Re: [PATCH v2 3/3] cgroup, docs: cpu controller interaction with various scheduling policies

[Tejun Heo <tj@kernel.org>]

Hello,

On Tue, May 20, 2025 at 11:07:47PM +0900, Shashank Balaji via B4 Relay wrote:
...
> +The interaction of a process with the cpu controller depends on its scheduling
> +policy. We have the following scheduling policies: ``SCHED_IDLE``, ``SCHED_BATCH``,
> +``SCHED_OTHER``, ``SCHED_EXT`` (if ``CONFIG_SCHED_CLASS_EXT`` is enabled), ``SCHED_FIFO``,
> +``SCHED_RR``, and ``SCHED_DEADLINE``. ``SCHED_{IDLE,BATCH,OTHER,EXT}`` can be scheduled
> +either by the fair-class scheduler or by a BPF scheduler::
> +
> +    CONFIG_SCHED_CLASS_EXT
> +    ├─ Disabled
> +    |   └─ SCHED_{IDLE,BATCH,OTHER} -> fair-class scheduler
> +    └─ Enabled
> +        ├─ BPF scheduler disabled
> +        |   └─ SCHED_{IDLE,BATCH,OTHER,EXT} -> fair-class scheduler
> +        ├─ BPF scheduler without SCX_OPS_SWITCH_PARTIAL enabled
> +        |   └─ SCHED_{IDLE,BATCH,OTHER,EXT} -> BPF scheduler
> +        └─ BPF scheduler with SCX_OPS_SWITCH_PARTIAL enabled
> +            ├─ SCHED_{IDLE,BATCH,OTHER} -> fair-class scheduler
> +            └─ SCHED_EXT -> BPF scheduler
> +
> +For more details on ``SCHED_EXT``, check out :ref:`Documentation/scheduler/sched-ext.rst. <sched-ext>`
> +From the point of view of the cpu controller, processes can be categorized as
> +follows:
> +
> +* Processes under the fair-class scheduler
> +* Processes under a BPF scheduler with the ``cgroup_set_weight`` callback
> +* Everything else: ``SCHED_{FIFO,RR,DEADLINE}`` and processes under a BPF scheduler
> +  without the ``cgroup_set_weight`` callback
> +
> +Note that the ``cgroup_*`` family of callbacks require ``CONFIG_EXT_GROUP_SCHED``
> +to be enabled. For each of the following interface files, the above categories
> +will be referred to. All time durations are in microseconds.

Can we document the above in sched_ext documentation and point to it from
here? Documenting sched_ext details here seems a bit out of place and prone
to becoming stale over time.

...
>    cpu.uclamp.min
> -        A read-write single value file which exists on non-root cgroups.
> -        The default is "0", i.e. no utilization boosting.
> +	A read-write single value file which exists on non-root cgroups.
> +	The default is "0", i.e. no utilization boosting.

Can you please separate out indentation changes to a separate patch? These
usually make reviewing tricky.

Thanks.

-- 
tejun