mlx4: Add alias_guid mechanism

For IB ports, we paravirtualize the GUID at index 0 on slaves.  The
GUID at index 0 seen by a slave is the actual GUID occupying the GUID
table at the slave-id index.

The driver, by default, requests at startup time that subnet manager
populate its entire guid table with GUIDs. These guids are then mapped
(paravirtualized) to the slaves, and appear for each slave as its GUID
at index 0.

Until each slave has such a guid, its port status is DOWN.

The guid table is cached to support special QP paravirtualization, and
event propagation to slaves on guid change (we test to see if the guid
really changed before propagating an event to the slave).

To support this caching, add capability to __mlx4_ib_query_gid() to
obtain the network view (i.e., physical view) gid at index X, not just
the host (paravirtualized) view.

Based on a patch from Erez Shitrit <erezsh@mellanox.com>
Signed-off-by: Jack Morgenstein <jackm@dev.mellanox.co.il>
Signed-off-by: Roland Dreier <roland@purestorage.com>

drivers/infiniband/hw/mlx4/Makefile

 obj-$(CONFIG_MLX4_INFINIBAND)	+= mlx4_ib.o
 
-mlx4_ib-y :=	ah.o cq.o doorbell.o mad.o main.o mr.o qp.o srq.o mcg.o cm.o
+mlx4_ib-y :=	ah.o cq.o doorbell.o mad.o main.o mr.o qp.o srq.o mcg.o cm.o alias_GUID.o

drivers/infiniband/hw/mlx4/mad.c

@@ -791,8 +791,10 @@ void mlx4_ib_mad_cleanup(struct mlx4_ib_dev *dev)
 
 static void handle_client_rereg_event(struct mlx4_ib_dev *dev, u8 port_num)
 {
-	/* re-configure the mcg's */
+	/* re-configure the alias-guid and mcg's */
 	if (mlx4_is_master(dev->dev)) {
+		mlx4_ib_invalidate_all_guid_record(dev, port_num);
+
 		if (!dev->sriov.is_going_down)
 			mlx4_ib_mcg_port_cleanup(&dev->sriov.demux[port_num - 1], 0);
 	}
@@ -1808,9 +1810,20 @@ int mlx4_ib_init_sriov(struct mlx4_ib_dev *dev)
 		return 0;
 	}
 
+	err = mlx4_ib_init_alias_guid_service(dev);
+	if (err) {
+		mlx4_ib_warn(&dev->ib_dev, "Failed init alias guid process.\n");
+		goto paravirt_err;
+	}
+
 	mlx4_ib_warn(&dev->ib_dev, "initializing demux service for %d qp1 clients\n",
 		     dev->dev->caps.sqp_demux);
 	for (i = 0; i < dev->num_ports; i++) {
+		union ib_gid gid;
+		err = __mlx4_ib_query_gid(&dev->ib_dev, i + 1, 0, &gid, 1);
+		if (err)
+			goto demux_err;
+		dev->sriov.demux[i].guid_cache[0] = gid.global.interface_id;
 		err = alloc_pv_object(dev, mlx4_master_func_num(dev->dev), i + 1,
 				      &dev->sriov.sqps[i]);
 		if (err)
@@ -1828,6 +1841,9 @@ int mlx4_ib_init_sriov(struct mlx4_ib_dev *dev)
 		mlx4_ib_free_demux_ctx(&dev->sriov.demux[i]);
 		--i;
 	}
+	mlx4_ib_destroy_alias_guid_service(dev);
+
+paravirt_err:
 	mlx4_ib_cm_paravirt_clean(dev, -1);
 
 	return err;
@@ -1854,5 +1870,6 @@ void mlx4_ib_close_sriov(struct mlx4_ib_dev *dev)
 		}
 
 		mlx4_ib_cm_paravirt_clean(dev, -1);
+		mlx4_ib_destroy_alias_guid_service(dev);
 	}
 }

drivers/infiniband/hw/mlx4/main.c

@@ -59,6 +59,10 @@ MODULE_DESCRIPTION("Mellanox ConnectX HCA InfiniBand driver");
 MODULE_LICENSE("Dual BSD/GPL");
 MODULE_VERSION(DRV_VERSION);
 
