drm/tegra: Parse device tree earlier

Parsing the device tree may cause probing to be deferred. Doing this as
early as possible prevents any other resources from being requested and
enabled, therefore reducing the need to cleanup on deferred probe while
at the same time not wasting precious CPU cycles determining if probing
needs to be deferred or not.
Signed-off-by: NThierry Reding <treding@nvidia.com>

drivers/gpu/host1x/drm/rgb.c

@@ -147,6 +147,13 @@ int tegra_dc_rgb_probe(struct tegra_dc *dc)
 	if (!rgb)
 		return -ENOMEM;
 
+	rgb->output.dev = dc->dev;
+	rgb->output.of_node = np;
+
+	err = tegra_output_parse_dt(&rgb->output);
+	if (err < 0)
+		return err;
+
 	rgb->clk = devm_clk_get(dc->dev, NULL);
 	if (IS_ERR(rgb->clk)) {
 		dev_err(dc->dev, "failed to get clock\n");
@@ -165,13 +172,6 @@ int tegra_dc_rgb_probe(struct tegra_dc *dc)
 		return err;
 	}
 
-	rgb->output.dev = dc->dev;
-	rgb->output.of_node = np;
-
-	err = tegra_output_parse_dt(&rgb->output);
-	if (err < 0)
-		return err;
-
 	dc->rgb = &rgb->output;
 
 	return 0;