#include <sys/un.h>
#include <sys/socket.h>
#include <linux/netlink.h>
+#include <linux/rtnetlink.h>
#include <sys/eventfd.h>
+#include <net/if.h>
+#include <linux/veth.h>
+
#ifdef HAVE_SELINUX
#include <selinux/selinux.h>
#endif
+#ifdef HAVE_SECCOMP
+#include <seccomp.h>
+#endif
+
#include "sd-daemon.h"
#include "sd-bus.h"
#include "sd-id128.h"
+#include "sd-rtnl.h"
#include "log.h"
#include "util.h"
#include "mkdir.h"
#include "bus-kernel.h"
#include "env-util.h"
#include "def.h"
+#include "rtnl-util.h"
+#include "udev-util.h"
typedef enum LinkJournal {
LINK_NO,
static bool arg_quiet = false;
static bool arg_share_system = false;
static bool arg_register = true;
+static bool arg_keep_unit = false;
+static char **arg_network_interfaces = NULL;
+static bool arg_network_veth = false;
static int help(void) {
"Spawn a minimal namespace container for debugging, testing and building.\n\n"
" -h --help Show this help\n"
" --version Print version string\n"
+ " -q --quiet Do not show status information\n"
" -D --directory=NAME Root directory for the container\n"
" -b --boot Boot up full system (i.e. invoke init)\n"
" -u --user=USER Run the command under specified user or uid\n"
- " --uuid=UUID Set a specific machine UUID for the container\n"
" -M --machine=NAME Set the machine name for the container\n"
+ " --uuid=UUID Set a specific machine UUID for the container\n"
" -S --slice=SLICE Place the container in the specified slice\n"
+ " --private-network Disable network in container\n"
+ " --network-interface=INTERFACE\n"
+ " Assign an existing network interface to the\n"
+ " container\n"
+ " --network-veth Add a a virtual ethernet connection between host\n"
+ " and container\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"
- " --private-network Disable network in container\n"
- " --share-system Share system namespaces with host\n"
- " --read-only Mount the root directory read-only\n"
" --capability=CAP In addition to the default, retain specified\n"
" capability\n"
" --drop-capability=CAP Drop the specified capability from the default set\n"
" --link-journal=MODE Link up guest journal, one of no, auto, guest, host\n"
" -j Equivalent to --link-journal=host\n"
+ " --read-only Mount the root directory read-only\n"
" --bind=PATH[:PATH] Bind mount a file or directory from the host into\n"
" the container\n"
" --bind-ro=PATH[:PATH] Similar, but creates a read-only bind mount\n"
" --setenv=NAME=VALUE Pass an environment variable to PID 1\n"
+ " --share-system Share system namespaces with host\n"
" --register=BOOLEAN Register container as machine\n"
- " -q --quiet Do not show status information\n",
+ " --keep-unit Do not register a scope for the machine, reuse\n"
+ " the service unit nspawn is running in\n",
program_invocation_short_name);
return 0;
ARG_BIND_RO,
ARG_SETENV,
ARG_SHARE_SYSTEM,
- ARG_REGISTER
+ ARG_REGISTER,
+ ARG_KEEP_UNIT,
+ ARG_NETWORK_INTERFACE,
+ ARG_NETWORK_VETH,
};
static const struct option options[] = {
- { "help", no_argument, NULL, 'h' },
- { "version", no_argument, NULL, ARG_VERSION },
- { "directory", required_argument, NULL, 'D' },
- { "user", required_argument, NULL, 'u' },
- { "private-network", no_argument, NULL, ARG_PRIVATE_NETWORK },
- { "boot", no_argument, NULL, 'b' },
- { "uuid", required_argument, NULL, ARG_UUID },
- { "read-only", no_argument, NULL, ARG_READ_ONLY },
