Hi Gloria, Greg,
> I interpret most of the discussion around limitations as ideas for
> future improvements rather than criticisms of the proposal
As far as I'm concerned, definitely!
My current understanding is that the main change/improvement that would
make sense here is restricting the whole v3 package's size (instead of
just the child) via committing to a specific value in the taproot annex
(also note that it's probably not just the v3 package's size, it should
be the whole unconfirmed package including potential v2 unconfirmed
ancestors).
While I think this would be very valuable and would like to see this
happen, I believe that can be done in a second, separate step since this
would make relay policy stricter (some v3 transactions that previously
propagated wouldn't propagate under this new rule). As long as you are
able to find a path to miners through upgraded peers that use this annex
approach, you should be able to resolve ACP pinning issues?
I'm curious to know how other people feel about that: is it ok to do
later or should we try to implement this for the first release of v3
transactions?
The other change mentioned (making OP_TRUE standard and allowing outputs
that are below dust) can be added later, as those won't be standard until
we start allowing them, so there shouldn't be any backwards-compatibility
issue with postponing this change. But maybe it's still worth having from
the get-go, even though it may take a bit more time? Again, I'm curious to
have other people's opinion here, I'd be happy to get all of those directly
in the first release of v3 transactions, but I don't know how much
implementation will have to go into that.
> For clarification, package RBF is ParentTx*s*(plural), and ChildTx(singular),
> so it might be a bit more complicated than we're thinking
Right, good catch, this does require new logic to handle this case.
As Gloria points out, this should be doable, and is definitely worth
adding (those CSV 1 on every other output are really hacky, glad to
find a way to get rid of them).
Thanks,
BastienLe lun. 26 sept. 2022 à 18:48, Gloria Zhao <gloriajzhao@gmail.com> a écrit :Hi Greg, Antoine, Bastien,
Thanks very much for the feedback! I interpret most of the discussion around limitations as ideas for future improvements rather than criticisms of the proposal (please correct me if I'm wrong). I'll try to respond to as much as possible.
Also I realize that I didn't contextualize this proposal clearly enough; it is very tailored for LN Penalty and definitely doesn't close all pinning attacks possible (sorry for confusing anyone). I also agree that some bits can be a little ugly or tack-on; I would definitely prefer a comprehensive RBF revamp to fix all our problems and enable other fee-bumping strategies such as sign-ANYONECANPAY-then-bring-your-own-fees-by-adding-inputs-at-broadcast. I was hoping to get some ideas with the "RBF Improvements" post in January, but it doesn't seem like we're much closer to a workable proposal. I think this is a minimally-invasive step that works for Lightning today, a small fix similar to CPFP carve out.
> As you likely know from previous discussions the biggest scenario this does not fix in my estimation is ANYONECANPAY situations. If the parent transaction can be "inflated" by tacking on additional inputs, this means the total weight of the parent tx lowers the effective feerate of the package.
(For more context to other readers I wrote an explanation for this in "SIGHASH_ANYONECANPAY Pinning" section of RBF ML post). Yes, this unfortunately doesn't fix any of the existing pinning attacks for single transaction RBF but also doesn't make them worse. This boils down to adding an incentive compatibility rule that ensures you can't replace a transaction with something that will confirm slower. Package RBF has an ancestor feerate-based rule for this (note it is quite conservative and not perfect).
So in the scenario above with the "inflated" parent that was signed ACP, the replacement would be rejected because the package ancestor feerate is lower than the feerate of what is being replaced. But it is imperfect (explained below) and thus I wouldn't recommend it for single transaction replacement. So that attack still exists for single transactions, yes.
The strategy of using ACP to bring-your-own-fees has its own challenges but hopefully has no current use cases as you say. AFAIK LN Penalty is not affected by this since it doesn't use ACP, though obviously I agree we should fix it for the future.
So when I said "this is intended for fee-bumping presigned txns in contracting protocols," I should have said "this is intended for fee-bumping presigned txns specifically using CPFP and anchor outputs." Apologies for forgetting to contextualize, I've been sitting on this for too long.
> The other scenario it doesn't really fix is where HTLC/commitment-like transactions are being resolved in a batch, but due to relative time constraints, you may want to accelerate some and not others. Now you must pay higher rates to replace all of the transaction bumps. This is a "self-pin" and "get good at utxos noob" type problem, but it's something that axing rule#3 in favor of a Replace-by-ancestor-feerate system would get us.
I understand you to mean "if you don't have enough UTXOs and you're forced to batch-bump, you over-pay because you need to bring them all to the highest target feerate." Isn't this kind of separate, wallet-related problem? Contracting or not, surely every wallet needs to have enough UTXOs to not batch transactions that shouldn't be batched... I don't see how a replace-by-ancestor-feerate policy would make any difference for this?
Also in general I'd like to reiterate that ancestor feerate is not a panacea to all our RBF incentive compatibility concerns. Like individual feerate, unless we run the mining algorithm, it cannot tell us exactly how quickly this transaction would be mined.
We're estimating the incentive compatibility of the original transaction(s) and replacement transaction(s), with the goal of not letting a transaction replace something that would have been more incentive compatible to mine. As such, we don't want to overestimate how good the replacement is, and we don't want to underestimate how good the original transactions are. This rule "The minimum between package feerate and ancestor feerate of the child is not lower than the individual feerates of all directly conflicting transactions and the ancestor feerates of all original transactions" is a conservative estimate.
> Would kind of be nice if package RBF would detect a "sibling output spend" conflict, and knock it out of the mempool via the other replacement rules? Getting rid of the requirement to 1 block csv lock every output would be quite nice from a smart contracting composability point of view.
Interesting, so when a transaction hits a mempool tx's descendant limit, we consider evicting one of its descendants in favor of this transaction, based on the RBF rules.
Cool idea! After chewing on this for a bit, I think this *also* just boils down to the fact that RBF should require replacements to be better mining candidates. As in, if we added this policy and it can make us evict the sibling and accept a transaction with a bunch of low-feerate ancestor junk, it would be a new pinning vector.
> If you're a miner and you receive a non-V3, second descendant of an unconfirmed V3 transaction, if the offered fee is in the top mempool backlog, I think you would have an interest to accept such a transaction.
> So I'm not sure if those two rules are compatible with miners incentives...
The same argument can be made for the 26th descendant of a mempool transaction; it's also not entirely incentive-compatible to reject it, but that is not the *only* design goal in mempool policy. Of course, the difference here is that the 25-descendant limit rule is a sensible DoS protection, while this 1-descendant limit rule is more of a "help the Bitcoin ecosystem" policy, just like CPFP carve-out, dust limit, etc. I can of course understand why not everyone would be in favor of this, but I do think it's worth it.
> > 4. A V3 transaction that has an unconfirmed V3 ancestor cannot be
> > larger than 1000 virtual bytes.
> If I understand correctly the 1000 vb upper bound rational, it would be to constraint the pinning counterparty to attach a high fee to a child due to the limited size, if they would like this transaction to be stuck in the network mempools. By doing so this child has high odds to confirm.
Yeah exactly, the "Rule 3 pin" is done by adding a child that's high-fee (so you have to pay that much to evict it). Because they *don't* want this tx to confirm, normally, this child would be really large. If they only have 1000vB for the child, they can't increase the replacement cost without also fee-bumping the transaction to make it confirm faster.
> As of today, I think yes you can already fingerprint LN transactions on the spec-defined amount value of the anchor outputs, 330 sats. There is always one of them on post-anchor commitment transactions. And sadly I would say we'll always have tricky fingerprints leaking from unilateral LN closures such as HTLC/PTLC timelocks...> I agree with you, this isn't worse than today, unilateral closes willprobably always be identifiable on-chain.
Great to hear that there is no privacy worsening!
