X-Git-Url: https://www.chiark.greenend.org.uk/ucgi/~ianmdlvl/git?a=blobdiff_plain;f=src%2Fnspawn%2Fnspawn.c;h=73e98b71ae80c08692d78ea6b6b8663ca7a12644;hb=31938a8560a664c32a9d72f1fc2d4347b232e6e9;hp=56d9cc68c669275d5a1d06ae90be15c236932e5f;hpb=1b6d7fa742e303611dff8d7ebfa86ee5fb8b7dc7;p=elogind.git diff --git a/src/nspawn/nspawn.c b/src/nspawn/nspawn.c index 56d9cc68c..73e98b71a 100644 --- a/src/nspawn/nspawn.c +++ b/src/nspawn/nspawn.c @@ -89,6 +89,7 @@ #include "copy.h" #include "base-filesystem.h" #include "barrier.h" +#include "event-util.h" #ifdef HAVE_SECCOMP #include "seccomp-util.h" @@ -123,6 +124,7 @@ static bool arg_private_network = false; static bool arg_read_only = false; static bool arg_boot = false; static LinkJournal arg_link_journal = LINK_AUTO; +static bool arg_link_journal_try = false; static uint64_t arg_retain = (1ULL << CAP_CHOWN) | (1ULL << CAP_DAC_OVERRIDE) | @@ -201,8 +203,9 @@ static void help(void) { " --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" + " --link-journal=MODE Link up guest journal, one of no, auto, guest, host,\n" + " try-guest, try-host\n" + " -j Equivalent to --link-journal=try-guest\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" @@ -427,6 +430,7 @@ static int parse_argv(int argc, char *argv[]) { case 'j': arg_link_journal = LINK_GUEST; + arg_link_journal_try = true; break; case ARG_LINK_JOURNAL: @@ -438,7 +442,13 @@ static int parse_argv(int argc, char *argv[]) { arg_link_journal = LINK_GUEST; else if (streq(optarg, "host")) arg_link_journal = LINK_HOST; - else { + else if (streq(optarg, "try-guest")) { + arg_link_journal = LINK_GUEST; + arg_link_journal_try = true; + } else if (streq(optarg, "try-host")) { + arg_link_journal = LINK_HOST; + arg_link_journal_try = true; + } else { log_error("Failed to parse link journal mode %s", optarg); return -EINVAL; } @@ -663,7 +673,7 @@ static int mount_all(const char *dest) { t = path_is_mount_point(where, true); if (t < 0) { - log_error("Failed to detect whether %s is a mount point: %s", where, strerror(-t)); + log_error_errno(t, "Failed to detect whether %s is a mount point: %m", where); if (r == 0) r = t; @@ -675,7 +685,18 @@ static int mount_all(const char *dest) { if (mount_table[k].what && t > 0) continue; - mkdir_p(where, 0755); + t = mkdir_p(where, 0755); + if (t < 0) { + if (mount_table[k].fatal) { + log_error_errno(t, "Failed to create directory %s: %m", where); + + if (r == 0) + r = t; + } else + log_warning_errno(t, "Failed to create directory %s: %m", where); + + continue; + } #ifdef HAVE_SELINUX if (arg_selinux_apifs_context && @@ -694,13 +715,15 @@ static int mount_all(const char *dest) { where, mount_table[k].type, mount_table[k].flags, - o) < 0 && - mount_table[k].fatal) { + o) < 0) { - log_error("mount(%s) failed: %m", where); + if (mount_table[k].fatal) { + log_error("mount(%s) failed: %m", where); - if (r == 0) - r = -errno; + if (r == 0) + r = -errno; + } else + log_warning("mount(%s) failed: %m", where); } } @@ -733,7 +756,7 @@ static int mount_binds(const char *dest, char **l, bool ro) { } else if (errno == ENOENT) { r = mkdir_parents_label(where, 0755); if (r < 0) { - log_error("Failed to bind mount %s: %s", *x, strerror(-r)); + log_error_errno(r, "Failed to bind mount %s: %m", *x); return r; } } else { @@ -743,15 +766,35 @@ static int mount_binds(const char *dest, char **l, bool ro) { /* Create the mount point, but be conservative -- refuse to create block * and char devices. */ - if (S_ISDIR(source_st.st_mode)) - mkdir_label(where, 0755); - else if (S_ISFIFO(source_st.st_mode)) - mkfifo(where, 0644); - else if (S_ISSOCK(source_st.st_mode)) - mknod(where, 0644 | S_IFSOCK, 0); - else if (S_ISREG(source_st.st_mode)) - touch(where); - else { + if (S_ISDIR(source_st.st_mode)) { + r = mkdir_label(where, 0755); + if (r < 0 && errno != EEXIST) { + log_error_errno(r, "Failed to create mount point %s: %m", where); + + return r; + } + } else if (S_ISFIFO(source_st.st_mode)) { + r = mkfifo(where, 0644); + if (r < 0 && errno != EEXIST) { + log_error("Failed to create mount point %s: %m", where); + + return -errno; + } + } else if (S_ISSOCK(source_st.st_mode)) { + r = mknod(where, 0644 | S_IFSOCK, 0); + if (r < 0 && errno != EEXIST) { + log_error("Failed to create mount point %s: %m", where); + + return -errno; + } + } else if (S_ISREG(source_st.st_mode)) { + r = touch(where); + if (r < 0) { + log_error_errno(r, "Failed to create mount point %s: %m", where); + + return r; + } + } else { log_error("Refusing to create mountpoint for file: %s", *x); return -ENOTSUP; } @@ -764,7 +807,7 @@ static int mount_binds(const char *dest, char **l, bool ro) { if (ro) { r = bind_remount_recursive(where, true); if (r < 0) { - log_error("Read-Only bind mount failed: %s", strerror(-r)); + log_error_errno(r, "Read-Only bind mount failed: %m"); return r; } } @@ -778,12 +821,18 @@ static int mount_tmpfs(const char *dest) { STRV_FOREACH_PAIR(i, o, arg_tmpfs) { _cleanup_free_ char *where = NULL; + int r; where = strappend(dest, *i); if (!where) return log_oom(); - mkdir_label(where, 0755); + r = mkdir_label(where, 0755); + if (r < 0 && errno != EEXIST) { + log_error_errno(r, "creating mount point for tmpfs %s failed: %m", where); + + return r; + } if (mount("tmpfs", where, "tmpfs", MS_NODEV|MS_STRICTATIME, *o) < 0) { log_error("tmpfs mount to %s failed: %m", where); @@ -844,8 +893,19 @@ static int setup_timezone(const char *dest) { if (!what) return log_oom(); - mkdir_parents(where, 0755); - unlink(where); + r = mkdir_parents(where, 0755); + if (r < 0) { + log_error_errno(r, "Failed to create directory for timezone info %s in container: %m", where); + + return 0; + } + + r = unlink(where); + if (r < 0 && errno != ENOENT) { + log_error("Failed to remove existing timezone info %s in container: %m", where); + + return 0; + } if (symlink(what, where) < 0) { log_error("Failed to correct timezone of container: %m"); @@ -857,6 +917,7 @@ static int setup_timezone(const char *dest) { static int setup_resolv_conf(const char *dest) { _cleanup_free_ char *where = NULL; + int r; assert(dest); @@ -870,8 +931,19 @@ static int setup_resolv_conf(const char *dest) { /* We don't really care for the results of this really. If it * fails, it fails, but meh... */ - mkdir_parents(where, 0755); - copy_file("/etc/resolv.conf", where, O_TRUNC|O_NOFOLLOW, 0644); + r = mkdir_parents(where, 0755); + if (r < 0) { + log_warning_errno(r, "Failed to create parent directory for resolv.conf %s: %m", where); + + return 0; + } + + r = copy_file("/etc/resolv.conf", where, O_TRUNC|O_NOFOLLOW, 0644); + if (r < 0) { + log_warning_errno(r, "Failed to copy /etc/resolv.conf to %s: %m", where); + + return 0; + } return 0; } @@ -890,12 +962,16 @@ static int setup_volatile_state(const char *directory) { r = bind_remount_recursive(directory, true); if (r < 0) { - log_error("Failed to remount %s read-only: %s", directory, strerror(-r)); + log_error_errno(r, "Failed to remount %s read-only: %m", directory); return r; } p = strappenda(directory, "/var"); - mkdir(p, 0755); + r = mkdir(p, 0755); + if (r < 0 && errno != EEXIST) { + log_error("Failed to create %s: %m", directory); + return -errno; + } if (mount("tmpfs", p, "tmpfs", MS_STRICTATIME, "mode=755") < 0) { log_error("Failed to mount tmpfs to /var: %m"); @@ -935,7 +1011,13 @@ static int setup_volatile(const char *directory) { f = strappenda(directory, "/usr"); t = strappenda(template, "/usr"); - mkdir(t, 0755); + r = mkdir(t, 0755); + if (r < 0 && errno != EEXIST) { + log_error("Failed to create %s: %m", t); + r = -errno; + goto fail; + } + if (mount(f, t, "bind", MS_BIND|MS_REC, NULL) < 0) { log_error("Failed to create /usr bind mount: %m"); r = -errno; @@ -946,7 +1028,7 @@ static int setup_volatile(const char *directory) { r = bind_remount_recursive(t, true); if (r < 0) { - log_error("Failed to remount %s read-only: %s", t, strerror(-r)); + log_error_errno(r, "Failed to remount %s read-only: %m", t); goto fail; } @@ -999,7 +1081,7 @@ static int setup_boot_id(const char *dest) { r = sd_id128_randomize(&rnd); if (r < 0) { - log_error("Failed to generate random boot id: %s", strerror(-r)); + log_error_errno(r, "Failed to generate random boot