react-native-executorch 0.5.6 → 0.5.8

package/third-party/include/executorch/runtime/core/named_data_map.h

@@ -22,8 +22,7 @@
 #include <executorch/runtime/platform/compiler.h>
 
 namespace executorch {
-namespace runtime {
-
+namespace ET_RUNTIME_NAMESPACE {
 /**
  * Interface to access and retrieve data via name.
  * See executorch/extension/flat_tensor/ for an example.
@@ -32,33 +31,34 @@ class ET_EXPERIMENTAL NamedDataMap {
 public:
   virtual ~NamedDataMap() = default;
   /**
-   * Get metadata by key.
+   * Get tensor_layout by key.
    *
    * @param key The name of the tensor.
-   * @return Result containing TensorLayout with tensor metadata.
+   * @return Result containing TensorLayout.
    */
-  ET_NODISCARD virtual Result<const executorch::runtime::TensorLayout>
-  get_metadata(const char *key) const = 0;
+  ET_NODISCARD virtual Result<const TensorLayout>
+  get_tensor_layout(executorch::aten::string_view key) const = 0;
   /**
    * Get data by key.
    *
    * @param key Name of the data.
-   * @return Result containing a FreeableBuffer with the tensor data.
+   * @return Result containing a FreeableBuffer.
    */
   ET_NODISCARD virtual Result<FreeableBuffer>
-  get_data(const char *key) const = 0;
+  get_data(executorch::aten::string_view key) const = 0;
 
   /**
    * Loads data corresponding to the key into the provided buffer.
    *
    * @param key The name of the data.
-   * @param size The number of bytes to load. Use `get_metadata` to retrieve the
-   * size of the data for a given key.
+   * @param size The number of bytes to load. Use `get_tensor_layout` to
+   * retrieve the size of the data for a given key.
    * @param buffer The buffer to load the data into. Must point to at least
    * `size` bytes of memory.
    * @returns an Error indicating if the load was successful.
    */
-  ET_NODISCARD virtual Error load_data_into(const char *key, void *buffer,
+  ET_NODISCARD virtual Error load_data_into(executorch::aten::string_view key,
+                                            void *buffer,
                                             size_t size) const = 0;
 
   /**
@@ -66,7 +66,7 @@ public:
    *
    * @return Result containing the number of keys.
    */
-  ET_NODISCARD virtual Result<size_t> get_num_keys() const = 0;
+  ET_NODISCARD virtual Result<uint32_t> get_num_keys() const = 0;
 
   /**
    * Get the key at the given index.
@@ -75,10 +75,10 @@ public:
    * @return Result containing the key at the given index. Note: the returned
    * pointer is only valid for the lifetime of the DataMap.
    */
-  ET_NODISCARD virtual Result<const char *> get_key(size_t index) const = 0;
+  ET_NODISCARD virtual Result<const char *> get_key(uint32_t index) const = 0;
 };
 
-} // namespace runtime
+} // namespace ET_RUNTIME_NAMESPACE
 } // namespace executorch
 
 #ifdef __GNUC__

package/third-party/include/executorch/runtime/core/portable_type/c10/c10/macros/Export.h

@@ -2,95 +2,11 @@
 #ifndef C10_MACROS_EXPORT_H_
 #define C10_MACROS_EXPORT_H_
 
-/* Header file to define the common scaffolding for exported symbols.
- *
- * Export is by itself a quite tricky situation to deal with, and if you are
- * hitting this file, make sure you start with the background here:
- * - Linux: https://gcc.gnu.org/wiki/Visibility
- * - Windows:
- * https://docs.microsoft.com/en-us/cpp/cpp/dllexport-dllimport?view=vs-2017
- *
- * Do NOT include this file directly. Instead, use c10/macros/Macros.h
- */
-
-// You do not need to edit this part of file unless you are changing the core
-// pytorch export abstractions.
-//
-// This part defines the C10 core export and import macros. This is controlled
-// by whether we are building shared libraries or not, which is determined
-// during build time and codified in c10/core/cmake_macros.h.
-// When the library is built as a shared lib, EXPORT and IMPORT will contain
-// visibility attributes. If it is being built as a static lib, then EXPORT
-// and IMPORT basically have no effect.
-
-// As a rule of thumb, you should almost NEVER mix static and shared builds for
-// libraries that depend on c10. AKA, if c10 is built as a static library, we
-// recommend everything dependent on c10 to be built statically. If c10 is built
-// as a shared library, everything dependent on it should be built as shared. In
-// the PyTorch project, all native libraries shall use the macro
-// C10_BUILD_SHARED_LIB to check whether pytorch is building shared or static
-// libraries.
-
-// For build systems that do not directly depend on CMake and directly build
-// from the source directory (such as Buck), one may not have a cmake_macros.h
-// file at all. In this case, the build system is responsible for providing
-// correct macro definitions corresponding to the cmake_macros.h.in file.
-//
-// In such scenarios, one should define the macro
-//     C10_USING_CUSTOM_GENERATED_MACROS
-// to inform this header that it does not need to include the cmake_macros.h
-// file.
-
 #ifndef C10_USING_CUSTOM_GENERATED_MACROS
 #include <c10/macros/cmake_macros.h>
 #endif // C10_USING_CUSTOM_GENERATED_MACROS
 
