Submitted by ladboy233, also found by Breeje
src/core/Pair.sol#L56
src/core/Pair.sol#L57
core/Lendgine.sol#L252
In the current codebase, FullMath.mulDiv is used, the function takes three parameters.
Basically FullMath.mulDIv(a, b, c) means a * b / c.
Then there are some operations which incur unnecessary precision loss because of division before multiplcation.
When accruing interest, the code below:
Note the line:
This basically equals to dilutionLPRequested = (borrowRate * totalLiquidityBorrowed / 1e18 * timeElapsed) / 365 days
The first part of division can greatly truncate the value borrowRate * totalLiquidityBorrowed / 1e18, the totalLiquidityBorrowed should be normalized and scaled by token precision when adding liqudity instead of division by 1e18 here.
Same preicision loss happens when computng the invariant
scale0 = (amount0 * 1e18 / liqudiity) * token0Scale
scale1 = (amount1 * 1e18 / liqudiity) * token1Scale
Whereas the amount0 and amount1 should be first be normalized by token0Scale and token1Scale and then divided by liquidity at last. If the liquidity is a larger number  amount0 * 1e18 / liqudity is already truncated to 0.
We recommend the protocol avoid divison before multiplication and always perform division operation at last.
kyscott18 (Numoen) confirmed
For this contest, 9 reports were submitted by wardens detailing low risk and non-critical issues. The report highlighted below by CodingNameKiki received the top score from the judge.
The following wardens also submitted reports: matrix_0wl, SleepingBugs, IllIllI, 0xAgro, 0xSmartContract, btk, chrisdior4, and Rolezn.
