Fix forced exit re-use #87
Conversation
test/zkbob/ZkBobPool.t.sol
Outdated
|
|
||
| skip(24 hours); | ||
|
|
||
| pool.cancelForcedExit(nullifier, user2, user2, 0.4 ether / D / denominator, exitStart, exitEnd); |
There was a problem hiding this comment.
Should it also be tested that executeForcedExit is not possible after cancelation?
There was a problem hiding this comment.
Since it will be already expired by then, it will be rejected anyway at the timestamp check, so this test won't add any new coverage.
There was a problem hiding this comment.
Got you, thanks! If it is assumed that the cancelation happens in order to commit the proof one more time, does it make sense to extend the method with the proof and if the proof/index/out_commit is not empty, the force exit will be re-commited? It will allow users to save on gas fees a bit.
There was a problem hiding this comment.
I re-worked code a bit, so it's no longer that simple. But anyway I think it should be possible to make cancel and commit in the same transaction using any multicall available on-chain, since they both do not depend on msg.sender.
k1rill-fedoseev
force-pushed
the
fix/forced-exit
branch
from
91e9c1c
to
6f35186
Compare
Previous implementation contained a bug, allowing anyone to infinitely resubmit the same forced exit proof again and again, effectively preventing exiting party from ever finalizing the withdrawal.
One way to fix that - disallow forced exit submission for nullifiers for which forced exists already exist. However, in order for exiting party to still be able to re-submit their forced exit after 24 hours, extra cancellation method is needed.