Make provably unsignable standard P2PK and P2MS outpoints unspendable.

russeree commented at 8:04 am on September 4, 2023: contributor

Overview

This pr introduces additional conditionals into IsUnspendable() to remove provably unspendable P2PK and P2MS tx outpoints from the UTXO set. This is done by using IsFullyValid() to check that the public key(s) for the outpoints script are valid. This trims nearly 20K outpoints from the UTXO set at height 805618. https://gist.github.com/russeree/85fb9519e0a1fc166177a6d5e0e15be2

A side effect of this PR is it removes the use case for uncompressed public keys through standard tx types to store arbitrary data in the UTXO set.

P2PK

P2PK outpoints with a single pubkey that is invalid can by flagged as unspendable becuase the public key does not exist on the SECP256K1 curve and thus no private key exists to make OP_CHECKSIG evaluate to true.

Script must be in the format of OP_PUSHBBYTES PUBKEY OP_CHECKSIG(VERIFY)

P2MS

P2MS outpoints that do not have enough valid public keys to meet the threshold. where (n-k < m) where k is the number of invalid public keys.

Script must be in the format of OP_(N) PUBKEY1 .... OP_(N) OP_CHECKMULTISIG(VERIFY)

DrahtBot commented at 8:04 am on September 4, 2023: contributor

The following sections might be updated with supplementary metadata relevant to reviewers and maintainers.

Code Coverage

For detailed information about the code coverage, see the test coverage report.

Reviews

See the guideline for information on the review process.

Type	Reviewers
Concept NACK	petertodd
Concept ACK	RandyMcMillan, dzyphr, dexX7, NicolasDorier

If your review is incorrectly listed, please react with 👎 to this comment and the bot will ignore it on the next update.

Conflicts

Reviewers, this pull request conflicts with the following ones:

#28728 (wallet: [bugfix] Mark CNoDestination and PubKeyDestination constructor explicit by maflcko)
#28690 (build: Introduce internal kernel library by TheCharlatan)
#28550 (Covenant tools softfork by jamesob)

If you consider this pull request important, please also help to review the conflicting pull requests. Ideally, start with the one that should be merged first.

DrahtBot added the label CI failed on Sep 4, 2023

theStack commented at 10:46 am on September 4, 2023: contributor

Changing the IsUnspendable logic in general seems very brittle, as it obviously leads to nodes with diverging UTXO sets. Even if all nodes would run this patch, they’d start to run it at different block heights (if between blocks M and N an UTXO enters the set that classifies for pruning, you’d already diverge), and you’d only end up with the same fully pruned UTXO set if all of them also did a -reindex once.

When the pruning of OP_RETURN outputs was implemented over 10 years ago (see #2791) this was seemingly less of a problem, but now that many projects depend on a normalized view of the UTXO set (muhash, assumeutxo, utreexo etc.), I think that would lead to much unwanted chaos.

Happy to hear other inputs how this could be solved, maybe I’m missing something.

russeree force-pushed on Sep 4, 2023

in src/script/script.cpp:324 in 5892815ced outdated

323@@ -258,6 +324,44 @@ bool CScript::IsPushOnly() const
324     return this->IsPushOnly(begin());
325 }
326 
327+
328+bool CScript::IsUnspendable() const
329+{
330+    if (size() > MAX_SCRIPT_SIZE) {

rot13maxi commented at 11:59 am on September 4, 2023:

did you find anything interesting when you added this check?

russeree commented at 12:08 pm on September 4, 2023:

No as it was previously defined here.

https://github.com/bitcoin/bitcoin/blob/6f03c45f6bb5a6edaa3051968b6a1ca4f84d2ccb/src/script/script.h#L551-L554

The reason I placed it at the top of the if else stack was to avoid doing work on a script that already is invalid due to length. If the script is too long just mark as unspendable and avoid any extra checking

ajtowns commented at 12:29 pm on September 4, 2023: contributor

Changing the IsUnspendable logic in general seems very brittle, as it obviously leads to nodes with diverging UTXO sets.

now that many projects depend on a normalized view of the UTXO set (muhash, assumeutxo, utreexo etc.), I think that would lead to much unwanted chaos.

