Smart contract applications that make use of the selfdestruct
functionality and their end-users.
The vulnerability has been classified as high
with a CVSS score of 8.2
. It has the potential to create a denial-of-service to all contracts that can invoke the selfdestruct
function to destroy a smart contract.
Due to the successfully coordinated security vulnerability disclosure, no smart contracts were impacted through the use of this vulnerability. Smart contract states and storage values are not affected by this vulnerability. User funds and balances are safe.
In Ethermint running versions before v0.17.2
, the contract selfdestruct
invocation permanently removes the corresponding bytecode from the internal database storage. However, due to a bug in the DeleteAccount
function, all contracts that used the identical bytecode (i.e shared the same CodeHash
) will also stop working once one contract invokes selfdestruct
, even though the other contracts did not invoke the selfdestruct
OPCODE.
The same contract bytecode can be deployed multiple times to create multiple contract instances. In the internal database, the bytecode is stored as a key-value entry bytecode hash --> bytecode
which is shared by those contracts. Unfortunately, when one of the contracts invokes selfdestruct
, it will remove the corresponding bytecode hash -> bytecode
entry, and thus it disables all the contracts that share the same bytecode.
The attack scenario is as follows:
selfdestruct
selfdestruct
operation on their redeployed contract, actively causing a DoS on the original and vulnerable contract. All transactions will fail until a workaround is used (see below).Has the problem been patched? What versions should users upgrade to?
This vulnerability has been patched in Ethermint versions ≥v0.18.0. The patch has state machine-breaking changes for applications using Ethermint so a coordinated upgrade procedure is required.
The patch removes the bytecode deletion logic, i.e. contract bytecodes are never deleted from the internal database after the patch.
At the moment, Ethermint does not track how many times each bytecode is used, and thus it cannot determine if it is safe to delete a particular bytecode on selfdestruct
invocations. This behavior is the same with go-ethereum.
Is there a way for users to fix or remediate the vulnerability without upgrading?
If a contract is subject to DoS due to this issue, the user can redeploy the same contract, i.e with identical bytecode, so that the original contract's code is recovered.
The new contract deployment restores the bytecode hash -> bytecode
entry in the internal state.
Are there any links users can visit to find out more?
If you have any questions or comments about this advisory:
Thanks to the
{ "nvd_published_at": "2022-08-05T13:15:00Z", "github_reviewed_at": "2022-08-18T19:04:47Z", "severity": "HIGH", "github_reviewed": true, "cwe_ids": [ "CWE-668" ] }