net/sched: Add precise drop reason for pfifo_fast queue overflows

[PATCH net-next] net/sched: Add precise drop reason for pfifo_fast queue overflows

Posted by fan.yu9@zte.com.cn 2 months, 1 week ago

From: Fan Yu <fan.yu9@zte.com.cn>

Currently, packets dropped by pfifo_fast due to queue overflow are
marked with a generic SKB_DROP_REASON_QDISC_DROP in __dev_xmit_skb().

This patch adds explicit drop reason SKB_DROP_REASON_QDISC_OVERLIMIT
for queue-full cases, providing better distinction from other qdisc drops.

Signed-off-by: Fan Yu <fan.yu9@zte.com.cn>
---
 net/sched/sch_generic.c | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/net/sched/sch_generic.c b/net/sched/sch_generic.c
index 16afb834fe4a..1e008a228ebd 100644
--- a/net/sched/sch_generic.c
+++ b/net/sched/sch_generic.c
@@ -740,6 +740,8 @@ static int pfifo_fast_enqueue(struct sk_buff *skb, struct Qdisc *qdisc,
 	err = skb_array_produce(q, skb);

 	if (unlikely(err)) {
+		tcf_set_drop_reason(skb, SKB_DROP_REASON_QDISC_OVERLIMIT);
+
 		if (qdisc_is_percpu_stats(qdisc))
 			return qdisc_drop_cpu(skb, qdisc, to_free);
 		else
-- 
2.25.1

Re: [PATCH net-next] net/sched: Add precise drop reason for pfifo_fast queue overflows

Posted by Cong Wang 2 months, 1 week ago

On Thu, Jul 24, 2025 at 09:28:37PM +0800, fan.yu9@zte.com.cn wrote:
> From: Fan Yu <fan.yu9@zte.com.cn>
> 
> Currently, packets dropped by pfifo_fast due to queue overflow are
> marked with a generic SKB_DROP_REASON_QDISC_DROP in __dev_xmit_skb().
> 
> This patch adds explicit drop reason SKB_DROP_REASON_QDISC_OVERLIMIT
> for queue-full cases, providing better distinction from other qdisc drops.
> 
> Signed-off-by: Fan Yu <fan.yu9@zte.com.cn>

Reviewed-by: Cong Wang <xiyou.wangcong@gmail.com>

BTW, it seems net-next is closed, you may need to resend it after it is
re-open.

Thanks.

Re: [PATCH net-next] net/sched: Add precise drop reason for pfifo_fast queue overflows

Posted by fan.yu9@zte.com.cn 2 months, 1 week ago

> BTW, it seems net-next is closed, you may need to resend it after it is> re-open.>> Thanks.

Hi Cong,  
Thank you for your review and the feedback!
Could you kindly share how to track the status of the net-next merge window?
Should I monitor the linux-netdev mailing list for announcements, or is there a specific schedule I can follow? 

Best Regards,  
Fan Yu

Re: [PATCH net-next] net/sched: Add precise drop reason for pfifo_fast queue overflows

Posted by Simon Horman 2 months, 1 week ago

On Mon, Jul 28, 2025 at 05:11:37PM +0800, fan.yu9@zte.com.cn wrote:
> > BTW, it seems net-next is closed, you may need to resend it after it is> re-open.>> Thanks.
> 
> Hi Cong,  
> Thank you for your review and the feedback!
> Could you kindly share how to track the status of the net-next merge window?
> Should I monitor the linux-netdev mailing list for announcements, or is there a specific schedule I can follow? 

Hi,

The merge-window opens when a new version of the Kernel is released, in
this case v6.16 was released yesterday. It remains open until the
subsequent rc1 release is made, in this case that will be v6.17-rc1. That
is typically two weeks after the merge-window opens.

You can observe the current Kernel release in a variety of ways,
including visiting https://www.kernel.org/

The timing of the merge window can be predicted with some accuracy by
following kernel rc releases: it usually occurs 1 week after an rc7
release. (If not, an rc8 release occurs, and the merge window will likely
open a week after that; and so on.)

net-next closes around the time that the merge-window opens.
But there is some variance in exactly when this occurs, due
to the schedules of the maintainers.

net-next re-opens around the time that the merge-window closes.
But again there is some variance.

The opening and closing of net-next is announced, with [ANN] in
the subject on the netdev mailing list.

https://lore.kernel.org/netdev/?q=ANN

There is some more information on this topic here:
https://docs.kernel.org/process/maintainer-netdev.html#development-cycle