I think muhash provides a pretty straightforward way of dealing with this. Namely: rather than only maintaining a single muhash for the utxo set you have, also maintain one for the outputs you’ve pruned as unspendable (or multiple indexes if you introduce more over time). Then you might have:

at block height 810,000
the spendable utxo set has muhash X
the OP_RETURN utxos have muhash Y
the bad pubkey utxos have muhash Z

if you compare with a node running current master, you then verify that your X+Z matches their X.

It’s easy to initialize these values on upgrade: Z=0, then iterate through your utxo set looking for bad pubkeys, adding them to Z as you find them, and removing them from the utxo set. Downgrading breaks things though.

I don’t think you can bootstrap utreexo from just the utxo set at an arbitrary height in the first place, so I don’t think its really affected by this. For assumeutxo, if you want to be able to distribute the utxo set snapshots over p2p, then peers that have pruned unspendable utxos won’t have them to share with peers that don’t realise they’re unspendable in the first place, but if you’re just producing signed snapshots, then you could probably just publish two versions until the old software goes out of its support window.

in src/script/script.cpp:215 in 5892815ced outdated

210+        return false;
211+    }
212+
213+    // Last byte is equal to OP_CHECKSIG / OP_CHECKSIGVERIFY
214+    const unsigned char last_byte = *(this->end() - 1);
215+    if (!(last_byte != OP_CHECKSIG ^ last_byte != OP_CHECKSIGVERIFY)) {

ajtowns commented at 12:37 pm on September 4, 2023:

if (back() != OP_CHECKSIG && back() != OP_CHECKSIGVERIFY) ?

in src/script/script.cpp:226 in 5892815ced outdated

221+    if (first_byte == OP_0 || first_byte >= OP_PUSHDATA1) {
222+        return false;
223+    }
224+
225+    // The script contains 2 ops and is equal to the script length
226+    if (first_byte + 2 == size){

ajtowns commented at 12:41 pm on September 4, 2023:

You’ve already checked size is between 2 and 77, so if first_byte + 2 == size, first_byte is between 0 and 75, so the check for OP_PUSHDATA1 is redundant (and the check for OP_0 could be subsumed if you checked size < 3).

Also, first_byte == front().

in src/script/script.cpp:327 in 5892815ced outdated

328+bool CScript::IsUnspendable() const
329+{
330+    if (size() > MAX_SCRIPT_SIZE) {
331+        return true;
332+    } else if (this->IsPayToPublicKey()) {
333+        std::vector<unsigned char> pubkey(this->begin() + 1, this->end() - 1);

ajtowns commented at 12:46 pm on September 4, 2023:

Should just be a span

theStack commented at 3:27 pm on September 4, 2023: contributor

@ajtowns: Neat idea with the additional MuHash! I agree this should work, but obviously at the cost of increased complexity (as there is much more to change than only the logic of a single method), and there should be a really good reason for doing it.

It’s easy to initialize these values on upgrade: Z=0, then iterate through your utxo set looking for bad pubkeys, adding them to Z as you find them, and removing them from the utxo set. Downgrading breaks things though.

Right, and you’d also need to somehow detect if an upgrade has already happened. In #2791 it was proposed to either scan for a particular output in the UTXO set (https://github.com/bitcoin/bitcoin/pull/2791#issuecomment-21256241) or to introduce a flag in the chainstate database (https://github.com/bitcoin/bitcoin/pull/2791#issuecomment-21292138). @russeree: Can you give some additional motivation? While I enjoy very much reasoning about these kind of topics, it’s still unclear to me what concrete problem this PR is trying to solve. The description claims that about 20k of outputs could be pruned. As of now (block 806205), that’s merely 0.016% of the total UTXO set size, freeing up about 1,24 MB of chainstate space (if we assume an average size of 65 bytes per UTXO). I’d argue that these numbers are way too low (even if you 10x them for the sake of projecting into the future) to justify the increased complexity in different areas that would need to be touched (i.e. differentiating between different IsUnspendable reasons and update all call-sites correctly, maintaining an extra MuHash, implementing an UTXO set upgrade mechanism, probably adapting gettxoutsetinfo RPC results to include both “new-pruned-MuHash” and “old-unpruned-MuHash”, dealing with different dumptxoutset/loadtxoutset result/behaviour on old vs new nodes for AssumeUTXO etc. etc.).

