Merge branch 'nfp-ring-reconfig-and-xdp-support'

Jakub Kicinski says:

====================
ring reconfiguration and XDP support

This set adds support for ethtool channel API and XDP.

I kick off with ethtool get_channels() implementation.
set_channels() needs some preparations to get right.  I follow
the prepare/commit paradigm and allocate all resources before
stopping the device.  It has already been done for ndo_change_mtu
and ethtool set_ringparam(), it makes sense now to consolidate all
the required logic in one place.

XDP support requires splitting TX rings into two classes -
for the stack and for XDP.  The ring structures are identical.
The differences are in how they are connected to IRQ vector
structs and how the completion/cleanup works.  When XDP is enabled
I switch from the frag allocator to page-per-packet and map buffers
BIDIRECTIONALly.

Last but not least XDP offload is added (the patch just takes
care of the small formal differences between cls_bpf and XDP).

There is a tiny & trivial DebugFS patch in the mix, I hope it can
be taken via net-next provided we have the right Acks.

Resending with improved commit message and CCing more people on patch 10.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>

drivers/net/ethernet/netronome/nfp/nfp_bpf.h

@@ -62,6 +62,7 @@ enum nfp_bpf_action_type {
 	NN_ACT_TC_DROP,
 	NN_ACT_TC_REDIR,
 	NN_ACT_DIRECT,
+	NN_ACT_XDP,
 };
 
 /* Software register representation, hardware encoding in asm.h */

drivers/net/ethernet/netronome/nfp/nfp_bpf_jit.c

@@ -1126,7 +1126,7 @@ static int data_ind_ld4(struct nfp_prog *nfp_prog, struct nfp_insn_meta *meta)
 				     meta->insn.src_reg * 2, true, 4);
 }
 
-static int mem_ldx4(struct nfp_prog *nfp_prog, struct nfp_insn_meta *meta)
+static int mem_ldx4_skb(struct nfp_prog *nfp_prog, struct nfp_insn_meta *meta)
 {
 	if (meta->insn.off == offsetof(struct sk_buff, len))
 		emit_alu(nfp_prog, reg_both(meta->insn.dst_reg * 2),
@@ -1134,12 +1134,42 @@ static int mem_ldx4(struct nfp_prog *nfp_prog, struct nfp_insn_meta *meta)
 	else
 		return -ENOTSUPP;
 
-	wrp_immed(nfp_prog, reg_both(meta->insn.dst_reg * 2 + 1), 0);
+	return 0;
+}
+
+static int mem_ldx4_xdp(struct nfp_prog *nfp_prog, struct nfp_insn_meta *meta)
+{
+	u32 dst = reg_both(meta->insn.dst_reg * 2);
+
+	if (meta->insn.off != offsetof(struct xdp_md, data) &&
+	    meta->insn.off != offsetof(struct xdp_md, data_end))
+		return -ENOTSUPP;
+
+	emit_alu(nfp_prog, dst, reg_none(), ALU_OP_NONE, NFP_BPF_ABI_PKT);
+
+	if (meta->insn.off == offsetof(struct xdp_md, data))
+		return 0;
+
+	emit_alu(nfp_prog, dst,	dst, ALU_OP_ADD, NFP_BPF_ABI_LEN);
 
 	return 0;
 }
 
-static int mem_stx4(struct nfp_prog *nfp_prog, struct nfp_insn_meta *meta)
+static int mem_ldx4(struct nfp_prog *nfp_prog, struct nfp_insn_meta *meta)
+{
+	int ret;
+
+	if (nfp_prog->act == NN_ACT_XDP)
+		ret = mem_ldx4_xdp(nfp_prog, meta);
+	else
+		ret = mem_ldx4_skb(nfp_prog, meta);
+
+	wrp_immed(nfp_prog, reg_both(meta->insn.dst_reg * 2 + 1), 0);
+
+	return ret;
+}
+
+static int mem_stx4_skb(struct nfp_prog *nfp_prog, struct nfp_insn_meta *meta)
 {
 	if (meta->insn.off == offsetof(struct sk_buff, mark))
 		return wrp_set_mark(nfp_prog, meta->insn.src_reg * 2);
@@ -1147,6 +1177,18 @@ static int mem_stx4(struct nfp_prog *nfp_prog, struct nfp_insn_meta *meta)
 	return -ENOTSUPP;
 }
 
+static int mem_stx4_xdp(struct nfp_prog *nfp_prog, struct nfp_insn_meta *meta)
+{
+	return -ENOTSUPP;
+}
+
+static int mem_stx4(struct nfp_prog *nfp_prog, struct nfp_insn_meta *meta)
+{
+	if (nfp_prog->act == NN_ACT_XDP)
+		return mem_stx4_xdp(nfp_prog, meta);
+	return mem_stx4_skb(nfp_prog, meta);
+}
+
 static int jump(struct nfp_prog *nfp_prog, struct nfp_insn_meta *meta)
 {
 	if (meta->insn.off < 0) /* TODO */
@@ -1530,6 +1572,47 @@ static void nfp_outro_tc_da(struct nfp_prog *nfp_prog)
 	emit_ld_field(nfp_prog, reg_a(0), 0xc, reg_b(2), SHF_SC_L_SHF, 16);
 }
 
+static void nfp_outro_xdp(struct nfp_prog *nfp_prog)
+{
+	/* XDP return codes:
+	 *   0 aborted  0x82 -> drop,  count as stat3
+	 *   1    drop  0x22 -> drop,  count as stat1
+	 *   2    pass  0x11 -> pass,  count as stat0
+	 *   3      tx  0x44 -> redir, count as stat2
+	 *   * unknown  0x82 -> drop,  count as stat3
+	 */
+	/* Target for aborts */
+	nfp_prog->tgt_abort = nfp_prog_current_offset(nfp_prog);
+
+	emit_br_def(nfp_prog, nfp_prog->tgt_done, 2);
+
+	emit_alu(nfp_prog, reg_a(0),
+		 reg_none(), ALU_OP_NONE, NFP_BPF_ABI_FLAGS);
+	emit_ld_field(nfp_prog, reg_a(0), 0xc, reg_imm(0x82), SHF_SC_L_SHF, 16);
+
+	/* Target for normal exits */
+	nfp_prog->tgt_out = nfp_prog_current_offset(nfp_prog);
+
+	/* if R0 > 3 jump to abort */
+	emit_alu(nfp_prog, reg_none(), reg_imm(3), ALU_OP_SUB, reg_b(0));
+	emit_br(nfp_prog, BR_BLO, nfp_prog->tgt_abort, 0);
+
+	wrp_immed(nfp_prog, reg_b(2), 0x44112282);
+
+	emit_shf(nfp_prog, reg_a(1),
+		 reg_none(), SHF_OP_NONE, reg_b(0), SHF_SC_L_SHF, 3);
+
+	emit_alu(nfp_prog, reg_none(), reg_a(1), ALU_OP_OR, reg_imm(0));
+	emit_shf(nfp_prog, reg_b(2),
+		 reg_imm(0xff), SHF_OP_AND, reg_b(2), SHF_SC_R_SHF, 0);
+
+	emit_br_def(nfp_prog, nfp_prog->tgt_done, 2);
+
+	emit_alu(nfp_prog, reg_a(0),
+		 reg_none(), ALU_OP_NONE, NFP_BPF_ABI_FLAGS);
+	emit_ld_field(nfp_prog, reg_a(0), 0xc, reg_b(2), SHF_SC_L_SHF, 16);
+}
+
 static void nfp_outro(struct nfp_prog *nfp_prog)
 {
 	switch (nfp_prog->act) {
@@ -1540,6 +1623,9 @@ static void nfp_outro(struct nfp_prog *nfp_prog)
 	case NN_ACT_TC_REDIR:
 		nfp_outro_tc_legacy(nfp_prog);
 		break;
+	case NN_ACT_XDP:
+		nfp_outro_xdp(nfp_prog);
+		break;
 	}
 }
 

drivers/net/ethernet/netronome/nfp/nfp_bpf_verifier.c

@@ -80,6 +80,9 @@ nfp_bpf_check_exit(struct nfp_prog *nfp_prog,
 {
 	const struct bpf_reg_state *reg0 = &env->cur_state.regs[0];
 
+	if (nfp_prog->act == NN_ACT_XDP)
+		return 0;
+
 	if (reg0->type != CONST_IMM) {
 		pr_info("unsupported exit state: %d, imm: %llx\n",
 			reg0->type, reg0->imm);

drivers/net/ethernet/netronome/nfp/nfp_net.h

@@ -171,7 +171,10 @@ struct nfp_net_tx_desc {
  *		on the head's buffer). Equal to skb->len for non-TSO packets.
  */
 struct nfp_net_tx_buf {
+	union {
 		struct sk_buff *skb;
+		void *frag;
+	};
 	dma_addr_t dma_addr;
 	short int fidx;
 	u16 pkt_cnt;
@@ -341,6 +344,7 @@ struct nfp_net_rx_ring {
  * @napi:           NAPI structure for this ring vec
  * @tx_ring:        Pointer to TX ring
  * @rx_ring:        Pointer to RX ring
+ * @xdp_ring:	    Pointer to an extra TX ring for XDP
  * @irq_idx:        Index into MSI-X table
  * @rx_sync:	    Seqlock for atomic updates of RX stats
  * @rx_pkts:        Number of received packets
@@ -384,6 +388,8 @@ struct nfp_net_r_vector {
 	u64 hw_csum_rx_inner_ok;
 	u64 hw_csum_rx_error;
 
+	struct nfp_net_tx_ring *xdp_ring;
+
 	struct u64_stats_sync tx_sync;
 	u64 tx_pkts;
 	u64 tx_bytes;
@@ -429,9 +435,11 @@ struct nfp_stat_pair {
  * @is_vf:              Is the driver attached to a VF?
  * @fw_loaded:          Is the firmware loaded?
  * @bpf_offload_skip_sw:  Offloaded BPF program will not be rerun by cls_bpf
+ * @bpf_offload_xdp:	Offloaded BPF program is XDP
  * @ctrl:               Local copy of the control register/word.
  * @fl_bufsz:           Currently configured size of the freelist buffers
  * @rx_offset:		Offset in the RX buffers where packet data starts
+ * @xdp_prog:		Installed XDP program
  * @cpp:                Pointer to the CPP handle
  * @nfp_dev_cpp:        Pointer to the NFP Device handle
  * @ctrl_area:          Pointer to the CPP area for the control BAR
@@ -451,6 +459,7 @@ struct nfp_stat_pair {
  * @max_tx_rings:       Maximum number of TX rings supported by the Firmware
  * @max_rx_rings:       Maximum number of RX rings supported by the Firmware
  * @num_tx_rings:       Currently configured number of TX rings
+ * @num_stack_tx_rings:	Number of TX rings used by the stack (not XDP)
  * @num_rx_rings:       Currently configured number of RX rings
  * @txd_cnt:            Size of the TX ring in number of descriptors
  * @rxd_cnt:            Size of the RX ring in number of descriptors
@@ -494,12 +503,15 @@ struct nfp_net {
 	unsigned is_vf:1;
 	unsigned fw_loaded:1;
 	unsigned bpf_offload_skip_sw:1;
+	unsigned bpf_offload_xdp:1;
 
 	u32 ctrl;
 	u32 fl_bufsz;
 
 	u32 rx_offset;
 
