Re: [bitcoindev] A "Free" Relay Attack Taking Advantage of The Lack of Full-RBF In Core

[Antoine Riard <antoine.riard@gmail.com>]

[-- Attachment #1.1: Type: text/plain, Size: 25607 bytes --]


Hi Dave,

Thanks for your thoughtful answer (even if its wasn't addressed to me 
primarly).

> I cannot imagine what would make you think that protocol developers are
> not concerned about attacks that could drive large numbers of relay
> nodes off the network for a cost easily affordable to any well-funded
> adversary.

From my experience code reviewing the wallet / mempool re-broadcast of few
years ago, free tx-relay / bandwidth waste attacks were far to be 
understood 
or plainly weighted by some contributors of a newer generations, including 
by
the own champion of the proposal. The proposal was finally abandonned when a
more senior dev came up with quantitative analysis of code changes [0].

[0] https://github.com/bitcoin/bitcoin/pull/21061#issuecomment-851563105

> In this case, you've found a specific instance (full-RBF vs signaled
> RBF) of a well-known general problem (optional policies leading to
> mempool inconsistencies, allowing free relay) and appear to be arguing
> that devs don't care about free relay because they refused to reverse a
> previous decision (to not change the RBF configuration default) that has
> been hotly debated multiple times.

I think what is more interesting and noteworhty in the whole line of 
reaosning
of Peter with the present disclosure is how much the adversial effect is 
favor
by the supermajority of miners running `mempoolfullrbf` [1].

[1] https://github.com/bitcoin/bitcoin/pull/28132#issue-1817178316

That Peter's sample calls for empirical measures of their own by other 
contributors 
that can only make the whole review process and consensus-building towards 
a change
or the status quo better. Being able to empirically measure how bitcoin 
full-node
software is practically used by end-users, be it miners, second-layers 
nodes, simple
coins wallet users, exchanges, and making that a protocol development 
standard I believe
that how we progress as an ecosystem.

> An alternative interpretation is that they (1) do care about free relay,
> (2) recognize that solving free relay is a hard problem that requires
> research and development, and (3) refuse to be forced into restarting
> debate on an already overdiscussed issue that gets nobody closer to
> fundamental solutions.

The strong statement we have from the bitcoin core about attacks and 
vulnerabilities
mitigation, which probably represents the viewpoint of many generation of
contributors is the following as old than few weeks ago:

"The project has historically done a poor job at publicly disclosing 
security-critical bugs,
whether externally reported or found by contributors. This has led to a 
situation where a lot
of users perceive Bitcoin Core as never having bugs. This perception is 
dangerous and, 
unfortunately, not accurate."

[2] https://groups.google.com/g/bitcoindev/c/Q2ZGit2wF7w

Under those conditions, where it took 9 years for the bitcoin core project 
to disclosre
some vulnerabilitires, personally I'm more likely to understand that the 
bitcoin core project
is under-staffed is competent experts to keep public disclosure in 
reasonable timeframe (-- at
least equivalent to the kernel world), and as corollorary to fully evaluate 
technical proposal
with all its strength and weaknesses.

Saying an "already overdiscussed issues that gets nobody closer to 
fundamental solutions" is
insulting for Peter, honestly.

> Many protocol developers have attempted to address the problem over the 
years, most recently just
> a few months ago with an updated proposal for using weak blocks as a 
first step to address
> "diverging mempool policies".[1]

With all my respect for the developers involved in the design of this 
proposal who are skilled bitcoin
engineers in themsleves, I think betting on the "weak blocks" idea is a 
dead end in itself as in relying
on the slow convergence idea, that it's at the heart of the ghost, block 
dag and other ideas we've seen
10 years ago to make mining income "more fair" among geographically 
distributed mining pools.

I conjecture, it's a dead end as you will more or less have to re-build a 
secondary consensus algorithm
in the block-relay stack or in the mining stack, which is going to run 
intou its own mempool convergence
issues again, in a recursive fashion.

I believe such proposal is more an instance of the ill-managed conflict of 
interest blockstream-style
syndrom that bitcoin protocol devs, who ever they care can be affected 
with. In the present instance:

"My employer has gained a commercial interest in mining business recently 
so now let's advocate a very
complex bitcoin p2p protocol changes, rather than humbly realize that 
mining economic units are more akin
to a traditional energy company which not really in the less than 2-decades 
corporate DNA of said employer,
and why actually the economic units results of our mining division are 
floppy and now we engage in bitcoin
core changes at our advantage" (in my own reasonable view -- though I can 
only invite the "weak block" employer
company corporate board to comment here if they diverge on my analysis).

