Hello Luca,

On Tue, May 19, 2026 at 01:08:43PM +0200, Luca Ceresoli wrote:
> On Fri, 01 May 2026 00:20:32 +0200, Sebastian Reichel <sebastian.reichel@collabora.com> wrote:
> > diff --git a/drivers/gpu/drm/bridge/synopsys/dw-dp.c b/drivers/gpu/drm/bridge/synopsys/dw-dp.c
> > index fc4ddb7792d7..ccc55e40e81c 100644
> > --- a/drivers/gpu/drm/bridge/synopsys/dw-dp.c
> > +++ b/drivers/gpu/drm/bridge/synopsys/dw-dp.c
> > @@ -2064,6 +2064,20 @@ struct dw_dp *dw_dp_bind(struct device *dev, struct drm_encoder *encoder,
> >  		goto unregister_aux;
> >  	}
> >  
> > +	next_bridge = devm_drm_of_get_bridge(dev, dev->of_node, 1, 0);
> 
> devm_drm_of_get_bridge() looks for either a panel or a bridge. Is it
> possible to have a panel connected here?
> 
> If it isn't, then using of_drm_get_bridge_by_endpoint() which has just been
> merged [0] would be much simpler and cleaner.

Technically this should always be connected to the USBDP PHY
on the Rockchip platforms using this IP (RK3588 & RK3576) and
there are no other upstream platforms using this code so far.

Having said that, we do have some upstream DTs, which directly
connect it to a DP connector instead (as the PHY does not yet
expose a DRM bridge, that's being handled in a different series).
But there is no upstream board connecting it to an eDP panel,
so I will switch over to of_drm_get_bridge_by_endpoint().

Thanks for the info.

Greetings,

-- Sebastian

> 
> [0] https://gitlab.freedesktop.org/drm/misc/kernel/-/commit/03d1078112fddd706b2c1e4a7d98cf18700eb5df
> 
> Luca
> 
> -- 
> Luca Ceresoli, Bootlin
> Embedded Linux and Kernel engineering
> https://bootlin.com
>