+	struct bpf_prog *xdp_prog;
+
 	struct nfp_net_tx_ring *tx_rings;
 	struct nfp_net_rx_ring *rx_rings;
 
@@ -532,6 +544,7 @@ struct nfp_net {
 	unsigned int max_rx_rings;
 
 	unsigned int num_tx_rings;
+	unsigned int num_stack_tx_rings;
 	unsigned int num_rx_rings;
 
 	int stride_tx;
@@ -583,6 +596,13 @@ struct nfp_net {
 	struct dentry *debugfs_dir;
 };
 
+struct nfp_net_ring_set {
+	unsigned int n_rings;
+	unsigned int mtu;
+	unsigned int dcnt;
+	void *rings;
+};
+
 /* Functions to read/write from/to a BAR
  * Performs any endian conversion necessary.
  */
@@ -771,7 +791,9 @@ void nfp_net_rss_write_key(struct nfp_net *nn);
 void nfp_net_coalesce_write_cfg(struct nfp_net *nn);
 int nfp_net_irqs_alloc(struct nfp_net *nn);
 void nfp_net_irqs_disable(struct nfp_net *nn);
-int nfp_net_set_ring_size(struct nfp_net *nn, u32 rxd_cnt, u32 txd_cnt);
+int
+nfp_net_ring_reconfig(struct nfp_net *nn, struct bpf_prog **xdp_prog,
+		      struct nfp_net_ring_set *rx, struct nfp_net_ring_set *tx);
 
 #ifdef CONFIG_NFP_NET_DEBUG
 void nfp_net_debugfs_create(void);
@@ -797,8 +819,6 @@ static inline void nfp_net_debugfs_adapter_del(struct nfp_net *nn)
 #endif /* CONFIG_NFP_NET_DEBUG */
 
 void nfp_net_filter_stats_timer(unsigned long data);
-int
-nfp_net_bpf_offload(struct nfp_net *nn, u32 handle, __be16 proto,
-		    struct tc_cls_bpf_offload *cls_bpf);
+int nfp_net_bpf_offload(struct nfp_net *nn, struct tc_cls_bpf_offload *cls_bpf);
 
 #endif /* _NFP_NET_H_ */

drivers/net/ethernet/netronome/nfp/nfp_net_common.c

@@ -41,6 +41,7 @@
  *          Chris Telfer <chris.telfer@netronome.com>
  */
 
+#include <linux/bpf.h>
 #include <linux/module.h>
 #include <linux/kernel.h>
 #include <linux/init.h>
@@ -490,11 +491,12 @@ static void nfp_net_irqs_assign(struct net_device *netdev)
 
 	nn->num_rx_rings = min(nn->num_r_vecs, nn->num_rx_rings);
 	nn->num_tx_rings = min(nn->num_r_vecs, nn->num_tx_rings);
+	nn->num_stack_tx_rings = nn->num_tx_rings;
 
 	nn->lsc_handler = nfp_net_irq_lsc;
 	nn->exn_handler = nfp_net_irq_exn;
 
-	for (r = 0; r < nn->num_r_vecs; r++) {
+	for (r = 0; r < nn->max_r_vecs; r++) {
 		r_vec = &nn->r_vecs[r];
 		r_vec->nfp_net = nn;
 		r_vec->handler = nfp_net_irq_rxtx;
@@ -713,6 +715,13 @@ static void nfp_net_tx_csum(struct nfp_net *nn, struct nfp_net_r_vector *r_vec,
 	u64_stats_update_end(&r_vec->tx_sync);
 }
 
+static void nfp_net_tx_xmit_more_flush(struct nfp_net_tx_ring *tx_ring)
+{
+	wmb();
+	nfp_qcp_wr_ptr_add(tx_ring->qcp_q, tx_ring->wr_ptr_add);
+	tx_ring->wr_ptr_add = 0;
+}
+
 /**
  * nfp_net_tx() - Main transmit entry point
  * @skb:    SKB to transmit
@@ -827,12 +836,8 @@ static int nfp_net_tx(struct sk_buff *skb, struct net_device *netdev)
 		nfp_net_tx_ring_stop(nd_q, tx_ring);
 
 	tx_ring->wr_ptr_add += nr_frags + 1;
-	if (!skb->xmit_more || netif_xmit_stopped(nd_q)) {
-		/* force memory write before we let HW know */
-		wmb();
-		nfp_qcp_wr_ptr_add(tx_ring->qcp_q, tx_ring->wr_ptr_add);
-		tx_ring->wr_ptr_add = 0;
-	}
+	if (!skb->xmit_more || netif_xmit_stopped(nd_q))
+		nfp_net_tx_xmit_more_flush(tx_ring);
 
 	skb_tx_timestamp(skb);
 
@@ -954,6 +959,56 @@ static void nfp_net_tx_complete(struct nfp_net_tx_ring *tx_ring)
 		  tx_ring->rd_p, tx_ring->wr_p, tx_ring->cnt);
 }
 
+static void nfp_net_xdp_complete(struct nfp_net_tx_ring *tx_ring)
+{
+	struct nfp_net_r_vector *r_vec = tx_ring->r_vec;
+	struct nfp_net *nn = r_vec->nfp_net;
+	u32 done_pkts = 0, done_bytes = 0;
+	int idx, todo;
+	u32 qcp_rd_p;
+
+	/* Work out how many descriptors have been transmitted */
+	qcp_rd_p = nfp_qcp_rd_ptr_read(tx_ring->qcp_q);
+
+	if (qcp_rd_p == tx_ring->qcp_rd_p)
+		return;
+
+	if (qcp_rd_p > tx_ring->qcp_rd_p)
+		todo = qcp_rd_p - tx_ring->qcp_rd_p;
+	else
+		todo = qcp_rd_p + tx_ring->cnt - tx_ring->qcp_rd_p;
+
+	while (todo--) {
+		idx = tx_ring->rd_p & (tx_ring->cnt - 1);
+		tx_ring->rd_p++;
+
+		if (!tx_ring->txbufs[idx].frag)
+			continue;
+
+		nfp_net_dma_unmap_rx(nn, tx_ring->txbufs[idx].dma_addr,
+				     nn->fl_bufsz, DMA_BIDIRECTIONAL);
+		__free_page(virt_to_page(tx_ring->txbufs[idx].frag));
+
+		done_pkts++;
+		done_bytes += tx_ring->txbufs[idx].real_len;
+
+		tx_ring->txbufs[idx].dma_addr = 0;
+		tx_ring->txbufs[idx].frag = NULL;
+		tx_ring->txbufs[idx].fidx = -2;
+	}
+
+	tx_ring->qcp_rd_p = qcp_rd_p;
+
+	u64_stats_update_begin(&r_vec->tx_sync);
+	r_vec->tx_bytes += done_bytes;
+	r_vec->tx_pkts += done_pkts;
+	u64_stats_update_end(&r_vec->tx_sync);
+
+	WARN_ONCE(tx_ring->wr_p - tx_ring->rd_p > tx_ring->cnt,
+		  "TX ring corruption rd_p=%u wr_p=%u cnt=%u\n",
+		  tx_ring->rd_p, tx_ring->wr_p, tx_ring->cnt);
+}
+
 /**
  * nfp_net_tx_ring_reset() - Free any untransmitted buffers and reset pointers
  * @nn:		NFP Net device
@@ -964,39 +1019,47 @@ static void nfp_net_tx_complete(struct nfp_net_tx_ring *tx_ring)
 static void
 nfp_net_tx_ring_reset(struct nfp_net *nn, struct nfp_net_tx_ring *tx_ring)
 {
+	struct nfp_net_r_vector *r_vec = tx_ring->r_vec;
 	const struct skb_frag_struct *frag;
-	struct netdev_queue *nd_q;
 	struct pci_dev *pdev = nn->pdev;
+	struct netdev_queue *nd_q;
 
 	while (tx_ring->rd_p != tx_ring->wr_p) {
-		int nr_frags, fidx, idx;
-		struct sk_buff *skb;
+		struct nfp_net_tx_buf *tx_buf;
+		int idx;
 
 		idx = tx_ring->rd_p & (tx_ring->cnt - 1);
-		skb = tx_ring->txbufs[idx].skb;
-		nr_frags = skb_shinfo(skb)->nr_frags;
-		fidx = tx_ring->txbufs[idx].fidx;
+		tx_buf = &tx_ring->txbufs[idx];
 
-		if (fidx == -1) {
+		if (tx_ring == r_vec->xdp_ring) {
+			nfp_net_dma_unmap_rx(nn, tx_buf->dma_addr,
+					     nn->fl_bufsz, DMA_BIDIRECTIONAL);
+			__free_page(virt_to_page(tx_ring->txbufs[idx].frag));
+		} else {
+			struct sk_buff *skb = tx_ring->txbufs[idx].skb;
+			int nr_frags = skb_shinfo(skb)->nr_frags;
+
+			if (tx_buf->fidx == -1) {
 				/* unmap head */
-			dma_unmap_single(&pdev->dev,
-					 tx_ring->txbufs[idx].dma_addr,
-					 skb_headlen(skb), DMA_TO_DEVICE);
+				dma_unmap_single(&pdev->dev, tx_buf->dma_addr,
+						 skb_headlen(skb),
+						 DMA_TO_DEVICE);
 			} else {
 				/* unmap fragment */
-			frag = &skb_shinfo(skb)->frags[fidx];
-			dma_unmap_page(&pdev->dev,
-				       tx_ring->txbufs[idx].dma_addr,
-				       skb_frag_size(frag), DMA_TO_DEVICE);
+				frag = &skb_shinfo(skb)->frags[tx_buf->fidx];
+				dma_unmap_page(&pdev->dev, tx_buf->dma_addr,
+					       skb_frag_size(frag),
+					       DMA_TO_DEVICE);
 			}
 
 			/* check for last gather fragment */
-		if (fidx == nr_frags - 1)
+			if (tx_buf->fidx == nr_frags - 1)
 				dev_kfree_skb_any(skb);
+		}
 
-		tx_ring->txbufs[idx].dma_addr = 0;
-		tx_ring->txbufs[idx].skb = NULL;
-		tx_ring->txbufs[idx].fidx = -2;
+		tx_buf->dma_addr = 0;
+		tx_buf->skb = NULL;
+		tx_buf->fidx = -2;
 
 		tx_ring->qcp_rd_p++;
 		tx_ring->rd_p++;
@@ -1008,6 +1071,9 @@ nfp_net_tx_ring_reset(struct nfp_net *nn, struct nfp_net_tx_ring *tx_ring)
 	tx_ring->qcp_rd_p = 0;
 	tx_ring->wr_ptr_add = 0;
 
