i40evf: add client interface

In preparation for upcoming RDMA-capable hardware, add a client
interface to the VF driver. This is a slightly-simplified version
of the PF client interface, with the names changed to protect the
innocent.

Due to the nature of the VF<->PF interactions, the client interface
sometimes needs to call back into itself to pass messages. Because
of this, we can't use the coarse-grained locking like the PF's
client interface uses. Instead, we handle all client interactions
in a separate thread so the watchdog can still run and process
virtual channel messages.
Signed-off-by: Mitch Williams <mitch.a.williams@intel.com>
Signed-off-by: Jesse Brandeburg <jesse.brandeburg@intel.com>
Signed-off-by: Anjali Singhai Jain <anjali.singhai@intel.com>
Signed-off-by: Avinash Dayanand <avinash.dayanand@intel.com>
Tested-by: Andrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>

drivers/net/ethernet/intel/i40evf/Makefile

@@ -32,5 +32,5 @@
 obj-$(CONFIG_I40EVF) += i40evf.o
 
 i40evf-objs :=	i40evf_main.o i40evf_ethtool.o i40evf_virtchnl.o \
-		i40e_txrx.o i40e_common.o i40e_adminq.o
+		i40e_txrx.o i40e_common.o i40e_adminq.o i40evf_client.o
 

drivers/net/ethernet/intel/i40evf/i40e_virtchnl.h

@@ -81,7 +81,9 @@ enum i40e_virtchnl_ops {
 	I40E_VIRTCHNL_OP_GET_STATS = 15,
 	I40E_VIRTCHNL_OP_FCOE = 16,
 	I40E_VIRTCHNL_OP_EVENT = 17, /* must ALWAYS be 17 */
+	I40E_VIRTCHNL_OP_IWARP = 20,
 	I40E_VIRTCHNL_OP_CONFIG_IWARP_IRQ_MAP = 21,
+	I40E_VIRTCHNL_OP_RELEASE_IWARP_IRQ_MAP = 22,
 	I40E_VIRTCHNL_OP_CONFIG_RSS_KEY = 23,
 	I40E_VIRTCHNL_OP_CONFIG_RSS_LUT = 24,
 	I40E_VIRTCHNL_OP_GET_RSS_HENA_CAPS = 25,
@@ -393,6 +395,37 @@ struct i40e_virtchnl_pf_event {
 	int severity;
 };
 
+/* I40E_VIRTCHNL_OP_CONFIG_IWARP_IRQ_MAP
+ * VF uses this message to request PF to map IWARP vectors to IWARP queues.
+ * The request for this originates from the VF IWARP driver through
+ * a client interface between VF LAN and VF IWARP driver.
+ * A vector could have an AEQ and CEQ attached to it although
+ * there is a single AEQ per VF IWARP instance in which case
+ * most vectors will have an INVALID_IDX for aeq and valid idx for ceq.
+ * There will never be a case where there will be multiple CEQs attached
+ * to a single vector.
+ * PF configures interrupt mapping and returns status.
+ */
+
+/* HW does not define a type value for AEQ; only for RX/TX and CEQ.
+ * In order for us to keep the interface simple, SW will define a
+ * unique type value for AEQ.
+ */
+#define I40E_QUEUE_TYPE_PE_AEQ  0x80
+#define I40E_QUEUE_INVALID_IDX  0xFFFF
+
+struct i40e_virtchnl_iwarp_qv_info {
+	u32 v_idx; /* msix_vector */
+	u16 ceq_idx;
+	u16 aeq_idx;
+	u8 itr_idx;
+};
+
+struct i40e_virtchnl_iwarp_qvlist_info {
+	u32 num_vectors;
+	struct i40e_virtchnl_iwarp_qv_info qv_info[1];
+};
+
 /* VF reset states - these are written into the RSTAT register:
  * I40E_VFGEN_RSTAT1 on the PF
  * I40E_VFGEN_RSTAT on the VF

drivers/net/ethernet/intel/i40evf/i40evf.h

@@ -60,6 +60,7 @@ struct i40e_vsi {
 	int base_vector;
 	u16 work_limit;
 	u16 qs_handle;
+	void *priv;     /* client driver data reference. */
 };
 
 /* How many Rx Buffers do we bundle into one write to the hardware ? */
@@ -169,6 +170,7 @@ enum i40evf_state_t {
 
 enum i40evf_critical_section_t {
 	__I40EVF_IN_CRITICAL_TASK,	/* cannot be interrupted */
+	__I40EVF_IN_CLIENT_TASK,
 };
 /* make common code happy */
 #define __I40E_DOWN __I40EVF_DOWN
@@ -178,6 +180,7 @@ struct i40evf_adapter {
 	struct timer_list watchdog_timer;
 	struct work_struct reset_task;
 	struct work_struct adminq_task;
+	struct delayed_work client_task;
 	struct delayed_work init_task;
 	struct i40e_q_vector *q_vectors;
 	struct list_head vlan_filter_list;
@@ -195,7 +198,10 @@ struct i40evf_adapter {
 	u64 hw_csum_rx_error;
 	u32 rx_desc_count;
 	int num_msix_vectors;
+	int num_iwarp_msix;
+	int iwarp_base_vector;
 	u32 client_pending;
+	struct i40e_client_instance *cinst;
 	struct msix_entry *msix_entries;
 
 	u32 flags;
@@ -211,8 +217,11 @@ struct i40evf_adapter {
 #define I40EVF_FLAG_OUTER_UDP_CSUM_CAPABLE	BIT(12)
 #define I40EVF_FLAG_ADDR_SET_BY_PF		BIT(13)
 #define I40EVF_FLAG_SERVICE_CLIENT_REQUESTED	BIT(14)
-#define I40EVF_FLAG_PROMISC_ON			BIT(15)
-#define I40EVF_FLAG_ALLMULTI_ON			BIT(16)
+#define I40EVF_FLAG_CLIENT_NEEDS_OPEN		BIT(15)
+#define I40EVF_FLAG_CLIENT_NEEDS_CLOSE		BIT(16)
+#define I40EVF_FLAG_CLIENT_NEEDS_L2_PARAMS	BIT(17)
+#define I40EVF_FLAG_PROMISC_ON			BIT(18)
+#define I40EVF_FLAG_ALLMULTI_ON			BIT(19)
 /* duplicates for common code */
 #define I40E_FLAG_FDIR_ATR_ENABLED		0
 #define I40E_FLAG_DCB_ENABLED			0
@@ -258,10 +267,11 @@ struct i40evf_adapter {
 	bool link_up;
 	enum i40e_aq_link_speed link_speed;
 	enum i40e_virtchnl_ops current_op;
-#define CLIENT_ENABLED(_a) ((_a)->vf_res ? \
+#define CLIENT_ALLOWED(_a) ((_a)->vf_res ? \
 			    (_a)->vf_res->vf_offload_flags & \
 				I40E_VIRTCHNL_VF_OFFLOAD_IWARP : \
 			    0)
+#define CLIENT_ENABLED(_a) ((_a)->cinst)
 /* RSS by the PF should be preferred over RSS via other methods. */
 #define RSS_PF(_a) ((_a)->vf_res->vf_offload_flags & \
 		    I40E_VIRTCHNL_VF_OFFLOAD_RSS_PF)
@@ -292,6 +302,12 @@ struct i40evf_adapter {
 
 /* Ethtool Private Flags */
 
+/* lan device */
+struct i40e_device {
+	struct list_head list;
+	struct i40evf_adapter *vf;
+};
+
 /* needed by i40evf_ethtool.c */
 extern char i40evf_driver_name[];
 extern const char i40evf_driver_version[];
@@ -337,4 +353,11 @@ void i40evf_virtchnl_completion(struct i40evf_adapter *adapter,
 				enum i40e_virtchnl_ops v_opcode,
 				i40e_status v_retval, u8 *msg, u16 msglen);
 int i40evf_config_rss(struct i40evf_adapter *adapter);
+int i40evf_lan_add_device(struct i40evf_adapter *adapter);
+int i40evf_lan_del_device(struct i40evf_adapter *adapter);
+void i40evf_client_subtask(struct i40evf_adapter *adapter);
+void i40evf_notify_client_message(struct i40e_vsi *vsi, u8 *msg, u16 len);
+void i40evf_notify_client_l2_params(struct i40e_vsi *vsi);
+void i40evf_notify_client_open(struct i40e_vsi *vsi);
+void i40evf_notify_client_close(struct i40e_vsi *vsi, bool reset);
 #endif /* _I40EVF_H_ */

drivers/net/ethernet/intel/i40evf/i40evf_client.c

+#include <linux/list.h>
+#include <linux/errno.h>
+
+#include "i40evf.h"
+#include "i40e_prototype.h"
+#include "i40evf_client.h"
+
+static
+const char i40evf_client_interface_version_str[] = I40EVF_CLIENT_VERSION_STR;
+static struct i40e_client *vf_registered_client;
+static LIST_HEAD(i40evf_devices);
+static DEFINE_MUTEX(i40evf_device_mutex);
+
+static u32 i40evf_client_virtchnl_send(struct i40e_info *ldev,
+				       struct i40e_client *client,
+				       u8 *msg, u16 len);
+
+static int i40evf_client_setup_qvlist(struct i40e_info *ldev,
+				      struct i40e_client *client,
+				      struct i40e_qvlist_info *qvlist_info);
+
+static struct i40e_ops i40evf_lan_ops = {
+	.virtchnl_send = i40evf_client_virtchnl_send,
+	.setup_qvlist = i40evf_client_setup_qvlist,
+};
+
+/**
+ * i40evf_notify_client_message - call the client message receive callback
+ * @vsi: the VSI associated with this client
+ * @msg: message buffer
+ * @len: length of message
+ *
+ * If there is a client to this VSI, call the client
+ **/
+void i40evf_notify_client_message(struct i40e_vsi *vsi, u8 *msg, u16 len)
+{
+	struct i40evf_adapter *adapter = vsi->back;
+	struct i40e_client_instance *cinst = adapter->cinst;
+
+	if (!vsi)
+		return;
+
+	if (!cinst || !cinst->client || !cinst->client->ops ||
+	    !cinst->client->ops->virtchnl_receive) {
+		dev_dbg(&vsi->back->pdev->dev,
+			"Cannot locate client instance virtchnl_receive function\n");
+		return;
+	}
+	cinst->client->ops->virtchnl_receive(&cinst->lan_info,  cinst->client,
+					     msg, len);
+}
+
+/**
+ * i40evf_notify_client_l2_params - call the client notify callback
+ * @vsi: the VSI with l2 param changes
+ *
+ * If there is a client to this VSI, call the client
+ **/
+void i40evf_notify_client_l2_params(struct i40e_vsi *vsi)
+{
+	struct i40evf_adapter *adapter = vsi->back;
+	struct i40e_client_instance *cinst = adapter->cinst;
+	struct i40e_params params;
+
+	if (!vsi)
+		return;
+	memset(&params, 0, sizeof(params));
+	params.mtu = vsi->netdev->mtu;
+	params.link_up = vsi->back->link_up;
+	params.qos.prio_qos[0].qs_handle = vsi->qs_handle;
+
+	if (!cinst || !cinst->client || !cinst->client->ops ||
+	    !cinst->client->ops->l2_param_change) {
+		dev_dbg(&vsi->back->pdev->dev,
+			"Cannot locate client instance l2_param_change function\n");
+		return;
+	}
+	cinst->client->ops->l2_param_change(&cinst->lan_info, cinst->client,
+					    &params);
+}
+
+/**
+ * i40evf_notify_client_open - call the client open callback
+ * @vsi: the VSI with netdev opened
+ *
+ * If there is a client to this netdev, call the client with open
+ **/
+void i40evf_notify_client_open(struct i40e_vsi *vsi)
+{
+	struct i40evf_adapter *adapter = vsi->back;
+	struct i40e_client_instance *cinst = adapter->cinst;
+	int ret;
+
+	if (!cinst || !cinst->client || !cinst->client->ops ||
+	    !cinst->client->ops->open) {
+		dev_dbg(&vsi->back->pdev->dev,
+			"Cannot locate client instance open function\n");
+		return;
+	}
+	if (!(test_bit(__I40E_CLIENT_INSTANCE_OPENED, &cinst->state))) {
+		ret = cinst->client->ops->open(&cinst->lan_info, cinst->client);
+		if (!ret)
+			set_bit(__I40E_CLIENT_INSTANCE_OPENED, &cinst->state);
+	}
+}
+
+/**
+ * i40evf_client_release_qvlist - send a message to the PF to release iwarp qv map
+ * @ldev: pointer to L2 context.
+ *
+ * Return 0 on success or < 0 on error
+ **/
+static int i40evf_client_release_qvlist(struct i40e_info *ldev)
+{
+	struct i40evf_adapter *adapter = ldev->vf;
+	i40e_status err;
+
+	if (adapter->aq_required)
+		return -EAGAIN;
+
+	err = i40e_aq_send_msg_to_pf(&adapter->hw,
+			I40E_VIRTCHNL_OP_RELEASE_IWARP_IRQ_MAP,
+			I40E_SUCCESS, NULL, 0, NULL);
+
+	if (err)
+		dev_err(&adapter->pdev->dev,
+			"Unable to send iWarp vector release message to PF, error %d, aq status %d\n",
+			err, adapter->hw.aq.asq_last_status);
+
+	return err;
+}
+
+/**
+ * i40evf_notify_client_close - call the client close callback
+ * @vsi: the VSI with netdev closed
+ * @reset: true when close called due to reset pending
+ *
+ * If there is a client to this netdev, call the client with close
+ **/
+void i40evf_notify_client_close(struct i40e_vsi *vsi, bool reset)
+{
+	struct i40evf_adapter *adapter = vsi->back;
+	struct i40e_client_instance *cinst = adapter->cinst;
+
+	if (!cinst || !cinst->client || !cinst->client->ops ||
+	    !cinst->client->ops->close) {
+		dev_dbg(&vsi->back->pdev->dev,
+			"Cannot locate client instance close function\n");
+		return;
+	}
+	cinst->client->ops->close(&cinst->lan_info, cinst->client, reset);
+	i40evf_client_release_qvlist(&cinst->lan_info);
+	clear_bit(__I40E_CLIENT_INSTANCE_OPENED, &cinst->state);
+}
+
+/**
+ * i40evf_client_add_instance - add a client instance to the instance list
+ * @adapter: pointer to the board struct
+ * @client: pointer to a client struct in the client list.
+ *
+ * Returns cinst ptr on success, NULL on failure
+ **/
+static struct i40e_client_instance *
+i40evf_client_add_instance(struct i40evf_adapter *adapter)
+{
+	struct i40e_client_instance *cinst = NULL;
+	struct netdev_hw_addr *mac = NULL;
+	struct i40e_vsi *vsi = &adapter->vsi;
+	int i;
+
+	if (!vf_registered_client)
+		goto out;
+
+	if (adapter->cinst) {
+		cinst = adapter->cinst;
+		goto out;
+	}
+
+	cinst = kzalloc(sizeof(*cinst), GFP_KERNEL);
+	if (!cinst)
+		goto out;
+
+	cinst->lan_info.vf = (void *)adapter;
+	cinst->lan_info.netdev = vsi->netdev;
+	cinst->lan_info.pcidev = adapter->pdev;
+	cinst->lan_info.fid = 0;
+	cinst->lan_info.ftype = I40E_CLIENT_FTYPE_VF;
+	cinst->lan_info.hw_addr = adapter->hw.hw_addr;
+	cinst->lan_info.ops = &i40evf_lan_ops;
+	cinst->lan_info.version.major = I40EVF_CLIENT_VERSION_MAJOR;
+	cinst->lan_info.version.minor = I40EVF_CLIENT_VERSION_MINOR;
+	cinst->lan_info.version.build = I40EVF_CLIENT_VERSION_BUILD;
+	set_bit(__I40E_CLIENT_INSTANCE_NONE, &cinst->state);
+
+	cinst->lan_info.msix_count = adapter->num_iwarp_msix;
+	cinst->lan_info.msix_entries =
+			&adapter->msix_entries[adapter->iwarp_base_vector];
+
+	for (i = 0; i < I40E_MAX_USER_PRIORITY; i++) {
+		cinst->lan_info.params.qos.prio_qos[i].tc = 0;
+		cinst->lan_info.params.qos.prio_qos[i].qs_handle =
+								vsi->qs_handle;
+	}
+
+	mac = list_first_entry(&cinst->lan_info.netdev->dev_addrs.list,
+			       struct netdev_hw_addr, list);
+	if (mac)
+		ether_addr_copy(cinst->lan_info.lanmac, mac->addr);
+	else
+		dev_err(&adapter->pdev->dev, "MAC address list is empty!\n");
+
+	cinst->client = vf_registered_client;
+	adapter->cinst = cinst;
+out:
+	return cinst;
+}
+
+/**
+ * i40evf_client_del_instance - removes a client instance from the list
+ * @adapter: pointer to the board struct
+ * @client: pointer to the client struct
+ *
+ **/
+static
+void i40evf_client_del_instance(struct i40evf_adapter *adapter)
+{
+	kfree(adapter->cinst);
+	adapter->cinst = NULL;
+}
+
+/**
+ * i40evf_client_subtask - client maintenance work
+ * @adapter: board private structure
+ **/
+void i40evf_client_subtask(struct i40evf_adapter *adapter)
+{
+	struct i40e_client *client = vf_registered_client;
+	struct i40e_client_instance *cinst;
+	int ret = 0;
+
+	if (adapter->state < __I40EVF_DOWN)
+		return;
+
+	/* first check client is registered */
+	if (!client)
+		return;
+
+	/* Add the client instance to the instance list */
+	cinst = i40evf_client_add_instance(adapter);
+	if (!cinst)
+		return;
+
+	dev_info(&adapter->pdev->dev, "Added instance of Client %s\n",
+		 client->name);
+
+	if (!test_bit(__I40E_CLIENT_INSTANCE_OPENED, &cinst->state)) {
+		/* Send an Open request to the client */
+
+		if (client->ops && client->ops->open)
+			ret = client->ops->open(&cinst->lan_info, client);
+		if (!ret)
+			set_bit(__I40E_CLIENT_INSTANCE_OPENED,
+				&cinst->state);
+		else
+			/* remove client instance */
+			i40evf_client_del_instance(adapter);
+	}
+}
+
+/**
+ * i40evf_lan_add_device - add a lan device struct to the list of lan devices
+ * @adapter: pointer to the board struct
+ *
+ * Returns 0 on success or none 0 on error
+ **/
+int i40evf_lan_add_device(struct i40evf_adapter *adapter)
+{
+	struct i40e_device *ldev;
+	int ret = 0;
+
+	mutex_lock(&i40evf_device_mutex);
+	list_for_each_entry(ldev, &i40evf_devices, list) {
+		if (ldev->vf == adapter) {
+			ret = -EEXIST;
+			goto out;
+		}
+	}
+	ldev = kzalloc(sizeof(*ldev), GFP_KERNEL);
+	if (!ldev) {
+		ret = -ENOMEM;
+		goto out;
+	}
+	ldev->vf = adapter;
+	INIT_LIST_HEAD(&ldev->list);
+	list_add(&ldev->list, &i40evf_devices);
+	dev_info(&adapter->pdev->dev, "Added LAN device bus=0x%02x dev=0x%02x func=0x%02x\n",
+		 adapter->hw.bus.bus_id, adapter->hw.bus.device,
+		 adapter->hw.bus.func);
+
+	/* Since in some cases register may have happened before a device gets
+	 * added, we can schedule a subtask to go initiate the clients.
+	 */
+	adapter->flags |= I40EVF_FLAG_SERVICE_CLIENT_REQUESTED;
+
+out:
+	mutex_unlock(&i40evf_device_mutex);
+	return ret;
+}
+
+/**
+ * i40evf_lan_del_device - removes a lan device from the device list
+ * @adapter: pointer to the board struct
+ *
+ * Returns 0 on success or non-0 on error
+ **/
+int i40evf_lan_del_device(struct i40evf_adapter *adapter)
+{
+	struct i40e_device *ldev, *tmp;
+	int ret = -ENODEV;
+
+	mutex_lock(&i40evf_device_mutex);
+	list_for_each_entry_safe(ldev, tmp, &i40evf_devices, list) {
+		if (ldev->vf == adapter) {
+			dev_info(&adapter->pdev->dev,
+				 "Deleted LAN device bus=0x%02x dev=0x%02x func=0x%02x\n",
+				 adapter->hw.bus.bus_id, adapter->hw.bus.device,
+				 adapter->hw.bus.func);
+			list_del(&ldev->list);
+			kfree(ldev);
+			ret = 0;
+			break;
+		}
+	}
+
+	mutex_unlock(&i40evf_device_mutex);
+	return ret;
+}
+
+/**
+ * i40evf_client_release - release client specific resources
+ * @client: pointer to the registered client
+ *
+ **/
+static void i40evf_client_release(struct i40e_client *client)
+{
+	struct i40e_client_instance *cinst;
+	struct i40e_device *ldev;
+	struct i40evf_adapter *adapter;
+
+	mutex_lock(&i40evf_device_mutex);
+	list_for_each_entry(ldev, &i40evf_devices, list) {
+		adapter = ldev->vf;
+		cinst = adapter->cinst;
+		if (!cinst)
+			continue;
+		if (test_bit(__I40E_CLIENT_INSTANCE_OPENED, &cinst->state)) {
+			if (client->ops && client->ops->close)
+				client->ops->close(&cinst->lan_info, client,
+						   false);
+			i40evf_client_release_qvlist(&cinst->lan_info);
+			clear_bit(__I40E_CLIENT_INSTANCE_OPENED, &cinst->state);
+
+			dev_warn(&adapter->pdev->dev,
+				 "Client %s instance closed\n", client->name);
+		}
+		/* delete the client instance */
+		i40evf_client_del_instance(adapter);
+		dev_info(&adapter->pdev->dev, "Deleted client instance of Client %s\n",
