From mboxrd@z Thu Jan 1 00:00:00 1970 From: James Bottomley Subject: Re: [PATCH] block: Free queue resources at blk_release_queue() Date: Wed, 28 Sep 2011 09:11:14 -0500 Message-ID: <1317219074.19034.4.camel@dabdike.hansenpartnership.com> References: <201109221318.p8MDIXPP020015@pentland.suse.de> <20110928004722.GA27421@kernel.dk> <4E82751C.9050908@kernel.dk> <1317183038.17578.9.camel@dabdike.hansenpartnership.com> <4E832A55.30709@kernel.dk> Mime-Version: 1.0 Content-Type: text/plain; charset="UTF-8" Content-Transfer-Encoding: 7bit Return-path: In-Reply-To: <4E832A55.30709@kernel.dk> Sender: linux-kernel-owner@vger.kernel.org To: Jens Axboe Cc: Linus Torvalds , Hannes Reinecke , James Bottomley , "linux-scsi@vger.kernel.org" , Linux Kernel List-Id: linux-scsi@vger.kernel.org On Wed, 2011-09-28 at 08:08 -0600, Jens Axboe wrote: > On 2011-09-27 22:10, James Bottomley wrote: > > On Tue, 2011-09-27 at 18:59 -0700, Linus Torvalds wrote: > >> On Tue, Sep 27, 2011 at 6:15 PM, Jens Axboe wrote: > >>>> > >>>> But if you forward the actual patch to me (the one I see on lkml seems > >>>> to be broken and doesn't compile cleanly because it's assiging a > >>>> structure to a pointer), I'll try it out anyway. > >>> > >>> Thanks, that would be great. It's inlined below. > >> > >> Well, I did several USB eject events, and nothing bad happened. > >> > >> But as mentioned, I don't think that means much. Have you tried this > >> with slub debugging and poisoning? It might be worth some extra > >> testing that way. > > > > I've been testing a simpler version: > > > > http://marc.info/?l=linux-kernel&m=131300594629839 > > > > For a long time now with great success. The only difference is the > > locking cleanup, but SCSI doesn't need that since it doesn't supply its > > own lock, so I've been voting for this as the final fix for a while now. > > The locking cleanup looks good, though, for devices that do use the > embedded lock. Exactly ... it's the missing piece; without it my patch is really only addressing SCSI. > But functionally they should be the same for SCSI, so if > you had great success with it, then that's a good data point. Right, I've run it through the memory debugger and USB ejection testing (with ext2, which seems to be the fs that triggers this problem the most). James