linux-kernel - Re: [PATCH] drm/nouveau: usif_ioctl: ensure returns are initialized

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:	Wed, 1 Jul 2015 18:59:57 +0100
From:	Emil Velikov <emil.l.velikov@...il.com>
To:	Ilia Mirkin <imirkin@...m.mit.edu>
Cc:	Colin Ian King <colin.king@...onical.com>,
	Ben Skeggs <bskeggs@...hat.com>,
	"dri-devel@...ts.freedesktop.org" <dri-devel@...ts.freedesktop.org>,
	"linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH] drm/nouveau: usif_ioctl: ensure returns are initialized

On 1 July 2015 at 18:37, Ilia Mirkin <imirkin@...m.mit.edu> wrote:
> On Wed, Jul 1, 2015 at 1:18 PM, Colin Ian King <colin.king@...onical.com> wrote:
>> On 01/07/15 18:12, Emil Velikov wrote:
>>> On 1 July 2015 at 17:56, Ilia Mirkin <imirkin@...m.mit.edu> wrote:
>>>> On Wed, Jul 1, 2015 at 12:51 PM, Colin King <colin.king@...onical.com> wrote:
>>>>> From: Colin Ian King <colin.king@...onical.com>
>>>>>
>>>>> Various usif_ioctl helper functions do not initialize the
>>>>> return variable ret and some of the error handling return
>>>>> paths just return garbage values that were on the stack (or
>>>>> in a register).  I believe that in all the cases, the
>>>>> initial ret variable should be set to -EINVAL and subsequent
>>>>> paths through these helper functions set it appropriately
>>>>> otherwise.
>>>>>
>>>>> Found via static analysis using cppcheck:
>>>>>
>>>>> [drivers/gpu/drm/nouveau/nouveau_usif.c:138]:
>>>>>     (error) Uninitialized variable: ret
>>>>
>>>> It sure would seem that way, wouldn't it?
>>>>
>>>> #define nvif_unpack(d,vl,vh,m) ({                                              \
>>>>         if ((vl) == 0 || ret == -ENOSYS) {                                     \
>>>>                 int _size = sizeof(d);                                         \
>>>>                 if (_size <= size && (d).version >= (vl) &&                    \
>>>>                                      (d).version <= (vh)) {                    \
>>>>                         data = (u8 *)data + _size;                             \
>>>>                         size = size - _size;                                   \
>>>>                         ret = ((m) || !size) ? 0 : -E2BIG;                     \
>>>>                 } else {                                                       \
>>>>                         ret = -ENOSYS;                                         \
>>>>                 }                                                              \
>>>>         }                                                                      \
>>>>         (ret == 0);                                                            \
>>>> })
>>>>
>>>> So actually it does get initialized, and I guess cppcheck doesn't know
>>>> about macros?
>>
>> Hrm, what about the case when ((vl) == 0 || ret == -ENOSYS) is false,
>> where is ret being set in that case?
>
> Is that actually the case for any of the callsites? gcc would complain
> about that...
There is one:

drm/nouveau/nvkm/engine/disp/nv50.c: if (nvif_unpack(args->v1, 1, 1, true))

Seems like a recent addition though,  I don't recall it with back when
was introduced.

-Emil
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/