From mboxrd@z Thu Jan 1 00:00:00 1970 From: Bernd Schubert Subject: Re: [ PATCH 0/3 ] mpt fusion enhancements Date: Fri, 22 Jun 2012 11:30:43 +0200 Message-ID: <4FE43B43.6030405@fastmail.fm> References: <200902061408.46652.bs@q-leap.de> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit Return-path: Received: from mailgw1.uni-kl.de ([131.246.120.220]:41021 "EHLO mailgw1.uni-kl.de" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1761856Ab2FVJar (ORCPT ); Fri, 22 Jun 2012 05:30:47 -0400 Received: from itwm2.itwm.fhg.de (itwm2.itwm.fhg.de [131.246.191.3]) by mailgw1.uni-kl.de (8.14.3/8.14.3/Debian-5+lenny1) with ESMTP id q5M9UiVT031032 (version=TLSv1/SSLv3 cipher=EDH-RSA-DES-CBC3-SHA bits=168 verify=NOT) for ; Fri, 22 Jun 2012 11:30:44 +0200 In-Reply-To: Sender: linux-scsi-owner@vger.kernel.org List-Id: linux-scsi@vger.kernel.org To: Ron Eisler Cc: linux-scsi@vger.kernel.org Hello Ron, On 06/21/2012 11:20 PM, Ron Eisler wrote: > Bernd Schubert q-leap.de> writes: > >> >> Here are 3 patches for the mpt fusion driver. >> > > Hello Bernd, > > I am looking mpt fusion backport fixes for the LSI 53C1030 that will run with > Linux kernel 2.6.27.62. Current version is 3.04.07. Our problem is that running > on VMWare VSphere, with the SAN heavily loaded, we get SCSI timeouts. > > Ex. mptscsih: ioc0: attempting task abort! (sc-ffff...) > sd 0:0:0:0: [sda] CDB: Read (10): 28 00 ... > mptscsih: ioc0: task abort: Success (sc=ffff...) > > The system hangs for 30 seconds on each one. > > I see comments/patches here: http://comments.gmane.org/gmane.linux.scsi/48215 > > Do you have source that would work for 2.6.27.62? > is there a chance you could use the the mpt-4.x driver? The other simple way would be to use the driver from the SLES10 kernel (SLES10 without any service pack, so 2.6.27). I think you could simply replace the drivers/message/funsion directory. I definitely cannot recommend to use a more recent kernel, as I know from Q-Leap that recent kernels run into DV (domain validation) races, which they 'solved' by introducing simple sleeps into the code. I'm still in contact with Q-Leap, but decided to take another job some time ago.... Cheers, Bernd