Hi all,
I have been running 4.18-rc8 kernel with enabled IP forwarding between
wired and wireless interfaces, where both interfaces
were configured as fq qdisc.
However after moving to 4.20-rc1 kernel the same configuration does not
work anymore: pass-through packets are not forwarded in both directions.
Forwarding starts working again only if I change qdisc of _both_ interfaces
to anything but fq/mq. For instance any combination of pfifo/fq_codel/noqueue
works fine.
Does it look like a regression or it is a known change in behavior ?
Regards,
Sergey
On 12/14/2018 04:02 AM, Sergey Matyukevich wrote:
> Hi all,
>
> I have been running 4.18-rc8 kernel with enabled IP forwarding between
> wired and wireless interfaces, where both interfaces
> were configured as fq qdisc.
>
> However after moving to 4.20-rc1 kernel the same configuration does not
> work anymore: pass-through packets are not forwarded in both directions.
> Forwarding starts working again only if I change qdisc of _both_ interfaces
> to anything but fq/mq. For instance any combination of pfifo/fq_codel/noqueue
> works fine.
>
> Does it look like a regression or it is a known change in behavior ?
>
> Regards,
> Sergey
>
Hi Sergey
I guess EDT model broke this use case.
I was under the impression skb->tstamp was cleared when forwarding packets, maybe I was wrong.
Can you try the following ?
diff --git a/net/ipv4/ip_forward.c b/net/ipv4/ip_forward.c
index 06ee4696703c0ce72ea914403b739839e60f1584..00ec819f949b5e76ea96be901a697f4e12d5cf4d 100644
--- a/net/ipv4/ip_forward.c
+++ b/net/ipv4/ip_forward.c
@@ -79,6 +79,7 @@ static int ip_forward_finish(struct net *net, struct sock *sk, struct sk_buff *s
if (unlikely(opt->optlen))
ip_forward_options(skb);
+ skb->tstamp = 0;
return dst_output(net, sk, skb);
}
> > Hi all,
> >
> > I have been running 4.18-rc8 kernel with enabled IP forwarding between
> > wired and wireless interfaces, where both interfaces
> > were configured as fq qdisc.
> >
> > However after moving to 4.20-rc1 kernel the same configuration does not
> > work anymore: pass-through packets are not forwarded in both directions.
> > Forwarding starts working again only if I change qdisc of _both_ interfaces
> > to anything but fq/mq. For instance any combination of pfifo/fq_codel/noqueue
> > works fine.
> >
> > Does it look like a regression or it is a known change in behavior ?
> >
> > Regards,
> > Sergey
> >
>
> Hi Sergey
>
> I guess EDT model broke this use case.
>
> I was under the impression skb->tstamp was cleared when forwarding packets, maybe I was wrong.
>
> Can you try the following ?
>
> diff --git a/net/ipv4/ip_forward.c b/net/ipv4/ip_forward.c
> index 06ee4696703c0ce72ea914403b739839e60f1584..00ec819f949b5e76ea96be901a697f4e12d5cf4d 100644
> --- a/net/ipv4/ip_forward.c
> +++ b/net/ipv4/ip_forward.c
> @@ -79,6 +79,7 @@ static int ip_forward_finish(struct net *net, struct sock *sk, struct sk_buff *s
> if (unlikely(opt->optlen))
> ip_forward_options(skb);
>
> + skb->tstamp = 0;
> return dst_output(net, sk, skb);
> }
Hi Eric,
This patch fixes the issue: ip forwarding works for fq/mq qdisc.
Thanks,
Sergey
On 12/14/2018 06:12 AM, Sergey Matyukevich wrote:
> Hi Eric,
>
> This patch fixes the issue: ip forwarding works for fq/mq qdisc.
>
Thanks for testing, I will send a formal patch for IPv4 and IPv6.