* Re: AF_XDP question
[not found] <CAGZFCEFQJSm2K3L-oVKddtvxJf0TiVupudSFzaL_eSCCF5-CQw@mail.gmail.com>
@ 2021-05-03 8:19 ` Magnus Karlsson
0 siblings, 0 replies; only message in thread
From: Magnus Karlsson @ 2021-05-03 8:19 UTC (permalink / raw)
To: ratheesh kannoth; +Cc: Jesper Dangaard Brouer, Xdp
On Thu, Apr 29, 2021 at 4:34 PM ratheesh kannoth <ratheesh.ksz@gmail.com> wrote:
>
> Hi,
>
> Suppose a DPI application running on a AF_XDP socket ( zero copy ) and after inspecting , say 100 initial packets of the flow, decides to accelerate the flow in HW.
Could you please give an example on what you mean with "accelerate the
flow in HW". Many things could be accelerated and in many different
ways.
> 1. Do we have use case and products/solutions in market which does the same ?
>
> Just thinking, since packet buffer is from umem, how another entity (hw) can accelerate.
It is possible to share the umem between AF_XDP sockets using
xsk_socket__create_shared() function in libbpf. It works between
netdevs, i.e. separate physical cards. But need the example first to
be able to help you in a better way.
/Magnus
> Ratheesh.
>
>
>
>
>
>
>
^ permalink raw reply [flat|nested] only message in thread
only message in thread, other threads:[~2021-05-03 8:19 UTC | newest]
Thread overview: (only message) (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
[not found] <CAGZFCEFQJSm2K3L-oVKddtvxJf0TiVupudSFzaL_eSCCF5-CQw@mail.gmail.com>
2021-05-03 8:19 ` AF_XDP question Magnus Karlsson
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).