From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1756092Ab2F2CUw (ORCPT ); Thu, 28 Jun 2012 22:20:52 -0400 Received: from szxga01-in.huawei.com ([119.145.14.64]:18732 "EHLO szxga01-in.huawei.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1754465Ab2F2CUu (ORCPT ); Thu, 28 Jun 2012 22:20:50 -0400 Message-ID: <4FED10DB.7040500@huawei.com> Date: Fri, 29 Jun 2012 10:20:11 +0800 From: Li Zefan User-Agent: Mozilla/5.0 (Windows NT 6.1; rv:11.0) Gecko/20120312 Thunderbird/11.0 MIME-Version: 1.0 To: "tj@kernel.org" CC: shyju pv , "linux-kernel@vger.kernel.org" , "cgroups@vger.kernel.org" , Sanil kumar , Masanari Iida Subject: Re: 3.5-rc3: BUG: Dentry still in use (1) [unmount of cgroup cgroup] References: <20120627182903.GP15811@google.com> <4FEBF4B7.2070105@huawei.com> <20120628180712.GC22641@google.com> In-Reply-To: <20120628180712.GC22641@google.com> Content-Type: text/plain; charset="ISO-8859-1" Content-Transfer-Encoding: 7bit X-Originating-IP: [10.166.88.115] X-CFilter-Loop: Reflected Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On 2012/6/29 2:07, tj@kernel.org wrote: > Hello, Li. > > On Thu, Jun 28, 2012 at 02:07:51PM +0800, Li Zefan wrote: >>> Hmm... fa980ca87d "cgroup: superblock can't be released with active >>> dentries" is supposed to have fixed that. Looking into it. >>> >> >> >> I think I know what happened here: >> >> umount >> deativate_super(sb) >> dput(subdir) >> subdir->d_count-- >> d_release(subdir) >> deactivate_super(sb) >> shrink_dcache_for_umount(sb) >> BUG(root->d_count)!! >> root->d_count-- > > Can you please elaborate a bit? I'm not really following? Where does > the last root->d_count-- come from? > >>From my limited knowledge about vfs internal, seems the parent's refcnt won't go down to 0 before its children. When mkdir, the parent's refcnt will be incremented, and after rmdir, dput(subdir) will drop subdir's refcnt and then drop parent's. So when dropping the subdir's refcnt and leading the superblock to be killed, the root's dentry is still > 0. >> I use this script to reproduce the bug: >> >> mount -t cgroup -o cpu xxx /mnt >> mkdir /mnt/sub >> sleep 100 < /mnt/sub & >> kill $! >> wait $! >> rmdir /mnt/sub >> umount /mnt > > Unfortunately, this doesn't reproduce the bug here either. :( > I can reproduce the bug reliably.. Try s/cpu/perf or s/cpu/net_cls, which have fewer cgroup files?