[SCTP]: Send only 1 window update SACK per message.
Tsutomu Fujii [Sun, 18 Jun 2006 05:58:28 +0000 (22:58 -0700)]
Right now, every time we increase our rwnd by more then MTU bytes, we
trigger a SACK.  When processing large messages, this will generate a
SACK for almost every other SCTP fragment. However since we are freeing
the entire message at the same time, we might as well collapse the SACK
generation to 1.

Signed-off-by: Tsutomu Fujii <t-fujii@nb.jp.nec.com>
Signed-off-by: Vlad Yasevich <vladislav.yasevich@hp.com>
Signed-off-by: Sridhar Samudrala <sri@us.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net/sctp/ulpevent.c

index ba97f97..ee23678 100644 (file)
@@ -51,6 +51,8 @@
 static void sctp_ulpevent_receive_data(struct sctp_ulpevent *event,
                                       struct sctp_association *asoc);
 static void sctp_ulpevent_release_data(struct sctp_ulpevent *event);
+static void sctp_ulpevent_release_frag_data(struct sctp_ulpevent *event);
+
 
 /* Initialize an ULP event from an given skb.  */
 SCTP_STATIC void sctp_ulpevent_init(struct sctp_ulpevent *event, int msg_flags)
@@ -883,6 +885,7 @@ static void sctp_ulpevent_receive_data(struct sctp_ulpevent *event,
 static void sctp_ulpevent_release_data(struct sctp_ulpevent *event)
 {
        struct sk_buff *skb, *frag;
+       unsigned int    len;
 
        /* Current stack structures assume that the rcv buffer is
         * per socket.   For UDP style sockets this is not true as
@@ -892,7 +895,30 @@ static void sctp_ulpevent_release_data(struct sctp_ulpevent *event)
         */
 
        skb = sctp_event2skb(event);
-       sctp_assoc_rwnd_increase(event->asoc, skb_headlen(skb));
+       len = skb->len;
+
+       if (!skb->data_len)
+               goto done;
+
+       /* Don't forget the fragments. */
+       for (frag = skb_shinfo(skb)->frag_list; frag; frag = frag->next) {
+               /* NOTE:  skb_shinfos are recursive. Although IP returns
+                * skb's with only 1 level of fragments, SCTP reassembly can
+                * increase the levels.
+                */
+               sctp_ulpevent_release_frag_data(sctp_skb2event(frag));
+       }
+
+done:
+       sctp_assoc_rwnd_increase(event->asoc, len);
+       sctp_ulpevent_release_owner(event);
+}
+
+static void sctp_ulpevent_release_frag_data(struct sctp_ulpevent *event)
+{
+       struct sk_buff *skb, *frag;
+
+       skb = sctp_event2skb(event);
 
        if (!skb->data_len)
                goto done;
@@ -903,7 +929,7 @@ static void sctp_ulpevent_release_data(struct sctp_ulpevent *event)
                 * skb's with only 1 level of fragments, SCTP reassembly can
                 * increase the levels.
                 */
-               sctp_ulpevent_release_data(sctp_skb2event(frag));
+               sctp_ulpevent_release_frag_data(sctp_skb2event(frag));
        }
 
 done: