Great Consensus Cleanup Revival

[Antoine Poinsot]

Hey all,

I've recently posted about the Great Consensus Cleanup there: https://delvingbitcoin.org/t/great-consensus-cleanup-revival/710.

I'm starting a thread on the mailing list as well to get comments and opinions from people who are not on Delving.

TL;DR:
- i think the worst block validation time is concerning. The mitigations proposed by Matt are effective, but i think we should also limit the maximum size of legacy transactions for an additional safety margin;
- i believe it's more important to fix the timewarp bug than people usually think;
- it would be nice to include a fix to make coinbase transactions unique once and for all, to avoid having to resort back to doing BIP30 validation after block 1,983,702;
- 64 bytes transactions should definitely be made invalid, but i don't think there is a strong case for making less than 64 bytes transactions invalid.

Anything in there that people disagree with conceptually?
Anything in there that people think shouldn't (or don't need to) be fixed?
Anything in there which can be improved (a simpler, or better fix)?
Anything NOT in there that people think should be fixed?


Antoine Poinsot

[Antoine Riard]

Hi Poinsot,

I think fixing the timewarp attack is a good idea, especially w.r.t safety implications of long-term timelocks usage.

The only beneficial case I can remember about the timewarp issue is "forwarding blocks" by maaku for on-chain scaling:

http://freico.in/forward-blocks-scalingbitcoin-paper.pdf

Shall we as a community completely disregard this approach for on-chain settlement throughput scaling ?

Personally, I think you can still design extension-block / side-chains like protocols by using other today available

Bitcoin Script mechanisms and get roughly (?) the same security / scalability trade-offs. Shall be fine to me to fix timewarp.

Worst-block validation time is concerning. I bet you can do worst than your examples if you're playing with other vectors like

low-level ECC tricks and micro-architectural layout of modern processors.

Consensus invalidation of old legacy scripts was quite controversial last time a consensus cleanup was proposed:

https://lists.linuxfoundation.org/pipermail/bitcoin-dev/2019-March/016714.html

Only making scripts invalid after a given block height (let's say the consensus cleanup activation height) is obviously a

way to solve the concern and any remaining sleeping DoSy unspent coins can be handled with  newly crafted and dedicated

transaction-relay rules (e.g at max 1000 DoSy coins can be spent per block for a given IBT span).

I think any consensus boundaries on the minimal transaction size would need to be done carefully and have all lightweight

clients update their own transaction acceptance logic to enforce the check to avoid years-long transitory massive double-spend

due to software incoordination. I doubt `MIN_STANDARD_TX_NON_WITNESS_SIZE` is implemented correctly by all transaction-relay

backends and it's a mess in this area. Quid if we have  < 64 bytes transaction where the only witness is enforced to be a minimal 1-byte

as witness elements are only used for higher layers protocols semantics  ? Shall get its own "only-after-height-X" exemption, I think.

Making coinbase unique by requesting the block height to be enforced in nLocktime, sounds more robust to take a monotonic counter

in the past in case of accidental or provoked shallow reorgs. I can see of you would have to re-compute a block template, loss a round-trip

compare to your mining competitors. Better if it doesn't introduce a new DoS vector at mining job distribution and control.

Beyond, I don't deny other mentioned issues (e.g UTXO entries growth limit) could be source of denial-of-service but a) I think it's hard

to tell if they're economically neutral on modern Bitcoin use-cases and their plausible evolvability and b) it's already a lot of careful consensus

code to get right :)

Best,

Antoine

[Antoine Poinsot]

Hi Poinsot,

Hi Riard,

The only beneficial case I can remember about the timewarp issue is "forwarding blocks" by maaku for on-chain scaling:

http://freico.in/forward-blocks-scalingbitcoin-paper.pdf

I would not qualify this hack of "beneficial". Besides the centralization pressure of an increased block frequency, leveraging the timewarp to achieve it would put the network constantly on the Brink of being seriously (fatally?) harmed. And this sets pernicious incentives too. Every individual user has a short-term incentive to get lower fees by the increased block space, at the expense of all users longer term. And every individual miner has an incentive to get more block reward at the expense of future miners. (And of course bigger miners benefit from an increased block frequency.)

I think any consensus boundaries on the minimal transaction size would need to be done carefully and have all lightweight

clients update their own transaction acceptance logic to enforce the check to avoid years-long transitory massive double-spend

due to software incoordination.

Note in my writeup i suggest we do not introduce a minimum transaction, but we instead only make 64 bytes transactions invalid. See https://delvingbitcoin.org/t/great-consensus-cleanup-revival/710#can-we-come-up-with-a-better-fix-10:

However the BIP proposes to also make less-than-64-bytes transactions invalid. Although they are of no (or little) use, such transactions are not harmful. I believe considering a type of transaction useless is not sufficient motivation for making them invalid through a soft fork.

Making (exactly) 64 bytes long transactions invalid is also what AJ implemented in his pull request to Bitcoin-inquisition.

I doubt `MIN_STANDARD_TX_NON_WITNESS_SIZE` is implemented correctly by all transaction-relay backends and it's a mess in this area.

What type of backend are you referring to here? Bitcoin full nodes reimplementations? These transactions have been non-standard in Bitcoin Core for the past 6 years (commit 7485488e907e236133a016ba7064c89bf9ab6da3).

Quid if we have < 64 bytes transaction where the only witness is enforced to be a minimal 1-byte

as witness elements are only used for higher layers protocols semantics ?

This restriction is on the size of the transaction serialized without witness. So this particular instance would not be affected and whatever the witness is isn't relevant.

Making coinbase unique by requesting the block height to be enforced in nLocktime, sounds more robust to take a monotonic counter

in the past in case of accidental or provoked shallow reorgs. I can see of you would have to re-compute a block template, loss a round-trip

compare to your mining competitors. Better if it doesn't introduce a new DoS vector at mining job distribution and control.

Could you clarify? Are you suggesting something else than to set the nLockTime in the coinbase transaction to the height of the block? If so, what exactly are you referring to by "monotonic counter in the past"?

At any rate in my writeup i suggested making the coinbase commitment mandatory (even when empty) instead for compatibility reasons.

That said, since we could make this rule kick in in 25 years from now, we might want to just do the Obvious Thing and just require the height in nLockTime.

 and b) it's already a lot of careful consensus

code to get right :)

Definitely. I just want to make sure we are not missing anything important if a soft fork gets proposed along these lines in the future.

Best,

Antoine

Le dimanche 24 mars 2024 à 19:06:57 UTC, Antoine Poinsot a écrit :

Hey all,

I've recently posted about the Great Consensus Cleanup there: https://delvingbitcoin.org/t/great-consensus-cleanup-revival/710.

I'm starting a thread on the mailing list as well to get comments and opinions from people who are not on Delving.

TL;DR:
- i think the worst block validation time is concerning. The mitigations proposed by Matt are effective, but i think we should also limit the maximum size of legacy transactions for an additional safety margin;
- i believe it's more important to fix the timewarp bug than people usually think;
- it would be nice to include a fix to make coinbase transactions unique once and for all, to avoid having to resort back to doing BIP30 validation after block 1,983,702;
- 64 bytes transactions should definitely be made invalid, but i don't think there is a strong case for making less than 64 bytes transactions invalid.

Anything in there that people disagree with conceptually?
Anything in there that people think shouldn't (or don't need to) be fixed?
Anything in there which can be improved (a simpler, or better fix)?
Anything NOT in there that people think should be fixed?


Antoine Poinsot

--
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+...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/bitcoindev/dc2cc46f-e697-4b14-91b3-34cf11de29a3n%40googlegroups.com.

[Antoine Riard]

Hi Darosior,

> I would not qualify this hack of "beneficial". Besides the centralization pressure of an increased block frequency, leveraging the timewarp to achieve it would put the network constantly on the Brink of being seriously (fatally?) harmed. And this sets pernicious incentives too. Every individual user has a short-term incentive to get lower fees by the increased block space, at the expense of all users longer term. And every individual miner has an incentive to get more block reward at the expense of future miners. (And of course bigger miners benefit from an increased block frequency.)

I'm not saying the hack is beneficial either. The "forward block" paper is just good to provide more context around timewarp.

> Note in my writeup i suggest we do not introduce a minimum transaction, but we instead only make 64 bytes transactions invalid

I think it's easier for the sake of analysis.

See this mailing list issue for 60-byte example transaction use-case: https://lists.linuxfoundation.org/pipermail/bitcoin-dev/2020-May/017883.html

Only I'm aware of to the best of my knowledge.

> What type of backend are you referring to here?

I can't find where `MIN_STANDARD_TX_NON_WITNESS_SIZE` is checked in btcd's `maybeAcceptTransaction()`.

> This restriction is on the size of the transaction serialized without witness. 

Oky.

> Could you clarify? Are you suggesting something else than to set the nLockTime in the coinbase transaction to the height of the block? If so, what exactly are you referring to by "monotonic counter in the past"?

