diff --git a/llvm/lib/ObjectYAML/DWARFYAML.cpp b/llvm/lib/ObjectYAML/DWARFYAML.cpp
--- a/llvm/lib/ObjectYAML/DWARFYAML.cpp
+++ b/llvm/lib/ObjectYAML/DWARFYAML.cpp
@@ -59,7 +59,7 @@
 DWARFYAML::Data::getAbbrevTableInfoByID(uint64_t ID) const {
   if (AbbrevTableInfoMap.empty()) {
     uint64_t AbbrevTableOffset = 0;
-    for (auto &AbbrevTable : enumerate(DebugAbbrev)) {
+    for (auto &&AbbrevTable : enumerate(DebugAbbrev)) {
       // If the abbrev table's ID isn't specified, we use the index as its ID.
       uint64_t AbbrevTableID =
           AbbrevTable.value().ID.value_or(AbbrevTable.index());
diff --git a/llvm/lib/ObjectYAML/MinidumpEmitter.cpp b/llvm/lib/ObjectYAML/MinidumpEmitter.cpp
--- a/llvm/lib/ObjectYAML/MinidumpEmitter.cpp
+++ b/llvm/lib/ObjectYAML/MinidumpEmitter.cpp
@@ -236,7 +236,7 @@
   Obj.Header.StreamDirectoryRVA = File.allocateArray(ArrayRef(StreamDirectory));
   Obj.Header.NumberOfStreams = StreamDirectory.size();
 
-  for (auto &Stream : enumerate(Obj.Streams))
+  for (auto &&Stream : enumerate(Obj.Streams))
     StreamDirectory[Stream.index()] = layout(File, *Stream.value());
 
   File.writeTo(Out);
diff --git a/llvm/lib/Transforms/Vectorize/SLPVectorizer.cpp b/llvm/lib/Transforms/Vectorize/SLPVectorizer.cpp
--- a/llvm/lib/Transforms/Vectorize/SLPVectorizer.cpp
+++ b/llvm/lib/Transforms/Vectorize/SLPVectorizer.cpp
@@ -3953,7 +3953,7 @@
         return std::get<1>(X) < std::get<1>(Y);
       });
       int InitialOffset = std::get<1>(Vec[0]);
-      AnyConsecutive |= all_of(enumerate(Vec), [InitialOffset](auto &P) {
+      AnyConsecutive |= all_of(enumerate(Vec), [InitialOffset](auto &&P) {
         return std::get<1>(P.value()) == int(P.index()) + InitialOffset;
       });
     }
diff --git a/llvm/tools/llvm-reduce/deltas/ReduceArguments.cpp b/llvm/tools/llvm-reduce/deltas/ReduceArguments.cpp
--- a/llvm/tools/llvm-reduce/deltas/ReduceArguments.cpp
+++ b/llvm/tools/llvm-reduce/deltas/ReduceArguments.cpp
@@ -101,7 +101,7 @@
       continue;
 
     std::set<int> ArgIndexesToKeep;
-    for (auto &Arg : enumerate(F->args()))
+    for (auto &&Arg : enumerate(F->args()))
       if (ArgsToKeep.count(&Arg.value()))
         ArgIndexesToKeep.insert(Arg.index());
 
diff --git a/llvm/unittests/DebugInfo/CodeView/RandomAccessVisitorTest.cpp b/llvm/unittests/DebugInfo/CodeView/RandomAccessVisitorTest.cpp
--- a/llvm/unittests/DebugInfo/CodeView/RandomAccessVisitorTest.cpp
+++ b/llvm/unittests/DebugInfo/CodeView/RandomAccessVisitorTest.cpp
@@ -280,7 +280,7 @@
 
   // 2, 4, 7
   EXPECT_EQ(3u, TestState->Callbacks.count());
-  for (auto &I : enumerate(IndicesToVisit))
+  for (auto &&I : enumerate(IndicesToVisit))
     EXPECT_TRUE(ValidateVisitedRecord(I.index(), I.value()));
 }
 
@@ -341,7 +341,7 @@
 
   // 5, 7
   EXPECT_EQ(2u, TestState->Callbacks.count());
-  for (auto &I : enumerate(IndicesToVisit))
+  for (auto &&I : enumerate(IndicesToVisit))
     EXPECT_TRUE(ValidateVisitedRecord(I.index(), I.value()));
 }
 
diff --git a/mlir/lib/Analysis/DataFlow/SparseAnalysis.cpp b/mlir/lib/Analysis/DataFlow/SparseAnalysis.cpp
--- a/mlir/lib/Analysis/DataFlow/SparseAnalysis.cpp
+++ b/mlir/lib/Analysis/DataFlow/SparseAnalysis.cpp
@@ -191,7 +191,7 @@
             dyn_cast<BranchOpInterface>(predecessor->getTerminator())) {
       SuccessorOperands operands =
           branch.getSuccessorOperands(it.getSuccessorIndex());
-      for (auto &it : llvm::enumerate(argLattices)) {
+      for (auto &&it : llvm::enumerate(argLattices)) {
         if (Value operand = operands[it.index()]) {
           join(it.value(), *getLatticeElementFor(block, operand));
         } else {
diff --git a/mlir/lib/Bytecode/Writer/IRNumbering.cpp b/mlir/lib/Bytecode/Writer/IRNumbering.cpp
--- a/mlir/lib/Bytecode/Writer/IRNumbering.cpp
+++ b/mlir/lib/Bytecode/Writer/IRNumbering.cpp
@@ -98,7 +98,7 @@
   }
 
