Re: [PATCH 2/2] CFS CGroup: Report usage

["Paul Menage" <menage@google.com>]

On 10/22/07, Srivatsa Vaddagiri <vatsa@linux.vnet.ibm.com> wrote:
> On Mon, Oct 22, 2007 at 05:49:39PM -0700, Paul Menage wrote:
> > +static u64 cpu_usage_read(struct cgroup *cgrp, struct cftype *cft)
> > +{
> > +     struct task_group *tg = cgroup_tg(cgrp);
> > +     int i;
> > +     u64 res = 0;
> > +     for_each_possible_cpu(i) {
> > +             unsigned long flags;
> > +             spin_lock_irqsave(&tg->cfs_rq[i]->rq->lock, flags);
>
> Is the lock absolutely required here?

I'm not sure, I was hoping you or Ingo could comment on this. But some
kind of locking seems to required at least on 32-bit platforms, since
sum_exec_runtime is a 64-bit number.

>
> Hmm .. I hope the cgroup code prevents a task group from being destroyed while
> we are still reading a task group's cpu usage. Is that so?

Good point - cgroups certainly prevents a cgroup itself from being
freed while a control file is being read in an RCU section, and
prevents a task group from being destroyed when that task group has
been read via a task's cgroups pointer and the reader is still in an
RCU section, but we need a generic protection for subsystem state
objects being accessed via control files too.

Using cgroup_mutex is certainly possible for now, although more
heavy-weight than I'd like long term. Using css_get isn't the right
approach, I think - we shouldn't be able to cause an rmdir to fail due
to a concurrent read.

Paul
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/