This differential implements a method to decrypt data in NotificationsCryptoModule. Additionally some CryptoModule code is
simplified. I don't see the reason the ckeck for matching inbound sessions every time we receive a message. Additionally we would
have to store keyserver identity keys which would make the code more complicated.
Details
Details
After succesfull log-in execute in the keyserver code to encrypt some mock data. Then hardcode this data in
NotificationsService and call NotificationsCryptoModule::decrypt() method on it. Build the app again and send a notif to trigger
decryption code execution. Examine logs from the NSE.
Diff Detail
Diff Detail
- Repository
- rCOMM Comm
- Branch
- marcin/eng-3026
- Lint
No Lint Coverage - Unit
No Test Coverage
Event Timeline
Comment Actions
I don't see the reason the ckeck for matching inbound sessions every time we receive a message.
This makes sense to me, but it'd be good to get @anunay's perspective on this as well. Is it safe for us to skip calling ::olm_matches_inbound_session_from when decrypting an inbound session?
I think it makes sense to remove the check... it doesn't make sense that we'd be checking the identity keys of the caller on each message, partly because we have no "proof" that the message is directly associated with the identity keys (outside of the standard chain-of-trust stuff to the chain key, which is already checked when we call ::olm_decrypt), and partly because the chain-of-trust should be enough "proof" for us.
| native/cpp/CommonCpp/CryptoTools/CryptoModule.cpp | ||
|---|---|---|
| 354 | Should we delete matchesInboundSession from the codebase? It appears we have no other uses | |
| native/cpp/CommonCpp/CryptoTools/CryptoModule.cpp | ||
|---|---|---|
| 354 | Since we recently we agreed on not to treat CryptoModule as a library, we can delete this methods that it is not used anywhere. | |