optimize gas usage with common techniques and refactor code structure #341

alex0207s · 2025-02-03T08:43:39Z

Enable the via-ir setting.
Increase optimizer_runs from 1,000 to 65,536.
Remove unused function parameters in the executeStrategy function.
Remove the logic that leaves a one-wei gap between operations in the SmartOrderStrategy contract, as it introduces unnecessary overhead.
Rewrite the getEIP712Hash function of EIP-712 using assembly code.
Refactor some interfaces to comply with linter rules.
Upgrade CI packages.

github-actions · 2025-02-03T15:15:48Z

Changes to gas cost

Generated at commit: 602387b586382295873ca62e3691488e02414418, compared to commit: b01d556be855a7ca334957b0e62ed27aa38a7b7e

🧾 Summary (20% most significant diffs)

Contract	Method	Avg (+/-)	%
AllowanceTarget	spendFromUserTo unpause	-1,176 ✅ -183 ✅	-2.29% -0.72%

Full diff report 👇

Contract	Deployment Cost (+/-)	Method	Min (+/-)	%	Avg (+/-)	%	Median (+/-)	%	Max (+/-)	%	# Calls (+/-)
AllowanceTarget	643,922 (+38,965)	pause spendFromUserTo unpause	23,485 (+65) 24,715 (-310) 23,375 (-46)	+0.28% -1.24% -0.20%	26,545 (-60) 50,069 (-1,176) 25,350 (-183)	-0.23% -2.29% -0.72%	27,565 (-102) 48,497 (-1,330) 25,350 (-183)	-0.37% -2.67% -0.72%	27,565 (-102) 89,862 (-1,714) 27,326 (-319)	-0.37% -1.87% -1.15%	4 (0) 8 (0) 2 (0)

github-actions · 2025-02-03T15:15:49Z

Changes to gas cost

Generated at commit: 602387b586382295873ca62e3691488e02414418, compared to commit: b01d556be855a7ca334957b0e62ed27aa38a7b7e

🧾 Summary (20% most significant diffs)

Contract	Method	Avg (+/-)	%
RFQ	feeCollector owner	+2,212 ❌ +2,101 ❌	+580.58% +521.34%
LimitOrderSwap	feeCollector isOrderCanceled orderHashToMakerTokenFilledAmount owner	+2,256 ❌ +2,234 ❌ +1,926 ❌ +2,123 ❌	+559.80% +443.25% +409.79% +526.80%
CoordinatedTaker	EIP712_DOMAIN_SEPARATOR coordinator owner	+314 ❌ +1,891 ❌ +2,179 ❌	+101.62% +479.95% +527.60%

