[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
[PULL 06/21] vhost: Always store new kick fd on vhost_svq_set_svq_kick_f
From: |
Jason Wang |
Subject: |
[PULL 06/21] vhost: Always store new kick fd on vhost_svq_set_svq_kick_fd |
Date: |
Fri, 2 Sep 2022 14:41:23 +0800 |
From: Eugenio Pérez <eperezma@redhat.com>
We can unbind twice a file descriptor if we call twice
vhost_svq_set_svq_kick_fd because of this. Since it comes from vhost and
not from SVQ, that file descriptor could be a different thing that
guest's vhost notifier.
Likewise, it can happens the same if a guest start and stop the device
multiple times.
Reported-by: Lei Yang <leiyang@redhat.com>
Fixes: dff4426fa6 ("vhost: Add Shadow VirtQueue kick forwarding capabilities")
Signed-off-by: Eugenio Pérez <eperezma@redhat.com>
Acked-by: Jason Wang <jasowang@redhat.com>
Signed-off-by: Jason Wang <jasowang@redhat.com>
---
hw/virtio/vhost-shadow-virtqueue.c | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/hw/virtio/vhost-shadow-virtqueue.c
b/hw/virtio/vhost-shadow-virtqueue.c
index e495672..82a784d 100644
--- a/hw/virtio/vhost-shadow-virtqueue.c
+++ b/hw/virtio/vhost-shadow-virtqueue.c
@@ -602,13 +602,13 @@ void vhost_svq_set_svq_kick_fd(VhostShadowVirtqueue *svq,
int svq_kick_fd)
event_notifier_set_handler(svq_kick, NULL);
}
+ event_notifier_init_fd(svq_kick, svq_kick_fd);
/*
* event_notifier_set_handler already checks for guest's notifications if
* they arrive at the new file descriptor in the switch, so there is no
* need to explicitly check for them.
*/
if (poll_start) {
- event_notifier_init_fd(svq_kick, svq_kick_fd);
event_notifier_set(svq_kick);
event_notifier_set_handler(svq_kick, vhost_handle_guest_kick_notifier);
}
@@ -655,7 +655,7 @@ void vhost_svq_start(VhostShadowVirtqueue *svq,
VirtIODevice *vdev,
*/
void vhost_svq_stop(VhostShadowVirtqueue *svq)
{
- event_notifier_set_handler(&svq->svq_kick, NULL);
+ vhost_svq_set_svq_kick_fd(svq, VHOST_FILE_UNBIND);
g_autofree VirtQueueElement *next_avail_elem = NULL;
if (!svq->vq) {
--
2.7.4
- [PULL 00/21] Net patches, Jason Wang, 2022/09/02
- [PULL 01/21] vdpa: Skip the maps not in the iova tree, Jason Wang, 2022/09/02
- [PULL 03/21] util: accept iova_tree_remove_parameter by value, Jason Wang, 2022/09/02
- [PULL 02/21] vdpa: do not save failed dma maps in SVQ iova tree, Jason Wang, 2022/09/02
- [PULL 05/21] vdpa: Make SVQ vring unmapping return void, Jason Wang, 2022/09/02
- [PULL 06/21] vhost: Always store new kick fd on vhost_svq_set_svq_kick_fd,
Jason Wang <=
- [PULL 07/21] vdpa: Use ring hwaddr at vhost_vdpa_svq_unmap_ring, Jason Wang, 2022/09/02
- [PULL 08/21] vhost: stop transfer elem ownership in vhost_handle_guest_kick, Jason Wang, 2022/09/02
- [PULL 10/21] vhost: Delete useless read memory barrier, Jason Wang, 2022/09/02
- [PULL 12/21] vhost_net: Add NetClientInfo start callback, Jason Wang, 2022/09/02
- [PULL 09/21] vhost: use SVQ element ndescs instead of opaque data for desc validation, Jason Wang, 2022/09/02
- [PULL 11/21] vhost: Do not depend on !NULL VirtQueueElement on vhost_svq_flush, Jason Wang, 2022/09/02
- [PULL 13/21] vhost_net: Add NetClientInfo stop callback, Jason Wang, 2022/09/02
- [PULL 16/21] vdpa: extract vhost_vdpa_net_cvq_add from vhost_vdpa_net_handle_ctrl_avail, Jason Wang, 2022/09/02
- [PULL 15/21] vdpa: Move command buffers map to start of net device, Jason Wang, 2022/09/02
- [PULL 14/21] vdpa: add net_vhost_vdpa_cvq_info NetClientInfo, Jason Wang, 2022/09/02