jail: add support for userns and cgroupsns
authorDaniel Golle <daniel@makrotopia.org>
Fri, 20 Mar 2020 18:20:51 +0000 (18:20 +0000)
committerDaniel Golle <daniel@makrotopia.org>
Thu, 9 Apr 2020 11:45:29 +0000 (12:45 +0100)
Add options to have jailed process inside new user namespace and
cgroups namespace.
Currently only the root user inside the container is mapped.
Also, mounting /proc currently still fails in the new user namespace
with permission denied for unknown reasons.

Signed-off-by: Daniel Golle <daniel@makrotopia.org>
jail/jail.c
service/instance.c
service/instance.h

index df1bda7666e3535d47c53a6ebfbb00949f84f5fb..78f0c6bbb6ceea6dcb150044df69fc0ed6036ef3 100644 (file)
 #include <libubus.h>
 
 #define STACK_SIZE     (1024 * 1024)
-#define OPT_ARGS       "S:C:n:h:r:w:d:psulocU:G:NR:"
-
-#define NAMESPACE_MOUNT                (1U << 0)
-#define NAMESPACE_IPC          (1U << 1)
-#define NAMESPACE_NET          (1U << 2)
-#define NAMESPACE_PID          (1U << 3)
-#define NAMESPACE_USER         (1U << 4)
-#define NAMESPACE_UTS          (1U << 5)
-#define NAMESPACE_CGROUP       (1U << 6)
+#define OPT_ARGS       "S:C:n:h:r:w:d:psulocU:G:NR:fF"
 
 static struct {
        char *name;
@@ -64,8 +56,12 @@ static struct {
        int procfs;
        int ronly;
        int sysfs;
+       int pw_uid;
+       int pw_gid;
+       int gr_gid;
 } opts;
 
+
 extern int pivot_root(const char *new_root, const char *put_old);
 
 int debug = 0;
@@ -166,7 +162,7 @@ static int build_jail_fs(void)
        }
 
        if (opts.extroot) {
-               if (mount(opts.extroot, jail_root, NULL, MS_BIND | MS_REC, NULL)) {
+               if (mount(opts.extroot, jail_root, NULL, MS_BIND, NULL)) {
                        ERROR("extroot mount failed %m\n");
                        return -1;
                }
@@ -187,7 +183,7 @@ static int build_jail_fs(void)
                return -1;
        }
 
-       if (opts.namespace & NAMESPACE_NET) {
+       if (opts.namespace & CLONE_NEWNET) {
                char hostdir[PATH_MAX], jailetc[PATH_MAX], jaillink[PATH_MAX];
 
                snprintf(hostdir, PATH_MAX, "/tmp/resolv.conf-%s.d", opts.name);
@@ -231,6 +227,99 @@ static int build_jail_fs(void)
        return 0;
 }
 
+static int write_uid_gid_map(pid_t child_pid, bool gidmap, int id)
+{
+       int map_file;
+       char map_path[64];
+       const char *map_format = "%d %d %d\n";
+       if (snprintf(map_path, sizeof(map_path), "/proc/%d/%s",
+               child_pid, gidmap?"gid_map":"uid_map") < 0)
+               return -1;
+
+       if ((map_file = open(map_path, O_WRONLY)) == -1)
+               return -1;
+
+       if (dprintf(map_file, map_format, 0, id, 1) == -1) {
+               close(map_file);
+               return -1;
+       }
+
+       close(map_file);
+       return 0;
+}
+
+static int write_setgroups(pid_t child_pid, bool allow)
+{
+       int setgroups_file;
+       char setgroups_path[64];
+
+       if (snprintf(setgroups_path, sizeof(setgroups_path), "/proc/%d/setgroups",
+               child_pid) < 0) {
+               return -1;
+       }
+
+       if ((setgroups_file = open(setgroups_path, O_WRONLY)) == -1) {
+               return -1;
+       }
+
+       if (dprintf(setgroups_file, allow?"allow":"deny") == -1) {
+               close(setgroups_file);
+               return -1;
+       }
+
+       close(setgroups_file);
+       return 0;
+}
+
+static void get_jail_user(int *user, int *user_gid, int *gr_gid)
+{
+       struct passwd *p = NULL;
+       struct group *g = NULL;
+
+       if (opts.user) {
+               p = getpwnam(opts.user);
+               if (!p) {
+                       ERROR("failed to get uid/gid for user %s: %d (%s)\n",
+                             opts.user, errno, strerror(errno));
+                       exit(EXIT_FAILURE);
+               }
+               *user = p->pw_uid;
+               *user_gid = p->pw_gid;
+       } else {
+               *user = -1;
+               *user_gid = -1;
+       }
+
+       if (opts.group) {
+               g = getgrnam(opts.group);
+               if (!g) {
+                       ERROR("failed to get gid for group %s: %m\n", opts.group);
+                       exit(EXIT_FAILURE);
+               }
+               *gr_gid = g->gr_gid;
+       } else {
+               *gr_gid = -1;
+       }
+};
+
+static void set_jail_user(int pw_uid, int user_gid, int gr_gid)
+{
+       if ((user_gid != -1) && initgroups(opts.user, user_gid)) {
+               ERROR("failed to initgroups() for user %s: %m\n", opts.user);
+               exit(EXIT_FAILURE);
+       }
+
+       if ((gr_gid != -1) && setregid(gr_gid, gr_gid)) {
+               ERROR("failed to set group id %d: %m\n", gr_gid);
+               exit(EXIT_FAILURE);
+       }
+
+       if ((pw_uid != -1) && setreuid(pw_uid, pw_uid)) {
+               ERROR("failed to set user id %d: %m\n", pw_uid);
+               exit(EXIT_FAILURE);
+       }
+}
+
 #define MAX_ENVP       8
 static char** build_envp(const char *seccomp)
 {
@@ -253,8 +342,7 @@ static char** build_envp(const char *seccomp)
                envp[count++] = preload_var;
        }
 
-       if (is_extroot)
-               envp[count++] = container_var;
+       envp[count++] = container_var;
 
        if (debug > 1)
                envp[count++] = debug_var;
@@ -292,62 +380,71 @@ ujail will not use namespace/build a jail,\n\
 and will only drop capabilities/apply seccomp filter.\n\n");
 }
 
-static int exec_jail(void *_notused)
+static int exec_jail(void *pipes_ptr)
 {
-       struct passwd *p = NULL;
-       struct group *g = NULL;
+       int *pipes = (int*)pipes_ptr;
+       char buf[1];
+       int pw_uid, pw_gid, gr_gid;
 
-       if (opts.capabilities && drop_capabilities(opts.capabilities))
-               exit(EXIT_FAILURE);
+       close(pipes[0]);
+       close(pipes[3]);
 
-       if (opts.no_new_privs && prctl(PR_SET_NO_NEW_PRIVS, 1, 0, 0, 0)) {
-                ERROR("prctl(PR_SET_NO_NEW_PRIVS) failed: %m\n");
+
+       buf[0] = 'i';
+       if (write(pipes[1], buf, 1) < 1) {
+               ERROR("can't write to parent\n");
                exit(EXIT_FAILURE);
        }
-
-       if (opts.namespace && opts.hostname && strlen(opts.hostname) > 0
-                       && sethostname(opts.hostname, strlen(opts.hostname))) {
-               ERROR("sethostname(%s) failed: %m\n", opts.hostname);
+       if (read(pipes[2], buf, 1) < 1) {
+               ERROR("can't read from parent\n");
                exit(EXIT_FAILURE);
        }
-
-       if (opts.namespace && build_jail_fs()) {
-               ERROR("failed to build jail fs\n");
+       if (buf[0] != 'O') {
+               ERROR("parent had an error, child exiting\n");
                exit(EXIT_FAILURE);
        }
 
-       if (opts.user) {
-               p = getpwnam(opts.user);
-               if (!p) {
-                       ERROR("failed to get uid/gid for user %s: %d (%s)\n",
-                             opts.user, errno, strerror(errno));
+       close(pipes[1]);
+       close(pipes[2]);
+
+       if (opts.namespace & CLONE_NEWUSER) {
+               if (setgid(0) < 0) {
+                       ERROR("setgid\n");
                        exit(EXIT_FAILURE);
                }
-       }
-
-       if (opts.group) {
-               g = getgrnam(opts.group);
-               if (!g) {
-                       ERROR("failed to get gid for group %s: %m\n", opts.group);
+               if (setuid(0) < 0) {
+                       ERROR("setuid\n");
                        exit(EXIT_FAILURE);
                }
+//             if (setgroups(0, NULL) < 0) {
+//                     ERROR("setgroups\n");
+//                     exit(EXIT_FAILURE);
+//             }
        }
 
