Index: lib/Transforms/Vectorize/SLPVectorizer.cpp
===================================================================
--- lib/Transforms/Vectorize/SLPVectorizer.cpp
+++ lib/Transforms/Vectorize/SLPVectorizer.cpp
@@ -4789,7 +4789,7 @@
     if (!V)
       continue;
     auto *Inst = dyn_cast<Instruction>(V);
-    if (!Inst || isa<PHINode>(Inst))
+    if (!Inst || isa<PHINode>(Inst) || Inst->getParent() != BB)
       continue;
     if (auto *BI = dyn_cast<BinaryOperator>(Inst)) {
       HorizontalReduction HorRdx;
@@ -4825,7 +4825,8 @@
     // Try to vectorize operands.
     if (++Level < RecursionMaxDepth)
       for (auto *Op : Inst->operand_values())
-        Stack.emplace_back(Op, Level);
+        if (VisitedInstrs.insert(Op).second)
+          Stack.emplace_back(Op, Level);
   }
   return Res;
 }
Index: test/Transforms/SLPVectorizer/X86/limit.ll
===================================================================
--- /dev/null
+++ test/Transforms/SLPVectorizer/X86/limit.ll
@@ -0,0 +1,70 @@
+; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
+; RUN: opt < %s --instcombine -slp-vectorizer -S | FileCheck %s
+target datalayout = "e-m:e-i64:64-f80:128-n8:16:32:64-S128"
+target triple = "x86_64-unknown-linux-gnu"
+
+@b = common global [4 x i32] zeroinitializer, align 16
+@c = common global [4 x i32] zeroinitializer, align 16
+@d = common global [4 x i32] zeroinitializer, align 16
+@e = common global [4 x i32] zeroinitializer, align 16
+@a = common global [4 x i32] zeroinitializer, align 16
+@fb = common global [4 x float] zeroinitializer, align 16
+@fc = common global [4 x float] zeroinitializer, align 16
+@fa = common global [4 x float] zeroinitializer, align 16
+@fd = common global [4 x float] zeroinitializer, align 16
+
+define void @addsub() {
+; CHECK-LABEL: @addsub(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    br label [[BB1:%.*]]
+; CHECK:       bb1:
+; CHECK-NEXT:    [[TMP0:%.*]] = load <4 x i32>, <4 x i32>* bitcast ([4 x i32]* @b to <4 x i32>*), align 16
+; CHECK-NEXT:    [[TMP1:%.*]] = load <4 x i32>, <4 x i32>* bitcast ([4 x i32]* @c to <4 x i32>*), align 16
+; CHECK-NEXT:    [[TMP2:%.*]] = add nsw <4 x i32> [[TMP0]], [[TMP1]]
+; CHECK-NEXT:    [[TMP3:%.*]] = load <4 x i32>, <4 x i32>* bitcast ([4 x i32]* @d to <4 x i32>*), align 16
+; CHECK-NEXT:    [[TMP4:%.*]] = load <4 x i32>, <4 x i32>* bitcast ([4 x i32]* @e to <4 x i32>*), align 16
+; CHECK-NEXT:    [[TMP5:%.*]] = add nsw <4 x i32> [[TMP3]], [[TMP4]]
+; CHECK-NEXT:    [[TMP6:%.*]] = add nsw <4 x i32> [[TMP2]], [[TMP5]]
+; CHECK-NEXT:    [[TMP7:%.*]] = sub nsw <4 x i32> [[TMP2]], [[TMP5]]
+; CHECK-NEXT:    [[TMP8:%.*]] = shufflevector <4 x i32> [[TMP6]], <4 x i32> [[TMP7]], <4 x i32> <i32 0, i32 5, i32 2, i32 7>
+; CHECK-NEXT:    store <4 x i32> [[TMP8]], <4 x i32>* bitcast ([4 x i32]* @a to <4 x i32>*), align 16
+; CHECK-NEXT:    ret void
+;
+entry:
+  %0 = load i32, i32* getelementptr inbounds ([4 x i32], [4 x i32]* @b, i32 0, i64 0), align 4
+  %1 = load i32, i32* getelementptr inbounds ([4 x i32], [4 x i32]* @c, i32 0, i64 0), align 4
+  %add = add nsw i32 %0, %1
+  br label %bb1
+bb1:                                              ; preds = %entry
+  %2 = load i32, i32* getelementptr inbounds ([4 x i32], [4 x i32]* @d, i32 0, i64 0), align 4
+  %3 = load i32, i32* getelementptr inbounds ([4 x i32], [4 x i32]* @e, i32 0, i64 0), align 4
+  %add1 = add nsw i32 %2, %3
+  %add2 = add nsw i32 %add, %add1
+  store i32 %add2, i32* getelementptr inbounds ([4 x i32], [4 x i32]* @a, i32 0, i64 0), align 4
+  %4 = load i32, i32* getelementptr inbounds ([4 x i32], [4 x i32]* @b, i32 0, i64 1), align 4
+  %5 = load i32, i32* getelementptr inbounds ([4 x i32], [4 x i32]* @c, i32 0, i64 1), align 4
+  %add3 = add nsw i32 %4, %5
+  %6 = load i32, i32* getelementptr inbounds ([4 x i32], [4 x i32]* @d, i32 0, i64 1), align 4
+  %7 = load i32, i32* getelementptr inbounds ([4 x i32], [4 x i32]* @e, i32 0, i64 1), align 4
+  %add4 = add nsw i32 %6, %7
+  %sub = sub nsw i32 %add3, %add4
+  store i32 %sub, i32* getelementptr inbounds ([4 x i32], [4 x i32]* @a, i32 0, i64 1), align 4
+  %8 = load i32, i32* getelementptr inbounds ([4 x i32], [4 x i32]* @b, i32 0, i64 2), align 4
+  %9 = load i32, i32* getelementptr inbounds ([4 x i32], [4 x i32]* @c, i32 0, i64 2), align 4
+  %add5 = add nsw i32 %8, %9
+  %10 = load i32, i32* getelementptr inbounds ([4 x i32], [4 x i32]* @d, i32 0, i64 2), align 4
+  %11 = load i32, i32* getelementptr inbounds ([4 x i32], [4 x i32]* @e, i32 0, i64 2), align 4
+  %add6 = add nsw i32 %10, %11
+  %add7 = add nsw i32 %add5, %add6
+  store i32 %add7, i32* getelementptr inbounds ([4 x i32], [4 x i32]* @a, i32 0, i64 2), align 4
+  %12 = load i32, i32* getelementptr inbounds ([4 x i32], [4 x i32]* @b, i32 0, i64 3), align 4
+  %13 = load i32, i32* getelementptr inbounds ([4 x i32], [4 x i32]* @c, i32 0, i64 3), align 4
+  %add8 = add nsw i32 %12, %13
+  %14 = load i32, i32* getelementptr inbounds ([4 x i32], [4 x i32]* @d, i32 0, i64 3), align 4
+  %15 = load i32, i32* getelementptr inbounds ([4 x i32], [4 x i32]* @e, i32 0, i64 3), align 4
+  %add9 = add nsw i32 %14, %15
+  %sub10 = sub nsw i32 %add8, %add9
+  store i32 %sub10, i32* getelementptr inbounds ([4 x i32], [4 x i32]* @a, i32 0, i64 3), align 4
+  ret void
+}
+