From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from smtp4.osuosl.org (smtp4.osuosl.org [140.211.166.137]) by lists.linuxfoundation.org (Postfix) with ESMTP id 23118C000D for ; Wed, 29 Sep 2021 11:56:43 +0000 (UTC) Received: from localhost (localhost [127.0.0.1]) by smtp4.osuosl.org (Postfix) with ESMTP id 07A034163C for ; Wed, 29 Sep 2021 11:56:43 +0000 (UTC) X-Virus-Scanned: amavisd-new at osuosl.org X-Spam-Flag: NO X-Spam-Score: -2.098 X-Spam-Level: X-Spam-Status: No, score=-2.098 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, DKIM_SIGNED=0.1, DKIM_VALID=-0.1, DKIM_VALID_AU=-0.1, DKIM_VALID_EF=-0.1, FREEMAIL_FROM=0.001, HTML_MESSAGE=0.001, RCVD_IN_DNSWL_NONE=-0.0001, SPF_HELO_NONE=0.001, SPF_PASS=-0.001] autolearn=ham autolearn_force=no Authentication-Results: smtp4.osuosl.org (amavisd-new); dkim=pass (2048-bit key) header.d=gmail.com Received: from smtp4.osuosl.org ([127.0.0.1]) by localhost (smtp4.osuosl.org [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id D5LHoPt_r7jn for ; Wed, 29 Sep 2021 11:56:38 +0000 (UTC) X-Greylist: whitelisted by SQLgrey-1.8.0 Received: from mail-yb1-xb2a.google.com (mail-yb1-xb2a.google.com [IPv6:2607:f8b0:4864:20::b2a]) by smtp4.osuosl.org (Postfix) with ESMTPS id C44724162E for ; Wed, 29 Sep 2021 11:56:37 +0000 (UTC) Received: by mail-yb1-xb2a.google.com with SMTP id 71so4721290ybe.6 for ; Wed, 29 Sep 2021 04:56:37 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=20210112; h=mime-version:references:in-reply-to:from:date:message-id:subject:to :cc; bh=5dRrZ2h1odOCcg3PmQbAqcepP8jmz+VUm4gTrMv3bXI=; b=iafCYtF6XtXaWaGnTmbyIifzu7BRlrj50jiEOmRsFvR8X2B6ekTG/3Z/vc3CeCCJBV yXBjZrHCNkdb8gyPeGd9u6nzchtk34FlW8uUxlUfoQPMsoMiHACIEpsfm6BDORbWJzkJ duYE3LpIdTbjuxcLioZocdP6rCnif+2l0bJa9jdliOh9hHB1ToqPj58hnLKCbX7Ivxem /NXCKzslzS6EuSvAdodWYxvU5AblRz+g3WdR0yJEb00/mxlb+MacC62d9zcXafcYF4Sn agSGSXW3AIXUvm2EaQqZtzAwqiOAFRmzPUXafKU4j0UXGwG89IgDKFjxlvbr2oiWeAmN PeWA== X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20210112; h=x-gm-message-state:mime-version:references:in-reply-to:from:date :message-id:subject:to:cc; bh=5dRrZ2h1odOCcg3PmQbAqcepP8jmz+VUm4gTrMv3bXI=; b=l1/Jels0JTQlmgWmfzb8zkZx1jrY8sm/G9bN8tM6RlXdyJBFJZ8iyDgHgTjwHaxebh 6hQw/ns8CmXz1yCq2emlay20wPZzIP+ewFnvqIjvTybZUapKd8dGF9QVJt68NUKSV03q iOOdeFheG19huiSknAayEMajIrzVCwBNz2qpeTdzPCy8cAhfCOal387m2nG9bM4WyJy8 ZM3szXf4r7WE9JCaBwyhdoPHOXUMVklqW5ITmT522bbtG0/gUJCiRa3UtrKouZ1MGv3e Jv+rw7HHvezWXTmoINoC3uTGPHIpErCs0J9VnbvVRKY4Pr545ksTKVRXKwYObDpL4R3m OxJA== X-Gm-Message-State: AOAM531A883Dk+lH+YSDpF+Ov7hJqLddsBiL5x5/A6h4JuQbD6Gt7X0l r3VoVAE24pxHv3UuQAW+0Jb6lJarZPduvVRJSODGJEHb4t0= X-Google-Smtp-Source: ABdhPJwdXDK4yuZ3NAujlFVoFAvvR/yVDcg5Ez7YHU4+rn9yC4tW6zuoDNZV4wsrIhmA+z9tZyH+dyevpk8EeYoxPBQ= X-Received: by 2002:a25:bd7:: with SMTP id 206mr4654367ybl.408.1632916595867; Wed, 29 Sep 2021 04:56:35 -0700 (PDT) MIME-Version: 1.0 References: In-Reply-To: From: Gloria Zhao Date: Wed, 29 Sep 2021 12:56:24 +0100 Message-ID: To: Antoine Riard Content-Type: multipart/alternative; boundary="000000000000a502f005cd2105e8" X-Mailman-Approved-At: Wed, 29 Sep 2021 12:05:36 +0000 Cc: Bitcoin Protocol Discussion Subject: Re: [bitcoin-dev] Proposal: Package Mempool Accept and Package RBF X-BeenThere: bitcoin-dev@lists.linuxfoundation.org X-Mailman-Version: 2.1.15 Precedence: list List-Id: Bitcoin Protocol Discussion List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Wed, 29 Sep 2021 11:56:43 -0000 --000000000000a502f005cd2105e8 Content-Type: text/plain; charset="UTF-8" Content-Transfer-Encoding: quoted-printable Hi Antoine and Bastien, > Yes 1) it would be good to have inputs of more potential users of package acceptance . And 2) I think it's more a matter of clearer wording of the proposal. (1) I'm leaning towards multi-parent-1-child and offering [#22674][0] up for review. If somebody feels very strongly about 1-parent-1-child, please let me know. (2) I'm glad this turned out to just be a wording problem. I've updated the proposal to [say][1] "If it meets feerate requirements, the package can replace mempool transactions if any of the parents conflict with mempool transactions. The child cannot conflict with any mempool transactions." Hopefully that is more *univoque*. Side note: I've also updated the proposal to contain a [section][2] on why submitting transactions individually before package validation is incentive-compatible. I think it's relevant to our conversation, but for those who just want to _use_ packages, it's just an implementation detail. On restricting packages to confirmed inputs only: > I think we could restrain package acceptance to only confirmed inputs for now and revisit later this point ? For LN-anchor, you can assume that the fee-bumping UTXO feeding the CPFP is already confirmed. Or are there currently-deployed use-cases which would benefit from your proposed Rule #2 ? I thought about this a lot this week, and wrote up a summary of why I don't think BIP125#2 helps us at all [here][3] on #23121. I see that you've already come across it :) > IIRC, the carve-out tolerance is only 2txn/10_000 vb. If one of your counterparties attach a junk branch on her own anchor output, are you allowed to chain your self-owned unconfirmed CPFP ? Yes, if your counterparty attaches a bunch of descendants to their anchor output to dominate the descendant limit of your shared commitment transaction, CPFP carve out allows you to add 1 extra transaction under 10KvB to your own anchor output. It's fine if it spends an unconfirmed input, as long as you aren't exceeding the descendant limits of that transaction. This shouldn't be the case; I think something is seriously wrong if all of your UTXOs are tied up in mempool transactions with big ancestor/descendant trees. I don't know much about L2 development so I'm just going to quote this: > I think constraining package acceptance to only confirmed inputs is very limiting and quite dangerous for L2 protocols. Since the restriction isn't helpful in simplifying the mempool code, makes things more complicated for application developers, and can be dangerous for L2, I'd prefer not to add this restriction for packages. On Antoine's question about our miner model: > Can you describe what miner model we are using ? Like the block construction strategy implemented by `addPackagesTxs` or also encompassing our current mempool acceptance policy, which I think rely on absolute fee over ancestor score in case of replacement ? Our current model for block construction is this: we sort our mempool by package ancestor score (total modified fees of a tx and its unconfirmed ancestors / total vsize as seen by our mempool) and add packages to a block until it's full. That's not to say this is the perfect miner policy, but mempool acceptance logic follows this model as closely as possible because it is, fundamentally, a cache that aids in block assembly performance. As another way of looking at this, imagine if our mempool was so small it could only store ~1 block's worth of transactions. It should always try to keep the highest-fees-within-1-block transactions, and obviously wouldn't evict small-but-valuable transations in favor or giant ones paying mediocre feerates. All fee-related mempool policies, including RBF, consider feerate. BIP125#3 is a rule on absolute fees, but it is always combined with BIP125#4, a rule on feerates. AFAIK, the reason it doesn't use ancestor score is that information wasn't cached in mempool entries at the time, and thus not readily available to use in mempool validation. That's why I don't think this is relevant to package validation. Commenting on the model itself: > Is this compatible with a model where a miner prioritizes absolute fees over ancestor score, in the case that mempools aren't full-enough to fulfill a block ? >> Yes, A+C+D pays 2500sat more in fees, but it is also 1000vB larger. A miner should prefer to utilize their block space more effectively. > If your mempool is empty and only composed of A+C+D or A+B, I think taking A+C+D is the most efficient block construction you can come up with as a miner ? > I think this point is worthy to discuss as otherwise we might downgrade the efficiency of our current block construction strategy in periods of near-empty mempools. A knowledge which could be discreetly leveraged by a miner to gain an advantage on the rest of the mining ecosystem. I believe this is suggesting "if our mempool has so few transactions that it wouldn't reach block capacity, prioritize any increase in absolute fees, even if the feerate is lower." I can see how this may result in a higher-fee block in a specific scenario such as the one highlighted above, but I don't think it is a sound model in general. It would be impossible to tell when we should use this model: we could simply be in IBD, restarted a node with an old/empty mempool.dat, and even if it's a low-transaction-volume time, we never know what transactions will trickle in between now and the next block. Going back to the tiny 1-block mempool scenario, i.e., if you _never_ wanted to keep transactions that you wouldn't put in the next block, would you ever switch strategies? Thanks again to everyone who's given their attention to the package mempool accept proposal. Best, Gloria [0]: https://github.com/bitcoin/bitcoin/pull/22674 [1]: https://gist.github.com/glozow/dc4e9d5c5b14ade7cdfac40f43adb18a#package-rbf [2]: https://gist.github.com/glozow/dc4e9d5c5b14ade7cdfac40f43adb18a#always-try-= individual-submission-first [3]: https://github.com/bitcoin/bitcoin/pull/23121#issuecomment-929475999 On Tue, Sep 28, 2021 at 11:59 PM Antoine Riard wrote: > Hi Bastien > > > In the case of LN, an attacker can game this and heavily restrict > your RBF attempts if you're only allowed to use confirmed inputs > and have many channels (and a limited number of confirmed inputs). > Otherwise you'll need node operators to pre-emptively split their > utxos into many small utxos just for fee bumping, which is inefficient... > > I share the concern about splitting utxos into smaller ones. > IIRC, the carve-out tolerance is only 2txn/10_000 vb. If one of your > counterparties attach a junk branch on her own anchor output, are you > allowed to chain your self-owned unconfirmed CPFP ? > I'm thinking about the topology "Chained CPFPs" exposed here : > https://github.com/rust-bitcoin/rust-lightning/issues/989. > Or if you have another L2 broadcast topology which could be safe w.r.t ou= r > current mempool logic :) ? > > > Le lun. 27 sept. 2021 =C3=A0 03:15, Bastien TEINTURIER = a > =C3=A9crit : > >> I think we could restrain package acceptance to only confirmed inputs fo= r >>> now and revisit later this point ? For LN-anchor, you can assume that t= he >>> fee-bumping UTXO feeding the CPFP is already >>> confirmed. Or are there currently-deployed use-cases which would benefi= t >>> from your proposed Rule #2 ? >>> >> >> I think constraining package acceptance to only confirmed inputs >> is very limiting and quite dangerous for L2 protocols. >> >> In the case of LN, an attacker can game this and heavily restrict >> your RBF attempts if you're only allowed to use confirmed inputs >> and have many channels (and a limited number of confirmed inputs). >> Otherwise you'll need node operators to pre-emptively split their >> utxos into many small utxos just for fee bumping, which is inefficient..= . >> >> Bastien >> >> Le lun. 27 sept. 2021 =C3=A0 00:27, Antoine Riard via bitcoin-dev < >> bitcoin-dev@lists.linuxfoundation.org> a =C3=A9crit : >> >>> Hi Gloria, >>> >>> Thanks for your answers, >>> >>> > In summary, it seems that the decisions that might still need >>> > attention/input from devs on this mailing list are: >>> > 1. Whether we should start with multiple-parent-1-child or >>> 1-parent-1-child. >>> > 2. Whether it's ok to require that the child not have conflicts with >>> > mempool transactions. >>> >>> Yes 1) it would be good to have inputs of more potential users of >>> package acceptance . And 2) I think it's more a matter of clearer wordi= ng >>> of the proposal. >>> >>> However, see my final point on the relaxation around "unconfirmed >>> inputs" which might in fact alter our current block construction strate= gy. >>> >>> > Right, the fact that we essentially always choose the first-seen >>> witness is >>> > an unfortunate limitation that exists already. Adding package mempool >>> > accept doesn't worsen this, but the procedure in the future is to >>> replace >>> > the witness when it makes sense economically. We can also add logic t= o >>> > allow package feerate to pay for witness replacements as well. This i= s >>> > pretty far into the future, though. >>> >>> Yes I agree package mempool doesn't worsen this. And it's not an issue >>> for current LN as you can't significantly inflate a spending witness fo= r >>> the 2-of-2 funding output. >>> However, it might be an issue for multi-party protocol where the >>> spending script has alternative branches with asymmetric valid witness >>> weights. Taproot should ease that kind of script so hopefully we would >>> deploy wtxid-replacement not too far in the future. >>> >>> > I could be misunderstanding, but an attacker wouldn't be able to >>> > batch-attack like this. Alice's package only conflicts with A' + D', >>> not A' >>> > + B' + C' + D'. She only needs to pay for evicting 2 transactions. >>> >>> Yeah I can be clearer, I think you have 2 pinning attacks scenarios to >>> consider. >>> >>> In LN, if you're trying to confirm a commitment transaction to time-out >>> or claim on-chain a HTLC and the timelock is near-expiration, you shoul= d be >>> ready to pay in commitment+2nd-stage HTLC transaction fees as much as t= he >>> value offered by the HTLC. >>> >>> Following this security assumption, an attacker can exploit it by >>> targeting together commitment transactions from different channels by >>> blocking them under a high-fee child, of which the fee value >>> is equal to the top-value HTLC + 1. Victims's fee-bumping logics won't >>> overbid as it's not worthy to offer fees beyond their competed HTLCs. A= part >>> from observing mempools state, victims can't learn they're targeted by = the >>> same attacker. >>> >>> To draw from the aforementioned topology, Mallory broadcasts A' + B' + >>> C' + D', where A' conflicts with Alice's P1, B' conflicts with Bob's P2= , C' >>> conflicts with Caroll's P3. Let's assume P1 is confirming the top-value >>> HTLC of the set. If D' fees is higher than P1 + 1, it won't be rational= for >>> Alice or Bob or Caroll to keep offering competing feerates. Mallory wil= l be >>> at loss on stealing P1, as she has paid more in fees but will realize a >>> gain on P2+P3. >>> >>> In this model, Alice is allowed to evict those 2 transactions (A' + D') >>> but as she is economically-bounded she won't succeed. >>> >>> Mallory is maliciously exploiting RBF rule 3 on absolute fee. I think >>> this 1st pinning scenario is correct and "lucractive" when you sum the >>> global gain/loss. >>> >>> There is a 2nd attack scenario where A + B + C + D, where D is the chil= d >>> of A,B,C. All those transactions are honestly issued by Alice. Once A += B + >>> C + D are propagated in network mempools, Mallory is able to replace A = + D >>> with A' + D' where D' is paying a higher fee. This package A' + D' wil= l >>> confirm soon if D feerate was compelling but Mallory succeeds in delayi= ng >>> the confirmation >>> of B + C for one or more blocks. As B + C are pre-signed commitments >>> with a low-fee rate they won't confirm without Alice issuing a new chil= d E. >>> Mallory can repeat the same trick by broadcasting >>> B' + E' and delay again the confirmation of C. >>> >>> If the remaining package pending HTLC has a higher-value than all the >>> malicious fees over-bid, Mallory should realize a gain. With this 2nd >>> pinning attack, the malicious entity buys confirmation delay of your >>> packaged-together commitments. >>> >>> Assuming those attacks are correct, I'm leaning towards being >>> conservative with the LDK broadcast backend. Though once again, other L= 2 >>> devs have likely other use-cases and opinions :) >>> >>> > B' only needs to pay for itself in this case. >>> >>> Yes I think it's a nice discount when UTXO is single-owned. In the >>> context of shared-owned UTXO (e.g LN), you might not if there is an >>> in-mempool package already spending the UTXO and have to assume the >>> worst-case scenario. I.e have B' committing enough fee to pay for A' >>> replacement bandwidth. I think we can't do that much for this case... >>> >>> > If a package meets feerate requirements as a >>> package, the parents in the transaction are allowed to replace-by-fee >>> mempool transactions. The child cannot replace mempool transactions." >>> >>> I agree with the Mallory-vs-Alice case. Though if Alice broadcasts A+B' >>> to replace A+B because the first broadcast isn't satisfying anymore due= to >>> mempool spikes ? Assuming B' fees is enough, I think that case as child= B' >>> replacing in-mempool transaction B. Which I understand going against "= The >>> child cannot replace mempool transactions". >>> >>> Maybe wording could be a bit clearer ? >>> >>> > While it would be nice to have full RBF, malleability of the child >>> won't >>> > block RBF here. If we're trying to replace A', we only require that A= ' >>> > signals replaceability, and don't mind if its child doesn't. >>> >>> Yes, it sounds good. >>> >>> > Yes, A+C+D pays 2500sat more in fees, but it is also 1000vB larger. A >>> miner >>> > should prefer to utilize their block space more effectively. >>> >>> If your mempool is empty and only composed of A+C+D or A+B, I think >>> taking A+C+D is the most efficient block construction you can come up w= ith >>> as a miner ? >>> >>> > No, because we don't use that model. >>> >>> Can you describe what miner model we are using ? Like the block >>> construction strategy implemented by `addPackagesTxs` or also encompass= ing >>> our current mempool acceptance policy, which I think rely on absolute f= ee >>> over ancestor score in case of replacement ? >>> >>> I think this point is worthy to discuss as otherwise we might downgrade >>> the efficiency of our current block construction strategy in periods of >>> near-empty mempools. A knowledge which could be discreetly leveraged by= a >>> miner to gain an advantage on the rest of the mining ecosystem. >>> >>> Note, I think we *might* have to go in this direction if we want to >>> replace replace-by-fee by replace-by-feerate or replace-by-ancestor and >>> solve in-depth pinning attacks. Though if we do so, >>> IMO we would need more thoughts. >>> >>> I think we could restrain package acceptance to only confirmed inputs >>> for now and revisit later this point ? For LN-anchor, you can assume th= at >>> the fee-bumping UTXO feeding the CPFP is already >>> confirmed. Or are there currently-deployed use-cases which would benefi= t >>> from your proposed Rule #2 ? >>> >>> Antoine >>> >>> Le jeu. 23 sept. 2021 =C3=A0 11:36, Gloria Zhao = a >>> =C3=A9crit : >>> >>>> Hi Antoine, >>>> >>>> Thanks as always for your input. I'm glad we agree on so much! >>>> >>>> In summary, it seems that the decisions that might still need >>>> attention/input from devs on this mailing list are: >>>> 1. Whether we should start with multiple-parent-1-child or >>>> 1-parent-1-child. >>>> 2. Whether it's ok to require that the child not have conflicts with >>>> mempool transactions. >>>> >>>> Responding to your comments... >>>> >>>> > IIUC, you have package A+B, during the dedup phase early in >>>> `AcceptMultipleTransactions` if you observe same-txid-different-wtixd = A' >>>> and A' is higher feerate than A, you trim A and replace by A' ? >>>> >>>> > I think this approach is safe, the one who appears unsafe to me is >>>> when A' has a _lower_ feerate, even if A' is already accepted by our >>>> mempool ? In that case iirc that would be a pinning. >>>> >>>> Right, the fact that we essentially always choose the first-seen >>>> witness