On Mon, Jan 12, 2026 at 08:35:37AM +0100, Eugenio Perez Martin wrote:
> On Sun, Jan 11, 2026 at 12:44 AM Michael S. Tsirkin <[email protected]> wrote:
> >
> > On Fri, Jan 09, 2026 at 04:24:20PM +0100, Eugenio Pérez wrote:
> > > @@ -1252,12 +1266,24 @@ static long vduse_dev_ioctl(struct file *file,
> > > unsigned int cmd,
> > > if (config.index >= dev->vq_num)
> > > break;
> > >
> > > - if (!is_mem_zero((const char *)config.reserved,
> > > - sizeof(config.reserved)))
> > > + if (dev->api_version < VDUSE_API_VERSION_1 && config.group)
> > > + break;
>
> (Bookmarking the piece of code above as [1] to reference later)
>
> > > +
> > > + if (dev->api_version >= VDUSE_API_VERSION_1) {
> > > + if (config.group >= dev->ngroups)
> > > + break;
> > > + if (dev->status & VIRTIO_CONFIG_S_DRIVER_OK)
> > > + break;
> > > + }
> > > +
> > > + if (config.reserved1 ||
> > > + !is_mem_zero((const char *)config.reserved2,
> > > + sizeof(config.reserved2)))
> >
> > Hmm but if api version is 0 then group should be 0 no?
> > We should validate.
> >
>
> The check (dev->api_version < VDUSE_API_VERSION_1 && config.group) is
> above this check in this set of changes [1], am I missing something?
> Would you prefer it to be reordered here or written differently?
Oh you are right. It's just not very clear that everything is covered.
if (dev->api_version < VDUSE_API_VERSION_1) {
if (config.group)
....
} else {
....
}
would be clearer.
BTW I don't really like this idiom of "break to return".
Just return -EINVAL would be more explicit.
But this is the way current code handles it, so I'm not demanding
it is changed as part of this patchset.
--
MST
