Bugzilla Bug 133743

/usr/share/hal/hal-system-power-suspend needs detection script for either vbe
tool 
or for hibernate-ram. (For advanced hibernation configuration). Works with
gnome-power-manager. (deleted other configuration added lines
/usr/sbin/hibernate-ram)

(In reply to comment #0)
> /usr/share/hal/hal-system-power-suspend needs detection script for either vbe
> tool 
> or for hibernate-ram. (For advanced hibernation configuration). Works with
> gnome-power-manager. (deleted other configuration added lines
> /usr/sbin/hibernate-ram)
> should have added script
elseif [ -x "/usr/sbin/hibernate-ram" ] ; then
/usr/sbin/hibernate-ram
RET=$?

gnome-power-manager does not auto reload config on boot. 
To fix it added 
> elseif [ -x "/usr/sbin/hibernate-ram" ] ; then
> /usr/sbin/hibernate-ram
> killall gnome-power-manager 
>RET=$?
> gnome-power-manager
now it reloads with the correct ammount of power (normally shows empty)

DBKoepke,

You're really not being clear on what the issue is... what needs to be changed.
etc.

basically the file /usr/share/hal/hal-system-power-suspend needs a detection
script for hibernate-ram, on my system and probably many others, suspend only
works with the hibernate-script, or vbetool. If you could add vbetool support
to the hald scripts than that would work, however it would be easier if you
just added the lines above so then it will execute 'hibernate-ram' instead of
'echo "mem" > /sys/power/state' (which is the default). This is really only
important for those people who have to use the suspend2-sources kernel for
suspend. 

Created an attachment (id=88413) [edit]
patch to add real suspend capability to gnome-power-manager

here is the patch, tested and working here, please consider adding it to the
gnome-power-manager ebuild.

The issue is with HAL, not gnome-power-manager

I agree, I should have directed this towards HAL and not gnome as this is a
global HAL problem. 

Fixed the patch and added it.

shall we report upstream as a patch for supporting gentoo ?

this is just a patch for people that have to use the hibernate-script in order
to suspend to ram. This is not a gentoo specific problem, anyone that uses the
hibernate script (even under another distrobution) will have the same problem.
It may be better to report this directly to the HAL project, however I do not
know if the hibernate-script is part of another distrobution or not. If someone
else agrees that this is neccesairy, then I will report it directly to hal, and
not as a gentoo-specific patch. 

At the moment, the patch doesn't work. I should have mentioned, the script must
go before is detects the raw kernel interface otherwise it will always run echo
mem /sys/power/state. 
it should look like this:
        elif [ -x "/usr/sbin/hibernate-ram" ] ; then
           # Use hibernate-ram if detected
           /usr/sbin/hibernate-ram
           RET=$?
        elif [ -w "/sys/power/state" ] ; then
            # Use the raw kernel sysfs interface
            echo "mem" > /sys/power/state
            RET=$?
        else
            # TODO: add other scripts support
            unsupported
            fi
sorry for the mix up

isn't it in hal-0.5.7-r3 already ?

(In reply to comment #12)
> isn't it in hal-0.5.7-r3 already ?

yes it is, however because it is placed after :
        elif [ -w "/sys/power/state" ] ; then
            # Use the raw kernel sysfs interface
            echo "mem" > /sys/power/state
            RET=$?
hibernate-ram never runs because the HAL always detects the raw kernel system
interface and executes echo "mem" > /sys/power/state. (in order for suspend2 to
work, you have to also have "mem" in /sys/power/state). The only way this will
work is if the hibernate-ram detection is placed before the raw kernel
interface is detected. 

