net/socket/socket_posix.cc

// Copyright 2014 The Chromium Authors. All rights reserved.
// Use of this source code is governed by a BSD-style license that can be
// found in the LICENSE file.

#include "net/socket/socket_posix.h"

#include <errno.h>
#include <netinet/in.h>
#include <sys/socket.h>
#include <utility>

#include "base/callback_helpers.h"
#include "base/files/file_util.h"
#include "base/logging.h"
#include "base/message_loop/message_loop.h"
#include "base/posix/eintr_wrapper.h"
#include "base/trace_event/trace_event.h"
#include "build/build_config.h"
#include "net/base/io_buffer.h"
#include "net/base/ip_endpoint.h"
#include "net/base/net_errors.h"
#include "net/base/sockaddr_storage.h"
#include "net/base/trace_constants.h"
#include "net/traffic_annotation/network_traffic_annotation.h"

#if defined(OS_FUCHSIA)
#include <poll.h>
#include <sys/ioctl.h>
#endif  // OS_FUCHSIA

namespace net {

namespace {

int MapAcceptError(int os_error) {
  switch (os_error) {
    // If the client aborts the connection before the server calls accept,
    // POSIX specifies accept should fail with ECONNABORTED. The server can
    // ignore the error and just call accept again, so we map the error to
    // ERR_IO_PENDING. See UNIX Network Programming, Vol. 1, 3rd Ed., Sec.
    // 5.11, "Connection Abort before accept Returns".
    case ECONNABORTED:
      return ERR_IO_PENDING;
    default:
      return MapSystemError(os_error);
  }
}

int MapConnectError(int os_error) {
  switch (os_error) {
    case EINPROGRESS:
      return ERR_IO_PENDING;
    case EACCES:
      return ERR_NETWORK_ACCESS_DENIED;
    case ETIMEDOUT:
      return ERR_CONNECTION_TIMED_OUT;
    default: {
      int net_error = MapSystemError(os_error);
      if (net_error == ERR_FAILED)
        return ERR_CONNECTION_FAILED;  // More specific than ERR_FAILED.
      return net_error;
    }
  }
}

}  // namespace

SocketPosix::SocketPosix()
    : socket_fd_(kInvalidSocket),
      accept_socket_watcher_(FROM_HERE),
      read_socket_watcher_(FROM_HERE),
      read_buf_len_(0),
      write_socket_watcher_(FROM_HERE),
      write_buf_len_(0),
      waiting_connect_(false) {}

SocketPosix::~SocketPosix() {
  Close();
}

int SocketPosix::Open(int address_family) {
  DCHECK(thread_checker_.CalledOnValidThread());
  DCHECK_EQ(kInvalidSocket, socket_fd_);
  DCHECK(address_family == AF_INET ||
         address_family == AF_INET6 ||
         address_family == AF_UNIX);

  socket_fd_ = CreatePlatformSocket(
      address_family,
      SOCK_STREAM,
      address_family == AF_UNIX ? 0 : IPPROTO_TCP);
  if (socket_fd_ < 0) {
    PLOG(ERROR) << "CreatePlatformSocket() returned an error, errno=" << errno;
    return MapSystemError(errno);
  }

  if (!base::SetNonBlocking(socket_fd_)) {
    int rv = MapSystemError(errno);
    Close();
    return rv;
  }

  return OK;
}

int SocketPosix::AdoptConnectedSocket(SocketDescriptor socket,
                                      const SockaddrStorage& address) {
  int rv = AdoptUnconnectedSocket(socket);
  if (rv != OK)
    return rv;

  SetPeerAddress(address);
  return OK;
}

int SocketPosix::AdoptUnconnectedSocket(SocketDescriptor socket) {
  DCHECK(thread_checker_.CalledOnValidThread());
  DCHECK_EQ(kInvalidSocket, socket_fd_);

  socket_fd_ = socket;

  if (!base::SetNonBlocking(socket_fd_)) {
    int rv = MapSystemError(errno);
    Close();
    return rv;
  }

  return OK;
}

SocketDescriptor SocketPosix::ReleaseConnectedSocket() {
  StopWatchingAndCleanUp();
  SocketDescriptor socket_fd = socket_fd_;
  socket_fd_ = kInvalidSocket;
  return socket_fd;
}

