[PATCH net] rds: tcp: compute m_ack_seq as offset from ->write_seq

[PATCH net] rds: tcp: compute m_ack_seq as offset from ->write_seq

[Sowmini Varadhan]

rds-tcp uses m_ack_seq to track the tcp ack# that indicates
that the peer has received a rds_message. The m_ack_seq is
used in rds_tcp_is_acked() to figure out when it is safe to
drop the rds_message from the RDS retransmit queue.

The m_ack_seq must be calculated as an offset from the right 
edge of the in-flight tcp buffer, i.e., it should be based on 
the ->write_seq, not the ->snd_nxt.

Signed-off-by: Sowmini Varadhan <sowmini.varadhan@oracle.com>
---
 net/rds/tcp.c      |    5 +++--
 net/rds/tcp.h      |    2 +-
 net/rds/tcp_send.c |    4 ++--
 3 files changed, 6 insertions(+), 5 deletions(-)

diff --git a/net/rds/tcp.c b/net/rds/tcp.c
index 6b7ee71..ab7356e 100644
--- a/net/rds/tcp.c
+++ b/net/rds/tcp.c
@@ -90,9 +90,10 @@ void rds_tcp_nonagle(struct socket *sock)
 			      sizeof(val));
 }
 
-u32 rds_tcp_snd_nxt(struct rds_tcp_connection *tc)
+u32 rds_tcp_write_seq(struct rds_tcp_connection *tc)
 {
-	return tcp_sk(tc->t_sock->sk)->snd_nxt;
+	/* seq# of the last byte of data in tcp send buffer */
+	return tcp_sk(tc->t_sock->sk)->write_seq;
 }
 
 u32 rds_tcp_snd_una(struct rds_tcp_connection *tc)
diff --git a/net/rds/tcp.h b/net/rds/tcp.h
index 1aafbf7..864ca7d 100644
--- a/net/rds/tcp.h
+++ b/net/rds/tcp.h
@@ -54,7 +54,7 @@ struct rds_tcp_statistics {
 void rds_tcp_reset_callbacks(struct socket *sock, struct rds_conn_path *cp);
 void rds_tcp_restore_callbacks(struct socket *sock,
 			       struct rds_tcp_connection *tc);
-u32 rds_tcp_snd_nxt(struct rds_tcp_connection *tc);
+u32 rds_tcp_write_seq(struct rds_tcp_connection *tc);
 u32 rds_tcp_snd_una(struct rds_tcp_connection *tc);
 u64 rds_tcp_map_seq(struct rds_tcp_connection *tc, u32 seq);
 extern struct rds_transport rds_tcp_transport;
diff --git a/net/rds/tcp_send.c b/net/rds/tcp_send.c
index dc860d1..9b76e0f 100644
--- a/net/rds/tcp_send.c
+++ b/net/rds/tcp_send.c
@@ -86,7 +86,7 @@ int rds_tcp_xmit(struct rds_connection *conn, struct rds_message *rm,
 		 * m_ack_seq is set to the sequence number of the last byte of
 		 * header and data.  see rds_tcp_is_acked().
 		 */
-		tc->t_last_sent_nxt = rds_tcp_snd_nxt(tc);
+		tc->t_last_sent_nxt = rds_tcp_write_seq(tc);
 		rm->m_ack_seq = tc->t_last_sent_nxt +
 				sizeof(struct rds_header) +
 				be32_to_cpu(rm->m_inc.i_hdr.h_len) - 1;
@@ -98,7 +98,7 @@ int rds_tcp_xmit(struct rds_connection *conn, struct rds_message *rm,
 			rm->m_inc.i_hdr.h_flags |= RDS_FLAG_RETRANSMITTED;
 
 		rdsdebug("rm %p tcp nxt %u ack_seq %llu\n",
-			 rm, rds_tcp_snd_nxt(tc),
+			 rm, rds_tcp_write_seq(tc),
 			 (unsigned long long)rm->m_ack_seq);
 	}
 
-- 
1.7.1

Re: [PATCH net] rds: tcp: compute m_ack_seq as offset from ->write_seq

[David Miller]

From: Sowmini Varadhan <sowmini.varadhan@oracle.com>
Date: Thu, 18 Jan 2018 13:11:07 -0800

> rds-tcp uses m_ack_seq to track the tcp ack# that indicates
> that the peer has received a rds_message. The m_ack_seq is
> used in rds_tcp_is_acked() to figure out when it is safe to
> drop the rds_message from the RDS retransmit queue.
> 
> The m_ack_seq must be calculated as an offset from the right 
> edge of the in-flight tcp buffer, i.e., it should be based on 
> the ->write_seq, not the ->snd_nxt.
> 
> Signed-off-by: Sowmini Varadhan <sowmini.varadhan@oracle.com>

Applied, thanks Sowmini.