diff --git a/llvm/include/llvm/Transforms/Scalar/LoopUnrollAndJamPass.h b/llvm/include/llvm/Transforms/Scalar/LoopUnrollAndJamPass.h
--- a/llvm/include/llvm/Transforms/Scalar/LoopUnrollAndJamPass.h
+++ b/llvm/include/llvm/Transforms/Scalar/LoopUnrollAndJamPass.h
@@ -15,9 +15,7 @@
 
 namespace llvm {
 
-class Loop;
-struct LoopStandardAnalysisResults;
-class LPMUpdater;
+class Function;
 
 /// A simple loop rotation transformation.
 class LoopUnrollAndJamPass : public PassInfoMixin<LoopUnrollAndJamPass> {
@@ -25,8 +23,7 @@
 
 public:
   explicit LoopUnrollAndJamPass(int OptLevel = 2) : OptLevel(OptLevel) {}
-  PreservedAnalyses run(Loop &L, LoopAnalysisManager &AM,
-                        LoopStandardAnalysisResults &AR, LPMUpdater &U);
+  PreservedAnalyses run(Function &F, FunctionAnalysisManager &AM);
 };
 
 } // end namespace llvm
diff --git a/llvm/lib/Passes/PassBuilder.cpp b/llvm/lib/Passes/PassBuilder.cpp
--- a/llvm/lib/Passes/PassBuilder.cpp
+++ b/llvm/lib/Passes/PassBuilder.cpp
@@ -970,8 +970,7 @@
   // across the loop nests.
   // We do UnrollAndJam in a separate LPM to ensure it happens before unroll
   if (EnableUnrollAndJam && PTO.LoopUnrolling) {
-    OptimizePM.addPass(
-        createFunctionToLoopPassAdaptor(LoopUnrollAndJamPass(Level)));
+    OptimizePM.addPass(LoopUnrollAndJamPass(Level));
   }
   OptimizePM.addPass(LoopUnrollPass(
       LoopUnrollOptions(Level, /*OnlyWhenForced=*/!PTO.LoopUnrolling,
diff --git a/llvm/lib/Passes/PassRegistry.def b/llvm/lib/Passes/PassRegistry.def
--- a/llvm/lib/Passes/PassRegistry.def
+++ b/llvm/lib/Passes/PassRegistry.def
@@ -235,6 +235,7 @@
 FUNCTION_PASS("sroa", SROA())
 FUNCTION_PASS("tailcallelim", TailCallElimPass())
 FUNCTION_PASS("unreachableblockelim", UnreachableBlockElimPass())
+FUNCTION_PASS("unroll-and-jam", LoopUnrollAndJamPass())
 FUNCTION_PASS("verify", VerifierPass())
 FUNCTION_PASS("verify<domtree>", DominatorTreeVerifierPass())
 FUNCTION_PASS("verify<loops>", LoopVerifierPass())
@@ -307,7 +308,6 @@
 LOOP_PASS("strength-reduce", LoopStrengthReducePass())
 LOOP_PASS("indvars", IndVarSimplifyPass())
 LOOP_PASS("irce", IRCEPass())
-LOOP_PASS("unroll-and-jam", LoopUnrollAndJamPass())
 LOOP_PASS("unroll-full", LoopFullUnrollPass())
 LOOP_PASS("print-access-info", LoopAccessInfoPrinterPass(dbgs()))
 LOOP_PASS("print<ddg>", DDGAnalysisPrinterPass(dbgs()))
diff --git a/llvm/lib/Transforms/Scalar/LoopUnrollAndJamPass.cpp b/llvm/lib/Transforms/Scalar/LoopUnrollAndJamPass.cpp
--- a/llvm/lib/Transforms/Scalar/LoopUnrollAndJamPass.cpp
+++ b/llvm/lib/Transforms/Scalar/LoopUnrollAndJamPass.cpp
@@ -427,51 +427,64 @@
   return UnrollResult;
 }
 
+static bool tryToUnrollAndJamLoop(Function &F, DominatorTree &DT, LoopInfo &LI,
+                                  ScalarEvolution &SE,
+                                  const TargetTransformInfo &TTI,
+                                  AssumptionCache &AC, DependenceInfo &DI,
+                                  OptimizationRemarkEmitter &ORE,
+                                  int OptLevel) {
+  SmallPriorityWorklist<Loop *, 4> Worklist;
+  internal::appendLoopsToWorklist(reverse(LI), Worklist);
+  bool DidSomething = false;
+  do {
+    Loop *L = Worklist.pop_back_val();
+    formLCSSA(*L, DT, &LI, &SE);
+    LoopUnrollResult Result =
+        tryToUnrollAndJamLoop(L, DT, &LI, SE, TTI, AC, DI, ORE, OptLevel);
+    if (Result != LoopUnrollResult::Unmodified)
+      DidSomething = true;
+  } while (!Worklist.empty());
+
+  return DidSomething;
+}
+
 namespace {
 
