broadcastTransaction - not enough gas #4602

novaknole · 2024-02-15T00:14:25Z

novaknole
Feb 15, 2024

Ethers Version

6.11.1

Search Terms

No response

Describe the Problem

I'm trying to let metamask sign a transaction using eth_sign and then broadcast it. I know eth_sign of metamask is dangerous, but just playing with it, nothing more.

Code Snippet

const txDetails = {
    from: "0x27A831Bb73811349537BA7A5121671189A40b82C",
    nonce: currentNonce + 1, 
    to: '0x75889e6cBE19B2Bc4CfC29EF5c78A03E1e100009',
    gasLimit: "0x249F0",
    value: ethers.utils.parseEther('0.0015').toHexString(),
    data: '0x', 
};

console.log(ethers.utils.parseEther('0.0015').toHexString())

const txMessage = JSON.stringify(txDetails);
const messageBytes = ethers.utils.toUtf8Bytes(txMessage);

// Hash the message using Keccak-256
const hash = ethers.utils.keccak256(messageBytes);

signer.provider.send("eth_sign", [hash, accounts[0]])
     .then((signature) => {

    // Remove the '0x' prefix if present
    const sig = signature.startsWith('0x') ? signature.slice(2) : signature;

    // Parse the signature components
    const r = '0x' + sig.slice(0, 64);
    const s = '0x' + sig.slice(64, 128);
    const v = '0x' + sig.slice(128, 130);

Then,

const Tx = require("ethereumjs-tx").Transaction;
const ethers = require('ethers');

const tx = new Tx({
    from: "0x27A831Bb73811349537BA7A5121671189A40b82C",
    nonce: 23, // this is correct
    to: "0x75889e6cBE19B2Bc4CfC29EF5c78A03E1e100009",
    value: "0x05543df729c000",
    gasLimit: "0x249F0",
    data: '0x',
    v: '0x1b',
    r: '0xb753cef2ef382bfd3fa0be57b935b1a2a4f58000d9ccf08a15986ac576762013',
    s: '0x37555651826e4f52d2e048ea64f182a7b13197ef2db26527f1b5df227a9d7092'
});

let provider = new ethers.JsonRpcProvider("https://goerli.infura.io/v3/a840be32fa314311992ca92da88145fe");

const serializedHex = "0x" + tx.serialize().toString("hex");
const { hash } = await provider.broadcastTransaction(
   serializedHex
);

The error is caused by broadcastTransaction, but I'm not sure what I'm doing wrong. I definitely got the balance on the from address and provider doesn't need any balance

Errors

Error: insufficient funds for intrinsic transaction cost (transaction="0xf8671780830249f09475889e6cbe19b2bc4cfc29ef5c78a03e1e1000098705543df729c000801ba0b753cef2ef382bfd3fa0be57b935b1a2a4f58000d9ccf08a15986ac576762013a037555651826e4f52d2e048ea64f182a7b13197ef2db26527f1b5df227a9d7092", info={ "error": { "code": -32000, "message": "insufficient funds for gas * price + value: balance 0, tx cost 1500000000000000, overshot 1500000000000000" } }, code=INSUFFICIENT_FUNDS, version=6.11.1)

Environment

Ethereum (mainnet/ropsten/rinkeby/goerli)

Environment (Other)

No response

ricmoo · 2024-02-15T00:38:07Z

ricmoo
Feb 15, 2024
Maintainer

That error means you don’t have a sufficient balance at the from address to send the transaction.

You must have enough funds to cover:

the value
the gasLimit * maxFeePerGas (or gasPrice for legacy transactions like yours above)
Each byte of data has an additional cost (bytes of 0 cost less than non-zero bytes)
If the to is null, you are performing an init transaction, which has an additional cost of 35k gas to create a new account
If there is an access list, there is a fee for each slot that needs to be “warmed”
If there are blobs, there is a fee per blob

And there are likely things I forgot to mention, but the intrinsic cost of a transaction must be met otherwise the transaction will be ignore (to prevent DoS) and will not be forwarded to the gossip network.

