1543 (define_insn xordi3_sp64_dbl): New pattern used for constant
1544 formation when crossing from 32-bit targets.
1545 (define_insn movsi_cc_reg_v8plus, v8plus_clear_high, and helper
1547 (define_insn addx, subx): Make visible and prettify.
1548 (define_insn adddi3_insn_sp32): Likewise and force split.
1549 (define_insn addx_extend, subx_extend, unnamed): New patterns for
1550 64bit scc split usage.
1551 (define_insn unnamed plusDI zero_extend, unnamed minusDI
1552 zero_extend, subdi3): Force and implement splits.
1554 * final.c (final_scan_insn): Don't output labels if target
1555 specifies ASM_OUTPUT_ADDR_{DIFF}_VEC. Do these macro operations
1558 * reorg.c (dbr_schedule): When taking on BR_PRED notes at the end,
1559 don't forget to walk inside SEQUENCESs too as these are what the
1560 delay slot scheduler will create.
1562 Mon Aug 10 01:21:01 1998 Richard Henderson <rth@cygnus.com>
1564 * alpha.md (extxl+1,+2): New patterns to work around
1567 Sat Aug 8 19:20:22 1998 Gary Thomas (gdt@linuxppc.org)
1569 * rs6000.c (rs6000_allocate_stack_space) Fix typo which
1570 caused bad assembly code to be generated.
1572 Sat Aug 8 18:53:28 1998 Jeffrey A Law (law@cygnus.com)
1574 * netbsd.h: Fix typo.
1576 Mon Aug 3 00:06:42 1998 Robert Lipe <robertl@dgii.com>
1578 * config.sub: Fix typo.
1580 Sun Aug 2 22:39:08 1998 Hans-Peter Nilsson <hp@axis.se>
1582 * invoke.texi (Environment Variables): Typo: Change "ascpects"
1584 (Running Protoize): Typo: Change "ther" into "other".
1586 Sun Aug 2 00:42:50 1998 Jeffrey A Law (law@cygnus.com)
1588 * i386/netbsd.h: Undo previous change to DWARF2_UNWIND_INFO.
1589 * m68k/netbsd.h: Likewise.
1590 * ns32k/netbsd.h: Likewise.
1591 * sparc/netbsd.h: Likewise.
1593 Sat Aug 1 17:59:30 1998 Richard Henderson <rth@cygnus.com>
1595 * ginclude/va-alpha.h (va_list): Use a typedef, not a define.
1596 * ginclude/va-clipper.h (va_list): Likewise.
1598 Fri Jul 31 20:22:02 1998 Michael Meissner <meissner@cygnus.com>
1600 * rs6000.c (rs6000_override_options): If big endian and -Os, use
1601 load/store multiple instructions unless user overrides.
1603 Fri Jul 31 17:08:59 1998 Jeffrey A Law (law@cygnus.com)
1605 * ns32k/netbsd.h: Fix typo.
1607 Fri Jul 31 10:23:55 1998 Doug Evans <devans@canuck.cygnus.com>
1609 * m32r/m32r.h (ASM_OUTPUT_SOURCE_LINE): Always output line number
1610 labels with .debugsym if no parallel insns.
1612 Thu Jul 30 19:15:53 1998 Richard Henderson <rth@cygnus.com>
1614 * alpha.md (fp cmp): Replicate patterns for ALPHA_TP_INSN.
1615 (fcmov): Remove ALPHA_TP_INSN patterns -- fcmov doesn't trap.
1617 Thu Jul 30 19:50:15 1998 David Edelsohn <edelsohn@mhpcc.edu>
1619 * rs6000/x-aix43 (AR_FOR_TARGET_FLAGS): Delete.
1620 (AR_FOR_TARGET): Define.
1622 Thu Jul 30 12:29:12 1998 Mark Mitchell <mark@markmitchell.com>
1624 * dyn-string.h: New file.
1625 * dyn-string.c: Likewise.
1626 * Makefile.in (OBJS): Add dyn-string.o.
