[PATCH 0/7] Miscellaneous fixes for mlx4 and mlx5

[PATCH 0/7] Miscellaneous fixes for mlx4 and mlx5

[Nelio Laranjeiro]

Various minor fixes for mlx4 (ConnectX-3) and mlx5 (ConnectX-4).

Adrien Mazarguil (4):
  mlx: ensure MTU update is effective
  mlx: retrieve mbuf size through proper API
  mlx5: fix RX VLAN stripping capability check
  mlx5: cosmetic changes (coding style)

Nelio Laranjeiro (3):
  mlx: remove unused memory region property
  mlx5: enhance SR-IOV detection
  mlx5: update documentation part related to features and limitations

 doc/guides/nics/mlx5.rst       |  2 +-
 drivers/net/mlx4/mlx4.c        | 43 ++++++++++++++++++++------------------
 drivers/net/mlx5/mlx5.c        | 14 +++++++------
 drivers/net/mlx5/mlx5.h        |  3 ++-
 drivers/net/mlx5/mlx5_ethdev.c | 47 ++++++++++++++++++++++++++++++++++++++++--
 drivers/net/mlx5/mlx5_rxq.c    | 23 +++++++--------------
 drivers/net/mlx5/mlx5_rxtx.c   |  7 +++----
 drivers/net/mlx5/mlx5_rxtx.h   | 16 ++++++--------
 drivers/net/mlx5/mlx5_txq.c    |  2 +-
 drivers/net/mlx5/mlx5_vlan.c   |  2 +-
 10 files changed, 98 insertions(+), 61 deletions(-)

-- 
2.1.4

[PATCH 1/7] mlx: remove unused memory region property

[Nelio Laranjeiro]

Memory regions are always local with raw Ethernet queues, drop the remote
property as it adds extra processing on the hardware side.

Signed-off-by: Nelio Laranjeiro <nelio.laranjeiro@6wind.com>
Signed-off-by: Adrien Mazarguil <adrien.mazarguil@6wind.com>
---
 drivers/net/mlx4/mlx4.c      | 4 ++--
 drivers/net/mlx5/mlx5_rxtx.c | 2 +-
 drivers/net/mlx5/mlx5_txq.c  | 2 +-
 3 files changed, 4 insertions(+), 4 deletions(-)

diff --git a/drivers/net/mlx4/mlx4.c b/drivers/net/mlx4/mlx4.c
index 9ed1491..661c49f 100644
--- a/drivers/net/mlx4/mlx4.c
+++ b/drivers/net/mlx4/mlx4.c
@@ -998,7 +998,7 @@ txq_alloc_elts(struct txq *txq, unsigned int elts_n)
 	}
 	mr_linear =
 		ibv_reg_mr(txq->priv->pd, elts_linear, sizeof(*elts_linear),
-			   (IBV_ACCESS_LOCAL_WRITE | IBV_ACCESS_REMOTE_WRITE));
+			   IBV_ACCESS_LOCAL_WRITE);
 	if (mr_linear == NULL) {
 		ERROR("%p: unable to configure MR, ibv_reg_mr() failed",
 		      (void *)txq);
@@ -1310,7 +1310,7 @@ mlx4_mp2mr(struct ibv_pd *pd, struct rte_mempool *mp)
 	return ibv_reg_mr(pd,
 			  (void *)start,
 			  end - start,
-			  IBV_ACCESS_LOCAL_WRITE | IBV_ACCESS_REMOTE_WRITE);
+			  IBV_ACCESS_LOCAL_WRITE);
 }
 
 /**
diff --git a/drivers/net/mlx5/mlx5_rxtx.c b/drivers/net/mlx5/mlx5_rxtx.c
index 29bfcec..7f02641 100644
--- a/drivers/net/mlx5/mlx5_rxtx.c
+++ b/drivers/net/mlx5/mlx5_rxtx.c
@@ -252,7 +252,7 @@ mlx5_mp2mr(struct ibv_pd *pd, struct rte_mempool *mp)
 	return ibv_reg_mr(pd,
 			  (void *)start,
 			  end - start,
-			  IBV_ACCESS_LOCAL_WRITE | IBV_ACCESS_REMOTE_WRITE);
+			  IBV_ACCESS_LOCAL_WRITE);
 }
 
 /**
diff --git a/drivers/net/mlx5/mlx5_txq.c b/drivers/net/mlx5/mlx5_txq.c
index 31ce53a..e20df21 100644
--- a/drivers/net/mlx5/mlx5_txq.c
+++ b/drivers/net/mlx5/mlx5_txq.c
@@ -95,7 +95,7 @@ txq_alloc_elts(struct txq *txq, unsigned int elts_n)
 	}
 	mr_linear =
 		ibv_reg_mr(txq->priv->pd, elts_linear, sizeof(*elts_linear),
-			   (IBV_ACCESS_LOCAL_WRITE | IBV_ACCESS_REMOTE_WRITE));
+			   IBV_ACCESS_LOCAL_WRITE);
 	if (mr_linear == NULL) {
 		ERROR("%p: unable to configure MR, ibv_reg_mr() failed",
 		      (void *)txq);
-- 
2.1.4

[PATCH 2/7] mlx: ensure MTU update is effective

[Nelio Laranjeiro]

From: Adrien Mazarguil <adrien.mazarguil@6wind.com>

There is no guarantee that the new MTU is effective after writing its value
to sysfs. Retrieve it to be sure.

Signed-off-by: Adrien Mazarguil <adrien.mazarguil@6wind.com>
---
 drivers/net/mlx4/mlx4.c        | 10 +++++++++-
 drivers/net/mlx5/mlx5_ethdev.c | 10 +++++++++-
 2 files changed, 18 insertions(+), 2 deletions(-)

diff --git a/drivers/net/mlx4/mlx4.c b/drivers/net/mlx4/mlx4.c
index 661c49f..6174e4b 100644
--- a/drivers/net/mlx4/mlx4.c
+++ b/drivers/net/mlx4/mlx4.c
@@ -659,7 +659,15 @@ priv_get_mtu(struct priv *priv, uint16_t *mtu)
 static int
 priv_set_mtu(struct priv *priv, uint16_t mtu)
 {
-	return priv_set_sysfs_ulong(priv, "mtu", mtu);
+	uint16_t new_mtu;
+
+	if (priv_set_sysfs_ulong(priv, "mtu", mtu) ||
+	    priv_get_mtu(priv, &new_mtu))
+		return -1;
+	if (new_mtu == mtu)
+		return 0;
+	errno = EINVAL;
+	return -1;
 }
 
 /**
diff --git a/drivers/net/mlx5/mlx5_ethdev.c b/drivers/net/mlx5/mlx5_ethdev.c
index 36b369e..25926cb 100644
--- a/drivers/net/mlx5/mlx5_ethdev.c
+++ b/drivers/net/mlx5/mlx5_ethdev.c
@@ -398,7 +398,15 @@ priv_get_mtu(struct priv *priv, uint16_t *mtu)
 static int
 priv_set_mtu(struct priv *priv, uint16_t mtu)
 {
-	return priv_set_sysfs_ulong(priv, "mtu", mtu);
+	uint16_t new_mtu;
+
+	if (priv_set_sysfs_ulong(priv, "mtu", mtu) ||
+	    priv_get_mtu(priv, &new_mtu))
+		return -1;
+	if (new_mtu == mtu)
+		return 0;
+	errno = EINVAL;
+	return -1;
 }
 
 /**
-- 
2.1.4

[PATCH 3/7] mlx: retrieve mbuf size through proper API

[Nelio Laranjeiro]

From: Adrien Mazarguil <adrien.mazarguil@6wind.com>

No need to allocate a mbuf for that.

Signed-off-by: Adrien Mazarguil <adrien.mazarguil@6wind.com>
---
 drivers/net/mlx4/mlx4.c        | 29 ++++++++++++-----------------
 drivers/net/mlx5/mlx5_ethdev.c |  5 ++++-
 drivers/net/mlx5/mlx5_rxq.c    | 20 ++++++--------------
 drivers/net/mlx5/mlx5_rxtx.c   |  1 -
 drivers/net/mlx5/mlx5_rxtx.h   |  1 -
 5 files changed, 22 insertions(+), 34 deletions(-)

diff --git a/drivers/net/mlx4/mlx4.c b/drivers/net/mlx4/mlx4.c
index 6174e4b..82b1c63 100644
--- a/drivers/net/mlx4/mlx4.c
+++ b/drivers/net/mlx4/mlx4.c
@@ -197,7 +197,6 @@ struct rxq {
 	unsigned int sp:1; /* Use scattered RX elements. */
 	unsigned int csum:1; /* Enable checksum offloading. */
 	unsigned int csum_l2tun:1; /* Same for L2 tunnels. */
-	uint32_t mb_len; /* Length of a mp-issued mbuf. */
 	struct mlx4_rxq_stats stats; /* RX queue counters. */
 	unsigned int socket; /* CPU socket ID for allocations. */
 	struct ibv_exp_res_domain *rd; /* Resource Domain. */
@@ -3160,7 +3159,6 @@ mlx4_rx_burst_sp(void *dpdk_rxq, struct rte_mbuf **pkts, uint16_t pkts_n)
 			rep->ol_flags = -1;
 #endif
 			assert(rep->buf_len == seg->buf_len);
-			assert(rep->buf_len == rxq->mb_len);
 			/* Reconfigure sge to use rep instead of seg. */
 			assert(sge->lkey == rxq->mr->lkey);
 			sge->addr = ((uintptr_t)rep->buf_addr + seg_headroom);
@@ -3581,6 +3579,7 @@ rxq_rehash(struct rte_eth_dev *dev, struct rxq *rxq)
 	unsigned int i, k;
 	struct ibv_exp_qp_attr mod;
 	struct ibv_recv_wr *bad_wr;
+	unsigned int mb_len;
 	int err;
 	int parent = (rxq == &priv->rxq_parent);
 
