X-Git-Url: https://www.chiark.greenend.org.uk/ucgi/~ianmdlvl/git?p=elogind.git;a=blobdiff_plain;f=src%2Fnspawn%2Fnspawn.c;h=92b67286762350c9b53fe555a404dd8f74235e3c;hp=ed13e813a7a293e2e433eb8bec46fc8d3989e38e;hb=13e8ceb84e56907d73b6b07418deb37faaf0e66d;hpb=1c03020cc48d41c94a51b5efbeeccb96e45d3e26 diff --git a/src/nspawn/nspawn.c b/src/nspawn/nspawn.c index ed13e813a..92b672867 100644 --- a/src/nspawn/nspawn.c +++ b/src/nspawn/nspawn.c @@ -41,13 +41,22 @@ #include #include #include -#if HAVE_SELINUX +#include +#include +#include + +#ifdef HAVE_SELINUX #include #endif +#ifdef HAVE_SECCOMP +#include +#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" @@ -68,6 +77,12 @@ #include "bus-kernel.h" #include "env-util.h" #include "def.h" +#include "rtnl-util.h" +#include "udev-util.h" + +#ifdef HAVE_SECCOMP +#include "seccomp-util.h" +#endif typedef enum LinkJournal { LINK_NO, @@ -80,8 +95,8 @@ static char *arg_directory = NULL; static char *arg_user = NULL; static sd_id128_t arg_uuid = {}; static char *arg_machine = NULL; -static char *arg_process_label = NULL; -static char *arg_file_label = NULL; +static const char *arg_selinux_context = NULL; +static const char *arg_selinux_apifs_context = NULL; static const char *arg_slice = NULL; static bool arg_private_network = false; static bool arg_read_only = false; @@ -117,6 +132,15 @@ static uint64_t arg_retain = static char **arg_bind = NULL; static char **arg_bind_ro = NULL; static char **arg_setenv = NULL; +static bool arg_quiet = false; +static bool arg_share_system = false; +static bool arg_register = true; +static bool arg_keep_unit = false; +static char **arg_network_interfaces = NULL; +static char **arg_network_macvlan = NULL; +static bool arg_network_veth = false; +static const char *arg_network_bridge = NULL; +static unsigned long arg_personality = 0xffffffffLU; static int help(void) { @@ -124,27 +148,46 @@ 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" - " -L --file-label=LABEL Set the MAC file label to be used by tmpfs file\n" - " systems in the container\n" - " -Z --process-label=LABEL Set the MAC label to be used by processes in\n" - " the container\n" " --private-network Disable network in container\n" - " --read-only Mount the root directory read-only\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-veth Add a virtual ethernet connection between host\n" + " and container\n" + " --network-bridge=INTERFACE\n" + " Add a virtual ethernet connection between host\n" + " and container and add it to an existing bridge on\n" + " the host\n" + " -Z --selinux-context=SECLABEL\n" + " Set the SELinux security context to be used by\n" + " processes in the container\n" + " -L --selinux-apifs-context=SECLABEL\n" + " Set the SELinux security context to be used by\n" + " API/tmpfs file systems in the container\n" " --capability=CAP In addition to the default, retain specified\n" " capability\n" " --drop-capability=CAP Drop the specified capability from the default set\n" " --link-journal=MODE Link up guest journal, one of no, auto, guest, host\n" " -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", + " --setenv=NAME=VALUE Pass an environment variable to PID 1\n" + " --share-system Share system namespaces with host\n" + " --register=BOOLEAN Register container as machine\n" + " --keep-unit Do not register a scope for the machine, reuse\n" + " the service unit nspawn is running