From e502b8b011294c293064ef1baeb04235ccc05439 Mon Sep 17 00:00:00 2001
From: Leon Romanovsky <leonro@mellanox.com>
Date: Tue, 8 Jan 2019 16:07:24 +0200
Subject: RDMA/core: Don't depend device ODP capabilities on kconfig option

Device capability bits are exposing what specific device supports from HW
perspective. Those bits are not dependent on kernel configurations and
RDMA/core should ensure that proper interfaces to users will be disabled
if CONFIG_INFINIBAND_ON_DEMAND_PAGING is not set.

Fixes: f4056bfd8ccf ("IB/core: Add on demand paging caps to ib_uverbs_ex_query_device")
Fixes: 8cdd312cfed7 ("IB/mlx5: Implement the ODP capability query verb")
Signed-off-by: Leon Romanovsky <leonro@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 drivers/infiniband/hw/mlx5/main.c | 10 +++++-----
 1 file changed, 5 insertions(+), 5 deletions(-)

(limited to 'drivers/infiniband/hw/mlx5/main.c')

diff --git a/drivers/infiniband/hw/mlx5/main.c b/drivers/infiniband/hw/mlx5/main.c
index 94fe253d4956..06ee1f0cb22d 100644
--- a/drivers/infiniband/hw/mlx5/main.c
+++ b/drivers/infiniband/hw/mlx5/main.c
@@ -923,11 +923,11 @@ static int mlx5_ib_query_device(struct ib_device *ibdev,
 	props->hca_core_clock = MLX5_CAP_GEN(mdev, device_frequency_khz);
 	props->timestamp_mask = 0x7FFFFFFFFFFFFFFFULL;
 
-#ifdef CONFIG_INFINIBAND_ON_DEMAND_PAGING
-	if (MLX5_CAP_GEN(mdev, pg))
-		props->device_cap_flags |= IB_DEVICE_ON_DEMAND_PAGING;
-	props->odp_caps = dev->odp_caps;
-#endif
+	if (IS_ENABLED(CONFIG_INFINIBAND_ON_DEMAND_PAGING)) {
+		if (MLX5_CAP_GEN(mdev, pg))
+			props->device_cap_flags |= IB_DEVICE_ON_DEMAND_PAGING;
+		props->odp_caps = dev->odp_caps;
+	}
 
 	if (MLX5_CAP_GEN(mdev, cd))
 		props->device_cap_flags |= IB_DEVICE_CROSS_CHANNEL;
-- 
cgit v1.2.3


From 13859d5df418ea535926e2b57c29d5161c522b9d Mon Sep 17 00:00:00 2001
From: Leon Romanovsky <leonro@mellanox.com>
Date: Tue, 8 Jan 2019 16:07:26 +0200
Subject: RDMA/mlx5: Embed into the code flow the ODP config option

Convert various places to more readable code, which embeds
CONFIG_INFINIBAND_ON_DEMAND_PAGING into the code flow.

Signed-off-by: Leon Romanovsky <leonro@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 drivers/infiniband/core/uverbs_cmd.c |  3 ---
 drivers/infiniband/hw/mlx5/main.c    | 41 ++++++++++++++++++------------------
 drivers/infiniband/hw/mlx5/mem.c     |  5 +----
 drivers/infiniband/hw/mlx5/mr.c      | 28 +++++++++++-------------
 include/rdma/ib_umem_odp.h           | 26 +++++++++++------------
 5 files changed, 46 insertions(+), 57 deletions(-)

(limited to 'drivers/infiniband/hw/mlx5/main.c')

diff --git a/drivers/infiniband/core/uverbs_cmd.c b/drivers/infiniband/core/uverbs_cmd.c
index 549d9eedf62e..d4f1a2ef5015 100644
--- a/drivers/infiniband/core/uverbs_cmd.c
+++ b/drivers/infiniband/core/uverbs_cmd.c
@@ -234,14 +234,11 @@ static int ib_uverbs_get_context(struct uverbs_attr_bundle *attrs)
 	ucontext->closing = false;
 	ucontext->cleanup_retryable = false;
 
-#ifdef CONFIG_INFINIBAND_ON_DEMAND_PAGING
 	mutex_init(&ucontext->per_mm_list_lock);
 	INIT_LIST_HEAD(&ucontext->per_mm_list);
 	if (!(ib_dev->attrs.device_cap_flags & IB_DEVICE_ON_DEMAND_PAGING))
 		ucontext->invalidate_range = NULL;
 
-#endif
-
 	resp.num_comp_vectors = file->device->num_comp_vectors;
 
 	ret = get_unused_fd_flags(O_CLOEXEC);
diff --git a/drivers/infiniband/hw/mlx5/main.c b/drivers/infiniband/hw/mlx5/main.c
index 06ee1f0cb22d..11e9783cefcc 100644
--- a/drivers/infiniband/hw/mlx5/main.c
+++ b/drivers/infiniband/hw/mlx5/main.c
@@ -1763,9 +1763,9 @@ static struct ib_ucontext *mlx5_ib_alloc_ucontext(struct ib_device *ibdev,
 	if (err)
 		goto out_sys_pages;
 
-#ifdef CONFIG_INFINIBAND_ON_DEMAND_PAGING
-	context->ibucontext.invalidate_range = &mlx5_ib_invalidate_range;
-#endif
+	if (ibdev->attrs.device_cap_flags & IB_DEVICE_ON_DEMAND_PAGING)
+		context->ibucontext.invalidate_range =
+			&mlx5_ib_invalidate_range;
 
 	if (req.flags & MLX5_IB_ALLOC_UCTX_DEVX) {
 		err = mlx5_ib_devx_create(dev, true);
@@ -1897,12 +1897,10 @@ static int mlx5_ib_dealloc_ucontext(struct ib_ucontext *ibcontext)
 	struct mlx5_ib_dev *dev = to_mdev(ibcontext->device);
 	struct mlx5_bfreg_info *bfregi;
 
-#ifdef CONFIG_INFINIBAND_ON_DEMAND_PAGING
 	/* All umem's must be destroyed before destroying the ucontext. */
 	mutex_lock(&ibcontext->per_mm_list_lock);
 	WARN_ON(!list_empty(&ibcontext->per_mm_list));
 	mutex_unlock(&ibcontext->per_mm_list_lock);
-#endif
 
 	bfregi = &context->bfregi;
 	mlx5_ib_dealloc_transport_domain(dev, context->tdn, context->devx_uid);
@@ -5722,11 +5720,11 @@ static struct ib_counters *mlx5_ib_create_counters(struct ib_device *device,
 void mlx5_ib_stage_init_cleanup(struct mlx5_ib_dev *dev)
 {
 	mlx5_ib_cleanup_multiport_master(dev);
-#ifdef CONFIG_INFINIBAND_ON_DEMAND_PAGING
-	cleanup_srcu_struct(&dev->mr_srcu);
-	drain_workqueue(dev->advise_mr_wq);
-	destroy_workqueue(dev->advise_mr_wq);
-#endif
+	if (IS_ENABLED(CONFIG_INFINIBAND_ON_DEMAND_PAGING)) {
+		cleanup_srcu_struct(&dev->mr_srcu);
+		drain_workqueue(dev->advise_mr_wq);
+		destroy_workqueue(dev->advise_mr_wq);
+	}
 	kfree(dev->port);
 }
 
@@ -5779,19 +5777,20 @@ int mlx5_ib_stage_init_init(struct mlx5_ib_dev *dev)
 	spin_lock_init(&dev->memic.memic_lock);
 	dev->memic.dev = mdev;
 
-#ifdef CONFIG_INFINIBAND_ON_DEMAND_PAGING
-	dev->advise_mr_wq = alloc_ordered_workqueue("mlx5_ib_advise_mr_wq", 0);
-	if (!dev->advise_mr_wq) {
-		err = -ENOMEM;
-		goto err_mp;
-	}
+	if (IS_ENABLED(CONFIG_INFINIBAND_ON_DEMAND_PAGING)) {
+		dev->advise_mr_wq =
+			alloc_ordered_workqueue("mlx5_ib_advise_mr_wq", 0);
+		if (!dev->advise_mr_wq) {
+			err = -ENOMEM;
+			goto err_mp;
+		}
 
-	err = init_srcu_struct(&dev->mr_srcu);
-	if (err) {
-		destroy_workqueue(dev->advise_mr_wq);
-		goto err_mp;
+		err = init_srcu_struct(&dev->mr_srcu);
+		if (err) {
+			destroy_workqueue(dev->advise_mr_wq);
+			goto err_mp;
+		}
 	}
-#endif
 
 	return 0;
 err_mp:
diff --git a/drivers/infiniband/hw/mlx5/mem.c b/drivers/infiniband/hw/mlx5/mem.c
index 549234988bb4..9f90be296ee0 100644
--- a/drivers/infiniband/hw/mlx5/mem.c
+++ b/drivers/infiniband/hw/mlx5/mem.c
@@ -111,7 +111,6 @@ void mlx5_ib_cont_pages(struct ib_umem *umem, u64 addr,
 	*count = i;
 }
 
-#ifdef CONFIG_INFINIBAND_ON_DEMAND_PAGING
 static u64 umem_dma_to_mtt(dma_addr_t umem_dma)
 {
 	u64 mtt_entry = umem_dma & ODP_DMA_ADDR_MASK;
@@ -123,7 +122,6 @@ static u64 umem_dma_to_mtt(dma_addr_t umem_dma)
 
 	return mtt_entry;
 }
-#endif
 
 /*
  * Populate the given array with bus addresses from the umem.
@@ -151,7 +149,7 @@ void __mlx5_ib_populate_pas(struct mlx5_ib_dev *dev, struct ib_umem *umem,
 	int len;
 	struct scatterlist *sg;
 	int entry;
-#ifdef CONFIG_INFINIBAND_ON_DEMAND_PAGING
+
 	if (umem->is_odp) {
 		WARN_ON(shift != 0);
 		WARN_ON(access_flags != (MLX5_IB_MTT_READ | MLX5_IB_MTT_WRITE));
@@ -164,7 +162,6 @@ void __mlx5_ib_populate_pas(struct mlx5_ib_dev *dev, struct ib_umem *umem,
 		}
 		return;
 	}
-#endif
 
 	i = 0;
 	for_each_sg(umem->sg_head.sgl, sg, umem->nmap, entry) {
diff --git a/drivers/infiniband/hw/mlx5/mr.c b/drivers/infiniband/hw/mlx5/mr.c
index c389750f771e..494a90f4348c 100644
--- a/drivers/infiniband/hw/mlx5/mr.c
+++ b/drivers/infiniband/hw/mlx5/mr.c
@@ -71,10 +71,9 @@ static int destroy_mkey(struct mlx5_ib_dev *dev, struct mlx5_ib_mr *mr)
 {
 	int err = mlx5_core_destroy_mkey(dev->mdev, &mr->mmkey);
 
-#ifdef CONFIG_INFINIBAND_ON_DEMAND_PAGING
-	/* Wait until all page fault handlers using the mr complete. */
-	synchronize_srcu(&dev->mr_srcu);
-#endif
+	if (IS_ENABLED(CONFIG_INFINIBAND_ON_DEMAND_PAGING))
+		/* Wait until all page fault handlers using the mr complete. */
+		synchronize_srcu(&dev->mr_srcu);
 
 	return err;
 }
@@ -254,9 +253,8 @@ static void remove_keys(struct mlx5_ib_dev *dev, int c, int num)
 		mlx5_core_destroy_mkey(dev->mdev, &mr->mmkey);
 	}
 
-#ifdef CONFIG_INFINIBAND_ON_DEMAND_PAGING
-	synchronize_srcu(&dev->mr_srcu);
-#endif
+	if (IS_ENABLED(CONFIG_INFINIBAND_ON_DEMAND_PAGING))
+		synchronize_srcu(&dev->mr_srcu);
 
 	list_for_each_entry_safe(mr, tmp_mr, &del_list, list) {
 		list_del(&mr->list);
@@ -1329,8 +1327,8 @@ struct ib_mr *mlx5_ib_reg_user_mr(struct ib_pd *pd, u64 start, u64 length,
 	mlx5_ib_dbg(dev, "start 0x%llx, virt_addr 0x%llx, length 0x%llx, access_flags 0x%x\n",
 		    start, virt_addr, length, access_flags);
 
-#ifdef CONFIG_INFINIBAND_ON_DEMAND_PAGING
-	if (!start && length == U64_MAX) {
+	if (IS_ENABLED(CONFIG_INFINIBAND_ON_DEMAND_PAGING) && !start &&
+	    length == U64_MAX) {
 		if (!(access_flags & IB_ACCESS_ON_DEMAND) ||
 		    !(dev->odp_caps.general_caps & IB_ODP_SUPPORT_IMPLICIT))
 			return ERR_PTR(-EINVAL);
@@ -1340,7 +1338,6 @@ struct ib_mr *mlx5_ib_reg_user_mr(struct ib_pd *pd, u64 start, u64 length,
 			return ERR_CAST(mr);
 		return &mr->ibmr;
 	}
-#endif
 
 	err = mr_umem_get(pd, start, length, access_flags, &umem, &npages,
 			   &page_shift, &ncont, &order);
@@ -1401,9 +1398,9 @@ struct ib_mr *mlx5_ib_reg_user_mr(struct ib_pd *pd, u64 start, u64 length,
 		}
 	}
 
-#ifdef CONFIG_INFINIBAND_ON_DEMAND_PAGING
-	mr->live = 1;
-#endif
+	if (IS_ENABLED(CONFIG_INFINIBAND_ON_DEMAND_PAGING))
+		mr->live = 1;
+
 	return &mr->ibmr;
 error:
 	ib_umem_release(umem);
@@ -1518,9 +1515,8 @@ int mlx5_ib_rereg_user_mr(struct ib_mr *ib_mr, int flags, u64 start,
 		}
 
 		mr->allocated_from_cache = 0;
-#ifdef CONFIG_INFINIBAND_ON_DEMAND_PAGING
-		mr->live = 1;
-#endif
+		if (IS_ENABLED(CONFIG_INFINIBAND_ON_DEMAND_PAGING))
+			mr->live = 1;
 	} else {
 		/*
 		 * Send a UMR WQE
diff --git a/include/rdma/ib_umem_odp.h b/include/rdma/ib_umem_odp.h
index 0b1446fe2fab..d3725cf13ecd 100644
--- a/include/rdma/ib_umem_odp.h
+++ b/include/rdma/ib_umem_odp.h
@@ -83,6 +83,19 @@ static inline struct ib_umem_odp *to_ib_umem_odp(struct ib_umem *umem)
 	return container_of(umem, struct ib_umem_odp, umem);
 }
 
+/*
+ * The lower 2 bits of the DMA address signal the R/W permissions for
+ * the entry. To upgrade the permissions, provide the appropriate
+ * bitmask to the map_dma_pages function.
+ *
+ * Be aware that upgrading a mapped address might result in change of
+ * the DMA address for the page.
+ */
+#define ODP_READ_ALLOWED_BIT  (1<<0ULL)
+#define ODP_WRITE_ALLOWED_BIT (1<<1ULL)
+
+#define ODP_DMA_ADDR_MASK (~(ODP_READ_ALLOWED_BIT | ODP_WRITE_ALLOWED_BIT))
+
 #ifdef CONFIG_INFINIBAND_ON_DEMAND_PAGING
 
 struct ib_ucontext_per_mm {
@@ -107,19 +120,6 @@ struct ib_umem_odp *ib_alloc_odp_umem(struct ib_ucontext_per_mm *per_mm,
 				      unsigned long addr, size_t size);
 void ib_umem_odp_release(struct ib_umem_odp *umem_odp);
 
-/*
- * The lower 2 bits of the DMA address signal the R/W permissions for
- * the entry. To upgrade the permissions, provide the appropriate
- * bitmask to the map_dma_pages function.
- *
- * Be aware that upgrading a mapped address might result in change of
- * the DMA address for the page.
- */
-#define ODP_READ_ALLOWED_BIT  (1<<0ULL)
-#define ODP_WRITE_ALLOWED_BIT (1<<1ULL)
-
-#define ODP_DMA_ADDR_MASK (~(ODP_READ_ALLOWED_BIT | ODP_WRITE_ALLOWED_BIT))
-
 int ib_umem_odp_map_dma_pages(struct ib_umem_odp *umem_odp, u64 start_offset,
 			      u64 bcnt, u64 access_mask,
 			      unsigned long current_seq);
-- 
cgit v1.2.3


From 8cbfaac3d0974ecf53b28db23ea8c105f491bb35 Mon Sep 17 00:00:00 2001
From: Leon Romanovsky <leonro@mellanox.com>
Date: Wed, 9 Jan 2019 20:15:58 +0200
Subject: RDMA: Clear PD objects during their allocation

As part of an audit process to update drivers to use rdma_restrack_add()
ensure that PD objects is cleared before access.

Signed-off-by: Leon Romanovsky <leonro@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 drivers/infiniband/hw/hns/hns_roce_pd.c         | 2 +-
 drivers/infiniband/hw/mlx5/main.c               | 2 +-
 drivers/infiniband/hw/mthca/mthca_provider.c    | 2 +-
 drivers/infiniband/hw/vmw_pvrdma/pvrdma_verbs.c | 2 +-
 drivers/infiniband/sw/rdmavt/pd.c               | 2 +-
 5 files changed, 5 insertions(+), 5 deletions(-)

(limited to 'drivers/infiniband/hw/mlx5/main.c')

diff --git a/drivers/infiniband/hw/hns/hns_roce_pd.c b/drivers/infiniband/hw/hns/hns_roce_pd.c
index e11c149da04d..4a29b2cb9bab 100644
--- a/drivers/infiniband/hw/hns/hns_roce_pd.c
+++ b/drivers/infiniband/hw/hns/hns_roce_pd.c
@@ -66,7 +66,7 @@ struct ib_pd *hns_roce_alloc_pd(struct ib_device *ib_dev,
 	struct hns_roce_pd *pd;
 	int ret;
 
-	pd = kmalloc(sizeof(*pd), GFP_KERNEL);
+	pd = kzalloc(sizeof(*pd), GFP_KERNEL);
 	if (!pd)
 		return ERR_PTR(-ENOMEM);
 
diff --git a/drivers/infiniband/hw/mlx5/main.c b/drivers/infiniband/hw/mlx5/main.c
index 11e9783cefcc..717e8535b41a 100644
--- a/drivers/infiniband/hw/mlx5/main.c
+++ b/drivers/infiniband/hw/mlx5/main.c
@@ -2263,7 +2263,7 @@ static struct ib_pd *mlx5_ib_alloc_pd(struct ib_device *ibdev,
 	u32 in[MLX5_ST_SZ_DW(alloc_pd_in)]   = {};
 	u16 uid = 0;
 
-	pd = kmalloc(sizeof(*pd), GFP_KERNEL);
+	pd = kzalloc(sizeof(*pd), GFP_KERNEL);
 	if (!pd)
 		return ERR_PTR(-ENOMEM);
 
diff --git a/drivers/infiniband/hw/mthca/mthca_provider.c b/drivers/infiniband/hw/mthca/mthca_provider.c
index 53fff6aed896..f87ee3058c65 100644
--- a/drivers/infiniband/hw/mthca/mthca_provider.c
+++ b/drivers/infiniband/hw/mthca/mthca_provider.c
@@ -381,7 +381,7 @@ static struct ib_pd *mthca_alloc_pd(struct ib_device *ibdev,
 	struct mthca_pd *pd;
 	int err;
 
-	pd = kmalloc(sizeof *pd, GFP_KERNEL);
+	pd = kzalloc(sizeof(*pd), GFP_KERNEL);
 	if (!pd)
 		return ERR_PTR(-ENOMEM);
 
diff --git a/drivers/infiniband/hw/vmw_pvrdma/pvrdma_verbs.c b/drivers/infiniband/hw/vmw_pvrdma/pvrdma_verbs.c
index 4d238d0e484b..65b6115caa55 100644
--- a/drivers/infiniband/hw/vmw_pvrdma/pvrdma_verbs.c
+++ b/drivers/infiniband/hw/vmw_pvrdma/pvrdma_verbs.c
@@ -457,7 +457,7 @@ struct ib_pd *pvrdma_alloc_pd(struct ib_device *ibdev,
 	if (!atomic_add_unless(&dev->num_pds, 1, dev->dsr->caps.max_pd))
 		return ERR_PTR(-ENOMEM);
 
-	pd = kmalloc(sizeof(*pd), GFP_KERNEL);
+	pd = kzalloc(sizeof(*pd), GFP_KERNEL);
 	if (!pd) {
 		ptr = ERR_PTR(-ENOMEM);
 		goto err;
diff --git a/drivers/infiniband/sw/rdmavt/pd.c b/drivers/infiniband/sw/rdmavt/pd.c
index 8a89afff3363..dcc1870b8d23 100644
--- a/drivers/infiniband/sw/rdmavt/pd.c
+++ b/drivers/infiniband/sw/rdmavt/pd.c
@@ -66,7 +66,7 @@ struct ib_pd *rvt_alloc_pd(struct ib_device *ibdev,
 	struct rvt_pd *pd;
 	struct ib_pd *ret;
 
-	pd = kmalloc(sizeof(*pd), GFP_KERNEL);
+	pd = kzalloc(sizeof(*pd), GFP_KERNEL);
 	if (!pd) {
 		ret = ERR_PTR(-ENOMEM);
 		goto bail;
-- 
cgit v1.2.3


From ea4baf7f116a18382df331db2123d98bc1c3cd83 Mon Sep 17 00:00:00 2001
From: Parav Pandit <parav@mellanox.com>
Date: Tue, 18 Dec 2018 14:28:30 +0200
Subject: RDMA: Rename port_callback to init_port

Most provider routines are callback routines which ib core invokes.
_callback suffix doesn't convey information about when such callback is
invoked. Therefore, rename port_callback to init_port.

Additionally, store the init_port function pointer in ib_device_ops, so
that it can be accessed in subsequent patches when binding rdma device to
net namespace.

Signed-off-by: Parav Pandit <parav@mellanox.com>
Signed-off-by: Leon Romanovsky <leonro@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 drivers/infiniband/core/core_priv.h            |  4 +---
 drivers/infiniband/core/device.c               |  7 +++----
 drivers/infiniband/core/sysfs.c                | 16 ++++++----------
 drivers/infiniband/hw/bnxt_re/main.c           |  2 +-
 drivers/infiniband/hw/cxgb3/iwch_provider.c    |  2 +-
 drivers/infiniband/hw/cxgb4/provider.c         |  2 +-
 drivers/infiniband/hw/hfi1/verbs.c             |  2 +-
 drivers/infiniband/hw/hns/hns_roce_main.c      |  2 +-
 drivers/infiniband/hw/i40iw/i40iw_verbs.c      |  2 +-
 drivers/infiniband/hw/mlx4/main.c              |  2 +-
 drivers/infiniband/hw/mlx5/main.c              |  2 +-
 drivers/infiniband/hw/mthca/mthca_provider.c   |  2 +-
 drivers/infiniband/hw/nes/nes_verbs.c          |  2 +-
 drivers/infiniband/hw/ocrdma/ocrdma_main.c     |  2 +-
 drivers/infiniband/hw/qedr/main.c              |  2 +-
 drivers/infiniband/hw/qib/qib_verbs.c          |  2 +-
 drivers/infiniband/hw/usnic/usnic_ib_main.c    |  2 +-
 drivers/infiniband/hw/vmw_pvrdma/pvrdma_main.c |  2 +-
 drivers/infiniband/sw/rdmavt/vt.c              |  5 ++---
 drivers/infiniband/sw/rxe/rxe_verbs.c          |  2 +-
 include/rdma/ib_verbs.h                        | 10 +++++++---
 include/rdma/rdma_vt.h                         |  3 ---
 22 files changed, 35 insertions(+), 42 deletions(-)

(limited to 'drivers/infiniband/hw/mlx5/main.c')

diff --git a/drivers/infiniband/core/core_priv.h b/drivers/infiniband/core/core_priv.h
index 3cd830d52967..aca75c74e451 100644
--- a/drivers/infiniband/core/core_priv.h
+++ b/drivers/infiniband/core/core_priv.h
@@ -54,9 +54,7 @@ struct pkey_index_qp_list {
 	struct list_head    qp_list;
 };
 
-int  ib_device_register_sysfs(struct ib_device *device,
-			      int (*port_callback)(struct ib_device *,
-						   u8, struct kobject *));
+int ib_device_register_sysfs(struct ib_device *device);
 void ib_device_unregister_sysfs(struct ib_device *device);
 int ib_device_rename(struct ib_device *ibdev, const char *name);
 
diff --git a/drivers/infiniband/core/device.c b/drivers/infiniband/core/device.c
index 8872453e26c0..66867e92ddea 100644
--- a/drivers/infiniband/core/device.c
+++ b/drivers/infiniband/core/device.c
@@ -574,9 +574,7 @@ port_cleanup:
  * callback for each device that is added. @device must be allocated
  * with ib_alloc_device().
  */
-int ib_register_device(struct ib_device *device, const char *name,
-		       int (*port_callback)(struct ib_device *, u8,
-					    struct kobject *))
+int ib_register_device(struct ib_device *device, const char *name)
 {
 	int ret;
 	struct ib_client *client;
@@ -613,7 +611,7 @@ int ib_register_device(struct ib_device *device, const char *name,
 		goto dev_cleanup;
 	}
 
-	ret = ib_device_register_sysfs(device, port_callback);
+	ret = ib_device_register_sysfs(device);
 	if (ret) {
 		dev_warn(&device->dev,
 			 "Couldn't register device with driver model\n");
@@ -1283,6 +1281,7 @@ void ib_set_device_ops(struct ib_device *dev, const struct ib_device_ops *ops)
 	SET_DEVICE_OP(dev_ops, get_vector_affinity);
 	SET_DEVICE_OP(dev_ops, get_vf_config);
 	SET_DEVICE_OP(dev_ops, get_vf_stats);
+	SET_DEVICE_OP(dev_ops, init_port);
 	SET_DEVICE_OP(dev_ops, map_mr_sg);
 	SET_DEVICE_OP(dev_ops, map_phys_fmr);
 	SET_DEVICE_OP(dev_ops, mmap);
diff --git a/drivers/infiniband/core/sysfs.c b/drivers/infiniband/core/sysfs.c
index 80f68eb0ba5c..7a5679933df6 100644
--- a/drivers/infiniband/core/sysfs.c
+++ b/drivers/infiniband/core/sysfs.c
@@ -1015,9 +1015,7 @@ err_free_stats:
 	return;
 }
 
-static int add_port(struct ib_device *device, int port_num,
-		    int (*port_callback)(struct ib_device *,
-					 u8, struct kobject *))
+static int add_port(struct ib_device *device, int port_num)
 {
 	struct ib_port *p;
 	struct ib_port_attr attr;
@@ -1113,8 +1111,8 @@ static int add_port(struct ib_device *device, int port_num,
 	if (ret)
 		goto err_free_pkey;
 
-	if (port_callback) {
-		ret = port_callback(device, port_num, &p->kobj);
+	if (device->ops.init_port) {
+		ret = device->ops.init_port(device, port_num, &p->kobj);
 		if (ret)
 			goto err_remove_pkey;
 	}
@@ -1308,9 +1306,7 @@ static void free_port_list_attributes(struct ib_device *device)
 	kobject_put(device->ports_kobj);
 }
 
-int ib_device_register_sysfs(struct ib_device *device,
-			     int (*port_callback)(struct ib_device *,
-						  u8, struct kobject *))
+int ib_device_register_sysfs(struct ib_device *device)
 {
 	struct device *class_dev = &device->dev;
 	int ret;
@@ -1330,12 +1326,12 @@ int ib_device_register_sysfs(struct ib_device *device,
 	}
 
 	if (rdma_cap_ib_switch(device)) {
-		ret = add_port(device, 0, port_callback);
+		ret = add_port(device, 0);
 		if (ret)
 			goto err_put;
 	} else {
 		for (i = 1; i <= device->phys_port_cnt; ++i) {
-			ret = add_port(device, i, port_callback);
+			ret = add_port(device, i);
 			if (ret)
 				goto err_put;
 		}
diff --git a/drivers/infiniband/hw/bnxt_re/main.c b/drivers/infiniband/hw/bnxt_re/main.c
index e7a997f2a537..797a3e943366 100644
--- a/drivers/infiniband/hw/bnxt_re/main.c
+++ b/drivers/infiniband/hw/bnxt_re/main.c
@@ -662,7 +662,7 @@ static int bnxt_re_register_ib(struct bnxt_re_dev *rdev)
 	rdma_set_device_sysfs_group(ibdev, &bnxt_re_dev_attr_group);
 	ibdev->driver_id = RDMA_DRIVER_BNXT_RE;
 	ib_set_device_ops(ibdev, &bnxt_re_dev_ops);
-	return ib_register_device(ibdev, "bnxt_re%d", NULL);
+	return ib_register_device(ibdev, "bnxt_re%d");
 }
 
 static void bnxt_re_dev_remove(struct bnxt_re_dev *rdev)
diff --git a/drivers/infiniband/hw/cxgb3/iwch_provider.c b/drivers/infiniband/hw/cxgb3/iwch_provider.c
index 92ee6761a3bd..ffdde3cca268 100644
--- a/drivers/infiniband/hw/cxgb3/iwch_provider.c
+++ b/drivers/infiniband/hw/cxgb3/iwch_provider.c
@@ -1409,7 +1409,7 @@ int iwch_register_device(struct iwch_dev *dev)
 	dev->ibdev.driver_id = RDMA_DRIVER_CXGB3;
 	rdma_set_device_sysfs_group(&dev->ibdev, &iwch_attr_group);
 	ib_set_device_ops(&dev->ibdev, &iwch_dev_ops);
-	ret = ib_register_device(&dev->ibdev, "cxgb3_%d", NULL);
+	ret = ib_register_device(&dev->ibdev, "cxgb3_%d");
 	if (ret)
 		kfree(dev->ibdev.iwcm);
 	return ret;
diff --git a/drivers/infiniband/hw/cxgb4/provider.c b/drivers/infiniband/hw/cxgb4/provider.c
index 586b0c37481f..0a99894b0160 100644
--- a/drivers/infiniband/hw/cxgb4/provider.c
+++ b/drivers/infiniband/hw/cxgb4/provider.c
@@ -634,7 +634,7 @@ void c4iw_register_device(struct work_struct *work)
 	rdma_set_device_sysfs_group(&dev->ibdev, &c4iw_attr_group);
 	dev->ibdev.driver_id = RDMA_DRIVER_CXGB4;
 	ib_set_device_ops(&dev->ibdev, &c4iw_dev_ops);
-	ret = ib_register_device(&dev->ibdev, "cxgb4_%d", NULL);
+	ret = ib_register_device(&dev->ibdev, "cxgb4_%d");
 	if (ret)
 		goto err_kfree_iwcm;
 	return;
diff --git a/drivers/infiniband/hw/hfi1/verbs.c b/drivers/infiniband/hw/hfi1/verbs.c
index ec582d86025f..8957adf58af7 100644
--- a/drivers/infiniband/hw/hfi1/verbs.c
+++ b/drivers/infiniband/hw/hfi1/verbs.c
@@ -1622,6 +1622,7 @@ static const struct ib_device_ops hfi1_dev_ops = {
 	.alloc_rdma_netdev = hfi1_vnic_alloc_rn,
 	.get_dev_fw_str = hfi1_get_dev_fw_str,
 	.get_hw_stats = get_hw_stats,
+	.init_port = hfi1_create_port_files,
 	.modify_device = modify_device,
 	/* keep process mad in the driver */
 	.process_mad = hfi1_process_mad,
@@ -1679,7 +1680,6 @@ int hfi1_register_ib_device(struct hfi1_devdata *dd)
 	/*
 	 * Fill in rvt info object.
 	 */
-	dd->verbs_dev.rdi.driver_f.port_callback = hfi1_create_port_files;
 	dd->verbs_dev.rdi.driver_f.get_pci_dev = get_pci_dev;
 	dd->verbs_dev.rdi.driver_f.check_ah = hfi1_check_ah;
 	dd->verbs_dev.rdi.driver_f.notify_new_ah = hfi1_notify_new_ah;
diff --git a/drivers/infiniband/hw/hns/hns_roce_main.c b/drivers/infiniband/hw/hns/hns_roce_main.c
index 44a07fce0617..46ede58ef3b8 100644
--- a/drivers/infiniband/hw/hns/hns_roce_main.c
+++ b/drivers/infiniband/hw/hns/hns_roce_main.c
@@ -564,7 +564,7 @@ static int hns_roce_register_device(struct hns_roce_dev *hr_dev)
 	ib_dev->driver_id = RDMA_DRIVER_HNS;
 	ib_set_device_ops(ib_dev, hr_dev->hw->hns_roce_dev_ops);
 	ib_set_device_ops(ib_dev, &hns_roce_dev_ops);
-	ret = ib_register_device(ib_dev, "hns_%d", NULL);
+	ret = ib_register_device(ib_dev, "hns_%d");
 	if (ret) {
 		dev_err(dev, "ib_register_device failed!\n");
 		return ret;
diff --git a/drivers/infiniband/hw/i40iw/i40iw_verbs.c b/drivers/infiniband/hw/i40iw/i40iw_verbs.c
index 80b66df95362..af66ab9d150b 100644
--- a/drivers/infiniband/hw/i40iw/i40iw_verbs.c
+++ b/drivers/infiniband/hw/i40iw/i40iw_verbs.c
@@ -2868,7 +2868,7 @@ int i40iw_register_rdma_device(struct i40iw_device *iwdev)
 	iwibdev = iwdev->iwibdev;
 	rdma_set_device_sysfs_group(&iwibdev->ibdev, &i40iw_attr_group);
 	iwibdev->ibdev.driver_id = RDMA_DRIVER_I40IW;
-	ret = ib_register_device(&iwibdev->ibdev, "i40iw%d", NULL);
+	ret = ib_register_device(&iwibdev->ibdev, "i40iw%d");
 	if (ret)
 		goto error;
 
diff --git a/drivers/infiniband/hw/mlx4/main.c b/drivers/infiniband/hw/mlx4/main.c
index 1f15ec3e2b83..c3f950d82ed0 100644
--- a/drivers/infiniband/hw/mlx4/main.c
+++ b/drivers/infiniband/hw/mlx4/main.c
@@ -2856,7 +2856,7 @@ static void *mlx4_ib_add(struct mlx4_dev *dev)
 
 	rdma_set_device_sysfs_group(&ibdev->ib_dev, &mlx4_attr_group);
 	ibdev->ib_dev.driver_id = RDMA_DRIVER_MLX4;
-	if (ib_register_device(&ibdev->ib_dev, "mlx4_%d", NULL))
+	if (ib_register_device(&ibdev->ib_dev, "mlx4_%d"))
 		goto err_diag_counters;
 
 	if (mlx4_ib_mad_init(ibdev))
diff --git a/drivers/infiniband/hw/mlx5/main.c b/drivers/infiniband/hw/mlx5/main.c
index 717e8535b41a..948617a60d44 100644
--- a/drivers/infiniband/hw/mlx5/main.c
+++ b/drivers/infiniband/hw/mlx5/main.c
@@ -6233,7 +6233,7 @@ int mlx5_ib_stage_ib_reg_init(struct mlx5_ib_dev *dev)
 		name = "mlx5_%d";
 	else
 		name = "mlx5_bond_%d";
-	return ib_register_device(&dev->ib_dev, name, NULL);
+	return ib_register_device(&dev->ib_dev, name);
 }
 
 void mlx5_ib_stage_pre_ib_reg_umr_cleanup(struct mlx5_ib_dev *dev)
diff --git a/drivers/infiniband/hw/mthca/mthca_provider.c b/drivers/infiniband/hw/mthca/mthca_provider.c
index 0a939ea5497b..3473c6c51b92 100644
--- a/drivers/infiniband/hw/mthca/mthca_provider.c
+++ b/drivers/infiniband/hw/mthca/mthca_provider.c
@@ -1338,7 +1338,7 @@ int mthca_register_device(struct mthca_dev *dev)
 
 	rdma_set_device_sysfs_group(&dev->ib_dev, &mthca_attr_group);
 	dev->ib_dev.driver_id = RDMA_DRIVER_MTHCA;
-	ret = ib_register_device(&dev->ib_dev, "mthca%d", NULL);
+	ret = ib_register_device(&dev->ib_dev, "mthca%d");
 	if (ret)
 		return ret;
 
diff --git a/drivers/infiniband/hw/nes/nes_verbs.c b/drivers/infiniband/hw/nes/nes_verbs.c
index feb4d259aab9..34601f0cbd74 100644
--- a/drivers/infiniband/hw/nes/nes_verbs.c
+++ b/drivers/infiniband/hw/nes/nes_verbs.c
@@ -3801,7 +3801,7 @@ int nes_register_ofa_device(struct nes_ib_device *nesibdev)
 
 	rdma_set_device_sysfs_group(&nesvnic->nesibdev->ibdev, &nes_attr_group);
 	nesvnic->nesibdev->ibdev.driver_id = RDMA_DRIVER_NES;
-	ret = ib_register_device(&nesvnic->nesibdev->ibdev, "nes%d", NULL);
+	ret = ib_register_device(&nesvnic->nesibdev->ibdev, "nes%d");
 	if (ret) {
 		return ret;
 	}
diff --git a/drivers/infiniband/hw/ocrdma/ocrdma_main.c b/drivers/infiniband/hw/ocrdma/ocrdma_main.c
index 1f393842453a..f45b996f617f 100644
--- a/drivers/infiniband/hw/ocrdma/ocrdma_main.c
+++ b/drivers/infiniband/hw/ocrdma/ocrdma_main.c
@@ -243,7 +243,7 @@ static int ocrdma_register_device(struct ocrdma_dev *dev)
 	}
 	rdma_set_device_sysfs_group(&dev->ibdev, &ocrdma_attr_group);
 	dev->ibdev.driver_id = RDMA_DRIVER_OCRDMA;
-	return ib_register_device(&dev->ibdev, "ocrdma%d", NULL);
+	return ib_register_device(&dev->ibdev, "ocrdma%d");
 }
 
 static int ocrdma_alloc_resources(struct ocrdma_dev *dev)
diff --git a/drivers/infiniband/hw/qedr/main.c b/drivers/infiniband/hw/qedr/main.c
index 75940e2a8791..8e5c76d06855 100644
--- a/drivers/infiniband/hw/qedr/main.c
+++ b/drivers/infiniband/hw/qedr/main.c
@@ -290,7 +290,7 @@ static int qedr_register_device(struct qedr_dev *dev)
 	ib_set_device_ops(&dev->ibdev, &qedr_dev_ops);
 
 	dev->ibdev.driver_id = RDMA_DRIVER_QEDR;
-	return ib_register_device(&dev->ibdev, "qedr%d", NULL);
+	return ib_register_device(&dev->ibdev, "qedr%d");
 }
 
 /* This function allocates fast-path status block memory */
diff --git a/drivers/infiniband/hw/qib/qib_verbs.c b/drivers/infiniband/hw/qib/qib_verbs.c
index 276304f611ab..ff8dab8e2344 100644
--- a/drivers/infiniband/hw/qib/qib_verbs.c
+++ b/drivers/infiniband/hw/qib/qib_verbs.c
@@ -1494,6 +1494,7 @@ static void qib_fill_device_attr(struct qib_devdata *dd)
 }
 
 static const struct ib_device_ops qib_dev_ops = {
+	.init_port = qib_create_port_files,
 	.modify_device = qib_modify_device,
 	.process_mad = qib_process_mad,
 };
@@ -1567,7 +1568,6 @@ int qib_register_ib_device(struct qib_devdata *dd)
 	/*
 	 * Fill in rvt info object.
 	 */
-	dd->verbs_dev.rdi.driver_f.port_callback = qib_create_port_files;
 	dd->verbs_dev.rdi.driver_f.get_pci_dev = qib_get_pci_dev;
 	dd->verbs_dev.rdi.driver_f.check_ah = qib_check_ah;
 	dd->verbs_dev.rdi.driver_f.setup_wqe = qib_check_send_wqe;
diff --git a/drivers/infiniband/hw/usnic/usnic_ib_main.c b/drivers/infiniband/hw/usnic/usnic_ib_main.c
index b2323a52a0dd..3201dd1899c7 100644
--- a/drivers/infiniband/hw/usnic/usnic_ib_main.c
+++ b/drivers/infiniband/hw/usnic/usnic_ib_main.c
@@ -422,7 +422,7 @@ static void *usnic_ib_device_add(struct pci_dev *dev)
 	us_ibdev->ib_dev.driver_id = RDMA_DRIVER_USNIC;
 	rdma_set_device_sysfs_group(&us_ibdev->ib_dev, &usnic_attr_group);
 
-	if (ib_register_device(&us_ibdev->ib_dev, "usnic_%d", NULL))
+	if (ib_register_device(&us_ibdev->ib_dev, "usnic_%d"))
 		goto err_fwd_dealloc;
 
 	usnic_fwd_set_mtu(us_ibdev->ufdev, us_ibdev->netdev->mtu);
diff --git a/drivers/infiniband/hw/vmw_pvrdma/pvrdma_main.c b/drivers/infiniband/hw/vmw_pvrdma/pvrdma_main.c
index eaa109dbc96a..1bc415483d9b 100644
--- a/drivers/infiniband/hw/vmw_pvrdma/pvrdma_main.c
+++ b/drivers/infiniband/hw/vmw_pvrdma/pvrdma_main.c
@@ -278,7 +278,7 @@ static int pvrdma_register_device(struct pvrdma_dev *dev)
 	spin_lock_init(&dev->srq_tbl_lock);
 	rdma_set_device_sysfs_group(&dev->ib_dev, &pvrdma_attr_group);
 
-	ret = ib_register_device(&dev->ib_dev, "vmw_pvrdma%d", NULL);
+	ret = ib_register_device(&dev->ib_dev, "vmw_pvrdma%d");
 	if (ret)
 		goto err_srq_free;
 
diff --git a/drivers/infiniband/sw/rdmavt/vt.c b/drivers/infiniband/sw/rdmavt/vt.c
index 8724a1817158..7de7389d0235 100644
--- a/drivers/infiniband/sw/rdmavt/vt.c
+++ b/drivers/infiniband/sw/rdmavt/vt.c
@@ -446,7 +446,7 @@ static noinline int check_support(struct rvt_dev_info *rdi, int verb)
 		 * These functions are not part of verbs specifically but are
 		 * required for rdmavt to function.
 		 */
-		if ((!rdi->driver_f.port_callback) ||
+		if ((!rdi->ibdev.ops.init_port) ||
 		    (!rdi->driver_f.get_pci_dev))
 			return -EINVAL;
 		break;
@@ -644,8 +644,7 @@ int rvt_register_device(struct rvt_dev_info *rdi, u32 driver_id)
 
 	rdi->ibdev.driver_id = driver_id;
 	/* We are now good to announce we exist */
-	ret = ib_register_device(&rdi->ibdev, dev_name(&rdi->ibdev.dev),
-				 rdi->driver_f.port_callback);
+	ret = ib_register_device(&rdi->ibdev, dev_name(&rdi->ibdev.dev));
 	if (ret) {
 		rvt_pr_err(rdi, "Failed to register driver with ib core.\n");
 		goto bail_wss;
diff --git a/drivers/infiniband/sw/rxe/rxe_verbs.c b/drivers/infiniband/sw/rxe/rxe_verbs.c
index b20e6e0415f5..43171148e9c5 100644
--- a/drivers/infiniband/sw/rxe/rxe_verbs.c
+++ b/drivers/infiniband/sw/rxe/rxe_verbs.c
@@ -1258,7 +1258,7 @@ int rxe_register_device(struct rxe_dev *rxe)
 
 	rdma_set_device_sysfs_group(dev, &rxe_attr_group);
 	dev->driver_id = RDMA_DRIVER_RXE;
-	err = ib_register_device(dev, "rxe%d", NULL);
+	err = ib_register_device(dev, "rxe%d");
 	if (err) {
 		pr_warn("%s failed with error %d\n", __func__, err);
 		goto err1;
diff --git a/include/rdma/ib_verbs.h b/include/rdma/ib_verbs.h
index aa1f126d3383..1d1902fd9f87 100644
--- a/include/rdma/ib_verbs.h
+++ b/include/rdma/ib_verbs.h
@@ -2504,6 +2504,12 @@ struct ib_device_ops {
 	 */
 	int (*get_hw_stats)(struct ib_device *device,
 			    struct rdma_hw_stats *stats, u8 port, int index);
+	/*
+	 * This function is called once for each port when a ib device is
+	 * registered.
+	 */
+	int (*init_port)(struct ib_device *device, u8 port_num,
+			 struct kobject *port_sysfs);
 };
 
 struct ib_device {
@@ -2620,9 +2626,7 @@ void ib_dealloc_device(struct ib_device *device);
 
 void ib_get_device_fw_str(struct ib_device *device, char *str);
 
-int ib_register_device(struct ib_device *device, const char *name,
-		       int (*port_callback)(struct ib_device *, u8,
-					    struct kobject *));
+int ib_register_device(struct ib_device *device, const char *name);
 void ib_unregister_device(struct ib_device *device);
 
 int ib_register_client   (struct ib_client *client);
diff --git a/include/rdma/rdma_vt.h b/include/rdma/rdma_vt.h
index dd0ed8048bb4..acb3bc96dfa7 100644
--- a/include/rdma/rdma_vt.h
+++ b/include/rdma/rdma_vt.h
@@ -250,9 +250,6 @@ struct rvt_driver_provided {
 	 */
 	void (*do_send)(struct rvt_qp *qp);
 
-	/* Passed to ib core registration. Callback to create syfs files */
-	int (*port_callback)(struct ib_device *, u8, struct kobject *);
-
 	/*
 	 * Returns a pointer to the undelying hardware's PCI device. This is
 	 * used to display information as to what hardware is being referenced
-- 
cgit v1.2.3


From 54747231150f0dddf68f2ee29ec2970fcc433909 Mon Sep 17 00:00:00 2001
From: Parav Pandit <parav@mellanox.com>
Date: Tue, 18 Dec 2018 14:15:56 +0200
Subject: RDMA: Introduce and use rdma_device_to_ibdev()

Introduce and use rdma_device_to_ibdev() API for those drivers which are
registering one sysfs group and also use in ib_core.

In subsequent patch, device->provider_ibdev one-to-one mapping is no
longer holds true during accessing sysfs entries.
Therefore, introduce an API rdma_device_to_ibdev() that provides such
information.

Signed-off-by: Parav Pandit <parav@mellanox.com>
Signed-off-by: Leon Romanovsky <leonro@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 drivers/infiniband/core/device.c             |  2 --
 drivers/infiniband/core/sysfs.c              | 12 ++++++------
 drivers/infiniband/hw/bnxt_re/main.c         |  6 ++++--
 drivers/infiniband/hw/cxgb3/iwch_provider.c  | 14 ++++++++------
 drivers/infiniband/hw/cxgb4/provider.c       | 14 ++++++++------
 drivers/infiniband/hw/hfi1/sysfs.c           | 16 ++++++++--------
 drivers/infiniband/hw/i40iw/i40iw_verbs.c    |  5 ++---
 drivers/infiniband/hw/mlx4/main.c            |  7 ++++---
 drivers/infiniband/hw/mlx5/main.c            | 13 ++++++++-----
 drivers/infiniband/hw/mthca/mthca_provider.c |  9 ++++++---
 drivers/infiniband/hw/nes/nes_verbs.c        |  2 +-
 drivers/infiniband/hw/ocrdma/ocrdma_main.c   |  6 ++++--
 drivers/infiniband/hw/qedr/main.c            |  3 ++-
 drivers/infiniband/hw/qib/qib_sysfs.c        | 18 +++++++++---------
 drivers/infiniband/hw/usnic/usnic_ib_sysfs.c | 26 +++++++++++---------------
 drivers/infiniband/sw/rxe/rxe_verbs.c        |  4 ++--
 include/rdma/ib_verbs.h                      | 23 +++++++++++++++++++++++
 17 files changed, 106 insertions(+), 74 deletions(-)

(limited to 'drivers/infiniband/hw/mlx5/main.c')

diff --git a/drivers/infiniband/core/device.c b/drivers/infiniband/core/device.c
index 66867e92ddea..f8180cf1a004 100644
--- a/drivers/infiniband/core/device.c
+++ b/drivers/infiniband/core/device.c
@@ -296,8 +296,6 @@ struct ib_device *ib_alloc_device(size_t size)
 	device->dev.class = &ib_class;
 	device_initialize(&device->dev);
 
-	dev_set_drvdata(&device->dev, device);
-
 	INIT_LIST_HEAD(&device->event_handler_list);
 	spin_lock_init(&device->event_handler_lock);
 	rwlock_init(&device->client_data_lock);
diff --git a/drivers/infiniband/core/sysfs.c b/drivers/infiniband/core/sysfs.c
index 7a5679933df6..c75692802da8 100644
--- a/drivers/infiniband/core/sysfs.c
+++ b/drivers/infiniband/core/sysfs.c
@@ -1187,7 +1187,7 @@ err_put:
 static ssize_t node_type_show(struct device *device,
 			      struct device_attribute *attr, char *buf)
 {
-	struct ib_device *dev = container_of(device, struct ib_device, dev);
+	struct ib_device *dev = rdma_device_to_ibdev(device);
 
 	switch (dev->node_type) {
 	case RDMA_NODE_IB_CA:	  return sprintf(buf, "%d: CA\n", dev->node_type);
@@ -1204,7 +1204,7 @@ static DEVICE_ATTR_RO(node_type);
 static ssize_t sys_image_guid_show(struct device *device,
 				   struct device_attribute *dev_attr, char *buf)
 {
-	struct ib_device *dev = container_of(device, struct ib_device, dev);
+	struct ib_device *dev = rdma_device_to_ibdev(device);
 
 	return sprintf(buf, "%04x:%04x:%04x:%04x\n",
 		       be16_to_cpu(((__be16 *) &dev->attrs.sys_image_guid)[0]),
@@ -1217,7 +1217,7 @@ static DEVICE_ATTR_RO(sys_image_guid);
 static ssize_t node_guid_show(struct device *device,
 			      struct device_attribute *attr, char *buf)
 {
-	struct ib_device *dev = container_of(device, struct ib_device, dev);
+	struct ib_device *dev = rdma_device_to_ibdev(device);
 
 	return sprintf(buf, "%04x:%04x:%04x:%04x\n",
 		       be16_to_cpu(((__be16 *) &dev->node_guid)[0]),
@@ -1230,7 +1230,7 @@ static DEVICE_ATTR_RO(node_guid);
 static ssize_t node_desc_show(struct device *device,
 			      struct device_attribute *attr, char *buf)
 {
-	struct ib_device *dev = container_of(device, struct ib_device, dev);
+	struct ib_device *dev = rdma_device_to_ibdev(device);
 
 	return sprintf(buf, "%.64s\n", dev->node_desc);
 }
@@ -1239,7 +1239,7 @@ static ssize_t node_desc_store(struct device *device,
 			       struct device_attribute *attr,
 			       const char *buf, size_t count)
 {
-	struct ib_device *dev = container_of(device, struct ib_device, dev);
+	struct ib_device *dev = rdma_device_to_ibdev(device);
 	struct ib_device_modify desc = {};
 	int ret;
 
@@ -1258,7 +1258,7 @@ static DEVICE_ATTR_RW(node_desc);
 static ssize_t fw_ver_show(struct device *device, struct device_attribute *attr,
 			   char *buf)
 {
-	struct ib_device *dev = container_of(device, struct ib_device, dev);
+	struct ib_device *dev = rdma_device_to_ibdev(device);
 
 	ib_get_device_fw_str(dev, buf);
 	strlcat(buf, "\n", IB_FW_VERSION_NAME_MAX);
diff --git a/drivers/infiniband/hw/bnxt_re/main.c b/drivers/infiniband/hw/bnxt_re/main.c
index 797a3e943366..16eecfa5882c 100644
--- a/drivers/infiniband/hw/bnxt_re/main.c
+++ b/drivers/infiniband/hw/bnxt_re/main.c
@@ -538,7 +538,8 @@ static struct bnxt_en_dev *bnxt_re_dev_probe(struct net_device *netdev)
 static ssize_t hw_rev_show(struct device *device, struct device_attribute *attr,
 			   char *buf)
 {
-	struct bnxt_re_dev *rdev = to_bnxt_re_dev(device, ibdev.dev);
+	struct bnxt_re_dev *rdev =
+		rdma_device_to_drv_device(device, struct bnxt_re_dev, ibdev);
 
 	return scnprintf(buf, PAGE_SIZE, "0x%x\n", rdev->en_dev->pdev->vendor);
 }
@@ -547,7 +548,8 @@ static DEVICE_ATTR_RO(hw_rev);
 static ssize_t hca_type_show(struct device *device,
 			     struct device_attribute *attr, char *buf)
 {
-	struct bnxt_re_dev *rdev = to_bnxt_re_dev(device, ibdev.dev);
+	struct bnxt_re_dev *rdev =
+		rdma_device_to_drv_device(device, struct bnxt_re_dev, ibdev);
 
 	return scnprintf(buf, PAGE_SIZE, "%s\n", rdev->ibdev.node_desc);
 }
diff --git a/drivers/infiniband/hw/cxgb3/iwch_provider.c b/drivers/infiniband/hw/cxgb3/iwch_provider.c
index ffdde3cca268..07c20cd07f33 100644
--- a/drivers/infiniband/hw/cxgb3/iwch_provider.c
+++ b/drivers/infiniband/hw/cxgb3/iwch_provider.c
@@ -1130,8 +1130,9 @@ static int iwch_query_port(struct ib_device *ibdev,
 static ssize_t hw_rev_show(struct device *dev,
 			   struct device_attribute *attr, char *buf)
 {
-	struct iwch_dev *iwch_dev = container_of(dev, struct iwch_dev,
-						 ibdev.dev);
+	struct iwch_dev *iwch_dev =
+			rdma_device_to_drv_device(dev, struct iwch_dev, ibdev);
+
 	pr_debug("%s dev 0x%p\n", __func__, dev);
 	return sprintf(buf, "%d\n", iwch_dev->rdev.t3cdev_p->type);
 }
@@ -1140,8 +1141,8 @@ static DEVICE_ATTR_RO(hw_rev);
 static ssize_t hca_type_show(struct device *dev,
 			     struct device_attribute *attr, char *buf)
 {
-	struct iwch_dev *iwch_dev = container_of(dev, struct iwch_dev,
-						 ibdev.dev);
+	struct iwch_dev *iwch_dev =
+			rdma_device_to_drv_device(dev, struct iwch_dev, ibdev);
 	struct ethtool_drvinfo info;
 	struct net_device *lldev = iwch_dev->rdev.t3cdev_p->lldev;
 
@@ -1154,8 +1155,9 @@ static DEVICE_ATTR_RO(hca_type);
 static ssize_t board_id_show(struct device *dev,
 			     struct device_attribute *attr, char *buf)
 {
-	struct iwch_dev *iwch_dev = container_of(dev, struct iwch_dev,
-						 ibdev.dev);
+	struct iwch_dev *iwch_dev =
+			rdma_device_to_drv_device(dev, struct iwch_dev, ibdev);
+
 	pr_debug("%s dev 0x%p\n", __func__, dev);
 	return sprintf(buf, "%x.%x\n", iwch_dev->rdev.rnic_info.pdev->vendor,
 		       iwch_dev->rdev.rnic_info.pdev->device);
diff --git a/drivers/infiniband/hw/cxgb4/provider.c b/drivers/infiniband/hw/cxgb4/provider.c
index 0a99894b0160..f977f8e7e162 100644
--- a/drivers/infiniband/hw/cxgb4/provider.c
+++ b/drivers/infiniband/hw/cxgb4/provider.c
@@ -376,8 +376,9 @@ static int c4iw_query_port(struct ib_device *ibdev, u8 port,
 static ssize_t hw_rev_show(struct device *dev,
 			   struct device_attribute *attr, char *buf)
 {
-	struct c4iw_dev *c4iw_dev = container_of(dev, struct c4iw_dev,
-						 ibdev.dev);
+	struct c4iw_dev *c4iw_dev =
+			rdma_device_to_drv_device(dev, struct c4iw_dev, ibdev);
+
 	pr_debug("dev 0x%p\n", dev);
 	return sprintf(buf, "%d\n",
 		       CHELSIO_CHIP_RELEASE(c4iw_dev->rdev.lldi.adapter_type));
@@ -387,8 +388,8 @@ static DEVICE_ATTR_RO(hw_rev);
 static ssize_t hca_type_show(struct device *dev,
 			     struct device_attribute *attr, char *buf)
 {
-	struct c4iw_dev *c4iw_dev = container_of(dev, struct c4iw_dev,
-						 ibdev.dev);
+	struct c4iw_dev *c4iw_dev =
+			rdma_device_to_drv_device(dev, struct c4iw_dev, ibdev);
 	struct ethtool_drvinfo info;
 	struct net_device *lldev = c4iw_dev->rdev.lldi.ports[0];
 
@@ -401,8 +402,9 @@ static DEVICE_ATTR_RO(hca_type);
 static ssize_t board_id_show(struct device *dev, struct device_attribute *attr,
 			     char *buf)
 {
-	struct c4iw_dev *c4iw_dev = container_of(dev, struct c4iw_dev,
-						 ibdev.dev);
+	struct c4iw_dev *c4iw_dev =
+			rdma_device_to_drv_device(dev, struct c4iw_dev, ibdev);
+
 	pr_debug("dev 0x%p\n", dev);
 	return sprintf(buf, "%x.%x\n", c4iw_dev->rdev.lldi.pdev->vendor,
 		       c4iw_dev->rdev.lldi.pdev->device);
diff --git a/drivers/infiniband/hw/hfi1/sysfs.c b/drivers/infiniband/hw/hfi1/sysfs.c
index 2be513d4c9da..90f62c4bddba 100644
--- a/drivers/infiniband/hw/hfi1/sysfs.c
+++ b/drivers/infiniband/hw/hfi1/sysfs.c
@@ -498,7 +498,7 @@ static ssize_t hw_rev_show(struct device *device, struct device_attribute *attr,
 			   char *buf)
 {
 	struct hfi1_ibdev *dev =
-		container_of(device, struct hfi1_ibdev, rdi.ibdev.dev);
+		rdma_device_to_drv_device(device, struct hfi1_ibdev, rdi.ibdev);
 
 	return sprintf(buf, "%x\n", dd_from_dev(dev)->minrev);
 }
@@ -508,7 +508,7 @@ static ssize_t board_id_show(struct device *device,
 			     struct device_attribute *attr, char *buf)
 {
 	struct hfi1_ibdev *dev =
-		container_of(device, struct hfi1_ibdev, rdi.ibdev.dev);
+		rdma_device_to_drv_device(device, struct hfi1_ibdev, rdi.ibdev);
 	struct hfi1_devdata *dd = dd_from_dev(dev);
 	int ret;
 
@@ -524,7 +524,7 @@ static ssize_t boardversion_show(struct device *device,
 				 struct device_attribute *attr, char *buf)
 {
 	struct hfi1_ibdev *dev =
-		container_of(device, struct hfi1_ibdev, rdi.ibdev.dev);
+		rdma_device_to_drv_device(device, struct hfi1_ibdev, rdi.ibdev);
 	struct hfi1_devdata *dd = dd_from_dev(dev);
 
 	/* The string printed here is already newline-terminated. */
@@ -536,7 +536,7 @@ static ssize_t nctxts_show(struct device *device,
 			   struct device_attribute *attr, char *buf)
 {
 	struct hfi1_ibdev *dev =
-		container_of(device, struct hfi1_ibdev, rdi.ibdev.dev);
+		rdma_device_to_drv_device(device, struct hfi1_ibdev, rdi.ibdev);
 	struct hfi1_devdata *dd = dd_from_dev(dev);
 
 	/*
@@ -555,7 +555,7 @@ static ssize_t nfreectxts_show(struct device *device,
 			       struct device_attribute *attr, char *buf)
 {
 	struct hfi1_ibdev *dev =
-		container_of(device, struct hfi1_ibdev, rdi.ibdev.dev);
+		rdma_device_to_drv_device(device, struct hfi1_ibdev, rdi.ibdev);
 	struct hfi1_devdata *dd = dd_from_dev(dev);
 
 	/* Return the number of free user ports (contexts) available. */
@@ -567,7 +567,7 @@ static ssize_t serial_show(struct device *device,
 			   struct device_attribute *attr, char *buf)
 {
 	struct hfi1_ibdev *dev =
-		container_of(device, struct hfi1_ibdev, rdi.ibdev.dev);
+		rdma_device_to_drv_device(device, struct hfi1_ibdev, rdi.ibdev);
 	struct hfi1_devdata *dd = dd_from_dev(dev);
 
 	return scnprintf(buf, PAGE_SIZE, "%s", dd->serial);
@@ -579,7 +579,7 @@ static ssize_t chip_reset_store(struct device *device,
 				size_t count)
 {
 	struct hfi1_ibdev *dev =
-		container_of(device, struct hfi1_ibdev, rdi.ibdev.dev);
+		rdma_device_to_drv_device(device, struct hfi1_ibdev, rdi.ibdev);
 	struct hfi1_devdata *dd = dd_from_dev(dev);
 	int ret;
 
@@ -609,7 +609,7 @@ static ssize_t tempsense_show(struct device *device,
 			      struct device_attribute *attr, char *buf)
 {
 	struct hfi1_ibdev *dev =
-		container_of(device, struct hfi1_ibdev, rdi.ibdev.dev);
+		rdma_device_to_drv_device(device, struct hfi1_ibdev, rdi.ibdev);
 	struct hfi1_devdata *dd = dd_from_dev(dev);
 	struct hfi1_temp temp;
 	int ret;
diff --git a/drivers/infiniband/hw/i40iw/i40iw_verbs.c b/drivers/infiniband/hw/i40iw/i40iw_verbs.c
index af66ab9d150b..12b31a8440be 100644
--- a/drivers/infiniband/hw/i40iw/i40iw_verbs.c
+++ b/drivers/infiniband/hw/i40iw/i40iw_verbs.c
@@ -2139,9 +2139,8 @@ static int i40iw_dereg_mr(struct ib_mr *ib_mr)
 static ssize_t hw_rev_show(struct device *dev,
 			   struct device_attribute *attr, char *buf)
 {
-	struct i40iw_ib_device *iwibdev = container_of(dev,
-						       struct i40iw_ib_device,
-						       ibdev.dev);
+	struct i40iw_ib_device *iwibdev =
+		rdma_device_to_drv_device(dev, struct i40iw_ib_device, ibdev);
 	u32 hw_rev = iwibdev->iwdev->sc_dev.hw_rev;
 
 	return sprintf(buf, "%x\n", hw_rev);
diff --git a/drivers/infiniband/hw/mlx4/main.c b/drivers/infiniband/hw/mlx4/main.c
index c3f950d82ed0..dc2ffd293a11 100644
--- a/drivers/infiniband/hw/mlx4/main.c
+++ b/drivers/infiniband/hw/mlx4/main.c
@@ -2043,7 +2043,7 @@ static ssize_t hca_type_show(struct device *device,
 			     struct device_attribute *attr, char *buf)
 {
 	struct mlx4_ib_dev *dev =
-		container_of(device, struct mlx4_ib_dev, ib_dev.dev);
+		rdma_device_to_drv_device(device, struct mlx4_ib_dev, ib_dev);
 	return sprintf(buf, "MT%d\n", dev->dev->persist->pdev->device);
 }
 static DEVICE_ATTR_RO(hca_type);
@@ -2052,7 +2052,7 @@ static ssize_t hw_rev_show(struct device *device,
 			   struct device_attribute *attr, char *buf)
 {
 	struct mlx4_ib_dev *dev =
-		container_of(device, struct mlx4_ib_dev, ib_dev.dev);
+		rdma_device_to_drv_device(device, struct mlx4_ib_dev, ib_dev);
 	return sprintf(buf, "%x\n", dev->dev->rev_id);
 }
 static DEVICE_ATTR_RO(hw_rev);
@@ -2061,7 +2061,8 @@ static ssize_t board_id_show(struct device *device,
 			     struct device_attribute *attr, char *buf)
 {
 	struct mlx4_ib_dev *dev =
-		container_of(device, struct mlx4_ib_dev, ib_dev.dev);
+		rdma_device_to_drv_device(device, struct mlx4_ib_dev, ib_dev);
+
 	return sprintf(buf, "%.*s\n", MLX4_BOARD_ID_LEN,
 		       dev->dev->board_id);
 }
diff --git a/drivers/infiniband/hw/mlx5/main.c b/drivers/infiniband/hw/mlx5/main.c
index 948617a60d44..4b1b56d54301 100644
--- a/drivers/infiniband/hw/mlx5/main.c
+++ b/drivers/infiniband/hw/mlx5/main.c
@@ -4104,7 +4104,7 @@ static ssize_t fw_pages_show(struct device *device,
 			     struct device_attribute *attr, char *buf)
 {
 	struct mlx5_ib_dev *dev =
-		container_of(device, struct mlx5_ib_dev, ib_dev.dev);
+		rdma_device_to_drv_device(device, struct mlx5_ib_dev, ib_dev);
 
 	return sprintf(buf, "%d\n", dev->mdev->priv.fw_pages);
 }
@@ -4114,7 +4114,7 @@ static ssize_t reg_pages_show(struct device *device,
 			      struct device_attribute *attr, char *buf)
 {
 	struct mlx5_ib_dev *dev =
-		container_of(device, struct mlx5_ib_dev, ib_dev.dev);
+		rdma_device_to_drv_device(device, struct mlx5_ib_dev, ib_dev);
 
 	return sprintf(buf, "%d\n", atomic_read(&dev->mdev->priv.reg_pages));
 }
@@ -4124,7 +4124,8 @@ static ssize_t hca_type_show(struct device *device,
 			     struct device_attribute *attr, char *buf)
 {
 	struct mlx5_ib_dev *dev =
-		container_of(device, struct mlx5_ib_dev, ib_dev.dev);
+		rdma_device_to_drv_device(device, struct mlx5_ib_dev, ib_dev);
+
 	return sprintf(buf, "MT%d\n", dev->mdev->pdev->device);
 }
 static DEVICE_ATTR_RO(hca_type);
@@ -4133,7 +4134,8 @@ static ssize_t hw_rev_show(struct device *device,
 			   struct device_attribute *attr, char *buf)
 {
 	struct mlx5_ib_dev *dev =
-		container_of(device, struct mlx5_ib_dev, ib_dev.dev);
+		rdma_device_to_drv_device(device, struct mlx5_ib_dev, ib_dev);
+
 	return sprintf(buf, "%x\n", dev->mdev->rev_id);
 }
 static DEVICE_ATTR_RO(hw_rev);
@@ -4142,7 +4144,8 @@ static ssize_t board_id_show(struct device *device,
 			     struct device_attribute *attr, char *buf)
 {
 	struct mlx5_ib_dev *dev =
-		container_of(device, struct mlx5_ib_dev, ib_dev.dev);
+		rdma_device_to_drv_device(device, struct mlx5_ib_dev, ib_dev);
+
 	return sprintf(buf, "%.*s\n", MLX5_BOARD_ID_LEN,
 		       dev->mdev->board_id);
 }
diff --git a/drivers/infiniband/hw/mthca/mthca_provider.c b/drivers/infiniband/hw/mthca/mthca_provider.c
index 3473c6c51b92..63003b4d2485 100644
--- a/drivers/infiniband/hw/mthca/mthca_provider.c
+++ b/drivers/infiniband/hw/mthca/mthca_provider.c
@@ -1081,7 +1081,8 @@ static ssize_t hw_rev_show(struct device *device,
 			   struct device_attribute *attr, char *buf)
 {
 	struct mthca_dev *dev =
-		container_of(device, struct mthca_dev, ib_dev.dev);
+		rdma_device_to_drv_device(device, struct mthca_dev, ib_dev);
+
 	return sprintf(buf, "%x\n", dev->rev_id);
 }
 static DEVICE_ATTR_RO(hw_rev);
@@ -1090,7 +1091,8 @@ static ssize_t hca_type_show(struct device *device,
 			     struct device_attribute *attr, char *buf)
 {
 	struct mthca_dev *dev =
-		container_of(device, struct mthca_dev, ib_dev.dev);
+		rdma_device_to_drv_device(device, struct mthca_dev, ib_dev);
+
 	switch (dev->pdev->device) {
 	case PCI_DEVICE_ID_MELLANOX_TAVOR:
 		return sprintf(buf, "MT23108\n");
@@ -1111,7 +1113,8 @@ static ssize_t board_id_show(struct device *device,
 			     struct device_attribute *attr, char *buf)
 {
 	struct mthca_dev *dev =
-		container_of(device, struct mthca_dev, ib_dev.dev);
+		rdma_device_to_drv_device(device, struct mthca_dev, ib_dev);
+
 	return sprintf(buf, "%.*s\n", MTHCA_BOARD_ID_LEN, dev->board_id);
 }
 static DEVICE_ATTR_RO(board_id);
diff --git a/drivers/infiniband/hw/nes/nes_verbs.c b/drivers/infiniband/hw/nes/nes_verbs.c
index 34601f0cbd74..034156f7e9ed 100644
--- a/drivers/infiniband/hw/nes/nes_verbs.c
+++ b/drivers/infiniband/hw/nes/nes_verbs.c
@@ -2560,7 +2560,7 @@ static ssize_t hw_rev_show(struct device *dev,
 			   struct device_attribute *attr, char *buf)
 {
 	struct nes_ib_device *nesibdev =
-			container_of(dev, struct nes_ib_device, ibdev.dev);
+		rdma_device_to_drv_device(dev, struct nes_ib_device, ibdev);
 	struct nes_vnic *nesvnic = nesibdev->nesvnic;
 
 	nes_debug(NES_DBG_INIT, "\n");
diff --git a/drivers/infiniband/hw/ocrdma/ocrdma_main.c b/drivers/infiniband/hw/ocrdma/ocrdma_main.c
index f45b996f617f..b0491b9ecfe4 100644
--- a/drivers/infiniband/hw/ocrdma/ocrdma_main.c
+++ b/drivers/infiniband/hw/ocrdma/ocrdma_main.c
@@ -118,7 +118,8 @@ static void get_dev_fw_str(struct ib_device *device, char *str)
 static ssize_t hw_rev_show(struct device *device,
 			   struct device_attribute *attr, char *buf)
 {
-	struct ocrdma_dev *dev = dev_get_drvdata(device);
+	struct ocrdma_dev *dev =
+		rdma_device_to_drv_device(device, struct ocrdma_dev, ibdev);
 
 	return scnprintf(buf, PAGE_SIZE, "0x%x\n", dev->nic_info.pdev->vendor);
 }
@@ -127,7 +128,8 @@ static DEVICE_ATTR_RO(hw_rev);
 static ssize_t hca_type_show(struct device *device,
 			     struct device_attribute *attr, char *buf)
 {
-	struct ocrdma_dev *dev = dev_get_drvdata(device);
+	struct ocrdma_dev *dev =
+		rdma_device_to_drv_device(device, struct ocrdma_dev, ibdev);
 
 	return scnprintf(buf, PAGE_SIZE, "%s\n", &dev->model_number[0]);
 }
diff --git a/drivers/infiniband/hw/qedr/main.c b/drivers/infiniband/hw/qedr/main.c
index 8e5c76d06855..f85e72b65a10 100644
--- a/drivers/infiniband/hw/qedr/main.c
+++ b/drivers/infiniband/hw/qedr/main.c
@@ -137,7 +137,8 @@ static int qedr_iw_port_immutable(struct ib_device *ibdev, u8 port_num,
 static ssize_t hw_rev_show(struct device *device, struct device_attribute *attr,
 			   char *buf)
 {
-	struct qedr_dev *dev = dev_get_drvdata(device);
+	struct qedr_dev *dev =
+		rdma_device_to_drv_device(device, struct qedr_dev, ibdev);
 
 	return scnprintf(buf, PAGE_SIZE, "0x%x\n", dev->pdev->vendor);
 }
diff --git a/drivers/infiniband/hw/qib/qib_sysfs.c b/drivers/infiniband/hw/qib/qib_sysfs.c
index 1cf4ca3f23e3..905206a0c2d5 100644
--- a/drivers/infiniband/hw/qib/qib_sysfs.c
+++ b/drivers/infiniband/hw/qib/qib_sysfs.c
@@ -555,7 +555,7 @@ static ssize_t hw_rev_show(struct device *device, struct device_attribute *attr,
 			   char *buf)
 {
 	struct qib_ibdev *dev =
-		container_of(device, struct qib_ibdev, rdi.ibdev.dev);
+		rdma_device_to_drv_device(device, struct qib_ibdev, rdi.ibdev);
 
 	return sprintf(buf, "%x\n", dd_from_dev(dev)->minrev);
 }
@@ -565,7 +565,7 @@ static ssize_t hca_type_show(struct device *device,
 			     struct device_attribute *attr, char *buf)
 {
 	struct qib_ibdev *dev =
-		container_of(device, struct qib_ibdev, rdi.ibdev.dev);
+		rdma_device_to_drv_device(device, struct qib_ibdev, rdi.ibdev);
 	struct qib_devdata *dd = dd_from_dev(dev);
 	int ret;
 
@@ -590,7 +590,7 @@ static ssize_t boardversion_show(struct device *device,
 				 struct device_attribute *attr, char *buf)
 {
 	struct qib_ibdev *dev =
-		container_of(device, struct qib_ibdev, rdi.ibdev.dev);
+		rdma_device_to_drv_device(device, struct qib_ibdev, rdi.ibdev);
 	struct qib_devdata *dd = dd_from_dev(dev);
 
 	/* The string printed here is already newline-terminated. */
@@ -602,7 +602,7 @@ static ssize_t localbus_info_show(struct device *device,
 				  struct device_attribute *attr, char *buf)
 {
 	struct qib_ibdev *dev =
-		container_of(device, struct qib_ibdev, rdi.ibdev.dev);
+		rdma_device_to_drv_device(device, struct qib_ibdev, rdi.ibdev);
 	struct qib_devdata *dd = dd_from_dev(dev);
 
 	/* The string printed here is already newline-terminated. */
@@ -614,7 +614,7 @@ static ssize_t nctxts_show(struct device *device,
 			   struct device_attribute *attr, char *buf)
 {
 	struct qib_ibdev *dev =
-		container_of(device, struct qib_ibdev, rdi.ibdev.dev);
+		rdma_device_to_drv_device(device, struct qib_ibdev, rdi.ibdev);
 	struct qib_devdata *dd = dd_from_dev(dev);
 
 	/* Return the number of user ports (contexts) available. */
@@ -630,7 +630,7 @@ static ssize_t nfreectxts_show(struct device *device,
 			       struct device_attribute *attr, char *buf)
 {
 	struct qib_ibdev *dev =
-		container_of(device, struct qib_ibdev, rdi.ibdev.dev);
+		rdma_device_to_drv_device(device, struct qib_ibdev, rdi.ibdev);
 	struct qib_devdata *dd = dd_from_dev(dev);
 
 	/* Return the number of free user ports (contexts) available. */
@@ -642,7 +642,7 @@ static ssize_t serial_show(struct device *device,
 			   struct device_attribute *attr, char *buf)
 {
 	struct qib_ibdev *dev =
-		container_of(device, struct qib_ibdev, rdi.ibdev.dev);
+		rdma_device_to_drv_device(device, struct qib_ibdev, rdi.ibdev);
 	struct qib_devdata *dd = dd_from_dev(dev);
 
 	buf[sizeof(dd->serial)] = '\0';
@@ -657,7 +657,7 @@ static ssize_t chip_reset_store(struct device *device,
 				size_t count)
 {
 	struct qib_ibdev *dev =
-		container_of(device, struct qib_ibdev, rdi.ibdev.dev);
+		rdma_device_to_drv_device(device, struct qib_ibdev, rdi.ibdev);
 	struct qib_devdata *dd = dd_from_dev(dev);
 	int ret;
 
@@ -679,7 +679,7 @@ static ssize_t tempsense_show(struct device *device,
 			      struct device_attribute *attr, char *buf)
 {
 	struct qib_ibdev *dev =
-		container_of(device, struct qib_ibdev, rdi.ibdev.dev);
+		rdma_device_to_drv_device(device, struct qib_ibdev, rdi.ibdev);
 	struct qib_devdata *dd = dd_from_dev(dev);
 	int ret;
 	int idx;
diff --git a/drivers/infiniband/hw/usnic/usnic_ib_sysfs.c b/drivers/infiniband/hw/usnic/usnic_ib_sysfs.c
index a7e4b2ccfaf8..c85d48ae7442 100644
--- a/drivers/infiniband/hw/usnic/usnic_ib_sysfs.c
+++ b/drivers/infiniband/hw/usnic/usnic_ib_sysfs.c
@@ -50,7 +50,7 @@ static ssize_t board_id_show(struct device *device,
 			     struct device_attribute *attr, char *buf)
 {
 	struct usnic_ib_dev *us_ibdev =
-		container_of(device, struct usnic_ib_dev, ib_dev.dev);
+		rdma_device_to_drv_device(device, struct usnic_ib_dev, ib_dev);
 	unsigned short subsystem_device_id;
 
 	mutex_lock(&us_ibdev->usdev_lock);
@@ -67,14 +67,13 @@ static DEVICE_ATTR_RO(board_id);
 static ssize_t
 config_show(struct device *device, struct device_attribute *attr, char *buf)
 {
-	struct usnic_ib_dev *us_ibdev;
+	struct usnic_ib_dev *us_ibdev =
+		rdma_device_to_drv_device(device, struct usnic_ib_dev, ib_dev);
 	char *ptr;
 	unsigned left;
 	unsigned n;
 	enum usnic_vnic_res_type res_type;
 
-	us_ibdev = container_of(device, struct usnic_ib_dev, ib_dev.dev);
-
 	/* Buffer space limit is 1 page */
 	ptr = buf;
 	left = PAGE_SIZE;
@@ -130,9 +129,8 @@ static DEVICE_ATTR_RO(config);
 static ssize_t
 iface_show(struct device *device, struct device_attribute *attr, char *buf)
 {
-	struct usnic_ib_dev *us_ibdev;
-
-	us_ibdev = container_of(device, struct usnic_ib_dev, ib_dev.dev);
+	struct usnic_ib_dev *us_ibdev =
+		rdma_device_to_drv_device(device, struct usnic_ib_dev, ib_dev);
 
 	return scnprintf(buf, PAGE_SIZE, "%s\n",
 			netdev_name(us_ibdev->netdev));
@@ -142,9 +140,8 @@ static DEVICE_ATTR_RO(iface);
 static ssize_t
 max_vf_show(struct device *device, struct device_attribute *attr, char *buf)
 {
-	struct usnic_ib_dev *us_ibdev;
-
-	us_ibdev = container_of(device, struct usnic_ib_dev, ib_dev.dev);
+	struct usnic_ib_dev *us_ibdev =
+		rdma_device_to_drv_device(device, struct usnic_ib_dev, ib_dev);
 
 	return scnprintf(buf, PAGE_SIZE, "%u\n",
 			kref_read(&us_ibdev->vf_cnt));
@@ -154,10 +151,10 @@ static DEVICE_ATTR_RO(max_vf);
 static ssize_t
 qp_per_vf_show(struct device *device, struct device_attribute *attr, char *buf)
 {
-	struct usnic_ib_dev *us_ibdev;
+	struct usnic_ib_dev *us_ibdev =
+		rdma_device_to_drv_device(device, struct usnic_ib_dev, ib_dev);
 	int qp_per_vf;
 
-	us_ibdev = container_of(device, struct usnic_ib_dev, ib_dev.dev);
 	qp_per_vf = max(us_ibdev->vf_res_cnt[USNIC_VNIC_RES_TYPE_WQ],
 			us_ibdev->vf_res_cnt[USNIC_VNIC_RES_TYPE_RQ]);
 
@@ -169,9 +166,8 @@ static DEVICE_ATTR_RO(qp_per_vf);
 static ssize_t
 cq_per_vf_show(struct device *device, struct device_attribute *attr, char *buf)
 {
-	struct usnic_ib_dev *us_ibdev;
-
-	us_ibdev = container_of(device, struct usnic_ib_dev, ib_dev.dev);
+	struct usnic_ib_dev *us_ibdev =
+		rdma_device_to_drv_device(device, struct usnic_ib_dev, ib_dev);
 
 	return scnprintf(buf, PAGE_SIZE, "%d\n",
 			us_ibdev->vf_res_cnt[USNIC_VNIC_RES_TYPE_CQ]);
diff --git a/drivers/infiniband/sw/rxe/rxe_verbs.c b/drivers/infiniband/sw/rxe/rxe_verbs.c
index 43171148e9c5..3d01247a28db 100644
--- a/drivers/infiniband/sw/rxe/rxe_verbs.c
+++ b/drivers/infiniband/sw/rxe/rxe_verbs.c
@@ -1129,8 +1129,8 @@ static int rxe_detach_mcast(struct ib_qp *ibqp, union ib_gid *mgid, u16 mlid)
 static ssize_t parent_show(struct device *device,
 			   struct device_attribute *attr, char *buf)
 {
-	struct rxe_dev *rxe = container_of(device, struct rxe_dev,
-					   ib_dev.dev);
+	struct rxe_dev *rxe =
+		rdma_device_to_drv_device(device, struct rxe_dev, ib_dev);
 
 	return snprintf(buf, 16, "%s\n", rxe_parent_name(rxe, 1));
 }
diff --git a/include/rdma/ib_verbs.h b/include/rdma/ib_verbs.h
index 1d1902fd9f87..94b6e1dd4dab 100644
--- a/include/rdma/ib_verbs.h
+++ b/include/rdma/ib_verbs.h
@@ -4241,4 +4241,27 @@ rdma_set_device_sysfs_group(struct ib_device *dev,
 	dev->groups[1] = group;
 }
 
+/**
+ * rdma_device_to_ibdev - Get ib_device pointer from device pointer
+ *
+ * @device:	device pointer for which ib_device pointer to retrieve
+ *
+ * rdma_device_to_ibdev() retrieves ib_device pointer from device.
+ *
+ */
+static inline struct ib_device *rdma_device_to_ibdev(struct device *device)
+{
+	return container_of(device, struct ib_device, dev);
+}
+
+/**
+ * rdma_device_to_drv_device - Helper macro to reach back to driver's
+ *			       ib_device holder structure from device pointer.
+ *
+ * NOTE: New drivers should not make use of this API; This API is only for
+ * existing drivers who have exposed sysfs entries using
+ * rdma_set_device_sysfs_group().
+ */
+#define rdma_device_to_drv_device(dev, drv_dev_struct, ibdev_member)           \
+	container_of(rdma_device_to_ibdev(dev), drv_dev_struct, ibdev_member)
 #endif /* IB_VERBS_H */
-- 
cgit v1.2.3


From 8e8aa1454205839030c9406febd1f6d4ece0d573 Mon Sep 17 00:00:00 2001
From: "Gustavo A. R. Silva" <gustavo@embeddedor.com>
Date: Tue, 15 Jan 2019 00:00:48 -0600
Subject: RDMA/mlx5: Replace kzalloc with kcalloc

Replace kzalloc() function with its 2-factor argument form, kcalloc().

This patch replaces cases of:

	kzalloc(a * b, gfp)

with:
	kcalloc(a, b, gfp)

This code was detected with the help of Coccinelle.

Signed-off-by: Gustavo A. R. Silva <gustavo@embeddedor.com>
Acked-by: Leon Romanovsky <leonro@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 drivers/infiniband/hw/mlx5/main.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

(limited to 'drivers/infiniband/hw/mlx5/main.c')

diff --git a/drivers/infiniband/hw/mlx5/main.c b/drivers/infiniband/hw/mlx5/main.c
index 4b1b56d54301..61064b7171fc 100644
--- a/drivers/infiniband/hw/mlx5/main.c
+++ b/drivers/infiniband/hw/mlx5/main.c
@@ -3823,7 +3823,7 @@ mlx5_ib_raw_fs_rule_add(struct mlx5_ib_dev *dev,
 	if (fs_matcher->priority > MLX5_IB_FLOW_LAST_PRIO)
 		return ERR_PTR(-ENOMEM);
 
-	dst = kzalloc(sizeof(*dst) * 2, GFP_KERNEL);
+	dst = kcalloc(2, sizeof(*dst), GFP_KERNEL);
 	if (!dst)
 		return ERR_PTR(-ENOMEM);
 
-- 
cgit v1.2.3


From 534fd7aac56a7994d16032f32123def9923e339f Mon Sep 17 00:00:00 2001
From: Yishai Hadas <yishaih@mellanox.com>
Date: Sun, 13 Jan 2019 16:01:17 +0200
Subject: IB/mlx5: Manage indirection mkey upon DEVX flow for ODP

Manage indirection mkey upon DEVX flow to support ODP.

To support a page fault event on the indirection mkey it needs to be part
of the device mkey radix tree.

Both the creation and the deletion flows for a DEVX object which is
indirection mkey were adapted to handle that.

Signed-off-by: Yishai Hadas <yishaih@mellanox.com>
Reviewed-by: Artemy Kovalyov <artemyko@mellanox.com>
Signed-off-by: Leon Romanovsky <leonro@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 drivers/infiniband/hw/mlx5/devx.c    | 89 +++++++++++++++++++++++++++++++++++-
 drivers/infiniband/hw/mlx5/main.c    |  1 +
 drivers/infiniband/hw/mlx5/mlx5_ib.h |  6 +++
 include/linux/mlx5/driver.h          |  1 +
 4 files changed, 96 insertions(+), 1 deletion(-)

(limited to 'drivers/infiniband/hw/mlx5/main.c')

diff --git a/drivers/infiniband/hw/mlx5/devx.c b/drivers/infiniband/hw/mlx5/devx.c
index b7ff2138ac2a..bbf9a26d8fa6 100644
--- a/drivers/infiniband/hw/mlx5/devx.c
+++ b/drivers/infiniband/hw/mlx5/devx.c
@@ -17,12 +17,18 @@
 #define UVERBS_MODULE_NAME mlx5_ib
 #include <rdma/uverbs_named_ioctl.h>
 
+enum devx_obj_flags {
+	DEVX_OBJ_FLAGS_INDIRECT_MKEY = 1 << 0,
+};
+
 #define MLX5_MAX_DESTROY_INBOX_SIZE_DW MLX5_ST_SZ_DW(delete_fte_in)
 struct devx_obj {
 	struct mlx5_core_dev	*mdev;
 	u64			obj_id;
 	u32			dinlen; /* destroy inbox length */
 	u32			dinbox[MLX5_MAX_DESTROY_INBOX_SIZE_DW];
+	u32			flags;
+	struct mlx5_ib_devx_mr	devx_mr;
 };
 
 struct devx_umem {
@@ -1011,6 +1017,36 @@ static void devx_obj_build_destroy_cmd(void *in, void *out, void *din,
 	}
 }
 
+static int devx_handle_mkey_indirect(struct devx_obj *obj,
+				     struct mlx5_ib_dev *dev,
+				     void *in, void *out)
+{
+	struct mlx5_mkey_table *table = &dev->mdev->priv.mkey_table;
+	struct mlx5_ib_devx_mr *devx_mr = &obj->devx_mr;
+	unsigned long flags;
+	struct mlx5_core_mkey *mkey;
+	void *mkc;
+	u8 key;
+	int err;
+
+	mkey = &devx_mr->mmkey;
+	mkc = MLX5_ADDR_OF(create_mkey_in, in, memory_key_mkey_entry);
+	key = MLX5_GET(mkc, mkc, mkey_7_0);
+	mkey->key = mlx5_idx_to_mkey(
+			MLX5_GET(create_mkey_out, out, mkey_index)) | key;
+	mkey->type = MLX5_MKEY_INDIRECT_DEVX;
+	mkey->iova = MLX5_GET64(mkc, mkc, start_addr);
+	mkey->size = MLX5_GET64(mkc, mkc, len);
+	mkey->pd = MLX5_GET(mkc, mkc, pd);
+	devx_mr->ndescs = MLX5_GET(mkc, mkc, translations_octword_size);
+
+	write_lock_irqsave(&table->lock, flags);
+	err = radix_tree_insert(&table->tree, mlx5_base_mkey(mkey->key),
+				mkey);
+	write_unlock_irqrestore(&table->lock, flags);
+	return err;
+}
+
 static int devx_handle_mkey_create(struct mlx5_ib_dev *dev,
 				   struct devx_obj *obj,
 				   void *in, int in_len)
@@ -1030,13 +1066,45 @@ static int devx_handle_mkey_create(struct mlx5_ib_dev *dev,
 	access_mode |= MLX5_GET(mkc, mkc, access_mode_4_2) << 2;
 
 	if (access_mode == MLX5_MKC_ACCESS_MODE_KLMS ||
-		access_mode == MLX5_MKC_ACCESS_MODE_KSM)
+		access_mode == MLX5_MKC_ACCESS_MODE_KSM) {
+		if (IS_ENABLED(CONFIG_INFINIBAND_ON_DEMAND_PAGING))
+			obj->flags |= DEVX_OBJ_FLAGS_INDIRECT_MKEY;
 		return 0;
+	}
 
 	MLX5_SET(create_mkey_in, in, mkey_umem_valid, 1);
 	return 0;
 }
 
+static void devx_free_indirect_mkey(struct rcu_head *rcu)
+{
+	kfree(container_of(rcu, struct devx_obj, devx_mr.rcu));
+}
+
+/* This function to delete from the radix tree needs to be called before
+ * destroying the underlying mkey. Otherwise a race might occur in case that
+ * other thread will get the same mkey before this one will be deleted,
+ * in that case it will fail via inserting to the tree its own data.
+ *
+ * Note:
+ * An error in the destroy is not expected unless there is some other indirect
+ * mkey which points to this one. In a kernel cleanup flow it will be just
+ * destroyed in the iterative destruction call. In a user flow, in case
+ * the application didn't close in the expected order it's its own problem,
+ * the mkey won't be part of the tree, in both cases the kernel is safe.
+ */
+static void devx_cleanup_mkey(struct devx_obj *obj)
+{
+	struct mlx5_mkey_table *table = &obj->mdev->priv.mkey_table;
+	struct mlx5_core_mkey *del_mkey;
+	unsigned long flags;
+
+	write_lock_irqsave(&table->lock, flags);
+	del_mkey = radix_tree_delete(&table->tree,
+				     mlx5_base_mkey(obj->devx_mr.mmkey.key));
+	write_unlock_irqrestore(&table->lock, flags);
+}
+
 static int devx_obj_cleanup(struct ib_uobject *uobject,
 			    enum rdma_remove_reason why)
 {
@@ -1044,10 +1112,21 @@ static int devx_obj_cleanup(struct ib_uobject *uobject,
 	struct devx_obj *obj = uobject->object;
 	int ret;
 
+	if (obj->flags & DEVX_OBJ_FLAGS_INDIRECT_MKEY)
+		devx_cleanup_mkey(obj);
+
 	ret = mlx5_cmd_exec(obj->mdev, obj->dinbox, obj->dinlen, out, sizeof(out));
 	if (ib_is_destroy_retryable(ret, why, uobject))
 		return ret;
 
+	if (obj->flags & DEVX_OBJ_FLAGS_INDIRECT_MKEY) {
+		struct mlx5_ib_dev *dev = to_mdev(uobject->context->device);
+
+		call_srcu(&dev->mr_srcu, &obj->devx_mr.rcu,
+			  devx_free_indirect_mkey);
+		return ret;
+	}
+
 	kfree(obj);
 	return ret;
 }
@@ -1108,6 +1187,12 @@ static int UVERBS_HANDLER(MLX5_IB_METHOD_DEVX_OBJ_CREATE)(
 				   &obj_id);
 	WARN_ON(obj->dinlen > MLX5_MAX_DESTROY_INBOX_SIZE_DW * sizeof(u32));
 
+	if (obj->flags & DEVX_OBJ_FLAGS_INDIRECT_MKEY) {
+		err = devx_handle_mkey_indirect(obj, dev, cmd_in, cmd_out);
+		if (err)
+			goto obj_destroy;
+	}
+
 	err = uverbs_copy_to(attrs, MLX5_IB_ATTR_DEVX_OBJ_CREATE_CMD_OUT, cmd_out, cmd_out_len);
 	if (err)
 		goto obj_destroy;
@@ -1116,6 +1201,8 @@ static int UVERBS_HANDLER(MLX5_IB_METHOD_DEVX_OBJ_CREATE)(
 	return 0;
 
 obj_destroy:
+	if (obj->flags & DEVX_OBJ_FLAGS_INDIRECT_MKEY)
+		devx_cleanup_mkey(obj);
 	mlx5_cmd_exec(obj->mdev, obj->dinbox, obj->dinlen, out, sizeof(out));
 obj_free:
 	kfree(obj);
diff --git a/drivers/infiniband/hw/mlx5/main.c b/drivers/infiniband/hw/mlx5/main.c
index 61064b7171fc..ae00f994673b 100644
--- a/drivers/infiniband/hw/mlx5/main.c
+++ b/drivers/infiniband/hw/mlx5/main.c
@@ -5724,6 +5724,7 @@ void mlx5_ib_stage_init_cleanup(struct mlx5_ib_dev *dev)
 {
 	mlx5_ib_cleanup_multiport_master(dev);
 	if (IS_ENABLED(CONFIG_INFINIBAND_ON_DEMAND_PAGING)) {
+		srcu_barrier(&dev->mr_srcu);
 		cleanup_srcu_struct(&dev->mr_srcu);
 		drain_workqueue(dev->advise_mr_wq);
 		destroy_workqueue(dev->advise_mr_wq);
diff --git a/drivers/infiniband/hw/mlx5/mlx5_ib.h b/drivers/infiniband/hw/mlx5/mlx5_ib.h
index b0a37ca2a714..819207190343 100644
--- a/drivers/infiniband/hw/mlx5/mlx5_ib.h
+++ b/drivers/infiniband/hw/mlx5/mlx5_ib.h
@@ -602,6 +602,12 @@ struct mlx5_ib_mw {
 	int			ndescs;
 };
 
+struct mlx5_ib_devx_mr {
+	struct mlx5_core_mkey	mmkey;
+	int			ndescs;
+	struct rcu_head		rcu;
+};
+
 struct mlx5_ib_umr_context {
 	struct ib_cqe		cqe;
 	enum ib_wc_status	status;
diff --git a/include/linux/mlx5/driver.h b/include/linux/mlx5/driver.h
index b6f5839f129a..619d6fee96a1 100644
--- a/include/linux/mlx5/driver.h
+++ b/include/linux/mlx5/driver.h
@@ -364,6 +364,7 @@ struct mlx5_core_sig_ctx {
 enum {
 	MLX5_MKEY_MR = 1,
 	MLX5_MKEY_MW,
+	MLX5_MKEY_INDIRECT_DEVX,
 };
 
 struct mlx5_core_mkey {
-- 
cgit v1.2.3


From 6113cc44015b41ca51c0a76fed82522b68108dac Mon Sep 17 00:00:00 2001
From: Maor Gottlieb <maorg@mellanox.com>
Date: Thu, 17 Jan 2019 20:08:15 +0200
Subject: IB/mlx5: Don't override existing ip_protocol

Two flow specifications can set the ip protocol field in
the flow table entry:

1) IB_FLOW_SPEC_TCP/UDP/GRE - set the ip protocol accordingly.
2) IB_FLOW_SPEC_IPV4/6 - has ip_protocol field for users
who want to receive specific L4 packets.

We need to avoid overriding of the ip_protocol with zeros,
in case that the user first put the L4 specification and
only then the L3.

Fixes: ca0d47538528b ('IB/mlx5: Add support in TOS and protocol to flow steering')
Signed-off-by: Maor Gottlieb <maorg@mellanox.com>
Signed-off-by: Leon Romanovsky <leonro@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 drivers/infiniband/hw/mlx5/main.c | 53 +++++++++++++++++++++++++++------------
 1 file changed, 37 insertions(+), 16 deletions(-)

(limited to 'drivers/infiniband/hw/mlx5/main.c')

diff --git a/drivers/infiniband/hw/mlx5/main.c b/drivers/infiniband/hw/mlx5/main.c
index ae00f994673b..b32b5c7f1cfd 100644
--- a/drivers/infiniband/hw/mlx5/main.c
+++ b/drivers/infiniband/hw/mlx5/main.c
@@ -2333,10 +2333,29 @@ static u8 get_match_criteria_enable(u32 *match_criteria)
 	return match_criteria_enable;
 }
 
-static void set_proto(void *outer_c, void *outer_v, u8 mask, u8 val)
+static int set_proto(void *outer_c, void *outer_v, u8 mask, u8 val)
 {
-	MLX5_SET(fte_match_set_lyr_2_4, outer_c, ip_protocol, mask);
-	MLX5_SET(fte_match_set_lyr_2_4, outer_v, ip_protocol, val);
+	u8 entry_mask;
+	u8 entry_val;
+	int err = 0;
+
+	if (!mask)
+		goto out;
+
+	entry_mask = MLX5_GET(fte_match_set_lyr_2_4, outer_c,
+			      ip_protocol);
+	entry_val = MLX5_GET(fte_match_set_lyr_2_4, outer_v,
+			     ip_protocol);
+	if (!entry_mask) {
+		MLX5_SET(fte_match_set_lyr_2_4, outer_c, ip_protocol, mask);
+		MLX5_SET(fte_match_set_lyr_2_4, outer_v, ip_protocol, val);
+		goto out;
+	}
+	/* Don't override existing ip protocol */
+	if (mask != entry_mask || val != entry_val)
+		err = -EINVAL;
+out:
+	return err;
 }
 
 static void set_flow_label(void *misc_c, void *misc_v, u32 mask, u32 val,
@@ -2570,8 +2589,10 @@ static int parse_flow_attr(struct mlx5_core_dev *mdev, u32 *match_c,
 		set_tos(headers_c, headers_v,
 			ib_spec->ipv4.mask.tos, ib_spec->ipv4.val.tos);
 
-		set_proto(headers_c, headers_v,
-			  ib_spec->ipv4.mask.proto, ib_spec->ipv4.val.proto);
+		if (set_proto(headers_c, headers_v,
+			      ib_spec->ipv4.mask.proto,
+			      ib_spec->ipv4.val.proto))
+			return -EINVAL;
 		break;
 	case IB_FLOW_SPEC_IPV6:
 		if (FIELDS_NOT_SUPPORTED(ib_spec->ipv6.mask, LAST_IPV6_FIELD))
@@ -2610,9 +2631,10 @@ static int parse_flow_attr(struct mlx5_core_dev *mdev, u32 *match_c,
 			ib_spec->ipv6.mask.traffic_class,
 			ib_spec->ipv6.val.traffic_class);
 
-		set_proto(headers_c, headers_v,
-			  ib_spec->ipv6.mask.next_hdr,
-			  ib_spec->ipv6.val.next_hdr);
+		if (set_proto(headers_c, headers_v,
+			      ib_spec->ipv6.mask.next_hdr,
+			      ib_spec->ipv6.val.next_hdr))
+			return -EINVAL;
 
 		set_flow_label(misc_params_c, misc_params_v,
 			       ntohl(ib_spec->ipv6.mask.flow_label),
@@ -2633,10 +2655,8 @@ static int parse_flow_attr(struct mlx5_core_dev *mdev, u32 *match_c,
 					 LAST_TCP_UDP_FIELD))
 			return -EOPNOTSUPP;
 
-		MLX5_SET(fte_match_set_lyr_2_4, headers_c, ip_protocol,
-			 0xff);
-		MLX5_SET(fte_match_set_lyr_2_4, headers_v, ip_protocol,
-			 IPPROTO_TCP);
+		if (set_proto(headers_c, headers_v, 0xff, IPPROTO_TCP))
+			return -EINVAL;
 
 		MLX5_SET(fte_match_set_lyr_2_4, headers_c, tcp_sport,
 			 ntohs(ib_spec->tcp_udp.mask.src_port));
@@ -2653,10 +2673,8 @@ static int parse_flow_attr(struct mlx5_core_dev *mdev, u32 *match_c,
 					 LAST_TCP_UDP_FIELD))
 			return -EOPNOTSUPP;
 
-		MLX5_SET(fte_match_set_lyr_2_4, headers_c, ip_protocol,
-			 0xff);
-		MLX5_SET(fte_match_set_lyr_2_4, headers_v, ip_protocol,
-			 IPPROTO_UDP);
+		if (set_proto(headers_c, headers_v, 0xff, IPPROTO_UDP))
+			return -EINVAL;
 
 		MLX5_SET(fte_match_set_lyr_2_4, headers_c, udp_sport,
 			 ntohs(ib_spec->tcp_udp.mask.src_port));
@@ -2672,6 +2690,9 @@ static int parse_flow_attr(struct mlx5_core_dev *mdev, u32 *match_c,
 		if (ib_spec->gre.mask.c_ks_res0_ver)
 			return -EOPNOTSUPP;
 
+		if (set_proto(headers_c, headers_v, 0xff, IPPROTO_GRE))
+			return -EINVAL;
+
 		MLX5_SET(fte_match_set_lyr_2_4, headers_c, ip_protocol,
 			 0xff);
 		MLX5_SET(fte_match_set_lyr_2_4, headers_v, ip_protocol,
-- 
cgit v1.2.3


From 73eb8f03f0ecfcb45ed1995990229f5a81e8e38f Mon Sep 17 00:00:00 2001
From: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Date: Tue, 22 Jan 2019 16:17:57 +0100
Subject: infiniband: mlx5: no need to check return value of debugfs_create
 functions

When calling debugfs functions, there is no need to ever check the
return value.  The function can work or not, but the code logic should
never do something different based on this.

Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Acked-by: Leon Romanovsky <leonro@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 drivers/infiniband/hw/mlx5/cong.c    | 15 ++++-------
 drivers/infiniband/hw/mlx5/main.c    |  9 +++----
 drivers/infiniband/hw/mlx5/mlx5_ib.h |  8 +-----
 drivers/infiniband/hw/mlx5/mr.c      | 51 +++++++-----------------------------
 4 files changed, 19 insertions(+), 64 deletions(-)

(limited to 'drivers/infiniband/hw/mlx5/main.c')

diff --git a/drivers/infiniband/hw/mlx5/cong.c b/drivers/infiniband/hw/mlx5/cong.c
index 7e4e358a4fd8..8ba439fabf7f 100644
--- a/drivers/infiniband/hw/mlx5/cong.c
+++ b/drivers/infiniband/hw/mlx5/cong.c
@@ -389,19 +389,19 @@ void mlx5_ib_cleanup_cong_debugfs(struct mlx5_ib_dev *dev, u8 port_num)
 	dev->port[port_num].dbg_cc_params = NULL;
 }
 
-int mlx5_ib_init_cong_debugfs(struct mlx5_ib_dev *dev, u8 port_num)
+void mlx5_ib_init_cong_debugfs(struct mlx5_ib_dev *dev, u8 port_num)
 {
 	struct mlx5_ib_dbg_cc_params *dbg_cc_params;
 	struct mlx5_core_dev *mdev;
 	int i;
 
 	if (!mlx5_debugfs_root)
-		goto out;
+		return;
 
 	/* Takes a 1-based port number */
 	mdev = mlx5_ib_get_native_port_mdev(dev, port_num + 1, NULL);
 	if (!mdev)
-		goto out;
+		return;
 
 	if (!MLX5_CAP_GEN(mdev, cc_query_allowed) ||
 	    !MLX5_CAP_GEN(mdev, cc_modify_allowed))
@@ -415,8 +415,6 @@ int mlx5_ib_init_cong_debugfs(struct mlx5_ib_dev *dev, u8 port_num)
 
 	dbg_cc_params->root = debugfs_create_dir("cc_params",
 						 mdev->priv.dbg_root);
-	if (!dbg_cc_params->root)
-		goto err;
 
 	for (i = 0; i < MLX5_IB_DBG_CC_MAX; i++) {
 		dbg_cc_params->params[i].offset = i;
@@ -427,14 +425,11 @@ int mlx5_ib_init_cong_debugfs(struct mlx5_ib_dev *dev, u8 port_num)
 					    0600, dbg_cc_params->root,
 					    &dbg_cc_params->params[i],
 					    &dbg_cc_fops);
-		if (!dbg_cc_params->params[i].dentry)
-			goto err;
 	}
 
 put_mdev:
 	mlx5_ib_put_native_port_mdev(dev, port_num + 1);
-out:
-	return 0;
+	return;
 
 err:
 	mlx5_ib_warn(dev, "cong debugfs failure\n");
@@ -445,5 +440,5 @@ err:
 	 * We don't want to fail driver if debugfs failed to initialize,
 	 * so we are not forwarding error to the user.
 	 */
-	return 0;
+	return;
 }
diff --git a/drivers/infiniband/hw/mlx5/main.c b/drivers/infiniband/hw/mlx5/main.c
index b32b5c7f1cfd..2aaa196c2653 100644
--- a/drivers/infiniband/hw/mlx5/main.c
+++ b/drivers/infiniband/hw/mlx5/main.c
@@ -5530,9 +5530,7 @@ static bool mlx5_ib_bind_slave_port(struct mlx5_ib_dev *ibdev,
 	mpi->mdev_events.notifier_call = mlx5_ib_event_slave_port;
 	mlx5_notifier_register(mpi->mdev, &mpi->mdev_events);
 
-	err = mlx5_ib_init_cong_debugfs(ibdev, port_num);
-	if (err)
-		goto unbind;
+	mlx5_ib_init_cong_debugfs(ibdev, port_num);
 
 	return true;
 
@@ -6207,8 +6205,9 @@ void mlx5_ib_stage_counters_cleanup(struct mlx5_ib_dev *dev)
 
 static int mlx5_ib_stage_cong_debugfs_init(struct mlx5_ib_dev *dev)
 {
-	return mlx5_ib_init_cong_debugfs(dev,
-					 mlx5_core_native_port_num(dev->mdev) - 1);
+	mlx5_ib_init_cong_debugfs(dev,
+				  mlx5_core_native_port_num(dev->mdev) - 1);
+	return 0;
 }
 
 static void mlx5_ib_stage_cong_debugfs_cleanup(struct mlx5_ib_dev *dev)
diff --git a/drivers/infiniband/hw/mlx5/mlx5_ib.h b/drivers/infiniband/hw/mlx5/mlx5_ib.h
index 819207190343..ea124af56dce 100644
--- a/drivers/infiniband/hw/mlx5/mlx5_ib.h
+++ b/drivers/infiniband/hw/mlx5/mlx5_ib.h
@@ -636,7 +636,6 @@ struct mlx5_cache_ent {
 	spinlock_t		lock;
 
 
-	struct dentry	       *dir;
 	char                    name[4];
 	u32                     order;
 	u32			xlt;
@@ -648,11 +647,6 @@ struct mlx5_cache_ent {
 	u32                     miss;
 	u32			limit;
 
-	struct dentry          *fsize;
-	struct dentry          *fcur;
-	struct dentry          *fmiss;
-	struct dentry          *flimit;
-
 	struct mlx5_ib_dev     *dev;
 	struct work_struct	work;
 	struct delayed_work	dwork;
@@ -1270,7 +1264,7 @@ __be16 mlx5_get_roce_udp_sport(struct mlx5_ib_dev *dev,
 			       const struct ib_gid_attr *attr);
 
 void mlx5_ib_cleanup_cong_debugfs(struct mlx5_ib_dev *dev, u8 port_num);
-int mlx5_ib_init_cong_debugfs(struct mlx5_ib_dev *dev, u8 port_num);
+void mlx5_ib_init_cong_debugfs(struct mlx5_ib_dev *dev, u8 port_num);
 
 /* GSI QP helper functions */
 struct ib_qp *mlx5_ib_gsi_create_qp(struct ib_pd *pd,
diff --git a/drivers/infiniband/hw/mlx5/mr.c b/drivers/infiniband/hw/mlx5/mr.c
index 6d763d6a189f..d78a94803e8c 100644
--- a/drivers/infiniband/hw/mlx5/mr.c
+++ b/drivers/infiniband/hw/mlx5/mr.c
@@ -606,52 +606,27 @@ static void mlx5_mr_cache_debugfs_cleanup(struct mlx5_ib_dev *dev)
 	dev->cache.root = NULL;
 }
 
-static int mlx5_mr_cache_debugfs_init(struct mlx5_ib_dev *dev)
+static void mlx5_mr_cache_debugfs_init(struct mlx5_ib_dev *dev)
 {
 	struct mlx5_mr_cache *cache = &dev->cache;
 	struct mlx5_cache_ent *ent;
+	struct dentry *dir;
 	int i;
 
 	if (!mlx5_debugfs_root || dev->rep)
-		return 0;
+		return;
 
 	cache->root = debugfs_create_dir("mr_cache", dev->mdev->priv.dbg_root);
-	if (!cache->root)
-		return -ENOMEM;
 
 	for (i = 0; i < MAX_MR_CACHE_ENTRIES; i++) {
 		ent = &cache->ent[i];
 		sprintf(ent->name, "%d", ent->order);
-		ent->dir = debugfs_create_dir(ent->name,  cache->root);
-		if (!ent->dir)
-			goto err;
-
-		ent->fsize = debugfs_create_file("size", 0600, ent->dir, ent,
-						 &size_fops);
-		if (!ent->fsize)
-			goto err;
-
-		ent->flimit = debugfs_create_file("limit", 0600, ent->dir, ent,
-						  &limit_fops);
-		if (!ent->flimit)
-			goto err;
-
-		ent->fcur = debugfs_create_u32("cur", 0400, ent->dir,
-					       &ent->cur);
-		if (!ent->fcur)
-			goto err;
-
-		ent->fmiss = debugfs_create_u32("miss", 0600, ent->dir,
-						&ent->miss);
-		if (!ent->fmiss)
-			goto err;
+		dir = debugfs_create_dir(ent->name, cache->root);
+		debugfs_create_file("size", 0600, dir, ent, &size_fops);
+		debugfs_create_file("limit", 0600, dir, ent, &limit_fops);
+		debugfs_create_u32("cur", 0400, dir, &ent->cur);
+		debugfs_create_u32("miss", 0600, dir, &ent->miss);
 	}
-
-	return 0;
-err:
-	mlx5_mr_cache_debugfs_cleanup(dev);
-
-	return -ENOMEM;
 }
 
 static void delay_time_func(struct timer_list *t)
@@ -665,7 +640,6 @@ int mlx5_mr_cache_init(struct mlx5_ib_dev *dev)
 {
 	struct mlx5_mr_cache *cache = &dev->cache;
 	struct mlx5_cache_ent *ent;
-	int err;
 	int i;
 
 	mutex_init(&dev->slow_path_mutex);
@@ -709,14 +683,7 @@ int mlx5_mr_cache_init(struct mlx5_ib_dev *dev)
 		queue_work(cache->wq, &ent->work);
 	}
 
-	err = mlx5_mr_cache_debugfs_init(dev);
-	if (err)
-		mlx5_ib_warn(dev, "cache debugfs failure\n");
-
-	/*
-	 * We don't want to fail driver if debugfs failed to initialize,
-	 * so we are not forwarding error to the user.
-	 */
+	mlx5_mr_cache_debugfs_init(dev);
 
 	return 0;
 }
-- 
cgit v1.2.3


From f3ffed0ce49a41cdbac5a85f42baeebe334ca9ab Mon Sep 17 00:00:00 2001
From: Kamal Heib <kamalheib1@gmail.com>
Date: Wed, 30 Jan 2019 16:13:42 +0200
Subject: IB/mlx5: Make mlx5_ib_stage_odp_cleanup() static

The function mlx5_ib_stage_odp_cleanup() is only used in main.c

Fixes: d5d284b829a6 ("{net,IB}/mlx5: Move Page fault EQ and ODP logic to RDMA")
Signed-off-by: Kamal Heib <kamalheib1@gmail.com>
Acked-by: Leon Romanovsky <leonro@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 drivers/infiniband/hw/mlx5/main.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

(limited to 'drivers/infiniband/hw/mlx5/main.c')

diff --git a/drivers/infiniband/hw/mlx5/main.c b/drivers/infiniband/hw/mlx5/main.c
index 2aaa196c2653..96bf1b2f9dd7 100644
--- a/drivers/infiniband/hw/mlx5/main.c
+++ b/drivers/infiniband/hw/mlx5/main.c
@@ -6176,7 +6176,7 @@ static int mlx5_ib_stage_odp_init(struct mlx5_ib_dev *dev)
 	return mlx5_ib_odp_init_one(dev);
 }
 
-void mlx5_ib_stage_odp_cleanup(struct mlx5_ib_dev *dev)
+static void mlx5_ib_stage_odp_cleanup(struct mlx5_ib_dev *dev)
 {
 	mlx5_ib_odp_cleanup_one(dev);
 }
-- 
cgit v1.2.3


From 459cc69fa4c17caf21de596693d8a07170820a58 Mon Sep 17 00:00:00 2001
From: Leon Romanovsky <leonro@mellanox.com>
Date: Wed, 30 Jan 2019 12:49:11 +0200
Subject: RDMA: Provide safe ib_alloc_device() function

All callers to ib_alloc_device() provide a larger size than struct
ib_device and rely on the fact that struct ib_device is embedded in their
driver specific structure as the first member.

Provide a safer variant of ib_alloc_device() that checks and enforces this
approach to make sure the drivers are using it right.

Signed-off-by: Leon Romanovsky <leonro@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 drivers/infiniband/core/device.c               | 6 +++---
 drivers/infiniband/hw/bnxt_re/main.c           | 2 +-
 drivers/infiniband/hw/cxgb3/iwch.c             | 2 +-
 drivers/infiniband/hw/cxgb4/device.c           | 2 +-
 drivers/infiniband/hw/hns/hns_roce_hw_v1.c     | 2 +-
 drivers/infiniband/hw/hns/hns_roce_hw_v2.c     | 2 +-
 drivers/infiniband/hw/i40iw/i40iw_verbs.c      | 2 +-
 drivers/infiniband/hw/mlx4/main.c              | 2 +-
 drivers/infiniband/hw/mlx5/ib_rep.c            | 2 +-
 drivers/infiniband/hw/mlx5/main.c              | 2 +-
 drivers/infiniband/hw/mthca/mthca_main.c       | 2 +-
 drivers/infiniband/hw/nes/nes_verbs.c          | 2 +-
 drivers/infiniband/hw/ocrdma/ocrdma_main.c     | 2 +-
 drivers/infiniband/hw/qedr/main.c              | 2 +-
 drivers/infiniband/hw/usnic/usnic_ib_main.c    | 2 +-
 drivers/infiniband/hw/vmw_pvrdma/pvrdma_main.c | 2 +-
 drivers/infiniband/sw/rdmavt/vt.c              | 2 +-
 drivers/infiniband/sw/rxe/rxe_net.c            | 2 +-
 include/rdma/ib_verbs.h                        | 8 +++++++-
 19 files changed, 27 insertions(+), 21 deletions(-)

(limited to 'drivers/infiniband/hw/mlx5/main.c')

diff --git a/drivers/infiniband/core/device.c b/drivers/infiniband/core/device.c
index 200431c540f2..b511cfa00bdb 100644
--- a/drivers/infiniband/core/device.c
+++ b/drivers/infiniband/core/device.c
@@ -268,7 +268,7 @@ static struct class ib_class = {
 };
 
 /**
- * ib_alloc_device - allocate an IB device struct
+ * _ib_alloc_device - allocate an IB device struct
  * @size:size of structure to allocate
  *
  * Low-level drivers should use ib_alloc_device() to allocate &struct
@@ -277,7 +277,7 @@ static struct class ib_class = {
  * ib_dealloc_device() must be used to free structures allocated with
  * ib_alloc_device().
  */
-struct ib_device *ib_alloc_device(size_t size)
+struct ib_device *_ib_alloc_device(size_t size)
 {
 	struct ib_device *device;
 
@@ -303,7 +303,7 @@ struct ib_device *ib_alloc_device(size_t size)
 
 	return device;
 }
-EXPORT_SYMBOL(ib_alloc_device);
+EXPORT_SYMBOL(_ib_alloc_device);
 
 /**
  * ib_dealloc_device - free an IB device struct
diff --git a/drivers/infiniband/hw/bnxt_re/main.c b/drivers/infiniband/hw/bnxt_re/main.c
index 16eecfa5882c..08777506d0b1 100644
--- a/drivers/infiniband/hw/bnxt_re/main.c
+++ b/drivers/infiniband/hw/bnxt_re/main.c
@@ -688,7 +688,7 @@ static struct bnxt_re_dev *bnxt_re_dev_add(struct net_device *netdev,
 	struct bnxt_re_dev *rdev;
 
 	/* Allocate bnxt_re_dev instance here */
-	rdev = (struct bnxt_re_dev *)ib_alloc_device(sizeof(*rdev));
+	rdev = ib_alloc_device(bnxt_re_dev, ibdev);
 	if (!rdev) {
 		dev_err(NULL, "%s: bnxt_re_dev allocation failure!",
 			ROCE_DRV_MODULE_NAME);
diff --git a/drivers/infiniband/hw/cxgb3/iwch.c b/drivers/infiniband/hw/cxgb3/iwch.c
index 591de319c178..fb03bc492ef7 100644
--- a/drivers/infiniband/hw/cxgb3/iwch.c
+++ b/drivers/infiniband/hw/cxgb3/iwch.c
@@ -146,7 +146,7 @@ static void open_rnic_dev(struct t3cdev *tdev)
 
 	pr_debug("%s t3cdev %p\n", __func__,  tdev);
 	pr_info_once("Chelsio T3 RDMA Driver - version %s\n", DRV_VERSION);
-	rnicp = (struct iwch_dev *)ib_alloc_device(sizeof(*rnicp));
+	rnicp = ib_alloc_device(iwch_dev, ibdev);
 	if (!rnicp) {
 		pr_err("Cannot allocate ib device\n");
 		return;
diff --git a/drivers/infiniband/hw/cxgb4/device.c b/drivers/infiniband/hw/cxgb4/device.c
index 9c10fff6dcfb..4b4e2464b705 100644
--- a/drivers/infiniband/hw/cxgb4/device.c
+++ b/drivers/infiniband/hw/cxgb4/device.c
@@ -966,7 +966,7 @@ static struct c4iw_dev *c4iw_alloc(const struct cxgb4_lld_info *infop)
 		pr_info("%s: On-Chip Queues not supported on this device\n",
 			pci_name(infop->pdev));
 
-	devp = (struct c4iw_dev *)ib_alloc_device(sizeof(*devp));
+	devp = ib_alloc_device(c4iw_dev, ibdev);
 	if (!devp) {
 		pr_err("Cannot allocate ib device\n");
 		return ERR_PTR(-ENOMEM);
diff --git a/drivers/infiniband/hw/hns/hns_roce_hw_v1.c b/drivers/infiniband/hw/hns/hns_roce_hw_v1.c
index b74c742b000c..fa08c22aad66 100644
--- a/drivers/infiniband/hw/hns/hns_roce_hw_v1.c
+++ b/drivers/infiniband/hw/hns/hns_roce_hw_v1.c
@@ -5002,7 +5002,7 @@ static int hns_roce_probe(struct platform_device *pdev)
 	struct hns_roce_dev *hr_dev;
 	struct device *dev = &pdev->dev;
 
-	hr_dev = (struct hns_roce_dev *)ib_alloc_device(sizeof(*hr_dev));
+	hr_dev = ib_alloc_device(hns_roce_dev, ib_dev);
 	if (!hr_dev)
 		return -ENOMEM;
 
diff --git a/drivers/infiniband/hw/hns/hns_roce_hw_v2.c b/drivers/infiniband/hw/hns/hns_roce_hw_v2.c
index 5c483b437bdd..48a5d6548cd4 100644
--- a/drivers/infiniband/hw/hns/hns_roce_hw_v2.c
+++ b/drivers/infiniband/hw/hns/hns_roce_hw_v2.c
@@ -6059,7 +6059,7 @@ static int hns_roce_hw_v2_init_instance(struct hnae3_handle *handle)
 	struct hns_roce_dev *hr_dev;
 	int ret;
 
-	hr_dev = (struct hns_roce_dev *)ib_alloc_device(sizeof(*hr_dev));
+	hr_dev = ib_alloc_device(hns_roce_dev, ib_dev);
 	if (!hr_dev)
 		return -ENOMEM;
 
diff --git a/drivers/infiniband/hw/i40iw/i40iw_verbs.c b/drivers/infiniband/hw/i40iw/i40iw_verbs.c
index 12b31a8440be..d4ab46dd9e6c 100644
--- a/drivers/infiniband/hw/i40iw/i40iw_verbs.c
+++ b/drivers/infiniband/hw/i40iw/i40iw_verbs.c
@@ -2762,7 +2762,7 @@ static struct i40iw_ib_device *i40iw_init_rdma_device(struct i40iw_device *iwdev
 	struct net_device *netdev = iwdev->netdev;
 	struct pci_dev *pcidev = (struct pci_dev *)iwdev->hw.dev_context;
 
-	iwibdev = (struct i40iw_ib_device *)ib_alloc_device(sizeof(*iwibdev));
+	iwibdev = ib_alloc_device(i40iw_ib_device, ibdev);
 	if (!iwibdev) {
 		i40iw_pr_err("iwdev == NULL\n");
 		return NULL;
diff --git a/drivers/infiniband/hw/mlx4/main.c b/drivers/infiniband/hw/mlx4/main.c
index dc2ffd293a11..d66002a31000 100644
--- a/drivers/infiniband/hw/mlx4/main.c
+++ b/drivers/infiniband/hw/mlx4/main.c
@@ -2635,7 +2635,7 @@ static void *mlx4_ib_add(struct mlx4_dev *dev)
 	if (num_ports == 0)
 		return NULL;
 
-	ibdev = (struct mlx4_ib_dev *) ib_alloc_device(sizeof *ibdev);
+	ibdev = ib_alloc_device(mlx4_ib_dev, ib_dev);
 	if (!ibdev) {
 		dev_err(&dev->persist->pdev->dev,
 			"Device struct alloc failed\n");
diff --git a/drivers/infiniband/hw/mlx5/ib_rep.c b/drivers/infiniband/hw/mlx5/ib_rep.c
index 46a9ddc8ca56..6d7b8bad4b61 100644
--- a/drivers/infiniband/hw/mlx5/ib_rep.c
+++ b/drivers/infiniband/hw/mlx5/ib_rep.c
@@ -70,7 +70,7 @@ mlx5_ib_vport_rep_load(struct mlx5_core_dev *dev, struct mlx5_eswitch_rep *rep)
 {
 	struct mlx5_ib_dev *ibdev;
 
-	ibdev = (struct mlx5_ib_dev *)ib_alloc_device(sizeof(*ibdev));
+	ibdev = ib_alloc_device(mlx5_ib_dev, ib_dev);
 	if (!ibdev)
 		return -ENOMEM;
 
diff --git a/drivers/infiniband/hw/mlx5/main.c b/drivers/infiniband/hw/mlx5/main.c
index 96bf1b2f9dd7..8161acda64e6 100644
--- a/drivers/infiniband/hw/mlx5/main.c
+++ b/drivers/infiniband/hw/mlx5/main.c
@@ -6508,7 +6508,7 @@ static void *mlx5_ib_add(struct mlx5_core_dev *mdev)
 	if (mlx5_core_is_mp_slave(mdev) && ll == IB_LINK_LAYER_ETHERNET)
 		return mlx5_ib_add_slave_port(mdev);
 
-	dev = (struct mlx5_ib_dev *)ib_alloc_device(sizeof(*dev));
+	dev = ib_alloc_device(mlx5_ib_dev, ib_dev);
 	if (!dev)
 		return NULL;
 
diff --git a/drivers/infiniband/hw/mthca/mthca_main.c b/drivers/infiniband/hw/mthca/mthca_main.c
index 92c49bff22bc..fe9654a7af71 100644
--- a/drivers/infiniband/hw/mthca/mthca_main.c
+++ b/drivers/infiniband/hw/mthca/mthca_main.c
@@ -961,7 +961,7 @@ static int __mthca_init_one(struct pci_dev *pdev, int hca_type)
 	/* We can handle large RDMA requests, so allow larger segments. */
 	dma_set_max_seg_size(&pdev->dev, 1024 * 1024 * 1024);
 
-	mdev = (struct mthca_dev *) ib_alloc_device(sizeof *mdev);
+	mdev = ib_alloc_device(mthca_dev, ib_dev);
 	if (!mdev) {
 		dev_err(&pdev->dev, "Device struct alloc failed, "
 			"aborting.\n");
diff --git a/drivers/infiniband/hw/nes/nes_verbs.c b/drivers/infiniband/hw/nes/nes_verbs.c
index 034156f7e9ed..6eb991d40035 100644
--- a/drivers/infiniband/hw/nes/nes_verbs.c
+++ b/drivers/infiniband/hw/nes/nes_verbs.c
@@ -3669,7 +3669,7 @@ struct nes_ib_device *nes_init_ofa_device(struct net_device *netdev)
 	struct nes_vnic *nesvnic = netdev_priv(netdev);
 	struct nes_device *nesdev = nesvnic->nesdev;
 
-	nesibdev = (struct nes_ib_device *)ib_alloc_device(sizeof(struct nes_ib_device));
+	nesibdev = ib_alloc_device(nes_ib_device, ibdev);
 	if (nesibdev == NULL) {
 		return NULL;
 	}
diff --git a/drivers/infiniband/hw/ocrdma/ocrdma_main.c b/drivers/infiniband/hw/ocrdma/ocrdma_main.c
index b0491b9ecfe4..88970a6bb555 100644
--- a/drivers/infiniband/hw/ocrdma/ocrdma_main.c
+++ b/drivers/infiniband/hw/ocrdma/ocrdma_main.c
@@ -297,7 +297,7 @@ static struct ocrdma_dev *ocrdma_add(struct be_dev_info *dev_info)
 	u8 lstate = 0;
 	struct ocrdma_dev *dev;
 
-	dev = (struct ocrdma_dev *)ib_alloc_device(sizeof(struct ocrdma_dev));
+	dev = ib_alloc_device(ocrdma_dev, ibdev);
 	if (!dev) {
 		pr_err("Unable to allocate ib device\n");
 		return NULL;
diff --git a/drivers/infiniband/hw/qedr/main.c b/drivers/infiniband/hw/qedr/main.c
index f85e72b65a10..878e9e23652b 100644
--- a/drivers/infiniband/hw/qedr/main.c
+++ b/drivers/infiniband/hw/qedr/main.c
@@ -853,7 +853,7 @@ static struct qedr_dev *qedr_add(struct qed_dev *cdev, struct pci_dev *pdev,
 	struct qedr_dev *dev;
 	int rc = 0;
 
-	dev = (struct qedr_dev *)ib_alloc_device(sizeof(*dev));
+	dev = ib_alloc_device(qedr_dev, ibdev);
 	if (!dev) {
 		pr_err("Unable to allocate ib device\n");
 		return NULL;
diff --git a/drivers/infiniband/hw/usnic/usnic_ib_main.c b/drivers/infiniband/hw/usnic/usnic_ib_main.c
index 3201dd1899c7..0c0a288cb585 100644
--- a/drivers/infiniband/hw/usnic/usnic_ib_main.c
+++ b/drivers/infiniband/hw/usnic/usnic_ib_main.c
@@ -372,7 +372,7 @@ static void *usnic_ib_device_add(struct pci_dev *dev)
 	usnic_dbg("\n");
 	netdev = pci_get_drvdata(dev);
 
-	us_ibdev = (struct usnic_ib_dev *)ib_alloc_device(sizeof(*us_ibdev));
+	us_ibdev = ib_alloc_device(usnic_ib_dev, ib_dev);
 	if (!us_ibdev) {
 		usnic_err("Device %s context alloc failed\n",
 				netdev_name(pci_get_drvdata(dev)));
diff --git a/drivers/infiniband/hw/vmw_pvrdma/pvrdma_main.c b/drivers/infiniband/hw/vmw_pvrdma/pvrdma_main.c
index a5f02276d903..e582beaf9430 100644
--- a/drivers/infiniband/hw/vmw_pvrdma/pvrdma_main.c
+++ b/drivers/infiniband/hw/vmw_pvrdma/pvrdma_main.c
@@ -795,7 +795,7 @@ static int pvrdma_pci_probe(struct pci_dev *pdev,
 	dev_dbg(&pdev->dev, "initializing driver %s\n", pci_name(pdev));
 
 	/* Allocate zero-out device */
-	dev = (struct pvrdma_dev *)ib_alloc_device(sizeof(*dev));
+	dev = ib_alloc_device(pvrdma_dev, ib_dev);
 	if (!dev) {
 		dev_err(&pdev->dev, "failed to allocate IB device\n");
 		return -ENOMEM;
diff --git a/drivers/infiniband/sw/rdmavt/vt.c b/drivers/infiniband/sw/rdmavt/vt.c
index 7de7389d0235..b3f0c5578925 100644
--- a/drivers/infiniband/sw/rdmavt/vt.c
+++ b/drivers/infiniband/sw/rdmavt/vt.c
@@ -91,7 +91,7 @@ struct rvt_dev_info *rvt_alloc_device(size_t size, int nports)
 {
 	struct rvt_dev_info *rdi;
 
-	rdi = (struct rvt_dev_info *)ib_alloc_device(size);
+	rdi = container_of(_ib_alloc_device(size), struct rvt_dev_info, ibdev);
 	if (!rdi)
 		return rdi;
 
diff --git a/drivers/infiniband/sw/rxe/rxe_net.c b/drivers/infiniband/sw/rxe/rxe_net.c
index 8fd03ae20efc..19f3c69916b1 100644
--- a/drivers/infiniband/sw/rxe/rxe_net.c
+++ b/drivers/infiniband/sw/rxe/rxe_net.c
@@ -555,7 +555,7 @@ struct rxe_dev *rxe_net_add(struct net_device *ndev)
 	int err;
 	struct rxe_dev *rxe = NULL;
 
-	rxe = (struct rxe_dev *)ib_alloc_device(sizeof(*rxe));
+	rxe = ib_alloc_device(rxe_dev, ib_dev);
 	if (!rxe)
 		return NULL;
 
diff --git a/include/rdma/ib_verbs.h b/include/rdma/ib_verbs.h
index 71ea144ec823..a1a1e710642c 100644
--- a/include/rdma/ib_verbs.h
+++ b/include/rdma/ib_verbs.h
@@ -2621,7 +2621,13 @@ struct ib_client {
 	struct list_head list;
 };
 
-struct ib_device *ib_alloc_device(size_t size);
+struct ib_device *_ib_alloc_device(size_t size);
+#define ib_alloc_device(drv_struct, member)                                    \
+	container_of(_ib_alloc_device(sizeof(struct drv_struct) +              \
+				      BUILD_BUG_ON_ZERO(offsetof(              \
+					      struct drv_struct, member))),    \
+		     struct drv_struct, member)
+
 void ib_dealloc_device(struct ib_device *device);
 
 void ib_get_device_fw_str(struct ib_device *device, char *str);
-- 
cgit v1.2.3


From cf34e1fe52540842aeae3c5cdaaedae97ee788dd Mon Sep 17 00:00:00 2001
From: Parav Pandit <parav@mellanox.com>
Date: Sun, 27 Jan 2019 20:35:50 +0200
Subject: IB/mlx5: Consider vlan of lower netdev for macvlan GID entries

When a given netdev of the GID entry is macvlan netdevice, and if the
lower netdevice is vlan device, GID entry for macvlan based IP address
needs to inherit the vlan of the lower netdevice.

Therefore, attempt to find out if the lower device exist and if so, if
it is vlan device and setup the vlan tag correctly.

Signed-off-by: Parav Pandit <parav@mellanox.com>
Reviewed-by: Mark Bloch <markb@mellanox.com>
Reviewed-by: Yuval Avnery <yuvalav@mellanox.com>
Signed-off-by: Leon Romanovsky <leonro@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 drivers/infiniband/hw/mlx5/main.c | 40 +++++++++++++++++++++++++++++++++------
 1 file changed, 34 insertions(+), 6 deletions(-)

(limited to 'drivers/infiniband/hw/mlx5/main.c')

diff --git a/drivers/infiniband/hw/mlx5/main.c b/drivers/infiniband/hw/mlx5/main.c
index 8161acda64e6..76d6c2557d0c 100644
--- a/drivers/infiniband/hw/mlx5/main.c
+++ b/drivers/infiniband/hw/mlx5/main.c
@@ -476,24 +476,51 @@ out:
 	return err;
 }
 
+struct mlx5_ib_vlan_info {
+	u16 vlan_id;
+	bool vlan;
+};
+
+static int get_lower_dev_vlan(struct net_device *lower_dev, void *data)
+{
+	struct mlx5_ib_vlan_info *vlan_info = data;
+
+	if (is_vlan_dev(lower_dev)) {
+		vlan_info->vlan = true;
+		vlan_info->vlan_id = vlan_dev_vlan_id(lower_dev);
+	}
+	/* We are interested only in first level vlan device, so
+	 * always return 1 to stop iterating over next level devices.
+	 */
+	return 1;
+}
+
 static int set_roce_addr(struct mlx5_ib_dev *dev, u8 port_num,
 			 unsigned int index, const union ib_gid *gid,
 			 const struct ib_gid_attr *attr)
 {
 	enum ib_gid_type gid_type = IB_GID_TYPE_IB;
+	struct mlx5_ib_vlan_info vlan_info = { };
 	u8 roce_version = 0;
 	u8 roce_l3_type = 0;
-	bool vlan = false;
 	u8 mac[ETH_ALEN];
-	u16 vlan_id = 0;
 
 	if (gid) {
 		gid_type = attr->gid_type;
 		ether_addr_copy(mac, attr->ndev->dev_addr);
 
 		if (is_vlan_dev(attr->ndev)) {
-			vlan = true;
-			vlan_id = vlan_dev_vlan_id(attr->ndev);
+			vlan_info.vlan = true;
+			vlan_info.vlan_id = vlan_dev_vlan_id(attr->ndev);
+		} else {
+			/* If the netdev is upper device and if it's lower
+			 * lower device is vlan device, consider vlan id of
+			 * the lower vlan device for this gid entry.
+			 */
+			rcu_read_lock();
+			netdev_walk_all_lower_dev_rcu(attr->ndev,
+					get_lower_dev_vlan, &vlan_info);
+			rcu_read_unlock();
 		}
 	}
 
@@ -514,8 +541,9 @@ static int set_roce_addr(struct mlx5_ib_dev *dev, u8 port_num,
 	}
 
 	return mlx5_core_roce_gid_set(dev->mdev, index, roce_version,
-				      roce_l3_type, gid->raw, mac, vlan,
-				      vlan_id, port_num);
+				      roce_l3_type, gid->raw, mac,
+				      vlan_info.vlan, vlan_info.vlan_id,
+				      port_num);
 }
 
 static int mlx5_ib_add_gid(const struct ib_gid_attr *attr,
-- 
cgit v1.2.3


From 21a428a019c9a6d133e745b529b9bf18c1187e70 Mon Sep 17 00:00:00 2001
From: Leon Romanovsky <leonro@mellanox.com>
Date: Sun, 3 Feb 2019 14:55:51 +0200
Subject: RDMA: Handle PD allocations by IB/core

The PD allocations in IB/core allows us to simplify drivers and their
error flows in their .alloc_pd() paths. The changes in .alloc_pd() go hand
in had with relevant update in .dealloc_pd().

We will use this opportunity and convert .dealloc_pd() to don't fail, as
it was suggested a long time ago, failures are not happening as we have
never seen a WARN_ON print.

