1 files changed, 184 insertions, 50 deletions

diff --git a/drivers/net/ethernet/marvell/octeontx2/nic/otx2_pf.c b/drivers/net/ethernet/marvell/octeontx2/nic/otx2_pf.c
index 53df7fff92c4..1e0d0c9c1dac 100644
--- a/drivers/net/ethernet/marvell/octeontx2/nic/otx2_pf.c
+++ b/drivers/net/ethernet/marvell/octeontx2/nic/otx2_pf.c
@@ -13,6 +13,8 @@
 #include <linux/if_vlan.h>
 #include <linux/iommu.h>
 #include <net/ip.h>
+#include <linux/bpf.h>
+#include <linux/bpf_trace.h>
 
 #include "otx2_reg.h"
 #include "otx2_common.h"
@@ -48,9 +50,15 @@ static int otx2_config_hw_rx_tstamp(struct otx2_nic *pfvf, bool enable);
 
 static int otx2_change_mtu(struct net_device *netdev, int new_mtu)
 {
+	struct otx2_nic *pf = netdev_priv(netdev);
 	bool if_up = netif_running(netdev);
 	int err = 0;
 
+	if (pf->xdp_prog && new_mtu > MAX_XDP_MTU) {
+		netdev_warn(netdev, "Jumbo frames not yet supported with XDP, current MTU %d.\n",
+			    netdev->mtu);
+		return -EINVAL;
+	}
 	if (if_up)
 		otx2_stop(netdev);
 
@@ -1180,7 +1188,7 @@ static irqreturn_t otx2_q_intr_handler(int irq, void *data)
 	}
 
 	/* SQ */
-	for (qidx = 0; qidx < pf->hw.tx_queues; qidx++) {
+	for (qidx = 0; qidx < pf->hw.tot_tx_queues; qidx++) {
 		ptr = otx2_get_regaddr(pf, NIX_LF_SQ_OP_INT);
 		val = otx2_atomic64_add((qidx << 44), ptr);
 		otx2_write64(pf, NIX_LF_SQ_OP_INT, (qidx << 44) |
@@ -1283,7 +1291,7 @@ static void otx2_free_sq_res(struct otx2_nic *pf)
 	otx2_ctx_disable(&pf->mbox, NIX_AQ_CTYPE_SQ, false);
 	/* Free SQB pointers */
 	otx2_sq_free_sqbs(pf);
-	for (qidx = 0; qidx < pf->hw.tx_queues; qidx++) {
+	for (qidx = 0; qidx < pf->hw.tot_tx_queues; qidx++) {
 		sq = &qset->sq[qidx];
 		qmem_free(pf->dev, sq->sqe);
 		qmem_free(pf->dev, sq->tso_hdrs);
@@ -1304,16 +1312,14 @@ static int otx2_get_rbuf_size(struct otx2_nic *pf, int mtu)
 	 * NIX transfers entire data using 6 segments/buffers and writes
 	 * a CQE_RX descriptor with those segment addresses. First segment
 	 * has additional data prepended to packet. Also software omits a
-	 * headroom of 128 bytes and sizeof(struct skb_shared_info) in
-	 * each segment. Hence the total size of memory needed
-	 * to receive a packet with 'mtu' is:
+	 * headroom of 128 bytes in each segment. Hence the total size of
+	 * memory needed to receive a packet with 'mtu' is:
 	 * frame size =  mtu + additional data;
-	 * memory = frame_size + (headroom + struct skb_shared_info size) * 6;
+	 * memory = frame_size + headroom * 6;
 	 * each receive buffer size = memory / 6;
 	 */
 	frame_size = mtu + OTX2_ETH_HLEN + OTX2_HW_TIMESTAMP_LEN;
-	total_size = frame_size + (OTX2_HEAD_ROOM +
-		     OTX2_DATA_ALIGN(sizeof(struct skb_shared_info))) * 6;
+	total_size = frame_size + OTX2_HEAD_ROOM * 6;
 	rbuf_size = total_size / 6;
 
 	return ALIGN(rbuf_size, 2048);
@@ -1332,10 +1338,11 @@ static int otx2_init_hw_resources(struct otx2_nic *pf)
 	 * so, aura count = pool count.
 	 */
 	hw->rqpool_cnt = hw->rx_queues;
-	hw->sqpool_cnt = hw->tx_queues;
+	hw->sqpool_cnt = hw->tot_tx_queues;
 	hw->pool_cnt = hw->rqpool_cnt + hw->sqpool_cnt;
 
