On Mon, 2006-07-31 at 10:50 -0400, Alex Polvi wrote:
> Proposed (trivial) patch to fix a NULL pointer deref in
> gss_pipe_release(). While this does seem to fix the problem, I'm not
> entirely sure it is the correct place to handle the NULL pointer.
> Included below is the script I used to recreate the problem, the oops,
> and the patch.
Sorry, but that is not the correct fix. The problem here is rather that
something is causing us to call rpc_close_pipes() on the file after the
call to gss_destroy(). That is supposed to be illegal.