id: %m"); return r; } @@ -1007,7 +1089,7 @@ static int setup_boot_id(const char *dest) { r = write_string_file(from, as_uuid); if (r < 0) { - log_error("Failed to write boot id: %s", strerror(-r)); + log_error_errno(r, "Failed to write boot id: %m"); return r; } @@ -1029,7 +1111,8 @@ static int copy_devnodes(const char *dest) { "full\0" "random\0" "urandom\0" - "tty\0"; + "tty\0" + "net/tun\0"; const char *d; int r = 0; @@ -1060,10 +1143,17 @@ static int copy_devnodes(const char *dest) { log_error("%s is not a char or block device, cannot copy", from); return -EIO; - } else if (mknod(to, st.st_mode, st.st_rdev) < 0) { + } else { + r = mkdir_parents(to, 0775); + if (r < 0) { + log_error_errno(r, "Failed to create parent directory of %s: %m", to); + return -r; + } - log_error("mknod(%s) failed: %m", dest); - return -errno; + if (mknod(to, st.st_mode, st.st_rdev) < 0) { + log_error("mknod(%s) failed: %m", dest); + return -errno; + } } } @@ -1103,7 +1193,7 @@ static int setup_dev_console(const char *dest, const char *console) { r = chmod_and_chown(console, 0600, 0, 0); if (r < 0) { - log_error("Failed to correct access mode for TTY: %s", strerror(-r)); + log_error_errno(r, "Failed to correct access mode for TTY: %m"); return r; } @@ -1166,7 +1256,7 @@ static int setup_kmsg(const char *dest, int kmsg_socket) { r = chmod_and_chown(from, 0600, 0, 0); if (r < 0) { - log_error("Failed to correct access mode for /dev/kmsg: %s", strerror(-r)); + log_error_errno(r, "Failed to correct access mode for /dev/kmsg: %m"); return r; } @@ -1209,7 +1299,7 @@ static int setup_hostname(void) { if (arg_share_system) return 0; - if (sethostname(arg_machine, strlen(arg_machine)) < 0) + if (sethostname_idempotent(arg_machine) < 0) return -errno; return 0; @@ -1229,7 +1319,7 @@ static int setup_journal(const char *directory) { if (r == -ENOENT && arg_link_journal == LINK_AUTO) return 0; else if (r < 0) { - log_error("Failed to read machine ID from %s: %s", p, strerror(-r)); + log_error_errno(r, "Failed to read machine ID from %s: %m", p); return r; } @@ -1240,13 +1330,13 @@ static int setup_journal(const char *directory) { /* Verify validity */ r = sd_id128_from_string(id, &machine_id); if (r < 0) { - log_error("Failed to parse machine ID from %s: %s", p, strerror(-r)); + log_error_errno(r, "Failed to parse machine ID from %s: %m", p); return r; } r = sd_id128_get_machine(&this_id); if (r < 0) { - log_error("Failed to retrieve machine ID: %s", strerror(-r)); + log_error_errno(r, "Failed to retrieve machine ID: %m"); return r; } @@ -1294,7 +1384,7 @@ static int setup_journal(const char *directory) { r = mkdir_p(q, 0755); if (r < 0) - log_warning("failed to create directory %s: %m", q); + log_warning("Failed to create directory %s: %m", q); return 0; } @@ -1323,21 +1413,33 @@ static int setup_journal(const char *directory) { if (arg_link_journal == LINK_GUEST) { if (symlink(q, p) < 0) { - log_error("Failed to symlink %s to %s: %m", q, p); - return -errno; + if (arg_link_journal_try) { + log_debug("Failed to symlink %s to %s, skipping journal setup: %m", q, p); + return 0; + } else { + log_error("Failed to symlink %s to %s: %m", q, p); + return -errno; + } } r = mkdir_p(q, 0755); if (r < 0) - log_warning("failed to create directory %s: %m", q); + log_warning("Failed to create directory %s: %m", q); return 0; } if (arg_link_journal == LINK_HOST) { - r = mkdir_p(p, 0755); + /* don't create parents here -- if the host doesn't have + * permanent journal set up, don't force it here */ + r = mkdir(p, 0755); if (r < 0) { - log_error("Failed to create %s: %m", p); - return r; + if (arg_link_journal_try) { + log_debug("Failed to create %s, skipping journal setup: %m", p); + return 0; + } else { + log_error("Failed to create %s: %m", p); + return r; + } } } else if (access(p, F_OK) < 0) @@ -1360,26 +1462,6 @@ static int setup_journal(const char *directory) { return 0; } -static int setup_kdbus(const char *dest, const char *path) { - const char *p; - - if (!path) - return 0; - - p = strappenda(dest, "/dev/kdbus"); - if (mkdir(p, 0755) < 0) { - log_error("Failed