Xen project Mailing List

Because the board is not a Intel One which just used the Q35 chips.
So if i want use the PCIE NVIDIA card with the VT-D enable.
I ought to follow that way.

http://www.nabble.com/Re:-Re:-VT-D-RMRR-is-incorrect-p22022061.html

Use this work around and (/* the whole “if” section */) and recompile Xen

I knew which was not a good way but i have no choice.Now I have modified the source code and recompiled. Everything seems good .

Domain0 works with VT-D. DomainU asigned the PCIE NVIDIA card and installed the nvidia-177-83 drivers.

But I am still face the tough question mentioned before that DomainU cann't be displaied on the second screen.

Booted the DomainU with console --> configured the driver for card --> startx

DomainU staied on the same screen then went into black screen but still alived.

2009/4/16 Ross Philipson <Ross.Philipson@xxxxxxxxxx>

Also a workaround was introduced a few months ago to work around RMRR mapping issues in dom0. You use the parameter “iommu_inclusive_mapping=1“ on the xen command line to map all reserved and conventional memory below 4GB for the iommu. Maybe this would help.

Thanks

Ross

From: xen-devel-bounces@xxxxxxxxxxxxxxxxxxx [mailto:xen-devel-bounces@xxxxxxxxxxxxxxxxxxx] On Behalf Of Cihula, Joseph
Sent: Wednesday, April 15, 2009 1:22 PM
To: Tim Moore; 'wei zhou'

Cc: xen-devel@xxxxxxxxxxxxxxxxxxx
Subject: [Xen-devel] RE: Some questions of VT-D graphic passthrough

And Everytime i use the PCIE card render for Domain0, the VT-D can't be Enable Because of the RMMR incorrect.

Are you using the latest BIOS version? There have been several BIOS releases that have improved VT-d and TXT support.

Joe

From: xen-devel-bounces@xxxxxxxxxxxxxxxxxxx [mailto:xen-devel-bounces@xxxxxxxxxxxxxxxxxxx] On Behalf Of Tim Moore
Sent: Wednesday, April 15, 2009 3:42 AM
To: 'wei zhou'
Cc: xen-devel@xxxxxxxxxxxxxxxxxxx
Subject: [Xen-devel] RE: Some questions of VT-D graphic passthrough

Answers inline below….

   I heard the good news you published from Xen Maillist.That really inspired me. Recently i was working on vt-d graphic passthrough experiment. That was a real tough job for me. I faced some problems that you may also met before. So i hope you can give me any advises or informaitons.
   Any reply appreciated!

HardWare: Q35 matherboard + PCIE NVIDIA 9400 + PCI ATI R128 + IGD Intel
                   Two Monitors connected with respective cards
OS           : Domain0 (Gentoo X86_64 2.6.28)
                   DomainU (Ubuntu i386 / windows xp / RedHat Ent4.5)
                   Xen3.3.1

At first ,I assumed the Matherboard integrated graphic card for Domain0 then passthrough the PCIE NVIDIA graphic card for DomainU.
But My Matherboard cann't support those two cards work together.
And Everytime i use the PCIE card render for Domain0, the VT-D can't be Enable Because of the RMMR incorrect.

See:

http://www.nabble.com/Re:-Re:-VT-D-RMRR-is-incorrect-p22022061.html

Use this work around and (/* the whole “if” section */) and recompile Xen

I think VT-d is needed for this to work …

Considered that reason , the PCIE card was aborted in my experiment.
Now I use the integrated card as the primary card ,the ATI PCI graphic card as the secondly.
Now i face the problem is that the DomainU asigned the ATI card can't displaied on the Monitor connected with the ATI card.

From you information on Xen maillist,
you passthrough the PCIE NVIDIA card to DomainU --> boot from console --> setup the proper driver for that card --> startx --> then the DomainU display on the second Monitor
Any other works ?
Is that right? You didn't change any source codes. The whole process are works about configurations.
I tried install a R128 driver in DomainU for the ATI card and Modified the xorg.conf(/etc/X11/xorg.conf) to tell the DomainU use the ATI card do render work.
whatever i do the DomainU always stay on the same screen.

Now i have none idea for the next step .
Hope you can give me some help.

Thanks so much !

--
Best regards ！
周魏
E-mail: weizhou.sir@xxxxxxxxx

[Xen-devel] Re: Some questions of VT-D graphic passthrough