in\n", program_invocation_short_name); return 0; @@ -163,36 +206,54 @@ static int parse_argv(int argc, char *argv[]) { ARG_BIND, ARG_BIND_RO, ARG_SETENV, + ARG_SHARE_SYSTEM, + ARG_REGISTER, + ARG_KEEP_UNIT, + ARG_NETWORK_INTERFACE, + ARG_NETWORK_MACVLAN, + ARG_NETWORK_VETH, + ARG_NETWORK_BRIDGE, + ARG_PERSONALITY, }; 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 }, - { "process-label", required_argument, NULL, 'Z' }, - { "file-label", required_argument, NULL, 'L' }, + { "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-macvlan", required_argument, NULL, ARG_NETWORK_MACVLAN }, + { "network-veth", no_argument, NULL, ARG_NETWORK_VETH }, + { "network-bridge", required_argument, NULL, ARG_NETWORK_BRIDGE }, + { "personality", required_argument, NULL, ARG_PERSONALITY }, {} }; int c, r; + uint64_t plus = 0, minus = 0; assert(argc >= 0); assert(argv); - while ((c = getopt_long(argc, argv, "+hD:u:bL:M:jS:Z:", options, NULL)) >= 0) { + while ((c = getopt_long(argc, argv, "+hD:u:bL:M:jS:Z:q", options, NULL)) >= 0) { switch (c) { @@ -222,6 +283,29 @@ static int parse_argv(int argc, char *argv[]) { break; + case ARG_NETWORK_BRIDGE: + arg_network_bridge = optarg; + + /* fall through */ + + case ARG_NETWORK_VETH: + arg_network_veth = true; + arg_private_network = true; + break; + + case ARG_NETWORK_INTERFACE: + if (strv_extend(&arg_network_interfaces, optarg) < 0) + return log_oom(); + + arg_private_network = true; + break; + + case ARG_NETWORK_MACVLAN: + if (strv_extend(&arg_network_macvlan, optarg) < 0) + return log_oom(); + + /* fall through */ + case ARG_PRIVATE_NETWORK: arg_private_network = true; break; @@ -239,31 +323,34 @@ static int parse_argv(int argc, char *argv[]) { break; case 'S': - arg_slice = strdup(optarg); - if (!arg_slice) - return log_oom(); - + arg_slice = optarg; break; case 'M': - if (!hostname_is_valid(optarg)) { - log_error("Invalid machine name: %s", optarg); - return -EINVAL; - } + if (isempty(optarg)) { + free(arg_machine); + arg_machine = NULL; + } else { - free(arg_machine); - arg_machine = strdup(optarg); - if (!arg_machine) - return log_oom(); + if (!hostname_is_valid(optarg)) { + log_error("Invalid machine name: %s", optarg); + return -EINVAL; + } - break; + free(arg_machine); + arg_machine = strdup(optarg); + if (!arg_machine) + return log_oom(); - case 'L': - arg_file_label = optarg; - break; + break; + } case 'Z': - arg_process_label = optarg; + arg_selinux_context = optarg; + break; + + case 'L': + arg_selinux_apifs_context = optarg; break; case ARG_READ_ONLY: @@ -276,25 +363,29 @@ static int parse_argv(int argc, char *argv[]) { 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; @@ -373,6 +464,38 @@ static int parse_argv(int argc, char *argv[]) { break; } + case 'q': + arg_quiet = true; + break; + + case ARG_SHARE_SYSTEM: + arg_share_system = true; + break; + + case ARG_REGISTER: + r = parse_boolean(optarg); + if (r < 0) { + log_error("Failed to parse --register= argument: %s", optarg); + return r; + } + + arg_register = r; + break; + + case ARG_KEEP_UNIT: + arg_keep_unit = true; + break; + + case ARG_PERSONALITY: + + arg_personality = personality_from_string(optarg); + if (arg_personality == 0xffffffffLU) { + log_error("Unknown or unsupported personality '%s'.", optarg); + return -EINVAL; + } + + break; + case '?': return -EINVAL; @@ -381,6 +504,21 @@ static int parse_argv(int argc, char *argv[]) { } } + if (arg_share_system) + arg_register = false; + + if (arg_boot && arg_share_system) { + log_error("--boot and --share-system may not be combined."); + return -EINVAL; + } + + if (arg_keep_unit && cg_pid_get_owner_uid(0, NULL) >= 0) { + log_error("--keep-unit may not be used when invoked from a user session."); + return -EINVAL; + } + + arg_retain = (arg_retain | plus | (arg_private_network ? 