Back out "[pytorch][PR] Move thnvrtc and DynamicLibrary to ATen" (pyt…

…orch#22749)

Summary:
Original commit changeset: add2ee8a8865

Pull Request resolved: pytorch#22749
ghstack-source-id: 86323899

Differential Revision: D16203552

fbshipit-source-id: 227df3b85316315c15d2cb7b6a5c884096a82e9e

aten/CMakeLists.txt

@@ -24,7 +24,6 @@ set(ATen_THIRD_PARTY_INCLUDE)
 set(ATen_CUDA_SRCS)
 set(ATen_CUDA_TEST_SRCS)
 set(ATen_CUDA_INCLUDE)
-set(ATen_NVRTC_STUB_SRCS)
 set(ATen_HIP_SRCS)
 set(ATen_HIP_TEST_SRCS)
 set(ATen_HIP_INCLUDE)
@@ -102,7 +101,6 @@ add_subdirectory(src/ATen)
 set(ATen_CPU_SRCS ${ATen_CPU_SRCS} PARENT_SCOPE)
 set(ATen_CUDA_SRCS ${ATen_CUDA_SRCS} PARENT_SCOPE)
 set(ATen_HIP_SRCS ${ATen_HIP_SRCS} PARENT_SCOPE)
-set(ATen_NVRTC_STUB_SRCS ${ATen_NVRTC_STUB_SRCS} PARENT_SCOPE)
 set(ATen_CPU_TEST_SRCS ${ATen_CPU_TEST_SRCS} PARENT_SCOPE)
 set(ATen_CUDA_TEST_SRCS ${ATen_CUDA_TEST_SRCS} PARENT_SCOPE)
 set(ATen_HIP_TEST_SRCS ${ATen_HIP_TEST_SRCS} PARENT_SCOPE)

aten/src/ATen/ATen.h

@@ -7,7 +7,6 @@
 #include <ATen/DeviceGuard.h>
 #include <ATen/DimVector.h>
 #include <ATen/Dispatch.h>
-#include <ATen/DynamicLibrary.h>
 #include <ATen/Formatting.h>
 #include <ATen/Functions.h>
 #ifdef BUILD_NAMEDTENSOR

aten/src/ATen/CMakeLists.txt

@@ -39,17 +39,13 @@ FILE(GLOB base_cpp "*.cpp" "detail/*.cpp" "cpu/*.cpp")
 add_subdirectory(core)
 FILE(GLOB cuda_h "cuda/*.h" "cuda/detail/*.h" "cuda/*.cuh" "cuda/detail/*.cuh")
 FILE(GLOB cuda_cpp "cuda/*.cpp" "cuda/detail/*.cpp")
-FILE(GLOB cuda_nvrtc_stub_h "cuda/nvrtc_stub/*.h")
-FILE(GLOB cuda_nvrtc_stub_cpp "cuda/nvrtc_stub/*.cpp")
 FILE(GLOB cuda_cu "cuda/*.cu" "cuda/detail/*.cu")
 FILE(GLOB cudnn_h "cudnn/*.h" "cudnn/*.cuh")
 FILE(GLOB cudnn_cpp "cudnn/*.cpp")
 
 FILE(GLOB hip_h "hip/*.h" "hip/detail/*.h" "hip/*.cuh" "hip/detail/*.cuh")
 FILE(GLOB hip_cpp "hip/*.cpp" "hip/detail/*.cpp" "hip/impl/*.cpp")
 FILE(GLOB hip_hip "hip/*.hip" "hip/detail/*.hip" "hip/impl/*.hip")
-FILE(GLOB hip_nvrtc_stub_h "hip/nvrtc_stub/*.h")
-FILE(GLOB hip_nvrtc_stub_cpp "hip/nvrtc_stub/*.cpp")
 FILE(GLOB miopen_h "miopen/*.h")
 FILE(GLOB miopen_cpp "miopen/*.cpp")
 
