From 230ce576f0735e2d4d14b86bb4b8f11f6f7394ff Mon Sep 17 00:00:00 2001
From: isaac <isaac>
Date: Mon, 7 Dec 2009 19:31:38 +0000
Subject: [PATCH] i=maxim,i=chris,b=21459: - should update lp_alive for
 non-router peers.

---
 lnet/ChangeLog                |  4 ++++
 lnet/include/lnet/lib-types.h |  2 ++
 lnet/lnet/lib-move.c          | 32 +++++++++++++++-----------------
 lnet/lnet/peer.c              |  2 +-
 lnet/lnet/router_proc.c       |  2 +-
 5 files changed, 23 insertions(+), 19 deletions(-)

diff --git a/lnet/ChangeLog b/lnet/ChangeLog
index 7fd91a5..d5e212d 100644
--- a/lnet/ChangeLog
+++ b/lnet/ChangeLog
@@ -12,6 +12,10 @@ tbd  Sun Microsystems, Inc.
 	  mxlnd     - MX 1.2.1 or later,
 	  ptllnd    - Portals 3.3 / UNICOS/lc 1.5.x, 2.0.x
 
+Severity   : minor
+Bugzilla   : 21459
+Description: should update lp_alive for non-router peers.
+
 Severity   : enhancement
 Bugzilla   : 15332
 Description: LNet router shuffler.
diff --git a/lnet/include/lnet/lib-types.h b/lnet/include/lnet/lib-types.h
index e321e25..f34ffb8 100644
--- a/lnet/include/lnet/lib-types.h
+++ b/lnet/include/lnet/lib-types.h
@@ -453,6 +453,8 @@ typedef struct lnet_peer {
         lnet_rc_data_t   *lp_rcd;               /* router checker state */
 } lnet_peer_t;
 
+#define lnet_peer_aliveness_enabled(lp) ((lp)->lp_ni->ni_peertimeout > 0)
+
 typedef struct {
 	struct list_head  lr_list;              /* chain on net */
         lnet_peer_t      *lr_gateway;           /* router node */
diff --git a/lnet/lnet/lib-move.c b/lnet/lnet/lib-move.c
index e3d6042..8646e3f 100644
--- a/lnet/lnet/lib-move.c
+++ b/lnet/lnet/lib-move.c
@@ -914,8 +914,7 @@ lnet_ni_peer_alive(lnet_peer_t *lp)
         cfs_time_t  last_alive = 0;
         lnet_ni_t  *ni = lp->lp_ni;
 
-        LASSERT (ni != NULL);
-        LASSERT (ni->ni_peertimeout > 0);
+        LASSERT (lnet_peer_aliveness_enabled(lp));
         LASSERT (ni->ni_lnd->lnd_query != NULL);
 
         LNET_UNLOCK();
@@ -933,12 +932,10 @@ lnet_ni_peer_alive(lnet_peer_t *lp)
 static inline int
 lnet_peer_is_alive (lnet_peer_t *lp, cfs_time_t now)
 {
-        lnet_ni_t  *ni = lp->lp_ni;
-        cfs_time_t  deadline;
-        int         alive;
+        int        alive;
+        cfs_time_t deadline;
 
-        LASSERT (ni != NULL);
-        LASSERT (ni->ni_peertimeout > 0);
+        LASSERT (lnet_peer_aliveness_enabled(lp));
 
         /* Trust lnet_notify() if it has more recent aliveness news, but
          * ignore the initial assumed death (see lnet_peers_start_down()).
@@ -948,12 +945,15 @@ lnet_peer_is_alive (lnet_peer_t *lp, cfs_time_t now)
                 return 0;
 
         deadline = cfs_time_add(lp->lp_last_alive,
-                                cfs_time_seconds(ni->ni_peertimeout));
+                                cfs_time_seconds(lp->lp_ni->ni_peertimeout));
         alive = cfs_time_after(deadline, now);
 
-        /* Update obsolete lp_alive */
-        if (alive && !lp->lp_alive && lp->lp_timestamp != 0 &&
-            cfs_time_before(lp->lp_timestamp, lp->lp_last_alive))
+        /* Update obsolete lp_alive except for routers assumed to be dead
+         * initially, because router checker would update aliveness in this
+         * case, and moreover lp_last_alive at peer creation is assumed.
+         */
+        if (alive && !lp->lp_alive &&
+            !(lnet_isrouter(lp) && lp->lp_alive_count == 0))
                 lnet_notify_locked(lp, 0, 1, lp->lp_last_alive);
 
         return alive;
@@ -965,12 +965,9 @@ lnet_peer_is_alive (lnet_peer_t *lp, cfs_time_t now)
 int
 lnet_peer_alive_locked (lnet_peer_t *lp)
 {
-        lnet_ni_t  *ni = lp->lp_ni;
-        cfs_time_t  now = cfs_time_current();
-
-        LASSERT (ni != NULL);
+        cfs_time_t now = cfs_time_current();
 
-        if (ni->ni_peertimeout <= 0)  /* disabled */
+        if (!lnet_peer_aliveness_enabled(lp))
                 return -ENODEV;
 
         if (lnet_peer_is_alive(lp, now))
@@ -991,7 +988,8 @@ lnet_peer_alive_locked (lnet_peer_t *lp)
                                       "%d < %d (%d/%d)\n",
                                       libcfs_nid2str(lp->lp_nid),
                                       (int)now, (int)next_query,
-                                      lnet_queryinterval, ni->ni_peertimeout);
+                                      lnet_queryinterval,
+                                      lp->lp_ni->ni_peertimeout);
                         return 0;
                 }
         }
diff --git a/lnet/lnet/peer.c b/lnet/lnet/peer.c
index 2e7b9a4..8baf199 100644
--- a/lnet/lnet/peer.c
+++ b/lnet/lnet/peer.c
@@ -253,7 +253,7 @@ lnet_debug_peer(lnet_nid_t nid)
                 return;
         }
 
-        if (lnet_isrouter(lp) || lp->lp_ni->ni_peertimeout > 0)
+        if (lnet_isrouter(lp) || lnet_peer_aliveness_enabled(lp))
                 aliveness = lp->lp_alive ? "up" : "down";
 
         CDEBUG(D_WARNING, "%-24s %4d %5s %5d %5d %5d %5d %5d %ld\n",
diff --git a/lnet/lnet/router_proc.c b/lnet/lnet/router_proc.c
index 76d82b3..f04f7ba 100644
--- a/lnet/lnet/router_proc.c
+++ b/lnet/lnet/router_proc.c
@@ -444,7 +444,7 @@ int LL_PROC_PROTO(proc_lnet_peers)
                         int        txqnob    = peer->lp_txqnob;
 
                         if (lnet_isrouter(peer) ||
-                            peer->lp_ni->ni_peertimeout > 0)
+                            lnet_peer_aliveness_enabled(peer))
                                 aliveness = peer->lp_alive ? "up" : "down";
 
                         s += snprintf(s, tmpstr + tmpsiz - s,
-- 
1.8.3.1