Merge branch 'dpaa_eth-next' of git://git.freescale.com/ppc/upstream/linux

Madalin Bucur says:

====================
QorIQ DPAA 1 updates

This patch set introduces a series of fixes and features to the DPAA 1
drivers. Besides activating hardware Rx checksum offloading, four traffic
classes are added for Tx traffic prioritisation.

changes from v1: added patch to enable context-A stashing
====================

Signed-off-by: David S. Miller <davem@davemloft.net>

5 files changed, 248 insertions, 47 deletions

diff --git a/drivers/net/ethernet/freescale/dpaa/dpaa_eth.c b/drivers/net/ethernet/freescale/dpaa/dpaa_eth.c
index e2ca107f9d94..aa769cbc7425 100644
--- a/drivers/net/ethernet/freescale/dpaa/dpaa_eth.c
+++ b/drivers/net/ethernet/freescale/dpaa/dpaa_eth.c
@@ -137,6 +137,13 @@ MODULE_PARM_DESC(tx_timeout, "The Tx timeout in ms");
 /* L4 Type field: TCP */
 #define FM_L4_PARSE_RESULT_TCP	0x20
 
+/* FD status field indicating whether the FM Parser has attempted to validate
+ * the L4 csum of the frame.
+ * Note that having this bit set doesn't necessarily imply that the checksum
+ * is valid. One would have to check the parse results to find that out.
+ */
+#define FM_FD_STAT_L4CV         0x00000004
+
 #define DPAA_SGT_MAX_ENTRIES 16 /* maximum number of entries in SG Table */
 #define DPAA_BUFF_RELEASE_MAX 8 /* maximum number of buffers released at once */
 
@@ -235,6 +242,7 @@ static int dpaa_netdev_init(struct net_device *net_dev,
 	 * For conformity, we'll still declare GSO explicitly.
 	 */
 	net_dev->features |= NETIF_F_GSO;
+	net_dev->features |= NETIF_F_RXCSUM;
 
 	net_dev->priv_flags |= IFF_LIVE_ADDR_CHANGE;
 	/* we do not want shared skbs on TX */
@@ -334,6 +342,41 @@ static void dpaa_get_stats64(struct net_device *net_dev,
 	}
 }
 
+static int dpaa_setup_tc(struct net_device *net_dev, u32 handle, __be16 proto,
+			 struct tc_to_netdev *tc)
+{
+	struct dpaa_priv *priv = netdev_priv(net_dev);
+	int i;
+
+	if (tc->type != TC_SETUP_MQPRIO)
+		return -EINVAL;
+
+	if (tc->tc == priv->num_tc)
+		return 0;
+
+	if (!tc->tc) {
+		netdev_reset_tc(net_dev);
+		goto out;
+	}
+
+	if (tc->tc > DPAA_TC_NUM) {
+		netdev_err(net_dev, "Too many traffic classes: max %d supported.\n",
+			   DPAA_TC_NUM);
+		return -EINVAL;
+	}
+
+	netdev_set_num_tc(net_dev, tc->tc);
+
+	for (i = 0; i < tc->tc; i++)
+		netdev_set_tc_queue(net_dev, i, DPAA_TC_TXQ_NUM,
+				    i * DPAA_TC_TXQ_NUM);
+
+out:
+	priv->num_tc = tc->tc ? tc->tc : 1;
+	netif_set_real_num_tx_queues(net_dev, priv->num_tc * DPAA_TC_TXQ_NUM);
+	return 0;
+}
+
 static struct mac_device *dpaa_mac_dev_get(struct platform_device *pdev)
 {
 	struct platform_device *of_dev;
@@ -557,16 +600,18 @@ static void dpaa_bps_free(struct dpaa_priv *priv)
 
