From mboxrd@z Thu Jan  1 00:00:00 1970
Received: from eggs.gnu.org ([2001:4830:134:3::10]:44297)
	by lists.gnu.org with esmtp (Exim 4.71)
	(envelope-from <mjt@tls.msk.ru>) id 1Wnu3V-0000EI-DE
	for qemu-devel@nongnu.org; Fri, 23 May 2014 14:19:03 -0400
Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71)
	(envelope-from <mjt@tls.msk.ru>) id 1Wnu3P-0003RE-8Q
	for qemu-devel@nongnu.org; Fri, 23 May 2014 14:18:57 -0400
Received: from isrv.corpit.ru ([86.62.121.231]:43546)
	by eggs.gnu.org with esmtp (Exim 4.71)
	(envelope-from <mjt@tls.msk.ru>) id 1Wnu3P-0003R8-19
	for qemu-devel@nongnu.org; Fri, 23 May 2014 14:18:51 -0400
Message-ID: <537F910A.2070902@msgid.tls.msk.ru>
Date: Fri, 23 May 2014 22:18:50 +0400
From: Michael Tokarev <mjt@tls.msk.ru>
MIME-Version: 1.0
References: <1400844279-1685-1-git-send-email-armbru@redhat.com>
	<1400844279-1685-7-git-send-email-armbru@redhat.com>
	<537F8EF4.9000803@tls.msk.ru> <537F906E.4010101@msgid.tls.msk.ru>
In-Reply-To: <537F906E.4010101@msgid.tls.msk.ru>
Content-Type: text/plain; charset=KOI8-R
Content-Transfer-Encoding: quoted-printable
Subject: Re: [Qemu-devel] [PATCH v2 6/7] libcacard/vcard_emul_nss: Assert
 vreaderOpt isn't null
List-Id: <qemu-devel.nongnu.org>
List-Unsubscribe: <https://lists.nongnu.org/mailman/options/qemu-devel>,
	<mailto:qemu-devel-request@nongnu.org?subject=unsubscribe>
List-Archive: <http://lists.nongnu.org/archive/html/qemu-devel>
List-Post: <mailto:qemu-devel@nongnu.org>
List-Help: <mailto:qemu-devel-request@nongnu.org?subject=help>
List-Subscribe: <https://lists.nongnu.org/mailman/listinfo/qemu-devel>,
	<mailto:qemu-devel-request@nongnu.org?subject=subscribe>
To: Markus Armbruster <armbru@redhat.com>, qemu-devel@nongnu.org
Cc: alevy@redhat.com

23.05.2014 22:16, Michael Tokarev =D0=C9=DB=C5=D4:
> 23.05.2014 22:09, Michael Tokarev wrote:
>> 23.05.2014 15:24, Markus Armbruster wrote:
>>> It's not locally obvious, and Coverity can't see it either.
>>>
>>> Signed-off-by: Markus Armbruster <armbru@redhat.com>
>>> Reviewed-by: Alon Levy <alevy@redhat.com>
>>> ---
>>>  libcacard/vcard_emul_nss.c | 1 +
>>>  1 file changed, 1 insertion(+)
>>>
>>> diff --git a/libcacard/vcard_emul_nss.c b/libcacard/vcard_emul_nss.c
>>> index 2048917..4f55e44 100644
>>> --- a/libcacard/vcard_emul_nss.c
>>> +++ b/libcacard/vcard_emul_nss.c
>>> @@ -1181,6 +1181,7 @@ vcard_emul_options(const char *args)
>>>                  vreaderOpt =3D g_renew(VirtualReaderOptions, opts->v=
reader,
>>>                                       reader_count);
>>>              }
>>> +            assert(vreaderOpt);
>>>              opts->vreader =3D vreaderOpt;
>>>              vreaderOpt =3D &vreaderOpt[opts->vreader_count];
>>>              vreaderOpt->name =3D g_strndup(name, name_length);
>>
>> Shouldn't the assignment be moved up one line into the if {}
>> statement instead?
>=20
> Actually it looks like this code is just buggy, it works for just
> one iteration.  Because at this place, vreaderOpts will be non-NULL
> only if we expanded the array.  If we didn't (we do that in
> READER_STEP increments), we'll be assigning NULL to opts->vreader,
> because vreaderOpt will _really_ be NULL here.

So, the real fix is:

1) drop =3D NULL at declaration of vreaderOpt;
2) do not mention vreaderOpt inside the if{} statement at
   all, we don't need indirection there;
3) drop this opts->vreader assignment

and ofcourse do not add the assert as in the patch above ;)

/mjt

> One good example when setting initial values like this (for vreaderOpts=
)
> is a Bad Idea (tm).  If it weren't needlessly NULL initially, compiler
> was able to tell us about using uninitialized variable.
>=20
> Thanks,
>=20
> /mjt
>=20