From owner-freebsd-hardware@FreeBSD.ORG Wed Oct 29 15:24:18 2008 Return-Path: Delivered-To: freebsd-hardware@freebsd.org Received: from mx1.freebsd.org (mx1.freebsd.org [IPv6:2001:4f8:fff6::34]) by hub.freebsd.org (Postfix) with ESMTP id 6EE051065686 for ; Wed, 29 Oct 2008 15:24:18 +0000 (UTC) (envelope-from groot@kde.org) Received: from smeltpunt.science.ru.nl (smeltpunt.science.ru.nl [131.174.16.145]) by mx1.freebsd.org (Postfix) with ESMTP id 060768FC1B for ; Wed, 29 Oct 2008 15:24:17 +0000 (UTC) (envelope-from groot@kde.org) Received: from n142207.science.ru.nl (n142207.science.ru.nl [131.174.142.207]) by smeltpunt.science.ru.nl (8.13.7/5.28) with ESMTP id m9TFOFSF006512 for ; Wed, 29 Oct 2008 16:24:16 +0100 (MET) From: Adriaan de Groot Organization: KDE e.V. To: freebsd-hardware@freebsd.org Date: Wed, 29 Oct 2008 16:32:06 +0100 User-Agent: KMail/1.9.7 References: <49074acebedc1@note2email.com> <20081029061606.GN1137@server.vk2pj.dyndns.org> In-Reply-To: <20081029061606.GN1137@server.vk2pj.dyndns.org> MIME-Version: 1.0 Content-Type: text/plain; charset="iso-8859-6" Content-Transfer-Encoding: quoted-printable Content-Disposition: inline Message-Id: <200810291632.06845.groot@kde.org> X-Spam-Score: -2.405 () ALL_TRUSTED,BAYES_20 X-Scanned-By: MIMEDefang 2.63 on 131.174.16.145 Subject: Re: PCI-X SATA Card + Server Recommendation X-BeenThere: freebsd-hardware@freebsd.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: General discussion of FreeBSD hardware List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Wed, 29 Oct 2008 15:24:18 -0000 On Wednesday 29 October 2008 07:16:06 am Peter Jeremy wrote: > > =A0When can we expect NCQ support for FreeBSD? > > When someone implements it. =A0If it's a serious issue for you, either > implement it or offer to pay someone to implement it for you. If I remember correctly -- this is from nearly a year ago -- Soren was goin= g=20 to do so for at least AHCI (Intel) things. It's a pretty big change in the= =20 way ata(4) works, because of the pervasive assumption that one ATA command= =20 results in one corresponding response. That assumption made doing FIS-based= =20 things (for SiI3124 support, oh irony in this thread) complicated as well. I really don't know what the status of this work on Soren's side is. Nor do= I=20 remember anything of my attempts to work NCQ into the SiI driver parts of=20 ata(4). I think I ran quite quickly into issues of simply remembering which= =20 slots were in use; it was going to be a horrific hack on my part in any cas= e. Note that drivers like the Areca one do do NCQ, but they hook in to the SCS= I=20 subsystem where that fits better with the design as a whole. [ade] =2D-=20 Adriaan de Groot