   // Assign the entry numbers based on the sort order.
-  for (auto &entry : llvm::enumerate(range))
+  for (auto &&entry : llvm::enumerate(range))
     entry.value()->number = entry.index();
 }
 
@@ -132,7 +132,7 @@
   // found, given that the number of dialects on average is small enough to fit
   // within a singly byte (128). If we ever have real world use cases that have
   // a huge number of dialects, this could be made more intelligent.
-  for (auto &it : llvm::enumerate(dialects))
+  for (auto &&it : llvm::enumerate(dialects))
     it.value().second->number = it.index();
 
   // Number each of the recorded components within each dialect.
@@ -245,7 +245,7 @@
 
   // Number the blocks within this region.
   size_t blockCount = 0;
-  for (auto &it : llvm::enumerate(region)) {
+  for (auto &&it : llvm::enumerate(region)) {
     blockIDs.try_emplace(&it.value(), it.index());
     number(it.value());
     ++blockCount;
diff --git a/mlir/lib/Conversion/AMDGPUToROCDL/AMDGPUToROCDL.cpp b/mlir/lib/Conversion/AMDGPUToROCDL/AMDGPUToROCDL.cpp
--- a/mlir/lib/Conversion/AMDGPUToROCDL/AMDGPUToROCDL.cpp
+++ b/mlir/lib/Conversion/AMDGPUToROCDL/AMDGPUToROCDL.cpp
@@ -200,7 +200,7 @@
 
     // Indexing (voffset)
     Value voffset;
-    for (auto &pair : llvm::enumerate(adaptor.getIndices())) {
+    for (auto &&pair : llvm::enumerate(adaptor.getIndices())) {
       size_t i = pair.index();
       Value index = pair.value();
       Value strideOp;
diff --git a/mlir/lib/Conversion/FuncToLLVM/FuncToLLVM.cpp b/mlir/lib/Conversion/FuncToLLVM/FuncToLLVM.cpp
--- a/mlir/lib/Conversion/FuncToLLVM/FuncToLLVM.cpp
+++ b/mlir/lib/Conversion/FuncToLLVM/FuncToLLVM.cpp
@@ -134,7 +134,7 @@
 
   SmallVector<Value, 8> args;
   size_t argOffset = resultStructType ? 1 : 0;
-  for (auto &[index, argType] : llvm::enumerate(type.getInputs())) {
+  for (auto [index, argType] : llvm::enumerate(type.getInputs())) {
     Value arg = wrapperFuncOp.getArgument(index + argOffset);
     if (auto memrefType = argType.dyn_cast<MemRefType>()) {
       Value loaded = rewriter.create<LLVM::LoadOp>(
@@ -222,7 +222,7 @@
 
   // Iterate over the inputs of the original function and pack values into
   // memref descriptors if the original type is a memref.
-  for (auto &en : llvm::enumerate(type.getInputs())) {
+  for (auto &&en : llvm::enumerate(type.getInputs())) {
     Value arg;
     int numToDrop = 1;
     auto memRefType = en.value().dyn_cast<MemRefType>();
@@ -677,7 +677,7 @@
         getTypeConverter()->packFunctionResults(op.getOperandTypes());
 
     Value packed = rewriter.create<LLVM::UndefOp>(loc, packedType);
-    for (auto &it : llvm::enumerate(updatedOperands)) {
+    for (auto &&it : llvm::enumerate(updatedOperands)) {
       packed = rewriter.create<LLVM::InsertValueOp>(loc, packed, it.value(),
                                                     it.index());
     }
diff --git a/mlir/lib/Conversion/LLVMCommon/TypeConverter.cpp b/mlir/lib/Conversion/LLVMCommon/TypeConverter.cpp
--- a/mlir/lib/Conversion/LLVMCommon/TypeConverter.cpp
+++ b/mlir/lib/Conversion/LLVMCommon/TypeConverter.cpp
@@ -228,7 +228,7 @@
                               ? barePtrFuncArgTypeConverter
                               : structFuncArgTypeConverter;
   // Convert argument types one by one and check for errors.
-  for (auto &en : llvm::enumerate(funcTy.getInputs())) {
+  for (auto &&en : llvm::enumerate(funcTy.getInputs())) {
     Type type = en.value();
     SmallVector<Type, 8> converted;
     if (failed(funcArgConverter(*this, type, converted)))
diff --git a/mlir/lib/Conversion/PDLToPDLInterp/PredicateTree.cpp b/mlir/lib/Conversion/PDLToPDLInterp/PredicateTree.cpp
--- a/mlir/lib/Conversion/PDLToPDLInterp/PredicateTree.cpp
+++ b/mlir/lib/Conversion/PDLToPDLInterp/PredicateTree.cpp
@@ -187,7 +187,7 @@
                       builder.getType(builder.getAllResults(opPos)));
   } else {
     bool foundVariableLength = false;
-    for (auto &resultIt : llvm::enumerate(types)) {
+    for (auto &&resultIt : llvm::enumerate(types)) {
       bool isVariadic = resultIt.value().getType().isa<pdl::RangeType>();
       foundVariableLength |= isVariadic;
 
