Re: [PATCH v3] mm: process_mrelease: introduce PROCESS_MRELEASE_REAP_KILL flag

[Oleg Nesterov]

On 05/15, Christian Brauner wrote:
>
> Right now we have clear and simple signal sending semantics for pidfds:
>
> * thread-specific pidfd -> thread-directed signal (unless signal is thread-group scoped by default)
> * thread-group pidfd -> thread-group directed signal

I don't even understand why this patch uses pidfd to send SIGKILL...

> > +	reap = reap_kill || task_will_free_mem(p);
> > +	if (!reap) {
> >  		/* Error only if the work has not been done already */
> >  		if (!mm_flags_test(MMF_OOM_SKIP, mm))
> >  			ret = -EINVAL;
> > +		task_unlock(p);
> > +		goto put_task;
> >  	}
> > +
> > +	mmgrab(mm);
> >  	task_unlock(p);
> >
> > -	if (!reap)
> > -		goto drop_mm;
> > +	if (reap_kill) {
> > +		ret = kill_all_shared_mm(task, mm);

If task_will_free_mem() above succeeded, we do not need to call
kill_all_shared_mm() ?

Oleg.