Thanks to the new debug infrastructure (part of 1.52 release), it becomes very 
easy to now figure out what each individual module is doing.

The patch I proposed in previous mail was already part of wireless-ipw-power 
module (ipw* is the wireless device that I use) but power saving never came 
into effect. And I was lazy to figure out.

With the new debug help, I have now known the problem. Reliance on iwconfig is 
going to work for the near future (till kernel breaks iwconfig). 

The problem was the I wasn't gathering the exit code of iwconfig properly. 
Fixed for 1.52. Works for ipw* devices. Will also work for iwl* devices.


PS: If you'd like to test, pm me. I can provide you a test deb.

Ritesh


On Saturday 19 Sep 2009 13:13:35 Ritesh Raj Sarraf wrote:
> Here's the final patch I've committed.
> 
> === modified file 'usr/share/laptop-mode-tools/modules/wireless-iwl-power'
> --- usr/share/laptop-mode-tools/modules/wireless-iwl-power      2009-09-18
> 05:26:58 +0000
> +++ usr/share/laptop-mode-tools/modules/wireless-iwl-power      2009-09-19
> 07:41:49 +0000
> @@ -35,6 +35,16 @@
>                          else
>                                  log "VERBOSE" "$DEVICE doesn't seem to be
> enabled. Radio Switched off?";
>                         fi
> +                else
> +                        # LP: #369113
> +                        # Kernel's 2.6.29 and above have been reported to
>  be missing
> +                        # the $DEVICE/wireless folder.
> +                        dev=`basename $DEVICE`
> +                        ret=`$IWCONFIG $dev >/dev/null 2>&1`;
> +                        if [ "$ret" = "0" ]; then
> +                               # add the interface name to the list
> +                               WIFI_IFNAMES="$WIFI_IFNAMES ${DEVICE##*/}"
> +                        fi
>                 fi
>         done
>  }
> 
> On Saturday 19 Sep 2009 13:10:15 Ritesh Raj Sarraf wrote:
> > Hello Darren,
> >
> > On Saturday 19 Sep 2009 04:28:40 Darren Hoo wrote:
> > > > Again, on my box with ipw2200 device, 2.6.30 still presents the
> > > > wireless folder.
> > >
> > > with my 3945ABG card, 2.6.30 does not have this folder either.
> > > Isn't there  a better way to check whether it is a wireless card?
> > >
> > > Also I want to know  how does wireless card power saving work?
> > > I know that I can set the signal power of the card using iwconfig like
> > > # iwconfig wlan0 txpower 10
> > >
> > > but changing /sys/class/net/*/device/power_level  does not change
> > > the tx-power that iwconfig reports.
> > >
> > > does this power_level have anything to do with that txpower?
> > > will reducing txpower by iwconfig  help power saving at all?
> >
> > Following is what I had done for the wireless-ipw-power module
> >
> > # LP: #369113
> >                                 # Kernel's 2.6.29 and above have been
> >  reported to be missing
> >                                 # the $DEVICE/wireless folder.
> >                                 dev=`basename $DEVICE`
> >                                 ret=`$IWCONFIG $dev >/dev/null 2>&1`;
> >                                 if [ "$ret" = "0" ]; then
> >                                         # add the interface name to the
> >  list WIFI_IFNAMES="$WIFI_IFNAMES ${DEVICE##*/}"
> >                                 fi
> >
> >
> > Perhaps, the same should be done for iwlwifi cards also. What we do here
> > is just run iwconfig on the device. If it is not a wireless device, it
> > would give you a bad exit status. That is what we are relying upon.
> >
> > I'll do the same for iwl-power module also.
> >
> > I'm not sure about the power saving. Wild guess is that if you know what
> >  your workload is (eg. server with a wireless card), you can power tune
> >  your device.
> >
> > Ritesh
> 

-- 
Ritesh Raj Sarraf
RESEARCHUT - http://www.researchut.com
"Necessity is the mother of invention."

Attachment: signature.asc
Description: This is a digitally signed message part.

Reply via email to