Additional support re: non-udev-managed devices

Is there a way to pass non-udev-managed devices into a container? For examples, the NVIDIA kernel driver writes these devices (which aren't monitored via udev, therefore the container doesn't start as udev cannot find them under DEVNAME):

crw-rw-rw- 1 root root 195,   0 Oct 20 13:36 /dev/nvidia0
crw-rw-rw- 1 root root 195,   1 Oct 20 13:36 /dev/nvidia1
crw-rw-rw- 1 root root 195, 255 Oct 20 13:36 /dev/nvidiactl
crw-rw-rw- 1 root root 507,   0 Oct 20 13:34 /dev/nvidia-uvm
crw-rw-rw- 1 root root 507,   1 Oct 20 13:34 /dev/nvidia-uvm-tools

/dev/nvidia-caps:
total 0
cr-------- 1 root root 234, 1 Oct 20 13:36 nvidia-cap1
cr--r--r-- 1 root root 234, 2 Oct 20 13:36 nvidia-cap2

Char nodes 195:(0/1/255) 507(0/1) and 234(1/2).

Please let me know if this request (to allow static/non-udev device nodes) is considered in-scope for this project,

Thank you!

首先，检查 container-hotplug 如何发现由 udev 管理的设备并将其传递给 Docker 容器。确定可以在哪里配置静态设备节点或非 udev 管理的设备节点，包括列出的 NVIDIA 字符设备 major 和 minor 编号；完成的标准是确定范围，并有办法验证这些设备在容器中可用。