Once VQs are filled with empty buffers and we kick the host, it can send
connection requests. If the_virtio_vsock is not initialized before,
replies are silently dropped and do not reach the host.
virtio_transport_send_pkt() can queue packets once the_virtio_vsock is
set, but they won't be processed until vsock->tx_run is set to true. We
queue vsock->send_pkt_work when initialization finishes to send those
packets queued earlier.
Fixes: 0deab087b16a ("vsock/virtio: use RCU to avoid use-after-free on the_virtio_vsock")
Signed-off-by: Alexandru Matei <[email protected]>
---
v4:
- moved queue_work for send_pkt_work in vqs_start and added comment explaining why
v3:
- renamed vqs_fill to vqs_start and moved tx_run initialization to it
- queued send_pkt_work at the end of initialization to send packets queued earlier
v2:
- split virtio_vsock_vqs_init in vqs_init and vqs_fill and moved
the_virtio_vsock initialization after vqs_init
net/vmw_vsock/virtio_transport.c | 18 +++++++++++++++++-
1 file changed, 17 insertions(+), 1 deletion(-)
diff --git a/net/vmw_vsock/virtio_transport.c b/net/vmw_vsock/virtio_transport.c
index e95df847176b..b80bf681327b 100644
--- a/net/vmw_vsock/virtio_transport.c
+++ b/net/vmw_vsock/virtio_transport.c
@@ -555,6 +555,11 @@ static int virtio_vsock_vqs_init(struct virtio_vsock *vsock)
virtio_device_ready(vdev);
+ return 0;
+}
+
+static void virtio_vsock_vqs_start(struct virtio_vsock *vsock)
+{
mutex_lock(&vsock->tx_lock);
vsock->tx_run = true;
mutex_unlock(&vsock->tx_lock);
@@ -569,7 +574,16 @@ static int virtio_vsock_vqs_init(struct virtio_vsock *vsock)
vsock->event_run = true;
mutex_unlock(&vsock->event_lock);
- return 0;
+ /* virtio_transport_send_pkt() can queue packets once
+ * the_virtio_vsock is set, but they won't be processed until
+ * vsock->tx_run is set to true. We queue vsock->send_pkt_work
+ * when initialization finishes to send those packets queued
+ * earlier.
+ * We don't need to queue the other workers (rx, event) because
+ * as long as we don't fill the queues with empty buffers, the
+ * host can't send us any notification.
+ */
+ queue_work(virtio_vsock_workqueue, &vsock->send_pkt_work);
}
static void virtio_vsock_vqs_del(struct virtio_vsock *vsock)
@@ -664,6 +678,7 @@ static int virtio_vsock_probe(struct virtio_device *vdev)
goto out;
rcu_assign_pointer(the_virtio_vsock, vsock);
+ virtio_vsock_vqs_start(vsock);
mutex_unlock(&the_virtio_vsock_mutex);
@@ -736,6 +751,7 @@ static int virtio_vsock_restore(struct virtio_device *vdev)
goto out;
rcu_assign_pointer(the_virtio_vsock, vsock);
+ virtio_vsock_vqs_start(vsock);
out:
mutex_unlock(&the_virtio_vsock_mutex);
--
2.25.1
On Tue, Oct 24, 2023 at 10:17:42PM +0300, Alexandru Matei wrote:
>Once VQs are filled with empty buffers and we kick the host, it can send
>connection requests. If the_virtio_vsock is not initialized before,
>replies are silently dropped and do not reach the host.
>
>virtio_transport_send_pkt() can queue packets once the_virtio_vsock is
>set, but they won't be processed until vsock->tx_run is set to true. We
>queue vsock->send_pkt_work when initialization finishes to send those
>packets queued earlier.
>
>Fixes: 0deab087b16a ("vsock/virtio: use RCU to avoid use-after-free on the_virtio_vsock")
>Signed-off-by: Alexandru Matei <[email protected]>
>---
>v4:
>- moved queue_work for send_pkt_work in vqs_start and added comment explaining why
>v3:
>- renamed vqs_fill to vqs_start and moved tx_run initialization to it
>- queued send_pkt_work at the end of initialization to send packets queued earlier
>v2:
>- split virtio_vsock_vqs_init in vqs_init and vqs_fill and moved
> the_virtio_vsock initialization after vqs_init
>
> net/vmw_vsock/virtio_transport.c | 18 +++++++++++++++++-
> 1 file changed, 17 insertions(+), 1 deletion(-)
LGTM!
Reviewed-by: Stefano Garzarella <[email protected]>
Hello:
This patch was applied to netdev/net.git (main)
by Jakub Kicinski <[email protected]>:
On Tue, 24 Oct 2023 22:17:42 +0300 you wrote:
> Once VQs are filled with empty buffers and we kick the host, it can send
> connection requests. If the_virtio_vsock is not initialized before,
> replies are silently dropped and do not reach the host.
>
> virtio_transport_send_pkt() can queue packets once the_virtio_vsock is
> set, but they won't be processed until vsock->tx_run is set to true. We
> queue vsock->send_pkt_work when initialization finishes to send those
> packets queued earlier.
>
> [...]
Here is the summary with links:
- [v4] vsock/virtio: initialize the_virtio_vsock before using VQs
https://git.kernel.org/netdev/net/c/53b08c498515
You are awesome, thank you!
--
Deet-doot-dot, I am a bot.
https://korg.docs.kernel.org/patchwork/pwbot.html