#include "base-filesystem.h"
#include "barrier.h"
#include "event-util.h"
+#include "cap-list.h"
+#include "btrfs-util.h"
#ifdef HAVE_SECCOMP
#include "seccomp-util.h"
} Volatile;
static char *arg_directory = NULL;
+static char *arg_template = NULL;
static char *arg_user = NULL;
static sd_id128_t arg_uuid = {};
static char *arg_machine = NULL;
static bool arg_private_network = false;
static bool arg_read_only = false;
static bool arg_boot = false;
+static bool arg_ephemeral = false;
static LinkJournal arg_link_journal = LINK_AUTO;
static bool arg_link_journal_try = false;
static uint64_t arg_retain =
static bool arg_network_veth = false;
static const char *arg_network_bridge = NULL;
static unsigned long arg_personality = 0xffffffffLU;
-static const char *arg_image = NULL;
+static char *arg_image = NULL;
static Volatile arg_volatile = VOLATILE_NO;
static void help(void) {
" --version Print version string\n"
" -q --quiet Do not show status information\n"
" -D --directory=PATH Root directory for the container\n"
- " -i --image=PATH File system device or image for the container\n"
+ " --template=PATH Initialize root directory from template directory,\n"
+ " if missing\n"
+ " -x --ephemeral Run container with snapshot of root directory, and\n"
+ " remove it after exit\n"
+ " -i --image=PATH File system device or disk image 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"
" -M --machine=NAME Set the machine name for the container\n"
program_invocation_short_name);
}
+static int set_sanitized_path(char **b, const char *path) {
+ char *p;
+
+ assert(b);
+ assert(path);
+
+ p = canonicalize_file_name(path);
+ if (!p) {
+ if (errno != ENOENT)
+ return -errno;
+
+ p = path_make_absolute_cwd(path);
+ if (!p)
+ return -ENOMEM;
+ }
+
+ free(*b);
+ *b = path_kill_slashes(p);
+ return 0;
+}
+
static int parse_argv(int argc, char *argv[]) {
enum {
ARG_NETWORK_BRIDGE,
ARG_PERSONALITY,
ARG_VOLATILE,
+ ARG_TEMPLATE,
};
static const struct option options[] = {
{ "help", no_argument, NULL, 'h' },
{ "version", no_argument, NULL, ARG_VERSION },
{ "directory", required_argument, NULL, 'D' },
+ { "template", required_argument, NULL, ARG_TEMPLATE },
+ { "ephemeral", no_argument, NULL, 'x' },
{ "user", required_argument, NULL, 'u' },
{ "private-network", no_argument, NULL, ARG_PRIVATE_NETWORK },
{ "boot", no_argument, NULL, 'b' },
assert(argc >= 0);
assert(argv);
- while ((c = getopt_long(argc, argv, "+hD:u:bL:M:jS:Z:qi:", options, NULL)) >= 0)
+ while ((c = getopt_long(argc, argv, "+hD:u:bL:M:jS:Z:qi:x", options, NULL)) >= 0)
switch (c) {
return 0;
case 'D':
- free(arg_directory);
- arg_directory = canonicalize_file_name(optarg);
- if (!arg_directory) {
- log_error_errno(errno, "Invalid root directory: %m");
- return -ENOMEM;
- }
+ r = set_sanitized_path(&arg_directory, optarg);
+ if (r < 0)
+ return log_error_errno(r, "Invalid root directory: %m");
+
+ break;
+
+ case ARG_TEMPLATE:
+ r = set_sanitized_path(&arg_template, optarg);
+ if (r < 0)
+ return log_error_errno(r, "Invalid template directory: %m");
break;
case 'i':
- arg_image = optarg;
+ r = set_sanitized_path(&arg_image, optarg);
+ if (r < 0)
+ return log_error_errno(r, "Invalid image path: %m");
+
+ break;
+
+ case 'x':
+ arg_ephemeral = true;
break;
case 'u':
free(arg_machine);
arg_machine = NULL;
} else {
-
- if (!hostname_is_valid(optarg)) {
+ if (!machine_name_is_valid(optarg)) {
log_error("Invalid machine name: %s", optarg);
return -EINVAL;
}
- free(arg_machine);
- arg_machine = strdup(optarg);
- if (!arg_machine)
+ r = free_and_strdup(&arg_machine, optarg);
+ if (r < 0)
return log_oom();
break;
FOREACH_WORD_SEPARATOR(word, length, optarg, ",", state) {
_cleanup_free_ char *t;
- cap_value_t cap;
t = strndup(word, length);
if (!t)
else
minus = (uint64_t) -1;
} else {
- if (cap_from_name(t, &cap) < 0) {
+ int cap;
+
+ cap = capability_from_name(t);
+ if (cap < 0) {
log_error("Failed to parse capability %s.", t);
return -EINVAL;
}
break;
case ARG_LINK_JOURNAL:
- if (streq(optarg, "auto"))
+ if (streq(optarg, "auto")) {
arg_link_journal = LINK_AUTO;
