#ifdef HAVE_BLKID

#include <blkid/blkid.h>

#endif

#include <errno.h>

#include <getopt.h>

#include <linux/loop.h>

#include <sched.h>

#ifdef HAVE_SECCOMP

#include <seccomp.h>

#endif

#ifdef HAVE_SELINUX

#include <selinux/selinux.h>

#endif

#include <signal.h>

#include <stdio.h>

#include <stdlib.h>

#include <string.h>

#include <sys/file.h>

#include <sys/mount.h>

#include <sys/personality.h>

#include <sys/prctl.h>

#include <sys/types.h>

#include <unistd.h>

#include "sd-daemon.h"

#include "sd-id128.h"

#include "barrier.h"

#include "base-filesystem.h"

#include "blkid-util.h"

#include "btrfs-util.h"

#include "build.h"

#include "cap-list.h"

#include "capability.h"

#include "cgroup-util.h"

#include "copy.h"

#include "dev-setup.h"

#include "env-util.h"

#include "event-util.h"

#include "fdset.h"

#include "fileio.h"

#include "formats-util.h"

#include "gpt.h"

#include "hostname-util.h"

#include "log.h"

#include "loopback-setup.h"

#include "machine-image.h"

#include "macro.h"

#include "missing.h"

#include "mkdir.h"

#include "netlink-util.h"

#include "path-util.h"

#include "process-util.h"

#include "ptyfwd.h"

#include "random-util.h"

#include "rm-rf.h"

#ifdef HAVE_SECCOMP

#include "seccomp-util.h"

#endif

#include "signal-util.h"

#include "strv.h"

#include "terminal-util.h"

#include "udev-util.h"

#include "util.h"

#include "nspawn-settings.h"

#include "nspawn-mount.h"

#include "nspawn-network.h"

#include "nspawn-expose-ports.h"

#include "nspawn-cgroup.h"

#include "nspawn-register.h"

#include "nspawn-setuid.h"

typedef enum ContainerStatus {

CONTAINER_TERMINATED,

CONTAINER_REBOOTED

} ContainerStatus;

typedef enum LinkJournal {

LINK_NO,

LINK_AUTO,

LINK_HOST,

LINK_GUEST

} LinkJournal;

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 char *arg_network_bridge = NULL;

static unsigned long arg_personality = PERSONALITY_INVALID;

static char *arg_image = NULL;

static VolatileMode arg_volatile_mode = 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 SettingsMask arg_settings_mask = 0;

static int arg_settings_trusted = -1;

static char **arg_parameters = NULL;

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"

" --settings=BOOLEAN Load additional settings from .nspawn file\n"

, program_invocation_short_name);

}

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,

ARG_SETTINGS,

};

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 },

{ "settings", required_argument, NULL, ARG_SETTINGS },

{}

};

int c, r;

uint64_t plus = 0, minus = 0;

bool mask_all_settings = false, mask_no_settings = false;

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();

arg_settings_mask |= SETTING_USER;

break;

case ARG_NETWORK_BRIDGE:

r = free_and_strdup(&arg_network_bridge, optarg);

if (r < 0)

return log_oom();

/* fall through */

case 'n':

arg_network_veth = true;

arg_private_network = true;

arg_settings_mask |= SETTING_NETWORK;

break;

case ARG_NETWORK_INTERFACE:

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

return log_oom();

arg_private_network = true;

arg_settings_mask |= SETTING_NETWORK;

break;

case ARG_NETWORK_MACVLAN:

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

return log_oom();

arg_private_network = true;

arg_settings_mask |= SETTING_NETWORK;

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;

arg_settings_mask |= SETTING_NETWORK;

break;

case 'b':

arg_boot = true;

arg_settings_mask |= SETTING_BOOT;

break;

case ARG_UUID:

r = sd_id128_from_string(optarg, &arg_uuid);

if (r < 0) {

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

return r;

}

arg_settings_mask |= SETTING_MACHINE_ID;

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;

arg_settings_mask |= SETTING_READ_ONLY;

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;

}

}

arg_settings_mask |= SETTING_CAPABILITY;

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:

r = bind_mount_parse(&arg_custom_mounts, &arg_n_custom_mounts, optarg, c == ARG_BIND_RO);

if (r < 0)

return log_error_errno(r, "Failed to parse --bind(-ro)= argument %s: %m", optarg);

arg_settings_mask |= SETTING_CUSTOM_MOUNTS;

break;

case ARG_TMPFS:

r = tmpfs_mount_parse(&arg_custom_mounts, &arg_n_custom_mounts, optarg);

if (r < 0)

return log_error_errno(r, "Failed to parse --tmpfs= argument %s: %m", optarg);

arg_settings_mask |= SETTING_CUSTOM_MOUNTS;

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(&arg_custom_mounts, &arg_n_custom_mounts, 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;

arg_settings_mask |= SETTING_CUSTOM_MOUNTS;

