bob@kahala.hig.hawaii.edu (Bob Cunningham) (12/15/88)
We received SunOS4.0.1 on Thursday, 1 December; I'm in the process of installing it on a series of various-model Sun3s here. It basically just a collection of bug fixes for SunOS4.0. It does not require a complete re-install of SunOS, but does require taking your machine down to single-user mode, running the tape and selecting which fixes you want (no, you cannot simply apply them all, the documentation says "this is for your protection"), rebuilding your kernal (including client kernals if you're installing on a server), and rebooting. One fix require sinstalling a new disk boot block before you reboot. In addition to reading the "READ ME FIRST" document that comes along with the tape, you will almost certainly want to use tar to copy off the README file on the tape itself and go through that carefully before you start the actual installation. It describes exactly what files are changed, and what each change fixes. If you run SunOS4.0 you definitely will want to install at least some of the fixes, especially the ones that fix "interesting" potential security loopholes. I'm installing all the fixes on all of our machines here. Here is a quick overview of most of the fixes for Sun3s: assertion failure panics some simple automount fixes bus eror watchdog resets several fixes to dbx format [also see note in READ ME FIRST about using this with MUNIX or from the miniroot!] fpaload fix for infinite loops in graph kernal crashes with streams NIT and IPC board kbdriver fixes fixing kadb ldcore fixes [this requires installing new boot block BEFORE reboot] fix for using kernals larger than 1MB various libc patches fix for security problem in login [you want this] lwp library linking fixes mb/mbugo fixes [kernal stuff] mmpeekio [more kernal stuff] various useful nfs fixes opreg [C float-to-double conversions] several rpclockd fixes an obscure fix to sa sdsci device driver fixes lots of serial driver fixes, including some ALM-II fixes sky fixes sysdiagp fixes TARGET fixes udpcksum/nfs fix uuxqt fix winlibs fixes ypbind fixes yppasswd fixs [security stuff you want] yppush Notable omissions include: None of the sendmail, fingerd, etc. patches that were broadcast in response to problems from last month's Internet "worm". No fixes to ypserv for the problems it has with nameservers, timing out, and single-character hosts (we're on our 2nd or 3rd round of such patches for ypserv from Sun, and it's likely that these fixes aren't yet complete). The only problem I had related to some diskless Sun 3/50s that we were running non-standard kernal configurations with. They wouldn't boot completely after installing SunOS4.0.1 on the server. Turns out there was no reason to actually run the unusual kernal configurations, and going back to an essentially vanilla DL50 configuration gave us kernals that booted quite nicely.
kim@uunet.uu.net (Kim Kempf) (12/30/88)
We've noticed that the problem with tty xon/xoff flow control being disabled when a tty is closed is still there. Tty spoolers must have an open() hanging on the port or the flow control is gone when the last job is printed. Also various "bus error" panics related to tar -f /dev/rst8 seem to occur quite regularly now and didn't before. Nothing for sure has been determined yet, but a certain tar dump that we regularly have to do will panic as follows: vmunix: tar: vmunix: trap address 0x8, pid 6679, pc = f06f7e4, sr = 2204, stkfmt b, context 6 vmunix: Bus Error Reg 20<TIMEOUT> vmunix: data fault address 11131113 faultc 0 faultb 0 dfault 1 rw 1 size 0 fcode 5 vmunix: KERNEL MODE vmunix: page map 20000000 pmgrp ff vmunix: D0-D7 2000 2200 5 0 0 f0b9ef8 2000 f0b9ef8 vmunix: A0-A7 11131113 0 f09f59a f09f56e f0b9f48 f0b9f60 ffff9630 ffff9610 vmunix: Begin traceback...fp = ffff9630, sp = ffff9610 vmunix: Called from f06f652, fp=ffff9654, args=f0b9f60 2000 0 0 vmunix: Called from f06ed86, fp=ffff9688, args=1208 0 0 2000 vmunix: Called from f06f200, fp=ffff96b0, args=1208 3 0 0 vmunix: Called from f042322, fp=ffff96d8, args=1208 3 ffff96d4 3 vmunix: Called from f039f76, fp=ffff973c, args=ffff9738 3 f1f5f44 0 vmunix: Called from f038dcc, fp=ffff9768, args=efffc79 0 603 1a4 vmunix: Called from f038d4a, fp=ffff9780, args=efffc79 603 1b6 ffff97b4 vmunix: Called from f064614, fp=ffff97a8, args=ffff9a18 efffc3c 0 0 vmunix: Called from f004768, fp=efffbf8, args=5 2 2d 5 vmunix: End traceback... vmunix: panic: Bus error vmunix: syncing file systems... [2] [2] done vmunix: vmunix: dumping to vp f0f7ba8, offset 37096 vmunix: 1534 total pages, dump succeeded vmunix: rebooting... vmunix: SunOS Release 4.0 (MCRWARE) #8: Tue Dec 6 19:25:25 CST 1988
prl@eiger.uucp (12/31/88)
> From: bob@kahala.hig.hawaii.edu (Bob Cunningham) ... > Here is a quick overview of most of the fixes for Sun3s: ... > yppasswd fixs [security stuff you want] Just so that you are not taken by surprise, the yppasswd fixes posted with 4.0.1 (RCS version 1.15 of yppasswdd.c) do * N O T * fix the security bug in yppasswdd. NEITHER does the later version which our local Sun office sent us (RCS version 1.16 of yppasswdd.c) when we complained about this! Peter Lamb uucp: uunet!mcvax!ethz!prl eunet: prl@ethz.uucp Tel: +411 256 5241 Integrated Systems Laboratory ETH-Zentrum, 8092 Zurich
zjat02@uunet.uu.net (Jon A. Tankersley) (01/12/89)
According to Sun, don't install the 4.0.1 automount fixes. It causes other things to fail. -tank-
jim@eda.com (Jim Budler) (01/18/89)
apctrc!zjat02@uunet.uu.net (Jon A. Tankersley) writes: > According to Sun, don't install the 4.0.1 automount fixes. It causes > other things to fail. Could we have a little more info, please? Bug report number? Thanks, jim -- Jim Budler address = uucp: ...!{decwrl,uunet}!eda!jim domain: jim@eda.com