p9y: bsd

[m6w6/libmemcached] / src / libmemcached / io.cc

/*
    +--------------------------------------------------------------------+
    | libmemcached - C/C++ Client Library for memcached                  |
    +--------------------------------------------------------------------+
    | Redistribution and use in source and binary forms, with or without |
    | modification, are permitted under the terms of the BSD license.    |
    | You should have received a copy of the license in a bundled file   |
    | named LICENSE; in case you did not receive a copy you can review   |
    | the terms online at: https://opensource.org/licenses/BSD-3-Clause  |
    +--------------------------------------------------------------------+
    | Copyright (c) 2006-2014 Brian Aker   https://datadifferential.com/ |
    | Copyright (c) 2020 Michael Wallner   <mike@php.net>                |
    +--------------------------------------------------------------------+
*/

#include "libmemcached/common.h"
#include "p9y/poll.hpp"

void initialize_binary_request(memcached_instance_st *server,
                               protocol_binary_request_header &header) {
  server->request_id++;
  header.request.magic = PROTOCOL_BINARY_REQ;
  header.request.opaque = htons(server->request_id);
}

enum memc_read_or_write { MEM_READ, MEM_WRITE };

/**
 * Try to fill the input buffer for a server with as much
 * data as possible.
 *
 * @param instance the server to pack
 */
static bool repack_input_buffer(memcached_instance_st *instance) {
  if (instance->read_ptr != instance->read_buffer) {
    /* Move all of the data to the beginning of the buffer so
     ** that we can fit more data into the buffer...
     */
    memmove(instance->read_buffer, instance->read_ptr, instance->read_buffer_length);
    instance->read_ptr = instance->read_buffer;
  }

  /* There is room in the buffer, try to fill it! */
  if (instance->read_buffer_length != MEMCACHED_MAX_BUFFER) {
    do {
      /* Just try a single read to grab what's available */
      ssize_t nr;
      if ((nr = ::recv(instance->fd, instance->read_ptr + instance->read_buffer_length,
                       MEMCACHED_MAX_BUFFER - instance->read_buffer_length, MSG_NOSIGNAL))
          <= 0)
      {
        if (nr == 0) {
          memcached_set_error(*instance, MEMCACHED_CONNECTION_FAILURE, MEMCACHED_AT);
        } else {
          switch (get_socket_errno()) {
          case EINTR:
            continue;

#if EWOULDBLOCK != EAGAIN
          case EWOULDBLOCK:
#endif
          case EAGAIN:
#ifdef __linux
          case ERESTART:
#endif
            break; // No IO is fine, we can just move on

          default:
            memcached_set_errno(*instance, get_socket_errno(), MEMCACHED_AT);
          }
        }

        break;
      } else // We read data, append to our read buffer
      {
        instance->read_buffer_length += size_t(nr);

        return true;
      }
    } while (false);
  }

  return false;
}

/**
 * If the we have callbacks connected to this server structure
 * we may start process the input queue and fire the callbacks
 * for the incomming messages. This function is _only_ called
 * when the input buffer is full, so that we _know_ that we have
 * at least _one_ message to process.
 *
 * @param instance the server to star processing iput messages for
 * @return true if we processed anything, false otherwise
 */
static bool process_input_buffer(memcached_instance_st *instance) {
  /*
   ** We might be able to process some of the response messages if we
   ** have a callback set up
   */
  if (instance->root->callbacks) {
    /*
     * We might have responses... try to read them out and fire
     * callbacks
     */
    memcached_callback_st cb = *instance->root->callbacks;

    memcached_set_processing_input((Memcached *) instance->root, true);

    char buffer[MEMCACHED_DEFAULT_COMMAND_SIZE];
    Memcached *root = (Memcached *) instance->root;
    memcached_return_t error = memcached_response(instance, buffer, sizeof(buffer), &root->result);

    memcached_set_processing_input(root, false);

    if (error == MEMCACHED_SUCCESS) {
      for (unsigned int x = 0; x < cb.number_of_callback; x++) {
        error = (*cb.callback[x])(instance->root, &root->result, cb.context);
        if (error != MEMCACHED_SUCCESS) {
          break;
        }
      }

