I am a beginner using Jetson and I am currently learning about PVA Authentication. I would like to ask some questions.
I saw in the doc that:
By default, only signed application code is allowed to execute on the PVA hardware. Software libraries provided by NVIDIA, such as VPI, will register signatures with the system when the libraries are installed by using a package manager on the target device
- May I ask if PVA Authentication is started by default? Or do I need to configure it?
- When VPI uses PVA, is the authentication process insensitive? What do I need to do?
Looking forward to your reply
May I ask if someone can help confirm? Thank。
Sorry for the late response, our team will do the investigation and provide suggestions soon. Thanks
Could share the document link so we can check with our internal team?
In general, we expect users to access PVA through VPI.
VPI takes care of the hardware setting so users don’t need to do it manually.
The doc is here: PVA Authentication — Jetson Linux Developer Guide documentation
Can we understand it this way,
- When using VPI to access PVA, VPI will automatically register a signature with the system, and I will directly call VPI without involving signature related operations.
- If I develop PVA programs myself, I need to sign the application before I can use PVA.
If that’s the case, how can I sign my own PVA application? I didn’t find an example in the wiki.
You will need to use VPI to access PVA.
We don’t have other API/software for users to directly access the hardware.
Can you help me answer this question?
Yes, you don’t need to handle the signature when using VPI.