RDMA/rxe: Fix missed IB reference counting in loopback

When the noted patch below extending the reference taken by
rxe_get_dev_from_net() in rxe_udp_encap_recv() until each skb is freed it
was not matched by a reference in the loopback path resulting in
underflows.

Fixes: 899aba891cab ("RDMA/rxe: Fix FIXME in rxe_udp_encap_recv()")
Link: https://lore.kernel.org/r/20210304192048.2958-1-rpearson@hpe.com
Signed-off-by: Bob Pearson <rpearsonhpe@gmail.com>
Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>

diff --git a/drivers/infiniband/sw/rxe/rxe_net.c b/drivers/infiniband/sw/rxe/rxe_net.c
index 0701bd1..0166272 100644 (file)
--- a/drivers/infiniband/sw/rxe/rxe_net.c
+++ b/drivers/infiniband/sw/rxe/rxe_net.c
@@ -407,14 +407,22 @@ int rxe_send(struct rxe_pkt_info *pkt, struct sk_buff *skb)
        return 0;
 }
 
+/* fix up a send packet to match the packets
+ * received from UDP before looping them back
+ */
 void rxe_loopback(struct sk_buff *skb)
 {
+       struct rxe_pkt_info *pkt = SKB_TO_PKT(skb);
+
        if (skb->protocol == htons(ETH_P_IP))
                skb_pull(skb, sizeof(struct iphdr));
        else
                skb_pull(skb, sizeof(struct ipv6hdr));
 
-       rxe_rcv(skb);
+       if (WARN_ON(!ib_device_try_get(&pkt->rxe->ib_dev)))
+               kfree_skb(skb);
+       else
+               rxe_rcv(skb);
 }
 
 struct sk_buff *rxe_init_packet(struct rxe_dev *rxe, struct rxe_av *av,