#include <signal.h>

#include <sched.h>

#include <unistd.h>

#include <sys/types.h>

#include <sys/mount.h>

#include <stdlib.h>

#include <string.h>

#include <stdio.h>

#include <errno.h>

#include <sys/prctl.h>

#include <getopt.h>

#include <grp.h>

#include <linux/fs.h>

#include <sys/socket.h>

#include <linux/netlink.h>

#include <net/if.h>

#include <linux/veth.h>

#include <sys/personality.h>

#include <linux/loop.h>

#include <sys/file.h>

#ifdef HAVE_SELINUX

#include <selinux/selinux.h>

#endif

#ifdef HAVE_SECCOMP

#include <seccomp.h>

#endif

#ifdef HAVE_BLKID

#include <blkid/blkid.h>

#endif

#include "sd-daemon.h"

#include "sd-bus.h"

#include "sd-id128.h"

#include "sd-netlink.h"

#include "random-util.h"

#include "log.h"

#include "util.h"

#include "mkdir.h"

#include "rm-rf.h"

#include "macro.h"

#include "missing.h"

#include "cgroup-util.h"

#include "strv.h"

#include "path-util.h"

#include "loopback-setup.h"

#include "dev-setup.h"

#include "fdset.h"

#include "build.h"

#include "fileio.h"

#include "bus-util.h"

#include "bus-error.h"

#include "ptyfwd.h"

#include "env-util.h"

#include "netlink-util.h"

#include "udev-util.h"

#include "blkid-util.h"

#include "gpt.h"

#include "siphash24.h"

#include "copy.h"

#include "base-filesystem.h"

#include "barrier.h"

#include "event-util.h"

#include "capability.h"

#include "cap-list.h"

#include "btrfs-util.h"

#include "machine-image.h"

#include "list.h"

#include "in-addr-util.h"

#include "firewall-util.h"

#include "local-addresses.h"

#include "formats-util.h"

#include "process-util.h"

#include "terminal-util.h"

#include "hostname-util.h"

#include "signal-util.h"

#ifdef HAVE_SECCOMP

#include "seccomp-util.h"

#endif

typedef struct ExposePort {

int protocol;

uint16_t host_port;

uint16_t container_port;

LIST_FIELDS(struct ExposePort, ports);

} ExposePort;

typedef enum ContainerStatus {

CONTAINER_TERMINATED,

CONTAINER_REBOOTED

} ContainerStatus;

typedef enum LinkJournal {

LINK_NO,

LINK_AUTO,

LINK_HOST,

LINK_GUEST

} LinkJournal;

typedef enum Volatile {

VOLATILE_NO,

VOLATILE_YES,

VOLATILE_STATE,

} Volatile;

typedef enum CustomMountType {

CUSTOM_MOUNT_BIND,

CUSTOM_MOUNT_TMPFS,

CUSTOM_MOUNT_OVERLAY,

} CustomMountType;

typedef struct CustomMount {

CustomMountType type;

bool read_only;

char *source; /* for overlayfs this is the upper directory */

char *destination;

char *options;

char *work_dir;

char **lower;

} CustomMount;

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 const char *arg_selinux_context = NULL;

static const char *arg_selinux_apifs_context = NULL;

static const char *arg_slice = 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 =

(1ULL << CAP_CHOWN) |

(1ULL << CAP_DAC_OVERRIDE) |

(1ULL << CAP_DAC_READ_SEARCH) |

(1ULL << CAP_FOWNER) |

(1ULL << CAP_FSETID) |

(1ULL << CAP_IPC_OWNER) |

(1ULL << CAP_KILL) |

(1ULL << CAP_LEASE) |

(1ULL << CAP_LINUX_IMMUTABLE) |

(1ULL << CAP_NET_BIND_SERVICE) |

(1ULL << CAP_NET_BROADCAST) |

(1ULL << CAP_NET_RAW) |

(1ULL << CAP_SETGID) |

(1ULL << CAP_SETFCAP) |

(1ULL << CAP_SETPCAP) |

(1ULL << CAP_SETUID) |

(1ULL << CAP_SYS_ADMIN) |

(1ULL << CAP_SYS_CHROOT) |

(1ULL << CAP_SYS_NICE) |

(1ULL << CAP_SYS_PTRACE) |

(1ULL << CAP_SYS_TTY_CONFIG) |

(1ULL << CAP_SYS_RESOURCE) |

(1ULL << CAP_SYS_BOOT) |

(1ULL << CAP_AUDIT_WRITE) |

(1ULL << CAP_AUDIT_CONTROL) |

(1ULL << CAP_MKNOD);

static CustomMount *arg_custom_mounts = NULL;

static unsigned arg_n_custom_mounts = 0;

static char **arg_setenv = NULL;

static bool arg_quiet = false;

static bool arg_share_system = false;

static bool arg_register = true;

static bool arg_keep_unit = false;

static char **arg_network_interfaces = NULL;

static char **arg_network_macvlan = NULL;

static char **arg_network_ipvlan = NULL;

static bool arg_network_veth = false;

static const char *arg_network_bridge = NULL;

static unsigned long arg_personality = PERSONALITY_INVALID;

static char *arg_image = NULL;

static Volatile arg_volatile = VOLATILE_NO;

static ExposePort *arg_expose_ports = NULL;

static char **arg_property = NULL;

static uid_t arg_uid_shift = UID_INVALID, arg_uid_range = 0x10000U;

static bool arg_userns = false;

static int arg_kill_signal = 0;

static bool arg_unified_cgroup_hierarchy = false;

static void 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"

" -q --quiet Do not show status information\n"

" -D --directory=PATH Root directory 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"

" --uuid=UUID Set a specific machine UUID for the container\n"

" -S --slice=SLICE Place the container in the specified slice\n"

" --property=NAME=VALUE Set scope unit property\n"

" --private-users[=UIDBASE[:NUIDS]]\n"

" Run within user namespace\n"

" --private-network Disable network in container\n"

" --network-interface=INTERFACE\n"

" Assign an existing network interface to the\n"

" container\n"

" --network-macvlan=INTERFACE\n"

" Create a macvlan network interface based on an\n"

" existing network interface to the container\n"

" --network-ipvlan=INTERFACE\n"

" Create a ipvlan network interface based on an\n"

" existing network interface to the container\n"

" -n --network-veth Add a virtual ethernet connection between host\n"

" and container\n"

" --network-bridge=INTERFACE\n"

" Add a virtual ethernet connection between host\n"

" and container and add it to an existing bridge on\n"

" the host\n"

" -p --port=[PROTOCOL:]HOSTPORT[:CONTAINERPORT]\n"

" Expose a container IP port on the host\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"

" --capability=CAP In addition to the default, retain specified\n"

" capability\n"

" --drop-capability=CAP Drop the specified capability from the default set\n"

" --kill-signal=SIGNAL Select signal to use for shutting down PID 1\n"

" --link-journal=MODE Link up guest journal, one of no, auto, guest, host,\n"

" try-guest, try-host\n"

" -j Equivalent to --link-journal=try-guest\n"

" --read-only Mount the root directory read-only\n"

" --bind=PATH[:PATH[:OPTIONS]]\n"

" Bind mount a file or directory from the host into\n"

" the container\n"

" --bind-ro=PATH[:PATH[:OPTIONS]\n"

" Similar, but creates a read-only bind mount\n"

" --tmpfs=PATH:[OPTIONS] Mount an empty tmpfs to the specified directory\n"

" --overlay=PATH[:PATH...]:PATH\n"

" Create an overlay mount from the host to \n"

" the container\n"

" --overlay-ro=PATH[:PATH...]:PATH\n"

" Similar, but creates a read-only overlay mount\n"

" --setenv=NAME=VALUE Pass an environment variable to PID 1\n"

" --share-system Share system namespaces with host\n"

" --register=BOOLEAN Register container as machine\n"

" --keep-unit Do not register a scope for the machine, reuse\n"

" the service unit nspawn is running in\n"

" --volatile[=MODE] Run the system in volatile mode\n"

, program_invocation_short_name);

}

static CustomMount* custom_mount_add(CustomMountType t) {

CustomMount *c, *ret;

c = realloc(arg_custom_mounts, (arg_n_custom_mounts + 1) * sizeof(CustomMount));

if (!c)

return NULL;

arg_custom_mounts = c;

ret = arg_custom_mounts + arg_n_custom_mounts;

arg_n_custom_mounts++;

*ret = (CustomMount) { .type = t };

return ret;

}

static void custom_mount_free_all(void) {

unsigned i;

for (i = 0; i < arg_n_custom_mounts; i++) {

CustomMount *m = &arg_custom_mounts[i];

free(m->source);

free(m->destination);

free(m->options);

if (m->work_dir) {

(void) rm_rf(m->work_dir, REMOVE_ROOT|REMOVE_PHYSICAL);

free(m->work_dir);

}

strv_free(m->lower);