is an unfortunate limitation that exists already. Adding packa= ge >>>> mempool accept doesn't worsen this, but the procedure in the future is= to >>>> replace the witness when it makes sense economically. We can also add = logic >>>> to allow package feerate to pay for witness replacements as well. This= is >>>> pretty far into the future, though. >>>> >>>> > It sounds uneconomical for an attacker but I think it's not when you >>>> consider than you can "batch" attack against multiple honest >>>> counterparties. E.g, Mallory broadcast A' + B' + C' + D' where A' conf= licts >>>> with Alice's honest package P1, B' conflicts with Bob's honest package= P2, >>>> C' conflicts with Caroll's honest package P3. And D' is a high-fee chi= ld of >>>> A' + B' + C'. >>>> >>>> > If D' is higher-fee than P1 or P2 or P3 but inferior to the sum of >>>> HTLCs confirmed by P1+P2+P3, I think it's lucrative for the attacker ? >>>> >>>> I could be misunderstanding, but an attacker wouldn't be able to >>>> batch-attack like this. Alice's package only conflicts with A' + D', n= ot A' >>>> + B' + C' + D'. She only needs to pay for evicting 2 transactions. >>>> >>>> > Do we assume that broadcasted packages are "honest" by default and >>>> that the parent(s) always need the child to pass the fee checks, that = way >>>> saving the processing of individual transactions which are expected to= fail >>>> in 99% of cases or more ad hoc composition of packages at relay ? >>>> > I think this point is quite dependent on the p2p packages >>>> format/logic we'll end up on and that we should feel free to revisit i= t >>>> later ? >>>> >>>> I think it's the opposite; there's no way for us to assume that p2p >>>> packages will be "honest." I'd like to have two things before we expos= e on >>>> P2P: (1) ensure that the amount of resources potentially allocated for >>>> package validation isn't disproportionately higher than that of single >>>> transaction validation and (2) only use package validation when we're >>>> unsatisifed with the single validation result, e.g. we might get bette= r >>>> fees. >>>> Yes, let's revisit this later :) >>>> >>>> > Yes, if you receive A+B, and A is already in-mempoo, I agree you ca= n >>>> discard its feerate as B should pay for all fees checked on its own. W= here >>>> I'm unclear is when you have in-mempool A+B and receive A+B'. Should B= ' >>>> have a fee high enough to cover the bandwidth penalty replacement >>>> (`PaysForRBF`, 2nd check) of both A+B' or only B' ? >>>> >>>> B' only needs to pay for itself in this case. >>>> >>>> > > Do we want the child to be able to replace mempool transactions as >>>> well? >>>> >>>> > If we mean when you have replaceable A+B then A'+B' try to replace >>>> with a higher-feerate ? I think that's exactly the case we need for >>>> Lightning as A+B is coming from Alice and A'+B' is coming from Bob :/ >>>> >>>> Let me clarify this because I can see that my wording was ambiguous, >>>> and then please let me know if it fits Lightning's needs? >>>> >>>> In my proposal, I wrote "If a package meets feerate requirements as a >>>> package, the parents in the transaction are allowed to replace-by-fee >>>> mempool transactions. The child cannot replace mempool transactions." = What >>>> I meant was: the package can replace mempool transactions if any of th= e >>>> parents conflict with mempool transactions. The child cannot not confl= ict >>>> with any mempool transactions. >>>> The Lightning use case this attempts to address is: Alice and Mallory >>>> are LN counterparties, and have packages A+B and A'+B', respectively. = A and >>>> A' are their commitment transactions and conflict with each other; the= y >>>> have shared inputs and different txids. >>>> B spends Alice's anchor output from A. B' spends Mallory's anchor >>>> output from A'. Thus, B and B' do not conflict with each other. >>>> Alice can broadcast her package, A+B, to replace Mallory's package, >>>> A'+B', since B doesn't conflict with the mempool. >>>> >>>> Would this be ok? >>>> >>>> > The second option, a child of A', In the LN case I think the CPFP is >>>> attached on one's anchor output. >>>> >>>> While it would be nice to have full RBF, malleability of the child >>>> won't block RBF here. If we're trying to replace A', we only require t= hat >>>> A' signals replaceability, and don't mind if its child doesn't. >>>> >>>> > > B has an ancestor score of 10sat/vb and D has an >>>> > > ancestor score of ~2.9sat/vb. Since D's ancestor score is lower >>>> than B's, >>>> > > it fails the proposed package RBF Rule #2, so this package would b= e >>>> > > rejected. Does this meet your expectations? >>>> >>>> > Well what sounds odd to me, in my example, we fail D even if it has = a >>>> higher-fee than B. Like A+B absolute fees are 2000 sats and A+C+D abso= lute >>>> fees are 4500 sats ? >>>> >>>> Yes, A+C+D pays 2500sat more in fees, but it is also 1000vB larger. A >>>> miner should prefer to utilize their block space more effectively. >>>> >>>> > Is this compatible with a model where a miner prioritizes absolute >>>> fees over ancestor score, in the case that mempools aren't full-enough= to >>>> fulfill a block ? >>>> >>>> No, because we don't use that model. >>>> >>>> Thanks, >>>> Gloria >>>> >>>> On Thu, Sep 23, 2021 at 5:29 AM Antoine Riard >>>> wrote: >>>> >>>>> > Correct, if B+C is too low feerate to be accepted, we will reject >>>>> it. I >>>>> > prefer this because it is incentive compatible: A can be mined by >>>>> itself, >>>>> > so there's no reason to prefer A+B+C instead of A. >>>>> > As another way of looking at this, consider the case where we do >>>>> accept >>>>> > A+B+C and it sits at the "bottom" of our mempool. If our mempool >>>>> reaches >>>>> > capacity, we evict the lowest descendant feerate transactions, whic= h >>>>> are >>>>> > B+C in this case. This gives us the same resulting mempool, with A >>>>> and not >>>>> > B+C. >>>>> >>>>> I agree here. Doing otherwise, we might evict other transactions >>>>> mempool in `MempoolAccept::Finalize` with a higher-feerate than B+C w= hile >>>>> those evicted transactions are the most compelling for block construc= tion. >>>>> >>>>> I thought at first missing this acceptance requirement would break a >>>>> fee-bumping scheme like Parent-Pay-For-Child where a high-fee parent = is >>>>> attached to a child signed with SIGHASH_ANYONECANPAY but in this case= the >>>>> child fee is capturing the parent value. I can't think of other fee-b= umping >>>>> schemes potentially affected. If they do exist I would say they're wr= ong in >>>>> their design assumptions. >>>>> >>>>> > If or when we have witness replacement, the logic is: if the >>>>> individual >>>>> > transaction is enough to replace the mempool one, the replacement >>>>> will >>>>> > happen during the preceding individual transaction acceptance, and >>>>> > deduplication logic will work. Otherwise, we will try to deduplicat= e >>>>> by >>>>> > wtxid, see that we need a package witness replacement, and use the >>>>> package >>>>> > feerate to evaluate whether this is economically rational. >>>>> >>>>> IIUC, you have package A+B, during the dedup phase early in >>>>> `AcceptMultipleTransactions` if you observe same-txid-different-wtixd= A' >>>>> and A' is higher feerate than A, you trim A and replace by A' ? >>>>> >>>>> I think this approach is safe, the one who appears unsafe to me is >>>>> when A' has a _lower_ feerate, even if A' is already accepted by our >>>>> mempool ? In that case iirc that would be a pinning. >>>>> >>>>> Good to see progress on witness replacement before we see usage of >>>>> Taproot tree in the context of multi-party, where a malicious counter= party >>>>> inflates its witness to jam a honest spending. >>>>> >>>>> (Note, the commit linked currently points nowhere :)) >>>>> >>>>> >>>>> > Please note that A may replace A' even if A' has higher fees than A >>>>> > individually, because the proposed package RBF utilizes the fees an= d >>>>> size >>>>> > of the entire package. This just requires E to pay enough fees, >>>>> although >>>>> > this can be pretty high if there are also potential B' and C' >>>>> competing >>>>> > commitment transactions that we don't know about. >>>>> >>>>> Ah right, if the package acceptance waives `PaysMoreThanConflicts` fo= r >>>>> the individual check on A, the honest package should replace the pinn= ing >>>>> attempt. I've not fully parsed the proposed implementation yet. >>>>> >>>>> Though note, I think it's still unsafe for a Lightning >>>>> multi-commitment-broadcast-as-one-package as a malicious A' might hav= e an >>>>> absolute fee higher than E. It sounds uneconomical for >>>>> an attacker but I think it's not when you consider than you can >>>>> "batch" attack against multiple honest counterparties. E.g, Mallory >>>>> broadcast A' + B' + C' + D' where A' conflicts with Alice's honest pa= ckage >>>>> P1, B' conflicts with Bob's honest package P2, C' conflicts with Caro= ll's >>>>> honest package P3. And D' is a high-fee child of A' + B' + C'. >>>>> >>>>> If D' is higher-fee than P1 or P2 or P3 but inferior to the sum of >>>>> HTLCs confirmed by P1+P2+P3, I think it's lucrative for the attacker = ? >>>>> >>>>> > So far, my understanding is that multi-parent-1-child is desired fo= r >>>>> > batched fee-bumping ( >>>>> > https://github.com/bitcoin/bitcoin/pull/22674#issuecomment-89795128= 9) >>>>> and >>>>> > I've also seen your response which I have less context on ( >>>>> > https://github.com/bitcoin/bitcoin/pull/22674#issuecomment-90035220= 2). >>>>> That >>>>> > being said, I am happy to create a new proposal for 1 parent + 1 >>>>> child >>>>> > (which would be slightly simpler) and plan for moving to >>>>> > multi-parent-1-child later if that is preferred. I am very >>>>> interested in >>>>> > hearing feedback on that approach. >>>>> >>>>> I think batched fee-bumping is okay as long as you don't have >>>>> time-sensitive outputs encumbering your commitment transactions. For = the >>>>> reasons mentioned above, I think that's unsafe. >>>>> >>>>> What I'm worried about is L2 developers, potentially not aware about >>>>> all the mempool subtleties blurring the difference and always batchin= g >>>>> their broadcast by default. >>>>> >>>>> IMO, a good thing by restraining to 1-parent + 1 child, we >>>>> artificially constraint L2 design space for now and minimize risks of >>>>> unsafe usage of the package API :) >>>>> >>>>> I think that's a point where it would be relevant to have the opinion >>>>> of more L2 devs. >>>>> >>>>> > I think there is a misunderstanding here - let me describe what I'm >>>>> > proposing we'd do in this situation: we'll try individual submissio= n >>>>> for A, >>>>> > see that it fails due to "insufficient fees." Then, we'll try packa= ge >>>>> > validation for A+B and use package RBF. If A+B pays enough, it can >>>>> still >>>>> > replace A'. If A fails for a bad signature, we won't look at B or >>>>> A+B. Does >>>>> > this meet your expectations? >>>>> >>>>> Yes there was a misunderstanding, I think this approach is correct, >>>>> it's more a question of performance. Do we assume that broadcasted pa= ckages >>>>> are "honest" by default and that the parent(s) always need the child = to >>>>> pass the fee checks, that way saving the processing of individual >>>>> transactions which are expected to fail in 99% of cases or more ad ho= c >>>>> composition of packages at relay ? >>>>> >>>>> I think this point is quite dependent on the p2p packages format/logi= c >>>>> we'll end up on and that we should feel free to revisit it later ? >>>>> >>>>> >>>>> > What problem are you trying to solve by the package feerate *after* >>>>> dedup >>>>> rule ? >>>>> > My understanding is that an in-package transaction might be already >>>>> in >>>>> the mempool. Therefore, to compute a correct RBF penalty replacement, >>>>> the >>>>> vsize of this transaction could be discarded lowering the cost of >>>>> package >>>>> RBF. >>>>> >>>>> > I'm proposing that, when a transaction has already been submitted t= o >>>>> > mempool, we would ignore both its fees and vsize when calculating >>>>> package >>>>> > feerate. >>>>> >>>>> Yes, if you receive A+B, and A is already in-mempoo, I agree you can >>>>> discard its feerate as B should pay for all fees checked on its own. = Where >>>>> I'm unclear is when you have in-mempool A+B and receive A+B'. Should = B' >>>>> have a fee high enough to cover the bandwidth penalty replacement >>>>> (`PaysForRBF`, 2nd check) of both A+B' or only B' ? >>>>> >>>>> If you have a second-layer like current Lightning, you might have a >>>>> counterparty commitment to replace and should always expect to have t= o pay >>>>> for parent replacement bandwidth. >>>>> >>>>> Where a potential discount sounds interesting is when you have an >>>>> univoque state on the first-stage of transactions. E.g DLC's funding >>>>> transaction which might be CPFP by any participant iirc. >>>>> >>>>> > Note that, if C' conflicts with C, it also conflicts with D, since = D >>>>> is a >>>>> > descendant of C and would thus need to be evicted along with it. >>>>> >>>>> Ah once again I think it's a misunderstanding without the code under >>>>> my eyes! If we do C' `PreChecks`, solve the conflicts provoked by it,= i.e >>>>> mark for potential eviction D and don't consider it for future confli= cts in >>>>> the rest of the package, I think D' `PreChecks` should be good ? >>>>> >>>>> > More generally, this example is surprising to me because I didn't >>>>> think >>>>> > packages would be used to fee-bump replaceable transactions. Do we >>>>> want the >>>>> > child to be able to replace mempool transactions as well? >>>>> >>>>> If we mean when you have replaceable A+B then A'+B' try to replace >>>>> with a higher-feerate ? I think that's exactly the case we need for >>>>> Lightning as A+B is coming from Alice and A'+B' is coming from Bob :/ >>>>> >>>>> > I'm not sure what you mean? Let's say we have a package of parent A >>>>> + child >>>>> > B, where A is supposed to replace a mempool transaction A'. Are you >>>>> saying >>>>> > that counterparties are able to malleate the package child B, or a >>>>> child of >>>>> > A'? >>>>> >>>>> The second option, a child of A', In the LN case I think the CPFP is >>>>> attached on one's anchor output. >>>>> >>>>> I think it's good if we assume the >>>>> solve-conflicts-after-parent's`'PreChecks` mentioned above or fixing >>>>> inherited signaling or full-rbf ? >>>>> >>>>> > Sorry, I don't understand what you mean by "preserve the package >>>>> > integrity?" Could you elaborate? >>>>> >>>>> After thinking the relaxation about the "new" unconfirmed input is no= t >>>>> linked to trimming but I would say more to the multi-parent support. >>>>> >>>>> Let's say you have A+B trying to replace C+D where B is also spending >>>>> already in-mempool E. To succeed, you need to waive the no-new-unconf= irmed >>>>> input as D isn't spending E. >>>>> >>>>> So good, I think we agree on the problem description here. >>>>> >>>>> > I am in agreement with your calculations but unsure if we disagree >>>>> on the >>>>> > expected outcome. Yes, B has an ancestor score of 10sat/vb and D ha= s >>>>> an >>>>> > ancestor score of ~2.9sat/vb. Since D's ancestor score is lower tha= n >>>>> B's, >>>>> > it fails the proposed package RBF Rule #2, so this package would be >>>>> > rejected. Does this meet your expectations? >>>>> >>>>> Well what sounds odd to me, in my example, we fail D even if it has a >>>>> higher-fee than B. Like A+B absolute fees are 2000 sats and A+C+D abs= olute >>>>> fees are 4500 sats ? >>>>> >>>>> Is this compatible with a model where a miner prioritizes absolute >>>>> fees over ancestor score, in the case that mempools aren't full-enoug= h to >>>>> fulfill a block ? >>>>> >>>>> Let me know if I can clarify a point. >>>>> >>>>> Antoine >>>>> >>>>> Le lun. 20 sept. 2021 =C3=A0 11:10, Gloria Zhao a >>>>> =C3=A9crit : >>>>> >>>>>> >>>>>> Hi Antoine, >>>>>> >>>>>> First of all, thank you for the thorough review. I appreciate your >>>>>> insight on LN requirements. >>>>>> >>>>>> > IIUC, you have a package A+B+C submitted for acceptance and A is >>>>>> already in your mempool. You trim out A from the package and then ev= aluate >>>>>> B+C. >>>>>> >>>>>> > I think this might be an issue if A is the higher-fee element of >>>>>> the ABC package. B+C package fees might be under the mempool min fee= and >>>>>> will be rejected, potentially breaking the acceptance expectations o= f the >>>>>> package issuer ? >>>>>> >>>>>> Correct, if B+C is too low feerate to be accepted, we will reject it= . >>>>>> I prefer this because it is incentive compatible: A can be mined by = itself, >>>>>> so there's no reason to prefer A+B+C instead of A. >>>>>> As another way of looking at this, consider the case where we do >>>>>> accept A+B+C and it sits at the "bottom" of our mempool. If our memp= ool >>>>>> reaches capacity, we evict the lowest descendant feerate transaction= s, >>>>>> which are B+C in this case. This gives us the same resulting mempool= , with >>>>>> A and not B+C. >>>>>> >>>>>> >>>>>> > Further, I think the dedup should be done on wtxid, as you might >>>>>> have multiple valid witnesses. Though with varying vsizes and as suc= h >>>>>> offering different feerates. >>>>>> >>>>>> I agree that variations of the same package with different witnesses >>>>>> is a case that must be handled. I consider witness replacement to be= a >>>>>> project that can be done in parallel to package mempool acceptance b= ecause >>>>>> being able to accept packages does not worsen the problem of a >>>>>> same-txid-different-witness "pinning" attack. >>>>>> >>>>>> If or when we have witness replacement, the logic is: if the >>>>>> individual transaction is enough to replace the mempool one, the >>>>>> replacement will happen during the preceding individual transaction >>>>>> acceptance, and deduplication logic will work. Otherwise, we will tr= y to >>>>>> deduplicate by wtxid, see that we need a package witness replacement= , and >>>>>> use the package feerate to evaluate whether this is economically rat= ional. >>>>>> >>>>>> See the #22290 "handle package transactions already in mempool" >>>>>> commit ( >>>>>> https://github.com/bitcoin/bitcoin/pull/22290/commits/fea75a2237b46c= f76145242fecad7e274bfcb5ff), >>>>>> which handles the case of same-txid-different-witness by simply usin= g the >>>>>> transaction in the mempool for now, with TODOs for what I just descr= ibed. >>>>>> >>>>>> >>>>>> > I'm not clearly understanding the accepted topologies. By "parent >>>>>> and child to share a parent", do you mean the set of transactions A,= B, C, >>>>>> where B is spending A and C is spending A and B would be correct ? >>>>>> >>>>>> Yes, that is what I meant. Yes, that would a valid package under >>>>>> these rules. >>>>>> >>>>>> > If yes, is there a width-limit introduced or we fallback on >>>>>> MAX_PACKAGE_COUNT=3D25 ? >>>>>> >>>>>> No, there is no limit on connectivity other than "child with all >>>>>> unconfirmed parents." We will enforce MAX_PACKAGE_COUNT=3D25 and chi= ld's >>>>>> in-mempool + in-package ancestor limits. >>>>>> >>>>>> >>>>>> > Considering the current Core's mempool acceptance rules, I think >>>>>> CPFP batching is unsafe for LN time-sensitive closure. A malicious t= x-relay >>>>>> jamming successful on one channel commitment transaction would conta= mine >>>>>> the remaining commitments sharing the same package. >>>>>> >>>>>> > E.g, you broadcast the package A+B+C+D+E where A,B,C,D are >>>>>> commitment transactions and E a shared CPFP. If a malicious A' trans= action >>>>>> has a better feerate than A, the whole package acceptance will fail.