core: implement eip-7981: Increase Access List Cost

[MariusVanDerWijden]

based on: #34748

spec: https://eips.ethereum.org/EIPS/eip-7981

[rjl493456442]

I am not sure how we can get rid of the overflow checking, but apparently the validation is missing here.

diff --git a/core/state_transition.go b/core/state_transition.go
index 3f7ec50f5b..e7b06a96ed 100644
--- a/core/state_transition.go
+++ b/core/state_transition.go
@@ -107,12 +107,31 @@ func IntrinsicGas(data []byte, accessList types.AccessList, authList []types.Set
 		}
 	}
 	if accessList != nil {
-		gas += uint64(len(accessList)) * params.TxAccessListAddressGas
-		gas += uint64(accessList.StorageKeys()) * params.TxAccessListStorageKeyGas
+		addresses := uint64(len(accessList))
+		storageKeys := uint64(accessList.StorageKeys())
+		if (math.MaxUint64-gas)/params.TxAccessListAddressGas < addresses {
+			return 0, ErrGasUintOverflow
+		}
+		gas += addresses * params.TxAccessListAddressGas
+		if (math.MaxUint64-gas)/params.TxAccessListStorageKeyGas < storageKeys {
+			return 0, ErrGasUintOverflow
+		}
+		gas += storageKeys * params.TxAccessListStorageKeyGas
+
 		// EIP-7981: access list data is charged in addition to the base charge.
 		if isAmsterdam {
-			gas += uint64(len(accessList)) * common.AddressLength * params.TxCostFloorPerToken7976 * params.TxTokenPerNonZeroByte
-			gas += uint64(accessList.StorageKeys()) * common.HashLength * params.TxCostFloorPerToken7976 * params.TxTokenPerNonZeroByte
+			const (
+				addressCost    = common.AddressLength * params.TxCostFloorPerToken7976 * params.TxTokenPerNonZeroByte
+				storageKeyCost = common.HashLength * params.TxCostFloorPerToken7976 * params.TxTokenPerNonZeroByte
+			)
+			if (math.MaxUint64-gas)/addressCost < addresses {
+				return 0, ErrGasUintOverflow
+			}
+			gas += addresses * addressCost
+			if (math.MaxUint64-gas)/storageKeyCost < storageKeys {
+				return 0, ErrGasUintOverflow
+			}
+			gas += storageKeys * storageKeyCost
 		}
 	}
 	if authList != nil {

[MariusVanDerWijden]

Yeah makes sense

[rjl493456442]

Please rebase and address the comments i left.

[fjl]

The overflow check here seems a bit over-protective. The overflow could only happen if len(data) is very large. But if we were worried about that happening, we would have to check for overflow also in the code above where we calculate based on len(data). Checking for overflow only here is just weird.

[rjl493456442]

LGTM