Index: llvm/lib/Transforms/Utils/CloneFunction.cpp
===================================================================
--- llvm/lib/Transforms/Utils/CloneFunction.cpp
+++ llvm/lib/Transforms/Utils/CloneFunction.cpp
@@ -765,37 +765,38 @@
   DT->addNewBlock(NewPH, LoopDomBB);
 
   for (Loop *CurLoop : OrigLoop->getLoopsInPreorder()) {
-    for (BasicBlock *BB : CurLoop->getBlocks()) {
-      if (CurLoop != LI->getLoopFor(BB))
-        continue;
+    Loop *&NewLoop = LMap[CurLoop];
+    if (!NewLoop) {
+      NewLoop = LI->AllocateLoop();
 
-      Loop *&NewLoop = LMap[CurLoop];
-      if (!NewLoop) {
-        NewLoop = LI->AllocateLoop();
+      // Establish the parent/child relationship.
+      Loop *OrigParent = CurLoop->getParentLoop();
+      assert(OrigParent && "Could not find the original parent loop");
+      Loop *NewParentLoop = LMap[OrigParent];
+      assert(NewParentLoop && "Could not find the new parent loop");
 
-        // Establish the parent/child relationship.
-        Loop *OrigParent = CurLoop->getParentLoop();
-        assert(OrigParent && "Could not find the original parent loop");
-        Loop *NewParentLoop = LMap[OrigParent];
-        assert(NewParentLoop && "Could not find the new parent loop");
+      NewParentLoop->addChildLoop(NewLoop);
+    }
+  }
 
-        NewParentLoop->addChildLoop(NewLoop);
-      }
+  for (BasicBlock *BB : OrigLoop->getBlocks()) {
+    Loop *CurLoop = LI->getLoopFor(BB);
+    Loop *&NewLoop = LMap[CurLoop];
+    assert(NewLoop && "Expecting new loop to be allocated");
 
-      BasicBlock *NewBB = CloneBasicBlock(BB, VMap, NameSuffix, F);
-      VMap[BB] = NewBB;
+    BasicBlock *NewBB = CloneBasicBlock(BB, VMap, NameSuffix, F);
+    VMap[BB] = NewBB;
 
-      // Update LoopInfo.
-      NewLoop->addBasicBlockToLoop(NewBB, *LI);
-      if (BB == CurLoop->getHeader())
-        NewLoop->moveToHeader(NewBB);
+    // Update LoopInfo.
+    NewLoop->addBasicBlockToLoop(NewBB, *LI);
+    if (BB == CurLoop->getHeader())
+      NewLoop->moveToHeader(NewBB);
 
-      // Add DominatorTree node. After seeing all blocks, update to correct
-      // IDom.
-      DT->addNewBlock(NewBB, NewPH);
+    // Add DominatorTree node. After seeing all blocks, update to correct
+    // IDom.
+    DT->addNewBlock(NewBB, NewPH);
 
-      Blocks.push_back(NewBB);
-    }
+    Blocks.push_back(NewBB);
   }
 
   for (BasicBlock *BB : OrigLoop->getBlocks()) {
Index: llvm/unittests/Transforms/Utils/CloningTest.cpp
===================================================================
--- llvm/unittests/Transforms/Utils/CloningTest.cpp
+++ llvm/unittests/Transforms/Utils/CloningTest.cpp
@@ -10,6 +10,8 @@
 #include "llvm/ADT/STLExtras.h"
 #include "llvm/ADT/SmallPtrSet.h"
 #include "llvm/Analysis/DomTreeUpdater.h"
+#include "llvm/Analysis/LoopInfo.h"
+#include "llvm/AsmParser/Parser.h"
 #include "llvm/IR/Argument.h"
 #include "llvm/IR/Constant.h"
 #include "llvm/IR/DIBuilder.h"
@@ -355,6 +357,91 @@
   delete F;
 }
 
