#include <sys/un.h>
#include <sys/socket.h>
#include <linux/netlink.h>
+#include <sys/eventfd.h>
+#ifdef HAVE_SELINUX
+#include <selinux/selinux.h>
+#endif
#include "sd-daemon.h"
#include "sd-bus.h"
#include "bus-error.h"
#include "ptyfwd.h"
#include "bus-kernel.h"
-
-#ifndef TTY_GID
-#define TTY_GID 5
-#endif
+#include "env-util.h"
+#include "def.h"
typedef enum LinkJournal {
LINK_NO,
static char *arg_user = NULL;
static sd_id128_t arg_uuid = {};
static char *arg_machine = NULL;
+static char *arg_selinux_context = NULL;
+static char *arg_selinux_apifs_context = NULL;
static const char *arg_slice = NULL;
static bool arg_private_network = false;
static bool arg_read_only = false;
(1ULL << CAP_SYS_RESOURCE) |
(1ULL << CAP_SYS_BOOT) |
(1ULL << CAP_AUDIT_WRITE) |
- (1ULL << CAP_AUDIT_CONTROL);
+ (1ULL << CAP_AUDIT_CONTROL) |
+ (1ULL << CAP_MKNOD);
static char **arg_bind = NULL;
static char **arg_bind_ro = NULL;
+static char **arg_setenv = NULL;
+static bool arg_quiet = false;
+static bool arg_share_system = false;
static int help(void) {
printf("%s [OPTIONS...] [PATH] [ARGUMENTS...]\n\n"
"Spawn a minimal namespace container for debugging, testing and building.\n\n"
- " -h --help Show this help\n"
- " --version Print version string\n"
- " -D --directory=NAME Root directory for the container\n"
- " -b --boot Boot up full system (i.e. invoke init)\n"
- " -u --user=USER Run the command under specified user or uid\n"
- " --uuid=UUID Set a specific machine UUID for the container\n"
- " -M --machine=NAME Set the machine name for the container\n"
- " -S --slice=SLICE Place the container in the specified slice\n"
- " --private-network Disable network in container\n"
- " --read-only Mount the root directory read-only\n"
- " --capability=CAP In addition to the default, retain specified\n"
- " capability\n"
- " --drop-capability=CAP Drop the specified capability from the default set\n"
- " --link-journal=MODE Link up guest journal, one of no, auto, guest, host\n"
- " -j Equivalent to --link-journal=host\n"
- " --bind=PATH[:PATH] Bind mount a file or directory from the host into\n"
- " the container\n"
- " --bind-ro=PATH[:PATH] Similar, but creates a read-only bind mount\n",
+ " -h --help Show this help\n"
+ " --version Print version string\n"
+ " -D --directory=NAME Root directory for the container\n"
+ " -b --boot Boot up full system (i.e. invoke init)\n"
+ " -u --user=USER Run the command under specified user or uid\n"
+ " --uuid=UUID Set a specific machine UUID for the container\n"
+ " -M --machine=NAME Set the machine name for the container\n"
+ " -S --slice=SLICE Place the container in the specified slice\n"
+ " -Z --selinux-context=SECLABEL\n"
+ " Set the SELinux security context to be used by\n"
+ " processes in the container\n"
+ " -L --selinux-apifs-context=SECLABEL\n"
+ " Set the SELinux security context to be used by\n"
+ " API/tmpfs file systems in the container\n"
+ " --private-network Disable network in container\n"
+ " --share-system Share system namespaces with host\n"
+ " --read-only Mount the root directory read-only\n"
+ " --capability=CAP In addition to the default, retain specified\n"
+ " capability\n"
+ " --drop-capability=CAP Drop the specified capability from the default set\n"
+ " --link-journal=MODE Link up guest journal, one of no, auto, guest, host\n"
+ " -j Equivalent to --link-journal=host\n"
+ " --bind=PATH[:PATH] Bind mount a file or directory from the host into\n"
+ " the container\n"
+ " --bind-ro=PATH[:PATH] Similar, but creates a read-only bind mount\n"
+ " --setenv=NAME=VALUE Pass an environment variable to PID 1\n"
+ " -q --quiet Do not show status information\n",
program_invocation_short_name);
return 0;
ARG_DROP_CAPABILITY,
ARG_LINK_JOURNAL,
ARG_BIND,
- ARG_BIND_RO
+ ARG_BIND_RO,
+ ARG_SETENV,
+ ARG_SHARE_SYSTEM
};
static const struct option options[] = {
- { "help", no_argument, NULL, 'h' },
- { "version", no_argument, NULL, ARG_VERSION },
- { "directory", required_argument, NULL, 'D' },
- { "user", required_argument, NULL, 'u' },
- { "private-network", no_argument, NULL, ARG_PRIVATE_NETWORK },
- { "boot", no_argument, NULL, 'b' },
- { "uuid", required_argument, NULL, ARG_UUID },
- { "read-only", no_argument, NULL, ARG_READ_ONLY },
- { "capability", required_argument, NULL, ARG_CAPABILITY },
- { "drop-capability", required_argument, NULL, ARG_DROP_CAPABILITY },
- { "link-journal", required_argument, NULL, ARG_LINK_JOURNAL },
- { "bind", required_argument, NULL, ARG_BIND },
- { "bind-ro", required_argument, NULL, ARG_BIND_RO },
- { "machine", required_argument, NULL, 'M' },
- { "slice", required_argument, NULL, 'S' },
+ { "help", no_argument, NULL, 'h' },
+ { "version", no_argument, NULL, ARG_VERSION },
+ { "directory", required_argument, NULL, 'D' },
+ { "user", required_argument, NULL, 'u' },
+ { "private-network", no_argument, NULL, ARG_PRIVATE_NETWORK },
+ { "boot", no_argument, NULL, 'b' },
+ { "uuid", required_argument, NULL, ARG_UUID },
+ { "read-only", no_argument, NULL, ARG_READ_ONLY },
+ { "capability", required_argument, NULL, ARG_CAPABILITY },
+ { "drop-capability", required_argument, NULL, ARG_DROP_CAPABILITY },
+ { "link-journal", required_argument, NULL, ARG_LINK_JOURNAL },
+ { "bind", required_argument, NULL, ARG_BIND },
+ { "bind-ro", required_argument, NULL, ARG_BIND_RO },
+ { "machine", required_argument, NULL, 'M' },
+ { "slice", required_argument, NULL, 'S' },
+ { "setenv", required_argument, NULL, ARG_SETENV },
+ { "selinux-context", required_argument, NULL, 'Z' },
+ { "selinux-apifs-context", required_argument, NULL, 'L' },
+ { "quiet", no_argument, NULL, 'q' },
+ { "share-system", no_argument, NULL, ARG_SHARE_SYSTEM },
{}
};
assert(argc >= 0);
assert(argv);
- while ((c = getopt_long(argc, argv, "+hD:u:bM:jS:", options, NULL)) >= 0) {
+ while ((c = getopt_long(argc, argv, "+hD:u:bL:M:jS:Z:q", options, NULL)) >= 0) {
switch (c) {
break;
+ case 'Z':
+ arg_selinux_context = optarg;
+ break;
+
+ case 'L':
+ arg_selinux_apifs_context = optarg;
+ break;
+
case ARG_READ_ONLY:
arg_read_only = true;
break;
break;
}
+ case ARG_SETENV: {
+ char **n;
+
+ if (!env_assignment_is_valid(optarg)) {
