WARNING - OLD ARCHIVES

This is an archived copy of the Xen.org mailing list, which we have preserved to ensure that existing links to archives are not broken. The live archive, which contains the latest emails, can be found at http://lists.xen.org/
   
 
 
Xen 
 
Home Products Support Community News
 
   
 

xen-users

[Xen-users] Problem with 802.1Q tagging in Xen domain 0. (May be HW vlan

To: xen-users@xxxxxxxxxxxxxxxxxxx
Subject: [Xen-users] Problem with 802.1Q tagging in Xen domain 0. (May be HW vlan acceleration related?)
From: Igor Chubin <igor@xxxxxxx>
Date: Mon, 9 Apr 2007 12:11:56 +0300
Delivery-date: Mon, 09 Apr 2007 02:10:43 -0700
Envelope-to: www-data@xxxxxxxxxxxxxxxxxx
List-help: <mailto:xen-users-request@lists.xensource.com?subject=help>
List-id: Xen user discussion <xen-users.lists.xensource.com>
List-post: <mailto:xen-users@lists.xensource.com>
List-subscribe: <http://lists.xensource.com/cgi-bin/mailman/listinfo/xen-users>, <mailto:xen-users-request@lists.xensource.com?subject=subscribe>
List-unsubscribe: <http://lists.xensource.com/cgi-bin/mailman/listinfo/xen-users>, <mailto:xen-users-request@lists.xensource.com?subject=unsubscribe>
Reply-to: Igor Chubin <igor@xxxxxxx>
Sender: xen-users-bounces@xxxxxxxxxxxxxxxxxxx
User-agent: Mutt/1.5.13 (2006-08-11)


Hello, all.



In the message 
[http://lists.xensource.com/archives/html/xen-users/2007-04/msg00190.html].
I have described the problem I've faced recently.

Now I want to give additional information about this.
I think that my problem is related to hardware VLAN acceleration.



So, here is my problem:


I have Xen dom 0 with several network interfaces.
I want to use 802.1Q VLAN's on one of the interfaces.



If I don't use xenbr0 (fig.1)
all works fine.



                     eth0    .---- eth0   VLAN1 [10.0.1.20]
============================<----- eth0.2 VLAN2 [10.0.2.20]
(tagged)                     .____ eth0.3 VLAN3 [10.0.3.20]

Figure 1.



      peth0 +------+ eth0    .---- eth0   VLAN1 [10.0.1.20]
============|xenbr0|========<----- eth0.2 VLAN2 [10.0.2.20]
(tagged)    +------+         .____ eth0.3 VLAN3 [10.0.3.20]

Figure 2.




But when I add the bridge xenbr0 (using network-bridge script)
only untagged interface (VLAN1, 10.0.1.20) works properly.



In instance,  I try to ping outside host 10.0.3.5 from 10.0.3.20 (VLAN3).
I have no responses.

During the pinging process I look into the peth0 (or xenbr0)
interface using tcpdump:


$ sudo tcpdump -i peth0 -e -n  arp
tcpdump: WARNING: peth0: no IPv4 address assigned
tcpdump: verbose output suppressed, use -v or -vv for full protocol decode
listening on peth0, link-type EN10MB (Ethernet), capture size 96 bytes
11:09:17.755752 00:16:3e:00:04:10 > 00:30:48:85:df:64, ethertype ARP (0x0806), 
length 56: arp reply 10.0.3.5 is-at 00:16:3e:00:04:10
11:09:18.763945 00:16:3e:00:04:10 > 00:30:48:85:df:64, ethertype ARP (0x0806), 
length 56: arp reply 10.0.3.5 is-at 00:16:3e:00:04:10
11:09:19.763942 00:16:3e:00:04:10 > 00:30:48:85:df:64, ethertype ARP (0x0806), 
length 56: arp reply 10.0.3.5 is-at 00:16:3e:00:04:10
11:09:20.764029 00:16:3e:00:04:10 > 00:30:48:85:df:64, ethertype ARP (0x0806), 
length 56: arp reply 10.0.3.5 is-at 00:16:3e:00:04:10


As you can see there are ARP-replyes, but there are no 802.Q tags in it!


The tags present in ARP-queries
(I have used tcpdump filter "vlan 3". 
You may notice also that tcpdump doesn't show replyes 
-- as you know for this moment -- replyes are not tagged):

$ sudo tcpdump -i xenbr0 -e -n vlan 3 
tcpdump: WARNING: xenbr0: no IPv4 address assigned
tcpdump: verbose output suppressed, use -v or -vv for full protocol decode
listening on tagged0, link-type EN10MB (Ethernet), capture size 96 bytes
10:59:32.790896 00:30:48:85:df:64 > ff:ff:ff:ff:ff:ff, ethertype 802.1Q 
(0x8100), length 46: ethertype ARP, arp who-has 10.0.3.5 tell 10.0.3.20
10:59:34.791017 00:30:48:85:df:64 > ff:ff:ff:ff:ff:ff, ethertype 802.1Q 
(0x8100), length 46: ethertype ARP, arp who-has 10.0.3.5 tell 10.0.3.20
10:59:35.791081 00:30:48:85:df:64 > ff:ff:ff:ff:ff:ff, ethertype 802.1Q 
(0x8100), length 46: ethertype ARP, arp who-has 10.0.3.5 tell 10.0.3.20
10:59:36.791144 00:30:48:85:df:64 > ff:ff:ff:ff:ff:ff, ethertype 802.1Q 
(0x8100), length 46: ethertype ARP, arp who-has 10.0.3.5 tell 10.0.3.20



I think that my problem is related to hardware VLAN acceleration
of Broadcom card.

The card model:
%# lspci | grep -i broadcom
04:00.0 Ethernet controller: Broadcom Corporation NetXtreme BCM5721 Gigabit 
Ethernet PCI Express (rev 11)
05:00.0 Ethernet controller: Broadcom Corporation NetXtreme BCM5721 Gigabit 
Ethernet PCI Express (rev 11)


(I have mentioned in my previous post that there are no problem with this 
configuration
on any of my other Xen installations. But I have no Broadcom card on these 
installations too).





Please tell me, 
do you think that my idea about hardware VLAN acceleration may be right?
And if so, what should I do?


Does anybody see something similar?


Can problem be solved using another (may be newer) driver?
Information about driver that I use:

$ sudo modinfo tg3
filename:       /lib/modules/2.6.18-4-xen-686/kernel/drivers/net/tg3.ko
author:         David S. Miller (davem@xxxxxxxxxx) and Jeff Garzik 
(jgarzik@xxxxxxxxx)
description:    Broadcom Tigon3 ethernet driver
license:        GPL
version:        3.65
vermagic:       2.6.18-4-xen-686 SMP mod_unload 686 REGPARM gcc-4.1



Is there any way to switch HW acceleration off (at least for debug purposes)?



Thank you for any ideas!


-- 
WBR, i.m.chubin


_______________________________________________
Xen-users mailing list
Xen-users@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-users

<Prev in Thread] Current Thread [Next in Thread>
  • [Xen-users] Problem with 802.1Q tagging in Xen domain 0. (May be HW vlan acceleration related?), Igor Chubin <=