 /* Use multiple WQs for FQ assignment:
  *	- Tx Confirmation queues go to WQ1.
- *	- Rx Error and Tx Error queues go to WQ2 (giving them a better chance
- *	  to be scheduled, in case there are many more FQs in WQ3).
- *	- Rx Default and Tx queues go to WQ3 (no differentiation between
- *	  Rx and Tx traffic).
+ *	- Rx Error and Tx Error queues go to WQ5 (giving them a better chance
+ *	  to be scheduled, in case there are many more FQs in WQ6).
+ *	- Rx Default goes to WQ6.
+ *	- Tx queues go to different WQs depending on their priority. Equal
+ *	  chunks of NR_CPUS queues go to WQ6 (lowest priority), WQ2, WQ1 and
+ *	  WQ0 (highest priority).
  * This ensures that Tx-confirmed buffers are timely released. In particular,
  * it avoids congestion on the Tx Confirm FQs, which can pile up PFDRs if they
  * are greatly outnumbered by other FQs in the system, while
  * dequeue scheduling is round-robin.
  */
-static inline void dpaa_assign_wq(struct dpaa_fq *fq)
+static inline void dpaa_assign_wq(struct dpaa_fq *fq, int idx)
 {
 	switch (fq->fq_type) {
 	case FQ_TYPE_TX_CONFIRM:
@@ -575,11 +620,33 @@ static inline void dpaa_assign_wq(struct dpaa_fq *fq)
 		break;
 	case FQ_TYPE_RX_ERROR:
 	case FQ_TYPE_TX_ERROR:
-		fq->wq = 2;
+		fq->wq = 5;
 		break;
 	case FQ_TYPE_RX_DEFAULT:
+		fq->wq = 6;
+		break;
 	case FQ_TYPE_TX:
-		fq->wq = 3;
+		switch (idx / DPAA_TC_TXQ_NUM) {
+		case 0:
+			/* Low priority (best effort) */
+			fq->wq = 6;
+			break;
+		case 1:
+			/* Medium priority */
+			fq->wq = 2;
+			break;
+		case 2:
+			/* High priority */
+			fq->wq = 1;
+			break;
+		case 3:
+			/* Very high priority */
+			fq->wq = 0;
+			break;
+		default:
+			WARN(1, "Too many TX FQs: more than %d!\n",
+			     DPAA_ETH_TXQ_NUM);
+		}
 		break;
 	default:
 		WARN(1, "Invalid FQ type %d for FQID %d!\n",
@@ -607,7 +674,7 @@ static struct dpaa_fq *dpaa_fq_alloc(struct device *dev,
 	}
 
 	for (i = 0; i < count; i++)
-		dpaa_assign_wq(dpaa_fq + i);
+		dpaa_assign_wq(dpaa_fq + i, i);
 
 	return dpaa_fq;
 }
@@ -985,7 +1052,8 @@ static int dpaa_fq_init(struct dpaa_fq *dpaa_fq, bool td_enable)
 		/* Initialization common to all ingress queues */
 		if (dpaa_fq->flags & QMAN_FQ_FLAG_NO_ENQUEUE) {
 			initfq.we_mask |= cpu_to_be16(QM_INITFQ_WE_CONTEXTA);
-			initfq.fqd.fq_ctrl |= cpu_to_be16(QM_FQCTRL_HOLDACTIVE);
+			initfq.fqd.fq_ctrl |= cpu_to_be16(QM_FQCTRL_HOLDACTIVE |
+						QM_FQCTRL_CTXASTASHING);
 			initfq.fqd.context_a.stashing.exclusive =
 				QM_STASHING_EXCL_DATA | QM_STASHING_EXCL_CTX |
 				QM_STASHING_EXCL_ANNOTATION;
@@ -1055,9 +1123,9 @@ static int dpaa_fq_free(struct device *dev, struct list_head *list)
 	return err;
 }
 
-static void dpaa_eth_init_tx_port(struct fman_port *port, struct dpaa_fq *errq,
-				  struct dpaa_fq *defq,
-				  struct dpaa_buffer_layout *buf_layout)
+static int dpaa_eth_init_tx_port(struct fman_port *port, struct dpaa_fq *errq,
+				 struct dpaa_fq *defq,
+				 struct dpaa_buffer_layout *buf_layout)
 {
 	struct fman_buffer_prefix_content buf_prefix_content;
 	struct fman_port_params params;
@@ -1076,23 +1144,29 @@ static void dpaa_eth_init_tx_port(struct fman_port *port, struct dpaa_fq *errq,
 	params.specific_params.non_rx_params.dflt_fqid = defq->fqid;
 
 	err = fman_port_config(port, &params);
-	if (err)
+	if (err) {
 		pr_err("%s: fman_port_config failed\n", __func__);
+		return err;
+	}
 
 	err = fman_port_cfg_buf_prefix_content(port, &buf_prefix_content);
-	if (err)
+	if (err) {
 		pr_err("%s: fman_port_cfg_buf_prefix_content failed\n",
 		       __func__);
+		return err;
+	}
 
 	err = fman_port_init(port);
 	if (err)
 		pr_err("%s: fm_port_init failed\n", __func__);
+
+	return err;
 }
 
