[PATCH net-next 0/2] vxlan, bnx2x offload

[PATCH net-next 0/2] vxlan, bnx2x offload

[Dmitry Kravkov]

Hi Dave,

The series fixes GSO flags in vxlan xmit code and
adds appropriate code for bnx2x driver to utilize
fw support for UDP tunneling protocols offload.

Netperf's results below show improvement in throughput
between two vxlan interfaces:
before:
MIGRATED TCP STREAM TEST from 0.0.0.0 (0.0.0.0) port 0 AF_INET to 9.0.0.2 (9.0.0.2) port 0 AF_INET
Recv   Send    Send                          
Socket Socket  Message  Elapsed              
Size   Size    Size     Time     Throughput  
bytes  bytes   bytes    secs.    10^6bits/sec  

 87380  16384  16384    10.20     228.73   

after:
MIGRATED TCP STREAM TEST from 0.0.0.0 (0.0.0.0) port 0 AF_INET to 9.0.0.2 (9.0.0.2) port 0 AF_INET
Recv   Send    Send                          
Socket Socket  Message  Elapsed              
Size   Size    Size     Time     Throughput  
bytes  bytes   bytes    secs.    10^6bits/sec  

 87380  16384  16384    10.00    3314.75   


Thanks,
Dmitry

[PATCH net-next 1/2] vxlan: do not set SKB_GSO_UDP

[Dmitry Kravkov]

Since SKB_GSO_* flags are set by appropriate gso_segment callback
in TCP/UDP layer.

CC: Pravin B Shelar <pshelar@nicira.com> 
Signed-off-by: Dmitry Kravkov <dmitry@broadcom.com>
Signed-off-by: Eilon Greenstein <eilong@broadcom.com>
---
 drivers/net/vxlan.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/net/vxlan.c b/drivers/net/vxlan.c
index a7fd9a0..b92c6a0 100644
--- a/drivers/net/vxlan.c
+++ b/drivers/net/vxlan.c
@@ -913,7 +913,7 @@ static int handle_offloads(struct sk_buff *skb)
 		if (unlikely(err))
 			return err;
 
-		skb_shinfo(skb)->gso_type |= (SKB_GSO_UDP_TUNNEL | SKB_GSO_UDP);
+		skb_shinfo(skb)->gso_type |= SKB_GSO_UDP_TUNNEL;
 	} else if (skb->ip_summed != CHECKSUM_PARTIAL)
 		skb->ip_summed = CHECKSUM_NONE;
 
-- 
1.8.1.4

[PATCH net-next 2/2] bnx2x: allow handling tunnel over udp offload

[Dmitry Kravkov]

CC: Pravin B Shelar <pshelar@nicira.com> 
Signed-off-by: Dmitry Kravkov <dmitry@broadcom.com>
Signed-off-by: Eilon Greenstein <eilong@broadcom.com>
---
 drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.c  | 5 +++++
 drivers/net/ethernet/broadcom/bnx2x/bnx2x_main.c | 4 ++--
 2 files changed, 7 insertions(+), 2 deletions(-)

diff --git a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.c b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.c
index fd20a4f..466b512 100644
--- a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.c
+++ b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.c
@@ -3530,6 +3530,11 @@ static void bnx2x_update_pbds_gso_enc(struct sk_buff *skb,
 			ETH_TX_PARSE_2ND_BD_IP_HDR_LEN_OUTER_W_SHIFT) |
 		((skb->protocol == cpu_to_be16(ETH_P_8021Q)) <<
 			ETH_TX_PARSE_2ND_BD_LLC_SNAP_EN_SHIFT);
+
+	if (ip_hdr(skb)->protocol == IPPROTO_UDP) {
+		SET_FLAG(*global_data, ETH_TX_PARSE_2ND_BD_TUNNEL_UDP_EXIST, 1);
+		pbd2->tunnel_udp_hdr_start_w = skb_transport_offset(skb) >> 1;
+	}
 }
 
 /* called with netif_tx_lock
diff --git a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_main.c b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_main.c
index 927f83a..91a0434 100644
--- a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_main.c
+++ b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_main.c
@@ -12043,11 +12043,11 @@ static int bnx2x_init_dev(struct bnx2x *bp, struct pci_dev *pdev,
 		NETIF_F_RXCSUM | NETIF_F_LRO | NETIF_F_GRO |
 		NETIF_F_RXHASH | NETIF_F_HW_VLAN_CTAG_TX;
 	if (!CHIP_IS_E1x(bp)) {
-		dev->hw_features |= NETIF_F_GSO_GRE;
+		dev->hw_features |= NETIF_F_GSO_GRE | NETIF_F_GSO_UDP_TUNNEL;
 		dev->hw_enc_features =
 			NETIF_F_IP_CSUM | NETIF_F_IPV6_CSUM | NETIF_F_SG |
 			NETIF_F_TSO | NETIF_F_TSO_ECN | NETIF_F_TSO6 |
-			NETIF_F_GSO_GRE;
+			NETIF_F_GSO_GRE | NETIF_F_GSO_UDP_TUNNEL;
 	}
 
 	dev->vlan_features = NETIF_F_SG | NETIF_F_IP_CSUM | NETIF_F_IPV6_CSUM |
-- 
1.8.1.4

Re: [PATCH net-next 1/2] vxlan: do not set SKB_GSO_UDP

[Pravin Shelar]

On Sun, Apr 28, 2013 at 11:16 AM, Dmitry Kravkov <dmitry@broadcom.com> wrote:
> Since SKB_GSO_* flags are set by appropriate gso_segment callback
> in TCP/UDP layer.
>
> CC: Pravin B Shelar <pshelar@nicira.com>
> Signed-off-by: Dmitry Kravkov <dmitry@broadcom.com>
> Signed-off-by: Eilon Greenstein <eilong@broadcom.com>
> ---

udp flag was set since we are using udp offload handler for vxlan gso.
But I think it forces to have udp offload for tunnel offloading.
Therefore the change looks right.

Acked-by: Pravin B Shelar <pshelar@nicira.com>

>  drivers/net/vxlan.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/drivers/net/vxlan.c b/drivers/net/vxlan.c
> index a7fd9a0..b92c6a0 100644
> --- a/drivers/net/vxlan.c
> +++ b/drivers/net/vxlan.c
> @@ -913,7 +913,7 @@ static int handle_offloads(struct sk_buff *skb)
>                 if (unlikely(err))
>                         return err;
>
> -               skb_shinfo(skb)->gso_type |= (SKB_GSO_UDP_TUNNEL | SKB_GSO_UDP);
> +               skb_shinfo(skb)->gso_type |= SKB_GSO_UDP_TUNNEL;
>         } else if (skb->ip_summed != CHECKSUM_PARTIAL)
>                 skb->ip_summed = CHECKSUM_NONE;
>
> --
> 1.8.1.4
>
>

Re: [PATCH net-next 0/2] vxlan, bnx2x offload

[David Miller]

From: "Dmitry Kravkov" <dmitry@broadcom.com>
Date: Sun, 28 Apr 2013 21:16:00 +0300

> The series fixes GSO flags in vxlan xmit code and
> adds appropriate code for bnx2x driver to utilize
> fw support for UDP tunneling protocols offload.

Series applied, thanks.