Re: [PATCH 4/4] ext4: call blkdev_issue_flush on fsync

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

On Tue, May 20 2008, Eric Sandeen wrote:
> Jamie Lokier wrote:
> > Theodore Tso wrote:
> >> On Fri, May 16, 2008 at 02:09:56PM -0500, Eric Sandeen wrote:
> >>> To ensure that bits are truly on-disk after an fsync,
> >>> we should call blkdev_issue_flush if barriers are supported.
> >> This patch isn't necessary, and in fact will cause a double flush.
> >> When you call fsync(), it calls ext4_force_commit(), and we do a the
> >> equivalent of a blkdev_issue_flush() today (which is what happenes
> >> when you do a submit_bh(WRITE_BARRIER, bh), which is what setting
> >> set_ordered_mode(bh) ends up causing.
> > 
> > ISTR fsync() on ext3 did not always force a commit, if in-place data
> > writes did not change any metadata.
> 
> I think that might still be true, but I'm still looking through it (in
> the background...)
> 
> I tried blktrace to see what was going on but I'm not sure what an "NB"
> in the RWBS field means, anyone know?

Eric already knows this now, but for the benefit of anyone else that may
be curious - it's an empty (data-less) barrier. 'N' is basically 'no
data' (eg not a read nor a write) and 'B' is barrier.

-- 
Jens Axboe