in seccomp/user-trap.c [201:375]
int main(void)
{
int sk_pair[2], ret = 1, status, listener;
pid_t worker = 0 , tracer = 0;
if (socketpair(PF_LOCAL, SOCK_SEQPACKET, 0, sk_pair) < 0) {
perror("socketpair");
return 1;
}
worker = fork();
if (worker < 0) {
perror("fork");
goto close_pair;
}
if (worker == 0) {
listener = user_trap_syscall(__NR_mount,
SECCOMP_FILTER_FLAG_NEW_LISTENER);
if (listener < 0) {
perror("seccomp");
exit(1);
}
/*
* Drop privileges. We definitely can't mount as uid 1000.
*/
if (setuid(1000) < 0) {
perror("setuid");
exit(1);
}
/*
* Send the listener to the parent; also serves as
* synchronization.
*/
if (send_fd(sk_pair[1], listener) < 0)
exit(1);
close(listener);
if (mkdir("/tmp/foo", 0755) < 0) {
perror("mkdir");
exit(1);
}
/*
* Try a bad mount just for grins.
*/
if (mount("/dev/sda", "/tmp/foo", NULL, 0, NULL) != -1) {
fprintf(stderr, "huh? mounted /dev/sda?\n");
exit(1);
}
if (errno != EPERM) {
perror("bad error from mount");
exit(1);
}
/*
* Ok, we expect this one to succeed.
*/
if (mount("/tmp/foo", "/tmp/foo", NULL, MS_BIND, NULL) < 0) {
perror("mount");
exit(1);
}
exit(0);
}
/*
* Get the listener from the child.
*/
listener = recv_fd(sk_pair[0]);
if (listener < 0)
goto out_kill;
/*
* Fork a task to handle the requests. This isn't strictly necessary,
* but it makes the particular writing of this sample easier, since we
* can just wait ofr the tracee to exit and kill the tracer.
*/
tracer = fork();
if (tracer < 0) {
perror("fork");
goto out_kill;
}
if (tracer == 0) {
struct seccomp_notif *req;
struct seccomp_notif_resp *resp;
struct seccomp_notif_sizes sizes;
if (seccomp(SECCOMP_GET_NOTIF_SIZES, 0, &sizes) < 0) {
perror("seccomp(GET_NOTIF_SIZES)");
goto out_close;
}
req = malloc(sizes.seccomp_notif);
if (!req)
goto out_close;
resp = malloc(sizes.seccomp_notif_resp);
if (!resp)
goto out_req;
memset(resp, 0, sizes.seccomp_notif_resp);
while (1) {
memset(req, 0, sizes.seccomp_notif);
if (ioctl(listener, SECCOMP_IOCTL_NOTIF_RECV, req)) {
perror("ioctl recv");
goto out_resp;
}
if (handle_req(req, resp, listener) < 0)
goto out_resp;
/*
* ENOENT here means that the task may have gotten a
* signal and restarted the syscall. It's up to the
* handler to decide what to do in this case, but for
* the sample code, we just ignore it. Probably
* something better should happen, like undoing the
* mount, or keeping track of the args to make sure we
* don't do it again.
*/
if (ioctl(listener, SECCOMP_IOCTL_NOTIF_SEND, resp) < 0 &&
errno != ENOENT) {
perror("ioctl send");
goto out_resp;
}
}
out_resp:
free(resp);
out_req:
free(req);
out_close:
close(listener);
exit(1);
}
close(listener);
if (waitpid(worker, &status, 0) != worker) {
perror("waitpid");
goto out_kill;
}
if (umount2("/tmp/foo", MNT_DETACH) < 0 && errno != EINVAL) {
perror("umount2");
goto out_kill;
}
if (remove("/tmp/foo") < 0 && errno != ENOENT) {
perror("remove");
exit(1);
}
if (!WIFEXITED(status) || WEXITSTATUS(status)) {
fprintf(stderr, "worker exited nonzero\n");
goto out_kill;
}
ret = 0;
out_kill:
if (tracer > 0)
kill(tracer, SIGKILL);
if (worker > 0)
kill(worker, SIGKILL);
close_pair:
close(sk_pair[0]);
close(sk_pair[1]);
return ret;
}