-static void dpaa_eth_init_rx_port(struct fman_port *port, struct dpaa_bp **bps,
-				  size_t count, struct dpaa_fq *errq,
-				  struct dpaa_fq *defq,
-				  struct dpaa_buffer_layout *buf_layout)
+static int dpaa_eth_init_rx_port(struct fman_port *port, struct dpaa_bp **bps,
+				 size_t count, struct dpaa_fq *errq,
+				 struct dpaa_fq *defq,
+				 struct dpaa_buffer_layout *buf_layout)
 {
 	struct fman_buffer_prefix_content buf_prefix_content;
 	struct fman_port_rx_params *rx_p;
@@ -1120,32 +1194,44 @@ static void dpaa_eth_init_rx_port(struct fman_port *port, struct dpaa_bp **bps,
 	}
 
 	err = fman_port_config(port, &params);
-	if (err)
+	if (err) {
 		pr_err("%s: fman_port_config failed\n", __func__);
+		return err;
+	}
 
 	err = fman_port_cfg_buf_prefix_content(port, &buf_prefix_content);
-	if (err)
+	if (err) {
 		pr_err("%s: fman_port_cfg_buf_prefix_content failed\n",
 		       __func__);
+		return err;
+	}
 
 	err = fman_port_init(port);
 	if (err)
 		pr_err("%s: fm_port_init failed\n", __func__);
+
+	return err;
 }
 
-static void dpaa_eth_init_ports(struct mac_device *mac_dev,
-				struct dpaa_bp **bps, size_t count,
-				struct fm_port_fqs *port_fqs,
-				struct dpaa_buffer_layout *buf_layout,
-				struct device *dev)
+static int dpaa_eth_init_ports(struct mac_device *mac_dev,
+			       struct dpaa_bp **bps, size_t count,
+			       struct fm_port_fqs *port_fqs,
+			       struct dpaa_buffer_layout *buf_layout,
+			       struct device *dev)
 {
 	struct fman_port *rxport = mac_dev->port[RX];
 	struct fman_port *txport = mac_dev->port[TX];
+	int err;
+
+	err = dpaa_eth_init_tx_port(txport, port_fqs->tx_errq,
+				    port_fqs->tx_defq, &buf_layout[TX]);
+	if (err)
+		return err;
+
+	err = dpaa_eth_init_rx_port(rxport, bps, count, port_fqs->rx_errq,
+				    port_fqs->rx_defq, &buf_layout[RX]);
 
-	dpaa_eth_init_tx_port(txport, port_fqs->tx_errq,
-			      port_fqs->tx_defq, &buf_layout[TX]);
-	dpaa_eth_init_rx_port(rxport, bps, count, port_fqs->rx_errq,
-			      port_fqs->rx_defq, &buf_layout[RX]);
+	return err;
 }
 
 static int dpaa_bman_release(const struct dpaa_bp *dpaa_bp,
@@ -1526,6 +1612,23 @@ static struct sk_buff *dpaa_cleanup_tx_fd(const struct dpaa_priv *priv,
 	return skb;
 }
 
+static u8 rx_csum_offload(const struct dpaa_priv *priv, const struct qm_fd *fd)
+{
+	/* The parser has run and performed L4 checksum validation.
+	 * We know there were no parser errors (and implicitly no
+	 * L4 csum error), otherwise we wouldn't be here.
+	 */
+	if ((priv->net_dev->features & NETIF_F_RXCSUM) &&
+	    (be32_to_cpu(fd->status) & FM_FD_STAT_L4CV))
+		return CHECKSUM_UNNECESSARY;
+
+	/* We're here because either the parser didn't run or the L4 checksum
+	 * was not verified. This may include the case of a UDP frame with
+	 * checksum zero or an L4 proto other than TCP/UDP
+	 */
+	return CHECKSUM_NONE;
+}
+
 /* Build a linear skb around the received buffer.
  * We are guaranteed there is enough room at the end of the data buffer to
  * accommodate the shared info area of the skb.
@@ -1556,7 +1659,7 @@ static struct sk_buff *contig_fd_to_skb(const struct dpaa_priv *priv,
 	skb_reserve(skb, fd_off);
 	skb_put(skb, qm_fd_get_length(fd));
 
