secp256k1 jets #18
secp256k1 jets #18
Conversation
In GCC you can "use" a parameter 'foo' with the statement '(void) foo;'. Using this trick we can eliminate the unused parameter warnings in the particular cases where we necessarily have unused parameters.
Specifically we add * sqrt for field elemnts * offsetPoint for group elements (adding a point in affine coordinates to a point in jacobian coordinates) * ecMult: Add a scalar multiple of a group element (in jacobian coordinates) to a scalar multiple of the secp256k1 generator point * schnorrVerify: Decide if a schnorr signature is valid for a given public key-message pair. These jets "exercise" the various domains of secp256k1. Later we will add a whole host of jets for secp256k1 functionality, but for now ecMult and schnorrVerify are the most useful functions.
|
What commit did you get these libsecp files from? and I guess you renamed |
|
oh, I see, you actually modified all the files |
| } ecmult_static; | ||
|
|
||
| /* This function will initialize the 'ecmult_static' global variable if it hasn't already been initialized. */ | ||
| static void ensure_ecmult_static(void) { |
There was a problem hiding this comment.
The thread-unsafety of this really makes me uncomfortable, though I can't think of any other way to do this. Maybe putting a secp context into txEnv?
There was a problem hiding this comment.
This is a more general problem with the entire library. I've created a separate issue to track it .
| static int secp256k1_eckey_pubkey_parse(secp256k1_ge *elem, const unsigned char *pub, size_t size); | ||
| static int secp256k1_eckey_pubkey_serialize(secp256k1_ge *elem, unsigned char *pub, size_t *size); | ||
|
|
||
| #if 0 |
There was a problem hiding this comment.
Is this #if 0 code meant to be committed?
| * It only operates on tables sized for WINDOW_A wnaf multiples. | ||
| * - secp256k1_ecmult_odd_multiples_table_storage_var, which converts its | ||
| * resulting point set to actually affine points, and stores those in pre. | ||
| * It operates on tables of any size, but uses heap-allocated temporaries. |
There was a problem hiding this comment.
FWIW this comment was updated in bitcoin-core/secp256k1#677 -- apparently at your advice.
|
|
||
| /* Converts the point encoded by a secp256k1_pubkey into its absolute value, | ||
| * i.e. keeps it as is if it is positive and otherwise negates it. Sign is set | ||
| * to 1 if the input point was negative and set to 0 otherwise. */ |
There was a problem hiding this comment.
This comment (and the function implementation) are also different upstream. See https://github.com/jonasnick/secp256k1/blob/schnorrsig/src/secp256k1.c#L736
There was a problem hiding this comment.
And in particular, I find the existing text which talks about a point being "positive" or "negative" to be confusing
|
Moved to #34. |
Review C/secp256k1/README.md for information about our copy of libsecp256k1.
Also included is a commit that eliminates unused parameter warnings.