-class LoopUnrollAndJam : public LoopPass {
+class LoopUnrollAndJam : public FunctionPass {
 public:
   static char ID; // Pass ID, replacement for typeid
   unsigned OptLevel;
 
-  LoopUnrollAndJam(int OptLevel = 2) : LoopPass(ID), OptLevel(OptLevel) {
+  LoopUnrollAndJam(int OptLevel = 2) : FunctionPass(ID), OptLevel(OptLevel) {
     initializeLoopUnrollAndJamPass(*PassRegistry::getPassRegistry());
   }
 
-  bool runOnLoop(Loop *L, LPPassManager &LPM) override {
-    if (skipLoop(L))
+  bool runOnFunction(Function &F) override {
+    if (skipFunction(F))
       return false;
 
-    Function &F = *L->getHeader()->getParent();
-
     auto &DT = getAnalysis<DominatorTreeWrapperPass>().getDomTree();
-    LoopInfo *LI = &getAnalysis<LoopInfoWrapperPass>().getLoopInfo();
+    LoopInfo &LI = getAnalysis<LoopInfoWrapperPass>().getLoopInfo();
     ScalarEvolution &SE = getAnalysis<ScalarEvolutionWrapperPass>().getSE();
     const TargetTransformInfo &TTI =
         getAnalysis<TargetTransformInfoWrapperPass>().getTTI(F);
     auto &AC = getAnalysis<AssumptionCacheTracker>().getAssumptionCache(F);
     auto &DI = getAnalysis<DependenceAnalysisWrapperPass>().getDI();
-    // For the old PM, we can't use OptimizationRemarkEmitter as an analysis
-    // pass.  Function analyses need to be preserved across loop transformations
-    // but ORE cannot be preserved (see comment before the pass definition).
-    OptimizationRemarkEmitter ORE(&F);
-
-    LoopUnrollResult Result =
-        tryToUnrollAndJamLoop(L, DT, LI, SE, TTI, AC, DI, ORE, OptLevel);
+    auto &ORE = getAnalysis<OptimizationRemarkEmitterWrapperPass>().getORE();
 
-    if (Result == LoopUnrollResult::FullyUnrolled)
-      LPM.markLoopAsDeleted(*L);
-
-    return Result != LoopUnrollResult::Unmodified;
+    return tryToUnrollAndJamLoop(F, DT, LI, SE, TTI, AC, DI, ORE, OptLevel);
   }
 
   /// This transformation requires natural loop information & requires that
   /// loop preheaders be inserted into the CFG...
   void getAnalysisUsage(AnalysisUsage &AU) const override {
-    AU.addRequired<AssumptionCacheTracker>();
+    AU.addRequired<DominatorTreeWrapperPass>();
+    AU.addRequired<LoopInfoWrapperPass>();
+    AU.addRequired<ScalarEvolutionWrapperPass>();
     AU.addRequired<TargetTransformInfoWrapperPass>();
+    AU.addRequired<AssumptionCacheTracker>();
     AU.addRequired<DependenceAnalysisWrapperPass>();
-    getLoopAnalysisUsage(AU);
+    AU.addRequired<OptimizationRemarkEmitterWrapperPass>();
   }
 };
 
