Bug 586237

+++ This bug was initially created as a clone of Bug #582237 +++

Pavel Emelyanov from Parallels Virtuozzo Containers team has found that 2.6.18-194.el5 kernel has a new feature "hung_task" backported from mainstream.
Unfortunately there is some important piece of code is missing:

The bad thing is that the following piece of code (marked with >>>-s) in the check_hung_uninterruptible_tasks() routine was lost:

        do_each_thread(g, t) {
                if (!max_count--)
                        goto unlock;
                if (!--batch_count) {
                        batch_count = HUNG_TASK_BATCHING;
                        rcu_lock_break(g, t);
>>> >>>                     /* Exit if t or g was unhashed during refresh. */
>>> >>>                     if (t->state == TASK_DEAD || g->state == TASK_DEAD)
>>> >>>                             goto unlock;
                }
                /* use "==" to skip the TASK_KILLABLE tasks waiting on NFS */
                if (t->state == TASK_UNINTERRUPTIBLE)
                        check_hung_task(t, timeout);
        } while_each_thread(g, t);


This code is required, since in case t or g goes dead the lists being scanned are no longer self-consistent.

Please, recheck and add the code.