[alsa-devel] Sharing PXA SSP ports
The current API for requesting an SSP port on PXA platforms assumes that the caller of pxa_ssp_request() is the only user of this port, so both input and output streams must be implemented in the same driver.
In a new setup I'm working on, this is not sufficient as we connect one codec to the input stream and another one to the output stream of the same CPU DAI "pxa-ssp-dai.0". I'm currently using a hack that disables the use count check in pxa_ssp_request(), but I certainly need a cleaner solution.
What would be a good approach? Would we need a multiplexer for this hardware bus, or can we extend the DAI driver in any clever way to achieve that? Or is there a solution already I just don't know about?
Thanks, Daniel
On Wed, Jun 22, 2011 at 02:44:10PM +0200, Daniel Mack wrote:
What would be a good approach? Would we need a multiplexer for this hardware bus, or can we extend the DAI driver in any clever way to achieve that? Or is there a solution already I just don't know about?
This seems like the DAI driver should deal with things. It needs to join up the input and output paths anyway as there's shared clocks and so on.
On Wed, Jun 22, 2011 at 3:05 PM, Mark Brown broonie@opensource.wolfsonmicro.com wrote:
On Wed, Jun 22, 2011 at 02:44:10PM +0200, Daniel Mack wrote:
What would be a good approach? Would we need a multiplexer for this hardware bus, or can we extend the DAI driver in any clever way to achieve that? Or is there a solution already I just don't know about?
This seems like the DAI driver should deal with things. It needs to join up the input and output paths anyway as there's shared clocks and so on.
So does it need modification, or do you say it should work already? I can care for fixing this up if necessary - is there any other DAI driver that does that and that could serve as an example?
Thanks, Daniel
On Wed, Jun 22, 2011 at 03:09:06PM +0200, Daniel Mack wrote:
On Wed, Jun 22, 2011 at 3:05 PM, Mark Brown
This seems like the DAI driver should deal with things. ?It needs to join up the input and output paths anyway as there's shared clocks and so on.
So does it need modification, or do you say it should work already? I can care for fixing this up if necessary - is there any other DAI driver that does that and that could serve as an example?
There's probably some of the i.MX drivers doing this. Depending on the driver and hardware implementation it can just work with no special code.
participants (2)
-
Daniel Mack -
Mark Brown