svn commit: r250609 - head/sys/dev/ath

[Ian Lepore]

On Wed, 2019-05-22 at 16:20 +0000, Alexey Dokuchaev wrote:
> On Mon, May 13, 2013 at 07:03:13PM +0000, Adrian Chadd wrote:
> > New Revision: 250609
> > URL: http://svnweb.freebsd.org/changeset/base/250609
> > 
> > Log:
> >   Since the node state is 100% back under the TX lock, just kill
> > the use
> >   of atomics.
> >   
> > Modified:
> >   head/sys/dev/ath/if_ath.c
> > 
> > @@ -6140,13 +6133,13 @@ ath_tx_update_tim(struct ath_softc *sc,
> >  		/*
> >  		 * Don't bother grabbing the lock unless the queue is
> > empty.
> >  		 */
> > -		if (atomic_load_acq_int(&an->an_swq_depth) != 0)
> > +		if (&an->an_swq_depth != 0)
> >  			return;
> >  
> >  		if (an->an_is_powersave &&
> >  		    an->an_stack_psq == 0 &&
> >  		    an->an_tim_set == 1 &&
> > -		    atomic_load_acq_int(&an->an_swq_depth) == 0) {
> > +		    an->an_swq_depth == 0) {
> 
> PVS Studio complains here: warning: V560 A part of conditional
> expression
> is always true: an->an_swq_depth == 0.  Which probably makes sense
> since
> you return earlier if it's != 0.
> 
> ./danfe
> 

You're replying to a six year old commit?

It doesn't check earlier whether that value is 0, it checks whether the
address of that value is 0.  That's probably a bug.

-- Ian