Re: [patch] O(1) sys_exit(), threading, scalable-exit-2.5.31-A6

[Ingo Molnar]

On Mon, 19 Aug 2002, Dave McCracken wrote:

> > No, you only need to make debugged children slightly pecial in wait4(), in
> > that the parent must never see their state, only the fact that they are
> > there (as if they were still running, in short, regardless of their _real_
> > state)
> 
> So does that mean we need something like a 'count of children stolen by
> debuggers' in the task struct that wait4() can check?

in fact we have this already, almost:

	if (!list_empty(&current->ptrace_children))

then block (or return -EAGAIN). Instead of the current -ENOCHLD.

	Ingo