- { "capability", required_argument, NULL, ARG_CAPABILITY },
- { "drop-capability", required_argument, NULL, ARG_DROP_CAPABILITY },
- { "link-journal", required_argument, NULL, ARG_LINK_JOURNAL },
- { "bind", required_argument, NULL, ARG_BIND },
- { "bind-ro", required_argument, NULL, ARG_BIND_RO },
- { "machine", required_argument, NULL, 'M' },
- { "slice", required_argument, NULL, 'S' },
- { "setenv", required_argument, NULL, ARG_SETENV },
- { "selinux-context", required_argument, NULL, 'Z' },
- { "selinux-apifs-context", required_argument, NULL, 'L' },
- { "quiet", no_argument, NULL, 'q' },
- { "share-system", no_argument, NULL, ARG_SHARE_SYSTEM },
- { "register", required_argument, NULL, ARG_REGISTER },
+ { "help", no_argument, NULL, 'h' },
+ { "version", no_argument, NULL, ARG_VERSION },
+ { "directory", required_argument, NULL, 'D' },
+ { "user", required_argument, NULL, 'u' },
+ { "private-network", no_argument, NULL, ARG_PRIVATE_NETWORK },
+ { "boot", no_argument, NULL, 'b' },
+ { "uuid", required_argument, NULL, ARG_UUID },
+ { "read-only", no_argument, NULL, ARG_READ_ONLY },
+ { "capability", required_argument, NULL, ARG_CAPABILITY },
+ { "drop-capability", required_argument, NULL, ARG_DROP_CAPABILITY },
+ { "link-journal", required_argument, NULL, ARG_LINK_JOURNAL },
+ { "bind", required_argument, NULL, ARG_BIND },
+ { "bind-ro", required_argument, NULL, ARG_BIND_RO },
+ { "machine", required_argument, NULL, 'M' },
+ { "slice", required_argument, NULL, 'S' },
+ { "setenv", required_argument, NULL, ARG_SETENV },
+ { "selinux-context", required_argument, NULL, 'Z' },
+ { "selinux-apifs-context", required_argument, NULL, 'L' },
+ { "quiet", no_argument, NULL, 'q' },
+ { "share-system", no_argument, NULL, ARG_SHARE_SYSTEM },
+ { "register", required_argument, NULL, ARG_REGISTER },
+ { "keep-unit", no_argument, NULL, ARG_KEEP_UNIT },
+ { "network-interface", required_argument, NULL, ARG_NETWORK_INTERFACE },
+ { "network-veth", no_argument, NULL, ARG_NETWORK_VETH },
{}
};
int c, r;
+ uint64_t plus = 0, minus = 0;
assert(argc >= 0);
assert(argv);
break;
+ case ARG_NETWORK_VETH:
+ arg_network_veth = true;
+ arg_private_network = true;
+ break;
+
+ case ARG_NETWORK_INTERFACE:
+ if (strv_push(&arg_network_interfaces, optarg) < 0)
+ return log_oom();
+
+ /* fall through */
+
case ARG_PRIVATE_NETWORK:
arg_private_network = true;
break;
size_t length;
FOREACH_WORD_SEPARATOR(word, length, optarg, ",", state) {
+ _cleanup_free_ char *t;
cap_value_t cap;
- char *t;
t = strndup(word, length);
if (!t)
return log_oom();
- if (cap_from_name(t, &cap) < 0) {
- log_error("Failed to parse capability %s.", t);
- free(t);
- return -EINVAL;
+ if (streq(t, "all")) {
+ if (c == ARG_CAPABILITY)
+ plus = (uint64_t) -1;
+ else
+ minus = (uint64_t) -1;
+ } else {
+ if (cap_from_name(t, &cap) < 0) {
+ log_error("Failed to parse capability %s.", t);
+ return -EINVAL;
+ }
+
+ if (c == ARG_CAPABILITY)
+ plus |= 1ULL << (uint64_t) cap;
+ else
+ minus |= 1ULL << (uint64_t) cap;
}
-
- free(t);
-
- if (c == ARG_CAPABILITY)
- arg_retain |= 1ULL << (uint64_t) cap;
- else
- arg_retain &= ~(1ULL << (uint64_t) cap);
}
break;
arg_register = r;
break;
+ case ARG_KEEP_UNIT:
+ arg_keep_unit = true;
+ break;
+
case '?':
return -EINVAL;
return -EINVAL;
}
+ if (arg_keep_unit && cg_pid_get_owner_uid(0, NULL) >= 0) {
+ log_error("--keep-unit may not be used when invoked from a user session.");
