Index: lib/CodeGen/SelectionDAG/DAGCombiner.cpp
===================================================================
--- lib/CodeGen/SelectionDAG/DAGCombiner.cpp
+++ lib/CodeGen/SelectionDAG/DAGCombiner.cpp
@@ -12678,7 +12678,11 @@
       SDValue Val = St->getValue();
       StoreInt <<= ElementSizeBits;
       if (ConstantSDNode *C = dyn_cast<ConstantSDNode>(Val)) {
-        StoreInt |= C->getAPIntValue().zextOrTrunc(SizeInBits);
+        // Since the store may be truncating, the stored immediate must be
+        // truncated to the width of the element.
+        APInt CElt(ElementSizeBits,
+               C->getAPIntValue().zextOrTrunc(ElementSizeBits).getZExtValue());
+        StoreInt |= CElt.zextOrTrunc(SizeInBits);
       } else if (ConstantFPSDNode *C = dyn_cast<ConstantFPSDNode>(Val)) {
         StoreInt |= C->getValueAPF().bitcastToAPInt().zextOrTrunc(SizeInBits);
       } else {
Index: test/CodeGen/SystemZ/DAGCombiner_stores.ll
===================================================================
--- /dev/null
+++ test/CodeGen/SystemZ/DAGCombiner_stores.ll
@@ -0,0 +1,327 @@
+; RUN: llc -mtriple=s390x-linux-gnu -mcpu=zEC12 < %s  | FileCheck %s
+;
+; This (reduced csmith) test case stores a lot of 16 bit constant -7 values
+; to adjacent locations, which the DAGCombiner will merge to wider stores. It
+; was discovered that the result of combining two or four such stores
+; resulted incorrectly in stores of 32-bit / 64-bit constants of value
+; -7. The correct constant then is not -7, but consists of two or four i16 -7
+; elements.
+;
+; CHECK-NOT: mvghi {{.*}} -7
+; CHECK-NOT: mvhi {{.*}} -7
+
+@0 = internal unnamed_addr global i16 1, align 2
+@1 = internal unnamed_addr global [10 x [7 x i16]] [[7 x i16] [i16 1, i16 -6729, i16 0, i16 0, i16 -6729, i16 1, i16 -4591], [7 x i16] [i16 -26203, i16 -6729, i16 9, i16 -4591, i16 0, i16 0, i16 0], [7 x i16] [i16 0, i16 0, i16 -10191, i16 0, i16 0, i16 -17444, i16 -26203], [7 x i16] [i16 22368, i16 -6729, i16 0, i16 1, i16 0, i16 -26203, i16 0], [7 x i16] [i16 1, i16 0, i16 0, i16 1, i16 0, i16 1, i16 22368], [7 x i16] [i16 22368, i16 1, i16 0, i16 1, i16 0, i16 0, i16 1], [7 x i16] [i16 0, i16 -26203, i16 0, i16 1, i16 0, i16 -6729, i16 22368], [7 x i16] [i16 -26203, i16 -17444, i16 0, i16 0, i16 -10191, i16 0, i16 0], [7 x i16] [i16 0, i16 0, i16 0, i16 -4591, i16 9, i16 -6729, i16 -26203], [7 x i16] [i16 -4591, i16 0, i16 0, i16 -6729, i16 -6729, i16 0, i16 0]], align 2
+@2 = private unnamed_addr constant [12 x i8] c"g_112[i][j]\00", align 2
+@3 = private unnamed_addr constant [9 x i8] c"g_653[i]\00", align 2
+@4 = internal unnamed_addr global i32 -1, align 4
