diff --git a/llvm/include/llvm/Transforms/Vectorize/LoopVectorizationLegality.h b/llvm/include/llvm/Transforms/Vectorize/LoopVectorizationLegality.h
--- a/llvm/include/llvm/Transforms/Vectorize/LoopVectorizationLegality.h
+++ b/llvm/include/llvm/Transforms/Vectorize/LoopVectorizationLegality.h
@@ -444,8 +444,9 @@
   /// Holds the widest induction type encountered.
   Type *WidestIndTy = nullptr;
 
-  /// Allowed outside users. This holds the induction and reduction
-  /// vars which can be accessed from outside the loop.
+  /// Allowed outside users. This holds the variables that can be accessed from
+  /// outside the loop (including induction and reduction vars and non-header
+  /// phis).
   SmallPtrSet<Value *, 4> AllowedExit;
 
   /// Can we assume the absence of NaNs.
diff --git a/llvm/lib/Transforms/Vectorize/LoopVectorizationLegality.cpp b/llvm/lib/Transforms/Vectorize/LoopVectorizationLegality.cpp
--- a/llvm/lib/Transforms/Vectorize/LoopVectorizationLegality.cpp
+++ b/llvm/lib/Transforms/Vectorize/LoopVectorizationLegality.cpp
@@ -595,6 +595,7 @@
           // Unsafe cyclic dependencies with header phis are identified during
           // legalization for reduction, induction and first order
           // recurrences.
+          AllowedExit.insert(&I);
           continue;
         }
 
diff --git a/llvm/test/Transforms/LoopVectorize/lv-fold-tail-by-masking-bug.ll b/llvm/test/Transforms/LoopVectorize/lv-fold-tail-by-masking-bug.ll
--- a/llvm/test/Transforms/LoopVectorize/lv-fold-tail-by-masking-bug.ll
+++ b/llvm/test/Transforms/LoopVectorize/lv-fold-tail-by-masking-bug.ll
@@ -32,41 +32,15 @@
 ;    build-all/bin/clang -O1 lv-bug.c -fvectorize && ./a.out
 ; the result is "BAD!"
 
-
-; FIXME: The result here is incorrect! See https://bugs.llvm.org/show_bug.cgi?id=43166
+; Used to miscompile (with clang 8.0.0), now we get
+;   loop not vectorized: Cannot fold tail by masking in the presence of live outs.
 define i64 @test1(i64 %y) {
 ; CHECK-LABEL: @test1(
 ; CHECK-NEXT:  entry:
-; CHECK-NEXT:    br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
-; CHECK:       vector.ph:
-; CHECK-NEXT:    [[BROADCAST_SPLATINSERT1:%.*]] = insertelement <4 x i64> undef, i64 [[Y:%.*]], i32 0
-; CHECK-NEXT:    [[BROADCAST_SPLAT2:%.*]] = shufflevector <4 x i64> [[BROADCAST_SPLATINSERT1]], <4 x i64> undef, <4 x i32> zeroinitializer
-; CHECK-NEXT:    br label [[VECTOR_BODY:%.*]]
-; CHECK:       vector.body:
-; CHECK-NEXT:    [[INDEX:%.*]] = phi i32 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT:    [[BROADCAST_SPLATINSERT:%.*]] = insertelement <4 x i32> undef, i32 [[INDEX]], i32 0
-; CHECK-NEXT:    [[BROADCAST_SPLAT:%.*]] = shufflevector <4 x i32> [[BROADCAST_SPLATINSERT]], <4 x i32> undef, <4 x i32> zeroinitializer
