We use the fact that cpuacct controller is implicitly enabled in cgroupv-2 by previous patch. So we can just do everyting the same as we do it on cgroup-v1 where we have both cpu and cpuacct controllers mounted together.
Note: Stats in cpu.proc.stat give per-cpu per-cgroup usages which are used in vcmmd to calculate per-numa cpu loads for it's balancing algorithm. This used both for Containers and VMs. https://virtuozzo.atlassian.net/browse/PSBM-154206 Signed-off-by: Pavel Tikhomirov <ptikhomi...@virtuozzo.com> Feature: sched: emulate virtual cpus for Containers --- kernel/sched/core.c | 4 ++++ 1 file changed, 4 insertions(+) diff --git a/kernel/sched/core.c b/kernel/sched/core.c index 98e9f78e19cf..2235f1565078 100644 --- a/kernel/sched/core.c +++ b/kernel/sched/core.c @@ -11835,6 +11835,10 @@ static struct cftype cpu_files[] = { .write = cpu_uclamp_max_write, }, #endif + { + .name = "proc.stat", + .seq_show = cpu_cgroup_proc_stat_show, + }, { } /* terminate */ }; -- 2.43.0 _______________________________________________ Devel mailing list Devel@openvz.org https://lists.openvz.org/mailman/listinfo/devel