+@5 = internal unnamed_addr global [256 x i32] zeroinitializer, align 4
+@6 = private unnamed_addr constant [36 x i8] c"...checksum after hashing %s : %lX\0A\00", align 2
+@7 = private unnamed_addr constant [15 x i8] c"checksum = %X\0A\00", align 2
+
+; Function Attrs: nounwind
+define signext i32 @main(i32 signext, i8** nocapture readonly) local_unnamed_addr #0 {
+  %3 = icmp eq i32 %0, 2
+  br i1 %3, label %4, label %20
+
+; <label>:4:                                      ; preds = %2
+  %5 = getelementptr inbounds i8*, i8** %1, i64 1
+  %6 = load i8*, i8** %5, align 8, !tbaa !2
+  %7 = load i8, i8* %6, align 1, !tbaa !6
+  %8 = zext i8 %7 to i32
+  %9 = sub nsw i32 49, %8
+  %10 = icmp eq i32 %9, 0
+  br i1 %10, label %11, label %16
+
+; <label>:11:                                     ; preds = %4
+  %12 = getelementptr inbounds i8, i8* %6, i64 1
+  %13 = load i8, i8* %12, align 1, !tbaa !6
+  %14 = zext i8 %13 to i32
+  %15 = sub nsw i32 0, %14
+  br label %16
+
+; <label>:16:                                     ; preds = %11, %4
+  %17 = phi i32 [ %15, %11 ], [ %9, %4 ]
+  %18 = icmp eq i32 %17, 0
+  %19 = zext i1 %18 to i32
+  br label %20
+
+; <label>:20:                                     ; preds = %16, %2
+  %21 = phi i32 [ 0, %2 ], [ %19, %16 ]
+  br label %22
+
+; <label>:22:                                     ; preds = %22, %20
+  %23 = phi i64 [ 0, %20 ], [ %66, %22 ]
+  %24 = trunc i64 %23 to i32
+  %25 = and i32 %24, 1
+  %26 = icmp eq i32 %25, 0
+  %27 = lshr i32 %24, 1
+  %28 = xor i32 %27, -306674912
+  %29 = select i1 %26, i32 %27, i32 %28
+  %30 = and i32 %29, 1
+  %31 = icmp eq i32 %30, 0
+  %32 = lshr i32 %29, 1
+  %33 = xor i32 %32, -306674912
+  %34 = select i1 %31, i32 %32, i32 %33
+  %35 = and i32 %34, 1
+  %36 = icmp eq i32 %35, 0
+  %37 = lshr i32 %34, 1
+  %38 = xor i32 %37, -306674912
+  %39 = select i1 %36, i32 %37, i32 %38
+  %40 = and i32 %39, 1
+  %41 = icmp eq i32 %40, 0
+  %42 = lshr i32 %39, 1
+  %43 = xor i32 %42, -306674912
+  %44 = select i1 %41, i32 %42, i32 %43
+  %45 = and i32 %44, 1
+  %46 = icmp eq i32 %45, 0
+  %47 = lshr i32 %44, 1
+  %48 = xor i32 %47, -306674912
+  %49 = select i1 %46, i32 %47, i32 %48
+  %50 = and i32 %49, 1
+  %51 = icmp eq i32 %50, 0
+  %52 = lshr i32 %49, 1
+  %53 = xor i32 %52, -306674912
+  %54 = select i1 %51, i32 %52, i32 %53
+  %55 = and i32 %54, 1
+  %56 = icmp eq i32 %55, 0
+  %57 = lshr i32 %54, 1
+  %58 = xor i32 %57, -306674912
+  %59 = select i1 %56, i32 %57, i32 %58
+  %60 = and i32 %59, 1
+  %61 = icmp eq i32 %60, 0
+  %62 = lshr i32 %59, 1
+  %63 = xor i32 %62, -306674912
+  %64 = select i1 %61, i32 %62, i32 %63
+  %65 = getelementptr inbounds [256 x i32], [256 x i32]* @5, i64 0, i64 %23
