selftests: cgroup: make sure reclaim target memcg is unprotected

Make sure that we ignore protection of a memcg that is the target of memcg
reclaim.

Link: https://lkml.kernel.org/r/20221202031512.1365483-4-yosryahmed@google.com
Signed-off-by: Yosry Ahmed <yosryahmed@google.com>
Reviewed-by: Roman Gushchin <roman.gushchin@linux.dev>
Cc: Chris Down <chris@chrisdown.name>
Cc: David Rientjes <rientjes@google.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Muchun Song <songmuchun@bytedance.com>
Cc: Shakeel Butt <shakeelb@google.com>
Cc: Tejun Heo <tj@kernel.org>
Cc: Vasily Averin <vasily.averin@linux.dev>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Yu Zhao <yuzhao@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
This commit is contained in:
Yosry Ahmed 2022-12-02 03:15:12 +00:00 committed by Andrew Morton
parent e5d64edac6
commit 1c74697776
1 changed files with 11 additions and 0 deletions

View File

@ -238,6 +238,8 @@ static int cg_test_proc_killed(const char *cgroup)
return -1;
}
static bool reclaim_until(const char *memcg, long goal);
/*
* First, this test creates the following hierarchy:
* A memory.min = 0, memory.max = 200M
@ -266,6 +268,12 @@ static int cg_test_proc_killed(const char *cgroup)
* unprotected memory in A available, and checks that:
* a) memory.min protects pagecache even in this case,
* b) memory.low allows reclaiming page cache with low events.
*
* Then we try to reclaim from A/B/C using memory.reclaim until its
* usage reaches 10M.
* This makes sure that:
* (a) We ignore the protection of the reclaim target memcg.
* (b) The previously calculated emin value (~29M) should be dismissed.
*/
static int test_memcg_protection(const char *root, bool min)
{
@ -385,6 +393,9 @@ static int test_memcg_protection(const char *root, bool min)
if (!values_close(cg_read_long(parent[1], "memory.current"), MB(50), 3))
goto cleanup;
if (!reclaim_until(children[0], MB(10)))
goto cleanup;
if (min) {
ret = KSFT_PASS;
goto cleanup;