Merge 3aa3f130c1 into f6793bbcf0

"nsenter -a" doesn't migrate the specified process into the target
cgroup (it really should). Thus the cgroup will remain in a cgroup
that is (due to cgroup ns) outside our visibility. The kernel will
report the cgroup path of such cgroups as starting with "/../". Detect
that and print a reasonably error message instead of trying to resolve
that.

src/basic/cgroup-util.c

@@ -803,6 +803,10 @@ int cg_pid_get_path(const char *controller, pid_t pid, char **ret_path) {
                 if (!path)
                         return -ENOMEM;
 
+                /* Refuse cgroup paths from outside our cgroup namespace */
+                if (startswith(path, "/../"))
+                        return -EUNATCH;
+
                 /* Truncate suffix indicating the process is a zombie */
                 e = endswith(path, " (deleted)");
                 if (e)

src/basic/process-util.c

@@ -102,8 +102,8 @@ int pid_get_comm(pid_t pid, char **ret) {
         _cleanup_free_ char *escaped = NULL, *comm = NULL;
         int r;
 
-        assert(ret);
         assert(pid >= 0);
+        assert(ret);
 
         if (pid == 0 || pid == getpid_cached()) {
                 comm = new0(char, TASK_COMM_LEN + 1); /* Must fit in 16 byte according to prctl(2) */
@@ -143,6 +143,9 @@ int pidref_get_comm(const PidRef *pid, char **ret) {
         if (!pidref_is_set(pid))
                 return -ESRCH;
 
+        if (pidref_is_remote(pid))
+                return -EREMOTE;
+
         r = pid_get_comm(pid->pid, &comm);
         if (r < 0)
                 return r;
@@ -289,6 +292,9 @@ int pidref_get_cmdline(const PidRef *pid, size_t max_columns, ProcessCmdlineFlag
         if (!pidref_is_set(pid))
                 return -ESRCH;
 
+        if (pidref_is_remote(pid))
+                return -EREMOTE;
+
         r = pid_get_cmdline(pid->pid, max_columns, flags, &s);
         if (r < 0)
                 return r;
@@ -331,6 +337,9 @@ int pidref_get_cmdline_strv(const PidRef *pid, ProcessCmdlineFlags flags, char *
         if (!pidref_is_set(pid))
                 return -ESRCH;
 
+        if (pidref_is_remote(pid))
+                return -EREMOTE;
+
         r = pid_get_cmdline_strv(pid->pid, flags, &args);
         if (r < 0)
                 return r;
@@ -477,6 +486,9 @@ int pidref_is_kernel_thread(const PidRef *pid) {
         if (!pidref_is_set(pid))
                 return -ESRCH;
 
+        if (pidref_is_remote(pid))
+                return -EREMOTE;
+
         result = pid_is_kernel_thread(pid->pid);
         if (result < 0)
                 return result;
@@ -594,6 +606,9 @@ int pidref_get_uid(const PidRef *pid, uid_t *ret) {
         if (!pidref_is_set(pid))
                 return -ESRCH;
 
+        if (pidref_is_remote(pid))
+                return -EREMOTE;
+
         r = pid_get_uid(pid->pid, &uid);
         if (r < 0)
                 return r;
@@ -794,6 +809,9 @@ int pidref_get_start_time(const PidRef *pid, usec_t *ret) {
         if (!pidref_is_set(pid))
                 return -ESRCH;
 
+        if (pidref_is_remote(pid))
+                return -EREMOTE;
+
         r = pid_get_start_time(pid->pid, ret ? &t : NULL);
         if (r < 0)
                 return r;
@@ -1093,6 +1111,9 @@ int pidref_is_my_child(const PidRef *pid) {
         if (!pidref_is_set(pid))
                 return -ESRCH;
 
+        if (pidref_is_remote(pid))
+                return -EREMOTE;
+
         result = pid_is_my_child(pid->pid);
         if (result < 0)
                 return result;
@@ -1128,6 +1149,9 @@ int pidref_is_unwaited(const PidRef *pid) {
         if (!pidref_is_set(pid))
                 return -ESRCH;
 
+        if (pidref_is_remote(pid))
+                return -EREMOTE;
+
         if (pid->pid == 1 || pidref_is_self(pid))
                 return true;
 
@@ -1169,6 +1193,9 @@ int pidref_is_alive(const PidRef *pidref) {
         if (!pidref_is_set(pidref))
                 return -ESRCH;
 
+        if (pidref_is_remote(pidref))
+                return -EREMOTE;
+
         result = pid_is_alive(pidref->pid);
         if (result < 0) {
                 assert(result != -ESRCH);

src/cryptenroll/cryptenroll-fido2.c

@@ -193,7 +193,7 @@ int enroll_fido2(
                 fflush(stdout);
 
                 fprintf(stderr,
-                        "\nPlease save this FIDO2 credential ID. It is required when unloocking the volume\n"
+                        "\nPlease save this FIDO2 credential ID. It is required when unlocking the volume\n"
                         "using the associated FIDO2 keyslot which we just created. To configure automatic\n"
                         "unlocking using this FIDO2 token, add an appropriate entry to your /etc/crypttab\n"
                         "file, see %s for details.\n", link);

src/shared/killall.c

@@ -46,13 +46,17 @@ static bool argv_has_at(pid_t pid) {
         return c == '@';
 }
 