+static void runWithLoopInfoAndDominatorTree(
+    Module &M, StringRef FuncName,
+    function_ref<void(Function &F, LoopInfo &LI, DominatorTree &DT)> Test) {
+  auto *F = M.getFunction(FuncName);
+  ASSERT_NE(F, nullptr) << "Could not find " << FuncName;
+
+  DominatorTree DT(*F);
+  LoopInfo LI(DT);
+
+  Test(*F, LI, DT);
+}
+
+static std::unique_ptr<Module> parseIR(LLVMContext &C, const char *IR) {
+  SMDiagnostic Err;
+  std::unique_ptr<Module> Mod = parseAssemblyString(IR, Err, C);
+  if (!Mod)
+    Err.print("CloneLoop", errs());
+  return Mod;
+}
+
+TEST(CloneLoop, CloneLoopNest) {
+  // Parse the module.
+  LLVMContext Context;
+
+  std::unique_ptr<Module> M = parseIR(
+    Context,
+    R"(define void @foo(i32* %A, i32 %ub) {
+entry:
+  %guardcmp = icmp slt i32 0, %ub
+  br i1 %guardcmp, label %for.outer.preheader, label %for.end
+for.outer.preheader:
+  br label %for.outer
+for.outer:
+  %j = phi i32 [ 0, %for.outer.preheader ], [ %inc.outer, %for.outer.latch ]
+  br i1 %guardcmp, label %for.inner.preheader, label %for.outer.latch
+for.inner.preheader:
+  br label %for.inner
+for.inner:
+  %i = phi i32 [ 0, %for.inner.preheader ], [ %inc, %for.inner ]
+  %idxprom = sext i32 %i to i64
+  %arrayidx = getelementptr inbounds i32, i32* %A, i64 %idxprom
+  store i32 %i, i32* %arrayidx, align 4
+  %inc = add nsw i32 %i, 1
+  %cmp = icmp slt i32 %inc, %ub
+  br i1 %cmp, label %for.inner, label %for.inner.exit
+for.inner.exit:
+  br label %for.outer.latch
+for.outer.latch:
+  %inc.outer = add nsw i32 %j, 1
+  %cmp.outer = icmp slt i32 %inc.outer, %ub
+  br i1 %cmp.outer, label %for.outer, label %for.outer.exit
+for.outer.exit:
+  br label %for.end
+for.end:
+  ret void
+})"
+    );
+
+  runWithLoopInfoAndDominatorTree(
+      *M, "foo", [&](Function &F, LoopInfo &LI, DominatorTree &DT) {
+        Function::iterator FI = F.begin();
+        // First basic block is entry - skip it.
+        BasicBlock *Preheader = &*(++FI);
+        BasicBlock *Header = &*(++FI);
+        assert(Header->getName() == "for.outer");
+        Loop *L = LI.getLoopFor(Header);
+        EXPECT_NE(L, nullptr);
+        EXPECT_EQ(Header, L->getHeader());
+        EXPECT_EQ(Preheader, L->getLoopPreheader());
+
+        ValueToValueMapTy VMap;
+        SmallVector<BasicBlock *, 4> ClonedLoopBlocks;
+        Loop *NewLoop = cloneLoopWithPreheader(Preheader, Preheader, L, VMap,
+                                               "", &LI, &DT, ClonedLoopBlocks);
+        EXPECT_NE(NewLoop, nullptr);
+        EXPECT_EQ(NewLoop->getSubLoops().size(), 1u);
+        Loop::block_iterator BI = NewLoop->block_begin();
+        EXPECT_TRUE((*BI)->getName().startswith("for.outer"));
+        EXPECT_TRUE((*(++BI))->getName().startswith("for.inner.preheader"));
+        EXPECT_TRUE((*(++BI))->getName().startswith("for.inner"));
+        EXPECT_TRUE((*(++BI))->getName().startswith("for.inner.exit"));
+        EXPECT_TRUE((*(++BI))->getName().startswith("for.outer.latch"));
+      });
+}
+
 class CloneFunc : public ::testing::Test {
 protected:
   void SetUp() override {