[mlir] Change ConvertOpToLLVMPattern::matchAndRewrite argument to concrete operand type.

[herhut]

Thank you for cleaning this up. I was not aware that the llvm conversion does not have Op specific patterns, yet.

mlir/include/mlir/Conversion/StandardToLLVM/ConvertStandardToLLVM.h

576

I wonder whether there is some C++ magic so that can be reused more directly from OpConversionPattern?

mlir/lib/Conversion/GPUCommon/ConvertLaunchFuncToRuntimeCalls.cpp

271

Is op really needed?

mlir/lib/Conversion/SPIRVToLLVM/ConvertLaunchFuncToLLVMCalls.cpp

156

Here, too, I think the uses can also be replaced.

mlir/lib/Conversion/StandardToLLVM/StandardToLLVM.cpp

2296–2297

Could this be callOp.operands()?

2314

Can this be newOp.getResult(0)?

2677

castOp.getAttrDictionary()

2745

reshapeOp.getAttrDictionary()

3254

op.operands()?

3270

op.operands().getTypes()?

[csigg]

Thanks Stephan for the review. These comments are all good points. This change is primarily about changing the interface. I would prefer to do the clean up in follow-up changes, which can be done in smaller chunks.

The concrete ops do not expose the same interface as Operation* (e.g. getAttrDictionary).
We could either add those, or add an 'Operation* operator->(). I think the latter would make sense because concrete ops already implicitly convert to Operation*`.
I will make that suggestion and see how it goes.

[herhut]

Cleaning this up in a follow-on is fine too. Please let @ftynse comment, too.

[ftynse]

Very nice, I should have done this longtime ago, thanks!

If you have time, maybe we can also consider passing in a SourceOp::Adaptor instead of raw operands?

mlir/include/mlir/Conversion/StandardToLLVM/ConvertStandardToLLVM.h

576

We can refactor (likely just rename) ConvertToLLVMPattern into ConvertToLLVMPatternBase, and have ConvertOpToLLVMPattern be public OpConversionPattern, ConvertToLLVMPattern and ConvertToLLVMPattern be public ConversionPattern, ConvertToLLVMPattern. The current base class is essentially a wrapper around TypeConverter + a couple of utility functions. Actually, ConversionPattern now also has a TypeConverter instance, so we could get rid of the instance we have in ConvertToLLVMPattern and just cast instead.

Not in this patch though.

mlir/lib/Conversion/StandardToLLVM/StandardToLLVM.cpp

3823–3824

I suppose this change is just because we never fail to match, but I ticked on it because everything else uses matchAndRewrite.

[csigg]

In D92111#2420580, @ftynse wrote:

If you have time, maybe we can also consider passing in a SourceOp::Adaptor instead of raw operands?

This wouldn't fit OpConversionPattern. I will leave it as is in this change.

mlir/lib/Conversion/StandardToLLVM/StandardToLLVM.cpp

3823–3824

I forgot to undo this. Done now.

[csigg]

Undo matchAndRewriter -> rewrite.

mlir/include/mlir/Conversion/StandardToLLVM/ConvertStandardToLLVM.h

 
 protected:
   /// Reference to the type converter, with potential extensions.
   LLVMTypeConverter &typeConverter;
 };
 
 /// Utility class for operation conversions targeting the LLVM dialect that
 /// match exactly one source operation.
-template <typename OpTy>
+template <typename SourceOp>
 class ConvertOpToLLVMPattern : public ConvertToLLVMPattern {
 public:
   ConvertOpToLLVMPattern(LLVMTypeConverter &typeConverter,
                          PatternBenefit benefit = 1)
-      : ConvertToLLVMPattern(OpTy::getOperationName(),
+      : ConvertToLLVMPattern(SourceOp::getOperationName(),
                              &typeConverter.getContext(), typeConverter,
                              benefit) {}
+
+  /// Wrappers around the RewritePattern methods that pass the derived op type.
+  void rewrite(Operation *op, ArrayRef<Value> operands,
+               ConversionPatternRewriter &rewriter) const final {
+    rewrite(cast<SourceOp>(op), operands, rewriter);
+  }
+  LogicalResult match(Operation *op) const final {
+    return match(cast<SourceOp>(op));
+  }
+  LogicalResult
+  matchAndRewrite(Operation *op, ArrayRef<Value> operands,
+                  ConversionPatternRewriter &rewriter) const final {
+    return matchAndRewrite(cast<SourceOp>(op), operands, rewriter);
+  }
+
+  /// Rewrite and Match methods that operate on the SourceOp type. These must be
+  /// overridden by the derived pattern class.
+  virtual void rewrite(SourceOp op, ArrayRef<Value> operands,
+                       ConversionPatternRewriter &rewriter) const {
+    llvm_unreachable("must override rewrite or matchAndRewrite");
+  }
+  virtual LogicalResult match(SourceOp op) const {
+    llvm_unreachable("must override match or matchAndRewrite");
+  }
+  virtual LogicalResult
+  matchAndRewrite(SourceOp op, ArrayRef<Value> operands,
+                  ConversionPatternRewriter &rewriter) const {
+    if (succeeded(match(op))) {
+      rewrite(op, operands, rewriter);
+      return success();
+    }
+    return failure();
+  }
 };
 
 namespace LLVM {
 namespace detail {
 /// Replaces the given operation "op" with a new operation of type "targetOp"
 /// and given operands.
 LogicalResult oneToOneRewrite(Operation *op, StringRef targetOp,
                               ValueRange operands,

 class OneToOneConvertToLLVMPattern : public ConvertOpToLLVMPattern<SourceOp> {
 public:
   using ConvertOpToLLVMPattern<SourceOp>::ConvertOpToLLVMPattern;
   using Super = OneToOneConvertToLLVMPattern<SourceOp, TargetOp>;
 
   /// Converts the type of the result to an LLVM type, pass operands as is,
   /// preserve attributes.
   LogicalResult
-  matchAndRewrite(Operation *op, ArrayRef<Value> operands,
+  matchAndRewrite(SourceOp op, ArrayRef<Value> operands,
                   ConversionPatternRewriter &rewriter) const override {
     return LLVM::detail::oneToOneRewrite(op, TargetOp::getOperationName(),
                                          operands, this->typeConverter,
                                          rewriter);
   }
 };
 
 /// Basic lowering implementation to rewrite Ops with just one result to the
 /// LLVM Dialect. This supports higher-dimensional vector types.
 template <typename SourceOp, typename TargetOp>
 class VectorConvertToLLVMPattern : public ConvertOpToLLVMPattern<SourceOp> {
 public:
   using ConvertOpToLLVMPattern<SourceOp>::ConvertOpToLLVMPattern;
   using Super = VectorConvertToLLVMPattern<SourceOp, TargetOp>;
 
   LogicalResult
-  matchAndRewrite(Operation *op, ArrayRef<Value> operands,
+  matchAndRewrite(SourceOp op, ArrayRef<Value> operands,
                   ConversionPatternRewriter &rewriter) const override {
     static_assert(
         std::is_base_of<OpTrait::OneResult<SourceOp>, SourceOp>::value,
         "expected single result op");
     static_assert(std::is_base_of<OpTrait::SameOperandsAndResultType<SourceOp>,
                                   SourceOp>::value,
                   "expected same operands and result type");
     return LLVM::detail::vectorOneToOneRewrite(op, TargetOp::getOperationName(),

mlir/lib/Conversion/GPUCommon/ConvertLaunchFuncToRuntimeCalls.cpp

 class ConvertHostRegisterOpToGpuRuntimeCallPattern
     : public ConvertOpToGpuRuntimeCallPattern<gpu::HostRegisterOp> {
 public:
   ConvertHostRegisterOpToGpuRuntimeCallPattern(LLVMTypeConverter &typeConverter)
       : ConvertOpToGpuRuntimeCallPattern<gpu::HostRegisterOp>(typeConverter) {}
 
 private:
   LogicalResult
-  matchAndRewrite(Operation *op, ArrayRef<Value> operands,
+  matchAndRewrite(gpu::HostRegisterOp hostRegisterOp, ArrayRef<Value> operands,
                   ConversionPatternRewriter &rewriter) const override;
 };
 
 /// A rewrite pattern to convert gpu.wait operations into a GPU runtime
 /// call. Currently it supports CUDA and ROCm (HIP).
 class ConvertWaitOpToGpuRuntimeCallPattern
     : public ConvertOpToGpuRuntimeCallPattern<gpu::WaitOp> {
 public:
   ConvertWaitOpToGpuRuntimeCallPattern(LLVMTypeConverter &typeConverter)
       : ConvertOpToGpuRuntimeCallPattern<gpu::WaitOp>(typeConverter) {}
 
 private:
   LogicalResult
-  matchAndRewrite(Operation *op, ArrayRef<Value> operands,
+  matchAndRewrite(gpu::WaitOp waitOp, ArrayRef<Value> operands,
                   ConversionPatternRewriter &rewriter) const override;
 };
 
 /// A rewrite pattern to convert gpu.wait async operations into a GPU runtime
 /// call. Currently it supports CUDA and ROCm (HIP).
 class ConvertWaitAsyncOpToGpuRuntimeCallPattern
     : public ConvertOpToGpuRuntimeCallPattern<gpu::WaitOp> {
 public:
   ConvertWaitAsyncOpToGpuRuntimeCallPattern(LLVMTypeConverter &typeConverter)
       : ConvertOpToGpuRuntimeCallPattern<gpu::WaitOp>(typeConverter) {}
 
 private:
   LogicalResult
-  matchAndRewrite(Operation *op, ArrayRef<Value> operands,
+  matchAndRewrite(gpu::WaitOp waitOp, ArrayRef<Value> operands,
                   ConversionPatternRewriter &rewriter) const override;
 };
 
 /// A rewrite patter to convert gpu.launch_func operations into a sequence of
 /// GPU runtime calls. Currently it supports CUDA and ROCm (HIP).
 ///
 /// In essence, a gpu.launch_func operations gets compiled into the following
 /// sequence of runtime calls:

 
 private:
   Value generateParamsArray(gpu::LaunchFuncOp launchOp,
                             ArrayRef<Value> operands, OpBuilder &builder) const;
   Value generateKernelNameConstant(StringRef moduleName, StringRef name,
                                    Location loc, OpBuilder &builder) const;
 
   LogicalResult
-  matchAndRewrite(Operation *op, ArrayRef<Value> operands,
+  matchAndRewrite(gpu::LaunchFuncOp launchOp, ArrayRef<Value> operands,
                   ConversionPatternRewriter &rewriter) const override;
 
   llvm::SmallString<32> gpuBinaryAnnotation;
 };
 
 class EraseGpuModuleOpPattern : public OpRewritePattern<gpu::GPUModuleOp> {
   using OpRewritePattern<gpu::GPUModuleOp>::OpRewritePattern;
 

 }
 
 // Returns whether value is of LLVM type.
 static bool isLLVMType(Value value) {
   return value.getType().isa<LLVM::LLVMType>();
 }
 
