RE: [PATCH v1 net 1/1] net/sched: sch_dualpi2: fix limit/memlimit enforcement when dequeueing L-queue

[Chia-Yu Chang (Nokia)]

> -----Original Message-----
> From: Victor Nogueira <victor@xxxxxxxxxxxx>
> Sent: Thursday, April 16, 2026 4:27 PM
> To: Chia-Yu Chang (Nokia) <chia-yu.chang@xxxxxxxxxxxxxxxxxxx>; linux-hardening@xxxxxxxxxxxxxxx; kees@xxxxxxxxxx; gustavoars@xxxxxxxxxx; jhs@xxxxxxxxxxxx; jiri@xxxxxxxxxxx; davem@xxxxxxxxxxxxx; edumazet@xxxxxxxxxx; kuba@xxxxxxxxxx; pabeni@xxxxxxxxxx; linux-kernel@xxxxxxxxxxxxxxx; netdev@xxxxxxxxxxxxxxx; horms@xxxxxxxxxx; ij@xxxxxxxxxx; ncardwell@xxxxxxxxxx; Koen De Schepper (Nokia) <koen.de_schepper@xxxxxxxxxxxxxxxxxxx>; g.white@xxxxxxxxxxxxx; ingemar.s.johansson@xxxxxxxxxxxx; mirja.kuehlewind@xxxxxxxxxxxx; cheshire@xxxxxxxxx; rs.ietf@xxxxxx; Jason_Livingood@xxxxxxxxxxx; vidhi_goel@xxxxxxxxx
> Subject: Re: [PATCH v1 net 1/1] net/sched: sch_dualpi2: fix limit/memlimit enforcement when dequeueing L-queue
>
>
> CAUTION: This is an external email. Please be very careful when clicking links or opening attachments. See the URL nok.it/ext for additional information.
>
>
>
> On 13/04/2026 13:37, chia-yu.chang@xxxxxxxxxxxxxxxxxxx wrote:
> > From: Chia-Yu Chang <chia-yu.chang@xxxxxxxxxxxxxxxxxxx>
> >
> > Fix dualpi2_change() to correctly enforce updated limit and memlimit
> > values after a configuration change of the dualpi2 qdisc.
> >
> > Before this patch, dualpi2_change() always attempted to dequeue
> > packets via the root qdisc (C-queue) when reducing backlog or memory
> > usage, and unconditionally assumed that a valid skb will be returned.
> > When traffic classification results in packets being queued in the
> > L-queue while the C-queue is empty, this leads to a NULL skb
> > dereference during limit or memlimit enforcement.
> >
> > This is fixed by first dequeuing from the C-queue path if it is non-empty.
> > Once the C-queue is empty, packets are dequeued directly from the
> > L-queue.s Return values from qdisc_dequeue_internal() are checked for
> > both queues. When dequeuing from the L-queue, the parent qdisc qlen
> > and backlog counters are updated explicitly to keep overall qdisc statistics consistent.
> > [...]
> > ---
> >   net/sched/sch_dualpi2.c | 24 +++++++++++++++++++-----
> >   1 file changed, 19 insertions(+), 5 deletions(-)
> >
> > diff --git a/net/sched/sch_dualpi2.c b/net/sched/sch_dualpi2.c index
> > 6d7e6389758d..56d4422970b6 100644
> > --- a/net/sched/sch_dualpi2.c
> > +++ b/net/sched/sch_dualpi2.c
> > @@ -872,11 +872,25 @@ static int dualpi2_change(struct Qdisc *sch, struct nlattr *opt,
> >       old_backlog = sch->qstats.backlog;
> >       while (qdisc_qlen(sch) > sch->limit ||
> >              q->memory_used > q->memory_limit) {
> > -             struct sk_buff *skb = qdisc_dequeue_internal(sch, true);
> > -
> > -             q->memory_used -= skb->truesize;
> > -             qdisc_qstats_backlog_dec(sch, skb);
> > -             rtnl_qdisc_drop(skb, sch);
> > +             int c_len = qdisc_qlen(sch) - qdisc_qlen(q->l_queue);
> > +             struct sk_buff *skb = NULL;
> > +
> > +             if (c_len) {
> > +                     skb = qdisc_dequeue_internal(sch, true);
> > +                     if (!skb)
> > +                             break;
> > +                     q->memory_used -= skb->truesize;
> > +                     rtnl_qdisc_drop(skb, sch);
> > +             } else if (qdisc_qlen(q->l_queue)) {
> > +                     skb = qdisc_dequeue_internal(q->l_queue, true);
> > +                     if (!skb)
> > +                             break;
> > +                     q->memory_used -= skb->truesize;
> > +                     rtnl_qdisc_drop(skb, q->l_queue);
> > +                     /* Keep the overall qdisc stats consistent */
> > +                     --sch->q.qlen;
> > +                     qdisc_qstats_backlog_dec(sch, skb);
>
> Sashiko is hallucinating saying this will cause a UAF, it won't.
> However it is good to maintain a consistent order here.
> For example, see how sch_choke is doing [1].
>
> [1]
> https://git.kernel.org/pub/scm/linux/kernel/git/netdev/net.git/tree/net/sched/sch_choke.c?id=1f5ffc672165ff851063a5fd044b727ab2517ae3#n394
>
> cheers,
> Victor

Hi Victor,

Thanks for the pointer to sch_choke, it follows the order: (1) qdisc_qstats_backlog_dec(), (2) reduce qlen, and (3) rtnl_qdisc_drop().

But I've also checked sch_codel, its order is: (1) reduce qlen, (2) qdisc_qstats_backlog_dec(), and (3) rtnl_qdisc_drop().

So, the key is to place rtnl_qdisc_drop() after the reduction of qstats_backlog as well as qlen.

Then, I will follow the same order for dualpi2 in next version:
1. qdisc_dequeue_internal(q->l_queue), including (a) --q->l_queue->q.qlen, and (2) qdisc_qstats_backlog_dec(q->l_queue)
2. --sch->q.qlen
3. qdisc_qstats_backlog_dec(sch)
4. rtnl_qdisc_drop(skb, q->l_queue), which will do "qdisc_qstats_drop(q->l_queue)"
5. qdisc_qstats_drop(sch)

Thanks,
Chia-Yu