1627 (dwarf2out.o): Add dyn-string.h dependency.
1628 (dyn-string.o): List dependencies.
1629 * dwarf2out.c: Include dyn-string.h.
1630 (ASM_NAME_TO_STRING): Use dyn_string_append, rather than strcpy.
1631 (addr_const_to_string): Take a dyn_string_t, not a char* as a
1632 prototype. Use dyn_string_append rather than strcat, throughout.
1633 (addr_to_string): Use dyn_string_t.
1635 Thu Jul 30 13:08:07 1998 Ken Raeburn <raeburn@cygnus.com>
1637 Function entry/exit profiling instrumentation:
1638 * expr.h (profile_function_entry_libfunc,
1639 profile_function_exit_libfunc): Declare new variables.
1640 * optabs.c: Define them here.
1641 (init_optabs): Initialize them.
1642 * tree.h (struct tree_decl): New flag
1643 no_instrument_function_entry_exit.
1644 (DECL_NO_INSTRUMENT_FUNCTION_ENTRY_EXIT): New accessor macro.
1645 * c-decl.c (duplicate_decls): Merge it.
1646 * c-common.c (enum attrs): New value A_NO_INSTRUMENT_FUNCTION.
1647 (init_attributes): Use it for "no_instrument_function".
1648 (decl_attributes): Handle it, for functions that have not yet been
1649 compiled. Set decl flag.
1650 * flags.h (flag_instrument_function_entry_exit): Declare new
1652 * toplev.c (flag_instrument_function_entry_exit): Define it here.
1653 (f_options): New option "instrument-functions".
1654 * function.h (struct function): New field instrument_entry_exit.
1655 * function.c (current_function_instrument_entry_exit): New
1657 (push_function_context_to, pop_function_context_from): Save and
1659 (expand_function_start): Set current_ variable, maybe emit return
1660 label and entry profile call.
1661 (expand_function_end): Maybe emit exit profile call.
1663 Thu Jul 30 00:58:34 1998 Jeffrey A Law (law@cygnus.com)
1665 * i386.md (movqi): When optimizing a load of (const_int 1) into a
1666 NON_QI_REG_P, pretend the register is SImode.
1668 Wed Jul 29 23:49:23 1998 Todd Vierling <tv@netbsd.org>
1670 * configure.in: Use xm-netbsd.h as the NetBSD xm file (not xm-siglist).
1671 Accept arm32 as arm, m68k4k as m68k, mipsle as mips-dec, and any
1672 manufacturer id for ns32k.
1673 * configure: Regenerated.
1674 * config/netbsd.h: When using ASM_WEAKEN_LABEL, make it global too.
1675 * config/t-netbsd: Don't compile libgcc1-test as the fns are in libc.
1676 * config/i386/netbsd.h: Undefine DWARF2_UNWIND_INFO, not define as 0.
1677 * config/m68k/netbsd.h: Same.
1678 * config/ns32k/netbsd.h: Same.
1679 * config/sparc/netbsd.h: Same.
1681 Wed Jul 29 22:39:21 1998 Jeffrey A Law (law@cygnus.com)
1683 * unroll.c (unroll_loop): Do not abort for an UNROLL_MODULO
1684 or UNROLL_COMPLETELY loop that starts with a jump to its
1687 Wed Jul 29 22:18:14 1998 David Edelsohn <edelsohn@mhpcc.edu>
1689 * rs6000/rs6000.md (absdi2 define_split): Swap operands of MINUS.
1690 * rs6000/rs6000.c (mask64_operand): Use HOST_BITS_PER_WIDE_INT.
1691 (print_operand, case 'B'): Don't fall through.
1692 (print_operand, case 'S'): Correct mask begin/end computation.
1693 Use HOST_BITS_PER_WIDE_INT.
1694 * rs6000/rs6000.h (CPP_PREDEFINES): Define _LONG_LONG.
1695 (CONDITIONAL_REGISTER_USAGE): GPR13 fixed if TARGET_64BIT.