}

arg_custom_mounts = mfree(arg_custom_mounts);

arg_n_custom_mounts = 0;

}

static int custom_mount_compare(const void *a, const void *b) {

const CustomMount *x = a, *y = b;

int r;

r = path_compare(x->destination, y->destination);

if (r != 0)

return r;

if (x->type < y->type)

return -1;

if (x->type > y->type)

return 1;

return 0;

}

static int custom_mounts_prepare(void) {

unsigned i;

int r;

/* Ensure the mounts are applied prefix first. */

qsort_safe(arg_custom_mounts, arg_n_custom_mounts, sizeof(CustomMount), custom_mount_compare);

/* Allocate working directories for the overlay file systems that need it */

for (i = 0; i < arg_n_custom_mounts; i++) {

CustomMount *m = &arg_custom_mounts[i];

if (arg_userns && arg_uid_shift == UID_INVALID && path_equal(m->destination, "/")) {

log_error("--private-users with automatic UID shift may not be combined with custom root mounts.");

return -EINVAL;

}

if (m->type != CUSTOM_MOUNT_OVERLAY)

continue;

if (m->work_dir)

continue;

if (m->read_only)

continue;

r = tempfn_random(m->source, NULL, &m->work_dir);

if (r < 0)

return log_error_errno(r, "Failed to generate work directory from %s: %m", m->source);

}

return 0;

}

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 detect_unified_cgroup_hierarchy(void) {

const char *e;

int r;

/* Allow the user to control whether the unified hierarchy is used */

e = getenv("UNIFIED_CGROUP_HIERARCHY");

if (e) {

r = parse_boolean(e);

if (r < 0)

return log_error_errno(r, "Failed to parse $UNIFIED_CGROUP_HIERARCHY.");

arg_unified_cgroup_hierarchy = r;

return 0;

}

/* Otherwise inherit the default from the host system */

r = cg_unified();

if (r < 0)

return log_error_errno(r, "Failed to determine whether the unified cgroups hierarchy is used: %m");

arg_unified_cgroup_hierarchy = r;

return 0;

}

static int parse_argv(int argc, char *argv[]) {

enum {

ARG_VERSION = 0x100,

ARG_PRIVATE_NETWORK,

ARG_UUID,

ARG_READ_ONLY,

ARG_CAPABILITY,

ARG_DROP_CAPABILITY,

ARG_LINK_JOURNAL,

ARG_BIND,

ARG_BIND_RO,

ARG_TMPFS,

ARG_OVERLAY,

ARG_OVERLAY_RO,

ARG_SETENV,

ARG_SHARE_SYSTEM,

ARG_REGISTER,

ARG_KEEP_UNIT,

ARG_NETWORK_INTERFACE,

ARG_NETWORK_MACVLAN,

ARG_NETWORK_IPVLAN,

ARG_NETWORK_BRIDGE,

ARG_PERSONALITY,

ARG_VOLATILE,

ARG_TEMPLATE,

ARG_PROPERTY,

ARG_PRIVATE_USERS,

ARG_KILL_SIGNAL,

};

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' },

{ "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 },

{ "tmpfs", required_argument, NULL, ARG_TMPFS },

{ "overlay", required_argument, NULL, ARG_OVERLAY },

{ "overlay-ro", required_argument, NULL, ARG_OVERLAY_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 },

{ "register", required_argument, NULL, ARG_REGISTER },

{ "keep-unit", no_argument, NULL, ARG_KEEP_UNIT },

{ "network-interface", required_argument, NULL, ARG_NETWORK_INTERFACE },

{ "network-macvlan", required_argument, NULL, ARG_NETWORK_MACVLAN },

{ "network-ipvlan", required_argument, NULL, ARG_NETWORK_IPVLAN },

{ "network-veth", no_argument, NULL, 'n' },

{ "network-bridge", required_argument, NULL, ARG_NETWORK_BRIDGE },

{ "personality", required_argument, NULL, ARG_PERSONALITY },

{ "image", required_argument, NULL, 'i' },

{ "volatile", optional_argument, NULL, ARG_VOLATILE },

{ "port", required_argument, NULL, 'p' },

{ "property", required_argument, NULL, ARG_PROPERTY },

{ "private-users", optional_argument, NULL, ARG_PRIVATE_USERS },

{ "kill-signal", required_argument, NULL, ARG_KILL_SIGNAL },

{}

};

int c, r;

uint64_t plus = 0, minus = 0;

assert(argc >= 0);

assert(argv);

while ((c = getopt_long(argc, argv, "+hD:u:bL:M:jS:Z:qi:xp:n", options, NULL)) >= 0)

switch (c) {

case 'h':

help();

return 0;

case ARG_VERSION:

puts(PACKAGE_STRING);

puts(SYSTEMD_FEATURES);

return 0;

case 'D':

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':

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':

r = free_and_strdup(&arg_user, optarg);

if (r < 0)

return log_oom();

break;

case ARG_NETWORK_BRIDGE:

arg_network_bridge = optarg;

/* fall through */

case 'n':

arg_network_veth = true;

arg_private_network = true;

break;

case ARG_NETWORK_INTERFACE:

if (strv_extend(&arg_network_interfaces, optarg) < 0)

return log_oom();

arg_private_network = true;

break;

case ARG_NETWORK_MACVLAN:

if (strv_extend(&arg_network_macvlan, optarg) < 0)

return log_oom();

arg_private_network = true;

break;

case ARG_NETWORK_IPVLAN:

if (strv_extend(&arg_network_ipvlan, optarg) < 0)

return log_oom();

/* fall through */

case ARG_PRIVATE_NETWORK:

arg_private_network = true;

break;

case 'b':

arg_boot = true;

break;

case ARG_UUID:

r = sd_id128_from_string(optarg, &arg_uuid);

if (r < 0) {

log_error("Invalid UUID: %s", optarg);

return r;

}

break;

case 'S':

arg_slice = optarg;

break;

case 'M':

if (isempty(optarg))

arg_machine = mfree(arg_machine);

else {

if (!machine_name_is_valid(optarg)) {

log_error("Invalid machine name: %s", optarg);

return -EINVAL;

}

r = free_and_strdup(&arg_machine, optarg);

if (r < 0)

return log_oom();

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;

case ARG_CAPABILITY:

case ARG_DROP_CAPABILITY: {

const char *state, *word;

size_t length;

FOREACH_WORD_SEPARATOR(word, length, optarg, ",", state) {

_cleanup_free_ char *t;

t = strndup(word, length);

if (!t)

return log_oom();

if (streq(t, "all")) {

if (c == ARG_CAPABILITY)

plus = (uint64_t) -1;

else

minus = (uint64_t) -1;

} else {

int cap;

cap = capability_from_name(t);

if (cap < 0) {

log_error("Failed to parse capability %s.", t);

return -EINVAL;

}

if (c == ARG_CAPABILITY)

plus |= 1ULL << (uint64_t) cap;

else

minus |= 1ULL << (uint64_t) cap;

}

}

break;

}

case 'j':

arg_link_journal = LINK_GUEST;

arg_link_journal_try = true;

break;

case ARG_LINK_JOURNAL:

if (streq(optarg, "auto")) {

arg_link_journal = LINK_AUTO;

arg_link_journal_try = false;

} else if (streq(optarg, "no")) {

arg_link_journal = LINK_NO;

arg_link_journal_try = false;

} else if (streq(optarg, "guest")) {

arg_link_journal = LINK_GUEST;

arg_link_journal_try = false;

} else if (streq(optarg, "host")) {

arg_link_journal = LINK_HOST;

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")) {

arg_link_journal = LINK_HOST;

arg_link_journal_try = true;

} else {

log_error("Failed to parse link journal mode %s", optarg);

return -EINVAL;

}

break;

case ARG_BIND:

case ARG_BIND_RO: {

const char *current = optarg;

_cleanup_free_ char *source = NULL, *destination = NULL, *opts = NULL;

CustomMount *m;

r = extract_many_words(&current, ":", EXTRACT_DONT_COALESCE_SEPARATORS, &source, &destination, &opts, NULL);

switch (r) {

case 1:

destination = strdup(source);

case 2:

case 3:

break;

case -ENOMEM:

return log_oom();

default:

log_error("Invalid bind mount specification: %s", optarg);

return -EINVAL;

}

if (!source || !destination)

return log_oom();

if (!path_is_absolute(source) || !path_is_absolute(destination)) {

log_error("Invalid bind mount specification: %s", optarg);

return -EINVAL;

}

m = custom_mount_add(CUSTOM_MOUNT_BIND);

if (!m)

return log_oom();

m->source = source;

m->destination = destination;

m->read_only = c == ARG_BIND_RO;

m->options = opts;

source = destination = opts = NULL;

break;

}

case ARG_TMPFS: {

const char *current = optarg;

_cleanup_free_ char *path = NULL, *opts = NULL;

