[vfio-users] Nvidia Code 43 on Fedora 24 host and Windows 10 guest.

[Rokas Kupstys]

Hey could you share /home/cocobo/scripts/single_head and
/home/cocobo/scripts/dual_head scripts? They sound very interesting if
they are what i think they are.

Rokas Kupstys

On 2016.10.05 17:35, Thomas Lindroth wrote:
> On 10/05/2016 09:28 AM, Dawid Osuchowski wrote:> I've ran dd if=NV287MH.2K0 of=strip bs=1 skip=1536, put the ROM file in
>> /usr/share/qemu, run restorecon on it and it started working! I'm so
>> thankful for both of you Alex and Thomas, i've been trying to get it to
>> work for at least 4-5 days and you've helped me in just one day. While
>> we are still here, do you have any recommendations on optimizing the VM
>> so it runs smoothly almost as a native OS and how should i interact with
>> the VM? I've seen an article on
>> https://rokups.github.io/blog/#!pages/kvm-hid.md that made a software
>> KVM switch, would that be good?
> https://www.redhat.com/archives/vfio-users/2016-September/msg00072.html
> this post explains ways of reducing latency for gaming but be aware that
> reducing latency also tends to reduce performance.
>
> I use synergy to share mouse and keyboard. Apparently they've switched to
> a pay model but you can still get betas for free: https://symless.com/nightly
> Using the "input-linux" driver for recent versions of qemu is also popular
> but you won't get shared clipboard with that.
> https://www.kraxel.org/blog/2016/04/linux-evdev-input-support-in-qemu-2-6/
>
> The virtual soundcards in qemu doesn't work so well. I get the audio out
> over hdmi. I then loop an audio cable from the audio out on my monitor
> to the line-in on my regular soundcard. That way the sound from the host
> and guest are mixed together and I can control the volume independently.
> There are some issues with hdmi audio which is explained here:
> https://wiki.archlinux.org/index.php/PCI_passthrough_via_OVMF#Slowed_down_audio_pumped_through_HDMI_on_the_video_card
>
>>>> On 10/05/2016 03:46 AM, Alex Williamson wrote:
>>>>> If you want to send me your ROM file, I'll compare to mine and see
>>>>> whether it works with my card (not sure whether it should or not given
>>>>> different vendors).  Thanks,  
>>>> This conversation from irc explains where that strip.rom file comes from.
>>>>
>>>> <tulphoon> Hello. I've extracted ROM file from my card and i would like to apply UEFI "patch" to it. How should i do it? I don't want to use nvflash directly on the card.
>>>> <tulphoon> My card is MSI GTX 660, and i've got the "patch" file from this thread https://forum-en.msi.com/index.php?topic=274331.0
>>>> <tulphoon> Basically i want to merge the two files (if that's even possible) and have single EFI compatible file in the end.
>>>> <tholin>   I see no mentioning of "patch" in that thread. It looks like just a .rom file. Use that rom instead of the one you dumped.
>>>> <tulphoon> That one doesn't qualify as rom tho.
>>>> <tulphoon> at least, EfiRom -d throws out error: 3001
>>>> <tholin>   looks like you are right, it's some metadata format
>>>> <tulphoon> but i don't want to nvflash it tho.
>>>> <tulphoon> Unless that's the last solution
>>>> <tholin>   the real vbios is embedded in it. It starts at byte offset 1536 and I suppose it's 128K big like most vbios.
>>>> <tulphoon> probably the nvflash thingy is able to strip the thign and make it work.
>>>> <tulphoon> the thing*
>>>> <tholin>   try running this on it "dd if=NV287MH.2K0 of=strip bs=1 skip=1536 count=131072"
>>>> <tulphoon> okay, gimme a sec.
>>>> <tholin>   the vbios rom-parser can parse the resultant file
>>>> <tulphoon> holy fuck
>>>> <tulphoon> you are a genius sir.
>>>> <tulphoon> wait.
>>>> <tulphoon> Okay, the rom is working and has EFI support.
>>>> <tulphoon> but i'm still getting Code 43 on my Win 10 VM.
>>>> <tulphoon> I'll investigate further tomorrow though. Huge thank you for help tholin :)
>>>> <tholin>   np
>>>>
>>>> Perhaps the real rom is larger than 128K? That seems unlikely.
>>>> At least you can recreate that stripping and get the same rom.
>>> Yes, it is more than 128K, mine and the one in that rar archive.  That
>>> could very well be the problem.  Dawid, repeat what you did above, but
>>> remove the count=131072 part of the dd command, which will just let it
>>> run to the end of the file.  Thanks Thomas, that could very well be the
>>> key.
>>>
>>> Alex
> I'm not such a genius after all. Glad you found an answer.
>
> _______________________________________________
> vfio-users mailing list
> vfio-users at redhat.com
> https://www.redhat.com/mailman/listinfo/vfio-users