> Does anyone out there knows about some things I could try
> to avoid those dead-scans?
Well, to at least avoid killing the Linux box you should check out
(1) SCSI may become happy again if you turn off the scanner when it hangs
You will see a lot of screaming from the kernel "Where did that go????"
but then it should settle down and you can reboot safely
(2) Magic SysRq is your friend, in 2.2.x at least you can configure the
Alt+SysRq sequence on and have a keypress which syncs and reboots to
avoid file corruption. Check out the appropriate kernel files.
AFAIK the SCSI mid-layer of Linux is unhelpful about reporting errors
such as "You sent too much/ not enough/ wrong kind of data" to upper
layers such as SCSI generic. Tinkering with mid-layer endangers SCSI
disks, tape backups etc. and only benefits SCSI generic users. So it
won't happen for a while yet.
Nick.
-- Source code, list archive, and docs: http://www.mostang.com/sane/ To unsubscribe: echo unsubscribe sane-devel | mail majordomo@mostang.com