- else if (streq(optarg, "no"))
+ arg_link_journal_try = false;
+ } else if (streq(optarg, "no")) {
arg_link_journal = LINK_NO;
- else if (streq(optarg, "guest"))
+ arg_link_journal_try = false;
+ } else if (streq(optarg, "guest")) {
arg_link_journal = LINK_GUEST;
- else if (streq(optarg, "host"))
+ arg_link_journal_try = false;
+ } else if (streq(optarg, "host")) {
arg_link_journal = LINK_HOST;
- else if (streq(optarg, "try-guest")) {
+ arg_link_journal_try = false;
+ } else if (streq(optarg, "try-guest")) {
arg_link_journal = LINK_GUEST;
arg_link_journal_try = true;
} else if (streq(optarg, "try-host")) {
return -EINVAL;
}
+ if (arg_template && arg_image) {
+ log_error("--template= and --image= may not be combined.");
+ return -EINVAL;
+ }
+
+ if (arg_template && !(arg_directory || arg_machine)) {
+ log_error("--template= needs --directory= or --machine=.");
+ return -EINVAL;
+ }
+
+ if (arg_ephemeral && arg_template) {
+ log_error("--ephemeral and --template= may not be combined.");
+ return -EINVAL;
+ }
+
+ if (arg_ephemeral && arg_image) {
+ log_error("--ephemeral and --image= may not be combined.");
+ return -EINVAL;
+ }
+
+ if (arg_ephemeral && !IN_SET(arg_link_journal, LINK_NO, LINK_AUTO)) {
+ log_error("--ephemeral and --link-journal= may not be combined.");
+ return -EINVAL;
+ }
+
if (arg_volatile != VOLATILE_NO && arg_read_only) {
log_error("Cannot combine --read-only with --volatile. Note that --volatile already implies a read-only base hierarchy.");
return -EINVAL;
struct stat source_st, dest_st;
int r;
- if (stat(*x, &source_st) < 0) {
- log_error_errno(errno, "Failed to stat %s: %m", *x);
- return -errno;
- }
+ if (stat(*x, &source_st) < 0)
+ return log_error_errno(errno, "Failed to stat %s: %m", *x);
where = strappend(dest, *y);
if (!where)
return log_error_errno(r, "Failed to create mount point %s: %m", where);
} else if (S_ISFIFO(source_st.st_mode)) {
r = mkfifo(where, 0644);
- if (r < 0 && errno != EEXIST) {
- log_error_errno(errno, "Failed to create mount point %s: %m", where);
-
- return -errno;
- }
+ if (r < 0 && errno != EEXIST)
+ return log_error_errno(errno, "Failed to create mount point %s: %m", where);
} else if (S_ISSOCK(source_st.st_mode)) {
r = mknod(where, 0644 | S_IFSOCK, 0);
- if (r < 0 && errno != EEXIST) {
- log_error_errno(errno, "Failed to create mount point %s: %m", where);
-
- return -errno;
- }
+ if (r < 0 && errno != EEXIST)
+ return log_error_errno(errno, "Failed to create mount point %s: %m", where);
} else if (S_ISREG(source_st.st_mode)) {
r = touch(where);
if (r < 0)
return -ENOTSUP;
}
- if (mount(*x, where, "bind", MS_BIND, NULL) < 0) {
- log_error_errno(errno, "mount(%s) failed: %m", where);
- return -errno;
- }
+ if (mount(*x, where, "bind", MS_BIND, NULL) < 0)
+ return log_error_errno(errno, "mount(%s) failed: %m", where);
if (ro) {
r = bind_remount_recursive(where, true);
return log_oom();
r = mkdir_label(where, 0755);
- if (r < 0 && errno != EEXIST)
- return log_error_errno(r, "creating mount point for tmpfs %s failed: %m", where);
+ if (r < 0 && r != -EEXIST)
+ return log_error_errno(r, "Creating mount point for tmpfs %s failed: %m", where);
- if (mount("tmpfs", where, "tmpfs", MS_NODEV|MS_STRICTATIME, *o) < 0) {
- log_error_errno(errno, "tmpfs mount to %s failed: %m", where);
- return -errno;
- }
+ if (mount("tmpfs", where, "tmpfs", MS_NODEV|MS_STRICTATIME, *o) < 0)
+ return log_error_errno(errno, "tmpfs mount to %s failed: %m", where);
}
return 0;
p = strappenda(directory, "/var");
r = mkdir(p, 0755);
- if (r < 0 && errno != EEXIST) {
- log_error_errno(errno, "Failed to create %s: %m", directory);
- return -errno;
- }
+ if (r < 0 && errno != EEXIST)
+ return log_error_errno(errno, "Failed to create %s: %m", directory);
- if (mount("tmpfs", p, "tmpfs", MS_STRICTATIME, "mode=755") < 0) {
- log_error_errno(errno, "Failed to mount tmpfs to /var: %m");
- return -errno;
- }
+ if (mount("tmpfs", p, "tmpfs", MS_STRICTATIME, "mode=755") < 0)
+ return log_error_errno(errno, "Failed to mount tmpfs to /var: %m");
return 0;
}
