Re: [RFC ABI V2 2/8] RDMA/core: Refactor IDR to be per-device

[Jason Gunthorpe <jgunthorpe-ePGOBjL8dl3ta4EC/59zMFaTQe2KTcn/@public.gmane.org>]

On Wed, Jul 20, 2016 at 06:20:37AM +0300, Leon Romanovsky wrote:
> > > diff --git a/include/rdma/ib_verbs.h b/include/rdma/ib_verbs.h
> > > index 432bed5..14bfe3b 100644
> > > +++ b/include/rdma/ib_verbs.h
> > > @@ -1706,6 +1706,10 @@ struct ib_device {
> > >
> > >  	struct iw_cm_verbs	     *iwcm;
> > >
> > > +	struct idr		idr;
> > > +	/* Global lock in use to safely release device IDR */
> > > +	spinlock_t		idr_lock;
> > 
> > Global? In what sense? This is a lock for the idrs for the particular
> > device right? The comment is a bit confusing. Maybe better remove it?
> 
> I'll fix it and will remove "global" word from the comment.
> It should be /* Lock in use to safety release device IDR */. I
> prefer to

That doesn't seem to be the right comment either.

+	spin_lock(&uobj->context->device->idr_lock);
+	ret = idr_alloc(&uobj->context->device->idr, uobj, 0, 0, GFP_NOWAIT);
+	spin_unlock(&uobj->context->device->idr_lock);

idr_lock appears to just be the standard spinlock protecting an IDR.

If it being re-used for something about device removal then that is
probably wrong....

Jason
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html