1696 * rs6000/aix41.h (CPP_PREDEFINES): Same.
1697 * rs6000/aix43.h (CPP_PREDEFINES): Same.
1699 Wed Jul 29 11:47:10 1998 Nick Clifton <nickc@cygnus.com>
1701 * config/arm/thumb.md (extendqisi2_insn): Remove earlyclobber
1702 constraint from second alternative.
1704 Tue Jul 28 23:29:04 1998 Jason Merrill <jason@yorick.cygnus.com>
1706 * configure.in: Fix --without/--disable cases for local-prefix,
1707 gxx-include-dir and checking.
1709 Tue Jul 28 22:01:23 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
1711 * configure.in (enable_haifa): Set by default for sparc64 too.
1714 Tue Jul 28 23:29:04 1998 Jason Merrill <jason@yorick.cygnus.com>
1716 * i386/cygwin32.h (VALID_MACHINE_TYPE_ATTRIBUTE): New macro.
1717 * i386/winnt.c (associated_type): New fn.
1718 (i386_pe_valid_type_attribute_p): New fn.
1719 (i386_pe_check_vtable_importexport): Remove.
1720 (i386_pe_dllexport_p): Use associated_type.
1721 (i386_pe_dllimport_p): Likewise.
1723 From Antonio M. O. Neto <anmendes@cruzeironet.com.br>:
1724 * i386.c (i386_valid_type_attribute_p): Also accept
1725 attributes for METHOD_TYPEs.
1727 Tue Jul 28 23:17:39 1998 Peter Gerwinski <peter@gerwinski.de>
1729 * tree.c (build_range_type): Copy TYPE_SIZE_UNIT.
1731 Tue Jul 28 22:31:12 1998 Craig Burley <burley@gnu.org>
1733 * gcc.c: Fix commentary describing %g, %u, %U, and %O.
1735 * gcc.c (do_spec_1): Fix handling of %g%O and %U%O to prevent
1736 them from generating a new base name for each occurence of
1739 1998-07-28 Vladimir N. Makarov <vmakarov@cygnus.com>
1741 * cse.c (cse_insn): Enable subsitution inside libcall only for REG,
1743 * rtlanal.c (replace_rtx): Prohibit replaces in CONST_DOUBLE.
1747 * cplus-dem.c (type_kind_t): New type.
1748 (demangle_template_value_parm): Add type_kind_t parameter. Rely
1749 on this paramter, rather than demangling the type again.
1750 (demangle_integral_value): Pass tk_integral.
1751 (demangle_template_: Pass the value returned from do_type.
1752 (do_type): Return a type_kind_t. Pass tk_integral to
1753 demangle_template_value_parm for array bounds.
1754 (demangle_fund_type): Likewise.
1756 Mon Jul 27 00:54:41 1998 Jason Merrill <jason@yorick.cygnus.com>
1758 * tree.c (simple_cst_equal, case CONSTRUCTOR): OK if the elts are
1761 Mon Jul 27 22:18:36 1998 Jeffrey A Law (law@cygnus.com)
1763 * pa.c (move_operand): Accept CONSTANT_P_RTX.
1765 Mon Jul 27 17:18:52 1998 Dave Brolley <brolley@cygnus.com>
1767 * stor-layout.c (layout_type): Handle arrays of bits, for Chill.
1769 * expr.c (get_inner_reference): Handle zero-based, unsigned, array
1772 Mon Jul 27 14:51:33 1998 Jeffrey A Law (law@cygnus.com)
1774 * mn10300.h (DEBUGGER_AUTO_OFFSET): Define.
1775 (DEBUGGER_ARG_OFFSET): Likewise.
1777 * mn10300.md (movsf): Remove last change. Not needed.
1779 Mon Jul 27 14:22:36 1998 Dave Brolley <brolley@cygnus.com>
1781 * c-lex.c (yylex): Fix boundary conditions in character literal and
1782 string literal loops.
1784 Mon Jul 27 11:43:54 1998 Stan Cox <scox@cygnus.com>
1786 * longlong.h (count_leading_zeros): Sparclite scan instruction was
1787 being invoked incorrectly.
