From be8822db62ddda6d316d2dd682679732ed2f0abf Mon Sep 17 00:00:00 2001
From: Mustafa Ismail <mustafa.ismail@intel.com>
Date: Fri, 23 Jun 2017 16:03:55 -0500
Subject: i40iw: Fix order of cleanup in close

The order for calling i40iw_destroy_pble_pool is incorrect.
Also, add PBLE_CHUNK_MEM init state to track pble pool
creation and destruction.

Signed-off-by: Mustafa Ismail <mustafa.ismail@intel.com>
Signed-off-by: Henry Orosco <henry.orosco@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
---
 drivers/infiniband/hw/i40iw/i40iw_main.c | 7 ++++---
 1 file changed, 4 insertions(+), 3 deletions(-)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/hw/i40iw/i40iw_main.c b/drivers/infiniband/hw/i40iw/i40iw_main.c
index e0f47cc2effc..8fc61b3bd223 100644
--- a/drivers/infiniband/hw/i40iw/i40iw_main.c
+++ b/drivers/infiniband/hw/i40iw/i40iw_main.c
@@ -1474,6 +1474,9 @@ static void i40iw_deinit_device(struct i40iw_device *iwdev, bool reset)
 			unregister_inet6addr_notifier(&i40iw_inetaddr6_notifier);
 		}
 		/* fallthrough */
+	case PBLE_CHUNK_MEM:
+		i40iw_destroy_pble_pool(dev, iwdev->pble_rsrc);
+		/* fallthrough */
 	case CEQ_CREATED:
 		i40iw_dele_ceqs(iwdev, reset);
 		/* fallthrough */
@@ -1489,9 +1492,6 @@ static void i40iw_deinit_device(struct i40iw_device *iwdev, bool reset)
 	case CCQ_CREATED:
 		i40iw_destroy_ccq(iwdev, reset);
 		/* fallthrough */
-	case PBLE_CHUNK_MEM:
-		i40iw_destroy_pble_pool(dev, iwdev->pble_rsrc);
-		/* fallthrough */
 	case HMC_OBJS_CREATED:
 		i40iw_del_hmc_objects(dev, dev->hmc_info, true, reset);
 		/* fallthrough */
@@ -1670,6 +1670,7 @@ static int i40iw_open(struct i40e_info *ldev, struct i40e_client *client)
 		status = i40iw_hmc_init_pble(&iwdev->sc_dev, iwdev->pble_rsrc);
 		if (status)
 			break;
+		iwdev->init_state = PBLE_CHUNK_MEM;
 		iwdev->virtchnl_wq = alloc_ordered_workqueue("iwvch", WQ_MEM_RECLAIM);
 		i40iw_register_notifiers();
 		iwdev->init_state = INET_NOTIFIER;
-- 
cgit v1.2.3


From 415920aa174666c0ac8c47eee974acc9f49efec4 Mon Sep 17 00:00:00 2001
From: Mustafa Ismail <mustafa.ismail@intel.com>
Date: Fri, 23 Jun 2017 16:03:56 -0500
Subject: i40iw: Do not poll CCQ after it is destroyed

Control Queue Pair (CQP) OPs, in this case - Update SDs,
cannot poll the Control Completion Queue (CCQ) after CCQ is
destroyed. Instead, poll via registers.

Signed-off-by: Mustafa Ismail <mustafa.ismail@intel.com>
Signed-off-by: Henry Orosco <henry.orosco@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
---
 drivers/infiniband/hw/i40iw/i40iw_ctrl.c | 2 ++
 1 file changed, 2 insertions(+)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/hw/i40iw/i40iw_ctrl.c b/drivers/infiniband/hw/i40iw/i40iw_ctrl.c
index a027e2072477..9ec1ae9a82c9 100644
--- a/drivers/infiniband/hw/i40iw/i40iw_ctrl.c
+++ b/drivers/infiniband/hw/i40iw/i40iw_ctrl.c
@@ -1970,6 +1970,8 @@ static enum i40iw_status_code i40iw_sc_ccq_destroy(struct i40iw_sc_cq *ccq,
 		ret_code = i40iw_cqp_poll_registers(cqp, tail, 1000);
 	}
 
+	cqp->process_cqp_sds = i40iw_update_sds_noccq;
+
 	return ret_code;
 }
 
-- 
cgit v1.2.3


From 6c1d94de4e75160d3ea5af3bf51d290341db1d44 Mon Sep 17 00:00:00 2001
From: Shiraz Saleem <shiraz.saleem@intel.com>
Date: Fri, 23 Jun 2017 16:03:57 -0500
Subject: i40iw: Utilize iwdev->reset during PCI function reset

Utilize iwdev->reset on a PCI function reset notification
instead of passing in reset flag for resource clean-up.

Signed-off-by: Shiraz Saleem <shiraz.saleem@intel.com>
Signed-off-by: Henry Orosco <henry.orosco@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
---
 drivers/infiniband/hw/i40iw/i40iw_main.c | 51 +++++++++++++++-----------------
 1 file changed, 24 insertions(+), 27 deletions(-)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/hw/i40iw/i40iw_main.c b/drivers/infiniband/hw/i40iw/i40iw_main.c
index 8fc61b3bd223..3bad7d967abe 100644
--- a/drivers/infiniband/hw/i40iw/i40iw_main.c
+++ b/drivers/infiniband/hw/i40iw/i40iw_main.c
@@ -274,13 +274,12 @@ static void i40iw_disable_irq(struct i40iw_sc_dev *dev,
 /**
  * i40iw_destroy_aeq - destroy aeq
  * @iwdev: iwarp device
- * @reset: true if called before reset
  *
  * Issue a destroy aeq request and
  * free the resources associated with the aeq
  * The function is called during driver unload
  */
-static void i40iw_destroy_aeq(struct i40iw_device *iwdev, bool reset)
+static void i40iw_destroy_aeq(struct i40iw_device *iwdev)
 {
 	enum i40iw_status_code status = I40IW_ERR_NOT_READY;
 	struct i40iw_sc_dev *dev = &iwdev->sc_dev;
@@ -288,7 +287,7 @@ static void i40iw_destroy_aeq(struct i40iw_device *iwdev, bool reset)
 
 	if (!iwdev->msix_shared)
 		i40iw_disable_irq(dev, iwdev->iw_msixtbl, (void *)iwdev);
-	if (reset)
+	if (iwdev->reset)
 		goto exit;
 
 	if (!dev->aeq_ops->aeq_destroy(&aeq->sc_aeq, 0, 1))
@@ -304,19 +303,17 @@ exit:
  * i40iw_destroy_ceq - destroy ceq
  * @iwdev: iwarp device
  * @iwceq: ceq to be destroyed
- * @reset: true if called before reset
  *
  * Issue a destroy ceq request and
  * free the resources associated with the ceq
  */
 static void i40iw_destroy_ceq(struct i40iw_device *iwdev,
-			      struct i40iw_ceq *iwceq,
-			      bool reset)
+			      struct i40iw_ceq *iwceq)
 {
 	enum i40iw_status_code status;
 	struct i40iw_sc_dev *dev = &iwdev->sc_dev;
 
-	if (reset)
+	if (iwdev->reset)
 		goto exit;
 
 	status = dev->ceq_ops->ceq_destroy(&iwceq->sc_ceq, 0, 1);
@@ -335,12 +332,11 @@ exit:
 /**
  * i40iw_dele_ceqs - destroy all ceq's
  * @iwdev: iwarp device
- * @reset: true if called before reset
  *
  * Go through all of the device ceq's and for each ceq
  * disable the ceq interrupt and destroy the ceq
  */
-static void i40iw_dele_ceqs(struct i40iw_device *iwdev, bool reset)
+static void i40iw_dele_ceqs(struct i40iw_device *iwdev)
 {
 	u32 i = 0;
 	struct i40iw_sc_dev *dev = &iwdev->sc_dev;
@@ -349,32 +345,31 @@ static void i40iw_dele_ceqs(struct i40iw_device *iwdev, bool reset)
 
 	if (iwdev->msix_shared) {
 		i40iw_disable_irq(dev, msix_vec, (void *)iwdev);
-		i40iw_destroy_ceq(iwdev, iwceq, reset);
+		i40iw_destroy_ceq(iwdev, iwceq);
 		iwceq++;
 		i++;
 	}
 
 	for (msix_vec++; i < iwdev->ceqs_count; i++, msix_vec++, iwceq++) {
 		i40iw_disable_irq(dev, msix_vec, (void *)iwceq);
-		i40iw_destroy_ceq(iwdev, iwceq, reset);
+		i40iw_destroy_ceq(iwdev, iwceq);
 	}
 }
 
 /**
  * i40iw_destroy_ccq - destroy control cq
  * @iwdev: iwarp device
- * @reset: true if called before reset
  *
  * Issue destroy ccq request and
  * free the resources associated with the ccq
  */
-static void i40iw_destroy_ccq(struct i40iw_device *iwdev, bool reset)
+static void i40iw_destroy_ccq(struct i40iw_device *iwdev)
 {
 	struct i40iw_sc_dev *dev = &iwdev->sc_dev;
 	struct i40iw_ccq *ccq = &iwdev->ccq;
 	enum i40iw_status_code status = 0;
 
-	if (!reset)
+	if (!iwdev->reset)
 		status = dev->ccq_ops->ccq_destroy(dev->ccq, 0, true);
 	if (status)
 		i40iw_pr_err("ccq destroy failed %d\n", status);
@@ -810,7 +805,7 @@ static enum i40iw_status_code i40iw_setup_ceqs(struct i40iw_device *iwdev,
 		iwceq->msix_idx = msix_vec->idx;
 		status = i40iw_configure_ceq_vector(iwdev, iwceq, ceq_id, msix_vec);
 		if (status) {
-			i40iw_destroy_ceq(iwdev, iwceq, false);
+			i40iw_destroy_ceq(iwdev, iwceq);
 			break;
 		}
 		i40iw_enable_intr(&iwdev->sc_dev, msix_vec->idx);
@@ -912,7 +907,7 @@ static enum i40iw_status_code i40iw_setup_aeq(struct i40iw_device *iwdev)
 
 	status = i40iw_configure_aeq_vector(iwdev);
 	if (status) {
-		i40iw_destroy_aeq(iwdev, false);
+		i40iw_destroy_aeq(iwdev);
 		return status;
 	}
 
@@ -1442,12 +1437,11 @@ static enum i40iw_status_code i40iw_save_msix_info(struct i40iw_device *iwdev,
 /**
  * i40iw_deinit_device - clean up the device resources
  * @iwdev: iwarp device
- * @reset: true if called before reset
  *
  * Destroy the ib device interface, remove the mac ip entry and ipv4/ipv6 addresses,
  * destroy the device queues and free the pble and the hmc objects
  */
-static void i40iw_deinit_device(struct i40iw_device *iwdev, bool reset)
+static void i40iw_deinit_device(struct i40iw_device *iwdev)
 {
 	struct i40e_info *ldev = iwdev->ldev;
 
@@ -1464,7 +1458,7 @@ static void i40iw_deinit_device(struct i40iw_device *iwdev, bool reset)
 		i40iw_destroy_rdma_device(iwdev->iwibdev);
 		/* fallthrough */
 	case IP_ADDR_REGISTERED:
-		if (!reset)
+		if (!iwdev->reset)
 			i40iw_del_macip_entry(iwdev, (u8)iwdev->mac_ip_table_idx);
 		/* fallthrough */
 	case INET_NOTIFIER:
@@ -1478,22 +1472,22 @@ static void i40iw_deinit_device(struct i40iw_device *iwdev, bool reset)
 		i40iw_destroy_pble_pool(dev, iwdev->pble_rsrc);
 		/* fallthrough */
 	case CEQ_CREATED:
-		i40iw_dele_ceqs(iwdev, reset);
+		i40iw_dele_ceqs(iwdev);
 		/* fallthrough */
 	case AEQ_CREATED:
-		i40iw_destroy_aeq(iwdev, reset);
+		i40iw_destroy_aeq(iwdev);
 		/* fallthrough */
 	case IEQ_CREATED:
-		i40iw_puda_dele_resources(&iwdev->vsi, I40IW_PUDA_RSRC_TYPE_IEQ, reset);
+		i40iw_puda_dele_resources(&iwdev->vsi, I40IW_PUDA_RSRC_TYPE_IEQ, iwdev->reset);
 		/* fallthrough */
 	case ILQ_CREATED:
-		i40iw_puda_dele_resources(&iwdev->vsi, I40IW_PUDA_RSRC_TYPE_ILQ, reset);
+		i40iw_puda_dele_resources(&iwdev->vsi, I40IW_PUDA_RSRC_TYPE_ILQ, iwdev->reset);
 		/* fallthrough */
 	case CCQ_CREATED:
-		i40iw_destroy_ccq(iwdev, reset);
+		i40iw_destroy_ccq(iwdev);
 		/* fallthrough */
 	case HMC_OBJS_CREATED:
-		i40iw_del_hmc_objects(dev, dev->hmc_info, true, reset);
+		i40iw_del_hmc_objects(dev, dev->hmc_info, true, iwdev->reset);
 		/* fallthrough */
 	case CQP_CREATED:
 		i40iw_destroy_cqp(iwdev, true);
@@ -1694,7 +1688,7 @@ static int i40iw_open(struct i40e_info *ldev, struct i40e_client *client)
 	} while (0);
 
 	i40iw_pr_err("status = %d last completion = %d\n", status, iwdev->init_state);
-	i40iw_deinit_device(iwdev, false);
+	i40iw_deinit_device(iwdev);
 	return -ERESTART;
 }
 
@@ -1775,9 +1769,12 @@ static void i40iw_close(struct i40e_info *ldev, struct i40e_client *client, bool
 	iwdev = &hdl->device;
 	iwdev->closing = true;
 
+	if (reset)
+		iwdev->reset = true;
+
 	i40iw_cm_disconnect_all(iwdev);
 	destroy_workqueue(iwdev->virtchnl_wq);
-	i40iw_deinit_device(iwdev, reset);
+	i40iw_deinit_device(iwdev);
 }
 
 /**
-- 
cgit v1.2.3


From 6327cb09dfda103f7255ef218ac18697b293554a Mon Sep 17 00:00:00 2001
From: Shiraz Saleem <shiraz.saleem@intel.com>
Date: Fri, 23 Jun 2017 16:03:58 -0500
Subject: i40iw: Release cm_id ref on PCI function reset

On PCI function reset, cm_id reference is not released
which causes an application hang, as it waits on the
cm_id to be released on rdma_destroy.