The error you are getting is coming directly from the node you are connected to. Ethers is just passing the error along to you; broadcastTransaction doesn’t do anything special. ;)

If you compare your balance in that account to the value + gasPrice * gasLimit I suspect you will see it isn’t enough to cover the cost of the transaction. If it is, include the from address and network and I can check the maths. :)

Does that make sense?

10 replies

ricmoo Feb 15, 2024
Maintainer

If you want to learn more about manually signing (which I assume is your goal?), you can use ethers.SigningKey(privately) to sign a digest directly.

novaknole Feb 15, 2024
Author

@ricmoo

I am using the latest version metamask. Just installed it yesterday.
I compared the signature given by eth_sign and personal_sign and they are both the same. What's interesting is that even if I use personal_sign, recoverAddress from ethers doesn't return the account that I signed the hash with, it returns a different address(where the hell did it get that unknown address from ? lol)
It's not about old/new metamask, what metamask docs says is the following: "the wallet user can re-enable eth_sign through advanced settings." (which I did) - https://docs.metamask.io/wallet/concepts/signing-methods/

ricmoo Feb 15, 2024
Maintainer

You need to use the hashMessage function to perform EIP-191 hashing. Make sure you read the documentation as it includes details about converting hashes to strings. I’m not sure which MetaMask expects.

If personal_sign and eth_sign are returning the same value, than eth_sign is not signing digests, so it isn’t performing the necessary signing you require for your above code. You need to sign the digest. It is signing an EIP-191 prefixed message. I imagine the enable you are doing just enables the end point, not the legacy (insecure) functionality.

ricmoo Feb 15, 2024
Maintainer

(if the hashes match, I think their documentation about re-enabling allows signing arbitrary data it is outdated)

novaknole Feb 15, 2024
Author

@ricmoo Finally, I can't believe I made it.

For future references, if you need this information, here are things how I made it.

in the browser, I was using ethers v5 version and somehow, I'm not sure why, eth_sign and personal_sign were behaving the same way. I updated to v6 and eth_sign actually works as it was working in the past while it was dangerous.

Also, I was using ethereumjs-tx and this package drove me crazy. No offence to their package owner, but I moved to ethers.transaction object and bravo, I made it work in 10 min.

Amazing package @ricmoo :)

novaknole · 2024-02-16T14:30:04Z

novaknole
Feb 16, 2024
Author

hey Richard. @ricmoo.

I wanted to ask a slightly different question, but probably related to this about hashes. I'm horrible at cryptography, so forgive me my noobness.

hacker creates a tx object:

const tx = ethers.Transaction.from({
       to: 'hackerAddress',
       signature: ''
  })

and in the signature, just puts some random signature directly and then broadcasts this tx.

What geth node does is receives this, decodes it(i.e unmarshalls) and tries to recover the address. geth uses tx.hash() and the signature to do this. See. What's interesting is even though signature in the tx is not the one that signed on the current tx, Ecrecover will still return an address(not the one that signed it), but some address(hell knows what). This is also happening in ethers.js - recoverAddress always returns an address. Now, in geth, there's no other validation and whatever address it recovered, it proceeds to transfer eth from it to hacker address. What if address, returned by recoverAddress, is actually someone's active address, in that case, it will succeed, even though that address didn't sign the tx at all. Am I making any sense ? The problem in smart contract is solved by including owner inside hash and then doing: require(signer == owner, "MyFunction: invalid signature");, but I don't see the same thing for geth and plain transactions. Note that from is not included in hash generation of tx in geth and for sure, not in ethers.

6 replies

novaknole Feb 16, 2024
Author

Oh.

in your first paragraph, what you're doing is pretty easy because you're the one that also recoversAddress to know what address it computed and then for sure, if you send eth on it, it already has funds and it will succeed.

