Messages in this thread | | | From | Song Liu <> | Date | Tue, 11 Jan 2022 13:47:49 -0800 | Subject | Re: [PATCH bpf-next] tcp: bpf: Add TCP_BPF_RCV_SSTHRESH for bpf_setsockopt |
| |
On Tue, Jan 11, 2022 at 11:29 AM Ivan Babrou <ivan@cloudflare.com> wrote: > > This patch adds bpf_setsockopt(TCP_BPF_RCV_SSTHRESH) to allow setting > rcv_ssthresh value for TCP connections. Combined with increased > window_clamp via tcp_rmem[1], it allows to advertise initial scaled > TCP window larger than 64k. This is useful for high BDP connections, > where it allows to push data with fewer roundtrips, reducing latency. > > For active connections the larger window is advertised in the first > non-SYN ACK packet as the part of the 3 way handshake. > > For passive connections the larger window is advertised whenever > there's any packet to send after the 3 way handshake. > > See: https://lkml.org/lkml/2021/12/22/652
I guess this is [1] mentioned above. Please use lore link instead, e.g.
[1] https://lore.kernel.org/all/CABWYdi0qBQ57OHt4ZbRxMtdSzhubzkPaPKkYzdNfu4+cgPyXCA@mail.gmail.com/
Can we add a selftests for this? Something similar to
tools/testing/selftests/bpf/progs/test_misc_tcp_hdr_options.c
Thanks, Song
[...]
| |