diff options
| author |   Sowmini Varadhan <sowmini.varadhan@oracle.com> | 2016-06-04 13:59:58 -0700 |
|---|---|---|
| committer |   David S. Miller <davem@davemloft.net> | 2016-06-07 15:10:15 -0700 |
| commit | 335b48d980f631fbc5b233cbb3625ac0c86d67cb (patch) | |
| tree | c09223b2cdbe5b6a9d876ad07a9ee29cebcb8588 /net/rds/tcp_listen.c | |
| parent | fq_codel: fix NET_XMIT_CN behavior (diff) | |
| download | linux-dev-335b48d980f631fbc5b233cbb3625ac0c86d67cb.tar.xz linux-dev-335b48d980f631fbc5b233cbb3625ac0c86d67cb.zip | |
RDS: TCP: Add/use rds_tcp_reset_callbacks to reset tcp socket safely
When rds_tcp_accept_one() has to replace the existing tcp socket
with a newer tcp socket (duelling-syn resolution), it must lock_sock()
to suppress the rds_tcp_data_recv() path while callbacks are being
changed. Also, existing RDS datagram reassembly state must be reset,
so that the next datagram on the new socket does not have corrupted
state. Similarly when resetting the newly accepted socket, appropriate
locks and synchronization is needed.
This commit ensures correct synchronization by invoking
kernel_sock_shutdown to reset a newly accepted sock, and by taking
appropriate lock_sock()s (for old and new sockets) when resetting
existing callbacks.
Signed-off-by: Sowmini Varadhan <sowmini.varadhan@oracle.com>
Acked-by: Santosh Shilimkar <santosh.shilimkar@oracle.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Diffstat (limited to 'net/rds/tcp_listen.c')
| -rw-r--r-- | net/rds/tcp_listen.c | 13 |
1 files changed, 4 insertions, 9 deletions
diff --git a/net/rds/tcp_listen.c b/net/rds/tcp_listen.c index 4bf4befe5066..d9fe53675d95 100644 --- a/net/rds/tcp_listen.c +++ b/net/rds/tcp_listen.c @@ -78,7 +78,6 @@ int rds_tcp_accept_one(struct socket *sock) struct inet_sock *inet; struct rds_tcp_connection *rs_tcp = NULL; int conn_state; - struct sock *nsk; if (!sock) /* module unload or netns delete in progress */ return -ENETUNREACH; @@ -139,23 +138,19 @@ int rds_tcp_accept_one(struct socket *sock) atomic_set(&conn->c_state, RDS_CONN_CONNECTING); wait_event(conn->c_waitq, !test_bit(RDS_IN_XMIT, &conn->c_flags)); - rds_tcp_restore_callbacks(rs_tcp->t_sock, rs_tcp); + rds_tcp_reset_callbacks(new_sock, conn); conn->c_outgoing = 0; } + } else { + rds_tcp_set_callbacks(new_sock, conn); } - rds_tcp_set_callbacks(new_sock, conn); rds_connect_complete(conn); /* marks RDS_CONN_UP */ new_sock = NULL; ret = 0; goto out; rst_nsk: /* reset the newly returned accept sock and bail */ - nsk = new_sock->sk; - rds_tcp_stats_inc(s_tcp_listen_closed_stale); - nsk->sk_user_data = NULL; - nsk->sk_prot->disconnect(nsk, 0); - tcp_done(nsk); - new_sock = NULL; + kernel_sock_shutdown(new_sock, SHUT_RDWR); ret = 0; out: if (rs_tcp) |