improve: Handle deposits referencing input/output/repayment tokens that do not map to PoolRebalanceRoutes #970
Conversation
…ap to PoolRebalanceRoutes Prerequisite changes needed to prepare for SpokePool that do not enforce deposit route whitelist checking The changes include: - changing the bundle data client's choice of repayment chain for a deposit where the inputToken and originChain do not map to a PoolRebalanceRoute - building running balances based on refunds for a token that isn't mapped to a pool rebalance route
nicholaspai
changed the title
nicholaspai
changed the title
…ation chain only if changing refund address to msg.sender and there are valid pool rebalance routes
| @@ -372,7 +373,8 @@ export class BundleDataClient { | |||
| if ( | |||
| fill.blockNumber < blockRanges[chainIndex][0] || | |||
| fill.blockNumber > blockRanges[chainIndex][1] || | |||
| isZeroValueFillOrSlowFillRequest(fill) | |||
| isZeroValueFillOrSlowFillRequest(fill) || | |||
| invalidOutputToken(fill) | |||
There was a problem hiding this comment.
What's the intention with this check? invalidOutputToken() is only checking against 0x0, but no fill should ever set that.
There was a problem hiding this comment.
My thinking was that technically you can send a fill with an output token equal to the zero address (contract allows it) and now it can match with a deposit with the zero address-- and we don't want that. Thoughts?
There was a problem hiding this comment.
Should this check even be in getApproximateRefundsForBlockRange? Would this not filter out valid deposits with 0x0 as the output token?
There was a problem hiding this comment.
@bmzig my thinking was that in getApproximateRefundsForBlockRange like in this loadDataFromScratch function, there is a possibility to match a fill with an outputToken = 0x0 with a deposit.
So that's why this check exists there now. WDYT?
There was a problem hiding this comment.
Yep. That makes sense. I think we do want this here, then.
src/utils/DepositUtils.ts
Outdated
| @@ -208,6 +208,11 @@ export function isZeroValueDeposit(deposit: Pick<Deposit, "inputAmount" | "messa | |||
| return deposit.inputAmount.eq(0) && isMessageEmpty(deposit.message); | |||
| } | |||
|
|
|||
| export function invalidOutputToken(deposit: Pick<DepositWithBlock, "outputToken">): boolean { | |||
| // If the output token is zero address, then it is invalid. | |||
| return deposit.outputToken === ZERO_ADDRESS; | |||
There was a problem hiding this comment.
Should this be ZERO_BYTES ? I think fwiw we already have a similar function (isZeroAddress()) used by the SpokePoolClient to determine whether it needs to sub in a valid address.
There was a problem hiding this comment.
This looks good to me. I couldn't see anything wrong. A minor note: it seems like getL1TokenForDeposit and getL2TokenForDeposit will break with the contract upgrade, but the only time we use it is when calculating lp fees, which is addressed by the changes here.
|
Yeah I think this PR catches all instances where we use getL1TokenForDeposit and getL2TokenForDeposit but that's definitely a worry of mine. Thanks for reviewing |
Prerequisite changes needed to prepare for SpokePool that do not enforce deposit route whitelist checking. In this upcoming paradigm, deposits can be sent where the
inputTokendoes not correspond to aPoolRebalanceRoutemapping. Implements this UMIP changeThis change should be backwards compatible as all tokens that can be deposited are mapped to PoolRebalanceRoutes
The changes include:
outputToken = 0x0but there are not sufficient PoolRebalanceRoutes to resolve the outputToken