[SE] Remove Platform*Handle classes

[jhen]

Summary

As pointed out by jprice, these classes don't serve a purpose. Instead,
we stay consistent with the way memory is managed and let the Stream and
Kernel classes directly hold opaque handles to device Stream and Kernel
instances, respectively.

[jlebar]

Seems simpler to me!

streamexecutor/lib/Kernel.cpp

27 ↗

(On Diff #70261)

Can we assert that PDevice and PlatformKernelHandle are not null? Otherwise we'll find out that PDevice was null only when we destroy the KernelBase.

streamexecutor/lib/Stream.cpp

21 ↗

(On Diff #70261)

Same for assertions here.

[jprice]

Looks good to me.

If there are never going to be any other platform interfaces I guess it might make sense to rename PlatformInterfaces.* to just PlatformDevice.*, but maybe it's too early for that.

[jhen]

• assert no nulls in ctors

[jhen]

In D24213#533985, @jprice wrote:

If there are never going to be any other platform interfaces I guess it might make sense to rename PlatformInterfaces.* to just PlatformDevice.*, but maybe it's too early for that.

I agree that PlatformDevice is a a better name now. I'll get a patch up for that after this one goes in.

[jhen]

• dummy platform stream in Stream test

parallel-libs/trunk/streamexecutor/include/streamexecutor/Device.h

public:
   explicit Device(PlatformDevice *PDevice);
   virtual ~Device();
 
   /// Creates a kernel object for this device.
   template <typename KernelT>
   Expected<typename std::enable_if<std::is_base_of<KernelBase, KernelT>::value,
                                    KernelT>::type>
   createKernel(const MultiKernelLoaderSpec &Spec) {
-    Expected<std::unique_ptr<PlatformKernelHandle>> MaybeKernelHandle =
-        PDevice->createKernel(Spec);
+    Expected<const void *> MaybeKernelHandle = PDevice->createKernel(Spec);
     if (!MaybeKernelHandle) {
       return MaybeKernelHandle.takeError();
     }
-    return KernelT(Spec.getKernelName(), std::move(*MaybeKernelHandle));
+    return KernelT(PDevice, *MaybeKernelHandle, Spec.getKernelName());
   }
 
   /// Creates a stream object for this device.
   Expected<Stream> createStream();
 
   /// Allocates an array of ElementCount entries of type T in device memory.
   template <typename T>
   Expected<GlobalDeviceMemory<T>> allocateDeviceMemory(size_t ElementCount) {

parallel-libs/trunk/streamexecutor/include/streamexecutor/Kernel.h

 
 #include "streamexecutor/KernelSpec.h"
 #include "streamexecutor/Utils/Error.h"
 
 #include <memory>
 
 namespace streamexecutor {
 
-class PlatformKernelHandle;
+class PlatformDevice;
 
 /// The base class for all kernel types.
 ///
 /// Stores the name of the kernel in both mangled and demangled forms.
 class KernelBase {
 public:
-  KernelBase(llvm::StringRef Name);
+  KernelBase(PlatformDevice *D, const void *PlatformKernelHandle,
+             llvm::StringRef Name);
 
+  KernelBase(const KernelBase &Other) = delete;
+  KernelBase &operator=(const KernelBase &Other) = delete;
+
+  KernelBase(KernelBase &&Other);
+  KernelBase &operator=(KernelBase &&Other);
+
+  ~KernelBase();
+
+  const void *getPlatformHandle() const { return PlatformKernelHandle; }
   const std::string &getName() const { return Name; }
   const std::string &getDemangledName() const { return DemangledName; }
 
 private:
+  PlatformDevice *PDevice;
+  const void *PlatformKernelHandle;
+
   std::string Name;
   std::string DemangledName;
 };
 
 /// A StreamExecutor kernel.
 ///
 /// The template parameters are the types of the parameters to the kernel
 /// function.
 template <typename... ParameterTs> class Kernel : public KernelBase {
 public:
-  Kernel(llvm::StringRef Name, std::unique_ptr<PlatformKernelHandle> PHandle)
-      : KernelBase(Name), PHandle(std::move(PHandle)) {}
+  Kernel(PlatformDevice *D, const void *PlatformKernelHandle,
+         llvm::StringRef Name)
+      : KernelBase(D, PlatformKernelHandle, Name) {}
 
