Skip to contentSkip to conversation

KNI broken again with 4.9 kernel

DISCUSSION Stephen Hemminger 5 participants 7 messages Dec 14, 2016 – Dec 15, 2016

lore
Share & tools
lore
#0 lore raw

/build/lib/librte_eal/linuxapp/kni/igb_main.c:2317:21: error: initialization from incompatible pointer type [-Werror=incompatible-pointer-types]
.ndo_set_vf_vlan = igb_ndo_set_vf_vlan,

                     ^~~~~~~~~~~~~~~~~~~

I am sure Ferruh Yigit will fix it.

Which raises a couple of questions:
1. Why is DPDK still keeping KNI support for Intel specific ethtool functionality. This always breaks, is code bloat, and means a 3rd copy of base code (Linux, DPDK PMD, + KNI)

2. Why is KNI not upstream?
If not acceptable due to security or supportablity then why does it still exist?

3. If not upstream, then maintainer should track upstream kernel changes and fix DPDK before kernel is released. The ABI is normally set early in the rc cycle weeks before release.

Tested-by Pablo de Lara
#1 lore raw

Hi Stephen,

I can't reproduce the error, I think it is not broken, that issue should be already fixed with commit [1].

Quoted 3 lines · Stephen Hemminger

/build/lib/librte_eal/linuxapp/kni/igb_main.c:2317:21: error: initialization from incompatible pointer type [-Werror=incompatible-pointer-types]
.ndo_set_vf_vlan = igb_ndo_set_vf_vlan,

                     ^~~~~~~~~~~~~~~~~~~

Thanks for the report, not everyone is working with latest kernel, so these reports are useful.

Quoted 2 lines

I am sure Ferruh Yigit will fix it.

I think I already did J, please check the commit [1], it is merged in v16.11, can you please double check the DPDK version.

[1]
commit 6445198f802d993c73f4b246353b2ceb2dfafc32
Refs: v16.11-rc1-2-g6445198
Author: Ferruh Yigit <[email protected]>
AuthorDate: Mon Oct 17 11:23:14 2016 +0100
Commit: Thomas Monjalon <[email protected]>
CommitDate: Tue Oct 25 16:20:43 2016 +0200

kni: fix build with kernel 4.9

compile error:
CC [M] .../lib/librte_eal/linuxapp/kni/igb_main.o
.../lib/librte_eal/linuxapp/kni/igb_main.c:2317:21:
error: initialization from incompatible pointer type
[-Werror=incompatible-pointer-types]
.ndo_set_vf_vlan = igb_ndo_set_vf_vlan,
^~~~~~~~~~~~~~~~~~~

Linux kernel 4.9 updates API for ndo_set_vf_vlan:
Linux: 79aab093a0b5 ("net: Update API for VF vlan protocol 802.1ad
support")

Use new API for Linux kernels >= 4.9

<...>
I will reply questions with a different subject...

#2 lore raw

Re: KNI Questions

In reply to Stephen Hemminger

Hi Stephen,

<...>

Quoted 4 lines

Which raises a couple of questions:
1. Why is DPDK still keeping KNI support for Intel specific ethtool functionality. This always breaks, is code bloat, and means a 3rd copy of base code (Linux, DPDK PMD, + KNI)

I agree on you comments related to the ethtool functionality, but right now that is a functionality that people may be using, I think we should not remove it without providing an alternative to it.

Quoted 3 lines

2. Why is KNI not upstream?
If not acceptable due to security or supportablity then why does it still exist?

I believe you are one of the most knowledgeable person in the mail list on upstreaming, any support is welcome.

Quoted 3 lines

3. If not upstream, then maintainer should track upstream kernel changes and fix DPDK before kernel is released. The ABI is normally set early in the rc cycle weeks before release.

I am trying to track as much as possible, any help appreciated.

Quoted 1 line
#3 lore raw

Re: KNI Questions

Quoted 14 lines · On Thu, 15 Dec 2016 11:53:59 +0000 Ferruh Yigit <[email protected]> wrote:

Hi Stephen,

<...>

Quoted 4 lines

Which raises a couple of questions:
1. Why is DPDK still keeping KNI support for Intel specific ethtool functionality. This always breaks, is code bloat, and means a 3rd copy of base code (Linux, DPDK PMD, + KNI)

I agree on you comments related to the ethtool functionality, but right now that is a functionality that people may be using, I think we should not remove it without providing an alternative to it.

Quoted 1 line
Quoted 5 lines
Quoted 2 lines

2. Why is KNI not upstream?
If not acceptable due to security or supportablity then why does it still exist?

I believe you are one of the most knowledgeable person in the mail list on upstreaming, any support is welcome.

It should be upstreamable but I doubt it would make it past the maintainer. Mostly because it supports DPDK which he is not in favor of but also since it is a specialized interface only usable by DPDK, ie. not a general infrastructure.

#4 lore raw

Re: KNI Questions

Quoted 20 lines · Stephen Hemminger
Quoted 17 lines · On Thu, 15 Dec 2016 11:53:59 +0000 Ferruh Yigit <[email protected]> wrote:

Hi Stephen,

<...>

Quoted 4 lines

Which raises a couple of questions:
1. Why is DPDK still keeping KNI support for Intel specific ethtool functionality. This always breaks, is code bloat, and means a 3rd copy of base code (Linux, DPDK PMD, + KNI)

I agree on you comments related to the ethtool functionality, but right now that is a functionality that people may be using, I think we should not remove it without providing an alternative to it.

