I was having problems with programs that access kernel memory, and was
coming up with error messages like "ps: /vmunix: can't find swapinfo". The
answer was to back out of patch 101686-01 and rebuild the kernel. Once I did
that, everything was fine again. I didn't receive any responses for about
18 hours, so I actually had it figured out before the first response came in,
but I truly appreciate the effort, and everyone was in the ball park.
Thanks go out to:
Kevin A. Noll (knoll@csl-emh2.army.mil)
Jeff LaCoursiere (z056716@uprc.com)
Nate Itkin (Nate-Itkin@ptdcs2.intel.com)
Kevin W. Thomas (kwthomas@nsslsun.nssl.uoknor.edu)
and anyone else who responds after this summary!
Original Question Follows:
> Hello Sun Managers,
>
> I've been stewing over a problem for a day a so now, and rather than start
> from scratch, I thought I would give you folks a crack at it. I had a
> SS20 which was running Generic 4.1.3_U1 successfully. I then installed the
> following patches, and everything seemed to be O.K, in fact, I have another
> SS20 that is still in this configuration, and it is running fine.
>
> 101434-03: lpr Jumbo Patch
> 101435-01: ypserv fix
> 101436-01: patch for mail executable
> 101437-01: accounting jumbo patch
> 101438-01: applications bind to same port if IP address supplied
> 101439-01: system crashed in tcp_Ercvconnect() from a NULL socket pointer
> 101440-01: (SECURITY) methods to exploit login/su
> 101441-01: syslog messages gets confused
> 101442-02: ps(1) with 'j' option may segmentation fault
> 101443-02: esp: SCSI Errors - 'ILLEGAL bit set','watchdog reset'
> 101444-01: 'at' jobs are not running; job files are being over-written
> 101445-01: quota exceeded console messages misleading
> 101446-01: iostat reports incorrect SCSI device names on sun4m
> 101447-01: tar xvpf does not restore the original ownerships on symbolic links
> 101449-01: PANICS WITH CLGET: NULL CLIENT
> 101450-01: two messages displayed when using FKFORMAT_TRACK ioctl
> 101453-01: /usr/ucb/Mail does not pass comma separated address as per RFC822
> 101454-01: vacation(1) does not recognize aliases of the type firstname.lastname
> 101455-01: Using compress/uncompress while over your quota can trash files
> 101456-01: diff -c dumps core
> 101460-02: SunOS consumes all file descriptors when open() returns an EINTR
> 101461-03: csh, sh jumbo patch
> 101462-01: tbl may dump core on sun-4
> 999999-01: homegrown patch to allow use of 8500C and 8505 8mm drives
>
> After running a for a while with that configuration, I checked the updated
> patch list, and I decided to install the following patches as well since they
> seem to apply to our setup:
>
> 101508-06: sun4m jumbo patch for kernel performance and memory bugs
> 101522-01: in.rexecd dumps core
> 101561-05: TTY settings change when rlogin into a 4.x system from a 5.x system.
> 101562-02: libcurses' dlescreen() causes getch to return bad char.
> 101587-01: security patch for mfree and icmp redirect
> 101618-01: inetd broken server detection breaks on fast machines
> 101621-02: Jumbo tty patch
> 101625-02: ftp does not prompt for account information
> 101664-01: rare panic due to serial port load
> 101665-02: sendmail jumbo patch
> 101679-01: Breach of security using modload
> 101686-01: libkvm/kvmopen leaks memory on sun4m machines
> 101696-01: data corruption when using cpio to archive data from NFS
> 101757-01: POINT PATCH fixes to ethernet driver (le.o)
> 101759-02: domestic libc jumbo patch
> 101764-01: coredump can overwrite core file with permsn 400 on NFS
> 101775-01: Cron consumes more memory as time goes on
> 101784-01: rpc.lockd jumbo patch
>
> Anyways, to make a long story short, after the second round of patches, and
> kernel rebuilds, I am having trouble running any program that accesses the
> kernel memory. Here are a couple of examples:
>
> updraft# /bin/ps -U
> ps: /vmunix: can't find swapinfo
> ps: could not read kernel VM
>
> updraft# ps -auwwx
> Segmentation fault
>
> updraft# /etc/pstat -T
> pstat: /vmunix: can't find swapinfo
>
> updraft# /bin/ipcs
> ipcs: /vmunix: can't find swapinfo
>
> And here is a dbx run of the SEGVing ps in case anyone can get anything from it:
>
> updraft# dbx /bin/ps
> Reading symbolic information...
> Read 148 symbols
> warning: main routine not compiled with the -g option
> (dbx) run
> Running: /bin/ps
> signal SEGV (segmentation violation) in _uncondense at 0xef7e6160
> _uncondense+0x40: ld [%i0 + 4], %l6
> (dbx)
>
> I've seen previous questions and SUMMARYS which deal with permission
> problems, but I think that I've done a fairly thorough comparison with the
> half-patched SS20 system that is not having any problems, and I don't see any
> inconsistencies, but then again, I'm the one that screwed it up in the first
> place.
>
> I didn't notice any problem until I completed all the patches, but it could
> have happened after any of them. I could have easily missed it, since the
> system seems to run fine otherwise.
>
> I'm hoping that someone has already come across this, or a similar problem, or
> that someone can figure out which stupid thing I did or didn't do. Anyways,
> any help or insight would be greatly appreciated.
>
> Thanks,
>
> Patrick Pawlak (patp@ll.mit.edu)
> Group 43, Weather Sensing
> M. I. T. Lincoln Lab
> Lexington, MA
This archive was generated by hypermail 2.1.2 : Fri Sep 28 2001 - 23:09:09 CDT