Re: kernel NULL pointer dereference, ucsi bug

From: Vladimir Yerilov <openmindead@gmail.com>
To: Heikki Krogerus <heikki.krogerus@linux.intel.com>,
	Greg KH <gregkh@linuxfoundation.org>
Cc: linux-usb@vger.kernel.org
Subject: Re: kernel NULL pointer dereference, ucsi bug
Date: Wed, 12 Jun 2019 22:23:56 +1000	[thread overview]
Message-ID: <CAB31r6WFwkJwnkqCy3Lw4f2ybZ63EoJFxAcLyDFvm_gHkLRrOQ@mail.gmail.com> (raw)
In-Reply-To: <20190612095532.GB19831@kuha.fi.intel.com>

Yes, it works.
I've built 5.2-rc4 with this patch and it works fine now, the problem is gone.
It is great that I didn't have to downgrade BIOS as a last resort in
my attempts to workaround this issue.

Thank you!

Cc to Greg in order to let him know that it is resolved now.

ср, 12 июн. 2019 г. в 19:55, Heikki Krogerus <heikki.krogerus@linux.intel.com>:
>
> On Tue, Jun 11, 2019 at 10:19:27PM +1000, Vladimir Yerilov wrote:
> > Hi Heikki,
> >
> > > What do you have connected to the USB Type-C connectors on the
> > machine when that happens?
> >
> > Now, on 5.2-rc4, it happens only during boot and when power cord is
> > connected to laptop's type-c charging port. Another port, which does
> > not support charging, does not cause this problem, I mean I can have
> > something connected to it or not, no issue is observed if charger is
> > disconnected. I even tried to connect my hub to the guilty charging
> > port, everything works fine until hub is powered from the charger, in
> > that case the bug happens again. This differs from my previous tests
> > when whatever I had connected to the charging port, caused failures,
> > even type C to A OTG cable. To summarize: now it happens when power
> > cord from the charger is connected during the boot, but if I connect
> > it after boot (e.g. after logging in to my X session), no crash
> > occurs.
> >
> > > Can you use the machine normally when nothing is connected to the USB Type-C connectors
> >
> > Yeah, everything is great when nothing is connected to the faulty
> > charging port during the boot, and charger may be connected later, but
> > it's kinda... wrong.
> >
> > > Is it possible to send complete dmesg output after that happened?
> >
> > You can find everything attached (done with journalctl -k -xb)
>
> Thanks. I don't see anything the driver is doing wrong. I'm attaching
> a patch. Can you test if it fixes the issue?
>
> Br,
>
> --
> heikki

-- 
----
Best regards,
Vladimir Yerilov