1ULL << CAP_NET_ADMIN : 0)) & ~minus; + return 1; } @@ -442,8 +580,9 @@ static int mount_all(const char *dest) { mkdir_p(where, 0755); #ifdef HAVE_SELINUX - if (arg_file_label && (streq_ptr(mount_table[k].what, "tmpfs") || streq_ptr(mount_table[k].what, "devpts"))) { - options = strjoin(mount_table[k].options, ",context=\"", arg_file_label, "\"", NULL); + if (arg_selinux_apifs_context && + (streq_ptr(mount_table[k].what, "tmpfs") || streq_ptr(mount_table[k].what, "devpts"))) { + options = strjoin(mount_table[k].options, ",context=\"", arg_selinux_apifs_context, "\"", NULL); if (!options) return log_oom(); @@ -612,12 +751,15 @@ static int setup_resolv_conf(const char *dest) { static int setup_boot_id(const char *dest) { _cleanup_free_ char *from = NULL, *to = NULL; - sd_id128_t rnd; + sd_id128_t rnd = {}; char as_uuid[37]; int r; assert(dest); + if (arg_share_system) + return 0; + /* Generate a new randomized boot ID, so that each boot-up of * the container gets a new one */ @@ -843,6 +985,9 @@ static int setup_kmsg(const char *dest, int kmsg_socket) { static int setup_hostname(void) { + if (arg_share_system) + return 0; + if (sethostname(arg_machine, strlen(arg_machine)) < 0) return -errno; @@ -1025,28 +1170,114 @@ static int register_machine(pid_t pid) { _cleanup_bus_unref_ sd_bus *bus = NULL; int r; + if (!arg_register) + return 0; + r = sd_bus_default_system(&bus); if (r < 0) { log_error("Failed to open system bus: %s", strerror(-r)); 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 { + _cleanup_bus_message_unref_ sd_bus_message *m = NULL; + + r = sd_bus_message_new_method_call( + bus, + &m, + "org.freedesktop.machine1", + "/org/freedesktop/machine1", + "org.freedesktop.machine1.Manager", + "CreateMachine"); + if (r < 0) { + log_error("Failed to create message: %s", strerror(-r)); + return r; + } + + r = sd_bus_message_append( + m, + "sayssus", + arg_machine, + SD_BUS_MESSAGE_APPEND_ID128(arg_uuid), + "nspawn", + "container", + (uint32_t) pid, + strempty(arg_directory)); + if (r < 0) { + log_error("Failed to append message arguments: %s", strerror(-r)); + return r; + } + + r = sd_bus_message_open_container(m, 'a', "(sv)"); + if (r < 0) { + log_error("Failed to open container: %s", strerror(-r)); + return r; + } + + if (!isempty(arg_slice)) { + r = sd_bus_message_append(m, "(sv)", "Slice", "s", arg_slice); + if (r < 0) { + log_error("Failed to append slice: %s", strerror(-r)); + return r; + } + } + + r = sd_bus_message_append(m, "(sv)", "DevicePolicy", "s", "strict"); + if (r < 0) { + log_error("Failed to add device policy: %s", strerror(-r)); + return r; + } + + r = sd_bus_message_append(m, "(sv)", "DeviceAllow", "a(ss)", 8, + /* Allow the container to + * access and create the API + * device nodes, so that + * PrivateDevices= in the + * container can work + * fine */ + "/dev/null", "rwm", + "/dev/zero", "rwm", + "/dev/full", "rwm", + "/dev/random", "rwm", + "/dev/urandom", "rwm", + "/dev/tty", "rwm", + /* Allow the container + * access to ptys. However, + * do not permit