diff --git a/mlir/lib/Dialect/Affine/Analysis/Utils.cpp b/mlir/lib/Dialect/Affine/Analysis/Utils.cpp
--- a/mlir/lib/Dialect/Affine/Analysis/Utils.cpp
+++ b/mlir/lib/Dialect/Affine/Analysis/Utils.cpp
@@ -126,7 +126,7 @@
     llvm::errs() << "\t\t" << iv << "\n";
 
   llvm::errs() << "\tLBs:\n";
-  for (auto &en : llvm::enumerate(lbs)) {
+  for (auto &&en : llvm::enumerate(lbs)) {
     llvm::errs() << "\t\t" << en.value() << "\n";
     llvm::errs() << "\t\tOperands:\n";
     for (Value lbOp : lbOperands[en.index()])
@@ -134,7 +134,7 @@
   }
 
   llvm::errs() << "\tUBs:\n";
-  for (auto &en : llvm::enumerate(ubs)) {
+  for (auto &&en : llvm::enumerate(ubs)) {
     llvm::errs() << "\t\t" << en.value() << "\n";
     llvm::errs() << "\t\tOperands:\n";
     for (Value ubOp : ubOperands[en.index()])
diff --git a/mlir/lib/Dialect/LLVMIR/IR/LLVMDialect.cpp b/mlir/lib/Dialect/LLVMIR/IR/LLVMDialect.cpp
--- a/mlir/lib/Dialect/LLVMIR/IR/LLVMDialect.cpp
+++ b/mlir/lib/Dialect/LLVMIR/IR/LLVMDialect.cpp
@@ -2408,7 +2408,7 @@
   // Canonicalize any dynamic indices of constant value to constant indices.
   bool changed = false;
   SmallVector<GEPArg> gepArgs;
-  for (auto &iter : llvm::enumerate(indices)) {
+  for (auto &&iter : llvm::enumerate(indices)) {
     auto integer = iter.value().dyn_cast_or_null<IntegerAttr>();
     // Constant indices can only be int32_t, so if integer does not fit we
     // are forced to keep it dynamic, despite being a constant.
diff --git a/mlir/lib/Dialect/Linalg/TransformOps/LinalgTransformOps.cpp b/mlir/lib/Dialect/Linalg/TransformOps/LinalgTransformOps.cpp
--- a/mlir/lib/Dialect/Linalg/TransformOps/LinalgTransformOps.cpp
+++ b/mlir/lib/Dialect/Linalg/TransformOps/LinalgTransformOps.cpp
@@ -2453,7 +2453,7 @@
   SmallVector<Operation *> tiled;
   SmallVector<SmallVector<Operation *, 4>, 4> loops;
   loops.resize(getLoops().size());
-  for (auto &[i, op] : llvm::enumerate(targets)) {
+  for (auto [i, op] : llvm::enumerate(targets)) {
     auto tilingInterface = dyn_cast<TilingInterface>(op);
     auto dpsInterface = dyn_cast<DestinationStyleOpInterface>(op);
     if (!tilingInterface || !dpsInterface) {
@@ -2856,7 +2856,7 @@
   SmallVector<Operation *> tiled;
   SmallVector<SmallVector<Operation *, 4>, 4> loops;
   loops.resize(getLoops().size());
-  for (auto &en : llvm::enumerate(targets)) {
+  for (auto &&en : llvm::enumerate(targets)) {
     auto tilingInterfaceOp = dyn_cast<TilingInterface>(en.value());
     if (!tilingInterfaceOp) {
       DiagnosedSilenceableFailure diag =
diff --git a/mlir/lib/Dialect/Linalg/Transforms/ElementwiseOpFusion.cpp b/mlir/lib/Dialect/Linalg/Transforms/ElementwiseOpFusion.cpp
--- a/mlir/lib/Dialect/Linalg/Transforms/ElementwiseOpFusion.cpp
+++ b/mlir/lib/Dialect/Linalg/Transforms/ElementwiseOpFusion.cpp
@@ -1394,7 +1394,7 @@
   //   i1 = (i_{folded} / d2) % d1
   //   i0 = i_{folded} / (d1 * d2)
   llvm::DenseMap<unsigned, Value> indexReplacementVals;
-  for (auto &foldedDims :
+  for (auto &&foldedDims :
        enumerate(collapsingInfo.getCollapsedOpToOrigOpMapping())) {
     ReassociationIndicesRef foldedDimsRef(foldedDims.value());
     Value newIndexVal =
diff --git a/mlir/lib/Dialect/Linalg/Transforms/SplitReduction.cpp b/mlir/lib/Dialect/Linalg/Transforms/SplitReduction.cpp
--- a/mlir/lib/Dialect/Linalg/Transforms/SplitReduction.cpp
+++ b/mlir/lib/Dialect/Linalg/Transforms/SplitReduction.cpp
@@ -156,7 +156,7 @@
   newMaps.push_back(AffineMap::get(oldOutputMap.getNumDims() + 1, 0, outputExpr,
                                    op.getContext()));
   SmallVector<utils::IteratorType> newIteratorTypes;
-  for (auto &it : llvm::enumerate(op.getIteratorTypesArray())) {
+  for (auto &&it : llvm::enumerate(op.getIteratorTypesArray())) {
     if (insertSplitDimension == it.index())
       newIteratorTypes.push_back(utils::IteratorType::parallel);
     newIteratorTypes.push_back(it.value());
diff --git a/mlir/lib/Dialect/Linalg/Transforms/Tiling.cpp b/mlir/lib/Dialect/Linalg/Transforms/Tiling.cpp
--- a/mlir/lib/Dialect/Linalg/Transforms/Tiling.cpp
+++ b/mlir/lib/Dialect/Linalg/Transforms/Tiling.cpp
@@ -89,7 +89,7 @@
     RewriterBase &b, LinalgOp op, SmallVectorImpl<Value> &ivs,
     const LoopIndexToRangeIndexMap &loopIndexToRangeIndex) {
   SmallVector<Value> allIvs(op.getNumLoops(), nullptr);
-  for (auto &en : enumerate(allIvs)) {
+  for (auto &&en : enumerate(allIvs)) {
     auto rangeIndex = loopIndexToRangeIndex.find(en.index());
     if (rangeIndex == loopIndexToRangeIndex.end())
       continue;
diff --git a/mlir/lib/Dialect/Linalg/Transforms/TilingInterfaceImpl.cpp b/mlir/lib/Dialect/Linalg/Transforms/TilingInterfaceImpl.cpp
--- a/mlir/lib/Dialect/Linalg/Transforms/TilingInterfaceImpl.cpp
+++ b/mlir/lib/Dialect/Linalg/Transforms/TilingInterfaceImpl.cpp
@@ -314,7 +314,7 @@
     AffineMap oldOutputMap =
         linalgOp.getMatchingIndexingMap(linalgOp.getDpsInitOperand(0));
     SmallVector<AffineExpr> outputExpr;
-    for (auto &[idx, expr] : llvm::enumerate(oldOutputMap.getResults())) {
+    for (auto [idx, expr] : llvm::enumerate(oldOutputMap.getResults())) {
       if (static_cast<int64_t>(idx) == insertSplitDimension) {
         outputExpr.push_back(b.getAffineDimExpr(reductionDims[0]));
       }
diff --git a/mlir/lib/Dialect/MemRef/IR/MemRefOps.cpp b/mlir/lib/Dialect/MemRef/IR/MemRefOps.cpp
--- a/mlir/lib/Dialect/MemRef/IR/MemRefOps.cpp
+++ b/mlir/lib/Dialect/MemRef/IR/MemRefOps.cpp
@@ -1871,7 +1871,7 @@
            << srcType << " and result memref type " << resultType;
 
