> > Rather post debug information from your kernel if available (as in: did
> > you get a panic message, and what did it say?).
>
> No, it's an immediate hard lockup. It occurs right after a file writefile,
> and the file that's supposed to have been written does not get written.
Are you sure it's related to the tty code only? Seems like a silly
question, but:
> Is there somewhere I can look after I reboot and it gets done fscking my
> drive? btw, about 20% of the time it does such serious damage that it puts
> me in maintenance mode on reboot. (I did a *lot* of rebooting to get the
> information for the bug.)
Real Serious Damage (tm) as in: loss of unrelated data, not just the file
nano didn't get around to writing, shouldn't happen here (the amount of
damage might be reduced by syncing a few times BTW, or using a
journaling filesystem). That's why I suspect the VFS might be part of the
problem.
Re: lot of rebooting with manual fsck and the like: Been there, done that
while debugging XFree86 4.0 on the Lombard. I'd try to get emergency sync
going with SysRq, failing that I'd try to get as much info out of the
crash by using serial console and hooking a panic() workalike to some
trigger (special key if the keyboard interrupt is still working; do
Powermacs have some sort of NMI interrupt switch??).
Michael