From a more pure technical viewpoint, I think bitcoin core compact blocks 
block-relay handling code, after
7 years of having been merged in bitcoin core has still `TODO` left in the 
codebase, and it could be an area
worhty of more testing / cleaning refactoring, I'm just saying by the way.

> You've previously argued against designing contract protocols to depend 
> CPFP-style fee bumping for their offchain transactions, however that is 
> what is widely used by LN today and it appears to be what LN and other 
> offchain protocol developers want to use in the future. If we accept 
> that, at least for the purposes of argument, what can we do to improve 
> CPFP fee bumping for LN? 

As an offchain protocol developers which has been involved in the majority 
of technical conversations,
implementations and deployment of the "anchor output" upgrade, I believe on 
the long-term CPFP-style fee-bumping
of contract protocol, including lighting, is not the most robust technical 
solutions. I think anyone can check
in the bitcoin optech anchor output glossary the numerous vulnerabilities 
that have plagued this fee-bumping 
solutions over the past years.

Pursuing on that trend will more only probably lead to a current state of 
what we have with DNS / BGP at the
internet-stack level, which are still constantly the sources of new 
security vulnerabilitiies and single-point
of-failure securiyt solutions like X509 certificates management (do we wish 
bitcoin to be plagued by crowdstyle
technical incident 15 years from now ? I'm not so sure).

> All we really need at this point is to enable package relay so that 
> parent transactions are no longer subject to a dynamic mempool minimum 
> when they're bundled with a high-feerate child. Preliminary support for 
> that should be released in the next major version of Bitcoin Core, with 
> improved support being worked on for later releases. 
> 
> Package relay is the only thing we need at this point because the 
> existing LN protocol makes use of CPFP carve-out, which minimizes 
> transaction pinning issues. 

I don't disagree on package relay deployment at this stage, with the some 
degree of technical skeptiscism
that shall accomaphny all p2p proposals. Let's remember things that have 
been rollback or slowly rollout like
bip37 or `MEMPOOL` p2p message.

> However, another substantial Bitcoin Core improvement, cluster mempool, 
> is incompatible with CPFP carve-out. TRUC is an alternative to CPFP 
> carve-out that is easy to reason about, easy to use, more general in 
> nature (good news for newer contract protocols), and which can possibly 
> be automatically applied to existing LN onchain transactions, allowing 
> cluster mempool to be deployed ASAP without requiring any significant 
> changes to anyone else's software. 

If you or anoyone think TRUC as an alternative to the CPFP as a transsction 
pinning mitigation as argued
in its merged BIP is easy to reason on, thanks to communicate your 
lightning node pubkey, with TRUC patch
applied and a GPG-signed message authorizing me to demonstrate the security 
properties of this solutions have
not been submitted to a fully contradictory evaluation.

I pointed out few years ago that relying on mempool-policy changes as a 
fully hardening solution was very
likely limited to mitigate econommical-based pinnings (i.e bip125 absolute 
fee rule), and this was before
the discovery of new transaction-relay jamming vector (a terminology 
catched by your himself in some
private conversation if my memory is correct) like replacement cycling 
attacks [4].

[4] 
https://lists.linuxfoundation.org/pipermail/bitcoin-dev/2021-June/019084.html 
"high-half" in this post
corresponds to what roughly is bip331 today.

> I don't think we need to be overly concerned about large numbers of LN 
users suddenly performing a malicious
> action that does not benefit them and does not put the network at risk.

Have a look on base-layer based exploitation description that could provoke 
a large number of LN users, without
their own involvement, that could put the base-network at risk [5].

[5] 
https://github.com/jamesob/mempool.work?tab=readme-ov-file#failure-one-mempool-to-rule-them-all

> The alternative to TRUC that you've proposed is replace-by-feerate 
> (RBFr). This is also compatible with existing LN transactions and it's 
> conceptually super simple (I assume the code is too), which is 
> wonderful. However, it's downsides are: 

About replace-by-feerate, I think it's a solution that have downside on its 
own, especially for
second-layers like lightning. I never took time to write a canonical post 
as infosec ethics would ask
me to notice lightning maininters to do first, even if the vulnerability 
idea has been vaguely sketched
out to few years ago.

On all your replace-by-feerate analysis, I think it deserves a thread on 
its own, as somehow the issue
can have interactions with degree of fulfillness of miner block templates, 
a conversation we had when V3
was proposed as solution [6].

