Commit Graph

68 Commits

Author SHA1 Message Date
Richard van der Hoff 5b4aedd4be Fix bug in verifying megolm event senders
1a03e534bd introduced a bug which mixed up the keys_proved and the
keys_claimed. Switch them around again so that megolm messages are correctly
tied back to the sending device.
2016-11-16 09:22:31 +00:00
Richard van der Hoff a5d857945a Retry decryption after receiving keys
m.room_keys may arrive after the messages themselves, so allow events to be
decrypted after the event (haha).
2016-11-14 15:13:02 +00:00
Richard van der Hoff 1a03e534bd Refactor decryption
Create the MatrixEvent wrapper before decryption, and then pass that into the
decryptors, which should update it.

Also remove the workaround that sends m.new_device messages when we get an
unknown session; it's just a bandaid which is obscuring more meaningful
problems.
2016-11-14 15:13:02 +00:00
Richard van der Hoff e623b539c4 persist DecryptionAlgorithm instances
It's useful to be able to keep state between events in the DecryptionAlgorithm,
so store them in a map.
2016-11-14 15:13:02 +00:00
Mark Haines bfffbea4a0 Merge remote-tracking branch 'origin/develop' into markjh/encrypted_voip 2016-11-11 10:20:40 +00:00
Mark Haines 777ef83378 Merge remote-tracking branch 'origin/develop' into markjh/encrypted_voip 2016-11-10 19:44:42 +00:00
Mark Haines 24283dcbd5 Encrypt all events, including 'm.call.*' 2016-11-10 19:42:16 +00:00
Richard van der Hoff 2113c83679 Ignore reshares of known megolm sessions
If we get a second key for a known megolm session, ignore it.

Fixes https://github.com/vector-im/vector-web/issues/2326, one hopes.
2016-11-10 19:28:08 +00:00
Richard van der Hoff e173d822e8 Log to the console on unknown session
This might help diagnose Erik/Matthew's comms breakdown.
2016-11-07 18:57:09 +00:00
Richard van der Hoff a4f192bc88 Sign one-time keys, and verify their signatures
We have decided that signing one-time keys is the lesser of two evils;
accordingly, use a new key algorithm type (`signed_curve25519`), sign the
one-time keys that we upload to the server, and verify the signatures on those
we download.

This will mean that develop won't be able to talk to master, but hey, we're in
beta.
2016-10-21 12:24:19 +01:00
Mark Haines 16b4865035 Merge pull request #241 from matrix-org/markjh/check_for_duplicate_message_ids
Check for duplicate message indexes for group messages
2016-10-21 09:55:38 +01:00
Mark Haines 20b310484b Document the format of the keys 2016-10-21 09:54:57 +01:00
Mark Haines 8b856b9d15 Wrap the longer lines 2016-10-20 18:02:48 +01:00
Mark Haines 3f7df0d15c Fiddle linebreaks 2016-10-20 17:59:15 +01:00
Mark Haines e0917d3c47 Check for duplicate message indexes for group messages 2016-10-20 17:49:37 +01:00
Richard van der Hoff 19c257703c Rotate megolm sessions
In order to mitigate backward-secrecy concerns, make sure that we rotate the
outbound megolm session at regular intervals (every week/100 msgs by default).
2016-10-20 15:42:06 +01:00
Richard van der Hoff 7a7f345f28 Merge pull request #239 from matrix-org/rav/fix_unknown_key
Check recipient and sender in Olm messages
2016-10-19 11:44:48 +01:00
Richard van der Hoff b5c7c700d5 Check recipient and sender in Olm messages
Embed the sender, recipient, and recipient keys in the plaintext of Olm
messages, and check those fields on receipt.

Fixes https://github.com/vector-im/vector-web/issues/2483
2016-10-19 11:24:59 +01:00
Richard van der Hoff aafb1ffdef Consistency checks for E2E device downloads
Check that the user_id and device_id in device query responses match those that
we expect.

This resolves an unknown-key attack whereby Eve can re-sign Bob's keys with her
own key, thus getting Alice to send her messages which she can then forward to
Bob, making Bob think that Alice sent the messages to him.
2016-10-18 13:40:13 +01:00
Richard van der Hoff faff057592 crypto: remove duplicate code
Only call SessionStore.storeEndToEndDevicesForUser once per user, rather than
once per device.

