Cake - FQ_codel the next generation
 help / color / mirror / Atom feed
From: "Toke Høiland-Jørgensen" <toke@toke.dk>
To: Xiang Mei <xmei5@asu.edu>, security@kernel.org
Cc: netdev@vger.kernel.org, xiyou.wangcong@gmail.com,
	cake@lists.bufferbloat.net, bestswngs@gmail.com,
	Xiang Mei <xmei5@asu.edu>
Subject: [Cake] Re: [PATCH net v5] net/sched: sch_cake: Fix incorrect qlen reduction in cake_drop
Date: Mon, 24 Nov 2025 11:48:14 +0100	[thread overview]
Message-ID: <87ms4bn1u9.fsf@toke.dk> (raw)
In-Reply-To: <20251121232735.1020046-1-xmei5@asu.edu>

Xiang Mei <xmei5@asu.edu> writes:

> In cake_drop(), qdisc_tree_reduce_backlog() is used to update the qlen
> and backlog of the qdisc hierarchy. Its caller, cake_enqueue(), assumes
> that the parent qdisc will enqueue the current packet. However, this
> assumption breaks when cake_enqueue() returns NET_XMIT_CN: the parent
> qdisc stops enqueuing current packet, leaving the tree qlen/backlog
> accounting inconsistent. This mismatch can lead to a NULL dereference
> (e.g., when the parent Qdisc is qfq_qdisc).
>
> This patch computes the qlen/backlog delta in a more robust way by
> observing the difference before and after the series of cake_drop()
> calls, and then compensates the qdisc tree accounting if cake_enqueue()
> returns NET_XMIT_CN.
>
> To ensure correct compensation when ACK thinning is enabled, a new
> variable is introduced to keep qlen unchanged.
>
> Fixes: 15de71d06a40 ("net/sched: Make cake_enqueue return NET_XMIT_CN when past buffer_limit")
> Signed-off-by: Xiang Mei <xmei5@asu.edu>
> ---
> v2: add missing cc
> v3: move qdisc_tree_reduce_backlog out of cake_drop
> v4: remove redundant variable and handle ack branch correctly
> v5: add the PoC as a test case

Please split the test case into its own patch and send both as a series.

Otherwise, the changes LGTM apart from the few nits below:

> ---
>  net/sched/sch_cake.c                          | 52 +++++++++++--------
>  .../tc-testing/tc-tests/qdiscs/cake.json      | 28 ++++++++++
>  2 files changed, 58 insertions(+), 22 deletions(-)
>
> diff --git a/net/sched/sch_cake.c b/net/sched/sch_cake.c
> index 32bacfc314c2..cf4d6454ca9c 100644
> --- a/net/sched/sch_cake.c
> +++ b/net/sched/sch_cake.c
> @@ -1597,7 +1597,6 @@ static unsigned int cake_drop(struct Qdisc *sch, struct sk_buff **to_free)
>  
>  	qdisc_drop_reason(skb, sch, to_free, SKB_DROP_REASON_QDISC_OVERLIMIT);
>  	sch->q.qlen--;
> -	qdisc_tree_reduce_backlog(sch, 1, len);
>  
>  	cake_heapify(q, 0);
>  
> @@ -1750,7 +1749,8 @@ static s32 cake_enqueue(struct sk_buff *skb, struct Qdisc *sch,
>  	ktime_t now = ktime_get();
>  	struct cake_tin_data *b;
>  	struct cake_flow *flow;
> -	u32 idx, tin;
> +	u32 idx, tin, prev_qlen, prev_backlog, drop_id;
> +	bool same_flow = false;

Please make sure to maintain the reverse x-mas tree ordering of the
variable declarations.