   // Match sizes in result memref type and in static_sizes attribute.
-  for (auto &en :
+  for (auto &&en :
        llvm::enumerate(llvm::zip(resultType.getShape(), getStaticSizes()))) {
     int64_t resultSize = std::get<0>(en.value());
     int64_t expectedSize = std::get<1>(en.value());
@@ -1900,7 +1900,7 @@
            << resultOffset << " instead of " << expectedOffset;
 
   // Match strides in result memref type and in static_strides attribute.
-  for (auto &en :
+  for (auto &&en :
        llvm::enumerate(llvm::zip(resultStrides, getStaticStrides()))) {
     int64_t resultStride = std::get<0>(en.value());
     int64_t expectedStride = std::get<1>(en.value());
diff --git a/mlir/lib/Dialect/SCF/IR/SCF.cpp b/mlir/lib/Dialect/SCF/IR/SCF.cpp
--- a/mlir/lib/Dialect/SCF/IR/SCF.cpp
+++ b/mlir/lib/Dialect/SCF/IR/SCF.cpp
@@ -3673,7 +3673,7 @@
 
   if (failed(verifyRegion(getDefaultRegion(), "default region")))
     return failure();
-  for (auto &[idx, caseRegion] : llvm::enumerate(getCaseRegions()))
+  for (auto [idx, caseRegion] : llvm::enumerate(getCaseRegions()))
     if (failed(verifyRegion(caseRegion, "case region #" + Twine(idx))))
       return failure();
 