      /* @todo what should I do with the error message??? */
    }
    /* @todo what should I do with other error messages?? */
    return true;
  }

  return false;
}

static memcached_return_t io_wait(memcached_instance_st *instance, const short events) {
  /*
   ** We are going to block on write, but at least on Solaris we might block
   ** on write if we haven't read anything from our input buffer..
   ** Try to purge the input buffer if we don't do any flow control in the
   ** application layer (just sending a lot of data etc)
   ** The test is moved down in the purge function to avoid duplication of
   ** the test.
   */
  if (events & POLLOUT) {
    if (memcached_purge(instance) == false) {
      return MEMCACHED_FAILURE;
    }
  }

  struct pollfd fds;
  fds.fd = instance->fd;
  fds.events = events;
  fds.revents = 0;

  if (fds.events & POLLOUT) /* write */ {
    instance->io_wait_count.write++;
  } else {
    instance->io_wait_count.read++;
  }

  if (instance->root->poll_timeout
      == 0) // Mimic 0 causes timeout behavior (not all platforms do this)
  {
    return memcached_set_error(*instance, MEMCACHED_TIMEOUT, MEMCACHED_AT,
                               memcached_literal_param("poll_timeout() was set to zero"));
  }

  size_t loop_max = 5;
  while (--loop_max) // While loop is for ERESTART or EINTR
  {
    int active_fd = poll(&fds, 1, instance->root->poll_timeout);

    if (active_fd >= 1) {
      assert_msg(active_fd == 1, "poll() returned an unexpected number of active file descriptors");
      if (fds.revents & POLLIN or fds.revents & POLLOUT) {
        return MEMCACHED_SUCCESS;
      }

      if (fds.revents & POLLHUP) {
        return memcached_set_error(*instance, MEMCACHED_CONNECTION_FAILURE, MEMCACHED_AT,
                                   memcached_literal_param("poll() detected hang up"));
      }

      if (fds.revents & POLLERR) {
        int local_errno = EINVAL;
        int err;
        socklen_t len = sizeof(err);
        if (getsockopt(instance->fd, SOL_SOCKET, SO_ERROR, (char *) &err, &len) == 0) {
          if (err == 0) // treat this as EINTR
          {
            continue;
          }
          local_errno = err;
        }
        memcached_quit_server(instance, true);
        return memcached_set_errno(*instance, local_errno, MEMCACHED_AT,
                                   memcached_literal_param("poll() returned POLLHUP"));
      }

      return memcached_set_error(
          *instance, MEMCACHED_FAILURE, MEMCACHED_AT,
          memcached_literal_param("poll() returned a value that was not dealt with"));
    }

    if (active_fd == 0) {
      return memcached_set_error(*instance, MEMCACHED_TIMEOUT, MEMCACHED_AT,
                                 memcached_literal_param("No active_fd were found"));
    }

    // Only an error should result in this code being called.
    int local_errno = get_socket_errno(); // We cache in case memcached_quit_server() modifies errno
    assert_msg(active_fd == -1, "poll() returned an unexpected value");
    switch (local_errno) {
#ifdef __linux
    case ERESTART:
#endif
    case EINTR:
      continue;

    case EFAULT:
    case ENOMEM:
      memcached_set_error(*instance, MEMCACHED_MEMORY_ALLOCATION_FAILURE, MEMCACHED_AT);
      break;

    case EINVAL:
      memcached_set_error(*instance, MEMCACHED_MEMORY_ALLOCATION_FAILURE, MEMCACHED_AT,
                          memcached_literal_param(
                              "RLIMIT_NOFILE exceeded, or if OSX the timeout value was invalid"));
      break;

    default:
      memcached_set_errno(*instance, local_errno, MEMCACHED_AT, memcached_literal_param("poll"));
    }

    break;
  }

  memcached_quit_server(instance, true);

  if (memcached_has_error(instance)) {
    return memcached_instance_error_return(instance);
  }

  return memcached_set_error(
      *instance, MEMCACHED_CONNECTION_FAILURE, MEMCACHED_AT,
      memcached_literal_param("number of attempts to call io_wait() failed"));
}

static bool io_flush(memcached_instance_st *instance, const bool with_flush,
                     memcached_return_t &error) {
  /*
   ** We might want to purge the input buffer if we haven't consumed
   ** any output yet... The test for the limits is the purge is inline
   ** in the purge function to avoid duplicating the logic..
   */
  {
    WATCHPOINT_ASSERT(instance->fd != INVALID_SOCKET);

    if (memcached_purge(instance) == false) {
      return false;
    }
  }
  char *local_write_ptr = instance->write_buffer;
  size_t write_length = instance->write_buffer_offset;

  error = MEMCACHED_SUCCESS;

