SHA-512 accelerated by x86#731

Open

MarekKnapek wants to merge 9 commits intolibtom:developfrom

MarekKnapek:sha512x86

Contributor

MarekKnapek commented Apr 17, 2026

SHA-512 accelerated by x86.

Contributor Author

MarekKnapek commented Apr 17, 2026

I have no idea what is the problem on ci with the amalgamed build. How do I run this myself?

karel-m reviewed

View reviewed changes

src/hashes/sha2/sha384_desc.c

karel-m reviewed

View reviewed changes

src/hashes/sha2/sha512_224_desc.c

karel-m reviewed

View reviewed changes

src/hashes/sha2/sha512_256_desc.c

karel-m reviewed

View reviewed changes

src/hashes/sha2/sha512_desc.c

MarekKnapek force-pushed the sha512x86 branch from 4d1344f to 7fbfdad Compare

April 23, 2026 06:37

karel-m reviewed

View reviewed changes

src/hashes/sha2/sha512_224_desc.c Outdated

karel-m reviewed

View reviewed changes

src/hashes/sha2/sha512_256_desc.c Outdated

MarekKnapek force-pushed the sha512x86 branch from ecbfe31 to ca377c2 Compare

April 23, 2026 07:39

MarekKnapek added 7 commits

April 23, 2026 12:34


          SHA-512 accelerated by x86

3485f56


          GCC >= 14 clang >= 17 MSVS >= 2022

9cfd41a


          LTC_S_X86_CPUIDEX

e1057cb


          SHA-512 constants table.

9a0c6f8


          Bring back unaligned reads.

8a896f5

We cannot force clients to align their buffers.


          LTC_UNUSED_PARAM

67e8811


          Fix from code review.

772d916

sjaeckel force-pushed the sha512x86 branch from ca377c2 to 772d916 Compare

April 23, 2026 10:35

sjaeckel requested changes

View reviewed changes

Member

sjaeckel left a comment

Basically LGTM, besides this one comment for now. I didn't look in further detail yet, but will do that tomorrow.

By merging #726 there was a merge conflict, which I've just resolved while rebasing this branch and therefore I've force-pushed.

src/hashes/sha2/sha512_x86.c Outdated

MarekKnapek added 2 commits

April 23, 2026 13:00


          Code review, alignment.

3542e52


          post-rebase-fix

c7bf504

sjaeckel mentioned this pull request

sha1_x86.c related failures with old compilers (Linux, FreeBSD) #738

Closed

sjaeckel requested changes

View reviewed changes

Member

sjaeckel left a comment •

edited

Loading

That last commit doesn't belong in this PR and requires further checks before applying it.

We started using inline functions declared in header files some time ago, but then @karel-m had some platforms where this doesn't work so we reverted back to not using them. We'd have to check again today. Please remove that commit and we'll do that in a separate PR.

MarekKnapek force-pushed the sha512x86 branch from 425f516 to c7bf504 Compare

April 25, 2026 00:46

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet