hwclock.sh should be called before/after hibernate

It's already called after hibernate, but not before.

(patch for acpi-support, not util-linux)

The attached patch modifies /etc/acpi/resume.d/50-time.sh to use the correct initscript to load hwclock state, instead of calling hwclock manually. Also, /etc//acpi/resume.d/88-time.sh was added, to save hwclock state before hibernation.

I chosed "88" for suspend to make sure the clock is the last step done before video is turned off.

This patch was successfully tested on a locally compiled acpi-support package, in two different computers - but it would be great if someone else could review/test it on different hardware.

Use case: clock set to local time (dual boot with Windows). A daylight savings transition happens, but the clock is not written out to hardware. Computer is suspended, and when resumed the timezone is correct, but the time has reverted to the time before the transition.

There are some relevant comments in bug 91418: the last comments by Andrew Bennetts in particular are actually related to this bug.

I was affected by this bug when daylight saving transition happened last week. Although I tried to update the clock it manually over and over again, it was reset everytime I suspended the computer (I usually never shut it off completely).

Stopping and starting hwclock.sh did the trick so I'm pretty sure the patch will solve the problem (although I didn't try it)

Use case (happened here right now):
Clock is off by X minutes for some reason. The running ntpd will reset it to the correct time after several minutes.
Then you suspend your laptop, resume it, clock is off by X minutes again.

The patch for this is attached to bug 193842.
You can get a test package for acpi-support from my PPA at
https://launchpad.net/~blueyed/+archive?field.name_filter=acpi-support (0.106blueyedppa5 or later)

@Daniel:
I couldn't test the whole package because i'm on Gutsy and Gutsy misses the dependecy package x11-xserver-utils.

But i tested the /etc/acpi/suspend.d/88-time.sh extracted from your package.
Works as expected!

root@laj:/etc/acpi/resume.d# date; hwclock --show
Don Mär 6 11:14:57 CET 2008
Don 06 Mär 2008 12:14:55 CET -0.122513 seconds
###SUSPEND-RESUME###
root@laj:/etc/acpi/resume.d# date; hwclock --show
Don Mär 6 11:15:29 CET 2008
Don 06 Mär 2008 11:15:30 CET -0.089833 seconds

This bug was fixed in the package acpi-support - 0.106

---------------
acpi-support (0.106) hardy; urgency=low

  * Cherrypicked bugfix patches from Launchpad (LP: #193842)
  * lib/state-funcs: make wireless led work with asus-laptop (the new module);
    patch from Nicolò Chieffo (LP: #189889)
  * Add events/asus-video to support asus switch video hotkey; patch from
    Nicolò Chieffo (LP: #138228)
  * events/asus-wireless-2, asus-wireless-2.sh:
    Support Fn+F2 on Asus A8Js; patch from Matteo Collina (LP: #105929)
  * Fix the value range in sonybright.sh (0-7 instead of 1-8); patch from unggnu
    (LP: #136380)
  * "lib/FUJITSU SIEMENS.config":
    - Enable suspend for Fujitsu-Siemens C1110 (LP: #52970)
    - Enable STR for Fujitsu-Siemens Lifebook S7020, S7110
  * resume.d/50-time.sh, suspend.d/88-time.sh:
    Fix handling of hardware clock before/after suspend; patch from Rodrigo
    Novo (LP: #36815)
  * events/asus-*: Fix event key names for ASUS (HOTK); patch from
    Jeroen van der Vegt (LP: #25784)
  * Fix names in comments for events/panasonic-{mute,volume-down,volume-up}
  * debian/rules, README.toshiba: Drop toshiba_acpi.modprobe, which only
    contained an invalid/obsolete option "hotkeys_over_acpi" (LP: #180678)
  * debian/control: Standards-Version 3.7.3
  * Drop thinkpad_acpi.modprobe: the workarounds therein are not needed for
    thinkpad_acpi in Hardy anymore; investigations by Jerone Young (LP: #194679)
  * Add events/lenovo-eject to handle the Thinkpad eject button (Fn+F9); patch
    from Jerone Young (LP: #194609)

 -- Daniel Hahler <email address hidden> Thu, 06 Mar 2008 02:15:54 +0100