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;

arg_settings_mask |= SETTING_ENVIRONMENT;

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;

}

arg_settings_mask |= SETTING_PERSONALITY;

break;

case ARG_VOLATILE:

if (!optarg)

arg_volatile_mode = VOLATILE_YES;

else {

VolatileMode m;

m = volatile_mode_from_string(optarg);

if (m < 0) {

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

return -EINVAL;

} else

arg_volatile_mode = m;

}

arg_settings_mask |= SETTING_VOLATILE_MODE;

break;

case 'p':

r = expose_port_parse(&arg_expose_ports, optarg);

if (r == -EEXIST)

return log_error_errno(r, "Duplicate port specification: %s", optarg);

if (r < 0)

return log_error_errno(r, "Failed to parse host port %s: %m", optarg);

arg_settings_mask |= SETTING_EXPOSE_PORTS;

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;

}

arg_settings_mask |= SETTING_KILL_SIGNAL;

break;

case ARG_SETTINGS:

/* no → do not read files

r = parse_boolean(optarg);

if (r < 0) {

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

mask_all_settings = false;

mask_no_settings = false;

arg_settings_trusted = true;

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

mask_all_settings = false;

mask_no_settings = true;

arg_settings_trusted = -1;

} else

return log_error_errno(r, "Failed to parse --settings= argument: %s", optarg);

} else if (r > 0) {

/* yes */

mask_all_settings = false;

mask_no_settings = false;

arg_settings_trusted = -1;

} else {

/* no */

mask_all_settings = true;

mask_no_settings = false;

arg_settings_trusted = false;

}

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_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.");

if (argc > optind) {

arg_parameters = strv_copy(argv + optind);

if (!arg_parameters)

return log_oom();

arg_settings_mask |= SETTING_BOOT;

}

/* Load all settings from .nspawn files */

if (mask_no_settings)

arg_settings_mask = 0;

/* Don't load any settings from .nspawn files */

if (mask_all_settings)

arg_settings_mask = _SETTINGS_MASK_ALL;

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

r = detect_unified_cgroup_hierarchy();

if (r < 0)

return r;

return 1;

}

static int verify_arguments(void) {

if (arg_volatile_mode != 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_boot && arg_kill_signal <= 0)

arg_kill_signal = SIGRTMIN+3;

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 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 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_port_execute(rtnl, arg_expose_ports, exposed);

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 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;

}

static int setup_seccomp(void) {

#ifdef HAVE_SECCOMP

static const struct {

uint64_t capability;

int syscall_num;

} blacklist[] = {

{ CAP_SYS_RAWIO, SCMP_SYS(iopl) },

{ CAP_SYS_RAWIO, SCMP_SYS(ioperm) },

{ CAP_SYS_BOOT, SCMP_SYS(kexec_load) },

{ CAP_SYS_ADMIN, SCMP_SYS(swapon) },

{ CAP_SYS_ADMIN, SCMP_SYS(swapoff) },

{ CAP_SYS_ADMIN, SCMP_SYS(open_by_handle_at) },

{ CAP_SYS_MODULE, SCMP_SYS(init_module) },

{ CAP_SYS_MODULE, SCMP_SYS(finit_module) },

{ CAP_SYS_MODULE, SCMP_SYS(delete_module) },

{ CAP_SYSLOG, SCMP_SYS(syslog) },

};

scmp_filter_ctx seccomp;

unsigned i;

int r;

seccomp = seccomp_init(SCMP_ACT_ALLOW);

if (!seccomp)

return log_oom();

r = seccomp_add_secondary_archs(seccomp);

if (r < 0) {

log_error_errno(r, "Failed to add secondary archs to seccomp filter: %m");

goto finish;

}

for (i = 0; i < ELEMENTSOF(blacklist); i++) {

if (arg_retain & (1ULL << blacklist[i].capability))

continue;

r = seccomp_rule_add(seccomp, SCMP_ACT_ERRNO(EPERM), blacklist[i].syscall_num, 0);

if (r == -EFAULT)

continue; /* unknown syscall */

if (r < 0) {

log_error_errno(r, "Failed to block syscall: %m");

goto finish;

}

}

/*

r = seccomp_rule_add(

seccomp,

SCMP_ACT_ERRNO(EAFNOSUPPORT),

SCMP_SYS(socket),

2,

SCMP_A0(SCMP_CMP_EQ, AF_NETLINK),

SCMP_A2(SCMP_CMP_EQ, NETLINK_AUDIT));

if (r < 0) {

log_error_errno(r, "Failed to add audit seccomp rule: %m");

goto finish;

}

r = seccomp_attr_set(seccomp, SCMP_FLTATR_CTL_NNP, 0);

if (r < 0) {

log_error_errno(r, "Failed to unset NO_NEW_PRIVS: %m");

goto finish;

