New netdev feature flags for XDP

.travis.yml

@@ -0,0 +1,33 @@
+sudo: required
+language: bash
+dist: bionic
+services:
+    - docker
+
+env:
+    global:
+        - PROJECT_NAME='libbpf'
+        - AUTHOR_EMAIL="$(git log -1 --pretty=\"%aE\")"
+        - REPO_ROOT="$TRAVIS_BUILD_DIR"
+        - CI_ROOT="$REPO_ROOT/travis-ci"
+        - VMTEST_ROOT="$CI_ROOT/vmtest"
+
+addons:
+    apt:
+        packages:
+            - qemu-kvm
+            - zstd
+            - binutils-dev
+            - elfutils
+            - libcap-dev
+            - libelf-dev
+            - libdw-dev
+            - python3-docutils
+
+jobs:
+    include:
+        - stage: Builds & Tests
+          name: Kernel LATEST + selftests
+          language: bash
+          env: KERNEL=LATEST
+          script:  $CI_ROOT/vmtest/run_vmtest.sh || travis_terminate 1

drivers/net/ethernet/broadcom/bnxt/bnxt.c

@@ -12604,6 +12604,7 @@ static int bnxt_init_one(struct pci_dev *pdev, const struct pci_device_id *ent)
 	dev->features |= dev->hw_features | NETIF_F_HIGHDMA;
 	if (dev->features & NETIF_F_GRO_HW)
 		dev->features &= ~NETIF_F_LRO;
+	xdp_set_feature_flag(&dev->features);
 	dev->priv_flags |= IFF_UNICAST_FLT;
 
 #ifdef CONFIG_BNXT_SRIOV

drivers/net/ethernet/freescale/dpaa2/dpaa2-eth.c

@@ -4014,6 +4014,7 @@ static int dpaa2_eth_netdev_init(struct net_device *net_dev)
 			    NETIF_F_SG | NETIF_F_HIGHDMA |
 			    NETIF_F_LLTX | NETIF_F_HW_TC;
 	net_dev->hw_features = net_dev->features;
+	xdp_set_feature_flag(&net_dev->features);
 
 	return 0;
 }

drivers/net/ethernet/intel/i40e/i40e_main.c

@@ -12873,6 +12873,8 @@ static int i40e_config_netdev(struct i40e_vsi *vsi)
 	netdev->hw_features |= hw_features;
 
 	netdev->features |= hw_features | NETIF_F_HW_VLAN_CTAG_FILTER;
+	xdp_set_feature_flag(&netdev->features);
+	xsk_set_feature_flag(&netdev->features);
 	netdev->hw_enc_features |= NETIF_F_TSO_MANGLEID;
 
 	if (vsi->type == I40E_VSI_MAIN) {

drivers/net/ethernet/intel/ice/ice_main.c

@@ -13,6 +13,7 @@
 #include "ice_dcb_lib.h"
 #include "ice_dcb_nl.h"
 #include "ice_devlink.h"
+#include <net/xdp_sock_drv.h>
 
 #define DRV_SUMMARY	"Intel(R) Ethernet Connection E800 Series Linux Driver"
 static const char ice_driver_string[] = DRV_SUMMARY;
@@ -2941,6 +2942,9 @@ static void ice_set_netdev_features(struct net_device *netdev)
 
 	/* enable features */
 	netdev->features |= netdev->hw_features;
+	xdp_set_feature_flag(&netdev->features);
+	xsk_set_feature_flag(&netdev->features);
+
 	/* encap and VLAN devices inherit default, csumo and tso features */
 	netdev->hw_enc_features |= dflt_features | csumo_features |
 				   tso_features;

drivers/net/ethernet/intel/igb/igb_main.c

@@ -3293,6 +3293,8 @@ static int igb_probe(struct pci_dev *pdev, const struct pci_device_id *ent)
 			    NETIF_F_HW_VLAN_CTAG_RX |
 			    NETIF_F_HW_VLAN_CTAG_TX;
 