 LogicalResult ConvertHostRegisterOpToGpuRuntimeCallPattern::matchAndRewrite(
-    Operation *op, ArrayRef<Value> operands,
+    gpu::HostRegisterOp hostRegisterOp, ArrayRef<Value> operands,
     ConversionPatternRewriter &rewriter) const {
+  auto *op = hostRegisterOp.getOperation();
   if (!llvm::all_of(operands, isLLVMType))
     return rewriter.notifyMatchFailure(
         op, "Cannot convert if operands aren't of LLVM type.");
 
   Location loc = op->getLoc();
 
-  auto memRefType = cast<gpu::HostRegisterOp>(op).value().getType();
+  auto memRefType = hostRegisterOp.value().getType();
   auto elementType = memRefType.cast<UnrankedMemRefType>().getElementType();
   auto elementSize = getSizeInBytes(loc, elementType, rewriter);
 
   auto arguments =
       typeConverter.promoteOperands(loc, op->getOperands(), operands, rewriter);
   arguments.push_back(elementSize);
   hostRegisterCallBuilder.create(loc, rewriter, arguments);
 
   rewriter.eraseOp(op);
   return success();
 }
 
 // Converts `gpu.wait` to runtime calls. The operands are all CUDA or ROCm
 // streams (i.e. void*). The converted op synchronizes the host with every
 // stream and then destroys it. That is, it assumes that the stream is not used
 // afterwards. In case this isn't correct, we will get a runtime error.
 // Eventually, we will have a pass that guarantees this property.
 LogicalResult ConvertWaitOpToGpuRuntimeCallPattern::matchAndRewrite(
-    Operation *op, ArrayRef<Value> operands,
+    gpu::WaitOp waitOp, ArrayRef<Value> operands,
     ConversionPatternRewriter &rewriter) const {
-  if (cast<gpu::WaitOp>(op).asyncToken())
-    return rewriter.notifyMatchFailure(op, "Cannot convert async op.");
+  if (waitOp.asyncToken())
+    return rewriter.notifyMatchFailure(waitOp, "Cannot convert async op.");
 
-  Location loc = op->getLoc();
+  Location loc = waitOp.getLoc();
 
   for (auto asyncDependency : operands)
     streamSynchronizeCallBuilder.create(loc, rewriter, {asyncDependency});
   for (auto asyncDependency : operands)
     streamDestroyCallBuilder.create(loc, rewriter, {asyncDependency});
 
-  rewriter.eraseOp(op);
+  rewriter.eraseOp(waitOp);
   return success();
 }
 
 // Converts `gpu.wait async` to runtime calls. The result is a new stream that
 // is synchronized with all operands, which are CUDA or ROCm streams (i.e.
 // void*). We create and record an event after the definition of the stream
 // and make the new stream wait on that event before destroying it again. This
 // assumes that there is no other use between the definition and this op, and
 // the plan is to have a pass that guarantees this property.
 LogicalResult ConvertWaitAsyncOpToGpuRuntimeCallPattern::matchAndRewrite(
-    Operation *op, ArrayRef<Value> operands,
+    gpu::WaitOp waitOp, ArrayRef<Value> operands,
     ConversionPatternRewriter &rewriter) const {
-  if (!cast<gpu::WaitOp>(op).asyncToken())
-    return rewriter.notifyMatchFailure(op, "Can only convert async op.");
+  if (!waitOp.asyncToken())
+    return rewriter.notifyMatchFailure(waitOp, "Can only convert async op.");
 
-  Location loc = op->getLoc();
+  Location loc = waitOp.getLoc();
 
   auto insertionPoint = rewriter.saveInsertionPoint();
   SmallVector<Value, 1> events;
-  for (auto pair : llvm::zip(op->getOperands(), operands)) {
+  for (auto pair : llvm::zip(waitOp.asyncDependencies(), operands)) {
     auto token = std::get<0>(pair);
     if (auto *defOp = token.getDefiningOp()) {
       rewriter.setInsertionPointAfter(defOp);
     } else {
       // If we can't find the defining op, we record the event at block start,
       // which is late and therefore misses parallelism, but still valid.
-      rewriter.setInsertionPointToStart(op->getBlock());
+      rewriter.setInsertionPointToStart(waitOp.getOperation()->getBlock());
     }
     auto event = eventCreateCallBuilder.create(loc, rewriter, {}).getResult(0);
     auto stream = std::get<1>(pair);
     eventRecordCallBuilder.create(loc, rewriter, {event, stream});
     events.push_back(event);
   }
   rewriter.restoreInsertionPoint(insertionPoint);
   auto stream = streamCreateCallBuilder.create(loc, rewriter, {}).getResult(0);
   for (auto event : events)
     streamWaitEventCallBuilder.create(loc, rewriter, {stream, event});
   for (auto event : events)
     eventDestroyCallBuilder.create(loc, rewriter, {event});
-  rewriter.replaceOp(op, {stream});
+  rewriter.replaceOp(waitOp, {stream});
 
   return success();
 }
 
 // Creates a struct containing all kernel parameters on the stack and returns
 // an array of type-erased pointers to the fields of the struct. The array can
 // then be passed to the CUDA / ROCm (HIP) kernel launch calls.
 // The generated code is essentially as follows:

 // call %launchKernel(%3, <launchOp operands 0..5>, 0, %4, %5, nullptr)
 // call %streamSynchronize(%4)
 // call %streamDestroy(%4)
 // call %moduleUnload(%1)
 //
 // If the op is async, the stream corresponds to the (single) async dependency
 // as well as the async token the op produces.
 LogicalResult ConvertLaunchFuncOpToGpuRuntimeCallPattern::matchAndRewrite(
-    Operation *op, ArrayRef<Value> operands,
+    gpu::LaunchFuncOp launchOp, ArrayRef<Value> operands,
     ConversionPatternRewriter &rewriter) const {
   if (!llvm::all_of(operands, isLLVMType))
     return rewriter.notifyMatchFailure(
-        op, "Cannot convert if operands aren't of LLVM type.");
-
-  auto launchOp = cast<gpu::LaunchFuncOp>(op);
+        launchOp, "Cannot convert if operands aren't of LLVM type.");
 
   if (launchOp.asyncDependencies().size() > 1)
     return rewriter.notifyMatchFailure(
-        op, "Cannot convert with more than one async dependency.");
+        launchOp, "Cannot convert with more than one async dependency.");
 
   // Fail when the synchronous version of the op has async dependencies. The
   // lowering destroys the stream, and we do not want to check that there is no
   // use of the stream after this op.
   if (!launchOp.asyncToken() && !launchOp.asyncDependencies().empty())
     return rewriter.notifyMatchFailure(
-        op, "Cannot convert non-async op with async dependencies.");
+        launchOp, "Cannot convert non-async op with async dependencies.");
 
   Location loc = launchOp.getLoc();
 
   // Create an LLVM global with CUBIN extracted from the kernel annotation and
   // obtain a pointer to the first byte in it.
   auto kernelModule = SymbolTable::lookupNearestSymbolFrom<gpu::GPUModuleOp>(
       launchOp, launchOp.getKernelModuleName());
   assert(kernelModule && "expected a kernel module");
LogicalResult ConvertLaunchFuncOpToGpuRuntimeCallPattern::matchAndRewrite(
   // Get the function from the module. The name corresponds to the name of
   // the kernel function.
   auto kernelName = generateKernelNameConstant(
       launchOp.getKernelModuleName(), launchOp.getKernelName(), loc, rewriter);
   auto function = moduleGetFunctionCallBuilder.create(
       loc, rewriter, {module.getResult(0), kernelName});
   auto zero = rewriter.create<LLVM::ConstantOp>(loc, llvmInt32Type,
                                                 rewriter.getI32IntegerAttr(0));
-  auto adaptor = gpu::LaunchFuncOpAdaptor(operands, op->getAttrDictionary());
+  auto adaptor = gpu::LaunchFuncOpAdaptor(
+      operands, launchOp.getOperation()->getAttrDictionary());
   Value stream =
       adaptor.asyncDependencies().empty()
           ? streamCreateCallBuilder.create(loc, rewriter, {}).getResult(0)
           : adaptor.asyncDependencies().front();
   // Create array of pointers to kernel arguments.
   auto kernelParams = generateParamsArray(launchOp, operands, rewriter);
   auto nullpointer = rewriter.create<LLVM::NullOp>(loc, llvmPointerPointerType);
-  launchKernelCallBuilder.create(
-      loc, rewriter,
-      {function.getResult(0), launchOp.gridSizeX(), launchOp.gridSizeY(),
-       launchOp.gridSizeZ(), launchOp.blockSizeX(), launchOp.blockSizeY(),
-       launchOp.blockSizeZ(), /*sharedMemBytes=*/zero, stream, kernelParams,
+  launchKernelCallBuilder.create(loc, rewriter,
+                                 {function.getResult(0), launchOp.gridSizeX(),
+                                  launchOp.gridSizeY(), launchOp.gridSizeZ(),
+                                  launchOp.blockSizeX(), launchOp.blockSizeY(),
+                                  launchOp.blockSizeZ(),
+                                  /*sharedMemBytes=*/zero, stream, kernelParams,
                                   /*extra=*/nullpointer});
 
   if (launchOp.asyncToken()) {
     // Async launch: make dependent ops use the same stream.
-    rewriter.replaceOp(op, {stream});
+    rewriter.replaceOp(launchOp, {stream});
   } else {
     // Synchronize with host and destroy stream. This must be the stream created
     // above (with no other uses) because we check that the synchronous version
     // does not have any async dependencies.
     streamSynchronizeCallBuilder.create(loc, rewriter, stream);
     streamDestroyCallBuilder.create(loc, rewriter, stream);
-    rewriter.eraseOp(op);
+    rewriter.eraseOp(launchOp);
   }
   moduleUnloadCallBuilder.create(loc, rewriter, module.getResult(0));
 
   return success();
 }
 
 std::unique_ptr<mlir::OperationPass<mlir::ModuleOp>>
 mlir::createGpuToLLVMConversionPass(StringRef gpuBinaryAnnotation) {

mlir/lib/Conversion/SPIRVToLLVM/ConvertLaunchFuncToLLVMCalls.cpp

 /// This pattern emulates a call to the kernel in LLVM dialect. For that, we
 /// copy the data to the global variable (emulating device side), call the
 /// kernel as a normal void LLVM function, and copy the data back (emulating the
 /// host side).
 class GPULaunchLowering : public ConvertOpToLLVMPattern<gpu::LaunchFuncOp> {
   using ConvertOpToLLVMPattern<gpu::LaunchFuncOp>::ConvertOpToLLVMPattern;
 
   LogicalResult
-  matchAndRewrite(Operation *op, ArrayRef<Value> operands,
+  matchAndRewrite(gpu::LaunchFuncOp launchOp, ArrayRef<Value> operands,
                   ConversionPatternRewriter &rewriter) const override {
-    gpu::LaunchFuncOp launchOp = cast<gpu::LaunchFuncOp>(op);
+    auto *op = launchOp.getOperation();
     MLIRContext *context = rewriter.getContext();
     auto module = launchOp.getParentOfType<ModuleOp>();
 
