Submitted by zhaojie, also found by MevSec and berndartmueller
Anyone can add InTxTracker has been carried out, the node/observer will still deal with executive InTxTracker, Tx will not be repeated(The status of the vote has been changed and the vote cannot pass) ,but an attacker can cause network congestion.
A new architecture use Permissionless Tx Validation Model, anyone can add an InTxTracker, but the AddToInTxTracker function does not verify that an InTxTracker has already been executed, so a Tx that has already been executed can be added again.
Since the executed Txs are normal Tx, they can be verified by proof and can be added by anyone.
The observer through GetInboundTrackersForChain access to all the InTxTracker, then parse the Tx content, then MsgVoteOnObservedInboundTx messages sent to a vote, after voting by creates CCTX, Finally execute OutBoundTx.
After the CCTX is created, the InTxTracker will be removed. If added again at this time, the observer still retrieves the Tx and creates a vote, since VoteOnObservedInboundTx uses msg.Digest as the index for voting, so duplicate messages cannot be voted, and eventually CCTX will not be created and will not be executed.
But heres the problem: failing InTxTracker are not deleted, InTxTracker is deleted after a VoteOnObservedInboundTx vote successfully creates a CCTX, but will not be deleted if a cctx is not created ,they are always available to observers, and they keep on executing and failing, causing congestion when there are a large number of these invalid InTxTracker on the network.
An attacker can add all the executed InTxTracker to the network and attack the network, which can make the network not work.
VScode
To verify that the InTxTracker has been executed.
DadeKuma (Lookout) commented:
lumtis (ZetaChain) confirmed
0xean (Judge) decreased severity to Medium commented:
