[v2] Add support for RDMA MAD

[Qemu-devel] [PATCH v2 03/22] hw/rdma: Return qpn 1 if ibqp is NULL

Posted by Yuval Shaia 7 years, 3 months ago

Device is not supporting QP0, only QP1.

Signed-off-by: Yuval Shaia <yuval.shaia@oracle.com>
---
 hw/rdma/rdma_backend.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/hw/rdma/rdma_backend.h b/hw/rdma/rdma_backend.h
index 86e8fe8ab6..3ccc9a2494 100644
--- a/hw/rdma/rdma_backend.h
+++ b/hw/rdma/rdma_backend.h
@@ -33,7 +33,7 @@ static inline union ibv_gid *rdma_backend_gid(RdmaBackendDev *dev)
 
 static inline uint32_t rdma_backend_qpn(const RdmaBackendQP *qp)
 {
-    return qp->ibqp ? qp->ibqp->qp_num : 0;
+    return qp->ibqp ? qp->ibqp->qp_num : 1;
 }
 
 static inline uint32_t rdma_backend_mr_lkey(const RdmaBackendMR *mr)
-- 
2.17.2

Re: [Qemu-devel] [PATCH v2 03/22] hw/rdma: Return qpn 1 if ibqp is NULL

Posted by Marcel Apfelbaum 7 years, 3 months ago

Hi Yuval,

On 11/8/18 6:07 PM, Yuval Shaia wrote:
> Device is not supporting QP0, only QP1.
>
> Signed-off-by: Yuval Shaia <yuval.shaia@oracle.com>
> ---
>   hw/rdma/rdma_backend.h | 2 +-
>   1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/hw/rdma/rdma_backend.h b/hw/rdma/rdma_backend.h
> index 86e8fe8ab6..3ccc9a2494 100644
> --- a/hw/rdma/rdma_backend.h
> +++ b/hw/rdma/rdma_backend.h
> @@ -33,7 +33,7 @@ static inline union ibv_gid *rdma_backend_gid(RdmaBackendDev *dev)
>   
>   static inline uint32_t rdma_backend_qpn(const RdmaBackendQP *qp)
>   {
> -    return qp->ibqp ? qp->ibqp->qp_num : 0;
> +    return qp->ibqp ? qp->ibqp->qp_num : 1;

Just to be sure, what are the cases we don't get  a qp_num?
Can we assume all of them are MADs?

Thanks,
Marcel

>   }
>   
>   static inline uint32_t rdma_backend_mr_lkey(const RdmaBackendMR *mr)

Re: [Qemu-devel] [PATCH v2 03/22] hw/rdma: Return qpn 1 if ibqp is NULL

Posted by Yuval Shaia 7 years, 3 months ago

On Sat, Nov 10, 2018 at 07:59:00PM +0200, Marcel Apfelbaum wrote:
> Hi Yuval,
> 
> On 11/8/18 6:07 PM, Yuval Shaia wrote:
> > Device is not supporting QP0, only QP1.
> > 
> > Signed-off-by: Yuval Shaia <yuval.shaia@oracle.com>
> > ---
> >   hw/rdma/rdma_backend.h | 2 +-
> >   1 file changed, 1 insertion(+), 1 deletion(-)
> > 
> > diff --git a/hw/rdma/rdma_backend.h b/hw/rdma/rdma_backend.h
> > index 86e8fe8ab6..3ccc9a2494 100644
> > --- a/hw/rdma/rdma_backend.h
> > +++ b/hw/rdma/rdma_backend.h
> > @@ -33,7 +33,7 @@ static inline union ibv_gid *rdma_backend_gid(RdmaBackendDev *dev)
> >   static inline uint32_t rdma_backend_qpn(const RdmaBackendQP *qp)
> >   {
> > -    return qp->ibqp ? qp->ibqp->qp_num : 0;
> > +    return qp->ibqp ? qp->ibqp->qp_num : 1;
> 
> Just to be sure, what are the cases we don't get  a qp_num?
> Can we assume all of them are MADs?
> 
> Thanks,
> Marcel

qp->ibqp is set only in case that QP type is not QP 1 (see
rdma_backend_create_qp()) so we can safely assume that this is QP 1.

> 
> >   }
> >   static inline uint32_t rdma_backend_mr_lkey(const RdmaBackendMR *mr)
>