David van Moolenbroek 27852ebe53 UDS: full rewrite
This new implementation of the UDS service is built on top of the
libsockevent library.  It thereby inherits all the advantages that
libsockevent brings.  However, the fundamental restructuring
required for that change also paved the way for resolution of a
number of other important open issues with the old UDS code.  Most
importantly, the rewrite brings the behavior of the service much
closer to POSIX compliance and NetBSD compatibility.  These are the
most important changes:

- due to the use of libsockevent, UDS now supports multiple suspending
  calls per socket and a large number of standard socket flags and
  options;
- socket address matching is now based on <device,inode> lookups
  instead of canonized path names, and socket addresses are no longer
  altered either due to canonization or at connect time;
- the socket state machine is now well defined, most importantly
  resolving the erroneous reset-on-EOF semantics of the old UDS, but
  also allowing socket reuse;
- sockets are now connected before being accepted instead of being
  held in connecting state, unless the LOCAL_CONNWAIT option is set
  on either the connecting or the listening socket;
- connect(2) on datagram sockets is now supported (needed by syslog),
  and proper datagram socket disconnect notification is provided;
- the receive queue now supports segmentation, associating ancillary
  data (in-flight file descriptors and credentials) with each segment
  instead of being kept fully separately; this is a POSIX requirement
  (and needed by tmux);
- as part of the segmentation support, the receive queue can now hold
  as many packets as can fit, instead of one;
- in addition to the flags supported by libsockevent, the MSG_PEEK,
  MSG_WAITALL, MSG_CMSG_CLOEXEC, MSG_TRUNC, and MSG_CTRUNC send and
  receive flags are now supported;
- the SO_PASSCRED and SO_PEERCRED socket options are replaced by
  LOCAL_CREDS and LOCAL_PEEREID respectively, now following NetBSD
  semantics and allowing use of NetBSD libc's getpeereid(3);
- memory usage is reduced by about 250 KB due to centralized in-flight
  file descriptor tracking, with a limit of OPEN_MAX total rather than
  of OPEN_MAX per socket;
- memory usage is reduced by another ~50 KB due to removal of state
  redundancy, despite the fact that socket path names may now be up to
  253 bytes rather than the previous 104 bytes;
- compared to the old UDS, there is now very little direct indexing on
  the static array of sockets, thus allowing dynamic allocation of
  sockets more easily in the future;
- the UDS service now has RMIB support for the net.local sysctl tree,
  implementing preliminary support for NetBSD netstat(1).

Change-Id: I4a9b6fe4aaeef0edf2547eee894e6c14403fcb32
2017-03-09 23:39:56 +00:00

549 lines
12 KiB
C

/* Id */
/*
* Copyright (c) 2007 Nicholas Marriott <nicm@users.sourceforge.net>
*
* Permission to use, copy, modify, and distribute this software for any
* purpose with or without fee is hereby granted, provided that the above
* copyright notice and this permission notice appear in all copies.
*
* THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
* WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
* MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
* ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
* WHATSOEVER RESULTING FROM LOSS OF MIND, USE, DATA OR PROFITS, WHETHER
* IN AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING
* OUT OF OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
*/
#include <sys/types.h>
#include <sys/ioctl.h>
#include <sys/socket.h>
#include <sys/stat.h>
#include <sys/un.h>
#include <sys/wait.h>
#include <errno.h>
#include <event.h>
#include <fcntl.h>
#include <signal.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <syslog.h>
#include <termios.h>
#include <time.h>
#include <unistd.h>
#include "tmux.h"
/*
* Main server functions.
