3 posts tagged with "Solidity"

The problem​

Source code verification requires compiling a contract written in a high-level language (e.g. Solidity, Vyper) to the bytecode, and comparing the compiled bytecode with the onchain bytecode. If there’s a match, we can say the given high-level code is the source-code of the contract at the given address.

The runtime bytecode of contracts by default also contain a special field at the end in CBOR encoding (auxdata). This field contains the hash of the contract metadata file (metadata hash), which acts as a fingerprint of the compilation. The metadata file has compiler settings, and source file hashes so the slightest change in the compiler settings or even a whitespace in any of the source files will cause a change in the metadata hash.

For a visual explanation of everything above, check out playground.sourcify.dev

Because of its sensitivity, some verifiers leave this field out in verification. In Sourcify’s case, if the recompiled bytecode and the onchain bytecodes match each other exactly (including the auxdata), it’s great. This will give us a “full match”. If not, we need to find the auxdatas and leave them out when comparing to be able to get at least a "partial match".

However this is not always trivial especially in these cases:

1. The creation bytecode of a contract does not necessarily have the CBOR encoded part at the very end of the bytecode. Although sometimes it’s found there, this field can be anywhere. In fact the only reason the CBOR encoded part is in the creation bytecode is because the runtime bytecode is embedded inside the creation bytecode as a whole.

   When executing the creation bytecode i.e. deploying the contract, the contract’s runtime bytecode needs to be returned. The runtime bytecode is already inside the creation bytecode so this part is extracted and returned by taking the offset and the length for the related bytecode and returning it. This can be anywhere inside the code. (Check this article for a comprehensive deep dive into contract creation)

2. The runtime bytecode has the CBOR encoded part always at the end of the contract (unless turned off with appendCbor: false). But the bytecode can contain other contract bytecodes nested inside, which also can have their own auxdatas, and these parts need to be ignored for a verification. This is found for example in factory contracts where a contract creates another contract and the child contract’s code is nested in the factory’s bytecode.

Now for other “special” parts of the bytecode, the compiler outputs the positions such as immutables in immutableReferences. Unfortunately this is not the case for auxdatas and we need to look elsewhere and find workarounds.

Workarounds​

If not the exact positions of the auxdatas, the compiler at least outputs the values. Inside the legacyAssembly object of the compiler output we can find the auxdata, which is under the key .auxdata

example legacyAssembly:

{
    ".code": [],
    ".data": {
        "0": {
            ".auxdata": "a26469706673582212203a05097003697b26b1da819218bcd95779598eaa90539e82a59ecbe4c09757e364736f6c63430007000033",
            ".code": [...]
        }
    }
}

At this point, one could think to do a simple string search in the bytecode for the auxdatas found in legacyAssembly, but it would be possible for an attacker to trick the search function and falsely ignore parts of the bytecode that are not supposed to be ignored.

The vulnerability​

Imagine we have the auxdata string from the compiler’s legacyAssembly above.

a26469706673582212203a05097003697b26b1da819218bcd95779598eaa90539e82a59ecbe4c09757e364736f6c63430007000033

This could be the auxdata of a simple child contract inside the whole bytecode that we know won’t be affected by the changes of our main contract.

For this specific example the attacker could embed these bytes inside the bytecode such a code in the main contract:

assembly {
    // Split the code from a push opcode:
    // a26469706673582212203a05097003697b26b1da819218bcd957
    // 79 (PUSH26)
    // 598eaa90539e82a59ecbe4c09757e364736f6c63430007000033

    mstore(0x598eaa90539e82a59ecbe4c09757e364736f6c63430007000033, 0xa26469706673582212203a05097003697b26b1da819218bcd957)
    // PUSH26 0xa26469706673582212203a05097003697b26b1da819218bcd957
    // PUSH26 0x598eaa90539e82a59ecbe4c09757e364736f6c63430007000033
    // MSTORE
}

By chance (really) this auxdata of 53 bytes is split into two exactly from the middle but this doesn’t have to be the case. Remember the large middle portion of the CBOR encoding contains the IPFS hash so one can salt and iterate it.