+			 client->name);
+	}
+	mutex_unlock(&i40evf_device_mutex);
+}
+
+/**
+ * i40evf_client_prepare - prepare client specific resources
+ * @client: pointer to the registered client
+ *
+ **/
+static void i40evf_client_prepare(struct i40e_client *client)
+{
+	struct i40e_device *ldev;
+	struct i40evf_adapter *adapter;
+
+	mutex_lock(&i40evf_device_mutex);
+	list_for_each_entry(ldev, &i40evf_devices, list) {
+		adapter = ldev->vf;
+		/* Signal the watchdog to service the client */
+		adapter->flags |= I40EVF_FLAG_SERVICE_CLIENT_REQUESTED;
+	}
+	mutex_unlock(&i40evf_device_mutex);
+}
+
+/**
+ * i40evf_client_virtchnl_send - send a message to the PF instance
+ * @ldev: pointer to L2 context.
+ * @client: Client pointer.
+ * @msg: pointer to message buffer
+ * @len: message length
+ *
+ * Return 0 on success or < 0 on error
+ **/
+static u32 i40evf_client_virtchnl_send(struct i40e_info *ldev,
+				       struct i40e_client *client,
+				       u8 *msg, u16 len)
+{
+	struct i40evf_adapter *adapter = ldev->vf;
+	i40e_status err;
+
+	if (adapter->aq_required)
+		return -EAGAIN;
+
+	err = i40e_aq_send_msg_to_pf(&adapter->hw, I40E_VIRTCHNL_OP_IWARP,
+				     I40E_SUCCESS, msg, len, NULL);
+	if (err)
+		dev_err(&adapter->pdev->dev, "Unable to send iWarp message to PF, error %d, aq status %d\n",
+			err, adapter->hw.aq.asq_last_status);
+
+	return err;
+}
+
+/**
+ * i40evf_client_setup_qvlist - send a message to the PF to setup iwarp qv map
+ * @ldev: pointer to L2 context.
+ * @client: Client pointer.
+ * @qv_info: queue and vector list
+ *
+ * Return 0 on success or < 0 on error
+ **/
+static int i40evf_client_setup_qvlist(struct i40e_info *ldev,
+				      struct i40e_client *client,
+				      struct i40e_qvlist_info *qvlist_info)
+{
+	struct i40e_virtchnl_iwarp_qvlist_info *v_qvlist_info;
+	struct i40evf_adapter *adapter = ldev->vf;
+	struct i40e_qv_info *qv_info;
+	i40e_status err;
+	u32 v_idx, i;
+	u32 msg_size;
+
+	if (adapter->aq_required)
+		return -EAGAIN;
+
+	/* A quick check on whether the vectors belong to the client */
+	for (i = 0; i < qvlist_info->num_vectors; i++) {
+		qv_info = &qvlist_info->qv_info[i];
+		if (!qv_info)
+			continue;
+		v_idx = qv_info->v_idx;
+		if ((v_idx >=
+		    (adapter->iwarp_base_vector + adapter->num_iwarp_msix)) ||
+		    (v_idx < adapter->iwarp_base_vector))
+			return -EINVAL;
+	}
+
+	v_qvlist_info = (struct i40e_virtchnl_iwarp_qvlist_info *)qvlist_info;
+	msg_size = sizeof(struct i40e_virtchnl_iwarp_qvlist_info) +
+			(sizeof(struct i40e_virtchnl_iwarp_qv_info) *
+			(v_qvlist_info->num_vectors - 1));
+
+	adapter->client_pending |= BIT(I40E_VIRTCHNL_OP_CONFIG_IWARP_IRQ_MAP);
+	err = i40e_aq_send_msg_to_pf(&adapter->hw,
+			I40E_VIRTCHNL_OP_CONFIG_IWARP_IRQ_MAP,
+			I40E_SUCCESS, (u8 *)v_qvlist_info, msg_size, NULL);
+
+	if (err) {
+		dev_err(&adapter->pdev->dev,
+			"Unable to send iWarp vector config message to PF, error %d, aq status %d\n",
+			err, adapter->hw.aq.asq_last_status);
+		goto out;
+	}
+
+	err = -EBUSY;
+	for (i = 0; i < 5; i++) {
+		msleep(100);
+		if (!(adapter->client_pending &
+		      BIT(I40E_VIRTCHNL_OP_CONFIG_IWARP_IRQ_MAP))) {
+			err = 0;
+			break;
+		}
+	}
+out:
+	return err;
+}
+
+/**
+ * i40evf_register_client - Register a i40e client driver with the L2 driver
+ * @client: pointer to the i40e_client struct
+ *
+ * Returns 0 on success or non-0 on error
+ **/
+int i40evf_register_client(struct i40e_client *client)
+{
+	int ret = 0;
+
+	if (!client) {
+		ret = -EIO;
+		goto out;
+	}
+
+	if (strlen(client->name) == 0) {
+		pr_info("i40evf: Failed to register client with no name\n");
+		ret = -EIO;
+		goto out;
+	}
+
+	if (vf_registered_client) {
+		pr_info("i40evf: Client %s has already been registered!\n",
+			client->name);
+		ret = -EEXIST;
+		goto out;
+	}
+
+	if ((client->version.major != I40EVF_CLIENT_VERSION_MAJOR) ||
+	    (client->version.minor != I40EVF_CLIENT_VERSION_MINOR)) {
+		pr_info("i40evf: Failed to register client %s due to mismatched client interface version\n",
+			client->name);
+		pr_info("Client is using version: %02d.%02d.%02d while LAN driver supports %s\n",
+			client->version.major, client->version.minor,
+			client->version.build,
+			i40evf_client_interface_version_str);
+		ret = -EIO;
+		goto out;
+	}
+
+	vf_registered_client = client;
+
+	i40evf_client_prepare(client);
+
+	pr_info("i40evf: Registered client %s with return code %d\n",
+		client->name, ret);
+out:
+	return ret;
+}
+EXPORT_SYMBOL(i40evf_register_client);
+
+/**
+ * i40evf_unregister_client - Unregister a i40e client driver with the L2 driver
+ * @client: pointer to the i40e_client struct
+ *
+ * Returns 0 on success or non-0 on error
+ **/
+int i40evf_unregister_client(struct i40e_client *client)
+{
+	int ret = 0;
+
+	/* When a unregister request comes through we would have to send
+	 * a close for each of the client instances that were opened.
+	 * client_release function is called to handle this.
+	 */
+	i40evf_client_release(client);
+
+	if (vf_registered_client != client) {
+		pr_info("i40evf: Client %s has not been registered\n",
+			client->name);
+		ret = -ENODEV;
+		goto out;
+	}
+	vf_registered_client = NULL;
+	pr_info("i40evf: Unregistered client %s\n", client->name);
+out:
+	return ret;
+}
+EXPORT_SYMBOL(i40evf_unregister_client);

