svn commit: r268660 - head/sys/amd64/amd64

[Ian Lepore]

On Tue, 2014-07-15 at 18:30 +0300, Konstantin Belousov wrote:
> On Tue, Jul 15, 2014 at 07:52:20AM -0600, Ian Lepore wrote:
> > >  		bcopy(a_cp, b_cp, cnt);
> > > +		if (__predict_false(pinned)) {
> > > +			sched_unpin();
> > > +			mtx_unlock(&cpage_lock);
> > > +			pinned = FALSE;
> > 
> > Should this pinned = FALSE be done under the cpage_lock to avoid a race?
> 
> What race ? Could you describe it in more details ?
> Note that pinned is local variable.

Ahh, so it is.  I just noticed that it was set under the lock and
cleared outside of the lock, and didn't look at the scope of the
variable.  Sorry for the noise.

-- Ian