(Probably also fixes a bug where, when a user removes all devices, the store
isn't updated)
2016-09-30 09:17:54 +01:00
Richard van der Hoff 1a3ee28d01 Log when we get an oh_hai message 2016-09-21 17:07:40 +01:00
Richard van der Hoff 669aecf4e6 E2E: Fix NPE in getEventSenderDeviceInfo 2016-09-21 15:05:27 +01:00
Richard van der Hoff 832559926f Fix the ed25519 key checking
Finish plumbing in the Ed25519 key checks. Make sure we store the claimed key
correctly in the megolm sessions, and keep them as a separate field in
MatrixEvent rather than stuffing them into _clearEvent
2016-09-20 20:42:08 +01:00
Richard van der Hoff 59411353b1 Add 'keys' to *all* olm messages
(including ones which just carry megolm keys)
2016-09-20 20:39:40 +01:00
Richard van der Hoff 83bd420cd5 Return null from decryptEvent if session is unknown
This just makes the shape of the API a bit saner.
2016-09-20 20:39:40 +01:00
Richard van der Hoff 78a0aa5d47 Add MatrixClient.getEventSenderDeviceInfo()
- a function to get information about the device which sent an event
2016-09-20 20:39:16 +01:00
Richard van der Hoff 6e31319294 Handle lack of one-time keys better
If a device had run out of one-time keys, we would send it an empty to_device
event, which it would then fail to decrypt with a "Not included in
recipients", which is all a bit pointless.
2016-09-18 22:58:35 +01:00
Richard van der Hoff cd0b19f93f Crypto: improve console logs
Attempt to make the console logs more helpful by reducing noise and adding
helpful debug info.
2016-09-18 21:55:38 +01:00
Richard van der Hoff 4f22610499 Megolm: clarify jsdoc
Clarify somewhat misleading jsdoc text
2016-09-18 13:52:48 +01:00
Matthew Hodgson 9e57a9352a Merge pull request #212 from matrix-org/rav/get_devicelist_on_join
Pull user device list on join
2016-09-17 19:21:17 +01:00
Richard van der Hoff f2e10e030d Unknown sessions: send oh_hai to all devices if device_id is unknown 2016-09-17 19:07:03 +01:00
Richard van der Hoff 266b7afc72 Fix sending of oh_hais on bad sessions
Fix a bunch of bugs in the code which tried to send an oh_hai message when we
got a message with an unknown megolm session.
2016-09-17 18:30:12 +01:00
Richard van der Hoff a15dffbb3a Pull user device list on join
When a new user joins a room, make sure we download their device list if we
don't already have it.

This should fix at least one cause of
https://github.com/vector-im/vector-web/issues/2249.
2016-09-17 17:44:15 +01:00
Matthew Hodgson a30c816cb6 typo 2016-09-17 15:44:55 +01:00
Matthew Hodgson 425f862cf8 Merge pull request #205 from matrix-org/markjh/megolm
Update the olm library version to 1.3.0
2016-09-16 17:30:26 +01:00
Mark Haines 5d6256bede Merge pull request #209 from matrix-org/markjh/comment_upload_key
Comment what the logic in uploadKeys does
2016-09-16 16:34:37 +01:00
Mark Haines ff5b923e6f Spelling: s/cliamed/claimed/ 2016-09-16 16:31:00 +01:00
Mark Haines af7a9a68b8 Merge pull request #210 from matrix-org/markjh/echo_keys_proved
Include keysProved and keysClaimed in the local echo for events we send.
2016-09-16 15:45:22 +01:00
Mark Haines 3bc56cf3f8 More comments on the local echo 2016-09-16 15:36:56 +01:00
Mark Haines c2a40572a5 Include keysProved and keysClaimed in the local echo for events we send. 2016-09-16 15:30:22 +01:00
Mark Haines ee7d4d0521 Explain what happens to the old keys in olm 2016-09-16 14:43:22 +01:00
Mark Haines 6ab410ef6a Comment what the logic in uploadKeys does 2016-09-16 14:38:26 +01:00
Mark Haines 460f20a4ce Merge pull request #207 from matrix-org/markjh/variable_scoping
Reset oneTimeKey to null on each loop iteration.
2016-09-16 11:23:58 +01:00
Mark Haines 9a98c3991a Reset onTimeKey to null on each loop iteration.
Otherwise we will use a value from a previous iteration of the loop.
2016-09-16 10:44:25 +01:00
Matthew Hodgson 2765720b76 unbreak NPE where megolm's decryptEvent doesn't return a result 2016-09-15 20:09:41 +01:00
Mark Haines 71f23ffce1 Merge branch 'develop' into markjh/megolm
Conflicts:
	lib/crypto/algorithms/megolm.js
2016-09-15 17:10:02 +01:00
Mark Haines 0d5d74674e Remove spurious senderKey argument 2016-09-15 16:46:28 +01:00
Mark Haines bde6a171f6 Add getKeysProved and getKeysClaimed methods to MatrixEvent.
These list the keys that sender of the event must have ownership
of and the keys of that the sender claims ownership of.

All olm and megolm messages prove ownership of a curve25519 key.
All new olm and megolm message will now claim ownership of a
ed25519 key.

This allows us to detect if an attacker claims ownership of a curve25519
key they don't own when advertising their device keys, because when we
receive an event from the original user it will have a different ed25519 key
to the attackers.
2016-09-15 16:26:43 +01:00
Mark Haines 2fbef8638f Fix grammar 2016-09-15 14:43:23 +01:00
Mark Haines 355b728a57 Remove unnecessary semicolon; 2016-09-15 14:23:30 +01:00