Re: give_terminal_to() / maybe_give_terminal_to() race

[Earl Chew]

On 2023-09-05 08:46, Chet Ramey wrote:
> Such as? If the parent doesn't need to change the terminal process group,
> synchronizes with the pipeline via waitpid(), and doesn't attempt to change
> the terminal process group back until all children in that process group
> have stopped or exited, what consequences are of concern?

I think the only other side-effects relate to IO (eg SIGTTIN, SIGTTOU),
but as far as I can tell these are not of concern since the parent stays
quiet until the foreground job either stops or completes.

> > Given that the children must synchronise before exec(), there is no
> > issue if all configure the terminal though it is sufficient
> > if only the first child does so.
>
> Not true; ...

You're right, I have that backwards. If only one child configures the
terminal, then the siblings must wait until that child has completed
doing so before they invoke exec(). Rather than do that, given that
subsequent invocations of give_terminal_to() (ie tcsetpgrp()) by siblings
will find the foreground group already configured, it makes sense to
simply let every sibling do so.

Earl