[anchor] pluggable anchor commitments #3821

halseth · 2019-12-11T14:53:43Z

Missing from end-to-end anchors working:

Commitment resolution for delayed to_remote output
Negotiating anchor type during funding flow

TODO:

Unit tests for anchor specific changes
One itest case using anchors: Pluggable anchors select itests #4001

Long term:

Automatically sweep anchors
whole itest package run using anchors
watchtower support
channel backups

Builds on #3829

input/script_utils.go

lnwallet/commitment.go

halseth · 2019-12-11T15:04:06Z

lnwallet/commitment.go

+		theirBalance -= commitFeeMSat
+	}
+
+	// If the commitment has anchors, we must also subtract the anchor


This is a bit awkward, but is a result of that the balances we store in the db has had their commit fee subtracted. should we do a db migration to change this?

You want instead to store the balance on disk with the anchor output size already subtracted?

joostjager · 2019-12-11T16:01:23Z

lnwallet/commitment.go

+// and commitment point. The keys are derived differently depending on the type
+// of channel, and whether the commitment transaction is ours or the remote
+// peer's.
+func DeriveCommitmentKeys(commitPoint *btcec.PublicKey,


Separate move from change in two commits, for easier review.

joostjager

This look pretty good. The word anchor disappeared completely from lnwallet.Channel. I also see that there may be more to encapsulate, but the commitment building is probably the most important thing.

I still have that ideal in mind where there is a CommitmentBuilder interface and a separate instance for the anchor commit format. Those instances would be thin and build on a lot of shared building blocks. It is probably too much for now and this is already a good step in that direction.

lnwallet/commitment.go

cfromknecht

Quick skim, cool pr :)

input/script_utils.go

input/size.go

lnwallet/reservation.go

joostjager

Some drive-by comments

lnwallet/commitment.go

joostjager · 2020-01-07T13:30:19Z

lnwallet/commitment.go

@@ -229,6 +229,18 @@ func CommitScriptToRemote(chanType channeldb.ChannelType, csvTimeout uint32,
 	}, nil
 }

+func InitiatorFee(chanType channeldb.ChannelType, commitWeight int64,


Would there be a better term for miner_fee + anchors then "fee"? To prevent introducing a second definition of fee. Cost, overhead, ..?

I like cost perhaps?

joostjager · 2020-01-07T13:32:14Z

lnwallet/channel.go

+		}
+
+		// Not found.
+		return nil


Could make this stricter and return an error if the channel type says there must be anchors.

lnwallet/channel.go

fundingmanager.go

lnwallet/commitment.go

input/script_utils.go

Roasbeef

Solid PR! We're soo close 🤗

The one thing that seems to be missing from my run through is ensuring that we re-account for the value that we deducted for anchors each time we go to create a new commitment.

lnwallet/commitment.go

Roasbeef · 2020-01-17T02:31:48Z

lnwallet/commitment.go

+		remoteBalance = theirBalance.ToSatoshis()
+	)
+	if cb.chanState.ChanType.HasAnchors() {
+		anchors := 2 * anchorSize


totalAnchorValue?

Also could very well just be lifted into a constant of its own.

Roasbeef · 2020-01-17T02:32:21Z

lnwallet/commitment.go

+		theirBalance -= commitFeeMSat
+	}
+
+	// If the commitment has anchors, we must also subtract the anchor


You want instead to store the balance on disk with the anchor output size already subtracted?

lnwallet/channel.go

Roasbeef · 2020-01-17T02:48:38Z

input/script_utils_test.go

+	// genSweepTx generates a swepp of the senderCommitTx, and sets the
+	// sequence if locktime is true.
+	genSweepTx := func(locktime bool) {
+		prevOut := &wire.OutPoint{


A worthy follow up PR would be also testing second-level HTLC aggregation.

input/script_utils.go

fundingmanager.go

lnwallet/reservation.go

Roasbeef · 2020-01-17T02:58:35Z

lnwallet/reservation.go

@@ -213,6 +219,16 @@ func NewChannelReservation(capacity, localFundingAmt btcutil.Amount,
 		)
 	}

+	// Similarly we ensure their balance is reasonable if we are not the
+	// initiator.
+	if !initiator && theirBalance.ToSatoshis() <= 2*DefaultDustLimit() {


Where does this requirement come from?

