[not for merge] (heavily wip) EIP4337 & EIP7702 #1438

popzxc · 2025-04-29T11:31:21Z

No description provided.

popzxc · 2025-05-14T11:18:32Z

system-contracts/bootloader/bootloader.yul

@@ -1910,6 +2033,39 @@ object "Bootloader" {


            <!-- @if BOOTLOADER_TYPE=='playground_batch' -->
+            function ethCallEvmConsturction(


This is needed to return deployment bytecode if eth_call is used to deploy contract.
Not directly related to this scope, but it was needed to make alto work, so I implemented it.

popzxc · 2025-05-14T11:20:28Z

system-contracts/contracts/AccountCodeStorage.sol

+            // EIP-2 still allows signature malleability for ecrecover(). Remove this possibility and make the signature
+            // unique. Appendix F in the Ethereum Yellow paper (https://ethereum.github.io/yellowpaper/paper.pdf), defines
+            // the valid range for s in (301): 0 < s < secp256k1n ÷ 2 + 1, and for v in (302): v ∈ {27, 28}. Most
+            // signatures from current libraries generate a unique signature with an s-value in the lower half order.
+            //
+            // If your library generates malleable signatures, such as s-values in the upper range, calculate a new s-value
+            // with 0xFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFEBAAEDCE6AF48A03BBFD25E8CD0364141 - s1 and flip v from 27 to 28 or
+            // vice versa. If your library also generates signatures with 0/1 for v instead 27/28, add 27 to v to accept
+            // these malleable signatures as well.
+            if (uint256(item.s) > 0x7FFFFFFFFFFFFFFFFFFFFFFFFFFFFFFF5D576E7357A4501DDFE92F46681B20A0) {
+                continue;
+            }


This part was copy-pasted from some other place, needs to be checked.

StanislavBreadless · 2025-05-15T15:47:55Z

system-contracts/contracts/AccountCodeStorage.sol

    bytes32 private constant EMPTY_STRING_KECCAK = 0xc5d2460186f7233c927e7db2dcc703c0e500b653ca82273b7bfad8045d85a470;

+    /// @notice Information about EIP-7702 delegated EOAs.
+    /// @dev Delegated EOAs.
+    mapping(address => address) private delegatedEOAs;


just in case, dont store anything in this contract except for code hashes for accounts, the role of code hash registry is used on the vm level and bad things will happen if unintended value is stored for some key

StanislavBreadless · 2025-05-15T15:49:39Z

system-contracts/bootloader/bootloader.yul

+
+                // Delegations are only processed if transaction is EIP7702 and
+                // authorization list is provided
+                let isEIP7702 := eq(getTxType(innerTxDataOffset), 4)


these checks should be included inside validateTypedTxStructure, if for the other types the reserverd dynamic is not zero, they should revert

StanislavBreadless · 2025-05-15T15:50:26Z

system-contracts/bootloader/bootloader.yul

+                // Delegations are only processed if transaction is EIP7702 and
+                // authorization list is provided
+                let isEIP7702 := eq(getTxType(innerTxDataOffset), 4)
+                let isDelegationProvided := gt(length, 0)


must not be zero:

"The transaction is considered invalid if the length of authorization_list is zero."

https://github.com/ethereum/EIPs/blob/master/EIPS/eip-7702.md

StanislavBreadless · 2025-05-15T15:54:24Z

system-contracts/contracts/AccountCodeStorage.sol

+            address authority = ecrecover(message, uint8(item.yParity + 27), bytes32(item.r), bytes32(item.s));
+
+            // ZKsync has native account abstraction, so we only allow delegation for EOAs.
+            if (this.getRawCodeHash(authority) != 0x00 && this.getAccountDelegation(authority) == address(0)) {


this. implies an external call to itself, it is better to avoid using this when it is not required for cheaper costs

StanislavBreadless · 2025-05-15T16:13:32Z

system-contracts/contracts/AccountCodeStorage.sol

+                delegatedEOAs[authority] = item.addr;
+
+                bytes32 codeHash = getRawCodeHash(item.addr);
+                _storeCodeHash(authority, codeHash); // TODO: Do we need additional checks here?


