From mboxrd@z Thu Jan 1 00:00:00 1970 From: Marc MERLIN Subject: Re: decoding PMP errors with Marvel SATA controller Date: Sun, 26 Jun 2011 17:06:14 -0700 Message-ID: <20110627000614.GA5362@merlins.org> References: <20110626070820.GC24867@merlins.org> <20110626090309.GA12200@mtj.dyndns.org> <4E076E9E.9080400@pobox.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Return-path: Received: from magic.merlins.org ([209.81.13.136]:43342 "EHLO mail1.merlins.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1755981Ab1F0AGX (ORCPT ); Sun, 26 Jun 2011 20:06:23 -0400 Content-Disposition: inline In-Reply-To: <4E076E9E.9080400@pobox.com> Sender: linux-ide-owner@vger.kernel.org List-Id: linux-ide@vger.kernel.org To: Mark Lord Cc: Tejun Heo , linux-ide@vger.kernel.org, Larry Li , hbailey@marvell.com Speaking of which, what sata chips are recommended nowadays for PMP outside sil24 cards? On Sun, Jun 26, 2011 at 01:38:38PM -0400, Mark Lord wrote: > > This sounds a lot like another recent PMP problem we had with marvell. > > It could be that FIS-based switching doesn't work quite as well for > > some reason. > > This all *used* to work with sata_mv. > But I also recently noticed my sata_mv cards no longer work with the > 4-port Marvell PMP here, and that combo definitely did work when I > finished with it all (back around 2.6.29 or so). I also used it but it was back 2-3 years ago now. I think it even worked back in 2.6.18. > I'm rather distracted here right now with broken internet connectivity, > and hunting for new providers etc., so not likely to dig around on it. > > Perhaps Marc MERLIN could retest with an older kernel, > say 2.6.32 or 2.6.29 even, and see what happens with those. > Hopefully it will work there for him, and we'll have a starting > point to bisect the issue. So it's a bit difficult for me to get too much data on this since it's a production system and downtime does come at somewhat of a premium. However, if whoever may be able to work on it and see what broke (because it almost works, I see all the drives on my PMP and I can read data somewhat, I just get a lot of somewhat scary errors now) lets me know, if he/she has no way to test him/herself, I'll see what I can do to help out and get to replicate this with older kernels on either that server, or maybe another machine I would have to build for it (trying repeatedly on production data, even with backups, not so good :) ). Thanks for the answer, Marc -- "A mouse is a device used to point at the xterm you want to type in" - A.S.R. Microsoft is to operating systems .... .... what McDonalds is to gourmet cooking Home page: http://marc.merlins.org/