#include <termios.h>
#include <sys/signalfd.h>
#include <grp.h>
+#include <linux/fs.h>
#include "log.h"
#include "util.h"
#include "cgroup-util.h"
#include "sd-daemon.h"
#include "strv.h"
+#include "loopback-setup.h"
static char *arg_directory = NULL;
static char *arg_user = NULL;
+static bool arg_private_network = false;
static int help(void) {
"Spawn a minimal namespace container for debugging, testing and building.\n\n"
" -h --help Show this help\n"
" -D --directory=NAME Root directory for the container\n"
- " -u --user=USER Run the command under specified user or uid\n",
+ " -u --user=USER Run the command under specified user or uid\n"
+ " --private-network Disable network in container\n",
program_invocation_short_name);
return 0;
static int parse_argv(int argc, char *argv[]) {
+ enum {
+ ARG_PRIVATE_NETWORK = 0x100
+ };
+
static const struct option options[] = {
- { "help", no_argument, NULL, 'h' },
- { "directory", required_argument, NULL, 'D' },
- { "user", optional_argument, NULL, 'u' },
- { NULL, 0, NULL, 0 }
+ { "help", no_argument, NULL, 'h' },
+ { "directory", required_argument, NULL, 'D' },
+ { "user", required_argument, NULL, 'u' },
+ { "private-network", no_argument, NULL, ARG_PRIVATE_NETWORK },
+ { NULL, 0, NULL, 0 }
};
int c;
break;
+ case ARG_PRIVATE_NETWORK:
+ arg_private_network = true;
+ break;
+
case '?':
return -EINVAL;
} MountPoint;
static const MountPoint mount_table[] = {
- { "proc", "/proc", "proc", NULL, MS_NOSUID|MS_NOEXEC|MS_NODEV, true },
- { "/proc/sys", "/proc/sys", "bind", NULL, MS_BIND, true }, /* Bind mount first */
- { "/proc/sys", "/proc/sys", "bind", NULL, MS_BIND|MS_RDONLY|MS_REMOUNT, true }, /* Then, make it r/o */
- { "sysfs", "/sys", "sysfs", NULL, MS_NOSUID|MS_NOEXEC|MS_NODEV|MS_RDONLY, true },
- { "tmpfs", "/dev", "tmpfs", "mode=755", MS_NOSUID, true },
- { "/dev/pts", "/dev/pts", "bind", NULL, MS_BIND, true },
- { "tmpfs", "/run", "tmpfs", "mode=755", MS_NOSUID|MS_NODEV, true },
+ { "proc", "/proc", "proc", NULL, MS_NOSUID|MS_NOEXEC|MS_NODEV, true },
+ { "/proc/sys", "/proc/sys", "bind", NULL, MS_BIND, true }, /* Bind mount first */
+ { "/proc/sys", "/proc/sys", "bind", NULL, MS_BIND|MS_RDONLY|MS_REMOUNT, true }, /* Then, make it r/o */
+ { "/sys", "/sys", "bind", NULL, MS_BIND, true }, /* Bind mount first */
+ { "/sys", "/sys", "bind", NULL, MS_BIND|MS_RDONLY|MS_REMOUNT, true }, /* Then, make it r/o */
+ { "tmpfs", "/dev", "tmpfs", "mode=755", MS_NOSUID, true },
+ { "/dev/pts", "/dev/pts", "bind", NULL, MS_BIND, true },
+ { "tmpfs", "/run", "tmpfs", "mode=755", MS_NOSUID|MS_NODEV, true },
#ifdef HAVE_SELINUX
- { "selinux", "/selinux", "selinuxfs", NULL, MS_NOSUID|MS_NOEXEC|MS_NODEV|MS_RDONLY, false },
+ { "/sys/fs/selinux", "/sys/fs/selinux", "bind", NULL, MS_BIND, false }, /* Bind mount first */
