* [PATCH] net_sched: prio: use qdisc_dequeue_peeked
@ 2011-08-09 12:04 Florian Westphal
2011-08-09 12:56 ` Eric Dumazet
2011-08-10 4:52 ` David Miller
0 siblings, 2 replies; 4+ messages in thread
From: Florian Westphal @ 2011-08-09 12:04 UTC (permalink / raw)
To: netdev; +Cc: Florian Westphal, Eric Dumazet
commit 07bd8df5df4369487812bf85a237322ff3569b77
(sch_sfq: fix peek() implementation) changed sfq to use generic
peek helper.
This makes HFSC complain about a non-work-conserving child qdisc, if
prio with sfq child is used within hfsc:
hfsc peeks into prio qdisc, which will then peek into sfq.
returned skb is stashed in sch->gso_skb.
Next, hfsc tries to dequeue from prio, but prio will call sfq dequeue
directly, which may return NULL instead of previously peeked-at skb.
Have prio call qdisc_dequeue_peeked, so sfq->dequeue() is
not called in this case.
Cc: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: Florian Westphal <fw@strlen.de>
---
Eric, does this look correct to you?
I am not sure if sfq needs fixing instead of this patch here.
diff --git a/net/sched/sch_prio.c b/net/sched/sch_prio.c
index 2a318f2..b5d56a2 100644
--- a/net/sched/sch_prio.c
+++ b/net/sched/sch_prio.c
@@ -112,7 +112,7 @@ static struct sk_buff *prio_dequeue(struct Qdisc *sch)
for (prio = 0; prio < q->bands; prio++) {
struct Qdisc *qdisc = q->queues[prio];
- struct sk_buff *skb = qdisc->dequeue(qdisc);
+ struct sk_buff *skb = qdisc_dequeue_peeked(qdisc);
if (skb) {
qdisc_bstats_update(sch, skb);
sch->q.qlen--;
--
1.7.3.4
^ permalink raw reply related [flat|nested] 4+ messages in thread* Re: [PATCH] net_sched: prio: use qdisc_dequeue_peeked
2011-08-09 12:04 [PATCH] net_sched: prio: use qdisc_dequeue_peeked Florian Westphal
@ 2011-08-09 12:56 ` Eric Dumazet
2011-08-09 13:45 ` Florian Westphal
2011-08-10 4:52 ` David Miller
1 sibling, 1 reply; 4+ messages in thread
From: Eric Dumazet @ 2011-08-09 12:56 UTC (permalink / raw)
To: Florian Westphal; +Cc: netdev
Le mardi 09 août 2011 à 14:04 +0200, Florian Westphal a écrit :
> commit 07bd8df5df4369487812bf85a237322ff3569b77
> (sch_sfq: fix peek() implementation) changed sfq to use generic
> peek helper.
>
> This makes HFSC complain about a non-work-conserving child qdisc, if
> prio with sfq child is used within hfsc:
>
> hfsc peeks into prio qdisc, which will then peek into sfq.
> returned skb is stashed in sch->gso_skb.
>
> Next, hfsc tries to dequeue from prio, but prio will call sfq dequeue
> directly, which may return NULL instead of previously peeked-at skb.
>
> Have prio call qdisc_dequeue_peeked, so sfq->dequeue() is
> not called in this case.
>
> Cc: Eric Dumazet <eric.dumazet@gmail.com>
> Signed-off-by: Florian Westphal <fw@strlen.de>
> ---
> Eric, does this look correct to you?
> I am not sure if sfq needs fixing instead of this patch here.
>
> diff --git a/net/sched/sch_prio.c b/net/sched/sch_prio.c
> index 2a318f2..b5d56a2 100644
> --- a/net/sched/sch_prio.c
> +++ b/net/sched/sch_prio.c
> @@ -112,7 +112,7 @@ static struct sk_buff *prio_dequeue(struct Qdisc *sch)
>
> for (prio = 0; prio < q->bands; prio++) {
> struct Qdisc *qdisc = q->queues[prio];
> - struct sk_buff *skb = qdisc->dequeue(qdisc);
> + struct sk_buff *skb = qdisc_dequeue_peeked(qdisc);
> if (skb) {
> qdisc_bstats_update(sch, skb);
> sch->q.qlen--;
Hi Florian
Are you sure this patch is still needed, after commit
e1738bd9cecc5c867b0e2996470c1ff20f66ba79
(sch_sfq: fix sfq_enqueue())
I am asking before fully reviewing the code, I dont know who
should/shouldnt call qdisc_dequeue_peeked() instead of qdisc->dequeue()
Thanks !
