Re: [PATCH v2 01,03,07,08,09/12] DHM Architecture and Cpuset Integration

[Qiliang Yuan]

I agree that extending the existing "isolated" partition in cpusets is a
more integrated approach. For v3, I will explore a new mode for the
cpuset.cpus.partition interface to support dynamic nohz_full/nocb transitions.

On Wed, 15 Apr 2026 12:39:09 +0200, Frederic Weisbecker wrote:
> We plan to assume that the CPU is offline while updating HK_TYPE_KERNEL_NOISE
> through cpusets. So you shouldn't need to care about offlining here.

The CPU offline logic in the RCU reconfiguration will be streamlined,
relying on the state managed by the cpuset/hotplug core.

On Wed, 15 Apr 2026 12:50:03 +0200, Frederic Weisbecker wrote:
> Unbound workqueues and kthreads are already handled by cpuset isolated
> partitions.

Affinity management for workqueues and domains will be removed from the
DHM patches, as these are indeed already addressed by the isolated
partition logic.

Best regards,
Qiliang

---

To: Waiman Long <longman@xxxxxxxxxx>
Subject: Re: [PATCH v2 01/12] sched/isolation: Use cases for fine-grained isolation
Date: Mon, 20 Apr 2026 16:15:00 +0800

Hi Waiman,