-static bool is_survivor_cgroup(const PidRef *pid) {
+static bool is_in_survivor_cgroup(const PidRef *pid) {
         _cleanup_free_ char *cgroup_path = NULL;
         int r;
 
         assert(pidref_is_set(pid));
 
         r = cg_pidref_get_path(/* root= */ NULL, pid, &cgroup_path);
+        if (r == -EUNATCH) {
+                log_warning_errno(r, "Process " PID_FMT " appears to originate in foreign namespace, ignoring.", pid->pid);
+                return true;
+        }
         if (r < 0) {
                 log_warning_errno(r, "Failed to get cgroup path of process " PID_FMT ", ignoring: %m", pid->pid);
                 return false;
@@ -86,7 +90,7 @@ static bool ignore_proc(const PidRef *pid, bool warn_rootfs) {
                 return true; /* also ignore processes where we can't determine this */
 
         /* Ignore processes that are part of a cgroup marked with the user.survive_final_kill_signal xattr */
-        if (is_survivor_cgroup(pid))
+        if (is_in_survivor_cgroup(pid))
                 return true;
 
         r = pidref_get_uid(pid, &uid);

src/test/test-audit-util.c

@@ -7,24 +7,26 @@ TEST(audit_loginuid_from_pid) {
         _cleanup_(pidref_done) PidRef self = PIDREF_NULL, pid1 = PIDREF_NULL;
         int r;
 
-        assert_se(pidref_set_self(&self) >= 0);
-        assert_se(pidref_set_pid(&pid1, 1) >= 0);
+        ASSERT_OK(pidref_set_self(&self));
+        ASSERT_OK(pidref_set_pid(&pid1, 1));
 
         uid_t uid;
         r = audit_loginuid_from_pid(&self, &uid);
-        assert_se(r >= 0 || r == -ENODATA);
+        if (r != -ENODATA)
+                ASSERT_OK(r);
         if (r >= 0)
                 log_info("self audit login uid: " UID_FMT, uid);
 
-        assert_se(audit_loginuid_from_pid(&pid1, &uid) == -ENODATA);
+        ASSERT_ERROR(audit_loginuid_from_pid(&pid1, &uid), ENODATA);
 
         uint32_t sessionid;
         r = audit_session_from_pid(&self, &sessionid);
-        assert_se(r >= 0 || r == -ENODATA);
+        if (r != -ENODATA)
+                ASSERT_OK(r);
         if (r >= 0)
                 log_info("self audit session id: %" PRIu32, sessionid);
 
-        assert_se(audit_session_from_pid(&pid1, &sessionid) == -ENODATA);
+        ASSERT_ERROR(audit_session_from_pid(&pid1, &sessionid), ENODATA);
 }
 
 static int intro(void) {

test/units/TEST-13-NSPAWN.machined.sh

@@ -22,6 +22,11 @@ trap at_exit EXIT
 
 systemctl service-log-level systemd-machined debug
 systemctl service-log-level systemd-importd debug
+# per request in https://github.com/systemd/systemd/pull/35117
+systemctl edit --runtime --stdin 'systemd-nspawn@.service' --drop-in=debug.conf <<EOF
+[Service]
+Environment=SYSTEMD_LOG_LEVEL=debug
+EOF
 
 # Mount temporary directory over /var/lib/machines to not pollute the image
 mkdir -p /var/lib/machines
@@ -278,13 +283,13 @@ varlinkctl call /run/systemd/machine/io.systemd.Machine io.systemd.Machine.List
 # sending TRAP signal
 rm -f /var/lib/machines/long-running/trap
 varlinkctl call /run/systemd/machine/io.systemd.Machine io.systemd.Machine.Kill '{"name":"long-running", "whom": "leader", "signal": 5}'
-timeout 30 bash -c "until test -e /var/lib/machines/long-running/trap; do sleep .5; done"
+timeout 120 bash -c "until test -e /var/lib/machines/long-running/trap; do sleep .5; done"
 
 # test io.systemd.Machine.Terminate
 long_running_machine_start
 rm -f /var/lib/machines/long-running/terminate
 varlinkctl call /run/systemd/machine/io.systemd.Machine io.systemd.Machine.Terminate '{"name":"long-running"}'
-timeout 10 bash -c "until test -e /var/lib/machines/long-running/terminate; do sleep .5; done"
+timeout 30 bash -c "until test -e /var/lib/machines/long-running/terminate; do sleep .5; done"
 timeout 30 bash -c "while varlinkctl call /run/systemd/machine/io.systemd.Machine io.systemd.Machine.List '{\"name\":\"long-running\"}'; do sleep 0.5; done"
 
 # test io.systemd.Machine.Register
@@ -356,7 +361,7 @@ journalctl --sync
 machinectl terminate container-without-os-release
 machinectl terminate long-running
 # wait for the container being stopped, otherwise acquiring image metadata by io.systemd.MachineImage.List may fail in the below.
-timeout 10 bash -c "while machinectl status long-running &>/dev/null; do sleep .5; done"
+timeout 30 bash -c "while machinectl status long-running &>/dev/null; do sleep .5; done"
 systemctl kill --signal=KILL systemd-nspawn@long-running.service || :
 
 (ip addr show lo | grep -q 192.168.1.100) || ip address add 192.168.1.100/24 dev lo