We’re working on passing through hardware from host device to virtual machine.
Devices we’re trying to passthrough are ethernet controller, GPU controller, USB controller, PCIe controller.
All this devices are dependent on BPMP, Nvidia Boot and Power Management processor.
Is it possible to make passthroughs and virtualization on BPMP?
Are there any official technical view/opinion/documentation exists on virtualization and hardware passthroughs?
Virtualization/passthrough orin hardware question
Would like to understand the query. Do you mean you don’t want to initialize the hardware blocks in BPMP stage, and all blocks are initialized in kernel stage?
My question if it is possible to split the device tree description of BPMP into parts and use some of this parts description in the vitrtual machine kernel so we can isolate some hardware driver (ethernet as example).
I’ll try to explain it a bit more.
We have a host ( nvidia orin) and would like to passthough the ethernet device (nvethernet) to the guest virtual machine on the host ( qemu-based virtual machine, virtio-platform passthrough).
To do this we need to passthough some parts of BPMP controller. Question if it is possible and how?
If it is not, if there is official NVIDIA approach to device passthroughs?
Have checked with our teams and confirmed this is not supported. BPMP part is prebuilt and cannot be customized.
This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.