Re: [PATCH] backing dev unplugging

[Jens Axboe <axboe@suse.de>]

On Wed, Mar 10 2004, Andrew Morton wrote:
> 
> (Please use reply-to-all)
> 
> "Miquel van Smoorenburg" <miquels@cistron.nl> wrote:
> >
> > In article <20040310210207.GL15087@suse.de>,
> > Jens Axboe  <axboe@suse.de> wrote:
> > >On Wed, Mar 10 2004, Andrew Morton wrote:
> > >> Jens Axboe <axboe@suse.de> wrote:
> > >> >
> > >> > Here's a first cut at killing global plugging of block devices to reduce
> > >> > the nasty contention blk_plug_lock caused.
> > >> 
> > >> Shouldn't we take read_lock(&md->map_lock) in dm_table_unplug_all()?
> > >
> > >Ugh yes, we certainly should.
> > 
> > With the latest patches from Joe it would be more like
> > 
> > 	map = dm_get_table(md);
> > 	if (map) {
> > 		dm_table_unplug_all(map);
> > 		dm_table_put(map);
> > 	}
> > 
> > No lock ranking issues, you just get a refcounted map (table, really).
> 
> Ah, OK.  Jens, you'll be needing this (on rc2-mm1):
> 
> dm.c: protect md->map with a rw spin lock rather than the md->lock
> semaphore.  Also ensure that everyone accesses md->map through
> dm_get_table(), rather than directly.

Neato, much better. I'll build on top of that.

-- 
Jens Axboe