EIP-7934 #3376
EIP-7934 #3376
Conversation
|
The same validation if applied to txpool can be tricky. |
| @@ -159,9 +159,13 @@ proc assembleBlock*( | |||
| blobsBundle = BlobsBundle( | |||
| wrapperVersion: getWrapperVersion(com, blk.header.timestamp) | |||
| ) | |||
| currentRlpSize = rlp.getEncodedLength(blk.header) + rlp.getEncodedLength(blk.withdrawals) | |||
There was a problem hiding this comment.
nimbus-eth1/execution_chain/core/tx_pool.nim
Line 178 in 6ce4568
withdrawals are fork dependant.
|
|
||
| for item in pst.packedTxs: | ||
| let tx = item.pooledTx | ||
| if currentRlpSize > MAX_RLP_BLOCK_SIZE: | ||
| break | ||
| currentRlpSize = currentRlpSize + rlp.getEncodedLength(tx.tx) |
There was a problem hiding this comment.
Calculating the rlp length like this might works to some extent. But considering how rlp algorithm works, there are and there will be edge cases where this calculation will produce invalid result. To produce a reliable result for all cases, we need a specially crafted getEncodedLength or we make a compromise to never exceeds the dangerous limit.
And that dangerous limit is MAX_RLP_BLOCK_SIZE - 5. The 5 represent the maximum number of bytes might be produced by the rlp for the length prefix which is 3 bytes for 10 MiB, and additional 2 bytes to compensate for missing rlp list marker when we skip the body. (This is my rough calculation, I may missed something), but you got the point.
And also a break from that loop must be executed before currentRlpSize exceeds MAX_RLP_BLOCK_SIZE - X.
No description provided.