CPU affinity behavior of liburcu call-rcu per-cpu worker threads

CPU affinity behavior of liburcu call-rcu per-cpu worker threads

[Mathieu Desnoyers via lttng-dev]

Hi Paul,

I have a question related to liburcu per-cpu call-rcu worker threads.

So far in the current liburcu releases no affinity is set when the
worker threads are started. We have to wait for 
SET_AFFINITY_CHECK_PERIOD_MASK grace periods before the affinity is set
(it takes about 2.5s in my benchmarks).

This means that short-lived test programs will get poor CPU affinity
for call-rcu worker threads at the beginning of their lifetime. So for
short lived programs, this means poor performance.

OTOH, if we have short-lived applications on a large machine, setting
the affinity immediately when the worker thread is starting means
we use CPU time on CPUs which may never be actually used by the
application (no call-rcu activity), which can be detrimental to other
use-cases as well.

So I was wondering: is the choice of skipping setting the affinity
on call-rcu worker thread startup done on purpose ? And if so,
should we perhaps consider setting the affinity as soon as the
worker thread is woken up for the first time rather than after
SET_AFFINITY_CHECK_PERIOD_MASK grace periods ?

Note that there is a commit in liburcu changing this behavior right
now, but it's not part of any release yet. I want to figure out the
wanted behavior before releasing this change:

commit 28f282af1905a1cf50ff9b5835b4cd9de4416ddf
Author: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Date:   Wed Jul 1 12:49:43 2026 -0400

     call_rcu: pin per-CPU worker at thread startup

Thanks,

Mathieu

-- 
Mathieu Desnoyers
EfficiOS Inc.
https://www.efficios.com

Re: CPU affinity behavior of liburcu call-rcu per-cpu worker threads

[Mathieu Desnoyers via lttng-dev]

On 2026-07-09 14:48, Mathieu Desnoyers wrote:
> Hi Paul,
> 
> I have a question related to liburcu per-cpu call-rcu worker threads.
> 
> So far in the current liburcu releases no affinity is set when the
> worker threads are started. We have to wait for 
> SET_AFFINITY_CHECK_PERIOD_MASK grace periods before the affinity is set
> (it takes about 2.5s in my benchmarks).
> 
> This means that short-lived test programs will get poor CPU affinity
> for call-rcu worker threads at the beginning of their lifetime. So for
> short lived programs, this means poor performance.
> 
> OTOH, if we have short-lived applications on a large machine, setting
> the affinity immediately when the worker thread is starting means
> we use CPU time on CPUs which may never be actually used by the
> application (no call-rcu activity), which can be detrimental to other
> use-cases as well.
> 
> So I was wondering: is the choice of skipping setting the affinity
> on call-rcu worker thread startup done on purpose ? And if so,
> should we perhaps consider setting the affinity as soon as the
> worker thread is woken up for the first time rather than after
> SET_AFFINITY_CHECK_PERIOD_MASK grace periods ?

I've implemented this compromise approach here:

https://review.lttng.org/c/userspace-rcu/+/18257 call-rcu worker: set CPU affinity on first non-empty dequeue

Feedback is welcome,

Thanks!

Mathieu

> 
> Note that there is a commit in liburcu changing this behavior right
> now, but it's not part of any release yet. I want to figure out the
> wanted behavior before releasing this change:
> 
> commit 28f282af1905a1cf50ff9b5835b4cd9de4416ddf
> Author: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
> Date:   Wed Jul 1 12:49:43 2026 -0400
> 
>      call_rcu: pin per-CPU worker at thread startup
> 
> Thanks,
> 
> Mathieu
> 


-- 
Mathieu Desnoyers
EfficiOS Inc.
https://www.efficios.com