X-Git-Url: http://www.chiark.greenend.org.uk/ucgi/~ianmdlvl/git?p=elogind.git;a=blobdiff_plain;f=src%2Fnspawn%2Fnspawn.c;h=abcee3fb98162fae7c7553a0e1e79f1e2d2bf9ee;hp=14774566a64dfe2c5f71946f88855a5c136003ad;hb=7e2270246b0906675c8f34bc278b1608b969e65c;hpb=04d39279245834494baccfdb9349db8bf80abd13 diff --git a/src/nspawn/nspawn.c b/src/nspawn/nspawn.c index 14774566a..abcee3fb9 100644 --- a/src/nspawn/nspawn.c +++ b/src/nspawn/nspawn.c @@ -33,8 +33,6 @@ #include #include #include -#include -#include #include #include #include @@ -42,10 +40,18 @@ #include #include #include +#include +#include +#include -#include -#include +#ifdef HAVE_SELINUX +#include +#endif +#include "sd-daemon.h" +#include "sd-bus.h" +#include "sd-id128.h" +#include "sd-rtnl.h" #include "log.h" #include "util.h" #include "mkdir.h" @@ -56,17 +62,18 @@ #include "strv.h" #include "path-util.h" #include "loopback-setup.h" -#include "sd-id128.h" #include "dev-setup.h" #include "fdset.h" #include "build.h" #include "fileio.h" #include "bus-util.h" +#include "bus-error.h" #include "ptyfwd.h" - -#ifndef TTY_GID -#define TTY_GID 5 -#endif +#include "bus-kernel.h" +#include "env-util.h" +#include "def.h" +#include "rtnl-util.h" +#include "udev-util.h" typedef enum LinkJournal { LINK_NO, @@ -79,6 +86,8 @@ static char *arg_directory = NULL; static char *arg_user = NULL; static sd_id128_t arg_uuid = {}; static char *arg_machine = NULL; +static char *arg_selinux_context = NULL; +static char *arg_selinux_apifs_context = NULL; static const char *arg_slice = NULL; static bool arg_private_network = false; static bool arg_read_only = false; @@ -109,31 +118,53 @@ static uint64_t arg_retain = (1ULL << CAP_SYS_RESOURCE) | (1ULL << CAP_SYS_BOOT) | (1ULL << CAP_AUDIT_WRITE) | - (1ULL << CAP_AUDIT_CONTROL); + (1ULL << CAP_AUDIT_CONTROL) | + (1ULL << CAP_MKNOD); static char **arg_bind = NULL; static char **arg_bind_ro = NULL; +static char **arg_setenv = NULL; +static bool arg_quiet = false; +static bool arg_share_system = false; +static bool arg_register = true; +static bool arg_keep_unit = false; +static char **arg_network_interfaces = NULL; static int help(void) { printf("%s [OPTIONS...] [PATH] [ARGUMENTS...]\n\n" "Spawn a minimal namespace container for debugging, testing and building.\n\n" - " -h --help Show this help\n" - " --version Print version string\n" - " -D --directory=NAME Root directory for the container\n" - " -b --boot Boot up full system (i.e. invoke init)\n" - " -u --user=USER Run the command under specified user or uid\n" - " --uuid=UUID Set a specific machine UUID for the container\n" - " -M --machine=NAME Set the machine name for the container\n" - " -S --slice=SLICE Place the container in the specified slice\n" - " --private-network Disable network in container\n" - " --read-only Mount the root directory read-only\n" - " --capability=CAP In addition to the default, retain specified\n" - " capability\n" - " --link-journal=MODE Link up guest journal, one of no, auto, guest, host\n" - " -j Equivalent to --link-journal=host\n" - " --bind=PATH[:PATH] Bind mount a file or directory from the host into\n" - " the container\n" - " --bind-ro=PATH[:PATH] Similar, but creates a read-only bind mount\n", + " -h --help Show this help\n" + " --version Print version string\n" + " -D --directory=NAME Root directory for the container\n" + " -b --boot Boot up full system (i.e. invoke init)\n" + " -u --user=USER Run the command under specified user or uid\n" + " --uuid=UUID Set a specific machine UUID for the container\n" + " -M --machine=NAME Set the machine name for the container\n" + " -S --slice=SLICE Place the container in the specified slice\n" + " -Z --selinux-context=SECLABEL\n" + " Set the SELinux security context to be used by\n" + " processes in the container\n" + " -L --selinux-apifs-context=SECLABEL\n" + " Set the SELinux security context to be used by\n" + " API/tmpfs file systems in the container\n" + " --private-network Disable network in container\n" + " --network-interface=INTERFACE\n" + " Assign an existing network interface to the container\n" + " --share-system Share system namespaces with host\n" + " --read-only Mount the root directory read-only\n" + " --capability=CAP In addition to the default, retain specified\n" + " capability\n" + " --drop-capability=CAP Drop the specified capability from the default set\n" + " --link-journal=MODE Link up guest journal, one of no, auto, guest, host\n" + " -j Equivalent to --link-journal=host\n" + " --bind=PATH[:PATH] Bind mount a file or directory from the host into\n" + " the container\n" + " --bind-ro=PATH[:PATH] Similar, but creates a read-only bind mount\n" + " --setenv=NAME=VALUE Pass an environment variable to PID 1\n" + " --register=BOOLEAN Register container as machine\n" + " --keep-unit Do not register a scope for the machine, reuse\n" + " the service unit nspawn is running in\n" + " -q --quiet Do not show status information\n", program_invocation_short_name); return 0; @@ -147,41 +178,56 @@ static int parse_argv(int argc, char *argv[]) { ARG_UUID, ARG_READ_ONLY, ARG_CAPABILITY, + ARG_DROP_CAPABILITY, ARG_LINK_JOURNAL, ARG_BIND, - ARG_BIND_RO + ARG_BIND_RO, + ARG_SETENV, + ARG_SHARE_SYSTEM, + ARG_REGISTER, + ARG_KEEP_UNIT, + ARG_NETWORK_INTERFACE }; static const struct option options[] = { - { "help", no_argument, NULL, 'h' }, - { "version", no_argument, NULL, ARG_VERSION }, - { "directory", required_argument, NULL, 'D' }, - { "user", required_argument, NULL, 'u' }, - { "private-network", no_argument, NULL, ARG_PRIVATE_NETWORK }, - { "boot", no_argument, NULL, 'b' }, - { "uuid", required_argument, NULL, ARG_UUID }, - { "read-only", no_argument, NULL, ARG_READ_ONLY }, - { "capability", required_argument, NULL, ARG_CAPABILITY }, - { "link-journal", required_argument, NULL, ARG_LINK_JOURNAL }, - { "bind", required_argument, NULL, ARG_BIND }, - { "bind-ro", required_argument, NULL, ARG_BIND_RO }, - { "machine", required_argument, NULL, 'M' }, - { "slice", required_argument, NULL, 'S' }, - { NULL, 0, NULL, 0 } + { "help", no_argument, NULL, 'h' }, + { "version", no_argument, NULL, ARG_VERSION }, + { "directory", required_argument, NULL, 'D' }, + { "user", required_argument, NULL, 'u' }, + { "private-network", no_argument, NULL, ARG_PRIVATE_NETWORK }, + { "boot", no_argument, NULL, 'b' }, + { "uuid", required_argument, NULL, ARG_UUID }, + { "read-only", no_argument, NULL, ARG_READ_ONLY }, + { "capability", required_argument, NULL, ARG_CAPABILITY }, + { "drop-capability", required_argument, NULL, ARG_DROP_CAPABILITY }, + { "link-journal", required_argument, NULL, ARG_LINK_JOURNAL }, + { "bind", required_argument, NULL, ARG_BIND }, + { "bind-ro", required_argument, NULL, ARG_BIND_RO }, + { "machine", required_argument, NULL, 'M' }, + { "slice", required_argument, NULL, 'S' }, + { "setenv", required_argument, NULL, ARG_SETENV }, + { "selinux-context", required_argument, NULL, 'Z' }, + { "selinux-apifs-context", required_argument, NULL, 'L' }, + { "quiet", no_argument, NULL, 'q' }, + { "share-system", no_argument, NULL, ARG_SHARE_SYSTEM }, + { "register", required_argument, NULL, ARG_REGISTER }, + { "keep-unit", no_argument, NULL, ARG_KEEP_UNIT }, + { "network-interface", required_argument, NULL, ARG_NETWORK_INTERFACE }, + {} }; int c, r; + uint64_t plus = 0, minus = 0; assert(argc >= 0); assert(argv); - while ((c = getopt_long(argc, argv, "+hD:u:bM:jS:", options, NULL)) >= 0) { + while ((c = getopt_long(argc, argv, "+hD:u:bL:M:jS:Z:q", options, NULL)) >= 0) { switch (c) { case 'h': - help(); - return 0; + return help(); case ARG_VERSION: puts(PACKAGE_STRING); @@ -192,7 +238,7 @@ static int parse_argv(int argc, char *argv[]) { free(arg_directory); arg_directory = canonicalize_file_name(optarg); if (!arg_directory) { - log_error("Failed to canonicalize root directory."); + log_error("Invalid root directory: %m"); return -ENOMEM; } @@ -206,6 +252,12 @@ static int parse_argv(int argc, char *argv[]) { break; + case ARG_NETWORK_INTERFACE: + if (strv_push(&arg_network_interfaces, optarg) < 0) + return log_oom(); + + /* fall through */ + case ARG_PRIVATE_NETWORK: arg_private_network = true; break; @@ -224,45 +276,71 @@ static int parse_argv(int argc, char *argv[]) { case 'S': arg_slice = strdup(optarg); + if (!arg_slice) + return log_oom(); + break; case 'M': - if (!hostname_is_valid(optarg)) { - log_error("Invalid machine name: %s", optarg); - return -EINVAL; + if (isempty(optarg)) { + free(arg_machine); + arg_machine = NULL; + } else { + + if (!hostname_is_valid(optarg)) { + log_error("Invalid machine name: %s", optarg); + return -EINVAL; + } + + free(arg_machine); + arg_machine = strdup(optarg); + if (!arg_machine) + return log_oom(); + + break; } - free(arg_machine); - arg_machine = strdup(optarg); - if (!arg_machine) - return log_oom(); + case 'Z': + arg_selinux_context = optarg; + break; + case 'L': + arg_selinux_apifs_context = optarg; break; case ARG_READ_ONLY: arg_read_only = true; break; - case ARG_CAPABILITY: { + case ARG_CAPABILITY: + case ARG_DROP_CAPABILITY: { char *state, *word; size_t length; FOREACH_WORD_SEPARATOR(word, length, optarg, ",", state) { + _cleanup_free_ char *t; cap_value_t cap; - char *t; t = strndup(word, length); if (!t) return log_oom(); - if (cap_from_name(t, &cap) < 0) { - log_error("Failed to parse capability %s.", t); - free(t); - return -EINVAL; + if (streq(t, "all")) { + if (c == ARG_CAPABILITY) + plus = (uint64_t) -1; + else + minus = (uint64_t) -1; + } else { + if (cap_from_name(t, &cap) < 0) { + log_error("Failed to parse capability %s.", t); + return -EINVAL; + } + + if (c == ARG_CAPABILITY) + plus |= 1ULL << (uint64_t) cap; + else + minus |= 1ULL << (uint64_t) cap; } - - free(t); - arg_retain |= 1ULL << (uint64_t) cap; } break; @@ -315,24 +393,77 @@ static int parse_argv(int argc, char *argv[]) { r = strv_extend(x, a); if (r < 0) - return r; + return log_oom(); r = strv_extend(x, b); if (r < 0) - return r; + return log_oom(); + + break; + } + + case ARG_SETENV: { + char **n; + + if (!env_assignment_is_valid(optarg)) { + log_error("Environment variable assignment '%s' is not valid.", optarg); + return -EINVAL; + } + + n = strv_env_set(arg_setenv, optarg); + if (!n) + return log_oom(); + strv_free(arg_setenv); + arg_setenv = n; break; } + case 'q': + arg_quiet = true; + break; + + case ARG_SHARE_SYSTEM: + arg_share_system = true; + break; + + case ARG_REGISTER: + r = parse_boolean(optarg); + if (r < 0) { + log_error("Failed to parse --register= argument: %s", optarg); + return r; + } + + arg_register = r; + break; + + case ARG_KEEP_UNIT: + arg_keep_unit = true; + break; + case '?': return -EINVAL; default: - log_error("Unknown option code %c", c); - return -EINVAL; + assert_not_reached("Unhandled option"); } } + if (arg_share_system) + arg_register = false; + + if (arg_boot && arg_share_system) { + log_error("--boot and --share-system may not be combined."); + return -EINVAL; + } + + if (arg_keep_unit && cg_pid_get_owner_uid(0, NULL) >= 0) { + log_error("--keep-unit may not be used when invoked from a user session."); + return -EINVAL; + } + + arg_retain = (arg_retain | plus | (arg_private_network ? 