Skip to content
Permalink
master
Go to file
 
 
Cannot retrieve contributors at this time
5557 lines (4520 sloc) 235 KB
/* SPDX-License-Identifier: LGPL-2.1+ */
#if HAVE_BLKID
#endif
#include <errno.h>
#include <getopt.h>
#include <linux/fs.h>
#include <linux/loop.h>
#if HAVE_SELINUX
#include <selinux/selinux.h>
#endif
#include <stdlib.h>
#include <sys/file.h>
#include <sys/ioctl.h>
#include <sys/personality.h>
#include <sys/prctl.h>
#include <sys/types.h>
#include <sys/wait.h>
#include <termios.h>
#include <unistd.h>
#include "sd-bus.h"
#include "sd-daemon.h"
#include "sd-id128.h"
#include "alloc-util.h"
#include "barrier.h"
#include "base-filesystem.h"
#include "blkid-util.h"
#include "btrfs-util.h"
#include "bus-error.h"
#include "bus-util.h"
#include "cap-list.h"
#include "capability-util.h"
#include "cgroup-util.h"
#include "copy.h"
#include "cpu-set-util.h"
#include "dev-setup.h"
#include "dissect-image.h"
#include "env-util.h"
#include "escape.h"
#include "fd-util.h"
#include "fdset.h"
#include "fileio.h"
#include "format-util.h"
#include "fs-util.h"
#include "gpt.h"
#include "hexdecoct.h"
#include "hostname-util.h"
#include "id128-util.h"
#include "io-util.h"
#include "log.h"
#include "loop-util.h"
#include "loopback-setup.h"
#include "machine-image.h"
#include "macro.h"
#include "main-func.h"
#include "missing_sched.h"
#include "mkdir.h"
#include "mount-util.h"
#include "mountpoint-util.h"
#include "namespace-util.h"
#include "netlink-util.h"
#include "nspawn-cgroup.h"
#include "nspawn-creds.h"
#include "nspawn-def.h"
#include "nspawn-expose-ports.h"
#include "nspawn-mount.h"
#include "nspawn-network.h"
#include "nspawn-oci.h"
#include "nspawn-patch-uid.h"
#include "nspawn-register.h"
#include "nspawn-seccomp.h"
#include "nspawn-settings.h"
#include "nspawn-setuid.h"
#include "nspawn-stub-pid1.h"
#include "nulstr-util.h"
#include "os-util.h"
#include "pager.h"
#include "parse-util.h"
#include "path-util.h"
#include "pretty-print.h"
#include "process-util.h"
#include "ptyfwd.h"
#include "random-util.h"
#include "raw-clone.h"
#include "resolve-util.h"
#include "rlimit-util.h"
#include "rm-rf.h"
#if HAVE_SECCOMP
#include "seccomp-util.h"
#endif
#include "selinux-util.h"
#include "signal-util.h"
#include "socket-util.h"
#include "stat-util.h"
#include "stdio-util.h"
#include "string-table.h"
#include "string-util.h"
#include "strv.h"
#include "sysctl-util.h"
#include "terminal-util.h"
#include "tmpfile-util.h"
#include "umask-util.h"
#include "unit-name.h"
#include "user-util.h"
#include "util.h"
/* The notify socket inside the container it can use to talk to nspawn using the sd_notify(3) protocol */
#define NSPAWN_NOTIFY_SOCKET_PATH "/run/host/notify"
#define EXIT_FORCE_RESTART 133
typedef enum ContainerStatus {
CONTAINER_TERMINATED,
CONTAINER_REBOOTED,
} ContainerStatus;
static char *arg_directory = NULL;
static char *arg_template = NULL;
static char *arg_chdir = NULL;
static char *arg_pivot_root_new = NULL;
static char *arg_pivot_root_old = NULL;
static char *arg_user = NULL;
static uid_t arg_uid = UID_INVALID;
static gid_t arg_gid = GID_INVALID;
static gid_t* arg_supplementary_gids = NULL;
static size_t arg_n_supplementary_gids = 0;
static sd_id128_t arg_uuid = {};
static char *arg_machine = NULL; /* The name used by the host to refer to this */
static char *arg_hostname = NULL; /* The name the payload sees by default */
static const char *arg_selinux_context = NULL;
static const char *arg_selinux_apifs_context = NULL;
static char *arg_slice = NULL;
static bool arg_private_network = false;
static bool arg_read_only = false;
static StartMode arg_start_mode = START_PID1;
static bool arg_ephemeral = false;
static LinkJournal arg_link_journal = LINK_AUTO;
static bool arg_link_journal_try = false;
static uint64_t arg_caps_retain =
(1ULL << CAP_AUDIT_CONTROL) |
(1ULL << CAP_AUDIT_WRITE) |
(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_MKNOD) |
(1ULL << CAP_NET_BIND_SERVICE) |
(1ULL << CAP_NET_BROADCAST) |
(1ULL << CAP_NET_RAW) |
(1ULL << CAP_SETFCAP) |
(1ULL << CAP_SETGID) |
(1ULL << CAP_SETPCAP) |
(1ULL << CAP_SETUID) |
(1ULL << CAP_SYS_ADMIN) |
(1ULL << CAP_SYS_BOOT) |
(1ULL << CAP_SYS_CHROOT) |
(1ULL << CAP_SYS_NICE) |
(1ULL << CAP_SYS_PTRACE) |
(1ULL << CAP_SYS_RESOURCE) |
(1ULL << CAP_SYS_TTY_CONFIG);
static CapabilityQuintet arg_full_capabilities = CAPABILITY_QUINTET_NULL;
static CustomMount *arg_custom_mounts = NULL;
static size_t arg_n_custom_mounts = 0;
static char **arg_setenv = NULL;
static bool arg_quiet = 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_veth_extra = NULL;
static char *arg_network_bridge = NULL;
static char *arg_network_zone = NULL;
static char *arg_network_namespace_path = NULL;
static PagerFlags arg_pager_flags = 0;
static unsigned long arg_personality = PERSONALITY_INVALID;
static char *arg_image = NULL;
static char *arg_oci_bundle = NULL;
static VolatileMode arg_volatile_mode = VOLATILE_NO;
static ExposePort *arg_expose_ports = NULL;
static char **arg_property = NULL;
static sd_bus_message *arg_property_message = NULL;
static UserNamespaceMode arg_userns_mode = USER_NAMESPACE_NO;
static uid_t arg_uid_shift = UID_INVALID, arg_uid_range = 0x10000U;
static bool arg_userns_chown = false;
static int arg_kill_signal = 0;
static CGroupUnified arg_unified_cgroup_hierarchy = CGROUP_UNIFIED_UNKNOWN;
static SettingsMask arg_settings_mask = 0;
static int arg_settings_trusted = -1;
static char **arg_parameters = NULL;
static const char *arg_container_service_name = "systemd-nspawn";
static bool arg_notify_ready = false;
static bool arg_use_cgns = true;
static unsigned long arg_clone_ns_flags = CLONE_NEWIPC|CLONE_NEWPID|CLONE_NEWUTS;
static MountSettingsMask arg_mount_settings = MOUNT_APPLY_APIVFS_RO|MOUNT_APPLY_TMPFS_TMP;
static VeritySettings arg_verity_settings = VERITY_SETTINGS_DEFAULT;
static char **arg_syscall_allow_list = NULL;
static char **arg_syscall_deny_list = NULL;
#if HAVE_SECCOMP
static scmp_filter_ctx arg_seccomp = NULL;
#endif
static struct rlimit *arg_rlimit[_RLIMIT_MAX] = {};
static bool arg_no_new_privileges = false;
static int arg_oom_score_adjust = 0;
static bool arg_oom_score_adjust_set = false;
static CPUSet arg_cpu_set = {};
static ResolvConfMode arg_resolv_conf = RESOLV_CONF_AUTO;
static TimezoneMode arg_timezone = TIMEZONE_AUTO;
static unsigned arg_console_width = (unsigned) -1, arg_console_height = (unsigned) -1;
static DeviceNode* arg_extra_nodes = NULL;
static size_t arg_n_extra_nodes = 0;
static char **arg_sysctl = NULL;
static ConsoleMode arg_console_mode = _CONSOLE_MODE_INVALID;
static Credential *arg_credentials = NULL;
static size_t arg_n_credentials = 0;
STATIC_DESTRUCTOR_REGISTER(arg_directory, freep);
STATIC_DESTRUCTOR_REGISTER(arg_template, freep);
STATIC_DESTRUCTOR_REGISTER(arg_chdir, freep);
STATIC_DESTRUCTOR_REGISTER(arg_pivot_root_new, freep);
STATIC_DESTRUCTOR_REGISTER(arg_pivot_root_old, freep);
STATIC_DESTRUCTOR_REGISTER(arg_user, freep);
STATIC_DESTRUCTOR_REGISTER(arg_supplementary_gids, freep);
STATIC_DESTRUCTOR_REGISTER(arg_machine, freep);
STATIC_DESTRUCTOR_REGISTER(arg_hostname, freep);
STATIC_DESTRUCTOR_REGISTER(arg_slice, freep);
STATIC_DESTRUCTOR_REGISTER(arg_setenv, strv_freep);
STATIC_DESTRUCTOR_REGISTER(arg_network_interfaces, strv_freep);
STATIC_DESTRUCTOR_REGISTER(arg_network_macvlan, strv_freep);
STATIC_DESTRUCTOR_REGISTER(arg_network_ipvlan, strv_freep);
STATIC_DESTRUCTOR_REGISTER(arg_network_veth_extra, strv_freep);
STATIC_DESTRUCTOR_REGISTER(arg_network_bridge, freep);
STATIC_DESTRUCTOR_REGISTER(arg_network_zone, freep);
STATIC_DESTRUCTOR_REGISTER(arg_network_namespace_path, freep);
STATIC_DESTRUCTOR_REGISTER(arg_image, freep);
STATIC_DESTRUCTOR_REGISTER(arg_oci_bundle, freep);
STATIC_DESTRUCTOR_REGISTER(arg_property, strv_freep);
STATIC_DESTRUCTOR_REGISTER(arg_property_message, sd_bus_message_unrefp);
STATIC_DESTRUCTOR_REGISTER(arg_parameters, strv_freep);
STATIC_DESTRUCTOR_REGISTER(arg_verity_settings, verity_settings_done);
STATIC_DESTRUCTOR_REGISTER(arg_syscall_allow_list, strv_freep);
STATIC_DESTRUCTOR_REGISTER(arg_syscall_deny_list, strv_freep);
#if HAVE_SECCOMP
STATIC_DESTRUCTOR_REGISTER(arg_seccomp, seccomp_releasep);
#endif
STATIC_DESTRUCTOR_REGISTER(arg_cpu_set, cpu_set_reset);
STATIC_DESTRUCTOR_REGISTER(arg_sysctl, strv_freep);
static int handle_arg_console(const char *arg) {
if (streq(arg, "help")) {
puts("autopipe\n"
"interactive\n"
"passive\n"
"pipe\n"
"read-only");
return 0;
}
if (streq(arg, "interactive"))
arg_console_mode = CONSOLE_INTERACTIVE;
else if (streq(arg, "read-only"))
arg_console_mode = CONSOLE_READ_ONLY;
else if (streq(arg, "passive"))
arg_console_mode = CONSOLE_PASSIVE;
else if (streq(arg, "pipe")) {
if (isatty(STDIN_FILENO) > 0 && isatty(STDOUT_FILENO) > 0)
log_full(arg_quiet ? LOG_DEBUG : LOG_NOTICE,
"Console mode 'pipe' selected, but standard input/output are connected to an interactive TTY. "
"Most likely you want to use 'interactive' console mode for proper interactivity and shell job control. "
"Proceeding anyway.");
arg_console_mode = CONSOLE_PIPE;
} else if (streq(arg, "autopipe")) {
if (isatty(STDIN_FILENO) > 0 && isatty(STDOUT_FILENO) > 0)
arg_console_mode = CONSOLE_INTERACTIVE;
else
arg_console_mode = CONSOLE_PIPE;
} else
return log_error_errno(SYNTHETIC_ERRNO(EINVAL), "Unknown console mode: %s", optarg);
arg_settings_mask |= SETTING_CONSOLE_MODE;
return 1;
}
static int help(void) {
_cleanup_free_ char *link = NULL;
int r;
(void) pager_open(arg_pager_flags);
r = terminal_urlify_man("systemd-nspawn", "1", &link);
if (r < 0)
return log_oom();
printf("%1$s [OPTIONS...] [PATH] [ARGUMENTS...]\n\n"
"%5$sSpawn a command or OS in a light-weight container.%6$s\n\n"
" -h --help Show this help\n"
" --version Print version string\n"
" -q --quiet Do not show status information\n"
" --no-pager Do not pipe output into a pager\n"
" --settings=BOOLEAN Load additional settings from .nspawn file\n\n"
"%3$sImage:%4$s\n"
" -D --directory=PATH Root directory for the container\n"
" --template=PATH Initialize root directory from template directory,\n"
" if missing\n"
" -x --ephemeral Run container with snapshot of root directory, and\n"
" remove it after exit\n"
" -i --image=PATH Root file system disk image (or device node) for\n"
" the container\n"
" --oci-bundle=PATH OCI bundle directory\n"
" --read-only Mount the root directory read-only\n"
" --volatile[=MODE] Run the system in volatile mode\n"
" --root-hash=HASH Specify verity root hash for root disk image\n"
" --root-hash-sig=SIG Specify pkcs7 signature of root hash for verity\n"
" as a DER encoded PKCS7, either as a path to a file\n"
" or as an ASCII base64 encoded string prefixed by\n"
" 'base64:'\n"
" --verity-data=PATH Specify hash device for verity\n"
" --pivot-root=PATH[:PATH]\n"
" Pivot root to given directory in the container\n\n"
"%3$sExecution:%4$s\n"
" -a --as-pid2 Maintain a stub init as PID1, invoke binary as PID2\n"