-#ifdef _WIN32
-#define C10_HIDDEN
-#if defined(C10_BUILD_SHARED_LIBS)
-#define C10_EXPORT __declspec(dllexport)
-#define C10_IMPORT __declspec(dllimport)
-#else
-#define C10_EXPORT
-#define C10_IMPORT
-#endif
-#else // _WIN32
-#if defined(__GNUC__)
-#define C10_EXPORT __attribute__((__visibility__("default")))
-#define C10_HIDDEN __attribute__((__visibility__("hidden")))
-#else // defined(__GNUC__)
-#define C10_EXPORT
-#define C10_HIDDEN
-#endif // defined(__GNUC__)
-#define C10_IMPORT C10_EXPORT
-#endif // _WIN32
-
-#ifdef NO_EXPORT
-#undef C10_EXPORT
-#define C10_EXPORT
-#endif
-
-// Definition of an adaptive XX_API macro, that depends on whether you are
-// building the library itself or not, routes to XX_EXPORT and XX_IMPORT.
-// Basically, you will need to do this for each shared library that you are
-// building, and the instruction is as follows: assuming that you are building
-// a library called libawesome.so. You should:
-// (1) for your cmake target (usually done by "add_library(awesome, ...)"),
-//     define a macro called AWESOME_BUILD_MAIN_LIB using
-//     target_compile_options.
-// (2) define the AWESOME_API macro similar to the one below.
-// And in the source file of your awesome library, use AWESOME_API to
-// annotate public symbols.
-
-// Here, for the C10 library, we will define the macro C10_API for both import
-// and export.
-
-// This one is being used by libc10.so
-#ifdef C10_BUILD_MAIN_LIB
-#define C10_API C10_EXPORT
-#else
-#define C10_API C10_IMPORT
-#endif
+#include <torch/headeronly/macros/Export.h>
 
 // This one is being used by libtorch.so
 #ifdef CAFFE2_BUILD_MAIN_LIB
@@ -160,4 +76,4 @@
 #define C10_API_ENUM
 #endif
 
-#endif // C10_MACROS_MACROS_H_
+#endif // C10_MACROS_EXPORT_H_

package/third-party/include/executorch/runtime/core/portable_type/c10/c10/macros/Macros.h

@@ -242,7 +242,7 @@ using namespace c10::xpu;
 #ifdef __HIPCC__
 // Unlike CUDA, HIP requires a HIP header to be included for __host__ to work.
 // We do this #include here so that C10_HOST_DEVICE and friends will Just Work.
-// See https://github.com/ROCm-Developer-Tools/HIP/issues/441
+// See https://github.com/ROCm/hip/issues/441
 #include <hip/hip_runtime.h>
 #endif
 
@@ -390,11 +390,24 @@ __host__ __device__
 #endif // __SYCL_DEVICE_ONLY__
 }
 #endif // NDEBUG
-// ROCm disable kernel assert by default
+// ROCm disables kernel assert by default for performance considerations.
+// Though ROCm supports __assert_fail, it uses kernel printf which has
+// a non-negligible performance impact even if the assert condition is
+// never triggered. We choose to use abort() instead which will still
+// terminate the application but without a more useful error message.
 #if !defined(C10_USE_ROCM_KERNEL_ASSERT) and defined(USE_ROCM)
-#define CUDA_KERNEL_ASSERT(cond)
-#define CUDA_KERNEL_ASSERT_MSG(cond, msg)
-#define SYCL_KERNEL_ASSERT(cond)
+#define CUDA_KERNEL_ASSERT(cond)                                               \
+  if C10_UNLIKELY (!(cond)) {                                                  \
+    abort();                                                                   \
+  }
+#define CUDA_KERNEL_ASSERT_MSG(cond, msg)                                      \
+  if C10_UNLIKELY (!(cond)) {                                                  \
+    abort();                                                                   \
+  }
+#define SYCL_KERNEL_ASSERT(cond)                                               \
+  if C10_UNLIKELY (!(cond)) {                                                  \
+    abort();                                                                   \
+  }
 #else
 #define CUDA_KERNEL_ASSERT(cond)                                               \
   if (C10_UNLIKELY(!(cond))) {                                                 \
@@ -494,4 +507,14 @@ __host__ __device__
 
