nvme-tcp: Use protocol specific operations while reading socket
authorPotnuri Bharat Teja <bharat@chelsio.com>
Mon, 8 Jul 2019 09:52:00 +0000 (15:22 +0530)
committerSagi Grimberg <sagi@grimberg.me>
Thu, 29 Aug 2019 19:55:00 +0000 (12:55 -0700)
Using socket specific read_sock() calls instead of directly calling
tcp_read_sock() helps lld module registered handlers if any, to be called
from nvme-tcp host.
This patch therefore replaces the tcp_read_sock() with socket specific
prot_ops.

Signed-off-by: Potnuri Bharat Teja <bharat@chelsio.com>
Acked-by: Sagi Grimberg <sagi@grimberg.me>
Signed-off-by: Sagi Grimberg <sagi@grimberg.me>
drivers/nvme/host/tcp.c

index 2c5df86..53c32f9 100644 (file)
@@ -1018,14 +1018,15 @@ done:
 
 static int nvme_tcp_try_recv(struct nvme_tcp_queue *queue)
 {
-       struct sock *sk = queue->sock->sk;
+       struct socket *sock = queue->sock;
+       struct sock *sk = sock->sk;
        read_descriptor_t rd_desc;
        int consumed;
 
        rd_desc.arg.data = queue;
        rd_desc.count = 1;
        lock_sock(sk);
-       consumed = tcp_read_sock(sk, &rd_desc, nvme_tcp_recv_skb);
+       consumed = sock->ops->read_sock(sk, &rd_desc, nvme_tcp_recv_skb);
        release_sock(sk);
        return consumed;
 }