" -b --boot Boot up full system (i.e. invoke init)\n"
" --chdir=PATH Set working directory in the container\n"
" -E --setenv=NAME=VALUE Pass an environment variable to PID 1\n"
" -u --user=USER Run the command under specified user or UID\n"
" --kill-signal=SIGNAL Select signal to use for shutting down PID 1\n"
" --notify-ready=BOOLEAN Receive notifications from the child init process\n\n"
"%3$sSystem Identity:%4$s\n"
" -M --machine=NAME Set the machine name for the container\n"
" --hostname=NAME Override the hostname for the container\n"
" --uuid=UUID Set a specific machine UUID for the container\n\n"
"%3$sProperties:%4$s\n"
" -S --slice=SLICE Place the container in the specified slice\n"
" --property=NAME=VALUE Set scope unit property\n"
" --register=BOOLEAN Register container as machine\n"
" --keep-unit Do not register a scope for the machine, reuse\n"
" the service unit nspawn is running in\n\n"
"%3$sUser Namespacing:%4$s\n"
" -U --private-users=pick Run within user namespace, autoselect UID/GID range\n"
" --private-users[=UIDBASE[:NUIDS]]\n"
" Similar, but with user configured UID/GID range\n"
" --private-users-chown Adjust OS tree ownership to private UID/GID range\n\n"
"%3$sNetworking:%4$s\n"
" --private-network Disable network in container\n"
" --network-interface=INTERFACE\n"
" Assign an existing network interface to the\n"
" container\n"
" --network-macvlan=INTERFACE\n"
" Create a macvlan network interface based on an\n"
" existing network interface to the container\n"
" --network-ipvlan=INTERFACE\n"
" Create a ipvlan network interface based on an\n"
" existing network interface to the container\n"
" -n --network-veth Add a virtual Ethernet connection between host\n"
" and container\n"
" --network-veth-extra=HOSTIF[:CONTAINERIF]\n"
" Add an additional virtual Ethernet link between\n"
" host and container\n"
" --network-bridge=INTERFACE\n"
" Add a virtual Ethernet connection to the container\n"
" and attach it to an existing bridge on the host\n"
" --network-zone=NAME Similar, but attach the new interface to an\n"
" an automatically managed bridge interface\n"
" --network-namespace-path=PATH\n"
" Set network namespace to the one represented by\n"
" the specified kernel namespace file node\n"
" -p --port=[PROTOCOL:]HOSTPORT[:CONTAINERPORT]\n"
" Expose a container IP port on the host\n\n"
"%3$sSecurity:%4$s\n"
" --capability=CAP In addition to the default, retain specified\n"
" capability\n"
" --drop-capability=CAP Drop the specified capability from the default set\n"
" --no-new-privileges Set PR_SET_NO_NEW_PRIVS flag for container payload\n"
" --system-call-filter=LIST|~LIST\n"
" Permit/prohibit specific system calls\n"
" -Z --selinux-context=SECLABEL\n"
" Set the SELinux security context to be used by\n"
" processes in the container\n"
" -L --selinux-apifs-context=SECLABEL\n"
" Set the SELinux security context to be used by\n"
" API/tmpfs file systems in the container\n\n"
"%3$sResources:%4$s\n"
" --rlimit=NAME=LIMIT Set a resource limit for the payload\n"
" --oom-score-adjust=VALUE\n"
" Adjust the OOM score value for the payload\n"
" --cpu-affinity=CPUS Adjust the CPU affinity of the container\n"
" --personality=ARCH Pick personality for this container\n\n"
"%3$sIntegration:%4$s\n"
" --resolv-conf=MODE Select mode of /etc/resolv.conf initialization\n"
" --timezone=MODE Select mode of /etc/localtime initialization\n"
" --link-journal=MODE Link up guest journal, one of no, auto, guest, \n"
" host, try-guest, try-host\n"
" -j Equivalent to --link-journal=try-guest\n\n"
"%3$sMounts:%4$s\n"
" --bind=PATH[:PATH[:OPTIONS]]\n"
" Bind mount a file or directory from the host into\n"
" the container\n"
" --bind-ro=PATH[:PATH[:OPTIONS]\n"
" Similar, but creates a read-only bind mount\n"
" --inaccessible=PATH Over-mount file node with inaccessible node to mask\n"
" it\n"
" --tmpfs=PATH:[OPTIONS] Mount an empty tmpfs to the specified directory\n"
" --overlay=PATH[:PATH...]:PATH\n"
" Create an overlay mount from the host to \n"
" the container\n"
" --overlay-ro=PATH[:PATH...]:PATH\n"
" Similar, but creates a read-only overlay mount\n\n"
"%3$sInput/Output:%4$s\n"
" --console=MODE Select how stdin/stdout/stderr and /dev/console are\n"
" set up for the container.\n"
" -P --pipe Equivalent to --console=pipe\n\n"
"%3$sCredentials:%4$s\n"
" --set-credential=ID:VALUE\n"
" Pass a credential with literal value to container.\n"
" --load-credential=ID:PATH\n"
" Load credential to pass to container from file or\n"
" AF_UNIX stream socket.\n"
"\nSee the %2$s for details.\n"
, program_invocation_short_name
, link
, ansi_underline(), ansi_normal()
, ansi_highlight(), ansi_normal()
);
return 0;
}
static int custom_mount_check_all(void) {
size_t i;
for (i = 0; i < arg_n_custom_mounts; i++) {
CustomMount *m = &arg_custom_mounts[i];
if (path_equal(m->destination, "/") && arg_userns_mode != USER_NAMESPACE_NO) {
if (arg_userns_chown)
return log_error_errno(SYNTHETIC_ERRNO(EINVAL),
"--private-users-chown may not be combined with custom root mounts.");
else if (arg_uid_shift == UID_INVALID)
return log_error_errno(SYNTHETIC_ERRNO(EINVAL),
"--private-users with automatic UID shift may not be combined with custom root mounts.");
}
}
return 0;
}
static int detect_unified_cgroup_hierarchy_from_environment(void) {
const char *e, *var = "SYSTEMD_NSPAWN_UNIFIED_HIERARCHY";
int r;
/* Allow the user to control whether the unified hierarchy is used */
e = getenv(var);
if (!e) {
/* $UNIFIED_CGROUP_HIERARCHY has been renamed to $SYSTEMD_NSPAWN_UNIFIED_HIERARCHY. */
var = "UNIFIED_CGROUP_HIERARCHY";
e = getenv(var);
}
if (!isempty(e)) {
r = parse_boolean(e);
if (r < 0)
return log_error_errno(r, "Failed to parse $%s: %m", var);
if (r > 0)
arg_unified_cgroup_hierarchy = CGROUP_UNIFIED_ALL;
else
arg_unified_cgroup_hierarchy = CGROUP_UNIFIED_NONE;
}
return 0;
}
static int detect_unified_cgroup_hierarchy_from_image(const char *directory) {
int r;
/* Let's inherit the mode to use from the host system, but let's take into consideration what systemd
* in the image actually supports. */
r = cg_all_unified();
if (r < 0)
return log_error_errno(r, "Failed to determine whether we are in all unified mode.");
if (r > 0) {
/* Unified cgroup hierarchy support was added in 230. Unfortunately the detection
* routine only detects 231, so we'll have a false negative here for 230. */
r = systemd_installation_has_version(directory, 230);
if (r < 0)
return log_error_errno(r, "Failed to determine systemd version in container: %m");
if (r > 0)
arg_unified_cgroup_hierarchy = CGROUP_UNIFIED_ALL;
else
arg_unified_cgroup_hierarchy = CGROUP_UNIFIED_NONE;
} else if (cg_unified_controller(SYSTEMD_CGROUP_CONTROLLER) > 0) {
/* Mixed cgroup hierarchy support was added in 233 */
r = systemd_installation_has_version(directory, 233);
if (r < 0)
return log_error_errno(r, "Failed to determine systemd version in container: %m");
if (r > 0)
arg_unified_cgroup_hierarchy = CGROUP_UNIFIED_SYSTEMD;
else
arg_unified_cgroup_hierarchy = CGROUP_UNIFIED_NONE;
} else
arg_unified_cgroup_hierarchy = CGROUP_UNIFIED_NONE;
log_debug("Using %s hierarchy for container.",
arg_unified_cgroup_hierarchy == CGROUP_UNIFIED_NONE ? "legacy" :
arg_unified_cgroup_hierarchy == CGROUP_UNIFIED_SYSTEMD ? "hybrid" : "unified");
return 0;
}
static int parse_capability_spec(const char *spec, uint64_t *ret_mask) {
uint64_t mask = 0;
int r;
for (;;) {
_cleanup_free_ char *t = NULL;
r = extract_first_word(&spec, &t, ",", 0);
if (r < 0)
return log_error_errno(r, "Failed to parse capability %s.", t);
if (r == 0)
break;
if (streq(t, "help")) {
for (int i = 0; i < capability_list_length(); i++) {
const char *name;
name = capability_to_name(i);
if (name)
puts(name);
}
return 0; /* quit */
}
if (streq(t, "all"))
mask = (uint64_t) -1;
else {
r = capability_from_name(t);
if (r < 0)
return log_error_errno(r, "Failed to parse capability %s.", t);
mask |= 1ULL << r;
}
}
*ret_mask = mask;
return 1; /* continue */
}
static int parse_share_ns_env(const char *name, unsigned long ns_flag) {
int r;
r = getenv_bool(name);
if (r == -ENXIO)
return 0;
if (r < 0)
return log_error_errno(r, "Failed to parse $%s: %m", name);
arg_clone_ns_flags = (arg_clone_ns_flags & ~ns_flag) | (r > 0 ? 0 : ns_flag);
arg_settings_mask |= SETTING_CLONE_NS_FLAGS;
return 0;
}
static int parse_mount_settings_env(void) {
const char *e;
int r;
r = getenv_bool("SYSTEMD_NSPAWN_TMPFS_TMP");
if (r < 0 && r != -ENXIO)
return log_error_errno(r, "Failed to parse $SYSTEMD_NSPAWN_TMPFS_TMP: %m");
if (r >= 0)
SET_FLAG(arg_mount_settings, MOUNT_APPLY_TMPFS_TMP, r > 0);
e = getenv("SYSTEMD_NSPAWN_API_VFS_WRITABLE");
if (streq_ptr(e, "network"))
arg_mount_settings |= MOUNT_APPLY_APIVFS_RO|MOUNT_APPLY_APIVFS_NETNS;
else if (e) {
r = parse_boolean(e);
if (r < 0)
return log_error_errno(r, "Failed to parse $SYSTEMD_NSPAWN_API_VFS_WRITABLE: %m");
SET_FLAG(arg_mount_settings, MOUNT_APPLY_APIVFS_RO, r == 0);
SET_FLAG(arg_mount_settings, MOUNT_APPLY_APIVFS_NETNS, false);
}
return 0;
}
static int parse_environment(void) {
const char *e;
int r;
r = parse_share_ns_env("SYSTEMD_NSPAWN_SHARE_NS_IPC", CLONE_NEWIPC);
if (r < 0)
return r;
r = parse_share_ns_env("SYSTEMD_NSPAWN_SHARE_NS_PID", CLONE_NEWPID);
if (r < 0)
return r;
r = parse_share_ns_env("SYSTEMD_NSPAWN_SHARE_NS_UTS", CLONE_NEWUTS);
if (r < 0)
return r;
r = parse_share_ns_env("SYSTEMD_NSPAWN_SHARE_SYSTEM", CLONE_NEWIPC|CLONE_NEWPID|CLONE_NEWUTS);
if (r < 0)
return r;
r = parse_mount_settings_env();
if (r < 0)
return r;
/* SYSTEMD_NSPAWN_USE_CGNS=0 can be used to disable CLONE_NEWCGROUP use,
* even if it is supported. If not supported, it has no effect. */
if (!cg_ns_supported())
arg_use_cgns = false;
else {
r = getenv_bool("SYSTEMD_NSPAWN_USE_CGNS");
if (r < 0) {
if (r != -ENXIO)
return log_error_errno(r, "Failed to parse $SYSTEMD_NSPAWN_USE_CGNS: %m");
arg_use_cgns = true;
} else {
arg_use_cgns = r > 0;
arg_settings_mask |= SETTING_USE_CGNS;
}
}
e = getenv("SYSTEMD_NSPAWN_CONTAINER_SERVICE");
if (e)
arg_container_service_name = e;
return detect_unified_cgroup_hierarchy_from_environment();
}
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_INACCESSIBLE,
ARG_SHARE_SYSTEM,
ARG_REGISTER,
ARG_KEEP_UNIT,
ARG_NETWORK_INTERFACE,
ARG_NETWORK_MACVLAN,
ARG_NETWORK_IPVLAN,
ARG_NETWORK_BRIDGE,
ARG_NETWORK_ZONE,
ARG_NETWORK_VETH_EXTRA,
ARG_NETWORK_NAMESPACE_PATH,
ARG_PERSONALITY,
ARG_VOLATILE,
ARG_TEMPLATE,
ARG_PROPERTY,
ARG_PRIVATE_USERS,
ARG_KILL_SIGNAL,
ARG_SETTINGS,
ARG_CHDIR,
ARG_PIVOT_ROOT,
ARG_PRIVATE_USERS_CHOWN,
ARG_NOTIFY_READY,
ARG_ROOT_HASH,
ARG_ROOT_HASH_SIG,
ARG_VERITY_DATA,
ARG_SYSTEM_CALL_FILTER,
ARG_RLIMIT,
ARG_HOSTNAME,
ARG_NO_NEW_PRIVILEGES,
ARG_OOM_SCORE_ADJUST,
ARG_CPU_AFFINITY,
ARG_RESOLV_CONF,
ARG_TIMEZONE,
ARG_CONSOLE,
ARG_PIPE,
ARG_OCI_BUNDLE,
ARG_NO_PAGER,
ARG_SET_CREDENTIAL,
ARG_LOAD_CREDENTIAL,
};
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 },
{ "as-pid2", no_argument, NULL, 'a' },
