Re: PATCH: job control debug

[Bart Schaefer <schaefer@xxxxxxxxxxxxxxxx>]

On Thu, Sep 6, 2018 at 2:22 AM, Peter Stephenson
<p.stephenson@xxxxxxxxxxx> wrote:
> Typo fixed, now with setpgrp debugging.
>
> Starting to think the key difference may be set_gleader(11) --- transfer
> of job ownership on exit, which happens too early for that to be
> possible in the bad case.

I don't think that can be it, at least not directly.  set_gleader()
only updates internal data structures.  Something has to be either
changing the tty process group too soon, or not soon enough, in order
for a TTOU signal to be sent.  The other possibility is that in the
"good" case the TTOU is sent but the process is ignoring it, whereas
in the bad case the process is allowing it.  You may need to add some
debugging for that.

In particular entersubsh() does some juggling of SIGTTOU -- which I
think predates this bug appearing, but there may be an interaction of
that with some later change.