mlx5: add device configure/start/stop

This commit adds the remaining missing callbacks to make mlx5 usable.
Like mlx4, device start and stop are implemented on top of MAC RX flows.

Signed-off-by: Adrien Mazarguil <adrien.mazarguil@6wind.com>
Signed-off-by: Nelio Laranjeiro <nelio.laranjeiro@6wind.com>
Signed-off-by: Francesco Santoro <francesco.santoro@6wind.com>
Signed-off-by: Didier Pallard <didier.pallard@6wind.com>

drivers/net/mlx5/Makefile

@@ -45,6 +45,7 @@ SRCS-$(CONFIG_RTE_LIBRTE_MLX5_PMD) += mlx5.c
 SRCS-$(CONFIG_RTE_LIBRTE_MLX5_PMD) += mlx5_rxq.c
 SRCS-$(CONFIG_RTE_LIBRTE_MLX5_PMD) += mlx5_txq.c
 SRCS-$(CONFIG_RTE_LIBRTE_MLX5_PMD) += mlx5_rxtx.c
+SRCS-$(CONFIG_RTE_LIBRTE_MLX5_PMD) += mlx5_trigger.c
 SRCS-$(CONFIG_RTE_LIBRTE_MLX5_PMD) += mlx5_ethdev.c
 SRCS-$(CONFIG_RTE_LIBRTE_MLX5_PMD) += mlx5_mac.c
 

drivers/net/mlx5/mlx5.c

@@ -129,7 +129,11 @@ mlx5_dev_close(struct rte_eth_dev *dev)
 }
 
 static const struct eth_dev_ops mlx5_dev_ops = {
+	.dev_configure = mlx5_dev_configure,
+	.dev_start = mlx5_dev_start,
+	.dev_stop = mlx5_dev_stop,
 	.dev_close = mlx5_dev_close,
+	.dev_infos_get = mlx5_dev_infos_get,
 	.rx_queue_setup = mlx5_rx_queue_setup,
 	.tx_queue_setup = mlx5_tx_queue_setup,
 	.rx_queue_release = mlx5_rx_queue_release,

drivers/net/mlx5/mlx5.h

@@ -160,6 +160,8 @@ int priv_get_ifname(const struct priv *, char (*)[IF_NAMESIZE]);
 int priv_ifreq(const struct priv *, int req, struct ifreq *);
 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 *);
+void mlx5_dev_infos_get(struct rte_eth_dev *, struct rte_eth_dev_info *);
 int mlx5_ibv_device_to_pci_addr(const struct ibv_device *,
 				struct rte_pci_addr *);
 
@@ -174,4 +176,9 @@ int priv_mac_addr_add(struct priv *, unsigned int,
 void mlx5_mac_addr_add(struct rte_eth_dev *, struct ether_addr *, uint32_t,
 		       uint32_t);
 
+/* mlx5_trigger.c */
+
+int mlx5_dev_start(struct rte_eth_dev *);
+void mlx5_dev_stop(struct rte_eth_dev *);
+
 #endif /* RTE_PMD_MLX5_H_ */

drivers/net/mlx5/mlx5_ethdev.c

@@ -32,6 +32,7 @@
  */
 
 #include <stddef.h>
+#include <assert.h>
 #include <unistd.h>
 #include <stdint.h>
 #include <stdio.h>
@@ -58,6 +59,7 @@
 #endif
 
 #include "mlx5.h"
+#include "mlx5_rxtx.h"
 #include "mlx5_utils.h"
 
 /**
@@ -369,6 +371,152 @@ priv_set_flags(struct priv *priv, unsigned int keep, unsigned int flags)
 	return priv_set_sysfs_ulong(priv, "flags", tmp);
 }
 
