Re: [PATCH] memcg: use ID in page_cgroup

From: Daisuke Nishimura
Date: Mon Aug 23 2010 - 01:53:10 EST


On Fri, 20 Aug 2010 19:01:32 +0900
KAMEZAWA Hiroyuki <kamezawa.hiroyu@xxxxxxxxxxxxxx> wrote:

>
> I have an idea to remove page_cgroup->page pointer, 8bytes reduction per page.
> But it will be after this work.
Another off topic. I think we can reduce the size of mem_cgroup by packing
some boolean members into one "unsinged long flags".

> @@ -300,12 +300,13 @@ static atomic_t mem_cgroup_num;
> #define NR_MEMCG_GROUPS (CONFIG_MEM_CGROUP_MAX_GROUPS + 1)
> static struct mem_cgroup *mem_cgroups[NR_MEMCG_GROUPS] __read_mostly;
>
> -/* Must be called under rcu_read_lock */
> -static struct mem_cgroup *id_to_memcg(unsigned short id)
> +/* Must be called under rcu_read_lock, set safe==true if under lock */
Do you mean, "Set safe==true if we can ensure by some locks that the id can be
safely dereferenced without rcu_read_lock", right ?

> +static struct mem_cgroup *id_to_memcg(unsigned short id, bool safe)
> {
> struct mem_cgroup *ret;
> /* see mem_cgroup_free() */
> - ret = rcu_dereference_check(mem_cgroups[id], rch_read_lock_held());
> + ret = rcu_dereference_check(mem_cgroups[id],
> + rch_read_lock_held() || safe);
> if (likely(ret && ret->valid))
> return ret;
> return NULL;

(snip)
> @@ -723,6 +729,11 @@ static inline bool mem_cgroup_is_root(st
> return (mem == root_mem_cgroup);
> }
>
> +static inline bool mem_cgroup_is_rootid(unsigned short id)
> +{
> + return (id == 1);
> +}
> +
It might be better to add

BUG_ON(newid->id != 1)

in cgroup.c::cgroup_init_idr().


Thanks,
Daisuke Nishimura.
--
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/