Thinking more, I believe it's okay to use the nLocktime in the coinbase transaction, as the wtxid of the coinbase is assumed to be 0x00.

To be checked if it doesn't break anything w.rt Stratum V2 / mining job distribution.

Best,

Antoine

  

[Mark F]

On Wednesday, March 27, 2024 at 4:00:34 AM UTC-7 Antoine Poinsot wrote:

The only beneficial case I can remember about the timewarp issue is "forwarding blocks" by maaku for on-chain scaling:

http://freico.in/forward-blocks-scalingbitcoin-paper.pdf

I would not qualify this hack of "beneficial". Besides the centralization pressure of an increased block frequency, leveraging the timewarp to achieve it would put the network constantly on the Brink of being seriously (fatally?) harmed. And this sets pernicious incentives too. Every individual user has a short-term incentive to get lower fees by the increased block space, at the expense of all users longer term. And every individual miner has an incentive to get more block reward at the expense of future miners. (And of course bigger miners benefit from an increased block frequency.)

 

Every single concern mentioned here is addressed prominently in the paper/presentation for Forward Blocks:

* Increased block frequency is only on the compatibility chain, where the content of blocks is deterministic anyway. There is no centralization pressure from the frequency of blocks on the compatibility chain, as the content of the blocks is not miner-editable in economically meaningful ways. Only the block frequency of the forward block chain matters, and here the block frequency is actually *reduced*, thereby decreasing centralization pressure.

* The elastic block size adjustment mechanism proposed in the paper is purposefully constructed so that users or miners wanting to increase the block size beyond what is currently provided for will have to pay significantly (multiple orders of magnitude) more than they could possibly acquire from larger blocks, and the block size would re-adjust downward shortly after the cessation of that artificial fee pressure.

* Increased block frequency of compatibility blocks has no effect on the total issuance, so miners are not rewarded by faster blocks.

You are free to criticize Forward Blocks, but please do so by actually addressing the content of the proposal. Let's please hold a standard of intellectual excellence on this mailing list in which ideas are debated based on content-level arguments rather than repeating inaccurate takes from Reddit/Twitter.

To the topic of the thread, disabling time-warp will close off an unlikely and difficult to pull off subsidy draining attack that to activate would necessarily require weeks of forewarning and could be easily countered in other ways, with the tradeoff of removing the only known mechanism for upgrading the bitcoin protocol to larger effective block sizes while staying 100% compatible with un-upgraded nodes (all nodes see all transactions).

I think we should keep our options open.

-Mark

[Antoine Poinsot]

You are free to criticize Forward Blocks, but please do so by actually addressing the content of the proposal. Let's please hold a standard of intellectual excellence on this mailing list in which ideas are debated based on content-level arguments rather than repeating inaccurate takes from Reddit/Twitter.

You are the one being dishonest here. Look, i understand you came up with a fun hack exploiting bugs in Bitcoin and you are biased against fixing them. Yet, the cost of not fixing timewarp objectively far exceeds the cost of making "forward blocks" impossible.

As already addressed in the DelvingBitcoin post:

1. The timewarp bug significantly changes the 51% attacker threat model. Without exploiting it a censoring miner needs to continuously keep more hashrate than the rest of the network combined for as long as he wants to prevent some people from using Bitcoin. By exploiting timewarp the attacker can prevent everybody from using Bitcoin within 40 days.
2. The timewarp bug allows an attacking miner to force on full nodes more block data than they agreed to. This is actually the attack leveraged by your proposal. I believe this variant of the attack is more likely to happen, simply for the reason that all participants of the system have a short term incentive to exploit this (yay lower fees! yay more block subsidy!), at the expense of the long term health of the system. As the block subsidy exponentially decreases miners are likely to start playing more games and that's a particularly attractive one. Given the level of mining centralization we are witnessing [0] i believe this is particularly worrisome.
3. I'm very skeptical of arguments about how "we" can stop an attack which requires "weeks of forewarning". Who's we? How do we proceed, all Bitcoin users coordinate and arbitrarily decide of the validity of a block? A few weeks is very little time if this is at all achievable. If you add on top of that the political implications of the previous point it gets particularly messy.

I've got better things to do than to play "you are being dishonest! -no it's you -no you" games. So unless you bring something new to the table this will be my last reply to your accusations.

Antoine

[0] https://x.com/0xB10C/status/1780611768081121700

--

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+...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/bitcoindev/62640263-077c-4ac7-98a6-d9c17913fca0n%40googlegroups.com.

[Antoine Riard]

Hi Maaku,

> Every single concern mentioned here is addressed prominently in the paper/presentation for Forward Blocks:

>

> * Increased block frequency is only on the compatibility chain, where the content of blocks is deterministic anyway. There is no centralization pressure from the frequency > of blocks on the compatibility chain, as the content of the blocks is not miner-editable in economically meaningful ways. Only the block frequency of the forward block > chain matters, and here the block frequency is actually *reduced*, thereby decreasing centralization pressure.

>

> * The elastic block size adjustment mechanism proposed in the paper is purposefully constructed so that users or miners wanting to increase the block size beyond what > is currently provided for will have to pay significantly (multiple orders of magnitude) more than they could possibly acquire from larger blocks, and the block size would re-> adjust downward shortly after the cessation of that artificial fee pressure.

> * Increased block frequency of compatibility blocks has no effect on the total issuance, so miners are not rewarded by faster blocks.

> You are free to criticize Forward Blocks, but please do so by actually addressing the content of the proposal. Let's please hold a standard of intellectual excellence on this > mailing list in which ideas are debated based on content-level arguments rather than repeating inaccurate takes from Reddit/Twitter.

> To the topic of the thread, disabling time-warp will close off an unlikely and difficult to pull off subsidy draining attack that to activate would necessarily require weeks of > forewarning and could be easily countered in other ways, with the tradeoff of removing the only known mechanism for upgrading the bitcoin protocol to larger effective > block sizes while staying 100% compatible with un-upgraded nodes (all nodes see all transactions).

> I think we should keep our options open.

Somehow, I'm sharing your concerns on preserving the long-term evolvability w.r.t scalability options

of bitcoin under the security model as very roughly describer in the paper. Yet, from my understanding

of the forwarding block proposal as described in your paper, I wonder if the forward block chain could

be re-pegged to the main bitcoin chain using the BIP141 extensible commitment structure (assuming

a future hypothetical soft-fork).

From my understanding, it's like doubly linked-list in C, you just need a pointer in the BIP141 extensible

commitment structure referencing back the forward chain headers. If one wishes no logically authoritative

cross-chain commitment, one could leverage some dynamic-membership multi-party signature. This

DMMS could even be backup by proof-of-work based schemes.

The forward block chain can have higher block-rate frequency and the number of block headers be

compressed in a merkle tree committed in the BIP141 extensible commitment structure. Compression

structure can only be defined by the forward chain consensus algorithm to allow more efficient accumulator

than merkle tree to be used".

The forward block chain can have elastic block size consensus-bounded by miners fees on long period

of time. Transaction elements can be just committed in the block headers themselves, so no centralization

pressure on the main chain. Increased block frequency or block size on the forward block chain have not

effect on the total issuance (modulo the game-theory limits of the known empirical effects of colored coins

on miners incentives).

I think the time-warp issues opens the door to economically non-null exploitation under some scenarios

over some considered time periods. If one can think to other ways to mitigate the issue in minimal and

non-invasive way w.r.t current Bitcoin consensus rules and respecting un-upgraded node ressources

consumption, I would say you're free to share them.

I can only share your take on maintaining a standard of intellectual excellence on the mailing list,

and avoid faltering in Reddit / Twitter-style "madness of the crowd"-like conversations.

Best,

Antoine

[Mark F]

Hi Antoine,

That's a reasonable suggestion, and one which has been discussed in the past under various names. Concrete ideas for a pegged extension-block side chain go back to 2014 at the very least. However there is one concrete way in which these proposals differ from forward blocks: the replay of transactions to the compatibility block chain. With forward blocks, even ancient versions of bitcoind that have been running since 2013 (picked as a cutoff because of the probabilistic fork caused by v0.8) will see all blocks, and have a complete listing of all UTXOs, and the content of transactions as they appear.

Does this matter? In principle you can just upgrade all nodes to understand the extension block, but in practice for a system as diverse as bitcoin support of older node versions is often required in critical infrastructure. Think of all the block explorer and mempool websites out there, for example, and various network monitoring and charting tools. Many of which are poorly maintained and probably running on two or three year old versions of Bitcoin Core.

The forward blocks proposal uses the timewarp bug to enable (1) a proof-of-work change, (2) sharding, (3) subsidy schedule smoothing, and (4) a flexible block size, all without forcing any non-mining nodes to *have* to upgrade in order to regain visibility into the network. Yes it's an everything-and-the-kitchen-sink straw man proposal, but that was on purpose to show that all these so-called “hard-fork” changes can in fact be done as a soft-fork on vanilla bitcoin, while supporting even the oldest still-running nodes.

