2da94a4c6f fuzz: add a fuzz target for Miniscript decoding from Script (Antoine Poinsot)
f8369996e7 Miniscript: ops limit and stack size computation (Pieter Wuille)
2e55e88f86 Miniscript: conversion from script (Pieter Wuille)
1ddaa66eae Miniscript: type system, script creation, text notation, tests (Pieter Wuille)
4fe29368c0 script: expose getter for CScriptNum, add a BuildScript helper (Antoine Poinsot)
f4e289f384 script: move CheckMinimalPush from interpreter to script.h (Antoine Poinsot)
31ec6ae92a script: make IsPushdataOp non-static (Antoine Poinsot)
Pull request description:
Miniscript is a language for writing (a subset of) Bitcoin Scripts in a structured way.
Miniscript permits:
- To safely extend the Output Descriptor language to many more scripting features thanks to the typing system (composition).
- Statical analysis of spending conditions, maximum spending cost of each branch, security properties, third-party malleability.
- General satisfaction of any correctly typed ("valid" [0]) Miniscript. The satisfaction itself is also analyzable.
- To extend the possibilities of external signers, because of all of the above and since it carries enough metadata.
Miniscript guarantees:
- That for any statically-analyzed as "safe" [0] Script, a witness can be constructed in the bounds of the consensus and standardness rules (standardness complete).
- That unless the conditions of the Miniscript are met, no witness can be created for the Script (consensus sound).
- Third-party malleability protection for the satisfaction of a sane Miniscript, which is too complex to summarize here.
For more details around Miniscript (including the specifications), please refer to the [website](https://bitcoin.sipa.be/miniscript/).
Miniscript was designed by Pieter Wuille, Andrew Poelstra and Sanket Kanjalkar.
This PR is an updated and rebased version of #16800. See [the commit history of the Miniscript repository](https://github.com/sipa/miniscript/commits/master) for details about the changes made since September 2019 (TL;DR: bugfixes, introduction of timelock conflicts in the type system, `pk()` and `pkh()` aliases, `thresh_m` renamed to `multi`, all recursive algorithms were made non-recursive).
This PR is also the first in a series of 3:
- The first one (here) integrates the backbone of Miniscript.
- The second one (#24148) introduces support for Miniscript in Output Descriptors, allowing for watch-only support of Miniscript Descriptors in the wallet.
- The third one (#24149) implements signing for these Miniscript Descriptors, using Miniscript's satisfaction algorithm.
Note to reviewers:
- Miniscript is currently defined only for P2WSH. No Taproot yet.
- Miniscript is different from the policy language (a high-level logical representation of a spending policy). A policy->Miniscript compiler is not included here.
- The fuzz target included here is more interestingly extended in the 3rd PR to check a script's satisfaction against `VerifyScript`. I think it could be further improved by having custom mutators as we now have for multisig (see https://github.com/bitcoin/bitcoin/issues/23105). A minified corpus of Miniscript Scripts is available at https://github.com/bitcoin-core/qa-assets/pull/85.
[0] We call "valid" any correctly-typed Miniscript. And "safe" any sane Miniscript, ie one whose satisfaction isn't malleable, which requires a key for any spending path, etc..
ACKs for top commit:
jb55:
ACK 2da94a4c6f
laanwj:
Light code review ACK 2da94a4c6f (mostly reviewed the changes to the existing code and build system)
Tree-SHA512: d3ef558436cfcc699a50ad13caf1e776f7d0addddb433ee28ef38f66ea5c3e581382d8c748ccac9b51768e4b95712ed7a6112b0e3281a6551e0f325331de9167
fa5d2e678c Remove unused char serialize (MarcoFalke)
fa24493d63 Use spans of std::byte in serialize (MarcoFalke)
fa65bbf217 span: Add BytePtr helper (MarcoFalke)
Pull request description:
This changes the serialize code (`.read()` and `.write()` functions) to take a `Span` instead of a pointer and size. This is a breaking change for the serialize interface, so at no additional cost we can also switch to `std::byte` (instead of using `char`).
The benefits of using `Span`:
* Less verbose and less fragile code when passing an already existing `Span`(-like) object to or from serialization
The benefits of using `std::byte`:
* `std::byte` can't accidentally be mistaken for an integer
The goal here is to only change serialize to use spans of `std::byte`. If needed, `AsBytes`, `MakeUCharSpan`, ... can be used (temporarily) to pass spans of the right type.
