[Swan-dev] unstable connection serial numbers

[Andrew Cagney]

On Tue, 27 Oct 2020 at 10:44, Paul Wouters <paul at nohats.ca> wrote:
>
> On Tue, 27 Oct 2020, Andrew Cagney wrote:
>
> > I've encountered another race.  Tests such as algparse-02-fips, which
> > start pluto only to immediately shut it down, can leak:
> >
> > leak-detective enabled
> > leak: fork pid, item size: 96
> > leak detective found 1 leaks, total size 96
> >
> > because the addconn child process is still running.
>
> I've changed it to ignore that allocation in leak-detective.

I'd think of this as the messenger -> pluto, even when exiting
cleanly, doesn't reap its children.