That changes if we "fix" the timewarp bug though. At the very least, the flexible block size and subsidy schedule smoothing can't be accomplished without exploiting the timewarp bug, as far as anyone can tell. Therefore fixing the timewarp bug will _permanently_ cutoff the bitcoin community from ever having the ability to scale on-chain in a backwards-compatible way, now or decades or centuries into the future.

Once thrown, this fuse switch can't be undone. We should be damn sure we will never, ever need that capability before giving it up.

Mark

[Antoine Riard]

Hi Maaku,

From reading back the "forward block" paper, while it effectively guarantees an on-chain settlment throughput increases without the
necessity to upgrade old clients, one could argue the proof-of-work change on the forward chain (unless it's a no-op double-sha256)
coupled with the subsidy schedule smoothing, constitutes a substantial change of the already-mined UTXO security model. You can
use a lot of hash functions as proof-of-work primitive, though it doesn't mean they are relying on as strong assumptions or level
of cryptanalysis.

In fine, you could have poorly picked up hash function for the forward chain resulting in a lowering of everyone coins security
(the > 100 TH/s of today is securing years old coins from block mined under < 1 TH/s). I hold the opinion that fundamental changes
affecting the security of everyone coins should be better to be opted-in by the super-economic majority of nodes, including non-mining
nodes. At the contrary, the "forward block" proposal sounds to make the point it's okay to update proof-of-work algorithm by a
combined set of mining nodes and upgraded non-mining nodes, which could hypothetically lead to a "security downgrade" due to weaker
proof-of-work algorithm used on the forward chain.

While your papers introduce formalization of both full-node cost of validation and censorship resistance concepts, one could also
add "hardness to change" as a property of the Bitcoin network we all cherishes. If tomorrow, 10% of the hahrate was able to enforce
proof-of-work upgrade to the broken SHA-1, I think we would all consider as a security downgrade.

Beyond, this is corect we have a diversity of old nodes used in the ecosystem, probably for block explorer and mempool websites.
Yet in practice, they're more certainly vectors of weakness for their end-users, as Bitcoin Core has sadly a limited security fixes
backport policy, which doesn't go as far as v0.8 for sure. That we can all deplore the lack of half-decade old LTS release policy for
Bitcoin Core, like done by the Linux kernel is a legitimate conversation to have (and it would be indeed make it easier with
libbitcoinkernel progress). I think we shall rather invite operators of oldest still-running nodes to upgrade to more recent
versions, before to ask them to go through the analytical process of weighting all the security / scalability trade-offs of a
proposal like "forward block".

Finally, on letting options open to bump block inter-val as a soft-fork on the compatibility chain, I think one could still have
a multi-stage "forward block" deployment, where a) a new difficutly adjustment algoritm with parameters is introduced bumping block
inter-val for upgraded mining nodes e.g a block every 400 s in average and the b) re-use this block inter-val capacity increase for
the forward chain flexible block size. Now why a miner would opt-in in such block-interval constraining soft-fork is a good question,
in a paradigm where they still get the same block subsidy distribution.

This is just a thought experiment aiming to invalidate the "as far as anyone can tell" statement on forclosing forever on-chain
settlement throughput increase, if we fix the timewarp bug.

Best,
Antoine

[Eric Voskuil]

Hi Antoine,

Regarding potential malleability pertaining to blocks with only 64 byte transactions, why is not a deserialization phase check for the coinbase input as a null point not sufficient mitigation (computational infeasibility) for any implementation that desires to perform permanent invalidity marking?

Best,
Eric

ref: Weaknesses in Bitcoin’s Merkle Root Construction

[Antoine Poinsot]

Hi Eric,

It is. This is what is implemented in Bitcoin Core, see this snippet and section 4.1 of the document you reference:

Another check that was also being done in CheckBlock() relates to the coinbase transaction: if the first transaction in a block fails the required structure of a coinbase – one input, with previous output hash of all zeros and index of all ones – then the block will fail validation. The side effect of this test being in CheckBlock() was that even though the block malleability discussed in section 3.1 was unknown, we were effectively protected against it – as described above, it would take at least 224 bits of work to produce a malleated block that passed the coinbase check.

Best,

Antoine

--

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+...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/bitcoindev/72e83c31-408f-4c13-bff5-bf0789302e23n%40googlegroups.com.

[Eric Voskuil]

Right, a fairly obvious resolution. My question is why is that not sufficient - especially given that a similar (context free) check is required for duplicated tx malleability? We'd just be swapping one trivial check (first input not null) for another (tx size not 64 bytes).

Best,
Eric

[Antoine Poinsot]

Making 64-bytes transactions invalid is indeed not the most pressing bug fix, but i believe it's still a very nice cleanup to include if such a soft fork ends up being seriously proposed.

As discussed here it would let node implementations cache block failures at an earlier stage of validation. Not a large gain, but still nice to have.

As discussed in the DelvingBitcoin post it would also be a small gain of bandwidth for SPV verifiers as they wouldn't have to query a merkle proof for the coinbase transaction in addition to the one for the transaction they're interested in. It would also avoid a large footgun for anyone implementing a software verifying an SPV proof verifier and not knowing the intricacies of the protocol which make such proofs not secure on their own today.

Finally, it would get rid of a large footgun in general. Certainly, unique block hashes would be a useful property for Bitcoin to have. It's not far-fetched to expect current or future Bitcoin-related software to rely on this.

Outlawing 64-bytes transactions is also a very narrow and straightforward change, with trivial confiscatory effect as any 64-bytes transactions would either be unspendable or an anyone-can-spend. Therefore i believe the benefits of making them illegal outweigh the costs.

Best,

Antoine

To view this discussion on the web visit https://groups.google.com/d/msgid/bitcoindev/5b0331a5-4e94-465d-a51d-02166e2c1937n%40googlegroups.com.

[Eric Voskuil]

Thanks for the responses Antoine.

>  As discussed here it would let node implementations cache block failures at an earlier stage of validation. Not a large gain, but still nice to have.

It is not clear to me how determining the coinbase size can be done at an earlier stage of validation than detection of the non-null coinbase. The former requires parsing the coinbase to determine its size, the latter requires parsing it to know if the point is null. Both of these can be performed as early as immediately following the socket read.

size check

(1) requires new consensus rule: 64 byte transactions (or coinbases?) are invalid.
(2) creates a consensus "seam"  (complexity) in txs, where < 64 bytes and > 64 bytes are potentially valid.
(3) can be limited to reading/skipping header (80 bytes) plus parsing 0 - 65 coinbase bytes.

point check

(1) requires no change.
(2) creates no consensus seam.
(3) can be limited to reading/skipping header (80 bytes) plus parsing 6 - 43 coinbase bytes.

Not only is this not a large (performance) gain, it's not one at all.

> It would also avoid a large footgun for anyone implementing a software verifying an SPV proof verifier and not knowing the intricacies of the protocol...

It seems to me that introducing an arbitrary tx size validity may create more potential implementation bugs than it resolves. And certainly anyone implementing such a verifier must know many intricacies of the protocol. This does not remove one, it introduces another - as there is not only a bifurcation around tx size but one around the question of whether this rule is active.

 
> Finally, it would get rid of a large footgun in general.

I do not see this. I see a very ugly perpetual seam which will likely result in unexpected complexities over time.

> Certainly, unique block hashes would be a useful property for Bitcoin to have. It's not far-fetched to expect current or future Bitcoin-related software to rely on this.

This does not produce unmalleable block hashes. Duplicate tx hash malleation remains in either case, to the same effect. Without a resolution to both issues this is an empty promise.

The only possible benefit that I can see here is the possible very small bandwidth savings pertaining to SPV proofs. I would have a very hard time justifying adding any consensus rule to achieve only that result.

Best,
Eric

[Antoine Poinsot]

It is not clear to me how determining the coinbase size can be done at an earlier stage of validation than detection of the non-null coinbase.

My point wasn't about checking the coinbase size, it was about being able to cache the hash of a (non-malleated) invalid block as permanently invalid to avoid re-downloading and re-validating it.

It seems to me that introducing an arbitrary tx size validity may create more potential implementation bugs than it resolves.

The potential for implementation bugs is a fair point to raise, but in this case i don't think it's a big concern. Verifying no transaction in a block is 64 bytes is as simple a check as you can get.

And certainly anyone implementing such a verifier must know many intricacies of the protocol.

They need to know some, but i don't think it's reasonable to expect them to realize the merkle tree construction is such that an inner node may be confused with a 64 bytes transaction.

I do not see this. I see a very ugly perpetual seam which will likely result in unexpected complexities over time.

What makes you think making 64 bytes transactions invalid could result in unexpected complexities? And why do you think it's likely?