+  store i32 %64, i32* %65, align 4, !tbaa !7
+  %66 = add nuw nsw i64 %23, 1
+  %67 = icmp eq i64 %66, 256
+  br i1 %67, label %68, label %22
+
+; <label>:68:                                     ; preds = %22
+  %69 = load i16, i16* @0, align 2, !tbaa !9
+  %70 = or i16 %69, 32606
+  store i16 %70, i16* @0, align 2, !tbaa !9
+  store i16 -7, i16* getelementptr inbounds ([10 x [7 x i16]], [10 x [7 x i16]]* @1, i64 0, i64 0, i64 0), align 2, !tbaa !9
+  store i16 -7, i16* getelementptr inbounds ([10 x [7 x i16]], [10 x [7 x i16]]* @1, i64 0, i64 0, i64 1), align 2, !tbaa !9
+  store i16 -7, i16* getelementptr inbounds ([10 x [7 x i16]], [10 x [7 x i16]]* @1, i64 0, i64 0, i64 2), align 2, !tbaa !9
+  store i16 -7, i16* getelementptr inbounds ([10 x [7 x i16]], [10 x [7 x i16]]* @1, i64 0, i64 0, i64 3), align 2, !tbaa !9
+  store i16 -7, i16* getelementptr inbounds ([10 x [7 x i16]], [10 x [7 x i16]]* @1, i64 0, i64 0, i64 4), align 2, !tbaa !9
+  store i16 -7, i16* getelementptr inbounds ([10 x [7 x i16]], [10 x [7 x i16]]* @1, i64 0, i64 0, i64 5), align 2, !tbaa !9
+  store i16 -7, i16* getelementptr inbounds ([10 x [7 x i16]], [10 x [7 x i16]]* @1, i64 0, i64 0, i64 6), align 2, !tbaa !9
+  store i16 -7, i16* getelementptr inbounds ([10 x [7 x i16]], [10 x [7 x i16]]* @1, i64 0, i64 1, i64 0), align 2, !tbaa !9
+  store i16 -7, i16* getelementptr inbounds ([10 x [7 x i16]], [10 x [7 x i16]]* @1, i64 0, i64 1, i64 1), align 2, !tbaa !9
+  store i16 -7, i16* getelementptr inbounds ([10 x [7 x i16]], [10 x [7 x i16]]* @1, i64 0, i64 1, i64 2), align 2, !tbaa !9
+  store i16 -7, i16* getelementptr inbounds ([10 x [7 x i16]], [10 x [7 x i16]]* @1, i64 0, i64 1, i64 3), align 2, !tbaa !9
+  store i16 -7, i16* getelementptr inbounds ([10 x [7 x i16]], [10 x [7 x i16]]* @1, i64 0, i64 1, i64 4), align 2, !tbaa !9
+  store i16 -7, i16* getelementptr inbounds ([10 x [7 x i16]], [10 x [7 x i16]]* @1, i64 0, i64 1, i64 5), align 2, !tbaa !9
+  store i16 -7, i16* getelementptr inbounds ([10 x [7 x i16]], [10 x [7 x i16]]* @1, i64 0, i64 1, i64 6), align 2, !tbaa !9
+  store i16 -7, i16* getelementptr inbounds ([10 x [7 x i16]], [10 x [7 x i16]]* @1, i64 0, i64 2, i64 0), align 2, !tbaa !9
+  store i16 -7, i16* getelementptr inbounds ([10 x [7 x i16]], [10 x [7 x i16]]* @1, i64 0, i64 2, i64 1), align 2, !tbaa !9
+  store i16 -7, i16* getelementptr inbounds ([10 x [7 x i16]], [10 x [7 x i16]]* @1, i64 0, i64 2, i64 2), align 2, !tbaa !9
+  store i16 -7, i16* getelementptr inbounds ([10 x [7 x i16]], [10 x [7 x i16]]* @1, i64 0, i64 2, i64 3), align 2, !tbaa !9
+  store i16 -7, i16* getelementptr inbounds ([10 x [7 x i16]], [10 x [7 x i16]]* @1, i64 0, i64 2, i64 4), align 2, !tbaa !9
