extending the fuzzers

Hi!
I have played around with the curl-fuzzers running in oss-fuzz. I am
impressed by how the existing fuzzers manage to emulate network traffic
and get such an extensive coverage!

I am doing some experiments with fuzzing the internals of curl, and so
far have found
 * doh buffer overwrite
 * doh resource leak
 * undefined behaviour in doh parsing
all of which have been fixed already (thanks Daniel!)
There is more in the pipe, but not yet fit for public discussion.

I am running the existing fuzzers, to look for under exposed parts of
curl. It is however quite slow to get the coverage up, so I wonder if I
could get a speed boost by receiving a copy of the fuzz corpus from
oss-fuzz?
I have spent something like 20 cpu days so far, and still finding new
paths. I assume the oss fuzz corpus must have had several cpu years by now.

This would help me focus my contributions to where it matters.

Thanks,
Paul
-------------------------------------------------------------------
Unsubscribe: https://cool.haxx.se/list/listinfo/curl-library
Etiquette: https://curl.haxx.se/mail/etiquette.html
Received on 2019-09-17