-; CHECK-NEXT:    [[INDUCTION:%.*]] = add <4 x i32> [[BROADCAST_SPLAT]], <i32 0, i32 1, i32 2, i32 3>
-; CHECK-NEXT:    [[TMP0:%.*]] = add i32 [[INDEX]], 0
-; CHECK-NEXT:    [[TMP1:%.*]] = icmp eq <4 x i64> [[BROADCAST_SPLAT2]], zeroinitializer
-; CHECK-NEXT:    [[TMP2:%.*]] = xor <4 x i64> <i64 3, i64 3, i64 3, i64 3>, [[BROADCAST_SPLAT2]]
-; CHECK-NEXT:    [[TMP3:%.*]] = icmp ule <4 x i32> [[INDUCTION]], <i32 2, i32 2, i32 2, i32 2>
-; CHECK-NEXT:    [[TMP4:%.*]] = xor <4 x i1> [[TMP1]], <i1 true, i1 true, i1 true, i1 true>
-; CHECK-NEXT:    [[TMP5:%.*]] = and <4 x i1> [[TMP4]], [[TMP3]]
-; CHECK-NEXT:    [[TMP6:%.*]] = and <4 x i1> [[TMP1]], [[TMP3]]
-; CHECK-NEXT:    [[PREDPHI:%.*]] = select <4 x i1> [[TMP6]], <4 x i64> <i64 77, i64 77, i64 77, i64 77>, <4 x i64> [[TMP2]]
-; CHECK-NEXT:    [[INDEX_NEXT]] = add i32 [[INDEX]], 4
-; CHECK-NEXT:    [[TMP7:%.*]] = icmp eq i32 [[INDEX_NEXT]], 4
-; CHECK-NEXT:    br i1 [[TMP7]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop !0
-; CHECK:       middle.block:
-; CHECK-NEXT:    [[TMP8:%.*]] = extractelement <4 x i64> [[PREDPHI]], i32 3
-; CHECK-NEXT:    br i1 true, label [[FOR_COND_CLEANUP:%.*]], label [[SCALAR_PH]]
-; CHECK:       scalar.ph:
-; CHECK-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i32 [ 4, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
 ; CHECK-NEXT:    br label [[FOR_BODY:%.*]]
 ; CHECK:       for.body:
-; CHECK-NEXT:    [[I:%.*]] = phi i32 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[INC:%.*]], [[COND_END:%.*]] ]
-; CHECK-NEXT:    [[CMP:%.*]] = icmp eq i64 [[Y]], 0
+; CHECK-NEXT:    [[I:%.*]] = phi i32 [ 0, [[ENTRY:%.*]] ], [ [[INC:%.*]], [[COND_END:%.*]] ]
+; CHECK-NEXT:    [[CMP:%.*]] = icmp eq i64 [[Y:%.*]], 0
 ; CHECK-NEXT:    br i1 [[CMP]], label [[COND_END]], label [[COND_FALSE:%.*]]
 ; CHECK:       cond.false:
 ; CHECK-NEXT:    [[DIV:%.*]] = xor i64 3, [[Y]]
@@ -75,9 +49,9 @@
 ; CHECK-NEXT:    [[COND:%.*]] = phi i64 [ [[DIV]], [[COND_FALSE]] ], [ 77, [[FOR_BODY]] ]
 ; CHECK-NEXT:    [[INC]] = add nuw nsw i32 [[I]], 1
 ; CHECK-NEXT:    [[EXITCOND:%.*]] = icmp eq i32 [[INC]], 3
-; CHECK-NEXT:    br i1 [[EXITCOND]], label [[FOR_COND_CLEANUP]], label [[FOR_BODY]], !llvm.loop !2
+; CHECK-NEXT:    br i1 [[EXITCOND]], label [[FOR_COND_CLEANUP:%.*]], label [[FOR_BODY]]
 ; CHECK:       for.cond.cleanup:
-; CHECK-NEXT:    [[COND_LCSSA:%.*]] = phi i64 [ [[COND]], [[COND_END]] ], [ [[TMP8]], [[MIDDLE_BLOCK]] ]
+; CHECK-NEXT:    [[COND_LCSSA:%.*]] = phi i64 [ [[COND]], [[COND_END]] ]
 ; CHECK-NEXT:    ret i64 [[COND_LCSSA]]
 ;
 entry:
@@ -102,40 +76,15 @@
   ret i64 %cond
 }
 
-
-; FIXME: The result here is incorrect! See https://bugs.llvm.org/show_bug.cgi?id=43166
+; Used to miscompile (with clang 8.0.0), now we get
+;   loop not vectorized: Cannot fold tail by masking in the presence of live outs.
