[Debian-eeepc-devel] rfkill, eeepc-laptop, and ath9k

[Alan Jenkins]

On 12/8/09, Kevin Goodsell <kevin-opensource at omegacrash.net> wrote:
> I know this isn't really an ACPI issue but it must affect the ACPI
> scripts, and I haven't been able to find any useful information anywhere
> else, so I'm wondering what if anything you guys know about it.
>
> The problem as I understand it, is that eeepc-laptop and ath9k both
> provide an rfkill interface for the wifi device. These two interfaces
> are different somehow and don't play nice together. On my 1005HA running
> Squeeze I had to write a new script to handle wifi toggling at least
> somewhat reliably. To turn off wifi, the script unloads ath9k and writes
> 0 to the eeepc-wlan rfkill interface. To turn wifi back on, it writes 1
> to the eeepc-wlan rfkill interface, modprobes ath9k, sleeps for 5
> seconds, then writes 1 to ath9k's rfkill interface.
>
> Does anyone have any idea why we have these two rfkill interfaces, what
> the "right" way to handle this is, and/or whether anyone is planning on
> fixing this? I find surprisingly little about the problem by googling.
>
> Thanks.

(Damyan explained why we get two rfkill devices)

[CC eeepc-laptop module list]

The eeepc-laptop module is known to have problems on the 1005HA, with
rfkill in particular.  There are a couple of ideas to avoid the known
problem where the _wired_ network adaptor disappears due to rfkill.
Corentin (the maintainer) is planning to write a new driver for the
1005HA.  But I don't think we've heard your problem before, so thanks
for the report :).

Unfortunately, even if you disable the acpi script entirely,  I think
the in-kernel rfkill-input will toggle the eeepc-laptop rfkill.  (The
kernel includes a default handler for KEY_WLAN).

I can only counsel patience.  (And leaving the wireless enabled :).
If you really need something working now, a quick hack might be to
write an acpi script to toggle the ath9k rfkill on fn+f6.

Regards
Alan