@@ -3589,6 +3588,7 @@ rxq_rehash(struct rte_eth_dev *dev, struct rxq *rxq)
 		      (void *)dev, (void *)rxq);
 		return EINVAL;
 	}
+	mb_len = rte_pktmbuf_data_room_size(rxq->mp);
 	DEBUG("%p: rehashing queue %p", (void *)dev, (void *)rxq);
 	/* Number of descriptors and mbufs currently allocated. */
 	desc_n = (tmpl.elts_n * (tmpl.sp ? MLX4_PMD_SGE_WR_N : 1));
@@ -3603,9 +3603,10 @@ rxq_rehash(struct rte_eth_dev *dev, struct rxq *rxq)
 		rxq->csum_l2tun = tmpl.csum_l2tun;
 	}
 	/* Enable scattered packets support for this queue if necessary. */
+	assert(mb_len >= RTE_PKTMBUF_HEADROOM);
 	if ((dev->data->dev_conf.rxmode.jumbo_frame) &&
 	    (dev->data->dev_conf.rxmode.max_rx_pkt_len >
-	     (tmpl.mb_len - RTE_PKTMBUF_HEADROOM))) {
+	     (mb_len - RTE_PKTMBUF_HEADROOM))) {
 		tmpl.sp = 1;
 		desc_n /= MLX4_PMD_SGE_WR_N;
 	} else
@@ -3796,7 +3797,7 @@ rxq_setup(struct rte_eth_dev *dev, struct rxq *rxq, uint16_t desc,
 	} attr;
 	enum ibv_exp_query_intf_status status;
 	struct ibv_recv_wr *bad_wr;
-	struct rte_mbuf *buf;
+	unsigned int mb_len;
 	int ret = 0;
 	int parent = (rxq == &priv->rxq_parent);
 
@@ -3812,31 +3813,22 @@ rxq_setup(struct rte_eth_dev *dev, struct rxq *rxq, uint16_t desc,
 		desc = 1;
 		goto skip_mr;
 	}
+	mb_len = rte_pktmbuf_data_room_size(mp);
 	if ((desc == 0) || (desc % MLX4_PMD_SGE_WR_N)) {
 		ERROR("%p: invalid number of RX descriptors (must be a"
 		      " multiple of %d)", (void *)dev, MLX4_PMD_SGE_WR_N);
 		return EINVAL;
 	}
-	/* Get mbuf length. */
-	buf = rte_pktmbuf_alloc(mp);
-	if (buf == NULL) {
-		ERROR("%p: unable to allocate mbuf", (void *)dev);
-		return ENOMEM;
-	}
-	tmpl.mb_len = buf->buf_len;
-	assert((rte_pktmbuf_headroom(buf) +
-		rte_pktmbuf_tailroom(buf)) == tmpl.mb_len);
-	assert(rte_pktmbuf_headroom(buf) == RTE_PKTMBUF_HEADROOM);
-	rte_pktmbuf_free(buf);
 	/* Toggle RX checksum offload if hardware supports it. */
 	if (priv->hw_csum)
 		tmpl.csum = !!dev->data->dev_conf.rxmode.hw_ip_checksum;
 	if (priv->hw_csum_l2tun)
 		tmpl.csum_l2tun = !!dev->data->dev_conf.rxmode.hw_ip_checksum;
 	/* Enable scattered packets support for this queue if necessary. */
+	assert(mb_len >= RTE_PKTMBUF_HEADROOM);
 	if ((dev->data->dev_conf.rxmode.jumbo_frame) &&
 	    (dev->data->dev_conf.rxmode.max_rx_pkt_len >
-	     (tmpl.mb_len - RTE_PKTMBUF_HEADROOM))) {
+	     (mb_len - RTE_PKTMBUF_HEADROOM))) {
 		tmpl.sp = 1;
 		desc /= MLX4_PMD_SGE_WR_N;
 	}
@@ -4873,6 +4865,7 @@ mlx4_dev_set_mtu(struct rte_eth_dev *dev, uint16_t mtu)
 	/* Reconfigure each RX queue. */
 	for (i = 0; (i != priv->rxqs_n); ++i) {
 		struct rxq *rxq = (*priv->rxqs)[i];
+		unsigned int mb_len;
 		unsigned int max_frame_len;
 		int sp;
 
