Re: [LTP] memcg/functional: fix memcg_failcnt intermittent timeout by disabling swap

[Li Wang <li.wang@linux.dev>]

Hi Xuewen, Andrea,

On Fri, Aug 07, 2026 at 05:34:49AM +0000, Andrea Cervesato wrote:
> Hi Li,
> 
> > Fair point that ignoring the return status defeats the purpose. But I'd
> > rather not TBROK on a swapoff failure, that's an environment limitation,
> > not a test failure. Especially swapoff here is an enhancement but not
> > a hard requirement for this test.
> 
> I agree.
> 
> > That is a bit over-engineering.
> > 
> > If an inactive fstab entry gets activated by swapon -a, that arguably
> > points to a misconfigured test host rather than something this test
> > should work around.
> 
> And also here. Unfortunately we don't have enough data for shell tests,
> so AI agent is not really working well in there..
> 
> Honestly I don't know how to fix this but adding more examples which
> we don't have..

No worries, AI agent being too strict in reviewing is not a bad thing.
It inspires me to think about something maybe in a different way.

Like this patch, I am now thinking if we can just limit it only
inside the cgroup:

--- a/testcases/kernel/controllers/memcg/functional/memcg_failcnt.sh
+++ b/testcases/kernel/controllers/memcg/functional/memcg_failcnt.sh
@@ -17,18 +17,17 @@ TST_CLEANUP=do_cleanup
 do_setup()
 {
        memcg_setup
-       swapoff -a
 }

 do_cleanup()
 {
-       swapon -a
        memcg_cleanup
 }

 do_test()
 {
        ROD echo $MEMORY_LIMIT \> memory.limit_in_bytes
+       ROD echo 0 \> memory.swappiness

        start_memcg_process $2 -s ${MEMORY_TO_ALLOCATE}
        ROD echo $MEMCG_PROCESS_PID \> tasks

@Xuewen, can you try this patch and see if it can solve your problem?

-- 
Regards,
Li Wang

-- 
Mailing list info: https://lists.linux.it/listinfo/ltp