   Kernel(Kernel &&Other) = default;
   Kernel &operator=(Kernel &&Other) = default;
-
-  /// Gets the underlying platform-specific handle for this kernel.
-  PlatformKernelHandle *getPlatformHandle() const { return PHandle.get(); }
-
-private:
-  std::unique_ptr<PlatformKernelHandle> PHandle;
 };
 
 } // namespace streamexecutor
 
 #endif // STREAMEXECUTOR_KERNEL_H

parallel-libs/trunk/streamexecutor/include/streamexecutor/PlatformInterfaces.h

 #include "streamexecutor/DeviceMemory.h"
 #include "streamexecutor/Kernel.h"
 #include "streamexecutor/LaunchDimensions.h"
 #include "streamexecutor/PackedKernelArgumentArray.h"
 #include "streamexecutor/Utils/Error.h"
 
 namespace streamexecutor {
 
-class PlatformDevice;
-
-/// Platform-specific kernel handle.
-class PlatformKernelHandle {
-public:
-  explicit PlatformKernelHandle(PlatformDevice *PDevice) : PDevice(PDevice) {}
-
-  virtual ~PlatformKernelHandle();
-
-  PlatformDevice *getDevice() { return PDevice; }
-
-private:
-  PlatformDevice *PDevice;
-};
-
-/// Platform-specific stream handle.
-class PlatformStreamHandle {
-public:
-  explicit PlatformStreamHandle(PlatformDevice *PDevice) : PDevice(PDevice) {}
-
-  virtual ~PlatformStreamHandle();
-
-  PlatformDevice *getDevice() { return PDevice; }
-
-private:
-  PlatformDevice *PDevice;
-};
-
 /// Raw executor methods that must be implemented by each platform.
 ///
 /// This class defines the platform interface that supports executing work on a
 /// device.
 ///
 /// The public Device and Stream classes have the type-safe versions of the
 /// functions in this interface.
 class PlatformDevice {
 public:
   virtual ~PlatformDevice();
 
   virtual std::string getName() const = 0;
 
   /// Creates a platform-specific kernel.
-  virtual Expected<std::unique_ptr<PlatformKernelHandle>>
+  virtual Expected<const void *>
   createKernel(const MultiKernelLoaderSpec &Spec) {
     return make_error("createKernel not implemented for platform " + getName());
   }
 
+  virtual Error destroyKernel(const void *Handle) {
+    return make_error("destroyKernel not implemented for platform " +
+                      getName());
+  }
+
   /// Creates a platform-specific stream.
-  virtual Expected<std::unique_ptr<PlatformStreamHandle>> createStream() {
+  virtual Expected<const void *> createStream() {
     return make_error("createStream not implemented for platform " + getName());
   }
 
+  virtual Error destroyStream(const void *Handle) {
+    return make_error("destroyStream not implemented for platform " +
+                      getName());
+  }
+
   /// Launches a kernel on the given stream.
-  virtual Error launch(PlatformStreamHandle *S, BlockDimensions BlockSize,
-                       GridDimensions GridSize, PlatformKernelHandle *K,
+  virtual Error launch(const void *PlatformStreamHandle,
+                       BlockDimensions BlockSize, GridDimensions GridSize,
+                       const void *PKernelHandle,
                        const PackedKernelArgumentArrayBase &ArgumentArray) {
     return make_error("launch not implemented for platform " + getName());
   }
 
   /// Copies data from the device to the host.
   ///
   /// HostDst should have been allocated by allocateHostMemory or registered
   /// with registerHostMemory.
-  virtual Error copyD2H(PlatformStreamHandle *S, const void *DeviceSrcHandle,
-                        size_t SrcByteOffset, void *HostDst,
-                        size_t DstByteOffset, size_t ByteCount) {
+  virtual Error copyD2H(const void *PlatformStreamHandle,
+                        const void *DeviceSrcHandle, size_t SrcByteOffset,
+                        void *HostDst, size_t DstByteOffset, size_t ByteCount) {
     return make_error("copyD2H not implemented for platform " + getName());
   }
 
