OpenThread RCP Dongle USB hang

[krbvroc1]

I am using NCS 2.6.1 and have programmed a nrf52840dongle (USB Dongle) with the build result of nrf/samples/openthread/coprocessor

The problem I am having is that normally if I power up the border router the device will be detected okay. However, if I reboot the border router I am using, the device is hung. I am suspecting it is a problem (bug?) with the openthread dongle firmware for a few reasons:

1) If I unplug and re-plug the device it enumerates without issue

2) If I press the side button on the dongle, the red light immediately starts pulsing and I see the device enumerate in the DFU bootloader mode

Here is the output on a reboot:

[   22.203362] usb 1-7: device descriptor read/64, error -110
[   37.819340] usb 1-7: device descriptor read/64, error -110
[   38.059182] usb 1-7: new full-speed USB device number 3 using xhci_hcd
[   53.435342] usb 1-7: device descriptor read/64, error -110
[   69.051332] usb 1-7: device descriptor read/64, error -110
[   69.163504] usb usb1-port7: attempt power cycle
[   69.579164] usb 1-7: new full-speed USB device number 4 using xhci_hcd
[   85.051516] usb 1-7: device descriptor read/8, error -110
[  100.667550] usb 1-7: device descriptor read/8, error -110
[  100.907174] usb 1-7: new full-speed USB device number 5 using xhci_hcd
[  116.283543] usb 1-7: device descriptor read/8, error -110

And it never recovers without removing power from the dongle. Of course this is not acceptable for use in a remote border router unit that will be rebooted from time to time.

The border router is a Linux box.

[Amanda Hsieh]

Hi, 

I am using NCS 2.6.1 and have programmed a nrf52840dongle (USB Dongle) with the build result of nrf/samples/openthread/coprocessor

Do you modify the sample? If so, what is that?

Could you build for debugging? In the build configuration menu, under Optimization level, select the option Optimize for debugging. And then provide the debug log. 

Regards,
Amanda H.

[krbvroc1]

I do not modify the example at all. Since this is a radio co-processor that is used by the OpenThread border router via the Spinel protocol, I have no idea how one would debug this. Where do debug messages even go? Is there some separate CDC ACM debug console made available? Of all problems, this seems like a disaster to debug.

I do not use VS Code.

[Amanda Hsieh]

See Logging extension and Rebooting to bootloader for nrf52840dongle. 

[krbvroc1]

Unfortunately I don't see how this is going to help. Since the hang only occurs when the host is rebooted, and the device no longer enumerates on the USB bus, there is no way the host can receive any log files. Since the JLink connector is a 'no legs' version of the TagConnect, I cannot even connect to that to attempt RTT log viewing. Are you aware of other reports of this issue? I am experiencing it with multiple nrf52840dongle and with multiple host computers. The only online reports of this issue I had seen was with early engineering versions of the 52840 chip. I wonder if this is a Zephyr ACM/CDC regression.

[Amanda Hsieh]

Is it 100% reproducible or does it only happen once a time? Could you elaborate on the border router used to reproduce (how is the thread stack built and how is it configured)?

[jn.chou]

Hi Amanda,

I'm encountering a consistent issue with the pca10059 ncs2.6.1 thread coprocessor sample on my setup, which includes:

- Linux pve 6.9.8-x64v3-xanmod1 #0~20240705.g77bcbc2 SMP PREEMPT_DYNAMIC Fri Jul 5 16:18:13 UTC x86_64 GNU/Linux
- An Intel NUC where I pass through the dongle to Home Assistant as a Thread dongle.

Whenever I use the ncs2.6.1 thread coprocessor sample, it frequently results in errors like these:
```
[10998.074483] usb 1-2: new full-speed USB device number 16 using xhci_hcd
[11013.454063] usb 1-2: device descriptor read/64, error -110
...
[11091.915835] usb 1-2: device descriptor read/8, error -110
[11092.023888] usb usb1-port2: unable to enumerate USB device
```
This problem occurs every time without fail.

However, when I switch to using github.com/.../ot-nrf528xx, there are no issues at all.