From: Jeffrey Altman <jaltman@secure-endpoints.com>
Date: Sat, 11 Jul 2009 00:43:13 +0000 (-0400)
Subject: Correct computation of rx packet resends and data packets sent.
X-Git-Tag: openafs-devel-1_5_61~137
X-Git-Url: https://git.openafs.org/?p=openafs.git;a=commitdiff_plain;h=c4c0c67fb6afc30a97c93552d00460b3291ac872;hp=5d4dabc50a0f313b95eb3ab6be7153d5f41fef5f

Correct computation of rx packet resends and data packets sent.

rx packet resends were being double counted.

rx data packets sent were one more per send list than it
should have been.

LICENSE MIT

Reviewed-on: http://gerrit.openafs.org/54
Verified-by: Russ Allbery <rra@stanford.edu>
Reviewed-by: Russ Allbery <rra@stanford.edu>
---

diff --git a/src/rx/rx.c b/src/rx/rx.c
index fdff269..63ed366 100644
--- a/src/rx/rx.c
+++ b/src/rx/rx.c
@@ -5052,7 +5052,7 @@ rxi_SendList(struct rx_call *call, struct rx_packet **list, int len,
     if (resending)
 	peer->reSends += len;
     if (rx_stats_active)
-        rx_MutexIncrement(rx_stats.dataPacketsSent, rx_stats_mutex);
+        rx_MutexAdd(rx_stats.dataPacketsSent, len, rx_stats_mutex);
     MUTEX_EXIT(&peer->peer_lock);
 
     if (list[len - 1]->header.flags & RX_LAST_PACKET) {
@@ -5099,14 +5099,6 @@ rxi_SendList(struct rx_call *call, struct rx_packet **list, int len,
 	    }
 	}
 
-	MUTEX_ENTER(&peer->peer_lock);
-	peer->nSent++;
-	if (resending)
-	    peer->reSends++;
-        if (rx_stats_active)
-            rx_MutexIncrement(rx_stats.dataPacketsSent, rx_stats_mutex);
-	MUTEX_EXIT(&peer->peer_lock);
-
 	/* Tag this packet as not being the last in this group,
 	 * for the receiver's benefit */
 	if (i < len - 1 || moreFlag) {