Index: lib/Target/NVPTX/NVPTXISelLowering.cpp
===================================================================
--- lib/Target/NVPTX/NVPTXISelLowering.cpp
+++ lib/Target/NVPTX/NVPTXISelLowering.cpp
@@ -180,6 +180,17 @@
     return;
   }
 
+  // Given a struct type, recursively traverse the elements with custom ComputePTXValueVTs.
+  if (StructType *STy = dyn_cast<StructType>(Ty)) {
+    const StructLayout *SL = DL.getStructLayout(STy);
+    for (StructType::element_iterator EB = STy->element_begin(), EI = EB,
+                                      EE = STy->element_end();
+         EI != EE; ++EI)
+      ComputePTXValueVTs(TLI, DL, *EI, ValueVTs, Offsets,
+                         StartingOffset + SL->getElementOffset(EI - EB));
+    return;
+  }
+
   ComputeValueVTs(TLI, DL, Ty, TempVTs, &TempOffsets, StartingOffset);
   for (unsigned i = 0, e = TempVTs.size(); i != e; ++i) {
     EVT VT = TempVTs[i];
Index: test/CodeGen/NVPTX/i128-struct.ll
===================================================================
--- /dev/null
+++ test/CodeGen/NVPTX/i128-struct.ll
@@ -0,0 +1,13 @@
+; RUN: llc < %s -O0 -march=nvptx64 -mcpu=sm_20 | FileCheck %s
+
+; CHECK-LABEL: .visible .func (.param .align 16 .b8 func_retval0[32]) foo(
+define { i128, i128 } @foo(i64 %a, i32 %b) {
+  %1 = sext i64 %a to i128
+  %2 = sext i32 %b to i128
+  %3 = insertvalue { i128, i128 } undef, i128 %1, 0
+  %4 = insertvalue { i128, i128 } %3, i128 %2, 1
+
+  ; CHECK: st.param.v2.b64 [func_retval0+0],  {%[[REG1:rd[0-9]+]], %[[REG2:rd[0-9]+]]};
+  ; CHECK: st.param.v2.b64 [func_retval0+16], {%[[REG3:rd[0-9]+]], %[[REG4:rd[0-9]+]]};
+  ret { i128, i128 } %4
+}