-	pf->max_frs = pf->netdev->mtu + OTX2_ETH_HLEN + OTX2_HW_TIMESTAMP_LEN;
+	/* Maximum hardware supported transmit length */
+	pf->tx_max_pktlen = pf->netdev->max_mtu + OTX2_ETH_HLEN;
 
 	pf->rbsize = otx2_get_rbuf_size(pf, pf->netdev->mtu);
 
@@ -1493,6 +1500,44 @@ static void otx2_free_hw_resources(struct otx2_nic *pf)
 	mutex_unlock(&mbox->lock);
 }
 
+static void otx2_do_set_rx_mode(struct otx2_nic *pf)
+{
+	struct net_device *netdev = pf->netdev;
+	struct nix_rx_mode *req;
+	bool promisc = false;
+
+	if (!(netdev->flags & IFF_UP))
+		return;
+
+	if ((netdev->flags & IFF_PROMISC) ||
+	    (netdev_uc_count(netdev) > OTX2_MAX_UNICAST_FLOWS)) {
+		promisc = true;
+	}
+
+	/* Write unicast address to mcam entries or del from mcam */
+	if (!promisc && netdev->priv_flags & IFF_UNICAST_FLT)
+		__dev_uc_sync(netdev, otx2_add_macfilter, otx2_del_macfilter);
+
+	mutex_lock(&pf->mbox.lock);
+	req = otx2_mbox_alloc_msg_nix_set_rx_mode(&pf->mbox);
+	if (!req) {
+		mutex_unlock(&pf->mbox.lock);
+		return;
+	}
+
+	req->mode = NIX_RX_MODE_UCAST;
+
+	if (promisc)
+		req->mode |= NIX_RX_MODE_PROMISC;
+	if (netdev->flags & (IFF_ALLMULTI | IFF_MULTICAST))
+		req->mode |= NIX_RX_MODE_ALLMULTI;
+
+	req->mode |= NIX_RX_MODE_USE_MCE;
+
+	otx2_sync_mbox_msg(&pf->mbox);
+	mutex_unlock(&pf->mbox.lock);
+}
+
 int otx2_open(struct net_device *netdev)
 {
 	struct otx2_nic *pf = netdev_priv(netdev);
@@ -1503,7 +1548,7 @@ int otx2_open(struct net_device *netdev)
 
 	netif_carrier_off(netdev);
 
-	pf->qset.cq_cnt = pf->hw.rx_queues + pf->hw.tx_queues;
+	pf->qset.cq_cnt = pf->hw.rx_queues + pf->hw.tot_tx_queues;
 	/* RQ and SQs are mapped to different CQs,
 	 * so find out max CQ IRQs (i.e CINTs) needed.
 	 */
@@ -1523,7 +1568,7 @@ int otx2_open(struct net_device *netdev)
 	if (!qset->cq)
 		goto err_free_mem;
 
-	qset->sq = kcalloc(pf->hw.tx_queues,
+	qset->sq = kcalloc(pf->hw.tot_tx_queues,
 			   sizeof(struct otx2_snd_queue), GFP_KERNEL);
 	if (!qset->sq)
 		goto err_free_mem;
@@ -1544,11 +1589,20 @@ int otx2_open(struct net_device *netdev)
 		/* RQ0 & SQ0 are mapped to CINT0 and so on..
 		 * 'cq_ids[0]' points to RQ's CQ and
 		 * 'cq_ids[1]' points to SQ's CQ and
+		 * 'cq_ids[2]' points to XDP's CQ and
 		 */
 		cq_poll->cq_ids[CQ_RX] =
 			(qidx <  pf->hw.rx_queues) ? qidx : CINT_INVALID_CQ;
 		cq_poll->cq_ids[CQ_TX] = (qidx < pf->hw.tx_queues) ?
 				      qidx + pf->hw.rx_queues : CINT_INVALID_CQ;
+		if (pf->xdp_prog)
+			cq_poll->cq_ids[CQ_XDP] = (qidx < pf->hw.xdp_queues) ?
+						  (qidx + pf->hw.rx_queues +
+						  pf->hw.tx_queues) :
+						  CINT_INVALID_CQ;
+		else
+			cq_poll->cq_ids[CQ_XDP] = CINT_INVALID_CQ;
+
 		cq_poll->dev = (void *)pf;
 		netif_napi_add(netdev, &cq_poll->napi,
 			       otx2_napi_handler, NAPI_POLL_WEIGHT);
@@ -1646,6 +1700,8 @@ int otx2_open(struct net_device *netdev)
 	if (err)
 		goto err_tx_stop_queues;
 
+	otx2_do_set_rx_mode(pf);
+
 	return 0;
 
 err_tx_stop_queues:
@@ -1750,7 +1806,7 @@ static netdev_tx_t otx2_xmit(struct sk_buff *skb, struct net_device *netdev)
 
 	/* Check for minimum and maximum packet length */
 	if (skb->len <= ETH_HLEN ||
-	    (!skb_shinfo(skb)->gso_size && skb->len > pf->max_frs)) {
+	    (!skb_shinfo(skb)->gso_size && skb->len > pf->tx_max_pktlen)) {
 		dev_kfree_skb(skb);
 		return NETDEV_TX_OK;
 	}
@@ -1791,43 +1847,11 @@ static void otx2_set_rx_mode(struct net_device *netdev)
 	queue_work(pf->otx2_wq, &pf->rx_mode_work);
 }
 
-static void otx2_do_set_rx_mode(struct work_struct *work)
+static void otx2_rx_mode_wrk_handler(struct work_struct *work)
 {
 	struct otx2_nic *pf = container_of(work, struct otx2_nic, rx_mode_work);
-	struct net_device *netdev = pf->netdev;
-	struct nix_rx_mode *req;
-	bool promisc = false;
-
-	if (!(netdev->flags & IFF_UP))
-		return;
-
-	if ((netdev->flags & IFF_PROMISC) ||
-	    (netdev_uc_count(netdev) > OTX2_MAX_UNICAST_FLOWS)) {
-		promisc = true;
-	}
-
-	/* Write unicast address to mcam entries or del from mcam */
-	if (!promisc && netdev->priv_flags & IFF_UNICAST_FLT)
-		__dev_uc_sync(netdev, otx2_add_macfilter, otx2_del_macfilter);
-
-	mutex_lock(&pf->mbox.lock);
-	req = otx2_mbox_alloc_msg_nix_set_rx_mode(&pf->mbox);
-	if (!req) {
-		mutex_unlock(&pf->mbox.lock);
-		return;
-	}
-
-	req->mode = NIX_RX_MODE_UCAST;
-
-	if (promisc)
-		req->mode |= NIX_RX_MODE_PROMISC;
-	if (netdev->flags & (IFF_ALLMULTI | IFF_MULTICAST))
-		req->mode |= NIX_RX_MODE_ALLMULTI;
 
-	req->mode |= NIX_RX_MODE_USE_MCE;
-
-	otx2_sync_mbox_msg(&pf->mbox);
-	mutex_unlock(&pf->mbox.lock);
+	otx2_do_set_rx_mode(pf);
 }
 
 static int otx2_set_features(struct net_device *netdev,
@@ -1967,7 +1991,7 @@ static int otx2_config_hw_tx_tstamp(struct otx2_nic *pfvf, bool enable)
 	return 0;
 }
 
-static int otx2_config_hwtstamp(struct net_device *netdev, struct ifreq *ifr)
+int otx2_config_hwtstamp(struct net_device *netdev, struct ifreq *ifr)
 {
 	struct otx2_nic *pfvf = netdev_priv(netdev);
 	struct hwtstamp_config config;
@@ -2023,8 +2047,9 @@ static int otx2_config_hwtstamp(struct net_device *netdev, struct ifreq *ifr)
 	return copy_to_user(ifr->ifr_data, &config,
 			    sizeof(config)) ? -EFAULT : 0;
 }
+EXPORT_SYMBOL(otx2_config_hwtstamp);
 
-static int otx2_ioctl(struct net_device *netdev, struct ifreq *req, int cmd)
+int otx2_ioctl(struct net_device *netdev, struct ifreq *req, int cmd)
 {
 	struct otx2_nic *pfvf = netdev_priv(netdev);
 	struct hwtstamp_config *cfg = &pfvf->tstamp;
@@ -2039,6 +2064,7 @@ static int otx2_ioctl(struct net_device *netdev, struct ifreq *req, int cmd)
 		return -EOPNOTSUPP;
 	}
 }
+EXPORT_SYMBOL(otx2_ioctl);
 
 static int otx2_do_set_vf_mac(struct otx2_nic *pf, int vf, const u8 *mac)
 {
@@ -2281,6 +2307,111 @@ static int otx2_get_vf_config(struct net_device *netdev, int vf,
 	return 0;
 }
 