*/
/* Client list. */
struct clients clients;
struct clients dead_clients;
int server_fd;
int server_shutdown;
struct event server_ev_accept;
struct event server_ev_second;
struct paste_stack global_buffers;
int server_create_socket(void);
void server_loop(void);
int server_should_shutdown(void);
void server_send_shutdown(void);
void server_clean_dead(void);
void server_accept_callback(int, short, void *);
void server_signal_callback(int, short, void *);
void server_child_signal(void);
void server_child_exited(pid_t, int);
void server_child_stopped(pid_t, int);
void server_second_callback(int, short, void *);
void server_lock_server(void);
void server_lock_sessions(void);
/* Create server socket. */
int
server_create_socket(void)
{
struct sockaddr_un sa;
size_t size;
mode_t mask;
int fd;
memset(&sa, 0, sizeof sa);
sa.sun_family = AF_UNIX;
size = strlcpy(sa.sun_path, socket_path, sizeof sa.sun_path);
if (size >= sizeof sa.sun_path) {
errno = ENAMETOOLONG;
fatal("socket failed");
}
unlink(sa.sun_path);
if ((fd = socket(AF_UNIX, SOCK_STREAM, 0)) == -1)
fatal("socket failed");
mask = umask(S_IXUSR|S_IXGRP|S_IRWXO);
if (bind(fd, (struct sockaddr *) &sa, SUN_LEN(&sa)) == -1)
fatal("bind failed");
umask(mask);
if (listen(fd, 16) == -1)
fatal("listen failed");
setblocking(fd, 0);
server_update_socket();
return (fd);
}
/* Fork new server. */
int
server_start(int lockfd, char *lockfile)
{
int pair[2];
struct timeval tv;
char *cause;
/* The first client is special and gets a socketpair; create it. */
if (socketpair(AF_UNIX, SOCK_STREAM, PF_UNSPEC, pair) != 0)
fatal("socketpair failed");
switch (fork()) {
case -1:
fatal("fork failed");
case 0:
break;
default:
close(pair[1]);
return (pair[0]);
}
close(pair[0]);
/*
* Must daemonise before loading configuration as the PID changes so
* $TMUX would be wrong for sessions created in the config file.
*/
if (daemon(1, 0) != 0)
fatal("daemon failed");
/* event_init() was called in our parent, need to reinit. */
if (event_reinit(ev_base) != 0)
fatal("event_reinit failed");
clear_signals(0);
logfile("server");
log_debug("server started, pid %ld", (long) getpid());
ARRAY_INIT(&windows);
RB_INIT(&all_window_panes);
ARRAY_INIT(&clients);
ARRAY_INIT(&dead_clients);
RB_INIT(&sessions);
RB_INIT(&dead_sessions);
TAILQ_INIT(&session_groups);
ARRAY_INIT(&global_buffers);
mode_key_init_trees();
key_bindings_init();
utf8_build();
start_time = time(NULL);
log_debug("socket path %s", socket_path);
#ifdef HAVE_SETPROCTITLE
setproctitle("server (%s)", socket_path);
#endif
server_fd = server_create_socket();
server_client_create(pair[1]);
unlink(lockfile);
free(lockfile);
close(lockfd);
cfg_cmd_q = cmdq_new(NULL);
cfg_cmd_q->emptyfn = cfg_default_done;
cfg_finished = 0;
cfg_references = 1;
ARRAY_INIT(&cfg_causes);
cfg_client = ARRAY_FIRST(&clients);
if (cfg_client != NULL)
cfg_client->references++;
if (access(TMUX_CONF, R_OK) == 0) {
if (load_cfg(TMUX_CONF, cfg_cmd_q, &cause) == -1) {
xasprintf(&cause, "%s: %s", TMUX_CONF, cause);
ARRAY_ADD(&cfg_causes, cause);
}
} else if (errno != ENOENT) {
xasprintf(&cause, "%s: %s", TMUX_CONF, strerror(errno));
ARRAY_ADD(&cfg_causes, cause);
}
if (cfg_file != NULL) {
if (load_cfg(cfg_file, cfg_cmd_q, &cause) == -1) {