1789 * i386.c (ix86_prologue): Added SUBTARGET_PROLOGUE invocation.
1790 * i386/cygwin32.h (STARTFILE_SPEC, LIB_SPEC, SUBTARGET_PROLOGUE):
1792 * i386/win32.h: New file. Hybrid mingw32.h/cygwin32.h configuration.
1793 * configure.in: Added i[34567]86-*-win32
1794 * config.sub: Ditto.
1795 * configure: Rebuilt.
1797 Sun Jul 26 01:11:12 1998 H.J. Lu (hjl@gnu.org)
1799 * i386.h (CONST_DOUBLE_OK_FOR_LETTER_P): Return 0 when eliminating
1800 the frame pointer and compiling PIC code and reload has not completed.
1802 * i386.c (output_to_reg): Add code to emulate non-popping DImode
1805 Sun Jul 26 01:01:32 1998 Jeffrey A Law (law@cygnus.com)
1807 * regmove.c (regmove_optimize): Fix typo initializing regmove_bb_head.
1809 Sat Jul 25 23:29:23 1998 Gerald Pfeifer <pfeifer@dbai.tuwien.ac.at>
1811 * Makefile.in (install-info): Only try to update the info
1812 directory file if it exists in the first place.
1814 Fri Jul 24 18:58:37 1998 Klaus Espenlaub <kespenla@student.informatik.uni-ulm.de>
1816 * rs6000.h (ASM_OUTPUT_CONSTRUCTOR, ASM_OUTPUT_DESTRUCTOR): Delete.
1818 Fri Jul 24 14:20:26 1998 Jeffrey A Law (law@cygnus.com)
1820 * mn10300.md (movqi, movhi, movsi, movsf): Correctly handle
1821 CONST_DOUBLE source.
1823 Fri Jul 24 11:17:04 1998 Nick Clifton <nickc@cygnus.com>
1825 * config/arm/thumb.c (thumb_print_operand): Decode %_ in asm
1826 strings as the insertion of USER_LABEL_PREFIX.
1827 * config/arm/thumb.h (PRINT_OPERAND_PUNCT_VALID_P): Accept _ as a
1829 * config/arm/thumb.md: Use %_ as a prefix to gcc library function
1832 Thu Jul 23 18:53:20 1998 Jim Wilson <wilson@cygnus.com>
1834 * dbxout.c (dbxout_range_type): Only call dbxout_type_index for
1835 already defined type.
1837 Thu Jul 23 13:49:41 1998 Jeffrey A Law (law@cygnus.com)
1839 * expr.c (check_max_integer_computation_mode): Allow conversions
1840 of constant integers to MAX_INTEGER_COMPUTATION_MODE.
1841 (expand_expr): Likewise.
1843 Thu Jul 23 11:12:06 1998 Alexandre Petit-Bianco <apbianco@cygnus.com>
1845 * expr.c (expand_expr): Expand RETURN_EXPR.
1847 Thu Jul 23 11:00:29 1998 Jim Wilson <wilson@cygnus.com>
1849 * dwarf2out.c (dwarf2out_finish): Call stripattributes on TEXT_SECTION.
1851 Wed Jul 22 19:10:00 1998 Catherine Moore <clm@cygnus.com>
1853 * dwarf2out.c (output_aranges): Call stripattributes
1854 for TEXT_SECTION references.
1855 (output_line_info): Likewise.
1857 Wed Jul 22 14:08:54 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
1859 * profile.c (branch_prob): Call allocate_reg_info after outputting
1860 profile rtl in instrument_arcs.
1862 Wed Jul 22 12:47:49 1998 Jim Wilson <wilson@cygnus.com>
1864 * fixinc.irix (math.h): Install wrapper instead of copying.
1866 Wed Jul 22 12:37:14 1998 Alexandre Petit-Bianco <apbianco@cygnus.com>
1868 * tree.def (EXPR_WITH_FILE_LOCATION): Defined as an 'e' expression
1869 so WFL are expanded correctly when contained in a COMPOUND_EXPR.
