Re: [PATCH] virtio: make virtio_add_queue() queue_size an unsigned int

On Thu, 30 Jul 2026 at 16:36, Stefan Hajnoczi <[email protected]> wrote:
>
> virtio_add_queue()'s queue_size argument is a signed int. Coverity is
> unhappy when that type is used as an argument to g_new0():
>
>   *** CID 1664271:         Error handling issues  (NEGATIVE_RETURNS)
>   /builds/qemu-project/qemu/hw/virtio/virtio.c: 2595             in virtio_add_queue()
>   2589         }
>   2590
>   2591         vdev->vq[i].vring.num = queue_size;
>   2592         vdev->vq[i].vring.num_default = queue_size;
>   2593         vdev->vq[i].vring.align = VIRTIO_PCI_VRING_ALIGN;
>   2594         vdev->vq[i].handle_output = handle_output;
>   >>>     CID 1664271:         Error handling issues  (NEGATIVE_RETURNS)
>   >>>     "__n" is passed to a parameter that cannot be negative.
>   2595         vdev->vq[i].used_elems = g_new0(VirtQueueElement, queue_size);

Though note that the reason Coverity thinks this value can be negative
is not because the function might be passed a negative value, but
because it incorrectly thinks that
        int override = object_property_get_int(OBJECT(qbus->parent),
                                               VIRTIO_QUEUE_SIZE_OVERRIDE,
                                               &error_abort);

can return a negative number. (It can't because of the error_abort
and because the property is a UINT16 one.)

It might be a good idea to enforce the VIRTQUEUE_MAX_SIZE bounds
check on the queue size we get from the override property. Since
I think that can be set by the end-user (though probably it won't
be in practice) ideally we would not abort() on a bad property
value, but that's awkward to do at this point...

thanks
-- PMM