Subject: Re: 4.99.36 very unstable for me(tm)
To: Thomas Klausner <wiz@NetBSD.org>
From: Greg Oster <oster@cs.usask.ca>
List: current-users
Date: 11/14/2007 16:01:10
Thomas Klausner writes:
> Hi!
> 
> 4.99.36/amd64 is really really unstable for me -- I had one hang in a
> bulk build yesterday (about which I already wrote before).
> 
> I rebooted and had a lockup after a reboot when I started X (no panic
> line in dmesg though, so I don't know why it hung), then I rebooted
> into 4.99.35 to compile a newer 4.99.36 to get the tty locking fixes;
> rebooted into 4.99.36 from today, started a bulk build and it hung in
> the first package again (perl). It was a uvm_fault. (Hardware
> problem?) I didn't copy down the message because I thought it would be
> in dmesg after reboot, but it wasn't.
> 
> After the next reboot, I started firefox and the machine hung.
> Firefox was compiled against 4.99.35 libraries and I had the 4.99.36
> ones installed. Is there an incompatibility or was I just unlucky
> again?
> 
> Is 4.99.36/amd64 stable for everyone else?

No.... I've had two hangs so far, mostly when trying to 'build.sh -j 16'.
The first one I didn't have a console window around, and couldn't get 
to any scrollback...  In the second case I got this on the console:

Mutex error: mutex_vector_exit: exiting unheld spin mutex

lock address : 0xffff800090a22730
current cpu  :                  3
current lwp  : 0xffff80009064a8a0
owner field  : 0xffff800000000000 wait/spin:                0/1

panic: lock error

but couldn't break into ddb at that point... (and ended up 
power-cycling the machine).  Unfortunately, I suspect the above 
information isn't enough to be useful either :( 

Later...

Greg Oster