From f4434529003522d72b314d26d65b18c06ea9307c Mon Sep 17 00:00:00 2001
From: Maor Gottlieb <maorg@mellanox.com>
Date: Tue, 23 Jun 2020 14:30:36 +0300
Subject: RDMA: Add dedicated MR resource tracker function

In order to avoid double multiplexing of the resource when it is a MR, add
a dedicated callback function.

Link: https://lore.kernel.org/r/20200623113043.1228482-5-leon@kernel.org
Signed-off-by: Maor Gottlieb <maorg@mellanox.com>
Signed-off-by: Leon Romanovsky <leonro@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
---
 drivers/infiniband/core/device.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

(limited to 'drivers/infiniband/core/device.c')

diff --git a/drivers/infiniband/core/device.c b/drivers/infiniband/core/device.c
index 905a2beaf885..ffdf9787e7f6 100644
--- a/drivers/infiniband/core/device.c
+++ b/drivers/infiniband/core/device.c
@@ -2618,7 +2618,8 @@ void ib_set_device_ops(struct ib_device *dev, const struct ib_device_ops *ops)
 	SET_DEVICE_OP(dev_ops, drain_sq);
 	SET_DEVICE_OP(dev_ops, enable_driver);
 	SET_DEVICE_OP(dev_ops, fill_res_entry);
-	SET_DEVICE_OP(dev_ops, fill_stat_entry);
+	SET_DEVICE_OP(dev_ops, fill_res_mr_entry);
+	SET_DEVICE_OP(dev_ops, fill_stat_mr_entry);
 	SET_DEVICE_OP(dev_ops, get_dev_fw_str);
 	SET_DEVICE_OP(dev_ops, get_dma_mr);
 	SET_DEVICE_OP(dev_ops, get_hw_stats);
-- 
cgit v1.2.3-59-g8ed1b


From 9e2a187a93c395f573ed38b888ba4bd731e70622 Mon Sep 17 00:00:00 2001
From: Maor Gottlieb <maorg@mellanox.com>
Date: Tue, 23 Jun 2020 14:30:37 +0300
Subject: RDMA: Add a dedicated CQ resource tracker function

In order to avoid double multiplexing of the resource when it is a CQ, add
a dedicated callback function.

Link: https://lore.kernel.org/r/20200623113043.1228482-6-leon@kernel.org
Signed-off-by: Maor Gottlieb <maorg@mellanox.com>
Signed-off-by: Leon Romanovsky <leonro@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
---
 drivers/infiniband/core/device.c              |  1 +
 drivers/infiniband/core/nldev.c               |  5 ++---
 drivers/infiniband/hw/cxgb4/iw_cxgb4.h        |  1 +
 drivers/infiniband/hw/cxgb4/provider.c        |  1 +
 drivers/infiniband/hw/cxgb4/restrack.c        |  5 +----
 drivers/infiniband/hw/hns/hns_roce_device.h   |  4 ++--
 drivers/infiniband/hw/hns/hns_roce_main.c     |  2 +-
 drivers/infiniband/hw/hns/hns_roce_restrack.c | 14 ++------------
 include/rdma/ib_verbs.h                       |  1 +
 9 files changed, 12 insertions(+), 22 deletions(-)

(limited to 'drivers/infiniband/core/device.c')

diff --git a/drivers/infiniband/core/device.c b/drivers/infiniband/core/device.c
index ffdf9787e7f6..9eeac8cb600e 100644
--- a/drivers/infiniband/core/device.c
+++ b/drivers/infiniband/core/device.c
@@ -2617,6 +2617,7 @@ void ib_set_device_ops(struct ib_device *dev, const struct ib_device_ops *ops)
 	SET_DEVICE_OP(dev_ops, drain_rq);
 	SET_DEVICE_OP(dev_ops, drain_sq);
 	SET_DEVICE_OP(dev_ops, enable_driver);
+	SET_DEVICE_OP(dev_ops, fill_res_cq_entry);
 	SET_DEVICE_OP(dev_ops, fill_res_entry);
 	SET_DEVICE_OP(dev_ops, fill_res_mr_entry);
 	SET_DEVICE_OP(dev_ops, fill_stat_mr_entry);
diff --git a/drivers/infiniband/core/nldev.c b/drivers/infiniband/core/nldev.c
index a4f3f838d6fe..707f724db1dd 100644
--- a/drivers/infiniband/core/nldev.c
+++ b/drivers/infiniband/core/nldev.c
@@ -598,9 +598,8 @@ static int fill_res_cq_entry(struct sk_buff *msg, bool has_cap_net_admin,
 	if (fill_res_name_pid(msg, res))
 		goto err;
 
-	if (fill_res_entry(dev, msg, res))
-		goto err;
-
+	if (dev->ops.fill_res_cq_entry)
+		return dev->ops.fill_res_cq_entry(msg, cq);
 	return 0;
 
 err:	return -EMSGSIZE;
diff --git a/drivers/infiniband/hw/cxgb4/iw_cxgb4.h b/drivers/infiniband/hw/cxgb4/iw_cxgb4.h
index 5b9884ca2f5e..18a2c1a44dcc 100644
--- a/drivers/infiniband/hw/cxgb4/iw_cxgb4.h
+++ b/drivers/infiniband/hw/cxgb4/iw_cxgb4.h
@@ -1056,6 +1056,7 @@ struct c4iw_wr_wait *c4iw_alloc_wr_wait(gfp_t gfp);
 typedef int c4iw_restrack_func(struct sk_buff *msg,
 			       struct rdma_restrack_entry *res);
 int c4iw_fill_res_mr_entry(struct sk_buff *msg, struct ib_mr *ibmr);
+int c4iw_fill_res_cq_entry(struct sk_buff *msg, struct ib_cq *ibcq);
 extern c4iw_restrack_func *c4iw_restrack_funcs[RDMA_RESTRACK_MAX];
 
 #endif
diff --git a/drivers/infiniband/hw/cxgb4/provider.c b/drivers/infiniband/hw/cxgb4/provider.c
index 36eeb595d41c..d6b20aa314a0 100644
--- a/drivers/infiniband/hw/cxgb4/provider.c
+++ b/drivers/infiniband/hw/cxgb4/provider.c
@@ -485,6 +485,7 @@ static const struct ib_device_ops c4iw_dev_ops = {
 	.destroy_cq = c4iw_destroy_cq,
 	.destroy_qp = c4iw_destroy_qp,
 	.destroy_srq = c4iw_destroy_srq,
+	.fill_res_cq_entry = c4iw_fill_res_cq_entry,
 	.fill_res_entry = fill_res_entry,
 	.fill_res_mr_entry = c4iw_fill_res_mr_entry,
 	.get_dev_fw_str = get_dev_fw_str,
diff --git a/drivers/infiniband/hw/cxgb4/restrack.c b/drivers/infiniband/hw/cxgb4/restrack.c
index 9a5ca9192c1c..ead2cd08793d 100644
--- a/drivers/infiniband/hw/cxgb4/restrack.c
+++ b/drivers/infiniband/hw/cxgb4/restrack.c
@@ -372,10 +372,8 @@ err:
 	return -EMSGSIZE;
 }
 
-static int fill_res_cq_entry(struct sk_buff *msg,
-			     struct rdma_restrack_entry *res)
+int c4iw_fill_res_cq_entry(struct sk_buff *msg, struct ib_cq *ibcq)
 {
-	struct ib_cq *ibcq = container_of(res, struct ib_cq, res);
 	struct c4iw_cq *chp = to_c4iw_cq(ibcq);
 	struct nlattr *table_attr;
 	struct t4_cqe hwcqes[2];
@@ -494,5 +492,4 @@ err:
 c4iw_restrack_func *c4iw_restrack_funcs[RDMA_RESTRACK_MAX] = {
 	[RDMA_RESTRACK_QP]	= fill_res_qp_entry,
 	[RDMA_RESTRACK_CM_ID]	= fill_res_ep_entry,
-	[RDMA_RESTRACK_CQ]	= fill_res_cq_entry,
 };
diff --git a/drivers/infiniband/hw/hns/hns_roce_device.h b/drivers/infiniband/hw/hns/hns_roce_device.h
index a77fa6730b2d..a61f0c4d4dbb 100644
--- a/drivers/infiniband/hw/hns/hns_roce_device.h
+++ b/drivers/infiniband/hw/hns/hns_roce_device.h
@@ -1266,6 +1266,6 @@ void hns_roce_handle_device_err(struct hns_roce_dev *hr_dev);
 int hns_roce_init(struct hns_roce_dev *hr_dev);
 void hns_roce_exit(struct hns_roce_dev *hr_dev);
 
-int hns_roce_fill_res_entry(struct sk_buff *msg,
-			    struct rdma_restrack_entry *res);
+int hns_roce_fill_res_cq_entry(struct sk_buff *msg,
+			       struct ib_cq *ib_cq);
 #endif /* _HNS_ROCE_DEVICE_H */
diff --git a/drivers/infiniband/hw/hns/hns_roce_main.c b/drivers/infiniband/hw/hns/hns_roce_main.c
index 50763cf4fa3d..5907cfd878a6 100644
--- a/drivers/infiniband/hw/hns/hns_roce_main.c
+++ b/drivers/infiniband/hw/hns/hns_roce_main.c
@@ -428,7 +428,7 @@ static const struct ib_device_ops hns_roce_dev_ops = {
 	.destroy_ah = hns_roce_destroy_ah,
 	.destroy_cq = hns_roce_destroy_cq,
 	.disassociate_ucontext = hns_roce_disassociate_ucontext,
-	.fill_res_entry = hns_roce_fill_res_entry,
+	.fill_res_cq_entry = hns_roce_fill_res_cq_entry,
 	.get_dma_mr = hns_roce_get_dma_mr,
 	.get_link_layer = hns_roce_get_link_layer,
 	.get_port_immutable = hns_roce_port_immutable,
diff --git a/drivers/infiniband/hw/hns/hns_roce_restrack.c b/drivers/infiniband/hw/hns/hns_roce_restrack.c
index 06871731ac43..259444c0a630 100644
--- a/drivers/infiniband/hw/hns/hns_roce_restrack.c
+++ b/drivers/infiniband/hw/hns/hns_roce_restrack.c
@@ -76,10 +76,9 @@ err:
 	return -EMSGSIZE;
 }
 