     // Get the SPIR-V module that represents the gpu kernel module. The module
     // is named:
     //   __spv__{kernel_module_name}
     // based on GPU to SPIR-V conversion.
     StringRef kernelModuleName = launchOp.getKernelModuleName();

mlir/lib/Conversion/StandardToLLVM/StandardToLLVM.cpp

 /// MemRef descriptors (LLVM struct data types) containing all the MemRef type
 /// information.
 static constexpr StringRef kEmitIfaceAttrName = "llvm.emit_c_interface";
 struct FuncOpConversion : public FuncOpConversionBase {
   FuncOpConversion(LLVMTypeConverter &converter)
       : FuncOpConversionBase(converter) {}
 
   LogicalResult
-  matchAndRewrite(Operation *op, ArrayRef<Value> operands,
+  matchAndRewrite(FuncOp funcOp, ArrayRef<Value> operands,
                   ConversionPatternRewriter &rewriter) const override {
-    auto funcOp = cast<FuncOp>(op);
-
     auto newFuncOp = convertFuncOpToLLVMFuncOp(funcOp, rewriter);
     if (!newFuncOp)
       return failure();
 
     if (typeConverter.getOptions().emitCWrappers ||
         funcOp.getAttrOfType<UnitAttr>(kEmitIfaceAttrName)) {
       if (newFuncOp.isExternal())
-        wrapExternalFunction(rewriter, op->getLoc(), typeConverter, funcOp,
+        wrapExternalFunction(rewriter, funcOp.getLoc(), typeConverter, funcOp,
                              newFuncOp);
       else
-        wrapForExternalCallers(rewriter, op->getLoc(), typeConverter, funcOp,
+        wrapForExternalCallers(rewriter, funcOp.getLoc(), typeConverter, funcOp,
                                newFuncOp);
     }
 
-    rewriter.eraseOp(op);
+    rewriter.eraseOp(funcOp);
     return success();
   }
 };
 
 /// FuncOp legalization pattern that converts MemRef arguments to bare pointers
 /// to the MemRef element type. This will impact the calling convention and ABI.
 struct BarePtrFuncOpConversion : public FuncOpConversionBase {
   using FuncOpConversionBase::FuncOpConversionBase;
 
   LogicalResult
-  matchAndRewrite(Operation *op, ArrayRef<Value> operands,
+  matchAndRewrite(FuncOp funcOp, ArrayRef<Value> operands,
                   ConversionPatternRewriter &rewriter) const override {
-    auto funcOp = cast<FuncOp>(op);
-
     // Store the type of memref-typed arguments before the conversion so that we
     // can promote them to MemRef descriptor at the beginning of the function.
     SmallVector<Type, 8> oldArgTypes =
         llvm::to_vector<8>(funcOp.getType().getInputs());
 
     auto newFuncOp = convertFuncOpToLLVMFuncOp(funcOp, rewriter);
     if (!newFuncOp)
       return failure();
     if (newFuncOp.getBody().empty()) {
-      rewriter.eraseOp(op);
+      rewriter.eraseOp(funcOp);
       return success();
     }
 
     // Promote bare pointers from memref arguments to memref descriptors at the
     // beginning of the function so that all the memrefs in the function have a
     // uniform representation.
     Block *entryBlock = &newFuncOp.getBody().front();
     auto blockArgs = entryBlock->getArguments();
for (auto it : llvm::zip(blockArgs, oldArgTypes)) {
         continue;
 
       // Replace barePtr with a placeholder (undef), promote barePtr to a ranked
       // or unranked memref descriptor and replace placeholder with the last
       // instruction of the memref descriptor.
       // TODO: The placeholder is needed to avoid replacing barePtr uses in the
       // MemRef descriptor instructions. We may want to have a utility in the
       // rewriter to properly handle this use case.
-      Location loc = op->getLoc();
+      Location loc = funcOp.getLoc();
       auto placeholder = rewriter.create<LLVM::UndefOp>(loc, memrefTy);
       rewriter.replaceUsesOfBlockArgument(arg, placeholder);
 
       Value desc = MemRefDescriptor::fromStaticShape(
           rewriter, loc, typeConverter, memrefTy, arg);
       rewriter.replaceOp(placeholder, {desc});
     }
 
-    rewriter.eraseOp(op);
+    rewriter.eraseOp(funcOp);
     return success();
   }
 };
 
 //////////////// Support for Lowering operations on n-D vectors ////////////////
 // Helper struct to "unroll" operations on n-D vectors in terms of operations on
 // 1-D LLVM vectors.
 struct NDVectorTypeInfo {

 /// Lower `std.assert`. The default lowering calls the `abort` function if the
 /// assertion is violated and has no effect otherwise. The failure message is
 /// ignored by the default lowering but should be propagated by any custom
 /// lowering.
 struct AssertOpLowering : public ConvertOpToLLVMPattern<AssertOp> {
   using ConvertOpToLLVMPattern<AssertOp>::ConvertOpToLLVMPattern;
 
   LogicalResult
-  matchAndRewrite(Operation *op, ArrayRef<Value> operands,
+  matchAndRewrite(AssertOp op, ArrayRef<Value> operands,
                   ConversionPatternRewriter &rewriter) const override {
-    auto loc = op->getLoc();
+    auto loc = op.getLoc();
     AssertOp::Adaptor transformed(operands);
 
     // Insert the `abort` declaration if necessary.
-    auto module = op->getParentOfType<ModuleOp>();
+    auto module = op.getParentOfType<ModuleOp>();
     auto abortFunc = module.lookupSymbol<LLVM::LLVMFuncOp>("abort");
     if (!abortFunc) {
       OpBuilder::InsertionGuard guard(rewriter);
       rewriter.setInsertionPointToStart(module.getBody());
       auto abortFuncTy =
           LLVM::LLVMType::getFunctionTy(getVoidType(), {}, /*isVarArg=*/false);
       abortFunc = rewriter.create<LLVM::LLVMFuncOp>(rewriter.getUnknownLoc(),
                                                     "abort", abortFuncTy);

 
 // Lowerings for operations on complex numbers.
 
 struct CreateComplexOpLowering
     : public ConvertOpToLLVMPattern<CreateComplexOp> {
   using ConvertOpToLLVMPattern<CreateComplexOp>::ConvertOpToLLVMPattern;
 
   LogicalResult
-  matchAndRewrite(Operation *op, ArrayRef<Value> operands,
+  matchAndRewrite(CreateComplexOp op, ArrayRef<Value> operands,
                   ConversionPatternRewriter &rewriter) const override {
     auto complexOp = cast<CreateComplexOp>(op);
     CreateComplexOp::Adaptor transformed(operands);
 
     // Pack real and imaginary part in a complex number struct.
-    auto loc = op->getLoc();
+    auto loc = op.getLoc();
     auto structType = typeConverter.convertType(complexOp.getType());
     auto complexStruct = ComplexStructBuilder::undef(rewriter, loc, structType);
     complexStruct.setReal(rewriter, loc, transformed.real());
     complexStruct.setImaginary(rewriter, loc, transformed.imaginary());
 
     rewriter.replaceOp(op, {complexStruct});
     return success();
   }
 };
 
 struct ReOpLowering : public ConvertOpToLLVMPattern<ReOp> {
   using ConvertOpToLLVMPattern<ReOp>::ConvertOpToLLVMPattern;
 
   LogicalResult
-  matchAndRewrite(Operation *op, ArrayRef<Value> operands,
+  matchAndRewrite(ReOp op, ArrayRef<Value> operands,
                   ConversionPatternRewriter &rewriter) const override {
     ReOp::Adaptor transformed(operands);
 
     // Extract real part from the complex number struct.
     ComplexStructBuilder complexStruct(transformed.complex());
-    Value real = complexStruct.real(rewriter, op->getLoc());
+    Value real = complexStruct.real(rewriter, op.getLoc());
     rewriter.replaceOp(op, real);
 
     return success();
   }
 };
 
 struct ImOpLowering : public ConvertOpToLLVMPattern<ImOp> {
   using ConvertOpToLLVMPattern<ImOp>::ConvertOpToLLVMPattern;
 
   LogicalResult
-  matchAndRewrite(Operation *op, ArrayRef<Value> operands,
+  matchAndRewrite(ImOp op, ArrayRef<Value> operands,
                   ConversionPatternRewriter &rewriter) const override {
     ImOp::Adaptor transformed(operands);
 
     // Extract imaginary part from the complex number struct.
     ComplexStructBuilder complexStruct(transformed.complex());
-    Value imaginary = complexStruct.imaginary(rewriter, op->getLoc());
+    Value imaginary = complexStruct.imaginary(rewriter, op.getLoc());
     rewriter.replaceOp(op, imaginary);
 
     return success();
   }
 };
 
 struct BinaryComplexOperands {
   std::complex<Value> lhs, rhs;
unpackBinaryComplexOperands(OpTy op, ArrayRef<Value> operands,
 
   return unpacked;
 }
 
 struct AddCFOpLowering : public ConvertOpToLLVMPattern<AddCFOp> {
   using ConvertOpToLLVMPattern<AddCFOp>::ConvertOpToLLVMPattern;
 
   LogicalResult
-  matchAndRewrite(Operation *operation, ArrayRef<Value> operands,
+  matchAndRewrite(AddCFOp op, ArrayRef<Value> operands,
                   ConversionPatternRewriter &rewriter) const override {
-    auto op = cast<AddCFOp>(operation);
     auto loc = op.getLoc();
     BinaryComplexOperands arg =
         unpackBinaryComplexOperands<AddCFOp>(op, operands, rewriter);
 
     // Initialize complex number struct for result.
     auto structType = this->typeConverter.convertType(op.getType());
     auto result = ComplexStructBuilder::undef(rewriter, loc, structType);
 
matchAndRewrite(AddCFOp op, ArrayRef<Value> operands,
     return success();
   }
 };
 
 struct SubCFOpLowering : public ConvertOpToLLVMPattern<SubCFOp> {
   using ConvertOpToLLVMPattern<SubCFOp>::ConvertOpToLLVMPattern;
 
   LogicalResult
-  matchAndRewrite(Operation *operation, ArrayRef<Value> operands,
+  matchAndRewrite(SubCFOp op, ArrayRef<Value> operands,
                   ConversionPatternRewriter &rewriter) const override {
-    auto op = cast<SubCFOp>(operation);
     auto loc = op.getLoc();
     BinaryComplexOperands arg =
         unpackBinaryComplexOperands<SubCFOp>(op, operands, rewriter);
 
     // Initialize complex number struct for result.
     auto structType = this->typeConverter.convertType(op.getType());
     auto result = ComplexStructBuilder::undef(rewriter, loc, structType);
 
matchAndRewrite(SubCFOp op, ArrayRef<Value> operands,
     return success();
   }
 };
 
 struct ConstantOpLowering : public ConvertOpToLLVMPattern<ConstantOp> {
   using ConvertOpToLLVMPattern<ConstantOp>::ConvertOpToLLVMPattern;
 
   LogicalResult
-  matchAndRewrite(Operation *operation, ArrayRef<Value> operands,
+  matchAndRewrite(ConstantOp op, ArrayRef<Value> operands,
                   ConversionPatternRewriter &rewriter) const override {
-    auto op = cast<ConstantOp>(operation);
     // If constant refers to a function, convert it to "addressof".
     if (auto symbolRef = op.getValue().dyn_cast<FlatSymbolRefAttr>()) {
       auto type = typeConverter.convertType(op.getResult().getType())
                       .dyn_cast_or_null<LLVM::LLVMType>();
       if (!type)
         return rewriter.notifyMatchFailure(op, "failed to convert result type");
 
