tcp_server_posix.c

/*
 *
 * Copyright 2015, Google Inc.
 * All rights reserved.
 *
 * Redistribution and use in source and binary forms, with or without
 * modification, are permitted provided that the following conditions are
 * met:
 *
 *     * Redistributions of source code must retain the above copyright
 * notice, this list of conditions and the following disclaimer.
 *     * Redistributions in binary form must reproduce the above
 * copyright notice, this list of conditions and the following disclaimer
 * in the documentation and/or other materials provided with the
 * distribution.
 *     * Neither the name of Google Inc. nor the names of its
 * contributors may be used to endorse or promote products derived from
 * this software without specific prior written permission.
 *
 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
 * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
 * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
 * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
 * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
 * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
 * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
 * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
 * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
 * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
 * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
 *
 */

/* FIXME: "posix" files shouldn't be depending on _GNU_SOURCE */
#ifndef _GNU_SOURCE
#define _GNU_SOURCE
#endif

#include <grpc/support/port_platform.h>

#ifdef GPR_POSIX_SOCKET

#include "src/core/iomgr/tcp_server.h"

#include <errno.h>
#include <fcntl.h>
#include <limits.h>
#include <netinet/in.h>
#include <netinet/tcp.h>
#include <stdio.h>
#include <string.h>
#include <sys/socket.h>
#include <sys/stat.h>
#include <sys/types.h>
#include <sys/un.h>
#include <unistd.h>

#include "src/core/iomgr/pollset_posix.h"
#include "src/core/iomgr/resolve_address.h"
#include "src/core/iomgr/sockaddr_utils.h"
#include "src/core/iomgr/socket_utils_posix.h"
#include "src/core/iomgr/tcp_posix.h"
#include "src/core/support/string.h"
#include <grpc/support/alloc.h>
#include <grpc/support/log.h>
#include <grpc/support/string_util.h>
#include <grpc/support/sync.h>
#include <grpc/support/time.h>

#define MIN_SAFE_ACCEPT_QUEUE_SIZE 100

static gpr_once s_init_max_accept_queue_size;
static int s_max_accept_queue_size;

/* one listening port */
struct grpc_tcp_listener {
  int fd;
  grpc_fd *emfd;
  grpc_tcp_server *server;
  union {
    gpr_uint8 untyped[GRPC_MAX_SOCKADDR_SIZE];
    struct sockaddr sockaddr;
    struct sockaddr_un un;
  } addr;
  size_t addr_len;
  int port;
  grpc_closure read_closure;
  grpc_closure destroyed_closure;
  gpr_refcount refs;
  struct grpc_tcp_listener *next;
  /* When we add a listener, more than one can be created, mainly because of
     IPv6. A sibling will still be in the normal list, but will be flagged
     as such. Any action, such as ref or unref, will affect all of the
     siblings in the list. */
  struct grpc_tcp_listener *sibling;
  int is_sibling;
};

static void unlink_if_unix_domain_socket(const struct sockaddr_un *un) {
  struct stat st;

  if (stat(un->sun_path, &st) == 0 && (st.st_mode & S_IFMT) == S_IFSOCK) {
    unlink(un->sun_path);
  }
}

/* the overall server */
struct grpc_tcp_server {
  /* Called whenever accept() succeeds on a server port. */
  grpc_tcp_server_cb on_accept_cb;
  void *on_accept_cb_arg;

  gpr_mu mu;

  /* active port count: how many ports are actually still listening */
  size_t active_ports;
  /* destroyed port count: how many ports are completely destroyed */
  size_t destroyed_ports;

  /* is this server shutting down? (boolean) */
  int shutdown;

  /* linked list of server ports */
  grpc_tcp_listener *head;
  unsigned nports;

  /* shutdown callback */
  grpc_closure *shutdown_complete;

