[Offload] Return error rather than dropping it

[RossBrunton]

No description provided.

[llvmbot]

@llvm/pr-subscribers-offload

Author: Ross Brunton (RossBrunton)

Changes

---

Full diff: https://github.com/llvm/llvm-project/pull/148609.diff

1 Files Affected:

• (modified) offload/plugins-nextgen/cuda/src/rtl.cpp (+1-1)

diff --git a/offload/plugins-nextgen/cuda/src/rtl.cpp b/offload/plugins-nextgen/cuda/src/rtl.cpp
index 15193de6ae430..b787376eb1770 100644
--- a/offload/plugins-nextgen/cuda/src/rtl.cpp
+++ b/offload/plugins-nextgen/cuda/src/rtl.cpp
@@ -1311,7 +1311,7 @@ Error CUDAKernelTy::launchImpl(GenericDeviceTy &GenericDevice,
   if (MaxDynCGroupMem >= MaxDynCGroupMemLimit) {
     CUresult AttrResult = cuFuncSetAttribute(
         Func, CU_FUNC_ATTRIBUTE_MAX_DYNAMIC_SHARED_SIZE_BYTES, MaxDynCGroupMem);
-    Plugin::check(
+    return Plugin::check(
         AttrResult,
         "Error in cuLaunchKernel while setting the memory limits: %s");
     MaxDynCGroupMemLimit = MaxDynCGroupMem;