ENGR00179636-01 - FEC : Enet RX FIFO overruns issue.

- Add NAPI methods.
  NAPI can improve the performance of high-speed networking,
  which can reduce the cpu loading of interrupt generate and
  drop packets.
- Enet RX FIFO overruns number has been reduced by NAPI method.

 Signed-off-by: Fugang Duan <B38611@freescale.com>

2 files changed, 206 insertions, 5 deletions

diff --git a/drivers/net/Kconfig b/drivers/net/Kconfig
index a6ccb2243b57..8e799247cd48 100644
--- a/drivers/net/Kconfig
+++ b/drivers/net/Kconfig
@@ -1951,6 +1951,13 @@ config FEC
 	  Say Y here if you want to use the built-in 10/100 Fast ethernet
 	  controller on some Motorola ColdFire and Freescale i.MX processors.
 
+config FEC_NAPI
+       bool "Use NAPI for FEC driver"
+       depends on FEC
+       help
+         This option enables NAPI support for the FEC's on-chip
+         Fast Ethernet Controller driver.
+
 config FEC_1588
 	bool "Enable FEC 1588 timestamping"
 	depends on FEC
diff --git a/drivers/net/fec.c b/drivers/net/fec.c
index 2078ef8d25f0..a93ac31e690f 100755
--- a/drivers/net/fec.c
+++ b/drivers/net/fec.c
@@ -230,10 +230,22 @@ struct fec_enet_private {
 
 	struct  fec_ptp_private *ptp_priv;
 	uint    ptimer_present;
+
+	struct napi_struct napi;
+	int	napi_weight;
+	bool	use_napi;
 };
 
+#define FEC_NAPI_WEIGHT 64
+#ifdef CONFIG_FEC_NAPI
+#define FEC_NAPI_ENABLE TRUE
+#else
+#define FEC_NAPI_ENABLE FALSE
+#endif
+
 static irqreturn_t fec_enet_interrupt(int irq, void * dev_id);
 static void fec_enet_tx(struct net_device *dev);
+static int fec_rx_poll(struct napi_struct *napi, int budget);
 static void fec_enet_rx(struct net_device *dev);
 static int fec_enet_close(struct net_device *dev);
 static void fec_restart(struct net_device *dev, int duplex);
@@ -390,6 +402,29 @@ fec_timeout(struct net_device *ndev)
 }
 
 static void
+fec_rx_int_is_enabled(struct net_device *ndev, bool enabled)
+{
+	struct fec_enet_private *fep = netdev_priv(ndev);
+	uint    int_events;
+
+	int_events = readl(fep->hwp + FEC_IMASK);
+	if (enabled)
+		int_events |= FEC_ENET_RXF;
+	else
+		int_events &= ~FEC_ENET_RXF;
+	writel(int_events, fep->hwp + FEC_IMASK);
+}
+
+#ifdef CONFIG_NET_POLL_CONTROLLER
+static void fec_enet_netpoll(struct net_device *ndev)
+{
+	disable_irq(ndev->irq);
+	fec_enet_interrupt(ndev->irq, ndev);
+	enable_irq(ndev->irq);
+}
+#endif
+
+static void
 fec_enet_tx(struct net_device *ndev)
 {
 	struct	fec_enet_private *fep;
@@ -478,6 +513,144 @@ fec_enet_tx(struct net_device *ndev)
 	spin_unlock(&fep->hw_lock);
 }
 