Full diff report 👇

Contract	Deployment Cost (+/-)	Method	Min (+/-)	%	Avg (+/-)	%	Median (+/-)	%	Max (+/-)	%	# Calls (+/-)
RFQ	2,628,391 (+208,452)	allowanceTarget cancelRFQOffer feeCollector fillRFQ fillRFQWithSig owner permit2 setFeeCollector weth	424 (+164) 24,027 (+34) 2,593 (+2,212) 31,518 (-291) 33,812 (-53) 2,504 (+2,101) 182 (-102) 23,854 (+64) 270 (+9)	+63.08% +0.14% +580.58% -0.91% -0.16% +521.34% -35.92% +0.27% +3.45%	424 (+164) 35,906 (-19) 2,593 (+2,212) 128,429 (-1,873) 111,785 (-1,579) 2,504 (+2,101) 182 (-102) 26,019 (+58) 270 (+9)	+63.08% -0.05% +580.58% -1.44% -1.39% +521.34% -35.92% +0.22% +3.45%	424 (+164) 27,978 (+9) 2,593 (+2,212) 145,801 (-2,081) 76,771 (-866) 2,504 (+2,101) 182 (-102) 24,069 (+74) 270 (+9)	+63.08% +0.03% +580.58% -1.41% -1.12% +521.34% -35.92% +0.31% +3.45%	424 (+164) 49,775 (-73) 2,593 (+2,212) 236,417 (-2,380) 224,772 (-3,820) 2,504 (+2,101) 182 (-102) 30,134 (+35) 270 (+9)	+63.08% -0.15% +580.58% -1.00% -1.67% +521.34% -35.92% +0.12% +3.45%	1 (0) 5 (0) 2 (0) 24 (0) 3 (0) 1 (0) 1 (0) 3 (0) 1 (0)
LimitOrderSwap	2,697,322 (+17,207)	allowanceTarget cancelOrder feeCollector fillLimitOrder fillLimitOrderFullOrKill fillLimitOrderGroup isOrderCanceled orderHashToMakerTokenFilledAmount owner permit2 setFeeCollector weth	446 (+141) 29,491 (+237) 2,659 (+2,256) 35,980 (-104) 47,919 (-345) 31,149 (-10) 2,738 (+2,234) 2,396 (+1,926) 2,526 (+2,123) 182 (-102) 23,898 (+108) 292 (+9)	+46.23% +0.81% +559.80% -0.29% -0.71% -0.03% +443.25% +409.79% +526.80% -35.92% +0.45% +3.18%	446 (+141) 40,897 (+233) 2,659 (+2,256) 1,177,683 (-1,721) 102,721 (-1,271) 172,112 (-2,492) 2,738 (+2,234) 2,396 (+1,926) 2,526 (+2,123) 182 (-102) 26,063 (+102) 292 (+9)	+46.23% +0.57% +559.80% -0.15% -1.22% -1.43% +443.25% +409.79% +526.80% -35.92% +0.39% +3.18%	446 (+141) 34,161 (+311) 2,659 (+2,256) 158,698 (-2,970) 47,919 (-345) 195,696 (-2,421) 2,738 (+2,234) 2,396 (+1,926) 2,526 (+2,123) 182 (-102) 24,113 (+118) 292 (+9)	+46.23% +0.92% +559.80% -1.84% -0.71% -1.22% +443.25% +409.79% +526.80% -35.92% +0.49% +3.18%	446 (+141) 52,957 (+204) 2,659 (+2,256) 29,534,754 (-123) 212,327 (-3,122) 287,118 (-4,631) 2,738 (+2,234) 2,396 (+1,926) 2,526 (+2,123) 182 (-102) 30,178 (+79) 292 (+9)	+46.23% +0.39% +559.80% -0.00% -1.45% -1.59% +443.25% +409.79% +526.80% -35.92% +0.26% +3.18%	8 (0) 7 (0) 8 (0) 28 (0) 3 (0) 9 (0) 1 (0) 1 (0) 8 (0) 8 (0) 3 (0) 8 (0)
CoordinatedTaker	2,404,834 (+84,175)	EIP712_DOMAIN_SEPARATOR allowanceTarget approveTokens coordinator limitOrderSwap owner permit2 setCoordinator submitLimitOrderFill weth	623 (+314) 534 (+263) 26,822 (-267) 2,285 (+1,891) 556 (+263) 2,592 (+2,179) 226 (-46) 23,898 (+158) 32,910 (-170) 314 (+20)	+101.62% +97.05% -0.99% +479.95% +89.76% +527.60% -16.91% +0.67% -0.51% +6.80%	623 (+314) 534 (+263) 166,962 (-2,228) 2,285 (+1,891) 556 (+263) 2,592 (+2,179) 226 (-46) 26,055 (+148) 126,645 (-1,729) 314 (+20)	+101.62% +97.05% -1.32% +479.95% +89.76% +527.60% -16.91% +0.57% -1.35% +6.80%	623 (+314) 534 (+263) 186,496 (-2,462) 2,285 (+1,891) 556 (+263) 2,592 (+2,179) 226 (-46) 24,101 (+162) 69,687 (-333) 314 (+20)	+101.62% +97.05% -1.30% +479.95% +89.76% +527.60% -16.91% +0.68% -0.48% +6.80%	623 (+314) 534 (+263) 186,496 (-2,462) 2,285 (+1,891) 556 (+263) 2,592 (+2,179) 226 (-46) 30,166 (+123) 252,405 (-4,068) 314 (+20)	+101.62% +97.05% -1.30% +479.95% +89.76% +527.60% -16.91% +0.41% -1.59% +6.80%	2 (0) 1 (0) 15 (0) 2 (0) 1 (0) 1 (0) 1 (0) 3 (0) 7 (0) 1 (0)
GenericSwap	1,726,929 (+5,457)	allowanceTarget executeSwap executeSwapWithSig permit2	314 (+54) 32,036 (-277) 39,061 (-446) 182 (-79)	+20.77% -0.86% -1.13% -30.27%	314 (+54) 109,248 (-8,370) 165,967 (-2,387) 182 (-79)	+20.77% -7.12% -1.42% -30.27%	314 (+54) 97,813 (-17,498) 172,667 (-2,409) 182 (-79)	+20.77% -15.17% -1.38% -30.27%	314 (+54) 248,141 (-3,768) 279,472 (-4,284) 182 (-79)	+20.77% -1.50% -1.51% -30.27%	1 (0) 13 (+1) 4 (0) 1 (0)
SmartOrderStrategy	1,163,069 (-36,334)	approveTokens executeStrategy	55,489 (-773) 22,468 (-758)	-1.37% -3.26%	173,320 (-11,682) 144,361 (+22,115)	-6.31% +18.09%	264,958 (-3,717) 148,816 (+58,046)	-1.38% +63.95%	264,970 (-3,717) 607,558 (-2,615)	-1.38% -0.43%	32 (-1) 14 (-4)

