Bounded fuzz baseline: 7515b79

This is Venom's first evidence-backed parser fuzz baseline. All five jobs passed in Scheduled Fuzzing run 30650839395 for commit 7515b79.

Target Executions Average exec/s New units Corpus files Peak RSS Crashes
HTTP 19,583,851 321,046 3,912 369 504 MiB 0
JSON 5,599,000 91,786 7,795 1,435 570 MiB 0
YAML 1,569,703 25,732 7,330 2,747 283 MiB 0
XML 3,919,957 64,261 5,389 778 553 MiB 0
Text / URL 1,828,203 29,970 8,253 3,067 564 MiB 0
Total 32,500,714 - 32,679 8,396 - 0

Each target received a 60-second budget; libFuzzer reported 61 seconds at the budget boundary. The wider 90-105 second workflow-step wall times include harness compilation and startup. Counts are per isolated target job and must not be interpreted as application throughput.

Environment and provenance

  • Runner: GitHub-hosted Ubuntu 24.04
  • Compiler: rustc 1.99.0-nightly (8ab9fdff5 2026-07-30)
  • Engine: cargo-fuzz with libFuzzer and AddressSanitizer
  • Machine-readable record: 7515b79.json
  • Raw libFuzzer logs and per-target summaries: workflow artifacts retained for 90 days; identifiers and digests are in the JSON record.

The preceding run 30650512500 failed while linking sanitizer coverage with workspace release LTO, before any target executed. Commit 7515b79 disables LTO and stripping only for fuzz harnesses; product release profiles are unchanged.

No crash or sanitizer finding was observed in this bounded run. That is a regression signal for these five harnesses at this exact commit, not proof that the parsers or Venom are safe.