This does not produce unmalleable block hashes. Duplicate tx hash malleation remains in either case, to the same effect. Without a resolution to both issues this is an empty promise.

Duplicate txids have been invalid since 2012 (CVE-2012-2459). If 64 bytes transactions are also made invalid, this would make it impossible for two valid blocks to have the same hash.

Best,
Antoine

--
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+...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/bitcoindev/be78e733-6e9f-4f4e-8dc2-67b79ddbf677n%40googlegroups.com.

[Eric Voskuil]

>> It is not clear to me how determining the coinbase size can be done at an earlier stage of validation than detection of the non-null coinbase.
> My point wasn't about checking the coinbase size, it was about being able to cache the hash of a (non-malleated) invalid block as permanently invalid to avoid re-downloading and re-validating it.

This I understood, but I think you misunderstood me. Your point was specifically that, "it would let node implementations cache block failures at an earlier stage of validation." Since you have not addressed that aspect I assume you agree with my assertion above that the proposed rule does not actually achieve this.

Regarding the question of checking coinbase size, the issue is of detecting (or preventing) hashes mallied via the 64 byte tx technique. A rule against 64 byte txs would allow this determination by checking the coinbase alone. If the coinbase is 64 bytes the block is invalid, if it is not the block hash cannot have been mallied (all txs must have been 64 bytes, see previous reference).

In that case if the block is invalid the invalidity can be cached. But block invalidity cannot actually be cached until the block is fully validated. A rule to prohibit *all* 64 byte txs is counterproductive as it only adds additional checks on typically thousands of txs per block, serving no purpose.

>> It seems to me that introducing an arbitrary tx size validity may create more potential implementation bugs than it resolves.
> The potential for implementation bugs is a fair point to raise, but in this case i don't think it's a big concern. Verifying no transaction in a block is 64 bytes is as simple a check as you can get.

You appear to be making the assumption that the check is performed after the block is fully parsed (contrary to your "earlier" criterion above). The only way to determine the tx sizes is to parse each tx for witness marker, input count, output count, input script sizes, output script sizes, witness sizes, and skipping over the header, several constants, and associated buffers. Doing this "early" to detect malleation is an extraordinarily complex and costly process. On the other hand, as I pointed out, a rational implementation would only do this early check for the coinbase.

Yet even determining the size of the coinbase is significantly more complex and costly than checking its first input point against null. That check (which is already necessary for validation) resolves the malleation question, can be performed on the raw unparsed block buffer by simply skipping header, version, reading input count and witness marker as necessary, offsetting to the 36 byte point buffer, and performing a byte comparison against [0000000000000000000000000000000000000000000000000000000000000000ffffffff].

This is:

(1) earlier
(2) faster
(3) simpler
(4) already consensus

>> And certainly anyone implementing such a verifier must know many intricacies of the protocol.
> They need to know some, but i don't think it's reasonable to expect them to realize the merkle tree construction is such that an inner node may be confused with a 64 bytes transaction.

A protocol developer needs to understand that the hash of an invalid block cannot be cached unless at least the coinbase has been restricted in size (under the proposal) -or- that the coinbase is a null point (presently or under the proposal). In the latter case the check is already performed in validation, so there is no way a block would presently be cached as invalid without checking it. The proposal adds a redundant check, even if limited to just the coinbase. [He must also understand the second type of malleability, discussed below.]

If this proposed rule was to activate we would implement it in a late stage tx.check, after txs/blocks had been fully deserialized. We would not check it an all in the case where the block is under checkpoint or milestone ("assume valid"). In this case we would retain the early null point malleation check (along with the hash duplication malleation check) that we presently have, would validate tx commitments, and commit the block. In other words, the proposal adds unnecessary late stage checks only. Implementing it otherwise would just add complexity and hurt performance.

>> I do not see this. I see a very ugly perpetual seam which will likely result in unexpected complexities over time.
> What makes you think making 64 bytes transactions invalid could result in unexpected complexities? And why do you think it's likely?

As described above, it's later, slower, more complex, unnecessarily broad, and a consensus change. Beyond that it creates an arbitrary size limit - not a lower or upper bound, but a slice out of the domain. Discontinuities are inherent complexities in computing. The "unexpected" part speaks for itself.

>> This does not produce unmalleable block hashes. Duplicate tx hash malleation remains in either case, to the same effect. Without a resolution to both issues this is an empty promise.
> Duplicate txids have been invalid since 2012 (CVE-2012-2459).

I think again here you may have misunderstood me. I was not making a point pertaining to BIP30. I was referring to the other form of block hash malleability, which results from duplicating sets of trailing txs in a single block (see previous reference). This malleation vector remains, even with invalid 64 byte txs. As I pointed out, this has the "same effect" as the 64 byte tx issue. Merkle hashing the set of txs is insufficient to determine identity. In one case the coinbase must be checked (null point or size) and in the other case the set of tx hashes must be checked for trailing duplicated sets. [Core performs this second check within the Merkle hashing algorithm (with far more comparisons than necessary), though this can be performed earlier and independently to avoid any hashing in the malleation case.]

I would also point out in the interest of correctness that Core reverted its BIP30 soft fork implementation as a consequence of the BIP90 hard fork, following and requiring the BIP34 soft fork that presumably precluded it but didn't, so it is no longer the case that duplicate tx hashes are invalid in implementation. As you have proposed in this rollup, this requires fixing again.

> If 64 bytes transactions are also made invalid, this would make it impossible for two valid blocks to have the same hash.

Aside from the BIP30/34/90 issue addressed above, it is already "impossible" (cannot be stronger than computationally infeasible) for two *valid* blocks to have the same hash. The proposal does not enable that objective, it is already the case. No malleated block is a valid block.

The proposal aims only to make it earlier or easier or faster to check for block hash malleation. And as I've pointed out above, it doesn't achieve those objectives. Possibly the perception that this would be the case is a consequence of implementation details, but as I have shown above, it is not in fact the case.

Given either type of malleation, the malleated block can be determined to be invalid by a context free check. But this knowledge cannot ever be cached against the block hash, since the same hash may be valid. Invalidity can only be cached once a non-mallied block is validated and determined to be invalid. Block hash malleations are and will remain invalid blocks with or without the proposal, and it will continue to be necessary to avoid caching invalid against the malleation. As you said:

> it was about being able to cache the hash of a (non-malleated) invalid block as permanently invalid to avoid re-downloading and re-validating it.

This is already the case, and requires validating the full non-malleated block. Adding a redundant invalidity check doesn't improve this in any way.

Best,
Eric

[Antoine Riard]

Hi Eric,

> It is not clear to me how determining the coinbase size can be done at an earlier stage of validation than

> detection of the non-null coinbase. The former requires parsing the coinbase to determine its size, the latter
> requires parsing it to know if the point is null. Both of these can be performed as early as immediately following the socket read.

If you have code in pure C with variables on the stack no malloc, doing a check of the coinbase size after the socket
read can be certainly more robust than checking a non-null pointer. And note the attacking game we're solving is a peer
passing a sequence of malleated blocks for which the headers have been already verified, so there we can only have weaker
assumptions on the computational infeasibility.

Introducing a discontinuity like ensuring that both leaf / non-leaf merkle tree nodes are belonging to different domains
can be obviously a source of additional software complexity, however from a security perspective discontinuities if they're
computational asymmetries at the advantage of validating nodes I think they can be worthy of considerations for soft-fork extensions.

After looking on the proposed implementation in bitcoin inquisition, I think this is correct that the efficiency
of the 64 byte technique transaction to check full block malleability is very implementation dependent. Sadly, I
cannot think about other directions to alleviate this dependence on the ordering of the block validation checks
from socket read.

In my reasonable opinion, it would be more constructive to come out with a full-fleshout "fast block malleability
validation" algorithm in the sense of SipHash (-- and see to have this implemented and benchmarked in core) before to
consider more the 64 byte transaction invalidity at the consensus level.

Best,

Antoine (the other one).

[Eric Voskuil]

Hello Antoine (other),

>  If you have code in pure C with variables on the stack no malloc, doing a check of the coinbase size after the socket read can be certainly more robust than checking a non-null pointer. 

Can you please clarify this for me? When you say "non-null pointer" do you mean C pointer or transaction input "null point" (sequence of 32 repeating 0x00 bytes and 4 0xff)? What do you mean by "more robust"?

Thanks,
Eric

[Antoine Riard]

Hi Eric,

I meant C pointer and by "more robust" any kind of memory / CPU DoS arising due to memory management (e.g. hypothetical rule checking the 64 bytes size for all block transactions).

In my understanding, the validation logic equivalent of core's CheckBlock is libbitcoin's block::check(): 

https://github.com/libbitcoin/libbitcoin-system/blob/master/src/chain/block.cpp#L751

Best,

Antoine

--
You received this message because you are subscribed to a topic in the Google Groups "Bitcoin Development Mailing List" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/bitcoindev/CAfm7D5ppjo/unsubscribe.
To unsubscribe from this group and all its topics, send an email to bitcoindev+...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/bitcoindev/26b7321b-cc64-44b9-bc95-a4d8feb701e5n%40googlegroups.com.