+  store i16 -7, i16* getelementptr inbounds ([10 x [7 x i16]], [10 x [7 x i16]]* @1, i64 0, i64 2, i64 5), align 2, !tbaa !9
+  store i16 -7, i16* getelementptr inbounds ([10 x [7 x i16]], [10 x [7 x i16]]* @1, i64 0, i64 2, i64 6), align 2, !tbaa !9
+  store i16 -7, i16* getelementptr inbounds ([10 x [7 x i16]], [10 x [7 x i16]]* @1, i64 0, i64 3, i64 0), align 2, !tbaa !9
+  store i16 -7, i16* getelementptr inbounds ([10 x [7 x i16]], [10 x [7 x i16]]* @1, i64 0, i64 3, i64 1), align 2, !tbaa !9
+  store i16 -7, i16* getelementptr inbounds ([10 x [7 x i16]], [10 x [7 x i16]]* @1, i64 0, i64 3, i64 2), align 2, !tbaa !9
+  store i16 -7, i16* getelementptr inbounds ([10 x [7 x i16]], [10 x [7 x i16]]* @1, i64 0, i64 3, i64 3), align 2, !tbaa !9
+  store i16 -7, i16* getelementptr inbounds ([10 x [7 x i16]], [10 x [7 x i16]]* @1, i64 0, i64 3, i64 4), align 2, !tbaa !9
+  store i16 -7, i16* getelementptr inbounds ([10 x [7 x i16]], [10 x [7 x i16]]* @1, i64 0, i64 3, i64 5), align 2, !tbaa !9
+  store i16 -7, i16* getelementptr inbounds ([10 x [7 x i16]], [10 x [7 x i16]]* @1, i64 0, i64 3, i64 6), align 2, !tbaa !9
+  store i16 -7, i16* getelementptr inbounds ([10 x [7 x i16]], [10 x [7 x i16]]* @1, i64 0, i64 4, i64 0), align 2, !tbaa !9
+  store i16 -7, i16* getelementptr inbounds ([10 x [7 x i16]], [10 x [7 x i16]]* @1, i64 0, i64 4, i64 1), align 2, !tbaa !9
+  store i16 -7, i16* getelementptr inbounds ([10 x [7 x i16]], [10 x [7 x i16]]* @1, i64 0, i64 4, i64 2), align 2, !tbaa !9
+  store i16 -7, i16* getelementptr inbounds ([10 x [7 x i16]], [10 x [7 x i16]]* @1, i64 0, i64 4, i64 3), align 2, !tbaa !9
+  store i16 -7, i16* getelementptr inbounds ([10 x [7 x i16]], [10 x [7 x i16]]* @1, i64 0, i64 4, i64 4), align 2, !tbaa !9
+  store i16 -7, i16* getelementptr inbounds ([10 x [7 x i16]], [10 x [7 x i16]]* @1, i64 0, i64 4, i64 5), align 2, !tbaa !9
+  store i16 -7, i16* getelementptr inbounds ([10 x [7 x i16]], [10 x [7 x i16]]* @1, i64 0, i64 4, i64 6), align 2, !tbaa !9
+  store i16 -7, i16* getelementptr inbounds ([10 x [7 x i16]], [10 x [7 x i16]]* @1, i64 0, i64 5, i64 0), align 2, !tbaa !9
+  store i16 -7, i16* getelementptr inbounds ([10 x [7 x i16]], [10 x [7 x i16]]* @1, i64 0, i64 5, i64 1), align 2, !tbaa !9
+  store i16 -7, i16* getelementptr inbounds ([10 x [7 x i16]], [10 x [7 x i16]]* @1, i64 0, i64 5, i64 2), align 2, !tbaa !9
+  store i16 -7, i16* getelementptr inbounds ([10 x [7 x i16]], [10 x [7 x i16]]* @1, i64 0, i64 5, i64 3), align 2, !tbaa !9
+  store i16 -7, i16* getelementptr inbounds ([10 x [7 x i16]], [10 x [7 x i16]]* @1, i64 0, i64 5, i64 4), align 2, !tbaa !9