It is not a new requirement, it was just something I noticed wasn't checked. Made it a separate commit.

Why 2 times the dust limit?

From 10 lines above

Isn't 10 lines above 2 times the anchor size?

a4841f3#diff-36eb42cbbf10393bf0c4e50d9b280ae4R208

this is likely wrong for LTC, but so are most of these checks...

With this commitment type, we'll add extra anchor outputs to the commitment transaction if the anchor channel type is active.

…witness script We use the fact that we can tell whether the commit is local or remote by inspecting the witness script. We cannot use the maturity delay anymore, as we can have delayed to_remote outputs also now. Co-authored-by: Joost Jager <joost.jager@gmail.com>

… type Based on the channel type, the commitment weight will be calculated.

If we are the initiator, we check that our starting balance after subtracting fees are not less than two times the default dust limit. This commit adds a similar check for the non-initiator case, checking that the remote party has a starting balance of reasonable size.

To prepare for adding more commit types to test for basic channel funding, we make the commit type an enum that gets its own set of subtests.

Since we are also going to use it for experimental new features.

Defaults to disabled.

If both nodes are signalling the feature, make all opened channels using this type.

Roasbeef

LGTM 📯

Excellent work @halseth and @joostjager!

We should also extend the integration tests to cover breach retribution for all the existing commitment types. This can be done as a follow up PR. I'll also put together a small follow up PR to enable SCB support as well.

Roasbeef · 2020-03-10T02:32:15Z

lnwire/features.go

+	// AnchorsRequired is a required feature bit that signals that the node
+	// requires channels to be made using commitments having anchor
+	// outputs.
+	AnchorsRequired FeatureBit = 1336


Rather it should be resolved before we cut a release with this included. Feature bits are easy to change after the fact.

Roasbeef · 2020-03-10T02:33:03Z

input/size.go

+	// HtlcConfirmedScriptOverhead is the extra length of an HTLC script
+	// that requires confirmation before it can be spent. These extra bytes
+	// is a result of the extra CSV check.
+	HtlcConfirmedScriptOverhead = 3


halseth commented Dec 11, 2019

View reviewed changes

input/script_utils.go Outdated Show resolved Hide resolved

halseth commented Dec 11, 2019

View reviewed changes

lnwallet/commitment.go Show resolved Hide resolved

halseth commented Dec 11, 2019

View reviewed changes

joostjager reviewed Dec 11, 2019

View reviewed changes

lnwallet/commitment.go Show resolved Hide resolved

cfromknecht reviewed Dec 12, 2019

View reviewed changes

halseth force-pushed the pluggable-anchors-lnwallet branch from 5f89aad to 819a5dd Compare December 12, 2019 13:04

halseth added the anchors label Dec 13, 2019

halseth mentioned this pull request Dec 13, 2019

[anchors] Pluggable anchor commitments #3798

Closed

halseth force-pushed the pluggable-anchors-lnwallet branch 2 times, most recently from b55825f to 0bb5f08 Compare December 13, 2019 10:58

halseth force-pushed the pluggable-anchors-lnwallet branch from 0bb5f08 to 62fe01f Compare January 7, 2020 12:18

joostjager reviewed Jan 7, 2020

View reviewed changes

halseth commented Jan 7, 2020

View reviewed changes

fundingmanager.go Outdated Show resolved Hide resolved

joostjager reviewed Jan 7, 2020

View reviewed changes

lnwallet/commitment.go Outdated Show resolved Hide resolved

halseth commented Jan 7, 2020

View reviewed changes

input/script_utils.go Outdated Show resolved Hide resolved

Roasbeef added this to the 0.10.0 milestone Jan 14, 2020

Roasbeef added the v0.10 label Jan 14, 2020

halseth force-pushed the pluggable-anchors-lnwallet branch 3 times, most recently from 3661f7e to d17272b Compare January 16, 2020 13:12

halseth marked this pull request as ready for review January 16, 2020 13:28

halseth requested a review from Roasbeef as a code owner January 16, 2020 13:28

halseth mentioned this pull request Jan 16, 2020

[anchors] anchors resolutions #3920

Closed

Roasbeef requested changes Jan 17, 2020

View reviewed changes

halseth force-pushed the pluggable-anchors-lnwallet branch 2 times, most recently from c472925 to fa13a63 Compare January 23, 2020 10:49

joostjager mentioned this pull request Jan 24, 2020

cnct: anchor output sweeping and fee bumping #3758

Merged

halseth and others added 23 commits March 9, 2020 12:10

input/script_utils test: add test cases for delayed HTLC receiver scrpts

a309132

input/script_utils: add delayed to_remote script + tests

990992c

input: add CommitmentToRemoteConfirmed witness type

6deb913

input/script_utils: add anchor scripts + tests

50199ae

lnwallet: add anchor commitmenttype

af68ff1

With this commitment type, we'll add extra anchor outputs to the commitment transaction if the anchor channel type is active.

input+lnwallet: use individual commit weight calculations for channel…

ea94dbb

… type Based on the channel type, the commitment weight will be calculated.

lnwallet/channel: add feerate sanity check

21c5a95

lnwallet: choose HTLC scripts based on channel type

6810912

input/test_utils: make mockSigner use SigHashType from sign descriptor

d1089fb

lnwallet: make second level sigs using sighash single|anyonecanpay

bddd3e1

lnwallet: set 2nd level sequence according to channel type

c5d58b4

lnwallet+nursery+input: set sequence=1 for direct HTLC spends

92af234

lnwallet+breacharbiter: record local csv delay

ad8e9f3

lnwallet+funding: create CommitmentType enum

f95a82b

itest: make commit type enum

51c5352

To prepare for adding more commit types to test for basic channel funding, we make the commit type an enum that gets its own set of subtests.

cfg: rename legacyprotocol to protocol

44756b5

Since we are also going to use it for experimental new features.

multi: optionally enable and signal anchor support

21126ab

Defaults to disabled.

config+link: disable watchtower for anchors

fd93c56

chanbackup: disable channel backup for anchor types

7adb1bc

fundingmanager+lnwallet: enable anchor commitments

ea2a58e

If both nodes are signalling the feature, make all opened channels using this type.

lnwallet+size: select HTLC fees based on channel type

b7885db

halseth force-pushed the pluggable-anchors-lnwallet branch from 00aa935 to b7885db Compare March 9, 2020 12:01

Roasbeef approved these changes Mar 10, 2020

View reviewed changes

Roasbeef merged commit 3dda93e into lightningnetwork:master Mar 10, 2020

cfromknecht mentioned this pull request Apr 21, 2020

lnwallet: absolute fee floor sanity check #4211

Merged

STAWKEYE mentioned this pull request Jun 29, 2020

LND nodes overpay quite significantly on mutual closing TX's #4413

Open

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

[anchor] pluggable anchor commitments #3821

[anchor] pluggable anchor commitments #3821

halseth commented Dec 11, 2019 •

edited

Loading

halseth Dec 11, 2019

Roasbeef Jan 17, 2020

joostjager Dec 11, 2019

joostjager left a comment

cfromknecht left a comment

joostjager left a comment

joostjager Jan 7, 2020

halseth Jan 16, 2020

joostjager Jan 7, 2020

Roasbeef left a comment

Roasbeef Jan 17, 2020

Roasbeef Jan 17, 2020

Roasbeef Jan 17, 2020

Roasbeef Jan 17, 2020

Roasbeef Jan 17, 2020

halseth Jan 24, 2020

joostjager Feb 14, 2020

halseth Feb 18, 2020

joostjager Feb 24, 2020

halseth Feb 25, 2020 •

edited

Loading

cfromknecht Feb 25, 2020

Roasbeef left a comment

Roasbeef Mar 10, 2020

Roasbeef Mar 10, 2020

[anchor] pluggable anchor commitments #3821

[anchor] pluggable anchor commitments #3821

Conversation

halseth commented Dec 11, 2019 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

joostjager left a comment

Choose a reason for hiding this comment

cfromknecht left a comment

Choose a reason for hiding this comment

joostjager left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Roasbeef left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

halseth Feb 25, 2020 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Roasbeef left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

halseth commented Dec 11, 2019 •

edited

Loading

halseth Feb 25, 2020 •

edited

Loading