1ULL << CAP_NET_ADMIN : 0)) & ~minus; + return 1; } @@ -367,6 +498,10 @@ static int mount_all(const char *dest) { for (k = 0; k < ELEMENTSOF(mount_table); k++) { _cleanup_free_ char *where = NULL; +#ifdef HAVE_SELINUX + _cleanup_free_ char *options = NULL; +#endif + const char *o; int t; where = strjoin(dest, "/", mount_table[k].where, NULL); @@ -389,11 +524,24 @@ static int mount_all(const char *dest) { mkdir_p(where, 0755); +#ifdef HAVE_SELINUX + if (arg_selinux_apifs_context && + (streq_ptr(mount_table[k].what, "tmpfs") || streq_ptr(mount_table[k].what, "devpts"))) { + options = strjoin(mount_table[k].options, ",context=\"", arg_selinux_apifs_context, "\"", NULL); + if (!options) + return log_oom(); + + o = options; + } else +#endif + o = mount_table[k].options; + + if (mount(mount_table[k].what, where, mount_table[k].type, mount_table[k].flags, - mount_table[k].options) < 0 && + o) < 0 && mount_table[k].fatal) { log_error("mount(%s) failed: %m", where); @@ -410,39 +558,46 @@ static int mount_binds(const char *dest, char **l, unsigned long flags) { char **x, **y; STRV_FOREACH_PAIR(x, y, l) { - _cleanup_free_ char *where = NULL; + char *where; struct stat source_st, dest_st; + int r; if (stat(*x, &source_st) < 0) { log_error("failed to stat %s: %m", *x); return -errno; } - where = strjoin(dest, "/", *y, NULL); - if (!where) - return log_oom(); - - if (stat(where, &dest_st) == 0) { + where = strappenda(dest, *y); + r = stat(where, &dest_st); + if (r == 0) { if ((source_st.st_mode & S_IFMT) != (dest_st.st_mode & S_IFMT)) { log_error("The file types of %s and %s do not match. Refusing bind mount", *x, where); return -EINVAL; } - } else { - /* Create the mount point, but be conservative -- refuse to create block - * and char devices. */ - if (S_ISDIR(source_st.st_mode)) - mkdir_p_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 { - log_error("Refusing to create mountpoint for file: %s", *x); - return -ENOTSUP; + } else if (errno == ENOENT) { + r = mkdir_parents_label(where, 0755); + if (r < 0) { + log_error("Failed to bind mount %s: %s", *x, strerror(-r)); + return r; } + } else { + log_error("Failed to bind mount %s: %s", *x, strerror(errno)); + return -errno; + } + /* 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 { + log_error("Refusing to create mountpoint for file: %s", *x); + return -ENOTSUP; } if (mount(*x, where, "bind", MS_BIND, NULL) < 0) { @@ -547,6 +702,9 @@ static int setup_boot_id(const char *dest) { assert(dest); + if (arg_share_system) + return 0; + /* Generate a new randomized boot ID, so that each boot-up of * the container gets a new one */ @@ -601,40 +759,30 @@ static int copy_devnodes(const char *dest) { u = umask(0000); NULSTR_FOREACH(d, devnodes) { - struct stat st; _cleanup_free_ char *from = NULL, *to = NULL; + struct stat st; - asprintf(&from, "/dev/%s", d); - asprintf(&to, "%s/dev/%s", dest, d); - - if (!from || !to) { - log_oom(); - - if (r == 0) - r = -ENOMEM; - - break; - } + from = strappend("/dev/", d); + to = strjoin(dest, "/dev/", d, NULL); + if (!from || !to) + return log_oom(); if (stat(from, &st) < 0) { if (errno != ENOENT) { log_error("Failed to stat %s: %m", from); - if (r == 0) - r = -errno; + return -errno; } } else if (!S_ISCHR(st.st_mode) && !S_ISBLK(st.st_mode)) { log_error("%s is not a char or block device, cannot copy", from); - if (r == 0) - r = -EIO; + return -EIO; } else if (mknod(to, st.st_mode, st.st_rdev) < 0) { log_error("mknod(%s) failed: %m", dest); - if (r == 0) - r = -errno; + return -errno; } } @@ -782,6 +930,9 @@ static int setup_kmsg(const char *dest, int kmsg_socket) { static int setup_hostname(void) { + if (arg_share_system) + return 0; + if (sethostname(arg_machine, strlen(arg_machine)) < 0) return -errno; @@ -789,14 +940,11 @@ static int setup_hostname(void) { } static int setup_journal(const char *directory) { - sd_id128_t machine_id; + sd_id128_t machine_id, this_id; _cleanup_free_ char *p = NULL, *b = NULL, *q = NULL, *d = NULL; char *id; int r; - if (arg_link_journal == LINK_NO) - return 0; - p = strappend(directory, "/etc/machine-id"); if (!p) return log_oom(); @@ -820,6 +968,24 @@ static int setup_journal(const char *directory) { return r; } + r = sd_id128_get_machine(&this_id); + if (r < 0) { + log_error("Failed to retrieve machine ID: %s", strerror(-r)); + return r; + } + + if (sd_id128_equal(machine_id, this_id)) { + log_full(arg_link_journal == LINK_AUTO ? LOG_WARNING : LOG_ERR, + "Host and machine ids are equal (%s): refusing to link journals", id); + if (arg_link_journal == LINK_AUTO) + return 0; + return + -EEXIST; + } + + if (arg_link_journal == LINK_NO) + return 0; + free(p); p = strappend("/var/log/journal/", id); q = strjoin(directory, "/var/log/journal/", id, NULL); @@ -920,16 +1086,98 @@ 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); } -static int register_machine(void) { +static int register_machine(pid_t pid) { + _cleanup_bus_error_free_ sd_bus_error error = SD_BUS_ERROR_NULL; + _cleanup_bus_unref_ sd_bus *bus = NULL; + int r; + + if (!arg_register) + return 0; + + r = sd_bus_default_system(&bus); + if (r < 0) { + log_error("Failed to open system bus: %s", strerror(-r)); + return r; + } + + if (arg_keep_unit) { + r = sd_bus_call_method( + bus, + "org.freedesktop.machine1", + "/org/freedesktop/machine1", + "org.freedesktop.machine1.Manager", + "RegisterMachine", + &error, + NULL, + "sayssus", + arg_machine, + SD_BUS_MESSAGE_APPEND_ID128(arg_uuid), + "nspawn", + "container", + (uint32_t) pid, + strempty(arg_directory)); + } else { + r = sd_bus_call_method( + bus, + "org.freedesktop.machine1", + "/org/freedesktop/machine1", + "org.freedesktop.machine1.Manager", + "CreateMachine", + &error, + NULL, + "sayssusa(sv)", + arg_machine, + SD_BUS_MESSAGE_APPEND_ID128(arg_uuid), + "nspawn", + "container", + (uint32_t) pid, + strempty(arg_directory), + !isempty(arg_slice), "Slice", "s", arg_slice); + } + + if (r < 0) { + log_error("Failed to register machine: %s", bus_error_message(&error, r)); + return r; + } + + return 0; +} + +static int terminate_machine(pid_t pid) { _cleanup_bus_error_free_ sd_bus_error error = SD_BUS_ERROR_NULL; + _cleanup_bus_message_unref_ sd_bus_message *reply = NULL; _cleanup_bus_unref_ sd_bus *bus = NULL; + const char *path; int r; - r = sd_bus_open_system(&bus); + if (!arg_register) + return 0; + + r = sd_bus_default_system(&bus); if (r < 0) { log_error("Failed to open system bus: %s", strerror(-r)); return r; @@ -940,45 +1188,153 @@ static int