= Even >>>>>> if A' confirms in the following block, >>>>>> the propagation and confirmation of B+C+D have been delayed. This >>>>>> could carry on a loss of funds. >>>>>> >>>>>> Please note that A may replace A' even if A' has higher fees than A >>>>>> individually, because the proposed package RBF utilizes the fees and= size >>>>>> of the entire package. This just requires E to pay enough fees, alth= ough >>>>>> this can be pretty high if there are also potential B' and C' compet= ing >>>>>> commitment transactions that we don't know about. >>>>>> >>>>>> >>>>>> > IMHO, I'm leaning towards deploying during a first phase >>>>>> 1-parent/1-child. I think it's the most conservative step still impr= oving >>>>>> second-layer safety. >>>>>> >>>>>> So far, my understanding is that multi-parent-1-child is desired for >>>>>> batched fee-bumping ( >>>>>> https://github.com/bitcoin/bitcoin/pull/22674#issuecomment-897951289= ) >>>>>> and I've also seen your response which I have less context on ( >>>>>> https://github.com/bitcoin/bitcoin/pull/22674#issuecomment-900352202= ). >>>>>> That being said, I am happy to create a new proposal for 1 parent + = 1 child >>>>>> (which would be slightly simpler) and plan for moving to >>>>>> multi-parent-1-child later if that is preferred. I am very intereste= d in >>>>>> hearing feedback on that approach. >>>>>> >>>>>> >>>>>> > If A+B is submitted to replace A', where A pays 0 sats, B pays 200 >>>>>> sats and A' pays 100 sats. If we apply the individual RBF on A, A+B >>>>>> acceptance fails. For this reason I think the individual RBF should = be >>>>>> bypassed and only the package RBF apply ? >>>>>> >>>>>> I think there is a misunderstanding here - let me describe what I'm >>>>>> proposing we'd do in this situation: we'll try individual submission= for A, >>>>>> see that it fails due to "insufficient fees." Then, we'll try packag= e >>>>>> validation for A+B and use package RBF. If A+B pays enough, it can s= till >>>>>> replace A'. If A fails for a bad signature, we won't look at B or A+= B. Does >>>>>> this meet your expectations? >>>>>> >>>>>> >>>>>> > What problem are you trying to solve by the package feerate *after= * >>>>>> dedup rule ? >>>>>> > My understanding is that an in-package transaction might be alread= y >>>>>> in the mempool. Therefore, to compute a correct RBF penalty replacem= ent, >>>>>> the vsize of this transaction could be discarded lowering the cost o= f >>>>>> package RBF. >>>>>> >>>>>> I'm proposing that, when a transaction has already been submitted to >>>>>> mempool, we would ignore both its fees and vsize when calculating pa= ckage >>>>>> feerate. In example G2, we shouldn't count M1 fees after its submiss= ion to >>>>>> mempool, since M1's fees have already been used to pay for its indiv= idual >>>>>> bandwidth, and it shouldn't be used again to pay for P2 and P3's ban= dwidth. >>>>>> We also shouldn't count its vsize, since it has already been paid fo= r. >>>>>> >>>>>> >>>>>> > I think this is a footgunish API, as if a package issuer send the >>>>>> multiple-parent-one-child package A,B,C,D where D is the child of A,= B,C. >>>>>> Then try to broadcast the higher-feerate C'+D' package, it should be >>>>>> rejected. So it's breaking the naive broadcaster assumption that a >>>>>> higher-feerate/higher-fee package always replaces ? >>>>>> >>>>>> Note that, if C' conflicts with C, it also conflicts with D, since D >>>>>> is a descendant of C and would thus need to be evicted along with it= . >>>>>> Implicitly, D' would not be in conflict with D. >>>>>> More generally, this example is surprising to me because I didn't >>>>>> think packages would be used to fee-bump replaceable transactions. D= o we >>>>>> want the child to be able to replace mempool transactions as well? T= his can >>>>>> be implemented with a bit of additional logic. >>>>>> >>>>>> > I think this is unsafe for L2s if counterparties have malleability >>>>>> of the child transaction. They can block your package replacement by >>>>>> opting-out from RBF signaling. IIRC, LN's "anchor output" presents s= uch an >>>>>> ability. >>>>>> >>>>>> I'm not sure what you mean? Let's say we have a package of parent A = + >>>>>> child B, where A is supposed to replace a mempool transaction A'. Ar= e you >>>>>> saying that counterparties are able to malleate the package child B,= or a >>>>>> child of A'? If they can malleate a child of A', that shouldn't matt= er as >>>>>> long as A' is signaling replacement. This would be handled identical= ly with >>>>>> full RBF and what Core currently implements. >>>>>> >>>>>> > I think this is an issue brought by the trimming during the dedup >>>>>> phase. If we preserve the package integrity, only re-using the tx-le= vel >>>>>> checks results of already in-mempool transactions to gain in CPU tim= e we >>>>>> won't have this issue. Package childs can add unconfirmed inputs as = long as >>>>>> they're in-package, the bip125 rule2 is only evaluated against paren= ts ? >>>>>> >>>>>> Sorry, I don't understand what you mean by "preserve the package >>>>>> integrity?" Could you elaborate? >>>>>> >>>>>> > Let's say you have in-mempool A, B where A pays 10 sat/vb for 100 >>>>>> vbytes and B pays 10 sat/vb for 100 vbytes. You have the candidate >>>>>> replacement D spending both A and C where D pays 15sat/vb for 100 vb= ytes >>>>>> and C pays 1 sat/vb for 1000 vbytes. >>>>>> >>>>>> > Package A + B ancestor score is 10 sat/vb. >>>>>> >>>>>> > D has a higher feerate/absolute fee than B. >>>>>> >>>>>> > Package A + C + D ancestor score is ~ 3 sat/vb ((A's 1000 sats + >>>>>> C's 1000 sats + D's 1500 sats) / A's 100 vb + C's 1000 vb + D's 100 = vb) >>>>>> >>>>>> I am in agreement with your calculations but unsure if we disagree o= n >>>>>> the expected outcome. Yes, B has an ancestor score of 10sat/vb and D= has an >>>>>> ancestor score of ~2.9sat/vb. Since D's ancestor score is lower than= B's, >>>>>> it fails the proposed package RBF Rule #2, so this package would be >>>>>> rejected. Does this meet your expectations? >>>>>> >>>>>> Thank you for linking to projects that might be interested in packag= e >>>>>> relay :) >>>>>> >>>>>> Thanks, >>>>>> Gloria >>>>>> >>>>>> On Mon, Sep 20, 2021 at 12:16 AM Antoine Riard < >>>>>> antoine.riard@gmail.com> wrote: >>>>>> >>>>>>> Hi Gloria, >>>>>>> >>>>>>> > A package may contain transactions that are already in the >>>>>>> mempool. We >>>>>>> > remove >>>>>>> > ("deduplicate") those transactions from the package for the >>>>>>> purposes of >>>>>>> > package >>>>>>> > mempool acceptance. If a package is empty after deduplication, we >>>>>>> do >>>>>>> > nothing. >>>>>>> >>>>>>> IIUC, you have a package A+B+C submitted for acceptance and A is >>>>>>> already in your mempool. You trim out A from the package and then e= valuate >>>>>>> B+C. >>>>>>> >>>>>>> I think this might be an issue if A is the higher-fee element of th= e >>>>>>> ABC package. B+C package fees might be under the mempool min fee an= d will >>>>>>> be rejected, potentially breaking the acceptance expectations of th= e >>>>>>> package issuer ? >>>>>>> >>>>>>> Further, I think the dedup should be done on wtxid, as you might >>>>>>> have multiple valid witnesses. Though with varying vsizes and as su= ch >>>>>>> offering different feerates. >>>>>>> >>>>>>> E.g you're going to evaluate the package A+B and A' is already in >>>>>>> your mempool with a bigger valid witness. You trim A based on txid,= then >>>>>>> you evaluate A'+B, which fails the fee checks. However, evaluating = A+B >>>>>>> would have been a success. >>>>>>> >>>>>>> AFAICT, the dedup rationale would be to save on CPU time/IO disk, t= o >>>>>>> avoid repeated signatures verification and parent UTXOs fetches ? C= an we >>>>>>> achieve the same goal by bypassing tx-level checks for already-in t= xn while >>>>>>> conserving the package integrity for package-level checks ? >>>>>>> >>>>>>> > Note that it's possible for the parents to be >>>>>>> > indirect >>>>>>> > descendants/ancestors of one another, or for parent and child to >>>>>>> share a >>>>>>> > parent, >>>>>>> > so we cannot make any other topology assumptions. >>>>>>> >>>>>>> I'm not clearly understanding the accepted topologies. By "parent >>>>>>> and child to share a parent", do you mean the set of transactions A= , B, C, >>>>>>> where B is spending A and C is spending A and B would be correct ? >>>>>>> >>>>>>> If yes, is there a width-limit introduced or we fallback on >>>>>>> MAX_PACKAGE_COUNT=3D25 ? >>>>>>> >>>>>>> IIRC, one rationale to come with this topology limitation was to >>>>>>> lower the DoS risks when potentially deploying p2p packages. >>>>>>> >>>>>>> Considering the current Core's mempool acceptance rules, I think >>>>>>> CPFP batching is unsafe for LN time-sensitive closure. A malicious = tx-relay >>>>>>> jamming successful on one channel commitment transaction would cont= amine >>>>>>> the remaining commitments sharing the same package. >>>>>>> >>>>>>> E.g, you broadcast the package A+B+C+D+E where A,B,C,D are >>>>>>> commitment transactions and E a shared CPFP. If a malicious A' tran= saction >>>>>>> has a better feerate than A, the whole package acceptance will fail= . Even >>>>>>> if A' confirms in the following block, >>>>>>> the propagation and confirmation of B+C+D have been delayed. This >>>>>>> could carry on a loss of funds. >>>>>>> >>>>>>> That said, if you're broadcasting commitment transactions without >>>>>>> time-sensitive HTLC outputs, I think the batching is effectively a = fee >>>>>>> saving as you don't have to duplicate the CPFP. >>>>>>> >>>>>>> IMHO, I'm leaning towards deploying during a first phase >>>>>>> 1-parent/1-child. I think it's the most conservative step still imp= roving >>>>>>> second-layer safety. >>>>>>> >>>>>>> > *Rationale*: It would be incorrect to use the fees of >>>>>>> transactions that are >>>>>>> > already in the mempool, as we do not want a transaction's fees to >>>>>>> be >>>>>>> > double-counted for both its individual RBF and package RBF. >>>>>>> >>>>>>> I'm unsure about the logical order of the checks proposed. >>>>>>> >>>>>>> If A+B is submitted to replace A', where A pays 0 sats, B pays 200 >>>>>>> sats and A' pays 100 sats. If we apply the individual RBF on A, A+B >>>>>>> acceptance fails. For this reason I think the individual RBF should= be >>>>>>> bypassed and only the package RBF apply ? >>>>>>> >>>>>>> Note this situation is plausible, with current LN design, your >>>>>>> counterparty can have a commitment transaction with a better fee ju= st by >>>>>>> selecting a higher `dust_limit_satoshis` than yours. >>>>>>> >>>>>>> > Examples F and G [14] show the same package, but P1 is submitted >>>>>>> > individually before >>>>>>> > the package in example G. In example F, we can see that the 300vB >>>>>>> package >>>>>>> > pays >>>>>>> > an additional 200sat in fees, which is not enough to pay for its >>>>>>> own >>>>>>> > bandwidth >>>>>>> > (BIP125#4). In example G, we can see that P1 pays enough to >>>>>>> replace M1, but >>>>>>> > using P1's fees again during package submission would make it loo= k >>>>>>> like a >>>>>>> > 300sat >>>>>>> > increase for a 200vB package. Even including its fees and size >>>>>>> would not be >>>>>>> > sufficient in this example, since the 300sat looks like enough fo= r >>>>>>> the 300vB >>>>>>> > package. The calculcation after deduplication is 100sat increase >>>>>>> for a >>>>>>> > package >>>>>>> > of size 200vB, which correctly fails BIP125#4. Assume all >>>>>>> transactions have >>>>>>> > a >>>>>>> > size of 100vB. >>>>>>> >>>>>>> What problem are you trying to solve by the package feerate *after* >>>>>>> dedup rule ? >>>>>>> >>>>>>> My understanding is that an in-package transaction might be already >>>>>>> in the mempool. Therefore, to compute a correct RBF penalty replace= ment, >>>>>>> the vsize of this transaction could be discarded lowering the cost = of >>>>>>> package RBF. >>>>>>> >>>>>>> If we keep a "safe" dedup mechanism (see my point above), I think >>>>>>> this discount is justified, as the validation cost of node operator= s is >>>>>>> paid for ? >>>>>>> >>>>>>> > The child cannot replace mempool transactions. >>>>>>> >>>>>>> Let's say you issue package A+B, then package C+B', where B' is a >>>>>>> child of both A and C. This rule fails the acceptance of C+B' ? >>>>>>> >>>>>>> I think this is a footgunish API, as if a package issuer send the >>>>>>> multiple-parent-one-child package A,B,C,D where D is the child of A= ,B,C. >>>>>>> Then try to broadcast the higher-feerate C'+D' package, it should b= e >>>>>>> rejected. So it's breaking the naive broadcaster assumption that a >>>>>>> higher-feerate/higher-fee package always replaces ? And it might be= unsafe >>>>>>> in protocols where states are symmetric. E.g a malicious counterpar= ty >>>>>>> broadcasts first S+A, then you honestly broadcast S+B, where B pays= better >>>>>>> fees. >>>>>>> >>>>>>> > All mempool transactions to be replaced must signal replaceabilit= y. >>>>>>> >>>>>>> I think this is unsafe for L2s if counterparties have malleability >>>>>>> of the child transaction. They can block your package replacement b= y >>>>>>> opting-out from RBF signaling. IIRC, LN's "anchor output" presents = such an >>>>>>> ability. >>>>>>> >>>>>>> I think it's better to either fix inherited signaling or move >>>>>>> towards full-rbf. >>>>>>> >>>>>>> > if a package parent has already been submitted, it would >>>>>>> > look >>>>>>> >like the child is spending a "new" unconfirmed input. >>>>>>> >>>>>>> I think this is an issue brought by the trimming during the dedup >>>>>>> phase. If we preserve the package integrity, only re-using the tx-l= evel >>>>>>> checks results of already in-mempool transactions to gain in CPU ti= me we >>>>>>> won't have this issue. Package childs can add unconfirmed inputs as= long as >>>>>>> they're in-package, the bip125 rule2 is only evaluated against pare= nts ? >>>>>>> >>>>>>> > However, we still achieve the same goal of requiring the >>>>>>> > replacement >>>>>>> > transactions to have a ancestor score at least as high as the >>>>>>> original >>>>>>> > ones. >>>>>>> >>>>>>> I'm not sure if this holds... >>>>>>> >>>>>>> Let's say you have in-mempool A, B where A pays 10 sat/vb for 100 >>>>>>> vbytes and B pays 10 sat/vb for 100 vbytes. You have the candidate >>>>>>> replacement D spending both A and C where D pays 15sat/vb for 100 v= bytes >>>>>>> and C pays 1 sat/vb for 1000 vbytes. >>>>>>> >>>>>>> Package A + B ancestor score is 10 sat/vb. >>>>>>> >>>>>>> D has a higher feerate/absolute fee than B. >>>>>>> >>>>>>> Package A + C + D ancestor score is ~ 3 sat/vb ((A's 1000 sats + C'= s >>>>>>> 1000 sats + D's 1500 sats) / >>>>>>> A's 100 vb + C's 1000 vb + D's 100 vb) >>>>>>> >>>>>>> Overall, this is a review through the lenses of LN requirements. I >>>>>>> think other L2 protocols/applications >>>>>>> could be candidates to using package accept/relay such as: >>>>>>> * https://github.com/lightninglabs/pool >>>>>>> * https://github.com/discreetlogcontracts/dlcspecs >>>>>>> * https://github.com/bitcoin-teleport/teleport-transactions/ >>>>>>> * https://github.com/sapio-lang/sapio >>>>>>> * >>>>>>> https://github.com/commerceblock/mercury/blob/master/doc/statechain= s.md >>>>>>> * https://github.com/revault/practical-revault >>>>>>> >>>>>>> Thanks for rolling forward the ball on this subject. >>>>>>> >>>>>>> Antoine >>>>>>> >>>>>>> Le jeu. 16 sept. 2021 =C3=A0 03:55, Gloria Zhao via bitcoin-dev < >>>>>>> bitcoin-dev@lists.linuxfoundation.org> a =C3=A9crit : >>>>>>> >>>>>>>> Hi there, >>>>>>>> >>>>>>>> I'm writing to propose a set of mempool policy changes to enable >>>>>>>> package >>>>>>>> validation (in preparation for package relay) in Bitcoin Core. >>>>>>>> These would not >>>>>>>> be consensus or P2P protocol changes. However, since mempool polic= y >>>>>>>> significantly affects transaction propagation, I believe this is >>>>>>>> relevant for >>>>>>>> the mailing list. >>>>>>>> >>>>>>>> My proposal enables packages consisting of multiple parents and 1 >>>>>>>> child. If you >>>>>>>> develop software that relies on specific transaction relay >>>>>>>> assumptions and/or >>>>>>>> are interested in using package relay in the future, I'm very >>>>>>>> interested to hear >>>>>>>> your feedback on the utility or restrictiveness of these package >>>>>>>> policies for >>>>>>>> your use cases. >>>>>>>> >>>>>>>> A draft implementation of this proposal can be found in [Bitcoin >>>>>>>> Core >>>>>>>> PR#22290][1]. >>>>>>>> >>>>>>>> An illustrated version of this post can be found at >>>>>>>> https://gist.github.com/glozow/dc4e9d5c5b14ade7cdfac40f43adb18a. >>>>>>>> I have also linked the images below. >>>>>>>> >>>>>>>> ## Background >>>>>>>> >>>>>>>> Feel free to skip this section if you are already familiar with >>>>>>>> mempool policy >>>>>>>> and package relay terminology. >>>>>>>> >>>>>>>> ### Terminology Clarifications >>>>>>>> >>>>>>>> * Package =3D an ordered list of related transactions, representab= le >>>>>>>> by a Directed >>>>>>>> Acyclic Graph. >>>>>>>> * Package Feerate =3D the total modified fees divided by the total >>>>>>>> virtual size of >>>>>>>> all transactions in the package. >>>>>>>> - Modified fees =3D a transaction's base fees + fee delta appl= ied >>>>>>>> by the user >>>>>>>> with `prioritisetransaction`. As such, we expect this to var= y >>>>>>>> across >>>>>>>> mempools. >>>>>>>> - Virtual Size =3D the maximum of virtual sizes calculated usi= ng >>>>>>>> [BIP141 >>>>>>>> virtual size][2] and sigop weight. [Implemented here in >>>>>>>> Bitcoin Core][3]. >>>>>>>> - Note that feerate is not necessarily based on the base fees >>>>>>>> and serialized >>>>>>>> size. >>>>>>>> >>>>>>>> * Fee-Bumping =3D user/wallet actions that take advantage of miner >>>>>>>> incentives to >>>>>>>> boost a transaction's candidacy for inclusion in a block, >>>>>>>> including Child Pays >>>>>>>> for Parent (CPFP) and [BIP125][12] Replace-by-Fee (RBF). Our >>>>>>>> intention in >>>>>>>> mempool policy is to recognize when the new transaction is more >>>>>>>> economical to >>>>>>>> mine than the original one(s) but not open DoS vectors, so there >>>>>>>> are some >>>>>>>> limitations. >>>>>>>> >>>>>>>> ### Policy >>>>>>>> >>>>>>>> The purpose of the mempool is to store the best (to be most >>>>>>>> incentive-compatible >>>>>>>> with miners, highest feerate) candidates for inclusion in a block. >>>>>>>> Miners use >>>>>>>> the mempool to build block templates. The mempool is also useful a= s >>>>>>>> a cache for >>>>>>>> boosting block relay and validation performance, aiding transactio= n >>>>>>>> relay, and >>>>>>>> generating feerate estimations. >>>>>>>> >>>>>>>> Ideally, all consensus-valid transactions paying reasonable fees >>>>>>>> should make it >>>>>>>> to miners through normal transaction relay, without any special >>>>>>>> connectivity or >>>>>>>> relationships with miners. On the other hand, nodes do not have >>>>>>>> unlimited >>>>>>>> resources, and a P2P network designed to let any honest node >>>>>>>> broadcast their >>>>>>>> transactions also exposes the transaction validation engine to DoS >>>>>>>> attacks from >>>>>>>> malicious peers. >>>>>>>> >>>>>>>> As such, for unconfirmed transactions we are considering for our >>>>>>>> mempool, we >>>>>>>> apply a set of validation rules in addition to consensus, primaril= y >>>>>>>> to protect >>>>>>>> us from resource exhaustion and aid our efforts to keep the highes= t >>>>>>>> fee >>>>>>>> transactions. We call this mempool _policy_: a set of (configurabl= e, >>>>>>>> node-specific) rules that transactions must abide by in order to b= e >>>>>>>> accepted >>>>>>>> into our mempool. Transaction "Standardness" rules and mempool >>>>>>>> restrictions such >>>>>>>> as "too-long-mempool-chain" are both examples of policy. >>>>>>>> >>>>>>>> ### Package Relay and Package Mempool Accept >>>>>>>> >>>>>>>> In transaction relay, we currently consider transactions one at a >>>>>>>> time for >>>>>>>> submission to the mempool. This creates a limitation in the node's >>>>>>>> ability to >>>>>>>> determine which transactions have the highest feerates, since we >>>>>>>> cannot take >>>>>>>> into account descendants (i.e. cannot use CPFP) until all the >>>>>>>> transactions are >>>>>>>> in the