Quoted 3 lines

2. Why is KNI not upstream?
If not acceptable due to security or supportablity then why does it still exist?

Quoted 7 lines
Quoted 2 lines

I believe you are one of the most knowledgeable person in the mail list on upstreaming, any support is welcome.

It should be upstreamable but I doubt it would make it past the maintainer. Mostly because it supports DPDK which he is not in favor of but also since it is a specialized interface only usable by DPDK, ie. not a general infrastructure.

I was looking to see if we could get more or less the same interface put in either af_packet or vhost directly. It would work nicely with the XDP patches where we want to forward a packet into userspace without having to build skb, etc. So it wouldn't be _just_ a DPDK interface at that point.

I was hoping to look into it in Jan/Feb I need to wrap a few other things up first.

.John

#5 lore raw
In reply to Stephen Hemminger
Quoted 22 lines

-----Original Message-----
From: dev [mailto:[email protected]] On Behalf Of Stephen Hemminger
Sent: Wednesday, December 14, 2016 11:41 PM
To: [email protected]
Subject: [dpdk-dev] KNI broken again with 4.9 kernel

/build/lib/librte_eal/linuxapp/kni/igb_main.c:2317:21: error:
initialization from incompatible pointer type [-Werror=incompatible- pointer-types]
.ndo_set_vf_vlan = igb_ndo_set_vf_vlan,

                     ^~~~~~~~~~~~~~~~~~~

I am sure Ferruh Yigit will fix it.

Which raises a couple of questions:
1. Why is DPDK still keeping KNI support for Intel specific ethtool
functionality.
This always breaks, is code bloat, and means a 3rd copy of base code
(Linux, DPDK PMD, + KNI)

Quoted 8 lines

2. Why is KNI not upstream?
If not acceptable due to security or supportablity then why does it
still exist?

3. If not upstream, then maintainer should track upstream kernel changes
and fix DPDK before
kernel is released. The ABI is normally set early in the rc cycle
weeks before release.

Hi Stephen,

On point 2: The feedback we have always received is that the KNI code isn't upstreamable. Do you think there is an upstream path?

Quoted 2 lines

If not acceptable due to security or supportablity then why does it still exist?

The most commonly expressed reason when we have asked this question in the past (and we did again at Userspace a few months ago) is that the people who use it want the performance.

On point 3: We do have an internal continuous integration system that runs nightly compiles of DPDK against the latest kernel and flags any issues.

John

#6 lore raw
Quoted 36 lines · John
Quoted 21 lines

-----Original Message-----
From: dev [mailto:[email protected]] On Behalf Of Stephen Hemminger
Sent: Wednesday, December 14, 2016 11:41 PM
To: [email protected]
Subject: [dpdk-dev] KNI broken again with 4.9 kernel

/build/lib/librte_eal/linuxapp/kni/igb_main.c:2317:21: error:
initialization from incompatible pointer type [-Werror=incompatible- pointer-types]
.ndo_set_vf_vlan = igb_ndo_set_vf_vlan,

                     ^~~~~~~~~~~~~~~~~~~

I am sure Ferruh Yigit will fix it.

Which raises a couple of questions:
1. Why is DPDK still keeping KNI support for Intel specific ethtool
functionality.
This always breaks, is code bloat, and means a 3rd copy of base code
(Linux, DPDK PMD, + KNI)

Quoted 8 lines

2. Why is KNI not upstream?
If not acceptable due to security or supportablity then why does it
still exist?

3. If not upstream, then maintainer should track upstream kernel changes
and fix DPDK before
kernel is released. The ABI is normally set early in the rc cycle
weeks before release.

Hi Stephen,

On point 2: The feedback we have always received is that the KNI code isn't upstreamable. Do you think there is an upstream path?

Quoted 7 lines
Quoted 2 lines

If not acceptable due to security or supportablity then why does it still exist?

The most commonly expressed reason when we have asked this question in the past (and we did again at Userspace a few months ago) is that the people who use it want the performance.

We use KNI in our product. In our case, it's because it allows "normal" non-DPDK apps in the control plane to interact with traffic on the fastpath as needed. Having everything under the sun live in DPDK's essentially flat memory space is not great for security or stability.

It helps time to market by being able to use existing programs that interface to the network via sockets instead of having to limit ourselves to the relatively tiny set of libraries out there that work directly DPDK.

Double bonus on the time-to-market argument since we can implement functionality in other higher-level languages as appropriate.

Performance-wise, KNI is "ok" but not great. It's not clear to me why it is so much slower than using a NIC normally (not via DPDK) via the Linux network stack. Copying data between sk_buf and mbuf is never going to be cheap, but comparing that to what happens through all the kernel network stack layers, the end result seems too slow.

That said, it's still faster than TAP/TUN interfaces and similar approaches.

Jay

On point 3: We do have an internal continuous integration system that runs

Quoted 4 lines

nightly compiles of DPDK against the latest kernel and flags any issues.

John