#include <signal.h>

#include <sched.h>

#include <unistd.h>

#include <sys/types.h>

#include <sys/syscall.h>

#include <sys/mount.h>

#include <sys/wait.h>

#include <stdlib.h>

#include <string.h>

#include <stdio.h>

#include <errno.h>

#include <sys/prctl.h>

#include <getopt.h>

#include <termios.h>

#include <sys/signalfd.h>

#include <grp.h>

#include <linux/fs.h>

#include <sys/un.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>

#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-rtnl.h"

#include "log.h"

#include "util.h"

#include "mkdir.h"

#include "macro.h"

#include "audit.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 "bus-kernel.h"

#include "env-util.h"

#include "def.h"

#include "rtnl-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 "fw-util.h"

#include "local-addresses.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;

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 char **arg_bind = NULL;

static char **arg_bind_ro = NULL;

static char **arg_tmpfs = NULL;

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 bool arg_network_veth = false;

static const char *arg_network_bridge = NULL;

static unsigned long arg_personality = 0xffffffffLU;

static char *arg_image = NULL;

static Volatile arg_volatile = VOLATILE_NO;

static ExposePort *arg_expose_ports = 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"

" --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"

" -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"

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

" the container\n"

" --bind-ro=PATH[:PATH] Similar, but creates a read-only bind mount\n"

" --tmpfs=PATH:[OPTIONS] Mount an empty tmpfs to the specified directory\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 int set_sanitized_path(char **b, const char *path) {

char *p;

assert(b);

assert(path);

p = canonicalize_file_name(path);

if (!p) {

if (errno != ENOENT)

return -errno;

p = path_make_absolute_cwd(path);

if (!p)

return -ENOMEM;

}

free(*b);

*b = path_kill_slashes(p);

return 0;

}

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

enum {

ARG_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_SETENV,

ARG_SHARE_SYSTEM,

ARG_REGISTER,

ARG_KEEP_UNIT,

ARG_NETWORK_INTERFACE,

ARG_NETWORK_MACVLAN,

ARG_NETWORK_BRIDGE,

ARG_PERSONALITY,

ARG_VOLATILE,

ARG_TEMPLATE,

};

static const struct option options[] = {

{ "help", no_argument, NULL, 'h' },

{ "version", no_argument, NULL, ARG_VERSION },

{ "directory", required_argument, NULL, 'D' },

{ "template", required_argument, NULL, ARG_TEMPLATE },

{ "ephemeral", no_argument, NULL, 'x' },

{ "user", required_argument, NULL, 'u' },

{ "private-network", no_argument, NULL, ARG_PRIVATE_NETWORK },

{ "boot", no_argument, NULL, 'b' },

{ "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 },

{ "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-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' },

{}

};

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':

free(arg_user);

arg_user = strdup(optarg);

if (!arg_user)

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();

/* 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)) {

free(arg_machine);

arg_machine = NULL;

} 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: {

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

char *e;

char ***x;

x = c == ARG_BIND ? &arg_bind : &arg_bind_ro;

e = strchr(optarg, ':');

if (e) {

a = strndup(optarg, e - optarg);

b = strdup(e + 1);

} else {

a = strdup(optarg);

b = strdup(optarg);

}

if (!a || !b)

return log_oom();

if (!path_is_absolute(a) || !path_is_absolute(b)) {

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

return -EINVAL;

}

r = strv_extend(x, a);

if (r < 0)

return log_oom();

r = strv_extend(x, b);

if (r < 0)

return log_oom();

break;

}

case ARG_TMPFS: {

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

char *e;

e = strchr(optarg, ':');

if (e) {

a = strndup(optarg, e - optarg);

b = strdup(e + 1);

} else {

a = strdup(optarg);

b = strdup("mode=0755");

}

if (!a || !b)

return log_oom();

if (!path_is_absolute(a)) {

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

return -EINVAL;

}

r = strv_push(&arg_tmpfs, a);

if (r < 0)

return log_oom();

a = NULL;

r = strv_push(&arg_tmpfs, b);

if (r < 0)

return log_oom();

b = 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 == 0xffffffffLU) {

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 '?':

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;

}

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

return 1;