mempool. Similarly, we cannot use a transaction's >>>>>>>> descendants when >>>>>>>> considering it for RBF. When an individual transaction does not >>>>>>>> meet the mempool >>>>>>>> minimum feerate and the user isn't able to create a replacement >>>>>>>> transaction >>>>>>>> directly, it will not be accepted by mempools. >>>>>>>> >>>>>>>> This limitation presents a security issue for applications and >>>>>>>> users relying on >>>>>>>> time-sensitive transactions. For example, Lightning and other >>>>>>>> protocols create >>>>>>>> UTXOs with multiple spending paths, where one counterparty's >>>>>>>> spending path opens >>>>>>>> up after a timelock, and users are protected from cheating >>>>>>>> scenarios as long as >>>>>>>> they redeem on-chain in time. A key security assumption is that al= l >>>>>>>> parties' >>>>>>>> transactions will propagate and confirm in a timely manner. This >>>>>>>> assumption can >>>>>>>> be broken if fee-bumping does not work as intended. >>>>>>>> >>>>>>>> The end goal for Package Relay is to consider multiple transaction= s >>>>>>>> at the same >>>>>>>> time, e.g. a transaction with its high-fee child. This may help us >>>>>>>> better >>>>>>>> determine whether transactions should be accepted to our mempool, >>>>>>>> especially if >>>>>>>> they don't meet fee requirements individually or are better RBF >>>>>>>> candidates as a >>>>>>>> package. A combination of changes to mempool validation logic, >>>>>>>> policy, and >>>>>>>> transaction relay allows us to better propagate the transactions >>>>>>>> with the >>>>>>>> highest package feerates to miners, and makes fee-bumping tools >>>>>>>> more powerful >>>>>>>> for users. >>>>>>>> >>>>>>>> The "relay" part of Package Relay suggests P2P messaging changes, >>>>>>>> but a large >>>>>>>> part of the changes are in the mempool's package validation logic. >>>>>>>> We call this >>>>>>>> *Package Mempool Accept*. >>>>>>>> >>>>>>>> ### Previous Work >>>>>>>> >>>>>>>> * Given that mempool validation is DoS-sensitive and complex, it >>>>>>>> would be >>>>>>>> dangerous to haphazardly tack on package validation logic. Many >>>>>>>> efforts have >>>>>>>> been made to make mempool validation less opaque (see [#16400][4], >>>>>>>> [#21062][5], >>>>>>>> [#22675][6], [#22796][7]). >>>>>>>> * [#20833][8] Added basic capabilities for package validation, tes= t >>>>>>>> accepts only >>>>>>>> (no submission to mempool). >>>>>>>> * [#21800][9] Implemented package ancestor/descendant limit checks >>>>>>>> for arbitrary >>>>>>>> packages. Still test accepts only. >>>>>>>> * Previous package relay proposals (see [#16401][10], [#19621][11]= ). >>>>>>>> >>>>>>>> ### Existing Package Rules >>>>>>>> >>>>>>>> These are in master as introduced in [#20833][8] and [#21800][9]. >>>>>>>> I'll consider >>>>>>>> them as "given" in the rest of this document, though they can be >>>>>>>> changed, since >>>>>>>> package validation is test-accept only right now. >>>>>>>> >>>>>>>> 1. A package cannot exceed `MAX_PACKAGE_COUNT=3D25` count and >>>>>>>> `MAX_PACKAGE_SIZE=3D101KvB` total size [8] >>>>>>>> >>>>>>>> *Rationale*: This is already enforced as mempool >>>>>>>> ancestor/descendant limits. >>>>>>>> Presumably, transactions in a package are all related, so exceedin= g >>>>>>>> this limit >>>>>>>> would mean that the package can either be split up or it wouldn't >>>>>>>> pass this >>>>>>>> mempool policy. >>>>>>>> >>>>>>>> 2. Packages must be topologically sorted: if any dependencies exis= t >>>>>>>> between >>>>>>>> transactions, parents must appear somewhere before children. [8] >>>>>>>> >>>>>>>> 3. A package cannot have conflicting transactions, i.e. none of >>>>>>>> them can spend >>>>>>>> the same inputs. This also means there cannot be duplicate >>>>>>>> transactions. [8] >>>>>>>> >>>>>>>> 4. When packages are evaluated against ancestor/descendant limits >>>>>>>> in a test >>>>>>>> accept, the union of all of their descendants and ancestors is >>>>>>>> considered. This >>>>>>>> is essentially a "worst case" heuristic where every transaction in >>>>>>>> the package >>>>>>>> is treated as each other's ancestor and descendant. [8] >>>>>>>> Packages for which ancestor/descendant limits are accurately >>>>>>>> captured by this >>>>>>>> heuristic: [19] >>>>>>>> >>>>>>>> There are also limitations such as the fact that CPFP carve out is >>>>>>>> not applied >>>>>>>> to package transactions. #20833 also disables RBF in package >>>>>>>> validation; this >>>>>>>> proposal overrides that to allow packages to use RBF. >>>>>>>> >>>>>>>> ## Proposed Changes >>>>>>>> >>>>>>>> The next step in the Package Mempool Accept project is to implemen= t >>>>>>>> submission >>>>>>>> to mempool, initially through RPC only. This allows us to test the >>>>>>>> submission >>>>>>>> logic before exposing it on P2P. >>>>>>>> >>>>>>>> ### Summary >>>>>>>> >>>>>>>> - Packages may contain already-in-mempool transactions. >>>>>>>> - Packages are 2 generations, Multi-Parent-1-Child. >>>>>>>> - Fee-related checks use the package feerate. This means that >>>>>>>> wallets can >>>>>>>> create a package that utilizes CPFP. >>>>>>>> - Parents are allowed to RBF mempool transactions with a set of >>>>>>>> rules similar >>>>>>>> to BIP125. This enables a combination of CPFP and RBF, where a >>>>>>>> transaction's descendant fees pay for replacing mempool conflicts. >>>>>>>> >>>>>>>> There is a draft implementation in [#22290][1]. It is WIP, but >>>>>>>> feedback is >>>>>>>> always welcome. >>>>>>>> >>>>>>>> ### Details >>>>>>>> >>>>>>>> #### Packages May Contain Already-in-Mempool Transactions >>>>>>>> >>>>>>>> A package may contain transactions that are already in the mempool= . >>>>>>>> We remove >>>>>>>> ("deduplicate") those transactions from the package for the >>>>>>>> purposes of package >>>>>>>> mempool acceptance. If a package is empty after deduplication, we >>>>>>>> do nothing. >>>>>>>> >>>>>>>> *Rationale*: Mempools vary across the network. It's possible for a >>>>>>>> parent to be >>>>>>>> accepted to the mempool of a peer on its own due to differences in >>>>>>>> policy and >>>>>>>> fee market fluctuations. We should not reject or penalize the >>>>>>>> entire package for >>>>>>>> an individual transaction as that could be a censorship vector. >>>>>>>> >>>>>>>> #### Packages Are Multi-Parent-1-Child >>>>>>>> >>>>>>>> Only packages of a specific topology are permitted. Namely, a >>>>>>>> package is exactly >>>>>>>> 1 child with all of its unconfirmed parents. After deduplication, >>>>>>>> the package >>>>>>>> may be exactly the same, empty, 1 child, 1 child with just some of >>>>>>>> its >>>>>>>> unconfirmed parents, etc. Note that it's possible for the parents >>>>>>>> to be indirect >>>>>>>> descendants/ancestors of one another, or for parent and child to >>>>>>>> share a parent, >>>>>>>> so we cannot make any other topology assumptions. >>>>>>>> >>>>>>>> *Rationale*: This allows for fee-bumping by CPFP. Allowing multipl= e >>>>>>>> parents >>>>>>>> makes it possible to fee-bump a batch of transactions. Restricting >>>>>>>> packages to a >>>>>>>> defined topology is also easier to reason about and simplifies the >>>>>>>> validation >>>>>>>> logic greatly. Multi-parent-1-child allows us to think of the >>>>>>>> package as one big >>>>>>>> transaction, where: >>>>>>>> >>>>>>>> - Inputs =3D all the inputs of parents + inputs of the child that >>>>>>>> come from >>>>>>>> confirmed UTXOs >>>>>>>> - Outputs =3D all the outputs of the child + all outputs of the >>>>>>>> parents that >>>>>>>> aren't spent by other transactions in the package >>>>>>>> >>>>>>>> Examples of packages that follow this rule (variations of example = A >>>>>>>> show some >>>>>>>> possibilities after deduplication): ![image][15] >>>>>>>> >>>>>>>> #### Fee-Related Checks Use Package Feerate >>>>>>>> >>>>>>>> Package Feerate =3D the total modified fees divided by the total >>>>>>>> virtual size of >>>>>>>> all transactions in the package. >>>>>>>> >>>>>>>> To meet the two feerate requirements of a mempool, i.e., the >>>>>>>> pre-configured >>>>>>>> minimum relay feerate (`minRelayTxFee`) and dynamic mempool minimu= m >>>>>>>> feerate, the >>>>>>>> total package feerate is used instead of the individual feerate. >>>>>>>> The individual >>>>>>>> transactions are allowed to be below feerate requirements if the >>>>>>>> package meets >>>>>>>> the feerate requirements. For example, the parent(s) in the packag= e >>>>>>>> can have 0 >>>>>>>> fees but be paid for by the child. >>>>>>>> >>>>>>>> *Rationale*: This can be thought of as "CPFP within a package," >>>>>>>> solving the >>>>>>>> issue of a parent not meeting minimum fees on its own. This allows >>>>>>>> L2 >>>>>>>> applications to adjust their fees at broadcast time instead of >>>>>>>> overshooting or >>>>>>>> risking getting stuck/pinned. >>>>>>>> >>>>>>>> We use the package feerate of the package *after deduplication*. >>>>>>>> >>>>>>>> *Rationale*: It would be incorrect to use the fees of transaction= s >>>>>>>> that are >>>>>>>> already in the mempool, as we do not want a transaction's fees to = be >>>>>>>> double-counted for both its individual RBF and package RBF. >>>>>>>> >>>>>>>> Examples F and G [14] show the same package, but P1 is submitted >>>>>>>> individually before >>>>>>>> the package in example G. In example F, we can see that the 300vB >>>>>>>> package pays >>>>>>>> an additional 200sat in fees, which is not enough to pay for its >>>>>>>> own bandwidth >>>>>>>> (BIP125#4). In example G, we can see that P1 pays enough to replac= e >>>>>>>> M1, but >>>>>>>> using P1's fees again during package submission would make it look >>>>>>>> like a 300sat >>>>>>>> increase for a 200vB package. Even including its fees and size >>>>>>>> would not be >>>>>>>> sufficient in this example, since the 300sat looks like enough for >>>>>>>> the 300vB >>>>>>>> package. The calculcation after deduplication is 100sat increase >>>>>>>> for a package >>>>>>>> of size 200vB, which correctly fails BIP125#4. Assume all >>>>>>>> transactions have a >>>>>>>> size of 100vB. >>>>>>>> >>>>>>>> #### Package RBF >>>>>>>> >>>>>>>> If a package meets feerate requirements as a package, the parents >>>>>>>> in the >>>>>>>> transaction are allowed to replace-by-fee mempool transactions. Th= e >>>>>>>> child cannot >>>>>>>> replace mempool transactions. Multiple transactions can replace th= e >>>>>>>> same >>>>>>>> transaction, but in order to be valid, none of the transactions ca= n >>>>>>>> try to >>>>>>>> replace an ancestor of another transaction in the same package >>>>>>>> (which would thus >>>>>>>> make its inputs unavailable). >>>>>>>> >>>>>>>> *Rationale*: Even if we are using package feerate, a package will >>>>>>>> not propagate >>>>>>>> as intended if RBF still requires each individual transaction to >>>>>>>> meet the >>>>>>>> feerate requirements. >>>>>>>> >>>>>>>> We use a set of rules slightly modified from BIP125 as follows: >>>>>>>> >>>>>>>> ##### Signaling (Rule #1) >>>>>>>> >>>>>>>> All mempool transactions to be replaced must signal replaceability= . >>>>>>>> >>>>>>>> *Rationale*: Package RBF signaling logic should be the same for >>>>>>>> package RBF and >>>>>>>> single transaction acceptance. This would be updated if single >>>>>>>> transaction >>>>>>>> validation moves to full RBF. >>>>>>>> >>>>>>>> ##### New Unconfirmed Inputs (Rule #2) >>>>>>>> >>>>>>>> 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. This is contrary to BIP125#2, which states "The >>>>>>>> replacement >>>>>>>> transaction may only include an unconfirmed input if that input wa= s >>>>>>>> included in >>>>>>>> one of the original transactions. (An unconfirmed input spends an >>>>>>>> output from a >>>>>>>> currently-unconfirmed transaction.)" >>>>>>>> >>>>>>>> *Rationale*: The purpose of BIP125#2 is to ensure that the >>>>>>>> replacement >>>>>>>> transaction has a higher ancestor score than the original >>>>>>>> transaction(s) (see >>>>>>>> [comment][13]). Example H [16] shows how adding a new unconfirmed >>>>>>>> input can lower the >>>>>>>> ancestor score of the replacement transaction. P1 is trying to >>>>>>>> replace M1, and >>>>>>>> spends an unconfirmed output of M2. P1 pays 800sat, M1 pays 600sat= , >>>>>>>> and M2 pays >>>>>>>> 100sat. Assume all transactions have a size of 100vB. While, in >>>>>>>> isolation, P1 >>>>>>>> looks like a better mining candidate than M1, it must be mined wit= h >>>>>>>> M2, so its >>>>>>>> ancestor feerate is actually 4.5sat/vB. This is lower than M1's >>>>>>>> ancestor >>>>>>>> feerate, which is 6sat/vB. >>>>>>>> >>>>>>>> In package RBF, the rule analogous to BIP125#2 would be "none of t= he >>>>>>>> transactions in the package can spend new unconfirmed inputs." >>>>>>>> Example J [17] shows >>>>>>>> why, if any of the package transactions have ancestors, package >>>>>>>> feerate is no >>>>>>>> longer accurate. Even though M2 and M3 are not ancestors of P1 >>>>>>>> (which is the >>>>>>>> replacement transaction in an RBF), we're actually interested in >>>>>>>> the entire >>>>>>>> package. A miner should mine M1 which is 5sat/vB instead of M2, M3= , >>>>>>>> P1, P2, and >>>>>>>> P3, which is only 4sat/vB. The Package RBF rule cannot be loosened >>>>>>>> to only allow >>>>>>>> the child to have new unconfirmed inputs, either, because it can >>>>>>>> still cause us >>>>>>>> to overestimate the package's ancestor score. >>>>>>>> >>>>>>>> However, enforcing a rule analogous to BIP125#2 would not only mak= e >>>>>>>> Package RBF >>>>>>>> less useful, but would also break Package RBF for packages with >>>>>>>> parents already >>>>>>>> in the mempool: if a package parent has already been submitted, it >>>>>>>> would look >>>>>>>> like the child is spending a "new" unconfirmed input. In example K >>>>>>>> [18], we're >>>>>>>> looking to replace M1 with the entire package including P1, P2, an= d >>>>>>>> P3. We must >>>>>>>> consider the case where one of the parents is already in the >>>>>>>> mempool (in this >>>>>>>> case, P2), which means we must allow P3 to have new unconfirmed >>>>>>>> inputs. However, >>>>>>>> M2 lowers the ancestor score of P3 to 4.3sat/vB, so we should not >>>>>>>> replace M1 >>>>>>>> with this package. >>>>>>>> >>>>>>>> Thus, the package RBF rule regarding new unconfirmed inputs is les= s >>>>>>>> strict than >>>>>>>> BIP125#2. However, we still achieve the same goal of requiring the >>>>>>>> replacement >>>>>>>> transactions to have a ancestor score at least as high as the >>>>>>>> original ones. As >>>>>>>> a result, the entire package is required to be a higher feerate >>>>>>>> mining candidate >>>>>>>> than each of the replaced transactions. >>>>>>>> >>>>>>>> Another note: the [comment][13] above the BIP125#2 code in the >>>>>>>> original RBF >>>>>>>> implementation suggests that the rule was intended to be temporary= . >>>>>>>> >>>>>>>> ##### Absolute Fee (Rule #3) >>>>>>>> >>>>>>>> The package must increase the absolute fee of the mempool, i.e. th= e >>>>>>>> total fees >>>>>>>> of the package must be higher than the absolute fees of the mempoo= l >>>>>>>> transactions >>>>>>>> it replaces. Combined with the CPFP rule above, this differs from >>>>>>>> BIP125 Rule #3 >>>>>>>> - an individual transaction in the package may have lower fees tha= n >>>>>>>> the >>>>>>>> transaction(s) it is replacing. In fact, it may have 0 fees, and >>>>>>>> the child >>>>>>>> pays for RBF. >>>>>>>> >>>>>>>> ##### Feerate (Rule #4) >>>>>>>> >>>>>>>> The package must pay for its own bandwidth; the package feerate >>>>>>>> must be higher >>>>>>>> than the replaced transactions by at least minimum relay feerate >>>>>>>> (`incrementalRelayFee`). Combined with the CPFP rule above, this >>>>>>>> differs from >>>>>>>> BIP125 Rule #4 - an individual transaction in the package can have >>>>>>>> a lower >>>>>>>> feerate than the transaction(s) it is replacing. In fact, it may >>>>>>>> have 0 fees, >>>>>>>> and the child pays for RBF. >>>>>>>> >>>>>>>> ##### Total Number of Replaced Transactions (Rule #5) >>>>>>>> >>>>>>>> The package cannot replace more than 100 mempool transactions. Thi= s >>>>>>>> is identical >>>>>>>> to BIP125 Rule #5. >>>>>>>> >>>>>>>> ### Expected FAQs >>>>>>>> >>>>>>>> 1. Is it possible for only some of the package to make it into the >>>>>>>> mempool? >>>>>>>> >>>>>>>> Yes, it is. However, since we evict transactions from the >>>>>>>> mempool by >>>>>>>> descendant score and the package child is supposed to be sponsorin= g >>>>>>>> the fees of >>>>>>>> its parents, the most common scenario would be all-or-nothing. Thi= s >>>>>>>> is >>>>>>>> incentive-compatible. In fact, to be conservative, package >>>>>>>> validation should >>>>>>>> begin by trying to submit all of the transactions individually, an= d >>>>>>>> only use the >>>>>>>> package mempool acceptance logic if the parents fail due to low >>>>>>>> feerate. >>>>>>>> >>>>>>>> 2. Should we allow packages to contain already-confirmed >>>>>>>> transactions? >>>>>>>> >>>>>>>> No, for practical reasons. In mempool validation, we actually >>>>>>>> aren't able to >>>>>>>> tell with 100% confidence if we are looking at a transaction that >>>>>>>> has already >>>>>>>> confirmed, because we look up inputs using a UTXO set. If we have >>>>>>>> historical >>>>>>>> block data, it's possible to look for it, but this is inefficient, >>>>>>>> not always >>>>>>>> possible for pruning nodes, and unnecessary because we're not goin= g >>>>>>>> to do >>>>>>>> anything with the transaction anyway. As such, we already have the >>>>>>>> expectation >>>>>>>> that transaction relay is somewhat "stateful" i.e. nobody should b= e >>>>>>>> relaying >>>>>>>> transactions that have already been confirmed. Similarly, we >>>>>>>> shouldn't be >>>>>>>> relaying packages that contain already-confirmed transactions. >>>>>>>> >>>>>>>> [1]: https://github.com/bitcoin/bitcoin/pull/22290 >>>>>>>> [2]: >>>>>>>> https://github.com/bitcoin/bips/blob/1f0b563738199ca60d32b4ba77979= 7fc97d040fe/bip-0141.mediawiki#transaction-size-calculations >>>>>>>> [3]: >>>>>>>> https://github.com/bitcoin/bitcoin/blob/94f83534e4b771944af7d9ed0f= 40746f392eb75e/src/policy/policy.cpp#L282 >>>>>>>> [4]: https://github.com/bitcoin/bitcoin/pull/16400 >>>>>>>> [5]: https://github.com/bitcoin/bitcoin/pull/21062 >>>>>>>> [6]: https://github.com/bitcoin/bitcoin/pull/22675 >>>>>>>> [7]: https://github.com/bitcoin/bitcoin/pull/22796 >>>>>>>> [8]: https://github.com/bitcoin/bitcoin/pull/20833 >>>>>>>> [9]: https://github.com/bitcoin/bitcoin/pull/21800 >>>>>>>> [10]: https://github.com/bitcoin/bitcoin/pull/16401 >>>>>>>> [11]: https://github.com/bitcoin/bitcoin/pull/19621 >>>>>>>> [12]: >>>>>>>> https://github.com/bitcoin/bips/blob/master/bip-0125.mediawiki >>>>>>>> [13]: >>>>>>>> https://github.com/bitcoin/bitcoin/pull/6871/files#diff-34d21af3c6= 14ea3cee120df276c9c4ae95053830d7f1d3deaf009a4625409ad2R1101-R1104 >>>>>>>> [14]: >>>>>>>> https://user-images.githubusercontent.com/25183001/133567078-075a9= 71c-0619-4339-9168-b41fd2b90c28.png >>>>>>>> [15]: >>>>>>>> https://user-images.githubusercontent.com/25183001/132856734-fc17d= a75-f875-44bb-b954-cb7a1725cc0d.png >>>>>>>> [16]: >>>>>>>> https://user-images.githubusercontent.com/25183001/133567347-a3e2e= 4a8-ae9c-49f8-abb9-81e8e0aba224.png >>>>>>>> [17]: >>>>>>>> https://user-images.githubusercontent.com/25183001/133567370-21566= d0e-36c8-4831-b1a8-706634540af3.png >>>>>>>> [18]: >>>>>>>> https://user-images.githubusercontent.com/25183001/133567444-bfff1= 142-439f-4547-800a-2ba2b0242bcb.png >>>>>>>> [19]: >>>>>>>> https://user-images.githubusercontent.com/25183001/133456219-0bb44= 7cb-dcb4-4a31-b9c1-7d86205b68bc.png >>>>>>>> [20]: >>>>>>>> https://user-images.githubusercontent.com/25183001/132857787-7b7c6= f56-af96-44c8-8d78-983719888c19.png >>>>>>>> _______________________________________________ >>>>>>>> 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 >>> >> --000000000000a502f005cd2105e8 Content-Type: text/html; charset="UTF-8" Content-Transfer-Encoding: quoted-printable
Hi Antoine and Bastien,

