From: Brett Creeley <bcreeley@amd.com>
To: Heng Qi <hengqi@linux.alibaba.com>,
netdev@vger.kernel.org, virtualization@lists.linux.dev
Cc: Jakub Kicinski <kuba@kernel.org>,
"David S. Miller" <davem@davemloft.net>,
Eric Dumazet <edumazet@google.com>,
Paolo Abeni <pabeni@redhat.com>, Jason Wang <jasowang@redhat.com>,
"Michael S. Tsirkin" <mst@redhat.com>,
Ratheesh Kannoth <rkannoth@marvell.com>,
Alexander Lobakin <aleksander.lobakin@intel.com>,
Xuan Zhuo <xuanzhuo@linux.alibaba.com>
Subject: Re: [PATCH net-next v6 2/4] ethtool: provide customized dim profile management
Date: Fri, 12 Apr 2024 08:33:56 -0700 [thread overview]
Message-ID: <c62052c7-a67c-40de-b100-e4469f219383@amd.com> (raw)
In-Reply-To: <99bc7a22-6fa4-49e1-b98b-4b6a46f1b9b8@linux.alibaba.com>
On 4/11/2024 7:07 PM, Heng Qi wrote:
> Caution: This message originated from an External Source. Use proper
> caution when opening attachments, clicking links, or responding.
>
>
> 在 2024/4/11 下午11:19, Brett Creeley 写道:
>>
>>
>> On 4/11/2024 7:12 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
>>> ========
>>> 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
>>> $ ethtool -c ethx
>>> ...
>>> rx-eqe-profile:
>>> {.usec = 1, .pkts = 1, .comps = 0,},
>>> {.usec = 2, .pkts = 2, .comps = 0,},
>>> {.usec = 3, .pkts = 3, .comps = 0,},
>>> {.usec = 4, .pkts = 4, .comps = 0,},
>>> {.usec = 5, .pkts = 5, .comps = 0,}
>>> rx-cqe-profile: n/a
>>> tx-eqe-profile: n/a
>>> tx-cqe-profile: n/a
>>>
>>> 3. Hint
>>> If the device does not support some type of customized dim
>>> profiles, the corresponding "n/a" will display.
>>
>> What if the user specifies a *-eqe-profile and *-cqe-profile for rx
>> and/or tx? Is that supported? If so, which one is the active profile?
>
>
> I think you mean GET? GET currently does not support any parameters, the
> working profile will be displayed.
Yeah, I meant the GET operation. As long as the currently in-use/working
profile is displayed that makes sense.
>
>>
>> Maybe I missed this, but it doesn't seem like the output from "ethtool
>> -c ethX" shows the active profile it just dumps the profile
>> configurations.
>
> Now it is required that dev->priv_flags is set to one of
> IFF_PROFILE_{USEC, PKTS, COMPS} (which means that the
> driver supports configurable profiles) before the profile can be queried
> or do you want to query without this restriction?
No, just as I mentioned above it seems okay.
>
> Thanks!
>
>>
>> Thanks,
>>
>> Brett
>>
>> [snip]
>
next prev parent reply other threads:[~2024-04-12 15:34 UTC|newest]
Thread overview: 17+ messages / expand[flat|nested] mbox.gz Atom feed top
2024-04-11 14:12 [PATCH net-next v6 0/4] ethtool: provide the dim profile fine-tuning channel Heng Qi
2024-04-11 14:12 ` [PATCH net-next v6 1/4] linux/dim: move useful macros to .h file Heng Qi
2024-04-11 14:12 ` [PATCH net-next v6 2/4] ethtool: provide customized dim profile management Heng Qi
2024-04-11 15:19 ` Brett Creeley
2024-04-12 2:07 ` Heng Qi
2024-04-12 15:33 ` Brett Creeley [this message]
2024-04-12 16:05 ` Heng Qi
2024-04-12 3:56 ` kernel test robot
2024-04-12 4:39 ` kernel test robot
2024-04-12 17:44 ` kernel test robot
2024-04-13 2:26 ` Jakub Kicinski
2024-04-13 10:14 ` Heng Qi
2024-04-13 2:27 ` Jakub Kicinski
2024-04-11 14:12 ` [PATCH net-next v6 3/4] virtio-net: refactor dim initialization/destruction Heng Qi
2024-04-11 14:12 ` [PATCH net-next v6 4/4] virtio-net: support dim profile fine-tuning Heng Qi
2024-04-11 15:23 ` Brett Creeley
2024-04-12 2:19 ` Heng Qi
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=c62052c7-a67c-40de-b100-e4469f219383@amd.com \
--to=bcreeley@amd.com \
--cc=aleksander.lobakin@intel.com \
--cc=davem@davemloft.net \
--cc=edumazet@google.com \
--cc=hengqi@linux.alibaba.com \
--cc=jasowang@redhat.com \
--cc=kuba@kernel.org \
--cc=mst@redhat.com \
--cc=netdev@vger.kernel.org \
--cc=pabeni@redhat.com \
--cc=rkannoth@marvell.com \
--cc=virtualization@lists.linux.dev \
--cc=xuanzhuo@linux.alibaba.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).