Should battery generate events?

Should battery generate events?

[Tim Dijkstra]

Hi,

Last week my laptop (Acer Travelmate 636Lci) run out of power resulting
in a unclean shutdown, disk corruption and me spending a few evenings to
get it to sanely work again.

I would want to avoid this happening again. Ideally I would want my
laptop to "shutdown -h now" if battery power is to low (well rather have
it go to S3 or softwaresuspend, but it seems hopeless to get that to
work :(  

>From some googl'ing I get the impression that ACPI should generate
battery events on a change of capacity or at least when it's low. My
idea was to have acpid call some script on battery events. 
The problem is that there doesn't seem to be such events on my machine. 

Should every bios generate these events?
Could it be a bug in my DSDT?
Does it have anything to do with
tdykstra@bonenpikster:/$ cat /proc/acpi/battery/BAT0/alarm
alarm:                   unsupported
?

FYI from my /proc/acpi:

tdykstra@bonenpikster:/$ cat /proc/acpi/battery/BAT0/state
present:                 yes
capacity state:          ok
charging state:          charging
present rate:            unknown
remaining capacity:      3840 mAh
present voltage:         14800 mV
tdykstra@bonenpikster:/$ cat /proc/acpi/battery/BAT0/info
present:                 yes
design capacity:         3920 mAh
last full capacity:      3920 mAh
battery technology:      rechargeable
design voltage:          14800 mV
design capacity warning: 30 mAh
design capacity low:     20 mAh
capacity granularity 1:  10 mAh
capacity granularity 2:  3470 mAh
model number:            Bat0
serial number:
battery type:            Lion
OEM info:                Acer


-------------------------------------------------------
This SF.net email is sponsored by: IT Product Guide on ITManagersJournal
Use IT products in your business? Tell us what you think of them. Give us
Your Opinions, Get Free ThinkGeek Gift Certificates! Click to find out more
http://productguide.itmanagersjournal.com/guidepromo.tmpl

Re: Should battery generate events?

[liste-9nAOAgdJVo4b1SvskN2V4Q]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Salut Tim,
On Tue, 5 Oct 2004, Tim Dijkstra wrote:
> On Tue, 5 Oct 2004 16:25:58 +0200 (CEST)
> liste-9nAOAgdJVo4b1SvskN2V4Q@public.gmane.org wrote:
>> But do you allways run X (i.e. a graphical desktop)? ...because most
>> monitoring applications are graphics driven: There are some solutions
>> that come with KDE and I had a look (some time ago) at kacpi...
>> One problem might be, that your DSDT does not calculate a "present
>> rate".
>
> It does show the correct state while discharging, so that's not the
> problem.
Sorry, but I doubt that, or you do not have the latest BIOS - and the 
introduced this later. The DSDT from r01c0O-BIOS for TM 63x does allways 
say, that the "present-rate" is unknown.
But that does not mean, that there is no solution for you - just that you 
might run into problems, with other's solution, that depend on a defined 
present-rate

> I just wouldn't want to depend on an gui. Several people can
> use this machine, and I do not want to configure their desktops and hope
> they keep it like that.
ok. that would be a cleaner solution. But if you like to nag them, when 
the state drops to "warning", you might want to popup an xterm or other 
x-window. To do that you need to steal the .Xauthority (cookie) from the 
one running the X-server; because root does not have access in a normal 
setup. Just as a hint.

>> If you do not use the proprietary Nvidia-X driver, swsusp2 worked for
>> me. But as the Nvidia-driver is so much faster, I am not able to do a
>> software-suspend too.
>
> I do use the nvidia drivers, but even without X running it won't suspend
> successfully. Did you do anything special to get it to work?
Sorry, it was a bit ago, when I last tried it. I took the swsusp2-patches 
and the latest 2.6-kernel.
If you do experiments with this, be sure to always do a "sync" right as 
the last command, before you try to susupend. This increases the 
likeliness, that the fsck (in case it crashed) to fix it automatically 
without you loosing data. But I said "likeliness". Hope you got the tune 
with this.

