Re: [PATCH] stop_machine: Disable preemption after queueing stopper threads

[Peter Zijlstra]

On Mon, Jul 30, 2018 at 12:20:57PM +0200, Thomas Gleixner wrote:
> On Tue, 24 Jul 2018, Sebastian Andrzej Siewior wrote:
> > On 2018-07-23 18:13:48 [-0700], isaacm@xxxxxxxxxxxxxx wrote:
> > > Hi all,
> > Hi,
> > 
> > > Are there any comments about this patch?
> > 
> > I haven't look in detail at this but your new preempt_disable() makes
> > things unbalanced for the err != 0 case.
> 
> It doesn't but that code is really an unreadable pile of ...

---
Subject: stop_machine: Reflow cpu_stop_queue_two_works()

The code flow in cpu_stop_queue_two_works() is a little arcane; fix
this by lifting the preempt_disable() to the top to create more natural
nesting wrt the spinlocks and make the wake_up_q() and preempt_enable()
unconditional at the end.

Furthermore, enable preemption in the -EDEADLK case, such that we
spin-wait with preemption enabled.

Suggested-by: Thomas Gleixner <tglx@xxxxxxxxxxxxx>
Signed-off-by: Peter Zijlstra (Intel) <peterz@xxxxxxxxxxxxx>
---
 kernel/stop_machine.c | 41 +++++++++++++++++++++++------------------
 1 file changed, 23 insertions(+), 18 deletions(-)

diff --git a/kernel/stop_machine.c b/kernel/stop_machine.c
index e190d1ef3a23..34b6652e8677 100644
--- a/kernel/stop_machine.c
+++ b/kernel/stop_machine.c
@@ -236,13 +236,24 @@ static int cpu_stop_queue_two_works(int cpu1, struct cpu_stop_work *work1,
 	struct cpu_stopper *stopper2 = per_cpu_ptr(&cpu_stopper, cpu2);
 	DEFINE_WAKE_Q(wakeq);
 	int err;
+
 retry:
+	/*
+	 * The waking up of stopper threads has to happen in the same
+	 * scheduling context as the queueing.  Otherwise, there is a
+	 * possibility of one of the above stoppers being woken up by another
+	 * CPU, and preempting us. This will cause us to not wake up the other
+	 * stopper forever.
+	 */
+	preempt_disable();
 	raw_spin_lock_irq(&stopper1->lock);
 	raw_spin_lock_nested(&stopper2->lock, SINGLE_DEPTH_NESTING);
 
-	err = -ENOENT;
-	if (!stopper1->enabled || !stopper2->enabled)
+	if (!stopper1->enabled || !stopper2->enabled) {
+		err = -ENOENT;
 		goto unlock;
+	}
+
 	/*
 	 * Ensure that if we race with __stop_cpus() the stoppers won't get
 	 * queued up in reverse order leading to system deadlock.
@@ -253,36 +264,30 @@ static int cpu_stop_queue_two_works(int cpu1, struct cpu_stop_work *work1,
 	 * It can be falsely true but it is safe to spin until it is cleared,
 	 * queue_stop_cpus_work() does everything under preempt_disable().
 	 */
-	err = -EDEADLK;
-	if (unlikely(stop_cpus_in_progress))
-			goto unlock;
+	if (unlikely(stop_cpus_in_progress)) {
+		err = -EDEADLK;
+		goto unlock;
+	}
 
 	err = 0;
 	__cpu_stop_queue_work(stopper1, work1, &wakeq);
 	__cpu_stop_queue_work(stopper2, work2, &wakeq);
-	/*
-	 * The waking up of stopper threads has to happen
-	 * in the same scheduling context as the queueing.
-	 * Otherwise, there is a possibility of one of the
-	 * above stoppers being woken up by another CPU,
-	 * and preempting us. This will cause us to n ot
-	 * wake up the other stopper forever.
-	 */
-	preempt_disable();
+
 unlock:
 	raw_spin_unlock(&stopper2->lock);
 	raw_spin_unlock_irq(&stopper1->lock);
 
 	if (unlikely(err == -EDEADLK)) {
+		preempt_enable();
+
 		while (stop_cpus_in_progress)
 			cpu_relax();
+
 		goto retry;
 	}
 
-	if (!err) {
-		wake_up_q(&wakeq);
-		preempt_enable();
-	}
+	wake_up_q(&wakeq);
+	preempt_enable();
 
 	return err;
 }