[PATCH 3/4] Use gssalloc_malloc()/gssalloc_free() withgss_buffer_t.

[Greg Hudson]

On Thu, 2011-10-06 at 15:39 -0400, Kevin Wasserman wrote:
> I am concerned that you say you get a crash when logfile is set.

I didn't see a crash experimentally; it was implied from my misreading
of the code.  I didn't realize that was a structure assignment.

> Again, though, I agree that this code is pretty badly structured
> and I'll clean it up along the lines you suggest.

Of course the code being badly structured predates your work; it's just
that the consequences aren't too severe until you introduce the
requirement that memory be freed according to how it was allocated.
Thanks for cleaning it up.