Message-ID: <CAF6AEGv_ekqmfFmLzYRn29eva6s2ZZQRW8xA-7+g5xVmwR2LZA@mail.gmail.com>
Date:   Tue, 22 Nov 2016 16:46:49 -0500
From:   Rob Clark <robdclark@...il.com>
To:     Liviu Dudau <Liviu.Dudau@....com>
Cc:     Sean Paul <seanpaul@...omium.org>,
        Ville Syrjälä <ville.syrjala@...ux.intel.com>,
        Jani Nikula <jani.nikula@...el.com>,
        Daniel Vetter <daniel.vetter@...ll.ch>,
        Eric Engestrom <eric@...estrom.ch>,
        LKML <linux-kernel@...r.kernel.org>,
        DRI devel <dri-devel@...ts.freedesktop.org>
Subject: Re: [PATCH] drm: check for NULL parameter in exported
 drm_get_format_name() function.

On Tue, Nov 22, 2016 at 1:47 PM, Liviu Dudau <Liviu.Dudau@....com> wrote:
> On Tue, Nov 22, 2016 at 01:15:08PM -0500, Sean Paul wrote:
>> On Tue, Nov 22, 2016 at 1:06 PM, Ville Syrjälä
>> <ville.syrjala@...ux.intel.com> wrote:
>> > On Tue, Nov 22, 2016 at 12:35:53PM -0500, Rob Clark wrote:
>> >> On Tue, Nov 22, 2016 at 12:31 PM, Ville Syrjälä
>> >> <ville.syrjala@...ux.intel.com> wrote:
>> >> > On Tue, Nov 22, 2016 at 12:23:59PM -0500, Rob Clark wrote:
>> >> >> On Tue, Nov 22, 2016 at 11:50 AM, Ville Syrjälä
>> >> >> <ville.syrjala@...ux.intel.com> wrote:
>> >> >> > On Tue, Nov 22, 2016 at 04:41:06PM +0000, Liviu Dudau wrote:
>> >> >> >> drm_get_format_name() de-references the buf parameter without checking
>> >> >> >> if the pointer was not NULL. Given that the function is EXPORT-ed, lets
>> >> >> >> sanitise the parameters before proceeding.
>> >> >> >>
>> >> >> >> Fixes: b3c11ac267d461d3d5 ("drm: move allocation out of drm_get_format_name())
>> >> >> >> Cc: Eric Engestrom <eric@...estrom.ch>
>> >> >> >> Cc: Rob Clark <robdclark@...il.com>
>> >> >> >> Cc: Jani Nikula <jani.nikula@...el.com>
>> >> >> >> Cc: Daniel Vetter <daniel.vetter@...ll.ch>
>> >> >> >>
>> >> >> >> Signed-off-by: Liviu Dudau <Liviu.Dudau@....com>
>> >> >> >> ---
>> >> >> >>  drivers/gpu/drm/drm_fourcc.c | 3 +++
>> >> >> >>  1 file changed, 3 insertions(+)
>> >> >> >>
>> >> >> >> diff --git a/drivers/gpu/drm/drm_fourcc.c b/drivers/gpu/drm/drm_fourcc.c
>> >> >> >> index 90d2cc8..0a3ff0b 100644
>> >> >> >> --- a/drivers/gpu/drm/drm_fourcc.c
>> >> >> >> +++ b/drivers/gpu/drm/drm_fourcc.c
>> >> >> >> @@ -85,6 +85,9 @@ EXPORT_SYMBOL(drm_mode_legacy_fb_format);
>> >> >> >>   */
>> >> >> >>  const char *drm_get_format_name(uint32_t format, struct drm_format_name_buf *buf)
>> >> >> >>  {
>> >> >> >> +     if (!buf)
>> >> >> >> +             return NULL;
>> >> >> >> +
>> >> >> >
>> >> >> > Seems rather pointless to me. Why would you ever pass NULL to this guy?
>
> I don't know, maybe because I'm a malicious driver writer that wants to crash the kernel?
> The function is exported, so it can be called by any module, even out-of-tree. Looks
> like an easy DoS to me.

if you are a kernel module, in tree or not, all bets are off..  an out
of tree module could just as easily call sprintf(NULL, "foo")
directly..

BR,
-R

>> >> >>
>> >> >> perhaps BUG_ON(!buf)...
>> >> >
>> >> > And how does that differ from just buf->foo?
>> >>
>> >> it gets you a file and line # in the error splat.. not that
>> >> drm_get_format_name() is such a big function that it would be
>> >> difficult to decipher the null deref crash, but if we added anything
>> >> it should be BUG_ON() to make it clear that passing null isn't a
>> >> caller error.
>> >
>> > Yeah, BUG_ON() at least documents the intent, so it's better than
>> > the null check. But for something like this even BUG_ON() is
>> > just wasted bytes IMO.
>>
>> +1
>>
>> The patch has a Fixes line, but AFAICT, the referenced patch didn't
>> introduce any NULL opportunities that weren't previously there.
>
> Really? Before the referenced patch the function only had one parameter,
> an uint32_t. True that kmalloc could return a NULL pointer and we would've
> crashed there as well, but at least we were running out of memory. Now
> I can crash the kernel any time I want with a badly written driver/module.
>
> Best regards,
> Liviu
>
>>
>> Sean
>>
>>
>> >
>> > BUG_ON() can be useful for those weird bugs where somewhere deep
>> > down you hit a null pointer and you can't figure out where the
>> > bad pointer came from. So you might sprinkle a few BUG_ONs()
>> > further up to catch it sooner. Esp. if you can't reproduce the
>> > bug yourself and have to rely on user(s) to find it for you.
>> >
>> > Even WARN_ON() w/ or w/o an early bailout might be a decent idea
>> > sometimes since it might have a slightly higher chance of keeping
>> > the kernel in working condition, but IMO just blindly throwing
>> > it around everywhere is not a good approach.
>> >
>> > --
>> > Ville Syrjälä
>> > Intel OTC
>> > _______________________________________________
>> > dri-devel mailing list
>> > dri-devel@...ts.freedesktop.org
>> > https://lists.freedesktop.org/mailman/listinfo/dri-devel
>
> --
> ====================
> | I would like to |
> | fix the world,  |
> | but they're not |
> | giving me the   |
>  \ source code!  /
>   ---------------
>     ¯\_(ツ)_/¯

Powered by blists - more mailing lists