My point was that if I am a hacker, what if my guessed signature actually is as if recoverAddress returns the already, existing, active user's address. In this case, boom, the user is doomed to lose funds even though he didn't sign anything at all.

Q1 : Do you agree with this exact analysis ?

So what you're saying is if I just picked a random signature and did the recoverAddress, the address it returns would never be an active user's address. correct ?

I think I understand what you explained. I'm not too experienced in the cryptography and this is a good sign that I will look into learning more in the future.

Thank you <3

ricmoo Feb 16, 2024
Maintainer

Right. What I’m saying, is you as a hacker cannot “guess” my signature.

If you set up a computer to do the following: tx = { to: hackerAddy }; while (tx.from != victimAddr) { tx.signature = randomSignature(); } it would never succeed, even if it ran until well beyond the end of time. The number is too big for you to ever guess.

It’s like asking “what if someone guesses my private key?”; it will never happen. There isn’t enough time or energy in the universe for you to guess. :)

(this all assumes you pick your private key in a truly random way; historically there have been hacks where a wallet was using an unsafe random source and funds were stolen)

novaknole Feb 16, 2024
Author

@ricmoo

I totally agree with that. I can't guess yours, but I will just choose a random signature and recoverAddress might return an address that belongs to someone. Assuming there's 2B ethereum's different addresses, what if my random signature that I guessed causes recoverAddress to return an address that belongs to someone in 2B people. Are you saying this is not possible either ?

ricmoo Feb 16, 2024
Maintainer

That’s also not possible (probabilistically). The numbers are astronomical. You will never guess any Ethereum address, even if you tried to the end of time. Any address. :)

The numbers are just very, very large. You won’t ever guess my private key, but you would also never guess anyone else’s private key. Even if every person on earth had trillions of private keys, it would have almost no effect on the odds.

The numbers are big. :)

novaknole Feb 16, 2024
Author

Makes sense. Thanks so much Richard.

Whenever you have time, I created a discussion a couple of days ago. No rush though - #4600

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

broadcastTransaction - not enough gas #4602

{{title}}

{{editor}}'s edit

{{editor}}'s edit

Replies: 2 comments 16 replies

{{title}}

{{title}}

{{title}}

{{title}}

{{title}}

{{title}}

{{editor}}'s edit

{{editor}}'s edit

{{title}}

{{editor}}'s edit

{{editor}}'s edit

{{title}}

{{title}}

{{title}}

{{title}}

{{title}}

Select a reply

broadcastTransaction - not enough gas #4602

novaknole Feb 15, 2024

Ethers Version

Search Terms

Describe the Problem

Code Snippet

Errors

Environment

Environment (Other)

Replies: 2 comments · 16 replies

ricmoo Feb 15, 2024 Maintainer

ricmoo Feb 15, 2024 Maintainer

novaknole Feb 15, 2024 Author

ricmoo Feb 15, 2024 Maintainer

ricmoo Feb 15, 2024 Maintainer

novaknole Feb 15, 2024 Author

novaknole Feb 16, 2024 Author

novaknole Feb 16, 2024 Author

ricmoo Feb 16, 2024 Maintainer

novaknole Feb 16, 2024 Author

ricmoo Feb 16, 2024 Maintainer

novaknole Feb 16, 2024 Author

novaknole
Feb 15, 2024

Replies: 2 comments 16 replies

ricmoo
Feb 15, 2024
Maintainer

ricmoo Feb 15, 2024
Maintainer

novaknole Feb 15, 2024
Author

ricmoo Feb 15, 2024
Maintainer

ricmoo Feb 15, 2024
Maintainer

novaknole Feb 15, 2024
Author

novaknole
Feb 16, 2024
Author

novaknole Feb 16, 2024
Author

ricmoo Feb 16, 2024
Maintainer

novaknole Feb 16, 2024
Author

ricmoo Feb 16, 2024
Maintainer

novaknole Feb 16, 2024
Author