}

static int mount_all(const char *dest) {

typedef struct MountPoint {

const char *what;

const char *where;

const char *type;

const char *options;

unsigned long flags;

bool fatal;

} MountPoint;

static const MountPoint mount_table[] = {

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

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

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

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

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

{ "devpts", "/dev/pts", "devpts","newinstance,ptmxmode=0666,mode=620,gid=" STRINGIFY(TTY_GID), MS_NOSUID|MS_NOEXEC, true },

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

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

#ifdef HAVE_SELINUX

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

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

#endif

};

unsigned k;

int r = 0;

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

_cleanup_free_ char *where = NULL;

#ifdef HAVE_SELINUX

_cleanup_free_ char *options = NULL;

#endif

const char *o;

int t;

where = strjoin(dest, "/", mount_table[k].where, NULL);

if (!where)

return log_oom();

t = path_is_mount_point(where, true);

if (t < 0) {

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

if (r == 0)

r = t;

continue;

}

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

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

continue;

t = mkdir_p(where, 0755);

if (t < 0) {

if (mount_table[k].fatal) {

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

if (r == 0)

r = t;

} else

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

continue;

}

#ifdef HAVE_SELINUX

if (arg_selinux_apifs_context &&

(streq_ptr(mount_table[k].what, "tmpfs") || streq_ptr(mount_table[k].what, "devpts"))) {

options = strjoin(mount_table[k].options, ",context=\"", arg_selinux_apifs_context, "\"", NULL);

if (!options)

return log_oom();

o = options;

} else

#endif

o = mount_table[k].options;

if (mount(mount_table[k].what,

where,

mount_table[k].type,

mount_table[k].flags,

o) < 0) {

if (mount_table[k].fatal) {

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

if (r == 0)

r = -errno;

} else

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

}

}

return r;

}

static int mount_binds(const char *dest, char **l, bool ro) {

char **x, **y;

STRV_FOREACH_PAIR(x, y, l) {

_cleanup_free_ char *where = NULL;

struct stat source_st, dest_st;

int r;

if (stat(*x, &source_st) < 0)

return log_error_errno(errno, "Failed to stat %s: %m", *x);

where = strappend(dest, *y);

if (!where)

return log_oom();

r = stat(where, &dest_st);

if (r == 0) {

if ((source_st.st_mode & S_IFMT) != (dest_st.st_mode & S_IFMT)) {

log_error("The file types of %s and %s do not match. Refusing bind mount", *x, where);

return -EINVAL;

}

} else if (errno == ENOENT) {

r = mkdir_parents_label(where, 0755);

if (r < 0)

return log_error_errno(r, "Failed to bind mount %s: %m", *x);

} else {

log_error_errno(errno, "Failed to bind mount %s: %m", *x);

return -errno;

}

/* Create the mount point, but be conservative -- refuse to create block

if (S_ISDIR(source_st.st_mode)) {

r = mkdir_label(where, 0755);

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

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

} else if (S_ISFIFO(source_st.st_mode)) {

r = mkfifo(where, 0644);

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

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

} else if (S_ISSOCK(source_st.st_mode)) {

r = mknod(where, 0644 | S_IFSOCK, 0);

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

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

} else if (S_ISREG(source_st.st_mode)) {

r = touch(where);

if (r < 0)

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

} else {

log_error("Refusing to create mountpoint for file: %s", *x);

return -ENOTSUP;

}

if (mount(*x, where, "bind", MS_BIND, NULL) < 0)

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

if (ro) {

r = bind_remount_recursive(where, true);

if (r < 0)

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

}

}

return 0;

}

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

char *to;

int r;

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

r = path_is_mount_point(to, false);

if (r < 0)

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

if (r > 0)

return 0;

mkdir_p(to, 0755);

if (mount("cgroup", to, "cgroup", MS_NOSUID|MS_NOEXEC|MS_NODEV|(read_only ? MS_RDONLY : 0), controller) < 0)

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

return 1;

}

static int mount_cgroup(const char *dest) {

_cleanup_set_free_free_ Set *controllers = NULL;

_cleanup_free_ char *own_cgroup_path = NULL;

const char *cgroup_root, *systemd_root, *systemd_own;

int r;