diff --git a/mlir/lib/Dialect/SCF/Transforms/TileUsingInterface.cpp b/mlir/lib/Dialect/SCF/Transforms/TileUsingInterface.cpp
--- a/mlir/lib/Dialect/SCF/Transforms/TileUsingInterface.cpp
+++ b/mlir/lib/Dialect/SCF/Transforms/TileUsingInterface.cpp
@@ -423,7 +423,7 @@
     return b.notifyMatchFailure(
         op, "only support ops with one reduction dimension.");
   int reductionDim;
-  for (auto &[idx, iteratorType] :
+  for (auto [idx, iteratorType] :
        llvm::enumerate(tilingInterfaceOp.getLoopIteratorTypes())) {
     if (iteratorType == utils::IteratorType::reduction) {
       reductionDim = idx;
diff --git a/mlir/lib/Dialect/SparseTensor/Transforms/SparseTensorCodegen.cpp b/mlir/lib/Dialect/SparseTensor/Transforms/SparseTensorCodegen.cpp
--- a/mlir/lib/Dialect/SparseTensor/Transforms/SparseTensorCodegen.cpp
+++ b/mlir/lib/Dialect/SparseTensor/Transforms/SparseTensorCodegen.cpp
@@ -1333,7 +1333,7 @@
                      {reader, dimSizes}, EmitCInterface::On)
           .getResult(0);
       ArrayRef<int64_t> dstShape = dstTp.getRankedTensorType().getShape();
-      for (auto &d : llvm::enumerate(dstShape)) {
+      for (auto &&d : llvm::enumerate(dstShape)) {
         if (d.value() == ShapedType::kDynamic) {
           dynSizes.push_back(rewriter.create<memref::LoadOp>(
               loc, dimSizes, constantIndex(rewriter, loc, d.index())));
diff --git a/mlir/lib/Dialect/SparseTensor/Transforms/SparseTensorRewriting.cpp b/mlir/lib/Dialect/SparseTensor/Transforms/SparseTensorRewriting.cpp
--- a/mlir/lib/Dialect/SparseTensor/Transforms/SparseTensorRewriting.cpp
+++ b/mlir/lib/Dialect/SparseTensor/Transforms/SparseTensorRewriting.cpp
@@ -377,7 +377,7 @@
       ArrayRef<int64_t> dstShape = dstTp.getShape();
       genReshapeDstShape(loc, rewriter, dstSizes, srcSizes, dstShape,
                          op.getReassociationIndices());
-      for (auto &d : llvm::enumerate(dstShape)) {
+      for (auto &&d : llvm::enumerate(dstShape)) {
         if (d.value() == ShapedType::kDynamic)
           dstDynSizes.push_back(dstSizes[d.index()]);
       }
diff --git a/mlir/lib/Dialect/Tensor/IR/TensorInferTypeOpInterfaceImpl.cpp b/mlir/lib/Dialect/Tensor/IR/TensorInferTypeOpInterfaceImpl.cpp
--- a/mlir/lib/Dialect/Tensor/IR/TensorInferTypeOpInterfaceImpl.cpp
+++ b/mlir/lib/Dialect/Tensor/IR/TensorInferTypeOpInterfaceImpl.cpp
@@ -91,7 +91,7 @@
                         .cast<AffineDimExpr>()
                         .getPosition();
   int64_t linearizedStaticDim = 1;
-  for (auto &d :
+  for (auto &&d :
        llvm::enumerate(dstStaticShape.slice(startPos, endPos - startPos + 1))) {
     if (d.index() + startPos == static_cast<unsigned>(dimIndex))
       continue;
diff --git a/mlir/lib/Dialect/Tensor/IR/TensorOps.cpp b/mlir/lib/Dialect/Tensor/IR/TensorOps.cpp
--- a/mlir/lib/Dialect/Tensor/IR/TensorOps.cpp
+++ b/mlir/lib/Dialect/Tensor/IR/TensorOps.cpp
@@ -2779,7 +2779,7 @@
     // zero-offset and zero-padding tensor::ExtractSliceOp, tensor::PadOp pair
     // exists.
     SmallVector<OpFoldResult> newOffsets(rank, rewriter.getIndexAttr(0));
-    for (auto &en : enumerate(newOffsets)) {
+    for (auto &&en : enumerate(newOffsets)) {
       OpFoldResult innerOffset = innerSliceOp.getMixedOffsets()[en.index()];
       OpFoldResult outerOffset = outerSliceOp.getMixedOffsets()[en.index()];
       if (!innerDims.test(en.index()) &&
@@ -2802,7 +2802,7 @@
     // tensor::ExtractSliceOp does not match the size of the padded dimension.
     // Otherwise, take the size of the inner tensor::ExtractSliceOp.
     SmallVector<OpFoldResult> newSizes = innerSliceOp.getMixedSizes();
-    for (auto &en : enumerate(newSizes)) {
+    for (auto &&en : enumerate(newSizes)) {
       if (!outerDims.test(en.index()))
         continue;
       OpFoldResult sliceSize = innerSliceOp.getMixedSizes()[en.index()];
@@ -2819,7 +2819,7 @@
 
     // Combine the high paddings of the two tensor::PadOps.
     SmallVector<OpFoldResult> newHighPad(rank, rewriter.getIndexAttr(0));
-    for (auto &en : enumerate(newHighPad)) {
+    for (auto &&en : enumerate(newHighPad)) {
       if (innerDims.test(en.index()))
         newHighPad[en.index()] = padOp.getMixedHighPad()[en.index()];
       if (outerDims.test(en.index()))
diff --git a/mlir/lib/Dialect/Tensor/Transforms/BufferizableOpInterfaceImpl.cpp b/mlir/lib/Dialect/Tensor/Transforms/BufferizableOpInterfaceImpl.cpp
--- a/mlir/lib/Dialect/Tensor/Transforms/BufferizableOpInterfaceImpl.cpp
+++ b/mlir/lib/Dialect/Tensor/Transforms/BufferizableOpInterfaceImpl.cpp
@@ -762,7 +762,7 @@
           return isConstantIntValue(ofr, 0);
         });
     bool sizesMatchDestSizes = llvm::all_of(
-        llvm::enumerate(insertSliceOp.getMixedSizes()), [&](auto &it) {
+        llvm::enumerate(insertSliceOp.getMixedSizes()), [&](auto &&it) {
           return getConstantIntValue(it.value()) ==
                  destType.getDimSize(it.index());
         });
diff --git a/mlir/lib/Dialect/Tosa/Transforms/TosaFoldConstantTranspose.cpp b/mlir/lib/Dialect/Tosa/Transforms/TosaFoldConstantTranspose.cpp
--- a/mlir/lib/Dialect/Tosa/Transforms/TosaFoldConstantTranspose.cpp
+++ b/mlir/lib/Dialect/Tosa/Transforms/TosaFoldConstantTranspose.cpp
@@ -55,11 +55,11 @@
     // Transpose the input constant. Because we don't know its rank in advance,
     // we need to loop over the range [0, element count) and delinearize the
     // index.
-    auto attrValues = inputValues.getValues<Attribute>();
+    auto attrValuesIt = inputValues.value_begin<Attribute>();
     ArrayRef<int64_t> outputShape = outputType.getShape();
-    for (const auto &it : llvm::enumerate(attrValues)) {
+    for (int64_t totalCount = 0, e = inputValues.size(); totalCount != e;
+         ++totalCount) {
       SmallVector<uint64_t, 6> srcIndices(inputType.getRank(), 0);
-      int totalCount = it.index();
       for (int dim = inputType.getRank() - 1; dim >= 0; --dim) {
         srcIndices[dim] = totalCount % inputShape[dim];
         totalCount /= inputShape[dim];
@@ -73,7 +73,7 @@
       for (int dim = 1; dim < outputType.getRank(); ++dim)
         dstLinearIndex = dstLinearIndex * outputShape[dim] + dstIndices[dim];
 
