netdev - Re: [PATCH net-next v9 2/4] ethtool: provide customized dim profile management

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <6509d9f4-09b1-43ee-ab38-8949f51a806d@linux.alibaba.com>
Date: Mon, 22 Apr 2024 17:04:06 +0800
From: Heng Qi <hengqi@...ux.alibaba.com>
To: Brett Creeley <bcreeley@....com>
Cc: Jakub Kicinski <kuba@...nel.org>, "David S . Miller"
 <davem@...emloft.net>, Eric Dumazet <edumazet@...gle.com>,
 Paolo Abeni <pabeni@...hat.com>, Jason Wang <jasowang@...hat.com>,
 "Michael S . Tsirkin" <mst@...hat.com>,
 Ratheesh Kannoth <rkannoth@...vell.com>,
 Alexander Lobakin <aleksander.lobakin@...el.com>,
 Xuan Zhuo <xuanzhuo@...ux.alibaba.com>, Tal Gilboa <talgi@...dia.com>,
 Jonathan Corbet <corbet@....net>, linux-doc@...r.kernel.org,
 Maxime Chevallier <maxime.chevallier@...tlin.com>,
 Jiri Pirko <jiri@...nulli.us>, Paul Greenwalt <paul.greenwalt@...el.com>,
 Ahmed Zaki <ahmed.zaki@...el.com>, Vladimir Oltean
 <vladimir.oltean@....com>, Kory Maincent <kory.maincent@...tlin.com>,
 Andrew Lunn <andrew@...n.ch>, "justinstitt@...gle.com"
 <justinstitt@...gle.com>,
 "open list:NETWORKING DRIVERS" <netdev@...r.kernel.org>,
 "open list:VIRTIO CORE AND NET DRIVERS" <virtualization@...ts.linux.dev>
Subject: Re: [PATCH net-next v9 2/4] ethtool: provide customized dim profile
 management

在 2024/4/21 下午11:53, Brett Creeley 写道:
> On 4/17/2024 8:55 AM, Heng Qi wrote:
>> Caution: This message originated from an External Source. Use proper 
>> caution when opening attachments, clicking links, or responding.
>>
>>
>> The NetDIM library, currently leveraged by an array of NICs, delivers
>> excellent acceleration benefits. Nevertheless, NICs vary significantly
>> in their dim profile list prerequisites.
>>
>> Specifically, virtio-net backends may present diverse sw or hw device
>> implementation, making a one-size-fits-all parameter list impractical.
>> On Alibaba Cloud, the virtio DPU's performance under the default DIM
>> profile falls short of expectations, partly due to a mismatch in
>> parameter configuration.
>>
>> I also noticed that ice/idpf/ena and other NICs have customized
>> profilelist or placed some restrictions on dim capabilities.
>>
>> Motivated by this, I tried adding new params for "ethtool -C" that 
>> provides
>> a per-device control to modify and access a device's interrupt 
>> parameters.
>>
>> Usage
>> ========
>> The target NIC is named ethx.
>>
>> Assume that ethx only declares support for 
>> ETHTOOL_COALESCE_RX_EQE_PROFILE
>> in ethtool_ops->supported_coalesce_params.
>>
>> 1. Query the currently customized list of the device
>>
>> $ ethtool -c ethx
>> ...
>> rx-eqe-profile:
>> {.usec =   1, .pkts = 256, .comps =   0,},
>> {.usec =   8, .pkts = 256, .comps =   0,},
>> {.usec =  64, .pkts = 256, .comps =   0,},
>> {.usec = 128, .pkts = 256, .comps =   0,},
>> {.usec = 256, .pkts = 256, .comps =   0,}
>> rx-cqe-profile:   n/a
>> tx-eqe-profile:   n/a
>> tx-cqe-profile:   n/a
>>
>> 2. Tune
>> $ ethtool -C ethx rx-eqe-profile 1,1,0_2,2,0_3,3,0_4,4,0_5,5,0
>
> With all of this work to support custom dim profiles (which I think is 
> a great idea FWIW), I wonder if it would be worth supporting a dynamic 
> number of profile entries instead of being hard-coded to 5?
>

In my practice, I have not found that fewer or more profile entries lead 
to better performance.

Thanks.

> Thanks,
>
> Brett
>
> <snip>