superfluous hardirq_endlock in bh_action

[Paul Mackerras]

It looks to me like the call to hardirq_endlock in bh_action is
superfluous, since it doesn't match up with any hardirq_trylock call
that I could see.  Perhaps it should be removed?

Paul.

diff -urN linux-2.5/kernel/softirq.c pmac-2.5/kernel/softirq.c
--- linux-2.5/kernel/softirq.c	Sun Jul 28 22:47:57 2002
+++ pmac-2.5/kernel/softirq.c	Mon Jul 29 12:29:34 2002
@@ -293,7 +293,6 @@
 	if (bh_base[nr])
 		bh_base[nr]();
 
-	hardirq_endlock();
 	spin_unlock(&global_bh_lock);
 	return;