On Fri, 17 Mar 2017 13:18:16 +0100
Andrew Jones <drjo...@redhat.com> wrote:
> On Fri, Mar 17, 2017 at 07:33:53PM +0800, Fam Zheng wrote:
> > For one thing we shouldn't continue if an error happened, for the other
> > two steps failing can cause an abort() in error_setg because we reuse
> > the same errp blindly.
> >
> > Add error handling checks to fix both issues.
> >
> > Signed-off-by: Fam Zheng <f...@redhat.com>
> > ---
> > hw/virtio/virtio-bus.c | 18 ++++++++++++++++--
> > 1 file changed, 16 insertions(+), 2 deletions(-)
> >
> > if (klass->device_plugged != NULL) {
> > klass->device_plugged(qbus->parent, errp);
>
> Did you intend to change this errp to &local_err, as well? If not,
> then the body of the below hunk is unreachable, as local_err can
> never be non-null there.
Oh, I managed to overlook this.
So my r-b applies if this is changed to use local_err :)
>
> Thanks,
> drew
>
> > }
> > + if (local_err) {
> > + error_propagate(errp, local_err);
> > + return;
> > + }
> >
> > if (klass->get_dma_as != NULL && has_iommu) {
> > virtio_add_feature(&vdev->host_features, VIRTIO_F_IOMMU_PLATFORM);
> > --
> > 2.9.3
> >
> >
>
>
