Re: [Qemu-devel] [qemu-s390x] [PATCH v2] s390: avoid potential null dereference in s390_pcihost_unplug()

[Halil Pasic <pasic@linux.ibm.com>]

On Fri, 4 Jan 2019 15:10:05 +0100
Cornelia Huck <cohuck@redhat.com> wrote:

> On Thu,  3 Jan 2019 07:16:12 -0800
> Li Qiang <liq3ea@163.com> wrote:
> 
> > When getting the 'pbdev', the if...else has no default branch.
> > From Coverity, the 'pbdev' maybe null when the 'dev' is not
> > the TYPE_PCI_BRIDGE/TYPE_PCI_DEVICE/TYPE_S390_PCI_DEVICE.
> > This patch adds a default branch for device plug and unplug.
> > 
> > Spotted by Coverity: CID 1398593
> > 
> > Signed-off-by: Li Qiang <liq3ea@163.com>
> > ---
> > Adds a default branch for device plug per Cornelia's review.
> > 
> >  hw/s390x/s390-pci-bus.c | 7 +++++++
> >  1 file changed, 7 insertions(+)
> > 
> > diff --git a/hw/s390x/s390-pci-bus.c b/hw/s390x/s390-pci-bus.c
> > index 15759b6514..fe48a36ff6 100644
> > --- a/hw/s390x/s390-pci-bus.c
> > +++ b/hw/s390x/s390-pci-bus.c
> > @@ -912,6 +912,9 @@ static void s390_pcihost_plug(HotplugHandler *hotplug_dev, DeviceState *dev,
> >          pbdev->fh = pbdev->idx;
> >          QTAILQ_INSERT_TAIL(&s->zpci_devs, pbdev, link);
> >          g_hash_table_insert(s->zpci_table, &pbdev->idx, pbdev);
> > +    } else {
> > +        error_setg(errp, "s390: device plug request for not supported device"
> > +                   " type: %s", object_get_typename(OBJECT(dev)));
> 
> Maybe make this "s390/pci: plugging device type <%s> is not supported"?
> 

Under what circumstances could/does this happen? I mean how can this
be triggered by the user?

Regards,
Halil

> >      }
> >  }
> >  
> > @@ -956,6 +959,10 @@ static void s390_pcihost_unplug(HotplugHandler *hotplug_dev, DeviceState *dev,
> >      } else if (object_dynamic_cast(OBJECT(dev), TYPE_S390_PCI_DEVICE)) {
> >          pbdev = S390_PCI_DEVICE(dev);
> >          pci_dev = pbdev->pdev;
> > +    } else {
> > +        error_setg(errp, "s390: device unplug request for not supported device"
> > +                   " type: %s", object_get_typename(OBJECT(dev)));
> 
> Halil has a point when he suggests an assert here. If we fence the
> device type in the plug function, I can't think of a way we would end
> up here.
> 
> > +        return;
> >      }
> >  
> >      switch (pbdev->state) {
> 
>