-static int hns_roce_fill_res_cq_entry(struct sk_buff *msg,
-				      struct rdma_restrack_entry *res)
+int hns_roce_fill_res_cq_entry(struct sk_buff *msg,
+			       struct ib_cq *ib_cq)
 {
-	struct ib_cq *ib_cq = container_of(res, struct ib_cq, res);
 	struct hns_roce_dev *hr_dev = to_hr_dev(ib_cq->device);
 	struct hns_roce_cq *hr_cq = to_hr_cq(ib_cq);
 	struct hns_roce_v2_cq_context *context;
@@ -119,12 +118,3 @@ err:
 	kfree(context);
 	return ret;
 }
-
-int hns_roce_fill_res_entry(struct sk_buff *msg,
-			    struct rdma_restrack_entry *res)
-{
-	if (res->type == RDMA_RESTRACK_CQ)
-		return hns_roce_fill_res_cq_entry(msg, res);
-
-	return 0;
-}
diff --git a/include/rdma/ib_verbs.h b/include/rdma/ib_verbs.h
index 117a0e802aa1..097b1d497d5f 100644
--- a/include/rdma/ib_verbs.h
+++ b/include/rdma/ib_verbs.h
@@ -2584,6 +2584,7 @@ struct ib_device_ops {
 	int (*fill_res_entry)(struct sk_buff *msg,
 			      struct rdma_restrack_entry *entry);
 	int (*fill_res_mr_entry)(struct sk_buff *msg, struct ib_mr *ibmr);
+	int (*fill_res_cq_entry)(struct sk_buff *msg, struct ib_cq *ibcq);
 
 	/* Device lifecycle callbacks */
 	/*
-- 
cgit v1.2.3-59-g8ed1b


From 5cc34116ccec60032dbaa92768f41e95ce2d8ec7 Mon Sep 17 00:00:00 2001
From: Maor Gottlieb <maorg@mellanox.com>
Date: Tue, 23 Jun 2020 14:30:38 +0300
Subject: RDMA: Add dedicated QP resource tracker function

In order to avoid double multiplexing of the resource when it is a QP, add
a dedicated callback function.

Link: https://lore.kernel.org/r/20200623113043.1228482-7-leon@kernel.org
Signed-off-by: Maor Gottlieb <maorg@mellanox.com>
Signed-off-by: Leon Romanovsky <leonro@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
---
 drivers/infiniband/core/device.c       | 1 +
 drivers/infiniband/core/nldev.c        | 5 ++---
 drivers/infiniband/hw/cxgb4/iw_cxgb4.h | 1 +
 drivers/infiniband/hw/cxgb4/restrack.c | 5 +----
 include/rdma/ib_verbs.h                | 1 +
 5 files changed, 6 insertions(+), 7 deletions(-)

(limited to 'drivers/infiniband/core/device.c')

diff --git a/drivers/infiniband/core/device.c b/drivers/infiniband/core/device.c
index 9eeac8cb600e..f94989274df5 100644
--- a/drivers/infiniband/core/device.c
+++ b/drivers/infiniband/core/device.c
@@ -2620,6 +2620,7 @@ void ib_set_device_ops(struct ib_device *dev, const struct ib_device_ops *ops)
 	SET_DEVICE_OP(dev_ops, fill_res_cq_entry);
 	SET_DEVICE_OP(dev_ops, fill_res_entry);
 	SET_DEVICE_OP(dev_ops, fill_res_mr_entry);
+	SET_DEVICE_OP(dev_ops, fill_res_qp_entry);
 	SET_DEVICE_OP(dev_ops, fill_stat_mr_entry);
 	SET_DEVICE_OP(dev_ops, get_dev_fw_str);
 	SET_DEVICE_OP(dev_ops, get_dma_mr);
diff --git a/drivers/infiniband/core/nldev.c b/drivers/infiniband/core/nldev.c
index 707f724db1dd..79d0980a75e0 100644
--- a/drivers/infiniband/core/nldev.c
+++ b/drivers/infiniband/core/nldev.c
@@ -507,9 +507,8 @@ static int fill_res_qp_entry(struct sk_buff *msg, bool has_cap_net_admin,
 	if (fill_res_name_pid(msg, res))
 		goto err;
 
-	if (fill_res_entry(dev, msg, res))
-		goto err;
-
+	if (dev->ops.fill_res_qp_entry)
+		return dev->ops.fill_res_qp_entry(msg, qp);
 	return 0;
 
 err:	return -EMSGSIZE;
diff --git a/drivers/infiniband/hw/cxgb4/iw_cxgb4.h b/drivers/infiniband/hw/cxgb4/iw_cxgb4.h
index 18a2c1a44dcc..c84aa7c937f1 100644
--- a/drivers/infiniband/hw/cxgb4/iw_cxgb4.h
+++ b/drivers/infiniband/hw/cxgb4/iw_cxgb4.h
@@ -1057,6 +1057,7 @@ typedef int c4iw_restrack_func(struct sk_buff *msg,
 			       struct rdma_restrack_entry *res);
 int c4iw_fill_res_mr_entry(struct sk_buff *msg, struct ib_mr *ibmr);
 int c4iw_fill_res_cq_entry(struct sk_buff *msg, struct ib_cq *ibcq);
+int c4iw_fill_res_qp_entry(struct sk_buff *msg, struct ib_qp *ibqp);
 extern c4iw_restrack_func *c4iw_restrack_funcs[RDMA_RESTRACK_MAX];
 
 #endif
diff --git a/drivers/infiniband/hw/cxgb4/restrack.c b/drivers/infiniband/hw/cxgb4/restrack.c
index ead2cd08793d..5144d3b67293 100644
--- a/drivers/infiniband/hw/cxgb4/restrack.c
+++ b/drivers/infiniband/hw/cxgb4/restrack.c
@@ -134,10 +134,8 @@ err:
 	return -EMSGSIZE;
 }
 
-static int fill_res_qp_entry(struct sk_buff *msg,
-			     struct rdma_restrack_entry *res)
+int c4iw_fill_res_qp_entry(struct sk_buff *msg, struct ib_qp *ibqp)
 {
-	struct ib_qp *ibqp = container_of(res, struct ib_qp, res);
 	struct t4_swsqe *fsp = NULL, *lsp = NULL;
 	struct c4iw_qp *qhp = to_c4iw_qp(ibqp);
 	u16 first_sq_idx = 0, last_sq_idx = 0;
@@ -490,6 +488,5 @@ err:
 }
 
 c4iw_restrack_func *c4iw_restrack_funcs[RDMA_RESTRACK_MAX] = {
-	[RDMA_RESTRACK_QP]	= fill_res_qp_entry,
 	[RDMA_RESTRACK_CM_ID]	= fill_res_ep_entry,
 };
diff --git a/include/rdma/ib_verbs.h b/include/rdma/ib_verbs.h
index 097b1d497d5f..4e8519ac7363 100644
--- a/include/rdma/ib_verbs.h
+++ b/include/rdma/ib_verbs.h
@@ -2585,6 +2585,7 @@ struct ib_device_ops {
 			      struct rdma_restrack_entry *entry);
 	int (*fill_res_mr_entry)(struct sk_buff *msg, struct ib_mr *ibmr);
 	int (*fill_res_cq_entry)(struct sk_buff *msg, struct ib_cq *ibcq);
+	int (*fill_res_qp_entry)(struct sk_buff *msg, struct ib_qp *ibqp);
 
 	/* Device lifecycle callbacks */
 	/*
-- 
cgit v1.2.3-59-g8ed1b


From 211cd9459fdabe9f556e539966f50825853bf262 Mon Sep 17 00:00:00 2001
From: Maor Gottlieb <maorg@mellanox.com>
Date: Tue, 23 Jun 2020 14:30:39 +0300
Subject: RDMA: Add dedicated CM_ID resource tracker function

In order to avoid double multiplexing of the resource when it is a cm id,
add a dedicated callback function. In addition remove fill_res_entry which
is not used anymore.

Link: https://lore.kernel.org/r/20200623113043.1228482-8-leon@kernel.org
Signed-off-by: Maor Gottlieb <maorg@mellanox.com>
Signed-off-by: Leon Romanovsky <leonro@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
---
 drivers/infiniband/core/device.c       |  2 +-
 drivers/infiniband/core/nldev.c        | 13 ++-----------
 drivers/infiniband/hw/cxgb4/iw_cxgb4.h |  4 +---
 drivers/infiniband/hw/cxgb4/provider.c |  9 +--------
 drivers/infiniband/hw/cxgb4/restrack.c |  9 ++-------
 include/rdma/ib_verbs.h                |  4 ++--
 6 files changed, 9 insertions(+), 32 deletions(-)

(limited to 'drivers/infiniband/core/device.c')

diff --git a/drivers/infiniband/core/device.c b/drivers/infiniband/core/device.c
index f94989274df5..cbe95e729cf1 100644
--- a/drivers/infiniband/core/device.c
+++ b/drivers/infiniband/core/device.c
@@ -2617,8 +2617,8 @@ void ib_set_device_ops(struct ib_device *dev, const struct ib_device_ops *ops)
 	SET_DEVICE_OP(dev_ops, drain_rq);
 	SET_DEVICE_OP(dev_ops, drain_sq);
 	SET_DEVICE_OP(dev_ops, enable_driver);
+	SET_DEVICE_OP(dev_ops, fill_res_cm_id_entry);
 	SET_DEVICE_OP(dev_ops, fill_res_cq_entry);
-	SET_DEVICE_OP(dev_ops, fill_res_entry);
 	SET_DEVICE_OP(dev_ops, fill_res_mr_entry);
 	SET_DEVICE_OP(dev_ops, fill_res_qp_entry);
 	SET_DEVICE_OP(dev_ops, fill_stat_mr_entry);
diff --git a/drivers/infiniband/core/nldev.c b/drivers/infiniband/core/nldev.c
index 79d0980a75e0..394e307c342c 100644
--- a/drivers/infiniband/core/nldev.c
+++ b/drivers/infiniband/core/nldev.c
@@ -446,14 +446,6 @@ static int fill_res_name_pid(struct sk_buff *msg,
 	return err ? -EMSGSIZE : 0;
 }
 
-static bool fill_res_entry(struct ib_device *dev, struct sk_buff *msg,
-			   struct rdma_restrack_entry *res)
-{
-	if (!dev->ops.fill_res_entry)
-		return false;
-	return dev->ops.fill_res_entry(msg, res);
-}
-
 static int fill_res_qp_entry(struct sk_buff *msg, bool has_cap_net_admin,
 			     struct rdma_restrack_entry *res, uint32_t port)
 {
@@ -559,9 +551,8 @@ static int fill_res_cm_id_entry(struct sk_buff *msg, bool has_cap_net_admin,
 	if (fill_res_name_pid(msg, res))
 		goto err;
 
