[LaMont Jones]
> I completely agree on the severity, and would love to hear from the
> people who convinced me this would work. again.
I guess that was me and Scott. :)
The idea was to allow the init.d scripts detect when udev would take
care of setting the system clock from the hardware clock when the rtc
device became available, and make sure the scripts became no-ops in
these cases to save 1-2 seconds during boot.
I was told that udev would run the udev rule to set the system clock
if it was enabled, and thus believed it would be sufficient to look
for the existence of /dev/.udev to decode when to disable the init.d
scripts. Obviously this assumtion is not quite right. I have asked
Marco d'Itri, the udev maintainer, about this and he believed such
udev rule would be called during boot even if no kernel module is
loaded.
Anyway, a workaround if the udev rule only work when rtc is a kernel
module, would be a patch like this, to only disable the init.d scripts
if rtc is a loaded module and udev is enabled.
diff -ur util-linux-2.16.1/debian/hwclock.sh
util-linux-2.16.1-pere/debian/hwclock.sh
--- util-linux-2.16.1/debian/hwclock.sh 2009-09-25 10:49:27.000000000 +0200
+++ util-linux-2.16.1-pere/debian/hwclock.sh 2009-09-25 10:54:22.000000000
+0200
@@ -67,7 +67,9 @@
case "$1" in
start)
- if [ -d /dev/.udev ]; then
+ # No need to set the system clock if udev take care of it.
+ # This happen when rtc was loaded as a module by udev.
+ if [ -d /dev/.udev ] && grep -qw ^rtc /proc/modules; then
return 0
fi
It should solve the described problems until we figure out why the
udev rule is not called for home built kernels.
Happy hacking,
--
Petter Reinholdtsen
--
To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
