Re: [PATCH v5 5/9] IB/isert: Replace semaphore sem with completion

[Sagi Grimberg]

> > > diff --git a/drivers/infiniband/ulp/isert/ib_isert.c b/drivers/infiniband/ulp/isert/ib_isert.c
> > > index 6dd43f6..de80f56 100644
> > > --- a/drivers/infiniband/ulp/isert/ib_isert.c
> > > +++ b/drivers/infiniband/ulp/isert/ib_isert.c
> > > @@ -619,7 +619,7 @@
> > >       mutex_unlock(&isert_np->mutex);
> > >
> > >       isert_info("np %p: Allow accept_np to continue\n", isert_np);
> > > -     up(&isert_np->sem);
> > > +     complete(&isert_np->comp);
> > >  }
> > >
> > >  static void
> > > @@ -2311,7 +2311,7 @@ struct rdma_cm_id *
> > >               isert_err("Unable to allocate struct isert_np\n");
> > >               return -ENOMEM;
> > >       }
> > > -     sema_init(&isert_np->sem, 0);
> > > +     init_completion(&isert_np->comp);
> >
> > This is still racy, a connect event can complete just before we
> > init the completion and *will* get lost...
> >
> > This code started off with a waitqueue which exposes the same
> > problem, see:
> > 531b7bf4bd79 Target/iser: Fix iscsit_accept_np and rdma_cm racy flow
> >
> > So, still NAK from me...
>
> I don't see what you mean here. The code using a waitqueue had no
> counter but the completion does.

The problem here is that init_completion sets the counter to zero
and between this thread wakes up and until it initializes comp->done
we might have another connect event completing it and I fail to
see how this is handled correctly.