Hi Roland! > Thanks, I applied this with some extra code in all the low-level > drivers to make sure that the create_flags are passed in as 0. Does > that make sense to everyone? Below changes make sense to me as I would have to check the flags when introducing LL QP flag for ehca later. BTW: If you have some minutes, please let us agree on the encoding scheme for qp_types and create_flags as discussed in this thread. Thanks! Nam > diff --git a/drivers/infiniband/hw/ehca/ehca_qp.c > b/drivers/infiniband/hw/ehca/ehca_qp.c > index a9fd419..3eb14a5 100644 > --- a/drivers/infiniband/hw/ehca/ehca_qp.c > +++ b/drivers/infiniband/hw/ehca/ehca_qp.c > @@ -421,6 +421,9 @@ static struct ehca_qp *internal_create_qp( > u32 swqe_size = 0, rwqe_size = 0, ib_qp_num; > unsigned long flags; > > + if (init_attr->create_flags) > + return ERR_PTR(-EINVAL); > + > memset(&parms, 0, sizeof(parms)); > qp_type = init_attr->qp_type; >
_______________________________________________ general mailing list [email protected] http://lists.openfabrics.org/cgi-bin/mailman/listinfo/general To unsubscribe, please visit http://openib.org/mailman/listinfo/openib-general
