[PATCH net-next 1/2] openvswitch: Fix double reporting of drops in dropwatch

[PATCH net-next 1/2] openvswitch: Fix double reporting of drops in dropwatch

[Mike Pattrick]

Frames sent to userspace can be reported as dropped in
ovs_dp_process_packet, however, if they are dropped in the netlink code
then netlink_attachskb will report the same frame as dropped.

This patch checks for error codes which indicate that the frame has
already been freed.

Bugzilla: https://bugzilla.redhat.com/show_bug.cgi?id=2109946
Signed-off-by: Mike Pattrick <mkp@redhat.com>
---
 net/openvswitch/datapath.c | 16 +++++++++++++---
 1 file changed, 13 insertions(+), 3 deletions(-)

diff --git a/net/openvswitch/datapath.c b/net/openvswitch/datapath.c
index 7e8a39a35627..029f9c3e1c28 100644
--- a/net/openvswitch/datapath.c
+++ b/net/openvswitch/datapath.c
@@ -252,10 +252,20 @@ void ovs_dp_process_packet(struct sk_buff *skb, struct sw_flow_key *key)
 
 		upcall.mru = OVS_CB(skb)->mru;
 		error = ovs_dp_upcall(dp, skb, key, &upcall, 0);
-		if (unlikely(error))
-			kfree_skb(skb);
-		else
+		switch (error) {
+		case 0:
+			fallthrough;
+		case -EAGAIN:
+			fallthrough;
+		case -ERESTARTSYS:
+			fallthrough;
+		case -EINTR:
 			consume_skb(skb);
+			break;
+		default:
+			kfree_skb(skb);
+			break;
+		}
 		stats_counter = &stats->n_missed;
 		goto out;
 	}
-- 
2.31.1

[PATCH net-next 2/2] openvswitch: Fix overreporting of drops in dropwatch

[Mike Pattrick]

Currently queue_userspace_packet will call kfree_skb for all frames,
whether or not an error occurred. This can result in a single dropped
frame being reported as multiple drops in dropwatch. This patch will
consume the skbs instead.

Bugzilla: https://bugzilla.redhat.com/show_bug.cgi?id=2109957
Signed-off-by: Mike Pattrick <mkp@redhat.com>
---
 net/openvswitch/datapath.c | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)

diff --git a/net/openvswitch/datapath.c b/net/openvswitch/datapath.c
index 029f9c3e1c28..3eee4b0a2005 100644
--- a/net/openvswitch/datapath.c
+++ b/net/openvswitch/datapath.c
@@ -561,8 +561,9 @@ static int queue_userspace_packet(struct datapath *dp, struct sk_buff *skb,
 out:
 	if (err)
 		skb_tx_error(skb);
-	kfree_skb(user_skb);
-	kfree_skb(nskb);
+	consume_skb(user_skb);
+	consume_skb(nskb);
+
 	return err;
 }
 
-- 
2.31.1

Re: [PATCH net-next 1/2] openvswitch: Fix double reporting of drops in dropwatch

[Jakub Kicinski]

On Thu, 28 Jul 2022 12:12:58 -0400 Mike Pattrick wrote:
> Frames sent to userspace can be reported as dropped in
> ovs_dp_process_packet, however, if they are dropped in the netlink code
> then netlink_attachskb will report the same frame as dropped.
> 
> This patch checks for error codes which indicate that the frame has
> already been freed.
> 
> Bugzilla: https://bugzilla.redhat.com/show_bug.cgi?id=2109946

Please remove the Bugzilla link, it doesn't seem to be public.
If it is then it should be a Link: tag, not a custom tag for bz.

> Signed-off-by: Mike Pattrick <mkp@redhat.com>
> ---
>  net/openvswitch/datapath.c | 16 +++++++++++++---
>  1 file changed, 13 insertions(+), 3 deletions(-)
> 
> diff --git a/net/openvswitch/datapath.c b/net/openvswitch/datapath.c
> index 7e8a39a35627..029f9c3e1c28 100644
> --- a/net/openvswitch/datapath.c
> +++ b/net/openvswitch/datapath.c
> @@ -252,10 +252,20 @@ void ovs_dp_process_packet(struct sk_buff *skb, struct sw_flow_key *key)
>  
>  		upcall.mru = OVS_CB(skb)->mru;
>  		error = ovs_dp_upcall(dp, skb, key, &upcall, 0);
> -		if (unlikely(error))
> -			kfree_skb(skb);
> -		else
> +		switch (error) {
> +		case 0:
> +			fallthrough;
> +		case -EAGAIN:
> +			fallthrough;
> +		case -ERESTARTSYS:
> +			fallthrough;

No need to add the fallthrough;s between two case statements.

> +		case -EINTR:
>  			consume_skb(skb);
> +			break;
> +		default:
> +			kfree_skb(skb);
> +			break;
> +		}
>  		stats_counter = &stats->n_missed;
>  		goto out;
>  	}