Advice on RCU for IPVS

Advice on RCU for IPVS

[Simon Horman]

Hi,

I have been looking at converting the rwlocks in IPVS over to use RCU.
A problem that I am facing is that the lblcr scheduler uses
a write lock on list A and then taking a write lock on list B.
Where list B is basically part of one of the elements of list A.

This problem is present in ip_vs_lblcr_schedule() and
the key code looks like this.


	/* First look in our cache */
	read_lock(&svc->sched_lock);
	en = ip_vs_lblcr_get(svc->af, tbl, &iph.daddr);
	if (en) {
		...

		/* Get the least loaded destination */
		read_lock(&en->set.lock);
		dest = ip_vs_dest_set_min(&en->set);
		read_unlock(&en->set.lock);

		...

			write_lock(&en->set.lock);
			m = ip_vs_dest_set_max(&en->set);
			if (m)
				ip_vs_dest_set_erase(&en->set, m);
			write_unlock(&en->set.lock);

		...

		/* Update our cache entry */
		write_lock(&en->set.lock);
		ip_vs_dest_set_insert(&en->set, dest);
		write_unlock(&en->set.lock);
	}
	read_unlock(&svc->sched_lock);

dest is referenced counted and doesn't seem to need to be guarded
by svc->sched_lock.

It seems to me that this is quite difficult to convert over to RCU
as there are write-side critical sections inside a read-side critical
section.

I investigated reference counting the return value of
ip_vs_lblcr_get() or the return value of ip_vs_dest_set_max() and
ip_vs_dest_set_insert(). But this seems to be difficult,
especially at rmmod time.

I also considered just making the whole thing a write-side critical section.
Which seems to be somewhat of a sledge-hammer and result in
a critical section that is much larger than I would like. Though
no bigger than the existing area covered by the read-lock on
svc->sched_lock.

Any suggestions would be appreciated.

Re: Advice on RCU for IPVS

[Eric Dumazet]

Le vendredi 26 février 2010 à 14:18 +1100, Simon Horman a écrit :
> Hi,
> 
> I have been looking at converting the rwlocks in IPVS over to use RCU.
> A problem that I am facing is that the lblcr scheduler uses
> a write lock on list A and then taking a write lock on list B.
> Where list B is basically part of one of the elements of list A.
> 
> This problem is present in ip_vs_lblcr_schedule() and
> the key code looks like this.
> 
> 
> 	/* First look in our cache */
> 	read_lock(&svc->sched_lock);
> 	en = ip_vs_lblcr_get(svc->af, tbl, &iph.daddr);
> 	if (en) {
> 		...
> 
> 		/* Get the least loaded destination */
> 		read_lock(&en->set.lock);
> 		dest = ip_vs_dest_set_min(&en->set);
> 		read_unlock(&en->set.lock);
> 
> 		...
> 
> 			write_lock(&en->set.lock);
> 			m = ip_vs_dest_set_max(&en->set);
> 			if (m)
> 				ip_vs_dest_set_erase(&en->set, m);
> 			write_unlock(&en->set.lock);
> 
> 		...
> 
> 		/* Update our cache entry */
> 		write_lock(&en->set.lock);
> 		ip_vs_dest_set_insert(&en->set, dest);
> 		write_unlock(&en->set.lock);
> 	}
> 	read_unlock(&svc->sched_lock);
> 
> dest is referenced counted and doesn't seem to need to be guarded
> by svc->sched_lock.
> 
> It seems to me that this is quite difficult to convert over to RCU
> as there are write-side critical sections inside a read-side critical
> section.
> 
> I investigated reference counting the return value of
> ip_vs_lblcr_get() or the return value of ip_vs_dest_set_max() and
> ip_vs_dest_set_insert(). But this seems to be difficult,
> especially at rmmod time.
> 
> I also considered just making the whole thing a write-side critical section.
> Which seems to be somewhat of a sledge-hammer and result in
> a critical section that is much larger than I would like. Though
> no bigger than the existing area covered by the read-lock on
> svc->sched_lock.
> 
> Any suggestions would be appreciated.

The code you copy/pasted seems really complex, I would suggest to make
it as simple as possible (using spinlocks for example instead of
rwlocks) before considering RCU conversion.