Olaf Kirch <okir@xxxxxxx> wrote:
>It seems when you create a VLAN configuration on top of a bonding
>device, any normal (non-VLAN) traffic stops completely. The reason
>is that bond_dev_queue_xmit() drops all outgoing frames that have
>no VLAN tag.
>
>The attached patch tries to fix this - is this the right way to do this,
>or does that break other areas of VLAN-over-bonding?
Logically this looks good. I tried it out on a non-VLAN aware
e100, and it seems to the do right thing. I tweaked the patch to update
the comments to reflect the new reality (and fix a typo while I was
there).
-J
---
-Jay Vosburgh, IBM Linux Technology Center, fubar@xxxxxxxxxx
Signed-off-by: Jay Vosburgh <fubar@xxxxxxxxxx>
--- linux-2.6.11-rc4-netdev1-virgin/drivers/net/bonding/bond_main.c
2005-02-15 11:31:40.000000000 -0800
+++ linux-2.6.11-rc4-netdev1/drivers/net/bonding/bond_main.c 2005-02-15
16:09:10.000000000 -0800
@@ -793,29 +793,20 @@
* @skb: hw accel VLAN tagged skb to transmit
* @slave_dev: slave that is supposed to xmit this skbuff
*
- * When the bond gets an skb to tarnsmit that is
+ * When the bond gets an skb to transmit that is
* already hardware accelerated VLAN tagged, and it
* needs to relay this skb to a slave that is not
* hw accel capable, the skb needs to be "unaccelerated",
* i.e. strip the hwaccel tag and re-insert it as part
* of the payload.
- *
- * Assumption - once a VLAN device is created over the bond device, all
- * packets are going to be hardware accelerated VLAN tagged since the IP
- * binding is done over the VLAN device
*/
int bond_dev_queue_xmit(struct bonding *bond, struct sk_buff *skb, struct
net_device *slave_dev)
{
unsigned short vlan_id;
- int res;
if (!list_empty(&bond->vlan_list) &&
- !(slave_dev->features & NETIF_F_HW_VLAN_TX)) {
- res = vlan_get_tag(skb, &vlan_id);
- if (res) {
- return -EINVAL;
- }
-
+ !(slave_dev->features & NETIF_F_HW_VLAN_TX) &&
+ vlan_get_tag(skb, &vlan_id) == 0) {
skb->dev = slave_dev;
skb = vlan_put_tag(skb, vlan_id);
if (!skb) {
|