diff --git a/llvm/lib/CodeGen/MachineBlockPlacement.cpp b/llvm/lib/CodeGen/MachineBlockPlacement.cpp
--- a/llvm/lib/CodeGen/MachineBlockPlacement.cpp
+++ b/llvm/lib/CodeGen/MachineBlockPlacement.cpp
@@ -2050,6 +2050,8 @@
   BlockChain &HeaderChain = *BlockToChain[OldTop];
   if (!LoopBlockSet.count(*HeaderChain.begin()))
     return OldTop;
+  if (OldTop != *HeaderChain.begin())
+    return OldTop;
 
   LLVM_DEBUG(dbgs() << "Finding best loop top for: " << getBlockName(OldTop)
                     << "\n");
diff --git a/llvm/test/CodeGen/ARM/mbp.ll b/llvm/test/CodeGen/ARM/mbp.ll
new file mode 100644
--- /dev/null
+++ b/llvm/test/CodeGen/ARM/mbp.ll
@@ -0,0 +1,65 @@
+; RUN: llc < %s | FileCheck %s
+target datalayout = "e-m:e-p:32:32-Fi8-i64:64-v128:64:128-a:0:32-n32-S64"
+target triple = "thumbv7-unknown-linux-gnueabihf"
+
+%Target = type { %Target*, %List* }
+%List = type { i32, i32* }
+
+; The entry block should be the first block of the function.
+; CHECK-LABEL: foo
+; CHECK:       %entry
+; CHECK:       %for.body
+; CHECK:       %for.inc
+; CHECK:       %if.then
+; CHECK:       %for.cond.i
+; CHECK:       %for.body.i
+; CHECK:       %return
+
+define i1 @foo(%Target** %ha, i32 %he) !prof !39 {
+entry:
+  %TargetPtr = load %Target*, %Target** %ha, align 4
+  %cmp1 = icmp eq %Target* %TargetPtr, null
+  br i1 %cmp1, label %return, label %for.body, !prof !50
+
+for.body:
+  %TargetPhi = phi %Target* [ %NextPtr, %for.inc ], [ %TargetPtr, %entry ]
+  %ListAddr = getelementptr inbounds %Target, %Target* %TargetPhi, i32 0, i32 1
+  %ListPtr = load %List*, %List** %ListAddr, align 4
+  %cmp2 = icmp eq %List* %ListPtr, null
+  br i1 %cmp2, label %for.inc, label %if.then, !prof !59
+
+if.then:
+  %lenAddr = getelementptr inbounds %List, %List* %ListPtr, i32 0, i32 0
+  %len = load i32, i32* %lenAddr, align 4
+  %ptr = getelementptr inbounds %List, %List* %ListPtr, i32 0, i32 1
+  %ptr2 = load i32*, i32** %ptr, align 4
+  br label %for.cond.i
+
+for.cond.i:
+  %i = phi i32 [ %len, %if.then ], [ %index, %for.body.i ]
+  %index = add nsw i32 %i, -1
+  %cmp3 = icmp sgt i32 %i, 0
+  br i1 %cmp3, label %for.body.i, label %for.inc, !prof !75
+
+for.body.i:
+  %ptr3 = getelementptr inbounds i32, i32* %ptr2, i32 %index
+  %data = load i32, i32* %ptr3, align 4
+  %cmp4 = icmp eq i32 %data, %he
+  br i1 %cmp4, label %return, label %for.cond.i, !prof !79
+
+for.inc:
+  %NextAddr = getelementptr inbounds %Target, %Target* %TargetPhi, i32 0, i32 0
+  %NextPtr = load %Target*, %Target** %NextAddr, align 4
+  %cmp5 = icmp eq %Target* %NextPtr, null
+  br i1 %cmp5, label %return, label %for.body, !prof !50
+
+return:
+  %retval = phi i1 [ false, %entry ], [ true, %for.body.i ], [ false, %for.inc ]
+  ret i1 %retval
+}
+
+!39 = !{!"function_entry_count", i64 226}
+!50 = !{!"branch_weights", i32 451, i32 1}
+!59 = !{!"branch_weights", i32 1502, i32 1}
+!75 = !{!"branch_weights", i32 301, i32 1}
+!79 = !{!"branch_weights", i32 1, i32 301}