drivers/net/ethernet/intel/i40evf/i40evf_client.h

+#ifndef _I40E_CLIENT_H_
+#define _I40E_CLIENT_H_
+
+#define I40EVF_CLIENT_STR_LENGTH 10
+
+/* Client interface version should be updated anytime there is a change in the
+ * existing APIs or data structures.
+ */
+#define I40EVF_CLIENT_VERSION_MAJOR 0
+#define I40EVF_CLIENT_VERSION_MINOR 01
+#define I40EVF_CLIENT_VERSION_BUILD 00
+#define I40EVF_CLIENT_VERSION_STR     \
+	__stringify(I40EVF_CLIENT_VERSION_MAJOR) "." \
+	__stringify(I40EVF_CLIENT_VERSION_MINOR) "." \
+	__stringify(I40EVF_CLIENT_VERSION_BUILD)
+
+struct i40e_client_version {
+	u8 major;
+	u8 minor;
+	u8 build;
+	u8 rsvd;
+};
+
+enum i40e_client_state {
+	__I40E_CLIENT_NULL,
+	__I40E_CLIENT_REGISTERED
+};
+
+enum i40e_client_instance_state {
+	__I40E_CLIENT_INSTANCE_NONE,
+	__I40E_CLIENT_INSTANCE_OPENED,
+};
+
+struct i40e_ops;
+struct i40e_client;
+
+/* HW does not define a type value for AEQ; only for RX/TX and CEQ.
+ * In order for us to keep the interface simple, SW will define a
+ * unique type value for AEQ.
+ */
+#define I40E_QUEUE_TYPE_PE_AEQ  0x80
+#define I40E_QUEUE_INVALID_IDX	0xFFFF
+
+struct i40e_qv_info {
+	u32 v_idx; /* msix_vector */
+	u16 ceq_idx;
+	u16 aeq_idx;
+	u8 itr_idx;
+};
+
+struct i40e_qvlist_info {
+	u32 num_vectors;
+	struct i40e_qv_info qv_info[1];
+};
+
+#define I40E_CLIENT_MSIX_ALL 0xFFFFFFFF
+
+/* set of LAN parameters useful for clients managed by LAN */
+
+/* Struct to hold per priority info */
+struct i40e_prio_qos_params {
+	u16 qs_handle; /* qs handle for prio */
+	u8 tc; /* TC mapped to prio */
+	u8 reserved;
+};
+
+#define I40E_CLIENT_MAX_USER_PRIORITY        8
+/* Struct to hold Client QoS */
+struct i40e_qos_params {
+	struct i40e_prio_qos_params prio_qos[I40E_CLIENT_MAX_USER_PRIORITY];
+};
+
+struct i40e_params {
+	struct i40e_qos_params qos;
+	u16 mtu;
+	u16 link_up; /* boolean */
+};
+
+/* Structure to hold LAN device info for a client device */
+struct i40e_info {
+	struct i40e_client_version version;
+	u8 lanmac[6];
+	struct net_device *netdev;
+	struct pci_dev *pcidev;
+	u8 __iomem *hw_addr;
+	u8 fid;	/* function id, PF id or VF id */
+#define I40E_CLIENT_FTYPE_PF 0
+#define I40E_CLIENT_FTYPE_VF 1
+	u8 ftype; /* function type, PF or VF */
+	void *vf; /* cast to i40evf_adapter */
+
+	/* All L2 params that could change during the life span of the device
+	 * and needs to be communicated to the client when they change
+	 */
+	struct i40e_params params;
+	struct i40e_ops *ops;
+
+	u16 msix_count;	 /* number of msix vectors*/
+	/* Array down below will be dynamically allocated based on msix_count */
+	struct msix_entry *msix_entries;
+	u16 itr_index; /* Which ITR index the PE driver is suppose to use */
+};
+
+struct i40e_ops {
+	/* setup_q_vector_list enables queues with a particular vector */
+	int (*setup_qvlist)(struct i40e_info *ldev, struct i40e_client *client,
+			    struct i40e_qvlist_info *qv_info);
+
+	u32 (*virtchnl_send)(struct i40e_info *ldev, struct i40e_client *client,
+			     u8 *msg, u16 len);
+
+	/* If the PE Engine is unresponsive, RDMA driver can request a reset.*/
+	void (*request_reset)(struct i40e_info *ldev,
+			      struct i40e_client *client);
+};
+
+struct i40e_client_ops {
+	/* Should be called from register_client() or whenever the driver is
+	 * ready to create a specific client instance.
+	 */
+	int (*open)(struct i40e_info *ldev, struct i40e_client *client);
+
+	/* Should be closed when netdev is unavailable or when unregister
+	 * call comes in. If the close happens due to a reset, set the reset
+	 * bit to true.
+	 */
+	void (*close)(struct i40e_info *ldev, struct i40e_client *client,
+		      bool reset);
+
+	/* called when some l2 managed parameters changes - mss */
+	void (*l2_param_change)(struct i40e_info *ldev,
+				struct i40e_client *client,
+				struct i40e_params *params);
+
+	/* called when a message is received from the PF */
+	int (*virtchnl_receive)(struct i40e_info *ldev,
+				struct i40e_client *client,
+				u8 *msg, u16 len);
+};
+
+/* Client device */
+struct i40e_client_instance {
+	struct list_head list;
+	struct i40e_info lan_info;
+	struct i40e_client *client;
+	unsigned long  state;
+};
+
+struct i40e_client {
+	struct list_head list;		/* list of registered clients */
+	char name[I40EVF_CLIENT_STR_LENGTH];
+	struct i40e_client_version version;
+	unsigned long state;		/* client state */
+	atomic_t ref_cnt;  /* Count of all the client devices of this kind */
+	u32 flags;
+#define I40E_CLIENT_FLAGS_LAUNCH_ON_PROBE	BIT(0)
+#define I40E_TX_FLAGS_NOTIFY_OTHER_EVENTS	BIT(2)
+	u8 type;
+#define I40E_CLIENT_IWARP 0
+	struct i40e_client_ops *ops;	/* client ops provided by the client */
+};
+
+/* used by clients */
+int i40evf_register_client(struct i40e_client *client);
+int i40evf_unregister_client(struct i40e_client *client);
+#endif /* _I40E_CLIENT_H_ */

