The Carloop library was covered by code review; it appears to not be the issue. See the earlier post here:
can.begin() is part of the Particle HAL. In the example code given, the next call was a Particle.process().
At the moment, the focus is on the can object in the system firmware along with how it might interact with the process function in the system thread, given the RTOS is multithreaded (application thread & system thread).
If you can setup a complete debugging environment and rebuild the complete firmware with debugging enabled, then we would welcome digging into the 0.7.0 system firmware.
The Carloop people are aware. However, I do not know if they have determined any cause yet for the SOS behaviour.