+	xdp_set_feature_flag(&netdev->features);
+
 	netdev->priv_flags |= IFF_SUPP_NOFCS;
 
 	netdev->priv_flags |= IFF_UNICAST_FLT;

drivers/net/ethernet/intel/ixgbe/ixgbe_main.c

@@ -10880,6 +10880,9 @@ static int ixgbe_probe(struct pci_dev *pdev, const struct pci_device_id *ent)
 	if (adapter->flags2 & IXGBE_FLAG2_RSC_ENABLED)
 		netdev->features |= NETIF_F_LRO;
 
+	xdp_set_feature_flag(&netdev->features);
+	xsk_set_feature_flag(&netdev->features);
+
 	if (ixgbe_check_fw_error(adapter)) {
 		err = -EIO;
 		goto err_sw_init;

drivers/net/ethernet/marvell/mvneta.c

@@ -5262,6 +5262,7 @@ static int mvneta_probe(struct platform_device *pdev)
 			NETIF_F_TSO | NETIF_F_RXCSUM;
 	dev->hw_features |= dev->features;
 	dev->vlan_features |= dev->features;
+	xdp_set_feature_flag(&dev->features);
 	dev->priv_flags |= IFF_LIVE_ADDR_CHANGE;
 	dev->gso_max_segs = MVNETA_MAX_TSO_SEGS;
 

drivers/net/ethernet/marvell/mvpp2/mvpp2_main.c

@@ -6475,6 +6475,7 @@ static int mvpp2_port_probe(struct platform_device *pdev,
 		mvpp2_set_hw_csum(port, port->pool_long->id);
 
 	dev->vlan_features |= features;
+	xdp_set_feature_flag(&dev->features);
 	dev->gso_max_segs = MVPP2_MAX_TSO_SEGS;
 	dev->priv_flags |= IFF_UNICAST_FLT;
 

drivers/net/ethernet/mellanox/mlx5/core/en_main.c

@@ -4980,6 +4980,8 @@ static void mlx5e_build_nic_netdev(struct net_device *netdev)
 
 	netdev->features         |= NETIF_F_HIGHDMA;
 	netdev->features         |= NETIF_F_HW_VLAN_STAG_FILTER;
+	xdp_set_feature_flag(&netdev->features);
+	xsk_set_feature_flag(&netdev->features);
 
 	netdev->priv_flags       |= IFF_UNICAST_FLT;
 

drivers/net/ethernet/qlogic/qede/qede_main.c

@@ -839,6 +839,7 @@ static void qede_init_ndev(struct qede_dev *edev)
 	ndev->features = hw_features | NETIF_F_RXHASH | NETIF_F_RXCSUM |
 			 NETIF_F_HW_VLAN_CTAG_RX | NETIF_F_HIGHDMA |
 			 NETIF_F_HW_VLAN_CTAG_FILTER | NETIF_F_HW_VLAN_CTAG_TX;
+	xdp_set_feature_flag(&ndev->features);
 
 	ndev->hw_features = hw_features;
 

drivers/net/ethernet/sfc/efx.c

@@ -1101,6 +1101,7 @@ static int efx_pci_probe(struct pci_dev *pci_dev,
 	efx = netdev_priv(net_dev);
 	efx->type = (const struct efx_nic_type *) entry->driver_data;
 	efx->fixed_features |= NETIF_F_HIGHDMA;
+	xdp_set_feature_flag(&efx->fixed_features);
 
 	pci_set_drvdata(pci_dev, efx);
 	SET_NETDEV_DEV(net_dev, &pci_dev->dev);

drivers/net/ethernet/socionext/netsec.c

@@ -2099,6 +2099,7 @@ static int netsec_probe(struct platform_device *pdev)
 	ndev->features |= NETIF_F_HIGHDMA | NETIF_F_RXCSUM | NETIF_F_GSO |
 				NETIF_F_IP_CSUM | NETIF_F_IPV6_CSUM;
 	ndev->hw_features = ndev->features;
+	xdp_set_feature_flag(&ndev->features);
 