CustomMount *m;

r = extract_first_word(&current, &path, ":", EXTRACT_DONT_COALESCE_SEPARATORS);

if (r == -ENOMEM)

return log_oom();

else if (r < 0) {

log_error("Invalid tmpfs specification: %s", optarg);

return r;

}

if (r)

opts = strdup(current);

else

opts = strdup("mode=0755");

if (!path || !opts)

return log_oom();

if (!path_is_absolute(path)) {

log_error("Invalid tmpfs specification: %s", optarg);

return -EINVAL;

}

m = custom_mount_add(CUSTOM_MOUNT_TMPFS);

if (!m)

return log_oom();

m->destination = path;

m->options = opts;

path = opts = NULL;

break;

}

case ARG_OVERLAY:

case ARG_OVERLAY_RO: {

_cleanup_free_ char *upper = NULL, *destination = NULL;

_cleanup_strv_free_ char **lower = NULL;

CustomMount *m;

unsigned n = 0;

char **i;

r = strv_split_extract(&lower, optarg, ":", EXTRACT_DONT_COALESCE_SEPARATORS);

if (r == -ENOMEM)

return log_oom();

else if (r < 0) {

log_error("Invalid overlay specification: %s", optarg);

return r;

}

STRV_FOREACH(i, lower) {

if (!path_is_absolute(*i)) {

log_error("Overlay path %s is not absolute.", *i);

return -EINVAL;

}

n++;

}

if (n < 2) {

log_error("--overlay= needs at least two colon-separated directories specified.");

return -EINVAL;

}

if (n == 2) {

/* If two parameters are specified,

upper = lower[1];

lower[1] = NULL;

destination = strdup(upper);

if (!destination)

return log_oom();

} else {

upper = lower[n - 2];

destination = lower[n - 1];

lower[n - 2] = NULL;

}

m = custom_mount_add(CUSTOM_MOUNT_OVERLAY);

if (!m)

return log_oom();

m->destination = destination;

m->source = upper;

m->lower = lower;

m->read_only = c == ARG_OVERLAY_RO;

upper = destination = NULL;

lower = NULL;

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 ARG_REGISTER:

r = parse_boolean(optarg);

if (r < 0) {

log_error("Failed to parse --register= argument: %s", optarg);

return r;

}

arg_register = r;

break;

case ARG_KEEP_UNIT:

arg_keep_unit = true;

break;

case ARG_PERSONALITY:

arg_personality = personality_from_string(optarg);

if (arg_personality == PERSONALITY_INVALID) {

log_error("Unknown or unsupported personality '%s'.", optarg);

return -EINVAL;

}

break;

case ARG_VOLATILE:

if (!optarg)

arg_volatile = VOLATILE_YES;

else {

r = parse_boolean(optarg);

if (r < 0) {

if (streq(optarg, "state"))

arg_volatile = VOLATILE_STATE;

else {

log_error("Failed to parse --volatile= argument: %s", optarg);

return r;

}

} else

arg_volatile = r ? VOLATILE_YES : VOLATILE_NO;

}

break;

case 'p': {

const char *split, *e;

uint16_t container_port, host_port;

int protocol;

ExposePort *p;

if ((e = startswith(optarg, "tcp:")))

protocol = IPPROTO_TCP;

else if ((e = startswith(optarg, "udp:")))

protocol = IPPROTO_UDP;

else {

e = optarg;

protocol = IPPROTO_TCP;

}

split = strchr(e, ':');

if (split) {

char v[split - e + 1];

memcpy(v, e, split - e);

v[split - e] = 0;

r = safe_atou16(v, &host_port);

if (r < 0 || host_port <= 0) {

log_error("Failed to parse host port: %s", optarg);

return -EINVAL;

}

r = safe_atou16(split + 1, &container_port);

} else {

r = safe_atou16(e, &container_port);

host_port = container_port;

}

if (r < 0 || container_port <= 0) {

log_error("Failed to parse host port: %s", optarg);

return -EINVAL;

}

LIST_FOREACH(ports, p, arg_expose_ports) {

if (p->protocol == protocol && p->host_port == host_port) {

log_error("Duplicate port specification: %s", optarg);

return -EINVAL;

}

}

p = new(ExposePort, 1);

if (!p)

return log_oom();

p->protocol = protocol;

p->host_port = host_port;

p->container_port = container_port;

LIST_PREPEND(ports, arg_expose_ports, p);

break;

}

case ARG_PROPERTY:

if (strv_extend(&arg_property, optarg) < 0)

return log_oom();

break;

case ARG_PRIVATE_USERS:

if (optarg) {

_cleanup_free_ char *buffer = NULL;

const char *range, *shift;

range = strchr(optarg, ':');

if (range) {

buffer = strndup(optarg, range - optarg);

if (!buffer)

return log_oom();

shift = buffer;

range++;

if (safe_atou32(range, &arg_uid_range) < 0 || arg_uid_range <= 0) {

log_error("Failed to parse UID range: %s", range);

return -EINVAL;

}

} else

shift = optarg;

if (parse_uid(shift, &arg_uid_shift) < 0) {

log_error("Failed to parse UID: %s", optarg);

return -EINVAL;

}

}

arg_userns = true;

break;

case ARG_KILL_SIGNAL:

arg_kill_signal = signal_from_string_try_harder(optarg);

if (arg_kill_signal < 0) {

log_error("Cannot parse signal: %s", optarg);

return -EINVAL;

}

break;

case '?':

return -EINVAL;

default:

assert_not_reached("Unhandled option");

}

if (arg_share_system)

arg_register = false;

if (arg_boot && arg_share_system) {

log_error("--boot and --share-system may not be combined.");

return -EINVAL;

}

if (arg_keep_unit && cg_pid_get_owner_uid(0, NULL) >= 0) {

log_error("--keep-unit may not be used when invoked from a user session.");

return -EINVAL;

}

if (arg_directory && arg_image) {

log_error("--directory= and --image= may not be combined.");

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;

}

if (arg_expose_ports && !arg_private_network) {

log_error("Cannot use --port= without private networking.");

return -EINVAL;

}

if (arg_userns && access("/proc/self/uid_map", F_OK) < 0)

return log_error_errno(EOPNOTSUPP, "--private-users= is not supported, kernel compiled without user namespace support.");

arg_retain = (arg_retain | plus | (arg_private_network ? 1ULL << CAP_NET_ADMIN : 0)) & ~minus;

if (arg_boot && arg_kill_signal <= 0)

arg_kill_signal = SIGRTMIN+3;

r = detect_unified_cgroup_hierarchy();

if (r < 0)

return r;

return 1;

}

static int tmpfs_patch_options(const char *options, char **ret) {

char *buf = NULL;

if (arg_userns && arg_uid_shift != 0) {

assert(arg_uid_shift != UID_INVALID);

if (options)

(void) asprintf(&buf, "%s,uid=" UID_FMT ",gid=" UID_FMT, options, arg_uid_shift, arg_uid_shift);

else

(void) asprintf(&buf, "uid=" UID_FMT ",gid=" UID_FMT, arg_uid_shift, arg_uid_shift);

if (!buf)

return -ENOMEM;

options = buf;

}

#ifdef HAVE_SELINUX

if (arg_selinux_apifs_context) {

char *t;

if (options)

t = strjoin(options, ",context=\"", arg_selinux_apifs_context, "\"", NULL);

else

t = strjoin("context=\"", arg_selinux_apifs_context, "\"", NULL);

if (!t) {

free(buf);

return -ENOMEM;

}

free(buf);

buf = t;

}

#endif

*ret = buf;

return !!buf;

}

static int mount_all(const char *dest, bool userns) {

typedef struct MountPoint {

const char *what;

const char *where;

const char *type;

const char *options;

unsigned long flags;

bool fatal;

bool userns;

} MountPoint;

static const MountPoint mount_table[] = {

{ "proc", "/proc", "proc", NULL, MS_NOSUID|MS_NOEXEC|MS_NODEV, true, true },

{ "/proc/sys", "/proc/sys", NULL, NULL, MS_BIND, true, true }, /* Bind mount first */

{ NULL, "/proc/sys", NULL, NULL, MS_BIND|MS_RDONLY|MS_NOSUID|MS_NOEXEC|MS_NODEV|MS_REMOUNT, true, true }, /* Then, make it r/o */

{ "sysfs", "/sys", "sysfs", NULL, MS_RDONLY|MS_NOSUID|MS_NOEXEC|MS_NODEV, true, false },

{ "tmpfs", "/dev", "tmpfs", "mode=755", MS_NOSUID|MS_STRICTATIME, true, false },

{ "tmpfs", "/dev/shm", "tmpfs", "mode=1777", MS_NOSUID|MS_NODEV|MS_STRICTATIME, true, false },

{ "tmpfs", "/run", "tmpfs", "mode=755", MS_NOSUID|MS_NODEV|MS_STRICTATIME, true, false },