/* --volatile=yes means we mount a tmpfs to the root dir, and
the original /usr to use inside it, and that read-only. */
- if (!mkdtemp(template)) {
- log_error_errno(errno, "Failed to create temporary directory: %m");
- return -errno;
- }
+ if (!mkdtemp(template))
+ return log_error_errno(errno, "Failed to create temporary directory: %m");
if (mount("tmpfs", template, "tmpfs", MS_STRICTATIME, "mode=755") < 0) {
log_error_errno(errno, "Failed to mount tmpfs for root directory: %m");
if (stat(from, &st) < 0) {
- if (errno != ENOENT) {
- log_error_errno(errno, "Failed to stat %s: %m", from);
- return -errno;
- }
+ if (errno != ENOENT)
+ return log_error_errno(errno, "Failed to stat %s: %m", from);
} else if (!S_ISCHR(st.st_mode) && !S_ISBLK(st.st_mode)) {
return -r;
}
- if (mknod(to, st.st_mode, st.st_rdev) < 0) {
- log_error_errno(errno, "mknod(%s) failed: %m", dest);
- return -errno;
- }
+ if (mknod(to, st.st_mode, st.st_rdev) < 0)
+ return log_error_errno(errno, "mknod(%s) failed: %m", dest);
}
}
if (!p)
return log_oom();
- if (symlink("pts/ptmx", p) < 0) {
- log_error_errno(errno, "Failed to create /dev/ptmx symlink: %m");
- return -errno;
- }
+ if (symlink("pts/ptmx", p) < 0)
+ return log_error_errno(errno, "Failed to create /dev/ptmx symlink: %m");
return 0;
}
u = umask(0000);
- if (stat("/dev/null", &st) < 0) {
- log_error_errno(errno, "Failed to stat /dev/null: %m");
- return -errno;
- }
+ if (stat("/dev/null", &st) < 0)
+ return log_error_errno(errno, "Failed to stat /dev/null: %m");
r = chmod_and_chown(console, 0600, 0, 0);
if (r < 0)
* matter here, since we mount it over anyway). */
to = strappenda(dest, "/dev/console");
- if (mknod(to, (st.st_mode & ~07777) | 0600, st.st_rdev) < 0) {
- log_error_errno(errno, "mknod() for /dev/console failed: %m");
- return -errno;
- }
+ if (mknod(to, (st.st_mode & ~07777) | 0600, st.st_rdev) < 0)
+ return log_error_errno(errno, "mknod() for /dev/console failed: %m");
- if (mount(console, to, "bind", MS_BIND, NULL) < 0) {
- log_error_errno(errno, "Bind mount for /dev/console failed: %m");
- return -errno;
- }
+ if (mount(console, to, "bind", MS_BIND, NULL) < 0)
+ return log_error_errno(errno, "Bind mount for /dev/console failed: %m");
return 0;
}
asprintf(&to, "%s/proc/kmsg", dest) < 0)
return log_oom();
- if (mkfifo(from, 0600) < 0) {
- log_error_errno(errno, "mkfifo() for /dev/kmsg failed: %m");
- return -errno;
- }
+ if (mkfifo(from, 0600) < 0)
+ return log_error_errno(errno, "mkfifo() for /dev/kmsg failed: %m");
r = chmod_and_chown(from, 0600, 0, 0);
if (r < 0)
return log_error_errno(r, "Failed to correct access mode for /dev/kmsg: %m");
- if (mount(from, to, "bind", MS_BIND, NULL) < 0) {
- log_error_errno(errno, "Bind mount for /proc/kmsg failed: %m");
- return -errno;
- }
+ if (mount(from, to, "bind", MS_BIND, NULL) < 0)
+ return log_error_errno(errno, "Bind mount for /proc/kmsg failed: %m");
fd = open(from, O_RDWR|O_NDELAY|O_CLOEXEC);
- if (fd < 0) {
- log_error_errno(errno, "Failed to open fifo: %m");
- return -errno;
- }
+ if (fd < 0)
+ return log_error_errno(errno, "Failed to open fifo: %m");
cmsg = CMSG_FIRSTHDR(&mh);
cmsg->cmsg_level = SOL_SOCKET;
k = sendmsg(kmsg_socket, &mh, MSG_DONTWAIT|MSG_NOSIGNAL);
safe_close(fd);
- if (k < 0) {
- log_error_errno(errno, "Failed to send FIFO fd: %m");
- return -errno;
- }
+ if (k < 0)
+ return log_error_errno(errno, "Failed to send FIFO fd: %m");
/* And now make the FIFO unavailable as /dev/kmsg... */
unlink(from);
char *id;
int r;
+ /* Don't link journals in ephemeral mode */
+ if (arg_ephemeral)
+ return 0;
+
p = strappend(directory, "/etc/machine-id");
if (!p)
return log_oom();
"Host and machine ids are equal (%s): refusing to link journals", id);
if (arg_link_journal == LINK_AUTO)
return 0;
- return
- -EEXIST;
+ return -EEXIST;
}
if (arg_link_journal == LINK_NO)
return 0;
}
- if (unlink(p) < 0) {
- log_error_errno(errno, "Failed to remove symlink %s: %m", p);
- return -errno;
- }
+ if (unlink(p) < 0)
+ return log_error_errno(errno, "Failed to remove symlink %s: %m", p);
} else if (r == -EINVAL) {
