Re: [libvirt-users] Sr-iov passthrough - no packet arrive to guest

Wednesday, 28 January 2015

I read a bit more about the card that i am using  and

The problem was that it filtered packet because it a trace that i replay
and not real traffic

So i changed the destination mac adress in the pcap to be the same as vm
and all packet got into the vm.

________________________________

From: libvirt-users-bounces(a)redhat.com
[mailto:libvirt-users-bounces@redhat.com] On Behalf Of Yaron Illouz
Sent: Wednesday, January 28, 2015 6:06 PM
To: libvirt-users(a)redhat.com
Subject: Re: [libvirt-users] Sr-iov passthrough - no packet arrive to
guest

I can see from different post that if working with sr-iov, i should work
with vlan

Is this an obligation to work with vlan if working with sr-iov?

If not according to which parameter will the different vf get the
traffic.

Let's say i declare max_vfs=7, how will the traffic be seperated between
the vm?

However till i get an answer i tried to work with vlan

And i still don't get traffic in vm, but i see packet arriving  in the
physical interface of the host

I tried to do the following to define the vlan

In virsh edit

<interface type='hostdev' managed='yes'>

      <mac address='52:54:00:60:7f:57'/>

      <source>

        <address type='pci' domain='0x0000' bus='0x15'
slot='0x10'
function='0x0'/>

      </source>

      <vlan>

        <tag id='44'/>

      </vlan>

      <alias name='hostdev0'/>

      <address type='pci' domain='0x0000' bus='0x00'
slot='0x07'
function='0x0'/>

    </interface>

In host ip link

23: eth6: <BROADCAST,MULTICAST,PROMISC,UP,LOWER_UP> mtu 1500 qdisc mq
state UP qlen 1000

    link/ether 38:ea:a7:8f:fa:98 brd ff:ff:ff:ff:ff:ff

    vf 0 MAC 52:54:00:60:7f:57, vlan 44

In guest

modprob 8021q

eth1      Link encap:Ethernet  HWaddr 52:54:00:60:7F:57

            inet6 addr: fe80::5054:ff:fe60:7f57/64 Scope:Link

            UP BROADCAST RUNNING PROMISC MULTICAST  MTU:1500  Metric:1

            RX packets:0 errors:0 dropped:0 overruns:0 frame:0

            TX packets:14 errors:0 dropped:0 overruns:0 carrier:0

            collisions:0 txqueuelen:1000

            RX bytes:0 (0.0 b)  TX bytes:1156 (1.1 KiB)

eth1.44   Link encap:Ethernet  HWaddr 52:54:00:60:7F:57

            inet6 addr: fe80::5054:ff:fe60:7f57/64 Scope:Link

            UP BROADCAST RUNNING MULTICAST  MTU:1500  Metric:1

            RX packets:0 errors:0 dropped:0 overruns:0 frame:0

            TX packets:7 errors:0 dropped:0 overruns:0 carrier:0

            collisions:0 txqueuelen:0

________________________________

From: Yaron Illouz 
Sent: Tuesday, January 27, 2015 9:56 PM
To: &#39;libvirt-users(a)redhat.com&#39;
Subject: Sr-iov passthrough - no packet arrive to guest

Hi 

I am trying to do passthrough to guest using sriov and libvirt

I tried to use both techniques from here

http://wiki.libvirt.org/page/Networking#Assignment_with_.3Chostdev.3E
<http://wiki.libvirt.org/page/Networking#Assignment_with_.3Chostdev.3E> 

Assignment from a pool of SRIOV VFs in a libvirt <network> definition

Assignment with <interface type='hostdev'> (SRIOV devices only)

<network>

   <name>passthrough-eth6</name>

   <forward mode='hostdev' managed='yes'>

     <pf dev='eth6'/>

   </forward>

 </network>

I launch the vm, there is a vm with virt-install .... --network
network=passthrough-eth6 

I inject traffic with a generator so I defined it in promisc mode
without an ip address.(ifconfig eth6 promisc up)

In my physical device eth6 i can see RX increasing. But in the guest
side nothing arrive in the RX? Why?