the + * container to ever create + * these device nodes. */ + "/dev/pts/ptmx", "rw", + "char-pts", "rw"); + if (r < 0) { + log_error("Failed to add device whitelist: %s", strerror(-r)); + return r; + } + + r = sd_bus_message_close_container(m); + if (r < 0) { + log_error("Failed to close container: %s", strerror(-r)); + return r; + } + + r = sd_bus_call(bus, m, 0, &error, NULL); + } + if (r < 0) { log_error("Failed to register machine: %s", bus_error_message(&error, r)); return r; @@ -1062,6 +1293,9 @@ static int terminate_machine(pid_t pid) { const char *path; int r; + if (!arg_register) + return 0; + r = sd_bus_default_system(&bus); if (r < 0) { log_error("Failed to open system bus: %s", strerror(-r)); @@ -1107,28 +1341,451 @@ static int terminate_machine(pid_t pid) { return 0; } -static bool audit_enabled(void) { - int fd; +static int reset_audit_loginuid(void) { + _cleanup_free_ char *p = NULL; + int r; + + if (arg_share_system) + return 0; + + r = read_one_line_file("/proc/self/loginuid", &p); + if (r == -ENOENT) + return 0; + if (r < 0) { + 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 0; +} + +static int setup_veth(pid_t pid, char iface_name[IFNAMSIZ]) { + _cleanup_rtnl_message_unref_ sd_rtnl_message *m = NULL; + _cleanup_rtnl_unref_ sd_rtnl *rtnl = NULL; + int r; + + if (!arg_private_network) + return 0; + + if (!arg_network_veth) + return 0; + + /* Use two different interface name prefixes depending whether + * we are in bridge mode or not. */ + if (arg_network_bridge) + memcpy(iface_name, "vb-", 3); + else + memcpy(iface_name, "ve-", 3); + + strncpy(iface_name+3, arg_machine, IFNAMSIZ - 3); + + r = sd_rtnl_open(&rtnl, 0); + if (r < 0) { + log_error("Failed to connect to netlink: %s", strerror(-r)); + return r; + } + + r = sd_rtnl_message_new_link(rtnl, &m, RTM_NEWLINK, 0); + if (r < 0) { + log_error("Failed to allocate netlink message: %s", strerror(-r)); + return r; + } + + r = sd_rtnl_message_append_string(m, IFLA_IFNAME, iface_name); + if (r < 0) { + log_error("Failed to add netlink interface name: %s", strerror(-r)); + return r; + } + + r = sd_rtnl_message_open_container(m, IFLA_LINKINFO); + 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); + 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); + if (r < 0) { + log_error("Failed to open netlink container: %s", strerror(-r)); + return r; + } + + r = sd_rtnl_message_append_string(m, IFLA_IFNAME, "host0"); + if (r < 0) { + log_error("Failed to add netlink interface name: %s", strerror(-r)); + return r; + } + + r = sd_rtnl_message_append_u32(m, IFLA_NET_NS_PID, pid); + 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 setup_bridge(const char veth_name[]) { + _cleanup_rtnl_message_unref_ sd_rtnl_message *m = NULL; + _cleanup_rtnl_unref_ sd_rtnl *rtnl = NULL; + int r, bridge; + + if (!arg_private_network) + return 0; + + if (!arg_network_veth) + return 0; + + if (!arg_network_bridge) + return 0; + + bridge = (int) if_nametoindex(arg_network_bridge); + if (bridge <= 0) { + log_error("Failed to resolve interface %s: %m", arg_network_bridge); + return -errno; + } + + r = sd_rtnl_open(&rtnl, 0); + if (r < 0) { + log_error("Failed to connect to netlink: %s", strerror(-r)); + return r; + } + + r = sd_rtnl_message_new_link(rtnl, &m, RTM_SETLINK, 0); + if (r < 0) { + log_error("Failed to allocate netlink message: %s", strerror(-r)); + return r; + } + + r = sd_rtnl_message_append_string(m, IFLA_IFNAME, veth_name); + if (r < 0) { + log_error("Failed to add netlink interface name field: %s", strerror(-r)); + return r; + } + + r = sd_rtnl_message_append_u32(m, IFLA_MASTER, bridge); + if (r < 0) { + log_error("Failed to add netlink master field: %s", strerror(-r)); + return r; + } + + r = sd_rtnl_call(rtnl, m, 0, NULL); + if (r < 0) { + log_error("Failed to add veth interface to bridge: %s", strerror(-r)); + return r; + } + + return 0; +} + +static int parse_interface(struct udev *udev, const char *name) { + _cleanup_udev_device_unref_ struct udev_device *d = NULL; + char ifi_str[2 + DECIMAL_STR_MAX(int)]; + int ifi; + + ifi = (int) if_nametoindex(name); + if (ifi <= 0) { + log_error("Failed to resolve interface %s: %m", name); + return -errno; + } - fd = socket(AF_NETLINK, SOCK_RAW, NETLINK_AUDIT); - if (fd >= 0) { - close_nointr_nofail(fd); - return true; + 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", name); + return -errno; } - return false; + + if (udev_device_get_is_initialized(d) <= 0) { + log_error("Network interface %s is not initialized yet.", name); + return -EBUSY; + } + + return ifi; +} + +static int move_network_interfaces(pid_t pid) { + _cleanup_udev_unref_ struct udev *udev = NULL; + _cleanup_rtnl_unref_ sd_rtnl *rtnl = NULL; + char **i; + int r; + + if (!arg_private_network) + return 0; + + if (strv_isempty(arg_network_interfaces)) + return 0; + + r = sd_rtnl_open(&rtnl, 0); + if (r < 0) { + 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; + int ifi; + + ifi = parse_interface(udev, *i); + if (ifi < 0) + return ifi; + + r = sd_rtnl_message_new_link(rtnl, &m, RTM_NEWLINK, ifi); + 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 setup_macvlan(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_macvlan)) + return 0; + + r = sd_rtnl_open(&rtnl, 0); + 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_macvlan) { + _cleanup_rtnl_message_unref_ sd_rtnl_message *m = NULL; + _cleanup_free_ char *n = NULL; + int ifi; + + ifi = parse_interface(udev, *i); + if (ifi < 0) + return ifi; + + r = sd_rtnl_message_new_link(rtnl, &m, RTM_NEWLINK, 0); + if (r < 0) { + log_error("Failed to allocate netlink message: %s", strerror(-r)); + return r; + } + + r = sd_rtnl_message_append_u32(m, IFLA_LINK, ifi); + if (r < 0) { + log_error("Failed to add netlink interface index: %s", strerror(-r)); + return r; + } + + n = strappend("mv-", *i); + if (!n) + return log_oom(); + + strshorten(n, IFNAMSIZ-1); + + r = sd_rtnl_message_append_string(m, IFLA_IFNAME, n); + if (r < 0) { + log_error("Failed to add netlink interface name: %s", strerror(-r)); + return r; + } + + r = sd_rtnl_message_append_u32(m, IFLA_NET_NS_PID, pid); + if (r < 0) { + log_error("Failed to add netlink namespace field: %s", strerror(-r)); + return r; + } + + r = sd_rtnl_message_open_container(m, IFLA_LINKINFO); + 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, "macvlan"); + 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); + if (r < 0) { + log_error("Failed to open netlink container: %s", strerror(-r)); + return r; + } + + r = sd_rtnl_message_append_u32(m, IFLA_MACVLAN_MODE, MACVLAN_MODE_BRIDGE); + if (r < 0) { + log_error("Failed to append macvlan mode: %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 macvlan interfaces: %s", 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_add_secondary_archs(seccomp); + if (r < 0) { + log_error("Failed to add secondary archs to seccomp filter: %s", strerror(-r)); + goto finish; + } + + r = seccomp_rule_add( + seccomp, + SCMP_ACT_ERRNO(EAFNOSUPPORT), + SCMP_SYS(socket), + 2, + SCMP_A0(SCMP_CMP_EQ, AF_NETLINK), + SCMP_A2(SCMP_CMP_EQ, NETLINK_AUDIT)); + if (r < 0) { + log_error("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; - int r = EXIT_FAILURE, k; + _cleanup_close_ int master = -1, kdbus_fd = -1, sync_fd = -1; - int n_fd_passed; - const char *console = NULL; - 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; - const char *ns; + _cleanup_fdset_free_ FDSet *fds = NULL; + const char *console = NULL; + int r = EXIT_FAILURE, k; + int n_fd_passed; + pid_t pid = 0; + sigset_t mask; + char veth_name[IFNAMSIZ]; log_parse_environment(); log_open(); @@ -1181,21 +1838,26 @@ int main(int argc, char *argv[]) { 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; } - if (path_is_os_tree(arg_directory) <= 0) { - log_error("Directory %s doesn't look like an OS root directory (/etc/os-release is missing). Refusing.", arg_directory); - goto finish; + if (arg_boot) { + if (path_is_os_tree(arg_directory) <= 0) { + log_error("Directory %s doesn't look like an OS root directory (/etc/os-release is missing). Refusing.", arg_directory); + goto finish; + } + } else { + const char *p; + + p = strappenda(arg_directory, + argc > optind && path_is_absolute(argv[optind]) ? argv[optind] : "/usr/bin/"); + if (access(p, F_OK) < 0) { + log_error("Directory %s lacks the binary to execute or doesn't look like a binary tree. Refusing.", arg_directory); + goto finish; + + } } log_close(); @@ -1222,19 +1884,33 @@ int main(int argc, char *argv[]) { goto finish; } - log_info("Spawning container %s on %s. Press ^] three times within 1s to abort execution.", arg_machine, arg_directory); + if (!arg_quiet) + log_info("Spawning container %s on %s. Press ^] three times within 1s to abort execution.", arg_machine, arg_directory); if (unlockpt(master) < 0) { log_error("Failed to unlock tty: %m"); goto finish; } - ns = strappenda("machine-", arg_machine); - kdbus_fd = bus_kernel_create_domain(ns, &kdbus_domain); - if (r < 0) - log_debug("Failed to create kdbus domain: %s", strerror(-r)); - else - log_debug("Successfully created kdbus domain as %s", kdbus_domain); + if (access("/dev/kdbus/control", F_OK) >= 0) { + + if (arg_share_system) { + kdbus_domain = strdup("/dev/kdbus"); + if (!kdbus_domain) { + log_oom(); + goto finish; + } + } else { + const char *ns; + + ns = strappenda("machine-", arg_machine); + kdbus_fd = bus_kernel_create_domain(ns, &kdbus_domain); + if (r < 0) + log_debug("Failed to create kdbus domain: %s", strerror(-r)); + else + log_debug("Successfully created kdbus domain as %s", kdbus_domain); + } + } if (socketpair(AF_UNIX, SOCK_DGRAM|SOCK_NONBLOCK|SOCK_CLOEXEC, 0, kmsg_socket_pair) < 0) { log_error("Failed to create kmsg socket pair: %m"); @@ -1256,7 +1932,10 @@ int main(int argc, char *argv[]) { goto finish; } - pid = syscall(__NR_clone, SIGCHLD|CLONE_NEWIPC|CLONE_NEWNS|CLONE_NEWPID|CLONE_NEWUTS|(arg_private_network ? CLONE_NEWNET : 0), NULL); + pid = syscall(__NR_clone, + SIGCHLD|CLONE_NEWNS| + (arg_share_system ? 