Re: kernel NULL pointer dereference, ucsi bug

[Heikki Krogerus <heikki.krogerus@linux.intel.com>]

On Wed, Jun 12, 2019 at 10:57:35PM +1000, Vladimir Yerilov wrote:
> Oh, you are right. I meant... You know what I meant :)
> I hope the fix will get there eventually in one way or another. Should
> you need any further tests from my side, just ask and I will make my
> faulty machine work on it.
> 
> Thank you guys again for your kind assistance.

The patch I gave you does not fix the root cause, but instead just
works around it. The change it brings is however useful, so I'm going
to send the patch to the linux-usb ml as a fix for this issue.

I think the root problem in this case is that the firmware is
reporting the alternate modes in a customized way. The problem is
actually not with the firmware itself. The problem is with UCSI
specification. The UCSI spec does not explain how exactly the
alternate modes should be handled, so every platform does it a bit
differently (a major operating system does not seem to care about the
alternate modes).

The ucsi driver we have in kernel already considers a number of
different ways the alternate modes could be handled, but clearly not
the way Kaby Lakes handle them. I'm going to continue debugging this
with a Kaby Lake board, and once/if I figure out how to manage the
alternate modes on it, I'll send the final fix. But for now, let's
workaround the problem.


thanks,

-- 
heikki