When this benchmark was imported from crypto-bench to *ring* and
ported from the libtest `#[bench]` framework to Criterion.rs, we
kept the macro-based structure from the original benchmarks. However,
Criterion.rs actually supports the kind of parameterized benchmarking
we do much more naturally, and so we don't need the macros. Get rid of
them.
Also remove distinction between TLS 1.2 and TLS 1.3 AAD. These
benchmarks were originally written long ago when the TLS 1.3 draft
specified a different AAD format.
I hope this will serve as a better example of how to write such
benchmarks than it previously did.
Add a new scalar base point multiplication test case generator that
where the points are *not* Montgomery-encoded. This way we don't need
to generate different test data files when the Montgomery encoding
for a curve isn't the same for 32-bit and 64-bit targets (P-521).
This version of the generator produces the test cases for all the
scalars that the current P-256 and P-384 tests generate, in the same
format; the only exception is that the point is not
Montgomery-encoded.
When I generated these test vectors, I gave all of them the same point: the
generator of the curve. Consequently these input files are 100% redundant
with the `point_mul_base.txt` input files. So just remove them and use the
`point_mul_base.txt` files instead.
Avoid using the P384_POINT type on the C side. It seems to work for all
the targets we support, for P-384, but this pattern probably doesn't
work in general. Especially due to alignment issues for 32-bit targets,
it is doubtful it would work for P-521.
Better match the Rust declarations of these functions. Prepare to
support more target platforms and more weird things (like P-521) that
by avoiding any kind of alignment assumptions at the language
boundary (or elsewhere).
When `elem_exp_consttime` replaced `BN_mod_exp_mont_consttime` I did
not fully understand the way the table was constructed in the original
function. Recent BoringSSL changes clarify the table construction. Do
it the same way, to restore performance to what it was previously.
This addresses the `// TODO: Optimize this to avoid gathering`.
When this code was written, it wasn't clear which assembly language
functions took a pointer to the entire state vs. just a pointer to
the accumulator (etc.). Now upstream clarified things and we can
clarify this code.
The current macro initializer is not uniform across architectures,
and doesn't work when the number of 64bit limbs is not exactly half
that of 32bit limbs, (i.e. 521 bit).
This change replaces all limbs! initializers with a new const
function with hex string initializers, which is more portable and
more ergonomic.
Enable `ring::agreement` for wasm32 targets using the small
implementation of Curve25519.
Use the small Curve25519 implementation of Curve25519 (and
P-256) for all targets except for Aarch64 and x86-64.
Besides being smaller, the small implementations are likely
more resistant to compiler-introduced side-channels, which
is especially important for the wasm32 virtual machine and
other targets that we don't QA as thoroughly as AAarch64
and x86-64.
Bring in the previously-removed definition of `fe_mul_llt`
from BoringSSL as of commit
8d71d244c0debac4079beeb02b5802fde59b94bd.
Benchmarks require Rust 1.65 or later currently due to benchmark-only
dependencies.
Also, benchmarks add a lot of downloading/build time because they have
many more dependencies. So, skipping them in these jobs makes them
faster.
Don't warn during `cargo build` or `cargo test`, but deny them
in `mk/clippy.sh` so that they are caught in CI. Limit `cargo build`
warnings to things that matter during the edit/test cycle.
