pid: Handle the exit of a multi-threaded init.
authorEric W. Biederman <ebiederm@xmission.com>
Tue, 26 Mar 2013 09:27:11 +0000 (02:27 -0700)
committerEric W. Biederman <ebiederm@xmission.com>
Tue, 26 Mar 2013 10:41:23 +0000 (03:41 -0700)
When a multi-threaded init exits and the initial thread is not the
last thread to exit the initial thread hangs around as a zombie
until the last thread exits.  In that case zap_pid_ns_processes
needs to wait until there are only 2 hashed pids in the pid
namespace not one.

v2. Replace thread_pid_vnr(me) == 1 with the test thread_group_leader(me)
    as suggested by Oleg.

Cc: stable@vger.kernel.org
Cc: Oleg Nesterov <oleg@redhat.com>
Reported-by: Caj Larsson <caj@omnicloud.com>
Signed-off-by: "Eric W. Biederman" <ebiederm@xmission.com>
kernel/pid_namespace.c

index c1c3dc1c60233f337a01ff13587f1a5a7f57f1cd..bea15bdf82b04c28d2f5d8c1a32a46b2621cb295 100644 (file)
@@ -181,6 +181,7 @@ void zap_pid_ns_processes(struct pid_namespace *pid_ns)
        int nr;
        int rc;
        struct task_struct *task, *me = current;
+       int init_pids = thread_group_leader(me) ? 1 : 2;
 
        /* Don't allow any more processes into the pid namespace */
        disable_pid_allocation(pid_ns);
@@ -230,7 +231,7 @@ void zap_pid_ns_processes(struct pid_namespace *pid_ns)
         */
        for (;;) {
                set_current_state(TASK_UNINTERRUPTIBLE);
-               if (pid_ns->nr_hashed == 1)
+               if (pid_ns->nr_hashed == init_pids)
                        break;
                schedule();
        }