Re: [PATCH 3/3] arm64: dts: update the cpu idle node

[Sudeep Holla]

(sorry for replying so late, just found this by accident when clearing
 old emails)

On 08/02/18 07:54, Ran Wang wrote:
> From: Yuantian Tang <andy.tang@xxxxxxx>
> 
> According to PSCI standard v0.2, for CPU_SUSPEND call, which is
> used by cpu idle framework, bit[16] of state parameter must be 0.
> So update bit[16] of property 'arm,psci-suspend-param', which is
> used as state parameter, to 0.
> 
This changelog makes zero sense to me.

1. PSCI specification makes no reference to "cpu idle framework".
   It's not a Linux specification to reference anything specific to
   Linux CPUIdle framework.

2. "for CPU_SUSPEND call,...bit[16] of state parameter must be 0."
   Where exactly to you see that ? Specification tells what value of
   0 or 1 in bit[16] means.

3. Is this retention state and doesn't need any save restore ? If not
   this change is wrong and will break if and when we use
   CPU_PM_CPU_IDLE_ENTER_RETENTION if this bit is not set.

-- 
Regards,
Sudeep