To fix this, call i40iw_cm_disconn during a PCI function
reset to clean-up resources and release cm_id reference.

Signed-off-by: Shiraz Saleem <shiraz.saleem@intel.com>
Signed-off-by: Henry Orosco <henry.orosco@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
---
 drivers/infiniband/hw/i40iw/i40iw_cm.c | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/hw/i40iw/i40iw_cm.c b/drivers/infiniband/hw/i40iw/i40iw_cm.c
index 6ae98aa7f74e..5a2fa743676c 100644
--- a/drivers/infiniband/hw/i40iw/i40iw_cm.c
+++ b/drivers/infiniband/hw/i40iw/i40iw_cm.c
@@ -3487,7 +3487,8 @@ static void i40iw_cm_disconn_true(struct i40iw_qp *iwqp)
 	if (((original_hw_tcp_state == I40IW_TCP_STATE_CLOSED) ||
 	     (original_hw_tcp_state == I40IW_TCP_STATE_TIME_WAIT) ||
 	     (last_ae == I40IW_AE_RDMAP_ROE_BAD_LLP_CLOSE) ||
-	     (last_ae == I40IW_AE_LLP_CONNECTION_RESET))) {
+	     (last_ae == I40IW_AE_LLP_CONNECTION_RESET) ||
+	      iwdev->reset)) {
 		issue_close = 1;
 		iwqp->cm_id = NULL;
 		if (!iwqp->flush_issued) {
@@ -4265,6 +4266,8 @@ void i40iw_cm_disconnect_all(struct i40iw_device *iwdev)
 		cm_node = container_of(list_node, struct i40iw_cm_node, connected_entry);
 		attr.qp_state = IB_QPS_ERR;
 		i40iw_modify_qp(&cm_node->iwqp->ibqp, &attr, IB_QP_STATE, NULL);
+		if (iwdev->reset)
+			i40iw_cm_disconn(cm_node->iwqp);
 		i40iw_rem_ref_cm_node(cm_node);
 	}
 }
-- 
cgit v1.2.3


From b5e452a04a10f12763f9836d3d3999f3bb1e56fb Mon Sep 17 00:00:00 2001
From: Shiraz Saleem <shiraz.saleem@intel.com>
Date: Fri, 23 Jun 2017 16:03:59 -0500
Subject: i40iw: Free QP resources on CQP destroy QP failure

Current flow leaves software QP structures in memory if
Control Queue Pair (CQP) destroy QP OP fails. To fix this,
free QP resources on fail of CQP destroy QP OP.

Signed-off-by: Shiraz Saleem <shiraz.saleem@intel.com>
Signed-off-by: Henry Orosco <henry.orosco@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
---
 drivers/infiniband/hw/i40iw/i40iw_utils.c | 8 ++++++--
 1 file changed, 6 insertions(+), 2 deletions(-)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/hw/i40iw/i40iw_utils.c b/drivers/infiniband/hw/i40iw/i40iw_utils.c
index 56d986924a4c..ded8e48ed3c8 100644
--- a/drivers/infiniband/hw/i40iw/i40iw_utils.c
+++ b/drivers/infiniband/hw/i40iw/i40iw_utils.c
@@ -546,8 +546,12 @@ void i40iw_rem_ref(struct ib_qp *ibqp)
 	cqp_info->in.u.qp_destroy.scratch = (uintptr_t)cqp_request;
 	cqp_info->in.u.qp_destroy.remove_hash_idx = true;
 	status = i40iw_handle_cqp_op(iwdev, cqp_request);
-	if (status)
-		i40iw_pr_err("CQP-OP Destroy QP fail");
+	if (!status)
+		return;
+
+	i40iw_rem_pdusecount(iwqp->iwpd, iwdev);
+	i40iw_free_qp_resources(iwdev, iwqp, qp_num);
+	i40iw_rem_devusecount(iwdev);
 }
 
 /**
-- 
cgit v1.2.3


From c5c9d27e6c79ab3ab36092fe67fb7f2c6a120171 Mon Sep 17 00:00:00 2001
From: Henry Orosco <henry.orosco@intel.com>
Date: Fri, 23 Jun 2017 16:04:00 -0500
Subject: i40iw: Add missing memory barrier

Add missing write memory barrier before writing the
header containing valid bit to the WQE in i40iw_puda_send.

Signed-off-by: Henry Orosco <henry.orosco@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
---
 drivers/infiniband/hw/i40iw/i40iw_puda.c | 3 +++
 1 file changed, 3 insertions(+)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/hw/i40iw/i40iw_puda.c b/drivers/infiniband/hw/i40iw/i40iw_puda.c
index db41ab40da9c..1bb16814560f 100644
--- a/drivers/infiniband/hw/i40iw/i40iw_puda.c
+++ b/drivers/infiniband/hw/i40iw/i40iw_puda.c
@@ -408,6 +408,9 @@ enum i40iw_status_code i40iw_puda_send(struct i40iw_sc_qp *qp,
 	set_64bit_val(wqe, 0, info->paddr);
 	set_64bit_val(wqe, 8, LS_64(info->len, I40IWQPSQ_FRAG_LEN));
 	set_64bit_val(wqe, 16, header[0]);
+
+	/* Ensure all data is written before writing valid bit */
+	wmb();
 	set_64bit_val(wqe, 24, header[1]);
 
 	i40iw_debug_buf(qp->dev, I40IW_DEBUG_PUDA, "PUDA SEND WQE", wqe, 32);
-- 
cgit v1.2.3


From c709d7f229a273c7c5664e9dfe5432b031842d0c Mon Sep 17 00:00:00 2001
From: Henry Orosco <henry.orosco@intel.com>
Date: Fri, 23 Jun 2017 16:04:01 -0500
Subject: i40iw: Update list correctly

To avoid infinite loop, in i40iw_ieq_handle_exception, update
plist inside while loop.

Signed-off-by: Henry Orosco <henry.orosco@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
---
 drivers/infiniband/hw/i40iw/i40iw_puda.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/hw/i40iw/i40iw_puda.c b/drivers/infiniband/hw/i40iw/i40iw_puda.c
index 1bb16814560f..71050c5d29a0 100644
--- a/drivers/infiniband/hw/i40iw/i40iw_puda.c
+++ b/drivers/infiniband/hw/i40iw/i40iw_puda.c
@@ -1414,10 +1414,10 @@ static void i40iw_ieq_handle_exception(struct i40iw_puda_rsrc *ieq,
 
 	if (!list_empty(rxlist)) {
 		tmpbuf = (struct i40iw_puda_buf *)rxlist->next;
-		plist = &tmpbuf->list;
 		while ((struct list_head *)tmpbuf != rxlist) {
 			if ((int)(buf->seqnum - tmpbuf->seqnum) < 0)
 				break;
+			plist = &tmpbuf->list;
 			tmpbuf = (struct i40iw_puda_buf *)plist->next;
 		}
 		/* Insert buf before tmpbuf */
-- 
cgit v1.2.3


From 44b99f88cdd5b47046c511aa64ae71ad2c9e5b1e Mon Sep 17 00:00:00 2001
From: Shiraz Saleem <shiraz.saleem@intel.com>
Date: Fri, 23 Jun 2017 16:04:02 -0500
Subject: i40iw: Avoid memory leak of CQP request objects

Control Queue Pair (CQP) request objects, which have
not received a completion upon interface close, remain
in memory.

To fix this, identify and free all pending CQP request
objects during destroy CQP OP.

Signed-off-by: Shiraz Saleem <shiraz.saleem@intel.com>
Signed-off-by: Henry Orosco <henry.orosco@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
---
 drivers/infiniband/hw/i40iw/i40iw.h       |  1 +
 drivers/infiniband/hw/i40iw/i40iw_main.c  |  2 ++
 drivers/infiniband/hw/i40iw/i40iw_utils.c | 52 +++++++++++++++++++++++++++++++
 3 files changed, 55 insertions(+)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/hw/i40iw/i40iw.h b/drivers/infiniband/hw/i40iw/i40iw.h
index da2eb5a281fa..9b1566468744 100644
--- a/drivers/infiniband/hw/i40iw/i40iw.h
+++ b/drivers/infiniband/hw/i40iw/i40iw.h
@@ -527,6 +527,7 @@ enum i40iw_status_code i40iw_add_mac_addr(struct i40iw_device *iwdev,
 int i40iw_modify_qp(struct ib_qp *, struct ib_qp_attr *, int, struct ib_udata *);
 void i40iw_cq_wq_destroy(struct i40iw_device *iwdev, struct i40iw_sc_cq *cq);
 
+void i40iw_cleanup_pending_cqp_op(struct i40iw_device *iwdev);
 void i40iw_rem_pdusecount(struct i40iw_pd *iwpd, struct i40iw_device *iwdev);
 void i40iw_add_pdusecount(struct i40iw_pd *iwpd);
 void i40iw_rem_devusecount(struct i40iw_device *iwdev);
diff --git a/drivers/infiniband/hw/i40iw/i40iw_main.c b/drivers/infiniband/hw/i40iw/i40iw_main.c
index 3bad7d967abe..ae8463ff59a7 100644
--- a/drivers/infiniband/hw/i40iw/i40iw_main.c
+++ b/drivers/infiniband/hw/i40iw/i40iw_main.c
@@ -243,6 +243,8 @@ static void i40iw_destroy_cqp(struct i40iw_device *iwdev, bool free_hwcqp)
 	if (free_hwcqp)
 		dev->cqp_ops->cqp_destroy(dev->cqp);
 
+	i40iw_cleanup_pending_cqp_op(iwdev);
+
 	i40iw_free_dma_mem(dev->hw, &cqp->sq);
 	kfree(cqp->scratch_array);
 	iwdev->cqp.scratch_array = NULL;
diff --git a/drivers/infiniband/hw/i40iw/i40iw_utils.c b/drivers/infiniband/hw/i40iw/i40iw_utils.c
index ded8e48ed3c8..e311ec559f4e 100644
--- a/drivers/infiniband/hw/i40iw/i40iw_utils.c
+++ b/drivers/infiniband/hw/i40iw/i40iw_utils.c
@@ -337,6 +337,7 @@ struct i40iw_cqp_request *i40iw_get_cqp_request(struct i40iw_cqp *cqp, bool wait
  */
 void i40iw_free_cqp_request(struct i40iw_cqp *cqp, struct i40iw_cqp_request *cqp_request)
 {
+	struct i40iw_device *iwdev = container_of(cqp, struct i40iw_device, cqp);
 	unsigned long flags;
 
 	if (cqp_request->dynamic) {
@@ -350,6 +351,7 @@ void i40iw_free_cqp_request(struct i40iw_cqp *cqp, struct i40iw_cqp_request *cqp
 		list_add_tail(&cqp_request->list, &cqp->cqp_avail_reqs);
 		spin_unlock_irqrestore(&cqp->req_lock, flags);
 	}
+	wake_up(&iwdev->close_wq);
 }
 
 /**
@@ -364,6 +366,56 @@ void i40iw_put_cqp_request(struct i40iw_cqp *cqp,
 		i40iw_free_cqp_request(cqp, cqp_request);
 }
 
+/**
+ * i40iw_free_pending_cqp_request -free pending cqp request objs
+ * @cqp: cqp ptr
+ * @cqp_request: to be put back in cqp list
+ */
+static void i40iw_free_pending_cqp_request(struct i40iw_cqp *cqp,
+					   struct i40iw_cqp_request *cqp_request)
+{
+	struct i40iw_device *iwdev = container_of(cqp, struct i40iw_device, cqp);
+
+	if (cqp_request->waiting) {
+		cqp_request->compl_info.error = true;
+		cqp_request->request_done = true;
+		wake_up(&cqp_request->waitq);
+	}
+	i40iw_put_cqp_request(cqp, cqp_request);
+	wait_event_timeout(iwdev->close_wq,
+			   !atomic_read(&cqp_request->refcount),
+			   1000);
+}
+
+/**
+ * i40iw_cleanup_pending_cqp_op - clean-up cqp with no completions
+ * @iwdev: iwarp device
+ */
+void i40iw_cleanup_pending_cqp_op(struct i40iw_device *iwdev)
+{
+	struct i40iw_sc_dev *dev = &iwdev->sc_dev;
+	struct i40iw_cqp *cqp = &iwdev->cqp;
+	struct i40iw_cqp_request *cqp_request = NULL;
+	struct cqp_commands_info *pcmdinfo = NULL;
+	u32 i, pending_work, wqe_idx;
+
+	pending_work = I40IW_RING_WORK_AVAILABLE(cqp->sc_cqp.sq_ring);
+	wqe_idx = I40IW_RING_GETCURRENT_TAIL(cqp->sc_cqp.sq_ring);
+	for (i = 0; i < pending_work; i++) {
+		cqp_request = (struct i40iw_cqp_request *)(unsigned long)cqp->scratch_array[wqe_idx];
+		if (cqp_request)
+			i40iw_free_pending_cqp_request(cqp, cqp_request);
+		wqe_idx = (wqe_idx + 1) % I40IW_RING_GETSIZE(cqp->sc_cqp.sq_ring);
+	}
+
+	while (!list_empty(&dev->cqp_cmd_head)) {
+		pcmdinfo = (struct cqp_commands_info *)i40iw_remove_head(&dev->cqp_cmd_head);
+		cqp_request = container_of(pcmdinfo, struct i40iw_cqp_request, info);
+		if (cqp_request)
+			i40iw_free_pending_cqp_request(cqp, cqp_request);
+	}
+}
+
 /**
  * i40iw_free_qp - callback after destroy cqp completes
  * @cqp_request: cqp request for destroy qp
-- 
cgit v1.2.3


From af56e53ccd29bda062a1ae75276dc9c0f8eedf47 Mon Sep 17 00:00:00 2001
From: Tatyana Nikolova <Tatyana.E.Nikolova@intel.com>
Date: Wed, 5 Jul 2017 21:25:33 -0500
Subject: i40iw: Free QP PBLEs when the QP is destroyed

If the physical buffer list entries (PBLEs) of a QP are freed
up at i40iw_dereg_mr, they can be assigned to a newly
created QP before the previous QP is destroyed. Fix this
by freeing PBLEs only when the QP is destroyed.