[Eric Voskuil]

> I meant C pointer and by "more robust" any kind of memory / CPU DoS arising due to memory management (e.g. hypothetical rule checking the 64 bytes size for all block transactions).

Ok, thanks for clarifying. I'm still not making the connection to "checking a non-null [C] pointer" but that's prob on me.

> In my understanding, the validation logic equivalent of core's CheckBlock is libbitcoin's block::check():
https://github.com/libbitcoin/libbitcoin-system/blob/master/src/chain/block.cpp#L751

Yes, a rough correlation but not necessarily equivalence. Note that block.check has context free and contextual overrides.

The 'bypass' parameter indicates a block under checkpoint or milestone ("assume valid"). In this case we must check Merkle root, witness commitment, and both types of malleation - as the purpose is to establish identity. Absent 'bypass' the typical checks are performed, and therefore a malleation check is not required here. The "type64" malleation is subsumed by the is_first_non_coinbase check and the "type32" malleation is subsumed by the is_internal_double_spend check.

I have some other thoughts on this that I'll post separately.

Best,
Eric

[Eric Voskuil]

Caching identity in the case of invalidity is more interesting question than it might seem.

Background: A fully-validated block has established identity in its block hash. However an invalid block message may include the same block header, producing the same hash, but with any kind of nonsense following the header. The purpose of the transaction and witness commitments is of course to establish this identity, so these two checks are therefore necessary even under checkpoint/milestone. And then of course the two Merkle tree issues complicate the tx commitment (the integrity of the witness commitment is assured by that of the tx commitment).

So what does it mean to speak of a block hash derived from:

(1) a block message with an unparseable header?
(2) a block message with parseable but invalid header?
(3) a block message with valid header but unparseable tx data?
(4) a block message with valid header but parseable invalid uncommitted tx data?
(5) a block message with valid header but parseable invalid malleated committed tx data?
(6) a block message with valid header but parseable invalid unmalleated committed tx data?
(7) a block message with valid header but uncommitted valid tx data?
(8) a block message with valid header but malleated committed valid tx data?
(9) a block message with valid header but unmalleated committed valid tx data?

Note that only the #9 p2p block message contains an actual Bitcoin block, the others are bogus messages. In all cases the message can be sha256 hashed to establish the identity of the *message*. And if one's objective is to reject repeating bogus messages, this might be a useful strategy. It's already part of the p2p protocol, is orders of magnitude cheaper to produce than a Merkle root, and has no identity issues.

The concept of Bitcoin block hash as unique identifier for invalid p2p block messages is problematic. Apart from the malleation question, what is the Bitcoin block hash for a message with unparseable data (#1 and #3)? Such messages are trivial to produce and have no block hash. What is the useful identifier for a block with malleated commitments (#5 and #8) or invalid commitments (#4 and #7) - valid txs or otherwise?

The stated objective for a consensus rule to invalidate all 64 byte txs is:

> being able to cache the hash of a (non-malleated) invalid block as permanently invalid to avoid re-downloading and re-validating it.

This seems reasonable at first glance, but given the list of scenarios above, which does it apply to? Presumably the invalid header (#2) doesn't get this far because of headers-first. That leaves just invalid blocks with useful block hash identifiers (#6). In all other cases the message is simply discarded. In this case the attempt is to move category #5 into category #6 by prohibiting 64 byte txs.

The requirement to "avoid re-downloading and re-validating it" is about performance, presumably minimizing initial block download/catch-up time. There is a computational cost to producing 64 byte malleations and none for any of the other bogus block message categories above, including the other form of malleation. Furthermore, 64 byte malleation has almost zero cost to preclude. No hashing and not even true header or tx parsing are required. Only a handful of bytes must be read from the raw message before it can be discarded presently.

That's actually far cheaper than any of the other scenarios that again, have no cost to produce. The other type of malleation requires parsing all of the txs in the block and hashing and comparing some or all of them. In other words, if there is an attack scenario, that must be addressed before this can be meaningful. In fact all of the other bogus message scenarios (with tx data) will remain more expensive to discard than this one.

The problem arises from trying to optimize dismissal by storing an identifier. Just *producing* the identifier is orders of magnitude more costly than simply dismissing this bogus message. I can't imagine why any implementation would want to compute and store and retrieve and recompute and compare hashes when the alterative is just dismissing the bogus messages with no hashing at all.

Bogus messages will arrive, they do not even have to be requested. The simplest are dealt with by parse failure. What defines a parse is entirely subjective. Generally it's "structural" but nothing precludes incorporating a requirement for a necessary leading pattern in the stream, sort of like how the witness pattern is identified. If we were going to prioritize early dismissal this is where we would put it.

However, there is a tradeoff in terms of early dismissal. Looking up invalid hashes is a costly tradeoff, which becomes multiplied by every block validated. For example, expending 1 millisecond in hash/lookup to save 1 second of validation time in the failure case seems like a reasonable tradeoff, until you multiply across the whole chain. 1 ms becomes 14 minutes across the chain, just to save a second for each mallied block encountered. That means you need to have encountered 840 such mallied blocks just to break even. Early dismissing the block for non-null coinbase point (without hashing anything) would be on the order of 1000x faster than that (breakeven at 1 encounter). So why the block hash cache requirement? It cannot be applied to many scenarios, and cannot be optimal in this one.

Eric

[Antoine Riard]

Hi Eric,

> Ok, thanks for clarifying. I'm still not making the connection to "checking a non-null [C] pointer" but that's prob on me.

A C pointer, which is a language idiome assigning to a memory address A the value o memory address B can be 0 (or NULL a standard macro defined in stddef.h).

Here a snippet example of linked list code checking the pointer (`*begin_list`) is non null before the comparison operation to find the target element list.

```
pointer_t       ft_list_find(pointer_t **start_list, void *data_ref, int (*cmp)())
{
        while (*start_list)
        {
                if (cmp((*start_list)->data, data_ref) == 0)
                        return (*start_list);
                *start_list = (*start_list)->next;
        }
        return (0);
}

```

While both libbitcoin and bitcoin core are both written in c++, you still have underlying pointer derefencing playing out to access the coinbase
transaction, and all underlying implications in terms of memory management.

> Yes, a rough correlation but not necessarily equivalence. Note that block.check has context free and contextual overrides.
>
> The 'bypass' parameter indicates a block under checkpoint or milestone ("assume valid"). In this case we must check Merkle root, witness commitment, and both types of malleation - as the purpose is to establish identity. Absent 'bypass' the typical checks are performed, and therefore a malleation check is not required here. The "type64" malleation is subsumed by the is_first_non_coinbase check and the "type32" malleation is subsumed by the is_internal_double_spend check.

Yes, I understand it's not a 1-to-1 compatibility, just a rough logical equivalence.

I think it's interesting to point out the two types of malleation that a bitcoin consensus validation logic should respect w.r.t block validity checks.

Like you said the first one on the merkle root committed in the headers's `hashMerkleRoot` due to the lack of domain separation between leaf and merkle tree nodes.
The second one is the bip141 wtxid commitment in one of the coinbase transaction `scriptpubkey` output, which is itself covered by a txid in the merkle tree.

> Caching identity in the case of invalidity is more interesting question than it might seem.
>
> Background: A fully-validated block has established identity in its block hash. However an invalid block message may include the same block header, producing the same hash, but with any kind of nonsense following the header. The purpose of the transaction and witness commitments is of course to establish this identity, so these two checks are therefore necessary even under checkpoint/milestone. And then of course the two Merkle tree issues complicate the tx commitment (the integrity of the witness commitment is assured by that of the tx commitment).
>
> So what does it mean to speak of a block hash derived from:
>
> (1) a block message with an unparseable header?
> (2) a block message with parseable but invalid header?
> (3) a block message with valid header but unparseable tx data?
> (4) a block message with valid header but parseable invalid uncommitted tx data?
> (5) a block message with valid header but parseable invalid malleated committed tx data?
> (6) a block message with valid header but parseable invalid unmalleated committed tx data?
> (7) a block message with valid header but uncommitted valid tx data?
> (8) a block message with valid header but malleated committed valid tx data?
> (9) a block message with valid header but unmalleated committed valid tx data?
>
> Note that only the #9 p2p block message contains an actual Bitcoin block, the others are bogus messages. In all cases the message can be sha256 hashed to establish the identity of the *message*. And if one's objective is to reject repeating bogus messages, this might be a useful strategy. It's already part of the p2p protocol, is orders of magnitude cheaper to produce than a Merkle root, and has no identity issues.