+int mlx4_ib_sm_guid_assign = 1;
+module_param_named(sm_guid_assign, mlx4_ib_sm_guid_assign, int, 0444);
+MODULE_PARM_DESC(sm_guid_assign, "Enable SM alias_GUID assignment if sm_guid_assign > 0 (Default: 1)");
+
 static const char mlx4_ib_version[] =
 	DRV_NAME ": Mellanox ConnectX InfiniBand driver v"
 	DRV_VERSION " (" DRV_RELDATE ")\n";
@@ -349,12 +353,15 @@ static int mlx4_ib_query_port(struct ib_device *ibdev, u8 port,
 	return __mlx4_ib_query_port(ibdev, port, props, 0);
 }
 
-static int __mlx4_ib_query_gid(struct ib_device *ibdev, u8 port, int index,
-			       union ib_gid *gid)
+int __mlx4_ib_query_gid(struct ib_device *ibdev, u8 port, int index,
+			union ib_gid *gid, int netw_view)
 {
 	struct ib_smp *in_mad  = NULL;
 	struct ib_smp *out_mad = NULL;
 	int err = -ENOMEM;
+	struct mlx4_ib_dev *dev = to_mdev(ibdev);
+	int clear = 0;
+	int mad_ifc_flags = MLX4_MAD_IFC_IGNORE_KEYS;
 
 	in_mad  = kzalloc(sizeof *in_mad, GFP_KERNEL);
 	out_mad = kmalloc(sizeof *out_mad, GFP_KERNEL);
@@ -365,18 +372,29 @@ static int __mlx4_ib_query_gid(struct ib_device *ibdev, u8 port, int index,
 	in_mad->attr_id  = IB_SMP_ATTR_PORT_INFO;
 	in_mad->attr_mod = cpu_to_be32(port);
 
-	err = mlx4_MAD_IFC(to_mdev(ibdev), MLX4_MAD_IFC_IGNORE_KEYS, port,
-			   NULL, NULL, in_mad, out_mad);
+	if (mlx4_is_mfunc(dev->dev) && netw_view)
+		mad_ifc_flags |= MLX4_MAD_IFC_NET_VIEW;
+
+	err = mlx4_MAD_IFC(dev, mad_ifc_flags, port, NULL, NULL, in_mad, out_mad);
 	if (err)
 		goto out;
 
 	memcpy(gid->raw, out_mad->data + 8, 8);
 
+	if (mlx4_is_mfunc(dev->dev) && !netw_view) {
+		if (index) {
+			/* For any index > 0, return the null guid */
+			err = 0;
+			clear = 1;
+			goto out;
+		}
+	}
+
 	init_query_mad(in_mad);
 	in_mad->attr_id  = IB_SMP_ATTR_GUID_INFO;
 	in_mad->attr_mod = cpu_to_be32(index / 8);
 
-	err = mlx4_MAD_IFC(to_mdev(ibdev), MLX4_MAD_IFC_IGNORE_KEYS, port,
+	err = mlx4_MAD_IFC(dev, mad_ifc_flags, port,
 			   NULL, NULL, in_mad, out_mad);
 	if (err)
 		goto out;
@@ -384,6 +402,8 @@ static int __mlx4_ib_query_gid(struct ib_device *ibdev, u8 port, int index,
 	memcpy(gid->raw + 8, out_mad->data + (index % 8) * 8, 8);
 
