X-Git-Url: https://www.chiark.greenend.org.uk/ucgi/~ianmdlvl/git?p=elogind.git;a=blobdiff_plain;f=src%2Fnspawn%2Fnspawn.c;h=379ea92355e6f512c15b06ad04fcf3b66fd63f7e;hp=97ef6c799d4281705e97e4533c6f60feb7bda604;hb=3302da4667640ac130956d8d90be7c39fb39222f;hpb=89f7c8465cd1ab37347dd0c15920bce31e8225df diff --git a/src/nspawn/nspawn.c b/src/nspawn/nspawn.c index 97ef6c799..379ea9235 100644 --- a/src/nspawn/nspawn.c +++ b/src/nspawn/nspawn.c @@ -41,13 +41,21 @@ #include #include #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 +76,12 @@ #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, @@ -121,6 +135,9 @@ 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 char *arg_network_bridge = NULL; static int help(void) { @@ -128,34 +145,43 @@ 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 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" - " --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" " --keep-unit Do not register a scope for the machine, reuse\n" - " the service unit nspawn is running in\n" - " -q --quiet Do not show status information\n", + " the service unit nspawn is running in\n", program_invocation_short_name); return 0; @@ -176,36 +202,43 @@ static int parse_argv(int argc, char *argv[]) { ARG_SETENV, ARG_SHARE_SYSTEM, ARG_REGISTER, - ARG_KEEP_UNIT + ARG_KEEP_UNIT, + ARG_NETWORK_INTERFACE, + ARG_NETWORK_VETH, + ARG_NETWORK_BRIDGE, }; 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 }, - { "keep-unit", no_argument, NULL, ARG_KEEP_UNIT }, + { "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 }, + { "network-bridge", required_argument, NULL, ARG_NETWORK_BRIDGE }, {} }; int c, r; + uint64_t plus = 0, minus = 0; assert(argc >= 0); assert(argv); @@ -240,6 +273,24 @@ static int parse_argv(int argc, char *argv[]) { break; + case ARG_NETWORK_BRIDGE: + arg_network_bridge = strdup(optarg); + if (!arg_network_bridge) + return log_oom(); + + /* fall through */ + + 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; @@ -300,25 +351,29 @@ 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; @@ -440,6 +495,8 @@ static int parse_argv(int argc, char *argv[]) { return -EINVAL; } + arg_retain = (arg_retain | plus | (arg_private_network ? 1ULL << CAP_NET_ADMIN : 0)) & ~minus; + return 1; } @@ -1198,27 +1255,325 @@ 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 == -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 0; +} + +static int setup_veth(pid_t pid, char iface_name[]) { + _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; + + strncpy(iface_name+3, arg_machine, IFNAMSIZ - 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(rtnl, 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, 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(0, &rtnl); + if (r < 0) { + log_error("Failed to connect to netlink: %s", strerror(-r)); + return r; + } + + r = sd_rtnl_message_new_link(rtnl, RTM_SETLINK, 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, 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 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(rtnl, 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(); - fd = socket(AF_NETLINK, SOCK_RAW, NETLINK_AUDIT); - if (fd >= 0) { - close_nointr_nofail(fd); - return true; + r = seccomp_add_secondary_archs(seccomp); + if (r < 0 && r != -EEXIST) { + log_error("Failed to add secondary archs to seccomp filter: %s", strerror(-r)); + goto finish; } - return false; + + 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; + _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] = "ve-"; log_parse_environment(); log_open(); @@ -1271,21 +1626,26 @@ 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 (arg_boot && 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(); @@ -1320,7 +1680,6 @@ int main(int argc, char *argv[]) { goto finish; } - if (access("/dev/kdbus/control", F_OK) >= 0) { if (arg_share_system) { @@ -1436,6 +1795,9 @@ 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; @@ -1472,6 +1834,9 @@ 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; @@ -1663,6 +2028,7 @@ 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"); @@ -1678,6 +2044,18 @@ 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; + eventfd_write(sync_fd, 1); close_nointr_nofail(sync_fd); sync_fd = -1; @@ -1748,6 +2126,7 @@ finish: free(arg_directory); free(arg_machine); free(arg_setenv); + free(arg_network_interfaces); return r; }