[dm-devel] [PATCH for-next] dm: fix missing bi_remaining accounting

[Mikulas Patocka]

On Mon, 4 Nov 2013, Mike Snitzer wrote:

> On Mon, Nov 04 2013 at 10:25am -0500,
> Mikulas Patocka <mpatocka at redhat.com> wrote:
> 
> > 
> > 
> > On Mon, 4 Nov 2013, Mike Snitzer wrote:
> > 
> > > On Mon, Nov 04 2013 at 10:06am -0500,
> > > Mikulas Patocka <mpatocka at redhat.com> wrote:
> > > > 
> > > > The code uses atomic_inc for restoration of bi_remaining. This patch
> > > > changes it to atomic_set(1) to avoid an interlocked instruction. In the
> > > > target's bi_endio routine we are sure that bi_remaining is zero
> > > > (otherwise, the bi_endio routine wouldn't be called) and there are no
> > > > concurrent users of the bio, so we can replace atomic_inc with
> > > > atomic_set(1).
> > > 
> > > This isn't DM-specific.  Shouldn't the other places in the tree that use
> > > atomic_inc on bi_remaining should really be converted at the same time?
> > 
> > There is no 'atomic_inc.*bi_remaining' in other drivers.
> 
> Wrong.  I know btrfs has at least one.  As does bcache afaik.

grep -r 'atomic_inc.*bi_remaining' * yilds no hits in btrfs or bcache (on 
git://git.kernel.org/pub/scm/linux/kernel/git/axboe/linux-block.git , 
branch remotes/origin/for-next). It only finds 
drivers/md/dm-cache-target.c, drivers/md/dm-verity.c, 
drivers/md/dm-snap.c, drivers/md/dm-thin.c. Maybe in other git trees there 
are more cases of this.

Mikulas