-      outputValues[dstLinearIndex] = it.value();
+      outputValues[dstLinearIndex] = *(attrValuesIt + totalCount);
     }
 
     rewriter.replaceOpWithNewOp<tosa::ConstOp>(
diff --git a/mlir/lib/Dialect/Transform/IR/TransformInterfaces.cpp b/mlir/lib/Dialect/Transform/IR/TransformInterfaces.cpp
--- a/mlir/lib/Dialect/Transform/IR/TransformInterfaces.cpp
+++ b/mlir/lib/Dialect/Transform/IR/TransformInterfaces.cpp
@@ -964,7 +964,7 @@
       continue;
     assert(transformOp->getNumResults() == partialResults.size() &&
            "expected as many partial results as op as results");
-    for (auto &[i, value] : llvm::enumerate(partialResults))
+    for (auto [i, value] : llvm::enumerate(partialResults))
       transposed[i].push_back(value);
   }
 
diff --git a/mlir/lib/Dialect/Vector/Transforms/VectorDistribute.cpp b/mlir/lib/Dialect/Vector/Transforms/VectorDistribute.cpp
--- a/mlir/lib/Dialect/Vector/Transforms/VectorDistribute.cpp
+++ b/mlir/lib/Dialect/Vector/Transforms/VectorDistribute.cpp
@@ -205,7 +205,7 @@
       indices.push_back(yieldValues.size() - 1);
     } else {
       // If the value already exit the region don't create a new output.
-      for (auto &yieldOperand : llvm::enumerate(yieldValues.getArrayRef())) {
+      for (auto &&yieldOperand : llvm::enumerate(yieldValues.getArrayRef())) {
         if (yieldOperand.value() == std::get<0>(newRet)) {
           indices.push_back(yieldOperand.index());
           break;
diff --git a/mlir/lib/Dialect/Vector/Transforms/VectorUnroll.cpp b/mlir/lib/Dialect/Vector/Transforms/VectorUnroll.cpp
--- a/mlir/lib/Dialect/Vector/Transforms/VectorUnroll.cpp
+++ b/mlir/lib/Dialect/Vector/Transforms/VectorUnroll.cpp
@@ -628,7 +628,7 @@
       SmallVector<int64_t> permutedOffsets(elementOffsets.size());
       SmallVector<int64_t> permutedShape(elementOffsets.size());
       // Compute the source offsets and shape.
-      for (auto &indices : llvm::enumerate(permutation)) {
+      for (auto &&indices : llvm::enumerate(permutation)) {
         permutedOffsets[indices.value()] = elementOffsets[indices.index()];
         permutedShape[indices.value()] = (*targetShape)[indices.index()];
       }
diff --git a/mlir/lib/Dialect/X86Vector/Transforms/AVXTranspose.cpp b/mlir/lib/Dialect/X86Vector/Transforms/AVXTranspose.cpp
--- a/mlir/lib/Dialect/X86Vector/Transforms/AVXTranspose.cpp
+++ b/mlir/lib/Dialect/X86Vector/Transforms/AVXTranspose.cpp
@@ -257,7 +257,7 @@
       return rewriter.notifyMatchFailure(op, "Unsupported vector element type");
 
     SmallVector<int64_t> srcGtOneDims;
-    for (auto &en : llvm::enumerate(srcType.getShape()))
+    for (auto &&en : llvm::enumerate(srcType.getShape()))
       if (en.value() > 1)
         srcGtOneDims.push_back(en.index());
 
