On Tue, Feb 9, 2021 at 5:27 PM Roman Gushchin [off-list ref] wrote:

On Tue, Feb 09, 2021 at 09:46:43AM -0800, Yang Shi wrote:

quoted

Use per memcg's nr_deferred for memcg aware shrinkers.  The shrinker's nr_deferred
will be used in the following cases:
    1. Non memcg aware shrinkers
    2. !CONFIG_MEMCG
    3. memcg is disabled by boot parameter

Signed-off-by: Yang Shi <redacted>
---
 mm/vmscan.c | 78 ++++++++++++++++++++++++++++++++++++++++++++---------
 1 file changed, 66 insertions(+), 12 deletions(-)

diff --git a/mm/vmscan.c b/mm/vmscan.c
index d4b030a0b2a9..748aa6e90f83 100644
--- a/mm/vmscan.c
+++ b/mm/vmscan.c

@@ -368,6 +368,24 @@ static void unregister_memcg_shrinker(struct shrinker *shrinker)
      up_write(&shrinker_rwsem);
 }

+static long count_nr_deferred_memcg(int nid, struct shrinker *shrinker,
+                                 struct mem_cgroup *memcg)

"Count" is not associated with xchg() semantics in my head, but I don't know
what's the better version. Maybe steal or pop?

It is used to retrieve the nr_deferred value. I don't think "steal" or
"pop" helps understand. Actually "count" is borrowed from
count_objects() method of shrinker.

Otherwise the patch looks good to me.

Thanks!