@@ -4882,7 +4875,9 @@ mlx4_dev_set_mtu(struct rte_eth_dev *dev, uint16_t mtu)
 		 * toggle scattered support (sp) if necessary. */
 		max_frame_len = (priv->mtu + ETHER_HDR_LEN +
 				 (ETHER_MAX_VLAN_FRAME_LEN - ETHER_MAX_LEN));
-		sp = (max_frame_len > (rxq->mb_len - RTE_PKTMBUF_HEADROOM));
+		mb_len = rte_pktmbuf_data_room_size(rxq->mp);
+		assert(mb_len >= RTE_PKTMBUF_HEADROOM);
+		sp = (max_frame_len > (mb_len - RTE_PKTMBUF_HEADROOM));
 		/* Provide new values to rxq_setup(). */
 		dev->data->dev_conf.rxmode.jumbo_frame = sp;
 		dev->data->dev_conf.rxmode.max_rx_pkt_len = max_frame_len;
diff --git a/drivers/net/mlx5/mlx5_ethdev.c b/drivers/net/mlx5/mlx5_ethdev.c
index 25926cb..ecbb49b 100644
--- a/drivers/net/mlx5/mlx5_ethdev.c
+++ b/drivers/net/mlx5/mlx5_ethdev.c
@@ -718,6 +718,7 @@ mlx5_dev_set_mtu(struct rte_eth_dev *dev, uint16_t mtu)
 	/* Reconfigure each RX queue. */
 	for (i = 0; (i != priv->rxqs_n); ++i) {
 		struct rxq *rxq = (*priv->rxqs)[i];
+		unsigned int mb_len;
 		unsigned int max_frame_len;
 		int sp;
 
@@ -727,7 +728,9 @@ mlx5_dev_set_mtu(struct rte_eth_dev *dev, uint16_t mtu)
 		 * toggle scattered support (sp) if necessary. */
 		max_frame_len = (priv->mtu + ETHER_HDR_LEN +
 				 (ETHER_MAX_VLAN_FRAME_LEN - ETHER_MAX_LEN));
-		sp = (max_frame_len > (rxq->mb_len - RTE_PKTMBUF_HEADROOM));
+		mb_len = rte_pktmbuf_data_room_size(rxq->mp);
+		assert(mb_len >= RTE_PKTMBUF_HEADROOM);
+		sp = (max_frame_len > (mb_len - RTE_PKTMBUF_HEADROOM));
 		/* Provide new values to rxq_setup(). */
 		dev->data->dev_conf.rxmode.jumbo_frame = sp;
 		dev->data->dev_conf.rxmode.max_rx_pkt_len = max_frame_len;
diff --git a/drivers/net/mlx5/mlx5_rxq.c b/drivers/net/mlx5/mlx5_rxq.c
index cbb017b..469ba98 100644
--- a/drivers/net/mlx5/mlx5_rxq.c
+++ b/drivers/net/mlx5/mlx5_rxq.c
@@ -973,6 +973,7 @@ rxq_rehash(struct rte_eth_dev *dev, struct rxq *rxq)
 	struct rte_mbuf **pool;
 	unsigned int i, k;
 	struct ibv_exp_wq_attr mod;
+	unsigned int mb_len = rte_pktmbuf_data_room_size(rxq->mp);
 	int err;
 