@@ -492,26 +505,18 @@
   return new LoopUnrollAndJam(OptLevel);
 }
 
-PreservedAnalyses LoopUnrollAndJamPass::run(Loop &L, LoopAnalysisManager &AM,
-                                            LoopStandardAnalysisResults &AR,
-                                            LPMUpdater &) {
-  const auto &FAM =
-      AM.getResult<FunctionAnalysisManagerLoopProxy>(L, AR).getManager();
-  Function *F = L.getHeader()->getParent();
-
-  auto *ORE = FAM.getCachedResult<OptimizationRemarkEmitterAnalysis>(*F);
-  // FIXME: This should probably be optional rather than required.
-  if (!ORE)
-    report_fatal_error(
-        "LoopUnrollAndJamPass: OptimizationRemarkEmitterAnalysis not cached at "
-        "a higher level");
-
-  DependenceInfo DI(F, &AR.AA, &AR.SE, &AR.LI);
-
-  LoopUnrollResult Result = tryToUnrollAndJamLoop(
-      &L, AR.DT, &AR.LI, AR.SE, AR.TTI, AR.AC, DI, *ORE, OptLevel);
-
-  if (Result == LoopUnrollResult::Unmodified)
+PreservedAnalyses LoopUnrollAndJamPass::run(Function &F,
+                                            FunctionAnalysisManager &AM) {
+  ScalarEvolution &SE = AM.getResult<ScalarEvolutionAnalysis>(F);
+  LoopInfo &LI = AM.getResult<LoopAnalysis>(F);
+  TargetTransformInfo &TTI = AM.getResult<TargetIRAnalysis>(F);
+  AssumptionCache &AC = AM.getResult<AssumptionAnalysis>(F);
+  DominatorTree &DT = AM.getResult<DominatorTreeAnalysis>(F);
+  DependenceInfo &DI = AM.getResult<DependenceAnalysis>(F);
+  OptimizationRemarkEmitter &ORE =
+      AM.getResult<OptimizationRemarkEmitterAnalysis>(F);
+
+  if (!tryToUnrollAndJamLoop(F, DT, LI, SE, TTI, AC, DI, ORE, OptLevel))
     return PreservedAnalyses::all();
 
   return getLoopPassPreservedAnalyses();
diff --git a/llvm/test/Transforms/LoopUnrollAndJam/dependencies.ll b/llvm/test/Transforms/LoopUnrollAndJam/dependencies.ll
--- a/llvm/test/Transforms/LoopUnrollAndJam/dependencies.ll
+++ b/llvm/test/Transforms/LoopUnrollAndJam/dependencies.ll
@@ -10,10 +10,13 @@
 define void @fore_aft_less(i32* noalias nocapture %A, i32 %N, i32* noalias nocapture readonly %B) {
 entry:
   %cmp = icmp sgt i32 %N, 0
-  br i1 %cmp, label %for.outer, label %cleanup
+  br i1 %cmp, label %for.outer.preheader, label %cleanup
+
+for.outer.preheader:
+  br label %for.outer
 
 for.outer:
-  %i = phi i32 [ %add7, %for.latch ], [ 0, %entry ]
+  %i = phi i32 [ %add7, %for.latch ], [ 0, %for.outer.preheader ]
   %arrayidx = getelementptr inbounds i32, i32* %A, i32 %i
   store i32 1, i32* %arrayidx, align 4
   br label %for.inner
@@ -35,7 +38,10 @@
   %arrayidx8 = getelementptr inbounds i32, i32* %A, i32 %add72
   store i32 %add, i32* %arrayidx8, align 4
   %exitcond29 = icmp eq i32 %add7, %N
-  br i1 %exitcond29, label %cleanup, label %for.outer
+  br i1 %exitcond29, label %cleanup.loopexit, label %for.outer
+
+cleanup.loopexit:
+  br label %cleanup
 
 cleanup:
   ret void
@@ -50,10 +56,13 @@
 define void @fore_aft_eq(i32* noalias nocapture %A, i32 %N, i32* noalias nocapture readonly %B) {
 entry:
   %cmp = icmp sgt i32 %N, 0
-  br i1 %cmp, label %for.outer, label %cleanup
+  br i1 %cmp, label %for.outer.preheader, label %cleanup
+
+for.outer.preheader:
+  br label %for.outer
 
 for.outer:
-  %i = phi i32 [ %add7, %for.latch ], [ 0, %entry ]
+  %i = phi i32 [ %add7, %for.latch ], [ 0, %for.outer.preheader ]
   %arrayidx = getelementptr inbounds i32, i32* %A, i32 %i
   store i32 1, i32* %arrayidx, align 4
   br label %for.inner
@@ -75,7 +84,10 @@
   %arrayidx8 = getelementptr inbounds i32, i32* %A, i32 %i
   store i32 %add, i32* %arrayidx8, align 4
   %exitcond29 = icmp eq i32 %add7, %N
-  br i1 %exitcond29, label %cleanup, label %for.outer
+  br i1 %exitcond29, label %cleanup.loopexit, label %for.outer
+
+cleanup.loopexit:
+  br label %cleanup
 
 cleanup:
   ret void
@@ -88,10 +100,13 @@
 define void @fore_aft_more(i32* noalias nocapture %A, i32 %N, i32* noalias nocapture readonly %B) {
 entry:
   %cmp = icmp sgt i32 %N, 0
-  br i1 %cmp, label %for.outer, label %cleanup
+  br i1 %cmp, label %for.outer.preheader, label %cleanup
+
+for.outer.preheader:
+  br label %for.outer
 
