Re: [PATCH] b43legacy: Fix machine check errors for PPC architecture with BCM4306/2

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

Michael Buesch wrote:
> With your patch you remove the "gmode" semantics, so you could
> as well remove the whole variable and always assume it's set.
> But I don't recommend to go that way. I'd suggest you find the
> place in the code where phy->gmode is not set but it should be.
> _that_ is the place where the real bug is.
> 
> Just want to make sure you understand the consequences of your patch.
> You remove a feature with that.
> If that feature is really needed in legacy; that is another question.
> It's needed if you want to drive APHY devices.

All of the devices that b43legacy drives have either a BPHY or a GPHY. The current logic always sets
gmode and that feature is meaningless. Anyone overriding the rules that select this driver in order
to operate with an APHY device will hit a BUG_ON in wireless_core_attach.

I have generated and tested a patch that changes back to the old 'connected' variable. That will
avoid the semantics argument and match the V3 specs.

Larry