================
@@ -14794,9 +14803,36 @@ void
ASTContext::getFunctionFeatureMap(llvm::StringMap<bool> &FeatureMap,
}
}
-static SYCLKernelInfo BuildSYCLKernelInfo(CanQualType KernelNameType,
+static SYCLKernelInfo BuildSYCLKernelInfo(ASTContext &Context,
+ CanQualType KernelNameType,
const FunctionDecl *FD) {
- return {KernelNameType, FD};
+ // Host and device compilation may use different ABIs and different ABIs
+ // may allocate name mangling discriminators differently. A discriminator
+ // override is used to ensure consistent discriminator allocation across
+ // host and device compilation.
+ auto DeviceDiscriminatorOverrider =
+ [](ASTContext &Ctx, const NamedDecl *ND) -> std::optional<unsigned> {
+ if (const auto *RD = dyn_cast<CXXRecordDecl>(ND))
+ if (RD->isLambda())
----------------
erichkeane wrote:
>Why do you believe that device mangling numbers are assigned differently for
>non-generic lambdas vs generic lambdas? I don't see any reason that they would
>be treated differently. Can you point to relevant code? The device mangling
>number applies to the (single) closure type produced by a (generic) lambda.
Just my memory of it when I was implementing `SYCLUniqueStableNameExpr` (or
actually, bugs found after and not fixed). I don't really recall where it
ended up though, so it seems worthy of investigation.
>Can you elaborate? I don't know what you're referring to. This code is nearly
>identical to SYCLUniqueStableNameExpr::ComputeName().
I mean in `TransfromLambdaExpr` and its descendants. So it would be there.
https://github.com/llvm/llvm-project/pull/133030
_______________________________________________
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