drivers/net/ethernet/intel/i40evf/i40evf_main.c

@@ -26,6 +26,7 @@
 
 #include "i40evf.h"
 #include "i40e_prototype.h"
+#include "i40evf_client.h"
 static int i40evf_setup_all_tx_resources(struct i40evf_adapter *adapter);
 static int i40evf_setup_all_rx_resources(struct i40evf_adapter *adapter);
 static int i40evf_close(struct net_device *netdev);
@@ -1058,6 +1059,8 @@ static void i40evf_up_complete(struct i40evf_adapter *adapter)
 	i40evf_napi_enable_all(adapter);
 
 	adapter->aq_required |= I40EVF_FLAG_AQ_ENABLE_QUEUES;
+	if (CLIENT_ENABLED(adapter))
+		adapter->flags |= I40EVF_FLAG_CLIENT_NEEDS_OPEN;
 	mod_timer_pending(&adapter->watchdog_timer, jiffies + 1);
 }
 
@@ -1685,6 +1688,7 @@ static void i40evf_watchdog_task(struct work_struct *work)
 		i40evf_set_promiscuous(adapter, 0);
 		goto watchdog_done;
 	}
+	schedule_delayed_work(&adapter->client_task, msecs_to_jiffies(5));
 
 	if (adapter->state == __I40EVF_RUNNING)
 		i40evf_request_stats(adapter);
@@ -1773,10 +1777,17 @@ static void i40evf_reset_task(struct work_struct *work)
 	u32 reg_val;
 	int i = 0, err;
 
-	while (test_and_set_bit(__I40EVF_IN_CRITICAL_TASK,
+	while (test_and_set_bit(__I40EVF_IN_CLIENT_TASK,
 				&adapter->crit_section))
 		usleep_range(500, 1000);
-
+	if (CLIENT_ENABLED(adapter)) {
+		adapter->flags &= ~(I40EVF_FLAG_CLIENT_NEEDS_OPEN |
+				    I40EVF_FLAG_CLIENT_NEEDS_CLOSE |
+				    I40EVF_FLAG_CLIENT_NEEDS_L2_PARAMS |
+				    I40EVF_FLAG_SERVICE_CLIENT_REQUESTED);
+		cancel_delayed_work_sync(&adapter->client_task);
+		i40evf_notify_client_close(&adapter->vsi, true);
+	}
 	i40evf_misc_irq_disable(adapter);
 	if (adapter->flags & I40EVF_FLAG_RESET_NEEDED) {
 		adapter->flags &= ~I40EVF_FLAG_RESET_NEEDED;
@@ -1819,6 +1830,7 @@ static void i40evf_reset_task(struct work_struct *work)
 		dev_err(&adapter->pdev->dev, "Reset never finished (%x)\n",
 			reg_val);
 		i40evf_disable_vf(adapter);
+		clear_bit(__I40EVF_IN_CLIENT_TASK, &adapter->crit_section);
 		return; /* Do not attempt to reinit. It's dead, Jim. */
 	}
 
@@ -1861,9 +1873,8 @@ static void i40evf_reset_task(struct work_struct *work)
 	}
 	adapter->aq_required |= I40EVF_FLAG_AQ_ADD_MAC_FILTER;
 	adapter->aq_required |= I40EVF_FLAG_AQ_ADD_VLAN_FILTER;
-	/* Open RDMA Client again */
-	adapter->aq_required |= I40EVF_FLAG_SERVICE_CLIENT_REQUESTED;
 	clear_bit(__I40EVF_IN_CRITICAL_TASK, &adapter->crit_section);
+	clear_bit(__I40EVF_IN_CLIENT_TASK, &adapter->crit_section);
 	i40evf_misc_irq_enable(adapter);
 
 	mod_timer(&adapter->watchdog_timer, jiffies + 2);
@@ -1979,6 +1990,48 @@ static void i40evf_adminq_task(struct work_struct *work)
 	i40evf_misc_irq_enable(adapter);
 }
 
+/**
+ * i40evf_client_task - worker thread to perform client work
+ * @work: pointer to work_struct containing our data
+ *
+ * This task handles client interactions. Because client calls can be
+ * reentrant, we can't handle them in the watchdog.
+ **/
+static void i40evf_client_task(struct work_struct *work)
+{
+	struct i40evf_adapter *adapter =
+		container_of(work, struct i40evf_adapter, client_task.work);
+
+	/* If we can't get the client bit, just give up. We'll be rescheduled
+	 * later.
+	 */
+
+	if (test_and_set_bit(__I40EVF_IN_CLIENT_TASK, &adapter->crit_section))
+		return;
+
+	if (adapter->flags & I40EVF_FLAG_SERVICE_CLIENT_REQUESTED) {
+		i40evf_client_subtask(adapter);
+		adapter->flags &= ~I40EVF_FLAG_SERVICE_CLIENT_REQUESTED;
+		goto out;
+	}
+	if (adapter->flags & I40EVF_FLAG_CLIENT_NEEDS_CLOSE) {
+		i40evf_notify_client_close(&adapter->vsi, false);
+		adapter->flags &= ~I40EVF_FLAG_CLIENT_NEEDS_CLOSE;
+		goto out;
+	}
+	if (adapter->flags & I40EVF_FLAG_CLIENT_NEEDS_OPEN) {
+		i40evf_notify_client_open(&adapter->vsi);
+		adapter->flags &= ~I40EVF_FLAG_CLIENT_NEEDS_OPEN;
+		goto out;
+	}
+	if (adapter->flags & I40EVF_FLAG_CLIENT_NEEDS_L2_PARAMS) {
+		i40evf_notify_client_l2_params(&adapter->vsi);
+		adapter->flags &= ~I40EVF_FLAG_CLIENT_NEEDS_L2_PARAMS;
+	}
+out:
+	clear_bit(__I40EVF_IN_CLIENT_TASK, &adapter->crit_section);
+}
+
 /**
  * i40evf_free_all_tx_resources - Free Tx Resources for All Queues
  * @adapter: board private structure
@@ -2148,6 +2201,8 @@ static int i40evf_close(struct net_device *netdev)
 
