Re: [V5 PATCH 08/26] memcontrol: use N_MEMORY instead N_HIGH_MEMORY

[Michal Hocko]

On Wed 31-10-12 15:03:36, Wen Congyang wrote:
> At 10/30/2012 04:46 AM, David Rientjes Wrote:
> > On Mon, 29 Oct 2012, Lai Jiangshan wrote:
[...]
> >> In one word, we need a N_MEMORY. We just intrude it as an alias to
> >> N_HIGH_MEMORY and fix all im-proper usages of N_HIGH_MEMORY in late patches.
> >>
> > 
> > If this is really that problematic (and it appears it's not given that 
> > there are many use cases of it and people tend to get it right), then why 
> > not simply rename N_HIGH_MEMORY instead of introducing yet another 
> > nodemask to the equation?
> 
> The reason is that we need a node which only contains movable memory. This
> feature is very important for node hotplug. So we will add a new nodemask
> for movable memory. N_MEMORY contains movable memory but N_HIGH_MEMORY
> doesn't contain it.

OK, so the N_MOVABLE_MEMORY (or how you will call it) requires that all
the allocations will be migrateable?
How do you want to achieve that with the page_cgroup descriptors? (see
bellow)

On Mon 29-10-12 23:20:58, Lai Jiangshan wrote:
[...]
> diff --git a/mm/page_cgroup.c b/mm/page_cgroup.c
> index 5ddad0c..c1054ad 100644
> --- a/mm/page_cgroup.c
> +++ b/mm/page_cgroup.c
> @@ -271,7 +271,7 @@ void __init page_cgroup_init(void)
>  	if (mem_cgroup_disabled())
>  		return;
>  
> -	for_each_node_state(nid, N_HIGH_MEMORY) {
> +	for_each_node_state(nid, N_MEMORY) {
>  		unsigned long start_pfn, end_pfn;
>  
>  		start_pfn = node_start_pfn(nid);

This will call init_section_page_cgroup(pfn, nid) later which allocates
page_cgroup descriptors which are not movable. Or is there any code in
your patchset that handles this?
-- 
Michal Hocko
SUSE Labs
--
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/