#include <sys/un.h>
#include <sys/socket.h>
#include <linux/netlink.h>
+#include <sys/eventfd.h>
+#if HAVE_SELINUX
+#include <selinux/selinux.h>
+#endif
#include "sd-daemon.h"
#include "sd-bus.h"
#include "ptyfwd.h"
#include "bus-kernel.h"
#include "env-util.h"
-
-#ifndef TTY_GID
-#define TTY_GID 5
-#endif
+#include "def.h"
typedef enum LinkJournal {
LINK_NO,
static char *arg_user = NULL;
static sd_id128_t arg_uuid = {};
static char *arg_machine = NULL;
+static char *arg_process_label = NULL;
+static char *arg_file_label = NULL;
static const char *arg_slice = NULL;
static bool arg_private_network = false;
static bool arg_read_only = false;
(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;
printf("%s [OPTIONS...] [PATH] [ARGUMENTS...]\n\n"
"Spawn a minimal namespace container for debugging, testing and building.\n\n"
- " -h --help Show this help\n"
- " --version Print version string\n"
- " -D --directory=NAME Root directory for the container\n"
- " -b --boot Boot up full system (i.e. invoke init)\n"
- " -u --user=USER Run the command under specified user or uid\n"
- " --uuid=UUID Set a specific machine UUID for the container\n"
- " -M --machine=NAME Set the machine name for the container\n"
- " -S --slice=SLICE Place the container in the specified slice\n"
- " --private-network Disable network in container\n"
- " --read-only Mount the root directory read-only\n"
- " --capability=CAP In addition to the default, retain specified\n"
- " capability\n"
- " --drop-capability=CAP Drop the specified capability from the default set\n"
- " --link-journal=MODE Link up guest journal, one of no, auto, guest, host\n"
- " -j Equivalent to --link-journal=host\n"
- " --bind=PATH[:PATH] Bind mount a file or directory from the host into\n"
- " the container\n"
- " --bind-ro=PATH[:PATH] Similar, but creates a read-only bind mount\n"
- " --setenv=NAME=VALUE Pass an environment variable to PID 1\n",
+ " -h --help Show this help\n"
+ " --version Print version string\n"
+ " -D --directory=NAME Root directory for the container\n"
+ " -b --boot Boot up full system (i.e. invoke init)\n"
+ " -u --user=USER Run the command under specified user or uid\n"
+ " --uuid=UUID Set a specific machine UUID for the container\n"
+ " -M --machine=NAME Set the machine name for the container\n"
+ " -S --slice=SLICE Place the container in the specified slice\n"
+ " -L --file-label=LABEL Set the MAC file label to be used by tmpfs file\n"
+ " systems in the container\n"
+ " -Z --process-label=LABEL Set the MAC label to be used by processes in\n"
+ " the container\n"
+ " --private-network Disable network in container\n"
+ " --read-only Mount the root directory read-only\n"
+ " --capability=CAP In addition to the default, retain specified\n"
+ " capability\n"
+ " --drop-capability=CAP Drop the specified capability from the default set\n"
+ " --link-journal=MODE Link up guest journal, one of no, auto, guest, host\n"
+ " -j Equivalent to --link-journal=host\n"
+ " --bind=PATH[:PATH] Bind mount a file or directory from the host into\n"
+ " the container\n"
+ " --bind-ro=PATH[:PATH] Similar, but creates a read-only bind mount\n"
+ " --setenv=NAME=VALUE Pass an environment variable to PID 1\n",
program_invocation_short_name);
return 0;
{ "machine", required_argument, NULL, 'M' },
{ "slice", required_argument, NULL, 'S' },
{ "setenv", required_argument, NULL, ARG_SETENV },
+ { "process-label", required_argument, NULL, 'Z' },
+ { "file-label", required_argument, NULL, 'L' },
{}
};
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:", options, NULL)) >= 0) {
switch (c) {
break;
+ case 'L':
+ arg_file_label = optarg;
+ break;
+
+ case 'Z':
+ arg_process_label = optarg;
+ break;
+
case ARG_READ_ONLY:
arg_read_only = true;
break;
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);
mkdir_p(where, 0755);
+#ifdef HAVE_SELINUX
+ if (arg_file_label && (streq_ptr(mount_table[k].what, "tmpfs") || streq_ptr(mount_table[k].what, "devpts"))) {
+ options = strjoin(mount_table[k].options, ",context=\"", arg_file_label, "\"", 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);
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;
}
}
}
if (mount(path, p, "bind", MS_BIND, NULL) < 0) {
- log_error("Failed to mount kdbus namespace path: %m");
+ log_error("Failed to mount kdbus domain path: %m");
return -errno;
}
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;
SD_BUS_MESSAGE_APPEND_ID128(arg_uuid),
"nspawn",
"container",
- (uint32_t) 0,
+ (uint32_t) pid,
strempty(arg_directory),
!isempty(arg_slice), "Slice", "s", arg_slice);
if (r < 0) {
int main(int argc, char *argv[]) {
pid_t pid = 0;
int r = EXIT_FAILURE, k;
- _cleanup_close_ int master = -1, kdbus_fd = -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_namespace = NULL;
+ _cleanup_free_ char *kdbus_domain = NULL;
const char *ns;
log_parse_environment();
}
ns = strappenda("machine-", arg_machine);
- kdbus_fd = bus_kernel_create_namespace(ns, &kdbus_namespace);
+ kdbus_fd = bus_kernel_create_domain(ns, &kdbus_domain);
if (r < 0)
- log_debug("Failed to create kdbus namespace: %s", strerror(-r));
+ log_debug("Failed to create kdbus domain: %s", strerror(-r));
else
- log_debug("Successfully created kdbus namespace as %s", kdbus_namespace);
+ 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;
}
for (;;) {
siginfo_t status;
+ 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_NEWIPC|CLONE_NEWNS|CLONE_NEWPID|CLONE_NEWUTS|(arg_private_network ? CLONE_NEWNET : 0), NULL);
if (pid < 0) {
if (errno == EINVAL)
NULL
};
char **env_use;
+ eventfd_t x;
envp[n_env] = strv_find_prefix(environ, "TERM=");
if (envp[n_env])
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. */
if (mount_binds(arg_directory, arg_bind_ro, MS_RDONLY) < 0)
goto child_fail;
- if (setup_kdbus(arg_directory, kdbus_namespace) < 0)
+ if (setup_kdbus(arg_directory, kdbus_domain) < 0)
goto child_fail;
if (chdir(arg_directory) < 0) {
setup_hostname();
+ eventfd_read(sync_fd, &x);
+ close_nointr_nofail(sync_fd);
+ sync_fd = -1;
+
if (!strv_isempty(arg_setenv)) {
char **n;
} else
env_use = (char**) envp;
+#if HAVE_SELINUX
+ if (arg_process_label)
+ if (setexeccon(arg_process_label) < 0)
+ log_error("setexeccon(\"%s\") failed: %m", arg_process_label);
+#endif
if (arg_boot) {
char **a;
size_t l;
fdset_free(fds);
fds = NULL;
+ r = register_machine(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;