Revision v1 of 2 in this series.
4 messages, 2 authors, 2016-10-12 · open the first message on its own page
From: Ursula Braun <hidden>
Date: 2016-10-07 13:51:57
Hi Dave,
here are 2 patches for the s390-only af_iucv socket family code.
Thanks, Ursula
Ursula Braun (2):
Subject: [PATCH] af_iucv: enable control sends in case of
SEND_SHUTDOWN
Subject: [PATCH] af_iucv: drop skbs rejected by filter
net/iucv/af_iucv.c | 34 ++++++++++++++++++++++++++--------
1 file changed, 26 insertions(+), 8 deletions(-)
--
2.8.4
From: Ursula Braun <hidden>
Date: 2016-10-07 13:51:56
A packet filter might be installed for instance with setsockopt SO_ATTACH_FILTER. af_iucv currently queues skbs rejected by filter into the backlog queue. This does not make sense, since packets rejected by filter can be dropped immediately. This patch adds separate sk_filter return code checking, and dropping of packets if applicable. Signed-off-by: Ursula Braun <redacted> Reported-by: Daniel Borkmann <daniel@iogearbox.net> --- net/iucv/af_iucv.c | 24 +++++++++++++++++------- 1 file changed, 17 insertions(+), 7 deletions(-)
diff --git a/net/iucv/af_iucv.c b/net/iucv/af_iucv.c
index 88d0eaa..cfb9e5f 100644
--- a/net/iucv/af_iucv.c
+++ b/net/iucv/af_iucv.c@@ -1323,8 +1323,13 @@ static void iucv_process_message(struct sock *sk, struct sk_buff *skb, } IUCV_SKB_CB(skb)->offset = 0; - if (sock_queue_rcv_skb(sk, skb)) - skb_queue_head(&iucv_sk(sk)->backlog_skb_q, skb); + if (sk_filter(sk, skb)) { + atomic_inc(&sk->sk_drops); /* skb rejected by filter */ + kfree_skb(skb); + return; + } + if (__sock_queue_rcv_skb(sk, skb)) /* handle rcv queue full */ + skb_queue_tail(&iucv_sk(sk)->backlog_skb_q, skb); } /* iucv_process_message_q() - Process outstanding IUCV messages
@@ -1438,13 +1443,13 @@ static int iucv_sock_recvmsg(struct socket *sock, struct msghdr *msg, rskb = skb_dequeue(&iucv->backlog_skb_q); while (rskb) { IUCV_SKB_CB(rskb)->offset = 0; - if (sock_queue_rcv_skb(sk, rskb)) { + if (__sock_queue_rcv_skb(sk, rskb)) { + /* handle rcv queue full */ skb_queue_head(&iucv->backlog_skb_q, rskb); break; - } else { - rskb = skb_dequeue(&iucv->backlog_skb_q); } + rskb = skb_dequeue(&iucv->backlog_skb_q); } if (skb_queue_empty(&iucv->backlog_skb_q)) { if (!list_empty(&iucv->message_q.list))
@@ -2124,12 +2129,17 @@ static int afiucv_hs_callback_rx(struct sock *sk, struct sk_buff *skb) skb_reset_transport_header(skb); skb_reset_network_header(skb); IUCV_SKB_CB(skb)->offset = 0; + if (sk_filter(sk, skb)) { + atomic_inc(&sk->sk_drops); /* skb rejected by filter */ + kfree_skb(skb); + return NET_RX_SUCCESS; + } + spin_lock(&iucv->message_q.lock); if (skb_queue_empty(&iucv->backlog_skb_q)) { - if (sock_queue_rcv_skb(sk, skb)) { + if (__sock_queue_rcv_skb(sk, skb)) /* handle rcv queue full */ skb_queue_tail(&iucv->backlog_skb_q, skb); - } } else skb_queue_tail(&iucv_sk(sk)->backlog_skb_q, skb); spin_unlock(&iucv->message_q.lock);
--
2.8.4
From: Ursula Braun <hidden>
Date: 2016-10-07 13:51:57
If a socket program has shut down the socket for sending, it can still receive an undetermined number of packets. The AF_IUCV protocol for HIPER transport requires sending of a WIN flag from time to time from the receiver to the sender, otherwise the peer cannot continue sending. That means sending of control flags must still work, even though the AF_IUCV socket is shutdown for sending data. sock_alloc_send_skb() returns with error EPIPE, if socket sk_shutdown is SEND_SHUTDOWN. Thus this patch temporarily removes the send shutdown attribute from the socket to enable transfer of control flags. Signed-off-by: Ursula Braun <redacted> --- net/iucv/af_iucv.c | 10 +++++++++- 1 file changed, 9 insertions(+), 1 deletion(-)
diff --git a/net/iucv/af_iucv.c b/net/iucv/af_iucv.c
index 02b45a8..88d0eaa 100644
--- a/net/iucv/af_iucv.c
+++ b/net/iucv/af_iucv.c@@ -453,19 +453,27 @@ static void iucv_sever_path(struct sock *sk, int with_user_data) } } -/* Send FIN through an IUCV socket for HIPER transport */ +/* Send controlling flags through an IUCV socket for HIPER transport */ static int iucv_send_ctrl(struct sock *sk, u8 flags) { int err = 0; int blen; struct sk_buff *skb; + u8 shutdown = 0; blen = sizeof(struct af_iucv_trans_hdr) + ETH_HLEN; + if (sk->sk_shutdown & SEND_SHUTDOWN) { + /* controlling flags should be sent anyway */ + shutdown = sk->sk_shutdown; + sk->sk_shutdown &= RCV_SHUTDOWN; + } skb = sock_alloc_send_skb(sk, blen, 1, &err); if (skb) { skb_reserve(skb, blen); err = afiucv_hs_send(NULL, sk, skb, flags); } + if (shutdown) + sk->sk_shutdown = shutdown; return err; }
--
2.8.4
From: David Miller <davem@davemloft.net>
Date: 2016-10-12 06:02:51
From: Ursula Braun <redacted> Date: Fri, 7 Oct 2016 15:51:46 +0200
here are 2 patches for the s390-only af_iucv socket family code.
Series applied to net-next, thanks.