Re: [RFC/PATCH 2/6] memcg: remove refcnt

From: KAMEZAWA Hiroyuki
Date: Wed May 14 2008 - 23:29:25 EST


On Thu, 15 May 2008 10:57:40 +0900
KAMEZAWA Hiroyuki <kamezawa.hiroyu@xxxxxxxxxxxxxx> wrote:
> > > #ifdef CONFIG_DEBUG_VM
> > > Index: linux-2.6.26-rc2/mm/shmem.c
> > > ===================================================================
> > > --- linux-2.6.26-rc2.orig/mm/shmem.c
> > > +++ linux-2.6.26-rc2/mm/shmem.c
> > > @@ -961,13 +961,14 @@ found:
> > > shmem_swp_unmap(ptr);
> > > spin_unlock(&info->lock);
> > > radix_tree_preload_end();
> > > -uncharge:
> > > - mem_cgroup_uncharge_page(page);
> > > out:
> > > unlock_page(page);
> > > page_cache_release(page);
> > > iput(inode); /* allows for NULL */
> > > return error;
> > > +uncharge:
> > > + mem_cgroup_uncharge_cache_page(page);
> > > + goto out;
> > > }
> > >
> >
> > Seems the logic is changed here. is it intended ?
> >
> intended. (if success, uncharge is not necessary because there is no refcnt.
> I'll add comment.
>
But, it seems patch 6/6 doesn't seem to be optimal in this case.
and have some troubles...I'll find a workaround.
It seems that shmem's memcg is complicated...

Thanks,
-Kame


--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/