+static int otx2_xdp_xmit_tx(struct otx2_nic *pf, struct xdp_frame *xdpf,
+			    int qidx)
+{
+	struct page *page;
+	u64 dma_addr;
+	int err = 0;
+
+	dma_addr = otx2_dma_map_page(pf, virt_to_page(xdpf->data),
+				     offset_in_page(xdpf->data), xdpf->len,
+				     DMA_TO_DEVICE);
+	if (dma_mapping_error(pf->dev, dma_addr))
+		return -ENOMEM;
+
+	err = otx2_xdp_sq_append_pkt(pf, dma_addr, xdpf->len, qidx);
+	if (!err) {
+		otx2_dma_unmap_page(pf, dma_addr, xdpf->len, DMA_TO_DEVICE);
+		page = virt_to_page(xdpf->data);
+		put_page(page);
+		return -ENOMEM;
+	}
+	return 0;
+}
+
+static int otx2_xdp_xmit(struct net_device *netdev, int n,
+			 struct xdp_frame **frames, u32 flags)
+{
+	struct otx2_nic *pf = netdev_priv(netdev);
+	int qidx = smp_processor_id();
+	struct otx2_snd_queue *sq;
+	int drops = 0, i;
+
+	if (!netif_running(netdev))
+		return -ENETDOWN;
+
+	qidx += pf->hw.tx_queues;
+	sq = pf->xdp_prog ? &pf->qset.sq[qidx] : NULL;
+
+	/* Abort xmit if xdp queue is not */
+	if (unlikely(!sq))
+		return -ENXIO;
+
+	if (unlikely(flags & ~XDP_XMIT_FLAGS_MASK))
+		return -EINVAL;
+
+	for (i = 0; i < n; i++) {
+		struct xdp_frame *xdpf = frames[i];
+		int err;
+
+		err = otx2_xdp_xmit_tx(pf, xdpf, qidx);
+		if (err)
+			drops++;
+	}
+	return n - drops;
+}
+
+static int otx2_xdp_setup(struct otx2_nic *pf, struct bpf_prog *prog)
+{
+	struct net_device *dev = pf->netdev;
+	bool if_up = netif_running(pf->netdev);
+	struct bpf_prog *old_prog;
+
+	if (prog && dev->mtu > MAX_XDP_MTU) {
+		netdev_warn(dev, "Jumbo frames not yet supported with XDP\n");
+		return -EOPNOTSUPP;
+	}
+
+	if (if_up)
+		otx2_stop(pf->netdev);
+
+	old_prog = xchg(&pf->xdp_prog, prog);
+
+	if (old_prog)
+		bpf_prog_put(old_prog);
+
+	if (pf->xdp_prog)
+		bpf_prog_add(pf->xdp_prog, pf->hw.rx_queues - 1);
+
+	/* Network stack and XDP shared same rx queues.
+	 * Use separate tx queues for XDP and network stack.
+	 */
+	if (pf->xdp_prog)
+		pf->hw.xdp_queues = pf->hw.rx_queues;
+	else
+		pf->hw.xdp_queues = 0;
+
+	pf->hw.tot_tx_queues += pf->hw.xdp_queues;
+
+	if (if_up)
+		otx2_open(pf->netdev);
+
+	return 0;
+}
+
+static int otx2_xdp(struct net_device *netdev, struct netdev_bpf *xdp)
+{
+	struct otx2_nic *pf = netdev_priv(netdev);
+
+	switch (xdp->command) {
+	case XDP_SETUP_PROG:
+		return otx2_xdp_setup(pf, xdp->prog);
+	default:
+		return -EINVAL;
+	}
+}
+
 static int otx2_set_vf_permissions(struct otx2_nic *pf, int vf,
 				   int req_perm)
 {
@@ -2348,6 +2479,8 @@ static const struct net_device_ops otx2_netdev_ops = {
 	.ndo_set_vf_mac		= otx2_set_vf_mac,
 	.ndo_set_vf_vlan	= otx2_set_vf_vlan,
 	.ndo_get_vf_config	= otx2_get_vf_config,
+	.ndo_bpf		= otx2_xdp,
+	.ndo_xdp_xmit           = otx2_xdp_xmit,
 	.ndo_setup_tc		= otx2_setup_tc,
 	.ndo_set_vf_trust	= otx2_ndo_set_vf_trust,
 };
@@ -2358,7 +2491,7 @@ static int otx2_wq_init(struct otx2_nic *pf)
 	if (!pf->otx2_wq)
 		return -ENOMEM;
 
-	INIT_WORK(&pf->rx_mode_work, otx2_do_set_rx_mode);
+	INIT_WORK(&pf->rx_mode_work, otx2_rx_mode_wrk_handler);
 	INIT_WORK(&pf->reset_task, otx2_reset_task);
 	return 0;
 }
@@ -2489,6 +2622,7 @@ static int otx2_probe(struct pci_dev *pdev, const struct pci_device_id *id)
 	hw->pdev = pdev;
 	hw->rx_queues = qcount;
 	hw->tx_queues = qcount;
+	hw->tot_tx_queues = qcount;
 	hw->max_queues = qcount;
 
 	num_vec = pci_msix_vec_count(pdev);