^ permalink raw reply [flat|nested] 4+ messages in thread* Re: [PATCH] net_sched: prio: use qdisc_dequeue_peeked
2011-08-09 12:56 ` Eric Dumazet
@ 2011-08-09 13:45 ` Florian Westphal
0 siblings, 0 replies; 4+ messages in thread
From: Florian Westphal @ 2011-08-09 13:45 UTC (permalink / raw)
To: Eric Dumazet; +Cc: Florian Westphal, netdev
Eric Dumazet <eric.dumazet@gmail.com> wrote:
> > Eric, does this look correct to you?
> > I am not sure if sfq needs fixing instead of this patch here.
> >
> > diff --git a/net/sched/sch_prio.c b/net/sched/sch_prio.c
> > index 2a318f2..b5d56a2 100644
> > --- a/net/sched/sch_prio.c
> > +++ b/net/sched/sch_prio.c
> > @@ -112,7 +112,7 @@ static struct sk_buff *prio_dequeue(struct Qdisc *sch)
> >
> > for (prio = 0; prio < q->bands; prio++) {
> > struct Qdisc *qdisc = q->queues[prio];
> > - struct sk_buff *skb = qdisc->dequeue(qdisc);
> > + struct sk_buff *skb = qdisc_dequeue_peeked(qdisc);
> > if (skb) {
> > qdisc_bstats_update(sch, skb);
> > sch->q.qlen--;
>
>
> Hi Florian
>
> Are you sure this patch is still needed, after commit
> e1738bd9cecc5c867b0e2996470c1ff20f66ba79
> (sch_sfq: fix sfq_enqueue())
Yes, I double checked. This patch is applied, also
I see HFSC complaints even without that code path touched by
"sch_sfq: fix sfq_enqueue" being hit.
^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: [PATCH] net_sched: prio: use qdisc_dequeue_peeked
2011-08-09 12:04 [PATCH] net_sched: prio: use qdisc_dequeue_peeked Florian Westphal
2011-08-09 12:56 ` Eric Dumazet
@ 2011-08-10 4:52 ` David Miller
1 sibling, 0 replies; 4+ messages in thread
From: David Miller @ 2011-08-10 4:52 UTC (permalink / raw)
To: fw; +Cc: netdev, eric.dumazet
From: Florian Westphal <fw@strlen.de>
Date: Tue, 9 Aug 2011 14:04:43 +0200
> commit 07bd8df5df4369487812bf85a237322ff3569b77
> (sch_sfq: fix peek() implementation) changed sfq to use generic
> peek helper.
>
> This makes HFSC complain about a non-work-conserving child qdisc, if
> prio with sfq child is used within hfsc:
>
> hfsc peeks into prio qdisc, which will then peek into sfq.
> returned skb is stashed in sch->gso_skb.
>
> Next, hfsc tries to dequeue from prio, but prio will call sfq dequeue
> directly, which may return NULL instead of previously peeked-at skb.
>
> Have prio call qdisc_dequeue_peeked, so sfq->dequeue() is
> not called in this case.
>
> Cc: Eric Dumazet <eric.dumazet@gmail.com>
> Signed-off-by: Florian Westphal <fw@strlen.de>
Applied and queued up for -stable, thanks!
^ permalink raw reply [flat|nested] 4+ messages in thread
end of thread, other threads:[~2011-08-10 4:52 UTC | newest]
Thread overview: 4+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2011-08-09 12:04 [PATCH] net_sched: prio: use qdisc_dequeue_peeked Florian Westphal
2011-08-09 12:56 ` Eric Dumazet
2011-08-09 13:45 ` Florian Westphal
2011-08-10 4:52 ` David Miller
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox