[API Implementation]: Support for Intel SHA extensions

[deeprobin]

Proposal implementation of #256 (closes#256)

Proposal

namespaceSystem.Runtime.Intrinsics.X86{publicabstractclassSha:Sse2{publicabstractclassX64:Sse2.X64{publicstaticboolIsSupported{get;}}publicstaticboolIsSupported{get;}publicstaticVector128<byte>Sha1MessageSchedule1(Vector128<byte>a,Vector128<byte>b);publicstaticVector128<byte>Sha1MessageSchedule2(Vector128<byte>a,Vector128<byte>b);publicstaticVector128<byte>Sha1NextE(Vector128<byte>a,Vector128<byte>b);publicstaticVector128<byte>Sha1FourRounds(Vector128<byte>state1,Vector128<byte>state2,bytefunc);publicstaticVector128<byte>Sha256MessageSchedule1(Vector128<byte>a,Vector128<byte>b);publicstaticVector128<byte>Sha256MessageSchedule2(Vector128<byte>a,Vector128<byte>b);publicstaticVector128<byte>Sha256TwoRounds(Vector128<byte>state1,Vector128<byte>state2,Vector128<byte>message);}}

Current state of implementation

• C++ CodeGen (JIT / NativeAOT) implementation (See JIT implementation for SHA instructions deeprobin/runtime#3)
• C++ Mono implementation
• C# Implementation
  • Ref Assembly
  • Intrinsic Implementation
• Doc comments
• Tests
  • HWIntrinsic Tests
  • Verification Tests

Used intrinsics

• Sha1
  • Sha1MessageSchedule1 – __m128i _mm_sha1msg1_epu32 (__m128i a, __m128i b)
  • Sha1MessageSchedule2 – __m128i _mm_sha1msg2_epu32 (__m128i a, __m128i b)
  • Sha1NextE – __m128i _mm_sha1nexte_epu32 (__m128i a, __m128i b)
  • Sha1FourRounds – __m128i _mm_sha1rnds4_epu32 (__m128i a, __m128i b, const int func)
• Sha256
  • Sha256MessageSchedule1 – __m128i _mm_sha256msg1_epu32 (__m128i a, __m128i b)
  • Sha256MessageSchedule2 – __m128i _mm_sha256msg2_epu32 (__m128i a, __m128i b)
  • Sha256TwoRounds – __m128i _mm_sha256rnds2_epu32 (__m128i a, __m128i b, __m128i k)

/cc @tannergooding

[ghost]

Note regarding the new-api-needs-documentation label:

This serves as a reminder for when your PR is modifying a ref *.cs file and adding/modifying public APIs, to please make sure the API implementation in the src *.cs file is documented with triple slash comments, so the PR reviewers can sign off that change.

[ghost]

I couldn't figure out the best area label to add to this PR. If you have write-permissions please help me learn by adding exactly one area label.

[ghost]

Tagging subscribers to this area: @dotnet/area-system-runtime-intrinsics
See info in area-owners.md if you want to be subscribed.

Issue Details

---

Proposal implementation of #256 (closes #256)

Proposal

namespaceSystem.Runtime.Intrinsics.X86{publicclassSha1{publicstaticboolIsSupported{get;}publicstaticVector128<byte>MessageSchedule1(Vector128<byte>a,Vector128<byte>b);publicstaticVector128<byte>MessageSchedule2(Vector128<byte>a,Vector128<byte>b);publicstaticVector128<byte>NextE(Vector128<byte>a,Vector128<byte>b);publicstaticVector128<byte>FourRounds(Vector128<byte>a,Vector128<byte>b,bytefunc);}publicclassSha256{publicstaticboolIsSupported{get;}publicstaticVector128<byte>MessageSchedule1(Vector128<byte>a,Vector128<byte>b);publicstaticVector128<byte>MessageSchedule2(Vector128<byte>a,Vector128<byte>b);publicstaticVector128<byte>TwoRounds(Vector128<byte>a,Vector128<byte>b,Vector128<byte>k);}}

Current state of implementation

• C++ CodeGen (JIT / NativeAOT) implementation (@tannergooding is this already implemented?)
• C# Implementation
  • Ref Assembly
  • Intrinsic Implementation
• Good documentation (suggestions?)

Used intrinsics

• Sha1
  • MessageSchedule1 – __m128i _mm_sha1msg1_epu32 (__m128i a, __m128i b)
  • MessageSchedule2 – __m128i _mm_sha1msg2_epu32 (__m128i a, __m128i b)
  • NextE – __m128i _mm_sha1nexte_epu32 (__m128i a, __m128i b)
  • FourRounds – __m128i _mm_sha1rnds4_epu32 (__m128i a, __m128i b, const int func)
• Sha256
  • MessageSchedule1 – __m128i _mm_sha256msg1_epu32 (__m128i a, __m128i b)
  • MessageSchedule2 – __m128i _mm_sha256msg2_epu32 (__m128i a, __m128i b)
  • TwoRounds – __m128i _mm_sha256rnds2_epu32 (__m128i a, __m128i b, __m128i k)

Tests

I think including tests for this is not relevant, since we trust the processor to execute the intrinsics correctly (there are only vector tests in System.Runtime.Intrinsics anyway).

/cc @tannergooding

Author:	deeprobin
Assignees:	-
Labels:	area-System.Runtime.Intrinsics, new-api-needs-documentation, community-contribution
Milestone:	-

[deeprobin]

I'm working on the JIT implementation for these intrinsics (see deeprobin#3). Happy to review 😄

[tannergooding]

I'm working on the JIT implementation for these intrinsics

Please let us know if you need any assistance here. There are a number of steps required to support new ISAs and you'll need to touch code in both the JIT and VM.

[deeprobin]

Please let us know if you need any assistance here. There are a number of steps required to support new ISAs and you'll need to touch code in both the JIT and VM.

Yes, I need some assistance.

Thats my current JIT change: deeprobin#3
I know that not everything will be right yet. However, it would be nice if you help me there a bit 😃 (What's missing, what's not quite right, ...).

[ghost]

Draft Pull Request was automatically closed for inactivity. Please let us know if you'd like to reopen it.

[deeprobin]

@tannergooding This was auto-closed. As soon as you have time to look at the bug in the wrong SHA encoding, feel free to reopen the PR :)