From 5a46d55fc8f3dc4a86f3994f96369d169c0b96bb Mon Sep 17 00:00:00 2001 From: Zbigniew Jędrzejewski-Szmek Date: Sun, 2 Oct 2016 15:51:27 +0200 Subject: path-util: add a function to peek into a container and guess systemd version This is a bit crude and only works for new systemd versions which have libsystemd-shared. --- src/basic/path-util.c | 68 +++++++++++++++++++++++++++++++++++++++++++++++ src/basic/path-util.h | 2 ++ src/test/test-path-util.c | 19 +++++++++++++ 3 files changed, 89 insertions(+) (limited to 'src') diff --git a/src/basic/path-util.c b/src/basic/path-util.c index b2fa81a294..c32e961af4 100644 --- a/src/basic/path-util.c +++ b/src/basic/path-util.c @@ -34,9 +34,11 @@ #include "alloc-util.h" #include "extract-word.h" #include "fs-util.h" +#include "glob-util.h" #include "log.h" #include "macro.h" #include "missing.h" +#include "parse-util.h" #include "path-util.h" #include "stat-util.h" #include "string-util.h" @@ -814,3 +816,69 @@ bool is_device_path(const char *path) { path_startswith(path, "/dev/") || path_startswith(path, "/sys/"); } + +int systemd_installation_has_version(const char *root, unsigned minimal_version) { + const char *pattern; + int r; + + /* Try to guess if systemd installation is later than the specified version. This + * is hacky and likely to yield false negatives, particularly if the installation + * is non-standard. False positives should be relatively rare. + */ + + NULSTR_FOREACH(pattern, + /* /lib works for systems without usr-merge, and for systems with a sane + * usr-merge, where /lib is a symlink to /usr/lib. /usr/lib is necessary + * for Gentoo which does a merge without making /lib a symlink. + */ + "lib/systemd/libsystemd-shared-*.so\0" + "usr/lib/systemd/libsystemd-shared-*.so\0") { + + _cleanup_strv_free_ char **names = NULL; + _cleanup_free_ char *path = NULL; + char *c, **name; + + path = prefix_root(root, pattern); + if (!path) + return -ENOMEM; + + r = glob_extend(&names, path); + if (r == -ENOENT) + continue; + if (r < 0) + return r; + + assert_se((c = endswith(path, "*.so"))); + *c = '\0'; /* truncate the glob part */ + + STRV_FOREACH(name, names) { + /* This is most likely to run only once, hence let's not optimize anything. */ + char *t, *t2; + unsigned version; + + t = startswith(*name, path); + if (!t) + continue; + + t2 = endswith(t, ".so"); + if (!t2) + continue; + + t2[0] = '\0'; /* truncate the suffix */ + + r = safe_atou(t, &version); + if (r < 0) { + log_debug_errno(r, "Found libsystemd shared at \"%s.so\", but failed to parse version: %m", *name); + continue; + } + + log_debug("Found libsystemd shared at \"%s.so\", version %u (%s).", + *name, version, + version >= minimal_version ? "OK" : "too old"); + if (version >= minimal_version) + return true; + } + } + + return false; +} diff --git a/src/basic/path-util.h b/src/basic/path-util.h index a27c13fcc3..78472f0961 100644 --- a/src/basic/path-util.h +++ b/src/basic/path-util.h @@ -125,3 +125,5 @@ char *file_in_same_dir(const char *path, const char *filename); bool hidden_or_backup_file(const char *filename) _pure_; bool is_device_path(const char *path); + +int systemd_installation_has_version(const char *root, unsigned minimal_version); diff --git a/src/test/test-path-util.c b/src/test/test-path-util.c index 164a10d8a8..0b10d8e25e 100644 --- a/src/test/test-path-util.c +++ b/src/test/test-path-util.c @@ -511,7 +511,24 @@ static void test_hidden_or_backup_file(void) { assert_se(!hidden_or_backup_file("test.dpkg-old.foo")); } +static void test_systemd_installation_has_version(const char *path) { + int r; + const unsigned versions[] = {0, 231, atoi(PACKAGE_VERSION), 999}; + unsigned i; + + for (i = 0; i < ELEMENTSOF(versions); i++) { + r = systemd_installation_has_version(path, versions[i]); + assert_se(r >= 0); + log_info("%s has systemd >= %u: %s", + path ?: "Current installation", versions[i], yes_no(r)); + } +} + int main(int argc, char **argv) { + log_set_max_level(LOG_DEBUG); + log_parse_environment(); + log_open(); + test_path(); test_find_binary(argv[0]); test_prefixes(); @@ -526,5 +543,7 @@ int main(int argc, char **argv) { test_filename_is_valid(); test_hidden_or_backup_file(); + test_systemd_installation_has_version(argv[1]); /* NULL is OK */ + return 0; } -- cgit v1.2.3-54-g00ecf From 98afd6af3a2022bd03543a6dea79120f689967b0 Mon Sep 17 00:00:00 2001 From: Zbigniew Jędrzejewski-Szmek Date: Sun, 2 Oct 2016 20:07:22 +0200 Subject: nspawn: check env var first, detect second If we are going to use the env var to override the detection result anyway, there is not point in doing the detection, especially that it can fail. --- src/nspawn/nspawn.c | 14 +++++++------- 1 file changed, 7 insertions(+), 7 deletions(-) (limited to 'src') diff --git a/src/nspawn/nspawn.c b/src/nspawn/nspawn.c index a08377b3a3..4ce0e2e0c1 100644 --- a/src/nspawn/nspawn.c +++ b/src/nspawn/nspawn.c @@ -320,13 +320,6 @@ static int detect_unified_cgroup_hierarchy(void) { const char *e; int r, all_unified, systemd_unified; - all_unified = cg_all_unified(); - systemd_unified = cg_unified(SYSTEMD_CGROUP_CONTROLLER); - - if (all_unified < 0 || systemd_unified < 0) - return log_error_errno(all_unified < 0 ? all_unified : systemd_unified, - "Failed to determine whether the unified cgroups hierarchy is used: %m"); - /* Allow the user to control whether the unified hierarchy is used */ e = getenv("UNIFIED_CGROUP_HIERARCHY"); if (e) { @@ -341,6 +334,13 @@ static int detect_unified_cgroup_hierarchy(void) { return 0; } + all_unified = cg_all_unified(); + systemd_unified = cg_unified(SYSTEMD_CGROUP_CONTROLLER); + + if (all_unified < 0 || systemd_unified < 0) + return log_error_errno(all_unified < 0 ? all_unified : systemd_unified, + "Failed to determine whether the unified cgroups hierarchy is used: %m"); + /* Otherwise inherit the default from the host system */ if (all_unified > 0) arg_unified_cgroup_hierarchy = CGROUP_UNIFIED_ALL; -- cgit v1.2.3-54-g00ecf From b006762524c76e8a3ef8d965bb2bf42c894196e4 Mon Sep 17 00:00:00 2001 From: Zbigniew Jędrzejewski-Szmek Date: Sun, 2 Oct 2016 21:32:38 +0200 Subject: nspawn: move the main loop body out to a new function The new function has 416 lines by itself! "return log_error_errno" is used to nicely reduce the volume of error handling code. A few minor issues are fixed on the way: - positive value was used as error value (EIO), causing systemd-nspawn to return success, even though it shouldn't. - In two places random values were used as error status, when the actual value was in an unusual place (etc_password_lock, notify_socket). Those are the only functional changes. There is another potential issue, which is marked with a comment, and left unresolved: the container can also return 133 by itself, causing a spurious reboot. --- src/nspawn/nspawn.c | 890 +++++++++++++++++++++++++--------------------------- 1 file changed, 430 insertions(+), 460 deletions(-) (limited to 'src') diff --git a/src/nspawn/nspawn.c b/src/nspawn/nspawn.c index 4ce0e2e0c1..d688d2eb64 100644 --- a/src/nspawn/nspawn.c +++ b/src/nspawn/nspawn.c @@ -3575,18 +3575,433 @@ static int load_settings(void) { return 0; } +static int run(int master, + const char* console, + const char *root_device, bool root_device_rw, + const char *home_device, bool home_device_rw, + const char *srv_device, bool srv_device_rw, + const char *esp_device, + bool interactive, + bool secondary, + FDSet *fds, + char veth_name[IFNAMSIZ], bool *veth_created, + union in_addr_union *exposed, + pid_t *pid, int *ret) { + + static const struct sigaction sa = { + .sa_handler = nop_signal_handler, + .sa_flags = SA_NOCLDSTOP, + }; + + _cleanup_release_lock_file_ LockFile uid_shift_lock = LOCK_FILE_INIT; + _cleanup_close_ int etc_passwd_lock = -1; + _cleanup_close_pair_ int + kmsg_socket_pair[2] = { -1, -1 }, + rtnl_socket_pair[2] = { -1, -1 }, + pid_socket_pair[2] = { -1, -1 }, + uuid_socket_pair[2] = { -1, -1 }, + notify_socket_pair[2] = { -1, -1 }, + uid_shift_socket_pair[2] = { -1, -1 }; + _cleanup_close_ int notify_socket= -1; + _cleanup_(barrier_destroy) Barrier barrier = BARRIER_NULL; + _cleanup_(sd_event_unrefp) sd_event *event = NULL; + _cleanup_(pty_forward_freep) PTYForward *forward = NULL; + _cleanup_(sd_netlink_unrefp) sd_netlink *rtnl = NULL; + ContainerStatus container_status = 0; + char last_char = 0; + int ifi = 0, r; + ssize_t l; + sigset_t mask_chld; + + assert_se(sigemptyset(&mask_chld) == 0); + assert_se(sigaddset(&mask_chld, SIGCHLD) == 0); + + if (arg_userns_mode == USER_NAMESPACE_PICK) { + /* When we shall pick the UID/GID range, let's first lock /etc/passwd, so that we can safely + * check with getpwuid() if the specific user already exists. Note that /etc might be + * read-only, in which case this will fail with EROFS. But that's really OK, as in that case we + * can be reasonably sure that no users are going to be added. Note that getpwuid() checks are + * really just an extra safety net. We kinda assume that the UID range we allocate from is + * really ours. */ + + etc_passwd_lock = take_etc_passwd_lock(NULL); + if (etc_passwd_lock < 0 && etc_passwd_lock != -EROFS) + return log_error_errno(etc_passwd_lock, "Failed to take /etc/passwd lock: %m"); + } + + r = barrier_create(&barrier); + if (r < 0) + return log_error_errno(r, "Cannot initialize IPC barrier: %m"); + + if (socketpair(AF_UNIX, SOCK_SEQPACKET|SOCK_CLOEXEC, 0, kmsg_socket_pair) < 0) + return log_error_errno(errno, "Failed to create kmsg socket pair: %m"); + + if (socketpair(AF_UNIX, SOCK_SEQPACKET|SOCK_CLOEXEC, 0, rtnl_socket_pair) < 0) + return log_error_errno(errno, "Failed to create rtnl socket pair: %m"); + + if (socketpair(AF_UNIX, SOCK_SEQPACKET|SOCK_CLOEXEC, 0, pid_socket_pair) < 0) + return log_error_errno(errno, "Failed to create pid socket pair: %m"); + + if (socketpair(AF_UNIX, SOCK_SEQPACKET|SOCK_CLOEXEC, 0, uuid_socket_pair) < 0) + return log_error_errno(errno, "Failed to create id socket pair: %m"); + + if (socketpair(AF_UNIX, SOCK_SEQPACKET|SOCK_CLOEXEC, 0, notify_socket_pair) < 0) + return log_error_errno(errno, "Failed to create notify socket pair: %m"); + + if (arg_userns_mode != USER_NAMESPACE_NO) + if (socketpair(AF_UNIX, SOCK_SEQPACKET|SOCK_CLOEXEC, 0, uid_shift_socket_pair) < 0) + return log_error_errno(errno, "Failed to create uid shift socket pair: %m"); + + /* Child can be killed before execv(), so handle SIGCHLD in order to interrupt + * parent's blocking calls and give it a chance to call wait() and terminate. */ + r = sigprocmask(SIG_UNBLOCK, &mask_chld, NULL); + if (r < 0) + return log_error_errno(errno, "Failed to change the signal mask: %m"); + + r = sigaction(SIGCHLD, &sa, NULL); + if (r < 0) + return log_error_errno(errno, "Failed to install SIGCHLD handler: %m"); + + *pid = raw_clone(SIGCHLD|CLONE_NEWNS); + if (*pid < 0) + return log_error_errno(errno, "clone() failed%s: %m", + errno == EINVAL ? + ", do you have namespace support enabled in your kernel? (You need UTS, IPC, PID and NET namespacing built in)" : ""); + + if (*pid == 0) { + /* The outer child only has a file system namespace. */ + barrier_set_role(&barrier, BARRIER_CHILD); + + master = safe_close(master); + + kmsg_socket_pair[0] = safe_close(kmsg_socket_pair[0]); + rtnl_socket_pair[0] = safe_close(rtnl_socket_pair[0]); + pid_socket_pair[0] = safe_close(pid_socket_pair[0]); + uuid_socket_pair[0] = safe_close(uuid_socket_pair[0]); + notify_socket_pair[0] = safe_close(notify_socket_pair[0]); + uid_shift_socket_pair[0] = safe_close(uid_shift_socket_pair[0]); + + (void) reset_all_signal_handlers(); + (void) reset_signal_mask(); + + r = outer_child(&barrier, + arg_directory, + console, + root_device, root_device_rw, + home_device, home_device_rw, + srv_device, srv_device_rw, + esp_device, + interactive, + secondary, + pid_socket_pair[1], + uuid_socket_pair[1], + notify_socket_pair[1], + kmsg_socket_pair[1], + rtnl_socket_pair[1], + uid_shift_socket_pair[1], + fds); + if (r < 0) + _exit(EXIT_FAILURE); + + _exit(EXIT_SUCCESS); + } + + barrier_set_role(&barrier, BARRIER_PARENT); + + fds = fdset_free(fds); + + kmsg_socket_pair[1] = safe_close(kmsg_socket_pair[1]); + rtnl_socket_pair[1] = safe_close(rtnl_socket_pair[1]); + pid_socket_pair[1] = safe_close(pid_socket_pair[1]); + uuid_socket_pair[1] = safe_close(uuid_socket_pair[1]); + notify_socket_pair[1] = safe_close(notify_socket_pair[1]); + uid_shift_socket_pair[1] = safe_close(uid_shift_socket_pair[1]); + + if (arg_userns_mode != USER_NAMESPACE_NO) { + /* The child just let us know the UID shift it might have read from the image. */ + l = recv(uid_shift_socket_pair[0], &arg_uid_shift, sizeof arg_uid_shift, 0); + if (l < 0) + return log_error_errno(errno, "Failed to read UID shift: %m"); + + if (l != sizeof arg_uid_shift) { + log_error("Short read while reading UID shift."); + return -EIO; + } + + if (arg_userns_mode == USER_NAMESPACE_PICK) { + /* If we are supposed to pick the UID shift, let's try to use the shift read from the + * image, but if that's already in use, pick a new one, and report back to the child, + * which one we now picked. */ + + r = uid_shift_pick(&arg_uid_shift, &uid_shift_lock); + if (r < 0) + return log_error_errno(r, "Failed to pick suitable UID/GID range: %m"); + + l = send(uid_shift_socket_pair[0], &arg_uid_shift, sizeof arg_uid_shift, MSG_NOSIGNAL); + if (l < 0) + return log_error_errno(errno, "Failed to send UID shift: %m"); + if (l != sizeof arg_uid_shift) { + log_error("Short write while writing UID shift."); + return -EIO; + } + } + } + + /* Wait for the outer child. */ + r = wait_for_terminate_and_warn("namespace helper", *pid, NULL); + if (r != 0) + return r < 0 ? r : -EIO; + + /* And now retrieve the PID of the inner child. */ + l = recv(pid_socket_pair[0], pid, sizeof *pid, 0); + if (l < 0) + return log_error_errno(errno, "Failed to read inner child PID: %m"); + if (l != sizeof *pid) { + log_error("Short read while reading inner child PID."); + return -EIO; + } + + /* We also retrieve container UUID in case it was generated by outer child */ + l = recv(uuid_socket_pair[0], &arg_uuid, sizeof arg_uuid, 0); + if (l < 0) + return log_error_errno(errno, "Failed to read container machine ID: %m"); + if (l != sizeof(arg_uuid)) { + log_error("Short read while reading container machined ID."); + return -EIO; + } + + /* We also retrieve the socket used for notifications generated by outer child */ + notify_socket = receive_one_fd(notify_socket_pair[0], 0); + if (notify_socket < 0) + return log_error_errno(notify_socket, + "Failed to receive notification socket from the outer child: %m"); + + log_debug("Init process invoked as PID "PID_FMT, *pid); + + if (arg_userns_mode != USER_NAMESPACE_NO) { + if (!barrier_place_and_sync(&barrier)) { /* #1 */ + log_error("Child died too early."); + return -ESRCH; + } + + r = setup_uid_map(*pid); + if (r < 0) + return r; + + (void) barrier_place(&barrier); /* #2 */ + } + + if (arg_private_network) { + + r = move_network_interfaces(*pid, arg_network_interfaces); + if (r < 0) + return r; + + if (arg_network_veth) { + r = setup_veth(arg_machine, *pid, veth_name, + arg_network_bridge || arg_network_zone); + if (r < 0) + return r; + else if (r > 0) + ifi = r; + + if (arg_network_bridge) { + /* Add the interface to a bridge */ + r = setup_bridge(veth_name, arg_network_bridge, false); + if (r < 0) + return r; + if (r > 0) + ifi = r; + } else if (arg_network_zone) { + /* Add the interface to a bridge, possibly creating it */ + r = setup_bridge(veth_name, arg_network_zone, true); + if (r < 0) + return r; + if (r > 0) + ifi = r; + } + } + + r = setup_veth_extra(arg_machine, *pid, arg_network_veth_extra); + if (r < 0) + return r; + + /* We created the primary and extra veth links now; let's remember this, so that we know to + remove them later on. Note that we don't bother with removing veth links that were created + here when their setup failed half-way, because in that case the kernel should be able to + remove them on its own, since they cannot be referenced by anything yet. */ + *veth_created = true; + + r = setup_macvlan(arg_machine, *pid, arg_network_macvlan); + if (r < 0) + return r; + + r = setup_ipvlan(arg_machine, *pid, arg_network_ipvlan); + if (r < 0) + return r; + } + + if (arg_register) { + r = register_machine( + arg_machine, + *pid, + arg_directory, + arg_uuid, + ifi, + arg_slice, + arg_custom_mounts, arg_n_custom_mounts, + arg_kill_signal, + arg_property, + arg_keep_unit, + arg_container_service_name); + if (r < 0) + return r; + } + + r = sync_cgroup(*pid, arg_unified_cgroup_hierarchy); + if (r < 0) + return r; + + if (arg_keep_unit) { + r = create_subcgroup(*pid, arg_unified_cgroup_hierarchy); + if (r < 0) + return r; + } + + r = chown_cgroup(*pid, arg_uid_shift); + if (r < 0) + return r; + + /* Notify the child that the parent is ready with all + * its setup (including cgroup-ification), and that + * the child can now hand over control to the code to + * run inside the container. */ + (void) barrier_place(&barrier); /* #3 */ + + /* Block SIGCHLD here, before notifying child. + * process_pty() will handle it with the other signals. */ + assert_se(sigprocmask(SIG_BLOCK, &mask_chld, NULL) >= 0); + + /* Reset signal to default */ + r = default_signals(SIGCHLD, -1); + if (r < 0) + return log_error_errno(r, "Failed to reset SIGCHLD: %m"); + + r = sd_event_new(&event); + if (r < 0) + return log_error_errno(r, "Failed to get default event source: %m"); + + r = setup_sd_notify_parent(event, notify_socket, PID_TO_PTR(*pid)); + if (r < 0) + return r; + + /* Let the child know that we are ready and wait that the child is completely ready now. */ + if (!barrier_place_and_sync(&barrier)) { /* #4 */ + log_error("Child died too early."); + return -ESRCH; + } + + /* At this point we have made use of the UID we picked, and thus nss-mymachines + * will make them appear in getpwuid(), thus we can release the /etc/passwd lock. */ + etc_passwd_lock = safe_close(etc_passwd_lock); + + sd_notifyf(false, + "STATUS=Container running.\n" + "X_NSPAWN_LEADER_PID=" PID_FMT, *pid); + if (!arg_notify_ready) + sd_notify(false, "READY=1\n"); + + if (arg_kill_signal > 0) { + /* Try to kill the init system on SIGINT or SIGTERM */ + sd_event_add_signal(event, NULL, SIGINT, on_orderly_shutdown, PID_TO_PTR(*pid)); + sd_event_add_signal(event, NULL, SIGTERM, on_orderly_shutdown, PID_TO_PTR(*pid)); + } else { + /* Immediately exit */ + sd_event_add_signal(event, NULL, SIGINT, NULL, NULL); + sd_event_add_signal(event, NULL, SIGTERM, NULL, NULL); + } + + /* simply exit on sigchld */ + sd_event_add_signal(event, NULL, SIGCHLD, NULL, NULL); + + if (arg_expose_ports) { + r = expose_port_watch_rtnl(event, rtnl_socket_pair[0], on_address_change, exposed, &rtnl); + if (r < 0) + return r; + + (void) expose_port_execute(rtnl, arg_expose_ports, exposed); + } + + rtnl_socket_pair[0] = safe_close(rtnl_socket_pair[0]); + + r = pty_forward_new(event, master, + PTY_FORWARD_IGNORE_VHANGUP | (interactive ? 0 : PTY_FORWARD_READ_ONLY), + &forward); + if (r < 0) + return log_error_errno(r, "Failed to create PTY forwarder: %m"); + + r = sd_event_loop(event); + if (r < 0) + return log_error_errno(r, "Failed to run event loop: %m"); + + pty_forward_get_last_char(forward, &last_char); + + forward = pty_forward_free(forward); + + if (!arg_quiet && last_char != '\n') + putc('\n', stdout); + + /* Kill if it is not dead yet anyway */ + if (arg_register && !arg_keep_unit) + terminate_machine(*pid); + + /* Normally redundant, but better safe than sorry */ + kill(*pid, SIGKILL); + + r = wait_for_container(*pid, &container_status); + *pid = 0; + + if (r < 0) + /* We failed to wait for the container, or the container exited abnormally. */ + return r; + if (r > 0 || container_status == CONTAINER_TERMINATED) { + /* r > 0 → The container exited with a non-zero status, + * otherwise → The container exited with zero status and reboot was not requested. */ + *ret = r; + // XXX: if the container returned 133, we will reboot! + return 0; /* finito */ + } + + /* CONTAINER_REBOOTED, loop again */ + + if (arg_keep_unit) { + /* Special handling if we are running as a service: instead of simply + * restarting the machine we want to restart the entire service, so let's + * inform systemd about this with the special exit code 133. The service + * file uses RestartForceExitStatus=133 so that this results in a full + * nspawn restart. This is necessary since we might have cgroup parameters + * set we want to have flushed out. */ + *ret = 0; + return 133; + } + + expose_port_flush(arg_expose_ports, exposed); + + (void) remove_veth_links(veth_name, arg_network_veth_extra); + *veth_created = false; + return 1; /* loop again */ +} + int main(int argc, char *argv[]) { _cleanup_free_ char *device_path = NULL, *root_device = NULL, *home_device = NULL, *srv_device = NULL, *esp_device = NULL, *console = NULL; bool root_device_rw = true, home_device_rw = true, srv_device_rw = true; _cleanup_close_ int master = -1, image_fd = -1; _cleanup_fdset_free_ FDSet *fds = NULL; - int r, n_fd_passed, loop_nr = -1; + int r, n_fd_passed, loop_nr = -1, ret = EXIT_FAILURE; char veth_name[IFNAMSIZ] = ""; bool secondary = false, remove_subvol = false; - sigset_t mask_chld; pid_t pid = 0; - int ret = EXIT_SUCCESS; union in_addr_union exposed = {}; _cleanup_release_lock_file_ LockFile tree_global_lock = LOCK_FILE_INIT, tree_local_lock = LOCK_FILE_INIT; bool interactive, veth_created = false; @@ -3802,470 +4217,25 @@ int main(int argc, char *argv[]) { assert_se(sigprocmask_many(SIG_BLOCK, NULL, SIGCHLD, SIGWINCH, SIGTERM, SIGINT, -1) >= 0); - assert_se(sigemptyset(&mask_chld) == 0); - assert_se(sigaddset(&mask_chld, SIGCHLD) == 0); - if (prctl(PR_SET_CHILD_SUBREAPER, 1) < 0) { r = log_error_errno(errno, "Failed to become subreaper: %m"); goto finish; } for (;;) { - static const struct sigaction sa = { - .sa_handler = nop_signal_handler, - .sa_flags = SA_NOCLDSTOP, - }; - - _cleanup_release_lock_file_ LockFile uid_shift_lock = LOCK_FILE_INIT; - _cleanup_close_ int etc_passwd_lock = -1; - _cleanup_close_pair_ int - kmsg_socket_pair[2] = { -1, -1 }, - rtnl_socket_pair[2] = { -1, -1 }, - pid_socket_pair[2] = { -1, -1 }, - uuid_socket_pair[2] = { -1, -1 }, - notify_socket_pair[2] = { -1, -1 }, - uid_shift_socket_pair[2] = { -1, -1 }; - _cleanup_close_ int notify_socket= -1; - _cleanup_(barrier_destroy) Barrier barrier = BARRIER_NULL; - _cleanup_(sd_event_unrefp) sd_event *event = NULL; - _cleanup_(pty_forward_freep) PTYForward *forward = NULL; - _cleanup_(sd_netlink_unrefp) sd_netlink *rtnl = NULL; - ContainerStatus container_status = 0; - char last_char = 0; - int ifi = 0; - ssize_t l; - - if (arg_userns_mode == USER_NAMESPACE_PICK) { - /* When we shall pick the UID/GID range, let's first lock /etc/passwd, so that we can safely - * check with getpwuid() if the specific user already exists. Note that /etc might be - * read-only, in which case this will fail with EROFS. But that's really OK, as in that case we - * can be reasonably sure that no users are going to be added. Note that getpwuid() checks are - * really just an extra safety net. We kinda assume that the UID range we allocate from is - * really ours. */ - - etc_passwd_lock = take_etc_passwd_lock(NULL); - if (etc_passwd_lock < 0 && etc_passwd_lock != -EROFS) { - log_error_errno(r, "Failed to take /etc/passwd lock: %m"); - goto finish; - } - } - - r = barrier_create(&barrier); - if (r < 0) { - log_error_errno(r, "Cannot initialize IPC barrier: %m"); - goto finish; - } - - if (socketpair(AF_UNIX, SOCK_SEQPACKET|SOCK_CLOEXEC, 0, kmsg_socket_pair) < 0) { - r = log_error_errno(errno, "Failed to create kmsg socket pair: %m"); - goto finish; - } - - if (socketpair(AF_UNIX, SOCK_SEQPACKET|SOCK_CLOEXEC, 0, rtnl_socket_pair) < 0) { - r = log_error_errno(errno, "Failed to create rtnl socket pair: %m"); - goto finish; - } - - if (socketpair(AF_UNIX, SOCK_SEQPACKET|SOCK_CLOEXEC, 0, pid_socket_pair) < 0) { - r = log_error_errno(errno, "Failed to create pid socket pair: %m"); - goto finish; - } - - if (socketpair(AF_UNIX, SOCK_SEQPACKET|SOCK_CLOEXEC, 0, uuid_socket_pair) < 0) { - r = log_error_errno(errno, "Failed to create id socket pair: %m"); - goto finish; - } - - if (socketpair(AF_UNIX, SOCK_SEQPACKET|SOCK_CLOEXEC, 0, notify_socket_pair) < 0) { - r = log_error_errno(errno, "Failed to create notify socket pair: %m"); - goto finish; - } - - if (arg_userns_mode != USER_NAMESPACE_NO) - if (socketpair(AF_UNIX, SOCK_SEQPACKET|SOCK_CLOEXEC, 0, uid_shift_socket_pair) < 0) { - r = log_error_errno(errno, "Failed to create uid shift socket pair: %m"); - goto finish; - } - - /* Child can be killed before execv(), so handle SIGCHLD - * in order to interrupt parent's blocking calls and - * give it a chance to call wait() and terminate. */ - r = sigprocmask(SIG_UNBLOCK, &mask_chld, NULL); - if (r < 0) { - r = log_error_errno(errno, "Failed to change the signal mask: %m"); - goto finish; - } - - r = sigaction(SIGCHLD, &sa, NULL); - if (r < 0) { - r = log_error_errno(errno, "Failed to install SIGCHLD handler: %m"); - goto finish; - } - - pid = raw_clone(SIGCHLD|CLONE_NEWNS); - if (pid < 0) { - if (errno == EINVAL) - r = log_error_errno(errno, "clone() failed, do you have namespace support enabled in your kernel? (You need UTS, IPC, PID and NET namespacing built in): %m"); - else - r = log_error_errno(errno, "clone() failed: %m"); - - goto finish; - } - - if (pid == 0) { - /* The outer child only has a file system namespace. */ - barrier_set_role(&barrier, BARRIER_CHILD); - - master = safe_close(master); - - kmsg_socket_pair[0] = safe_close(kmsg_socket_pair[0]); - rtnl_socket_pair[0] = safe_close(rtnl_socket_pair[0]); - pid_socket_pair[0] = safe_close(pid_socket_pair[0]); - uuid_socket_pair[0] = safe_close(uuid_socket_pair[0]); - notify_socket_pair[0] = safe_close(notify_socket_pair[0]); - uid_shift_socket_pair[0] = safe_close(uid_shift_socket_pair[0]); - - (void) reset_all_signal_handlers(); - (void) reset_signal_mask(); - - r = outer_child(&barrier, - arg_directory, - console, - root_device, root_device_rw, - home_device, home_device_rw, - srv_device, srv_device_rw, - esp_device, - interactive, - secondary, - pid_socket_pair[1], - uuid_socket_pair[1], - notify_socket_pair[1], - kmsg_socket_pair[1], - rtnl_socket_pair[1], - uid_shift_socket_pair[1], - fds); - if (r < 0) - _exit(EXIT_FAILURE); - - _exit(EXIT_SUCCESS); - } - - barrier_set_role(&barrier, BARRIER_PARENT); - - fds = fdset_free(fds); - - kmsg_socket_pair[1] = safe_close(kmsg_socket_pair[1]); - rtnl_socket_pair[1] = safe_close(rtnl_socket_pair[1]); - pid_socket_pair[1] = safe_close(pid_socket_pair[1]); - uuid_socket_pair[1] = safe_close(uuid_socket_pair[1]); - notify_socket_pair[1] = safe_close(notify_socket_pair[1]); - uid_shift_socket_pair[1] = safe_close(uid_shift_socket_pair[1]); - - if (arg_userns_mode != USER_NAMESPACE_NO) { - /* The child just let us know the UID shift it might have read from the image. */ - l = recv(uid_shift_socket_pair[0], &arg_uid_shift, sizeof(arg_uid_shift), 0); - if (l < 0) { - r = log_error_errno(errno, "Failed to read UID shift: %m"); - goto finish; - } - if (l != sizeof(arg_uid_shift)) { - log_error("Short read while reading UID shift."); - r = EIO; - goto finish; - } - - if (arg_userns_mode == USER_NAMESPACE_PICK) { - /* If we are supposed to pick the UID shift, let's try to use the shift read from the - * image, but if that's already in use, pick a new one, and report back to the child, - * which one we now picked. */ - - r = uid_shift_pick(&arg_uid_shift, &uid_shift_lock); - if (r < 0) { - log_error_errno(r, "Failed to pick suitable UID/GID range: %m"); - goto finish; - } - - l = send(uid_shift_socket_pair[0], &arg_uid_shift, sizeof(arg_uid_shift), MSG_NOSIGNAL); - if (l < 0) { - r = log_error_errno(errno, "Failed to send UID shift: %m"); - goto finish; - } - if (l != sizeof(arg_uid_shift)) { - log_error("Short write while writing UID shift."); - r = -EIO; - goto finish; - } - } - } - - /* Wait for the outer child. */ - r = wait_for_terminate_and_warn("namespace helper", pid, NULL); - if (r < 0) - goto finish; - if (r != 0) { - r = -EIO; - goto finish; - } - pid = 0; - - /* And now retrieve the PID of the inner child. */ - l = recv(pid_socket_pair[0], &pid, sizeof(pid), 0); - if (l < 0) { - r = log_error_errno(errno, "Failed to read inner child PID: %m"); - goto finish; - } - if (l != sizeof(pid)) { - log_error("Short read while reading inner child PID."); - r = EIO; - goto finish; - } - - /* We also retrieve container UUID in case it was generated by outer child */ - l = recv(uuid_socket_pair[0], &arg_uuid, sizeof(arg_uuid), 0); - if (l < 0) { - r = log_error_errno(errno, "Failed to read container machine ID: %m"); - goto finish; - } - if (l != sizeof(arg_uuid)) { - log_error("Short read while reading container machined ID."); - r = EIO; - goto finish; - } - - /* We also retrieve the socket used for notifications generated by outer child */ - notify_socket = receive_one_fd(notify_socket_pair[0], 0); - if (notify_socket < 0) { - r = log_error_errno(errno, "Failed to receive notification socket from the outer child: %m"); - goto finish; - } - - log_debug("Init process invoked as PID " PID_FMT, pid); - - if (arg_userns_mode != USER_NAMESPACE_NO) { - if (!barrier_place_and_sync(&barrier)) { /* #1 */ - log_error("Child died too early."); - r = -ESRCH; - goto finish; - } - - r = setup_uid_map(pid); - if (r < 0) - goto finish; - - (void) barrier_place(&barrier); /* #2 */ - } - - if (arg_private_network) { - - r = move_network_interfaces(pid, arg_network_interfaces); - if (r < 0) - goto finish; - - if (arg_network_veth) { - r = setup_veth(arg_machine, pid, veth_name, - arg_network_bridge || arg_network_zone); - if (r < 0) - goto finish; - else if (r > 0) - ifi = r; - - if (arg_network_bridge) { - /* Add the interface to a bridge */ - r = setup_bridge(veth_name, arg_network_bridge, false); - if (r < 0) - goto finish; - if (r > 0) - ifi = r; - } else if (arg_network_zone) { - /* Add the interface to a bridge, possibly creating it */ - r = setup_bridge(veth_name, arg_network_zone, true); - if (r < 0) - goto finish; - if (r > 0) - ifi = r; - } - } - - r = setup_veth_extra(arg_machine, pid, arg_network_veth_extra); - if (r < 0) - goto finish; - - /* We created the primary and extra veth links now; let's remember this, so that we know to - remove them later on. Note that we don't bother with removing veth links that were created - here when their setup failed half-way, because in that case the kernel should be able to - remove them on its own, since they cannot be referenced by anything yet. */ - veth_created = true; - - r = setup_macvlan(arg_machine, pid, arg_network_macvlan); - if (r < 0) - goto finish; - - r = setup_ipvlan(arg_machine, pid, arg_network_ipvlan); - if (r < 0) - goto finish; - } - - if (arg_register) { - r = register_machine( - arg_machine, - pid, - arg_directory, - arg_uuid, - ifi, - arg_slice, - arg_custom_mounts, arg_n_custom_mounts, - arg_kill_signal, - arg_property, - arg_keep_unit, - arg_container_service_name); - if (r < 0) - goto finish; - } - - r = sync_cgroup(pid, arg_unified_cgroup_hierarchy); - if (r < 0) - goto finish; - - if (arg_keep_unit) { - r = create_subcgroup(pid, arg_unified_cgroup_hierarchy); - if (r < 0) - goto finish; - } - - r = chown_cgroup(pid, arg_uid_shift); - if (r < 0) - goto finish; - - /* Notify the child that the parent is ready with all - * its setup (including cgroup-ification), and that - * the child can now hand over control to the code to - * run inside the container. */ - (void) barrier_place(&barrier); /* #3 */ - - /* Block SIGCHLD here, before notifying child. - * process_pty() will handle it with the other signals. */ - assert_se(sigprocmask(SIG_BLOCK, &mask_chld, NULL) >= 0); - - /* Reset signal to default */ - r = default_signals(SIGCHLD, -1); - if (r < 0) { - log_error_errno(r, "Failed to reset SIGCHLD: %m"); - goto finish; - } - - r = sd_event_new(&event); - if (r < 0) { - log_error_errno(r, "Failed to get default event source: %m"); - goto finish; - } - - r = setup_sd_notify_parent(event, notify_socket, PID_TO_PTR(pid)); - if (r < 0) - goto finish; - - /* Let the child know that we are ready and wait that the child is completely ready now. */ - if (!barrier_place_and_sync(&barrier)) { /* #4 */ - log_error("Child died too early."); - r = -ESRCH; - goto finish; - } - - /* At this point we have made use of the UID we picked, and thus nss-mymachines will make them appear - * in getpwuid(), thus we can release the /etc/passwd lock. */ - etc_passwd_lock = safe_close(etc_passwd_lock); - - sd_notifyf(false, - "STATUS=Container running.\n" - "X_NSPAWN_LEADER_PID=" PID_FMT, pid); - if (!arg_notify_ready) - sd_notify(false, "READY=1\n"); - - if (arg_kill_signal > 0) { - /* Try to kill the init system on SIGINT or SIGTERM */ - sd_event_add_signal(event, NULL, SIGINT, on_orderly_shutdown, PID_TO_PTR(pid)); - sd_event_add_signal(event, NULL, SIGTERM, on_orderly_shutdown, PID_TO_PTR(pid)); - } else { - /* Immediately exit */ - sd_event_add_signal(event, NULL, SIGINT, NULL, NULL); - sd_event_add_signal(event, NULL, SIGTERM, NULL, NULL); - } - - /* simply exit on sigchld */ - sd_event_add_signal(event, NULL, SIGCHLD, NULL, NULL); - - if (arg_expose_ports) { - r = expose_port_watch_rtnl(event, rtnl_socket_pair[0], on_address_change, &exposed, &rtnl); - if (r < 0) - goto finish; - - (void) expose_port_execute(rtnl, arg_expose_ports, &exposed); - } - - rtnl_socket_pair[0] = safe_close(rtnl_socket_pair[0]); - - r = pty_forward_new(event, master, PTY_FORWARD_IGNORE_VHANGUP | (interactive ? 0 : PTY_FORWARD_READ_ONLY), &forward); - if (r < 0) { - log_error_errno(r, "Failed to create PTY forwarder: %m"); - goto finish; - } - - r = sd_event_loop(event); - if (r < 0) { - log_error_errno(r, "Failed to run event loop: %m"); - goto finish; - } - - pty_forward_get_last_char(forward, &last_char); - - forward = pty_forward_free(forward); - - if (!arg_quiet && last_char != '\n') - putc('\n', stdout); - - /* Kill if it is not dead yet anyway */ - if (arg_register && !arg_keep_unit) - terminate_machine(pid); - - /* Normally redundant, but better safe than sorry */ - kill(pid, SIGKILL); - - r = wait_for_container(pid, &container_status); - pid = 0; - - if (r < 0) - /* We failed to wait for the container, or the - * container exited abnormally */ - goto finish; - else if (r > 0 || container_status == CONTAINER_TERMINATED) { - /* The container exited with a non-zero - * status, or with zero status and no reboot - * was requested. */ - ret = r; - break; - } - - /* CONTAINER_REBOOTED, loop again */ - - if (arg_keep_unit) { - /* Special handling if we are running as a - * service: instead of simply restarting the - * machine we want to restart the entire - * service, so let's inform systemd about this - * with the special exit code 133. The service - * file uses RestartForceExitStatus=133 so - * that this results in a full nspawn - * restart. This is necessary since we might - * have cgroup parameters set we want to have - * flushed out. */ - ret = 133; - r = 0; + r = run(master, + console, + root_device, root_device_rw, + home_device, home_device_rw, + srv_device, srv_device_rw, + esp_device, + interactive, secondary, + fds, + veth_name, &veth_created, + &exposed, + &pid, &ret); + if (r <= 0) break; - } - - expose_port_flush(arg_expose_ports, &exposed); - - (void) remove_veth_links(veth_name, arg_network_veth_extra); - veth_created = false; } finish: -- cgit v1.2.3-54-g00ecf From 27e29a1e4390415e2d5b709f3d8c2d9481b50987 Mon Sep 17 00:00:00 2001 From: Zbigniew Jędrzejewski-Szmek Date: Fri, 7 Oct 2016 10:31:47 -0400 Subject: nspawn: fix spurious reboot if container process returns 133 --- src/nspawn/nspawn.c | 10 +++++++--- 1 file changed, 7 insertions(+), 3 deletions(-) (limited to 'src') diff --git a/src/nspawn/nspawn.c b/src/nspawn/nspawn.c index d688d2eb64..12d6eab8be 100644 --- a/src/nspawn/nspawn.c +++ b/src/nspawn/nspawn.c @@ -3965,10 +3965,14 @@ static int run(int master, /* We failed to wait for the container, or the container exited abnormally. */ return r; if (r > 0 || container_status == CONTAINER_TERMINATED) { - /* r > 0 → The container exited with a non-zero status, - * otherwise → The container exited with zero status and reboot was not requested. */ + /* r > 0 → The container exited with a non-zero status. + * As a special case, we need to replace 133 with a different value, + * because 133 is special-cased in the service file to reboot the container. + * otherwise → The container exited with zero status and a reboot was not requested. + */ + if (r == 133) + r = EXIT_FAILURE; /* replace 133 with the general failure code */ *ret = r; - // XXX: if the container returned 133, we will reboot! return 0; /* finito */ } -- cgit v1.2.3-54-g00ecf From 0fd9563fdeff88458f1a533ac50445abc89e56d8 Mon Sep 17 00:00:00 2001 From: Zbigniew Jędrzejewski-Szmek Date: Sat, 8 Oct 2016 02:18:26 -0400 Subject: nspawn: use mixed cgroup hierarchy only when container has new systemd systemd-soon-to-be-released-232 is able to deal with the mixed hierarchy. So make an educated guess, and use the mixed hierarchy in that case. Tested by running the host with mixed hierarchy (i.e. simply using a recent kernel with systemd from git), and booting first a container with older systemd, and then one with a newer systemd. Fixes #4008. --- src/nspawn/nspawn.c | 23 +++++++++++++++-------- 1 file changed, 15 insertions(+), 8 deletions(-) (limited to 'src') diff --git a/src/nspawn/nspawn.c b/src/nspawn/nspawn.c index 12d6eab8be..a775d1ac4f 100644 --- a/src/nspawn/nspawn.c +++ b/src/nspawn/nspawn.c @@ -316,7 +316,7 @@ static int custom_mounts_prepare(void) { return 0; } -static int detect_unified_cgroup_hierarchy(void) { +static int detect_unified_cgroup_hierarchy(const char *directory) { const char *e; int r, all_unified, systemd_unified; @@ -344,9 +344,16 @@ static int detect_unified_cgroup_hierarchy(void) { /* Otherwise inherit the default from the host system */ if (all_unified > 0) arg_unified_cgroup_hierarchy = CGROUP_UNIFIED_ALL; - else if (systemd_unified > 0) - arg_unified_cgroup_hierarchy = CGROUP_UNIFIED_SYSTEMD; - else + else if (systemd_unified > 0) { + /* mixed cgroup hierarchy support was added in 232 */ + r = systemd_installation_has_version(directory, 232); + 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; return 0; @@ -1125,10 +1132,6 @@ static int parse_argv(int argc, char *argv[]) { arg_caps_retain = (arg_caps_retain | plus | (arg_private_network ? 1ULL << CAP_NET_ADMIN : 0)) & ~minus; - r = detect_unified_cgroup_hierarchy(); - if (r < 0) - return r; - e = getenv("SYSTEMD_NSPAWN_CONTAINER_SERVICE"); if (e) arg_container_service_name = e; @@ -2970,6 +2973,10 @@ static int outer_child( if (r < 0) return r; + r = detect_unified_cgroup_hierarchy(directory); + if (r < 0) + return r; + if (arg_userns_mode != USER_NAMESPACE_NO) { /* Let the parent know which UID shift we read from the image */ l = send(uid_shift_socket, &arg_uid_shift, sizeof(arg_uid_shift), MSG_NOSIGNAL); -- cgit v1.2.3-54-g00ecf From a8725a06e6c5073325f2f6a7d6ef115c11d30a51 Mon Sep 17 00:00:00 2001 From: Zbigniew Jędrzejewski-Szmek Date: Sat, 8 Oct 2016 19:03:53 -0400 Subject: nspawn: also fall back to legacy cgroup hierarchy for old containers Current systemd version detection routine cannot detect systemd 230, only systmed >= 231. This means that we'll still use the legacy hierarchy in some cases where we wouldn't have too. If somebody figures out a nice way to detect systemd 230 this can be later improved. --- src/nspawn/nspawn.c | 16 ++++++++++++---- 1 file changed, 12 insertions(+), 4 deletions(-) (limited to 'src') diff --git a/src/nspawn/nspawn.c b/src/nspawn/nspawn.c index a775d1ac4f..dbf94b48c3 100644 --- a/src/nspawn/nspawn.c +++ b/src/nspawn/nspawn.c @@ -342,10 +342,18 @@ static int detect_unified_cgroup_hierarchy(const char *directory) { "Failed to determine whether the unified cgroups hierarchy is used: %m"); /* Otherwise inherit the default from the host system */ - if (all_unified > 0) - arg_unified_cgroup_hierarchy = CGROUP_UNIFIED_ALL; - else if (systemd_unified > 0) { - /* mixed cgroup hierarchy support was added in 232 */ + if (all_unified > 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 (systemd_unified > 0) { + /* Mixed cgroup hierarchy support was added in 232 */ r = systemd_installation_has_version(directory, 232); if (r < 0) return log_error_errno(r, "Failed to determine systemd version in container: %m"); -- cgit v1.2.3-54-g00ecf