On 9 July 2014 20:08, Santosh Shilimkar santosh.shilimkar@ti.com wrote:
I might be wrong but this API seems to bit short cited. We should probably handle it bit better since there are more cases instead of just checking CPU tuple. More than two CPUs may share the clock so discovering that in one iteration is better. I also think this is closely related to CPU topology.
- CPUs part of 'a cluster' shares same clock.
- Multiple clusters may share same clock
- Every CPU might be clocked from separate PLL.
All these configurations are currently supported by this series.