Best,
GloriaOn Mon, Sep 26, 2022 at 5:02 PM Greg Sanders <gsanders87@gmail.com> wrote:Bastien,> This may be already covered by the current package RBF logic, in thatscenario we are simply replacing [ParentTx, ChildTx1] with
[ParentTx, ChildTx2] that pays more fees, right?For clarification, package RBF is ParentTx*s*(plural), and ChildTx(singular), so it might be a bit more complicated than we're thinking, and currently the V3 proposal would first de-duplicate the ParentTx based on what is in the mempool, then look at the "rest" of the transactions as a package, then individually. Not the same, not sure how different. I'll defer to experts.Best,GregOn Mon, Sep 26, 2022 at 11:48 AM Bastien TEINTURIER via bitcoin-dev <bitcoin-dev@lists.linuxfoundation.org> wrote:Thanks Gloria for this great post.
This is very valuable work for L2 contracts, and will greatly improve
their security model.
> "Only 1 anchor output? What if I need to bump counterparty's commitment tx in mempool?"
> You won't need to fee-bump a counterparty's commitment tx using CPFP.
> You would just package RBF it by attaching a high-feerate child to
> your commitment tx.
Note that we can also very easily make that single anchor spendable by
both participants (or even anyone), so if you see your counterparty's
commitment in your mempool, you can bump it without publishing your
own commitment, which is quite desirable (your own commitment tx has
CSV delays on your outputs, whereas your counterparty's commitment tx
doesn't).
> "Is this a privacy issue, i.e. doesn't it allow fingerprinting LN
transactions based on nVersion?"
I agree with you, this isn't worse than today, unilateral closes will
probably always be identifiable on-chain.
> Would kind of be nice if package RBF would detect a "sibling output spend"
> conflict, and knock it out of the mempool via the other replacement rules?
> Getting rid of the requirement to 1 block csv lock every output would be
> quite nice from a smart contracting composability point of view.
+1, that would be very neat!
This may be already covered by the current package RBF logic, in that
scenario we are simply replacing [ParentTx, ChildTx1] with
[ParentTx, ChildTx2] that pays more fees, right?
> 1) I do think that we should seriously consider allowing OP_TRUE to become
> a standard script type as part of this policy update. If pinning is solved,
> then there's no reason to require all those extra bytes for "binding" an
> anchor to a specific wallet/user. We can save quite a few bytes by having
> the input be empty of witness data.
> 2) If we allow for a single dust-value(0 on up) output which is immediately
> spent by the package, anchors become even easier to to design. No value has
> to be "sapped" from contract participants to make an anchor output. There's
> more complications for this, such as making sure the parent transaction is
> dropped if the child spend is dropped, but maybe it's worth the squeeze.
I also think both of these could be quite useful. This would probably always
be used in combination with a parent transaction that pays 0 fees, so the
0-value output would always be spent in the same block.
But this means we could end up with 0-value outputs in the utxo set, if for
some reason the parent tx is CPFP-ed via another output than the 0-value one,
which would be a utxo set bloat issue. But I'd argue that we're probably
already creating utxo set bloat with the 330 sat anchor outputs (especially
since we use two of them, but only one is usually spent), so it would
probably be *better* than what we're doing today.
Thanks,
Bastien_______________________________________________Le lun. 26 sept. 2022 à 03:22, Antoine Riard via bitcoin-dev <bitcoin-dev@lists.linuxfoundation.org> a écrit :Hi Gloria,
Thanks for the progress on package RBF, few early questions.
> 2. Any descendant of an unconfirmed V3 transaction must also be V3.
> 3. An unconfirmed V3 transaction cannot have more than 1 descendant.
If you're a miner and you receive a non-V3, second descendant of an unconfirmed V3 transaction, if the offered fee is in the top mempool backlog, I think you would have an interest to accept such a transaction.
So I'm not sure if those two rules are compatible with miners incentives...
> 4. A V3 transaction that has an unconfirmed V3 ancestor cannot be
> larger than 1000 virtual bytes.
If I understand correctly the 1000 vb upper bound rational, it would be to constraint the pinning counterparty to attach a high fee to a child due to the limited size, if they would like this transaction to be stuck in the network mempools. By doing so this child has high odds to confirm.
I still wonder if this compatible with miner incentives in period of empty mempools, in the sense that if you've already a V3 transaction of size 100Kvb offering 2 sat/vb, it's more interesting than a V3 replacement candidate of size 1000 vb offering 10 sat/vb. It could be argued the former should be conserved.
(That said, the hard thing with any replacement strategy we might evict a parent transaction *now* to which is attached a high-feerate child *latter* making for a utxo considered the best ancestor set. Maybe in the long-term miners should keep every transaction ever accepted...)
> (Lower bound) the smaller this limit, the fewer UTXOs a child may use
> to fund this fee-bump. For example, only allowing the V3 child to have
> 2 inputs would require L2 protocols to manage a wallet with high-value
> UTXOs and make batched fee-bumping impossible. However, as the
> fee-bumping child only needs to fund fees (as opposed to payments),
> just a few UTXOs should suffice.
Reminder for L2 devs, batched fee-bumping of time-sensitive confirmations of commitment transactions is unsafe, as the counterparty could enter in a "cat-and-mouse" game to replace one of the batch element at each block to delay confirmation of the remaining elements in the batch, I think.
On the other hand, I wonder if we wouldn't want a higher bound. LN wallets are likely to have one big UTXO in their fee-bumping reserve pool, as the cost of acquiring UTXO is non-null and in the optimistic case, you don't need to do unilateral closure. Let's say you close dozens of channels at the same time, a UTXO pool management strategy might be to fan-out the first spends UTXOs in N fan-out outputs ready to feed the remaining in-flight channels.
> 1. The rule around unconfirmed inputs was
> originally "A package may include new unconfirmed inputs, but the
> ancestor feerate of the child must be at least as high as the ancestor
> feerates of every transaction being replaced."
Note, I think we would like this new RBF rule to also apply to single transaction package, e.g second-stage HTLC transactions, where a counterparty pins a HTLC-preimage by abusing rule 3. In that case, the honest LN node should be able to broadcast a "at least as high ancestor feerate" HTLC-timeout transaction. With `option_anchor_outputs" there is no unconfirmed ancestor to replace, as the commitment transaction, whatever the party it is originating from, should already be confirmed.
> "Is this a privacy issue, i.e. doesn't it allow fingerprinting LN
transactions based on nVersion?"
As of today, I think yes you can already fingerprint LN transactions on the spec-defined amount value of the anchor outputs, 330 sats. There is always one of them on post-anchor commitment transactions. And sadly I would say we'll always have tricky fingerprints leaking from unilateral LN closures such as HTLC/PTLC timelocks...
> "Can a V2 transaction replace a V3 transaction and vice versa?"
IIUC, a V3 package could replace a V2 package, with the benefit of the new package RBF rules applied. I think this would be a significant advantage for LN, as for the current ~85k of opened channels, the old V2 states shouldn't be pinning vectors. Currently, commitment transactions signal replaceability._______________________________________________Le ven. 23 sept. 2022 à 11:26, Gloria Zhao via bitcoin-dev <bitcoin-dev@lists.linuxfoundation.org> a écrit :Hi everyone,_______________________________________________
I'm writing to propose a very simple set of mempool/transaction relay
policies intended to aid L2/contract protocols. I realized that
the previously proposed Package Mempool Accept package RBF [1]had a few remaining problems after digging into the RBF logic more [2].This additional set of policies solves them without requiring a huge RBF overhaul.
I've written an implementation (and docs) for Bitcoin Core:
https://github.com/bitcoin/bitcoin/pull/25038
(You may notice that this proposal incorporates feedback on the PR - thanks Suhas Daftuar, Gregory Sanders, Bastien Teinturier, Anthony Towns, and others.)
If you are interested in using package RBF/relay to bump presigned
transactions, I think you may be interested in reviewing this proposal.
This should solve Rule 3 pinning and perhaps allow us
to get rid of CPFP carve-out (yay!). I'm keen to hear if people find
the 1-anchor-output, 1000vB child limit too restrictive. Also, if you find a
pinning attack or something that makes it unusable for you, I would
really really like to know.
Note that transactions with nVersion=3 ("V3 transactions") are
currently non-standard in Bitcoin Core. That means **anything that was
standard before this policy change would still be standard
afterwards.** If you don't want your transactions to be subject to
these rules, just continue whatever you're doing and don't use
nVersion=3. AFAICT this shouldn't break anything, but let me know if
this would be disruptive for you?