 for.outer:
-  %i = phi i32 [ %add7, %for.latch ], [ 0, %entry ]
+  %i = phi i32 [ %add7, %for.latch ], [ 0, %for.outer.preheader ]
   %arrayidx = getelementptr inbounds i32, i32* %A, i32 %i
   store i32 1, i32* %arrayidx, align 4
   br label %for.inner
@@ -113,7 +128,10 @@
   %arrayidx8 = getelementptr inbounds i32, i32* %A, i32 %add72
   store i32 %add, i32* %arrayidx8, align 4
   %exitcond29 = icmp eq i32 %add7, %N
-  br i1 %exitcond29, label %cleanup, label %for.outer
+  br i1 %exitcond29, label %cleanup.loopexit, label %for.outer
+
+cleanup.loopexit:
+  br label %cleanup
 
 cleanup:
   ret void
@@ -128,10 +146,13 @@
 define void @fore_sub_less(i32* noalias nocapture %A, i32 %N, i32* noalias nocapture readonly %B) {
 entry:
   %cmp = icmp sgt i32 %N, 0
-  br i1 %cmp, label %for.outer, label %cleanup
+  br i1 %cmp, label %for.outer.preheader, label %cleanup
+
+for.outer.preheader:
+  br label %for.outer
 
 for.outer:
-  %i = phi i32 [ %add7, %for.latch ], [ 0, %entry ]
+  %i = phi i32 [ %add7, %for.latch ], [ 0, %for.outer.preheader ]
   %arrayidx = getelementptr inbounds i32, i32* %A, i32 %i
   store i32 1, i32* %arrayidx, align 4
   br label %for.inner
@@ -153,7 +174,10 @@
 for.latch:
   %add7 = add nuw nsw i32 %i, 1
   %exitcond29 = icmp eq i32 %add7, %N
-  br i1 %exitcond29, label %cleanup, label %for.outer
+  br i1 %exitcond29, label %cleanup.loopexit, label %for.outer
+
+cleanup.loopexit:
+  br label %cleanup
 
 cleanup:
   ret void
@@ -168,10 +192,13 @@
 define void @fore_sub_eq(i32* noalias nocapture %A, i32 %N, i32* noalias nocapture readonly %B) {
 entry:
   %cmp = icmp sgt i32 %N, 0
-  br i1 %cmp, label %for.outer, label %cleanup
+  br i1 %cmp, label %for.outer.preheader, label %cleanup
+
+for.outer.preheader:
+  br label %for.outer
 
 for.outer:
-  %i = phi i32 [ %add7, %for.latch ], [ 0, %entry ]
+  %i = phi i32 [ %add7, %for.latch ], [ 0, %for.outer.preheader ]
   %arrayidx = getelementptr inbounds i32, i32* %A, i32 %i
   store i32 1, i32* %arrayidx, align 4
   br label %for.inner
@@ -193,7 +220,10 @@
 for.latch:
   %add7 = add nuw nsw i32 %i, 1
   %exitcond29 = icmp eq i32 %add7, %N
-  br i1 %exitcond29, label %cleanup, label %for.outer
+  br i1 %exitcond29, label %cleanup.loopexit, label %for.outer
+
+cleanup.loopexit:
+  br label %cleanup
 
 cleanup:
   ret void
@@ -206,10 +236,13 @@
 define void @fore_sub_more(i32* noalias nocapture %A, i32 %N, i32* noalias nocapture readonly %B) {
 entry:
   %cmp = icmp sgt i32 %N, 0
-  br i1 %cmp, label %for.outer, label %cleanup
+  br i1 %cmp, label %for.outer.preheader, label %cleanup
+
+for.outer.preheader:
+  br label %for.outer
 
 for.outer:
-  %i = phi i32 [ %add7, %for.latch ], [ 0, %entry ]
+  %i = phi i32 [ %add7, %for.latch ], [ 0, %for.outer.preheader ]
   %arrayidx = getelementptr inbounds i32, i32* %A, i32 %i
   store i32 1, i32* %arrayidx, align 4
   br label %for.inner
@@ -231,7 +264,10 @@
 for.latch:
   %add7 = add nuw nsw i32 %i, 1
   %exitcond29 = icmp eq i32 %add7, %N
-  br i1 %exitcond29, label %cleanup, label %for.outer
+  br i1 %exitcond29, label %cleanup.loopexit, label %for.outer
+
+cleanup.loopexit:
+  br label %cleanup
 
 cleanup:
   ret void
@@ -246,10 +282,13 @@
 define void @sub_aft_less(i32* noalias nocapture %A, i32 %N, i32* noalias nocapture readonly %B) {
 entry:
   %cmp = icmp sgt i32 %N, 0
-  br i1 %cmp, label %for.outer, label %cleanup
+  br i1 %cmp, label %for.outer.preheader, label %cleanup
+
+for.outer.preheader:
+  br label %for.outer
 
