On 2/5/2026 9:22 PM, Krzysztof Kozlowski wrote:

On Thu, Jan 29, 2026 at 12:05:32PM +0800, Joey Lu wrote:

quoted

Add DRM driver support for the Display Control Unit (DCU)
found in Nuvoton MA35D1 SoCs.

Signed-off-by: Joey Lu <redacted>
---
  drivers/gpu/drm/Kconfig                  |   1 +
  drivers/gpu/drm/Makefile                 |   1 +
  drivers/gpu/drm/nuvoton/Kconfig          |  21 +
  drivers/gpu/drm/nuvoton/Makefile         |   7 +
  drivers/gpu/drm/nuvoton/ma35_crtc.c      | 372 ++++++++++++++
  drivers/gpu/drm/nuvoton/ma35_crtc.h      |  67 +++
  drivers/gpu/drm/nuvoton/ma35_drm.c       | 371 ++++++++++++++
  drivers/gpu/drm/nuvoton/ma35_drm.h       |  48 ++
  drivers/gpu/drm/nuvoton/ma35_interface.c | 193 ++++++++
  drivers/gpu/drm/nuvoton/ma35_interface.h |  30 ++
  drivers/gpu/drm/nuvoton/ma35_plane.c     | 603 +++++++++++++++++++++++
  drivers/gpu/drm/nuvoton/ma35_plane.h     | 115 +++++
  drivers/gpu/drm/nuvoton/ma35_regs.h      |  88 ++++

No maintainers? Why would we want to take unmaintained code?

I'll add an entry in MAINTAINERS file.

quoted

+static void ma35_mode_fini(struct ma35_drm *priv)
+{
+	struct drm_device *drm_dev = &priv->drm_dev;
+
+	drm_kms_helper_poll_fini(drm_dev);
+}
+
+static int ma35_clocks_prepare(struct ma35_drm *priv)
+{
+	struct drm_device *drm_dev = &priv->drm_dev;
+	struct device *dev = drm_dev->dev;
+	int ret;
+
+	priv->dcuclk = devm_clk_get(dev, "dcu_gate");
+	if (IS_ERR(priv->dcuclk)) {
+		dev_err(dev, "Failed to get display core clock\n");

Don't spam logs on defers. Syntax is in entire probe path: return
dev_err_probe

quoted

+		return PTR_ERR(priv->dcuclk);
+	}
+
+	ret = clk_prepare_enable(priv->dcuclk);

Why this cannot be devm_clk_get_enabled?

I'll fix it.

quoted

+	if (ret) {
+		dev_err(dev, "Failed to enable display core clock\n");
+		return ret;
+	}
+
+	priv->dcupclk = devm_clk_get(dev, "dcup_div");
+	if (IS_ERR(priv->dcupclk)) {
+		dev_err(dev, "Failed to get display pixel clock\n");
+		return PTR_ERR(priv->dcupclk);
+	}
+
+	ret = clk_prepare_enable(priv->dcupclk);
+	if (ret) {
+		dev_err(dev, "Failed to enable display pixel clock\n");
+		return ret;
+	}
+
+	return 0;
+}
+
+static int ma35_clocks_unprepare(struct ma35_drm *priv)
+{
+	struct clk **clocks[] = {
+		&priv->dcuclk,
+		&priv->dcupclk,
+	};
+	unsigned int i;
+
+	for (i = 0; i < ARRAY_SIZE(clocks); i++) {
+		if (!*clocks[i])
+			continue;
+
+		clk_disable_unprepare(*clocks[i]);
+		*clocks[i] = NULL;

Huh, pretty complicated and pointless code. This should be devm and bulk
API...

I'll use memory safe helpers instead.

quoted

+	}
+
+	return 0;
+}
+
+static int ma35_drm_probe(struct platform_device *pdev)
+{
+	struct device *dev = &pdev->dev;
+	struct ma35_drm *priv;
+	struct drm_device *drm_dev;
+	void __iomem *base;
+	struct regmap *regmap = NULL;
+	int irq;
+	int ret;
+
+	ret = of_reserved_mem_device_init(dev);
+	if (ret && ret != -ENODEV) {
+		dev_err(dev, "Failed to get optional reserved memory: %d\n", ret);
+		return ret;
+	}
+
+	base = devm_platform_ioremap_resource(pdev, 0);
+	if (IS_ERR(base)) {
+		dev_err(dev, "Failed to map I/O base\n");

Why aren't you using dev_err_probe?

quoted

+		ret = PTR_ERR(base);
+		goto error_reserved_mem;
+	}
+	regmap = devm_regmap_init_mmio(dev, base, &ma35_drm_regmap_config);
+	if (IS_ERR(regmap)) {
+		dev_err(dev, "Failed to create regmap for I/O\n");
+		ret = PTR_ERR(regmap);
+		goto error_reserved_mem;
+	}
+
+	irq = platform_get_irq(pdev, 0);
+	if (irq < 0) {
+		ret = -ENODEV;
+		goto error_reserved_mem;
+	}
+
+	priv = devm_drm_dev_alloc(dev, &ma35_drm_driver,
+				     struct ma35_drm, drm_dev);
+	if (IS_ERR(priv)) {
+		ret = PTR_ERR(priv);
+		goto error_reserved_mem;
+	}
+
+	platform_set_drvdata(pdev, priv);
+	drm_dev = &priv->drm_dev;
+	priv->regmap = regmap;
+	INIT_LIST_HEAD(&priv->layers_list);
+
+	ret = ma35_clocks_prepare(priv);
+	if (ret) {
+		drm_err(drm_dev, "Failed to prepare clocks\n");

Why do you print error twice? Once in the function, second time here?

quoted

+		goto error_reserved_mem;
+	}
+
+	ret = devm_request_irq(dev, irq, ma35_drm_irq_handler, 0,
+			       dev_name(dev), priv);
+	if (ret) {
+		drm_err(drm_dev, "Failed to request IRQ\n");
+		goto error_clocks;
+	}
+
+	/* modeset */
+	ret = ma35_mode_init(priv);
+	if (ret) {
+		drm_err(drm_dev, "Failed to initialize KMS\n");
+		goto error_clocks;
+	}
+
+	/* plane */
+	ret = ma35_plane_init(priv);
+	if (ret) {
+		drm_err(drm_dev, "Failed to initialize layers\n");
+		goto error_clocks;
+	}
+
+	/* crtc */
+	ret = ma35_crtc_init(priv);
+	if (ret) {
+		drm_err(drm_dev, "Failed to initialize CRTC\n");
+		goto error_clocks;
+	}
+
+	/* interface */
+	ret = ma35_interface_init(priv);
+	if (ret) {
+		if (ret != -EPROBE_DEFER)
+			drm_err(drm_dev, "Failed to initialize interface\n");
+
+		goto error_clocks;
+	}
+
+	drm_mode_config_reset(drm_dev);
+
+	ret = drm_dev_register(drm_dev, 0);
+	if (ret) {
+		drm_err(drm_dev, "Failed to register DRM device\n");
+		goto error_mode;
+	}
+
+	drm_client_setup(drm_dev, NULL);

Best regards,
Krzysztof

I'll  return raw error codes and let probe wrap them with dev_err_probe().

Thanks for the review.

Best regards,

Joey