Signed-off-by: Leon Romanovsky <leonro@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 drivers/infiniband/core/device.c                |  2 +
 drivers/infiniband/core/uverbs_cmd.c            | 15 ++--
 drivers/infiniband/core/uverbs_std_types.c      |  2 +-
 drivers/infiniband/core/verbs.c                 | 27 +++++---
 drivers/infiniband/hw/bnxt_re/ib_verbs.c        | 37 ++++------
 drivers/infiniband/hw/bnxt_re/ib_verbs.h        |  9 ++-
 drivers/infiniband/hw/bnxt_re/main.c            |  1 +
 drivers/infiniband/hw/cxgb3/iwch_provider.c     | 25 +++----
 drivers/infiniband/hw/cxgb4/provider.c          | 25 +++----
 drivers/infiniband/hw/hns/hns_roce_device.h     |  7 +-
 drivers/infiniband/hw/hns/hns_roce_hw_v1.c      | 27 +++++---
 drivers/infiniband/hw/hns/hns_roce_main.c       |  1 +
 drivers/infiniband/hw/hns/hns_roce_pd.c         | 25 +++----
 drivers/infiniband/hw/i40iw/i40iw_utils.c       |  1 -
 drivers/infiniband/hw/i40iw/i40iw_verbs.c       | 32 ++++-----
 drivers/infiniband/hw/mlx4/main.c               | 36 ++++------
 drivers/infiniband/hw/mlx5/main.c               | 48 +++++++------
 drivers/infiniband/hw/mthca/mthca_provider.c    | 29 +++-----
 drivers/infiniband/hw/nes/nes_verbs.c           | 32 +++------
 drivers/infiniband/hw/ocrdma/ocrdma_main.c      |  1 +
 drivers/infiniband/hw/ocrdma/ocrdma_verbs.c     | 92 +++++++++++--------------
 drivers/infiniband/hw/ocrdma/ocrdma_verbs.h     |  6 +-
 drivers/infiniband/hw/qedr/main.c               |  1 +
 drivers/infiniband/hw/qedr/verbs.c              | 34 +++------
 drivers/infiniband/hw/qedr/verbs.h              |  6 +-
 drivers/infiniband/hw/usnic/usnic_ib_main.c     |  1 +
 drivers/infiniband/hw/usnic/usnic_ib_verbs.c    | 26 ++-----
 drivers/infiniband/hw/usnic/usnic_ib_verbs.h    |  7 +-
 drivers/infiniband/hw/vmw_pvrdma/pvrdma_main.c  |  1 +
 drivers/infiniband/hw/vmw_pvrdma/pvrdma_verbs.c | 43 ++++--------
 drivers/infiniband/hw/vmw_pvrdma/pvrdma_verbs.h |  7 +-
 drivers/infiniband/sw/rdmavt/pd.c               | 29 +++-----
 drivers/infiniband/sw/rdmavt/pd.h               |  7 +-
 drivers/infiniband/sw/rdmavt/vt.c               |  1 +
 drivers/infiniband/sw/rxe/rxe_pool.c            | 60 +++++++++++++---
 drivers/infiniband/sw/rxe/rxe_pool.h            |  4 ++
 drivers/infiniband/sw/rxe/rxe_verbs.c           | 16 ++---
 drivers/infiniband/sw/rxe/rxe_verbs.h           |  2 +-
 include/rdma/ib_verbs.h                         |  9 +--
 39 files changed, 325 insertions(+), 409 deletions(-)

(limited to 'drivers/infiniband/hw/mlx5/main.c')

diff --git a/drivers/infiniband/core/device.c b/drivers/infiniband/core/device.c
index d806a5c7b202..57e1e177921e 100644
--- a/drivers/infiniband/core/device.c
+++ b/drivers/infiniband/core/device.c
@@ -1319,6 +1319,8 @@ void ib_set_device_ops(struct ib_device *dev, const struct ib_device_ops *ops)
 	SET_DEVICE_OP(dev_ops, set_vf_guid);
 	SET_DEVICE_OP(dev_ops, set_vf_link_state);
 	SET_DEVICE_OP(dev_ops, unmap_fmr);
+
+	SET_OBJ_SIZE(dev_ops, ib_pd);
 }
 EXPORT_SYMBOL(ib_set_device_ops);
 
diff --git a/drivers/infiniband/core/uverbs_cmd.c b/drivers/infiniband/core/uverbs_cmd.c
index aa260cafbd85..5ac143f22df0 100644
--- a/drivers/infiniband/core/uverbs_cmd.c
+++ b/drivers/infiniband/core/uverbs_cmd.c
@@ -407,9 +407,9 @@ static int ib_uverbs_alloc_pd(struct uverbs_attr_bundle *attrs)
 	if (IS_ERR(uobj))
 		return PTR_ERR(uobj);
 
-	pd = ib_dev->ops.alloc_pd(ib_dev, uobj->context, &attrs->driver_udata);
-	if (IS_ERR(pd)) {
-		ret = PTR_ERR(pd);
+	pd = rdma_zalloc_drv_obj(ib_dev, ib_pd);
+	if (!pd) {
+		ret = -ENOMEM;
 		goto err;
 	}
 
@@ -417,11 +417,15 @@ static int ib_uverbs_alloc_pd(struct uverbs_attr_bundle *attrs)
 	pd->uobject = uobj;
 	pd->__internal_mr = NULL;
 	atomic_set(&pd->usecnt, 0);
+	pd->res.type = RDMA_RESTRACK_PD;
+
+	ret = ib_dev->ops.alloc_pd(pd, uobj->context, &attrs->driver_udata);
+	if (ret)
+		goto err_alloc;
 
 	uobj->object = pd;
 	memset(&resp, 0, sizeof resp);
 	resp.pd_handle = uobj->id;
-	pd->res.type = RDMA_RESTRACK_PD;
 	rdma_restrack_uadd(&pd->res);
 
 	ret = uverbs_response(attrs, &resp, sizeof(resp));
@@ -432,7 +436,8 @@ static int ib_uverbs_alloc_pd(struct uverbs_attr_bundle *attrs)
 
 err_copy:
 	ib_dealloc_pd(pd);
-
+err_alloc:
+	kfree(pd);
 err:
 	uobj_alloc_abort(uobj);
 	return ret;
diff --git a/drivers/infiniband/core/uverbs_std_types.c b/drivers/infiniband/core/uverbs_std_types.c
index cbc72312eb41..f224cb727224 100644
--- a/drivers/infiniband/core/uverbs_std_types.c
+++ b/drivers/infiniband/core/uverbs_std_types.c
@@ -188,7 +188,7 @@ static int uverbs_free_pd(struct ib_uobject *uobject,
 	if (ret)
 		return ret;
 
-	ib_dealloc_pd((struct ib_pd *)uobject->object);
+	ib_dealloc_pd(pd);
 	return 0;
 }
 
diff --git a/drivers/infiniband/core/verbs.c b/drivers/infiniband/core/verbs.c
index 3220fb42ecce..de5d895a5054 100644
--- a/drivers/infiniband/core/verbs.c
+++ b/drivers/infiniband/core/verbs.c
@@ -254,10 +254,11 @@ struct ib_pd *__ib_alloc_pd(struct ib_device *device, unsigned int flags,
 {
 	struct ib_pd *pd;
 	int mr_access_flags = 0;
+	int ret;
 
-	pd = device->ops.alloc_pd(device, NULL, NULL);
-	if (IS_ERR(pd))
-		return pd;
+	pd = rdma_zalloc_drv_obj(device, ib_pd);
+	if (!pd)
+		return ERR_PTR(-ENOMEM);
 
 	pd->device = device;
 	pd->uobject = NULL;
@@ -265,6 +266,16 @@ struct ib_pd *__ib_alloc_pd(struct ib_device *device, unsigned int flags,
 	atomic_set(&pd->usecnt, 0);
 	pd->flags = flags;
 
+	pd->res.type = RDMA_RESTRACK_PD;
+	rdma_restrack_set_task(&pd->res, caller);
+
+	ret = device->ops.alloc_pd(pd, NULL, NULL);
+	if (ret) {
+		kfree(pd);
+		return ERR_PTR(ret);
+	}
+	rdma_restrack_kadd(&pd->res);
+
 	if (device->attrs.device_cap_flags & IB_DEVICE_LOCAL_DMA_LKEY)
 		pd->local_dma_lkey = device->local_dma_lkey;
 	else
@@ -275,10 +286,6 @@ struct ib_pd *__ib_alloc_pd(struct ib_device *device, unsigned int flags,
 		mr_access_flags |= IB_ACCESS_REMOTE_READ | IB_ACCESS_REMOTE_WRITE;
 	}
 
-	pd->res.type = RDMA_RESTRACK_PD;
-	rdma_restrack_set_task(&pd->res, caller);
-	rdma_restrack_kadd(&pd->res);
-
 	if (mr_access_flags) {
 		struct ib_mr *mr;
 
@@ -329,10 +336,8 @@ void ib_dealloc_pd(struct ib_pd *pd)
 	WARN_ON(atomic_read(&pd->usecnt));
 
 	rdma_restrack_del(&pd->res);
-	/* Making delalloc_pd a void return is a WIP, no driver should return
-	   an error here. */
-	ret = pd->device->ops.dealloc_pd(pd);
-	WARN_ONCE(ret, "Infiniband HW driver failed dealloc_pd");
+	pd->device->ops.dealloc_pd(pd);
+	kfree(pd);
 }
 EXPORT_SYMBOL(ib_dealloc_pd);
 
diff --git a/drivers/infiniband/hw/bnxt_re/ib_verbs.c b/drivers/infiniband/hw/bnxt_re/ib_verbs.c
index 1d7469e23cde..1606571af63d 100644
--- a/drivers/infiniband/hw/bnxt_re/ib_verbs.c
+++ b/drivers/infiniband/hw/bnxt_re/ib_verbs.c
@@ -563,41 +563,29 @@ fail:
 }
 
 /* Protection Domains */
-int bnxt_re_dealloc_pd(struct ib_pd *ib_pd)
+void bnxt_re_dealloc_pd(struct ib_pd *ib_pd)
 {
 	struct bnxt_re_pd *pd = container_of(ib_pd, struct bnxt_re_pd, ib_pd);
 	struct bnxt_re_dev *rdev = pd->rdev;
-	int rc;
 
 	bnxt_re_destroy_fence_mr(pd);
 
-	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 PD");
-	}
-
-	kfree(pd);
-	return 0;
+	if (pd->qplib_pd.id)
+		bnxt_qplib_dealloc_pd(&rdev->qplib_res, &rdev->qplib_res.pd_tbl,
+				      &pd->qplib_pd);
 }
 
-struct ib_pd *bnxt_re_alloc_pd(struct ib_device *ibdev,
-			       struct ib_ucontext *ucontext,
-			       struct ib_udata *udata)
+int bnxt_re_alloc_pd(struct ib_pd *ibpd, struct ib_ucontext *ucontext,
+		     struct ib_udata *udata)
 {
+	struct ib_device *ibdev = ibpd->device;
 	struct bnxt_re_dev *rdev = to_bnxt_re_dev(ibdev, ibdev);
 	struct bnxt_re_ucontext *ucntx = container_of(ucontext,
 						      struct bnxt_re_ucontext,
 						      ib_uctx);
-	struct bnxt_re_pd *pd;
+	struct bnxt_re_pd *pd = container_of(ibpd, struct bnxt_re_pd, ib_pd);
 	int rc;
 
-	pd = kzalloc(sizeof(*pd), GFP_KERNEL);
-	if (!pd)
-		return ERR_PTR(-ENOMEM);
-
 	pd->rdev = rdev;
 	if (bnxt_qplib_alloc_pd(&rdev->qplib_res.pd_tbl, &pd->qplib_pd)) {
 		dev_err(rdev_to_dev(rdev), "Failed to allocate HW PD");
@@ -637,13 +625,12 @@ struct ib_pd *bnxt_re_alloc_pd(struct ib_device *ibdev,
 		if (bnxt_re_create_fence_mr(pd))
 			dev_warn(rdev_to_dev(rdev),
 				 "Failed to create Fence-MR\n");
-	return &pd->ib_pd;
+	return 0;
 dbfail:
-	(void)bnxt_qplib_dealloc_pd(&rdev->qplib_res, &rdev->qplib_res.pd_tbl,
-				    &pd->qplib_pd);
+	bnxt_qplib_dealloc_pd(&rdev->qplib_res, &rdev->qplib_res.pd_tbl,
+			      &pd->qplib_pd);
 fail:
-	kfree(pd);
-	return ERR_PTR(rc);
+	return rc;
 }
 
 /* Address Handles */
diff --git a/drivers/infiniband/hw/bnxt_re/ib_verbs.h b/drivers/infiniband/hw/bnxt_re/ib_verbs.h
index c4af72604b4f..c7cca803cfa3 100644
--- a/drivers/infiniband/hw/bnxt_re/ib_verbs.h
+++ b/drivers/infiniband/hw/bnxt_re/ib_verbs.h
@@ -56,8 +56,8 @@ struct bnxt_re_fence_data {
 };
 
 struct bnxt_re_pd {
+	struct ib_pd            ib_pd;
 	struct bnxt_re_dev	*rdev;
-	struct ib_pd		ib_pd;
 	struct bnxt_qplib_pd	qplib_pd;
 	struct bnxt_re_fence_data fence;
 };
@@ -163,10 +163,9 @@ int bnxt_re_query_gid(struct ib_device *ibdev, u8 port_num,
 		      int index, union ib_gid *gid);
 enum rdma_link_layer bnxt_re_get_link_layer(struct ib_device *ibdev,
 					    u8 port_num);
-struct ib_pd *bnxt_re_alloc_pd(struct ib_device *ibdev,
-			       struct ib_ucontext *context,
-			       struct ib_udata *udata);
-int bnxt_re_dealloc_pd(struct ib_pd *pd);
+int bnxt_re_alloc_pd(struct ib_pd *pd, struct ib_ucontext *context,
+		     struct ib_udata *udata);
+void bnxt_re_dealloc_pd(struct ib_pd *pd);
 struct ib_ah *bnxt_re_create_ah(struct ib_pd *pd,
 				struct rdma_ah_attr *ah_attr,
 				u32 flags,
diff --git a/drivers/infiniband/hw/bnxt_re/main.c b/drivers/infiniband/hw/bnxt_re/main.c
index 0d40a930c192..0a89ef6e5754 100644
--- a/drivers/infiniband/hw/bnxt_re/main.c
+++ b/drivers/infiniband/hw/bnxt_re/main.c
@@ -637,6 +637,7 @@ static const struct ib_device_ops bnxt_re_dev_ops = {
 	.query_srq = bnxt_re_query_srq,
 	.reg_user_mr = bnxt_re_reg_user_mr,
 	.req_notify_cq = bnxt_re_req_notify_cq,
+	INIT_RDMA_OBJ_SIZE(ib_pd, bnxt_re_pd, ib_pd),
 };
 
 static int bnxt_re_register_ib(struct bnxt_re_dev *rdev)
diff --git a/drivers/infiniband/hw/cxgb3/iwch_provider.c b/drivers/infiniband/hw/cxgb3/iwch_provider.c
index 804c1fc7bfc1..4cc9a6ae2139 100644
--- a/drivers/infiniband/hw/cxgb3/iwch_provider.c
+++ b/drivers/infiniband/hw/cxgb3/iwch_provider.c
@@ -370,7 +370,7 @@ static int iwch_mmap(struct ib_ucontext *context, struct vm_area_struct *vma)
 	return ret;
 }
 
-static int iwch_deallocate_pd(struct ib_pd *pd)
+static void iwch_deallocate_pd(struct ib_pd *pd)
 {
 	struct iwch_dev *rhp;
 	struct iwch_pd *php;
@@ -379,15 +379,13 @@ static int iwch_deallocate_pd(struct ib_pd *pd)
 	rhp = php->rhp;
 	pr_debug("%s ibpd %p pdid 0x%x\n", __func__, pd, php->pdid);
 	cxio_hal_put_pdid(rhp->rdev.rscp, php->pdid);
-	kfree(php);
-	return 0;
 }
 
-static struct ib_pd *iwch_allocate_pd(struct ib_device *ibdev,
-			       struct ib_ucontext *context,
-			       struct ib_udata *udata)
+static int iwch_allocate_pd(struct ib_pd *pd, struct ib_ucontext *context,
+			    struct ib_udata *udata)
 {
-	struct iwch_pd *php;
+	struct iwch_pd *php = to_iwch_pd(pd);
+	struct ib_device *ibdev = pd->device;
 	u32 pdid;
 	struct iwch_dev *rhp;
 
@@ -395,12 +393,8 @@ static struct ib_pd *iwch_allocate_pd(struct ib_device *ibdev,
 	rhp = (struct iwch_dev *) ibdev;
 	pdid = cxio_hal_get_pdid(rhp->rdev.rscp);
 	if (!pdid)
-		return ERR_PTR(-EINVAL);
-	php = kzalloc(sizeof(*php), GFP_KERNEL);
-	if (!php) {
-		cxio_hal_put_pdid(rhp->rdev.rscp, pdid);
-		return ERR_PTR(-ENOMEM);
-	}
+		return -EINVAL;
+
 	php->pdid = pdid;
 	php->rhp = rhp;
 	if (context) {
@@ -408,11 +402,11 @@ static struct ib_pd *iwch_allocate_pd(struct ib_device *ibdev,
 
 		if (ib_copy_to_udata(udata, &resp, sizeof(resp))) {
 			iwch_deallocate_pd(&php->ibpd);
-			return ERR_PTR(-EFAULT);
+			return -EFAULT;
 		}
 	}
 	pr_debug("%s pdid 0x%0x ptr 0x%p\n", __func__, pdid, php);
-	return &php->ibpd;
+	return 0;
 }
 
 static int iwch_dereg_mr(struct ib_mr *ib_mr)
@@ -1350,6 +1344,7 @@ static const struct ib_device_ops iwch_dev_ops = {
 	.reg_user_mr = iwch_reg_user_mr,
 	.req_notify_cq = iwch_arm_cq,
 	.resize_cq = iwch_resize_cq,
+	INIT_RDMA_OBJ_SIZE(ib_pd, iwch_pd, ibpd),
 };
 
 int iwch_register_device(struct iwch_dev *dev)
diff --git a/drivers/infiniband/hw/cxgb4/provider.c b/drivers/infiniband/hw/cxgb4/provider.c
index f59bf7e5a589..680b5e98491d 100644
--- a/drivers/infiniband/hw/cxgb4/provider.c
+++ b/drivers/infiniband/hw/cxgb4/provider.c
@@ -209,7 +209,7 @@ static int c4iw_mmap(struct ib_ucontext *context, struct vm_area_struct *vma)
 	return ret;
 }
 
-static int c4iw_deallocate_pd(struct ib_pd *pd)
+static void c4iw_deallocate_pd(struct ib_pd *pd)
 {
 	struct c4iw_dev *rhp;
 	struct c4iw_pd *php;
@@ -221,15 +221,13 @@ static int c4iw_deallocate_pd(struct ib_pd *pd)
 	mutex_lock(&rhp->rdev.stats.lock);
 	rhp->rdev.stats.pd.cur--;
 	mutex_unlock(&rhp->rdev.stats.lock);
-	kfree(php);
-	return 0;
 }
 
-static struct ib_pd *c4iw_allocate_pd(struct ib_device *ibdev,
-				      struct ib_ucontext *context,
-				      struct ib_udata *udata)
+static int c4iw_allocate_pd(struct ib_pd *pd, struct ib_ucontext *context,
+			    struct ib_udata *udata)
 {
-	struct c4iw_pd *php;
+	struct c4iw_pd *php = to_c4iw_pd(pd);
+	struct ib_device *ibdev = pd->device;
 	u32 pdid;
 	struct c4iw_dev *rhp;
 
@@ -237,12 +235,8 @@ static struct ib_pd *c4iw_allocate_pd(struct ib_device *ibdev,
 	rhp = (struct c4iw_dev *) ibdev;
 	pdid =  c4iw_get_resource(&rhp->rdev.resource.pdid_table);
 	if (!pdid)
-		return ERR_PTR(-EINVAL);
-	php = kzalloc(sizeof(*php), GFP_KERNEL);
-	if (!php) {
-		c4iw_put_resource(&rhp->rdev.resource.pdid_table, pdid);
-		return ERR_PTR(-ENOMEM);
-	}
+		return -EINVAL;
+
 	php->pdid = pdid;
 	php->rhp = rhp;
 	if (context) {
@@ -250,7 +244,7 @@ static struct ib_pd *c4iw_allocate_pd(struct ib_device *ibdev,
 
 		if (ib_copy_to_udata(udata, &uresp, sizeof(uresp))) {
 			c4iw_deallocate_pd(&php->ibpd);
-			return ERR_PTR(-EFAULT);
+			return -EFAULT;
 		}
 	}
 	mutex_lock(&rhp->rdev.stats.lock);
@@ -259,7 +253,7 @@ static struct ib_pd *c4iw_allocate_pd(struct ib_device *ibdev,
 		rhp->rdev.stats.pd.max = rhp->rdev.stats.pd.cur;
 	mutex_unlock(&rhp->rdev.stats.lock);
 	pr_debug("pdid 0x%0x ptr 0x%p\n", pdid, php);
-	return &php->ibpd;
+	return 0;
 }
 
 static int c4iw_query_pkey(struct ib_device *ibdev, u8 port, u16 index,
@@ -570,6 +564,7 @@ static const struct ib_device_ops c4iw_dev_ops = {
 	.query_qp = c4iw_ib_query_qp,
 	.reg_user_mr = c4iw_reg_user_mr,
 	.req_notify_cq = c4iw_arm_cq,
+	INIT_RDMA_OBJ_SIZE(ib_pd, c4iw_pd, ibpd),
 };
 
 void c4iw_register_device(struct work_struct *work)
diff --git a/drivers/infiniband/hw/hns/hns_roce_device.h b/drivers/infiniband/hw/hns/hns_roce_device.h
index 8ca8d74dfb6a..9ee86daf1700 100644
--- a/drivers/infiniband/hw/hns/hns_roce_device.h
+++ b/drivers/infiniband/hw/hns/hns_roce_device.h
@@ -1114,10 +1114,9 @@ struct ib_ah *hns_roce_create_ah(struct ib_pd *pd,
 int hns_roce_query_ah(struct ib_ah *ibah, struct rdma_ah_attr *ah_attr);
 int hns_roce_destroy_ah(struct ib_ah *ah, u32 flags);
 
-struct ib_pd *hns_roce_alloc_pd(struct ib_device *ib_dev,
-				struct ib_ucontext *context,
-				struct ib_udata *udata);
-int hns_roce_dealloc_pd(struct ib_pd *pd);
+int hns_roce_alloc_pd(struct ib_pd *pd, struct ib_ucontext *context,
+		      struct ib_udata *udata);
+void hns_roce_dealloc_pd(struct ib_pd *pd);
 
 struct ib_mr *hns_roce_get_dma_mr(struct ib_pd *pd, int acc);
 struct ib_mr *hns_roce_reg_user_mr(struct ib_pd *pd, u64 start, u64 length,
diff --git a/drivers/infiniband/hw/hns/hns_roce_hw_v1.c b/drivers/infiniband/hw/hns/hns_roce_hw_v1.c
index fa08c22aad66..a18b88c95995 100644
--- a/drivers/infiniband/hw/hns/hns_roce_hw_v1.c
+++ b/drivers/infiniband/hw/hns/hns_roce_hw_v1.c
@@ -711,13 +711,14 @@ static int hns_roce_v1_rsv_lp_qp(struct hns_roce_dev *hr_dev)
 	struct ib_qp_attr attr = { 0 };
 	struct hns_roce_v1_priv *priv;
 	struct hns_roce_qp *hr_qp;
+	struct ib_device *ibdev;
 	struct ib_cq *cq;
 	struct ib_pd *pd;
 	union ib_gid dgid;
 	u64 subnet_prefix;
 	int attr_mask = 0;
+	int ret = -ENOMEM;
 	int i, j;
-	int ret;
 	u8 queue_en[HNS_ROCE_V1_RESV_QP] = { 0 };
 	u8 phy_port;
 	u8 port = 0;
@@ -742,12 +743,16 @@ static int hns_roce_v1_rsv_lp_qp(struct hns_roce_dev *hr_dev)
 	free_mr->mr_free_cq->ib_cq.cq_context		= NULL;
 	atomic_set(&free_mr->mr_free_cq->ib_cq.usecnt, 0);
 
-	pd = hns_roce_alloc_pd(&hr_dev->ib_dev, NULL, NULL);
-	if (IS_ERR(pd)) {
-		dev_err(dev, "Create pd for reserved loop qp failed!");
-		ret = -ENOMEM;
+	ibdev = &hr_dev->ib_dev;
+	pd = rdma_zalloc_drv_obj(ibdev, ib_pd);
+	if (pd)
+		goto alloc_mem_failed;
+
+	pd->device  = ibdev;
+	ret = hns_roce_alloc_pd(pd, NULL, NULL);
+	if (ret)
 		goto alloc_pd_failed;
-	}
+
 	free_mr->mr_free_pd = to_hr_pd(pd);
 	free_mr->mr_free_pd->ibpd.device  = &hr_dev->ib_dev;
 	free_mr->mr_free_pd->ibpd.uobject = NULL;
@@ -854,10 +859,12 @@ create_lp_qp_failed:
 			dev_err(dev, "Destroy qp %d for mr free failed!\n", i);
 	}
 
-	if (hns_roce_dealloc_pd(pd))
-		dev_err(dev, "Destroy pd for create_lp_qp failed!\n");
+	hns_roce_dealloc_pd(pd);
 
 alloc_pd_failed:
+	kfree(pd);
+
+alloc_mem_failed:
 	if (hns_roce_ib_destroy_cq(cq))
 		dev_err(dev, "Destroy cq for create_lp_qp failed!\n");
 
@@ -891,9 +898,7 @@ static void hns_roce_v1_release_lp_qp(struct hns_roce_dev *hr_dev)
 	if (ret)
 		dev_err(dev, "Destroy cq for mr_free failed(%d)!\n", ret);
 
-	ret = hns_roce_dealloc_pd(&free_mr->mr_free_pd->ibpd);
-	if (ret)
-		dev_err(dev, "Destroy pd for mr_free failed(%d)!\n", ret);
+	hns_roce_dealloc_pd(&free_mr->mr_free_pd->ibpd);
 }
 
 static int hns_roce_db_init(struct hns_roce_dev *hr_dev)
diff --git a/drivers/infiniband/hw/hns/hns_roce_main.c b/drivers/infiniband/hw/hns/hns_roce_main.c
index 67a8c4333f4f..ccf10622586c 100644
--- a/drivers/infiniband/hw/hns/hns_roce_main.c
+++ b/drivers/infiniband/hw/hns/hns_roce_main.c
@@ -472,6 +472,7 @@ static const struct ib_device_ops hns_roce_dev_ops = {
 	.query_pkey = hns_roce_query_pkey,
 	.query_port = hns_roce_query_port,
 	.reg_user_mr = hns_roce_reg_user_mr,
+	INIT_RDMA_OBJ_SIZE(ib_pd, hns_roce_pd, ibpd),
 };
 
 static const struct ib_device_ops hns_roce_dev_mr_ops = {
diff --git a/drivers/infiniband/hw/hns/hns_roce_pd.c b/drivers/infiniband/hw/hns/hns_roce_pd.c
index 4a29b2cb9bab..b9b97c5e97e6 100644
--- a/drivers/infiniband/hw/hns/hns_roce_pd.c
+++ b/drivers/infiniband/hw/hns/hns_roce_pd.c
@@ -57,24 +57,19 @@ void hns_roce_cleanup_pd_table(struct hns_roce_dev *hr_dev)
 	hns_roce_bitmap_cleanup(&hr_dev->pd_bitmap);
 }
 
-struct ib_pd *hns_roce_alloc_pd(struct ib_device *ib_dev,
-				struct ib_ucontext *context,
-				struct ib_udata *udata)
+int hns_roce_alloc_pd(struct ib_pd *ibpd, struct ib_ucontext *context,
+		      struct ib_udata *udata)
 {
+	struct ib_device *ib_dev = ibpd->device;
 	struct hns_roce_dev *hr_dev = to_hr_dev(ib_dev);
 	struct device *dev = hr_dev->dev;
-	struct hns_roce_pd *pd;
+	struct hns_roce_pd *pd = to_hr_pd(ibpd);
 	int ret;
 
-	pd = kzalloc(sizeof(*pd), GFP_KERNEL);
-	if (!pd)
-		return ERR_PTR(-ENOMEM);
-
 	ret = hns_roce_pd_alloc(to_hr_dev(ib_dev), &pd->pdn);
 	if (ret) {
-		kfree(pd);
 		dev_err(dev, "[alloc_pd]hns_roce_pd_alloc failed!\n");
-		return ERR_PTR(ret);
+		return ret;
 	}
 
 	if (context) {
@@ -83,21 +78,17 @@ struct ib_pd *hns_roce_alloc_pd(struct ib_device *ib_dev,
 		if (ib_copy_to_udata(udata, &uresp, sizeof(uresp))) {
 			hns_roce_pd_free(to_hr_dev(ib_dev), pd->pdn);
 			dev_err(dev, "[alloc_pd]ib_copy_to_udata failed!\n");
-			kfree(pd);
-			return ERR_PTR(-EFAULT);
+			return -EFAULT;
 		}
 	}
 
-	return &pd->ibpd;
+	return 0;
 }
 EXPORT_SYMBOL_GPL(hns_roce_alloc_pd);
 
-int hns_roce_dealloc_pd(struct ib_pd *pd)
+void hns_roce_dealloc_pd(struct ib_pd *pd)
 {
 	hns_roce_pd_free(to_hr_dev(pd->device), to_hr_pd(pd)->pdn);
-	kfree(to_hr_pd(pd));
-
-	return 0;
 }
 EXPORT_SYMBOL_GPL(hns_roce_dealloc_pd);
 
diff --git a/drivers/infiniband/hw/i40iw/i40iw_utils.c b/drivers/infiniband/hw/i40iw/i40iw_utils.c
index 59e978141ad4..c5a881172524 100644
--- a/drivers/infiniband/hw/i40iw/i40iw_utils.c
+++ b/drivers/infiniband/hw/i40iw/i40iw_utils.c
@@ -601,7 +601,6 @@ void i40iw_rem_pdusecount(struct i40iw_pd *iwpd, struct i40iw_device *iwdev)
 	if (!atomic_dec_and_test(&iwpd->usecount))
 		return;
 	i40iw_free_resource(iwdev, iwdev->allocated_pds, iwpd->sc_pd.pd_id);
-	kfree(iwpd);
 }
 
 /**
diff --git a/drivers/infiniband/hw/i40iw/i40iw_verbs.c b/drivers/infiniband/hw/i40iw/i40iw_verbs.c
index d4ab46dd9e6c..28449ad57b37 100644
--- a/drivers/infiniband/hw/i40iw/i40iw_verbs.c
+++ b/drivers/infiniband/hw/i40iw/i40iw_verbs.c
@@ -312,16 +312,15 @@ static void i40iw_dealloc_push_page(struct i40iw_device *iwdev, struct i40iw_sc_
 
 /**
  * i40iw_alloc_pd - allocate protection domain
- * @ibdev: device pointer from stack
+ * @pd: PD pointer
  * @context: user context created during alloc
  * @udata: user data
  */
-static struct ib_pd *i40iw_alloc_pd(struct ib_device *ibdev,
-				    struct ib_ucontext *context,
-				    struct ib_udata *udata)
+static int i40iw_alloc_pd(struct ib_pd *pd, struct ib_ucontext *context,
+			  struct ib_udata *udata)
 {
-	struct i40iw_pd *iwpd;
-	struct i40iw_device *iwdev = to_iwdev(ibdev);
+	struct i40iw_pd *iwpd = to_iwpd(pd);
+	struct i40iw_device *iwdev = to_iwdev(pd->device);
 	struct i40iw_sc_dev *dev = &iwdev->sc_dev;
 	struct i40iw_alloc_pd_resp uresp;
 	struct i40iw_sc_pd *sc_pd;
@@ -330,19 +329,13 @@ static struct ib_pd *i40iw_alloc_pd(struct ib_device *ibdev,
 	int err;
 
 	if (iwdev->closing)
-		return ERR_PTR(-ENODEV);
+		return -ENODEV;
 
 	err = i40iw_alloc_resource(iwdev, iwdev->allocated_pds,
 				   iwdev->max_pd, &pd_id, &iwdev->next_pd);
 	if (err) {
 		i40iw_pr_err("alloc resource failed\n");
-		return ERR_PTR(err);
-	}
-
-	iwpd = kzalloc(sizeof(*iwpd), GFP_KERNEL);
-	if (!iwpd) {
-		err = -ENOMEM;
-		goto free_res;
+		return err;
 	}
 
 	sc_pd = &iwpd->sc_pd;
@@ -361,25 +354,23 @@ static struct ib_pd *i40iw_alloc_pd(struct ib_device *ibdev,
 	}
 
 	i40iw_add_pdusecount(iwpd);
-	return &iwpd->ibpd;
+	return 0;
+
 error:
-	kfree(iwpd);
-free_res:
 	i40iw_free_resource(iwdev, iwdev->allocated_pds, pd_id);
-	return ERR_PTR(err);
+	return err;
 }
 
 /**
  * i40iw_dealloc_pd - deallocate pd
  * @ibpd: ptr of pd to be deallocated
  */
-static int i40iw_dealloc_pd(struct ib_pd *ibpd)
+static void i40iw_dealloc_pd(struct ib_pd *ibpd)
 {
 	struct i40iw_pd *iwpd = to_iwpd(ibpd);
 	struct i40iw_device *iwdev = to_iwdev(ibpd->device);
 
 	i40iw_rem_pdusecount(iwpd, iwdev);
-	return 0;
 }
 
 /**
@@ -2750,6 +2741,7 @@ static const struct ib_device_ops i40iw_dev_ops = {
 	.query_qp = i40iw_query_qp,
 	.reg_user_mr = i40iw_reg_user_mr,
 	.req_notify_cq = i40iw_req_notify_cq,
+	INIT_RDMA_OBJ_SIZE(ib_pd, i40iw_pd, ibpd),
 };
 
 /**
diff --git a/drivers/infiniband/hw/mlx4/main.c b/drivers/infiniband/hw/mlx4/main.c
index d66002a31000..c0f6aea7ed7c 100644
--- a/drivers/infiniband/hw/mlx4/main.c
+++ b/drivers/infiniband/hw/mlx4/main.c
@@ -1186,38 +1186,27 @@ static int mlx4_ib_mmap(struct ib_ucontext *context, struct vm_area_struct *vma)
 	}
 }
 
-static struct ib_pd *mlx4_ib_alloc_pd(struct ib_device *ibdev,
-				      struct ib_ucontext *context,
-				      struct ib_udata *udata)
+static int mlx4_ib_alloc_pd(struct ib_pd *ibpd, struct ib_ucontext *context,
+			    struct ib_udata *udata)
 {
-	struct mlx4_ib_pd *pd;
+	struct mlx4_ib_pd *pd = to_mpd(ibpd);
+	struct ib_device *ibdev = ibpd->device;
 	int err;
 
-	pd = kzalloc(sizeof(*pd), GFP_KERNEL);
-	if (!pd)
-		return ERR_PTR(-ENOMEM);
-
 	err = mlx4_pd_alloc(to_mdev(ibdev)->dev, &pd->pdn);
-	if (err) {
-		kfree(pd);
-		return ERR_PTR(err);
-	}
+	if (err)
+		return err;
 
-	if (context)
-		if (ib_copy_to_udata(udata, &pd->pdn, sizeof (__u32))) {
-			mlx4_pd_free(to_mdev(ibdev)->dev, pd->pdn);
-			kfree(pd);
-			return ERR_PTR(-EFAULT);
-		}
-	return &pd->ibpd;
+	if (context && ib_copy_to_udata(udata, &pd->pdn, sizeof(__u32))) {
+		mlx4_pd_free(to_mdev(ibdev)->dev, pd->pdn);
+		return -EFAULT;
+	}
+	return 0;
 }
 
-static int mlx4_ib_dealloc_pd(struct ib_pd *pd)
+static void mlx4_ib_dealloc_pd(struct ib_pd *pd)
 {
 	mlx4_pd_free(to_mdev(pd->device)->dev, to_mpd(pd)->pdn);
-	kfree(pd);
-
-	return 0;
 }
 
 static struct ib_xrcd *mlx4_ib_alloc_xrcd(struct ib_device *ibdev,
@@ -2580,6 +2569,7 @@ static const struct ib_device_ops mlx4_ib_dev_ops = {
 	.req_notify_cq = mlx4_ib_arm_cq,
 	.rereg_user_mr = mlx4_ib_rereg_user_mr,
 	.resize_cq = mlx4_ib_resize_cq,
+	INIT_RDMA_OBJ_SIZE(ib_pd, mlx4_ib_pd, ibpd),
 };
 
 static const struct ib_device_ops mlx4_ib_dev_wq_ops = {
diff --git a/drivers/infiniband/hw/mlx5/main.c b/drivers/infiniband/hw/mlx5/main.c
index 76d6c2557d0c..f9cddc6f2ab6 100644
--- a/drivers/infiniband/hw/mlx5/main.c
+++ b/drivers/infiniband/hw/mlx5/main.c
@@ -2280,30 +2280,24 @@ int mlx5_ib_dealloc_dm(struct ib_dm *ibdm)
 	return 0;
 }
 
-static struct ib_pd *mlx5_ib_alloc_pd(struct ib_device *ibdev,
-				      struct ib_ucontext *context,
-				      struct ib_udata *udata)
+static int mlx5_ib_alloc_pd(struct ib_pd *ibpd, struct ib_ucontext *context,
+			    struct ib_udata *udata)
 {
+	struct mlx5_ib_pd *pd = to_mpd(ibpd);
+	struct ib_device *ibdev = ibpd->device;
 	struct mlx5_ib_alloc_pd_resp resp;
-	struct mlx5_ib_pd *pd;
 	int err;
 	u32 out[MLX5_ST_SZ_DW(alloc_pd_out)] = {};
 	u32 in[MLX5_ST_SZ_DW(alloc_pd_in)]   = {};
 	u16 uid = 0;
 
-	pd = kzalloc(sizeof(*pd), GFP_KERNEL);
-	if (!pd)
-		return ERR_PTR(-ENOMEM);
-
 	uid = context ? to_mucontext(context)->devx_uid : 0;
 	MLX5_SET(alloc_pd_in, in, opcode, MLX5_CMD_OP_ALLOC_PD);
 	MLX5_SET(alloc_pd_in, in, uid, uid);
 	err = mlx5_cmd_exec(to_mdev(ibdev)->mdev, in, sizeof(in),
 			    out, sizeof(out));
-	if (err) {
-		kfree(pd);
-		return ERR_PTR(err);
-	}
+	if (err)
+		return err;
 
 	pd->pdn = MLX5_GET(alloc_pd_out, out, pd);
 	pd->uid = uid;
@@ -2311,23 +2305,19 @@ static struct ib_pd *mlx5_ib_alloc_pd(struct ib_device *ibdev,
 		resp.pdn = pd->pdn;
 		if (ib_copy_to_udata(udata, &resp, sizeof(resp))) {
 			mlx5_cmd_dealloc_pd(to_mdev(ibdev)->mdev, pd->pdn, uid);
-			kfree(pd);
-			return ERR_PTR(-EFAULT);
+			return -EFAULT;
 		}
 	}
 
-	return &pd->ibpd;
+	return 0;
 }
 
-static int mlx5_ib_dealloc_pd(struct ib_pd *pd)
+static void mlx5_ib_dealloc_pd(struct ib_pd *pd)
 {
 	struct mlx5_ib_dev *mdev = to_mdev(pd->device);
 	struct mlx5_ib_pd *mpd = to_mpd(pd);
 
 	mlx5_cmd_dealloc_pd(mdev->mdev, mpd->pdn, mpd->uid);
-	kfree(mpd);
-
-	return 0;
 }
 
 enum {
@@ -4680,23 +4670,28 @@ static int create_dev_resources(struct mlx5_ib_resources *devr)
 {
 	struct ib_srq_init_attr attr;
 	struct mlx5_ib_dev *dev;
+	struct ib_device *ibdev;
 	struct ib_cq_init_attr cq_attr = {.cqe = 1};
 	int port;
 	int ret = 0;
 
 	dev = container_of(devr, struct mlx5_ib_dev, devr);
+	ibdev = &dev->ib_dev;
 
 	mutex_init(&devr->mutex);
 
-	devr->p0 = mlx5_ib_alloc_pd(&dev->ib_dev, NULL, NULL);
-	if (IS_ERR(devr->p0)) {
-		ret = PTR_ERR(devr->p0);
-		goto error0;
-	}
-	devr->p0->device  = &dev->ib_dev;
+	devr->p0 = rdma_zalloc_drv_obj(ibdev, ib_pd);
+	if (!devr->p0)
+		return -ENOMEM;
+
+	devr->p0->device  = ibdev;
 	devr->p0->uobject = NULL;
 	atomic_set(&devr->p0->usecnt, 0);
 
+	ret = mlx5_ib_alloc_pd(devr->p0, NULL, NULL);
+	if (ret)
+		goto error0;
+
 	devr->c0 = mlx5_ib_create_cq(&dev->ib_dev, &cq_attr, NULL, NULL);
 	if (IS_ERR(devr->c0)) {
 		ret = PTR_ERR(devr->c0);
@@ -4794,6 +4789,7 @@ error2:
 error1:
 	mlx5_ib_dealloc_pd(devr->p0);
 error0:
+	kfree(devr->p0);
 	return ret;
 }
 
@@ -4809,6 +4805,7 @@ static void destroy_dev_resources(struct mlx5_ib_resources *devr)
 	mlx5_ib_dealloc_xrcd(devr->x1);
 	mlx5_ib_destroy_cq(devr->c0);
 	mlx5_ib_dealloc_pd(devr->p0);
+	kfree(devr->p0);
 
 	/* Make sure no change P_Key work items are still executing */
 	for (port = 0; port < dev->num_ports; ++port)
@@ -5938,6 +5935,7 @@ static const struct ib_device_ops mlx5_ib_dev_ops = {
 	.req_notify_cq = mlx5_ib_arm_cq,
 	.rereg_user_mr = mlx5_ib_rereg_user_mr,
 	.resize_cq = mlx5_ib_resize_cq,
+	INIT_RDMA_OBJ_SIZE(ib_pd, mlx5_ib_pd, ibpd),
 };
 
 static const struct ib_device_ops mlx5_ib_dev_flow_ipsec_ops = {
diff --git a/drivers/infiniband/hw/mthca/mthca_provider.c b/drivers/infiniband/hw/mthca/mthca_provider.c
index 1bb67562c8c8..2c754bc226f3 100644
--- a/drivers/infiniband/hw/mthca/mthca_provider.c
+++ b/drivers/infiniband/hw/mthca/mthca_provider.c
@@ -374,40 +374,30 @@ static int mthca_mmap_uar(struct ib_ucontext *context,
 	return 0;
 }
 
-static struct ib_pd *mthca_alloc_pd(struct ib_device *ibdev,
-				    struct ib_ucontext *context,
-				    struct ib_udata *udata)
+static int mthca_alloc_pd(struct ib_pd *ibpd, struct ib_ucontext *context,
+			  struct ib_udata *udata)
 {
-	struct mthca_pd *pd;
+	struct ib_device *ibdev = ibpd->device;
+	struct mthca_pd *pd = to_mpd(ibpd);
 	int err;
 
-	pd = kzalloc(sizeof(*pd), GFP_KERNEL);
-	if (!pd)
-		return ERR_PTR(-ENOMEM);
-
 	err = mthca_pd_alloc(to_mdev(ibdev), !context, pd);
-	if (err) {
-		kfree(pd);
-		return ERR_PTR(err);
-	}
+	if (err)
+		return err;
 
 	if (context) {
 		if (ib_copy_to_udata(udata, &pd->pd_num, sizeof (__u32))) {
 			mthca_pd_free(to_mdev(ibdev), pd);
-			kfree(pd);
-			return ERR_PTR(-EFAULT);
+			return -EFAULT;
 		}
 	}
 
-	return &pd->ibpd;
+	return 0;
 }
 
-static int mthca_dealloc_pd(struct ib_pd *pd)
+static void mthca_dealloc_pd(struct ib_pd *pd)
 {
 	mthca_pd_free(to_mdev(pd->device), to_mpd(pd));
-	kfree(pd);
-
-	return 0;
 }
 
 static struct ib_ah *mthca_ah_create(struct ib_pd *pd,
@@ -1228,6 +1218,7 @@ static const struct ib_device_ops mthca_dev_ops = {
 	.query_qp = mthca_query_qp,
 	.reg_user_mr = mthca_reg_user_mr,
 	.resize_cq = mthca_resize_cq,
+	INIT_RDMA_OBJ_SIZE(ib_pd, mthca_pd, ibpd),
 };
 
 static const struct ib_device_ops mthca_dev_arbel_srq_ops = {
diff --git a/drivers/infiniband/hw/nes/nes_verbs.c b/drivers/infiniband/hw/nes/nes_verbs.c
index 6eb991d40035..f18b28ae4bd9 100644
--- a/drivers/infiniband/hw/nes/nes_verbs.c
+++ b/drivers/infiniband/hw/nes/nes_verbs.c
@@ -658,10 +658,11 @@ static int nes_mmap(struct ib_ucontext *context, struct vm_area_struct *vma)
 /**
  * nes_alloc_pd
  */
-static struct ib_pd *nes_alloc_pd(struct ib_device *ibdev,
-		struct ib_ucontext *context, struct ib_udata *udata)
+static int nes_alloc_pd(struct ib_pd *pd, struct ib_ucontext *context,
+			struct ib_udata *udata)
 {
-	struct nes_pd *nespd;
+	struct ib_device *ibdev = pd->device;
+	struct nes_pd *nespd = to_nespd(pd);
 	struct nes_vnic *nesvnic = to_nesvnic(ibdev);
 	struct nes_device *nesdev = nesvnic->nesdev;
 	struct nes_adapter *nesadapter = nesdev->nesadapter;
@@ -676,15 +677,8 @@ static struct ib_pd *nes_alloc_pd(struct ib_device *ibdev,
 
 	err = nes_alloc_resource(nesadapter, nesadapter->allocated_pds,
 			nesadapter->max_pd, &pd_num, &nesadapter->next_pd, NES_RESOURCE_PD);
-	if (err) {
-		return ERR_PTR(err);
-	}
-
-	nespd = kzalloc(sizeof (struct nes_pd), GFP_KERNEL);
-	if (!nespd) {
-		nes_free_resource(nesadapter, nesadapter->allocated_pds, pd_num);
-		return ERR_PTR(-ENOMEM);
-	}
+	if (err)
+		return err;
 
 	nes_debug(NES_DBG_PD, "Allocating PD (%p) for ib device %s\n",
 			nespd, dev_name(&nesvnic->nesibdev->ibdev.dev));
@@ -700,16 +694,14 @@ static struct ib_pd *nes_alloc_pd(struct ib_device *ibdev,
 		if (nespd->mmap_db_index >= NES_MAX_USER_DB_REGIONS) {
 			nes_debug(NES_DBG_PD, "mmap_db_index > MAX\n");
 			nes_free_resource(nesadapter, nesadapter->allocated_pds, pd_num);
-			kfree(nespd);
-			return ERR_PTR(-ENOMEM);
+			return -ENOMEM;
 		}
 
 		uresp.pd_id = nespd->pd_id;
 		uresp.mmap_db_index = nespd->mmap_db_index;
 		if (ib_copy_to_udata(udata, &uresp, sizeof (struct nes_alloc_pd_resp))) {
 			nes_free_resource(nesadapter, nesadapter->allocated_pds, pd_num);
-			kfree(nespd);
-			return ERR_PTR(-EFAULT);
+			return -EFAULT;
 		}
 
 		set_bit(nespd->mmap_db_index, nesucontext->allocated_doorbells);
@@ -718,14 +710,14 @@ static struct ib_pd *nes_alloc_pd(struct ib_device *ibdev,
 	}
 
 	nes_debug(NES_DBG_PD, "PD%u structure located @%p.\n", nespd->pd_id, nespd);
-	return &nespd->ibpd;
+	return 0;
 }
 
 
 /**
  * nes_dealloc_pd
  */
-static int nes_dealloc_pd(struct ib_pd *ibpd)
+static void nes_dealloc_pd(struct ib_pd *ibpd)
 {
 	struct nes_ucontext *nesucontext;
 	struct nes_pd *nespd = to_nespd(ibpd);
@@ -748,9 +740,6 @@ static int nes_dealloc_pd(struct ib_pd *ibpd)
 			nespd->pd_id, nespd);
 	nes_free_resource(nesadapter, nesadapter->allocated_pds,
 			(nespd->pd_id-nesadapter->base_pd)>>(PAGE_SHIFT-12));
-	kfree(nespd);
-
-	return 0;
 }
 