   /// Copies data from the host to the device.
   ///
   /// HostSrc should have been allocated by allocateHostMemory or registered
   /// with registerHostMemory.
-  virtual Error copyH2D(PlatformStreamHandle *S, const void *HostSrc,
+  virtual Error copyH2D(const void *PlatformStreamHandle, const void *HostSrc,
                         size_t SrcByteOffset, const void *DeviceDstHandle,
                         size_t DstByteOffset, size_t ByteCount) {
     return make_error("copyH2D not implemented for platform " + getName());
   }
 
   /// Copies data from one device location to another.
-  virtual Error copyD2D(PlatformStreamHandle *S, const void *DeviceSrcHandle,
-                        size_t SrcByteOffset, const void *DeviceDstHandle,
-                        size_t DstByteOffset, size_t ByteCount) {
+  virtual Error copyD2D(const void *PlatformStreamHandle,
+                        const void *DeviceSrcHandle, size_t SrcByteOffset,
+                        const void *DeviceDstHandle, size_t DstByteOffset,
+                        size_t ByteCount) {
     return make_error("copyD2D not implemented for platform " + getName());
   }
 
   /// Blocks the host until the given stream completes all the work enqueued up
   /// to the point this function is called.
-  virtual Error blockHostUntilDone(PlatformStreamHandle *S) {
+  virtual Error blockHostUntilDone(const void *PlatformStreamHandle) {
     return make_error("blockHostUntilDone not implemented for platform " +
                       getName());
   }
 
   /// Allocates untyped device memory of a given size in bytes.
   virtual Expected<void *> allocateDeviceMemory(size_t ByteCount) {
     return make_error("allocateDeviceMemory not implemented for platform " +
                       getName());

parallel-libs/trunk/streamexecutor/include/streamexecutor/Stream.h

 /// the execution of the stream.
 ///
 /// If any given operation fails when entraining work for the stream, isOK()
 /// will indicate that an error has occurred and getStatus() will get the first
 /// error that occurred on the stream. There is no way to clear the error state
 /// of a stream once it is in an error state.
 class Stream {
 public:
-  explicit Stream(std::unique_ptr<PlatformStreamHandle> PStream);
+  Stream(PlatformDevice *D, const void *PlatformStreamHandle);
 
-  Stream(Stream &&Other) = default;
-  Stream &operator=(Stream &&Other) = default;
+  Stream(const Stream &Other) = delete;
+  Stream &operator=(const Stream &Other) = delete;
+
+  Stream(Stream &&Other);
+  Stream &operator=(Stream &&Other);
 
   ~Stream();
 
   /// Returns whether any error has occurred while entraining work on this
   /// stream.
   bool isOK() const {
     llvm::sys::ScopedReader ReaderLock(*ErrorMessageMutex);
     return !ErrorMessage;
else
       return Error::success();
   }
 
   // Blocks the calling host thread until all work enqueued on this Stream
   // completes.
   //
   // Returns the result of getStatus() after the Stream work completes.
   Error blockHostUntilDone() {
-    setError(PDevice->blockHostUntilDone(ThePlatformStream.get()));
+    setError(PDevice->blockHostUntilDone(PlatformStreamHandle));
     return getStatus();
   }
 
   /// Entrains onto the stream of operations a kernel launch with the given
   /// arguments.
   ///
   /// These arguments can be device memory types like GlobalDeviceMemory<T> and
   /// SharedDeviceMemory<T>, or they can be primitive types such as int. The
   /// allowable argument types are determined by the template parameters to the
   /// Kernel argument.
   template <typename... ParameterTs>
   Stream &thenLaunch(BlockDimensions BlockSize, GridDimensions GridSize,
                      const Kernel<ParameterTs...> &K,
                      const ParameterTs &... Arguments) {
     auto ArgumentArray =
         make_kernel_argument_pack<ParameterTs...>(Arguments...);
-    setError(PDevice->launch(ThePlatformStream.get(), BlockSize, GridSize,
+    setError(PDevice->launch(PlatformStreamHandle, BlockSize, GridSize,
                              K.getPlatformHandle(), ArgumentArray));
     return *this;
   }
 
   /// \name Device memory copying functions
   ///
   /// These methods enqueue a device memory copy operation on the stream and
   /// return without waiting for the operation to complete.
Stream &thenCopyD2H(GlobalDeviceMemorySlice<T> Src,
     if (ElementCount > Src.getElementCount())
       setError("copying too many elements, " + llvm::Twine(ElementCount) +
                ", from a device array of element count " +
                llvm::Twine(Src.getElementCount()));
     else if (ElementCount > Dst.size())
       setError("copying too many elements, " + llvm::Twine(ElementCount) +
                ", to a host array of element count " + llvm::Twine(Dst.size()));
     else
-      setError(PDevice->copyD2H(ThePlatformStream.get(),
+      setError(PDevice->copyD2H(PlatformStreamHandle,
                                 Src.getBaseMemory().getHandle(),
                                 Src.getElementOffset() * sizeof(T), Dst.data(),
                                 0, ElementCount * sizeof(T)));
     return *this;
   }
 
   template <typename T>
   Stream &thenCopyD2H(GlobalDeviceMemorySlice<T> Src,
if (ElementCount > Src.size())
       setError("copying too many elements, " + llvm::Twine(ElementCount) +
                ", from a host array of element count " +
                llvm::Twine(Src.size()));
     else if (ElementCount > Dst.getElementCount())
       setError("copying too many elements, " + llvm::Twine(ElementCount) +
                ", to a device array of element count " +
                llvm::Twine(Dst.getElementCount()));
     else
-      setError(PDevice->copyH2D(ThePlatformStream.get(), Src.data(), 0,
-                                Dst.getBaseMemory().getHandle(),
-                                Dst.getElementOffset() * sizeof(T),
-                                ElementCount * sizeof(T)));
+      setError(PDevice->copyH2D(
+          PlatformStreamHandle, Src.data(), 0, Dst.getBaseMemory().getHandle(),
+          Dst.getElementOffset() * sizeof(T), ElementCount * sizeof(T)));
     return *this;
   }
 
   template <typename T>
   Stream &thenCopyH2D(llvm::ArrayRef<T> Src, GlobalDeviceMemorySlice<T> Dst) {
     if (Src.size() != Dst.getElementCount())
       setError("array size mismatch for H2D, host source has element count " +
                llvm::Twine(Src.size()) +
if (ElementCount > Src.getElementCount())
                ", from a device array of element count " +
                llvm::Twine(Src.getElementCount()));
     else if (ElementCount > Dst.getElementCount())
       setError("copying too many elements, " + llvm::Twine(ElementCount) +
                ", to a device array of element count " +
                llvm::Twine(Dst.getElementCount()));
     else
       setError(PDevice->copyD2D(
-          ThePlatformStream.get(), Src.getBaseMemory().getHandle(),
+          PlatformStreamHandle, Src.getBaseMemory().getHandle(),
           Src.getElementOffset() * sizeof(T), Dst.getBaseMemory().getHandle(),
           Dst.getElementOffset() * sizeof(T), ElementCount * sizeof(T)));
     return *this;
   }
 
   template <typename T>
   Stream &thenCopyD2D(GlobalDeviceMemorySlice<T> Src,
                       GlobalDeviceMemorySlice<T> Dst) {
void setError(llvm::Twine Message) {
     if (!ErrorMessage)
       ErrorMessage = Message.str();
   }
 
   /// The PlatformDevice that supports the operations of this stream.
   PlatformDevice *PDevice;
 
   /// The platform-specific stream handle for this instance.
-  std::unique_ptr<PlatformStreamHandle> ThePlatformStream;
+  const void *PlatformStreamHandle;
 
   /// Mutex that guards the error state flags.
   std::unique_ptr<llvm::sys::RWMutex> ErrorMessageMutex;
 
