[plug] kill them dead

[Adrian Chadd]

On Fri, Oct 13, 2006, Gavin Chester wrote:
> On Fri, 2006-10-13 at 23:15 +0800, Adrian Chadd wrote:
> > It doesn't hvae to be a kernel bug.
> > 
> > You can't kill -9 a process locked waiting for NFS IO to complete, IIRC.
> > There's more examples of that. Its certainly not a bug. :)
> > 
> The process in question (k9copy) usually behaves very well, but when it
> gets a disk that it can't handle it causes it to lock up jsut that
> process.  Then the dvd drive is locked for any other signal or process
> and no matter how many times I issue 'kill -9' the process
> stays/persists.  I sometimes have luck with trying to start another
> instance of k9copy and then killing the second instance followed by the
> first, otherwise it's a reboot.  BTW: it seems disney discs are the most
> troublesome ;-) 

Well, the driver could certainly bit a bit better behaved and not just
sit there waiting for IO to (never) complete (successfully or not.)

Yes, signals won't always work when the process is stuck in kernel mode
trying to achieve something thats not interruptable for various reasons.
NFS is a good example (unless mounted with nointr.) Doing non-interruptable
device IO is another. Its not a bug that its not interruptable but, as I
said, the driver writer could've been a bit nicer and made things
interruptable..

> I'm worried by the things that you guys are telling me - it doesn't
> sound good :-(

Heh. Welcome to the world of free unix. Its certainly not perfect.
But you can always take a look and try to fix it. :)



Adrian