tx_memory_pool: quick abort on block templates for too-young FCMP txs#220
tx_memory_pool: quick abort on block templates for too-young FCMP txs#220jeffro256 wants to merge 2 commits intoseraphis-migration:fcmp++-stagefrom
Conversation
src/cryptonote_core/tx_pool.cpp
Outdated
| if (txd.hard_height_requirement) | ||
| { | ||
| const uint64_t curr_height = m_blockchain.get_current_blockchain_height(); | ||
| if (txd.max_used_block_height + CRYPTONOTE_DEFAULT_TX_SPENDABLE_AGE > curr_height) |
There was a problem hiding this comment.
Idea: if you have max_used_block_height=reference_block for FCMP++ txs, you can just check txd.max_used_block_height >= curr_height here
We wouldn't need the hard_height_requirement field, and can call this for any tx. It also avoids any issues surrounding off-by-1 logic
It introduces a difference on what max_used_block_height means for pre and post FCMP++ txs (which affects the if for // enforce min output age in blockchain.cpp), so I would accept if you prefer not to do that
There was a problem hiding this comment.
We wouldn't need the hard_height_requirement field, and can call this for any tx. It also avoids any issues surrounding off-by-1 logic
We can't use that height without re-checking for ring signature transactions because a reorg can change the height of referenced ring members. It's different for FCMP++ txs because the reference height is explicitly a height value and isn't allowed by consensus any lower than at least 1 block after the reference block height.
For example. if the top ring member gets re-orged from height H to height H-1, this code would skip over this transaction in the block template for the next block. Eventually, this would be eligible again, so maybe it's okay?
It introduces a difference on what max_used_block_height means for pre and post FCMP++ txs (which affects the if for // enforce min output age in blockchain.cpp), so I would accept if you prefer not to do that
This is also one of the main reasons why I chose to do it this way: I wanted to behavior to match what's happening with the ring signature transactions.
There was a problem hiding this comment.
Good point, you're right. As is is ok with me, will review closer for off-by-1
There was a problem hiding this comment.
Looks good, always tricky
j-berman
left a comment
j-berman
left a comment
There was a problem hiding this comment.
Small nit on passing db height into is_transaction_meta_ready_to_go, otherwise LGTM
src/cryptonote_core/tx_pool.cpp
Outdated
| return ret; | ||
| } | ||
| //--------------------------------------------------------------------------------- | ||
| bool tx_memory_pool::is_transaction_meta_ready_to_go(const txpool_tx_meta_t& txd) const |
There was a problem hiding this comment.
Could pass in the db height to prevent the db read for height for all txs in a loop
| *pmax_used_block_height = tx.rct_signatures.p.reference_block | ||
| - std::min<std::uint64_t>(tx.rct_signatures.p.reference_block, CRYPTONOTE_DEFAULT_TX_SPENDABLE_AGE - 1); |
There was a problem hiding this comment.
Note for self:
I use a similar pattern in db_lmdb.cpp here
I think it would be good to use a helper function for this logic just so it's consolidated somewhere / makes it easier to manage if there is some change down the road
I can do it in a separate PR
src/cryptonote_core/tx_pool.cpp
Outdated
| if (txd.hard_height_requirement) | ||
| { | ||
| const uint64_t curr_height = m_blockchain.get_current_blockchain_height(); | ||
| if (txd.max_used_block_height + CRYPTONOTE_DEFAULT_TX_SPENDABLE_AGE > curr_height) |
There was a problem hiding this comment.
Looks good, always tricky
2 participants
No description provided.