+	if (tx_ring == r_vec->xdp_ring)
+		return;
+
 	nd_q = netdev_get_tx_queue(nn->netdev, tx_ring->idx);
 	netdev_tx_reset_queue(nd_q);
 }
@@ -1017,7 +1083,7 @@ static void nfp_net_tx_timeout(struct net_device *netdev)
 	struct nfp_net *nn = netdev_priv(netdev);
 	int i;
 
-	for (i = 0; i < nn->num_tx_rings; i++) {
+	for (i = 0; i < nn->netdev->real_num_tx_queues; i++) {
 		if (!netif_tx_queue_stopped(netdev_get_tx_queue(netdev, i)))
 			continue;
 		nn_warn(nn, "TX timeout on ring: %d\n", i);
@@ -1045,11 +1111,21 @@ nfp_net_calc_fl_bufsz(struct nfp_net *nn, unsigned int mtu)
 	return fl_bufsz;
 }
 
+static void
+nfp_net_free_frag(void *frag, bool xdp)
+{
+	if (!xdp)
+		skb_free_frag(frag);
+	else
+		__free_page(virt_to_page(frag));
+}
+
 /**
  * nfp_net_rx_alloc_one() - Allocate and map page frag for RX
  * @rx_ring:	RX ring structure of the skb
  * @dma_addr:	Pointer to storage for DMA address (output param)
  * @fl_bufsz:	size of freelist buffers
+ * @xdp:	Whether XDP is enabled
  *
  * This function will allcate a new page frag, map it for DMA.
  *
@@ -1057,20 +1133,26 @@ nfp_net_calc_fl_bufsz(struct nfp_net *nn, unsigned int mtu)
  */
 static void *
 nfp_net_rx_alloc_one(struct nfp_net_rx_ring *rx_ring, dma_addr_t *dma_addr,
-		     unsigned int fl_bufsz)
+		     unsigned int fl_bufsz, bool xdp)
 {
 	struct nfp_net *nn = rx_ring->r_vec->nfp_net;
+	int direction;
 	void *frag;
 
+	if (!xdp)
 		frag = netdev_alloc_frag(fl_bufsz);
+	else
+		frag = page_address(alloc_page(GFP_KERNEL | __GFP_COLD));
 	if (!frag) {
 		nn_warn_ratelimit(nn, "Failed to alloc receive page frag\n");
 		return NULL;
 	}
 
-	*dma_addr = nfp_net_dma_map_rx(nn, frag, fl_bufsz, DMA_FROM_DEVICE);
+	direction = xdp ? DMA_BIDIRECTIONAL : DMA_FROM_DEVICE;
+
+	*dma_addr = nfp_net_dma_map_rx(nn, frag, fl_bufsz, direction);
 	if (dma_mapping_error(&nn->pdev->dev, *dma_addr)) {
-		skb_free_frag(frag);
+		nfp_net_free_frag(frag, xdp);
 		nn_warn_ratelimit(nn, "Failed to map DMA RX buffer\n");
 		return NULL;
 	}
@@ -1078,19 +1160,23 @@ nfp_net_rx_alloc_one(struct nfp_net_rx_ring *rx_ring, dma_addr_t *dma_addr,
 	return frag;
 }
 
-static void *nfp_net_napi_alloc_one(struct nfp_net *nn, dma_addr_t *dma_addr)
+static void *
+nfp_net_napi_alloc_one(struct nfp_net *nn, int direction, dma_addr_t *dma_addr)
 {
 	void *frag;
 
+	if (!nn->xdp_prog)
 		frag = napi_alloc_frag(nn->fl_bufsz);
+	else
+		frag = page_address(alloc_page(GFP_ATOMIC | __GFP_COLD));
 	if (!frag) {
 		nn_warn_ratelimit(nn, "Failed to alloc receive page frag\n");
 		return NULL;
 	}
 
-	*dma_addr = nfp_net_dma_map_rx(nn, frag, nn->fl_bufsz, DMA_FROM_DEVICE);
+	*dma_addr = nfp_net_dma_map_rx(nn, frag, nn->fl_bufsz, direction);
 	if (dma_mapping_error(&nn->pdev->dev, *dma_addr)) {
-		skb_free_frag(frag);
+		nfp_net_free_frag(frag, nn->xdp_prog);
 		nn_warn_ratelimit(nn, "Failed to map DMA RX buffer\n");
 		return NULL;
 	}
@@ -1161,14 +1247,17 @@ static void nfp_net_rx_ring_reset(struct nfp_net_rx_ring *rx_ring)
  * nfp_net_rx_ring_bufs_free() - Free any buffers currently on the RX ring
  * @nn:		NFP Net device
  * @rx_ring:	RX ring to remove buffers from
+ * @xdp:	Whether XDP is enabled
  *
  * Assumes that the device is stopped and buffers are in [0, ring->cnt - 1)
  * entries.  After device is disabled nfp_net_rx_ring_reset() must be called
  * to restore required ring geometry.
  */
 static void
-nfp_net_rx_ring_bufs_free(struct nfp_net *nn, struct nfp_net_rx_ring *rx_ring)
+nfp_net_rx_ring_bufs_free(struct nfp_net *nn, struct nfp_net_rx_ring *rx_ring,
+			  bool xdp)
 {
+	int direction = xdp ? DMA_BIDIRECTIONAL : DMA_FROM_DEVICE;
 	unsigned int i;
 
 	for (i = 0; i < rx_ring->cnt - 1; i++) {
@@ -1180,8 +1269,8 @@ nfp_net_rx_ring_bufs_free(struct nfp_net *nn, struct nfp_net_rx_ring *rx_ring)
 			continue;
 
 		nfp_net_dma_unmap_rx(nn, rx_ring->rxbufs[i].dma_addr,
-				     rx_ring->bufsz, DMA_FROM_DEVICE);
-		skb_free_frag(rx_ring->rxbufs[i].frag);
+				     rx_ring->bufsz, direction);
+		nfp_net_free_frag(rx_ring->rxbufs[i].frag, xdp);
 		rx_ring->rxbufs[i].dma_addr = 0;
 		rx_ring->rxbufs[i].frag = NULL;
 	}
@@ -1191,9 +1280,11 @@ nfp_net_rx_ring_bufs_free(struct nfp_net *nn, struct nfp_net_rx_ring *rx_ring)
  * nfp_net_rx_ring_bufs_alloc() - Fill RX ring with buffers (don't give to FW)
  * @nn:		NFP Net device
  * @rx_ring:	RX ring to remove buffers from
+ * @xdp:	Whether XDP is enabled
  */
 static int
-nfp_net_rx_ring_bufs_alloc(struct nfp_net *nn, struct nfp_net_rx_ring *rx_ring)
+nfp_net_rx_ring_bufs_alloc(struct nfp_net *nn, struct nfp_net_rx_ring *rx_ring,
+			   bool xdp)
 {
 	struct nfp_net_rx_buf *rxbufs;
 	unsigned int i;
@@ -1203,9 +1294,9 @@ nfp_net_rx_ring_bufs_alloc(struct nfp_net *nn, struct nfp_net_rx_ring *rx_ring)
 	for (i = 0; i < rx_ring->cnt - 1; i++) {
 		rxbufs[i].frag =
 			nfp_net_rx_alloc_one(rx_ring, &rxbufs[i].dma_addr,
-					     rx_ring->bufsz);
+					     rx_ring->bufsz, xdp);
 		if (!rxbufs[i].frag) {
-			nfp_net_rx_ring_bufs_free(nn, rx_ring);
+			nfp_net_rx_ring_bufs_free(nn, rx_ring, xdp);
 			return -ENOMEM;
 		}
 	}
@@ -1368,6 +1459,68 @@ nfp_net_rx_drop(struct nfp_net_r_vector *r_vec, struct nfp_net_rx_ring *rx_ring,
 		dev_kfree_skb_any(skb);
 }
 
+static void
+nfp_net_tx_xdp_buf(struct nfp_net *nn, struct nfp_net_rx_ring *rx_ring,
+		   struct nfp_net_tx_ring *tx_ring,
+		   struct nfp_net_rx_buf *rxbuf, unsigned int pkt_off,
+		   unsigned int pkt_len)
+{
+	struct nfp_net_tx_buf *txbuf;
+	struct nfp_net_tx_desc *txd;
+	dma_addr_t new_dma_addr;
+	void *new_frag;
+	int wr_idx;
+
+	if (unlikely(nfp_net_tx_full(tx_ring, 1))) {
+		nfp_net_rx_drop(rx_ring->r_vec, rx_ring, rxbuf, NULL);
+		return;
+	}
+
+	new_frag = nfp_net_napi_alloc_one(nn, DMA_BIDIRECTIONAL, &new_dma_addr);
+	if (unlikely(!new_frag)) {
+		nfp_net_rx_drop(rx_ring->r_vec, rx_ring, rxbuf, NULL);
+		return;
+	}
+	nfp_net_rx_give_one(rx_ring, new_frag, new_dma_addr);
+
+	wr_idx = tx_ring->wr_p & (tx_ring->cnt - 1);
+
+	/* Stash the soft descriptor of the head then initialize it */
+	txbuf = &tx_ring->txbufs[wr_idx];
+	txbuf->frag = rxbuf->frag;
+	txbuf->dma_addr = rxbuf->dma_addr;
+	txbuf->fidx = -1;
+	txbuf->pkt_cnt = 1;
+	txbuf->real_len = pkt_len;
+
+	dma_sync_single_for_device(&nn->pdev->dev, rxbuf->dma_addr + pkt_off,
+				   pkt_len, DMA_TO_DEVICE);
+
+	/* Build TX descriptor */
+	txd = &tx_ring->txds[wr_idx];
+	txd->offset_eop = PCIE_DESC_TX_EOP;
+	txd->dma_len = cpu_to_le16(pkt_len);
+	nfp_desc_set_dma_addr(txd, rxbuf->dma_addr + pkt_off);
+	txd->data_len = cpu_to_le16(pkt_len);
+
+	txd->flags = 0;
+	txd->mss = 0;
+	txd->l4_offset = 0;
+
+	tx_ring->wr_p++;
+	tx_ring->wr_ptr_add++;
+}
+
+static int nfp_net_run_xdp(struct bpf_prog *prog, void *data, unsigned int len)
+{
+	struct xdp_buff xdp;
+
+	xdp.data = data;
+	xdp.data_end = data + len;
+
+	return BPF_PROG_RUN(prog, (void *)&xdp);
+}
+
 /**
  * nfp_net_rx() - receive up to @budget packets on @rx_ring
  * @rx_ring:   RX ring to receive from
@@ -1383,16 +1536,27 @@ static int nfp_net_rx(struct nfp_net_rx_ring *rx_ring, int budget)
 {
 	struct nfp_net_r_vector *r_vec = rx_ring->r_vec;
 	struct nfp_net *nn = r_vec->nfp_net;
-	unsigned int data_len, meta_len;
-	struct nfp_net_rx_buf *rxbuf;
-	struct nfp_net_rx_desc *rxd;
-	dma_addr_t new_dma_addr;
+	struct nfp_net_tx_ring *tx_ring;
+	struct bpf_prog *xdp_prog;
+	unsigned int true_bufsz;
 	struct sk_buff *skb;
 	int pkts_polled = 0;
-	void *new_frag;
+	int rx_dma_map_dir;
 	int idx;
 