xasprintf(&cause, "%s: %s", cfg_file, cause);
ARRAY_ADD(&cfg_causes, cause);
}
}
cmdq_continue(cfg_cmd_q);
server_add_accept(0);
memset(&tv, 0, sizeof tv);
tv.tv_sec = 1;
evtimer_set(&server_ev_second, server_second_callback, NULL);
evtimer_add(&server_ev_second, &tv);
set_signals(server_signal_callback);
server_loop();
exit(0);
}
/* Main server loop. */
void
server_loop(void)
{
while (!server_should_shutdown()) {
event_loop(EVLOOP_ONCE);
server_window_loop();
server_client_loop();
key_bindings_clean();
server_clean_dead();
}
}
/* Check if the server should be shutting down (no more clients or sessions). */
int
server_should_shutdown(void)
{
u_int i;
if (!options_get_number(&global_options, "exit-unattached")) {
if (!RB_EMPTY(&sessions))
return (0);
}
for (i = 0; i < ARRAY_LENGTH(&clients); i++) {
if (ARRAY_ITEM(&clients, i) != NULL)
return (0);
}
return (1);
}
/* Shutdown the server by killing all clients and windows. */
void
server_send_shutdown(void)
{
struct client *c;
struct session *s, *next_s;
u_int i;
for (i = 0; i < ARRAY_LENGTH(&clients); i++) {
c = ARRAY_ITEM(&clients, i);
if (c != NULL) {
if (c->flags & (CLIENT_BAD|CLIENT_SUSPENDED))
server_client_lost(c);
else
server_write_client(c, MSG_SHUTDOWN, NULL, 0);
c->session = NULL;
}
}
s = RB_MIN(sessions, &sessions);
while (s != NULL) {
next_s = RB_NEXT(sessions, &sessions, s);
session_destroy(s);
s = next_s;
}
}
/* Free dead, unreferenced clients and sessions. */
void
server_clean_dead(void)
{
struct session *s, *next_s;
struct client *c;
u_int i;
s = RB_MIN(sessions, &dead_sessions);
while (s != NULL) {
next_s = RB_NEXT(sessions, &dead_sessions, s);
if (s->references == 0) {
RB_REMOVE(sessions, &dead_sessions, s);
free(s->name);
free(s);
}
s = next_s;
}
for (i = 0; i < ARRAY_LENGTH(&dead_clients); i++) {
c = ARRAY_ITEM(&dead_clients, i);
if (c == NULL || c->references != 0)
continue;
ARRAY_SET(&dead_clients, i, NULL);
free(c);
}
}
/* Update socket execute permissions based on whether sessions are attached. */
void
server_update_socket(void)
{
struct session *s;
static int last = -1;
int n, mode;
struct stat sb;
n = 0;
RB_FOREACH(s, sessions, &sessions) {
if (!(s->flags & SESSION_UNATTACHED)) {
n++;
break;
}
}
if (n != last) {
last = n;
if (stat(socket_path, &sb) != 0)
return;
mode = sb.st_mode;
if (n != 0) {
if (mode & S_IRUSR)
mode |= S_IXUSR;
if (mode & S_IRGRP)
mode |= S_IXGRP;
if (mode & S_IROTH)
mode |= S_IXOTH;
} else
mode &= ~(S_IXUSR|S_IXGRP|S_IXOTH);
chmod(socket_path, mode);
}
}
/* Callback for server socket. */
void
server_accept_callback(int fd, short events, unused void *data)
{
struct sockaddr_storage sa;
socklen_t slen = sizeof sa;
int newfd;
server_add_accept(0);
if (!(events & EV_READ))
return;
newfd = accept(fd, (struct sockaddr *) &sa, &slen);
if (newfd == -1) {
if (errno == EAGAIN || errno == EINTR || errno == ECONNABORTED)
return;
if (errno == ENFILE || errno == EMFILE) {
/* Delete and don't try again for 1 second. */
server_add_accept(1);
return;
}
fatal("accept failed");
}
if (server_shutdown) {
close(newfd);
return;
}
server_client_create(newfd);
}
/*
* Add accept event. If timeout is nonzero, add as a timeout instead of a read
* event - used to backoff when running out of file descriptors.
