In cm_destroy_id(), when the cm_id state is IB_CM_REQ_SENT or
IB_CM_MRA_REQ_RCVD, an attempt is made to send a REJ with
IB_CM_REJ_TIMEOUT as the reject reason.
However, in said states, we have no remote_id. For the REQ_SENT case,
we simply haven't received anything from our peer, for the
MRA_REQ_RCVD case, the cm_rma_handler() doesn't pick up the remote_id.
Therefore, it is no reason to send this REJ, since it simply will be
tossed at the peer's CM layer (if it reaches it). If running in CX-3
virtualized and having the pr_debug enabled in the mlx4_ib driver, we
will see:
mlx4_ib_demux_cm_handler: Couldn't find an entry for pv_cm_id 0x0
because the proxy de-muxer is unable to determine which slave this MAD
packet should be sent to.
Signed-off-by: Håkon Bugge <haakon.bugge@xxxxxxxxxx>
---
drivers/infiniband/core/cm.c | 4 ----
1 file changed, 4 deletions(-)
diff --git a/drivers/infiniband/core/cm.c b/drivers/infiniband/core/cm.c
index 4794113ecd59..ed80e5b56e2b 100644
--- a/drivers/infiniband/core/cm.c
+++ b/drivers/infiniband/core/cm.c
@@ -1058,10 +1058,6 @@ static void cm_destroy_id(struct ib_cm_id *cm_id, int err)
case IB_CM_REQ_SENT:
case IB_CM_MRA_REQ_RCVD:
ib_cancel_mad(cm_id_priv->av.port->mad_agent, cm_id_priv->msg);
- cm_send_rej_locked(cm_id_priv, IB_CM_REJ_TIMEOUT,
- &cm_id_priv->id.device->node_guid,
- sizeof(cm_id_priv->id.device->node_guid),
- NULL, 0);
break;
case IB_CM_REQ_RCVD:
if (err == -ENOMEM) {
--
2.20.1