{ "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 },
{ "no-new-privileges", required_argument, NULL, ARG_NO_NEW_PRIVILEGES },
{ "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 },
{ "inaccessible", required_argument, NULL, ARG_INACCESSIBLE },
{ "machine", required_argument, NULL, 'M' },
{ "hostname", required_argument, NULL, ARG_HOSTNAME },
{ "slice", required_argument, NULL, 'S' },
{ "setenv", required_argument, NULL, 'E' },
{ "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 }, /* not documented */
{ "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-veth-extra", required_argument, NULL, ARG_NETWORK_VETH_EXTRA },
{ "network-bridge", required_argument, NULL, ARG_NETWORK_BRIDGE },
{ "network-zone", required_argument, NULL, ARG_NETWORK_ZONE },
{ "network-namespace-path", required_argument, NULL, ARG_NETWORK_NAMESPACE_PATH },
{ "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 },
{ "private-users-chown", optional_argument, NULL, ARG_PRIVATE_USERS_CHOWN },
{ "kill-signal", required_argument, NULL, ARG_KILL_SIGNAL },
{ "settings", required_argument, NULL, ARG_SETTINGS },
{ "chdir", required_argument, NULL, ARG_CHDIR },
{ "pivot-root", required_argument, NULL, ARG_PIVOT_ROOT },
{ "notify-ready", required_argument, NULL, ARG_NOTIFY_READY },
{ "root-hash", required_argument, NULL, ARG_ROOT_HASH },
{ "root-hash-sig", required_argument, NULL, ARG_ROOT_HASH_SIG },
{ "verity-data", required_argument, NULL, ARG_VERITY_DATA },
{ "system-call-filter", required_argument, NULL, ARG_SYSTEM_CALL_FILTER },
{ "rlimit", required_argument, NULL, ARG_RLIMIT },
{ "oom-score-adjust", required_argument, NULL, ARG_OOM_SCORE_ADJUST },
{ "cpu-affinity", required_argument, NULL, ARG_CPU_AFFINITY },
{ "resolv-conf", required_argument, NULL, ARG_RESOLV_CONF },
{ "timezone", required_argument, NULL, ARG_TIMEZONE },
{ "console", required_argument, NULL, ARG_CONSOLE },
{ "pipe", no_argument, NULL, ARG_PIPE },
{ "oci-bundle", required_argument, NULL, ARG_OCI_BUNDLE },
{ "no-pager", no_argument, NULL, ARG_NO_PAGER },
{ "set-credential", required_argument, NULL, ARG_SET_CREDENTIAL },
{ "load-credential", required_argument, NULL, ARG_LOAD_CREDENTIAL },
{}
};
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:abL:M:jS:Z:qi:xp:nUE:P", options, NULL)) >= 0)
switch (c) {
case 'h':
return help();
case ARG_VERSION:
return version();
case 'D':
r = parse_path_argument_and_warn(optarg, false, &arg_directory);
if (r < 0)
return r;
arg_settings_mask |= SETTING_DIRECTORY;
break;
case ARG_TEMPLATE:
r = parse_path_argument_and_warn(optarg, false, &arg_template);
if (r < 0)
return r;
arg_settings_mask |= SETTING_DIRECTORY;
break;
case 'i':
r = parse_path_argument_and_warn(optarg, false, &arg_image);
if (r < 0)
return r;
arg_settings_mask |= SETTING_DIRECTORY;
break;
case ARG_OCI_BUNDLE:
r = parse_path_argument_and_warn(optarg, false, &arg_oci_bundle);
if (r < 0)
return r;
break;
case 'x':
arg_ephemeral = true;
arg_settings_mask |= SETTING_EPHEMERAL;
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_ZONE: {
char *j;
j = strjoin("vz-", optarg);
if (!j)
return log_oom();
if (!ifname_valid(j)) {
log_error("Network zone name not valid: %s", j);
free(j);
return -EINVAL;
}
free_and_replace(arg_network_zone, j);
arg_network_veth = true;
arg_private_network = true;
arg_settings_mask |= SETTING_NETWORK;
break;
}
case ARG_NETWORK_BRIDGE:
if (!ifname_valid(optarg))
return log_error_errno(SYNTHETIC_ERRNO(EINVAL),
"Bridge interface name not valid: %s", optarg);
r = free_and_strdup(&arg_network_bridge, optarg);
if (r < 0)
return log_oom();
_fallthrough_;
case 'n':
arg_network_veth = true;
arg_private_network = true;
arg_settings_mask |= SETTING_NETWORK;
break;
case ARG_NETWORK_VETH_EXTRA:
r = veth_extra_parse(&arg_network_veth_extra, optarg);
if (r < 0)
return log_error_errno(r, "Failed to parse --network-veth-extra= parameter: %s", optarg);
arg_private_network = true;
arg_settings_mask |= SETTING_NETWORK;
break;
case ARG_NETWORK_INTERFACE:
if (!ifname_valid(optarg))
return log_error_errno(SYNTHETIC_ERRNO(EINVAL),
"Network interface name not valid: %s", optarg);
r = test_network_interface_initialized(optarg);
if (r < 0)
return r;
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 (!ifname_valid(optarg))
return log_error_errno(SYNTHETIC_ERRNO(EINVAL),
"MACVLAN network interface name not valid: %s", optarg);
r = test_network_interface_initialized(optarg);
if (r < 0)
return r;
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 (!ifname_valid(optarg))
return log_error_errno(SYNTHETIC_ERRNO(EINVAL),
"IPVLAN network interface name not valid: %s", optarg);
r = test_network_interface_initialized(optarg);
if (r < 0)
return r;
if (strv_extend(&arg_network_ipvlan, optarg) < 0)
return log_oom();
_fallthrough_;
case ARG_PRIVATE_NETWORK:
arg_private_network = true;
arg_settings_mask |= SETTING_NETWORK;
break;
case ARG_NETWORK_NAMESPACE_PATH:
r = parse_path_argument_and_warn(optarg, false, &arg_network_namespace_path);
if (r < 0)
return r;
arg_settings_mask |= SETTING_NETWORK;
break;
case 'b':
if (arg_start_mode == START_PID2)
return log_error_errno(SYNTHETIC_ERRNO(EINVAL),
"--boot and --as-pid2 may not be combined.");
arg_start_mode = START_BOOT;
arg_settings_mask |= SETTING_START_MODE;
break;
case 'a':
if (arg_start_mode == START_BOOT)
return log_error_errno(SYNTHETIC_ERRNO(EINVAL),
"--boot and --as-pid2 may not be combined.");
arg_start_mode = START_PID2;
arg_settings_mask |= SETTING_START_MODE;
break;
case ARG_UUID:
r = sd_id128_from_string(optarg, &arg_uuid);
if (r < 0)
return log_error_errno(r, "Invalid UUID: %s", optarg);
if (sd_id128_is_null(arg_uuid))
return log_error_errno(SYNTHETIC_ERRNO(EINVAL),
"Machine UUID may not be all zeroes.");
arg_settings_mask |= SETTING_MACHINE_ID;
break;
case 'S': {
_cleanup_free_ char *mangled = NULL;
r = unit_name_mangle_with_suffix(optarg, NULL, UNIT_NAME_MANGLE_WARN, ".slice", &mangled);
if (r < 0)
return log_oom();
free_and_replace(arg_slice, mangled);
arg_settings_mask |= SETTING_SLICE;
break;
}
case 'M':
if (isempty(optarg))
arg_machine = mfree(arg_machine);
else {
if (!machine_name_is_valid(optarg))
return log_error_errno(SYNTHETIC_ERRNO(EINVAL),
"Invalid machine name: %s", optarg);
r = free_and_strdup(&arg_machine, optarg);
if (r < 0)
return log_oom();
}
break;
case ARG_HOSTNAME:
if (isempty(optarg))
arg_hostname = mfree(arg_hostname);
else {
if (!hostname_is_valid(optarg, false))
return log_error_errno(SYNTHETIC_ERRNO(EINVAL),
"Invalid hostname: %s", optarg);
r = free_and_strdup(&arg_hostname, optarg);
if (r < 0)
return log_oom();
}
arg_settings_mask |= SETTING_HOSTNAME;
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: {
uint64_t m;
r = parse_capability_spec(optarg, &m);
if (r <= 0)
return r;
if (c == ARG_CAPABILITY)
plus |= m;
else
minus |= m;
arg_settings_mask |= SETTING_CAPABILITY;
break;
}
case ARG_NO_NEW_PRIVILEGES:
r = parse_boolean(optarg);
if (r < 0)
return log_error_errno(r, "Failed to parse --no-new-privileges= argument: %s", optarg);
arg_no_new_privileges = r;
arg_settings_mask |= SETTING_NO_NEW_PRIVILEGES;
break;
case 'j':
arg_link_journal = LINK_GUEST;
arg_link_journal_try = true;
arg_settings_mask |= SETTING_LINK_JOURNAL;
break;
case ARG_LINK_JOURNAL:
r = parse_link_journal(optarg, &arg_link_journal, &arg_link_journal_try);
if (r < 0)
return log_error_errno(r, "Failed to parse link journal mode %s", optarg);
arg_settings_mask |= SETTING_LINK_JOURNAL;
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:
r = overlay_mount_parse(&arg_custom_mounts, &arg_n_custom_mounts, optarg, c == ARG_OVERLAY_RO);
if (r == -EADDRNOTAVAIL)
return log_error_errno(r, "--overlay(-ro)= needs at least two colon-separated directories specified.");
if (r < 0)
return log_error_errno(r, "Failed to parse --overlay(-ro)= argument %s: %m", optarg);
arg_settings_mask |= SETTING_CUSTOM_MOUNTS;
break;
case ARG_INACCESSIBLE:
r = inaccessible_mount_parse(&arg_custom_mounts, &arg_n_custom_mounts, optarg);
if (r < 0)
return log_error_errno(r, "Failed to parse --inaccessible= argument %s: %m", optarg);
arg_settings_mask |= SETTING_CUSTOM_MOUNTS;
break;
case 'E': {
char **n;
if (!env_assignment_is_valid(optarg))
return log_error_errno(SYNTHETIC_ERRNO(EINVAL),
"Environment variable assignment '%s' is not valid.", optarg);
n = strv_env_set(arg_setenv, optarg);
if (!n)
return log_oom();
strv_free_and_replace(arg_setenv, n);
arg_settings_mask |= SETTING_ENVIRONMENT;
break;
}
case 'q':
arg_quiet = true;
break;
case ARG_SHARE_SYSTEM:
/* We don't officially support this anymore, except for compat reasons. People should use the
* $SYSTEMD_NSPAWN_SHARE_* environment variables instead. */
log_warning("Please do not use --share-system anymore, use $SYSTEMD_NSPAWN_SHARE_* instead.");
arg_clone_ns_flags = 0;
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)
return log_error_errno(SYNTHETIC_ERRNO(EINVAL),
"Unknown or unsupported personality '%s'.", optarg);
arg_settings_mask |= SETTING_PERSONALITY;
break;
case ARG_VOLATILE:
if (!optarg)
arg_volatile_mode = VOLATILE_YES;
else if (streq(optarg, "help")) {
DUMP_STRING_TABLE(volatile_mode, VolatileMode, _VOLATILE_MODE_MAX);
return 0;
} else {
VolatileMode m;
m = volatile_mode_from_string(optarg);
if (m < 0)
return log_error_errno(SYNTHETIC_ERRNO(EINVAL),
"Failed to parse --volatile= argument: %s", optarg);
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: {
int boolean = -1;
if (!optarg)
boolean = true;
else if (!in_charset(optarg, DIGITS))
/* do *not* parse numbers as booleans */
boolean = parse_boolean(optarg);
if (boolean == false) {
/* no: User namespacing off */
arg_userns_mode = USER_NAMESPACE_NO;
arg_uid_shift = UID_INVALID;
arg_uid_range = UINT32_C(0x10000);
} else if (boolean == true) {
/* yes: User namespacing on, UID range is read from root dir */
arg_userns_mode = USER_NAMESPACE_FIXED;
arg_uid_shift = UID_INVALID;
arg_uid_range = UINT32_C(0x10000);
} else if (streq(optarg, "pick")) {
/* pick: User namespacing on, UID range is picked randomly */
arg_userns_mode = USER_NAMESPACE_PICK;
arg_uid_shift = UID_INVALID;
arg_uid_range = UINT32_C(0x10000);
} else {
_cleanup_free_ char *buffer = NULL;
const char *range, *shift;
/* anything else: User namespacing on, UID range is explicitly configured */
range = strchr(optarg, ':');
if (range) {
buffer = strndup(optarg, range - optarg);
if (!buffer)
return log_oom();
shift = buffer;
range++;
r = safe_atou32(range, &arg_uid_range);
if (r < 0)
return log_error_errno(r, "Failed to parse UID range \"%s\": %m", range);
} else
shift = optarg;
r = parse_uid(shift, &arg_uid_shift);
if (r < 0)
return log_error_errno(r, "Failed to parse UID \"%s\": %m", optarg);
arg_userns_mode = USER_NAMESPACE_FIXED;
}
if (arg_uid_range <= 0)
return log_error_errno(SYNTHETIC_ERRNO(EINVAL),
"UID range cannot be 0.");
arg_settings_mask |= SETTING_USERNS;
break;
}
case 'U':
if (userns_supported()) {
arg_userns_mode = USER_NAMESPACE_PICK;
arg_uid_shift = UID_INVALID;
