On Sun, 2010-07-11 at 15:33 +0200, Mike Galbraith wrote: > On Sat, 2010-07-10 at 21:41 +0200, Mike Galbraith wrote: > diff --git a/kernel/futex.c b/kernel/futex.c > index a6cec32..ef489f3 100644 > --- a/kernel/futex.c > +++ b/kernel/futex.c > @@ -2255,7 +2255,14 @@ static int futex_wait_requeue_pi(u32 __user *uaddr, int fshared, > /* Queue the futex_q, drop the hb lock, wait for wakeup. */ > futex_wait_queue_me(hb, &q, to); > > - spin_lock(&hb->lock); > + /* > + * Non-blocking synchronization point with futex_requeue(). > + * > + * We dare not block here because this will alter PI state, possibly > + * before our waker finishes modifying same in wakeup_next_waiter(). > + */ > + while(!spin_trylock(&hb->lock)) > + cpu_relax(); I agree that this would work. But I wonder if this should have an: #ifdef PREEMPT_RT [...] #else spin_lock(&hb->lock); #endif around it. Or encapsulate this lock in a macro that does the same thing (just to keep the actual code cleaner) -- Steve > ret = handle_early_requeue_pi_wakeup(hb, &q, &key2, to); > spin_unlock(&hb->lock); > if (ret) > > -- To unsubscribe from this list: send the line "unsubscribe linux-rt-users" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html