Imagine the source code of the attacker compiles to the code below. Putting new lines to demonstrate the (allegedly) auxdata part:

0x6080...b732b960691b604482015260640160405180910390fd5b5f80546040516001600160a01b03808516939216917f342827c97908e5e2f71151c08502a66d44b6f758e3ac2f1de95f02eb95f0a73591a35f80546001600160a01b0319166001600160a01b039290921691909117905579
  a26469706673582212203a05097003697b26b1da819218bcd95779598eaa90539e82a59ecbe4c09757e364736f6c63430007000033
  52565b5f6a636f6e736f6c652e6c6f6790505f80835160208501845afa505050565b6101756101a4565b565b5f60208284031215610187575f80fd5b81356001600160a01b038116811461019d575f80fd5b9392505050565b634e487b7160e01b5f52605160045260245ffdfea2646970667358221220212c0514e8c0db310d02690fc2def199f4fba3828f2401ec2b8d7104e450b8b164736f6c63430008180033

This is what we get from the source code the attacker gives us to verify. So we go: “Oh right there's an auxdata a26469706673582212203a05097003697b26b1da819218bcd95779598eaa90539e82a59ecbe4c09757e364736f6c63430007000033 in this bytecode. We should ignore the corresponding part in the (onchain) bytecode to have a partial match.”

Oops now we are ignoring a part in the bytecode that we're not supposed to. These code parts are only meant for non-executable code whereas we embedded this with an assembly block.

In the attacker’s onchain bytecode (what actually will be executed vs. the verified code) the attacker could have placed anything in this assembly block for 53 bytes. I leave it up to your imagination what can be done with this ignored bytecode block.

The gist is, we need to make sure these to-be-ignored blocks are actually auxdatas and not coming for an executable code block. How do we do it?

The solution(s)​

Well, we know that the IPFS hash inside the auxdata is the hash of the metadata file and the metadata file contains the source file hashes. So we can touch all source files to change their hashes, e.g. by adding a whitespace at the end of each. By touching every single source file, we make sure the nested auxdatas will be modified as well. If we compile again, we will have the exact same bytecode just with differences at the metadata hashes. Then we can locate the metadata hashes by comparing the original and edited bytecodes side by side.

But we need one more thing: Now we know where the metadata hashes are but that is just a substring of the whole CBOR auxdata. So we need to figure out where the CBOR auxdata starts and ends.

Blockscout solution​

One way to do this is to start at the metadata hash positions we've found by comparing and go extend the byte substring byte-by-byte and each time try to decode the whole byte string in CBOR. If at one point successful, we know that the auxdata ends here. Remember that right after the CBOR encoding you'll find the length of the encoded part, so we know where it starts as well.

Indeed this is how Blockscout finds the auxdata positions.

Sourcify solution​

The way we approach this in Sourcify is by again making use of the legacyAssembly.

These are roughly the steps:

1. Use bytecodes: Compare the original bytecode to the whitespaced (edited) contract’s bytecode. This will give us the positions of the metadata hashes, remember not the whole auxdata.
2. Use legacyAssembly: Compare the auxdatas from legacyAssembly s of both contracts. We will get a auxdataDiff between each auxdata (1st auxdata in original vs 1st in edited etc.). The diff will not exactly be the whole metadata hashes because CIDv0 IPFS hashes start with Qm but the rest of the hash. The other parts of the auxdatas will be the same. We also keep the position of the diff inside the whole auxdata diffStart:

   interface AuxdataDiff {
       real: string;
       diffStart: number;
       diff: string;
   }
   
3. Remember these are the metadata hashes. If they are equal, we can now find where the whole auxdata starts with:

   for (const position of positions) {
       for (const auxdataDiff of auxdataDiffs) {
           // Compare if the diff from raw bytecode is equal the diff from `legacyAssembly` auxdatas
           if (editedBytecode.substring(position + auxdataDiff.diff.length) === auxdataDiff.diff)
               return originalBytecode.substring(position - auxdataDiff.diffStart, position + auxdataDiff.diff.length);
       }
   }
   

Original:

0x6080...                CBOR auxdata     
1909117905579a26469706673582212203a05097003697b26b1da819218bcd95779598eaa90539e82a59ecbe4c09757e364736f6c6343000700003352565b5f6a636f6e736f6c652e6c6

Edited:

0x6080...                CBOR auxdata     
1909117905579a2646970667358221220dceca8706b29e917dacf25fceef95acac8d90d765ac926663ce4096195952b6164736f6c6343000700003352565b5f6a636f6e736f6c652e6c6
             └──────────────────┘↑
                diffStart        position

An Alternative​

1. Start with a string search inside the bytecode for the auxdatas from legacyAssembly of the contract. Now we have the positions of potential auxdatas of the original contract.
2. Next we whitespace the source files and compile the contract again. Let’s call it the edited contract.
3. Finally we check if the bytecode substrings from the original contract and the edited contract have changed at the positions we found at the 1st step. We expect these to change if they indeed contain a real auxdata and not some custom bytecode.

