On Tue, Dec 07, 2010 at 11:10:22AM +0900, Jassi Brar wrote:
On Mon, Dec 6, 2010 at 9:46 PM, Mark Brown
In order to do this you need to tell the clock framework about the ASoC virtual device - once you're doing stuff like that you may as well just fix the S/PDIF driver to let you do what you need.
Assuming I understand you correctly, the problem still remains of having the S/PDIF driver to setup clock hierarchies -- which is the job of the board init code, IMHO. So until we have proper clock subsystem and board init code to handle such scenarios we have to do that in ASoC machine driver in such hackish way.
It may be that what's needed in the S/PDIF driver is to provide a way to get the device for the S/PDIF controller so that the machine driver can fiddle with it. Actually, you can do that anyway as the device will be associated with the DAI - the machine driver can just look at dai->dev to get the struct device for the S/PDIF controller.