Replace deprecated use of .cast/dyn_cast/isa with the mlir freestandi…

…ng functions and misc cleanups.

Also remove some unecessary semicolons and standardize to use mlir instead of llvm casts

src/Conversion/ONNXToTOSA/DialectBuilder.cpp

@@ -176,7 +176,8 @@ std::optional<Value> TosaBuilder::gather(Value resultValue, Value inputValue,
     Value indicesValue, int32_t batchDims, int32_t axis) {
   return tosa::convertGatherOp(rewriter(), loc(), resultValue, inputValue,
       indicesValue, batchDims, axis);
-};
+}
+
 Value TosaBuilder::reshape(mlir::Value value, llvm::ArrayRef<int64_t> shape) {
   auto shapeAttr = rewriter().getDenseI64ArrayAttr(shape);
   auto valueType = mlir::cast<ShapedType>(value.getType());
@@ -246,7 +247,7 @@ template Value TosaBuilder::binaryOp<mlir::tosa::PowOp>(
 
 template <typename T>
 Value TosaBuilder::unaryOp(mlir::Value &input) {
-  auto inputType = input.getType().cast<ShapedType>();
+  auto inputType = cast<ShapedType>(input.getType());
   Type newValueType = RankedTensorType::get(
       llvm::SmallVector<int64_t, 4>(inputType.getRank(), ShapedType::kDynamic),
       inputType.getElementType());
@@ -304,7 +305,7 @@ Value TosaBuilder::select(
     lhs = valueVec[1];
     rhs = valueVec[2];
   }
-  auto lhsType = lhs.getType().cast<ShapedType>();
+  auto lhsType = cast<ShapedType>(lhs.getType());
   Type newValueType = RankedTensorType::get(
       llvm::SmallVector<int64_t, 4>(lhsType.getRank(), ShapedType::kDynamic),
       lhsType.getElementType());
@@ -326,7 +327,7 @@ mlir::Value TosaBuilder::castToNewTensorElementType(
 }
 
 Value TosaBuilder::sqrt(mlir::Value &input) {
-  auto inputType = input.getType().cast<ShapedType>();
+  auto inputType = cast<ShapedType>(input.getType());
   auto oneHalf = this->getSplattedConst(
       0.5, inputType.getElementType(), inputType.getShape());
   return this->binaryOp<mlir::tosa::PowOp>(input, oneHalf);

src/Conversion/ONNXToTOSA/Math/Conv2D.cpp

@@ -101,7 +101,7 @@ class ONNXConvOpLoweringToTOSA : public ConversionPattern {
       ConversionPatternRewriter &rewriter) const final {
     OpAdaptor adaptor(operands, op->getAttrDictionary());
     auto loc = op->getLoc();
-    auto convOp = llvm::cast<ONNXConvOp>(op);
+    auto convOp = mlir::cast<ONNXConvOp>(op);
 
     TosaBuilder tosaBuilder(rewriter, loc);
 

src/Conversion/ONNXToTOSA/Math/Elementwise.cpp

@@ -80,12 +80,12 @@ LogicalResult checkBasicTosaRequirementsForBinaryOps(
     ConversionPatternRewriter &rewriter, Operation *op, OpAdaptorT adaptor,
     Type resultType) {
   Value lhs = adaptor.getOperands()[0];
-  auto lhsType = lhs.getType().dyn_cast<TensorType>();
+  auto lhsType = dyn_cast<TensorType>(lhs.getType());
 
   Value rhs = adaptor.getOperands()[1];
-  auto rhsType = rhs.getType().dyn_cast<TensorType>();
+  auto rhsType = dyn_cast<TensorType>(rhs.getType());
 
-  auto resultTensorType = resultType.dyn_cast<TensorType>();
+  auto resultTensorType = dyn_cast<TensorType>(resultType);
   if (!lhsType || !rhsType || !resultTensorType) {
     return rewriter.notifyMatchFailure(op, "Tosa only supports TensorTypes");
   }
@@ -121,9 +121,9 @@ class ONNXElementwiseUnaryOpLoweringToTOSA
       ConversionPatternRewriter &rewriter) const override {
 
