[API Proposal]: Arm64: FEAT_SM4

[a74nh]

FEAT_SM4 provides instruction to accelerate the SM4 encryption algorithm, the standard Chinese encryption
algorithm. This set of instructions is described as the SM4 instructions.

The SM4 algorithm is 128-bit wide block cipher. The SM4E instruction accelerates a single round of encryption or decryption, and the SM4EKEY instruction accelerates a single
round of key generation:

namespace System.Runtime.Intrinsics.Arm;

/// VectorT Summary
public abstract partial class Sm4 : AdvSimd /// Feature: FEAT_SM4
{

 public static unsafe Vector128<uint> Sm4Encode(Vector128<uint> a, Vector128<uint> b); // SM4E

 public static unsafe Vector128<uint> Sm4KeyUpdate(Vector128<uint> a, Vector128<uint> b); // SM4EKEY

 /// total method signatures: 2

}

SM4 Encode takes input data as a 128-bit vector from the first source SIMD&FP register, and four iterations of the round key held as the elements of the 128-bit vector in the second source SIMD&FP register. It encrypts the data by four rounds, in accordance with the SM4 standard, returning the 128-bit result to the destination SIMD&FP register.

SM4 Key takes an input as a 128-bit vector from the first source SIMD&FP register and a 128-bit constant from the second SIMD&FP register. It derives four iterations of the output key, in accordance with the SM4 standard, returning the 128-bit result to the destination SIMD&FP register.

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

Issue Details

---

namespace System.Runtime.Intrinsics.Arm;

/// VectorT Summary
public abstract partial class Sm4 : AdvSimd /// Feature: FEAT_SM4
{

 public static unsafe Vector128<uint> Sm4EncryptionAndDecryption(Vector128<uint> a, Vector128<uint> b); // SM4E

 public static unsafe Vector128<uint> Sm4KeyUpdates(Vector128<uint> a, Vector128<uint> b); // SM4EKEY

 /// total method signatures: 2

}

Author:	a74nh
Assignees:	-
Labels:	area-System.Runtime.Intrinsics
Milestone:	-

[a74nh]

/// Full API
public abstract partial class Sm4 : AdvSimd /// Feature: FEAT_SM4
{
 /// Sm4EncryptionAndDecryption : SM4 Encode takes input data as a 128-bit vector from the first source SIMD&FP register, and four iterations of the round key held as the elements of the 128-bit vector in the second source SIMD&FP register. It encrypts the data by four rounds, in accordance with the SM4 standard, returning the 128-bit result to the destination SIMD&FP register.

 /// uint32x4_t vsm4eq_u32(uint32x4_t a, uint32x4_t b) : "SM4E Vd.4S,Vn.4S"
 public static unsafe Vector128<uint> Sm4EncryptionAndDecryption(Vector128<uint> a, Vector128<uint> b);


 /// Sm4KeyUpdates : SM4 Key takes an input as a 128-bit vector from the first source SIMD&FP register and a 128-bit constant from the second SIMD&FP register. It derives four iterations of the output key, in accordance with the SM4 standard, returning the 128-bit result to the destination SIMD&FP register.

 /// uint32x4_t vsm4ekeyq_u32(uint32x4_t a, uint32x4_t b) : "SM4EKEY Vd.4S,Vn.4S,Vm.4S"
 public static unsafe Vector128<uint> Sm4KeyUpdates(Vector128<uint> a, Vector128<uint> b);


 /// total method signatures: 2
 /// total method names: 2
}


 /// Total ACLE covered across API: 2

[a74nh]

This adds the API for FEAT_SM4. It complements the FEAT_SVE_SM4 API which is specified in #94426.

Note this is a NEON (AdvSimd) extension. It is not a SVE extension.

[a74nh]

@tannergooding : Also As requested, the API for SM4. Identical to the SVE version except for VectorT vs Vector128