DavidKorczynski
DavidKorczynski
`unblob` is not resolved, but the problem here looks to be a failing coverage build
Apologies for the delay! Will land this once the CI is green!
/gcbrun skip
/gcbrun skip
/gcbrun skip
This happens because the OSS-Fuzz build for `flex` patches the `main` function to enable analysis: https://github.com/google/oss-fuzz/blob/74e3c9413ac02ea72bc3be161096a390f05de846/projects/flex/build.sh#L26 As such, there is a `main2` function in the binary artifacts build by the...
I think the issue highlighted by this PR may need a more substantial fix. There is mixing of dynamically allocated memory and non-dynamically allocated memory, and in order to achieve...
> I do have some reservations with regards to the fuzzing harness. The input string is required to be null-terminated and padded with ZONE_BLOCK_SIZE. This is one of the assumptions...
> Null-termination should happen before the padding. So, given `foo bar baz\0...`, to determine the length of the final token, i.e. `baz`, the pointer to `baz\0...` is subtracted from the...
Corpus is nice, but due to code size I'm not sure it's a concern as such in this case. I think increasing the amount of seconds spend fuzzing may have...