 	DEBUG("%p: rehashing queue %p", (void *)dev, (void *)rxq);
@@ -989,9 +990,10 @@ rxq_rehash(struct rte_eth_dev *dev, struct rxq *rxq)
 		rxq->csum_l2tun = tmpl.csum_l2tun;
 	}
 	/* Enable scattered packets support for this queue if necessary. */
+	assert(mb_len >= RTE_PKTMBUF_HEADROOM);
 	if ((dev->data->dev_conf.rxmode.jumbo_frame) &&
 	    (dev->data->dev_conf.rxmode.max_rx_pkt_len >
-	     (tmpl.mb_len - RTE_PKTMBUF_HEADROOM))) {
+	     (mb_len - RTE_PKTMBUF_HEADROOM))) {
 		tmpl.sp = 1;
 		desc_n /= MLX5_PMD_SGE_WR_N;
 	} else
@@ -1156,7 +1158,7 @@ rxq_setup(struct rte_eth_dev *dev, struct rxq *rxq, uint16_t desc,
 		struct ibv_exp_wq_init_attr wq;
 	} attr;
 	enum ibv_exp_query_intf_status status;
-	struct rte_mbuf *buf;
+	unsigned int mb_len = rte_pktmbuf_data_room_size(mp);
 	int ret = 0;
 	unsigned int i;
 	unsigned int cq_size = desc;
@@ -1167,26 +1169,16 @@ rxq_setup(struct rte_eth_dev *dev, struct rxq *rxq, uint16_t desc,
 		      " multiple of %d)", (void *)dev, MLX5_PMD_SGE_WR_N);
 		return EINVAL;
 	}
-	/* Get mbuf length. */
-	buf = rte_pktmbuf_alloc(mp);
-	if (buf == NULL) {
-		ERROR("%p: unable to allocate mbuf", (void *)dev);
-		return ENOMEM;
-	}
-	tmpl.mb_len = buf->buf_len;
-	assert((rte_pktmbuf_headroom(buf) +
-		rte_pktmbuf_tailroom(buf)) == tmpl.mb_len);
-	assert(rte_pktmbuf_headroom(buf) == RTE_PKTMBUF_HEADROOM);
-	rte_pktmbuf_free(buf);
 	/* Toggle RX checksum offload if hardware supports it. */
 	if (priv->hw_csum)
 		tmpl.csum = !!dev->data->dev_conf.rxmode.hw_ip_checksum;
 	if (priv->hw_csum_l2tun)
 		tmpl.csum_l2tun = !!dev->data->dev_conf.rxmode.hw_ip_checksum;
 	/* Enable scattered packets support for this queue if necessary. */
+	assert(mb_len >= RTE_PKTMBUF_HEADROOM);
 	if ((dev->data->dev_conf.rxmode.jumbo_frame) &&
 	    (dev->data->dev_conf.rxmode.max_rx_pkt_len >
-	     (tmpl.mb_len - RTE_PKTMBUF_HEADROOM))) {
+	     (mb_len - RTE_PKTMBUF_HEADROOM))) {
 		tmpl.sp = 1;
 		desc /= MLX5_PMD_SGE_WR_N;
 	}
diff --git a/drivers/net/mlx5/mlx5_rxtx.c b/drivers/net/mlx5/mlx5_rxtx.c
index 7f02641..3881c30 100644
--- a/drivers/net/mlx5/mlx5_rxtx.c
+++ b/drivers/net/mlx5/mlx5_rxtx.c
@@ -1005,7 +1005,6 @@ mlx5_rx_burst_sp(void *dpdk_rxq, struct rte_mbuf **pkts, uint16_t pkts_n)
 			rep->ol_flags = -1;
 #endif
 			assert(rep->buf_len == seg->buf_len);
-			assert(rep->buf_len == rxq->mb_len);
 			/* Reconfigure sge to use rep instead of seg. */
 			assert(sge->lkey == rxq->mr->lkey);
 			sge->addr = ((uintptr_t)rep->buf_addr + seg_headroom);
diff --git a/drivers/net/mlx5/mlx5_rxtx.h b/drivers/net/mlx5/mlx5_rxtx.h
index d522f70..dd3003c 100644
--- a/drivers/net/mlx5/mlx5_rxtx.h
+++ b/drivers/net/mlx5/mlx5_rxtx.h
@@ -121,7 +121,6 @@ struct rxq {
 		struct rxq_elt_sp (*sp)[]; /* Scattered RX elements. */
 		struct rxq_elt (*no_sp)[]; /* RX elements. */
 	} elts;
-	uint32_t mb_len; /* Length of a mp-issued mbuf. */
 	unsigned int socket; /* CPU socket ID for allocations. */
 	struct mlx5_rxq_stats stats; /* RX queue counters. */
 	struct ibv_exp_res_domain *rd; /* Resource Domain. */
-- 
2.1.4

[PATCH 4/7] mlx5: fix RX VLAN stripping capability check

[Nelio Laranjeiro]

From: Adrien Mazarguil <adrien.mazarguil@6wind.com>

A hardware capability check is missing before enabling RX VLAN stripping
during queue setup.

Also, while dev_conf.rxmode.hw_vlan_strip is currently a single bit that
can be stored in priv->hw_vlan_strip directly, it should be interpreted as
a boolean value for safety.

Fixes: f3db9489188a ("mlx5: support Rx VLAN stripping")

Signed-off-by: Adrien Mazarguil <adrien.mazarguil@6wind.com>
---
 drivers/net/mlx5/mlx5_rxq.c  | 3 ++-
 drivers/net/mlx5/mlx5_vlan.c | 2 +-
 2 files changed, 3 insertions(+), 2 deletions(-)

diff --git a/drivers/net/mlx5/mlx5_rxq.c b/drivers/net/mlx5/mlx5_rxq.c
index 469ba98..0bcf55b 100644
--- a/drivers/net/mlx5/mlx5_rxq.c
+++ b/drivers/net/mlx5/mlx5_rxq.c
@@ -1222,7 +1222,8 @@ rxq_setup(struct rte_eth_dev *dev, struct rxq *rxq, uint16_t desc,
 	DEBUG("priv->device_attr.max_sge is %d",
 	      priv->device_attr.max_sge);
 	/* Configure VLAN stripping. */
-	tmpl.vlan_strip = dev->data->dev_conf.rxmode.hw_vlan_strip;
+	tmpl.vlan_strip = (priv->hw_vlan_strip &&
+			   !!dev->data->dev_conf.rxmode.hw_vlan_strip);
 	attr.wq = (struct ibv_exp_wq_init_attr){
 		.wq_context = NULL, /* Could be useful in the future. */
 		.wq_type = IBV_EXP_WQT_RQ,
diff --git a/drivers/net/mlx5/mlx5_vlan.c b/drivers/net/mlx5/mlx5_vlan.c
index ea7af1e..ff40538 100644
--- a/drivers/net/mlx5/mlx5_vlan.c
+++ b/drivers/net/mlx5/mlx5_vlan.c
@@ -218,7 +218,7 @@ mlx5_vlan_offload_set(struct rte_eth_dev *dev, int mask)
 	unsigned int i;
 
 	if (mask & ETH_VLAN_STRIP_MASK) {
-		int hw_vlan_strip = dev->data->dev_conf.rxmode.hw_vlan_strip;
+		int hw_vlan_strip = !!dev->data->dev_conf.rxmode.hw_vlan_strip;
 
 		if (!priv->hw_vlan_strip) {
 			ERROR("VLAN stripping is not supported");
-- 
2.1.4

[PATCH 5/7] mlx5: cosmetic changes (coding style)

[Nelio Laranjeiro]

From: Adrien Mazarguil <adrien.mazarguil@6wind.com>

Add consistency to mlx5_rxtx.h.