  WATCHPOINT_ASSERT(instance->fd != INVALID_SOCKET);

  /* Looking for memory overflows */
#if defined(DEBUG)
  if (write_length == MEMCACHED_MAX_BUFFER)
    WATCHPOINT_ASSERT(instance->write_buffer == local_write_ptr);
  WATCHPOINT_ASSERT((instance->write_buffer + MEMCACHED_MAX_BUFFER)
                    >= (local_write_ptr + write_length));
#endif

  while (write_length) {
    WATCHPOINT_ASSERT(instance->fd != INVALID_SOCKET);
    WATCHPOINT_ASSERT(write_length > 0);

    int flags;
    if (with_flush) {
      flags = MSG_NOSIGNAL;
    } else {
      flags = MSG_NOSIGNAL | MSG_MORE;
    }

    ssize_t sent_length = ::send(instance->fd, local_write_ptr, write_length, flags);
    int local_errno = get_socket_errno(); // We cache in case memcached_quit_server() modifies errno

    if (sent_length == SOCKET_ERROR) {
#if 0 // @todo I should look at why we hit this bit of code hard frequently
      WATCHPOINT_ERRNO(get_socket_errno());
      WATCHPOINT_NUMBER(get_socket_errno());
#endif
      switch (get_socket_errno()) {
      case ENOBUFS:
        continue;

#if EWOULDBLOCK != EAGAIN
      case EWOULDBLOCK:
#endif
      case EAGAIN: {
        /*
         * We may be blocked on write because the input buffer
         * is full. Let's check if we have room in our input
         * buffer for more data and retry the write before
         * waiting..
         */
        if (repack_input_buffer(instance) or process_input_buffer(instance)) {
          continue;
        }

        memcached_return_t rc = io_wait(instance, POLLOUT);
        if (memcached_success(rc)) {
          continue;
        } else if (rc == MEMCACHED_TIMEOUT) {
          return false;
        }

        memcached_quit_server(instance, true);
        error = memcached_set_errno(*instance, local_errno, MEMCACHED_AT);
        return false;
      }
      case ENOTCONN:
      case EPIPE:
      default:
        memcached_quit_server(instance, true);
        error = memcached_set_errno(*instance, local_errno, MEMCACHED_AT);
        WATCHPOINT_ASSERT(instance->fd == INVALID_SOCKET);
        return false;
      }
    }

    instance->io_bytes_sent += uint32_t(sent_length);

    local_write_ptr += sent_length;
    write_length -= uint32_t(sent_length);
  }

  WATCHPOINT_ASSERT(write_length == 0);
  instance->write_buffer_offset = 0;

  return true;
}

memcached_return_t memcached_io_wait_for_write(memcached_instance_st *instance) {
  return io_wait(instance, POLLOUT);
}

memcached_return_t memcached_io_wait_for_read(memcached_instance_st *instance) {
  return io_wait(instance, POLLIN);
}

static memcached_return_t _io_fill(memcached_instance_st *instance) {
  ssize_t data_read;
  do {
    data_read = ::recv(instance->fd, instance->read_buffer, MEMCACHED_MAX_BUFFER, MSG_NOSIGNAL);
    int local_errno = get_socket_errno(); // We cache in case memcached_quit_server() modifies errno

    if (data_read == SOCKET_ERROR) {
      switch (get_socket_errno()) {
      case EINTR: // We just retry
        continue;

      case ETIMEDOUT: // OSX
#if EWOULDBLOCK != EAGAIN
      case EWOULDBLOCK:
#endif
      case EAGAIN:
#ifdef __linux
      case ERESTART:
#endif
      {
        memcached_return_t io_wait_ret;
        if (memcached_success(io_wait_ret = io_wait(instance, POLLIN))) {
          continue;
        }

