From mboxrd@z Thu Jan 1 00:00:00 1970 From: Andreas Werner Subject: Re: libata-core: devslp fails on ATP mSATA Date: Wed, 2 Dec 2015 10:33:10 +0100 Message-ID: <20151202093310.GE10207@awelinux> References: <20151201092234.GB29791@awelinux> <20151201161715.GA12922@mtj.duckdns.org> <20151201190410.GA21252@mtj.duckdns.org> Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Return-path: Received: from mail1.bemta14.messagelabs.com ([193.109.254.117]:13907 "EHLO mail1.bemta14.messagelabs.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1756930AbbLBJlp (ORCPT ); Wed, 2 Dec 2015 04:41:45 -0500 Content-Disposition: inline In-Reply-To: <20151201190410.GA21252@mtj.duckdns.org> Sender: linux-ide-owner@vger.kernel.org List-Id: linux-ide@vger.kernel.org To: Tejun Heo Cc: Andreas Werner , linux-kernel@vger.kernel.org, linux-ide@vger.kernel.org, Hannes Reinecke , "Martin K. Petersen" On Tue, Dec 01, 2015 at 02:04:10PM -0500, Tejun Heo wrote: > On Tue, Dec 01, 2015 at 11:17:15AM -0500, Tejun Heo wrote: > > So, I suppose this is a fallout from 9faa643855df ("libata: use > > READ_LOG_DMA_EXT"). The description just says "we should try use it" > > but is there any benefit from using NCQ variant of read log page? I > > mean, it's used during config and error handling, so it's not like > > queueing is gonna help anything. It ends up issuing NCQ commands on > > controllers which don't support NCQ and causes failures on devices > > which lie about supporting the feature. > > Hmmm... so, the controller locks up on READ_LOG_EXT issued as pio? > That's just weird. I suppose you can blacklist the controller so that > any attempt to read the log page fails without actually issuing the > command. > > Thanks. > > -- > tejun Hi, yes since i am using 3.14 where the READ_LOG_DMA_EXT is not it, it seems that the controller locks up even on READ_LOG_EXT as PIO. I have checked the current ERRATA sheet of the CPU but there are no issues regarding this problem. Test on a Freescale T4240 show the same behaviour as on P1013/P1022. It seems that some FSL Cpus are affected. I have asked my FAE at Freescale if he has some information about that kind of issue. Blacklisting the controller would be a solution yes, but I just wanna wait the answer from the FAE to be sure we really have a problem. What kind of identifier can I use for blacklisting? The driver name from the host driver? (fsl-sata) I did not find any register in the sata controller to read out an ID. Regards Andy