 	priv->rx_cksum_offload_flag = true;
 

drivers/net/ethernet/ti/cpsw.c

@@ -1475,6 +1475,7 @@ static int cpsw_probe_dual_emac(struct cpsw_priv *priv)
 	priv_sl2->emac_port = 1;
 	cpsw->slaves[1].ndev = ndev;
 	ndev->features |= NETIF_F_HW_VLAN_CTAG_FILTER | NETIF_F_HW_VLAN_CTAG_RX;
+	xdp_set_feature_flag(&ndev->features);
 
 	ndev->netdev_ops = &cpsw_netdev_ops;
 	ndev->ethtool_ops = &cpsw_ethtool_ops;
@@ -1654,6 +1655,7 @@ static int cpsw_probe(struct platform_device *pdev)
 	cpsw->slaves[0].ndev = ndev;
 
 	ndev->features |= NETIF_F_HW_VLAN_CTAG_FILTER | NETIF_F_HW_VLAN_CTAG_RX;
+	xdp_set_feature_flag(&ndev->features);
 
 	ndev->netdev_ops = &cpsw_netdev_ops;
 	ndev->ethtool_ops = &cpsw_ethtool_ops;

drivers/net/ethernet/ti/cpsw_new.c

@@ -1416,6 +1416,8 @@ static int cpsw_create_ports(struct cpsw_common *cpsw)
 		ndev->features |= NETIF_F_HW_VLAN_CTAG_FILTER |
 				  NETIF_F_HW_VLAN_CTAG_RX | NETIF_F_NETNS_LOCAL;
 
+		xdp_set_feature_flag(&ndev->features);
+
 		ndev->netdev_ops = &cpsw_netdev_ops;
 		ndev->ethtool_ops = &cpsw_ethtool_ops;
 		SET_NETDEV_DEV(ndev, dev);

drivers/net/tun.c

@@ -2721,6 +2721,10 @@ static int tun_set_iff(struct net *net, struct file *file, struct ifreq *ifr)
 				     ~(NETIF_F_HW_VLAN_CTAG_TX |
 				       NETIF_F_HW_VLAN_STAG_TX);
 
+		/* Currently tap does not support XDP, only tun does. */
+		if (tun->flags == IFF_TUN)
+			xdp_set_feature_flag(&dev->features);
+
 		tun->flags = (tun->flags & ~TUN_FEATURES) |
 			      (ifr->ifr_flags & TUN_FEATURES);
 

drivers/net/veth.c

@@ -1266,6 +1266,7 @@ static void veth_setup(struct net_device *dev)
 	dev->hw_features = VETH_FEATURES;
 	dev->hw_enc_features = VETH_FEATURES;
 	dev->mpls_features = NETIF_F_HW_CSUM | NETIF_F_GSO_SOFTWARE;
+	xdp_set_feature_flag(&dev->features);
 }
 
 /*

drivers/net/virtio_net.c

@@ -3017,6 +3017,7 @@ static int virtnet_probe(struct virtio_device *vdev)
 		dev->hw_features |= NETIF_F_LRO;
 
 	dev->vlan_features = dev->features;
+	xdp_set_feature_flag(&dev->features);
 
