IB/usnic: Add UDP support in u*verbs.c, u*main.c and u*util.h

Add supports for:
	1) Parsing the socket file descriptor pass down from userspace.
	2) IP notifiers
	3) Encoding the IP in the GID
	4) Other aux. changes to support UDP

Signed-off-by: Upinder Malhi <umalhi@cisco.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>

diff --git a/drivers/infiniband/hw/usnic/usnic_common_util.h b/drivers/infiniband/hw/usnic/usnic_common_util.h
index 128550a..afd8bfa 100644 (file)
--- a/drivers/infiniband/hw/usnic/usnic_common_util.h
+++ b/drivers/infiniband/hw/usnic/usnic_common_util.h
@@ -36,6 +36,23 @@ usnic_mac_to_gid(const char *const mac, char *raw_gid)
 }
 
 static inline void
+usnic_mac_ip_to_gid(const char *const mac, const uint32_t inaddr, char *raw_gid)
+{
+       raw_gid[0] = 0xfe;
+       raw_gid[1] = 0x80;
+       memset(&raw_gid[2], 0, 2);
+       memcpy(&raw_gid[4], &inaddr, 4);
+       raw_gid[8] = mac[0]^2;
+       raw_gid[9] = mac[1];
+       raw_gid[10] = mac[2];
+       raw_gid[11] = 0xff;
+       raw_gid[12] = 0xfe;
+       raw_gid[13] = mac[3];
+       raw_gid[14] = mac[4];
+       raw_gid[15] = mac[5];
+}
+
+static inline void
 usnic_write_gid_if_id_from_mac(char *mac, char *raw_gid)
 {
        raw_gid[8] = mac[0]^2;

diff --git a/drivers/infiniband/hw/usnic/usnic_ib_main.c b/drivers/infiniband/hw/usnic/usnic_ib_main.c
index 6ab0b41..3b7e8bd 100644 (file)
--- a/drivers/infiniband/hw/usnic/usnic_ib_main.c
+++ b/drivers/infiniband/hw/usnic/usnic_ib_main.c
@@ -25,6 +25,7 @@
  */
 
 #include <linux/module.h>
+#include <linux/inetdevice.h>
 #include <linux/init.h>
 #include <linux/slab.h>
 #include <linux/errno.h>
@@ -236,13 +237,79 @@ static struct notifier_block usnic_ib_netdevice_notifier = {
 };
 /* End of netdev section */
 
+/* Start of inet section */
+static int usnic_ib_handle_inet_event(struct usnic_ib_dev *us_ibdev,
+                                       unsigned long event, void *ptr)
+{
+       struct in_ifaddr *ifa = ptr;
+       struct ib_event ib_event;
+
+       mutex_lock(&us_ibdev->usdev_lock);
+
+       switch (event) {
+       case NETDEV_DOWN:
+               usnic_info("%s via ip notifiers",
+                               usnic_ib_netdev_event_to_string(event));
+               usnic_fwd_del_ipaddr(us_ibdev->ufdev);
+               usnic_ib_qp_grp_modify_active_to_err(us_ibdev);
+               ib_event.event = IB_EVENT_GID_CHANGE;
+               ib_event.device = &us_ibdev->ib_dev;
+               ib_event.element.port_num = 1;
+               ib_dispatch_event(&ib_event);
+               break;
+       case NETDEV_UP:
+               usnic_fwd_add_ipaddr(us_ibdev->ufdev, ifa->ifa_address);
+               usnic_info("%s via ip notifiers: ip %pI4",
+                               usnic_ib_netdev_event_to_string(event),
+                               &us_ibdev->ufdev->inaddr);
+               ib_event.event = IB_EVENT_GID_CHANGE;
+               ib_event.device = &us_ibdev->ib_dev;
+               ib_event.element.port_num = 1;
+               ib_dispatch_event(&ib_event);
+               break;
+       default:
+               usnic_info("Ignorning event %s on %s",
+                               usnic_ib_netdev_event_to_string(event),
+                               us_ibdev->ib_dev.name);
+       }
+       mutex_unlock(&us_ibdev->usdev_lock);
+
+       return NOTIFY_DONE;
+}
+
+static int usnic_ib_inetaddr_event(struct notifier_block *notifier,
+                                       unsigned long event, void *ptr)
+{
+       struct usnic_ib_dev *us_ibdev;
+       struct in_ifaddr *ifa = ptr;
+       struct net_device *netdev = ifa->ifa_dev->dev;
+
+       mutex_lock(&usnic_ib_ibdev_list_lock);
+       list_for_each_entry(us_ibdev, &usnic_ib_ibdev_list, ib_dev_link) {
+               if (us_ibdev->netdev == netdev) {
+                       usnic_ib_handle_inet_event(us_ibdev, event, ptr);
+                       break;
+               }
+       }
+       mutex_unlock(&usnic_ib_ibdev_list_lock);
+
+       return NOTIFY_DONE;
+}
+static struct notifier_block usnic_ib_inetaddr_notifier = {
+       .notifier_call = usnic_ib_inetaddr_event
+};
+/* End of inet section*/
+
 /* Start of PF discovery section */
 static void *usnic_ib_device_add(struct pci_dev *dev)
 {
        struct usnic_ib_dev *us_ibdev;
        union ib_gid gid;
+       struct in_ifaddr *in;
+       struct net_device *netdev;
 
