Skip to contentSkip to conversation

ice: Add mpls+tso support

PATCH New Joe Damato 2 participants 2 messages Mar 18, 2022 – Apr 5, 2022

Subsystems (2)
lore
Share & tools
lore

Joe Damato

author
#0 lore raw

PATCH v2 1/1 ice: Add mpls+tso support

Attempt to add mpls+tso support.

I don't have ice hardware available to test myself, but I just implemented this feature in i40e and thought it might be useful to implement for ice while this is fresh in my brain.

Hoping some one at intel will be able to test this on my behalf.

v1 -> v2:
- In ice_tx_csum, only set ICE_TX_FLAGS_IPV6 if ip v6 is detected explicitly. All cases other than ipv4 or ipv6 leave first->tx_flags unset, which allows for ice_tx_csum to return -1 to indicate that csum offloading is not supported.

---
 drivers/net/ethernet/intel/ice/ice_main.c |  4 +++-
 drivers/net/ethernet/intel/ice/ice_txrx.c | 29 +++++++++++++++++++++--------
 2 files changed, 24 insertions(+), 9 deletions(-)
drivers/net/ethernet/intel/ice/ice_main.c +3 -1
index 285b0cd..3032eab 100644
@@ -3454,7 +3454,9 @@ static void ice_set_netdev_features(struct net_device *netdev)
 			      vlano_features | tso_features;
 
 	/* add support for HW_CSUM on packets with MPLS header */
-	netdev->mpls_features =  NETIF_F_HW_CSUM;
+	netdev->mpls_features =  NETIF_F_HW_CSUM |
+				 NETIF_F_TSO     |
+				 NETIF_F_TSO6;
 
 	/* enable features */
 	netdev->features |= netdev->hw_features;
drivers/net/ethernet/intel/ice/ice_txrx.c +21 -8
index 2dd3e1b..ca41453 100644
@@ -9,6 +9,7 @@
 #include <linux/bpf_trace.h>
 #include <net/busy_poll.h>
 #include <net/dsfield.h>
+#include <net/mpls.h>
 #include <net/xdp.h>
 #include "ice_txrx_lib.h"
 #include "ice_lib.h"
@@ -1910,18 +1911,24 @@ int ice_tx_csum(struct ice_tx_buf *first, struct ice_tx_offload_params *off)
 	if (skb->ip_summed != CHECKSUM_PARTIAL)
 		return 0;
 
-	ip.hdr = skb_network_header(skb);
-	l4.hdr = skb_transport_header(skb);
+	protocol = vlan_get_protocol(skb);
+
+	if (eth_p_mpls(protocol))
+		ip.hdr = skb_inner_network_header(skb);
+	else
+		ip.hdr = skb_network_header(skb);
+	l4.hdr = skb_checksum_start(skb);
 
 	/* compute outer L2 header size */
 	l2_len = ip.hdr - skb->data;
 	offset = (l2_len / 2) << ICE_TX_DESC_LEN_MACLEN_S;
 
-	protocol = vlan_get_protocol(skb);
-
-	if (protocol == htons(ETH_P_IP))
+	/* set the tx_flags to indicate the IP protocol type. this is
+	 * required so that checksum header computation below is accurate.
+	 */
+	if (ip.v4->version == 4)
 		first->tx_flags |= ICE_TX_FLAGS_IPV4;
-	else if (protocol == htons(ETH_P_IPV6))
+	else if (ip.v6->version == 6)
 		first->tx_flags |= ICE_TX_FLAGS_IPV6;
 
 	if (skb->encapsulation) {
@@ -2122,6 +2129,7 @@ int ice_tso(struct ice_tx_buf *first, struct ice_tx_offload_params *off)
 	u32 paylen;
 	u8 l4_start;
 	int err;
+	__be16 protocol;
 
 	if (skb->ip_summed != CHECKSUM_PARTIAL)
 		return 0;
@@ -2134,8 +2142,13 @@ int ice_tso(struct ice_tx_buf *first, struct ice_tx_offload_params *off)
 		return err;
 
 	/* cppcheck-suppress unreadVariable */
-	ip.hdr = skb_network_header(skb);
-	l4.hdr = skb_transport_header(skb);
+	protocol = vlan_get_protocol(skb);
+
+	if (eth_p_mpls(protocol))
+		ip.hdr = skb_inner_network_header(skb);
+	else
+		ip.hdr = skb_network_header(skb);
+	l4.hdr = skb_checksum_start(skb);
 
 	/* initialize outer IP header fields */
 	if (ip.v4->version == 4) {
Signature · 1 line
2.7.4
#1 lore raw

PATCH v2 1/1 ice: Add mpls+tso support

Quoted 26 lines

-----Original Message-----
From: Joe Damato <[email protected]>
Sent: Friday, March 18, 2022 9:42 AM
To: intel-wired-lan at lists.osuosl.org; kuba at kernel.org; Nguyen, Anthony L <[email protected]>; Brandeburg, Jesse <[email protected]>; Kubalewski, Arkadiusz <[email protected]>
Cc: Joe Damato <[email protected]>
Subject: [next-queue v2] ice: Add mpls+tso support

Attempt to add mpls+tso support.

I don't have ice hardware available to test myself, but I just implemented this feature in i40e and thought it might be useful to implement for ice while this is fresh in my brain.

Hoping some one at intel will be able to test this on my behalf.

v1 -> v2:
- In ice_tx_csum, only set ICE_TX_FLAGS_IPV6 if ip v6 is detected explicitly. All cases other than ipv4 or ipv6 leave first->tx_flags unset, which allows for ice_tx_csum to return -1 to indicate that csum offloading is not supported.

Quoted 7 lines
---
 drivers/net/ethernet/intel/ice/ice_main.c |  4 +++-
drivers/net/ethernet/intel/ice/ice_txrx.c | 29 +++++++++++++++++++++----
----
 2 files changed, 24 insertions(+), 9 deletions(-)