Use official splice messages

We replace our experimental version of `splice_init`, `splice_ack` and
`splice_locked` by their official version. If our peer is using the
experimental feature bit, we convert our outgoing messages to use the
experimental encoding and incoming messages to the official messages.

We also change the TLV fields added to `tx_add_input`, `tx_signatures`
and `splice_locked` to match the spec version. We always write both the
official and experimental TLV to updated nodes (because the experimental
one is odd and will be ignored) but we drop the official TLV if our
peer is using the experimental feature, because it won't understand the
even TLV field.

We do the same thing for the `commit_sig` TLV. For peers who support the
official splicing version, we insert the `start_batch` message before
the batch of `commit_sig` messages.

This guarantees backwards-compatibility with peers who only support the
experimental feature.

Author: t-bast

docs/release-notes/eclair-vnext.md

@@ -6,6 +6,39 @@
 
 <insert changes>
 
+### Channel Splicing
+
+With this release, we add support for the final version of [splicing](https://github.com/lightning/bolts/pull/1160) that was recently added to the BOLTs.
+Splicing allows node operators to change the size of their existing channels, which makes it easier and more efficient to allocate liquidity where it is most needed.
+Most node operators can now have a single channel with each of their peer, which costs less on-chain fees and resources, and makes path-finding easier.
+
+The size of an existing channel can be increased with the `splicein` API:
+
+```sh
+eclair-cli splicein --channelId=<channel_id> --amountIn=<amount_satoshis>
+```
+
+Once that transaction confirms, the additional liquidity can be used to send outgoing payments.
+If the transaction doesn't confirm, the node operator can speed up confirmation with the `rbfsplice` API:
+
+```sh
+eclair-cli rbfsplice --channelId=<channel_id> --targetFeerateSatByte=<feerate_satoshis_per_byte> --fundingFeeBudgetSatoshis=<maximum_on_chain_fee_satoshis>
+```
+
+If the node operator wants to reduce the size of a channel, or send some of the channel funds to an on-chain address, they can use the `spliceout` API:
+
+```sh
+eclair-cli spliceout --channelId=<channel_id> --amountOut=<amount_satoshis> --scriptPubKey=<on_chain_address>
+```
+
+That operation can also be RBF-ed with the `rbfsplice` API to speed up confirmation if necessary.
+
+Note that when 0-conf is used for the channel, it is not possible to RBF splice transactions.
+Node operators should instead create a new splice transaction (with `splicein` or `spliceout`) to CPFP the previous transaction.
+
+Note that eclair had already introduced support for a splicing prototype in v0.9.0, which helped improve the BOLT proposal.
+We're removing support for the previous splicing prototype feature: users that depended on this protocol must upgrade to create official splice transactions.
+
 ### Package relay
 
 With Bitcoin Core 28.1, eclair starts relying on the `submitpackage` RPC during channel force-close.

eclair-core/src/main/resources/reference.conf

@@ -88,6 +88,7 @@ eclair {
     option_zeroconf = disabled
     keysend = disabled
     option_simple_close=optional
+    option_splice = optional
     trampoline_payment_prototype = disabled
     async_payment_prototype = disabled
     on_the_fly_funding = disabled
@@ -109,7 +110,7 @@ eclair {
     funding {
       // Each RBF attempt adds more data that we need to store and process, so we want to limit our peers to a reasonable use of RBF.
       remote-rbf-limits {
-        max-attempts = 5 // maximum number of RBF attempts our peer is allowed to make
+        max-attempts = 10 // maximum number of RBF attempts our peer is allowed to make
         attempt-delta-blocks = 6 // minimum number of blocks between RBF attempts
       }
       // Duration after which we abort a channel creation. If our peer seems unresponsive and doesn't complete the

eclair-core/src/main/scala/fr/acinq/eclair/Features.scala

@@ -264,8 +264,7 @@ object Features {
     val mandatory = 28
   }
 
-  // TODO: this should also extend NodeFeature once the spec is finalized
-  case object Quiescence extends Feature with InitFeature {
+  case object Quiescence extends Feature with InitFeature with NodeFeature {
     val rfcName = "option_quiesce"
     val mandatory = 34
   }
@@ -314,6 +313,11 @@ object Features {
     val mandatory = 60
   }
 
+  case object Splicing extends Feature with InitFeature with NodeFeature {
+    val rfcName = "option_splice"
+    val mandatory = 62
+  }
+
   /** This feature bit indicates that the node is a mobile wallet that can be woken up via push notifications. */
   case object WakeUpNotificationClient extends Feature with InitFeature {
     val rfcName = "wake_up_notification_client"
@@ -337,12 +341,6 @@ object Features {
     val mandatory = 152
   }
 
-  // TODO: @pm47 custom splices implementation for phoenix, to be replaced once splices is spec-ed (currently reserved here: https://github.com/lightning/bolts/issues/605)
-  case object SplicePrototype extends Feature with InitFeature {
-    val rfcName = "splice_prototype"
-    val mandatory = 154
-  }
-
   /**
    * Activate this feature to provide on-the-fly funding to remote nodes, as specified in bLIP 36: https://github.com/lightning/blips/blob/master/blip-0036.md.
    * TODO: add NodeFeature once bLIP is merged.
@@ -386,10 +384,10 @@ object Features {
     ZeroConf,
     KeySend,
     SimpleClose,
+    Splicing,
     WakeUpNotificationClient,
     TrampolinePaymentPrototype,
     AsyncPaymentPrototype,
-    SplicePrototype,
     OnTheFlyFunding,
     FundingFeeCredit
   )
@@ -406,7 +404,6 @@ object Features {
     KeySend -> (VariableLengthOnion :: Nil),
     SimpleClose -> (ShutdownAnySegwit :: Nil),
     AsyncPaymentPrototype -> (TrampolinePaymentPrototype :: Nil),
-    OnTheFlyFunding -> (SplicePrototype :: Nil),
     FundingFeeCredit -> (OnTheFlyFunding :: Nil)
   )
 

eclair-core/src/main/scala/fr/acinq/eclair/channel/Commitments.scala

@@ -231,7 +231,7 @@ case class RemoteCommit(index: Long, spec: CommitmentSpec, txid: TxId, remotePer
     params.commitmentFormat match {
       case _: SegwitV0CommitmentFormat =>
         val sig = remoteCommitTx.sign(fundingKey, remoteFundingPubKey).sig
-        CommitSig(params.channelId, sig, htlcSigs.toList)
+        CommitSig(params.channelId, sig, htlcSigs.toList, TlvStream(CommitSigTlv.FundingTx(commitInput.outPoint.txid)))
       case _: SimpleTaprootChannelCommitmentFormat => ???
     }
   }
@@ -658,7 +658,8 @@ case class Commitment(fundingTxIndex: Long,
     Metrics.recordHtlcsInFlight(spec, remoteCommit.spec)
 
     val tlvs = Set(
-      if (batchSize > 1) Some(CommitSigTlv.BatchTlv(batchSize)) else None
+      Some(CommitSigTlv.FundingTx(fundingTxId)),
+      if (batchSize > 1) Some(CommitSigTlv.ExperimentalBatchTlv(batchSize)) else None,
     ).flatten[CommitSigTlv]
     val commitSig = params.commitmentFormat match {
       case _: SegwitV0CommitmentFormat =>
@@ -1042,8 +1043,10 @@ case class Commitments(params: ChannelParams,
       case commitSig: CommitSig => Seq(commitSig)
     }
     val commitKeys = LocalCommitmentKeys(params, channelKeys, localCommitIndex + 1)
-    // Signatures are sent in order (most recent first), calling `zip` will drop trailing sigs that are for deactivated/pruned commitments.
-    val active1 = active.zip(sigs).map { case (commitment, commit) =>
+    val active1 = active.zipWithIndex.map { case (commitment, idx) =>
+      // If the funding_txid isn't provided, we assume that signatures are sent in order (most recent first).
+      // This matches the behavior of peers who only support the experimental version of splicing.
+      val commit = sigs.find(_.fundingTxId_opt.contains(commitment.fundingTxId)).getOrElse(sigs(idx))
       commitment.receiveCommit(params, channelKeys, commitKeys, changes, commit) match {
         case Left(f) => return Left(f)
         case Right(commitment1) => commitment1

eclair-core/src/main/scala/fr/acinq/eclair/channel/fsm/Channel.scala

@@ -883,7 +883,7 @@ class Channel(val nodeParams: NodeParams, val channelKeys: ChannelKeys, val wall
       }
 
