Reject Redundant Tx Antedecorator

modules/core/04-channel/types/errors.go

@@ -30,4 +30,7 @@ var (
 
 	// ORDERED channel error
 	ErrPacketSequenceOutOfOrder = sdkerrors.Register(SubModuleName, 21, "packet sequence is out of order")
+
+	// Antehandler error
+	ErrRedundantTx = sdkerrors.Register(SubModuleName, 22, "packet messages are redundant")
 )

modules/core/ante/ante.go

@@ -0,0 +1,72 @@
+package ante
+
+import (
+	sdk "github.com/cosmos/cosmos-sdk/types"
+	clienttypes "github.com/cosmos/ibc-go/modules/core/02-client/types"
+	channelkeeper "github.com/cosmos/ibc-go/modules/core/04-channel/keeper"
+	channeltypes "github.com/cosmos/ibc-go/modules/core/04-channel/types"
+)
+
+type AnteDecorator struct {
+	k channelkeeper.Keeper
+}
+
+func NewAnteDecorator(k channelkeeper.Keeper) AnteDecorator {
+	return AnteDecorator{k: k}
+}
+
+// AnteDecorator returns an error if a multiMsg tx only contains packet messages (Recv, Ack, Timeout) and additional update messages and all packet messages
+// are redundant. If the transaction is just a single UpdateClient message, or the multimsg transaction contains some other message type, then the antedecorator returns no error
+// and continues processing to ensure these transactions are included.
+// This will ensure that relayers do not waste fees on multiMsg transactions when another relayer has already submitted all packets, by rejecting the tx at the mempool layer.
+func (ad AnteDecorator) AnteHandle(ctx sdk.Context, tx sdk.Tx, simulate bool, next sdk.AnteHandler) (sdk.Context, error) {
+	// do not run redundancy check on DeliverTx or simulate
+	if (ctx.IsCheckTx() || ctx.IsReCheckTx()) && !simulate {
+		// keep track of total packet messages and number of redundancies across `RecvPacket`, `AcknowledgePacket`, and `TimeoutPacket/OnClose`
+		redundancies := 0
+		packetMsgs := 0
+		for _, m := range tx.GetMsgs() {
+			switch msg := m.(type) {
+			case *channeltypes.MsgRecvPacket:
+				if _, found := ad.k.GetPacketReceipt(ctx, msg.Packet.GetDestPort(), msg.Packet.GetDestChannel(), msg.Packet.GetSequence()); found {
+					redundancies += 1
+				}
+				packetMsgs += 1
+
+			case *channeltypes.MsgAcknowledgement:
+				if commitment := ad.k.GetPacketCommitment(ctx, msg.Packet.GetSourcePort(), msg.Packet.GetSourceChannel(), msg.Packet.GetSequence()); len(commitment) == 0 {
+					redundancies += 1
+				}
+				packetMsgs += 1
+
+			case *channeltypes.MsgTimeout:
+				if commitment := ad.k.GetPacketCommitment(ctx, msg.Packet.GetSourcePort(), msg.Packet.GetSourceChannel(), msg.Packet.GetSequence()); len(commitment) == 0 {
+					redundancies += 1
+				}
+				packetMsgs += 1
+
+			case *channeltypes.MsgTimeoutOnClose:
+				if commitment := ad.k.GetPacketCommitment(ctx, msg.Packet.GetSourcePort(), msg.Packet.GetSourceChannel(), msg.Packet.GetSequence()); len(commitment) == 0 {
+					redundancies += 1
+				}
+				packetMsgs += 1
+
+			case *clienttypes.MsgUpdateClient:
+				// do nothing here, as we want to avoid updating clients if it is batched with only redundant messages
+
+			default:
+				// if the multiMsg tx has a msg that is not a packet msg or update msg, then we will not return error
+				// regardless of if all packet messages are redundant. This ensures that non-packet messages get processed
+				// even if they get batched with redundant packet messages.
+				return next(ctx, tx, simulate)
+			}
+
+		}
+
+		// only return error if all packet messages are redundant
+		if redundancies == packetMsgs && packetMsgs > 0 {
+			return ctx, channeltypes.ErrRedundantTx
+		}
+	}
+	return next(ctx, tx, simulate)
+}