From: Quentin Young <qlyoung@cumulusnetworks.com>
Date: Thu, 6 Apr 2017 01:09:33 +0000 (+0000)
Subject: bgpd: transfer packets from peer stub to actual peer
X-Git-Tag: frr-4.0-dev~120^2~52
X-Git-Url: https://git.puffer.fish/?a=commitdiff_plain;h=727c4f870b0e6079b761afc525f6a1f5262bcf9f;p=mirror%2Ffrr.git

bgpd: transfer packets from peer stub to actual peer

During transition from OpenConfirm -> Established, we wipe the peer stub's
output buffer. Because thread.c prioritizes I/O operations over regular
background threads and events, in a single threaded environment this ordering
meant that the output buffer would be happily empty at wipe time.  In MT-land,
this convenient coincidence is no longer true; thus we need to make sure that
any packets remaining on the peer stub get transferred over to the peer proper.

Also removes misleading comment indicating that bgp_establish() sends a
keepalive packet. It does not.

Signed-off-by: Quentin Young <qlyoung@cumulusnetworks.com>
---

diff --git a/bgpd/bgp_fsm.c b/bgpd/bgp_fsm.c
index aa883f7364..3344bf3126 100644
--- a/bgpd/bgp_fsm.c
+++ b/bgpd/bgp_fsm.c
@@ -138,17 +138,22 @@ static struct peer *peer_xfer_conn(struct peer *from_peer)
 	from_peer->fd = fd;
 	stream_reset(peer->ibuf);
 
+	// At this point in time, it is possible that there are packets pending
+	// on
+	// from_peer->obuf. These need to be transferred to the new peer struct.
 	pthread_mutex_lock(&peer->obuf_mtx);
+	pthread_mutex_lock(&from_peer->obuf_mtx);
 	{
+		// wipe new peer's packet queue
 		stream_fifo_clean(peer->obuf);
-	}
-	pthread_mutex_unlock(&peer->obuf_mtx);
 
-	pthread_mutex_lock(&from_peer->obuf_mtx);
-	{
-		stream_fifo_clean(from_peer->obuf);
+		// copy each packet from old peer's queue to new peer's queue
+		while (from_peer->obuf->head)
+			stream_fifo_push(peer->obuf,
+					 stream_fifo_pop(from_peer->obuf));
 	}
 	pthread_mutex_unlock(&from_peer->obuf_mtx);
+	pthread_mutex_unlock(&peer->obuf_mtx);
 
 	peer->as = from_peer->as;
 	peer->v_holdtime = from_peer->v_holdtime;
@@ -1398,8 +1403,12 @@ static int bgp_fsm_holdtime_expire(struct peer *peer)
 	return bgp_stop_with_notify(peer, BGP_NOTIFY_HOLD_ERR, 0);
 }
 
-/* Status goes to Established.  Send keepalive packet then make first
-   update information. */
+/**
+ * Transition to Established state.
+ *
+ * Convert peer from stub to full fledged peer, set some timers, and generate
+ * initial updates.
+ */
 static int bgp_establish(struct peer *peer)
 {
 	afi_t afi;