}

r = seccomp_load(seccomp);

if (r == -EINVAL) {

log_debug_errno(r, "Kernel is probably not configured with CONFIG_SECCOMP. Disabling seccomp audit filter: %m");

r = 0;

goto finish;

}

if (r < 0) {

log_error_errno(r, "Failed to install seccomp audit filter: %m");

goto finish;

}

finish:

seccomp_release(seccomp);

return r;

#else

return 0;

#endif

}

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 = strjoina("/run/systemd/nspawn/propagate/", arg_machine);

(void) mkdir_p(p, 0600);

if (userns_mkdir(root, "/run/systemd", 0755, 0, 0) < 0)

return log_error_errno(errno, "Failed to create /run/systemd: %m");

if (userns_mkdir(root, "/run/systemd/nspawn", 0755, 0, 0) < 0)

return log_error_errno(errno, "Failed to create /run/systemd/nspawn: %m");

if (userns_mkdir(root, "/run/systemd/nspawn/incoming", 0600, 0, 0) < 0)

return log_error_errno(errno, "Failed to create /run/systemd/nspawn/incoming: %m");

q = prefix_roota(root, "/run/systemd/nspawn/incoming");

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

};

_cleanup_close_ int fd = -1, control = -1, loop = -1;

_cleanup_free_ char* loopdev = NULL;

struct stat st;

int r, nr;

assert(device_path);

assert(loop_nr);

assert(arg_image);

fd = open(arg_image, O_CLOEXEC|(arg_read_only ? O_RDONLY : O_RDWR)|O_NONBLOCK|O_NOCTTY);

if (fd < 0)

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

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;

p = strdup(arg_image);

if (!p)

return log_oom();

*device_path = p;

*loop_nr = -1;

r = fd;

fd = -1;

return r;

}

if (!S_ISREG(st.st_mode)) {

log_error_errno(errno, "%s is not a regular file or block device: %m", arg_image);

return -EINVAL;

}

control = open("/dev/loop-control", O_RDWR|O_CLOEXEC|O_NOCTTY|O_NONBLOCK);

if (control < 0)

return log_error_errno(errno, "Failed to open /dev/loop-control: %m");

nr = ioctl(control, LOOP_CTL_GET_FREE);

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)

return log_error_errno(errno, "Failed to open loop device %s: %m", loopdev);

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)

return log_error_errno(errno, "Failed to set loopback settings on %s: %m", loopdev);

*device_path = loopdev;

loopdev = NULL;

*loop_nr = nr;

r = loop;

loop = -1;

return r;

}

#define PARTITION_TABLE_BLURB \

"Note that the disk image needs to either contain only a single MBR partition of\n" \

"type 0x83 that is marked bootable, or a single GPT partition of type " \

"0FC63DAF-8483-4772-8E79-3D69D8477DE4 or follow\n" \

" http://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec/\n" \

"to be bootable with systemd-nspawn."

static int dissect_image(

int fd,

char **root_device, bool *root_device_rw,

char **home_device, bool *home_device_rw,

char **srv_device, bool *srv_device_rw,

bool *secondary) {

#ifdef HAVE_BLKID

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, *generic = NULL;

_cleanup_udev_enumerate_unref_ struct udev_enumerate *e = NULL;

_cleanup_udev_device_unref_ struct udev_device *d = NULL;

_cleanup_blkid_free_probe_ blkid_probe b = NULL;

_cleanup_udev_unref_ struct udev *udev = NULL;

struct udev_list_entry *first, *item;

bool home_rw = true, root_rw = true, secondary_root_rw = true, srv_rw = true, generic_rw = true;

bool is_gpt, is_mbr, multiple_generic = false;

const char *pttype = NULL;

blkid_partlist pl;

struct stat st;

unsigned i;

int r;

assert(fd >= 0);

assert(root_device);

assert(home_device);

assert(srv_device);

assert(secondary);

assert(arg_image);

b = blkid_new_probe();

if (!b)

return log_oom();

errno = 0;

r = blkid_probe_set_device(b, fd, 0, 0);

if (r != 0) {

if (errno == 0)

return log_oom();

log_error_errno(errno, "Failed to set device on blkid probe: %m");

return -errno;

}

blkid_probe_enable_partitions(b, 1);

blkid_probe_set_partitions_flags(b, BLKID_PARTS_ENTRY_DETAILS);

errno = 0;

r = blkid_do_safeprobe(b);

if (r == -2 || r == 1) {

log_error("Failed to identify any partition table on\n"

" %s\n"

PARTITION_TABLE_BLURB, arg_image);

return -EINVAL;

} else if (r != 0) {

if (errno == 0)

errno = EIO;

log_error_errno(errno, "Failed to probe: %m");

return -errno;

}

(void) blkid_probe_lookup_value(b, "PTTYPE", &pttype, NULL);

is_gpt = streq_ptr(pttype, "gpt");

is_mbr = streq_ptr(pttype, "dos");

if (!is_gpt && !is_mbr) {

log_error("No GPT or MBR partition table discovered on\n"

" %s\n"

PARTITION_TABLE_BLURB, arg_image);

return -EINVAL;

}

errno = 0;

pl = blkid_probe_get_partitions(b);

if (!pl) {

if (errno == 0)

return log_oom();

log_error("Failed to list partitions of %s", arg_image);

return -errno;

}

udev = udev_new();

if (!udev)

return log_oom();

if (fstat(fd, &st) < 0)

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 log_oom();

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

int n, m;

if (i >= 10) {

log_error("Kernel partitions never appeared.");

return -ENXIO;

}

e = udev_enumerate_new(udev);

if (!e)

return log_oom();

r = udev_enumerate_add_match_parent(e, d);

if (r < 0)

return log_oom();

r = udev_enumerate_scan_devices(e);

if (r < 0)

return log_error_errno(r, "Failed to scan for partition devices of %s: %m", arg_image);

/* Count the partitions enumerated by the kernel */

n = 0;

first = udev_enumerate_get_list_entry(e);

udev_list_entry_foreach(item, first)

n++;

/* Count the partitions enumerated by blkid */

m = blkid_partlist_numof_partitions(pl);

if (n == m + 1)

break;

if (n > m + 1) {

log_error("blkid and kernel partition list do not match.");

return -EIO;

}

if (n < m + 1) {

unsigned j;

/* The kernel has probed fewer partitions than

for (j = 0; j < 20; j++) {

r = ioctl(fd, BLKRRPART, 0);

if (r < 0)

r = -errno;

if (r >= 0 || r != -EBUSY)

break;

/* If something else has the device

usleep(50 * USEC_PER_MSEC);

}

if (r < 0)

return log_error_errno(r, "Failed to reread partition table: %m");

}

e = udev_enumerate_unref(e);

}

first = udev_enumerate_get_list_entry(e);

udev_list_entry_foreach(item, first) {

_cleanup_udev_device_unref_ struct udev_device *q;

const char *node;

unsigned long long flags;

blkid_partition pp;

dev_t qn;

int nr;

errno = 0;

q = udev_device_new_from_syspath(udev, udev_list_entry_get_name(item));

if (!q) {

if (!errno)

errno = ENOMEM;

log_error_errno(errno, "Failed to get partition device of %s: %m", arg_image);

return -errno;

}

qn = udev_device_get_devnum(q);

if (major(qn) == 0)

continue;

if (st.st_rdev == qn)

continue;

node = udev_device_get_devnode(q);

if (!node)

continue;

pp = blkid_partlist_devno_to_partition(pl, qn);

if (!pp)

continue;

flags = blkid_partition_get_flags(pp);

nr = blkid_partition_get_partno(pp);

if (nr < 0)

continue;

if (is_gpt) {

sd_id128_t type_id;

const char *stype;

if (flags & GPT_FLAG_NO_AUTO)

continue;

stype = blkid_partition_get_type_string(pp);

if (!stype)

continue;

if (sd_id128_from_string(stype, &type_id) < 0)

continue;

if (sd_id128_equal(type_id, GPT_HOME)) {

if (home && nr >= home_nr)

continue;

home_nr = nr;

home_rw = !(flags & GPT_FLAG_READ_ONLY);

r = free_and_strdup(&home, node);

if (r < 0)

return log_oom();

} else if (sd_id128_equal(type_id, GPT_SRV)) {

if (srv && nr >= srv_nr)

continue;

srv_nr = nr;

srv_rw = !(flags & GPT_FLAG_READ_ONLY);

r = free_and_strdup(&srv, node);

if (r < 0)

return log_oom();

}

#ifdef GPT_ROOT_NATIVE

else if (sd_id128_equal(type_id, GPT_ROOT_NATIVE)) {

if (root && nr >= root_nr)

continue;

root_nr = nr;

root_rw = !(flags & GPT_FLAG_READ_ONLY);

r = free_and_strdup(&root, node);

if (r < 0)

return log_oom();

}

#endif

#ifdef GPT_ROOT_SECONDARY

else if (sd_id128_equal(type_id, GPT_ROOT_SECONDARY)) {

if (secondary_root && nr >= secondary_root_nr)

continue;

secondary_root_nr = nr;

secondary_root_rw = !(flags & GPT_FLAG_READ_ONLY);

r = free_and_strdup(&secondary_root, node);

if (r < 0)

return log_oom();

}

#endif

else if (sd_id128_equal(type_id, GPT_LINUX_GENERIC)) {

if (generic)

multiple_generic = true;

else {

generic_rw = !(flags & GPT_FLAG_READ_ONLY);

r = free_and_strdup(&generic, node);

if (r < 0)

return log_oom();

}

}

} else if (is_mbr) {

int type;

if (flags != 0x80) /* Bootable flag */

continue;

type = blkid_partition_get_type(pp);

if (type != 0x83) /* Linux partition */

continue;

if (generic)

multiple_generic = true;

else {

generic_rw = true;

r = free_and_strdup(&root, node);

if (r < 0)

return log_oom();

