On Mon, May 27, 2019 at 11:34 PM Herbert Xu <herb...@gondor.apana.org.au> wrote:
>
> Hi Eric:
>
> Eric Dumazet <eduma...@google.com> wrote:
> >
> > +void fqdir_exit(struct fqdir *fqdir)
> > +{
> > + fqdir->high_thresh = 0; /* prevent creation of new frags */
> > +
> > + /* paired with READ_ONCE() in inet_frag_kill() :
> > + * We want to prevent rhashtable_remove_fast() calls
> > + */
> > + smp_store_release(&fqdir->dead, true);
> > +
> > + INIT_RCU_WORK(&fqdir->destroy_rwork, fqdir_rwork_fn);
> > + queue_rcu_work(system_wq, &fqdir->destroy_rwork);
> > +
> > +}
>
> What is the smp_store_release supposed to protect here? If it's
> meant to separate the setting of dead and the subsequent destruction
> work then it doesn't work because the barrier only protects the code
> preceding it, not after.
>
This smp_store_release() is a left over of the first version of the patch, where
there was no rcu grace period enforcement.
I do not believe there is harm letting this, but if you disagree
please send a patch ;)
Thanks
