Commit 2df6d895 authored by Shiraz Saleem's avatar Shiraz Saleem Committed by Jason Gunthorpe
Browse files

RDMA/irdma: Reduce iWARP QP destroy time 

QP destroy is synchronous and waits for its refcnt to be decremented in
irdma_cm_node_free_cb (for iWARP) which fires after the RCU grace period
elapses.

Applications running a large number of connections are exposed to high
wait times on destroy QP for events like SIGABORT.

The long pole for this wait time is the firing of the call_rcu callback
during a CM node destroy which can be slow. It holds the QP reference
count and blocks the destroy QP from completing.

call_rcu only needs to make sure that list walkers have a reference to the
cm_node object before freeing it and thus need to wait for grace period
elapse. The rest of the connection teardown in irdma_cm_node_free_cb is
moved out of the grace period wait in irdma_destroy_connection. Also,
replace call_rcu with a simple kfree_rcu as it just needs to do a kfree on
the cm_node

Fixes: 146b9756 ("RDMA/irdma: Add connection manager")
Link: https://lore.kernel.org/r/20220425181703.1634-3-shiraz.saleem@intel.com


Signed-off-by: default avatarShiraz Saleem <shiraz.saleem@intel.com>
Signed-off-by: default avatarJason Gunthorpe <jgg@nvidia.com>
parent 7b8943b8

drivers/infiniband/hw/irdma/cm.c

+4 −6
Original line number Diff line number Diff line
@@ -2308,10 +2308,8 @@ irdma_make_cm_node(struct irdma_cm_core *cm_core, struct irdma_device *iwdev, 
	return NULL;

}



static void irdma_cm_node_free_cb(struct rcu_head *rcu_head)

static void irdma_destroy_connection(struct irdma_cm_node *cm_node)

{

	struct irdma_cm_node *cm_node =

			    container_of(rcu_head, struct irdma_cm_node, rcu_head);

	struct irdma_cm_core *cm_core = cm_node->cm_core;

	struct irdma_qp *iwqp;

	struct irdma_cm_info nfo;
@@ -2359,7 +2357,6 @@ static void irdma_cm_node_free_cb(struct rcu_head *rcu_head) 
	}



	cm_core->cm_free_ah(cm_node);

	kfree(cm_node);

}



/**
@@ -2387,8 +2384,9 @@ void irdma_rem_ref_cm_node(struct irdma_cm_node *cm_node) 


	spin_unlock_irqrestore(&cm_core->ht_lock, flags);



	/* wait for all list walkers to exit their grace period */

	call_rcu(&cm_node->rcu_head, irdma_cm_node_free_cb);

	irdma_destroy_connection(cm_node);



	kfree_rcu(cm_node, rcu_head);

}



/**