+ log_error("Environment variable assignment '%s' is not valid.", optarg);
+ return -EINVAL;
+ }
+
+ n = strv_env_set(arg_setenv, optarg);
+ if (!n)
+ return log_oom();
+
+ strv_free(arg_setenv);
+ arg_setenv = n;
+ break;
+ }
+
+ case 'q':
+ arg_quiet = true;
+ break;
+
+ case ARG_SHARE_SYSTEM:
+ arg_share_system = true;
+ break;
+
case '?':
return -EINVAL;
for (k = 0; k < ELEMENTSOF(mount_table); k++) {
_cleanup_free_ char *where = NULL;
+#ifdef HAVE_SELINUX
+ _cleanup_free_ char *options = NULL;
+#endif
+ const char *o;
int t;
where = strjoin(dest, "/", mount_table[k].where, NULL);
mkdir_p(where, 0755);
+#ifdef HAVE_SELINUX
+ if (arg_selinux_apifs_context &&
+ (streq_ptr(mount_table[k].what, "tmpfs") || streq_ptr(mount_table[k].what, "devpts"))) {
+ options = strjoin(mount_table[k].options, ",context=\"", arg_selinux_apifs_context, "\"", NULL);
+ if (!options)
+ return log_oom();
+
+ o = options;
+ } else
+#endif
+ o = mount_table[k].options;
+
+
if (mount(mount_table[k].what,
where,
mount_table[k].type,
mount_table[k].flags,
- mount_table[k].options) < 0 &&
+ o) < 0 &&
mount_table[k].fatal) {
log_error("mount(%s) failed: %m", where);
u = umask(0000);
NULSTR_FOREACH(d, devnodes) {
- struct stat st;
_cleanup_free_ char *from = NULL, *to = NULL;
+ struct stat st;
- asprintf(&from, "/dev/%s", d);
- asprintf(&to, "%s/dev/%s", dest, d);
-
- if (!from || !to) {
- log_oom();
-
- if (r == 0)
- r = -ENOMEM;
-
- break;
- }
+ from = strappend("/dev/", d);
+ to = strjoin(dest, "/dev/", d, NULL);
+ if (!from || !to)
+ return log_oom();
if (stat(from, &st) < 0) {
if (errno != ENOENT) {
log_error("Failed to stat %s: %m", from);
- if (r == 0)
- r = -errno;
+ return -errno;
}
} else if (!S_ISCHR(st.st_mode) && !S_ISBLK(st.st_mode)) {
log_error("%s is not a char or block device, cannot copy", from);
- if (r == 0)
- r = -EIO;
+ return -EIO;
} else if (mknod(to, st.st_mode, st.st_rdev) < 0) {
log_error("mknod(%s) failed: %m", dest);
- if (r == 0)
- r = -errno;
+ return -errno;
}
}
}
static int setup_journal(const char *directory) {
- sd_id128_t machine_id;
+ sd_id128_t machine_id, this_id;
_cleanup_free_ char *p = NULL, *b = NULL, *q = NULL, *d = NULL;
char *id;
int r;
- if (arg_link_journal == LINK_NO)
- return 0;
-
p = strappend(directory, "/etc/machine-id");
if (!p)
return log_oom();
return r;
}
+ r = sd_id128_get_machine(&this_id);
+ if (r < 0) {
+ log_error("Failed to retrieve machine ID: %s", strerror(-r));
+ return r;
+ }
+
+ if (sd_id128_equal(machine_id, this_id)) {
+ log_full(arg_link_journal == LINK_AUTO ? LOG_WARNING : LOG_ERR,
+ "Host and machine ids are equal (%s): refusing to link journals", id);
+ if (arg_link_journal == LINK_AUTO)
+ return 0;
+ return
+ -EEXIST;
+ }
+
+ if (arg_link_journal == LINK_NO)
+ return 0;
+
free(p);
p = strappend("/var/log/journal/", id);
q = strjoin(directory, "/var/log/journal/", id, NULL);
}
if (mount(path, p, "bind", MS_BIND, NULL) < 0) {
- log_error("Failed to mount kdbus namespace path: %m");
+ log_error("Failed to mount kdbus domain path: %m");
return -errno;
