When a task calls the function wait_task_interactive(), it waits on the task to change its tasks state. But it does not take into account that spin_locks() in RT will change the real state of the task, but keep a mirror copy of the state, that it resets to when the task unlocks the spin lock. The wait_task_interactive() may get the wrong result because it sees the temporary state caused by the RT spin lock and not the real state.
Created attachment 567100 [details] Have wait_task_interactive() check saved state of task In RT, spin locks store the task's state in a shadow copy because the wakeup of the spin lock will change it. On unlock the shadow copy is moved to the original tasks state. But this can confuse wait_task_interactive() because it doesn't see the copy of it. This patch makes wait_task_interactive() aware of the copy as well.
The fix here is equivalent to c7004d0 git describe --contains c7004d0 v3.2.14-rt24~8 Documenting this in the changelog section of kernel-rt.spec
Technical note added. If any revisions are required, please edit the "Technical Notes" field accordingly. All revisions will be proofread by the Engineering Content Services team. New Contents: Cause: wait_task_interactive() ignores the rt_spin_locks state Consequence: wait_task_interactive() can return with an incorrect result if the task it is waiting for is blocked on a rt_spin_lock() and is waking up. Fix: wait_task_interactive() must also test the saved_state field saved by rt_spin_locks Result: The correct result is returned from wait_task_interactive()
Since the problem described in this bug report should be resolved in a recent advisory, it has been closed with a resolution of ERRATA. For information on the advisory, and where to find the updated files, follow the link below. If the solution does not work for you, open a new bug report. http://rhn.redhat.com/errata/RHSA-2012-1282.html