From owner-freebsd-scsi@FreeBSD.ORG Tue Dec 4 18:42:55 2012 Return-Path: Delivered-To: freebsd-scsi@freebsd.org Received: from mx1.freebsd.org (mx1.freebsd.org [69.147.83.52]) by hub.freebsd.org (Postfix) with ESMTP id E77E1429 for ; Tue, 4 Dec 2012 18:42:55 +0000 (UTC) (envelope-from jhb@freebsd.org) Received: from bigwig.baldwin.cx (bigknife-pt.tunnel.tserv9.chi1.ipv6.he.net [IPv6:2001:470:1f10:75::2]) by mx1.freebsd.org (Postfix) with ESMTP id B92D28FC0C for ; Tue, 4 Dec 2012 18:42:55 +0000 (UTC) Received: from pakbsde14.localnet (unknown [38.105.238.108]) by bigwig.baldwin.cx (Postfix) with ESMTPSA id 1298FB93A; Tue, 4 Dec 2012 13:42:55 -0500 (EST) From: John Baldwin To: "Steven Hartland" Subject: Re: Changing mfi max_cmds for new tbolt based cards? Date: Tue, 4 Dec 2012 11:00:25 -0500 User-Agent: KMail/1.13.5 (FreeBSD/8.2-CBSD-20110714-p22; KDE/4.5.5; amd64; ; ) References: <93175830FB964224A4A0B2C4A6C662B3@multiplay.co.uk> In-Reply-To: <93175830FB964224A4A0B2C4A6C662B3@multiplay.co.uk> MIME-Version: 1.0 Content-Type: Text/Plain; charset="iso-8859-1" Content-Transfer-Encoding: 7bit Message-Id: <201212041100.25693.jhb@freebsd.org> X-Greylist: Sender succeeded SMTP AUTH, not delayed by milter-greylist-4.2.7 (bigwig.baldwin.cx); Tue, 04 Dec 2012 13:42:55 -0500 (EST) Cc: freebsd-scsi@freebsd.org X-BeenThere: freebsd-scsi@freebsd.org X-Mailman-Version: 2.1.14 Precedence: list List-Id: SCSI subsystem List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Tue, 04 Dec 2012 18:42:56 -0000 On Friday, November 09, 2012 9:18:17 am Steven Hartland wrote: > I've been doing a lot of work over the past few weeks find and fixing bugs > in the mfi driver. One of these fixes was to ensure that mfi_max_cmds is > used consistently throughout the driver to ensure validation checks work > as expected. > > During this I noticed that mfi_max_cmds is very low for current generation > controllers. In my case 2208 based controller reports 1008 cmd slots, so > quite a jump from the previous max which I believe was 256? > > I was wondering if we should look to change the how our sysctl is setup? > > Recently John you made the following comment: > > Mess with mfi_max_cmds at your own risk. The limit was added to work around > > broken mfi(4) firmware revisions that would lock up when the entire command > > queue (256) was used. Just a suggestion to be cautious. It is probably > > safe to use more than 128, but I would be wary of using all of the slots on > > your adapter. (A verbose boost will show you the number of command slots your > > firmware supports.) > > http://lists.freebsd.org/pipermail/freebsd-current/2012-September/036639.html > > Do you have an information about this specifically:- > 1. Which generation of controller? > 2. Which controller firmware was effected? > 3. Was there a reproducible scenario which triggered this behaviour? > > I'm going to test stability here but my current thinking is to change the > default to -1 (controller limit) for new tbolt based controllers. > > N.B. Resent from correct list account, sorry for the duplicate John. I do not have those details, no. I believe it was seen in earlier mfi(4) adapters such as the parts shipped in Dell PE2950s (though knowing Dell that probably covers a wide range of mfi(4) parts at this point). Scott Long is the one who added the limit in r178968 in May of 2008 (so that gives you a bit of the timeframe for which parts were in vogue then). I do think that the adapters where we saw this hang used 256 command slots (may also help narrow the relevant set of boards down). -- John Baldwin