}
return capability_bounding_set_drop(~arg_retain, false);
}
-static int register_machine(void) {
+static int register_machine(pid_t pid) {
_cleanup_bus_error_free_ sd_bus_error error = SD_BUS_ERROR_NULL;
_cleanup_bus_unref_ sd_bus *bus = NULL;
int r;
- r = sd_bus_open_system(&bus);
+ r = sd_bus_default_system(&bus);
if (r < 0) {
log_error("Failed to open system bus: %s", strerror(-r));
return r;
SD_BUS_MESSAGE_APPEND_ID128(arg_uuid),
"nspawn",
"container",
- (uint32_t) 0,
+ (uint32_t) pid,
strempty(arg_directory),
!isempty(arg_slice), "Slice", "s", arg_slice);
if (r < 0) {
int main(int argc, char *argv[]) {
pid_t pid = 0;
int r = EXIT_FAILURE, k;
- _cleanup_close_ int master = -1, kdbus_fd = -1;
+ _cleanup_close_ int master = -1, kdbus_fd = -1, sync_fd = -1;
int n_fd_passed;
const char *console = NULL;
sigset_t mask;
_cleanup_close_pipe_ int kmsg_socket_pair[2] = { -1, -1 };
_cleanup_fdset_free_ FDSet *fds = NULL;
- _cleanup_free_ char *kdbus_namespace = NULL;
+ _cleanup_free_ char *kdbus_domain = NULL;
const char *ns;
log_parse_environment();
goto finish;
}
- if (path_is_os_tree(arg_directory) <= 0) {
+ if (arg_boot && path_is_os_tree(arg_directory) <= 0) {
log_error("Directory %s doesn't look like an OS root directory (/etc/os-release is missing). Refusing.", arg_directory);
goto finish;
}
goto finish;
}
- log_info("Spawning container %s on %s. Press ^] three times within 1s to abort execution.", arg_machine, arg_directory);
+ if (!arg_quiet)
+ log_info("Spawning container %s on %s. Press ^] three times within 1s to abort execution.", arg_machine, arg_directory);
if (unlockpt(master) < 0) {
log_error("Failed to unlock tty: %m");
}
ns = strappenda("machine-", arg_machine);
- kdbus_fd = bus_kernel_create_namespace(ns, &kdbus_namespace);
+ kdbus_fd = bus_kernel_create_domain(ns, &kdbus_domain);
if (r < 0)
- log_debug("Failed to create kdbus namespace: %s", strerror(-r));
+ log_debug("Failed to create kdbus domain: %s", strerror(-r));
else
- log_debug("Successfully created kdbus namespace as %s", kdbus_namespace);
+ log_debug("Successfully created kdbus domain as %s", kdbus_domain);
if (socketpair(AF_UNIX, SOCK_DGRAM|SOCK_NONBLOCK|SOCK_CLOEXEC, 0, kmsg_socket_pair) < 0) {
- log_error("Failed to create kmsg socket pair.");
+ log_error("Failed to create kmsg socket pair: %m");
goto finish;
}
for (;;) {
siginfo_t status;
- pid = syscall(__NR_clone, SIGCHLD|CLONE_NEWIPC|CLONE_NEWNS|CLONE_NEWPID|CLONE_NEWUTS|(arg_private_network ? CLONE_NEWNET : 0), NULL);
+ sync_fd = eventfd(0, EFD_CLOEXEC);
+ if (sync_fd < 0) {
+ log_error("Failed to create event fd: %m");
+ goto finish;
+ }
+
+ pid = syscall(__NR_clone,
+ SIGCHLD|CLONE_NEWNS|
+ (arg_share_system ? 0 : CLONE_NEWIPC|CLONE_NEWPID|CLONE_NEWUTS)|
+ (arg_private_network ? CLONE_NEWNET : 0), NULL);
if (pid < 0) {
if (errno == EINVAL)
log_error("clone() failed, do you have namespace support enabled in your kernel? (You need UTS, IPC, PID and NET namespacing built in): %m");
gid_t gid = (gid_t) -1;
unsigned n_env = 2;
const char *envp[] = {
- "PATH=/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin",
+ "PATH=" DEFAULT_PATH_SPLIT_USR,
"container=systemd-nspawn", /* LXC sets container=lxc, so follow the scheme here */
NULL, /* TERM */
NULL, /* HOME */
NULL, /* LISTEN_PID */
NULL
};
+ char **env_use;
+ eventfd_t x;
envp[n_env] = strv_find_prefix(environ, "TERM=");
if (envp[n_env])
goto child_fail;
}
- r = register_machine();
- if (r < 0)
- goto finish;
-
/* Mark everything as slave, so that we still
* receive mounts from the real root, but don't
* propagate mounts to the real root. */
if (mount_binds(arg_directory, arg_bind_ro, MS_RDONLY) < 0)
goto child_fail;
- if (setup_kdbus(arg_directory, kdbus_namespace) < 0)
+ if (setup_kdbus(arg_directory, kdbus_domain) < 0)
goto child_fail;
if (chdir(arg_directory) < 0) {
setup_hostname();
+ eventfd_read(sync_fd, &x);
+ close_nointr_nofail(sync_fd);
+ sync_fd = -1;
+
+ if (!strv_isempty(arg_setenv)) {
+ char **n;
+
+ n = strv_env_merge(2, envp, arg_setenv);
+ if (!n) {
+ log_oom();
+ goto child_fail;
+ }
+
+ env_use = n;
+ } else
+ env_use = (char**) envp;
+
+#ifdef HAVE_SELINUX
+ if (arg_selinux_context)
+ if (setexeccon(arg_selinux_context) < 0)
+ log_error("setexeccon(\"%s\") failed: %m", arg_selinux_context);
+#endif
if (arg_boot) {
char **a;
size_t l;
memcpy(a + 1, argv + optind, l * sizeof(char*));
a[0] = (char*) "/usr/lib/systemd/systemd";
- execve(a[0], a, (char**) envp);
+ execve(a[0], a, env_use);
a[0] = (char*) "/lib/systemd/systemd";
- execve(a[0], a, (char**) envp);
+ execve(a[0], a, env_use);
a[0] = (char*) "/sbin/init";
- execve(a[0], a, (char**) envp);
+ execve(a[0], a, env_use);
} else if (argc > optind)
- execvpe(argv[optind], argv + optind, (char**) envp);
+ execvpe(argv[optind], argv + optind, env_use);
else {
chdir(home ? home : "/root");
- execle("/bin/bash", "-bash", NULL, (char**) envp);
+ execle("/bin/bash", "-bash", NULL, env_use);
}
log_error("execv() failed: %m");
fdset_free(fds);
fds = NULL;
+ r = register_machine(pid);
+ if (r < 0)
+ goto finish;
+
+ eventfd_write(sync_fd, 1);
+ close_nointr_nofail(sync_fd);
+ sync_fd = -1;
+
k = process_pty(master, &mask, arg_boot ? pid : 0, SIGRTMIN+3);
if (k < 0) {
r = EXIT_FAILURE;
break;
}
- putc('\n', stdout);
+ if (!arg_quiet)
+ putc('\n', stdout);
/* Kill if it is not dead yet anyway */
terminate_machine(pid);
break;
}
- log_debug("Container %s exited successfully.", arg_machine);
+ if (!arg_quiet)
+ log_debug("Container %s exited successfully.", arg_machine);
break;
} else if (status.si_code == CLD_KILLED &&
status.si_status == SIGINT) {
- log_info("Container %s has been shut down.", arg_machine);
+
+ if (!arg_quiet)
+ log_info("Container %s has been shut down.", arg_machine);
r = 0;
break;
} else if (status.si_code == CLD_KILLED &&
status.si_status == SIGHUP) {
- log_info("Container %s is being rebooted.", arg_machine);
+
+ if (!arg_quiet)
+ log_info("Container %s is being rebooted.", arg_machine);
continue;
} else if (status.si_code == CLD_KILLED ||
status.si_code == CLD_DUMPED) {
free(arg_directory);
free(arg_machine);
+ free(arg_setenv);
return r;
}