1870 * tree.h (EXPR_WFL_EMIT_LINE_NOTE): Change macro not to use
1871 lang_flag_0. Added documentation in the flag table.
1873 Tue Jul 21 23:28:35 1998 Klaus Kaempf <kkaempf@rmi.de>
1875 * cccp.c (do_include): Fix vax c style include handling.
1877 Tue Jul 21 13:28:19 1998 Jason Merrill <jason@yorick.cygnus.com>
1879 * cplus-dem.c (do_type): Use demangle_template_value_parm for arrays.
1881 Sun Jul 12 01:27:05 1998 Jason Merrill <jason@yorick.cygnus.com>
1883 * fold-const.c (non_lvalue): Don't deal with null pointer
1885 (fold, case COMPOUND_EXPR): Wrap a constant 0 in a NOP_EXPR.
1887 Tue Jul 21 15:49:31 1998 David Edelsohn <edelsohn@mhpcc.edu>
1889 * rs6000.h (PREDICATE_CODES): Add CONSTANT_P_RTX.
1890 * rs6000.md (movsi, movdi): Add CONSTANT_P_RTX.
1891 * rs6000.c (short_cint_operand): Add CONSTANT_P_RTX.
1892 (u_short_cint_operand): Same.
1893 (reg_or_cint_operand): Same.
1894 (logical_operand): Same.
1895 (input_operand): Same.
1896 (reg_or_short_operand): Use u_short_cint_operand.
1898 Tue Jul 21 08:56:42 1998 Richard Henderson <rth@cygnus.com>
1900 * alpha.md (fix_truncdfsi2, fix_truncsfsi2): Remove the define_expands,
1901 but keep the insns and splits. Adjust so when the ultimate destination
1902 is memory, use cvtql.
1904 Tue Jul 21 08:55:09 1998 Richard Henderson <rth@cygnus.com>
1906 * flow.c (regno_uninitialized): Fixed regs are never uninitialized.
1908 Tue Jul 21 00:31:01 1998 Jeffrey A Law (law@cygnus.com)
1910 * gcc.c (do_spec): Call "error" not "warning".
1912 * configure.in: Fix minor problems with gas feature detection code.
1913 * configure: Rebuilt.
1915 * gcc.c (do_spec): Issue a warning for '%[]' usage.
1918 * gcc.c: Delete %[spec] support.
1919 (do_spec_1, case '('): Likewise.
1920 (do_spec_1, case '['): Call error.
1922 Mon Jul 20 22:34:17 1998 Richard Henderson <rth@cygnus.com>
1924 * alpha.h (CPP_SPEC): Tidy. Hook to cpp_cpu and cpp_subtarget.
1925 (CPP_SUBTARGET_SPEC): Default to empty string.
1926 (CPP_AM_*, CPP_IM_*, CPP_CPU_*, CPP_CPU_SPEC): New.
1927 (EXTRA_SPECS, SUBTARGET_EXTRA_SPECS): New.
1928 * alpha/elf.h (LD_SPEC): Use %(elf_dynamic_linker).
1929 * alpha/linux-elf.h (SUBTARGET_EXTRA_SPECS): New.
1931 * alpha/linux.h (CPP_PREDEFINES): Tidy.
1932 * alpha/netbsd-elf.h (SUBTARGET_EXTRA_SPECS): New.
1933 * alpha/netbsd.h (CPP_PREDEFINES): Tidy.
1934 * alpha/osf.h (CPP_PREDEFINES): Remove bits subsumed by CPP_CPU_SPEC.
1935 * alpha/win-nt.h (CPP_PREDEFINES): Likewise.
1936 * alpha/vsf.h (CPP_PREDEFINES): Likewise.
1937 (CPP_SUBTARGET_SPEC): New. Do this instead of overriding CPP_SPEC.
1938 * alpha/vxworks.h: Likewise.
