In the field 5x52 asm for x86_64, stack variables are provided as outputs.
The existing inputs are all forcibly allocated to registers, so cannot
coincide, but mark them as early clobber anyway to make this clearer.
In the existing code, the compiler is allowed to allocate the RSI register
for outputs m0, m1, or m2, which are written to before the input in RSI is
read from. Fix this by marking them as early clobber.
Reported by ehoffman2 in https://github.com/bitcoin-core/secp256k1/issues/766
- secp256k1_scalar_cadd_bit
- secp256k1_modinvXX_normalize_YY
- secp256k1_modinvXX_divsteps_ZZ
- ECMULT_CONST_TABLE_GET_GE
Even though those code loations are not problematic right now
(with current compilers).
f3126fdfec norm arg: remove prove edge tests which are now covered by vectors (Jonas Nick)
847ed9ecb2 norm arg: add verification to prove vectors (Jonas Nick)
cf797ed2a4 norm arg: add prove test vectors (Jonas Nick)
095c1e749c norm arg: add prove_const to tests (Jonas Nick)
bf7bf8a64f norm arg: split norm_arg_zero into prove_edge and verify_zero_len (Jonas Nick)
a70c4d4a8a norm arg: add test vector for |n| = 0 (Jonas Nick)
f5e4b16f0f norm arg: add test vector for sign bit malleability (Jonas Nick)
c0de361fc5 norm arg: allow X and R to be the point at infinity (Jonas Nick)
f22834f202 norm arg: add verify vector for n = [0], l = [0] (Jonas Nick)
d8e7f3763b musig: move ge_{serialize,parse}_ext to module-independent file (Jonas Nick)
Pull request description:
ACKs for top commit:
Liam-Eagen:
ACK f3126fd
Tree-SHA512: 1aad86521fce74435beabe7690c7fcc38ad9ae7a884ddcab69ef825b573433f700723a7672d29df1b4465bc33d5957b6a46f657f988cfd2cc73fa94a3472357d
Apparently clang 15 is able to compile our cmov code into a branch,
at least for fe_cmov and fe_storage_cmov. This commit makes the
condition volatile in all cmov implementations (except ge but that
one only calls into the fe impls).
This is just a quick fix. We should still look into other methods,
e.g., asm and #457. We should also consider not caring about
constant-time in scalar_low_impl.h
We should also consider testing on very new compilers in nightly CI,
see https://github.com/bitcoin-core/secp256k1/pull/864#issuecomment-769211867
b43dd83b43 musig: add missing static keyword to function (Jonas Nick)
068e6a036a musig: add test vectors from BIP MuSig (Jonas Nick)
36621d13be musig: update to BIP v1.0.0-rc.2 "Add ''pk'' arg to ''NonceGen''" (Jonas Nick)
d717a4980b musig: update to BIP v0.8 "Switch from X-only to plain pk inputs." (Jonas Nick)
304f1bc96d extrakeys: add pubkey_sort test vectors from BIP MuSig2 (Jonas Nick)
ae89051547 extrakeys: replace xonly_sort with pubkey_sort (Jonas Nick)
98242fcdd9 extrakeys: add secp256k1_pubkey_cmp (Jonas Nick)
73d5b6654d musig: update to BIP v0.7.0 (NonceGen) (Jonas Nick)
060887e9d7 musig: update to BIP v0.5.1 "Rename ordinary tweaking to plain" (Jonas Nick)
cbe2815633 musig: update to BIP v0.4 "Allow the output of NonceAgg to be inf" (Jonas Nick)
206017d67d musig: update to BIP v0.3 (NonceGen) (Jonas Nick)
d800dd55db musig: remove test vectors (Jonas Nick)
Pull request description:
Version 1.0.0-rc.3 of BIP MuSig2 can be found [here](https://github.com/jonasnick/bips/pull/75). This PR does _not_ implement the following optional features that have been added to BIP MuSig2:
- variable length messages
- deterministic signing
- identifiable aborts
The PR also does _not_ yet change the `secnonce` structure to also contain the signer's public key (which would also imply changing the seckey argument in `sign` to a keypair). Additionally, we may want to rename some things in the future to be more consistent with the BIP (e.g. keyagg_cache vs. keyagg_ctx, applytweak vs. tweak_add).
ACKs for top commit:
ariard:
Light Code Review ACK b43dd83b, mostly looks on how the user API will make sense for Lightning, thanks for the answers!
real-or-random:
ACK b43dd83b43
Tree-SHA512: 9b1410951b55a1b0e6590b8c302052996d1fb6d9771765498b4282ff68b44ab0d6add8144c9330217b682ec5a93508b5546099db9a1f2c865f99253010dd76f4
- 0.7.0: Change ''NonceGen'' such that output when message is not present is different from when message is present but has length 0.
- 0.6.0: Change order of arguments and serialization of the message in the ''NonceGen'' hash function