diff --git a/mlir/lib/ExecutionEngine/ExecutionEngine.cpp b/mlir/lib/ExecutionEngine/ExecutionEngine.cpp
--- a/mlir/lib/ExecutionEngine/ExecutionEngine.cpp
+++ b/mlir/lib/ExecutionEngine/ExecutionEngine.cpp
@@ -196,7 +196,7 @@
     llvm::Value *argList = interfaceFunc->arg_begin();
     SmallVector<llvm::Value *, 8> args;
     args.reserve(llvm::size(func.args()));
-    for (auto &indexedArg : llvm::enumerate(func.args())) {
+    for (auto &&indexedArg : llvm::enumerate(func.args())) {
       llvm::Value *argIndex = llvm::Constant::getIntegerValue(
           builder.getInt64Ty(), APInt(64, indexedArg.index()));
       llvm::Value *argPtrPtr =
diff --git a/mlir/lib/IR/AsmPrinter.cpp b/mlir/lib/IR/AsmPrinter.cpp
--- a/mlir/lib/IR/AsmPrinter.cpp
+++ b/mlir/lib/IR/AsmPrinter.cpp
@@ -3274,7 +3274,7 @@
   // One value might be used as the operand of an operation more than once.
   // Only print the operations results once in that case.
   SmallPtrSet<Operation *, 1> userSet;
-  for (auto &indexedUser : enumerate(value.getUsers())) {
+  for (auto &&indexedUser : enumerate(value.getUsers())) {
     if (userSet.insert(indexedUser.value()).second)
       printUserIDs(indexedUser.value(), indexedUser.index());
   }
diff --git a/mlir/lib/Pass/PassStatistics.cpp b/mlir/lib/Pass/PassStatistics.cpp
--- a/mlir/lib/Pass/PassStatistics.cpp
+++ b/mlir/lib/Pass/PassStatistics.cpp
@@ -73,7 +73,7 @@
         for (Pass::Statistic *it : pass->getStatistics())
           passEntry.push_back({it->getName(), it->getDesc(), it->getValue()});
       } else {
-        for (auto &it : llvm::enumerate(pass->getStatistics()))
+        for (auto &&it : llvm::enumerate(pass->getStatistics()))
           passEntry[it.index()].value += it.value()->getValue();
       }
 #endif
diff --git a/mlir/lib/TableGen/Operator.cpp b/mlir/lib/TableGen/Operator.cpp
--- a/mlir/lib/TableGen/Operator.cpp
+++ b/mlir/lib/TableGen/Operator.cpp
@@ -408,7 +408,7 @@
 
   // For all results whose types are buildable, initialize their type inference
   // nodes with an edge to themselves. Mark those nodes are fully-inferred.
-  for (auto &[idx, infer] : llvm::enumerate(inference)) {
+  for (auto &&[idx, infer] : llvm::enumerate(inference)) {
     if (getResult(idx).constraint.getBuilderCall()) {
       infer.sources.emplace_back(InferredResultType::mapResultIndex(idx),
                                  "$_self");
diff --git a/mlir/lib/Target/LLVMIR/ModuleTranslation.cpp b/mlir/lib/Target/LLVMIR/ModuleTranslation.cpp
--- a/mlir/lib/Target/LLVMIR/ModuleTranslation.cpp
+++ b/mlir/lib/Target/LLVMIR/ModuleTranslation.cpp
@@ -520,7 +520,7 @@
     auto phis = llvmBB->phis();
     auto numArguments = bb.getNumArguments();
     assert(numArguments == std::distance(phis.begin(), phis.end()));
-    for (auto &numberedPhiNode : llvm::enumerate(phis)) {
+    for (auto &&numberedPhiNode : llvm::enumerate(phis)) {
       auto &phiNode = numberedPhiNode.value();
       unsigned index = numberedPhiNode.index();
       for (auto *pred : bb.getPredecessors()) {
diff --git a/mlir/lib/Transforms/TopologicalSort.cpp b/mlir/lib/Transforms/TopologicalSort.cpp
--- a/mlir/lib/Transforms/TopologicalSort.cpp
+++ b/mlir/lib/Transforms/TopologicalSort.cpp
@@ -24,7 +24,7 @@
   void runOnOperation() override {
     // Topologically sort the regions of the operation without SSA dominance.
     getOperation()->walk([](RegionKindInterface op) {
-      for (auto &it : llvm::enumerate(op->getRegions())) {
+      for (auto &&it : llvm::enumerate(op->getRegions())) {
         if (op.hasSSADominance(it.index()))
           continue;
         for (Block &block : it.value())
diff --git a/mlir/lib/Transforms/Utils/ControlFlowSinkUtils.cpp b/mlir/lib/Transforms/Utils/ControlFlowSinkUtils.cpp
--- a/mlir/lib/Transforms/Utils/ControlFlowSinkUtils.cpp
+++ b/mlir/lib/Transforms/Utils/ControlFlowSinkUtils.cpp
@@ -136,7 +136,7 @@
                                           SmallVectorImpl<Region *> &regions) {
   // Collect constant operands.
   SmallVector<Attribute> operands(branch->getNumOperands(), Attribute());
-  for (auto &it : llvm::enumerate(branch->getOperands()))
+  for (auto &&it : llvm::enumerate(branch->getOperands()))
     (void)matchPattern(it.value(), m_Constant(&operands[it.index()]));
 
