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

Re: [Xen-devel] xl/SR-IOV: disposition of VFs when PF disappears?



>>> On 28.10.14 at 15:00, <konrad.wilk@xxxxxxxxxx> wrote:
> On Tue, Oct 28, 2014 at 10:25:37AM +0000, Jan Beulich wrote:
>> >>> On 27.10.14 at 19:07, <konrad.wilk@xxxxxxxxxx> wrote:
>> > 2). PF removal. Currently if you try to unload the PF and the VFs
>> >    are in use (pciback owns them), the unloading will not happen. Until
>> >    all of the VFs have been de-assigned.
>> >    Is the "bug" here that the reporter (Intel?) wants the VFs to be
>> >    automatically yanked out of a guest when the system admin wants to
>> >    unload the PF?
>> 
>> No. They see the VF still listed as assignable after the guest terminated
>> and the PF got made assignable. The PF driver unloading, however,
>> should trigger all VFs (which are about to disappear) to get removed
>> from pciback. I hope to find some time later today to look into why (if
>> at all) this isn't happening right now.
> 
> That is due to PCI_DEV_FLAGS_ASSIGNED being set on the PF device which
> inhibits it from unloading until that flag has been cleared.
> 
> That is odd that Intel is reporting it to you as Intel added that
> flag as part of the SR-IOV NICs to inhibit the PF unloading when VFs
> are still in-use.

That flag doesn't inhibit the PF driver unloading. It only suppresses
it freeing the VFs in that process (and afaics nothing will ever allow
them to be released subsequently in that case). I'm about to test a
change to pciback (it's building right now).

Jan


_______________________________________________
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®.