runtime-sdk: add p384 signing & verification

[nhynes]

No description provided.

[codecov]

Codecov Report

Merging #1519 (5e530c0) into main (6590ae6) will increase coverage by 0.04%.
The diff coverage is 64.64%.

@@            Coverage Diff             @@
##             main    #1519      +/-   ##
==========================================
+ Coverage   60.45%   60.50%   +0.04%     
==========================================
  Files         138      139       +1     
  Lines        9894     9957      +63     
==========================================
+ Hits         5981     6024      +43     
- Misses       3871     3891      +20     
  Partials       42       42              

Files	Coverage Δ
...ime-sdk/modules/evm/src/precompile/confidential.rs	80.64% <100.00%> (+0.31%)	⬆️
runtime-sdk/src/crypto/signature/secp256k1.rs	66.66% <100.00%> (-0.94%)	⬇️
runtime-sdk/src/crypto/signature/secp256r1.rs	53.70% <33.33%> (+5.21%)	⬆️
runtime-sdk/src/crypto/signature/mod.rs	72.48% <77.77%> (+0.61%)	⬆️
runtime-sdk/src/crypto/signature/secp384r1.rs	53.70% <53.70%> (ø)

[kostko]

If you want this to be accessible from EVM, make sure to also update the gas costs in the confidential precompiles. Ideally you would add some benchmarks to establish proper costs (see how it was done for the others).

Otherwise looks good.

[nhynes]

The benches look like

test precompile::confidential::test::bench_sign_secp256r1_prehashed_sha256   ... bench:     288,916 ns/iter (+/- 1,086)
test precompile::confidential::test::bench_sign_secp384r1_prehashed_sha384   ... bench:   1,214,954 ns/iter (+/- 6,736)
test precompile::confidential::test::bench_verify_secp256r1_prehashed_sha256 ... bench:     277,456 ns/iter (+/- 1,579
)
test precompile::confidential::test::bench_verify_secp384r1_prehashed_sha384 ... bench:   1,178,009 ns/iter (+/- 4,977
)

and

test precompile::confidential::test::bench_sign_secp256r1_prehashed_sha256   ... bench:     248,305 ns/iter (+/- 184)
test precompile::confidential::test::bench_sign_secp384r1_prehashed_sha384   ... bench:   1,163,157 ns/iter (+/- 3,854)
test precompile::confidential::test::bench_verify_secp256r1_prehashed_sha256 ... bench:     233,778 ns/iter (+/- 1,733)
test precompile::confidential::test::bench_verify_secp384r1_prehashed_sha384 ... bench:   1,127,003 ns/iter (+/- 25,798)

and so forth, with about 4.8x being the max I've seen

[peternose]

so about 4.8x as slow

Did you mean 4.2x?

[nhynes]

Did you mean 4.2x?

It looks that way from the one sample, but it's highly variable, and sometimes I get 4.8x, so I took the max so that we can safely reduce it later.