29 points by r9295 7 hours ago | 4 comments
mmastrac 6 hours ago
I've used AFL but never managed to create a consistent fuzzing process using it.
This _should_ allow me to add some more extensive fuzz tests than I've currently been using, even though the grammar is binary since we're really just "parsing" and "rendering" text that happens to be [u8] rather than str.
I don't think there's been any real discussion of the Postgres message format philosophy, but it's a very regular syntax: messages are composed of primitives and lists. Lists are either length-prefixed or zero-suffixed. Everything else is just a combination of these building blocks.
r9295 5 hours ago
vrm 5 hours ago
r9295 4 hours ago
https://github.com/oxidecomputer/typify may help for starters. Please create an issue if you need further help with integration!
trumpisafag 4 hours ago