Patchwork codel: refine one condition to avoid a nul rec_inv_sqrt

login
register
mail settings
Submitter Eric Dumazet
Date July 30, 2012, 6:52 a.m.
Message ID <1343631141.2626.13293.camel@edumazet-glaptop>
Download mbox | patch
Permalink /patch/173968/
State Accepted
Delegated to: David Miller
Headers show

Comments

Eric Dumazet - July 30, 2012, 6:52 a.m.
From: Eric Dumazet <edumazet@google.com>

One condition before codel_Newton_step() was not good if
we never left the dropping state for a flow. As a result
rec_inv_sqrt was 0, instead of the ~0 initial value.

codel control law was then set to a very aggressive mode, dropping
many packets before reaching 'target' and recovering from this problem.

To keep codel_vars_init() as efficient as possible, refine
the condition to make sure rec_inv_sqrt initial value is correct

Many thanks to Anton Mich for discovering the issue and suggesting
a fix.

Reported-by: Anton Mich <lp2s1h@gmail.com>
Signed-off-by: Eric Dumazet <edumazet@google.com>
---
 include/net/codel.h |    8 ++++++--
 1 file changed, 6 insertions(+), 2 deletions(-)



--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
David Miller - July 30, 2012, 9:54 p.m.
From: Eric Dumazet <eric.dumazet@gmail.com>
Date: Mon, 30 Jul 2012 08:52:21 +0200

> From: Eric Dumazet <edumazet@google.com>
> 
> One condition before codel_Newton_step() was not good if
> we never left the dropping state for a flow. As a result
> rec_inv_sqrt was 0, instead of the ~0 initial value.
> 
> codel control law was then set to a very aggressive mode, dropping
> many packets before reaching 'target' and recovering from this problem.
> 
> To keep codel_vars_init() as efficient as possible, refine
> the condition to make sure rec_inv_sqrt initial value is correct
> 
> Many thanks to Anton Mich for discovering the issue and suggesting
> a fix.
> 
> Reported-by: Anton Mich <lp2s1h@gmail.com>
> Signed-off-by: Eric Dumazet <edumazet@google.com>

Applied and queued up for -stable.
--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Eric Dumazet - Aug. 10, 2012, 3:16 p.m.
On Mon, 2012-07-30 at 14:54 -0700, David Miller wrote:
> From: Eric Dumazet <eric.dumazet@gmail.com>
> Date: Mon, 30 Jul 2012 08:52:21 +0200
> 
> > From: Eric Dumazet <edumazet@google.com>
> > 
> > One condition before codel_Newton_step() was not good if
> > we never left the dropping state for a flow. As a result
> > rec_inv_sqrt was 0, instead of the ~0 initial value.
> > 
> > codel control law was then set to a very aggressive mode, dropping
> > many packets before reaching 'target' and recovering from this problem.
> > 
> > To keep codel_vars_init() as efficient as possible, refine
> > the condition to make sure rec_inv_sqrt initial value is correct
> > 
> > Many thanks to Anton Mich for discovering the issue and suggesting
> > a fix.
> > 
> > Reported-by: Anton Mich <lp2s1h@gmail.com>
> > Signed-off-by: Eric Dumazet <edumazet@google.com>
> 
> Applied and queued up for -stable.

Hi David, I think this patch got lost somehow ?

Thanks


--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
David Miller - Aug. 10, 2012, 11:54 p.m.
From: Eric Dumazet <eric.dumazet@gmail.com>
Date: Fri, 10 Aug 2012 17:16:11 +0200

> On Mon, 2012-07-30 at 14:54 -0700, David Miller wrote:
>> From: Eric Dumazet <eric.dumazet@gmail.com>
>> Date: Mon, 30 Jul 2012 08:52:21 +0200
>> 
>> > From: Eric Dumazet <edumazet@google.com>
>> > 
>> > One condition before codel_Newton_step() was not good if
>> > we never left the dropping state for a flow. As a result
>> > rec_inv_sqrt was 0, instead of the ~0 initial value.
>> > 
>> > codel control law was then set to a very aggressive mode, dropping
>> > many packets before reaching 'target' and recovering from this problem.
>> > 
>> > To keep codel_vars_init() as efficient as possible, refine
>> > the condition to make sure rec_inv_sqrt initial value is correct
>> > 
>> > Many thanks to Anton Mich for discovering the issue and suggesting
>> > a fix.
>> > 
>> > Reported-by: Anton Mich <lp2s1h@gmail.com>
>> > Signed-off-by: Eric Dumazet <edumazet@google.com>
>> 
>> Applied and queued up for -stable.
> 
> Hi David, I think this patch got lost somehow ?

Thanks for catching this, I've corrected it now.
--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Patch

diff --git a/include/net/codel.h b/include/net/codel.h
index 550debf..389cf62 100644
--- a/include/net/codel.h
+++ b/include/net/codel.h
@@ -305,6 +305,8 @@  static struct sk_buff *codel_dequeue(struct Qdisc *sch,
 			}
 		}
 	} else if (drop) {
+		u32 delta;
+
 		if (params->ecn && INET_ECN_set_ce(skb)) {
 			stats->ecn_mark++;
 		} else {
@@ -320,9 +322,11 @@  static struct sk_buff *codel_dequeue(struct Qdisc *sch,
 		 * assume that the drop rate that controlled the queue on the
 		 * last cycle is a good starting point to control it now.
 		 */
-		if (codel_time_before(now - vars->drop_next,
+		delta = vars->count - vars->lastcount;
+		if (delta > 1 &&
+		    codel_time_before(now - vars->drop_next,
 				      16 * params->interval)) {
-			vars->count = (vars->count - vars->lastcount) | 1;
+			vars->count = delta;
 			/* we dont care if rec_inv_sqrt approximation
 			 * is not very precise :
 			 * Next Newton steps will correct it quadratically.