X_LINK_UNBOOTED on CM4 - RuntimeError: No available devices

BWixted

Hi,

I've got a remote CM4 that is stuck with the above error.

Attempting to call any method that would try and load a device returns:

[warning] skipping X_LINK_UNBOOTED device with name "1.1"
[warning] skipping X_LINK_UNBOOTED device having name "1.1"

udev rules are correct. device_manager.py suffers the same issue. lsusb contains the "Intel Movidius MyriadX" device

dai.Device.getFirstAvailableDevice(skipInvalidDevices=False)
returns:
True, DeviceInfo(name=1.1, mxid=, X_LINK_UNBOOTED, X_LINK_USB_VSC, X_LINK_MYRIAD_X, X_LINK_ERROR)

Rebooting the pi does not solve the issue.

This device has been successfully working continuously for months. This issue has appeared after remotely accessing the pi to run some image/data collection code (which was unchanged from previous runs). The only thing I can think of that could have caused the issue was closing an IDE terminal window while either the device or pipeline was initializing.

All of the code has been run on an identical device (as identical as possible - same batch, same os version, same libraries, etc.) with no issue.

BWixted

I've already replicated this, and I think I've realized what's happened - I'd moved the dai.Device() call into a class that sets up the camera pipeline so that I could grab calibration data directly off the camera and set the camera focus. Immediately after initializing that class, its reference to device is used properly (with camera.device as device: ). Closing the code in the (tiny) window between the initialization and the with: appears to be the cause. Closing it after the with: doesn't cause any problems. Closing it prior to the dai.Device() call also doesn't cause any problems. These calls are almost immediately adjacent. Power cycling fixes it, but I don't expect there's any way to do that without physical access.

jakaskerl

BWixted

BWixted Power cycling fixes it, but I don't expect there's any way to do that without physical access.

It the SOM that is the issue so restarting it would probably work. On CM4, GPIO1
is module reset. Pulling that low should reset the SOM.

Thanks,
Jaka