        return io_wait_ret;
      }

        /* fall through */

      case ENOTCONN: // Programmer Error
        WATCHPOINT_ASSERT(0);
        // fall through
      case ENOTSOCK:
        WATCHPOINT_ASSERT(0);
        // fall through
      case EBADF:
        assert_msg(instance->fd != INVALID_SOCKET, "Programmer error, invalid socket");
        /* fall through */
      case EINVAL:
      case EFAULT:
      case ECONNREFUSED:
      default:
        memcached_quit_server(instance, true);
        memcached_set_errno(*instance, local_errno, MEMCACHED_AT);
        break;
      }

      return memcached_instance_error_return(instance);
    } else if (data_read == 0) {
      /*
        EOF. Any data received so far is incomplete
        so discard it. This always reads by byte in case of TCP
        and protocol enforcement happens at memcached_response()
        looking for '\n'. We do not care for UDB which requests 8 bytes
        at once. Generally, this means that connection went away. Since
        for blocking I/O we do not return 0 and for non-blocking case
        it will return EGAIN if data is not immediatly available.
      */
      memcached_quit_server(instance, true);
      return memcached_set_error(
          *instance, MEMCACHED_CONNECTION_FAILURE, MEMCACHED_AT,
          memcached_literal_param("::rec() returned zero, server has disconnected"));
    }
    instance->io_wait_count._bytes_read += data_read;
  } while (data_read <= 0);

  instance->io_bytes_sent = 0;
  instance->read_buffer_length = (size_t) data_read;
  instance->read_ptr = instance->read_buffer;

  return MEMCACHED_SUCCESS;
}

memcached_return_t memcached_io_read(memcached_instance_st *instance, void *buffer, size_t length,
                                     ssize_t &nread) {
  assert(memcached_is_udp(instance->root) == false);
  assert_msg(
      instance,
      "Programmer error, memcached_io_read() recieved an invalid Instance"); // Programmer error
  char *buffer_ptr = static_cast<char *>(buffer);

  if (instance->fd == INVALID_SOCKET) {
#if 0
    assert_msg(int(instance->state) <= int(MEMCACHED_SERVER_STATE_ADDRINFO), "Programmer error, invalid socket state");
#endif
    return MEMCACHED_CONNECTION_FAILURE;
  }

  while (length) {
    if (instance->read_buffer_length == 0) {
      memcached_return_t io_fill_ret;
      if (memcached_fatal(io_fill_ret = _io_fill(instance))) {
        nread = -1;
        return io_fill_ret;
      }
    }

    if (length > 1) {
      size_t difference =
          (length > instance->read_buffer_length) ? instance->read_buffer_length : length;

      memcpy(buffer_ptr, instance->read_ptr, difference);
      length -= difference;
      instance->read_ptr += difference;
      instance->read_buffer_length -= difference;
      buffer_ptr += difference;
    } else {
      *buffer_ptr = *instance->read_ptr;
      instance->read_ptr++;
      instance->read_buffer_length--;
      buffer_ptr++;
      break;
    }
  }

  nread = ssize_t(buffer_ptr - (char *) buffer);

  return MEMCACHED_SUCCESS;
}

memcached_return_t memcached_io_slurp(memcached_instance_st *instance) {
  assert_msg(instance, "Programmer error, invalid Instance");
  assert(memcached_is_udp(instance->root) == false);

  if (instance->fd == INVALID_SOCKET) {
    assert_msg(int(instance->state) <= int(MEMCACHED_SERVER_STATE_ADDRINFO),
               "Invalid socket state");
    return MEMCACHED_CONNECTION_FAILURE;
  }

  ssize_t data_read;
  char buffer[MEMCACHED_MAX_BUFFER];
  do {
    data_read = ::recv(instance->fd, instance->read_buffer, sizeof(buffer), MSG_NOSIGNAL);
    if (data_read == SOCKET_ERROR) {
      switch (get_socket_errno()) {
      case EINTR: // We just retry
        continue;

      case ETIMEDOUT: // OSX
#if EWOULDBLOCK != EAGAIN
      case EWOULDBLOCK:
#endif
      case EAGAIN:
#ifdef __linux
      case ERESTART:
#endif
        if (memcached_success(io_wait(instance, POLLIN))) {
          continue;
        }
        return MEMCACHED_IN_PROGRESS;

