OSS-Fuzz revision
Switch to the exact oss-fuzz_sha recorded for the case.
Fuzzing build failure repair benchmark
A reproducible benchmark dataset for studying automated repair of OSS-Fuzz build failures.
cases
projects
dataset version
last generated
Each case records the metadata needed to initialize the OSS-Fuzz environment and reproduce the observed failing build, including the OSS-Fuzz revision, upstream revision, base image digest, fuzzing engine, sanitizer, architecture, and archived build log.
Failure categories follow the 25 fine-grained root causes reported in My Fuzzers Won't Build: An Empirical Study of Fuzzing Build Failures.
Computed from the current dataset file.
Fine-grained RC labels normalized before aggregation.
A case is reproducible only when the source tree and build environment are pinned together. Use the values in each record as immutable inputs before running the failing build.
Switch to the exact oss-fuzz_sha recorded for the case.
Clone software_repo_url, then check out the exact software_sha.
Inspect the project Dockerfile in OSS-Fuzz or upstream. Pin FROM to the recorded base_image_digest.
Update any Dockerfile git clone URL and checkout to match software_repo_url and software_sha.
Search by project, language, repository, SHA, error category, or build log. Links and hashes are shortened in the table and expanded in the details view.
| Project | Language | Failure Date | OSS-Fuzz SHA | Build Log | Upstream Repo | Engine | Sanitizer | Arch | Error Category |
|---|
Public records omit the repair outcome field and retain the reproducibility metadata needed for auditing and benchmark execution.