Created by: jeromebrunet
Hi, I'm working on platform on which several agent might be using the same clock. In this configuration, it is crucial that a call to set_state() with MOD_CLOCK_STATE_STOPPED does not stop the clock if another agent still use the clock.
I have discussed this topic with Sudeep Holla by mail. He asked me to raise a bug but I gave it a shot a tried to implement something.
I have tested this on Juno by hacking Linux CCF to manually send set_state() request to the SCP firmware. It seems to work as expected:
- Call to set_state() will increment or decrement the run_count depending on the requested state.
- Calls to the underlying clock driver is only made when the run count transition from 0 to 1 or 1 to 0.
What is missing: On start, the initial clock state is check to intitialize the run_count to 0 or 1. This init does not support delayed response ATM. It will be added before the actual submission
But first, lets see if this proposal is going in the right direction :)
Signed-off-by: Jerome Brunet jbrunet@baylibre.com