 
@@ -3658,6 +3647,7 @@ static const struct ib_device_ops nes_dev_ops = {
 	.query_qp = nes_query_qp,
 	.reg_user_mr = nes_reg_user_mr,
 	.req_notify_cq = nes_req_notify_cq,
+	INIT_RDMA_OBJ_SIZE(ib_pd, nes_pd, ibpd),
 };
 
 /**
diff --git a/drivers/infiniband/hw/ocrdma/ocrdma_main.c b/drivers/infiniband/hw/ocrdma/ocrdma_main.c
index 88970a6bb555..0de83c92691f 100644
--- a/drivers/infiniband/hw/ocrdma/ocrdma_main.c
+++ b/drivers/infiniband/hw/ocrdma/ocrdma_main.c
@@ -179,6 +179,7 @@ static const struct ib_device_ops ocrdma_dev_ops = {
 	.reg_user_mr = ocrdma_reg_user_mr,
 	.req_notify_cq = ocrdma_arm_cq,
 	.resize_cq = ocrdma_resize_cq,
+	INIT_RDMA_OBJ_SIZE(ib_pd, ocrdma_pd, ibpd),
 };
 
 static const struct ib_device_ops ocrdma_dev_srq_ops = {
diff --git a/drivers/infiniband/hw/ocrdma/ocrdma_verbs.c b/drivers/infiniband/hw/ocrdma/ocrdma_verbs.c
index 2a62936bef4d..980ba97188ff 100644
--- a/drivers/infiniband/hw/ocrdma/ocrdma_verbs.c
+++ b/drivers/infiniband/hw/ocrdma/ocrdma_verbs.c
@@ -367,17 +367,12 @@ static int ocrdma_get_pd_num(struct ocrdma_dev *dev, struct ocrdma_pd *pd)
 	return status;
 }
 
-static struct ocrdma_pd *_ocrdma_alloc_pd(struct ocrdma_dev *dev,
-					  struct ocrdma_ucontext *uctx,
-					  struct ib_udata *udata)
+static int _ocrdma_alloc_pd(struct ocrdma_dev *dev, struct ocrdma_pd *pd,
+			    struct ocrdma_ucontext *uctx,
+			    struct ib_udata *udata)
 {
-	struct ocrdma_pd *pd = NULL;
 	int status;
 
-	pd = kzalloc(sizeof(*pd), GFP_KERNEL);
-	if (!pd)
-		return ERR_PTR(-ENOMEM);
-
 	if (udata && uctx && dev->attr.max_dpp_pds) {
 		pd->dpp_enabled =
 			ocrdma_get_asic_type(dev) == OCRDMA_ASIC_GEN_SKH_R;
@@ -386,15 +381,8 @@ static struct ocrdma_pd *_ocrdma_alloc_pd(struct ocrdma_dev *dev,
 					   dev->attr.wqe_size) : 0;
 	}
 
-	if (dev->pd_mgr->pd_prealloc_valid) {
-		status = ocrdma_get_pd_num(dev, pd);
-		if (status == 0) {
-			return pd;
-		} else {
-			kfree(pd);
-			return ERR_PTR(status);
-		}
-	}
+	if (dev->pd_mgr->pd_prealloc_valid)
+		return ocrdma_get_pd_num(dev, pd);
 
 retry:
 	status = ocrdma_mbx_alloc_pd(dev, pd);
@@ -403,13 +391,11 @@ retry:
 			pd->dpp_enabled = false;
 			pd->num_dpp_qp = 0;
 			goto retry;
-		} else {
-			kfree(pd);
-			return ERR_PTR(status);
 		}
+		return status;
 	}
 
-	return pd;
+	return 0;
 }
 
 static inline int is_ucontext_pd(struct ocrdma_ucontext *uctx,
@@ -418,30 +404,33 @@ static inline int is_ucontext_pd(struct ocrdma_ucontext *uctx,
 	return (uctx->cntxt_pd == pd);
 }
 
-static int _ocrdma_dealloc_pd(struct ocrdma_dev *dev,
+static void _ocrdma_dealloc_pd(struct ocrdma_dev *dev,
 			      struct ocrdma_pd *pd)
 {
-	int status;
-
 	if (dev->pd_mgr->pd_prealloc_valid)
-		status = ocrdma_put_pd_num(dev, pd->id, pd->dpp_enabled);
+		ocrdma_put_pd_num(dev, pd->id, pd->dpp_enabled);
 	else
-		status = ocrdma_mbx_dealloc_pd(dev, pd);
-
-	kfree(pd);
-	return status;
+		ocrdma_mbx_dealloc_pd(dev, pd);
 }
 
 static int ocrdma_alloc_ucontext_pd(struct ocrdma_dev *dev,
 				    struct ocrdma_ucontext *uctx,
 				    struct ib_udata *udata)
 {
-	int status = 0;
+	struct ib_device *ibdev = &dev->ibdev;
+	struct ib_pd *pd;
+	int status;
+
+	pd = rdma_zalloc_drv_obj(ibdev, ib_pd);
+	if (!pd)
+		return -ENOMEM;
+
+	pd->device  = ibdev;
+	uctx->cntxt_pd = get_ocrdma_pd(pd);
 
-	uctx->cntxt_pd = _ocrdma_alloc_pd(dev, uctx, udata);
-	if (IS_ERR(uctx->cntxt_pd)) {
-		status = PTR_ERR(uctx->cntxt_pd);
-		uctx->cntxt_pd = NULL;
+	status = _ocrdma_alloc_pd(dev, uctx->cntxt_pd, uctx, udata);
+	if (status) {
+		kfree(uctx->cntxt_pd);
 		goto err;
 	}
 
@@ -460,6 +449,7 @@ static int ocrdma_dealloc_ucontext_pd(struct ocrdma_ucontext *uctx)
 		pr_err("%s(%d) Freeing in use pdid=0x%x.\n",
 		       __func__, dev->id, pd->id);
 	}
+	kfree(uctx->cntxt_pd);
 	uctx->cntxt_pd = NULL;
 	(void)_ocrdma_dealloc_pd(dev, pd);
 	return 0;
@@ -537,6 +527,7 @@ struct ib_ucontext *ocrdma_alloc_ucontext(struct ib_device *ibdev,
 	return &ctx->ibucontext;
 
 cpy_err:
+	ocrdma_dealloc_ucontext_pd(ctx);
 pd_err:
 	ocrdma_del_mmap(ctx, ctx->ah_tbl.pa, ctx->ah_tbl.len);
 map_err:
@@ -658,10 +649,10 @@ dpp_map_err:
 	return status;
 }
 
-struct ib_pd *ocrdma_alloc_pd(struct ib_device *ibdev,
-			      struct ib_ucontext *context,
-			      struct ib_udata *udata)
+int ocrdma_alloc_pd(struct ib_pd *ibpd, struct ib_ucontext *context,
+		    struct ib_udata *udata)
 {
+	struct ib_device *ibdev = ibpd->device;
 	struct ocrdma_dev *dev = get_ocrdma_dev(ibdev);
 	struct ocrdma_pd *pd;
 	struct ocrdma_ucontext *uctx = NULL;
@@ -677,11 +668,10 @@ struct ib_pd *ocrdma_alloc_pd(struct ib_device *ibdev,
 		}
 	}
 
-	pd = _ocrdma_alloc_pd(dev, uctx, udata);
-	if (IS_ERR(pd)) {
-		status = PTR_ERR(pd);
+	pd = get_ocrdma_pd(ibpd);
+	status = _ocrdma_alloc_pd(dev, pd, uctx, udata);
+	if (status)
 		goto exit;
-	}
 
 pd_mapping:
 	if (udata && context) {
@@ -689,25 +679,22 @@ pd_mapping:
 		if (status)
 			goto err;
 	}
-	return &pd->ibpd;
+	return 0;
 
 err:
-	if (is_uctx_pd) {
+	if (is_uctx_pd)
 		ocrdma_release_ucontext_pd(uctx);
-	} else {
-		if (_ocrdma_dealloc_pd(dev, pd))
-			pr_err("%s: _ocrdma_dealloc_pd() failed\n", __func__);
-	}
+	else
+		_ocrdma_dealloc_pd(dev, pd);
 exit:
-	return ERR_PTR(status);
+	return status;
 }
 
-int ocrdma_dealloc_pd(struct ib_pd *ibpd)
+void ocrdma_dealloc_pd(struct ib_pd *ibpd)
 {
 	struct ocrdma_pd *pd = get_ocrdma_pd(ibpd);
 	struct ocrdma_dev *dev = get_ocrdma_dev(ibpd->device);
 	struct ocrdma_ucontext *uctx = NULL;
-	int status = 0;
 	u64 usr_db;
 
 	uctx = pd->uctx;
@@ -721,11 +708,10 @@ int ocrdma_dealloc_pd(struct ib_pd *ibpd)
 
 		if (is_ucontext_pd(uctx, pd)) {
 			ocrdma_release_ucontext_pd(uctx);
-			return status;
+			return;
 		}
 	}
-	status = _ocrdma_dealloc_pd(dev, pd);
-	return status;
+	_ocrdma_dealloc_pd(dev, pd);
 }
 
 static int ocrdma_alloc_lkey(struct ocrdma_dev *dev, struct ocrdma_mr *mr,
diff --git a/drivers/infiniband/hw/ocrdma/ocrdma_verbs.h b/drivers/infiniband/hw/ocrdma/ocrdma_verbs.h
index b69cfdce7970..1fd66721c930 100644
--- a/drivers/infiniband/hw/ocrdma/ocrdma_verbs.h
+++ b/drivers/infiniband/hw/ocrdma/ocrdma_verbs.h
@@ -70,9 +70,9 @@ int ocrdma_dealloc_ucontext(struct ib_ucontext *);
 
 int ocrdma_mmap(struct ib_ucontext *, struct vm_area_struct *vma);
 
-struct ib_pd *ocrdma_alloc_pd(struct ib_device *,
-			      struct ib_ucontext *, struct ib_udata *);
-int ocrdma_dealloc_pd(struct ib_pd *pd);
+int ocrdma_alloc_pd(struct ib_pd *pd, struct ib_ucontext *uctx,
+		    struct ib_udata *udata);
+void ocrdma_dealloc_pd(struct ib_pd *pd);
 
 struct ib_cq *ocrdma_create_cq(struct ib_device *ibdev,
 			       const struct ib_cq_init_attr *attr,
diff --git a/drivers/infiniband/hw/qedr/main.c b/drivers/infiniband/hw/qedr/main.c
index 878e9e23652b..44ce4989dcef 100644
--- a/drivers/infiniband/hw/qedr/main.c
+++ b/drivers/infiniband/hw/qedr/main.c
@@ -239,6 +239,7 @@ static const struct ib_device_ops qedr_dev_ops = {
 	.reg_user_mr = qedr_reg_user_mr,
 	.req_notify_cq = qedr_arm_cq,
 	.resize_cq = qedr_resize_cq,
+	INIT_RDMA_OBJ_SIZE(ib_pd, qedr_pd, ibpd),
 };
 
 static int qedr_register_device(struct qedr_dev *dev)
diff --git a/drivers/infiniband/hw/qedr/verbs.c b/drivers/infiniband/hw/qedr/verbs.c
index 989f08633fbe..a06d2258394a 100644
--- a/drivers/infiniband/hw/qedr/verbs.c
+++ b/drivers/infiniband/hw/qedr/verbs.c
@@ -450,11 +450,12 @@ int qedr_mmap(struct ib_ucontext *context, struct vm_area_struct *vma)
 				  vma->vm_page_prot);
 }
 
-struct ib_pd *qedr_alloc_pd(struct ib_device *ibdev,
-			    struct ib_ucontext *context, struct ib_udata *udata)
+int qedr_alloc_pd(struct ib_pd *ibpd, struct ib_ucontext *context,
+		  struct ib_udata *udata)
 {
+	struct ib_device *ibdev = ibpd->device;
 	struct qedr_dev *dev = get_qedr_dev(ibdev);
-	struct qedr_pd *pd;
+	struct qedr_pd *pd = get_qedr_pd(ibpd);
 	u16 pd_id;
 	int rc;
 
@@ -463,16 +464,12 @@ struct ib_pd *qedr_alloc_pd(struct ib_device *ibdev,
 
 	if (!dev->rdma_ctx) {
 		DP_ERR(dev, "invalid RDMA context\n");
-		return ERR_PTR(-EINVAL);
+		return -EINVAL;
 	}
 
-	pd = kzalloc(sizeof(*pd), GFP_KERNEL);
-	if (!pd)
-		return ERR_PTR(-ENOMEM);
-
 	rc = dev->ops->rdma_alloc_pd(dev->rdma_ctx, &pd_id);
 	if (rc)
-		goto err;
+		return rc;
 
 	pd->pd_id = pd_id;
 
@@ -485,36 +482,23 @@ struct ib_pd *qedr_alloc_pd(struct ib_device *ibdev,
 		if (rc) {
 			DP_ERR(dev, "copy error pd_id=0x%x.\n", pd_id);
 			dev->ops->rdma_dealloc_pd(dev->rdma_ctx, pd_id);
-			goto err;
+			return rc;
 		}
 
 		pd->uctx = get_qedr_ucontext(context);
 		pd->uctx->pd = pd;
 	}
 
-	return &pd->ibpd;
-
-err:
-	kfree(pd);
-	return ERR_PTR(rc);
+	return 0;
 }
 
-int qedr_dealloc_pd(struct ib_pd *ibpd)
+void qedr_dealloc_pd(struct ib_pd *ibpd)
 {
 	struct qedr_dev *dev = get_qedr_dev(ibpd->device);
 	struct qedr_pd *pd = get_qedr_pd(ibpd);
 
-	if (!pd) {
-		pr_err("Invalid PD received in dealloc_pd\n");
-		return -EINVAL;
-	}
-
 	DP_DEBUG(dev, QEDR_MSG_INIT, "Deallocating PD %d\n", pd->pd_id);
 	dev->ops->rdma_dealloc_pd(dev->rdma_ctx, pd->pd_id);
-
-	kfree(pd);
-
-	return 0;
 }
 
 static void qedr_free_pbl(struct qedr_dev *dev,
diff --git a/drivers/infiniband/hw/qedr/verbs.h b/drivers/infiniband/hw/qedr/verbs.h
index 1852b7012bf4..97a6ff3f9afb 100644
--- a/drivers/infiniband/hw/qedr/verbs.h
+++ b/drivers/infiniband/hw/qedr/verbs.h
@@ -47,9 +47,9 @@ struct ib_ucontext *qedr_alloc_ucontext(struct ib_device *, struct ib_udata *);
 int qedr_dealloc_ucontext(struct ib_ucontext *);
 
 int qedr_mmap(struct ib_ucontext *, struct vm_area_struct *vma);
-struct ib_pd *qedr_alloc_pd(struct ib_device *,
-			    struct ib_ucontext *, struct ib_udata *);
-int qedr_dealloc_pd(struct ib_pd *pd);
+int qedr_alloc_pd(struct ib_pd *pd, struct ib_ucontext *uctx,
+		  struct ib_udata *udata);
+void qedr_dealloc_pd(struct ib_pd *pd);
 
 struct ib_cq *qedr_create_cq(struct ib_device *ibdev,
 			     const struct ib_cq_init_attr *attr,
diff --git a/drivers/infiniband/hw/usnic/usnic_ib_main.c b/drivers/infiniband/hw/usnic/usnic_ib_main.c
index 1ec155823716..256ad2f236c8 100644
--- a/drivers/infiniband/hw/usnic/usnic_ib_main.c
+++ b/drivers/infiniband/hw/usnic/usnic_ib_main.c
@@ -352,6 +352,7 @@ static const struct ib_device_ops usnic_dev_ops = {
 	.query_port = usnic_ib_query_port,
 	.query_qp = usnic_ib_query_qp,
 	.reg_user_mr = usnic_ib_reg_mr,
+	INIT_RDMA_OBJ_SIZE(ib_pd, usnic_ib_pd, ibpd),
 };
 
 /* Start of PF discovery section */
diff --git a/drivers/infiniband/hw/usnic/usnic_ib_verbs.c b/drivers/infiniband/hw/usnic/usnic_ib_verbs.c
index 9dea18106247..0ced89b51448 100644
--- a/drivers/infiniband/hw/usnic/usnic_ib_verbs.c
+++ b/drivers/infiniband/hw/usnic/usnic_ib_verbs.c
@@ -456,37 +456,23 @@ int usnic_ib_query_pkey(struct ib_device *ibdev, u8 port, u16 index,
 	return 0;
 }
 
-struct ib_pd *usnic_ib_alloc_pd(struct ib_device *ibdev,
-					struct ib_ucontext *context,
-					struct ib_udata *udata)
+int usnic_ib_alloc_pd(struct ib_pd *ibpd, struct ib_ucontext *context,
+		      struct ib_udata *udata)
 {
-	struct usnic_ib_pd *pd;
+	struct usnic_ib_pd *pd = to_upd(ibpd);
 	void *umem_pd;
 
-	usnic_dbg("\n");
-
-	pd = kzalloc(sizeof(*pd), GFP_KERNEL);
-	if (!pd)
-		return ERR_PTR(-ENOMEM);
-
 	umem_pd = pd->umem_pd = usnic_uiom_alloc_pd();
 	if (IS_ERR_OR_NULL(umem_pd)) {
-		kfree(pd);
-		return ERR_PTR(umem_pd ? PTR_ERR(umem_pd) : -ENOMEM);
+		return umem_pd ? PTR_ERR(umem_pd) : -ENOMEM;
 	}
 
-	usnic_info("domain 0x%p allocated for context 0x%p and device %s\n",
-		   pd, context, dev_name(&ibdev->dev));
-	return &pd->ibpd;
+	return 0;
 }
 
-int usnic_ib_dealloc_pd(struct ib_pd *pd)
+void usnic_ib_dealloc_pd(struct ib_pd *pd)
 {
-	usnic_info("freeing domain 0x%p\n", pd);
-
 	usnic_uiom_dealloc_pd((to_upd(pd))->umem_pd);
-	kfree(pd);
-	return 0;
 }
 
 struct ib_qp *usnic_ib_create_qp(struct ib_pd *pd,
diff --git a/drivers/infiniband/hw/usnic/usnic_ib_verbs.h b/drivers/infiniband/hw/usnic/usnic_ib_verbs.h
index 99a6d81c2bcd..44a9d2f82bf5 100644
--- a/drivers/infiniband/hw/usnic/usnic_ib_verbs.h
+++ b/drivers/infiniband/hw/usnic/usnic_ib_verbs.h
@@ -51,10 +51,9 @@ int usnic_ib_query_gid(struct ib_device *ibdev, u8 port, int index,
 struct net_device *usnic_get_netdev(struct ib_device *device, u8 port_num);
 int usnic_ib_query_pkey(struct ib_device *ibdev, u8 port, u16 index,
 				u16 *pkey);
-struct ib_pd *usnic_ib_alloc_pd(struct ib_device *ibdev,
-				struct ib_ucontext *context,
-				struct ib_udata *udata);
-int usnic_ib_dealloc_pd(struct ib_pd *pd);
+int usnic_ib_alloc_pd(struct ib_pd *ibpd, struct ib_ucontext *context,
+		      struct ib_udata *udata);
+void usnic_ib_dealloc_pd(struct ib_pd *pd);
 struct ib_qp *usnic_ib_create_qp(struct ib_pd *pd,
 					struct ib_qp_init_attr *init_attr,
 					struct ib_udata *udata);
diff --git a/drivers/infiniband/hw/vmw_pvrdma/pvrdma_main.c b/drivers/infiniband/hw/vmw_pvrdma/pvrdma_main.c
index e582beaf9430..47e653d2495c 100644
--- a/drivers/infiniband/hw/vmw_pvrdma/pvrdma_main.c
+++ b/drivers/infiniband/hw/vmw_pvrdma/pvrdma_main.c
@@ -195,6 +195,7 @@ static const struct ib_device_ops pvrdma_dev_ops = {
 	.query_qp = pvrdma_query_qp,
 	.reg_user_mr = pvrdma_reg_user_mr,
 	.req_notify_cq = pvrdma_req_notify_cq,
+	INIT_RDMA_OBJ_SIZE(ib_pd, pvrdma_pd, ibpd),
 };
 
 static const struct ib_device_ops pvrdma_dev_srq_ops = {
diff --git a/drivers/infiniband/hw/vmw_pvrdma/pvrdma_verbs.c b/drivers/infiniband/hw/vmw_pvrdma/pvrdma_verbs.c
index fafb2add3b44..f44220f72e05 100644
--- a/drivers/infiniband/hw/vmw_pvrdma/pvrdma_verbs.c
+++ b/drivers/infiniband/hw/vmw_pvrdma/pvrdma_verbs.c
@@ -438,37 +438,29 @@ int pvrdma_mmap(struct ib_ucontext *ibcontext, struct vm_area_struct *vma)
 
 /**
  * pvrdma_alloc_pd - allocate protection domain
- * @ibdev: the IB device
+ * @ibpd: PD pointer
  * @context: user context
  * @udata: user data
  *
  * @return: the ib_pd protection domain pointer on success, otherwise errno.
  */
-struct ib_pd *pvrdma_alloc_pd(struct ib_device *ibdev,
-			      struct ib_ucontext *context,
-			      struct ib_udata *udata)
+int pvrdma_alloc_pd(struct ib_pd *ibpd, struct ib_ucontext *context,
+		    struct ib_udata *udata)
 {
-	struct pvrdma_pd *pd;
+	struct ib_device *ibdev = ibpd->device;
+	struct pvrdma_pd *pd = to_vpd(ibpd);
 	struct pvrdma_dev *dev = to_vdev(ibdev);
-	union pvrdma_cmd_req req;
-	union pvrdma_cmd_resp rsp;
+	union pvrdma_cmd_req req = {};
+	union pvrdma_cmd_resp rsp = {};
 	struct pvrdma_cmd_create_pd *cmd = &req.create_pd;
 	struct pvrdma_cmd_create_pd_resp *resp = &rsp.create_pd_resp;
 	struct pvrdma_alloc_pd_resp pd_resp = {0};
 	int ret;
-	void *ptr;
 
 	/* Check allowed max pds */
 	if (!atomic_add_unless(&dev->num_pds, 1, dev->dsr->caps.max_pd))
-		return ERR_PTR(-ENOMEM);
+		return -ENOMEM;
 
-	pd = kzalloc(sizeof(*pd), GFP_KERNEL);
-	if (!pd) {
-		ptr = ERR_PTR(-ENOMEM);
-		goto err;
-	}
-
-	memset(cmd, 0, sizeof(*cmd));
 	cmd->hdr.cmd = PVRDMA_CMD_CREATE_PD;
 	cmd->ctx_handle = (context) ? to_vucontext(context)->ctx_handle : 0;
 	ret = pvrdma_cmd_post(dev, &req, &rsp, PVRDMA_CMD_CREATE_PD_RESP);
@@ -476,8 +468,7 @@ struct ib_pd *pvrdma_alloc_pd(struct ib_device *ibdev,
 		dev_warn(&dev->pdev->dev,
 			 "failed to allocate protection domain, error: %d\n",
 			 ret);
-		ptr = ERR_PTR(ret);
-		goto freepd;
+		goto err;
 	}
 
 	pd->privileged = !context;
@@ -490,18 +481,16 @@ struct ib_pd *pvrdma_alloc_pd(struct ib_device *ibdev,
 			dev_warn(&dev->pdev->dev,
 				 "failed to copy back protection domain\n");
 			pvrdma_dealloc_pd(&pd->ibpd);
-			return ERR_PTR(-EFAULT);
+			return -EFAULT;
 		}
 	}
 
 	/* u32 pd handle */
-	return &pd->ibpd;
+	return 0;
 
-freepd:
-	kfree(pd);
 err:
 	atomic_dec(&dev->num_pds);
-	return ptr;
+	return ret;
 }
 
 /**
@@ -510,14 +499,13 @@ err:
  *
  * @return: 0 on success, otherwise errno.
  */
-int pvrdma_dealloc_pd(struct ib_pd *pd)
+void pvrdma_dealloc_pd(struct ib_pd *pd)
 {
 	struct pvrdma_dev *dev = to_vdev(pd->device);
-	union pvrdma_cmd_req req;
+	union pvrdma_cmd_req req = {};
 	struct pvrdma_cmd_destroy_pd *cmd = &req.destroy_pd;
 	int ret;
 
-	memset(cmd, 0, sizeof(*cmd));
 	cmd->hdr.cmd = PVRDMA_CMD_DESTROY_PD;
 	cmd->pd_handle = to_vpd(pd)->pd_handle;
 
@@ -527,10 +515,7 @@ int pvrdma_dealloc_pd(struct ib_pd *pd)
 			 "could not dealloc protection domain, error: %d\n",
 			 ret);
 
-	kfree(to_vpd(pd));
 	atomic_dec(&dev->num_pds);
-
-	return 0;
 }
 
 /**
diff --git a/drivers/infiniband/hw/vmw_pvrdma/pvrdma_verbs.h b/drivers/infiniband/hw/vmw_pvrdma/pvrdma_verbs.h
index f7f758d60110..ed91baad1ffa 100644
--- a/drivers/infiniband/hw/vmw_pvrdma/pvrdma_verbs.h
+++ b/drivers/infiniband/hw/vmw_pvrdma/pvrdma_verbs.h
@@ -399,10 +399,9 @@ int pvrdma_mmap(struct ib_ucontext *context, struct vm_area_struct *vma);
 struct ib_ucontext *pvrdma_alloc_ucontext(struct ib_device *ibdev,
 					  struct ib_udata *udata);
 int pvrdma_dealloc_ucontext(struct ib_ucontext *context);
-struct ib_pd *pvrdma_alloc_pd(struct ib_device *ibdev,
-			      struct ib_ucontext *context,
-			      struct ib_udata *udata);
-int pvrdma_dealloc_pd(struct ib_pd *ibpd);
+int pvrdma_alloc_pd(struct ib_pd *pd, struct ib_ucontext *context,
+		    struct ib_udata *udata);
+void pvrdma_dealloc_pd(struct ib_pd *ibpd);
 struct ib_mr *pvrdma_get_dma_mr(struct ib_pd *pd, int acc);
 struct ib_mr *pvrdma_reg_user_mr(struct ib_pd *pd, u64 start, u64 length,
 				 u64 virt_addr, int access_flags,
diff --git a/drivers/infiniband/sw/rdmavt/pd.c b/drivers/infiniband/sw/rdmavt/pd.c
index dcc1870b8d23..6033054b22fa 100644
--- a/drivers/infiniband/sw/rdmavt/pd.c
+++ b/drivers/infiniband/sw/rdmavt/pd.c
@@ -50,7 +50,7 @@
 
 /**
  * rvt_alloc_pd - allocate a protection domain
- * @ibdev: ib device
+ * @ibpd: PD
  * @context: optional user context
  * @udata: optional user data
  *
@@ -58,19 +58,14 @@
  *
  * Return: 0 on success
  */
-struct ib_pd *rvt_alloc_pd(struct ib_device *ibdev,
-			   struct ib_ucontext *context,
-			   struct ib_udata *udata)
+int rvt_alloc_pd(struct ib_pd *ibpd, struct ib_ucontext *context,
+		 struct ib_udata *udata)
 {
+	struct ib_device *ibdev = ibpd->device;
 	struct rvt_dev_info *dev = ib_to_rvt(ibdev);
-	struct rvt_pd *pd;
-	struct ib_pd *ret;
+	struct rvt_pd *pd = ibpd_to_rvtpd(ibpd);
+	int ret = 0;
 
-	pd = kzalloc(sizeof(*pd), GFP_KERNEL);
-	if (!pd) {
-		ret = ERR_PTR(-ENOMEM);
-		goto bail;
-	}
 	/*
 	 * While we could continue allocating protecetion domains, being
 	 * constrained only by system resources. The IBTA spec defines that
@@ -81,8 +76,7 @@ struct ib_pd *rvt_alloc_pd(struct ib_device *ibdev,
 	spin_lock(&dev->n_pds_lock);
 	if (dev->n_pds_allocated == dev->dparms.props.max_pd) {
 		spin_unlock(&dev->n_pds_lock);
-		kfree(pd);
-		ret = ERR_PTR(-ENOMEM);
+		ret = -ENOMEM;
 		goto bail;
 	}
 
@@ -92,8 +86,6 @@ struct ib_pd *rvt_alloc_pd(struct ib_device *ibdev,
 	/* ib_alloc_pd() will initialize pd->ibpd. */
 	pd->user = !!udata;
 
-	ret = &pd->ibpd;
-
 bail:
 	return ret;
 }
@@ -104,16 +96,11 @@ bail:
  *
  * Return: always 0
  */
-int rvt_dealloc_pd(struct ib_pd *ibpd)
+void rvt_dealloc_pd(struct ib_pd *ibpd)
 {
-	struct rvt_pd *pd = ibpd_to_rvtpd(ibpd);
 	struct rvt_dev_info *dev = ib_to_rvt(ibpd->device);
 
 	spin_lock(&dev->n_pds_lock);
 	dev->n_pds_allocated--;
 	spin_unlock(&dev->n_pds_lock);
-
-	kfree(pd);
-
-	return 0;
 }
diff --git a/drivers/infiniband/sw/rdmavt/pd.h b/drivers/infiniband/sw/rdmavt/pd.h
index 1892ca4a9746..7a887e4a45e7 100644
--- a/drivers/infiniband/sw/rdmavt/pd.h
+++ b/drivers/infiniband/sw/rdmavt/pd.h
@@ -50,9 +50,8 @@
 
 #include <rdma/rdma_vt.h>
 
-struct ib_pd *rvt_alloc_pd(struct ib_device *ibdev,
-			   struct ib_ucontext *context,
-			   struct ib_udata *udata);
-int rvt_dealloc_pd(struct ib_pd *ibpd);
+int rvt_alloc_pd(struct ib_pd *pd, struct ib_ucontext *context,
+		 struct ib_udata *udata);
+void rvt_dealloc_pd(struct ib_pd *ibpd);
 
 #endif          /* DEF_RDMAVTPD_H */
diff --git a/drivers/infiniband/sw/rdmavt/vt.c b/drivers/infiniband/sw/rdmavt/vt.c
index b3f0c5578925..a19832c73d5a 100644
--- a/drivers/infiniband/sw/rdmavt/vt.c
+++ b/drivers/infiniband/sw/rdmavt/vt.c
@@ -436,6 +436,7 @@ static const struct ib_device_ops rvt_dev_ops = {
 	.req_notify_cq = rvt_req_notify_cq,
 	.resize_cq = rvt_resize_cq,
 	.unmap_fmr = rvt_unmap_fmr,
+	INIT_RDMA_OBJ_SIZE(ib_pd, rvt_pd, ibpd),
 };
 
 static noinline int check_support(struct rvt_dev_info *rdi, int verb)
diff --git a/drivers/infiniband/sw/rxe/rxe_pool.c b/drivers/infiniband/sw/rxe/rxe_pool.c
index b5c91df22047..cd3f14629ba8 100644
--- a/drivers/infiniband/sw/rxe/rxe_pool.c
+++ b/drivers/infiniband/sw/rxe/rxe_pool.c
@@ -46,6 +46,7 @@ struct rxe_type_info rxe_type_info[RXE_NUM_TYPES] = {
 	[RXE_TYPE_PD] = {
 		.name		= "rxe-pd",
 		.size		= sizeof(struct rxe_pd),
+		.flags		= RXE_POOL_NO_ALLOC,
 	},
 	[RXE_TYPE_AH] = {
 		.name		= "rxe-ah",
@@ -119,8 +120,10 @@ static void rxe_cache_clean(size_t cnt)
 
 	for (i = 0; i < cnt; i++) {
 		type = &rxe_type_info[i];
-		kmem_cache_destroy(type->cache);
-		type->cache = NULL;
+		if (!(type->flags & RXE_POOL_NO_ALLOC)) {
+			kmem_cache_destroy(type->cache);
+			type->cache = NULL;
+		}
 	}
 }
 
@@ -134,14 +137,17 @@ int rxe_cache_init(void)
 	for (i = 0; i < RXE_NUM_TYPES; i++) {
 		type = &rxe_type_info[i];
 		size = ALIGN(type->size, RXE_POOL_ALIGN);
-		type->cache = kmem_cache_create(type->name, size,
-				RXE_POOL_ALIGN,
-				RXE_POOL_CACHE_FLAGS, NULL);
-		if (!type->cache) {
-			pr_err("Unable to init kmem cache for %s\n",
-			       type->name);
-			err = -ENOMEM;
-			goto err1;
+		if (!(type->flags & RXE_POOL_NO_ALLOC)) {
+			type->cache =
+				kmem_cache_create(type->name, size,
+						  RXE_POOL_ALIGN,
+						  RXE_POOL_CACHE_FLAGS, NULL);
+			if (!type->cache) {
+				pr_err("Unable to init kmem cache for %s\n",
+				       type->name);
+				err = -ENOMEM;
+				goto err1;
+			}
 		}
 	}
 
@@ -415,6 +421,37 @@ out_put_pool:
 	return NULL;
 }
 
+int rxe_add_to_pool(struct rxe_pool *pool, struct rxe_pool_entry *elem)
+{
+	unsigned long flags;
+
+	might_sleep_if(!(pool->flags & RXE_POOL_ATOMIC));
+
+	read_lock_irqsave(&pool->pool_lock, flags);
+	if (pool->state != RXE_POOL_STATE_VALID) {
+		read_unlock_irqrestore(&pool->pool_lock, flags);
+		return -EINVAL;
+	}
+	kref_get(&pool->ref_cnt);
+	read_unlock_irqrestore(&pool->pool_lock, flags);
+
+	kref_get(&pool->rxe->ref_cnt);
+
+	if (atomic_inc_return(&pool->num_elem) > pool->max_elem)
+		goto out_put_pool;
+
+	elem->pool = pool;
+	kref_init(&elem->ref_cnt);
+
+	return 0;
+
+out_put_pool:
+	atomic_dec(&pool->num_elem);
+	rxe_dev_put(pool->rxe);
+	rxe_pool_put(pool);
+	return -EINVAL;
+}
+
 void rxe_elem_release(struct kref *kref)
 {
 	struct rxe_pool_entry *elem =
@@ -424,7 +461,8 @@ void rxe_elem_release(struct kref *kref)
 	if (pool->cleanup)
 		pool->cleanup(elem);
 
-	kmem_cache_free(pool_cache(pool), elem);
+	if (!(pool->flags & RXE_POOL_NO_ALLOC))
+		kmem_cache_free(pool_cache(pool), elem);
 	atomic_dec(&pool->num_elem);
 	rxe_dev_put(pool->rxe);
 	rxe_pool_put(pool);
diff --git a/drivers/infiniband/sw/rxe/rxe_pool.h b/drivers/infiniband/sw/rxe/rxe_pool.h
index 72968c29e01f..2f2cff1cbe43 100644
--- a/drivers/infiniband/sw/rxe/rxe_pool.h
+++ b/drivers/infiniband/sw/rxe/rxe_pool.h
@@ -41,6 +41,7 @@ enum rxe_pool_flags {
 	RXE_POOL_ATOMIC		= BIT(0),
 	RXE_POOL_INDEX		= BIT(1),
 	RXE_POOL_KEY		= BIT(2),
+	RXE_POOL_NO_ALLOC	= BIT(4),
 };
 
 enum rxe_elem_type {
@@ -131,6 +132,9 @@ void rxe_pool_cleanup(struct rxe_pool *pool);
 /* allocate an object from pool */
 void *rxe_alloc(struct rxe_pool *pool);
 
+/* connect already allocated object to pool */
+int rxe_add_to_pool(struct rxe_pool *pool, struct rxe_pool_entry *elem);
+
 /* assign an index to an indexed object and insert object into
  *  pool's rb tree
  */
diff --git a/drivers/infiniband/sw/rxe/rxe_verbs.c b/drivers/infiniband/sw/rxe/rxe_verbs.c
index cc5a05124ece..051c3930e808 100644
--- a/drivers/infiniband/sw/rxe/rxe_verbs.c
+++ b/drivers/infiniband/sw/rxe/rxe_verbs.c
@@ -191,23 +191,20 @@ static int rxe_port_immutable(struct ib_device *dev, u8 port_num,
 	return 0;
 }
 
-static struct ib_pd *rxe_alloc_pd(struct ib_device *dev,
-				  struct ib_ucontext *context,
-				  struct ib_udata *udata)
+static int rxe_alloc_pd(struct ib_pd *ibpd, struct ib_ucontext *context,
+			struct ib_udata *udata)
 {
-	struct rxe_dev *rxe = to_rdev(dev);
-	struct rxe_pd *pd;
+	struct rxe_dev *rxe = to_rdev(ibpd->device);
+	struct rxe_pd *pd = to_rpd(ibpd);
 
-	pd = rxe_alloc(&rxe->pd_pool);
-	return pd ? &pd->ibpd : ERR_PTR(-ENOMEM);
+	return rxe_add_to_pool(&rxe->pd_pool, &pd->pelem);
 }
 
-static int rxe_dealloc_pd(struct ib_pd *ibpd)
+static void rxe_dealloc_pd(struct ib_pd *ibpd)
 {
 	struct rxe_pd *pd = to_rpd(ibpd);
 
 	rxe_drop_ref(pd);
-	return 0;
 }
 
 static struct ib_ah *rxe_create_ah(struct ib_pd *ibpd,
@@ -1183,6 +1180,7 @@ static const struct ib_device_ops rxe_dev_ops = {
 	.reg_user_mr = rxe_reg_user_mr,
 	.req_notify_cq = rxe_req_notify_cq,
 	.resize_cq = rxe_resize_cq,
+	INIT_RDMA_OBJ_SIZE(ib_pd, rxe_pd, ibpd),
 };
 
 int rxe_register_device(struct rxe_dev *rxe)
diff --git a/drivers/infiniband/sw/rxe/rxe_verbs.h b/drivers/infiniband/sw/rxe/rxe_verbs.h
index 74e04801d34d..70839d3f55d9 100644
--- a/drivers/infiniband/sw/rxe/rxe_verbs.h
+++ b/drivers/infiniband/sw/rxe/rxe_verbs.h
@@ -66,8 +66,8 @@ struct rxe_ucontext {
 };
 
 struct rxe_pd {
+	struct ib_pd            ibpd;
 	struct rxe_pool_entry	pelem;
-	struct ib_pd		ibpd;
 };
 
 struct rxe_ah {
diff --git a/include/rdma/ib_verbs.h b/include/rdma/ib_verbs.h
index e29eae4aec84..854d7816787c 100644
--- a/include/rdma/ib_verbs.h
+++ b/include/rdma/ib_verbs.h
@@ -2385,10 +2385,9 @@ struct ib_device_ops {
 	int (*dealloc_ucontext)(struct ib_ucontext *context);
 	int (*mmap)(struct ib_ucontext *context, struct vm_area_struct *vma);
 	void (*disassociate_ucontext)(struct ib_ucontext *ibcontext);
-	struct ib_pd *(*alloc_pd)(struct ib_device *device,
-				  struct ib_ucontext *context,
-				  struct ib_udata *udata);
-	int (*dealloc_pd)(struct ib_pd *pd);
+	int (*alloc_pd)(struct ib_pd *pd, struct ib_ucontext *context,
+			struct ib_udata *udata);
+	void (*dealloc_pd)(struct ib_pd *pd);
 	struct ib_ah *(*create_ah)(struct ib_pd *pd,
 				   struct rdma_ah_attr *ah_attr, u32 flags,
 				   struct ib_udata *udata);
@@ -2530,6 +2529,8 @@ struct ib_device_ops {
 	 */
 	int (*fill_res_entry)(struct sk_buff *msg,
 			      struct rdma_restrack_entry *entry);
+
+	DECLARE_RDMA_OBJ_SIZE(ib_pd);
 };
 
 struct ib_device {
-- 
cgit v1.2.3


From f0666f1f22b5748e7caf87481a0869689f9f513d Mon Sep 17 00:00:00 2001
From: Bodong Wang <bodong@mellanox.com>
Date: Tue, 12 Feb 2019 22:55:34 -0800
Subject: IB/mlx5: Use unified register/load function for uplink and VF vports

IB driver maintains different registration and load function calls
for uplink and VF vports. This is not necessary as they only differ
with each other on their profiles.

This patch doesn't change any functionality.