Also, calling IsFullyValid is significantly more expensive than just doing the usual Solver matching and length checks, so one would also need to analyze if this has a noticable negative performance impact on standardness checks.

in src/script/script.cpp:332 in 5892815ced outdated

333+        std::vector<unsigned char> pubkey(this->begin() + 1, this->end() - 1);
334+        if(!CPubKey(pubkey).IsFullyValid()){
335+            return true;
336+        }
337+    } else if (this->IsPayToMultisig()) {
338+        CScript::const_iterator pc = this->begin() + 1;

ajtowns commented at 10:58 pm on September 4, 2023:

Seems like it would be better to have this->IsInvalidMultisig() and move this code all into one place? Something like:

 0if (size() < 3) return false;
 1if (back() != OP_CHECKMULTISIG && back() != OP_CHECKMULTISIGVERIFY) return false;
 2int n;
 3if (!IsOpN(*(end() - 2), n ) return false;
 4int m;
 5if (!IsOpN(front(), m) return false;
 6std::vector<Span<unsigned char>> keys;
 7keys.resize(n);
 8// parse into keys, finish checking it's well-formed
 9if (m > n) return true;
10int good = 0;
11int max_good = n;
12for (k : keys) {
13    if (CPubKey(k).IsFullyValid()) {
14        ++good;
15    } else {
16        --max_good;
17    }
18    if (good >= m) return false;
19    if (max_good < m) return true;
20}
21return false; // unreachable

in src/script/script.cpp:205 in 5892815ced outdated

201@@ -201,6 +202,71 @@ unsigned int CScript::GetSigOpCount(const CScript& scriptSig) const
202     return subscript.GetSigOpCount(true);
203 }
204 
205+bool CScript::IsPayToPublicKey() const

Randy808 commented at 12:45 pm on September 5, 2023:

It might be simpler to follow the style of the existing script template checks like ‘IsPayToScriptHash’. You can also use the existing constants to check against the compressed and uncompressed sizes

in src/script/script.cpp:445 in 5892815ced outdated

446@@ -343,6 +447,40 @@ bool IsOpSuccess(const opcodetype& opcode)
447            (opcode >= 187 && opcode <= 254);
448 }
449 
450+bool IsOpN(unsigned char op_code){
451+    return OP_0 || (OP_1 <= op_code && op_code <= OP_16);

Randy808 commented at 12:55 pm on September 5, 2023:

I think you meant: return op_code == OP_0 || (OP_1 <= op_code && op_code <= OP_16);

in src/script/script.cpp:448 in 5892815ced outdated

446@@ -343,6 +447,40 @@ bool IsOpSuccess(const opcodetype& opcode)
447            (opcode >= 187 && opcode <= 254);
448 }
449 
450+bool IsOpN(unsigned char op_code){
451+    return OP_0 || (OP_1 <= op_code && op_code <= OP_16);
452+}
453+
454+bool IsOpN(unsigned char op_code, unsigned char& value){

Randy808 commented at 12:59 pm on September 5, 2023:

Might be more concise if you used the above function in this one:

 0 bool IsOpN(unsigned char op_code, unsigned char &value) {
 1   if (IsOpN(op_code)) {
 2     if (op_code == OP_0) {
 3       value = 0;
 4     } else {
 5       value = op_code - 0x50;
 6     }
 7     return true;
 8   }
 9   return false;
10 }

russeree commented at 1:48 am on September 11, 2023: contributor

Sorry for the lack of work on this PR last week, was at TABConf working on #27260 . Work has resumed and I will update this thread with changes over the next few days.

russeree force-pushed on Sep 12, 2023

Make unsignable standard P2PK and P2MS outpoints unspendable.

