[2748] in linux-scsi channel archive
Re: CD-Rom mounting oddities
daemon@ATHENA.MIT.EDU (Gerard Roudier)
Mon Nov 3 17:04:13 1997
Date: Mon, 3 Nov 1997 22:56:30 +0100 (MET)
From: Gerard Roudier <groudier@club-internet.fr>
To: Chris Chiappa <griffon+@cmu.edu>
cc: linux-scsi@vger.rutgers.edu
In-Reply-To: <19971103042616.31865@KeyserSoze.snurgle.org>
On Mon, 3 Nov 1997, Chris Chiappa wrote:
> Often, when mounting a CD-Rom on my SCSI CD-Rom drive I, rather than getting
> a proper mount, get something like this:
> -r-xr-xr-x 1 root root 0 Dec 31 1969 8*
^^
Is this a smiley?
The inode of the mounted point seems obviously corrupted.
The missing 'd' flag seems very strange to me.
Donnot have any idea about io problems causing this.
> Note that 8 is the mount point (ie a directory), yet after mounting it shows
> up as an executable file. The messages logged during the mount operation
> show nothing out of the usual. Usually unmounting, ejecting, and remounting
> solves this although occasionally it takes two or three tries. The drive
I noticed that you are using a module for the fs.
Did you give a try with the CD fs linked into the kernel?
> works fine once it has been successfully mounted. Note that the only
> CD-Roms I generally use are CD-R's. I don't know if this has any relevance.
> This has occured to me both on 2.0.30 and 2.0.31. The drive is a Toshiba
> XM-3801TA SCSI drive attached to a generic NCR 810a based SCSI board. Also
> on the the chain are a Seagate Barracuda (ST32550N) and a Nakamichi MBR-7 7
> disk changer. Also in the machine is an NCR 875 with a Quantum IMP 2100S
> drive and a Philips(IMS) CDD-2000 CD recorder. This happens consistenty
> with the 2.5* and 2.6* ncr53c8xx "FreeBSD-port" drivers. The SCSI buses are
> set to operate at 10MB/sAny ideas? I've attached the output of the various
> proc files at the end.
>
> As a side note, the Barracuda(firmware revision 0022) is on the 810 because
> when attached to the 875 I get bus timeouts/crashes similar to those
> described by someone else recently on this list(aborting command due to
> timeout messages a few times followed by a complete scsi bus reset etc), but
> my problems weren't solved by an upgrade to the 2.6 driver. I've tried
> multiple cables and have disabled command qeueing but not disconnection.
> However, while I had at one point another Barracuda which worked fine on the
> 875, I'm not 100% confident that there isn't some weird jumpering problem
> somewhere on my chain so...(I need to go read up to learn about the
> different types of termination etc...the drives tend to have a jumpers for
> things which mean nothing to me. :) ).
Hmmm ...
My experience with SCSI is that you must be not less than 100% sure of
all settings if you really want a working SCSI system.
The most important jumpers are related to terminations and SCSI id.
Recent disks allow to enable/disable SCAM protocol. Given than the driver
resets the bus at start-up and does not implement SCAM protocol, I would
recommend to avoid using SCAM and so to use jumpered (or default) SCSI ids
for all devices.
Other jumpers should generally not cause problems.
Mixing CD family devices with fast HDs on a signle SCSI bus, is not,
IMO, the best SCSI configuration for your system.
My recommendation would be to look into the docs of your hard disks and
your 875 and to try to be 100% sure, on paper, about the following
settings: (I assume all your 2 disks are narrow and you are using a single
narrow cable).
1 - Both narrow and wide portion of terminations enabled on the
controller side.
2 - One end of the cable connected to the controller.
3 - A disk that has active terminator connected at the other end of
the cable with its terminator enabled.
4 - The other disk connected somewhere on the cable with its terminator
disabled.
Once you are _sure_, try it and let me know.
Good luck!
Gerard.
