+ r = seccomp_attr_set(seccomp, SCMP_FLTATR_CTL_NNP, 0);
+ if (r < 0) {
+ log_error("Failed to unset NO_NEW_PRIVS: %s", strerror(-r));
+ goto finish;
+ }
+
+ r = seccomp_load(seccomp);
+ if (r < 0)
+ log_error("Failed to install seccomp audit filter: %s", strerror(-r));
+
+finish:
+ seccomp_release(seccomp);
+ return r;
+#else
+ return 0;
+#endif
+
+}
+
+static int setup_image(char **device_path, int *loop_nr) {
+ struct loop_info64 info = {
+ .lo_flags = LO_FLAGS_AUTOCLEAR|LO_FLAGS_PARTSCAN
+ };
+ _cleanup_close_ int fd = -1, control = -1, loop = -1;
+ _cleanup_free_ char* loopdev = NULL;
+ struct stat st;
+ int r, nr;
+
+ assert(device_path);
+ assert(loop_nr);
+
+ fd = open(arg_image, O_CLOEXEC|(arg_read_only ? O_RDONLY : O_RDWR)|O_NONBLOCK|O_NOCTTY);
+ if (fd < 0) {
+ log_error("Failed to open %s: %m", arg_image);
+ return -errno;
+ }
+
+ if (fstat(fd, &st) < 0) {
+ log_error("Failed to stat %s: %m", arg_image);
+ return -errno;
+ }
+
+ if (S_ISBLK(st.st_mode)) {
+ char *p;
+
+ p = strdup(arg_image);
+ if (!p)
+ return log_oom();
+
+ *device_path = p;
+
+ *loop_nr = -1;
+
+ r = fd;
+ fd = -1;
+
+ return r;
+ }
+
+ if (!S_ISREG(st.st_mode)) {
+ log_error("%s is not a regular file or block device: %m", arg_image);
+ return -EINVAL;
+ }
+
+ control = open("/dev/loop-control", O_RDWR|O_CLOEXEC|O_NOCTTY|O_NONBLOCK);
+ if (control < 0) {
+ log_error("Failed to open /dev/loop-control: %m");
+ return -errno;
+ }
+
+ nr = ioctl(control, LOOP_CTL_GET_FREE);
+ if (nr < 0) {
+ log_error("Failed to allocate loop device: %m");
+ return -errno;
+ }
+
+ if (asprintf(&loopdev, "/dev/loop%i", nr) < 0)
+ return log_oom();
+
+ loop = open(loopdev, O_CLOEXEC|(arg_read_only ? O_RDONLY : O_RDWR)|O_NONBLOCK|O_NOCTTY);
+ if (loop < 0) {
+ log_error("Failed to open loop device %s: %m", loopdev);
+ return -errno;
+ }
+
+ if (ioctl(loop, LOOP_SET_FD, fd) < 0) {
+ log_error("Failed to set loopback file descriptor on %s: %m", loopdev);
+ return -errno;
+ }
+
+ if (arg_read_only)
+ info.lo_flags |= LO_FLAGS_READ_ONLY;
+
+ if (ioctl(loop, LOOP_SET_STATUS64, &info) < 0) {
+ log_error("Failed to set loopback settings on %s: %m", loopdev);
+ return -errno;
+ }
+
+ *device_path = loopdev;
+ loopdev = NULL;
+
+ *loop_nr = nr;
+
+ r = loop;
+ loop = -1;
+
+ return r;
+}
+
+static int dissect_image(
+ int fd,
+ char **root_device,
+ char **home_device,
+ char **srv_device,
+ bool *secondary) {
+
+#ifdef HAVE_BLKID
+ int home_nr = -1, root_nr = -1, secondary_root_nr = -1, srv_nr = -1;
+ _cleanup_free_ char *home = NULL, *root = NULL, *secondary_root = NULL, *srv = NULL;
+ _cleanup_udev_enumerate_unref_ struct udev_enumerate *e = NULL;
+ _cleanup_udev_device_unref_ struct udev_device *d = NULL;
+ _cleanup_blkid_free_probe_ blkid_probe b = NULL;
+ _cleanup_udev_unref_ struct udev *udev = NULL;
+ struct udev_list_entry *first, *item;
+ const char *pttype = NULL;
+ blkid_partlist pl;
+ struct stat st;
+ int r;
+
+ assert(fd >= 0);
+ assert(root_device);
+ assert(home_device);
+ assert(srv_device);
+ assert(secondary);
+
+ b = blkid_new_probe();
+ if (!b)
+ return log_oom();
+
+ errno = 0;
+ r = blkid_probe_set_device(b, fd, 0, 0);
+ if (r != 0) {
+ if (errno == 0)
+ return log_oom();
+
+ log_error("Failed to set device on blkid probe: %m");
+ return -errno;
+ }
+
+ blkid_probe_enable_partitions(b, 1);
+ blkid_probe_set_partitions_flags(b, BLKID_PARTS_ENTRY_DETAILS);
+
+ errno = 0;
+ r = blkid_do_safeprobe(b);
+ if (r == -2 || r == 1) {
+ log_error("Failed to identify any partition table on %s.\n"
+ "Note that the disk image needs to follow http://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec/ to be supported by systemd-nspawn.", arg_image);
+ return -EINVAL;
+ } else if (r != 0) {
+ if (errno == 0)
+ errno = EIO;
+ log_error("Failed to probe: %m");
+ return -errno;
+ }
+
+ blkid_probe_lookup_value(b, "PTTYPE", &pttype, NULL);
+ if (!streq_ptr(pttype, "gpt")) {
+ log_error("Image %s does not carry a GUID Partition Table.\n"
+ "Note that the disk image needs to follow http://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec/ to be supported by systemd-nspawn.", arg_image);
+ return -EINVAL;
+ }
+
+ errno = 0;
+ pl = blkid_probe_get_partitions(b);
+ if (!pl) {
+ if (errno == 0)
+ return log_oom();
+
+ log_error("Failed to list partitions of %s", arg_image);
+ return -errno;
+ }
+
+ udev = udev_new();
+ if (!udev)
+ return log_oom();
+
+ if (fstat(fd, &st) < 0) {
+ log_error("Failed to stat block device: %m");
+ return -errno;
+ }
+
+ d = udev_device_new_from_devnum(udev, 'b', st.st_rdev);
+ if (!d)
+ return log_oom();
+
+ e = udev_enumerate_new(udev);
+ if (!e)
+ return log_oom();
+
+ r = udev_enumerate_add_match_parent(e, d);
+ if (r < 0)
+ return log_oom();
+
+ r = udev_enumerate_scan_devices(e);
+ if (r < 0) {
+ log_error("Failed to scan for partition devices of %s: %s", arg_image, strerror(-r));
+ return r;
+ }
+
+ first = udev_enumerate_get_list_entry(e);
+ udev_list_entry_foreach(item, first) {
+ _cleanup_udev_device_unref_ struct udev_device *q;
+ const char *stype, *node;
+ sd_id128_t type_id;
+ blkid_partition pp;
+ dev_t qn;
+ int nr;
+
+ errno = 0;
+ q = udev_device_new_from_syspath(udev, udev_list_entry_get_name(item));
+ if (!q) {
+ if (!errno)
+ errno = ENOMEM;
+
+ log_error("Failed to get partition device of %s: %m", arg_image);
+ return -errno;
+ }
+
+ qn = udev_device_get_devnum(q);
+ if (major(qn) == 0)
+ continue;
+
+ if (st.st_rdev == qn)
+ continue;
+
+ node = udev_device_get_devnode(q);
+ if (!node)
+ continue;
+
+ pp = blkid_partlist_devno_to_partition(pl, qn);
+ if (!pp)
+ continue;
+
+ nr = blkid_partition_get_partno(pp);
+ if (nr < 0)
+ continue;
+
+ stype = blkid_partition_get_type_string(pp);
+ if (!stype)