-	if (fill_res_entry(dev, msg, res))
-		goto err;
-
+	if (dev->ops.fill_res_cm_id_entry)
+		return dev->ops.fill_res_cm_id_entry(msg, cm_id);
 	return 0;
 
 err: return -EMSGSIZE;
diff --git a/drivers/infiniband/hw/cxgb4/iw_cxgb4.h b/drivers/infiniband/hw/cxgb4/iw_cxgb4.h
index c84aa7c937f1..27da0705c88a 100644
--- a/drivers/infiniband/hw/cxgb4/iw_cxgb4.h
+++ b/drivers/infiniband/hw/cxgb4/iw_cxgb4.h
@@ -1053,11 +1053,9 @@ int c4iw_post_srq_recv(struct ib_srq *ibsrq, const struct ib_recv_wr *wr,
 		       const struct ib_recv_wr **bad_wr);
 struct c4iw_wr_wait *c4iw_alloc_wr_wait(gfp_t gfp);
 
-typedef int c4iw_restrack_func(struct sk_buff *msg,
-			       struct rdma_restrack_entry *res);
 int c4iw_fill_res_mr_entry(struct sk_buff *msg, struct ib_mr *ibmr);
 int c4iw_fill_res_cq_entry(struct sk_buff *msg, struct ib_cq *ibcq);
 int c4iw_fill_res_qp_entry(struct sk_buff *msg, struct ib_qp *ibqp);
-extern c4iw_restrack_func *c4iw_restrack_funcs[RDMA_RESTRACK_MAX];
+int c4iw_fill_res_cm_id_entry(struct sk_buff *msg, struct rdma_cm_id *cm_id);
 
 #endif
diff --git a/drivers/infiniband/hw/cxgb4/provider.c b/drivers/infiniband/hw/cxgb4/provider.c
index d6b20aa314a0..1d3ff59e4060 100644
--- a/drivers/infiniband/hw/cxgb4/provider.c
+++ b/drivers/infiniband/hw/cxgb4/provider.c
@@ -458,13 +458,6 @@ static void get_dev_fw_str(struct ib_device *dev, char *str)
 		 FW_HDR_FW_VER_BUILD_G(c4iw_dev->rdev.lldi.fw_vers));
 }
 
-static int fill_res_entry(struct sk_buff *msg, struct rdma_restrack_entry *res)
-{
-	return (res->type < ARRAY_SIZE(c4iw_restrack_funcs) &&
-		c4iw_restrack_funcs[res->type]) ?
-		c4iw_restrack_funcs[res->type](msg, res) : 0;
-}
-
 static const struct ib_device_ops c4iw_dev_ops = {
 	.owner = THIS_MODULE,
 	.driver_id = RDMA_DRIVER_CXGB4,
@@ -486,7 +479,7 @@ static const struct ib_device_ops c4iw_dev_ops = {
 	.destroy_qp = c4iw_destroy_qp,
 	.destroy_srq = c4iw_destroy_srq,
 	.fill_res_cq_entry = c4iw_fill_res_cq_entry,
-	.fill_res_entry = fill_res_entry,
+	.fill_res_cm_id_entry = c4iw_fill_res_cm_id_entry,
 	.fill_res_mr_entry = c4iw_fill_res_mr_entry,
 	.get_dev_fw_str = get_dev_fw_str,
 	.get_dma_mr = c4iw_get_dma_mr,
diff --git a/drivers/infiniband/hw/cxgb4/restrack.c b/drivers/infiniband/hw/cxgb4/restrack.c
index 5144d3b67293..b32e6516d65f 100644
--- a/drivers/infiniband/hw/cxgb4/restrack.c
+++ b/drivers/infiniband/hw/cxgb4/restrack.c
@@ -193,10 +193,9 @@ union union_ep {
 	struct c4iw_ep ep;
 };
 
-static int fill_res_ep_entry(struct sk_buff *msg,
-			     struct rdma_restrack_entry *res)
+int c4iw_fill_res_cm_id_entry(struct sk_buff *msg,
+			      struct rdma_cm_id *cm_id)
 {
-	struct rdma_cm_id *cm_id = rdma_res_to_id(res);
 	struct nlattr *table_attr;
 	struct c4iw_ep_common *epcp;
 	struct c4iw_listen_ep *listen_ep = NULL;
@@ -486,7 +485,3 @@ err_cancel_table:
 err:
 	return -EMSGSIZE;
 }
-
-c4iw_restrack_func *c4iw_restrack_funcs[RDMA_RESTRACK_MAX] = {
-	[RDMA_RESTRACK_CM_ID]	= fill_res_ep_entry,
-};
diff --git a/include/rdma/ib_verbs.h b/include/rdma/ib_verbs.h
index 4e8519ac7363..9127cffafccd 100644
--- a/include/rdma/ib_verbs.h
+++ b/include/rdma/ib_verbs.h
@@ -75,6 +75,7 @@ struct ib_umem_odp;
 struct ib_uqp_object;
 struct ib_usrq_object;
 struct ib_uwq_object;
+struct rdma_cm_id;
 
 extern struct workqueue_struct *ib_wq;
 extern struct workqueue_struct *ib_comp_wq;
@@ -2581,11 +2582,10 @@ struct ib_device_ops {
 	/**
 	 * Allows rdma drivers to add their own restrack attributes.
 	 */
-	int (*fill_res_entry)(struct sk_buff *msg,
-			      struct rdma_restrack_entry *entry);
 	int (*fill_res_mr_entry)(struct sk_buff *msg, struct ib_mr *ibmr);
 	int (*fill_res_cq_entry)(struct sk_buff *msg, struct ib_cq *ibcq);
 	int (*fill_res_qp_entry)(struct sk_buff *msg, struct ib_qp *ibqp);
+	int (*fill_res_cm_id_entry)(struct sk_buff *msg, struct rdma_cm_id *id);
 
 	/* Device lifecycle callbacks */
 	/*
-- 
cgit v1.2.3-59-g8ed1b


From 65959522f8060659e308977f09f3eb7b7af5e43f Mon Sep 17 00:00:00 2001
From: Maor Gottlieb <maorg@mellanox.com>
Date: Tue, 23 Jun 2020 14:30:40 +0300
Subject: RDMA: Add support to dump resource tracker in RAW format

Add support to get resource dump in raw format. It enable drivers to
return the entire device specific QP/CQ/MR context without a need from the
driver to set each field separately.

The raw query returns only the device specific data, general data is still
returned by using the existing queries.

Example:

$ rdma res show mr dev mlx5_1 mrn 2 -r -j
[{"ifindex":7,"ifname":"mlx5_1",
"data":[0,4,255,254,0,0,0,0,0,0,0,0,16,28,0,216,...]}]

Link: https://lore.kernel.org/r/20200623113043.1228482-9-leon@kernel.org
Signed-off-by: Maor Gottlieb <maorg@mellanox.com>
Signed-off-by: Leon Romanovsky <leonro@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
---
 drivers/infiniband/core/device.c |   3 +
 drivers/infiniband/core/nldev.c  | 180 +++++++++++++++++++++++++--------------
 include/rdma/ib_verbs.h          |   3 +
 include/uapi/rdma/rdma_netlink.h |   8 ++
 4 files changed, 132 insertions(+), 62 deletions(-)

(limited to 'drivers/infiniband/core/device.c')

diff --git a/drivers/infiniband/core/device.c b/drivers/infiniband/core/device.c
index cbe95e729cf1..1335ed1f1e4a 100644
--- a/drivers/infiniband/core/device.c
+++ b/drivers/infiniband/core/device.c
@@ -2619,8 +2619,11 @@ void ib_set_device_ops(struct ib_device *dev, const struct ib_device_ops *ops)
 	SET_DEVICE_OP(dev_ops, enable_driver);
 	SET_DEVICE_OP(dev_ops, fill_res_cm_id_entry);
 	SET_DEVICE_OP(dev_ops, fill_res_cq_entry);
+	SET_DEVICE_OP(dev_ops, fill_res_cq_entry_raw);
 	SET_DEVICE_OP(dev_ops, fill_res_mr_entry);
+	SET_DEVICE_OP(dev_ops, fill_res_mr_entry_raw);
 	SET_DEVICE_OP(dev_ops, fill_res_qp_entry);
+	SET_DEVICE_OP(dev_ops, fill_res_qp_entry_raw);
 	SET_DEVICE_OP(dev_ops, fill_stat_mr_entry);
 	SET_DEVICE_OP(dev_ops, get_dev_fw_str);
 	SET_DEVICE_OP(dev_ops, get_dma_mr);
diff --git a/drivers/infiniband/core/nldev.c b/drivers/infiniband/core/nldev.c
index 394e307c342c..1051b5622b62 100644
--- a/drivers/infiniband/core/nldev.c
+++ b/drivers/infiniband/core/nldev.c
@@ -114,6 +114,7 @@ static const struct nla_policy nldev_policy[RDMA_NLDEV_ATTR_MAX] = {
 	[RDMA_NLDEV_ATTR_RES_PS]		= { .type = NLA_U32 },
 	[RDMA_NLDEV_ATTR_RES_QP]		= { .type = NLA_NESTED },
 	[RDMA_NLDEV_ATTR_RES_QP_ENTRY]		= { .type = NLA_NESTED },
+	[RDMA_NLDEV_ATTR_RES_RAW]		= { .type = NLA_BINARY },
 	[RDMA_NLDEV_ATTR_RES_RKEY]		= { .type = NLA_U32 },
 	[RDMA_NLDEV_ATTR_RES_RQPN]		= { .type = NLA_U32 },
 	[RDMA_NLDEV_ATTR_RES_RQ_PSN]		= { .type = NLA_U32 },
@@ -446,11 +447,11 @@ static int fill_res_name_pid(struct sk_buff *msg,
 	return err ? -EMSGSIZE : 0;
 }
 
-static int fill_res_qp_entry(struct sk_buff *msg, bool has_cap_net_admin,
-			     struct rdma_restrack_entry *res, uint32_t port)
+static int fill_res_qp_entry_query(struct sk_buff *msg,
+				   struct rdma_restrack_entry *res,
+				   struct ib_device *dev,
+				   struct ib_qp *qp)
 {
-	struct ib_qp *qp = container_of(res, struct ib_qp, res);
-	struct ib_device *dev = qp->device;
 	struct ib_qp_init_attr qp_init_attr;
 	struct ib_qp_attr qp_attr;
 	int ret;
@@ -459,16 +460,6 @@ static int fill_res_qp_entry(struct sk_buff *msg, bool has_cap_net_admin,
 	if (ret)
 		return ret;
 