I think I mostly agree with the identity issue as laid out so far, there is one caveat to add if you're considering identity caching as the problem solved.
A validation node might have to consider differently block messages processed if they connect on the longest most PoW valid chain for which all blocks have been validated. Or alternatively if they have to be added on a candidate longest most PoW valid chain.

> The concept of Bitcoin block hash as unique identifier for invalid p2p block messages is problematic. Apart from the malleation question, what is the Bitcoin block
> hash for a message with unparseable data (#1 and #3)? Such messages are trivial to produce and have no block hash.

For reasons, bitcoin core has the concept of outbound `BLOCK_RELAY` (in `src/node/connection_types.h`) where some preferential peering policy is applied in matters of block messages download.

> What is the useful identifier for a block with malleated commitments (#5 and #8) or invalid commitments (#4 and #7) - valid txs or otherwise?

The block header, as it commits to the transaction identifier tree can be useful as much for #4 and #5. On the bitcoin core side, about #7 the uncommitted valid tx data can be already present in the validation cache from mempool acceptance. About #8, the malleaed committed valid transactions shall be also committed in the merkle root in headers.

> This seems reasonable at first glance, but given the list of scenarios above, which does it apply to?

> This seems reasonable at first glance, but given the list of scenarios above, which does it apply to? Presumably the invalid header (#2) doesn't get this far because of headers-first.
> That leaves just invalid blocks with useful block hash identifiers (#6). In all other cases the message is simply discarded. In this case the attempt is to move category #5 into category #6 by prohibiting 64 byte txs.

Yes, it's moving from the category #5 to the category #6. Note, transaction malleability can be a distinct issue than lack of domain separation.

> The requirement to "avoid re-downloading and re-validating it" is about performance, presumably minimizing initial block download/catch-up time. There is a > computational cost to producing 64 byte malleations and none for any of the other bogus block message categories above, including the other form of malleation. > Furthermore, 64 byte malleation has almost zero cost to preclude. No hashing and not even true header or tx parsing are required. Only a handful of bytes must be read > from the raw message before it can be discarded presently.

> That's actually far cheaper than any of the other scenarios that again, have no cost to produce. The other type of malleation requires parsing all of the txs in the block and > hashing and comparing some or all of them. In other words, if there is an attack scenario, that must be addressed before this can be meaningful. In fact all of the other

> bogus message scenarios (with tx data) will remain more expensive to discard than this one.

In practice on the bitcoin core side, the bogus block message categories from #4 to #6 are already mitigated by validation caching for transactions that have been received early. While libbitcoin has no mempool (at least in earlier versions) transactions buffering can be done by bip152's HeadersAndShortIds message.

About #7 and #8, introducing a domain separation where 64 bytes transactions are rejected and making it harder to exploit #7 and #8 categories of bogus block messages.

This is correct that bitcoin core might accept valid transaction data before the merkle tree commitment has been verified.

> The problem arises from trying to optimize dismissal by storing an identifier. Just *producing* the identifier is orders of magnitude more costly than simply dismissing this > bogus message. I can't imagine why any implementation would want to compute and store and retrieve and recompute and compare hashes when the alterative is just

> dismissing the bogus messages with no hashing at all.

> Bogus messages will arrive, they do not even have to be requested. The simplest are dealt with by parse failure. What defines a parse is entirely subjective. Generally it's

> "structural" but nothing precludes incorporating a requirement for a necessary leading pattern in the stream, sort of like how the witness pattern is identified. If we were

> going to prioritize early dismissal this is where we would put it.

I don't think this is that simple - While producing an identifier comes with a computational cost (e.g fixed 64-byte structured coinbase transaction), if the full node have a hierarchy of validation cache like bitcoin core has already, the cost of bogus block messages can be slashed down. On the other hand, just dealing with parse failure on the spot by introducing a leading pattern in the stream just inflates the size of p2p messages, and the transaction-relay bandwidth cost.

> However, there is a tradeoff in terms of early dismissal. Looking up invalid hashes is a costly tradeoff, which becomes multiplied by every block validated. For example,

> expending 1 millisecond in hash/lookup to save 1 second of validation time in the failure case seems like a reasonable tradeoff, until you multiply across the whole chain. > 1 ms becomes 14 minutes across the chain, just to save a second for each mallied block encountered. That means you need to have encountered 840 such mallied blocks > just to break even. Early dismissing the block for non-null coinbase point (without hashing anything) would be on the order of 1000x faster than that (breakeven at 1 > encounter). So why the block hash cache requirement? It cannot be applied to many scenarios, and cannot be optimal in this one.

I think what you're describing is more a classic time-space tradeoff which is well-known in classic computer science litterature. In my reasonable opinion, one should more reason under what is the security paradigm we wish for bitcoin block-relay network and perduring decentralization, i.e one where it's easy to verify block messages proofs which could have been generated on specialized hardware with an asymmetric cost. Obviously encountering 840 such malliead blocks to make it break even doesn't make the math up to save on hash lookup, unless you can reduce the attack scenario in terms of adversaries capabilities.

Best,

Antoine 

[Antoine Poinsot]

>> This does not produce unmalleable block hashes. Duplicate tx hash malleation remains in either case, to the same effect. Without a resolution to both issues this is an empty promise.

> Duplicate txids have been invalid since 2012 (CVE-2012-2459).

I think again here you may have misunderstood me. I was not making a point pertaining to BIP30.

No, in fact you did. CVE-2012-2459 is unrelated to BIP30, it's the duplicate txids malleability found by forrestv in 2012. It's the one you are talking about thereafter and the one relevant for the purpose of this discussion.

For future reference, the full disclosure of CVE-2012-2459 can be found there: https://bitcointalk.org/?topic=102395.

The proposal does not enable that objective, it is already the case. No malleated block is a valid block.

You are right. The advantage i initially mentioned about how making 64-bytes transactions invalid could help caching block failures at an earlier stage is incorrect.

Best,
Antoine Poinsot

--

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+...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/bitcoindev/9a4c4151-36ed-425a-a535-aa2837919a04n%40googlegroups.com.

[Eric Voskuil]

>>>> This does not produce unmalleable block hashes. Duplicate tx hash malleation remains in either case, to the same effect. Without a resolution to both issues this is an empty promise.

>>> Duplicate txids have been invalid since 2012 (CVE-2012-2459).

>> I think again here you may have misunderstood me. I was not making a point pertaining to BIP30.

> No, in fact you did. CVE-2012-2459 is unrelated to BIP30, it's the duplicate txids malleability found by forrestv in 2012. It's the one you are talking about thereafter and the one relevant for the purpose of this discussion.

Yes, my mistake. I didn't look up the CVE because malleability has no affect on consensus rules (validity). Without BIP30/34/90 a duplicated tx/txid (in a given chain) would still be valid (and under the caveats previously mentioned, still is). So I assumed you were referring to it/them. Malleability pertains strictly to validation implementation shortcuts (checkpoints, milestones, invalidity caching), not what is actually valid.

>> The proposal does not enable that objective, it is already the case. No malleated block is a valid block.

> You are right. The advantage i initially mentioned about how making 64-bytes transactions invalid could help caching block failures at an earlier stage is incorrect.

Hopefully the discussion leads to simpler and more performant implementation. As I mentioned previously, the usefulness (i.e. performance improving outcome) of block hash invalidity caching is very limited.

Libbitcoin implements an append-only store. And we write a checkpointed, milestoned, or current/strong header chains before obtaining blocks. So in the case where an invalid block corresponds to a stored header we must store the header's invalidity. Obviously this is guarded by PoW and therefore extremely rare, but must be accounted for. Otherwise we do not under any circumstances store invalidity. This is far more effective than storing it, even under heavy/constant "attack".

Given the PoW guard, the worst case scenario is where the witness commitment is invalid (it is performed after tx commitment, because it relies on the coinbase tx commit). Next worse is where the tx commitment is invalid. Neither present any cost to the attacker and neither rely on Merkle tree malleability. The latter requires hashing every tx and performing the Merkle root calculation. The former requires doing this twice. For a block with 4096 txs, that's [2 * (4096 + 4095) = 16382] tx hashes.

While that's nothing to sneeze at, in our implementation this constitutes 1-2% of total sync time on my 7 year old machine (no shani and no avx512). But what if we were to cache every invalid hash? Let's say we're under constant attack (despite dropping any peer that provides an invalid/unrequested block/message). The smart attacker doesn't use malleation, since he knows this is mitigated and cheaper in both cases to guard against. He just sends block messages with requested headers and a maximal set of valid txs (maybe from that actual block) and modifies one byte of any witness (or of any script for non-witness blocks). Every time sending a unique block, of which he can produce an effectively unlimited quantity. With or without caching this requires computation of all 16382 hashes for each bogus block that includes a requested header (unrequested are dismissed at the cost of just one hash).

