While testing a clock synchronization software on DragonFly I observed a
problem with the adjtime(2) system call. The retrieved olddelta values
appeared to be quite incorrect, and the reason for this eventually turned
out to be an elementary bug which can be fixed by the patch below. Please
have a look at it, I'd think this case is pretty clear and needs no further
description.
Frank Josellis
--- patch begins here ---
--- sys/kern/kern_time.c.orig 2005-10-08 14:24:26.000000000 +0200
+++ sys/kern/kern_time.c 2006-03-18 23:53:37.000000000 +0100
@@ -547,7 +547,7 @@
if (uap->olddelta) {
atv.tv_sec = odelta / 1000000000;
- atv.tv_usec = odelta % 1000000 / 1000;
+ atv.tv_usec = odelta % 1000000000 / 1000;
(void) copyout((caddr_t)&atv, (caddr_t)uap->olddelta,
sizeof(struct timeval));
}
--- patch ends here ---