> Yes 1) it would be goo= d to have inputs of more potential users of package acceptance . And 2) I t= hink it's more a matter of clearer wording of the proposal.

(1) = I'm leaning towards multi-parent-1-child and offering [#22674][0] up fo= r review. If somebody feels very strongly about 1-parent-1-child, please le= t me know.

(2) I'm glad this turned out to just be a wording pro= blem. I've updated the proposal to [say][1] "If it meets feerate r= equirements, the package can replace mempool transactions if any of the par= ents conflict with mempool transactions. The child cannot conflict with any= mempool transactions." Hopefully that is more *univoque*.

Side= note: I've also updated the proposal to contain a [section][2] on why = submitting transactions individually before package validation is incentive= -compatible. I think it's relevant to our conversation, but for those w= ho just want to _use_ packages, it's just an implementation detail.
=
On restricting packages to confirmed inputs only:

> I think w= e could restrain package acceptance to only confirmed inputs for now and re= visit later this point ? For LN-anchor, you can assume that the fee-bumping= UTXO feeding the CPFP is already
confirmed. Or are there currently-depl= oyed use-cases which would benefit from your proposed Rule #2 ?

I th= ought about this a lot this week, and wrote up a summary of why I don't= think BIP125#2 helps us at all [here][3] on #23121. I see that you've = already come across it :)

> IIRC, the carve-out t= olerance is only 2txn/10_000 vb. If one of your=20 counterparties attach a junk branch on her own anchor output, are you=20 allowed to chain your self-owned unconfirmed CPFP ?

Yes, if your counterparty attaches a bunch of descendants to their anchor= output to dominate the descendant limit of your shared commitment transact= ion, CPFP carve out allows you to add 1 extra transaction under 10KvB to yo= ur own anchor output. It's fine if it spends an unconfirmed input, as l= ong as you aren't exceeding the descendant limits of that transaction. = This shouldn't be the case; I think something is seriously wrong if all= of your UTXOs are tied up in mempool transactions with big ancestor/descen= dant trees.

I don't know much about L2 dev= elopment so I'm just going to quote this:

> I think constrain= ing package acceptance to only confirmed inputs is very limiting and quite = dangerous for L2 protocols.

Since the restriction isn&#= 39;t helpful in simplifying the mempool code, makes things more complicated= for application developers, and can be dangerous for L2, I'd prefer no= t to add this restriction for packages.

On Antoine's question ab= out our miner model:

> Can you describe what miner model we are u= sing ? Like the block construction strategy implemented by `addPackagesTxs`= or also encompassing our current mempool acceptance policy, which I think = rely on absolute fee over ancestor score in case of replacement ?

Ou= r current model for block construction is this: we sort our mempool by pack= age ancestor score (total modified fees of a tx and its unconfirmed ancesto= rs / total vsize as seen by our mempool) and add packages to a block until = it's full. That's not to say this is the perfect miner policy, but = mempool acceptance logic follows this model as closely as possible because = it is, fundamentally, a cache that aids in block assembly performance. As a= nother way of looking at this, imagine if our mempool was so small it could= only store ~1 block's worth of transactions. It should always try to k= eep the highest-fees-within-1-block transactions, and obviously wouldn'= t evict small-but-valuable transations in favor or giant ones paying medioc= re feerates. All fee-related mempool policies, including RBF, consider feer= ate. BIP125#3 is a rule on absolute fees, but it is always combined with BI= P125#4, a rule on feerates. AFAIK, the reason it doesn't use ancestor s= core is that information wasn't cached in mempool entries at the time, = and thus not readily available to use in mempool validation.

That= 9;s why I don't think this is relevant to package validation. Commentin= g on the model itself:

> Is this compatible with a model where a = miner prioritizes absolute fees over ancestor score, in the case that mempo= ols aren't full-enough to fulfill a block ?
>> Yes, A+C+D pays= 2500sat more in fees, but it is also 1000vB larger. A miner should prefer = to utilize their block space more effectively.
> If your mempool is e= mpty and only composed of A+C+D or A+B, I think taking A+C+D is the most ef= ficient block construction you can come up with as a miner ?
> I thin= k this point is worthy to discuss as otherwise we might downgrade the effic= iency of our current block construction strategy in periods of near-empty m= empools. A knowledge which could be discreetly leveraged by a miner to gain= an advantage on the rest of the mining ecosystem.

I believe this is= suggesting "if our mempool has so few transactions that it wouldn'= ;t reach block capacity, prioritize any increase in absolute fees, even if = the feerate is lower." I can see how this may result in a higher-fee b= lock in a specific scenario such as the one highlighted above, but I don= 9;t think it is a sound model in general. It would be impossible to tell wh= en we should use this model: we could simply be in IBD, restarted a node wi= th an old/empty mempool.dat, and even if it's a low-transaction-volume = time, we never know what transactions will trickle in between now and the n= ext block. Going back to the tiny 1-block mempool scenario, i.e., if you _n= ever_ wanted to keep transactions that you wouldn't put in the next blo= ck, would you ever switch strategies?

Thanks again to everyone who&#= 39;s given their attention to the package mempool accept proposal.

B= est,
Gloria

[0]: https://github.com/bitcoin/bitcoin/pull/22674
[1]: https://gist.github.com/glozow/dc4e9d5c5b14ade7cdfac40f43adb18a#packa= ge-rbf
[2]: https://gist.git= hub.com/glozow/dc4e9d5c5b14ade7cdfac40f43adb18a#always-try-individual-submi= ssion-first
[3]: https://github.com/bitcoin/bitcoin/pull/23121= #issuecomment-929475999
On Tue, Sep 28, 2021 at 11:59 PM Antoine Riard &= lt;antoine.riard@gmail.com&g= t; wrote:
Hi Bastien

> In the case of LN, an attacker can g= ame this and heavily restrict
your RBF attempts if you're only allo= wed to use confirmed inputs
and have many channels (and a limited= number of confirmed inputs).
Otherwise you'll need node oper= ators to pre-emptively split their
utxos into many small utxos ju= st for fee bumping, which is inefficient...

I share the c= oncern about splitting utxos into smaller ones.
IIRC, the carve-out tole= rance is only 2txn/10_000 vb. If one of your counterparties attach a junk b= ranch on her own anchor output, are you allowed to chain your self-owned un= confirmed CPFP ?
I'm thinking about the topology "Chained CPFPs= " exposed here : https://github.com/rust-bitcoin/rust-lig= htning/issues/989.
Or if you have another L2 broadcast to= pology which could be safe w.r.t our current mempool logic :) ?
<= div>

Le=C2=A0lun. 27 sept. 2021 =C3=A0=C2=A003:15, Bastien T= EINTURIER <bastien= @acinq.fr> a =C3=A9crit=C2=A0:
I think we could restrain package acceptance to only confirme= d inputs for now and revisit later this point ? For LN-anchor, you can assu= me that the fee-bumping UTXO feeding the CPFP is already
confirmed. Or a= re there currently-deployed use-cases which would benefit from your propose= d Rule #2 ?

I think constraining packag= e acceptance to only confirmed inputs
is very limiting and quite = dangerous for L2 protocols.