}

}

}

if (root) {

*root_device = root;

root = NULL;

*root_device_rw = root_rw;

*secondary = false;

} else if (secondary_root) {

*root_device = secondary_root;

secondary_root = NULL;

*root_device_rw = secondary_root_rw;

*secondary = true;

} else if (generic) {

/* There were no partitions with precise meanings

if (multiple_generic) {

log_error("Identified multiple bootable Linux partitions on\n"

" %s\n"

PARTITION_TABLE_BLURB, arg_image);

return -EINVAL;

}

*root_device = generic;

generic = NULL;

*root_device_rw = generic_rw;

*secondary = false;

} else {

log_error("Failed to identify root partition in disk image\n"

" %s\n"

PARTITION_TABLE_BLURB, arg_image);

return -EINVAL;

}

if (home) {

*home_device = home;

home = NULL;

*home_device_rw = home_rw;

}

if (srv) {

*srv_device = srv;

srv = NULL;

*srv_device_rw = srv_rw;

}

return 0;

#else

log_error("--image= is not supported, compiled without blkid support.");

return -EOPNOTSUPP;

#endif

}

static int mount_device(const char *what, const char *where, const char *directory, bool rw) {

#ifdef HAVE_BLKID

_cleanup_blkid_free_probe_ blkid_probe b = NULL;

const char *fstype, *p;

int r;

assert(what);

assert(where);

if (arg_read_only)

rw = false;

if (directory)

p = strjoina(where, directory);

else

p = where;

errno = 0;

b = blkid_new_probe_from_filename(what);

if (!b) {

if (errno == 0)

return log_oom();

log_error_errno(errno, "Failed to allocate prober for %s: %m", what);

return -errno;

}

blkid_probe_enable_superblocks(b, 1);

blkid_probe_set_superblocks_flags(b, BLKID_SUBLKS_TYPE);

errno = 0;

r = blkid_do_safeprobe(b);

if (r == -1 || r == 1) {

log_error("Cannot determine file system type of %s", what);

return -EINVAL;

} else if (r != 0) {

if (errno == 0)

errno = EIO;

log_error_errno(errno, "Failed to probe %s: %m", what);

return -errno;

}

errno = 0;

if (blkid_probe_lookup_value(b, "TYPE", &fstype, NULL) < 0) {

if (errno == 0)

errno = EINVAL;

log_error("Failed to determine file system type of %s", what);

return -errno;

}

if (streq(fstype, "crypto_LUKS")) {

log_error("nspawn currently does not support LUKS disk images.");

return -EOPNOTSUPP;

}

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

log_error("--image= is not supported, compiled without blkid support.");

return -EOPNOTSUPP;

#endif

}

static int mount_devices(

const char *where,

const char *root_device, bool root_device_rw,

const char *home_device, bool home_device_rw,

const char *srv_device, bool srv_device_rw) {

int r;

assert(where);

if (root_device) {

r = mount_device(root_device, arg_directory, NULL, root_device_rw);

if (r < 0)

return log_error_errno(r, "Failed to mount root directory: %m");

}

if (home_device) {

r = mount_device(home_device, arg_directory, "/home", home_device_rw);

if (r < 0)

return log_error_errno(r, "Failed to mount home directory: %m");

}

if (srv_device) {

r = mount_device(srv_device, arg_directory, "/srv", srv_device_rw);

if (r < 0)

return log_error_errno(r, "Failed to mount server data directory: %m");

}

return 0;

}

static void loop_remove(int nr, int *image_fd) {

_cleanup_close_ int control = -1;

int r;

if (nr < 0)

return;

if (image_fd && *image_fd >= 0) {

r = ioctl(*image_fd, LOOP_CLR_FD);

if (r < 0)

log_debug_errno(errno, "Failed to close loop image: %m");

*image_fd = safe_close(*image_fd);

}

control = open("/dev/loop-control", O_RDWR|O_CLOEXEC|O_NOCTTY|O_NONBLOCK);

if (control < 0) {

log_warning_errno(errno, "Failed to open /dev/loop-control: %m");

return;

}

r = ioctl(control, LOOP_CTL_REMOVE, nr);

if (r < 0)

log_debug_errno(errno, "Failed to remove loop %d: %m", nr);

}

static int wait_for_container(pid_t pid, ContainerStatus *container) {

siginfo_t status;

int r;

r = wait_for_terminate(pid, &status);

if (r < 0)

return log_warning_errno(r, "Failed to wait for container: %m");

switch (status.si_code) {

case CLD_EXITED:

if (status.si_status == 0) {

log_full(arg_quiet ? LOG_DEBUG : LOG_INFO, "Container %s exited successfully.", arg_machine);

} else

log_full(arg_quiet ? LOG_DEBUG : LOG_INFO, "Container %s failed with error code %i.", arg_machine, status.si_status);

*container = CONTAINER_TERMINATED;

return status.si_status;

case CLD_KILLED:

if (status.si_status == SIGINT) {

log_full(arg_quiet ? LOG_DEBUG : LOG_INFO, "Container %s has been shut down.", arg_machine);

*container = CONTAINER_TERMINATED;

return 0;

} else if (status.si_status == SIGHUP) {

log_full(arg_quiet ? LOG_DEBUG : LOG_INFO, "Container %s is being rebooted.", arg_machine);

*container = CONTAINER_REBOOTED;

return 0;

}

/* CLD_KILLED fallthrough */

case CLD_DUMPED:

log_error("Container %s terminated by signal %s.", arg_machine, signal_to_string(status.si_status));

return -EIO;

default:

log_error("Container %s failed due to unknown reason.", arg_machine);

return -EIO;

}

return r;

}

static void nop_handler(int sig) {}

static int on_orderly_shutdown(sd_event_source *s, const struct signalfd_siginfo *si, void *userdata) {

pid_t pid;

pid = PTR_TO_UINT32(userdata);

if (pid > 0) {

if (kill(pid, arg_kill_signal) >= 0) {

log_info("Trying to halt container. Send SIGTERM again to trigger immediate termination.");

sd_event_source_set_userdata(s, NULL);

return 0;

}

}

sd_event_exit(sd_event_source_get_event(s), 0);

return 0;

}

static int determine_names(void) {

int r;

if (arg_template && !arg_directory && arg_machine) {

/* If --template= was specified then we should not

arg_directory = strjoin("/var/lib/machines/", arg_machine, NULL);

if (!arg_directory)

return log_oom();

}

if (!arg_image && !arg_directory) {

if (arg_machine) {

_cleanup_(image_unrefp) Image *i = NULL;

r = image_find(arg_machine, &i);

if (r < 0)

return log_error_errno(r, "Failed to find image for machine '%s': %m", arg_machine);

else if (r == 0) {

log_error("No image for machine '%s': %m", arg_machine);

return -ENOENT;

}

if (i->type == IMAGE_RAW)

r = set_sanitized_path(&arg_image, i->path);

else

r = set_sanitized_path(&arg_directory, i->path);

if (r < 0)

return log_error_errno(r, "Invalid image directory: %m");

if (!arg_ephemeral)

arg_read_only = arg_read_only || i->read_only;

} else

arg_directory = get_current_dir_name();

if (!arg_directory && !arg_machine) {

log_error("Failed to determine path, please use -D or -i.");

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);

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

if (asprintf(&b, "%s-%016" PRIx64, arg_machine, random_u64()) < 0)

return log_oom();

free(arg_machine);

arg_machine = b;

}

}

return 0;

}

static int determine_uid_shift(const char *directory) {

int r;

if (!arg_userns) {

arg_uid_shift = 0;

return 0;

}

if (arg_uid_shift == UID_INVALID) {

struct stat st;

r = stat(directory, &st);

if (r < 0)

return log_error_errno(errno, "Failed to determine UID base of %s: %m", directory);

arg_uid_shift = st.st_uid & UINT32_C(0xffff0000);

if (arg_uid_shift != (st.st_gid & UINT32_C(0xffff0000))) {

log_error("UID and GID base of %s don't match.", directory);

return -EINVAL;

}

arg_uid_range = UINT32_C(0x10000);

}

if (arg_uid_shift > (uid_t) -1 - arg_uid_range) {

log_error("UID base too high for UID range.");

return -EINVAL;

}

log_info("Using user namespaces with base " UID_FMT " and range " UID_FMT ".", arg_uid_shift, arg_uid_range);

return 0;

}

static int inner_child(

Barrier *barrier,

const char *directory,

bool secondary,

int kmsg_socket,

int rtnl_socket,

FDSet *fds) {

_cleanup_free_ char *home = NULL;

unsigned n_env = 2;

const char *envp[] = {

"PATH=" DEFAULT_PATH_SPLIT_USR,

"container=systemd-nspawn", /* LXC sets container=lxc, so follow the scheme here */

NULL, /* TERM */

NULL, /* HOME */

NULL, /* USER */

NULL, /* LOGNAME */

NULL, /* container_uuid */

NULL, /* LISTEN_FDS */

NULL, /* LISTEN_PID */

NULL

};

_cleanup_strv_free_ char **env_use = NULL;

int r;

assert(barrier);

assert(directory);

assert(kmsg_socket >= 0);

cg_unified_flush();

if (arg_userns) {

/* Tell the parent, that it now can write the UID map. */

(void) barrier_place(barrier); /* #1 */

/* Wait until the parent wrote the UID map */

if (!barrier_place_and_sync(barrier)) { /* #2 */

log_error("Parent died too early");

return -ESRCH;

