Re: [PATCH] ACPI: emits change uevents to all physical companion devices of container's children

[Rafael J. Wysocki]

On Mon, Apr 3, 2017 at 5:55 PM, Lee, Chun-Yi <joeyli.kernel@xxxxxxxxx> wrote:
> The caa73ea1 patch, "ACPI / hotplug / driver core: Handle containers
> in a special way", introduced the offline callback of acpi container.
> In the patch description, it mentions:
>
>     For ACPI containers that callback simply walks the list of ACPI
>     device objects right below the container object (its children) and
>     checks if all of their physical companion devices are offline.  If
>     that's not the case, it returns -EBUSY and the container system
>     devivce cannot be put offline.  Consequently, to put the container
>     system device offline, it is necessary to put all of the physical
>     devices depending on its ACPI companion object offline beforehand.
>
> Looks that it means acpi_container_offline() should walks all physical
> companion devices of container's children and checks their offline
> state. And, the comment in source code is "Check all of the dependent
> devices' physical companions", which means it should checks _all_
> physical companions.
>
> But, the checking code just stops at the first not-offlined physical
> companion device of the first not-offlined child, then kernel only
> emits KOBJ_CHANGE uevent to the one device. It doesn't really walk
> all children's all physical companion devices and doesn't send change
> uevent to them.

It is unclear to me from the description whether or not this is a
practical issue.

Also there is an alternative, which is not to send KOBJ_CHANGE uevents
to any children at all.

Why is the approach you chose better?

Thanks,
Rafael