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;
MIME-Version: 1.0
References: <20200521100952.GA5360@willie-the-truck> <20200521173738.GA29590@e121166-lin.cambridge.arm.com>
 <20200526202157.GE2206@willie-the-truck> <20200527134104.GA16115@e121166-lin.cambridge.arm.com>
 <20200601070459.GB8601@willie-the-truck> <CAKwvOdmXmxOdW_TJQmYBYDY8gDOacjDTcpSWQGATb2p_85tFAQ@mail.gmail.com>
 <CAMj1kXFQzBaZO+RGKs2iJOzW6rdEiAjdVc8PJ4U+KMWgCD9a6w@mail.gmail.com>
In-Reply-To: <CAMj1kXFQzBaZO+RGKs2iJOzW6rdEiAjdVc8PJ4U+KMWgCD9a6w@mail.gmail.com>
From:   Nick Desaulniers <ndesaulniers@google.com>
Date:   Mon, 1 Jun 2020 15:19:23 -0700
Message-ID: <CAKwvOdnz-=GD9-taLQt6LDhs2Q-xgWmywCUA6skB0NJhubB+nw@mail.gmail.com>
Subject: Re: arm64/acpi: NULL dereference reports from UBSAN at boot
To:     Ard Biesheuvel <ardb@kernel.org>
Cc:     Will Deacon <will@kernel.org>,
        Lorenzo Pieralisi <lorenzo.pieralisi@arm.com>,
        Hanjun Guo <guohanjun@huawei.com>,
        "Rafael J. Wysocki" <rjw@rjwysocki.net>,
        Linux ARM <linux-arm-kernel@lists.infradead.org>,
        LKML <linux-kernel@vger.kernel.org>,
        Mark Rutland <mark.rutland@arm.com>,
        Dmitry Vyukov <dvyukov@google.com>,
        Alexander Potapenko <glider@google.com>,
        Peter Collingbourne <pcc@google.com>
Content-Type: text/plain; charset="UTF-8"
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

On Mon, Jun 1, 2020 at 2:57 PM Ard Biesheuvel <ardb@kernel.org> wrote:
>
> On Mon, 1 Jun 2020 at 23:52, Nick Desaulniers <ndesaulniers@google.com> wrote:
> >
> > Anyways, it looks like the address of member from NULL subexpression
> > looks problematic.  I wonder if offsetof can be used here?
> >
> > #define ACPI_OFFSET(d, f) ACPI_PTR_DIFF (offsetof(d, f), (void *) 0)
> >
> > Seems to work in my basic test case.  Untested in the kernel.
> >
> > IIUC, ACPI_OFFSET is trying to calculate the difference between the
> > offset of a member of a struct and 0?  Isn't that the tautology `x - 0
> > == x`?
>
> No. ACPI_OFFSET() is just a poor person's version of offsetof().
>
> (Note that it calculates the difference between &(((d *) 0)->f) and
> (void *)0x0, so the 0x0 term is there on both sides)

Got it. So we're trying to avoid including stddef.h?  Can
__builtin_offsetof be used here?
#define ACPI_OFFSET(d, f) ACPI_PTR_DIFF (__builtin_offsetof(d, f), (void *) 0)
The oldest version of GCC in godbolt.org (4.1) supports this builtin.
-- 
Thanks,
~Nick Desaulniers