+ { "/sys/fs/selinux", "/sys/fs/selinux", "bind", NULL, MS_BIND|MS_RDONLY|MS_REMOUNT, false }, /* Then, make it r/o */
#endif
};
break;
}
- if ((t = path_is_mount_point(where)) < 0) {
+ if ((t = path_is_mount_point(where, false)) < 0) {
log_error("Failed to detect whether %s is a mount point: %s", where, strerror(-t));
free(where);
/* Fix the timezone, if possible */
if (asprintf(&where, "%s/%s", dest, "/etc/localtime") >= 0) {
- mount("/etc/localtime", where, "bind", MS_BIND, NULL);
- mount("/etc/localtime", where, "bind", MS_BIND|MS_REMOUNT|MS_RDONLY, NULL);
+
+ if (mount("/etc/localtime", where, "bind", MS_BIND, NULL) >= 0)
+ mount("/etc/localtime", where, "bind", MS_BIND|MS_REMOUNT|MS_RDONLY, NULL);
+
free(where);
}
}
finish:
-
umask(u);
return r;
unsigned long l;
- for (l = 0; l <= MAX(63LU, (unsigned long) CAP_LAST_CAP); l++) {
+ for (l = 0; l <= cap_last_cap(); l++) {
unsigned i;
for (i = 0; i < ELEMENTSOF(retain); i++)
continue;
if (prctl(PR_CAPBSET_DROP, l) < 0) {
-
- /* If this capability is not known, EINVAL
- * will be returned, let's ignore this. */
- if (errno == EINVAL)
- break;
-
log_error("PR_CAPBSET_DROP failed: %m");
return -errno;
}
sigset_add_many(&mask, SIGCHLD, SIGWINCH, SIGTERM, SIGINT, -1);
assert_se(sigprocmask(SIG_BLOCK, &mask, NULL) == 0);
- if ((pid = syscall(__NR_clone, SIGCHLD|CLONE_NEWIPC|CLONE_NEWNS|CLONE_NEWPID|CLONE_NEWUTS, NULL)) < 0) {
+ if ((pid = syscall(__NR_clone, SIGCHLD|CLONE_NEWIPC|CLONE_NEWNS|CLONE_NEWPID|CLONE_NEWUTS|(arg_private_network ? CLONE_NEWNET : 0), NULL)) < 0) {
log_error("clone() failed: %m");
goto finish;
}
gid_t gid = (gid_t) -1;
const char *envp[] = {
"PATH=/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin",
+ "container=systemd-nspawn", /* LXC sets container=lxc, so follow the scheme here */
NULL, /* TERM */
NULL, /* HOME */
NULL, /* USER */
NULL
};
- envp[1] = strv_find_prefix(environ, "TERM=");
+ envp[2] = strv_find_prefix(environ, "TERM=");
close_nointr_nofail(master);
if (prctl(PR_SET_PDEATHSIG, SIGKILL) < 0)
goto child_fail;
+ /* Mark / as private, in case somebody marked it shared */
+ if (mount(NULL, "/", NULL, MS_PRIVATE|MS_REC, NULL) < 0)
+ goto child_fail;
+
if (mount_all(arg_directory) < 0)
goto child_fail;
goto child_fail;
}
- umask(0002);
+ umask(0022);
+
+ loopback_setup();
if (drop_capabilities() < 0)
goto child_fail;
}
}
- if ((asprintf((char**)(envp + 2), "HOME=%s", home? home: "/root") < 0) ||
- (asprintf((char**)(envp + 3), "USER=%s", arg_user? arg_user : "root") < 0) ||
- (asprintf((char**)(envp + 4), "LOGNAME=%s", arg_user? arg_user : "root") < 0)) {
+ if ((asprintf((char**)(envp + 3), "HOME=%s", home? home: "/root") < 0) ||
+ (asprintf((char**)(envp + 4), "USER=%s", arg_user? arg_user : "root") < 0) ||
+ (asprintf((char**)(envp + 5), "LOGNAME=%s", arg_user? arg_user : "root") < 0)) {
log_error("Out of memory");
goto child_fail;
}