Commit fca4d518 authored by Jon Mason's avatar Jon Mason
Browse files

NTB: Fix ntb_transport link down race 



A WARN_ON is being hit in ntb_qp_link_work due to the NTB transport link
being down while the ntb qp link is still active.  This is caused by the
transport link being brought down prior to the qp link worker thread
being terminated.  To correct this, shutdown the qp's prior to bringing
the transport link down.  Also, only call the qp worker thread if it is
in interrupt context, otherwise call the function directly.

Signed-off-by: default avatarJon Mason <jon.mason@intel.com>
parent 97390473

drivers/ntb/ntb_transport.c

+28 −15
Original line number Diff line number Diff line
@@ -584,11 +584,8 @@ static int ntb_set_mw(struct ntb_transport *nt, int num_mw, unsigned int size) 
	return 0;

}



static void ntb_qp_link_cleanup(struct work_struct *work)

static void ntb_qp_link_cleanup(struct ntb_transport_qp *qp)

{

	struct ntb_transport_qp *qp = container_of(work,

						   struct ntb_transport_qp,

						   link_cleanup);

	struct ntb_transport *nt = qp->transport;

	struct pci_dev *pdev = ntb_query_pdev(nt->ndev);
@@ -602,6 +599,16 @@ static void ntb_qp_link_cleanup(struct work_struct *work) 


	dev_info(&pdev->dev, "qp %d: Link Down\n", qp->qp_num);

	qp->qp_link = NTB_LINK_DOWN;

}



static void ntb_qp_link_cleanup_work(struct work_struct *work)

{

	struct ntb_transport_qp *qp = container_of(work,

						   struct ntb_transport_qp,

						   link_cleanup);

	struct ntb_transport *nt = qp->transport;



	ntb_qp_link_cleanup(qp);



	if (nt->transport_link == NTB_LINK_UP)

		schedule_delayed_work(&qp->link_work,
@@ -613,22 +620,20 @@ static void ntb_qp_link_down(struct ntb_transport_qp *qp) 
	schedule_work(&qp->link_cleanup);

}



static void ntb_transport_link_cleanup(struct work_struct *work)

static void ntb_transport_link_cleanup(struct ntb_transport *nt)

{

	struct ntb_transport *nt = container_of(work, struct ntb_transport,

						link_cleanup);

	int i;



	/* Pass along the info to any clients */

	for (i = 0; i < nt->max_qps; i++)

		if (!test_bit(i, &nt->qp_bitmap))

			ntb_qp_link_cleanup(&nt->qps[i]);



	if (nt->transport_link == NTB_LINK_DOWN)

		cancel_delayed_work_sync(&nt->link_work);

	else

		nt->transport_link = NTB_LINK_DOWN;



	/* Pass along the info to any clients */

	for (i = 0; i < nt->max_qps; i++)

		if (!test_bit(i, &nt->qp_bitmap))

			ntb_qp_link_down(&nt->qps[i]);



	/* The scratchpad registers keep the values if the remote side

	 * goes down, blast them now to give them a sane value the next

	 * time they are accessed
@@ -637,6 +642,14 @@ static void ntb_transport_link_cleanup(struct work_struct *work) 
		ntb_write_local_spad(nt->ndev, i, 0);

}



static void ntb_transport_link_cleanup_work(struct work_struct *work)

{

	struct ntb_transport *nt = container_of(work, struct ntb_transport,

						link_cleanup);



	ntb_transport_link_cleanup(nt);

}



static void ntb_transport_event_callback(void *data, enum ntb_hw_event event)

{

	struct ntb_transport *nt = data;
@@ -880,7 +893,7 @@ static int ntb_transport_init_queue(struct ntb_transport *nt, 
	}



	INIT_DELAYED_WORK(&qp->link_work, ntb_qp_link_work);

	INIT_WORK(&qp->link_cleanup, ntb_qp_link_cleanup);

	INIT_WORK(&qp->link_cleanup, ntb_qp_link_cleanup_work);



	spin_lock_init(&qp->ntb_rx_pend_q_lock);

	spin_lock_init(&qp->ntb_rx_free_q_lock);
@@ -936,7 +949,7 @@ int ntb_transport_init(struct pci_dev *pdev) 
	}



	INIT_DELAYED_WORK(&nt->link_work, ntb_transport_link_work);

	INIT_WORK(&nt->link_cleanup, ntb_transport_link_cleanup);

	INIT_WORK(&nt->link_cleanup, ntb_transport_link_cleanup_work);



	rc = ntb_register_event_callback(nt->ndev,

					 ntb_transport_event_callback);
@@ -972,7 +985,7 @@ void ntb_transport_free(void *transport) 
	struct ntb_device *ndev = nt->ndev;

	int i;



	nt->transport_link = NTB_LINK_DOWN;

	ntb_transport_link_cleanup(nt);



	/* verify that all the qp's are freed */

	for (i = 0; i < nt->max_qps; i++) {

CRA Git | Maintained and supported by SUSTech CRA and CCSE