On 15.08.2013 15:39, Mark Brown wrote:
On Thu, Aug 15, 2013 at 12:51:31PM +0200, Daniel Mack wrote:
Hmm, and you don't like the approach either of having upstream ports and assign users to it via phandles?
I don't know what that means, sorry.
Well, we have the 'real' devices registered in DT like this:
ssp0: ssp@41000000 { compatible = "mrvl,pxa3xx-ssp"; reg = <0x41000000 0x40>; interrupts = <24>; clock-names = "pxa27x-ssp.0"; dmas = <&dma 13 &dma 14>; dma-names = "rx", "tx"; };
And users, in that case, a ssp DAI, just reference a port like so:
ssp_dai0: ssp_dai@0 { compatible = "mrvl,pxa-ssp-dai"; port = <&ssp0>; };
Is that something we can live with?
Also note that the last patch in that series ("ASoC: pxa: add DT bindings for pxa2xx-pcm") can be omitted, as Lars pointed out. I'll have to care for a proper solution at some point.
I applied it since we were already adding a dummy device for the DAI. I can revert it if this work without that.
I'll send a cleanup patch once I have a better solution.
Daniel