     Value input = *adaptor.getODSOperands(0).begin();
-    auto inputType = input.getType().dyn_cast<TensorType>();
+    auto inputType = dyn_cast<TensorType>(input.getType());
     Value output = op.getResult();
-    auto outputType = output.getType().dyn_cast<TensorType>();
+    auto outputType = dyn_cast<TensorType>(output.getType());
 
     if (!inputType || !outputType) {
       return rewriter.notifyMatchFailure(op, "Tosa only supports TensorTypes");
@@ -248,7 +248,7 @@ class ONNXLeakyReluOpLoweringToTOSA
   LogicalResult matchAndRewrite(ONNXLeakyReluOp op, OpAdaptor adaptor,
       ConversionPatternRewriter &rewriter) const override {
 
-    auto outputType = op.getResult().getType().cast<TensorType>();
+    auto outputType = cast<TensorType>(op.getResult().getType());
     if (failed(IsIntOrFloat::checkType(
             rewriter, outputType.getElementType(), op))) {
       return failure();
@@ -279,15 +279,13 @@ class ONNXComparisonOpLoweringToTOSA : public OpConversionPattern<OnnxCompOp> {
       ConversionPatternRewriter &rewriter) const override {
 
     Value input1 = adaptor.getA();
-    auto input1ElemType =
-        input1.getType().template cast<TensorType>().getElementType();
+    auto input1ElemType = cast<TensorType>(input1.getType()).getElementType();
     if (failed(IsIntOrFloat::checkType(rewriter, input1ElemType, op))) {
       return failure();
     }
 
     Value input2 = adaptor.getB();
-    auto input2ElemType =
-        input2.getType().template cast<TensorType>().getElementType();
+    auto input2ElemType = cast<TensorType>(input2.getType()).getElementType();
     if (input1ElemType != input2ElemType) {
       return failure();
     }
@@ -432,7 +430,7 @@ class ONNXSqrtOpLoweringToTOSA : public OpConversionPattern<ONNXSqrtOp> {
   LogicalResult matchAndRewrite(ONNXSqrtOp op, OpAdaptor adaptor,
       ConversionPatternRewriter &rewriter) const override {
 
-    auto resultTensorType = op.getResult().getType().cast<TensorType>();
+    auto resultTensorType = cast<TensorType>(op.getResult().getType());
     if (failed(IsFloat::checkType(
             rewriter, resultTensorType.getElementType(), op))) {
       return failure();
@@ -454,7 +452,7 @@ class ONNXEluOpLoweringToTOSA : public OpConversionPattern<ONNXEluOp> {
     // ELU(x) = x                     if x >= 0
     //         alpha * (exp(x) - 1.)  if x <  0
 
-    auto resultTensorType = op.getResult().getType().cast<TensorType>();
+    auto resultTensorType = cast<TensorType>(op.getResult().getType());
     if (failed(IsFloat::checkType(
             rewriter, resultTensorType.getElementType(), op))) {
       return failure();
@@ -496,7 +494,7 @@ class ONNXHardSigmoidOpLoweringToTOSA
     // - tosa.mul(clamp, alpha)
     Value input = adaptor.getX();
 
-    auto resultType = op.getResult().getType().template cast<TensorType>();
+    auto resultType = cast<TensorType>(op.getResult().getType());
     auto resultElementType = resultType.getElementType();
 
     TosaBuilder tosaBuilder(rewriter, op->getLoc());
@@ -536,7 +534,7 @@ class ONNXPReluOpLoweringToTOSA : public OpConversionPattern<ONNXPReluOp> {
   LogicalResult matchAndRewrite(ONNXPReluOp op, OpAdaptor adaptor,
       ConversionPatternRewriter &rewriter) const override {
 
