[pamungkaski]

Stacked on #365 — targets feat/add-max-bid-price-validation-hook, so the diff here is just this change.

The degenerate case

MaxBidPriceValidationHook(0) is a brick as written. CCA requires every bid to price strictly above the clearing price:

// ContinuousClearingAuction.sol
if (_maxPriceQ96 <= _checkpoint.clearingPrice) revert BidMustBeAboveClearingPrice();

and the clearing price starts at a floor of at least ConstantsLib.MIN_FLOOR_PRICE (2**32 + 1), rising from there. So every bid that could otherwise be placed has maxPrice > 0 and reverts with MaxBidPriceExceeded. A bid at maxPrice == 0 passes the hook and then fails on BidMustBeAboveClearingPrice. The result is an auction that can never accept a bid.

The proposal

Since nobody would deploy that deliberately, spend the degenerate value on the more useful meaning:

• Document maxBidPrice == 0 as "no effective cap" in IMaxBidPriceValidationHook, so it's part of the interface contract rather than one implementation's quirk.
• Guard the reference implementation with if (maxBidPrice != 0 && maxPrice > maxBidPrice).

For this hook it changes nothing anyone would deploy: every reachable cap is >= MIN_FLOOR_PRICE, and behavior at those values is identical. What it buys is a natural "unset" for integrators building the hook from config — today a missing/zero value yields an auction that silently accepts no bids, which is a bad failure mode for a value that's easy to leave at its default.

Tests

The BTT tree gains a when maxBidPrice is 0 branch, and the existing revert case is scoped under when maxBidPrice is not 0:

├── validate
│   ├── when maxBidPrice is 0
│   │   └── it does not revert
│   └── when maxBidPrice is not 0
│       ├── when maxPrice GT maxBidPrice
│       │   └── it reverts with {MaxBidPriceExceeded}
│       └── when maxPrice LTE maxBidPrice
│           ├── ...

Note test_WhenMaxPriceGTMaxBidPrice previously fuzzed _maxBidPrice from 0, so it would have caught this change — its lower bound moves to 1, and the zero case is asserted explicitly instead.

7/7 pass at 2500 fuzz runs. Nothing outside these three files references the hook.

Happy to drop this if you'd rather keep 0 meaning a literal cap — the interface note is the part I think is worth having either way, since the current behavior at 0 is surprising enough to deserve a line of natspec.

[zhongeric]

thank you!