How to configure MIPI lanes for non-I2C camera sensor on Jetson Orin Nano

Issue Overview

Users are experiencing difficulties configuring and capturing frames from a non-I2C camera sensor connected to the Jetson Orin Nano developer kit. The camera is connected to the J20 or J21 connector and only uses MIPI data lanes without an I2C interface. The goal is to receive MIPI data through tegra-capture-vi, but users are encountering various issues, including kernel crashes, frame capture failures, and MIPI-related errors.

Possible Causes

1. Incorrect device tree configuration: The device tree may not be properly configured for a non-I2C camera sensor, leading to communication issues.

2. Mismatched lane configuration: The lane configuration in the device tree may not match the physical connection or camera capabilities.

3. Timing and synchronization issues: Incorrect clock settings or timing parameters could cause frame capture failures.

4. Driver compatibility: The existing camera driver may not be fully compatible with the non-I2C sensor, requiring modifications.

5. MIPI signal integrity: Issues with signal integrity, such as deskew requirements, may affect data reception.

6. Kernel module conflicts: Incompatibilities between the custom camera module and the Jetson Orin Nano’s kernel could lead to crashes or malfunctions.

Troubleshooting Steps, Solutions & Fixes

1. Modify the device tree:

   • Create a custom device tree file for your camera sensor.
   • Remove all I2C-related configurations.
   • Set the correct number of lanes, typically 2 for this setup.
   • Add the lane_polarity=6 property to the device tree.
   • Example snippet:

     mode3 {
         mclk_khz = "104250";
         num_lanes = "2";
         tegra_sinterface = "serial_c";
         phy_mode = "DPHY";
         discontinuous_clk = "yes";
         dpcm_enable = "false";
         cil_settletime = "0";
         dynamic_pixel_bit_depth = "12";
         csi_pixel_bit_depth = "12";
         mode_type = "bayer";
         pixel_phase = "bggr";
         active_w = "640";
         active_h = "1152";
         readout_orientation = "0";
         line_length = "640";
         inherent_gain = "1";
         mclk_multiplier = "4";
         pix_clk_hz = "416666666";
         // ... other properties ...
     };
     

2. Adjust the camera driver:

   • Use a reference sensor driver as a template.
   • Remove all I2C read/write code from the driver.
   • Modify the driver to report fewer lines if encountering CHANSEL_SHORT_FRAME errors.
   • For debugging, remove all modes except the one being tested.

3. Verify MIPI lane configuration:

   • Ensure the camera is connected to the correct connector (J20 or J21).
   • Confirm that the connector supports the required lane configuration (1×2 or 1×4).
   • Double-check the physical connections match the device tree configuration.

4. Calculate and set the correct pixel clock:

   • Use the formula: pixel_clk_hz = (sensor data rate per lane * number of lanes) / bits per pixel
   • For 2.5 Gbps, 2 lanes, and 12 bits per pixel: pixel_clk_hz = (2500 * 2) / 12 = 416666666
   • Update the pix_clk_hz value in the device tree accordingly.

5. Address deskew requirements:

   • Consult the camera vendor for specific deskew signal requirements.
   • Look for register configurations in the camera datasheet that may control deskew settings.

6. Debugging and logging:

   • Use v4l2-ctl to list supported formats and verify driver configuration:

     v4l2-ctl -d /dev/video0 --list-formats-ext
     

   • Capture and analyze trace logs for detailed error information:

     # Refer to https://elinux.org/Jetson/l4t/Camera_BringUp for trace log capture instructions
     

   • Check dmesg logs for kernel-level errors.

7. Investigate MIPI errors:

   • For PHY_INTR0 register information, consult the Tegra Reference Manual (TRM).
   • Check the NVCSI_PHY_0_CILA_INTR_0_STATUS_CILA_0 register in the TRM for status code meanings.

8. Handle specific errors:

   • For CHANSEL_SHORT_FRAME errors, verify that the output size (height) matches what the driver reports.
   • For CHANSEL_NOMATCH errors, ensure the camera’s output format matches the driver’s expectations.

9. Kernel module management:

   • If kernel crashes occur, try rebuilding the kernel module with debug symbols.
   • Use insmod and rmmod to load and unload the module for testing.

10. MIPI signal integrity:

    • Verify that the camera’s MIPI CSI-2 data speed (e.g., 2.5 Gbps) is supported by the Jetson Orin Nano.
    • Check cable quality and length to ensure signal integrity.

If issues persist after trying these steps, consider reaching out to NVIDIA developer support or consulting the camera manufacturer for specific configuration requirements.