Re: [Qemu-devel] Isuue assiging devices using VFIO on x86

[Alex Williamson]

On Wed, 2012-08-29 at 07:11 +0000, Bhushan Bharat-R65777 wrote:
> 
> > -----Original Message-----
> > From: Alex Williamson [mailto:address@hidden
> > Sent: Wednesday, August 29, 2012 12:16 PM
> > To: Bhushan Bharat-R65777
> > Cc: address@hidden; Avi Kivity; address@hidden
> > Subject: Re: Isuue assiging devices using VFIO on x86
> > 
> > On Tue, 2012-08-28 at 17:10 +0000, Bhushan Bharat-R65777 wrote:
> > >
> > > > -----Original Message-----
> > > > From: Alex Williamson [mailto:address@hidden
> > > > Sent: Tuesday, August 28, 2012 9:27 PM
> > > > To: Bhushan Bharat-R65777
> > > > Cc: address@hidden; Avi Kivity; address@hidden
> > > > Subject: Re: Isuue assiging devices using VFIO on x86
> > > >
> > > > On Tue, 2012-08-28 at 09:23 +0000, Bhushan Bharat-R65777 wrote:
> > > > > Hi Alex,
> > > > >
> > > > > In my susyem I have following devices:
> > > > >
> > > > > I tried assigning a following PCI devices:
> > > > > 00:03.0 Communication controller: Intel Corporation 4 Series
> > > > > Chipset HECI
> > > > Controller (rev 03)
> > > > > 00:03.2 IDE interface: Intel Corporation 4 Series Chipset PT IDER
> > > > > Controller
> > > > (rev 03)
> > > > > 00:03.3 Serial controller: Intel Corporation 4 Series Chipset
> > > > > Serial KT
> > > > Controller (rev 03)
> Hello Alex,
> I am using " ./configure --target-list=x86_64-softmmu --enable-kvm " .
> QEMU repo is at:
> commit 3b4672f6614e07f9dc0a85a12a8c89d480a2493c
> Author: Alex Williamson <address@hidden>
> Date:   Tue Aug 14 14:07:19 2012 -0600
> 
>     vfio: Enable vfio-pci and mark supported
>     
>     Signed-off-by: Alex Williamson <address@hidden>
> 
> Linux Repo is at :
> commit d9875690d9b89a866022ff49e3fcea892345ad92
> Author: Linus Torvalds <address@hidden>
> Date:   Thu Aug 16 14:51:24 2012 -0700
> 
>     Linux 3.6-rc2
> 
> ---------------------------
> 
> Then I bound 3 pci devices with VFIO and run below command to launch VM
> qemu-system-x86_64 -enable-kvm  -nographic  -kernel /boot/vmlinuz-3.6.0-rc2+ 
> -initrd /boot/initramfs-3.6.0-rc2+.img -append "root=/dev/sda1" -m 1024 
> -drive file=/home/kvmdev/debian_squeeze_amd64_standard.qcow2 -device 
> vfio-pci,host=0000:00:03.0 -device vfio-pci,host=0000:00:03.2 -device 
> vfio-pci,host=0000:00:03.3
> 
> Then I was getting below errors: 
> qemu-system-x86_64: -device vfio-pci,host=0000:00:03.0: vfio: failed to set 
> iommu for container: Operation not permitted
> 
> qemu-system-x86_64: -device vfio-pci,host=0000:00:03.0: vfio: failed to setup 
> container for group 2
> 
> qemu-system-x86_64: -device vfio-pci,host=0000:00:03.0: vfio: failed to get 
> group 2
> 
> qemu-system-x86_64: -device vfio-pci,host=0000:00:03.0: Device 'vfio-pci' 
> could not be initialized
> ---------
> 
> Then I set /sys/module/vfio_iommu_type1/parameters/allow_unsafe_interrupts = 1
> 
> And getting below error:
> 
> qemu-system-x86_64: -device vfio-pci,host=0000:00:03.0: Warning, device 
> 0000:00:03.0 does not support reset
> 
> qemu-system-x86_64: -device vfio-pci,host=0000:00:03.2: Warning, device 
> 0000:00:03.2 does not support reset
> 
> qemu-system-x86_64: -device vfio-pci,host=0000:00:03.3: Warning, device 
> 0000:00:03.3 does not support reset
> 
> qemu: hardware error: register_ioport_read: invalid opaque for address 0x3f6
> CPU #0:
> EAX=80000003 EBX=3ffe0e80 ECX=80000003 EDX=00000cfc
> ESI=00002800 EDI=80002804 EBP=00006fc0 ESP=00006f38
> EIP=3ffec005 EFL=00000086 [--S--P-] CPL=0 II=0 A20=1 SMM=0 HLT=0
> ES =0010 00000000 ffffffff 00c09300 DPL=0 DS   [-WA]
> CS =0008 00000000 ffffffff 00c09b00 DPL=0 CS32 [-RA]
> SS =0010 00000000 ffffffff 00c09300 DPL=0 DS   [-WA]
> DS =0010 00000000 ffffffff 00c09300 DPL=0 DS   [-WA]
> FS =0010 00000000 ffffffff 00c09300 DPL=0 DS   [-WA]
> GS =0010 00000000 ffffffff 00c09300 DPL=0 DS   [-WA]
> LDT=0000 00000000 0000ffff 00008200 DPL=0 LDT
> TR =0000 00000000 0000ffff 00008b00 DPL=0 TSS32-busy
> GDT=     000fcd68 00000037
> IDT=     000fdb60 00000000
> CR0=00000011 CR2=00000000 CR3=00000000 CR4=00000000
> DR0=0000000000000000 DR1=0000000000000000 DR2=0000000000000000 
> DR3=0000000000000000 
> DR6=00000000ffff0ff0 DR7=0000000000000400
> EFER=0000000000000000
> FCW=037f FSW=0000 [ST=0] FTW=00 MXCSR=00001f80
> FPR0=0000000000000000 0000 FPR1=0000000000000000 0000
> FPR2=0000000000000000 0000 FPR3=0000000000000000 0000
> FPR4=0000000000000000 0000 FPR5=0000000000000000 0000
> FPR6=0000000000000000 0000 FPR7=0000000000000000 0000
> XMM00=00000000000000000000000000000000 XMM01=00000000000000000000000000000000
> XMM02=00000000000000000000000000000000 XMM03=00000000000000000000000000000000
> XMM04=00000000000000000000000000000000 XMM05=00000000000000000000000000000000
> XMM06=00000000000000000000000000000000 XMM07=00000000000000000000000000000000
> Aborted (core dumped)
> 
> Thanks a lot for your help

You're assigning a device that's getting programmed to to overlap ioport
space with another device.  Can I ask why you're trying to assigning
rather sensitive chipset components to a guest?  I can see you might
want to assign the IDE controller, but the serial KT controller and HECI
controller really have no business being exposed to the guest.  There
are certainly going to be combinations of assigning devices that don't
make sense, but I don't know how to generically prevent them.  Those
kind of sanity checks are probably better left for a libvirt interfaces
than for qemu itself.  What happens if you only expose 3.2?  Thanks,

Alex