diff --git a/llvm/lib/Transforms/Instrumentation/MemorySanitizer.cpp b/llvm/lib/Transforms/Instrumentation/MemorySanitizer.cpp
--- a/llvm/lib/Transforms/Instrumentation/MemorySanitizer.cpp
+++ b/llvm/lib/Transforms/Instrumentation/MemorySanitizer.cpp
@@ -3264,11 +3264,6 @@
     Value *Mask = I.getArgOperand(3);
     Value *Shadow = getShadow(V);
 
-    Value *ShadowPtr;
-    Value *OriginPtr;
-    std::tie(ShadowPtr, OriginPtr) = getShadowOriginPtr(
-        Addr, IRB, Shadow->getType(), Alignment, /*isStore*/ true);
-
     if (ClCheckAccessAddress) {
       insertShadowCheck(Addr, &I);
       // Uninitialized mask is kind of like uninitialized address, but not as
@@ -3276,6 +3271,11 @@
       insertShadowCheck(Mask, &I);
     }
 
+    Value *ShadowPtr;
+    Value *OriginPtr;
+    std::tie(ShadowPtr, OriginPtr) = getShadowOriginPtr(
+        Addr, IRB, Shadow->getType(), Alignment, /*isStore*/ true);
+
     IRB.CreateMaskedStore(Shadow, ShadowPtr, Alignment, Mask);
 
     if (MS.TrackOrigins) {
@@ -3294,6 +3294,11 @@
     Value *Mask = I.getArgOperand(2);
     Value *PassThru = I.getArgOperand(3);
 
+    if (ClCheckAccessAddress) {
+      insertShadowCheck(Addr, &I);
+      insertShadowCheck(Mask, &I);
+    }
+
     Type *ShadowTy = getShadowTy(&I);
     Value *ShadowPtr, *OriginPtr;
     if (PropagateShadow) {
@@ -3305,11 +3310,6 @@
       setShadow(&I, getCleanShadow(&I));
     }
 
-    if (ClCheckAccessAddress) {
-      insertShadowCheck(Addr, &I);
-      insertShadowCheck(Mask, &I);
-    }
-
     if (MS.TrackOrigins) {
       if (PropagateShadow) {
         // Choose between PassThru's and the loaded value's origins.
diff --git a/llvm/test/Instrumentation/MemorySanitizer/masked-store-load.ll b/llvm/test/Instrumentation/MemorySanitizer/masked-store-load.ll
--- a/llvm/test/Instrumentation/MemorySanitizer/masked-store-load.ll
+++ b/llvm/test/Instrumentation/MemorySanitizer/masked-store-load.ll
@@ -97,16 +97,16 @@
 ;
 ; ADDR-LABEL: @Load(
 ; ADDR-NEXT:  entry:
-; ADDR-NEXT:    [[TMP0:%.*]] = load <4 x i64>, <4 x i64>* inttoptr (i64 add (i64 ptrtoint ([100 x i64]* @__msan_param_tls to i64), i64 8) to <4 x i64>*), align 8
-; ADDR-NEXT:    [[TMP1:%.*]] = load i64, i64* getelementptr inbounds ([100 x i64], [100 x i64]* @__msan_param_tls, i32 0, i32 0), align 8
-; ADDR-NEXT:    [[TMP2:%.*]] = load <4 x i1>, <4 x i1>* inttoptr (i64 add (i64 ptrtoint ([100 x i64]* @__msan_param_tls to i64), i64 40) to <4 x i1>*), align 8
+; ADDR-NEXT:    [[TMP0:%.*]] = load i64, i64* getelementptr inbounds ([100 x i64], [100 x i64]* @__msan_param_tls, i32 0, i32 0), align 8
+; ADDR-NEXT:    [[TMP1:%.*]] = load <4 x i1>, <4 x i1>* inttoptr (i64 add (i64 ptrtoint ([100 x i64]* @__msan_param_tls to i64), i64 40) to <4 x i1>*), align 8
+; ADDR-NEXT:    [[TMP2:%.*]] = load <4 x i64>, <4 x i64>* inttoptr (i64 add (i64 ptrtoint ([100 x i64]* @__msan_param_tls to i64), i64 8) to <4 x i64>*), align 8
 ; ADDR-NEXT:    call void @llvm.donothing()
 ; ADDR-NEXT:    [[TMP3:%.*]] = ptrtoint <4 x double>* [[P:%.*]] to i64
 ; ADDR-NEXT:    [[TMP4:%.*]] = xor i64 [[TMP3]], 87960930222080
 ; ADDR-NEXT:    [[TMP5:%.*]] = inttoptr i64 [[TMP4]] to <4 x i64>*
-; ADDR-NEXT:    [[_MSMASKEDLD:%.*]] = call <4 x i64> @llvm.masked.load.v4i64.p0v4i64(<4 x i64>* [[TMP5]], i32 1, <4 x i1> [[MASK:%.*]], <4 x i64> [[TMP0]])
-; ADDR-NEXT:    [[_MSCMP:%.*]] = icmp ne i64 [[TMP1]], 0
-; ADDR-NEXT:    [[TMP6:%.*]] = bitcast <4 x i1> [[TMP2]] to i4
+; ADDR-NEXT:    [[_MSMASKEDLD:%.*]] = call <4 x i64> @llvm.masked.load.v4i64.p0v4i64(<4 x i64>* [[TMP5]], i32 1, <4 x i1> [[MASK:%.*]], <4 x i64> [[TMP2]])
+; ADDR-NEXT:    [[_MSCMP:%.*]] = icmp ne i64 [[TMP0]], 0
+; ADDR-NEXT:    [[TMP6:%.*]] = bitcast <4 x i1> [[TMP1]] to i4
 ; ADDR-NEXT:    [[_MSCMP1:%.*]] = icmp ne i4 [[TMP6]], 0
 ; ADDR-NEXT:    [[_MSOR:%.*]] = or i1 [[_MSCMP]], [[_MSCMP1]]
 ; ADDR-NEXT:    br i1 [[_MSOR]], label [[TMP7:%.*]], label [[TMP8:%.*]], !prof [[PROF0]]