diff --git a/llvm/lib/Transforms/Scalar/ScalarizeMaskedMemIntrin.cpp b/llvm/lib/Transforms/Scalar/ScalarizeMaskedMemIntrin.cpp
--- a/llvm/lib/Transforms/Scalar/ScalarizeMaskedMemIntrin.cpp
+++ b/llvm/lib/Transforms/Scalar/ScalarizeMaskedMemIntrin.cpp
@@ -94,6 +94,11 @@
   return true;
 }
 
+static unsigned adjustForEndian(const DataLayout &DL, unsigned VectorWidth,
+                                unsigned Idx) {
+  return DL.isBigEndian() ? VectorWidth - 1 - Idx : Idx;
+}
+
 // Translate a masked load intrinsic like
 // <16 x i32 > @llvm.masked.load( <16 x i32>* %addr, i32 align,
 //                               <16 x i1> %mask, <16 x i32> %passthru)
@@ -126,7 +131,8 @@
 //  %10 = extractelement <16 x i1> %mask, i32 2
 //  br i1 %10, label %cond.load4, label %else5
 //
-static void scalarizeMaskedLoad(CallInst *CI, bool &ModifiedDT) {
+static void scalarizeMaskedLoad(const DataLayout &DL, CallInst *CI,
+                                bool &ModifiedDT) {
   Value *Ptr = CI->getArgOperand(0);
   Value *Alignment = CI->getArgOperand(1);
   Value *Mask = CI->getArgOperand(2);
@@ -195,7 +201,8 @@
     //
     Value *Predicate;
     if (VectorWidth != 1) {
-      Value *Mask = Builder.getInt(APInt::getOneBitSet(VectorWidth, Idx));
+      Value *Mask = Builder.getInt(APInt::getOneBitSet(
+          VectorWidth, adjustForEndian(DL, VectorWidth, Idx)));
       Predicate = Builder.CreateICmpNE(Builder.CreateAnd(SclrMask, Mask),
                                        Builder.getIntN(VectorWidth, 0));
     } else {
@@ -265,7 +272,8 @@
 //   store i32 %6, i32* %7
 //   br label %else2
 //   . . .
-static void scalarizeMaskedStore(CallInst *CI, bool &ModifiedDT) {
+static void scalarizeMaskedStore(const DataLayout &DL, CallInst *CI,
+                                 bool &ModifiedDT) {
   Value *Src = CI->getArgOperand(0);
   Value *Ptr = CI->getArgOperand(1);
   Value *Alignment = CI->getArgOperand(2);
@@ -327,7 +335,8 @@
     //
     Value *Predicate;
     if (VectorWidth != 1) {
-      Value *Mask = Builder.getInt(APInt::getOneBitSet(VectorWidth, Idx));
+      Value *Mask = Builder.getInt(APInt::getOneBitSet(
+          VectorWidth, adjustForEndian(DL, VectorWidth, Idx)));
       Predicate = Builder.CreateICmpNE(Builder.CreateAnd(SclrMask, Mask),
                                        Builder.getIntN(VectorWidth, 0));
     } else {
@@ -391,7 +400,8 @@
 // . . .
 // %Result = select <16 x i1> %Mask, <16 x i32> %res.phi.select, <16 x i32> %Src
 // ret <16 x i32> %Result
-static void scalarizeMaskedGather(CallInst *CI, bool &ModifiedDT) {
+static void scalarizeMaskedGather(const DataLayout &DL, CallInst *CI,
+                                  bool &ModifiedDT) {
   Value *Ptrs = CI->getArgOperand(0);
   Value *Alignment = CI->getArgOperand(1);
   Value *Mask = CI->getArgOperand(2);
@@ -446,7 +456,8 @@
 
     Value *Predicate;
     if (VectorWidth != 1) {
-      Value *Mask = Builder.getInt(APInt::getOneBitSet(VectorWidth, Idx));
+      Value *Mask = Builder.getInt(APInt::getOneBitSet(
+          VectorWidth, adjustForEndian(DL, VectorWidth, Idx)));
       Predicate = Builder.CreateICmpNE(Builder.CreateAnd(SclrMask, Mask),
                                        Builder.getIntN(VectorWidth, 0));
     } else {
@@ -515,7 +526,8 @@
 // store i32 %Elt1, i32* %Ptr1, align 4
 // br label %else2
 //   . . .
-static void scalarizeMaskedScatter(CallInst *CI, bool &ModifiedDT) {
+static void scalarizeMaskedScatter(const DataLayout &DL, CallInst *CI,
+                                   bool &ModifiedDT) {
   Value *Src = CI->getArgOperand(0);
   Value *Ptrs = CI->getArgOperand(1);
   Value *Alignment = CI->getArgOperand(2);
@@ -568,7 +580,8 @@
     //
     Value *Predicate;
     if (VectorWidth != 1) {
-      Value *Mask = Builder.getInt(APInt::getOneBitSet(VectorWidth, Idx));
+      Value *Mask = Builder.getInt(APInt::getOneBitSet(
+          VectorWidth, adjustForEndian(DL, VectorWidth, Idx)));
       Predicate = Builder.CreateICmpNE(Builder.CreateAnd(SclrMask, Mask),
                                        Builder.getIntN(VectorWidth, 0));
     } else {
@@ -601,7 +614,8 @@
   ModifiedDT = true;
 }
 
