[vfio-users] PCI Passthrough of GTX 980 on guest getting "unable to determine the device handle"

[Kevin Vasko]

I haven't seen any other error messages in any of the logs. 

Sorry, what is ovmf?

-Kevin

> On Dec 15, 2016, at 11:02 PM, Zachary Boley <zboley00 at gmail.com> wrote:
> 
> All I can say is try vfio-pci if you haven't already, may get a 1070 but would love to know if those work, and the guest is Linux with ovmf? Not sure if it'd help but do you have the nvidia error 43 stuff in the libvirt config
> 
>> On Dec 7, 2016 11:35 PM, "Kevin Vasko" <kvasko at gmail.com> wrote:
>> No, the problem is that the drivers don't work on the guest (which would allow me to use the card on the guest system). I am speculating it is a passthrough issue or a GPU driver bug.
>> 
>> On the host the device is being recognized with lspci. I used pci-stub (since its 3.x kernel) to make sure the host system didn't attach a driver to the GPU. If I launch a guest VM to use the GPU, the guest system boots properly and sees the GPU with the lspci command (e.g. 00:05.0 VGA compatible controller: NVIDIA Corporation GM204 [GeForce GTX 980] (rev a1)). However, when I install the GPU drivers on the guest (the installation of the driver completes successfully), but when I run the nvidia-smi command (to look at the device on the guest and to make sure the guest can communicate with the GPU via passthrough) I get the error "Unable to determine the device handle for GPU 0000:00:05.0: Unknown Error" and see the error messages in dmesg (which I posted in original).
>> 
>> I have tried several different drivers, two different cards (1080 and 980), different kernels on the host and guest and still get the same error. Those error messages don't seem to lead me anywhere on google either unfortunately.
>> 
>> 
>> 
>>> On Wed, Dec 7, 2016 at 8:01 PM, Zachary Boley <zboley00 at gmail.com> wrote:
>>> are you saying its not recognized in the host? but it works in the guest perfectly? ooor? I'm not quite understanding the problem sorry
>>> 
>>>> On Wed, Dec 7, 2016 at 1:45 PM, Kevin Vasko <kvasko at gmail.com> wrote:
>>>> I am doing passthrough from Ubuntu 14.04 3.19 (Host) to an Ubuntu 14.04 guest 3.19. I have been testing with a GeForce GTX 980 and GTX 1080. I have seen people do PCI Passthrough without a problem with GeForce cards.
>>>> 
>>>> I can see device inside the guest VM:
>>>> 
>>>>     lspci -vnn | grep VGA
>>>> 
>>>>  00:05.0 VGA compatible controller: NVIDIA Corporation Device 1b80 (rev a1)
>>>> 
>>>> OR if I have the GTX 980 passed through.
>>>> 
>>>> 00:05.0 VGA compatible controller: NVIDIA Corporation GM204 [GeForce GTX 980] (rev a1)
>>>> 
>>>> I was able to successfully install the driver (370.2, latest driver). I have tried with several other driver versions with same results. It installs, but is not recognized by nvidia-smi (same result both cards):
>>>> 
>>>>     nvidia-smi 
>>>> 
>>>> > Unable to determine the device handle for GPU 0000:00:05.0: Unknown
>>>> > Error
>>>> 
>>>> 
>>>> Looking in dmesg I see the following error message
>>>> 
>>>> > [   29.535583] nvidia 0000:00:05.0: irq 45 for MSI/MSI-X [  
>>>> > 29.577727] NVRM: RmInitAdapter failed! (0x23:0x56:458) [   29.577807] NVRM: rm_init_adapter failed for device bearing minor number 0
>>>> 
>>>> Another person on the NVidia forums had the SAME exact issue as me (but no answer).
>>>> 
>>>> https://devtalk.nvidia.com/default/topic/957757/gtx-1080-amp-kvm-pci-passthrough-to-guest/?offset=2#5033118
>>>> 
>>>> Is there anyway to debug this further? I'm currently trying to narrow this down to either drivers or passthrough causing it.  
>>>> 
>>>> 
>>>> 
>>>> _______________________________________________
>>>> vfio-users mailing list
>>>> vfio-users at redhat.com
>>>> https://www.redhat.com/mailman/listinfo/vfio-users
>>>> 
>>> 
>> 
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/vfio-users/attachments/20161216/2c59cf79/attachment.htm>