if (arg_link_journal == LINK_GUEST &&
return r;
}
- if (mount(p, q, "bind", MS_BIND, NULL) < 0) {
- log_error_errno(errno, "Failed to bind mount journal from host into guest: %m");
- return -errno;
- }
+ if (mount(p, q, "bind", MS_BIND, NULL) < 0)
+ return log_error_errno(errno, "Failed to bind mount journal from host into guest: %m");
return 0;
}
#define HOST_HASH_KEY SD_ID128_MAKE(1a,37,6f,c7,46,ec,45,0b,ad,a3,d5,31,06,60,5d,b1)
#define CONTAINER_HASH_KEY SD_ID128_MAKE(c3,c4,f9,19,b5,57,b2,1c,e6,cf,14,27,03,9c,ee,a2)
+#define MACVLAN_HASH_KEY SD_ID128_MAKE(00,13,6d,bc,66,83,44,81,bb,0c,f9,51,1f,24,a6,6f)
-static int generate_mac(struct ether_addr *mac, sd_id128_t hash_key) {
- int r;
-
+static int generate_mac(struct ether_addr *mac, sd_id128_t hash_key, uint64_t idx) {
uint8_t result[8];
size_t l, sz;
- uint8_t *v;
+ uint8_t *v, *i;
+ int r;
l = strlen(arg_machine);
sz = sizeof(sd_id128_t) + l;
+ if (idx > 0)
+ sz += sizeof(idx);
+
v = alloca(sz);
/* fetch some persistent data unique to the host */
/* combine with some data unique (on this host) to this
* container instance */
- memcpy(v + sizeof(sd_id128_t), arg_machine, l);
+ i = mempcpy(v + sizeof(sd_id128_t), arg_machine, l);
+ if (idx > 0) {
+ idx = htole64(idx);
+ memcpy(i, &idx, sizeof(idx));
+ }
/* Let's hash the host machine ID plus the container name. We
* use a fixed, but originally randomly created hash key here. */
snprintf(iface_name, IFNAMSIZ - 1, "%s-%s",
arg_network_bridge ? "vb" : "ve", arg_machine);
- r = generate_mac(&mac_container, CONTAINER_HASH_KEY);
- if (r < 0) {
- log_error("Failed to generate predictable MAC address for container side");
- return r;
- }
+ r = generate_mac(&mac_container, CONTAINER_HASH_KEY, 0);
+ if (r < 0)
+ return log_error_errno(r, "Failed to generate predictable MAC address for container side: %m");
- r = generate_mac(&mac_host, HOST_HASH_KEY);
- if (r < 0) {
- log_error("Failed to generate predictable MAC address for host side");
- return r;
- }
+ r = generate_mac(&mac_host, HOST_HASH_KEY, 0);
+ if (r < 0)
+ return log_error_errno(r, "Failed to generate predictable MAC address for host side: %m");
r = sd_rtnl_open(&rtnl, 0);
if (r < 0)
return log_error_errno(r, "Failed to add new veth interfaces: %m");
i = (int) if_nametoindex(iface_name);
- if (i <= 0) {
- log_error_errno(errno, "Failed to resolve interface %s: %m", iface_name);
- return -errno;
- }
+ if (i <= 0)
+ return log_error_errno(errno, "Failed to resolve interface %s: %m", iface_name);
*ifi = i;
return 0;
bridge = (int) if_nametoindex(arg_network_bridge);
- if (bridge <= 0) {
- log_error_errno(errno, "Failed to resolve interface %s: %m", arg_network_bridge);
- return -errno;
- }
+ if (bridge <= 0)
+ return log_error_errno(errno, "Failed to resolve interface %s: %m", arg_network_bridge);
*ifi = bridge;
int ifi;
ifi = (int) if_nametoindex(name);
- if (ifi <= 0) {
- log_error_errno(errno, "Failed to resolve interface %s: %m", name);
- return -errno;
- }
+ if (ifi <= 0)
+ return log_error_errno(errno, "Failed to resolve interface %s: %m", name);
sprintf(ifi_str, "n%i", ifi);
d = udev_device_new_from_device_id(udev, ifi_str);
- if (!d) {
- log_error_errno(errno, "Failed to get udev device for interface %s: %m", name);
- return -errno;
- }
+ if (!d)
+ return log_error_errno(errno, "Failed to get udev device for interface %s: %m", name);
if (udev_device_get_is_initialized(d) <= 0) {
log_error("Network interface %s is not initialized yet.", name);
static int setup_macvlan(pid_t pid) {
_cleanup_udev_unref_ struct udev *udev = NULL;
_cleanup_rtnl_unref_ sd_rtnl *rtnl = NULL;
+ unsigned idx = 0;
char **i;
int r;
STRV_FOREACH(i, arg_network_macvlan) {
_cleanup_rtnl_message_unref_ sd_rtnl_message *m = NULL;
_cleanup_free_ char *n = NULL;
+ struct ether_addr mac;
int ifi;
ifi = parse_interface(udev, *i);
if (ifi < 0)
return ifi;
+ r = generate_mac(&mac, MACVLAN_HASH_KEY, idx++);
+ if (r < 0)
+ return log_error_errno(r, "Failed to create MACVLAN MAC address: %m");
+
r = sd_rtnl_message_new_link(rtnl, &m, RTM_NEWLINK, 0);
if (r < 0)