-	if (port && port != qp_attr.port_num)
-		return -EAGAIN;
-
-	/* In create_qp() port is not set yet */
-	if (qp_attr.port_num &&
-	    nla_put_u32(msg, RDMA_NLDEV_ATTR_PORT_INDEX, qp_attr.port_num))
-		goto err;
-
-	if (nla_put_u32(msg, RDMA_NLDEV_ATTR_RES_LQPN, qp->qp_num))
-		goto err;
 	if (qp->qp_type == IB_QPT_RC || qp->qp_type == IB_QPT_UC) {
 		if (nla_put_u32(msg, RDMA_NLDEV_ATTR_RES_RQPN,
 				qp_attr.dest_qp_num))
@@ -492,13 +483,6 @@ static int fill_res_qp_entry(struct sk_buff *msg, bool has_cap_net_admin,
 	if (nla_put_u8(msg, RDMA_NLDEV_ATTR_RES_STATE, qp_attr.qp_state))
 		goto err;
 
-	if (!rdma_is_kernel_res(res) &&
-	    nla_put_u32(msg, RDMA_NLDEV_ATTR_RES_PDN, qp->pd->res.id))
-		goto err;
-
-	if (fill_res_name_pid(msg, res))
-		goto err;
-
 	if (dev->ops.fill_res_qp_entry)
 		return dev->ops.fill_res_qp_entry(msg, qp);
 	return 0;
@@ -506,6 +490,48 @@ static int fill_res_qp_entry(struct sk_buff *msg, bool has_cap_net_admin,
 err:	return -EMSGSIZE;
 }
 
+static int fill_res_qp_entry(struct sk_buff *msg, bool has_cap_net_admin,
+			     struct rdma_restrack_entry *res, uint32_t port)
+{
+	struct ib_qp *qp = container_of(res, struct ib_qp, res);
+	struct ib_device *dev = qp->device;
+	int ret;
+
+	if (port && port != qp->port)
+		return -EAGAIN;
+
+	/* In create_qp() port is not set yet */
+	if (qp->port && nla_put_u32(msg, RDMA_NLDEV_ATTR_PORT_INDEX, qp->port))
+		return -EINVAL;
+
+	ret = nla_put_u32(msg, RDMA_NLDEV_ATTR_RES_LQPN, qp->qp_num);
+	if (ret)
+		return -EMSGSIZE;
+
+	if (!rdma_is_kernel_res(res) &&
+	    nla_put_u32(msg, RDMA_NLDEV_ATTR_RES_PDN, qp->pd->res.id))
+		return -EMSGSIZE;
+
+	ret = fill_res_name_pid(msg, res);
+	if (ret)
+		return -EMSGSIZE;
+
+	return fill_res_qp_entry_query(msg, res, dev, qp);
+}
+
+static int fill_res_qp_raw_entry(struct sk_buff *msg, bool has_cap_net_admin,
+				 struct rdma_restrack_entry *res, uint32_t port)
+{
+	struct ib_qp *qp = container_of(res, struct ib_qp, res);
+	struct ib_device *dev = qp->device;
+
+	if (port && port != qp->port)
+		return -EAGAIN;
+	if (!dev->ops.fill_res_qp_entry_raw)
+		return -EINVAL;
+	return dev->ops.fill_res_qp_entry_raw(msg, qp);
+}
+
 static int fill_res_cm_id_entry(struct sk_buff *msg, bool has_cap_net_admin,
 				struct rdma_restrack_entry *res, uint32_t port)
 {
@@ -565,34 +591,42 @@ static int fill_res_cq_entry(struct sk_buff *msg, bool has_cap_net_admin,
 	struct ib_device *dev = cq->device;
 
 	if (nla_put_u32(msg, RDMA_NLDEV_ATTR_RES_CQE, cq->cqe))
-		goto err;
+		return -EMSGSIZE;
 	if (nla_put_u64_64bit(msg, RDMA_NLDEV_ATTR_RES_USECNT,
 			      atomic_read(&cq->usecnt), RDMA_NLDEV_ATTR_PAD))
-		goto err;
+		return -EMSGSIZE;
 
 	/* Poll context is only valid for kernel CQs */
 	if (rdma_is_kernel_res(res) &&
 	    nla_put_u8(msg, RDMA_NLDEV_ATTR_RES_POLL_CTX, cq->poll_ctx))
-		goto err;
+		return -EMSGSIZE;
 
 	if (nla_put_u8(msg, RDMA_NLDEV_ATTR_DEV_DIM, (cq->dim != NULL)))
-		goto err;
+		return -EMSGSIZE;
 
 	if (nla_put_u32(msg, RDMA_NLDEV_ATTR_RES_CQN, res->id))
-		goto err;
+		return -EMSGSIZE;
 	if (!rdma_is_kernel_res(res) &&
 	    nla_put_u32(msg, RDMA_NLDEV_ATTR_RES_CTXN,
 			cq->uobject->uevent.uobject.context->res.id))
-		goto err;
+		return -EMSGSIZE;
 
 	if (fill_res_name_pid(msg, res))
-		goto err;
+		return -EMSGSIZE;
 
-	if (dev->ops.fill_res_cq_entry)
-		return dev->ops.fill_res_cq_entry(msg, cq);
-	return 0;
+	return (dev->ops.fill_res_cq_entry) ?
+		dev->ops.fill_res_cq_entry(msg, cq) : 0;
+}
 
-err:	return -EMSGSIZE;
+static int fill_res_cq_raw_entry(struct sk_buff *msg, bool has_cap_net_admin,
+				 struct rdma_restrack_entry *res, uint32_t port)
+{
+	struct ib_cq *cq = container_of(res, struct ib_cq, res);
+	struct ib_device *dev = cq->device;
+
+	if (!dev->ops.fill_res_cq_entry_raw)
+		return -EINVAL;
+	return dev->ops.fill_res_cq_entry_raw(msg, cq);
 }
 
 static int fill_res_mr_entry(struct sk_buff *msg, bool has_cap_net_admin,
@@ -603,30 +637,39 @@ static int fill_res_mr_entry(struct sk_buff *msg, bool has_cap_net_admin,
 
 	if (has_cap_net_admin) {
 		if (nla_put_u32(msg, RDMA_NLDEV_ATTR_RES_RKEY, mr->rkey))
-			goto err;
+			return -EMSGSIZE;
 		if (nla_put_u32(msg, RDMA_NLDEV_ATTR_RES_LKEY, mr->lkey))
-			goto err;
+			return -EMSGSIZE;
 	}
 
 	if (nla_put_u64_64bit(msg, RDMA_NLDEV_ATTR_RES_MRLEN, mr->length,
 			      RDMA_NLDEV_ATTR_PAD))
-		goto err;
+		return -EMSGSIZE;
 
 	if (nla_put_u32(msg, RDMA_NLDEV_ATTR_RES_MRN, res->id))
-		goto err;
+		return -EMSGSIZE;
 
 	if (!rdma_is_kernel_res(res) &&
 	    nla_put_u32(msg, RDMA_NLDEV_ATTR_RES_PDN, mr->pd->res.id))
-		goto err;
+		return -EMSGSIZE;
 
 	if (fill_res_name_pid(msg, res))
-		goto err;
+		return -EMSGSIZE;
 
-	if (dev->ops.fill_res_mr_entry)
-		return dev->ops.fill_res_mr_entry(msg, mr);
-	return 0;
+	return (dev->ops.fill_res_mr_entry) ?
+		       dev->ops.fill_res_mr_entry(msg, mr) :
+		       0;
+}
 
-err:	return -EMSGSIZE;
+static int fill_res_mr_raw_entry(struct sk_buff *msg, bool has_cap_net_admin,
+				 struct rdma_restrack_entry *res, uint32_t port)
+{
+	struct ib_mr *mr = container_of(res, struct ib_mr, res);
+	struct ib_device *dev = mr->pd->device;
+
+	if (!dev->ops.fill_res_mr_entry_raw)
+		return -EINVAL;
+	return dev->ops.fill_res_mr_entry_raw(msg, mr);
 }
 
 static int fill_res_pd_entry(struct sk_buff *msg, bool has_cap_net_admin,
@@ -1149,7 +1192,6 @@ static int nldev_res_get_dumpit(struct sk_buff *skb,
 
 struct nldev_fill_res_entry {
 	enum rdma_nldev_attr nldev_attr;
-	enum rdma_nldev_command nldev_cmd;
 	u8 flags;
 	u32 entry;
 	u32 id;
@@ -1161,40 +1203,34 @@ enum nldev_res_flags {
 
 static const struct nldev_fill_res_entry fill_entries[RDMA_RESTRACK_MAX] = {
 	[RDMA_RESTRACK_QP] = {
-		.nldev_cmd = RDMA_NLDEV_CMD_RES_QP_GET,
 		.nldev_attr = RDMA_NLDEV_ATTR_RES_QP,
 		.entry = RDMA_NLDEV_ATTR_RES_QP_ENTRY,
 		.id = RDMA_NLDEV_ATTR_RES_LQPN,
 	},
 	[RDMA_RESTRACK_CM_ID] = {
-		.nldev_cmd = RDMA_NLDEV_CMD_RES_CM_ID_GET,
 		.nldev_attr = RDMA_NLDEV_ATTR_RES_CM_ID,
 		.entry = RDMA_NLDEV_ATTR_RES_CM_ID_ENTRY,
 		.id = RDMA_NLDEV_ATTR_RES_CM_IDN,
 	},
 	[RDMA_RESTRACK_CQ] = {
-		.nldev_cmd = RDMA_NLDEV_CMD_RES_CQ_GET,
 		.nldev_attr = RDMA_NLDEV_ATTR_RES_CQ,
 		.flags = NLDEV_PER_DEV,
 		.entry = RDMA_NLDEV_ATTR_RES_CQ_ENTRY,
 		.id = RDMA_NLDEV_ATTR_RES_CQN,
 	},
 	[RDMA_RESTRACK_MR] = {
-		.nldev_cmd = RDMA_NLDEV_CMD_RES_MR_GET,
 		.nldev_attr = RDMA_NLDEV_ATTR_RES_MR,
 		.flags = NLDEV_PER_DEV,
 		.entry = RDMA_NLDEV_ATTR_RES_MR_ENTRY,
 		.id = RDMA_NLDEV_ATTR_RES_MRN,
 	},
 	[RDMA_RESTRACK_PD] = {
-		.nldev_cmd = RDMA_NLDEV_CMD_RES_PD_GET,
 		.nldev_attr = RDMA_NLDEV_ATTR_RES_PD,
 		.flags = NLDEV_PER_DEV,
 		.entry = RDMA_NLDEV_ATTR_RES_PD_ENTRY,
 		.id = RDMA_NLDEV_ATTR_RES_PDN,
 	},
 	[RDMA_RESTRACK_COUNTER] = {
-		.nldev_cmd = RDMA_NLDEV_CMD_STAT_GET,
 		.nldev_attr = RDMA_NLDEV_ATTR_STAT_COUNTER,
 		.entry = RDMA_NLDEV_ATTR_STAT_COUNTER_ENTRY,
 		.id = RDMA_NLDEV_ATTR_STAT_COUNTER_ID,
@@ -1253,7 +1289,8 @@ static int res_get_common_doit(struct sk_buff *skb, struct nlmsghdr *nlh,
 	}
 
 	nlh = nlmsg_put(msg, NETLINK_CB(skb).portid, nlh->nlmsg_seq,
-			RDMA_NL_GET_TYPE(RDMA_NL_NLDEV, fe->nldev_cmd),
+			RDMA_NL_GET_TYPE(RDMA_NL_NLDEV,
+					 RDMA_NL_GET_OP(nlh->nlmsg_type)),
 			0, 0);
 
 	if (fill_nldev_handle(msg, device)) {
@@ -1331,7 +1368,8 @@ static int res_get_common_dumpit(struct sk_buff *skb,
 	}
 
 	nlh = nlmsg_put(skb, NETLINK_CB(cb->skb).portid, cb->nlh->nlmsg_seq,
-			RDMA_NL_GET_TYPE(RDMA_NL_NLDEV, fe->nldev_cmd),
+			RDMA_NL_GET_TYPE(RDMA_NL_NLDEV,
+					 RDMA_NL_GET_OP(cb->nlh->nlmsg_type)),
 			0, NLM_F_MULTI);
 
 	if (fill_nldev_handle(skb, device)) {
@@ -1413,26 +1451,29 @@ err_index:
 	return ret;
 }
 
