broken scsi_reset (patch)

Manfred Spraul (masp0008@stud.uni-sb.de)
Sat, 9 Jan 1999 23:55:41 +0100

During the introduction of "hostt->use_new_eh_code" the "old" code was
changed as well.
But this broke hostt->scsi_reset() for all drivers that where not updated to
the "modified old" error system.
At least the
ncr53c8xx & 53c7,8xx drivers are affected
possibly some other drivers. (but I have only one hostadapter)

I've attached a patch that 1) prints a warning message, 2) prevents the
computer from crashing during scsi_reset()'s, 3) fixes a bug in the current
implementation 4) should not affect correct drivers at all.

Could you please check that code with your host adapter?
Please note that the code only affects the 'OLD' error handling, and that it
only affects scsi_reset().

Regards,
Manfred
P.S.: If you have external scsi devices it is simple to provoke a
scsi_reset(): just remove the cable for a fraction of a second while doing
'tar c /mnt/sda > /dev/null'. (but mount the volume read-only before you do
that.)
I found this bug because my scsi cdrom doesn't handle lun's correctly, so
the bug isn't totally artificial.

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/