Skip to contentSkip to conversation

8021q: set hard_header_len when VLAN offload features are toggled

RFC PATCH In review John Fastabend 2 participants 4 messages Oct 26, 2010 – Oct 27, 2010

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

RFC PATCH v2 1/1 8021q: set hard_header_len when VLAN offload features are toggled

Toggling the vlan tx|rx hw offloads needs to set the hard_header_len as well otherwise we end up using LL_RESERVED_SPACE incorrectly. This results in pskb_expand_head() being used unnecessarily.

This add a check in vlan_transfer_features to catch the ETH_FLAG_TXVLAN flag and set the header length. This requires drivers to add the ETH_FLAG_TXVLAN to vlan_features.

---

 net/8021q/vlan.c |   10 ++++++++++
 1 files changed, 10 insertions(+), 0 deletions(-)
net/8021q/vlan.c +10 -0
index 05b867e..825011b 100644
@@ -334,6 +334,16 @@ static void vlan_transfer_features(struct net_device *dev,
 	vlandev->features &= ~dev->vlan_features;
 	vlandev->features |= dev->features & dev->vlan_features;
 	vlandev->gso_max_size = dev->gso_max_size;
+
+	/* is ETH_FLAGS_TXVLAN being toggled */
+	if ((vlandev->features & ETH_FLAG_TXVLAN) ^
+	    (old_features & ETH_FLAG_TXVLAN)) {
+		if (vlandev->features & ETH_FLAG_TXVLAN)
+			vlandev->hard_header_len -= VLAN_HLEN;
+		else
+			vlandev->hard_header_len += VLAN_HLEN;
+	}
+
 #if defined(CONFIG_FCOE) || defined(CONFIG_FCOE_MODULE)
 	vlandev->fcoe_ddp_xid = dev->fcoe_ddp_xid;
 #endif
#1 lore raw
Quoted 9 lines · John Fastabend

Toggling the vlan tx|rx hw offloads needs to set the hard_header_len as well otherwise we end up using LL_RESERVED_SPACE incorrectly. This results in pskb_expand_head() being used unnecessarily.

This add a check in vlan_transfer_features  to catch the ETH_FLAG_TXVLAN flag and set the header length. This requires drivers to add the ETH_FLAG_TXVLAN to vlan_features.

I think this addresses all of the original problems. However, I don't think that we want to have drivers claim to support vlan offloading as a feature for vlan packets. That implies some type of QinQ functionality to me. In addition, if the vlan device claims to support offloading and a second vlan device is stacked on top of it, then the two will clobber skb->vlan_tci. It's probably simpler to just keep track of whether vlan offloading is currently enabled so we can find out whether it changed.

Quoted 12 lines

---

 net/8021q/vlan.c |   10 ++++++++++

 1 files changed, 10 insertions(+), 0 deletions(-)
net/8021q/vlan.c +0 -0
index 05b867e..825011b 100644
@@ -334,6 +334,16 @@ static void vlan_transfer_features(struct net_device *dev,
        vlandev->features &= ~dev->vlan_features;
        vlandev->features |= dev->features & dev->vlan_features;

net/8021q/vlan.c static void vlan_transfer_features(struct net_device *dev, · line 345

Quoted 10 lines
net/8021q/vlan.c +9 -0
        vlandev->gso_max_size = dev->gso_max_size;
+
+       /* is ETH_FLAGS_TXVLAN being toggled */
+       if ((vlandev->features & ETH_FLAG_TXVLAN) ^
+           (old_features & ETH_FLAG_TXVLAN)) {
+               if (vlandev->features & ETH_FLAG_TXVLAN)
+                       vlandev->hard_header_len -= VLAN_HLEN;
+               else
+                       vlandev->hard_header_len += VLAN_HLEN;
+       }

The correct flag for dev->features is NETIF_F_HW_VLAN_TX. ETH_FLAGS_TXVLAN is an ethtool construct (that happens to have the same value).

Thanks.

#2 lore raw
Quoted 12 lines · Jesse Gross
Quoted 9 lines · John Fastabend

Toggling the vlan tx|rx hw offloads needs to set the hard_header_len as well otherwise we end up using LL_RESERVED_SPACE incorrectly. This results in pskb_expand_head() being used unnecessarily.

This add a check in vlan_transfer_features to catch the ETH_FLAG_TXVLAN flag and set the header length. This requires drivers to add the ETH_FLAG_TXVLAN to vlan_features.

Quoted 9 lines

I think this addresses all of the original problems. However, I don't think that we want to have drivers claim to support vlan offloading as a feature for vlan packets. That implies some type of QinQ functionality to me. In addition, if the vlan device claims to support offloading and a second vlan device is stacked on top of it, then the two will clobber skb->vlan_tci. It's probably simpler to just keep track of whether vlan offloading is currently enabled so we can find out whether it changed.

Agreed. Rather then trying to be clever this is probably the easiest.

net/8021q/vlan.c +6 -0
@@ -334,6 +334,12 @@ Hunk #1, a/net/8021q/vlan.c static void vlan_transfer_features(struct net_device *dev,
        vlandev->features &= ~dev->vlan_features;
        vlandev->features |= dev->features & dev->vlan_features;
        vlandev->gso_max_size = dev->gso_max_size;
+
+       if (dev->features & NETIF_F_HW_VLAN_TX)
+               vlandev->hard_header_len = dev->hard_header_len;
+       else
+               vlandev->hard_header_len = dev->hard_header_len + VLAN_HLEN;
+
 #if defined(CONFIG_FCOE) || defined(CONFIG_FCOE_MODULE)
        vlandev->fcoe_ddp_xid = dev->fcoe_ddp_xid;
 #endif
Quoted 32 lines
Quoted 12 lines

---

 net/8021q/vlan.c |   10 ++++++++++
 1 files changed, 10 insertions(+), 0 deletions(-)
net/8021q/vlan.c +0 -0
index 05b867e..825011b 100644
@@ -334,6 +334,16 @@ static void vlan_transfer_features(struct net_device *dev,
       vlandev->features &= ~dev->vlan_features;
       vlandev->features |= dev->features & dev->vlan_features;

net/8021q/vlan.c static void vlan_transfer_features(struct net_device *dev, · line 345

Quoted 10 lines
net/8021q/vlan.c +9 -0
       vlandev->gso_max_size = dev->gso_max_size;
+
+       /* is ETH_FLAGS_TXVLAN being toggled */
+       if ((vlandev->features & ETH_FLAG_TXVLAN) ^
+           (old_features & ETH_FLAG_TXVLAN)) {
+               if (vlandev->features & ETH_FLAG_TXVLAN)
+                       vlandev->hard_header_len -= VLAN_HLEN;
+               else
+                       vlandev->hard_header_len += VLAN_HLEN;
+       }

The correct flag for dev->features is NETIF_F_HW_VLAN_TX. ETH_FLAGS_TXVLAN is an ethtool construct (that happens to have the same value).

Thanks.
--
To unsubscribe from this list: send the line "unsubscribe netdev" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html

#3 lore raw
Quoted 28 lines · John Fastabend
Quoted 12 lines · Jesse Gross
Quoted 9 lines · John Fastabend

Toggling the vlan tx|rx hw offloads needs to set the hard_header_len as well otherwise we end up using LL_RESERVED_SPACE incorrectly. This results in pskb_expand_head() being used unnecessarily.

This add a check in vlan_transfer_features  to catch the ETH_FLAG_TXVLAN flag and set the header length. This requires drivers to add the ETH_FLAG_TXVLAN to vlan_features.

Quoted 9 lines

I think this addresses all of the original problems.  However, I don't think that we want to have drivers claim to support vlan offloading as a feature for vlan packets.  That implies some type of QinQ functionality to me.  In addition, if the vlan device claims to support offloading and a second vlan device is stacked on top of it, then the two will clobber skb->vlan_tci.  It's probably simpler to just keep track of whether vlan offloading is currently enabled so we can find out whether it changed.

Agreed. Rather then trying to be clever this is probably the easiest.

net/8021q/vlan.c Hunk #1, a/net/8021q/vlan.c static void vlan_transfer_features(struct net_device *dev, · line 342

Quoted 9 lines
net/8021q/vlan.c +6 -0
@@ -334,6 +334,12 @@ Hunk #1, a/net/8021q/vlan.c static void vlan_transfer_features(struct net_device *dev,
        vlandev->features &= ~dev->vlan_features;
        vlandev->features |= dev->features & dev->vlan_features;
        vlandev->gso_max_size = dev->gso_max_size;
+
+       if (dev->features & NETIF_F_HW_VLAN_TX)
+               vlandev->hard_header_len = dev->hard_header_len;
+       else
+               vlandev->hard_header_len = dev->hard_header_len + VLAN_HLEN;
+

Great, that's even simpler than I was thinking.

I think this series is ready to go.