I don't know in which log to look at to understand the problem

I defined the  network-script/eth1 file in the guest is defined as
static (not dhcp) as the interface it is connected to in host have no
ip.

I have a another interface on the host connected via ovs and libvirt and
it is working fine - you can see it in log as vnet0(thanks to  this
mailing-list help).

In the guest part I attached the log but I think interesting part is 

Jan 27 17:03:21 localhost kernel: ixgbevf 0000:00:04.0: NIC Link is Up,
10 Gbps

I also attached dmesg output from guest

Here is the log from the host side, I can't see an error.
/var/log/message     

Jan 27 12:02:29 1235 avahi-daemon[2598]: Withdrawing address record for
fe80::fc54:ff:fe92:7352 on vnet0.

Jan 27 12:02:29 1235 kernel: device vnet0 left promiscuous mode

Jan 27 12:02:30 1235 kernel: ixgbe 0000:15:00.0: setting MAC
a2:fc:a8:89:2a:8f on VF 0

Jan 27 12:02:30 1235 ovs-vsctl: 00001|vsctl|INFO|Called as ovs-vsctl
--timeout=5 -- --if-exists del-port vnet0

Jan 27 12:02:30 1235 ovs-vswitchd: 00180|bridge|ERR|bridge virbr0:
mirror mirror-1-to-3 does not specify output; ignoring

Jan 27 12:02:31 1235 ntpd[2905]: Deleting interface #28 vnet0,
fe80::fc54:ff:fe92:7352#123, interface stats: received=0, sent=0,
dropped=0, active_time=981 secs

Jan 27 12:02:31 1235 ntpd[2905]: peers refreshed

Jan 27 12:02:43 1235 kernel: ixgbe 0000:15:00.0: Reload the VF driver to
make this change effective.

Jan 27 12:02:43 1235 kernel: kvm: 7957: cpu0 unhandled rdmsr: 0x345

Jan 27 12:02:43 1235 kernel: kvm: 7957: cpu0 unhandled wrmsr: 0x680 data
0

...

Jan 27 12:02:51 1235 ovs-vsctl: 00001|vsctl|INFO|Called as ovs-vsctl
--timeout=5 -- --may-exist add-port br3 vnet0 -- set Interface vnet0
"external-ids:attached-mac=\"52:54:00:43:BF:8C\"" -- set Interface
vnet0
"external-ids:iface-id=\"309ed431-417f-2843-89b8-179a066aa595\"" --
set
Interface vnet0
"external-ids:vm-id=\"870e3b65-953b-01ee-84a3-c2d05300c2d9\"" -- set
Interface vnet0 external-ids:iface-status=active

Jan 27 12:02:51 1235 ovs-vswitchd: 00183|bridge|ERR|bridge virbr0:
mirror mirror-1-to-3 does not specify output; ignoring

Jan 27 12:02:51 1235 kernel: device vnet0 entered promiscuous mode

Jan 27 12:02:51 1235 kernel: ixgbe 0000:15:00.0: setting MAC
52:54:00:d9:f1:bd on VF 0

Jan 27 12:02:51 1235 kernel: ixgbe 0000:15:00.0: Reload the VF driver to
make this change effective.

Jan 27 12:02:51 1235 kernel: pci-stub 0000:15:10.0: enabling device
(0000 -> 0002)

Jan 27 12:02:52 1235 avahi-daemon[2598]: Registering new address record
for fe80::fc54:ff:fe43:bf8c on vnet0.*.

Jan 27 12:02:54 1235 ntpd[2905]: Listen normally on 29 vnet0
fe80::fc54:ff:fe43:bf8c UDP 123

Jan 27 12:02:54 1235 ntpd[2905]: peers refreshed

Jan 27 12:03:17 1235 kernel: __ratelimit: 23 callbacks suppressed

Jan 27 12:03:17 1235 kernel: kvm: 8006: cpu0 disabled perfctr wrmsr:
0xc1 data 0xabcd

Jan 27 12:03:20 1235 kernel: ixgbe 0000:15:00.0: eth6: VF Reset msg
received from vf 0

And this is the /var/log/libvirt/qemu/probe8.log from the host

2015-01-27 17:02:51.762+0000: starting up

LC_ALL=C PATH=/sbin:/usr/sbin:/bin:/usr/bin QEMU_AUDIO_DRV=none
/usr/libexec/qemu-kvm -name probe8 -S -M rhel6.6.0 -enable-kvm -m 4096
-realtime mlock=off -smp 4,sockets=4,cores=1,threads=1 -uuid
870e3b65-953b-01ee-84a3-c2d05300c2d9 -nodefconfig -nodefaults -chardev
socket,id=charmonitor,path=/var/lib/libvirt/qemu/probe8.monitor,server,n
owait -mon chardev=charmonitor,id=monitor,mode=control -rtc base=utc
-no-shutdown -device ich9-usb-ehci1,id=usb,bus=pci.0,addr=0x5.0x7
-device
ich9-usb-uhci1,masterbus=usb.0,firstport=0,bus=pci.0,multifunction=on,ad
dr=0x5 -device
ich9-usb-uhci2,masterbus=usb.0,firstport=2,bus=pci.0,addr=0x5.0x1
-device
ich9-usb-uhci3,masterbus=usb.0,firstport=4,bus=pci.0,addr=0x5.0x2 -drive
file=/data/KVM/comp_VNF_MaveriQ_Probe2-cloud8.qcow2,if=none,id=drive-ide
0-0-0,format=qcow2,cache=none -device
ide-drive,bus=ide.0,unit=0,drive=drive-ide0-0-0,id=ide0-0-0,bootindex=1
-drive
file=/var/lib/libvirt/images/configuration8.iso,if=none,id=drive-virtio-
disk0,format=raw,cache=none -device
virtio-blk-pci,scsi=off,bus=pci.0,addr=0x6,drive=drive-virtio-disk0,id=v
irtio-disk0 -netdev tap,fd=25,id=hostnet0 -device
rtl8139,netdev=hostnet0,id=net0,mac=52:54:00:43:bf:8c,bus=pci.0,addr=0x3
-chardev pty,id=charserial0 -device
isa-serial,chardev=charserial0,id=serial0 -vnc 0.0.0.0:0 -vga cirrus
-device
pci-assign,host=15:10.0,id=hostdev0,configfd=26,bus=pci.0,addr=0x4
-device virtio-balloon-pci,id=balloon0,bus=pci.0,addr=0x7 -msg
timestamp=on

char device redirected to /dev/pts/1

I can see that virtual function succeeded because I got the following
output

virsh net-dumpxml  passthrough-eth6

<network connections='1'>

  <name>passthrough-eth6</name>

  <uuid>f8fd37cd-3215-e4b0-17b1-0b5515140db9</uuid>

  <forward mode='hostdev' managed='yes'>

    <pf dev='eth6'/>

    <address type='pci' domain='0x0000' bus='0x15'
slot='0x10'
function='0x0'/>

    <address type='pci' domain='0x0000' bus='0x15'
slot='0x10'
function='0x2'/>

    <address type='pci' domain='0x0000' bus='0x15'
slot='0x10'
function='0x4'/>

    <address type='pci' domain='0x0000' bus='0x15'
slot='0x10'
function='0x6'/>

    ...

  </forward>

</network>

lspci | grep Intel

15:00.0 Ethernet controller: Intel Corporation 82599ES 10-Gigabit
SFI/SFP+ Network Connection (rev 01)

15:00.1 Ethernet controller: Intel Corporation 82599ES 10-Gigabit
SFI/SFP+ Network Connection (rev 01)

15:10.0 Ethernet controller: Intel Corporation 82599 Ethernet Controller
Virtual Function (rev 01)

15:10.1 Ethernet controller: Intel Corporation 82599 Ethernet Controller
Virtual Function (rev 01)

....

And Intel-IOMMU: enabled in the dmesg

...
From host dmesg 

ixgbe 0000:15:00.0: setting MAC 52:54:00:97:c6:86 on VF 0