crypto: replace handwritten assembly with native Go SIMD

[thatnealpatel]

The current state of the SIMD API (#73787) provides a great opportunity to improve the safety of divergent assembly implementations in crypto/ through better maintainability and readability. In some cases, we can also improve the performance of these implementations.

The primary targets are SHA1 and SHA256; however, others may follow as the compiler evolves.

[gopherbot]

Change https://go.dev/cl/778100 mentions this issue: crypto/internal/fips140/sha256: rewrite SHA256 SHA-NI using simd/archsimd

[gopherbot]

Change https://go.dev/cl/778102 mentions this issue: crypto/sha1: rewrite SHA1 SHA-NI using simd/archsimd

[gabyhelp]

Related Issues

• proposal: crypto: replace assembly implementations with internal instrinsics #64634
• crypto/sha1: add native SHA1 instruction implementation for AMD64 #27443
• crypto/sha,crypto/sha256: consider dropping extension-less assembly implementations #71525 (closed)
• x/crypto/sha3: add SHA3 assembly implementation for ARMv7 #28148
• crypto/md5: remove assembly cores #49248
• crypto: new assembly policy #37168 (closed)
• wiki: clarify Assembly Policy for crypto #28126 (closed)
• crypto: drop pre-AVX2 amd64 assembly #69587
• math, crypto, hash: Write power64 versions of assembler routines where applicable #8913

Related Code Changes

• crypto/sha: implement SHA1 & SHA256 acceleration using Intel SHA extensions

_{(Emoji vote if this was helpful or unhelpful; more detailed feedback welcome in this discussion.)}

[gopherbot]

Change https://go.dev/cl/778120 mentions this issue: crypto/internal/fips140/sha256: rewrite SHA256 AVX2 using simd/archsimd

[gopherbot]

Change https://go.dev/cl/778260 mentions this issue: crypto/internal/fips140/aes: rewrite AES-NI using simd/archsimd

[gopherbot]

Change https://go.dev/cl/778420 mentions this issue: crypto/internal/fips140/mlkem: add accelerated simd/archsimd implementation