(In reply to comment #13)
> (In reply to comment #12)
> > isn't it in hal-0.5.7-r3 already ?
> 
> yes it is, however because it is placed after :
>         elif [ -w "/sys/power/state" ] ; then
>             # Use the raw kernel sysfs interface
>             echo "mem" > /sys/power/state
>             RET=$?
> hibernate-ram never runs because the HAL always detects the raw kernel system
> interface and executes echo "mem" > /sys/power/state. (in order for suspend2 to
> work, you have to also have "mem" in /sys/power/state). The only way this will
> work is if the hibernate-ram detection is placed before the raw kernel
> interface is detected. 
> 

I'm very supprised that this bug is still there:(

Nobody cares about it?

Created an attachment (id=97825) [edit]
revised suspend patch for hibernate-script

Can we please patch the existing, stable (x86) package, and push it out?  It's
BROKEN for me on may laptop (#@%$! ATI).  This should fix the problem, since I
*must* use hibernate-ram to have a shot at having anything but funky lines show
up on my screen after suspend.

*** Bug 141694 has been marked as a duplicate of this bug. ***

*** Bug 152829 has been marked as a duplicate of this bug. ***

*** Bug 152163 has been marked as a duplicate of this bug. ***

*** Bug 158004 has been marked as a duplicate of this bug. ***

*** Bug 158178 has been marked as a duplicate of this bug. ***

*** Bug 161131 has been marked as a duplicate of this bug. ***

*** Bug 161131 has been marked as a duplicate of this bug. ***

I think hibernate-ram was wrongly added to hal-system-power-hibernate too.

I noticed the new script /usr/share/hal/scripts/hal-system-power-hibernate in
sys-apps/hal-0.5.7.1-r2 contains (before the "hibernate section"):

[...]
elif [ -x "/usr/sbin/hibernate-ram" ] ; then
  /usr/sbin/hibernate-ram
  RET=$?
[...]

We need to drop them out from the hibernate script (and leave it only in the
suspend script).
For example, with the actual version, if I try to hibernate the system from the
gnome-power-manager my pc going to suspend.

*** Bug 162325 has been marked as a duplicate of this bug. ***

Note that the two patches here are almost identical but different in an
important way: the older patch tries hibernate-ram *after* the dumb kernel
suspend, which means hibernate-ram never gets used. We want the second patch,
instead.

(From update of attachment 88413 [edit])
To make this crystal clear, I'm marking my patch as obsolete.

Fixed in hal-0.5.7.1-r3

The ebuild, as of now, fails to unpack:

 * Failed Patch: hal-0.5.7.1-hibernate-fix-r1.patch !
 *  ( /usr/portage/sys-apps/hal/files/hal-0.5.7.1-hibernate-fix-r1.patch )

Well since neither of the patches actually apply to anything, you guys are
going to have to deal with it broken for some more time. Until someone can
provide a working patch.

Notice bug 162624 about the fact that the patch from comment 15 won't apply.
There is a conflict between that one and hal-0.5.7-hald-scripts.patch.
So if this patch is what you want, make those two patches play nice.

Should hibernate-ram be before or after this block?

elif  [ "$HAL_PROP_POWER_MANAGEMENT_TYPE" = "pmu" ]; then
  #PMU systems cannot use /sys/power/state yet, so use a helper to issue an
ioctl
  hal-system-power-pmu sleep

Created an attachment (id=107396) [edit]
hal-0.5.7-hald-scripts.patch

Drop in replacement for current hal-0.5.7-hald-scripts.patch, this one should
do the right thing (tm),

you should remove :
hal-0.5.7-hibernate.patch
hal-0.5.7.1-hibernate-fix.patch

which are either obsoleted by this patch or don't do what they are supposed to
do.

It seems this trivial bug will never be fixed :(

I've attempted to fix it 4 times. I don't use hibernate support at all,
unfortunately, NO one can provide a functional patch ever.

I will try Gilles latest patch for the next revision.

Doug, thanks for taking care of this. I hope I can help with hibernate stuff as
soon as I'll be a gentoo dev (which I hope should be soon).

*** Bug 158527 has been marked as a duplicate of this bug. ***

Still does not apply.

Fixed for hal-0.5.7.1-r4

yes, the scripts are good now, thanks you so much :)