Here the semantics that you imply is not "I delegate to this address", but more like "I use the same bytecode as another address".

It has the following differences from EIP7702:

In EIP7702 extcodesize/extcodehash dont return the exact bytecode info. This is not really an important issue

In EIP7702 in case of delegation loops (i.e. default account pointing to another account), this wont work, but in your case it will.

There is a (potential) use-case of delegating to an address that has not yet been deployed, e.g. an account may want to initialize its implementation as a first-ever transaction. Not sure how important is it. It does not work in your case, since the bytecode copied is 0.

I would personally make it more like EIP7702: reset DefaultAccount to some default code that has another delegation as an immutable. Note, that this implementation will still have to check that another delegator is not a delegator.

But my approach has drawbacks:

It is is just harder to implement

In EIP7702 no actual delegatecall is done, but in my case it will be.

It will be more expensive since before executing the call I need to retrieve whether another address is a delegator.

So yeah.. not sure

StanislavBreadless · 2025-05-15T16:39:07Z

system-contracts/contracts/erc4337/EntryPointV01.sol

+contract EntryPointV01 is IEntryPoint, SystemContractBase {
+    using TransactionHelper for *;
+
+    function handleUserOps(PackedUserOperation[] calldata _ops) external {


One more difference from a normal tx: getCurrentPubdataSpent() does not behave the same way.

We use this variable so that each tx can check how much pubdata it has spent so far. If we want to, we can deprecate its behavior and say that it returns how much pubdata has been spent in the batch yet, but I am not sure whether it wont leak unnecessary information

StanislavBreadless · 2025-05-15T16:41:10Z

system-contracts/contracts/erc4337/EntryPointV01.sol

+                bootloaderBalanceAfter - bootloaderBalanceBefore >= _tx.gasLimit * _tx.maxFeePerGas,
+                "Transaction payment amount mismatch"
+            );
+            // TODO: should we send back excessive funds like bootloader does?


StanislavBreadless · 2025-05-15T16:42:29Z

system-contracts/contracts/erc4337/EntryPointV01.sol

+        );
+
+        // execute transaction
+        // TODO: do we need to do it through MsgValueSimulator?


in this case no

StanislavBreadless · 2025-05-15T16:43:22Z

system-contracts/contracts/erc4337/EntryPointV01.sol

+        }
+    }
+
+    function _executeTransaction(bytes32 _txHash, bytes32 _suggestedSignedHash, Transaction memory _tx) private {


this function does not handle user tx failing, though I guess it is the case in the initial iteration only

StanislavBreadless · 2025-05-15T16:46:32Z

system-contracts/contracts/erc4337/EntryPointV01.sol

+contract EntryPointV01 is IEntryPoint, SystemContractBase {
+    using TransactionHelper for *;
+
+    function handleUserOps(PackedUserOperation[] calldata _ops) external {


to not forget: as of now, the function will work for EOAs as well, I dont think this is what we want (the ability to start tx out of an empty account inside the transaction

StanislavBreadless · 2025-05-15T16:54:31Z

system-contracts/bootloader/bootloader.yul

@@ -1465,6 +1586,8 @@ object "Bootloader" {
                    setTxOrigin(BOOTLOADER_FORMAL_ADDR())
                }

+                processDelegations(innerTxDataOffset)


everything here would be reverted if the execution reverts.

from the doc:

Note, if transaction execution results in failure (e.g. any exceptional condition or code reverting), the processed delegation indicators is not rolled back.