+/*NAPI polling Receive packets */
+static int fec_rx_poll(struct napi_struct *napi, int budget)
+{
+	struct  fec_enet_private *fep =
+		container_of(napi, struct fec_enet_private, napi);
+	struct net_device *ndev = napi->dev;
+	struct  fec_ptp_private *fpp = fep->ptp_priv;
+	const struct platform_device_id *id_entry =
+				platform_get_device_id(fep->pdev);
+	int pkt_received = 0;
+	struct bufdesc *bdp;
+	unsigned short status;
+	struct	sk_buff	*skb;
+	ushort	pkt_len;
+	__u8 *data;
+
+	if (fep->use_napi)
+		WARN_ON(!budget);
+
+#ifdef CONFIG_M532x
+	flush_cache_all();
+#endif
+
+	/* First, grab all of the stats for the incoming packet.
+	 * These get messed up if we get called due to a busy condition.
+	 */
+	bdp = fep->cur_rx;
+
+	while (!((status = bdp->cbd_sc) & BD_ENET_RX_EMPTY)) {
+		if (pkt_received >= budget)
+			break;
+		pkt_received++;
+
+		/* Since we have allocated space to hold a complete frame,
+		 * the last indicator should be set.
+		 */
+		if ((status & BD_ENET_RX_LAST) == 0)
+			dev_err(&ndev->dev, "FEC ENET: rcv is not +last\n");
+
+		if (!fep->opened)
+			goto rx_processing_done;
+
+		/* Check for errors. */
+		if (status & (BD_ENET_RX_LG | BD_ENET_RX_SH | BD_ENET_RX_NO |
+			   BD_ENET_RX_CR | BD_ENET_RX_OV)) {
+			ndev->stats.rx_errors++;
+			if (status & (BD_ENET_RX_LG | BD_ENET_RX_SH)) {
+				/* Frame too long or too short. */
+				ndev->stats.rx_length_errors++;
+			}
+			if (status & BD_ENET_RX_NO)	/* Frame alignment */
+				ndev->stats.rx_frame_errors++;
+			if (status & BD_ENET_RX_CR)	/* CRC Error */
+				ndev->stats.rx_crc_errors++;
+			if (status & BD_ENET_RX_OV)	/* FIFO overrun */
+				ndev->stats.rx_fifo_errors++;
+		}
+
+		/* Report late collisions as a frame error.
+		 * On this error, the BD is closed, but we don't know what we
+		 * have in the buffer.  So, just drop this frame on the floor.
+		 */
+		if (status & BD_ENET_RX_CL) {
+			ndev->stats.rx_errors++;
+			ndev->stats.rx_frame_errors++;
+			goto rx_processing_done;
+		}
+
+		/* Process the incoming frame. */
+		ndev->stats.rx_packets++;
+		pkt_len = bdp->cbd_datlen;
+		ndev->stats.rx_bytes += pkt_len;
+		data = (__u8 *)__va(bdp->cbd_bufaddr);
+
+		if (bdp->cbd_bufaddr)
+			dma_unmap_single(&ndev->dev, bdp->cbd_bufaddr,
+				FEC_ENET_RX_FRSIZE, DMA_FROM_DEVICE);
+
+		if (id_entry->driver_data & FEC_QUIRK_SWAP_FRAME)
+			swap_buffer(data, pkt_len);
+
+		/* This does 16 byte alignment, exactly what we need.
+		 * The packet length includes FCS, but we don't want to
+		 * include that when passing upstream as it messes up
+		 * bridging applications.
+		 */
+		skb = dev_alloc_skb(pkt_len - 4 + NET_IP_ALIGN);
+
+		if (unlikely(!skb)) {
+			dev_err(&ndev->dev,
+			"%s: Memory squeeze, dropping packet.\n", ndev->name);
+			ndev->stats.rx_dropped++;
+		} else {
+			skb_reserve(skb, NET_IP_ALIGN);
+			skb_put(skb, pkt_len - 4);	/* Make room */
+			skb_copy_to_linear_data(skb, data, pkt_len - 4);
+			/* 1588 messeage TS handle */
+			if (fep->ptimer_present)
+				fec_ptp_store_rxstamp(fpp, skb, bdp);
+			skb->protocol = eth_type_trans(skb, ndev);
+			netif_receive_skb(skb);
+		}
+
+		bdp->cbd_bufaddr = dma_map_single(&ndev->dev, data,
+				FEC_ENET_RX_FRSIZE, DMA_FROM_DEVICE);
+rx_processing_done:
+		/* Clear the status flags for this buffer */
+		status &= ~BD_ENET_RX_STATS;
+
+		/* Mark the buffer empty */
+		status |= BD_ENET_RX_EMPTY;
+		bdp->cbd_sc = status;
+#ifdef CONFIG_ENHANCED_BD
+		bdp->cbd_esc = BD_ENET_RX_INT;
+		bdp->cbd_prot = 0;
+		bdp->cbd_bdu = 0;
+#endif
+
+		/* Update BD pointer to next entry */
+		if (status & BD_ENET_RX_WRAP)
+			bdp = fep->rx_bd_base;
+		else
+			bdp++;
+		/* Doing this here will keep the FEC running while we process
+		 * incoming frames.  On a heavily loaded network, we should be
+		 * able to keep up at the expense of system resources.
+		 */
+		writel(0, fep->hwp + FEC_R_DES_ACTIVE);
+	}
+	fep->cur_rx = bdp;
+
+	if (pkt_received < budget) {
+		napi_complete(napi);
+		fec_rx_int_is_enabled(ndev, true);
+	}
+
+	return pkt_received;
+}
 