+	rcu_read_lock();
+	xdp_prog = READ_ONCE(nn->xdp_prog);
+	rx_dma_map_dir = xdp_prog ? DMA_BIDIRECTIONAL : DMA_FROM_DEVICE;
+	true_bufsz = xdp_prog ? PAGE_SIZE : nn->fl_bufsz;
+	tx_ring = r_vec->xdp_ring;
+
 	while (pkts_polled < budget) {
+		unsigned int meta_len, data_len, data_off, pkt_len, pkt_off;
+		struct nfp_net_rx_buf *rxbuf;
+		struct nfp_net_rx_desc *rxd;
+		dma_addr_t new_dma_addr;
+		void *new_frag;
+
 		idx = rx_ring->rd_p & (rx_ring->cnt - 1);
 
 		rxd = &rx_ring->rxds[idx];
@@ -1408,22 +1572,6 @@ static int nfp_net_rx(struct nfp_net_rx_ring *rx_ring, int budget)
 		pkts_polled++;
 
 		rxbuf =	&rx_ring->rxbufs[idx];
-		skb = build_skb(rxbuf->frag, nn->fl_bufsz);
-		if (unlikely(!skb)) {
-			nfp_net_rx_drop(r_vec, rx_ring, rxbuf, NULL);
-			continue;
-		}
-		new_frag = nfp_net_napi_alloc_one(nn, &new_dma_addr);
-		if (unlikely(!new_frag)) {
-			nfp_net_rx_drop(r_vec, rx_ring, rxbuf, skb);
-			continue;
-		}
-
-		nfp_net_dma_unmap_rx(nn, rx_ring->rxbufs[idx].dma_addr,
-				     nn->fl_bufsz, DMA_FROM_DEVICE);
-
-		nfp_net_rx_give_one(rx_ring, new_frag, new_dma_addr);
-
 		/*         < meta_len >
 		 *  <-- [rx_offset] -->
 		 *  ---------------------------------------------------------
@@ -1438,20 +1586,66 @@ static int nfp_net_rx(struct nfp_net_rx_ring *rx_ring, int budget)
 		 */
 		meta_len = rxd->rxd.meta_len_dd & PCIE_DESC_RX_META_LEN_MASK;
 		data_len = le16_to_cpu(rxd->rxd.data_len);
+		pkt_len = data_len - meta_len;
 
 		if (nn->rx_offset == NFP_NET_CFG_RX_OFFSET_DYNAMIC)
-			skb_reserve(skb, NFP_NET_RX_BUF_HEADROOM + meta_len);
+			pkt_off = meta_len;
 		else
-			skb_reserve(skb,
-				    NFP_NET_RX_BUF_HEADROOM + nn->rx_offset);
-		skb_put(skb, data_len - meta_len);
+			pkt_off = nn->rx_offset;
+		data_off = NFP_NET_RX_BUF_HEADROOM + pkt_off;
 
 		/* Stats update */
 		u64_stats_update_begin(&r_vec->rx_sync);
 		r_vec->rx_pkts++;
-		r_vec->rx_bytes += skb->len;
+		r_vec->rx_bytes += pkt_len;
 		u64_stats_update_end(&r_vec->rx_sync);
 
+		if (xdp_prog && !(rxd->rxd.flags & PCIE_DESC_RX_BPF &&
+				  nn->bpf_offload_xdp)) {
+			int act;
+
+			dma_sync_single_for_cpu(&nn->pdev->dev,
+						rxbuf->dma_addr + pkt_off,
+						pkt_len, DMA_FROM_DEVICE);
+			act = nfp_net_run_xdp(xdp_prog, rxbuf->frag + data_off,
+					      pkt_len);
+			switch (act) {
+			case XDP_PASS:
+				break;
+			case XDP_TX:
+				nfp_net_tx_xdp_buf(nn, rx_ring, tx_ring, rxbuf,
+						   pkt_off, pkt_len);
+				continue;
+			default:
+				bpf_warn_invalid_xdp_action(act);
+			case XDP_ABORTED:
+			case XDP_DROP:
+				nfp_net_rx_give_one(rx_ring, rxbuf->frag,
+						    rxbuf->dma_addr);
+				continue;
+			}
+		}
+
+		skb = build_skb(rxbuf->frag, true_bufsz);
+		if (unlikely(!skb)) {
+			nfp_net_rx_drop(r_vec, rx_ring, rxbuf, NULL);
+			continue;
+		}
+		new_frag = nfp_net_napi_alloc_one(nn, rx_dma_map_dir,
+						  &new_dma_addr);
+		if (unlikely(!new_frag)) {
+			nfp_net_rx_drop(r_vec, rx_ring, rxbuf, skb);
+			continue;
+		}
+
+		nfp_net_dma_unmap_rx(nn, rxbuf->dma_addr, nn->fl_bufsz,
+				     rx_dma_map_dir);
+
+		nfp_net_rx_give_one(rx_ring, new_frag, new_dma_addr);
+
+		skb_reserve(skb, data_off);
+		skb_put(skb, pkt_len);
+
 		if (nn->fw_ver.major <= 3) {
 			nfp_net_set_hash_desc(nn->netdev, skb, rxd);
 		} else if (meta_len) {
@@ -1477,6 +1671,10 @@ static int nfp_net_rx(struct nfp_net_rx_ring *rx_ring, int budget)
 		napi_gro_receive(&rx_ring->r_vec->napi, skb);
 	}
 
+	if (xdp_prog && tx_ring->wr_ptr_add)
+		nfp_net_tx_xmit_more_flush(tx_ring);
+	rcu_read_unlock();
+
 	return pkts_polled;
 }
 
@@ -1495,8 +1693,11 @@ static int nfp_net_poll(struct napi_struct *napi, int budget)
 
 	if (r_vec->tx_ring)
 		nfp_net_tx_complete(r_vec->tx_ring);
-	if (r_vec->rx_ring)
+	if (r_vec->rx_ring) {
 		pkts_polled = nfp_net_rx(r_vec->rx_ring, budget);
+		if (r_vec->xdp_ring)
+			nfp_net_xdp_complete(r_vec->xdp_ring);
+	}
 
 	if (pkts_polled < budget) {
 		napi_complete_done(napi, pkts_polled);
@@ -1536,10 +1737,12 @@ static void nfp_net_tx_ring_free(struct nfp_net_tx_ring *tx_ring)
  * nfp_net_tx_ring_alloc() - Allocate resource for a TX ring
  * @tx_ring:   TX Ring structure to allocate
  * @cnt:       Ring buffer count
+ * @is_xdp:    True if ring will be used for XDP
  *
  * Return: 0 on success, negative errno otherwise.
  */
-static int nfp_net_tx_ring_alloc(struct nfp_net_tx_ring *tx_ring, u32 cnt)
+static int
+nfp_net_tx_ring_alloc(struct nfp_net_tx_ring *tx_ring, u32 cnt, bool is_xdp)
 {
 	struct nfp_net_r_vector *r_vec = tx_ring->r_vec;
 	struct nfp_net *nn = r_vec->nfp_net;
@@ -1559,11 +1762,14 @@ static int nfp_net_tx_ring_alloc(struct nfp_net_tx_ring *tx_ring, u32 cnt)
 	if (!tx_ring->txbufs)
 		goto err_alloc;
 
-	netif_set_xps_queue(nn->netdev, &r_vec->affinity_mask, tx_ring->idx);
+	if (!is_xdp)
+		netif_set_xps_queue(nn->netdev, &r_vec->affinity_mask,
+				    tx_ring->idx);
 
-	nn_dbg(nn, "TxQ%02d: QCidx=%02d cnt=%d dma=%#llx host=%p\n",
+	nn_dbg(nn, "TxQ%02d: QCidx=%02d cnt=%d dma=%#llx host=%p %s\n",
 	       tx_ring->idx, tx_ring->qcidx,
-	       tx_ring->cnt, (unsigned long long)tx_ring->dma, tx_ring->txds);
+	       tx_ring->cnt, (unsigned long long)tx_ring->dma, tx_ring->txds,
+	       is_xdp ? "XDP" : "");
 
 	return 0;
 
@@ -1573,23 +1779,29 @@ static int nfp_net_tx_ring_alloc(struct nfp_net_tx_ring *tx_ring, u32 cnt)
 }
 
 static struct nfp_net_tx_ring *
-nfp_net_shadow_tx_rings_prepare(struct nfp_net *nn, u32 buf_cnt)
+nfp_net_tx_ring_set_prepare(struct nfp_net *nn, struct nfp_net_ring_set *s,
+			    unsigned int num_stack_tx_rings)
 {
 	struct nfp_net_tx_ring *rings;
 	unsigned int r;
 
-	rings = kcalloc(nn->num_tx_rings, sizeof(*rings), GFP_KERNEL);
+	rings = kcalloc(s->n_rings, sizeof(*rings), GFP_KERNEL);
 	if (!rings)
 		return NULL;
 
-	for (r = 0; r < nn->num_tx_rings; r++) {
-		nfp_net_tx_ring_init(&rings[r], nn->tx_rings[r].r_vec, r);
+	for (r = 0; r < s->n_rings; r++) {
+		int bias = 0;
+
+		if (r >= num_stack_tx_rings)
+			bias = num_stack_tx_rings;
+
+		nfp_net_tx_ring_init(&rings[r], &nn->r_vecs[r - bias], r);
 
-		if (nfp_net_tx_ring_alloc(&rings[r], buf_cnt))
+		if (nfp_net_tx_ring_alloc(&rings[r], s->dcnt, bias))
 			goto err_free_prev;
 	}
 
-	return rings;
+	return s->rings = rings;
 
 err_free_prev:
 	while (r--)
@@ -1598,28 +1810,27 @@ nfp_net_shadow_tx_rings_prepare(struct nfp_net *nn, u32 buf_cnt)
 	return NULL;
 }
 
-static struct nfp_net_tx_ring *
-nfp_net_shadow_tx_rings_swap(struct nfp_net *nn, struct nfp_net_tx_ring *rings)
+static void
+nfp_net_tx_ring_set_swap(struct nfp_net *nn, struct nfp_net_ring_set *s)
 {
-	struct nfp_net_tx_ring *old = nn->tx_rings;
-	unsigned int r;
+	struct nfp_net_ring_set new = *s;
 
