return StatusFromTF_Status(plugin_status.get());
}

Status RegisterFilesystemPlugin(const std::string& dso_path) {
  // Step 1: Load plugin
  Env* env = Env::Default();
  void* dso_handle;
  TF_RETURN_IF_ERROR(env->LoadDynamicLibrary(dso_path.c_str(), &dso_handle));

  // Step 2: Load symbol for `TF_InitPlugin`
  void* dso_symbol;
  TF_RETURN_WITH_CONTEXT_IF_ERROR(

// only once. Hence, this function keeping another static pointer to a vector
// which contains only the schemes under test.
//
// Without this additional step, when there are schemes available but the user
// only requests schemes that don't exist, first instantiation of the test would
// filter out all the user provided schemes (as they are not registered) but

// Some TF ops need a step container to be set to limit the lifetime of some
// resources (mostly TensorArray and Stack, used in while loop gradients in
// graph mode). Calling this on a context tells it to start a step.
TF_CAPI_EXPORT extern void TFE_ContextStartStep(TFE_Context* ctx);

// Ends a step. When there is no active step (that is, every started step has

index 3f9112e..8e60197 100644
--- a/iconv/gconv.h
+++ b/iconv/gconv.h
@@ -174,7 +174,7 @@ typedef struct __gconv_info
 {
   size_t __nsteps;
   struct __gconv_step *__steps;
-  __extension__ struct __gconv_step_data __data __flexarr;
+  __extension__ struct __gconv_step_data __data[0];
 } *__gconv_t;
 
 #endif /* gconv.h */
diff --git a/include/libc-symbols.h b/include/libc-symbols.h
index c555bf2..143b26d 100644

  # Find all one-step dependencies of those tests which are from //tensorflow
  # (since external deps will come from Python-level pip dependencies),
  # excluding dependencies and files that are known to be unneccessary.

      CancellationManager* cancellation_manager) = 0;

  // Returns the stack trace set by `SetStackTrace` if exists.
  virtual absl::optional<ManagedStackTrace> GetStackTrace() = 0;

  virtual void SetStepId(int64_t step_id) = 0;

  // For LLVM style RTTI.
  static bool classof(const AbstractOperation* ptr) {
    return ptr->getKind() == kEager || ptr->getKind() == kTfrt;
  }

 protected:

}

#undef CONFIG_TOGGLE

absl::Status InitGraphPlugin(void* dso_handle) {
  tsl::Env* env = tsl::Env::Default();

  // Step 1: Load symbol for `TF_InitPlugin`
  void* dso_symbol;
  TF_RETURN_IF_ERROR(
      env->GetSymbolFromLibrary(dso_handle, "TF_InitGraph", &dso_symbol));

  // Step 2: Call `TF_InitPlugin`
  auto init_fn = reinterpret_cast<TFInitGraphPluginFn>(dso_symbol);
  return InitGraphPlugin(init_fn);
}

  virtual void ClearCachesAndThreadExecutors() = 0;

  // Initialize the step resource container for a training step. This is used
  // in current TF runtime. For tfrt, it is used by fallback op handler.
  virtual void StartStep() = 0;
  // Destroy the step resource container for a training step.
  virtual void EndStep() = 0;

  // Return the Eager Executor for current thread. Please note that Eager

  // Step 2: Initialize the opaque filesystem structure
  auto filesystem = std::make_unique<TF_Filesystem>();
  TF_Status* c_status = TF_NewStatus();
  Status status = OkStatus();
  core_filesystem_ops->init(filesystem.get(), c_status);
  status = Status(c_status->status);
  TF_DeleteStatus(c_status);
  if (!status.ok()) return status;

  // Step 3: Actual registration

        `(batch_size, 1)`. This enables `Model` subclasses to process scalar
        data in their `train_step()`/`test_step()`/`predict_step()` methods. \
        Note that this change may break certain subclassed models. You can
        revert back to the previous behavior by adding upranking yourself in the
        `train_step()`/`test_step()`/`predict_step()` methods, e.g. `if
        x.shape.rank == 1: x = tf.expand_dims(x, axis=-1)`. Functional models as