 define i64 @test2(i64 %y) {
 ; CHECK-LABEL: @test2(
 ; CHECK-NEXT:  entry:
-; CHECK-NEXT:    br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
-; CHECK:       vector.ph:
-; CHECK-NEXT:    [[BROADCAST_SPLATINSERT1:%.*]] = insertelement <4 x i64> undef, i64 [[Y:%.*]], i32 0
-; CHECK-NEXT:    [[BROADCAST_SPLAT2:%.*]] = shufflevector <4 x i64> [[BROADCAST_SPLATINSERT1]], <4 x i64> undef, <4 x i32> zeroinitializer
-; CHECK-NEXT:    br label [[VECTOR_BODY:%.*]]
-; CHECK:       vector.body:
-; CHECK-NEXT:    [[INDEX:%.*]] = phi i32 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT:    [[BROADCAST_SPLATINSERT:%.*]] = insertelement <4 x i32> undef, i32 [[INDEX]], i32 0
-; CHECK-NEXT:    [[BROADCAST_SPLAT:%.*]] = shufflevector <4 x i32> [[BROADCAST_SPLATINSERT]], <4 x i32> undef, <4 x i32> zeroinitializer
-; CHECK-NEXT:    [[INDUCTION:%.*]] = add <4 x i32> [[BROADCAST_SPLAT]], <i32 0, i32 1, i32 2, i32 3>
-; CHECK-NEXT:    [[TMP0:%.*]] = add i32 [[INDEX]], 0
-; CHECK-NEXT:    [[TMP1:%.*]] = icmp eq <4 x i64> [[BROADCAST_SPLAT2]], zeroinitializer
-; CHECK-NEXT:    [[TMP2:%.*]] = icmp ule <4 x i32> [[INDUCTION]], <i32 2, i32 2, i32 2, i32 2>
-; CHECK-NEXT:    [[TMP3:%.*]] = xor <4 x i1> [[TMP1]], <i1 true, i1 true, i1 true, i1 true>
-; CHECK-NEXT:    [[TMP4:%.*]] = and <4 x i1> [[TMP3]], [[TMP2]]
-; CHECK-NEXT:    [[TMP5:%.*]] = and <4 x i1> [[TMP1]], [[TMP2]]
-; CHECK-NEXT:    [[PREDPHI:%.*]] = select <4 x i1> [[TMP5]], <4 x i64> <i64 77, i64 77, i64 77, i64 77>, <4 x i64> <i64 55, i64 55, i64 55, i64 55>
-; CHECK-NEXT:    [[INDEX_NEXT]] = add i32 [[INDEX]], 4
-; CHECK-NEXT:    [[TMP6:%.*]] = icmp eq i32 [[INDEX_NEXT]], 4
-; CHECK-NEXT:    br i1 [[TMP6]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop !4
-; CHECK:       middle.block:
-; CHECK-NEXT:    [[TMP7:%.*]] = extractelement <4 x i64> [[PREDPHI]], i32 3
-; CHECK-NEXT:    br i1 true, label [[FOR_COND_CLEANUP:%.*]], label [[SCALAR_PH]]
-; CHECK:       scalar.ph:
-; CHECK-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i32 [ 4, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
 ; CHECK-NEXT:    br label [[FOR_BODY:%.*]]
 ; CHECK:       for.body:
-; CHECK-NEXT:    [[I:%.*]] = phi i32 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[INC:%.*]], [[COND_END:%.*]] ]
-; CHECK-NEXT:    [[CMP:%.*]] = icmp eq i64 [[Y]], 0
+; CHECK-NEXT:    [[I:%.*]] = phi i32 [ 0, [[ENTRY:%.*]] ], [ [[INC:%.*]], [[COND_END:%.*]] ]
+; CHECK-NEXT:    [[CMP:%.*]] = icmp eq i64 [[Y:%.*]], 0
 ; CHECK-NEXT:    br i1 [[CMP]], label [[COND_END]], label [[COND_FALSE:%.*]]
 ; CHECK:       cond.false:
 ; CHECK-NEXT:    br label [[COND_END]]
@@ -143,9 +92,9 @@
 ; CHECK-NEXT:    [[COND:%.*]] = phi i64 [ 55, [[COND_FALSE]] ], [ 77, [[FOR_BODY]] ]
 ; CHECK-NEXT:    [[INC]] = add nuw nsw i32 [[I]], 1
 ; CHECK-NEXT:    [[EXITCOND:%.*]] = icmp eq i32 [[INC]], 3
-; CHECK-NEXT:    br i1 [[EXITCOND]], label [[FOR_COND_CLEANUP]], label [[FOR_BODY]], !llvm.loop !5
+; CHECK-NEXT:    br i1 [[EXITCOND]], label [[FOR_COND_CLEANUP:%.*]], label [[FOR_BODY]]
 ; CHECK:       for.cond.cleanup:
-; CHECK-NEXT:    [[COND_LCSSA:%.*]] = phi i64 [ [[COND]], [[COND_END]] ], [ [[TMP7]], [[MIDDLE_BLOCK]] ]
+; CHECK-NEXT:    [[COND_LCSSA:%.*]] = phi i64 [ [[COND]], [[COND_END]] ]
 ; CHECK-NEXT:    ret i64 [[COND_LCSSA]]
 ;
 entry:
@@ -169,38 +118,15 @@
   ret i64 %cond
 }
 
-
-; FIXME: The result here is incorrect! See https://bugs.llvm.org/show_bug.cgi?id=43166
+; Used to miscompile (with clang 8.0.0), now we get
+;   loop not vectorized: Cannot fold tail by masking in the presence of live outs.