-	for (r = 0; r < nn->num_tx_rings; r++)
-		old[r].r_vec->tx_ring = &rings[r];
+	s->dcnt = nn->txd_cnt;
+	s->rings = nn->tx_rings;
+	s->n_rings = nn->num_tx_rings;
 
-	nn->tx_rings = rings;
-	return old;
+	nn->txd_cnt = new.dcnt;
+	nn->tx_rings = new.rings;
+	nn->num_tx_rings = new.n_rings;
 }
 
 static void
-nfp_net_shadow_tx_rings_free(struct nfp_net *nn, struct nfp_net_tx_ring *rings)
+nfp_net_tx_ring_set_free(struct nfp_net *nn, struct nfp_net_ring_set *s)
 {
+	struct nfp_net_tx_ring *rings = s->rings;
 	unsigned int r;
 
-	if (!rings)
-		return;
-
-	for (r = 0; r < nn->num_tx_rings; r++)
+	for (r = 0; r < s->n_rings; r++)
 		nfp_net_tx_ring_free(&rings[r]);
 
 	kfree(rings);
@@ -1691,31 +1902,32 @@ nfp_net_rx_ring_alloc(struct nfp_net_rx_ring *rx_ring, unsigned int fl_bufsz,
 }
 
 static struct nfp_net_rx_ring *
-nfp_net_shadow_rx_rings_prepare(struct nfp_net *nn, unsigned int fl_bufsz,
-				u32 buf_cnt)
+nfp_net_rx_ring_set_prepare(struct nfp_net *nn, struct nfp_net_ring_set *s,
+			    bool xdp)
 {
+	unsigned int fl_bufsz =	nfp_net_calc_fl_bufsz(nn, s->mtu);
 	struct nfp_net_rx_ring *rings;
 	unsigned int r;
 
-	rings = kcalloc(nn->num_rx_rings, sizeof(*rings), GFP_KERNEL);
+	rings = kcalloc(s->n_rings, sizeof(*rings), GFP_KERNEL);
 	if (!rings)
 		return NULL;
 
-	for (r = 0; r < nn->num_rx_rings; r++) {
-		nfp_net_rx_ring_init(&rings[r], nn->rx_rings[r].r_vec, r);
+	for (r = 0; r < s->n_rings; r++) {
+		nfp_net_rx_ring_init(&rings[r], &nn->r_vecs[r], r);
 
-		if (nfp_net_rx_ring_alloc(&rings[r], fl_bufsz, buf_cnt))
+		if (nfp_net_rx_ring_alloc(&rings[r], fl_bufsz, s->dcnt))
 			goto err_free_prev;
 
-		if (nfp_net_rx_ring_bufs_alloc(nn, &rings[r]))
+		if (nfp_net_rx_ring_bufs_alloc(nn, &rings[r], xdp))
 			goto err_free_ring;
 	}
 
-	return rings;
+	return s->rings = rings;
 
 err_free_prev:
 	while (r--) {
-		nfp_net_rx_ring_bufs_free(nn, &rings[r]);
+		nfp_net_rx_ring_bufs_free(nn, &rings[r], xdp);
 err_free_ring:
 		nfp_net_rx_ring_free(&rings[r]);
 	}
@@ -1723,35 +1935,50 @@ nfp_net_shadow_rx_rings_prepare(struct nfp_net *nn, unsigned int fl_bufsz,
 	return NULL;
 }
 
-static struct nfp_net_rx_ring *
-nfp_net_shadow_rx_rings_swap(struct nfp_net *nn, struct nfp_net_rx_ring *rings)
+static void
+nfp_net_rx_ring_set_swap(struct nfp_net *nn, struct nfp_net_ring_set *s)
 {
-	struct nfp_net_rx_ring *old = nn->rx_rings;
-	unsigned int r;
+	struct nfp_net_ring_set new = *s;
 
-	for (r = 0; r < nn->num_rx_rings; r++)
-		old[r].r_vec->rx_ring = &rings[r];
+	s->mtu = nn->netdev->mtu;
+	s->dcnt = nn->rxd_cnt;
+	s->rings = nn->rx_rings;
+	s->n_rings = nn->num_rx_rings;
 
-	nn->rx_rings = rings;
-	return old;
+	nn->netdev->mtu = new.mtu;
+	nn->fl_bufsz = nfp_net_calc_fl_bufsz(nn, new.mtu);
+	nn->rxd_cnt = new.dcnt;
+	nn->rx_rings = new.rings;
+	nn->num_rx_rings = new.n_rings;
 }
 
 static void
-nfp_net_shadow_rx_rings_free(struct nfp_net *nn, struct nfp_net_rx_ring *rings)
+nfp_net_rx_ring_set_free(struct nfp_net *nn, struct nfp_net_ring_set *s,
+			 bool xdp)
 {
+	struct nfp_net_rx_ring *rings = s->rings;
 	unsigned int r;
 
-	if (!rings)
-		return;
-
-	for (r = 0; r < nn->num_rx_rings; r++) {
-		nfp_net_rx_ring_bufs_free(nn, &rings[r]);
+	for (r = 0; r < s->n_rings; r++) {
+		nfp_net_rx_ring_bufs_free(nn, &rings[r], xdp);
 		nfp_net_rx_ring_free(&rings[r]);
 	}
 
 	kfree(rings);
 }
 
+static void
+nfp_net_vector_assign_rings(struct nfp_net *nn, struct nfp_net_r_vector *r_vec,
+			    int idx)
+{
+	r_vec->rx_ring = idx < nn->num_rx_rings ? &nn->rx_rings[idx] : NULL;
+	r_vec->tx_ring =
+		idx < nn->num_stack_tx_rings ? &nn->tx_rings[idx] : NULL;
+
+	r_vec->xdp_ring = idx < nn->num_tx_rings - nn->num_stack_tx_rings ?
+		&nn->tx_rings[nn->num_stack_tx_rings + idx] : NULL;
+}
+
 static int
 nfp_net_prepare_vector(struct nfp_net *nn, struct nfp_net_r_vector *r_vec,
 		       int idx)
@@ -1759,33 +1986,20 @@ nfp_net_prepare_vector(struct nfp_net *nn, struct nfp_net_r_vector *r_vec,
 	struct msix_entry *entry = &nn->irq_entries[r_vec->irq_idx];
 	int err;
 
-	if (idx < nn->num_tx_rings) {
-		r_vec->tx_ring = &nn->tx_rings[idx];
-		nfp_net_tx_ring_init(r_vec->tx_ring, r_vec, idx);
-	} else {
-		r_vec->tx_ring = NULL;
-	}
-
-	if (idx < nn->num_rx_rings) {
-		r_vec->rx_ring = &nn->rx_rings[idx];
-		nfp_net_rx_ring_init(r_vec->rx_ring, r_vec, idx);
-	} else {
-		r_vec->rx_ring = NULL;
-	}
+	/* Setup NAPI */
+	netif_napi_add(nn->netdev, &r_vec->napi,
+		       nfp_net_poll, NAPI_POLL_WEIGHT);
 
 	snprintf(r_vec->name, sizeof(r_vec->name),
 		 "%s-rxtx-%d", nn->netdev->name, idx);
 	err = request_irq(entry->vector, r_vec->handler, 0, r_vec->name, r_vec);
 	if (err) {
+		netif_napi_del(&r_vec->napi);
 		nn_err(nn, "Error requesting IRQ %d\n", entry->vector);
 		return err;
 	}
 	disable_irq(entry->vector);
 
-	/* Setup NAPI */
-	netif_napi_add(nn->netdev, &r_vec->napi,
-		       nfp_net_poll, NAPI_POLL_WEIGHT);
-
 	irq_set_affinity_hint(entry->vector, &r_vec->affinity_mask);
 
 	nn_dbg(nn, "RV%02d: irq=%03d/%03d\n", idx, entry->vector, entry->entry);
@@ -1913,9 +2127,9 @@ static void nfp_net_clear_config_and_disable(struct nfp_net *nn)
 		nn_err(nn, "Could not disable device: %d\n", err);
 
 	for (r = 0; r < nn->num_rx_rings; r++)
-		nfp_net_rx_ring_reset(nn->r_vecs[r].rx_ring);
+		nfp_net_rx_ring_reset(&nn->rx_rings[r]);
 	for (r = 0; r < nn->num_tx_rings; r++)
-		nfp_net_tx_ring_reset(nn, nn->r_vecs[r].tx_ring);
+		nfp_net_tx_ring_reset(nn, &nn->tx_rings[r]);
 	for (r = 0; r < nn->num_r_vecs; r++)
 		nfp_net_vec_clear_ring_data(nn, r);
 
@@ -1993,7 +2207,7 @@ static int __nfp_net_set_config_and_enable(struct nfp_net *nn)
 	nn->ctrl = new_ctrl;
 
 	for (r = 0; r < nn->num_rx_rings; r++)
-		nfp_net_rx_ring_fill_freelist(nn->r_vecs[r].rx_ring);
+		nfp_net_rx_ring_fill_freelist(&nn->rx_rings[r]);
 
 	/* Since reconfiguration requests while NFP is down are ignored we
 	 * have to wipe the entire VXLAN configuration and reinitialize it.
@@ -2044,6 +2258,15 @@ static void nfp_net_open_stack(struct nfp_net *nn)
 static int nfp_net_netdev_open(struct net_device *netdev)
 {
 	struct nfp_net *nn = netdev_priv(netdev);
+	struct nfp_net_ring_set rx = {
+		.n_rings = nn->num_rx_rings,
+		.mtu = nn->netdev->mtu,
+		.dcnt = nn->rxd_cnt,
+	};
+	struct nfp_net_ring_set tx = {
+		.n_rings = nn->num_tx_rings,
+		.dcnt = nn->txd_cnt,
+	};
 	int err, r;
 
 	if (nn->ctrl & NFP_NET_CFG_CTRL_ENABLE) {
@@ -2068,41 +2291,29 @@ static int nfp_net_netdev_open(struct net_device *netdev)
 		goto err_free_exn;
 	disable_irq(nn->irq_entries[NFP_NET_IRQ_LSC_IDX].vector);
 
