Skip to content

Latest commit

 

History

History
269 lines (162 loc) · 13.9 KB

Metalabel-security-review.md

File metadata and controls

269 lines (162 loc) · 13.9 KB
Raw

Introduction

A time-boxed security review of the Metalabel protocol was done by pashov, with a focus on the security aspects of the application's implementation.

Disclaimer

A smart contract security review can never verify the complete absence of vulnerabilities. This is a time, resource and expertise bound effort where I try to find as many vulnerabilities as possible. I can not guarantee 100% security after the review or even if the review will find any problems with your smart contracts.

Protocol Overview

Metalabel is a release club protocol. Groups of people with similar interests can gather and drop work together as collaborators. The protocol allows the creators to split the economic rewards that their "metalabel" has received amongst them. It has multiple technical abstractions like:

  • Resources

    • Collection - ERC721 contract that mints tokens (records)
    • Split - a payment logic contract that has different percentage allocations for different contributors
    • Waterfall - a payment logic contract that enforces paying one party a certain amount before paying another party

  • Accounts - for a user to unlock the functionalities of a protocol he needs to register an account (account creation is gated at first, later it becomes permissionless)

  • Node - an ownable abstraction that groups Records and Resources and allows the owner account or a controller address to manage them

  • Engine - a contract for dropping a new Collection, it manages mints, royalties, rendering (tokenURI) of the ERC721

The protocol is well-tested, as it has 100% code coverage (line, branch, function).

More docs

Threat Model

System Actors

  • Account - can create a new Node
  • Node owner - can manage nodes (configure collections, their mints, their royalties and price) and add controllers
  • Controller - can manage nodes but can't add controllers
  • Mint Authority - can mint permissioned sequences

External functions:

  • AccountRegistry all methods - callable by anyone to register an account (unless owner is set)
  • NodeRegistry::createNode - callable by anyone that creates an account
  • DropEngine::mint - callable by anyone (unless mintAuthorities mapping is set for the sequence)

Q: What in the protocol has value in the market?

A: ERC721 token mints can be paid with ETH, so ETH value and also the ERC721 tokens themselves.

Q: What is the worst thing that can happen to the protocol?

  1. Node ownership stolen
  2. An attacker sets himself as mint payments' recipient
  3. Exploiting the mint functionality so it allows free or unlimited mints

Interesting/unexpected design choices:

The owner of a group node can set multiple controllers for all other nodes in the group.

The controller of a group node can manage all other nodes in the group.

Controllers can be smart contracts, accounts can be smart contracts as well.

Controller of a node has the same rights as it's owner (apart from adding more controllers).

Severity classification

Severity Impact: High Impact: Medium Impact: Low
Likelihood: High Critical High Medium
Likelihood: Medium High Medium Low
Likelihood: Low Medium Low Low

Security Assessment Summary

review commit hash - effe2e89996d6809c5ec6c6da10c663246f91fc1

Scope

The following smart contracts were in scope of the audit:

  • AccountRegistry
  • Collection
  • CollectionFactory
  • DropEngine
  • NodeRegistry
  • Resource
  • ResourceFactory
  • SplitFactory
  • WaterfallFactory
  • interfaces/**

The following number of issues were found, categorized by their severity:

  • High: 0 issues
  • Medium: 4 issues
  • Low: 2 issues
  • Informational: 8 issues

Findings Summary

ID Title Severity
[M-01] Insufficient input data validation for configureSequence Medium
[M-02] Using the transfer function of address payable is discouraged Medium
[M-03] Role transfer actions done in a single-step manner are dangerous Medium
[M-04] Records minted to an address that is a smart contract that can't handle ERC721 tokens will be stuck forever Medium
[L-01] resolveId will return the 0 ID even if an address does not have an account Low
[L-02] Inconsistent tokenData.owner validation Low
[I-01] Use a more abstract name for function parameter Informational
[I-02] NodeData.nodeType is used as an existence check and nothing else Informational
[I-03] NatSpec docs are incomplete Informational
[I-04] Unused import Informational
[I-05] Copy-pasted comments in WaterfallFactory Informational
[I-06] Missing override keyword Informational
[I-07] Use a safe pragma statement Informational
[I-08] Typos and grammatical errors Informational

Detailed Findings

[M-01] Insufficient input data validation for configureSequence

Severity

Impact: High, as some of those can result in a DoS or too big of a royalty payment

Likelihood: Low, as it requires a configuration error or a malicious actor

Description

An authorized address for a node can call Collection::configureSequence where most of the input is not validated properly. The _sequence parameter of the method is of type SequenceData which fields are not validated. Missing checks are the following:

  1. sealedBeforeTimestamp is bigger than block.timestamp
  2. sealedAfterTimestamp is always bigger than sealedBeforeTimestamp
  3. The difference between sealedBeforeTimestamp and sealedAfterTimestamp is at least 2 days for example
  4. The difference between sealedBeforeTimestamp and sealedAfterTimestamp is not more than 500 days for example
  5. The difference between sealedBeforeTimestamp and block.timestamp is not more than 10 days for example

Also in DropEngine::configureSequence the royaltyBps is not validated that it is not more than 100% (a value of 10000). I suggest you add a lower royaltyBps upper bound.

