- bugfix: resolved potential segfault condition on HUP (extremely unlikely

    to happen in practice), for details see tracker:
    http://bugzilla.adiscon.com/show_bug.cgi?id=38

1 files changed, 12 insertions, 0 deletions

diff --git a/modules.c b/modules.c
index f04c4a97..406cf32a 100644
--- a/modules.c
+++ b/modules.c
@@ -378,6 +378,18 @@ rsRetVal modUnloadAndDestructDynamic(void)
 		}
 	}
 
+	/* Note: the last modules pNext pointer is now invalid
+	 * (except if the last module was not touched, what is highly
+	 * unlikely. We simply fix this be setting it to NULL. After all, 
+	 * it is the last module ;). This bug had some severe effects in
+	 * v3, but none in v2 because in v2 the list was never again
+	 * traversed before a new one was added. But even in v2 it may cause
+	 * a segfault if the number of loaded modules changed between HUPs.
+	 * rgerhards, 2008-02-26
+	 */
+	if(pLoadedModulesLast != NULL)
+		pLoadedModulesLast->pNext = NULL;
+
 	return iRet;
 }
 /*