How to support a MIPI camera with both 8 and 16 bit formats (2 and 4 lanes) with a single device tree

Robotics & Edge Computing Jetson & Embedded Systems Jetson Orin Nano
1

Using a custom MIPI camera that supports both 8 bit and 16 bit per pixel imaging. The 16 bit format needs 4 MIPI lanes while the 8 bit format only needs two. I can capture both formats by using different device trees because I needed to set “bus-width” to 4 for 4 lanes and to 2 for 2 lanes. Is there a way to have a single device tree that can support both the 8 bit and 16 bit per pixel formats (i.e., either programmatically via the camera driver or by somehow confining the format differences to the sensor “modes”)?

3

hello james.basso,

it should be possible by defining those 2-lane/4-lane settings within different sensor modes.
you should also adding use_sensor_mode_id device tree property, please refer to developer guide, Device Properties.
here’s an example of mode selection.
$ v4l2-ctl -d /dev/video0 --set-fmt-video=width=1280,height=720,pixelformat=RG10 --set-ctrl bypass_mode=0 --set-ctrl sensor_mode=4 --stream-mmap --stream-count=100

4

The port definitions all have a definition of “bus-width” that is not sensor-mode-specific. I need to set “bus-width” to 4 to get 4-lanes, 16 bit to work. However, if I leave the “bus-width” = 4 then 8 bit doesn’t work (even if I change the num lanes and bit depth in modes)

For example:

                        tegra-capture-vi  {
                                num-channels = <2>;
                                ports {
                                        #address-cells = <1>;
                                        #size-cells = <0>;
                                        vi_port0: port@0 {
                                                reg = <0>;
                                                custom_cam_vi_in0: endpoint {
                                                        port-index = <1>;
                                                        bus-width = <4>;
                                                        remote-endpoint = <&custom_cam_csi_out0>;
                                                };
                                        };

                                        port@1 {
                                                reg = <1>;
                                                custom_cam_vi_in1: endpoint {
                                                        port-index = <2>;
                                                        bus-width = <4>;
                                                        remote-endpoint = <&custom_cam_csi_out1>;
                                                };
                                        };
                                };
                        };
5

hello james.basso,

you may revise the 8 bit format as 4-lane setting by lower the pixel clock rate.

7

Unfortunately our camera only uses 2 lanes when in 8 bit pixel mode

8

hello james.basso,

camera software stack loads the device tree settings to initialize camera.
currently, it doesn’t support same camera device with different lane configuration.

9

Could you point me to the code that uses the “bus-width” (i.e., number of lanes). Maybe I can override the dev-tree on the fly?

10

hello james.basso,

you may refer to below for sending numlanes to camera software stack,
$public_sources/kernel_src/kernel/nvidia/drivers/media/platform/tegra/camera/camera_common.c

int camera_common_parse_ports(..)
{
        err = of_property_read_u32(ep, "bus-width", &bus_width);
        s_data->numlanes = bus_width;
11

Thanks Jerry. I will take a look but I will need to find the CSI register that actually uses this to make sure it gets updated. Do you know which register it is? Is it NVCSI_PHY__NVCSI_CIL_CONFIG_0?

12

hello james.basso,

yes.

or,
may I double check which L4T version you’re using?
you may refer to Topic 285027, which shared debug version rce-fw on JP-5.1.1/r35.3.1
it reports NVCSI stream configuration upon running with debug version rce-fw.

13

Thanks Jerry, We are using r36.3

14

hello james.basso,

here’s debug version rce-fw of JP-6.0/r36.3, Topic308937_Oct18.zip (249.8 KB).

you may follow that previous mentioned topic, Topic 285027 for the update steps on AGX Orin.
if you’re based-on Orin NX/Orin Nano series, please check Topic 260583 for steps to execute with l4t_initrd_flash.sh for replacing the camera debug firmware.

16

This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.