  /* all pollsets interested in new connections */
  grpc_pollset **pollsets;
  /* number of pollsets in the pollsets array */
  size_t pollset_count;
};

grpc_tcp_server *grpc_tcp_server_create(void) {
  grpc_tcp_server *s = gpr_malloc(sizeof(grpc_tcp_server));
  gpr_mu_init(&s->mu);
  s->active_ports = 0;
  s->destroyed_ports = 0;
  s->shutdown = 0;
  s->on_accept_cb = NULL;
  s->on_accept_cb_arg = NULL;
  s->head = NULL;
  s->nports = 0;
  return s;
}

static void finish_shutdown(grpc_exec_ctx *exec_ctx, grpc_tcp_server *s) {
  grpc_exec_ctx_enqueue(exec_ctx, s->shutdown_complete, 1);

  gpr_mu_destroy(&s->mu);

  while (s->head) {
    grpc_tcp_listener *sp = s->head;
    s->head = sp->next;
    grpc_tcp_listener_unref(sp);
  }

  gpr_free(s);
}

static void destroyed_port(grpc_exec_ctx *exec_ctx, void *server, int success) {
  grpc_tcp_server *s = server;
  gpr_mu_lock(&s->mu);
  s->destroyed_ports++;
  if (s->destroyed_ports == s->nports) {
    gpr_mu_unlock(&s->mu);
    finish_shutdown(exec_ctx, s);
  } else {
    GPR_ASSERT(s->destroyed_ports < s->nports);
    gpr_mu_unlock(&s->mu);
  }
}

/* called when all listening endpoints have been shutdown, so no further
   events will be received on them - at this point it's safe to destroy
   things */
static void deactivated_all_ports(grpc_exec_ctx *exec_ctx, grpc_tcp_server *s) {
  /* delete ALL the things */
  gpr_mu_lock(&s->mu);

  if (!s->shutdown) {
    gpr_mu_unlock(&s->mu);
    return;
  }

  if (s->head) {
    grpc_tcp_listener *sp;
    for (sp = s->head; sp; sp = sp->next) {
      if (sp->addr.sockaddr.sa_family == AF_UNIX) {
        unlink_if_unix_domain_socket(&sp->addr.un);
      }
      sp->destroyed_closure.cb = destroyed_port;
      sp->destroyed_closure.cb_arg = s;
      grpc_fd_orphan(exec_ctx, sp->emfd, &sp->destroyed_closure, NULL,
                     "tcp_listener_shutdown");
    }
    gpr_mu_unlock(&s->mu);
  } else {
    gpr_mu_unlock(&s->mu);
    finish_shutdown(exec_ctx, s);
  }
}

void grpc_tcp_server_destroy(grpc_exec_ctx *exec_ctx, grpc_tcp_server *s,
                             grpc_closure *closure) {
  gpr_mu_lock(&s->mu);

  GPR_ASSERT(!s->shutdown);
  s->shutdown = 1;

  s->shutdown_complete = closure;

  /* shutdown all fd's */
  if (s->active_ports) {
    grpc_tcp_listener *sp;
    for (sp = s->head; sp; sp = sp->next) {
      grpc_fd_shutdown(exec_ctx, sp->emfd);
    }
    gpr_mu_unlock(&s->mu);
  } else {
    gpr_mu_unlock(&s->mu);
    deactivated_all_ports(exec_ctx, s);
  }
}

/* get max listen queue size on linux */
static void init_max_accept_queue_size(void) {
  int n = SOMAXCONN;
  char buf[64];
  FILE *fp = fopen("/proc/sys/net/core/somaxconn", "r");
  if (fp == NULL) {
    /* 2.4 kernel. */
    s_max_accept_queue_size = SOMAXCONN;
    return;
  }
  if (fgets(buf, sizeof buf, fp)) {
    char *end;
    long i = strtol(buf, &end, 10);
    if (i > 0 && i <= INT_MAX && end && *end == 0) {
      n = (int)i;
    }
  }
  fclose(fp);
  s_max_accept_queue_size = n;

  if (s_max_accept_queue_size < MIN_SAFE_ACCEPT_QUEUE_SIZE) {
    gpr_log(GPR_INFO,
            "Suspiciously small accept queue (%d) will probably lead to "
            "connection drops",
            s_max_accept_queue_size);
  }
}

static int get_max_accept_queue_size(void) {
  gpr_once_init(&s_init_max_accept_queue_size, init_max_accept_queue_size);
  return s_max_accept_queue_size;
}