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");

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");

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

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

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

for (;;) {

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

controller = set_steal_first(controllers);

if (!controller)

break;

origin = strappend("/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_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 = strjoin(dest, "/sys/fs/cgroup/", controller, NULL);

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_cgroup_hierarchy(dest, combined, combined, true);

if (r < 0)

return r;

if (symlink(combined, target) < 0)

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

}

}

r = mount_cgroup_hierarchy(dest, "name=systemd", "systemd", false);

if (r < 0)

return r;

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

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

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 */

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

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");

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_tmpfs(const char *dest) {

char **i, **o;

STRV_FOREACH_PAIR(i, o, arg_tmpfs) {

_cleanup_free_ char *where = NULL;

int r;

where = strappend(dest, *i);

if (!where)

return log_oom();

r = mkdir_label(where, 0755);

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

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

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

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

}

return 0;

}

static int setup_timezone(const char *dest) {

_cleanup_free_ char *where = NULL, *p = NULL, *q = NULL, *check = NULL, *what = NULL;

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 = strappend(dest, "/etc/localtime");

if (!where)

return log_oom();

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 = strjoin(dest, "/usr/share/zoneinfo/", z, NULL);

if (!check)

return log_oom();

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

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

return 0;

}

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

if (!what)

return log_oom();

r = mkdir_parents(where, 0755);

if (r < 0) {

log_error_errno(r, "Failed to create directory for timezone info %s in container: %m", where);

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;

}

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

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

return 0;

}

return 0;

}

static int setup_resolv_conf(const char *dest) {

_cleanup_free_ char *where = NULL;

int r;

assert(dest);

if (arg_private_network)

return 0;

/* Fix resolv.conf, if possible */

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

if (!where)

return log_oom();

/* We don't really care for the results of this really. If it

r = mkdir_parents(where, 0755);

if (r < 0) {

log_warning_errno(r, "Failed to create parent directory for resolv.conf %s: %m", where);

return 0;

}

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

if (r < 0) {

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

return 0;

}

return 0;

}

static int setup_volatile_state(const char *directory) {

const char *p;

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 = strappenda(directory, "/var");

r = mkdir(p, 0755);

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

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

if (mount("tmpfs", p, "tmpfs", MS_STRICTATIME, "mode=755") < 0)

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";

const char *f, *t;

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");

if (mount("tmpfs", template, "tmpfs", MS_STRICTATIME, "mode=755") < 0) {

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

r = -errno;

goto fail;

}

tmpfs_mounted = true;

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

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

r = mkdir(t, 0755);

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

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

r = -errno;

goto fail;

}

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

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

r = -errno;

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) {

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

r = -errno;

goto fail;

}

rmdir(template);

return 0;

fail:

if (bind_mounted)

umount(t);

if (tmpfs_mounted)

umount(template);

rmdir(template);

return r;

}

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

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) {

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

sd_id128_t rnd = {};

char as_uuid[37];

int r;

assert(dest);

if (arg_share_system)

return 0;

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

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

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

if (!from || !to)

return log_oom();

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);

if (r < 0)

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

if (mount(from, to, "bind", MS_BIND, NULL) < 0) {

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

r = -errno;

} else if (mount(from, to, "bind", MS_BIND|MS_REMOUNT|MS_RDONLY, NULL))

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);

NULSTR_FOREACH(d, devnodes) {

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

struct stat st;

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

to = strjoin(dest, "/dev/", d, NULL);

if (!from || !to)

return log_oom();

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 {

r = mkdir_parents(to, 0775);

if (r < 0) {

log_error_errno(r, "Failed to create parent directory of %s: %m", to);

return -r;

}

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

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

}

}

return r;

}

static int setup_ptmx(const char *dest) {

_cleanup_free_ char *p = NULL;

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

if (!p)

return log_oom();

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

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

return 0;

}

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

_cleanup_umask_ mode_t u;

const char *to;

struct stat st;

int r;

assert(dest);

assert(console);

u = umask(0000);

if (stat("/dev/null", &st) < 0)

return log_error_errno(errno, "Failed to stat /dev/null: %m");

r = chmod_and_chown(console, 0600, 0, 0);

if (r < 0)

