Re: [Xen-users] Problem with VT-d on Mainboard Intel DQ45CB

["Todd Deshane" <deshantm@xxxxxxxxx>]

Hi.

Some questions and comments inline.

On Thu, Oct 23, 2008 at 3:25 AM, Franz Regensburger
<Franz.Regensburger@xxxxxxxx> wrote:
> Hello all,
>
> this mail is related to a post by  Axel Schöner that I found at
>        http://article.gmane.org/gmane.comp.emulators.xen.user/40621
> when I was searching the WEB about solutions to my problem.
>
> I would like to confirm that VT-d with XEN 3.3 kernel 2.6.18.8 on mainboard
> Intel DQ45CB seems to be broken.
>
> Survey (details are down below)
> ---------------------------------------
>
> Whenever I boot my Xen 2.6.18.8 (Xen 3.3) with kernel parameter iommu=1
> the boot process freezes.
>
> I see:
> ...
> (XEN) Intel VT-d has been enabled
> ...
> (XEN) HVM: VMX enabled
> ...
> (XEN) I/O virtualisation enabled
> (XEN) I/O virtualisation for PV guests disabled
> Here the boot process freezes
>

I am surprised it freezes here.

A serial console debug might be of help here see:
http://xen.markmail.org/search/?q=serial%20console%20debugging#query:serial%20console%20debugging+page:1+mid:scjkellgexgofu6d+state:results

> Does anybody else suffer from the same problem (VT-d with Intel board DQ45CB)?
> Any suggestions about a solution?
>

This VT-d stuff is quite new and the Intel developers really need help
testing and debugging.


>From the VTdHowTo on the Xen wiki I see:

 kernel /boot/xen.gz com1=115200,8n1 console=com1 iommu=1
        module /boot/vmlinuz-2.6.18.8-xen root=LABEL=/ ro xencons=ttyS
console=tty0 console=ttyS0, pciback.hide=(01:00.0)(03:00.0)
        module /boot/initrd-2.6.18-xen.img

There are some differences in the console's used with yours.

> title           Xen 3.3.0 / kernel 2.6.18.8-xen backhide the ATI Radeon
> root            (hd0,0)
> #kernel          /xen-3.3.0.gz console=vga noreboot max_loop=128 
> dom0_mem=1024M iommu=1
> kernel          /xen-3.3.0.gz console=vga noreboot max_loop=128 dom0_mem=1024M
> module          /vmlinuz-2.6.18.8-xen root=/dev/mapper/RootVG-root ro 
> console=tty0 pciback.hide=(01:00.0)(01:00.1)
> module          /initrd-2.6.18.8-xen.img
> savedefault


> Additional problem respectively disappointments with the board DB45CB
> ---------------------------------------------------------------------------------------
>
> Now, I just need to get rid of my frustrations. Of course, some or all of 
> them might be caused by my
> personal lack of Linux or Xen experience ...
>
> - I am not able to configure the onboard GB ethernet adapter on Linux.
>  Even the e1000e driver

This works for me in Ubuntu (not Xen in my experience yet).

>    susi:/lib/modules/2.6.26-bpo.1-amd64/kernel/drivers/net/e1000e
>  of newer kernels does not seam to support that adapter.
>  As a fix I run addtional NICs in PCI and PCIe slots (inetl 1000 pro, various 
> brands)
>  The onboard LAN is unused.
>


> - I am not able to get Xorg running with the onboard graphics adapter.
>  There is no support for the Q45 Chipset (at least not in the 2.6.18-6, 
> 2.6.24 and of course not in the xen kernels)
>

I would try the latest Ubuntu --Intrepid is at RC now-- (even just the
liveCD here for testing)


> In summary I am rather disappointed by the board DB45CB as a platform for XEN.
>        - VT-d broken. No PCI backend hiding with HVM guest possible
>        - no onboard LAN on linux
>        - no remote management from linux
>        - no onboard graphics on linux
>        - PCIe x16 graphics with Xen kernel broken
>

Living on the cutting edge is like this sometimes, it will only get better.

> On the positive side I can state the following results:
>
> - Xen host without X11 (xorg) runs perfectly (especially fast).
> - I was able to configure and run a WinXP HVM guest (although without VT-d 
> and PCI backend hiding)
>
> Is there someone out who can show me light at the end of the tunnel?
>

Here are my suggestions and things to watch for.

The Intel developers are quite good and responsive on xen-devel,
provided that you are willing to do
some testing/debugging and applying patches etc.

To get to the point where you can be helpful, you will need to get the
latest xen-unstable.

hg clone http://xenbits.xensource.com/xen-unstable.hg
hg clone http://xenbits.xensource.com/linux-2.6.18-xen.hg
cd xen-unstable
make world

Follow these again
http://wiki.xensource.com/xenwiki/VTdHowTo

Do you testing again and send a shorter, less verbose email to xen-devel.
Give the key results and the most descriptive and useful snippets of
the appropriate logs.

They will work likely work with you, especially since there are quite
a few others that
are running up against this problem and are also likely to help with testing.

The Light at the end of the tunnel on the Xen/Linux side of things
will happen as
the Xen dom0 gets into mainline. Progress on this seems to be going pretty well.
see:
http://wiki.xensource.com/xenwiki/XenParavirtOps
http://xenbits.xensource.com/paravirt_ops/patches.hg


Hope that gives you a better sense and gets the debugging process underway.

Best Regards,
Todd

-- 
Todd Deshane
http://todddeshane.net
http://runningxen.com

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