I wrote a patch (attached) for a standalone program that takes a packet on stdin and passes it to parse_packet, for the purpose of fuzzing. Please let me know if I have approximately the right approach.
I've been running this with afl-fuzz (american fuzzy lop) for a few days on a single CPU; so far no crashes.
I wrote a patch (attached) for a standalone program that takes a packet
on stdin and passes it to parse_packet, for the purpose of fuzzing.
Please let me know if I have approximately the right approach.
Thank you for the patch, I put it in the contrib/ directory in the
source tree. Where people can apply it to compile the unbound-fuzzme
program for fuzzing. But it is not included in the main codeline, since
most distros do not need to ship it to normal users.
I am fine with you testing the code and appreciate the reports that come
out of that.
Thank you for the patch, I put it in the contrib/ directory in the
source tree. Where people can apply it to compile the unbound-fuzzme
program for fuzzing. But it is not included in the main codeline, since
most distros do not need to ship it to normal users.
Excellent, thanks. I'll proceed with submitting it to OSS-Fuzz.
In reading the OSS-Fuzz instructions, I noticed they recommend that the fuzzing binary be built and run regularly as part of "make test." If I refactored this to be part of "make test," would you accept it as part of the main codeline?