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 = strappenda(dest, "/dev/console");

if (mknod(to, (st.st_mode & ~07777) | 0600, st.st_rdev) < 0)

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

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

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

return 0;

}

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

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

_cleanup_umask_ mode_t u;

int r, 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(dest);

assert(kmsg_socket >= 0);

u = umask(0000);

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

if (asprintf(&from, "%s/dev/kmsg", dest) < 0 ||

asprintf(&to, "%s/proc/kmsg", dest) < 0)

return log_oom();

if (mkfifo(from, 0600) < 0)

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

r = chmod_and_chown(from, 0600, 0, 0);

if (r < 0)

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

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

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 /dev/kmsg... */

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_rtnl *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_rtnl *rtnl, sd_rtnl_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_rtnl **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_rtnl_unref_ sd_rtnl *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_rtnl_open_fd(&rtnl, fd, 1, RTNLGRP_IPV4_IFADDR);

if (r < 0) {

safe_close(fd);

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

}

r = sd_rtnl_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_rtnl_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_rtnl_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 *p = NULL, *b = NULL, *q = NULL, *d = NULL;

char *id;

int r;

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

if (arg_ephemeral)

return 0;

p = strappend(directory, "/etc/machine-id");

if (!p)

return log_oom();

r = read_one_line_file(p, &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", p);

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", p);

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;

free(p);

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

q = strjoin(directory, "/var/log/journal/", id, NULL);

if (!p || !q)

return log_oom();

if (path_is_mount_point(p, false) > 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, false) > 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 = mkdir_p(q, 0755);

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 = mkdir_p(q, 0755);

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 = mkdir_p(q, 0755);

if (r < 0) {

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

return r;

}

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

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

return 0;

}

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_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;

r = sd_bus_message_new_method_call(

bus,

&m,

"org.freedesktop.machine1",

"/org/freedesktop/machine1",

"org.freedesktop.machine1.Manager",

"CreateMachineWithNetwork");

if (r < 0)

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

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

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

if (r < 0)

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

if (!isempty(arg_slice)) {

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

if (r < 0)

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

}

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

if (r < 0)

return log_error_errno(r, "Failed to add device policy: %m");

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 log_error_errno(r, "Failed to add device whitelist: %m");

r = sd_bus_message_close_container(m);

if (r < 0)

return log_error_errno(r, "Failed to close container: %m");

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_close_unref_ sd_bus *bus = NULL;

const char *path;

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");

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");

if (r < 0) {

log_error("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... (%s)\n", strerror(-r));

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);

assert_cc(ETH_ALEN <= sizeof(result));

memcpy(mac->ether_addr_octet, result, ETH_ALEN);

/* see eth_random_addr in the kernel */

mac->ether_addr_octet[0] &= 0xfe; /* clear multicast bit */

mac->ether_addr_octet[0] |= 0x02; /* set local assignment bit (IEEE802) */

return 0;

}

static int setup_veth(pid_t pid, char iface_name[IFNAMSIZ], int *ifi) {

_cleanup_rtnl_message_unref_ sd_rtnl_message *m = NULL;

_cleanup_rtnl_unref_ sd_rtnl *rtnl = NULL;

struct ether_addr mac_host, mac_container;

int r, i;

if (!arg_private_network)

return 0;

if (!arg_network_veth)

return 0;

/* Use two different interface name prefixes depending whether

snprintf(iface_name, IFNAMSIZ - 1, "%s-%s",

arg_network_bridge ? "vb" : "ve", arg_machine);

r = generate_mac(&mac_container, CONTAINER_HASH_KEY, 0);

if (r < 0)

return log_error_errno(r, "Failed to generate predictable MAC address for container side: %m");

r = generate_mac(&mac_host, HOST_HASH_KEY, 0);

if (r < 0)

return log_error_errno(r, "Failed to generate predictable MAC address for host side: %m");

r = sd_rtnl_open(&rtnl, 0);

if (r < 0)

return log_error_errno(r, "Failed to connect to netlink: %m");

r = sd_rtnl_message_new_link(rtnl, &m, RTM_NEWLINK, 0);

if (r < 0)

return log_error_errno(r, "Failed to allocate netlink message: %m");

r = sd_rtnl_message_append_string(m, IFLA_IFNAME, iface_name);

if (r < 0)

return log_error_errno(r, "Failed to add netlink interface name: %m");

r = sd_rtnl_message_append_ether_addr(m, IFLA_ADDRESS, &mac_host);

if (r < 0)

return log_error_errno(r, "Failed to add netlink MAC address: %m");

r = sd_rtnl_message_open_container(m, IFLA_LINKINFO);

if (r < 0)