+/**
+ * Ethernet device configuration.
+ *
+ * Prepare the driver for a given number of TX and RX queues.
+ * Allocate parent RSS queue when several RX queues are requested.
+ *
+ * @param dev
+ *   Pointer to Ethernet device structure.
+ *
+ * @return
+ *   0 on success, errno value on failure.
+ */
+static int
+dev_configure(struct rte_eth_dev *dev)
+{
+	struct priv *priv = dev->data->dev_private;
+	unsigned int rxqs_n = dev->data->nb_rx_queues;
+	unsigned int txqs_n = dev->data->nb_tx_queues;
+	unsigned int tmp;
+	int ret;
+
+	priv->rxqs = (void *)dev->data->rx_queues;
+	priv->txqs = (void *)dev->data->tx_queues;
+	if (txqs_n != priv->txqs_n) {
+		INFO("%p: TX queues number update: %u -> %u",
+		     (void *)dev, priv->txqs_n, txqs_n);
+		priv->txqs_n = txqs_n;
+	}
+	if (rxqs_n == priv->rxqs_n)
+		return 0;
+	INFO("%p: RX queues number update: %u -> %u",
+	     (void *)dev, priv->rxqs_n, rxqs_n);
+	/* If RSS is enabled, disable it first. */
+	if (priv->rss) {
+		unsigned int i;
+
+		/* Only if there are no remaining child RX queues. */
+		for (i = 0; (i != priv->rxqs_n); ++i)
+			if ((*priv->rxqs)[i] != NULL)
+				return EINVAL;
+		rxq_cleanup(&priv->rxq_parent);
+		priv->rss = 0;
+		priv->rxqs_n = 0;
+	}
+	if (rxqs_n <= 1) {
+		/* Nothing else to do. */
+		priv->rxqs_n = rxqs_n;
+		return 0;
+	}
+	/* Allocate a new RSS parent queue if supported by hardware. */
+	if (!priv->hw_rss) {
+		ERROR("%p: only a single RX queue can be configured when"
+		      " hardware doesn't support RSS",
+		      (void *)dev);
+		return EINVAL;
+	}
+	/* Fail if hardware doesn't support that many RSS queues. */
+	if (rxqs_n >= priv->max_rss_tbl_sz) {
+		ERROR("%p: only %u RX queues can be configured for RSS",
+		      (void *)dev, priv->max_rss_tbl_sz);
+		return EINVAL;
+	}
+	priv->rss = 1;
+	tmp = priv->rxqs_n;
+	priv->rxqs_n = rxqs_n;
+	ret = rxq_setup(dev, &priv->rxq_parent, 0, 0, NULL, NULL);
+	if (!ret)
+		return 0;
+	/* Failure, rollback. */
+	priv->rss = 0;
+	priv->rxqs_n = tmp;
+	assert(ret > 0);
+	return ret;
+}
+
+/**
+ * DPDK callback for Ethernet device configuration.
+ *
+ * @param dev
+ *   Pointer to Ethernet device structure.
+ *
+ * @return
+ *   0 on success, negative errno value on failure.
+ */
+int
+mlx5_dev_configure(struct rte_eth_dev *dev)
+{
+	struct priv *priv = dev->data->dev_private;
+	int ret;
+
+	priv_lock(priv);
+	ret = dev_configure(dev);
+	assert(ret >= 0);
+	priv_unlock(priv);
+	return -ret;
+}
+
+/**
+ * DPDK callback to get information about the device.
+ *
+ * @param dev
+ *   Pointer to Ethernet device structure.
+ * @param[out] info
+ *   Info structure output buffer.
+ */
+void
+mlx5_dev_infos_get(struct rte_eth_dev *dev, struct rte_eth_dev_info *info)
+{
+	struct priv *priv = dev->data->dev_private;
+	unsigned int max;
+	char ifname[IF_NAMESIZE];
+
+	priv_lock(priv);
+	/* FIXME: we should ask the device for these values. */
+	info->min_rx_bufsize = 32;
+	info->max_rx_pktlen = 65536;
+	/*
+	 * Since we need one CQ per QP, the limit is the minimum number
+	 * between the two values.
+	 */
+	max = ((priv->device_attr.max_cq > priv->device_attr.max_qp) ?
+	       priv->device_attr.max_qp : priv->device_attr.max_cq);
+	/* If max >= 65535 then max = 0, max_rx_queues is uint16_t. */
+	if (max >= 65535)
+		max = 65535;
+	info->max_rx_queues = max;
+	info->max_tx_queues = max;
+	/* Last array entry is reserved for broadcast. */
+	info->max_mac_addrs = (RTE_DIM(priv->mac) - 1);
+	info->rx_offload_capa =
+		(priv->hw_csum ?
+		 (DEV_RX_OFFLOAD_IPV4_CKSUM |
+		  DEV_RX_OFFLOAD_UDP_CKSUM |
+		  DEV_RX_OFFLOAD_TCP_CKSUM) :
+		 0);
+	info->tx_offload_capa =
+		(priv->hw_csum ?
+		 (DEV_TX_OFFLOAD_IPV4_CKSUM |
+		  DEV_TX_OFFLOAD_UDP_CKSUM |
+		  DEV_TX_OFFLOAD_TCP_CKSUM) :
+		 0);
+	if (priv_get_ifname(priv, &ifname) == 0)
+		info->if_index = if_nametoindex(ifname);
+	priv_unlock(priv);
+}
+
 /**
  * Get PCI information from struct ibv_device.
  *