Signed-off-by: Adrien Mazarguil <adrien.mazarguil@6wind.com>
---
 drivers/net/mlx5/mlx5_rxtx.h | 15 ++++++---------
 1 file changed, 6 insertions(+), 9 deletions(-)

diff --git a/drivers/net/mlx5/mlx5_rxtx.h b/drivers/net/mlx5/mlx5_rxtx.h
index dd3003c..47f6299 100644
--- a/drivers/net/mlx5/mlx5_rxtx.h
+++ b/drivers/net/mlx5/mlx5_rxtx.h
@@ -231,7 +231,8 @@ struct hash_rxq {
 	struct ibv_qp *qp; /* Hash RX QP. */
 	enum hash_rxq_type type; /* Hash RX queue type. */
 	/* MAC flow steering rules, one per VLAN ID. */
-	struct ibv_exp_flow *mac_flow[MLX5_MAX_MAC_ADDRESSES][MLX5_MAX_VLAN_IDS];
+	struct ibv_exp_flow *mac_flow
+		[MLX5_MAX_MAC_ADDRESSES][MLX5_MAX_VLAN_IDS];
 	struct ibv_exp_flow *special_flow
 		[MLX5_MAX_SPECIAL_FLOWS][MLX5_MAX_VLAN_IDS];
 };
@@ -322,21 +323,17 @@ int rxq_setup(struct rte_eth_dev *, struct rxq *, uint16_t, unsigned int,
 int mlx5_rx_queue_setup(struct rte_eth_dev *, uint16_t, uint16_t, unsigned int,
 			const struct rte_eth_rxconf *, struct rte_mempool *);
 void mlx5_rx_queue_release(void *);
-uint16_t mlx5_rx_burst_secondary_setup(void *dpdk_rxq, struct rte_mbuf **pkts,
-			      uint16_t pkts_n);
-
+uint16_t mlx5_rx_burst_secondary_setup(void *, struct rte_mbuf **, uint16_t);
 
 /* mlx5_txq.c */
 
 void txq_cleanup(struct txq *);
-int txq_setup(struct rte_eth_dev *dev, struct txq *txq, uint16_t desc,
-	  unsigned int socket, const struct rte_eth_txconf *conf);
-
+int txq_setup(struct rte_eth_dev *, struct txq *, uint16_t, unsigned int,
+	      const struct rte_eth_txconf *);
 int mlx5_tx_queue_setup(struct rte_eth_dev *, uint16_t, uint16_t, unsigned int,
 			const struct rte_eth_txconf *);
 void mlx5_tx_queue_release(void *);
-uint16_t mlx5_tx_burst_secondary_setup(void *dpdk_txq, struct rte_mbuf **pkts,
-			      uint16_t pkts_n);
+uint16_t mlx5_tx_burst_secondary_setup(void *, struct rte_mbuf **, uint16_t);
 
 /* mlx5_rxtx.c */
 
-- 
2.1.4

[PATCH 6/7] mlx5: enhance SR-IOV detection

[Nelio Laranjeiro]

SR-IOV mode is currently set when dealing with VF devices. PF devices must
be taken into account as well if they have active VFs.

Signed-off-by: Nelio Laranjeiro <nelio.laranjeiro@6wind.com>
Signed-off-by: Adrien Mazarguil <adrien.mazarguil@6wind.com>
---
 drivers/net/mlx5/mlx5.c        | 14 ++++++++------
 drivers/net/mlx5/mlx5.h        |  3 ++-
 drivers/net/mlx5/mlx5_ethdev.c | 32 ++++++++++++++++++++++++++++++++
 drivers/net/mlx5/mlx5_rxtx.c   |  4 ++--
 4 files changed, 44 insertions(+), 9 deletions(-)

