[PATCH] memcg: do not expose uninitialized mem_cgroup_per_node to world

[Igor Mammedov]

On 06/08/2011 11:09 PM, Andrew Morton wrote:
> The original patch:
>
> --- a/mm/memcontrol.c
> +++ b/mm/memcontrol.c
> @@ -4707,7 +4707,6 @@ static int alloc_mem_cgroup_per_zone_info(struct mem_cgroup *mem, int node)
>   	if (!pn)
>   		return 1;
>
> -	mem->info.nodeinfo[node] = pn;
>   	for (zone = 0; zone<  MAX_NR_ZONES; zone++) {
>   		mz =&pn->zoneinfo[zone];
>   		for_each_lru(l)
> @@ -4716,6 +4715,7 @@ static int alloc_mem_cgroup_per_zone_info(struct mem_cgroup *mem, int node)
>   		mz->on_tree = false;
>   		mz->mem = mem;
>   	}
> +	mem->info.nodeinfo[node] = pn;
>   	return 0;
>   }
>
> looks like a really good idea.  But it needs a new changelog and I'd be
> a bit reluctant to merge it as it appears that the aptch removes our
> only known way of reproducing a bug.
>
> So for now I think I'll queue the patch up unchangelogged so the issue
> doesn't get forgotten about.
>
Problem was in rhel's xen hv.
It was missing fix for imul emulation.
Details here 
http://lists.xensource.com/archives/html/xen-devel/2011-06/msg00801.html
Thanks to Tim Deegan and everyone who was involved in the discussion.

-- 
Thanks,
  Igor