On Mon, May 04, 2015 at 09:13:48PM +0800, Jun Nie wrote:

2015-05-01 18:55 GMT+08:00 Mark Brown broonie@kernel.org:

...

This is adding a new device with DT bindings but there is no bindings document. All new DT bindings must be documented.

Plan to add a separate patch to device tree. Or you prefer to add device tree patch in this patch set and loop device tree maintainer?

You *always* need to document new device tree bindings at the same time as adding new code.

There are more than one SPDIF controller on ZTE SOC. One of them support HDMI output, while the other SPDIF controller support direct output. So I want to separate SPDIF and HDMI and add SPDIF id 0 test

So this special HDMI S/PDIF controller is directly integrated with the HDMI IP in the SoC and not usable as a generic S/PDIF controller? That seems to mirror system designs which have an external HDMI encoder which use S/PDIF or I2S to connect the audio portion to an external device.

Do you have any suggestion on what driver to look at for the interface between audio and video?

Half the problem right now is that the people working on HDMI don't appear to be talking to each other very much so everything is very inconsistent with no real code sharing. I don't really feel I've got a good enough picture of what the hardware looks like in general to be 100% clear on what the best way is forward beyond the fact that we should be sharing the EDID parsing code.

I'd suggest for future versions reorganizing things so that you add the S/PDIF driver with only S/PDIF support and then have a separate patch that layers on the HDMI support.