* Re: eBPF support for TCP payload append feature
[not found] <CABX6iNpgX0EHWNJxb5Vd45Dc6BmS7P7Fjz5o0AyHU3Jq0ORYTw@mail.gmail.com>
@ 2021-07-08 22:06 ` Alexei Starovoitov
0 siblings, 0 replies; only message in thread
From: Alexei Starovoitov @ 2021-07-08 22:06 UTC (permalink / raw)
To: Sandesh.DhawaskarSathyanarayana; +Cc: bpf, Alexei Starovoitov, Daniel Borkmann
On Wed, Jul 7, 2021 at 8:29 AM Sandesh Dhawaskar Sathyanarayana
<Sandesh.DhawaskarSathyanarayana@colorado.edu> wrote:
>
>
> Hence to do such modification for TCP telemetry in eBPF with software switches, we need to add new eBPF support as kernel workaround can create csum issues for TCP.
For TCP telemetry use case a custom TCP hdr option would fit better.
Sender and receiver TCP stacks won't get confused and middle boxes
should handle it fine.
https://lore.kernel.org/bpf/20200820190008.2883500-1-kafai@fb.com/
Once such option (read: space in the packet) is produced at the sender side
the middle box can fill in the extra data running at XDP layer, since space
in the packet is already reserved.
I believe UDP is already used for switch telemetry in such a fashion.
^ permalink raw reply [flat|nested] only message in thread
only message in thread, other threads:[~2021-07-08 22:06 UTC | newest]
Thread overview: (only message) (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
[not found] <CABX6iNpgX0EHWNJxb5Vd45Dc6BmS7P7Fjz5o0AyHU3Jq0ORYTw@mail.gmail.com>
2021-07-08 22:06 ` eBPF support for TCP payload append feature Alexei Starovoitov
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.