wake event?

wake event?

[Brian Perkins]

I've started playing around with the acpid socket for writing power
management clients.

One deficiency that I've noticed is that events don't arrive while the 
machine is sleeping (not too surprising).  This means that things like 
ac status monitors will be incorrect if the ac status changes while we 
are sleeping.

It might be nice to deliver a wake event to /proc/acpi/event so clients
can know that they can update properly.

I thought about writing a patch to do this, but I had trouble deciding
what arguments ought to be given to acpi_bus_generate_event (device 
e.g.), and where it should be done.  Perhaps this is because this 
feature is somewhat outside the scope of what was originally intended 
the interface.

Any thoughts?

--
Brian Perkins				bperkins-5L9LcNmrgCtAfugRpC6u6w@public.gmane.org




-------------------------------------------------------
This SF.net email is sponsored by: SF.net Giveback Program.
Does SourceForge.net help you be more productive?  Does it
help you create better code?  SHARE THE LOVE, and help us help
YOU!  Click Here: http://sourceforge.net/donate/

RE: wake event?

[Yu, Luming]

>I've started playing around with the acpid socket for writing power
>management clients.

Could you please give me some detail about your power management client?

>One deficiency that I've noticed is that events don't arrive while the 
>machine is sleeping (not too surprising).  

Is it a reasonable requirement?  when sleeping, only wake up event should
be noticed.  If wake up event can not wake up machine, we need fix it!

--Luming



-------------------------------------------------------
This SF.net email is sponsored by: SF.net Giveback Program.
Does SourceForge.net help you be more productive?  Does it
help you create better code?  SHARE THE LOVE, and help us help
YOU!  Click Here: http://sourceforge.net/donate/

Re: wake event?

[Brian Perkins]

Yu, Luming wrote:
> Could you please give me some detail about your power management client?
> 

Well, one is good old (bad old) battstat, which has been recently 
patched to use the acpid unix socket to receive battery and AC events, 
instead of polling once per second.  This is a pretty big win on my 
machine, and an even bigger win on other machines.

I also wrote a simple perl script to manage cpufreq transitions based on 
AC status, mostly for fun.

If the laptop is unplugged while sleeping, the battstat applet is 
incorrect on waking, and my cpufreq daemon doesn't go to low power mode.

I can't think of a way of dealing with this, other than polling, which 
is dumb and bad.

I suppose this could be considered a userspace problem, but the current 
design seems to make this a little complicated, since acpid doesn't 
really know what its actions do.


-------------------------------------------------------
This SF.net email is sponsored by: SF.net Giveback Program.
Does SourceForge.net help you be more productive?  Does it
help you create better code?  SHARE THE LOVE, and help us help
YOU!  Click Here: http://sourceforge.net/donate/

RE: wake event?

[Yu, Luming]

>If the laptop is unplugged while sleeping, the battstat applet is 
>incorrect on waking, and my cpufreq daemon doesn't go to low power mode.

After waking from sleeping, your applet should refresh its devices status information through /proc/acpi/ 
interfaces


-------------------------------------------------------
This SF.net email is sponsored by: SF.net Giveback Program.
Does SourceForge.net help you be more productive?  Does it
help you create better code?  SHARE THE LOVE, and help us help
YOU!  Click Here: http://sourceforge.net/donate/