return log_error_errno(r, "Failed to allocate netlink message: %m");
if (r < 0)
return log_error_errno(r, "Failed to add netlink interface name: %m");
+ r = sd_rtnl_message_append_ether_addr(m, IFLA_ADDRESS, &mac);
+ if (r < 0)
+ return log_error_errno(r, "Failed to add netlink MAC address: %m");
+
r = sd_rtnl_message_append_u32(m, IFLA_NET_NS_PID, pid);
if (r < 0)
return log_error_errno(r, "Failed to add netlink namespace field: %m");
}
+static int setup_propagate(const char *root) {
+ const char *p, *q;
+
+ (void) mkdir_p("/run/systemd/nspawn/", 0755);
+ (void) mkdir_p("/run/systemd/nspawn/propagate", 0600);
+ p = strappenda("/run/systemd/nspawn/propagate/", arg_machine);
+ (void) mkdir_p(p, 0600);
+
+ q = strappenda(root, "/run/systemd/nspawn/incoming");
+ mkdir_parents(q, 0755);
+ mkdir_p(q, 0600);
+
+ if (mount(p, q, NULL, MS_BIND, NULL) < 0)
+ return log_error_errno(errno, "Failed to install propagation bind mount.");
+
+ if (mount(NULL, q, NULL, MS_BIND|MS_REMOUNT|MS_RDONLY, NULL) < 0)
+ return log_error_errno(errno, "Failed to make propagation mount read-only");
+
+ return 0;
+}
+
static int setup_image(char **device_path, int *loop_nr) {
struct loop_info64 info = {
.lo_flags = LO_FLAGS_AUTOCLEAR|LO_FLAGS_PARTSCAN
assert(device_path);
assert(loop_nr);
+ assert(arg_image);
fd = open(arg_image, O_CLOEXEC|(arg_read_only ? O_RDONLY : O_RDWR)|O_NONBLOCK|O_NOCTTY);
- if (fd < 0) {
- log_error_errno(errno, "Failed to open %s: %m", arg_image);
- return -errno;
- }
+ if (fd < 0)
+ return log_error_errno(errno, "Failed to open %s: %m", arg_image);
- if (fstat(fd, &st) < 0) {
- log_error_errno(errno, "Failed to stat %s: %m", arg_image);
- return -errno;
- }
+ if (fstat(fd, &st) < 0)
+ return log_error_errno(errno, "Failed to stat %s: %m", arg_image);
if (S_ISBLK(st.st_mode)) {
char *p;
}
control = open("/dev/loop-control", O_RDWR|O_CLOEXEC|O_NOCTTY|O_NONBLOCK);
- if (control < 0) {
- log_error_errno(errno, "Failed to open /dev/loop-control: %m");
- return -errno;
- }
+ if (control < 0)
+ return log_error_errno(errno, "Failed to open /dev/loop-control: %m");
nr = ioctl(control, LOOP_CTL_GET_FREE);
- if (nr < 0) {
- log_error_errno(errno, "Failed to allocate loop device: %m");
- return -errno;
- }
+ if (nr < 0)
+ return log_error_errno(errno, "Failed to allocate loop device: %m");
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_errno(errno, "Failed to open loop device %s: %m", loopdev);
- return -errno;
- }
+ if (loop < 0)
+ return log_error_errno(errno, "Failed to open loop device %s: %m", loopdev);
- if (ioctl(loop, LOOP_SET_FD, fd) < 0) {
- log_error_errno(errno, "Failed to set loopback file descriptor on %s: %m", loopdev);
- return -errno;
- }
+ if (ioctl(loop, LOOP_SET_FD, fd) < 0)
+ return log_error_errno(errno, "Failed to set loopback file descriptor on %s: %m", loopdev);
if (arg_read_only)
info.lo_flags |= LO_FLAGS_READ_ONLY;
- if (ioctl(loop, LOOP_SET_STATUS64, &info) < 0) {
- log_error_errno(errno, "Failed to set loopback settings on %s: %m", loopdev);
- return -errno;
- }
+ if (ioctl(loop, LOOP_SET_STATUS64, &info) < 0)
+ return log_error_errno(errno, "Failed to set loopback settings on %s: %m", loopdev);
*device_path = loopdev;
loopdev = NULL;
bool *secondary) {
#ifdef HAVE_BLKID
- int home_nr = -1, root_nr = -1, secondary_root_nr = -1, srv_nr = -1;
+ int home_nr = -1, srv_nr = -1;
+#ifdef GPT_ROOT_NATIVE
+ int root_nr = -1;
+#endif
+#ifdef GPT_ROOT_SECONDARY
+ int secondary_root_nr = -1;
+#endif
+
_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;
assert(home_device);
assert(srv_device);
assert(secondary);
+ assert(arg_image);
b = blkid_new_probe();
if (!b)
if (!udev)
return log_oom();
- if (fstat(fd, &st) < 0) {
- log_error_errno(errno, "Failed to stat block device: %m");
- return -errno;
- }
+ if (fstat(fd, &st) < 0)
+ return log_error_errno(errno, "Failed to stat block device: %m");
d = udev_device_new_from_devnum(udev, 'b', st.st_rdev);
if (!d)
return -ENOTSUP;
}
- if (mount(what, p, fstype, MS_NODEV|(rw ? 0 : MS_RDONLY), NULL) < 0) {
