xen/pvcalls: implement bind command
authorStefano Stabellini <sstabellini@kernel.org>
Mon, 30 Oct 2017 22:40:56 +0000 (15:40 -0700)
committerBoris Ostrovsky <boris.ostrovsky@oracle.com>
Tue, 31 Oct 2017 13:05:53 +0000 (09:05 -0400)
Send PVCALLS_BIND to the backend. Introduce a new structure, part of
struct sock_mapping, to store information specific to passive sockets.

Introduce a status field to keep track of the status of the passive
socket.

Signed-off-by: Stefano Stabellini <stefano@aporeto.com>
Reviewed-by: Boris Ostrovsky <boris.ostrovsky@oracle.com>
CC: boris.ostrovsky@oracle.com
CC: jgross@suse.com
Signed-off-by: Boris Ostrovsky <boris.ostrovsky@oracle.com>
drivers/xen/pvcalls-front.c
drivers/xen/pvcalls-front.h

index 8d4a43e6aa4628aefa56655e988e33a783a9260b..6e70ec3ea796d781455eff2e1030a0dd98b881d4 100644 (file)
@@ -70,6 +70,13 @@ struct sock_mapping {
 
                        wait_queue_head_t inflight_conn_req;
                } active;
+               struct {
+               /* Socket status */
+#define PVCALLS_STATUS_UNINITALIZED  0
+#define PVCALLS_STATUS_BIND          1
+#define PVCALLS_STATUS_LISTEN        2
+                       uint8_t status;
+               } passive;
        };
 };
 
@@ -346,6 +353,65 @@ int pvcalls_front_connect(struct socket *sock, struct sockaddr *addr,
        return ret;
 }
 
+int pvcalls_front_bind(struct socket *sock, struct sockaddr *addr, int addr_len)
+{
+       struct pvcalls_bedata *bedata;
+       struct sock_mapping *map = NULL;
+       struct xen_pvcalls_request *req;
+       int notify, req_id, ret;
+
+       if (addr->sa_family != AF_INET || sock->type != SOCK_STREAM)
+               return -EOPNOTSUPP;
+
+       pvcalls_enter();
+       if (!pvcalls_front_dev) {
+               pvcalls_exit();
+               return -ENOTCONN;
+       }
+       bedata = dev_get_drvdata(&pvcalls_front_dev->dev);
+
+       map = (struct sock_mapping *) sock->sk->sk_send_head;
+       if (map == NULL) {
+               pvcalls_exit();
+               return -ENOTSOCK;
+       }
+
+       spin_lock(&bedata->socket_lock);
+       ret = get_request(bedata, &req_id);
+       if (ret < 0) {
+               spin_unlock(&bedata->socket_lock);
+               pvcalls_exit();
+               return ret;
+       }
+       req = RING_GET_REQUEST(&bedata->ring, req_id);
+       req->req_id = req_id;
+       map->sock = sock;
+       req->cmd = PVCALLS_BIND;
+       req->u.bind.id = (uintptr_t)map;
+       memcpy(req->u.bind.addr, addr, sizeof(*addr));
+       req->u.bind.len = addr_len;
+
+       map->active_socket = false;
+
+       bedata->ring.req_prod_pvt++;
+       RING_PUSH_REQUESTS_AND_CHECK_NOTIFY(&bedata->ring, notify);
+       spin_unlock(&bedata->socket_lock);
+       if (notify)
+               notify_remote_via_irq(bedata->irq);
+
+       wait_event(bedata->inflight_req,
+                  READ_ONCE(bedata->rsp[req_id].req_id) == req_id);
+
+       /* read req_id, then the content */
+       smp_rmb();
+       ret = bedata->rsp[req_id].ret;
+       bedata->rsp[req_id].req_id = PVCALLS_INVALID_ID;
+
+       map->passive.status = PVCALLS_STATUS_BIND;
+       pvcalls_exit();
+       return 0;
+}
+
 static const struct xenbus_device_id pvcalls_front_ids[] = {
        { "pvcalls" },
        { "" }
index 63b0417c31d37233c46837a9cc3c5ba931fcf17d..8b0a2749d4b8fe95aa5447d240dbda8ae49da014 100644 (file)
@@ -6,5 +6,8 @@
 int pvcalls_front_socket(struct socket *sock);
 int pvcalls_front_connect(struct socket *sock, struct sockaddr *addr,
                          int addr_len, int flags);
+int pvcalls_front_bind(struct socket *sock,
+                      struct sockaddr *addr,
+                      int addr_len);
 
 #endif