[7678] in linux-scsi channel archive
Re: How do I invoke a re-probe from kernel space?
daemon@ATHENA.MIT.EDU (Eric Youngdale)
Mon Dec 13 08:39:09 1999
Date: Mon, 13 Dec 1999 08:28:42 -0500 (EST)
From: Eric Youngdale <eric@andante.org>
To: mkwasigr@INTERCOPE.COM
Cc: Alan Cox <alan@lxorguk.ukuu.org.uk>, linux-scsi@vger.rutgers.edu
In-Reply-To: <C1256846.00444CE6.00@NS.INTERCOPE.COM>
Message-ID: <Pine.LNX.4.10.9912130827150.25848-100000@gwyn.tux.org>
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
On Mon, 13 Dec 1999 mkwasigr@INTERCOPE.COM wrote:
> Hi Alan,
>
> you said that the "scsi add-single-device" code is horrible and gets
> overhauled...
>
>
> Some weeks ago I noticed that I loose tagged command queueing (TCQ) on my
> harddisks when I dynamically add a device with "scsi add-single-device" and
> posted it to linux-scsi but nobody answered... sigh. I have this with
> 2.2.1[123] on an adatptec 2940. Although the harddisks use TQC with a queue
> depth of eight (at least /proc/scsi/aic7xxx/0 reports it) the rescan on
> "scsi add-single-device" bumps the disk-queues to a depth of one. :-(
>
> Will the "overhaul" cure this?
My guess is that the problem is that when you do an
add-single-device that it calls down to the low-level driver to select the
queue depth again. The problem is (as the comment describes) that the
low-level drivers may not expect being called to do this task a second
time.
-Eric
-
To unsubscribe from this list: send the line "unsubscribe linux-scsi" in
the body of a message to majordomo@vger.rutgers.edu