+  store i16 -7, i16* getelementptr inbounds ([10 x [7 x i16]], [10 x [7 x i16]]* @1, i64 0, i64 5, i64 5), align 2, !tbaa !9
+  store i16 -7, i16* getelementptr inbounds ([10 x [7 x i16]], [10 x [7 x i16]]* @1, i64 0, i64 5, i64 6), align 2, !tbaa !9
+  store i16 -7, i16* getelementptr inbounds ([10 x [7 x i16]], [10 x [7 x i16]]* @1, i64 0, i64 6, i64 0), align 2, !tbaa !9
+  store i16 -7, i16* getelementptr inbounds ([10 x [7 x i16]], [10 x [7 x i16]]* @1, i64 0, i64 6, i64 1), align 2, !tbaa !9
+  store i16 -7, i16* getelementptr inbounds ([10 x [7 x i16]], [10 x [7 x i16]]* @1, i64 0, i64 6, i64 2), align 2, !tbaa !9
+  store i16 -7, i16* getelementptr inbounds ([10 x [7 x i16]], [10 x [7 x i16]]* @1, i64 0, i64 6, i64 3), align 2, !tbaa !9
+  store i16 -7, i16* getelementptr inbounds ([10 x [7 x i16]], [10 x [7 x i16]]* @1, i64 0, i64 6, i64 4), align 2, !tbaa !9
+  store i16 -7, i16* getelementptr inbounds ([10 x [7 x i16]], [10 x [7 x i16]]* @1, i64 0, i64 6, i64 5), align 2, !tbaa !9
+  store i16 -7, i16* getelementptr inbounds ([10 x [7 x i16]], [10 x [7 x i16]]* @1, i64 0, i64 6, i64 6), align 2, !tbaa !9
+  store i16 -7, i16* getelementptr inbounds ([10 x [7 x i16]], [10 x [7 x i16]]* @1, i64 0, i64 7, i64 0), align 2, !tbaa !9
+  store i16 -7, i16* getelementptr inbounds ([10 x [7 x i16]], [10 x [7 x i16]]* @1, i64 0, i64 7, i64 1), align 2, !tbaa !9
+  store i16 -7, i16* getelementptr inbounds ([10 x [7 x i16]], [10 x [7 x i16]]* @1, i64 0, i64 7, i64 2), align 2, !tbaa !9
+  store i16 -7, i16* getelementptr inbounds ([10 x [7 x i16]], [10 x [7 x i16]]* @1, i64 0, i64 7, i64 3), align 2, !tbaa !9
+  store i16 -7, i16* getelementptr inbounds ([10 x [7 x i16]], [10 x [7 x i16]]* @1, i64 0, i64 7, i64 4), align 2, !tbaa !9
+  store i16 -7, i16* getelementptr inbounds ([10 x [7 x i16]], [10 x [7 x i16]]* @1, i64 0, i64 7, i64 5), align 2, !tbaa !9
+  store i16 -7, i16* getelementptr inbounds ([10 x [7 x i16]], [10 x [7 x i16]]* @1, i64 0, i64 7, i64 6), align 2, !tbaa !9
+  store i16 -7, i16* getelementptr inbounds ([10 x [7 x i16]], [10 x [7 x i16]]* @1, i64 0, i64 8, i64 0), align 2, !tbaa !9
+  store i16 -7, i16* getelementptr inbounds ([10 x [7 x i16]], [10 x [7 x i16]]* @1, i64 0, i64 8, i64 1), align 2, !tbaa !9
+  store i16 -7, i16* getelementptr inbounds ([10 x [7 x i16]], [10 x [7 x i16]]* @1, i64 0, i64 8, i64 2), align 2, !tbaa !9
+  store i16 -7, i16* getelementptr inbounds ([10 x [7 x i16]], [10 x [7 x i16]]* @1, i64 0, i64 8, i64 3), align 2, !tbaa !9
+  store i16 -7, i16* getelementptr inbounds ([10 x [7 x i16]], [10 x [7 x i16]]* @1, i64 0, i64 8, i64 4), align 2, !tbaa !9
