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:   Wed, 20 Apr 2022 11:11:26 -0700
From:   Kees Cook <keescook@chromium.org>
To:     Jann Horn <jannh@google.com>
Cc:     Johan Hovold <johan@kernel.org>, kernel test robot <lkp@intel.com>,
        Greg Kroah-Hartman <gregkh@linuxfoundation.org>,
        linux-usb@vger.kernel.org, stable@vger.kernel.org,
        linux-kernel@vger.kernel.org, linux-hardening@vger.kernel.org
Subject: Re: [PATCH] USB: serial: Fix heap overflow in WHITEHEAT_GET_DTR_RTS
Message-ID: <202204201056.5A1A6BAE04@keescook>
References: <20220419041742.4117026-1-keescook@chromium.org>
 <Yl+8K++wZUJCthMj@hovoldconsulting.com>
 <CAG48ez2Pikm5g2RfJxae=jz1C7KSCWc99sCa7fXFBKvDOPJubA@mail.gmail.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <CAG48ez2Pikm5g2RfJxae=jz1C7KSCWc99sCa7fXFBKvDOPJubA@mail.gmail.com>
Precedence: bulk

On Wed, Apr 20, 2022 at 02:33:06PM +0200, Jann Horn wrote:
> On Wed, Apr 20, 2022 at 10:14 AM Johan Hovold <johan@kernel.org> wrote:
> > On Mon, Apr 18, 2022 at 09:17:42PM -0700, Kees Cook wrote:
> > > This looks like it's harmless, as both the source and the destinations are
> > > currently the same allocation size (4 bytes) and don't use their padding,
> > > but if anything were to ever be added after the "mcr" member in "struct
> > > whiteheat_private", it would be overwritten. The structs both have a
> > > single u8 "mcr" member, but are 4 bytes in padded size. The memcpy()
> > > destination was explicitly targeting the u8 member (size 1) with the
> > > length of the whole structure (size 4), triggering the memcpy buffer
> > > overflow warning:
> >
> > Ehh... No. The size of a structure with a single u8 is 1, not 4. There's
> > nothing wrong with the current code even if the use of memcpy for this
> > is a bit odd.

I thought that was surprising too, and suspected it was something
specific to the build (as Jann also suggested). I tracked it down[1] to
"-mabi=apcs-gnu", which is from CONFIG_AEABI=n.

whiteheat_private {
        __u8                       mcr;                  /*     0     1 */

        /* size: 4, cachelines: 1, members: 1 */
        /* padding: 3 */
        /* last cacheline: 4 bytes */
};


> >
> > > In file included from include/linux/string.h:253,
> > >                  from include/linux/bitmap.h:11,
> > >                  from include/linux/cpumask.h:12,
> > >                  from include/linux/smp.h:13,
> > >                  from include/linux/lockdep.h:14,
> > >                  from include/linux/spinlock.h:62,
> > >                  from include/linux/mmzone.h:8,
> > >                  from include/linux/gfp.h:6,
> > >                  from include/linux/slab.h:15,
> > >                  from drivers/usb/serial/whiteheat.c:17:
> > > In function 'fortify_memcpy_chk',
> > >     inlined from 'firm_send_command' at drivers/usb/serial/whiteheat.c:587:4:
> > > include/linux/fortify-string.h:328:25: warning: call to '__write_overflow_field' declared with attribute warning: detected write beyond size of field (1st parameter); maybe use struct_group()? [-Wattribute-warning]
> > >   328 |                         __write_overflow_field(p_size_field, size);
> > >       |                         ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
> >
> > So something is confused here.
> 
> So something's going wrong in fortify_memcpy_chk()? It looks like it
> is called with constant "size" equal to 1, and the condition

sizeof(info->mcr) is 1.
sizeof(struct whiteheat_dr_info) (with CONFIG_AEABI=n) is 4.

Given nothing actually uses "struct whiteheat_dr_info", except for the
sizing (har har), I suspect the better solution is just to do:

	info->mcr = command_info->result_buffer[0];

-Kees

[1] https://godbolt.org/z/3YvM1MYWW

-- 
Kees Cook