From mboxrd@z Thu Jan 1 00:00:00 1970 From: Thomas Graf Subject: [PATCH] sctp: ABORT if receive queue is not empty while closing socket Date: Thu, 30 Jun 2011 09:31:22 -0400 Message-ID: <20110630133122.GB24074@canuck.infradead.org> References: <20110629135704.GB10085@canuck.infradead.org> <4E0B3491.1060603@hp.com> <20110629143649.GC10085@canuck.infradead.org> <4E0B3DA1.9060200@hp.com> <20110629154814.GD10085@canuck.infradead.org> <4E0B4F71.4020108@hp.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Cc: netdev@vger.kernel.org, davem@davemloft.net, Wei Yongjun , Sridhar Samudrala , linux-sctp@vger.kernel.org To: Vladislav Yasevich Return-path: Received: from merlin.infradead.org ([205.233.59.134]:38405 "EHLO merlin.infradead.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751057Ab1F3NbZ (ORCPT ); Thu, 30 Jun 2011 09:31:25 -0400 Content-Disposition: inline In-Reply-To: <4E0B4F71.4020108@hp.com> Sender: netdev-owner@vger.kernel.org List-ID: On Wed, Jun 29, 2011 at 12:14:41PM -0400, Vladislav Yasevich wrote: > Right. The lack of ABORT from the receive of data is a bug. I was trying to point out > that instead of modified the sender of data to send the ABORT, you modify the receiver > to send the ABORT when it is being closed while having data queued. Is this what you had in mind? Trigger user ABORT when a socket is closed which has skbs sitting on the receive queue. If data was lost, there is no point in doing a graceful shutdown. This is consistent with TCP behaviour. This also resolves the situation when a receiver cannot reopen its rwnd and the sender continues retransmission attempts indefinitely before initiating the shutdown. Signed-off-by: Thomas Graf diff --git a/include/net/sctp/ulpevent.h b/include/net/sctp/ulpevent.h index 99b027b..ca4693b 100644 --- a/include/net/sctp/ulpevent.h +++ b/include/net/sctp/ulpevent.h @@ -80,7 +80,7 @@ static inline struct sctp_ulpevent *sctp_skb2event(struct sk_buff *skb) void sctp_ulpevent_free(struct sctp_ulpevent *); int sctp_ulpevent_is_notification(const struct sctp_ulpevent *); -void sctp_queue_purge_ulpevents(struct sk_buff_head *list); +unsigned int sctp_queue_purge_ulpevents(struct sk_buff_head *list); struct sctp_ulpevent *sctp_ulpevent_make_assoc_change( const struct sctp_association *asoc, diff --git a/net/sctp/socket.c b/net/sctp/socket.c index 6766913..958253a 100644 --- a/net/sctp/socket.c +++ b/net/sctp/socket.c @@ -1384,6 +1384,7 @@ SCTP_STATIC void sctp_close(struct sock *sk, long timeout) struct sctp_endpoint *ep; struct sctp_association *asoc; struct list_head *pos, *temp; + unsigned int data_was_unread; SCTP_DEBUG_PRINTK("sctp_close(sk: 0x%p, timeout:%ld)\n", sk, timeout); @@ -1393,6 +1394,10 @@ SCTP_STATIC void sctp_close(struct sock *sk, long timeout) ep = sctp_sk(sk)->ep; + /* Clean up any skbs sitting on the receive queue. */ + data_was_unread = sctp_queue_purge_ulpevents(&sk->sk_receive_queue); + data_was_unread += sctp_queue_purge_ulpevents(&sctp_sk(sk)->pd_lobby); + /* Walk all associations on an endpoint. */ list_for_each_safe(pos, temp, &ep->asocs) { asoc = list_entry(pos, struct sctp_association, asocs); @@ -1410,7 +1415,8 @@ SCTP_STATIC void sctp_close(struct sock *sk, long timeout) } } - if (sock_flag(sk, SOCK_LINGER) && !sk->sk_lingertime) { + if (data_was_unread || + (sock_flag(sk, SOCK_LINGER) && !sk->sk_lingertime)) { struct sctp_chunk *chunk; chunk = sctp_make_abort_user(asoc, NULL, 0); @@ -1420,10 +1426,6 @@ SCTP_STATIC void sctp_close(struct sock *sk, long timeout) sctp_primitive_SHUTDOWN(asoc, NULL); } - /* Clean up any skbs sitting on the receive queue. */ - sctp_queue_purge_ulpevents(&sk->sk_receive_queue); - sctp_queue_purge_ulpevents(&sctp_sk(sk)->pd_lobby); - /* On a TCP-style socket, block for at most linger_time if set. */ if (sctp_style(sk, TCP) && timeout) sctp_wait_for_close(sk, timeout); diff --git a/net/sctp/ulpevent.c b/net/sctp/ulpevent.c index e70e5fc..aab3184 100644 --- a/net/sctp/ulpevent.c +++ b/net/sctp/ulpevent.c @@ -1081,9 +1081,19 @@ void sctp_ulpevent_free(struct sctp_ulpevent *event) } /* Purge the skb lists holding ulpevents. */ -void sctp_queue_purge_ulpevents(struct sk_buff_head *list) +unsigned int sctp_queue_purge_ulpevents(struct sk_buff_head *list) { struct sk_buff *skb; - while ((skb = skb_dequeue(list)) != NULL) + unsigned int data_unread = 0; + + while ((skb = skb_dequeue(list)) != NULL) { + struct sctp_ulpevent *event = sctp_skb2event(skb); + + if (!sctp_ulpevent_is_notification(event)) + data_unread += skb->len; + sctp_ulpevent_free(sctp_skb2event(skb)); + } + + return data_unread; }