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

r = sd_rtnl_message_open_container_union(m, IFLA_INFO_DATA, "veth");

if (r < 0)

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

r = sd_rtnl_message_open_container(m, VETH_INFO_PEER);

if (r < 0)

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

r = sd_rtnl_message_append_string(m, IFLA_IFNAME, "host0");

if (r < 0)

return log_error_errno(r, "Failed to add netlink interface name: %m");

r = sd_rtnl_message_append_ether_addr(m, IFLA_ADDRESS, &mac_container);

if (r < 0)

return log_error_errno(r, "Failed to add netlink MAC address: %m");

r = sd_rtnl_message_append_u32(m, IFLA_NET_NS_PID, pid);

if (r < 0)

return log_error_errno(r, "Failed to add netlink namespace field: %m");

r = sd_rtnl_message_close_container(m);

if (r < 0)

return log_error_errno(r, "Failed to close netlink container: %m");

r = sd_rtnl_message_close_container(m);

if (r < 0)

return log_error_errno(r, "Failed to close netlink container: %m");

r = sd_rtnl_message_close_container(m);

if (r < 0)

return log_error_errno(r, "Failed to close netlink container: %m");

r = sd_rtnl_call(rtnl, m, 0, NULL);

if (r < 0)

return log_error_errno(r, "Failed to add new veth interfaces: %m");

i = (int) if_nametoindex(iface_name);

if (i <= 0)

return log_error_errno(errno, "Failed to resolve interface %s: %m", iface_name);

*ifi = i;

return 0;

}

static int setup_bridge(const char veth_name[], int *ifi) {

_cleanup_rtnl_message_unref_ sd_rtnl_message *m = NULL;

_cleanup_rtnl_unref_ sd_rtnl *rtnl = NULL;

int r, bridge;

if (!arg_private_network)

return 0;

if (!arg_network_veth)

return 0;

if (!arg_network_bridge)

return 0;

bridge = (int) if_nametoindex(arg_network_bridge);

if (bridge <= 0)

return log_error_errno(errno, "Failed to resolve interface %s: %m", arg_network_bridge);

*ifi = bridge;

r = sd_rtnl_open(&rtnl, 0);

if (r < 0)

return log_error_errno(r, "Failed to connect to netlink: %m");

r = sd_rtnl_message_new_link(rtnl, &m, RTM_SETLINK, 0);

if (r < 0)

return log_error_errno(r, "Failed to allocate netlink message: %m");

r = sd_rtnl_message_link_set_flags(m, IFF_UP, IFF_UP);

if (r < 0)

return log_error_errno(r, "Failed to set IFF_UP flag: %m");

r = sd_rtnl_message_append_string(m, IFLA_IFNAME, veth_name);

if (r < 0)

return log_error_errno(r, "Failed to add netlink interface name field: %m");

r = sd_rtnl_message_append_u32(m, IFLA_MASTER, bridge);

if (r < 0)

return log_error_errno(r, "Failed to add netlink master field: %m");

r = sd_rtnl_call(rtnl, m, 0, NULL);

if (r < 0)

return log_error_errno(r, "Failed to add veth interface to bridge: %m");

return 0;

}

static int parse_interface(struct udev *udev, const char *name) {

_cleanup_udev_device_unref_ struct udev_device *d = NULL;

char ifi_str[2 + DECIMAL_STR_MAX(int)];

int ifi;

ifi = (int) if_nametoindex(name);

if (ifi <= 0)

return log_error_errno(errno, "Failed to resolve interface %s: %m", name);

sprintf(ifi_str, "n%i", ifi);

d = udev_device_new_from_device_id(udev, ifi_str);

if (!d)

return log_error_errno(errno, "Failed to get udev device for interface %s: %m", name);

if (udev_device_get_is_initialized(d) <= 0) {

log_error("Network interface %s is not initialized yet.", name);

return -EBUSY;