[6] 
https://lists.linuxfoundation.org/pipermail/bitcoin-dev/2022-January/019817.html 
and following answers

> but the deployment of cluster mempool requires removal of CPFP carve-out, 
and removal of CPFP carve-out requires
> either the upgrade of thousands of LN nodes and channels or a drop-in 
solution (ideally one that can be analyzed 
> independently from cluster mempool, like TRUC)

And as I observed on one of the V3 PR threads and corresponding 
conversations, the CPFP carve-out do not provide
security to lightning implementation as (a) old revoked commitment 
transactions can be used to fail the CPFP bump,
V3 or not V3 and (b) there is no base-layer p2p protocol discovery of txids 
[6]. And I won't observe that when I looked
on lightning implementations code, only Eclair have minimal correct 
implementation of remote valid commitment transactions
broadcast, not the other one as it's not even in the bolt spec.

[6] https://github.com/bitcoin/bitcoin/pull/28948

So unless someone argued to the contrary, saying TRUC was needed to then 
deploy cluster mempool is an intellectual
fallacy, of which the origin could likely arises from the back macket of PR 
reviews trading happening in the bitcoin core
project for people to advance their own pet projects. At the very least, 
from my own code review of the folks bitcoin
core PR related to this line of deployment you're exposing, I'm not sure 
they're understanding so well cross-layers issues [7].

[7] https://github.com/bitcoin/bitcoin/pull/29427

I think you're rather making the points that the current group of most 
active bitcoin core contributors, are indeed
skilled bitcoin engineers, vetted with project-management know-how and 
informed about bitcoin technical history and know
how to stay polite and civil (most of the time) to create a fruitful 
technical conversation atmosphere. They're not necessarily
the most astute and adversarial minds to spot on the flight the weakness of 
new technical proposals, neither with the character
to say so _at the cost_ of sometimes being rude and negative in the 
communication tone. And this in line with my point above
recalling that the bitcoin core group of people themselves estimate 
themsleves they don't do a good job in term of security
vulnerabilties handling.

> There have already been multiple public discussions about how improved 
> RBF policies can be implemented once cluster mempool is available, many 
> of which bring us closer to something like RBFr in a way that's easier 
> to prove won't enable free relay, and perusing that seems to me like a 
> productive outlet if you are interested. 

I already reviewed some parts of cluster mempool. Fundamentally, economical 
mempool pinnings for second-layers (bip125 absolute
fee) with pre-signed time-sensitive transactions arises from a world where 
there is (a) an asynchronicity of mempools and (b) one
cannot guess feerates at block + 1 (-- let's say in a deterministic fashion 
as understood by traditional cryptographic litterature
when doing cryptanalysis). Better RBF policies won't solve that, including 
RBFr.

> After several hours of writing and thinking, I don't see any problems
> with the current approach using TRUC or with the general lack of
> interest in RBFr solutions at this time. I've tried to explain how I
> arrived at my conclusions at each step and I welcome any corrections.

On my personal perspective, and after careful considerations of all the 
technical points you've raised. I still think TRUC
has a lot of problems. RBRFr too has technical problems. About TRUC I think 
it's an acceptable temporary solution to minimize
the pinning surface of lightning implementations, pending for the design of 
a better solution (more likely at the consensus-level,
here consistently as I pointed out 3 years ago). Though we shall be honest 
with ourselves and not engage in overdue security theater
about TRUC properties.

To finish, I'll re-assert what is my appreciation. Currently, I don't think 
the bitcoin core folks handling the security list has
necessarily the proven level of competency to fully evaluate the issues 
reported by Peter, however I don't believe in the present
case they have done anything wrong infosec ethically-wise.

Best,
Antoine
ots hash: 001e8dc7fbced4cb9ea30c8a3b7fa22dc1b07939e7125cb46e7a3d65b256caa7

Le samedi 20 juillet 2024 à 07:45:24 UTC+1, David A. Harding a écrit :

