Re: [PATCH 2/2] mm: call complete_vfork_done() after clearingchild_tid and flushing rss-counters

[Andrew Morton]

On Thu, 12 Apr 2012 12:09:53 +0400
Konstantin Khlebnikov <khlebnikov@xxxxxxxxxx> wrote:

> Child should wake ups parent from vfork() only after finishing all operations with
> shared mm. There is no sense to use CLONE_CHILD_CLEARTID together with CLONE_VFORK,
> but it looks more accurate now.
> 
> ...
>
> --- a/kernel/fork.c
> +++ b/kernel/fork.c
> @@ -728,9 +728,6 @@ void mm_release(struct task_struct *tsk, struct mm_struct *mm)
>  	/* Get rid of any cached register state */
>  	deactivate_mm(tsk, mm);
>  
> -	if (tsk->vfork_done)
> -		complete_vfork_done(tsk);
> -
>  	/*
>  	 * If we're exiting normally, clear a user-space tid field if
>  	 * requested.  We leave this alone when dying by signal, to leave
> @@ -759,6 +756,13 @@ void mm_release(struct task_struct *tsk, struct mm_struct *mm)
>  	 */
>  	if (mm)
>  		sync_mm_rss(mm);
> +
> +	/*
> +	 * All done, finally we can wake up parent and return this mm to him.
> +	 * Also kthread_stop() uses this completion for synchronization.
> +	 */
> +	if (tsk->vfork_done)
> +		complete_vfork_done(tsk);
>  }

That does look a bit racy.

But are we really sure that the patch really does fix something? 
Because it does increase vfork() latency a tiny bit.

I'm going to call this a patch against the fork subsystem, not the mm
subsystem.

I believe that this patch is unrelated to "mm: set task exit code
before complete_vfork_done()", yes?
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/