-	skb->ip_summed = CHECKSUM_NONE;
+	skb->ip_summed = rx_csum_offload(priv, fd);
 
 	return skb;
 
@@ -1616,7 +1719,7 @@ static struct sk_buff *sg_fd_to_skb(const struct dpaa_priv *priv,
 			if (WARN_ON(unlikely(!skb)))
 				goto free_buffers;
 
-			skb->ip_summed = CHECKSUM_NONE;
+			skb->ip_summed = rx_csum_offload(priv, fd);
 
 			/* Make sure forwarded skbs will have enough space
 			 * on Tx, if extra headers are added.
@@ -2093,7 +2196,7 @@ static enum qman_cb_dqrr_result rx_default_dqrr(struct qman_portal *portal,
 	dma_addr_t addr = qm_fd_addr(fd);
 	enum qm_fd_format fd_format;
 	struct net_device *net_dev;
-	u32 fd_status = fd->status;
+	u32 fd_status;
 	struct dpaa_bp *dpaa_bp;
 	struct dpaa_priv *priv;
 	unsigned int skb_len;
@@ -2350,6 +2453,7 @@ static const struct net_device_ops dpaa_ops = {
 	.ndo_validate_addr = eth_validate_addr,
 	.ndo_set_rx_mode = dpaa_set_rx_mode,
 	.ndo_do_ioctl = dpaa_ioctl,
+	.ndo_setup_tc = dpaa_setup_tc,
 };
 
 static int dpaa_napi_add(struct net_device *net_dev)
@@ -2624,8 +2728,10 @@ static int dpaa_eth_probe(struct platform_device *pdev)
 	priv->rx_headroom = dpaa_get_headroom(&priv->buf_layout[RX]);
 
 	/* All real interfaces need their ports initialized */
-	dpaa_eth_init_ports(mac_dev, dpaa_bps, DPAA_BPS_NUM, &port_fqs,
-			    &priv->buf_layout[0], dev);
+	err = dpaa_eth_init_ports(mac_dev, dpaa_bps, DPAA_BPS_NUM, &port_fqs,
+				  &priv->buf_layout[0], dev);
+	if (err)
+		goto init_ports_failed;
 
 	priv->percpu_priv = devm_alloc_percpu(dev, *priv->percpu_priv);
 	if (!priv->percpu_priv) {
@@ -2638,6 +2744,9 @@ static int dpaa_eth_probe(struct platform_device *pdev)
 		memset(percpu_priv, 0, sizeof(*percpu_priv));
 	}
 
+	priv->num_tc = 1;
+	netif_set_real_num_tx_queues(net_dev, priv->num_tc * DPAA_TC_TXQ_NUM);
+
 	/* Initialize NAPI */
 	err = dpaa_napi_add(net_dev);
 	if (err < 0)
@@ -2658,6 +2767,7 @@ netdev_init_failed:
 napi_add_failed:
 	dpaa_napi_del(net_dev);
 alloc_percpu_failed:
+init_ports_failed:
 	dpaa_fq_free(dev, &priv->dpaa_fq_list);
 fq_alloc_failed:
 	qman_delete_cgr_safe(&priv->ingress_cgr);
diff --git a/drivers/net/ethernet/freescale/dpaa/dpaa_eth.h b/drivers/net/ethernet/freescale/dpaa/dpaa_eth.h
index 1f9aebf3f3c5..9941a7866ebe 100644
--- a/drivers/net/ethernet/freescale/dpaa/dpaa_eth.h
+++ b/drivers/net/ethernet/freescale/dpaa/dpaa_eth.h
@@ -39,7 +39,12 @@
 #include "mac.h"
 #include "dpaa_eth_trace.h"
 
-#define DPAA_ETH_TXQ_NUM	NR_CPUS
+/* Number of prioritised traffic classes */
+#define DPAA_TC_NUM		4
+/* Number of Tx queues per traffic class */
+#define DPAA_TC_TXQ_NUM		NR_CPUS
+/* Total number of Tx queues */
+#define DPAA_ETH_TXQ_NUM	(DPAA_TC_NUM * DPAA_TC_TXQ_NUM)
 