       MutableDictionaryAttr attrs(op.getAttrs());

 // passes the pointer to the MemRef across function boundaries.
 template <typename CallOpType>
 struct CallOpInterfaceLowering : public ConvertOpToLLVMPattern<CallOpType> {
   using ConvertOpToLLVMPattern<CallOpType>::ConvertOpToLLVMPattern;
   using Super = CallOpInterfaceLowering<CallOpType>;
   using Base = ConvertOpToLLVMPattern<CallOpType>;
 
   LogicalResult
-  matchAndRewrite(Operation *op, ArrayRef<Value> operands,
+  matchAndRewrite(CallOpType callOp, ArrayRef<Value> operands,
                   ConversionPatternRewriter &rewriter) const override {
     typename CallOpType::Adaptor transformed(operands);
-    auto callOp = cast<CallOpType>(op);
 
     // Pack the result types into a struct.
     Type packedResult = nullptr;
     unsigned numResults = callOp.getNumResults();
     auto resultTypes = llvm::to_vector<4>(callOp.getResultTypes());
 
     if (numResults != 0) {
       if (!(packedResult =
                 this->typeConverter.packFunctionResults(resultTypes)))
         return failure();
     }
 
     auto promoted = this->typeConverter.promoteOperands(
-        op->getLoc(), /*opOperands=*/op->getOperands(), operands, rewriter);
+        callOp.getLoc(), /*opOperands=*/callOp.getOperation()->getOperands(),
+        operands, rewriter);
     auto newOp = rewriter.create<LLVM::CallOp>(
-        op->getLoc(), packedResult ? TypeRange(packedResult) : TypeRange(),
-        promoted, op->getAttrs());
+        callOp.getLoc(), packedResult ? TypeRange(packedResult) : TypeRange(),
+        promoted, callOp.getAttrs());
 
     SmallVector<Value, 4> results;
     if (numResults < 2) {
       // If < 2 results, packing did not do anything and we can just return.
       results.append(newOp.result_begin(), newOp.result_end());
     } else {
       // Otherwise, it had been converted to an operation producing a structure.
       // Extract individual results from the structure and return them as list.
       results.reserve(numResults);
       for (unsigned i = 0; i < numResults; ++i) {
-        auto type = this->typeConverter.convertType(op->getResult(i).getType());
+        auto type =
+            this->typeConverter.convertType(callOp.getResult(i).getType());
         results.push_back(rewriter.create<LLVM::ExtractValueOp>(
-            op->getLoc(), type, newOp.getOperation()->getResult(0),
+            callOp.getLoc(), type, newOp.getOperation()->getResult(0),
             rewriter.getI64ArrayAttr(i)));
       }
     }
 
     if (this->typeConverter.getOptions().useBarePtrCallConv) {
       // For the bare-ptr calling convention, promote memref results to
       // descriptors.
       assert(results.size() == resultTypes.size() &&
              "The number of arguments and types doesn't match");
-      this->typeConverter.promoteBarePtrsToDescriptors(rewriter, op->getLoc(),
-                                                       resultTypes, results);
-    } else if (failed(copyUnrankedDescriptors(rewriter, op->getLoc(),
+      this->typeConverter.promoteBarePtrsToDescriptors(
+          rewriter, callOp.getLoc(), resultTypes, results);
+    } else if (failed(copyUnrankedDescriptors(rewriter, callOp.getLoc(),
                                               this->typeConverter, resultTypes,
                                               results,
                                               /*toDynamic=*/false))) {
       return failure();
     }
 
-    rewriter.replaceOp(op, results);
+    rewriter.replaceOp(callOp, results);
     return success();
   }
 };
 
 struct CallOpLowering : public CallOpInterfaceLowering<CallOp> {
   using Super::Super;
 };
 
 struct CallIndirectOpLowering : public CallOpInterfaceLowering<CallIndirectOp> {
   using Super::Super;
 };
 
 // A `dealloc` is converted into a call to `free` on the underlying data buffer.
 // The memref descriptor being an SSA value, there is no need to clean it up
 // in any way.
 struct DeallocOpLowering : public ConvertOpToLLVMPattern<DeallocOp> {
   using ConvertOpToLLVMPattern<DeallocOp>::ConvertOpToLLVMPattern;
 
   explicit DeallocOpLowering(LLVMTypeConverter &converter)
       : ConvertOpToLLVMPattern<DeallocOp>(converter) {}
 
   LogicalResult
-  matchAndRewrite(Operation *op, ArrayRef<Value> operands,
+  matchAndRewrite(DeallocOp op, ArrayRef<Value> operands,
                   ConversionPatternRewriter &rewriter) const override {
     assert(operands.size() == 1 && "dealloc takes one operand");
     DeallocOp::Adaptor transformed(operands);
 
     // Insert the `free` declaration if it is not already present.
     auto freeFunc =
-        op->getParentOfType<ModuleOp>().lookupSymbol<LLVM::LLVMFuncOp>("free");
+        op.getParentOfType<ModuleOp>().lookupSymbol<LLVM::LLVMFuncOp>("free");
     if (!freeFunc) {
       OpBuilder::InsertionGuard guard(rewriter);
       rewriter.setInsertionPointToStart(
-          op->getParentOfType<ModuleOp>().getBody());
+          op.getParentOfType<ModuleOp>().getBody());
       freeFunc = rewriter.create<LLVM::LLVMFuncOp>(
           rewriter.getUnknownLoc(), "free",
           LLVM::LLVMType::getFunctionTy(getVoidType(), getVoidPtrType(),
                                         /*isVarArg=*/false));
     }
 
     MemRefDescriptor memref(transformed.memref());
     Value casted = rewriter.create<LLVM::BitcastOp>(
-        op->getLoc(), getVoidPtrType(),
-        memref.allocatedPtr(rewriter, op->getLoc()));
+        op.getLoc(), getVoidPtrType(),
+        memref.allocatedPtr(rewriter, op.getLoc()));
     rewriter.replaceOpWithNewOp<LLVM::CallOp>(
         op, TypeRange(), rewriter.getSymbolRefAttr(freeFunc), casted);
     return success();
   }
 };
 
 /// Returns the LLVM type of the global variable given the memref type `type`.
 static LLVM::LLVMType
convertGlobalMemrefTypeToLLVM(MemRefType type,
   return arrayTy;
 }
 
 /// GlobalMemrefOp is lowered to a LLVM Global Variable.
 struct GlobalMemrefOpLowering : public ConvertOpToLLVMPattern<GlobalMemrefOp> {
   using ConvertOpToLLVMPattern<GlobalMemrefOp>::ConvertOpToLLVMPattern;
 
   LogicalResult
-  matchAndRewrite(Operation *op, ArrayRef<Value> operands,
+  matchAndRewrite(GlobalMemrefOp global, ArrayRef<Value> operands,
                   ConversionPatternRewriter &rewriter) const override {
-    auto global = cast<GlobalMemrefOp>(op);
     MemRefType type = global.type().cast<MemRefType>();
     if (!isSupportedMemRefType(type))
       return failure();
 
     LLVM::LLVMType arrayTy = convertGlobalMemrefTypeToLLVM(type, typeConverter);
 
     LLVM::Linkage linkage =
         global.isPublic() ? LLVM::Linkage::External : LLVM::Linkage::Private;
 
     Attribute initialValue = nullptr;
     if (!global.isExternal() && !global.isUninitialized()) {
       auto elementsAttr = global.initial_value()->cast<ElementsAttr>();
       initialValue = elementsAttr;
 
       // For scalar memrefs, the global variable created is of the element type,
       // so unpack the elements attribute to extract the value.
       if (type.getRank() == 0)
         initialValue = elementsAttr.getValue({});
     }
 
     rewriter.replaceOpWithNewOp<LLVM::GlobalOp>(
-        op, arrayTy, global.constant(), linkage, global.sym_name(),
+        global, arrayTy, global.constant(), linkage, global.sym_name(),
         initialValue, type.getMemorySpace());
     return success();
   }
 };
 
 /// GetGlobalMemrefOp is lowered into a Memref descriptor with the pointer to
 /// the first element stashed into the descriptor. This reuses
 /// `AllocLikeOpLowering` to reuse the Memref descriptor construction.
struct GetGlobalMemrefOpLowering : public AllocLikeOpLowering {
   }
 };
 
 // A `rsqrt` is converted into `1 / sqrt`.
 struct RsqrtOpLowering : public ConvertOpToLLVMPattern<RsqrtOp> {
   using ConvertOpToLLVMPattern<RsqrtOp>::ConvertOpToLLVMPattern;
 
   LogicalResult
-  matchAndRewrite(Operation *op, ArrayRef<Value> operands,
+  matchAndRewrite(RsqrtOp op, ArrayRef<Value> operands,
                   ConversionPatternRewriter &rewriter) const override {
     RsqrtOp::Adaptor transformed(operands);
     auto operandType =
         transformed.operand().getType().dyn_cast<LLVM::LLVMType>();
 
     if (!operandType)
       return failure();
 
-    auto loc = op->getLoc();
-    auto resultType = *op->result_type_begin();
+    auto loc = op.getLoc();
+    auto resultType = op.getResult().getType();
     auto floatType = getElementTypeOrSelf(resultType).cast<FloatType>();
     auto floatOne = rewriter.getFloatAttr(floatType, 1.0);
 
     if (!operandType.isArrayTy()) {
       LLVM::ConstantOp one;
       if (operandType.isVectorTy()) {
         one = rewriter.create<LLVM::ConstantOp>(
             loc, operandType,
             SplatElementsAttr::get(resultType.cast<ShapedType>(), floatOne));
       } else {
         one = rewriter.create<LLVM::ConstantOp>(loc, operandType, floatOne);
       }
       auto sqrt = rewriter.create<LLVM::SqrtOp>(loc, transformed.operand());
       rewriter.replaceOpWithNewOp<LLVM::FDivOp>(op, operandType, one, sqrt);
       return success();
     }
 
     auto vectorType = resultType.dyn_cast<VectorType>();
     if (!vectorType)
       return failure();
 
     return handleMultidimensionalVectors(
-        op, operands, typeConverter,
+        op.getOperation(), operands, typeConverter,
         [&](LLVM::LLVMType llvmVectorTy, ValueRange operands) {
           auto splatAttr = SplatElementsAttr::get(
               mlir::VectorType::get({llvmVectorTy.getVectorNumElements()},
                                     floatType),
               floatOne);
           auto one =
               rewriter.create<LLVM::ConstantOp>(loc, llvmVectorTy, splatAttr);
           auto sqrt =
               rewriter.create<LLVM::SqrtOp>(loc, llvmVectorTy, operands[0]);
           return rewriter.create<LLVM::FDivOp>(loc, llvmVectorTy, one, sqrt);
         },
         rewriter);
   }
 };
 
 struct MemRefCastOpLowering : public ConvertOpToLLVMPattern<MemRefCastOp> {
   using ConvertOpToLLVMPattern<MemRefCastOp>::ConvertOpToLLVMPattern;
 