Signed-off-by: Tatyana Nikolova <tatyana.e.nikolova@intel.com>
Signed-off-by: Faisal Latif <faisal.latif@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
---
 drivers/infiniband/hw/i40iw/i40iw_verbs.c | 15 +++++++++++----
 drivers/infiniband/hw/i40iw/i40iw_verbs.h |  2 +-
 2 files changed, 12 insertions(+), 5 deletions(-)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/hw/i40iw/i40iw_verbs.c b/drivers/infiniband/hw/i40iw/i40iw_verbs.c
index 4dbe61ec7a77..4aa0264ccb3a 100644
--- a/drivers/infiniband/hw/i40iw/i40iw_verbs.c
+++ b/drivers/infiniband/hw/i40iw/i40iw_verbs.c
@@ -426,9 +426,13 @@ void i40iw_free_qp_resources(struct i40iw_device *iwdev,
 			     struct i40iw_qp *iwqp,
 			     u32 qp_num)
 {
+	struct i40iw_pbl *iwpbl = &iwqp->iwpbl;
+
 	i40iw_dealloc_push_page(iwdev, &iwqp->sc_qp);
 	if (qp_num)
 		i40iw_free_resource(iwdev, iwdev->allocated_qps, qp_num);
+	if (iwpbl->pbl_allocated)
+		i40iw_free_pble(iwdev->pble_rsrc, &iwpbl->pble_alloc);
 	i40iw_free_dma_mem(iwdev->sc_dev.hw, &iwqp->q2_ctx_mem);
 	i40iw_free_dma_mem(iwdev->sc_dev.hw, &iwqp->kqp.dma_mem);
 	kfree(iwqp->kqp.wrid_mem);
@@ -483,7 +487,7 @@ static int i40iw_setup_virt_qp(struct i40iw_device *iwdev,
 			       struct i40iw_qp *iwqp,
 			       struct i40iw_qp_init_info *init_info)
 {
-	struct i40iw_pbl *iwpbl = iwqp->iwpbl;
+	struct i40iw_pbl *iwpbl = &iwqp->iwpbl;
 	struct i40iw_qp_mr *qpmr = &iwpbl->qp_mr;
 
 	iwqp->page = qpmr->sq_page;
@@ -688,19 +692,22 @@ static struct ib_qp *i40iw_create_qp(struct ib_pd *ibpd,
 			ucontext = to_ucontext(ibpd->uobject->context);
 
 			if (req.user_wqe_buffers) {
+				struct i40iw_pbl *iwpbl;
+
 				spin_lock_irqsave(
 				    &ucontext->qp_reg_mem_list_lock, flags);
-				iwqp->iwpbl = i40iw_get_pbl(
+				iwpbl = i40iw_get_pbl(
 				    (unsigned long)req.user_wqe_buffers,
 				    &ucontext->qp_reg_mem_list);
 				spin_unlock_irqrestore(
 				    &ucontext->qp_reg_mem_list_lock, flags);
 
-				if (!iwqp->iwpbl) {
+				if (!iwpbl) {
 					err_code = -ENODATA;
 					i40iw_pr_err("no pbl info\n");
 					goto error;
 				}
+				memcpy(&iwqp->iwpbl, iwpbl, sizeof(iwqp->iwpbl));
 			}
 		}
 		err_code = i40iw_setup_virt_qp(iwdev, iwqp, &init_info);
@@ -2063,7 +2070,7 @@ static int i40iw_dereg_mr(struct ib_mr *ib_mr)
 			ucontext = to_ucontext(ibpd->uobject->context);
 			i40iw_del_memlist(iwmr, ucontext);
 		}
-		if (iwpbl->pbl_allocated)
+		if (iwpbl->pbl_allocated && iwmr->type != IW_MEMREG_TYPE_QP)
 			i40iw_free_pble(iwdev->pble_rsrc, palloc);
 		kfree(iwmr);
 		return 0;
diff --git a/drivers/infiniband/hw/i40iw/i40iw_verbs.h b/drivers/infiniband/hw/i40iw/i40iw_verbs.h
index 07c3fec77de6..9067443cd311 100644
--- a/drivers/infiniband/hw/i40iw/i40iw_verbs.h
+++ b/drivers/infiniband/hw/i40iw/i40iw_verbs.h
@@ -170,7 +170,7 @@ struct i40iw_qp {
 	struct i40iw_qp_kmode kqp;
 	struct i40iw_dma_mem host_ctx;
 	struct timer_list terminate_timer;
-	struct i40iw_pbl *iwpbl;
+	struct i40iw_pbl iwpbl;
 	struct i40iw_dma_mem q2_ctx_mem;
 	struct i40iw_dma_mem ietf_mem;
 	struct completion sq_drained;
-- 
cgit v1.2.3


From 653f0a71daf1a71d55d3af368c28c8114c11c607 Mon Sep 17 00:00:00 2001
From: Dan Carpenter <dan.carpenter@oracle.com>
Date: Mon, 10 Jul 2017 10:22:47 +0300
Subject: RDMA/bnxt_re: checking for NULL instead of IS_ERR()

bnxt_re_alloc_mw() doesn't return NULL, it returns error pointers.

Fixes: 9152e0b722b2 ("RDMA/bnxt_re: HW workarounds for handling specific conditions")
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Acked-by: Selvin Xavier <selvin.xavier@broadcom.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
---
 drivers/infiniband/hw/bnxt_re/ib_verbs.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/hw/bnxt_re/ib_verbs.c b/drivers/infiniband/hw/bnxt_re/ib_verbs.c
index c7bd68311d0c..e794b0fa4ae6 100644
--- a/drivers/infiniband/hw/bnxt_re/ib_verbs.c
+++ b/drivers/infiniband/hw/bnxt_re/ib_verbs.c
@@ -588,10 +588,10 @@ static int bnxt_re_create_fence_mr(struct bnxt_re_pd *pd)
 
 	/* Create a fence MW only for kernel consumers */
 	mw = bnxt_re_alloc_mw(&pd->ib_pd, IB_MW_TYPE_1, NULL);
-	if (!mw) {
+	if (IS_ERR(mw)) {
 		dev_err(rdev_to_dev(rdev),
 			"Failed to create fence-MW for PD: %p\n", pd);
-		rc = -EINVAL;
+		rc = PTR_ERR(mw);
 		goto fail;
 	}
 	fence->mw = mw;
-- 
cgit v1.2.3


From 5c8857b653e71a9850a02837e1268e3198abbd1a Mon Sep 17 00:00:00 2001
From: Dan Carpenter <dan.carpenter@oracle.com>
Date: Thu, 13 Jul 2017 10:45:48 +0300
Subject: IB/IPoIB: Fix error code in ipoib_add_port()

We accidentally don't see the error code on some of these error paths.
It means we return ERR_PTR(0) which is NULL and it results in a NULL
dereference in the caller.

This bug dates to pre-git days.

Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Reviewed-by: Leon Romanovsky <leonro@mellanox.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
---
 drivers/infiniband/ulp/ipoib/ipoib_main.c | 1 +
 1 file changed, 1 insertion(+)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/ulp/ipoib/ipoib_main.c b/drivers/infiniband/ulp/ipoib/ipoib_main.c
index 70dacaf9044e..4ce315c92b48 100644
--- a/drivers/infiniband/ulp/ipoib/ipoib_main.c
+++ b/drivers/infiniband/ulp/ipoib/ipoib_main.c
@@ -2239,6 +2239,7 @@ static struct net_device *ipoib_add_port(const char *format,
 		goto register_failed;
 	}
 
+	result = -ENOMEM;
 	if (ipoib_cm_add_mode_attr(priv->dev))
 		goto sysfs_failed;
 	if (ipoib_add_pkey_attr(priv->dev))
-- 
cgit v1.2.3


From 6031e079aa4656743298ea235b894ee883f45c71 Mon Sep 17 00:00:00 2001
From: Dan Carpenter <dan.carpenter@oracle.com>
Date: Thu, 13 Jul 2017 10:47:22 +0300
Subject: IB/i40iw: Fix error code in i40iw_create_cq()

We accidentally forgot to set the error code if ib_copy_from_udata()
fails.  It means we return ERR_PTR(0) which is NULL and results in a
NULL dereference in the callers.

Fixes: d37498417947 ("i40iw: add files for iwarp interface")
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Acked-by: Shiraz Saleem <shiraz.saleem@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
---
 drivers/infiniband/hw/i40iw/i40iw_verbs.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/hw/i40iw/i40iw_verbs.c b/drivers/infiniband/hw/i40iw/i40iw_verbs.c
index 4aa0264ccb3a..02d871db7ca5 100644
--- a/drivers/infiniband/hw/i40iw/i40iw_verbs.c
+++ b/drivers/infiniband/hw/i40iw/i40iw_verbs.c
@@ -1168,8 +1168,10 @@ static struct ib_cq *i40iw_create_cq(struct ib_device *ibdev,
 		memset(&req, 0, sizeof(req));
 		iwcq->user_mode = true;
 		ucontext = to_ucontext(context);
-		if (ib_copy_from_udata(&req, udata, sizeof(struct i40iw_create_cq_req)))
+		if (ib_copy_from_udata(&req, udata, sizeof(struct i40iw_create_cq_req))) {
+			err_code = -EFAULT;
 			goto cq_free_resources;
+		}
 
 		spin_lock_irqsave(&ucontext->cq_reg_mem_list_lock, flags);
 		iwpbl = i40iw_get_pbl((unsigned long)req.user_cq_buffer,
-- 
cgit v1.2.3


From 6ebedacbb44602d4dec3348dee5ec31dd9b09521 Mon Sep 17 00:00:00 2001
From: Dan Carpenter <dan.carpenter@oracle.com>
Date: Thu, 13 Jul 2017 10:47:40 +0300
Subject: cxgb4: Fix error codes in c4iw_create_cq()

If one of these kmalloc() calls fails then we return ERR_PTR(0) which is
NULL.  It results in a NULL dereference in the callers.

Fixes: cfdda9d76436 ("RDMA/cxgb4: Add driver for Chelsio T4 RNIC")
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Acked-by: Steve Wise <swise@opengridcomputing.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
---
 drivers/infiniband/hw/cxgb4/cq.c | 1 +
 1 file changed, 1 insertion(+)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/hw/cxgb4/cq.c b/drivers/infiniband/hw/cxgb4/cq.c
index e16fcaf6b5a3..be07da1997e6 100644
--- a/drivers/infiniband/hw/cxgb4/cq.c
+++ b/drivers/infiniband/hw/cxgb4/cq.c
@@ -963,6 +963,7 @@ struct ib_cq *c4iw_create_cq(struct ib_device *ibdev,
 		goto err3;
 
 	if (ucontext) {
+		ret = -ENOMEM;
 		mm = kmalloc(sizeof *mm, GFP_KERNEL);
 		if (!mm)
 			goto err4;
-- 
cgit v1.2.3


From 9064d6055c14f700aa13f7c72fd3e63d12bee643 Mon Sep 17 00:00:00 2001
From: Dan Carpenter <dan.carpenter@oracle.com>
Date: Thu, 13 Jul 2017 10:48:00 +0300
Subject: IB/cxgb3: Fix error codes in iwch_alloc_mr()

We accidentally don't set the error code on some error paths.  It means
return ERR_PTR(0) which is NULL and results in a NULL dereference in the
caller.