 #define DPAA_BPS_NUM 3 /* number of bpools per interface */
 
@@ -152,6 +157,7 @@ struct dpaa_priv {
 	u16 channel;
 	struct list_head dpaa_fq_list;
 
+	u8 num_tc;
 	u32 msg_enable;	/* net_device message level */
 
 	struct {
diff --git a/drivers/net/ethernet/freescale/fman/fman.c b/drivers/net/ethernet/freescale/fman/fman.c
index f60845f0c6ca..4aefe2438969 100644
--- a/drivers/net/ethernet/freescale/fman/fman.c
+++ b/drivers/net/ethernet/freescale/fman/fman.c
@@ -59,6 +59,7 @@
 #define DMA_OFFSET		0x000C2000
 #define FPM_OFFSET		0x000C3000
 #define IMEM_OFFSET		0x000C4000
+#define HWP_OFFSET		0x000C7000
 #define CGP_OFFSET		0x000DB000
 
 /* Exceptions bit map */
@@ -218,6 +219,9 @@
 
 #define QMI_GS_HALT_NOT_BUSY		0x00000002
 
+/* HWP defines */
+#define HWP_RPIMAC_PEN			0x00000001
+
 /* IRAM defines */
 #define IRAM_IADD_AIE			0x80000000
 #define IRAM_READY			0x80000000
@@ -475,6 +479,12 @@ struct fman_dma_regs {
 	u32 res00e0[0x400 - 56];
 };
 
+struct fman_hwp_regs {
+	u32 res0000[0x844 / 4];		/* 0x000..0x843 */
+	u32 fmprrpimac;	/* FM Parser Internal memory access control */
+	u32 res[(0x1000 - 0x848) / 4];	/* 0x848..0xFFF */
+};
+
 /* Structure that holds current FMan state.
  * Used for saving run time information.
  */
@@ -606,6 +616,7 @@ struct fman {
 	struct fman_bmi_regs __iomem *bmi_regs;
 	struct fman_qmi_regs __iomem *qmi_regs;
 	struct fman_dma_regs __iomem *dma_regs;
+	struct fman_hwp_regs __iomem *hwp_regs;
 	fman_exceptions_cb *exception_cb;
 	fman_bus_error_cb *bus_error_cb;
 	/* Spinlock for FMan use */
@@ -999,6 +1010,12 @@ static void qmi_init(struct fman_qmi_regs __iomem *qmi_rg,
 	iowrite32be(tmp_reg, &qmi_rg->fmqm_ien);
 }
 
+static void hwp_init(struct fman_hwp_regs __iomem *hwp_rg)
+{
+	/* enable HW Parser */
+	iowrite32be(HWP_RPIMAC_PEN, &hwp_rg->fmprrpimac);
+}
+
 static int enable(struct fman *fman, struct fman_cfg *cfg)
 {
 	u32 cfg_reg = 0;
@@ -1195,7 +1212,7 @@ static int fill_soc_specific_params(struct fman_state_struct *state)
 		state->max_num_of_open_dmas	= 32;
 		state->fm_port_num_of_cg	= 256;
 		state->num_of_rx_ports	= 6;
-		state->total_fifo_size	= 122 * 1024;
+		state->total_fifo_size	= 136 * 1024;
 		break;
 
 	case 2:
@@ -1793,6 +1810,7 @@ static int fman_config(struct fman *fman)
 	fman->bmi_regs = base_addr + BMI_OFFSET;
 	fman->qmi_regs = base_addr + QMI_OFFSET;
 	fman->dma_regs = base_addr + DMA_OFFSET;
+	fman->hwp_regs = base_addr + HWP_OFFSET;
 	fman->base_addr = base_addr;
 
 	spin_lock_init(&fman->spinlock);
@@ -2062,6 +2080,9 @@ static int fman_init(struct fman *fman)
 	/* Init QMI Registers */
 	qmi_init(fman->qmi_regs, fman->cfg);
 
