Many thanks for your explanations Carlo. I'm now using WMaker 0.92 (Debian lenny) and your patch semmes to work fine!

Regards,

Jean-Paul Lacquement


Carlos R. Mafra wrote:

Jean-Paul Lacquement escreveu:
Sorry for replying so late. I am interested in the patch you're talking about. Could you pease tell me how?

OK, nice! Just grab Window Maker 0.92.0 from http://www.windowmaker.info
and apply my humble patch against it. It is available here:
http://www.ift.unesp.br/users/crmafra/patch_wmaker-0.92.0-m3

Note that wmaker 0.92.0 doesn't compile with gcc 4.1. I had to
compile it with gcc 3.3. There is a patch from Mandriva to enable
gcc 4 compilation, but I didn't test it. This patch is also at
the directory "crmafra" from the above link, just in case you
want to use gcc 4.

Well, my patch should be taken "cum grano salis" (reasons on my
previous email). If you are a programmer, I would like to know what exactly I am missing by not executing "delayedAction". I removed the line which creates a timer which calls this function.

The experimental fact is that my patched Window Maker is working without
any problems for a few days now and that it doesn't wake up the cpu
anymore.

Regards,
Carlos R. Mafra

Carlos R. Mafra wrote:

I am not a programmer, but the Window Maker development list is down and the ex-developers are all busy and so far I have not been able to contact any of them. So, as this list is all about wakeups and related
things, I've decided to write this email.

I realized that Window Maker was waking up the cpu at 3-4 Hz when idle,
and Magnus Fromreide alerted me that this is probably a bug. After that
I tried to find out the reason and if I could make my beloved wmaker
reach the 0 wakeups mark.

I have made an amateur patch [1] which does that, and after using this
patched Window Maker without getting my laptop exploded and without
noticing anything strange with its behaviour, I am quite convinced
that it indeed was a bug.

The patch does two things:
1) It removes three "timers" which were called frequently and were the responsables for wmaker showing up in powertop. One of them
was called to check if configuration files had changed, the other was
called to "synchronize" the configuration files (not sure why one needs
that), and the worst of them (called each 500 millisecs) to take care of
some "delayedActions" (I don't understand this one).

2) Adds a dnotify mechanism to tell wmaker that the configuration files have changed, in substitution of the default
polling behaviour. I copied the code from Documentation/dnotify.txt
from the linux kernel source :-)

With my patch, wmaker no longer appears in powertop's list and I believe
that the 0 wakeups when idle goal was reached. Now I can rest in peace.

I am not sure if there is anyone else who cares about wmaker anymore, but if such people still exists and also take care of needless wakeups,
then I would like them to check my humble and very amateur patch.

Regards to all,
Carlos R. Mafra

[1] http://www.ift.unesp.br/users/crmafra/patch_wmaker-0.92.0-m3

Magnus Fromreide escreveu:
On lör, 2007-06-09 at 20:16 -0300, Carlos R. Mafra wrote:
2007/6/9, Magnus Fromreide <[EMAIL PROTECTED]>:
On lör, 2007-06-09 at 16:23 -0300, Carlos R. Mafra wrote:
I just compiled kernel 2.6.21 (with acpi, cpu frequency scaling and so on and HPET patch) on my debian laptop. However, which browser, MUA, desktop manager, etc do you use ? I plain to use a battery optimized
system but don't really know how to.
I use Window Maker for the window manager, which is responsable for 4 wakeups. If I use it with the --no-polling option, then it reduces to 3 wakeups.
Give it a try!
Eh, what? Why should the _window_manager_ generate any wakeups at all?
Well, I don´t really know...so do you think this is a kind of bug in
Window Maker?

It depends - right now it is waking up 4 (or 3) times/s, if it is doing
something useful when it wakes up then it sn't a bug but rather a
feature but if it just wakes up, looks around confusedly and then
decides there is nothing to do then it is a bug.

I thought it was ok to generate some wakeups (to check if some window was moved or something like that), but I am not a programer and I don´t have the
knowledge to judge this behaviour.

That isn't OK, the WM gets a notification whenever there is a window
event that it is subscribed to so there is no need for that.

If this is a bug, maybe the Window Maker developers will start coding
a little bit again :-)

One could hope so. You could probably increase the chances for that to
happen by reporting the bug.

And now I realize that you asked about a _desktop_ manager, not a
window manager...so I am sorry for my suggestion.

Yes, the original poster did so, and that probably do make a difference.
Maybe you meant to send this to the list as well.

However, if this is a bug I hope it gets corrected!

As I said, try reporting it.

/MF



_______________________________________________
Power mailing list
[email protected]
http://www.bughost.org/mailman/listinfo/power






_______________________________________________
Power mailing list
[email protected]
http://www.bughost.org/mailman/listinfo/power

Reply via email to