-    auto outputType = op.getResult().getType().cast<TensorType>();
+    auto outputType = cast<TensorType>(op.getResult().getType());
     if (failed(IsIntOrFloat::checkType(
             rewriter, outputType.getElementType(), op))) {
       return failure();
@@ -554,7 +552,7 @@ class ONNXSoftplusOpLoweringToTOSA
   LogicalResult matchAndRewrite(ONNXSoftplusOp op, OpAdaptor adaptor,
       ConversionPatternRewriter &rewriter) const override {
 
-    auto outputType = op.getResult().getType().cast<TensorType>();
+    auto outputType = cast<TensorType>(op.getResult().getType());
     if (failed(IsFloat::checkType(rewriter, outputType.getElementType(), op))) {
       return failure();
     }
@@ -579,7 +577,7 @@ class ONNXSeluOpLoweringToTOSA : public OpConversionPattern<ONNXSeluOp> {
   LogicalResult matchAndRewrite(ONNXSeluOp op, OpAdaptor adaptor,
       ConversionPatternRewriter &rewriter) const override {
 
-    auto outputType = op.getResult().getType().cast<TensorType>();
+    auto outputType = cast<TensorType>(op.getResult().getType());
     if (failed(IsFloat::checkType(rewriter, outputType.getElementType(), op))) {
       return failure();
     }
@@ -618,7 +616,7 @@ class ONNXThresholdedReluOpLoweringToTOSA
   LogicalResult matchAndRewrite(ONNXThresholdedReluOp op, OpAdaptor adaptor,
       ConversionPatternRewriter &rewriter) const override {
 
-    auto outputType = op.getResult().getType().cast<TensorType>();
+    auto outputType = cast<TensorType>(op.getResult().getType());
     if (failed(IsIntOrFloat::checkType(
             rewriter, outputType.getElementType(), op))) {
       return failure();

src/Conversion/ONNXToTOSA/Math/MatMul.cpp

@@ -50,7 +50,7 @@ Value reshapeUpTo3DTensor(Value tensor, TosaBuilder &builder) {
   }
 
   return builder.reshape(tensor, newShape);
-};
+}
 
 // Obtaining the rank broadcasted shapes of tensors makes it easier to
 // construct the input and output reshaping logic.

src/Conversion/ONNXToTOSA/Math/Reduce.cpp

@@ -43,7 +43,7 @@ DenseIntElementsAttr getAxesLatestsVersionAttr(ONNXReduceOp op) {
     if (noOpIfAxesEmpty == 0) {
       // Default behaviour when "axes" is none and "noop_with_empty_axes" is
       // set to false, it is to reduce all dims
-      const int64_t numberOfAxes = input.getType().cast<ShapedType>().getRank();
+      const int64_t numberOfAxes = cast<ShapedType>(input.getType()).getRank();
       auto iotaRange =
           llvm::iota_range<int64_t>(0, numberOfAxes, /*Inclusive=*/false);
       targetAxes = SmallVector<int64_t>(iotaRange.begin(), iotaRange.end());
@@ -86,7 +86,7 @@ DenseIntElementsAttr getAxesLegacyVersionAttr(ONNXReduceOp op) {
   SmallVector<int64_t> targetAxes;
   if (!axes) {
     // if not present all axes are reduced
-    const int64_t numberOfAxes = input.getType().cast<ShapedType>().getRank();
+    const int64_t numberOfAxes = cast<ShapedType>(input.getType()).getRank();
     auto iotaRange =
         llvm::iota_range<int64_t>(0, numberOfAxes, /*Inclusive=*/false);
     targetAxes = SmallVector<int64_t>(iotaRange.begin(), iotaRange.end());
@@ -111,14 +111,12 @@ class ONNXReduceOpLoweringToTOSA : public OpConversionPattern<ONNXReduceOp> {
   LogicalResult matchAndRewrite(ONNXReduceOp op, OpAdaptor adaptor,
       ConversionPatternRewriter &rewriter) const override {
 
-    auto inputType =
-        adaptor.getData().getType().template dyn_cast<RankedTensorType>();
+    auto inputType = dyn_cast<RankedTensorType>(adaptor.getData().getType());
     if (!inputType)
       return rewriter.notifyMatchFailure(op, "input type not a ranked tensor.");
 
