Hi,
I started working on WPA3 decryption support. Some parts of it has already been merged. While working on it I've noticed a couple of things that I'd like to bring up here before spending too much time.
For WPA3 enterprise support keys and mic are no longer a fixed size. This will have quite big impact on the dot11crypt code as there are plenty of decisions taken based on fixed offsets into data frames. This now must be dynamically calculated based on AKM (authentication and key management) and cipher suite selected for current connection.
To determine AKM & cipher suite today the dot11crypt engine searches through 4-way handshake EAPOL frames. Unfortunately even the format of EAPOL frames (MIC length) is based on current AKMS. A "chicken or the egg" kind of problem that can be solved by also tracking association req/resp frames. These frames are at the moment ignored by dot11crypt. The IEEE 802.11 dissector however already dissect both assoc and EAPOL frames perfectly fine so should be possible to reuse.
The dot11crypt engine duplicate quite a lot IEEE 802.11 dissector functionality but now also lack certain parts. Instead of adding further duplication I'd like to propose the following changes:
- Replace the scan for keys functionality from dot11decrypt engine with a new SetKey(from, to, key_index, key) function that the IEEE802.11 dissector can call when a key is found.
Removing scan for keys means the following can be removed in dot11decrypt:
- Check done on each and every 802.11 frame for certain type/format
- 4-way handshake frame parsing
- TDLS frame parsing
- EAP keydata parsing
- RSN tag parsing
When this is done next steps to support the new key derivation and decryption methods needed for enterprise decryption should be easier.
Any objections, comments? If anyone else is doing larger work in this area maybe we can sync the activities.
Best regards,
Mikael?