Author:    russeree <git@qrsnap.io>

@ajtowns - Remove OP_O case from IsPayToPublicKey()

d6841940c5

russeree force-pushed on Sep 12, 2023

russeree marked this as a draft on Sep 13, 2023

russeree force-pushed on Sep 17, 2023

p2pk unit tests 3733665990

russeree force-pushed on Sep 26, 2023

petertodd commented at 8:47 am on September 26, 2023: contributor

This trims nearly 20K outpoints from the UTXO set at height 805618.

Concept NACK.

At the moment the UTXO set has 130 million entries. Reducing it by 0.015% isn’t worth the technical risk.

By comparison, there have been 53 million OP_Return outputs. Removing those from the UTXO set was a not-so-trivial ~30% reduction.

DrahtBot added the label Needs rebase on Oct 26, 2023

DrahtBot commented at 4:42 pm on October 26, 2023: contributor

🐙 This pull request conflicts with the target branch and needs rebase.

RandyMcMillan commented at 7:55 pm on December 14, 2023: contributor

Concept ACK

BTCMcBoatface commented at 11:32 pm on January 2, 2024: none

I have no business adding comments here but I may humbly add that, despite this only affecting 0.016% of the UTXOs, it is a smart preemptive defense against arbitrary data on chain using a method that is far more damaging than inscriptions.

DrahtBot commented at 10:42 am on January 5, 2024: contributor

There hasn’t been much activity lately and the patch still needs rebase. What is the status here?

Is it still relevant? ➡️ Please solve the conflicts to make it ready for review and to ensure the CI passes.
Is it no longer relevant? ➡️ Please close.
Did the author lose interest or time to work on this? ➡️ Please close it and mark it ‘Up for grabs’ with the label, so that it can be picked up in the future.

owenstrevor commented at 7:12 pm on January 28, 2024: none

So this would make all current and future STAMPS unspendable? Or just future ones? How would this affect other potential legitimate use cases?

What is the projected growth of node requirements if this situation were to get worse and how does that track with Moore’s Law of the increase in computer performance? Or is there a better way to think about this?

It’s worth noting, making these unspendable may not stop their proliferation. For example, solutions that allow you to trade private keys off-chain, in the event that there is market demand for STAMPS, would still incentivize their creation to the same degree even if they are unspendable.

dzyphr commented at 11:12 pm on January 28, 2024: none

Concept ACK

Responses:

So this would make all current and future STAMPS unspendable? Or just future ones? How would this affect other potential legitimate use cases?

What is the projected growth of node requirements if this situation were to get worse and how does that track with Moore’s Law of the increase in computer performance? Or is there a better way to think about this?

It’s worth noting, making these unspendable may not stop their proliferation. For example, solutions that allow you to trade private keys off-chain, in the event that there is market demand for STAMPS, would still incentivize their creation to the same degree even if they are unspendable. @owenstrevor it does nothing of the sort. You should read more carefully before asserting a slippery slope economic argument about STAMPS.

to remove provably unspendable P2PK and P2MS tx outpoints from the UTXO set.

These outputs are already unspendable mathematically, and the person spending to them agreed to that condition upon signing.

This trims nearly 20K outpoints from the UTXO set at height 805618.

Concept NACK.

At the moment the UTXO set has 130 million entries. Reducing it by 0.015% isn’t worth the technical risk.

By comparison, there have been 53 million OP_Return outputs. Removing those from the UTXO set was a not-so-trivial ~30% reduction.

OP_RETURN outputs are not provably unspendable though, they have valid spending paths. So the comparison is irrelevant, the point is why would you incentivize the storage of something that is basically a burned output? Sure right now the storage space is negligible, however you would have to have a completely non-adversarial mindset to suggest that no one would ever exploit this further.

theStack commented at 2:51 am on January 29, 2024: contributor

Fun challenge for everyone: can you link to a specific bare multisig UTXO that has been created within the past year (let’s say, since block 769785) where the prunable detection in this PR would hit? Looking at an UTXO snapshot from now (block 827855), from all the 749496 P2MS UTXOs that have been created since the start of the year 2023, I haven’t managed to find a single one which is provably unspendable. // EDIT: nevermind, my script had a bug.

