On 13/02/2020 14:00, Sudeep Holla wrote:
On Thu, Feb 13, 2020 at 11:52:09AM +0000, John Garry wrote:
[...]
Hi Sudeep,
As for when it's not set, it's unclear. My understanding is that the ACPI
processor id should still be used as the non-leaf node identifier, but it
would not match a UID for a processor container (as it may not exist).
I can't infer anything that matches your understanding from the spec in
this regard. If it's not set, then it is left to OSPM.
ACPI Processor ID valid just means that there is an associated processor
container entry which has a UID which matches the ACPI Processor ID for
this node.
I can't see anything to say that if the ACPI Processor ID valid flag is
unset then the ACPI processor ID itself is not still a valid identifier.
As such, it's implied that it is still valid. But the spec should be
clarified here.
The kernel does have behave according to this.
According to what in the specification ?
That's just based on my interpretation, above. That being, the ACPI
Processor ID is always valid, independent of whether ACPI Processor ID
valid flag is set.
So how I am misinterpreting this?
May be you are not, it is just ambiguous in the spec, worth checking and
fixing if it is an issue.
Absolutely agree on all of this.
I reiterate we don't want to generate anything
in the OS for this purpose.
I'm ok with that.
Cheers,
John
