IB/mlx5: Avoid passing an invalid QP type to firmware

During QP creation, the mlx5 driver translates the QP type to an
internal value which is passed on to FW. There was no check to make
sure that the translated value is valid, and -EINVAL was coerced into
the mailbox command.

Current firmware refuses this as an invalid QP type, but future/past
firmware may do something else.

Fixes: 09a7d9eca1a6c ('{net,IB}/mlx5: QP/XRCD commands via mlx5 ifc')
Reviewed-by: Ilya Lesokhin <ilyal@mellanox.com>
Signed-off-by: Noa Osherovich <noaos@mellanox.com>
Signed-off-by: Leon Romanovsky <leon@kernel.org>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>

diff --git a/drivers/infiniband/hw/mlx5/qp.c b/drivers/infiniband/hw/mlx5/qp.c
index 39d24bf694a864788f2d3dc806c3019bb4b12b8b..e8d7eaf0670ca5aa109c934e1f09bf0020302a57 100644 (file)
--- a/drivers/infiniband/hw/mlx5/qp.c
+++ b/drivers/infiniband/hw/mlx5/qp.c
@@ -1584,6 +1584,7 @@ static int create_qp_common(struct mlx5_ib_dev *dev, struct ib_pd *pd,
        u32 uidx = MLX5_IB_DEFAULT_UIDX;
        struct mlx5_ib_create_qp ucmd;
        struct mlx5_ib_qp_base *base;
+       int mlx5_st;
        void *qpc;
        u32 *in;
        int err;
@@ -1592,6 +1593,10 @@ static int create_qp_common(struct mlx5_ib_dev *dev, struct ib_pd *pd,
        spin_lock_init(&qp->sq.lock);
        spin_lock_init(&qp->rq.lock);
 
+       mlx5_st = to_mlx5_st(init_attr->qp_type);
+       if (mlx5_st < 0)
+               return -EINVAL;
+
        if (init_attr->rwq_ind_tbl) {
                if (!udata)
                        return -ENOSYS;
@@ -1753,7 +1758,7 @@ static int create_qp_common(struct mlx5_ib_dev *dev, struct ib_pd *pd,
 
        qpc = MLX5_ADDR_OF(create_qp_in, in, qpc);
 
-       MLX5_SET(qpc, qpc, st, to_mlx5_st(init_attr->qp_type));
+       MLX5_SET(qpc, qpc, st, mlx5_st);
        MLX5_SET(qpc, qpc, pm_state, MLX5_QP_PM_MIGRATED);
 
        if (init_attr->qp_type != MLX5_IB_QPT_REG_UMR)