Add Splicing (and Quiescence) wire message definitions

[optout21]

Add Splicing and Quiescence wire messages.

The messages are defined in the LN spec PR 863 (draft). Note that the spec PR is currently a draft, it needs a rebase, and does not reflect the latest state of discussions.

The following messages are added:

• stfu (2)
• splice (75* -- see notes below)
• splice_ack (76)
• splice_locked (77)

Fixes#2542 .

Note that current spec PR specifies type 74 for splice message, but that value is already used (by tx_abort), so value 75 is used as of now in this PR.

TODO:

• msg definitions
• msg encoding
• send&handle infrastructure
• Finalize type for splice msg
• fuzzing

[codecov-commenter]

Codecov Report

Attention: 129 lines in your changes are missing coverage. Please review.

Comparison is base (96e7d7a) 88.85% compared to head (649129d) 88.68%.

❗ Your organization needs to install the Codecov GitHub app to enable full functionality.

Additional details and impacted files

@@ Coverage Diff @@## main #2544 +/- ##
==========================================
- Coverage 88.85% 88.68% -0.17% 
==========================================
Files 113 113 Lines 89170 89344 +174 Branches 89170 89344 +174 ==========================================
+ Hits 79228 79238 +10 - Misses 7699 7859 +160 - Partials 2243 2247 +4 

Files	Coverage Δ
lightning/src/events/mod.rs	33.11% <ø> (ø)
lightning-net-tokio/src/lib.rs	73.94% <0.00%> (-1.69%)	⬇️
lightning/src/ln/functional_test_utils.rs	90.71% <0.00%> (-0.35%)	⬇️
lightning/src/ln/wire.rs	55.81% <0.00%> (-2.02%)	⬇️
lightning/src/util/test_utils.rs	76.18% <0.00%> (-1.02%)	⬇️
lightning/src/ln/channelmanager.rs	78.46% <0.00%> (-0.29%)	⬇️
lightning/src/ln/msgs.rs	76.83% <64.28%> (-0.40%)	⬇️
lightning/src/ln/peer_handler.rs	57.28% <0.00%> (-1.52%)	⬇️

... and 6 files with indirect coverage changes

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

[dunxen]

Thanks! Yeah, due to the drafty nature of the spec, there's no rush to merge this, but then again it can always be reworked as they're also not yet used in release.

Note that current spec PR specifies type 74 for splice message, but that value is already used (by tx_abort), so value 84 is used as of now in this PR.

Oh, really? Spec PR probably hasn't been updated in while. Thanks for tracking it. Then we'll definitely keep this draft until there's a clear type.

[optout21]

Did some changes to be in sync with CLN, as opposed to the outdated version of the spec (but this will need confirmation):

• type of splice msg is 75 (74 is taken by tx_abort; the 3 splicing msgs are thus 75, 76, 77)
• funding parameter is now relative, signed value, relative_satoshis
• reorder first two parameters channel_id and chain_hash

Note: definition in CLN https://github.com/ElementsProject/lightning/blob/master/wire/peer_wire.csv#L209

[optout21]

This is on hold, due to the spec not being updated. Above changes are in sync with the CLN implementation, but not the spec. Specifically, using 75 for splice message.
The spec PR has not been updated, but the changes are clearly mentioned as TODO in a comment by @ddustin.

Therefore I suggest to take this as a confirmation, and do not necessarily wait for spec update with this PR.

I will wrap up with adding fuzzing tests and propose for review.

[moneyball]

Out of curiosity what do Eclair/Phoenix do?

[optout21]

Out of curiosity what do Eclair/Phoenix do?

They are currently using nonstandard 3700x numerical types for splice messages (which means that their implementation is not compatible with others).
(https://github.com/ACINQ/eclair/blob/master/eclair-core/src/main/scala/fr/acinq/eclair/wire/protocol/LightningMessageCodecs.scala#L483)

[dunxen]

Nice. I think we can squash the fix ups and formatting commits now :)

[optout21]

Squashed and rebased

[dunxen]

LGTM.

[TheBlueMatt]

I'll trust @dunxen on the message encodings and contents.