net: lan966x: Add FDMA functionality

Ethernet frames can be extracted or injected to or from the device's
DDR memory. There is one channel for injection and one channel for
extraction. Each of these channels contain a linked list of DCBs which
contains DB. The DCB contains only 1 DB for both the injection and
extraction. Each DB contains a frame. Every time when a frame is received
or transmitted an interrupt is generated.

It is not possible to use both the FDMA and the manual
injection/extraction of the frames. Therefore the FDMA has priority over
the manual because of better performance values.

FDMA:
iperf -c 192.168.1.1
[  5]   0.00-10.02  sec   420 MBytes   352 Mbits/sec    0 sender
[  5]   0.00-10.03  sec   420 MBytes   351 Mbits/sec      receiver

iperf -c 192.168.1.1 -R
[  5]   0.00-10.01  sec   528 MBytes   442 Mbits/sec    0 sender
[  5]   0.00-10.00  sec   524 MBytes   440 Mbits/sec      receiver

Manual:
iperf -c 192.168.1.1
[  5]   0.00-10.02  sec  93.8 MBytes  78.5 Mbits/sec    0 sender
[  5]   0.00-10.03  sec  93.8 MBytes  78.4 Mbits/sec      receiver

ipers -c 192.168.1.1 -R
[  5]   0.00-10.03  sec   121 MBytes   101 Mbits/sec    0 sender
[  5]   0.00-10.01  sec   118 MBytes  99.0 Mbits/sec      receiver
Signed-off-by: Horatiu Vultur <horatiu.vultur@microchip.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

drivers/net/ethernet/microchip/lan966x/Makefile

@@ -8,4 +8,4 @@ obj-$(CONFIG_LAN966X_SWITCH) += lan966x-switch.o
 lan966x-switch-objs  := lan966x_main.o lan966x_phylink.o lan966x_port.o \
 			lan966x_mac.o lan966x_ethtool.o lan966x_switchdev.o \
 			lan966x_vlan.o lan966x_fdb.o lan966x_mdb.o \
-			lan966x_ptp.o
+			lan966x_ptp.o lan966x_fdma.o

drivers/net/ethernet/microchip/lan966x/lan966x_main.c

@@ -341,7 +341,10 @@ static int lan966x_port_xmit(struct sk_buff *skb, struct net_device *dev)
 	}
 
 	spin_lock(&lan966x->tx_lock);
-	err = lan966x_port_ifh_xmit(skb, ifh, dev);
+	if (port->lan966x->fdma)
+		err = lan966x_fdma_xmit(skb, ifh, dev);
+	else
+		err = lan966x_port_ifh_xmit(skb, ifh, dev);
 	spin_unlock(&lan966x->tx_lock);
 
 	return err;
@@ -643,6 +646,9 @@ static void lan966x_cleanup_ports(struct lan966x *lan966x)
 		if (port->dev)
 			unregister_netdev(port->dev);
 
+		if (lan966x->fdma && lan966x->fdma_ndev == port->dev)
+			lan966x_fdma_netdev_deinit(lan966x, port->dev);
+
 		if (port->phylink) {
 			rtnl_lock();
 			lan966x_port_stop(port->dev);
@@ -662,6 +668,9 @@ static void lan966x_cleanup_ports(struct lan966x *lan966x)
 		disable_irq(lan966x->ana_irq);
 		lan966x->ana_irq = -ENXIO;
 	}
+
+	if (lan966x->fdma)
+		devm_free_irq(lan966x->dev, lan966x->fdma_irq, lan966x);
 }
 
 static int lan966x_probe_port(struct lan966x *lan966x, u32 p,
@@ -790,12 +799,12 @@ static void lan966x_init(struct lan966x *lan966x)
 	/* Do byte-swap and expect status after last data word
 	 * Extraction: Mode: manual extraction) | Byte_swap
 	 */
-	lan_wr(QS_XTR_GRP_CFG_MODE_SET(1) |
+	lan_wr(QS_XTR_GRP_CFG_MODE_SET(lan966x->fdma ? 2 : 1) |
 	       QS_XTR_GRP_CFG_BYTE_SWAP_SET(1),
 	       lan966x, QS_XTR_GRP_CFG(0));
 
 	/* Injection: Mode: manual injection | Byte_swap */
-	lan_wr(QS_INJ_GRP_CFG_MODE_SET(1) |
+	lan_wr(QS_INJ_GRP_CFG_MODE_SET(lan966x->fdma ? 2 : 1) |
 	       QS_INJ_GRP_CFG_BYTE_SWAP_SET(1),
 	       lan966x, QS_INJ_GRP_CFG(0));
 
@@ -1017,6 +1026,17 @@ static int lan966x_probe(struct platform_device *pdev)
 		lan966x->ptp = 1;
 	}
 
+	lan966x->fdma_irq = platform_get_irq_byname(pdev, "fdma");
+	if (lan966x->fdma_irq > 0) {
+		err = devm_request_irq(&pdev->dev, lan966x->fdma_irq,
+				       lan966x_fdma_irq_handler, 0,
+				       "fdma irq", lan966x);
+		if (err)
+			return dev_err_probe(&pdev->dev, err, "Unable to use fdma irq");
+
+		lan966x->fdma = true;
+	}
+
 	/* init switch */
 	lan966x_init(lan966x);
 	lan966x_stats_init(lan966x);
@@ -1055,8 +1075,15 @@ static int lan966x_probe(struct platform_device *pdev)
 	if (err)
 		goto cleanup_fdb;
 
+	err = lan966x_fdma_init(lan966x);
+	if (err)
+		goto cleanup_ptp;
+
 	return 0;
 
+cleanup_ptp:
+	lan966x_ptp_deinit(lan966x);
+
 cleanup_fdb:
 	lan966x_fdb_deinit(lan966x);
 
@@ -1076,6 +1103,7 @@ static int lan966x_remove(struct platform_device *pdev)
 {
 	struct lan966x *lan966x = platform_get_drvdata(pdev);
 
+	lan966x_fdma_deinit(lan966x);
 	lan966x_cleanup_ports(lan966x);
 
 	cancel_delayed_work_sync(&lan966x->stats_work);

drivers/net/ethernet/microchip/lan966x/lan966x_main.h

@@ -61,6 +61,22 @@
 #define IFH_REW_OP_ONE_STEP_PTP		0x3
 #define IFH_REW_OP_TWO_STEP_PTP		0x4
 
+#define FDMA_RX_DCB_MAX_DBS		1
+#define FDMA_TX_DCB_MAX_DBS		1
+#define FDMA_DCB_INFO_DATAL(x)		((x) & GENMASK(15, 0))
+
+#define FDMA_DCB_STATUS_BLOCKL(x)	((x) & GENMASK(15, 0))
+#define FDMA_DCB_STATUS_SOF		BIT(16)
+#define FDMA_DCB_STATUS_EOF		BIT(17)
+#define FDMA_DCB_STATUS_INTR		BIT(18)
+#define FDMA_DCB_STATUS_DONE		BIT(19)
+#define FDMA_DCB_STATUS_BLOCKO(x)	(((x) << 20) & GENMASK(31, 20))
+#define FDMA_DCB_INVALID_DATA		0x1
+
+#define FDMA_XTR_CHANNEL		6
+#define FDMA_INJ_CHANNEL		0
+#define FDMA_DCB_MAX			512
+
 /* MAC table entry types.
  * ENTRYTYPE_NORMAL is subject to aging.
  * ENTRYTYPE_LOCKED is not subject to aging.
@@ -76,6 +92,83 @@ enum macaccess_entry_type {
 
 struct lan966x_port;
 