}

return ifi;

}

static int move_network_interfaces(pid_t pid) {

_cleanup_udev_unref_ struct udev *udev = NULL;

_cleanup_rtnl_unref_ sd_rtnl *rtnl = NULL;

char **i;

int r;

if (!arg_private_network)

return 0;

if (strv_isempty(arg_network_interfaces))

return 0;

r = sd_rtnl_open(&rtnl, 0);

if (r < 0)

return log_error_errno(r, "Failed to connect to netlink: %m");

udev = udev_new();

if (!udev) {

log_error("Failed to connect to udev.");

return -ENOMEM;

}

STRV_FOREACH(i, arg_network_interfaces) {

_cleanup_rtnl_message_unref_ sd_rtnl_message *m = NULL;

int ifi;

ifi = parse_interface(udev, *i);

if (ifi < 0)

return ifi;

r = sd_rtnl_message_new_link(rtnl, &m, RTM_SETLINK, ifi);

if (r < 0)

return log_error_errno(r, "Failed to allocate netlink message: %m");

r = sd_rtnl_message_append_u32(m, IFLA_NET_NS_PID, pid);

if (r < 0)

return log_error_errno(r, "Failed to append namespace PID to netlink message: %m");

r = sd_rtnl_call(rtnl, m, 0, NULL);

if (r < 0)

return log_error_errno(r, "Failed to move interface %s to namespace: %m", *i);

}

return 0;

}

static int setup_macvlan(pid_t pid) {

_cleanup_udev_unref_ struct udev *udev = NULL;

_cleanup_rtnl_unref_ sd_rtnl *rtnl = NULL;

unsigned idx = 0;

char **i;

int r;

if (!arg_private_network)

return 0;

if (strv_isempty(arg_network_macvlan))

return 0;

r = sd_rtnl_open(&rtnl, 0);

if (r < 0)

return log_error_errno(r, "Failed to connect to netlink: %m");

udev = udev_new();

if (!udev) {

log_error("Failed to connect to udev.");

return -ENOMEM;

}

STRV_FOREACH(i, arg_network_macvlan) {

_cleanup_rtnl_message_unref_ sd_rtnl_message *m = NULL;

_cleanup_free_ char *n = NULL;

struct ether_addr mac;

int ifi;

ifi = parse_interface(udev, *i);

if (ifi < 0)

return ifi;

r = generate_mac(&mac, MACVLAN_HASH_KEY, idx++);

if (r < 0)

return log_error_errno(r, "Failed to create MACVLAN MAC address: %m");

r = sd_rtnl_message_new_link(rtnl, &m, RTM_NEWLINK, 0);

if (r < 0)

return log_error_errno(r, "Failed to allocate netlink message: %m");

r = sd_rtnl_message_append_u32(m, IFLA_LINK, ifi);

if (r < 0)

return log_error_errno(r, "Failed to add netlink interface index: %m");

n = strappend("mv-", *i);

if (!n)

return log_oom();

strshorten(n, IFNAMSIZ-1);

r = sd_rtnl_message_append_string(m, IFLA_IFNAME, n);

if (r < 0)

return log_error_errno(r, "Failed to add netlink interface name: %m");

r = sd_rtnl_message_append_ether_addr(m, IFLA_ADDRESS, &mac);

if (r < 0)

return log_error_errno(r, "Failed to add netlink MAC address: %m");

r = sd_rtnl_message_append_u32(m, IFLA_NET_NS_PID, pid);

if (r < 0)

return log_error_errno(r, "Failed to add netlink namespace field: %m");

r = sd_rtnl_message_open_container(m, IFLA_LINKINFO);

if (r < 0)

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

r = sd_rtnl_message_open_container_union(m, IFLA_INFO_DATA, "macvlan");

if (r < 0)