@@ -360,7 +356,6 @@ endif()
 
 if(USE_CUDA)
   set(ATen_CUDA_SRCS ${all_cuda_cpp})
-  set(ATen_NVRTC_STUB_SRCS ${cuda_nvrtc_stub_cpp})
   if(AT_LINK_STYLE STREQUAL "INTERFACE")
     # Source code can't be added to an interface library, so it is
     # passed back to be compiled into the containing library
@@ -373,9 +368,6 @@ endif()
 
 if(USE_ROCM)
   set(ATen_HIP_SRCS ${all_hip_cpp})
-  # caffe2_nvrtc's stubs to driver APIs are useful for HIP.
-  # See NOTE [ ATen NVRTC Stub and HIP ]
-  set(ATen_NVRTC_STUB_SRCS ${hip_nvrtc_stub_cpp})
   if(AT_LINK_STYLE STREQUAL "INTERFACE")
     # Source code can't be added to an interface library, so it is
     # passed back to be compiled into the containing library
@@ -447,7 +439,6 @@ endif()
 set(ATen_CORE_SRCS ${ATen_CORE_SRCS} PARENT_SCOPE)
 set(ATen_CPU_SRCS ${ATen_CPU_SRCS} PARENT_SCOPE)
 set(ATen_CUDA_SRCS ${ATen_CUDA_SRCS} PARENT_SCOPE)
-set(ATen_NVRTC_STUB_SRCS ${ATen_NVRTC_STUB_SRCS} PARENT_SCOPE)
 set(ATen_HIP_SRCS ${ATen_HIP_SRCS} PARENT_SCOPE)
 set(ATen_QUANTIZED_SRCS ${ATen_QUANTIZED_SRCS} PARENT_SCOPE)
 set(ATen_CPU_TEST_SRCS ${ATen_CPU_TEST_SRCS} PARENT_SCOPE)

aten/src/ATen/Context.h

@@ -77,9 +77,7 @@ class CAFFE2_API Context {
     });
     return thh_state.get();
   }
-  const at::cuda::NVRTC& getNVRTC() {
-    return detail::getCUDAHooks().nvrtc();
-  }
+
   THCState* getTHCState() {
     // AT_ASSERT(thc_state);
     return thc_state.get();

aten/src/ATen/Utils.h

@@ -24,10 +24,6 @@
 #define __ubsan_ignore_vptr__
 #endif
 
-#define AT_DISALLOW_COPY_AND_ASSIGN(TypeName) \
-  TypeName(const TypeName&) = delete; \
-  void operator=(const TypeName&) = delete
-
 namespace at {
 
 CAFFE2_API int _crash_if_asan(int);

aten/src/ATen/cuda/CUDAContext.h

@@ -3,8 +3,8 @@
 #include <ATen/core/ATenGeneral.h>
 #include <ATen/Context.h>
 #include <c10/cuda/CUDAStream.h>
-#include <c10/cuda/CUDAFunctions.h>
 #include <ATen/cuda/Exceptions.h>
+#include <c10/cuda/CUDAFunctions.h>
 
 #include <cstdint>
 

aten/src/ATen/cuda/Exceptions.h

@@ -1,6 +1,5 @@
 #pragma once
 
-#include <ATen/Context.h>
 #include <c10/util/Exception.h>
 #include <c10/cuda/CUDAException.h>
 
@@ -21,57 +20,3 @@
   } while (0)
 