+struct lan966x_db {
+	u64 dataptr;
+	u64 status;
+};
+
+struct lan966x_rx_dcb {
+	u64 nextptr;
+	u64 info;
+	struct lan966x_db db[FDMA_RX_DCB_MAX_DBS];
+};
+
+struct lan966x_tx_dcb {
+	u64 nextptr;
+	u64 info;
+	struct lan966x_db db[FDMA_TX_DCB_MAX_DBS];
+};
+
+struct lan966x_rx {
+	struct lan966x *lan966x;
+
+	/* Pointer to the array of hardware dcbs. */
+	struct lan966x_rx_dcb *dcbs;
+
+	/* Pointer to the last address in the dcbs. */
+	struct lan966x_rx_dcb *last_entry;
+
+	/* For each DB, there is a page */
+	struct page *page[FDMA_DCB_MAX][FDMA_RX_DCB_MAX_DBS];
+
+	/* Represents the db_index, it can have a value between 0 and
+	 * FDMA_RX_DCB_MAX_DBS, once it reaches the value of FDMA_RX_DCB_MAX_DBS
+	 * it means that the DCB can be reused.
+	 */
+	int db_index;
+
+	/* Represents the index in the dcbs. It has a value between 0 and
+	 * FDMA_DCB_MAX
+	 */
+	int dcb_index;
+
+	/* Represents the dma address to the dcbs array */
+	dma_addr_t dma;
+
+	/* Represents the page order that is used to allocate the pages for the
+	 * RX buffers. This value is calculated based on max MTU of the devices.
+	 */
+	u8 page_order;
+
+	u8 channel_id;
+};
+
+struct lan966x_tx_dcb_buf {
+	struct net_device *dev;
+	struct sk_buff *skb;
+	dma_addr_t dma_addr;
+	bool used;
+	bool ptp;
+};
+
+struct lan966x_tx {
+	struct lan966x *lan966x;
+
+	/* Pointer to the dcb list */
+	struct lan966x_tx_dcb *dcbs;
+	u16 last_in_use;
+
+	/* Represents the DMA address to the first entry of the dcb entries. */
+	dma_addr_t dma;
+
+	/* Array of dcbs that are given to the HW */
+	struct lan966x_tx_dcb_buf *dcbs_buf;
+
+	u8 channel_id;
+
+	bool activated;
+};
+
 struct lan966x_stat_layout {
 	u32 offset;
 	char name[ETH_GSTRING_LEN];
@@ -137,6 +230,7 @@ struct lan966x {
 	int xtr_irq;
 	int ana_irq;
 	int ptp_irq;
+	int fdma_irq;
 
 	/* worqueue for fdb */
 	struct workqueue_struct *fdb_work;
@@ -153,6 +247,13 @@ struct lan966x {
 	spinlock_t ptp_ts_id_lock; /* lock for ts_id */
 	struct mutex ptp_lock; /* lock for ptp interface state */
 	u16 ptp_skbs;
+
+	/* fdma */
+	bool fdma;
+	struct net_device *fdma_ndev;
+	struct lan966x_rx rx;
+	struct lan966x_tx tx;
+	struct napi_struct napi;
 };
 
 struct lan966x_port_config {
@@ -292,6 +393,13 @@ void lan966x_ptp_txtstamp_release(struct lan966x_port *port,
 				  struct sk_buff *skb);
 irqreturn_t lan966x_ptp_irq_handler(int irq, void *args);
 
+int lan966x_fdma_xmit(struct sk_buff *skb, __be32 *ifh, struct net_device *dev);
+void lan966x_fdma_netdev_init(struct lan966x *lan966x, struct net_device *dev);
+void lan966x_fdma_netdev_deinit(struct lan966x *lan966x, struct net_device *dev);
+int lan966x_fdma_init(struct lan966x *lan966x);
+void lan966x_fdma_deinit(struct lan966x *lan966x);
+irqreturn_t lan966x_fdma_irq_handler(int irq, void *args);
+
 static inline void __iomem *lan_addr(void __iomem *base[],
 				     int id, int tinst, int tcnt,
 				     int gbase, int ginst,

drivers/net/ethernet/microchip/lan966x/lan966x_port.c

@@ -393,6 +393,9 @@ void lan966x_port_init(struct lan966x_port *port)
 
 	lan966x_port_config_down(port);
 
+	if (lan966x->fdma)
+		lan966x_fdma_netdev_init(lan966x, port->dev);
+
 	if (config->portmode != PHY_INTERFACE_MODE_QSGMII)
 		return;