Submitted by cccz
In LinearDistributor.declareReward, distributed represents the reward to distribute and is calculated using netVest(currentlyVested - previouslyVested).
At the same time, distributed cannot exceed balance, which means that if balance < linearBondedValue /ast netVest / vestingBondedValue, part of the rewards in netVest will be lost.
At the end of the function, previouslyVested is directly assigned to currentlyVested instead of using the Vested adjusted according to distributed, which means that the previously lost rewards will also be skipped in the next distribution.
Also, in the next distribution, bufferRequirement will be small because distributed is small, so it may increase the number of forfeits.
https://github.com/code-423n4/2023-02-malt/blob/700f9b468f9cf8c9c5cffaa1eba1b8dea40503f9/contracts/RewardSystem/LinearDistributor.sol#L111-L153
Consider adapting previouslyVested based on distributed:
0xScotch (Malt) confirmed and commented:
