[PATCH net-next] vlan: Add ability to always enable TSO/UFO

[PATCH net-next] vlan: Add ability to always enable TSO/UFO

[Toshiaki Makita]

Since the real device can segment packets by software, a vlan device
can set TSO/UFO even when the real device doesn't have those features.
Unlike GSO, this allows packets to be segmented after Qdisc.

Signed-off-by: Toshiaki Makita <makita.toshiaki@lab.ntt.co.jp>
---
 net/8021q/vlan_dev.c | 7 ++++---
 1 file changed, 4 insertions(+), 3 deletions(-)

diff --git a/net/8021q/vlan_dev.c b/net/8021q/vlan_dev.c
index 3768050..1189564 100644
--- a/net/8021q/vlan_dev.c
+++ b/net/8021q/vlan_dev.c
@@ -579,11 +579,12 @@ static int vlan_dev_init(struct net_device *dev)
 		      (1<<__LINK_STATE_PRESENT);
 
 	dev->hw_features = NETIF_F_ALL_CSUM | NETIF_F_SG |
-			   NETIF_F_FRAGLIST | NETIF_F_ALL_TSO |
+			   NETIF_F_FRAGLIST | NETIF_F_GSO_SOFTWARE |
 			   NETIF_F_HIGHDMA | NETIF_F_SCTP_CSUM |
 			   NETIF_F_ALL_FCOE;
 
-	dev->features |= real_dev->vlan_features | NETIF_F_LLTX;
+	dev->features |= real_dev->vlan_features | NETIF_F_LLTX |
+			 NETIF_F_GSO_SOFTWARE;
 	dev->gso_max_size = real_dev->gso_max_size;
 	if (dev->features & NETIF_F_VLAN_FEATURES)
 		netdev_warn(real_dev, "VLAN features are set incorrectly.  Q-in-Q configurations may not work correctly.\n");
@@ -648,7 +649,7 @@ static netdev_features_t vlan_dev_fix_features(struct net_device *dev,
 	features |= NETIF_F_RXCSUM;
 	features = netdev_intersect_features(features, real_dev->features);
 
-	features |= old_features & NETIF_F_SOFT_FEATURES;
+	features |= old_features & (NETIF_F_SOFT_FEATURES | NETIF_F_GSO_SOFTWARE);
 	features |= NETIF_F_LLTX;
 
 	return features;
-- 
1.8.1.2

[PATCH net-next] bridge: Add ability to always enable TSO/UFO

[Toshiaki Makita]

Currently a bridge device turns off TSO/UFO features if no bridge ports
support it. We can always enable them, since packets can be segmented
on ports by software as well as on the bridge device.
This will reduce the number of packets processed in the bridge.

Signed-off-by: Toshiaki Makita <makita.toshiaki@lab.ntt.co.jp>
---
 net/bridge/br_if.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/net/bridge/br_if.c b/net/bridge/br_if.c
index ed307db..e93bf17 100644
--- a/net/bridge/br_if.c
+++ b/net/bridge/br_if.c
@@ -418,7 +418,7 @@ netdev_features_t br_features_recompute(struct net_bridge *br,
 		return features;
 
 	mask = features;
-	features &= ~NETIF_F_ONE_FOR_ALL;
+	features &= ~NETIF_F_ONE_FOR_ALL | NETIF_F_GSO_SOFTWARE;
 
 	list_for_each_entry(p, &br->port_list, list) {
 		features = netdev_increment_features(features,
-- 
1.8.1.2

Re: [PATCH net-next] bridge: Add ability to always enable TSO/UFO

[David Miller]

From: Toshiaki Makita <makita.toshiaki@lab.ntt.co.jp>
Date: Wed, 10 Dec 2014 11:43:14 +0900

> -	features &= ~NETIF_F_ONE_FOR_ALL;
> +	features &= ~NETIF_F_ONE_FOR_ALL | NETIF_F_GSO_SOFTWARE;

I don't think this is the expression you intend to use.