+ return -EINVAL;
+ }
+
+ arg_retain = (arg_retain | plus | (arg_private_network ? 1ULL << CAP_NET_ADMIN : 0)) & ~minus;
+
return 1;
}
return r;
}
- r = sd_bus_call_method(
- bus,
- "org.freedesktop.machine1",
- "/org/freedesktop/machine1",
- "org.freedesktop.machine1.Manager",
- "CreateMachine",
- &error,
- NULL,
- "sayssusa(sv)",
- arg_machine,
- SD_BUS_MESSAGE_APPEND_ID128(arg_uuid),
- "nspawn",
- "container",
- (uint32_t) pid,
- strempty(arg_directory),
- !isempty(arg_slice), "Slice", "s", arg_slice);
+ if (arg_keep_unit) {
+ r = sd_bus_call_method(
+ bus,
+ "org.freedesktop.machine1",
+ "/org/freedesktop/machine1",
+ "org.freedesktop.machine1.Manager",
+ "RegisterMachine",
+ &error,
+ NULL,
+ "sayssus",
+ arg_machine,
+ SD_BUS_MESSAGE_APPEND_ID128(arg_uuid),
+ "nspawn",
+ "container",
+ (uint32_t) pid,
+ strempty(arg_directory));
+ } else {
+ r = sd_bus_call_method(
+ bus,
+ "org.freedesktop.machine1",
+ "/org/freedesktop/machine1",
+ "org.freedesktop.machine1.Manager",
+ "CreateMachine",
+ &error,
+ NULL,
+ "sayssusa(sv)",
+ arg_machine,
+ SD_BUS_MESSAGE_APPEND_ID128(arg_uuid),
+ "nspawn",
+ "container",
+ (uint32_t) pid,
+ strempty(arg_directory),
+ !isempty(arg_slice), "Slice", "s", arg_slice);
+ }
+
if (r < 0) {
log_error("Failed to register machine: %s", bus_error_message(&error, r));
return r;
return 0;
}
-static bool audit_enabled(void) {
- int fd;
+static int reset_audit_loginuid(void) {
+ _cleanup_free_ char *p = NULL;
+ int r;
- fd = socket(AF_NETLINK, SOCK_RAW, NETLINK_AUDIT);
- if (fd >= 0) {
- close_nointr_nofail(fd);
- return true;
+ if (arg_share_system)
+ return 0;
+
+ r = read_one_line_file("/proc/self/loginuid", &p);
+ if (r == -EEXIST)
+ return 0;
+ if (r < 0) {
+ log_error("Failed to read /proc/self/loginuid: %s", strerror(-r));
+ return r;
+ }
+
+ /* Already reset? */
+ if (streq(p, "4294967295"))
+ return 0;
+
+ r = write_string_file("/proc/self/loginuid", "4294967295");
+ if (r < 0) {
+ log_error("Failed to reset audit login UID. This probably means that your kernel is too\n"
+ "old and you have audit enabled. Note that the auditing subsystem is known to\n"
+ "be incompatible with containers on old kernels. Please make sure to upgrade\n"
+ "your kernel or to off auditing with 'audit=0' on the kernel command line before\n"
+ "using systemd-nspawn. Sleeping for 5s... (%s)\n", strerror(-r));
+
+ sleep(5);
}
- return false;
+
+ return 0;
+}
+
+static int setup_veth(int netns_fd) {
+ _cleanup_rtnl_message_unref_ sd_rtnl_message *m = NULL;
+ _cleanup_rtnl_unref_ sd_rtnl *rtnl = NULL;
+ char iface_name[IFNAMSIZ] = "ve-";
+ int r;
+
+ if (!arg_private_network)
+ return 0;
+
+ if (!arg_network_veth)
+ return 0;
+
+ strncpy(iface_name+3, arg_machine, sizeof(iface_name) - 3);
+
+ r = sd_rtnl_open(0, &rtnl);
+ if (r < 0) {
+ log_error("Failed to connect to netlink: %s", strerror(-r));
+ return r;
+ }
+
+ r = sd_rtnl_message_new_link(RTM_NEWLINK, 0, &m);
+ if (r < 0) {
+ log_error("Failed to allocate netlink message: %s", strerror(-r));
+ return r;
+ }
+
+ r = sd_rtnl_message_append_string(m, IFLA_IFNAME, "host0");
+ if (r < 0) {
+ log_error("Failed to append netlink kind: %s", strerror(-r));
+ return r;
+ }
+
+ r = sd_rtnl_message_open_container(m, IFLA_LINKINFO, 0);
+ if (r < 0) {
+ log_error("Failed to open netlink container: %s", strerror(-r));
+ return r;
+ }
+
+ r = sd_rtnl_message_append_string(m, IFLA_INFO_KIND, "veth");
+ if (r < 0) {
+ log_error("Failed to append netlink kind: %s", strerror(-r));
+ return r;
+ }
+
+ r = sd_rtnl_message_open_container(m, IFLA_INFO_DATA, 0);
+ if (r < 0) {
+ log_error("Failed to open netlink container: %s", strerror(-r));
+ return r;
+ }
+
+ r = sd_rtnl_message_open_container(m, VETH_INFO_PEER, sizeof(struct ifinfomsg));
+ if (r < 0) {
+ log_error("z Failed to open netlink container: %s", strerror(-r));
+ return r;
+ }
+
+ r = sd_rtnl_message_append_string(m, IFLA_IFNAME, iface_name);
+ if (r < 0) {
+ log_error("Failed to append netlink kind: %s", strerror(-r));
+ return r;
+ }
+
+ r = sd_rtnl_message_append_u32(m, IFLA_NET_NS_FD, netns_fd);
+ if (r < 0) {
+ log_error("Failed to add netlink namespace field: %s", strerror(-r));
+ return r;
+ }
+
+ r = sd_rtnl_message_close_container(m);
+ if (r < 0) {
+ log_error("Failed to close netlink container: %s", strerror(-r));
+ return r;
+ }
+
+ r = sd_rtnl_message_close_container(m);
+ if (r < 0) {
+ log_error("Failed to close netlink container: %s", strerror(-r));
+ return r;
+ }
+
+ r = sd_rtnl_message_close_container(m);
+ if (r < 0) {
+ log_error("Failed to close netlink container: %s", strerror(-r));
+ return r;
+ }
+
+ r = sd_rtnl_call(rtnl, m, 0, NULL);
+ if (r < 0) {
+ log_error("Failed to add new veth interfaces: %s", strerror(-r));
+ return r;
+ }
+
+ return 0;
+}
+
+static int move_network_interfaces(pid_t pid) {
+ _cleanup_udev_unref_ struct udev *udev = NULL;
+ _cleanup_rtnl_unref_ sd_rtnl *rtnl = NULL;
+ char **i;
+ int r;
+
+ if (!arg_private_network)
+ return 0;
+
+ if (strv_isempty(arg_network_interfaces))
+ return 0;
+
+ r = sd_rtnl_open(0, &rtnl);
+ if (r < 0) {
+ log_error("Failed to connect to netlink: %s", strerror(-r));
+ return r;
+ }
+
+ udev = udev_new();
+ if (!udev) {
+ log_error("Failed to connect to udev.");
+ return -ENOMEM;
+ }
+
+ STRV_FOREACH(i, arg_network_interfaces) {
+ _cleanup_rtnl_message_unref_ sd_rtnl_message *m = NULL;
+ _cleanup_udev_device_unref_ struct udev_device *d = NULL;
+ char ifi_str[2 + DECIMAL_STR_MAX(int)];
+ int ifi;
+
+ ifi = (int) if_nametoindex(*i);
+ if (ifi <= 0) {
+ log_error("Failed to resolve interface %s: %m", *i);
+ return -errno;
+ }
+
+ sprintf(ifi_str, "n%i", ifi);
+ d = udev_device_new_from_device_id(udev, ifi_str);
+ if (!d) {
+ log_error("Failed to get udev device for interface %s: %m", *i);
+ return -errno;
+ }
+
+ if (udev_device_get_is_initialized(d) <= 0) {
+ log_error("Network interface %s is not initialized yet.", *i);
+ return -EBUSY;
+ }
+
+ r = sd_rtnl_message_new_link(RTM_NEWLINK, ifi, &m);
+ if (r < 0) {
+ log_error("Failed to allocate netlink message: %s", strerror(-r));
+ return r;
+ }
+
+ r = sd_rtnl_message_append_u32(m, IFLA_NET_NS_PID, pid);
+ if (r < 0) {
+ log_error("Failed to append namespace PID to netlink message: %s", strerror(-r));
+ return r;
+ }
+
+ r = sd_rtnl_call(rtnl, m, 0, NULL);
+ if (r < 0) {
+ log_error("Failed to move interface %s to namespace: %s", *i, strerror(-r));
+ return r;
+ }
+ }
+
+ return 0;
+}
+
+static int audit_still_doesnt_work_in_containers(void) {
+
+#ifdef HAVE_SECCOMP
+ scmp_filter_ctx seccomp;
+ int r;
+
+ /*
+ Audit is broken in containers, much of the userspace audit
+ hookup will fail if running inside a container. We don't
+ care and just turn off creation of audit sockets.
