> > > Synopsis: df hangs the system after shutting down network with a NFS
> > 
> > is this considered as a bug?
> 
> well, I'd think that it would be better for df to fail in this case.
> or at least sleep interruptibly.

there is no way for kernel or df to know which's a desiable behaviour.

> I just tried this to see what would happen, and I got an infinite stream
> of "nfs_timer: ignoring error 50" on the console.  I had to hard-reset
> the machine to get control back.  so yea, I'd call that a bug.

"DEBUG vs verbosity" is another problem, i think.

YAMAMOTO Takashi