Date: Sat, 25 Sep 2010 11:40:39 +0200 From: Marius Strobl <marius@alchemy.franken.de> To: Lawrence Stewart <lstewart@freebsd.org> Cc: svn-src-head@freebsd.org, svn-src-all@freebsd.org, src-committers@freebsd.org Subject: Re: svn commit: r213105 - head/sys/dev/mpt Message-ID: <20100925094039.GS46946@alchemy.franken.de> In-Reply-To: <4C9D7068.2010100@freebsd.org> References: <201009241640.o8OGelKS031911@svn.freebsd.org> <4C9D7068.2010100@freebsd.org>
next in thread | previous in thread | raw e-mail | index | archive | help
On Sat, Sep 25, 2010 at 01:45:44PM +1000, Lawrence Stewart wrote: > Hi Marius, > > On 09/25/10 02:40, Marius Strobl wrote: > > Author: marius > > Date: Fri Sep 24 16:40:46 2010 > > New Revision: 213105 > > URL: http://svn.freebsd.org/changeset/base/213105 > > > > Log: > > Improve r56796; the reply handler actually may remove the request from > > the chain in which case it shouldn't be removed twice. > > Reported by: Staale Kristoffersen > > Can you please provide a bit of extra commentary about the symptoms of > the issue this patch addresses? Like Staale, I'm running a FreeBSD > server which uses mpt for its disk subsystem (SunFire X4100 with 2 x > SATA 2.5" disks in RAID-1) and I've had a series of long running > problems with the performance of the controller. I'm keeping a close eye > on all mpt related commits and hope to find the cause sometime. > In case the reply handler also removed the request this triggered a sanity check in TAILQ_REMOVE with INVARIANTS compiled in, causing a panic. I don't think it had a negative impact with INVARIANTS disabled. Also, in order to hit the problem you first had to run into a timeout leading to an IOC rest like here: http://folk.uio.no/stalk/mpt/timeout.txt i.e. I don't think this is related to your performance issue. Marius
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20100925094039.GS46946>