feat: E2E gas profiling native and erc20 by shamit05 · Pull Request #122 · coinbase/smart-wallet

shamit05 · 2025-09-05T17:40:14Z

Gas profiling done for native and erc20 transfers. Results in snapshot folder.
https://docs.google.com/document/d/1G-ReRP_SQ4SvbHSjytj_ubdRwXTTMTdlwPAYSyr7LF4/edit?tab=t.0

cb-heimdall · 2025-09-05T17:40:18Z

🟡 Heimdall Review Status

Requirement Status More Info

Reviews

🟡 0/1

Denominator calculation

Show calculation

1 if user is bot 0

1 if user is external 0

2 if repo is sensitive 0

From .codeflow.yml 1

Additional review requirements

Show calculation

Max	0

0

From CODEOWNERS 0

Global minimum 0

Max 1

1

1 if commit is unverified 1

Sum 2

shamit05 · 2025-09-05T17:41:29Z

Addressing this previous comment:

does _getUserOpWithSignature use userOpCalldata as calldata for the op?

Yes, it does since it calls internal _getUserOp which uses userOpCalldata.

ilikesymmetry

Good pass, handful of comments on cleaning up the conventions

ilikesymmetry · 2025-09-05T18:40:09Z

test/gas/EndToEnd.t.sol

+    function test_transfer_native() public {
+        userOpCalldata = abi.encodeCall(CoinbaseSmartWallet.execute, (address(0x1234), 1 ether, ""));


Recommend fuzzing in general which can remove hardcoded values.

Also there's an increased cost when updating the storage for a users balance from zero to nonzero so we can cover for that by dusting the address first.

Suggested change

function test_transfer_native() public {

userOpCalldata = abi.encodeCall(CoinbaseSmartWallet.execute, (address(0x1234), 1 ether, ""));

function test_transfer_native(address recipient, uint256 amount) public {

vm.assume(amount < 1 ether);

// Dust recipient to control for gas increase for first non-zero balance

vm.deal(recipient, 1 wei);

usdc.mint(eoaUser, 1 wei);

// Prepare userOp

userOpCalldata = abi.encodeCall(CoinbaseSmartWallet.execute, (recipient, amount, "");

ilikesymmetry · 2025-09-05T18:41:43Z

test/gas/EndToEnd.t.sol

+        bytes memory handleOpsCalldata = abi.encodeCall(entryPoint.handleOps, (_makeOpsArray(op), payable(bundler)));
+        console2.log("test_transfer_native ERC-4337 calldata size:", handleOpsCalldata.length);
+
+        vm.startSnapshotGas("e2e_transfer_native_4337");


naming nit

Suggested change

vm.startSnapshotGas("e2e_transfer_native_4337");

vm.startSnapshotGas("e2e_transfer_native_baseAccount");

ilikesymmetry · 2025-09-05T18:42:11Z

test/gas/EndToEnd.t.sol

+
+        vm.prank(eoaUser);
+        vm.startSnapshotGas("e2e_transfer_native_eoa");
+        payable(address(0x1234)).transfer(1 ether);


if using fuzzing from earlier comment

Suggested change

payable(address(0x1234)).transfer(1 ether);

payable(address(recipient)).transfer(1 ether);

ilikesymmetry · 2025-09-05T18:42:37Z

test/gas/EndToEnd.t.sol

+        UserOperation memory op = _getUserOpWithSignature();
+
+        bytes memory handleOpsCalldata = abi.encodeCall(entryPoint.handleOps, (_makeOpsArray(op), payable(bundler)));
+        console2.log("test_transfer_native ERC-4337 calldata size:", handleOpsCalldata.length);


also thinking we should log the calldata size for eoa execution as well for easy diff

ilikesymmetry · 2025-09-05T18:46:52Z

test/gas/EndToEnd.t.sol

+    function test_transfer_erc20() public {
+        userOpCalldata = abi.encodeCall(
+            CoinbaseSmartWallet.execute,
+            (address(usdc), 0, abi.encodeCall(usdc.transfer, (address(0x5678), 100e6)))
+        );


same thing here as previous function with:

fuzzing instead of hardcoding values

dust recipients to remove variability from zero->nonzero storage transition

ilikesymmetry · 2025-09-05T18:47:18Z

test/gas/EndToEnd.t.sol

+        UserOperation memory op = _getUserOpWithSignature();
+
+        bytes memory handleOpsCalldata = abi.encodeCall(entryPoint.handleOps, (_makeOpsArray(op), payable(bundler)));
+        console2.log("test_transfer_erc20 ERC-4337 calldata size:", handleOpsCalldata.length);


also add eoa calldata size for comparison

ilikesymmetry · 2025-09-05T18:47:26Z

test/gas/EndToEnd.t.sol

+        bytes memory handleOpsCalldata = abi.encodeCall(entryPoint.handleOps, (_makeOpsArray(op), payable(bundler)));
+        console2.log("test_transfer_erc20 ERC-4337 calldata size:", handleOpsCalldata.length);
+
+        vm.startSnapshotGas("e2e_transfer_erc20_4337");