     case Event(cmd: CMD_SPLICE, d: DATA_NORMAL) =>
-      if (!d.commitments.params.remoteParams.initFeatures.hasFeature(Features.SplicePrototype)) {
+      if (!d.commitments.params.remoteParams.initFeatures.hasFeature(Features.Splicing)) {
         log.warning("cannot initiate splice, peer doesn't support splicing")
         cmd.replyTo ! RES_FAILURE(cmd, CommandUnavailableInThisState(d.channelId, "splice", NORMAL))
         stay()
@@ -2327,7 +2327,8 @@ class Channel(val nodeParams: NodeParams, val channelKeys: ChannelKeys, val wall
         }
         case _ => Set.empty
       }
-      val lastFundingLockedTlvs: Set[ChannelReestablishTlv] = if (d.commitments.params.remoteParams.initFeatures.hasFeature(Features.SplicePrototype)) {
+      val remoteFeatures = d.commitments.params.remoteParams.initFeatures
+      val lastFundingLockedTlvs: Set[ChannelReestablishTlv] = if (remoteFeatures.hasFeature(Features.Splicing) || remoteFeatures.unknown.contains(UnknownFeature(154)) || remoteFeatures.unknown.contains(UnknownFeature(155))) {
         d.commitments.lastLocalLocked_opt.map(c => ChannelReestablishTlv.MyCurrentFundingLockedTlv(c.fundingTxId)).toSet ++
           d.commitments.lastRemoteLocked_opt.map(c => ChannelReestablishTlv.YourLastFundingLockedTlv(c.fundingTxId)).toSet
       } else Set.empty
@@ -2456,7 +2457,8 @@ class Channel(val nodeParams: NodeParams, val channelKeys: ChannelKeys, val wall
             // We only send channel_ready for initial funding transactions.
             case Some(c) if c.fundingTxIndex != 0 => ()
             case Some(c) =>
-              val remoteSpliceSupport = d.commitments.params.remoteParams.initFeatures.hasFeature(Features.SplicePrototype)
+              val remoteFeatures = d.commitments.params.remoteParams.initFeatures
+              val remoteSpliceSupport = remoteFeatures.hasFeature(Features.Splicing) || remoteFeatures.unknown.contains(UnknownFeature(154)) || remoteFeatures.unknown.contains(UnknownFeature(155))
               // If our peer has not received our channel_ready, we retransmit it.
               val notReceivedByRemote = remoteSpliceSupport && channelReestablish.yourLastFundingLocked_opt.isEmpty
               // If next_local_commitment_number is 1 in both the channel_reestablish it sent and received, then the node

eclair-core/src/main/scala/fr/acinq/eclair/channel/fund/InteractiveTxBuilder.scala

@@ -855,7 +855,7 @@ private class InteractiveTxBuilder(replyTo: ActorRef[InteractiveTxBuilder.Respon
           case _: SegwitV0CommitmentFormat =>
             val localSigOfRemoteTx = remoteCommitTx.sign(localFundingKey, fundingParams.remoteFundingPubKey).sig
             val htlcSignatures = sortedHtlcTxs.map(_.sign(remoteCommitmentKeys, channelParams.commitmentFormat)).toList
-            val localCommitSig = CommitSig(fundingParams.channelId, localSigOfRemoteTx, htlcSignatures)
+            val localCommitSig = CommitSig(fundingParams.channelId, localSigOfRemoteTx, htlcSignatures, TlvStream(CommitSigTlv.FundingTx(fundingTx.txid)))
             val localCommit = UnsignedLocalCommit(purpose.localCommitIndex, localSpec, localCommitTx, htlcTxs = Nil)
             val remoteCommit = RemoteCommit(purpose.remoteCommitIndex, remoteSpec, remoteCommitTx.tx.txid, purpose.remotePerCommitmentPoint)
             signFundingTx(completeTx, localCommitSig, localCommit, remoteCommit)

eclair-core/src/main/scala/fr/acinq/eclair/io/PeerConnection.scala

@@ -28,7 +28,7 @@ import fr.acinq.eclair.remote.EclairInternalsSerializer.RemoteTypes
 import fr.acinq.eclair.router.Router._
 import fr.acinq.eclair.wire.protocol
 import fr.acinq.eclair.wire.protocol._
-import fr.acinq.eclair.{FSMDiagnosticActorLogging, Features, InitFeature, Logs, TimestampMilli, TimestampSecond}
+import fr.acinq.eclair.{FSMDiagnosticActorLogging, Features, InitFeature, Logs, TimestampMilli, TimestampSecond, UnknownFeature}
 import scodec.Attempt
 import scodec.bits.ByteVector
 