+  store i16 -7, i16* getelementptr inbounds ([10 x [7 x i16]], [10 x [7 x i16]]* @1, i64 0, i64 8, i64 5), align 2, !tbaa !9
+  store i16 -7, i16* getelementptr inbounds ([10 x [7 x i16]], [10 x [7 x i16]]* @1, i64 0, i64 8, i64 6), align 2, !tbaa !9
+  store i16 -7, i16* getelementptr inbounds ([10 x [7 x i16]], [10 x [7 x i16]]* @1, i64 0, i64 9, i64 0), align 2, !tbaa !9
+  store i16 -7, i16* getelementptr inbounds ([10 x [7 x i16]], [10 x [7 x i16]]* @1, i64 0, i64 9, i64 1), align 2, !tbaa !9
+  store i16 -7, i16* getelementptr inbounds ([10 x [7 x i16]], [10 x [7 x i16]]* @1, i64 0, i64 9, i64 2), align 2, !tbaa !9
+  store i16 -7, i16* getelementptr inbounds ([10 x [7 x i16]], [10 x [7 x i16]]* @1, i64 0, i64 9, i64 3), align 2, !tbaa !9
+  store i16 -7, i16* getelementptr inbounds ([10 x [7 x i16]], [10 x [7 x i16]]* @1, i64 0, i64 9, i64 4), align 2, !tbaa !9
+  store i16 -7, i16* getelementptr inbounds ([10 x [7 x i16]], [10 x [7 x i16]]* @1, i64 0, i64 9, i64 5), align 2, !tbaa !9
+  store i16 -7, i16* getelementptr inbounds ([10 x [7 x i16]], [10 x [7 x i16]]* @1, i64 0, i64 9, i64 6), align 2, !tbaa !9
+  br label %71
+
+; <label>:71:                                     ; preds = %90, %68
+  %72 = phi i16 [ -7, %68 ], [ %92, %90 ]
+  %73 = phi i64 [ 0, %68 ], [ %88, %90 ]
+  %74 = getelementptr inbounds [10 x [7 x i16]], [10 x [7 x i16]]* @1, i64 0, i64 %73, i64 5
+  %75 = sext i16 %72 to i64
+  tail call fastcc void @8(i64 %75, i8* getelementptr inbounds ([12 x i8], [12 x i8]* @2, i64 0, i64 0), i32 signext %21)
+  %76 = load i16, i16* %74, align 2, !tbaa !9
+  %77 = sext i16 %76 to i64
+  tail call fastcc void @8(i64 %77, i8* getelementptr inbounds ([12 x i8], [12 x i8]* @2, i64 0, i64 0), i32 signext %21)
+  %78 = load i16, i16* %74, align 2, !tbaa !9
+  %79 = sext i16 %78 to i64
+  tail call fastcc void @8(i64 %79, i8* getelementptr inbounds ([12 x i8], [12 x i8]* @2, i64 0, i64 0), i32 signext %21)
+  %80 = load i16, i16* %74, align 2, !tbaa !9
+  %81 = sext i16 %80 to i64
+  tail call fastcc void @8(i64 %81, i8* getelementptr inbounds ([12 x i8], [12 x i8]* @2, i64 0, i64 0), i32 signext %21)
+  %82 = load i16, i16* %74, align 2, !tbaa !9
+  %83 = sext i16 %82 to i64
+  tail call fastcc void @8(i64 %83, i8* getelementptr inbounds ([12 x i8], [12 x i8]* @2, i64 0, i64 0), i32 signext %21)
+  %84 = load i16, i16* %74, align 2, !tbaa !9
+  %85 = sext i16 %84 to i64
+  tail call fastcc void @8(i64 %85, i8* getelementptr inbounds ([12 x i8], [12 x i8]* @2, i64 0, i64 0), i32 signext %21)
+  %86 = load i16, i16* %74, align 2, !tbaa !9
+  %87 = sext i16 %86 to i64
+  tail call fastcc void @8(i64 %87, i8* getelementptr inbounds ([12 x i8], [12 x i8]* @2, i64 0, i64 0), i32 signext %21)
