summaryrefslogtreecommitdiff
path: root/drivers/infiniband/sw/rxe/rxe_net.c
diff options
context:
space:
mode:
Diffstat (limited to 'drivers/infiniband/sw/rxe/rxe_net.c')
-rw-r--r--drivers/infiniband/sw/rxe/rxe_net.c146
1 files changed, 96 insertions, 50 deletions
diff --git a/drivers/infiniband/sw/rxe/rxe_net.c b/drivers/infiniband/sw/rxe/rxe_net.c
index cd59666158b1..0195d361e5e3 100644
--- a/drivers/infiniband/sw/rxe/rxe_net.c
+++ b/drivers/infiniband/sw/rxe/rxe_net.c
@@ -20,6 +20,54 @@
static struct rxe_recv_sockets recv_sockets;
+#ifdef CONFIG_DEBUG_LOCK_ALLOC
+/*
+ * lockdep can detect false positive circular dependencies
+ * when there are user-space socket API users or in kernel
+ * users switching between a tcp and rdma transport.
+ * Maybe also switching between siw and rxe may cause
+ * problems as per default sockets are only classified
+ * by family and not by ip protocol. And there might
+ * be different locks used between the application
+ * and the low level sockets.
+ *
+ * Problems were seen with ksmbd.ko and cifs.ko,
+ * switching transports, use git blame to find
+ * more details.
+ */
+static struct lock_class_key rxe_recv_sk_key[2];
+static struct lock_class_key rxe_recv_slock_key[2];
+#endif /* CONFIG_DEBUG_LOCK_ALLOC */
+
+static inline void rxe_reclassify_recv_socket(struct socket *sock)
+{
+#ifdef CONFIG_DEBUG_LOCK_ALLOC
+ struct sock *sk = sock->sk;
+
+ if (WARN_ON_ONCE(!sock_allow_reclassification(sk)))
+ return;
+
+ switch (sk->sk_family) {
+ case AF_INET:
+ sock_lock_init_class_and_name(sk,
+ "slock-AF_INET-RDMA-RXE-RECV",
+ &rxe_recv_slock_key[0],
+ "sk_lock-AF_INET-RDMA-RXE-RECV",
+ &rxe_recv_sk_key[0]);
+ break;
+ case AF_INET6:
+ sock_lock_init_class_and_name(sk,
+ "slock-AF_INET6-RDMA-RXE-RECV",
+ &rxe_recv_slock_key[1],
+ "sk_lock-AF_INET6-RDMA-RXE-RECV",
+ &rxe_recv_sk_key[1]);
+ break;
+ default:
+ WARN_ON_ONCE(1);
+ }
+#endif /* CONFIG_DEBUG_LOCK_ALLOC */
+}
+
static struct dst_entry *rxe_find_route4(struct rxe_qp *qp,
struct net_device *ndev,
struct in_addr *saddr,
@@ -192,6 +240,7 @@ static struct socket *rxe_setup_udp_tunnel(struct net *net, __be16 port,
err = udp_sock_create(net, &udp_cfg, &sock);
if (err < 0)
return ERR_PTR(err);
+ rxe_reclassify_recv_socket(sock);
tnl_cfg.encap_type = 1;
tnl_cfg.encap_rcv = rxe_udp_encap_recv;
@@ -345,46 +394,35 @@ int rxe_prepare(struct rxe_av *av, struct rxe_pkt_info *pkt,
static void rxe_skb_tx_dtor(struct sk_buff *skb)
{
- struct sock *sk = skb->sk;
- struct rxe_qp *qp = sk->sk_user_data;
- int skb_out = atomic_dec_return(&qp->skb_out);
+ struct rxe_qp *qp = skb->sk->sk_user_data;
+ int skb_out;
+ skb_out = atomic_dec_return(&qp->skb_out);
if (unlikely(qp->need_req_skb &&
- skb_out < RXE_INFLIGHT_SKBS_PER_QP_LOW))
- rxe_sched_task(&qp->req.task);
+ skb_out < RXE_INFLIGHT_SKBS_PER_QP_LOW))
+ rxe_sched_task(&qp->send_task);
rxe_put(qp);
+ sock_put(skb->sk);
}
static int rxe_send(struct sk_buff *skb, struct rxe_pkt_info *pkt)
{
int err;
+ struct sock *sk = pkt->qp->sk->sk;
+ sock_hold(sk);
+ skb->sk = sk;
skb->destructor = rxe_skb_tx_dtor;
- skb->sk = pkt->qp->sk->sk;
-
rxe_get(pkt->qp);
atomic_inc(&pkt->qp->skb_out);
- if (skb->protocol == htons(ETH_P_IP)) {
+ if (skb->protocol == htons(ETH_P_IP))
err = ip_local_out(dev_net(skb_dst(skb)->dev), skb->sk, skb);
- } else if (skb->protocol == htons(ETH_P_IPV6)) {
+ else
err = ip6_local_out(dev_net(skb_dst(skb)->dev), skb->sk, skb);
- } else {
- rxe_dbg_qp(pkt->qp, "Unknown layer 3 protocol: %d\n",
- skb->protocol);
- atomic_dec(&pkt->qp->skb_out);
- rxe_put(pkt->qp);
- kfree_skb(skb);
- return -EINVAL;
- }
-
- if (unlikely(net_xmit_eval(err))) {
