On Thu, Nov 06, 2014 at 01:46:00PM +0000, Mark Brown wrote:
On Thu, Nov 06, 2014 at 06:41:42PM +0530, Vinod Koul wrote:
On Thu, Nov 06, 2014 at 12:48:54PM +0000, Mark Brown wrote:
Don't bother with the wrapper functions, they're not adding anything. Why are we using prepare() and complete() here - other machine drivers don't need to do that? Comments might be helpful...
due to I2C. We have seen that codec is resumed but I2C is still not ready causing i2c failures, so moving to complete and prepare helps. I will add this comment. Will remove wrappers.
That doesn't sound right - this would affect almost all systems. Deferred probe should ensure that the machine driver can't load until after all the component devices (including the CODEC driver) are instantiated and we should be suspending in the reverse order that we probe the devices.
Yes defer probe is the right solution here. But in machine driver loading how does it ensure that codec and platform are already loaded, should the sound card registration be treated as failure and deferred?
Thanks