 define i32 @test3(i64 %y) {
 ; CHECK-LABEL: @test3(
 ; CHECK-NEXT:  entry:
-; CHECK-NEXT:    br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
-; CHECK:       vector.ph:
-; CHECK-NEXT:    [[BROADCAST_SPLATINSERT:%.*]] = insertelement <4 x i64> undef, i64 [[Y:%.*]], i32 0
-; CHECK-NEXT:    [[BROADCAST_SPLAT:%.*]] = shufflevector <4 x i64> [[BROADCAST_SPLATINSERT]], <4 x i64> undef, <4 x i32> zeroinitializer
-; CHECK-NEXT:    br label [[VECTOR_BODY:%.*]]
-; CHECK:       vector.body:
-; CHECK-NEXT:    [[INDEX:%.*]] = phi i32 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT:    [[VEC_IND:%.*]] = phi <4 x i32> [ <i32 0, i32 1, i32 2, i32 3>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT:    [[TMP0:%.*]] = icmp eq <4 x i64> [[BROADCAST_SPLAT]], zeroinitializer
-; CHECK-NEXT:    [[TMP1:%.*]] = icmp ule <4 x i32> [[VEC_IND]], <i32 2, i32 2, i32 2, i32 2>
-; CHECK-NEXT:    [[TMP2:%.*]] = xor <4 x i1> [[TMP0]], <i1 true, i1 true, i1 true, i1 true>
-; CHECK-NEXT:    [[TMP3:%.*]] = and <4 x i1> [[TMP2]], [[TMP1]]
-; CHECK-NEXT:    [[TMP4:%.*]] = and <4 x i1> [[TMP0]], [[TMP1]]
-; CHECK-NEXT:    [[PREDPHI:%.*]] = select <4 x i1> [[TMP4]], <4 x i32> [[VEC_IND]], <4 x i32> <i32 55, i32 55, i32 55, i32 55>
-; CHECK-NEXT:    [[INDEX_NEXT]] = add i32 [[INDEX]], 4
-; CHECK-NEXT:    [[VEC_IND_NEXT]] = add <4 x i32> [[VEC_IND]], <i32 4, i32 4, i32 4, i32 4>
-; CHECK-NEXT:    [[TMP5:%.*]] = icmp eq i32 [[INDEX_NEXT]], 4
-; CHECK-NEXT:    br i1 [[TMP5]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop !6
-; CHECK:       middle.block:
-; CHECK-NEXT:    [[TMP6:%.*]] = extractelement <4 x i32> [[PREDPHI]], i32 3
-; CHECK-NEXT:    br i1 true, label [[FOR_COND_CLEANUP:%.*]], label [[SCALAR_PH]]
-; CHECK:       scalar.ph:
-; CHECK-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i32 [ 4, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
 ; CHECK-NEXT:    br label [[FOR_BODY:%.*]]
 ; CHECK:       for.body:
-; CHECK-NEXT:    [[I:%.*]] = phi i32 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[INC:%.*]], [[COND_END:%.*]] ]
-; CHECK-NEXT:    [[CMP:%.*]] = icmp eq i64 [[Y]], 0
+; CHECK-NEXT:    [[I:%.*]] = phi i32 [ 0, [[ENTRY:%.*]] ], [ [[INC:%.*]], [[COND_END:%.*]] ]
+; CHECK-NEXT:    [[CMP:%.*]] = icmp eq i64 [[Y:%.*]], 0
 ; CHECK-NEXT:    br i1 [[CMP]], label [[COND_END]], label [[COND_FALSE:%.*]]
 ; CHECK:       cond.false:
 ; CHECK-NEXT:    br label [[COND_END]]
@@ -208,9 +134,9 @@
 ; CHECK-NEXT:    [[COND:%.*]] = phi i32 [ 55, [[COND_FALSE]] ], [ [[I]], [[FOR_BODY]] ]
 ; CHECK-NEXT:    [[INC]] = add nuw nsw i32 [[I]], 1
 ; CHECK-NEXT:    [[EXITCOND:%.*]] = icmp eq i32 [[INC]], 3
-; CHECK-NEXT:    br i1 [[EXITCOND]], label [[FOR_COND_CLEANUP]], label [[FOR_BODY]], !llvm.loop !7
+; CHECK-NEXT:    br i1 [[EXITCOND]], label [[FOR_COND_CLEANUP:%.*]], label [[FOR_BODY]]
 ; CHECK:       for.cond.cleanup:
-; CHECK-NEXT:    [[COND_LCSSA:%.*]] = phi i32 [ [[COND]], [[COND_END]] ], [ [[TMP6]], [[MIDDLE_BLOCK]] ]
+; CHECK-NEXT:    [[COND_LCSSA:%.*]] = phi i32 [ [[COND]], [[COND_END]] ]
 ; CHECK-NEXT:    ret i32 [[COND_LCSSA]]
 ;
 entry: