X-Git-Url: https://www.chiark.greenend.org.uk/ucgi/~ianmdlvl/git?a=blobdiff_plain;f=src%2Fnspawn%2Fnspawn.c;h=a1d77244f8c38edab93f1139cff95a3af6355f1a;hb=e866af3acc30fcd1183a028ea3ef552b7237cc55;hp=7c3d7b89beb03890db7fc73a77f66f7206514c05;hpb=f418f31d5042398344740f2d8ac4dc1c2583151c;p=elogind.git diff --git a/src/nspawn/nspawn.c b/src/nspawn/nspawn.c index 7c3d7b89b..a1d77244f 100644 --- a/src/nspawn/nspawn.c +++ b/src/nspawn/nspawn.c @@ -84,6 +84,7 @@ #include "def.h" #include "rtnl-util.h" #include "udev-util.h" +#include "eventfd-util.h" #include "blkid-util.h" #include "gpt.h" #include "siphash24.h" @@ -92,6 +93,11 @@ #include "seccomp-util.h" #endif +typedef enum ContainerStatus { + CONTAINER_TERMINATED, + CONTAINER_REBOOTED +} ContainerStatus; + typedef enum LinkJournal { LINK_NO, LINK_AUTO, @@ -657,7 +663,7 @@ static int mount_binds(const char *dest, char **l, unsigned long flags) { return r; } } else { - log_error("Failed to bind mount %s: %s", *x, strerror(errno)); + log_error("Failed to bind mount %s: %m", *x); return -errno; } /* Create the mount point, but be conservative -- refuse to create block @@ -769,6 +775,15 @@ static int setup_resolv_conf(const char *dest) { return 0; } +static char* id128_format_as_uuid(sd_id128_t id, char s[37]) { + + snprintf(s, 37, + "%02x%02x%02x%02x-%02x%02x-%02x%02x-%02x%02x-%02x%02x%02x%02x%02x%02x", + SD_ID128_FORMAT_VAL(id)); + + return s; +} + static int setup_boot_id(const char *dest) { _cleanup_free_ char *from = NULL, *to = NULL; sd_id128_t rnd = {}; @@ -794,10 +809,7 @@ static int setup_boot_id(const char *dest) { return r; } - snprintf(as_uuid, sizeof(as_uuid), - "%02x%02x%02x%02x-%02x%02x-%02x%02x-%02x%02x-%02x%02x%02x%02x%02x%02x", - SD_ID128_FORMAT_VAL(rnd)); - char_array_0(as_uuid); + id128_format_as_uuid(rnd, as_uuid); r = write_string_file(from, as_uuid); if (r < 0) { @@ -986,7 +998,7 @@ static int setup_kmsg(const char *dest, int kmsg_socket) { /* Store away the fd in the socket, so that it stays open as * long as we run the child */ k = sendmsg(kmsg_socket, &mh, MSG_DONTWAIT|MSG_NOSIGNAL); - close_nointr_nofail(fd); + safe_close(fd); if (k < 0) { log_error("Failed to send FIFO fd: %m"); @@ -1137,10 +1149,8 @@ static int setup_journal(const char *directory) { } else if (access(p, F_OK) < 0) return 0; - if (dir_is_empty(q) == 0) { - log_error("%s not empty.", q); - return -ENOTEMPTY; - } + if (dir_is_empty(q) == 0) + log_warning("%s is not empty, proceeding anyway.", q); r = mkdir_p(q, 0755); if (r < 0) { @@ -1486,13 +1496,7 @@ static int setup_veth(pid_t pid, char iface_name[IFNAMSIZ]) { 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); + r = sd_rtnl_message_open_container_union(m, IFLA_INFO_DATA, "veth"); if (r < 0) { log_error("Failed to open netlink container: %s", strerror(-r)); return r; @@ -1581,6 +1585,12 @@ static int setup_bridge(const char veth_name[]) { return r; } + r = sd_rtnl_message_link_set_flags(m, IFF_UP, IFF_UP); + if (r < 0) { + log_error("Failed to set IFF_UP flag: %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)); @@ -1751,13 +1761,7 @@ static int setup_macvlan(pid_t pid) { 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); + r = sd_rtnl_message_open_container_union(m, IFLA_INFO_DATA, "macvlan"); if (r < 0) { log_error("Failed to open netlink container: %s", strerror(-r)); return r; @@ -2298,8 +2302,7 @@ static void loop_remove(int nr, int *image_fd) { if (image_fd && *image_fd >= 0) { ioctl(*image_fd, LOOP_CLR_FD); - close_nointr_nofail(*image_fd); - *image_fd = -1; + *image_fd = safe_close(*image_fd); } control = open("/dev/loop-control", O_RDWR|O_CLOEXEC|O_NOCTTY|O_NONBLOCK); @@ -2334,9 +2337,9 @@ static int spawn_getent(const char *database, const char *key, pid_t *rpid) { _exit(EXIT_FAILURE); if (pipe_fds[0] > 2) - close_nointr_nofail(pipe_fds[0]); + safe_close(pipe_fds[0]); if (pipe_fds[1] > 2) - close_nointr_nofail(pipe_fds[1]); + safe_close(pipe_fds[1]); nullfd = open("/dev/null", O_RDWR); if (nullfd < 0) @@ -2349,7 +2352,7 @@ static int spawn_getent(const char *database, const char *key, pid_t *rpid) { _exit(EXIT_FAILURE); if (nullfd > 2) - close_nointr_nofail(nullfd); + safe_close(nullfd); reset_all_signal_handlers(); close_all_fds(NULL, 0); @@ -2359,8 +2362,7 @@ static int spawn_getent(const char *database, const char *key, pid_t *rpid) { _exit(EXIT_FAILURE); } - close_nointr_nofail(pipe_fds[1]); - pipe_fds[1] = -1; + pipe_fds[1] = safe_close(pipe_fds[1]); *rpid = pid; @@ -2368,15 +2370,13 @@ static int spawn_getent(const char *database, const char *key, pid_t *rpid) { } static int change_uid_gid(char **_home) { - - _cleanup_strv_free_ char **passwd = NULL; char line[LINE_MAX], *w, *x, *state, *u, *g, *h; _cleanup_free_ uid_t *uids = NULL; _cleanup_free_ char *home = NULL; _cleanup_fclose_ FILE *f = NULL; _cleanup_close_ int fd = -1; unsigned n_uids = 0; - size_t sz, l; + size_t sz = 0, l; uid_t uid; gid_t gid; pid_t pid; @@ -2575,19 +2575,89 @@ static int change_uid_gid(char **_home) { return 0; } +/* + * Return 0 in case the container is being rebooted, has been shut + * down or exited successfully. On failures a negative value is + * returned. + * + * The status of the container "CONTAINER_TERMINATED" or + * "CONTAINER_REBOOTED" will be saved in the container argument + */ +static int wait_for_container(pid_t pid, ContainerStatus *container) { + int r; + siginfo_t status; + + r = wait_for_terminate(pid, &status); + if (r < 0) + return r; + + switch (status.si_code) { + case CLD_EXITED: + r = status.si_status; + if (r == 0) { + if (!arg_quiet) + log_debug("Container %s exited successfully.", + arg_machine); + + *container = CONTAINER_TERMINATED; + } else { + log_error("Container %s failed with error code %i.", + arg_machine, status.si_status); + r = -1; + } + break; + + case CLD_KILLED: + if (status.si_status == SIGINT) { + if (!arg_quiet) + log_info("Container %s has been shut down.", + arg_machine); + + *container = CONTAINER_TERMINATED; + r = 0; + break; + } else if (status.si_status == SIGHUP) { + if (!arg_quiet) + log_info("Container %s is being rebooted.", + arg_machine); + + *container = CONTAINER_REBOOTED; + r = 0; + break; + } + /* CLD_KILLED fallthrough */ + + case CLD_DUMPED: + log_error("Container %s terminated by signal %s.", + arg_machine, signal_to_string(status.si_status)); + r = -1; + break; + + default: + log_error("Container %s failed due to unknown reason.", + arg_machine); + r = -1; + break; + } + + return r; +} + +static void nop_handler(int sig) {} + int main(int argc, char *argv[]) { _cleanup_free_ char *kdbus_domain = NULL, *device_path = NULL, *root_device = NULL, *home_device = NULL, *srv_device = NULL; bool root_device_rw = true, home_device_rw = true, srv_device_rw = true; _cleanup_close_ int master = -1, kdbus_fd = -1, image_fd = -1; - _cleanup_close_pipe_ int kmsg_socket_pair[2] = { -1, -1 }; + _cleanup_close_pair_ int kmsg_socket_pair[2] = { -1, -1 }; _cleanup_fdset_free_ FDSet *fds = NULL; int r = EXIT_FAILURE, k, n_fd_passed, loop_nr = -1; const char *console = NULL; char veth_name[IFNAMSIZ]; bool secondary = false; + sigset_t mask, mask_chld; pid_t pid = 0; - sigset_t mask; log_parse_environment(); log_open(); @@ -2749,36 +2819,44 @@ int main(int argc, char *argv[]) { sd_notify(0, "READY=1"); assert_se(sigemptyset(&mask) == 0); + assert_se(sigemptyset(&mask_chld) == 0); + sigaddset(&mask_chld, SIGCHLD); sigset_add_many(&mask, SIGCHLD, SIGWINCH, SIGTERM, SIGINT, -1); assert_se(sigprocmask(SIG_BLOCK, &mask, NULL) == 0); for (;;) { - int parent_ready_fd = -1, child_ready_fd = -1; - siginfo_t status; - eventfd_t x; - - parent_ready_fd = eventfd(0, EFD_CLOEXEC); - if (parent_ready_fd < 0) { - log_error("Failed to create event fd: %m"); + ContainerStatus container_status; + int eventfds[2] = { -1, -1 }; + struct sigaction sa = { + .sa_handler = nop_handler, + .sa_flags = SA_NOCLDSTOP, + }; + + /* 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) { + log_error("Failed to change the signal mask: %m"); goto finish; } - child_ready_fd = eventfd(0, EFD_CLOEXEC); - if (child_ready_fd < 0) { - log_error("Failed to create event fd: %m"); + r = sigaction(SIGCHLD, &sa, NULL); + if (r < 0) { + log_error("Failed to install SIGCHLD handler: %m"); goto finish; } - pid = syscall(__NR_clone, - SIGCHLD|CLONE_NEWNS| - (arg_share_system ? 0 : CLONE_NEWIPC|CLONE_NEWPID|CLONE_NEWUTS)| - (arg_private_network ? CLONE_NEWNET : 0), NULL); + pid = clone_with_eventfd(SIGCHLD|CLONE_NEWNS| + (arg_share_system ? 0 : CLONE_NEWIPC|CLONE_NEWPID|CLONE_NEWUTS)| + (arg_private_network ? CLONE_NEWNET : 0), eventfds); 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"); else log_error("clone() failed: %m"); + r = pid; goto finish; } @@ -2804,15 +2882,13 @@ int main(int argc, char *argv[]) { if (envp[n_env]) n_env ++; - close_nointr_nofail(master); - master = -1; + master = safe_close(master); close_nointr(STDIN_FILENO); close_nointr(STDOUT_FILENO); close_nointr(STDERR_FILENO); - close_nointr_nofail(kmsg_socket_pair[0]); - kmsg_socket_pair[0] = -1; + kmsg_socket_pair[0] = safe_close(kmsg_socket_pair[0]); reset_all_signal_handlers(); @@ -2822,7 +2898,7 @@ int main(int argc, char *argv[]) { k = open_terminal(console, O_RDWR); if (k != STDIN_FILENO) { if (k >= 0) { - close_nointr_nofail(k); + safe_close(k); k = -EINVAL; } @@ -2895,8 +2971,7 @@ int main(int argc, char *argv[]) { if (setup_kmsg(arg_directory, kmsg_socket_pair[1]) < 0) goto child_fail; - close_nointr_nofail(kmsg_socket_pair[1]); - kmsg_socket_pair[1] = -1; + kmsg_socket_pair[1] = safe_close(kmsg_socket_pair[1]); if (setup_boot_id(arg_directory) < 0) goto child_fail; @@ -2922,9 +2997,11 @@ int main(int argc, char *argv[]) { /* Tell the parent that we are ready, and that * it can cgroupify us to that we lack access * to certain devices and resources. */ - eventfd_write(child_ready_fd, 1); - close_nointr_nofail(child_ready_fd); - child_ready_fd = -1; + r = eventfd_send_state(eventfds[1], + EVENTFD_CHILD_SUCCEEDED); + eventfds[1] = safe_close(eventfds[1]); + if (r < 0) + goto child_fail; if (chdir(arg_directory) < 0) { log_error("chdir(%s) failed: %m", arg_directory); @@ -2968,7 +3045,9 @@ int main(int argc, char *argv[]) { } if (!sd_id128_equal(arg_uuid, SD_ID128_NULL)) { - if (asprintf((char**)(envp + n_env++), "container_uuid=" SD_ID128_FORMAT_STR, SD_ID128_FORMAT_VAL(arg_uuid)) < 0) { + char as_uuid[37]; + + if (asprintf((char**)(envp + n_env++), "container_uuid=%s", id128_format_as_uuid(arg_uuid, as_uuid)) < 0) { log_oom(); goto child_fail; } @@ -3024,9 +3103,10 @@ int main(int argc, char *argv[]) { env_use = (char**) envp; /* Wait until the parent is ready with the setup, too... */ - eventfd_read(parent_ready_fd, &x); - close_nointr_nofail(parent_ready_fd); - parent_ready_fd = -1; + r = eventfd_parent_succeeded(eventfds[0]); + eventfds[0] = safe_close(eventfds[0]); + if (r < 0) + goto child_fail; if (arg_boot) { char **a; @@ -3057,17 +3137,27 @@ int main(int argc, char *argv[]) { log_error("execv() failed: %m"); child_fail: + /* Tell the parent that the setup failed, so he + * can clean up resources and terminate. */ + if (eventfds[1] != -1) + eventfd_send_state(eventfds[1], + EVENTFD_CHILD_FAILED); _exit(EXIT_FAILURE); } fdset_free(fds); fds = NULL; - /* Wait until the child reported that it is ready with - * all it needs to do with priviliges. After we got - * the notification we can make the process join its - * cgroup which might limit what it can do */ - eventfd_read(child_ready_fd, &x); + /* Wait for the child event: + * If EVENTFD_CHILD_FAILED, the child will terminate soon. + * If EVENTFD_CHILD_SUCCEEDED, the child is reporting that + * it is ready with all it needs to do with priviliges. + * After we got the notification we can make the process + * join its cgroup which might limit what it can do */ + r = eventfd_child_succeeded(eventfds[1]); + eventfds[1] = safe_close(eventfds[1]); + if (r < 0) + goto check_container_status; r = register_machine(pid); if (r < 0) @@ -3089,10 +3179,25 @@ int main(int argc, char *argv[]) { if (r < 0) goto finish; + /* Block SIGCHLD here, before notifying child. + * process_pty() will handle it with the other signals. */ + r = sigprocmask(SIG_BLOCK, &mask_chld, NULL); + if (r < 0) + goto finish; + + /* Reset signal to default */ + r = default_signals(SIGCHLD, -1); + if (r < 0) + goto finish; + /* Notify the child that the parent is ready with all - * its setup, and thtat the child can now hand over + * its setup, and that the child can now hand over * control to the code to run inside the container. */ - eventfd_write(parent_ready_fd, 1); + r = eventfd_send_state(eventfds[0], + EVENTFD_PARENT_SUCCEEDED); + eventfds[0] = safe_close(eventfds[0]); + if (r < 0) + goto finish; k = process_pty(master, &mask, arg_boot ? pid : 0, SIGRTMIN+3); if (k < 0) { @@ -3106,51 +3211,20 @@ int main(int argc, char *argv[]) { /* Kill if it is not dead yet anyway */ terminate_machine(pid); +check_container_status: /* Redundant, but better safe than sorry */ kill(pid, SIGKILL); - k = wait_for_terminate(pid, &status); + r = wait_for_container(pid, &container_status); pid = 0; - if (k < 0) { + if (r < 0) { r = EXIT_FAILURE; break; - } - - if (status.si_code == CLD_EXITED) { - r = status.si_status; - if (status.si_status != 0) { - log_error("Container %s failed with error code %i.", arg_machine, status.si_status); - break; - } - - if (!arg_quiet) - log_debug("Container %s exited successfully.", arg_machine); + } else if (container_status == CONTAINER_TERMINATED) break; - } else if (status.si_code == CLD_KILLED && - status.si_status == SIGINT) { - 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) { - - 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)); - r = EXIT_FAILURE; - break; - } else { - log_error("Container %s failed due to unknown reason.", arg_machine); - r = EXIT_FAILURE; - break; - } + /* CONTAINER_REBOOTED, loop again */ } finish: