Re: [uml-devel] [PATCH 1/6] UML - Console locking fixes

[Jeff Dike]

On Wed, Jan 03, 2007 at 04:07:34PM +0100, Blaisorblade wrote:
> > +	spin_lock(&line->count_lock);
> > +	if(!line->valid)
> > +		goto out_unlock;
> > +
> > +	err = 0;
> > +	if(tty->count > 1)
> > +		goto out_unlock;
> >
> > -	/* The IRQ which takes this lock is not yet enabled and won't be run
> > -	 * before the end, so we don't need to use spin_lock_irq.*/
> > -	spin_lock(&line->lock);
> > +	mutex_lock(&line->open_mutex);
> > +	spin_unlock(&line->count_lock);
> 
> This is an obnoxious thing to do unless you specifically prove otherwise. 

Didn't I?  
The proof goes like this:
	we only take the semaphore if tty->count == 1, in which case
we are opening the device for the first time and there can't be anyone
else looking at it, so the mutex_lock won't sleep.

However, now that you're making me think about it again, I'm wondering
about the sanity of introducing a mutex which is guaranteed not to
sleep.

This is starting to make sense, with (tty->count > 1) being the
OPENING flag:

> In the first solution, you can create a OPENING flag (via a state variable), 
> and add the rule that (unlike the count) nobody but the original setter is 
> allowed to change it, and that who finds it set (say a concurrent open) must 
> return without touching it.

Then, I think the mutex can just be thrown away.

				Jeff

-- 
Work email - jdike at linux dot intel dot com
-
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/