 for.outer:
-  %i = phi i32 [ %add7, %for.latch ], [ 0, %entry ]
+  %i = phi i32 [ %add7, %for.latch ], [ 0, %for.outer.preheader ]
   br label %for.inner
 
 for.inner:
@@ -271,7 +310,10 @@
   %arrayidx8 = getelementptr inbounds i32, i32* %A, i32 %add72
   store i32 %add, i32* %arrayidx8, align 4
   %exitcond29 = icmp eq i32 %add7, %N
-  br i1 %exitcond29, label %cleanup, label %for.outer
+  br i1 %exitcond29, label %cleanup.loopexit, label %for.outer
+
+cleanup.loopexit:
+  br label %cleanup
 
 cleanup:
   ret void
@@ -286,10 +328,13 @@
 define void @sub_aft_eq(i32* noalias nocapture %A, i32 %N, i32* noalias nocapture readonly %B) {
 entry:
   %cmp = icmp sgt i32 %N, 0
-  br i1 %cmp, label %for.outer, label %cleanup
+  br i1 %cmp, label %for.outer.preheader, label %cleanup
+
+for.outer.preheader:
+  br label %for.outer
 
 for.outer:
-  %i = phi i32 [ %add7, %for.latch ], [ 0, %entry ]
+  %i = phi i32 [ %add7, %for.latch ], [ 0, %for.outer.preheader ]
   br label %for.inner
 
 for.inner:
@@ -311,7 +356,10 @@
   %arrayidx8 = getelementptr inbounds i32, i32* %A, i32 %i
   store i32 %add, i32* %arrayidx8, align 4
   %exitcond29 = icmp eq i32 %add7, %N
-  br i1 %exitcond29, label %cleanup, label %for.outer
+  br i1 %exitcond29, label %cleanup.loopexit, label %for.outer
+
+cleanup.loopexit:
+  br label %cleanup
 
 cleanup:
   ret void
@@ -324,10 +372,13 @@
 define void @sub_aft_more(i32* noalias nocapture %A, i32 %N, i32* noalias nocapture readonly %B) {
 entry:
   %cmp = icmp sgt i32 %N, 0
-  br i1 %cmp, label %for.outer, label %cleanup
+  br i1 %cmp, label %for.outer.preheader, label %cleanup
+
+for.outer.preheader:
+  br label %for.outer
 
 for.outer:
-  %i = phi i32 [ %add7, %for.latch ], [ 0, %entry ]
+  %i = phi i32 [ %add7, %for.latch ], [ 0, %for.outer.preheader ]
   br label %for.inner
 
 for.inner:
@@ -349,7 +400,10 @@
   %arrayidx8 = getelementptr inbounds i32, i32* %A, i32 %add72
   store i32 %add, i32* %arrayidx8, align 4
   %exitcond29 = icmp eq i32 %add7, %N
-  br i1 %exitcond29, label %cleanup, label %for.outer
+  br i1 %exitcond29, label %cleanup.loopexit, label %for.outer
+
+cleanup.loopexit:
+  br label %cleanup
 
 cleanup:
   ret void
@@ -362,10 +416,13 @@
 define void @sub_sub_less(i32* noalias nocapture %A, i32 %N, i32* noalias nocapture readonly %B) {
 entry:
   %cmp = icmp sgt i32 %N, 0
-  br i1 %cmp, label %for.outer, label %cleanup
+  br i1 %cmp, label %for.outer.preheader, label %cleanup
+
+for.outer.preheader:
+  br label %for.outer
 
 for.outer:
-  %i = phi i32 [ %add7, %for.latch ], [ 0, %entry ]
+  %i = phi i32 [ %add7, %for.latch ], [ 0, %for.outer.preheader ]
   br label %for.inner
 
 for.inner:
@@ -387,7 +444,10 @@
 for.latch:
   %add7 = add nuw nsw i32 %i, 1
   %exitcond29 = icmp eq i32 %add7, %N
-  br i1 %exitcond29, label %cleanup, label %for.outer
+  br i1 %exitcond29, label %cleanup.loopexit, label %for.outer
+
+cleanup.loopexit:
+  br label %cleanup
 
 cleanup:
   ret void
@@ -400,10 +460,13 @@
 define void @sub_sub_eq(i32* noalias nocapture %A, i32 %N, i32* noalias nocapture readonly %B) {
 entry:
   %cmp = icmp sgt i32 %N, 0
-  br i1 %cmp, label %for.outer, label %cleanup
+  br i1 %cmp, label %for.outer.preheader, label %cleanup
+
+for.outer.preheader:
+  br label %for.outer
 