Fixes: 13a239330abd ("RDMA/cxgb3: Don't ignore insert_handle() failures")
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Acked-by: Steve Wise <swise@opengridcomputing.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
---
 drivers/infiniband/hw/cxgb3/iwch_provider.c | 9 ++++-----
 1 file changed, 4 insertions(+), 5 deletions(-)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/hw/cxgb3/iwch_provider.c b/drivers/infiniband/hw/cxgb3/iwch_provider.c
index 29d30744d6c9..0cd0c1fa27d4 100644
--- a/drivers/infiniband/hw/cxgb3/iwch_provider.c
+++ b/drivers/infiniband/hw/cxgb3/iwch_provider.c
@@ -718,7 +718,7 @@ static struct ib_mr *iwch_alloc_mr(struct ib_pd *pd,
 	struct iwch_mr *mhp;
 	u32 mmid;
 	u32 stag = 0;
-	int ret = 0;
+	int ret = -ENOMEM;
 
 	if (mr_type != IB_MR_TYPE_MEM_REG ||
 	    max_num_sg > T3_MAX_FASTREG_DEPTH)
@@ -731,10 +731,8 @@ static struct ib_mr *iwch_alloc_mr(struct ib_pd *pd,
 		goto err;
 
 	mhp->pages = kcalloc(max_num_sg, sizeof(u64), GFP_KERNEL);
-	if (!mhp->pages) {
-		ret = -ENOMEM;
+	if (!mhp->pages)
 		goto pl_err;
-	}
 
 	mhp->rhp = rhp;
 	ret = iwch_alloc_pbl(mhp, max_num_sg);
@@ -751,7 +749,8 @@ static struct ib_mr *iwch_alloc_mr(struct ib_pd *pd,
 	mhp->attr.state = 1;
 	mmid = (stag) >> 8;
 	mhp->ibmr.rkey = mhp->ibmr.lkey = stag;
-	if (insert_handle(rhp, &rhp->mmidr, mhp, mmid))
+	ret = insert_handle(rhp, &rhp->mmidr, mhp, mmid);
+	if (ret)
 		goto err3;
 
 	pr_debug("%s mmid 0x%x mhp %p stag 0x%x\n", __func__, mmid, mhp, stag);
-- 
cgit v1.2.3


From dd75cfa6d3216c79c695f5af13e52208afe374ad Mon Sep 17 00:00:00 2001
From: Dan Carpenter <dan.carpenter@oracle.com>
Date: Thu, 13 Jul 2017 10:46:14 +0300
Subject: RDMA/ocrdma: Fix an error code in ocrdma_alloc_pd()

We should preserve the original "status" error code instead of resetting
it to zero.  Returning ERR_PTR(0) is the same as NULL and results in a
NULL dereference in the callers.  I added a printk() on error instead.

Fixes: 45e86b33ec8b ("RDMA/ocrdma: Cache recv DB until QP moved to RTR")
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
---
 drivers/infiniband/hw/ocrdma/ocrdma_verbs.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/hw/ocrdma/ocrdma_verbs.c b/drivers/infiniband/hw/ocrdma/ocrdma_verbs.c
index 2f30bda8457a..cc317e858040 100644
--- a/drivers/infiniband/hw/ocrdma/ocrdma_verbs.c
+++ b/drivers/infiniband/hw/ocrdma/ocrdma_verbs.c
@@ -744,7 +744,8 @@ err:
 	if (is_uctx_pd) {
 		ocrdma_release_ucontext_pd(uctx);
 	} else {
-		status = _ocrdma_dealloc_pd(dev, pd);
+		if (_ocrdma_dealloc_pd(dev, pd))
+			pr_err("%s: _ocrdma_dealloc_pd() failed\n", __func__);
 	}
 exit:
 	return ERR_PTR(status);
-- 
cgit v1.2.3


From f0c6e88288d65c93bbc7da4fb6f7d51b2733228a Mon Sep 17 00:00:00 2001
From: Dan Carpenter <dan.carpenter@oracle.com>
Date: Thu, 13 Jul 2017 10:46:49 +0300
Subject: RDMA/ocrdma: Fix error codes in ocrdma_create_srq()

If either of these allocations fail then we return ERR_PTR(0).  That's
equivalent to NULL and results in a NULL pointer dereference in the
caller.

Fixes: fe2caefcdf58 ("RDMA/ocrdma: Add driver for Emulex OneConnect IBoE RDMA adapter")
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
---
 drivers/infiniband/hw/ocrdma/ocrdma_verbs.c | 1 +
 1 file changed, 1 insertion(+)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/hw/ocrdma/ocrdma_verbs.c b/drivers/infiniband/hw/ocrdma/ocrdma_verbs.c
index cc317e858040..27d5e8d9f08d 100644
--- a/drivers/infiniband/hw/ocrdma/ocrdma_verbs.c
+++ b/drivers/infiniband/hw/ocrdma/ocrdma_verbs.c
@@ -1902,6 +1902,7 @@ struct ib_srq *ocrdma_create_srq(struct ib_pd *ibpd,
 		goto err;
 
 	if (udata == NULL) {
+		status = -ENOMEM;
 		srq->rqe_wr_id_tbl = kzalloc(sizeof(u64) * srq->rq.max_cnt,
 			    GFP_KERNEL);
 		if (srq->rqe_wr_id_tbl == NULL)
-- 
cgit v1.2.3


From 396551eb00e46aa8f843c448bced0c76971ec58c Mon Sep 17 00:00:00 2001
From: Dan Carpenter <dan.carpenter@oracle.com>
Date: Wed, 14 Jun 2017 13:20:09 +0300
Subject: IB/mlx5: Fix a warning message

"umem" is a valid pointer.  We intended to print "*umem" or even just
"err" instead.

Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Acked-by: Leon Romanovsky <leonro@mellanox.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
---
 drivers/infiniband/hw/mlx5/mr.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/hw/mlx5/mr.c b/drivers/infiniband/hw/mlx5/mr.c
index 8ab2f1360a45..2c40a2e989d2 100644
--- a/drivers/infiniband/hw/mlx5/mr.c
+++ b/drivers/infiniband/hw/mlx5/mr.c
@@ -835,7 +835,7 @@ static int mr_umem_get(struct ib_pd *pd, u64 start, u64 length,
 			    access_flags, 0);
 	err = PTR_ERR_OR_ZERO(*umem);
 	if (err < 0) {
-		mlx5_ib_err(dev, "umem get failed (%ld)\n", PTR_ERR(umem));
+		mlx5_ib_err(dev, "umem get failed (%d)\n", err);
 		return err;
 	}
 
-- 
cgit v1.2.3


From b3b2c7c5506bd8d15214136f8b11a2e6c9728033 Mon Sep 17 00:00:00 2001
From: Devesh Sharma <devesh.sharma@broadcom.com>
Date: Thu, 29 Jun 2017 12:28:08 -0700
Subject: RDMA/bnxt_re: Free doorbell page index (DPI) during dealloc ucontext

The driver must free the DPI during the dealloc_ucontext
instead of freeing it during dealloc_pd. However, the DPI
allocation scheme remains unchanged.

Signed-off-by: Devesh Sharma <devesh.sharma@broadcom.com>
Signed-off-by: Selvin Xavier <selvin.xavier@broadcom.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
---
 drivers/infiniband/hw/bnxt_re/ib_verbs.c | 58 ++++++++++++++++----------------
 drivers/infiniband/hw/bnxt_re/ib_verbs.h |  3 +-
 2 files changed, 30 insertions(+), 31 deletions(-)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/hw/bnxt_re/ib_verbs.c b/drivers/infiniband/hw/bnxt_re/ib_verbs.c
index e794b0fa4ae6..e743ffd392c6 100644
--- a/drivers/infiniband/hw/bnxt_re/ib_verbs.c
+++ b/drivers/infiniband/hw/bnxt_re/ib_verbs.c
@@ -612,30 +612,13 @@ int bnxt_re_dealloc_pd(struct ib_pd *ib_pd)
 	int rc;
 
 	bnxt_re_destroy_fence_mr(pd);
-	if (ib_pd->uobject && pd->dpi.dbr) {
-		struct ib_ucontext *ib_uctx = ib_pd->uobject->context;
-		struct bnxt_re_ucontext *ucntx;
 
-		/* Free DPI only if this is the first PD allocated by the
-		 * application and mark the context dpi as NULL
-		 */
-		ucntx = container_of(ib_uctx, struct bnxt_re_ucontext, ib_uctx);
-
-		rc = bnxt_qplib_dealloc_dpi(&rdev->qplib_res,
-					    &rdev->qplib_res.dpi_tbl,
-					    &pd->dpi);
+	if (pd->qplib_pd.id) {
+		rc = bnxt_qplib_dealloc_pd(&rdev->qplib_res,
+					   &rdev->qplib_res.pd_tbl,
+					   &pd->qplib_pd);
 		if (rc)
-			dev_err(rdev_to_dev(rdev), "Failed to deallocate HW DPI");
-			/* Don't fail, continue*/
-		ucntx->dpi = NULL;
-	}
-
-	rc = bnxt_qplib_dealloc_pd(&rdev->qplib_res,
-				   &rdev->qplib_res.pd_tbl,
-				   &pd->qplib_pd);
-	if (rc) {
-		dev_err(rdev_to_dev(rdev), "Failed to deallocate HW PD");
-		return rc;
+			dev_err(rdev_to_dev(rdev), "Failed to deallocate HW PD");
 	}
 
 	kfree(pd);
@@ -667,23 +650,22 @@ struct ib_pd *bnxt_re_alloc_pd(struct ib_device *ibdev,
 	if (udata) {
 		struct bnxt_re_pd_resp resp;
 
-		if (!ucntx->dpi) {
+		if (!ucntx->dpi.dbr) {
 			/* Allocate DPI in alloc_pd to avoid failing of
 			 * ibv_devinfo and family of application when DPIs
 			 * are depleted.
 			 */
 			if (bnxt_qplib_alloc_dpi(&rdev->qplib_res.dpi_tbl,
-						 &pd->dpi, ucntx)) {
+						 &ucntx->dpi, ucntx)) {
 				rc = -ENOMEM;
 				goto dbfail;
 			}
-			ucntx->dpi = &pd->dpi;
 		}
 
 		resp.pdid = pd->qplib_pd.id;
 		/* Still allow mapping this DBR to the new user PD. */
-		resp.dpi = ucntx->dpi->dpi;
-		resp.dbr = (u64)ucntx->dpi->umdbr;
+		resp.dpi = ucntx->dpi.dpi;
+		resp.dbr = (u64)ucntx->dpi.umdbr;
 
 		rc = ib_copy_to_udata(udata, &resp, sizeof(resp));
 		if (rc) {
@@ -960,7 +942,7 @@ static int bnxt_re_init_user_qp(struct bnxt_re_dev *rdev, struct bnxt_re_pd *pd,
 		qplib_qp->rq.nmap = umem->nmap;
 	}
 
-	qplib_qp->dpi = cntx->dpi;
+	qplib_qp->dpi = &cntx->dpi;
 	return 0;
 rqfail:
 	ib_umem_release(qp->sumem);
@@ -2403,7 +2385,7 @@ struct ib_cq *bnxt_re_create_cq(struct ib_device *ibdev,
 		}
 		cq->qplib_cq.sghead = cq->umem->sg_head.sgl;
 		cq->qplib_cq.nmap = cq->umem->nmap;
-		cq->qplib_cq.dpi = uctx->dpi;
+		cq->qplib_cq.dpi = &uctx->dpi;
 	} else {
 		cq->max_cql = min_t(u32, entries, MAX_CQL_PER_POLL);
 		cq->cql = kcalloc(cq->max_cql, sizeof(struct bnxt_qplib_cqe),
@@ -3388,8 +3370,26 @@ int bnxt_re_dealloc_ucontext(struct ib_ucontext *ib_uctx)
 	struct bnxt_re_ucontext *uctx = container_of(ib_uctx,
 						   struct bnxt_re_ucontext,
 						   ib_uctx);
+
+	struct bnxt_re_dev *rdev = uctx->rdev;
+	int rc = 0;
+
 	if (uctx->shpg)
 		free_page((unsigned long)uctx->shpg);
+
+	if (uctx->dpi.dbr) {
+		/* Free DPI only if this is the first PD allocated by the
+		 * application and mark the context dpi as NULL
+		 */
+		rc = bnxt_qplib_dealloc_dpi(&rdev->qplib_res,
+					    &rdev->qplib_res.dpi_tbl,
+					    &uctx->dpi);
+		if (rc)
+			dev_err(rdev_to_dev(rdev), "Deallocte HW DPI failed!");
+			/* Don't fail, continue*/
+		uctx->dpi.dbr = NULL;
+	}
+
 	kfree(uctx);
 	return 0;
 }
diff --git a/drivers/infiniband/hw/bnxt_re/ib_verbs.h b/drivers/infiniband/hw/bnxt_re/ib_verbs.h
index 6c160f6a5398..a0bb7e33d7ca 100644
--- a/drivers/infiniband/hw/bnxt_re/ib_verbs.h
+++ b/drivers/infiniband/hw/bnxt_re/ib_verbs.h
@@ -59,7 +59,6 @@ struct bnxt_re_pd {
 	struct bnxt_re_dev	*rdev;
 	struct ib_pd		ib_pd;
 	struct bnxt_qplib_pd	qplib_pd;
-	struct bnxt_qplib_dpi	dpi;
 	struct bnxt_re_fence_data fence;
 };
 
@@ -127,7 +126,7 @@ struct bnxt_re_mw {
 struct bnxt_re_ucontext {
 	struct bnxt_re_dev	*rdev;
 	struct ib_ucontext	ib_uctx;
-	struct bnxt_qplib_dpi	*dpi;
+	struct bnxt_qplib_dpi	dpi;
 	void			*shpg;
 	spinlock_t		sh_lock;	/* protect shpg */
 };
-- 
cgit v1.2.3


From ab69d4c8da38024191f3514c1296e9e8deea4e98 Mon Sep 17 00:00:00 2001
From: Somnath Kotur <somnath.kotur@broadcom.com>
Date: Thu, 29 Jun 2017 12:28:09 -0700
Subject: RDMA/bnxt_re: Fix WQE Size posted to HW to prevent it from throwing
 error

Posting WQE size of 2 results in a WQE_FORMAT_ERROR
thrown by the HW as it requires host to supply WQE Size with room
for atleast one SGE so that the resulting WQE size be atleast 3.

