mirror of https://github.com/google/oss-fuzz.git
9aa1893d2f
The fuzzer's entire purpose is to use a libFuzzer flag that has no equivalent in AFL. So only build it for libFuzzer. |
||
|---|---|---|
| .. | ||
| BUILD.gn.diff | ||
| Dockerfile | ||
| README.md | ||
| animated_image_decode.options | ||
| api_fuzzers.options | ||
| build.sh | ||
| image_decode.options | ||
| image_filter_deserialize.options | ||
| image_filter_deserialize_width.options | ||
| path_deserialize.options | ||
| project.yaml | ||
| region_deserialize.options | ||
| region_set_path.options | ||
| skia.diff | ||
| textblob_deserialize.options | ||
README.md
When bench testing image_filter_deserialize, it may be useful to have malloc_limit_mb = 500, and timeout = 10 to find actionable OOM culprits sooner.
When reproducing, instead of running python infra/helper.py reproduce ... try running
docker run --rm -i --privileged -v $OSS_DIR/build/out/skia:/out -v [/path/to/testcase]:/testcase -t gcr.io/oss-fuzz-base/base-runner reproduce image_filter_deserialize -runs=100 -malloc_limit_mb=100
So the OOM can be better located.