On Thu, 19 Nov 1998, Felix Egli wrote:

> > Fundamentally, the kernel is hung. It is not smart enough to
> > recognize the fault and use the other swap partition.
> 
> The kernel can do absolutly nothing when it wants to read back some
> pages (from swap to RAM) that ar no longer readable.
> 
I would think it could at least ID the task the ram belongs to as trashed 
and sigfault the task, mark the disk partition as faulty and continue. 
The overhead to do that should be minimal. Oh well.... I'm wandering off 
topic.

Reply via email to