diff --git a/llvm/lib/Target/NVPTX/NVPTXISelLowering.cpp b/llvm/lib/Target/NVPTX/NVPTXISelLowering.cpp
--- a/llvm/lib/Target/NVPTX/NVPTXISelLowering.cpp
+++ b/llvm/lib/Target/NVPTX/NVPTXISelLowering.cpp
@@ -4273,6 +4273,11 @@
         assert(MDN->getNumOperands() == 3);
 
         const Metadata *MD0 = MDN->getOperand(0).get();
+        // Allow degenerated annotations created by OpenMP offloading with old
+        // CUDA versions.
+        if (!MD0)
+          continue;
+
         const auto *MDV0 = cast<ConstantAsMetadata>(MD0)->getValue();
         const auto *MDFn = cast<Function>(MDV0);
         if (MDFn != F)
diff --git a/llvm/test/CodeGen/NVPTX/param-vectorize-device.ll b/llvm/test/CodeGen/NVPTX/param-vectorize-device.ll
--- a/llvm/test/CodeGen/NVPTX/param-vectorize-device.ll
+++ b/llvm/test/CodeGen/NVPTX/param-vectorize-device.ll
@@ -799,3 +799,7 @@
   %oldret = insertvalue [4 x i32] %7, i32 %4, 3
   ret [4 x i32] %oldret
 }
+
+!nvvm.annotations = !{!0}
+
+!0 = !{null, !"foobar", i32 0}