[PATCH] of: Add MIPI DSI bus device tree bindings

Bert Kenward bert.kenward at broadcom.com
Tue Dec 24 00:31:08 PST 2013


Hello. Late to the party...

On Monday December 2 2013 at 21:05, Thierry Reding wrote:
> On Mon, Dec 02, 2013 at 08:57:20PM +0100, Tomasz Figa wrote:
> > In general, this looks good to me as a starter, so we could have support
> > for DSI bus merged. IMHO we should consider adding some generic bus
> > properties in future, though.
> 
> To be honest this looked somewhat minimal to me too at first, and then I
> tried to come up with any other properties but couldn't think of any. Of
> course anything that we add later on has the potential to break ABI
> compatibility.
> 
> A few of the things I had in mind that might be added as properties were
> the number of lanes or the video format. But those will already be
> implied by the compatible value and therefore don't really belong in the
> DT.
> 
> But if anyone can think of other properties that would be useful for DSI
> host or peripherals, by all means, let me know.

I've been working through the DSI bus patch set in conjunction with this one. There are two properties that are properties of the board rather than the host or the connected panels, so could fit in the DT. The first is the number of lanes connected - the compatible string can only provide the maximum number of lanes. Having said that, the panel will specify how many lanes it uses - if they're not all connected up you'll hopefully have noticed during the board layout...

The other property (that may actually be useful) is the maximum HS clock speed. The host and panel drivers will specify one, along with an implicit minimum from the panel driver for the data rate requirement. The board may also impose limits on the clock speed. For burst mode video or command mode some flexibility in bus clock speed is helpful, with the actual used clock rate frequently depending on RF concerns.

Bert.


More information about the dri-devel mailing list