I think you meant:

	features &= ~(NETIF_F_ONE_FOR_ALL | NETIF_F_GSO_SOFTWARE);

Or:

	features = ~NETIF_F_ONE_FOR_ALL;
	features |= NETIF_F_GSO_SOFTWARE;

Re: [PATCH net-next] bridge: Add ability to always enable TSO/UFO

[Toshiaki Makita]

On 2014/12/11 4:50, David Miller wrote:
> From: Toshiaki Makita <makita.toshiaki@lab.ntt.co.jp>
> Date: Wed, 10 Dec 2014 11:43:14 +0900
> 
>> -	features &= ~NETIF_F_ONE_FOR_ALL;
>> +	features &= ~NETIF_F_ONE_FOR_ALL | NETIF_F_GSO_SOFTWARE;
> 
> I don't think this is the expression you intend to use.

Thank you, but this is really my intended expression.

"features &= ~NETIF_F_ONE_FOR_ALL" drops all of ONE_FOR_ALL bits
including GSO_SOFTWARE.
But I want to leave GSO_SOFTWARE bits here.

> 
> I think you meant:
> 
> 	features &= ~(NETIF_F_ONE_FOR_ALL | NETIF_F_GSO_SOFTWARE);

ONE_FOR_ALL already includes GSO_SOFTWARE.

> 
> Or:
> 
> 	features = ~NETIF_F_ONE_FOR_ALL;
> 	features |= NETIF_F_GSO_SOFTWARE;

This way, users cannot drop TSO/UFO by ethtool.

Thanks,
Toshiaki Makita

Re: [PATCH net-next] bridge: Add ability to always enable TSO/UFO

[David Miller]

From: Toshiaki Makita <makita.toshiaki@lab.ntt.co.jp>
Date: Thu, 11 Dec 2014 11:04:44 +0900

> On 2014/12/11 4:50, David Miller wrote:
>> From: Toshiaki Makita <makita.toshiaki@lab.ntt.co.jp>
>> Date: Wed, 10 Dec 2014 11:43:14 +0900
>> 
>>> -	features &= ~NETIF_F_ONE_FOR_ALL;
>>> +	features &= ~NETIF_F_ONE_FOR_ALL | NETIF_F_GSO_SOFTWARE;
>> 
>> I don't think this is the expression you intend to use.
> 
> Thank you, but this is really my intended expression.
> 
> "features &= ~NETIF_F_ONE_FOR_ALL" drops all of ONE_FOR_ALL bits
> including GSO_SOFTWARE.
> But I want to leave GSO_SOFTWARE bits here.

It is clearer to say this as:

	~(NETIF_F_ONE_FOR_ALL & ~NETIF_F_GSO_SOFTWARE)

Or create a new NETIF_F_* macro to express this idea succinctly.

Re: [PATCH net-next] bridge: Add ability to always enable TSO/UFO

[Toshiaki Makita]

On 2014/12/11 11:13, David Miller wrote:
> From: Toshiaki Makita <makita.toshiaki@lab.ntt.co.jp>
> Date: Thu, 11 Dec 2014 11:04:44 +0900
> 
>> On 2014/12/11 4:50, David Miller wrote:
>>> From: Toshiaki Makita <makita.toshiaki@lab.ntt.co.jp>
>>> Date: Wed, 10 Dec 2014 11:43:14 +0900
>>>
>>>> -	features &= ~NETIF_F_ONE_FOR_ALL;
>>>> +	features &= ~NETIF_F_ONE_FOR_ALL | NETIF_F_GSO_SOFTWARE;
>>>
>>> I don't think this is the expression you intend to use.
>>
>> Thank you, but this is really my intended expression.
>>
>> "features &= ~NETIF_F_ONE_FOR_ALL" drops all of ONE_FOR_ALL bits
>> including GSO_SOFTWARE.
>> But I want to leave GSO_SOFTWARE bits here.
> 
> It is clearer to say this as:
> 
> 	~(NETIF_F_ONE_FOR_ALL & ~NETIF_F_GSO_SOFTWARE)
> 
> Or create a new NETIF_F_* macro to express this idea succinctly.

OK, I will. (once net-next is reopened)

Thanks,
Toshiaki Makita

Re: [PATCH net-next] vlan: Add ability to always enable TSO/UFO

[David Miller]

From: Toshiaki Makita <makita.toshiaki@lab.ntt.co.jp>
Date: Wed, 10 Dec 2014 11:43:13 +0900

> Since the real device can segment packets by software, a vlan device
> can set TSO/UFO even when the real device doesn't have those features.
> Unlike GSO, this allows packets to be segmented after Qdisc.
> 
> Signed-off-by: Toshiaki Makita <makita.toshiaki@lab.ntt.co.jp>

Applied.