[arch-dev-public] [signoffs] openvpn, mkinitcpio, cryptsetup, initscripts

[Gerhard Brauer]

Am Freitag, den 31.07.2009, 01:31 +0300 schrieb Roman Kyrylych:
> On Thu, Jul 30, 2009 at 23:22, Gerhard Brauer<gerhard.brauer at web.de> wrote:
>
> > Generally you're right, but this package contains only this commit:
> > http://projects.archlinux.org/?p=initscripts.git;a=commit;h=92cbc62e1c1b22c6cc251628e8d51a45607998b6
> 
> I hope this commit was added too:
> http://projects.archlinux.org/?p=initscripts.git;a=commitdiff;h=28d34292a9a8a03ea78a909ccb1d8d50ee21d0bb
> ;-)

That is a little confusing ;-) I never committed to initscripts, i sent
a patch in FS#15263. I think Aaron wrote the commit and had a typo.
In the modified initscripts 2009.03-2 we use my patch from the flyspray
report.

> I don't get it - the change exists in the core package anyway.
> What am I missing?

No, it's not in core (initscripts 2009.03-2).
Therein we have:
--------
HWCLOCK_PARAMS="--systohc"
if [ "$HARDWARECLOCK" = "UTC" ]; then
        HWCLOCK_PARAMS="$HWCLOCK_PARAMS --utc"
else
        HWCLOCK_PARAMS="$HWCLOCK_PARAMS --localtime"
fi
-------------
And this is the problem: We assume automatic localtime if $HARDWARECLOCK
is not UTC. But during boot of the LiveCD $HARDWARECLOCK isn't set to
any value by default - only the user could set it by cmdline.

This lead to the problem for UTC bios clock users...

initscripts 2009.07-3 handle this now better (also for normal use in a
normal system): It changes hwclock *only* if a defined value got parsed
from rc.conf - either localtime or UTC. Otherwise bios clock is not
touched.
And the live CD benefits most from this change: it avoids that a user
found his hardware was modified against the state it was before using
the LiveCD.

Gerhard