On Sun, 10 Mar 2002, Arnd Bergmann wrote:

At 13:40 10.03.2002 +0000, Chris Clayton wrote:

...

...

...

EIP; c012d67f <===== Trace; c01c5d46 <__make_request+86/5b0> Trace; c01c6308 Trace; d089c93f <[pktcdvd]pkt_read_bh+3b/50> I don't have access to any proper Linux workstation right now, but this looks like it could a problem with highmem support.

Do you have support for more than 1GB of physical RAM compiled in your kernel and does changing the option make a difference?

Yes, this is indeed a highmem problem. This patch should fix it. --- linux/drivers/block/pktcdvd.c.old Sun Mar 10 16:10:00 2002 +++ linux/drivers/block/pktcdvd.c Sun Mar 10 16:01:45 2002 @@ -264,6 +264,7 @@ kmem_cache_free(bh_cachep, bh); break; } + bh->b_page = virt_to_page(bh->b_data); spin_lock_irq(&pd->lock); bh->b_pprev = &cdrw->bhlist; -- Peter Osterlund - petero2@telia.com http://w1.894.telia.com/~u89404340