Re: [PULL 1/5] hw/pcie-root-port: Fix hotplug for PCI devices requiring IO

[Daniel P . Berrangé]

On Mon, Sep 27, 2021 at 05:49:15AM -0400, Michael S. Tsirkin wrote:
> On Mon, Sep 27, 2021 at 10:33:42AM +0100, Daniel P. Berrangé wrote:
> > On Tue, Aug 03, 2021 at 04:52:03PM -0400, Michael S. Tsirkin wrote:
> > > From: Marcel Apfelbaum <marcel.apfelbaum@gmail.com>
> > > 
> > > Q35 has now ACPI hotplug enabled by default for PCI(e) devices.
> > > As opposed to native PCIe hotplug, guests like Fedora 34
> > > will not assign IO range to pcie-root-ports not supporting
> > > native hotplug, resulting into a regression.
> > > 
> > > Reproduce by:
> > >     qemu-bin -M q35 -device pcie-root-port,id=p1 -monitor stdio
> > >     device_add e1000,bus=p1
> > > In the Guest OS the respective pcie-root-port will have the IO range
> > > disabled.
> > > 
> > > Fix it by setting the "reserve-io" hint capability of the
> > > pcie-root-ports so the firmware will allocate the IO range instead.
> > > 
> > > Acked-by: Igor Mammedov <imammedo@redhat.com>
> > > Signed-off-by: Marcel Apfelbaum <marcel@redhat.com>
> > > Message-Id: <20210802090057.1709775-1-marcel@redhat.com>
> > > Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
> > > Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
> > > ---
> > >  hw/pci-bridge/gen_pcie_root_port.c | 5 +++++
> > >  1 file changed, 5 insertions(+)
> > 
> > This change, when combined with the switch to ACPI based hotplug by
> > default, is responsible for a significant regression in QEMU 6.1.0
> > 
> > It is no longer possible to have more than 15 pcie-root-port devices
> > added to a q35 VM in 6.1.0.  Before this I've had as many as 80+ devices
> > present before I stopped trying to add more.
> > 
> >   https://gitlab.com/qemu-project/qemu/-/issues/641
> > 
> > This regression is significant, because it has broken the out of the
> > box default configuration that OpenStack uses for booting all VMs.
> > They add 16 pcie-root-ports by defalt to allow empty slots for device
> > hotplug under q35 [1].
> 
> 
> Indeed, oops. Thanks for the report!

We're at soft freeze now and this is still broken in git master.
I don't recall seeing a fix for this problem on list and no one has
commented on the bug report.

Is anyone actively working on a fix for this release ?

If not, I'm going to post a patch to revert to PCI native
hotplug, because this was a significant regression in 6.1 that
breaks openstack out of the box and we can't leave it broken
again for 6.2.

Regards,
Daniel
-- 
|: https://berrange.com      -o-    https://www.flickr.com/photos/dberrange :|
|: https://libvirt.org         -o-            https://fstop138.berrange.com :|
|: https://entangle-photo.org    -o-    https://www.instagram.com/dberrange :|