NIC619 · 2025-02-05T03:46:42Z

contracts/abstracts/EIP712.sol

+        assembly {
+            // Compute the digest.
+            mstore(0x00, 0x1901000000000000000000000000000000000000000000000000000000000000) // Store "\x19\x01".
+            mstore(0x2, digest) // Store the domain separator.


Maybe align with other hex values:

Suggested change

mstore(0x2, digest) // Store the domain separator.

mstore(0x02, digest) // Store the domain separator.

NIC619 · 2025-02-05T03:48:12Z

contracts/abstracts/EIP712.sol

+        digest = _getDomainSeparator();
+
+        // solhint-disable no-inline-assembly
+        assembly {


Are those inline assemblies copied from some standard library? How much gas does it save?

Is it possible to use OZ's 712 implementation?

Are those inline assemblies copied from some standard library?

The assembly code was referenced from soladay's implementaion and 1inch's implementation (line 1204~1209)

How much gas does it save?

Basically, it saves 135 gas when testing the getEIP712Hash function. However, the main value of using assembly here is that it helps avoid quadratic memory expansion costs, as we reuse the same memory slot for hashing.

reference: https://x.com/optimizoor/status/1825913380244435408?mx=2

Is it possible to use OZ's 712 implementation?

Our implementation declares some variables as public, while the OZ's implementation declares them as private. I'm not sure whether this difference might cause issues in the front-end or back-end. I will switch to OZ's (or slady) implementation after verifying that it does not break anything on either end

Perhaps add a comment referencing the code copied from

NIC619 · 2025-02-05T03:58:43Z

test/libraries/SignatureValidator.t.sol

        assertFalse(SignatureValidator.validateSignature(vm.addr(walletAdminPrivateKey), otherDigest, signature));
    }

+    /// forge-config: default.allow_internal_expect_revert = true


What is the setting for? Where is it checked?

Same reason as mentioned above.

Add this config setting to resolve the following error:
[FAIL: call didn't revert at a lower depth than cheatcode call depth]

see this: https://book.getfoundry.sh/announcements#-notes

NIC619 · 2025-02-05T04:00:34Z

test/forkMainnet/LimitOrderSwap/Fill.t.sol

        {
            bytes memory makerSpecificData = abi.encode(defaultAMMPath);
-            bytes memory strategyData = abi.encode(UNISWAP_SWAP_ROUTER_02_ADDRESS, makerSpecificData);
+            bytes memory strategyData = abi.encode(UNISWAP_SWAP_ROUTER_02_ADDRESS, order.makerToken, order.makerTokenAmount - fee, makerSpecificData);


Why added these two parameters? Is it because the change of executeStrategy interface?

