I’ve thought in dividing the input into two i2c entries (CAM_A@24 and CAM_B@24), each of them with num_lanes = “2”;. Then, connect each of them to a NVCSI port, and then merge the output of both port outputs, to a single input port in the VI block. Would this be correct? Simplified, something like this:
I took that as reference, but yet it is connected to 6 VI channels (NVCSI channel <-> VI channel). As my video source is only one, they’ll have to merge somewhere. Would it be possible to merge it as I depicted in the image of the post in a single channel of the VI block? Or they don’t need to be explicitly merged?
But how do you manage then a 4-lane camera, with a single slave address, and without the I2C mux? It is connected to two interfaces (CSI_A and CSI_B). Is it possible?
Just 1 sensor, which hardware is configured to send data through CSI_A and CSI_B (with CSI_A clock) as specified in the documentation. Anyway, as it is a single sensor, it has only one slave address. The solution proposed in comment #8 would be valid?
For a single sensor only you can reference to tegra186-camera-imx274-a00.dtsi
the csi-port and bus-width do the things. YOu should have a check the document for the details.