Skip to contentSkip to conversation

arm64: dts: rockchip: Drop data-mapping from Engicam PX30 Core

PATCH Applied Fabio Estevam 3 participants 3 messages Jul 27 – Jul 29

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 arm64: dts: rockchip: Drop data-mapping from Engicam PX30 Core

The panel-simple schema only allows data-mapping overrides for panels whose mapping is not fixed by their compatible. Consequently, the property on the Ampire AM-1280800N3TZQW-T00H panel fails validation:

(ampire,am-1280800n3tzqw-t00h): False schema does not allow ['vesa-24']

The panel descriptor already selects 8 bits per color and MEDIA_BUS_FMT_RGB888_1X7X4_SPWG. The DRM OF helpers translate "vesa-24" to that exact bus format, and panel-simple falls back to the descriptor's format when the property is absent. Removing the property therefore has no functional effect.

Drop the redundant property to satisfy the binding.

---
 .../boot/dts/rockchip/px30-engicam-px30-core-ctouch2-of10.dts    | 1 -
 1 file changed, 1 deletion(-)
arch/arm64/boot/dts/rockchip/px30-engicam-px30-core-ctouch2-of10.dts +0 -1
index 47aa30505a42..aa636ce45e3a 100644
@@ -28,7 +28,6 @@ panel {
 		compatible = "ampire,am-1280800n3tzqw-t00h";
 		backlight = <&backlight>;
 		power-supply = <&vcc3v3_lcd>;
-		data-mapping = "vesa-24";
 
 		port {
 			panel_in_lvds: endpoint {
Signature · 1 line
2.43.0
#1 lore raw
Thank you for your contribution! Sashiko AI review found 2 potential issue(s) to consider:

Thank you for your contribution! Sashiko AI review found 2 potential issue(s) to consider:

New issues:
- [High] Removing the data-mapping property causes a regression in the Rockchip LVDS controller, silently defaulting the PHY format to 18-bit VESA instead of 24-bit VESA.

Pre-existing issues:
- [High] The px30_lvds_grf_config function incorrectly passes a dynamic value as the mask to regmap_update_bits, preventing it from clearing format configuration bits. --

--- Patch [1]: [PATCH] arm64: dts: rockchip: Drop data-mapping from Engicam PX30 Core --- commit 9e01384e585886fe7dfe0bda8fec7f830eade509
Author: Fabio Estevam <[email protected]>

arm64: dts: rockchip: Drop data-mapping from Engicam PX30 Core

The commit removes the data-mapping property from the panel node to satisfy the panel-simple DT schema. The commit message states this has no functional effect because the DRM OF helpers translate the bus format, falling back to the descriptor's format when the property is absent.

arch/arm64/boot/dts/rockchip/px30-engicam-px30-core-ctouch2-of10.dts panel { · line 31

Quoted 9 lines
arch/arm64/boot/dts/rockchip/px30-engicam-px30-core-ctouch2-of10.dts +0 -1
index 47aa30505a42e..aa636ce45e3a9 100644
@@ -28,7 +28,6 @@ panel {
 		compatible = "ampire,am-1280800n3tzqw-t00h";
 		backlight = <&backlight>;
 		power-supply = <&vcc3v3_lcd>;
-		data-mapping = "vesa-24";

[Severity: High]
Does removing this property inadvertently break the host LVDS controller configuration?

The commit message assumes only the panel-simple driver uses this property, but the rockchip LVDS host controller driver bypasses the framework and reads it directly from the remote panel device tree node:

drivers/gpu/drm/rockchip/rockchip_lvds.c:rockchip_lvds_bind() {
    ...
    if (of_property_read_string(remote, "data-mapping", &name))
        /* default set it as format vesa 18 */
        lvds->format = LVDS_VESA_18;
    ...
}

By removing the property, will this fallback to LVDS_VESA_18 be triggered, causing a mismatch with the 24-bit VESA panel and corrupting the display output?

[Severity: High]
This isn't a bug introduced by this patch, but while reviewing the rockchip LVDS driver's format configuration, I noticed a potential issue with how the register bits are cleared:

drivers/gpu/drm/rockchip/rockchip_lvds.c:px30_lvds_grf_config() {
    ...
    return regmap_update_bits(lvds->grf, PX30_LVDS_GRF_PD_VO_CON1,
                              PX30_LVDS_FORMAT(lvds->format),
                              PX30_LVDS_FORMAT(lvds->format));
}

Does this incorrectly pass the dynamic value PX30_LVDS_FORMAT(lvds->format) as both the mask and the value?

Because regmap_update_bits() uses the mask argument to determine which bits to clear in the lower 16 bits, passing the value as the mask means any 0-bits in the format will be 0 in the mask and ignored instead of cleared.

If the bootloader left the format in a different state (e.g., LVDS_JEIDA_18), will the kernel fail to clear the necessary bits to switch it to LVDS_VESA_24?

Quoted 3 lines
arch/arm64/boot/dts/rockchip/px30-engicam-px30-core-ctouch2-of10.dts +0 -0
 		port {
 			panel_in_lvds: endpoint {
Signature · 1 line
Sashiko AI review · https://sashiko.dev/#/patchset/[email protected]?part=1
maintainer Applied
#2 lore raw
In reply to Fabio Estevam
Quoted 6 lines · Fabio Estevam

The panel-simple schema only allows data-mapping overrides for panels whose mapping is not fixed by their compatible. Consequently, the property on the Ampire AM-1280800N3TZQW-T00H panel fails validation:

(ampire,am-1280800n3tzqw-t00h): False schema does not allow ['vesa-24']

Quoted 7 lines

The panel descriptor already selects 8 bits per color and MEDIA_BUS_FMT_RGB888_1X7X4_SPWG. The DRM OF helpers translate "vesa-24" to that exact bus format, and panel-simple falls back to the descriptor's format when the property is absent. Removing the property therefore has no functional effect.

[...]

Applied, thanks!

[1/1] arm64: dts: rockchip: Drop data-mapping from Engicam PX30 Core

      commit: 9f23009c60a3787c9bdb1eedf725a691537dfd5c

Best regards,

Signature · 1 line
Heiko Stuebner <[email protected]>