Re: [Qemu-devel] Using virtio-net and vhost_net on an ARM machine using qemu-system-arm & KVM

[Barak Wasserstrom]

Ying-Shiuan Pan,

Thanks again - please see few questions below.

Regards,

Barak

On Tue, Jan 14, 2014 at 5:37 AM, Ying-Shiuan Pan <address@hidden> wrote:

Hi, Barak,

Hope the following info can help you

1.

HOST:

http://git.linaro.org/people/christoffer.dall/linux-kvm-arm.git

branch: v3.10-arndale

config: arch/arm/configs/exynos5_arndale_defconfig
dtb: arch/arm/boot/dts/exynos5250-arndale.dtb

rootfs: Ubuntu 13.10

GUEST:

Official 3.12

config: arch/arm/configs/vexpress_defconfig  with virtio-devices enabled
dtb: arch/arm/boot/dts/vexpress-v2p-ca15-tc1.dtb

rootfs: Ubuntu 12.04

2.

We are still developing it in progress and will try to open source asap.

The main purpose of that patch is to introduce the ioeventfd into kvm-arm

[Barak] Do you have any estimation about when you can release these patches? 

[Barak] Is this required for enabling vhost-net?

 

3. as mentioned in 1.

4. qemu-1.6.0

5. We ported part of guest/host notifiers of virtio-pci to virtio-mmio

[Barak] Any patches available for this?

[Barak] Is this required for enabling vhost-net?

 

6. /usr/bin/qemu-system-arm -enable-kvm -kernel /root/nfs/zImage -m 128 --machine vexpress-a15 -cpu cortex-a15 -drive file=/root/nfs/guest-1G-precise-vm1.img,id=virtio-blk,if=none,cache=none -device virtio-blk-device,drive=virtio-blk -append "earlyprintk=ttyAMA0 console=ttyAMA0 root=/dev/vda rw ip=192.168.101.101::192.168.101.1:vm1:eth0:off --no-log" -dtb /root/nfs/vexpress-v2p-ca15-tc1.dtb --nographic -chardev socket,id=mon,path=/root/vm1.monitor,server,nowait -mon chardev=mon,id=monitor,mode=readline -device virtio-net-device,netdev=net0,mac="52:54:00:12:34:01" -netdev type=tap,id=net0,script=/root/nfs/net.sh,downscript=no,vhost=off

[Barak] Could you share "/root/nfs/net.sh" with me?

[Barak] In the guest i can see that eth0 has all offload features disabled and cannot be enabled. I suspect this is related to the tap configuration in the host. Do you have any ideas?

 

vhost-net could be truned on by changing the last parameter vhost=on.

[Barak] When enabling vhost i get errors in qemu, do you know what might be the reason?

[Barak] qemu-system-arm: binding does not support guest notifiers

[Barak] qemu-system-arm: unable to start vhost net: 38: falling back on userspace virtio

 

--
Ying-Shiuan Pan,
H Div., CCMA, ITRI, TW

----
Best Regards,
潘穎軒Ying-Shiuan Pan

2014/1/13 Barak Wasserstrom <address@hidden>

Ying-Shiuan Pan,

Your experiments with arndale Exynos-5250 board can help me greatly and i would really appreciate if you share with me the following information:
1. Which Linux kernel did you use for the host and for the guest?

2. Which Linux kernel patches did you use for KVM?

3. Which config files did you use for both the host and guest?

4. Which QEMU did you use?

5. Which QEMU patches did you use?

6. What is the exact command line you used for invoking the guest, with and without vhost-net?

Many thanks in advance!

Regards,

Barak

On Mon, Jan 13, 2014 at 5:47 AM, Ying-Shiuan Pan <address@hidden> wrote:

Hi, Barak,

We've tried vhost-net in kvm-arm on arndale Exynos-5250 board (it requires some patches in qemu and kvm, of course). It works (without irqfd support), however, the performance does not increase much. The throughput (iperf) of virtio-net and vhost-net are 93.5Mbps and 93.6Mbps respectively. I thought the result are because both virtio-net and vhost-net almost reached the limitation of 100Mbps Ethernet.

The good news is that we even ported vhost-net in our kvm-a9 hypervisor (refer: http://academic.odysci.com/article/1010113020064758/evaluation-of-a-server-grade-software-only-arm-hypervisor), and the throughput of vhost-net on that platform (with 1Gbps Ethernet) increased from 323Mbps to 435Mbps.

--
Ying-Shiuan Pan,
H Div., CCMA, ITRI, TW

----
Best Regards,
潘穎軒Ying-Shiuan Pan

2014/1/13 Peter Maydell <address@hidden>

On 12 January 2014 21:49, Barak Wasserstrom <address@hidden> wrote:
> Thanks - I got virtio-net-device running now, but performance is terrible.
> When i look at the guest's ethernet interface features (ethtool -k eth0) i
> see all offload features are disabled.
> I'm using a virtual tap on the host (tap0 bridged to eth3).
> On the tap i also see all offload features are disabled, while on br0 and
> eth3 i see the expected offload features.
> Can this explain the terrible performance i'm facing?
> If so, how can this be changed?
> If not, what else can cause such bad performance?
> Do you know if vhost_net can be used on ARM Cortex A15 host/guest, even
> though the guest doesn't support PCI & MSIX?

I have no idea, I'm afraid. I don't have enough time available to
investigate performance issues at the moment; if you find anything
specific you can submit patches...

thanks
-- PMM