17.06.2014 21:46, Sylvain Munaut пишет:
Bunch of fucking idiots who chose as test data completely invalid FN count values that can't possibly happen in a GSM system ...
I presume the idea was not to test gsm use cases but to cover all the internal states of the cipher.
- Either just use different test vector you generate
I would rather stick to official test vectors published in standard.
- Or use a marker bit like (1<<31) to indicate to _a5_{3,4} not to
convert the values for testing
I really like my initial proposal: - it works already - it's simpler
Could you emphasize - why exactly introducing private header is such a disaster that we have to waste time and efforts trying to not let it happen?