Hello.
I'm just flushing the simple patches to make memcontrol selftests check the
events behavior we had consensus about (test_memcg_low fails). (I've dropped to
goto macros for now.)
(test_memcg_reclaim, test_memcg_swap_max fail for me now but it's present
even before the refactoring.)
The only bigger change is adjustment of the protected values to make tests
succeed with the given tolerance.
It's based on mm-stable [1] commit e240ac52f7da. AFAIC, the fixup and partial
reverts may be folded into respective commits.
Let me know if it should be (re)based on something else.
Thanks,
Michal
[1] https://git.kernel.org/pub/scm/linux/kernel/git/akpm/mm.git/log/tools/testing/selftests/cgroup?h=mm-stable
Michal Koutný (4):
selftests: memcg: Fix compilation
selftests: memcg: Expect no low events in unprotected sibling
selftests: memcg: Adjust expected reclaim values of protected cgroups
selftests: memcg: Remove protection from top level memcg
.../selftests/cgroup/test_memcontrol.c | 59 +++++++++----------
1 file changed, 29 insertions(+), 30 deletions(-)
--
2.35.3
This fixes mis-applied changes from commit 72b1e03aa725 ("cgroup:
account for memory_localevents in test_memcg_oom_group_leaf_events()").
Signed-off-by: Michal Koutný <[email protected]>
---
.../selftests/cgroup/test_memcontrol.c | 25 +++++++++++--------
1 file changed, 14 insertions(+), 11 deletions(-)
diff --git a/tools/testing/selftests/cgroup/test_memcontrol.c b/tools/testing/selftests/cgroup/test_memcontrol.c
index 6ab94317c87b..4958b42201a9 100644
--- a/tools/testing/selftests/cgroup/test_memcontrol.c
+++ b/tools/testing/selftests/cgroup/test_memcontrol.c
@@ -1241,7 +1241,16 @@ static int test_memcg_oom_group_leaf_events(const char *root)
if (cg_read_key_long(child, "memory.events", "oom_kill ") <= 0)
goto cleanup;
- if (cg_read_key_long(parent, "memory.events", "oom_kill ") <= 0)
+ parent_oom_events = cg_read_key_long(
+ parent, "memory.events", "oom_kill ");
+ /*
+ * If memory_localevents is not enabled (the default), the parent should
+ * count OOM events in its children groups. Otherwise, it should not
+ * have observed any events.
+ */
+ if (has_localevents && parent_oom_events != 0)
+ goto cleanup;
+ else if (!has_localevents && parent_oom_events <= 0)
goto cleanup;
ret = KSFT_PASS;
@@ -1349,20 +1358,14 @@ static int test_memcg_oom_group_score_events(const char *root)
if (!cg_run(memcg, alloc_anon, (void *)MB(100)))
goto cleanup;
- parent_oom_events = cg_read_key_long(
- parent, "memory.events", "oom_kill ");
- /*
- * If memory_localevents is not enabled (the default), the parent should
- * count OOM events in its children groups. Otherwise, it should not
- * have observed any events.
- */
- if ((has_localevents && parent_oom_events == 0) ||
- parent_oom_events > 0)
- ret = KSFT_PASS;
+ if (cg_read_key_long(memcg, "memory.events", "oom_kill ") != 3)
+ FAIL(cleanup);
if (kill(safe_pid, SIGKILL))
goto cleanup;
+ ret = KSFT_PASS;
+
cleanup:
if (memcg)
cg_destroy(memcg);
--
2.35.3
Apologies for spam due to botched sending.
Please disregard this (old) series.
The replacement should come in
https://lore.kernel.org/r/[email protected]
(That one is also not 100% correct, it's missing a Subject: therefore
may not be pass through some filters.)
The 1st patch of that v2 series is at
https://lore.kernel.org/r/[email protected]/
And I copy the cover letter here to be sure (and not to spam even more).
---->----
Subject: [PATCH v2 0/5] memcontrol selftests fixups
Hello.
I'm just flushing the patches to make memcontrol selftests check the
events behavior we had consensus about (test_memcg_low fails).
(test_memcg_reclaim, test_memcg_swap_max fail for me now but it's present
even before the refactoring.)
The two bigger changes are:
- adjustment of the protected values to make tests succeed with the given
tolerance,
- both test_memcg_low and test_memcg_min check protection of memory in
populated cgroups (actually as per Documentation/admin-guide/cgroup-v2.rst
memory.min should not apply to empty cgroups, which is not the case
currently. Therefore I unified tests with the populated case in order to to
bring more broken tests).
Thanks,
Michal
Changes from v1 (https://lore.kernel.org/r/[email protected]/)
- fixed mis-rebase in compilation fix patch,
- added review, ack tags from v1,
- applied feedback from v1 (Octave script in git tree),
- added one more patch extracting common parts,
- rebased on mm-stable bbe832b9db2e.
----<----