-  LogicalResult match(Operation *op) const override {
-    auto memRefCastOp = cast<MemRefCastOp>(op);
+  LogicalResult match(MemRefCastOp memRefCastOp) const override {
     Type srcType = memRefCastOp.getOperand().getType();
     Type dstType = memRefCastOp.getType();
 
     // MemRefCastOp reduce to bitcast in the ranked MemRef case and can be used
     // for type erasure. For now they must preserve underlying element type and
     // require source and result type to have the same rank. Therefore, perform
     // a sanity check that the underlying structs are the same. Once op
     // semantics are relaxed we can revisit.
     if (srcType.isa<MemRefType>() && dstType.isa<MemRefType>())
       return success(typeConverter.convertType(srcType) ==
                      typeConverter.convertType(dstType));
 
     // At least one of the operands is unranked type
     assert(srcType.isa<UnrankedMemRefType>() ||
            dstType.isa<UnrankedMemRefType>());
 
     // Unranked to unranked cast is disallowed
     return !(srcType.isa<UnrankedMemRefType>() &&
              dstType.isa<UnrankedMemRefType>())
                ? success()
                : failure();
   }
 
-  void rewrite(Operation *op, ArrayRef<Value> operands,
+  void rewrite(MemRefCastOp memRefCastOp, ArrayRef<Value> operands,
                ConversionPatternRewriter &rewriter) const override {
-    auto memRefCastOp = cast<MemRefCastOp>(op);
     MemRefCastOp::Adaptor transformed(operands);
 
     auto srcType = memRefCastOp.getOperand().getType();
     auto dstType = memRefCastOp.getType();
     auto targetStructType = typeConverter.convertType(memRefCastOp.getType());
-    auto loc = op->getLoc();
+    auto loc = memRefCastOp.getLoc();
 
     // For ranked/ranked case, just keep the original descriptor.
     if (srcType.isa<MemRefType>() && dstType.isa<MemRefType>())
-      return rewriter.replaceOp(op, {transformed.source()});
+      return rewriter.replaceOp(memRefCastOp, {transformed.source()});
 
     if (srcType.isa<MemRefType>() && dstType.isa<UnrankedMemRefType>()) {
       // Casting ranked to unranked memref type
       // Set the rank in the destination from the memref type
       // Allocate space on the stack and copy the src memref descriptor
       // Set the ptr in the destination to the stack space
       auto srcMemRefType = srcType.cast<MemRefType>();
       int64_t rank = srcMemRefType.getRank();
if (srcType.isa<MemRefType>() && dstType.isa<UnrankedMemRefType>()) {
           rewriter.getI64IntegerAttr(rank));
       // undef = UndefOp
       UnrankedMemRefDescriptor memRefDesc =
           UnrankedMemRefDescriptor::undef(rewriter, loc, targetStructType);
       // d1 = InsertValueOp undef, rank, 0
       memRefDesc.setRank(rewriter, loc, rankVal);
       // d2 = InsertValueOp d1, voidptr, 1
       memRefDesc.setMemRefDescPtr(rewriter, loc, voidPtr);
-      rewriter.replaceOp(op, (Value)memRefDesc);
+      rewriter.replaceOp(memRefCastOp, (Value)memRefDesc);
 
     } else if (srcType.isa<UnrankedMemRefType>() && dstType.isa<MemRefType>()) {
       // Casting from unranked type to ranked.
       // The operation is assumed to be doing a correct cast. If the destination
       // type mismatches the unranked the type, it is undefined behavior.
       UnrankedMemRefDescriptor memRefDesc(transformed.source());
       // ptr = ExtractValueOp src, 1
       auto ptr = memRefDesc.memRefDescPtr(rewriter, loc);
       // castPtr = BitCastOp i8* to structTy*
       auto castPtr =
           rewriter
               .create<LLVM::BitcastOp>(
                   loc, targetStructType.cast<LLVM::LLVMType>().getPointerTo(),
                   ptr)
               .getResult();
       // struct = LoadOp castPtr
       auto loadOp = rewriter.create<LLVM::LoadOp>(loc, castPtr);
-      rewriter.replaceOp(op, loadOp.getResult());
+      rewriter.replaceOp(memRefCastOp, loadOp.getResult());
     } else {
       llvm_unreachable("Unsupported unranked memref to unranked memref cast");
     }
   }
 };
 
 /// Extracts allocated, aligned pointers and offset from a ranked or unranked
 /// memref type. In unranked case, the fields are extracted from the underlying
static void extractPointersAndOffset(Location loc,
   }
 }
 
 struct MemRefReinterpretCastOpLowering
     : public ConvertOpToLLVMPattern<MemRefReinterpretCastOp> {
   using ConvertOpToLLVMPattern<MemRefReinterpretCastOp>::ConvertOpToLLVMPattern;
 
   LogicalResult
-  matchAndRewrite(Operation *op, ArrayRef<Value> operands,
+  matchAndRewrite(MemRefReinterpretCastOp castOp, ArrayRef<Value> operands,
                   ConversionPatternRewriter &rewriter) const override {
-    auto castOp = cast<MemRefReinterpretCastOp>(op);
-    MemRefReinterpretCastOp::Adaptor adaptor(operands, op->getAttrDictionary());
+    MemRefReinterpretCastOp::Adaptor adaptor(
+        operands, castOp.getOperation()->getAttrDictionary());
     Type srcType = castOp.source().getType();
 
     Value descriptor;
     if (failed(convertSourceMemRefToDescriptor(rewriter, srcType, castOp,
                                                adaptor, &descriptor)))
       return failure();
-    rewriter.replaceOp(op, {descriptor});
+    rewriter.replaceOp(castOp, {descriptor});
     return success();
   }
 
 private:
   LogicalResult
   convertSourceMemRefToDescriptor(ConversionPatternRewriter &rewriter,
                                   Type srcType, MemRefReinterpretCastOp castOp,
                                   MemRefReinterpretCastOp::Adaptor adaptor,
private:
   }
 };
 
 struct MemRefReshapeOpLowering
     : public ConvertOpToLLVMPattern<MemRefReshapeOp> {
   using ConvertOpToLLVMPattern<MemRefReshapeOp>::ConvertOpToLLVMPattern;
 
   LogicalResult
-  matchAndRewrite(Operation *op, ArrayRef<Value> operands,
+  matchAndRewrite(MemRefReshapeOp reshapeOp, ArrayRef<Value> operands,
                   ConversionPatternRewriter &rewriter) const override {
-    auto reshapeOp = cast<MemRefReshapeOp>(op);
-
+    auto *op = reshapeOp.getOperation();
     MemRefReshapeOp::Adaptor adaptor(operands, op->getAttrDictionary());
     Type srcType = reshapeOp.source().getType();
 
     Value descriptor;
     if (failed(convertSourceMemRefToDescriptor(rewriter, srcType, reshapeOp,
                                                adaptor, &descriptor)))
       return failure();
     rewriter.replaceOp(op, {descriptor});
     return success();
private:
   }
 };
 
 struct DialectCastOpLowering
     : public ConvertOpToLLVMPattern<LLVM::DialectCastOp> {
   using ConvertOpToLLVMPattern<LLVM::DialectCastOp>::ConvertOpToLLVMPattern;
 
   LogicalResult
-  matchAndRewrite(Operation *op, ArrayRef<Value> operands,
+  matchAndRewrite(LLVM::DialectCastOp castOp, ArrayRef<Value> operands,
                   ConversionPatternRewriter &rewriter) const override {
-    auto castOp = cast<LLVM::DialectCastOp>(op);
     LLVM::DialectCastOp::Adaptor transformed(operands);
     if (transformed.in().getType() !=
         typeConverter.convertType(castOp.getType())) {
       return failure();
     }
-    rewriter.replaceOp(op, transformed.in());
+    rewriter.replaceOp(castOp, transformed.in());
     return success();
   }
 };
 
 // A `dim` is converted to a constant for static sizes and to an access to the
 // size stored in the memref descriptor for dynamic sizes.
 struct DimOpLowering : public ConvertOpToLLVMPattern<DimOp> {
   using ConvertOpToLLVMPattern<DimOp>::ConvertOpToLLVMPattern;
 
   LogicalResult
-  matchAndRewrite(Operation *op, ArrayRef<Value> operands,
+  matchAndRewrite(DimOp dimOp, ArrayRef<Value> operands,
                   ConversionPatternRewriter &rewriter) const override {
-    auto dimOp = cast<DimOp>(op);
     Type operandType = dimOp.memrefOrTensor().getType();
     if (operandType.isa<UnrankedMemRefType>()) {
-      rewriter.replaceOp(op, {extractSizeOfUnrankedMemRef(operandType, dimOp,
-                                                          operands, rewriter)});
+      rewriter.replaceOp(dimOp, {extractSizeOfUnrankedMemRef(
+                                    operandType, dimOp, operands, rewriter)});
 
       return success();
     }
     if (operandType.isa<MemRefType>()) {
-      rewriter.replaceOp(op, {extractSizeOfRankedMemRef(operandType, dimOp,
-                                                        operands, rewriter)});
+      rewriter.replaceOp(dimOp, {extractSizeOfRankedMemRef(
+                                    operandType, dimOp, operands, rewriter)});
       return success();
     }
     return failure();
   }
 
 private:
   Value extractSizeOfUnrankedMemRef(Type operandType, DimOp dimOp,
                                     ArrayRef<Value> operands,
Value extractSizeOfRankedMemRef(Type operandType, DimOp dimOp,
     return memrefDescriptor.size(rewriter, loc, index, rank);
   }
 };
 
 struct RankOpLowering : public ConvertOpToLLVMPattern<RankOp> {
   using ConvertOpToLLVMPattern<RankOp>::ConvertOpToLLVMPattern;
 
   LogicalResult
-  matchAndRewrite(Operation *op, ArrayRef<Value> operands,
+  matchAndRewrite(RankOp op, ArrayRef<Value> operands,
                   ConversionPatternRewriter &rewriter) const override {
-    Location loc = op->getLoc();
-    Type operandType = cast<RankOp>(op).memrefOrTensor().getType();
+    Location loc = op.getLoc();
+    Type operandType = op.memrefOrTensor().getType();
     if (auto unrankedMemRefType = operandType.dyn_cast<UnrankedMemRefType>()) {
       UnrankedMemRefDescriptor desc(RankOp::Adaptor(operands).memrefOrTensor());
       rewriter.replaceOp(op, {desc.rank(rewriter, loc)});
       return success();
     }
     if (auto rankedMemRefType = operandType.dyn_cast<MemRefType>()) {
       rewriter.replaceOp(
           op, {createIndexConstant(rewriter, loc, rankedMemRefType.getRank())});
       return success();
     }
     return failure();
   }
 };
 
 // Common base for load and store operations on MemRefs.  Restricts the match
 // to supported MemRef types.  Provides functionality to emit code accessing a
 // specific element of the underlying data buffer.
 template <typename Derived>
 struct LoadStoreOpLowering : public ConvertOpToLLVMPattern<Derived> {
   using ConvertOpToLLVMPattern<Derived>::ConvertOpToLLVMPattern;
   using ConvertOpToLLVMPattern<Derived>::isSupportedMemRefType;
   using Base = LoadStoreOpLowering<Derived>;
 
