crypto: x86/aes-gcm - clean up AVX512 code to assume 512-bit vectors

Browse source

aes-gcm-vaes-avx512.S (originally aes-gcm-avx10-x86_64.S) was designed
to support multiple maximum vector lengths, while still utilizing AVX512
/ AVX10 features such as the increased number of vector registers.
However, the support for multiple maximum vector lengths turned out to
not be useful.  Support for maximum vector lengths other than 512 bits
was removed from the AVX10 specification, which leaves "avoiding
overly-eager downclocking" as the only remaining use case for limiting
AVX512 / AVX10 code to 256-bit vectors.  But this issue has gone away in
new CPUs, and the separate VAES+AVX2 code which I ended up having to
write anyway provides nearly as good 256-bit support.

Therefore, clean up this code to not be written in terms of a generic
vector length, but rather just assume 512-bit vectors.

Acked-by: Ard Biesheuvel <ardb@kernel.org>
Tested-by: Ard Biesheuvel <ardb@kernel.org>
Link: https://lore.kernel.org/r/20251002023117.37504-5-ebiggers@kernel.org
Signed-off-by: Eric Biggers <ebiggers@kernel.org>

...

This commit is contained in:

Eric Biggers 2025-10-01 19:31:13 -07:00

parent 12beec21c5

commit 4b582e0fb3

1 changed files with 153 additions and 200 deletions

Show all changes Ignore whitespace when comparing lines Ignore changes in amount of whitespace Ignore changes in whitespace at EOL

Download patch file Download diff file Expand all files Collapse all files

353

arch/x86/crypto/aes-gcm-vaes-avx512.S

View file