-	nn->rx_rings = kcalloc(nn->num_rx_rings, sizeof(*nn->rx_rings),
-			       GFP_KERNEL);
-	if (!nn->rx_rings) {
-		err = -ENOMEM;
-		goto err_free_lsc;
-	}
-	nn->tx_rings = kcalloc(nn->num_tx_rings, sizeof(*nn->tx_rings),
-			       GFP_KERNEL);
-	if (!nn->tx_rings) {
-		err = -ENOMEM;
-		goto err_free_rx_rings;
-	}
-
 	for (r = 0; r < nn->num_r_vecs; r++) {
 		err = nfp_net_prepare_vector(nn, &nn->r_vecs[r], r);
 		if (err)
 			goto err_cleanup_vec_p;
 	}
-	for (r = 0; r < nn->num_tx_rings; r++) {
-		err = nfp_net_tx_ring_alloc(nn->r_vecs[r].tx_ring, nn->txd_cnt);
-		if (err)
-			goto err_free_tx_ring_p;
+
+	nn->rx_rings = nfp_net_rx_ring_set_prepare(nn, &rx, nn->xdp_prog);
+	if (!nn->rx_rings) {
+		err = -ENOMEM;
+		goto err_cleanup_vec;
 	}
-	for (r = 0; r < nn->num_rx_rings; r++) {
-		err = nfp_net_rx_ring_alloc(nn->r_vecs[r].rx_ring,
-					    nn->fl_bufsz, nn->rxd_cnt);
-		if (err)
-			goto err_flush_free_rx_ring_p;
 
-		err = nfp_net_rx_ring_bufs_alloc(nn, nn->r_vecs[r].rx_ring);
-		if (err)
-			goto err_free_rx_ring_p;
+	nn->tx_rings = nfp_net_tx_ring_set_prepare(nn, &tx,
+						   nn->num_stack_tx_rings);
+	if (!nn->tx_rings) {
+		err = -ENOMEM;
+		goto err_free_rx_rings;
 	}
 
-	err = netif_set_real_num_tx_queues(netdev, nn->num_tx_rings);
+	for (r = 0; r < nn->max_r_vecs; r++)
+		nfp_net_vector_assign_rings(nn, &nn->r_vecs[r], r);
+
+	err = netif_set_real_num_tx_queues(netdev, nn->num_stack_tx_rings);
 	if (err)
 		goto err_free_rings;
 
@@ -2132,25 +2343,14 @@ static int nfp_net_netdev_open(struct net_device *netdev)
 	return 0;
 
 err_free_rings:
-	r = nn->num_rx_rings;
-err_flush_free_rx_ring_p:
-	while (r--) {
-		nfp_net_rx_ring_bufs_free(nn, nn->r_vecs[r].rx_ring);
-err_free_rx_ring_p:
-		nfp_net_rx_ring_free(nn->r_vecs[r].rx_ring);
-	}
-	r = nn->num_tx_rings;
-err_free_tx_ring_p:
-	while (r--)
-		nfp_net_tx_ring_free(nn->r_vecs[r].tx_ring);
+	nfp_net_tx_ring_set_free(nn, &tx);
+err_free_rx_rings:
+	nfp_net_rx_ring_set_free(nn, &rx, nn->xdp_prog);
+err_cleanup_vec:
 	r = nn->num_r_vecs;
 err_cleanup_vec_p:
 	while (r--)
 		nfp_net_cleanup_vector(nn, &nn->r_vecs[r]);
-	kfree(nn->tx_rings);
-err_free_rx_rings:
-	kfree(nn->rx_rings);
-err_free_lsc:
 	nfp_net_aux_irq_free(nn, NFP_NET_CFG_LSC, NFP_NET_IRQ_LSC_IDX);
 err_free_exn:
 	nfp_net_aux_irq_free(nn, NFP_NET_CFG_EXN, NFP_NET_IRQ_EXN_IDX);
@@ -2186,11 +2386,11 @@ static void nfp_net_close_free_all(struct nfp_net *nn)
 	unsigned int r;
 
 	for (r = 0; r < nn->num_rx_rings; r++) {
-		nfp_net_rx_ring_bufs_free(nn, nn->r_vecs[r].rx_ring);
-		nfp_net_rx_ring_free(nn->r_vecs[r].rx_ring);
+		nfp_net_rx_ring_bufs_free(nn, &nn->rx_rings[r], nn->xdp_prog);
+		nfp_net_rx_ring_free(&nn->rx_rings[r]);
 	}
 	for (r = 0; r < nn->num_tx_rings; r++)
-		nfp_net_tx_ring_free(nn->r_vecs[r].tx_ring);
+		nfp_net_tx_ring_free(&nn->tx_rings[r]);
 	for (r = 0; r < nn->num_r_vecs; r++)
 		nfp_net_cleanup_vector(nn, &nn->r_vecs[r]);
 
@@ -2255,89 +2455,137 @@ static void nfp_net_set_rx_mode(struct net_device *netdev)
 	nn->ctrl = new_ctrl;
 }
 
-static int nfp_net_change_mtu(struct net_device *netdev, int new_mtu)
+static void nfp_net_rss_init_itbl(struct nfp_net *nn)
 {
-	unsigned int old_mtu, old_fl_bufsz, new_fl_bufsz;
-	struct nfp_net *nn = netdev_priv(netdev);
-	struct nfp_net_rx_ring *tmp_rings;
-	int err;
-
-	old_mtu = netdev->mtu;
-	old_fl_bufsz = nn->fl_bufsz;
-	new_fl_bufsz = nfp_net_calc_fl_bufsz(nn, new_mtu);
-
-	if (!netif_running(netdev)) {
-		netdev->mtu = new_mtu;
-		nn->fl_bufsz = new_fl_bufsz;
-		return 0;
-	}
+	int i;
 
-	/* Prepare new rings */
-	tmp_rings = nfp_net_shadow_rx_rings_prepare(nn, new_fl_bufsz,
-						    nn->rxd_cnt);
-	if (!tmp_rings)
-		return -ENOMEM;
+	for (i = 0; i < sizeof(nn->rss_itbl); i++)
+		nn->rss_itbl[i] =
+			ethtool_rxfh_indir_default(i, nn->num_rx_rings);
+}
 
-	/* Stop device, swap in new rings, try to start the firmware */
-	nfp_net_close_stack(nn);
-	nfp_net_clear_config_and_disable(nn);
+static int
+nfp_net_ring_swap_enable(struct nfp_net *nn, unsigned int *num_vecs,
+			 unsigned int *stack_tx_rings,
+			 struct bpf_prog **xdp_prog,
+			 struct nfp_net_ring_set *rx,
+			 struct nfp_net_ring_set *tx)
+{
+	unsigned int r;
+	int err;
 
-	tmp_rings = nfp_net_shadow_rx_rings_swap(nn, tmp_rings);
+	if (rx)
+		nfp_net_rx_ring_set_swap(nn, rx);
+	if (tx)
+		nfp_net_tx_ring_set_swap(nn, tx);
 
-	netdev->mtu = new_mtu;
-	nn->fl_bufsz = new_fl_bufsz;
+	swap(*num_vecs, nn->num_r_vecs);
+	swap(*stack_tx_rings, nn->num_stack_tx_rings);
+	*xdp_prog = xchg(&nn->xdp_prog, *xdp_prog);
 
-	err = nfp_net_set_config_and_enable(nn);
-	if (err) {
-		const int err_new = err;
+	for (r = 0; r <	nn->max_r_vecs; r++)
+		nfp_net_vector_assign_rings(nn, &nn->r_vecs[r], r);
 
-		/* Try with old configuration and old rings */
-		tmp_rings = nfp_net_shadow_rx_rings_swap(nn, tmp_rings);
+	if (nn->netdev->real_num_rx_queues != nn->num_rx_rings) {
+		if (!netif_is_rxfh_configured(nn->netdev))
+			nfp_net_rss_init_itbl(nn);
 
-		netdev->mtu = old_mtu;
-		nn->fl_bufsz = old_fl_bufsz;
+		err = netif_set_real_num_rx_queues(nn->netdev,
+						   nn->num_rx_rings);
+		if (err)
+			return err;
+	}
 
-		err = __nfp_net_set_config_and_enable(nn);
+	if (nn->netdev->real_num_tx_queues != nn->num_stack_tx_rings) {
+		err = netif_set_real_num_tx_queues(nn->netdev,
+						   nn->num_stack_tx_rings);
 		if (err)
-			nn_err(nn, "Can't restore MTU - FW communication failed (%d,%d)\n",
-			       err_new, err);
+			return err;
 	}
 
-	nfp_net_shadow_rx_rings_free(nn, tmp_rings);
+	return __nfp_net_set_config_and_enable(nn);
+}
 
-	nfp_net_open_stack(nn);
+static int
+nfp_net_check_config(struct nfp_net *nn, struct bpf_prog *xdp_prog,
+		     struct nfp_net_ring_set *rx, struct nfp_net_ring_set *tx)
+{
+	/* XDP-enabled tests */
+	if (!xdp_prog)
+		return 0;
+	if (rx && nfp_net_calc_fl_bufsz(nn, rx->mtu) > PAGE_SIZE) {
+		nn_warn(nn, "MTU too large w/ XDP enabled\n");
+		return -EINVAL;
+	}
+	if (tx && tx->n_rings > nn->max_tx_rings) {
+		nn_warn(nn, "Insufficient number of TX rings w/ XDP enabled\n");
+		return -EINVAL;
+	}
 
-	return err;
+	return 0;
 }
 
-int nfp_net_set_ring_size(struct nfp_net *nn, u32 rxd_cnt, u32 txd_cnt)
-{
-	struct nfp_net_tx_ring *tx_rings = NULL;
-	struct nfp_net_rx_ring *rx_rings = NULL;
-	u32 old_rxd_cnt, old_txd_cnt;
+static void
+nfp_net_ring_reconfig_down(struct nfp_net *nn, struct bpf_prog **xdp_prog,
+			   struct nfp_net_ring_set *rx,
+			   struct nfp_net_ring_set *tx,
+			   unsigned int stack_tx_rings, unsigned int num_vecs)
+{
+	nn->netdev->mtu = rx ? rx->mtu : nn->netdev->mtu;
+	nn->fl_bufsz = nfp_net_calc_fl_bufsz(nn, nn->netdev->mtu);
+	nn->rxd_cnt = rx ? rx->dcnt : nn->rxd_cnt;
+	nn->txd_cnt = tx ? tx->dcnt : nn->txd_cnt;
+	nn->num_rx_rings = rx ? rx->n_rings : nn->num_rx_rings;
+	nn->num_tx_rings = tx ? tx->n_rings : nn->num_tx_rings;
+	nn->num_stack_tx_rings = stack_tx_rings;
+	nn->num_r_vecs = num_vecs;
+	*xdp_prog = xchg(&nn->xdp_prog, *xdp_prog);
+
+	if (!netif_is_rxfh_configured(nn->netdev))
+		nfp_net_rss_init_itbl(nn);
+}
+
+int
+nfp_net_ring_reconfig(struct nfp_net *nn, struct bpf_prog **xdp_prog,
+		      struct nfp_net_ring_set *rx, struct nfp_net_ring_set *tx)
+{
+	unsigned int stack_tx_rings, num_vecs, r;
 	int err;
 
+	stack_tx_rings = tx ? tx->n_rings : nn->num_tx_rings;
+	if (*xdp_prog)
+		stack_tx_rings -= rx ? rx->n_rings : nn->num_rx_rings;
+
+	num_vecs = max(rx ? rx->n_rings : nn->num_rx_rings, stack_tx_rings);
+
+	err = nfp_net_check_config(nn, *xdp_prog, rx, tx);
+	if (err)
+		return err;
+
 	if (!netif_running(nn->netdev)) {
-		nn->rxd_cnt = rxd_cnt;
-		nn->txd_cnt = txd_cnt;
+		nfp_net_ring_reconfig_down(nn, xdp_prog, rx, tx,
+					   stack_tx_rings, num_vecs);
 		return 0;
 	}
 
-	old_rxd_cnt = nn->rxd_cnt;
-	old_txd_cnt = nn->txd_cnt;
-
 	/* Prepare new rings */
-	if (nn->rxd_cnt != rxd_cnt) {
-		rx_rings = nfp_net_shadow_rx_rings_prepare(nn, nn->fl_bufsz,
-							   rxd_cnt);
-		if (!rx_rings)
-			return -ENOMEM;
+	for (r = nn->num_r_vecs; r < num_vecs; r++) {
+		err = nfp_net_prepare_vector(nn, &nn->r_vecs[r], r);
+		if (err) {
+			num_vecs = r;
+			goto err_cleanup_vecs;
 		}
-	if (nn->txd_cnt != txd_cnt) {
-		tx_rings = nfp_net_shadow_tx_rings_prepare(nn, txd_cnt);
-		if (!tx_rings) {
-			nfp_net_shadow_rx_rings_free(nn, rx_rings);
-			return -ENOMEM;
+	}
+	if (rx) {
+		if (!nfp_net_rx_ring_set_prepare(nn, rx, *xdp_prog)) {
+			err = -ENOMEM;
+			goto err_cleanup_vecs;
+		}
+	}
+	if (tx) {
+		if (!nfp_net_tx_ring_set_prepare(nn, tx, stack_tx_rings)) {
+			err = -ENOMEM;
+			goto err_free_rx;
 		}
 	}
 
@@ -2345,39 +2593,51 @@ int nfp_net_set_ring_size(struct nfp_net *nn, u32 rxd_cnt, u32 txd_cnt)
 	nfp_net_close_stack(nn);
 	nfp_net_clear_config_and_disable(nn);
 
-	if (rx_rings)
-		rx_rings = nfp_net_shadow_rx_rings_swap(nn, rx_rings);
-	if (tx_rings)
-		tx_rings = nfp_net_shadow_tx_rings_swap(nn, tx_rings);
-
-	nn->rxd_cnt = rxd_cnt;
-	nn->txd_cnt = txd_cnt;
-
-	err = nfp_net_set_config_and_enable(nn);
+	err = nfp_net_ring_swap_enable(nn, &num_vecs, &stack_tx_rings,
+				       xdp_prog, rx, tx);
 	if (err) {
-		const int err_new = err;
-
-		/* Try with old configuration and old rings */
-		if (rx_rings)
-			rx_rings = nfp_net_shadow_rx_rings_swap(nn, rx_rings);
-		if (tx_rings)
-			tx_rings = nfp_net_shadow_tx_rings_swap(nn, tx_rings);
+		int err2;
 
-		nn->rxd_cnt = old_rxd_cnt;
-		nn->txd_cnt = old_txd_cnt;
+		nfp_net_clear_config_and_disable(nn);
 
-		err = __nfp_net_set_config_and_enable(nn);
-		if (err)
+		/* Try with old configuration and old rings */
+		err2 = nfp_net_ring_swap_enable(nn, &num_vecs, &stack_tx_rings,
+						xdp_prog, rx, tx);
+		if (err2)
 			nn_err(nn, "Can't restore ring config - FW communication failed (%d,%d)\n",
-			       err_new, err);
+			       err, err2);
 	}
+	for (r = num_vecs - 1; r >= nn->num_r_vecs; r--)
+		nfp_net_cleanup_vector(nn, &nn->r_vecs[r]);
 
-	nfp_net_shadow_rx_rings_free(nn, rx_rings);
-	nfp_net_shadow_tx_rings_free(nn, tx_rings);
+	if (rx)
+		nfp_net_rx_ring_set_free(nn, rx, *xdp_prog);
+	if (tx)
+		nfp_net_tx_ring_set_free(nn, tx);
 
 	nfp_net_open_stack(nn);
 
 	return err;
+
+err_free_rx:
+	if (rx)
+		nfp_net_rx_ring_set_free(nn, rx, *xdp_prog);
+err_cleanup_vecs:
+	for (r = num_vecs - 1; r >= nn->num_r_vecs; r--)
+		nfp_net_cleanup_vector(nn, &nn->r_vecs[r]);
+	return err;
+}
+
+static int nfp_net_change_mtu(struct net_device *netdev, int new_mtu)
+{
+	struct nfp_net *nn = netdev_priv(netdev);
+	struct nfp_net_ring_set rx = {
+		.n_rings = nn->num_rx_rings,
+		.mtu = new_mtu,
+		.dcnt = nn->rxd_cnt,
+	};
+
+	return nfp_net_ring_reconfig(nn, &nn->xdp_prog, &rx, NULL);
 }
 
 static struct rtnl_link_stats64 *nfp_net_stat64(struct net_device *netdev,
@@ -2434,8 +2694,12 @@ nfp_net_setup_tc(struct net_device *netdev, u32 handle, __be16 proto,
 	if (proto != htons(ETH_P_ALL))
 		return -ENOTSUPP;
 
-	if (tc->type == TC_SETUP_CLSBPF && nfp_net_ebpf_capable(nn))
-		return nfp_net_bpf_offload(nn, handle, proto, tc->cls_bpf);
+	if (tc->type == TC_SETUP_CLSBPF && nfp_net_ebpf_capable(nn)) {
+		if (!nn->bpf_offload_xdp)
+			return nfp_net_bpf_offload(nn, tc->cls_bpf);
+		else
+			return -EBUSY;
+	}
 
 	return -EINVAL;
 }
@@ -2643,6 +2907,87 @@ static void nfp_net_del_vxlan_port(struct net_device *netdev,
 		nfp_net_set_vxlan_port(nn, idx, 0);
 }
 
+static int nfp_net_xdp_offload(struct nfp_net *nn, struct bpf_prog *prog)
+{
+	struct tc_cls_bpf_offload cmd = {
+		.prog = prog,
+	};
+	int ret;
+
+	if (!nfp_net_ebpf_capable(nn))
+		return -EINVAL;
+
+	if (nn->ctrl & NFP_NET_CFG_CTRL_BPF) {
+		if (!nn->bpf_offload_xdp)
+			return prog ? -EBUSY : 0;
+		cmd.command = prog ? TC_CLSBPF_REPLACE : TC_CLSBPF_DESTROY;
+	} else {
+		if (!prog)
+			return 0;
+		cmd.command = TC_CLSBPF_ADD;
+	}
+
+	ret = nfp_net_bpf_offload(nn, &cmd);
+	/* Stop offload if replace not possible */
+	if (ret && cmd.command == TC_CLSBPF_REPLACE)
+		nfp_net_xdp_offload(nn, NULL);
+	nn->bpf_offload_xdp = prog && !ret;
+	return ret;
+}
+
+static int nfp_net_xdp_setup(struct nfp_net *nn, struct bpf_prog *prog)
+{
+	struct nfp_net_ring_set rx = {
+		.n_rings = nn->num_rx_rings,
+		.mtu = nn->netdev->mtu,
+		.dcnt = nn->rxd_cnt,
+	};
+	struct nfp_net_ring_set tx = {
+		.n_rings = nn->num_tx_rings,
+		.dcnt = nn->txd_cnt,
+	};
+	int err;
+
+	if (!prog && !nn->xdp_prog)
+		return 0;
+	if (prog && nn->xdp_prog) {
+		prog = xchg(&nn->xdp_prog, prog);
+		bpf_prog_put(prog);
+		nfp_net_xdp_offload(nn, nn->xdp_prog);
+		return 0;
+	}
+
+	tx.n_rings += prog ? nn->num_rx_rings : -nn->num_rx_rings;
+
+	/* We need RX reconfig to remap the buffers (BIDIR vs FROM_DEV) */
+	err = nfp_net_ring_reconfig(nn, &prog, &rx, &tx);
+	if (err)
+		return err;
+
+	/* @prog got swapped and is now the old one */
+	if (prog)
+		bpf_prog_put(prog);
+
+	nfp_net_xdp_offload(nn, nn->xdp_prog);
+
+	return 0;
+}
+
+static int nfp_net_xdp(struct net_device *netdev, struct netdev_xdp *xdp)
+{
+	struct nfp_net *nn = netdev_priv(netdev);
+
+	switch (xdp->command) {
+	case XDP_SETUP_PROG:
+		return nfp_net_xdp_setup(nn, xdp->prog);
+	case XDP_QUERY_PROG:
+		xdp->prog_attached = !!nn->xdp_prog;
+		return 0;
+	default:
+		return -EINVAL;
+	}
+}
+
 static const struct net_device_ops nfp_net_netdev_ops = {
 	.ndo_open		= nfp_net_netdev_open,
 	.ndo_stop		= nfp_net_netdev_close,
@@ -2657,6 +3002,7 @@ static const struct net_device_ops nfp_net_netdev_ops = {
 	.ndo_features_check	= nfp_net_features_check,
 	.ndo_udp_tunnel_add	= nfp_net_add_vxlan_port,
 	.ndo_udp_tunnel_del	= nfp_net_del_vxlan_port,
+	.ndo_xdp		= nfp_net_xdp,
 };
 
 /**
@@ -2763,13 +3109,9 @@ void nfp_net_netdev_free(struct nfp_net *nn)
  */
 static void nfp_net_rss_init(struct nfp_net *nn)
 {
-	int i;
-
 	netdev_rss_key_fill(nn->rss_key, NFP_NET_CFG_RSS_KEY_SZ);
 
-	for (i = 0; i < sizeof(nn->rss_itbl); i++)
-		nn->rss_itbl[i] =
-			ethtool_rxfh_indir_default(i, nn->num_rx_rings);
+	nfp_net_rss_init_itbl(nn);
 
 	/* Enable IPv4/IPv6 TCP by default */
 	nn->rss_cfg = NFP_NET_CFG_RSS_IPV4_TCP |
@@ -2923,5 +3265,11 @@ int nfp_net_netdev_init(struct net_device *netdev)
  */
 void nfp_net_netdev_clean(struct net_device *netdev)
 {
-	unregister_netdev(netdev);
+	struct nfp_net *nn = netdev_priv(netdev);
+
+	if (nn->xdp_prog)
+		bpf_prog_put(nn->xdp_prog);
+	if (nn->bpf_offload_xdp)
+		nfp_net_xdp_offload(nn, NULL);
+	unregister_netdev(nn->netdev);
 }

drivers/net/ethernet/netronome/nfp/nfp_net_debugfs.c

@@ -114,6 +114,16 @@ static const struct file_operations nfp_rx_q_fops = {
 	.llseek = seq_lseek
 };
 