 	/* MTU range: 68 - 65535 */
 	dev->min_mtu = MIN_MTU;

drivers/net/xen-netfront.c

@@ -1555,6 +1555,7 @@ static struct net_device *xennet_create_dev(struct xenbus_device *dev)
          * negotiate with the backend regarding supported features.
          */
 	netdev->features |= netdev->hw_features;
+	xdp_set_feature_flag(&netdev->features);
 
 	netdev->ethtool_ops = &xennet_ethtool_ops;
 	netdev->min_mtu = ETH_MIN_MTU;

include/linux/netdev_features.h

@@ -85,6 +85,8 @@ enum {
 
 	NETIF_F_HW_MACSEC_BIT,		/* Offload MACsec operations */
 
+	NETIF_F_XDP_BIT,		/* XDP support */
+	NETIF_F_AF_XDP_ZC_BIT,		/* AF_XDP zero-copy support */
 	/*
 	 * Add your fresh new feature above and remember to update
 	 * netdev_features_strings[] in net/core/ethtool.c and maybe
@@ -157,6 +159,9 @@ enum {
 #define NETIF_F_GRO_FRAGLIST	__NETIF_F(GRO_FRAGLIST)
 #define NETIF_F_GSO_FRAGLIST	__NETIF_F(GSO_FRAGLIST)
 #define NETIF_F_HW_MACSEC	__NETIF_F(HW_MACSEC)
+#define NETIF_F_XDP		__NETIF_F(XDP)
+#define NETIF_F_AF_XDP_ZC	__NETIF_F(AF_XDP_ZC)
+
 
 /* Finds the next feature with the highest number of the range of start till 0.
  */
@@ -182,6 +187,7 @@ static inline int find_next_netdev_feature(u64 feature, unsigned long start)
 /* Features valid for ethtool to change */
 /* = all defined minus driver/device-class-related */
 #define NETIF_F_NEVER_CHANGE	(NETIF_F_VLAN_CHALLENGED | \
+				 NETIF_F_XDP | NETIF_F_AF_XDP_ZC | \
 				 NETIF_F_LLTX | NETIF_F_NETNS_LOCAL)
 
 /* remember that ((t)1 << t_BITS) is undefined in C99 */

include/net/xdp.h

@@ -254,6 +254,19 @@ struct xdp_attachment_info {
 	u32 flags;
 };
 
+#if defined(CONFIG_NET) && defined(CONFIG_BPF_SYSCALL)
+static __always_inline void
+xdp_set_feature_flag(netdev_features_t *features)
+{
+	*features |= NETIF_F_XDP;
+}
+#else
+static __always_inline void
+xdp_set_feature_flag(netdev_features_t *features)
+{
+}
+#endif
+
 struct netdev_bpf;
 bool xdp_attachment_flags_ok(struct xdp_attachment_info *info,
 			     struct netdev_bpf *bpf);

include/net/xdp_sock_drv.h

@@ -22,6 +22,12 @@ void xsk_clear_rx_need_wakeup(struct xsk_buff_pool *pool);
 void xsk_clear_tx_need_wakeup(struct xsk_buff_pool *pool);
 bool xsk_uses_need_wakeup(struct xsk_buff_pool *pool);
 
+static __always_inline void
+xsk_set_feature_flag(netdev_features_t *features)
+{
+	*features |= NETIF_F_AF_XDP_ZC;
+}
+
 static inline u32 xsk_pool_get_headroom(struct xsk_buff_pool *pool)
 {
 	return XDP_PACKET_HEADROOM + pool->headroom;
@@ -235,6 +241,11 @@ static inline void xsk_buff_raw_dma_sync_for_device(struct xsk_buff_pool *pool,
 {
 }
 
+static __always_inline void
+xsk_set_feature_flag(netdev_features_t *features)
+{
+}
+
 #endif /* CONFIG_XDP_SOCKETS */
 
 #endif /* _LINUX_XDP_SOCK_DRV_H */

include/uapi/linux/if_xdp.h

@@ -25,6 +25,7 @@
  * application.
  */
 #define XDP_USE_NEED_WAKEUP (1 << 3)
+#define XDP_CHECK_NATIVE_MODE (1 << 4)
 