arg_uid_range = UINT32_C(0x10000);
arg_settings_mask |= SETTING_USERNS;
}
break;
case ARG_PRIVATE_USERS_CHOWN:
arg_userns_chown = true;
arg_settings_mask |= SETTING_USERNS;
break;
case ARG_KILL_SIGNAL:
if (streq(optarg, "help")) {
DUMP_STRING_TABLE(signal, int, _NSIG);
return 0;
}
arg_kill_signal = signal_from_string(optarg);
if (arg_kill_signal < 0)
return log_error_errno(SYNTHETIC_ERRNO(EINVAL),
"Cannot parse signal: %s", optarg);
arg_settings_mask |= SETTING_KILL_SIGNAL;
break;
case ARG_SETTINGS:
/* no → do not read files
* yes → read files, do not override cmdline, trust only subset
* override → read files, override cmdline, trust only subset
* trusted → read files, do not override cmdline, trust all
*/
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 ARG_CHDIR:
if (!path_is_absolute(optarg))
return log_error_errno(SYNTHETIC_ERRNO(EINVAL),
"Working directory %s is not an absolute path.", optarg);
r = free_and_strdup(&arg_chdir, optarg);
if (r < 0)
return log_oom();
arg_settings_mask |= SETTING_WORKING_DIRECTORY;
break;
case ARG_PIVOT_ROOT:
r = pivot_root_parse(&arg_pivot_root_new, &arg_pivot_root_old, optarg);
if (r < 0)
return log_error_errno(r, "Failed to parse --pivot-root= argument %s: %m", optarg);
arg_settings_mask |= SETTING_PIVOT_ROOT;
break;
case ARG_NOTIFY_READY:
r = parse_boolean(optarg);
if (r < 0)
return log_error_errno(SYNTHETIC_ERRNO(EINVAL),
"%s is not a valid notify mode. Valid modes are: yes, no, and ready.", optarg);
arg_notify_ready = r;
arg_settings_mask |= SETTING_NOTIFY_READY;
break;
case ARG_ROOT_HASH: {
_cleanup_free_ void *k = NULL;
size_t l;
r = unhexmem(optarg, strlen(optarg), &k, &l);
if (r < 0)
return log_error_errno(r, "Failed to parse root hash: %s", optarg);
if (l < sizeof(sd_id128_t))
return log_error_errno(SYNTHETIC_ERRNO(EINVAL), "Root hash must be at least 128bit long: %s", optarg);
free_and_replace(arg_verity_settings.root_hash, k);
arg_verity_settings.root_hash_size = l;
break;
}
case ARG_ROOT_HASH_SIG: {
char *value;
size_t l;
void *p;
if ((value = startswith(optarg, "base64:"))) {
r = unbase64mem(value, strlen(value), &p, &l);
if (r < 0)
return log_error_errno(r, "Failed to parse root hash signature '%s': %m", optarg);
} else {
r = read_full_file(optarg, (char**) &p, &l);
if (r < 0)
return log_error_errno(r, "Failed parse root hash signature file '%s': %m", optarg);
}
free_and_replace(arg_verity_settings.root_hash_sig, p);
arg_verity_settings.root_hash_sig_size = l;
break;
}
case ARG_VERITY_DATA:
r = parse_path_argument_and_warn(optarg, false, &arg_verity_settings.data_path);
if (r < 0)
return r;
break;
case ARG_SYSTEM_CALL_FILTER: {
bool negative;
const char *items;
negative = optarg[0] == '~';
items = negative ? optarg + 1 : optarg;
for (;;) {
_cleanup_free_ char *word = NULL;
r = extract_first_word(&items, &word, NULL, 0);
if (r == 0)
break;
if (r == -ENOMEM)
return log_oom();
if (r < 0)
return log_error_errno(r, "Failed to parse system call filter: %m");
if (negative)
r = strv_extend(&arg_syscall_deny_list, word);
else
r = strv_extend(&arg_syscall_allow_list, word);
if (r < 0)
return log_oom();
}
arg_settings_mask |= SETTING_SYSCALL_FILTER;
break;
}
case ARG_RLIMIT: {
const char *eq;
_cleanup_free_ char *name = NULL;
int rl;
if (streq(optarg, "help")) {
DUMP_STRING_TABLE(rlimit, int, _RLIMIT_MAX);
return 0;
}
eq = strchr(optarg, '=');
if (!eq)
return log_error_errno(SYNTHETIC_ERRNO(EINVAL),
"--rlimit= expects an '=' assignment.");
name = strndup(optarg, eq - optarg);
if (!name)
return log_oom();
rl = rlimit_from_string_harder(name);
if (rl < 0)
return log_error_errno(SYNTHETIC_ERRNO(EINVAL),
"Unknown resource limit: %s", name);
if (!arg_rlimit[rl]) {
arg_rlimit[rl] = new0(struct rlimit, 1);
if (!arg_rlimit[rl])
return log_oom();
}
r = rlimit_parse(rl, eq + 1, arg_rlimit[rl]);
if (r < 0)
return log_error_errno(r, "Failed to parse resource limit: %s", eq + 1);
arg_settings_mask |= SETTING_RLIMIT_FIRST << rl;
break;
}
case ARG_OOM_SCORE_ADJUST:
r = parse_oom_score_adjust(optarg, &arg_oom_score_adjust);
if (r < 0)
return log_error_errno(r, "Failed to parse --oom-score-adjust= parameter: %s", optarg);
arg_oom_score_adjust_set = true;
arg_settings_mask |= SETTING_OOM_SCORE_ADJUST;
break;
case ARG_CPU_AFFINITY: {
CPUSet cpuset;
r = parse_cpu_set(optarg, &cpuset);
if (r < 0)
return log_error_errno(r, "Failed to parse CPU affinity mask %s: %m", optarg);
cpu_set_reset(&arg_cpu_set);
arg_cpu_set = cpuset;
arg_settings_mask |= SETTING_CPU_AFFINITY;
break;
}
case ARG_RESOLV_CONF:
if (streq(optarg, "help")) {
DUMP_STRING_TABLE(resolv_conf_mode, ResolvConfMode, _RESOLV_CONF_MODE_MAX);
return 0;
}
arg_resolv_conf = resolv_conf_mode_from_string(optarg);
if (arg_resolv_conf < 0)
return log_error_errno(SYNTHETIC_ERRNO(EINVAL),
"Failed to parse /etc/resolv.conf mode: %s", optarg);
arg_settings_mask |= SETTING_RESOLV_CONF;
break;
case ARG_TIMEZONE:
if (streq(optarg, "help")) {
DUMP_STRING_TABLE(timezone_mode, TimezoneMode, _TIMEZONE_MODE_MAX);
return 0;
}
arg_timezone = timezone_mode_from_string(optarg);
if (arg_timezone < 0)
return log_error_errno(SYNTHETIC_ERRNO(EINVAL),
"Failed to parse /etc/localtime mode: %s", optarg);
arg_settings_mask |= SETTING_TIMEZONE;
break;
case ARG_CONSOLE:
r = handle_arg_console(optarg);
if (r <= 0)
return r;
break;
case 'P':
case ARG_PIPE:
r = handle_arg_console("pipe");
if (r <= 0)
return r;
break;
case ARG_NO_PAGER:
arg_pager_flags |= PAGER_DISABLE;
break;
case ARG_SET_CREDENTIAL: {
_cleanup_free_ char *word = NULL, *data = NULL;
const char *p = optarg;
Credential *a;
size_t i;
int l;
r = extract_first_word(&p, &word, ":", EXTRACT_DONT_COALESCE_SEPARATORS);
if (r == -ENOMEM)
return log_oom();
if (r < 0)
return log_error_errno(r, "Failed to parse --set-credential= parameter: %m");
if (r == 0 || !p)
return log_error_errno(SYNTHETIC_ERRNO(EINVAL), "Missing value for --set-credential=: %s", optarg);
if (!credential_name_valid(word))
return log_error_errno(SYNTHETIC_ERRNO(EINVAL), "Credential name is not valid: %s", word);
for (i = 0; i < arg_n_credentials; i++)
if (streq(arg_credentials[i].id, word))
return log_error_errno(SYNTHETIC_ERRNO(EEXIST), "Duplicate credential '%s', refusing.", word);
l = cunescape(p, UNESCAPE_ACCEPT_NUL, &data);
if (l < 0)
return log_error_errno(l, "Failed to unescape credential data: %s", p);
a = reallocarray(arg_credentials, arg_n_credentials + 1, sizeof(Credential));
if (!a)
return log_oom();
a[arg_n_credentials++] = (Credential) {
.id = TAKE_PTR(word),
.data = TAKE_PTR(data),
.size = l,
};
arg_credentials = a;
arg_settings_mask |= SETTING_CREDENTIALS;
break;
}
case ARG_LOAD_CREDENTIAL: {
ReadFullFileFlags flags = READ_FULL_FILE_SECURE;
_cleanup_(erase_and_freep) char *data = NULL;
_cleanup_free_ char *word = NULL, *j = NULL;
const char *p = optarg;
Credential *a;
size_t size, i;
r = extract_first_word(&p, &word, ":", EXTRACT_DONT_COALESCE_SEPARATORS);
if (r == -ENOMEM)
return log_oom();
if (r < 0)
return log_error_errno(r, "Failed to parse --set-credential= parameter: %m");
if (r == 0 || !p)
return log_error_errno(SYNTHETIC_ERRNO(EINVAL), "Missing value for --set-credential=: %s", optarg);
if (!credential_name_valid(word))
return log_error_errno(SYNTHETIC_ERRNO(EINVAL), "Credential name is not valid: %s", word);
for (i = 0; i < arg_n_credentials; i++)
if (streq(arg_credentials[i].id, word))
return log_error_errno(SYNTHETIC_ERRNO(EEXIST), "Duplicate credential '%s', refusing.", word);
if (path_is_absolute(p))
flags |= READ_FULL_FILE_CONNECT_SOCKET;
else {
const char *e;
e = getenv("CREDENTIALS_DIRECTORY");
if (!e)
return log_error_errno(SYNTHETIC_ERRNO(EINVAL), "Credential not available (no credentials passed at all): %s", word);
j = path_join(e, p);
if (!j)
return log_oom();
}
r = read_full_file_full(AT_FDCWD, j ?: p, flags, &data, &size);
if (r < 0)
return log_error_errno(r, "Failed to read credential '%s': %m", j ?: p);
a = reallocarray(arg_credentials, arg_n_credentials + 1, sizeof(Credential));
if (!a)
return log_oom();
a[arg_n_credentials++] = (Credential) {
.id = TAKE_PTR(word),
.data = TAKE_PTR(data),
.size = size,
};
arg_credentials = a;
arg_settings_mask |= SETTING_CREDENTIALS;
break;
}
case '?':
return -EINVAL;
default:
assert_not_reached("Unhandled option");
}
if (argc > optind) {
strv_free(arg_parameters);
arg_parameters = strv_copy(argv + optind);
if (!arg_parameters)
return log_oom();
arg_settings_mask |= SETTING_START_MODE;
}
if (arg_ephemeral && arg_template && !arg_directory)
/* User asked for ephemeral execution but specified --template= instead of --directory=. Semantically
* such an invocation makes some sense, see https://github.com/systemd/systemd/issues/3667. Let's
* accept this here, and silently make "--ephemeral --template=" equivalent to "--ephemeral
* --directory=". */
arg_directory = TAKE_PTR(arg_template);
arg_caps_retain = (arg_caps_retain | plus | (arg_private_network ? UINT64_C(1) << CAP_NET_ADMIN : 0)) & ~minus;
/* Make sure to parse environment before we reset the settings mask below */
r = parse_environment();
if (r < 0)
return r;
/* 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;
return 1;
}
static int verify_arguments(void) {
int r;
if (arg_start_mode == START_PID2 && arg_unified_cgroup_hierarchy == CGROUP_UNIFIED_UNKNOWN) {
/* If we are running the stub init in the container, we don't need to look at what the init
* in the container supports, because we are not using it. Let's immediately pick the right
* setting based on the host system configuration.
*
* We only do this, if the user didn't use an environment variable to override the detection.
*/
r = cg_all_unified();
if (r < 0)
return log_error_errno(r, "Failed to determine whether we are in all unified mode.");
if (r > 0)
arg_unified_cgroup_hierarchy = CGROUP_UNIFIED_ALL;
else if (cg_unified_controller(SYSTEMD_CGROUP_CONTROLLER) > 0)
arg_unified_cgroup_hierarchy = CGROUP_UNIFIED_SYSTEMD;
else
arg_unified_cgroup_hierarchy = CGROUP_UNIFIED_NONE;
}
if (arg_userns_mode != USER_NAMESPACE_NO)
arg_mount_settings |= MOUNT_USE_USERNS;
if (arg_private_network)
arg_mount_settings |= MOUNT_APPLY_APIVFS_NETNS;
if (!(arg_clone_ns_flags & CLONE_NEWPID) ||
!(arg_clone_ns_flags & CLONE_NEWUTS)) {
arg_register = false;
if (arg_start_mode != START_PID1)
return log_error_errno(SYNTHETIC_ERRNO(EINVAL), "--boot cannot be used without namespacing.");
}
if (arg_userns_mode == USER_NAMESPACE_PICK)
arg_userns_chown = true;
if (arg_start_mode == START_BOOT && arg_kill_signal <= 0)
arg_kill_signal = SIGRTMIN+3;
if (arg_volatile_mode != VOLATILE_NO) /* Make sure all file systems contained in the image are mounted read-only if we are in volatile mode */
arg_read_only = true;
if (has_custom_root_mount(arg_custom_mounts, arg_n_custom_mounts))
arg_read_only = true;
if (arg_keep_unit && arg_register && cg_pid_get_owner_uid(0, NULL) >= 0)
/* Save the user from accidentally registering either user-$SESSION.scope or user@.service.