   // Get the invocation bounds.
diff --git a/mlir/test/lib/Analysis/DataFlow/TestDenseDataFlowAnalysis.cpp b/mlir/test/lib/Analysis/DataFlow/TestDenseDataFlowAnalysis.cpp
--- a/mlir/test/lib/Analysis/DataFlow/TestDenseDataFlowAnalysis.cpp
+++ b/mlir/test/lib/Analysis/DataFlow/TestDenseDataFlowAnalysis.cpp
@@ -254,7 +254,7 @@
       const LastModification *lastMods =
           solver.lookupState<LastModification>(op);
       assert(lastMods && "expected a dense lattice");
-      for (auto &it : llvm::enumerate(op->getOperands())) {
+      for (auto &&it : llvm::enumerate(op->getOperands())) {
         os << " operand #" << it.index() << "\n";
         Value value = getMostUnderlyingValue(it.value(), [&](Value value) {
           return solver.lookupState<UnderlyingValueLattice>(value);
diff --git a/mlir/tools/mlir-linalg-ods-gen/mlir-linalg-ods-yaml-gen.cpp b/mlir/tools/mlir-linalg-ods-gen/mlir-linalg-ods-yaml-gen.cpp
--- a/mlir/tools/mlir-linalg-ods-gen/mlir-linalg-ods-yaml-gen.cpp
+++ b/mlir/tools/mlir-linalg-ods-gen/mlir-linalg-ods-yaml-gen.cpp
@@ -869,7 +869,7 @@
           if (arg.kind != LinalgOperandDefKind::IndexAttr)
             continue;
           assert(arg.indexAttrMap);
-          for (auto &en :
+          for (auto &&en :
                llvm::enumerate(arg.indexAttrMap->affineMap().getResults())) {
             if (auto symbol = en.value().dyn_cast<AffineSymbolExpr>()) {
               std::string argName = arg.name;
diff --git a/mlir/tools/mlir-tblgen/AttrOrTypeFormatGen.cpp b/mlir/tools/mlir-tblgen/AttrOrTypeFormatGen.cpp
--- a/mlir/tools/mlir-tblgen/AttrOrTypeFormatGen.cpp
+++ b/mlir/tools/mlir-tblgen/AttrOrTypeFormatGen.cpp
@@ -951,7 +951,7 @@
 LogicalResult DefFormatParser::verify(SMLoc loc,
                                       ArrayRef<FormatElement *> elements) {
   // Check that all parameters are referenced in the format.
-  for (auto &it : llvm::enumerate(def.getParameters())) {
+  for (auto &&it : llvm::enumerate(def.getParameters())) {
     if (it.value().isOptional())
       continue;
     if (!seenParams.test(it.index())) {
diff --git a/mlir/tools/mlir-tblgen/EnumsGen.cpp b/mlir/tools/mlir-tblgen/EnumsGen.cpp
--- a/mlir/tools/mlir-tblgen/EnumsGen.cpp
+++ b/mlir/tools/mlir-tblgen/EnumsGen.cpp
@@ -123,7 +123,7 @@
   // case value to determine when to print in the string form.
   if (nonKeywordCases.any()) {
     os << "  switch (value) {\n";
-    for (auto &it : llvm::enumerate(cases)) {
+    for (auto &&it : llvm::enumerate(cases)) {
       if (nonKeywordCases.test(it.index()))
         continue;
       StringRef symbol = it.value().getSymbol();
diff --git a/mlir/tools/mlir-tblgen/OpDefinitionsGen.cpp b/mlir/tools/mlir-tblgen/OpDefinitionsGen.cpp
--- a/mlir/tools/mlir-tblgen/OpDefinitionsGen.cpp
+++ b/mlir/tools/mlir-tblgen/OpDefinitionsGen.cpp
@@ -934,7 +934,7 @@
   // Generate the <attr>AttrName methods, that expose the attribute names to
   // users.
   const char *attrNameMethodBody = "  return getAttributeNameForIndex({0});";
-  for (auto &attrIt : llvm::enumerate(llvm::make_first_range(attributes))) {
+  for (auto &&attrIt : llvm::enumerate(llvm::make_first_range(attributes))) {
     std::string name = op.getGetterName(attrIt.value());
     std::string methodName = name + "AttrName";
 
@@ -2801,7 +2801,7 @@
 
   body << "  {\n    unsigned index = 0; (void)index;\n";
 
-  for (auto &it : llvm::enumerate(successors)) {
+  for (auto &&it : llvm::enumerate(successors)) {
     const auto &successor = it.value();
     if (canSkip(successor))
       continue;
diff --git a/mlir/tools/mlir-tblgen/OpFormatGen.cpp b/mlir/tools/mlir-tblgen/OpFormatGen.cpp
--- a/mlir/tools/mlir-tblgen/OpFormatGen.cpp
+++ b/mlir/tools/mlir-tblgen/OpFormatGen.cpp
@@ -1827,7 +1827,7 @@
   // Get a string containing all of the cases that can't be represented with a
   // keyword.
   BitVector nonKeywordCases(cases.size());
-  for (auto &it : llvm::enumerate(cases)) {
+  for (auto &&it : llvm::enumerate(cases)) {
     if (!canFormatStringAsKeyword(it.value().getStr()))
       nonKeywordCases.set(it.index());
   }
@@ -1836,7 +1836,7 @@
   // overlap with other cases. For simplicity sake, only allow cases with a
   // single bit value.
   if (enumAttr.isBitEnum()) {
-    for (auto &it : llvm::enumerate(cases)) {
+    for (auto &&it : llvm::enumerate(cases)) {
       int64_t value = it.value().getValue();
       if (value < 0 || !llvm::isPowerOf2_64(value))
         nonKeywordCases.set(it.index());
@@ -1849,7 +1849,7 @@
     body << "    switch (caseValue) {\n";
     StringRef cppNamespace = enumAttr.getCppNamespace();
     StringRef enumName = enumAttr.getEnumClassName();
-    for (auto &it : llvm::enumerate(cases)) {
+    for (auto &&it : llvm::enumerate(cases)) {
       if (nonKeywordCases.test(it.index()))
         continue;
       StringRef symbol = it.value().getSymbol();