/* Prepare a recently-created socket for listening. */
static int prepare_socket(int fd, const struct sockaddr *addr,
                          size_t addr_len) {
  struct sockaddr_storage sockname_temp;
  socklen_t sockname_len;

  if (fd < 0) {
    goto error;
  }

  if (!grpc_set_socket_nonblocking(fd, 1) || !grpc_set_socket_cloexec(fd, 1) ||
      (addr->sa_family != AF_UNIX && (!grpc_set_socket_low_latency(fd, 1) ||
                                      !grpc_set_socket_reuse_addr(fd, 1))) ||
      !grpc_set_socket_no_sigpipe_if_possible(fd)) {
    gpr_log(GPR_ERROR, "Unable to configure socket %d: %s", fd,
            strerror(errno));
    goto error;
  }

  GPR_ASSERT(addr_len < ~(socklen_t)0);
  if (bind(fd, addr, (socklen_t)addr_len) < 0) {
    char *addr_str;
    grpc_sockaddr_to_string(&addr_str, addr, 0);
    gpr_log(GPR_ERROR, "bind addr=%s: %s", addr_str, strerror(errno));
    gpr_free(addr_str);
    goto error;
  }

  if (listen(fd, get_max_accept_queue_size()) < 0) {
    gpr_log(GPR_ERROR, "listen: %s", strerror(errno));
    goto error;
  }

  sockname_len = sizeof(sockname_temp);
  if (getsockname(fd, (struct sockaddr *)&sockname_temp, &sockname_len) < 0) {
    goto error;
  }

  return grpc_sockaddr_get_port((struct sockaddr *)&sockname_temp);

error:
  if (fd >= 0) {
    close(fd);
  }
  return -1;
}

/* event manager callback when reads are ready */
static void on_read(grpc_exec_ctx *exec_ctx, void *arg, int success) {
  grpc_tcp_listener *sp = arg;
  grpc_fd *fdobj;
  size_t i;

  if (!success) {
    goto error;
  }

  /* loop until accept4 returns EAGAIN, and then re-arm notification */
  for (;;) {
    struct sockaddr_storage addr;
    socklen_t addrlen = sizeof(addr);
    char *addr_str;
    char *name;
    /* Note: If we ever decide to return this address to the user, remember to
       strip off the ::ffff:0.0.0.0/96 prefix first. */
    int fd = grpc_accept4(sp->fd, (struct sockaddr *)&addr, &addrlen, 1, 1);
    if (fd < 0) {
      switch (errno) {
        case EINTR:
          continue;
        case EAGAIN:
          grpc_fd_notify_on_read(exec_ctx, sp->emfd, &sp->read_closure);
          return;
        default:
          gpr_log(GPR_ERROR, "Failed accept4: %s", strerror(errno));
          goto error;
      }
    }

    grpc_set_socket_no_sigpipe_if_possible(fd);

    addr_str = grpc_sockaddr_to_uri((struct sockaddr *)&addr);
    gpr_asprintf(&name, "tcp-server-connection:%s", addr_str);

    if (grpc_tcp_trace) {
      gpr_log(GPR_DEBUG, "SERVER_CONNECT: incoming connection: %s", addr_str);
    }

    fdobj = grpc_fd_create(fd, name);
    /* TODO(ctiller): revise this when we have server-side sharding
       of channels -- we certainly should not be automatically adding every
       incoming channel to every pollset owned by the server */
    for (i = 0; i < sp->server->pollset_count; i++) {
      grpc_pollset_add_fd(exec_ctx, sp->server->pollsets[i], fdobj);
    }
    sp->server->on_accept_cb(
        exec_ctx, sp->server->on_accept_cb_arg,
        grpc_tcp_create(fdobj, GRPC_TCP_DEFAULT_READ_SLICE_SIZE, addr_str));