Cheers
   hartwig felger

Hartwig Felger informatics

- -- 
1024D/339FD693 Hartwig Felger <hgfelger-9nAOAgdJVo4b1SvskN2V4Q@public.gmane.org>
Key fingerprint = FB2F 3EE9 345A D55B 6FF2  0EC1 F5B0 684F 339F D693
For the pulic keys, please visit my page.
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.2 (GNU/Linux)

iD8DBQFBYxAf9bBoTzOf1pMRAk/AAJ93GLK0QKd4SdccQ1GnQpb/tN1XFwCg7l8G
9OTMFduJAgmm4ElEEwDZuDM=
=I2jH
-----END PGP SIGNATURE-----


-------------------------------------------------------
This SF.net email is sponsored by: IT Product Guide on ITManagersJournal
Use IT products in your business? Tell us what you think of them. Give us
Your Opinions, Get Free ThinkGeek Gift Certificates! Click to find out more
http://productguide.itmanagersjournal.com/guidepromo.tmpl

Re: Should battery generate events?

[Tim Dijkstra]

On Tue, 5 Oct 2004 23:20:22 +0200 (CEST)
liste-9nAOAgdJVo4b1SvskN2V4Q@public.gmane.org wrote:

> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
> 
> Salut Tim,
> On Tue, 5 Oct 2004, Tim Dijkstra wrote:
> > On Tue, 5 Oct 2004 16:25:58 +0200 (CEST)
> > It does show the correct state while discharging, so that's not the
> > problem.
> Sorry, but I doubt that, or you do not have the latest BIOS - and the 
> introduced this later. The DSDT from r01c0O-BIOS for TM 63x does
> allways say, that the "present-rate" is unknown.

Ah, yes present rate is always unknown, but what seemed more important
to me is 'remaining capacity', that does represents a usefull value.
Also the ACPI spec says that the battery/bios could signal that the
battery is almost empty by generating an event if capacity goes under a
certain value (see other mail from me in this thread).

grts TIm


-------------------------------------------------------
This SF.net email is sponsored by: IT Product Guide on ITManagersJournal
Use IT products in your business? Tell us what you think of them. Give us
Your Opinions, Get Free ThinkGeek Gift Certificates! Click to find out more
http://productguide.itmanagersjournal.com/guidepromo.tmpl

Re: Should battery generate events?

[David Bronaugh]

Tim Dijkstra wrote:

>Hi,
>
>Last week my laptop (Acer Travelmate 636Lci) run out of power resulting
>in a unclean shutdown, disk corruption and me spending a few evenings to
>get it to sanely work again.
>
>I would want to avoid this happening again. Ideally I would want my
>laptop to "shutdown -h now" if battery power is to low (well rather have
>it go to S3 or softwaresuspend, but it seems hopeless to get that to
>work :(  
>
>>From some googl'ing I get the impression that ACPI should generate
>battery events on a change of capacity or at least when it's low. My
>idea was to have acpid call some script on battery events. 
>The problem is that there doesn't seem to be such events on my machine. 
>  
>
Whenever polling is required, this is expected to be done in userspace, 
not in kernelspace. There's a long history of people asking "why doesn't 
battery status changes generate events" and a long history of people 
saying "because it would require polling in either acpid or the kernel, 
which is a Bad Thing(tm)".

If you'd like something to watch the battery capacity, perhaps you 
should use something like KLaptop -- you can configure it to run a 
command (like 'echo -n "disk" > /sys/power/state') on low and/or on 
critical battery level warnings. Since it tracks this kind of state, you 
don't have to.

Of course, KLaptop's not the only option -- it's just what I have here.

David Bronaugh


-------------------------------------------------------
This SF.net email is sponsored by: IT Product Guide on ITManagersJournal
Use IT products in your business? Tell us what you think of them. Give us
Your Opinions, Get Free ThinkGeek Gift Certificates! Click to find out more
http://productguide.itmanagersjournal.com/guidepromo.tmpl

Re: Should battery generate events?

[Tim Dijkstra]

On Mon, 04 Oct 2004 15:14:45 -0700
David Bronaugh <dbronaugh-Jp3n8lUXroSX6QiC4yPwbg@public.gmane.org> wrote:

> Tim Dijkstra wrote:
>   
> >From some googl'ing I get the impression that ACPI should generate
> >battery events on a change of capacity or at least when it's low. My
> >idea was to have acpid call some script on battery events. 
> >The problem is that there doesn't seem to be such events on my
> >machine. 
> >  
> >
> Whenever polling is required, this is expected to be done in
> userspace, not in kernelspace. There's a long history of people asking
> "why doesn't battery status changes generate events" and a long
> history of people saying "because it would require polling in either
> acpid or the kernel, which is a Bad Thing(tm)".

Hmm, sounds reasonable ... But what do these entries in
/proc/acpi/battery/BAT0/info mean:

design capacity warning: 30 mAh
design capacity low:     20 mAh

it seems to imply to me that there should be a warning of some kind when
the capacity hits 30 mAh. But if I read your post correctly there is no
such thing?

