re active the network mesh when suspends the system

[Martin Abente]

On Fri, Dec 17, 2010 at 5:35 PM, Paul Fox <pgf at laptop.org> wrote:

> martin wrote:
>  > I have done a few test in 3 different builds, the tests are:
>  >
>  > A. Killing the mesh with "echo 0 > /sys/class/net/eth0/lbs_mesh", then
>  > suspend and then check if the mesh is re-activated after resume.
>  > B. While the mesh is active, add a script in postresume.d with the same
> line
>  > to kill the mesh (granting it run permission), then suspend and then
> check
>  > if the mesh is de-activated after resume.
>  >
>  > The builds I used for the test are:
>  >
>  > * olpc os852 (powerd 26, kernel 2.6.31_xo1.20100823.1641.1.olpc)
>  > * olpc os359 (powerd 32, kernel 2.6.31_xo1-20101119.1249.1.olpc)
>  > * last dextrose 2 (powerd 32, kernel 2.6.31_xo1-20101216.1250.1.olpc)
>  >
>  > Results:
>  >
>  > For build os582:
>  > A: the mesh is _not_ being re-activated (this is the behaviour we need!)
>  > B: postresume.d scripts are not supported. (but if does not matter
> because
>  > with A is enough)
>
> there was a bug in os852 which prevented us from ever powering off
> the wlan during suspend.  it was as if config_MESH_DURING_SUSPEND was
> always turned on.
>
> have your script sleep for 10 or 20 seconds before disabling mesh, and
> see if that helps.  there may be a better solution, but that would give
> some more data.
>
>
I wrote this script that is a little bit better than just waiting N seconds,
and it seems to work fine:

LBS_MESH=/sys/class/net/eth0/lbs_mesh

while [ ! -f $LBS_MESH ] || [ $(cat $LBS_MESH) = "0x0" ]
do
    sleep 0.5s
done

echo 0 > $LBS_MESH

But I like Martin's idea better, just not sure how to make it work atm.



> paul
>
>  >
>  > For build os359:
>  > A: the mesh is re-activated. :(
>  > B: the mesh is not being de-activated (even though the script
> effectively
>  > runs, something else must be happening _after_ postresume.d script that
>  > re-activates the mesh)
>  >
>  > For build last dextrose 2:
>  > A: the mesh is re-activated. :(
>  > B: the mesh is not being de-activated (it is probably the same problem
> as
>  > with os359)
>  >
>  > Comments:
>  >
>  > I am not sure why the os852 build behaves as we need, while any of the
>  > newest builds does not. Could it be a regression in powerd?
>
> =---------------------
>  paul fox, pgf at laptop.org
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.laptop.org/pipermail/devel/attachments/20101217/4a525d59/attachment.html>