Suggested change

vm.startSnapshotGas("e2e_transfer_erc20_4337");

vm.startSnapshotGas("e2e_transfer_erc20_baseAccount");

ilikesymmetry · 2025-09-05T18:47:40Z

test/gas/EndToEnd.t.sol

+
+        vm.prank(eoaUser);
+        vm.startSnapshotGas("e2e_transfer_erc20_eoa");
+        usdc.transfer(address(0x5678), 100e6);


use fuzzing

Suggested change

usdc.transfer(address(0x5678), 100e6);

usdc.transfer(recipient, amount);

ilikesymmetry · 2025-09-05T18:48:31Z

test/gas/EndToEnd.t.sol

+        return ops;
+    }
+
+    function _sign(UserOperation memory userOp) internal view override returns (bytes memory signature) {


Not seeing where this is being used. Does the parent test contract require use to override this?

ilikesymmetry · 2025-09-05T18:49:25Z

foundry.lock

Think it would be good to separate the lockfile and lib/forge-std diff into its own tiny PR for clarity and merge that first. Might not be worth the effort though, but generally touching build stuff is high enough importance to be isolated.

ilikesymmetry

Thanks for adding benchmarks :)

cb-heimdall · 2025-09-08T20:10:33Z

Review Error for ilikesymmetry @ 2025-09-08 20:10:32 UTC
User failed mfa authentication, see go/mfa-help

cb-heimdall · 2025-09-08T22:34:52Z

Review Error for ilikesymmetry @ 2025-09-08 22:34:51 UTC
User failed mfa authentication, see go/mfa-help

shamit05 added 4 commits September 4, 2025 10:37

init

edc248b

updates

da81d34

fixes

9ba2ff1

erc20 and native transfer

1e99582

ilikesymmetry reviewed Sep 5, 2025

View reviewed changes

done

b51effe

ilikesymmetry self-requested a review September 8, 2025 20:09

ilikesymmetry approved these changes Sep 8, 2025

View reviewed changes

shamit05 marked this pull request as ready for review September 8, 2025 20:09

ilikesymmetry self-requested a review September 8, 2025 22:34

ilikesymmetry approved these changes Sep 8, 2025

View reviewed changes

shamit05 closed this Sep 8, 2025

		function test_transfer_native() public {
		userOpCalldata = abi.encodeCall(CoinbaseSmartWallet.execute, (address(0x1234), 1 ether, ""));

-    function test_transfer_native() public {
-        userOpCalldata = abi.encodeCall(CoinbaseSmartWallet.execute, (address(0x1234), 1 ether, ""));
+    function test_transfer_native(address recipient, uint256 amount) public {
+        vm.assume(amount < 1 ether);
+        // Dust recipient to control for gas increase for first non-zero balance
+        vm.deal(recipient, 1 wei);
+        usdc.mint(eoaUser, 1 wei);
+        // Prepare userOp
+        userOpCalldata = abi.encodeCall(CoinbaseSmartWallet.execute, (recipient, amount, "");

	vm.startSnapshotGas("e2e_transfer_native_4337");
	vm.startSnapshotGas("e2e_transfer_native_baseAccount");

	payable(address(0x1234)).transfer(1 ether);
	payable(address(recipient)).transfer(1 ether);

	vm.startSnapshotGas("e2e_transfer_erc20_4337");
	vm.startSnapshotGas("e2e_transfer_erc20_baseAccount");

	usdc.transfer(address(0x5678), 100e6);
	usdc.transfer(recipient, amount);

Conversation

shamit05 commented Sep 5, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

cb-heimdall commented Sep 5, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

🟡 Heimdall Review Status

Uh oh!

shamit05 commented Sep 5, 2025

Uh oh!

ilikesymmetry left a comment

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!

ilikesymmetry left a comment

Choose a reason for hiding this comment

Uh oh!

cb-heimdall commented Sep 8, 2025

Uh oh!

cb-heimdall commented Sep 8, 2025

Uh oh!

Reviewers

Assignees

Labels

Milestone

Development

Uh oh!

3 participants

shamit05 commented Sep 5, 2025 •

edited

Loading

cb-heimdall commented Sep 5, 2025 •

edited

Loading