-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
On 14/12/09 00:05, David Sommerseth wrote:
> On 13/12/09 21:18, Daniel Johnson wrote:
>> David Sommerseth wrote:
>>> - - const char *return_value = NULL;
>>> + aresp[i].resp = NULL;
>
>>> This I didn't think about, but I saw another potential leak.
>>> However, your idea here is clever! But I would probably
>>> suggest:
>
>>> if( aresp[i].resp ) {
>>> free(aresp[i].resp);
>>> aresp[i].resp = NULL;
>>> }
>
>>> But! The potential culprit here is if the aresp[i] struct is not
>>> zeroed before usage. If it is not and this pointer is pointing
>>> at something it do not own, your patch is definitely more the
>>> correct approach.
>
>> It was already nulled at post-patch line 558. I think that the
>> additional NULLing makes the code more readable but strictly
>> speaking it is not necessary.
>
> Yes, you are right. It's calloc'ed earlier in my_conv(). I just double
> checked the whole code now (I just looked at the patch itself earlier
> today).
>
> I'd say this patch is ready for some real testing now, possibly worth
> considering for inclusion to the next development tree. I find it
> valuable to include it at least.
Hi,
Just wanted to inform that this patch has been included into the
bugfix2.1 branch in the openvpn-testing.git tree. It will be merged
into the 'allmerged' branch soon when more patches are ready.
commit 0439701630e257d7abf0e58c204210a6459307da
<http://openvpn.git.sourceforge.net/git/gitweb.cgi?p=openvpn/openvpn-testing.git;a=commitdiff;h=0439701630e257d7abf0e58c204210a6459307da>
Thank you very much, and I'm sorry about the delay getting it in.
kind regards,
David Sommerseth
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.10 (GNU/Linux)
Comment: Using GnuPG with Fedora - http://enigmail.mozdev.org/
iEYEARECAAYFAkuyA70ACgkQDC186MBRfrqhYQCfbh8RjV+tFgrw0JnvYadMOqkE
uE4AnjZyD1w2lmBf0UKc3e5e523A51lv
=uGsN
-----END PGP SIGNATURE-----