dzyphr commented at 3:06 am on January 29, 2024: none

Fun challenge for everyone: can you link to a specific bare multisig UTXO that has been created within the past year (let’s say, since block 769785) where the prunable detection in this PR would hit? Looking at an UTXO snapshot from now (block 827855), from all the 749496 P2MS UTXOs that have been created since the start of the year 2023, I haven’t managed to find a single one which is provably unspendable.

they have provided a list: https://gist.github.com/russeree/85fb9519e0a1fc166177a6d5e0e15be2

ajtowns commented at 7:22 am on January 29, 2024: contributor

Fun challenge for everyone: can you link to a specific bare multisig UTXO that has been created within the past year (let’s say, since block 769785) where the prunable detection in this PR would hit? Looking at an UTXO snapshot from now (block 827855), from all the 749496 P2MS UTXOs that have been created since the start of the year 2023, I haven’t managed to find a single one which is provably unspendable.

they have provided a list: https://gist.github.com/russeree/85fb9519e0a1fc166177a6d5e0e15be2

The most recently confirmed txs in that list are from block 792783 (June 2023), namely c18fe6…, 74c96e… and a6062e…. All but ~50 of the txs are over 300k blocks ago (ie 6+ years). The ones that aren’t are:

 0c18fe6a7e72d75def4dde08982eb217e6d940253c62e8edfd91d027e42ea305e
 174c96e6dcd6ceb02b37c82384a199a896acbddc67b26afb575f9ed94ac643a42
 2a6062e218659a9b08fd345116aaa0b70e1fdce9c44b394e8a983afb9c858f0a9
 344500c98c7c50a95490e9b949e29ce0748b6184f63035672fc5901094f6bb56c
 4d697f7081adecdf9459c336d8c6a5b0ae0397f84d24261d1e818a93157f5b206
 581f7dfeb5b9f32fab76bf473dcebe2b386aefbff441f0cd822d2c6b3bed31f2f
 63fdd82de9be8a8d85120035819bd8cfac04c455245d59f6966d9a21f2095c734
 74ce4431ac29acf2ec0c404d37798b8288a1f0684c34c98bf7d10932c0b0959cd
 8d53fc95fe481f7268ac9ee64ecbed99f0b4b9b6744af4f1fc8f1fca7940ed26c
 9cbc47c01f6a7446fba8dda5d644fe5921b706337be64a4375ede3c9db6f3dcff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ajtowns commented at 7:29 am on January 29, 2024: contributor

So this would make all current and future STAMPS unspendable? Or just future ones? How would this affect other potential legitimate use cases?

Most STAMPS txs are spendable and are not affected by this PR. As I understand it, that protocol sets up a 1-of-3 multisig, where of the keys is valid (providing a spendable path) and the other two are used for data, and are thus provably unspendable about 50% of the time as the data doesn’t match a valid secp point. The only times those txs become unspendable are when the one valid key is replaced by an invalid key – eg the txs linked in my previous comment have a pubkey 030303030303030303030303030303030303030303030303030303030303030303 which is neither a valid point nor carries any data.

dexX7 commented at 8:44 am on January 29, 2024: contributor

Concept ACK.

Note, however, that pubkeys can be modified to be valid and still be used to store data, for example by using a byte that is shuffled until the pubkey becomes valid.

NicolasDorier commented at 12:13 pm on January 29, 2024: contributor

Concept ACK.

However, note that from the time this PR run on a node, its UTXO Set hash will start to differ from any peer that didn’t start using this PR at the same time.

This can be solved by cleaning up the UTXO Set when the node starts. Unsure how expensive would it be.

It would be nice to know how much space is saved running this PR from genesis.

We are facing a problem in BTCPay: A large number of users have only 20GB of space to store the UTXO Set. Now that the UTXO Set is reaching 10GB and increasing, those users might have their node crashing sooner or later. So we are very interested into solutions to drop down the UTXO set size.