+  %88 = add nuw nsw i64 %73, 1
+  %89 = icmp eq i64 %88, 10
+  br i1 %89, label %93, label %90
+
+; <label>:90:                                     ; preds = %71
+  %91 = getelementptr inbounds [10 x [7 x i16]], [10 x [7 x i16]]* @1, i64 0, i64 %88, i64 5
+  %92 = load i16, i16* %91, align 2, !tbaa !9
+  br label %71
+
+; <label>:93:                                     ; preds = %71
+  tail call fastcc void @8(i64 1430488437, i8* getelementptr inbounds ([9 x i8], [9 x i8]* @3, i64 0, i64 0), i32 signext %21)
+  %94 = load i32, i32* @4, align 4, !tbaa !7
+  %95 = xor i32 %94, -1
+  %96 = tail call signext i32 (i8*, ...) @printf(i8* getelementptr inbounds ([15 x i8], [15 x i8]* @7, i64 0, i64 0), i32 zeroext %95) #2
+  ret i32 0
+}
+
+; Function Attrs: nounwind
+define internal fastcc void @8(i64, i8*, i32 signext) unnamed_addr #0 {
+  %4 = trunc i64 %0 to i32
+  %5 = load i32, i32* @4, align 4, !tbaa !7
+  %6 = lshr i32 %5, 8
+  %7 = xor i32 %5, %4
+  %8 = and i32 %7, 255
+  %9 = zext i32 %8 to i64
+  %10 = getelementptr inbounds [256 x i32], [256 x i32]* @5, i64 0, i64 %9
+  %11 = load i32, i32* %10, align 4, !tbaa !7
+  %12 = xor i32 %6, %11
+  %13 = lshr i64 %0, 8
+  %14 = trunc i64 %13 to i32
+  %15 = lshr i32 %12, 8
+  %16 = xor i32 %12, %14
+  %17 = and i32 %16, 255
+  %18 = zext i32 %17 to i64
+  %19 = getelementptr inbounds [256 x i32], [256 x i32]* @5, i64 0, i64 %18
+  %20 = load i32, i32* %19, align 4, !tbaa !7
+  %21 = xor i32 %15, %20
+  %22 = lshr i64 %0, 16
+  %23 = trunc i64 %22 to i32
+  %24 = lshr i32 %21, 8
+  %25 = xor i32 %21, %23
+  %26 = and i32 %25, 255
+  %27 = zext i32 %26 to i64
+  %28 = getelementptr inbounds [256 x i32], [256 x i32]* @5, i64 0, i64 %27
+  %29 = load i32, i32* %28, align 4, !tbaa !7
+  %30 = xor i32 %24, %29
+  %31 = lshr i64 %0, 24
+  %32 = trunc i64 %31 to i32
+  %33 = lshr i32 %30, 8
+  %34 = xor i32 %30, %32
+  %35 = and i32 %34, 255
+  %36 = zext i32 %35 to i64
+  %37 = getelementptr inbounds [256 x i32], [256 x i32]* @5, i64 0, i64 %36
+  %38 = load i32, i32* %37, align 4, !tbaa !7
+  %39 = xor i32 %33, %38
+  %40 = lshr i64 %0, 32
+  %41 = trunc i64 %40 to i32
+  %42 = lshr i32 %39, 8
+  %43 = xor i32 %39, %41
+  %44 = and i32 %43, 255
+  %45 = zext i32 %44 to i64
+  %46 = getelementptr inbounds [256 x i32], [256 x i32]* @5, i64 0, i64 %45
+  %47 = load i32, i32* %46, align 4, !tbaa !7
+  %48 = xor i32 %42, %47
+  %49 = lshr i64 %0, 40
+  %50 = trunc i64 %49 to i32
+  %51 = lshr i32 %48, 8
+  %52 = xor i32 %48, %50
+  %53 = and i32 %52, 255
+  %54 = zext i32 %53 to i64
+  %55 = getelementptr inbounds [256 x i32], [256 x i32]* @5, i64 0, i64 %54
+  %56 = load i32, i32* %55, align 4, !tbaa !7
