[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Xen-devel] Inconsistent use of libxl__device_pci_reset() when adding/removing all functions of a device



On Tue, Feb 04, 2014 at 07:19:16PM +0100, Martin Öhrling wrote:
> I tried to use vga passthrough with an AMD card and ran into the issues
> with dom0 crash on domU reboot. My intention is to debug the issue and I
> have started to read up on the code for pci passthrough. The following
> observations may not be an error but perhaps someone could shed some light
> over the design intentions.
> 
> My first impression was that libxl__device_pci_reset() is a function level
> reset function. It is called each time a single function of a device is
> added or removed. A device reset would break other domU:s if other
> functions of the device are passed through to these domU:s.
> 
> The strange thing is that there is only a single libxl__device_pci_reset()
> call when pcidev->vfunc_mask is set to LIBXL_PCI_FUNC_ALL. I'm getting the
> impression that the function is assumed to be a device reset function.
> 
> Is the attempt to reset a function, when all other functions belongs to
> pciback, detected and replaced by a device reset?

Yes with these patches:
https://git.kernel.org/cgit/linux/kernel/git/konrad/xen.git/log/?h=devel/xen-pciback.slot_and_bus.v0

But the last one seems to hang pciback when the device is unbound.

> 
> Best Regards,
> Martin

> _______________________________________________
> Xen-devel mailing list
> Xen-devel@xxxxxxxxxxxxx
> http://lists.xen.org/xen-devel


_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxx
http://lists.xen.org/xen-devel


 


Rackspace

Lists.xenproject.org is hosted with RackSpace, monitoring our
servers 24x7x365 and backed by RackSpace's Fanatical Support®.