-    auto outputType = this->getTypeConverter()
-                          ->convertType(op.getResult().getType())
-                          .template cast<RankedTensorType>();
+    auto outputType = cast<RankedTensorType>(
+        this->getTypeConverter()->convertType(op.getResult().getType()));
 
     return (*lowerFn)(op, inputType, outputType, rewriter);
   }

src/Conversion/ONNXToTOSA/NN/AveragePool.cpp

@@ -41,14 +41,14 @@ LogicalResult handleIncludePadAttr(
     return rewriter.notifyMatchFailure(op, "Could not infer shapes");
   }
 
-  auto inputType = input.getType().cast<mlir::TensorType>();
+  auto inputType = cast<mlir::TensorType>(input.getType());
   if (inputType.getShape().size() != 4) {
     return rewriter.notifyMatchFailure(op, "TOSA only supports 2d pooling");
   }
 
   llvm::SmallVector<int64_t, 4> pads =
       tosa::createOrderedPadAttrForWindowBasedOps(rewriter,
-          input.getType().cast<mlir::TensorType>().getShape(), shapeHelper,
+          cast<mlir::TensorType>(input.getType()).getShape(), shapeHelper,
           /*ceilMode*/ 0, {0, 1, 2, 3});
 
   // Create Padding and ConstPad tosa::ConstOp's

src/Conversion/ONNXToTOSA/NN/MaxPoolSingleOut.cpp

@@ -36,7 +36,7 @@ class ONNXMaxPoolSingleOutOpLoweringToTOSA : public ConversionPattern {
   using OpAdaptor = typename ONNXMaxPoolSingleOutOp::Adaptor;
   LogicalResult matchAndRewrite(Operation *op, ArrayRef<Value> operands,
       ConversionPatternRewriter &rewriter) const final {
-    auto maxpoolOp = llvm::cast<ONNXMaxPoolSingleOutOp>(op);
+    auto maxpoolOp = mlir::cast<ONNXMaxPoolSingleOutOp>(op);
     OpAdaptor adaptor(operands, op->getAttrDictionary());
 
     Value input = adaptor.getX();

src/Conversion/ONNXToTOSA/ONNXToTOSACommon.cpp

@@ -57,9 +57,9 @@ std::optional<Value> convertGatherOp(PatternRewriter &rewriter, Location loc,
 
   TosaBuilder tosaBuilder(rewriter, loc);
 
-  auto resultType = resultValue.getType().dyn_cast<ShapedType>();
-  auto inputType = inputValue.getType().dyn_cast<RankedTensorType>();
-  auto indicesType = indicesValue.getType().dyn_cast<RankedTensorType>();
+  auto resultType = dyn_cast<ShapedType>(resultValue.getType());
+  auto inputType = dyn_cast<RankedTensorType>(inputValue.getType());
+  auto indicesType = dyn_cast<RankedTensorType>(indicesValue.getType());
 
   if (!resultType || !inputType || !indicesType)
     return std::nullopt;
@@ -143,7 +143,7 @@ std::optional<Value> convertGatherOp(PatternRewriter &rewriter, Location loc,
   // onnx allows i64 indices, but tosa does not.
   if (indicesType.getElementType().isInteger(64)) {
     indicesType =
-        indicesType.clone(rewriter.getI32Type()).dyn_cast<RankedTensorType>();
+        dyn_cast<RankedTensorType>(indicesType.clone(rewriter.getI32Type()));
     indicesValue = CreateOpAndInfer<mlir::tosa::CastOp>(
         rewriter, loc, indicesType, indicesValue)
                        .getResult();
@@ -293,7 +293,7 @@ std::optional<Value> convertReduceOpCommon(PatternRewriter &rewriter,
     bool is_quantized, double input_scale, int64_t input_zp,
     double output_scale, int64_t output_zp) {
   RankedTensorType input_type =
-      input_value.getType().dyn_cast<RankedTensorType>();
+      dyn_cast<RankedTensorType>(input_value.getType());
   if (!input_type)
     return std::nullopt;
 