diff --git a/drivers/net/mlx5/mlx5.c b/drivers/net/mlx5/mlx5.c
index 041cfc3..67a541c 100644
--- a/drivers/net/mlx5/mlx5.c
+++ b/drivers/net/mlx5/mlx5.c
@@ -260,7 +260,7 @@ mlx5_pci_devinit(struct rte_pci_driver *pci_drv, struct rte_pci_device *pci_dev)
 	int err = 0;
 	struct ibv_context *attr_ctx = NULL;
 	struct ibv_device_attr device_attr;
-	unsigned int vf;
+	unsigned int sriov;
 	unsigned int mps;
 	int idx;
 	int i;
@@ -303,17 +303,17 @@ mlx5_pci_devinit(struct rte_pci_driver *pci_drv, struct rte_pci_device *pci_dev)
 		    (pci_dev->addr.devid != pci_addr.devid) ||
 		    (pci_dev->addr.function != pci_addr.function))
 			continue;
-		vf = ((pci_dev->id.device_id ==
+		sriov = ((pci_dev->id.device_id ==
 		       PCI_DEVICE_ID_MELLANOX_CONNECTX4VF) ||
 		      (pci_dev->id.device_id ==
 		       PCI_DEVICE_ID_MELLANOX_CONNECTX4LXVF));
 		/* Multi-packet send is only supported by ConnectX-4 Lx PF. */
 		mps = (pci_dev->id.device_id ==
 		       PCI_DEVICE_ID_MELLANOX_CONNECTX4LX);
-		INFO("PCI information matches, using device \"%s\" (VF: %s,"
-		     " MPS: %s)",
+		INFO("PCI information matches, using device \"%s\""
+		     " (SR-IOV: %s, MPS: %s)",
 		     list[i]->name,
-		     vf ? "true" : "false",
+		     sriov ? "true" : "false",
 		     mps ? "true" : "false");
 		attr_ctx = ibv_open_device(list[i]);
 		err = errno;
@@ -351,6 +351,7 @@ mlx5_pci_devinit(struct rte_pci_driver *pci_drv, struct rte_pci_device *pci_dev)
 		struct ibv_exp_device_attr exp_device_attr;
 #endif /* HAVE_EXP_QUERY_DEVICE */
 		struct ether_addr mac;
+		uint16_t num_vfs = 0;
 
 #ifdef HAVE_EXP_QUERY_DEVICE
 		exp_device_attr.comp_mask =
@@ -464,7 +465,8 @@ mlx5_pci_devinit(struct rte_pci_driver *pci_drv, struct rte_pci_device *pci_dev)
 		priv->ind_table_max_size = RSS_INDIRECTION_TABLE_SIZE;
 #endif /* HAVE_EXP_QUERY_DEVICE */
 
-		priv->vf = vf;
+		priv_get_num_vfs(priv, &num_vfs);
+		priv->sriov = (num_vfs || sriov);
 		priv->mps = mps;
 		/* Allocate and register default RSS hash keys. */
 		priv->rss_conf = rte_calloc(__func__, hash_rxq_init_n,
diff --git a/drivers/net/mlx5/mlx5.h b/drivers/net/mlx5/mlx5.h
index 2487662..dccc18d 100644
--- a/drivers/net/mlx5/mlx5.h
+++ b/drivers/net/mlx5/mlx5.h
@@ -105,7 +105,7 @@ struct priv {
 	unsigned int hw_vlan_strip:1; /* VLAN stripping is supported. */
 	unsigned int hw_fcs_strip:1; /* FCS stripping is supported. */
 	unsigned int hw_padding:1; /* End alignment padding is supported. */
-	unsigned int vf:1; /* This is a VF device. */
+	unsigned int sriov:1; /* This is a VF or PF with VF devices. */
 	unsigned int mps:1; /* Whether multi-packet send is supported. */
 	unsigned int pending_alarm:1; /* An alarm is pending. */
 	/* RX/TX queues. */
@@ -173,6 +173,7 @@ struct priv *mlx5_get_priv(struct rte_eth_dev *dev);
 int mlx5_is_secondary(void);
 int priv_get_ifname(const struct priv *, char (*)[IF_NAMESIZE]);
 int priv_ifreq(const struct priv *, int req, struct ifreq *);
+int priv_get_num_vfs(struct priv *, uint16_t *);
 int priv_get_mtu(struct priv *, uint16_t *);
 int priv_set_flags(struct priv *, unsigned int, unsigned int);
 int mlx5_dev_configure(struct rte_eth_dev *);
diff --git a/drivers/net/mlx5/mlx5_ethdev.c b/drivers/net/mlx5/mlx5_ethdev.c
index ecbb49b..d2a63b8 100644
--- a/drivers/net/mlx5/mlx5_ethdev.c
+++ b/drivers/net/mlx5/mlx5_ethdev.c
@@ -363,6 +363,38 @@ priv_ifreq(const struct priv *priv, int req, struct ifreq *ifr)
 }
 
