Fixed JOIN notify timeout handling, fixes a crash.

[silc.git] / lib / silcclient / client_notify.c

diff --git a/lib/silcclient/client_notify.c b/lib/silcclient/client_notify.c
index 5d431954d602b0882b0583d64b746451812b4b87..64c041eaf11a6371551c4b4a757b64a82a96a83c 100644 (file)
--- a/lib/silcclient/client_notify.c
+++ b/lib/silcclient/client_notify.c
@@ -56,9 +56,14 @@ static void silc_client_notify_resolved(SilcClient client,
   /* If entry is still invalid, resolving failed.  Finish notify processing. */
   if (notify->client_entry && !notify->client_entry->internal.valid) {
     /* If resolving timedout try it again many times. */
-    if (status != SILC_STATUS_ERR_TIMEDOUT || ++notify->resolve_retry > 1000)
+    if (status != SILC_STATUS_ERR_TIMEDOUT || ++notify->resolve_retry > 1000) {
       silc_fsm_next(notify->fsm, silc_client_notify_processed);
-    silc_client_unref_client(client, conn, notify->client_entry);
+
+      /* Unref client only in case of non-timeout error.  In case of timeout
+        occurred, the routine reprocessing the notify is expected not to
+        create new references of the entry. */
+      silc_client_unref_client(client, conn, notify->client_entry);
+    }
   }
 
   /* If no entries found, just finish the notify processing */