 #endif
 
+// This macro is used to find older C++ compilers
+// that don't support move optimization for return values.
+
+#if (defined(__GNUC__) && __GNUC__ < 13) ||                                    \
+    (defined(__clang_major__) && __clang_major__ < 13)
+#define C10_RETURN_MOVE_IF_OLD_COMPILER 1
+#else
+#define C10_RETURN_MOVE_IF_OLD_COMPILER 0
+#endif
+
 #endif // C10_MACROS_MACROS_H_

package/third-party/include/executorch/runtime/core/portable_type/c10/c10/util/BFloat16-inl.h

@@ -10,14 +10,11 @@ C10_CLANG_DIAGNOSTIC_PUSH()
 C10_CLANG_DIAGNOSTIC_IGNORE("-Wimplicit-int-float-conversion")
 #endif
 
-#if defined(SYCL_EXT_ONEAPI_BFLOAT16_MATH_FUNCTIONS)
 #if defined(CL_SYCL_LANGUAGE_VERSION)
 #include <CL/sycl.hpp> // for SYCL 1.2.1
-#else
+#elif defined(SYCL_LANGUAGE_VERSION)
 #include <sycl/sycl.hpp> // for SYCL 2020
 #endif
-#include <ext/oneapi/bfloat16.hpp>
-#endif
 
 namespace c10 {
 

package/third-party/include/executorch/runtime/core/portable_type/c10/c10/util/BFloat16.h

@@ -14,14 +14,11 @@
 #include <cuda_bf16.h>
 #endif
 
-#if defined(SYCL_EXT_ONEAPI_BFLOAT16_MATH_FUNCTIONS)
 #if defined(CL_SYCL_LANGUAGE_VERSION)
 #include <CL/sycl.hpp> // for SYCL 1.2.1
-#else
+#elif defined(SYCL_LANGUAGE_VERSION)
 #include <sycl/sycl.hpp> // for SYCL 2020
 #endif
-#include <ext/oneapi/bfloat16.hpp>
-#endif
 
 namespace c10 {
 
@@ -31,7 +28,7 @@ inline C10_HOST_DEVICE float f32_from_bits(uint16_t src) {
   uint32_t tmp = src;
   tmp <<= 16;
 
-#if defined(USE_ROCM)
+#if defined(USE_ROCM) && defined(__HIPCC__)
   float *tempRes;
 
   // We should be using memcpy in order to respect the strict aliasing rule
@@ -48,7 +45,7 @@ inline C10_HOST_DEVICE float f32_from_bits(uint16_t src) {
 inline C10_HOST_DEVICE uint16_t bits_from_f32(float src) {
   uint32_t res = 0;
 
-#if defined(USE_ROCM)
+#if defined(USE_ROCM) && defined(__HIPCC__)
   // We should be using memcpy in order to respect the strict aliasing rule
   // but it fails in the HIP environment.
   uint32_t *tempRes = reinterpret_cast<uint32_t *>(&src);
@@ -61,7 +58,7 @@ inline C10_HOST_DEVICE uint16_t bits_from_f32(float src) {
 }
 
 inline C10_HOST_DEVICE uint16_t round_to_nearest_even(float src) {
-#if defined(USE_ROCM)
+#if defined(USE_ROCM) && defined(__HIPCC__)
   if (src != src) {
 #elif defined(_MSC_VER)
   if (isnan(src)) {
@@ -87,7 +84,7 @@ struct alignas(2) BFloat16 {
   uint16_t x;
 
   // HIP wants __host__ __device__ tag, CUDA does not
-#if defined(USE_ROCM)
+#if defined(USE_ROCM) && defined(__HIPCC__)
   C10_HOST_DEVICE BFloat16() = default;
 #else
   BFloat16() = default;

package/third-party/include/executorch/runtime/core/portable_type/c10/c10/util/Half.h

@@ -242,7 +242,12 @@ C10_HOST_DEVICE inline float fp16_ieee_to_fp32_value(uint16_t h) {
   // const float exp_scale = 0x1.0p-112f;
   constexpr uint32_t scale_bits = (uint32_t)15 << 23;
   float exp_scale_val = 0;
+#if defined(_MSC_VER) && defined(__clang__)
+  __builtin_memcpy(&exp_scale_val, &scale_bits, sizeof(exp_scale_val));
+#else
   std::memcpy(&exp_scale_val, &scale_bits, sizeof(exp_scale_val));
+#endif
+
   const float exp_scale = exp_scale_val;
   const float normalized_value =
       fp32_from_bits((two_w >> 4) + exp_offset) * exp_scale;

package/third-party/include/executorch/runtime/core/portable_type/c10/c10/util/bit_cast.h

@@ -3,7 +3,7 @@
 #include <cstring>
 #include <type_traits>
 
-#if __has_include(<bit>) && (__cplusplus >= 202002L || (defined(__cpp_lib_bit_cast) && __cpp_lib_bit_cast >= 201806L))
+#if __has_include(<bit>) && (defined(__cpp_lib_bit_cast) && __cpp_lib_bit_cast >= 201806L)
 #include <bit>
 #define C10_HAVE_STD_BIT_CAST 1
 #else