Re: [RFC PATCH v7 00/29] Hierarchical Constant Bandwidth Server

From: Tejun Heo

Date: Sat Nov 21 2026 - 12:24:44 EST


Hello, Luca.

On Fri, Nov 20, 2026 at 09:24:18PM +0100, luca abeni wrote:
> But the deadline entities serving the real-time tasks in a cgroup are
> pinned to cores (one dl entity serves the cgroup's RT rq for a core).
> So, they cannot just be added to the sum mentioned above (which works
> for dl entities which are free to migrate).
> This issue was solved by reserving a fraction of the CPU time (and the
> root cgroup parameters indicated this fraction) to RT cgropus, so that
> the SCHED_DEADLINE admission control is still the same (we just decrease
> "limit" by the fraction reserved to RT cgroups).

I see. Can the fraction be the maximum across the CPUs then? With the
root at "max", the kernel knows how much the cgroup servers have
reserved on each CPU. If "limit" is decreased by the largest of those
on all CPUs, every CPU has at least what its servers need set aside
and the admission control for tasks stays the same.

I'm not familiar with the admission test, so I may be missing
something.

Thanks.

--
tejun