[PATCH v12] drm/bridge: imx: Add i.MX93 parallel display format configuration support

[PATCH v12] drm/bridge: imx: Add i.MX93 parallel display format configuration support

[Liu Ying]

NXP i.MX93 mediamix blk-ctrl contains one DISPLAY_MUX register which
configures parallel display format by using the "PARALLEL_DISP_FORMAT"
field. Add a DRM bridge driver to support the display format configuration.

[m.felsch@pengutronix.de: add review feedback (Alexander)]
[m.felsch@pengutronix.de: fix to short Kconfig description (checkpath)]
[m.felsch@pengutronix.de: use "GPL" instead of "GPL v2" (checkpatch)]
[m.felsch@pengutronix.de: add bus-width support]
[ Liu Ying: Add nxp,imx91-pdfc compatible string ]
[ Liu Ying: Call drm_bridge_get() for next_bridge ]

Reviewed-by: Luca Ceresoli <luca.ceresoli@bootlin.com>
[m.felsch@pengutronix.de: port to v7.0-rc1]
Signed-off-by: Marco Felsch <m.felsch@pengutronix.de>
[ Liu Ying: Port to drm-misc/drm-misc-next(v7.2-rc2 based) ]
Signed-off-by: Liu Ying <victor.liu@nxp.com>
---
Hi,

I'm taking Marco Felsch's v11 work over and trying to add the imx93-pdfc.c
driver.

DT binding patch and DT patch in v11 have already landed, so only need
to send the driver patch out for review.
---
Changes in v12:
- Link to v11: https://lore.kernel.org/r/20260303-v6-18-topic-imx93-parallel-display-v11-0-1b03733c8461@pengutronix.de
- Rebase onto latest drm-misc/drm-misc-next
- Drop DT binding patch and DT patch because they have landed
- Add nxp,imx91-pdfc compatible string
- Call drm_bridge_get() for next_bridge
- Cc Marco

Changes in v11:
- Link to v10: https://lore.kernel.org/r/20260302-v6-18-topic-imx93-parallel-display-v10-0-634fe2778c7a@pengutronix.de
- Fix double space (Liu)
- Drop 'const' (Liu)
- Drop GFP_KERNEL from kmalloc_obj() (Liu)
- Add r-b (Luca)

Changes in v10:
- Link to v9: https://lore.kernel.org/r/20260115-v6-18-topic-imx93-parallel-display-v9-0-2c5051e4b144@pengutronix.de
- Add MEDIA_BUS_FMT_FIXED (Liu)
- Drop next_bridge from driver struct and use bridge.next_bridge (Liu)
- Drop linux/of_address.h include (Liu)
- imx93-pdfc: drop bridge.driver_private usage++ (Liu)
- Make use of kmalloc_obj() (Liu)

Changes in v9:
- Link to v8: https://lore.kernel.org/r/20260113-v6-18-topic-imx93-parallel-display-v8-0-4abccdc473a5@pengutronix.de
- dt-bindings: drop unncessary changes (Frank)
- imx93-pdfc: drop bridge.driver_private usage (Luca)
- Kconfig: Adapt Kconfig symbol and prompt (Luca)

Changes in v8:
- Link to v7: https://lore.kernel.org/r/20251202-v6-18-topic-imx93-parallel-display-v7-0-2cce31d64608@pengutronix.de
- dt-bindings: add nxp,imx91-pdfc compatible (Liu)
- dt-bindings: use video-interfaces.yaml# (Liu)
- dt-bindings: s/data lanes/data lines/ (Liu)
- dt-bindings: drop 'reg' poperty
- dt-bindings: drop #address-cells, #size-cells
- imx93-pdfc: drop drm/drm_print.h include (Liu)
- imx93-pdfc: s/exist/exists/ (Liu)
- imx93-pdfc: drop MEDIA_BUS_FMT_FIXED from imx93_pdfc_bus_output_fmts
- imx93-pdfc: imx93_pdfc_bus_output_fmt_supported: make fmt const
- imx93-pdfc: Rework input-fmt selection to always fallback to a sane
              default.
- imx93-pdfc: imx93_pdfc_bridge_atomic_check: make use of
	      imx93_pdfc_bus_output_fmt_supported()
- imx93-pdfc: drop 'reg' dt-property usage
- imx93-pdfc: imx93_pdfc_bridge_probe: pass -1 for endpoint reg value (Liu)