@@ -206,9 +206,19 @@ class PeerConnection(keyPair: KeyPair, conf: PeerConnection.Conf, switchboard: A
         stay()
 
       case Event(msg: LightningMessage, d: ConnectedData) if sender() != d.transport => // if the message doesn't originate from the transport, it is an outgoing message
+        val useExperimentalSplice = d.remoteInit.features.unknown.contains(UnknownFeature(154)) || d.remoteInit.features.unknown.contains(UnknownFeature(155))
         msg match {
-          case batch: CommitSigBatch => batch.messages.foreach(msg => d.transport forward msg)
-          case msg => d.transport forward msg
+          // If our peer is using the experimental splice version, we convert splice messages.
+          case msg: SpliceInit if useExperimentalSplice => d.transport forward ExperimentalSpliceInit.from(msg)
+          case msg: SpliceAck if useExperimentalSplice => d.transport forward ExperimentalSpliceAck.from(msg)
+          case msg: SpliceLocked if useExperimentalSplice => d.transport forward ExperimentalSpliceLocked.from(msg)
+          case msg: TxAddInput if useExperimentalSplice => d.transport forward msg.copy(tlvStream = TlvStream(msg.tlvStream.records.filterNot(_.isInstanceOf[TxAddInputTlv.SharedInputTxId])))
+          case msg: TxSignatures if useExperimentalSplice => d.transport forward msg.copy(tlvStream = TlvStream(msg.tlvStream.records.filterNot(_.isInstanceOf[TxSignaturesTlv.PreviousFundingTxSig])))
+          case batch: CommitSigBatch if useExperimentalSplice => batch.messages.foreach(msg => d.transport forward msg.copy(tlvStream = TlvStream(msg.tlvStream.records.filterNot(_.isInstanceOf[CommitSigTlv.FundingTx]))))
+          case batch: CommitSigBatch =>
+            d.transport forward StartBatch.commitSigBatch(batch.channelId, batch.batchSize)
+            batch.messages.foreach(msg => d.transport forward msg.copy(tlvStream = TlvStream(msg.tlvStream.records.filterNot(_.isInstanceOf[CommitSigTlv.ExperimentalBatchTlv]))))
+          case _ => d.transport forward msg
         }
         msg match {
           // If we send any channel management message to this peer, the connection should be persistent.
@@ -348,8 +358,51 @@ class PeerConnection(keyPair: KeyPair, conf: PeerConnection.Conf, switchboard: A
         // We immediately forward messages to the peer, unless they are part of a batch, in which case we wait to
         // receive the whole batch before forwarding.
         msg match {
+          case msg: StartBatch =>
+            if (!msg.messageType_opt.contains(132)) {
+              log.debug("ignoring start_batch: we only support batching commit_sig messages")
+              d.transport ! Warning(msg.channelId, "invalid start_batch message: we only support batching commit_sig messages")
+              stay()
+            } else if (msg.batchSize > 20) {
+              log.debug("ignoring start_batch with batch_size = {} > 20", msg.batchSize)
+              d.transport ! Warning(msg.channelId, "invalid start_batch message: batch_size must not be greater than 20")
+              stay()
+            } else {
+              log.debug("starting commit_sig batch of size {} for channel_id={}", msg.batchSize, msg.channelId)
+              d.commitSigBatch_opt match {
+                case Some(pending) if pending.received.nonEmpty =>
+                  log.warning("starting batch with incomplete previous batch ({}/{} received)", pending.received.size, pending.batchSize)
+                  // This is a spec violation from our peer: this will likely lead to a force-close.
+                  d.transport ! Warning(msg.channelId, "invalid start_batch message: the previous batch is not done yet")
+                  d.peer ! CommitSigBatch(pending.received)
+                case _ => ()
+              }
+              stay() using d.copy(commitSigBatch_opt = Some(PendingCommitSigBatch(msg.channelId, msg.batchSize, Nil)))
+            }
+          case msg: HasChannelId if d.commitSigBatch_opt.nonEmpty =>
+            // We only support batches of commit_sig messages: other messages will simply be relayed individually.
+            val pending = d.commitSigBatch_opt.get
+            msg match {
+              case msg: CommitSig if msg.channelId == pending.channelId =>
+                val received1 = pending.received :+ msg
+                if (received1.size == pending.batchSize) {
+                  log.debug("received last commit_sig in batch for channel_id={}", msg.channelId)
+                  d.peer ! CommitSigBatch(received1)
+                  stay() using d.copy(commitSigBatch_opt = None)
+                } else {
+                  log.debug("received commit_sig {}/{} in batch for channel_id={}", received1.size, pending.batchSize, msg.channelId)
+                  stay() using d.copy(commitSigBatch_opt = Some(pending.copy(received = received1)))
+                }
+              case _ =>
+                log.warning("received {} as part of a batch: we don't support batching that kind of messages", msg.getClass.getSimpleName)
+                if (pending.received.nonEmpty) d.peer ! CommitSigBatch(pending.received)
+                d.peer ! msg
+                stay() using d.copy(commitSigBatch_opt = None)
+            }
           case msg: CommitSig =>
-            msg.tlvStream.get[CommitSigTlv.BatchTlv].map(_.size) match {
+            // We keep supporting the experimental version of splicing that older Phoenix wallets use.
+            // Once we're confident that enough Phoenix users have upgraded, we should remove this branch.
+            msg.tlvStream.get[CommitSigTlv.ExperimentalBatchTlv].map(_.size) match {
               case Some(batchSize) if batchSize > 25 =>
                 log.warning("received legacy batch of commit_sig exceeding our threshold ({} > 25), processing messages individually", batchSize)
                 // We don't want peers to be able to exhaust our memory by sending batches of dummy messages that we keep in RAM.
@@ -381,6 +434,16 @@ class PeerConnection(keyPair: KeyPair, conf: PeerConnection.Conf, switchboard: A
                 d.peer ! msg
                 stay()
             }
+          // If our peer is using the experimental splice version, we convert splice messages.
+          case msg: ExperimentalSpliceInit =>
+            d.peer ! msg.toSpliceInit()
+            stay()
+          case msg: ExperimentalSpliceAck =>
+            d.peer ! msg.toSpliceAck()
+            stay()
+          case msg: ExperimentalSpliceLocked =>
+            d.peer ! msg.toSpliceLocked()
+            stay()
           case _ =>
             d.peer ! msg
             stay()
@@ -613,6 +676,7 @@ object PeerConnection {
                            gossipTimestampFilter: Option[GossipTimestampFilter] = None,
                            behavior: Behavior = Behavior(),
                            expectedPong_opt: Option[ExpectedPong] = None,
+                           commitSigBatch_opt: Option[PendingCommitSigBatch] = None,
                            legacyCommitSigBatch_opt: Option[PendingCommitSigBatch] = None,
                            isPersistent: Boolean) extends Data with HasTransport
 

