struct sockaddr_un prepare_address(const char *path) {
    struct sockaddr_un address;
    address.sun_family = AF_UNIX;
    strcpy(address.sun_path, path);
    return address;
  }

  // Implemented based on https://man7.org/linux/man-pages/man7/unix.7.html
  size_t address_length(struct sockaddr_un address) {

            

Reported by FlawFinder.

                }

  void register_allocation(AllocInfo &info) {
    char buffer[3] = {0, 0, 0};
    send(&info, sizeof(info));
    recv(buffer, 2);
    if (strcmp(buffer, "OK") != 0)
      throw std::runtime_error("Shared memory manager didn't respond with an OK");
  }

            

Reported by FlawFinder.

              
  // Implemented based on https://man7.org/linux/man-pages/man7/unix.7.html
  size_t address_length(struct sockaddr_un address) {
    return offsetof(sockaddr_un, sun_path) + strlen(address.sun_path) + 1;
  }

  void recv(void *_buffer, size_t num_bytes) {
    char *buffer = (char*)_buffer;
    size_t bytes_received = 0;

            

Reported by FlawFinder.

                  while (bytes_received < num_bytes) {
      SYSCHECK_ERR_RETURN_NEG1(poll(&pfd, 1, 1000));
      if (pfd.revents & POLLIN) {
        SYSCHECK_ERR_RETURN_NEG1(step_received = ::read(socket_fd, buffer, num_bytes - bytes_received));
        if (step_received == 0)
          throw std::runtime_error("Other end has closed the connection");
        bytes_received += step_received;
        buffer += step_received;
      } else if (pfd.revents & (POLLERR | POLLHUP)) {

            

Reported by FlawFinder.

                  SYSCHECK_ERR_RETURN_NEG1(close(pipe_ends[0]));
    SYSCHECK_ERR_RETURN_NEG1(dup2(pipe_ends[1], 1)); // Replace stdout
    SYSCHECK_ERR_RETURN_NEG1(close(pipe_ends[1]));
    execl(manager_executable_path.c_str(), "torch_shm_manager", NULL);

    std::string msg("ERROR: execl failed: ");
    msg += std::strerror(errno);
    msg += '\n';
    write(1, msg.c_str(), msg.size());

            

Reported by FlawFinder.

                if (len >= sizeof(info.filename)) {
    throw std::runtime_error("MapAllocatorContext_filename too long");
  }
  memcpy(info.filename, filename, len + 1);
  return info;
}

void start_manager() {
  std::array<int, 2> pipe_ends;

            

Reported by FlawFinder.

                AllocInfo info = {0};
  info.pid = getpid();
  info.free = false;
  size_t len = strlen(filename);
  if (len >= sizeof(info.filename)) {
    throw std::runtime_error("MapAllocatorContext_filename too long");
  }
  memcpy(info.filename, filename, len + 1);
  return info;

            

Reported by FlawFinder.

                std::array<char, MAX_BUFFER_SIZE> buffer;
  std::string handle;
  while(handle.empty() || handle.back() != '\n') {
    const auto bytes_read = read(pipe_ends[0], buffer.data(), buffer.size());
    SYSCHECK_ERR_RETURN_NEG1(bytes_read);
    if (bytes_read == 0) {
      break;
    }
    handle.append(buffer.data(), bytes_read);

            

Reported by FlawFinder.

              import torch.utils.data.graph


def apply_sharding(datapipe, num_of_instances, instance_id):
    graph = torch.utils.data.graph.traverse(datapipe)

    def traverse_graph(graph):
        results = set()
        for datapipe, sub_graph in graph.items():

            

Reported by Pylint.

              import torch.utils.data.graph


def apply_sharding(datapipe, num_of_instances, instance_id):
    graph = torch.utils.data.graph.traverse(datapipe)

    def traverse_graph(graph):
        results = set()
        for datapipe, sub_graph in graph.items():

            

Reported by Pylint.

                          if pipe.is_shardable():
                if hasattr(pipe, 'apply_sharding'):
                    if already_applied_to is not None:
                        raise RuntimeError('This implementation of sharding can be only applied once per instance of DataPipeline.',
                                           'Already applied to', already_applied_to, 'while trying to apply to', pipe)
                    pipe.apply_sharding(num_of_instances, instance_id)
                    already_applied_to = pipe

            

Reported by Pylint.

                              if hasattr(pipe, 'apply_sharding'):
                    if already_applied_to is not None:
                        raise RuntimeError('This implementation of sharding can be only applied once per instance of DataPipeline.',
                                           'Already applied to', already_applied_to, 'while trying to apply to', pipe)
                    pipe.apply_sharding(num_of_instances, instance_id)
                    already_applied_to = pipe

            

Reported by Pylint.

              
def worker_init_fn(worker_id):
    info = torch.utils.data.get_worker_info()
    num_workers = info.num_workers
    datapipe = info.dataset
    torch.utils.data.sharding.apply_sharding(datapipe, num_workers, worker_id)

            

Reported by Pylint.

              def worker_init_fn(worker_id):
    info = torch.utils.data.get_worker_info()
    num_workers = info.num_workers
    datapipe = info.dataset
    torch.utils.data.sharding.apply_sharding(datapipe, num_workers, worker_id)

            

Reported by Pylint.

              import torch.utils.data.sharding


def worker_init_fn(worker_id):
    info = torch.utils.data.get_worker_info()
    num_workers = info.num_workers
    datapipe = info.dataset
    torch.utils.data.sharding.apply_sharding(datapipe, num_workers, worker_id)

            

Reported by Pylint.