In the case of LN, an = attacker can game this and heavily restrict
your RBF attempts if = you're only allowed to use confirmed inputs
and have many cha= nnels (and a limited number of confirmed inputs).
Otherwise you&#= 39;ll need node operators to pre-emptively split their
utxos into= many small utxos just for fee bumping, which is inefficient...
<= br>
Bastien

Le=C2=A0lun. 27 sept. 2021 =C3=A0=C2=A000:27, An= toine Riard via bitcoin-dev <bitcoin-dev@lists.linuxfoundation.org&g= t; a =C3=A9crit=C2=A0:
Hi Gloria,

Thanks for your answers,

&= gt; In summary, it seems that the decisions that might still need
> a= ttention/input from devs on this mailing list are:
> 1. Whether we sh= ould start with multiple-parent-1-child or 1-parent-1-child.
> 2. Whe= ther it's ok to require that the child not have conflicts with
> = mempool transactions.

Yes 1) it would be good to have inputs of more= potential users of package acceptance . And 2) I think it's more a mat= ter of clearer wording of the proposal.

However, see my final point = on the relaxation around "unconfirmed inputs" which might in fact= alter our current block construction strategy.

> Right, the fact= that we essentially always choose the first-seen witness is
> an unf= ortunate limitation that exists already. Adding package mempool
> acc= ept doesn't worsen this, but the procedure in the future is to replace<= br>> the witness when it makes sense economically. We can also add logic= to
> allow package feerate to pay for witness replacements as well. = This is
> pretty far into the future, though.

Yes I agree pack= age mempool doesn't worsen this. And it's not an issue for current = LN as you can't significantly inflate a spending witness for the 2-of-2= funding output.
However, it might be an issue for multi-party protocol = where the spending script has alternative branches with asymmetric valid wi= tness weights. Taproot should ease that kind of script so hopefully we woul= d deploy wtxid-replacement not too far in the future.

> I could b= e misunderstanding, but an attacker wouldn't be able to
> batch-a= ttack like this. Alice's package only conflicts with A' + D', n= ot A'
> + B' + C' + D'. She only needs to pay for evi= cting 2 transactions.

Yeah I can be clearer, I think you have 2 pinn= ing attacks scenarios to consider.

In LN, if you're trying to co= nfirm a commitment transaction to time-out or claim on-chain a HTLC and the= timelock is near-expiration, you should be ready to pay in commitment+2nd-= stage HTLC transaction fees as much as the value offered by the HTLC.
Following this security assumption, an attacker can exploit it by targeti= ng together commitment transactions from different channels by blocking the= m under a high-fee child, of which the fee value
is equal to the top-val= ue HTLC + 1. Victims's fee-bumping logics won't overbid as it's= not worthy to offer fees beyond their competed HTLCs. Apart from observing= mempools state, victims can't learn they're targeted by the same a= ttacker.

To draw from the aforementioned topology, Mallory broadcast= s A' + B' + C' + D', where A' conflicts with Alice'= s P1, B' conflicts with Bob's P2, C' conflicts with Caroll'= s P3. Let's assume P1 is confirming the top-value HTLC of the set. If D= ' fees is higher than P1 + 1, it won't be rational for Alice or Bob= or Caroll to keep offering competing feerates. Mallory will be at loss on = stealing P1, as she has paid more in fees but will realize a gain on P2+P3.=

In this model, Alice is allowed to evict those 2 transactions (A= 9; + D') but as she is economically-bounded she won't succeed.
<= br>Mallory is maliciously exploiting RBF rule 3 on absolute fee. I think th= is 1st pinning scenario is correct and "lucractive" when you sum = the global gain/loss.

There is a 2nd attack scenario where A + B + C= + D, where D is the child of A,B,C. All those transactions are honestly is= sued by Alice. Once A + B + C + D are propagated in network mempools, Mallo= ry is able to replace A + D with =C2=A0A' + D' where D' is payi= ng a higher fee. This package A' + D' will confirm soon if D feerat= e was compelling but Mallory succeeds in delaying the confirmation
of B = + C for one or more blocks. As B + C are pre-signed commitments with a low-= fee rate they won't confirm without Alice issuing a new child E. Mallor= y can repeat the same trick by broadcasting
B' + E' and delay ag= ain the confirmation of C.

If the remaining package pending HTLC has= a higher-value than all the malicious fees over-bid, Mallory should realiz= e a gain. With this 2nd pinning attack, the malicious entity buys confirmat= ion delay of your packaged-together commitments.

Assuming those atta= cks are correct, I'm leaning towards being conservative with the LDK br= oadcast backend. Though once again, other L2 devs have likely other use-cas= es and opinions :)

> =C2=A0B' only needs to pay for itself in= this case.

Yes I think it's a nice discount when UTXO is single= -owned. In the context of shared-owned UTXO (e.g LN), you might not if ther= e is an in-mempool package already spending the UTXO and have to assume the= worst-case scenario. I.e have B' committing enough fee to pay for A= 9; replacement bandwidth. I think we can't do that much for this case..= .

> If a package meets feerate requirements as a
package, the = parents in the transaction are allowed to replace-by-fee
mempool transac= tions. The child cannot replace mempool transactions."

I agree = with the Mallory-vs-Alice case. Though if Alice broadcasts A+B' to repl= ace A+B because the first broadcast isn't satisfying anymore due to mem= pool spikes ? Assuming B' fees is enough, I think that case as child B&= #39; replacing in-mempool transaction B. Which I understand going against= =C2=A0 "The child cannot replace mempool transactions".

Ma= ybe wording could be a bit clearer ?

> While it would be nice to = have full RBF, malleability of the child won't
> block RBF here. = If we're trying to replace A', we only require that A'
> = signals replaceability, and don't mind if its child doesn't.
Yes, it sounds good.

> Yes, A+C+D pays 2500sat more in fees, but= it is also 1000vB larger. A miner
> should prefer to utilize their b= lock space more effectively.

If your mempool is empty and only compo= sed of A+C+D or A+B, I think taking A+C+D is the most efficient block const= ruction you can come up with as a miner ?

> No, because we don= 9;t use that model.

Can you describe what miner model we are using ?= Like the block construction strategy implemented by `addPackagesTxs` or al= so encompassing our current mempool acceptance policy, which I think rely o= n absolute fee over ancestor score in case of replacement ?

I think = this point is worthy to discuss as otherwise we might downgrade the efficie= ncy of our current block construction strategy in periods of near-empty mem= pools. A knowledge which could be discreetly leveraged by a miner to gain a= n advantage on the rest of the mining ecosystem.

Note, I think we *m= ight* have to go in this direction if we want to replace replace-by-fee by = replace-by-feerate or replace-by-ancestor and solve in-depth pinning attack= s. Though if we do so,
IMO we would need more thoughts.

I think = we could restrain package acceptance to only confirmed inputs for now and r= evisit later this point ? For LN-anchor, you can assume that the fee-bumpin= g UTXO feeding the CPFP is already
confirmed. Or are there currently-dep= loyed use-cases which would benefit from your proposed Rule #2 ?

Ant= oine

Le=C2=A0jeu. 23 sept. 2021 =C3=A0=C2=A011:36, Gloria Zhao <gloriajzhao@gmail.com= > a =C3=A9crit=C2=A0:
Hi Antoine,

Thanks as always for = your input. I'm glad we agree on so much!

In s= ummary, it seems that the decisions that might still need attention/input f= rom devs on this mailing list are:
1. Whether we should start with mul= tiple-parent-1-child or 1-parent-1-child.
2. Whether it's ok to requ= ire that the child not have conflicts with mempool transactions.

Responding to your comments...

> IIUC, you have pa= ckage A+B, during the dedup phase early in `AcceptMultipleTransactions` if = you observe same-txid-different-wtixd A' and A' is higher feerate t= han A, you trim A and replace by A' ?

> I think this approach= is safe, the one who appears unsafe to me is when A' has a _lower_ fee= rate, even if A' is already accepted by our mempool ? In that case iirc= that would be a pinning.

Right, the fact that we essentially always= choose the first-seen witness is an unfortunate limitation that exists alr= eady. Adding package mempool accept doesn't worsen this, but the proced= ure in the future is to replace the witness when it makes sense economicall= y. We can also add logic to allow package feerate to pay for witness replac= ements as well. This is pretty far into the future, though.

> It = sounds uneconomical for an attacker but I think it's not when you consi= der than you can "batch" attack against multiple honest counterpa= rties. E.g, Mallory broadcast A' + B' + C' + D' where A'= ; conflicts with Alice's honest package P1, B' conflicts with Bob&#= 39;s honest package P2, C' conflicts with Caroll's honest package P= 3. And D' is a high-fee child of A' + B' + C'.

> = If D' is higher-fee than P1 or P2 or P3 but inferior to the sum of HTLC= s confirmed by P1+P2+P3, I think it's lucrative for the attacker ?
<= br>I could be misunderstanding, but an attacker wouldn't be able to bat= ch-attack like this. Alice's package only conflicts with A' + D'= ;, not A' + B' + C' + D'. She only needs to pay for evictin= g 2 transactions.

> Do we assume that broadcasted packages are &q= uot;honest" by default and that the parent(s) always need the child to= pass the fee checks, that way saving the processing of individual transact= ions which are expected to fail in 99% of cases or more ad hoc composition = of packages at relay ?
> I think this point is quite dependent on the= p2p packages format/logic we'll end up on and that we should feel free= to revisit it later ?

I think it's the opposite; there's no= way for us to assume that p2p packages will be "honest." I'd= like to have two things before we expose on P2P: (1) ensure that the amoun= t of resources potentially allocated for package validation isn't dispr= oportionately higher than that of single transaction validation and (2) onl= y use package validation when we're unsatisifed with the single validat= ion result, e.g. we might get better fees.
Yes, let's revisit this l= ater :)
=C2=A0
=C2=A0> Yes, if you receive A+B, and A is already i= n-mempoo, I agree you can discard its feerate as B should pay for all fees = checked on its own. Where I'm unclear is when you have in-mempool A+B a= nd receive A+B'. Should B' have a fee high enough to cover the band= width penalty replacement (`PaysForRBF`, 2nd check) of both A+B' or onl= y B' ?
=C2=A0
=C2=A0B' only needs to pay for itself in this c= ase.
=C2=A0
> > Do we want the child to be able to replace memp= ool transactions as well?

> If we mean when you have replaceable = A+B then A'+B' try to replace with a higher-feerate ? I think that&= #39;s exactly the case we need for Lightning as A+B is coming from Alice an= d A'+B' is coming from Bob :/

Let me clarify this becau= se I can see that my wording was ambiguous, and then please let me know if = it fits Lightning's needs?

In my proposal, I wrote = "If a package meets feerate requirements as a package, the parents in = the transaction are allowed to replace-by-fee mempool transactions. The chi= ld cannot replace mempool transactions." What I meant was: the package= can replace mempool transactions if any of the parents conflict with mempo= ol transactions. The child cannot not conflict with any mempool transaction= s.
The Lightning use case this attempts to address is: Alice and Mallory= are LN counterparties, and have packages A+B and A'+B', respective= ly. A and A' are their commitment transactions and conflict with each o= ther; they have shared inputs and different txids.
B spends Alice's = anchor output from A. B' spends Mallory's anchor output from A'= . Thus, B and B' do not conflict with each other.
Alice can broadcas= t her package, A+B, to replace Mallory's package, A'+B', since = B doesn't conflict with the mempool.

Would this be ok?

&g= t; The second option, a child of A', In the LN case I think the CPFP is= attached on one's anchor output.

While it would be nice to have= full RBF, malleability of the child won't block RBF here. If we're= trying to replace A', we only require that A' signals replaceabili= ty, and don't mind if its child doesn't.

> > B has an = ancestor score of 10sat/vb and D has an
> > ancestor score of ~2.9= sat/vb. Since D's ancestor score is lower than B's,
> > it= fails the proposed package RBF Rule #2, so this package would be
> &= gt; rejected. Does this meet your expectations?

> Well what sound= s odd to me, in my example, we fail D even if it has a higher-fee than B. L= ike A+B absolute fees are 2000 sats and A+C+D absolute fees are 4500 sats ?=

Yes, A+C+D pays 2500sat more in fees, but it is also 1000vB larger.= A miner should prefer to utilize their block space more effectively.
> Is this compatible with a model where a miner prioritizes absolute f= ees over ancestor score, in the case that mempools aren't full-enough t= o fulfill a block ?

No, because we don't use that model.

= Thanks,
Gloria

On Thu, Sep 23, 2021 at 5:29 AM Antoine Riard <antoine.riard@gmail.c= om> wrote:
> Correct, if B+C is too low feerate to be accepted, = we will reject it. I
> prefer this because it is incentive compatible= : A can be mined by itself,
> so there's no reason to prefer A+B+= C instead of A.
> As another way of looking at this, consider the cas= e where we do accept
> A+B+C and it sits at the "bottom" of= our mempool. If our mempool reaches
> capacity, we evict the lowest = descendant feerate transactions, which are
> B+C in this case. This g= ives us the same resulting mempool, with A and not
> B+C.

I ag= ree here. Doing otherwise, we might evict other transactions mempool in `Me= mpoolAccept::Finalize` with a higher-feerate than B+C while those evicted t= ransactions are the most compelling for block construction.

I though= t at first missing this acceptance requirement would break a fee-bumping sc= heme like Parent-Pay-For-Child where a high-fee parent is attached to a chi= ld signed with SIGHASH_ANYONECANPAY but in this case the child fee is captu= ring the parent value. I can't think of other fee-bumping schemes poten= tially affected. If they do exist I would say they're wrong in their de= sign assumptions.

> If or when we have witness replacement, the l= ogic is: if the individual
> transaction is enough to replace the mem= pool one, the replacement will
> happen during the preceding individu= al transaction acceptance, and
> deduplication logic will work. Other= wise, we will try to deduplicate by
> wtxid, see that we need a packa= ge witness replacement, and use the package
> feerate to evaluate whe= ther this is economically rational.

IIUC, you have package A+B, duri= ng the dedup phase early in `AcceptMultipleTransactions` if you observe sam= e-txid-different-wtixd A' and A' is higher feerate than A, you trim= A and replace by A' ?

I think this approach is safe, the one wh= o appears unsafe to me is when A' has a _lower_ feerate, even if A'= is already accepted by our mempool ? In that case iirc that would be a pin= ning.

Good to see progress on witness replacement before we see usag= e of Taproot tree in the context of multi-party, where a malicious counterp= arty inflates its witness to jam a honest spending.

(Note, the commi= t linked currently points nowhere :))


> Please note that A ma= y replace A' even if A' has higher fees than A
> individually= , because the proposed package RBF utilizes the fees and size
> of th= e entire package. This just requires E to pay enough fees, although
>= this can be pretty high if there are also potential B' and C' comp= eting
> commitment transactions that we don't know about.

= Ah right, if the package acceptance waives `PaysMoreThanConflicts` for the = individual check on A, the honest package should replace the pinning attemp= t. I've not fully parsed the proposed implementation yet.

Though= note, I think it's still unsafe for a Lightning multi-commitment-broad= cast-as-one-package as a malicious A' might have an absolute fee higher= than E. It sounds uneconomical for
an attacker but I think it's not= when you consider than you can "batch" attack against multiple h= onest counterparties. E.g, Mallory broadcast A' + B' + C' + D&#= 39; where A' conflicts with Alice's honest package P1, B' confl= icts with Bob's honest package P2, C' conflicts with Caroll's h= onest package P3. And D' is a high-fee child of A' + B' + C'= ;.

If D' is higher-fee than P1 or P2 or P3 but inferior to the s= um of HTLCs confirmed by P1+P2+P3, I think it's lucrative for the attac= ker ?

> So far, my understanding is that multi-parent-1-child is = desired for
> batched fee-bumping (
> h= ttps://github.com/bitcoin/bitcoin/pull/22674#issuecomment-897951289) an= d
> I've also seen your response which I have less context on (> https://github.com/bitcoin/bitcoin/pull/2267= 4#issuecomment-900352202). That
> being said, I am happy to creat= e a new proposal for 1 parent + 1 child
> (which would be slightly si= mpler) and plan for moving to
> multi-parent-1-child later if that is= preferred. I am very interested in
> hearing feedback on that approa= ch.

I think batched fee-bumping is okay as long as you don't hav= e time-sensitive outputs encumbering your commitment transactions. For the = reasons mentioned above, I think that's unsafe.

What I'm wor= ried about is=C2=A0 L2 developers, potentially not aware about all the memp= ool subtleties blurring the difference and always batching their broadcast = by default.

IMO, a good thing by restraining to 1-parent + 1 child, = =C2=A0we artificially constraint L2 design space for now and minimize risks= of unsafe usage of the package API :)

I think that's a point wh= ere it would be relevant to have the opinion of more L2 devs.

> I= think there is a misunderstanding here - let me describe what I'm
&= gt; proposing we'd do in this situation: we'll try individual submi= ssion for A,
> see that it fails due to "insufficient fees."= ; Then, we'll try package
> validation for A+B and use package RB= F. If A+B pays enough, it can still
> replace A'. If A fails for = a bad signature, we won't look at B or A+B. Does
> this meet your= expectations?

Yes there was a misunderstanding, I think this approa= ch is correct, it's more a question of performance. Do we assume that b= roadcasted packages are "honest" by default and that the parent(s= ) always need the child to pass the fee checks, that way saving the process= ing of individual transactions which are expected to fail in 99% of cases o= r more ad hoc composition of packages at relay ?

I think this point = is quite dependent on the p2p packages format/logic we'll end up on and= that we should feel free to revisit it later ?


> What proble= m are you trying to solve by the package feerate *after* dedup
rule ?> My understanding is that an in-package transaction might be already i= n
the mempool. Therefore, to compute a correct RBF penalty replacement, = the
vsize of this transaction could be discarded lowering the cost of pa= ckage
RBF.

> I'm proposing that, when a transaction has al= ready been submitted to
> mempool, we would ignore both its fees and = vsize when calculating package
> feerate.

Yes, if you receive= A+B, and A is already in-mempoo, I agree you can discard its feerate as B = should pay for all fees checked on its own. Where I'm unclear is when y= ou have in-mempool A+B and receive A+B'. Should B' have a fee high = enough to cover the bandwidth penalty replacement (`PaysForRBF`, 2nd check)= of both A+B' or only B' ?

If you have a second-layer like c= urrent Lightning, you might have a counterparty commitment to replace and s= hould always expect to have to pay for parent replacement bandwidth.
Where a potential discount sounds interesting is when you have an univoque= state on the first-stage of transactions. E.g DLC's funding transactio= n which might be CPFP by any participant iirc.

> Note that, if C&= #39; conflicts with C, it also conflicts with D, since D is a
> desce= ndant of C and would thus need to be evicted along with it.