> On 2024-07-18 05:56, Peter Todd wrote:
> > I disclosed it to the bitcoin-security mailing list as a test: does
> > Bitcoin Core actually care about free relay attacks?
>
> They do. Several free relay attacks that were present in earlier
> versions of Bitcoin were eliminated in later versions. New proposals
> are evaluated for their potential to create new permanent free relay
> vectors. The discovery of free relay is almost always reason enough to
> reject a proposal.
>
> The free relay attack you describe in your email and the type of free
> relay enabled by your replace-by-feerate (RBFr) proposal can allow an
> attacker to 10x to 100x the amount of bandwidth used network wide by
> relay nodes for a cost of $10,000 to $50,000 a day (or, as you mention,
> effectively for free if they were going to send a bunch of transactions
> anyway).
>
> I cannot imagine what would make you think that protocol developers are
> not concerned about attacks that could drive large numbers of relay
> nodes off the network for a cost easily affordable to any well-funded
> adversary.
>
> In this case, you've found a specific instance (full-RBF vs signaled
> RBF) of a well-known general problem (optional policies leading to
> mempool inconsistencies, allowing free relay) and appear to be arguing
> that devs don't care about free relay because they refused to reverse a
> previous decision (to not change the RBF configuration default) that has
> been hotly debated multiple times.
>
> An alternative interpretation is that they (1) do care about free relay,
> (2) recognize that solving free relay is a hard problem that requires
> research and development, and (3) refuse to be forced into restarting
> debate on an already overdiscussed issue that gets nobody closer to
> fundamental solutions.
>
> > I believe the authors of that BIP are fully aware of the fact that
> > "free" relay is an unavoidable problem, making their rational for
> > TRUC/V3 bogus
>
> Differences in node policy leading to mempool inconsistencies (which
> allows free relay) is a well known problem that's the result of Bitcoin
> being an open protocol based on free/libre software (two things I think
> we all want). Many protocol developers have attempted to address the
> problem over the years, most recently just a few months ago with an
> updated proposal for using weak blocks as a first step to address
> "diverging mempool policies".[1]
>
> A currently unsolved problem is not necessarily an unavoidable problem
> and it seems reasonable to me for devs to be skeptical of proposals like
> RBFr that add to the list of things that enable free relay.
>
> > I believe the authors of [BIP431...] don't want to admit that they've
> > wasted a large amount of engineering time on a bad proposal.
>
> You've previously argued against designing contract protocols to depend
> CPFP-style fee bumping for their offchain transactions, however that is
> what is widely used by LN today and it appears to be what LN and other
> offchain protocol developers want to use in the future. If we accept
> that, at least for the purposes of argument, what can we do to improve
> CPFP fee bumping for LN?
>
> All we really need at this point is to enable package relay so that
> parent transactions are no longer subject to a dynamic mempool minimum
> when they're bundled with a high-feerate child. Preliminary support for
> that should be released in the next major version of Bitcoin Core, with
> improved support being worked on for later releases.
>
> Package relay is the only thing we need at this point because the
> existing LN protocol makes use of CPFP carve-out, which minimizes
> transaction pinning issues.
>
> However, another substantial Bitcoin Core improvement, cluster mempool,
> is incompatible with CPFP carve-out. TRUC is an alternative to CPFP
> carve-out that is easy to reason about, easy to use, more general in
> nature (good news for newer contract protocols), and which can possibly
> be automatically applied to existing LN onchain transactions, allowing
> cluster mempool to be deployed ASAP without requiring any significant
> changes to anyone else's software.
>
> As you've shown[2], the main downside of TRUC transactions (if we're
> going to depend on CPFP-style fee bumping anyway) is that users of TRUC
> transactions who have a malicious counterparty might need to pay a
> slightly higher onchain feerate than they would need to pay under the
> same situation with CPFP carve-out. However, the increase is small
> enough that most honest parties should be able to afford it, so there's
> no incentive for a counterparty to do it. I don't think we need to be
> overly concerned about large numbers of LN users suddenly performing a
> malicious action that does not benefit them and does not put the network
> at risk.
>
> The alternative to TRUC that you've proposed is replace-by-feerate
> (RBFr). This is also compatible with existing LN transactions and it's
> conceptually super simple (I assume the code is too), which is
> wonderful. However, it's downsides are:
>
> 1. It fundamentally enables a significant amount of free relay. I'll
> sketch a super basic attack at the end of this email that costs 0.55
> BTC per day ($67,000 USD) to 100x the amount of bandwidth used by
> relay nodes. I'm sure more efficient attacks are possible.
>
> An attacker who is able to consume an excessive amount of relay node
> bandwidth at relatively low cost may be able to create both
> short-term and long-term problems for all Bitcoin users. If the
> created problems result in a rapid increase in user feerates, then
> free relay attacks become cheaper due to low feerate transactions
> being automatically evicted from the bottom of the mempool.
>
> 2. It may allow empting the mempool at relatively low cost. An attacker
> sending just 750 transactions at the top mempool feerate can
> guarantee eviction of every honest user's transactions. The attacker
> can then replace 300 MB of transactions with a single 43,179 vbyte
> transaction. Even if the replacement transaction pays 100
> sats/vbyte, when you compare that to the 1,000,000 vbytes of
> next-block transactions each miner lost, the miner is only earning an
> effective feerate of 4.3 sats/vbyte.
>
> Further, it's easy to imagine situations where evicting
> time-sensitive transactions from mempools might allow theft of funds
> in excess of a few thousand dollars (my immediate thoughts go to
> situations involving watchtowers).
>
> 3. Limiting the worst-case free relay and excessive mempool eviction
> requires additional rules (e.g. one-shot RBFr) that are challenging
> to implement and analyze at present, as several devs have noted[3].
> Both implementation and analysis should become much easier if cluster
> mempool is deployed (also noted by devs), but the deployment of
> cluster mempool requires removal of CPFP carve-out, and removal of
> CPFP carve-out requires either the upgrade of thousands of LN nodes
> and channels or a drop-in solution (ideally one that can be analyzed
> independently from cluster mempool, like TRUC).
>
> To me, TRUC seems like an excellent approach. It's something that can
> be evaluated independently of cluster mempool but which can help allow
> that deployment to proceed (in addition to the other previously
> described benefits that TRUC brings).
>
> There have already been multiple public discussions about how improved
> RBF policies can be implemented once cluster mempool is available, many
> of which bring us closer to something like RBFr in a way that's easier
> to prove won't enable free relay, and perusing that seems to me like a
> productive outlet if you are interested.
>
> > this is quite an odd case of Core politics
>
> I began writing this reply to force me to examine your claims for
> myself. You have a long history of noticing things other people missed.
> I was worried that some compelling point of yours was being ignored as
> the result of past controversies.
>
> After several hours of writing and thinking, I don't see any problems
> with the current approach using TRUC or with the general lack of
> interest in RBFr solutions at this time. I've tried to explain how I
> arrived at my conclusions at each step and I welcome any corrections.
>
> Thanks,
>
> -Dave
>
> [1] https://delvingbitcoin.org/t/second-look-at-weak-blocks/805
> [2] 
>
> https://lists.linuxfoundation.org/pipermail/bitcoin-dev/2023-December/022211.html
> [3] 
>
> https://bitcoinops.org/en/newsletters/2024/02/07/#proposal-for-replace-by-feerate-to-escape-pinning
>
> ---
> A simple free relay attack using RBFr
>
> ## Constants
>
> 100,000 vbytes == ~400,000 bytes
> A 1-input, 1-output P2TR scriptpath spend with the maximum amount
> of witness data allowed by Bitcoin Core defaults
>
> 111 vbytes == 162 bytes
> A 1-input, 1-output P2TR keypath spend
>
> ## Attack
>
> - Attacker obtains 500,000 UTXOs
>
> - For each UTXO
>
> - At the dynamic mempool minimum, broadcasts a 100,000 vbyte (400,000
> byte) transacton.
>
> - Waits for it to propagate.
>
> - At 1.25x the dynamic mempool minimum, broadcasts a RBFr replacement
> that is 111 vbytes (162 bytes).
>
> ## Analysis
>
> - At 162 bytes each, 500,000 transactions is 81 MB. I think that will
> fit in a default-sized mempool.
>
> - The total amount of transaction data relayed is 500_000 * (400_000 +
> 162), or about 200 GB.
>
> - The typical daily bandwidth requirement of a blocks-only node is
> roughly 2.5 MB * 144, or about 0.36 GB per day. Ideally a relaying
> node is about the same due to compact blocks, but realistically, it's
> a small multiple of that. Call it 2 GB per day.
>
> - This implies the attack push each RBFr relay node to use 100x a
> non-RBFr relay node.
>
> - At 111 vbytes each, 500,000 transactions is 55.5 million vbytes. At
> my nodes current mempoolminfee (1 sat/vb), that's 55.5 million sats,
> or 0.55 BTC (about $37,000 USD).
>
> - This analysis ignores the cost of obtaining the UTXOs and possibly
> later consolidating them, but it also ignores some easy optimizations
> such as batching the replacements.
>

-- 
You received this message because you are subscribed to the Google Groups "Bitcoin Development Mailing List" group.
To unsubscribe from this group and stop receiving emails from it, send an email to bitcoindev+unsubscribe@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/bitcoindev/99f8b3b5-996e-41a4-bca8-eb1ddcba4ef3n%40googlegroups.com.

[-- Attachment #1.2: Type: text/html, Size: 28823 bytes --]