fprobe-ng uses a weak hash function (based on CRC16). It is likely that it is subject to the usual DoS attacks.
Hello Florian Weimer wrote: Even though it's probably upstream's choice how he wants to fix this, do you have any suggestions for a proper hash function (md5 or sha1)? Radu
* Radu Spineanu:
Personally, I think the computational overhead of really safe hash
functions is too high, and you should use balanced binary trees
instead. However, this view is not universally agreed upon.
The Linux networking code uses Bob Jenkins' hash function ("jhash"),
and the fprobe package follows this example. I don't know of any
serious attempts of cryptanalysis, but my gut feeling is that jhash is
still to weak. It's better than CRC, though.
Has there been any progress in the last 6.5 years?
Cheers,
Moritz