   /// First error message for an operation in this stream or empty if there have
   /// been no errors.
   llvm::Optional<std::string> ErrorMessage;
-
-  Stream(const Stream &) = delete;
-  void operator=(const Stream &) = delete;
 };
 
 } // namespace streamexecutor
 
 #endif // STREAMEXECUTOR_STREAM_H

parallel-libs/trunk/streamexecutor/lib/Device.cpp

 
 namespace streamexecutor {
 
 Device::Device(PlatformDevice *PDevice) : PDevice(PDevice) {}
 
 Device::~Device() = default;
 
 Expected<Stream> Device::createStream() {
-  Expected<std::unique_ptr<PlatformStreamHandle>> MaybePlatformStream =
-      PDevice->createStream();
+  Expected<const void *> MaybePlatformStream = PDevice->createStream();
   if (!MaybePlatformStream) {
     return MaybePlatformStream.takeError();
   }
-  assert((*MaybePlatformStream)->getDevice() == PDevice &&
-         "an executor created a stream with a different stored executor");
-  return Stream(std::move(*MaybePlatformStream));
+  return Stream(PDevice, *MaybePlatformStream);
 }
 
 } // namespace streamexecutor

parallel-libs/trunk/streamexecutor/lib/Kernel.cpp

 //===-- Kernel.cpp - General kernel implementation ------------------------===//
 //
 //                     The LLVM Compiler Infrastructure
 //
 // This file is distributed under the University of Illinois Open Source
 // License. See LICENSE.TXT for details.
 //
 //===----------------------------------------------------------------------===//
 ///
 /// \file
 /// This file contains the implementation details for kernel types.
 ///
 //===----------------------------------------------------------------------===//
 
-#include "streamexecutor/Kernel.h"
+#include <cassert>
+
 #include "streamexecutor/Device.h"
+#include "streamexecutor/Kernel.h"
 #include "streamexecutor/PlatformInterfaces.h"
 
 #include "llvm/DebugInfo/Symbolize/Symbolize.h"
 
 namespace streamexecutor {
 
-KernelBase::KernelBase(llvm::StringRef Name)
-    : Name(Name), DemangledName(llvm::symbolize::LLVMSymbolizer::DemangleName(
-                      Name, nullptr)) {}
+KernelBase::KernelBase(PlatformDevice *D, const void *PlatformKernelHandle,
+                       llvm::StringRef Name)
+    : PDevice(D), PlatformKernelHandle(PlatformKernelHandle), Name(Name),
+      DemangledName(
+          llvm::symbolize::LLVMSymbolizer::DemangleName(Name, nullptr)) {
+  assert(D != nullptr &&
+         "cannot construct a kernel object with a null platform device");
+  assert(PlatformKernelHandle != nullptr &&
+         "cannot construct a kernel object with a null platform kernel handle");
+}
+
+KernelBase::KernelBase(KernelBase &&Other)
+    : PDevice(Other.PDevice), PlatformKernelHandle(Other.PlatformKernelHandle),
+      Name(std::move(Other.Name)),
+      DemangledName(std::move(Other.DemangledName)) {
+  Other.PDevice = nullptr;
+  Other.PlatformKernelHandle = nullptr;
+}
+
+KernelBase &KernelBase::operator=(KernelBase &&Other) {
+  PDevice = Other.PDevice;
+  PlatformKernelHandle = Other.PlatformKernelHandle;
+  Name = std::move(Other.Name);
+  DemangledName = std::move(Other.DemangledName);
+  Other.PDevice = nullptr;
+  Other.PlatformKernelHandle = nullptr;
+  return *this;
+}
+
+KernelBase::~KernelBase() {
+  if (PlatformKernelHandle)
+    // TODO(jhen): Handle the error here.
+    consumeError(PDevice->destroyKernel(PlatformKernelHandle));
+}
 
 } // namespace streamexecutor

parallel-libs/trunk/streamexecutor/lib/PlatformInterfaces.cpp

 /// Implementation file for PlatformInterfaces.h.
 ///
 //===----------------------------------------------------------------------===//
 