register_machine(void) { "org.freedesktop.machine1", "/org/freedesktop/machine1", "org.freedesktop.machine1.Manager", - "CreateMachine", + "GetMachineByPID", + &error, + &reply, + "u", + (uint32_t) pid); + if (r < 0) { + /* Note that the machine might already have been + * cleaned up automatically, hence don't consider it a + * failure if we cannot get the machine object. */ + log_debug("Failed to get machine: %s", bus_error_message(&error, r)); + return 0; + } + + r = sd_bus_message_read(reply, "o", &path); + if (r < 0) + return bus_log_parse_error(r); + + r = sd_bus_call_method( + bus, + "org.freedesktop.machine1", + path, + "org.freedesktop.machine1.Machine", + "Terminate", &error, NULL, - "sayssusa(sv)", - arg_machine, - SD_BUS_MESSAGE_APPEND_ID128(arg_uuid), - "nspawn", - "container", - (uint32_t) 0, - strempty(arg_directory), - !isempty(arg_slice), "Slice", "s", arg_slice); + NULL); + if (r < 0) { + log_debug("Failed to terminate machine: %s", bus_error_message(&error, r)); + return 0; + } + + return 0; +} + +static int reset_audit_loginuid(void) { + _cleanup_free_ char *p = NULL; + int r; + + if (arg_share_system) + return 0; + + r = read_one_line_file("/proc/self/loginuid", &p); + if (r == -EEXIST) + return 0; if (r < 0) { - log_error("Failed to register machine: %s", error.message ? error.message : strerror(-r)); + log_error("Failed to read /proc/self/loginuid: %s", strerror(-r)); return r; } + /* Already reset? */ + if (streq(p, "4294967295")) + return 0; + + r = write_string_file("/proc/self/loginuid", "4294967295"); + if (r < 0) { + log_error("Failed to reset audit login UID. This probably means that your kernel is too\n" + "old and you have audit enabled. Note that the auditing subsystem is known to\n" + "be incompatible with containers on old kernels. Please make sure to upgrade\n" + "your kernel or to off auditing with 'audit=0' on the kernel command line before\n" + "using systemd-nspawn. Sleeping for 5s... (%s)\n", strerror(-r)); + + sleep(5); + } + return 0; } -static bool audit_enabled(void) { - int fd; +static int move_network_interfaces(pid_t pid) { + _cleanup_rtnl_unref_ sd_rtnl *rtnl = NULL; + _cleanup_udev_unref_ struct udev *udev = NULL; + char **i; + int r; - fd = socket(AF_NETLINK, SOCK_RAW, NETLINK_AUDIT); - if (fd >= 0) { - close_nointr_nofail(fd); - return true; + if (!arg_private_network) + return 0; + + if (strv_isempty(arg_network_interfaces)) + return 0; + + r = sd_rtnl_open(0, &rtnl); + if (r < 0) { + log_error("Failed to connect to netlink: %s", strerror(-r)); + return r; + } + + udev = udev_new(); + if (!udev) { + log_error("Failed to connect to udev."); + return -ENOMEM; } - return false; + + STRV_FOREACH(i, arg_network_interfaces) { + _cleanup_rtnl_message_unref_ sd_rtnl_message *m = NULL; + _cleanup_udev_device_unref_ struct udev_device *d = NULL; + char ifi_str[2 + DECIMAL_STR_MAX(int)]; + int ifi; + + ifi = (int) if_nametoindex(*i); + if (ifi <= 0) { + log_error("Failed to resolve interface %s: %m", *i); + return -errno; + } + + sprintf(ifi_str, "n%i", ifi); + d = udev_device_new_from_device_id(udev, ifi_str); + if (!d) { + log_error("Failed to get udev device for interface %s: %m", *i); + return -errno; + } + + if (udev_device_get_is_initialized(d) <= 0) { + log_error("Network interface %s is not initialized yet.", *i); + return -EBUSY; + } + + r = sd_rtnl_message_new_link(RTM_NEWLINK, ifi, &m); + if (r < 0) { + log_error("Failed to allocate netlink message: %s", strerror(-r)); + return r; + } + + r = sd_rtnl_message_append_u32(m, IFLA_NET_NS_PID, pid); + if (r < 0) { + log_error("Failed to append namespace PID to