-       if (p && p->pw_gid && initgroups(opts.user, p->pw_gid)) {
-               ERROR("failed to initgroups() for user %s: %m\n", opts.user);
+       if (opts.namespace && opts.hostname && strlen(opts.hostname) > 0
+                       && sethostname(opts.hostname, strlen(opts.hostname))) {
+               ERROR("sethostname(%s) failed: %m\n", opts.hostname);
                exit(EXIT_FAILURE);
        }
 
-       if (g && g->gr_gid && setgid(g->gr_gid)) {
-               ERROR("failed to set group id %d: %m\n", g?g->gr_gid:p->pw_gid);
+       if (opts.namespace && build_jail_fs()) {
+               ERROR("failed to build jail fs\n");
                exit(EXIT_FAILURE);
        }
 
-       if (p && p->pw_uid && setuid(p->pw_uid)) {
-               ERROR("failed to set user id %d: %m\n", p->pw_uid);
+       if (opts.capabilities && drop_capabilities(opts.capabilities))
+               exit(EXIT_FAILURE);
+
+       if (opts.no_new_privs && prctl(PR_SET_NO_NEW_PRIVS, 1, 0, 0, 0)) {
+                ERROR("prctl(PR_SET_NO_NEW_PRIVS) failed: %m\n");
                exit(EXIT_FAILURE);
        }
 
+       if (!(opts.namespace & CLONE_NEWUSER)) {
+               get_jail_user(&pw_uid, &pw_gid, &gr_gid);
+               set_jail_user(pw_uid, pw_gid, gr_gid);
+       }
 
        char **envp = build_envp(opts.seccomp);
        if (!envp)
@@ -398,12 +495,20 @@ static void jail_handle_signal(int signo)
        kill(jail_process.pid, signo);
 }
 
-static void netns_updown(bool start)
+static int netns_open_pid(const pid_t target_ns)
+{
+       char pid_net_path[PATH_MAX];
+
+       snprintf(pid_net_path, sizeof(pid_net_path), "/proc/%u/ns/net", target_ns);
+
+       return open(pid_net_path, O_RDONLY);
+}
+
+static void netns_updown(pid_t pid, bool start)
 {
        struct ubus_context *ctx = ubus_connect(NULL);
        static struct blob_buf req;
        uint32_t id;
-       pid_t pid = getpid();
 
        if (!ctx)
                return;
@@ -428,6 +533,9 @@ int main(int argc, char **argv)
        char log[] = "/dev/log";
        char ubus[] = "/var/run/ubus.sock";
        int ch, i;
+       int pipes[4];
+       char sig_buf[1];
+       int netns_fd;
 
        if (uid) {
                ERROR("not root, aborting: %m\n");
@@ -444,19 +552,24 @@ int main(int argc, char **argv)
                        debug = atoi(optarg);
                        break;
                case 'p':
-                       opts.namespace |= NAMESPACE_MOUNT;
+                       opts.namespace |= CLONE_NEWNS;
                        opts.procfs = 1;
                        break;
                case 'o':
-                       opts.namespace |= NAMESPACE_MOUNT;
+                       opts.namespace |= CLONE_NEWNS;
                        opts.ronly = 1;
                        break;
+               case 'f':
+                       opts.namespace |= CLONE_NEWUSER;
+                       break;
+               case 'F':
+                       opts.namespace |= CLONE_NEWCGROUP;
+                       break;
                case 'R':
-                       opts.namespace |= NAMESPACE_MOUNT | NAMESPACE_UTS;
                        opts.extroot = optarg;
                        break;
                case 's':
-                       opts.namespace |= NAMESPACE_MOUNT;
+                       opts.namespace |= CLONE_NEWNS;
                        opts.sysfs = 1;
                        break;
                case 'S':
@@ -473,25 +586,26 @@ int main(int argc, char **argv)
                        opts.name = optarg;
                        break;
                case 'N':
-                       opts.namespace |= NAMESPACE_NET;
+                       opts.namespace |= CLONE_NEWNET;
                        break;
                case 'h':
+                       opts.namespace |= CLONE_NEWUTS;
                        opts.hostname = optarg;
                        break;
                case 'r':
-                       opts.namespace |= NAMESPACE_MOUNT;
+                       opts.namespace |= CLONE_NEWNS;
                        add_path_and_deps(optarg, 1, 0, 0);
                        break;
                case 'w':
-                       opts.namespace |= NAMESPACE_MOUNT;
+                       opts.namespace |= CLONE_NEWNS;
                        add_path_and_deps(optarg, 0, 0, 0);
                        break;
                case 'u':
-                       opts.namespace |= NAMESPACE_MOUNT;
+                       opts.namespace |= CLONE_NEWNS;
                        add_mount(ubus, 0, -1);
                        break;
                case 'l':
-                       opts.namespace |= NAMESPACE_MOUNT;
+                       opts.namespace |= CLONE_NEWNS;
                        add_mount(log, 0, -1);
                        break;
                case 'U':
@@ -503,6 +617,9 @@ int main(int argc, char **argv)
                }
        }
 
+       if (opts.namespace)
+               opts.namespace |= CLONE_NEWIPC | CLONE_NEWPID;
+
        /* no <binary> param found */
        if (argc - optind < 1) {
                usage();
@@ -513,13 +630,15 @@ int main(int argc, char **argv)
                usage();
                return EXIT_FAILURE;
        }
-       DEBUG("Using namespaces(%d), capabilities(%d), seccomp(%d)\n",
+       DEBUG("Using namespaces(0x%08x), capabilities(%d), seccomp(%d)\n",
                opts.namespace,
                opts.capabilities != 0,
                opts.seccomp != 0);
 
        opts.jail_argv = &argv[optind];
 
+       get_jail_user(&opts.pw_uid, &opts.pw_gid, &opts.gr_gid);
+
        if (!opts.extroot) {
                if (opts.namespace && add_path_and_deps(*opts.jail_argv, 1, -1, 0)) {
                        ERROR("failed to load dependencies\n");
@@ -551,10 +670,7 @@ int main(int argc, char **argv)
        }
 
        if (opts.namespace) {
-               int flags = SIGCHLD | CLONE_NEWPID | CLONE_NEWIPC;
-
-               if (opts.namespace & NAMESPACE_MOUNT) {
-                       flags |= CLONE_NEWNS;
+               if (opts.namespace & CLONE_NEWNS) {
                        add_mount("/dev/full", 0, -1);
                        add_mount("/dev/null", 0, -1);
                        add_mount("/dev/random", 0, -1);
@@ -563,7 +679,7 @@ int main(int argc, char **argv)
                        add_mount("/dev/zero", 0, -1);
                        add_mount("/dev/console", 0, -1);
 
-                       if (opts.user || opts.group) {
+                       if (!opts.extroot && (opts.user || opts.group)) {
                                add_mount("/etc/passwd", 0, -1);
                                add_mount("/etc/group", 0, -1);
                        }
@@ -573,21 +689,50 @@ int main(int argc, char **argv)
                        }
                }
 
-               if (opts.hostname)
-                       flags |= CLONE_NEWUTS;
-
-               if (opts.namespace & NAMESPACE_NET) {
-                       unshare(CLONE_NEWNET);
-                       netns_updown(true);
-               };
+               if (pipe(&pipes[0]) < 0 || pipe(&pipes[2]) < 0)
+                       return -1;
 
-               jail_process.pid = clone(exec_jail, child_stack + STACK_SIZE, flags, NULL);
+               jail_process.pid = clone(exec_jail, child_stack + STACK_SIZE, SIGCHLD | opts.namespace, &pipes);
        } else {
                jail_process.pid = fork();
        }
 
