Re: [PATCH v9] nvmet: add cgroup_id to charge namespace I/O to a cgroup

[Tejun Heo <tj@kernel.org>]

Hello, Peng.

On Thu, Oct 22, 2026 at 09:34:23PM -0700, Peng Yu wrote:
> +	/*
> +	 * The I/O commands of a namespace with a cgroup, except Reservation
> +	 * commands, are executed from ->blkcg_work on ->blkcg_wq, so the
> +	 * cgroup's throttling doesn't put a transport context to sleep.
> +	 */
> +	struct workqueue_struct	*blkcg_wq;
> +	struct llist_head	blkcg_reqs;
> +	struct work_struct	blkcg_work;

All commands of the namespace go through the single blkcg_work, so they
execute one at a time on one worker. Can't each command embed its own
work item and be queued on blkcg_wq instead? For an unbound workqueue,
max_active applies to the whole system, so blkcg_wq's max_active can
just be set to the number of CPUs. That would be simpler and might
remove the performance degradation too.

Thanks.

--
tejun