![[BACK]](/cvsweb/icons/back.gif){kind=icon}
Return to Log CVS log ![[TXT]](/cvsweb/icons/text.gif){kind=icon}
![[DIR]](/cvsweb/icons/dir.gif){kind=icon}
Up to [MW Coherent from dump] / coherent / b / STREAMS / i386|
Return to Log CVS log | Up to [MW Coherent from dump] / coherent / b / STREAMS / i386 |
1.1 root 1: Changes to kernel and related files for version #80:
2:
3: NOTE: The SCSI-problems reported by Nigel in R79 seem to be related to
4: hardware problems on his 486, as the HAI driver fails similarly on that
5: hardware, and the AHA drivers works fine everywhere else.
6:
7: The following headers have been deleted:
8: access.h, action.h, disp.h, ebcdic.h, larges.h, timef.h
9: sys/ascii.h, sys/fun.h, sys/chars.h, sys/dir.h sys/malloc.h
10:
11: The following headers have been moved to /usr/include/kernel, and are no
12: longer accessible to user-level code (more will follow):
13: sys/alloc.h, sys/const.h, sys/fakeff.h, sys/machine.h, sys/param.h,
14: sys/reg.h, sys/typed.h
15:
16: The above header changes plus the following changes from Nigel have required
17: global modification to almost every single kernel module.
18: The global u.u_regl in the U area has been removed, requiring vast
19: changes to system-call handling, signalling, and subtle changes to
20: much system-call code, in addition to changing the U area.
21:
22: The stack-access constants from <kernel/reg.h> moved to <ieeefp.h>,
23: and are no longer used in the kernel.
24:
25: The u.u_args global was removed from the U area, requiring further
26: changes to system calls and signalling.
27:
28: The u.u_direct and u.u_io global structures removed from the U area,
29: requiring changes to all filesystem-related system calls, and
30: considerably changing the calling protocol of ftoi ().
31:
32: 08/01 curses.h: Colour modifications.
33: 08/01 sys/patch.h: Added, ioctl () interface to patch driver.
34:
35: Before R80 was turned over to Nigel:
36: 07/19 haiscsi.h,haiinst.h,hai154x.c,haicfg.c,haict.c,haisd.c,haiscsi.c:
37: revision 2 from Chris Hilton
38: 07/19 rlock.c: return EACCES, not EAGAIN when locking fails.
39: 07/19 reg.h,as.s: make setspace() stuff part of saved context.
40: 07/19 sys5.c: io_seg was uninitialized in dirio.
41: 07/19 sys2.c: upoll() needed to use fdget().
42:
43: Changes to kernel and related files for version #79:
44:
45: NOTE (Nigel): The outstanding issue is the [pre-Chris-Hilton] Adaptec
46: driver; switching virtual terminals (or sometimes, doing console printf
47: ()'s while debugging) cause I/O requests to fail mysteriously and/or go
48: into oblivion (causing deadlocks); it is not clear whether this is
49: caused by the console drivers, or is a race condition exacerbated by
50: them.
51:
52: 07/16 timeb.h,sys1632.c: ftime() is again a system call. oldtimeb.h
53: is obsolete.
54: 07/16 Chris Hilton's Adaptec (hai) SCSI disk and tape driver
55: 07/16 as.s, trap.c: hlt instruction in reboot no longer causes gp fault
56: 07/16 kb.c,kb_d.c,kb_f.c,nkb.c,vtkb.c,vtkb_d.c,vtkb_f.c,vtnkb.c -
57: X11 mods from RTR.
58: 07/16 kb.c,kb_d.c,kb_f.c,nkb.c,vtkb.c,vtkb_d.c,vtkb_f.c,vtnkb.c -
59: Function key update (Word Perfect compatibility?) from Louis.
60: 07/16 fl386.c - New floppy minor devices skip cylinder 0, autosense format.