+	/* Init HW Parser */
+	hwp_init(fman->hwp_regs);
+
 	err = enable(fman, cfg);
 	if (err != 0)
 		return err;
diff --git a/drivers/net/ethernet/freescale/fman/fman.h b/drivers/net/ethernet/freescale/fman/fman.h
index 57aae8d17d77..f53e1473dbcc 100644
--- a/drivers/net/ethernet/freescale/fman/fman.h
+++ b/drivers/net/ethernet/freescale/fman/fman.h
@@ -134,14 +134,14 @@ enum fman_exceptions {
 struct fman_prs_result {
 	u8 lpid;		/* Logical port id */
 	u8 shimr;		/* Shim header result  */
-	u16 l2r;		/* Layer 2 result */
-	u16 l3r;		/* Layer 3 result */
+	__be16 l2r;		/* Layer 2 result */
+	__be16 l3r;		/* Layer 3 result */
 	u8 l4r;		/* Layer 4 result */
 	u8 cplan;		/* Classification plan id */
-	u16 nxthdr;		/* Next Header  */
-	u16 cksum;		/* Running-sum */
+	__be16 nxthdr;		/* Next Header  */
+	__be16 cksum;		/* Running-sum */
 	/* Flags&fragment-offset field of the last IP-header */
-	u16 flags_frag_off;
+	__be16 flags_frag_off;
 	/* Routing type field of a IPV6 routing extension header */
 	u8 route_type;
 	/* Routing Extension Header Present; last bit is IP valid */
diff --git a/drivers/net/ethernet/freescale/fman/fman_port.c b/drivers/net/ethernet/freescale/fman/fman_port.c
index 9f3bb50a2365..57bf44fa16a1 100644
--- a/drivers/net/ethernet/freescale/fman/fman_port.c
+++ b/drivers/net/ethernet/freescale/fman/fman_port.c
@@ -62,6 +62,7 @@
 
 #define BMI_PORT_REGS_OFFSET				0
 #define QMI_PORT_REGS_OFFSET				0x400
+#define HWP_PORT_REGS_OFFSET				0x800
 
 /* Default values */
 #define DFLT_PORT_BUFFER_PREFIX_CONTEXT_DATA_ALIGN		\
@@ -182,7 +183,7 @@
 #define NIA_ENG_BMI					0x00500000
 #define NIA_ENG_QMI_ENQ					0x00540000
 #define NIA_ENG_QMI_DEQ					0x00580000
-
+#define NIA_ENG_HWP					0x00440000
 #define NIA_BMI_AC_ENQ_FRAME				0x00000002
 #define NIA_BMI_AC_TX_RELEASE				0x000002C0
 #define NIA_BMI_AC_RELEASE				0x000000C0
@@ -317,6 +318,19 @@ struct fman_port_qmi_regs {
 	u32 fmqm_pndcc;		/* PortID n Dequeue Confirm Counter */
 };
 
+#define HWP_HXS_COUNT 16
+#define HWP_HXS_PHE_REPORT 0x00000800
+#define HWP_HXS_PCAC_PSTAT 0x00000100
+#define HWP_HXS_PCAC_PSTOP 0x00000001
+struct fman_port_hwp_regs {
+	struct {
+		u32 ssa; /* Soft Sequence Attachment */
+		u32 lcv; /* Line-up Enable Confirmation Mask */
+	} pmda[HWP_HXS_COUNT]; /* Parse Memory Direct Access Registers */
+	u32 reserved080[(0x3f8 - 0x080) / 4]; /* (0x080-0x3f7) */
+	u32 fmpr_pcac; /* Configuration Access Control */
+};
+
 /* QMI dequeue prefetch modes */
 enum fman_port_deq_prefetch {
 	FMAN_PORT_DEQ_NO_PREFETCH, /* No prefetch mode */
@@ -436,6 +450,7 @@ struct fman_port {
 
 	union fman_port_bmi_regs __iomem *bmi_regs;
 	struct fman_port_qmi_regs __iomem *qmi_regs;
+	struct fman_port_hwp_regs __iomem *hwp_regs;
 
 	struct fman_sp_buffer_offsets buffer_offsets;
 
@@ -521,9 +536,12 @@ static int init_bmi_rx(struct fman_port *port)
 	/* NIA */
 	tmp = (u32)cfg->rx_fd_bits << BMI_NEXT_ENG_FD_BITS_SHIFT;
 
