[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <CACSApvbAXTN9VLcB0BphbJffZ8YXpxm=JF8FteKBVkBArwcd3A@mail.gmail.com>
Date: Sun, 13 Dec 2020 21:22:22 -0500
From: Soheil Hassas Yeganeh <soheil@...gle.com>
To: Victor Stewart <v@...etag.social>
Cc: io-uring <io-uring@...r.kernel.org>,
netdev <netdev@...r.kernel.org>, Jann Horn <jannh@...gle.com>
Subject: Re: [PATCH v3] Allow UDP cmsghdrs through io_uring
On Sun, Dec 13, 2020 at 3:43 PM Victor Stewart <v@...etag.social> wrote:
>
> here we go, figured it out. sorry about that.
>
> This patch adds PROTO_CMSG_DATA_ONLY to inet_dgram_ops and inet6_dgram_ops so that UDP_SEGMENT (GSO) and UDP_GRO can be used through io_uring.
The v3 patch doesn't have "Signed-off-by". You can see all the checks
on patchwork:
https://patchwork.kernel.org/project/netdevbpf/patch/20201213204339.24445-2-v@nametag.social/
For patches to netdev, please include the target. Here it should be net-next:
https://www.kernel.org/doc/Documentation/networking/netdev-FAQ.txt
git format-patch --subject-prefix='PATCH net-next v4' ...
Also, since this patch affects udp, it's preferred to include "udp: "
in your commit's subject line. In other words, the v4 of this patch
should look like the following:
[PATCH net-next v2] udp: allow UDP cmsghdrs through io_uring
Thank you
> GSO and GRO are vital to bring QUIC servers on par with TCP throughputs, and together offer a higher
> throughput gain than io_uring alone (rate of data transit
> considering), thus io_uring is presently the lesser performance choice.
>
> RE http://vger.kernel.org/lpc_net2018_talks/willemdebruijn-lpc2018-udpgso-paper-DRAFT-1.pdf,
> GSO is about +~63% and GRO +~82%.
>
> this patch closes that loophole.
>
> net/ipv4/af_inet.c | 1 +
> net/ipv6/af_inet6.c | 1 +
> net/socket.c | 8 +++++---
> 3 files changed, 7 insertions(+), 3 deletions(-)
>
>
Powered by blists - more mailing lists