Burning our stablecoins before redeeming in fuzz testing

[Podziuu]

Hello,

While writing fuzz tests for the Stablecoin project, I discovered an edge case where the redeemCollateral function reverts due to the user’s health factor being broken. Upon investigation, I found that the user had minted some coins, and the bound function selected almost the maximum amount to redeem collateral. This caused the revert, which is expected behavior.

To handle this, I modified the redeemCollateral function to pass all cases. Here is the updated function:

function redeemCollateral(uint256 collateralSeed, uint256 amountCollateral) public {
        ERC20Mock collateral = _getCollateralFromSeed(collateralSeed);
        uint256 maxAmountToRedeem = engine.getCollateralBalance(msg.sender, address(collateral));
        (uint256 mintedPdn,) = engine.getAccountInformation(msg.sender);

        amountCollateral = bound(amountCollateral, 0, maxAmountToRedeem);

        if (amountCollateral == 0) {
            return;
        }
        vm.startPrank(msg.sender);
        if (mintedPdn > 0) { // burn all minted coins to secure redeeming
            pdn.approve(address(engine), mintedPdn);
            engine.burnPdn(mintedPdn);
        }
        engine.redeemCollateral(address(collateral), amountCollateral);
        vm.stopPrank();
    }

Is this a good approach, or am I missing something?

[EngrPips]

Hey @Podziuu, This seems like a good approach. We usually want our invariant test suite to break our protocol in a way that we don't foresee, so we should write the invariant test suite to prevent all reverts we know of.

[Podziuu]

Thanks for the confirmation @EngrPips

[EngrPips]

You are welcome, Buddy. Keep crushing it.