* The latter is not technically a user session, but we don't need to labour the point. */
return log_error_errno(SYNTHETIC_ERRNO(EINVAL), "--keep-unit --register=yes may not be used when invoked from a user session.");
if (arg_directory && arg_image)
return log_error_errno(SYNTHETIC_ERRNO(EINVAL), "--directory= and --image= may not be combined.");
if (arg_template && arg_image)
return log_error_errno(SYNTHETIC_ERRNO(EINVAL), "--template= and --image= may not be combined.");
if (arg_template && !(arg_directory || arg_machine))
return log_error_errno(SYNTHETIC_ERRNO(EINVAL), "--template= needs --directory= or --machine=.");
if (arg_ephemeral && arg_template)
return log_error_errno(SYNTHETIC_ERRNO(EINVAL), "--ephemeral and --template= may not be combined.");
if (arg_ephemeral && !IN_SET(arg_link_journal, LINK_NO, LINK_AUTO))
return log_error_errno(SYNTHETIC_ERRNO(EINVAL), "--ephemeral and --link-journal= may not be combined.");
if (arg_userns_mode != USER_NAMESPACE_NO && !userns_supported())
return log_error_errno(SYNTHETIC_ERRNO(EOPNOTSUPP), "--private-users= is not supported, kernel compiled without user namespace support.");
if (arg_userns_chown && arg_read_only)
return log_error_errno(SYNTHETIC_ERRNO(EINVAL),
"--read-only and --private-users-chown may not be combined.");
/* We don't support --private-users-chown together with any of the volatile modes since we couldn't
* change the read-only part of the tree (i.e. /usr) anyway, or because it would trigger a massive
* copy-up (in case of overlay) making the entire exercise pointless. */
if (arg_userns_chown && arg_volatile_mode != VOLATILE_NO)
return log_error_errno(SYNTHETIC_ERRNO(EINVAL), "--volatile= and --private-users-chown may not be combined.");
/* If --network-namespace-path is given with any other network-related option (except --private-network),
* we need to error out, to avoid conflicts between different network options. */
if (arg_network_namespace_path &&
(arg_network_interfaces || arg_network_macvlan ||
arg_network_ipvlan || arg_network_veth_extra ||
arg_network_bridge || arg_network_zone ||
arg_network_veth))
return log_error_errno(SYNTHETIC_ERRNO(EINVAL), "--network-namespace-path= cannot be combined with other network options.");
if (arg_network_bridge && arg_network_zone)
return log_error_errno(SYNTHETIC_ERRNO(EINVAL),
"--network-bridge= and --network-zone= may not be combined.");
if (arg_userns_mode != USER_NAMESPACE_NO && (arg_mount_settings & MOUNT_APPLY_APIVFS_NETNS) && !arg_private_network)
return log_error_errno(SYNTHETIC_ERRNO(EINVAL), "Invalid namespacing settings. Mounting sysfs with --private-users requires --private-network.");
if (arg_userns_mode != USER_NAMESPACE_NO && !(arg_mount_settings & MOUNT_APPLY_APIVFS_RO))
return log_error_errno(SYNTHETIC_ERRNO(EINVAL), "Cannot combine --private-users with read-write mounts.");
if (arg_expose_ports && !arg_private_network)
return log_error_errno(SYNTHETIC_ERRNO(EINVAL), "Cannot use --port= without private networking.");
#if ! HAVE_LIBIPTC
if (arg_expose_ports)
return log_error_errno(SYNTHETIC_ERRNO(EOPNOTSUPP), "--port= is not supported, compiled without libiptc support.");
#endif
r = custom_mount_check_all();
if (r < 0)
return r;
return 0;
}
static int userns_lchown(const char *p, uid_t uid, gid_t gid) {
assert(p);
if (arg_userns_mode == USER_NAMESPACE_NO)
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;
int r;
q = prefix_roota(root, path);
r = mkdir_errno_wrapper(q, mode);
if (r == -EEXIST)
return 0;
if (r < 0)
return r;
return userns_lchown(q, uid, gid);
}
static const char *timezone_from_path(const char *path) {
return PATH_STARTSWITH_SET(
path,
"../usr/share/zoneinfo/",
"/usr/share/zoneinfo/");
}
static bool etc_writable(void) {
return !arg_read_only || IN_SET(arg_volatile_mode, VOLATILE_YES, VOLATILE_OVERLAY);
}
static int setup_timezone(const char *dest) {
_cleanup_free_ char *p = NULL, *etc = NULL;
const char *where, *check;
TimezoneMode m;
int r;
assert(dest);
if (IN_SET(arg_timezone, TIMEZONE_AUTO, TIMEZONE_SYMLINK)) {
r = readlink_malloc("/etc/localtime", &p);
if (r == -ENOENT && arg_timezone == TIMEZONE_AUTO)
m = etc_writable() ? TIMEZONE_DELETE : TIMEZONE_OFF;
else if (r == -EINVAL && arg_timezone == TIMEZONE_AUTO) /* regular file? */
m = etc_writable() ? TIMEZONE_COPY : TIMEZONE_BIND;
else if (r < 0) {
log_warning_errno(r, "Failed to read host's /etc/localtime symlink, not updating container timezone: %m");
/* To handle warning, delete /etc/localtime and replace it with a symbolic link to a time zone data
* file.
*
* Example:
* ln -s /usr/share/zoneinfo/UTC /etc/localtime
*/
return 0;
} else if (arg_timezone == TIMEZONE_AUTO)
m = etc_writable() ? TIMEZONE_SYMLINK : TIMEZONE_BIND;
else
m = arg_timezone;
} else
m = arg_timezone;
if (m == TIMEZONE_OFF)
return 0;
r = chase_symlinks("/etc", dest, CHASE_PREFIX_ROOT, &etc, NULL);
if (r < 0) {
log_warning_errno(r, "Failed to resolve /etc path in container, ignoring: %m");
return 0;
}
where = strjoina(etc, "/localtime");
switch (m) {
case TIMEZONE_DELETE:
if (unlink(where) < 0)
log_full_errno(errno == ENOENT ? LOG_DEBUG : LOG_WARNING, errno, "Failed to remove '%s', ignoring: %m", where);
return 0;
case TIMEZONE_SYMLINK: {
_cleanup_free_ char *q = NULL;
const char *z, *what;
z = timezone_from_path(p);
if (!z) {
log_warning("/etc/localtime does not point into /usr/share/zoneinfo/, not updating container timezone.");
return 0;
}
r = readlink_malloc(where, &q);
if (r >= 0 && streq_ptr(timezone_from_path(q), z))
return 0; /* Already pointing to the right place? Then do nothing .. */
check = strjoina(dest, "/usr/share/zoneinfo/", z);
r = chase_symlinks(check, dest, 0, NULL, NULL);
if (r < 0)
log_debug_errno(r, "Timezone %s does not exist (or is not accessible) in container, not creating symlink: %m", z);
else {
if (unlink(where) < 0 && errno != ENOENT) {
log_full_errno(IN_SET(errno, EROFS, EACCES, EPERM) ? LOG_DEBUG : LOG_WARNING, /* Don't complain on read-only images */
errno, "Failed to remove existing timezone info %s in container, ignoring: %m", where);
return 0;
}
what = strjoina("../usr/share/zoneinfo/", z);
if (symlink(what, where) < 0) {
log_full_errno(IN_SET(errno, EROFS, EACCES, EPERM) ? LOG_DEBUG : LOG_WARNING,
errno, "Failed to correct timezone of container, ignoring: %m");
return 0;
}
break;
}
_fallthrough_;
}
case TIMEZONE_BIND: {
_cleanup_free_ char *resolved = NULL;
int found;
found = chase_symlinks(where, dest, CHASE_NONEXISTENT, &resolved, NULL);
if (found < 0) {
log_warning_errno(found, "Failed to resolve /etc/localtime path in container, ignoring: %m");
return 0;
}
if (found == 0) /* missing? */
(void) touch(resolved);
r = mount_nofollow_verbose(LOG_WARNING, "/etc/localtime", resolved, NULL, MS_BIND, NULL);
if (r >= 0)
return mount_nofollow_verbose(LOG_ERR, NULL, resolved, NULL, MS_BIND|MS_REMOUNT|MS_RDONLY|MS_NOSUID|MS_NODEV, NULL);
_fallthrough_;
}
case TIMEZONE_COPY:
/* If mounting failed, try to copy */
r = copy_file_atomic("/etc/localtime", where, 0644, 0, 0, COPY_REFLINK|COPY_REPLACE);
if (r < 0) {
log_full_errno(IN_SET(r, -EROFS, -EACCES, -EPERM) ? LOG_DEBUG : LOG_WARNING, r,
"Failed to copy /etc/localtime to %s, ignoring: %m", where);
return 0;
}
break;
default:
assert_not_reached("unexpected mode");
}
/* Fix permissions of the symlink or file copy we just created */
r = userns_lchown(where, 0, 0);
if (r < 0)
log_warning_errno(r, "Failed to chown /etc/localtime, ignoring: %m");
return 0;
}
static int have_resolv_conf(const char *path) {
assert(path);
if (access(path, F_OK) < 0) {
if (errno == ENOENT)
return 0;
return log_debug_errno(errno, "Failed to determine whether '%s' is available: %m", path);
}
return 1;
}
static int resolved_listening(void) {
_cleanup_(sd_bus_error_free) sd_bus_error error = SD_BUS_ERROR_NULL;
_cleanup_(sd_bus_flush_close_unrefp) sd_bus *bus = NULL;
_cleanup_free_ char *dns_stub_listener_mode = NULL;
int r;
/* Check if resolved is listening */
r = sd_bus_open_system(&bus);
if (r < 0)
return log_debug_errno(r, "Failed to open system bus: %m");
r = bus_name_has_owner(bus, "org.freedesktop.resolve1", NULL);
if (r < 0)
return log_debug_errno(r, "Failed to check whether the 'org.freedesktop.resolve1' bus name is taken: %m");
if (r == 0)
return 0;
r = sd_bus_get_property_string(bus,
"org.freedesktop.resolve1",
"/org/freedesktop/resolve1",
"org.freedesktop.resolve1.Manager",
"DNSStubListener",
&error,
&dns_stub_listener_mode);
if (r < 0)
return log_debug_errno(r, "Failed to query DNSStubListener property: %s", bus_error_message(&error, r));
return STR_IN_SET(dns_stub_listener_mode, "udp", "yes");
}
static int setup_resolv_conf(const char *dest) {
_cleanup_free_ char *etc = NULL;
const char *where, *what;
ResolvConfMode m;
int r;
assert(dest);
if (arg_resolv_conf == RESOLV_CONF_AUTO) {
if (arg_private_network)
m = RESOLV_CONF_OFF;
else if (have_resolv_conf(PRIVATE_STUB_RESOLV_CONF) > 0 && resolved_listening() > 0)
m = etc_writable() ? RESOLV_CONF_COPY_STUB : RESOLV_CONF_BIND_STUB;
else if (have_resolv_conf("/etc/resolv.conf") > 0)
m = etc_writable() ? RESOLV_CONF_COPY_HOST : RESOLV_CONF_BIND_HOST;
else
m = etc_writable() ? RESOLV_CONF_DELETE : RESOLV_CONF_OFF;
} else
m = arg_resolv_conf;
if (m == RESOLV_CONF_OFF)
return 0;
r = chase_symlinks("/etc", dest, CHASE_PREFIX_ROOT, &etc, NULL);
if (r < 0) {
log_warning_errno(r, "Failed to resolve /etc path in container, ignoring: %m");
return 0;
}
where = strjoina(etc, "/resolv.conf");
if (m == RESOLV_CONF_DELETE) {
if (unlink(where) < 0)
log_full_errno(errno == ENOENT ? LOG_DEBUG : LOG_WARNING, errno, "Failed to remove '%s', ignoring: %m", where);
return 0;
}
if (IN_SET(m, RESOLV_CONF_BIND_STATIC, RESOLV_CONF_REPLACE_STATIC, RESOLV_CONF_COPY_STATIC))
what = PRIVATE_STATIC_RESOLV_CONF;
else if (IN_SET(m, RESOLV_CONF_BIND_UPLINK, RESOLV_CONF_REPLACE_UPLINK, RESOLV_CONF_COPY_UPLINK))
what = PRIVATE_UPLINK_RESOLV_CONF;
else if (IN_SET(m, RESOLV_CONF_BIND_STUB, RESOLV_CONF_REPLACE_STUB, RESOLV_CONF_COPY_STUB))
what = PRIVATE_STUB_RESOLV_CONF;
else
what = "/etc/resolv.conf";
if (IN_SET(m, RESOLV_CONF_BIND_HOST, RESOLV_CONF_BIND_STATIC, RESOLV_CONF_BIND_UPLINK, RESOLV_CONF_BIND_STUB)) {
_cleanup_free_ char *resolved = NULL;
int found;
found = chase_symlinks(where, dest, CHASE_NONEXISTENT, &resolved, NULL);
if (found < 0) {
log_warning_errno(found, "Failed to resolve /etc/resolv.conf path in container, ignoring: %m");
return 0;
}
if (found == 0) /* missing? */
(void) touch(resolved);
r = mount_nofollow_verbose(LOG_WARNING, what, resolved, NULL, MS_BIND, NULL);
if (r >= 0)
return mount_nofollow_verbose(LOG_ERR, NULL, resolved, NULL, MS_BIND|MS_REMOUNT|MS_RDONLY|MS_NOSUID|MS_NODEV, NULL);
/* If that didn't work, let's copy the file */
}
if (IN_SET(m, RESOLV_CONF_REPLACE_HOST, RESOLV_CONF_REPLACE_STATIC, RESOLV_CONF_REPLACE_UPLINK, RESOLV_CONF_REPLACE_STUB))
r = copy_file_atomic(what, where, 0644, 0, 0, COPY_REFLINK|COPY_REPLACE);
else
r = copy_file(what, where, O_TRUNC|O_NOFOLLOW, 0644, 0, 0, COPY_REFLINK);
if (r < 0) {
/* If the file already exists as symlink, let's suppress the warning, under the assumption that
* resolved or something similar runs inside and the symlink points there.