Changes in v7:
- Link to v6: https://lore.kernel.org/r/20251201-v6-18-topic-imx93-parallel-display-v6-0-7b056e1e5b1e@pengutronix.de
- Add missing bits.h and bitfield.h headers (lkp)

Changes in v6:
- Link to v5: https://lore.kernel.org/all/20250304082434.834031-1-victor.liu@nxp.com/
- Add bus-width support
- rebase onto v6.18-rc1
- add review feedback (Alexander)
- driver license "GPL v2" -> "GPL" (checkpatch)
- make use of reg of-property
- fix to short Kconfig description (checkpath)
- add OF integration

To: Liu Ying <victor.liu@nxp.com>
To: Andrzej Hajda <andrzej.hajda@intel.com>
To: Neil Armstrong <neil.armstrong@linaro.org>
To: Robert Foss <rfoss@kernel.org>
To: Laurent Pinchart <Laurent.pinchart@ideasonboard.com>
To: Jonas Karlman <jonas@kwiboo.se>
To: Jernej Skrabec <jernej.skrabec@gmail.com>
To: Luca Ceresoli <luca.ceresoli@bootlin.com>
To: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
To: Maxime Ripard <mripard@kernel.org>
To: Thomas Zimmermann <tzimmermann@suse.de>
To: David Airlie <airlied@gmail.com>
To: Simona Vetter <simona@ffwll.ch>
To: Frank Li <Frank.Li@nxp.com>
To: Sascha Hauer <s.hauer@pengutronix.de>
To: Pengutronix Kernel Team <kernel@pengutronix.de>
To: Fabio Estevam <festevam@gmail.com>
Cc: linux-kernel@vger.kernel.org
Cc: dri-devel@lists.freedesktop.org
Cc: imx@lists.linux.dev
Cc: linux-arm-kernel@lists.infradead.org
Cc: Marco Felsch <m.felsch@pengutronix.de>
---
 drivers/gpu/drm/bridge/imx/Kconfig      |  11 ++
 drivers/gpu/drm/bridge/imx/Makefile     |   1 +
 drivers/gpu/drm/bridge/imx/imx93-pdfc.c | 226 ++++++++++++++++++++++++++++++++
 3 files changed, 238 insertions(+)

diff --git a/drivers/gpu/drm/bridge/imx/Kconfig b/drivers/gpu/drm/bridge/imx/Kconfig
index 005a745e5c47..49fcecfee38e 100644
--- a/drivers/gpu/drm/bridge/imx/Kconfig
+++ b/drivers/gpu/drm/bridge/imx/Kconfig
@@ -107,4 +107,15 @@ config DRM_IMX93_MIPI_DSI
 	  Choose this to enable MIPI DSI controller found in Freescale i.MX93
 	  processor.
 
+config DRM_IMX93_PARALLEL_DISP_FMT_CONVERTER
+	tristate "NXP i.MX91/i.MX93 parallel display format converter"
+	depends on OF
+	select DRM_KMS_HELPER
+	help
+	  On i.MX93 and i.MX91 SoCs the parallel display format output is
+	  controlled via the MEDIAMIX BLK-CTRL DISPLAY_MUX.
+
+	  Say 'Y' or 'M' if you use the parallel display output path on a
+	  i.MX93 or i.MX91 SoC.
+
 endif # ARCH_MXC || COMPILE_TEST
diff --git a/drivers/gpu/drm/bridge/imx/Makefile b/drivers/gpu/drm/bridge/imx/Makefile
index 94ac8c40ebe9..99f28cd013ae 100644
--- a/drivers/gpu/drm/bridge/imx/Makefile
+++ b/drivers/gpu/drm/bridge/imx/Makefile
@@ -10,3 +10,4 @@ obj-$(CONFIG_DRM_IMX8QXP_PIXEL_COMBINER) += imx8qxp-pixel-combiner.o
 obj-$(CONFIG_DRM_IMX8QXP_PIXEL_LINK) += imx8qxp-pixel-link.o
 obj-$(CONFIG_DRM_IMX8QXP_PIXEL_LINK_TO_DPI) += imx8qxp-pxl2dpi.o
 obj-$(CONFIG_DRM_IMX93_MIPI_DSI) += imx93-mipi-dsi.o
+obj-$(CONFIG_DRM_IMX93_PARALLEL_DISP_FMT_CONVERTER) += imx93-pdfc.o
diff --git a/drivers/gpu/drm/bridge/imx/imx93-pdfc.c b/drivers/gpu/drm/bridge/imx/imx93-pdfc.c
new file mode 100644
index 000000000000..b084409d22ef
--- /dev/null
+++ b/drivers/gpu/drm/bridge/imx/imx93-pdfc.c
@@ -0,0 +1,226 @@
+// SPDX-License-Identifier: GPL-2.0+
+
+/*
+ * Copyright 2022-2025 NXP
+ */
+
+#include <linux/bitfield.h>
+#include <linux/bits.h>
+#include <linux/media-bus-format.h>
+#include <linux/mfd/syscon.h>
+#include <linux/module.h>
+#include <linux/of.h>
+#include <linux/of_graph.h>
+#include <linux/platform_device.h>
+#include <linux/regmap.h>
+
+#include <drm/drm_atomic_state_helper.h>
+#include <drm/drm_bridge.h>
+
+#define IMX93_DISPLAY_MUX_REG		0x60
+#define PARALLEL_DISP_FORMAT		GENMASK(10, 8)
+#define FORMAT_RGB888_TO_RGB888		FIELD_PREP(PARALLEL_DISP_FORMAT, 0)
+#define FORMAT_RGB888_TO_RGB666		FIELD_PREP(PARALLEL_DISP_FORMAT, 1)
+#define FORMAT_RGB565_TO_RGB565		FIELD_PREP(PARALLEL_DISP_FORMAT, 2)
+
+struct imx93_pdfc {
+	struct drm_bridge bridge;
+	struct device *dev;
+	struct regmap *regmap;
+	u32 phy_bus_width;
+};
+
+static struct imx93_pdfc *bridge_to_imx93_pdfc(struct drm_bridge *bridge)
+{
+	return container_of(bridge, struct imx93_pdfc, bridge);
+}
+
+static int
+imx93_pdfc_bridge_attach(struct drm_bridge *bridge, struct drm_encoder *encoder,
+			 enum drm_bridge_attach_flags flags)
+{
+	return drm_bridge_attach(bridge->encoder, bridge->next_bridge, bridge, flags);
+}
+
+static void imx93_pdfc_bridge_atomic_enable(struct drm_bridge *bridge,
+					    struct drm_atomic_commit *state)
+{
+	struct imx93_pdfc *pdfc = bridge_to_imx93_pdfc(bridge);
+	const struct drm_bridge_state *bridge_state;
+	unsigned int mask = PARALLEL_DISP_FORMAT;
+	unsigned int val;
+
+	bridge_state = drm_atomic_get_new_bridge_state(state, bridge);
+
+	switch (bridge_state->output_bus_cfg.format) {
+	case MEDIA_BUS_FMT_RGB888_1X24:
+	case MEDIA_BUS_FMT_FIXED:
+		val = FORMAT_RGB888_TO_RGB888;
+		if (pdfc->phy_bus_width == 18) {
+			/*
+			 * Can be valid if physical bus limitation exists,
+			 * therefore use dev_dbg().
+			 */
+			dev_dbg(pdfc->dev, "Truncate two LSBs from each color\n");
+			val = FORMAT_RGB888_TO_RGB666;
+		}
+		break;
+	case MEDIA_BUS_FMT_RGB666_1X18:
+		val = FORMAT_RGB888_TO_RGB666;
+		break;
+	case MEDIA_BUS_FMT_RGB565_1X16:
+		val = FORMAT_RGB565_TO_RGB565;
+		break;
+	}
+
+	regmap_update_bits(pdfc->regmap, IMX93_DISPLAY_MUX_REG, mask, val);
+}
+
+/* TODO: Add YUV formats */
+static const u32 imx93_pdfc_bus_output_fmts[] = {
+	MEDIA_BUS_FMT_FIXED,
+	MEDIA_BUS_FMT_RGB888_1X24,
+	MEDIA_BUS_FMT_RGB666_1X18,
+	MEDIA_BUS_FMT_RGB565_1X16,
+};
+
+static bool imx93_pdfc_bus_output_fmt_supported(u32 fmt)
+{
+	int i;
+
+	for (i = 0; i < ARRAY_SIZE(imx93_pdfc_bus_output_fmts); i++) {
+		if (imx93_pdfc_bus_output_fmts[i] == fmt)
+			return true;
+	}
+
+	return false;
+}
+
+static u32 *
+imx93_pdfc_bridge_atomic_get_input_bus_fmts(struct drm_bridge *bridge,
+					    struct drm_bridge_state *bridge_state,
+					    struct drm_crtc_state *crtc_state,
+					    struct drm_connector_state *conn_state,
+					    u32 output_fmt,
+					    unsigned int *num_input_fmts)
+{
+	struct imx93_pdfc *pdfc = bridge_to_imx93_pdfc(bridge);
+	u32 *input_fmts;
+
+	*num_input_fmts = 0;
+
+	input_fmts = kmalloc_obj(*input_fmts);
+	if (!input_fmts)
+		return NULL;
+
+	*num_input_fmts = 1;
+
+	if (!imx93_pdfc_bus_output_fmt_supported(output_fmt)) {
+		dev_dbg(pdfc->dev, "No valid output bus-fmt detected, fallback to MEDIA_BUS_FMT_RGB888_1X24\n");
+		input_fmts[0] = MEDIA_BUS_FMT_RGB888_1X24;
+		return input_fmts;
+	}
+
+	switch (output_fmt) {
+	case MEDIA_BUS_FMT_RGB888_1X24:
+	case MEDIA_BUS_FMT_RGB565_1X16:
+		input_fmts[0] = output_fmt;
+		break;
+	case MEDIA_BUS_FMT_RGB666_1X18:
+	case MEDIA_BUS_FMT_FIXED:
+		input_fmts[0] = MEDIA_BUS_FMT_RGB888_1X24;
+		break;
+	}
+
+	return input_fmts;
+}
+
+static int imx93_pdfc_bridge_atomic_check(struct drm_bridge *bridge,
+					  struct drm_bridge_state *bridge_state,
+					  struct drm_crtc_state *crtc_state,
+					  struct drm_connector_state *conn_state)
+{
+	struct imx93_pdfc *pdfc = bridge_to_imx93_pdfc(bridge);
+	u32 format = bridge_state->output_bus_cfg.format;
+
+	if (imx93_pdfc_bus_output_fmt_supported(format))
+		return 0;
+
+	dev_warn(pdfc->dev, "Unsupported output bus format: 0x%x\n", format);
+
+	return -EINVAL;
+}
+
+static const struct drm_bridge_funcs funcs = {
+	.attach			= imx93_pdfc_bridge_attach,
+	.atomic_enable		= imx93_pdfc_bridge_atomic_enable,
+	.atomic_duplicate_state	= drm_atomic_helper_bridge_duplicate_state,
+	.atomic_destroy_state	= drm_atomic_helper_bridge_destroy_state,
+	.atomic_get_input_bus_fmts	= imx93_pdfc_bridge_atomic_get_input_bus_fmts,
+	.atomic_check		= imx93_pdfc_bridge_atomic_check,
+	.atomic_create_state	= drm_atomic_helper_bridge_create_state,
+};
+
+static int imx93_pdfc_bridge_probe(struct platform_device *pdev)
+{
+	struct device *dev = &pdev->dev;
+	struct drm_bridge *next_bridge;
+	struct imx93_pdfc *pdfc;
+	struct device_node *ep;
+	int err;
+
+	pdfc = devm_drm_bridge_alloc(dev, struct imx93_pdfc, bridge, &funcs);
+	if (IS_ERR(pdfc))
+		return PTR_ERR(pdfc);
+
+	pdfc->regmap = syscon_node_to_regmap(dev->of_node->parent);
+	if (IS_ERR(pdfc->regmap))
+		return dev_err_probe(dev, PTR_ERR(pdfc->regmap),
+				     "failed to get regmap\n");
+
+	/* No limits per default */
+	pdfc->phy_bus_width = 24;
+
+	/* Get output ep (port1/endpoint) */
+	ep = of_graph_get_endpoint_by_regs(dev->of_node, 1, -1);
+	if (ep) {
+		err = of_property_read_u32(ep, "bus-width", &pdfc->phy_bus_width);
+		of_node_put(ep);
+
+		/* bus-width is optional but it must have valid data if present */
+		if (err && err != -EINVAL)
+			return dev_err_probe(dev, err,
+					     "failed to query bus-width\n");
+	}
+
+	next_bridge = devm_drm_of_get_bridge(dev, dev->of_node, 1, 0);
+	if (IS_ERR(next_bridge))
+		return dev_err_probe(dev, PTR_ERR(next_bridge),
+				     "failed to get next bridge\n");
+	pdfc->dev = dev;
+	pdfc->bridge.of_node = dev->of_node;
+	pdfc->bridge.type = DRM_MODE_CONNECTOR_DPI;
+	pdfc->bridge.next_bridge = drm_bridge_get(next_bridge);
+
+	return devm_drm_bridge_add(dev, &pdfc->bridge);
+}
+
+static const struct of_device_id imx93_pdfc_dt_ids[] = {
+	{ .compatible = "nxp,imx91-pdfc", },
+	{ .compatible = "nxp,imx93-pdfc", },
+	{ /* sentinel */ }
+};
+MODULE_DEVICE_TABLE(of, imx93_pdfc_dt_ids);
+
+static struct platform_driver imx93_pdfc_bridge_driver = {
+	.probe	= imx93_pdfc_bridge_probe,
+	.driver	= {
+		.of_match_table = imx93_pdfc_dt_ids,
+		.name = "imx93_pdfc",
+	},
+};
+module_platform_driver(imx93_pdfc_bridge_driver);
+
+MODULE_DESCRIPTION("NXP i.MX93 parallel display format configuration driver");
+MODULE_AUTHOR("Liu Ying <victor.liu@nxp.com>");
+MODULE_LICENSE("GPL");