In this case there is never a cache hit. Each bogus block is unique, but "valid enough" to force full double Merkle root computations. Storing the cached invalid hash then absorbs additional time and 32 bytes of space plus indexation, and achieves nothing. It's as if the hope is that the attacker is dumb and just keeps sending the same invalid block. But what's actually happening as (1) deoptimization, (2) unnecessary complexity, and (3) exposure to a disk-full attack vector which must then also be mitigated.

The other scenarios where parse fails cannot rely on invalidity caching, since they don't produce valid commitments, and are dismissed cheaply. That leaves only malleability. This comes in two forms, the 64 byte form ("type64") and what we call "type32" (hashes are 32 bytes and in this form they are duplicated). Type64 malleation is the cheapest form of dismissal, very early in parse (as discussed). Type32 malleation is far more expensive, but no more so than the worst case scenario above. In the Core implementation this detection adds a constant (and unnecessarily high) cost to the Merkle root computation. This makes it *more* expensive to detect than the worst case non-witness scenario above (and its discovery cannot be cached). It is possible to reduce this cost significantly by relying on some simple math operating over the tx count. So even this scenario is not inherently worst case.

So unless one is caching invalidity under PoW and due to an append-only store, I can see no reason to ever do it. Getting rid of it would improve both performance and security while reducing complexity. Optimally dismissing both types of malleation as described would improve performance, but is neutral regarding security.

e

[Larry Ruane]

On Monday, July 1, 2024 at 11:03:33 PM UTC-6 Antoine Riard wrote:

Here a snippet example of linked list code checking the pointer (`*begin_list`) is non null before the comparison operation to find the target element list.

```
pointer_t       ft_list_find(pointer_t **start_list, void *data_ref, int (*cmp)())
{
        while (*start_list)
        {
                if (cmp((*start_list)->data, data_ref) == 0)
                        return (*start_list);
                *start_list = (*start_list)->next;
        }
        return (0);
}

```

I assume this function lets you search for an element starting in the middle of a single-linked list (the middle because you could call `ft_list_find(&p-next, data_ref)` where `p` points to any element in the middle of the list, including possibly the last item in the list, in which case the loop body wouldn't run). If so, I don't think this does what's intended. This actually unlinks (and memory-leaks) elements up to where the match is found. I think you want to advance `start_list` this way (I didn't test this):

```

    start_list = &(*start_list)->next;

```

Larry

[Eric Voskuil]

Hi Antoine R,

>> Ok, thanks for clarifying. I'm still not making the connection to "checking a non-null [C] pointer" but that's prob on me.

> A C pointer, which is a language idiome assigning to a memory address A the value o memory address B can be 0 (or NULL a standard macro defined in stddef.h).
> Here a snippet example of linked list code checking the pointer (`*begin_list`) is non null before the comparison operation to find the target element list.

> ...

> While both libbitcoin and bitcoin core are both written in c++, you still have underlying pointer derefencing playing out to access the coinbase transaction, and all underlying implications in terms of memory management.

I'm familiar with pointers ;).

While at some level the block message buffer would generally be referenced by one or more C pointers, the difference between a valid coinbase input (i.e. with a "null point") and any other input, is not nullptr vs. !nullptr. A "null point" is a 36 byte value, 32 0x00 byes followed by 4 0xff bytes. In his infinite wisdom Satoshi decided it was better (or easier) to serialize a first block tx (coinbase) with an input containing an unusable script and pointing to an invalid [tx:index] tuple (input point) as opposed to just not having any input. That invalid input point is called a "null point", and of course cannot be pointed to by a "null pointer". The coinbase must be identified by comparing those 36 bytes to the well-known null point value (and if this does not match the Merkle hash cannot have been type64 malleated).

> I think it's interesting to point out the two types of malleation that a bitcoin consensus validation logic should respect w.r.t block validity checks. Like you said the first one on the merkle root committed in the headers's `hashMerkleRoot` due to the lack of domain separation between leaf and merkle tree nodes.

We call this type64 malleability (or malleation where it is not only possible but occurs).

> The second one is the bip141 wtxid commitment in one of the coinbase transaction `scriptpubkey` output, which is itself covered by a txid in the merkle tree.

While symmetry seems to imply that the witness commitment would be malleable, just as the txs commitment, this is not the case. If the tx commitment is correct it is computationally infeasible for the witness commitment to be malleated, as the witness commitment incorporates each full tx (with witness, sentinel, and marker). As such the block identifier, which relies only on the header and tx commitment, is a sufficient identifier. Yet it remains necessary to validate the witness commitment to ensure that the correct witness data has been provided in the block message.

The second type of malleability, in addition to type64, is what we call type32. This is the consequence of duplicated trailing sets of txs (and therefore tx hashes) in a block message. This is applicable to some but not all blocks, as a function of the number of txs contained.

>> Caching identity in the case of invalidity is more interesting question than it might seem.
>> Background: A fully-validated block has established identity in its block hash. However an invalid block message may include the same block header, producing the same hash, but with any kind of nonsense following the header. The purpose of the transaction and witness commitments is of course to establish this identity, so these two checks are therefore necessary even under checkpoint/milestone. And then of course the two Merkle tree issues complicate the tx commitment (the integrity of the witness commitment is assured by that of the tx commitment).
>>
>> So what does it mean to speak of a block hash derived from:
>> (1) a block message with an unparseable header?
>> (2) a block message with parseable but invalid header?
>> (3) a block message with valid header but unparseable tx data?
>> (4) a block message with valid header but parseable invalid uncommitted tx data?
>> (5) a block message with valid header but parseable invalid malleated committed tx data?
>> (6) a block message with valid header but parseable invalid unmalleated committed tx data?
>> (7) a block message with valid header but uncommitted valid tx data?
>> (8) a block message with valid header but malleated committed valid tx data?
>> (9) a block message with valid header but unmalleated committed valid tx data?
>>
>> Note that only the #9 p2p block message contains an actual Bitcoin block, the others are bogus messages. In all cases the message can be sha256 hashed to establish the identity of the *message*. And if one's objective is to reject repeating bogus messages, this might be a useful strategy. It's already part of the p2p protocol, is orders of magnitude cheaper to produce than a Merkle root, and has no identity issues.

> I think I mostly agree with the identity issue as laid out so far, there is one caveat to add if you're considering identity caching as the problem solved. A validation node might have to consider differently block messages processed if they connect on the longest most PoW valid chain for which all blocks have been validated. Or alternatively if they have to be added on a candidate longest most PoW valid chain.

Certainly an important consideration. We store both types. Once there is a stronger candidate header chain we store the headers and proceed to obtaining the blocks (if we don't already have them). The blocks are stored in the same table; the confirmed vs. candidate indexes simply point to them as applicable. It is feasible (and has happened twice) for two blocks to share the very same coinbase tx, even with either/all bip30/34/90 active (and setting aside future issues here for the sake of simplicity). This remains only because two competing branches can have blocks at the same height, and bip34 requires only height in the coinbase input script. This therefore implies the same transaction but distinct blocks. It is however infeasible for one block to exist in multiple distinct chains. In order for this to happen two blocks at the same height must have the same coinbase (ok), and also the same parent (ok). But this then means that they either (1) have distinct identity due to another header property deviation, or (2) are the same block with the same parent and are therefore in just one chain. So I don't see an actual caveat. I'm not certain if this is the ambiguity that you were referring to. If not please feel free to clarify.

>> The concept of Bitcoin block hash as unique identifier for invalid p2p block messages is problematic. Apart from the malleation question, what is the Bitcoin block hash for a message with unparseable data (#1 and #3)? Such messages are trivial to produce and have no block hash.

> For reasons, bitcoin core has the concept of outbound `BLOCK_RELAY` (in `src/node/connection_types.h`) where some preferential peering policy is applied in matters of block messages download.

We don't do this and I don't see how it would be relevant. If a peer provides any invalid message or otherwise violates the protocol it is simply dropped.

The "problematic" that I'm referring to is the reliance on the block hash as a message identifier, because it does not identify the message and cannot be useful in an effectively unlimited number of zero-cost cases.

>> What is the useful identifier for a block with malleated commitments (#5 and #8) or invalid commitments (#4 and #7) - valid txs or otherwise?

> The block header, as it commits to the transaction identifier tree can be useful as much for #4 and #5.

#4 and #5 refer to "uncommitted" and "malleated committed". It may not be clear, but "uncommitted" means that the tx commitment is not valid (Merkle root doesn't match the header's value) and "malleated committed" means that the (matching) commitment cannot be relied upon because the txs represent malleation, invalidating the identifier. So neither of these are usable identifiers.

> On the bitcoin core side, about #7 the uncommitted valid tx data can be already present in the validation cache from mempool acceptance. About #8, the malleaed committed valid transactions shall be also committed in the merkle root in headers.

It seems you may be referring to "unconfirmed" txs as opposed to "uncommitted" txs. This doesn't pertain to tx storage or identifiers. Neither #7 nor #8 are usable for the same reasons.

