[PATCH RFC] Specifying default-disabled devices

[Oliver O'Halloran]

On Fri, Sep 10, 2021 at 1:30 PM Zev Weiss <zweiss at equinix.com> wrote:
>
> The other alternative I've considered (though not actually implemented thus
> far) is to start using the "reserved" status value defined in the device-tree
> spec (section 2.3.4, [0]) to mean this -- from the wording in the spec it seems
> like a not-terribly-unreasonable interpretation, and the existing kernel &
> u-boot fdt code don't seem to make any use of it as far as I can see (though I
> don't know what might be out there in device-tree implementations floating
> around in other projects).

Nothing explicitly looks for status = "reserved" because Linux (and
u-boot apparently) ignore the device unless it's marked as okay. I
added the reserved state to the DTS spec since OpenPower platform
firmware was using it to mark devices that are owned by firmware
rather than the OS. I'd say the idea of having a bus device
instantiated for "reserved" devices is largely compatible with that
usage.

The one thing I'd be concerned about is what happens when the bus
itself can modify the device. For example, during PCI device probing
Linux will do some basic device configuration (enabling / disabling
various features in the PCIe caps) so you could have the OS modifying
a "reserved" device even if a driver is never attached to it. That's a
slightly academic problem for PCI since it doesn't even check the
status in the DT node, but it might be an issue for other busses.

Oliver