-#define RES_GET_FUNCS(name, type)					       \
-	static int nldev_res_get_##name##_dumpit(struct sk_buff *skb,	       \
+#define RES_GET_FUNCS(name, type)                                              \
+	static int nldev_res_get_##name##_dumpit(struct sk_buff *skb,          \
 						 struct netlink_callback *cb)  \
-	{								       \
-		return res_get_common_dumpit(skb, cb, type,		       \
-					     fill_res_##name##_entry);	       \
-	}								       \
-	static int nldev_res_get_##name##_doit(struct sk_buff *skb,	       \
-					       struct nlmsghdr *nlh,	       \
+	{                                                                      \
+		return res_get_common_dumpit(skb, cb, type,                    \
+					     fill_res_##name##_entry);         \
+	}                                                                      \
+	static int nldev_res_get_##name##_doit(struct sk_buff *skb,            \
+					       struct nlmsghdr *nlh,           \
 					       struct netlink_ext_ack *extack) \
-	{								       \
-		return res_get_common_doit(skb, nlh, extack, type,	       \
-					   fill_res_##name##_entry);	       \
+	{                                                                      \
+		return res_get_common_doit(skb, nlh, extack, type,             \
+					   fill_res_##name##_entry);           \
 	}
 
 RES_GET_FUNCS(qp, RDMA_RESTRACK_QP);
+RES_GET_FUNCS(qp_raw, RDMA_RESTRACK_QP);
 RES_GET_FUNCS(cm_id, RDMA_RESTRACK_CM_ID);
 RES_GET_FUNCS(cq, RDMA_RESTRACK_CQ);
+RES_GET_FUNCS(cq_raw, RDMA_RESTRACK_CQ);
 RES_GET_FUNCS(pd, RDMA_RESTRACK_PD);
 RES_GET_FUNCS(mr, RDMA_RESTRACK_MR);
+RES_GET_FUNCS(mr_raw, RDMA_RESTRACK_MR);
 RES_GET_FUNCS(counter, RDMA_RESTRACK_COUNTER);
 
 static LIST_HEAD(link_ops);
@@ -2117,6 +2158,21 @@ static const struct rdma_nl_cbs nldev_cb_table[RDMA_NLDEV_NUM_OPS] = {
 		.doit = nldev_stat_del_doit,
 		.flags = RDMA_NL_ADMIN_PERM,
 	},
+	[RDMA_NLDEV_CMD_RES_QP_GET_RAW] = {
+		.doit = nldev_res_get_qp_raw_doit,
+		.dump = nldev_res_get_qp_raw_dumpit,
+		.flags = RDMA_NL_ADMIN_PERM,
+	},
+	[RDMA_NLDEV_CMD_RES_CQ_GET_RAW] = {
+		.doit = nldev_res_get_cq_raw_doit,
+		.dump = nldev_res_get_cq_raw_dumpit,
+		.flags = RDMA_NL_ADMIN_PERM,
+	},
+	[RDMA_NLDEV_CMD_RES_MR_GET_RAW] = {
+		.doit = nldev_res_get_mr_raw_doit,
+		.dump = nldev_res_get_mr_raw_dumpit,
+		.flags = RDMA_NL_ADMIN_PERM,
+	},
 };
 
 void __init nldev_init(void)
diff --git a/include/rdma/ib_verbs.h b/include/rdma/ib_verbs.h
index 9127cffafccd..77106ff3cd26 100644
--- a/include/rdma/ib_verbs.h
+++ b/include/rdma/ib_verbs.h
@@ -2583,8 +2583,11 @@ struct ib_device_ops {
 	 * Allows rdma drivers to add their own restrack attributes.
 	 */
 	int (*fill_res_mr_entry)(struct sk_buff *msg, struct ib_mr *ibmr);
+	int (*fill_res_mr_entry_raw)(struct sk_buff *msg, struct ib_mr *ibmr);
 	int (*fill_res_cq_entry)(struct sk_buff *msg, struct ib_cq *ibcq);
+	int (*fill_res_cq_entry_raw)(struct sk_buff *msg, struct ib_cq *ibcq);
 	int (*fill_res_qp_entry)(struct sk_buff *msg, struct ib_qp *ibqp);
+	int (*fill_res_qp_entry_raw)(struct sk_buff *msg, struct ib_qp *ibqp);
 	int (*fill_res_cm_id_entry)(struct sk_buff *msg, struct rdma_cm_id *id);
 
 	/* Device lifecycle callbacks */
diff --git a/include/uapi/rdma/rdma_netlink.h b/include/uapi/rdma/rdma_netlink.h
index 8e277783fa96..3826143d420d 100644
--- a/include/uapi/rdma/rdma_netlink.h
+++ b/include/uapi/rdma/rdma_netlink.h
@@ -287,6 +287,12 @@ enum rdma_nldev_command {
 
 	RDMA_NLDEV_CMD_STAT_DEL,
 
+	RDMA_NLDEV_CMD_RES_QP_GET_RAW,
+
+	RDMA_NLDEV_CMD_RES_CQ_GET_RAW,
+
+	RDMA_NLDEV_CMD_RES_MR_GET_RAW,
+
 	RDMA_NLDEV_NUM_OPS
 };
 
@@ -525,6 +531,8 @@ enum rdma_nldev_attr {
 	 */
 	RDMA_NLDEV_ATTR_DEV_DIM,                /* u8 */
 
+	RDMA_NLDEV_ATTR_RES_RAW,	/* binary */
+
 	/*
 	 * Always the end
 	 */
-- 
cgit v1.2.3-59-g8ed1b


From 0cb42c0265837fafa2b4f302c8a7fed2631d7869 Mon Sep 17 00:00:00 2001
From: Jason Gunthorpe <jgg@nvidia.com>
Date: Fri, 26 Jun 2020 14:49:10 -0300
Subject: RDMA/core: Fix bogus WARN_ON during ib_unregister_device_queued()

ib_unregister_device_queued() can only be used by drivers using the new
dealloc_device callback flow, and it has a safety WARN_ON to ensure
drivers are using it properly.

However, if unregister and register are raced there is a special
destruction path that maintains the uniform error handling semantic of
'caller does ib_dealloc_device() on failure'. This requires disabling the
dealloc_device callback which triggers the WARN_ON.

Instead of using NULL to disable the callback use a special function
pointer so the WARN_ON does not trigger.

Fixes: d0899892edd0 ("RDMA/device: Provide APIs from the core code to help unregistration")
Link: https://lore.kernel.org/r/0-v1-a36d512e0a99+762-syz_dealloc_driver_jgg@nvidia.com
Reported-by: syzbot+4088ed905e4ae2b0e13b@syzkaller.appspotmail.com
Suggested-by: Hillf Danton <hdanton@sina.com>
Reviewed-by: Leon Romanovsky <leonro@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
---
 drivers/infiniband/core/device.c | 11 ++++++++---
 1 file changed, 8 insertions(+), 3 deletions(-)

(limited to 'drivers/infiniband/core/device.c')

diff --git a/drivers/infiniband/core/device.c b/drivers/infiniband/core/device.c
index 1335ed1f1e4a..40cf07129f66 100644
--- a/drivers/infiniband/core/device.c
+++ b/drivers/infiniband/core/device.c
@@ -1339,6 +1339,10 @@ out:
 	return ret;
 }
 
+static void prevent_dealloc_device(struct ib_device *ib_dev)
+{
+}
+
 /**
  * ib_register_device - Register an IB device with IB core
  * @device: Device to register
@@ -1409,11 +1413,11 @@ int ib_register_device(struct ib_device *device, const char *name)
 		 * possibility for a parallel unregistration along with this
 		 * error flow. Since we have a refcount here we know any
 		 * parallel flow is stopped in disable_device and will see the
-		 * NULL pointers, causing the responsibility to
+		 * special dealloc_driver pointer, causing the responsibility to
 		 * ib_dealloc_device() to revert back to this thread.
 		 */
 		dealloc_fn = device->ops.dealloc_driver;
-		device->ops.dealloc_driver = NULL;
+		device->ops.dealloc_driver = prevent_dealloc_device;
 		ib_device_put(device);
 		__ib_unregister_device(device);
 		device->ops.dealloc_driver = dealloc_fn;
@@ -1462,7 +1466,8 @@ static void __ib_unregister_device(struct ib_device *ib_dev)
 	 * Drivers using the new flow may not call ib_dealloc_device except
 	 * in error unwind prior to registration success.
 	 */
-	if (ib_dev->ops.dealloc_driver) {
+	if (ib_dev->ops.dealloc_driver &&
+	    ib_dev->ops.dealloc_driver != prevent_dealloc_device) {
 		WARN_ON(kref_read(&ib_dev->dev.kobj.kref) <= 1);
 		ib_dealloc_device(ib_dev);
 	}
-- 
cgit v1.2.3-59-g8ed1b


From 1c8fb1ea5a1dbd2159e78fa580aaffb001794cfa Mon Sep 17 00:00:00 2001
From: Yishai Hadas <yishaih@mellanox.com>
Date: Tue, 30 Jun 2020 12:39:12 +0300
Subject: IB/uverbs: Expose UAPI to query ucontext

Expose UAPI to query ucontext, this will let user space application that
didn't allocate the ucontext but has access to by owning the matching
command FD to retrieve the ucontext information.