int SocketPosix::Bind(const SockaddrStorage& address) {
  DCHECK(thread_checker_.CalledOnValidThread());
  DCHECK_NE(kInvalidSocket, socket_fd_);

  int rv = bind(socket_fd_, address.addr, address.addr_len);
  if (rv < 0) {
    PLOG(ERROR) << "bind() returned an error, errno=" << errno;
    return MapSystemError(errno);
  }

  return OK;
}

int SocketPosix::Listen(int backlog) {
  DCHECK(thread_checker_.CalledOnValidThread());
  DCHECK_NE(kInvalidSocket, socket_fd_);
  DCHECK_LT(0, backlog);

  int rv = listen(socket_fd_, backlog);
  if (rv < 0) {
    PLOG(ERROR) << "listen() returned an error, errno=" << errno;
    return MapSystemError(errno);
  }

  return OK;
}

int SocketPosix::Accept(std::unique_ptr<SocketPosix>* socket,
                        const CompletionCallback& callback) {
  DCHECK(thread_checker_.CalledOnValidThread());
  DCHECK_NE(kInvalidSocket, socket_fd_);
  DCHECK(accept_callback_.is_null());
  DCHECK(socket);
  DCHECK(!callback.is_null());

  int rv = DoAccept(socket);
  if (rv != ERR_IO_PENDING)
    return rv;

  if (!base::MessageLoopForIO::current()->WatchFileDescriptor(
          socket_fd_, true, base::MessagePumpForIO::WATCH_READ,
          &accept_socket_watcher_, this)) {
    PLOG(ERROR) << "WatchFileDescriptor failed on accept, errno " << errno;
    return MapSystemError(errno);
  }

  accept_socket_ = socket;
  accept_callback_ = callback;
  return ERR_IO_PENDING;
}

int SocketPosix::Connect(const SockaddrStorage& address,
                         const CompletionCallback& callback) {
  DCHECK(thread_checker_.CalledOnValidThread());
  DCHECK_NE(kInvalidSocket, socket_fd_);
  DCHECK(!waiting_connect_);
  DCHECK(!callback.is_null());

  SetPeerAddress(address);

  int rv = DoConnect();
  if (rv != ERR_IO_PENDING)
    return rv;

  if (!base::MessageLoopForIO::current()->WatchFileDescriptor(
          socket_fd_, true, base::MessagePumpForIO::WATCH_WRITE,
          &write_socket_watcher_, this)) {
    PLOG(ERROR) << "WatchFileDescriptor failed on connect, errno " << errno;
    return MapSystemError(errno);
  }

  // There is a race-condition in the above code if the kernel receive a RST
  // packet for the "connect" call before the registration of the socket file
  // descriptor to the message loop pump. On most platform it is benign as the
  // message loop pump is awakened for that socket in an error state, but on
  // iOS this does not happens. Check the status of the socket at this point
  // and if in error, consider the connection as failed.
  int os_error = 0;
  socklen_t len = sizeof(os_error);
  if (getsockopt(socket_fd_, SOL_SOCKET, SO_ERROR, &os_error, &len) == 0) {
    // TCPSocketPosix expects errno to be set.
    errno = os_error;
  }

  rv = MapConnectError(errno);
  if (rv != OK && rv != ERR_IO_PENDING) {
    write_socket_watcher_.StopWatchingFileDescriptor();
    return rv;
  }

  write_callback_ = callback;
  waiting_connect_ = true;
  return ERR_IO_PENDING;
}

bool SocketPosix::IsConnected() const {
  DCHECK(thread_checker_.CalledOnValidThread());

  if (socket_fd_ == kInvalidSocket || waiting_connect_)
    return false;

#if !defined(OS_FUCHSIA)
  // Checks if connection is alive.
  char c;
  int rv = HANDLE_EINTR(recv(socket_fd_, &c, 1, MSG_PEEK));
  if (rv == 0)
    return false;
  if (rv == -1 && errno != EAGAIN && errno != EWOULDBLOCK)
    return false;

  return true;

