Submitted by bin2chen, also found by ladboy233 and said
Currently, there are two ways to retrieve Liquidity
No matter which one, if there is a profit in the end, it needs to be refunded to the borrower.
In this way, if the borrower enters the token0 or token1 blacklist, such as USDC and the token always has a profit, then refund() -> TransferHelper.safeTransfer() will definitely revert, causing _closePositon() to always revert, and LPs Liquidity is locked in the contract.
If the borrower enters the token blacklist and always has  a profit, LP may not be able to retrieve Liquidity.
Add a new claims[token] mechanism.
If refund()-> transfer() fails, record claims[token]+= (amountExpected - amountActual).
And provide methods to support borrower to claim().
wukong-particle (Particle) commented:
0xleastwood (Judge) commented:
wukong-particle (Particle) confirmed, but disagreed with severity and commented:
0xleastwood (Judge) commented:
wukong-particle (Particle) commented:
However, after revising the contract, our team decides to skip this black issue for now for 2 reasons. 
(1) this should be a rare situation: a borrower was not in a blacklist, opens a position, then enters the blacklist before closing/liquidating the position. If this really happens once or twice, our protocol will have an insurance fund to repay the LP. If we observe it to be a frequently occurred issue, we will upgrade the contract with the suggested try-catch-claim pattern. 
(2) the liquidator can control the amountSwap to reduce the refund amount of blacklisted token to 0, only refunding the other token (basically skipping the swap for this token). This cant solve the problem if both tokens blacklist the borrower. 
So we will update the tag to sponsor-acknowledged. And we maintain that this shouldnt be a high severity issue since it should happen in very rare situation and s not widespread. Thanks!