Link: https://lore.kernel.org/r/20200630093916.332097-4-leon@kernel.org
Signed-off-by: Yishai Hadas <yishaih@mellanox.com>
Signed-off-by: Leon Romanovsky <leonro@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
---
 drivers/infiniband/core/device.c                  |  1 +
 drivers/infiniband/core/uverbs_std_types_device.c | 41 ++++++++++++++++++++++-
 include/rdma/ib_verbs.h                           |  4 +++
 include/uapi/rdma/ib_user_ioctl_cmds.h            |  6 ++++
 4 files changed, 51 insertions(+), 1 deletion(-)

(limited to 'drivers/infiniband/core/device.c')

diff --git a/drivers/infiniband/core/device.c b/drivers/infiniband/core/device.c
index 40cf07129f66..1900c0df3c8a 100644
--- a/drivers/infiniband/core/device.c
+++ b/drivers/infiniband/core/device.c
@@ -2674,6 +2674,7 @@ void ib_set_device_ops(struct ib_device *dev, const struct ib_device_ops *ops)
 	SET_DEVICE_OP(dev_ops, query_port);
 	SET_DEVICE_OP(dev_ops, query_qp);
 	SET_DEVICE_OP(dev_ops, query_srq);
+	SET_DEVICE_OP(dev_ops, query_ucontext);
 	SET_DEVICE_OP(dev_ops, rdma_netdev_get_params);
 	SET_DEVICE_OP(dev_ops, read_counters);
 	SET_DEVICE_OP(dev_ops, reg_dm_mr);
diff --git a/drivers/infiniband/core/uverbs_std_types_device.c b/drivers/infiniband/core/uverbs_std_types_device.c
index ae4a59d6f9b1..8e58605a17be 100644
--- a/drivers/infiniband/core/uverbs_std_types_device.c
+++ b/drivers/infiniband/core/uverbs_std_types_device.c
@@ -229,6 +229,37 @@ static int UVERBS_HANDLER(UVERBS_METHOD_GET_CONTEXT)(
 	return 0;
 }
 
+static int UVERBS_HANDLER(UVERBS_METHOD_QUERY_CONTEXT)(
+	struct uverbs_attr_bundle *attrs)
+{
+	u64 core_support = IB_UVERBS_CORE_SUPPORT_OPTIONAL_MR_ACCESS;
+	struct ib_ucontext *ucontext;
+	struct ib_device *ib_dev;
+	u32 num_comp;
+	int ret;
+
+	ucontext = ib_uverbs_get_ucontext(attrs);
+	if (IS_ERR(ucontext))
+		return PTR_ERR(ucontext);
+	ib_dev = ucontext->device;
+
+	if (!ib_dev->ops.query_ucontext)
+		return -EOPNOTSUPP;
+
+	num_comp = attrs->ufile->device->num_comp_vectors;
+	ret = uverbs_copy_to(attrs, UVERBS_ATTR_QUERY_CONTEXT_NUM_COMP_VECTORS,
+			     &num_comp, sizeof(num_comp));
+	if (IS_UVERBS_COPY_ERR(ret))
+		return ret;
+
+	ret = uverbs_copy_to(attrs, UVERBS_ATTR_QUERY_CONTEXT_CORE_SUPPORT,
+			     &core_support, sizeof(core_support));
+	if (IS_UVERBS_COPY_ERR(ret))
+		return ret;
+
+	return ucontext->device->ops.query_ucontext(ucontext, attrs);
+}
+
 DECLARE_UVERBS_NAMED_METHOD(
 	UVERBS_METHOD_GET_CONTEXT,
 	UVERBS_ATTR_PTR_OUT(UVERBS_ATTR_GET_CONTEXT_NUM_COMP_VECTORS,
@@ -237,6 +268,13 @@ DECLARE_UVERBS_NAMED_METHOD(
 			    UVERBS_ATTR_TYPE(u64), UA_OPTIONAL),
 	UVERBS_ATTR_UHW());
 
+DECLARE_UVERBS_NAMED_METHOD(
+	UVERBS_METHOD_QUERY_CONTEXT,
+	UVERBS_ATTR_PTR_OUT(UVERBS_ATTR_QUERY_CONTEXT_NUM_COMP_VECTORS,
+			    UVERBS_ATTR_TYPE(u32), UA_OPTIONAL),
+	UVERBS_ATTR_PTR_OUT(UVERBS_ATTR_QUERY_CONTEXT_CORE_SUPPORT,
+			    UVERBS_ATTR_TYPE(u64), UA_OPTIONAL));
+
 DECLARE_UVERBS_NAMED_METHOD(
 	UVERBS_METHOD_INFO_HANDLES,
 	/* Also includes any device specific object ids */
@@ -260,7 +298,8 @@ DECLARE_UVERBS_GLOBAL_METHODS(UVERBS_OBJECT_DEVICE,
 			      &UVERBS_METHOD(UVERBS_METHOD_GET_CONTEXT),
 			      &UVERBS_METHOD(UVERBS_METHOD_INVOKE_WRITE),
 			      &UVERBS_METHOD(UVERBS_METHOD_INFO_HANDLES),
-			      &UVERBS_METHOD(UVERBS_METHOD_QUERY_PORT));
+			      &UVERBS_METHOD(UVERBS_METHOD_QUERY_PORT),
+			      &UVERBS_METHOD(UVERBS_METHOD_QUERY_CONTEXT));
 
 const struct uapi_definition uverbs_def_obj_device[] = {
 	UAPI_DEF_CHAIN_OBJ_TREE_NAMED(UVERBS_OBJECT_DEVICE),
diff --git a/include/rdma/ib_verbs.h b/include/rdma/ib_verbs.h
index 20c801730fed..6c72bb194148 100644
--- a/include/rdma/ib_verbs.h
+++ b/include/rdma/ib_verbs.h
@@ -2645,6 +2645,10 @@ struct ib_device_ops {
 	 */
 	int (*fill_stat_mr_entry)(struct sk_buff *msg, struct ib_mr *ibmr);
 
+	/* query driver for its ucontext properties */
+	int (*query_ucontext)(struct ib_ucontext *context,
+			      struct uverbs_attr_bundle *attrs);
+
 	DECLARE_RDMA_OBJ_SIZE(ib_ah);
 	DECLARE_RDMA_OBJ_SIZE(ib_cq);
 	DECLARE_RDMA_OBJ_SIZE(ib_pd);
diff --git a/include/uapi/rdma/ib_user_ioctl_cmds.h b/include/uapi/rdma/ib_user_ioctl_cmds.h
index 4961d5e858eb..83b6e71ea216 100644
--- a/include/uapi/rdma/ib_user_ioctl_cmds.h
+++ b/include/uapi/rdma/ib_user_ioctl_cmds.h
@@ -69,6 +69,7 @@ enum uverbs_methods_device {
 	UVERBS_METHOD_INFO_HANDLES,
 	UVERBS_METHOD_QUERY_PORT,
 	UVERBS_METHOD_GET_CONTEXT,
+	UVERBS_METHOD_QUERY_CONTEXT,
 };
 
 enum uverbs_attrs_invoke_write_cmd_attr_ids {
@@ -87,6 +88,11 @@ enum uverbs_attrs_get_context_attr_ids {
 	UVERBS_ATTR_GET_CONTEXT_CORE_SUPPORT,
 };
 
+enum uverbs_attrs_query_context_attr_ids {
+	UVERBS_ATTR_QUERY_CONTEXT_NUM_COMP_VECTORS,
+	UVERBS_ATTR_QUERY_CONTEXT_CORE_SUPPORT,
+};
+
 enum uverbs_attrs_create_cq_cmd_attr_ids {
 	UVERBS_ATTR_CREATE_CQ_HANDLE,
 	UVERBS_ATTR_CREATE_CQ_CQE,
-- 
cgit v1.2.3-59-g8ed1b


From 3b023e1b680a56e84c22d43486875a5aa4c78afe Mon Sep 17 00:00:00 2001
From: Leon Romanovsky <leonro@mellanox.com>
Date: Tue, 30 Jun 2020 13:18:52 +0300
Subject: RDMA/core: Create and destroy counters in the ib_core

Move allocation and destruction of counters under ib_core responsibility

Link: https://lore.kernel.org/r/20200630101855.368895-2-leon@kernel.org
Signed-off-by: Leon Romanovsky <leonro@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
---
 drivers/infiniband/core/device.c                    |  1 +
 drivers/infiniband/core/uverbs_std_types_counters.c | 17 +++++++++--------
 drivers/infiniband/hw/mlx5/main.c                   | 20 ++++++--------------
 include/rdma/ib_verbs.h                             |  7 ++++---
 4 files changed, 20 insertions(+), 25 deletions(-)

(limited to 'drivers/infiniband/core/device.c')

diff --git a/drivers/infiniband/core/device.c b/drivers/infiniband/core/device.c
index 1900c0df3c8a..0a259f475e89 100644
--- a/drivers/infiniband/core/device.c
+++ b/drivers/infiniband/core/device.c
@@ -2687,6 +2687,7 @@ void ib_set_device_ops(struct ib_device *dev, const struct ib_device_ops *ops)
 	SET_DEVICE_OP(dev_ops, set_vf_link_state);
 
 	SET_OBJ_SIZE(dev_ops, ib_ah);
+	SET_OBJ_SIZE(dev_ops, ib_counters);
 	SET_OBJ_SIZE(dev_ops, ib_cq);
 	SET_OBJ_SIZE(dev_ops, ib_pd);
 	SET_OBJ_SIZE(dev_ops, ib_srq);
diff --git a/drivers/infiniband/core/uverbs_std_types_counters.c b/drivers/infiniband/core/uverbs_std_types_counters.c
index 9f013304e677..c7e7438752bc 100644
--- a/drivers/infiniband/core/uverbs_std_types_counters.c
+++ b/drivers/infiniband/core/uverbs_std_types_counters.c
@@ -46,7 +46,9 @@ static int uverbs_free_counters(struct ib_uobject *uobject,
 	if (ret)
 		return ret;
 
-	return counters->device->ops.destroy_counters(counters);
+	counters->device->ops.destroy_counters(counters);
+	kfree(counters);
+	return 0;
 }
 
 static int UVERBS_HANDLER(UVERBS_METHOD_COUNTERS_CREATE)(
@@ -66,20 +68,19 @@ static int UVERBS_HANDLER(UVERBS_METHOD_COUNTERS_CREATE)(
 	if (!ib_dev->ops.create_counters)
 		return -EOPNOTSUPP;
 
-	counters = ib_dev->ops.create_counters(ib_dev, attrs);
-	if (IS_ERR(counters)) {
-		ret = PTR_ERR(counters);
-		goto err_create_counters;
-	}
+	counters = rdma_zalloc_drv_obj(ib_dev, ib_counters);
+	if (!counters)
+		return -ENOMEM;
 
 	counters->device = ib_dev;
 	counters->uobject = uobj;
 	uobj->object = counters;
 	atomic_set(&counters->usecnt, 0);
 
-	return 0;
+	ret = ib_dev->ops.create_counters(counters, attrs);
+	if (ret)
+		kfree(counters);
 
-err_create_counters:
 	return ret;
 }
 
diff --git a/drivers/infiniband/hw/mlx5/main.c b/drivers/infiniband/hw/mlx5/main.c
index 324a98c77ad5..6e6e126c39ef 100644
--- a/drivers/infiniband/hw/mlx5/main.c
+++ b/drivers/infiniband/hw/mlx5/main.c
@@ -6455,7 +6455,7 @@ err_bound:
 	return ret;
 }
 
-static int mlx5_ib_destroy_counters(struct ib_counters *counters)
+static void mlx5_ib_destroy_counters(struct ib_counters *counters)
 {
 	struct mlx5_ib_mcounters *mcounters = to_mcounters(counters);
 
@@ -6463,24 +6463,15 @@ static int mlx5_ib_destroy_counters(struct ib_counters *counters)
 	if (mcounters->hw_cntrs_hndl)
 		mlx5_fc_destroy(to_mdev(counters->device)->mdev,
 				mcounters->hw_cntrs_hndl);
-
-	kfree(mcounters);
-
-	return 0;
 }
 
-static struct ib_counters *mlx5_ib_create_counters(struct ib_device *device,
-						   struct uverbs_attr_bundle *attrs)
+static int mlx5_ib_create_counters(struct ib_counters *counters,
+				   struct uverbs_attr_bundle *attrs)
 {
-	struct mlx5_ib_mcounters *mcounters;
-
-	mcounters = kzalloc(sizeof(*mcounters), GFP_KERNEL);
-	if (!mcounters)
-		return ERR_PTR(-ENOMEM);
+	struct mlx5_ib_mcounters *mcounters = to_mcounters(counters);
 
 	mutex_init(&mcounters->mcntrs_mutex);
-
-	return &mcounters->ibcntrs;
+	return 0;
 }
 
 static void mlx5_ib_stage_init_cleanup(struct mlx5_ib_dev *dev)
@@ -6648,6 +6639,7 @@ static const struct ib_device_ops mlx5_ib_dev_ops = {
 	.resize_cq = mlx5_ib_resize_cq,
 
 	INIT_RDMA_OBJ_SIZE(ib_ah, mlx5_ib_ah, ibah),
+	INIT_RDMA_OBJ_SIZE(ib_counters, mlx5_ib_mcounters, ibcntrs),
 	INIT_RDMA_OBJ_SIZE(ib_cq, mlx5_ib_cq, ibcq),
 	INIT_RDMA_OBJ_SIZE(ib_pd, mlx5_ib_pd, ibpd),
 	INIT_RDMA_OBJ_SIZE(ib_srq, mlx5_ib_srq, ibsrq),
diff --git a/include/rdma/ib_verbs.h b/include/rdma/ib_verbs.h
index 6c72bb194148..ecaf299e0789 100644
--- a/include/rdma/ib_verbs.h
+++ b/include/rdma/ib_verbs.h
@@ -2540,9 +2540,9 @@ struct ib_device_ops {
 	struct ib_mr *(*reg_dm_mr)(struct ib_pd *pd, struct ib_dm *dm,
 				   struct ib_dm_mr_attr *attr,
 				   struct uverbs_attr_bundle *attrs);
-	struct ib_counters *(*create_counters)(
-		struct ib_device *device, struct uverbs_attr_bundle *attrs);
-	int (*destroy_counters)(struct ib_counters *counters);
+	int (*create_counters)(struct ib_counters *counters,
+			       struct uverbs_attr_bundle *attrs);
+	void (*destroy_counters)(struct ib_counters *counters);
 	int (*read_counters)(struct ib_counters *counters,
 			     struct ib_counters_read_attr *counters_read_attr,
 			     struct uverbs_attr_bundle *attrs);
@@ -2650,6 +2650,7 @@ struct ib_device_ops {
 			      struct uverbs_attr_bundle *attrs);
 
 	DECLARE_RDMA_OBJ_SIZE(ib_ah);
+	DECLARE_RDMA_OBJ_SIZE(ib_counters);
 	DECLARE_RDMA_OBJ_SIZE(ib_cq);
 	DECLARE_RDMA_OBJ_SIZE(ib_pd);
 	DECLARE_RDMA_OBJ_SIZE(ib_srq);
-- 
cgit v1.2.3-59-g8ed1b


From 28ad5f65c314ffdd5888d6afa61772d3032a332c Mon Sep 17 00:00:00 2001
From: Leon Romanovsky <leonro@mellanox.com>
Date: Tue, 30 Jun 2020 13:18:54 +0300
Subject: RDMA: Move XRCD to be under ib_core responsibility

Update the code to allocate and free ib_xrcd structure in the
ib_core instead of inside drivers.

Link: https://lore.kernel.org/r/20200630101855.368895-4-leon@kernel.org
Signed-off-by: Leon Romanovsky <leonro@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
---
 drivers/infiniband/core/device.c     |  1 +
 drivers/infiniband/core/verbs.c      | 28 ++++++++++++++++++---------
 drivers/infiniband/hw/mlx4/main.c    | 37 +++++++++++++++---------------------
 drivers/infiniband/hw/mlx5/main.c    |  2 ++
 drivers/infiniband/hw/mlx5/mlx5_ib.h |  5 ++---
 drivers/infiniband/hw/mlx5/qp.c      | 32 +++++++------------------------
 include/rdma/ib_verbs.h              |  6 +++---
 7 files changed, 49 insertions(+), 62 deletions(-)

(limited to 'drivers/infiniband/core/device.c')

diff --git a/drivers/infiniband/core/device.c b/drivers/infiniband/core/device.c
index 0a259f475e89..b2d617e599a1 100644
--- a/drivers/infiniband/core/device.c
+++ b/drivers/infiniband/core/device.c
@@ -2692,6 +2692,7 @@ void ib_set_device_ops(struct ib_device *dev, const struct ib_device_ops *ops)
 	SET_OBJ_SIZE(dev_ops, ib_pd);
 	SET_OBJ_SIZE(dev_ops, ib_srq);
 	SET_OBJ_SIZE(dev_ops, ib_ucontext);
+	SET_OBJ_SIZE(dev_ops, ib_xrcd);
 }
 EXPORT_SYMBOL(ib_set_device_ops);
 
diff --git a/drivers/infiniband/core/verbs.c b/drivers/infiniband/core/verbs.c
index b1b6cc21ca96..a92783105cea 100644
--- a/drivers/infiniband/core/verbs.c
+++ b/drivers/infiniband/core/verbs.c
@@ -2293,20 +2293,28 @@ struct ib_xrcd *ib_alloc_xrcd_user(struct ib_device *device,
 				   struct inode *inode, struct ib_udata *udata)
 {
 	struct ib_xrcd *xrcd;
+	int ret;
 
 	if (!device->ops.alloc_xrcd)
 		return ERR_PTR(-EOPNOTSUPP);
 
-	xrcd = device->ops.alloc_xrcd(device, udata);
-	if (!IS_ERR(xrcd)) {
-		xrcd->device = device;
-		xrcd->inode = inode;
-		atomic_set(&xrcd->usecnt, 0);
-		init_rwsem(&xrcd->tgt_qps_rwsem);
-		xa_init(&xrcd->tgt_qps);
-	}
+	xrcd = rdma_zalloc_drv_obj(device, ib_xrcd);
+	if (!xrcd)
+		return ERR_PTR(-ENOMEM);
 
+	xrcd->device = device;
+	xrcd->inode = inode;
+	atomic_set(&xrcd->usecnt, 0);
+	init_rwsem(&xrcd->tgt_qps_rwsem);
+	xa_init(&xrcd->tgt_qps);
+
+	ret = device->ops.alloc_xrcd(xrcd, udata);
+	if (ret)
+		goto err;
 	return xrcd;
+err:
+	kfree(xrcd);
+	return ERR_PTR(ret);
 }
 EXPORT_SYMBOL(ib_alloc_xrcd_user);
 
@@ -2321,7 +2329,9 @@ int ib_dealloc_xrcd_user(struct ib_xrcd *xrcd, struct ib_udata *udata)
 		return -EBUSY;
 
 	WARN_ON(!xa_empty(&xrcd->tgt_qps));
-	return xrcd->device->ops.dealloc_xrcd(xrcd, udata);
+	xrcd->device->ops.dealloc_xrcd(xrcd, udata);
+	kfree(xrcd);
+	return 0;
 }
 EXPORT_SYMBOL(ib_dealloc_xrcd_user);
 
diff --git a/drivers/infiniband/hw/mlx4/main.c b/drivers/infiniband/hw/mlx4/main.c
index 816d28854a8e..5e7910a517da 100644
--- a/drivers/infiniband/hw/mlx4/main.c
+++ b/drivers/infiniband/hw/mlx4/main.c
@@ -1219,56 +1219,47 @@ static void mlx4_ib_dealloc_pd(struct ib_pd *pd, struct ib_udata *udata)
 	mlx4_pd_free(to_mdev(pd->device)->dev, to_mpd(pd)->pdn);
 }
 
-static struct ib_xrcd *mlx4_ib_alloc_xrcd(struct ib_device *ibdev,
-					  struct ib_udata *udata)
+static int mlx4_ib_alloc_xrcd(struct ib_xrcd *ibxrcd, struct ib_udata *udata)
 {
-	struct mlx4_ib_xrcd *xrcd;
+	struct mlx4_ib_dev *dev = to_mdev(ibxrcd->device);
+	struct mlx4_ib_xrcd *xrcd = to_mxrcd(ibxrcd);
 	struct ib_cq_init_attr cq_attr = {};
 	int err;
 
-	if (!(to_mdev(ibdev)->dev->caps.flags & MLX4_DEV_CAP_FLAG_XRC))
-		return ERR_PTR(-ENOSYS);
-
-	xrcd = kmalloc(sizeof *xrcd, GFP_KERNEL);
-	if (!xrcd)
-		return ERR_PTR(-ENOMEM);
+	if (!(dev->dev->caps.flags & MLX4_DEV_CAP_FLAG_XRC))
+		return -EOPNOTSUPP;
 
-	err = mlx4_xrcd_alloc(to_mdev(ibdev)->dev, &xrcd->xrcdn);
+	err = mlx4_xrcd_alloc(dev->dev, &xrcd->xrcdn);
 	if (err)
-		goto err1;
+		return err;
 
-	xrcd->pd = ib_alloc_pd(ibdev, 0);
+	xrcd->pd = ib_alloc_pd(ibxrcd->device, 0);
 	if (IS_ERR(xrcd->pd)) {
 		err = PTR_ERR(xrcd->pd);
 		goto err2;
 	}
 
 	cq_attr.cqe = 1;
-	xrcd->cq = ib_create_cq(ibdev, NULL, NULL, xrcd, &cq_attr);
+	xrcd->cq = ib_create_cq(ibxrcd->device, NULL, NULL, xrcd, &cq_attr);
 	if (IS_ERR(xrcd->cq)) {
 		err = PTR_ERR(xrcd->cq);
 		goto err3;
 	}
 
-	return &xrcd->ibxrcd;
+	return 0;
 
 err3:
 	ib_dealloc_pd(xrcd->pd);
 err2:
-	mlx4_xrcd_free(to_mdev(ibdev)->dev, xrcd->xrcdn);
-err1:
-	kfree(xrcd);
-	return ERR_PTR(err);
+	mlx4_xrcd_free(dev->dev, xrcd->xrcdn);
+	return err;
 }
 
-static int mlx4_ib_dealloc_xrcd(struct ib_xrcd *xrcd, struct ib_udata *udata)
+static void mlx4_ib_dealloc_xrcd(struct ib_xrcd *xrcd, struct ib_udata *udata)
 {
 	ib_destroy_cq(to_mxrcd(xrcd)->cq);
 	ib_dealloc_pd(to_mxrcd(xrcd)->pd);
 	mlx4_xrcd_free(to_mdev(xrcd->device)->dev, to_mxrcd(xrcd)->xrcdn);
-	kfree(xrcd);
-
-	return 0;
 }
 
 static int add_gid_entry(struct ib_qp *ibqp, union ib_gid *gid)
@@ -2607,6 +2598,8 @@ static const struct ib_device_ops mlx4_ib_dev_mw_ops = {
 static const struct ib_device_ops mlx4_ib_dev_xrc_ops = {
 	.alloc_xrcd = mlx4_ib_alloc_xrcd,
 	.dealloc_xrcd = mlx4_ib_dealloc_xrcd,
+
+	INIT_RDMA_OBJ_SIZE(ib_xrcd, mlx4_ib_xrcd, ibxrcd),
 };
 
 static const struct ib_device_ops mlx4_ib_dev_fs_ops = {
diff --git a/drivers/infiniband/hw/mlx5/main.c b/drivers/infiniband/hw/mlx5/main.c
index 6e6e126c39ef..2da592054a3f 100644
--- a/drivers/infiniband/hw/mlx5/main.c
+++ b/drivers/infiniband/hw/mlx5/main.c
@@ -6671,6 +6671,8 @@ static const struct ib_device_ops mlx5_ib_dev_mw_ops = {
 static const struct ib_device_ops mlx5_ib_dev_xrc_ops = {
 	.alloc_xrcd = mlx5_ib_alloc_xrcd,
 	.dealloc_xrcd = mlx5_ib_dealloc_xrcd,
+
+	INIT_RDMA_OBJ_SIZE(ib_xrcd, mlx5_ib_xrcd, ibxrcd),
 };
 
 static const struct ib_device_ops mlx5_ib_dev_dm_ops = {
diff --git a/drivers/infiniband/hw/mlx5/mlx5_ib.h b/drivers/infiniband/hw/mlx5/mlx5_ib.h
index 27b069ef63d9..88e30ff77c16 100644
--- a/drivers/infiniband/hw/mlx5/mlx5_ib.h
+++ b/drivers/infiniband/hw/mlx5/mlx5_ib.h
@@ -1224,9 +1224,8 @@ int mlx5_ib_process_mad(struct ib_device *ibdev, int mad_flags, u8 port_num,
 			const struct ib_wc *in_wc, const struct ib_grh *in_grh,
 			const struct ib_mad *in, struct ib_mad *out,
 			size_t *out_mad_size, u16 *out_mad_pkey_index);
-struct ib_xrcd *mlx5_ib_alloc_xrcd(struct ib_device *ibdev,
-				   struct ib_udata *udata);
-int mlx5_ib_dealloc_xrcd(struct ib_xrcd *xrcd, struct ib_udata *udata);
+int mlx5_ib_alloc_xrcd(struct ib_xrcd *xrcd, struct ib_udata *udata);
+void mlx5_ib_dealloc_xrcd(struct ib_xrcd *xrcd, struct ib_udata *udata);
 int mlx5_ib_get_buf_offset(u64 addr, int page_shift, u32 *offset);
 int mlx5_query_ext_port_caps(struct mlx5_ib_dev *dev, u8 port);
 int mlx5_query_mad_ifc_smp_attr_node_info(struct ib_device *ibdev,
diff --git a/drivers/infiniband/hw/mlx5/qp.c b/drivers/infiniband/hw/mlx5/qp.c
index 7489b0479e4f..c6a4db9f6a0b 100644
--- a/drivers/infiniband/hw/mlx5/qp.c
+++ b/drivers/infiniband/hw/mlx5/qp.c
@@ -4700,41 +4700,23 @@ out:
 	return err;
 }
 
-struct ib_xrcd *mlx5_ib_alloc_xrcd(struct ib_device *ibdev,
-				   struct ib_udata *udata)
+int mlx5_ib_alloc_xrcd(struct ib_xrcd *ibxrcd, struct ib_udata *udata)
 {
-	struct mlx5_ib_dev *dev = to_mdev(ibdev);
-	struct mlx5_ib_xrcd *xrcd;
-	int err;
+	struct mlx5_ib_dev *dev = to_mdev(ibxrcd->device);
+	struct mlx5_ib_xrcd *xrcd = to_mxrcd(ibxrcd);
 
 	if (!MLX5_CAP_GEN(dev->mdev, xrc))
-		return ERR_PTR(-ENOSYS);
-
-	xrcd = kmalloc(sizeof(*xrcd), GFP_KERNEL);
-	if (!xrcd)
-		return ERR_PTR(-ENOMEM);
-
-	err = mlx5_cmd_xrcd_alloc(dev->mdev, &xrcd->xrcdn, 0);
-	if (err) {
-		kfree(xrcd);
-		return ERR_PTR(-ENOMEM);
-	}
+		return -EOPNOTSUPP;
 
-	return &xrcd->ibxrcd;
+	return mlx5_cmd_xrcd_alloc(dev->mdev, &xrcd->xrcdn, 0);
 }
 
-int mlx5_ib_dealloc_xrcd(struct ib_xrcd *xrcd, struct ib_udata *udata)
+void mlx5_ib_dealloc_xrcd(struct ib_xrcd *xrcd, struct ib_udata *udata)
 {
 	struct mlx5_ib_dev *dev = to_mdev(xrcd->device);
 	u32 xrcdn = to_mxrcd(xrcd)->xrcdn;
-	int err;
 
-	err = mlx5_cmd_xrcd_dealloc(dev->mdev, xrcdn, 0);
-	if (err)
-		mlx5_ib_warn(dev, "failed to dealloc xrcdn 0x%x\n", xrcdn);
-
-	kfree(xrcd);
-	return 0;
+	mlx5_cmd_xrcd_dealloc(dev->mdev, xrcdn, 0);
 }
 
 static void mlx5_ib_wq_event(struct mlx5_core_qp *core_qp, int type)
diff --git a/include/rdma/ib_verbs.h b/include/rdma/ib_verbs.h
index ecaf299e0789..087f001fd020 100644
--- a/include/rdma/ib_verbs.h
+++ b/include/rdma/ib_verbs.h
@@ -2494,9 +2494,8 @@ struct ib_device_ops {
 	int (*dealloc_mw)(struct ib_mw *mw);
 	int (*attach_mcast)(struct ib_qp *qp, union ib_gid *gid, u16 lid);
 	int (*detach_mcast)(struct ib_qp *qp, union ib_gid *gid, u16 lid);
-	struct ib_xrcd *(*alloc_xrcd)(struct ib_device *device,
-				      struct ib_udata *udata);
-	int (*dealloc_xrcd)(struct ib_xrcd *xrcd, struct ib_udata *udata);
+	int (*alloc_xrcd)(struct ib_xrcd *xrcd, struct ib_udata *udata);
+	void (*dealloc_xrcd)(struct ib_xrcd *xrcd, struct ib_udata *udata);
 	struct ib_flow *(*create_flow)(struct ib_qp *qp,
 				       struct ib_flow_attr *flow_attr,
 				       int domain, struct ib_udata *udata);
@@ -2655,6 +2654,7 @@ struct ib_device_ops {
 	DECLARE_RDMA_OBJ_SIZE(ib_pd);
 	DECLARE_RDMA_OBJ_SIZE(ib_srq);
 	DECLARE_RDMA_OBJ_SIZE(ib_ucontext);
+	DECLARE_RDMA_OBJ_SIZE(ib_xrcd);
 };
 
 struct ib_core_device {
-- 
cgit v1.2.3-59-g8ed1b


From ab75a6cb8cb27ff7f65ab66962975be30507ee29 Mon Sep 17 00:00:00 2001
From: Kamal Heib <kamalheib1@gmail.com>
Date: Tue, 14 Jul 2020 21:34:10 +0300
Subject: RDMA/core: Remove query_pkey from the mandatory ops

The query_pkey() isn't mandatory for the iwarp providers, so remove this
requirement from the RDMA core.

Link: https://lore.kernel.org/r/20200714183414.61069-4-kamalheib1@gmail.com
Signed-off-by: Kamal Heib <kamalheib1@gmail.com>
Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
---
 drivers/infiniband/core/device.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

(limited to 'drivers/infiniband/core/device.c')

diff --git a/drivers/infiniband/core/device.c b/drivers/infiniband/core/device.c
index b2d617e599a1..d293b826acbc 100644
--- a/drivers/infiniband/core/device.c
+++ b/drivers/infiniband/core/device.c
@@ -272,7 +272,6 @@ static void ib_device_check_mandatory(struct ib_device *device)
 	} mandatory_table[] = {
 		IB_MANDATORY_FUNC(query_device),
 		IB_MANDATORY_FUNC(query_port),
-		IB_MANDATORY_FUNC(query_pkey),
 		IB_MANDATORY_FUNC(alloc_pd),
 		IB_MANDATORY_FUNC(dealloc_pd),
 		IB_MANDATORY_FUNC(create_qp),
@@ -2362,6 +2361,9 @@ int ib_query_pkey(struct ib_device *device,
 	if (!rdma_is_port_valid(device, port_num))
 		return -EINVAL;
 
+	if (!device->ops.query_pkey)
+		return -EOPNOTSUPP;
+
 	return device->ops.query_pkey(device, port_num, index, pkey);
 }
 EXPORT_SYMBOL(ib_query_pkey);
-- 
cgit v1.2.3-59-g8ed1b