        if (jail_process.pid > 0) {
+               seteuid(0);
                /* parent process */
+               close(pipes[1]);
+               close(pipes[2]);
+               if (read(pipes[0], sig_buf, 1) < 1) {
+                       ERROR("can't read from child\n");
+                       return -1;
+               }
+               close(pipes[0]);
+               if (opts.namespace & CLONE_NEWUSER) {
+                       bool has_gr = (opts.gr_gid != -1);
+                       if (write_setgroups(jail_process.pid, false)) {
+                               ERROR("can't write setgroups\n");
+                               return -1;
+                       }
+                       if (opts.pw_uid != -1) {
+                               write_uid_gid_map(jail_process.pid, 0, opts.pw_uid);
+                               write_uid_gid_map(jail_process.pid, 1, has_gr?opts.gr_gid:opts.pw_gid);
+                       } else {
+                               write_uid_gid_map(jail_process.pid, 0, 65534);
+                               write_uid_gid_map(jail_process.pid, 1, has_gr?opts.gr_gid:65534);
+                       }
+               }
+
+               if (opts.namespace & CLONE_NEWNET) {
+                       netns_fd = netns_open_pid(jail_process.pid);
+                       netns_updown(jail_process.pid, true);
+               }
+
+               sig_buf[0] = 'O';
+               if (write(pipes[3], sig_buf, 1) < 0) {
+                       ERROR("can't write to child\n");
+                       return -1;
+               }
+               close(pipes[3]);
                uloop_process_add(&jail_process);
                uloop_run();
                if (jail_running) {
@@ -597,9 +742,11 @@ int main(int argc, char **argv)
                        uloop_run();
                }
                uloop_done();
-               if (opts.namespace & NAMESPACE_NET)
-                       netns_updown(false);
-
+               if (opts.namespace & CLONE_NEWNET) {
+                       setns(netns_fd, CLONE_NEWNET);
+                       netns_updown(getpid(), false);
+                       close(netns_fd);
+               }
                return jail_return_code;
        } else if (jail_process.pid == 0) {
                /* fork child process */
index 47b7d09dae2a5080203dd952aad346bfa074f828..d29aa0da53e249b5fa946772fb0ec3359615eef9 100644 (file)
@@ -62,6 +62,7 @@ enum {
        INSTANCE_ATTR_RELOADSIG,
        INSTANCE_ATTR_TERMTIMEOUT,
        INSTANCE_ATTR_FACILITY,
+       INSTANCE_ATTR_EXTROOT,
        __INSTANCE_ATTR_MAX
 };
 
@@ -89,6 +90,7 @@ static const struct blobmsg_policy instance_attr[__INSTANCE_ATTR_MAX] = {
        [INSTANCE_ATTR_RELOADSIG] = { "reload_signal", BLOBMSG_TYPE_INT32 },
        [INSTANCE_ATTR_TERMTIMEOUT] = { "term_timeout", BLOBMSG_TYPE_INT32 },
        [INSTANCE_ATTR_FACILITY] = { "facility", BLOBMSG_TYPE_STRING },
+       [INSTANCE_ATTR_EXTROOT] = { "extroot", BLOBMSG_TYPE_STRING },
 };
 
 enum {
@@ -101,8 +103,9 @@ enum {
        JAIL_ATTR_RONLY,
        JAIL_ATTR_MOUNT,
        JAIL_ATTR_NETNS,
+       JAIL_ATTR_USERNS,
+       JAIL_ATTR_CGROUPSNS,
        JAIL_ATTR_REQUIREJAIL,
-       JAIL_ATTR_EXTROOT,
        __JAIL_ATTR_MAX,
 };
 
@@ -116,8 +119,9 @@ static const struct blobmsg_policy jail_attr[__JAIL_ATTR_MAX] = {
        [JAIL_ATTR_RONLY] = { "ronly", BLOBMSG_TYPE_BOOL },
        [JAIL_ATTR_MOUNT] = { "mount", BLOBMSG_TYPE_TABLE },
        [JAIL_ATTR_NETNS] = { "netns", BLOBMSG_TYPE_BOOL },
+       [JAIL_ATTR_USERNS] = { "userns", BLOBMSG_TYPE_BOOL },
+       [JAIL_ATTR_CGROUPSNS] = { "cgroupsns", BLOBMSG_TYPE_BOOL },
        [JAIL_ATTR_REQUIREJAIL] = { "requirejail", BLOBMSG_TYPE_BOOL },
-       [JAIL_ATTR_EXTROOT] = { "extroot", BLOBMSG_TYPE_STRING },
 };
 
 struct instance_netdev {
@@ -260,9 +264,15 @@ jail_run(struct service_instance *in, char **argv)
        if (jail->netns)
                argv[argc++] = "-N";
 