0 : CLONE_NEWIPC|CLONE_NEWPID|CLONE_NEWUTS)| + (arg_private_network ? CLONE_NEWNET : 0), NULL); if (pid < 0) { if (errno == EINVAL) log_error("clone() failed, do you have namespace support enabled in your kernel? (You need UTS, IPC, PID and NET namespacing built in): %m"); @@ -1328,6 +2007,9 @@ int main(int argc, char *argv[]) { 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; @@ -1364,6 +2046,9 @@ int main(int argc, char *argv[]) { dev_setup(arg_directory); + if (audit_still_doesnt_work_in_containers() < 0) + goto child_fail; + if (setup_dev_console(arg_directory, console) < 0) goto child_fail; @@ -1416,7 +2101,8 @@ int main(int argc, char *argv[]) { umask(0022); - loopback_setup(); + if (arg_private_network) + loopback_setup(); if (drop_capabilities() < 0) { log_error("drop_capabilities() failed: %m"); @@ -1509,6 +2195,13 @@ int main(int argc, char *argv[]) { setup_hostname(); + if (arg_personality != 0xffffffffLU) { + if (personality(arg_personality) < 0) { + log_error("personality() failed: %m"); + goto child_fail; + } + } + eventfd_read(sync_fd, &x); close_nointr_nofail(sync_fd); sync_fd = -1; @@ -1526,10 +2219,10 @@ int main(int argc, char *argv[]) { } else env_use = (char**) envp; -#if HAVE_SELINUX - if (arg_process_label) - if (setexeccon(arg_process_label) < 0) - log_error("setexeccon(\"%s\") failed: %m", arg_process_label); +#ifdef HAVE_SELINUX + if (arg_selinux_context) + if (setexeccon((security_context_t) arg_selinux_context) < 0) + log_error("setexeccon(\"%s\") failed: %m", arg_selinux_context); #endif if (arg_boot) { char **a; @@ -1554,6 +2247,7 @@ int main(int argc, char *argv[]) { else { chdir(home ? home : "/root"); execle("/bin/bash", "-bash", NULL, env_use); + execle("/bin/sh", "-sh", NULL, env_use); } log_error("execv() failed: %m"); @@ -1569,6 +2263,22 @@ int main(int argc, char *argv[]) { if (r < 0) goto finish; + r = move_network_interfaces(pid); + if (r < 0) + goto finish; + + r = setup_veth(pid, veth_name); + if (r < 0) + goto finish; + + r = setup_bridge(veth_name); + if (r < 0) + goto finish; + + r = setup_macvlan(pid); + if (r < 0) + goto finish; + eventfd_write(sync_fd, 1); close_nointr_nofail(sync_fd); sync_fd = -1; @@ -1579,7 +2289,8 @@ int main(int argc, char *argv[]) { break; } - putc('\n', stdout); + if (!arg_quiet) + putc('\n', stdout); /* Kill if it is not dead yet anyway */ terminate_machine(pid); @@ -1602,21 +2313,26 @@ int main(int argc, char *argv[]) { break; } - log_debug("Container %s exited successfully.", arg_machine); + if (!arg_quiet) + log_debug("Container %s exited successfully.", arg_machine); break; } else if (status.si_code == CLD_KILLED && status.si_status == SIGINT) { - log_info("Container %s has been shut down.", arg_machine); + + if (!arg_quiet) + log_info("Container %s has been shut down.", arg_machine); r = 0; break; } else if (status.si_code == CLD_KILLED && status.si_status == SIGHUP) { - log_info("Container %s is being rebooted.", arg_machine); + + if (!arg_quiet) + log_info("Container %s is being rebooted.", arg_machine); continue; } else if (status.si_code == CLD_KILLED || status.si_code == CLD_DUMPED) { - log_error("Container %s terminated by signal %s.", arg_machine, signal_to_string(status.si_status)); + log_error("Container %s terminated by signal %s.", arg_machine, signal_to_string(status.si_status)); r = EXIT_FAILURE; break; } else { @@ -1632,7 +2348,12 @@ finish: free(arg_directory); free(arg_machine); - free(arg_setenv); + free(arg_user); + strv_free(arg_setenv); + strv_free(arg_network_interfaces); + strv_free(arg_network_macvlan); + strv_free(arg_bind); + strv_free(arg_bind_ro); return r; }