Re: [PATCH 11/11] drm/bridge: tc358768: Add DRM_BRIDGE_ATTACH_NO_CONNECTOR support

[Tomi Valkeinen]

Hi Sam,

On 14/08/2023 13:10, Sam Ravnborg wrote:
> Hi Tomi,
>
> >  From c13c691bd8826b978325575be9a87f577b83b86b Mon Sep 17 00:00:00 2001
> > From: Tomi Valkeinen <tomi.valkeinen@xxxxxxxxxxxxxxxx>
> > Date: Mon, 14 Aug 2023 13:02:23 +0300
> > Subject: [PATCH] drm/bridge: tc358768: fix 'Add DRM_BRIDGE_ATTACH_NO_CONNECTOR
> >   support'
> >
> > Signed-off-by: Tomi Valkeinen <tomi.valkeinen@xxxxxxxxxxxxxxxx>
> > ---
> >   drivers/gpu/drm/bridge/tc358768.c | 56 +++++++++++++------------------
> >   1 file changed, 24 insertions(+), 32 deletions(-)
> >
> > diff --git a/drivers/gpu/drm/bridge/tc358768.c b/drivers/gpu/drm/bridge/tc358768.c
> > index 82ea4d9a814a..9705ce1bd028 100644
> > --- a/drivers/gpu/drm/bridge/tc358768.c
> > +++ b/drivers/gpu/drm/bridge/tc358768.c
> > @@ -455,8 +455,6 @@ static int tc358768_dsi_host_detach(struct mipi_dsi_host *host,
> >   	struct tc358768_priv *priv = dsi_host_to_tc358768(host);
> >   
> >   	drm_bridge_remove(&priv->bridge);
> > -	if (priv->output.panel)
> > -		drm_panel_bridge_remove(priv->output.next_bridge);
> >   
> >   	return 0;
> >   }
> > @@ -531,49 +529,42 @@ static int tc358768_bridge_attach(struct drm_bridge *bridge,
> >   				  enum drm_bridge_attach_flags flags)
> >   {
> >   	struct tc358768_priv *priv = bridge_to_tc358768(bridge);
> > +	struct drm_bridge *next_bridge;
> > +	struct drm_panel *panel;
> > +	int ret;
> >   
> >   	if (!drm_core_check_feature(bridge->dev, DRIVER_ATOMIC)) {
> >   		dev_err(priv->dev, "needs atomic updates support\n");
> >   		return -ENOTSUPP;
> >   	}
> >   
> > -	if (flags & DRM_BRIDGE_ATTACH_NO_CONNECTOR) {
> > -		struct device_node *node;
> > -
> > -		/* Get the next bridge, connected to port@1. */
> > -		node = of_graph_get_remote_node(priv->dev->of_node, 1, -1);
> > -		if (!node)
> > -			return -ENODEV;
> > -
> > -		priv->output.next_bridge = of_drm_find_bridge(node);
> > -		of_node_put(node);
> > -		if (!priv->output.next_bridge)
> > -			return -EPROBE_DEFER;
> > -	} else {
> > -		struct drm_bridge *bridge;
> > -		struct drm_panel *panel;
> > -		int ret;
> > -
> > -		ret = drm_of_find_panel_or_bridge(priv->dev->of_node, 1, 0,
> > -						  &panel, &bridge);
> > -		if (ret)
> > -			return ret;
> > -
> > -		if (panel) {
> > -			bridge = drm_panel_bridge_add_typed(panel,
> > -				DRM_MODE_CONNECTOR_DSI);
> > -			if (IS_ERR(bridge))
> > -				return PTR_ERR(bridge);
> > -		}
> > +	ret = drm_of_find_panel_or_bridge(priv->dev->of_node, 1, -1, &panel,
> > +					  &next_bridge);
>
> I think the right way is to wrap the panel in a bridge,
> so something like:
>
> 	next_bridge = devm_drm_of_get_bridge(dev, priv->dev->of_node, 1, -1)
>
> 	if (IS_ERR(next_bridge))
> 		return ...
> 	priv->output.next_bridge = next_bridge;

I tried that, but I had trouble with the cleanup side.

In the fixup patch I attached in my reply to Maxim I used 
drm_of_find_panel_or_bridge() + drm_panel_bridge_add_typed(), and on 
bridge_detach callback I used drm_panel_bridge_remove() (if there is a 
panel). This worked for me, but it does feel like a bit too much work 
for a driver to do.

 Tomi