Re: [PATCH] drm: bridge: ldb: add support for using channel 1 only

[Luca Ceresoli <luca.ceresoli@bootlin.com>]

Hi Marek,

thanks for the quick and detailed review!

On Wed, 5 Apr 2023 05:28:16 +0200
Marek Vasut <marex@denx.de> wrote:

> On 4/4/23 09:37, Luca Ceresoli wrote:
> 
> [...]
> 
> > @@ -177,28 +183,25 @@ static void fsl_ldb_atomic_enable(struct drm_bridge *bridge,
> >   	clk_prepare_enable(fsl_ldb->clk);
> >   
> >   	/* Program LDB_CTRL */
> > -	reg = LDB_CTRL_CH0_ENABLE;
> > -
> > -	if (fsl_ldb->lvds_dual_link)
> > -		reg |= LDB_CTRL_CH1_ENABLE | LDB_CTRL_SPLIT_MODE;
> > -
> > -	if (lvds_format_24bpp) {
> > -		reg |= LDB_CTRL_CH0_DATA_WIDTH;
> > -		if (fsl_ldb->lvds_dual_link)
> > -			reg |= LDB_CTRL_CH1_DATA_WIDTH;
> > -	}
> > -
> > -	if (lvds_format_jeida) {
> > -		reg |= LDB_CTRL_CH0_BIT_MAPPING;
> > -		if (fsl_ldb->lvds_dual_link)
> > -			reg |= LDB_CTRL_CH1_BIT_MAPPING;
> > -	}
> > -
> > -	if (mode->flags & DRM_MODE_FLAG_PVSYNC) {
> > -		reg |= LDB_CTRL_DI0_VSYNC_POLARITY;
> > -		if (fsl_ldb->lvds_dual_link)
> > -			reg |= LDB_CTRL_DI1_VSYNC_POLARITY;
> > -	}
> > +	reg =  
> 
> Cosmetic nit, do we need the newline here , can't we just move the first 
> '(fsl_ldb->ch0_enabled ? LDB_CTRL_CH0_ENABLE : 0) |' on the same line as 
> 'reg =' ? It might need a bit of indent with spaces, but that should be OK.

Sure. Maybe 'reg =<tab>(fsl...' would be even better, it checkpatch
allows.

> > @@ -311,10 +314,23 @@ static int fsl_ldb_probe(struct platform_device *pdev)
> >   	if (IS_ERR(fsl_ldb->regmap))
> >   		return PTR_ERR(fsl_ldb->regmap);
> >   
> > -	/* Locate the panel DT node. */
> > -	panel_node = of_graph_get_remote_node(dev->of_node, 1, 0);
> > -	if (!panel_node)
> > -		return -ENXIO;
> > +	/* Locate the remote ports and the panel node */
> > +	remote1 = of_graph_get_remote_node(dev->of_node, 1, 0);
> > +	remote2 = of_graph_get_remote_node(dev->of_node, 2, 0);
> > +	fsl_ldb->ch0_enabled = (remote1 != NULL);
> > +	fsl_ldb->ch1_enabled = (remote2 != NULL);
> > +	panel_node = of_node_get(remote1 ? remote1 : remote2);  
> 
> You can even do this without the middle 'remote1' I think:
> 
> panel_node = of_node_get(remote1 ? : remote2);

Apparently, but honestly with such short expressions clearly having no
side effects I think it's not helping readability.

> > +	of_node_put(remote1);
> > +	of_node_put(remote2);
> > +
> > +	if (!fsl_ldb->ch0_enabled && !fsl_ldb->ch1_enabled) {
> > +		of_node_put(panel_node);
> > +		return dev_err_probe(dev, -ENXIO, "No panel node found");
> > +	}
> > +
> > +	dev_dbg(dev, "Using %s\n",
> > +		fsl_ldb_is_dual(fsl_ldb) ? "dual mode" :  
> 
> I think this is called "dual-link mode" , maybe update the string .

I was using the terms from the NXP docs, but indeed in the kernel it
seems that "dual-link" is the common name. Updating that.

> > @@ -325,20 +341,26 @@ static int fsl_ldb_probe(struct platform_device *pdev)
> >   	if (IS_ERR(fsl_ldb->panel_bridge))
> >   		return PTR_ERR(fsl_ldb->panel_bridge);
> >   
> > -	/* Determine whether this is dual-link configuration */
> > -	port1 = of_graph_get_port_by_id(dev->of_node, 1);
> > -	port2 = of_graph_get_port_by_id(dev->of_node, 2);
> > -	dual_link = drm_of_lvds_get_dual_link_pixel_order(port1, port2);
> > -	of_node_put(port1);
> > -	of_node_put(port2);
> >   
> > -	if (dual_link == DRM_LVDS_DUAL_LINK_EVEN_ODD_PIXELS) {
> > -		dev_err(dev, "LVDS channel pixel swap not supported.\n");
> > -		return -EINVAL;
> > -	}
> > +	if (fsl_ldb_is_dual(fsl_ldb)) {
> > +		struct device_node *port1, *port2;
> > +
> > +		port1 = of_graph_get_port_by_id(dev->of_node, 1);
> > +		port2 = of_graph_get_port_by_id(dev->of_node, 2);
> > +		dual_link = drm_of_lvds_get_dual_link_pixel_order(port1, port2);
> > +		of_node_put(port1);
> > +		of_node_put(port2);
> >   
> > -	if (dual_link == DRM_LVDS_DUAL_LINK_ODD_EVEN_PIXELS)
> > -		fsl_ldb->lvds_dual_link = true;
> > +		if (dual_link < 0)
> > +			return dev_err_probe(dev, dual_link,
> > +					     "Error getting dual link configuration");  
> 
> Does this need a trailing '\n' in the formatting string or not ? I think 
> yes.

Oops, good catch.

> The rest looks good, with the few details fixed:
> 
> Reviewed-by: Marek Vasut <marex@denx.de>

Thanks!

-- 
Luca Ceresoli, Bootlin
Embedded Linux and Kernel engineering
https://bootlin.com