From mboxrd@z Thu Jan 1 00:00:00 1970 From: Al Viro Subject: Re: linux-next: Tree for January 15 (Call Trace in fs/dcache.c + autofs4) Date: Sat, 15 Jan 2011 11:08:48 +0000 Message-ID: <20110115110848.GL19804@ZenIV.linux.org.uk> References: <20110115075723.GJ19804@ZenIV.linux.org.uk> <20110115090702.GK19804@ZenIV.linux.org.uk> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Cc: Stephen Rothwell , linux-next@vger.kernel.org, LKML , linux-fsdevel To: sedat.dilek@gmail.com Return-path: Content-Disposition: inline In-Reply-To: Sender: linux-kernel-owner@vger.kernel.org List-Id: linux-fsdevel.vger.kernel.org On Sat, Jan 15, 2011 at 12:00:54PM +0100, Sedat Dilek wrote: > > Argh... ??In __do_follow_link() replace > > > > ?? ?? ?? ??if (link->mnt != nd->path.mnt) > > with > > ?? ?? ?? ??if (link->mnt == nd->path.mnt) > > > > Mismerge yesterday ;-/ ??I've pushed fix for that in for-next, will fold > > shortly. ??As for autofs4 breakage, I've a preliminary fix, testing it > > now. > > > > Hey, cool and thanks. > > Not sure if you catched them all, I have noticed on my latest > ("buggy") linux-next kernel these Call Traces when doing an > update-grub. That might be vfsmount being dropped when it shouldn't or dentry leaked. And seeing that it's umount(8), I would suspect the latter... Anyway, with the latest from dhowells we probably should have d_set_d_op() mess on autofs4 under control (in #for-next). Whether it's enough to actually fix the sucker is a separate question, of course - there might very well be more crap. I've instrumented mntput() et.al. here; hopefully that'll make catching the remaining turds easier. As for dcache leaks... ouch. Could you try to reproduce that one on the mainline kernel? At least that'd isolate things a bit.