 /**
+ * Return the number of active VFs for the current device.
+ *
+ * @param[in] priv
+ *   Pointer to private structure.
+ * @param[out] num_vfs
+ *   Number of active VFs.
+ *
+ * @return
+ *   0 on success, -1 on failure and errno is set.
+ */
+int
+priv_get_num_vfs(struct priv *priv, uint16_t *num_vfs)
+{
+	/* The sysfs entry name depends on the operating system. */
+	const char **name = (const char *[]){
+		"device/sriov_numvfs",
+		"device/mlx5_num_vfs",
+		NULL,
+	};
+	int ret;
+
+	do {
+		unsigned long ulong_num_vfs;
+
+		ret = priv_get_sysfs_ulong(priv, *name, &ulong_num_vfs);
+		if (!ret)
+			*num_vfs = ulong_num_vfs;
+	} while (*(++name) && ret);
+	return ret;
+}
+
+/**
  * Get device MTU.
  *
  * @param priv
diff --git a/drivers/net/mlx5/mlx5_rxtx.c b/drivers/net/mlx5/mlx5_rxtx.c
index 3881c30..9cb1dfa 100644
--- a/drivers/net/mlx5/mlx5_rxtx.c
+++ b/drivers/net/mlx5/mlx5_rxtx.c
@@ -507,7 +507,7 @@ tx_burst_sg(struct txq *txq, unsigned int segs, struct txq_elt *elt,
 		}
 		/* Update SGE. */
 		sge->addr = rte_pktmbuf_mtod(buf, uintptr_t);
-		if (txq->priv->vf)
+		if (txq->priv->sriov)
 			rte_prefetch0((volatile void *)
 				      (uintptr_t)sge->addr);
 		sge->length = DATA_LEN(buf);
@@ -662,7 +662,7 @@ mlx5_tx_burst(void *dpdk_txq, struct rte_mbuf **pkts, uint16_t pkts_n)
 			length = DATA_LEN(buf);
 			/* Update element. */
 			elt->buf = buf;
-			if (txq->priv->vf)
+			if (txq->priv->sriov)
 				rte_prefetch0((volatile void *)
 					      (uintptr_t)addr);
 			/* Prefetch next buffer data. */
-- 
2.1.4

[PATCH 7/7] mlx5: update documentation part related to features and limitations

[Nelio Laranjeiro]

Signed-off-by: Nelio Laranjeiro <nelio.laranjeiro@6wind.com>
---
 doc/guides/nics/mlx5.rst | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/doc/guides/nics/mlx5.rst b/doc/guides/nics/mlx5.rst
index b6f91e6..d9196d1 100644
--- a/doc/guides/nics/mlx5.rst
+++ b/doc/guides/nics/mlx5.rst
@@ -86,11 +86,11 @@ Features
 - Hardware checksum offloads.
 - Flow director (RTE_FDIR_MODE_PERFECT and RTE_FDIR_MODE_PERFECT_MAC_VLAN).
 - Secondary process TX is supported.
+- KVM and VMware ESX SR-IOV modes are supported.
 
 Limitations
 -----------
 
-- KVM and VMware ESX SR-IOV modes are not supported yet.
 - Inner RSS for VXLAN frames is not supported yet.
 - Port statistics through software counters only.
 - Hardware checksum offloads for VXLAN inner header are not supported yet.
-- 
2.1.4

Re: [PATCH 7/7] mlx5: update documentation part related to features and limitations

[Mcnamara, John]

> -----Original Message-----
> From: dev [mailto:dev-bounces@dpdk.org] On Behalf Of Nelio Laranjeiro
> Sent: Wednesday, June 8, 2016 10:44 AM
> To: dev@dpdk.org
> Subject: [dpdk-dev] [PATCH 7/7] mlx5: update documentation part related to
> features and limitations
> 
> Signed-off-by: Nelio Laranjeiro <nelio.laranjeiro@6wind.com>

Acked-by: John McNamara <john.mcnamara@intel.com>

Re: [PATCH 0/7] Miscellaneous fixes for mlx4 and mlx5

[Bruce Richardson]

On Wed, Jun 08, 2016 at 11:43:24AM +0200, Nelio Laranjeiro wrote:
> Various minor fixes for mlx4 (ConnectX-3) and mlx5 (ConnectX-4).
> 
> Adrien Mazarguil (4):
>   mlx: ensure MTU update is effective
>   mlx: retrieve mbuf size through proper API
>   mlx5: fix RX VLAN stripping capability check
>   mlx5: cosmetic changes (coding style)
> 
> Nelio Laranjeiro (3):
>   mlx: remove unused memory region property
>   mlx5: enhance SR-IOV detection
>   mlx5: update documentation part related to features and limitations
> 
>  doc/guides/nics/mlx5.rst       |  2 +-
>  drivers/net/mlx4/mlx4.c        | 43 ++++++++++++++++++++------------------
>  drivers/net/mlx5/mlx5.c        | 14 +++++++------
>  drivers/net/mlx5/mlx5.h        |  3 ++-
>  drivers/net/mlx5/mlx5_ethdev.c | 47 ++++++++++++++++++++++++++++++++++++++++--
>  drivers/net/mlx5/mlx5_rxq.c    | 23 +++++++--------------
>  drivers/net/mlx5/mlx5_rxtx.c   |  7 +++----
>  drivers/net/mlx5/mlx5_rxtx.h   | 16 ++++++--------
>  drivers/net/mlx5/mlx5_txq.c    |  2 +-
>  drivers/net/mlx5/mlx5_vlan.c   |  2 +-
>  10 files changed, 98 insertions(+), 61 deletions(-)
>
Applied to dpdk-next-net/rel_16_07

/Bruce