+ continue;
+
+ if (sd_id128_from_string(stype, &type_id) < 0)
+ continue;
+
+ if (sd_id128_equal(type_id, GPT_HOME)) {
+
+ if (home && nr >= home_nr)
+ continue;
+
+ home_nr = nr;
+ free(home);
+ home = strdup(node);
+ if (!home)
+ return log_oom();
+ } else if (sd_id128_equal(type_id, GPT_SRV)) {
+
+ if (srv && nr >= srv_nr)
+ continue;
+
+ srv_nr = nr;
+ free(srv);
+ srv = strdup(node);
+ if (!srv)
+ return log_oom();
+ }
+#ifdef GPT_ROOT_NATIVE
+ else if (sd_id128_equal(type_id, GPT_ROOT_NATIVE)) {
+
+ if (root && nr >= root_nr)
+ continue;
+
+ root_nr = nr;
+ free(root);
+ root = strdup(node);
+ if (!root)
+ return log_oom();
+ }
+#endif
+#ifdef GPT_ROOT_SECONDARY
+ else if (sd_id128_equal(type_id, GPT_ROOT_SECONDARY)) {
+
+ if (secondary_root && nr >= secondary_root_nr)
+ continue;
+
+ secondary_root_nr = nr;
+ free(secondary_root);
+ secondary_root = strdup(node);
+ if (!secondary_root)
+ return log_oom();
+ }
+#endif
+ }
+
+ if (!root && !secondary_root) {
+ log_error("Failed to identify root partition in disk image %s.\n"
+ "Note that the disk image needs to follow http://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec/ to be supported by systemd-nspawn.", arg_image);
+ return -EINVAL;
+ }
+
+ if (root) {
+ *root_device = root;
+ root = NULL;
+ *secondary = false;
+ } else if (secondary_root) {
+ *root_device = secondary_root;
+ secondary_root = NULL;
+ *secondary = true;
+ }
+
+ if (home) {
+ *home_device = home;
+ home = NULL;
+ }
+
+ if (srv) {
+ *srv_device = srv;
+ srv = NULL;
+ }
+
+ return 0;
+#else
+ log_error("--image= is not supported, compiled without blkid support.");
+ return -ENOTSUP;
+#endif
+}
+
+static int mount_device(const char *what, const char *where, const char *directory) {
+#ifdef HAVE_BLKID
+ _cleanup_blkid_free_probe_ blkid_probe b = NULL;
+ const char *fstype, *p;
+ int r;
+
+ assert(what);
+ assert(where);
+
+ if (directory)
+ p = strappenda(where, directory);
+ else
+ p = where;
+
+ errno = 0;
+ b = blkid_new_probe_from_filename(what);
+ if (!b) {
+ if (errno == 0)
+ return log_oom();
+ log_error("Failed to allocate prober for %s: %m", what);
+ return -errno;
+ }
+
+ blkid_probe_enable_superblocks(b, 1);
+ blkid_probe_set_superblocks_flags(b, BLKID_SUBLKS_TYPE);
+
+ errno = 0;
+ r = blkid_do_safeprobe(b);
+ if (r == -1 || r == 1) {
+ log_error("Cannot determine file system type of %s", what);
+ return -EINVAL;
+ } else if (r != 0) {
+ if (errno == 0)
+ errno = EIO;
+ log_error("Failed to probe %s: %m", what);
+ return -errno;
+ }
+
+ errno = 0;
+ if (blkid_probe_lookup_value(b, "TYPE", &fstype, NULL) < 0) {
+ if (errno == 0)
+ errno = EINVAL;
+ log_error("Failed to determine file system type of %s", what);
+ return -errno;
+ }
+
+ if (streq(fstype, "crypto_LUKS")) {
+ log_error("nspawn currently does not support LUKS disk images.");
+ return -ENOTSUP;
+ }
+
+ if (mount(what, p, fstype, arg_read_only ? MS_NODEV|MS_RDONLY : 0, NULL) < 0) {
+ log_error("Failed to mount %s: %m", what);
+ return -errno;
+ }
+
+ return 0;
+#else
+ log_error("--image= is not supported, compiled without blkid support.");
+ return -ENOTSUP;
+#endif
+}
+
+static int mount_devices(const char *where, const char *root_device, const char *home_device, const char *srv_device) {
+ int r;
+
+ assert(where);
+
+ if (root_device) {
+ r = mount_device(root_device, arg_directory, NULL);
+ if (r < 0) {
+ log_error("Failed to mount root directory: %s", strerror(-r));
+ return r;
+ }
+ }
+
+ if (home_device) {
+ r = mount_device(home_device, arg_directory, "/home");
+ if (r < 0) {
+ log_error("Failed to mount home directory: %s", strerror(-r));
+ return r;
+ }
+ }
+
+ if (srv_device) {
+ r = mount_device(srv_device, arg_directory, "/srv");
+ if (r < 0) {
+ log_error("Failed to mount server data directory: %s", strerror(-r));
+ return r;
+ }
+ }
+
+ return 0;
+}
+
+static void loop_remove(int nr, int *image_fd) {
+ _cleanup_close_ int control = -1;
+
+ if (nr < 0)
+ return;
+
+ if (image_fd && *image_fd >= 0) {
+ ioctl(*image_fd, LOOP_CLR_FD);
+ close_nointr_nofail(*image_fd);
+ *image_fd = -1;
+ }
+
+ control = open("/dev/loop-control", O_RDWR|O_CLOEXEC|O_NOCTTY|O_NONBLOCK);
+ if (control < 0)
+ return;
+
+ ioctl(control, LOOP_CTL_REMOVE, nr);
+}
+
+int main(int argc, char *argv[]) {
+
+ _cleanup_free_ char *kdbus_domain = NULL, *device_path = NULL, *root_device = NULL, *home_device = NULL, *srv_device = NULL;
+ _cleanup_close_ int master = -1, kdbus_fd = -1, image_fd = -1;
+ _cleanup_close_pipe_ int kmsg_socket_pair[2] = { -1, -1 };
+ _cleanup_fdset_free_ FDSet *fds = NULL;
+ int r = EXIT_FAILURE, k, n_fd_passed, loop_nr = -1;
+ const char *console = NULL;
+ char veth_name[IFNAMSIZ];
+ bool secondary = false;
+ pid_t pid = 0;
+ sigset_t mask;
+
+ log_parse_environment();
+ log_open();
+
+ k = parse_argv(argc, argv);
+ if (k < 0)
+ goto finish;
+ else if (k == 0) {
+ r = EXIT_SUCCESS;
+ goto finish;
+ }
+
+ if (!arg_image) {
+ if (arg_directory) {
+ char *p;
+
+ p = path_make_absolute_cwd(arg_directory);
+ free(arg_directory);
+ arg_directory = p;
+ } else
+ arg_directory = get_current_dir_name();
+
+ if (!arg_directory) {
+ log_error("Failed to determine path, please use -D.");
+ goto finish;
+ }
+ path_kill_slashes(arg_directory);
+ }
+
+ if (!arg_machine) {
+ arg_machine = strdup(basename(arg_image ? arg_image : arg_directory));
+ if (!arg_machine) {
+ log_oom();
+ goto finish;
+ }
+
+ hostname_cleanup(arg_machine, false);
+ if (isempty(arg_machine)) {
+ log_error("Failed to determine machine name automatically, please use -M.");
+ goto finish;
+ }
+ }
+
+ if (geteuid() != 0) {
+ log_error("Need to be root.");
+ goto finish;
+ }
+
+ if (sd_booted() <= 0) {
+ log_error("Not running on a systemd system.");