> I did the following modification to the EATA driver (but the same thing
> is possible with almost all scsi drivers) in order to track down
> the scsi problems with 2.1.97-SMP:
> at the beginning of the queuecommand routine I added the test
> if (SCpnt->host_scribble) panic ("SCpnt %p in use!\n", SCpnt);
Hi Dario,
let me just add a pretty stupid "me too" to that topic. I have the same
problems here with a twin channel adaptec card. I think the whole
spin_lock_* calls in scsi.c are simply not working. I worked the last
couple of days to change the code back to the previous cli/sti approach,
but I had not much luck.
If I had a patch ready I'd really like to send it to the list, but I messed
up the current code so badly that I cannot provide one. I was wondering
that nobody else reported that problem, because it's pretty easy for me to
lock up my system with a simple tar command.
I hope someone with deeper knowledge of the whole spinlock think could fix
this, because it's pretty annoying ...
... I'll keep on trying (Patrick)
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu