FreeTDS Development Group

[John <ionutz22 AT gmail.com>]

>> 
>> > > 
>> > > I upgraded FreeTSD to 0.64RC2 and my scripts runs OK (I get my
>> > > results) but I get the following error/warning:
>> > > 
>> > > <snip>
>> > > done
>> > > *** glibc detected *** double free or corruption (fasttop): 
>> > > 0x08b6c9e0 ***
>> > > Aborted
>> > > </snip>
>> > > 
>> > > I get the same error if I use FreeTDS current snapshot. I 
>> > > also get the error
>> > > with the following configuration.
>> > > 
>> > > OS          - CentOS 4.3 x86_64
>> > > DBI         - 1.40
>> > > DBD::Sybase - 1.07
>> > > FreeTDS     - 0.64RC2 or current snapshot
>> > > ASE         - Adaptive Server Enterprise/12.5.3/EBF 13204 
>> > > ESD#6/P/Linux
>> > > 
>> > > If I comment out the line containing "$dbh->disconnect;" 
>> the script
>> > > doesn't generate this warning/error.
>> > > 
>> > > Did anyone encountered this kind of problem ? Can it be fixed ?
>> > > Thanks for any direction you can provide.
>> > > 
>> > 
>> > Adding a 
>> > 
>> >   $sth = undef;
>> > 
>> > before disconnect will remove the problem. The reason is that
>> > $dbh->prepare allocate a CS_COMMAND which is destroyed on 
>> ct_con_drop
>> > (called at $dbh->disconnect) so when $sth is freed is still 
>> > use a freed
>> > pointer... Now the problem is; what's the right behavior? I think
>> > however that right behavior is Sybase one...
>> > 
>> 
>> The DBI documentation
>> (http://search.cpan.org/~timb/DBI-1.50/DBI.pm#finish) suggests using
>> $sth->finish on all statement handles before disconnecting. I didn't
>> make the connection to the statement handle earlier because I've never
>> seen that glibc error message.
>> 
>> Regards,
>> Paul

> Could you try patch at
> http://freetds.sourceforge.net/post64rc2.diff.gz?
> Is a patch for 0.64RC2 (as name suggests).

> freddy77
> _______________________________________________
> FreeTDS mailing list
> FreeTDS AT lists.ibiblio.org
> http://lists.ibiblio.org/mailman/listinfo/freetds

I applied the patch and it's working and I don't get that
error/warning. It works on both configurations (i386 and x86_64).
Thanks for the fix.

-- 
Best regards,
John.