+static int nfp_net_debugfs_tx_q_open(struct inode *inode, struct file *f);
+
+static const struct file_operations nfp_tx_q_fops = {
+	.owner = THIS_MODULE,
+	.open = nfp_net_debugfs_tx_q_open,
+	.release = single_release,
+	.read = seq_read,
+	.llseek = seq_lseek
+};
+
 static int nfp_net_debugfs_tx_q_read(struct seq_file *file, void *data)
 {
 	struct nfp_net_r_vector *r_vec = file->private;
@@ -126,10 +136,13 @@ static int nfp_net_debugfs_tx_q_read(struct seq_file *file, void *data)
 
 	rtnl_lock();
 
-	if (!r_vec->nfp_net || !r_vec->tx_ring)
+	if (debugfs_real_fops(file->file) == &nfp_tx_q_fops)
+		tx_ring = r_vec->tx_ring;
+	else
+		tx_ring = r_vec->xdp_ring;
+	if (!r_vec->nfp_net || !tx_ring)
 		goto out;
 	nn = r_vec->nfp_net;
-	tx_ring = r_vec->tx_ring;
 	if (!netif_running(nn->netdev))
 		goto out;
 
@@ -148,9 +161,14 @@ static int nfp_net_debugfs_tx_q_read(struct seq_file *file, void *data)
 			   txd->vals[2], txd->vals[3]);
 
 		skb = READ_ONCE(tx_ring->txbufs[i].skb);
-		if (skb)
+		if (skb) {
+			if (tx_ring == r_vec->tx_ring)
 				seq_printf(file, " skb->head=%p skb->data=%p",
 					   skb->head, skb->data);
+			else
+				seq_printf(file, " frag=%p", skb);
+		}
+
 		if (tx_ring->txbufs[i].dma_addr)
 			seq_printf(file, " dma_addr=%pad",
 				   &tx_ring->txbufs[i].dma_addr);
@@ -176,7 +194,7 @@ static int nfp_net_debugfs_tx_q_open(struct inode *inode, struct file *f)
 	return single_open(f, nfp_net_debugfs_tx_q_read, inode->i_private);
 }
 
-static const struct file_operations nfp_tx_q_fops = {
+static const struct file_operations nfp_xdp_q_fops = {
 	.owner = THIS_MODULE,
 	.open = nfp_net_debugfs_tx_q_open,
 	.release = single_release,
@@ -186,7 +204,7 @@ static const struct file_operations nfp_tx_q_fops = {
 
 void nfp_net_debugfs_adapter_add(struct nfp_net *nn)
 {
-	struct dentry *queues, *tx, *rx;
+	struct dentry *queues, *tx, *rx, *xdp;
 	char int_name[16];
 	int i;
 
@@ -204,16 +222,19 @@ void nfp_net_debugfs_adapter_add(struct nfp_net *nn)
 
 	rx = debugfs_create_dir("rx", queues);
 	tx = debugfs_create_dir("tx", queues);
-	if (IS_ERR_OR_NULL(rx) || IS_ERR_OR_NULL(tx))
+	xdp = debugfs_create_dir("xdp", queues);
+	if (IS_ERR_OR_NULL(rx) || IS_ERR_OR_NULL(tx) || IS_ERR_OR_NULL(xdp))
 		return;
 
-	for (i = 0; i < nn->num_rx_rings; i++) {
+	for (i = 0; i < min(nn->max_rx_rings, nn->max_r_vecs); i++) {
 		sprintf(int_name, "%d", i);
 		debugfs_create_file(int_name, S_IRUSR, rx,
 				    &nn->r_vecs[i], &nfp_rx_q_fops);
+		debugfs_create_file(int_name, S_IRUSR, xdp,
+				    &nn->r_vecs[i], &nfp_xdp_q_fops);
 	}
 
-	for (i = 0; i < nn->num_tx_rings; i++) {
+	for (i = 0; i < min(nn->max_tx_rings, nn->max_r_vecs); i++) {
 		sprintf(int_name, "%d", i);
 		debugfs_create_file(int_name, S_IRUSR, tx,
 				    &nn->r_vecs[i], &nfp_tx_q_fops);

drivers/net/ethernet/netronome/nfp/nfp_net_ethtool.c

@@ -158,6 +158,28 @@ static void nfp_net_get_ringparam(struct net_device *netdev,
 	ring->tx_pending = nn->txd_cnt;
 }
 
+static int nfp_net_set_ring_size(struct nfp_net *nn, u32 rxd_cnt, u32 txd_cnt)
+{
+	struct nfp_net_ring_set *reconfig_rx = NULL, *reconfig_tx = NULL;
+	struct nfp_net_ring_set rx = {
+		.n_rings = nn->num_rx_rings,
+		.mtu = nn->netdev->mtu,
+		.dcnt = rxd_cnt,
+	};
+	struct nfp_net_ring_set tx = {
+		.n_rings = nn->num_tx_rings,
+		.dcnt = txd_cnt,
+	};
+
+	if (nn->rxd_cnt != rxd_cnt)
+		reconfig_rx = ℞
+	if (nn->txd_cnt != txd_cnt)
+		reconfig_tx = &tx;
+
+	return nfp_net_ring_reconfig(nn, &nn->xdp_prog,
+				     reconfig_rx, reconfig_tx);
+}
+
 static int nfp_net_set_ringparam(struct net_device *netdev,
 				 struct ethtool_ringparam *ring)
 {
@@ -614,6 +636,76 @@ static int nfp_net_set_coalesce(struct net_device *netdev,
 	return nfp_net_reconfig(nn, NFP_NET_CFG_UPDATE_IRQMOD);
 }
 
+static void nfp_net_get_channels(struct net_device *netdev,
+				 struct ethtool_channels *channel)
+{
+	struct nfp_net *nn = netdev_priv(netdev);
+	unsigned int num_tx_rings;
+
+	num_tx_rings = nn->num_tx_rings;
+	if (nn->xdp_prog)
+		num_tx_rings -= nn->num_rx_rings;
+
+	channel->max_rx = min(nn->max_rx_rings, nn->max_r_vecs);
+	channel->max_tx = min(nn->max_tx_rings, nn->max_r_vecs);
+	channel->max_combined = min(channel->max_rx, channel->max_tx);
+	channel->max_other = NFP_NET_NON_Q_VECTORS;
+	channel->combined_count = min(nn->num_rx_rings, num_tx_rings);
+	channel->rx_count = nn->num_rx_rings - channel->combined_count;
+	channel->tx_count = num_tx_rings - channel->combined_count;
+	channel->other_count = NFP_NET_NON_Q_VECTORS;
+}
+
+static int nfp_net_set_num_rings(struct nfp_net *nn, unsigned int total_rx,
+				 unsigned int total_tx)
+{
+	struct nfp_net_ring_set *reconfig_rx = NULL, *reconfig_tx = NULL;
+	struct nfp_net_ring_set rx = {
+		.n_rings = total_rx,
+		.mtu = nn->netdev->mtu,
+		.dcnt = nn->rxd_cnt,
+	};
+	struct nfp_net_ring_set tx = {
+		.n_rings = total_tx,
+		.dcnt = nn->txd_cnt,
+	};
+
+	if (nn->num_rx_rings != total_rx)
+		reconfig_rx = ℞
+	if (nn->num_stack_tx_rings != total_tx ||
+	    (nn->xdp_prog && reconfig_rx))
+		reconfig_tx = &tx;
+
+	/* nfp_net_check_config() will catch tx.n_rings > nn->max_tx_rings */
+	if (nn->xdp_prog)
+		tx.n_rings += total_rx;
+
+	return nfp_net_ring_reconfig(nn, &nn->xdp_prog,
+				     reconfig_rx, reconfig_tx);
+}
+
+static int nfp_net_set_channels(struct net_device *netdev,
+				struct ethtool_channels *channel)
+{
+	struct nfp_net *nn = netdev_priv(netdev);
+	unsigned int total_rx, total_tx;
+
+	/* Reject unsupported */
+	if (!channel->combined_count ||
+	    channel->other_count != NFP_NET_NON_Q_VECTORS ||
+	    (channel->rx_count && channel->tx_count))
+		return -EINVAL;
+
+	total_rx = channel->combined_count + channel->rx_count;
+	total_tx = channel->combined_count + channel->tx_count;
+
+	if (total_rx > min(nn->max_rx_rings, nn->max_r_vecs) ||
+	    total_tx > min(nn->max_tx_rings, nn->max_r_vecs))
+		return -EINVAL;
+
+	return nfp_net_set_num_rings(nn, total_rx, total_tx);
+}
+
 static const struct ethtool_ops nfp_net_ethtool_ops = {
 	.get_drvinfo		= nfp_net_get_drvinfo,
 	.get_link		= ethtool_op_get_link,
@@ -632,6 +724,8 @@ static const struct ethtool_ops nfp_net_ethtool_ops = {
 	.get_regs		= nfp_net_get_regs,
 	.get_coalesce           = nfp_net_get_coalesce,
 	.set_coalesce           = nfp_net_set_coalesce,
+	.get_channels		= nfp_net_get_channels,
+	.set_channels		= nfp_net_set_channels,
 };
 
 void nfp_net_set_ethtool_ops(struct net_device *netdev)

drivers/net/ethernet/netronome/nfp/nfp_net_offload.c

@@ -111,6 +111,9 @@ nfp_net_bpf_get_act(struct nfp_net *nn, struct tc_cls_bpf_offload *cls_bpf)
 	const struct tc_action *a;
 	LIST_HEAD(actions);
 
+	if (!cls_bpf->exts)
+		return NN_ACT_XDP;
+
 	/* TC direct action */
 	if (cls_bpf->exts_integrated) {
 		if (tc_no_actions(cls_bpf->exts))
@@ -233,9 +236,7 @@ static int nfp_net_bpf_stop(struct nfp_net *nn)
 	return nfp_net_reconfig(nn, NFP_NET_CFG_UPDATE_GEN);
 }
 
-int
-nfp_net_bpf_offload(struct nfp_net *nn, u32 handle, __be16 proto,
-		    struct tc_cls_bpf_offload *cls_bpf)
+int nfp_net_bpf_offload(struct nfp_net *nn, struct tc_cls_bpf_offload *cls_bpf)
 {
 	struct nfp_bpf_result res;
 	dma_addr_t dma_addr;

include/linux/debugfs.h

@@ -52,7 +52,8 @@ extern struct srcu_struct debugfs_srcu;
  * Must only be called under the protection established by
  * debugfs_use_file_start().
  */
-static inline const struct file_operations *debugfs_real_fops(struct file *filp)
+static inline const struct file_operations *
+debugfs_real_fops(const struct file *filp)
 	__must_hold(&debugfs_srcu)
 {
 	/*