core: add namespace_fork() helper, that forks, joins a set of namespaces and forks again

This helper is useful to ensure pidns/userns joining is properly
executed (as that requires a fork after the setns()). This is
particularly important when it comes to /proc/self/ access or
SCM_CREDENTIALS, but is generally the safer mode of operation.

2 files changed, 56 insertions, 0 deletions

diff --git a/src/basic/process-util.c b/src/basic/process-util.c
index b2aab853e2..572d7eb38d 100644
--- a/src/basic/process-util.c
+++ b/src/basic/process-util.c
@@ -1409,6 +1409,60 @@ int safe_fork_full(
         return 0;
 }
 
+int namespace_fork(
+                const char *outer_name,
+                const char *inner_name,
+                const int except_fds[],
+                size_t n_except_fds,
+                ForkFlags flags,
+                int pidns_fd,
+                int mntns_fd,
+                int netns_fd,
+                int userns_fd,
+                int root_fd,
+                pid_t *ret_pid) {
+
+        int r;
+
+        /* This is much like safe_fork(), but forks twice, and joins the specified namespaces in the middle
+         * process. This ensures that we are fully a member of the destination namespace, with pidns an all, so that
+         * /proc/self/fd works correctly. */
+
+        r = safe_fork_full(outer_name, except_fds, n_except_fds, (flags|FORK_DEATHSIG) & ~(FORK_REOPEN_LOG|FORK_NEW_MOUNTNS|FORK_MOUNTNS_SLAVE), ret_pid);
+        if (r < 0)
+                return r;
+        if (r == 0) {
+                pid_t pid;
+
+                /* Child */
+
+                r = namespace_enter(pidns_fd, mntns_fd, netns_fd, userns_fd, root_fd);
+                if (r < 0) {
+                        log_full_errno(FLAGS_SET(flags, FORK_LOG) ? LOG_ERR : LOG_DEBUG, r, "Failed to join namespace: %m");
+                        _exit(EXIT_FAILURE);
+                }
+
+                /* We mask a few flags here that either make no sense for the grandchild, or that we don't have to do again */
+                r = safe_fork_full(inner_name, except_fds, n_except_fds, flags & ~(FORK_WAIT|FORK_RESET_SIGNALS|FORK_CLOSE_ALL_FDS|FORK_NULL_STDIO), &pid);
+                if (r < 0)
+                        _exit(EXIT_FAILURE);
+                if (r == 0) {
+                        /* Child */
+                        if (ret_pid)
+                                *ret_pid = pid;
+                        return 0;
+                }
+
+                r = wait_for_terminate_and_check(inner_name, pid, FLAGS_SET(flags, FORK_LOG) ? WAIT_LOG : 0);
+                if (r < 0)
+                        _exit(EXIT_FAILURE);
+
+                _exit(r);
+        }
+
+        return 1;
+}
+
 int fork_agent(const char *name, const int except[], size_t n_except, pid_t *ret_pid, const char *path, ...) {
         bool stdout_is_tty, stderr_is_tty;
         size_t n, i;
diff --git a/src/basic/process-util.h b/src/basic/process-util.h
index ca4e4401a9..af47513fab 100644
--- a/src/basic/process-util.h
+++ b/src/basic/process-util.h
@@ -159,6 +159,8 @@ static inline int safe_fork(const char *name, ForkFlags flags, pid_t *ret_pid) {
         return safe_fork_full(name, NULL, 0, flags, ret_pid);
 }
 
+int namespace_fork(const char *outer_name, const char *inner_name, const int except_fds[], size_t n_except_fds, ForkFlags flags, int pidns_fd, int mntns_fd, int netns_fd, int userns_fd, int root_fd, pid_t *ret_pid);
+
 int fork_agent(const char *name, const int except[], size_t n_except, pid_t *pid, const char *path, ...) _sentinel_;
 
 int set_oom_score_adjust(int value);