[libvirt] [PATCH 0/8] Various KVM PCI device assignment improvements

[Mark McLoughlin]

On Tue, 2009-08-18 at 12:23 +0200, Mirko Raasch wrote:
> Mark McLoughlin schrieb:
> > On Sun, 2009-08-16 at 22:13 +0200, Mirko Raasch wrote:
> >   
> >> ==8823== Invalid read of size 8
> >> ==8823==    at 0x420C04: qemuCheckPciHostDevice (qemu_driver.c:1337)
> >> ==8823==    by 0x4E48CA7: pciResetDevice (pci.c:606)
> >> ==8823==    by 0x4229D5: qemudShutdownVMDaemon (qemu_driver.c:1479)
> >> ==8823==    by 0x43150B: qemudStartup (qemu_driver.c:357)
> >>     
> >
> > Ah, qemudShutdownVMDaemon() is passed a NULL connection and then
> > qemuCheckPciHostDevice de-references the NULL pointer.
> >
> > I have a bunch of patches to re-work all this code which I will post
> > later. Hopefully they will work better for you.
> >
> > Thanks,
> > Mark.
> >
> >   
> Hi,
> 
> i tried your patches and now it works, no segfault anymore.

Thanks for testing.

> But the other problem is, that i tried a lot of pci device and passed 
> them through in the guest, but none
> of them were usable in the guest.
> 
> I tried three WLAN-PCI-cards, installed the driver in the guest, but i 
> had no success to get a signal from my WLAN-router.
> In the host, the cards works well, with same driver and same kernel.
> 
> The same problem do i have with my dvb-cards, in the host they works 
> very well, i can create a channels.conf with scanning the transponders,
> but in the guest, i dont get any signal.
> 
> I think, its more a problem of qemu then libvirt.

Does your machine have Intel VT-d support and your kernel have it
enabled? (e.g. dmesg | grep DMAR)

If so, it's probably best to supply details to the qemu-kvm mailing list
with lspci -vvv etc.

Cheers,
Mark.