【Nvidia】How to Activate a Camera on Nvidia Platform in Details

# 0. Pre

There are very rich documents to help us to develop a driver on Nvidia platform. Further more, there are also camera driver development guide like `Jetson GMSL Camera Framework Guide nv.pdf`. [annexe 4]

But if you do not have much experience on driver development, It could be really hard to know how to start the task. So maybe this blog could be helpful for you not only understand the architecture about how to do it but also gives a best prectice about active two camera ==imx390c(max9295a+GW5200+GSML2)== and == OX03C10+max96717F+GSML2== on ==Jetson Orin Nano 4G== with ==a CSI-2-GSML board==.

``` mindmap

# hardware
## camera
### imx390C(with ISP GW5200 on camera)
### OX03C10+max96717F(without ISP on camera)
## Jetson Orin Nano 4G (Part 1. Nvidia Jetson Platform )
## csi 2 gsml board (Part 2.)
## FPC
## GSML Line
# Software
## Nvidia Platform Software(Part 3.)
### SDKManager
### Jetpack, L4t, Supported SDK
### Kernel Compile
## Understand the Architect of the driver development especially camera driver on Nvidia(Part 4.)
###	Device Tree
### Driver
### Register Config & Activation File
### ISP(optinonal)
```

# 1. Nvidia Jetson Platform

![](/media/202408/2024-08-27_174745_4157010.5320168426563784.png)

* when you get an nvidia embedded platform, this file could help you to distinguish which type of platform that you are working on.

* In the filename `/boot/dtb/kernel_tegra234-p3767-0004-p3768-0000-a0.dtb`, the terms p3767 and p3768 refer to specific `hardware platform` and `carrier board` configurations for NVIDIA Jetson devices.
	* `p3767`: This typically represents the module or the system-on-chip (SoC) variant. For example, it could refer to the Jetson Orin Nano or another variant in the Jetson Orin family.
	* `0004`: This represent the different parameter of the SoC. For example, 0004 (orin nano 4G), 0003 (orin nano 8G), 0000 (orin nx)
	* `p3768`: This generally represents the carrier board or baseboard to which the module is connected. It defines the specific layout and connectivity features that the module will interact with.

![](/media/202408/2024-08-27_175526_3888160.5014229862003617.png)

* AGX Orin 64GB / 32GB / Industrial: p3701 (for AGX Orin series modules)
* Orin NX 16GB / 8GB: p3767 (for Orin NX series modules)
* Orin Nano 8GB / 4GB: p3767 (for Orin Nano series modules)
* AGX Xavier Industrial / 64GB / 32GB: p2888 (for AGX Xavier modules)
* Xavier NX 16GB / 8GB: p3509 (for Xavier NX modules)
* TX2 NX: p3489 (for TX2 NX module)
* Jetson Nano: p3448 (for Jetson Nano module)

![](/media/202408/2024-08-27_175641_8397890.4310180255826216.png)

# 2. Architecture of the physical connections

* csi1, csi2, cs3
* 22pin, 15pin, 30pin, 40pin
* pin15

![](/media/202409/2024-09-05_174840_6639750.05357545217873516.png)

![](/media/202409/2024-09-13_143413_0001950.2723714524056694.png)
![](/media/202409/2024-09-13_143413_2318270.7046382340937376.png)
![](/media/202409/2024-09-13_143413_4459520.34260866820001656.png)

# 3. SDKManager, Jetpack SDK, L4T(Linux for Tegra), Toolchain, Supported SDK

## 3.1 SDKManager
* SDK Manager is used to install Jetpack & Jetpack supported on the Hardware Platform (Devkit or clb256)

![](/media/202408/2024-08-27_181248_8905320.30314551247577026.png)

![](/media/202408/2024-08-27_180628_8359920.1727327236452998.png)

``` bash
sparkmind@sparkmind-S14-72:~$ docker images
REPOSITORY   TAG                        IMAGE ID       CREATED        SIZE
sdkmanager   2.1.0.11698-Ubuntu_20.04   cf06d9276abf   3 weeks ago    966MB
sdkmanager   2.1.0.11698-Ubuntu_18.04   b48827454275   3 weeks ago    908MB
sdkmanager   2.1.0.11682-Ubuntu_22.04   45897d7a36d1   2 months ago   970MB

# ~/.bashrc
alias nv18='docker run -it --privileged --network=host --rm -v /home/sparkmind/Nvidia/:/home/nvidia/Downloads/nvidia/ -v /home/sparkmind/nvidia:/home/nvidia/nvidia sdkmanager:2.1.0.11698-Ubuntu_18.04 --cli'
alias nv20='docker run -it --privileged --network=host --rm -v /home/sparkmind/Nvidia/:/home/nvidia/Downloads/nvidia/ -v /home/sparkmind/nvidia:/home/nvidia/nvidia sdkmanager:2.1.0.11698-Ubuntu_20.04 --cli'
alias nv22='docker run -it --privileged --network=host --rm -v /home/sparkmind/Nvidia/:/home/nvidia/Downloads/nvidia/ -v /home/sparkmind/nvidia:/home/nvidia/nvidia sdkmanager:2.1.0.11682-Ubuntu_22.04 --cli'

```
`Each version of nvidia sdkmanager has its correspondant ubuntu version, but docker could be used no matter on which ubuntu.(Bug exists.) `

## 3.2 Jetpack

![](/media/202408/2024-08-27_181435_7369840.2012800962132607.png)
![](/media/202408/2024-08-27_180210_4795530.40108344501847926.png)

## 3.3 L4T

![](/media/202408/2024-08-27_180432_7432840.29501461370076465.png)

## 3.4 Kernel Customization

Follow the link below it will help you to customize the kernal. dtb and Image/.ko file.
https://docs.nvidia.com/jetson/archives/r35.4.1/DeveloperGuide/text/SD/Kernel/KernelCustomization.html?highlight=nvbuild%20sh

# 4. Camera Architecture Stack

The picture below gives you whole picture of camera Architecture Stack. 
The main part of the camera driver development are:
* Device Tree Create [sensor device tree](https://docs.nvidia.com/jetson/archives/r35.4.1/DeveloperGuide/text/SD/CameraDevelopment/SensorSoftwareDriverProgramming.html#device-tree)
* Implement the custom camera driver with `Camera Core/Tegra Drivers` structure.
* Verify the register Config with the activate Ser/Des file or the documents from the sensor Company.
* Use V4l2 to test the result

![](/media/202408/2024-08-27_181751_5463240.6298154109833365.png)

* Config the Register of the Ser,Des,Sensor by I2C.

![](/media/202409/2024-09-13_115618_8139670.4553539620854772.png)

![](/media/202408/2024-08-27_183003_3455080.5871638661817971.png)

![](/media/202408/2024-08-27_182725_5970940.8774663697145837.png)

![](/media/202408/2024-08-28_150732_4471920.828899657857988.png)

* Config the Device Tree

![](/media/202408/2024-08-27_183104_0485740.23889118832708267.png)

* Write the driver of the camera by the configuration of the Register

![](/media/202408/2024-08-27_183202_8853290.979786151795299.png)

# Annexe

* Jetpack Archive: https://developer.nvidia.com/embedded/jetpack-archive 
* Jetson Linux Archive: https://developer.nvidia.com/embedded/jetson-linux-archive
* sdkmanager: https://developer.nvidia.com/sdk-manager

![](/media/202409/2024-09-18_112755_0509510.9698585452027437.png)