Uh oh!
There was an error while loading. Please reload this page.
Move pre-signed channel transactions generation in OnchainTxHandler - #540
Move pre-signed channel transactions generation in OnchainTxHandler#540ariard wants to merge 7 commits into
Conversation
TheBlueMatt
commented
Mar 10, 2020
Note: failed borrowck on 1.22. |
Codecov Report
@@ Coverage Diff @@## master #540 +/- ##
==========================================
- Coverage 90.23% 90.21% -0.03%
==========================================
Files 34 34 Lines 19182 19120 -62 ==========================================
- Hits 17309 17249 -60 + Misses 1873 1871 -2
Continue to review full report at Codecov.
|
Uh oh!
There was an error while loading. Please reload this page.
| if let Some((new_timer, new_feerate, tx)) = self.generate_claim_tx(height, &claim_material, &*fee_estimator) { | ||
| claim_material.height_timer = new_timer; | ||
| claim_material.feerate_previous = new_feerate; | ||
| if claim_material.height_timer != ::std::u32::MAX { |
There was a problem hiding this comment.
Please use Option<> instead of a sentinal value.
There was a problem hiding this comment.
Yeah, spendable outputs is reworked in next PR, so just temporary but switch to a Option<> ofc
| }, | ||
| Funding { | ||
| local_tx_remote_signed: LocalCommitmentTransaction, | ||
| funding_key: SecretKey, |
There was a problem hiding this comment.
Don't pass the SecretKeys - you can get them out of the ChannelKeys already with ChannelKeys::funding_key() or ChannelKeys::htlc_base_key().
There was a problem hiding this comment.
Hmmmm what do you suggest exactly ? To introduce yet ChannelKeys in OnchainTxHandler ?
There was a problem hiding this comment.
Yea, just pass in a reference to the ChannelMonitor ChannelKeys in OnchainTxHandler::block_connected() and pass it down. We can rework it later, but its so awkward to copy private keys around like this.
Previously, we would regenerate this class of txn twice due to block-rescan triggered by new watching outputs registered. This commmit doesn't change behavior, it only tweaks TestBroadcaster to ensure we modify cleanly tests anticipating next commit refactor.
Watchtower will be supported through external signer interface where a watchtower implementation may differ from a local one by the scope of key access and pre-signed datas.
@TheBlueMatt okay added 2 new commits : one to drop Watchtower mode from Storage, tower support is planned to be implemented as an external signer implem, so remove it from ChannelMonitor which is for detection. The other to move duplicate keys storage in OnchainTxHanndler, to avoid passing back and forth stuff in InputMaterial. Next PRs should keep drying up key usage in ChannelMonitor and pass Storage behind ChanSigner. |
TheBlueMatt
left a comment
There was a problem hiding this comment.
Note this currently does not build.
| assert_eq!(node_txn[3].input.len(), 1); | ||
| let witness_script = node_txn[3].input[0].witness.last().unwrap(); | ||
| assert_eq!(witness_script.len(), OFFERED_HTLC_SCRIPT_WEIGHT); //Spending an offered htlc output | ||
| check_spends!(node_txn[3], node_txn[2].clone()); |
There was a problem hiding this comment.
Rebase error - you dont need the clone()s anymore. Also in a few other places.
Going further between splitting detection and transaction generation, we endow OnchainTxHandler with keys access. That way, in latter commits, we may remove keys entirely from ChannelMonitor.
Local Commitment Transaction can't be bumped without anchor outputs so their generation is one-time for now. We move them in OnchainTxHandler for simplifying ChannelMonitor and to prepare storage of keys material behind one external signer interface. Some tests break due to change in transaction broadcast order but number of transactions broadcast should stay the same.
HTLC Transaction can't be bumped without sighash changes so their gneeration is one-time for nwo. We move them in OnchainTxHandler for simplifying ChannelMonitor and to prepare storage of keys material behind one external signer interface. Some tests break due to change in transaction broadcaster order. Number of transactions may vary because of temporary anti-duplicata tweak can't dissociate between 2- broadcast from different origins (ChannelMonitor, ChannelManager) and 2-broadcast from same component.
| assert_eq!(node_txn.len(), 3); | ||
| check_spends!(node_txn[1], node_txn[0]); | ||
| check_spends!(node_txn[2], node_txn[0]); | ||
| check_spends!(node_txn[0], node_txn[2].clone()); |
There was a problem hiding this comment.
Rebase error - clone is not needed here.
| // If would_broadcast_at_height determine this local tx should be broadcast, absolute_timelock is set to current_height, | ||
| // because it indicates that this tx confirmation is urgent. It's not going to change anything because we can't bump | ||
| // local_commitment before anchor_outputs, should be rethought afterwards | ||
| claimable_outpoints.push(ClaimRequest { absolute_timelock: height, aggregable: false, outpoint: BitcoinOutPoint { txid: self.key_storage.funding_info.as_ref().unwrap().0.txid.clone(), vout: self.key_storage.funding_info.as_ref().unwrap().0.index as u32 }, witness_data: InputMaterial::Funding { local_tx_remote_signed: cur_local_tx.tx.clone(), channel_value: self.channel_value_satoshis.unwrap() }}); |
There was a problem hiding this comment.
clone() ing the cur_local_tx wholesale here really sucks. a) you need to drop the add_local_sig line above, since we should now be signing + broadcasting this tx in onchain wholesale, and b) can we take() the cur_local_tx.tx and pass it to OnChainTxHandler? We should never be doing anything with it aside from sign, then broadcast, so can it happen in onchain?
| // Check the pair local commitment and HTLC-timeout broadcast due to HTLC expiration | ||
| assert_eq!(node_txn[3].input.len(), 1); | ||
| check_spends!(node_txn[3], chan_1.3.clone()); |
There was a problem hiding this comment.
Same here - can you do a pass for clone() in check_spends? Seems like you reintroduced a bunch of them.
TheBlueMatt
commented
Mar 20, 2020
Needs rebase. |
ariard
commented
Mar 24, 2020
Replaced by #559 |
Next step in moving private keys behind an external signer interface, moving pre-signed channel transactions (local commitment, HTLC-Success, HTLC-Timeout) in OnchainTxHandler. There is no bump logic, as without option_simplified_commitment changes we can't CPFP/BYOF them.
After this, code should be ready to move key storage inside OnchainTxHandler and some chan_utils methods in signer. We need also to think about bumping logic, if we make it an interface accessible to the signer.