Re: [PATCH for mmotm 0414] vmscan,memcg: reintroduce sc->may_swap

[Johannes Weiner <hannes@cmpxchg.org>]

Hi KOSAKI-san,

On Sat, Apr 18, 2009 at 03:26:02PM +0900, KOSAKI Motohiro wrote:
> Subject: vmscan,memcg: reintroduce sc->may_swap
> 
> vmscan-rename-scmay_swap-to-may_unmap.patch removed may_swap flag,
> but memcg had used it as a flag for "we need to use swap?", as the
> name indicate.
> 
> And in current implementation, memcg cannot reclaim mapped file caches
> when mem+swap hits the limit.
> 
> re-introduce may_swap flag and handle it at get_scan_ratio().
> This patch doesn't influence any scan_control users other than memcg.
> 
> Signed-off-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
> Signed-off-by: Daisuke Nishimura <nishimura@mxp.nes.nec.co.jp>
> --
>  mm/vmscan.c |   12 ++++++++++--
>  1 file changed, 10 insertions(+), 2 deletions(-)
> 
> Index: b/mm/vmscan.c
> ===================================================================
> --- a/mm/vmscan.c	2009-04-16 21:25:41.000000000 +0900
> +++ b/mm/vmscan.c	2009-04-16 21:56:54.000000000 +0900
> @@ -64,6 +64,9 @@ struct scan_control {
>  	/* Can mapped pages be reclaimed? */
>  	int may_unmap;
>  
> +	/* Can pages be swapped as part of reclaim? */
> +	int may_swap;
> +
>  	/* This context's SWAP_CLUSTER_MAX. If freeing memory for
>  	 * suspend, we effectively ignore SWAP_CLUSTER_MAX.
>  	 * In this context, it doesn't matter that we scan the
> @@ -1387,7 +1390,7 @@ static void get_scan_ratio(struct zone *
>  	struct zone_reclaim_stat *reclaim_stat = get_reclaim_stat(zone, sc);
>  
>  	/* If we have no swap space, do not bother scanning anon pages. */
> -	if (nr_swap_pages <= 0) {
> +	if (!sc->may_swap || (nr_swap_pages <= 0)) {

I wonder if may_swap is a good name for this effect.  See below the
__zone_reclaim() comments.

>  		percent[0] = 0;
>  		percent[1] = 100;
>  		return;
> @@ -1704,6 +1707,7 @@ unsigned long try_to_free_pages(struct z
>  		.may_writepage = !laptop_mode,
>  		.swap_cluster_max = SWAP_CLUSTER_MAX,
>  		.may_unmap = 1,
> +		.may_swap = 1,
>  		.swappiness = vm_swappiness,
>  		.order = order,
>  		.mem_cgroup = NULL,
> @@ -1724,6 +1728,7 @@ unsigned long try_to_free_mem_cgroup_pag
>  	struct scan_control sc = {
>  		.may_writepage = !laptop_mode,
>  		.may_unmap = 1,
> +		.may_swap = 1,
>  		.swap_cluster_max = SWAP_CLUSTER_MAX,
>  		.swappiness = swappiness,
>  		.order = 0,
> @@ -1734,7 +1739,7 @@ unsigned long try_to_free_mem_cgroup_pag
>  	struct zonelist *zonelist;
>  
>  	if (noswap)
> -		sc.may_unmap = 0;
> +		sc.may_swap = 0;

Can this be directly initialized?

struct scan_control sc = {
	...
	.may_swap = !noswap,
	...
};

>  	sc.gfp_mask = (gfp_mask & GFP_RECLAIM_MASK) |
>  			(GFP_HIGHUSER_MOVABLE & ~GFP_RECLAIM_MASK);
> @@ -1774,6 +1779,7 @@ static unsigned long balance_pgdat(pg_da
>  	struct scan_control sc = {
>  		.gfp_mask = GFP_KERNEL,
>  		.may_unmap = 1,
> +		.may_swap = 1,
>  		.swap_cluster_max = SWAP_CLUSTER_MAX,
>  		.swappiness = vm_swappiness,
>  		.order = order,
> @@ -2120,6 +2126,7 @@ unsigned long shrink_all_memory(unsigned
>  	struct scan_control sc = {
>  		.gfp_mask = GFP_KERNEL,
>  		.may_unmap = 0,
> +		.may_swap = 1,

shrink_all_memory() is not a user of shrink_zone() -> get_scan_ratio()
and therefor not affected by this flag.  I think it's better not to
set it here (just like sc->swappiness).

>  		.may_writepage = 1,
>  		.isolate_pages = isolate_pages_global,
>  	};
> @@ -2304,6 +2311,7 @@ static int __zone_reclaim(struct zone *z
>  	struct scan_control sc = {
>  		.may_writepage = !!(zone_reclaim_mode & RECLAIM_WRITE),
>  		.may_unmap = !!(zone_reclaim_mode & RECLAIM_SWAP),
> +		.may_swap = 1,

Shouldn't this be set to !!(zone_reclaim_mode & RECLAIM_SWAP) as well?

With set to 1, zone_reclaim() will also reclaim unmapped swap cache
pages (without swapping) and it might be desirable to do that.  But
then may_swap is a confusing name.  may_anon?  may_scan_anon?
scan_anon?

	Hannes