Re: [PATCH rdma v2] RDMA: Add rdma_connect_locked()

[Jason Gunthorpe <jgg@nvidia.com>]

On Tue, Oct 27, 2020 at 03:19:36PM +0200, Leon Romanovsky wrote:

> > +int rdma_connect_locked(struct rdma_cm_id *id,
> > +			struct rdma_conn_param *conn_param)
> >  {
> >  	struct rdma_id_private *id_priv =
> >  		container_of(id, struct rdma_id_private, id);
> >  	int ret;
> >
> > -	mutex_lock(&id_priv->handler_mutex);
> >  	if (!cma_comp_exch(id_priv, RDMA_CM_ROUTE_RESOLVED, RDMA_CM_CONNECT)) {
> >  		ret = -EINVAL;
> >  		goto err_unlock;
> 
> Not a big deal, but his label is not correct anymore.

Oh, yep

diff --git a/drivers/infiniband/core/cma.c b/drivers/infiniband/core/cma.c
index 193c8902b9db26..f58d19881524dc 100644
--- a/drivers/infiniband/core/cma.c
+++ b/drivers/infiniband/core/cma.c
@@ -4053,10 +4053,8 @@ int rdma_connect_locked(struct rdma_cm_id *id,
 		container_of(id, struct rdma_id_private, id);
 	int ret;
 
-	if (!cma_comp_exch(id_priv, RDMA_CM_ROUTE_RESOLVED, RDMA_CM_CONNECT)) {
-		ret = -EINVAL;
-		goto err_unlock;
-	}
+	if (!cma_comp_exch(id_priv, RDMA_CM_ROUTE_RESOLVED, RDMA_CM_CONNECT))
+		return -EINVAL;
 
 	if (!id->qp) {
 		id_priv->qp_num = conn_param->qp_num;
@@ -4078,7 +4076,6 @@ int rdma_connect_locked(struct rdma_cm_id *id,
 	return 0;
 err_state:
 	cma_comp_exch(id_priv, RDMA_CM_CONNECT, RDMA_CM_ROUTE_RESOLVED);
-err_unlock:
 	return ret;
 }
 EXPORT_SYMBOL(rdma_connect_locked);