Re: [PATCH 2/2] memcg fix stale swap cache account leak v6

[Ingo Molnar <mingo@elte.hu>]

x
* KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com> wrote:

> +struct swapio_check {
> +	spinlock_t	lock;
> +	void		*swap_bio_list;
> +	struct delayed_work work;
> +} stale_swap_check;

Small nit. It's nice that you lined up the first two fields, but it 
would be nice to line up the third one too:

struct swapio_check {
	spinlock_t		lock;
	void			*swap_bio_list;
	struct delayed_work	work;
} stale_swap_check;

> +	while (nr--) {
> +		cond_resched();
> +		spin_lock_irq(&sc->lock);
> +		bio = sc->swap_bio_list;

> @@ -66,6 +190,7 @@ static void end_swap_bio_write(struct bi
>  				(unsigned long long)bio->bi_sector);
>  		ClearPageReclaim(page);
>  	}
> +	mem_cgroup_swapio_check_again(bio, page);

Hm, this patch adds quite a bit of scanning overhead to 
end_swap_bio_write(), to work around artifacts of a global LRU not 
working well with a partitioned system's per-partition LRU needs.

Isnt the right solution to have a better LRU that is aware of this, 
instead of polling around in the hope of cleaning up stale entries?

	Ingo

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>