 out:
+	if (clear)
+		memset(gid->raw + 8, 0, 8);
 	kfree(in_mad);
 	kfree(out_mad);
 	return err;
@@ -403,7 +423,7 @@ static int mlx4_ib_query_gid(struct ib_device *ibdev, u8 port, int index,
 			     union ib_gid *gid)
 {
 	if (rdma_port_get_link_layer(ibdev, port) == IB_LINK_LAYER_INFINIBAND)
-		return __mlx4_ib_query_gid(ibdev, port, index, gid);
+		return __mlx4_ib_query_gid(ibdev, port, index, gid, 0);
 	else
 		return iboe_query_gid(ibdev, port, index, gid);
 }
@@ -1566,6 +1586,11 @@ static void mlx4_ib_event(struct mlx4_dev *dev, void *ibdev_ptr,
 	case MLX4_DEV_EVENT_PORT_UP:
 		if (p > ibdev->num_ports)
 			return;
+		if (mlx4_is_master(dev) &&
+		    rdma_port_get_link_layer(&ibdev->ib_dev, p) ==
+			IB_LINK_LAYER_INFINIBAND) {
+			mlx4_ib_invalidate_all_guid_record(ibdev, p);
+		}
 		ibev.event = IB_EVENT_PORT_ACTIVE;
 		break;
 

drivers/infiniband/hw/mlx4/mlx4_ib.h

@@ -42,6 +42,7 @@
 #include <rdma/ib_verbs.h>
 #include <rdma/ib_umem.h>
 #include <rdma/ib_mad.h>
+#include <rdma/ib_sa.h>
 
 #include <linux/mlx4/device.h>
 #include <linux/mlx4/doorbell.h>
@@ -64,6 +65,9 @@ enum {
 #define MLX4_IB_SQ_HEADROOM(shift)	((MLX4_IB_MAX_HEADROOM >> (shift)) + 1)
 #define MLX4_IB_SQ_MAX_SPARE		(MLX4_IB_SQ_HEADROOM(MLX4_IB_SQ_MIN_WQE_SHIFT))
 
+/*module param to indicate if SM assigns the alias_GUID*/
+extern int mlx4_ib_sm_guid_assign;
+
 struct mlx4_ib_ucontext {
 	struct ib_ucontext	ibucontext;
 	struct mlx4_uar		uar;
@@ -277,6 +281,57 @@ struct mlx4_ib_ah {
 	union mlx4_ext_av       av;
 };
 
+/****************************************/
+/* alias guid support */
+/****************************************/
+#define NUM_PORT_ALIAS_GUID		2
+#define NUM_ALIAS_GUID_IN_REC		8
+#define NUM_ALIAS_GUID_REC_IN_PORT	16
+#define GUID_REC_SIZE			8
+#define NUM_ALIAS_GUID_PER_PORT		128
+#define MLX4_NOT_SET_GUID		(0x00LL)
+#define MLX4_GUID_FOR_DELETE_VAL	(~(0x00LL))
+
+enum mlx4_guid_alias_rec_status {
+	MLX4_GUID_INFO_STATUS_IDLE,
+	MLX4_GUID_INFO_STATUS_SET,
+	MLX4_GUID_INFO_STATUS_PENDING,
+};
+
+enum mlx4_guid_alias_rec_ownership {
+	MLX4_GUID_DRIVER_ASSIGN,
+	MLX4_GUID_SYSADMIN_ASSIGN,
+	MLX4_GUID_NONE_ASSIGN, /*init state of each record*/
+};
+
+enum mlx4_guid_alias_rec_method {
+	MLX4_GUID_INFO_RECORD_SET	= IB_MGMT_METHOD_SET,
+	MLX4_GUID_INFO_RECORD_DELETE	= IB_SA_METHOD_DELETE,
+};
+
+struct mlx4_sriov_alias_guid_info_rec_det {
+	u8 all_recs[GUID_REC_SIZE * NUM_ALIAS_GUID_IN_REC];
+	ib_sa_comp_mask guid_indexes; /*indicates what from the 8 records are valid*/
+	enum mlx4_guid_alias_rec_status status; /*indicates the administraively status of the record.*/
+	u8 method; /*set or delete*/
+	enum mlx4_guid_alias_rec_ownership ownership; /*indicates who assign that alias_guid record*/
+};
+
+struct mlx4_sriov_alias_guid_port_rec_det {
+	struct mlx4_sriov_alias_guid_info_rec_det all_rec_per_port[NUM_ALIAS_GUID_REC_IN_PORT];
+	struct workqueue_struct *wq;
+	struct delayed_work alias_guid_work;
+	u8 port;
+	struct mlx4_sriov_alias_guid *parent;
+	struct list_head cb_list;
+};
+
+struct mlx4_sriov_alias_guid {
+	struct mlx4_sriov_alias_guid_port_rec_det ports_guid[MLX4_MAX_PORTS];
+	spinlock_t ag_work_lock;
+	struct ib_sa_client *sa_client;
+};
+
 struct mlx4_ib_demux_work {
 	struct work_struct	work;
 	struct mlx4_ib_dev     *dev;
@@ -349,6 +404,8 @@ struct mlx4_ib_sriov {
 	spinlock_t going_down_lock;
 	int is_going_down;
 
