Thank you for your reply.
Well, I am not an expert for network cards or the networking stack in
the kernel as both of you are. So I can't (and don't want to) decide if
it is a driver or a kernel module problem. I think I could characterize
the problem relatively precisely and think it should be fixed because it
is a relatively common network card in (older) servers.
Since I know almost nothing about kernel programming, it would be great,
if one of you could provide me with a patch I could test. If further
debug output is needed which I could generate or anything else I can do
(in a limited amount of time) I would like to help you as much as I can.
Thanks,
Timm
On 09/11/2012 06:32 PM, Jesse Gross wrote:
OVS doesn't clear NETIF_F_HW_VLAN_RX but it's possible that something
else does. In any case, the driver must support stripping and
insertion independently and if the hardware doesn't support it then it
must set the feature flags so that things get emulated appropriately.
On Tue, Sep 11, 2012 at 1:02 AM, Timm Essigke
<timm.essi...@uni-bayreuth.de <mailto:timm.essi...@uni-bayreuth.de>>
wrote:
Dear OVS developers,
some days ago (09/04/12) I reported my problems with Intel PCI-X
NICs on this mailing list. Jesse Gross considered it as a driver
problem and pointed me to the e1000 mailing list.
Jesse Brandeburg (Intel) kindly analyzed the problem and plays the
ball back to your side (see below).
Can you please join the discussion?
Thank you!
Timm
-------- Original Message --------
Subject: [e1000:bugs] #357 Intel 82546GB chip does not work with
OpenVSwitch
Date: Tue, 11 Sep 2012 01:11:46 +0000
From: Jesse Brandeburg <go_je...@users.sf.net>
<mailto:go_je...@users.sf.net>
Reply-To: [e1000:bugs] <3...@bugs.e1000.p.re.sf.net>
<mailto:3...@bugs.e1000.p.re.sf.net>
To: [e1000:bugs] <3...@bugs.e1000.p.re.sf.net>
<mailto:3...@bugs.e1000.p.re.sf.net>
I found why you don't see any transmits. CTRL.VME is not set
(0x40000000) and needs to be for the hardware to insert the vlan
tag on transmit. So your hardware is currently transmitting
untagged frames when openvswitch enabled the vlan. This appears to
be (unverified) because openvswitch is actively clearing the
netdev->features flag NETIF_F_HW_VLAN_RX, and then doesn't read
the features mask back afterward to find out that our e1000
hardware no longer supports tag insertion.
------------------------------------------------------------------------
*bugs:357 Intel 82546GB chip does not work with OpenVSwitch*
*Status:* open *Labels:* e1000 *Created:* Mon Sep 10, 2012 09:45
AM UTC by T. Essigke *Last Updated:* Tue Sep 11, 2012 12:29 AM UTC
*Owner:* Jesse Brandeburg
The 82546GB chip does not work with OpenVSwitch, while similar
chips do. It works with the old vlan/bridge code. In the
attachement is a script to reproduce the problem and the output of
ethregs.
Here is a summary of the results:
vlan/bridge openvswitch
ping ethregs ping ethregs
80003ES2 + - + -
80003ES2_iomem + + + +
80003ES2_iomem_run2 + + + +
82546GB + + - +
82546GB_iomem + + - +
82546GB_iomem_run2 - + - +
Maybe it is an initialization issue of the card.
For more details see my mails on the e1000-devel mailing list.
------------------------------------------------------------------------
Sent from sourceforge.net <http://sourceforge.net> because you
indicated interest in https://sourceforge.net/p/e1000/bugs/357/
To unsubscribe from further messages, please visit
https://sourceforge.net/auth/prefs/
--
========================================
Dr. Timm Essigke
Structural Biology/Bioinformatics
University of Bayreuth
Universitätsstr. 30, BGI
95447 Bayreuth
Germany
Telephone: +49-921-55-3547
Fax: +49-921-55-3071
e-mail: timm.essi...@uni-bayreuth.de
http://www.bisb.uni-bayreuth.de/
========================================
_______________________________________________
discuss mailing list
discuss@openvswitch.org
http://openvswitch.org/mailman/listinfo/discuss