    gpr_free(name);
    gpr_free(addr_str);
  }

  GPR_UNREACHABLE_CODE(return );

error:
  gpr_mu_lock(&sp->server->mu);
  if (0 == --sp->server->active_ports) {
    gpr_mu_unlock(&sp->server->mu);
    deactivated_all_ports(exec_ctx, sp->server);
  } else {
    gpr_mu_unlock(&sp->server->mu);
  }
}

static grpc_tcp_listener *add_socket_to_server(grpc_tcp_server *s, int fd,
                                               const struct sockaddr *addr,
                                               size_t addr_len) {
  grpc_tcp_listener *sp = NULL;
  int port;
  char *addr_str;
  char *name;

  port = prepare_socket(fd, addr, addr_len);
  if (port >= 0) {
    grpc_sockaddr_to_string(&addr_str, (struct sockaddr *)&addr, 1);
    gpr_asprintf(&name, "tcp-server-listener:%s", addr_str);
    gpr_mu_lock(&s->mu);
    s->nports++;
    GPR_ASSERT(!s->on_accept_cb && "must add ports before starting server");
    sp = gpr_malloc(sizeof(grpc_tcp_listener));
    sp->next = s->head;
    s->head = sp;
    sp->server = s;
    sp->fd = fd;
    sp->emfd = grpc_fd_create(fd, name);
    memcpy(sp->addr.untyped, addr, addr_len);
    sp->addr_len = addr_len;
    sp->port = port;
    sp->is_sibling = 0;
    sp->sibling = NULL;
    gpr_ref_init(&sp->refs, 1);
    GPR_ASSERT(sp->emfd);
    gpr_mu_unlock(&s->mu);
    gpr_free(addr_str);
    gpr_free(name);
  }

  return sp;
}

grpc_tcp_listener *grpc_tcp_server_add_port(grpc_tcp_server *s,
                                            const void *addr, size_t addr_len) {
  grpc_tcp_listener *sp;
  grpc_tcp_listener *sp2 = NULL;
  int fd;
  grpc_dualstack_mode dsmode;
  struct sockaddr_in6 addr6_v4mapped;
  struct sockaddr_in wild4;
  struct sockaddr_in6 wild6;
  struct sockaddr_in addr4_copy;
  struct sockaddr *allocated_addr = NULL;
  struct sockaddr_storage sockname_temp;
  socklen_t sockname_len;
  int port;

  if (((struct sockaddr *)addr)->sa_family == AF_UNIX) {
    unlink_if_unix_domain_socket(addr);
  }

  /* Check if this is a wildcard port, and if so, try to keep the port the same
     as some previously created listener. */
  if (grpc_sockaddr_get_port(addr) == 0) {
    for (sp = s->head; sp; sp = sp->next) {
      sockname_len = sizeof(sockname_temp);
      if (0 == getsockname(sp->fd, (struct sockaddr *)&sockname_temp,
                           &sockname_len)) {
        port = grpc_sockaddr_get_port((struct sockaddr *)&sockname_temp);
        if (port > 0) {
          allocated_addr = malloc(addr_len);
          memcpy(allocated_addr, addr, addr_len);
          grpc_sockaddr_set_port(allocated_addr, port);
          addr = allocated_addr;
          break;
        }
      }
    }
  }

  sp = NULL;

  if (grpc_sockaddr_to_v4mapped(addr, &addr6_v4mapped)) {
    addr = (const struct sockaddr *)&addr6_v4mapped;
    addr_len = sizeof(addr6_v4mapped);
  }

  /* Treat :: or 0.0.0.0 as a family-agnostic wildcard. */
  if (grpc_sockaddr_is_wildcard(addr, &port)) {
    grpc_sockaddr_make_wildcards(port, &wild4, &wild6);