---
base-commit: 9499d3bc41a2c4e6a1badfb55f943b9d0c34f034
change-id: 20251201-v6-18-topic-imx93-parallel-display-95f9234bf6cc

Best regards,
--  
Regards,
Liu Ying

Re: [PATCH v12] drm/bridge: imx: Add i.MX93 parallel display format configuration support

[sashiko-bot]

Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
- [High] Returning a fallback format for unsupported output formats breaks DRM bridge format negotiation.
--

--- Patch [1]: [PATCH v12] drm/bridge: imx: Add i.MX93 parallel display format configuration support ---
commit 62def9c6e85686d63c6204b71b54383955f8f9d1
Author: Liu Ying <victor.liu@nxp.com>
drm/bridge: imx: Add i.MX93 parallel display format configuration support

This commit introduces a new DRM bridge driver to support parallel display
format configuration on the NXP i.MX93 and i.MX91 SoCs. It handles format
conversions for the MEDIAMIX BLK-CTRL DISPLAY_MUX register.

> diff --git a/drivers/gpu/drm/bridge/imx/imx93-pdfc.c b/drivers/gpu/drm/bridge/imx/imx93-pdfc.c
> new file mode 100644
> index 0000000000000..b084409d22ef2
> --- /dev/null
> +++ b/drivers/gpu/drm/bridge/imx/imx93-pdfc.c
[ ... ]
> +static u32 *
> +imx93_pdfc_bridge_atomic_get_input_bus_fmts(struct drm_bridge *bridge,
> +					    struct drm_bridge_state *bridge_state,
> +					    struct drm_crtc_state *crtc_state,
> +					    struct drm_connector_state *conn_state,
> +					    u32 output_fmt,
> +					    unsigned int *num_input_fmts)
> +{
> +	struct imx93_pdfc *pdfc = bridge_to_imx93_pdfc(bridge);
> +	u32 *input_fmts;
> +
> +	*num_input_fmts = 0;
> +
> +	input_fmts = kmalloc_obj(*input_fmts);
> +	if (!input_fmts)
> +		return NULL;
> +
> +	*num_input_fmts = 1;
> +
> +	if (!imx93_pdfc_bus_output_fmt_supported(output_fmt)) {
> +		dev_dbg(pdfc->dev, "No valid output bus-fmt detected, fallback to MEDIA_BUS_FMT_RGB888_1X24\n");
> +		input_fmts[0] = MEDIA_BUS_FMT_RGB888_1X24;
> +		return input_fmts;
> +	}

[Severity: High]
Does returning a fallback format here break DRM bridge format negotiation?

