FreeTDS Development Group

[Daniel Parnell <me AT danielparnell.com>]

Hi James,

Looks like I've got things working on my machine.
What I had to do was replace any calls to strdup with a new function  
tds_strdup.
I also have to replace the call to asprintf in odbc.c with sprintf  
into a character array and a tds_strdup (not too happy about that one).

It's very VERY strange that I should need to do this.  I've tried  
replicating the problem with a smaller application without success.

Daniel

On 08/03/2009, at 3:10 AM, James K. Lowden wrote:

> Hi Daniel,
>
> > I put some printfs in around the free and it appears to be pointing  
> > to
> > a valid date format.
>
> That's interesting. I don't know anything about OS X heap management  
> or
> how an executable is arranged in memory.  My instinct says 0x16 is  
> too low
> for the heap, and free(3) seems to agree.
>
> Looking at the code, there are only a few places where data_fmt is  
> set,
> all of them output of calloc(3) or strdup(3).  I'm sure you see the  
> same
> yourself.  I don't see how FreeTDS could have set that pointer to  
> anything
> that can't be freed.
>
> > If I just comment out the free it then seems to fail on the next free
> > the code hits (and of course leaks;)
> > It is almost as if the heap is getting corrupted
>
> The OS X tool vmmap(1) might be of some help here.  I haven't used it
> myself but it's in the right neighborhood.
>
> I still suspect someone somewhere is trimming off the top 4 bytes of a
> pointer and then having a party.
>
> HTH.
>
> --jkl
> _______________________________________________
> FreeTDS mailing list
> FreeTDS AT lists.ibiblio.org
> http://lists.ibiblio.org/mailman/listinfo/freetds