to create kdbus path: %m"); - return -errno; - } - - if (mount(path, p, "bind", MS_BIND, NULL) < 0) { - log_error("Failed to mount kdbus domain path: %m"); - return -errno; - } - - return 0; -} - static int drop_capabilities(void) { return capability_bounding_set_drop(~arg_retain, false); } @@ -1394,7 +1476,7 @@ static int register_machine(pid_t pid, int local_ifindex) { r = sd_bus_default_system(&bus); if (r < 0) { - log_error("Failed to open system bus: %s", strerror(-r)); + log_error_errno(r, "Failed to open system bus: %m"); return r; } @@ -1426,7 +1508,7 @@ static int register_machine(pid_t pid, int local_ifindex) { "org.freedesktop.machine1.Manager", "CreateMachineWithNetwork"); if (r < 0) { - log_error("Failed to create message: %s", strerror(-r)); + log_error_errno(r, "Failed to create message: %m"); return r; } @@ -1441,31 +1523,31 @@ static int register_machine(pid_t pid, int local_ifindex) { strempty(arg_directory), local_ifindex > 0 ? 1 : 0, local_ifindex); if (r < 0) { - log_error("Failed to append message arguments: %s", strerror(-r)); + log_error_errno(r, "Failed to append message arguments: %m"); return r; } r = sd_bus_message_open_container(m, 'a', "(sv)"); if (r < 0) { - log_error("Failed to open container: %s", strerror(-r)); + log_error_errno(r, "Failed to open container: %m"); return r; } if (!isempty(arg_slice)) { r = sd_bus_message_append(m, "(sv)", "Slice", "s", arg_slice); if (r < 0) { - log_error("Failed to append slice: %s", strerror(-r)); + log_error_errno(r, "Failed to append slice: %m"); return r; } } r = sd_bus_message_append(m, "(sv)", "DevicePolicy", "s", "strict"); if (r < 0) { - log_error("Failed to add device policy: %s", strerror(-r)); + log_error_errno(r, "Failed to add device policy: %m"); return r; } - r = sd_bus_message_append(m, "(sv)", "DeviceAllow", "a(ss)", 10, + r = sd_bus_message_append(m, "(sv)", "DeviceAllow", "a(ss)", 9, /* Allow the container to * access and create the API * device nodes, so that @@ -1478,32 +1560,22 @@ static int register_machine(pid_t pid, int local_ifindex) { "/dev/random", "rwm", "/dev/urandom", "rwm", "/dev/tty", "rwm", + "/dev/net/tun", "rwm", /* Allow the container * access to ptys. However, * do not permit the * container to ever create * these device nodes. */ "/dev/pts/ptmx", "rw", - "char-pts", "rw", - /* Allow the container - * access to all kdbus - * devices. Again, the - * container cannot create - * these nodes, only use - * them. We use a pretty - * open match here, so that - * the kernel API can still - * change. */ - "char-kdbus", "rw", - "char-kdbus/*", "rw"); + "char-pts", "rw"); if (r < 0) { - log_error("Failed to add device whitelist: %s", strerror(-r)); + log_error_errno(r, "Failed to add device whitelist: %m"); return r; } r = sd_bus_message_close_container(m); if (r < 0) { - log_error("Failed to close container: %s", strerror(-r)); + log_error_errno(r, "Failed to close container: %m"); return r; } @@ -1530,7 +1602,7 @@ static int terminate_machine(pid_t pid) { r = sd_bus_default_system(&bus); if (r < 0) { - log_error("Failed to open system bus: %s", strerror(-r)); + log_error_errno(r, "Failed to open system bus: %m"); return r; } @@ -1584,7 +1656,7 @@ static int reset_audit_loginuid(void) { if (r == -ENOENT) return 0; if (r < 0) { - log_error("Failed to read /proc/self/loginuid: %s", strerror(-r)); + log_error_errno(r, "Failed to read /proc/self/loginuid: %m"); return r; } @@ -1657,7 +1729,7 @@ static int setup_veth(pid_t pid, char iface_name[IFNAMSIZ], int *ifi) { /* Use two different interface name prefixes depending whether * we are in bridge mode or not. */ - snprintf(iface_name, IFNAMSIZ, "%s-%s", + snprintf(iface_name, IFNAMSIZ - 1, "%s-%s", arg_network_bridge ? "vb" : "ve", arg_machine); r = generate_mac(&mac_container, CONTAINER_HASH_KEY); @@ -1674,85 +1746,85 @@ static int setup_veth(pid_t pid, char iface_name[IFNAMSIZ], int *ifi) { r = sd_rtnl_open(&rtnl, 0); if (r < 0) { - log_error("Failed to connect to netlink: %s", strerror(-r)); + log_error_errno(r, "Failed to connect to netlink: %m"); return r; } r = sd_rtnl_message_new_link(rtnl, &m, RTM_NEWLINK, 0); if (r < 0) { - log_error("Failed to allocate netlink message: %s", strerror(-r)); + log_error_errno(r, "Failed to allocate netlink message: %m"); 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)); + log_error_errno(r, "Failed to add netlink interface name: %m"); return r; } r = sd_rtnl_message_append_ether_addr(m, IFLA_ADDRESS, &mac_host); if (r < 0) { - log_error("Failed to add netlink MAC address: %s", strerror(-r)); + log_error_errno(r, "Failed to add netlink MAC address: %m"); return r; } r = sd_rtnl_message_open_container(m, IFLA_LINKINFO); if (r < 0) { - log_error("Failed to open netlink container: %s", strerror(-r)); + log_error_errno(r, "Failed to open netlink container: %m"); return r; } 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)); + log_error_errno(r, "Failed to open netlink container: %m"); 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)); + log_error_errno(r, "Failed to open netlink container: %m"); 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)); + log_error_errno(r, "Failed to add netlink interface name: %m"); return r; } r = sd_rtnl_message_append_ether_addr(m, IFLA_ADDRESS, &mac_container); if (r < 0) { - log_error("Failed to add netlink MAC address: %s", strerror(-r)); + log_error_errno(r, "Failed to add netlink MAC address: %m"); 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)); + log_error_errno(r, "Failed to add netlink namespace field: %m"); return r; } r = sd_rtnl_message_close_container(m); if (r < 0) { - log_error("Failed to close netlink container: %s", strerror(-r)); + log_error_errno(r, "Failed to close netlink container: %m"); return r; } r = sd_rtnl_message_close_container(m); if (r < 0) { - log_error("Failed to close netlink container: %s", strerror(-r)); + log_error_errno(r, "Failed to close netlink container: %m"); return r; } r = sd_rtnl_message_close_container(m); if (r < 0) { - log_error("Failed to close netlink container: %s", strerror(-r)); + log_error_errno(r, "Failed to close netlink container: %m"); return r; } r = sd_rtnl_call(rtnl, m, 0, NULL); if (r < 0) { - log_error("Failed to add new veth interfaces: %s", strerror(-r)); + log_error_errno(r, "Failed to add new veth interfaces: %m"); return r; } @@ -1791,37 +1863,37 @@ static int setup_bridge(const char veth_name[], int *ifi) { r = sd_rtnl_open(&rtnl, 0); if (r < 0) { - log_error("Failed to connect to netlink: %s", strerror(-r)); + log_error_errno(r, "Failed to connect to netlink: %m"); return r; } r = sd_rtnl_message_new_link(rtnl, &m, RTM_SETLINK, 0); if (r < 0) { - log_error("Failed to allocate netlink message: %s", strerror(-r)); + log_error_errno(r, "Failed to allocate netlink message: %m"); 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)); + log_error_errno(r, "Failed to set IFF_UP flag: %m"); 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)); + log_error_errno(r, "Failed to add netlink interface name field: %m"); 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)); + log_error_errno(r, "Failed to add netlink master field: %m"); 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)); + log_error_errno(r, "Failed to add veth interface to bridge: %m"); return r; } @@ -1868,7 +1940,7 @@ static int move_network_interfaces(pid_t pid) { r = sd_rtnl_open(&rtnl, 0); if (r < 0) { - log_error("Failed to connect to netlink: %s", strerror(-r)); + log_error_errno(r, "Failed to connect to netlink: %m"); return r; } @@ -1886,21 +1958,21 @@ static int move_network_interfaces(pid_t pid) { if (ifi < 0) return ifi; - r = sd_rtnl_message_new_link(rtnl, &m, RTM_NEWLINK, ifi); + r = sd_rtnl_message_new_link(rtnl, &m, RTM_SETLINK, ifi); if (r < 0) { - log_error("Failed to allocate netlink message: %s", strerror(-r)); + log_error_errno(r, "Failed to allocate netlink message: %m"); 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)); + log_error_errno(r, "Failed to append namespace PID to netlink message: %m"); 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)); + log_error_errno(r, "Failed to move interface %s to namespace: %m", *i); return r; } } @@ -1922,7 +1994,7 @@ static int setup_macvlan(pid_t pid) { r = sd_rtnl_open(&rtnl, 0); if (r < 0) { - log_error("Failed to connect to netlink: %s", strerror(-r)); + log_error_errno(r, "Failed to connect to netlink: %m"); return r; } @@ -1943,13 +2015,13 @@ static int setup_macvlan(pid_t pid) { r = sd_rtnl_message_new_link(rtnl, &m, RTM_NEWLINK, 0); if (r < 0) { - log_error("Failed to allocate netlink message: %s", strerror(-r)); + log_error_errno(r, "Failed to allocate netlink message: %m"); return r; } r = sd_rtnl_message_append_u32(m, IFLA_LINK, ifi); if (r < 0) { - log_error("Failed to add netlink interface index: %s", strerror(-r)); + log_error_errno(r, "Failed to add netlink interface index: %m"); return r; } @@ -1961,49 +2033,49 @@ static int setup_macvlan(pid_t pid) { r = sd_rtnl_message_append_string(m, IFLA_IFNAME, n); if (r < 0) { - log_error("Failed to add netlink interface name: %s", strerror(-r)); + log_error_errno(r, "Failed to add netlink interface name: %m"); 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)); + log_error_errno(r, "Failed to add netlink namespace field: %m"); return r; } r = sd_rtnl_message_open_container(m, IFLA_LINKINFO); if (r < 0) { - log_error("Failed to open netlink container: %s", strerror(-r)); + log_error_errno(r, "Failed to open netlink container: %m"); return r; } 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)); + log_error_errno(r, "Failed to open netlink container: %m"); return r; } r = sd_rtnl_message_append_u32(m, IFLA_MACVLAN_MODE, MACVLAN_MODE_BRIDGE); if (r < 0) { - log_error("Failed to append macvlan mode: %s", strerror(-r)); + log_error_errno(r, "Failed to append macvlan mode: %m"); return r; } r = sd_rtnl_message_close_container(m); if (r < 0) { - log_error("Failed to close netlink container: %s", strerror(-r)); + log_error_errno(r, "Failed to close netlink container: %m"); return r; } r = sd_rtnl_message_close_container(m); if (r < 0) { - log_error("Failed to close netlink container: %s", strerror(-r)); + log_error_errno(r, "Failed to close netlink container: %m"); return r; } r = sd_rtnl_call(rtnl, m, 0, NULL); if (r < 0) { - log_error("Failed to add new macvlan interfaces: %s", strerror(-r)); + log_error_errno(r, "Failed to add new macvlan interfaces: %m"); return r; } } @@ -2036,7 +2108,7 @@ static int setup_seccomp(void) { r = seccomp_add_secondary_archs(seccomp); if (r < 0) { - log_error("Failed to add secondary archs to seccomp filter: %s", strerror(-r)); + log_error_errno(r, "Failed to add secondary archs to seccomp filter: %m"); goto finish; } @@ -2045,7 +2117,7 @@ static int setup_seccomp(void) { if (r == -EFAULT) continue; /* unknown syscall */ if (r < 0) { - log_error("Failed to block syscall: %s", strerror(-r)); + log_error_errno(r, "Failed to block syscall: %m"); goto finish; } } @@ -2068,19 +2140,19 @@ static int setup_seccomp(void) { 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)); + log_error_errno(r, "Failed to add audit seccomp rule: %m"); 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)); + log_error_errno(r, "Failed to unset NO_NEW_PRIVS: %m"); goto finish; } r = seccomp_load(seccomp); if (r < 0) - log_error("Failed to install seccomp audit filter: %s", strerror(-r)); + log_error_errno(r, "Failed to install seccomp audit filter: %m"); finish: seccomp_release(seccomp); @@ -2278,7 +2350,7 @@ static int dissect_image( r = udev_enumerate_scan_devices(e); if (r < 0) { - log_error("Failed to scan for partition devices of %s: %s", arg_image, strerror(-r)); + log_error_errno(r, "Failed to scan for partition devices of %s: %m", arg_image); return r; } @@ -2509,7 +2581,7 @@ static int mount_devices( if (root_device) { r = mount_device(root_device, arg_directory, NULL, root_device_rw); if (r < 0) { - log_error("Failed to mount root directory: %s", strerror(-r)); + log_error_errno(r, "Failed to mount root directory: %m"); return r; } } @@ -2517,7 +2589,7 @@ static int mount_devices( if (home_device) { r = mount_device(home_device, arg_directory, "/home", home_device_rw); if (r < 0) { - log_error("Failed to mount home