-       if (jail->extroot) {
+       if (jail->userns)
+               argv[argc++] = "-f";
+
+       if (jail->cgroupsns)
+               argv[argc++] = "-F";
+
+       if (in->extroot) {
                argv[argc++] = "-R";
-               argv[argc++] = jail->extroot;
+               argv[argc++] = in->extroot;
        }
 
        blobmsg_list_for_each(&jail->mount, var) {
@@ -870,9 +880,13 @@ instance_jail_parse(struct service_instance *in, struct blob_attr *attr)
                jail->netns = blobmsg_get_bool(tb[JAIL_ATTR_NETNS]);
                jail->argc++;
        }
-       if (tb[JAIL_ATTR_EXTROOT]) {
-               jail->extroot = strdup(blobmsg_get_string(tb[JAIL_ATTR_EXTROOT]));
-               jail->argc += 2;
+       if (tb[JAIL_ATTR_USERNS]) {
+               jail->userns = blobmsg_get_bool(tb[JAIL_ATTR_USERNS]);
+               jail->argc++;
+       }
+       if (tb[JAIL_ATTR_CGROUPSNS]) {
+               jail->cgroupsns = blobmsg_get_bool(tb[JAIL_ATTR_CGROUPSNS]);
+               jail->argc++;
        }
 
        if (tb[JAIL_ATTR_MOUNT]) {
@@ -892,6 +906,10 @@ instance_jail_parse(struct service_instance *in, struct blob_attr *attr)
        if (in->group)
                jail->argc += 2;
 
+       if (in->extroot) {
+               jail->argc += 2;
+       }
+
        if (in->no_new_privs)
                jail->argc++;
 
@@ -1003,6 +1021,9 @@ instance_config_parse(struct service_instance *in)
        if (!in->trace && tb[INSTANCE_ATTR_SECCOMP])
                in->seccomp = strdup(blobmsg_get_string(tb[INSTANCE_ATTR_SECCOMP]));
 
+       if (tb[INSTANCE_ATTR_EXTROOT])
+               in->extroot = strdup(blobmsg_get_string(tb[INSTANCE_ATTR_EXTROOT]));
+
        if (tb[INSTANCE_ATTR_PIDFILE]) {
                char *pidfile = blobmsg_get_string(tb[INSTANCE_ATTR_PIDFILE]);
                if (pidfile)
@@ -1151,7 +1172,7 @@ instance_free(struct service_instance *in)
        free(in->config);
        free(in->user);
        free(in->group);
-       free(in->jail.extroot);
+       free(in->extroot);
        free(in->jail.name);
        free(in->jail.hostname);
        free(in->seccomp);
@@ -1275,14 +1296,16 @@ void instance_dump(struct blob_buf *b, struct service_instance *in, int verbose)
                        blobmsg_add_string(b, "name", in->jail.name);
                if (in->jail.hostname)
                        blobmsg_add_string(b, "hostname", in->jail.hostname);
-               if (in->jail.extroot)
-                       blobmsg_add_string(b, "extroot", in->jail.extroot);
+               if (in->extroot)
+                       blobmsg_add_string(b, "extroot", in->extroot);
                blobmsg_add_u8(b, "procfs", in->jail.procfs);
                blobmsg_add_u8(b, "sysfs", in->jail.sysfs);
                blobmsg_add_u8(b, "ubus", in->jail.ubus);
                blobmsg_add_u8(b, "log", in->jail.log);
                blobmsg_add_u8(b, "ronly", in->jail.ronly);
                blobmsg_add_u8(b, "netns", in->jail.netns);
+               blobmsg_add_u8(b, "userns", in->jail.userns);
+               blobmsg_add_u8(b, "cgroupsns", in->jail.cgroupsns);
                blobmsg_close_table(b, r);
                if (!avl_is_empty(&in->jail.mount.avl)) {
                        struct blobmsg_list_node *var;
index 29406c93ecbc71b0ffa0f91c015ffe1ded24fc84..7e80c61946516112bef33b6f2a96761f52d5d826 100644 (file)
@@ -30,9 +30,10 @@ struct jail {
        bool log;
        bool ronly;
        bool netns;
+       bool userns;
+       bool cgroupsns;
        char *name;
        char *hostname;
-       char *extroot;
        struct blobmsg_list mount;
        int argc;
 };
@@ -65,6 +66,7 @@ struct service_instance {
        struct jail jail;
        char *seccomp;
        char *pidfile;
+       char *extroot;
        int syslog_facility;
        int exit_code;