Re: [patch 1/3] ipc/sem: fix -rt livelock

[Manfred Spraul <manfred@colorfullife.com>]

On 10/04/2013 03:55 PM, Mike Galbraith wrote:
> On Fri, 2013-10-04 at 13:28 +0200, Sebastian Andrzej Siewior wrote:
>> On 10/04/2013 01:16 PM, Mike Galbraith wrote:
>>> Manfred's race fix also kills loop, and thereby the -rt livelock, so the
>>> only question is does -rt want to use his completion wakeup scheme, and
>>> does it want to do something about spin_unlock_wait() unconditionally
>>> grabbing/releasing every lock in the array.
>> So his patches went into v3.11 or so right? So all -RT released post
>> this point will have it fixed right?
> Most fixes have gone into master.  There's still at least one in the
> pipe, ipc/sem.c: synchronize semop and semctl with IPC_RMID, and I
> _think_ there's still one open issue.
>
> <quoting Manfred>
> And now new:
> 1) ipc/namespace.c:
>       free_ipcs() still assumes the "old style" free calls:
>       rcu_lock and ipc_lock dropped within the callback.
>
>       freeary() was converted - but free_ipcs was not updated.
>
>       Thus:
>       Closing a namespace with sem arrays and threads that are waiting
> on
> the array with semtimedop() and bad timing can deadlock the semtimedop
> thread.
>       (i.e.: spin_lock() waiting forever).
> </quoting Manfred>
That one was a false alarm, i.e. everything is fixed in -mm.

--
	Manfred