Skip site navigation (1)Skip section navigation (2)
Date:      Wed, 28 Apr 1999 21:08:36 +0200
From:      J Wunsch <j@uriah.heep.sax.de>
To:        freebsd-scsi@FreeBSD.ORG
Subject:   Re: QIC tape problems on -stable (was: hanging `tar xfvR /dev/nrst0' process, can i debug it?)
Message-ID:  <19990428210836.47084@uriah.heep.sax.de>
In-Reply-To: <Pine.LNX.4.04.9904280824090.24720-100000@feral.com>; from Matthew Jacob on Wed, Apr 28, 1999 at 08:39:34AM -0700
References:  <19990428091622.29861@uriah.heep.sax.de> <Pine.LNX.4.04.9904280824090.24720-100000@feral.com>

next in thread | previous in thread | raw e-mail | index | archive | help
As Matthew Jacob wrote:

> These are *amazingly* broken drives then. The SCSI specification
> says that a write of *zero* filemarks is to flush any pending
> writes.

Well, but it doesn't say nothing must be written. ;-)

The SCSI spec's a little thin in describing what needs to be done upon
receipt of a WRITE FILEMARKS with a zero count, and what must not be
done.  However, i agree that's a bug in the Tandberg since it violates
the Tandberg SCSI reference as well which indeed claims nothing would
be written.

However, trying to issue a WRITE FILEMARKS on a tape that has never
been written so far is IMHO an mistake, and the SCSI specs don't seem
to forbid that the drive refuses the command at all iff the medium is
write-protected.  So if at all, this `flush buffers' should IMHO only
be done after the tape has actually been written to (which is already
being tracked inside the driver).

However(2), while i was at it, i traced the CAM CCBs, and i couldn't
think of any scenario where a still-buffered write operation could get
into your way at all.  Whatever you're doing, either the application
that holds the tape open for writing hasn't finished yet, in which
case any further attempt to open the tape will be denied, or said
application already caused a real filemark to be written (which by
SCSI definition causes any buffered data to be written out before the
WRITE FILEMARKS (with a count of >= 1) returns).  So what's the
scenario where you think you really need another WRITE FILEMARKS with
a count of 0?

>  I believe that the writers of QIC f/w should be sent to
> Redmond for the rest of their miserable lives.

Nah, c'mon, it's just one bug in the firmware.  By your terms, you
could easily throw away almost any SCSI hardware since they often
suffer more serious firmware bugs...

-- 
cheers, J"org

joerg_wunsch@uriah.heep.sax.de -- http://www.sax.de/~joerg/ -- NIC: JW11-RIPE
Never trust an operating system you don't have sources for. ;-)


To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-scsi" in the body of the message




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?19990428210836.47084>