[PATCH net-next-2.6] vlan: adds drops accounting

[PATCH net-next-2.6] vlan: adds drops accounting

[Eric Dumazet]

Its hard to tell if vlans are dropping frames, since
every frame given to vlan_???_start_xmit() functions
is accounted as fully transmitted by lower device.

We can test dev_queue_xmit() return values to
properly account for dropped frames.

Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
---
 net/8021q/vlan_dev.c |   29 ++++++++++++++++++++++-------
 1 files changed, 22 insertions(+), 7 deletions(-)

diff --git a/net/8021q/vlan_dev.c b/net/8021q/vlan_dev.c
index 3938c3e..ca1f6a5 100644
--- a/net/8021q/vlan_dev.c
+++ b/net/8021q/vlan_dev.c
@@ -294,6 +294,8 @@ static netdev_tx_t vlan_dev_hard_start_xmit(struct sk_buff *skb,
 	int i = skb_get_queue_mapping(skb);
 	struct netdev_queue *txq = netdev_get_tx_queue(dev, i);
 	struct vlan_ethhdr *veth = (struct vlan_ethhdr *)(skb->data);
+	unsigned int len;
+	int ret;
 
 	/* Handle non-VLAN frames if they are sent to us, for example by DHCP.
 	 *
@@ -319,11 +321,17 @@ static netdev_tx_t vlan_dev_hard_start_xmit(struct sk_buff *skb,
 			vlan_dev_info(dev)->cnt_inc_headroom_on_tx++;
 	}
 
-	txq->tx_packets++;
-	txq->tx_bytes += skb->len;
 
 	skb->dev = vlan_dev_info(dev)->real_dev;
-	dev_queue_xmit(skb);
+	len = skb->len;
+	ret = dev_queue_xmit(skb);
+
+	if (likely(ret == NET_XMIT_SUCCESS)) {
+		txq->tx_packets++;
+		txq->tx_bytes += len;
+	} else
+		txq->tx_dropped++;
+
 	return NETDEV_TX_OK;
 }
 
@@ -333,16 +341,23 @@ static netdev_tx_t vlan_dev_hwaccel_hard_start_xmit(struct sk_buff *skb,
 	int i = skb_get_queue_mapping(skb);
 	struct netdev_queue *txq = netdev_get_tx_queue(dev, i);
 	u16 vlan_tci;
+	unsigned int len;
+	int ret;
 
 	vlan_tci = vlan_dev_info(dev)->vlan_id;
 	vlan_tci |= vlan_dev_get_egress_qos_mask(dev, skb);
 	skb = __vlan_hwaccel_put_tag(skb, vlan_tci);
 
-	txq->tx_packets++;
-	txq->tx_bytes += skb->len;
-
 	skb->dev = vlan_dev_info(dev)->real_dev;
-	dev_queue_xmit(skb);
+	len = skb->len;
+	ret = dev_queue_xmit(skb);
+
+	if (likely(ret == NET_XMIT_SUCCESS)) {
+		txq->tx_packets++;
+		txq->tx_bytes += len;
+	} else
+		txq->tx_dropped++;
+
 	return NETDEV_TX_OK;
 }
 

Re: [PATCH net-next-2.6] vlan: adds drops accounting

[David Miller]

From: Eric Dumazet <eric.dumazet@gmail.com>
Date: Thu, 03 Sep 2009 12:39:16 +0200

> Its hard to tell if vlans are dropping frames, since
> every frame given to vlan_???_start_xmit() functions
> is accounted as fully transmitted by lower device.
> 
> We can test dev_queue_xmit() return values to
> properly account for dropped frames.
> 
> Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>

Applied.

But, how inappropriate would it be to keep doing the accounting
but pass the status back to the caller?  Or will that screw up
the qdisc running engine?

Re: [PATCH net-next-2.6] vlan: adds drops accounting

[Patrick McHardy]

David Miller wrote:
> From: Eric Dumazet <eric.dumazet@gmail.com>
> Date: Thu, 03 Sep 2009 12:39:16 +0200
> 
>> Its hard to tell if vlans are dropping frames, since
>> every frame given to vlan_???_start_xmit() functions
>> is accounted as fully transmitted by lower device.
>>
>> We can test dev_queue_xmit() return values to
>> properly account for dropped frames.
>>
>> Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
> 
> Applied.
> 
> But, how inappropriate would it be to keep doing the accounting
> but pass the status back to the caller?  Or will that screw up
> the qdisc running engine?

Its currently not possible to return the dev_queue_xmit() status
from ndo_start_xmit(). My patch to extend the possible return
values will allow that. I hope I'm able to take care of it today.