-
Notifications
You must be signed in to change notification settings - Fork 267
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
[WIP] Add bip352 silentpayments module #721
base: master
Are you sure you want to change the base?
Conversation
This modification allows running the command `./vendor-libsecp.sh -f pull/1519/head` to obtain MuSig2 components from `bitcoin-core/secp256k1`. This commit may be discarded after merging PR #1519.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Cool! Concept ACK.
ec34fda
to
c217147
Compare
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Thanks for picking this up!
src/constants.rs
Outdated
@@ -33,6 +33,9 @@ pub const KEY_PAIR_SIZE: usize = 96; | |||
/// The size of a full ElligatorSwift encoding. | |||
pub const ELLSWIFT_ENCODING_SIZE: usize = 64; | |||
|
|||
/// The size of a silent payment public data. |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
in 0e7e062 ("Add Silent Payments module"):
nit: "silent payments"
/// The size of a silent payment public data. | |
/// The size of a silent payments public data object. |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Thanks for the review. Done !
pub fn silentpayments_sender_create_outputs<C: Verification>( | ||
secp: &Secp256k1<C>, | ||
recipients: &[SilentpaymentsRecipient], | ||
smallest_outpoint: &[u8; 36], |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
in 0e7e062 ("lib: Add Silent Payments module"):
Would it make sense to use an Outpoint
type here (assuming one exists)? On the one hand, it makes things more clear and less foot gunny for the caller, on the other hand I can see an argument for having the rust-secp256k1 crate be unaware of Bitcoin protocol specifics.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Sounds like we should wait for bitcoin-primitives
to be released and gate this behind an optional dependency on it.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I had a massive brainfart here. We want to optionally use crypto
in primitives
which will optionally depend on secp256k1
, so secp256k1
cannot (optionally) depend on primitives
as it'd create a cycle if all features are on. I'm not sure what the best solution is but having silent payments implemented outside secp256k1
sounds most appealing at first thought.
src/silentpayments.rs
Outdated
/// Creates an `SilentpaymentsPublicData` object from a 98-byte array. | ||
pub fn from_array(data: [u8; 98]) -> SilentpaymentsPublicData { | ||
SilentpaymentsPublicData(ffi::SilentpaymentsPublicData::from_array(data)) | ||
} | ||
|
||
/// Returns the 64-byte array representation of this `SilentpaymentsPublicData` object. | ||
pub fn to_array(&self) -> [u8; 98] { self.0.to_array() } |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
in 0e7e062 ("lib: Add Silent Payments module"):
How do you foresee the to_array/from_array functions being used? public_data
is meant to be an opaque data type, so it seems better to me to not have these methods and instead have callers always use the create
and serialize
functions, but perhaps I'm missing something.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Good catch ! Done.
c217147
to
36b2c6b
Compare
Removed dynamically sized vector in most functions and simplified the code.
|
unsafe { | ||
// Get a pointer to the inner ffi::PublicKey | ||
let ffi_pubkey_ptr: *const ffi::PublicKey = pubkey.as_c_ptr(); | ||
|
||
// Dereference the pointer to get the ffi::PublicKey | ||
// Then create a copy of it | ||
(*ffi_pubkey_ptr).clone() | ||
} |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Unnecessary unsafe
and clone
, PublicKey
wraps ffi::PublicKey
and they are both Copy
.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Thanks. Done in 6fbb059
src/silentpayments.rs
Outdated
for (i, pubkey) in out_pubkeys.iter_mut().enumerate() { | ||
out_pubkeys_ptrs[i] = pubkey as *mut _; | ||
} |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Looks redundant, doesn't out_pubkeys.iter_mut().map(|k| k as *mut _).collect()
(line 81) do the same?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Thanks. Done in 6fbb059
|
||
let (ffi_taproot_seckeys, n_taproot_seckeys) = match taproot_seckeys { | ||
Some(keys) => { | ||
let ffi_keys: &[*const ffi::Keypair] = transmute::<&[&Keypair], &[*const ffi::Keypair]>(taproot_seckeys.unwrap()); |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Pointer casting is recommended over transmuting pointers and references, but both approaches are unsound if Keypair
is not guaranteed to have the same layout as ffi::Keypair
. Adding #[repr(transparent)]
to Keypair
fixes this.