- rxe_dbg_qp(pkt->qp, "error sending packet: %d\n", err);
- return -EAGAIN;
- }
- return 0;
+ return err;
}
/* fix up a send packet to match the packets
@@ -392,8 +430,16 @@ static int rxe_send(struct sk_buff *skb, struct rxe_pkt_info *pkt)
*/
static int rxe_loopback(struct sk_buff *skb, struct rxe_pkt_info *pkt)
{
+ struct sock *sk = pkt->qp->sk->sk;
+
memcpy(SKB_TO_PKT(skb), pkt, sizeof(*pkt));
+ sock_hold(sk);
+ skb->sk = sk;
+ skb->destructor = rxe_skb_tx_dtor;
+ rxe_get(pkt->qp);
+ atomic_inc(&pkt->qp->skb_out);
+
if (skb->protocol == htons(ETH_P_IP))
skb_pull(skb, sizeof(struct iphdr));
else
@@ -440,12 +486,6 @@ int rxe_xmit_packet(struct rxe_qp *qp, struct rxe_pkt_info *pkt,
return err;
}
- if ((qp_type(qp) != IB_QPT_RC) &&
- (pkt->mask & RXE_END_MASK)) {
- pkt->wqe->state = wqe_state_done;
- rxe_sched_task(&qp->comp.task);
- }
-
rxe_counter_inc(rxe, RXE_CNT_SENT_PKTS);
goto done;
@@ -490,6 +530,9 @@ struct sk_buff *rxe_init_packet(struct rxe_dev *rxe, struct rxe_av *av,
goto out;
}
+ /* Add time stamp to skb. */
+ skb->tstamp = ktime_get();
+
skb_reserve(skb, hdr_len + LL_RESERVED_SPACE(ndev));
/* FIXME: hold reference to this netdev until life of this skb. */
@@ -517,7 +560,16 @@ out:
*/
const char *rxe_parent_name(struct rxe_dev *rxe, unsigned int port_num)
{
- return rxe->ndev->name;
+ struct net_device *ndev;
+ char *ndev_name;
+
+ ndev = rxe_ib_device_get_netdev(&rxe->ib_dev);
+ if (!ndev)
+ return NULL;
+ ndev_name = ndev->name;
+ dev_put(ndev);
+
+ return ndev_name;
}
int rxe_net_add(const char *ibdev_name, struct net_device *ndev)
@@ -529,9 +581,9 @@ int rxe_net_add(const char *ibdev_name, struct net_device *ndev)
if (!rxe)
return -ENOMEM;
- rxe->ndev = ndev;
+ ib_mark_name_assigned_by_user(&rxe->ib_dev);
- err = rxe_add(rxe, ndev->mtu, ibdev_name);
+ err = rxe_add(rxe, ndev->mtu, ibdev_name, ndev);
if (err) {
ib_dealloc_device(&rxe->ib_dev);
return err;
@@ -555,11 +607,6 @@ static void rxe_port_event(struct rxe_dev *rxe,
/* Caller must hold net_info_lock */
void rxe_port_up(struct rxe_dev *rxe)
{
- struct rxe_port *port;
-
- port = &rxe->port;
- port->attr.state = IB_PORT_ACTIVE;
-
rxe_port_event(rxe, IB_EVENT_PORT_ACTIVE);
dev_info(&rxe->ib_dev.dev, "set active\n");
}
@@ -567,11 +614,6 @@ void rxe_port_up(struct rxe_dev *rxe)
/* Caller must hold net_info_lock */
void rxe_port_down(struct rxe_dev *rxe)
{
- struct rxe_port *port;
-
- port = &rxe->port;
- port->attr.state = IB_PORT_DOWN;
-
rxe_port_event(rxe, IB_EVENT_PORT_ERR);
rxe_counter_inc(rxe, RXE_CNT_LINK_DOWNED);
dev_info(&rxe->ib_dev.dev, "set down\n");
@@ -579,10 +621,18 @@ void rxe_port_down(struct rxe_dev *rxe)
void rxe_set_port_state(struct rxe_dev *rxe)
{
- if (netif_running(rxe->ndev) && netif_carrier_ok(rxe->ndev))
+ struct net_device *ndev;
+
+ ndev = rxe_ib_device_get_netdev(&rxe->ib_dev);
+ if (!ndev)
+ return;
+
+ if (ib_get_curr_port_state(ndev) == IB_PORT_ACTIVE)
rxe_port_up(rxe);
else
rxe_port_down(rxe);
+
+ dev_put(ndev);
}
static int rxe_notify(struct notifier_block *not_blk,
@@ -599,18 +649,14 @@ static int rxe_notify(struct notifier_block *not_blk,
case NETDEV_UNREGISTER:
ib_unregister_device_queued(&rxe->ib_dev);
break;
- case NETDEV_UP:
- rxe_port_up(rxe);
- break;
- case NETDEV_DOWN:
- rxe_port_down(rxe);
- break;
case NETDEV_CHANGEMTU:
rxe_dbg_dev(rxe, "%s changed mtu to %d\n", ndev->name, ndev->mtu);
rxe_set_mtu(rxe, ndev->mtu);
break;
+ case NETDEV_DOWN:
case NETDEV_CHANGE:
- rxe_set_port_state(rxe);
+ if (ib_get_curr_port_state(ndev) == IB_PORT_DOWN)
+ rxe_counter_inc(rxe, RXE_CNT_LINK_DOWNED);
break;
case NETDEV_REBOOT:
case NETDEV_GOING_DOWN:
generated by cgit (git 2.34.1) at 2025-12-25 05:26:32 +0000