-  LogicalResult match(Operation *op) const override {
-    MemRefType type = cast<Derived>(op).getMemRefType();
+  LogicalResult match(Derived op) const override {
+    MemRefType type = op.getMemRefType();
     return isSupportedMemRefType(type) ? success() : failure();
   }
 };
 
 // Load operation is lowered to obtaining a pointer to the indexed element
 // and loading it.
 struct LoadOpLowering : public LoadStoreOpLowering<LoadOp> {
   using Base::Base;
 
   LogicalResult
-  matchAndRewrite(Operation *op, ArrayRef<Value> operands,
+  matchAndRewrite(LoadOp loadOp, ArrayRef<Value> operands,
                   ConversionPatternRewriter &rewriter) const override {
-    auto loadOp = cast<LoadOp>(op);
     LoadOp::Adaptor transformed(operands);
     auto type = loadOp.getMemRefType();
 
     Value dataPtr =
-        getStridedElementPtr(op->getLoc(), type, transformed.memref(),
+        getStridedElementPtr(loadOp.getLoc(), type, transformed.memref(),
                              transformed.indices(), rewriter);
-    rewriter.replaceOpWithNewOp<LLVM::LoadOp>(op, dataPtr);
+    rewriter.replaceOpWithNewOp<LLVM::LoadOp>(loadOp, dataPtr);
     return success();
   }
 };
 
 // Store operation is lowered to obtaining a pointer to the indexed element,
 // and storing the given value to it.
 struct StoreOpLowering : public LoadStoreOpLowering<StoreOp> {
   using Base::Base;
 
   LogicalResult
-  matchAndRewrite(Operation *op, ArrayRef<Value> operands,
+  matchAndRewrite(StoreOp op, ArrayRef<Value> operands,
                   ConversionPatternRewriter &rewriter) const override {
-    auto type = cast<StoreOp>(op).getMemRefType();
+    auto type = op.getMemRefType();
     StoreOp::Adaptor transformed(operands);
 
     Value dataPtr =
-        getStridedElementPtr(op->getLoc(), type, transformed.memref(),
+        getStridedElementPtr(op.getLoc(), type, transformed.memref(),
                              transformed.indices(), rewriter);
     rewriter.replaceOpWithNewOp<LLVM::StoreOp>(op, transformed.value(),
                                                dataPtr);
     return success();
   }
 };
 
 // The prefetch operation is lowered in a way similar to the load operation
 // except that the llvm.prefetch operation is used for replacement.
 struct PrefetchOpLowering : public LoadStoreOpLowering<PrefetchOp> {
   using Base::Base;
 
   LogicalResult
-  matchAndRewrite(Operation *op, ArrayRef<Value> operands,
+  matchAndRewrite(PrefetchOp prefetchOp, ArrayRef<Value> operands,
                   ConversionPatternRewriter &rewriter) const override {
-    auto prefetchOp = cast<PrefetchOp>(op);
     PrefetchOp::Adaptor transformed(operands);
     auto type = prefetchOp.getMemRefType();
+    auto loc = prefetchOp.getLoc();
 
-    Value dataPtr =
-        getStridedElementPtr(op->getLoc(), type, transformed.memref(),
+    Value dataPtr = getStridedElementPtr(loc, type, transformed.memref(),
                                          transformed.indices(), rewriter);
 
     // Replace with llvm.prefetch.
     auto llvmI32Type = typeConverter.convertType(rewriter.getIntegerType(32));
     auto isWrite = rewriter.create<LLVM::ConstantOp>(
-        op->getLoc(), llvmI32Type,
-        rewriter.getI32IntegerAttr(prefetchOp.isWrite()));
+        loc, llvmI32Type, rewriter.getI32IntegerAttr(prefetchOp.isWrite()));
     auto localityHint = rewriter.create<LLVM::ConstantOp>(
-        op->getLoc(), llvmI32Type,
+        loc, llvmI32Type,
         rewriter.getI32IntegerAttr(prefetchOp.localityHint()));
     auto isData = rewriter.create<LLVM::ConstantOp>(
-        op->getLoc(), llvmI32Type,
-        rewriter.getI32IntegerAttr(prefetchOp.isDataCache()));
+        loc, llvmI32Type, rewriter.getI32IntegerAttr(prefetchOp.isDataCache()));
 
-    rewriter.replaceOpWithNewOp<LLVM::Prefetch>(op, dataPtr, isWrite,
+    rewriter.replaceOpWithNewOp<LLVM::Prefetch>(prefetchOp, dataPtr, isWrite,
                                                 localityHint, isData);
     return success();
   }
 };
 
 // The lowering of index_cast becomes an integer conversion since index becomes
 // an integer.  If the bit width of the source and target integer types is the
 // same, just erase the cast.  If the target type is wider, sign-extend the
 // value, otherwise truncate it.
 struct IndexCastOpLowering : public ConvertOpToLLVMPattern<IndexCastOp> {
   using ConvertOpToLLVMPattern<IndexCastOp>::ConvertOpToLLVMPattern;
 
   LogicalResult
-  matchAndRewrite(Operation *op, ArrayRef<Value> operands,
+  matchAndRewrite(IndexCastOp indexCastOp, ArrayRef<Value> operands,
                   ConversionPatternRewriter &rewriter) const override {
     IndexCastOpAdaptor transformed(operands);
-    auto indexCastOp = cast<IndexCastOp>(op);
 
     auto targetType =
         this->typeConverter.convertType(indexCastOp.getResult().getType())
             .cast<LLVM::LLVMType>();
     auto sourceType = transformed.in().getType().cast<LLVM::LLVMType>();
     unsigned targetBits = targetType.getIntegerBitWidth();
     unsigned sourceBits = sourceType.getIntegerBitWidth();
 
     if (targetBits == sourceBits)
-      rewriter.replaceOp(op, transformed.in());
+      rewriter.replaceOp(indexCastOp, transformed.in());
     else if (targetBits < sourceBits)
-      rewriter.replaceOpWithNewOp<LLVM::TruncOp>(op, targetType,
+      rewriter.replaceOpWithNewOp<LLVM::TruncOp>(indexCastOp, targetType,
                                                  transformed.in());
     else
-      rewriter.replaceOpWithNewOp<LLVM::SExtOp>(op, targetType,
+      rewriter.replaceOpWithNewOp<LLVM::SExtOp>(indexCastOp, targetType,
                                                 transformed.in());
     return success();
   }
 };
 
 // Convert std.cmp predicate into the LLVM dialect CmpPredicate.  The two
 // enums share the numerical values so just cast.
 template <typename LLVMPredType, typename StdPredType>
 static LLVMPredType convertCmpPredicate(StdPredType pred) {
   return static_cast<LLVMPredType>(pred);
 }
 
 struct CmpIOpLowering : public ConvertOpToLLVMPattern<CmpIOp> {
   using ConvertOpToLLVMPattern<CmpIOp>::ConvertOpToLLVMPattern;
 
   LogicalResult
-  matchAndRewrite(Operation *op, ArrayRef<Value> operands,
+  matchAndRewrite(CmpIOp cmpiOp, ArrayRef<Value> operands,
                   ConversionPatternRewriter &rewriter) const override {
-    auto cmpiOp = cast<CmpIOp>(op);
     CmpIOpAdaptor transformed(operands);
 
     rewriter.replaceOpWithNewOp<LLVM::ICmpOp>(
-        op, typeConverter.convertType(cmpiOp.getResult().getType()),
+        cmpiOp, typeConverter.convertType(cmpiOp.getResult().getType()),
         rewriter.getI64IntegerAttr(static_cast<int64_t>(
             convertCmpPredicate<LLVM::ICmpPredicate>(cmpiOp.getPredicate()))),
         transformed.lhs(), transformed.rhs());
 
     return success();
   }
 };
 
 struct CmpFOpLowering : public ConvertOpToLLVMPattern<CmpFOp> {
   using ConvertOpToLLVMPattern<CmpFOp>::ConvertOpToLLVMPattern;
 
   LogicalResult
-  matchAndRewrite(Operation *op, ArrayRef<Value> operands,
+  matchAndRewrite(CmpFOp cmpfOp, ArrayRef<Value> operands,
                   ConversionPatternRewriter &rewriter) const override {
-    auto cmpfOp = cast<CmpFOp>(op);
     CmpFOpAdaptor transformed(operands);
 
     rewriter.replaceOpWithNewOp<LLVM::FCmpOp>(
-        op, typeConverter.convertType(cmpfOp.getResult().getType()),
+        cmpfOp, typeConverter.convertType(cmpfOp.getResult().getType()),
         rewriter.getI64IntegerAttr(static_cast<int64_t>(
             convertCmpPredicate<LLVM::FCmpPredicate>(cmpfOp.getPredicate()))),
         transformed.lhs(), transformed.rhs());
 
     return success();
   }
 };
 

 // Base class for LLVM IR lowering terminator operations with successors.
 template <typename SourceOp, typename TargetOp>
 struct OneToOneLLVMTerminatorLowering
     : public ConvertOpToLLVMPattern<SourceOp> {
   using ConvertOpToLLVMPattern<SourceOp>::ConvertOpToLLVMPattern;
   using Super = OneToOneLLVMTerminatorLowering<SourceOp, TargetOp>;
 
   LogicalResult
-  matchAndRewrite(Operation *op, ArrayRef<Value> operands,
+  matchAndRewrite(SourceOp op, ArrayRef<Value> operands,
                   ConversionPatternRewriter &rewriter) const override {
-    rewriter.replaceOpWithNewOp<TargetOp>(op, operands, op->getSuccessors(),
-                                          op->getAttrs());
+    rewriter.replaceOpWithNewOp<TargetOp>(
+        op, operands, op.getOperation()->getSuccessors(), op.getAttrs());
     return success();
   }
 };
 
 // Special lowering pattern for `ReturnOps`.  Unlike all other operations,
 // `ReturnOp` interacts with the function signature and must have as many
 // operands as the function has return values.  Because in LLVM IR, functions
 // can only return 0 or 1 value, we pack multiple values into a structure type.
 // Emit `UndefOp` followed by `InsertValueOp`s to create such structure if
 // necessary before returning it
 struct ReturnOpLowering : public ConvertOpToLLVMPattern<ReturnOp> {
   using ConvertOpToLLVMPattern<ReturnOp>::ConvertOpToLLVMPattern;
 
   LogicalResult
-  matchAndRewrite(Operation *op, ArrayRef<Value> operands,
+  matchAndRewrite(ReturnOp op, ArrayRef<Value> operands,
                   ConversionPatternRewriter &rewriter) const override {
-    Location loc = op->getLoc();
-    unsigned numArguments = op->getNumOperands();
+    Location loc = op.getLoc();
+    unsigned numArguments = op.getNumOperands();
     SmallVector<Value, 4> updatedOperands;
 
     if (typeConverter.getOptions().useBarePtrCallConv) {
       // For the bare-ptr calling convention, extract the aligned pointer to
       // be returned from the memref descriptor.
-      for (auto it : llvm::zip(op->getOperands(), operands)) {
+      for (auto it : llvm::zip(op.getOperation()->getOperands(), operands)) {
         Type oldTy = std::get<0>(it).getType();
         Value newOperand = std::get<1>(it);
         if (oldTy.isa<MemRefType>()) {
           MemRefDescriptor memrefDesc(newOperand);
           newOperand = memrefDesc.alignedPtr(rewriter, loc);
         } else if (oldTy.isa<UnrankedMemRefType>()) {
           // Unranked memref is not supported in the bare pointer calling
           // convention.
           return failure();
         }
         updatedOperands.push_back(newOperand);
       }
     } else {
       updatedOperands = llvm::to_vector<4>(operands);
       copyUnrankedDescriptors(rewriter, loc, typeConverter,
-                              op->getOperands().getTypes(), updatedOperands,
+                              op.getOperands().getTypes(), updatedOperands,
                               /*toDynamic=*/true);
     }
 
