From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from ipmail06.adl2.internode.on.net ([150.101.137.129]:34238 "EHLO ipmail06.adl2.internode.on.net" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751133AbcCBXyc (ORCPT ); Wed, 2 Mar 2016 18:54:32 -0500 Date: Thu, 3 Mar 2016 10:52:07 +1100 From: Dave Chinner To: Al Viro Cc: Frank Sorenson , linux-fsdevel@vger.kernel.org, dwysocha@redhat.com, lvaz@redhat.com Subject: Re: [PATCH 0/5] Add trace events for filesystem freeze/thaw events Message-ID: <20160302235207.GR29057@dastard> References: <1456945319-16283-1-git-send-email-sorenson@redhat.com> <20160302214744.GN17997@ZenIV.linux.org.uk> <20160302224730.GP29057@dastard> <20160302232254.GO17997@ZenIV.linux.org.uk> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20160302232254.GO17997@ZenIV.linux.org.uk> Sender: linux-fsdevel-owner@vger.kernel.org List-ID: On Wed, Mar 02, 2016 at 11:22:54PM +0000, Al Viro wrote: > On Thu, Mar 03, 2016 at 09:47:30AM +1100, Dave Chinner wrote: > > > > So let's get it right. Preferably - without need for boot-to-shutdown > > > tracking just to mirror the state. What do we really want? > > > > > > * an ioctl to query the state (frozen/freezing/not frozen) for something in > > > util-linux to use? > > > > > > * /proc/fs/freezing and /proc/fs/frozen, with ->s_id of affected filesystems > > > or, pehaps, one file with (frozen|freezing) + ->s_id? > > > > > > * ability to audit on state changes? That'd need some thought re what to > > > do when some joker freezes the fs syslogd is logging to... > > > > > > * something else entirely? > > > > Such as: > > > > https://lkml.org/lkml/2015/6/16/456 > > IIRC, there had been a weird use of vfsmounts as proxies for superblocks and > even more strange requirements along the lines "you should just pick one and > use only it in case if there's more than one mountpoint for this fs". It's > been a while, though, so I might've forgotten if that stuff got resolved in > later thread... > > Lifetime rules are really odd there; the object is killed as we are about > to gut vfsmount for good, even though the code in there appears to assume > that it would stay connected to mount tree all the way until that point. > Again, that might have been resolved later; I really can't recall right now... Agreed, there were lots of unresolved problems with that patch set (e.g. the duplicate space accounting infrastructure, rather than just calling vfs_statfs() to grab the free/used space from the filesystems). I was just making the point that we really need a generic fs event mechanism, not just something specific to freeze/unfreeze. I've kind of been waiting for the kdbus stuff to work itself out, because if there's going to be a generic kernel-wide event mechanism added to the kernel, we should just be using that... Cheers, Dave. -- Dave Chinner david@fromorbit.com