> Tuomas Heino wrote:
>
> > After the first OOPS most programs that try to write something
> > to the fs's will OOPS... and I up'd my puter from .64 to .72 because it had
> > (less serious) OOPSes... after a few OOPSes all the progs which try to
> > read will "hang" ... the odd thing is that sysklogd got some of the
> > OOPSes saved ;)
>
> Once something has trashed the dentry list, the system has no good way
> to recover. So the first oops will lead to many more to follow ...
>
> I haven't been able to find out what's causing the list to be trashed.
> If you spot a pattern to the problem (e.g. "every time I use ppp I get
> an oops within 10 minutes") I'd appreciate any further clues.
>
Dec 11 17:02:20 bx1 kernel: registered device ppp0
Dec 11 19:32:58 bx1 kernel: Unable to handle kernel paging request at virtual address 842340c1
Dec 12 21:54:20 bx1 kernel: registered device ppp0
Dec 12 23:02:08 bx1 kernel: Unable to handle kernel paging request at virtual address 53565708
To me that doesn't look like ppp causing it - or does it? ;)
... well there are no kernel messages between those two ...
and in the first case pppd was killed at 17:32... it _could_ have tried
to unload the ppp module at 19:32... (picking up ppp just 'cause you said
it AND 'cause it screws up with sysklogd... meaning sysklogd sends
"emergency" messages to ppl log'd in with "null" (inlcudes only the hostname)
message body... and my old dialup script doesn't work with recent
kernels/pppds either... (that script used multiple processes...)