    /* Try listening on IPv6 first. */
    addr = (struct sockaddr *)&wild6;
    addr_len = sizeof(wild6);
    fd = grpc_create_dualstack_socket(addr, SOCK_STREAM, 0, &dsmode);
    sp = add_socket_to_server(s, fd, addr, addr_len);
    if (fd >= 0 && dsmode == GRPC_DSMODE_DUALSTACK) {
      goto done;
    }

    /* If we didn't get a dualstack socket, also listen on 0.0.0.0. */
    if (port == 0 && sp != NULL) {
      grpc_sockaddr_set_port((struct sockaddr *)&wild4, sp->port);
      sp2 = sp;
    }
    addr = (struct sockaddr *)&wild4;
    addr_len = sizeof(wild4);
  }

  fd = grpc_create_dualstack_socket(addr, SOCK_STREAM, 0, &dsmode);
  if (fd < 0) {
    gpr_log(GPR_ERROR, "Unable to create socket: %s", strerror(errno));
  }
  if (dsmode == GRPC_DSMODE_IPV4 &&
      grpc_sockaddr_is_v4mapped(addr, &addr4_copy)) {
    addr = (struct sockaddr *)&addr4_copy;
    addr_len = sizeof(addr4_copy);
  }
  sp = add_socket_to_server(s, fd, addr, addr_len);
  if (sp != NULL) sp->sibling = sp2;
  if (sp2 != NULL) sp2->is_sibling = 1;

done:
  gpr_free(allocated_addr);
  return sp;
}

int grpc_tcp_server_get_fd(grpc_tcp_server *s, unsigned port_index) {
  grpc_tcp_listener *sp;
  for (sp = s->head; sp && port_index != 0; sp = sp->next, port_index--)
    ;
  if (port_index == 0 && sp) {
    return sp->fd;
  } else {
    return -1;
  }
}

void grpc_tcp_server_start(grpc_exec_ctx *exec_ctx, grpc_tcp_server *s,
                           grpc_pollset **pollsets, size_t pollset_count,
                           grpc_tcp_server_cb on_accept_cb,
                           void *on_accept_cb_arg) {
  size_t i;
  grpc_tcp_listener *sp;
  GPR_ASSERT(on_accept_cb);
  gpr_mu_lock(&s->mu);
  GPR_ASSERT(!s->on_accept_cb);
  GPR_ASSERT(s->active_ports == 0);
  s->on_accept_cb = on_accept_cb;
  s->on_accept_cb_arg = on_accept_cb_arg;
  s->pollsets = pollsets;
  s->pollset_count = pollset_count;
  for (sp = s->head; sp; sp = sp->next) {
    for (i = 0; i < pollset_count; i++) {
      grpc_pollset_add_fd(exec_ctx, pollsets[i], sp->emfd);
    }
    sp->read_closure.cb = on_read;
    sp->read_closure.cb_arg = sp;
    grpc_fd_notify_on_read(exec_ctx, sp->emfd, &sp->read_closure);
    s->active_ports++;
  }
  gpr_mu_unlock(&s->mu);
}

int grpc_tcp_listener_get_port(grpc_tcp_listener *listener) {
  grpc_tcp_listener *sp = listener;
  return sp->port;
}

void grpc_tcp_listener_ref(grpc_tcp_listener *listener) {
  grpc_tcp_listener *sp = listener;
  gpr_ref(&sp->refs);
}

void grpc_tcp_listener_unref(grpc_tcp_listener *listener) {
  grpc_tcp_listener *sp = listener;
  if (sp->is_sibling) return;
  if (gpr_unref(&sp->refs)) {
    grpc_tcp_listener *sibling = sp->sibling;
    while (sibling) {
      sp = sibling;
      sibling = sp->sibling;
      gpr_free(sp);
    }
    gpr_free(listener);
  }
}

#endif