[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <73b28f6e-e6de-41a8-8622-3d58832e1582@kernel.org>
Date: Tue, 5 Dec 2023 09:08:05 -0700
From: David Ahern <dsahern@...nel.org>
To: Coco Li <lixiaoyan@...gle.com>, Jakub Kicinski <kuba@...nel.org>,
Eric Dumazet <edumazet@...gle.com>, Neal Cardwell <ncardwell@...gle.com>,
Mubashir Adnan Qureshi <mubashirq@...gle.com>,
Paolo Abeni <pabeni@...hat.com>, Andrew Lunn <andrew@...n.ch>,
Jonathan Corbet <corbet@....net>, Daniel Borkmann <daniel@...earbox.net>
Cc: netdev@...r.kernel.org, Chao Wu <wwchao@...gle.com>,
Wei Wang <weiwan@...gle.com>, Pradeep Nemavat <pnemavat@...gle.com>
Subject: Re: [PATCH v1 net-next 2/2] tcp: reorganize tcp_sock fast path
variables
On 12/4/23 1:12 PM, Coco Li wrote:
> The variables are organized according in the following way:
>
> - TX read-mostly hotpath cache lines
> - TXRX read-mostly hotpath cache lines
> - RX read-mostly hotpath cache lines
> - TX read-write hotpath cache line
> - TXRX read-write hotpath cache line
> - RX read-write hotpath cache line
>
> Fastpath cachelines end after rcvq_space.
>
> Cache line boundaries are enforced only between read-mostly and
> read-write. That is, if read-mostly tx cachelines bleed into
> read-mostly txrx cachelines, we do not care. We care about the
> boundaries between read and write cachelines because we want
> to prevent false sharing.
>
> Fast path variables span cache lines before change: 12
> Fast path variables span cache lines after change: 8
>
> Suggested-by: Eric Dumazet <edumazet@...gle.com>
> Reviewed-by: Wei Wang <weiwan@...gle.com>
> Signed-off-by: Coco Li <lixiaoyan@...gle.com>
> ---
> include/linux/tcp.h | 248 ++++++++++++++++++++++++--------------------
> net/ipv4/tcp.c | 93 +++++++++++++++++
> 2 files changed, 227 insertions(+), 114 deletions(-)
>
Reviewed-by: David Ahern <dsahern@...nel.org>
Powered by blists - more mailing lists