        usnic_dbg("\n");
+       netdev = pci_get_drvdata(dev);
 
        us_ibdev = (struct usnic_ib_dev *)ib_alloc_device(sizeof(*us_ibdev));
        if (IS_ERR_OR_NULL(us_ibdev)) {
@@ -326,6 +393,12 @@ static void *usnic_ib_device_add(struct pci_dev *dev)
        if (netif_carrier_ok(us_ibdev->netdev))
                usnic_fwd_carrier_up(us_ibdev->ufdev);
 
+       in = ((struct in_device *)(netdev->ip_ptr))->ifa_list;
+       if (in != NULL)
+               usnic_fwd_add_ipaddr(us_ibdev->ufdev, in->ifa_address);
+
+       usnic_mac_ip_to_gid(us_ibdev->netdev->perm_addr,
+                               us_ibdev->ufdev->inaddr, &gid.raw[0]);
        memcpy(&us_ibdev->ib_dev.node_guid, &gid.global.interface_id,
                sizeof(gid.global.interface_id));
        kref_init(&us_ibdev->vf_cnt);
@@ -555,16 +628,24 @@ static int __init usnic_ib_init(void)
                goto out_pci_unreg;
        }
 
+       err = register_inetaddr_notifier(&usnic_ib_inetaddr_notifier);
+       if (err) {
+               usnic_err("Failed to register inet addr notifier\n");
+               goto out_unreg_netdev_notifier;
+       }
+
        err = usnic_transport_init();
        if (err) {
                usnic_err("Failed to initialize transport\n");
-               goto out_unreg_netdev_notifier;
+               goto out_unreg_inetaddr_notifier;
        }
 
        usnic_debugfs_init();
 
        return 0;
 
+out_unreg_inetaddr_notifier:
+       unregister_inetaddr_notifier(&usnic_ib_inetaddr_notifier);
 out_unreg_netdev_notifier:
        unregister_netdevice_notifier(&usnic_ib_netdevice_notifier);
 out_pci_unreg:
@@ -580,6 +661,7 @@ static void __exit usnic_ib_destroy(void)
        usnic_dbg("\n");
        usnic_debugfs_exit();
        usnic_transport_fini();
+       unregister_inetaddr_notifier(&usnic_ib_inetaddr_notifier);
        unregister_netdevice_notifier(&usnic_ib_netdevice_notifier);
        pci_unregister_driver(&usnic_ib_pci_driver);
        usnic_uiom_fini();

diff --git a/drivers/infiniband/hw/usnic/usnic_ib_verbs.c b/drivers/infiniband/hw/usnic/usnic_ib_verbs.c
index 2217bc0..937113f 100644 (file)
--- a/drivers/infiniband/hw/usnic/usnic_ib_verbs.c
+++ b/drivers/infiniband/hw/usnic/usnic_ib_verbs.c
@@ -230,6 +230,15 @@ static void eth_speed_to_ib_speed(int speed, u8 *active_speed,
        }
 }
 
