Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
Date:   Fri, 11 Sep 2020 12:09:11 -0500
From:   Josh Poimboeuf <jpoimboe@redhat.com>
To:     Logan Gunthorpe <logang@deltatee.com>
Cc:     LKML <linux-kernel@vger.kernel.org>, x86@kernel.org,
        Thomas Gleixner <tglx@linutronix.de>,
        Ingo Molnar <mingo@kernel.org>,
        Andy Lutomirski <luto@kernel.org>,
        Naresh Kamboju <naresh.kamboju@linaro.org>,
        Borislav Petkov <bp@alien8.de>
Subject: Re: WARNING: Kernel stack regs has bad 'bp' value
Message-ID: <20200911170911.mrkorgsplrrwp4rv@treble>
References: <c0ba7077-7977-0155-d7be-d4133ebaee5c@deltatee.com>
 <20200911160053.w66xit3imcqsn33g@treble>
 <5a66f512-960b-7f24-068a-01c043ca9877@deltatee.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Disposition: inline
In-Reply-To: <5a66f512-960b-7f24-068a-01c043ca9877@deltatee.com>
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

On Fri, Sep 11, 2020 at 10:14:41AM -0600, Logan Gunthorpe wrote:
> 
> 
> On 2020-09-11 10:00 a.m., Josh Poimboeuf wrote:
> > Hi Logan,
> > 
> > Thanks for the bug report.  (Sorry I missed the first one, Naresh.)
> > 
> > The problem is that ret_from_fork() is no longer in .entry.text, so the
> > following check in the FP unwinder doesn't work when ret_from_fork()
> > gets interrupted.
> > 
> > 	/*
> > 	 * Don't warn if the unwinder got lost due to an interrupt in entry
> > 	 * code or in the C handler before the first frame pointer got set up:
> > 	 */
> > 	if (state->got_irq && in_entry_code(state->ip))
> > 		goto the_end;
> > 
> > If you have the ability to recreate, can you try the following patch?
> 
> Sorry, but I can not reliably hit this bug. I hit it randomly twice last
> week however, despite doing a bunch more runs this week, I haven't hit
> it again.
> 
> I can add the patch to my testing next week but I ca not give you a
> definitive answer on whether this fixes the bug.
> 
> Perhaps Naresh's test is a better reproducer.

That's ok.  I forced a reproducer by putting an infinite loop in
ret_from_fork to force an APIC interrupt with lockdep unwind.

The patch fixes it, I'll post an official version after some more
testing.

-- 
Josh