Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Date:   Tue, 10 May 2022 15:11:24 +0100
From:   Will Deacon <will@kernel.org>
To:     Arnd Bergmann <arnd@arndb.de>
Cc:     "Verdun, Jean-Marie" <verdun@hpe.com>,
        "Hawkins, Nick" <nick.hawkins@hpe.com>,
        Russell King - ARM Linux <linux@armlinux.org.uk>,
        Mark Rutland <mark.rutland@arm.com>,
        Linux ARM <linux-arm-kernel@lists.infradead.org>,
        Linux Kernel Mailing List <linux-kernel@vger.kernel.org>,
        Joel Stanley <joel@jms.id.au>
Subject: Re: [PATCH v1] ARM: A9: Add ARM ERRATA 764319 workaround
Message-ID: <20220510141124.GB28104@willie-the-truck>
References: <20220506192957.24889-1-nick.hawkins@hpe.com>
 <CAK8P3a045Di_zRomezeah0ZoSGPw0Z6YoYkZtoxx1qOXAtKbbw@mail.gmail.com>
 <7C103AEB-3111-4AE6-9645-CF590388A879@hpe.com>
 <CAK8P3a0OS+4XTG9VmfPwbuQoT+_G5-fSatbJ0g8Y7Y+O6-3YLQ@mail.gmail.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <CAK8P3a0OS+4XTG9VmfPwbuQoT+_G5-fSatbJ0g8Y7Y+O6-3YLQ@mail.gmail.com>
User-Agent: Mutt/1.10.1 (2018-07-13)
Precedence: bulk

On Tue, May 10, 2022 at 03:01:26PM +0200, Arnd Bergmann wrote:
> On Tue, May 10, 2022 at 1:53 PM Verdun, Jean-Marie <verdun@hpe.com> wrote:
> >
> > Hi Arnd,
> >
> > >    Hi Nick,
> >
> > >    This seems a bit more complex than necessary. Can't you just use a custom
> > >    inline asm with an ex_table entry to catch the fault? Have a look at
> > >    __get_user_asm() for an example.
> > >
> > >           Arnd
> >
> > We got inspired from debug_reg_hook within the same source file
> >( ./arch/arm/kernel/hw_breakpoint.c ). We chose that path to keep coherency
> > within the source code. We can implement the same fix by using an ex_table
> > entry, but this will create two different ways at catching unknown instruction
> > within the same source file. Will that be ok ?
> 
> I got a little lost trying to find where the breakpoint instruction comes
> from that gets trapped here, but I would guess that they had to do this
> using an undef_hook because the ex_table approach does not work
> there for some reason.
> 
> I would still pick the ex_table method here if that works.

IIRC, the ex_table handlers are called only for data aborts and are intended
to be used to handle cases where we take a fault on a memory access (e.g.
translation fault). In this case, we're taking an undefined instruction
exception on a cp14 access and so the undef_hook is the right thing to use.

Will