When the downstream bridge requests an unsupported output format first,
drm_atomic_bridge_chain_select_bus_fmts() queries atomic_get_input_bus_fmts()
to check if the format is supported. By returning MEDIA_BUS_FMT_RGB888_1X24
instead of 0 formats, this driver falsely claims it can support the requested
output_fmt.

The core then proceeds with the unsupported format, which is correctly
rejected by imx93_pdfc_bridge_atomic_check(), aborting the modeset with
-EINVAL instead of gracefully negotiating a fallback.

Should this instead set *num_input_fmts = 0, free the array, and return NULL
when the output format is unsupported?

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20260722-v6-18-topic-imx93-parallel-display-v12-1-a59dbb4dbc13@nxp.com?part=1

Re: [PATCH v12] drm/bridge: imx: Add i.MX93 parallel display format configuration support

[Liu Ying]

On Wed, Jul 22, 2026 at 08:23:59AM +0000, sashiko-bot@kernel.org wrote:
> Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
> - [High] Returning a fallback format for unsupported output formats breaks DRM bridge format negotiation.
> --
> 
> --- Patch [1]: [PATCH v12] drm/bridge: imx: Add i.MX93 parallel display format configuration support ---
> commit 62def9c6e85686d63c6204b71b54383955f8f9d1
> Author: Liu Ying <victor.liu@nxp.com>
> drm/bridge: imx: Add i.MX93 parallel display format configuration support
> 
> This commit introduces a new DRM bridge driver to support parallel display
> format configuration on the NXP i.MX93 and i.MX91 SoCs. It handles format
> conversions for the MEDIAMIX BLK-CTRL DISPLAY_MUX register.
> 
> > diff --git a/drivers/gpu/drm/bridge/imx/imx93-pdfc.c b/drivers/gpu/drm/bridge/imx/imx93-pdfc.c
> > new file mode 100644
> > index 0000000000000..b084409d22ef2
> > --- /dev/null
> > +++ b/drivers/gpu/drm/bridge/imx/imx93-pdfc.c
> [ ... ]
> > +static u32 *
> > +imx93_pdfc_bridge_atomic_get_input_bus_fmts(struct drm_bridge *bridge,
> > +					    struct drm_bridge_state *bridge_state,
> > +					    struct drm_crtc_state *crtc_state,
> > +					    struct drm_connector_state *conn_state,
> > +					    u32 output_fmt,
> > +					    unsigned int *num_input_fmts)
> > +{
> > +	struct imx93_pdfc *pdfc = bridge_to_imx93_pdfc(bridge);
> > +	u32 *input_fmts;
> > +
> > +	*num_input_fmts = 0;
> > +
> > +	input_fmts = kmalloc_obj(*input_fmts);
> > +	if (!input_fmts)
> > +		return NULL;
> > +
> > +	*num_input_fmts = 1;
> > +
> > +	if (!imx93_pdfc_bus_output_fmt_supported(output_fmt)) {
> > +		dev_dbg(pdfc->dev, "No valid output bus-fmt detected, fallback to MEDIA_BUS_FMT_RGB888_1X24\n");
> > +		input_fmts[0] = MEDIA_BUS_FMT_RGB888_1X24;
> > +		return input_fmts;
> > +	}
> 
> [Severity: High]
> Does returning a fallback format here break DRM bridge format negotiation?
> 
> When the downstream bridge requests an unsupported output format first,
> drm_atomic_bridge_chain_select_bus_fmts() queries atomic_get_input_bus_fmts()
> to check if the format is supported. By returning MEDIA_BUS_FMT_RGB888_1X24
> instead of 0 formats, this driver falsely claims it can support the requested
> output_fmt.
> 
> The core then proceeds with the unsupported format, which is correctly
> rejected by imx93_pdfc_bridge_atomic_check(), aborting the modeset with
> -EINVAL instead of gracefully negotiating a fallback.

Makes sense.  The reported issue was introduced by v8.  My earlier v5 does
the negotiation correctly.

> 
> Should this instead set *num_input_fmts = 0, free the array, and return NULL
> when the output format is unsupported?

No need to free any array.  Just need to set *num_input_fmts = 0 and then
return NULL when the output format is unsupported.

> 
> -- 
> Sashiko AI review · https://sashiko.dev/#/patchset/20260722-v6-18-topic-imx93-parallel-display-v12-1-a59dbb4dbc13@nxp.com?part=1

-- 
Regards,
Liu Ying