Signed-off-by: Bodong Wang <bodong@mellanox.com>
Reviewed-by: Mark Bloch <markb@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
---
 drivers/infiniband/hw/mlx5/ib_rep.c | 73 +++++++++----------------------------
 drivers/infiniband/hw/mlx5/ib_rep.h | 10 +++--
 drivers/infiniband/hw/mlx5/main.c   | 26 ++++++-------
 3 files changed, 37 insertions(+), 72 deletions(-)

(limited to 'drivers/infiniband/hw/mlx5/main.c')

diff --git a/drivers/infiniband/hw/mlx5/ib_rep.c b/drivers/infiniband/hw/mlx5/ib_rep.c
index 46a9ddc8ca56..3fb22967c098 100644
--- a/drivers/infiniband/hw/mlx5/ib_rep.c
+++ b/drivers/infiniband/hw/mlx5/ib_rep.c
@@ -6,7 +6,7 @@
 #include "ib_rep.h"
 #include "srq.h"
 
-static const struct mlx5_ib_profile rep_profile = {
+static const struct mlx5_ib_profile vf_rep_profile = {
 	STAGE_CREATE(MLX5_IB_STAGE_INIT,
 		     mlx5_ib_stage_init_init,
 		     mlx5_ib_stage_init_cleanup),
@@ -45,31 +45,18 @@ static const struct mlx5_ib_profile rep_profile = {
 		     NULL),
 };
 
-static int
-mlx5_ib_nic_rep_load(struct mlx5_core_dev *dev, struct mlx5_eswitch_rep *rep)
-{
-	struct mlx5_ib_dev *ibdev;
-
-	ibdev = mlx5_ib_rep_to_dev(rep);
-	if (!__mlx5_ib_add(ibdev, ibdev->profile))
-		return -EINVAL;
-	return 0;
-}
-
-static void
-mlx5_ib_nic_rep_unload(struct mlx5_eswitch_rep *rep)
-{
-	struct mlx5_ib_dev *ibdev;
-
-	ibdev = mlx5_ib_rep_to_dev(rep);
-	__mlx5_ib_remove(ibdev, ibdev->profile, MLX5_IB_STAGE_MAX);
-}
-
 static int
 mlx5_ib_vport_rep_load(struct mlx5_core_dev *dev, struct mlx5_eswitch_rep *rep)
 {
+#define FDB_UPLINK_VPORT 0xffff
+	const struct mlx5_ib_profile *profile;
 	struct mlx5_ib_dev *ibdev;
 
+	if (rep->vport == FDB_UPLINK_VPORT)
+		profile = &uplink_rep_profile;
+	else
+		profile = &vf_rep_profile;
+
 	ibdev = (struct mlx5_ib_dev *)ib_alloc_device(sizeof(*ibdev));
 	if (!ibdev)
 		return -ENOMEM;
@@ -78,7 +65,7 @@ mlx5_ib_vport_rep_load(struct mlx5_core_dev *dev, struct mlx5_eswitch_rep *rep)
 	ibdev->mdev = dev;
 	ibdev->num_ports = max(MLX5_CAP_GEN(dev, num_ports),
 			       MLX5_CAP_GEN(dev, num_vhca_ports));
-	if (!__mlx5_ib_add(ibdev, &rep_profile))
+	if (!__mlx5_ib_add(ibdev, profile))
 		return -EINVAL;
 
 	rep->rep_if[REP_IB].priv = ibdev;
@@ -105,15 +92,14 @@ static void *mlx5_ib_vport_get_proto_dev(struct mlx5_eswitch_rep *rep)
 	return mlx5_ib_rep_to_dev(rep);
 }
 
-static void mlx5_ib_rep_register_vf_vports(struct mlx5_ib_dev *dev)
+void mlx5_ib_register_vport_reps(struct mlx5_core_dev *mdev)
 {
-	struct mlx5_eswitch *esw   = dev->mdev->priv.eswitch;
-	int total_vfs = MLX5_TOTAL_VPORTS(dev->mdev);
+	struct mlx5_eswitch *esw = mdev->priv.eswitch;
+	int total_vports = MLX5_TOTAL_VPORTS(mdev);
+	struct mlx5_eswitch_rep_if rep_if = {};
 	int vport;
 
-	for (vport = 1; vport < total_vfs; vport++) {
-		struct mlx5_eswitch_rep_if rep_if = {};
-
+	for (vport = 0; vport < total_vports; vport++) {
 		rep_if.load = mlx5_ib_vport_rep_load;
 		rep_if.unload = mlx5_ib_vport_rep_unload;
 		rep_if.get_proto_dev = mlx5_ib_vport_get_proto_dev;
@@ -121,39 +107,16 @@ static void mlx5_ib_rep_register_vf_vports(struct mlx5_ib_dev *dev)
 	}
 }
 
-static void mlx5_ib_rep_unregister_vf_vports(struct mlx5_ib_dev *dev)
+void mlx5_ib_unregister_vport_reps(struct mlx5_core_dev *mdev)
 {
-	struct mlx5_eswitch *esw   = dev->mdev->priv.eswitch;
-	int total_vfs = MLX5_TOTAL_VPORTS(dev->mdev);
+	struct mlx5_eswitch *esw = mdev->priv.eswitch;
+	int total_vports = MLX5_TOTAL_VPORTS(mdev);
 	int vport;
 
-	for (vport = 1; vport < total_vfs; vport++)
+	for (vport = total_vports - 1; vport >= 0; vport--)
 		mlx5_eswitch_unregister_vport_rep(esw, vport, REP_IB);
 }
 
-void mlx5_ib_register_vport_reps(struct mlx5_ib_dev *dev)
-{
-	struct mlx5_eswitch *esw = dev->mdev->priv.eswitch;
-	struct mlx5_eswitch_rep_if rep_if = {};
-
-	rep_if.load = mlx5_ib_nic_rep_load;
-	rep_if.unload = mlx5_ib_nic_rep_unload;
-	rep_if.get_proto_dev = mlx5_ib_vport_get_proto_dev;
-	rep_if.priv = dev;
-
-	mlx5_eswitch_register_vport_rep(esw, 0, &rep_if, REP_IB);
-
-	mlx5_ib_rep_register_vf_vports(dev);
-}
-
-void mlx5_ib_unregister_vport_reps(struct mlx5_ib_dev *dev)
-{
-	struct mlx5_eswitch *esw   = dev->mdev->priv.eswitch;
-
-	mlx5_ib_rep_unregister_vf_vports(dev); /* VFs vports */
-	mlx5_eswitch_unregister_vport_rep(esw, 0, REP_IB); /* UPLINK PF*/
-}
-
 u8 mlx5_ib_eswitch_mode(struct mlx5_eswitch *esw)
 {
 	return mlx5_eswitch_mode(esw);
diff --git a/drivers/infiniband/hw/mlx5/ib_rep.h b/drivers/infiniband/hw/mlx5/ib_rep.h
index 2ba73636a2fb..798d41e61fb4 100644
--- a/drivers/infiniband/hw/mlx5/ib_rep.h
+++ b/drivers/infiniband/hw/mlx5/ib_rep.h
@@ -10,14 +10,16 @@
 #include "mlx5_ib.h"
 
 #ifdef CONFIG_MLX5_ESWITCH
+extern const struct mlx5_ib_profile uplink_rep_profile;
+
 u8 mlx5_ib_eswitch_mode(struct mlx5_eswitch *esw);
 struct mlx5_ib_dev *mlx5_ib_get_rep_ibdev(struct mlx5_eswitch *esw,
 					  int vport_index);
 struct mlx5_ib_dev *mlx5_ib_get_uplink_ibdev(struct mlx5_eswitch *esw);
 struct mlx5_eswitch_rep *mlx5_ib_vport_rep(struct mlx5_eswitch *esw,
 					   int vport_index);
-void mlx5_ib_register_vport_reps(struct mlx5_ib_dev *dev);
-void mlx5_ib_unregister_vport_reps(struct mlx5_ib_dev *dev);
+void mlx5_ib_register_vport_reps(struct mlx5_core_dev *mdev);
+void mlx5_ib_unregister_vport_reps(struct mlx5_core_dev *mdev);
 int create_flow_rule_vport_sq(struct mlx5_ib_dev *dev,
 			      struct mlx5_ib_sq *sq);
 struct net_device *mlx5_ib_get_rep_netdev(struct mlx5_eswitch *esw,
@@ -48,8 +50,8 @@ struct mlx5_eswitch_rep *mlx5_ib_vport_rep(struct mlx5_eswitch *esw,
 	return NULL;
 }
 
-static inline void mlx5_ib_register_vport_reps(struct mlx5_ib_dev *dev) {}
-static inline void mlx5_ib_unregister_vport_reps(struct mlx5_ib_dev *dev) {}
+static inline void mlx5_ib_register_vport_reps(struct mlx5_core_dev *mdev) {}
+static inline void mlx5_ib_unregister_vport_reps(struct mlx5_core_dev *mdev) {}
 static inline int create_flow_rule_vport_sq(struct mlx5_ib_dev *dev,
 					    struct mlx5_ib_sq *sq)
 {
diff --git a/drivers/infiniband/hw/mlx5/main.c b/drivers/infiniband/hw/mlx5/main.c
index 94fe253d4956..87ce62e44898 100644
--- a/drivers/infiniband/hw/mlx5/main.c
+++ b/drivers/infiniband/hw/mlx5/main.c
@@ -6386,7 +6386,7 @@ static const struct mlx5_ib_profile pf_profile = {
 		     mlx5_ib_stage_delay_drop_cleanup),
 };
 
-static const struct mlx5_ib_profile nic_rep_profile = {
+const struct mlx5_ib_profile uplink_rep_profile = {
 	STAGE_CREATE(MLX5_IB_STAGE_INIT,
 		     mlx5_ib_stage_init_init,
 		     mlx5_ib_stage_init_cleanup),
@@ -6479,6 +6479,12 @@ static void *mlx5_ib_add(struct mlx5_core_dev *mdev)
 
 	printk_once(KERN_INFO "%s", mlx5_version);
 
+	if (MLX5_ESWITCH_MANAGER(mdev) &&
+	    mlx5_ib_eswitch_mode(mdev->priv.eswitch) == SRIOV_OFFLOADS) {
+		mlx5_ib_register_vport_reps(mdev);
+		return mdev;
+	}
+
 	port_type_cap = MLX5_CAP_GEN(mdev, port_type);
 	ll = mlx5_port_type_cap_to_rdma_ll(port_type_cap);
 
@@ -6493,14 +6499,6 @@ static void *mlx5_ib_add(struct mlx5_core_dev *mdev)
 	dev->num_ports = max(MLX5_CAP_GEN(mdev, num_ports),
 			     MLX5_CAP_GEN(mdev, num_vhca_ports));
 
-	if (MLX5_ESWITCH_MANAGER(mdev) &&
-	    mlx5_ib_eswitch_mode(mdev->priv.eswitch) == SRIOV_OFFLOADS) {
-		dev->rep = mlx5_ib_vport_rep(mdev->priv.eswitch, 0);
-		dev->profile = &nic_rep_profile;
-		mlx5_ib_register_vport_reps(dev);
-		return dev;
-	}
-
 	return __mlx5_ib_add(dev, &pf_profile);
 }
 
@@ -6509,6 +6507,11 @@ static void mlx5_ib_remove(struct mlx5_core_dev *mdev, void *context)
 	struct mlx5_ib_multiport_info *mpi;
 	struct mlx5_ib_dev *dev;
 
+	if (MLX5_ESWITCH_MANAGER(mdev) && context == mdev) {
+		mlx5_ib_unregister_vport_reps(mdev);
+		return;
+	}
+
 	if (mlx5_core_is_mp_slave(mdev)) {
 		mpi = context;
 		mutex_lock(&mlx5_ib_multiport_mutex);
@@ -6520,10 +6523,7 @@ static void mlx5_ib_remove(struct mlx5_core_dev *mdev, void *context)
 	}
 
 	dev = context;
-	if (dev->profile == &nic_rep_profile)
-		mlx5_ib_unregister_vport_reps(dev);
-	else
-		__mlx5_ib_remove(dev, dev->profile, MLX5_IB_STAGE_MAX);
+	__mlx5_ib_remove(dev, dev->profile, MLX5_IB_STAGE_MAX);
 
 	ib_dealloc_device((struct ib_device *)dev);
 }
-- 
cgit v1.2.3


From bc4e12ffefdd886057eabe38135515690d0756a6 Mon Sep 17 00:00:00 2001
From: Aya Levin <ayal@mellanox.com>
Date: Tue, 12 Feb 2019 22:55:43 -0800
Subject: net/mlx5: Refactor queries to speed fields in Port Type and Speed
 register

This patch fascicles queries to speed related fields in Port Type and
Speed register (PTYS) into a single API. I addition, this patch
refactors functions which serves only Ethernet driver: remove the
protocol type as an input parameter, move code from 'core' directory
into 'en' directory and add 'eth' prefix to the function's name. The
patch also encapsulates functions that are not used outside the Ethernet
driver removes redundant include files.

Signed-off-by: Aya Levin <ayal@mellanox.com>
Reviewed-by: Eran Ben Elisha <eranbe@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
---
 drivers/infiniband/hw/mlx5/main.c                  |   6 +-
 drivers/net/ethernet/mellanox/mlx5/core/en/port.c  |  75 +++++++++++++--
 drivers/net/ethernet/mellanox/mlx5/core/en/port.h  |  12 +++
 .../net/ethernet/mellanox/mlx5/core/en_ethtool.c   |  23 ++---
 drivers/net/ethernet/mellanox/mlx5/core/port.c     | 106 ---------------------
 include/linux/mlx5/port.h                          |  11 ---
 6 files changed, 91 insertions(+), 142 deletions(-)

(limited to 'drivers/infiniband/hw/mlx5/main.c')

diff --git a/drivers/infiniband/hw/mlx5/main.c b/drivers/infiniband/hw/mlx5/main.c
index 87ce62e44898..efd08b41126c 100644
--- a/drivers/infiniband/hw/mlx5/main.c
+++ b/drivers/infiniband/hw/mlx5/main.c
@@ -393,6 +393,7 @@ static int mlx5_query_port_roce(struct ib_device *device, u8 port_num,
 				struct ib_port_attr *props)
 {
 	struct mlx5_ib_dev *dev = to_mdev(device);
+	u32 out[MLX5_ST_SZ_DW(ptys_reg)] = {0};
 	struct mlx5_core_dev *mdev;
 	struct net_device *ndev, *upper;
 	enum ib_mtu ndev_ib_mtu;
@@ -416,10 +417,11 @@ static int mlx5_query_port_roce(struct ib_device *device, u8 port_num,
 	/* Possible bad flows are checked before filling out props so in case
 	 * of an error it will still be zeroed out.
 	 */
-	err = mlx5_query_port_eth_proto_oper(mdev, &eth_prot_oper,
-					     mdev_port_num);
+	err = mlx5_query_port_ptys(mdev, out, sizeof(out), MLX5_PTYS_EN,
+				   mdev_port_num);
 	if (err)
 		goto out;
+	eth_prot_oper = MLX5_GET(ptys_reg, out, eth_proto_oper);
 
 	props->active_width     = IB_WIDTH_4X;
 	props->active_speed     = IB_SPEED_QDR;
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/port.c b/drivers/net/ethernet/mellanox/mlx5/core/en/port.c
index 4a37713023be..9a1c2b2f87d8 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en/port.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/port.c
@@ -63,6 +63,67 @@ static const u32 mlx5e_link_speed[MLX5E_LINK_MODES_NUMBER] = {
 	[MLX5E_50GBASE_KR2]       = 50000,
 };
 
+int mlx5_port_query_eth_proto(struct mlx5_core_dev *dev, u8 port,
+			      struct mlx5e_port_eth_proto *eproto)
+{
+	u32 out[MLX5_ST_SZ_DW(ptys_reg)];
+	int err;
+
+	if (!eproto)
+		return -EINVAL;
+
+	err = mlx5_query_port_ptys(dev, out, sizeof(out), MLX5_PTYS_EN, port);
+	if (err)
+		return err;
+
+	eproto->cap   = MLX5_GET(ptys_reg, out, eth_proto_capability);
+	eproto->admin = MLX5_GET(ptys_reg, out, eth_proto_admin);
+	eproto->oper  = MLX5_GET(ptys_reg, out, eth_proto_oper);
+	return 0;
+}
+
+void mlx5_port_query_eth_autoneg(struct mlx5_core_dev *dev, u8 *an_status,
+				 u8 *an_disable_cap, u8 *an_disable_admin)
+{
+	u32 out[MLX5_ST_SZ_DW(ptys_reg)];
+
+	*an_status = 0;
+	*an_disable_cap = 0;
+	*an_disable_admin = 0;
+
+	if (mlx5_query_port_ptys(dev, out, sizeof(out), MLX5_PTYS_EN, 1))
+		return;
+
+	*an_status = MLX5_GET(ptys_reg, out, an_status);
+	*an_disable_cap = MLX5_GET(ptys_reg, out, an_disable_cap);
+	*an_disable_admin = MLX5_GET(ptys_reg, out, an_disable_admin);
+}
+
+int mlx5_port_set_eth_ptys(struct mlx5_core_dev *dev, bool an_disable,
+			   u32 proto_admin)
+{
+	u32 out[MLX5_ST_SZ_DW(ptys_reg)];
+	u32 in[MLX5_ST_SZ_DW(ptys_reg)];
+	u8 an_disable_admin;
+	u8 an_disable_cap;
+	u8 an_status;
+
+	mlx5_port_query_eth_autoneg(dev, &an_status, &an_disable_cap,
+				    &an_disable_admin);
+	if (!an_disable_cap && an_disable)
+		return -EPERM;
+
+	memset(in, 0, sizeof(in));
+
+	MLX5_SET(ptys_reg, in, local_port, 1);
+	MLX5_SET(ptys_reg, in, an_disable_admin, an_disable);
+	MLX5_SET(ptys_reg, in, proto_mask, MLX5_PTYS_EN);
+	MLX5_SET(ptys_reg, in, eth_proto_admin, proto_admin);
+
+	return mlx5_core_access_reg(dev, in, sizeof(in), out,
+			    sizeof(out), MLX5_REG_PTYS, 0, 1);
+}
+
 u32 mlx5e_port_ptys2speed(u32 eth_proto_oper)
 {
 	unsigned long temp = eth_proto_oper;
@@ -78,16 +139,14 @@ u32 mlx5e_port_ptys2speed(u32 eth_proto_oper)
 
 int mlx5e_port_linkspeed(struct mlx5_core_dev *mdev, u32 *speed)
 {
-	u32 out[MLX5_ST_SZ_DW(ptys_reg)] = {};
-	u32 eth_proto_oper;
+	struct mlx5e_port_eth_proto eproto;
 	int err;
 
-	err = mlx5_query_port_ptys(mdev, out, sizeof(out), MLX5_PTYS_EN, 1);
+	err = mlx5_port_query_eth_proto(mdev, 1, &eproto);
 	if (err)
 		return err;
 
-	eth_proto_oper = MLX5_GET(ptys_reg, out, eth_proto_oper);
-	*speed = mlx5e_port_ptys2speed(eth_proto_oper);
+	*speed = mlx5e_port_ptys2speed(eproto.oper);
 	if (!(*speed))
 		err = -EINVAL;
 
@@ -96,17 +155,17 @@ int mlx5e_port_linkspeed(struct mlx5_core_dev *mdev, u32 *speed)
 
 int mlx5e_port_max_linkspeed(struct mlx5_core_dev *mdev, u32 *speed)
 {
+	struct mlx5e_port_eth_proto eproto;
 	u32 max_speed = 0;
-	u32 proto_cap;
 	int err;
 	int i;
 
-	err = mlx5_query_port_proto_cap(mdev, &proto_cap, MLX5_PTYS_EN);
+	err = mlx5_port_query_eth_proto(mdev, 1, &eproto);
 	if (err)
 		return err;
 
 	for (i = 0; i < MLX5E_LINK_MODES_NUMBER; ++i)
-		if (proto_cap & MLX5E_PROT_MASK(i))
+		if (eproto.cap & MLX5E_PROT_MASK(i))
 			max_speed = max(max_speed, mlx5e_link_speed[i]);
 
 	*speed = max_speed;
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/port.h b/drivers/net/ethernet/mellanox/mlx5/core/en/port.h
index cd2160b8c9bf..4bdab8be10af 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en/port.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/port.h
@@ -36,6 +36,18 @@
 #include <linux/mlx5/driver.h>
 #include "en.h"
 
+struct mlx5e_port_eth_proto {
+	u32 cap;
+	u32 admin;
+	u32 oper;
+};
+
+int mlx5_port_query_eth_proto(struct mlx5_core_dev *dev, u8 port,
+			      struct mlx5e_port_eth_proto *eproto);
+void mlx5_port_query_eth_autoneg(struct mlx5_core_dev *dev, u8 *an_status,
+				 u8 *an_disable_cap, u8 *an_disable_admin);
+int mlx5_port_set_eth_ptys(struct mlx5_core_dev *dev, bool an_disable,
+			   u32 proto_admin);
 u32 mlx5e_port_ptys2speed(u32 eth_proto_oper);
 int mlx5e_port_linkspeed(struct mlx5_core_dev *mdev, u32 *speed);
 int mlx5e_port_max_linkspeed(struct mlx5_core_dev *mdev, u32 *speed);
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_ethtool.c b/drivers/net/ethernet/mellanox/mlx5/core/en_ethtool.c
index c9df08133718..c29e141d72fb 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_ethtool.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_ethtool.c
@@ -882,7 +882,7 @@ int mlx5e_ethtool_set_link_ksettings(struct mlx5e_priv *priv,
 				     const struct ethtool_link_ksettings *link_ksettings)
 {
 	struct mlx5_core_dev *mdev = priv->mdev;
-	u32 eth_proto_cap, eth_proto_admin;
+	struct mlx5e_port_eth_proto eproto;
 	bool an_changes = false;
 	u8 an_disable_admin;
 	u8 an_disable_cap;
@@ -898,14 +898,14 @@ int mlx5e_ethtool_set_link_ksettings(struct mlx5e_priv *priv,
 		mlx5e_ethtool2ptys_adver_link(link_ksettings->link_modes.advertising) :
 		mlx5e_port_speed2linkmodes(speed);
 
-	err = mlx5_query_port_proto_cap(mdev, &eth_proto_cap, MLX5_PTYS_EN);
+	err = mlx5_port_query_eth_proto(mdev, 1, &eproto);
 	if (err) {
-		netdev_err(priv->netdev, "%s: query port eth proto cap failed: %d\n",
+		netdev_err(priv->netdev, "%s: query port eth proto failed: %d\n",
 			   __func__, err);
 		goto out;
 	}
 
-	link_modes = link_modes & eth_proto_cap;
+	link_modes = link_modes & eproto.cap;
 	if (!link_modes) {
 		netdev_err(priv->netdev, "%s: Not supported link mode(s) requested",
 			   __func__);
@@ -913,24 +913,17 @@ int mlx5e_ethtool_set_link_ksettings(struct mlx5e_priv *priv,
 		goto out;
 	}
 
-	err = mlx5_query_port_proto_admin(mdev, &eth_proto_admin, MLX5_PTYS_EN);
-	if (err) {
-		netdev_err(priv->netdev, "%s: query port eth proto admin failed: %d\n",
-			   __func__, err);
-		goto out;
-	}
-
-	mlx5_query_port_autoneg(mdev, MLX5_PTYS_EN, &an_status,
-				&an_disable_cap, &an_disable_admin);
+	mlx5_port_query_eth_autoneg(mdev, &an_status, &an_disable_cap,
+				    &an_disable_admin);
 
 	an_disable = link_ksettings->base.autoneg == AUTONEG_DISABLE;
 	an_changes = ((!an_disable && an_disable_admin) ||
 		      (an_disable && !an_disable_admin));
 
-	if (!an_changes && link_modes == eth_proto_admin)
+	if (!an_changes && link_modes == eproto.admin)
 		goto out;
 
-	mlx5_set_port_ptys(mdev, an_disable, link_modes, MLX5_PTYS_EN);
+	mlx5_port_set_eth_ptys(mdev, an_disable, link_modes);
 	mlx5_toggle_port_link(mdev);
 
 out:
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/port.c b/drivers/net/ethernet/mellanox/mlx5/core/port.c
index 2b82f35f4c35..b81542820528 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/port.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/port.c
@@ -30,10 +30,7 @@
  * SOFTWARE.
  */
 
-#include <linux/module.h>
-#include <linux/mlx5/driver.h>
 #include <linux/mlx5/port.h>
-#include <linux/mlx5/cmd.h>
 #include "mlx5_core.h"
 
 int mlx5_core_access_reg(struct mlx5_core_dev *dev, void *data_in,
@@ -157,44 +154,6 @@ int mlx5_set_port_beacon(struct mlx5_core_dev *dev, u16 beacon_duration)
 				    sizeof(out), MLX5_REG_MLCR, 0, 1);
 }
 
-int mlx5_query_port_proto_cap(struct mlx5_core_dev *dev,
-			      u32 *proto_cap, int proto_mask)
-{
-	u32 out[MLX5_ST_SZ_DW(ptys_reg)];
-	int err;
-
-	err = mlx5_query_port_ptys(dev, out, sizeof(out), proto_mask, 1);
-	if (err)
-		return err;
-
-	if (proto_mask == MLX5_PTYS_EN)
-		*proto_cap = MLX5_GET(ptys_reg, out, eth_proto_capability);
-	else
-		*proto_cap = MLX5_GET(ptys_reg, out, ib_proto_capability);
-
-	return 0;
-}
-EXPORT_SYMBOL_GPL(mlx5_query_port_proto_cap);
-
-int mlx5_query_port_proto_admin(struct mlx5_core_dev *dev,
-				u32 *proto_admin, int proto_mask)
-{
-	u32 out[MLX5_ST_SZ_DW(ptys_reg)];
-	int err;
-
-	err = mlx5_query_port_ptys(dev, out, sizeof(out), proto_mask, 1);
-	if (err)
-		return err;
-
-	if (proto_mask == MLX5_PTYS_EN)
-		*proto_admin = MLX5_GET(ptys_reg, out, eth_proto_admin);
-	else
-		*proto_admin = MLX5_GET(ptys_reg, out, ib_proto_admin);
-
-	return 0;
-}
-EXPORT_SYMBOL_GPL(mlx5_query_port_proto_admin);
-
 int mlx5_query_port_link_width_oper(struct mlx5_core_dev *dev,
 				    u8 *link_width_oper, u8 local_port)
 {
@@ -211,23 +170,6 @@ int mlx5_query_port_link_width_oper(struct mlx5_core_dev *dev,
 }
 EXPORT_SYMBOL_GPL(mlx5_query_port_link_width_oper);
 
-int mlx5_query_port_eth_proto_oper(struct mlx5_core_dev *dev,
-				   u32 *proto_oper, u8 local_port)
-{
-	u32 out[MLX5_ST_SZ_DW(ptys_reg)];
-	int err;
-
-	err = mlx5_query_port_ptys(dev, out, sizeof(out), MLX5_PTYS_EN,
-				   local_port);
-	if (err)
-		return err;
-
-	*proto_oper = MLX5_GET(ptys_reg, out, eth_proto_oper);
-
-	return 0;
-}
-EXPORT_SYMBOL(mlx5_query_port_eth_proto_oper);
-
 int mlx5_query_port_ib_proto_oper(struct mlx5_core_dev *dev,
 				  u8 *proto_oper, u8 local_port)
 {
@@ -245,35 +187,6 @@ int mlx5_query_port_ib_proto_oper(struct mlx5_core_dev *dev,
 }
 EXPORT_SYMBOL(mlx5_query_port_ib_proto_oper);
 
-int mlx5_set_port_ptys(struct mlx5_core_dev *dev, bool an_disable,
-		       u32 proto_admin, int proto_mask)
-{
-	u32 out[MLX5_ST_SZ_DW(ptys_reg)];
-	u32 in[MLX5_ST_SZ_DW(ptys_reg)];
-	u8 an_disable_admin;
-	u8 an_disable_cap;
-	u8 an_status;
-
-	mlx5_query_port_autoneg(dev, proto_mask, &an_status,
-				&an_disable_cap, &an_disable_admin);
-	if (!an_disable_cap && an_disable)
-		return -EPERM;
-
-	memset(in, 0, sizeof(in));
-
-	MLX5_SET(ptys_reg, in, local_port, 1);
-	MLX5_SET(ptys_reg, in, an_disable_admin, an_disable);
-	MLX5_SET(ptys_reg, in, proto_mask, proto_mask);
-	if (proto_mask == MLX5_PTYS_EN)
-		MLX5_SET(ptys_reg, in, eth_proto_admin, proto_admin);
-	else
-		MLX5_SET(ptys_reg, in, ib_proto_admin, proto_admin);
-
-	return mlx5_core_access_reg(dev, in, sizeof(in), out,
-				    sizeof(out), MLX5_REG_PTYS, 0, 1);
-}
-EXPORT_SYMBOL_GPL(mlx5_set_port_ptys);
-
 /* This function should be used after setting a port register only */
 void mlx5_toggle_port_link(struct mlx5_core_dev *dev)
 {
@@ -606,25 +519,6 @@ int mlx5_query_port_pfc(struct mlx5_core_dev *dev, u8 *pfc_en_tx, u8 *pfc_en_rx)
 }
 EXPORT_SYMBOL_GPL(mlx5_query_port_pfc);
 
-void mlx5_query_port_autoneg(struct mlx5_core_dev *dev, int proto_mask,
-			     u8 *an_status,
-			     u8 *an_disable_cap, u8 *an_disable_admin)
-{
-	u32 out[MLX5_ST_SZ_DW(ptys_reg)];
-
-	*an_status = 0;
-	*an_disable_cap = 0;
-	*an_disable_admin = 0;
-
-	if (mlx5_query_port_ptys(dev, out, sizeof(out), proto_mask, 1))
-		return;
-
-	*an_status = MLX5_GET(ptys_reg, out, an_status);
-	*an_disable_cap = MLX5_GET(ptys_reg, out, an_disable_cap);
-	*an_disable_admin = MLX5_GET(ptys_reg, out, an_disable_admin);
-}
-EXPORT_SYMBOL_GPL(mlx5_query_port_autoneg);
-
 int mlx5_max_tc(struct mlx5_core_dev *mdev)
 {
 	u8 num_tc = MLX5_CAP_GEN(mdev, max_tc) ? : 8;
diff --git a/include/linux/mlx5/port.h b/include/linux/mlx5/port.h
index bf4bc01ffb0c..5be7eefa6d75 100644
--- a/include/linux/mlx5/port.h
+++ b/include/linux/mlx5/port.h
@@ -110,27 +110,16 @@ enum mlx5e_connector_type {
 int mlx5_set_port_caps(struct mlx5_core_dev *dev, u8 port_num, u32 caps);
 int mlx5_query_port_ptys(struct mlx5_core_dev *dev, u32 *ptys,
 			 int ptys_size, int proto_mask, u8 local_port);
-int mlx5_query_port_proto_cap(struct mlx5_core_dev *dev,
-			      u32 *proto_cap, int proto_mask);
-int mlx5_query_port_proto_admin(struct mlx5_core_dev *dev,
-				u32 *proto_admin, int proto_mask);
 int mlx5_query_port_link_width_oper(struct mlx5_core_dev *dev,
 				    u8 *link_width_oper, u8 local_port);
 int mlx5_query_port_ib_proto_oper(struct mlx5_core_dev *dev,
 				  u8 *proto_oper, u8 local_port);
-int mlx5_query_port_eth_proto_oper(struct mlx5_core_dev *dev,
-				   u32 *proto_oper, u8 local_port);
-int mlx5_set_port_ptys(struct mlx5_core_dev *dev, bool an_disable,
-		       u32 proto_admin, int proto_mask);
 void mlx5_toggle_port_link(struct mlx5_core_dev *dev);
 int mlx5_set_port_admin_status(struct mlx5_core_dev *dev,
 			       enum mlx5_port_status status);
 int mlx5_query_port_admin_status(struct mlx5_core_dev *dev,
 				 enum mlx5_port_status *status);
 int mlx5_set_port_beacon(struct mlx5_core_dev *dev, u16 beacon_duration);
-void mlx5_query_port_autoneg(struct mlx5_core_dev *dev, int proto_mask,
-			     u8 *an_status,
-			     u8 *an_disable_cap, u8 *an_disable_admin);
 
 int mlx5_set_port_mtu(struct mlx5_core_dev *dev, u16 mtu, u8 port);
 void mlx5_query_port_max_mtu(struct mlx5_core_dev *dev, u16 *max_mtu, u8 port);
-- 
cgit v1.2.3


From a08b4ed1373dc59e3e15029bc6f135ba0f53c9a7 Mon Sep 17 00:00:00 2001
From: Aya Levin <ayal@mellanox.com>
Date: Tue, 12 Feb 2019 22:55:45 -0800
Subject: net/mlx5: Add support to ext_* fields introduced in Port Type and
 Speed register

This patch exposes new link modes (including 50Gbps per lane), and ext_*
fields which describes the new link modes in Port Type and Speed
register (PTYS).
Access functions, translation functions (speed <-> HW bits) and
link max speed function were modified.

Signed-off-by: Aya Levin <ayal@mellanox.com>
Reviewed-by: Eran Ben Elisha <eranbe@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
---
 drivers/infiniband/hw/mlx5/main.c                  |  3 +-
 drivers/net/ethernet/mellanox/mlx5/core/en/port.c  | 85 ++++++++++++++++------
 drivers/net/ethernet/mellanox/mlx5/core/en/port.h  |  8 +-
 .../net/ethernet/mellanox/mlx5/core/en_ethtool.c   |  9 ++-
 include/linux/mlx5/port.h                          | 19 +++++
 5 files changed, 94 insertions(+), 30 deletions(-)

(limited to 'drivers/infiniband/hw/mlx5/main.c')

diff --git a/drivers/infiniband/hw/mlx5/main.c b/drivers/infiniband/hw/mlx5/main.c
index efd08b41126c..3677c00fa3bb 100644
--- a/drivers/infiniband/hw/mlx5/main.c
+++ b/drivers/infiniband/hw/mlx5/main.c
@@ -421,7 +421,8 @@ static int mlx5_query_port_roce(struct ib_device *device, u8 port_num,
 				   mdev_port_num);
 	if (err)
 		goto out;
-	eth_prot_oper = MLX5_GET(ptys_reg, out, eth_proto_oper);
+	eth_prot_oper = MLX5_GET_ETH_PROTO(ptys_reg, out, false,
+					   eth_proto_oper);
 
 	props->active_width     = IB_WIDTH_4X;
 	props->active_speed     = IB_SPEED_QDR;
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/port.c b/drivers/net/ethernet/mellanox/mlx5/core/en/port.c
index 9a1c2b2f87d8..122927f3a600 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en/port.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/port.c
@@ -63,7 +63,31 @@ static const u32 mlx5e_link_speed[MLX5E_LINK_MODES_NUMBER] = {
 	[MLX5E_50GBASE_KR2]       = 50000,
 };
 
-int mlx5_port_query_eth_proto(struct mlx5_core_dev *dev, u8 port,
+static const u32 mlx5e_ext_link_speed[MLX5E_EXT_LINK_MODES_NUMBER] = {
+	[MLX5E_SGMII_100M]			= 100,
+	[MLX5E_1000BASE_X_SGMII]		= 1000,
+	[MLX5E_5GBASE_R]			= 5000,
+	[MLX5E_10GBASE_XFI_XAUI_1]		= 10000,
+	[MLX5E_40GBASE_XLAUI_4_XLPPI_4]		= 40000,
+	[MLX5E_25GAUI_1_25GBASE_CR_KR]		= 25000,
+	[MLX5E_50GAUI_2_LAUI_2_50GBASE_CR2_KR2]	= 50000,
+	[MLX5E_50GAUI_1_LAUI_1_50GBASE_CR_KR]	= 50000,
+	[MLX5E_CAUI_4_100GBASE_CR4_KR4]		= 100000,
+	[MLX5E_200GAUI_4_200GBASE_CR4_KR4]	= 200000,
+	[MLX5E_400GAUI_8]			= 400000,
+};
+
+static void mlx5e_port_get_speed_arr(struct mlx5_core_dev *mdev,
+				     const u32 **arr, u32 *size)
+{
+	bool ext = MLX5_CAP_PCAM_FEATURE(mdev, ptys_extended_ethernet);
+
+	*size = ext ? ARRAY_SIZE(mlx5e_ext_link_speed) :
+		      ARRAY_SIZE(mlx5e_link_speed);
+	*arr  = ext ? mlx5e_ext_link_speed : mlx5e_link_speed;
+}
+
+int mlx5_port_query_eth_proto(struct mlx5_core_dev *dev, u8 port, bool ext,
 			      struct mlx5e_port_eth_proto *eproto)
 {
 	u32 out[MLX5_ST_SZ_DW(ptys_reg)];
@@ -72,13 +96,17 @@ int mlx5_port_query_eth_proto(struct mlx5_core_dev *dev, u8 port,
 	if (!eproto)
 		return -EINVAL;
 
+	if (ext !=  MLX5_CAP_PCAM_FEATURE(dev, ptys_extended_ethernet))
+		return -EOPNOTSUPP;
+
 	err = mlx5_query_port_ptys(dev, out, sizeof(out), MLX5_PTYS_EN, port);
 	if (err)
 		return err;
 
-	eproto->cap   = MLX5_GET(ptys_reg, out, eth_proto_capability);
-	eproto->admin = MLX5_GET(ptys_reg, out, eth_proto_admin);
-	eproto->oper  = MLX5_GET(ptys_reg, out, eth_proto_oper);
+	eproto->cap   = MLX5_GET_ETH_PROTO(ptys_reg, out, ext,
+					   eth_proto_capability);
+	eproto->admin = MLX5_GET_ETH_PROTO(ptys_reg, out, ext, eth_proto_admin);
+	eproto->oper  = MLX5_GET_ETH_PROTO(ptys_reg, out, ext, eth_proto_oper);
 	return 0;
 }
 
@@ -100,7 +128,7 @@ void mlx5_port_query_eth_autoneg(struct mlx5_core_dev *dev, u8 *an_status,
 }
 
 int mlx5_port_set_eth_ptys(struct mlx5_core_dev *dev, bool an_disable,
-			   u32 proto_admin)
+			   u32 proto_admin, bool ext)
 {
 	u32 out[MLX5_ST_SZ_DW(ptys_reg)];
 	u32 in[MLX5_ST_SZ_DW(ptys_reg)];
@@ -118,38 +146,46 @@ int mlx5_port_set_eth_ptys(struct mlx5_core_dev *dev, bool an_disable,
 	MLX5_SET(ptys_reg, in, local_port, 1);
 	MLX5_SET(ptys_reg, in, an_disable_admin, an_disable);
 	MLX5_SET(ptys_reg, in, proto_mask, MLX5_PTYS_EN);
-	MLX5_SET(ptys_reg, in, eth_proto_admin, proto_admin);
+	if (ext)
+		MLX5_SET(ptys_reg, in, ext_eth_proto_admin, proto_admin);
+	else
+		MLX5_SET(ptys_reg, in, eth_proto_admin, proto_admin);
 
 	return mlx5_core_access_reg(dev, in, sizeof(in), out,
 			    sizeof(out), MLX5_REG_PTYS, 0, 1);
 }
 
-u32 mlx5e_port_ptys2speed(u32 eth_proto_oper)
+u32 mlx5e_port_ptys2speed(struct mlx5_core_dev *mdev, u32 eth_proto_oper)
 {
 	unsigned long temp = eth_proto_oper;
+	const u32 *table;
 	u32 speed = 0;
+	u32 max_size;
 	int i;
 
-	i = find_first_bit(&temp, MLX5E_LINK_MODES_NUMBER);
-	if (i < MLX5E_LINK_MODES_NUMBER)
-		speed = mlx5e_link_speed[i];
-
+	mlx5e_port_get_speed_arr(mdev, &table, &max_size);
+	i = find_first_bit(&temp, max_size);
+	if (i < max_size)
+		speed = table[i];
 	return speed;
 }
 
 int mlx5e_port_linkspeed(struct mlx5_core_dev *mdev, u32 *speed)
 {
 	struct mlx5e_port_eth_proto eproto;
+	bool ext;
 	int err;
 
-	err = mlx5_port_query_eth_proto(mdev, 1, &eproto);
+	ext = MLX5_CAP_PCAM_FEATURE(mdev, ptys_extended_ethernet);
+	err = mlx5_port_query_eth_proto(mdev, 1, ext, &eproto);
 	if (err)
-		return err;
+		goto out;
 
-	*speed = mlx5e_port_ptys2speed(eproto.oper);
+	*speed = mlx5e_port_ptys2speed(mdev, eproto.oper);
 	if (!(*speed))
 		err = -EINVAL;
 
+out:
 	return err;
 }
 
@@ -157,31 +193,38 @@ int mlx5e_port_max_linkspeed(struct mlx5_core_dev *mdev, u32 *speed)
 {
 	struct mlx5e_port_eth_proto eproto;
 	u32 max_speed = 0;
+	const u32 *table;
+	u32 max_size;
+	bool ext;
 	int err;
 	int i;
 
-	err = mlx5_port_query_eth_proto(mdev, 1, &eproto);
+	ext = MLX5_CAP_PCAM_FEATURE(mdev, ptys_extended_ethernet);
+	err = mlx5_port_query_eth_proto(mdev, 1, ext, &eproto);
 	if (err)
 		return err;
 
-	for (i = 0; i < MLX5E_LINK_MODES_NUMBER; ++i)
+	mlx5e_port_get_speed_arr(mdev, &table, &max_size);
+	for (i = 0; i < max_size; ++i)
 		if (eproto.cap & MLX5E_PROT_MASK(i))
-			max_speed = max(max_speed, mlx5e_link_speed[i]);
+			max_speed = max(max_speed, table[i]);
 
 	*speed = max_speed;
 	return 0;
 }
 
-u32 mlx5e_port_speed2linkmodes(u32 speed)
+u32 mlx5e_port_speed2linkmodes(struct mlx5_core_dev *mdev, u32 speed)
 {
 	u32 link_modes = 0;
+	const u32 *table;
+	u32 max_size;
 	int i;
 
-	for (i = 0; i < MLX5E_LINK_MODES_NUMBER; ++i) {
-		if (mlx5e_link_speed[i] == speed)
+	mlx5e_port_get_speed_arr(mdev, &table, &max_size);
+	for (i = 0; i < max_size; ++i) {
+		if (table[i] == speed)
 			link_modes |= MLX5E_PROT_MASK(i);
 	}
-
 	return link_modes;
 }
 
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/port.h b/drivers/net/ethernet/mellanox/mlx5/core/en/port.h
index 4bdab8be10af..70f536ec51c4 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en/port.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/port.h
@@ -42,16 +42,16 @@ struct mlx5e_port_eth_proto {
 	u32 oper;
 };
 
-int mlx5_port_query_eth_proto(struct mlx5_core_dev *dev, u8 port,
+int mlx5_port_query_eth_proto(struct mlx5_core_dev *dev, u8 port, bool ext,
 			      struct mlx5e_port_eth_proto *eproto);
 void mlx5_port_query_eth_autoneg(struct mlx5_core_dev *dev, u8 *an_status,
 				 u8 *an_disable_cap, u8 *an_disable_admin);
 int mlx5_port_set_eth_ptys(struct mlx5_core_dev *dev, bool an_disable,
-			   u32 proto_admin);
-u32 mlx5e_port_ptys2speed(u32 eth_proto_oper);
+			   u32 proto_admin, bool ext);
+u32 mlx5e_port_ptys2speed(struct mlx5_core_dev *mdev, u32 eth_proto_oper);
 int mlx5e_port_linkspeed(struct mlx5_core_dev *mdev, u32 *speed);
 int mlx5e_port_max_linkspeed(struct mlx5_core_dev *mdev, u32 *speed);
-u32 mlx5e_port_speed2linkmodes(u32 speed);
+u32 mlx5e_port_speed2linkmodes(struct mlx5_core_dev *mdev, u32 speed);
 
 int mlx5e_port_query_pbmc(struct mlx5_core_dev *mdev, void *out);
 int mlx5e_port_set_pbmc(struct mlx5_core_dev *mdev, void *in);
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_ethtool.c b/drivers/net/ethernet/mellanox/mlx5/core/en_ethtool.c
index c29e141d72fb..8343cf7d292c 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_ethtool.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_ethtool.c
@@ -695,13 +695,14 @@ static void get_speed_duplex(struct net_device *netdev,
 			     u32 eth_proto_oper,
 			     struct ethtool_link_ksettings *link_ksettings)
 {
+	struct mlx5e_priv *priv = netdev_priv(netdev);
 	u32 speed = SPEED_UNKNOWN;
 	u8 duplex = DUPLEX_UNKNOWN;
 
 	if (!netif_carrier_ok(netdev))
 		goto out;
 
-	speed = mlx5e_port_ptys2speed(eth_proto_oper);
+	speed = mlx5e_port_ptys2speed(priv->mdev, eth_proto_oper);
 	if (!speed) {
 		speed = SPEED_UNKNOWN;
 		goto out;
@@ -896,9 +897,9 @@ int mlx5e_ethtool_set_link_ksettings(struct mlx5e_priv *priv,
 
 	link_modes = link_ksettings->base.autoneg == AUTONEG_ENABLE ?
 		mlx5e_ethtool2ptys_adver_link(link_ksettings->link_modes.advertising) :
-		mlx5e_port_speed2linkmodes(speed);
+		mlx5e_port_speed2linkmodes(mdev, speed);
 
-	err = mlx5_port_query_eth_proto(mdev, 1, &eproto);
+	err = mlx5_port_query_eth_proto(mdev, 1, false, &eproto);
 	if (err) {
 		netdev_err(priv->netdev, "%s: query port eth proto failed: %d\n",
 			   __func__, err);
@@ -923,7 +924,7 @@ int mlx5e_ethtool_set_link_ksettings(struct mlx5e_priv *priv,
 	if (!an_changes && link_modes == eproto.admin)
 		goto out;
 
-	mlx5_port_set_eth_ptys(mdev, an_disable, link_modes);
+	mlx5_port_set_eth_ptys(mdev, an_disable, link_modes, false);
 	mlx5_toggle_port_link(mdev);
 
 out:
diff --git a/include/linux/mlx5/port.h b/include/linux/mlx5/port.h
index 5be7eefa6d75..814fa194663b 100644
--- a/include/linux/mlx5/port.h
+++ b/include/linux/mlx5/port.h
@@ -92,6 +92,22 @@ enum mlx5e_link_mode {
 	MLX5E_LINK_MODES_NUMBER,
 };
 
+enum mlx5e_ext_link_mode {
+	MLX5E_SGMII_100M			= 0,
+	MLX5E_1000BASE_X_SGMII			= 1,
+	MLX5E_5GBASE_R				= 3,
+	MLX5E_10GBASE_XFI_XAUI_1		= 4,
+	MLX5E_40GBASE_XLAUI_4_XLPPI_4		= 5,
+	MLX5E_25GAUI_1_25GBASE_CR_KR		= 6,
+	MLX5E_50GAUI_2_LAUI_2_50GBASE_CR2_KR2	= 7,
+	MLX5E_50GAUI_1_LAUI_1_50GBASE_CR_KR	= 8,
+	MLX5E_CAUI_4_100GBASE_CR4_KR4		= 9,
+	MLX5E_100GAUI_2_100GBASE_CR2_KR2	= 10,
+	MLX5E_200GAUI_4_200GBASE_CR4_KR4	= 12,
+	MLX5E_400GAUI_8				= 15,
+	MLX5E_EXT_LINK_MODES_NUMBER,
+};
+
 enum mlx5e_connector_type {
 	MLX5E_PORT_UNKNOWN	= 0,
 	MLX5E_PORT_NONE			= 1,
@@ -106,6 +122,9 @@ enum mlx5e_connector_type {
 };
 
 #define MLX5E_PROT_MASK(link_mode) (1 << link_mode)
+#define MLX5_GET_ETH_PROTO(reg, out, ext, field)	\
+	(ext ? MLX5_GET(reg, out, ext_##field) :	\
+	MLX5_GET(reg, out, field))
 
 int mlx5_set_port_caps(struct mlx5_core_dev *dev, u8 port_num, u32 caps);
 int mlx5_query_port_ptys(struct mlx5_core_dev *dev, u32 *ptys,
-- 
cgit v1.2.3


From 08e8676f1607925adf36e399f0faa8ce3b10bb86 Mon Sep 17 00:00:00 2001
From: Aya Levin <ayal@mellanox.com>
Date: Tue, 12 Feb 2019 22:55:46 -0800
Subject: IB/mlx5: Add support for 50Gbps per lane link modes

Driver now supports new link modes: 50Gbps per lane support for
50G/100G/200G. This patch reads the correct field (legacy vs. extended)
based on a FW indication bit, and adds a translation function (link
modes to IB width and speed) to the new link modes.

Signed-off-by: Aya Levin <ayal@mellanox.com>
Reviewed-by: Leon Romanovsky <leonro@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
---
 drivers/infiniband/hw/mlx5/main.c | 66 ++++++++++++++++++++++++++++++++++++---
 1 file changed, 61 insertions(+), 5 deletions(-)

(limited to 'drivers/infiniband/hw/mlx5/main.c')

diff --git a/drivers/infiniband/hw/mlx5/main.c b/drivers/infiniband/hw/mlx5/main.c
index 3677c00fa3bb..581ae11e2fc9 100644
--- a/drivers/infiniband/hw/mlx5/main.c
+++ b/drivers/infiniband/hw/mlx5/main.c
@@ -331,8 +331,8 @@ out:
 	spin_unlock(&port->mp.mpi_lock);
 }
 
-static int translate_eth_proto_oper(u32 eth_proto_oper, u8 *active_speed,
-				    u8 *active_width)
+static int translate_eth_legacy_proto_oper(u32 eth_proto_oper, u8 *active_speed,
+					   u8 *active_width)
 {
 	switch (eth_proto_oper) {
 	case MLX5E_PROT_MASK(MLX5E_1000BASE_CX_SGMII):
@@ -389,6 +389,61 @@ static int translate_eth_proto_oper(u32 eth_proto_oper, u8 *active_speed,
 	return 0;
 }
 
+static int translate_eth_ext_proto_oper(u32 eth_proto_oper, u8 *active_speed,
+					u8 *active_width)
+{
+	switch (eth_proto_oper) {
+	case MLX5E_PROT_MASK(MLX5E_SGMII_100M):
+	case MLX5E_PROT_MASK(MLX5E_1000BASE_X_SGMII):
+		*active_width = IB_WIDTH_1X;
+		*active_speed = IB_SPEED_SDR;
+		break;
+	case MLX5E_PROT_MASK(MLX5E_5GBASE_R):
+		*active_width = IB_WIDTH_1X;
+		*active_speed = IB_SPEED_DDR;
+		break;
+	case MLX5E_PROT_MASK(MLX5E_10GBASE_XFI_XAUI_1):
+		*active_width = IB_WIDTH_1X;
+		*active_speed = IB_SPEED_QDR;
+		break;
+	case MLX5E_PROT_MASK(MLX5E_40GBASE_XLAUI_4_XLPPI_4):
+		*active_width = IB_WIDTH_4X;
+		*active_speed = IB_SPEED_QDR;
+		break;
+	case MLX5E_PROT_MASK(MLX5E_25GAUI_1_25GBASE_CR_KR):
+		*active_width = IB_WIDTH_1X;
+		*active_speed = IB_SPEED_EDR;
+		break;
+	case MLX5E_PROT_MASK(MLX5E_50GAUI_2_LAUI_2_50GBASE_CR2_KR2):
+	case MLX5E_PROT_MASK(MLX5E_50GAUI_1_LAUI_1_50GBASE_CR_KR):
+		*active_width = IB_WIDTH_1X;
+		*active_speed = IB_SPEED_HDR;
+		break;
+	case MLX5E_PROT_MASK(MLX5E_100GAUI_2_100GBASE_CR2_KR2):
+		*active_width = IB_WIDTH_2X;
+		*active_speed = IB_SPEED_HDR;
+		break;
+	case MLX5E_PROT_MASK(MLX5E_200GAUI_4_200GBASE_CR4_KR4):
+		*active_width = IB_WIDTH_4X;
+		*active_speed = IB_SPEED_HDR;
+		break;
+	default:
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+static int translate_eth_proto_oper(u32 eth_proto_oper, u8 *active_speed,
+				    u8 *active_width, bool ext)
+{
+	return ext ?
+		translate_eth_ext_proto_oper(eth_proto_oper, active_speed,
+					     active_width) :
+		translate_eth_legacy_proto_oper(eth_proto_oper, active_speed,
+						active_width);
+}
+
 static int mlx5_query_port_roce(struct ib_device *device, u8 port_num,
 				struct ib_port_attr *props)
 {
@@ -401,6 +456,7 @@ static int mlx5_query_port_roce(struct ib_device *device, u8 port_num,
 	u16 qkey_viol_cntr;
 	u32 eth_prot_oper;
 	u8 mdev_port_num;
+	bool ext;
 	int err;
 
 	mdev = mlx5_ib_get_native_port_mdev(dev, port_num, &mdev_port_num);
@@ -421,14 +477,14 @@ static int mlx5_query_port_roce(struct ib_device *device, u8 port_num,
 				   mdev_port_num);
 	if (err)
 		goto out;
-	eth_prot_oper = MLX5_GET_ETH_PROTO(ptys_reg, out, false,
-					   eth_proto_oper);
+	ext = MLX5_CAP_PCAM_FEATURE(dev->mdev, ptys_extended_ethernet);
+	eth_prot_oper = MLX5_GET_ETH_PROTO(ptys_reg, out, ext, eth_proto_oper);
 
 	props->active_width     = IB_WIDTH_4X;
 	props->active_speed     = IB_SPEED_QDR;
 
 	translate_eth_proto_oper(eth_prot_oper, &props->active_speed,
-				 &props->active_width);
+				 &props->active_width, ext);
 
 	props->port_cap_flags |= IB_PORT_CM_SUP;
 	props->ip_gids = true;
-- 
cgit v1.2.3


From a6bc3875f176f52c4a247c341e80d52dd4f5e356 Mon Sep 17 00:00:00 2001
From: Moni Shoua <monis@mellanox.com>
Date: Sun, 17 Feb 2019 16:08:22 +0200
Subject: IB/mlx5: Protect against prefetch of invalid MR

When deferring a prefetch request we need to protect against MR or PD
being destroyed while the request is still enqueued.

The first step is to validate that PD owns the lkey that describes the MR
and that the MR that the lkey refers to is owned by that PD.

The second step is to dequeue all requests when MR is destroyed.

Since PD can't be destroyed while it owns MRs it is guaranteed that when a
worker wakes up the request it refers to is still valid.

Now, it is possible to refrain from taking a reference on the device since
it is assured to be present as pd.

While that, replace the dedicated ordered workqueue with the system
unbound workqueue to reuse an existing resource and improve
performance. This will also fix a bug of queueing to the wrong workqueue.

Fixes: 813e90b1aeaa ("IB/mlx5: Add advise_mr() support")
Reported-by: Parav Pandit <parav@mellanox.com>
Signed-off-by: Moni Shoua <monis@mellanox.com>
Signed-off-by: Leon Romanovsky <leonro@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 drivers/infiniband/hw/mlx5/main.c    | 13 +----
 drivers/infiniband/hw/mlx5/mlx5_ib.h |  2 +-
 drivers/infiniband/hw/mlx5/mr.c      | 14 +++++-
 drivers/infiniband/hw/mlx5/odp.c     | 96 +++++++++++++++++++++++++++++++++---
 4 files changed, 102 insertions(+), 23 deletions(-)

(limited to 'drivers/infiniband/hw/mlx5/main.c')

diff --git a/drivers/infiniband/hw/mlx5/main.c b/drivers/infiniband/hw/mlx5/main.c
index 45afe1d9e125..614b2acbc621 100644
--- a/drivers/infiniband/hw/mlx5/main.c
+++ b/drivers/infiniband/hw/mlx5/main.c
@@ -5829,8 +5829,6 @@ void mlx5_ib_stage_init_cleanup(struct mlx5_ib_dev *dev)
 	if (IS_ENABLED(CONFIG_INFINIBAND_ON_DEMAND_PAGING)) {
 		srcu_barrier(&dev->mr_srcu);
 		cleanup_srcu_struct(&dev->mr_srcu);
-		drain_workqueue(dev->advise_mr_wq);
-		destroy_workqueue(dev->advise_mr_wq);
 	}
 	kfree(dev->port);
 }
@@ -5885,18 +5883,9 @@ int mlx5_ib_stage_init_init(struct mlx5_ib_dev *dev)
 	dev->memic.dev = mdev;
 
 	if (IS_ENABLED(CONFIG_INFINIBAND_ON_DEMAND_PAGING)) {
-		dev->advise_mr_wq =
-			alloc_ordered_workqueue("mlx5_ib_advise_mr_wq", 0);
-		if (!dev->advise_mr_wq) {
-			err = -ENOMEM;
-			goto err_mp;
-		}
-
 		err = init_srcu_struct(&dev->mr_srcu);
-		if (err) {
-			destroy_workqueue(dev->advise_mr_wq);
+		if (err)
 			goto err_mp;
-		}
 	}
 
 	return 0;
diff --git a/drivers/infiniband/hw/mlx5/mlx5_ib.h b/drivers/infiniband/hw/mlx5/mlx5_ib.h
index f9817284c7a3..4a617d78eae1 100644
--- a/drivers/infiniband/hw/mlx5/mlx5_ib.h
+++ b/drivers/infiniband/hw/mlx5/mlx5_ib.h
@@ -589,6 +589,7 @@ struct mlx5_ib_mr {
 	atomic_t		num_leaf_free;
 	wait_queue_head_t       q_leaf_free;
 	struct mlx5_async_work  cb_work;
+	atomic_t		num_pending_prefetch;
 };
 
 static inline bool is_odp_mr(struct mlx5_ib_mr *mr)
@@ -929,7 +930,6 @@ struct mlx5_ib_dev {
 	 */
 	struct srcu_struct      mr_srcu;
 	u32			null_mkey;
-	struct workqueue_struct *advise_mr_wq;
 	struct mlx5_ib_flow_db	*flow_db;
 	/* protect resources needed as part of reset flow */
 	spinlock_t		reset_flow_resource_lock;
diff --git a/drivers/infiniband/hw/mlx5/mr.c b/drivers/infiniband/hw/mlx5/mr.c
index 705a79cd21da..c85f00255884 100644
--- a/drivers/infiniband/hw/mlx5/mr.c
+++ b/drivers/infiniband/hw/mlx5/mr.c
@@ -1339,8 +1339,10 @@ struct ib_mr *mlx5_ib_reg_user_mr(struct ib_pd *pd, u64 start, u64 length,
 		}
 	}
 
-	if (IS_ENABLED(CONFIG_INFINIBAND_ON_DEMAND_PAGING))
+	if (IS_ENABLED(CONFIG_INFINIBAND_ON_DEMAND_PAGING)) {
 		mr->live = 1;
+		atomic_set(&mr->num_pending_prefetch, 0);
+	}
 
 	return &mr->ibmr;
 error:
@@ -1576,8 +1578,16 @@ static void dereg_mr(struct mlx5_ib_dev *dev, struct mlx5_ib_mr *mr)
 	if (is_odp_mr(mr)) {
 		struct ib_umem_odp *umem_odp = to_ib_umem_odp(umem);
 
-		/* Prevent new page faults from succeeding */
+		/* Prevent new page faults and
+		 * prefetch requests from succeeding
+		 */
 		mr->live = 0;
+
+		/* dequeue pending prefetch requests for the mr */
+		if (atomic_read(&mr->num_pending_prefetch))
+			flush_workqueue(system_unbound_wq);
+		WARN_ON(atomic_read(&mr->num_pending_prefetch));
+
 		/* Wait for all running page-fault handlers to finish. */
 		synchronize_srcu(&dev->mr_srcu);
 		/* Destroy all page mappings */
diff --git a/drivers/infiniband/hw/mlx5/odp.c b/drivers/infiniband/hw/mlx5/odp.c
index 335fd0c6ea2a..a1346978702f 100644
--- a/drivers/infiniband/hw/mlx5/odp.c
+++ b/drivers/infiniband/hw/mlx5/odp.c
@@ -535,6 +535,7 @@ struct mlx5_ib_mr *mlx5_ib_alloc_implicit_mr(struct mlx5_ib_pd *pd,
 	imr->umem = umem;
 	init_waitqueue_head(&imr->q_leaf_free);
 	atomic_set(&imr->num_leaf_free, 0);
+	atomic_set(&imr->num_pending_prefetch, 0);
 
 	return imr;
 }
@@ -1660,24 +1661,91 @@ struct prefetch_mr_work {
 	struct ib_sge sg_list[0];
 };
 
+static void num_pending_prefetch_dec(struct mlx5_ib_dev *dev,
+				     struct ib_sge *sg_list, u32 num_sge,
+				     u32 from)
+{
+	u32 i;
+	int srcu_key;
+
+	srcu_key = srcu_read_lock(&dev->mr_srcu);
+
+	for (i = from; i < num_sge; ++i) {
+		struct mlx5_core_mkey *mmkey;
+		struct mlx5_ib_mr *mr;
+
+		mmkey = __mlx5_mr_lookup(dev->mdev,
+					 mlx5_base_mkey(sg_list[i].lkey));
+		mr = container_of(mmkey, struct mlx5_ib_mr, mmkey);
+		atomic_dec(&mr->num_pending_prefetch);
+	}
+
+	srcu_read_unlock(&dev->mr_srcu, srcu_key);
+}
+
+static bool num_pending_prefetch_inc(struct ib_pd *pd,
+				     struct ib_sge *sg_list, u32 num_sge)
+{
+	struct mlx5_ib_dev *dev = to_mdev(pd->device);
+	bool ret = true;
+	u32 i;
+
+	for (i = 0; i < num_sge; ++i) {
+		struct mlx5_core_mkey *mmkey;
+		struct mlx5_ib_mr *mr;
+
+		mmkey = __mlx5_mr_lookup(dev->mdev,
+					 mlx5_base_mkey(sg_list[i].lkey));
+		if (!mmkey || mmkey->key != sg_list[i].lkey) {
+			ret = false;
+			break;
+		}
+
+		if (mmkey->type != MLX5_MKEY_MR) {
+			ret = false;
+			break;
+		}
+
+		mr = container_of(mmkey, struct mlx5_ib_mr, mmkey);
+
+		if (mr->ibmr.pd != pd) {
+			ret = false;
+			break;
+		}
+
+		if (!mr->live) {
+			ret = false;
+			break;
+		}
+
+		atomic_inc(&mr->num_pending_prefetch);
+	}
+
+	if (!ret)
+		num_pending_prefetch_dec(dev, sg_list, i, 0);
+
+	return ret;
+}
+
 static int mlx5_ib_prefetch_sg_list(struct mlx5_ib_dev *dev, u32 pf_flags,
 				    struct ib_sge *sg_list, u32 num_sge)
 {
-	int i;
+	u32 i;
+	int ret = 0;
 
 	for (i = 0; i < num_sge; ++i) {
 		struct ib_sge *sg = &sg_list[i];
 		int bytes_committed = 0;
-		int ret;
 
 		ret = pagefault_single_data_segment(dev, sg->lkey, sg->addr,
 						    sg->length,
 						    &bytes_committed, NULL,
 						    pf_flags);
 		if (ret < 0)
-			return ret;
+			break;
 	}
-	return 0;
+
+	return ret < 0 ? ret : 0;
 }
 
 static void mlx5_ib_prefetch_mr_work(struct work_struct *work)
@@ -1690,7 +1758,8 @@ static void mlx5_ib_prefetch_mr_work(struct work_struct *work)
 					 w->num_sge);
 		ib_device_put(&w->dev->ib_dev);
 	}
-	put_device(&w->dev->ib_dev.dev);
+
+	num_pending_prefetch_dec(w->dev, w->sg_list, w->num_sge, 0);
 	kfree(w);
 }
 
@@ -1701,6 +1770,8 @@ int mlx5_ib_advise_mr_prefetch(struct ib_pd *pd,
 	struct mlx5_ib_dev *dev = to_mdev(pd->device);
 	u32 pf_flags = MLX5_PF_FLAGS_PREFETCH;
 	struct prefetch_mr_work *work;
+	bool valid_req;
+	int srcu_key;
 
 	if (advice == IB_UVERBS_ADVISE_MR_ADVICE_PREFETCH)
 		pf_flags |= MLX5_PF_FLAGS_DOWNGRADE;
@@ -1715,12 +1786,21 @@ int mlx5_ib_advise_mr_prefetch(struct ib_pd *pd,
 
 	memcpy(work->sg_list, sg_list, num_sge * sizeof(struct ib_sge));
 
-	get_device(&dev->ib_dev.dev);
 	work->dev = dev;
 	work->pf_flags = pf_flags;
 	work->num_sge = num_sge;
 
 	INIT_WORK(&work->work, mlx5_ib_prefetch_mr_work);
-	schedule_work(&work->work);
-	return 0;
+
+	srcu_key = srcu_read_lock(&dev->mr_srcu);
+
+	valid_req = num_pending_prefetch_inc(pd, sg_list, num_sge);
+	if (valid_req)
+		queue_work(system_unbound_wq, &work->work);
+	else
+		kfree(work);
+
+	srcu_read_unlock(&dev->mr_srcu, srcu_key);
+
+	return valid_req ? 0 : -EINVAL;
 }
-- 
cgit v1.2.3


From a2a074ef396f8738d9ee08ceefa8811381a4fe4f Mon Sep 17 00:00:00 2001
From: Leon Romanovsky <leonro@mellanox.com>
Date: Tue, 12 Feb 2019 20:39:16 +0200
Subject: RDMA: Handle ucontext allocations by IB/core

Following the PD conversion patch, do the same for ucontext allocations.

Signed-off-by: Leon Romanovsky <leonro@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 drivers/infiniband/core/device.c                |  1 +
 drivers/infiniband/core/rdma_core.c             |  9 +----
 drivers/infiniband/core/uverbs_cmd.c            | 24 +++++++-----
 drivers/infiniband/hw/bnxt_re/ib_verbs.c        | 32 +++++-----------
 drivers/infiniband/hw/bnxt_re/ib_verbs.h        |  7 ++--
 drivers/infiniband/hw/bnxt_re/main.c            |  1 +
 drivers/infiniband/hw/cxgb3/iwch_provider.c     | 17 ++++-----
 drivers/infiniband/hw/cxgb4/provider.c          | 26 +++++--------
 drivers/infiniband/hw/hns/hns_roce_main.c       | 26 +++++--------
 drivers/infiniband/hw/i40iw/i40iw_verbs.c       | 50 +++++++------------------
 drivers/infiniband/hw/mlx4/main.c               | 30 ++++++---------
 drivers/infiniband/hw/mlx5/main.c               | 35 +++++++----------
 drivers/infiniband/hw/mthca/mthca_provider.c    | 39 +++++++------------
 drivers/infiniband/hw/nes/nes_verbs.c           | 32 ++++++----------
 drivers/infiniband/hw/ocrdma/ocrdma_main.c      |  1 +
 drivers/infiniband/hw/ocrdma/ocrdma_verbs.c     | 38 +++++++------------
 drivers/infiniband/hw/ocrdma/ocrdma_verbs.h     |  5 +--
 drivers/infiniband/hw/qedr/main.c               |  1 +
 drivers/infiniband/hw/qedr/verbs.c              | 34 +++++------------
 drivers/infiniband/hw/qedr/verbs.h              |  4 +-
 drivers/infiniband/hw/usnic/usnic_ib_main.c     |  1 +
 drivers/infiniband/hw/usnic/usnic_ib_verbs.c    | 18 +++------
 drivers/infiniband/hw/usnic/usnic_ib_verbs.h    |  5 +--
 drivers/infiniband/hw/vmw_pvrdma/pvrdma_main.c  |  1 +
 drivers/infiniband/hw/vmw_pvrdma/pvrdma_verbs.c | 50 ++++++++-----------------
 drivers/infiniband/hw/vmw_pvrdma/pvrdma_verbs.h |  5 +--
 drivers/infiniband/sw/rdmavt/vt.c               | 22 ++++-------
 drivers/infiniband/sw/rxe/rxe_pool.c            |  1 +
 drivers/infiniband/sw/rxe/rxe_verbs.c           | 14 +++----
 drivers/infiniband/sw/rxe/rxe_verbs.h           |  2 +-
 include/rdma/ib_verbs.h                         |  7 ++--
 31 files changed, 198 insertions(+), 340 deletions(-)

(limited to 'drivers/infiniband/hw/mlx5/main.c')

diff --git a/drivers/infiniband/core/device.c b/drivers/infiniband/core/device.c
index bf2a215d94dd..a9f29156e486 100644
--- a/drivers/infiniband/core/device.c
+++ b/drivers/infiniband/core/device.c
@@ -1832,6 +1832,7 @@ void ib_set_device_ops(struct ib_device *dev, const struct ib_device_ops *ops)
 	SET_DEVICE_OP(dev_ops, unmap_fmr);
 
 	SET_OBJ_SIZE(dev_ops, ib_pd);
+	SET_OBJ_SIZE(dev_ops, ib_ucontext);
 }
 EXPORT_SYMBOL(ib_set_device_ops);
 
diff --git a/drivers/infiniband/core/rdma_core.c b/drivers/infiniband/core/rdma_core.c
index 96f919fe86e7..778375ff664e 100644
--- a/drivers/infiniband/core/rdma_core.c
+++ b/drivers/infiniband/core/rdma_core.c
@@ -844,7 +844,6 @@ static void ufile_destroy_ucontext(struct ib_uverbs_file *ufile,
 {
 	struct ib_ucontext *ucontext = ufile->ucontext;
 	struct ib_device *ib_dev = ucontext->device;
-	int ret;
 
 	/*
 	 * If we are closing the FD then the user mmap VMAs must have
@@ -862,12 +861,8 @@ static void ufile_destroy_ucontext(struct ib_uverbs_file *ufile,
 
 	rdma_restrack_del(&ucontext->res);
 
-	/*
-	 * FIXME: Drivers are not permitted to fail dealloc_ucontext, remove
-	 * the error return.
-	 */
-	ret = ib_dev->ops.dealloc_ucontext(ucontext);
-	WARN_ON(ret);
+	ib_dev->ops.dealloc_ucontext(ucontext);
+	kfree(ucontext);
 
 	ufile->ucontext = NULL;
 }
diff --git a/drivers/infiniband/core/uverbs_cmd.c b/drivers/infiniband/core/uverbs_cmd.c
index e44ac718f1cd..3128821ca36e 100644
--- a/drivers/infiniband/core/uverbs_cmd.c
+++ b/drivers/infiniband/core/uverbs_cmd.c
@@ -224,12 +224,13 @@ static int ib_uverbs_get_context(struct uverbs_attr_bundle *attrs)
 	if (ret)
 		goto err;
 
-	ucontext = ib_dev->ops.alloc_ucontext(ib_dev, &attrs->driver_udata);
-	if (IS_ERR(ucontext)) {
-		ret = PTR_ERR(ucontext);
+	ucontext = rdma_zalloc_drv_obj(ib_dev, ib_ucontext);
+	if (!ucontext) {
+		ret = -ENOMEM;
 		goto err_alloc;
 	}
 
+	ucontext->res.type = RDMA_RESTRACK_CTX;
 	ucontext->device = ib_dev;
 	ucontext->cg_obj = cg_obj;
 	/* ufile is required when some objects are released */
@@ -240,10 +241,6 @@ static int ib_uverbs_get_context(struct uverbs_attr_bundle *attrs)
 
 	mutex_init(&ucontext->per_mm_list_lock);
 	INIT_LIST_HEAD(&ucontext->per_mm_list);
-	if (!(ib_dev->attrs.device_cap_flags & IB_DEVICE_ON_DEMAND_PAGING))
-		ucontext->invalidate_range = NULL;
-
-	resp.num_comp_vectors = file->device->num_comp_vectors;
 
 	ret = get_unused_fd_flags(O_CLOEXEC);
 	if (ret < 0)
@@ -256,15 +253,22 @@ static int ib_uverbs_get_context(struct uverbs_attr_bundle *attrs)
 		goto err_fd;
 	}
 
+	resp.num_comp_vectors = file->device->num_comp_vectors;
+
 	ret = uverbs_response(attrs, &resp, sizeof(resp));
 	if (ret)
 		goto err_file;
 
-	fd_install(resp.async_fd, filp);
+	ret = ib_dev->ops.alloc_ucontext(ucontext, &attrs->driver_udata);
+	if (ret)
+		goto err_file;
+	if (!(ib_dev->attrs.device_cap_flags & IB_DEVICE_ON_DEMAND_PAGING))
+		ucontext->invalidate_range = NULL;
 
-	ucontext->res.type = RDMA_RESTRACK_CTX;
 	rdma_restrack_uadd(&ucontext->res);
 
+	fd_install(resp.async_fd, filp);
+
 	/*
 	 * Make sure that ib_uverbs_get_ucontext() sees the pointer update
 	 * only after all writes to setup the ucontext have completed
@@ -283,7 +287,7 @@ err_fd:
 	put_unused_fd(resp.async_fd);
 
 err_free:
-	ib_dev->ops.dealloc_ucontext(ucontext);
+	kfree(ucontext);
 
 err_alloc:
 	ib_rdmacg_uncharge(&cg_obj, ib_dev, RDMACG_RESOURCE_HCA_HANDLE);
diff --git a/drivers/infiniband/hw/bnxt_re/ib_verbs.c b/drivers/infiniband/hw/bnxt_re/ib_verbs.c
index f29f29aae537..24092911c2ac 100644
--- a/drivers/infiniband/hw/bnxt_re/ib_verbs.c
+++ b/drivers/infiniband/hw/bnxt_re/ib_verbs.c
@@ -3671,13 +3671,14 @@ free_mr:
 	return ERR_PTR(rc);
 }
 
-struct ib_ucontext *bnxt_re_alloc_ucontext(struct ib_device *ibdev,
-					   struct ib_udata *udata)
+int bnxt_re_alloc_ucontext(struct ib_ucontext *ctx, struct ib_udata *udata)
 {
+	struct ib_device *ibdev = ctx->device;
+	struct bnxt_re_ucontext *uctx =
+		container_of(ctx, struct bnxt_re_ucontext, ib_uctx);
 	struct bnxt_re_dev *rdev = to_bnxt_re_dev(ibdev, ibdev);
 	struct bnxt_qplib_dev_attr *dev_attr = &rdev->dev_attr;
 	struct bnxt_re_uctx_resp resp;
-	struct bnxt_re_ucontext *uctx;
 	u32 chip_met_rev_num = 0;
 	int rc;
 
@@ -3687,13 +3688,9 @@ struct ib_ucontext *bnxt_re_alloc_ucontext(struct ib_device *ibdev,
 	if (ibdev->uverbs_abi_ver != BNXT_RE_ABI_VERSION) {
 		dev_dbg(rdev_to_dev(rdev), " is different from the device %d ",
 			BNXT_RE_ABI_VERSION);
-		return ERR_PTR(-EPERM);
+		return -EPERM;
 	}
 
-	uctx = kzalloc(sizeof(*uctx), GFP_KERNEL);
-	if (!uctx)
-		return ERR_PTR(-ENOMEM);
-
 	uctx->rdev = rdev;
 
 	uctx->shpg = (void *)__get_free_page(GFP_KERNEL);
@@ -3727,23 +3724,21 @@ struct ib_ucontext *bnxt_re_alloc_ucontext(struct ib_device *ibdev,
 		goto cfail;
 	}
 
-	return &uctx->ib_uctx;
+	return 0;
 cfail:
 	free_page((unsigned long)uctx->shpg);
 	uctx->shpg = NULL;
 fail:
-	kfree(uctx);
-	return ERR_PTR(rc);
+	return rc;
 }
 
-int bnxt_re_dealloc_ucontext(struct ib_ucontext *ib_uctx)
+void 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);
@@ -3752,17 +3747,10 @@ int bnxt_re_dealloc_ucontext(struct ib_ucontext *ib_uctx)
 		/* 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), "Deallocate HW DPI failed!");
-			/* Don't fail, continue*/
+		bnxt_qplib_dealloc_dpi(&rdev->qplib_res,
+				       &rdev->qplib_res.dpi_tbl, &uctx->dpi);
 		uctx->dpi.dbr = NULL;
 	}
-
-	kfree(uctx);
-	return 0;
 }
 
 /* Helper function to mmap the virtual memory from user app */
diff --git a/drivers/infiniband/hw/bnxt_re/ib_verbs.h b/drivers/infiniband/hw/bnxt_re/ib_verbs.h
index c7cca803cfa3..e45465ed4eee 100644
--- a/drivers/infiniband/hw/bnxt_re/ib_verbs.h
+++ b/drivers/infiniband/hw/bnxt_re/ib_verbs.h
@@ -135,8 +135,8 @@ struct bnxt_re_mw {
 };
 
 struct bnxt_re_ucontext {
+	struct ib_ucontext      ib_uctx;
 	struct bnxt_re_dev	*rdev;
-	struct ib_ucontext	ib_uctx;
 	struct bnxt_qplib_dpi	dpi;
 	void			*shpg;
 	spinlock_t		sh_lock;	/* protect shpg */
@@ -215,9 +215,8 @@ int bnxt_re_dealloc_mw(struct ib_mw *mw);
 struct ib_mr *bnxt_re_reg_user_mr(struct ib_pd *pd, u64 start, u64 length,
 				  u64 virt_addr, int mr_access_flags,
 				  struct ib_udata *udata);
-struct ib_ucontext *bnxt_re_alloc_ucontext(struct ib_device *ibdev,
-					   struct ib_udata *udata);
-int bnxt_re_dealloc_ucontext(struct ib_ucontext *context);
+int bnxt_re_alloc_ucontext(struct ib_ucontext *ctx, struct ib_udata *udata);
+void bnxt_re_dealloc_ucontext(struct ib_ucontext *context);
 int bnxt_re_mmap(struct ib_ucontext *context, struct vm_area_struct *vma);
 
 unsigned long bnxt_re_lock_cqs(struct bnxt_re_qp *qp);
diff --git a/drivers/infiniband/hw/bnxt_re/main.c b/drivers/infiniband/hw/bnxt_re/main.c
index 0a89ef6e5754..2bd24ac45ee4 100644
--- a/drivers/infiniband/hw/bnxt_re/main.c
+++ b/drivers/infiniband/hw/bnxt_re/main.c
@@ -638,6 +638,7 @@ static const struct ib_device_ops bnxt_re_dev_ops = {
 	.reg_user_mr = bnxt_re_reg_user_mr,
 	.req_notify_cq = bnxt_re_req_notify_cq,
 	INIT_RDMA_OBJ_SIZE(ib_pd, bnxt_re_pd, ib_pd),
+	INIT_RDMA_OBJ_SIZE(ib_ucontext, bnxt_re_ucontext, ib_uctx),
 };
 
 static int bnxt_re_register_ib(struct bnxt_re_dev *rdev)
diff --git a/drivers/infiniband/hw/cxgb3/iwch_provider.c b/drivers/infiniband/hw/cxgb3/iwch_provider.c
index b74fd90a22dc..4accf7b3dcf2 100644
--- a/drivers/infiniband/hw/cxgb3/iwch_provider.c
+++ b/drivers/infiniband/hw/cxgb3/iwch_provider.c
@@ -62,7 +62,7 @@
 #include <rdma/cxgb3-abi.h>
 #include "common.h"
 
-static int iwch_dealloc_ucontext(struct ib_ucontext *context)
+static void iwch_dealloc_ucontext(struct ib_ucontext *context)
 {
 	struct iwch_dev *rhp = to_iwch_dev(context->device);
 	struct iwch_ucontext *ucontext = to_iwch_ucontext(context);
@@ -72,24 +72,20 @@ static int iwch_dealloc_ucontext(struct ib_ucontext *context)
 	list_for_each_entry_safe(mm, tmp, &ucontext->mmaps, entry)
 		kfree(mm);
 	cxio_release_ucontext(&rhp->rdev, &ucontext->uctx);
-	kfree(ucontext);
-	return 0;
 }
 
-static struct ib_ucontext *iwch_alloc_ucontext(struct ib_device *ibdev,
-					struct ib_udata *udata)
+static int iwch_alloc_ucontext(struct ib_ucontext *ucontext,
+			       struct ib_udata *udata)
 {
-	struct iwch_ucontext *context;
+	struct ib_device *ibdev = ucontext->device;
+	struct iwch_ucontext *context = to_iwch_ucontext(ucontext);
 	struct iwch_dev *rhp = to_iwch_dev(ibdev);
 
 	pr_debug("%s ibdev %p\n", __func__, ibdev);
-	context = kzalloc(sizeof(*context), GFP_KERNEL);
-	if (!context)
-		return ERR_PTR(-ENOMEM);
 	cxio_init_ucontext(&rhp->rdev, &context->uctx);
 	INIT_LIST_HEAD(&context->mmaps);
 	spin_lock_init(&context->mmap_lock);
-	return &context->ibucontext;
+	return 0;
 }
 
 static int iwch_destroy_cq(struct ib_cq *ib_cq)
@@ -1342,6 +1338,7 @@ static const struct ib_device_ops iwch_dev_ops = {
 	.req_notify_cq = iwch_arm_cq,
 	.resize_cq = iwch_resize_cq,
 	INIT_RDMA_OBJ_SIZE(ib_pd, iwch_pd, ibpd),
+	INIT_RDMA_OBJ_SIZE(ib_ucontext, iwch_ucontext, ibucontext),
 };
 
 int iwch_register_device(struct iwch_dev *dev)
diff --git a/drivers/infiniband/hw/cxgb4/provider.c b/drivers/infiniband/hw/cxgb4/provider.c
index 81fcffb597ab..507c54572cc9 100644
--- a/drivers/infiniband/hw/cxgb4/provider.c
+++ b/drivers/infiniband/hw/cxgb4/provider.c
@@ -58,7 +58,7 @@ static int fastreg_support = 1;
 module_param(fastreg_support, int, 0644);
 MODULE_PARM_DESC(fastreg_support, "Advertise fastreg support (default=1)");
 
-static int c4iw_dealloc_ucontext(struct ib_ucontext *context)
+static void c4iw_dealloc_ucontext(struct ib_ucontext *context)
 {
 	struct c4iw_ucontext *ucontext = to_c4iw_ucontext(context);
 	struct c4iw_dev *rhp;
@@ -70,26 +70,19 @@ static int c4iw_dealloc_ucontext(struct ib_ucontext *context)
 	list_for_each_entry_safe(mm, tmp, &ucontext->mmaps, entry)
 		kfree(mm);
 	c4iw_release_dev_ucontext(&rhp->rdev, &ucontext->uctx);
-	kfree(ucontext);
-	return 0;
 }
 
-static struct ib_ucontext *c4iw_alloc_ucontext(struct ib_device *ibdev,
-					       struct ib_udata *udata)
+static int c4iw_alloc_ucontext(struct ib_ucontext *ucontext,
+			       struct ib_udata *udata)
 {
-	struct c4iw_ucontext *context;
+	struct ib_device *ibdev = ucontext->device;
+	struct c4iw_ucontext *context = to_c4iw_ucontext(ucontext);
 	struct c4iw_dev *rhp = to_c4iw_dev(ibdev);
 	struct c4iw_alloc_ucontext_resp uresp;
 	int ret = 0;
 	struct c4iw_mm_entry *mm = NULL;
 
 	pr_debug("ibdev %p\n", ibdev);
-	context = kzalloc(sizeof(*context), GFP_KERNEL);
-	if (!context) {
-		ret = -ENOMEM;
-		goto err;
-	}
-
 	c4iw_init_dev_ucontext(&rhp->rdev, &context->uctx);
 	INIT_LIST_HEAD(&context->mmaps);
 	spin_lock_init(&context->mmap_lock);
@@ -101,7 +94,7 @@ static struct ib_ucontext *c4iw_alloc_ucontext(struct ib_device *ibdev,
 		mm = kmalloc(sizeof(*mm), GFP_KERNEL);
 		if (!mm) {
 			ret = -ENOMEM;
-			goto err_free;
+			goto err;
 		}
 
 		uresp.status_page_size = PAGE_SIZE;
@@ -121,13 +114,11 @@ static struct ib_ucontext *c4iw_alloc_ucontext(struct ib_device *ibdev,
 		mm->len = PAGE_SIZE;
 		insert_mmap(context, mm);
 	}
-	return &context->ibucontext;
+	return 0;
 err_mm:
 	kfree(mm);
-err_free:
-	kfree(context);
 err:
-	return ERR_PTR(ret);
+	return ret;
 }
 
 static int c4iw_mmap(struct ib_ucontext *context, struct vm_area_struct *vma)
@@ -555,6 +546,7 @@ static const struct ib_device_ops c4iw_dev_ops = {
 	.reg_user_mr = c4iw_reg_user_mr,
 	.req_notify_cq = c4iw_arm_cq,
 	INIT_RDMA_OBJ_SIZE(ib_pd, c4iw_pd, ibpd),
+	INIT_RDMA_OBJ_SIZE(ib_ucontext, c4iw_ucontext, ibucontext),
 };
 
 void c4iw_register_device(struct work_struct *work)
diff --git a/drivers/infiniband/hw/hns/hns_roce_main.c b/drivers/infiniband/hw/hns/hns_roce_main.c
index 29fb4fbba5ba..c929125da84b 100644
--- a/drivers/infiniband/hw/hns/hns_roce_main.c
+++ b/drivers/infiniband/hw/hns/hns_roce_main.c
@@ -335,23 +335,19 @@ static int hns_roce_modify_port(struct ib_device *ib_dev, u8 port_num, int mask,
 	return 0;
 }
 
-static struct ib_ucontext *hns_roce_alloc_ucontext(struct ib_device *ib_dev,
-						   struct ib_udata *udata)
+static int hns_roce_alloc_ucontext(struct ib_ucontext *uctx,
+				   struct ib_udata *udata)
 {
 	int ret = 0;
-	struct hns_roce_ucontext *context;
+	struct hns_roce_ucontext *context = to_hr_ucontext(uctx);
 	struct hns_roce_ib_alloc_ucontext_resp resp = {};
-	struct hns_roce_dev *hr_dev = to_hr_dev(ib_dev);
+	struct hns_roce_dev *hr_dev = to_hr_dev(uctx->device);
 
 	if (!hr_dev->active)
-		return ERR_PTR(-EAGAIN);
+		return -EAGAIN;
 
 	resp.qp_tab_size = hr_dev->caps.num_qps;
 
-	context = kzalloc(sizeof(*context), GFP_KERNEL);
-	if (!context)
-		return ERR_PTR(-ENOMEM);
-
 	ret = hns_roce_uar_alloc(hr_dev, &context->uar);
 	if (ret)
 		goto error_fail_uar_alloc;
@@ -365,25 +361,20 @@ static struct ib_ucontext *hns_roce_alloc_ucontext(struct ib_device *ib_dev,
 	if (ret)
 		goto error_fail_copy_to_udata;
 
-	return &context->ibucontext;
+	return 0;
 
 error_fail_copy_to_udata:
 	hns_roce_uar_free(hr_dev, &context->uar);
 
 error_fail_uar_alloc:
-	kfree(context);
-
-	return ERR_PTR(ret);
+	return ret;
 }
 
-static int hns_roce_dealloc_ucontext(struct ib_ucontext *ibcontext)
+static void hns_roce_dealloc_ucontext(struct ib_ucontext *ibcontext)
 {
 	struct hns_roce_ucontext *context = to_hr_ucontext(ibcontext);
 
 	hns_roce_uar_free(to_hr_dev(ibcontext->device), &context->uar);
-	kfree(context);
-
-	return 0;
 }
 
 static int hns_roce_mmap(struct ib_ucontext *context,
@@ -478,6 +469,7 @@ static const struct ib_device_ops hns_roce_dev_ops = {
 	.query_port = hns_roce_query_port,
 	.reg_user_mr = hns_roce_reg_user_mr,
 	INIT_RDMA_OBJ_SIZE(ib_pd, hns_roce_pd, ibpd),
+	INIT_RDMA_OBJ_SIZE(ib_ucontext, hns_roce_ucontext, ibucontext),
 };
 
 static const struct ib_device_ops hns_roce_dev_mr_ops = {
diff --git a/drivers/infiniband/hw/i40iw/i40iw_verbs.c b/drivers/infiniband/hw/i40iw/i40iw_verbs.c
index 76b4d1218696..a8352e3ca23d 100644
--- a/drivers/infiniband/hw/i40iw/i40iw_verbs.c
+++ b/drivers/infiniband/hw/i40iw/i40iw_verbs.c
@@ -121,78 +121,55 @@ static int i40iw_query_port(struct ib_device *ibdev,
 
 /**
  * i40iw_alloc_ucontext - Allocate the user context data structure
- * @ibdev: device pointer from stack
+ * @uctx: Uverbs context pointer from stack
  * @udata: user data
  *
  * This keeps track of all objects associated with a particular
  * user-mode client.
  */
-static struct ib_ucontext *i40iw_alloc_ucontext(struct ib_device *ibdev,
-						struct ib_udata *udata)
+static int i40iw_alloc_ucontext(struct ib_ucontext *uctx,
+				struct ib_udata *udata)
 {
+	struct ib_device *ibdev = uctx->device;
 	struct i40iw_device *iwdev = to_iwdev(ibdev);
 	struct i40iw_alloc_ucontext_req req;
-	struct i40iw_alloc_ucontext_resp uresp;
-	struct i40iw_ucontext *ucontext;
+	struct i40iw_alloc_ucontext_resp uresp = {};
+	struct i40iw_ucontext *ucontext = to_ucontext(uctx);
 
 	if (ib_copy_from_udata(&req, udata, sizeof(req)))
-		return ERR_PTR(-EINVAL);
+		return -EINVAL;
 
 	if (req.userspace_ver < 4 || req.userspace_ver > I40IW_ABI_VER) {
 		i40iw_pr_err("Unsupported provider library version %u.\n", req.userspace_ver);
-		return ERR_PTR(-EINVAL);
+		return -EINVAL;
 	}
 
-	memset(&uresp, 0, sizeof(uresp));
 	uresp.max_qps = iwdev->max_qp;
 	uresp.max_pds = iwdev->max_pd;
 	uresp.wq_size = iwdev->max_qp_wr * 2;
 	uresp.kernel_ver = req.userspace_ver;
 
-	ucontext = kzalloc(sizeof(*ucontext), GFP_KERNEL);
-	if (!ucontext)
-		return ERR_PTR(-ENOMEM);
-
 	ucontext->iwdev = iwdev;
 	ucontext->abi_ver = req.userspace_ver;
 
-	if (ib_copy_to_udata(udata, &uresp, sizeof(uresp))) {
-		kfree(ucontext);
-		return ERR_PTR(-EFAULT);
-	}
+	if (ib_copy_to_udata(udata, &uresp, sizeof(uresp)))
+		return -EFAULT;
 
 	INIT_LIST_HEAD(&ucontext->cq_reg_mem_list);
 	spin_lock_init(&ucontext->cq_reg_mem_list_lock);
 	INIT_LIST_HEAD(&ucontext->qp_reg_mem_list);
 	spin_lock_init(&ucontext->qp_reg_mem_list_lock);
 
-	return &ucontext->ibucontext;
+	return 0;
 }
 
 /**
  * i40iw_dealloc_ucontext - deallocate the user context data structure
  * @context: user context created during alloc
  */
-static int i40iw_dealloc_ucontext(struct ib_ucontext *context)
+static void i40iw_dealloc_ucontext(struct ib_ucontext *context)
 {
-	struct i40iw_ucontext *ucontext = to_ucontext(context);
-	unsigned long flags;
-
-	spin_lock_irqsave(&ucontext->cq_reg_mem_list_lock, flags);
-	if (!list_empty(&ucontext->cq_reg_mem_list)) {
-		spin_unlock_irqrestore(&ucontext->cq_reg_mem_list_lock, flags);
-		return -EBUSY;
-	}
-	spin_unlock_irqrestore(&ucontext->cq_reg_mem_list_lock, flags);
-	spin_lock_irqsave(&ucontext->qp_reg_mem_list_lock, flags);
-	if (!list_empty(&ucontext->qp_reg_mem_list)) {
-		spin_unlock_irqrestore(&ucontext->qp_reg_mem_list_lock, flags);
-		return -EBUSY;
-	}
-	spin_unlock_irqrestore(&ucontext->qp_reg_mem_list_lock, flags);
-
-	kfree(ucontext);
-	return 0;
+	return;
 }
 
 /**
@@ -2740,6 +2717,7 @@ static const struct ib_device_ops i40iw_dev_ops = {
 	.reg_user_mr = i40iw_reg_user_mr,
 	.req_notify_cq = i40iw_req_notify_cq,
 	INIT_RDMA_OBJ_SIZE(ib_pd, i40iw_pd, ibpd),
+	INIT_RDMA_OBJ_SIZE(ib_ucontext, i40iw_ucontext, ibucontext),
 };
 
 /**
diff --git a/drivers/infiniband/hw/mlx4/main.c b/drivers/infiniband/hw/mlx4/main.c
index c0f6aea7ed7c..733f7bbd5901 100644
--- a/drivers/infiniband/hw/mlx4/main.c
+++ b/drivers/infiniband/hw/mlx4/main.c
@@ -1076,17 +1076,18 @@ out:
 	return err;
 }
 
-static struct ib_ucontext *mlx4_ib_alloc_ucontext(struct ib_device *ibdev,
-						  struct ib_udata *udata)
+static int mlx4_ib_alloc_ucontext(struct ib_ucontext *uctx,
+				  struct ib_udata *udata)
 {
+	struct ib_device *ibdev = uctx->device;
 	struct mlx4_ib_dev *dev = to_mdev(ibdev);
-	struct mlx4_ib_ucontext *context;
+	struct mlx4_ib_ucontext *context = to_mucontext(uctx);
 	struct mlx4_ib_alloc_ucontext_resp_v3 resp_v3;
 	struct mlx4_ib_alloc_ucontext_resp resp;
 	int err;
 
 	if (!dev->ib_active)
-		return ERR_PTR(-EAGAIN);
+		return -EAGAIN;
 
 	if (ibdev->uverbs_abi_ver == MLX4_IB_UVERBS_NO_DEV_CAPS_ABI_VERSION) {
 		resp_v3.qp_tab_size      = dev->dev->caps.num_qps;
@@ -1100,15 +1101,9 @@ static struct ib_ucontext *mlx4_ib_alloc_ucontext(struct ib_device *ibdev,
 		resp.cqe_size	      = dev->dev->caps.cqe_size;
 	}
 
-	context = kzalloc(sizeof(*context), GFP_KERNEL);
-	if (!context)
-		return ERR_PTR(-ENOMEM);
-
 	err = mlx4_uar_alloc(to_mdev(ibdev)->dev, &context->uar);
-	if (err) {
-		kfree(context);
-		return ERR_PTR(err);
-	}
+	if (err)
+		return err;
 
 	INIT_LIST_HEAD(&context->db_page_list);
 	mutex_init(&context->db_page_mutex);
@@ -1123,21 +1118,17 @@ static struct ib_ucontext *mlx4_ib_alloc_ucontext(struct ib_device *ibdev,
 
 	if (err) {
 		mlx4_uar_free(to_mdev(ibdev)->dev, &context->uar);
-		kfree(context);
-		return ERR_PTR(-EFAULT);
+		return -EFAULT;
 	}
 
-	return &context->ibucontext;
+	return err;
 }
 
-static int mlx4_ib_dealloc_ucontext(struct ib_ucontext *ibcontext)
+static void mlx4_ib_dealloc_ucontext(struct ib_ucontext *ibcontext)
 {
 	struct mlx4_ib_ucontext *context = to_mucontext(ibcontext);
 
 	mlx4_uar_free(to_mdev(ibcontext->device)->dev, &context->uar);
-	kfree(context);
-
-	return 0;
 }
 
 static void mlx4_ib_disassociate_ucontext(struct ib_ucontext *ibcontext)
@@ -2570,6 +2561,7 @@ static const struct ib_device_ops mlx4_ib_dev_ops = {
 	.rereg_user_mr = mlx4_ib_rereg_user_mr,
 	.resize_cq = mlx4_ib_resize_cq,
 	INIT_RDMA_OBJ_SIZE(ib_pd, mlx4_ib_pd, ibpd),
+	INIT_RDMA_OBJ_SIZE(ib_ucontext, mlx4_ib_ucontext, ibucontext),
 };
 
 static const struct ib_device_ops mlx4_ib_dev_wq_ops = {
diff --git a/drivers/infiniband/hw/mlx5/main.c b/drivers/infiniband/hw/mlx5/main.c
index 614b2acbc621..994c19d01211 100644
--- a/drivers/infiniband/hw/mlx5/main.c
+++ b/drivers/infiniband/hw/mlx5/main.c
@@ -1745,14 +1745,15 @@ static void mlx5_ib_dealloc_transport_domain(struct mlx5_ib_dev *dev, u32 tdn,
 	mlx5_ib_disable_lb(dev, true, false);
 }
 
-static struct ib_ucontext *mlx5_ib_alloc_ucontext(struct ib_device *ibdev,
-						  struct ib_udata *udata)
+static int mlx5_ib_alloc_ucontext(struct ib_ucontext *uctx,
+				  struct ib_udata *udata)
 {
+	struct ib_device *ibdev = uctx->device;
 	struct mlx5_ib_dev *dev = to_mdev(ibdev);
 	struct mlx5_ib_alloc_ucontext_req_v2 req = {};
 	struct mlx5_ib_alloc_ucontext_resp resp = {};
 	struct mlx5_core_dev *mdev = dev->mdev;
-	struct mlx5_ib_ucontext *context;
+	struct mlx5_ib_ucontext *context = to_mucontext(uctx);
 	struct mlx5_bfreg_info *bfregi;
 	int ver;
 	int err;
@@ -1762,29 +1763,29 @@ static struct ib_ucontext *mlx5_ib_alloc_ucontext(struct ib_device *ibdev,
 	bool lib_uar_4k;
 
 	if (!dev->ib_active)
-		return ERR_PTR(-EAGAIN);
+		return -EAGAIN;
 
 	if (udata->inlen == sizeof(struct mlx5_ib_alloc_ucontext_req))
 		ver = 0;
 	else if (udata->inlen >= min_req_v2)
 		ver = 2;
 	else
-		return ERR_PTR(-EINVAL);
+		return -EINVAL;
 
 	err = ib_copy_from_udata(&req, udata, min(udata->inlen, sizeof(req)));
 	if (err)
-		return ERR_PTR(err);
+		return err;
 
 	if (req.flags & ~MLX5_IB_ALLOC_UCTX_DEVX)
-		return ERR_PTR(-EOPNOTSUPP);
+		return -EOPNOTSUPP;
 
 	if (req.comp_mask || req.reserved0 || req.reserved1 || req.reserved2)
-		return ERR_PTR(-EOPNOTSUPP);
+		return -EOPNOTSUPP;
 
 	req.total_num_bfregs = ALIGN(req.total_num_bfregs,
 				    MLX5_NON_FP_BFREGS_PER_UAR);
 	if (req.num_low_latency_bfregs > req.total_num_bfregs - 1)
-		return ERR_PTR(-EINVAL);
+		return -EINVAL;
 
 	resp.qp_tab_size = 1 << MLX5_CAP_GEN(dev->mdev, log_max_qp);
 	if (mlx5_core_is_pf(dev->mdev) && MLX5_CAP_GEN(dev->mdev, bf))
@@ -1817,10 +1818,6 @@ static struct ib_ucontext *mlx5_ib_alloc_ucontext(struct ib_device *ibdev,
 		/* MLX5_USER_ALLOC_UCONTEXT_FLOW_ACTION_FLAGS_ESP_AES_GCM_FULL_OFFLOAD is currently always 0 */
 	}
 
-	context = kzalloc(sizeof(*context), GFP_KERNEL);
-	if (!context)
-		return ERR_PTR(-ENOMEM);
-
 	lib_uar_4k = req.lib_caps & MLX5_LIB_CAP_4K_UAR;
 	bfregi = &context->bfregi;
 
@@ -1955,7 +1952,7 @@ static struct ib_ucontext *mlx5_ib_alloc_ucontext(struct ib_device *ibdev,
 				   1, &dev->roce[port].tx_port_affinity));
 	}
 
-	return &context->ibucontext;
+	return 0;
 
 out_mdev:
 	mlx5_ib_dealloc_transport_domain(dev, context->tdn, context->devx_uid);
@@ -1973,12 +1970,10 @@ out_count:
 	kfree(bfregi->count);
 
 out_ctx:
-	kfree(context);
-
-	return ERR_PTR(err);
+	return err;
 }
 
-static int mlx5_ib_dealloc_ucontext(struct ib_ucontext *ibcontext)
+static void mlx5_ib_dealloc_ucontext(struct ib_ucontext *ibcontext)
 {
 	struct mlx5_ib_ucontext *context = to_mucontext(ibcontext);
 	struct mlx5_ib_dev *dev = to_mdev(ibcontext->device);
@@ -1998,9 +1993,6 @@ static int mlx5_ib_dealloc_ucontext(struct ib_ucontext *ibcontext)
 	deallocate_uars(dev, context);
 	kfree(bfregi->sys_pages);
 	kfree(bfregi->count);
-	kfree(context);
-
-	return 0;
 }
 
 static phys_addr_t uar_index2pfn(struct mlx5_ib_dev *dev,
@@ -5984,6 +5976,7 @@ static const struct ib_device_ops mlx5_ib_dev_ops = {
 	.rereg_user_mr = mlx5_ib_rereg_user_mr,
 	.resize_cq = mlx5_ib_resize_cq,
 	INIT_RDMA_OBJ_SIZE(ib_pd, mlx5_ib_pd, ibpd),
+	INIT_RDMA_OBJ_SIZE(ib_ucontext, mlx5_ib_ucontext, ibucontext),
 };
 
 static const struct ib_device_ops mlx5_ib_dev_flow_ipsec_ops = {
diff --git a/drivers/infiniband/hw/mthca/mthca_provider.c b/drivers/infiniband/hw/mthca/mthca_provider.c
index 80c3af217d96..d063d7a37762 100644
--- a/drivers/infiniband/hw/mthca/mthca_provider.c
+++ b/drivers/infiniband/hw/mthca/mthca_provider.c
@@ -301,17 +301,16 @@ static int mthca_query_gid(struct ib_device *ibdev, u8 port,
 	return err;
 }
 
-static struct ib_ucontext *mthca_alloc_ucontext(struct ib_device *ibdev,
-						struct ib_udata *udata)
+static int mthca_alloc_ucontext(struct ib_ucontext *uctx,
+				struct ib_udata *udata)
 {
-	struct mthca_alloc_ucontext_resp uresp;
-	struct mthca_ucontext           *context;
+	struct ib_device *ibdev = uctx->device;
+	struct mthca_alloc_ucontext_resp uresp = {};
+	struct mthca_ucontext *context = to_mucontext(uctx);
 	int                              err;
 
 	if (!(to_mdev(ibdev)->active))
-		return ERR_PTR(-EAGAIN);
-
-	memset(&uresp, 0, sizeof uresp);
+		return -EAGAIN;
 
 	uresp.qp_tab_size = to_mdev(ibdev)->limits.num_qps;
 	if (mthca_is_memfree(to_mdev(ibdev)))
@@ -319,44 +318,33 @@ static struct ib_ucontext *mthca_alloc_ucontext(struct ib_device *ibdev,
 	else
 		uresp.uarc_size = 0;
 
-	context = kzalloc(sizeof(*context), GFP_KERNEL);
-	if (!context)
-		return ERR_PTR(-ENOMEM);
-
 	err = mthca_uar_alloc(to_mdev(ibdev), &context->uar);
-	if (err) {
-		kfree(context);
-		return ERR_PTR(err);
-	}
+	if (err)
+		return err;
 
 	context->db_tab = mthca_init_user_db_tab(to_mdev(ibdev));
 	if (IS_ERR(context->db_tab)) {
 		err = PTR_ERR(context->db_tab);
 		mthca_uar_free(to_mdev(ibdev), &context->uar);
-		kfree(context);
-		return ERR_PTR(err);
+		return err;
 	}
 
-	if (ib_copy_to_udata(udata, &uresp, sizeof uresp)) {
+	if (ib_copy_to_udata(udata, &uresp, sizeof(uresp))) {
 		mthca_cleanup_user_db_tab(to_mdev(ibdev), &context->uar, context->db_tab);
 		mthca_uar_free(to_mdev(ibdev), &context->uar);
-		kfree(context);
-		return ERR_PTR(-EFAULT);
+		return -EFAULT;
 	}
 
 	context->reg_mr_warned = 0;
 
-	return &context->ibucontext;
+	return 0;
 }
 
-static int mthca_dealloc_ucontext(struct ib_ucontext *context)
+static void mthca_dealloc_ucontext(struct ib_ucontext *context)
 {
 	mthca_cleanup_user_db_tab(to_mdev(context->device), &to_mucontext(context)->uar,
 				  to_mucontext(context)->db_tab);
 	mthca_uar_free(to_mdev(context->device), &to_mucontext(context)->uar);
-	kfree(to_mucontext(context));
-
-	return 0;
 }
 
 static int mthca_mmap_uar(struct ib_ucontext *context,
@@ -1213,6 +1201,7 @@ static const struct ib_device_ops mthca_dev_ops = {
 	.reg_user_mr = mthca_reg_user_mr,
 	.resize_cq = mthca_resize_cq,
 	INIT_RDMA_OBJ_SIZE(ib_pd, mthca_pd, ibpd),
+	INIT_RDMA_OBJ_SIZE(ib_ucontext, mthca_ucontext, ibucontext),
 };
 
 static const struct ib_device_ops mthca_dev_arbel_srq_ops = {
diff --git a/drivers/infiniband/hw/nes/nes_verbs.c b/drivers/infiniband/hw/nes/nes_verbs.c
index b23956aa45b8..828e4af3f951 100644
--- a/drivers/infiniband/hw/nes/nes_verbs.c
+++ b/drivers/infiniband/hw/nes/nes_verbs.c
@@ -529,27 +529,27 @@ static int nes_query_gid(struct ib_device *ibdev, u8 port,
  * nes_alloc_ucontext - Allocate the user context data structure. This keeps track
  * of all objects associated with a particular user-mode client.
  */
-static struct ib_ucontext *nes_alloc_ucontext(struct ib_device *ibdev,
-		struct ib_udata *udata)
+static int nes_alloc_ucontext(struct ib_ucontext *uctx, struct ib_udata *udata)
 {
+	struct ib_device *ibdev = uctx->device;
 	struct nes_vnic *nesvnic = to_nesvnic(ibdev);
 	struct nes_device *nesdev = nesvnic->nesdev;
 	struct nes_adapter *nesadapter = nesdev->nesadapter;
 	struct nes_alloc_ucontext_req req;
 	struct nes_alloc_ucontext_resp uresp = {};
-	struct nes_ucontext *nes_ucontext;
+	struct nes_ucontext *nes_ucontext = to_nesucontext(uctx);
 	struct nes_ib_device *nesibdev = nesvnic->nesibdev;
 
 
 	if (ib_copy_from_udata(&req, udata, sizeof(struct nes_alloc_ucontext_req))) {
 		printk(KERN_ERR PFX "Invalid structure size on allocate user context.\n");
-		return ERR_PTR(-EINVAL);
+		return -EINVAL;
 	}
 
 	if (req.userspace_ver != NES_ABI_USERSPACE_VER) {
 		printk(KERN_ERR PFX "Invalid userspace driver version detected. Detected version %d, should be %d\n",
 			req.userspace_ver, NES_ABI_USERSPACE_VER);
-		return ERR_PTR(-EINVAL);
+		return -EINVAL;
 	}
 
 
@@ -559,10 +559,6 @@ static struct ib_ucontext *nes_alloc_ucontext(struct ib_device *ibdev,
 	uresp.virtwq = nesadapter->virtwq;
 	uresp.kernel_ver = NES_ABI_KERNEL_VER;
 
-	nes_ucontext = kzalloc(sizeof *nes_ucontext, GFP_KERNEL);
-	if (!nes_ucontext)
-		return ERR_PTR(-ENOMEM);
-
 	nes_ucontext->nesdev = nesdev;
 	nes_ucontext->mmap_wq_offset = uresp.max_pds;
 	nes_ucontext->mmap_cq_offset = nes_ucontext->mmap_wq_offset +
@@ -570,29 +566,22 @@ static struct ib_ucontext *nes_alloc_ucontext(struct ib_device *ibdev,
 			PAGE_SIZE;
 
 
-	if (ib_copy_to_udata(udata, &uresp, sizeof uresp)) {
-		kfree(nes_ucontext);
-		return ERR_PTR(-EFAULT);
-	}
+	if (ib_copy_to_udata(udata, &uresp, sizeof(uresp)))
+		return -EFAULT;
 
 	INIT_LIST_HEAD(&nes_ucontext->cq_reg_mem_list);
 	INIT_LIST_HEAD(&nes_ucontext->qp_reg_mem_list);
-	return &nes_ucontext->ibucontext;
+	return 0;
 }
 
-
 /**
  * nes_dealloc_ucontext
  */
-static int nes_dealloc_ucontext(struct ib_ucontext *context)
+static void nes_dealloc_ucontext(struct ib_ucontext *context)
 {
-	struct nes_ucontext *nes_ucontext = to_nesucontext(context);
-
-	kfree(nes_ucontext);
-	return 0;
+	return;
 }
 
-
 /**
  * nes_mmap
  */
@@ -3599,6 +3588,7 @@ static const struct ib_device_ops nes_dev_ops = {
 	.reg_user_mr = nes_reg_user_mr,
 	.req_notify_cq = nes_req_notify_cq,
 	INIT_RDMA_OBJ_SIZE(ib_pd, nes_pd, ibpd),
+	INIT_RDMA_OBJ_SIZE(ib_ucontext, nes_ucontext, ibucontext),
 };
 
 /**
diff --git a/drivers/infiniband/hw/ocrdma/ocrdma_main.c b/drivers/infiniband/hw/ocrdma/ocrdma_main.c
index 0de83c92691f..b9e10d55a58e 100644
--- a/drivers/infiniband/hw/ocrdma/ocrdma_main.c
+++ b/drivers/infiniband/hw/ocrdma/ocrdma_main.c
@@ -180,6 +180,7 @@ static const struct ib_device_ops ocrdma_dev_ops = {
 	.req_notify_cq = ocrdma_arm_cq,
 	.resize_cq = ocrdma_resize_cq,
 	INIT_RDMA_OBJ_SIZE(ib_pd, ocrdma_pd, ibpd),
+	INIT_RDMA_OBJ_SIZE(ib_ucontext, ocrdma_ucontext, ibucontext),
 };
 
 static const struct ib_device_ops ocrdma_dev_srq_ops = {
diff --git a/drivers/infiniband/hw/ocrdma/ocrdma_verbs.c b/drivers/infiniband/hw/ocrdma/ocrdma_verbs.c
index ed5da67b693d..b4e1777c2c97 100644
--- a/drivers/infiniband/hw/ocrdma/ocrdma_verbs.c
+++ b/drivers/infiniband/hw/ocrdma/ocrdma_verbs.c
@@ -440,7 +440,7 @@ err:
 	return status;
 }
 
-static int ocrdma_dealloc_ucontext_pd(struct ocrdma_ucontext *uctx)
+static void ocrdma_dealloc_ucontext_pd(struct ocrdma_ucontext *uctx)
 {
 	struct ocrdma_pd *pd = uctx->cntxt_pd;
 	struct ocrdma_dev *dev = get_ocrdma_dev(pd->ibpd.device);
@@ -451,8 +451,7 @@ static int ocrdma_dealloc_ucontext_pd(struct ocrdma_ucontext *uctx)
 	}
 	kfree(uctx->cntxt_pd);
 	uctx->cntxt_pd = NULL;
-	(void)_ocrdma_dealloc_pd(dev, pd);
-	return 0;
+	_ocrdma_dealloc_pd(dev, pd);
 }
 
 static struct ocrdma_pd *ocrdma_get_ucontext_pd(struct ocrdma_ucontext *uctx)
@@ -476,33 +475,28 @@ static void ocrdma_release_ucontext_pd(struct ocrdma_ucontext *uctx)
 	mutex_unlock(&uctx->mm_list_lock);
 }
 
-struct ib_ucontext *ocrdma_alloc_ucontext(struct ib_device *ibdev,
-					  struct ib_udata *udata)
+int ocrdma_alloc_ucontext(struct ib_ucontext *uctx, struct ib_udata *udata)
 {
+	struct ib_device *ibdev = uctx->device;
 	int status;
-	struct ocrdma_ucontext *ctx;
-	struct ocrdma_alloc_ucontext_resp resp;
+	struct ocrdma_ucontext *ctx = get_ocrdma_ucontext(uctx);
+	struct ocrdma_alloc_ucontext_resp resp = {};
 	struct ocrdma_dev *dev = get_ocrdma_dev(ibdev);
 	struct pci_dev *pdev = dev->nic_info.pdev;
 	u32 map_len = roundup(sizeof(u32) * 2048, PAGE_SIZE);
 
 	if (!udata)
-		return ERR_PTR(-EFAULT);
-	ctx = kzalloc(sizeof(*ctx), GFP_KERNEL);
-	if (!ctx)
-		return ERR_PTR(-ENOMEM);
+		return -EFAULT;
 	INIT_LIST_HEAD(&ctx->mm_head);
 	mutex_init(&ctx->mm_list_lock);
 
 	ctx->ah_tbl.va = dma_alloc_coherent(&pdev->dev, map_len,
 					    &ctx->ah_tbl.pa, GFP_KERNEL);
-	if (!ctx->ah_tbl.va) {
-		kfree(ctx);
-		return ERR_PTR(-ENOMEM);
-	}
+	if (!ctx->ah_tbl.va)
+		return -ENOMEM;
+
 	ctx->ah_tbl.len = map_len;
 
-	memset(&resp, 0, sizeof(resp));
 	resp.ah_tbl_len = ctx->ah_tbl.len;
 	resp.ah_tbl_page = virt_to_phys(ctx->ah_tbl.va);
 
@@ -524,7 +518,7 @@ struct ib_ucontext *ocrdma_alloc_ucontext(struct ib_device *ibdev,
 	status = ib_copy_to_udata(udata, &resp, sizeof(resp));
 	if (status)
 		goto cpy_err;
-	return &ctx->ibucontext;
+	return 0;
 
 cpy_err:
 	ocrdma_dealloc_ucontext_pd(ctx);
@@ -533,19 +527,17 @@ pd_err:
 map_err:
 	dma_free_coherent(&pdev->dev, ctx->ah_tbl.len, ctx->ah_tbl.va,
 			  ctx->ah_tbl.pa);
-	kfree(ctx);
-	return ERR_PTR(status);
+	return status;
 }
 
-int ocrdma_dealloc_ucontext(struct ib_ucontext *ibctx)
+void ocrdma_dealloc_ucontext(struct ib_ucontext *ibctx)
 {
-	int status;
 	struct ocrdma_mm *mm, *tmp;
 	struct ocrdma_ucontext *uctx = get_ocrdma_ucontext(ibctx);
 	struct ocrdma_dev *dev = get_ocrdma_dev(ibctx->device);
 	struct pci_dev *pdev = dev->nic_info.pdev;
 
-	status = ocrdma_dealloc_ucontext_pd(uctx);
+	ocrdma_dealloc_ucontext_pd(uctx);
 
 	ocrdma_del_mmap(uctx, uctx->ah_tbl.pa, uctx->ah_tbl.len);
 	dma_free_coherent(&pdev->dev, uctx->ah_tbl.len, uctx->ah_tbl.va,
@@ -555,8 +547,6 @@ int ocrdma_dealloc_ucontext(struct ib_ucontext *ibctx)
 		list_del(&mm->entry);
 		kfree(mm);
 	}
-	kfree(uctx);
-	return status;
 }
 
 int ocrdma_mmap(struct ib_ucontext *context, struct vm_area_struct *vma)
diff --git a/drivers/infiniband/hw/ocrdma/ocrdma_verbs.h b/drivers/infiniband/hw/ocrdma/ocrdma_verbs.h
index 1fd66721c930..4c04ab40798e 100644
--- a/drivers/infiniband/hw/ocrdma/ocrdma_verbs.h
+++ b/drivers/infiniband/hw/ocrdma/ocrdma_verbs.h
@@ -64,9 +64,8 @@ void ocrdma_get_guid(struct ocrdma_dev *, u8 *guid);
 struct net_device *ocrdma_get_netdev(struct ib_device *device, u8 port_num);
 int ocrdma_query_pkey(struct ib_device *, u8 port, u16 index, u16 *pkey);
 
-struct ib_ucontext *ocrdma_alloc_ucontext(struct ib_device *,
-					  struct ib_udata *);
-int ocrdma_dealloc_ucontext(struct ib_ucontext *);
+int ocrdma_alloc_ucontext(struct ib_ucontext *uctx, struct ib_udata *udata);
+void ocrdma_dealloc_ucontext(struct ib_ucontext *uctx);
 
 int ocrdma_mmap(struct ib_ucontext *, struct vm_area_struct *vma);
 
diff --git a/drivers/infiniband/hw/qedr/main.c b/drivers/infiniband/hw/qedr/main.c
index 44ce4989dcef..996d9ecd93e0 100644
--- a/drivers/infiniband/hw/qedr/main.c
+++ b/drivers/infiniband/hw/qedr/main.c
@@ -240,6 +240,7 @@ static const struct ib_device_ops qedr_dev_ops = {
 	.req_notify_cq = qedr_arm_cq,
 	.resize_cq = qedr_resize_cq,
 	INIT_RDMA_OBJ_SIZE(ib_pd, qedr_pd, ibpd),
+	INIT_RDMA_OBJ_SIZE(ib_ucontext, qedr_ucontext, ibucontext),
 };
 
 static int qedr_register_device(struct qedr_dev *dev)
diff --git a/drivers/infiniband/hw/qedr/verbs.c b/drivers/infiniband/hw/qedr/verbs.c
index d51bc3ede9d1..59ad4202422c 100644
--- a/drivers/infiniband/hw/qedr/verbs.c
+++ b/drivers/infiniband/hw/qedr/verbs.c
@@ -316,28 +316,24 @@ static bool qedr_search_mmap(struct qedr_ucontext *uctx, u64 phy_addr,
 	return found;
 }
 
-struct ib_ucontext *qedr_alloc_ucontext(struct ib_device *ibdev,
-					struct ib_udata *udata)
+int qedr_alloc_ucontext(struct ib_ucontext *uctx, struct ib_udata *udata)
 {
+	struct ib_device *ibdev = uctx->device;
 	int rc;
-	struct qedr_ucontext *ctx;
-	struct qedr_alloc_ucontext_resp uresp;
+	struct qedr_ucontext *ctx = get_qedr_ucontext(uctx);
+	struct qedr_alloc_ucontext_resp uresp = {};
 	struct qedr_dev *dev = get_qedr_dev(ibdev);
 	struct qed_rdma_add_user_out_params oparams;
 
 	if (!udata)
-		return ERR_PTR(-EFAULT);
-
-	ctx = kzalloc(sizeof(*ctx), GFP_KERNEL);
-	if (!ctx)
-		return ERR_PTR(-ENOMEM);
+		return -EFAULT;
 
 	rc = dev->ops->rdma_add_user(dev->rdma_ctx, &oparams);
 	if (rc) {
 		DP_ERR(dev,
 		       "failed to allocate a DPI for a new RoCE application, rc=%d. To overcome this consider to increase the number of DPIs, increase the doorbell BAR size or just close unnecessary RoCE applications. In order to increase the number of DPIs consult the qedr readme\n",
 		       rc);
-		goto err;
+		return rc;
 	}
 
 	ctx->dpi = oparams.dpi;
@@ -347,8 +343,6 @@ struct ib_ucontext *qedr_alloc_ucontext(struct ib_device *ibdev,
 	INIT_LIST_HEAD(&ctx->mm_head);
 	mutex_init(&ctx->mm_list_lock);
 
-	memset(&uresp, 0, sizeof(uresp));
-
 	uresp.dpm_enabled = dev->user_dpm_enabled;
 	uresp.wids_enabled = 1;
 	uresp.wid_count = oparams.wid_count;
@@ -364,28 +358,23 @@ struct ib_ucontext *qedr_alloc_ucontext(struct ib_device *ibdev,
 
 	rc = qedr_ib_copy_to_udata(udata, &uresp, sizeof(uresp));
 	if (rc)
-		goto err;
+		return rc;
 
 	ctx->dev = dev;
 
 	rc = qedr_add_mmap(ctx, ctx->dpi_phys_addr, ctx->dpi_size);
 	if (rc)
-		goto err;
+		return rc;
 
 	DP_DEBUG(dev, QEDR_MSG_INIT, "Allocating user context %p\n",
 		 &ctx->ibucontext);
-	return &ctx->ibucontext;
-
-err:
-	kfree(ctx);
-	return ERR_PTR(rc);
+	return 0;
 }
 
-int qedr_dealloc_ucontext(struct ib_ucontext *ibctx)
+void qedr_dealloc_ucontext(struct ib_ucontext *ibctx)
 {
 	struct qedr_ucontext *uctx = get_qedr_ucontext(ibctx);
 	struct qedr_mm *mm, *tmp;
-	int status = 0;
 
 	DP_DEBUG(uctx->dev, QEDR_MSG_INIT, "Deallocating user context %p\n",
 		 uctx);
@@ -398,9 +387,6 @@ int qedr_dealloc_ucontext(struct ib_ucontext *ibctx)
 		list_del(&mm->entry);
 		kfree(mm);
 	}
-
-	kfree(uctx);
-	return status;
 }
 
 int qedr_mmap(struct ib_ucontext *context, struct vm_area_struct *vma)
diff --git a/drivers/infiniband/hw/qedr/verbs.h b/drivers/infiniband/hw/qedr/verbs.h
index 97a6ff3f9afb..f0c05f4771ac 100644
--- a/drivers/infiniband/hw/qedr/verbs.h
+++ b/drivers/infiniband/hw/qedr/verbs.h
@@ -43,8 +43,8 @@ int qedr_iw_query_gid(struct ib_device *ibdev, u8 port,
 
 int qedr_query_pkey(struct ib_device *, u8 port, u16 index, u16 *pkey);
 
-struct ib_ucontext *qedr_alloc_ucontext(struct ib_device *, struct ib_udata *);
-int qedr_dealloc_ucontext(struct ib_ucontext *);
+int qedr_alloc_ucontext(struct ib_ucontext *uctx, struct ib_udata *udata);
+void qedr_dealloc_ucontext(struct ib_ucontext *uctx);
 
 int qedr_mmap(struct ib_ucontext *, struct vm_area_struct *vma);
 int qedr_alloc_pd(struct ib_pd *pd, struct ib_ucontext *uctx,
diff --git a/drivers/infiniband/hw/usnic/usnic_ib_main.c b/drivers/infiniband/hw/usnic/usnic_ib_main.c
index be6468021a9a..d88d9f8a7f9a 100644
--- a/drivers/infiniband/hw/usnic/usnic_ib_main.c
+++ b/drivers/infiniband/hw/usnic/usnic_ib_main.c
@@ -350,6 +350,7 @@ static const struct ib_device_ops usnic_dev_ops = {
 	.query_qp = usnic_ib_query_qp,
 	.reg_user_mr = usnic_ib_reg_mr,
 	INIT_RDMA_OBJ_SIZE(ib_pd, usnic_ib_pd, ibpd),
+	INIT_RDMA_OBJ_SIZE(ib_ucontext, usnic_ib_ucontext, ibucontext),
 };
 
 /* Start of PF discovery section */
diff --git a/drivers/infiniband/hw/usnic/usnic_ib_verbs.c b/drivers/infiniband/hw/usnic/usnic_ib_verbs.c
index 7549ae23027e..bd4521b2cc5f 100644
--- a/drivers/infiniband/hw/usnic/usnic_ib_verbs.c
+++ b/drivers/infiniband/hw/usnic/usnic_ib_verbs.c
@@ -653,37 +653,31 @@ int usnic_ib_dereg_mr(struct ib_mr *ibmr)
 	return 0;
 }
 
-struct ib_ucontext *usnic_ib_alloc_ucontext(struct ib_device *ibdev,
-							struct ib_udata *udata)
+int usnic_ib_alloc_ucontext(struct ib_ucontext *uctx, struct ib_udata *udata)
 {
-	struct usnic_ib_ucontext *context;
+	struct ib_device *ibdev = uctx->device;
+	struct usnic_ib_ucontext *context = to_ucontext(uctx);
 	struct usnic_ib_dev *us_ibdev = to_usdev(ibdev);
 	usnic_dbg("\n");
 
-	context = kzalloc(sizeof(*context), GFP_KERNEL);
-	if (!context)
-		return ERR_PTR(-ENOMEM);
-
 	INIT_LIST_HEAD(&context->qp_grp_list);
 	mutex_lock(&us_ibdev->usdev_lock);
 	list_add_tail(&context->link, &us_ibdev->ctx_list);
 	mutex_unlock(&us_ibdev->usdev_lock);
 
-	return &context->ibucontext;
+	return 0;
 }
 
-int usnic_ib_dealloc_ucontext(struct ib_ucontext *ibcontext)
+void usnic_ib_dealloc_ucontext(struct ib_ucontext *ibcontext)
 {
 	struct usnic_ib_ucontext *context = to_uucontext(ibcontext);
 	struct usnic_ib_dev *us_ibdev = to_usdev(ibcontext->device);
 	usnic_dbg("\n");
 
 	mutex_lock(&us_ibdev->usdev_lock);
-	BUG_ON(!list_empty(&context->qp_grp_list));
+	WARN_ON_ONCE(!list_empty(&context->qp_grp_list));
 	list_del(&context->link);
 	mutex_unlock(&us_ibdev->usdev_lock);
-	kfree(context);
-	return 0;
 }
 
 int usnic_ib_mmap(struct ib_ucontext *context,
diff --git a/drivers/infiniband/hw/usnic/usnic_ib_verbs.h b/drivers/infiniband/hw/usnic/usnic_ib_verbs.h
index 2a87650949f6..c40e89b6246f 100644
--- a/drivers/infiniband/hw/usnic/usnic_ib_verbs.h
+++ b/drivers/infiniband/hw/usnic/usnic_ib_verbs.h
@@ -68,9 +68,8 @@ struct ib_mr *usnic_ib_reg_mr(struct ib_pd *pd, u64 start, u64 length,
 				u64 virt_addr, int access_flags,
 				struct ib_udata *udata);
 int usnic_ib_dereg_mr(struct ib_mr *ibmr);
-struct ib_ucontext *usnic_ib_alloc_ucontext(struct ib_device *ibdev,
-						struct ib_udata *udata);
-int usnic_ib_dealloc_ucontext(struct ib_ucontext *ibcontext);
+int usnic_ib_alloc_ucontext(struct ib_ucontext *uctx, struct ib_udata *udata);
+void usnic_ib_dealloc_ucontext(struct ib_ucontext *ibcontext);
 int usnic_ib_mmap(struct ib_ucontext *context,
 			struct vm_area_struct *vma);
 #endif /* !USNIC_IB_VERBS_H */
diff --git a/drivers/infiniband/hw/vmw_pvrdma/pvrdma_main.c b/drivers/infiniband/hw/vmw_pvrdma/pvrdma_main.c
index 47e653d2495c..6d8b3e0de57a 100644
--- a/drivers/infiniband/hw/vmw_pvrdma/pvrdma_main.c
+++ b/drivers/infiniband/hw/vmw_pvrdma/pvrdma_main.c
@@ -196,6 +196,7 @@ static const struct ib_device_ops pvrdma_dev_ops = {
 	.reg_user_mr = pvrdma_reg_user_mr,
 	.req_notify_cq = pvrdma_req_notify_cq,
 	INIT_RDMA_OBJ_SIZE(ib_pd, pvrdma_pd, ibpd),
+	INIT_RDMA_OBJ_SIZE(ib_ucontext, pvrdma_ucontext, ibucontext),
 };
 
 static const struct ib_device_ops pvrdma_dev_srq_ops = {
diff --git a/drivers/infiniband/hw/vmw_pvrdma/pvrdma_verbs.c b/drivers/infiniband/hw/vmw_pvrdma/pvrdma_verbs.c
index f44220f72e05..42fe821f8d58 100644
--- a/drivers/infiniband/hw/vmw_pvrdma/pvrdma_verbs.c
+++ b/drivers/infiniband/hw/vmw_pvrdma/pvrdma_verbs.c
@@ -306,41 +306,32 @@ out:
 
 /**
  * pvrdma_alloc_ucontext - allocate ucontext
- * @ibdev: the IB device
+ * @uctx: the uverbs countext
  * @udata: user data
  *
- * @return: the ib_ucontext pointer on success, otherwise errno.
+ * @return:  zero on success, otherwise errno.
  */
-struct ib_ucontext *pvrdma_alloc_ucontext(struct ib_device *ibdev,
-					  struct ib_udata *udata)
+int pvrdma_alloc_ucontext(struct ib_ucontext *uctx, struct ib_udata *udata)
 {
+	struct ib_device *ibdev = uctx->device;
 	struct pvrdma_dev *vdev = to_vdev(ibdev);
-	struct pvrdma_ucontext *context;
-	union pvrdma_cmd_req req;
-	union pvrdma_cmd_resp rsp;
+	struct pvrdma_ucontext *context = to_vucontext(uctx);
+	union pvrdma_cmd_req req = {};
+	union pvrdma_cmd_resp rsp = {};
 	struct pvrdma_cmd_create_uc *cmd = &req.create_uc;
 	struct pvrdma_cmd_create_uc_resp *resp = &rsp.create_uc_resp;
-	struct pvrdma_alloc_ucontext_resp uresp = {0};
+	struct pvrdma_alloc_ucontext_resp uresp = {};
 	int ret;
-	void *ptr;
 
 	if (!vdev->ib_active)
-		return ERR_PTR(-EAGAIN);
-
-	context = kzalloc(sizeof(*context), GFP_KERNEL);
-	if (!context)
-		return ERR_PTR(-ENOMEM);
+		return -EAGAIN;
 
 	context->dev = vdev;
 	ret = pvrdma_uar_alloc(vdev, &context->uar);
-	if (ret) {
-		kfree(context);
-		return ERR_PTR(-ENOMEM);
-	}
+	if (ret)
+		return -ENOMEM;
 
 	/* get ctx_handle from host */
-	memset(cmd, 0, sizeof(*cmd));
-
 	if (vdev->dsr_version < PVRDMA_PPN64_VERSION)
 		cmd->pfn = context->uar.pfn;
 	else
@@ -351,7 +342,6 @@ struct ib_ucontext *pvrdma_alloc_ucontext(struct ib_device *ibdev,
 	if (ret < 0) {
 		dev_warn(&vdev->pdev->dev,
 			 "could not create ucontext, error: %d\n", ret);
-		ptr = ERR_PTR(ret);
 		goto err;
 	}
 
@@ -362,33 +352,28 @@ struct ib_ucontext *pvrdma_alloc_ucontext(struct ib_device *ibdev,
 	ret = ib_copy_to_udata(udata, &uresp, sizeof(uresp));
 	if (ret) {
 		pvrdma_uar_free(vdev, &context->uar);
-		context->ibucontext.device = ibdev;
 		pvrdma_dealloc_ucontext(&context->ibucontext);
-		return ERR_PTR(-EFAULT);
+		return -EFAULT;
 	}
 
-	return &context->ibucontext;
+	return 0;
 
 err:
 	pvrdma_uar_free(vdev, &context->uar);
-	kfree(context);
-	return ptr;
+	return ret;
 }
 
 /**
  * pvrdma_dealloc_ucontext - deallocate ucontext
  * @ibcontext: the ucontext
- *
- * @return: 0 on success, otherwise errno.
  */
-int pvrdma_dealloc_ucontext(struct ib_ucontext *ibcontext)
+void pvrdma_dealloc_ucontext(struct ib_ucontext *ibcontext)
 {
 	struct pvrdma_ucontext *context = to_vucontext(ibcontext);
-	union pvrdma_cmd_req req;
+	union pvrdma_cmd_req req = {};
 	struct pvrdma_cmd_destroy_uc *cmd = &req.destroy_uc;
 	int ret;
 
-	memset(cmd, 0, sizeof(*cmd));
 	cmd->hdr.cmd = PVRDMA_CMD_DESTROY_UC;
 	cmd->ctx_handle = context->ctx_handle;
 
@@ -399,9 +384,6 @@ int pvrdma_dealloc_ucontext(struct ib_ucontext *ibcontext)
 
 	/* Free the UAR even if the device command failed */
 	pvrdma_uar_free(to_vdev(ibcontext->device), &context->uar);
-	kfree(context);
-
-	return ret;
 }
 
 /**
diff --git a/drivers/infiniband/hw/vmw_pvrdma/pvrdma_verbs.h b/drivers/infiniband/hw/vmw_pvrdma/pvrdma_verbs.h
index ed91baad1ffa..607aa131d67c 100644
--- a/drivers/infiniband/hw/vmw_pvrdma/pvrdma_verbs.h
+++ b/drivers/infiniband/hw/vmw_pvrdma/pvrdma_verbs.h
@@ -396,9 +396,8 @@ int pvrdma_modify_device(struct ib_device *ibdev, int mask,
 int pvrdma_modify_port(struct ib_device *ibdev, u8 port,
 		       int mask, struct ib_port_modify *props);
 int pvrdma_mmap(struct ib_ucontext *context, struct vm_area_struct *vma);
-struct ib_ucontext *pvrdma_alloc_ucontext(struct ib_device *ibdev,
-					  struct ib_udata *udata);
-int pvrdma_dealloc_ucontext(struct ib_ucontext *context);
+int pvrdma_alloc_ucontext(struct ib_ucontext *uctx, struct ib_udata *udata);
+void pvrdma_dealloc_ucontext(struct ib_ucontext *context);
 int pvrdma_alloc_pd(struct ib_pd *pd, struct ib_ucontext *context,
 		    struct ib_udata *udata);
 void pvrdma_dealloc_pd(struct ib_pd *ibpd);
diff --git a/drivers/infiniband/sw/rdmavt/vt.c b/drivers/infiniband/sw/rdmavt/vt.c
index ee9c82cb3b6b..42c9d35f832d 100644
--- a/drivers/infiniband/sw/rdmavt/vt.c
+++ b/drivers/infiniband/sw/rdmavt/vt.c
@@ -292,28 +292,21 @@ static inline struct rvt_ucontext *to_iucontext(struct ib_ucontext
 
 /**
  * rvt_alloc_ucontext - Allocate a user context
- * @ibdev: Verbs IB dev
+ * @uctx: Verbs context
  * @udata: User data allocated
  */
-static struct ib_ucontext *rvt_alloc_ucontext(struct ib_device *ibdev,
-					      struct ib_udata *udata)
+static int rvt_alloc_ucontext(struct ib_ucontext *uctx, struct ib_udata *udata)
 {
-	struct rvt_ucontext *context;
-
-	context = kzalloc(sizeof(*context), GFP_KERNEL);
-	if (!context)
-		return ERR_PTR(-ENOMEM);
-	return &context->ibucontext;
+	return 0;
 }
 
 /**
- *rvt_dealloc_ucontext - Free a user context
- *@context - Free this
+ * rvt_dealloc_ucontext - Free a user context
+ * @context - Free this
  */
-static int rvt_dealloc_ucontext(struct ib_ucontext *context)
+static void rvt_dealloc_ucontext(struct ib_ucontext *context)
 {
-	kfree(to_iucontext(context));
-	return 0;
+	return;
 }
 
 static int rvt_get_port_immutable(struct ib_device *ibdev, u8 port_num,
@@ -433,6 +426,7 @@ static const struct ib_device_ops rvt_dev_ops = {
 	.resize_cq = rvt_resize_cq,
 	.unmap_fmr = rvt_unmap_fmr,
 	INIT_RDMA_OBJ_SIZE(ib_pd, rvt_pd, ibpd),
+	INIT_RDMA_OBJ_SIZE(ib_ucontext, rvt_ucontext, ibucontext),
 };
 
 static noinline int check_support(struct rvt_dev_info *rdi, int verb)
diff --git a/drivers/infiniband/sw/rxe/rxe_pool.c b/drivers/infiniband/sw/rxe/rxe_pool.c
index c9b8b8c6bfb5..120fa9005954 100644
--- a/drivers/infiniband/sw/rxe/rxe_pool.c
+++ b/drivers/infiniband/sw/rxe/rxe_pool.c
@@ -42,6 +42,7 @@ struct rxe_type_info rxe_type_info[RXE_NUM_TYPES] = {
 	[RXE_TYPE_UC] = {
 		.name		= "rxe-uc",
 		.size		= sizeof(struct rxe_ucontext),
+		.flags          = RXE_POOL_NO_ALLOC,
 	},
 	[RXE_TYPE_PD] = {
 		.name		= "rxe-pd",
diff --git a/drivers/infiniband/sw/rxe/rxe_verbs.c b/drivers/infiniband/sw/rxe/rxe_verbs.c
index bd6a379b79d3..6ecf28570ff0 100644
--- a/drivers/infiniband/sw/rxe/rxe_verbs.c
+++ b/drivers/infiniband/sw/rxe/rxe_verbs.c
@@ -142,22 +142,19 @@ static enum rdma_link_layer rxe_get_link_layer(struct ib_device *dev,
 	return rxe_link_layer(rxe, port_num);
 }
 
-static struct ib_ucontext *rxe_alloc_ucontext(struct ib_device *dev,
-					      struct ib_udata *udata)
+static int rxe_alloc_ucontext(struct ib_ucontext *uctx, struct ib_udata *udata)
 {
-	struct rxe_dev *rxe = to_rdev(dev);
-	struct rxe_ucontext *uc;
+	struct rxe_dev *rxe = to_rdev(uctx->device);
+	struct rxe_ucontext *uc = to_ruc(uctx);
 
-	uc = rxe_alloc(&rxe->uc_pool);
-	return uc ? &uc->ibuc : ERR_PTR(-ENOMEM);
+	return rxe_add_to_pool(&rxe->uc_pool, &uc->pelem);
 }
 
-static int rxe_dealloc_ucontext(struct ib_ucontext *ibuc)
+static void rxe_dealloc_ucontext(struct ib_ucontext *ibuc)
 {
 	struct rxe_ucontext *uc = to_ruc(ibuc);
 
 	rxe_drop_ref(uc);
-	return 0;
 }
 
 static int rxe_port_immutable(struct ib_device *dev, u8 port_num,
@@ -1180,6 +1177,7 @@ static const struct ib_device_ops rxe_dev_ops = {
 	.req_notify_cq = rxe_req_notify_cq,
 	.resize_cq = rxe_resize_cq,
 	INIT_RDMA_OBJ_SIZE(ib_pd, rxe_pd, ibpd),
+	INIT_RDMA_OBJ_SIZE(ib_ucontext, rxe_ucontext, ibuc),
 };
 
 int rxe_register_device(struct rxe_dev *rxe, const char *ibdev_name)
diff --git a/drivers/infiniband/sw/rxe/rxe_verbs.h b/drivers/infiniband/sw/rxe/rxe_verbs.h
index d02eb75ef282..157e51aeb1e1 100644
--- a/drivers/infiniband/sw/rxe/rxe_verbs.h
+++ b/drivers/infiniband/sw/rxe/rxe_verbs.h
@@ -61,8 +61,8 @@ static inline int psn_compare(u32 psn_a, u32 psn_b)
 }
 
 struct rxe_ucontext {
+	struct ib_ucontext ibuc;
 	struct rxe_pool_entry	pelem;
-	struct ib_ucontext	ibuc;
 };
 
 struct rxe_pd {
diff --git a/include/rdma/ib_verbs.h b/include/rdma/ib_verbs.h
index 225cb76d469f..9b9e17bcc201 100644
--- a/include/rdma/ib_verbs.h
+++ b/include/rdma/ib_verbs.h
@@ -2389,9 +2389,9 @@ struct ib_device_ops {
 	int (*del_gid)(const struct ib_gid_attr *attr, void **context);
 	int (*query_pkey)(struct ib_device *device, u8 port_num, u16 index,
 			  u16 *pkey);
-	struct ib_ucontext *(*alloc_ucontext)(struct ib_device *device,
-					      struct ib_udata *udata);
-	int (*dealloc_ucontext)(struct ib_ucontext *context);
+	int (*alloc_ucontext)(struct ib_ucontext *context,
+			      struct ib_udata *udata);
+	void (*dealloc_ucontext)(struct ib_ucontext *context);
 	int (*mmap)(struct ib_ucontext *context, struct vm_area_struct *vma);
 	void (*disassociate_ucontext)(struct ib_ucontext *ibcontext);
 	int (*alloc_pd)(struct ib_pd *pd, struct ib_ucontext *context,
@@ -2551,6 +2551,7 @@ struct ib_device_ops {
 	void (*dealloc_driver)(struct ib_device *dev);
 
 	DECLARE_RDMA_OBJ_SIZE(ib_pd);
+	DECLARE_RDMA_OBJ_SIZE(ib_ucontext);
 };
 
 struct rdma_restrack_root;
-- 
cgit v1.2.3


From cd27287562d69629c5f007c6f64c27b6ff15f3e5 Mon Sep 17 00:00:00 2001
From: Aya Levin <ayal@mellanox.com>
Date: Mon, 11 Mar 2019 14:35:58 +0200
Subject: IB/mlx5: Fix mapping of link-mode to IB width and speed

Add mapping of link mode: CAUI4 100Gbps CR4/KR4 with 4 lines and 25Gbps.
Fix mapping of link mode: GAUI2 50Gbps CR2/KR2 to be 2 lines with 25Gbps.

Fixes: 08e8676f1607 ("IB/mlx5: Add support for 50Gbps per lane link modes")
Signed-off-by: Aya Levin <ayal@mellanox.com>
Signed-off-by: Leon Romanovsky <leonro@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 drivers/infiniband/hw/mlx5/main.c | 7 +++++++
 1 file changed, 7 insertions(+)

(limited to 'drivers/infiniband/hw/mlx5/main.c')

diff --git a/drivers/infiniband/hw/mlx5/main.c b/drivers/infiniband/hw/mlx5/main.c
index 994c19d01211..531ff20b32ad 100644
--- a/drivers/infiniband/hw/mlx5/main.c
+++ b/drivers/infiniband/hw/mlx5/main.c
@@ -415,10 +415,17 @@ static int translate_eth_ext_proto_oper(u32 eth_proto_oper, u8 *active_speed,
 		*active_speed = IB_SPEED_EDR;
 		break;
 	case MLX5E_PROT_MASK(MLX5E_50GAUI_2_LAUI_2_50GBASE_CR2_KR2):
+		*active_width = IB_WIDTH_2X;
+		*active_speed = IB_SPEED_EDR;
+		break;
 	case MLX5E_PROT_MASK(MLX5E_50GAUI_1_LAUI_1_50GBASE_CR_KR):
 		*active_width = IB_WIDTH_1X;
 		*active_speed = IB_SPEED_HDR;
 		break;
+	case MLX5E_PROT_MASK(MLX5E_CAUI_4_100GBASE_CR4_KR4):
+		*active_width = IB_WIDTH_4X;
+		*active_speed = IB_SPEED_EDR;
+		break;
 	case MLX5E_PROT_MASK(MLX5E_100GAUI_2_100GBASE_CR2_KR2):
 		*active_width = IB_WIDTH_2X;
 		*active_speed = IB_SPEED_HDR;
-- 
cgit v1.2.3