[8641] in linux-scsi channel archive
Re: reading blocks > 32K with generic device
daemon@ATHENA.MIT.EDU (Kai Makisara)
Sat Apr 15 17:40:32 2000
Date: Sun, 16 Apr 2000 00:37:33 +0300 (EET DST)
From: Kai Makisara <makisara@metla.fi>
Reply-To: Kai.Makisara@metla.fi
To: Matthew Jacob <mjacob@feral.com>
Cc: Eric Lee Green <eric@estinc.com>,
Jim Hammack <HammackJ@navo.navy.mil>, linux-scsi@vger.rutgers.edu
In-Reply-To: <Pine.BSF.4.10.10004150958570.88842-100000@beppo.feral.com>
Message-ID: <Pine.OSF.4.10.10004160015180.9062-100000@abies.metla.fi>
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
On Sat, 15 Apr 2000, Matthew Jacob wrote:
>
> > Should we perhaps have turned that into *TWO* SCSI Write() commands?
>
> Never, ever, ever, ever do this unless the tape is in fixed block mode to
> begin with.
>
I agree. One write() in variable block mode must only write one tape
block.
> Also, you can issue the READ BLOCK LIMITS command to get the range of allowed
> block sizes for a device. The high end AMPEX drives don't get up to speed
> until you start writing > 256KB blocks.
>
The driver reads the block limits when the device is opened and it checks
that the block size for fixed block mode does not exceed these limits. In
variable block mode the check is left for the drive. The user sees the
error anyway.
If the driver would check the byte count against the limits, the driver
could return EINVAL instead of EIO. But would this added code actually
help anyone?
Kai
-
To unsubscribe from this list: send the line "unsubscribe linux-scsi" in
the body of a message to majordomo@vger.rutgers.edu