Signed-off-by: Somnath Kotur <somnath.kotur@broadcom.com>
Signed-off-by: Selvin Xavier <selvin.xavier@broadcom.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
---
 drivers/infiniband/hw/bnxt_re/qplib_fp.c | 10 ++++++++++
 1 file changed, 10 insertions(+)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/hw/bnxt_re/qplib_fp.c b/drivers/infiniband/hw/bnxt_re/qplib_fp.c
index f05500bcdcf1..8ef39df2f44f 100644
--- a/drivers/infiniband/hw/bnxt_re/qplib_fp.c
+++ b/drivers/infiniband/hw/bnxt_re/qplib_fp.c
@@ -1128,6 +1128,11 @@ int bnxt_qplib_post_send(struct bnxt_qplib_qp *qp,
 		}
 		/* Each SGE entry = 1 WQE size16 */
 		wqe_size16 = wqe->num_sge;
+		/* HW requires wqe size has room for atleast one SGE even if
+		 * none was supplied by ULP
+		 */
+		if (!wqe->num_sge)
+			wqe_size16++;
 	}
 
 	/* Specifics */
@@ -1364,6 +1369,11 @@ int bnxt_qplib_post_recv(struct bnxt_qplib_qp *qp,
 	rqe->flags = wqe->flags;
 	rqe->wqe_size = wqe->num_sge +
 			((offsetof(typeof(*rqe), data) + 15) >> 4);
+	/* HW requires wqe size has room for atleast one SGE even if none
+	 * was supplied by ULP
+	 */
+	if (!wqe->num_sge)
+		rqe->wqe_size++;
 
 	/* Supply the rqe->wr_id index to the wr_id_tbl for now */
 	rqe->wr_id[0] = cpu_to_le32(sw_prod);
-- 
cgit v1.2.3


From 4a62c5e9e2e1c15ceb1654715d9284d97f921119 Mon Sep 17 00:00:00 2001
From: Selvin Xavier <selvin.xavier@broadcom.com>
Date: Thu, 29 Jun 2017 12:28:11 -0700
Subject: RDMA/bnxt_re: Do not free the ctx_tbl entry if delete GID fails

This fix is added only to avoid system crash in some a
specific scenario. When bnxt_re driver is loaded and if
user tries to change interface mac address, delete GID
fails because QP1 is still associated with existing MAC
(default GID). If the above command fails GID tables are
not modified in the h/w or driver, but the GID context memory
is freed. Now, if the user changes the mac back to the original
value, another add_gid comes to the driver where the driver
reports that the GID is already present in its table
and tries to access the context which was already freed.

So, in this case, in order to  avoid NULL pointer de-reference,
this patch removes the context memory free  if delete_gid fails
and the same context memory is re-used in new add_gid.
Memory cleanup will be taken care during driver unload, while
deleting the GID table.

Signed-off-by: Kalesh AP <kalesh-anakkur.purayil@broadcom.com>
Signed-off-by: Selvin Xavier <selvin.xavier@broadcom.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
---
 drivers/infiniband/hw/bnxt_re/ib_verbs.c | 16 +++++++++-------
 1 file changed, 9 insertions(+), 7 deletions(-)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/hw/bnxt_re/ib_verbs.c b/drivers/infiniband/hw/bnxt_re/ib_verbs.c
index e743ffd392c6..0cd8372989ce 100644
--- a/drivers/infiniband/hw/bnxt_re/ib_verbs.c
+++ b/drivers/infiniband/hw/bnxt_re/ib_verbs.c
@@ -390,15 +390,17 @@ int bnxt_re_del_gid(struct ib_device *ibdev, u8 port_num,
 			return -EINVAL;
 		ctx->refcnt--;
 		if (!ctx->refcnt) {
-			rc = bnxt_qplib_del_sgid
-					(sgid_tbl,
-					 &sgid_tbl->tbl[ctx->idx], true);
-			if (rc)
+			rc = bnxt_qplib_del_sgid(sgid_tbl,
+						 &sgid_tbl->tbl[ctx->idx],
+						 true);
+			if (rc) {
 				dev_err(rdev_to_dev(rdev),
 					"Failed to remove GID: %#x", rc);
-			ctx_tbl = sgid_tbl->ctx;
-			ctx_tbl[ctx->idx] = NULL;
-			kfree(ctx);
+			} else {
+				ctx_tbl = sgid_tbl->ctx;
+				ctx_tbl[ctx->idx] = NULL;
+				kfree(ctx);
+			}
 		}
 	} else {
 		return -EINVAL;
-- 
cgit v1.2.3


From 58d4a671d0eac45db1c7f27c8684c277249ac127 Mon Sep 17 00:00:00 2001
From: Selvin Xavier <selvin.xavier@broadcom.com>
Date: Thu, 29 Jun 2017 12:28:12 -0700
Subject: RDMA/bnxt_re: Report supported value to IB stack in query_device

- Report supported value for max_mr_size to IB stack in query_device.
   Also, check and log if MR size requested by application in
   reg_user_mr() is greater than value currently supported by driver.
 - Report only 4K page size support for now
 - Fix Max_QP value returned by ibv_devinfo -vv.
   In case of PF, FW reserves 129 QPs for creating QP1s of VFs
   and PF. So the max_qp value reported by FW for PF doesn'tt include
   the QP1. Fixing this issue by adding 1 with the value reported
   by FW.

Signed-off-by: Selvin Xavier <selvin.xavier@broadcom.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
---
 drivers/infiniband/hw/bnxt_re/bnxt_re.h  |  2 ++
 drivers/infiniband/hw/bnxt_re/ib_verbs.c | 12 ++++++++----
 drivers/infiniband/hw/bnxt_re/qplib_sp.c |  2 ++
 3 files changed, 12 insertions(+), 4 deletions(-)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/hw/bnxt_re/bnxt_re.h b/drivers/infiniband/hw/bnxt_re/bnxt_re.h
index 08772836fded..4cb257dc61be 100644
--- a/drivers/infiniband/hw/bnxt_re/bnxt_re.h
+++ b/drivers/infiniband/hw/bnxt_re/bnxt_re.h
@@ -51,6 +51,8 @@
 #define BNXT_RE_PAGE_SIZE_8M		BIT(23)
 #define BNXT_RE_PAGE_SIZE_1G		BIT(30)
 
+#define BNXT_RE_MAX_MR_SIZE		BIT(30)
+
 #define BNXT_RE_MAX_QPC_COUNT		(64 * 1024)
 #define BNXT_RE_MAX_MRW_COUNT		(64 * 1024)
 #define BNXT_RE_MAX_SRQC_COUNT		(64 * 1024)
diff --git a/drivers/infiniband/hw/bnxt_re/ib_verbs.c b/drivers/infiniband/hw/bnxt_re/ib_verbs.c
index 0cd8372989ce..4e3e5b91d855 100644
--- a/drivers/infiniband/hw/bnxt_re/ib_verbs.c
+++ b/drivers/infiniband/hw/bnxt_re/ib_verbs.c
@@ -145,10 +145,8 @@ int bnxt_re_query_device(struct ib_device *ibdev,
 	ib_attr->fw_ver = (u64)(unsigned long)(dev_attr->fw_ver);
 	bnxt_qplib_get_guid(rdev->netdev->dev_addr,
 			    (u8 *)&ib_attr->sys_image_guid);
-	ib_attr->max_mr_size = ~0ull;
-	ib_attr->page_size_cap = BNXT_RE_PAGE_SIZE_4K | BNXT_RE_PAGE_SIZE_8K |
-				 BNXT_RE_PAGE_SIZE_64K | BNXT_RE_PAGE_SIZE_2M |
-				 BNXT_RE_PAGE_SIZE_8M | BNXT_RE_PAGE_SIZE_1G;
+	ib_attr->max_mr_size = BNXT_RE_MAX_MR_SIZE;
+	ib_attr->page_size_cap = BNXT_RE_PAGE_SIZE_4K;
 
 	ib_attr->vendor_id = rdev->en_dev->pdev->vendor;
 	ib_attr->vendor_part_id = rdev->en_dev->pdev->device;
@@ -3229,6 +3227,12 @@ struct ib_mr *bnxt_re_reg_user_mr(struct ib_pd *ib_pd, u64 start, u64 length,
 	struct scatterlist *sg;
 	int entry;
 
+	if (length > BNXT_RE_MAX_MR_SIZE) {
+		dev_err(rdev_to_dev(rdev), "MR Size: %lld > Max supported:%ld\n",
+			length, BNXT_RE_MAX_MR_SIZE);
+		return ERR_PTR(-ENOMEM);
+	}
+
 	mr = kzalloc(sizeof(*mr), GFP_KERNEL);
 	if (!mr)
 		return ERR_PTR(-ENOMEM);
diff --git a/drivers/infiniband/hw/bnxt_re/qplib_sp.c b/drivers/infiniband/hw/bnxt_re/qplib_sp.c
index fde18cf0e406..5827573875d4 100644
--- a/drivers/infiniband/hw/bnxt_re/qplib_sp.c
+++ b/drivers/infiniband/hw/bnxt_re/qplib_sp.c
@@ -81,6 +81,8 @@ int bnxt_qplib_get_dev_attr(struct bnxt_qplib_rcfw *rcfw,
 
 	/* Extract the context from the side buffer */
 	attr->max_qp = le32_to_cpu(sb->max_qp);
+	/* max_qp value reported by FW for PF doesn't include the QP1 for PF */
+	attr->max_qp += 1;
 	attr->max_qp_rd_atom =
 		sb->max_qp_rd_atom > BNXT_QPLIB_MAX_OUT_RD_ATOM ?
 		BNXT_QPLIB_MAX_OUT_RD_ATOM : sb->max_qp_rd_atom;
-- 
cgit v1.2.3


From a25d112fe9c8e8817cde1df17a82aee472c55993 Mon Sep 17 00:00:00 2001
From: Eddie Wai <eddie.wai@broadcom.com>
Date: Thu, 29 Jun 2017 12:28:13 -0700
Subject: RDMA/bnxt_re: Fixed the max_rd_atomic support for initiator and
 destination QP

There's a couple of bugs in the support of max_rd_atomic and
max_dest_rd_atomic. In the modify_qp, if the requested max_rd_atomic,
which is the ORRQ size, is greater than what the chip can support,
then we have to cap the request to chip max as we can't have the HW
overflow the ORRQ. Capping the max_rd_atomic support internally is okay
to do as the remaining read/atomic WRs will still be sitting in the SQ.
However, for the max_dest_rd_atomic, the driver has to error out as
this dictates the IRRQ size and we can't control what the remote
side sends.

Signed-off-by: Eddie Wai <eddie.wai@broadcom.com>
Signed-off-by: Selvin Xavier <selvin.xavier@broadcom.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
---
 drivers/infiniband/hw/bnxt_re/ib_verbs.c | 15 +++++++++++++--
 1 file changed, 13 insertions(+), 2 deletions(-)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/hw/bnxt_re/ib_verbs.c b/drivers/infiniband/hw/bnxt_re/ib_verbs.c
index 4e3e5b91d855..4d3cdca03c02 100644
--- a/drivers/infiniband/hw/bnxt_re/ib_verbs.c
+++ b/drivers/infiniband/hw/bnxt_re/ib_verbs.c
@@ -172,7 +172,7 @@ int bnxt_re_query_device(struct ib_device *ibdev,
 	ib_attr->max_mr = dev_attr->max_mr;
 	ib_attr->max_pd = dev_attr->max_pd;
 	ib_attr->max_qp_rd_atom = dev_attr->max_qp_rd_atom;
-	ib_attr->max_qp_init_rd_atom = dev_attr->max_qp_rd_atom;
+	ib_attr->max_qp_init_rd_atom = dev_attr->max_qp_init_rd_atom;
 	ib_attr->atomic_cap = IB_ATOMIC_HCA;
 	ib_attr->masked_atomic_cap = IB_ATOMIC_HCA;
 
@@ -1512,13 +1512,24 @@ int bnxt_re_modify_qp(struct ib_qp *ib_qp, struct ib_qp_attr *qp_attr,
 	if (qp_attr_mask & IB_QP_MAX_QP_RD_ATOMIC) {
 		qp->qplib_qp.modify_flags |=
 				CMDQ_MODIFY_QP_MODIFY_MASK_MAX_RD_ATOMIC;
-		qp->qplib_qp.max_rd_atomic = qp_attr->max_rd_atomic;
+		/* Cap the max_rd_atomic to device max */
+		qp->qplib_qp.max_rd_atomic = min_t(u32, qp_attr->max_rd_atomic,
+						   dev_attr->max_qp_rd_atom);
 	}
 	if (qp_attr_mask & IB_QP_SQ_PSN) {
 		qp->qplib_qp.modify_flags |= CMDQ_MODIFY_QP_MODIFY_MASK_SQ_PSN;
 		qp->qplib_qp.sq.psn = qp_attr->sq_psn;
 	}
 	if (qp_attr_mask & IB_QP_MAX_DEST_RD_ATOMIC) {
+		if (qp_attr->max_dest_rd_atomic >
+		    dev_attr->max_qp_init_rd_atom) {
+			dev_err(rdev_to_dev(rdev),
+				"max_dest_rd_atomic requested%d is > dev_max%d",
+				qp_attr->max_dest_rd_atomic,
+				dev_attr->max_qp_init_rd_atom);
+			return -EINVAL;
+		}
+
 		qp->qplib_qp.modify_flags |=
 				CMDQ_MODIFY_QP_MODIFY_MASK_MAX_DEST_RD_ATOMIC;
 		qp->qplib_qp.max_dest_rd_atomic = qp_attr->max_dest_rd_atomic;
-- 
cgit v1.2.3


From 536f092805cecc5c2c0ba7f051a7552619bd4491 Mon Sep 17 00:00:00 2001
From: Somnath Kotur <somnath.kotur@broadcom.com>
Date: Thu, 29 Jun 2017 12:28:14 -0700
Subject: RDMA/bnxt_re: Specify RDMA component when allocating stats context

Starting FW version 20.6.47, firmware is keeping separate statistics
for L2 and RDMA. However, driver needs to specify RDMA or not when
allocating stat_ctx.