*
* If the disk image is read-only, there's also no point in complaining.
*/
log_full_errno(!IN_SET(RESOLV_CONF_COPY_HOST, RESOLV_CONF_COPY_STATIC, RESOLV_CONF_COPY_UPLINK, RESOLV_CONF_COPY_STUB) &&
IN_SET(r, -ELOOP, -EROFS, -EACCES, -EPERM) ? LOG_DEBUG : LOG_WARNING, r,
"Failed to copy /etc/resolv.conf to %s, ignoring: %m", where);
return 0;
}
r = userns_lchown(where, 0, 0);
if (r < 0)
log_warning_errno(r, "Failed to chown /etc/resolv.conf, ignoring: %m");
return 0;
}
static int setup_boot_id(void) {
_cleanup_(unlink_and_freep) char *from = NULL;
_cleanup_free_ char *path = NULL;
sd_id128_t rnd = SD_ID128_NULL;
const char *to;
int r;
/* Generate a new randomized boot ID, so that each boot-up of the container gets a new one */
r = tempfn_random_child("/run", "proc-sys-kernel-random-boot-id", &path);
if (r < 0)
return log_error_errno(r, "Failed to generate random boot ID path: %m");
r = sd_id128_randomize(&rnd);
if (r < 0)
return log_error_errno(r, "Failed to generate random boot id: %m");
r = id128_write(path, ID128_UUID, rnd, false);
if (r < 0)
return log_error_errno(r, "Failed to write boot id: %m");
from = TAKE_PTR(path);
to = "/proc/sys/kernel/random/boot_id";
r = mount_nofollow_verbose(LOG_ERR, from, to, NULL, MS_BIND, NULL);
if (r < 0)
return r;
return mount_nofollow_verbose(LOG_ERR, NULL, to, NULL, MS_BIND|MS_REMOUNT|MS_RDONLY|MS_NOSUID|MS_NOEXEC|MS_NODEV, NULL);
}
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";
_cleanup_umask_ mode_t u;
const char *d;
int r = 0;
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 = path_join("/dev/", d);
if (!from)
return log_oom();
to = path_join(dest, from);
if (!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))
return log_error_errno(SYNTHETIC_ERRNO(EIO),
"%s is not a char or block device, cannot copy.", from);
else {
_cleanup_free_ char *sl = NULL, *prefixed = NULL, *dn = NULL, *t = NULL;
if (mknod(to, st.st_mode, st.st_rdev) < 0) {
/* Explicitly warn the user when /dev is already populated. */
if (errno == EEXIST)
log_notice("%s/dev is pre-mounted and pre-populated. If a pre-mounted /dev is provided it needs to be an unpopulated file system.", dest);
if (errno != EPERM)
return log_error_errno(errno, "mknod(%s) failed: %m", to);
/* Some systems abusively restrict mknod but allow bind mounts. */
r = touch(to);
if (r < 0)
return log_error_errno(r, "touch (%s) failed: %m", to);
r = mount_nofollow_verbose(LOG_DEBUG, from, to, NULL, MS_BIND, NULL);
if (r < 0)
return log_error_errno(r, "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);
dn = path_join("/dev", S_ISCHR(st.st_mode) ? "char" : "block");
if (!dn)
return log_oom();
r = userns_mkdir(dest, dn, 0755, 0, 0);
if (r < 0)
return log_error_errno(r, "Failed to create '%s': %m", dn);
if (asprintf(&sl, "%s/%u:%u", dn, major(st.st_rdev), minor(st.st_rdev)) < 0)
return log_oom();
prefixed = path_join(dest, sl);
if (!prefixed)
return log_oom();
t = path_join("..", d);
if (!t)
return log_oom();
if (symlink(t, prefixed) < 0)
log_debug_errno(errno, "Failed to symlink '%s' to '%s': %m", t, prefixed);
}
}
return r;
}
static int make_extra_nodes(const char *dest) {
_cleanup_umask_ mode_t u;
size_t i;
int r;
u = umask(0000);
for (i = 0; i < arg_n_extra_nodes; i++) {
_cleanup_free_ char *path = NULL;
DeviceNode *n = arg_extra_nodes + i;
path = path_join(dest, n->path);
if (!path)
return log_oom();
if (mknod(path, n->mode, S_ISCHR(n->mode) || S_ISBLK(n->mode) ? makedev(n->major, n->minor) : 0) < 0)
return log_error_errno(errno, "Failed to create device node '%s': %m", path);
r = chmod_and_chown(path, n->mode, n->uid, n->gid);
if (r < 0)
return log_error_errno(r, "Failed to adjust device node ownership of '%s': %m", path);
}
return 0;
}
static int setup_pts(const char *dest) {
_cleanup_free_ char *options = NULL;
const char *p;
int r;
#if 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");
r = mkdir_errno_wrapper(p, 0755);
if (r < 0)
return log_error_errno(r, "Failed to create /dev/pts: %m");
r = mount_nofollow_verbose(LOG_ERR, "devpts", p, "devpts", MS_NOSUID|MS_NOEXEC, options);
if (r < 0)
return r;
r = userns_lchown(p, 0, 0);
if (r < 0)
return log_error_errno(r, "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");
r = userns_lchown(p, 0, 0);
if (r < 0)
return log_error_errno(r, "Failed to chown /dev/ptmx: %m");
/* And fix /dev/pts/ptmx ownership */
p = prefix_roota(dest, "/dev/pts/ptmx");
r = userns_lchown(p, 0, 0);
if (r < 0)
return log_error_errno(r, "Failed to chown /dev/pts/ptmx: %m");
return 0;
}
static int setup_stdio_as_dev_console(void) {
_cleanup_close_ int terminal = -1;
int r;
/* We open the TTY in O_NOCTTY mode, so that we do not become controller yet. We'll do that later
* explicitly, if we are configured to. */
terminal = open_terminal("/dev/console", O_RDWR|O_NOCTTY);
if (terminal < 0)
return log_error_errno(terminal, "Failed to open console: %m");
/* Make sure we can continue logging to the original stderr, even if
* stderr points elsewhere now */
r = log_dup_console();
if (r < 0)
return log_error_errno(r, "Failed to duplicate stderr: %m");
/* invalidates 'terminal' on success and failure */
r = rearrange_stdio(terminal, terminal, terminal);
TAKE_FD(terminal);
if (r < 0)
return log_error_errno(r, "Failed to move console to stdin/stdout/stderr: %m");
return 0;
}
static int setup_dev_console(const char *console) {
_cleanup_free_ char *p = NULL;
int r;
/* Create /dev/console symlink */
r = path_make_relative("/dev", console, &p);
if (r < 0)
return log_error_errno(r, "Failed to create relative path: %m");
if (symlink(p, "/dev/console") < 0)
return log_error_errno(errno, "Failed to create /dev/console symlink: %m");
return 0;
}
static int setup_keyring(void) {
key_serial_t keyring;
/* Allocate a new session keyring for the container. This makes sure the keyring of the session
* systemd-nspawn was invoked from doesn't leak into the container. Note that by default we block
* keyctl() and request_key() anyway via seccomp so doing this operation isn't strictly necessary,
* but in case people explicitly allow-list these system calls let's make sure we don't leak anything
* into the container. */
keyring = keyctl(KEYCTL_JOIN_SESSION_KEYRING, 0, 0, 0, 0);
if (keyring == -1) {
if (errno == ENOSYS)
log_debug_errno(errno, "Kernel keyring not supported, ignoring.");
else if (ERRNO_IS_PRIVILEGE(errno))
log_debug_errno(errno, "Kernel keyring access prohibited, ignoring.");
else
return log_error_errno(errno, "Setting up kernel keyring failed: %m");
}
return 0;
}
static int setup_credentials(const char *root) {
const char *q;
int r;
if (arg_n_credentials <= 0)
return 0;
r = userns_mkdir(root, "/run/host", 0755, 0, 0);
if (r < 0)
return log_error_errno(r, "Failed to create /run/host: %m");
r = userns_mkdir(root, "/run/host/credentials", 0700, 0, 0);
if (r < 0)
return log_error_errno(r, "Failed to create /run/host/credentials: %m");
q = prefix_roota(root, "/run/host/credentials");
r = mount_nofollow_verbose(LOG_ERR, NULL, q, "ramfs", MS_NOSUID|MS_NOEXEC|MS_NODEV, "mode=0700");
if (r < 0)
return r;
for (size_t i = 0; i < arg_n_credentials; i++) {
_cleanup_free_ char *j = NULL;
_cleanup_close_ int fd = -1;
j = path_join(q, arg_credentials[i].id);
if (!j)
return log_oom();
fd = open(j, O_CREAT|O_EXCL|O_WRONLY|O_CLOEXEC|O_NOFOLLOW, 0600);
if (fd < 0)
return log_error_errno(errno, "Failed to create credential file %s: %m", j);
r = loop_write(fd, arg_credentials[i].data, arg_credentials[i].size, /* do_poll= */ false);
if (r < 0)
return log_error_errno(r, "Failed to write credential to file %s: %m", j);
if (fchmod(fd, 0400) < 0)
return log_error_errno(errno, "Failed to adjust access mode of %s: %m", j);
if (arg_userns_mode != USER_NAMESPACE_NO) {
if (fchown(fd, arg_uid_shift, arg_uid_shift) < 0)
return log_error_errno(errno, "Failed to adjust ownership of %s: %m", j);
}
}
if (chmod(q, 0500) < 0)
return log_error_errno(errno, "Failed to adjust access mode of %s: %m", q);
r = userns_lchown(q, 0, 0);
if (r < 0)
return r;
/* Make both mount and superblock read-only now */
r = mount_nofollow_verbose(LOG_ERR, NULL, q, NULL, MS_REMOUNT|MS_BIND|MS_RDONLY|MS_NOSUID|MS_NOEXEC|MS_NODEV, NULL);
if (r < 0)
return r;
return mount_nofollow_verbose(LOG_ERR, NULL, q, NULL, MS_REMOUNT|MS_RDONLY|MS_NOSUID|MS_NOEXEC|MS_NODEV, "mode=0500");
}
static int setup_kmsg(int kmsg_socket) {
_cleanup_(unlink_and_freep) char *from = NULL;
_cleanup_free_ char *fifo = NULL;
_cleanup_close_ int fd = -1;
_cleanup_umask_ mode_t u;
int r;
assert(kmsg_socket >= 0);
u = umask(0000);
/* We create the kmsg FIFO as as temporary file in /run, but immediately delete it after bind mounting it to
* /proc/kmsg. While FIFOs on the reading side behave very similar to /proc/kmsg, their writing side behaves
* differently from /dev/kmsg in that writing blocks when nothing is reading. In order to avoid any problems
* with containers deadlocking due to this we simply make /dev/kmsg unavailable to the container. */
r = tempfn_random_child("/run", "proc-kmsg", &fifo);
if (r < 0)
return log_error_errno(r, "Failed to generate kmsg path: %m");
if (mkfifo(fifo, 0600) < 0)
return log_error_errno(errno, "mkfifo() for /run/kmsg failed: %m");
from = TAKE_PTR(fifo);
r = mount_nofollow_verbose(LOG_ERR, from, "/proc/kmsg", NULL, MS_BIND, NULL);
if (r < 0)
return r;
fd = open(from, O_RDWR|O_NONBLOCK|O_CLOEXEC);
if (fd < 0)
return log_error_errno(errno, "Failed to open fifo: %m");
/* Store away the fd in the socket, so that it stays open as long as we run the child */
r = send_one_fd(kmsg_socket, fd, 0);
if (r < 0)
return log_error_errno(r, "Failed to send FIFO fd: %m");
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) {
int r;
if ((arg_clone_ns_flags & CLONE_NEWUTS) == 0)
return 0;
r = sethostname_idempotent(arg_hostname ?: arg_machine);
if (r < 0)
return log_error_errno(r, "Failed to set hostname: %m");
return 0;
}
static int setup_journal(const char *directory) {
_cleanup_free_ char *d = NULL;
char id[SD_ID128_STRING_MAX];
const char *dirname, *p, *q;
sd_id128_t this_id;
bool try;
int r;
/* Don't link journals in ephemeral mode */
if (arg_ephemeral)
return 0;
if (arg_link_journal == LINK_NO)
return 0;
try = arg_link_journal_try || arg_link_journal == LINK_AUTO;
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(arg_uuid, this_id)) {
log_full(try ? LOG_WARNING : LOG_ERR,
"Host and machine ids are equal (%s): refusing to link journals", sd_id128_to_string(arg_uuid, id));
if (try)
return 0;
return -EEXIST;
}
FOREACH_STRING(dirname, "/var", "/var/log", "/var/log/journal") {
r = userns_mkdir(directory, dirname, 0755, 0, 0);
if (r < 0) {
bool ignore = r == -EROFS && try;
log_full_errno(ignore ? LOG_DEBUG : LOG_ERR, r,
"Failed to create %s%s: %m", dirname, ignore ? ", ignoring" : "");
return ignore ? 0 : r;
}
}
(void) sd_id128_to_string(arg_uuid, id);
p = strjoina("/var/log/journal/", id);
q = prefix_roota(directory, p);
if (path_is_mount_point(p, NULL, 0) > 0) {
if (try)
return 0;
return log_error_errno(SYNTHETIC_ERRNO(EEXIST),
"%s: already a mount point, refusing to use for journal", p);
}
if (path_is_mount_point(q, NULL, 0) > 0) {
if (try)
return 0;
return log_error_errno(SYNTHETIC_ERRNO(EEXIST),
"%s: already a mount point, refusing to use for journal", q);
}
r = readlink_and_make_absolute(p, &d);
if (r >= 0) {
if (IN_SET(arg_link_journal, LINK_GUEST, LINK_AUTO) &&
path_equal(d, q)) {
r = userns_mkdir(directory, p, 0755, 0, 0);
if (r < 0)
log_warning_errno(r, "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
return log_error_errno(errno, "Failed to remove %s: %m", p);
}
} else if (r != -ENOENT)
return log_error_errno(r, "readlink(%s) failed: %m", p);
if (arg_link_journal == LINK_GUEST) {
if (symlink(q, p) < 0) {
if (try) {
log_debug_errno(errno, "Failed to symlink %s to %s, skipping journal setup: %m", q, p);
return 0;
} else
return log_error_errno(errno, "Failed to symlink %s to %s: %m", q, p);
}
r = userns_mkdir(directory, p, 0755, 0, 0);
if (r < 0)
log_warning_errno(r, "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
* permanent journal set up, don't force it here */
r = mkdir_errno_wrapper(p, 0755);
if (r < 0 && r != -EEXIST) {
if (try) {
log_debug_errno(r, "Failed to create %s, skipping journal setup: %m", p);
return 0;
} else
return log_error_errno(r, "Failed to create %s: %m", p);
}
} 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)
return log_error_errno(r, "Failed to create %s: %m", q);
r = mount_nofollow_verbose(LOG_DEBUG, p, q, NULL, MS_BIND, NULL);
if (r < 0)
return log_error_errno(errno, "Failed to bind mount journal from host into guest: %m");
