For some reason, there's a local python3 installation on the
oss-fuzz-base image with a pretty old pip version, 19.2.3, which doesn't
have up-to-date wheel platform tags, so, when installing SPDK
dependencies, it tries to compile them instead of using the wheels. The
compilation fails on the grpcio package for some reason.
So, to resolve this, upgrade the local pip version when building the
SPDK image, so that it's able to fetch the pre-compiled packages. This
is a workaround for #7914.
* Add more fuzzer
Fix build script
* Fix formatting for fuzzers
Fix length checking for nonce in fuzz_aead
* Fix fuzzer
- Add more randomization to data input
* Swap from `compile_go_fuzzer` to `compile_native_go_fuzzer`
* Move building to an internal script `fuzz/oss-fuzz-build.sh` to ease addition of new fuzzing methods
* Update `auto_ccs`
* ghostscript: extend fuzzing
- Add two new fuzzers
- One new fuzzer that explore multiple colorings
- One new fuzzer that focuses on PDF exploration
- Add dictionary to PDF fuzzer
- Add small sized PDF seed to PDF fuzzer
- Ensure max_len on multiple coloring fuzzing
* Add myself to auto_ccs
* Fix typo
* set indentation
* add options file
* Address review: split fuzzers and have dict + seed locally
* nit
* clarify prepending of single byte
* fix styling
* clarify comments in all_colors fuzzer
* Addess final review
Using bfd_check_format_matches requires that a returned matching
vector (when return is false and bfd_get_error() is
bfd_error_file_ambiguously_recognized) be freed. Use bfd_check_format
instead.
to make it easier to catch issues like
https://github.com/google/honggfuzz/issues/465.
With -V when unreproducible crashes are triggered honggfuzz shows something like
```
[2022-06-10T15:21:12+0000][W][2373157] subproc_checkTimeLimit():532 pid=2374490 took too much time (limit 30 s). Killing it with SIGKILL
Sz:280 Tm:30,021,052us (i/b/h/e/p/c) New:0/0/0/0/0/3, Cur:0/0/0/0/0/591
Launching verifier for HASH: 1b859677dd (iteration: 1 out of 5)
[2022-06-10T15:21:13+0000][E][2373157] fuzz_runVerifier():280 Verifier stack mismatch: (original) 1b859677dd != (new) 0
```
* croniter: instrument fuzzing function
The Croniter build is having some troubles with coverage and I'm not
entirely sure why. It seems like the corpus is not there and I'm not
entirely sure if this has to do with some instrumentation underlyings.
Locally the end-to-end process of running, collecting seeds and
generating coverage works for me. The effort in this PR is to make the
setup similar to pyyalm where the coverage visualisation works -- the
only difference I could spot between fuzzers from the two projects is
that croniter does not have its fuzzer entry function instrumented.
* Instrument all. This is similar to protobuf-python
Kevin Backhouse