[PATCH] Coverity fixes.

[Jeremy Allison]

On Sat, Nov 10, 2018 at 12:05:31AM +0100, Stefan Metzmacher via samba-technical wrote:
> Am 10.11.18 um 00:02 schrieb Stefan Metzmacher via samba-technical:
> > Am 09.11.18 um 22:25 schrieb Jeremy Allison via samba-technical:
> >> Subject: [PATCH 1/2] s4: lib: messaging. Fix dead code. Coverity CID: 1437973
> >>
> >> Signed-off-by: Jeremy Allison <jra at samba.org>
> >> ---
> >>  source4/lib/messaging/messaging.c | 4 ----
> >>  1 file changed, 4 deletions(-)
> >>
> >> diff --git a/source4/lib/messaging/messaging.c b/source4/lib/messaging/messaging.c
> >> index 935951f3fba..dada565d1be 100644
> >> --- a/source4/lib/messaging/messaging.c
> >> +++ b/source4/lib/messaging/messaging.c
> >> @@ -470,10 +470,6 @@ static void imessaging_post_handler(struct tevent_context *ev,
> >>  	imessaging_dgm_recv(ev, state->buf, state->buf_len, NULL, 0,
> >>  			    state->msg_ctx);
> >>  
> >> -	if (state == NULL) {
> >> -		return;
> >> -	}
> >> -
> >>  	state->busy_ref = NULL;
> >>  	TALLOC_FREE(state);
> >>  }
> > 
> > This is wrong! See a08ab2940051ae47ce71149087a24d060227ef19
> > 
> > imessaging_post_state_destructor() may set state = NULL,
> > via the state->busy_ref pointer.
> 
> Fixing Coverity or compiler warnings is *much* harder than
> it sometimes seems and they need very careful review!

Well not *all* of them :-). That one was particularly
nasty, as it involved a destructor reaching into local
varaible state and modifying it.

That's pretty rare (or at least I'd hope so). Most
of them are much more straightforward than that.