Signed-off-by: Selvin Xavier <selvin.xavier@broadcom.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
---
 drivers/infiniband/hw/bnxt_re/main.c | 1 +
 1 file changed, 1 insertion(+)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/hw/bnxt_re/main.c b/drivers/infiniband/hw/bnxt_re/main.c
index 1fce5e73216b..ceae2d92fb08 100644
--- a/drivers/infiniband/hw/bnxt_re/main.c
+++ b/drivers/infiniband/hw/bnxt_re/main.c
@@ -333,6 +333,7 @@ static int bnxt_re_net_stats_ctx_alloc(struct bnxt_re_dev *rdev,
 	bnxt_re_init_hwrm_hdr(rdev, (void *)&req, HWRM_STAT_CTX_ALLOC, -1, -1);
 	req.update_period_ms = cpu_to_le32(1000);
 	req.stats_dma_addr = cpu_to_le64(dma_map);
+	req.stat_ctx_flags = STAT_CTX_ALLOC_REQ_STAT_CTX_FLAGS_ROCE;
 	bnxt_re_fill_fw_msg(&fw_msg, (void *)&req, sizeof(req), (void *)&resp,
 			    sizeof(resp), DFLT_HWRM_CMD_TIMEOUT);
 	rc = en_dev->en_ops->bnxt_send_fw_msg(en_dev, BNXT_ROCE_ULP, &fw_msg);
-- 
cgit v1.2.3


From 254cd2590d408f2c0375fbf73ef8250d47529e7b Mon Sep 17 00:00:00 2001
From: Devesh Sharma <devesh.sharma@broadcom.com>
Date: Thu, 29 Jun 2017 12:28:16 -0700
Subject: RDMA/bnxt_re: Enable atomics only if host bios supports

Driver shall check if the host system bios has enabled
Atomic operations capability in PCI Device Control 2
register of the pci-device. Expose the ATOMIC_HCA
flag only if the Atomic operations capability is set.

Signed-off-by: Devesh Sharma <devesh.sharma@broadcom.com>
Signed-off-by: Selvin Xavier <selvin.xavier@broadcom.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
---
 drivers/infiniband/hw/bnxt_re/ib_verbs.c |  6 ++++--
 drivers/infiniband/hw/bnxt_re/qplib_sp.c | 14 ++++++++++++++
 drivers/infiniband/hw/bnxt_re/qplib_sp.h |  3 +++
 3 files changed, 21 insertions(+), 2 deletions(-)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/hw/bnxt_re/ib_verbs.c b/drivers/infiniband/hw/bnxt_re/ib_verbs.c
index 4d3cdca03c02..7b17030d2696 100644
--- a/drivers/infiniband/hw/bnxt_re/ib_verbs.c
+++ b/drivers/infiniband/hw/bnxt_re/ib_verbs.c
@@ -173,8 +173,10 @@ int bnxt_re_query_device(struct ib_device *ibdev,
 	ib_attr->max_pd = dev_attr->max_pd;
 	ib_attr->max_qp_rd_atom = dev_attr->max_qp_rd_atom;
 	ib_attr->max_qp_init_rd_atom = dev_attr->max_qp_init_rd_atom;
-	ib_attr->atomic_cap = IB_ATOMIC_HCA;
-	ib_attr->masked_atomic_cap = IB_ATOMIC_HCA;
+	if (dev_attr->is_atomic) {
+		ib_attr->atomic_cap = IB_ATOMIC_HCA;
+		ib_attr->masked_atomic_cap = IB_ATOMIC_HCA;
+	}
 
 	ib_attr->max_ee_rd_atom = 0;
 	ib_attr->max_res_rd_atom = 0;
diff --git a/drivers/infiniband/hw/bnxt_re/qplib_sp.c b/drivers/infiniband/hw/bnxt_re/qplib_sp.c
index 5827573875d4..ef91ab786dd4 100644
--- a/drivers/infiniband/hw/bnxt_re/qplib_sp.c
+++ b/drivers/infiniband/hw/bnxt_re/qplib_sp.c
@@ -51,6 +51,19 @@ const struct bnxt_qplib_gid bnxt_qplib_gid_zero = {{ 0, 0, 0, 0, 0, 0, 0, 0,
 						     0, 0, 0, 0, 0, 0, 0, 0 } };
 
 /* Device */
+
+static bool bnxt_qplib_is_atomic_cap(struct bnxt_qplib_rcfw *rcfw)
+{
+	int rc;
+	u16 pcie_ctl2;
+
+	rc = pcie_capability_read_word(rcfw->pdev, PCI_EXP_DEVCTL2,
+				       &pcie_ctl2);
+	if (rc)
+		return false;
+	return !!(pcie_ctl2 & PCI_EXP_DEVCTL2_ATOMIC_REQ);
+}
+
 int bnxt_qplib_get_dev_attr(struct bnxt_qplib_rcfw *rcfw,
 			    struct bnxt_qplib_dev_attr *attr)
 {
@@ -131,6 +144,7 @@ int bnxt_qplib_get_dev_attr(struct bnxt_qplib_rcfw *rcfw,
 		attr->tqm_alloc_reqs[i * 4 + 3] = *(++tqm_alloc);
 	}
 
+	attr->is_atomic = bnxt_qplib_is_atomic_cap(rcfw);
 bail:
 	bnxt_qplib_rcfw_free_sbuf(rcfw, sbuf);
 	return rc;
diff --git a/drivers/infiniband/hw/bnxt_re/qplib_sp.h b/drivers/infiniband/hw/bnxt_re/qplib_sp.h
index a543f959098b..2ce7e2a32cf0 100644
--- a/drivers/infiniband/hw/bnxt_re/qplib_sp.h
+++ b/drivers/infiniband/hw/bnxt_re/qplib_sp.h
@@ -42,6 +42,8 @@
 
 #define BNXT_QPLIB_RESERVED_QP_WRS	128
 
+#define PCI_EXP_DEVCTL2_ATOMIC_REQ      0x0040
+
 struct bnxt_qplib_dev_attr {
 	char				fw_ver[32];
 	u16				max_sgid;
@@ -70,6 +72,7 @@ struct bnxt_qplib_dev_attr {
 	u32				max_inline_data;
 	u32				l2_db_size;
 	u8				tqm_alloc_reqs[MAX_TQM_ALLOC_REQ];
+	bool				is_atomic;
 };
 
 struct bnxt_qplib_pd {
-- 
cgit v1.2.3


From 10d1dedf9b428ed776f244eb8b43f28974da3fb0 Mon Sep 17 00:00:00 2001
From: Devesh Sharma <devesh.sharma@broadcom.com>
Date: Thu, 29 Jun 2017 12:28:17 -0700
Subject: RDMA/bnxt_re: Fix return value of poll routine

Fix the incorrect reporting of number of polled
entries by taking into account the max CQ depth
in the driver.

Signed-off-by: Devesh Sharma <devesh.sharma@broadcom.com>
Signed-off-by: Selvin Xavier <selvin.xavier@broadcom.com>
Reviewed-by: Leon Romanovsky <leonro@mellanox.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
---
 drivers/infiniband/hw/bnxt_re/ib_verbs.c | 1 +
 1 file changed, 1 insertion(+)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/hw/bnxt_re/ib_verbs.c b/drivers/infiniband/hw/bnxt_re/ib_verbs.c
index 7b17030d2696..b92a06d17186 100644
--- a/drivers/infiniband/hw/bnxt_re/ib_verbs.c
+++ b/drivers/infiniband/hw/bnxt_re/ib_verbs.c
@@ -2900,6 +2900,7 @@ int bnxt_re_poll_cq(struct ib_cq *ib_cq, int num_entries, struct ib_wc *wc)
 
 	spin_lock_irqsave(&cq->cq_lock, flags);
 	budget = min_t(u32, num_entries, cq->max_cql);
+	num_entries = budget;
 	if (!cq->cql) {
 		dev_err(rdev_to_dev(cq->rdev), "POLL CQ : no CQL to use");
 		goto exit;
-- 
cgit v1.2.3


From 499e456981d376ef614e257d5dcb280581db9ac6 Mon Sep 17 00:00:00 2001
From: Selvin Xavier <selvin.xavier@broadcom.com>
Date: Thu, 29 Jun 2017 12:28:18 -0700
Subject: RDMA/bnxt_re: Report MISSED_EVENTS in req_notify_cq

While invoking the req_notify_cq hook, ULPs can request
whether the CQs have any CQEs pending. If CQEs are pending,
drivers can indicate  it by returning 1 for req_notify_cq.
The stack will poll CQ again till CQ is empty.

This patch peeks the CQ for any valid entries and return accordingly.

Signed-off-by: Selvin Xavier <selvin.xavier@broadcom.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
---
 drivers/infiniband/hw/bnxt_re/ib_verbs.c |  5 +++++
 drivers/infiniband/hw/bnxt_re/qplib_fp.c | 19 +++++++++++++++++++
 drivers/infiniband/hw/bnxt_re/qplib_fp.h |  1 +
 3 files changed, 25 insertions(+)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/hw/bnxt_re/ib_verbs.c b/drivers/infiniband/hw/bnxt_re/ib_verbs.c
index b92a06d17186..d5aa5a1a7e00 100644
--- a/drivers/infiniband/hw/bnxt_re/ib_verbs.c
+++ b/drivers/infiniband/hw/bnxt_re/ib_verbs.c
@@ -3027,6 +3027,11 @@ int bnxt_re_req_notify_cq(struct ib_cq *ib_cq,
 	else if (ib_cqn_flags & IB_CQ_SOLICITED)
 		type = DBR_DBR_TYPE_CQ_ARMSE;
 
+	/* Poll to see if there are missed events */
+	if ((ib_cqn_flags & IB_CQ_REPORT_MISSED_EVENTS) &&
+	    !(bnxt_qplib_is_cq_empty(&cq->qplib_cq)))
+		return 1;
+
 	bnxt_qplib_req_notify_cq(&cq->qplib_cq, type);
 
 	return 0;
diff --git a/drivers/infiniband/hw/bnxt_re/qplib_fp.c b/drivers/infiniband/hw/bnxt_re/qplib_fp.c
index 8ef39df2f44f..9af1514e5944 100644
--- a/drivers/infiniband/hw/bnxt_re/qplib_fp.c
+++ b/drivers/infiniband/hw/bnxt_re/qplib_fp.c
@@ -1895,6 +1895,25 @@ flush_rq:
 	return rc;
 }
 
+bool bnxt_qplib_is_cq_empty(struct bnxt_qplib_cq *cq)
+{
+	struct cq_base *hw_cqe, **hw_cqe_ptr;
+	unsigned long flags;
+	u32 sw_cons, raw_cons;
+	bool rc = true;
+
+	spin_lock_irqsave(&cq->hwq.lock, flags);
+	raw_cons = cq->hwq.cons;
+	sw_cons = HWQ_CMP(raw_cons, &cq->hwq);
+	hw_cqe_ptr = (struct cq_base **)cq->hwq.pbl_ptr;
+	hw_cqe = &hw_cqe_ptr[CQE_PG(sw_cons)][CQE_IDX(sw_cons)];
+
+	 /* Check for Valid bit. If the CQE is valid, return false */
+	rc = !CQE_CMP_VALID(hw_cqe, raw_cons, cq->hwq.max_elements);
+	spin_unlock_irqrestore(&cq->hwq.lock, flags);
+	return rc;
+}
+
 static int bnxt_qplib_cq_process_res_raweth_qp1(struct bnxt_qplib_cq *cq,
 						struct cq_res_raweth_qp1 *hwcqe,
 						struct bnxt_qplib_cqe **pcqe,
diff --git a/drivers/infiniband/hw/bnxt_re/qplib_fp.h b/drivers/infiniband/hw/bnxt_re/qplib_fp.h
index 36b7b7db0e3f..19176e06c98a 100644
--- a/drivers/infiniband/hw/bnxt_re/qplib_fp.h
+++ b/drivers/infiniband/hw/bnxt_re/qplib_fp.h
@@ -449,6 +449,7 @@ int bnxt_qplib_create_cq(struct bnxt_qplib_res *res, struct bnxt_qplib_cq *cq);
 int bnxt_qplib_destroy_cq(struct bnxt_qplib_res *res, struct bnxt_qplib_cq *cq);
 int bnxt_qplib_poll_cq(struct bnxt_qplib_cq *cq, struct bnxt_qplib_cqe *cqe,
 		       int num, struct bnxt_qplib_qp **qp);
+bool bnxt_qplib_is_cq_empty(struct bnxt_qplib_cq *cq);
 void bnxt_qplib_req_notify_cq(struct bnxt_qplib_cq *cq, u32 arm_type);
 void bnxt_qplib_free_nq(struct bnxt_qplib_nq *nq);
 int bnxt_qplib_alloc_nq(struct pci_dev *pdev, struct bnxt_qplib_nq *nq);
-- 
cgit v1.2.3