Ah once = again I think it's a misunderstanding without the code under my eyes! I= f we do C' `PreChecks`, solve the conflicts provoked by it, i.e mark fo= r potential eviction D and don't consider it for future conflicts in th= e rest of the package, I think D' `PreChecks` should be good ?

&= gt; More generally, this example is surprising to me because I didn't t= hink
> packages would be used to fee-bump replaceable transactions. D= o we want the
> child to be able to replace mempool transactions as w= ell?

If we mean when you have replaceable A+B then A'+B' try= to replace with a higher-feerate ? I think that's exactly the case we = need for Lightning as A+B is coming from Alice and A'+B' is coming = from Bob :/

> I'm not sure what you mean? Let's say we ha= ve a package of parent A + child
> B, where A is supposed to replace = a mempool transaction A'. Are you saying
> that counterparties ar= e able to malleate the package child B, or a child of
> A'?
<= br>The second option, a child of A', In the LN case I think the CPFP is= attached on one's anchor output.

I think it's good if we as= sume the solve-conflicts-after-parent's`'PreChecks` mentioned above= or fixing inherited signaling or full-rbf ?

> Sorry, I don't= understand what you mean by "preserve the package
> integrity?&= quot; Could you elaborate?

After thinking the relaxation about the &= quot;new" unconfirmed input is not linked to trimming but I would say = more to the multi-parent support.

Let's say you have A+B trying = to replace C+D where B is also spending already in-mempool E. To succeed, y= ou need to waive the no-new-unconfirmed input as D isn't spending E.
So good, I think we agree on the problem description here.

>= I am in agreement with your calculations but unsure if we disagree on the<= br>> expected outcome. Yes, B has an ancestor score of 10sat/vb and D ha= s an
> ancestor score of ~2.9sat/vb. Since D's ancestor score is = lower than B's,
> it fails the proposed package RBF Rule #2, so t= his package would be
> rejected. Does this meet your expectations?
Well what sounds odd to me, in my example, we fail D even if it has a = higher-fee than B. Like A+B absolute fees are 2000 sats and A+C+D absolute = fees are 4500 sats ?

Is this compatible with a model where a miner p= rioritizes absolute fees over ancestor score, in the case that mempools are= n't full-enough to fulfill a block ?

Let me know if I can clarif= y a point.

Antoine

Le=C2=A0lun. 20 sept. 2021 =C3=A0=C2=A011:10,= Gloria Zhao <gloriajzhao@gmail.com> a =C3=A9crit=C2=A0:

Hi Antoine,

= First of all, thank you for the thorough review. I appreciate your insight = on LN requirements.

> IIUC, you have a package A+B+C submitted fo= r acceptance and A is already in your mempool. You trim out A from the pack= age and then evaluate B+C.

> I think this might be an issue if A = is the higher-fee element of the ABC package. B+C package fees might be und= er the mempool min fee and will be rejected, potentially breaking the accep= tance expectations of the package issuer ?

Correct, if B+C is too lo= w feerate to be accepted, we will reject it. I prefer this because it is in= centive compatible: A can be mined by itself, so there's no reason to p= refer A+B+C instead of A.
As another way of looking at this, consider th= e case where we do accept A+B+C and it sits at the "bottom" of ou= r mempool. If our mempool reaches capacity, we evict the lowest descendant = feerate transactions, which are B+C in this case. This gives us the same re= sulting mempool, with A and not B+C.


> = Further, I think the dedup should be done on wtxid, as you might have multi= ple valid witnesses. Though with varying vsizes and as such offering differ= ent feerates.

I agree that variations of the same package with diffe= rent witnesses is a case that must be handled. I consider witness replaceme= nt to be a project that can be done in parallel to package mempool acceptan= ce because being able to accept packages does not worsen the problem of a s= ame-txid-different-witness "pinning" attack.

If or when we= have witness replacement, the logic is: if the individual transaction is e= nough to replace the mempool one, the replacement will happen during the pr= eceding individual transaction acceptance, and deduplication logic will wor= k. Otherwise, we will try to deduplicate by wtxid, see that we need a packa= ge witness replacement, and use the package feerate to evaluate whether thi= s is economically rational.

See the #22290 "handle package tran= sactions already in mempool" commit (https://github.com/bitcoin/bitcoin/pull/22290/commits/fea7= 5a2237b46cf76145242fecad7e274bfcb5ff), which handles the case of same-t= xid-different-witness by simply using the transaction in the mempool for no= w, with TODOs for what I just described.


> I'm not clearl= y understanding the accepted topologies. By "parent and child to share= a parent", do you mean the set of transactions A, B, C, where B is sp= ending A and C is spending A and B would be correct ?

Yes, that is w= hat I meant. Yes, that would a valid package under these rules.

>= If yes, is there a width-limit introduced or we fallback on MAX_PACKAGE_CO= UNT=3D25 ?

No, there is no limit on connectivity other than "ch= ild with all unconfirmed parents." We will enforce MAX_PACKAGE_COUNT= =3D25 and child's in-mempool + in-package ancestor limits.


&= gt; Considering the current Core's mempool acceptance rules, I think CP= FP batching is unsafe for LN time-sensitive closure. A malicious tx-relay j= amming successful on one channel commitment transaction would contamine the= remaining commitments sharing the same package.

> E.g, you broad= cast the package A+B+C+D+E where A,B,C,D are commitment transactions and E = a shared CPFP. If a malicious A' transaction has a better feerate than = A, the whole package acceptance will fail. Even if A' confirms in the f= ollowing block,
the propagation and confirmation of B+C+D have been dela= yed. This could carry on a loss of funds.

Please note that A may rep= lace A' even if A' has higher fees than A individually, because the= proposed package RBF utilizes the fees and size of the entire package. Thi= s just requires E to pay enough fees, although this can be pretty high if t= here are also potential B' and C' competing commitment transactions= that we don't know about.


> IMHO, I'm leaning toward= s deploying during a first phase 1-parent/1-child. I think it's the mos= t conservative step still improving second-layer safety.

So far, my = understanding is that multi-parent-1-child is desired for batched fee-bumpi= ng (https://github.com/bitcoin/bitcoin/pull/22674#i= ssuecomment-897951289) and I've also seen your response which I hav= e less context on (https://github.com/bitcoin/bitco= in/pull/22674#issuecomment-900352202). That being said, I am happy to c= reate a new proposal for 1 parent + 1 child (which would be slightly simple= r) and plan for moving to multi-parent-1-child later if that is preferred. = I am very interested in hearing feedback on that approach.


> = If A+B is submitted to replace A', where A pays 0 sats, B pays 200 sats= and A' pays 100 sats. If we apply the individual RBF on A, A+B accepta= nce fails. For this reason I think the individual RBF should be bypassed an= d only the package RBF apply ?

I think there is a misunderstanding h= ere - let me describe what I'm proposing we'd do in this situation:= we'll try individual submission for A, see that it fails due to "= insufficient fees." Then, we'll try package validation for A+B and= use package RBF. If A+B pays enough, it can still replace A'. If A fai= ls for a bad signature, we won't look at B or A+B. Does this meet your = expectations?


> What problem are you trying to solve by the p= ackage feerate *after* dedup rule ?
> My understanding is that an in-= package transaction might be already in the mempool. Therefore, to compute = a correct RBF penalty replacement, the vsize of this transaction could be d= iscarded lowering the cost of package RBF.

I'm proposing that, w= hen a transaction has already been submitted to mempool, we would ignore bo= th its fees and vsize when calculating package feerate. In example G2, we s= houldn't count M1 fees after its submission to mempool, since M1's = fees have already been used to pay for its individual bandwidth, and it sho= uldn't be used again to pay for P2 and P3's bandwidth. We also shou= ldn't count its vsize, since it has already been paid for.


&= gt; I think this is a footgunish API, as if a package issuer send the multi= ple-parent-one-child package A,B,C,D where D is the child of A,B,C. Then tr= y to broadcast the higher-feerate C'+D' package, it should be rejec= ted. So it's breaking the naive broadcaster assumption that a higher-fe= erate/higher-fee package always replaces ?

Note that, if C'= conflicts with C, it also conflicts with D, since D is a descendant of C a= nd would thus need to be evicted along with it. Implicitly, D' would no= t be in conflict with D.
More generally, this example is surp= rising to me because I didn't think packages would be used to fee-bump = replaceable transactions. Do we want the child to be able to replace mempoo= l transactions as well? This can be implemented with a bit of additional lo= gic.

> I think this is unsafe for L2s if counterparties have ma= lleability of the child transaction. They can block your package replacemen= t by opting-out from RBF signaling. IIRC, LN's "anchor output"= ; presents such an ability.

I'm not sure what you mean? Let'= s say we have a package of parent A + child B, where A is supposed to repla= ce a mempool transaction A'. Are you saying that counterparties are abl= e to malleate the package child B, or a child of A'? If they can mallea= te a child of A', that shouldn't matter as long as A' is signal= ing replacement. This would be handled identically with full RBF and what C= ore currently implements.

> I think this is an issue brought by t= he trimming during the dedup phase. If we preserve the package integrity, o= nly re-using the tx-level checks results of already in-mempool transactions= to gain in CPU time we won't have this issue. Package childs can add u= nconfirmed inputs as long as they're in-package, the bip125 rule2 is on= ly evaluated against parents ?

Sorry, I don't understand what yo= u mean by "preserve the package integrity?" Could you elaborate?<= br>
> Let's say you have in-mempool A, B where A pays 10 sat/vb f= or 100 vbytes and B pays 10 sat/vb for 100 vbytes. You have the candidate r= eplacement D spending both A and C where D pays 15sat/vb for 100 vbytes and= C pays 1 sat/vb for 1000 vbytes.

> Package A + B ancestor score = is 10 sat/vb.

> D has a higher feerate/absolute fee than B.
> Package A + C + D ancestor score is ~ 3 sat/vb ((A's 1000 sats += C's 1000 sats + D's 1500 sats) / A's 100 vb + C's 1000 vb = + D's 100 vb)

I am in agreement with your calculations but = unsure if we disagree on the expected outcome. Yes, B has an ancestor score= of 10sat/vb and D has an ancestor score of ~2.9sat/vb. Since D's ances= tor score is lower than B's, it fails the proposed package RBF Rule #2,= so this package would be rejected. Does this meet your expectations?
=

Thank you for linking to projects that might be interes= ted in package relay :)

Thanks,
Gloria

On Mon, Sep 20, 2= 021 at 12:16 AM Antoine Riard <antoine.riard@gmail.com> wrote:
Hi Gloria,
<= br>> A package may contain transactions that are already in the mempool.= We
> remove
> ("deduplicate") those transactions fro= m the package for the purposes of
> package
> mempool acceptanc= e. If a package is empty after deduplication, we do
> nothing.
IIUC, you have a package A+B+C submitted for acceptance and A is already i= n your mempool. You trim out A from the package and then evaluate B+C.
<= br>I think this might be an issue if A is the higher-fee element of the ABC= package. B+C package fees might be under the mempool min fee and will be r= ejected, potentially breaking the acceptance expectations of the package is= suer ?

Further, I think the dedup should be done on wtxid, as you mi= ght have multiple valid witnesses. Though with varying vsizes and as such o= ffering different feerates.

E.g you're going to evaluate the pac= kage A+B and A' is already in your mempool with a bigger valid witness.= You trim A based on txid, then you evaluate A'+B, which fails the fee = checks. However, evaluating A+B would have been a success.

AFAICT, t= he dedup rationale would be to save on CPU time/IO disk, to avoid repeated = signatures verification and parent UTXOs fetches ? Can we achieve the same = goal by bypassing tx-level checks for already-in txn while conserving the p= ackage integrity for package-level checks ?

> Note that it's = possible for the parents to be
> indirect
> descendants/ancesto= rs of one another, or for parent and child to share a
> parent,
&g= t; so we cannot make any other topology assumptions.

I'm not cle= arly understanding the accepted topologies. By "parent and child to sh= are a parent", do you mean the set of transactions A, B, C, where B is= spending A and C is spending A and B would be correct ?

If yes, is = there a width-limit introduced or we fallback on MAX_PACKAGE_COUNT=3D25 ?
IIRC, one rationale to come with this topology limitation was to lowe= r the DoS risks when potentially deploying p2p packages.

Considering= the current Core's mempool acceptance rules, I think CPFP batching is = unsafe for LN time-sensitive closure. A malicious tx-relay jamming successf= ul on one channel commitment transaction would contamine the remaining comm= itments sharing the same package.

E.g, you broadcast the package A+B= +C+D+E where A,B,C,D are commitment transactions and E a shared CPFP. If a = malicious A' transaction has a better feerate than A, the whole package= acceptance will fail. Even if A' confirms in the following block,
= the propagation and confirmation of B+C+D have been delayed. This could car= ry on a loss of funds.

That said, if you're broadcasting commitm= ent transactions without time-sensitive HTLC outputs, I think the batching = is effectively a fee saving as you don't have to duplicate the CPFP.
IMHO, I'm leaning towards deploying during a first phase 1-parent/= 1-child. I think it's the most conservative step still improving second= -layer safety.

> *Rationale*: =C2=A0It would be incorrect to use = the fees of transactions that are
> already in the mempool, as we do = not want a transaction's fees to be
> double-counted for both its= individual RBF and package RBF.

I'm unsure about the logical or= der of the checks proposed.

If A+B is submitted to replace A', w= here A pays 0 sats, B pays 200 sats and A' pays 100 sats. If we apply t= he individual RBF on A, A+B acceptance fails. For this reason I think the i= ndividual RBF should be bypassed and only the package RBF apply ?

N= ote this situation is plausible, with current LN design, your counterparty = can have a commitment transaction with a better fee just by selecting a hig= her `dust_limit_satoshis` than yours.

> Examples F and G [14] sho= w the same package, but P1 is submitted
> individually before
>= the package in example G. In example F, we can see that the 300vB package<= br>> pays
> an additional 200sat in fees, which is not enough to p= ay for its own
> bandwidth
> (BIP125#4). In example G, we can s= ee that P1 pays enough to replace M1, but
> using P1's fees again= during package submission would make it look like a
> 300sat
>= increase for a 200vB package. Even including its fees and size would not b= e
> sufficient in this example, since the 300sat looks like enough fo= r the 300vB
> package. The calculcation after deduplication is 100sat= increase for a
> package
> of size 200vB, which correctly fail= s BIP125#4. Assume all transactions have
> a
> size of 100vB.
What problem are you trying to solve by the package feerate *after* d= edup rule ?

My understanding is that an in-package transaction might= be already in the mempool. Therefore, to compute a correct RBF penalty rep= lacement, the vsize of this transaction could be discarded lowering the cos= t of package RBF.

If we keep a "safe" dedup mechanism (see= my point above), I think this discount is justified, as the validation cos= t of node operators is paid for ?

> The child cannot replace memp= ool transactions.

Let's say you issue package A+B, then package = C+B', where B' is a child of both A and C. This rule fails the acce= ptance of C+B' ?

I think this is a footgunish API, as if a packa= ge issuer send the multiple-parent-one-child package A,B,C,D where D is the= child of A,B,C. Then try to broadcast the higher-feerate C'+D' pac= kage, it should be rejected. So it's breaking the naive broadcaster ass= umption that a higher-feerate/higher-fee package always replaces ? And it m= ight be unsafe in protocols where states are symmetric. E.g a malicious cou= nterparty broadcasts first S+A, then you honestly broadcast S+B, where B pa= ys better fees.

> All mempool transactions to be replaced must si= gnal replaceability.

I think this is unsafe for L2s if counterpartie= s have malleability of the child transaction. They can block your package r= eplacement by opting-out from RBF signaling. IIRC, LN's "anchor ou= tput" presents such an ability.

I think it's better to eith= er fix inherited signaling or move towards full-rbf.

> if a packa= ge parent has already been submitted, it would
> look
>like the= child is spending a "new" unconfirmed input.

I think this= is an issue brought by the trimming during the dedup phase. If we preserve= the package integrity, only re-using the tx-level checks results of alread= y in-mempool transactions to gain in CPU time we won't have this issue.= Package childs can add unconfirmed inputs as long as they're in-packag= e, the bip125 rule2 is only evaluated against parents ?

> However= , we still achieve the same goal of requiring the
> replacement
&g= t; transactions to have a ancestor score at least as high as the original> ones.

I'm not sure if this holds...

Let's say = you have in-mempool A, B where A pays 10 sat/vb for 100 vbytes and B pays 1= 0 sat/vb for 100 vbytes. You have the candidate replacement D spending both= A and C where D pays 15sat/vb for 100 vbytes and C pays 1 sat/vb for 1000 = vbytes.

Package A + B ancestor score is 10 sat/vb.

D has a hi= gher feerate/absolute fee than B.

Package A + C + D ancestor score i= s ~ 3 sat/vb ((A's 1000 sats + C's 1000 sats + D's 1500 sats) /=
A's 100 vb + C's 1000 vb + D's 100 vb)

Overall, thi= s is a review through the lenses of LN requirements. I think other L2 proto= cols/applications
could be candidates to using package accept/relay such= as:
* https://github.com/lightninglabs/pool
* https://github.com/di= screetlogcontracts/dlcspecs
* https://github.com/bitco= in-teleport/teleport-transactions/
* https://github.com/sapio-lang/sapio<= br>* https://github.com/commerceblock/mercury/bl= ob/master/doc/statechains.md
* https://github.com/revault/practical-= revault

Thanks for rolling forward the ball on this subject.
=
Antoine

Le=C2=A0jeu. 16 sept. 2021 =C3=A0=C2=A003:55, Gloria Zhao v= ia bitcoin-dev <bitcoin-dev@lists.linuxfoundation.org> a =C3=A9cr= it=C2=A0:
Hi there,

I'm writing to propose a set of mempool pol= icy changes to enable package
validation (in preparation for package rel= ay) in Bitcoin Core. These would not
be consensus or P2P protocol change= s. However, since mempool policy
significantly affects transaction propa= gation, I believe this is relevant for
the mailing list.

My propo= sal enables packages consisting of multiple parents and 1 child. If you
= develop software that relies on specific transaction relay assumptions and/= or
are interested in using package relay in the future, I'm very int= erested to hear
your feedback on the utility or restrictiveness of these= package policies for
your use cases.

A draft implementation of t= his proposal can be found in [Bitcoin Core
PR#22290][1].

An illus= trated version of this post can be found at
= I have also linked the images below.

## Background

Feel fre= e to skip this section if you are already familiar with mempool policy
a= nd package relay terminology.

