void rds_hb_worker(struct work_struct *work)
{
struct rds_connection *conn = container_of(work, struct rds_connection, c_hb_w.work);
unsigned long now = get_seconds();
int ret;
if (!rds_conn_hb_timeout || conn->c_loopback)
return;
if (rds_conn_state(conn) == RDS_CONN_UP) {
if (!conn->c_hb_start) {
ret = rds_send_hb(conn, 0);
if (ret) {
rdsdebug("RDS/IB: rds_hb_worker: failed %d\n", ret);
return;
}
conn->c_hb_start = now;
} else if (now - conn->c_hb_start > rds_conn_hb_timeout) {
printk(KERN_NOTICE
"RDS/IB: connection <%u.%u.%u.%u,%u.%u.%u.%u,%d> "
"timed out (0x%lx,0x%lx)..disconnecting and reconnecting\n",
NIPQUAD(conn->c_laddr),
NIPQUAD(conn->c_faddr), conn->c_tos,
conn->c_hb_start, now);
rds_conn_drop(conn);
return;
}
queue_delayed_work(rds_wq, &conn->c_hb_w, HZ);
}
}
static int rds_tcp_accept_one(struct socket *sock)
{
struct socket *new_sock = NULL;
struct rds_connection *conn;
int ret;
struct inet_sock *inet;
ret = sock_create_lite(sock->sk->sk_family, sock->sk->sk_type,
sock->sk->sk_protocol, &new_sock);
if (ret)
goto out;
new_sock->type = sock->type;
new_sock->ops = sock->ops;
ret = sock->ops->accept(sock, new_sock, O_NONBLOCK);
if (ret < 0)
goto out;
rds_tcp_tune(new_sock);
inet = inet_sk(new_sock->sk);
rdsdebug("accepted tcp %pI4:%u -> %pI4:%u\n",
&inet->inet_saddr, ntohs(inet->inet_sport),
&inet->inet_daddr, ntohs(inet->inet_dport));
conn = rds_conn_create(inet->inet_saddr, inet->inet_daddr,
&rds_tcp_transport, GFP_KERNEL);
if (IS_ERR(conn)) {
ret = PTR_ERR(conn);
goto out;
}
/*
* see the comment above rds_queue_delayed_reconnect()
*/
if (!rds_conn_transition(conn, RDS_CONN_DOWN, RDS_CONN_CONNECTING)) {
if (rds_conn_state(conn) == RDS_CONN_UP)
rds_tcp_stats_inc(s_tcp_listen_closed_stale);
else
rds_tcp_stats_inc(s_tcp_connect_raced);
rds_conn_drop(conn);
ret = 0;
goto out;
}
rds_tcp_set_callbacks(new_sock, conn);
rds_connect_complete(conn);
new_sock = NULL;
ret = 0;
out:
if (new_sock)
sock_release(new_sock);
return ret;
}
void rds_recv_worker(struct work_struct *work)
{
struct rds_connection *conn = container_of(work, struct rds_connection, c_recv_w.work);
int ret;
if (rds_conn_state(conn) == RDS_CONN_UP) {
ret = conn->c_trans->recv(conn);
rdsdebug("conn %p ret %d\n", conn, ret);
switch (ret) {
case -EAGAIN:
rds_stats_inc(s_recv_immediate_retry);
queue_delayed_work(rds_wq, &conn->c_recv_w, 0);
break;
case -ENOMEM:
rds_stats_inc(s_recv_delayed_retry);
queue_delayed_work(rds_wq, &conn->c_recv_w, 2);
default:
break;
}
}
}
void rds_send_worker(struct work_struct *work)
{
struct rds_connection *conn = container_of(work, struct rds_connection, c_send_w.work);
int ret;
if (rds_conn_state(conn) == RDS_CONN_UP) {
clear_bit(RDS_LL_SEND_FULL, &conn->c_flags);
ret = rds_send_xmit(conn);
cond_resched();
rdsdebug("conn %p ret %d\n", conn, ret);
switch (ret) {
case -EAGAIN:
rds_stats_inc(s_send_immediate_retry);
queue_delayed_work(rds_wq, &conn->c_send_w, 0);
break;
case -ENOMEM:
rds_stats_inc(s_send_delayed_retry);
queue_delayed_work(rds_wq, &conn->c_send_w, 2);
default:
break;
}
}
}
int rds_rdma_cm_event_handler(struct rdma_cm_id *cm_id,
struct rdma_cm_event *event)
{
/* this can be null in the listening path */
struct rds_connection *conn = cm_id->context;
struct rds_transport *trans;
int ret = 0;
rdsdebug("conn %p id %p handling event %u (%s)\n", conn, cm_id,
event->event, rdma_event_msg(event->event));
if (cm_id->device->node_type == RDMA_NODE_RNIC)
trans = &rds_iw_transport;
else
trans = &rds_ib_transport;
/* Prevent shutdown from tearing down the connection
* while we're executing. */
if (conn) {
mutex_lock(&conn->c_cm_lock);
/* If the connection is being shut down, bail out
* right away. We return 0 so cm_id doesn't get
* destroyed prematurely */
if (rds_conn_state(conn) == RDS_CONN_DISCONNECTING) {