StanislavBreadless · 2025-05-15T16:59:18Z

system-contracts/bootloader/bootloader.yul

                success := callAccountMethod({{PRE_PAYMASTER_SELECTOR}}, account, txDataOffset)
            }

            /// @dev Calls the `validateAndPayForPaymasterTransaction` method of a paymaster
            function validateAndPayForPaymasterTransaction(paymaster, txDataOffset) -> success {
+                // TODO: should we allow delegated accounts to use native paymasters?
+                // TODO: Gut feeling is that the answer is "NO" as we're deprecating EIP-712 txs


may need to do vibe check in the community for it, though I dont have strong opinions

StanislavBreadless · 2025-05-15T17:05:35Z

system-contracts/bootloader/bootloader.yul

+                        let innerTxDataOffset := add(txDataOffset, 32)
+                        let calldataPtr := getDataPtr(innerTxDataOffset)
+                        let value := getValue(innerTxDataOffset)
+                        ret := msgValueSimulatorMimicCall(from, from, value, calldataPtr)


Suggested change

ret := msgValueSimulatorMimicCall(from, from, value, calldataPtr)

ret := msgValueSimulatorMimicCall(to, from, value, calldataPtr)

StanislavBreadless · 2025-05-15T17:12:11Z

system-contracts/bootloader/bootloader.yul

+                let delegation := getDelegationAddress(from)
+                let rawCodeHash := 0
+                if gt(delegation, 0) {
+                    rawCodeHash := getRawCodeHash(delegation, true)


I have a somewhat bad feeling about this hack when it is coupled with a paymaster:

Paymaster can do any calls

But the world is in a strange state where the account actually has a delegator but the bytecodes are different

To make it safer one option could be similar to what I described in another thread: keep DefaultAccount code roughly same, but also store the delegate address.

Another alternative if we decide to stick to the bytecode-based approach is to disallow paymasters

popzxc added 3 commits April 24, 2025 16:08

Basic implementation

c6d283e

Align EntryPoint interface with L1

6b40b7a

7702 early draft

7fbb437

popzxc requested review from Deniallugo, vladbochok and StanislavBreadless as code owners April 29, 2025 11:31

popzxc changed the title ~~[not for merge] (heavily wip) EIP4339 & EIP7702~~ [not for merge] (heavily wip) EIP4337 & EIP7702 Apr 29, 2025

Invoke processDelegations

5a256d9

popzxc force-pushed the popzxc-erc4337 branch from 7444870 to 5a256d9 Compare April 30, 2025 06:19

popzxc added 5 commits April 30, 2025 15:41

7702 verification works

94f5a9b

Delegation works (hopefully)

cb98f1b

Delegation works

a7890d8

Delegate 7702 through mimiccall

1eee592

Merge remote-tracking branch 'origin/draft-v28' into popzxc-erc4337

eb9855c

popzxc changed the base branch from draft-v29 to draft-v28 May 6, 2025 07:32

Fixes after merge

55b1891

popzxc marked this pull request as draft May 7, 2025 08:16

Set code hash for delegated account

a204c5f

popzxc force-pushed the popzxc-erc4337 branch from 941fad9 to a204c5f Compare May 12, 2025 11:45

Support EVM deployments in eth_call

2f6735e

popzxc commented May 14, 2025

View reviewed changes

StanislavBreadless reviewed May 15, 2025

View reviewed changes

		@@ -1910,6 +2033,39 @@ object "Bootloader" {


		<!-- @if BOOTLOADER_TYPE=='playground_batch' -->
		function ethCallEvmConsturction(

	ret := msgValueSimulatorMimicCall(from, from, value, calldataPtr)
	ret := msgValueSimulatorMimicCall(to, from, value, calldataPtr)

[not for merge] (heavily wip) EIP4337 & EIP7702 #1438

Are you sure you want to change the base?

[not for merge] (heavily wip) EIP4337 & EIP7702 #1438

Uh oh!

Conversation

popzxc commented Apr 29, 2025

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

StanislavBreadless May 15, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

StanislavBreadless May 15, 2025 •

edited

Loading