return 0;
}
static int drop_capabilities(uid_t uid) {
CapabilityQuintet q;
/* Let's initialize all five capability sets to something valid. If the quintet was configured via
* OCI use that, but fill in missing bits. If it wasn't then derive the quintet in full from
* arg_caps_retain. */
if (capability_quintet_is_set(&arg_full_capabilities)) {
q = arg_full_capabilities;
if (q.bounding == (uint64_t) -1)
q.bounding = uid == 0 ? arg_caps_retain : 0;
if (q.effective == (uint64_t) -1)
q.effective = uid == 0 ? q.bounding : 0;
if (q.inheritable == (uint64_t) -1)
q.inheritable = uid == 0 ? q.bounding : 0;
if (q.permitted == (uint64_t) -1)
q.permitted = uid == 0 ? q.bounding : 0;
if (q.ambient == (uint64_t) -1 && ambient_capabilities_supported())
q.ambient = 0;
if (capability_quintet_mangle(&q))
return log_error_errno(SYNTHETIC_ERRNO(EPERM), "Cannot set capabilities that are not in the current bounding set.");
} else {
q = (CapabilityQuintet) {
.bounding = arg_caps_retain,
.effective = uid == 0 ? arg_caps_retain : 0,
.inheritable = uid == 0 ? arg_caps_retain : 0,
.permitted = uid == 0 ? arg_caps_retain : 0,
.ambient = ambient_capabilities_supported() ? 0 : (uint64_t) -1,
};
/* If we're not using OCI, proceed with mangled capabilities (so we don't error out)
* in order to maintain the same behavior as systemd < 242. */
if (capability_quintet_mangle(&q))
log_full(arg_quiet ? LOG_DEBUG : LOG_WARNING,
"Some capabilities will not be set because they are not in the current bounding set.");
}
return capability_quintet_enforce(&q);
}
static int reset_audit_loginuid(void) {
_cleanup_free_ char *p = NULL;
int r;
if ((arg_clone_ns_flags & CLONE_NEWPID) == 0)
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", WRITE_STRING_FILE_DISABLE_BUFFER);
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_propagate(const char *root) {
const char *p, *q;
int r;
(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);
r = userns_mkdir(root, "/run/host", 0755, 0, 0);
if (r < 0)
return log_error_errno(r, "Failed to create /run/host: %m");
r = userns_mkdir(root, "/run/host/incoming", 0600, 0, 0);
if (r < 0)
return log_error_errno(r, "Failed to create /run/host/incoming: %m");
q = prefix_roota(root, "/run/host/incoming");
r = mount_nofollow_verbose(LOG_ERR, p, q, NULL, MS_BIND, NULL);
if (r < 0)
return r;
r = mount_nofollow_verbose(LOG_ERR, NULL, q, NULL, MS_BIND|MS_REMOUNT|MS_RDONLY, NULL);
if (r < 0)
return r;
/* machined will MS_MOVE into that directory, and that's only supported for non-shared mounts. */
return mount_nofollow_verbose(LOG_ERR, NULL, q, NULL, MS_SLAVE, NULL);
}
static int setup_machine_id(const char *directory) {
const char *etc_machine_id;
sd_id128_t id;
int r;
/* If the UUID in the container is already set, then that's what counts, and we use. If it isn't set, and the
* caller passed --uuid=, then we'll pass it in the $container_uuid env var to PID 1 of the container. The
* assumption is that PID 1 will then write it to /etc/machine-id to make it persistent. If --uuid= is not
* passed we generate a random UUID, and pass it via $container_uuid. In effect this means that /etc/machine-id
* in the container and our idea of the container UUID will always be in sync (at least if PID 1 in the
* container behaves nicely). */
etc_machine_id = prefix_roota(directory, "/etc/machine-id");
r = id128_read(etc_machine_id, ID128_PLAIN_OR_UNINIT, &id);
if (r < 0) {
if (!IN_SET(r, -ENOENT, -ENOMEDIUM)) /* If the file is missing or empty, we don't mind */
return log_error_errno(r, "Failed to read machine ID from container image: %m");
if (sd_id128_is_null(arg_uuid)) {
r = sd_id128_randomize(&arg_uuid);
if (r < 0)
return log_error_errno(r, "Failed to acquire randomized machine UUID: %m");
}
} else {
if (sd_id128_is_null(id))
return log_error_errno(SYNTHETIC_ERRNO(EINVAL),
"Machine ID in container image is zero, refusing.");
arg_uuid = id;
}
return 0;
}
static int recursive_chown(const char *directory, uid_t shift, uid_t range) {
int r;
assert(directory);
if (arg_userns_mode == USER_NAMESPACE_NO || !arg_userns_chown)
return 0;
r = path_patch_uid(directory, arg_uid_shift, arg_uid_range);
if (r == -EOPNOTSUPP)
return log_error_errno(r, "Automatic UID/GID adjusting is only supported for UID/GID ranges starting at multiples of 2^16 with a range of 2^16.");
if (r == -EBADE)
return log_error_errno(r, "Upper 16 bits of root directory UID and GID do not match.");
if (r < 0)
return log_error_errno(r, "Failed to adjust UID/GID shift of OS tree: %m");
if (r == 0)
log_debug("Root directory of image is already owned by the right UID/GID range, skipping recursive chown operation.");
else
log_debug("Patched directory tree to match UID/GID range.");
return r;
}
/*
* Return values:
* < 0 : wait_for_terminate() failed to get the state of the
* container, the container was terminated by a signal, or
* failed for an unknown reason. No change is made to the
* container argument.
* > 0 : The program executed in the container terminated with an
* error. The exit code of the program executed in the
* container is returned. The container argument has been set
* to CONTAINER_TERMINATED.
* 0 : The container is being rebooted, has been shut down or exited
* successfully. The container argument has been set to either
* CONTAINER_TERMINATED or CONTAINER_REBOOTED.
*
* That is, success is indicated by a return value of zero, and an
* error is indicated by a non-zero value.
*/
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;
}
_fallthrough_;
case CLD_DUMPED:
return log_error_errno(SYNTHETIC_ERRNO(EIO),
"Container %s terminated by signal %s.", arg_machine, signal_to_string(status.si_status));
default:
return log_error_errno(SYNTHETIC_ERRNO(EIO),
"Container %s failed due to unknown reason.", arg_machine);
}
}
static int on_orderly_shutdown(sd_event_source *s, const struct signalfd_siginfo *si, void *userdata) {
pid_t pid;
pid = PTR_TO_PID(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 on_sigchld(sd_event_source *s, const struct signalfd_siginfo *ssi, void *userdata) {
pid_t pid;
assert(s);
assert(ssi);
pid = PTR_TO_PID(userdata);
for (;;) {
siginfo_t si = {};
if (waitid(P_ALL, 0, &si, WNOHANG|WNOWAIT|WEXITED) < 0)
return log_error_errno(errno, "Failed to waitid(): %m");
if (si.si_pid == 0) /* No pending children. */
break;
if (si.si_pid == pid) {
/* The main process we care for has exited. Return from
* signal handler but leave the zombie. */
sd_event_exit(sd_event_source_get_event(s), 0);
break;
}
/* Reap all other children. */
(void) waitid(P_PID, si.si_pid, &si, WNOHANG|WEXITED);
}
return 0;
}
static int on_request_stop(sd_bus_message *m, void *userdata, sd_bus_error *error) {
pid_t pid;
assert(m);
pid = PTR_TO_PID(userdata);
if (arg_kill_signal > 0) {
log_info("Container termination requested. Attempting to halt container.");
(void) kill(pid, arg_kill_signal);
} else {
log_info("Container termination requested. Exiting.");
sd_event_exit(sd_bus_get_event(sd_bus_message_get_bus(m)), 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
* search for a machine, but instead create a new one
* in /var/lib/machine. */
arg_directory = path_join("/var/lib/machines", arg_machine);
if (!arg_directory)
return log_oom();
}
if (!arg_image && !arg_directory) {
if (arg_machine) {
_cleanup_(image_unrefp) Image *i = NULL;
r = image_find(IMAGE_MACHINE, arg_machine, &i);
if (r == -ENOENT)
return log_error_errno(r, "No image for machine '%s'.", arg_machine);
if (r < 0)
return log_error_errno(r, "Failed to find image for machine '%s': %m", arg_machine);
if (IN_SET(i->type, IMAGE_RAW, IMAGE_BLOCK))
r = free_and_strdup(&arg_image, i->path);
else
r = free_and_strdup(&arg_directory, i->path);
if (r < 0)
return log_oom();
if (!arg_ephemeral)
arg_read_only = arg_read_only || i->read_only;
} else {
r = safe_getcwd(&arg_directory);
if (r < 0)
return log_error_errno(r, "Failed to determine current directory: %m");
}
if (!arg_directory && !arg_image)
return log_error_errno(SYNTHETIC_ERRNO(EINVAL), "Failed to determine path, please use -D or -i.");
}
if (!arg_machine) {
if (arg_directory && path_equal(arg_directory, "/"))
arg_machine = gethostname_malloc();
else {
if (arg_image) {
char *e;
arg_machine = strdup(basename(arg_image));
/* Truncate suffix if there is one */
e = endswith(arg_machine, ".raw");
if (e)
*e = 0;
} else
arg_machine = strdup(basename(arg_directory));
}
if (!arg_machine)
return log_oom();
hostname_cleanup(arg_machine);
if (!machine_name_is_valid(arg_machine))
return log_error_errno(SYNTHETIC_ERRNO(EINVAL), "Failed to determine machine name automatically, please use -M.");
if (arg_ephemeral) {
char *b;
/* Add a random suffix when this is an
* ephemeral machine, so that we can run many
* instances at once without manually having
* to specify -M each time. */
if (asprintf(&b, "%s-%016" PRIx64, arg_machine, random_u64()) < 0)
return log_oom();
free(arg_machine);
arg_machine = b;
}
}
return 0;
}
static int chase_symlinks_and_update(char **p, unsigned flags) {
char *chased;
int r;
assert(p);
if (!*p)
return 0;
r = chase_symlinks(*p, NULL, flags, &chased, NULL);
if (r < 0)
return log_error_errno(r, "Failed to resolve path %s: %m", *p);
return free_and_replace(*p, chased);
}
static int determine_uid_shift(const char *directory) {
int r;
if (arg_userns_mode == USER_NAMESPACE_NO) {
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)))
return log_error_errno(SYNTHETIC_ERRNO(EINVAL),
"UID and GID base of %s don't match.", directory);
arg_uid_range = UINT32_C(0x10000);
}
if (arg_uid_shift > (uid_t) -1 - arg_uid_range)
return log_error_errno(SYNTHETIC_ERRNO(EINVAL),
"UID base too high for UID range.");
return 0;
}
static unsigned long effective_clone_ns_flags(void) {
unsigned long flags = arg_clone_ns_flags;
if (arg_private_network)
flags |= CLONE_NEWNET;
if (arg_use_cgns)
flags |= CLONE_NEWCGROUP;
if (arg_userns_mode != USER_NAMESPACE_NO)
flags |= CLONE_NEWUSER;
return flags;
}
static int patch_sysctl(void) {
/* This table is inspired by runc's sysctl() function */
static const struct {
const char *key;
bool prefix;
unsigned long clone_flags;
} safe_sysctl[] = {
{ "kernel.hostname", false, CLONE_NEWUTS },
{ "kernel.domainname", false, CLONE_NEWUTS },
{ "kernel.msgmax", false, CLONE_NEWIPC },
{ "kernel.msgmnb", false, CLONE_NEWIPC },
{ "kernel.msgmni", false, CLONE_NEWIPC },
{ "kernel.sem", false, CLONE_NEWIPC },
{ "kernel.shmall", false, CLONE_NEWIPC },
{ "kernel.shmmax", false, CLONE_NEWIPC },
{ "kernel.shmmni", false, CLONE_NEWIPC },
{ "fs.mqueue.", true, CLONE_NEWIPC },
{ "net.", true, CLONE_NEWNET },
};
unsigned long flags;
char **k, **v;
int r;
flags = effective_clone_ns_flags();
STRV_FOREACH_PAIR(k, v, arg_sysctl) {
bool good = false;
size_t i;
for (i = 0; i < ELEMENTSOF(safe_sysctl); i++) {
if (!FLAGS_SET(flags, safe_sysctl[i].clone_flags))
continue;
if (safe_sysctl[i].prefix)
good = startswith(*k, safe_sysctl[i].key);
else
good = streq(*k, safe_sysctl[i].key);
if (good)
break;
}
if (!good)
return log_error_errno(SYNTHETIC_ERRNO(EPERM), "Refusing to write to sysctl '%s', as it is not safe in the selected namespaces.", *k);
r = sysctl_write(*k, *v);
if (r < 0)
return log_error_errno(r, "Failed to write sysctl '%s': %m", *k);
}
return 0;
}
static int inner_child(
Barrier *barrier,
const char *directory,
bool secondary,
int kmsg_socket,
int rtnl_socket,
int master_pty_socket,
FDSet *fds,
char **os_release_pairs) {
_cleanup_free_ char *home = NULL;
char as_uuid[ID128_UUID_STRING_MAX];
size_t n_env = 1;
const char *envp[] = {
"PATH=" DEFAULT_PATH_COMPAT,
NULL, /* container */
NULL, /* TERM */
NULL, /* HOME */
NULL, /* USER */
NULL, /* LOGNAME */
NULL, /* container_uuid */
NULL, /* LISTEN_FDS */
NULL, /* LISTEN_PID */
NULL, /* NOTIFY_SOCKET */
NULL, /* CREDENTIALS_DIRECTORY */
NULL
};
const char *exec_target;
_cleanup_strv_free_ char **env_use = NULL;
int r, which_failed;
/* This is the "inner" child process, i.e. the one forked off by the "outer" child process, which is the one
* the container manager itself forked off. At the time of clone() it gained its own CLONE_NEWNS, CLONE_NEWPID,
* CLONE_NEWUTS, CLONE_NEWIPC, CLONE_NEWUSER namespaces. Note that it has its own CLONE_NEWNS namespace,
* separate from the CLONE_NEWNS created for the "outer" child, and also separate from the host's CLONE_NEWNS
* namespace. The reason for having two levels of CLONE_NEWNS namespaces is that the "inner" one is owned by
* the CLONE_NEWUSER namespace of the container, while the "outer" one is owned by the host's CLONE_NEWUSER
* namespace.