}

}

r = mount_all(NULL, true, arg_uid_shift, arg_uid_range, arg_selinux_apifs_context);

if (r < 0)

return r;

/* Wait until we are cgroup-ified, so that we

if (!barrier_place_and_sync(barrier)) { /* #3 */

log_error("Parent died too early");

return -ESRCH;

}

r = mount_systemd_cgroup_writable("", arg_unified_cgroup_hierarchy);

if (r < 0)

return r;

r = reset_uid_gid();

if (r < 0)

return log_error_errno(r, "Couldn't become new root: %m");

r = setup_boot_id(NULL);

if (r < 0)

return r;

r = setup_kmsg(NULL, kmsg_socket);

if (r < 0)

return r;

kmsg_socket = safe_close(kmsg_socket);

umask(0022);

if (setsid() < 0)

return log_error_errno(errno, "setsid() failed: %m");

if (arg_private_network)

loopback_setup();

if (arg_expose_ports) {

r = expose_port_send_rtnl(rtnl_socket);

if (r < 0)

return r;

rtnl_socket = safe_close(rtnl_socket);

}

if (drop_capabilities() < 0)

return log_error_errno(errno, "drop_capabilities() failed: %m");

setup_hostname();

if (arg_personality != PERSONALITY_INVALID) {

if (personality(arg_personality) < 0)

return log_error_errno(errno, "personality() failed: %m");

} else if (secondary) {

if (personality(PER_LINUX32) < 0)

return log_error_errno(errno, "personality() failed: %m");

}

#ifdef HAVE_SELINUX

if (arg_selinux_context)

if (setexeccon((security_context_t) arg_selinux_context) < 0)

return log_error_errno(errno, "setexeccon(\"%s\") failed: %m", arg_selinux_context);

#endif

r = change_uid_gid(arg_user, &home);

if (r < 0)

return r;

envp[n_env] = strv_find_prefix(environ, "TERM=");

if (envp[n_env])

n_env ++;

if ((asprintf((char**)(envp + n_env++), "HOME=%s", home ? home: "/root") < 0) ||

(asprintf((char**)(envp + n_env++), "USER=%s", arg_user ? arg_user : "root") < 0) ||

(asprintf((char**)(envp + n_env++), "LOGNAME=%s", arg_user ? arg_user : "root") < 0))

return log_oom();

if (!sd_id128_equal(arg_uuid, SD_ID128_NULL)) {

char as_uuid[37];

if (asprintf((char**)(envp + n_env++), "container_uuid=%s", id128_format_as_uuid(arg_uuid, as_uuid)) < 0)

return log_oom();

}

if (fdset_size(fds) > 0) {

r = fdset_cloexec(fds, false);

if (r < 0)

return log_error_errno(r, "Failed to unset O_CLOEXEC for file descriptors.");

if ((asprintf((char **)(envp + n_env++), "LISTEN_FDS=%u", fdset_size(fds)) < 0) ||

(asprintf((char **)(envp + n_env++), "LISTEN_PID=1") < 0))

return log_oom();

}

env_use = strv_env_merge(2, envp, arg_setenv);

if (!env_use)

return log_oom();

/* Let the parent know that we are ready and

if (!barrier_place_and_sync(barrier)) { /* #4 */

log_error("Parent died too early");

return -ESRCH;

}

/* Now, explicitly close the log, so that we

log_close();

(void) fdset_close_others(fds);

if (arg_boot) {

char **a;

size_t m;

/* Automatically search for the init system */

m = 1 + strv_length(arg_parameters);

a = newa(char*, m + 1);

if (strv_isempty(arg_parameters))

a[1] = NULL;

else

memcpy(a + 1, arg_parameters, m * sizeof(char*));

a[0] = (char*) "/usr/lib/systemd/systemd";

execve(a[0], a, env_use);

a[0] = (char*) "/lib/systemd/systemd";

execve(a[0], a, env_use);

a[0] = (char*) "/sbin/init";

execve(a[0], a, env_use);

} else if (!strv_isempty(arg_parameters))

execvpe(arg_parameters[0], arg_parameters, env_use);

else {

chdir(home ?: "/root");

execle("/bin/bash", "-bash", NULL, env_use);

execle("/bin/sh", "-sh", NULL, env_use);

}

(void) log_open();

return log_error_errno(errno, "execv() failed: %m");

}

static int outer_child(

Barrier *barrier,

const char *directory,

const char *console,

const char *root_device, bool root_device_rw,

const char *home_device, bool home_device_rw,

const char *srv_device, bool srv_device_rw,

bool interactive,

bool secondary,

int pid_socket,

int kmsg_socket,

int rtnl_socket,

int uid_shift_socket,

FDSet *fds) {

pid_t pid;

ssize_t l;

int r;

assert(barrier);

assert(directory);

assert(console);

assert(pid_socket >= 0);

assert(kmsg_socket >= 0);

cg_unified_flush();

if (prctl(PR_SET_PDEATHSIG, SIGKILL) < 0)

return log_error_errno(errno, "PR_SET_PDEATHSIG failed: %m");

if (interactive) {

close_nointr(STDIN_FILENO);

close_nointr(STDOUT_FILENO);

close_nointr(STDERR_FILENO);

r = open_terminal(console, O_RDWR);

if (r != STDIN_FILENO) {

if (r >= 0) {

safe_close(r);

r = -EINVAL;

