On 12/20/2012 07:14 PM, Stephen Warren wrote:
>
> What's wrong with just having each device ask the host1x (its parent)
> for a pointer to the (dummy) tegradrm device. That seems extremely
>

We are talking about creating a dummy device because:
- we need to give something for drm_platform_init(),
- drm related data should be stored somewhere,
- some data must be passed to all driver probe() functions. This is not 
hw-related data, but just some lists that are used to ensure that all 
drivers have been initialised before calling drm_platform_init().

All these issues are purely tegra-drm related and solving them elsewhere 
(in host1x) would be just wrong! host1x would not even use the dummy 
device for anything so creating it there seems bizarre.

- Arto