Re: gfs2 iomap dealock, IOMAP_F_UNBALANCED

From: Christoph Hellwig <hch@lst.de>
To: Andreas Gruenbacher <agruenba@redhat.com>
Cc: "Christoph Hellwig" <hch@lst.de>,
	cluster-devel <cluster-devel@redhat.com>,
	"Dave Chinner" <david@fromorbit.com>,
	"Ross Lagerwall" <ross.lagerwall@citrix.com>,
	"Mark Syms" <Mark.Syms@citrix.com>,
	"Edwin Török" <edvin.torok@citrix.com>,
	linux-fsdevel <linux-fsdevel@vger.kernel.org>,
	"Jan Kara" <jack@suse.cz>,
	linux-mm@kvack.org
Subject: Re: gfs2 iomap dealock, IOMAP_F_UNBALANCED
Date: Sun, 7 Apr 2019 09:32:13 +0200	[thread overview]
Message-ID: <20190407073213.GA9509@lst.de> (raw)
In-Reply-To: <CAHc6FU49oBdo8mAq7hb1greR+B1C_Fpy5JU7RBHfRYACt1S4wA@mail.gmail.com>

[adding Jan and linux-mm]

On Fri, Mar 29, 2019 at 11:13:00PM +0100, Andreas Gruenbacher wrote:
> > But what is the requirement to do this in writeback context?  Can't
> > we move it out into another context instead?
> 
> Indeed, this isn't for data integrity in this case but because the
> dirty limit is exceeded. What other context would you suggest to move
> this to?
> 
> (The iomap flag I've proposed would save us from getting into this
> situation in the first place.)

Your patch does two things:

 - it only calls balance_dirty_pages_ratelimited once per write
   operation instead of once per page.  In the past btrfs did
   hacks like that, but IIRC they caused VM balancing issues.
   That is why everyone now calls balance_dirty_pages_ratelimited
   one per page.  If calling it at a coarse granularity would
   be fine we should do it everywhere instead of just in gfs2
   in journaled mode
 - it artifically reduces the size of writes to a low value,
   which I suspect is going to break real life application

So I really think we need to fix this properly.  And if that means
that you can't make use of the iomap batching for gfs2 in journaled
mode that is still a better option.  But I really think you need
to look into the scope of your flush_log and figure out a good way
to reduce that as solve the root cause.