Re: vga passthrough // vfio // qemu bridge

[Martin Wolf <mwolf@adiumentum.com>]

thank you for the quick response alex,

but i still need your help ;)
i cloned both git trees
(http://lists.gnu.org/archive/html/qemu-devel/2013-05/msg00432.html)
that was the easy part for me (it boots like a charm ...) ... then i
built the qemu tree and found out
that it is just 1.4.50 and something was missing
so qemu would not start up ...

would you be so kind alex and supply me with
the patches you meant yesterday i would need for qemu?

ty in advance



Am 16.07.2013 16:25, schrieb Alex Williamson:
> On Tue, 2013-07-16 at 14:35 +0200, Martin Wolf wrote:
>> Early 2012 i tested the old vga passthrough capabilities of KVM and was
>> partly successful.
>> now with the new vfio driver i tried again according to alex's hints and
>> this guide:
>> https://bbs.archlinux.org/viewtopic.php?id=162768
>>
>> since im primarily using ubuntu i used the daily build of saucy.
>> it ships qemu 1.5 and seabios 1.7.3 so the requirements are met.
>>
>> according to the guide i prepared the vga card (amd 7870)
>>
>> [    0.000000] Command line: BOOT_IMAGE=/boot/vmlinuz-3.10.0-2-generic
>> root=UUID=26fed560-a972-499d-ab14-7fec6439fd3d ro intel_iommu=on
>> pci-stub.ids=1002:6818,1002:aab0 quiet splash vt.handoff=7
>> [    0.000000] Kernel command line:
>> BOOT_IMAGE=/boot/vmlinuz-3.10.0-2-generic
>> root=UUID=26fed560-a972-499d-ab14-7fec6439fd3d ro intel_iommu=on
>> pci-stub.ids=1002:6818,1002:aab0 quiet splash vt.handoff=7
>> [    0.569977] pci-stub: add 1002:6818 sub=FFFFFFFF:FFFFFFFF
>> cls=00000000/00000000
>> [    0.569987] pci-stub 0000:01:00.0: claimed by stub
>> [    0.569994] pci-stub: add 1002:AAB0 sub=FFFFFFFF:FFFFFFFF
>> cls=00000000/00000000
>> [    0.569998] pci-stub 0000:01:00.1: claimed by stub
>>
>> then did this just to be sure:
>> echo "options vfio_iommu_type1 allow_unsafe_interrupts=1" >
>> /etc/modprobe.d/vfio_iommu_type1.conf
>> (or was that wrong?)
>> im using a z87 haswell mainboard
> Hopefully not needed, only use this option if you need to.  vfio will
> print an error to dmesg and qemu will fail to start if you need it.
>
>> after that i binded the two devices to vfio-pci with:
>> vfio-bind 0000:01:00.0 0000:01:00.1 (the script in the guide)
>>
>> afterwards i was able to start the kvm with
>> qemu-system-x86_64 -enable-kvm -M q35 -m 8192 -cpu host \
>> -smp 8,sockets=1,cores=4,threads=8 \
>> -bios /usr/share/qemu/bios.bin -vga none \
>> -device
>> ioh3420,bus=pcie.0,addr=1c.0,multifunction=on,port=1,chassis=1,id=root.1 \
>> -device
>> vfio-pci,host=01:00.0,bus=root.1,addr=00.0,multifunction=on,x-vga=on \
>> -device vfio-pci,host=01:00.1,bus=root.1,addr=00.1 \
>> -device ahci,bus=pcie.0,id=ahci \
>> -drive file=/home/martin/windows.img,if=none,id=disk,format=raw -device
>> virtio-blk-pci,drive=disk \
>> -drive file=/home/martin/X17-59885.iso,id=isocd -device
>> ide-cd,bus=ahci.0,drive=isocd \
>> -net nic,model=virtio \
>> -net user \
>> -usb -usbdevice host:1532:000c \
>> -drive file=/home/martin/Downloads/virtio-win-0.1-59.iso,id=isocd1
>> -device ide-cd,bus=ahci.1,drive=isocd1
>>
>> to my surprise i instantly got the windows installation running
>> installed the virtio drivers for nic and storage
>> and had 15 mins later a working win7 installation.
>> now i installed the amd driver (13.4) and rebooted.
>> i got a bluescreen. similar to my old expiriences so i thought do a
>> clean host reboot and try again.
>> but still the same. so i tried to load the bios.rom for the card (found
>> it on techpowerup) again no luck.
>> maybe someone knows a hint?
> I think most of the folks using the guide you reference are using my
> vfio-vga-reset branches of qemu & kernel (or patches extracted from
> them).  These add an important step for reproducibility, by being able
> to reset the bus for the graphics card, giving us a way to reset the
> device.  The other thing in the qemu branch are improved quirks.  I've
> just posted these to qemu-devel and plan to get them pulled for 1.6.
> Note that for ATI/AMD cards, a critical quirks is intercepting the byte
> at I/O port address 0x3c3.  Without this, the VGA BIOS can't bootstrap
> itself.  The vfio-vga-reset branch has a conditional replacement of
> this, which doesn't seem to work for everyone.  I believe the version I
> posted to qemu-devel yesterday is a better implementation of that quirk.
>
>> -------
>>
>> about qemu bridge
>> i tried to set up a bridge with the config but qemu always told me that
>> qemu-bridge-helper is not present.
>> all i found out that it propably got removed from the package because of
>> the lack of control over the tap
>> devices.
>> now my question how can i still bridge the vm into my network without
>> that helper?
> I don't know what qemu-bridge-helper is/was, but you're probably better
> off asking bridge questions in a separate thread instead of buried here.
> Thanks,
>
> Alex
>

-- 
Adiumentum GmbH
Gf. Martin Wolf
Banderbacherstraße 76
90513 Zirndorf

0911 / 9601470
mwolf@adiumentum.com