Il 14/01/2014 15:24, Eric Farman ha scritto:
> In some cases, an unplug can cause events to be dropped, which
> leads to an assertion failure when preparing to notify the guest
> kernel. This merely accommodates both variations of the "no event"
> value that could occur in this codepath.
>
> Signed-off-by: Eric Farman <far...@linux.vnet.ibm.com>
> ---
> hw/scsi/virtio-scsi.c | 3 ++-
> 1 file changed, 2 insertions(+), 1 deletion(-)
>
> diff --git a/hw/scsi/virtio-scsi.c b/hw/scsi/virtio-scsi.c
> index 49a9576..f8e3632 100644
> --- a/hw/scsi/virtio-scsi.c
> +++ b/hw/scsi/virtio-scsi.c
> @@ -521,7 +521,8 @@ static void virtio_scsi_push_event(VirtIOSCSI *s,
> SCSIDevice *dev,
> evt->event = event;
> evt->reason = reason;
> if (!dev) {
> - assert(event == VIRTIO_SCSI_T_NO_EVENT);
> + assert(event == VIRTIO_SCSI_T_NO_EVENT ||
> + event == VIRTIO_SCSI_T_EVENTS_MISSED);
> } else {
> evt->lun[0] = 1;
> evt->lun[1] = dev->id;
>
Ugh, you're right. In fact, you can only have event ==
VIRTIO_SCSI_T_EVENTS_MISSED at this point. Can you make v2 that
completely replaces the condition?
Paolo
