Re: [PATCH 03/16] signal: remove superflous try_to_freeze() loop in do_signal_stop()

[Oleg Nesterov <oleg@redhat.com>]

On 12/06, Tejun Heo wrote:
>
> do_signal_stop() is used only by get_signal_to_deliver() and after a
> successful signal stop, it always calls try_to_freeze(), so the
> try_to_freeze() loop around schedule() in do_signal_stop() is
> superflous and confusing.  Remove it.

I think the patch is obvioulsy fine.

Acked-by: Oleg Nesterov <oleg@redhat.com>



But I am a bit confused, a couple of off-topic questions to Rafael.

> --- a/kernel/signal.c
> +++ b/kernel/signal.c
> @@ -1781,9 +1781,7 @@ static int do_signal_stop(int signr)
>  	}
>
>  	/* Now we don't run again until woken by SIGCONT or SIGKILL */
> -	do {
> -		schedule();
> -	} while (try_to_freeze());
> +	schedule();

I am wondering what was the purpose of this do/while loop. Probably
this was just oversight. We always return in TASK_RUNNING state from
schedule, try_to_freeze() should return the task in this state too.

My question is: refrigerator() tries to preserve the caller's state.
Why? I think it is just wrong to call refrigerator() unless the
task is TASK_RUNNING, no? If no, then the games with saving/restoring
->state look obviously racy/wrong.

Oleg.