[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <7beeecae-8d6f-d4c2-db43-ddf468469ca0@gmail.com>
Date: Wed, 7 Dec 2022 20:53:51 +0200
From: Tariq Toukan <ttoukan.linux@...il.com>
To: Eric Dumazet <edumazet@...gle.com>,
"David S . Miller" <davem@...emloft.net>,
Jakub Kicinski <kuba@...nel.org>,
Paolo Abeni <pabeni@...hat.com>
Cc: Tariq Toukan <tariqt@...dia.com>,
Leon Romanovsky <leonro@...dia.com>, netdev@...r.kernel.org,
eric.dumazet@...il.com, Wei Wang <weiwan@...gle.com>
Subject: Re: [PATCH v2 net-next 2/3] net/mlx4: MLX4_TX_BOUNCE_BUFFER_SIZE
depends on MAX_SKB_FRAGS
On 12/7/2022 4:12 PM, Eric Dumazet wrote:
> Google production kernel has increased MAX_SKB_FRAGS to 45
> for BIG-TCP rollout.
>
> Unfortunately mlx4 TX bounce buffer is not big enough whenever
> an skb has up to 45 page fragments.
>
> This can happen often with TCP TX zero copy, as one frag usually
> holds 4096 bytes of payload (order-0 page).
>
> Tested:
> Kernel built with MAX_SKB_FRAGS=45
> ip link set dev eth0 gso_max_size 185000
> netperf -t TCP_SENDFILE
>
> I made sure that "ethtool -G eth0 tx 64" was properly working,
> ring->full_size being set to 15.
>
> Signed-off-by: Eric Dumazet <edumazet@...gle.com>
> Reported-by: Wei Wang <weiwan@...gle.com>
> Cc: Tariq Toukan <tariqt@...dia.com>
Reviewed-by: Tariq Toukan <tariqt@...dia.com>
Powered by blists - more mailing lists