Re: [PATCH v4 03/11] cgroup/cpuset: Drive kernel-noise housekeeping updates from isolated partitions

[Frederic Weisbecker <frederic@kernel.org>]

Le Fri, Jul 10, 2026 at 11:28:14AM +0800, Jing Wu a écrit :
> An isolated cpuset partition already updates the HK_TYPE_DOMAIN
> housekeeping mask.  Extend it to also update the kernel-noise masks
> (HK_TYPE_KERNEL_NOISE and HK_TYPE_MANAGED_IRQ) so that creating or
> destroying an isolated partition reconfigures the full set of
> housekeeping cpumasks.
> 
> The sched domain mask is updated first because the workqueue flush and
> timer migration paths depend on it; the kernel-noise masks are updated
> afterwards via housekeeping_update_types().
> 
> housekeeping_update() and housekeeping_update_types() are called after
> dropping cpus_read_lock and cpuset_mutex, with only cpuset_top_mutex held
> for mutual exclusion.
> 
> Co-developed-by: Qiliang Yuan <yuanql9@chinatelecom.cn>
> Signed-off-by: Qiliang Yuan <yuanql9@chinatelecom.cn>
> Signed-off-by: Jing Wu <realwujing@gmail.com>
> ---
>  kernel/cgroup/cpuset.c | 23 +++++++++++++++++++++--
>  1 file changed, 21 insertions(+), 2 deletions(-)
> 
> diff --git a/kernel/cgroup/cpuset.c b/kernel/cgroup/cpuset.c
> index 5c33ab20cc208..80f43a24d3c8a 100644
> --- a/kernel/cgroup/cpuset.c
> +++ b/kernel/cgroup/cpuset.c
> @@ -1347,17 +1347,36 @@ static void cpuset_update_sd_hk_unlock(void)
>  		rebuild_sched_domains_locked();
>  
>  	if (update_housekeeping) {
> +		static const unsigned long noise_types =
> +			BIT(HK_TYPE_KERNEL_NOISE) | BIT(HK_TYPE_MANAGED_IRQ);
> +
>  		update_housekeeping = false;
>  		cpumask_copy(isolated_hk_cpus, isolated_cpus);
>  
> +		mutex_unlock(&cpuset_mutex);
> +		cpus_read_unlock();
> +
>  		/*
>  		 * housekeeping_update() is now called without holding
>  		 * cpus_read_lock and cpuset_mutex. Only cpuset_top_mutex
>  		 * is still being held for mutual exclusion.
>  		 */
> -		mutex_unlock(&cpuset_mutex);
> -		cpus_read_unlock();
> +
> +		/*
> +		 * Update the sched domain mask first; it must succeed
> +		 * before the kernel-noise types because workqueue flush
> +		 * and timer migration depend on the sched domain mask.
> +		 */
>  		WARN_ON_ONCE(housekeeping_update(isolated_hk_cpus));
> +
> +		/*
> +		 * Update the kernel-noise housekeeping masks
> +		 * (HK_TYPE_KERNEL_NOISE and HK_TYPE_MANAGED_IRQ).  The tick,
> +		 * RCU and managed-interrupt state is reconfigured as the
> +		 * affected CPUs are cycled through the CPU hotplug machinery.
> +		 */
> +		WARN_ON_ONCE(housekeeping_update_types(noise_types,
> +						       isolated_hk_cpus));

We rely on the fact that target CPUs are offline in order to set them
as HK_TYPE_KERNEL_NOISE, right? So, should we keep cpus_read_lock held
to prevent from concurrent hotplug events?

Thanks.

>  		mutex_unlock(&cpuset_top_mutex);
>  	} else {
>  		cpuset_full_unlock();
> 
> -- 
> 2.43.0
> 

-- 
Frederic Weisbecker
SUSE Labs