rfkill vs. interface up

rfkill vs. interface up

[Johannes Berg]

[-- Attachment #1: Type: text/plain, Size: 1237 bytes --]

Hi,

More thoughts on rfkill ...

So we were thinking it would be sensible to just force interfaces down
on rfkill, which is of course possible, and then reject attempts to set
the interface UP while killed.

There's just one problem with that -- when you un-rfkill, does the
kernel set interfaces UP again?

If not -- how would we possibly handle 'iwconfig .. txpower off'? We're
mostly handling it like rfkill now afaict, but 'iwconfig .. txpower on'
wouldn't be able to do anything properly.

If yes -- how do we know which interfaces to set UP? However we turn it,
userspace can then not disable an interface regardless of rfkill state,
which gets really confusing.

Therefore, I don't think we can simply set interfaces down on rfkill
with the current scheme.

On the other hand, the interfaces really are dysfunctional in rfkill and
we really need more integration.


I'm happy to give up on 'iwconfig wlan0 txpower on/off' entirely, simply
refuse supporting it (return -EINVAL if txpower.disabled) and then use
the first scheme where rfkill forces the interface down but userspace is
responsible for enabling it again. This is asymmetric, but I don't see
what else to do.

Thoughts?

johannes


[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 801 bytes --]

Re: rfkill vs. interface up

[Dan Williams]

On Tue, 2009-05-19 at 14:32 +0200, Johannes Berg wrote:
> Hi,
> 
> More thoughts on rfkill ...
> 
> So we were thinking it would be sensible to just force interfaces down
> on rfkill, which is of course possible, and then reject attempts to set
> the interface UP while killed.
> 
> There's just one problem with that -- when you un-rfkill, does the
> kernel set interfaces UP again?
> 
> If not -- how would we possibly handle 'iwconfig .. txpower off'? We're
> mostly handling it like rfkill now afaict, but 'iwconfig .. txpower on'
> wouldn't be able to do anything properly.
> 
> If yes -- how do we know which interfaces to set UP? However we turn it,
> userspace can then not disable an interface regardless of rfkill state,
> which gets really confusing.
> 
> Therefore, I don't think we can simply set interfaces down on rfkill
> with the current scheme.
> 
> On the other hand, the interfaces really are dysfunctional in rfkill and
> we really need more integration.
> 
> 
> I'm happy to give up on 'iwconfig wlan0 txpower on/off' entirely, simply
> refuse supporting it (return -EINVAL if txpower.disabled) and then use
> the first scheme where rfkill forces the interface down but userspace is
> responsible for enabling it again. This is asymmetric, but I don't see
> what else to do.

That proposal sounds fine to me from a userspace perspective.  Sane
implementations don't assume the interface is IFF_UP when they configure
the device anyway (since it's not necessarily up at boot time or after
hibernate for example), and since you need a reconfigure after rfkill,
this seems reasonable.

You wouldn't have to give up on txpower either, you could simply map
'txpower off' to SW-rfkill, and 'txpower on' to un-SW-rfkill, where of
course the interface would be !IFF_UP after 'txpower on' just like
flipping the killswitch would.

Dan

Re: rfkill vs. interface up

[Johannes Berg]

[-- Attachment #1: Type: text/plain, Size: 834 bytes --]

On Thu, 2009-05-21 at 11:39 -0400, Dan Williams wrote:

> That proposal sounds fine to me from a userspace perspective.  Sane
> implementations don't assume the interface is IFF_UP when they configure
> the device anyway (since it's not necessarily up at boot time or after
> hibernate for example), and since you need a reconfigure after rfkill,
> this seems reasonable.
> 
> You wouldn't have to give up on txpower either, you could simply map
> 'txpower off' to SW-rfkill, and 'txpower on' to un-SW-rfkill, where of
> course the interface would be !IFF_UP after 'txpower on' just like
> flipping the killswitch would.

Yeah, good point, that sounds better than having to give up on it -- the
key though is that we can't recover IFF_UP state when that happens; if
that's ok this should be fairly simple.

johannes

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 801 bytes --]

Re: rfkill vs. interface up

[Dan Williams]

On Thu, 2009-05-21 at 18:05 +0200, Johannes Berg wrote:
> On Thu, 2009-05-21 at 11:39 -0400, Dan Williams wrote:
> 
> > That proposal sounds fine to me from a userspace perspective.  Sane
> > implementations don't assume the interface is IFF_UP when they configure
> > the device anyway (since it's not necessarily up at boot time or after
> > hibernate for example), and since you need a reconfigure after rfkill,
> > this seems reasonable.
> > 
> > You wouldn't have to give up on txpower either, you could simply map
> > 'txpower off' to SW-rfkill, and 'txpower on' to un-SW-rfkill, where of
> > course the interface would be !IFF_UP after 'txpower on' just like
> > flipping the killswitch would.
> 
> Yeah, good point, that sounds better than having to give up on it -- the
> key though is that we can't recover IFF_UP state when that happens; if
> that's ok this should be fairly simple.

I don't think recovering IFF_UP needs to be done by the kernel.  rfkill
already hoses the card state anyway, requiring a full setup by
userspace, which usually includes ensuring the device is up (which, for
example, both NM and wpa_supplicant have done since the beginning of
time).

Dan