 /* During a receive, the cur_rx points to the current incoming buffer.
  * When we update through the ring, if the next incoming buffer has
@@ -501,8 +674,6 @@ fec_enet_rx(struct net_device *ndev)
 	flush_cache_all();
 #endif
 
-	spin_lock(&fep->hw_lock);
-
 	/* First, grab all of the stats for the incoming packet.
 	 * These get messed up if we get called due to a busy condition.
 	 */
@@ -607,8 +778,6 @@ rx_processing_done:
 		writel(0, fep->hwp + FEC_R_DES_ACTIVE);
 	}
 	fep->cur_rx = bdp;
-
-	spin_unlock(&fep->hw_lock);
 }
 
 static irqreturn_t
@@ -618,6 +787,7 @@ fec_enet_interrupt(int irq, void *dev_id)
 	struct fec_enet_private *fep = netdev_priv(ndev);
 	struct fec_ptp_private *fpp = fep->ptp_priv;
 	uint int_events;
+	ulong flags;
 	irqreturn_t ret = IRQ_NONE;
 
 	do {
@@ -626,7 +796,18 @@ fec_enet_interrupt(int irq, void *dev_id)
 
 		if (int_events & FEC_ENET_RXF) {
 			ret = IRQ_HANDLED;
-			fec_enet_rx(ndev);
+			spin_lock_irqsave(&fep->hw_lock, flags);
+
+			if (fep->use_napi) {
+				/* Disable the RX interrupt */
+				if (napi_schedule_prep(&fep->napi)) {
+					fec_rx_int_is_enabled(ndev, false);
+					__napi_schedule(&fep->napi);
+				}
+			} else
+				fec_enet_rx(ndev);
+
+			spin_unlock_irqrestore(&fep->hw_lock, flags);
 		}
 
 		/* Transmit OK, or non-fatal error. Update the buffer
@@ -1106,6 +1287,9 @@ fec_enet_open(struct net_device *ndev)
 	struct fec_platform_data *pdata = fep->pdev->dev.platform_data;
 	int ret;
 
+	if (fep->use_napi)
+		napi_enable(&fep->napi);
+
 	/* I should reset the ring buffers here, but I don't yet know
 	 * a simple way to do that.
 	 */
@@ -1266,6 +1450,9 @@ static const struct net_device_ops fec_netdev_ops = {
 	.ndo_tx_timeout		= fec_timeout,
 	.ndo_set_mac_address	= fec_set_mac_address,
 	.ndo_do_ioctl		= fec_enet_ioctl,
+#ifdef CONFIG_NET_POLL_CONTROLLER
+	.ndo_poll_controller    = fec_enet_netpoll,
+#endif
 };
 
  /*
@@ -1303,6 +1490,13 @@ static int fec_enet_init(struct net_device *ndev)
 	ndev->netdev_ops = &fec_netdev_ops;
 	ndev->ethtool_ops = &fec_enet_ethtool_ops;
 
+	fep->use_napi = FEC_NAPI_ENABLE;
+	fep->napi_weight = FEC_NAPI_WEIGHT;
+	if (fep->use_napi) {
+		fec_rx_int_is_enabled(ndev, false);
+		netif_napi_add(ndev, &fep->napi, fec_rx_poll, fep->napi_weight);
+	}
+
 	/* Initialize the receive buffer descriptors. */
 	bdp = fep->rx_bd_base;
 	for (i = 0; i < RX_RING_SIZE; i++) {