Authenticate blinded path contexts with a secret AAD in the MAC #3845
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
`Poly1305::raw_result` copies the output into a slice, for some reason allowing any length sice. This isn't a great API, so while we're here we change it to return the 16-byte tag instead.
`ChaChaPolyReadAdapter` decodes an arbitrary object and checks the poly1305 tag. In the coming commits, we'll need a variant of this which allows for an *optional* AAD in the poly1305 tag, accepting either tag as valid, but indicating to the caller whether the AAD was used. We could use the actual AAD setup in poly1305, which puts the AAD first in the MAC (and then pads it out to a multiple of 16 bytes), but since we're gonna check both with and without, its nice to instead put the AAD at the end, enabling us to only calculate most of the hash once before cloning its state and adding the AAD block. We do this by swapping the AAD and the data being MAC'd in the AAD-containing MAC check (but leaving them where they belong for the non-AAD-containing MAC check).
|
👋 Hi! I see this is a draft PR. |
TheBlueMatt
force-pushed
the
2025-06-internally-auth-contexts
branch
from
921cacf to
21fab54
Compare
When we receive an onion message, we often want to make sure it was sent through a blinded path we constructed. This protects us from various deanonymization attacks where someone can send a message to every node on the network until they find us, effectively unwrapping the blinded path and identifying its recipient. We generally do so by adding authentication tags to our `MessageContext` variants. Because the contexts themselves are encrypted (and MAC'd) to us, we only have to ensure that they cannot be forged, which is trivially accomplished with a simple nonce and a MAC covering it. This logic has ended up being repeated in nearly all of our onion message handlers, and has gotten quite repetitive. Instead, here, we simply authenticate the blinded path contexts using the MAC that's already there, but tweaking it with an additional secret as the AAD in Poly1305. This prevents forgery as the secret is now required to make the MAC check pass. Ultimately this means that no one can ever build a blinded path which terminates at an LDK node that we'll accept, but over time we've come to recognize this as a useful property, rather than something to fight. Here we finally break from the spec fully in our context encryption (not just the contents thereof). This will save a bit of space in some of our `MessageContext`s, though sadly not in the blinded path we include in `Bolt12Offer`s, so they're generally not in space-sensitive blinded paths. We can apply the same logic in our blinded payment paths as well, but we do not do so here. This commit only adds the required changes to the cryptography, for now it uses a constant key of `[41; 32]`.
TheBlueMatt
force-pushed
the
2025-06-internally-auth-contexts
branch
from
21fab54 to
1d3985b
Compare
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
This is basically done, the only step left is to change the public API to take in a new key, rather than using the constant
[41; 32]. Someone with an LLM agent setup properly can probably make it do the work.