From 601577b7d14cba5848b4c44ef719881f8538f702 Mon Sep 17 00:00:00 2001
From: Selvin Xavier <selvin.xavier@broadcom.com>
Date: Thu, 29 Jun 2017 12:28:19 -0700
Subject: RDMA/bnxt_re: Fix the value reported for local ack delay

Local ack delay exposed by the driver is 0 which means infinite QP
timeout. Reporting the default value to 16 (approx 260ms)

Signed-off-by: Selvin Xavier <selvin.xavier@broadcom.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
---
 drivers/infiniband/hw/bnxt_re/bnxt_re.h  | 7 +++++++
 drivers/infiniband/hw/bnxt_re/ib_verbs.c | 2 +-
 2 files changed, 8 insertions(+), 1 deletion(-)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/hw/bnxt_re/bnxt_re.h b/drivers/infiniband/hw/bnxt_re/bnxt_re.h
index 4cb257dc61be..85527532c49d 100644
--- a/drivers/infiniband/hw/bnxt_re/bnxt_re.h
+++ b/drivers/infiniband/hw/bnxt_re/bnxt_re.h
@@ -62,6 +62,13 @@
 
 #define BNXT_RE_RQ_WQE_THRESHOLD	32
 
+/*
+ * Setting the default ack delay value to 16, which means
+ * the default timeout is approx. 260ms(4 usec * 2 ^(timeout))
+ */
+
+#define BNXT_RE_DEFAULT_ACK_DELAY	16
+
 struct bnxt_re_work {
 	struct work_struct	work;
 	unsigned long		event;
diff --git a/drivers/infiniband/hw/bnxt_re/ib_verbs.c b/drivers/infiniband/hw/bnxt_re/ib_verbs.c
index d5aa5a1a7e00..f0e01b3ac711 100644
--- a/drivers/infiniband/hw/bnxt_re/ib_verbs.c
+++ b/drivers/infiniband/hw/bnxt_re/ib_verbs.c
@@ -201,7 +201,7 @@ int bnxt_re_query_device(struct ib_device *ibdev,
 	ib_attr->max_fast_reg_page_list_len = MAX_PBL_LVL_1_PGS;
 
 	ib_attr->max_pkeys = 1;
-	ib_attr->local_ca_ack_delay = 0;
+	ib_attr->local_ca_ack_delay = BNXT_RE_DEFAULT_ACK_DELAY;
 	return 0;
 }
 
-- 
cgit v1.2.3


From 266098b841d48f7f0db40424bdbc072e4db14e9b Mon Sep 17 00:00:00 2001
From: Matan Barak <matanb@mellanox.com>
Date: Thu, 8 Jun 2017 17:30:47 +0300
Subject: IB/core: Fix sparse warnings

Delete unused variables to prevent sparse warnings.

Fixes: db1b5ddd5336 ("IB/core: Rename uverbs event file structure")
Fixes: fd3c7904db6e ("IB/core: Change idr objects to use the new schema")

Signed-off-by: Doug Ledford <dledford@redhat.com>
---
 drivers/infiniband/core/uverbs_cmd.c | 10 ----------
 1 file changed, 10 deletions(-)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/core/uverbs_cmd.c b/drivers/infiniband/core/uverbs_cmd.c
index 3f55d18a3791..71451eae42de 100644
--- a/drivers/infiniband/core/uverbs_cmd.c
+++ b/drivers/infiniband/core/uverbs_cmd.c
@@ -1296,7 +1296,6 @@ ssize_t ib_uverbs_destroy_cq(struct ib_uverbs_file *file,
 	struct ib_uobject		*uobj;
 	struct ib_cq               	*cq;
 	struct ib_ucq_object        	*obj;
-	struct ib_uverbs_event_queue	*ev_queue;
 	int                        	 ret = -EINVAL;
 
 	if (copy_from_user(&cmd, buf, sizeof cmd))
@@ -1313,7 +1312,6 @@ ssize_t ib_uverbs_destroy_cq(struct ib_uverbs_file *file,
 	 */
 	uverbs_uobject_get(uobj);
 	cq      = uobj->object;
-	ev_queue = cq->cq_context;
 	obj     = container_of(cq->uobject, struct ib_ucq_object, uobject);
 
 	memset(&resp, 0, sizeof(resp));
@@ -2088,7 +2086,6 @@ ssize_t ib_uverbs_destroy_qp(struct ib_uverbs_file *file,
 	struct ib_uverbs_destroy_qp      cmd;
 	struct ib_uverbs_destroy_qp_resp resp;
 	struct ib_uobject		*uobj;
-	struct ib_qp               	*qp;
 	struct ib_uqp_object        	*obj;
 	int                        	 ret = -EINVAL;
 
@@ -2102,7 +2099,6 @@ ssize_t ib_uverbs_destroy_qp(struct ib_uverbs_file *file,
 	if (IS_ERR(uobj))
 		return PTR_ERR(uobj);
 
-	qp  = uobj->object;
 	obj = container_of(uobj, struct ib_uqp_object, uevent.uobject);
 	/*
 	 * Make sure we don't free the memory in remove_commit as we still
@@ -3004,7 +3000,6 @@ int ib_uverbs_ex_destroy_wq(struct ib_uverbs_file *file,
 {
 	struct ib_uverbs_ex_destroy_wq	cmd = {};
 	struct ib_uverbs_ex_destroy_wq_resp	resp = {};
-	struct ib_wq			*wq;
 	struct ib_uobject		*uobj;
 	struct ib_uwq_object		*obj;
 	size_t required_cmd_sz;
@@ -3038,7 +3033,6 @@ int ib_uverbs_ex_destroy_wq(struct ib_uverbs_file *file,
 	if (IS_ERR(uobj))
 		return PTR_ERR(uobj);
 
-	wq = uobj->object;
 	obj = container_of(uobj, struct ib_uwq_object, uevent.uobject);
 	/*
 	 * Make sure we don't free the memory in remove_commit as we still
@@ -3728,10 +3722,8 @@ ssize_t ib_uverbs_destroy_srq(struct ib_uverbs_file *file,
 	struct ib_uverbs_destroy_srq      cmd;
 	struct ib_uverbs_destroy_srq_resp resp;
 	struct ib_uobject		 *uobj;
-	struct ib_srq               	 *srq;
 	struct ib_uevent_object        	 *obj;
 	int                         	  ret = -EINVAL;
-	enum ib_srq_type		  srq_type;
 
 	if (copy_from_user(&cmd, buf, sizeof cmd))
 		return -EFAULT;
@@ -3741,9 +3733,7 @@ ssize_t ib_uverbs_destroy_srq(struct ib_uverbs_file *file,
 	if (IS_ERR(uobj))
 		return PTR_ERR(uobj);
 
-	srq = uobj->object;
 	obj = container_of(uobj, struct ib_uevent_object, uobject);
-	srq_type = srq->srq_type;
 	/*
 	 * Make sure we don't free the memory in remove_commit as we still
 	 * needs the uobject memory to create the response.
-- 
cgit v1.2.3


From a25ce4270bfdd522207b02f81a594c7d1746b697 Mon Sep 17 00:00:00 2001
From: Kaike Wan <kaike.wan@intel.com>
Date: Sat, 17 Jun 2017 10:37:26 -0700
Subject: IB/rdmavt: Setting of QP timeout can overflow jiffies computation

Current computation of qp->timeout_jiffies in rvt_modify_qp() will cause
overflow due to the fact that the input to the function usecs_to_jiffies
is only 32-bit ( unsigned int). Overflow will occur when attr->timeout is
equal to or greater than 30. The consequence is unnecessarily excessive
retry and thus degradation of the system performance.

This patch fixes the problem by limiting the input to 5-bit and calling
usecs_to_jiffies() before multiplying the scaling factor.

Reviewed-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Signed-off-by: Kaike Wan <kaike.wan@intel.com>
Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
---
 drivers/infiniband/sw/rdmavt/qp.c |  4 +---
 include/rdma/rdmavt_qp.h          | 14 ++++++++++++++
 2 files changed, 15 insertions(+), 3 deletions(-)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/sw/rdmavt/qp.c b/drivers/infiniband/sw/rdmavt/qp.c
index 459865439a0b..8876ee7bc326 100644
--- a/drivers/infiniband/sw/rdmavt/qp.c
+++ b/drivers/infiniband/sw/rdmavt/qp.c
@@ -1258,9 +1258,7 @@ int rvt_modify_qp(struct ib_qp *ibqp, struct ib_qp_attr *attr,
 
 	if (attr_mask & IB_QP_TIMEOUT) {
 		qp->timeout = attr->timeout;
-		qp->timeout_jiffies =
-			usecs_to_jiffies((4096UL * (1UL << qp->timeout)) /
-				1000UL);
+		qp->timeout_jiffies = rvt_timeout_to_jiffies(qp->timeout);
 	}
 
 	if (attr_mask & IB_QP_QKEY)
diff --git a/include/rdma/rdmavt_qp.h b/include/rdma/rdmavt_qp.h
index be6472e5b06b..d664d2e76280 100644
--- a/include/rdma/rdmavt_qp.h
+++ b/include/rdma/rdmavt_qp.h
@@ -647,6 +647,20 @@ static inline u32 rvt_div_mtu(struct rvt_qp *qp, u32 len)
 	return len >> qp->log_pmtu;
 }
 
+/**
+ * rvt_timeout_to_jiffies - Convert a ULP timeout input into jiffies
+ * @timeout - timeout input(0 - 31).
+ *
+ * Return a timeout value in jiffies.
+ */
+static inline unsigned long rvt_timeout_to_jiffies(u8 timeout)
+{
+	if (timeout > 31)
+		timeout = 31;
+
+	return usecs_to_jiffies(1U << timeout) * 4096UL / 1000UL;
+}
+
 extern const int  ib_rvt_state_ops[];
 
 struct rvt_dev_info;
-- 
cgit v1.2.3


From 4542e3c79a2c5a167cbeb4f4190d5f705d272002 Mon Sep 17 00:00:00 2001
From: Håkon Bugge <haakon.bugge@oracle.com>
Date: Tue, 20 Jun 2017 14:07:50 +0200
Subject: IB/mlx4: Fix CM REQ retries in paravirt mode
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

CM REQs cannot be successfully retried, because a new pv_cm_id is
created for each request, without checking if one already exists.

By checking if an id exists before creating one, the bug is fixed.

This bug can be provoked by running an RDMA CM user-land application,
but inserting a five seconds delay before the rdma_accept() call on
the passive side. This delay is larger than the default CMA timeout,
and triggers a retry from the active side. The retried REQ will use
another pv_cm_id (the cm_id on the wire). This confuses the CM
protocol and two REJs are sent from the passive side.

Here is an excerpt from ibdump running without the patch:

3.285092       LID: 4 -> LID: 4       SDP 290 CM: ConnectRequest(SDP Hello)
7.382711       LID: 4 -> LID: 4       SDP 290 CM: ConnectRequest(SDP Hello)
7.382861       LID: 4 -> LID: 4       InfiniBand 290 CM: ConnectReject
7.387644       LID: 4 -> LID: 4       InfiniBand 290 CM: ConnectReject

and here is the same with bug fix applied:

3.251010       LID: 4 -> LID: 4       SDP 290 CM: ConnectRequest(SDP Hello)
7.349387       LID: 4 -> LID: 4       SDP 290 CM: ConnectRequest(SDP Hello)
8.258443       LID: 4 -> LID: 4       SDP 290 CM: ConnectReply(SDP Hello)
8.259890       LID: 4 -> LID: 4       InfiniBand 290 CM: ReadyToUse

Suggested-by: Venkat Venkatsubra <venkat.x.venkatsubra@oracle.com>
Signed-off-by: Håkon Bugge <haakon.bugge@oracle.com>
Reported-by: Wei Lin Guay <wei.lin.guay@oracle.com>
Tested-by: Wei Lin Guay <wei.lin.guay@oracle.com>
Reviewed-by: Yuval Shaia <yuval.shaia@oracle.com>
Acked-by: Jack Morgenstein <jackm@dev.mellanox.co.il>
Signed-off-by: Doug Ledford <dledford@redhat.com>
---
 drivers/infiniband/hw/mlx4/cm.c | 4 ++++
 1 file changed, 4 insertions(+)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/hw/mlx4/cm.c b/drivers/infiniband/hw/mlx4/cm.c
index 1e6c526450d9..fedaf8260105 100644
--- a/drivers/infiniband/hw/mlx4/cm.c
+++ b/drivers/infiniband/hw/mlx4/cm.c
@@ -323,6 +323,9 @@ int mlx4_ib_multiplex_cm_handler(struct ib_device *ibdev, int port, int slave_id
 			mad->mad_hdr.attr_id == CM_REP_ATTR_ID ||
 			mad->mad_hdr.attr_id == CM_SIDR_REQ_ATTR_ID) {
 		sl_cm_id = get_local_comm_id(mad);
+		id = id_map_get(ibdev, &pv_cm_id, slave_id, sl_cm_id);
+		if (id)
+			goto cont;
 		id = id_map_alloc(ibdev, slave_id, sl_cm_id);
 		if (IS_ERR(id)) {
 			mlx4_ib_warn(ibdev, "%s: id{slave: %d, sl_cm_id: 0x%x} Failed to id_map_alloc\n",
@@ -343,6 +346,7 @@ int mlx4_ib_multiplex_cm_handler(struct ib_device *ibdev, int port, int slave_id
 		return -EINVAL;
 	}
 
+cont:
 	set_local_comm_id(mad, id->pv_cm_id);
 
 	if (mad->mad_hdr.attr_id == CM_DREQ_ATTR_ID)
-- 
cgit v1.2.3


From 720336c42e41a917002fcae3aa14e30f5022bbb7 Mon Sep 17 00:00:00 2001
From: Ganesh Goudar <ganeshgr@chelsio.com>
Date: Wed, 21 Jun 2017 19:55:43 +0530
Subject: iw_cxgb4: don't use WR keys/addrs for 0 byte reads

Only use the read sge lkey/addr and the remote rkey/addr if the
length of the read is not zero. Otherwise the read response might
be treated as the RTR read response and not delivered to the
application. Or worse Terminator hardware will fail a 0B read
if the STAG is 0 even if the read length is 0.