>  
>  	/* choose flow to insert into */
>  	idx = cake_classify(sch, &b, skb, q->flow_mode, &ret);
> @@ -1823,6 +1823,8 @@ static s32 cake_enqueue(struct sk_buff *skb, struct Qdisc *sch,
>  		consume_skb(skb);
>  	} else {
>  		/* not splitting */
> +		int ack_pkt_len = 0;
> +
>  		cobalt_set_enqueue_time(skb, now);
>  		get_cobalt_cb(skb)->adjusted_len = cake_overhead(q, skb);
>  		flow_queue_add(flow, skb);
> @@ -1834,7 +1836,7 @@ static s32 cake_enqueue(struct sk_buff *skb, struct Qdisc *sch,
>  			b->ack_drops++;
>  			sch->qstats.drops++;
>  			b->bytes += qdisc_pkt_len(ack);
> -			len -= qdisc_pkt_len(ack);
> +			ack_pkt_len = qdisc_pkt_len(ack);

There's a qdisc_tree_reduce_backlog() that uses qdisc_pkt_len(ack) just
below this; let's also change that to use ack_pkt_len while we're at it.

>  			q->buffer_used += skb->truesize - ack->truesize;
>  			if (q->rate_flags & CAKE_FLAG_INGRESS)
>  				cake_advance_shaper(q, b, ack, now, true);
> @@ -1848,11 +1850,11 @@ static s32 cake_enqueue(struct sk_buff *skb, struct Qdisc *sch,
>  
>  		/* stats */
>  		b->packets++;
> -		b->bytes	    += len;
> -		b->backlogs[idx]    += len;
> -		b->tin_backlog      += len;
> -		sch->qstats.backlog += len;
> -		q->avg_window_bytes += len;
> +		b->bytes	    += len - ack_pkt_len;
> +		b->backlogs[idx]    += len - ack_pkt_len;
> +		b->tin_backlog      += len - ack_pkt_len;
> +		sch->qstats.backlog += len - ack_pkt_len;
> +		q->avg_window_bytes += len - ack_pkt_len;
>  	}
>  
>  	if (q->overflow_timeout)
> @@ -1927,24 +1929,30 @@ static s32 cake_enqueue(struct sk_buff *skb, struct Qdisc *sch,
>  	if (q->buffer_used > q->buffer_max_used)
>  		q->buffer_max_used = q->buffer_used;
>  
> -	if (q->buffer_used > q->buffer_limit) {
> -		bool same_flow = false;
> -		u32 dropped = 0;
> -		u32 drop_id;
> +	if (q->buffer_used <= q->buffer_limit)
> +		return NET_XMIT_SUCCESS;
>  
> -		while (q->buffer_used > q->buffer_limit) {
> -			dropped++;
> -			drop_id = cake_drop(sch, to_free);
> +	prev_qlen = sch->q.qlen;
> +	prev_backlog = sch->qstats.backlog;
>  
> -			if ((drop_id >> 16) == tin &&
> -			    (drop_id & 0xFFFF) == idx)
> -				same_flow = true;
> -		}
> -		b->drop_overlimit += dropped;
> +	while (q->buffer_used > q->buffer_limit) {
> +		drop_id = cake_drop(sch, to_free);
> +		if ((drop_id >> 16) == tin &&
> +		    (drop_id & 0xFFFF) == idx)
> +			same_flow = true;
> +	}
> +
> +	/* Compute the droppped qlen and pkt length */
> +	prev_qlen -= sch->q.qlen;
> +	prev_backlog -= sch->qstats.backlog;
> +	b->drop_overlimit += prev_backlog;

drop_overlimit was accounted in packets before, so this should be += prev_qlen.

-Toke

      reply	other threads:[~2025-11-24 10:48 UTC|newest]

Thread overview: 2+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2025-11-21 22:27 [Cake] " Xiang Mei
2025-11-24 10:48 ` Toke Høiland-Jørgensen [this message]

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

  List information: https://lists.bufferbloat.net/postorius/lists/cake.lists.bufferbloat.net/

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=87ms4bn1u9.fsf@toke.dk \
    --to=toke@toke.dk \
    --cc=bestswngs@gmail.com \
    --cc=cake@lists.bufferbloat.net \
    --cc=netdev@vger.kernel.org \
    --cc=security@kernel.org \
    --cc=xiyou.wangcong@gmail.com \
    --cc=xmei5@asu.edu \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox