With the nocodec topology, the existing code generates an oops on insert. It looks like the cleanups are not done right with race conditions and sysfs entries inconsistent. The patch in #255 solves these issues by adding a second unregistration of the machine driver - which works but makes absolutely no sense... Clearly a timing issue that needs to be handled with a proper fix.
dmesg_before_remove.log
dmesg_after_remove.log
dmesg_after_insert.log