Re: [PATCH net-next 2/2] net/sched: fq_codel: local packets no longer count against memory limit

["Toke Høiland-Jørgensen" <toke@toke.dk>]

Eric Dumazet <edumazet@google.com> writes:

> Commit 95b58430abe7 ("fq_codel: add memory limitation per queue")
> claimed that the 32Mb default was "reasonable even for heavy duty usages."
>
> In practice, this is not the case.

Well, the assumption lasted a decade, so that's pretty good? :)

> Packets that are associated with local sockets sk_wmem_alloc
> do not really need additional memory control.
>
> Signed-off-by: Eric Dumazet <edumazet@google.com>
> ---
>  net/sched/sch_fq_codel.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/net/sched/sch_fq_codel.c b/net/sched/sch_fq_codel.c
> index 1b1de693d4c64a1f5f4e9e788371829dea91740e..71107dc52be799a14f370f2ad74d2eadd93992c1 100644
> --- a/net/sched/sch_fq_codel.c
> +++ b/net/sched/sch_fq_codel.c
> @@ -212,7 +212,7 @@ static int fq_codel_enqueue(struct sk_buff *skb, struct Qdisc *sch,
>  		q->new_flow_count++;
>  		WRITE_ONCE(flow->deficit, q->quantum);
>  	}
> -	get_codel_cb(skb)->mem_usage = skb->truesize;
> +	get_codel_cb(skb)->mem_usage = is_skb_wmem(skb) ? 0 : skb->truesize;
>  	q->memory_usage += get_codel_cb(skb)->mem_usage;

Only one concern here: q->memory_usage is exposed to userspace in the
stats, so this will look like the packets queued are zero-length to
anyone watching, which may end up confusing folks? Also, there will be
no way to see how many bytes are actually in the qdisc.

Should we keep a separate counter so we can still accurately report the
memory usage to userspace?

-Toke