Signed-off-by: Steve Wise <swise@opengridcomputing.com>
Signed-off-by: Ganesh Goudar <ganeshgr@chelsio.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
---
 drivers/infiniband/hw/cxgb4/qp.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/hw/cxgb4/qp.c b/drivers/infiniband/hw/cxgb4/qp.c
index bfc77596acbe..cb7fc0d35d1d 100644
--- a/drivers/infiniband/hw/cxgb4/qp.c
+++ b/drivers/infiniband/hw/cxgb4/qp.c
@@ -569,7 +569,7 @@ static int build_rdma_read(union t4_wr *wqe, struct ib_send_wr *wr, u8 *len16)
 {
 	if (wr->num_sge > 1)
 		return -EINVAL;
-	if (wr->num_sge) {
+	if (wr->num_sge && wr->sg_list[0].length) {
 		wqe->read.stag_src = cpu_to_be32(rdma_wr(wr)->rkey);
 		wqe->read.to_src_hi = cpu_to_be32((u32)(rdma_wr(wr)->remote_addr
 							>> 32));
-- 
cgit v1.2.3


From c75d3ec8c0ee469de79ae83c1a827d753603e49f Mon Sep 17 00:00:00 2001
From: "Amrani, Ram" <Ram.Amrani@cavium.com>
Date: Mon, 26 Jun 2017 19:05:04 +0300
Subject: RDMA/qedr: Prevent memory overrun in verbs' user responses

Wrap ib_copy_to_udata with a function that ensures that the data
being copied over to user space isn't longer than the allowed.

Fixes: cecbcddf6461 ("qedr: Add support for QP verbs")
Fixes: a7efd7773e31 ("qedr: Add support for PD,PKEY and CQ verbs")
Fixes: ac1b36e55a51 ("qedr: Add support for user context verbs")
Signed-off-by: Ram Amrani <Ram.Amrani@cavium.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
---
 drivers/infiniband/hw/qedr/verbs.c | 16 ++++++++++++----
 1 file changed, 12 insertions(+), 4 deletions(-)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/hw/qedr/verbs.c b/drivers/infiniband/hw/qedr/verbs.c
index 548e4d1e998f..2ae71b8f1ba8 100644
--- a/drivers/infiniband/hw/qedr/verbs.c
+++ b/drivers/infiniband/hw/qedr/verbs.c
@@ -53,6 +53,14 @@
 
 #define DB_ADDR_SHIFT(addr)		((addr) << DB_PWM_ADDR_OFFSET_SHIFT)
 
+static inline int qedr_ib_copy_to_udata(struct ib_udata *udata, void *src,
+					size_t len)
+{
+	size_t min_len = min_t(size_t, len, udata->outlen);
+
+	return ib_copy_to_udata(udata, src, min_len);
+}
+
 int qedr_query_pkey(struct ib_device *ibdev, u8 port, u16 index, u16 *pkey)
 {
 	if (index > QEDR_ROCE_PKEY_TABLE_LEN)
@@ -378,7 +386,7 @@ struct ib_ucontext *qedr_alloc_ucontext(struct ib_device *ibdev,
 	uresp.sges_per_srq_wr = dev->attr.max_srq_sge;
 	uresp.max_cqes = QEDR_MAX_CQES;
 
-	rc = ib_copy_to_udata(udata, &uresp, sizeof(uresp));
+	rc = qedr_ib_copy_to_udata(udata, &uresp, sizeof(uresp));
 	if (rc)
 		goto err;
 
@@ -499,7 +507,7 @@ struct ib_pd *qedr_alloc_pd(struct ib_device *ibdev,
 
 		uresp.pd_id = pd_id;
 
-		rc = ib_copy_to_udata(udata, &uresp, sizeof(uresp));
+		rc = qedr_ib_copy_to_udata(udata, &uresp, sizeof(uresp));
 		if (rc) {
 			DP_ERR(dev, "copy error pd_id=0x%x.\n", pd_id);
 			dev->ops->rdma_dealloc_pd(dev->rdma_ctx, pd_id);
@@ -729,7 +737,7 @@ static int qedr_copy_cq_uresp(struct qedr_dev *dev,
 	uresp.db_offset = DB_ADDR_SHIFT(DQ_PWM_OFFSET_UCM_RDMA_CQ_CONS_32BIT);
 	uresp.icid = cq->icid;
 
-	rc = ib_copy_to_udata(udata, &uresp, sizeof(uresp));
+	rc = qedr_ib_copy_to_udata(udata, &uresp, sizeof(uresp));
 	if (rc)
 		DP_ERR(dev, "copy error cqid=0x%x.\n", cq->icid);
 
@@ -1238,7 +1246,7 @@ static int qedr_copy_qp_uresp(struct qedr_dev *dev,
 	uresp.atomic_supported = dev->atomic_cap != IB_ATOMIC_NONE;
 	uresp.qp_id = qp->qp_id;
 
-	rc = ib_copy_to_udata(udata, &uresp, sizeof(uresp));
+	rc = qedr_ib_copy_to_udata(udata, &uresp, sizeof(uresp));
 	if (rc)
 		DP_ERR(dev,
 		       "create qp: failed a copy to user space with qp icid=0x%x.\n",
-- 
cgit v1.2.3


From 1217197142d1681a8b8aaa88cdf4b245b76974cd Mon Sep 17 00:00:00 2001
From: Vijay Immanuel <vijayi@attalasystems.com>
Date: Tue, 27 Jun 2017 12:19:38 +0300
Subject: rxe: fix broken receive queue draining

If we modified the qp to ERROR state, and
drained the recieve queue, post_recv must
trigger the responder task to complete
the drain work request.

Cc: Bart Van Assche <bart.vanassche@sandisk.com>
Signed-off-by: Vijay Immanuel <vijayi@attalasystems.com>
Signed-off-by: Sagi Grimberg <sagi@grimberg.me>
Reviewed-by: Bart Van Assche <bart.vanassche@wdc.com>--
Signed-off-by: Doug Ledford <dledford@redhat.com>
---
 drivers/infiniband/sw/rxe/rxe_resp.c  | 3 +++
 drivers/infiniband/sw/rxe/rxe_verbs.c | 3 +++
 2 files changed, 6 insertions(+)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/sw/rxe/rxe_resp.c b/drivers/infiniband/sw/rxe/rxe_resp.c
index be944d5aa9af..a958ee918a49 100644
--- a/drivers/infiniband/sw/rxe/rxe_resp.c
+++ b/drivers/infiniband/sw/rxe/rxe_resp.c
@@ -1219,6 +1219,9 @@ void rxe_drain_req_pkts(struct rxe_qp *qp, bool notify)
 		kfree_skb(skb);
 	}
 
+	if (notify)
+		return;
+
 	while (!qp->srq && qp->rq.queue && queue_head(qp->rq.queue))
 		advance_consumer(qp->rq.queue);
 }
diff --git a/drivers/infiniband/sw/rxe/rxe_verbs.c b/drivers/infiniband/sw/rxe/rxe_verbs.c
index 07511718d98d..af90a7d42b96 100644
--- a/drivers/infiniband/sw/rxe/rxe_verbs.c
+++ b/drivers/infiniband/sw/rxe/rxe_verbs.c
@@ -914,6 +914,9 @@ static int rxe_post_recv(struct ib_qp *ibqp, struct ib_recv_wr *wr,
 
 	spin_unlock_irqrestore(&rq->producer_lock, flags);
 
+	if (qp->resp.state == QP_STATE_ERROR)
+		rxe_run_task(&qp->resp.task, 1);
+
 err1:
 	return err;
 }
-- 
cgit v1.2.3


From e6e52aec494900912fedd7b595b8827ba70a670d Mon Sep 17 00:00:00 2001
From: Sagi Grimberg <sagi@grimberg.me>
Date: Thu, 6 Jul 2017 10:21:36 +0300
Subject: RDMA/iser: don't send an rkey if all data is written as
 immadiate-data

We might get some bogus error completions in case the target will
remotely invalidate the rkey and the HCA will need to retransmit
from this buffer.

Signed-off-by: Sagi Grimberg <sagi@grimberg.me>
Signed-off-by: Doug Ledford <dledford@redhat.com>
---
 drivers/infiniband/ulp/iser/iser_initiator.c | 6 ++++--
 1 file changed, 4 insertions(+), 2 deletions(-)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/ulp/iser/iser_initiator.c b/drivers/infiniband/ulp/iser/iser_initiator.c
index 12ed62ce9ff7..2a07692007bd 100644
--- a/drivers/infiniband/ulp/iser/iser_initiator.c
+++ b/drivers/infiniband/ulp/iser/iser_initiator.c
@@ -137,8 +137,10 @@ iser_prepare_write_cmd(struct iscsi_task *task,
 
 	if (unsol_sz < edtl) {
 		hdr->flags     |= ISER_WSV;
-		hdr->write_stag = cpu_to_be32(mem_reg->rkey);
-		hdr->write_va   = cpu_to_be64(mem_reg->sge.addr + unsol_sz);
+		if (buf_out->data_len > imm_sz) {
+			hdr->write_stag = cpu_to_be32(mem_reg->rkey);
+			hdr->write_va = cpu_to_be64(mem_reg->sge.addr + unsol_sz);
+		}
 
 		iser_dbg("Cmd itt:%d, WRITE tags, RKEY:%#.4X "
 			 "VA:%#llX + unsol:%d\n",
-- 
cgit v1.2.3


From 5a7a88f1b488e4ee49eb3d5b82612d4d9ffdf2c3 Mon Sep 17 00:00:00 2001
From: "Ismail, Mustafa" <mustafa.ismail@intel.com>
Date: Fri, 14 Jul 2017 09:41:30 -0500
Subject: RDMA/uverbs: Fix the check for port number

The port number is only valid if IB_QP_PORT is set in the mask.
So only check port number if it is valid to prevent modify_qp from
failing due to an invalid port number.

Fixes: 5ecce4c9b17b("Check port number supplied by user verbs cmds")
Cc: <stable@vger.kernel.org> # v2.6.14+
Reviewed-by: Steve Wise <swise@opengridcomputing.com>
Signed-off-by: Mustafa Ismail <mustafa.ismail@intel.com>
Tested-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
---
 drivers/infiniband/core/uverbs_cmd.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/core/uverbs_cmd.c b/drivers/infiniband/core/uverbs_cmd.c
index 71451eae42de..2c98533a0203 100644
--- a/drivers/infiniband/core/uverbs_cmd.c
+++ b/drivers/infiniband/core/uverbs_cmd.c
@@ -1933,7 +1933,8 @@ static int modify_qp(struct ib_uverbs_file *file,
 		goto out;
 	}
 
-	if (!rdma_is_port_valid(qp->device, cmd->base.port_num)) {
+	if ((cmd->base.attr_mask & IB_QP_PORT) &&
+	    !rdma_is_port_valid(qp->device, cmd->base.port_num)) {
 		ret = -EINVAL;
 		goto release_qp;
 	}
-- 
cgit v1.2.3


From a62ab66b13a0f9bcb17b7b761f6670941ed5cd62 Mon Sep 17 00:00:00 2001
From: "Ismail, Mustafa" <mustafa.ismail@intel.com>
Date: Fri, 14 Jul 2017 09:41:31 -0500
Subject: RDMA/core: Initialize port_num in qp_attr

Initialize the port_num for iWARP in rdma_init_qp_attr.

Fixes: 5ecce4c9b17b("Check port number supplied by user verbs cmds")
Cc: <stable@vger.kernel.org> # v2.6.14+
Reviewed-by: Steve Wise <swise@opengridcomputing.com>
Signed-off-by: Mustafa Ismail <mustafa.ismail@intel.com>
Tested-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
---
 drivers/infiniband/core/cma.c | 2 ++
 1 file changed, 2 insertions(+)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/core/cma.c b/drivers/infiniband/core/cma.c
index 11aff923b633..0eb393237ba2 100644
--- a/drivers/infiniband/core/cma.c
+++ b/drivers/infiniband/core/cma.c
@@ -1033,6 +1033,8 @@ int rdma_init_qp_attr(struct rdma_cm_id *id, struct ib_qp_attr *qp_attr,
 		} else
 			ret = iw_cm_init_qp_attr(id_priv->cm_id.iw, qp_attr,
 						 qp_attr_mask);
+		qp_attr->port_num = id_priv->id.port_num;
+		*qp_attr_mask |= IB_QP_PORT;
 	} else
 		ret = -ENOSYS;
 
-- 
cgit v1.2.3