[Debian-eeepc-devel] suspend.sh in eeepc-acpi-scripts 1.02 and NetworkManager

[elijah r.]

>  Most of the cases, ot all "just works".
>
>  When it does not, I think the problem is somewhere in the madwifi thing,
>  as turning the wireless off and then back on does not succeed. The blue
>  LED lights up and then shuts off. I guess this is because the ath0
>  interface cannot be brought up.

I have had this problem occasionally as well, where I have to try to
turn wireless back on twice before it works.
The first time, the blue LED lights up, and then it turns off again.

>
>  Note I use a modified version of wireless.sh that does not fiddle with
>  pciehp at all (because this seems to be of no good).
>
>  So, when I resume and wireless seems to be not working (no results from
>  scanning for example), I try off->on. If this does not work, I remove
>  the pciehp module and then re-load it. After this, the wireless script
>  seems to work ok.

I'll have to play around with reloading the pciehp module.  I seem to
remember having some issues with suspend and wireless on another
laptop about a year ago,
and in the ACPI settings all I had to do to fix it was have some
modules unload before suspend and reload after resuming.  This could
be more elegant in terms of the NetworkManager issue I am having,
where I have been restarting the entirety of dbus, which adds a few
extra seconds to the whole process.

>  My very untested impression is that if the period when the Eee was
>  suspended is "sufficiently long", it all works. The problems come when
>  it is short (< 10 min for example).
>
>  I don't know how much sense this all make, I'd just share it in case it
>  rings some bells.
>
>  As for the wireless.sh/pciehp, I guess I can come with some patch that
>  would handle pciehp re-loading in case if otherwise ifconfig up fails.
>
>  On a general note, I think that shutting down wireless on suspend and
>  bringing it up on resume looks like a sane idea.
>

I appreciate your feedback.  I'm still looking for a way to do this
without restarting dbus, and unloading/reloading modules might just be
the ticket.

Elijah R.