]> git.proxmox.com Git - mirror_ubuntu-bionic-kernel.git/commitdiff
be2net: csum, tso and rss steering offload support for VxLAN
authorSathya Perla <sathya.perla@emulex.com>
Thu, 27 Mar 2014 05:16:19 +0000 (10:46 +0530)
committerDavid S. Miller <davem@davemloft.net>
Fri, 28 Mar 2014 18:24:54 +0000 (14:24 -0400)
This patch mainly implements the add/del_vxlan_port() methods by invoking
the needed FW cmds for supporting VxLAN offloads for Skyhawk-R.

Signed-off-by: Sathya Perla <sathya.perla@emulex.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
drivers/net/ethernet/emulex/benet/be.h
drivers/net/ethernet/emulex/benet/be_hw.h
drivers/net/ethernet/emulex/benet/be_main.c

index a587c8aa27ed865efa5a9a537c33043c7d745303..8ccaa2520dc3e9f154ae10b5e866e99173531274 100644 (file)
@@ -296,6 +296,7 @@ struct be_rx_compl_info {
        u8 qnq;
        u8 pkt_type;
        u8 ip_frag;
+       u8 tunneled;
 };
 
 struct be_rx_obj {
@@ -371,10 +372,11 @@ enum vf_state {
 #define BE_FLAGS_WORKER_SCHEDULED              (1 << 3)
 #define BE_FLAGS_VLAN_PROMISC                  (1 << 4)
 #define BE_FLAGS_NAPI_ENABLED                  (1 << 9)
-#define BE_UC_PMAC_COUNT               30
-#define BE_VF_UC_PMAC_COUNT            2
 #define BE_FLAGS_QNQ_ASYNC_EVT_RCVD            (1 << 11)
+#define BE_FLAGS_VXLAN_OFFLOADS                        (1 << 12)
 
+#define BE_UC_PMAC_COUNT                       30
+#define BE_VF_UC_PMAC_COUNT                    2
 /* Ethtool set_dump flags */
 #define LANCER_INITIATE_FW_DUMP                        0x1
 
@@ -494,6 +496,7 @@ struct be_adapter {
        u32 sli_family;
        u8 hba_port_num;
        u16 pvid;
+       __be16 vxlan_port;
        struct phy_info phy;
        u8 wol_cap;
        bool wol_en;
index 28ac8dd0beaab35b178b9e6c911c823d421b8613..3bd198550edbb95a64602e28535e8eb56758a220 100644 (file)
@@ -407,7 +407,8 @@ struct amap_eth_rx_compl_v1 {
        u8 vntagp;              /* dword 2 */
        u8 header_len[8];       /* dword 2 */
        u8 header_split[2];     /* dword 2 */
-       u8 rsvd1[13];           /* dword 2 */
+       u8 rsvd1[12];           /* dword 2 */
+       u8 tunneled;
        u8 valid;               /* dword 2 */
        u8 rsshash[32];         /* dword 3 */
 } __packed;
index 8790e5497ba0b2738f241d1acc93a4c9dd3258df..c89dc85ad8d60439f02039f96f7d756591415bf6 100644 (file)
@@ -23,6 +23,7 @@
 #include <linux/aer.h>
 #include <linux/if_bridge.h>
 #include <net/busy_poll.h>
+#include <net/vxlan.h>
 
 MODULE_VERSION(DRV_VER);
 MODULE_DEVICE_TABLE(pci, be_dev_ids);
@@ -718,10 +719,23 @@ static inline u16 be_get_tx_vlan_tag(struct be_adapter *adapter,
        return vlan_tag;
 }
 
+/* Used only for IP tunnel packets */
+static u16 skb_inner_ip_proto(struct sk_buff *skb)
+{
+       return (inner_ip_hdr(skb)->version == 4) ?
+               inner_ip_hdr(skb)->protocol : inner_ipv6_hdr(skb)->nexthdr;
+}
+
+static u16 skb_ip_proto(struct sk_buff *skb)
+{
+       return (ip_hdr(skb)->version == 4) ?
+               ip_hdr(skb)->protocol : ipv6_hdr(skb)->nexthdr;
+}
+
 static void wrb_fill_hdr(struct be_adapter *adapter, struct be_eth_hdr_wrb *hdr,
                struct sk_buff *skb, u32 wrb_cnt, u32 len, bool skip_hw_vlan)
 {
-       u16 vlan_tag;
+       u16 vlan_tag, proto;
 
        memset(hdr, 0, sizeof(*hdr));
 
@@ -734,9 +748,15 @@ static void wrb_fill_hdr(struct be_adapter *adapter, struct be_eth_hdr_wrb *hdr,
                if (skb_is_gso_v6(skb) && !lancer_chip(adapter))
                        AMAP_SET_BITS(struct amap_eth_hdr_wrb, lso6, hdr, 1);
        } else if (skb->ip_summed == CHECKSUM_PARTIAL) {
-               if (is_tcp_pkt(skb))
+               if (skb->encapsulation) {
+                       AMAP_SET_BITS(struct amap_eth_hdr_wrb, ipcs, hdr, 1);
+                       proto = skb_inner_ip_proto(skb);
+               } else {
+                       proto = skb_ip_proto(skb);
+               }
+               if (proto == IPPROTO_TCP)
                        AMAP_SET_BITS(struct amap_eth_hdr_wrb, tcpcs, hdr, 1);
-               else if (is_udp_pkt(skb))
+               else if (proto == IPPROTO_UDP)
                        AMAP_SET_BITS(struct amap_eth_hdr_wrb, udpcs, hdr, 1);
        }
 
@@ -1467,9 +1487,10 @@ static void be_rx_stats_update(struct be_rx_obj *rxo,
 static inline bool csum_passed(struct be_rx_compl_info *rxcp)
 {
        /* L4 checksum is not reliable for non TCP/UDP packets.
-        * Also ignore ipcksm for ipv6 pkts */
+        * Also ignore ipcksm for ipv6 pkts
+        */
        return (rxcp->tcpf || rxcp->udpf) && rxcp->l4_csum &&
-                               (rxcp->ip_csum || rxcp->ipv6);
+               (rxcp->ip_csum || rxcp->ipv6) && !rxcp->err;
 }
 
 static struct be_rx_page_info *get_rx_page_info(struct be_rx_obj *rxo)
@@ -1612,6 +1633,8 @@ static void be_rx_compl_process(struct be_rx_obj *rxo, struct napi_struct *napi,
        skb_record_rx_queue(skb, rxo - &adapter->rx_obj[0]);
        if (netdev->features & NETIF_F_RXHASH)
                skb_set_hash(skb, rxcp->rss_hash, PKT_HASH_TYPE_L3);
+
+       skb->encapsulation = rxcp->tunneled;
        skb_mark_napi_id(skb, napi);
 
        if (rxcp->vlanf)
@@ -1668,6 +1691,8 @@ static void be_rx_compl_process_gro(struct be_rx_obj *rxo,
        skb_record_rx_queue(skb, rxo - &adapter->rx_obj[0]);
        if (adapter->netdev->features & NETIF_F_RXHASH)
                skb_set_hash(skb, rxcp->rss_hash, PKT_HASH_TYPE_L3);
+
+       skb->encapsulation = rxcp->tunneled;
        skb_mark_napi_id(skb, napi);
 
        if (rxcp->vlanf)
@@ -1704,6 +1729,8 @@ static void be_parse_rx_compl_v1(struct be_eth_rx_compl *compl,
                                               compl);
        }
        rxcp->port = AMAP_GET_BITS(struct amap_eth_rx_compl_v1, port, compl);
+       rxcp->tunneled =
+               AMAP_GET_BITS(struct amap_eth_rx_compl_v1, tunneled, compl);
 }
 
 static void be_parse_rx_compl_v0(struct be_eth_rx_compl *compl,
@@ -2834,6 +2861,9 @@ static int be_open(struct net_device *netdev)
 
        netif_tx_start_all_queues(netdev);
        be_roce_dev_open(adapter);
+
+       if (skyhawk_chip(adapter))
+               vxlan_get_rx_port(netdev);
        return 0;
 err:
        be_close(adapter->netdev);
@@ -2989,6 +3019,19 @@ static void be_mac_clear(struct be_adapter *adapter)
        }
 }
 
+static void be_disable_vxlan_offloads(struct be_adapter *adapter)
+{
+       if (adapter->flags & BE_FLAGS_VXLAN_OFFLOADS)
+               be_cmd_manage_iface(adapter, adapter->if_handle,
+                                   OP_CONVERT_TUNNEL_TO_NORMAL);
+
+       if (adapter->vxlan_port)
+               be_cmd_set_vxlan_port(adapter, 0);
+
+       adapter->flags &= ~BE_FLAGS_VXLAN_OFFLOADS;
+       adapter->vxlan_port = 0;
+}
+
 static int be_clear(struct be_adapter *adapter)
 {
        be_cancel_worker(adapter);
@@ -2996,6 +3039,8 @@ static int be_clear(struct be_adapter *adapter)
        if (sriov_enabled(adapter))
                be_vf_clear(adapter);
 
+       be_disable_vxlan_offloads(adapter);
+
        /* delete the primary mac along with the uc-mac list */
        be_mac_clear(adapter);
 
@@ -4120,6 +4165,65 @@ static int be_ndo_bridge_getlink(struct sk_buff *skb, u32 pid, u32 seq,
                                       BRIDGE_MODE_VEPA : BRIDGE_MODE_VEB);
 }
 
+static void be_add_vxlan_port(struct net_device *netdev, sa_family_t sa_family,
+                             __be16 port)
+{
+       struct be_adapter *adapter = netdev_priv(netdev);
+       struct device *dev = &adapter->pdev->dev;
+       int status;
+
+       if (lancer_chip(adapter) || BEx_chip(adapter))
+               return;
+
+       if (adapter->flags & BE_FLAGS_VXLAN_OFFLOADS) {
+               dev_warn(dev, "Cannot add UDP port %d for VxLAN offloads\n",
+                        be16_to_cpu(port));
+               dev_info(dev,
+                        "Only one UDP port supported for VxLAN offloads\n");
+               return;
+       }
+
+       status = be_cmd_manage_iface(adapter, adapter->if_handle,
+                                    OP_CONVERT_NORMAL_TO_TUNNEL);
+       if (status) {
+               dev_warn(dev, "Failed to convert normal interface to tunnel\n");
+               goto err;
+       }
+
+       status = be_cmd_set_vxlan_port(adapter, port);
+       if (status) {
+               dev_warn(dev, "Failed to add VxLAN port\n");
+               goto err;
+       }
+       adapter->flags |= BE_FLAGS_VXLAN_OFFLOADS;
+       adapter->vxlan_port = port;
+
+       dev_info(dev, "Enabled VxLAN offloads for UDP port %d\n",
+                be16_to_cpu(port));
+       return;
+err:
+       be_disable_vxlan_offloads(adapter);
+       return;
+}
+
+static void be_del_vxlan_port(struct net_device *netdev, sa_family_t sa_family,
+                             __be16 port)
+{
+       struct be_adapter *adapter = netdev_priv(netdev);
+
+       if (lancer_chip(adapter) || BEx_chip(adapter))
+               return;
+
+       if (adapter->vxlan_port != port)
+               return;
+
+       be_disable_vxlan_offloads(adapter);
+
+       dev_info(&adapter->pdev->dev,
+                "Disabled VxLAN offloads for UDP port %d\n",
+                be16_to_cpu(port));
+}
+
 static const struct net_device_ops be_netdev_ops = {
        .ndo_open               = be_open,
        .ndo_stop               = be_close,
@@ -4142,14 +4246,22 @@ static const struct net_device_ops be_netdev_ops = {
        .ndo_bridge_setlink     = be_ndo_bridge_setlink,
        .ndo_bridge_getlink     = be_ndo_bridge_getlink,
 #ifdef CONFIG_NET_RX_BUSY_POLL
-       .ndo_busy_poll          = be_busy_poll
+       .ndo_busy_poll          = be_busy_poll,
 #endif
+       .ndo_add_vxlan_port     = be_add_vxlan_port,
+       .ndo_del_vxlan_port     = be_del_vxlan_port,
 };
 
 static void be_netdev_init(struct net_device *netdev)
 {
        struct be_adapter *adapter = netdev_priv(netdev);
 
+       if (skyhawk_chip(adapter)) {
+               netdev->hw_enc_features |= NETIF_F_IP_CSUM | NETIF_F_IPV6_CSUM |
+                                          NETIF_F_TSO | NETIF_F_TSO6 |
+                                          NETIF_F_GSO_UDP_TUNNEL;
+               netdev->hw_features |= NETIF_F_GSO_UDP_TUNNEL;
+       }
        netdev->hw_features |= NETIF_F_SG | NETIF_F_TSO | NETIF_F_TSO6 |
                NETIF_F_IP_CSUM | NETIF_F_IPV6_CSUM | NETIF_F_RXCSUM |
                NETIF_F_HW_VLAN_CTAG_TX;