- log_error_errno(errno, "Failed to mount %s: %m", what);
- return -errno;
- }
+ if (mount(what, p, fstype, MS_NODEV|(rw ? 0 : MS_RDONLY), NULL) < 0)
+ return log_error_errno(errno, "Failed to mount %s: %m", what);
return 0;
#else
assert(key);
assert(rpid);
- if (pipe2(pipe_fds, O_CLOEXEC) < 0) {
- log_error_errno(errno, "Failed to allocate pipe: %m");
- return -errno;
- }
+ if (pipe2(pipe_fds, O_CLOEXEC) < 0)
+ return log_error_errno(errno, "Failed to allocate pipe: %m");
pid = fork();
- if (pid < 0) {
- log_error_errno(errno, "Failed to fork getent child: %m");
- return -errno;
- } else if (pid == 0) {
+ if (pid < 0)
+ return log_error_errno(errno, "Failed to fork getent child: %m");
+ else if (pid == 0) {
int nullfd;
char *empty_env = NULL;
if (!arg_user || streq(arg_user, "root") || streq(arg_user, "0")) {
/* Reset everything fully to 0, just in case */
- if (setgroups(0, NULL) < 0) {
- log_error_errno(errno, "setgroups() failed: %m");
- return -errno;
- }
+ if (setgroups(0, NULL) < 0)
+ return log_error_errno(errno, "setgroups() failed: %m");
- if (setresgid(0, 0, 0) < 0) {
- log_error_errno(errno, "setregid() failed: %m");
- return -errno;
- }
+ if (setresgid(0, 0, 0) < 0)
+ return log_error_errno(errno, "setregid() failed: %m");
- if (setresuid(0, 0, 0) < 0) {
- log_error_errno(errno, "setreuid() failed: %m");
- return -errno;
- }
+ if (setresuid(0, 0, 0) < 0)
+ return log_error_errno(errno, "setreuid() failed: %m");
*_home = NULL;
return 0;
truncate_nl(line);
- wait_for_terminate_and_warn("getent passwd", pid);
+ wait_for_terminate_and_warn("getent passwd", pid, true);
x = strchr(line, ':');
if (!x) {
truncate_nl(line);
- wait_for_terminate_and_warn("getent initgroups", pid);
+ wait_for_terminate_and_warn("getent initgroups", pid, true);
/* Skip over the username and subsequent separator whitespace */
x = line;
fchown(STDOUT_FILENO, uid, gid);
fchown(STDERR_FILENO, uid, gid);
- if (setgroups(n_uids, uids) < 0) {
- log_error_errno(errno, "Failed to set auxiliary groups: %m");
- return -errno;
- }
+ if (setgroups(n_uids, uids) < 0)
+ return log_error_errno(errno, "Failed to set auxiliary groups: %m");
- if (setresgid(gid, gid, gid) < 0) {
- log_error_errno(errno, "setregid() failed: %m");
- return -errno;
- }
+ if (setresgid(gid, gid, gid) < 0)
+ return log_error_errno(errno, "setregid() failed: %m");
- if (setresuid(uid, uid, uid) < 0) {
- log_error_errno(errno, "setreuid() failed: %m");
- return -errno;
- }
+ if (setresuid(uid, uid, uid) < 0)
+ return log_error_errno(errno, "setreuid() failed: %m");
if (_home) {
*_home = home;
return 0;
}
+static int determine_names(void) {
+
+ if (!arg_image && !arg_directory) {
+ if (arg_machine)
+ arg_directory = strappend("/var/lib/container/", arg_machine);
+ else
+ arg_directory = get_current_dir_name();
+
+ if (!arg_directory) {
+ log_error("Failed to determine path, please use -D.");
+ return -EINVAL;
+ }
+ }
+
+ if (!arg_machine) {
+ if (arg_directory && path_equal(arg_directory, "/"))
+ arg_machine = gethostname_malloc();
+ else
+ arg_machine = strdup(basename(arg_image ?: arg_directory));
+
+ if (!arg_machine)
+ return log_oom();
+
+ hostname_cleanup(arg_machine, false);
+ if (!machine_name_is_valid(arg_machine)) {
+ log_error("Failed to determine machine name automatically, please use -M.");
+ return -EINVAL;
+ }
+
+ if (arg_ephemeral) {
+ char *b;
+
+ /* Add a random suffix when this is an
+ * ephemeral machine, so that we can run many
+ * instances at once without manually having
+ * to specify -M each time. */
+
+ if (asprintf(&b, "%s-%016" PRIx64, arg_machine, random_u64()) < 0)
+ return log_oom();
+
+ free(arg_machine);
+ arg_machine = b;
+ }
+ }
+
+ return 0;
+}
+
int main(int argc, char *argv[]) {
_cleanup_free_ char *device_path = NULL, *root_device = NULL, *home_device = NULL, *srv_device = NULL;
_cleanup_close_ int master = -1, image_fd = -1;
_cleanup_close_pair_ int kmsg_socket_pair[2] = { -1, -1 };
_cleanup_fdset_free_ FDSet *fds = NULL;
- int r = EXIT_FAILURE, k, n_fd_passed, loop_nr = -1;
+ int r, n_fd_passed, loop_nr = -1;
const char *console = NULL;
char veth_name[IFNAMSIZ];
- bool secondary = false;
+ bool secondary = false, remove_subvol = false;
sigset_t mask, mask_chld;
pid_t pid = 0;
+ int ret = EXIT_SUCCESS;
log_parse_environment();
log_open();
- k = parse_argv(argc, argv);
- if (k < 0)
- goto finish;
- else if (k == 0) {
- r = EXIT_SUCCESS;
+ r = parse_argv(argc, argv);
+ if (r <= 0)
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;
- }
- }
+ r = determine_names();
+ if (r < 0)
+ goto finish;
if (geteuid() != 0) {
log_error("Need to be root.");
+ r = -EPERM;
goto finish;
}
if (sd_booted() <= 0) {
log_error("Not running on a systemd system.");
+ r = -EINVAL;
goto finish;
}
log_close();
n_fd_passed = sd_listen_fds(false);
if (n_fd_passed > 0) {
- k = fdset_new_listen_fds(&fds, false);
- if (k < 0) {
- log_error_errno(k, "Failed to collect file descriptors: %m");
+ r = fdset_new_listen_fds(&fds, false);
+ if (r < 0) {
+ log_error_errno(r, "Failed to collect file descriptors: %m");
goto finish;
}
}
log_open();
if (arg_directory) {
- if (path_equal(arg_directory, "/")) {
- log_error("Spawning container on root directory not supported.");
+ assert(!arg_image);
+
+ if (path_equal(arg_directory, "/") && !arg_ephemeral) {
+ log_error("Spawning container on root directory is not supported. Consider using --ephemeral.");
+ r = -EINVAL;
goto finish;
}
+ if (arg_template) {
+ r = btrfs_subvol_snapshot(arg_template, arg_directory, arg_read_only, true);
+ if (r == -EEXIST) {
+ if (!arg_quiet)
+ log_info("Directory %s already exists, not populating from template %s.", arg_directory, arg_template);
+ } else if (r < 0) {
+ log_error_errno(r, "Couldn't create snapshort %s from %s: %m", arg_directory, arg_template);
+ goto finish;
+ } else {
+ if (!arg_quiet)
+ log_info("Populated %s from template %s.", arg_directory, arg_template);
+ }
+
+ } else if (arg_ephemeral) {
+ char *np;
+
+ /* If the specified path is a mount point we
+ * generate the new snapshot immediately
+ * inside it under a random name. However if
+ * the specified is not a mount point we
+ * create the new snapshot in the parent
+ * directory, just next to it. */
+ r = path_is_mount_point(arg_directory, false);
+ if (r < 0) {
+ log_error_errno(r, "Failed to determine whether directory %s is mount point: %m", arg_directory);
+ goto finish;
+ }
+ if (r > 0)
+ r = tempfn_random_child(arg_directory, &np);
+ else
+ r = tempfn_random(arg_directory, &np);
+ if (r < 0) {
+ log_error_errno(r, "Failed to generate name for snapshot: %m");
+ goto finish;
+ }
+
+ r = btrfs_subvol_snapshot(arg_directory, np, arg_read_only, true);
+ if (r < 0) {
+ free(np);
+ log_error_errno(r, "Failed to create snapshot %s from %s: %m", np, arg_directory);
+ goto finish;
+ }
+
+ free(arg_directory);
+ arg_directory = np;
+
+ remove_subvol = true;
+ }
+
if (arg_boot) {
if (path_is_os_tree(arg_directory) <= 0) {
log_error("Directory %s doesn't look like an OS root directory (os-release file is missing). Refusing.", arg_directory);
+ r = -EINVAL;
goto finish;
}
} else {
argc > optind && path_is_absolute(argv[optind]) ? argv[optind] : "/usr/bin/");
if (access(p, F_OK) < 0) {
log_error("Directory %s lacks the binary to execute or doesn't look like a binary tree. Refusing.", arg_directory);
+ r = -EINVAL;
goto finish;
-
}
}
+
} else {
char template[] = "/tmp/nspawn-root-XXXXXX";
+ assert(arg_image);
+ assert(!arg_template);
+
if (!mkdtemp(template)) {
log_error_errno(errno, "Failed to create temporary directory: %m");
r = -errno;
master = posix_openpt(O_RDWR|O_NOCTTY|O_CLOEXEC|O_NDELAY);
if (master < 0) {
- log_error_errno(errno, "Failed to acquire pseudo tty: %m");
+ r = log_error_errno(errno, "Failed to acquire pseudo tty: %m");
goto finish;
}
console = ptsname(master);
if (!console) {
- log_error_errno(errno, "Failed to determine tty name: %m");
+ r = log_error_errno(errno, "Failed to determine tty name: %m");
goto finish;
}
if (!arg_quiet)
log_info("Spawning container %s on %s.\nPress ^] three times within 1s to kill container.",
- arg_machine, arg_image ? arg_image : arg_directory);