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

r = sd_rtnl_message_append_u32(m, IFLA_MACVLAN_MODE, MACVLAN_MODE_BRIDGE);

if (r < 0)

return log_error_errno(r, "Failed to append macvlan mode: %m");

r = sd_rtnl_message_close_container(m);

if (r < 0)

return log_error_errno(r, "Failed to close netlink container: %m");

r = sd_rtnl_message_close_container(m);

if (r < 0)

return log_error_errno(r, "Failed to close netlink container: %m");

r = sd_rtnl_call(rtnl, m, 0, NULL);

if (r < 0)

return log_error_errno(r, "Failed to add new macvlan interfaces: %m");

}

return 0;

}

static int setup_seccomp(void) {

#ifdef HAVE_SECCOMP

static const int blacklist[] = {

SCMP_SYS(kexec_load),

SCMP_SYS(open_by_handle_at),

SCMP_SYS(init_module),

SCMP_SYS(finit_module),

SCMP_SYS(delete_module),

SCMP_SYS(iopl),

SCMP_SYS(ioperm),

SCMP_SYS(swapon),

SCMP_SYS(swapoff),

};

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++) {

r = seccomp_rule_add(seccomp, SCMP_ACT_ERRNO(EPERM), blacklist[i], 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 < 0)

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

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

(void) mkdir_p(p, 0600);

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

mkdir_parents(q, 0755);

mkdir_p(q, 0600);

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

return log_error_errno(errno, "Failed to install propagation bind mount.");

if (mount(NULL, q, NULL, MS_BIND|MS_REMOUNT|MS_RDONLY, NULL) < 0)

return log_error_errno(errno, "Failed to make propagation mount read-only");

return 0;

}

static int setup_image(char **device_path, int *loop_nr) {

struct loop_info64 info = {

.lo_flags = LO_FLAGS_AUTOCLEAR|LO_FLAGS_PARTSCAN

};

_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;

}

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;

_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;

const char *pttype = NULL;

blkid_partlist pl;

struct stat st;

int r;

assert(fd >= 0);

assert(root_device);

assert(home_device);

assert(srv_device);

assert(secondary);

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 %s.\n"

"Note that the disk image needs to follow http://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec/ to be supported by systemd-nspawn.", arg_image);

return -EINVAL;

} else if (r != 0) {

if (errno == 0)

errno = EIO;

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

return -errno;

}

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

if (!streq_ptr(pttype, "gpt")) {

log_error("Image %s does not carry a GUID Partition Table.\n"

"Note that the disk image needs to follow http://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec/ to be supported by systemd-nspawn.", arg_image);

return -EINVAL;

}

errno = 0;

pl = blkid_probe_get_partitions(b);

if (!pl) {

if (errno == 0)

return log_oom();

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

return -errno;

}

udev = udev_new();

if (!udev)

return log_oom();

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

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();

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);

first = udev_enumerate_get_list_entry(e);

udev_list_entry_foreach(item, first) {

_cleanup_udev_device_unref_ struct udev_device *q;

const char *stype, *node;

unsigned long long flags;

sd_id128_t type_id;

blkid_partition pp;

dev_t qn;

int nr;

errno = 0;

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

if (!q) {

if (!errno)

errno = ENOMEM;

log_error_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);

if (flags & GPT_FLAG_NO_AUTO)

continue;

nr = blkid_partition_get_partno(pp);

if (nr < 0)

continue;

stype = blkid_partition_get_type_string(pp);

if (!stype)

continue;

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

continue;

if (sd_id128_equal(type_id, GPT_HOME)) {

if (home && nr >= home_nr)

continue;

home_nr = nr;

home_rw = !(flags & GPT_FLAG_READ_ONLY);

free(home);

home = strdup(node);

if (!home)

return log_oom();

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

if (srv && nr >= srv_nr)

continue;

srv_nr = nr;

srv_rw = !(flags & GPT_FLAG_READ_ONLY);

free(srv);

srv = strdup(node);

if (!srv)

return log_oom();

}

#ifdef GPT_ROOT_NATIVE

else if (sd_id128_equal(type_id, GPT_ROOT_NATIVE)) {

if (root && nr >= root_nr)

continue;

root_nr = nr;

root_rw = !(flags & GPT_FLAG_READ_ONLY);