}

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

}

if (dup2(STDIN_FILENO, STDOUT_FILENO) != STDOUT_FILENO ||

dup2(STDIN_FILENO, STDERR_FILENO) != STDERR_FILENO)

return log_error_errno(errno, "Failed to duplicate console: %m");

}

r = reset_audit_loginuid();

if (r < 0)

return r;

/* Mark everything as slave, so that we still

if (mount(NULL, "/", NULL, MS_SLAVE|MS_REC, NULL) < 0)

return log_error_errno(errno, "MS_SLAVE|MS_REC failed: %m");

r = mount_devices(directory,

root_device, root_device_rw,

home_device, home_device_rw,

srv_device, srv_device_rw);

if (r < 0)

return r;

r = determine_uid_shift(directory);

if (r < 0)

return r;

if (arg_userns) {

l = send(uid_shift_socket, &arg_uid_shift, sizeof(arg_uid_shift), MSG_NOSIGNAL);

if (l < 0)

return log_error_errno(errno, "Failed to send UID shift: %m");

if (l != sizeof(arg_uid_shift)) {

log_error("Short write while sending UID shift.");

return -EIO;

}

}

/* Turn directory into bind mount */

if (mount(directory, directory, NULL, MS_BIND|MS_REC, NULL) < 0)

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

r = setup_volatile(directory, arg_volatile_mode, arg_userns, arg_uid_shift, arg_uid_range, arg_selinux_context);

if (r < 0)

return r;

r = setup_volatile_state(directory, arg_volatile_mode, arg_userns, arg_uid_shift, arg_uid_range, arg_selinux_context);

if (r < 0)

return r;

r = base_filesystem_create(directory, arg_uid_shift, (gid_t) arg_uid_shift);

if (r < 0)

return r;

if (arg_read_only) {

r = bind_remount_recursive(directory, true);

if (r < 0)

return log_error_errno(r, "Failed to make tree read-only: %m");

}

r = mount_all(directory, false, arg_uid_shift, arg_uid_range, arg_selinux_apifs_context);

if (r < 0)

return r;

r = copy_devnodes(directory);

if (r < 0)

return r;

dev_setup(directory, arg_uid_shift, arg_uid_shift);

r = setup_pts(directory);

if (r < 0)

return r;

r = setup_propagate(directory);

if (r < 0)

return r;

r = setup_dev_console(directory, console);

if (r < 0)

return r;

r = setup_seccomp();

if (r < 0)

return r;

r = setup_timezone(directory);

if (r < 0)

return r;

r = setup_resolv_conf(directory);

if (r < 0)

return r;

r = setup_journal(directory);

if (r < 0)

return r;

r = mount_custom(directory, arg_custom_mounts, arg_n_custom_mounts, arg_userns, arg_uid_shift, arg_uid_range, arg_selinux_apifs_context);

if (r < 0)

return r;

r = mount_cgroups(directory, arg_unified_cgroup_hierarchy, arg_userns, arg_uid_shift, arg_uid_range, arg_selinux_apifs_context);

if (r < 0)

return r;

r = mount_move_root(directory);

if (r < 0)

return log_error_errno(r, "Failed to move root directory: %m");

pid = raw_clone(SIGCHLD|CLONE_NEWNS|

(arg_share_system ? 0 : CLONE_NEWIPC|CLONE_NEWPID|CLONE_NEWUTS) |

(arg_private_network ? CLONE_NEWNET : 0) |

(arg_userns ? CLONE_NEWUSER : 0),

NULL);

if (pid < 0)

return log_error_errno(errno, "Failed to fork inner child: %m");

if (pid == 0) {

pid_socket = safe_close(pid_socket);

uid_shift_socket = safe_close(uid_shift_socket);

/* The inner child has all namespaces that are

r = inner_child(barrier, directory, secondary, kmsg_socket, rtnl_socket, fds);

if (r < 0)

_exit(EXIT_FAILURE);

_exit(EXIT_SUCCESS);

}

l = send(pid_socket, &pid, sizeof(pid), MSG_NOSIGNAL);

if (l < 0)

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

if (l != sizeof(pid)) {

log_error("Short write while sending PID.");

return -EIO;

}

pid_socket = safe_close(pid_socket);

return 0;

}

static int setup_uid_map(pid_t pid) {

char uid_map[strlen("/proc//uid_map") + DECIMAL_STR_MAX(uid_t) + 1], line[DECIMAL_STR_MAX(uid_t)*3+3+1];