Other changes that are included here:
* [#22167](https://github.com/bitcoin/bitcoin/pull/22167) (refactor: Remove char serialize by MarcoFalke)
* [#21906](https://github.com/bitcoin/bitcoin/pull/21906) (Preserve const in cast on CTransactionSignatureSerializer by promag)
ACKs for top commit:
laanwj:
Concept and code review ACK fa5d2e678c
sipa:
re-utACK fa5d2e678c
Tree-SHA512: 08ee9eced5fb777cedae593b11e33660bed9a3e1711a7451a87b835089a96c99ce0632918bb4666a4e859c4d020f88fb50f2dd734216b0c3d1a9a704967ece6f
As XOnlyPubKey has a Span-based constructor, that can be used directly
without needing to first convert the byte sequence into a vector, only
to convert that to a uint256, which only then can then be passed as a
span to the constructor.
f9e37f33ce doc: IsFinalTx comment about nSequence & OP_CLTV (Yuval Kogman)
Pull request description:
It's somewhat surprising that a transaction's `nLockTime` field is ignored
when all `nSequence` fields are final, so this change aims to clarify this
behavior and cross reference relevant details of `OP_CHECKLOCKTIMEVERIFY`.
ACKs for top commit:
MarcoFalke:
ACK f9e37f33ce
Tree-SHA512: 88460dacbe4b8115fb1948715f09b21d4f34ba1da9e88d52f0b774a969f845e9eddc5940e7fee66eacdd3062dc40d6d44c3f282b0e5144411fd47eb2320b44f5
At verification time, the to be precomputed data can be inferred from
the transaction itself. For signing, the necessary witnesses don't
exist yet, so just permit precomputing everything in that case.
That results in a much safer interface (making the tweak commit
to the key implicitly using a fixed tag means it can't be used for
unrelated tweaking).
We were previously ruling out 17-20 pubkeys multisig, while they are
only invalid under P2SH context.
This makes multisigs with up to 20 keys be detected as valid by the
solver. This is however *not* a policy change as it would only apply
to bare multisigs, which are already limited to 3 pubkeys.
Note that this does not change the sigOpCount calculation (as it would
break consensus). Therefore 1-16 keys multisigs are counted as 1-16 sigops
and 17-20 keys multisigs are counted as 20 sigops.
Signed-off-by: Antoine Poinsot <darosior@protonmail.com>
It's somewhat surprising that a transaction's nLockTime field is ignored
when all nSequence fields are final, so this change aims to clarify this
behavior and cross reference relevant details of OP_CHECKLOCKTIMEVERIFY.
Historically lack of amount data has been treated as amount==-1. Change
this and treat it as missing data, as introduced in the previous commits.
To be minimally invasive, do this at SignatureHash() call sites rather
than inside SignatureHash() (which currently has no means or returning
a failure code).
This allows specifying how *TransactionSignatureChecker will behave when
presented with missing transaction data such as amounts spent, BIP341 data,
or spent outputs.
As all call sites still (implicitly) use MissingDataBehavior::ASSERT_FAIL,
this commit introduces no change in behavior.
This adds a new `SigVersion::TAPSCRIPT`, makes the necessary interpreter
changes to make it implement BIP342, and uses them for leaf version 0xc0
in Taproot script path spends.
This includes key path spending and script path spending, but not the
Tapscript execution implementation (leaf 0xc0 remains unemcumbered in
this commit).
Includes constants for various aspects of the consensus rules suggested
by Jeremy Rubin.
This enables the schnorrsig module in libsecp256k1, adds the relevant types
and functions to src/pubkey, as well as in higher-level `SignatureChecker`
classes. The (verification side of the) BIP340 test vectors is also added.
This implements the new sighashing scheme from BIP341, with all relevant
whole-transaction values precomputed once and cached.
Includes changes to PrecomputedTransactionData by Pieter Wuille.
A BIP-341 signature message may commit to the scriptPubKeys and amounts
of all spent outputs (including other ones than the input being signed
for spends), so keep them available to signature hashing code.
In preparation for adding Schnorr versions of `CheckSig`, `VerifySignature`, and
`ComputeEntry`, give them an ECDSA specific name.
-BEGIN VERIFY SCRIPT-
sed -i 's/CheckSig(/CheckECDSASignature(/g' $(git grep -l CheckSig ./src)
sed -i 's/VerifySignature(/VerifyECDSASignature(/g' $(git grep -l VerifySignature ./src)
sed -i 's/ComputeEntry(/ComputeEntryECDSA(/g' $(git grep -l ComputeEntry ./src)
-END VERIFY SCRIPT-
f63dec189c [REFACTOR] Initialize PrecomputedTransactionData in CheckInputScripts (Pieter Wuille)
Pull request description:
This is a single commit taken from the Schnorr/Taproot PR #17977.
Add a default constructor to `PrecomputedTransactionData`, which doesn't initialize the struct's members. Instead they're initialized inside the `CheckInputScripts()` function. This allows a later commit to add the spent UTXOs to that structure. The spent UTXOs are required for the schnorr signature hash, since it commits to the scriptPubKeys. See https://github.com/bitcoin/bips/blob/master/bip-0341.mediawiki#common-signature-message for details.
By itself, this isn't really an improvement to the code, but I think it makes sense to separate out the refactor/moveonly commits from PR #17977 so that PR is only the logical changes needed for Schnorr/Taproot.
ACKs for top commit:
jonatack:
Re-ACK f63dec1 `git diff 851908d f63dec1` shows no change since last ACK.
sipa:
utACK f63dec189c
theStack:
re-ACK f63dec189c
fjahr:
Re-ACK f63dec189c
ariard:
Code Review ACK f63dec1
Tree-SHA512: ecf9154077824ae4c274b4341e985797f3648c0cb0c31cb25ce382163b923a3acbc7048683720be4ae3663501801129cd0f48c441a36f049cc304ebe9f30994e
Add a default constructor to `PrecomputedTransactionData`, which doesn't
initialize the struct's members. Instead they're initialized inside the
`CheckInputScripts()` function. This allows a later commit to add the
spent UTXOs to that structure.
14e8cf974a [consensus] MOVEONLY: Move single-sig checking EvalScript code to EvalChecksig (Pieter Wuille)
Pull request description:
This is another small refactor pulled out of the Schnorr/Taproot PR #17977.
This is in preparation for adding different signature verification rules,
specifically tapscript (BIP 342), which interprets opcode 0xac and 0xad
as Schnorr signature verifications.
ACKs for top commit:
sipa:
ACK 14e8cf974a, verified move-only.
MarcoFalke:
ACK 14e8cf974a, reviewed with "git show 14e8cf974a --color-moved=dimmed-zebra --color-moved-ws=ignore-all-space -W" 👆
fjahr:
Code-review ACK 14e8cf974a, verified that it's move-only.
instagibbs:
code review ACK 14e8cf974a, verified move-only
theStack:
Code-Review ACK 14e8cf974a
jonatack:
ACK 14e8cf974a
Tree-SHA512: af2efce9ae39d5ec01db5b9ef0ff383fe252ef5f33b3483927308ae17d91a619266cb45951f32ea1ce54807a4c0f052bcdefb47e244465d3a726393221c227b1
This is in preparation for adding different signature verification rules,
specifically tapscript (BIP 342), which interprets opcode 0xac and 0xad
as Schnorr signature verifications.
e6e622e5a0 Implement O(1) OP_IF/NOTIF/ELSE/ENDIF logic (Pieter Wuille)
d0e8f4d5d8 [refactor] interpreter: define interface for vfExec (Anthony Towns)
89fb241c54 Benchmark script verification with 100 nested IFs (Pieter Wuille)
Pull request description:
While investigating what mechanisms are possible to maximize the per-opcode verification cost of scripts, I noticed that the logic for determining whether a particular opcode is to be executed is O(n) in the nesting depth. This issue was also pointed out by Sergio Demian Lerner in https://bitslog.wordpress.com/2017/04/17/new-quadratic-delays-in-bitcoin-scripts/, and this PR implements a variant of the O(1) algorithm suggested there.
This is not a problem currently, because even with a nesting depth of 100 (the maximum possible right now due to the 201 ops limit), the slowdown caused by this on my machine is around 70 ns per opcode (or 0.25 s per block) at worst, far lower than what is possible with other opcodes.
This PR mostly serves as a proof of concept that it's possible to avoid it, which may be relevant in discussions around increasing the opcode limits in future script versions. Without it, the execution time of scripts can grow quadratically with the nesting depth, which very quickly becomes unreasonable.
This improves upon #14245 by completely removing the `vfExec` vector.
ACKs for top commit:
jnewbery:
Code review ACK e6e622e5a0
MarcoFalke:
ACK e6e622e5a0🐴
fjahr:
ACK e6e622e5a0
ajtowns:
ACK e6e622e5a0
laanwj:
concept and code review ACK e6e622e5a0
jonatack:
ACK e6e622e5a0 code review, build, benches, fuzzing
Tree-SHA512: 1dcfac3411ff04773de461959298a177f951cb5f706caa2734073bcec62224d7cd103767cfeef85cd129813e70c14c74fa8f1e38e4da70ec38a0f615aab1f7f7