1940 Mon Jul 20 22:51:57 1998 Ken Raeburn <raeburn@cygnus.com>
1942 * mips.md (reload_outsi): Added missing REGNO call.
1943 (smulsi3_highpart, umulsi3_highpart): Provide prototype for
1945 (mul_acc_di, mul_acc_64bit_di): Don't use match_op_dup, use
1946 another match_operator and compare the codes.
1948 * mips.h (MASK_DEBUG_E, MASK_DEBUG_I): Set to zero.
1950 * MIPS multiply pattern fixes:
1951 * mips.h (enum reg_class, REG_CLASS_NAMES, REG_CLASS_CONTENTS):
1952 Add union classes for HI, LO, or HILO plus general registers.
1953 (GENERATE_MADD): Deleted.
1954 * mips.md (mulsi3_mult3): Don't disparage output-LO alternative.
1955 Add TARGET_MAD to condition.
1956 (mulsi3): Test HAVE_mulsi3_mult3, not specific flags.
1957 (mul_acc_si): Expand GENERATE_MADD here; it's the only use. Use
1958 "*d" for accumulator, to give preference to LO initially but not
1961 Mon Jul 20 16:16:38 1998 Dave Brolley <brolley@cygnus.com>
1963 * configure.in (enable_c_mbchar): New configure option.
1964 (extra_cpp_objs): Always available now.
1966 * cexp.y (mbchar.h): #include it.
1967 (yylex): Handle Multibyte characters in character literals.
1969 * cccp.c (mbchar.h): #include it.
1970 (main): Set character set based on LANG environment variable.
1971 (rescan): Handle multibyte characters in comments.
1972 (skip_if_group): See above.
1973 (validate_else): See above.
1974 (skip_to_end_of_comment): See above.
1975 (macarg1): See above.
1976 (discard_comments): See above.
1977 (rescan): Handle multibyte characters in string and character literals.
1978 (collect_expansion): See above.
1979 (skip_quoted_string): See above.
1980 (macroexpand): See above.
1981 (macarg1): See above.
1982 (discard_comments): See above.
1983 (change_newlines): See above.
1985 * c-lex.c (mbchar.h): #include it.
1986 (GET_ENVIRONMENT): New macro.
1987 (init_lex): Set character set based on LANG environment variable.
1988 (yylex): Handle multibyte characters in character literals.
1989 (yylex): Handle multibyte characters in string literals.
1991 * Makefile.in (mbchar.o): New target.
1992 (cccp$(exeext)): @extra_cpp_objs@ is always available.
1993 (cppmain$(exeext)): @extra_cpp_objs@ is always available.
1995 * mbchar.[ch]: New files for multibyte character handling.
1997 Mon Jul 20 01:11:11 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
1999 * jump.c (jump_optimize): When simplifying noop moves and
2000 PUSH_ROUNDING, fix thinko so we use same criterion for identifying
2001 the PUSHes to rewrite in second loop as we did in the first.
2003 Sun Jul 19 08:23:53 1998 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
2005 * cplus-dem.c (demangle_nested_args): Make function definition
2006 static to match the prototype.
2008 Fri Jul 17 14:58:44 1998 Richard Henderson <rth@cygnus.com>
2010 * alloca.c: Respect USE_C_ALLOCA.
2011 * gencheck.c (xmalloc): Ignore __GNUC__ for definition.
2012 * gengenrtl.c (xmalloc): Likewise.
2014 Fri Jul 17 14:18:14 1998 Richard Henderson <rth@cygnus.com>
2016 * loop.h (struct induction): Add no_const_addval.
2017 * loop.c (the_movables, reg_address_cost): New variables.
2018 (init_loop): Init reg_address_cost.
2019 (loop_optimize): Call end_alias_analysis.
2020 (scan_loop): Init the_movables.
2021 (record_giv): Init induction->no_const_addval.
2022 (basic_induction_var) [PLUS]: Use rtx_equal_p instead of ==.
2023 [REG]: Rearrange loop search test to catch more cases.