netlink message: %s", strerror(-r)); + return r; + } + + r = sd_rtnl_call(rtnl, m, 0, NULL); + if (r < 0) { + log_error("Failed to move interface %s to namespace: %s", *i, strerror(-r)); + return r; + } + } + + return 0; } int main(int argc, char *argv[]) { pid_t pid = 0; int r = EXIT_FAILURE, k; - _cleanup_close_ int master = -1; + _cleanup_close_ int master = -1, kdbus_fd = -1, sync_fd = -1; int n_fd_passed; const char *console = NULL; sigset_t mask; _cleanup_close_pipe_ int kmsg_socket_pair[2] = { -1, -1 }; _cleanup_fdset_free_ FDSet *fds = NULL; + _cleanup_free_ char *kdbus_domain = NULL; log_parse_environment(); log_open(); @@ -1008,7 +1364,7 @@ int main(int argc, char *argv[]) { path_kill_slashes(arg_directory); if (!arg_machine) { - arg_machine = strdup(path_get_file_name(arg_directory)); + arg_machine = strdup(basename(arg_directory)); if (!arg_machine) { log_oom(); goto finish; @@ -1031,19 +1387,12 @@ int main(int argc, char *argv[]) { goto finish; } - if (arg_boot && audit_enabled()) { - log_warning("The kernel auditing subsystem is known to be incompatible with containers.\n" - "Please make sure to turn off auditing with 'audit=0' on the kernel command\n" - "line before using systemd-nspawn. Sleeping for 5s...\n"); - sleep(5); - } - if (path_equal(arg_directory, "/")) { log_error("Spawning container on root directory not supported."); goto finish; } - if (path_is_os_tree(arg_directory) <= 0) { + if (arg_boot && path_is_os_tree(arg_directory) <= 0) { log_error("Directory %s doesn't look like an OS root directory (/etc/os-release is missing). Refusing.", arg_directory); goto finish; } @@ -1072,15 +1421,37 @@ int main(int argc, char *argv[]) { goto finish; } - log_info("Spawning container %s on %s. Press ^] three times within 1s to abort execution.", arg_machine, arg_directory); + if (!arg_quiet) + log_info("Spawning container %s on %s. Press ^] three times within 1s to abort execution.", arg_machine, arg_directory); if (unlockpt(master) < 0) { log_error("Failed to unlock tty: %m"); goto finish; } + + 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."); + log_error("Failed to create kmsg socket pair: %m"); goto finish; } @@ -1093,7 +1464,16 @@ int main(int argc, char *argv[]) { for (;;) { siginfo_t status; - pid = syscall(__NR_clone, SIGCHLD|CLONE_NEWIPC|CLONE_NEWNS|CLONE_NEWPID|CLONE_NEWUTS|(arg_private_network ? CLONE_NEWNET : 0), NULL); + sync_fd = eventfd(0, EFD_CLOEXEC); + if (sync_fd < 0) { + log_error("Failed to create event fd: %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); 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"); @@ -1110,7 +1490,7 @@ int main(int argc, char *argv[]) { gid_t gid = (gid_t) -1; unsigned n_env = 2; const char *envp[] = { - "PATH=/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin", + "PATH=" DEFAULT_PATH_SPLIT_USR, "container=systemd-nspawn", /* LXC sets container=lxc, so follow the scheme here */ NULL, /* TERM */ NULL, /* HOME */ @@ -1121,6 +1501,8 @@ int main(int argc, char *argv[]) { NULL, /* LISTEN_PID */ NULL }; + char **env_use; + eventfd_t x; envp[n_env] = strv_find_prefix(environ, "TERM="); if (envp[n_env]) @@ -1163,15 +1545,14 @@ int main(int argc, char *argv[]) { goto child_fail; } + if (reset_audit_loginuid() < 0) + goto child_fail; + if (prctl(PR_SET_PDEATHSIG, SIGKILL) < 0) { log_error("PR_SET_PDEATHSIG failed: %m"); goto child_fail; } - r = register_machine(); - if (r < 0) - goto finish; - /* Mark everything as slave, so that we still * receive mounts from the real root, but don't * propagate mounts to