              import torch.utils.data.sharding


def worker_init_fn(worker_id):
    info = torch.utils.data.get_worker_info()
    num_workers = info.num_workers
    datapipe = info.dataset
    torch.utils.data.sharding.apply_sharding(datapipe, num_workers, worker_id)

            

Reported by Pylint.

              from .linear_relu import LinearReLU
from .conv_relu import ConvReLU1d, ConvReLU2d, ConvReLU3d
from .bn_relu import BNReLU2d, BNReLU3d

__all__ = [
    'LinearReLU',
    'ConvReLU1d',
    'ConvReLU2d',
    'ConvReLU3d',

            

Reported by Pylint.

              from .linear_relu import LinearReLU
from .conv_relu import ConvReLU1d, ConvReLU2d, ConvReLU3d
from .bn_relu import BNReLU2d, BNReLU3d

__all__ = [
    'LinearReLU',
    'ConvReLU1d',
    'ConvReLU2d',
    'ConvReLU3d',

            

Reported by Pylint.

              from .linear_relu import LinearReLU
from .conv_relu import ConvReLU1d, ConvReLU2d, ConvReLU3d
from .bn_relu import BNReLU2d, BNReLU3d

__all__ = [
    'LinearReLU',
    'ConvReLU1d',
    'ConvReLU2d',
    'ConvReLU3d',

            

Reported by Pylint.

              from .linear_relu import LinearReLU
from .conv_relu import ConvReLU1d, ConvReLU2d, ConvReLU3d
from .bn_relu import BNReLU2d, BNReLU3d

__all__ = [
    'LinearReLU',
    'ConvReLU1d',
    'ConvReLU2d',
    'ConvReLU3d',

            

Reported by Pylint.

              from .linear import Linear
from .conv import Conv2d
from .conv import Conv3d

__all__ = [
    "Linear",
    "Conv2d",
    "Conv3d",
]

            

Reported by Pylint.

              from .linear import Linear
from .conv import Conv2d
from .conv import Conv3d

__all__ = [
    "Linear",
    "Conv2d",
    "Conv3d",
]

            

Reported by Pylint.

              from .linear import Linear
from .conv import Conv2d
from .conv import Conv3d

__all__ = [
    "Linear",
    "Conv2d",
    "Conv3d",
]

            

Reported by Pylint.

              from .linear import Linear
from .conv import Conv2d
from .conv import Conv3d

__all__ = [
    "Linear",
    "Conv2d",
    "Conv3d",
]

            

Reported by Pylint.

              from .prepare import prepare
from .convert import convert
from .fuse import Fuser

            

Reported by Pylint.

              from .prepare import prepare
from .convert import convert
from .fuse import Fuser

            

Reported by Pylint.

              from .prepare import prepare
from .convert import convert
from .fuse import Fuser

            

Reported by Pylint.

              from .prepare import prepare
from .convert import convert
from .fuse import Fuser

            

Reported by Pylint.

              import inspect
import sys

from .dispatcher import Dispatcher, MethodDispatcher

global_namespace = dict()  # type: ignore[var-annotated]


def dispatch(*types, **kwargs):

            

Reported by Pylint.

                      return signature.parameters.get('self', None) is not None
    else:
        if sys.version_info.major < 3:
            spec = inspect.getargspec(func)
        else:
            spec = inspect.getfullargspec(func)  # type: ignore[union-attr, assignment]
        return spec and spec.args and spec.args[0] == 'self'

            

Reported by Pylint.

              import inspect
import sys

from .dispatcher import Dispatcher, MethodDispatcher

global_namespace = dict()  # type: ignore[var-annotated]


def dispatch(*types, **kwargs):

            

Reported by Pylint.

                  Note that this has to work as the method is defined but before the class is
    defined.  At this stage methods look like functions.
    """
    if hasattr(inspect, "signature"):
        signature = inspect.signature(func)
        return signature.parameters.get('self', None) is not None
    else:
        if sys.version_info.major < 3:
            spec = inspect.getargspec(func)

            

Reported by Pylint.

              from types import ModuleType
from typing import Any

from .._mangling import is_mangled


def is_from_package(obj: Any) -> bool:
    """
    Return whether an object was loaded from a package.

            

Reported by Pylint.

              from types import ModuleType
from typing import Any

from .._mangling import is_mangled


def is_from_package(obj: Any) -> bool:
    """
    Return whether an object was loaded from a package.

            

Reported by Pylint.

              
    Note: packaged objects from externed modules will return ``False``.
    """
    if type(obj) == ModuleType:
        return is_mangled(obj.__name__)
    else:
        return is_mangled(type(obj).__module__)

            

Reported by Pylint.

              
    Note: packaged objects from externed modules will return ``False``.
    """
    if type(obj) == ModuleType:
        return is_mangled(obj.__name__)
    else:
        return is_mangled(type(obj).__module__)

            

Reported by Pylint.

              import torch


def make_np(x):
    """
    Args:
      x: An instance of torch tensor or caffe blob name

    Returns:

            

Reported by Pylint.

                      'Got {}, but numpy array, torch tensor, or caffe2 blob name are expected.'.format(type(x)))


def _prepare_pytorch(x):
    if isinstance(x, torch.autograd.Variable):
        x = x.data
    x = x.cpu().numpy()
    return x


            

Reported by Pylint.

                  return x


def _prepare_caffe2(x):
    from caffe2.python import workspace
    x = workspace.FetchBlob(x)
    return x

            

Reported by Pylint.

              

def _prepare_caffe2(x):
    from caffe2.python import workspace
    x = workspace.FetchBlob(x)
    return x

            

Reported by Pylint.