Background: For nvidia's jetson orin camera, how is it implemented internally? How are the hardware pipelines related, and what camera-related modules are there inside? For these modules, how to implement them in software? How does the device tree abstract these modules? After the analysis, give us a camera sensor, how to bring up? This article will make an analysis.
The previous article analyzed the code related to nvidia tegra-capture-vi in detail, so starting from the article, we will analyze the code related to nvcsi
As can be seen from the figure below, nvcsi is the input of nvidia about the camera, and the output of nvcsi reaches the vi (video input) module. The code of vi has been analyzed in the previous article, but when analyzing the previous article, it is found that in the last v4l2_async_notifier When registering, you need to rely on the specific v4l2_subdev, so when was v4l2_subdev registered? Through the analysis (code call relationship), we can know that there are related registrations in nvcsi and related registrations in sensor, so it is logical to lead to the analysis of nvcsi in this article as article two, and sensor related leads to the next article three. After the analysis n