> If you'd like something to watch the battery capacity, perhaps you 
> should use something like KLaptop -- you can configure it to run a 
> command (like 'echo -n "disk" > /sys/power/state') on low and/or on 
> critical battery level warnings. 

Yeah, but several people log in on this laptop (wife, kid (well not
yet), brothers, sister ...) and that means I have to set up their
desktops and hope they don't accidentally disable it. It seemed a much
cleaner solution to have a gui/user independent process handle it.

It seems I have to hack a little script that polls the battery status
ones in a while and takes appropriate action...

grts Tim 


-------------------------------------------------------
This SF.net email is sponsored by: IT Product Guide on ITManagersJournal
Use IT products in your business? Tell us what you think of them. Give us
Your Opinions, Get Free ThinkGeek Gift Certificates! Click to find out more
http://productguide.itmanagersjournal.com/guidepromo.tmpl

Re: Should battery generate events?

[Tim Dijkstra]

On Tue, 5 Oct 2004 20:28:44 +0200
Tim Dijkstra <newsuser-TpnWJeB+oO0hKRfeEwdQzg@public.gmane.org> wrote:

> On Mon, 04 Oct 2004 15:14:45 -0700
> David Bronaugh <dbronaugh-Jp3n8lUXroSX6QiC4yPwbg@public.gmane.org> wrote:
> 
> > Tim Dijkstra wrote:
> >   
> > >From some googl'ing I get the impression that ACPI should generate
> > >battery events on a change of capacity or at least when it's low.
> > >My idea was to have acpid call some script on battery events. 
> > >The problem is that there doesn't seem to be such events on my
> > >machine. 
> > >  
> > >
> > Whenever polling is required, this is expected to be done in
> > userspace, not in kernelspace. There's a long history of people
> > asking"why doesn't battery status changes generate events" and a
> > long history of people saying "because it would require polling in
> > either acpid or the kernel, which is a Bad Thing(tm)".
> 
> Hmm, sounds reasonable ... But what do these entries in
> /proc/acpi/battery/BAT0/info mean:
> 
> design capacity warning: 30 mAh
> design capacity low:     20 mAh
> 
> it seems to imply to me that there should be a warning of some kind
> when the capacity hits 30 mAh. But if I read your post correctly there
> is no such thing?

Replying to my self here ... 

Readying the ACPI spec I learned there _is_ such a thing, but it's not
required. And browsing through my dsdt I see there's no _BTP method,
which probably means my battery doesn't support that:( 
After reading battery.c I can further reply to myself in my first
post:

> > > Does it have anything to do with
> > > tdykstra@bonenpikster:/$ cat /proc/acpi/battery/BAT0/alarm
> > > alarm:                   unsupported

Yes, this means there's no _BTP and hence no warning events.

So polling is the only option for me ...


grts Tim


-------------------------------------------------------
This SF.net email is sponsored by: IT Product Guide on ITManagersJournal
Use IT products in your business? Tell us what you think of them. Give us
Your Opinions, Get Free ThinkGeek Gift Certificates! Click to find out more
http://productguide.itmanagersjournal.com/guidepromo.tmpl

Re: Should battery generate events?

[Stefan Seyfried]

On Mon, Oct 04, 2004 at 11:25:57PM +0200, Tim Dijkstra wrote:
 
> I would want to avoid this happening again. Ideally I would want my
> laptop to "shutdown -h now" if battery power is to low (well rather have
> it go to S3 or softwaresuspend, but it seems hopeless to get that to
> work :(  

Well, there are user space tools that will do this for you.
SUSE provides powersaved which will poll your battery if it doesn't support
events and take apropriate action (default is shutdown if battery is <2%).
It is also caching the data so that your battery applet does not have to
poll the battery by itself (which is ridiculously expensive on many
machines) but it only has to ask powersaved.

Regarding suwsusp: this should work on most i386 machines out of the box
with upcoming SUSE 9.2.


> From some googl'ing I get the impression that ACPI should generate
> battery events on a change of capacity or at least when it's low. My
> idea was to have acpid call some script on battery events. 
> The problem is that there doesn't seem to be such events on my machine. 

well, if you had bought a (e.g.) hp nx5000 you would get events for every
charge state change (every remaining capacity change, not only alarms).
On the Dell D600, i get at least alarms but have to poll the battery for
the actual level and on acer you get nothing. Go bug acer.

