crypto: Implemented sharing room keys for past messages (MSC3061)

[Michael-Hollister]

Fixes #580

Overview

I have created an implementation for MSC 3061 and wanted to share the changes should the team want to merge into the SDK.

Note this follows the implementation suggested in the MSC where keys are forwarded upon invitation of a new user. Based on past discussion in #580 however, this approach was was not preferred due to some flaws. I thought I would still share this implementation however for the following reasons:

1. MSC 3061 has officially been merged into the spec per my understanding: MSC3061: Sharing room keys for past messages matrix-spec-proposals#3061
2. The proposed changes discussed in Implement MSC3061: Sharing room keys for past messages  #580 sounded to me that it requires a new MSC proposal for adding a new event type and improving the key sharing process.
3. This implementation provides a working solution that is equivalent to what is offered currently in the other Matrix SDKs, which can be used until the preferred solution is developed.

Any feedback welcome.

PR Dependencies:

• vodozemac: Added ExportedSessionKey conversion trait for SessionKey vodozemac#114
• ruma: Added shared_history flag to room key events ruma/ruma#1669

Testing:

Tested key sharing with verified and unverified devices with multiple clients including:

• FUTO Circles iOS / Android
• iamb
• Element / Cinny (interoperability testing since these don't use the crypto FFI)

Todo:

• Update Cargo.toml to point to upstream SDK instead of fork with the ruma/vodozemac changes
• Docs: Update room key sharing diagram indicating how HistoryVisibility is taken into account when sharing keys (https://raw.githubusercontent.com/matrix-org/matrix-rust-sdk/main/contrib/key-sharing-algorithm/model.png)
• Build and deploy a new crypto FFI release?
• Public API changes documented in changelogs (optional)

Signed-off-by: Michael Hollister michael@futo.org

[codecov]

Codecov Report

Attention: 57 lines in your changes are missing coverage. Please review.

Comparison is base (a433d51) 78.74% compared to head (ada3e82) 78.51%.
Report is 85 commits behind head on main.

❗ Current head ada3e82 differs from pull request most recent head 68fd193. Consider uploading reports for the commit 68fd193 to get more accurate results

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #2650      +/-   ##
==========================================
- Coverage   78.74%   78.51%   -0.23%     
==========================================
  Files         200      200              
  Lines       20360    20445      +85     
==========================================
+ Hits        16032    16053      +21     
- Misses       4328     4392      +64     

Files	Coverage Δ
...trix-sdk-crypto/src/olm/group_sessions/outbound.rs	93.61% <100.00%> (+0.10%)	⬆️
...tes/matrix-sdk-crypto/src/types/events/room_key.rs	63.88% <100.00%> (+1.03%)	⬆️
...es/matrix-sdk-crypto/src/types/events/to_device.rs	61.40% <ø> (ø)
...es/matrix-sdk-crypto/src/olm/group_sessions/mod.rs	74.35% <66.66%> (-0.65%)	⬇️
crates/matrix-sdk/src/room/mod.rs	82.79% <90.00%> (+0.14%)	⬆️
...-sdk-crypto/src/types/events/forwarded_room_key.rs	33.33% <0.00%> (-2.16%)	⬇️
...atrix-sdk-crypto/src/olm/group_sessions/inbound.rs	81.39% <60.00%> (-1.03%)	⬇️
crates/matrix-sdk-crypto/src/gossiping/machine.rs	63.17% <36.36%> (-2.10%)	⬇️
crates/matrix-sdk-crypto/src/machine.rs	78.59% <14.63%> (-4.86%)	⬇️

... and 3 files with indirect coverage changes

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[poljar]

To manage expectations a bit, I sadly won't have the time to take a proper look at this before the end of the month.

[Michael-Hollister]

To manage expectations a bit, I sadly won't have the time to take a proper look at this before the end of the month.

No problem, thanks for letting me know and taking the time to review this when ready.

[Michael-Hollister]

If this PR is of interest being reviewed/merged, let me know and I can fix conflicts and push a few fixes to issues that exist in this old version. Otherwise, my impression from the discussion in matrix-org/matrix-spec#1655 is that the PR will be not be merged, so I will be holding off updating until needed.