Plans, runs, scores and reports fair comparisons of Redline with origin/main, with other Redline builds and with Fuzzilli. Covers the three measurements of Redline, which are bug finding on re-injected bugs, change-based fuzzing against normal mode, and…
Reproduces, heals (e.g., converting browser/HTML POCs to d8), minimizes, and classifies whether a V8 POC is a security vulnerability or a regular bug.
Guides the initial analysis and impact assessment of a V8 security report, strictly excluding implementation or fixing.
Enforces the [P0]-[P4] CL priority prefix workflow for V8 Gerrit CLs. Always use whenever drafting commit messages, running git commit, uploading CLs with git cl upload, or preparing CLs for submission (Commit-Queue+2, git cl set-commit, or git cl land). Do…
Builds d8 binaries that contain a known V8 bug again, to measure how Redline and Fuzzilli find bugs. Covers the choice of fixes, the training and held-out sets, the revert and the two d8 builds. Also covers the change-based fuzzing targets, the check of the…
Runs and checks the change-based fuzzing of Redline, which gives more fuzzing time to the code of one change. Covers the targets file, the status line, the weights and the focus. Lists the causes of a session without targets, or with targets that no run…
Explains how to write the text of experimental/redline in ASD-STE100 Simplified Technical English. The text includes the Markdown files, the code comments, the commit messages and the CL descriptions. Shows how to check the text with the linter of the…
Runs and reports the checks that a change to experimental/redline must pass before upload. The checks cover formatting, lint, the build, the selftests, the parser round trip and the golden traces. They also cover the Markdown links, the ASD-STE100 language of…