>> This seems reasonable at first glance, but given the list of scenarios above, which does it apply to?

>> This seems reasonable at first glance, but given the list of scenarios above, which does it apply to? Presumably the invalid header (#2) doesn't get this far because of headers-first.
>> That leaves just invalid blocks with useful block hash identifiers (#6). In all other cases the message is simply discarded. In this case the attempt is to move category #5 into category #6 by prohibiting 64 byte txs.

> Yes, it's moving from the category #5 to the category #6. Note, transaction malleability can be a distinct issue than lack of domain separation.

I'm making no reference to tx malleability. This concerns only Merkle tree (block hash) malleability, the two types described in detail in the paper I referenced earlier, here again:

https://lists.linuxfoundation.org/pipermail/bitcoin-dev/attachments/20190225/a27d8837/attachment-0001.pdf

>> The requirement to "avoid re-downloading and re-validating it" is about performance, presumably minimizing initial block download/catch-up time. There is a > computational cost to producing 64 byte malleations and none for any of the other bogus block message categories above, including the other form of malleation. > Furthermore, 64 byte malleation has almost zero cost to preclude. No hashing and not even true header or tx parsing are required. Only a handful of bytes must be read > from the raw message before it can be discarded presently.

>> That's actually far cheaper than any of the other scenarios that again, have no cost to produce. The other type of malleation requires parsing all of the txs in the block and > hashing and comparing some or all of them. In other words, if there is an attack scenario, that must be addressed before this can be meaningful. In fact all of the other bogus message scenarios (with tx data) will remain more expensive to discard than this one.

> In practice on the bitcoin core side, the bogus block message categories from #4 to #6 are already mitigated by validation caching for transactions that have been received early. While libbitcoin has no mempool (at least in earlier versions) transactions buffering can be done by bip152's HeadersAndShortIds message.

Again, this has no relation to tx hashes/identifiers. Libbitcoin has a tx pool, we just don't store them in RAM (memory).

> About #7 and #8, introducing a domain separation where 64 bytes transactions are rejected and making it harder to exploit #7 and #8 categories of bogus block messages. This is correct that bitcoin core might accept valid transaction data before the merkle tree commitment has been verified.

I don't follow this. An invalid 64 byte tx consensus rule would definitely not make it harder to exploit block message invalidity. In fact it would just slow down validation by adding a redundant rule. Furthermore, as I have detailed in a previous message, caching invalidity does absolutely nothing to increase protection. In fact it makes the situation materially worse.

>> The problem arises from trying to optimize dismissal by storing an identifier. Just *producing* the identifier is orders of magnitude more costly than simply dismissing this > bogus message. I can't imagine why any implementation would want to compute and store and retrieve and recompute and compare hashes when the alterative is just dismissing the bogus messages with no hashing at all.

>> Bogus messages will arrive, they do not even have to be requested. The simplest are dealt with by parse failure. What defines a parse is entirely subjective. Generally it's
>> "structural" but nothing precludes incorporating a requirement for a necessary leading pattern in the stream, sort of like how the witness pattern is identified. If we were
>> going to prioritize early dismissal this is where we would put it.

> I don't think this is that simple - While producing an identifier comes with a computational cost (e.g fixed 64-byte structured coinbase transaction), if the full node have a hierarchy of validation cache like bitcoin core has already, the cost of bogus block messages can be slashed down.

No, this is not the case. As I detailed in my previous message, there is no possible scenario where invalidation caching does anything but make the situation materially worse.

> On the other hand, just dealing with parse failure on the spot by introducing a leading pattern in the stream just inflates the size of p2p messages, and the transaction-relay bandwidth cost.

I think you misunderstood me. I am suggesting no change to serialization. I can see how it might be unclear, but I said, "nothing precludes incorporating a requirement for a necessary leading pattern in the stream." I meant that the parser can simply incorporate the *requirement* that the byte stream starts with a null input point. That identifies the malleation or invalidity without a single hash operation and while only reading a handful of bytes. No change to any messages.

>> However, there is a tradeoff in terms of early dismissal. Looking up invalid hashes is a costly tradeoff, which becomes multiplied by every block validated. For example, expending 1 millisecond in hash/lookup to save 1 second of validation time in the failure case seems like a reasonable tradeoff, until you multiply across the whole chain. > 1 ms becomes 14 minutes across the chain, just to save a second for each mallied block encountered. That means you need to have encountered 840 such mallied blocks > just to break even. Early dismissing the block for non-null coinbase point (without hashing anything) would be on the order of 1000x faster than that (breakeven at 1 > encounter). So why the block hash cache requirement? It cannot be applied to many scenarios, and cannot be optimal in this one.

> I think what you're describing is more a classic time-space tradeoff which is well-known in classic computer science litterature. In my reasonable opinion, one should more reason under what is the security paradigm we wish for bitcoin block-relay network and perduring decentralization, i.e one where it's easy to verify block messages proofs which could have been generated on specialized hardware with an asymmetric cost. Obviously encountering 840 such malliead blocks to make it break even doesn't make the math up to save on hash lookup, unless you can reduce the attack scenario in terms of adversaries capabilities.

I'm referring to DoS mitigation (the only relevant security consideration here). I'm pointing out that invalidity caching is pointless in all cases, and in this case is the most pointless as type64 malleation is the cheapest of all invalidity to detect. I would prefer that all bogus blocks sent to my node are of this type. The worst types of invalidity detection have no mitigation and from a security standpoint are counterproductive to cache. I'm describing what overall is actually not a tradeoff. It's all negative and no positive.

Best,
Eric

[Eric Voskuil]

This is why we don't use C - unsafe, unclear, unnecessary.

e

[Antoine Riard]

> I assume this function lets you search for an element starting in the middle of a single-linked list (the middle because you could call `ft_list_find(&p-next, data_ref)` where `p` points to any element in the

> middle of the list, including possibly the last item in the list, in which case the loop body wouldn't run). If so, I don't think this does what's intended. This actually unlinks (and memory-leaks) elements up to > where the match is found. I think you want to advance `start_list` this way (I didn't test this):

Note the usage of a pointer to pointer so the correct way to call the code is : `pointer_t * list_ptr ; list_ptr = first_list_element ; ft_list_find(list_ptr, data_rf, cmp);`.

This is correct that if you point to the last item in the list, the loop body wouldn't run (which is the expected behavior). When there is a match, the pointer `*start_list` takes as value the memory address of the next element in the list, the contained structure pointer is not changed. The code has been tested a while back, though it's indeed clearer if a typedef `pointer_t` for list is fully given: 

```

typedef struct                 s_list

{

                         void     *content;

                         size_t  content_size;

                         struct  s_list. *next;

}                                      pointer_t

```

> This is why we don't use C - unsafe, unclear, unnecessary.

Actually, I think libbitcoin is using its own maintained fork of secp256k1, which is written in C.

For sure, I wouldn't recommend using C across a whole codebase as it's not memory-safe (euphemism) though it's still un-match if you wish to understand low-level memory management in hot paths.

It can be easier to use C++ or Rust, though it doesn't mean it will be as (a) perf optimal and (b) hardened against side-channels.

I have not read in detail the last Eric's email on the whole caching identity in case of invalidity discussion, though I'll do so.

Best,

Antoine

Le jeu. 4 juil. 2024 à 00:57, Eric Voskuil <er...@voskuil.org> a écrit :

This is why we don't use C - unsafe, unclear, unnecessary.

e

--

You received this message because you are subscribed to a topic in the Google Groups "Bitcoin Development Mailing List" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/bitcoindev/CAfm7D5ppjo/unsubscribe.
To unsubscribe from this group and all its topics, send an email to bitcoindev+...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/bitcoindev/33dfd007-ac28-44a5-acee-cec4b381e854n%40googlegroups.com.

[Eric Voskuil]

> This is why we don't use C - unsafe, unclear, unnecessary.

Actually, I think libbitcoin is using its own maintained fork of secp256k1, which is written in C.

We do not maintain secp256k1 code. For years that library carried the same version, despite regular breaking changes to its API. This compelled us to place these different versions on distinct git branches. When it finally became versioned we started phasing this unfortunate practice out.

Out of the 10 repositories and at least half million lines of code, apart from an embedded copy of qrencode that we don’t independently maintain, I believe there is only one .c file in use in the entire project - the database mmap.c implementation for msvc builds. This includes hash functions, with vectorization optimizations, etc.

 

For sure, I wouldn't recommend using C across a whole codebase as it's not memory-safe (euphemism) though it's still un-match if you wish to understand low-level memory management in hot paths.

This is a commonly held misperception.

It can be easier to use C++ or Rust, though it doesn't mean it will be as (a) perf optimal and (b) hardened against side-channels.

Rust has its own set of problems. No need to get into a language Jihad here. My point was to clarify that the particular question was not about a C (or C++) null pointer value, either on the surface or underneath an abstraction.

e