Same comment for SecretKey
a few lines down, and some more in this file, please check all uses of transmute
and use pointer casting where possible, and review if the source and destination type are layout-compatible.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
src/silentpayments.rs
Outdated
); | ||
|
||
if res == 1 { | ||
Ok(out_pubkeys.into_iter().map(XOnlyPublicKey::from).collect()) |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
This can be optimized if XOnlyPublicKey
is #[repr(transparent)]
. (using Vec::from_raw_parts)
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
src/silentpayments.rs
Outdated
return Ok(LabelTweakResult { pubkey, label_tweak }); | ||
} else { | ||
return Err(LabelTweakError::Failure); | ||
} |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Syntax could be more "rusty" by removing the 'return' keywords (this is a clippy warning, see other clippy warnings for more simple improvements like this)
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Thanks. Done in 6fbb059
src/silentpayments.rs
Outdated
unsafe { | ||
|
||
let empty_data = [0u8; constants::SILENT_PAYMENTS_PUBLIC_DATA_SIZE]; | ||
let mut silentpayments_public_data = ffi::SilentpaymentsPublicData::from_array(empty_data); |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
This can use Self::new()
, and as_mut_c_ptr()
for the c function. It then does not need to be wrapped before returning.
Same comment for line 308.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Thanks. Done in 6fbb059
src/silentpayments.rs
Outdated
let mut result = vec![SilentpaymentsFoundOutput::empty(); n_found_outputs]; | ||
|
||
for i in 0..n_found_outputs { | ||
result[i] = SilentpaymentsFoundOutput(out_found_output[i]); | ||
|
||
} |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Can be optimized to use Vec::from_raw_parts
if SilentpaymentsFoundOutput
is made #[repr(transparent)]
.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Thanks. Done in 6fbb059
label_lookup: SilentpaymentsLabelLookupFunction, | ||
label_context: L, |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
This requires users to use unsafe, and is maybe even unsound to accept a function that can return pointers to a "safe" (not annotated unsafe) function.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
This is also unsound on Rust < 1.81, see https://blog.rust-lang.org/2024/09/05/Rust-1.81.0.html#abort-on-uncaught-panics-in-extern-c-functions. To fix that you have to catch_unwind
and abort
manually when using older versions.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I think I managed to create a safe abstraction for this callback (given the current restrictions, which might change as bitcoin-core/secp256k1#1519 is not merged yet). I added it (2b3f633) to a branch on my fork, feel free to add (may be modified) it here.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Thanks @antonilol for all your contributions, here and to other projects like electrs! They are very helpful.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Thanks for working on this and proposing a better approach.
But I haven't added the commit yet because it still has some TODO and the example file is not using the modified function.
I would like to better understand your proposal and how we can implement the TODOs and call the function from the example file.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
First TODO: the C library allows you to pass a null pointer for the label lookup function, but the function signature on the Rust side does not, this type definition must be wrapped in Option<...>
to allow null pointers.
Second TODO: the C library checks that the context pointer is non null when the function pointer is non null, in my opinion the C library should treat it as opaque and 100% managed by the user so should not check it, just pass it to the callback (comment thread, same link as in the code). Note this will not be an issue for the Rust side anymore when TODO 1 is fixed, but at first I thought that could be harder because bindings are normally generated.
Function pointer + context pointer is a C thing, in Rust a closure automatically captures the variables it needs. To allow users to pass closures, the closure needs to be "converted" to a function pointer and context pointer for the C side to use.
This safe wrapper stores the state part of the closure (a Rust closure can be seen as a combination of a function and state) on the stack and passes the pointer to that as the context pointer. In the callback the function and the state are but "back together" and called.
To add to this, the C side expects a pointer to the return value. The C function uses this after the callback returned (obviously), so this pointer needs to live for longer than the callback, but as long as the safe wrapper is enough, so that return value is stored on the stack as well. This storage (the variable named storage
) is passed with the closure state to the C side as the context pointer.
Calling from the example (or from any user code) should be straightforward, these should (roughly) be the changes needed to the example:
- make the cache a Vec for simplicity
- change the callback function (in the argument list) to
|label| cache.iter().find(|cache_entry| cache_entry.label == label).map(|cache_entry| cache_entry.label_tweak)
cache
will be a captured variable and the safe wrapper will take care of getting it to the C side.
} | ||
} | ||
|
||
impl SilentpaymentsFoundOutput { |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Maybe we should implement some getter functions too?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
What would be the use cases for the getter functions?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I may be missing something, but in the tests we're just printing the SilentpaymentsFoundOutput
returned by silentpayments_recipient_scan_outputs
, and I can't figure out how I would take the x only pubkey to update my wallet state, let alone use the tweak to actually create the spending key for that output.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I made a quick commit just to show you what I mean https://github.com/Sosthene00/rust-secp256k1/tree/bip352-add-getters
36b2c6b
to
9566ad2
Compare
This PR adds bip352
silentpayments
module based on bitcoin-core/secp256k1#1519.Currently, it creates bindings for all silent payment structs and functions and also adds the high-level API to access them.
examples/silentpayments.rs
implements the same example asexamples/silentpayments.c
from the original repository and shows how to use all functions.The example can be run with
cargo run --example silentpayments --features="rand std"
.Although it implements all the features of the
silentpayments
module, this is still a work in progress, as the current module only works with the standard library.For review purposes, only the last three commits matter. The others may be discarded after the module is merged into the original repository and the
depend
folder updated.