#else   // OS_FUCHSIA
  // Fuchsia currently doesn't support MSG_PEEK flag in recv(), so the code
  // above doesn't work on Fuchsia. IsConnected() must return true if the
  // connection is alive or if it was terminated but there is still data pending
  // in the incoming buffer.
  //   1. Check if the connection is alive using poll(POLLRDHUP).
  //   2. If closed, then use ioctl(FIONREAD) to check if there is data to be
  //   read.
  // TODO(bug 738275): Remove once MSG_PEEK is implemented on Fuchsia.
  struct pollfd pollfd;
  pollfd.fd = socket_fd_;
  pollfd.events = POLLRDHUP;
  pollfd.revents = 0;
  const int poll_result = HANDLE_EINTR(poll(&pollfd, 1, 0));

  if (poll_result == 1) {
    int bytes_available;
    int ioctl_result =
        HANDLE_EINTR(ioctl(socket_fd_, FIONREAD, &bytes_available));
    return ioctl_result == 0 && bytes_available > 0;
  }

  return poll_result == 0;
#endif  // OS_FUCHSIA
}

bool SocketPosix::IsConnectedAndIdle() const {
  DCHECK(thread_checker_.CalledOnValidThread());

  if (socket_fd_ == kInvalidSocket || waiting_connect_)
    return false;

#if !defined(OS_FUCHSIA)
  // Check if connection is alive and we haven't received any data
  // unexpectedly.
  char c;
  int rv = HANDLE_EINTR(recv(socket_fd_, &c, 1, MSG_PEEK));
  if (rv >= 0)
    return false;
  if (errno != EAGAIN && errno != EWOULDBLOCK)
    return false;

  return true;

#else   // OS_FUCHSIA
  // Fuchsia currently doesn't support MSG_PEEK flag in recv(), so the code
  // above doesn't work on Fuchsia. Use poll(POLLIN) to check state of the
  // socket. POLLIN is signaled if the socket is readable or if it was closed by
  // the peer, i.e. the socket is connected and idle if and only if POLLIN is
  // not signaled.
  // TODO(bug 738275): Remove once MSG_PEEK is implemented.
  struct pollfd pollfd;
  pollfd.fd = socket_fd_;
  pollfd.events = POLLIN;
  pollfd.revents = 0;
  const int poll_result = HANDLE_EINTR(poll(&pollfd, 1, 0));
  return poll_result == 0;
#endif  // OS_FUCHSIA
}

int SocketPosix::Read(IOBuffer* buf,
                      int buf_len,
                      const CompletionCallback& callback) {
  // Use base::Unretained() is safe here because OnFileCanReadWithoutBlocking()
  // won't be called if |this| is gone.
  int rv =
      ReadIfReady(buf, buf_len,
                  base::Bind(&SocketPosix::RetryRead, base::Unretained(this)));
  if (rv == ERR_IO_PENDING) {
    read_buf_ = buf;
    read_buf_len_ = buf_len;
    read_callback_ = callback;
  }
  return rv;
}

int SocketPosix::ReadIfReady(IOBuffer* buf,
                             int buf_len,
                             const CompletionCallback& callback) {
  DCHECK(thread_checker_.CalledOnValidThread());
  DCHECK_NE(kInvalidSocket, socket_fd_);
  DCHECK(!waiting_connect_);
  CHECK(read_if_ready_callback_.is_null());
  DCHECK(!callback.is_null());
  DCHECK_LT(0, buf_len);

  int rv = DoRead(buf, buf_len);
  if (rv != ERR_IO_PENDING)
    return rv;

  if (!base::MessageLoopForIO::current()->WatchFileDescriptor(
          socket_fd_, true, base::MessagePumpForIO::WATCH_READ,
          &read_socket_watcher_, this)) {
    PLOG(ERROR) << "WatchFileDescriptor failed on read, errno " << errno;
    return MapSystemError(errno);
  }

  read_if_ready_callback_ = callback;
  return ERR_IO_PENDING;
}

int SocketPosix::Write(
    IOBuffer* buf,
    int buf_len,
    const CompletionCallback& callback,
    const NetworkTrafficAnnotationTag& /* traffic_annotation */) {
  DCHECK(thread_checker_.CalledOnValidThread());
  DCHECK_NE(kInvalidSocket, socket_fd_);
  DCHECK(!waiting_connect_);
  CHECK(write_callback_.is_null());
  // Synchronous operation not supported
  DCHECK(!callback.is_null());
  DCHECK_LT(0, buf_len);

  int rv = DoWrite(buf, buf_len);
  if (rv == ERR_IO_PENDING)
    rv = WaitForWrite(buf, buf_len, callback);
  return rv;
}

int SocketPosix::WaitForWrite(IOBuffer* buf,
                              int buf_len,
                              const CompletionCallback& callback) {
  DCHECK(thread_checker_.CalledOnValidThread());
  DCHECK_NE(kInvalidSocket, socket_fd_);
  DCHECK(write_callback_.is_null());
  // Synchronous operation not supported
  DCHECK(!callback.is_null());
  DCHECK_LT(0, buf_len);

  if (!base::MessageLoopForIO::current()->WatchFileDescriptor(
          socket_fd_, true, base::MessagePumpForIO::WATCH_WRITE,
          &write_socket_watcher_, this)) {
    PLOG(ERROR) << "WatchFileDescriptor failed on write, errno " << errno;
    return MapSystemError(errno);
  }

  write_buf_ = buf;
  write_buf_len_ = buf_len;
  write_callback_ = callback;
  return ERR_IO_PENDING;
}

int SocketPosix::GetLocalAddress(SockaddrStorage* address) const {
  DCHECK(thread_checker_.CalledOnValidThread());
  DCHECK(address);

  if (getsockname(socket_fd_, address->addr, &address->addr_len) < 0)
    return MapSystemError(errno);
  return OK;
}

int SocketPosix::GetPeerAddress(SockaddrStorage* address) const {
  DCHECK(thread_checker_.CalledOnValidThread());
  DCHECK(address);

  if (!HasPeerAddress())
    return ERR_SOCKET_NOT_CONNECTED;

  *address = *peer_address_;
  return OK;
}

void SocketPosix::SetPeerAddress(const SockaddrStorage& address) {
  DCHECK(thread_checker_.CalledOnValidThread());
  // |peer_address_| will be non-NULL if Connect() has been called. Unless
  // Close() is called to reset the internal state, a second call to Connect()
  // is not allowed.
  // Please note that we don't allow a second Connect() even if the previous
  // Connect() has failed. Connecting the same |socket_| again after a
  // connection attempt failed results in unspecified behavior according to
  // POSIX.
  DCHECK(!peer_address_);
  peer_address_.reset(new SockaddrStorage(address));
}

bool SocketPosix::HasPeerAddress() const {
  DCHECK(thread_checker_.CalledOnValidThread());
  return peer_address_ != NULL;
}

void SocketPosix::Close() {
  DCHECK(thread_checker_.CalledOnValidThread());

  StopWatchingAndCleanUp();

  if (socket_fd_ != kInvalidSocket) {
    if (IGNORE_EINTR(close(socket_fd_)) < 0)
      PLOG(ERROR) << "close() returned an error, errno=" << errno;
    socket_fd_ = kInvalidSocket;
  }
}

void SocketPosix::DetachFromThread() {
  thread_checker_.DetachFromThread();
}

void SocketPosix::OnFileCanReadWithoutBlocking(int fd) {
  TRACE_EVENT0(kNetTracingCategory,
               "SocketPosix::OnFileCanReadWithoutBlocking");
  if (!accept_callback_.is_null()) {
    AcceptCompleted();
  } else {
    DCHECK(!read_if_ready_callback_.is_null());
    ReadCompleted();
  }
}

void SocketPosix::OnFileCanWriteWithoutBlocking(int fd) {
  DCHECK(!write_callback_.is_null());
  if (waiting_connect_) {
    ConnectCompleted();
  } else {
    WriteCompleted();
  }
}

int SocketPosix::DoAccept(std::unique_ptr<SocketPosix>* socket) {
  SockaddrStorage new_peer_address;
  int new_socket = HANDLE_EINTR(accept(socket_fd_,
                                       new_peer_address.addr,
                                       &new_peer_address.addr_len));
  if (new_socket < 0)
    return MapAcceptError(errno);

  std::unique_ptr<SocketPosix> accepted_socket(new SocketPosix);
  int rv = accepted_socket->AdoptConnectedSocket(new_socket, new_peer_address);
  if (rv != OK)
    return rv;

  *socket = std::move(accepted_socket);
  return OK;
}

void SocketPosix::AcceptCompleted() {
  DCHECK(accept_socket_);
  int rv = DoAccept(accept_socket_);
  if (rv == ERR_IO_PENDING)
    return;

  bool ok = accept_socket_watcher_.StopWatchingFileDescriptor();
  DCHECK(ok);
  accept_socket_ = NULL;
  base::ResetAndReturn(&accept_callback_).Run(rv);
}

int SocketPosix::DoConnect() {
  int rv = HANDLE_EINTR(connect(socket_fd_,
                                peer_address_->addr,
                                peer_address_->addr_len));
  DCHECK_GE(0, rv);
  return rv == 0 ? OK : MapConnectError(errno);
}

void SocketPosix::ConnectCompleted() {
  // Get the error that connect() completed with.
  int os_error = 0;
  socklen_t len = sizeof(os_error);
  if (getsockopt(socket_fd_, SOL_SOCKET, SO_ERROR, &os_error, &len) == 0) {
    // TCPSocketPosix expects errno to be set.
    errno = os_error;
  }

  int rv = MapConnectError(errno);
  if (rv == ERR_IO_PENDING)
    return;

  bool ok = write_socket_watcher_.StopWatchingFileDescriptor();
  DCHECK(ok);
  waiting_connect_ = false;
  base::ResetAndReturn(&write_callback_).Run(rv);
}

int SocketPosix::DoRead(IOBuffer* buf, int buf_len) {
  int rv = HANDLE_EINTR(read(socket_fd_, buf->data(), buf_len));
  return rv >= 0 ? rv : MapSystemError(errno);
}

void SocketPosix::RetryRead(int rv) {
  DCHECK(read_callback_);
  DCHECK(read_buf_);
  DCHECK_LT(0, read_buf_len_);

  if (rv == OK) {
    rv = ReadIfReady(
        read_buf_.get(), read_buf_len_,
        base::Bind(&SocketPosix::RetryRead, base::Unretained(this)));
    if (rv == ERR_IO_PENDING)
      return;
  }
  read_buf_ = nullptr;
  read_buf_len_ = 0;
  base::ResetAndReturn(&read_callback_).Run(rv);
}

void SocketPosix::ReadCompleted() {
  DCHECK(read_if_ready_callback_);

  bool ok = read_socket_watcher_.StopWatchingFileDescriptor();
  DCHECK(ok);
  base::ResetAndReturn(&read_if_ready_callback_).Run(OK);
}

int SocketPosix::DoWrite(IOBuffer* buf, int buf_len) {
#if defined(OS_LINUX) || defined(OS_ANDROID)
  // Disable SIGPIPE for this write. Although Chromium globally disables
  // SIGPIPE, the net stack may be used in other consumers which do not do
  // this. MSG_NOSIGNAL is a Linux-only API. On OS X, this is a setsockopt on
  // socket creation.
  int rv = HANDLE_EINTR(send(socket_fd_, buf->data(), buf_len, MSG_NOSIGNAL));
#else
  int rv = HANDLE_EINTR(write(socket_fd_, buf->data(), buf_len));
#endif
  return rv >= 0 ? rv : MapSystemError(errno);
}

void SocketPosix::WriteCompleted() {
  int rv = DoWrite(write_buf_.get(), write_buf_len_);
  if (rv == ERR_IO_PENDING)
    return;

  bool ok = write_socket_watcher_.StopWatchingFileDescriptor();
  DCHECK(ok);
  write_buf_ = NULL;
  write_buf_len_ = 0;
  base::ResetAndReturn(&write_callback_).Run(rv);
}

void SocketPosix::StopWatchingAndCleanUp() {
  bool ok = accept_socket_watcher_.StopWatchingFileDescriptor();
  DCHECK(ok);
  ok = read_socket_watcher_.StopWatchingFileDescriptor();
  DCHECK(ok);
  ok = write_socket_watcher_.StopWatchingFileDescriptor();
  DCHECK(ok);

  if (!accept_callback_.is_null()) {
    accept_socket_ = NULL;
    accept_callback_.Reset();
  }

  if (!read_callback_.is_null()) {
    read_buf_ = NULL;
    read_buf_len_ = 0;
    read_callback_.Reset();
  }

  read_if_ready_callback_.Reset();

  if (!write_callback_.is_null()) {
    write_buf_ = NULL;
    write_buf_len_ = 0;
    write_callback_.Reset();
  }

  waiting_connect_ = false;
  peer_address_.reset();
}

}  // namespace net