+static int create_qp_validate_user_data(struct usnic_ib_create_qp_cmd cmd)
+{
+       if (cmd.spec.trans_type <= USNIC_TRANSPORT_UNKNOWN ||
+                       cmd.spec.trans_type >= USNIC_TRANSPORT_MAX)
+               return -EINVAL;
+
+       return 0;
+}
+
 /* Start of ib callback functions */
 
 enum rdma_link_layer usnic_ib_port_link_layer(struct ib_device *device,
@@ -252,7 +261,8 @@ int usnic_ib_query_device(struct ib_device *ibdev,
        us_ibdev->netdev->ethtool_ops->get_drvinfo(us_ibdev->netdev, &info);
        us_ibdev->netdev->ethtool_ops->get_settings(us_ibdev->netdev, &cmd);
        memset(props, 0, sizeof(*props));
-       usnic_mac_to_gid(us_ibdev->ufdev->mac, &gid.raw[0]);
+       usnic_mac_ip_to_gid(us_ibdev->ufdev->mac, us_ibdev->ufdev->inaddr,
+                       &gid.raw[0]);
        memcpy(&props->sys_image_guid, &gid.global.interface_id,
                sizeof(gid.global.interface_id));
        usnic_ib_fw_string_to_u64(&info.fw_version[0], &props->fw_ver);
@@ -310,12 +320,15 @@ int usnic_ib_query_port(struct ib_device *ibdev, u8 port,
        props->sm_lid = 0;
        props->sm_sl = 0;
 
-       if (us_ibdev->ufdev->link_up) {
-               props->state = IB_PORT_ACTIVE;
-               props->phys_state = 5;
-       } else {
+       if (!us_ibdev->ufdev->link_up) {
                props->state = IB_PORT_DOWN;
                props->phys_state = 3;
+       } else if (!us_ibdev->ufdev->inaddr) {
+               props->state = IB_PORT_INIT;
+               props->phys_state = 4;
+       } else {
+               props->state = IB_PORT_ACTIVE;
+               props->phys_state = 5;
        }
 
        props->port_cap_flags = 0;
@@ -385,7 +398,8 @@ int usnic_ib_query_gid(struct ib_device *ibdev, u8 port, int index,
 
        mutex_lock(&us_ibdev->usdev_lock);
        memset(&(gid->raw[0]), 0, sizeof(gid->raw));
-       usnic_mac_to_gid(us_ibdev->ufdev->mac, &gid->raw[0]);
+       usnic_mac_ip_to_gid(us_ibdev->ufdev->mac, us_ibdev->ufdev->inaddr,
+                       &gid->raw[0]);
        mutex_unlock(&us_ibdev->usdev_lock);
 
        return 0;
@@ -444,6 +458,7 @@ struct ib_qp *usnic_ib_create_qp(struct ib_pd *pd,
        struct usnic_ib_ucontext *ucontext;
        int cq_cnt;
        struct usnic_vnic_res_spec res_spec;
+       struct usnic_ib_create_qp_cmd cmd;
        struct usnic_transport_spec trans_spec;
 
        usnic_dbg("\n");
@@ -451,14 +466,27 @@ struct ib_qp *usnic_ib_create_qp(struct ib_pd *pd,
        ucontext = to_uucontext(pd->uobject->context);
        us_ibdev = to_usdev(pd->device);
 
+       err = ib_copy_from_udata(&cmd, udata, sizeof(cmd));
+       if (err) {
+               usnic_err("%s: cannot copy udata for create_qp\n",
+                               us_ibdev->ib_dev.name);
+               return ERR_PTR(-EINVAL);
+       }
+
+       err = create_qp_validate_user_data(cmd);
+       if (err) {
+               usnic_err("%s: Failed to validate user data\n",
+                               us_ibdev->ib_dev.name);
+               return ERR_PTR(-EINVAL);
+       }
+
        if (init_attr->qp_type != IB_QPT_UD) {
                usnic_err("%s asked to make a non-UD QP: %d\n",
                                us_ibdev->ib_dev.name, init_attr->qp_type);
                return ERR_PTR(-EINVAL);
        }
 
-       memset(&trans_spec, 0, sizeof(trans_spec));
-       trans_spec.trans_type = USNIC_TRANSPORT_ROCE_CUSTOM;
+       trans_spec = cmd.spec;
        mutex_lock(&us_ibdev->usdev_lock);
        cq_cnt = (init_attr->send_cq == init_attr->recv_cq) ? 1 : 2;
        res_spec = min_transport_spec[trans_spec.trans_type];