Hi Markus, Thanks for the patches! I separately committed a fix for the buffering=0 issue, just so it's separate and not part of the other changes. > rfkill.py: migrate to python3 Applied, but I made some cleanups. > rfkill: update rfkill.h > rfkill: support hard block reason in python code > rfkill: support hard block reason in C code This is broken, unfortunately. Updating rfkill.h as is causes issues because we read more data than we are willing to accept, which you've fixed up in the support for hard block reasons patch. Seems that should partially be in the original patch - e.g. checking "size >= V1" instead of "size == V1". Additionally, the hard block reasons in C patch is wrong because if the read was short, the field is not initialized. Need to clear the event to zeroes before reading it, or at least at the beginnning (I guess we can assume the kernel will not change size between two reads). The python patch I applied with some fixups. johannes