**New Policies:**
This includes:
- a set of additional policy rules applying to V3 transactions
- modifications to package RBF rules
**V3 transactions:**
Existing standardness rules apply to V3 (e.g. min/max tx weight,
standard output types, cleanstack, etc.). The following additional
rules apply to V3:
1. A V3 transaction can be replaced, even if it does not signal BIP125
replaceability. (It must also meet the other RBF rules around fees,
etc. for replacement to happen).
2. Any descendant of an unconfirmed V3 transaction must also be V3.
*Rationale*: Combined with Rule 1, this gives us the property of
"inherited" replaceability signaling when descendants of unconfirmed
transactions are created. Additionally, checking whether a transaction
signals replaceability this way does not require mempool traversal,
and does not change based on what transactions are mined. It also
makes subsequent rules about descendant limits much easier to check.
*Note*: The descendant of a *confirmed* V3 transaction does not need to be V3.
3. An unconfirmed V3 transaction cannot have more than 1 descendant.
*Rationale*: (Upper bound) the larger the descendant limit, the more
transactions may need to be replaced. This is a problematic pinning
attack, i.e., a malicious counterparty prevents the transaction from
being replaced by adding many descendant transactions that aren't
fee-bumping.
(Lower bound) at least 1 descendant is required to allow CPFP of the
presigned transaction. The contract protocol can create presigned
transactions paying 0 fees and 1 output for attaching a CPFP at
broadcast time ("anchor output"). Without package RBF, multiple anchor
outputs would be required to allow each counterparty to fee-bump any
presigned transaction. With package RBF, since the presigned
transactions can replace each other, 1 anchor output is sufficient.
4. A V3 transaction that has an unconfirmed V3 ancestor cannot be
larger than 1000 virtual bytes.
*Rationale*: (Upper bound) the larger the descendant size limit, the
more vbytes may need to be replaced. With default limits, if the child
is e.g. 100,000vB, that might be an additional 100,000sats (at
1sat/vbyte) or more, depending on the feerate.
(Lower bound) the smaller this limit, the fewer UTXOs a child may use
to fund this fee-bump. For example, only allowing the V3 child to have
2 inputs would require L2 protocols to manage a wallet with high-value
UTXOs and make batched fee-bumping impossible. However, as the
fee-bumping child only needs to fund fees (as opposed to payments),
just a few UTXOs should suffice.
With a limit of 1000 virtual bytes, depending on the output types, the
child can have 6-15 UTXOs, which should be enough to fund a fee-bump
without requiring a carefully-managed UTXO pool. With 1000 virtual
bytes as the descendant limit, the cost to replace a V3 transaction
has much lower variance.
*Rationale*: This makes the rule very easily "tacked on" to existing
logic for policy and wallets. A transaction may be up to 100KvB on its
own (`MAX_STANDARD_TX_WEIGHT`) and 101KvB with descendants
(`DEFAULT_DESCENDANT_SIZE_LIMIT_KVB`). If an existing V3 transaction
in the mempool is 100KvB, its descendant can only be 1000vB, even if
the policy is 10KvB.
**Package RBF modifications:**
1. The rule around unconfirmed inputs was
originally "A package may include new unconfirmed inputs, but the
ancestor feerate of the child must be at least as high as the ancestor
feerates of every transaction being replaced."
The package may still include new unconfirmed inputs. However,
the new rule is modified to be "The minimum between package feerate
and ancestor feerate of the child is not lower than the individual
feerates of all directly conflicting transactions and the ancestor
feerates of all original transactions."
*Rationale*: We are attempting to ensure that the replacement
transactions are not less incentive-compatible to mine. However, a
package/transaction's ancestor feerate is not perfectly representative
of its incentive compatibility; it may overestimate (some subset of
the ancestors could be included by itself if it has other high-feerate
descendants or are themselves higher feerate than this
package/transaction). Instead, we use the minimum between the package
feerate and ancestor feerate of the child as a more conservative value
than what was proposed originally.
