From: Gordan Bobic <gordan@bobich.net>
To: John Wargo <wargo.john25@gmail.com>,
"xen-devel@lists.xen.org" <xen-devel@lists.xen.org>
Subject: Re: Windows 8 VGA Passthrough
Date: Sun, 27 Oct 2013 19:35:45 +0000 [thread overview]
Message-ID: <526D6B11.5080803@bobich.net> (raw)
In-Reply-To: <526D3D91.5090906@gmail.com>
On 10/27/2013 04:21 PM, John Wargo wrote:
> On 10/27/2013 09:27 AM, John Wargo wrote:
>> On 10/27/2013 04:30 AM, Gordan Bobic wrote:
>>> On 10/26/2013 09:58 PM, John Wargo wrote:
>>>> Geniuses,
>>>>
>>>> I've been trying to get VGA pass-through working on Windows 8, however
>>>> I've been running into a few issues.
>>>> Currently I can get Windows 8 running flawlessly without VGA
>>>> pass-through, however I'm planning on using it for gaming, so this wont
>>>> work.
>>>>
>>>> When I start windows 8, I am able to get past the install portion of
>>>> the
>>>> DVD, however it hangs on "Getting devices ready.. (n%)".
>>>>
>>>> Hardware Information :
>>>> AMD Fx6300
>>>> ASUS M5a99x evo r2.0
>>>> 16GB RAM
>>>> 6670 Radeon HD - Dom0
>>>> 7850 Radeon HD - Win8
>>>
>>> I have recently (recently as in over the past 48 hours) re-tried to
>>> get ATI VGA passthrough working, and the results are... interesting.
>>>
>>> HD4850 dom0 + HD7450 domU worked fine - or as fine as ATI passthrough
>>> ever seems to work (slowdown + white lines when dragging windows
>>> after a domU reboot). I am using XP64 domU so I can't eject the GPU
>>> (it says it is being used). But shutting down the domU, doing echo 1
>>> > reset for the GPU from dom0 (I don't know how this is implemented
>>> and whether it does anything) and then starting it up again sometimes
>>> recovers it. GPU-Z runs and reports the GPU correctly.
>>>
>>> HD4850 dom0 + HD7970 domU results in the entire host locking up as
>>> soon as the driver is loaded in domU. My solution is to swap the dom0
>>> GPU for an 8800GT, and the problem now doesn't occur. If I had to
>>> guess, BIOS on the HD7970 does something bizzare, finds the HD4850 on
>>> the bus, tries to do something to it, and ends up crashing the host.
>>> This could also be exacerbated by the fact that I have to run with
>>> relaxed checking and no PCI ACS due to NF200 bridges on my
>>> motherboard. Even so, however, GPU-Z crashes the domU and it's GPU
>>> _hard_ (but not the host, thankfully). It crashes it so hard, that
>>> the only cure is to reboot the host. Reseting the GPU via /sys/
>>> doesn't seem to do anything to help.
>>>
>>> Are you using gfx_passthru=1 in your config? The way most people have
>>> this kind of thing working is with gfx_passthru=0, i.e. secondary
>>> passthru. Don't assign the PCI device in the config when installing.
>>> Once the domU install is complete, shut it down, and add the GPU you
>>> want to pass through. Then fire up domU in VNC, install the driver,
>>> and with a bit of luck it will come up with output on the external
>>> monitor.
>>>
>>> Also make sure that radeon or radeonfb kernel drivers DO NOT load
>>> before the device is assigned to xen-pciback. The exact way to ensure
>>> this will depend on what distro and dom0 kernel you are using.
>>>
>>> Gordan
>> Status update:
>>
>> The catalyst drivers appear to be working perfectly fine, device
>> manager shows no errors, and everything works great, except that when
>> I plug my monitor into the second card, I get "No Signal" on the
>> monitor, then it shuts off.
>>
>> I though that secondary passthrough meant that it would pass the
>> virtual adapter in as primary, but I could use my GPU that I passed
>> through, so where am I wrong, and how do I do that?
>>
>> -John
> I've been doing some research into my VGA passthrough, so I've gone
> through the various debugging steps, and the only thing that I've found
> out of place is with "xl dmesg" (attached).
>
> Where do I go from here?
Sounds like your IOMMU may be buggy. You may need to set permissive=1 on
your xen-pciback module.
I cannot remember if xl uses the same configuration, but for xm you
could try putting this in your /etc/xen/xend-config.sxp:
(pci-passthrough-strict-check no)
(pci-dev-assign-strict-check no)
Gordan
next prev parent reply other threads:[~2013-10-27 19:35 UTC|newest]
Thread overview: 20+ messages / expand[flat|nested] mbox.gz Atom feed top
2013-10-26 20:58 Windows 8 VGA Passthrough John Wargo
2013-10-27 9:30 ` Gordan Bobic
2013-10-27 14:27 ` John Wargo
2013-10-27 16:21 ` John Wargo
2013-10-27 17:21 ` John Wargo
2013-10-28 17:07 ` Konrad Rzeszutek Wilk
2013-10-28 21:32 ` John Wargo
2013-10-29 13:09 ` Zytaruk, Kelly
2013-10-28 19:30 ` Zytaruk, Kelly
2013-10-28 20:31 ` John Wargo
2013-10-28 20:46 ` Gordan Bobic
2013-10-28 21:29 ` John Wargo
2013-10-27 19:35 ` Gordan Bobic [this message]
2013-10-27 19:45 ` John Wargo
2013-10-27 20:05 ` John Wargo
2013-10-27 21:01 ` John Wargo
2013-10-27 22:51 ` Gordan Bobic
2013-10-28 11:06 ` John Wargo
2013-10-27 19:31 ` Gordan Bobic
[not found] ` <526D1FBA.10708@gmail.com>
2013-10-27 19:28 ` Gordan Bobic
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=526D6B11.5080803@bobich.net \
--to=gordan@bobich.net \
--cc=wargo.john25@gmail.com \
--cc=xen-devel@lists.xen.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).