Proposal 220 suggests that we pin RSA and Ed25519 identity keys to one another authority-side. Roger suggested to me that we also consider doing client-side identity pinning.
To upload designs, you'll need to enable LFS and have an admin enable hashed storage. More information
Child items
0
Show closed items
No child items are currently assigned. Use child items to break down this issue into smaller parts.
Linked items
0
Link issues together to show that they're related.
Learn more.
I started writing up a proposal draft here, but I'm not currently seeing the point of it. If a client has a correct consensus, it should get the correct RSA1024<->Ed25519 mappings unless the authorities are lying. But if the authorities are lying, they can poison the clients in lots of other ways too.
Similarly, for stuff like bridges, we can export the ed25519 key in the bridge line, and we don't need to remember the RSA1024 key at all. That's probably a better idea than pinning in the first place, right?
For guards, we should remember every public key we've seen for the guard, and only connect if all the keys are good.
So, what's the value here? What's the threat model it helps for?