2. A new rule is added, requiring that all package transactions with
mempool conflicts to be V3. This also means the "sponsoring"
child transaction must be V3.
*Note*: Combined with the V3 rules, this means the package must be
a child-with-parents package. Since package validation is only
attempted if the transactions do not pay sufficient fees to be
accepted on their own, this effectively means that only V3
transactions can pay to replace their ancestors' conflicts, and only
V3 transactions' replacements may be paid for by a descendant.
*Rationale*: The fee-related rules are economically rational for
ancestor packages, but not necessarily other types of packages.
A child-with-parents package is a type of ancestor package. It
may be fine to allow any ancestor package, but it's more difficult
to account for all of the possibilities. For example, it gets much
harder to see that we're applying the descendant limits correctly if
the package has a gnarly, many-generation, non-tree shape. I'm also
not sure if this policy is 100% incentive-compatible if the sponsor
is not a direct descendant of the sponsee.
Please see doc/policy/version3_transactions.md and
doc/policy/packages.md in the PR for the full set of rules.
**Intended usage for LN:**
Commitment transactions should be V3 and have 1 anchor output. They
can be signed with 0 fees (or 1sat/vbyte) once package relay is deployed
on a significant portion of the network. If the commitment tx must
be broadcast, determine the desired feerate at broadcast time and
spend the anchor output in a high feerate transaction. I'm going to
call the broadcasted commitment tx "the parent" and the attached
fee-bumping tx "the child."
- This child must be V3.
- This child must be at most 1000vB. Note this restricts the
number of inputs you can use to fund the fee bump. Depending
on the output types, this is around 6-15.
- One child may fund fees for multiple commitment tx ("batched
fee-bumping").
- To do a second fee-bump to add more fees, replace the
*child* with a higher-feerate tx. Do not try to attach a grandchild.
Otherwise, never try to spend from an unconfirmed V3 transaction. The
descendant limits for V3 transactions are very restrictive.
**Expected Questions:**
"Does this fix Rule 3 Pinning?"
Yes. The V3 descendant limit restricts both you and your counterparty.
Assuming nodes adopted this policy, you may reasonably assume that you
only need to replace the commitment transaction + up to 1000vB.
"Only 1 anchor output? What if I need to bump counterparty's commitment tx in mempool?"You won't need to fee-bump a counterparty's commitment tx using CPFP.You would just package RBF it by attaching a high-feerate child toyour commitment tx.
"Is this a privacy issue, i.e. doesn't it allow fingerprinting LN
transactions based on nVersion?"
Indeed it may be unrealistic to assume V3 transactions will be in
widespread use outside of L2. IIUC, unilateral closes are already
obvious LN transactions because of the HTLC inputs. For e.g.
cooperative closes and opens, I think it makes sense to continue using
V2. So, unless I'm missing something, this shouldn't make it worse.
"So a V3 transaction that doesn't signal BIP125 replaceability is
replaceable? Is that a backward compatibility issue?"
Yes it's replaceable. It's not an issue AFAICT because,
under previous policy, the V3 transaction wouldn't have been
in the mempool in the first place.
"Can a V2 transaction replace a V3 transaction and vice versa?"
Yes, otherwise someone can use V3 transactions to censor V2
transactions spending shared inputs. Note if the
original V3 transaction has an unconfirmed V3 parent, this would
violate the "inherited V3" rule and would be rejected.
Thanks for reading! Feedback and review would be much appreciated.
[1]: https://lists.linuxfoundation.org/pipermail/bitcoin-dev/2021-September/019464.htmlBest,
Gloria
bitcoin-dev mailing list
bitcoin-dev@lists.linuxfoundation.org
https://lists.linuxfoundation.org/mailman/listinfo/bitcoin-dev
bitcoin-dev mailing list
bitcoin-dev@lists.linuxfoundation.org
https://lists.linuxfoundation.org/mailman/listinfo/bitcoin-dev
bitcoin-dev mailing list
bitcoin-dev@lists.linuxfoundation.org
https://lists.linuxfoundation.org/mailman/listinfo/bitcoin-dev