Recommendations

Add sensible constraints and validations for all user input mentioned above.

[M-02] Using the transfer function of address payable is discouraged

Severity

Impact: Medium, as sequence won't be usable as mints will revert

Likelihood: Medium, as it happens any time the recipient is a smart contract or a multisig wallet that has a receive function taking up more than 2300 gas

Description

The mint function in DropEngine uses the transfer method of address payable to transfer native asset funds to an address. This address is set by a node owner and is possible to be a smart contract that has a receive or fallback function that takes up more than the 2300 gas which is the limit of transfer. Examples are some smart contract wallets or multi-sig wallets, so usage of transfer is discouraged.

Recommendations

Use a call with value instead of transfer. There is no risk from reentrancy in the mint method as it has a check for the caller to be an EOA. When this is done you can remove the payable keyword from the revenueRecipient variable.

[M-03] Role transfer actions done in a single-step manner are dangerous

Severity

Impact: High, as important protocol functionality would become unusable

Likelihood: Low, as it requires an admin/owner error

Description

This is a common problem where transferring a role or admin rights to a different address can go wrong if this address is wrong and not actually controlled by any user. This is taken into consideration in NodeRegistry where the node ownership transfer is a two-step operation. Not the same approach is used in AccountRegistry though, where the contract inherits from Owned which has a single-step ownership transfer pattern and also the transferAccount logic in it is also using a single-step pattern.

Recommendations

Use a two-step ownership/rights transfer pattern in both the AccountRegistry ownership and in the transferAccount method, you can reuse the approach you used in NodeRegistry.

[M-04] Records minted to an address that is a smart contract that can't handle ERC721 tokens will be stuck forever

Severity

Impact: High, as records will be stuck forever

Likelihood: Low, as it requires the engine to allow smart contracts as minters and that contracts should not support handling of ERC721 tokens

Description

Both mintRecord methods in Collection use the _mint method of ERC721 which is missing a check if the recipient is a smart contract that can actually handle ERC721 tokens. If the case is that the recipient can't handle ERC721 tokens then they will be stuck forever. For this particular problem the safe methods were added to the ERC721 standard and Solmate has added the _safeMint method to check handle this problem in a minting context. This is actually not a problem in DropEngine because it allows only EOAs to mint, but since users can freely implement Engines then this is a valid problem.

Recommendations

Prefer using _safeMint over _mint for ERC721 tokens, but do this very carefully, because this opens up a reentrancy attack vector. It's best to add a nonReentrant modifier in the method that is calling _safeMint because of this.

[L-01] resolveId will return the 0 ID even if an address does not have an account

This is error prone and requires all clients of the resolveId function to do checks for != 0, as we can see in NodeRegistry where this method is called. A better and safer approach is to just revert in the resolveId method when the subject does not have an account, then if the client of the method wants to catch the error he can do this.

[L-02] Inconsistent tokenData.owner validation

The mintRecord functionality in Collection allows the caller to mint tokens to any address since the to argument is not validated in any way. The problem is that the getTokenData inherited function has the following check:

require(data.owner != address(0), "NOT_MINTED");

which is not actually true, because a token could be minted to the zero address, even due to a mistake. I suggest to add zero-address checks for the to argument in both mintRecord functions in Collection for consistency.

[I-01] Use a more abstract name for function parameter

The broadcast and broadcastAndStore methods in ResourceFactory have a waterfall parameter but some other type of resource might be used - for example a Collection or a Split. Rename parameter from waterfall to resource in both methods.

[I-02] NodeData.nodeType is used as an existence check and nothing else

The only thing nodeType is used for in the system is to check if a node exists. This can be done with a simple boolean or if you actually need a "type" then use an enum where the first value is NON_EXISTENT for example.

[I-03] NatSpec docs are incomplete

In almost all methods the NatSpec documentation is incomplete as it is missing parameters and return variables in it. NatSpec documentation is essential for better understanding of the code by developers and auditors and is strongly recommended. Please refer to the NatSpec format and follow the guidelines outlined there.

[I-04] Unused import

The IERC721 import in Collection is not used. Remove it from the code.

[I-05] Copy-pasted comments in WaterfallFactory

Almost all methods reference a split instead of a waterfall in the contract, I expect they were copy-pasted from SplitFactory. Replace split with waterfall in all of the comments in WaterfallFactory.

[I-06] Missing override keyword

Most methods in contracts that are implemented and inherited from an interface are missing the override keyword. Go through each method and apply the keyword where it is right to do so. Some examples are the mintRecord and royaltyInfo methods.

[I-07] Use a safe pragma statement

Always use stable pragma statement to lock the compiler version and to have deterministic compilation to bytecode. Keep in mind 0.8.13 has a bug when using assembly, even though the codebase does not do that.

[I-08] Typos and grammatical errors

A number of typos in the codebase that need to be fixed:

AccountTransfered -> AccountTransferred

has be -> has been

Inheritting -> Inheriting

Otherise -> Otherwise

managaeable -> manageable

so long as their is -> as long as there is

entites -> entities

Otherise -> Otherwise

additonal -> additional

seqeuence -> sequence

funcionality -> functionality

reliquish -> relinquish