Received-SPF: pass (google.com: best guess record for domain of linux-kernel-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67;
Date:   Thu, 30 Aug 2018 12:24:24 -0700
From:   Guenter Roeck <linux@roeck-us.net>
To:     Joerg Roedel <jroedel@suse.de>
Cc:     Linus Torvalds <torvalds@linux-foundation.org>,
        Linux Kernel Mailing List <linux-kernel@vger.kernel.org>,
        Thomas Gleixner <tglx@linutronix.de>,
        Michal Hocko <mhocko@suse.com>,
        Andi Kleen <ak@linux.intel.com>,
        the arch/x86 maintainers <x86@kernel.org>,
        Dave Hansen <dave.hansen@intel.com>,
        Pavel Machek <pavel@ucw.cz>
Subject: Re: efi boot failures due to PTI with 32 bit builds and Intel CPUs
Message-ID: <20180830192424.GC6752@roeck-us.net>
References: <20180829201659.GA29801@roeck-us.net>
 <20180830071951.dv6wgnn6ikfhh7yt@suse.de>
 <f6d9bdcd-98a6-fcf2-5308-936ffd714ea7@roeck-us.net>
 <20180830113059.qj6yvw4ggktc7yex@suse.de>
 <915dca84-c98a-fc63-3210-b21dd926b3d4@roeck-us.net>
 <20180830180849.nmh646ehnolphq4v@suse.de>
 <CA+55aFzQwC59pkeA+T3E0G4SZe4dQsYcmnf1eKGYnYpcPTKhSQ@mail.gmail.com>
 <20180830184639.6iv5itjex6y45w3j@suse.de>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20180830184639.6iv5itjex6y45w3j@suse.de>
User-Agent: Mutt/1.5.24 (2015-08-30)
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

On Thu, Aug 30, 2018 at 08:46:39PM +0200, Joerg Roedel wrote:
> On Thu, Aug 30, 2018 at 11:21:49AM -0700, Linus Torvalds wrote:
> > On Thu, Aug 30, 2018 at 11:08 AM Joerg Roedel <jroedel@suse.de> wrote:
> > >
> > > Without a mapped GDT the #PF and #DF handlers also can't be started, so
> > > the machine triple-faults. Below diff fixes it for me, I'll send a
> > > proper patch tomorrow.
> > 
> > Hmm. Is there any reason why this code doesn't just use
> > 
> >         load_fixmap_gdt(0);
> 
> No idea, probably the function didn't exist when the code was written?
> I can change that when writing the patch.
> 
> > and shouldn't it do it after loading the new %cr3?
> 
> That seems more robust, yes. No sure if the old %cr3
> (initial_page_table) has the fixmap gdt mapped at all.

All three variants (hardcoded, call load_fixmap_gdt(0) first, call
load_fixmap_gdt(0) after load_cr3()) work for me. Feel free to add

Tested-by: Guenter Roeck <linux@roeck-us.net>

when you submit the patch.

Thanks a lot for tracking this down!

Guenter