### Terminology Clarifications

= * Package =3D an ordered list of related transactions, representable by a D= irected
=C2=A0 Acyclic Graph.
* Package Feerate =3D the total modifie= d fees divided by the total virtual size of
=C2=A0 all transactions in t= he package.
=C2=A0 =C2=A0 - Modified fees =3D a transaction's base f= ees + fee delta applied by the user
=C2=A0 =C2=A0 =C2=A0 with `prioritis= etransaction`. As such, we expect this to vary across
mempools.
=C2= =A0 =C2=A0 - Virtual Size =3D the maximum of virtual sizes calculated using= [BIP141
=C2=A0 =C2=A0 =C2=A0 virtual size][2] and sigop weight. [Implem= ented here in Bitcoin Core][3].
=C2=A0 =C2=A0 - Note that feerate is not= necessarily based on the base fees and serialized
=C2=A0 =C2=A0 =C2=A0 = size.

* Fee-Bumping =3D user/wallet actions that take advantage of m= iner incentives to
=C2=A0 boost a transaction's candidacy for inclus= ion in a block, including Child Pays
for Parent (CPFP) and [BIP125][12] = Replace-by-Fee (RBF). Our intention in
mempool policy is to recognize wh= en the new transaction is more economical to
mine than the original one(= s) but not open DoS vectors, so there are some
limitations.

### P= olicy

The purpose of the mempool is to store the best (to be most in= centive-compatible
with miners, highest feerate) candidates for inclusio= n in a block. Miners use
the mempool to build block templates. The mempo= ol is also useful as a cache for
boosting block relay and validation per= formance, aiding transaction relay, and
generating feerate estimations.<= br>
Ideally, all consensus-valid transactions paying reasonable fees sho= uld make it
to miners through normal transaction relay, without any spec= ial connectivity or
relationships with miners. On the other hand, nodes = do not have unlimited
resources, and a P2P network designed to let any h= onest node broadcast their
transactions also exposes the transaction val= idation engine to DoS attacks from
malicious peers.

As such, for = unconfirmed transactions we are considering for our mempool, we
apply a = set of validation rules in addition to consensus, primarily to protect
u= s from resource exhaustion and aid our efforts to keep the highest fee
t= ransactions. We call this mempool _policy_: a set of (configurable,
node= -specific) rules that transactions must abide by in order to be acceptedinto our mempool. Transaction "Standardness" rules and mempool r= estrictions such
as "too-long-mempool-chain" are both examples= of policy.

### Package Relay and Package Mempool Accept

In t= ransaction relay, we currently consider transactions one at a time for
s= ubmission to the mempool. This creates a limitation in the node's abili= ty to
determine which transactions have the highest feerates, since we c= annot take
into account descendants (i.e. cannot use CPFP) until all the= transactions are
in the mempool. Similarly, we cannot use a transaction= 's descendants when
considering it for RBF. When an individual trans= action does not meet the mempool
minimum feerate and the user isn't = able to create a replacement transaction
directly, it will not be accept= ed by mempools.

This limitation presents a security issue for applic= ations and users relying on
time-sensitive transactions. For example, Li= ghtning and other protocols create
UTXOs with multiple spending paths, w= here one counterparty's spending path opens
up after a timelock, and= users are protected from cheating scenarios as long as
they redeem on-c= hain in time. A key security assumption is that all parties'
transac= tions will propagate and confirm in a timely manner. This assumption canbe broken if fee-bumping does not work as intended.

The end goal fo= r Package Relay is to consider multiple transactions at the same
time, e= .g. a transaction with its high-fee child. This may help us better
deter= mine whether transactions should be accepted to our mempool, especially if<= br>they don't meet fee requirements individually or are better RBF cand= idates as a
package. A combination of changes to mempool validation logi= c, policy, and
transaction relay allows us to better propagate the trans= actions with the
highest package feerates to miners, and makes fee-bumpi= ng tools more powerful
for users.

The "relay" part of P= ackage Relay suggests P2P messaging changes, but a large
part of the cha= nges are in the mempool's package validation logic. We call this
*Pa= ckage Mempool Accept*.

### Previous Work

* Given that mempool= validation is DoS-sensitive and complex, it would be
=C2=A0 dangerous t= o haphazardly tack on package validation logic. Many efforts have
been m= ade to make mempool validation less opaque (see [#16400][4], [#21062][5],[#22675][6], [#22796][7]).
* [#20833][8] Added basic capabilities for = package validation, test accepts only
=C2=A0 (no submission to mempool).=
* [#21800][9] Implemented package ancestor/descendant limit checks for = arbitrary
=C2=A0 packages. Still test accepts only.
* Previous packag= e relay proposals (see [#16401][10], [#19621][11]).

### Existing Pac= kage Rules

These are in master as introduced in [#20833][8] and [#21= 800][9]. I'll consider
them as "given" in the rest of this= document, though they can be changed, since
package validation is test-= accept only right now.

1. A package cannot exceed `MAX_PACKAGE_COUNT= =3D25` count and
`MAX_PACKAGE_SIZE=3D101KvB` total size [8]

=C2= =A0 =C2=A0*Rationale*: This is already enforced as mempool ancestor/descend= ant limits.
Presumably, transactions in a package are all related, so ex= ceeding this limit
would mean that the package can either be split up or= it wouldn't pass this
mempool policy.

2. Packages must be to= pologically sorted: if any dependencies exist between
transactions, pare= nts must appear somewhere before children. [8]

3. A package cannot h= ave conflicting transactions, i.e. none of them can spend
the same = inputs. This also means there cannot be duplicate transactions. [8]

4. When packages are evaluated against ancestor/descendant lim= its in a test
accept, the union of all of their descendants and ancestor= s is considered. This
is essentially a "worst case" heuristic = where every transaction in the package
is treated as each other's an= cestor and descendant. [8]
Packages for which ancestor/descendant limit= s are accurately captured by this
heuristic: [19]

There ar= e also limitations such as the fact that CPFP carve out is not applied
t= o package transactions. #20833 also disables RBF in package validation; thi= s
proposal overrides that to allow packages to use RBF.

## Propos= ed Changes

The next step in the Package Mempool Accept project is to= implement submission
to mempool, initially through RPC only. This allow= s us to test the submission
logic before exposing it on P2P.

### = Summary

- Packages may contain already-in-mempool transactions.
-= Packages are 2 generations, Multi-Parent-1-Child.
- Fee-related checks = use the package feerate. This means that wallets can
create a package th= at utilizes CPFP.
- Parents are allowed to RBF mempool transactions with= a set of rules similar
=C2=A0 to BIP125. This enables a combination of = CPFP and RBF, where a
transaction's descendant fees pay for replacin= g mempool conflicts.

There is a draft implementation in [#22290][1].= It is WIP, but feedback is
always welcome.

### Details

##= ## Packages May Contain Already-in-Mempool Transactions

A package ma= y contain transactions that are already in the mempool. We remove
("= ;deduplicate") those transactions from the package for the purposes of= package
mempool acceptance. If a package is empty after deduplication, = we do nothing.

*Rationale*: Mempools vary across the network. It'= ;s possible for a parent to be
accepted to the mempool of a peer on its = own due to differences in policy and
fee market fluctuations. We should = not reject or penalize the entire package for
an individual transaction = as that could be a censorship vector.

#### Packages Are Multi-Parent= -1-Child

Only packages of a specific topology are permitted. Namely,= a package is exactly
1 child with all of its unconfirmed parents. After= deduplication, the package
may be exactly the same, empty, 1 child, 1 c= hild with just some of its
unconfirmed parents, etc. Note that it's = possible for the parents to be indirect
descendants/ancestors of one ano= ther, or for parent and child to share a parent,
so we cannot make any o= ther topology assumptions.

*Rationale*: This allows for fee-bumping = by CPFP. Allowing multiple parents
makes it possible to fee-bump a batch= of transactions. Restricting packages to a
defined topology is also eas= ier to reason about and simplifies the validation
logic greatly. Multi-p= arent-1-child allows us to think of the package as one big
transaction, = where:

- Inputs =3D all the inputs of parents + inputs of the child = that come from
=C2=A0 confirmed UTXOs
- Outputs =3D all the outputs o= f the child + all outputs of the parents that
=C2=A0 aren't spent by= other transactions in the package

Examples of packages that follow = this rule (variations of example A show some
possibilities after dedupli= cation): ![image][15]

#### Fee-Related Checks Use Package Feerate
Package Feerate =3D the total modified fees divided by the total virtu= al size of
all transactions in the package.

To meet the two feera= te requirements of a mempool, i.e., the pre-configured
minimum relay fee= rate (`minRelayTxFee`) and dynamic mempool minimum feerate, the
total pa= ckage feerate is used instead of the individual feerate. The individual
= transactions are allowed to be below feerate requirements if the package me= ets
the feerate requirements. For example, the parent(s) in the package = can have 0
fees but be paid for by the child.

*Rationale*: This c= an be thought of as "CPFP within a package," solving the
issue= of a parent not meeting minimum fees on its own. This allows L2
applica= tions to adjust their fees at broadcast time instead of overshooting or
= risking getting stuck/pinned.

We use the package feerate of the pack= age *after deduplication*.

*Rationale*: =C2=A0It would be incorrect = to use the fees of transactions that are
already in the mempool, as we d= o not want a transaction's fees to be
double-counted for both its in= dividual RBF and package RBF.

Examples F and G [14] show the same pa= ckage, but P1 is submitted individually before
the package in example G.= In example F, we can see that the 300vB package pays
an additional 200s= at in fees, which is not enough to pay for its own bandwidth
(BIP125#4).= In example G, we can see that P1 pays enough to replace M1, but
using P= 1's fees again during package submission would make it look like a 300s= at
increase for a 200vB package. Even including its fees and size would = not be
sufficient in this example, since the 300sat looks like enough fo= r the 300vB
package. The calculcation after deduplication is 100sat incr= ease for a package
of size 200vB, which correctly fails BIP125#4. Assume= all transactions have a
size of 100vB.

#### Package RBF

I= f a package meets feerate requirements as a package, the parents in the
= transaction are allowed to replace-by-fee mempool transactions. The child c= annot
replace mempool transactions. Multiple transactions can replace th= e same
transaction, but in order to be valid, none of the transactions c= an try to
replace an ancestor of another transaction in the same package= (which would thus
make its inputs unavailable).

*Rationale*: Eve= n if we are using package feerate, a package will not propagate
as inten= ded if RBF still requires each individual transaction to meet the
feerat= e requirements.

We use a set of rules slightly modified from BIP125 = as follows:

##### Signaling (Rule #1)

All mempool transaction= s to be replaced must signal replaceability.

*Rationale*: Package RB= F signaling logic should be the same for package RBF and
single transact= ion acceptance. This would be updated if single transaction
validation m= oves to full RBF.

##### New Unconfirmed Inputs (Rule #2)

A pa= ckage may include new unconfirmed inputs, but the ancestor feerate of thechild must be at least as high as the ancestor feerates of every transact= ion
being replaced. This is contrary to BIP125#2, which states "The= replacement
transaction may only include an unconfirmed input if that i= nput was included in
one of the original transactions. (An unconfirmed i= nput spends an output from a
currently-unconfirmed transaction.)"
*Rationale*: The purpose of BIP125#2 is to ensure that the replacemen= t
transaction has a higher ancestor score than the original transaction(= s) (see
[comment][13]). Example H [16] shows how adding a new unconfirme= d input can lower the
ancestor score of the replacement transaction. P1 = is trying to replace M1, and
spends an unconfirmed output of M2. P1 pays= 800sat, M1 pays 600sat, and M2 pays
100sat. Assume all transactions hav= e a size of 100vB. While, in isolation, P1
looks like a better mining ca= ndidate than M1, it must be mined with M2, so its
ancestor feerate is ac= tually 4.5sat/vB.=C2=A0 This is lower than M1's ancestor
feerate, wh= ich is 6sat/vB.

In package RBF, the rule analogous to BIP125#2 would= be "none of the
transactions in the package can spend new unconfir= med inputs." Example J [17] shows
why, if any of the package transa= ctions have ancestors, package feerate is no
longer accurate. Even thoug= h M2 and M3 are not ancestors of P1 (which is the
replacement transactio= n in an RBF), we're actually interested in the entire
package. A min= er should mine M1 which is 5sat/vB instead of M2, M3, P1, P2, and
P3, wh= ich is only 4sat/vB. The Package RBF rule cannot be loosened to only allow<= br>the child to have new unconfirmed inputs, either, because it can still c= ause us
to overestimate the package's ancestor score.

However= , enforcing a rule analogous to BIP125#2 would not only make Package RBFless useful, but would also break Package RBF for packages with parents al= ready
in the mempool: if a package parent has already been submitted, it= would look
like the child is spending a "new" unconfirmed inp= ut. In example K [18], we're
looking to replace M1 with the entire p= ackage including P1, P2, and P3. We must
consider the case where one of = the parents is already in the mempool (in this
case, P2), which means we= must allow P3 to have new unconfirmed inputs. However,
M2 lowers the an= cestor score of P3 to 4.3sat/vB, so we should not replace M1
with this p= ackage.

Thus, the package RBF rule regarding new unconfirmed inputs = is less strict than
BIP125#2. However, we still achieve the same goal of= requiring the replacement
transactions to have a ancestor score at leas= t as high as the original ones. As
a result, the entire package is requi= red to be a higher feerate mining candidate
than each of the replaced tr= ansactions.

Another note: the [comment][13] above the BIP125#2 code = in the original RBF
implementation suggests that the rule was intended t= o be temporary.

##### Absolute Fee (Rule #3)

The package must= increase the absolute fee of the mempool, i.e. the total fees
of the pa= ckage must be higher than the absolute fees of the mempool transactions
= it replaces. Combined with the CPFP rule above, this differs from BIP125 Ru= le #3
- an individual transaction in the package may have lower fees tha= n the
=C2=A0 transaction(s) it is replacing. In fact, it may have 0 fees= , and the child
pays for RBF.

##### Feerate (Rule #4)

The = package must pay for its own bandwidth; the package feerate must be higher<= br>than the replaced transactions by at least minimum relay feerate
(`in= crementalRelayFee`). Combined with the CPFP rule above, this differs fromBIP125 Rule #4 - an individual transaction in the package can have a lowe= r
feerate than the transaction(s) it is replacing. In fact, it may have = 0 fees,
and the child pays for RBF.

##### Total Number of Replace= d Transactions (Rule #5)

The package cannot replace more than 100 me= mpool transactions. This is identical
to BIP125 Rule #5.

### Expe= cted FAQs

1. Is it possible for only some of the package to make it = into the mempool?

=C2=A0 =C2=A0Yes, it is. However, since we evict t= ransactions from the mempool by
descendant score and the package child i= s supposed to be sponsoring the fees of
its parents, the most common sce= nario would be all-or-nothing. This is
incentive-compatible. In fact, to= be conservative, package validation should
begin by trying to submit al= l of the transactions individually, and only use the
package mempool acc= eptance logic if the parents fail due to low feerate.

2. Should we a= llow packages to contain already-confirmed transactions?

=C2=A0 =C2= =A0 No, for practical reasons. In mempool validation, we actually aren'= t able to
tell with 100% confidence if we are looking at a transaction t= hat has already
confirmed, because we look up inputs using a UTXO set. I= f we have historical
block data, it's possible to look for it, but t= his is inefficient, not always
possible for pruning nodes, and unnecessa= ry because we're not going to do
anything with the transaction anywa= y. As such, we already have the expectation
that transaction relay is so= mewhat "stateful" i.e. nobody should be relaying
transactions = that have already been confirmed. Similarly, we shouldn't be
relayin= g packages that contain already-confirmed transactions.

[1]: https:= //github.com/bitcoin/bitcoin/pull/22290
[2]: https://github.c= om/bitcoin/bips/blob/1f0b563738199ca60d32b4ba779797fc97d040fe/bip-0141.medi= awiki#transaction-size-calculations
[3]: https://github.com/bitcoin/bitcoin/blob= /94f83534e4b771944af7d9ed0f40746f392eb75e/src/policy/policy.cpp#L282[4]: https://github.com/bitcoin/bitcoin/pull/16400
[5]: https://g= ithub.com/bitcoin/bitcoin/pull/21062
[6]: https://github.com/bitcoi= n/bitcoin/pull/22675
[7]: https://github.com/bitcoin/bitcoin/pull/2= 2796
[8]: https://github.com/bitcoin/bitcoin/pull/20833
[9]:= https://github.com/bitcoin/bitcoin/pull/21800
[10]: https://githu= b.com/bitcoin/bitcoin/pull/16401
[11]: https://github.com/bitcoin/b= itcoin/pull/19621
[12]: https://github.com/bitcoin= /bips/blob/master/bip-0125.mediawiki
[13]: https://githu= b.com/bitcoin/bitcoin/pull/6871/files#diff-34d21af3c614ea3cee120df276c9c4ae= 95053830d7f1d3deaf009a4625409ad2R1101-R1104
[14]: https://user-images.githubuserconten= t.com/25183001/133567078-075a971c-0619-4339-9168-b41fd2b90c28.png
[1= 5]: https://user-i= mages.githubusercontent.com/25183001/132856734-fc17da75-f875-44bb-b954-cb7a= 1725cc0d.png
[16]: https://user-images.githubusercontent.com/25183001/133567347-a3e2e4= a8-ae9c-49f8-abb9-81e8e0aba224.png
[17]: https://user-images.githubusercontent.com/251= 83001/133567370-21566d0e-36c8-4831-b1a8-706634540af3.png
[18]: https://user-images.git= hubusercontent.com/25183001/133567444-bfff1142-439f-4547-800a-2ba2b0242bcb.= png
[19]: h= ttps://user-images.githubusercontent.com/25183001/133456219-0bb447cb-dcb4-4= a31-b9c1-7d86205b68bc.png
[20]: https://user-images.githubusercontent.com/25183001/132= 857787-7b7c6f56-af96-44c8-8d78-983719888c19.png
_______________________________________________
bitcoin-dev mailing list
= bitcoin-dev@lists.linuxfoundation.org
https://lists.linuxfoundation.org/mail= man/listinfo/bitcoin-dev
_______________________________________________
bitcoin-dev mailing list
= bitcoin-dev@lists.linuxfoundation.org
https://lists.linuxfoundation.org/mail= man/listinfo/bitcoin-dev
--000000000000a502f005cd2105e8--