 for.outer:
-  %i = phi i32 [ %add7, %for.latch ], [ 0, %entry ]
+  %i = phi i32 [ %add7, %for.latch ], [ 0, %for.outer.preheader ]
   br label %for.inner
 
 for.inner:
@@ -425,7 +488,10 @@
 for.latch:
   %add7 = add nuw nsw i32 %i, 1
   %exitcond29 = icmp eq i32 %add7, %N
-  br i1 %exitcond29, label %cleanup, label %for.outer
+  br i1 %exitcond29, label %cleanup.loopexit, label %for.outer
+
+cleanup.loopexit:
+  br label %cleanup
 
 cleanup:
   ret void
@@ -438,10 +504,13 @@
 define void @sub_sub_more(i32* noalias nocapture %A, i32 %N, i32* noalias nocapture readonly %B) {
 entry:
   %cmp = icmp sgt i32 %N, 0
-  br i1 %cmp, label %for.outer, label %cleanup
+  br i1 %cmp, label %for.outer.preheader, label %cleanup
+
+for.outer.preheader:
+  br label %for.outer
 
 for.outer:
-  %i = phi i32 [ %add7, %for.latch ], [ 0, %entry ]
+  %i = phi i32 [ %add7, %for.latch ], [ 0, %for.outer.preheader ]
   br label %for.inner
 
 for.inner:
@@ -463,7 +532,10 @@
 for.latch:
   %add7 = add nuw nsw i32 %i, 1
   %exitcond29 = icmp eq i32 %add7, %N
-  br i1 %exitcond29, label %cleanup, label %for.outer
+  br i1 %exitcond29, label %cleanup.loopexit, label %for.outer
+
+cleanup.loopexit:
+  br label %cleanup
 
 cleanup:
   ret void
diff --git a/llvm/test/Transforms/LoopUnrollAndJam/disable.ll b/llvm/test/Transforms/LoopUnrollAndJam/disable.ll
--- a/llvm/test/Transforms/LoopUnrollAndJam/disable.ll
+++ b/llvm/test/Transforms/LoopUnrollAndJam/disable.ll
@@ -44,7 +44,10 @@
   %inc8 = add nuw nsw i32 %b.028, 1
   %add10 = add nuw nsw i32 %i.029, 1
   %exitcond30 = icmp eq i32 %add10, %I
-  br i1 %exitcond30, label %return, label %for.outer
+  br i1 %exitcond30, label %return.loopexit, label %for.outer
+
+return.loopexit:
+  br label %return
 
 return:
   ret void
@@ -71,11 +74,14 @@
   %arrayidx = getelementptr inbounds i32, i32* %B, i32 %i.032
   %0 = load i32, i32* %arrayidx, align 4
   %tobool = icmp eq i32 %0, 0
-  br i1 %tobool, label %for.latch, label %for.inner
+  br i1 %tobool, label %for.latch, label %for.inner.preheader
+
+for.inner.preheader:
+  br label %for.inner
 
 for.inner:
-  %j.030 = phi i32 [ %inc, %for.inner ], [ 0, %for.outer ]
-  %sum1.029 = phi i32 [ %sum1.1, %for.inner ], [ 0, %for.outer ]
+  %j.030 = phi i32 [ %inc, %for.inner ], [ 0, %for.inner.preheader ]
+  %sum1.029 = phi i32 [ %sum1.1, %for.inner ], [ 0, %for.inner.preheader ]
   %arrayidx6 = getelementptr inbounds i32, i32* %B, i32 %j.030
   %1 = load i32, i32* %arrayidx6, align 4
   %tobool7 = icmp eq i32 %1, 0
@@ -84,15 +90,21 @@
   %sum1.1 = select i1 %tobool7, i32 %sum1.029, i32 %add
   %inc = add nuw i32 %j.030, 1
   %exitcond = icmp eq i32 %inc, %J
-  br i1 %exitcond, label %for.latch, label %for.inner
+  br i1 %exitcond, label %for.latch.loopexit, label %for.inner
+
+for.latch.loopexit:
+  br label %for.latch
 
 for.latch:
-  %sum1.1.lcssa = phi i32 [ 0, %for.outer ], [ %sum1.1, %for.inner ]
+  %sum1.1.lcssa = phi i32 [ 0, %for.outer ], [ %sum1.1, %for.latch.loopexit ]
   %arrayidx11 = getelementptr inbounds i32, i32* %A, i32 %i.032
   store i32 %sum1.1.lcssa, i32* %arrayidx11, align 4
   %add13 = add nuw i32 %i.032, 1
   %exitcond33 = icmp eq i32 %add13, %I
-  br i1 %exitcond33, label %for.end14, label %for.outer
+  br i1 %exitcond33, label %for.end14.loopexit, label %for.outer
+
+for.end14.loopexit:
+  br label %for.end14
 
 for.end14:
   ret void
@@ -190,6 +202,9 @@
   %exitcond24 = icmp eq i32 %indvars.iv.next, %I
   br i1 %exitcond24, label %for.end9, label %for.outer
 
+for.end9.loopexit:
+  br label %for.end9
+
 for.end9:
   ret void
 }
@@ -306,6 +321,9 @@
   %exitcond = icmp eq i32 %add9, %J
   br i1 %exitcond, label %for.cond3.for.cond.cleanup5_crit_edge, label %for.body6
 
+for.end12.loopexit:
+  br label %for.end12
+
 for.end12:
   ret void
 }
@@ -359,6 +377,9 @@
   %exitcond = icmp eq i32 %add13, %J
   br i1 %exitcond, label %for.latch, label %for.inner
 
+for.end18.loopexit:
+  br label %for.end18
+
 for.end18:
   ret void
 }
@@ -576,11 +597,14 @@
   %add = add i32 %0, %sum1
   %inc = add nuw i32 %j, 1
   %exitcond = icmp eq i32 %inc, %J
-  br i1 %exitcond, label %for.inner2, label %for.inner
+  br i1 %exitcond, label %for.inner2.preheader, label %for.inner
+
+for.inner2.preheader:
+  br label %for.inner2
 
 for.inner2:
-  %j2 = phi i32 [ 0, %for.inner ], [ %inc2, %for.inner2 ]
-  %sum12 = phi i32 [ 0, %for.inner ], [ %add2, %for.inner2 ]
+  %j2 = phi i32 [ %inc2, %for.inner2 ], [ 0, %for.inner2.preheader ]
+  %sum12 = phi i32 [ %add2, %for.inner2 ], [ 0, %for.inner2.preheader ]
   %arrayidx2 = getelementptr inbounds i32, i32* %B, i32 %j2
   %l0 = load i32, i32* %arrayidx2, align 4
   %add2 = add i32 %l0, %sum12
@@ -622,11 +646,14 @@
   %i = phi i32 [ %add8, %for.latch ], [ 0, %for.outer.preheader ]
   %add8 = add nuw i32 %i, 1
   %exitcond23 = icmp eq i32 %add8, %I
-  br i1 %exitcond23, label %for.end.loopexit, label %for.inner
+  br i1 %exitcond23, label %for.end.loopexit, label %for.inner.preheader
+
+for.inner.preheader:
+  br label %for.inner
 
 for.inner:
-  %j = phi i32 [ 0, %for.outer ], [ %inc, %for.inner ]
-  %sum1 = phi i32 [ 0, %for.outer ], [ %add, %for.inner ]
+  %j = phi i32 [ %inc, %for.inner ], [ 0, %for.inner.preheader ]
+  %sum1 = phi i32 [ %add, %for.inner ], [ 0, %for.inner.preheader ]
   %arrayidx = getelementptr inbounds i32, i32* %B, i32 %j
   %0 = load i32, i32* %arrayidx, align 4
   %add = add i32 %0, %sum1
@@ -667,11 +694,14 @@
   %i = phi i32 [ %add8, %for.latch ], [ 0, %for.outer.preheader ]
   %add8 = add nuw i32 %i, 1
   %exitcond25 = icmp eq i32 %add8, %I
-  br i1 %exitcond25, label %for.end.loopexit, label %for.inner
+  br i1 %exitcond25, label %for.end.loopexit, label %for.inner.preheader
+
+for.inner.preheader:
+  br label %for.inner
 
 for.inner:
-  %j = phi i32 [ 0, %for.outer ], [ %inc, %for.inner ]
-  %sum1 = phi i32 [ 0, %for.outer ], [ %add, %for.inner ]
+  %j = phi i32 [ %inc, %for.inner ], [ 0, %for.inner.preheader ]
+  %sum1 = phi i32 [ %add, %for.inner ], [ 0, %for.inner.preheader ]
   %arrayidx = getelementptr inbounds i32, i32* %B, i32 %j
   %0 = load i32, i32* %arrayidx, align 4
   %add = add i32 %0, %sum1
diff --git a/llvm/test/Transforms/LoopUnrollAndJam/pragma-explicit.ll b/llvm/test/Transforms/LoopUnrollAndJam/pragma-explicit.ll
--- a/llvm/test/Transforms/LoopUnrollAndJam/pragma-explicit.ll
+++ b/llvm/test/Transforms/LoopUnrollAndJam/pragma-explicit.ll
@@ -64,9 +64,12 @@
   %add.ptr.us = getelementptr inbounds i8, i8* %dst.addr.054.us, i64 %idx.ext
   %inc30.us = add nuw nsw i32 %y.055.us, 1
   %exitcond58 = icmp eq i32 %inc30.us, %height
-  br i1 %exitcond58, label %for.cond.cleanup, label %for.cond1.preheader.us, !llvm.loop !5
+  br i1 %exitcond58, label %for.cond.cleanup.loopexit, label %for.cond1.preheader.us, !llvm.loop !5
 
-for.cond.cleanup:                                 ; preds = %for.cond1.for.cond.cleanup3_crit_edge.us, %for.cond1.preheader.lr.ph, %entry
+for.cond.cleanup.loopexit:                        ; preds = %for.cond1.for.cond.cleanup3_crit_edge.us
+  br label %for.cond.cleanup
+
+for.cond.cleanup:                                 ; preds = %for.cond.cleanup.loopexit, %for.cond1.preheader.lr.ph, %entry
   ret void
 }
 
@@ -132,9 +135,12 @@
   %add.ptr.us = getelementptr inbounds i8, i8* %dst.addr.054.us, i64 %idx.ext
   %inc30.us = add nuw nsw i32 %y.055.us, 1
   %exitcond58 = icmp eq i32 %inc30.us, %height
-  br i1 %exitcond58, label %for.cond.cleanup, label %for.cond1.preheader.us, !llvm.loop !7
+  br i1 %exitcond58, label %for.cond.cleanup.loopexit, label %for.cond1.preheader.us, !llvm.loop !7
+
+for.cond.cleanup.loopexit:                        ; preds = %for.cond1.for.cond.cleanup3_crit_edge.us
+  br label %for.cond.cleanup
 
-for.cond.cleanup:                                 ; preds = %for.cond1.for.cond.cleanup3_crit_edge.us, %for.cond1.preheader.lr.ph, %entry
+for.cond.cleanup:                                 ; preds = %for.cond.cleanup.loopexit, %for.cond1.preheader.lr.ph, %entry
   ret void
 }
 
diff --git a/llvm/test/Transforms/LoopUnrollAndJam/unroll-and-jam.ll b/llvm/test/Transforms/LoopUnrollAndJam/unroll-and-jam.ll
--- a/llvm/test/Transforms/LoopUnrollAndJam/unroll-and-jam.ll
+++ b/llvm/test/Transforms/LoopUnrollAndJam/unroll-and-jam.ll
@@ -297,7 +297,10 @@
   store i32 %add, i32* %arrayidx6, align 4, !tbaa !5
   %add8 = add nuw nsw i32 %i, 1
   %exitcond23 = icmp eq i32 %add8, 4
-  br i1 %exitcond23, label %for.end, label %for.outer
+  br i1 %exitcond23, label %for.end.loopexit, label %for.outer
+
+for.end.loopexit:
+  br label %for.end
 
 for.end:
   ret void
@@ -344,7 +347,10 @@
   store i32 %add, i32* %arrayidx6, align 4, !tbaa !5
   %add8 = add nuw nsw i32 %i, 1
   %exitcond23 = icmp eq i32 %add8, 1
-  br i1 %exitcond23, label %for.end, label %for.outer
+  br i1 %exitcond23, label %for.end.loopexit, label %for.outer
+
+for.end.loopexit:
+  br label %for.end
 
 for.end:
   ret void
@@ -509,7 +515,7 @@
 for.latch:
   store i32 %add9, i32* %arrayidx, align 4, !tbaa !5
   %exitcond30 = icmp eq i32 %add, %I
-  br i1 %exitcond30, label %for.end, label %for.outer
+  br i1 %exitcond30, label %for.end.loopexit, label %for.outer
 
 for.inner:
   %sum = phi i32 [ 0, %for.outer ], [ %add9, %for.inner ]
@@ -521,6 +527,9 @@
   %exitcond = icmp eq i32 %add10, %J
   br i1 %exitcond, label %for.latch, label %for.inner
 
+for.end.loopexit:
+  br label %for.end
+
 for.end:
   ret void
 }
@@ -597,7 +606,10 @@
 for.cleanup:
   %inc = add nuw nsw i32 %x.038, 1
   %exitcond41 = icmp eq i32 %inc, 5
-  br i1 %exitcond41, label %for.end, label %for.outest
+  br i1 %exitcond41, label %for.end.loopexit, label %for.outest
+
+for.end.loopexit:
+  br label %for.end
 
 for.end:
   ret void