Hello.
On Wed, Jul 10, 2024 at 10:51:53PM GMT, Waiman Long <longman@xxxxxxxxxx> wrote:
> As cgroup v2 had deprecated the use of /proc/cgroups, the hierarchical
> cgroup.stat file is now being extended to show the number of live and
> dying CSSes associated with all the non-inhibited cgroup subsystems
> that have been bound to cgroup v2 as long as it is not zero. The number
> includes CSSes in the current cgroup as well as in all the descendants
> underneath it. This will help us pinpoint which subsystems are
> responsible for the increasing number of dying (nr_dying_descendants)
> cgroups.
This implementation means every onlining/offlining (only additionally)
contends in root's css updates (even when stats aren't ever read).
There's also 'debug' subsys. Have you looked at (extending) that wrt
dying csses troubleshooting?
It'd be good to document here why you decided against it.
> --- a/kernel/cgroup/cgroup.c
> +++ b/kernel/cgroup/cgroup.c
> @@ -3669,12 +3669,36 @@ static int cgroup_events_show(struct seq_file *seq, void *v)
> static int cgroup_stat_show(struct seq_file *seq, void *v)
> {
> struct cgroup *cgroup = seq_css(seq)->cgroup;
> + struct cgroup_subsys_state *css;
> + int ssid;
>
> seq_printf(seq, "nr_descendants %d\n",
> cgroup->nr_descendants);
> seq_printf(seq, "nr_dying_descendants %d\n",
> cgroup->nr_dying_descendants);
>
> + /*
> + * Show the number of live and dying csses associated with each of
> + * non-inhibited cgroup subsystems bound to cgroup v2 if non-zero.
> + *
> + * Without proper lock protection, racing is possible. So the
> + * numbers may not be consistent when that happens.
> + */
> + rcu_read_lock();
> + for_each_css(css, ssid, cgroup) {
> + if ((BIT(ssid) & cgrp_dfl_inhibit_ss_mask) ||
> + (cgroup_subsys[ssid]->root != &cgrp_dfl_root))
> + continue;
Is this taken? (Given cgroup.stat is only on the default hierarchy.)
Thanks,
Michal
Attachment:
signature.asc
Description: PGP signature
