X-Git-Url: https://www.chiark.greenend.org.uk/ucgi/~ianmdlvl/git?p=elogind.git;a=blobdiff_plain;f=src%2Fnspawn%2Fnspawn.c;h=160b50b3ed83c8042c025da6bd46246ab11d7800;hp=5352b95ec69776b20c103ac6a1a4736dfe870fa8;hb=aa28aefe61c5406c5cb631f3e82457b6d1bcc967;hpb=40ddbdf85b592add401a0fbc963cec93559def54 diff --git a/src/nspawn/nspawn.c b/src/nspawn/nspawn.c index 5352b95ec..160b50b3e 100644 --- a/src/nspawn/nspawn.c +++ b/src/nspawn/nspawn.c @@ -40,11 +40,18 @@ #include #include #include +#include #include +#include + +#ifdef HAVE_SELINUX +#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" @@ -65,6 +72,7 @@ #include "bus-kernel.h" #include "env-util.h" #include "def.h" +#include "rtnl-util.h" typedef enum LinkJournal { LINK_NO, @@ -77,6 +85,8 @@ static char *arg_directory = NULL; static char *arg_user = NULL; static sd_id128_t arg_uuid = {}; static char *arg_machine = NULL; +static char *arg_selinux_context = NULL; +static char *arg_selinux_apifs_context = NULL; static const char *arg_slice = NULL; static bool arg_private_network = false; static bool arg_read_only = false; @@ -112,30 +122,48 @@ 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 int help(void) { printf("%s [OPTIONS...] [PATH] [ARGUMENTS...]\n\n" "Spawn a minimal namespace container for debugging, testing and building.\n\n" - " -h --help Show this help\n" - " --version Print version string\n" - " -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" - " -S --slice=SLICE Place the container in the specified slice\n" - " --private-network Disable network in container\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" - " --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", + " -h --help Show this help\n" + " --version Print version string\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" + " -S --slice=SLICE Place the container in the specified slice\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" + " --network-interface=INTERFACE\n" + " Assign an existing network interface to the 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" + " --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" + " --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", program_invocation_short_name); return 0; @@ -154,25 +182,36 @@ 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 }; 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 }, + { "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 }, {} }; @@ -181,7 +220,7 @@ static int parse_argv(int argc, char *argv[]) { assert(argc >= 0); assert(argv); - while ((c = getopt_long(argc, argv, "+hD:u:bM:jS:", options, NULL)) >= 0) { + while ((c = getopt_long(argc, argv, "+hD:u:bL:M:jS:Z:q", options, NULL)) >= 0) { switch (c) { @@ -211,6 +250,12 @@ static int parse_argv(int argc, char *argv[]) { 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; @@ -235,16 +280,30 @@ static int parse_argv(int argc, char *argv[]) { 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 { + + if (!hostname_is_valid(optarg)) { + log_error("Invalid machine name: %s", optarg); + return -EINVAL; + } + + free(arg_machine); + arg_machine = strdup(optarg); + if (!arg_machine) + return log_oom(); + + break; } - free(arg_machine); - arg_machine = strdup(optarg); - if (!arg_machine) - return log_oom(); + case 'Z': + arg_selinux_context = optarg; + break; + case 'L': + arg_selinux_apifs_context = optarg; break; case ARG_READ_ONLY: @@ -257,25 +316,29 @@ 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) + arg_retain = (uint64_t) -1; + else + arg_retain = 0; + } else { + if (cap_from_name(t, &cap) < 0) { + log_error("Failed to parse capability %s.", t); + return -EINVAL; + } + + if (c == ARG_CAPABILITY) + arg_retain |= 1ULL << (uint64_t) cap; + else + arg_retain &= ~(1ULL << (uint64_t) cap); } - - free(t); - - if (c == ARG_CAPABILITY) - arg_retain |= 1ULL << (uint64_t) cap; - else - arg_retain &= ~(1ULL << (uint64_t) cap); } break; @@ -354,6 +417,28 @@ 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 '?': return -EINVAL; @@ -362,6 +447,19 @@ 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; + } + return 1; } @@ -396,6 +494,10 @@ static int mount_all(const char *dest) { for (k = 0; k < ELEMENTSOF(mount_table); k++) { _cleanup_free_ char *where = NULL; +#ifdef HAVE_SELINUX + _cleanup_free_ char *options = NULL; +#endif + const char *o; int t; where = strjoin(dest, "/", mount_table[k].where, NULL); @@ -418,11 +520,24 @@ static int mount_all(const char *dest) { mkdir_p(where, 0755); +#ifdef HAVE_SELINUX + if (arg_selinux_apifs_context && + (streq_ptr(mount_table[k].what, "tmpfs") || streq_ptr(mount_table[k].what, "devpts"))) { + options = strjoin(mount_table[k].options, ",context=\"", arg_selinux_apifs_context, "\"", NULL); + if (!options) + return log_oom(); + + o = options; + } else +#endif + o = mount_table[k].options; + + if (mount(mount_table[k].what, where, mount_table[k].type, mount_table[k].flags, - mount_table[k].options) < 0 && + o) < 0 && mount_table[k].fatal) { log_error("mount(%s) failed: %m", where); @@ -583,6 +698,9 @@ static int setup_boot_id(const char *dest) { 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 */ @@ -808,6 +926,9 @@ 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; @@ -974,7 +1095,7 @@ static int setup_kdbus(const char *dest, const char *path) { } if (mount(path, p, "bind", MS_BIND, NULL) < 0) { - log_error("Failed to mount kdbus namespace path: %m"); + log_error("Failed to mount kdbus domain path: %m"); return -errno; } @@ -990,28 +1111,50 @@ static int register_machine(pid_t pid) { _cleanup_bus_unref_ sd_bus *bus = NULL; int r; - r = sd_bus_open_system(&bus); + 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 { + 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; @@ -1027,6 +1170,9 @@ 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)); @@ -1072,15 +1218,86 @@ 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 move_network_interfaces(pid_t pid) { + _cleanup_sd_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(NETLINK_ROUTE, &rtnl); + if (r < 0) { + log_error("Failed to connect to netlink: %s", strerror(-r)); + return r; + } + + STRV_FOREACH(i, arg_network_interfaces) { + _cleanup_sd_rtnl_message_unref_ sd_rtnl_message *m = NULL; + unsigned ifi; + + ifi = if_nametoindex(*i); + if (ifi == 0) { + log_error("Failed to resolve interface %s: %m", *i); + return -errno; + } + + r = sd_rtnl_message_link_new(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; + } - fd = socket(AF_NETLINK, SOCK_RAW, NETLINK_AUDIT); - if (fd >= 0) { - close_nointr_nofail(fd); - return true; + r = sd_rtnl_call(rtnl, m, 0, NULL); + if (r < 0) { + log_error("Failed to move interface to namespace: %s", strerror(-r)); + return r; + } } - return false; + + return 0; } int main(int argc, char *argv[]) { @@ -1092,8 +1309,7 @@ int main(int argc, char *argv[]) { sigset_t mask; _cleanup_close_pipe_ int kmsg_socket_pair[2] = { -1, -1 }; _cleanup_fdset_free_ FDSet *fds = NULL; - _cleanup_free_ char *kdbus_namespace = NULL; - const char *ns; + _cleanup_free_ char *kdbus_domain = NULL; log_parse_environment(); log_open(); @@ -1146,19 +1362,12 @@ 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) { + 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; } @@ -1187,19 +1396,34 @@ 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_namespace(ns, &kdbus_namespace); - if (r < 0) - log_debug("Failed to create kdbus namespace: %s", strerror(-r)); - else - log_debug("Successfully created kdbus namespace as %s", kdbus_namespace); + + 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"); @@ -1221,7 +1445,10 @@ 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"); @@ -1293,6 +1520,9 @@ 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; @@ -1356,7 +1586,7 @@ int main(int argc, char *argv[]) { if (mount_binds(arg_directory, arg_bind_ro, MS_RDONLY) < 0) goto child_fail; - if (setup_kdbus(arg_directory, kdbus_namespace) < 0) + if (setup_kdbus(arg_directory, kdbus_domain) < 0) goto child_fail; if (chdir(arg_directory) < 0) { @@ -1381,7 +1611,8 @@ 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"); @@ -1491,6 +1722,11 @@ int main(int argc, char *argv[]) { } else env_use = (char**) envp; +#ifdef HAVE_SELINUX + if (arg_selinux_context) + if (setexeccon(arg_selinux_context) < 0) + log_error("setexeccon(\"%s\") failed: %m", arg_selinux_context); +#endif if (arg_boot) { char **a; size_t l; @@ -1529,6 +1765,10 @@ int main(int argc, char *argv[]) { 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; @@ -1539,7 +1779,8 @@ 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); @@ -1562,21 +1803,26 @@ 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 { @@ -1593,6 +1839,7 @@ finish: free(arg_directory); free(arg_machine); free(arg_setenv); + free(arg_network_interfaces); return r; }