On Thu, May 09, 2013 at 10:06:19AM +0400, Glauber Costa wrote:
> In very low free kernel memory situations, it may be the case that we
> have less objects to free than our initial batch size. If this is the
> case, it is better to shrink those, and open space for the new workload
> then to keep them and fail the new allocations. For the purpose of
> defining what "very low memory" means, we will purposefuly exclude
> kswapd runs.
>
> More specifically, this happens because we encode this in a loop with
> the condition: "while (total_scan >= batch_size)". So if we are in such
> a case, we'll not even enter the loop.
>
> This patch modifies turns it into a do () while {} loop, that will
> guarantee that we scan it at least once, while keeping the behaviour
> exactly the same for the cases in which total_scan > batch_size.
>
> [ v5: differentiate no-scan case, don't do this for kswapd ]
>
> Signed-off-by: Glauber Costa <glommer@xxxxxxxxxx>
> Reviewed-by: Dave Chinner <david@xxxxxxxxxxxxx>
> Reviewed-by: Carlos Maiolino <cmaiolino@xxxxxxxxxx>
> CC: "Theodore Ts'o" <tytso@xxxxxxx>
> CC: Al Viro <viro@xxxxxxxxxxxxxxxxxx>
> ---
> mm/vmscan.c | 24 +++++++++++++++++++++---
> 1 file changed, 21 insertions(+), 3 deletions(-)
>
> diff --git a/mm/vmscan.c b/mm/vmscan.c
> index fa6a853..49691da 100644
> --- a/mm/vmscan.c
> +++ b/mm/vmscan.c
> @@ -281,12 +281,30 @@ unsigned long shrink_slab(struct shrink_control *shrink,
> nr_pages_scanned, lru_pages,
> max_pass, delta, total_scan);
>
> - while (total_scan >= batch_size) {
> + do {
> int nr_before;
>
> + /*
> + * When we are kswapd, there is no need for us to go
> + * desperate and try to reclaim any number of objects
> + * regardless of batch size. Direct reclaim, OTOH, may
> + * benefit from freeing objects in any quantities. If
> + * the workload is actually stressing those objects,
> + * this may be the difference between succeeding or
> + * failing an allocation.
> + */
> + if ((total_scan < batch_size) && current_is_kswapd())
> + break;
> + /*
> + * Differentiate between "few objects" and "no objects"
> + * as returned by the count step.
> + */
> + if (!total_scan)
> + break;
> +
To reduce the risk of slab reclaiming the world in the reasonable cases
I outlined after the leader mail, I would go further than this and either
limit it to memcg after shrinkers are memcg aware or only do the full scan
if direct reclaim and priority == 0.
What do you think?
--
Mel Gorman
SUSE Labs
--
To unsubscribe from this list: send the line "unsubscribe linux-fsdevel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
