How to catch and handle M4 watchdog crashes on Linux side?

[Mario Luzeiro]

I managed to make the M4 WWDOG to work.

I got on Linux cmd line:

remoteproc remoteproc0: crash detected in m4: type watchdog

How can I catch and handle this events programmatically on a shell script, C or Python ?

[ArnaudP]

Hello Mario,

Different strategies are possible.

1) The polling method

• disable the auto recovery

echo disabled >/sys/class/remoteproc/remoteproc0/recovery

• poll the remoteproc state, looking for the "crashed" stated

cat /sys/class/remoteproc/remoteproc0/state

• recover the M4

echo recover >/sys/class/remoteproc/remoteproc0/recovery 

2) Use coredump trigger

An event is sent to the userspace application on crash to save the coredump, this can also be used to detect the crash

Details and link to udev script are available on this page:

https://wiki.st.com/stm32mpu/wiki/How_to_retrieve_Cortex-M4_logs_after_crash

3) Create your own udev trigger based on virtio device add/remove ( work only if you use RPMsg)

this page provide details on how to implement it:

https://www.tecmint.com/udev-for-device-detection-management-in-linux/

The rules to implement is:

SUBSYSTEM=="virtio", ACTION=="add", RUN+="/bin/device_added.sh"
SUBSYSTEM=="virtio", ACTION=="remove", RUN+="/bin/device_removed.sh"

4) tips to ease your test

you can simulate a watchdog crash using following command:

echo 1 >/sys/kernel/debug/remoteproc/remoteproc0/crash

Regards,

Arnaud