dot dot is dead

--- linux-2.6.5-rc2.orig/fs/reiser4/txnmgr.c Thu Jan 1 03:00:00 1970
+++ linux-2.6.5-rc2.new/fs/reiser4/txnmgr.c Thu Mar 25 20:29:14 2004
@@ -0,0 +1,4421 @@
+/* Copyright 2001, 2002, 2003 by Hans Reiser, licensing governed by
+ * reiser4/README */
+
+/* Joshua MacDonald wrote the first draft of this code. */
+
+/* ZAM-LONGTERM-FIXME-HANS: The locking in this file is badly designed, and
+a filesystem scales only as well as its worst locking design. You need to
+substantially restructure this code. Josh was not as experienced a programmer
+as you. Particularly review how the locking style differs from what you did
+for znodes usingt hi-lo priority locking, and present to me an opinion on
+whether the differences are well founded. */
+
+/* I cannot help but to disagree with the sentiment above. Locking of
+ * transaction manager is _not_ badly designed, and, at the very least, is not
+ * the scaling bottleneck. Scaling bottleneck is _exactly_ hi-lo priority
+ * locking on znodes, especially on the root node of the tree. --nikita,
+ * 2003.10.13 */

Headache: txnmgr.c
Migraine: flush.c

This post is dedicated to Zam, Nikita, Hans & Co.