+
+ This will make socket(AF_NETLINK, *, NETLINK_AUDIT) fail
+ with EAFNOSUPPORT which audit userspace uses as indication
+ that audit is disabled in the kernel.
+ */
+
+ seccomp = seccomp_init(SCMP_ACT_ALLOW);
+ if (!seccomp)
+ return log_oom();
+
+ r = seccomp_rule_add_exact(
+ seccomp,
+ SCMP_ACT_ERRNO(EAFNOSUPPORT),
+ SCMP_SYS(socket),
+ 2,
+ SCMP_A0(SCMP_CMP_EQ, AF_NETLINK),
+ SCMP_A2(SCMP_CMP_EQ, NETLINK_AUDIT));
+ if (r < 0) {
+ log_error("Failed to add audit seccomp rule: %s", strerror(-r));
+ goto finish;
+ }
+
+ r = seccomp_attr_set(seccomp, SCMP_FLTATR_CTL_NNP, 0);
+ if (r < 0) {
+ log_error("Failed to unset NO_NEW_PRIVS: %s", strerror(-r));
+ goto finish;
+ }
+
+ r = seccomp_load(seccomp);
+ if (r < 0)
+ log_error("Failed to install seccomp audit filter: %s", strerror(-r));
+
+finish:
+ seccomp_release(seccomp);
+ return r;
+#else
+ return 0;
+#endif
+
}
int main(int argc, char *argv[]) {
- pid_t pid = 0;
+
+ _cleanup_close_ int master = -1, kdbus_fd = -1, sync_fd = -1, netns_fd = -1;
+ _cleanup_close_pipe_ int kmsg_socket_pair[2] = { -1, -1 };
+ _cleanup_free_ char *kdbus_domain = NULL;
+ _cleanup_fdset_free_ FDSet *fds = NULL;
+ const char *console = NULL;
int r = EXIT_FAILURE, k;
- _cleanup_close_ int master = -1, kdbus_fd = -1, sync_fd = -1;
int n_fd_passed;
- const char *console = NULL;
+ pid_t pid = 0;
sigset_t mask;
- _cleanup_close_pipe_ int kmsg_socket_pair[2] = { -1, -1 };
- _cleanup_fdset_free_ FDSet *fds = NULL;
- _cleanup_free_ char *kdbus_domain = NULL;
log_parse_environment();
log_open();
goto finish;
}
- if (arg_boot && audit_enabled()) {
- log_warning("The kernel auditing subsystem is known to be incompatible with containers.\n"
- "Please make sure to turn off auditing with 'audit=0' on the kernel command\n"
- "line before using systemd-nspawn. Sleeping for 5s...\n");
- sleep(5);
- }
-
if (path_equal(arg_directory, "/")) {
log_error("Spawning container on root directory not supported.");
goto finish;
goto finish;
}
+ if (arg_network_veth) {
+ netns_fd = open("/proc/self/ns/net", O_RDWR|O_CLOEXEC);
+ if (netns_fd < 0) {
+ log_error("Failed to open network namespace fd: %m");
+ goto finish;
+ }
+ }
if (access("/dev/kdbus/control", F_OK) >= 0) {
goto child_fail;
}
+ if (reset_audit_loginuid() < 0)
+ goto child_fail;
+
if (prctl(PR_SET_PDEATHSIG, SIGKILL) < 0) {
log_error("PR_SET_PDEATHSIG failed: %m");
goto child_fail;
dev_setup(arg_directory);
+ if (setup_veth(netns_fd) < 0)
+ goto child_fail;
+
+ if (netns_fd >= 0) {
+ close_nointr_nofail(netns_fd);
+ netns_fd = -1;
+ }
+
+ if (audit_still_doesnt_work_in_containers() < 0)
+ goto child_fail;
+
if (setup_dev_console(arg_directory, console) < 0)
goto child_fail;
if (r < 0)
goto finish;
+ r = move_network_interfaces(pid);
+ if (r < 0)
+ goto finish;
+
eventfd_write(sync_fd, 1);
close_nointr_nofail(sync_fd);
sync_fd = -1;
free(arg_directory);
free(arg_machine);
free(arg_setenv);
+ free(arg_network_interfaces);
return r;
}