+ arg_machine, arg_image ?: arg_directory);
if (unlockpt(master) < 0) {
- log_error_errno(errno, "Failed to unlock tty: %m");
+ r = log_error_errno(errno, "Failed to unlock tty: %m");
goto finish;
}
if (socketpair(AF_UNIX, SOCK_DGRAM|SOCK_NONBLOCK|SOCK_CLOEXEC, 0, kmsg_socket_pair) < 0) {
- log_error_errno(errno, "Failed to create kmsg socket pair: %m");
+ r = log_error_errno(errno, "Failed to create kmsg socket pair: %m");
goto finish;
}
* give it a chance to call wait() and terminate. */
r = sigprocmask(SIG_UNBLOCK, &mask_chld, NULL);
if (r < 0) {
- log_error_errno(errno, "Failed to change the signal mask: %m");
+ r = log_error_errno(errno, "Failed to change the signal mask: %m");
goto finish;
}
r = sigaction(SIGCHLD, &sa, NULL);
if (r < 0) {
- log_error_errno(errno, "Failed to install SIGCHLD handler: %m");
+ r = log_error_errno(errno, "Failed to install SIGCHLD handler: %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);
+ pid = raw_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_errno(errno, "clone() failed, do you have namespace support enabled in your kernel? (You need UTS, IPC, PID and NET namespacing built in): %m");
+ r = log_error_errno(errno, "clone() failed, do you have namespace support enabled in your kernel? (You need UTS, IPC, PID and NET namespacing built in): %m");
else
- log_error_errno(errno, "clone() failed: %m");
+ r = log_error_errno(errno, "clone() failed: %m");
- r = pid;
goto finish;
}
reset_all_signal_handlers();
reset_signal_mask();
- k = open_terminal(console, O_RDWR);
- if (k != STDIN_FILENO) {
- if (k >= 0) {
- safe_close(k);
- k = -EINVAL;
+ r = open_terminal(console, O_RDWR);
+ if (r != STDIN_FILENO) {
+ if (r >= 0) {
+ safe_close(r);
+ r = -EINVAL;
}
- log_error_errno(k, "Failed to open console: %m");
+ log_error_errno(r, "Failed to open console: %m");
_exit(EXIT_FAILURE);
}
_exit(EXIT_FAILURE);
if (arg_read_only) {
- k = bind_remount_recursive(arg_directory, true);
- if (k < 0) {
- log_error_errno(k, "Failed to make tree read-only: %m");
+ r = bind_remount_recursive(arg_directory, true);
+ if (r < 0) {
+ log_error_errno(r, "Failed to make tree read-only: %m");
_exit(EXIT_FAILURE);
}
}
dev_setup(arg_directory);
+ if (setup_propagate(arg_directory) < 0)
+ _exit(EXIT_FAILURE);
+
if (setup_seccomp() < 0)
_exit(EXIT_FAILURE);
}
if (fdset_size(fds) > 0) {
- k = fdset_cloexec(fds, false);
- if (k < 0) {
- log_error("Failed to unset O_CLOEXEC for file descriptors.");
+ r = fdset_cloexec(fds, false);
+ if (r < 0) {
+ log_error_errno(r, "Failed to unset O_CLOEXEC for file descriptors.");
_exit(EXIT_FAILURE);
}
}
r = sd_event_loop(event);
- if (r < 0)
- return log_error_errno(r, "Failed to run event loop: %m");
+ if (r < 0) {
+ log_error_errno(r, "Failed to run event loop: %m");
+ goto finish;
+ }
forward = pty_forward_free(forward);
r = wait_for_container(pid, &container_status);
pid = 0;
- if (r < 0) {
+ if (r < 0)
/* We failed to wait for the container, or the
* container exited abnormally */
- r = EXIT_FAILURE;
- break;
- } else if (r > 0 || container_status == CONTAINER_TERMINATED)
+ goto finish;
+ else if (r > 0 || container_status == CONTAINER_TERMINATED){
/* The container exited with a non-zero
* status, or with zero status and no reboot
* was requested. */
+ ret = r;
break;
+ }
/* CONTAINER_REBOOTED, loop again */
* restart. This is necessary since we might
* have cgroup parameters set we want to have
* flushed out. */
- r = 133;
+ ret = 133;
+ r = 0;
break;
}
}
if (pid > 0)
kill(pid, SIGKILL);
+ if (remove_subvol && arg_directory) {
+ int k;
+
+ k = btrfs_subvol_remove(arg_directory);
+ if (k < 0)
+ log_warning_errno(k, "Cannot remove subvolume '%s', ignoring: %m", arg_directory);
+ }
+
+ if (arg_machine) {
+ const char *p;
+
+ p = strappenda("/run/systemd/nspawn/propagate", arg_machine);
+ (void) rm_rf(p, false, true, false);
+ }
+
free(arg_directory);
+ free(arg_template);
+ free(arg_image);
free(arg_machine);
free(arg_user);
strv_free(arg_setenv);
strv_free(arg_bind_ro);
strv_free(arg_tmpfs);
- return r;
+ return r < 0 ? EXIT_FAILURE : ret;
}