[8705] in linux-scsi channel archive
Re: max_segments
daemon@ATHENA.MIT.EDU (Eric Youngdale)
Mon Apr 24 20:29:37 2000
Message-ID: <001901bfae4d$0e9e1860$0f17a8c0@eric.home>
From: "Eric Youngdale" <eric@andante.org>
To: "Andrea Arcangeli" <andrea@suse.de>, "Jens Axboe" <axboe@suse.de>
Cc: <linux-scsi@vger.rutgers.edu>,
"Alan Cox" <alan@lxorguk.ukuu.org.uk>
Date: Mon, 24 Apr 2000 20:27:36 -0400
MIME-Version: 1.0
Content-Type: text/plain;
charset="iso-8859-1"
Content-Transfer-Encoding: 7bit
> The original discussion was the 64 limit imposed by the SCSI merge
> functions.
The limit got added between 2.3.44 and 2.3.46. I don't know who added
it - my guess is that it is PCI related, since the comments imply a problem
in pci_map_sg() if the limit grows larger than this.
Folks - the question before the table is this. Who added the code in
scsi_merge.c:
if (max_segments > 64)
max_segments = 64;
This is being used to limit scatter-gather table length to an absolute
hardcoded maximum of 64 segments. It strikes me as wrong to impose a
unilateral 64-segment limit, but we need to know what the problem was that
the author was trying to solve.
-Eric
-
To unsubscribe from this list: send the line "unsubscribe linux-scsi" in
the body of a message to majordomo@vger.rutgers.edu
