Jaromir Dolecek wrote:
> 
> Guenther Grau wrote:
> > I don't think so. IMHO, it would be more usable to
> > have an umount -f(orce) which forces the unmount to be done.
> 
> It does forced unmount, the problem is that the locked-up filesystem
> (ntfs in my tests) never actually completes the unmount.

Maybe I am talking complete nonsense now, but why not change
the ntfs code (in fact maybe the whole fs-layer to not block
for too long in the unmount case? Is it not possible to set up
a kernel watchdog thread?

On my way home now... Have a nice weekend

  Guenther