[Clang][bytecode] Add interp__builtin_elementwise_triop to handle general 3-operand integer intrinsics #156944
+20
−31
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
…eral 3-operand integer intrinsics Refactor interp__builtin_elementwise_fsh into something similar to interp__builtin_elementwise_int_binop with a callback function argument to allow reuse with other intrinsics This will allow reuse with some upcoming x86 intrinsics
|
@llvm/pr-subscribers-clang Author: Simon Pilgrim (RKSimon) ChangesRefactor interp__builtin_elementwise_fsh into something similar to interp__builtin_elementwise_int_binop with a callback function argument to allow reuse with other intrinsics This will allow reuse with some upcoming x86 intrinsics We can flesh out handling for mixed vector/scalar args as the need arises Full diff: https://github.com/llvm/llvm-project/pull/156944.diff 1 Files Affected:
diff --git a/clang/lib/AST/ByteCode/InterpBuiltin.cpp b/clang/lib/AST/ByteCode/InterpBuiltin.cpp
index ff6ef5a1f6864..5713afb7162a8 100644
--- a/clang/lib/AST/ByteCode/InterpBuiltin.cpp
+++ b/clang/lib/AST/ByteCode/InterpBuiltin.cpp
@@ -2829,9 +2829,10 @@ static bool interp__builtin_select(InterpState &S, CodePtr OpPC,
return true;
}
-static bool interp__builtin_elementwise_fsh(InterpState &S, CodePtr OpPC,
- const CallExpr *Call,
- unsigned BuiltinID) {
+static bool interp__builtin_elementwise_triop(
+ InterpState &S, CodePtr OpPC, const CallExpr *Call, unsigned BuiltinID,
+ llvm::function_ref<APInt(const APSInt &, const APSInt &, const APSInt &)>
+ Fn) {
assert(Call->getNumArgs() == 3);
QualType Arg0Type = Call->getArg(0)->getType();
@@ -2840,17 +2841,10 @@ static bool interp__builtin_elementwise_fsh(InterpState &S, CodePtr OpPC,
// Non-vector integer types.
if (!Arg0Type->isVectorType()) {
- const APSInt &Shift =
- popToAPSInt(S.Stk, *S.getContext().classify(Arg2Type));
- const APSInt &Lo = popToAPSInt(S.Stk, *S.getContext().classify(Arg1Type));
- const APSInt &Hi = popToAPSInt(S.Stk, *S.getContext().classify(Arg0Type));
- APSInt Result;
- if (BuiltinID == Builtin::BI__builtin_elementwise_fshl)
- Result = APSInt(llvm::APIntOps::fshl(Hi, Lo, Shift), Hi.isUnsigned());
- else if (BuiltinID == Builtin::BI__builtin_elementwise_fshr)
- Result = APSInt(llvm::APIntOps::fshr(Hi, Lo, Shift), Hi.isUnsigned());
- else
- llvm_unreachable("Wrong builtin ID");
+ const APSInt &Op2 = popToAPSInt(S.Stk, *S.getContext().classify(Arg2Type));
+ const APSInt &Op1 = popToAPSInt(S.Stk, *S.getContext().classify(Arg1Type));
+ const APSInt &Op0 = popToAPSInt(S.Stk, *S.getContext().classify(Arg0Type));
+ APSInt Result = APSInt(Fn(Op0, Op1, Op2), Op0.isUnsigned());
pushInteger(S, Result, Call->getType());
return true;
}
@@ -2860,26 +2854,18 @@ static bool interp__builtin_elementwise_fsh(InterpState &S, CodePtr OpPC,
const PrimType &ElemT = *S.getContext().classify(VecT->getElementType());
unsigned NumElems = VecT->getNumElements();
- const Pointer &VecShift = S.Stk.pop<Pointer>();
- const Pointer &VecLo = S.Stk.pop<Pointer>();
- const Pointer &VecHi = S.Stk.pop<Pointer>();
+ const Pointer &Op2 = S.Stk.pop<Pointer>();
+ const Pointer &Op1 = S.Stk.pop<Pointer>();
+ const Pointer &Op0 = S.Stk.pop<Pointer>();
const Pointer &Dst = S.Stk.peek<Pointer>();
for (unsigned I = 0; I != NumElems; ++I) {
- APSInt Hi;
- APSInt Lo;
- APSInt Shift;
+ APSInt Val0, Val1, Val2;
INT_TYPE_SWITCH_NO_BOOL(ElemT, {
- Hi = VecHi.elem<T>(I).toAPSInt();
- Lo = VecLo.elem<T>(I).toAPSInt();
- Shift = VecShift.elem<T>(I).toAPSInt();
+ Val0 = Op0.elem<T>(I).toAPSInt();
+ Val1 = Op1.elem<T>(I).toAPSInt();
+ Val2 = Op2.elem<T>(I).toAPSInt();
});
- APSInt Result;
- if (BuiltinID == Builtin::BI__builtin_elementwise_fshl)
- Result = APSInt(llvm::APIntOps::fshl(Hi, Lo, Shift), Hi.isUnsigned());
- else if (BuiltinID == Builtin::BI__builtin_elementwise_fshr)
- Result = APSInt(llvm::APIntOps::fshr(Hi, Lo, Shift), Hi.isUnsigned());
- else
- llvm_unreachable("Wrong builtin ID");
+ APSInt Result = APSInt(Fn(Val0, Val1, Val2), Val0.isUnsigned());
INT_TYPE_SWITCH_NO_BOOL(ElemT,
{ Dst.elem<T>(I) = static_cast<T>(Result); });
}
@@ -3453,8 +3439,11 @@ bool InterpretBuiltin(InterpState &S, CodePtr OpPC, const CallExpr *Call,
return interp__builtin_select(S, OpPC, Call);
case Builtin::BI__builtin_elementwise_fshl:
+ return interp__builtin_elementwise_triop(S, OpPC, Call, BuiltinID,
+ llvm::APIntOps::fshl);
case Builtin::BI__builtin_elementwise_fshr:
- return interp__builtin_elementwise_fsh(S, OpPC, Call, BuiltinID);
+ return interp__builtin_elementwise_triop(S, OpPC, Call, BuiltinID,
+ llvm::APIntOps::fshr);
default:
S.FFDiag(S.Current->getLocation(OpPC),
|
tbaederr
reviewed
Sep 4, 2025
tbaederr
approved these changes
Sep 5, 2025
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Refactor interp__builtin_elementwise_fsh into something similar to interp__builtin_elementwise_int_binop with a callback function argument to allow reuse with other intrinsics
This will allow reuse with some upcoming x86 intrinsics
We can flesh out handling for mixed vector/scalar args as the need arises