-- 
Stefan Seyfried



-------------------------------------------------------
This SF.net email is sponsored by: IT Product Guide on ITManagersJournal
Use IT products in your business? Tell us what you think of them. Give us
Your Opinions, Get Free ThinkGeek Gift Certificates! Click to find out more
http://productguide.itmanagersjournal.com/guidepromo.tmpl

Re: Re: Should battery generate events?

[Tim Dijkstra]

On Fri, 8 Oct 2004 18:10:46 +0200
Stefan Seyfried <seife-T+3qhDtrTvDv8lqoKu5MV4HBZQ1bDw0s@public.gmane.org> wrote:

> On Mon, Oct 04, 2004 at 11:25:57PM +0200, Tim Dijkstra wrote:
>  
> > I would want to avoid this happening again. Ideally I would want my
> > laptop to "shutdown -h now" if battery power is to low (well rather
> > have it go to S3 or softwaresuspend, but it seems hopeless to get
> > that to work :(  
> 
> Well, there are user space tools that will do this for you.
> SUSE provides powersaved which will poll your battery if it doesn't
> support events and take apropriate action (default is shutdown if
> battery is <2%). It is also caching the data so that your battery
> applet does not have to poll the battery by itself (which is
> ridiculously expensive on many machines) but it only has to ask
> powersaved.

I could not find a lot about it on the web, then downloaded the src rpm
from suse. Looks interesting ... I do not use suse myself but maybe I
can get (parts of it) to work on my system.

> Regarding suwsusp: this should work on most i386 machines out of the
> box with upcoming SUSE 9.2.

I find that a bold statement if I look at the number of posts to this
and the swsusp2 list in which people tell they can't get suspend to
work. 

grts Tim


-------------------------------------------------------
This SF.net email is sponsored by: IT Product Guide on ITManagersJournal
Use IT products in your business? Tell us what you think of them. Give us
Your Opinions, Get Free ThinkGeek Gift Certificates! Click to find out more
http://productguide.itmanagersjournal.com/guidepromo.tmpl

Re: Re: Should battery generate events?

[Stefan Seyfried]

On Tue, Oct 12, 2004 at 08:20:10PM +0200, Tim Dijkstra wrote:
> On Fri, 8 Oct 2004 18:10:46 +0200

> I could not find a lot about it on the web, then downloaded the src rpm
> from suse. Looks interesting ... I do not use suse myself but maybe I
> can get (parts of it) to work on my system.

we are currently working on getting it more portable, it is pretty
SUSE-centric right now but this should change soon. If it builds and
runs on other distributions, tarballs and packages will be available
from forge.novell.com (the CVS is hosted there).
 
> > Regarding suwsusp: this should work on most i386 machines out of the
> > box with upcoming SUSE 9.2.
> 
> I find that a bold statement if I look at the number of posts to this
> and the swsusp2 list in which people tell they can't get suspend to
> work. 

It worked at least on every i386 notebook i could get my hands on during
beta and that were a lot of machines. I did not test many desktops (i don't
have many of them :-) but i really believe it should work out of the box
on most i386 notebooks. Of course there are known show-stoppers like the
binary NVidia drivers but these should fail gracefully and there is not
much we can do about them.
-- 
Stefan Seyfried



-------------------------------------------------------
This SF.net email is sponsored by: IT Product Guide on ITManagersJournal
Use IT products in your business? Tell us what you think of them. Give us
Your Opinions, Get Free ThinkGeek Gift Certificates! Click to find out more
http://productguide.itmanagersjournal.com/guidepromo.tmpl

Re: Re: Should battery generate events?

[Pavel Machek]

Hi!

> > Regarding suwsusp: this should work on most i386 machines out of the
> > box with upcoming SUSE 9.2.
> 
> I find that a bold statement if I look at the number of posts to this
> and the swsusp2 list in which people tell they can't get suspend to
> work. 

SuSE 9.2 uses swsusp1, similar to version in 2.6.9-rc4, with some
support in userland (module unloading). It tends to work very well.
								Pavel
-- 
People were complaining that M$ turns users into beta-testers...
...jr ghea gurz vagb qrirybcref, naq gurl frrz gb yvxr vg gung jnl!


-------------------------------------------------------
This SF.net email is sponsored by: IT Product Guide on ITManagersJournal
Use IT products in your business? Tell us what you think of them. Give us
Your Opinions, Get Free ThinkGeek Gift Certificates! Click to find out more
http://productguide.itmanagersjournal.com/guidepromo.tmpl