Re: [Qemu-devel] [PATCH for-2.10 0/3] qdev/vfio: defer DEVICE_DEL to avoid races with libvirt

[Michael Roth]

Quoting David Gibson (2017-07-27 05:53:48)
> On Thu, Jul 27, 2017 at 10:11:48AM +0100, Peter Maydell wrote:
> > On 27 July 2017 at 02:30, Michael Roth <address@hidden> wrote:
> > > In particular, Mellanox CX4 adapters on PowerNV hosts might not be fully
> > > quiesced by vfio-pci's finalize() routine until up to 6s after the
> > > DEVICE_DELETED was emitted, leading to detach-device on the libvirt side 
> > > pretty
> > > much always crashing the host.
> > 
> > My initial naive thought is that if the host kernel can crash then
> > this is a host kernel bug... shouldn't the host kernel refuse
> > the subsequent libvirt rebind if it would cause a crash ?
> 
> I think so too, but I haven't been able to convince Alex.  Nor
> find time to fix it in the kernel myself.

In the thread I linked to Alex had mentioned he was pursuing something on the
kernel side, but my understanding what that we'd simply have the kernel fail
more gracefully when attempting to rebind in this situation.

But that still leaves the matter of libvirt failing to rebind the device to
the host. This series addresses that aspect of it, so I think the 2 approaches
are complementary.

> 
> -- 
> David Gibson                    | I'll have my music baroque, and my code
> david AT gibson.dropbear.id.au  | minimalist, thank you.  NOT _the_ _other_
>                                 | _way_ _around_!
> http://www.ozlabs.org/~dgibson