+	struct mlx4_sriov_alias_guid alias_guid;
+
 	/* CM paravirtualization fields */
 	struct list_head cm_list;
 	spinlock_t id_map_lock;
@@ -555,6 +612,9 @@ int __mlx4_ib_query_port(struct ib_device *ibdev, u8 port,
 int __mlx4_ib_query_pkey(struct ib_device *ibdev, u8 port, u16 index,
 			 u16 *pkey, int netw_view);
 
+int __mlx4_ib_query_gid(struct ib_device *ibdev, u8 port, int index,
+			union ib_gid *gid, int netw_view);
+
 int mlx4_ib_resolve_grh(struct mlx4_ib_dev *dev, const struct ib_ah_attr *ah_attr,
 			u8 *mac, int *is_mcast, u8 port);
 
@@ -606,4 +666,18 @@ int mlx4_ib_multiplex_cm_handler(struct ib_device *ibdev, int port, int slave_id
 void mlx4_ib_cm_paravirt_init(struct mlx4_ib_dev *dev);
 void mlx4_ib_cm_paravirt_clean(struct mlx4_ib_dev *dev, int slave_id);
 
+/* alias guid support */
+void mlx4_ib_init_alias_guid_work(struct mlx4_ib_dev *dev, int port);
+int mlx4_ib_init_alias_guid_service(struct mlx4_ib_dev *dev);
+void mlx4_ib_destroy_alias_guid_service(struct mlx4_ib_dev *dev);
+void mlx4_ib_invalidate_all_guid_record(struct mlx4_ib_dev *dev, int port);
+
+void mlx4_ib_notify_slaves_on_guid_change(struct mlx4_ib_dev *dev,
+					  int block_num,
+					  u8 port_num, u8 *p_data);
+
+void mlx4_ib_update_cache_on_guid_change(struct mlx4_ib_dev *dev,
+					 int block_num, u8 port_num,
+					 u8 *p_data);
+
 #endif /* MLX4_IB_H */

drivers/net/ethernet/mellanox/mlx4/cmd.c

@@ -674,8 +674,10 @@ static int get_full_pkey_table(struct mlx4_dev *dev, u8 port, u16 *table,
 
 static enum ib_port_state vf_port_state(struct mlx4_dev *dev, int port, int vf)
 {
-	/* will be modified when add alias_guid feature */
-	return IB_PORT_DOWN;
+	if (mlx4_get_slave_port_state(dev, vf, port) == SLAVE_PORT_UP)
+		return IB_PORT_ACTIVE;
+	else
+		return IB_PORT_DOWN;
 }
 
 static int mlx4_MAD_IFC_wrapper(struct mlx4_dev *dev, int slave,