Small leak in kadm5_get_init_creds

[Jerry James]

Tom Yu <tlyu at MIT.EDU>, on Fri, 09 Mar 2007 at 16:16:24 -0500 you wrote:
>>>>>> "Jerry" == Jerry James <Jerry.James at usu.edu> writes:
>
> Jerry> I found this on Fedora Core 6, which is using 1.5, but I downloaded the
> Jerry> sources and found the same leak is present in 1.6.  The function
> Jerry> kadm5_get_init_creds fails to clean up a principal before exiting.  This
> Jerry> patch plugs the leak.
>
> [...]
>
> Jerry> I first sent this to kerberos at mit.edu, but the list management software
> Jerry> classified the patch (which was attached) as a binary and quarantined
> Jerry> it, and also threw away the text part of my message.  Is it configured
> Jerry> that way on purpose?
>
> Both your text and patch arrived intact as far as I could see on
> kerberos at mit.edu, but I haven't checked to see if the archives have it
> intact.

I'm glad the mail arrived intact.  The archives don't have the text
portion:

http://mailman.mit.edu/pipermail/kerberos/2007-March/011365.html

>          Is this bug related to ticket #1000 in our bug database?
>
> http://krbdev.mit.edu/rt/Ticket/Display.html?id=1000
>
> It looks similar, but it seems like it covers a different set of error
> cases.

It is in the same region of code, but it appears to be a different bug.
The leak I reported always happens; ticket #1000 is for a leak that only
occurs when there is an error.  Plus, the structure being leaked is a
principal in my case, but is a context and a handle in the case of
ticket #1000.

Regards,
-- 
Jerry James, Assistant Professor        Jerry.James at usu.edu
Computer Science Department             http://www.cs.usu.edu/~jerry/
Utah State University