Re: [PATCH v2] memcg: reduce the size of struct memcg 244-fold.

[Andrew Morton <akpm-de/tnXTf+JLsfHDXvbKv3WD2FQJk+8+b@public.gmane.org>]

On Wed, 23 Jan 2013 23:50:31 -0800
Greg Thelen <gthelen-hpIqsD4AKlfQT0dZR+AlfA@public.gmane.org> wrote:

> > --- a/mm/memcontrol.c
> > +++ b/mm/memcontrol.c
> > @@ -172,7 +172,7 @@ struct mem_cgroup_per_node {
> >  };
> >  
> >  struct mem_cgroup_lru_info {
> > -	struct mem_cgroup_per_node *nodeinfo[MAX_NUMNODES];
> > +	struct mem_cgroup_per_node *nodeinfo[0];
> 
> It seems like a VM_BUG_ON() in mem_cgroup_zoneinfo() asserting that the
> nid index is less than nr_node_ids would be good at catching illegal
> indexes.  I don't see any illegal indexes in your patch, but I fear that
> someday a MAX_NUMNODES based for() loop might sneak in.

Can't hurt ;)

> Tangential question: why use inline here?  I figure that modern
> compilers are good at making inlining decisions.

And that'll save some disk space.

This?

--- a/mm/memcontrol.c~memcg-reduce-the-size-of-struct-memcg-244-fold-fix
+++ a/mm/memcontrol.c
@@ -381,7 +381,7 @@ enum {
 		((1 << KMEM_ACCOUNTED_ACTIVE) | (1 << KMEM_ACCOUNTED_ACTIVATED))
 
 #ifdef CONFIG_MEMCG_KMEM
-static inline void memcg_kmem_set_active(struct mem_cgroup *memcg)
+static void memcg_kmem_set_active(struct mem_cgroup *memcg)
 {
 	set_bit(KMEM_ACCOUNTED_ACTIVE, &memcg->kmem_account_flags);
 }
@@ -645,6 +645,7 @@ static void drain_all_stock_async(struct
 static struct mem_cgroup_per_zone *
 mem_cgroup_zoneinfo(struct mem_cgroup *memcg, int nid, int zid)
 {
+	VM_BUG_ON((unsigned)nid >= nr_node_ids);
 	return &memcg->info.nodeinfo[nid]->zoneinfo[zid];
 }
 
_

Glauber, could you please cc me on patches more often?  It's a bit of a
pita having to go back to the mailing list to see if there has been
more dicussion and I may end up missing late review comments and acks.