/* Reject incoming connections while we're tearing
* down an existing one. */
if (event->event == RDMA_CM_EVENT_CONNECT_REQUEST)
ret = 1;
goto out;
}
}
switch (event->event) {
case RDMA_CM_EVENT_CONNECT_REQUEST:
ret = trans->cm_handle_connect(cm_id, event);
break;
case RDMA_CM_EVENT_ADDR_RESOLVED:
/* XXX do we need to clean up if this fails? */
ret = rdma_resolve_route(cm_id,
RDS_RDMA_RESOLVE_TIMEOUT_MS);
break;
case RDMA_CM_EVENT_ROUTE_RESOLVED:
/* Connection could have been dropped so make sure the
* cm_id is valid before proceeding
*/
if (conn) {
struct rds_ib_connection *ibic;
ibic = conn->c_transport_data;
if (ibic && ibic->i_cm_id == cm_id)
ret = trans->cm_initiate_connect(cm_id);
else
rds_conn_drop(conn);
}
break;
case RDMA_CM_EVENT_ESTABLISHED:
trans->cm_connect_complete(conn, event);
break;
case RDMA_CM_EVENT_ADDR_ERROR:
case RDMA_CM_EVENT_ROUTE_ERROR:
case RDMA_CM_EVENT_CONNECT_ERROR:
case RDMA_CM_EVENT_UNREACHABLE:
case RDMA_CM_EVENT_REJECTED:
case RDMA_CM_EVENT_DEVICE_REMOVAL:
case RDMA_CM_EVENT_ADDR_CHANGE:
if (conn)
rds_conn_drop(conn);
break;
case RDMA_CM_EVENT_DISCONNECTED:
rdsdebug("DISCONNECT event - dropping connection "
"%pI4->%pI4\n", &conn->c_laddr,
&conn->c_faddr);
rds_conn_drop(conn);
break;
default:
/* things like device disconnect? */
printk(KERN_ERR "RDS: unknown event %u (%s)!\n",
event->event, rdma_event_msg(event->event));
break;
}
out:
if (conn)
mutex_unlock(&conn->c_cm_lock);
rdsdebug("id %p event %u (%s) handling ret %d\n", cm_id, event->event,
rdma_event_msg(event->event), ret);
return ret;
}
int rds_rdma_cm_event_handler(struct rdma_cm_id *cm_id,
struct rdma_cm_event *event)
{
/* this can be null in the listening path */
struct rds_connection *conn = cm_id->context;
struct rds_transport *trans;
int ret = 0;
rdsdebug("conn %p id %p handling event %u\n", conn, cm_id,
event->event);
if (cm_id->device->node_type == RDMA_NODE_RNIC)
trans = &rds_iw_transport;
else
trans = &rds_ib_transport;
/* Prevent shutdown from tearing down the connection
* while we're executing. */
if (conn) {
mutex_lock(&conn->c_cm_lock);
/* If the connection is being shut down, bail out
* right away. We return 0 so cm_id doesn't get
* destroyed prematurely */
if (rds_conn_state(conn) == RDS_CONN_DISCONNECTING) {
/* Reject incoming connections while we're tearing
* down an existing one. */
if (event->event == RDMA_CM_EVENT_CONNECT_REQUEST)
ret = 1;
goto out;
}
}
switch (event->event) {
case RDMA_CM_EVENT_CONNECT_REQUEST:
ret = trans->cm_handle_connect(cm_id, event);
break;
case RDMA_CM_EVENT_ADDR_RESOLVED:
/* XXX do we need to clean up if this fails? */
ret = rdma_resolve_route(cm_id,
RDS_RDMA_RESOLVE_TIMEOUT_MS);
break;
case RDMA_CM_EVENT_ROUTE_RESOLVED:
/* XXX worry about racing with listen acceptance */
ret = trans->cm_initiate_connect(cm_id);
break;
case RDMA_CM_EVENT_ESTABLISHED:
trans->cm_connect_complete(conn, event);
break;
case RDMA_CM_EVENT_ADDR_ERROR:
case RDMA_CM_EVENT_ROUTE_ERROR:
case RDMA_CM_EVENT_CONNECT_ERROR:
case RDMA_CM_EVENT_UNREACHABLE:
case RDMA_CM_EVENT_REJECTED:
case RDMA_CM_EVENT_DEVICE_REMOVAL:
case RDMA_CM_EVENT_ADDR_CHANGE:
if (conn)
rds_conn_drop(conn);
break;
case RDMA_CM_EVENT_DISCONNECTED:
printk(KERN_WARNING "RDS/RDMA: DISCONNECT event - dropping connection "
"%u.%u.%u.%u->%u.%u.%u.%u\n", NIPQUAD(conn->c_laddr),
NIPQUAD(conn->c_faddr));
rds_conn_drop(conn);
break;
default:
/* things like device disconnect? */
printk(KERN_ERR "unknown event %u\n", event->event);
BUG();
break;
}
out:
if (conn) {
//struct rds_iw_connection *ic = conn->c_transport_data;
/* If we return non-zero, we must to hang on to the cm_id */
//BUG_ON(ic->i_cm_id == cm_id && ret);
mutex_unlock(&conn->c_cm_lock);
}
rdsdebug("id %p event %u handling ret %d\n", cm_id, event->event, ret);
return ret;
}
