Re: BUG in 2.6.22-rc2-mm1: NIC module b44.c broken (Broadcom 4400)

[Andrew Morton <akpm@linux-foundation.org>]

On Sat, 26 May 2007 17:50:48 +0200 Michael Buesch <mb@bu3sch.de> wrote:

> Andrew,
> 
> I am going to ignore Uwe from now on. It's simply impossible
> to debug the problem the way he is responding.
> Well, I'm not the first person in the Linux community adding
> him to the killfile, ... .

Well yes, there are some personality issues here ;) But the main thing is
to struggle on and fix this bug, wherever it lies.

> I ask to try wireless-dev, as the driver works perfectly fine
> for me there, but he refuses to try it, too. So I'm stuck.

I don't think he knows how to obtain it.

Uwe, http://userweb.kernel.org/~akpm/git-wireless.patch.gz is the current
wireless tree.  That's a patch against 2.6.22-rc3.  Could you please test
that?  If that works then we know that the bug probably lies outside the
b44 driver (or it was subsequently fixed).

> I already tried your -mm kernel, but it crashes on my machine
> for other reasons. (Yeah, I should look into them, too :P )

err, please do.  Just the oops trace would be a start.

> Andrew, is it possible that the breakage was introduced in the
> merge process somehow? Didn't the patch apply cleanly? Are there
> other changes to b44 I should know about in your tree?

Only git-wireless.net modifies b44.c but if we're having IRQ assignment
problems then we'd need to look elsewhere.  I guess you could diff
rc2-mm1's b44.c against the expected version.


> If it is really impossible to debug this problem, I'd like to
> suggest you to drop the b44-ssb port completely.

Well we don't know if that'll fix it.

I believe that Uwe said that reverting the b44.c changes from rc2-mm1 fixes
things for him?  Odd, but it still doesn't rule out acip/pci/platform
changes as being the cause.