On Mon, Oct 07, 2019 at 04:18:26PM +0200, Dmitry Vyukov wrote:
> On Mon, Oct 7, 2019 at 4:14 PM Andrea Parri <parri.andrea@xxxxxxxxx> wrote:
> >
> > > > > static struct taskstats *taskstats_tgid_alloc(struct task_struct *tsk)
> > > > > {
> > > > > struct signal_struct *sig = tsk->signal;
> > > > > - struct taskstats *stats;
> > > > > + struct taskstats *stats_new, *stats;
> > > > >
> > > > > - if (sig->stats || thread_group_empty(tsk))
> > > > > - goto ret;
> > > > > + /* Pairs with smp_store_release() below. */
> > > > > + stats = READ_ONCE(sig->stats);
> > > >
> > > > This pairing suggests that the READ_ONCE() is heading an address
> > > > dependency, but I fail to identify it: what is the target memory
> > > > access of such a (putative) dependency?
> > >
> > > I would assume callers of this function access *stats. So the
> > > dependency is between loading stats and accessing *stats.
> >
> > AFAICT, the only caller of the function in 5.4-rc2 is taskstats_exit(),
> > which 'casts' the return value to a boolean (so I really don't see how
> > any address dependency could be carried over/relied upon here).
>
> This does not make sense.
>
> But later taskstats_exit does:
>
> memcpy(stats, tsk->signal->stats, sizeof(*stats));
>
> Perhaps it's supposed to use stats returned by taskstats_tgid_alloc?
Seems reasonable to me. If so, replacing the READ_ONCE() in question
with an smp_load_acquire() might be the solution. Thoughts?
Andrea
