[Devel] Re: [RFC/PATCH] cgroup swap subsystem

[KAMEZAWA Hiroyuki]

On Wed, 05 Mar 2008 14:59:05 +0900
Daisuke Nishimura <nishimura at mxp.nes.nec.co.jp> wrote:

> +int swap_cgroup_charge(struct page *page,
> +			struct swap_info_struct *si,
> +			unsigned long offset)
> +{
> +	int ret;
> +	struct page_cgroup *pc;
> +	struct mm_struct *mm;
> +	struct swap_cgroup *swap;
> +
> +	BUG_ON(!page);
> +
> +	/*
> +	 * Pages to be swapped out should have been charged by memory cgroup,
> +	 * but very rarely, pc would be NULL (pc is not reliable without lock,
> +	 * so I should fix here).
> +	 * In such cases, we charge the init_mm now.
> +	 */
> +	pc = page_get_page_cgroup(page);
> +	if (WARN_ON(!pc))
> +		mm = &init_mm;
> +	else
> +		mm = pc->pc_mm;
> +	BUG_ON(!mm);
> +
> +	rcu_read_lock();
> +	swap = rcu_dereference(mm->swap_cgroup);
> +	rcu_read_unlock();
> +	BUG_ON(!swap);
> +
> +	ret = res_counter_charge(&swap->res, PAGE_SIZE);
> +	if (!ret) {
> +		css_get(&swap->css);
> +		si->swap_cgroup[offset] = swap;
> +	}
> +
I think it's better to reclaim swap_entry used for SwapCache but not in Harddisk
before failure.

Thanks,
-Kame

_______________________________________________
Containers mailing list
Containers at lists.linux-foundation.org
https://lists.linux-foundation.org/mailman/listinfo/containers