 
 	set_bit(__I40E_DOWN, &adapter->vsi.state);
+	if (CLIENT_ENABLED(adapter))
+		adapter->flags |= I40EVF_FLAG_CLIENT_NEEDS_CLOSE;
 
 	i40evf_down(adapter);
 	adapter->state = __I40EVF_DOWN_PENDING;
@@ -2188,6 +2243,10 @@ static int i40evf_change_mtu(struct net_device *netdev, int new_mtu)
 	struct i40evf_adapter *adapter = netdev_priv(netdev);
 
 	netdev->mtu = new_mtu;
+	if (CLIENT_ENABLED(adapter)) {
+		i40evf_notify_client_l2_params(&adapter->vsi);
+		adapter->flags |= I40EVF_FLAG_SERVICE_CLIENT_REQUESTED;
+	}
 	adapter->flags |= I40EVF_FLAG_RESET_NEEDED;
 	schedule_work(&adapter->reset_task);
 
@@ -2581,6 +2640,12 @@ static void i40evf_init_task(struct work_struct *work)
 	adapter->netdev_registered = true;
 
 	netif_tx_stop_all_queues(netdev);
+	if (CLIENT_ALLOWED(adapter)) {
+		err = i40evf_lan_add_device(adapter);
+		if (err)
+			dev_info(&pdev->dev, "Failed to add VF to client API service list: %d\n",
+				 err);
+	}
 
 	dev_info(&pdev->dev, "MAC address: %pM\n", adapter->hw.mac.addr);
 	if (netdev->features & NETIF_F_GRO)
@@ -2745,6 +2810,7 @@ static int i40evf_probe(struct pci_dev *pdev, const struct pci_device_id *ent)
 	INIT_WORK(&adapter->reset_task, i40evf_reset_task);
 	INIT_WORK(&adapter->adminq_task, i40evf_adminq_task);
 	INIT_WORK(&adapter->watchdog_task, i40evf_watchdog_task);
+	INIT_DELAYED_WORK(&adapter->client_task, i40evf_client_task);
 	INIT_DELAYED_WORK(&adapter->init_task, i40evf_init_task);
 	schedule_delayed_work(&adapter->init_task,
 			      msecs_to_jiffies(5 * (pdev->devfn & 0x07)));
@@ -2857,14 +2923,21 @@ static void i40evf_remove(struct pci_dev *pdev)
 	struct i40evf_adapter *adapter = netdev_priv(netdev);
 	struct i40evf_mac_filter *f, *ftmp;
 	struct i40e_hw *hw = &adapter->hw;
+	int err;
 
 	cancel_delayed_work_sync(&adapter->init_task);
 	cancel_work_sync(&adapter->reset_task);
-
+	cancel_delayed_work_sync(&adapter->client_task);
 	if (adapter->netdev_registered) {
 		unregister_netdev(netdev);
 		adapter->netdev_registered = false;
 	}
+	if (CLIENT_ALLOWED(adapter)) {
+		err = i40evf_lan_del_device(adapter);
+		if (err)
+			dev_warn(&pdev->dev, "Failed to delete client device: %d\n",
+				 err);
+	}
 
 	/* Shut down all the garbage mashers on the detention level */
 	adapter->state = __I40EVF_REMOVE;

drivers/net/ethernet/intel/i40evf/i40evf_virtchnl.c

@@ -26,6 +26,7 @@
 
 #include "i40evf.h"
 #include "i40e_prototype.h"
+#include "i40evf_client.h"
 
 /* busy wait delay in msec */
 #define I40EVF_BUSY_WAIT_DELAY 10
@@ -999,6 +1000,16 @@ void i40evf_virtchnl_completion(struct i40evf_adapter *adapter,
 		if (v_opcode != adapter->current_op)
 			return;
 		break;
+	case I40E_VIRTCHNL_OP_IWARP:
+		/* Gobble zero-length replies from the PF. They indicate that
+		 * a previous message was received OK, and the client doesn't
+		 * care about that.
+		 */
+		if (msglen && CLIENT_ENABLED(adapter))
+			i40evf_notify_client_message(&adapter->vsi,
+						     msg, msglen);
+		break;
+
 	case I40E_VIRTCHNL_OP_CONFIG_IWARP_IRQ_MAP:
 		adapter->client_pending &=
 				~(BIT(I40E_VIRTCHNL_OP_CONFIG_IWARP_IRQ_MAP));
@@ -1014,7 +1025,7 @@ void i40evf_virtchnl_completion(struct i40evf_adapter *adapter,
 		}
 		break;
 	default:
-		if (v_opcode != adapter->current_op)
+		if (adapter->current_op && (v_opcode != adapter->current_op))
 			dev_warn(&adapter->pdev->dev, "Expected response %d from PF, received %d\n",
 				 adapter->current_op, v_opcode);
 		break;