Yes, the strategyData here is for the mockLimitOrderTaker, which inherits an interface that defines the executeStrategy function. To minimize breaking changes for the mock contract, we simply encode these two parameters in strategyData.

NIC619 · 2025-02-05T04:01:13Z

Can you still run forge coverage after enabled via-ir?

alex0207s · 2025-02-06T02:43:56Z

Can you still run forge coverage after enabled via-ir?

Yes, it works fine after enabling the via-ir setting.

alex0207s · 2025-02-03T10:19:11Z

contracts/SmartOrderStrategy.sol

        if (ops.length == 0) revert EmptyOps();

-        // wrap ETH to WETH if inputToken is ETH
-        if (Asset.isETH(inputToken)) {


This if statement duplicates the logic in the GenericSwap contract: https://github.com/consenlabs/tokenlon-contracts/blob/v6.0.1/contracts/GenericSwap.sol#L68-L75

alex0207s · 2025-02-03T10:40:06Z

contracts/SmartOrderStrategy.sol

-            if (msg.value != inputAmount) revert InvalidMsgValue();
-            // the coverage report indicates that the following line causes this branch to not be covered by our tests
-            // even though we tried all possible success and revert scenarios
-            IWETH(weth).deposit{ value: inputAmount }();


We don't wrap ETH directly, as it could prevent SmartOrderStrategy from interacting with functions that require ETH.

alex0207s · 2025-02-03T10:42:29Z

contracts/abstracts/EIP712.sol

 /// @dev This contract provides functions to handle EIP-712 domain separator and hash calculation.
 abstract contract EIP712 {
-    // EIP-191 Header
-    string public constant EIP191_HEADER = "\x19\x01";


This variable is no longer needed after rewriting the getEIP712Hash function

alex0207s · 2025-02-03T10:49:13Z

contracts/libraries/Asset.sol

-            if (!success) {
-                assembly {
-                    revert(add(result, 32), mload(result))
+        if (amount > 0) {


Using nested if statements is more gas-efficient.

alex0207s · 2025-02-03T10:56:46Z

test/libraries/SignatureValidator.t.sol

        assertFalse(SignatureValidator.validateSignature(vm.addr(walletAdminPrivateKey), otherDigest, signature));
    }

+    /// forge-config: default.allow_internal_expect_revert = true


Add this config setting to resolve the following error:
[FAIL: call didn't revert at a lower depth than cheatcode call depth]

see this: https://book.getfoundry.sh/announcements#-notes

alex0207s · 2025-02-03T13:04:54Z

foundry.toml

 optimizer = true # enable or disable the solc optimizer
-optimizer_runs = 1000 # the number of optimizer runs
+optimizer_runs = 65536 # the number of optimizer runs
+via_ir = true # enable or disable the compilation pipeline for the new IR optimizer


Enabling via-IR: true improves Solidity compilation by introducing an Intermediate Representation (IR) pipeline. This enhances bytecode efficiency, reduces gas costs, and optimizes contract size. It also makes compilation more predictable and future-proof, aligning with Solidity's long-term roadmap

alex0207s · 2025-02-04T08:13:23Z

test/forkMainnet/SmartOrderStrategy/IntegrationV6.t.sol

        bytes memory opsData = abi.encode(operations);

        vm.startPrank(genericSwap);
-        vm.deal(address(smartOrderStrategy), rfqOffer.takerTokenAmount);


The deal cheatcode attempts to mock GenericSwap transferring takerToken to the SmartOrderStrategy contract, but since takerToken is ETH, which is sent during executeStrategy, deal should not be used here.

alex0207s · 2025-02-06T02:18:45Z

test/libraries/SignatureValidator.t.sol

        assertFalse(SignatureValidator.validateSignature(vm.addr(walletAdminPrivateKey), otherDigest, signature));
    }

+    /// forge-config: default.allow_internal_expect_revert = true


Same reason as mentioned above.

