From mboxrd@z Thu Jan  1 00:00:00 1970
From: Lukas Wunner <lukas@wunner.de>
Subject: Re: [PATCH] Revert "apple-gmux: lock iGP IO to protect from vgaarb
 changes"
Date: Thu, 25 Jan 2018 21:25:30 +0100
Message-ID: <20180125202530.GA7175@wunner.de>
References: <3671c3081a0ea3226dbb4c04e780ad7b6be0f201.1516793407.git.lukas@wunner.de>
 <20180125201048.x3tr3xpovo5tzduz@chaos.in-kiel.de>
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Return-path: <platform-driver-x86-owner@vger.kernel.org>
Received: from bmailout2.hostsharing.net ([83.223.90.240]:33821 "EHLO
        bmailout2.hostsharing.net" rhost-flags-OK-OK-OK-OK) by vger.kernel.org
        with ESMTP id S1751109AbeAYUZd (ORCPT
        <rfc822;platform-driver-x86@vger.kernel.org>);
        Thu, 25 Jan 2018 15:25:33 -0500
Content-Disposition: inline
In-Reply-To: <20180125201048.x3tr3xpovo5tzduz@chaos.in-kiel.de>
Sender: platform-driver-x86-owner@vger.kernel.org
List-ID: <platform-driver-x86.vger.kernel.org>
To: Wilfried Klaebe <linux-kernel@lebenslange-mailadresse.de>
Cc: platform-driver-x86@vger.kernel.org, Darren Hart <dvhart@infradead.org>, Andy Shevchenko <andy@infradead.org>, Petri Hodju <petrihodju@yahoo.com>, Bjorn Helgaas <bhelgaas@google.com>, Bruno Premont <bonbons@linux-vserver.org>, Andy Ritger <aritger@nvidia.com>, Ronald Tschalaer <ronald@innovation.ch>

On Thu, Jan 25, 2018 at 09:10:48PM +0100, Wilfried Klaebe wrote:
> On Wed, Jan 24, 2018 at 07:35:45PM +0100, Lukas Wunner wrote:
> > * Wilfried Klaebe reports a deadlock when launching Xorg because it
> >   opens /dev/vga_arbiter and calls vga_get(), but apple-gmux is holding
> >   a lock on I/O space indefinitely.  It looks like apple-gmux' current
> >   behavior is an abuse of the vgaarb API as locks are not meant to be
> >   held for longer periods:
> >   https://bugzilla.kernel.org/show_bug.cgi?id=88861#c11
> >   https://bugzilla.kernel.org/attachment.cgi?id=217541
[...]
> I see warnings about possible circular locking regarding vgasr_mutex,
> but haven't had a deadlock yet. I'd be happy to provide more information
> as timely as job and family allow.

Could you attach dmesg output with that lockdep splat to the
above-quoted bugzilla entry?

Thanks,

Lukas