From owner-freebsd-scsi Tue May 26 14:37:28 1998 Return-Path: Received: (from majordom@localhost) by hub.freebsd.org (8.8.8/8.8.8) id OAA25107 for freebsd-scsi-outgoing; Tue, 26 May 1998 14:37:28 -0700 (PDT) (envelope-from owner-freebsd-scsi@FreeBSD.ORG) Received: from panzer.plutotech.com (ken@panzer.plutotech.com [206.168.67.125]) by hub.freebsd.org (8.8.8/8.8.8) with ESMTP id OAA25082; Tue, 26 May 1998 14:37:16 -0700 (PDT) (envelope-from ken@panzer.plutotech.com) Received: (from ken@localhost) by panzer.plutotech.com (8.8.8/8.8.5) id PAA00792; Tue, 26 May 1998 15:37:04 -0600 (MDT) From: "Kenneth D. Merry" Message-Id: <199805262137.PAA00792@panzer.plutotech.com> Subject: Re: DPT install problem In-Reply-To: from Simon Shapiro at "May 26, 98 12:58:53 pm" To: shimon@simon-shapiro.org Date: Tue, 26 May 1998 15:37:03 -0600 (MDT) Cc: rzig@verio.net, freebsd-scsi@FreeBSD.ORG, freebsd-stable@FreeBSD.ORG, tom@sdf.com, owensc@enc.edu X-Mailer: ELM [version 2.4ME+ PL28s (25)] MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-freebsd-scsi@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.org Simon Shapiro wrote... > On 22-May-98 Raul Zighelboim wrote: > > As a sidebar; the DOS configuration software can only create arrays with > > up to 1 mbyte interleave. Is this a limitation on the hardware or the > > software ? Is (will there be) a way to create an array with a larger > > interleave (32 megs comes to mind). > > An interleave factor larger than 1MB will be very inefficient: > > a. The SCSI bus will be used (locked) for long periods. This will > cause poor multi-tasking response. > > b. The cache memory will be able to contain very few stripes. > > Also notice that in the context of FreeBSD, anything larger than 64K is > sort of mute, as the O/S limits all SCSI transfers to that size. I do not > know yet how CAM will do, but suspect it to be the same. It will be the same. The 64K limitation is a generic FreeBSD limitation, so CAM won't change it. CAM won't limit it, though. Here at Pluto, we've been running large chunks (> 64K) through CAM and the Adaptec driver by passing physical memory addresses in through the passthrough driver. (and therefore you bypass the whole struct buf thing and physio) If we ever get a buffer chaining (or similar) scheme in place, it'll be possible to do any size I/O without jumping through hoops. > > Thanks, Simon, for a job well done. > > Thanx. I am going to get started on CAM port in the next few days. Good deal. Contact me via private email about getting access to the CAM source repository. Ken -- Kenneth Merry ken@plutotech.com To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-scsi" in the body of the message