@@ -362,14 +362,14 @@ std::optional<Value> convertReduceMeanOp(PatternRewriter &rewriter,
   // op2 = mul(op1, 1.0 / num_elements_on_reduced_axis)
 
   RankedTensorType input_type =
-      input_value.getType().dyn_cast<RankedTensorType>();
+      dyn_cast<RankedTensorType>(input_value.getType());
   if (!input_type)
     return std::nullopt;
 
   bool input_is_qtype =
-      input_type.getElementType().isa<mlir::quant::UniformQuantizedType>();
+      isa<mlir::quant::UniformQuantizedType>(input_type.getElementType());
   bool output_is_qtype =
-      output_type.getElementType().isa<mlir::quant::UniformQuantizedType>();
+      isa<mlir::quant::UniformQuantizedType>(output_type.getElementType());
 
   if (input_is_qtype != output_is_qtype) {
     op->emitOpError("ConvertReduceSumOp: input/output tensor should "
@@ -378,7 +378,7 @@ std::optional<Value> convertReduceMeanOp(PatternRewriter &rewriter,
   }
 
   // Only supports float type mean() if it's non-quantized
-  if (!input_is_qtype && !output_type.getElementType().isa<mlir::FloatType>()) {
+  if (!input_is_qtype && !isa<mlir::FloatType>(output_type.getElementType())) {
     op->emitWarning(
         "Failed convertReduceMean: input unquantized type but output element "
         "not FloatType!");
@@ -403,9 +403,9 @@ std::optional<Value> convertReduceMeanOp(PatternRewriter &rewriter,
 
   if (input_is_qtype) {
     auto input_qtype =
-        input_type.getElementType().cast<mlir::quant::UniformQuantizedType>();
+        cast<mlir::quant::UniformQuantizedType>(input_type.getElementType());
     auto output_qtype =
-        output_type.getElementType().cast<mlir::quant::UniformQuantizedType>();
+        cast<mlir::quant::UniformQuantizedType>(output_type.getElementType());
 
     // Combine 'div_scale' as part of output rescale
     output_scale = div_scale * input_qtype.getScale() / output_qtype.getScale();

src/Conversion/ONNXToTOSA/ONNXToTOSACommon.hpp

@@ -87,12 +87,12 @@ namespace onnx_mlir {
 //===----------------------------------------------------------------------===//
 
 inline bool isTOSABool(mlir::Type type) {
-  mlir::IntegerType intType = type.dyn_cast<mlir::IntegerType>();
+  mlir::IntegerType intType = mlir::dyn_cast<mlir::IntegerType>(type);
   return intType && intType.isSignless() && intType.getWidth() == 1;
 }
 
 inline bool isTOSAInt(mlir::Type type) {
-  mlir::IntegerType intType = type.dyn_cast<mlir::IntegerType>();
+  mlir::IntegerType intType = mlir::dyn_cast<mlir::IntegerType>(type);
   std::set<unsigned> intWidth{1, 8, 16, 32, 48, 64};
   return intType && (intType.isSignless() || intType.isUnsignedInteger()) &&
          (intWidth.find(intType.getWidth()) != intWidth.end());

src/Conversion/ONNXToTOSA/ONNXToTOSACommon.hpp.inc

@@ -126,13 +126,13 @@ llvm::SmallVector<int64_t, 4> createOrderedPadAttrForWindowBasedOps(
 
 inline mlir::LogicalResult getAvgPool2dAccType(mlir::PatternRewriter &rewriter,
     mlir::Value input, mlir::TypeAttr &accType) {
-  auto inputTy = llvm::dyn_cast<mlir::ShapedType>(input.getType());
+  auto inputTy = mlir::dyn_cast<mlir::ShapedType>(input.getType());
   if (!inputTy)
     return mlir::failure();
   auto inputETy = inputTy.getElementType();
 
   if (auto quantType =
-          llvm::dyn_cast<mlir::quant::UniformQuantizedType>(inputETy))
+          mlir::dyn_cast<mlir::quant::UniformQuantizedType>(inputETy))
     inputETy = quantType.getStorageType();
 