the real root. */ @@ -1230,6 +1611,9 @@ int main(int argc, char *argv[]) { if (mount_binds(arg_directory, arg_bind_ro, MS_RDONLY) < 0) goto child_fail; + if (setup_kdbus(arg_directory, kdbus_domain) < 0) + goto child_fail; + if (chdir(arg_directory) < 0) { log_error("chdir(%s) failed: %m", arg_directory); goto child_fail; @@ -1252,7 +1636,8 @@ int main(int argc, char *argv[]) { umask(0022); - loopback_setup(); + if (arg_private_network) + loopback_setup(); if (drop_capabilities() < 0) { log_error("drop_capabilities() failed: %m"); @@ -1345,6 +1730,28 @@ int main(int argc, char *argv[]) { setup_hostname(); + eventfd_read(sync_fd, &x); + close_nointr_nofail(sync_fd); + sync_fd = -1; + + if (!strv_isempty(arg_setenv)) { + char **n; + + n = strv_env_merge(2, envp, arg_setenv); + if (!n) { + log_oom(); + goto child_fail; + } + + env_use = n; + } else + env_use = (char**) envp; + +#ifdef HAVE_SELINUX + if (arg_selinux_context) + if (setexeccon(arg_selinux_context) < 0) + log_error("setexeccon(\"%s\") failed: %m", arg_selinux_context); +#endif if (arg_boot) { char **a; size_t l; @@ -1356,18 +1763,18 @@ int main(int argc, char *argv[]) { memcpy(a + 1, argv + optind, l * sizeof(char*)); a[0] = (char*) "/usr/lib/systemd/systemd"; - execve(a[0], a, (char**) envp); + execve(a[0], a, env_use); a[0] = (char*) "/lib/systemd/systemd"; - execve(a[0], a, (char**) envp); + execve(a[0], a, env_use); a[0] = (char*) "/sbin/init"; - execve(a[0], a, (char**) envp); + execve(a[0], a, env_use); } else if (argc > optind) - execvpe(argv[optind], argv + optind, (char**) envp); + execvpe(argv[optind], argv + optind, env_use); else { chdir(home ? home : "/root"); - execle("/bin/bash", "-bash", NULL, (char**) envp); + execle("/bin/bash", "-bash", NULL, env_use); } log_error("execv() failed: %m"); @@ -1379,15 +1786,31 @@ int main(int argc, char *argv[]) { fdset_free(fds); fds = NULL; + r = register_machine(pid); + if (r < 0) + goto finish; + + r = move_network_interfaces(pid); + if (r < 0) + goto finish; + + eventfd_write(sync_fd, 1); + close_nointr_nofail(sync_fd); + sync_fd = -1; + k = process_pty(master, &mask, arg_boot ? pid : 0, SIGRTMIN+3); if (k < 0) { r = EXIT_FAILURE; break; } - putc('\n', stdout); + if (!arg_quiet) + putc('\n', stdout); /* Kill if it is not dead yet anyway */ + terminate_machine(pid); + + /* Redundant, but better safe than sorry */ kill(pid, SIGKILL); k = wait_for_terminate(pid, &status); @@ -1405,21 +1828,26 @@ int main(int argc, char *argv[]) { break; } - log_debug("Container %s exited successfully.", arg_machine); + if (!arg_quiet) + log_debug("Container %s exited successfully.", arg_machine); break; } else if (status.si_code == CLD_KILLED && status.si_status == SIGINT) { - log_info("Container %s has been shut down.", arg_machine); + + if (!arg_quiet) + log_info("Container %s has been shut down.", arg_machine); r = 0; break; } else if (status.si_code == CLD_KILLED && status.si_status == SIGHUP) { - log_info("Container %s is being rebooted.", arg_machine); + + if (!arg_quiet) + log_info("Container %s is being rebooted.", arg_machine); continue; } else if (status.si_code == CLD_KILLED || status.si_code == CLD_DUMPED) { - log_error("Container %s terminated by signal %s.", arg_machine, signal_to_string(status.si_status)); + log_error("Container %s terminated by signal %s.", arg_machine, signal_to_string(status.si_status)); r = EXIT_FAILURE; break; } else { @@ -1435,6 +1863,8 @@ finish: free(arg_directory); free(arg_machine); + free(arg_setenv); + free(arg_network_interfaces); return r; }