[llvm] [Offload][AMDGPU] Correctly handle variable implicit argument sizes (PR #142199)

[Joseph Huber via llvm-commits]

================
@@ -3365,7 +3363,7 @@ Error AMDGPUKernelTy::launchImpl(GenericDeviceTy &GenericDevice,
                                  KernelLaunchParamsTy LaunchParams,
                                  AsyncInfoWrapperTy &AsyncInfoWrapper) const {
   if (ArgsSize != LaunchParams.Size &&
-      ArgsSize != LaunchParams.Size + getImplicitArgsSize())
+      ArgsSize > LaunchParams.Size + getImplicitArgsSize())
     return Plugin::error(ErrorCode::INVALID_ARGUMENT,
                          "mismatch of kernel arguments size");
----------------
jhuber6 wrote:

done

https://github.com/llvm/llvm-project/pull/142199