Subject: Re: vmware suddenly crashing the system...
To: None <smb@research.att.com>
From: Brad Spencer <brad@anduin.eldar.org>
List: port-i386
Date: 08/01/2001 20:40:37
vmware has suddenly started hanging or crashing my machine. I had been
running 1.5.1b2; in a vain attempt to solve the problem, I upgraded
this morning to the latest kernel in the 1.5 branch, which identifies
itself is 1.5.2_ALPHA.
The symptom is that most user-level programs (including my window
manager) go non-responsive when vmware is trying to sync its redo file
(I use undoable virtual disks). ssh from another host hung, too, but I
could get a response from 'ntpq'. Sometimes -- but not always -- I've
seen TCP connection attempts from outside stay in SYN_SENT state (i.e.,
it never got an answer from an interrupt-level function on the hung
machine), but ping succeeded. I've *never* seen that happen.
At least once, the machine panicked, leaving behind the following
mesage on reboot:
Jul 31 10:56:45 berkshire savecore: reboot after panic: uvm_pagedeactivate: caller did not check wire count
Jul 31 10:56:45 berkshire savecore: no dump, not enough free space in /var/crash
I moved /var/crash to /usr, but (of course) I haven't seen that failure
since then.
The only substantive thing I changed between when vmware had been
working and when it started failing is that I removed a 128M SIMM. I
had been getting sig11 during gcc compilations, which (according to ma
ny folks) indicate memory errors. I've seen no such failures since
removing the SIMM -- which still leaves me with 256M -- but that could
mean that the bad spot is now somewhere in the kernel. (The IBM
diagnostics haven't found anything wrong...)
Any suggestions would be gratefully appreciated.
--Steve Bellovin, http://www.research.att.com/~smb
This sort of thing has *always* happened to me when I run MS-WINDOWS 2000
in a virtual machine.
Basically, it appears that if things have to start to page that there is a
chance that the machine will hang. I can almost always trigger this by
starting two vmware sessions at the same time.
And, for reasons that are not clear, when you shut down a virtual machine
that has been running MS-WINDOWS 2000, the vmware process will hang,
followed by the system.
Brad Spencer - brad@anduin.eldar.org
http://anduin.eldar.org - & - http://anduin.ipv6.eldar.org [IPv6 only]
[finger brad@anduin.eldar.org for PGP public key]