        /* fall through */

      case ENOTCONN: // Programmer Error
      case ENOTSOCK:
        assert(0);
        /* fall through */
      case EBADF:
        assert_msg(instance->fd != INVALID_SOCKET, "Invalid socket state");
        /* fall through */
      case EINVAL:
      case EFAULT:
      case ECONNREFUSED:
      default:
        return MEMCACHED_CONNECTION_FAILURE; // We want this!
      }
    }
  } while (data_read > 0);

  return MEMCACHED_CONNECTION_FAILURE;
}

static bool _io_write(memcached_instance_st *instance, const void *buffer, size_t length,
                      bool with_flush, size_t &written) {
  assert(instance->fd != INVALID_SOCKET);
  assert(memcached_is_udp(instance->root) == false);

  const char *buffer_ptr = static_cast<const char *>(buffer);

  const size_t original_length = length;

  while (length) {
    char *write_ptr;
    size_t buffer_end = MEMCACHED_MAX_BUFFER;
    size_t should_write = buffer_end - instance->write_buffer_offset;
    should_write = (should_write < length) ? should_write : length;

    write_ptr = instance->write_buffer + instance->write_buffer_offset;
    memcpy(write_ptr, buffer_ptr, should_write);
    instance->write_buffer_offset += should_write;
    buffer_ptr += should_write;
    length -= should_write;

    if (instance->write_buffer_offset == buffer_end) {
      WATCHPOINT_ASSERT(instance->fd != INVALID_SOCKET);

      memcached_return_t rc;
      if (io_flush(instance, with_flush, rc) == false) {
        written = original_length - length;
        return false;
      }
    }
  }

  if (with_flush) {
    memcached_return_t rc;
    WATCHPOINT_ASSERT(instance->fd != INVALID_SOCKET);
    if (io_flush(instance, with_flush, rc) == false) {
      written = original_length - length;
      return false;
    }
  }

  written = original_length - length;

  return true;
}

bool memcached_io_write(memcached_instance_st *instance) {
  size_t written;
  return _io_write(instance, NULL, 0, true, written);
}

ssize_t memcached_io_write(memcached_instance_st *instance, const void *buffer, const size_t length,
                           const bool with_flush) {
  size_t written;

  if (_io_write(instance, buffer, length, with_flush, written) == false) {
    return -1;
  }

  return ssize_t(written);
}

bool memcached_io_writev(memcached_instance_st *instance, libmemcached_io_vector_st vector[],
                         const size_t number_of, const bool with_flush) {
  ssize_t complete_total = 0;
  ssize_t total = 0;

  for (size_t x = 0; x < number_of; x++, vector++) {
    complete_total += vector->length;
    if (vector->length) {
      size_t written;
      if ((_io_write(instance, vector->buffer, vector->length, false, written)) == false) {
        return false;
      }
      total += written;
    }
  }

  if (with_flush) {
    if (memcached_io_write(instance) == false) {
      return false;
    }
  }

  return (complete_total == total);
}

void memcached_instance_st::start_close_socket() {
  if (fd != INVALID_SOCKET) {
    shutdown(fd, SHUT_WR);
    options.is_shutting_down = true;
  }
}

void memcached_instance_st::reset_socket() {
  if (fd != INVALID_SOCKET) {
    (void) closesocket(fd);
    fd = INVALID_SOCKET;
  }
}

void memcached_instance_st::close_socket() {
  if (fd != INVALID_SOCKET) {
    int shutdown_options = SHUT_RD;
    if (options.is_shutting_down == false) {
      shutdown_options = SHUT_RDWR;
    }

    /* in case of death shutdown to avoid blocking at close() */
    if (shutdown(fd, shutdown_options) == SOCKET_ERROR and get_socket_errno() != ENOTCONN) {
      WATCHPOINT_NUMBER(fd);
      WATCHPOINT_ERRNO(get_socket_errno());
      WATCHPOINT_ASSERT(get_socket_errno());
    }

    reset_socket();
    state = MEMCACHED_SERVER_STATE_NEW;
  }

  state = MEMCACHED_SERVER_STATE_NEW;
  cursor_active_ = 0;
  io_bytes_sent = 0;
  write_buffer_offset = size_t(root and memcached_is_udp(root) ? UDP_DATAGRAM_HEADER_LENGTH : 0);
  read_buffer_length = 0;
  read_ptr = read_buffer;
  options.is_shutting_down = false;
  memcached_server_response_reset(this);