 #include "streamexecutor/PlatformInterfaces.h"
 
 namespace streamexecutor {
 
-PlatformStreamHandle::~PlatformStreamHandle() = default;
-
 PlatformDevice::~PlatformDevice() = default;
 
 } // namespace streamexecutor

parallel-libs/trunk/streamexecutor/lib/Stream.cpp

 //===-- Stream.cpp - General stream implementation ------------------------===//
 //
 //                     The LLVM Compiler Infrastructure
 //
 // This file is distributed under the University of Illinois Open Source
 // License. See LICENSE.TXT for details.
 //
 //===----------------------------------------------------------------------===//
 ///
 /// \file
 /// This file contains the implementation details for a general stream object.
 ///
 //===----------------------------------------------------------------------===//
 
+#include <cassert>
+
 #include "streamexecutor/Stream.h"
 
 namespace streamexecutor {
 
-Stream::Stream(std::unique_ptr<PlatformStreamHandle> PStream)
-    : PDevice(PStream->getDevice()), ThePlatformStream(std::move(PStream)),
-      ErrorMessageMutex(llvm::make_unique<llvm::sys::RWMutex>()) {}
+Stream::Stream(PlatformDevice *D, const void *PlatformStreamHandle)
+    : PDevice(D), PlatformStreamHandle(PlatformStreamHandle),
+      ErrorMessageMutex(llvm::make_unique<llvm::sys::RWMutex>()) {
+  assert(D != nullptr &&
+         "cannot construct a stream object with a null platform device");
+  assert(PlatformStreamHandle != nullptr &&
+         "cannot construct a stream object with a null platform stream handle");
+}
+
+Stream::Stream(Stream &&Other)
+    : PDevice(Other.PDevice), PlatformStreamHandle(Other.PlatformStreamHandle),
+      ErrorMessageMutex(std::move(Other.ErrorMessageMutex)),
+      ErrorMessage(std::move(Other.ErrorMessage)) {
+  Other.PDevice = nullptr;
+  Other.PlatformStreamHandle = nullptr;
+}
+
+Stream &Stream::operator=(Stream &&Other) {
+  PDevice = Other.PDevice;
+  PlatformStreamHandle = Other.PlatformStreamHandle;
+  ErrorMessageMutex = std::move(Other.ErrorMessageMutex);
+  ErrorMessage = std::move(Other.ErrorMessage);
+  Other.PDevice = nullptr;
+  Other.PlatformStreamHandle = nullptr;
+  return *this;
+}
 
-Stream::~Stream() = default;
+Stream::~Stream() {
+  if (PlatformStreamHandle)
+    // TODO(jhen): Handle error condition here.
+    consumeError(PDevice->destroyStream(PlatformStreamHandle));
+}
 
 } // namespace streamexecutor

parallel-libs/trunk/streamexecutor/lib/unittests/SimpleHostPlatformDevice.h

 /// host platform.
 ///
 /// The allocate and copy methods are simple wrappers for std::malloc and
 /// std::memcpy.
 class SimpleHostPlatformDevice : public streamexecutor::PlatformDevice {
 public:
   std::string getName() const override { return "SimpleHostPlatformDevice"; }
 
-  streamexecutor::Expected<
-      std::unique_ptr<streamexecutor::PlatformStreamHandle>>
-  createStream() override {
+  streamexecutor::Expected<const void *> createStream() override {
     return nullptr;
   }
 
   streamexecutor::Expected<void *>
   allocateDeviceMemory(size_t ByteCount) override {
     return std::malloc(ByteCount);
   }
 
streamexecutor::Error registerHostMemory(void *Memory,
                                            size_t ByteCount) override {
     return streamexecutor::Error::success();
   }
 
   streamexecutor::Error unregisterHostMemory(void *Memory) override {
     return streamexecutor::Error::success();
   }
 
-  streamexecutor::Error copyD2H(streamexecutor::PlatformStreamHandle *S,
+  streamexecutor::Error copyD2H(const void *StreamHandle,
                                 const void *DeviceHandleSrc,
                                 size_t SrcByteOffset, void *HostDst,
                                 size_t DstByteOffset,
                                 size_t ByteCount) override {
     std::memcpy(static_cast<char *>(HostDst) + DstByteOffset,
                 static_cast<const char *>(DeviceHandleSrc) + SrcByteOffset,
                 ByteCount);
     return streamexecutor::Error::success();
   }
 
-  streamexecutor::Error copyH2D(streamexecutor::PlatformStreamHandle *S,
-                                const void *HostSrc, size_t SrcByteOffset,
+  streamexecutor::Error copyH2D(const void *StreamHandle, const void *HostSrc,
+                                size_t SrcByteOffset,
                                 const void *DeviceHandleDst,
                                 size_t DstByteOffset,
                                 size_t ByteCount) override {
     std::memcpy(static_cast<char *>(const_cast<void *>(DeviceHandleDst)) +
                     DstByteOffset,
                 static_cast<const char *>(HostSrc) + SrcByteOffset, ByteCount);
     return streamexecutor::Error::success();
   }
 
   streamexecutor::Error
-  copyD2D(streamexecutor::PlatformStreamHandle *S, const void *DeviceHandleSrc,
+  copyD2D(const void *StreamHandle, const void *DeviceHandleSrc,
           size_t SrcByteOffset, const void *DeviceHandleDst,
           size_t DstByteOffset, size_t ByteCount) override {
     std::memcpy(static_cast<char *>(const_cast<void *>(DeviceHandleDst)) +
                     DstByteOffset,
                 static_cast<const char *>(DeviceHandleSrc) + SrcByteOffset,
                 ByteCount);
     return streamexecutor::Error::success();
   }

parallel-libs/trunk/streamexecutor/lib/unittests/StreamTest.cpp

 
 const auto &getDeviceValue =
     se::test::SimpleHostPlatformDevice::getDeviceValue<int>;
 
 /// Test fixture to hold objects used by tests.
 class StreamTest : public ::testing::Test {
 public:
   StreamTest()
-      : Device(&PDevice),
-        Stream(llvm::make_unique<se::PlatformStreamHandle>(&PDevice)),
-        HostA5{0, 1, 2, 3, 4}, HostB5{5, 6, 7, 8, 9},
-        HostA7{10, 11, 12, 13, 14, 15, 16}, HostB7{17, 18, 19, 20, 21, 22, 23},
-        Host5{24, 25, 26, 27, 28}, Host7{29, 30, 31, 32, 33, 34, 35},
+      : DummyPlatformStream(1), Device(&PDevice),
+        Stream(&PDevice, &DummyPlatformStream), HostA5{0, 1, 2, 3, 4},
+        HostB5{5, 6, 7, 8, 9}, HostA7{10, 11, 12, 13, 14, 15, 16},
+        HostB7{17, 18, 19, 20, 21, 22, 23}, Host5{24, 25, 26, 27, 28},
+        Host7{29, 30, 31, 32, 33, 34, 35},
         DeviceA5(getOrDie(Device.allocateDeviceMemory<int>(5))),
         DeviceB5(getOrDie(Device.allocateDeviceMemory<int>(5))),
         DeviceA7(getOrDie(Device.allocateDeviceMemory<int>(7))),
         DeviceB7(getOrDie(Device.allocateDeviceMemory<int>(7))) {
     se::dieIfError(Device.synchronousCopyH2D<int>(HostA5, DeviceA5));
     se::dieIfError(Device.synchronousCopyH2D<int>(HostB5, DeviceB5));
     se::dieIfError(Device.synchronousCopyH2D<int>(HostA7, DeviceA7));
     se::dieIfError(Device.synchronousCopyH2D<int>(HostB7, DeviceB7));
   }
 
 protected:
+  int DummyPlatformStream; // Mimicking a platform where the platform stream
+                           // handle is just a stream number.
   se::test::SimpleHostPlatformDevice PDevice;
   se::Device Device;
   se::Stream Stream;
 
   // Device memory is matched by host arrays.
   int HostA5[5];
   int HostB5[5];
   int HostA7[7];