Hello,
Unfortunately mails to <bugnumber>-forwarded get archived, but are not
relayed to the list. Therefore I resend this for reference.
Frank
"Mark A. Wicks" <[EMAIL PROTECTED]> schrieb:
> Prabhu has definitely uncovered a bug, but changing the length of the
> string isn't the right fix. That field should never be more than 24
> characters (including the terminating null character).
>
> I'm working on a fix. It looks like there's more than one thing
> broken in asn_date().
>
> Mark
>
> On Sun, 23 Jan 2005, [iso-8859-1] Frank Küster wrote:
>
>> Dear Mark,
>>
>> Prabhu Rmachandran, a user of the Debian teTeX package that contains
>> dvipdfm has reported a segfault in dvipdfm, and also found a fix that
>> corrects the problem.
>>
>> The strange thing is that other people were not able to reproduce the
>> segfault with the same minimal LaTeX or dvi file as Prabhu did. We would
>> be glad if you could look into this.
>>
>> Regards, Frank
>>
>>
>> Prabhu Ramachandran <[EMAIL PROTECTED]> wrote:
>>
>>>
>>> OK, I've found the bug. Basically, the asn_date function in
>>> texk/dvipdfm/pdfdoc.c was writing to a string but did not allocate
>>> enough memory for the string before writing to it. Attached is a
>>> trivial patch for the file texk/dvipdfm/pdfdoc.c.
>>>
>>> With this patch dvipdfm works fine for me and does not segfault
>>> anymore.
>>>
>>> HTH.
>>> prabhu
>>>
>>> --- pdfdoc.c.orig 2005-01-22 16:12:30.000000000 +0530
>>> +++ pdfdoc.c 2005-01-22 16:12:38.000000000 +0530
>>> @@ -239,7 +239,7 @@
>>> #define timezone 0l
>>> #endif /* TM_GM_TOFF */
>>> #endif /* HAVE_TIMEZONE */
>>> - static char date_string[24];
>>> + static char date_string[32];
>>> time_t current_time;
>>> struct tm *bd_time;
>>> if (debug) {
>>
>> --
>> Frank Küster
>> Inst. f. Biochemie der Univ. Zürich
>> Debian Developer
>>
--
Frank Küster
Inst. f. Biochemie der Univ. Zürich
Debian Developer