Add this config setting to resolve the following error:
[FAIL: call didn't revert at a lower depth than cheatcode call depth]

see this: https://book.getfoundry.sh/announcements#-notes

alex0207s · 2025-02-06T02:40:52Z

test/forkMainnet/LimitOrderSwap/Fill.t.sol

        {
            bytes memory makerSpecificData = abi.encode(defaultAMMPath);
-            bytes memory strategyData = abi.encode(UNISWAP_SWAP_ROUTER_02_ADDRESS, makerSpecificData);
+            bytes memory strategyData = abi.encode(UNISWAP_SWAP_ROUTER_02_ADDRESS, order.makerToken, order.makerTokenAmount - fee, makerSpecificData);


Yes, the strategyData here is for the mockLimitOrderTaker, which inherits an interface that defines the executeStrategy function. To minimize breaking changes for the mock contract, we simply encode these two parameters in strategyData.

alex0207s · 2025-02-06T03:38:33Z

contracts/abstracts/EIP712.sol

+        digest = _getDomainSeparator();
+
+        // solhint-disable no-inline-assembly
+        assembly {


Are those inline assemblies copied from some standard library?

The assembly code was referenced from soladay's implementaion and 1inch's implementation (line 1204~1209)

How much gas does it save?

Basically, it saves 135 gas when testing the getEIP712Hash function. However, the main value of using assembly here is that it helps avoid quadratic memory expansion costs, as we reuse the same memory slot for hashing.

reference: https://x.com/optimizoor/status/1825913380244435408?mx=2

Is it possible to use OZ's 712 implementation?

Our implementation declares some variables as public, while the OZ's implementation declares them as private. I'm not sure whether this difference might cause issues in the front-end or back-end. I will switch to OZ's (or slady) implementation after verifying that it does not break anything on either end

…n assembly

alex0207s added 4 commits February 3, 2025 16:38

optimize gas usage with common techniques and refactor code structure

f337815

add forge config to allow internal expect revert in tests

d0674a1

upgrade checkout from v3 to v4 to resolve CI issues

f3f2162

fix gas diff ci

0eaf592

alex0207s requested a review from NIC619 February 4, 2025 08:15

add error handling for zero swap amount in GenericSwap

ca01cee

NIC619 reviewed Feb 5, 2025

View reviewed changes

alex0207s commented Feb 18, 2025

View reviewed changes

alex0207s added 2 commits April 8, 2025 15:41

refine by PR comments

9e87299

add references in comments for EIP-712 hash calculation implemented i…

da74666

…n assembly

NIC619 approved these changes Apr 9, 2025

View reviewed changes

alex0207s merged commit c3ef880 into master Apr 10, 2025
2 checks passed

	mstore(0x2, digest) // Store the domain separator.
	mstore(0x02, digest) // Store the domain separator.

optimize gas usage with common techniques and refactor code structure #341

optimize gas usage with common techniques and refactor code structure #341

Uh oh!

Conversation

alex0207s commented Feb 3, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

github-actions bot commented Feb 3, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Changes to gas cost

🧾 Summary (20% most significant diffs)

Uh oh!

github-actions bot commented Feb 3, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Changes to gas cost

🧾 Summary (20% most significant diffs)

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!

NIC619 commented Feb 5, 2025

Uh oh!

alex0207s commented Feb 6, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

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!

Reviewers

Assignees

Labels

Milestone

Development

Uh oh!

3 participants

alex0207s commented Feb 3, 2025 •

edited

Loading

github-actions bot commented Feb 3, 2025 •

edited

Loading

github-actions bot commented Feb 3, 2025 •

edited

Loading

alex0207s commented Feb 6, 2025 •

edited

Loading