     // If ReturnOp has 0 or 1 operand, create it and return immediately.
     if (numArguments == 0) {
       rewriter.replaceOpWithNewOp<LLVM::ReturnOp>(op, TypeRange(), ValueRange(),
-                                                  op->getAttrs());
+                                                  op.getAttrs());
       return success();
     }
     if (numArguments == 1) {
       rewriter.replaceOpWithNewOp<LLVM::ReturnOp>(
-          op, TypeRange(), updatedOperands, op->getAttrs());
+          op, TypeRange(), updatedOperands, op.getAttrs());
       return success();
     }
 
     // Otherwise, we need to pack the arguments into an LLVM struct type before
     // returning.
     auto packedType = typeConverter.packFunctionResults(
-        llvm::to_vector<4>(op->getOperandTypes()));
+        llvm::to_vector<4>(op.getOperandTypes()));
 
     Value packed = rewriter.create<LLVM::UndefOp>(loc, packedType);
     for (unsigned i = 0; i < numArguments; ++i) {
       packed = rewriter.create<LLVM::InsertValueOp>(
           loc, packedType, packed, updatedOperands[i],
           rewriter.getI64ArrayAttr(i));
     }
     rewriter.replaceOpWithNewOp<LLVM::ReturnOp>(op, TypeRange(), packed,
-                                                op->getAttrs());
+                                                op.getAttrs());
     return success();
   }
 };
 
 // FIXME: this should be tablegen'ed as well.
 struct BranchOpLowering
     : public OneToOneLLVMTerminatorLowering<BranchOp, LLVM::BrOp> {
   using Super::Super;
 };
 struct CondBranchOpLowering
     : public OneToOneLLVMTerminatorLowering<CondBranchOp, LLVM::CondBrOp> {
   using Super::Super;
 };
 
 // The Splat operation is lowered to an insertelement + a shufflevector
 // operation. Splat to only 1-d vector result types are lowered.
 struct SplatOpLowering : public ConvertOpToLLVMPattern<SplatOp> {
   using ConvertOpToLLVMPattern<SplatOp>::ConvertOpToLLVMPattern;
 
   LogicalResult
-  matchAndRewrite(Operation *op, ArrayRef<Value> operands,
+  matchAndRewrite(SplatOp splatOp, ArrayRef<Value> operands,
                   ConversionPatternRewriter &rewriter) const override {
-    auto splatOp = cast<SplatOp>(op);
     VectorType resultType = splatOp.getType().dyn_cast<VectorType>();
     if (!resultType || resultType.getRank() != 1)
       return failure();
 
     // First insert it into an undef vector so we can shuffle it.
     auto vectorType = typeConverter.convertType(splatOp.getType());
-    Value undef = rewriter.create<LLVM::UndefOp>(op->getLoc(), vectorType);
+    Value undef = rewriter.create<LLVM::UndefOp>(splatOp.getLoc(), vectorType);
     auto zero = rewriter.create<LLVM::ConstantOp>(
-        op->getLoc(), typeConverter.convertType(rewriter.getIntegerType(32)),
+        splatOp.getLoc(),
+        typeConverter.convertType(rewriter.getIntegerType(32)),
         rewriter.getZeroAttr(rewriter.getIntegerType(32)));
 
     auto v = rewriter.create<LLVM::InsertElementOp>(
-        op->getLoc(), vectorType, undef, splatOp.getOperand(), zero);
+        splatOp.getLoc(), vectorType, undef, splatOp.getOperand(), zero);
 
     int64_t width = splatOp.getType().cast<VectorType>().getDimSize(0);
     SmallVector<int32_t, 4> zeroValues(width, 0);
 
     // Shuffle the value across the desired number of elements.
     ArrayAttr zeroAttrs = rewriter.getI32ArrayAttr(zeroValues);
-    rewriter.replaceOpWithNewOp<LLVM::ShuffleVectorOp>(op, v, undef, zeroAttrs);
+    rewriter.replaceOpWithNewOp<LLVM::ShuffleVectorOp>(splatOp, v, undef,
+                                                       zeroAttrs);
     return success();
   }
 };
 
 // The Splat operation is lowered to an insertelement + a shufflevector
 // operation. Splat to only 2+-d vector result types are lowered by the
 // SplatNdOpLowering, the 1-d case is handled by SplatOpLowering.
 struct SplatNdOpLowering : public ConvertOpToLLVMPattern<SplatOp> {
   using ConvertOpToLLVMPattern<SplatOp>::ConvertOpToLLVMPattern;
 
   LogicalResult
-  matchAndRewrite(Operation *op, ArrayRef<Value> operands,
+  matchAndRewrite(SplatOp splatOp, ArrayRef<Value> operands,
                   ConversionPatternRewriter &rewriter) const override {
-    auto splatOp = cast<SplatOp>(op);
     SplatOp::Adaptor adaptor(operands);
     VectorType resultType = splatOp.getType().dyn_cast<VectorType>();
     if (!resultType || resultType.getRank() == 1)
       return failure();
 
     // First insert it into an undef vector so we can shuffle it.
-    auto loc = op->getLoc();
+    auto loc = splatOp.getLoc();
     auto vectorTypeInfo = extractNDVectorTypeInfo(resultType, typeConverter);
     auto llvmArrayTy = vectorTypeInfo.llvmArrayTy;
     auto llvmVectorTy = vectorTypeInfo.llvmVectorTy;
     if (!llvmArrayTy || !llvmVectorTy)
       return failure();
 
     // Construct returned value.
     Value desc = rewriter.create<LLVM::UndefOp>(loc, llvmArrayTy);
matchAndRewrite(SplatOp splatOp, ArrayRef<Value> operands,
     v = rewriter.create<LLVM::ShuffleVectorOp>(loc, v, v, zeroAttrs);
 
     // Iterate of linear index, convert to coords space and insert splatted 1-D
     // vector in each position.
     nDVectorIterate(vectorTypeInfo, rewriter, [&](ArrayAttr position) {
       desc = rewriter.create<LLVM::InsertValueOp>(loc, llvmArrayTy, desc, v,
                                                   position);
     });
-    rewriter.replaceOp(op, desc);
+    rewriter.replaceOp(splatOp, desc);
     return success();
   }
 };
 
 /// Helper function extracts int64_t from the assumedArrayAttr of IntegerAttr.
 static SmallVector<int64_t, 4> extractFromI64ArrayAttr(Attribute attr) {
   return llvm::to_vector<4>(
       llvm::map_range(attr.cast<ArrayAttr>(), [](Attribute a) -> int64_t {
         return a.cast<IntegerAttr>().getInt();
       }));
 }
 
 /// Conversion pattern that transforms a subview op into:
 ///   1. An `llvm.mlir.undef` operation to create a memref descriptor
 ///   2. Updates to the descriptor to introduce the data ptr, offset, size
 ///      and stride.
 /// The subview op is replaced by the descriptor.
 struct SubViewOpLowering : public ConvertOpToLLVMPattern<SubViewOp> {
   using ConvertOpToLLVMPattern<SubViewOp>::ConvertOpToLLVMPattern;
 
   LogicalResult
-  matchAndRewrite(Operation *op, ArrayRef<Value> operands,
+  matchAndRewrite(SubViewOp subViewOp, ArrayRef<Value> operands,
                   ConversionPatternRewriter &rewriter) const override {
-    auto loc = op->getLoc();
-    auto subViewOp = cast<SubViewOp>(op);
+    auto loc = subViewOp.getLoc();
 
     auto sourceMemRefType = subViewOp.source().getType().cast<MemRefType>();
     auto sourceElementTy =
         typeConverter.convertType(sourceMemRefType.getElementType())
             .dyn_cast_or_null<LLVM::LLVMType>();
 
     auto viewMemRefType = subViewOp.getType();
     auto inferredType = SubViewOp::inferResultType(
for (int i = inferredShapeRank - 1, j = resultShapeRank - 1;
                       loc, llvmIndexType,
                       rewriter.getI64IntegerAttr(subViewOp.getStaticStride(i)));
         stride = rewriter.create<LLVM::MulOp>(loc, stride, strideValues[i]);
       }
       targetMemRef.setStride(rewriter, loc, j, stride);
       j--;
     }
 
-    rewriter.replaceOp(op, {targetMemRef});
+    rewriter.replaceOp(subViewOp, {targetMemRef});
     return success();
   }
 };
 
 /// Conversion pattern that transforms a transpose op into:
 ///   1. A function entry `alloca` operation to allocate a ViewDescriptor.
 ///   2. A load of the ViewDescriptor from the pointer allocated in 1.
 ///   3. Updates to the ViewDescriptor to introduce the data ptr, offset, size
 ///      and stride. Size and stride are permutations of the original values.
 ///   4. A store of the resulting ViewDescriptor to the alloca'ed pointer.
 /// The transpose op is replaced by the alloca'ed pointer.
 class TransposeOpLowering : public ConvertOpToLLVMPattern<TransposeOp> {
 public:
   using ConvertOpToLLVMPattern<TransposeOp>::ConvertOpToLLVMPattern;
 
   LogicalResult
-  matchAndRewrite(Operation *op, ArrayRef<Value> operands,
+  matchAndRewrite(TransposeOp transposeOp, ArrayRef<Value> operands,
                   ConversionPatternRewriter &rewriter) const override {
-    auto loc = op->getLoc();
+    auto loc = transposeOp.getLoc();
     TransposeOpAdaptor adaptor(operands);
     MemRefDescriptor viewMemRef(adaptor.in());
 
-    auto transposeOp = cast<TransposeOp>(op);
     // No permutation, early exit.
     if (transposeOp.permutation().isIdentity())
-      return rewriter.replaceOp(op, {viewMemRef}), success();
+      return rewriter.replaceOp(transposeOp, {viewMemRef}), success();
 
     auto targetMemRef = MemRefDescriptor::undef(
         rewriter, loc, typeConverter.convertType(transposeOp.getShapedType()));
 
     // Copy the base and aligned pointers from the old descriptor to the new
     // one.
     targetMemRef.setAllocatedPtr(rewriter, loc,
                                  viewMemRef.allocatedPtr(rewriter, loc));
     targetMemRef.setAlignedPtr(rewriter, loc,
                                viewMemRef.alignedPtr(rewriter, loc));
 
     // Copy the offset pointer from the old descriptor to the new one.
     targetMemRef.setOffset(rewriter, loc, viewMemRef.offset(rewriter, loc));
 
