[unixODBC-dev] Segfault when utilizing libodbc with Asterisk

Leif Madsen leif.madsen at gmail.com
Fri May 18 16:02:24 BST 2007

On 5/18/07, Nick Gorham <nick.gorham at easysoft.com> wrote:
> > I apologize now if I'm utilizing the wrong mailing list.
> No, this will do :-)

Excellent :)

> > I am utilizing unixODBC package (unixODBC and unixODBC-devel version
> > 2.2.12-1.el4s1.1) on CentOS 4.4, and have been seeing some random
> > segfaults of Asterisk when utilizing the libodbc libraries. The
> > developers of Asterisk have stated they believe it to be a problem in
> > the libodbc driver, and the developers of the Progress (not
> > postgresql) database have fixed a similar issue apparently for another
> > implementor I speak with regularly and his crashes seem to have become
> > resolved. (I do not have the fix he received from the vendor, but if
> > you think it would be of use, then I can follow up with that
> > information.)
> Ok, I don't know exactly how (or what) 2.2.12-1.el4s1 is, I guess its
> based on the 2.2.12 tree, but as I know people started producing builds
> before it was released your guess is as good as mine. I could do with a
> bit more information, what odbc driver are you using under libodbc? Yes
> any details from the postgres folk could help.

I'm using the PostgreSQL ODBC driver version 08.01.0200. It looks like
Stefan replied before I woke up and was able to provide some more
information about the package.

> The backtrace show two failures in different places, that makes me
> suspect memory corruption somewhere. May be unixODBC, may be the odbc
> driver under it, may be something else.
> I would get either the real released 2.2.12 or the current 2.2.13 from
> the ftp site, and build from source (making sure its building with
> threads), try that, and see if it gives you any more information.

OK, I will give that a shot next and see if I can't try to find some
more information...

> Failing that, maybe something like valgrind would let you find the place
> the actual corruption happened, I suspect the crash you are seing is the
> result of a problem that happened at some point before the actual crash.

Since I'm not very familiar with valgrind, would you mind giving some
pointers how I might utilize that to help track the issue?

Thanks so much for your quick response!

Leif Madsen.

More information about the unixODBC-dev mailing list