Re: [PATCH] bdi_sync_writeback should WB_SYNC_NONE first

[Jens Axboe <jens.axboe@oracle.com>]

On Sun, Sep 27 2009, Andrew Morton wrote:
> On Sun, 27 Sep 2009 18:44:32 +0200 Jens Axboe <jens.axboe@oracle.com> wrote:
> 
> > On Sun, Sep 27 2009, Andrew Morton wrote:
> > > On Fri, 25 Sep 2009 10:10:14 -0400 Chris Mason <chris.mason@oracle.com> wrote:
> > > 
> > > > diff --git a/fs/fs-writeback.c b/fs/fs-writeback.c
> > > > index 8e1e5e1..27f8e0e 100644
> > > > --- a/fs/fs-writeback.c
> > > > +++ b/fs/fs-writeback.c
> > > > @@ -225,7 +225,7 @@ static void bdi_sync_writeback(struct backing_dev_info *bdi,
> > > >  {
> > > >  	struct wb_writeback_args args = {
> > > >  		.sb		= sb,
> > > > -		.sync_mode	= WB_SYNC_ALL,
> > > > +		.sync_mode	= WB_SYNC_NONE,
> > > >  		.nr_pages	= LONG_MAX,
> > > >  		.range_cyclic	= 0,
> > > >  	};
> > > > @@ -236,6 +236,13 @@ static void bdi_sync_writeback(struct backing_dev_info *bdi,
> > > >  
> > > >  	bdi_queue_work(bdi, &work);
> > > >  	bdi_wait_on_work_clear(&work);
> > > > +
> > > > +	args.sync_mode = WB_SYNC_ALL;
> > > > +	args.nr_pages = LONG_MAX;
> > > > +
> > > > +	work.state = WS_USED | WS_ONSTACK;
> > > > +	bdi_queue_work(bdi, &work);
> > > > +	bdi_wait_on_work_clear(&work);
> > > >  }
> > > 
> > > Those LONG_MAX's are a worry.  What prevents a very long
> > > almost-livelock from occurring if userspace is concurrently dirtying
> > > pagecache at a high rate?
> > 
> > Not sure whether Chris' system is back up again, but I discussed this
> > with him on irc. Since the WB_SYNC_ALL writeback should be queued behind
> > the WB_SYNC_NONE that the non-wait sync already issued, not sure why
> > this patch makes a difference. It's definitely not the right approach.
> > 
> 
> I wasn't referring to this patch actually.  The code as it stands in
> Linus's tree right now attempts to write back up to 2^63 pages...  

I agree, it could make the fs sync take a looong time. This is not a new
issue, though.

-- 
Jens Axboe