Thanks to Rim from Blockscout for pointing out this alternative.

Making life easier for verifiers​

To avoid doing all these nitty workarounds we just proposed the Solidity compiler to output the positions of the auxdatas, similar to the immutableReferences field: https://github.com/ethereum/solidity/issues/14827

We are still going to need to do this for the compiler versions before this gets implemented but still it would be less work in verification, particularly not having to compile contracts twice.

Since we edited the original source code with whitespaces and compiled the contract, we also have the legacyAssembly for the edited contract, which contain auxdatas. If we compare all the auxdatas extracted from legacyAssembly s of both, we will get a diff of each auxdata field which will be the metadata hashes. The rest of the auxdatas will be the same.

Introduction​

Solidity compiler has a feature, not known by everyone, that appends the IPFS hash of the contract metadata to the contract bytecode. This hash effectively acts as a fingerprint of the compilation, and when deployed, goes onchain. With that, we can verify the contracts "perfectly" and fetch the contract source code from IPFS. One of our missions at Sourcify is to make this feature more known and used, but not everyone is a fan of it.

(If you don't fully understand the metadata hash check out our playground to see it in action.)

I argue this is the only foolproof way to verify contracts. Languages and tooling should come together and come up with a common standard. We should look back at what worked and what didn't, and come up with a better next version.

Runtime code vs Creation code​

In source-code verification you compare a bytecode to a high-level code (Solidity, Vyper).

When you compile a contract you get two bytecodes:

Runtime bytecode is the code of the contract living on the blockchain. This is what really gets executed when you call a contract. You'll find it if you look at the bytecode of an unverified contract in a block explorer or when you call eth_getCode(address) on the contract.

Creation bytecode is the code that will be executed by the EVM when the contract is being deployed, which will store the runtime code at contract's address.

Since the terms are not well defined, some terminology:

• "code" = "bytecode" in this context. Sometimes people just call it "runtime code", or "creation code".
• "Init code" = "Creation bytecode". This is usually used in create2 context.
• "Deployed Bytecode" = "Runtime Bytecode". This is another common way to refer to the runtime bytecode by the Solidity compiler and frameworks. I refrain from using this as sometimes the contract is not deployed and "runtime code" is more accurate.
• evm.bytecode = "Creation bytecode". The Solidity compiler refers to it as this in the output.
• evm.deployedBytecode = "Runtime bytecode". Same as above.

Which bytecode?​

Let's go back to the source code verification. The problem we are trying to solve is we have a contract, and we want to see the original source code of it. Because we humans, can't really read bytecodes.

However, a contract has two bytecodes, which one should we compare the source code to?

Verifying with Creation Bytecode​

One can say that the bytecode counterparty of a contract written in a high level language is the creation bytecode. Because, in a typical contract deployment this is what you give to the EVM to execute.

The problem with the creation bytecode is that it's not always stored onchain. The only time you see this is when you deploy a contract from an Externally Owned Account (EOA) by putting the creation bytecode in the tx.data and setting the receiver tx.to to null. In that case you'll see the creation bytecode if you look at the transaction.

However, for contracts created by other contracts (e.g. factories) it is executed once and then discarded. So someone needs to index and save the creation bytecodes somewhere and you need to trust them. Whereas the runtime bytecode is stored onchain and you can request it from your node with eth_getCode.

On the other hand, the creation bytecode of a contract is not necessarily what the compiler outputs. The creation bytecode can be any code that will execute and store the runtime bytecode at the contract address. See @ricmoo's CREATE2 example. He demonstrates how to deploy and SELFDESTRUCT a contract, and finally deploy a completely different contract at the same address, even though CREATE2 addreses depend on the init code. In this case the init code is the same but it dynamically gets and writes the contract code from somewhere else. If you change the code where it's dynamically fetched from, you deploy a different contract at the same address. So for this contract, even if we knew its original source code, we can't compile and compare against its creation code.

Verifying with the Runtime Bytecode​

The runtime bytecode is the actual code of the contract and is readily available at eth_getCode. The compiler also outputs the runtime bytecode so one can verify contracts with the runtime bytecode too. With that, you can easily verify a contract on the "edge" (i.e. on your machine) trustlessly by getting the bytecode from your execution client.

The compiler output can be different than the onchain one as during deployment the runtime bytecode can be modified by writing the immutable values and the linked libraries in the placeholders. It's ok because, for Solidity, the compiler outputs the immutableReferences and libraries have a __$ placeholder, so we know where these are positioned in the bytecode.

The problem is, not everything in high-level contract code is represented in the runtime bytecode. Imagine this contract excerpt:

    constructor() {
        owner = msg.sender;
        emit OwnerSet(address(0), owner);
    }

I can deploy this contract but verify it with a slightly different contract with the following constructor, which can have huge implications:

    constructor() {
        owner = tx.origin;
        emit OwnerSet(address(0), owner);
    }

This is because this constructor code part will not be included in the runtime bytecode, and the owner value is not stored inside the bytecode but in the contract's storage.

Verifying with the Runtime Bytecode + Metadata Hash​

There's a way around this problem. If you verify a contract with its metadata hash appended to the runtime bytecode, you'll get a full match. This means the source code you are looking at is exactly the same as the one that was originally compiled, because if you change anything about the contract (even a whitespace), the metadata hash will change and you will not get a "full match" but a "partial match".

This, I'd argue, is the only foolproof way to verify a contract's source code. This method covers all the cases above and the ones I haven't mentioned or we don't know about yet. By being based on the runtime code, this also removes the need to trust a third party to index the creation bytecode, and instead you can get the bytecode from your own execution client's JSON RPC interface.

Problems with the Metadata Hash​

The main critisism of this feature is that the hash is too sensitive. It's both a bug and a feature that the hash changes even with a whitespace change.

A bigger problem is with the paths of the .sources.

  ...
  "sources": {
    "myDirectory/myFile.sol": {
      "keccak256": "0x123...",
      "license": "MIT",
      "urls": [ "bzz-raw://7d7a...", "dweb:/ipfs/QmN..." ]
    }
  }

The keys here are actually not file paths but source-unit names, meaning they can be arbitrary strings. This is especially a problem for projects deploying with CREATE2, where the address of the contract depends on the init code. Any difference in "path" will be a different metadata hash --> diferent bytecode --> different contract address. As a result, most of them just turn off this feature.

It's a bigger problem if the same codebase does not compile to the same bytecode on different platforms. The differences caused by comments/whitespaces are not that big of a deal if we can verify contracts at the deployment pipeline i.e. right at the point when they are deployed. This also means we need to stop flattening contracts. Ideally you never drag and drop any files to a website, but use a verification plugin on your tooling (Foundry, Hardhat) or IDE (Remix). No medium size contract would manually be verified.

What would be a more clever way to do this? If we are able get this right, we solve most of the problems.

Conclusion​

The two bytecodes associated with a contract are not always sufficient to correctly verify a contract. The only foolproof and decentralized way to do it is to use the runtime bytecode with the metadata hash appended to it. I believe this needs to be the default way to verify contracts, and only when you can't do it (like this bug), you should fall back to the partial match. Although at Sourcify we base our verification on this, most of the ecosystem don't make the partial vs full match distinction or are just aware of it.

As an outcome of this article I'd really want to see:

1. Other cases where a runtime bytecode or creation bytecode fails to correctly verify a contract.
2. Counter-arguments to the usefulness of the metadata hash.
3. Clever ways to mitigate the problems with the metadata hash.
4. Languages other than Solidity adopting this feature, and coming up with a standard for it.

Do have anything to add for these points above? Please reach out to me on Twitter or add your remarks in the discussion issue for this article (I'll link). I'll also be updating this article with the feedback I get, and be linking to discussions. This will be a living document.

In an ecosystem with the core values of transparency, security, and trust (and trustlessness); it is expected from all contract developers to publish their source code. If you're even slightly familiar with Ethereum, there is no need for further explaination.

But if I give you a source code, how do you make sure the published source code really is the source code of the contract? That's where source code verification comes into play.

What is source code verification?​

First thing first, all the smart contracts on blockchain are stored in bytecode. Just like our physical machines that only speak bits and bytes, Ethereum Virtual Machine also only understands bytes. If you ask the Ethereum blockchain the code of a contract, you only get a byte string.

So, let's say I give you a contract in Solidity and claim that this is the code behind the contract at "0xabcdef...". To verify, you need to make sure this code compiles to the same bytecode as the claimed contract at "0xabcdef...". This is the basic idea behind the smart contract verification: we compile a contract and check if the bytecode matches the one on blockchain.

You have probably made use of contract verification before. For many users this is the green checkmark in Etherscan:

You see the green checkmark and you are happy!

But is it really exactly the same code that is deployed?

The answer is, you don't know 🤷

In fact, no one else would be able to know except the contract developer, and he/she can't really prove it. The reason is, when compiling the contract i.e. translating the human-readable source code (in Solidity or any other higher-level language) to machine-readable bytecode, some information is lost. These include internal variable names, internal function names, names of contracts etc.

So yes this is functionally the same code as deployed: it compiles to the same bytecode as the original mysterious source code 🕵.

And you might be thinking, sure this is good enough. But:

• Someone can insert misleading comments, (internal) function or variable names
• Whoever verifies a contract first is chosen as the matching result, not the "authentic" one
• We can't verify things other than the contract's code itself (i.e. metadata)

In fact when not verified properly, it is possible to inject code that would be shown in the verified source code.

Enough bad news... There's actually a way to verify Solidity contracts that would cryptographically ensure the exactness of the source files and it is already here: It's called Sourcify!

This way of verifying contracts is what we call a perfect verification, (in contrast to partial verification). This is enabled by the Solidity contract metadata, and that the hash of it is appended to the contract's bytecode. The metadata hash acts as a fingerprint of the whole compilation and with the information in the metadata file we can completely reproduce the contract compilation.

Contract Metadata​

The Solidity compiler by default appends some information to the contract's bytecode in CBOR encoding. This special field, I like referring to as auxdata, usually contains the "Solidity version", the "metadata hash", and occasionally the "experimental" flag. The encoded data and it's decoding looks like this:

You can actually inspect this field and see the decoding in action for any contract in playground.sourcify.dev.

To see how this cryptographically ensures the exactness of the source files we need to look into the contents of the metadata file. The metadata file is a JSON document that looks like this and contains information on two things:

1. How to interact with the contract: ABI, documentation
2. How to reproduce a contract compilation: compiler version and settings, source file information

The latter is the relevant field for our purposes. Specifically, the fact that the metadata file contains source file hashes. To illustrate this, let's walk through what happens when you compile a contract and what happens when you change a source file.

When you compile a contract, the compiler computes the hashes of the source files and embeds this information in the metadata file. On the right side, you see the relevant fields of the metadata file:

Then the compiler takes the hash of this whole file:

And encodes it in the auxdata at the end of the bytecode:

So if you were to decode the auxdata you'd see:

What happens when we change something in the source files? Say we change a variable name or a comment in the new MyContract-diff.sol file. In turn the hash of the file changes, as well as the hash in the metadata:

...and of course the hash of the metadata file changes:

...and the auxdata changes:

Sooo, if we match both the bytecode + the appended auxdata, we have byte-by-byte exactly the same source code and compilation settings of the original deployed contract. This is a perfect verification.

If the bytecode matches but not the auxdata (which includes the metadata hash), we have a partial verification.

Did you notice?​

If you are familiar with IPFS and paid attention, you might ask: Can't we already get everything from the bytecode itself?

And yes, if published on IPFS, you can actually fetch the source code from the bytecode of a contract, because all the information is already there:

• The metadata IPFS hash is appended to the bytecode so (if published) you can fetch the metadata file.
• The metadata file contains (alongside the normal keccak256) the IPFS hashes of the source files so you can fetch the complete source code from IPFS.

So there's only one thing that you need to do as a contract developer: Publish your source files and metadata on IPFS.

Why do you need verification then? Isn't the source file already out there?

Although unlikely since the compiler does it automatically, someone can change the auxdata of the contract before deploying it and show you a different random source code. We make sure it really is the same code by doing a whole recompilation of the provided files and comparing the resulting bytecodes. Plus, we share all verified contracts in our repository on IPFS to make sure it's available.

Conclusion​

Perfect verification enables more secure and transparent verification on contracts, as well as other useful things such as decoding tx's and enabling human-readable contract interactions, but this is a topic for another article.

Next level smart contract verification is already here. We just need to adopt this way of verifying contracts as a community. Obviously, we need a lot of tooling, integrations, and more awareness. Let's step up and make this the standard way of verifying contracts!

(This article is a summary of my recent talks about Sourcify. If you are interested in learning more, check out one of the latest talks)