Re: [PATCH v4 bpf-next 00/10] bpf: Add bpf_tcp_ops hooks for TCP AutoLOWAT.

[Jakub Sitnicki <jakub@cloudflare.com>]

On Tue, Oct 06, 2026 at 07:24 PM GMT, Kuniyuki Iwashima wrote:
> This series introduces two callbacks for bpf_tcp_ops:
>
>   .enqueue_rcvq(): invoked when TCP stack enqueues skb to
>                    sk->sk_receive_queue
>
>   .dequeue_rcvq(): invoked in tcp_cleanup_rbuf() after data
>                    is dequeued from sk->sk_receive_queue
>
> Those callbacks can be enabled on a per-socket basis by
> a new kfunc bpf_tcp_ops_set_flags():
>
>   bpf_tcp_ops_set_flags((struct tcp_sock *)sk,
>                         BPF_TCP_OPS_FLAG_RCVQ, 0);
>
> This allows the BPF prog to dynamically adjust sk->sk_rcvlowat,
> suppressing unnecessary EPOLLIN wakeups until sufficient data
> is available in the receive queue.
>
> This functionality, which we call "TCP AutoLOWAT", was originally
> developed in 2020 by Tenzin Ukyab with the help of Soheil Hassas
> Yeganeh, Arjun Roy, and Eric Dumazet.  It has served Google RPC
> workloads for more than 5 years.
>
> Combined with TCP RX zerocopy, this typically allows us to read an
> entire RPC frame with just a single wakeup and a single system call.

Nice! Just yesterday we saw at LPC that even with SK_SKB/SOCKMAP
overhead, you can observe an RPS gain from waking up once there is a
full L7 request to read out [1].

[1] slide 72, https://lpc.events/event/20/contributions/2428/attachments/2064/4781/Beeper.pdf

[...]