 /* Flags for xsk_umem_config flags */
 #define XDP_UMEM_UNALIGNED_CHUNK_FLAG (1 << 0)

net/ethtool/common.c

@@ -68,6 +68,8 @@ const char netdev_features_strings[NETDEV_FEATURE_COUNT][ETH_GSTRING_LEN] = {
 	[NETIF_F_HW_TLS_RX_BIT] =	 "tls-hw-rx-offload",
 	[NETIF_F_GRO_FRAGLIST_BIT] =	 "rx-gro-list",
 	[NETIF_F_HW_MACSEC_BIT] =	 "macsec-hw-offload",
+	[NETIF_F_XDP_BIT] =              "xdp",
+	[NETIF_F_AF_XDP_ZC_BIT] =        "af-xdp-zc",
 };
 
 const char

net/xdp/xsk.c

@@ -658,7 +658,7 @@ static int xsk_bind(struct socket *sock, struct sockaddr *addr, int addr_len)
 
 	flags = sxdp->sxdp_flags;
 	if (flags & ~(XDP_SHARED_UMEM | XDP_COPY | XDP_ZEROCOPY |
-		      XDP_USE_NEED_WAKEUP))
+		      XDP_USE_NEED_WAKEUP | XDP_CHECK_NATIVE_MODE))
 		return -EINVAL;
 
 	rtnl_lock();
@@ -686,7 +686,7 @@ static int xsk_bind(struct socket *sock, struct sockaddr *addr, int addr_len)
 		struct socket *sock;
 
 		if ((flags & XDP_COPY) || (flags & XDP_ZEROCOPY) ||
-		    (flags & XDP_USE_NEED_WAKEUP)) {
+		    (flags & XDP_USE_NEED_WAKEUP) || (flags & XDP_CHECK_NATIVE_MODE)) {
 			/* Cannot specify flags for shared sockets. */
 			err = -EINVAL;
 			goto out_unlock;

net/xdp/xsk_buff_pool.c

@@ -123,18 +123,32 @@ static void xp_disable_drv_zc(struct xsk_buff_pool *pool)
 static int __xp_assign_dev(struct xsk_buff_pool *pool,
 			   struct net_device *netdev, u16 queue_id, u16 flags)
 {
-	bool force_zc, force_copy;
+	bool force_zc, force_copy, force_check;
 	struct netdev_bpf bpf;
 	int err = 0;
 
 	ASSERT_RTNL();
 
 	force_zc = flags & XDP_ZEROCOPY;
 	force_copy = flags & XDP_COPY;
+	force_check = flags & XDP_CHECK_NATIVE_MODE;
+
 
 	if (force_zc && force_copy)
 		return -EINVAL;
 
+	if (!(flags & XDP_SHARED_UMEM)) {
+		if (force_check) {
+			/* forbid driver mode without full XDP support */
+			if (!(NETIF_F_XDP & netdev->features))
+				return -EOPNOTSUPP;
+		} else {
+			/* forbid skb mode and zero copy */
+			if (force_zc)
+				return -EINVAL;
+		}
+	}
+
 	if (xsk_get_pool_from_qid(netdev, queue_id))
 		return -EBUSY;
 
@@ -159,8 +173,8 @@ static int __xp_assign_dev(struct xsk_buff_pool *pool,
 		/* For copy-mode, we are done. */
 		return 0;
 
-	if (!netdev->netdev_ops->ndo_bpf ||
-	    !netdev->netdev_ops->ndo_xsk_wakeup) {
+	if (!(NETIF_F_XDP & netdev->features &&
+	      NETIF_F_AF_XDP_ZC & netdev->features)) {
 		err = -EOPNOTSUPP;
 		goto err_unreg_pool;
 	}
@@ -206,6 +220,7 @@ int xp_assign_dev_shared(struct xsk_buff_pool *pool, struct xdp_umem *umem,
 		return -EINVAL;
 
 	flags = umem->zc ? XDP_ZEROCOPY : XDP_COPY;
+	flags |= XDP_SHARED_UMEM;
 	if (pool->uses_need_wakeup)
 		flags |= XDP_USE_NEED_WAKEUP;