*/
void
server_add_accept(int timeout)
{
struct timeval tv = { timeout, 0 };
if (event_initialized(&server_ev_accept))
event_del(&server_ev_accept);
if (timeout == 0) {
event_set(&server_ev_accept,
server_fd, EV_READ, server_accept_callback, NULL);
event_add(&server_ev_accept, NULL);
} else {
event_set(&server_ev_accept,
server_fd, EV_TIMEOUT, server_accept_callback, NULL);
event_add(&server_ev_accept, &tv);
}
}
/* Signal handler. */
void
server_signal_callback(int sig, unused short events, unused void *data)
{
switch (sig) {
case SIGTERM:
server_shutdown = 1;
server_send_shutdown();
break;
case SIGCHLD:
server_child_signal();
break;
case SIGUSR1:
event_del(&server_ev_accept);
close(server_fd);
server_fd = server_create_socket();
server_add_accept(0);
break;
}
}
/* Handle SIGCHLD. */
void
server_child_signal(void)
{
int status;
pid_t pid;
for (;;) {
switch (pid = waitpid(WAIT_ANY, &status, WNOHANG|WUNTRACED)) {
case -1:
if (errno == ECHILD)
return;
fatal("waitpid failed");
case 0:
return;
}
if (WIFSTOPPED(status))
server_child_stopped(pid, status);
else if (WIFEXITED(status) || WIFSIGNALED(status))
server_child_exited(pid, status);
}
}
/* Handle exited children. */
void
server_child_exited(pid_t pid, int status)
{
struct window *w;
struct window_pane *wp;
struct job *job;
u_int i;
for (i = 0; i < ARRAY_LENGTH(&windows); i++) {
if ((w = ARRAY_ITEM(&windows, i)) == NULL)
continue;
TAILQ_FOREACH(wp, &w->panes, entry) {
if (wp->pid == pid) {
server_destroy_pane(wp);
break;
}
}
}
LIST_FOREACH(job, &all_jobs, lentry) {
if (pid == job->pid) {
job_died(job, status); /* might free job */
break;
}
}
}
/* Handle stopped children. */
void
server_child_stopped(pid_t pid, int status)
{
struct window *w;
struct window_pane *wp;
u_int i;
if (WSTOPSIG(status) == SIGTTIN || WSTOPSIG(status) == SIGTTOU)
return;
for (i = 0; i < ARRAY_LENGTH(&windows); i++) {
if ((w = ARRAY_ITEM(&windows, i)) == NULL)
continue;
TAILQ_FOREACH(wp, &w->panes, entry) {
if (wp->pid == pid) {
if (killpg(pid, SIGCONT) != 0)
kill(pid, SIGCONT);
}
}
}
}
/* Handle once-per-second timer events. */
void
server_second_callback(unused int fd, unused short events, unused void *arg)
{
struct window *w;
struct window_pane *wp;
struct timeval tv;
u_int i;
if (options_get_number(&global_s_options, "lock-server"))
server_lock_server();
else
server_lock_sessions();
for (i = 0; i < ARRAY_LENGTH(&windows); i++) {
w = ARRAY_ITEM(&windows, i);
if (w == NULL)
continue;
TAILQ_FOREACH(wp, &w->panes, entry) {
if (wp->mode != NULL && wp->mode->timer != NULL)
wp->mode->timer(wp);
}
}
server_client_status_timer();
evtimer_del(&server_ev_second);
memset(&tv, 0, sizeof tv);
tv.tv_sec = 1;
evtimer_add(&server_ev_second, &tv);
}
/* Lock the server if ALL sessions have hit the time limit. */
void
server_lock_server(void)
{
struct session *s;
int timeout;
time_t t;
t = time(NULL);
RB_FOREACH(s, sessions, &sessions) {
if (s->flags & SESSION_UNATTACHED)
continue;
timeout = options_get_number(&s->options, "lock-after-time");
if (timeout <= 0 || t <= s->activity_time.tv_sec + timeout)
return; /* not timed out */
}
server_lock();
recalculate_sizes();
}
/* Lock any sessions which have timed out. */
void
server_lock_sessions(void)
{
struct session *s;
int timeout;
time_t t;
t = time(NULL);
RB_FOREACH(s, sessions, &sessions) {
if (s->flags & SESSION_UNATTACHED)
continue;
timeout = options_get_number(&s->options, "lock-after-time");
if (timeout > 0 && t > s->activity_time.tv_sec + timeout) {
server_lock_session(s);
recalculate_sizes();
}
}
}