Thank you for your contribution! Sashiko AI review found 4 potential issues to consider: - [Medium] dt-bindings: display: cix,sky1-dptx: missing output port and required phys - [Low] dt-bindings: display: cix,sky1-dptx: non-standard cix,dp-max-rate property - [Low] dt-bindings: display: cix,sky1-dptx: instance index properties in binding - [Low] dt-bindings: display: cix,sky1-dptx: unused dp0 label in DTS example --
commit 8c60d6c22ca4c6d5157c09f9c5eab3f993001d1b Author: Luo Liu <[email protected]> dt-bindings: display: add CIX Sky1 DisplayPort transmitter This commit adds the Devicetree binding schema and MAINTAINERS entry for the CIX Sky1 DisplayPort transmitter controller. > diff --git a/Documentation/devicetree/bindings/display/cix,sky1-dptx.yaml > b/Documentation/devicetree/bindings/display/cix,sky1-dptx.yaml > new file mode 100644 > index 0000000000000..9087b5edd9f02 > --- /dev/null > +++ b/Documentation/devicetree/bindings/display/cix,sky1-dptx.yaml > @@ -0,0 +1,108 @@ [ ... ] > + phys: > + maxItems: 1 > + description: DisplayPort PHY the controller transmits through. > + > + port: > + $ref: /schemas/graph.yaml#/properties/port > + description: > + Video input from the display pipeline that drives this controller. [Severity: Medium] Should this binding define ports with both an input port from the display pipeline and an output port to the downstream DisplayPort connector, eDP panel, or Type-C bridge, and mark phys and ports as required? With only a single input port and additionalProperties: false, board DTS files that connect the transmitter output to a connector or panel via ports will fail dtbs_check validation. In addition, cix_dp_parse_dt() unconditionally calls devm_phy_get(dev, NULL) and fails probe if phys is absent, yet phys and the OF graph port are omitted from both required and the example. > + > + cix,adapter-port: > + description: > + RCSU video adapter port of this transmitter. Sky1 numbers its five > + transmitters from 1 to 5. Port 3 is the native DP/eDP output. > + $ref: /schemas/types.yaml#/definitions/uint32 > + minimum: 1 > + maximum: 5 > + > + cix,dp-max-rate: > + description: > + Maximum link rate the board is wired for, in units of 10 kbit/s. > + $ref: /schemas/types.yaml#/definitions/uint32 > + enum: [162000, 270000, 540000, 810000] > + default: 810000 [Severity: Low] Could this use the standard link-frequencies property from video-interfaces.yaml on the output endpoint instead of a custom cix,dp-max-rate property? The values 162000, 270000, 540000, and 810000 in 10 kbit/s units come from Linux DRM's internal drm_dp_bw_code_to_link_rate() representation and lack a standard unit suffix, whereas DisplayPort bindings normally specify board-supported link rates in Hz via link-frequencies on the output endpoint. > + > + cix,stream-id: > + description: > + Index of the video stream this controller uses. Sky1 has two streams, > + numbered 0 and 1. > + $ref: /schemas/types.yaml#/definitions/uint32 > + minimum: 0 > + maximum: 1 > + default: 0 [Severity: Low] Should cix,adapter-port and cix,stream-id be avoided in favor of describing hardware differences via compatible strings or OF graph endpoints? In cix_dp_cfg_adapter(), each instance already maps its own rcsu region from reg, and cix,adapter-port is only checked against port 3 because the native DP/eDP port uses a different RCSU register layout (offset 0x320) than the USB-C DP ports (offsets 0x300 and 0x304). Similarly, cix,stream-id encodes the display pipeline stream index as a custom vendor property rather than representing the stream routing through OF graph ports or endpoints. [ ... ] > +examples: > + - | > + #include <dt-bindings/interrupt-controller/arm-gic.h> > + > + dp0: dp@14064000 { [Severity: Low] Can the unreferenced dp0 label be dropped from the example? It is never referenced within the example and triggers an unused-labels warning from scripts/dtc/dt-check-style in strict mode. -- Sashiko AI review · https://sashiko.dev/#/patchset/[email protected]?part=1
