[2462] in linux-scsi channel archive
ncr53c8xx experimental patch.
daemon@ATHENA.MIT.EDU (Gerard Roudier)
Sun Sep 14 18:13:20 1997
Date: Sun, 14 Sep 1997 22:58:17 +0200 (MET DST)
From: Gerard Roudier <groudier@club-internet.fr>
To: linux-scsi@vger.rutgers.edu, ncr53c810@colorado.edu
Hi,
I just made available an experimental version of the ncr53c8xx driver:
ftp://linux.wauug.org/pub/roudier/Testing/ncr53c8xx-2.5b-to-2.6.patch.gz
The patch is against driver version 2.5b and I've tested it under
linux-2.0.29 (only).
All the changes in this patch are related to error detection and
recovery (when it is possible).
This patch will very probably not make difference with flawless SCSI
hardware (It worked just fine for my testings, uploading the patch
and sending this mail).
For the moment I am not able to trigger the modified code.
It is always possible to make the devices have some unexpected behaviour
by sending to it bad messages or commands, but injecting parity errors
is difficult without hacking the hardware. Since I hate hardware that
fails (and I pay for it), I only use a screwdriver when I absolutely
need to. As you probably guess, I do prefer some other sorts of driver.
I have decided to make these changes in the driver after having looked into
the list of failure reports that have'nt been clearly explained.
This patch is _experimental_ and I will reverse any change that does not
really improve error detection and recovery.
I need help, not from a psychiatric (I hope), but for testing this patch.
If you usually get errors using the ncr53c8xx driver with your
hardware, you can give this patch a try and report me if it makes
differences.
Common errors are unexpected phase changes, scsi timeouts, ...
Gerard.
