Re: [Xen-users] using physical NIC

[Ryan <hap9@xxxxxxxxxxxxxx>]

On Mon, 2006-04-10 at 09:48 -0600, Glen Davis wrote:
> Thanks, the  'echo - n 0000:02:06.0 > /sys/bus/pci/drivers/pciback/bind'
> was what I was missing.  After doing that it all worked. 
> 
> One more question, when do devices normally get bound to a backend driver?  
> Is this something that I should always manually do or should this have 
> already been taken care of during the install of the server?
> 

Glad that fixed your problem! (I hope you don't mind but I'm posting
this back to xen-users so the solution to your problem is archived)

Normally, devices get bound when the kernel boots. You specify them on
the kernel command-line with "pciback.hide". Then, the pciback driver
loads itself first on the kernel's list of PCI drivers so that it gets
to be the first driver to "seize" a particular PCI device (this way, it
gets the devices before the device driver for that device can seize it).

There could be a couple reasons why this didn't work for you. One could
be that the pciback module isn't correctly parsing your command-line
parameter and seizing it. Try looking through the logs from when your
kernel booted (dmesg) and see if you see any statements about the
pciback driver. If not, you could try and re-compile your kernel with
PCI Backend Debugging turned on (it's under the Xen menu) and see what's
happening. Another problem could be that pciback is compiled as a module
instead of being built-in to the kernel. This means that pciback can't
load itself as first in the list of PCI drivers and it may not be able
to seize all devices (other device drivers might get the device first
because they're ahead of the pciback driver on the kernel's pci driver
linked list). If the pciback module shows up in "lsmod", then this is
most likely your problem and you'll have to do the manual bind every
time (after first unbinding the device from any device driver it might
have bound to first).

If you find out what's wrong, please let me know as I'm curious to know
what the real issue is.

Ryan


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