+  %57 = xor i32 %51, %56
+  %58 = lshr i64 %0, 48
+  %59 = trunc i64 %58 to i32
+  %60 = lshr i32 %57, 8
+  %61 = xor i32 %57, %59
+  %62 = and i32 %61, 255
+  %63 = zext i32 %62 to i64
+  %64 = getelementptr inbounds [256 x i32], [256 x i32]* @5, i64 0, i64 %63
+  %65 = load i32, i32* %64, align 4, !tbaa !7
+  %66 = xor i32 %60, %65
+  %67 = lshr i64 %0, 56
+  %68 = trunc i64 %67 to i32
+  %69 = lshr i32 %66, 8
+  %70 = and i32 %66, 255
+  %71 = xor i32 %70, %68
+  %72 = zext i32 %71 to i64
+  %73 = getelementptr inbounds [256 x i32], [256 x i32]* @5, i64 0, i64 %72
+  %74 = load i32, i32* %73, align 4, !tbaa !7
+  %75 = xor i32 %69, %74
+  store i32 %75, i32* @4, align 4, !tbaa !7
+  %76 = icmp eq i32 %2, 0
+  br i1 %76, label %81, label %77
+
+; <label>:77:                                     ; preds = %3
+  %78 = xor i32 %75, -1
+  %79 = zext i32 %78 to i64
+  %80 = tail call signext i32 (i8*, ...) @printf(i8* getelementptr inbounds ([36 x i8], [36 x i8]* @6, i64 0, i64 0), i8* %1, i64 %79)
+  br label %81
+
+; <label>:81:                                     ; preds = %77, %3
+  ret void
+}
+
+; Function Attrs: nounwind
+declare signext i32 @printf(i8* nocapture readonly, ...) local_unnamed_addr #1
+
+attributes #0 = { nounwind "correctly-rounded-divide-sqrt-fp-math"="false" "disable-tail-calls"="false" "less-precise-fpmad"="false" "no-frame-pointer-elim"="false" "no-infs-fp-math"="false" "no-jump-tables"="false" "no-nans-fp-math"="false" "no-signed-zeros-fp-math"="false" "no-trapping-math"="false" "stack-protector-buffer-size"="8" "target-cpu"="zEC12" "target-features"="+transactional-execution" "unsafe-fp-math"="false" "use-soft-float"="false" }
+attributes #1 = { nounwind "correctly-rounded-divide-sqrt-fp-math"="false" "disable-tail-calls"="false" "less-precise-fpmad"="false" "no-frame-pointer-elim"="false" "no-infs-fp-math"="false" "no-nans-fp-math"="false" "no-signed-zeros-fp-math"="false" "no-trapping-math"="false" "stack-protector-buffer-size"="8" "target-cpu"="zEC12" "target-features"="+transactional-execution" "unsafe-fp-math"="false" "use-soft-float"="false" }
+attributes #2 = { nounwind }
+
+!llvm.module.flags = !{!0}
+!llvm.ident = !{!1}
+
+!0 = !{i32 1, !"wchar_size", i32 4}
+!1 = !{!"clang version 6.0.0 (http://llvm.org/git/clang.git dd9e3a32396ac1f1807e7c14084e1233f36d3918) (http://llvm.org/git/llvm.git 470e6707258cbe72d1c2ebf9fcfd810dcf175906)"}
+!2 = !{!3, !3, i64 0}
+!3 = !{!"any pointer", !4, i64 0}
+!4 = !{!"omnipotent char", !5, i64 0}
+!5 = !{!"Simple C/C++ TBAA"}
+!6 = !{!4, !4, i64 0}
+!7 = !{!8, !8, i64 0}
+!8 = !{!"int", !4, i64 0}
+!9 = !{!10, !10, i64 0}
+!10 = !{!"short", !4, i64 0}