61: 07/16 syscoh.c - int11 now available to user code for installation (Louis)
62: 07/12 fs1.c,sys[235].c: exec perms for root on directories fixed (Vlad)
63: 07/12 msg386.c: Pemission fix from Vlad
64: 07/12 Nigel: Signal handling (almost) totally rewritten, and is now
65: built around "sigaction" structures rather than an array of
66: pointers and legion bitmasks... all code affecting signal state
67: properly centralized, requiring numerous tiny driver mods.
68: 07/12 Nigel: Worst of the u.u_regl stupidity removed, fixing nasty
69: signalling bugs.
70: 07/12 Nigel: Limits on the user code and data chopped to 0x7FFFFFF
71: rather than all of linear space; this exposed numerous small
72: problems in e.g. ustime () where kernel data was being treated as
73: user data via kucopy ()/ukcopy ().
74: 07/12 Nigel: Everything to do with user-kernel copying (mainly in as.s
75: and fakedma.c) changed; now %es == %ds in the kernel, which is
76: vital for using normal C library routines, most important of
77: which are the GCC inlines. Now, %fs in the kernel holds an RPL 3
78: user data selelector, and %gs is scratch. Much simplified, these
79: should run much faster now (fakedma.c is now about half the size
80: it was).
81: 07/12 Nigel: The bogus kclear () in kalloc () (actually, alloc () in
82: coh.386/alloc.c) removed, exposing numerous problems where kernel
83: data structures were not fully initialised, including the buffer
84: cache, in-core inode table, SCSI requests in io.386/scsi.c and more.
85: 07/12 Nigel: The bogus memory-clear at startup has been removed, with
86: no noticeable effect other than exacerbating some of the problems
87: mentioned above, which have hopefully all been fixed.
88: 07/12 Nigel: <sys/ktty.h> defined numerous predicates which dug bits
89: out of tty control structures by implicitly referencing local
90: variables of the routines where they were embedded via magic
91: names. All the predicates have sensible names and take parameters
92: now, which meant changing the tty code and every function that
93: called it to use the sane predicates, including the serial
94: drivers and the legion *kb* "modules".
95: 07/12 Nigel: Removed dead code and bogus FFCOPY () from io.386/vt*kb*.c
96: 07/12 Nigel: selkcopy () removed from trap.c (changed to ffbyte ()
97: instead), and the SDUMP ()/RDUMP () hacks changed to use a real
98: routine that uses a real register-set structure (created for the
99: signalling work) and can do proper stack backtracing in the
100: kernel.
101: 07/12 Nigel: Several small changes to coh.386/fs[23].c aimed at
102: increasing performance, along with a change to the locking in
103: bclaim (), all to negligible effect on the results as reported by
104: iozone ().
105: 07/12 exec.c,fd.c: close on exec bug fixed (Nigel)
106: 07/12 tty.c,pty.c,pipe.c: return 0 if O_NDELAY and -1/EAGAIN if
107: O_NONBLOCK when no input available (Nigel)
108: 07/01 sig.c: fail with EFAULT if trying to access nonexistent NDP state
109: 07/01 shm.h,shm0.c,shm1.c: fixed define/int conflict with SHMMAX
110: 07/01 sig.c: botched variable in sigdump() broke core file writes
111: 06/17 fl386.c,fdc.c,fdc765.h: change CMD... to FDC_CMD_...
112: 06/14 as.s,xdt.s: call gate for hlt instruction to fix reboot panic
113:
114: Changes to kernel and related files for version #78:
115:
116: NOTES:
117: New libc.a needed for support of nap() system call.
118: New kbmain.o is needed, (just recompile current kbmain.c) and new
119: /conf/kbd/us, etc as well, due to change in kb.h (or else you won't
120: be able to switch sessions with virtual consoles).
121: New db needed from Steve since uproc.h changed.
122: File name changes in kernel file set:
123: lp.c is replaced by lp386.c
124: fl.c is replaced by fdc.c, fl386.c, and <sys/fdc765.h>.
125:
126: 06/10 fl386.c: some open failures left diskette drive locked
127: 06/04 sys2.c, var.c: add nap() system call
128: 06/03 exec.c: vlad's fix - kernel blew up in pexece() when out of memory
129: 05/26 uproc.h: change u_error from char to int so errnos > 127 come out ok
130: 05/24 xl.h,xlfdc.h,xlft.h,xlft.c,xl_dec.c,xla.s: Archive floppy tape driver
131: 05/24 seg.h,shm0.c,shm1.c: vlad - allow deferred removal of live shm segments
132: 05/24 as.s: increase i/o map upper limit (3ff->1fff) for newer vga cards
133: 05/19 mmu.c: add getDmaMem for aligned contiguous memory allocation
134: 05/18 sig.c: patchable variable CATCH_SEGV for nonportable software
135: 05/18 sig.c: signal to traced process wrongly caused core dump
136: 05/14 sys3.c: read on file set for APPEND should not be forced to end
137: 05/14 mmu.c: growing segments beyond 4 megabytes caused a panic
138: 05/14 kb.h: Add lots more function keys; change vt keydefs. (mlk)
139: 05/11 sig.c, signal.h, msig.c: sigpause() and reliable signals
140: 05/11 sys1.c: upause() should set EINTR.
141: 05/11 sys1.c: part of waitpid(), done by Michael
142: 05/07 lp386.c: fix interrupt mask and wakeup problems
143: 05/07 tty.c: fix sign extension bug in vmin/vtime handling
144: 05/07 fl386.c: delete scratch_buffer code - had ugly bug [mlk found this]
145: 05/07 fl386.c,fdc.c,fdc765.h: split floppy driver for floppy tape
146: 05/03 ndp.c: delete "Bad Em write..." printf, just do SIGSEGV
147: 04/27 tty.c: change defer(wakeup,...) to wakeup(...) since all wakeups
148: are deferred. This should cut down on defer overflows.
149: 04/27 sig.c: set SRFDUMP flag for all dumped segments
150: 04/27 move acct.h from /usr/include/ to /usr/include/sys
151: 04/26 reg.h,md.c: setivec return status added, for floppy tape support.
152: 04/20 fl.c: another FL_DSK_CH_PROB change from Michael
153:
154: Changes to kernel and related files for version #77:
155:
156: 04/19 proc.h,sem.h,proc.c,sem386.c: sem_undo and other semaphore final test
157: changes from Vlad.
158: 04/20 fl.c: FL_AUTO_PARM added by Michael for PS/2-L40 compatibility.
159:
160: Changes to kernel and related files for version #76:
161:
162: NOTES:
163: "dcore" utility groks r75 and later core files
164: "dlout" utility is like "cdmp" for l.out's
165: --
166: 04/15 fakedma.c: dmaout() sometimes wrote garbage over the start of a click
167: 04/14 sig.c: DUMP_TEXT if patched causes text segment to appear in core files
168:
169: Changes to kernel and related files for version #75:
170:
171: NOTES:
172: New "file" command groks fixed core file format.
173: New kbmain.o needed for Greek keyboards.
174: New db needed from steve (later than 93/04/08) for revised core file format.
175: --
176: 04/08 sem386.c,Makefiles,bld: vlad's new, improved semaphores
177: 04/08 sig.c: put magic number back at start of core file
178: 04/08 vtnkb.c,kbmain.c,kb.h: Greek keyboard, with patchable variable VTGREEK
179: 04/07 various headers: the definitive (!) paddr_t cleanup
180: 04/07 ndpas.s: change fdiv to fdivr to reflect change in assembler
181: 04/07 sys/kb.h: added function key definitions from mlk
182: 04/07 fl.c: further PS/1 fixes from mlk
183: 03/22 exec.c: exsread(): do only one iread() per section
184: 03/20 a.out.h,exec.c: allow more than 1 text section for coff executables
185: 03/18 var.c: change argument count for uwait() from 0 to 3 for waitpid()
186: 03/18 exec.c: delete redundant kkcopy()
187: 03/18 aha.c: nigel's fix to prevent defer overflows
188: 03/18 sys/__paddr.h added to header suite
189:
190: Changes to kernel and related files for version #74:
191:
192: NOTES:
193: --
194: 03/10 fl.c: workaround for incorrect detection of nspt on some 3-1/2" drives
195: 03/05 aha.c: nigel's interrupt fix
196: 03/05 fs1.c: verbose panic output in case idetach panic
197: 03/05 fs2.c: once again, restore verbose inode busy diagnostic
198: 03/04 alloc.c: verbose panic output in case of bad free
199:
200:
201: Changes to kernel and related files for version #73:
202:
203: NOTES:
204:
205: 1. You will need a new libc.a to use waitpid() and chsize().
206: 2. There is a new fsck from epstein.
207: 3. "putq()" and "getq()" from prior kernels are now "cltputq()" and "cltgetq()"
208: 4. sleep() and v_sleep() have been superseded by x_sleep() - see the comments
209: in proc.c for argument list and return values.
210: 5. new ps needed
211:
212: 03/01 sys1.c,proc.c,sig.c: numerous fixes to sleep/wakeup/wait().
213: waitpid() is still a stub.
214: 02/23 fs1.c,sys2.c,sys5.c,timeout.c,var.c,fs2.c,pipe.c,sys3.c,atcon.c,
215: trap.c,usswan.c,vtnkb.c: epstein's file system changes - pipe fix,
216: chsize(), enforce ulimit on file size...,allow Swan kb support
217: by patching VTSWAN=1.
218: 02/18 types.h et al - Nigel replaced vaddr_t with caddr_t and made other
219: header changes in preparation for STREAMS.
220: 02/17 fd.c: Nigel's fix - open was creating files even when it failed
221: 02/12 tty.c: make sure VEOF is initialized properly on first opens.
222: 02/10 sys2.c: allow upoll() to succeed when number of fd's is 0
223: 02/09 seg.h,mmu.c,shm0.c: allow individual shm seg references to be RW/RO
224: 02/04 reg.h,con.h,bio.c,fs2.c,fs3.c,sys3.c,sys5.c,md.c,as.s,clock.c: Nigel's
225: fixes (1) call STREAMS at key places via macros; (2) add extra
226: parameters to dopen()/dclose()/dioctl(); (3) support mask register
227: for interrupts.
228: 02/04 shm1.c: don't allow attempts to create shm segments of size 0
229: 02/02 fs2.c: restore verbose inode busy diagnostic
230: 02/02 as.inc,as.s,clist.c,tty.c: replace getq() and putq() with cltgetq()
231: and cltputq() since the old names conflict with standard STREAMS
232: externals.
233: 02/02 coherent.h,proc.c,rlock.c,alx.c,asy.c,dmareq.c,hs.c,lf.c,msg386.c,sem.c:
234: stop passing 2nd and 3rd magic arguments to sleep as they were always
235: ignored. sleep() vanishes; v_sleep() becomes a macro calling
236: w_sleep(char * event, int cl, char * msg)
237: 01/22 sys1.c: utimes() - check for null or otherwise invalid user pointer
238: 01/22 fs1.c: imode() - access() as superuser returns exec perm set only
239: if the file is executable.
240: 01/22 fs3.c: iopen() - a non-superuser trying to write to a directory
241: gets EISDIR, not EPERM.
242: 01/22 tty.c: sgtty-type TIOCSETx ioctl's no longer set ISTRIP.
243: 01/22 sys1.c: add interface for waitpid(); fix uwait return value for
244: stopped processes (i.e., processes that hit a breakpoint)
245:
246: Changes to kernel and related files for version #72:
247:
248: NOTES:
249:
250: 1. You will need an upgraded libc.a or at least shmemul.o to use S5
251: shared memory.
252: 2. You will need tboot 1.2.6 or later if you are using FP emulation.
253: 3. Epstein has given us another fsck.
254:
255: 01/17 sys5.c: sysi86(SI86FPHW, &data) supported to get ndp type
256: 01/17 sig.c: ptrace can examine modify fp state in emulator
257: 01/14 fs2.c,fs3.c,sys2.c,pipe.c,sys/inode.h: epstein's pollable pipes
258: 01/14 main.c,at.c,fl.c: michael's PS1 mods
259: 01/11 ndp.c,mmu.c,bld: hooks for linking Linux/cef floating point emulator
260: 12/29 new shm1.c from vlad, after testing preliminary ipcs
261: 12/29 mmu.c: fix bug allocating ram disk slots in page directory
262: 12/29 mmu.c,ndp.c,uproc.h: fp emulator state save/restore
263: 12/23 typed.h and .c files containing it: remove use of T_NULL which
264: conflicts with use of that identifier in coff.h
265: 12/23 rm.c,mmu.c: move start of RAM disk to virtual 0x8800_0000 for
266: compatibility with shm as assigned on other systems.
267: 12/23 shm1.c,sys5.c,shm.h,Makefiles,bld: S5 shared memory
268: 12/23 mmu.c: leave room for ndp state in user stack, per cef's request
269: 12/21 fs1.c,fs2.c: epstein's fix to inode busy bug
270:
271: Changes to kernel and related files for version #71:
272:
273: 12/18 fd.c, fs1.c, fs3.c, pipe.c, sys2.c, sys3.c: merge epstein's pipe
274: fixes.
275: 12/17 shm0.c: minor cleanup, add shmAtt() for vlad
276: 12/10 sys[235].c,tty.c: enhanced useracc(), ttioctl no longer allowed to
277: overwrite kernel text
278: 12/10 ndp.c: fix interrupt 13 attachment in case of 287
279: 12/10 shm0.c: add shmDetachP() in support of ipcrm et al
280: 12/10 syscoh.c(): add cohcall(COH_WTEXT,dest,src,numBytes) so a process can
281: write into its text segment. This is mainly for cef's NDP emulator
282: thunks.
283: 12/10 sys5.c,atcon.c,main.c,uproc.h: add code for ulimit(); add kernel
284: patchable variable BPFMAX = max # of 512 byte blocks when writing
285: a file. Warning: ulimit(2,..) is not yet enforced! ulimit(3) returns
286: the maximum possible break value (with 64 kbytes of cushion).
287: 12/09 sys1632.c: make 286 stime() work as before
288: 12/09 sys5.c: fix return values from ustatfs(),ufstatfs()
289: 12/08 null.c,as.s,work.c,fakedma.c,ndp.c,shm0.c,as.inc,mmu.h: replace
290: use of WORK0 and WORK1 with calls to workAlloc() and workFree().
291: 12/07 sig.c,ndp.c: allow ptrace() to transfer NDP state
292: 12/04 sys1.c: stime() gets value, not pointer (undocumented part of BCS
293: compatibility)
294: 12/02 shm0.c,bio.c,sys3.c,proc.c,syscoh.c,mmu.c,exec.c: kernel support
295: for shared memory
296:
297: Changes to kernel and related files for version #70:
298:
299: 11/30 sys2.c,pty.c: fix typo in POLLOUT polling
300: 11/25 ktty.h,tty.c: XTABS under sgtty becomes TABDLY=TAB3 under termio
301: 11/25 sys1.c: ulseek() once again rejects absolute offsets outside
302: the range 0..0x7ffff_ffff.
303: 11/25 null.c: /dev/kmemhi
304: 11/25 null.c: can now read/write /dev/mem more than 4096 bytes
305: 11/25 tty.h,ktty.h,asy.c,*kb*.c: support IXANY
306: 11/24 tty.c: block before first character if -icanon, vmin > 0.
307:
308: New /bin/stty, accepts decimal min/time values, ^? for DEL, ^- for <undef>.
309: New /conf/patch needed since we can no longer seek past 0x7ffff_ffff!
310:
311: Changes to kernel and related files for version #69:
312:
313: Warning: old COHERENT 286-style (was never part of S5) TIOC[C]BREAD is
314: no longer supported. It conflicts with termio-style VMIN/VTIME. Norm
315: said it was ok to drop this.
316:
317: 11/20 fl.c: fix at least one endless retry condition
318: 11/20 asy.c: setting BAUD 0 now hangs up the line
319: 11/19 asy.c: TCSBRK now sends 0.25 second break
320: 11/19 tty.c,asy: TCSETAF,TCSBRK weren't draining output
321: 11/19 tty.c: TCFLSH was ignoring third argument
322: 11/18 sys1.c: fix improper u_error setting in setuid()/setgid()
323: 11/17 mmu.c: fix startup arithmetic bug
324: 11/17 misc.c: use strchirp() in panic() if paging not enabled
325: 11/17 as.s: fix comparison in paging()
326: 11/12 mmu.c: add DV() macro for debugging
327: 11/12 asy.c: drain port fully on TCSETAW/TIOCSETP
328: 11/12 tty.c: add support for OCRNL, VMIN, VTIME
329:
330: Changes to kernel and related files for version #68:
331:
332: New /bin/ps needed!
333: setfpe no longer supported.
334:
335: 11/10 mmu.c,md.c,syscoh.c,as.s: add window server support (iomapOr(),
336: iomapAnd, kiopriv(), mapPhysUser()).
337: 11/10 msg386.c: vlad's permission 0 fix.
338: ... ker 67 was in here somewhere but had fatal character processing bug...
339: 11/06 null.c: improved ps command, uses new /dev/ps
340: 11/04 sys/ktty.h,tty.c: fix ONLCR/ICRNL check in tty module
341: 11/04 at.c,misc.c,as.s: atbsyw() and atdrqw() no longer depend on CPU speed
342: 11/04 sys/ktty.h,*kb*.c: fix ctrl-s/ctrl-q test to check IXON flag
343: 10/29 sys1.c: fix utimes() return value.
344: 10/28 287 support (untested)
345: 10/25 Fix floating point exception handling for 387. Add NDP sensing.
346:
347: Changes to kernel and related files for version #66:
348:
349: New /bin/ps needed!
350:
351: NOTICE: /usr/include/sys/dirent.h and /usr/include/sys/dir.h are
352: obsolete and should be removed from the /usr/include/sys directory!
353: All references to them should include /usr/include/dirent.h instead.
354:
355: 10/19 trap.c - stub for floating point exceptions
356: 10/20 as.s,mmu.h,uproc.h - reserve space just below u area for NDP state
357: 10/20 trap.c - device not available trap rewritten for NDP
358: 10/22 floating point context switching, exceptions, and signal handlers
359:
360: Changes to kernel and related files for version #65:
361: 10/08 fl.c - fix open/close count imbalance
362: 10/11 as.s,trap.c - fix runaway resulting from bad user pointer in system call
363:
364: Changes to kernel and related files for version #64:
365: New /etc/reboot (source in ker386/tools).
366: 09/24 sig.c: ptrace fix, going from single step to full speed
367: 10/01 trap.c: revise retry policy for iret bug
368: 10/01 syscoh.c,param.h: superuser-callable reboot function
369: 10/02 as.s: revise IODELAY to solve install problems on some systems
370: 10/02 syscoh.c: update breakpoint code for new trap handler
371: 10/02 msg386.c,sys2.c: vlad's new msg handler (no longer a device driver)
372: 10/02 die.c: shorter code for print8/16/32, without need for data seg.
373: 10/05 mmu.c: add PHYS_MEM and getPhysMem() for NCM and others (allocate
374: blocks of contiguous physical memory).
375: 10/06 fl.c: flopen() would occasionally hang the system
376: 10/06 as.s,as.inc,xdt.s: mmu update call gate to speed interrupt response
377:
378: Changes to kernel and related files for version #63:
379:
380: 09/08 sys5.c: fix permission problem in mkdir()
381: 09/08 fl.c: fix end-of-volume problem
382: 09/08 uproc.h, sys1.c, exec.c: S5-compatible setuid()/setgid()
383: 09/09 coherent.h,misc.c: replace kkcopy(), kclear() with macro calls to
384: memcpy and memset.
385: 09/09 tty.c: TCSETAW was losing buffer contents; add ttrtp() call.
386: 09/10 seg.c: vlad's fix for recursive exec's
387: 09/16 bio.c,tioc.c,al.c,hs.c,asy.c,pty.c,*kb*.c: get rid of tioc286()
388: kluge in device drivers; move tioc() call into dioctl()
389: 09/17 null.c: /dev/proc for leaner ps, courtesy of La Monte.
390: 09/17 uproc.h,md.c,sig.c,sys1.c,trap.c: fixes for db support
391: 09/22 ptrace.h: add constants for ptrace() commands (no BCS standard for these)
392: 09/22 fs2.c: Steve's conjectured fix for inode busy
393: 09/22 as.s,xdt.s,as.inc: further changes for db support
394: 09/24 fdioctl.h,sdioctl.h,mtioctl.h,buf.h,fl.c: clean up ioctl name space junk
395:
396: Related to kernel changes above:
397: new bin/ps must be used due to yet another change in u area
398: watch for new cron from vlad that doesn't spawn lots more crons
399: new db has at least some working features
400:
401: still to do: ptrace(), floating point
402: clean up __cinit()/__putchar() stuff
403:
404: Changes to kernel and related files for version #62:
405:
406: 08/28 memset.s: yet another segment botch fixed (sorry!)
407: 08/31 fl.c: Jack's improved floppy driver; much quicker on dd; warns on
408: attempting to mount write-protected floppy for read/write.
409: 09/01 stdio.h,sys/param.h,sys/uproc.h,sys/select.h,libc.a,misc.tar.Z
410: (bedaemon.c, select.c),bin/ps,exec.c,fd.c,sys2.c,sys1632.c: up
411: to 60 open files per process. If you are building kernels from
412: #62 sources, best rm ALL older .o files and start over due to
413: change in value of NOFILE in sys/param.h.
414: 09/01 conf/kgen - allows you to link a new kernel with same configuration
415: as your original 4.0.0/4.0.1 installation. This script is NFG on
416: systems not built by doing a full COH install of 4.0.0/4.0.1. Its
417: main purpose is to allow developers & customers to replace one or
418: more kernel binary files, then make a new kernel with all the right
419: drivers and patches for their systems.
420:
421: Changes to kernel for version #61:
422:
423: 08/18 new memset.s,memcpy.s
424: 08/21 vtkb.c - vlad
425: 08/21 null.c - fix bug with /dev/null input
426: 08/24 vtkb_d.c,vtkb_f.c - vlad
427: 08/25 trap.c,as.s: fix IRET GP fault bug
428: 08/25 trap.c: separate GP fault handler
429: 08/25 trap.c,as.s,syscoh.c,param.h: kernel breakpoints
430:
431: Changes to kernel for version #60:
432:
433: 08/05 as.s: set IOPL to 1, not 3. Requires new /etc/ATclock.
434: 08/05 xdt.s: source cleanup
435: 08/05 mmu.h,as.inc: add SEG_RNG0_STK, SEG_RNG0_TXT, SEG_RNG1_STK
436: 08/14 null.c: set seek limit of 256 on /dev/cmos. Add /dev/clock.
437: 08/18 as.s,as.inc,xdt.s,mmu.h,mmu.c,vtnkb.c,mmas.s: Ring 1 kernel (use new ps)
438: 08/18 mmu.c: replace zero_fill() with memset()
439:
440: Changes to kernel for version #59:
441:
442: 07/24 mmu.c: zero_fill now uses register variables and long* for speed
443: 07/20 kb_d.c, kb_f.c: fix Alt-Gr bug.
444: ????? limit.h, coff.h cleaned up by steve
445: 07/24 typed.h,mmu.h fixed
446: 07/24 mmu.c: vaddr() checks MAX_VADDR; round CMOS RAM amt down to click bdry
447: 07/24 die.c: hex output
448: 07/24 main.c: t_piggy stuff
This archive runs on limited infrastructure. Preserving old code on modern bandwidth. Automated agents are requested to crawl responsibly.