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: <20210111125702.360745-1-geert+renesas@glider.be> <X/0nmQ/XBpj6PJAh@pendragon.ideasonboard.com>
In-Reply-To: <X/0nmQ/XBpj6PJAh@pendragon.ideasonboard.com>
From:   Geert Uytterhoeven <geert@linux-m68k.org>
Date:   Tue, 12 Jan 2021 09:46:56 +0100
Message-ID: <CAMuHMdVaJgHv0kRSLaFfPzD5hiJKQCg30D=7SAv0cPwR9j5DZw@mail.gmail.com>
Subject: Re: [PATCH] drm/bridge: nwl-dsi: Avoid potential multiplication
 overflow on 32-bit
To:     Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Cc:     Andrzej Hajda <a.hajda@samsung.com>,
        Neil Armstrong <narmstrong@baylibre.com>,
        Jonas Karlman <jonas@kwiboo.se>,
        Jernej Skrabec <jernej.skrabec@siol.net>,
        David Airlie <airlied@linux.ie>,
        Daniel Vetter <daniel@ffwll.ch>,
        DRI Development <dri-devel@lists.freedesktop.org>,
        Linux Kernel Mailing List <linux-kernel@vger.kernel.org>
Content-Type: text/plain; charset="UTF-8"
Precedence: bulk

Hi Laurent,

On Tue, Jan 12, 2021 at 5:38 AM Laurent Pinchart
<laurent.pinchart@ideasonboard.com> wrote:
> On Mon, Jan 11, 2021 at 01:57:02PM +0100, Geert Uytterhoeven wrote:
> > As nwl_dsi.lanes is u32, and NSEC_PER_SEC is 1000000000L, the second
> > multiplication in
> >
> >     dsi->lanes * 8 * NSEC_PER_SEC
> >
> > will overflow on a 32-bit platform.  Fix this by making the constant
> > unsigned long long, forcing 64-bit arithmetic.
> >
> > While iMX8 is arm64, this driver is currently used on 64-bit platforms
> > only, where long is 64-bit, so this cannot happen.  But the issue may
> > start to happen when the driver is reused for a 32-bit SoC, or when code
> > is copied for a new driver.
> >
> > Signed-off-by: Geert Uytterhoeven <geert+renesas@glider.be>

> > --- a/drivers/gpu/drm/bridge/nwl-dsi.c
> > +++ b/drivers/gpu/drm/bridge/nwl-dsi.c
> > @@ -195,7 +195,7 @@ static u32 ps2bc(struct nwl_dsi *dsi, unsigned long long ps)
> >       u32 bpp = mipi_dsi_pixel_format_to_bpp(dsi->format);
> >
> >       return DIV64_U64_ROUND_UP(ps * dsi->mode.clock * bpp,
> > -                               dsi->lanes * 8 * NSEC_PER_SEC);
> > +                               dsi->lanes * 8ULL * NSEC_PER_SEC);
>
> I wonder if we could get rid of a whole class of bugs by turning
> NSEC_PER_SEC into a ULL, but I suppose there are valid cases where a
> 32-bit integer is enough.

Indeed, and 64-bit arithmetic is more expensive on 32-bit platforms.
I considered that change, but doing so would require updates all over
the place (e.g. printing a value derived from NSEC_PER_SEC, divisions
 that need to be changed to do_div or div_u64(), ...)

Note that the selftests already use such a definition.

> Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>

Thanks!

> How did you come across this by the way ?

https://lore.kernel.org/linux-renesas-soc/CAMuHMdXQvPY_mYicjPKjDSCwdO_rP-9PJOvqD0J6=S3Opr1ycg@mail.gmail.com/
and of course I grepped for similar use patterns...

Gr{oetje,eeting}s,

                        Geert

-- 
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org

In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
                                -- Linus Torvalds