[snip snip]
It appears that the new reset code in scsi.c is responsible for
this. That is, I've been doing diffs between .90 and .94, and
that's the only place where anything that looks remotely related
to the issue is changed. There doesn't seem to be a name attached
to the patches, so I don't know who to thank, so keep up the good
work, whoever you are.
I'm glad to hear it. Actually, this was a big enough edit that I should have
added a comment with my name attached, so I've sent off a patch to do so. The
full details of these changes were announced on the linux-scsi list last week,
though with the reliability of the vger lists, who knows if people actually saw
it.
[snip snip]
The machine I'm using for testing this is going away in a few
days :-(, so if anyone wants to see any of my debugging output
(which, thanks to the lack of kernel panics, now gets saved to
disk :-), please let me know and I'll be absolutely pleased to
oblige.
I would like to see it, but I have to admit that it sounds like you're using
the NCR 53c810 driver, and I suspect that I won't be able to make heads or
tails of it.
Thanks again!
You're most welcome.
Leonard