directory: %s", strerror(-r)); + log_error_errno(r, "Failed to mount home directory: %m"); return r; } } @@ -2525,7 +2597,7 @@ static int mount_devices( if (srv_device) { r = mount_device(srv_device, arg_directory, "/srv", srv_device_rw); if (r < 0) { - log_error("Failed to mount server data directory: %s", strerror(-r)); + log_error_errno(r, "Failed to mount server data directory: %m"); return r; } } @@ -2535,20 +2607,27 @@ static int mount_devices( static void loop_remove(int nr, int *image_fd) { _cleanup_close_ int control = -1; + int r; if (nr < 0) return; if (image_fd && *image_fd >= 0) { - ioctl(*image_fd, LOOP_CLR_FD); + r = ioctl(*image_fd, LOOP_CLR_FD); + if (r < 0) + log_warning("Failed to close loop image: %m"); *image_fd = safe_close(*image_fd); } control = open("/dev/loop-control", O_RDWR|O_CLOEXEC|O_NOCTTY|O_NONBLOCK); - if (control < 0) + if (control < 0) { + log_warning("Failed to open /dev/loop-control: %m"); return; + } - ioctl(control, LOOP_CTL_REMOVE, nr); + r = ioctl(control, LOOP_CTL_REMOVE, nr); + if (r < 0) + log_warning("Failed to remove loop %d: %m", nr); } static int spawn_getent(const char *database, const char *key, pid_t *rpid) { @@ -2778,13 +2857,13 @@ static int change_uid_gid(char **_home) { r = mkdir_parents(home, 0775); if (r < 0) { - log_error("Failed to make home root directory: %s", strerror(-r)); + log_error_errno(r, "Failed to make home root directory: %m"); return r; } r = mkdir_safe(home, 0755, uid, gid); if (r < 0 && r != -EEXIST) { - log_error("Failed to make home directory: %s", strerror(-r)); + log_error_errno(r, "Failed to make home directory: %m"); return r; } @@ -2823,8 +2902,8 @@ static int change_uid_gid(char **_home) { * container argument. * > 0 : The program executed in the container terminated with an * error. The exit code of the program executed in the - * container is returned. No change is made to the container - * argument. + * container is returned. The container argument has been set + * to CONTAINER_TERMINATED. * 0 : The container is being rebooted, has been shut down or exited * successfully. The container argument has been set to either * CONTAINER_TERMINATED or CONTAINER_REBOOTED. @@ -2833,61 +2912,50 @@ static int change_uid_gid(char **_home) { * error is indicated by a non-zero value. */ static int wait_for_container(pid_t pid, ContainerStatus *container) { - int r; siginfo_t status; + int r; r = wait_for_terminate(pid, &status); if (r < 0) { - log_warning("Failed to wait for container: %s", strerror(-r)); + log_warning_errno(r, "Failed to wait for container: %m"); 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); + if (status.si_status == 0) { + log_full(arg_quiet ? LOG_DEBUG : LOG_INFO, "Container %s exited successfully.", arg_machine); - *container = CONTAINER_TERMINATED; - } else { - log_error("Container %s failed with error code %i.", - arg_machine, status.si_status); - } - break; + } else + log_full(arg_quiet ? LOG_DEBUG : LOG_INFO, "Container %s failed with error code %i.", arg_machine, status.si_status); + + *container = CONTAINER_TERMINATED; + return status.si_status; case CLD_KILLED: if (status.si_status == SIGINT) { - if (!arg_quiet) - log_info("Container %s has been shut down.", - arg_machine); + log_full(arg_quiet ? LOG_DEBUG : LOG_INFO, "Container %s has been shut down.", arg_machine); *container = CONTAINER_TERMINATED; - r = 0; - break; + return 0; + } else if (status.si_status == SIGHUP) { - if (!arg_quiet) - log_info("Container %s is being rebooted.", - arg_machine); + log_full(arg_quiet ? LOG_DEBUG : LOG_INFO, "Container %s is being rebooted.", arg_machine); *container = CONTAINER_REBOOTED; - r = 0; - break; + return 0; } + /* 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; + log_error("Container %s terminated by signal %s.", arg_machine, signal_to_string(status.si_status)); + return -EIO; default: - log_error("Container %s failed due to unknown reason.", - arg_machine); - r = -1; - break; + log_error("Container %s failed due to unknown reason.", arg_machine); + return -EIO; } return r; @@ -2895,11 +2963,27 @@ static int