{ "tmpfs", "/tmp", "tmpfs", "mode=1777", MS_STRICTATIME, true, false },

#ifdef HAVE_SELINUX

{ "/sys/fs/selinux", "/sys/fs/selinux", NULL, NULL, MS_BIND, false, false }, /* Bind mount first */

{ NULL, "/sys/fs/selinux", NULL, NULL, MS_BIND|MS_RDONLY|MS_NOSUID|MS_NOEXEC|MS_NODEV|MS_REMOUNT, false, false }, /* Then, make it r/o */

#endif

};

unsigned k;

int r;

for (k = 0; k < ELEMENTSOF(mount_table); k++) {

_cleanup_free_ char *where = NULL, *options = NULL;

const char *o;

if (userns != mount_table[k].userns)

continue;

where = prefix_root(dest, mount_table[k].where);

if (!where)

return log_oom();

r = path_is_mount_point(where, AT_SYMLINK_FOLLOW);

if (r < 0 && r != -ENOENT)

return log_error_errno(r, "Failed to detect whether %s is a mount point: %m", where);

/* Skip this entry if it is not a remount. */

if (mount_table[k].what && r > 0)

continue;

r = mkdir_p(where, 0755);

if (r < 0) {

if (mount_table[k].fatal)

return log_error_errno(r, "Failed to create directory %s: %m", where);

log_warning_errno(r, "Failed to create directory %s: %m", where);

continue;

}

o = mount_table[k].options;

if (streq_ptr(mount_table[k].type, "tmpfs")) {

r = tmpfs_patch_options(o, &options);

if (r < 0)

return log_oom();

if (r > 0)

o = options;

}

if (mount(mount_table[k].what,

where,

mount_table[k].type,

mount_table[k].flags,

o) < 0) {

if (mount_table[k].fatal)

return log_error_errno(errno, "mount(%s) failed: %m", where);

log_warning_errno(errno, "mount(%s) failed, ignoring: %m", where);

}

}

return 0;

}

static int parse_mount_bind_options(const char *options, unsigned long *mount_flags, char **mount_opts) {

const char *p = options;

unsigned long flags = *mount_flags;

char *opts = NULL;

assert(options);

for (;;) {

_cleanup_free_ char *word = NULL;

int r = extract_first_word(&p, &word, ",", 0);

if (r < 0)

return log_error_errno(r, "Failed to extract mount option: %m");

if (r == 0)

break;

if (streq(word, "rbind"))

flags |= MS_REC;

else if (streq(word, "norbind"))

flags &= ~MS_REC;

else {

log_error("Invalid bind mount option: %s", word);

return -EINVAL;

}

}

*mount_flags = flags;

/* in the future mount_opts will hold string options for mount(2) */

*mount_opts = opts;

return 0;

}

static int mount_bind(const char *dest, CustomMount *m) {

struct stat source_st, dest_st;

const char *where;

unsigned long mount_flags = MS_BIND | MS_REC;

_cleanup_free_ char *mount_opts = NULL;

int r;

assert(m);

if (m->options) {

r = parse_mount_bind_options(m->options, &mount_flags, &mount_opts);

if (r < 0)

return r;

}

if (stat(m->source, &source_st) < 0)

return log_error_errno(errno, "Failed to stat %s: %m", m->source);

where = prefix_roota(dest, m->destination);

if (stat(where, &dest_st) >= 0) {

if (S_ISDIR(source_st.st_mode) && !S_ISDIR(dest_st.st_mode)) {

log_error("Cannot bind mount directory %s on file %s.", m->source, where);

return -EINVAL;

}

if (!S_ISDIR(source_st.st_mode) && S_ISDIR(dest_st.st_mode)) {

log_error("Cannot bind mount file %s on directory %s.", m->source, where);

return -EINVAL;

}

} else if (errno == ENOENT) {

r = mkdir_parents_label(where, 0755);

if (r < 0)

return log_error_errno(r, "Failed to make parents of %s: %m", where);

} else {

log_error_errno(errno, "Failed to stat %s: %m", where);

return -errno;

}

/* Create the mount point. Any non-directory file can be

if (S_ISDIR(source_st.st_mode))

r = mkdir_label(where, 0755);

else

r = touch(where);

if (r < 0 && r != -EEXIST)

return log_error_errno(r, "Failed to create mount point %s: %m", where);

if (mount(m->source, where, NULL, mount_flags, mount_opts) < 0)

return log_error_errno(errno, "mount(%s) failed: %m", where);

if (m->read_only) {

r = bind_remount_recursive(where, true);

if (r < 0)

return log_error_errno(r, "Read-only bind mount failed: %m");

}

return 0;

}

static int mount_tmpfs(const char *dest, CustomMount *m) {

const char *where, *options;

_cleanup_free_ char *buf = NULL;

int r;

assert(dest);

assert(m);

where = prefix_roota(dest, m->destination);

r = mkdir_p_label(where, 0755);

if (r < 0 && r != -EEXIST)

return log_error_errno(r, "Creating mount point for tmpfs %s failed: %m", where);

r = tmpfs_patch_options(m->options, &buf);

if (r < 0)

return log_oom();

options = r > 0 ? buf : m->options;

if (mount("tmpfs", where, "tmpfs", MS_NODEV|MS_STRICTATIME, options) < 0)

return log_error_errno(errno, "tmpfs mount to %s failed: %m", where);

return 0;

}

static char *joined_and_escaped_lower_dirs(char * const *lower) {

_cleanup_strv_free_ char **sv = NULL;

sv = strv_copy(lower);

if (!sv)

return NULL;

strv_reverse(sv);

if (!strv_shell_escape(sv, ",:"))

return NULL;

return strv_join(sv, ":");

}

static int mount_overlay(const char *dest, CustomMount *m) {

_cleanup_free_ char *lower = NULL;

const char *where, *options;

int r;

assert(dest);

assert(m);

where = prefix_roota(dest, m->destination);

r = mkdir_label(where, 0755);

if (r < 0 && r != -EEXIST)

return log_error_errno(r, "Creating mount point for overlay %s failed: %m", where);

(void) mkdir_p_label(m->source, 0755);

lower = joined_and_escaped_lower_dirs(m->lower);

if (!lower)

return log_oom();

if (m->read_only) {

_cleanup_free_ char *escaped_source = NULL;

escaped_source = shell_escape(m->source, ",:");

if (!escaped_source)

return log_oom();

options = strjoina("lowerdir=", escaped_source, ":", lower);

} else {

_cleanup_free_ char *escaped_source = NULL, *escaped_work_dir = NULL;

assert(m->work_dir);

(void) mkdir_label(m->work_dir, 0700);

escaped_source = shell_escape(m->source, ",:");

if (!escaped_source)

return log_oom();

escaped_work_dir = shell_escape(m->work_dir, ",:");

if (!escaped_work_dir)

return log_oom();

options = strjoina("lowerdir=", lower, ",upperdir=", escaped_source, ",workdir=", escaped_work_dir);

}

if (mount("overlay", where, "overlay", m->read_only ? MS_RDONLY : 0, options) < 0)

return log_error_errno(errno, "overlay mount to %s failed: %m", where);

return 0;

}

static int mount_custom(const char *dest) {

unsigned i;

int r;

assert(dest);

for (i = 0; i < arg_n_custom_mounts; i++) {

CustomMount *m = &arg_custom_mounts[i];

switch (m->type) {

case CUSTOM_MOUNT_BIND:

r = mount_bind(dest, m);

break;

case CUSTOM_MOUNT_TMPFS:

r = mount_tmpfs(dest, m);

break;

case CUSTOM_MOUNT_OVERLAY:

r = mount_overlay(dest, m);

break;

default:

assert_not_reached("Unknown custom mount type");

}

if (r < 0)

return r;

}

return 0;

}

static int mount_legacy_cgroup_hierarchy(const char *dest, const char *controller, const char *hierarchy, bool read_only) {

char *to;

int r;

to = strjoina(dest, "/sys/fs/cgroup/", hierarchy);

r = path_is_mount_point(to, 0);

if (r < 0 && r != -ENOENT)

return log_error_errno(r, "Failed to determine if %s is mounted already: %m", to);

if (r > 0)

return 0;

mkdir_p(to, 0755);

/* The superblock mount options of the mount point need to be

if (mount("cgroup", to, "cgroup", MS_NOSUID|MS_NOEXEC|MS_NODEV, controller) < 0)

return log_error_errno(errno, "Failed to mount to %s: %m", to);

/* ... hence let's only make the bind mount read-only, not the

if (read_only) {

if (mount(NULL, to, NULL, MS_BIND|MS_REMOUNT|MS_NOSUID|MS_NOEXEC|MS_NODEV|MS_RDONLY, NULL) < 0)

return log_error_errno(errno, "Failed to remount %s read-only: %m", to);

}

return 1;

}

static int mount_legacy_cgroups(const char *dest) {

_cleanup_set_free_free_ Set *controllers = NULL;

const char *cgroup_root;

int r;

cgroup_root = prefix_roota(dest, "/sys/fs/cgroup");

/* Mount a tmpfs to /sys/fs/cgroup if it's not mounted there yet. */

r = path_is_mount_point(cgroup_root, AT_SYMLINK_FOLLOW);

if (r < 0)

return log_error_errno(r, "Failed to determine if /sys/fs/cgroup is already mounted: %m");

if (r == 0) {

_cleanup_free_ char *options = NULL;

r = tmpfs_patch_options("mode=755", &options);

if (r < 0)

return log_oom();

if (mount("tmpfs", cgroup_root, "tmpfs", MS_NOSUID|MS_NOEXEC|MS_NODEV|MS_STRICTATIME, options) < 0)

return log_error_errno(errno, "Failed to mount /sys/fs/cgroup: %m");

}

if (cg_unified() > 0)

goto skip_controllers;

controllers = set_new(&string_hash_ops);

if (!controllers)

return log_oom();

r = cg_kernel_controllers(controllers);

if (r < 0)

return log_error_errno(r, "Failed to determine cgroup controllers: %m");

for (;;) {

_cleanup_free_ char *controller = NULL, *origin = NULL, *combined = NULL;

controller = set_steal_first(controllers);

if (!controller)

break;

origin = prefix_root("/sys/fs/cgroup/", controller);

if (!origin)

return log_oom();

r = readlink_malloc(origin, &combined);

if (r == -EINVAL) {

/* Not a symbolic link, but directly a single cgroup hierarchy */

r = mount_legacy_cgroup_hierarchy(dest, controller, controller, true);

if (r < 0)

return r;

} else if (r < 0)

return log_error_errno(r, "Failed to read link %s: %m", origin);

else {

_cleanup_free_ char *target = NULL;

target = prefix_root(dest, origin);

if (!target)

return log_oom();

/* A symbolic link, a combination of controllers in one hierarchy */

if (!filename_is_valid(combined)) {

log_warning("Ignoring invalid combined hierarchy %s.", combined);

continue;

}

r = mount_legacy_cgroup_hierarchy(dest, combined, combined, true);

if (r < 0)

return r;

r = symlink_idempotent(combined, target);

if (r == -EINVAL) {

log_error("Invalid existing symlink for combined hierarchy");

return r;

}

if (r < 0)

return log_error_errno(r, "Failed to create symlink for combined hierarchy: %m");

}

}

skip_controllers:

r = mount_legacy_cgroup_hierarchy(dest, "none,name=systemd,xattr", "systemd", false);

if (r < 0)

return r;

if (mount(NULL, cgroup_root, NULL, MS_REMOUNT|MS_NOSUID|MS_NOEXEC|MS_NODEV|MS_STRICTATIME|MS_RDONLY, "mode=755") < 0)

return log_error_errno(errno, "Failed to remount %s read-only: %m", cgroup_root);

return 0;

}

static int mount_unified_cgroups(const char *dest) {

const char *p;

int r;

assert(dest);

p = strjoina(dest, "/sys/fs/cgroup");

r = path_is_mount_point(p, AT_SYMLINK_FOLLOW);

if (r < 0)

return log_error_errno(r, "Failed to determine if %s is mounted already: %m", p);

if (r > 0) {

p = strjoina(dest, "/sys/fs/cgroup/cgroup.procs");

if (access(p, F_OK) >= 0)

return 0;

if (errno != ENOENT)

return log_error_errno(errno, "Failed to determine if mount point %s contains the unified cgroup hierarchy: %m", p);

log_error("%s is already mounted but not a unified cgroup hierarchy. Refusing.", p);

return -EINVAL;

}

if (mount("cgroup", p, "cgroup", MS_NOSUID|MS_NOEXEC|MS_NODEV, "__DEVEL__sane_behavior") < 0)

return log_error_errno(errno, "Failed to mount unified cgroup hierarchy to %s: %m", p);

return 0;

}

static int mount_cgroups(const char *dest) {

if (arg_unified_cgroup_hierarchy)

return mount_unified_cgroups(dest);

else

return mount_legacy_cgroups(dest);

}

static int mount_systemd_cgroup_writable(const char *dest) {

_cleanup_free_ char *own_cgroup_path = NULL;

const char *systemd_root, *systemd_own;

int r;

assert(dest);

r = cg_pid_get_path(NULL, 0, &own_cgroup_path);

if (r < 0)

return log_error_errno(r, "Failed to determine our own cgroup path: %m");

/* If we are living in the top-level, then there's nothing to do... */

if (path_equal(own_cgroup_path, "/"))

return 0;

if (arg_unified_cgroup_hierarchy) {

systemd_own = strjoina(dest, "/sys/fs/cgroup", own_cgroup_path);

systemd_root = prefix_roota(dest, "/sys/fs/cgroup");

} else {

systemd_own = strjoina(dest, "/sys/fs/cgroup/systemd", own_cgroup_path);

systemd_root = prefix_roota(dest, "/sys/fs/cgroup/systemd");

}

/* Make our own cgroup a (writable) bind mount */

if (mount(systemd_own, systemd_own, NULL, MS_BIND, NULL) < 0)

return log_error_errno(errno, "Failed to turn %s into a bind mount: %m", own_cgroup_path);

/* And then remount the systemd cgroup root read-only */

if (mount(NULL, systemd_root, NULL, MS_BIND|MS_REMOUNT|MS_NOSUID|MS_NOEXEC|MS_NODEV|MS_RDONLY, NULL) < 0)

return log_error_errno(errno, "Failed to mount cgroup root read-only: %m");

return 0;

}

static int userns_lchown(const char *p, uid_t uid, gid_t gid) {

assert(p);

if (!arg_userns)

return 0;

if (uid == UID_INVALID && gid == GID_INVALID)

return 0;

if (uid != UID_INVALID) {

uid += arg_uid_shift;

if (uid < arg_uid_shift || uid >= arg_uid_shift + arg_uid_range)

return -EOVERFLOW;

}

if (gid != GID_INVALID) {

gid += (gid_t) arg_uid_shift;

if (gid < (gid_t) arg_uid_shift || gid >= (gid_t) (arg_uid_shift + arg_uid_range))

return -EOVERFLOW;

}

if (lchown(p, uid, gid) < 0)

return -errno;

return 0;

}

static int userns_mkdir(const char *root, const char *path, mode_t mode, uid_t uid, gid_t gid) {

const char *q;

q = prefix_roota(root, path);

if (mkdir(q, mode) < 0) {

if (errno == EEXIST)

return 0;

return -errno;

}

return userns_lchown(q, uid, gid);

}

static int setup_timezone(const char *dest) {

_cleanup_free_ char *p = NULL, *q = NULL;

const char *where, *check, *what;

char *z, *y;

int r;

assert(dest);

/* Fix the timezone, if possible */

r = readlink_malloc("/etc/localtime", &p);

if (r < 0) {

log_warning("/etc/localtime is not a symlink, not updating container timezone.");

return 0;

}

z = path_startswith(p, "../usr/share/zoneinfo/");

if (!z)

z = path_startswith(p, "/usr/share/zoneinfo/");

if (!z) {

log_warning("/etc/localtime does not point into /usr/share/zoneinfo/, not updating container timezone.");

return 0;

}

where = prefix_roota(dest, "/etc/localtime");

r = readlink_malloc(where, &q);

if (r >= 0) {

y = path_startswith(q, "../usr/share/zoneinfo/");

if (!y)

y = path_startswith(q, "/usr/share/zoneinfo/");

/* Already pointing to the right place? Then do nothing .. */

if (y && streq(y, z))

return 0;

}

check = strjoina("/usr/share/zoneinfo/", z);

check = prefix_root(dest, check);

if (laccess(check, F_OK) < 0) {

log_warning("Timezone %s does not exist in container, not updating container timezone.", z);

return 0;

}

r = unlink(where);

if (r < 0 && errno != ENOENT) {

log_error_errno(errno, "Failed to remove existing timezone info %s in container: %m", where);

return 0;

}

what = strjoina("../usr/share/zoneinfo/", z);

if (symlink(what, where) < 0) {

log_error_errno(errno, "Failed to correct timezone of container: %m");

return 0;

}

r = userns_lchown(where, 0, 0);

if (r < 0)

return log_warning_errno(r, "Failed to chown /etc/localtime: %m");

return 0;

}

static int setup_resolv_conf(const char *dest) {

const char *where = NULL;

int r;

assert(dest);

if (arg_private_network)

return 0;

/* Fix resolv.conf, if possible */

where = prefix_roota(dest, "/etc/resolv.conf");

r = copy_file("/etc/resolv.conf", where, O_TRUNC|O_NOFOLLOW, 0644, 0);

if (r < 0) {

/* If the file already exists as symlink, let's

log_full_errno(IN_SET(r, -ELOOP, -EROFS) ? LOG_DEBUG : LOG_WARNING, r,

"Failed to copy /etc/resolv.conf to %s: %m", where);

return 0;

}

r = userns_lchown(where, 0, 0);

if (r < 0)

log_warning_errno(r, "Failed to chown /etc/resolv.conf: %m");

return 0;

}

static int setup_volatile_state(const char *directory) {

_cleanup_free_ char *buf = NULL;

const char *p, *options;

int r;

assert(directory);

if (arg_volatile != VOLATILE_STATE)

return 0;

/* --volatile=state means we simply overmount /var

r = bind_remount_recursive(directory, true);

if (r < 0)

return log_error_errno(r, "Failed to remount %s read-only: %m", directory);

p = prefix_roota(directory, "/var");

r = mkdir(p, 0755);

if (r < 0 && errno != EEXIST)

return log_error_errno(errno, "Failed to create %s: %m", directory);

options = "mode=755";

r = tmpfs_patch_options(options, &buf);

if (r < 0)

return log_oom();

if (r > 0)

options = buf;

if (mount("tmpfs", p, "tmpfs", MS_STRICTATIME, options) < 0)

return log_error_errno(errno, "Failed to mount tmpfs to /var: %m");

return 0;

}

static int setup_volatile(const char *directory) {

bool tmpfs_mounted = false, bind_mounted = false;

char template[] = "/tmp/nspawn-volatile-XXXXXX";

_cleanup_free_ char *buf = NULL;

const char *f, *t, *options;

int r;

assert(directory);

if (arg_volatile != VOLATILE_YES)

return 0;

/* --volatile=yes means we mount a tmpfs to the root dir, and

if (!mkdtemp(template))

return log_error_errno(errno, "Failed to create temporary directory: %m");

options = "mode=755";

r = tmpfs_patch_options(options, &buf);

if (r < 0)

return log_oom();

if (r > 0)

options = buf;

if (mount("tmpfs", template, "tmpfs", MS_STRICTATIME, options) < 0) {

r = log_error_errno(errno, "Failed to mount tmpfs for root directory: %m");

goto fail;

}

tmpfs_mounted = true;

f = prefix_roota(directory, "/usr");

t = prefix_roota(template, "/usr");

r = mkdir(t, 0755);

if (r < 0 && errno != EEXIST) {

r = log_error_errno(errno, "Failed to create %s: %m", t);

goto fail;

}

if (mount(f, t, NULL, MS_BIND|MS_REC, NULL) < 0) {

r = log_error_errno(errno, "Failed to create /usr bind mount: %m");

goto fail;

}

bind_mounted = true;

r = bind_remount_recursive(t, true);

if (r < 0) {

log_error_errno(r, "Failed to remount %s read-only: %m", t);

goto fail;

}

if (mount(template, directory, NULL, MS_MOVE, NULL) < 0) {

r = log_error_errno(errno, "Failed to move root mount: %m");

goto fail;

}

(void) rmdir(template);

return 0;

fail:

if (bind_mounted)

(void) umount(t);

if (tmpfs_mounted)

(void) umount(template);

(void) rmdir(template);

return r;

}

static char* id128_format_as_uuid(sd_id128_t id, char s[37]) {

assert(s);

snprintf(s, 37,

"%02x%02x%02x%02x-%02x%02x-%02x%02x-%02x%02x-%02x%02x%02x%02x%02x%02x",

SD_ID128_FORMAT_VAL(id));

return s;

}

static int setup_boot_id(const char *dest) {

const char *from, *to;

sd_id128_t rnd = {};

char as_uuid[37];

int r;

if (arg_share_system)

return 0;

/* Generate a new randomized boot ID, so that each boot-up of

from = prefix_roota(dest, "/run/proc-sys-kernel-random-boot-id");

to = prefix_roota(dest, "/proc/sys/kernel/random/boot_id");

r = sd_id128_randomize(&rnd);

if (r < 0)

return log_error_errno(r, "Failed to generate random boot id: %m");

id128_format_as_uuid(rnd, as_uuid);

r = write_string_file(from, as_uuid, WRITE_STRING_FILE_CREATE);

if (r < 0)

return log_error_errno(r, "Failed to write boot id: %m");

if (mount(from, to, NULL, MS_BIND, NULL) < 0)

r = log_error_errno(errno, "Failed to bind mount boot id: %m");

else if (mount(NULL, to, NULL, MS_BIND|MS_REMOUNT|MS_RDONLY|MS_NOSUID|MS_NODEV, NULL) < 0)

log_warning_errno(errno, "Failed to make boot id read-only: %m");

unlink(from);

return r;

}

static int copy_devnodes(const char *dest) {

static const char devnodes[] =

"null\0"

"zero\0"

"full\0"

"random\0"

"urandom\0"

"tty\0"

"net/tun\0";

const char *d;

int r = 0;

_cleanup_umask_ mode_t u;

assert(dest);

u = umask(0000);

/* Create /dev/net, so that we can create /dev/net/tun in it */

if (userns_mkdir(dest, "/dev/net", 0755, 0, 0) < 0)

return log_error_errno(r, "Failed to create /dev/net directory: %m");

NULSTR_FOREACH(d, devnodes) {

_cleanup_free_ char *from = NULL, *to = NULL;

struct stat st;

from = strappend("/dev/", d);

to = prefix_root(dest, from);

if (stat(from, &st) < 0) {

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)) {

log_error("%s is not a char or block device, cannot copy.", from);

return -EIO;

} else {

if (mknod(to, st.st_mode, st.st_rdev) < 0) {

if (errno != EPERM)

return log_error_errno(errno, "mknod(%s) failed: %m", to);

/* Some systems abusively restrict mknod but

r = touch(to);

if (r < 0)

return log_error_errno(r, "touch (%s) failed: %m", to);

if (mount(from, to, NULL, MS_BIND, NULL) < 0)

return log_error_errno(errno, "Both mknod and bind mount (%s) failed: %m", to);

}

r = userns_lchown(to, 0, 0);

if (r < 0)

return log_error_errno(r, "chown() of device node %s failed: %m", to);

}

}

return r;

}

static int setup_pts(const char *dest) {

_cleanup_free_ char *options = NULL;

const char *p;

#ifdef HAVE_SELINUX

if (arg_selinux_apifs_context)

(void) asprintf(&options,

"newinstance,ptmxmode=0666,mode=620,gid=" GID_FMT ",context=\"%s\"",

arg_uid_shift + TTY_GID,

arg_selinux_apifs_context);

else

#endif

(void) asprintf(&options,

"newinstance,ptmxmode=0666,mode=620,gid=" GID_FMT,

arg_uid_shift + TTY_GID);

if (!options)

return log_oom();

/* Mount /dev/pts itself */

p = prefix_roota(dest, "/dev/pts");

if (mkdir(p, 0755) < 0)

return log_error_errno(errno, "Failed to create /dev/pts: %m");

if (mount("devpts", p, "devpts", MS_NOSUID|MS_NOEXEC, options) < 0)

return log_error_errno(errno, "Failed to mount /dev/pts: %m");

if (userns_lchown(p, 0, 0) < 0)

return log_error_errno(errno, "Failed to chown /dev/pts: %m");

/* Create /dev/ptmx symlink */

p = prefix_roota(dest, "/dev/ptmx");

if (symlink("pts/ptmx", p) < 0)

return log_error_errno(errno, "Failed to create /dev/ptmx symlink: %m");

if (userns_lchown(p, 0, 0) < 0)

return log_error_errno(errno, "Failed to chown /dev/ptmx: %m");

/* And fix /dev/pts/ptmx ownership */

p = prefix_roota(dest, "/dev/pts/ptmx");

if (userns_lchown(p, 0, 0) < 0)

return log_error_errno(errno, "Failed to chown /dev/pts/ptmx: %m");

return 0;

}

static int setup_dev_console(const char *dest, const char *console) {

_cleanup_umask_ mode_t u;

const char *to;

int r;

assert(dest);

assert(console);

u = umask(0000);

r = chmod_and_chown(console, 0600, arg_uid_shift, arg_uid_shift);

if (r < 0)

return log_error_errno(r, "Failed to correct access mode for TTY: %m");

/* We need to bind mount the right tty to /dev/console since

to = prefix_roota(dest, "/dev/console");

r = touch(to);

if (r < 0)

return log_error_errno(r, "touch() for /dev/console failed: %m");

if (mount(console, to, NULL, MS_BIND, NULL) < 0)

return log_error_errno(errno, "Bind mount for /dev/console failed: %m");

return 0;

}

static int setup_kmsg(const char *dest, int kmsg_socket) {

const char *from, *to;

_cleanup_umask_ mode_t u;

int fd, k;

union {

struct cmsghdr cmsghdr;

uint8_t buf[CMSG_SPACE(sizeof(int))];

} control = {};

struct msghdr mh = {

.msg_control = &control,

.msg_controllen = sizeof(control),

};

struct cmsghdr *cmsg;

assert(kmsg_socket >= 0);

u = umask(0000);

/* We create the kmsg FIFO as /run/kmsg, but immediately

from = prefix_roota(dest, "/run/kmsg");

to = prefix_roota(dest, "/proc/kmsg");

if (mkfifo(from, 0600) < 0)

return log_error_errno(errno, "mkfifo() for /run/kmsg failed: %m");

if (mount(from, to, NULL, 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)

return log_error_errno(errno, "Failed to open fifo: %m");

cmsg = CMSG_FIRSTHDR(&mh);

cmsg->cmsg_level = SOL_SOCKET;

cmsg->cmsg_type = SCM_RIGHTS;

cmsg->cmsg_len = CMSG_LEN(sizeof(int));

memcpy(CMSG_DATA(cmsg), &fd, sizeof(int));

mh.msg_controllen = cmsg->cmsg_len;

/* Store away the fd in the socket, so that it stays open as

k = sendmsg(kmsg_socket, &mh, MSG_NOSIGNAL);

safe_close(fd);

if (k < 0)

return log_error_errno(errno, "Failed to send FIFO fd: %m");

/* And now make the FIFO unavailable as /run/kmsg... */

(void) unlink(from);

return 0;

}

static int send_rtnl(int send_fd) {

union {

struct cmsghdr cmsghdr;

uint8_t buf[CMSG_SPACE(sizeof(int))];

} control = {};

struct msghdr mh = {

.msg_control = &control,

.msg_controllen = sizeof(control),

};

struct cmsghdr *cmsg;

_cleanup_close_ int fd = -1;

ssize_t k;

assert(send_fd >= 0);

if (!arg_expose_ports)

return 0;

fd = socket(PF_NETLINK, SOCK_RAW|SOCK_CLOEXEC|SOCK_NONBLOCK, NETLINK_ROUTE);

if (fd < 0)

return log_error_errno(errno, "Failed to allocate container netlink: %m");

cmsg = CMSG_FIRSTHDR(&mh);

cmsg->cmsg_level = SOL_SOCKET;

cmsg->cmsg_type = SCM_RIGHTS;

cmsg->cmsg_len = CMSG_LEN(sizeof(int));

memcpy(CMSG_DATA(cmsg), &fd, sizeof(int));

mh.msg_controllen = cmsg->cmsg_len;

/* Store away the fd in the socket, so that it stays open as

k = sendmsg(send_fd, &mh, MSG_NOSIGNAL);

if (k < 0)

return log_error_errno(errno, "Failed to send netlink fd: %m");

return 0;

}

static int flush_ports(union in_addr_union *exposed) {

ExposePort *p;

int r, af = AF_INET;

assert(exposed);

if (!arg_expose_ports)

return 0;

if (in_addr_is_null(af, exposed))

return 0;

log_debug("Lost IP address.");

LIST_FOREACH(ports, p, arg_expose_ports) {

r = fw_add_local_dnat(false,

af,

p->protocol,

NULL,

NULL, 0,

NULL, 0,

p->host_port,

exposed,

p->container_port,

NULL);

if (r < 0)

log_warning_errno(r, "Failed to modify firewall: %m");

}

*exposed = IN_ADDR_NULL;

return 0;

}

static int expose_ports(sd_netlink *rtnl, union in_addr_union *exposed) {

_cleanup_free_ struct local_address *addresses = NULL;

_cleanup_free_ char *pretty = NULL;

union in_addr_union new_exposed;

ExposePort *p;

bool add;

int af = AF_INET, r;

assert(exposed);

/* Invoked each time an address is added or removed inside the

if (!arg_expose_ports)

return 0;

r = local_addresses(rtnl, 0, af, &addresses);

if (r < 0)

return log_error_errno(r, "Failed to enumerate local addresses: %m");

add = r > 0 &&

addresses[0].family == af &&

addresses[0].scope < RT_SCOPE_LINK;

if (!add)

return flush_ports(exposed);

new_exposed = addresses[0].address;

if (in_addr_equal(af, exposed, &new_exposed))

return 0;

in_addr_to_string(af, &new_exposed, &pretty);

log_debug("New container IP is %s.", strna(pretty));

LIST_FOREACH(ports, p, arg_expose_ports) {

r = fw_add_local_dnat(true,

af,

p->protocol,

NULL,

NULL, 0,

NULL, 0,

p->host_port,

&new_exposed,

p->container_port,

in_addr_is_null(af, exposed) ? NULL : exposed);

if (r < 0)

log_warning_errno(r, "Failed to modify firewall: %m");

}

*exposed = new_exposed;

return 0;

}

static int on_address_change(sd_netlink *rtnl, sd_netlink_message *m, void *userdata) {

union in_addr_union *exposed = userdata;

assert(rtnl);

assert(m);

assert(exposed);

expose_ports(rtnl, exposed);

return 0;

}

static int watch_rtnl(sd_event *event, int recv_fd, union in_addr_union *exposed, sd_netlink **ret) {

union {

struct cmsghdr cmsghdr;

uint8_t buf[CMSG_SPACE(sizeof(int))];

} control = {};

struct msghdr mh = {

.msg_control = &control,

.msg_controllen = sizeof(control),

};

struct cmsghdr *cmsg;

_cleanup_netlink_unref_ sd_netlink *rtnl = NULL;

int fd, r;

ssize_t k;

assert(event);

assert(recv_fd >= 0);

assert(ret);

if (!arg_expose_ports)

return 0;

k = recvmsg(recv_fd, &mh, MSG_NOSIGNAL);

if (k < 0)

return log_error_errno(errno, "Failed to recv netlink fd: %m");

cmsg = CMSG_FIRSTHDR(&mh);

assert(cmsg->cmsg_level == SOL_SOCKET);

assert(cmsg->cmsg_type == SCM_RIGHTS);

assert(cmsg->cmsg_len == CMSG_LEN(sizeof(int)));

memcpy(&fd, CMSG_DATA(cmsg), sizeof(int));

r = sd_netlink_open_fd(&rtnl, fd);

if (r < 0) {

safe_close(fd);

return log_error_errno(r, "Failed to create rtnl object: %m");

}

r = sd_netlink_add_match(rtnl, RTM_NEWADDR, on_address_change, exposed);

if (r < 0)

return log_error_errno(r, "Failed to subscribe to RTM_NEWADDR messages: %m");

r = sd_netlink_add_match(rtnl, RTM_DELADDR, on_address_change, exposed);

if (r < 0)

return log_error_errno(r, "Failed to subscribe to RTM_DELADDR messages: %m");

r = sd_netlink_attach_event(rtnl, event, 0);

if (r < 0)

return log_error_errno(r, "Failed to add to even loop: %m");

*ret = rtnl;

rtnl = NULL;

return 0;

}

static int setup_hostname(void) {

if (arg_share_system)

return 0;

if (sethostname_idempotent(arg_machine) < 0)

return -errno;

return 0;

}

static int setup_journal(const char *directory) {

sd_id128_t machine_id, this_id;

_cleanup_free_ char *b = NULL, *d = NULL;

const char *etc_machine_id, *p, *q;

char *id;

int r;

/* Don't link journals in ephemeral mode */

if (arg_ephemeral)

return 0;

etc_machine_id = prefix_roota(directory, "/etc/machine-id");

r = read_one_line_file(etc_machine_id, &b);

if (r == -ENOENT && arg_link_journal == LINK_AUTO)

return 0;

else if (r < 0)

return log_error_errno(r, "Failed to read machine ID from %s: %m", etc_machine_id);

id = strstrip(b);

if (isempty(id) && arg_link_journal == LINK_AUTO)

return 0;

/* Verify validity */

r = sd_id128_from_string(id, &machine_id);

if (r < 0)

return log_error_errno(r, "Failed to parse machine ID from %s: %m", etc_machine_id);

r = sd_id128_get_machine(&this_id);

if (r < 0)

return log_error_errno(r, "Failed to retrieve machine ID: %m");

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;

r = userns_mkdir(directory, "/var", 0755, 0, 0);

if (r < 0)

return log_error_errno(r, "Failed to create /var: %m");

r = userns_mkdir(directory, "/var/log", 0755, 0, 0);

if (r < 0)

return log_error_errno(r, "Failed to create /var/log: %m");

r = userns_mkdir(directory, "/var/log/journal", 0755, 0, 0);

if (r < 0)

return log_error_errno(r, "Failed to create /var/log/journal: %m");

p = strjoina("/var/log/journal/", id);

q = prefix_roota(directory, p);

if (path_is_mount_point(p, 0) > 0) {

if (arg_link_journal != LINK_AUTO) {

log_error("%s: already a mount point, refusing to use for journal", p);

return -EEXIST;

}

return 0;

}

if (path_is_mount_point(q, 0) > 0) {

if (arg_link_journal != LINK_AUTO) {

log_error("%s: already a mount point, refusing to use for journal", q);

return -EEXIST;

}

return 0;

