Re: [PATCH v1 3/5] dt-bindings: arm: Add cpu-idle-states to Tegra194 CPU nodes

[Sudeep Holla <sudeep.holla@arm.com>]

On Fri, Mar 12, 2021 at 02:27:30PM -0800, Sowjanya Komatineni wrote:
> Hi Sudeep,
>
> To make our driver PSCI compliant below are few updates to be done
>
> 1. Standardize passing residency time run-time thru PSCI to ATF.
>

Yes that was my initial understanding, but your previous response was
confusing. I should have read this first.

>     From PSCI specification I only see PSCI_STAT_RESIDENCY and
> PSCI_STAT_COUNT optional functions for PSCI1.0/PSCI1.1
>

Indeed, we don't have any support to pass run-time residency hints in PSCI
today.

>    Can you please help add right people to explore on possible ways to add
> PSCI function for passing corresponding state residency time to ATF?
>

Before we jump to implementation in TF-A we need to get agreement to get this
added to the specification to support in OSPM/Linux. TF-A is just one
implementation of PSCI and may not be only one.

Formally you can raise any specification related queries to
https://developer.arm.com/support or support@arm.com. I will ask the author
of PSCI specification internally to take a look at this thread and chime in
if they can.

> 2. Tegra CPU Idle support definitely need to pass deepest cluster state and
> threshold to MCE firmware from DT and looks like we can move this
> implementation to ATF
>

Yes, I just asked the same question as response to your earlier email. Thanks
for confirming that it can be moved out of OSPM/Linux and DT

>      With both of the above implementation changes Tegra194 driver will be
> PSCI compliant.
>

We still need to get agreement on the specification front 😉.

--
Regards,
Sudeep