wait_for_container(pid_t pid, ContainerStatus *container) { static void nop_handler(int sig) {} +static int on_orderly_shutdown(sd_event_source *s, const struct signalfd_siginfo *si, void *userdata) { + pid_t pid; + + pid = PTR_TO_UINT32(userdata); + if (pid > 0) { + if (kill(pid, SIGRTMIN+3) >= 0) { + log_info("Trying to halt container. Send SIGTERM again to trigger immediate termination."); + sd_event_source_set_userdata(s, NULL); + return 0; + } + } + + sd_event_exit(sd_event_source_get_event(s), 0); + return 0; +} + int main(int argc, char *argv[]) { - _cleanup_free_ char *kdbus_domain = NULL, *device_path = NULL, *root_device = NULL, *home_device = NULL, *srv_device = NULL; + _cleanup_free_ char *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_ int master = -1, image_fd = -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; @@ -2966,7 +3050,7 @@ int main(int argc, char *argv[]) { if (n_fd_passed > 0) { k = fdset_new_listen_fds(&fds, false); if (k < 0) { - log_error("Failed to collect file descriptors: %s", strerror(-k)); + log_error_errno(k, "Failed to collect file descriptors: %m"); goto finish; } } @@ -3046,26 +3130,6 @@ int main(int argc, char *argv[]) { goto finish; } - 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"); goto finish; @@ -3076,11 +3140,12 @@ int main(int argc, char *argv[]) { "STATUS=Container running."); 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); + assert_se(sigemptyset(&mask_chld) == 0); + assert_se(sigaddset(&mask_chld, SIGCHLD) == 0); + for (;;) { ContainerStatus container_status; _cleanup_(barrier_destroy) Barrier barrier = BARRIER_NULL; @@ -3091,7 +3156,7 @@ int main(int argc, char *argv[]) { r = barrier_create(&barrier); if (r < 0) { - log_error("Cannot initialize IPC barrier: %s", strerror(-r)); + log_error_errno(r, "Cannot initialize IPC barrier: %m"); goto finish; } @@ -3165,7 +3230,7 @@ int main(int argc, char *argv[]) { k = -EINVAL; } - log_error("Failed to open console: %s", strerror(-k)); + log_error_errno(k, "Failed to open console: %m"); _exit(EXIT_FAILURE); } @@ -3222,7 +3287,7 @@ int main(int argc, char *argv[]) { if (arg_read_only) { k = bind_remount_recursive(arg_directory, true); if (k < 0) { - log_error("Failed to make tree read-only: %s", strerror(-k)); + log_error_errno(k, "Failed to make tree read-only: %m"); _exit(EXIT_FAILURE); } } @@ -3270,13 +3335,10 @@ int main(int argc, char *argv[]) { if (mount_tmpfs(arg_directory) < 0) _exit(EXIT_FAILURE); - if (setup_kdbus(arg_directory, kdbus_domain) < 0) - _exit(EXIT_FAILURE); - /* Tell the parent that we are ready, and that * it can cgroupify us to that we lack access * to certain devices and resources. */ - barrier_place(&barrier); + (void)barrier_place(&barrier); if (chdir(arg_directory) < 0) { log_error("chdir(%s) failed: %m", arg_directory); @@ -3417,6 +3479,8 @@ int main(int argc, char *argv[]) { /* wait for child-setup to be done */ if (barrier_place_and_sync(&barrier)) { + _cleanup_event_unref_ sd_event *event = NULL; + _cleanup_(pty_forward_freep) PTYForward *forward = NULL; int ifi = 0; r = move_network_interfaces(pid); @@ -3453,14 +3517,41 @@ int main(int argc, char *argv[]) { /* Notify the child that the parent is ready with all * its setup, and that the child can now hand over * control to the code to run inside the container. */ - barrier_place(&barrier); + (void)barrier_place(&barrier); - k = process_pty(master, &mask, arg_boot ? pid : 0, SIGRTMIN+3); - if (k < 0) { - r = EXIT_FAILURE; - break; + r = sd_event_new(&event); + if (r < 0) { + log_error_errno(r, "Failed to get default event source: %m"); + goto finish; + } + + if (arg_boot) { + /* Try to kill the init system on SIGINT or SIGTERM */ + sd_event_add_signal(event, NULL, SIGINT, on_orderly_shutdown, UINT32_TO_PTR(pid)); + sd_event_add_signal(event, NULL, SIGTERM, on_orderly_shutdown, UINT32_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); + + r = pty_forward_new(event, master, &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"); + return r; + } + + forward = pty_forward_free(forward); + if (!arg_quiet) putc('\n', stdout);