On Thu, Jul 21, 2016 at 03:11:01PM +0100, Mel Gorman wrote:
> Page reclaim determines whether a pgdat is unreclaimable by examining how
> many pages have been scanned since a page was freed and comparing that to
> the LRU sizes. Skipped pages are not reclaim candidates but contribute to
> scanned. This can prematurely mark a pgdat as unreclaimable and trigger
> an OOM kill.
>
> This patch accounts for skipped pages as a partial scan so that an
> unreclaimable pgdat will still be marked as such but by scaling the cost
> of a skip, it'll avoid the pgdat being marked prematurely.
>
> Signed-off-by: Mel Gorman <mgorman@xxxxxxxxxxxxxxxxxxx>
> ---
> mm/vmscan.c | 20 ++++++++++++++++++--
> 1 file changed, 18 insertions(+), 2 deletions(-)
>
> diff --git a/mm/vmscan.c b/mm/vmscan.c
> index 6810d81f60c7..e5af357dd4ac 100644
> --- a/mm/vmscan.c
> +++ b/mm/vmscan.c
> @@ -1424,7 +1424,7 @@ static unsigned long isolate_lru_pages(unsigned long nr_to_scan,
> LIST_HEAD(pages_skipped);
>
> for (scan = 0; scan < nr_to_scan && nr_taken < nr_to_scan &&
> - !list_empty(src); scan++) {
> + !list_empty(src);) {
> struct page *page;
>
> page = lru_to_page(src);
> @@ -1438,6 +1438,12 @@ static unsigned long isolate_lru_pages(unsigned long nr_to_scan,
> continue;
> }
>
> + /*
> + * Account for scanned and skipped separetly to avoid the pgdat
> + * being prematurely marked unreclaimable by pgdat_reclaimable.
> + */
> + scan++;
> +
> switch (__isolate_lru_page(page, mode)) {
> case 0:
> nr_pages = hpage_nr_pages(page);
> @@ -1465,14 +1471,24 @@ static unsigned long isolate_lru_pages(unsigned long nr_to_scan,
> */
> if (!list_empty(&pages_skipped)) {
> int zid;
> + unsigned long total_skipped = 0;
>
> - list_splice(&pages_skipped, src);
> for (zid = 0; zid < MAX_NR_ZONES; zid++) {
> if (!nr_skipped[zid])
> continue;
>
> __count_zid_vm_events(PGSCAN_SKIP, zid, nr_skipped[zid]);
> + total_skipped += nr_skipped[zid];
> }
> +
> + /*
> + * Account skipped pages as a partial scan as the pgdat may be
> + * close to unreclaimable. If the LRU list is empty, account
> + * skipped pages as a full scan.
> + */
node-lru made OOM detection lengthy because a freeing of any zone will
reset NR_PAGES_SCANNED easily so that it's hard to meet a situation
pgdat_reclaimable returns *false*.
When I perform stress test, it seems I encounter the situation easily
although I have no number now.
Anyway, this patch makes sense to me because it's better than now.
About accounting scan, I supports this idea.
But still, I doubt it's okay to continue skipping pages under
irq-disabled-spin lock without any condition.
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@xxxxxxxxx. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@xxxxxxxxx";> email@xxxxxxxxx </a>