simonb@NetBSD.org wrote:

> > splclock()/splx() unmatch?
> 
> Yes indeed.  I was a bit hasty commiting that.
> 
> Please commit your fix.

Some people would hate goto (though I think it's okay for "break" case).
Which is better?
---
Izumi Tsutsui


Index: kern_tc.c
===================================================================
RCS file: /cvsroot/src/sys/kern/kern_tc.c,v
retrieving revision 1.11
diff -u -r1.11 kern_tc.c
--- kern_tc.c	7 Sep 2006 15:48:14 -0000	1.11
+++ kern_tc.c	10 Sep 2006 06:54:43 -0000
@@ -454,17 +454,15 @@
 	 * Even though we run on the dummy counter, switching here may be
 	 * worse since this timecounter may not be monotonous.
 	 */
-	if (tc->tc_quality < 0)
-		return;
-	if (tc->tc_quality < timecounter->tc_quality)
-		return;
-	if (tc->tc_quality == timecounter->tc_quality &&
-	    tc->tc_frequency < timecounter->tc_frequency)
-		return;
-	(void)tc->tc_get_timecount(tc);
-	(void)tc->tc_get_timecount(tc);
-	timecounter = tc;
-	tc_windup();
+	if (tc->tc_quality >= 0 &&
+	    tc->tc_quality >= timecounter->tc_quality &&
+	    (tc->tc_quality > timecounter->tc_quality ||
+	     tc->tc_frequency >= timecounter->tc_frequency)) {
+		(void)tc->tc_get_timecount(tc);
+		(void)tc->tc_get_timecount(tc);
+		timecounter = tc;
+		tc_windup();
+	}
 
 	splx(s);
 }