Alexey,
Your patch is a little bit overcautious. You could test just in front of the
line
memcpy(out, user, ulen);
In the lines before that ulen gets decremented.
Lars
-----Ursprüngliche Nachricht-----
Von: Alexey Melnikov [mailto:alexey.melnikov@xxxxxxxxx]
Gesendet: Donnerstag, 10. Juni 2010 12:46
An: Howard Chu
Cc: Lars Duesing; cyrus-sasl@xxxxxxxxxxxxxxxxxxxx
Betreff: Re: Bug in ldapdb_plugin - No check if memory is exhausted in
ldapdb_canon_client
Howard Chu wrote:
> Lars Duesing wrote:
>
>> Hi List,
>>
>> I used the ldapdb_plugin as a template for my sql_plugin-enhancements.
>>
>> While reading through the code there is one problem coming to my mind:
>>
>> In ldapdb_canon_client there is NO check whether ulen is greater than
>> out_umax ? maybe it is only a minor issue because the string user is
>> only truncated, but I didn?t have a look if there could be any
>> situation where the size of the string user could be greater than
>> out_umax.
>
> Yeah, didn't seem to be a likely case. Still probably ought to be fixed.
>
>> Patch would be:
>>
>>> if (ulen>out_umax) return SASL_NOMEM;
>>
> Should use SASL_BUFOVER actually.
Agreed.
Committed.
>
>> Just in front of the memcpy.
>>
>> Lars
>