     // Iterate over the dimensions and apply size/stride permutation.
     for (auto en : llvm::enumerate(transposeOp.permutation().getResults())) {
       int sourcePos = en.index();
       int targetPos = en.value().cast<AffineDimExpr>().getPosition();
       targetMemRef.setSize(rewriter, loc, targetPos,
                            viewMemRef.size(rewriter, loc, sourcePos));
       targetMemRef.setStride(rewriter, loc, targetPos,
                              viewMemRef.stride(rewriter, loc, sourcePos));
     }
 
-    rewriter.replaceOp(op, {targetMemRef});
+    rewriter.replaceOp(transposeOp, {targetMemRef});
     return success();
   }
 };
 
 /// Conversion pattern that transforms an op into:
 ///   1. An `llvm.mlir.undef` operation to create a memref descriptor
 ///   2. Updates to the descriptor to introduce the data ptr, offset, size
 ///      and stride.
if (nextSize)
       return runningStride
                  ? rewriter.create<LLVM::MulOp>(loc, runningStride, nextSize)
                  : nextSize;
     assert(!runningStride);
     return createIndexConstant(rewriter, loc, 1);
   }
 
   LogicalResult
-  matchAndRewrite(Operation *op, ArrayRef<Value> operands,
+  matchAndRewrite(ViewOp viewOp, ArrayRef<Value> operands,
                   ConversionPatternRewriter &rewriter) const override {
-    auto loc = op->getLoc();
-    auto viewOp = cast<ViewOp>(op);
+    auto loc = viewOp.getLoc();
     ViewOpAdaptor adaptor(operands);
 
     auto viewMemRefType = viewOp.getType();
     auto targetElementTy =
         typeConverter.convertType(viewMemRefType.getElementType())
             .dyn_cast<LLVM::LLVMType>();
     auto targetDescTy =
         typeConverter.convertType(viewMemRefType).dyn_cast<LLVM::LLVMType>();
     if (!targetDescTy)
-      return op->emitWarning("Target descriptor type not converted to LLVM"),
+      return viewOp.emitWarning("Target descriptor type not converted to LLVM"),
              failure();
 
     int64_t offset;
     SmallVector<int64_t, 4> strides;
     auto successStrides = getStridesAndOffset(viewMemRefType, strides, offset);
     if (failed(successStrides))
-      return op->emitWarning("cannot cast to non-strided shape"), failure();
+      return viewOp.emitWarning("cannot cast to non-strided shape"), failure();
     assert(offset == 0 && "expected offset to be 0");
 
     // Create the descriptor.
     MemRefDescriptor sourceMemRef(adaptor.source());
     auto targetMemRef = MemRefDescriptor::undef(rewriter, loc, targetDescTy);
 
     // Field 1: Copy the allocated pointer, used for malloc/free.
     Value allocatedPtr = sourceMemRef.allocatedPtr(rewriter, loc);
matchAndRewrite(ViewOp viewOp, ArrayRef<Value> operands,
     // Field 3: The offset in the resulting type must be 0. This is because of
     // the type change: an offset on srcType* may not be expressible as an
     // offset on dstType*.
     targetMemRef.setOffset(rewriter, loc,
                            createIndexConstant(rewriter, loc, offset));
 
     // Early exit for 0-D corner case.
     if (viewMemRefType.getRank() == 0)
-      return rewriter.replaceOp(op, {targetMemRef}), success();
+      return rewriter.replaceOp(viewOp, {targetMemRef}), success();
 
     // Fields 4 and 5: Update sizes and strides.
     if (strides.back() != 1)
-      return op->emitWarning("cannot cast to non-contiguous shape"), failure();
+      return viewOp.emitWarning("cannot cast to non-contiguous shape"),
+             failure();
     Value stride = nullptr, nextSize = nullptr;
     for (int i = viewMemRefType.getRank() - 1; i >= 0; --i) {
       // Update size.
       Value size =
           getSize(rewriter, loc, viewMemRefType.getShape(), adaptor.sizes(), i);
       targetMemRef.setSize(rewriter, loc, i, size);
       // Update stride.
       stride = getStride(rewriter, loc, strides, nextSize, stride, i);
       targetMemRef.setStride(rewriter, loc, i, stride);
       nextSize = size;
     }
 
-    rewriter.replaceOp(op, {targetMemRef});
+    rewriter.replaceOp(viewOp, {targetMemRef});
     return success();
   }
 };
 
 struct AssumeAlignmentOpLowering
     : public ConvertOpToLLVMPattern<AssumeAlignmentOp> {
   using ConvertOpToLLVMPattern<AssumeAlignmentOp>::ConvertOpToLLVMPattern;
 
   LogicalResult
-  matchAndRewrite(Operation *op, ArrayRef<Value> operands,
+  matchAndRewrite(AssumeAlignmentOp op, ArrayRef<Value> operands,
                   ConversionPatternRewriter &rewriter) const override {
     AssumeAlignmentOp::Adaptor transformed(operands);
     Value memref = transformed.memref();
-    unsigned alignment = cast<AssumeAlignmentOp>(op).alignment();
+    unsigned alignment = op.alignment();
+    auto loc = op.getLoc();
 
     MemRefDescriptor memRefDescriptor(memref);
     Value ptr = memRefDescriptor.alignedPtr(rewriter, memref.getLoc());
 
     // Emit llvm.assume(memref.alignedPtr & (alignment - 1) == 0). Notice that
     // the asserted memref.alignedPtr isn't used anywhere else, as the real
     // users like load/store/views always re-extract memref.alignedPtr as they
     // get lowered.
     //
     // This relies on LLVM's CSE optimization (potentially after SROA), since
     // after CSE all memref.alignedPtr instances get de-duplicated into the same
     // pointer SSA value.
     auto intPtrType =
         getIntPtrType(memRefDescriptor.getElementPtrType().getAddressSpace());
-    Value zero = createIndexAttrConstant(rewriter, op->getLoc(), intPtrType, 0);
-    Value mask = createIndexAttrConstant(rewriter, op->getLoc(), intPtrType,
-                                         alignment - 1);
-    Value ptrValue =
-        rewriter.create<LLVM::PtrToIntOp>(op->getLoc(), intPtrType, ptr);
+    Value zero = createIndexAttrConstant(rewriter, loc, intPtrType, 0);
+    Value mask =
+        createIndexAttrConstant(rewriter, loc, intPtrType, alignment - 1);
+    Value ptrValue = rewriter.create<LLVM::PtrToIntOp>(loc, intPtrType, ptr);
     rewriter.create<LLVM::AssumeOp>(
-        op->getLoc(),
-        rewriter.create<LLVM::ICmpOp>(
-            op->getLoc(), LLVM::ICmpPredicate::eq,
-            rewriter.create<LLVM::AndOp>(op->getLoc(), ptrValue, mask), zero));
+        loc, rewriter.create<LLVM::ICmpOp>(
+                 loc, LLVM::ICmpPredicate::eq,
+                 rewriter.create<LLVM::AndOp>(loc, ptrValue, mask), zero));
 
     rewriter.eraseOp(op);
     return success();
   }
 };
 
 } // namespace
 

 }
 
 namespace {
 
 struct AtomicRMWOpLowering : public LoadStoreOpLowering<AtomicRMWOp> {
   using Base::Base;
 
   LogicalResult
-  matchAndRewrite(Operation *op, ArrayRef<Value> operands,
+  matchAndRewrite(AtomicRMWOp atomicOp, ArrayRef<Value> operands,
                   ConversionPatternRewriter &rewriter) const override {
-    auto atomicOp = cast<AtomicRMWOp>(op);
+    if (failed(match(atomicOp)))
+      return failure();
     auto maybeKind = matchSimpleAtomicOp(atomicOp);
     if (!maybeKind)
       return failure();
     AtomicRMWOp::Adaptor adaptor(operands);
     auto resultType = adaptor.value().getType();
     auto memRefType = atomicOp.getMemRefType();
     auto dataPtr =
-        getStridedElementPtr(op->getLoc(), memRefType, adaptor.memref(),
+        getStridedElementPtr(atomicOp.getLoc(), memRefType, adaptor.memref(),
                              adaptor.indices(), rewriter);
     rewriter.replaceOpWithNewOp<LLVM::AtomicRMWOp>(
-        op, resultType, *maybeKind, dataPtr, adaptor.value(),
+        atomicOp, resultType, *maybeKind, dataPtr, adaptor.value(),
         LLVM::AtomicOrdering::acq_rel);
     return success();
   }
 };
 
 /// Wrap a llvm.cmpxchg operation in a while loop so that the operation can be
 /// retried until it succeeds in atomically storing a new value into memory.
 ///

 ///      |   <code after the AtomicRMWOp> |
 ///      +--------------------------------+
 ///
 struct GenericAtomicRMWOpLowering
     : public LoadStoreOpLowering<GenericAtomicRMWOp> {
   using Base::Base;
 
   LogicalResult
-  matchAndRewrite(Operation *op, ArrayRef<Value> operands,
+  matchAndRewrite(GenericAtomicRMWOp atomicOp, ArrayRef<Value> operands,
                   ConversionPatternRewriter &rewriter) const override {
-    auto atomicOp = cast<GenericAtomicRMWOp>(op);
 
-    auto loc = op->getLoc();
+    auto loc = atomicOp.getLoc();
     GenericAtomicRMWOp::Adaptor adaptor(operands);
     LLVM::LLVMType valueType =
         typeConverter.convertType(atomicOp.getResult().getType())
             .cast<LLVM::LLVMType>();
 
     // Split the block into initial, loop, and ending parts.
     auto *initBlock = rewriter.getInsertionBlock();
     auto *loopBlock =
matchAndRewrite(GenericAtomicRMWOp atomicOp, ArrayRef<Value> operands,
     rewriter.create<LLVM::CondBrOp>(loc, ok, endBlock, ArrayRef<Value>(),
                                     loopBlock, newLoaded);
 
     rewriter.setInsertionPointToEnd(endBlock);
     moveOpsRange(atomicOp.getResult(), newLoaded, std::next(opsToMoveStart),
                  std::next(opsToMoveEnd), rewriter);
 
     // The 'result' of the atomic_rmw op is the newly loaded value.
-    rewriter.replaceOp(op, {newLoaded});
+    rewriter.replaceOp(atomicOp, {newLoaded});
 
     return success();
   }
 
 private:
   // Clones a segment of ops [start, end) and erases the original.
   void moveOpsRange(ValueRange oldResult, ValueRange newResult,
                     Block::iterator start, Block::iterator end,

mlir/test/lib/Transforms/TestConvertCallOp.cpp

 
 class TestTypeProducerOpConverter
     : public ConvertOpToLLVMPattern<test::TestTypeProducerOp> {
 public:
   using ConvertOpToLLVMPattern<
       test::TestTypeProducerOp>::ConvertOpToLLVMPattern;
 
   LogicalResult
-  matchAndRewrite(Operation *op, ArrayRef<Value> operands,
+  matchAndRewrite(test::TestTypeProducerOp op, ArrayRef<Value> operands,
                   ConversionPatternRewriter &rewriter) const override {
     rewriter.replaceOpWithNewOp<LLVM::NullOp>(op, getVoidPtrType());
     return success();
   }
 };
 
 class TestConvertCallOp
     : public PassWrapper<TestConvertCallOp, OperationPass<ModuleOp>> {