drivers/net/mlx5/mlx5_trigger.c

@@ -0,0 +1,146 @@
+/*-
+ *   BSD LICENSE
+ *
+ *   Copyright 2015 6WIND S.A.
+ *   Copyright 2015 Mellanox.
+ *
+ *   Redistribution and use in source and binary forms, with or without
+ *   modification, are permitted provided that the following conditions
+ *   are met:
+ *
+ *     * Redistributions of source code must retain the above copyright
+ *       notice, this list of conditions and the following disclaimer.
+ *     * Redistributions in binary form must reproduce the above copyright
+ *       notice, this list of conditions and the following disclaimer in
+ *       the documentation and/or other materials provided with the
+ *       distribution.
+ *     * Neither the name of 6WIND S.A. nor the names of its
+ *       contributors may be used to endorse or promote products derived
+ *       from this software without specific prior written permission.
+ *
+ *   THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ *   "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ *   LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+ *   A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+ *   OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ *   SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ *   LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ *   DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+ *   THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ *   (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+ *   OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+/* DPDK headers don't like -pedantic. */
+#ifdef PEDANTIC
+#pragma GCC diagnostic ignored "-pedantic"
+#endif
+#include <rte_ether.h>
+#include <rte_ethdev.h>
+#ifdef PEDANTIC
+#pragma GCC diagnostic error "-pedantic"
+#endif
+
+#include "mlx5.h"
+#include "mlx5_rxtx.h"
+#include "mlx5_utils.h"
+
+/**
+ * DPDK callback to start the device.
+ *
+ * Simulate device start by attaching all configured flows.
+ *
+ * @param dev
+ *   Pointer to Ethernet device structure.
+ *
+ * @return
+ *   0 on success, negative errno value on failure.
+ */
+int
+mlx5_dev_start(struct rte_eth_dev *dev)
+{
+	struct priv *priv = dev->data->dev_private;
+	unsigned int i = 0;
+	unsigned int r;
+	struct rxq *rxq;
+
+	priv_lock(priv);
+	if (priv->started) {
+		priv_unlock(priv);
+		return 0;
+	}
+	DEBUG("%p: attaching configured flows to all RX queues", (void *)dev);
+	priv->started = 1;
+	if (priv->rss) {
+		rxq = &priv->rxq_parent;
+		r = 1;
+	} else {
+		rxq = (*priv->rxqs)[0];
+		r = priv->rxqs_n;
+	}
+	/* Iterate only once when RSS is enabled. */
+	do {
+		int ret;
+
+		/* Ignore nonexistent RX queues. */
+		if (rxq == NULL)
+			continue;
+		ret = rxq_mac_addrs_add(rxq);
+		if (!ret)
+			continue;
+		WARN("%p: QP flow attachment failed: %s",
+		     (void *)dev, strerror(ret));
+		/* Rollback. */
+		while (i != 0) {
+			rxq = (*priv->rxqs)[--i];
+			if (rxq != NULL) {
+				rxq_mac_addrs_del(rxq);
+			}
+		}
+		priv->started = 0;
+		priv_unlock(priv);
+		return -ret;
+	} while ((--r) && ((rxq = (*priv->rxqs)[++i]), i));
+	priv_unlock(priv);
+	return 0;
+}
+
+/**
+ * DPDK callback to stop the device.
+ *
+ * Simulate device stop by detaching all configured flows.
+ *
+ * @param dev
+ *   Pointer to Ethernet device structure.
+ */
+void
+mlx5_dev_stop(struct rte_eth_dev *dev)
+{
+	struct priv *priv = dev->data->dev_private;
+	unsigned int i = 0;
+	unsigned int r;
+	struct rxq *rxq;
+
+	priv_lock(priv);
+	if (!priv->started) {
+		priv_unlock(priv);
+		return;
+	}
+	DEBUG("%p: detaching flows from all RX queues", (void *)dev);
+	priv->started = 0;
+	if (priv->rss) {
+		rxq = &priv->rxq_parent;
+		r = 1;
+	} else {
+		rxq = (*priv->rxqs)[0];
+		r = priv->rxqs_n;
+	}
+	/* Iterate only once when RSS is enabled. */
+	do {
+		/* Ignore nonexistent RX queues. */
+		if (rxq == NULL)
+			continue;
+		rxq_mac_addrs_del(rxq);
+	} while ((--r) && ((rxq = (*priv->rxqs)[++i]), i));
+	priv_unlock(priv);
+}