Received-SPF: pass (google.com: domain of linux-kernel+bounces-142777-linux.lists.archive=gmail.com@vger.kernel.org designates 2604:1380:45d1:ec00::1 as permitted sender) client-ip=2604:1380:45d1:ec00::1;
Message-ID: <16790d82-e134-4a3b-b67e-d56c041362c0@baylibre.com>
Date: Fri, 12 Apr 2024 15:56:39 +0200
Precedence: bulk
MIME-Version: 1.0
User-Agent: Mozilla Thunderbird
Subject: Re: [PATCH net-next v8 3/3] net: ethernet: ti: am65-cpsw: Add minimal
 XDP support
To: Jakub Kicinski <kuba@kernel.org>
Cc: "David S. Miller" <davem@davemloft.net>,
 Eric Dumazet <edumazet@google.com>, Paolo Abeni <pabeni@redhat.com>,
 Russell King <linux@armlinux.org.uk>, Alexei Starovoitov <ast@kernel.org>,
 Daniel Borkmann <daniel@iogearbox.net>,
 Jesper Dangaard Brouer <hawk@kernel.org>,
 John Fastabend <john.fastabend@gmail.com>,
 Sumit Semwal <sumit.semwal@linaro.org>,
 =?UTF-8?Q?Christian_K=C3=B6nig?= <christian.koenig@amd.com>,
 Simon Horman <horms@kernel.org>, Andrew Lunn <andrew@lunn.ch>,
 Ratheesh Kannoth <rkannoth@marvell.com>,
 Naveen Mamindlapalli <naveenm@marvell.com>, danishanwar@ti.com,
 yuehaibing@huawei.com, rogerq@kernel.org, netdev@vger.kernel.org,
 linux-kernel@vger.kernel.org, bpf@vger.kernel.org,
 linux-media@vger.kernel.org, dri-devel@lists.freedesktop.org,
 linaro-mm-sig@lists.linaro.org
References: <20240223-am65-cpsw-xdp-basic-v8-0-f3421b58da09@baylibre.com>
 <20240223-am65-cpsw-xdp-basic-v8-3-f3421b58da09@baylibre.com>
 <20240409174928.58a7c3f0@kernel.org>
 <9cb903df-3e83-409a-aa4b-218742804cc6@baylibre.com>
 <20240410080225.2e024f7c@kernel.org>
Content-Language: en-US
From: Julien Panis <jpanis@baylibre.com>
In-Reply-To: <20240410080225.2e024f7c@kernel.org>
Content-Type: text/plain; charset=UTF-8; format=flowed
Content-Transfer-Encoding: 7bit

On 4/10/24 17:02, Jakub Kicinski wrote:
> On Wed, 10 Apr 2024 16:02:00 +0200 Julien Panis wrote:
>>> You shouldn't build the skb upfront any more. Give the page to the HW,
>>> once HW sends you a completion - build the skbs. If build fails
>>> (allocation failure) just give the page back to HW. If it succeeds,
>>> however, you'll get a skb which is far more likely to be cache hot.
>> Not sure I get this point.
>>
>> "Give the page to the HW" = Do you mean that I should dma_map_single()
>> a full page (|PAGE_SIZE|) in am65_cpsw_nuss_rx_push() ?
> Yes, I think so. I think that's what you effectively do now anyway,
> you just limit the len and wrap it in an skb. But
> am65_cpsw_nuss_rx_push() will effectively get that page back from
> skb->data and map it.

That's much better indeed, with the implementation you suggest: 600 Mbits/sec
instead of 500. I did not expect such improvement.
I'll send a new version when I finish retesting.