Subject: Re: status of the thorpej_scsipi code
To: Luke Mewburn <firstname.lastname@example.org>
From: Matthew Jacob <email@example.com>
Date: 04/03/2001 18:43:01
On Wed, 4 Apr 2001, Luke Mewburn wrote:
> On Tue, Mar 27, 2001 at 06:40:13PM +0200, Manuel Bouyer wrote:
> > Hi,
> > This is a status update about the thorpej_scsipi branch (and request for
> > testers :)
> A question about the thorpej_scsipi work; are SCSI tape drives better
> supported ?
> At the moment our st driver has no idea about the current file or
> block number, and not knowing where you are on the tape is very
[ I suppose I'm last owner of record for tape - although you wouldn't know it
from lack of working on it ]
This is somewhat orthogonal to the thorpej-scsipi
The tape position is hard to implement safely- not that as soon as you user
hardware or logical block positioning, logical tape position is 'unknownn'.
I did it for the FreeBSD driver, and could probably do something for NetBSD as
well. You can look up how it works there.
> Also, there doesn't seem to be any feedback in "mt status"
> for settings such as "mt compress", etc.
Some of this can be changed. It also depends on how 'compatible' you want to
keep the MTIOCGET structure. Note that this has implications (wierd ones) for
> Finally, the driver has been
> flaky at times, especially in the case of drive or tape error.
Yeah. There are some pending issues.
> I was under the impression that work such as the thorpej_scsipi should
> make improving the above problems much easier, especially with respect
> to the error reporting.
Why don't you collect the PRs about tape and a wish list and send it to me. I
thought I would have time for this over Christmas, but it didn't happen. I can
lump this together with some more FreeBSD work I need to do (to make a driver
that an NDMP agent will be happy to use).