}

r = readlink_and_make_absolute(p, &d);

if (r >= 0) {

if ((arg_link_journal == LINK_GUEST ||

arg_link_journal == LINK_AUTO) &&

path_equal(d, q)) {

r = userns_mkdir(directory, p, 0755, 0, 0);

if (r < 0)

log_warning_errno(errno, "Failed to create directory %s: %m", q);

return 0;

}

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 &&

rmdir(p) < 0) {

if (errno == ENOTDIR) {

log_error("%s already exists and is neither a symlink nor a directory", p);

return r;

} else {

log_error_errno(errno, "Failed to remove %s: %m", p);

return -errno;

}

}

} else if (r != -ENOENT) {

log_error_errno(errno, "readlink(%s) failed: %m", p);

return r;

}

if (arg_link_journal == LINK_GUEST) {

if (symlink(q, p) < 0) {

if (arg_link_journal_try) {

log_debug_errno(errno, "Failed to symlink %s to %s, skipping journal setup: %m", q, p);

return 0;

} else {

log_error_errno(errno, "Failed to symlink %s to %s: %m", q, p);

return -errno;

}

}

r = userns_mkdir(directory, p, 0755, 0, 0);

if (r < 0)

log_warning_errno(errno, "Failed to create directory %s: %m", q);

return 0;

}

if (arg_link_journal == LINK_HOST) {

/* don't create parents here -- if the host doesn't have

r = mkdir(p, 0755);

if (r < 0) {

if (arg_link_journal_try) {

log_debug_errno(errno, "Failed to create %s, skipping journal setup: %m", p);

return 0;

} else {

log_error_errno(errno, "Failed to create %s: %m", p);

return r;

}

}

} else if (access(p, F_OK) < 0)

return 0;

if (dir_is_empty(q) == 0)

log_warning("%s is not empty, proceeding anyway.", q);

r = userns_mkdir(directory, p, 0755, 0, 0);

if (r < 0) {

log_error_errno(errno, "Failed to create %s: %m", q);

return r;

}

if (mount(p, q, NULL, MS_BIND, NULL) < 0)

return log_error_errno(errno, "Failed to bind mount journal from host into guest: %m");

return 0;

}

static int drop_capabilities(void) {

return capability_bounding_set_drop(~arg_retain, false);

}

static int register_machine(pid_t pid, int local_ifindex) {

_cleanup_bus_error_free_ sd_bus_error error = SD_BUS_ERROR_NULL;

_cleanup_bus_flush_close_unref_ sd_bus *bus = NULL;

int r;

if (!arg_register)

return 0;

r = sd_bus_default_system(&bus);

if (r < 0)

return log_error_errno(r, "Failed to open system bus: %m");

if (arg_keep_unit) {

r = sd_bus_call_method(

bus,

"org.freedesktop.machine1",

"/org/freedesktop/machine1",

"org.freedesktop.machine1.Manager",

"RegisterMachineWithNetwork",

&error,

NULL,

"sayssusai",

arg_machine,

SD_BUS_MESSAGE_APPEND_ID128(arg_uuid),

"nspawn",

"container",

(uint32_t) pid,

strempty(arg_directory),

local_ifindex > 0 ? 1 : 0, local_ifindex);

} else {

_cleanup_bus_message_unref_ sd_bus_message *m = NULL;

char **i;

unsigned j;

r = sd_bus_message_new_method_call(

bus,

&m,

"org.freedesktop.machine1",

"/org/freedesktop/machine1",

"org.freedesktop.machine1.Manager",

"CreateMachineWithNetwork");

if (r < 0)

return bus_log_create_error(r);

r = sd_bus_message_append(

m,

"sayssusai",

arg_machine,

SD_BUS_MESSAGE_APPEND_ID128(arg_uuid),

"nspawn",

"container",

(uint32_t) pid,

strempty(arg_directory),

local_ifindex > 0 ? 1 : 0, local_ifindex);

if (r < 0)

return bus_log_create_error(r);

r = sd_bus_message_open_container(m, 'a', "(sv)");

if (r < 0)

return bus_log_create_error(r);

if (!isempty(arg_slice)) {

r = sd_bus_message_append(m, "(sv)", "Slice", "s", arg_slice);

if (r < 0)

return bus_log_create_error(r);

}

r = sd_bus_message_append(m, "(sv)", "DevicePolicy", "s", "strict");

if (r < 0)

return bus_log_create_error(r);

/* If you make changes here, also make sure to update

r = sd_bus_message_append(m, "(sv)", "DeviceAllow", "a(ss)", 9,

/* Allow the container to

"/dev/null", "rwm",

"/dev/zero", "rwm",

"/dev/full", "rwm",

"/dev/random", "rwm",

"/dev/urandom", "rwm",

"/dev/tty", "rwm",

"/dev/net/tun", "rwm",

/* Allow the container

"/dev/pts/ptmx", "rw",

"char-pts", "rw");

if (r < 0)

return bus_log_create_error(r);

for (j = 0; j < arg_n_custom_mounts; j++) {

CustomMount *cm = &arg_custom_mounts[j];

if (cm->type != CUSTOM_MOUNT_BIND)

continue;

r = is_device_node(cm->source);

if (r < 0)

return log_error_errno(r, "Failed to stat %s: %m", cm->source);

if (r) {

r = sd_bus_message_append(m, "(sv)", "DeviceAllow", "a(ss)", 1,

cm->source, cm->read_only ? "r" : "rw");

if (r < 0)

return log_error_errno(r, "Failed to append message arguments: %m");

}

}

if (arg_kill_signal != 0) {

r = sd_bus_message_append(m, "(sv)", "KillSignal", "i", arg_kill_signal);

if (r < 0)

return bus_log_create_error(r);

r = sd_bus_message_append(m, "(sv)", "KillMode", "s", "mixed");

if (r < 0)

return bus_log_create_error(r);

}

STRV_FOREACH(i, arg_property) {

r = sd_bus_message_open_container(m, 'r', "sv");

if (r < 0)

return bus_log_create_error(r);

r = bus_append_unit_property_assignment(m, *i);

if (r < 0)

return r;

r = sd_bus_message_close_container(m);

if (r < 0)

return bus_log_create_error(r);

}

r = sd_bus_message_close_container(m);

if (r < 0)

return bus_log_create_error(r);

r = sd_bus_call(bus, m, 0, &error, NULL);

}

if (r < 0) {

log_error("Failed to register machine: %s", bus_error_message(&error, r));

return r;

}

return 0;

}

static int terminate_machine(pid_t pid) {

_cleanup_bus_error_free_ sd_bus_error error = SD_BUS_ERROR_NULL;

_cleanup_bus_message_unref_ sd_bus_message *reply = NULL;

_cleanup_bus_flush_close_unref_ sd_bus *bus = NULL;

const char *path;

int r;

if (!arg_register)

return 0;

/* If we are reusing the unit, then just exit, systemd will do

if (arg_keep_unit)

return 0;

r = sd_bus_default_system(&bus);

if (r < 0)

return log_error_errno(r, "Failed to open system bus: %m");

r = sd_bus_call_method(

bus,

"org.freedesktop.machine1",

"/org/freedesktop/machine1",

"org.freedesktop.machine1.Manager",

"GetMachineByPID",

&error,

&reply,

"u",

(uint32_t) pid);

if (r < 0) {

/* Note that the machine might already have been

log_debug("Failed to get machine: %s", bus_error_message(&error, r));

return 0;

}

r = sd_bus_message_read(reply, "o", &path);

if (r < 0)

return bus_log_parse_error(r);

r = sd_bus_call_method(

bus,

"org.freedesktop.machine1",

path,

"org.freedesktop.machine1.Machine",

"Terminate",

&error,

NULL,

NULL);

if (r < 0) {

log_debug("Failed to terminate machine: %s", bus_error_message(&error, r));

return 0;

}

return 0;

}

static int reset_audit_loginuid(void) {

_cleanup_free_ char *p = NULL;

int r;

if (arg_share_system)

return 0;

r = read_one_line_file("/proc/self/loginuid", &p);

if (r == -ENOENT)

return 0;

if (r < 0)

return log_error_errno(r, "Failed to read /proc/self/loginuid: %m");

/* Already reset? */

if (streq(p, "4294967295"))

return 0;

r = write_string_file("/proc/self/loginuid", "4294967295", 0);

if (r < 0) {

log_error_errno(r,

"Failed to reset audit login UID. This probably means that your kernel is too\n"

"old and you have audit enabled. Note that the auditing subsystem is known to\n"

"be incompatible with containers on old kernels. Please make sure to upgrade\n"

"your kernel or to off auditing with 'audit=0' on the kernel command line before\n"

"using systemd-nspawn. Sleeping for 5s... (%m)");

sleep(5);

}

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, uint64_t idx) {

uint8_t result[8];

size_t l, sz;

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 */

r = sd_id128_get_machine((sd_id128_t*) v);

if (r < 0)

return r;

/* combine with some data unique (on this host) to this

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

siphash24(result, v, sz, hash_key.bytes);