-	tmp |= NIA_ENG_BMI | NIA_BMI_AC_ENQ_FRAME;
+	tmp |= NIA_ENG_HWP;
 	iowrite32be(tmp, &regs->fmbm_rfne);
 
+	/* Parser Next Engine NIA */
+	iowrite32be(NIA_ENG_BMI | NIA_BMI_AC_ENQ_FRAME, &regs->fmbm_rfpne);
+
 	/* Enqueue NIA */
 	iowrite32be(NIA_ENG_QMI_ENQ | NIA_ORDER_RESTOR, &regs->fmbm_rfene);
 
@@ -665,6 +683,50 @@ static int init_qmi(struct fman_port *port)
 	return 0;
 }
 
+static void stop_port_hwp(struct fman_port *port)
+{
+	struct fman_port_hwp_regs __iomem *regs = port->hwp_regs;
+	int cnt = 100;
+
+	iowrite32be(HWP_HXS_PCAC_PSTOP, &regs->fmpr_pcac);
+
+	while (cnt-- > 0 &&
+	       (ioread32be(&regs->fmpr_pcac) & HWP_HXS_PCAC_PSTAT))
+		udelay(10);
+	if (!cnt)
+		pr_err("Timeout stopping HW Parser\n");
+}
+
+static void start_port_hwp(struct fman_port *port)
+{
+	struct fman_port_hwp_regs __iomem *regs = port->hwp_regs;
+	int cnt = 100;
+
+	iowrite32be(0, &regs->fmpr_pcac);
+
+	while (cnt-- > 0 &&
+	       !(ioread32be(&regs->fmpr_pcac) & HWP_HXS_PCAC_PSTAT))
+		udelay(10);
+	if (!cnt)
+		pr_err("Timeout starting HW Parser\n");
+}
+
+static void init_hwp(struct fman_port *port)
+{
+	struct fman_port_hwp_regs __iomem *regs = port->hwp_regs;
+	int i;
+
+	stop_port_hwp(port);
+
+	for (i = 0; i < HWP_HXS_COUNT; i++) {
+		/* enable HXS error reporting into FD[STATUS] PHE */
+		iowrite32be(0x00000000, &regs->pmda[i].ssa);
+		iowrite32be(0xffffffff, &regs->pmda[i].lcv);
+	}
+
+	start_port_hwp(port);
+}
+
 static int init(struct fman_port *port)
 {
 	int err;
@@ -673,6 +735,8 @@ static int init(struct fman_port *port)
 	switch (port->port_type) {
 	case FMAN_PORT_TYPE_RX:
 		err = init_bmi_rx(port);
+		if (!err)
+			init_hwp(port);
 		break;
 	case FMAN_PORT_TYPE_TX:
 		err = init_bmi_tx(port);
@@ -686,7 +750,8 @@ static int init(struct fman_port *port)
 
 	/* Init QMI registers */
 	err = init_qmi(port);
-	return err;
+	if (err)
+		return err;
 
 	return 0;
 }
@@ -1247,7 +1312,7 @@ int fman_port_config(struct fman_port *port, struct fman_port_params *params)
 	/* Allocate the FM driver's parameters structure */
 	port->cfg = kzalloc(sizeof(*port->cfg), GFP_KERNEL);
 	if (!port->cfg)
-		goto err_params;
+		return -EINVAL;
 
 	/* Initialize FM port parameters which will be kept by the driver */
 	port->port_type = port->dts_params.type;
@@ -1276,6 +1341,7 @@ int fman_port_config(struct fman_port *port, struct fman_port_params *params)
 	/* set memory map pointers */
 	port->bmi_regs = base_addr + BMI_PORT_REGS_OFFSET;
 	port->qmi_regs = base_addr + QMI_PORT_REGS_OFFSET;
+	port->hwp_regs = base_addr + HWP_PORT_REGS_OFFSET;
 
 	port->max_frame_length = DFLT_PORT_MAX_FRAME_LENGTH;
 	/* resource distribution. */
@@ -1327,8 +1393,6 @@ int fman_port_config(struct fman_port *port, struct fman_port_params *params)
 
 err_port_cfg:
 	kfree(port->cfg);
-err_params:
-	kfree(port);
 	return -EINVAL;
 }
 EXPORT_SYMBOL(fman_port_config);