EDIT: It turns out that our problem wouldn’t be solved see #28400 (comment)

russeree commented at 12:47 pm on January 29, 2024: contributor

and are thus provably unspendable about 50% of the time as the data doesn’t match a valid secp point.

Sorry for not understanding this, but why 50%? I thought compressed pubkeys covered almost everything on the X axis because the y is derived? Also many of the pruned outpoints are not compressed pubkeys but instead uncompressed.

theStack commented at 5:11 pm on January 29, 2024: contributor

and are thus provably unspendable about 50% of the time as the data doesn’t match a valid secp point.

Sorry for not understanding this, but why 50%? I thought compressed pubkeys covered almost everything on the X axis because the y is derived? Also many of the pruned outpoints are not compressed pubkeys but instead uncompressed.

In my own non-cryptographer words: for about half of all possible field elements x, there exists no field element y such that the secp256k1 equation $y^2 = x^3 + 7$ holds (more precisely, you would get the coordinate y via $y = \sqrt{x^3 + 7}$, but for those x values, the expression $x^3 + 7$ doesn’t have a square root).

ajtowns commented at 5:12 pm on January 29, 2024: contributor

and are thus provably unspendable about 50% of the time as the data doesn’t match a valid secp point.

Sorry for not understanding this, but why 50%? I thought compressed pubkeys covered almost everything on the X axis because the y is derived? Also many of the pruned outpoints are not compressed pubkeys but instead uncompressed.

The formula is y^2 = x^3 + 7, so you derive x^3+7 for any x, but only half of those will be valid squares – every number 0 < i < p/2 squares to the same result as the different number p-i does, so you only use up half the possible numbers as squares; presuming x^3+7 is just randomly choosing a number between 1 and p, then half the time it won’t be a square. You get back to having ~2^256 points despite half of the ~2^256 x coords being invalid because each of the valid coords gives you two points (y and p-y).

1440000bytes commented at 6:25 pm on January 29, 2024: none

We are facing a problem in BTCPay: A large number of users have only 20GB of space to store the UTXO Set. Now that the UTXO Set is reaching 10GB and increasing, those users might have their node crashing sooner or later. So we are very interested into solutions to drop down the UTXO set size.

This pull request won’t make any difference. Top 5 output types in UTXO set are P2PKH, P2WPKH, P2TR, P2SH and P2WSH.

https://txstats.com/d/000000054/utxo-set-repartition-by-output-type?orgId=1

NicolasDorier commented at 11:39 pm on January 29, 2024: contributor

@1440000bytes the repartition you are showing is on outputs count, not size. Since ordinal, utxoset size grew from 5GB to 10GB. Even if they are marginal in term of output count, in term of size, it takes around 50% of the UTXO Set at the moment.

I haven’t looked how they put JPEG and JSON files on the chain exactly, I assumed it was with invalid P2PK, but didn’t look closely into it.

achow101 commented at 11:49 pm on January 29, 2024: member

I haven’t looked how they put JPEG and JSON files on the chain exactly, I assumed it was with invalid P2PK, but didn’t look closely into it.

Ordinal inscriptions don’t use P2PK or bare multisig, they use P2TR with tapscripts. In order for an inscription to be in the blockchain, it must create and then spend a UTXO, so inscriptions themselves have no impact on the UTXO set. Ordinals may have an impact as there are people creating small outputs for their “rare” sats, but those are all perfectly valid P2TR outputs, so nothing can be done there.

Stamps is using the old Counterparty protocol that uses bare multisigs. These are 1-of-3 multisigs where one key is valid, and the other two are the data. These are also spendable so this PR would not remove them.

OP_RETURN outputs are not provably unspendable though, they have valid spending paths

The point of OP_RETURN outputs (defined as the ones following the template OP_RETURN <data>) is to have a provably unspendable output that can be, and already are, pruned from the UTXO set. Yes, technically there are scripts that can contain OP_RETURN that are valid as long as OP_RETURN is not executed, but that’s not what people generally refer to when saying “OP_RETURN output”.

russeree commented at 11:20 pm on March 7, 2024: contributor

Closing because the fragility of this PR does not justify it’s limited impact.