DPDK patches and discussions
 help / color / mirror / Atom feed
From: Levend Sayar <levendsayar@gmail.com>
To: "Guo, Junfeng" <junfeng.guo@intel.com>
Cc: Ferruh Yigit <ferruh.yigit@amd.com>,
	dev@dpdk.org, Jeroen de Borst <jeroendb@google.com>,
	Rushil Gupta <rushilg@google.com>,
	Jordan Kimbrough <jrkim@google.com>
Subject: Re: Google Virtual NIC (GVE) PMD
Date: Tue, 31 Jan 2023 11:17:59 +0300	[thread overview]
Message-ID: <9B795DFE-EE56-4F51-9896-8DEFDB63E85B@gmail.com> (raw)
In-Reply-To: <DM6PR11MB37231B91661145BCD3B4721DE7D09@DM6PR11MB3723.namprd11.prod.outlook.com>

Thanks for the reply Junfeng.

I will check the document you mentioned.
I don't know the command 'cksum set ip hw 0', but it looks like to me as "hardware will NOT do checksum'.

There is a gateway between VMs at GCP as you mentioned. My test application generates UDP packets.
I am checking the offloading capacity of the NIC and behaving accordingly. Since GVE says IP checksum capability, I am offloading that.
My packets have 0 as an IP checksum. In this case, I can not see any packet at the destination with tcpdump.
Since I do not have a chance to see what is happening at the gateway, I only check packet transmission at the destination VM.

If I calculate IP checksums by myself, I see the packets at the destination. So according to my observations, the gateway does not do such an IP checksum correction.

My second observation is GVE does UDP checksum if you offload. Normally I am not using a checksum at UDP and use zero as the checksum.

Best,
Levend





Telefonumdan gönderildi

> Guo, Junfeng <junfeng.guo@intel.com> şunları yazdı (31 Oca 2023 09:52):
> 
> 
> 
>> -----Original Message-----
>> From: Ferruh Yigit <ferruh.yigit@amd.com>
>> Sent: Wednesday, January 18, 2023 23:32
>> To: Levend Sayar <levendsayar@gmail.com>; Guo, Junfeng
>> <junfeng.guo@intel.com>
>> Cc: dev@dpdk.org; Jeroen de Borst <jeroendb@google.com>; Rushil
>> Gupta <rushilg@google.com>; Jordan Kimbrough <jrkim@google.com>
>> Subject: Re: Google Virtual NIC (GVE) PMD
>> 
>>> On 1/18/2023 1:47 PM, Levend Sayar wrote:
>>> Hi all.
>>> 
>>> PMD for Google Virtual NIC says it is capable of IPV4 TX checksum
>>> offloading.
>>> 
>>> 
>> https://github.com/DPDK/dpdk/blob/main/drivers/net/gve/gve_ethdev.c
>> #L285
>>> 
>> <https://github.com/DPDK/dpdk/blob/main/drivers/net/gve/gve_ethdev.
>> c#L285>
>>> 
>>> But according to my tests on Google Cloud, it is not doing that ipv4
>>> checksum tx offload.
>>> I only managed to send a packet via DPDK if I calculate the checksum
>> myself.
>>> 
>>> Do you have any idea about that?
> 
> Thanks for the feedback!
> 
> I tried to use this loopback typo in csum fwd mode based on this doc:
> "https://doc.dpdk.org/dts/test_plans/checksum_offload_test_plan.html"
> During the dpdk ports forwarding, the received pkts at tester side can have
> correct csum after setting with 'csum set ip hw 0'.
> 
> Also, on GCP, the gateway between VMs will correct the wrong L3 chksum
> values by default when passing the pkts.
> 
> Could you please provide more details for your test cases? 
> So that we can have more information to verify this. Thanks!
> 
> 
>>> 
>> 
>> cc'ed gve maintainers.
> 
> Thanks Ferruh for the forwarding!
> 

      reply	other threads:[~2023-01-31  8:18 UTC|newest]

Thread overview: 4+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2023-01-18 13:47 Levend Sayar
2023-01-18 15:31 ` Ferruh Yigit
2023-01-31  6:52   ` Guo, Junfeng
2023-01-31  8:17     ` Levend Sayar [this message]

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=9B795DFE-EE56-4F51-9896-8DEFDB63E85B@gmail.com \
    --to=levendsayar@gmail.com \
    --cc=dev@dpdk.org \
    --cc=ferruh.yigit@amd.com \
    --cc=jeroendb@google.com \
    --cc=jrkim@google.com \
    --cc=junfeng.guo@intel.com \
    --cc=rushilg@google.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).