[SVE2] CreateWhileWriteAfterRead* and SaturatingExtract*

[dhartglassMSFT]

Reamining missing sve2 mask instructions.

CreateWhileWriteAfterRead*:

/// T: float, double, sbyte, short, int, long, byte, ushort, uint, ulong
public static unsafe Vector<T> CreateWhileWriteAfterReadMask(T* left, T* right); // WHILEWR

Same as CreateWhileReadAfterWriteMask, except this doesn't use Abs(...) to compute ptrdiff. So behavior differs if op2ptr=op1ptr-1, added a test for that.

 

SaturatingExtract*:

/// T: [sbyte, short], [short, int], [int, long], [byte, ushort], [ushort, uint], [uint, ulong]
public static unsafe Vector<T> SaturatingExtractNarrowingLower(Vector<T2> value); // SQXTNB or UQXTNB

/// T: [sbyte, short], [short, int], [int, long], [byte, ushort], [ushort, uint], [uint, ulong]
public static unsafe Vector<T> SaturatingExtractNarrowingUpper(Vector<T> even, Vector<T2> op); // SQXTNT or UQXTNT

/// T: [byte, short], [ushort, int], [uint, long]
public static unsafe Vector<T> SaturatingExtractUnsignedNarrowingLower(Vector<T2> value); // SQXTUNB

/// T: [byte, short], [ushort, int], [uint, long]
public static unsafe Vector<T> SaturatingExtractUnsignedNarrowingUpper(Vector<T> even, Vector<T2> op); // SQXTUNT

Lower* variants are boilerplate. Upper variants take implicit out param, so it needs special RMW handling. We do similar things for other instructions like AddHighNarrowingOdd (INS_sve_addhnt)

[dotnet-policy-service]

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

[Copilot]

Pull request overview

Adds the SVE2 CreateWhileWriteAfterReadMask* (WHILEWR) intrinsic family alongside the existing CreateWhileReadAfterWriteMask* (WHILERW) family, and extends the shared test template to cover both, including a new "reverse offset by one" scenario that exercises the asymmetric behavior between the two instructions.

Changes:

• Adds 10 new CreateWhileWriteAfterReadMask{Type} public APIs (Sve2) in ref assembly, the throwing PlatformNotSupported stubs, and the recursive forwarding stubs in Sve2.cs.
• Wires up the new intrinsics in hwintrinsiclistarm64sve.h (mapping each base type to INS_sve_whilewr) and extends the EA_8BYTE pointer-operand codegen case in hwintrinsiccodegenarm64.cpp to cover the new IDs.
• Renames the shared SVE RW test template to SveCreateWhileDependencyMaskTest, parameterizes it with IsWriteAfterRead, adds a RunReverseOffsetByOneScenario distinguishing WHILEWR vs WHILERW, and adds 10 generator entries for the new WHILEWR tests.

Reviewed changes

Copilot reviewed 8 out of 8 changed files in this pull request and generated no comments.

Show a summary per file

File	Description
src/libraries/System.Runtime.Intrinsics/ref/System.Runtime.Intrinsics.cs	Adds 10 new ref APIs for CreateWhileWriteAfterReadMask*.
src/libraries/System.Private.CoreLib/src/System/Runtime/Intrinsics/Arm/Sve2.PlatformNotSupported.cs	Adds throwing stubs and XML docs for the new intrinsics.
src/libraries/System.Private.CoreLib/src/System/Runtime/Intrinsics/Arm/Sve2.cs	Adds the runtime intrinsic forwarding entry points.
src/coreclr/jit/hwintrinsiclistarm64sve.h	Registers the 10 new intrinsic IDs mapped to INS_sve_whilewr per base type.
src/coreclr/jit/hwintrinsiccodegenarm64.cpp	Routes the new IDs through the existing 64-bit pointer codegen path (comment updated to WHILERW/WHILEWR).
src/tests/Common/GenerateHWIntrinsicTests/Arm/Templates.cs	Renames the template config to SveCreateWhileDependencyMaskTest.
src/tests/Common/GenerateHWIntrinsicTests/Arm/Sve2Tests.cs	Updates RW entries with IsWriteAfterRead=false and adds 10 WHILEWR entries with IsWriteAfterRead=true.
src/tests/JIT/HardwareIntrinsics/Arm/Shared/SveCreateWhileDependencyMaskTest.template	Renames the test class, adds RunReverseOffsetByOneScenario, and documents the WHILEWR/WHILERW spec semantics.

[Copilot]

Pull request overview

Copilot reviewed 10 out of 10 changed files in this pull request and generated 6 comments.

[a74nh]

Other than the comments above, everything else LGTM