  // We reset the version so that if we end up talking to a different server
  // we don't have stale server version information.
  major_version = minor_version = micro_version = UINT8_MAX;
}

memcached_instance_st *memcached_io_get_readable_server(Memcached *memc, memcached_return_t &) {
#define MAX_SERVERS_TO_POLL 100
  struct pollfd fds[MAX_SERVERS_TO_POLL];
  nfds_t host_index = 0;

  for (uint32_t x = 0; x < memcached_server_count(memc) and host_index < MAX_SERVERS_TO_POLL; ++x) {
    memcached_instance_st *instance = memcached_instance_fetch(memc, x);

    if (instance->read_buffer_length > 0) /* I have data in the buffer */ {
      return instance;
    }

    if (instance->response_count() > 0) {
      fds[host_index].events = POLLIN;
      fds[host_index].revents = 0;
      fds[host_index].fd = instance->fd;
      ++host_index;
    }
  }

  if (host_index < 2) {
    /* We have 0 or 1 server with pending events.. */
    for (uint32_t x = 0; x < memcached_server_count(memc); ++x) {
      memcached_instance_st *instance = memcached_instance_fetch(memc, x);

      if (instance->response_count() > 0) {
        return instance;
      }
    }

    return NULL;
  }

  int error = poll(fds, host_index, memc->poll_timeout);
  switch (error) {
  case -1:
    memcached_set_errno(*memc, get_socket_errno(), MEMCACHED_AT);
    /* FALLTHROUGH */
  case 0:
    break;

  default:
    for (nfds_t x = 0; x < host_index; ++x) {
      if (fds[x].revents & POLLIN) {
        for (uint32_t y = 0; y < memcached_server_count(memc); ++y) {
          memcached_instance_st *instance = memcached_instance_fetch(memc, y);

          if (instance->fd == fds[x].fd) {
            return instance;
          }
        }
      }
    }
  }

  return NULL;
}

/*
  Eventually we will just kill off the server with the problem.
*/
void memcached_io_reset(memcached_instance_st *instance) {
  memcached_quit_server(instance, true);
}

/**
 * Read a given number of bytes from the server and place it into a specific
 * buffer. Reset the IO channel on this server if an error occurs.
 */
memcached_return_t memcached_safe_read(memcached_instance_st *instance, void *dta,
                                       const size_t size) {
  size_t offset = 0;
  char *data = static_cast<char *>(dta);

  while (offset < size) {
    ssize_t nread;
    memcached_return_t rc;

    while (
        memcached_continue(rc = memcached_io_read(instance, data + offset, size - offset, nread))) {
    };

    if (memcached_failed(rc)) {
      return rc;
    }

    offset += size_t(nread);
  }

  return MEMCACHED_SUCCESS;
}

memcached_return_t memcached_io_readline(memcached_instance_st *instance, char *buffer_ptr,
                                         size_t size, size_t &total_nr) {
  total_nr = 0;
  bool line_complete = false;

  while (line_complete == false) {
    if (instance->read_buffer_length == 0) {
      /*
       * We don't have any data in the buffer, so let's fill the read
       * buffer. Call the standard read function to avoid duplicating
       * the logic.
       */
      ssize_t nread;
      memcached_return_t rc = memcached_io_read(instance, buffer_ptr, 1, nread);
      if (memcached_failed(rc) and rc == MEMCACHED_IN_PROGRESS) {
        memcached_quit_server(instance, true);
        return memcached_set_error(*instance, rc, MEMCACHED_AT);
      } else if (memcached_failed(rc)) {
        return rc;
      }

      if (*buffer_ptr == '\n') {
        line_complete = true;
      }

      ++buffer_ptr;
      ++total_nr;
    }

    /* Now let's look in the buffer and copy as we go! */
    while (instance->read_buffer_length and total_nr < size and line_complete == false) {
      *buffer_ptr = *instance->read_ptr;
      if (*buffer_ptr == '\n') {
        line_complete = true;
      }
      --instance->read_buffer_length;
      ++instance->read_ptr;
      ++total_nr;
      ++buffer_ptr;
    }

    if (total_nr == size) {
      return MEMCACHED_PROTOCOL_ERROR;
    }
  }

  return MEMCACHED_SUCCESS;
}