*
* Note at this point we have no CLONE_NEWNET namespace yet. We'll acquire that one later through
* unshare(). See below. */
assert(barrier);
assert(directory);
assert(kmsg_socket >= 0);
log_debug("Inner child is initializing.");
if (arg_userns_mode != USER_NAMESPACE_NO) {
/* 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 */
return log_error_errno(SYNTHETIC_ERRNO(ESRCH), "Parent died too early");
/* Become the new root user inside our namespace */
r = reset_uid_gid();
if (r < 0)
return log_error_errno(r, "Couldn't become new root: %m");
/* Creating a new user namespace means all MS_SHARED mounts become MS_SLAVE. Let's put them
* back to MS_SHARED here, since that's what we want as defaults. (This will not reconnect
* propagation, but simply create new peer groups for all our mounts). */
r = mount_follow_verbose(LOG_ERR, NULL, "/", NULL, MS_SHARED|MS_REC, NULL);
if (r < 0)
return r;
}
r = mount_all(NULL,
arg_mount_settings | MOUNT_IN_USERNS,
arg_uid_shift,
arg_selinux_apifs_context);
if (r < 0)
return r;
if (!arg_network_namespace_path && arg_private_network) {
r = unshare(CLONE_NEWNET);
if (r < 0)
return log_error_errno(errno, "Failed to unshare network namespace: %m");
/* Tell the parent that it can setup network interfaces. */
(void) barrier_place(barrier); /* #3 */
}
r = mount_sysfs(NULL, arg_mount_settings);
if (r < 0)
return r;
/* Wait until we are cgroup-ified, so that we
* can mount the right cgroup path writable */
if (!barrier_place_and_sync(barrier)) /* #4 */
return log_error_errno(SYNTHETIC_ERRNO(ESRCH),
"Parent died too early");
if (arg_use_cgns) {
r = unshare(CLONE_NEWCGROUP);
if (r < 0)
return log_error_errno(errno, "Failed to unshare cgroup namespace: %m");
r = mount_cgroups(
"",
arg_unified_cgroup_hierarchy,
arg_userns_mode != USER_NAMESPACE_NO,
arg_uid_shift,
arg_uid_range,
arg_selinux_apifs_context,
true);
} else
r = mount_systemd_cgroup_writable("", arg_unified_cgroup_hierarchy);
if (r < 0)
return r;
r = setup_boot_id();
if (r < 0)
return r;
r = setup_kmsg(kmsg_socket);
if (r < 0)
return r;
kmsg_socket = safe_close(kmsg_socket);
r = mount_custom(
"/",
arg_custom_mounts,
arg_n_custom_mounts,
0,
arg_selinux_apifs_context,
MOUNT_NON_ROOT_ONLY | MOUNT_IN_USERNS);
if (r < 0)
return r;
if (setsid() < 0)
return log_error_errno(errno, "setsid() failed: %m");
if (arg_private_network)
(void) 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 (arg_console_mode != CONSOLE_PIPE) {
_cleanup_close_ int master = -1;
_cleanup_free_ char *console = NULL;
/* Allocate a pty and make it available as /dev/console. */
master = openpt_allocate(O_RDWR|O_NONBLOCK, &console);
if (master < 0)
return log_error_errno(master, "Failed to allocate a pty: %m");
r = setup_dev_console(console);
if (r < 0)
return log_error_errno(r, "Failed to set up /dev/console: %m");
r = send_one_fd(master_pty_socket, master, 0);
if (r < 0)
return log_error_errno(r, "Failed to send master fd: %m");
master_pty_socket = safe_close(master_pty_socket);
r = setup_stdio_as_dev_console();
if (r < 0)
return r;
}
r = patch_sysctl();
if (r < 0)
return r;
if (arg_oom_score_adjust_set) {
r = set_oom_score_adjust(arg_oom_score_adjust);
if (r < 0)
return log_error_errno(r, "Failed to adjust OOM score: %m");
}
if (arg_cpu_set.set)
if (sched_setaffinity(0, arg_cpu_set.allocated, arg_cpu_set.set) < 0)
return log_error_errno(errno, "Failed to set CPU affinity: %m");
(void) setup_hostname();
if (arg_personality != PERSONALITY_INVALID) {
r = safe_personality(arg_personality);
if (r < 0)
return log_error_errno(r, "personality() failed: %m");
} else if (secondary) {
r = safe_personality(PER_LINUX32);
if (r < 0)
return log_error_errno(r, "personality() failed: %m");
}
r = setrlimit_closest_all((const struct rlimit *const*) arg_rlimit, &which_failed);
if (r < 0)
return log_error_errno(r, "Failed to apply resource limit RLIMIT_%s: %m", rlimit_to_string(which_failed));
#if HAVE_SECCOMP
if (arg_seccomp) {
if (is_seccomp_available()) {
r = seccomp_load(arg_seccomp);
if (ERRNO_IS_SECCOMP_FATAL(r))
return log_error_errno(r, "Failed to install seccomp filter: %m");
if (r < 0)
log_debug_errno(r, "Failed to install seccomp filter: %m");
}
} else
#endif
{
r = setup_seccomp(arg_caps_retain, arg_syscall_allow_list, arg_syscall_deny_list);
if (r < 0)
return r;
}
#if HAVE_SELINUX
if (arg_selinux_context)
if (setexeccon(arg_selinux_context) < 0)
return log_error_errno(errno, "setexeccon(\"%s\") failed: %m", arg_selinux_context);
#endif
/* Make sure we keep the caps across the uid/gid dropping, so that we can retain some selected caps
* if we need to later on. */
if (prctl(PR_SET_KEEPCAPS, 1) < 0)
return log_error_errno(errno, "Failed to set PR_SET_KEEPCAPS: %m");
if (uid_is_valid(arg_uid) || gid_is_valid(arg_gid))
r = change_uid_gid_raw(arg_uid, arg_gid, arg_supplementary_gids, arg_n_supplementary_gids, arg_console_mode != CONSOLE_PIPE);
else
r = change_uid_gid(arg_user, arg_console_mode != CONSOLE_PIPE, &home);
if (r < 0)
return r;
r = drop_capabilities(getuid());
if (r < 0)
return log_error_errno(r, "Dropping capabilities failed: %m");
if (arg_no_new_privileges)
if (prctl(PR_SET_NO_NEW_PRIVS, 1, 0, 0, 0) < 0)
return log_error_errno(errno, "Failed to disable new privileges: %m");
/* LXC sets container=lxc, so follow the scheme here */
envp[n_env++] = strjoina("container=", arg_container_service_name);
envp[n_env] = strv_find_prefix(environ, "TERM=");
if (envp[n_env])
n_env++;
if (home || !uid_is_valid(arg_uid) || arg_uid == 0)
if (asprintf((char**)(envp + n_env++), "HOME=%s", home ?: "/root") < 0)
return log_oom();
if (arg_user || !uid_is_valid(arg_uid) || arg_uid == 0)
if (asprintf((char**)(envp + n_env++), "USER=%s", arg_user ?: "root") < 0 ||
asprintf((char**)(envp + n_env++), "LOGNAME=%s", arg_user ? arg_user : "root") < 0)
return log_oom();
assert(!sd_id128_is_null(arg_uuid));
if (asprintf((char**)(envp + n_env++), "container_uuid=%s", id128_to_uuid_string(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();
}
if (asprintf((char **)(envp + n_env++), "NOTIFY_SOCKET=%s", NSPAWN_NOTIFY_SOCKET_PATH) < 0)
return log_oom();
if (arg_n_credentials > 0) {
envp[n_env] = strdup("CREDENTIALS_DIRECTORY=/run/host/credentials");
if (!envp[n_env])
return log_oom();
n_env++;
}
env_use = strv_env_merge(3, envp, os_release_pairs, arg_setenv);
if (!env_use)
return log_oom();
/* Let the parent know that we are ready and
* wait until the parent is ready with the
* setup, too... */
if (!barrier_place_and_sync(barrier)) /* #5 */
return log_error_errno(SYNTHETIC_ERRNO(ESRCH), "Parent died too early");
if (arg_chdir)
if (chdir(arg_chdir) < 0)
return log_error_errno(errno, "Failed to change to specified working directory %s: %m", arg_chdir);
if (arg_start_mode == START_PID2) {
r = stub_pid1(arg_uuid);
if (r < 0)
return r;
}
if (arg_console_mode != CONSOLE_PIPE) {
/* So far our pty wasn't controlled by any process. Finally, it's time to change that, if we
* are configured for that. Acquire it as controlling tty. */
if (ioctl(STDIN_FILENO, TIOCSCTTY) < 0)
return log_error_errno(errno, "Failed to acquire controlling TTY: %m");
}
log_debug("Inner child completed, invoking payload.");
/* Now, explicitly close the log, so that we then can close all remaining fds. Closing the log explicitly first
* has the benefit that the logging subsystem knows about it, and is thus ready to be reopened should we need
* it again. Note that the other fds closed here are at least the locking and barrier fds. */
log_close();
log_set_open_when_needed(true);
(void) fdset_close_others(fds);
if (arg_start_mode == START_BOOT) {
char **a;
size_t m;
/* Automatically search for the init system */
m = strv_length(arg_parameters);
a = newa(char*, m + 2);
memcpy_safe(a + 1, arg_parameters, m * sizeof(char*));
a[1 + m] = NULL;
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);
exec_target = "/usr/lib/systemd/systemd, /lib/systemd/systemd, /sbin/init";
} else if (!strv_isempty(arg_parameters)) {
const char *dollar_path;
exec_target = arg_parameters[0];
/* Use the user supplied search $PATH if there is one, or DEFAULT_PATH_COMPAT if not to search the
* binary. */
dollar_path = strv_env_get(env_use, "PATH");
if (dollar_path) {
if (setenv("PATH", dollar_path, 1) < 0)
return log_error_errno(errno, "Failed to update $PATH: %m");
}
execvpe(arg_parameters[0], arg_parameters, env_use);
} else {
if (!arg_chdir)
/* If we cannot change the directory, we'll end up in /, that is expected. */
(void) chdir(home ?: "/root");
execle("/bin/bash", "-bash", NULL, env_use);
execle("/bin/sh", "-sh", NULL, env_use);
exec_target = "/bin/bash, /bin/sh";
}
return log_error_errno(errno, "execv(%s) failed: %m", exec_target);
}
static int setup_notify_child(void) {
_cleanup_close_ int fd = -1;
union sockaddr_union sa = {
.un.sun_family = AF_UNIX,
.un.sun_path = NSPAWN_NOTIFY_SOCKET_PATH,
};
int r;
fd = socket(AF_UNIX, SOCK_DGRAM|SOCK_CLOEXEC|SOCK_NONBLOCK, 0);
if (fd < 0)
return log_error_errno(errno, "Failed to allocate notification socket: %m");
(void) mkdir_parents(NSPAWN_NOTIFY_SOCKET_PATH, 0755);
(void) sockaddr_un_unlink(&sa.un);
r = bind(fd, &sa.sa, SOCKADDR_UN_LEN(sa.un));
if (r < 0)
return log_error_errno(errno, "bind(" NSPAWN_NOTIFY_SOCKET_PATH ") failed: %m");
r = userns_lchown(NSPAWN_NOTIFY_SOCKET_PATH, 0, 0);
if (r < 0)
return log_error_errno(r, "Failed to chown " NSPAWN_NOTIFY_SOCKET_PATH ": %m");
r = setsockopt_int(fd, SOL_SOCKET, SO_PASSCRED, true);