No-Montgomery inverse is actuallly faster for the field prime on Curve25519

Expose TurboSHAKE128 API publicly

And another!

Fixed some typos

Fixed slow modular reductions

Merge pull request 'Optimized and fixed UB in ChaCha20-Poly1305' (#1) from optimize-chacha20-simd into main

Reviewed-on: https://gitea.eyes-like-fire.org/juniper/TinyCrypT/pulls/1

Revert "Testing something"

This reverts commit d3bd61b876270c9366b24991118095f000661baf.

Testing something

Fixed buffer overflow in portable Ed25519, oopsies

Fixed failing tests on portable Ed25519

Started work on ESP32-S3 port

Disabled tests for x25519 by default (now that all pass)

Fixed builds on Raspberry Pi

X25519 optimizations passing tests

Started on ESP32S3 port

Reorganizations

X25519 optimizations fail test vector 2 from the RFC

Optimized X25519

Yet more gloating

Gigabit+ throughput on CC20-P1305

Fixed issue with cross compilations

More gloating

Forgot to stage all the changes

45% of Bernstein et. al.'s performance

Satisfied with Ed25519 performance

Forgot to delete test file

Ed25519 signatures are under 2ms

Cleanup

Ed25519 testing and optimization

Gloating

SHA2 implementations pass all 586 NIST tests and valgrind

Forgot to delete the Makefile

Set up meson, ninja, and GoogleTest

Yay things work

Libraries and vanity, as always

Made all the red squigglies from clangd go away

Started work on testing

Ed25519 implemented and working

Tweak x2

Tweak

Added SHA-2 implementations

Tweak

X25519 is working

KangarooTwelve128 works for arbitrary input

KangarooTwelve128 working for small inputs

Further work (X25519 still doesn't work)

Dev environment setup

Second commit

Initial commit

Optimized and fixed UB in ChaCha20-Poly1305