Re: Regression in 2.6.27-rcX caused by commit bc19d6e0b74ef03a3baf035412c95192b54dfc6f

[Larry Finger <Larry.Finger@lwfinger.net>]

Henrique de Moraes Holschuh wrote:
> 
> However, rfkill_force_state() is NOT updating LED states (I just checked).
> I will sleep on the issue, and send in a patch tomorrow.
> 
> This probably means a small patch to rfkill + Matthew's fixed patch to use
> rfkill_force_state() in b43 will fix the regression in the right way.
> 
> I don't care either way which kind of fix goes to 2.6.27, though.
> 
> The proper fix for rfkill will be in two stages. A small fix now, and a
> complete change on the LED handling to use the blocking notifier chain
> instead later on (which will clean up rfkill code somewhat).
> 

I do not dispute that rfkill-handling in b43 is broken; however, prior
to the commit in question, it worked.

I also think we can agree that we need to get it working before 2.6.27
is released. If the small fix now is the reversion of bc19d6e, then I
think this is the correct path. We will then have a couple of weeks to
get the code working correctly before the 2.6.28 merge starts.

I admit that I never tested any of the RFKILL patches as they went in.
One of the reasons is that the development process seemed rather
untidy to an outsider, and I wasn't sure that any of the code would
ever be in the kernel. As such, it snuck up on me. I'll not let that
happen again. After the reversion, I will again test any suggested
code changes, but do not expect me to work out any of the changes. I
have enough to do.

Larry