 #define AT_CUDA_CHECK(EXPR) C10_CUDA_CHECK(EXPR)
-
-// For CUDA Driver API
-//
-// This is here instead of in c10 because NVRTC is loaded dynamically via a stub
-// in ATen, and we need to use its nvrtcGetErrorString.
-// See NOTE [ USE OF NVRTC AND DRIVER API ].
-#ifndef __HIP_PLATFORM_HCC__
-
-#define AT_CUDA_DRIVER_CHECK(EXPR)                                                                               \
-  do {                                                                                                           \
-    CUresult __err = EXPR;                                                                                       \
-    if (__err != CUDA_SUCCESS) {                                                                                 \
-      const char* err_str;                                                                                       \
-      CUresult get_error_str_err C10_UNUSED = at::globalContext().getNVRTC().cuGetErrorString(__err, &err_str);  \
-      if (get_error_str_err != CUDA_SUCCESS) {                                                                   \
-        AT_ERROR("CUDA driver error: unknown error");                                                            \
-      } else {                                                                                                   \
-        AT_ERROR("CUDA driver error: ", err_str);                                                                \
-      }                                                                                                          \
-    }                                                                                                            \
-  } while (0)
-
-#else
-
-#define AT_CUDA_DRIVER_CHECK(EXPR)                                                \
-  do {                                                                            \
-    CUresult __err = EXPR;                                                        \
-    if (__err != CUDA_SUCCESS) {                                                  \
-      AT_ERROR("CUDA driver error: ", static_cast<int>(__err));                   \
-    }                                                                             \
-  } while (0)
-
-#endif
-
-// For CUDA NVRTC
-//
-// Note: As of CUDA 10, nvrtc error code 7, NVRTC_ERROR_BUILTIN_OPERATION_FAILURE,
-// incorrectly produces the error string "NVRTC unknown error."
-// The following maps it correctly.
-//
-// This is here instead of in c10 because NVRTC is loaded dynamically via a stub
-// in ATen, and we need to use its nvrtcGetErrorString.
-// See NOTE [ USE OF NVRTC AND DRIVER API ].
-#define AT_CUDA_NVRTC_CHECK(EXPR)                                                                   \
-  do {                                                                                              \
-    nvrtcResult __err = EXPR;                                                                       \
-    if (__err != NVRTC_SUCCESS) {                                                                   \
-      if (static_cast<int>(__err) != 7) {                                                           \
-        AT_ERROR("CUDA NVRTC error: ", at::globalContext().getNVRTC().nvrtcGetErrorString(__err));  \
-      } else {                                                                                      \
-        AT_ERROR("CUDA NVRTC error: NVRTC_ERROR_BUILTIN_OPERATION_FAILURE");                        \
-      }                                                                                             \
-    }                                                                                               \
-  } while (0)

aten/src/ATen/cuda/detail/CUDAHooks.cpp

@@ -2,11 +2,9 @@
 
 #include <ATen/CUDAGenerator.h>
 #include <ATen/Context.h>
-#include <ATen/DynamicLibrary.h>
 #include <ATen/cuda/CUDAConfig.h>
 #include <ATen/cuda/CUDADevice.h>
 #include <ATen/cuda/PinnedMemoryAllocator.h>
-#include <ATen/cuda/nvrtc_stub/ATenNVRTC.h>
 #include <ATen/detail/CUDAHooksInterface.h>
 #include <ATen/native/cuda/CuFFTPlanCache.h>
 #include <c10/util/Exception.h>
@@ -79,32 +77,6 @@ bool CUDAHooks::hasCuDNN() const {
   return AT_CUDNN_ENABLED();
 }
 
-#ifdef USE_DIRECT_NVRTC
-static std::pair<std::unique_ptr<at::DynamicLibrary>, at::cuda::NVRTC*> load_nvrtc() {
-  return std::make_pair(nullptr, at::cuda::load_nvrtc());
-}
-#else
-static std::pair<std::unique_ptr<at::DynamicLibrary>, at::cuda::NVRTC*> load_nvrtc() {
-#if defined(_WIN32)
-  std::string libcaffe2_nvrtc = "caffe2_nvrtc.dll";
-#elif defined(__APPLE__)
-  std::string libcaffe2_nvrtc = "libcaffe2_nvrtc.dylib";
-#else
-  std::string libcaffe2_nvrtc = "libcaffe2_nvrtc.so";
-#endif
-  std::unique_ptr<at::DynamicLibrary> libnvrtc_stub(
-      new at::DynamicLibrary(libcaffe2_nvrtc.c_str()));
-  auto fn = (at::cuda::NVRTC * (*)()) libnvrtc_stub->sym("load_nvrtc");
-  return std::make_pair(std::move(libnvrtc_stub), fn());
-}
-#endif
-
-const at::cuda::NVRTC& CUDAHooks::nvrtc() const {
-  // must hold onto DynamicLibrary otherwise it will unload
-  static auto handle = load_nvrtc();
-  return *handle.second;
-}
-
 int64_t CUDAHooks::current_device() const {
   int device;
   cudaError_t err = cudaGetDevice(&device);

aten/src/ATen/cuda/detail/CUDAHooks.h

@@ -16,7 +16,6 @@ struct CUDAHooks : public at::CUDAHooksInterface {
   bool hasCUDA() const override;
   bool hasMAGMA() const override;
   bool hasCuDNN() const override;
-  const at::cuda::NVRTC& nvrtc() const override;
   int64_t current_device() const override;
   Allocator* getPinnedMemoryAllocator() const override;
   bool compiledWithCuDNN() const override;