Commit 82dfb540aeb2 ("VSOCK: Add virtio vsock vsockmon hooks") added
virtio_transport_deliver_tap_pkt() for handing packets to the
vsockmon device. However, in virtio_transport_send_pkt_work(),
the function is called before actually sending the packet (i.e.
before placing it in the virtqueue with virtqueue_add_sgs() and checking
whether it returned successfully).
Queuing the packet in the virtqueue can fail even multiple times.
However, in virtio_transport_deliver_tap_pkt() we deliver the packet
to the monitoring tap interface only the first time we call it.
This certainly avoids seeing the same packet replicated multiple times
in the monitoring interface, but it can show the packet sent with the
wrong timestamp or even before we succeed to queue it in the virtqueue.
Move virtio_transport_deliver_tap_pkt() after calling virtqueue_add_sgs()
and making sure it returned successfully.
Fixes: 82dfb540aeb2 ("VSOCK: Add virtio vsock vsockmon hooks")
Cc: [email protected]
Signed-off-by: Marco Pinna <[email protected]>
---
net/vmw_vsock/virtio_transport.c | 3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)
diff --git a/net/vmw_vsock/virtio_transport.c b/net/vmw_vsock/virtio_transport.c
index 1748268e0694..ee5d306a96d0 100644
--- a/net/vmw_vsock/virtio_transport.c
+++ b/net/vmw_vsock/virtio_transport.c
@@ -120,7 +120,6 @@ virtio_transport_send_pkt_work(struct work_struct *work)
if (!skb)
break;
- virtio_transport_deliver_tap_pkt(skb);
reply = virtio_vsock_skb_reply(skb);
sgs = vsock->out_sgs;
sg_init_one(sgs[out_sg], virtio_vsock_hdr(skb),
@@ -170,6 +169,8 @@ virtio_transport_send_pkt_work(struct work_struct *work)
break;
}
+ virtio_transport_deliver_tap_pkt(skb);
+
if (reply) {
struct virtqueue *rx_vq = vsock->vqs[VSOCK_VQ_RX];
int val;
--
2.44.0
On Fri, Mar 29, 2024 at 05:12:59PM +0100, Marco Pinna wrote:
>Commit 82dfb540aeb2 ("VSOCK: Add virtio vsock vsockmon hooks") added
>virtio_transport_deliver_tap_pkt() for handing packets to the
>vsockmon device. However, in virtio_transport_send_pkt_work(),
>the function is called before actually sending the packet (i.e.
>before placing it in the virtqueue with virtqueue_add_sgs() and checking
>whether it returned successfully).
>Queuing the packet in the virtqueue can fail even multiple times.
>However, in virtio_transport_deliver_tap_pkt() we deliver the packet
>to the monitoring tap interface only the first time we call it.
>This certainly avoids seeing the same packet replicated multiple times
>in the monitoring interface, but it can show the packet sent with the
>wrong timestamp or even before we succeed to queue it in the virtqueue.
>
>Move virtio_transport_deliver_tap_pkt() after calling virtqueue_add_sgs()
>and making sure it returned successfully.
>
>Fixes: 82dfb540aeb2 ("VSOCK: Add virtio vsock vsockmon hooks")
>Cc: [email protected]
>Signed-off-by: Marco Pinna <[email protected]>
>---
> net/vmw_vsock/virtio_transport.c | 3 ++-
> 1 file changed, 2 insertions(+), 1 deletion(-)
Reviewed-by: Stefano Garzarella <[email protected]>
>
>diff --git a/net/vmw_vsock/virtio_transport.c b/net/vmw_vsock/virtio_transport.c
>index 1748268e0694..ee5d306a96d0 100644
>--- a/net/vmw_vsock/virtio_transport.c
>+++ b/net/vmw_vsock/virtio_transport.c
>@@ -120,7 +120,6 @@ virtio_transport_send_pkt_work(struct work_struct *work)
> if (!skb)
> break;
>
>- virtio_transport_deliver_tap_pkt(skb);
> reply = virtio_vsock_skb_reply(skb);
> sgs = vsock->out_sgs;
> sg_init_one(sgs[out_sg], virtio_vsock_hdr(skb),
>@@ -170,6 +169,8 @@ virtio_transport_send_pkt_work(struct work_struct *work)
> break;
> }
>
>+ virtio_transport_deliver_tap_pkt(skb);
>+
> if (reply) {
> struct virtqueue *rx_vq = vsock->vqs[VSOCK_VQ_RX];
> int val;
>--
>2.44.0
>
Hello:
This patch was applied to netdev/net.git (main)
by Jakub Kicinski <[email protected]>:
On Fri, 29 Mar 2024 17:12:59 +0100 you wrote:
> Commit 82dfb540aeb2 ("VSOCK: Add virtio vsock vsockmon hooks") added
> virtio_transport_deliver_tap_pkt() for handing packets to the
> vsockmon device. However, in virtio_transport_send_pkt_work(),
> the function is called before actually sending the packet (i.e.
> before placing it in the virtqueue with virtqueue_add_sgs() and checking
> whether it returned successfully).
> Queuing the packet in the virtqueue can fail even multiple times.
> However, in virtio_transport_deliver_tap_pkt() we deliver the packet
> to the monitoring tap interface only the first time we call it.
> This certainly avoids seeing the same packet replicated multiple times
> in the monitoring interface, but it can show the packet sent with the
> wrong timestamp or even before we succeed to queue it in the virtqueue.
>
> [...]
Here is the summary with links:
- [net,v2] vsock/virtio: fix packet delivery to tap device
https://git.kernel.org/netdev/net/c/b32a09ea7c38
You are awesome, thank you!
--
Deet-doot-dot, I am a bot.
https://korg.docs.kernel.org/patchwork/pwbot.html