Re: safety of *mutex_unlock() (Was: [BUG] signal: sighand unprotected when accessed by /proc)

[Paul E. McKenney]

On Wed, Jun 18, 2014 at 06:43:59PM +0200, Oleg Nesterov wrote:
> On 06/17, Paul E. McKenney wrote:
> >
> > +		if (drop_boost_mutex) {
> > +			rt_mutex_unlock(&rnp->boost_mtx);
> >  			complete(&rnp->boost_completion);
> 
> Well, I still do not understand this ->boost_completion...
> 
> > -	/* Wait until boostee is done accessing mtx before reinitializing. */
> > +	/* Wait for boostee to be done w/boost_mtx before reinitializing. */
> >  	wait_for_completion(&rnp->boost_completion);
> 
> OK, at least we have a comment.
> 
> But let me repeat. Thomas has already fixed rt_mutex, unlock is atomic.
> It doesn't touch this memory after it makes another lock() possible.
> 
> And (contrary to what I said initially) we can rely on this because -rt
> converts spinlock_t into rt_mutex ?

Well, perhaps I should rein in my paranoia on this one.  That said, the
cost of my paranoia is minimal in this slowpath.

							Thanx, Paul

--
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/