   // Tosa supports FP16 and FP32 accumulator type for FP16 input. When the time
@@ -180,7 +180,7 @@ mlir::FailureOr<mlir::Value> convertPoolOp(
   llvm::SmallVector<int64_t, 4> kernelShapeVec;
   llvm::transform(kernelShape, std::back_inserter(kernelShapeVec),
       [](const mlir::Attribute &pad) {
-        return pad.cast<mlir::IntegerAttr>().getInt();
+        return  mlir::cast<mlir::IntegerAttr>(pad).getInt();
       });
 
   const int64_t ceilMode = adaptor.getCeilMode();
@@ -216,7 +216,7 @@ mlir::FailureOr<mlir::Value> convertPoolOp(
       pads[0], pads[2] + ceilConstants[0], pads[1], pads[3] + ceilConstants[1]};
 
   mlir::FailureOr<mlir::Value> resizedInput = tosaBuilder.resizeWindowBasedOps(
-      input, input.getType().cast<mlir::RankedTensorType>().getShape(),
+      input, mlir::cast<mlir::RankedTensorType>(input.getType()).getShape(),
       {kernelShapeVec[0], kernelShapeVec[1]}, reorderedPads,
       shapeHelper.strides, shapeHelper.dilations);
 
@@ -257,4 +257,4 @@ mlir::FailureOr<mlir::Value> convertPoolOp(
   // Construct the old result shape out of the new one
   mlir::Value transpose = tosaBuilder.transpose(input, {0, 3, 1, 2});
   return transpose;
-};
+}

src/Conversion/ONNXToTOSA/ONNXToTOSALegalizeUtils.cpp

@@ -99,7 +99,7 @@ Value buildRescale(PatternRewriter &rewriter, Operation *op,
 Value buildRescaleToInt32(PatternRewriter &rewriter, Operation *op,
     Value input_val, double input_scale, int64_t input_zp) {
   // Output is always int32 type
-  auto input_type = input_val.getType().dyn_cast<mlir::ShapedType>();
+  auto input_type = dyn_cast<mlir::ShapedType>(input_val.getType());
   assert(input_type);
   auto output_type = input_type.clone(rewriter.getI32Type());
 

src/Conversion/ONNXToTOSA/ONNXToTOSALegalizeUtils.hpp

@@ -63,11 +63,11 @@ mlir::Value expandShape(mlir::PatternRewriter &rewriter, mlir::Location loc,
 // op. This allows shape inference during the framework to TOSA lowering.
 template <typename TosaOp, typename... Args>
 TosaOp CreateOpAndInfer(mlir::PatternRewriter &rewriter, mlir::Location loc,
-    mlir::Type result_ty, Args &&... args) {
+    mlir::Type result_ty, Args &&...args) {
   auto op = rewriter.create<TosaOp>(loc, result_ty, args...);
 
   mlir::InferShapedTypeOpInterface shapeInterface =
-      llvm::dyn_cast<mlir::InferShapedTypeOpInterface>(op.getOperation());
+      mlir::dyn_cast<mlir::InferShapedTypeOpInterface>(op.getOperation());
   if (!shapeInterface)
     return op;
 
@@ -92,7 +92,7 @@ TosaOp CreateOpAndInfer(mlir::PatternRewriter &rewriter, mlir::Location loc,
 
 template <typename TosaOp, typename... Args>
 void CreateReplaceOpAndInfer(mlir::PatternRewriter &rewriter,
-    mlir::Operation *op, mlir::Type result_ty, Args &&... args) {
+    mlir::Operation *op, mlir::Type result_ty, Args &&...args) {
   auto result =
       CreateOpAndInfer<TosaOp>(rewriter, op->getLoc(), result_ty, args...);
   rewriter.replaceOp(op, result->getResults());