2024 (general_induction_var): Return success not benefit; take an extra
2025 argument for that. Change all callers.
2026 (simplify_giv_expr) [PLUS]: Always combine invariants. Use sge_plus.
2027 [MULT]: Use rtx_equal_p instead of ==. Combine simple invariants.
2028 [default]: Search the_movables for additional combinations.
2029 (sge_plus_constant, sge_plus): New functions.
2030 (express_from_1): New function.
2031 (express_from): Always define. Rewrite using express_from_1.
2032 (combine_givs_p): Handle more cases. Ignore address cost.
2033 (cmp_combine_givs_stats): New function.
2034 (combine_givs_used_once, combine_givs_benefit_from): New functions.
2035 (combine_givs): Rewrite to do best-fit combination.
2037 * fold-const.c (operand_equal_p): Handle RTL_EXPR.
2038 (fold): Do a complete (A*C)+(B*C) association check.
2040 Fri Jul 17 11:21:55 1998 Jim Wilson <wilson@cygnus.com>
2042 * function.c (fixup_var_refs_insns): Handle CLOBBER of a CONCAT.
2044 Fri Jul 17 11:48:55 1998 Jeffrey A Law (law@cygnus.com)
2046 * mn10300.c (MODES_TIEABLE_P): Fix typo.
2048 Fri Jul 17 03:26:12 1998 Rihcard Earnshaw (rearnsha@arm.com)
2050 * tree.c (valid_machine_attribute): Only create a new type variant if
2051 there is a decl to use it.
2053 Thu Jul 16 14:48:04 1998 Nick Clifton <nickc@cygnus.com>
2055 * gcc.c (do_spec_1): Cope with %g/%u/%U options which do not have
2058 Fri Jul 17 03:24:40 1998 Hans-Peter Nilsson <hp@axis.se>
2060 * extend.texi (Explicit Reg Vars): Typo: change "may deleted" into "may
2063 Thu Jul 16 14:48:47 1998 Jeffrey A Law (law@cygnus.com)
2065 * mn10300.c (count_tst_insns): New arg oreg_countp. Callers changed.
2066 Simplify tests for clearing an address register.
2067 (expand_prologue): Corresponding changes.
2069 * mn10300.md (movXX patterns): Make sure the destination is an
2070 ADDRESS_REG when substituting "zero_areg" for (const_int 0).
2071 (logical patterns): Split into expanders + patterns
2072 (zero and sign extension patterns): Similarly.
2073 (shift patterns): Similarly.
2075 Thu Jul 16 01:17:44 1998 Richard Henderson <rth@cygnus.com>
2077 * loop.c (emit_iv_add_mult): Scan the entire insn list generated
2078 for the sequence, recording base values.
2080 Wed Jul 15 10:49:55 1998 Richard Henderson <rth@cygnus.com>
2082 * i386.h (CPP_CPU_SPEC): Remove -Asystem(unix).
2084 Tue Jul 14 14:15:30 1998 Nick Clifton <nickc@cygnus.com>
2086 * gcc.c: Remove ANSI-C ism from --help code.
2088 * toplev.c: Support --help with USE_CPPLIB.
2090 Tue Jul 14 14:46:08 1998 Jeffrey A Law (law@cygnus.com)
2092 * configure.in: Rework gas feature code to work with symlink based
2095 * extend.texi: Clarify some issues related to local variables
2096 assigned to explicit registers.
2098 * mn10300.md (mulsi): Turn into expander + pattern.
2100 * mn10300.md (movsi, movsf, movdi, movdf): Remove "x" from I -> a
2103 Tue Jul 14 07:41:59 1998 Richard Earnshaw (rearnsha@arm.com)
2105 * arm/tcoff.h (USER_LABEL_PREFIX): Make it empty to match coff.h.
2107 Tue Jul 14 03:02:44 1998 Jeffrey A Law (law@cygnus.com)
2109 * version.c: Bump again to distinguish mainline tree from the
2112 See ChangeLog.0 for earlier changes.
2115 add-log-time-format: current-time-string