On Tue, Feb 25, 2014 at 12:09:47PM +0900, Mark Brown wrote:
On Tue, Feb 25, 2014 at 10:38:41AM +0800, Nicolin Chen wrote:
Hmm...my words might not be so clear last time: we have to handle the dividers of ESAI in ESAI driver because the dividers is in the ESAI's IP, not in the SoC clock controlling unit. So it's hard to get them visible in the clock tree.
We should be able to arrange to have the ESAI be a clock provider shouldn't we? If the clocks need to interface to other things (and they do) then we should be able to use the standard interface we have to clocks.
Excuse me, I think I just don't get the approach how to use clock API to standardize the MCLK derived from ESAI IP: it's surely not a fixed-clock so we couldn't register in DT and it's not SoC internal clock which means we couldn't register it in the SoC clock driver either.
Could you please shed me some light on it?
Thank you, Nicolin Chen