Re: VLAN I/F's and TX queue.

[Eric Dumazet <eric.dumazet@gmail.com>]

Le lundi 10 mai 2010 à 16:41 +0200, Patrick McHardy a écrit :
> Eric Dumazet wrote:
> > Le lundi 10 mai 2010 à 16:26 +0200, Patrick McHardy a écrit :
> > 
> >> Is the intention just to avoid accounting the packet as dropped?
> >> That seems fine to me since in case of NET_XMIT_CN its actually
> >> not the currently transmitted packet that was dropped.
> >>
> >> But part of the intention of the above mentioned patch was actually
> >> to inform higher layers of congestion so they can take action if
> >> desired, which would be defeated by this patch.
> >>
> > 
> > I see, so maybe we want following patch instead ?
> > 
> > (letting NET_XMIT_CN be given to caller, but accounting current packet
> > as transmitted ?)
> 
> Perfect, thanks. I'd suggest to change macvlan in a similar fashion
> for consistency though.
> 
> In any case please feel free to add my
> 
> Acked-by: Patrick McHardy <kaber@trash.net>

Indeed, thanks !

[PATCH net-next-2.6] net: congestion notifications are not dropped packets

vlan/macvlan start_xmit() can inform caller of congestion with
NET_XMIT_CN return value. This doesnt mean packet was dropped.
Increment normal stat counters instead of tx_dropped.

Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Acked-by: Patrick McHardy <kaber@trash.net>
---
diff --git a/drivers/net/macvlan.c b/drivers/net/macvlan.c
index 9a939d8..0f8cd95 100644
--- a/drivers/net/macvlan.c
+++ b/drivers/net/macvlan.c
@@ -243,7 +243,7 @@ netdev_tx_t macvlan_start_xmit(struct sk_buff *skb,
 	int ret;
 
 	ret = macvlan_queue_xmit(skb, dev);
-	if (likely(ret == NET_XMIT_SUCCESS)) {
+	if (likely(ret == NET_XMIT_SUCCESS || ret == NET_XMIT_CN)) {
 		txq->tx_packets++;
 		txq->tx_bytes += len;
 	} else
diff --git a/net/8021q/vlan_dev.c b/net/8021q/vlan_dev.c
index b5249c5..55be908 100644
--- a/net/8021q/vlan_dev.c
+++ b/net/8021q/vlan_dev.c
@@ -327,7 +327,7 @@ static netdev_tx_t vlan_dev_hard_start_xmit(struct sk_buff *skb,
 	len = skb->len;
 	ret = dev_queue_xmit(skb);
 
-	if (likely(ret == NET_XMIT_SUCCESS)) {
+	if (likely(ret == NET_XMIT_SUCCESS || ret == NET_XMIT_CN)) {
 		txq->tx_packets++;
 		txq->tx_bytes += len;
 	} else
@@ -353,7 +353,7 @@ static netdev_tx_t vlan_dev_hwaccel_hard_start_xmit(struct sk_buff *skb,
 	len = skb->len;
 	ret = dev_queue_xmit(skb);
 
-	if (likely(ret == NET_XMIT_SUCCESS)) {
+	if (likely(ret == NET_XMIT_SUCCESS || ret == NET_XMIT_CN)) {
 		txq->tx_packets++;
 		txq->tx_bytes += len;
 	} else