eclair-core/src/main/scala/fr/acinq/eclair/wire/protocol/ChannelTlv.scala

@@ -234,8 +234,16 @@ sealed trait ChannelReestablishTlv extends Tlv
 
 object ChannelReestablishTlv {
 
+  /**
+   * When disconnected in the middle of an interactive-tx session, this field is used to request a retransmission of
+   * [[TxSignatures]] for the given [[txId]].
+   */
   case class NextFundingTlv(txId: TxId) extends ChannelReestablishTlv
+
+  /** The txid of the last [[ChannelReady]] or [[SpliceLocked]] message received before disconnecting, if any. */
   case class YourLastFundingLockedTlv(txId: TxId) extends ChannelReestablishTlv
+
+  /** The txid of our latest outgoing [[ChannelReady]] or [[SpliceLocked]] for this channel. */
   case class MyCurrentFundingLockedTlv(txId: TxId) extends ChannelReestablishTlv
 
   object NextFundingTlv {
@@ -245,6 +253,7 @@ object ChannelReestablishTlv {
   object YourLastFundingLockedTlv {
     val codec: Codec[YourLastFundingLockedTlv] = tlvField("your_last_funding_locked_txid" | txIdAsHash)
   }
+
   object MyCurrentFundingLockedTlv {
     val codec: Codec[MyCurrentFundingLockedTlv] = tlvField("my_current_funding_locked_txid" | txIdAsHash)
   }
@@ -301,3 +310,14 @@ object ClosingTlv {
   )
 
 }
+
+sealed trait StartBatchTlv extends Tlv
+
+object StartBatchTlv {
+  /** Type of [[LightningMessage]] that is included in the batch, when batching a single message type. */
+  case class MessageType(tag: Int) extends StartBatchTlv
+
+  val startBatchTlvCodec: Codec[TlvStream[StartBatchTlv]] = tlvStream(discriminated[StartBatchTlv].by(varint)
+    .typecase(UInt64(1), tlvField(uint16.as[MessageType]))
+  )
+}