-static void scalarizeMaskedExpandLoad(CallInst *CI, bool &ModifiedDT) {
+static void scalarizeMaskedExpandLoad(const DataLayout &DL, CallInst *CI,
+                                      bool &ModifiedDT) {
   Value *Ptr = CI->getArgOperand(0);
   Value *Mask = CI->getArgOperand(1);
   Value *PassThru = CI->getArgOperand(2);
@@ -669,7 +683,8 @@
 
     Value *Predicate;
     if (VectorWidth != 1) {
-      Value *Mask = Builder.getInt(APInt::getOneBitSet(VectorWidth, Idx));
+      Value *Mask = Builder.getInt(APInt::getOneBitSet(
+          VectorWidth, adjustForEndian(DL, VectorWidth, Idx)));
       Predicate = Builder.CreateICmpNE(Builder.CreateAnd(SclrMask, Mask),
                                        Builder.getIntN(VectorWidth, 0));
     } else {
@@ -725,7 +740,8 @@
   ModifiedDT = true;
 }
 
-static void scalarizeMaskedCompressStore(CallInst *CI, bool &ModifiedDT) {
+static void scalarizeMaskedCompressStore(const DataLayout &DL, CallInst *CI,
+                                         bool &ModifiedDT) {
   Value *Src = CI->getArgOperand(0);
   Value *Ptr = CI->getArgOperand(1);
   Value *Mask = CI->getArgOperand(2);
@@ -775,7 +791,8 @@
     //
     Value *Predicate;
     if (VectorWidth != 1) {
-      Value *Mask = Builder.getInt(APInt::getOneBitSet(VectorWidth, Idx));
+      Value *Mask = Builder.getInt(APInt::getOneBitSet(
+          VectorWidth, adjustForEndian(DL, VectorWidth, Idx)));
       Predicate = Builder.CreateICmpNE(Builder.CreateAnd(SclrMask, Mask),
                                        Builder.getIntN(VectorWidth, 0));
     } else {
@@ -895,14 +912,14 @@
               CI->getType(),
               cast<ConstantInt>(CI->getArgOperand(1))->getAlignValue()))
         return false;
-      scalarizeMaskedLoad(CI, ModifiedDT);
+      scalarizeMaskedLoad(DL, CI, ModifiedDT);
       return true;
     case Intrinsic::masked_store:
       if (TTI.isLegalMaskedStore(
               CI->getArgOperand(0)->getType(),
               cast<ConstantInt>(CI->getArgOperand(2))->getAlignValue()))
         return false;
-      scalarizeMaskedStore(CI, ModifiedDT);
+      scalarizeMaskedStore(DL, CI, ModifiedDT);
       return true;
     case Intrinsic::masked_gather: {
       unsigned AlignmentInt =
@@ -912,7 +929,7 @@
           DL.getValueOrABITypeAlignment(MaybeAlign(AlignmentInt), LoadTy);
       if (TTI.isLegalMaskedGather(LoadTy, Alignment))
         return false;
-      scalarizeMaskedGather(CI, ModifiedDT);
+      scalarizeMaskedGather(DL, CI, ModifiedDT);
       return true;
     }
     case Intrinsic::masked_scatter: {
@@ -923,18 +940,18 @@
           DL.getValueOrABITypeAlignment(MaybeAlign(AlignmentInt), StoreTy);
       if (TTI.isLegalMaskedScatter(StoreTy, Alignment))
         return false;
-      scalarizeMaskedScatter(CI, ModifiedDT);
+      scalarizeMaskedScatter(DL, CI, ModifiedDT);
       return true;
     }
     case Intrinsic::masked_expandload:
       if (TTI.isLegalMaskedExpandLoad(CI->getType()))
         return false;
-      scalarizeMaskedExpandLoad(CI, ModifiedDT);
+      scalarizeMaskedExpandLoad(DL, CI, ModifiedDT);
       return true;
     case Intrinsic::masked_compressstore:
       if (TTI.isLegalMaskedCompressStore(CI->getArgOperand(0)->getType()))
         return false;
-      scalarizeMaskedCompressStore(CI, ModifiedDT);
+      scalarizeMaskedCompressStore(DL, CI, ModifiedDT);
       return true;
     }
   }