Merge tag 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/rdma/rdma

Pull rdma updates from Jason Gunthorpe:
 "A very quiet cycle with few notable changes. Mostly the usual list of
  one or two patches to drivers changing something that isn't quite rc
  worthy. The subsystem seems to be seeing a larger number of rework and
  cleanup style patches right now, I feel that several vendors are
  prepping their drivers for new silicon.

  Summary:

   - Driver updates and cleanup for qedr, bnxt_re, hns, siw, mlx5, mlx4,
     rxe, i40iw

   - Larger series doing cleanup and rework for hns and hfi1.

   - Some general reworking of the CM code to make it a little more
     understandable

   - Unify the different code paths connected to the uverbs FD scheme

   - New UAPI ioctls conversions for get context and get async fd

   - Trace points for CQ and CM portions of the RDMA stack

   - mlx5 driver support for virtio-net formatted rings as RDMA raw
     ethernet QPs

   - verbs support for setting the PCI-E relaxed ordering bit on DMA
     traffic connected to a MR

   - A couple of bug fixes that came too late to make rc7"

* tag 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/rdma/rdma: (108 commits)
  RDMA/core: Make the entire API tree static
  RDMA/efa: Mask access flags with the correct optional range
  RDMA/cma: Fix unbalanced cm_id reference count during address resolve
  RDMA/umem: Fix ib_umem_find_best_pgsz()
  IB/mlx4: Fix leak in id_map_find_del
  IB/opa_vnic: Spelling correction of 'erorr' to 'error'
  IB/hfi1: Fix logical condition in msix_request_irq
  RDMA/cm: Remove CM message structs
  RDMA/cm: Use IBA functions for complex structure members
  RDMA/cm: Use IBA functions for simple structure members
  RDMA/cm: Use IBA functions for swapping get/set acessors
  RDMA/cm: Use IBA functions for simple get/set acessors
  RDMA/cm: Add SET/GET implementations to hide IBA wire format
  RDMA/cm: Add accessors for CM_REQ transport_type
  IB/mlx5: Return the administrative GUID if exists
  RDMA/core: Ensure that rdma_user_mmap_entry_remove() is a fence
  IB/mlx4: Fix memory leak in add_gid error flow
  IB/mlx5: Expose RoCE accelerator counters
  RDMA/mlx5: Set relaxed ordering when requested
  RDMA/core: Add the core support field to METHOD_GET_CONTEXT
  ...

16 files changed, 899 insertions, 101 deletions

diff --git a/include/linux/mlx4/cq.h b/include/linux/mlx4/cq.h
index 508e8cc5ee86..653d2a0aa44c 100644
--- a/include/linux/mlx4/cq.h
+++ b/include/linux/mlx4/cq.h
@@ -130,6 +130,11 @@ enum {
 	MLX4_CQE_STATUS_IPOK		= 1 << 12,
 };
 
+/* L4_CSUM is logically part of status, but has to checked against badfcs_enc */
+enum {
+	MLX4_CQE_STATUS_L4_CSUM		= 1 << 2,
+};
+
 enum {
 	MLX4_CQE_LLC                     = 1,
 	MLX4_CQE_SNAP                    = 1 << 1,
diff --git a/include/linux/mlx5/driver.h b/include/linux/mlx5/driver.h
index 22bd0d5024c8..277a51d3ec40 100644
--- a/include/linux/mlx5/driver.h
+++ b/include/linux/mlx5/driver.h
@@ -463,6 +463,11 @@ struct mlx5_vf_context {
 	int	enabled;
 	u64	port_guid;
 	u64	node_guid;
+	/* Valid bits are used to validate administrative guid only.
+	 * Enabled after ndo_set_vf_guid
+	 */
+	u8	port_guid_valid:1;
+	u8	node_guid_valid:1;
 	enum port_state_policy	policy;
 };
 
diff --git a/include/linux/mlx5/mlx5_ifc.h b/include/linux/mlx5/mlx5_ifc.h
index ee0a34d66c7c..032cd6630720 100644
--- a/include/linux/mlx5/mlx5_ifc.h
+++ b/include/linux/mlx5/mlx5_ifc.h
@@ -1188,7 +1188,8 @@ struct mlx5_ifc_cmd_hca_cap_bits {
 	u8         log_max_cq[0x5];
 
 	u8         log_max_eq_sz[0x8];
-	u8         reserved_at_e8[0x2];
+	u8         relaxed_ordering_write[0x1];
+	u8         relaxed_ordering_read[0x1];
 	u8         log_max_mkey[0x6];
 	u8         reserved_at_f0[0x8];
 	u8         dump_fill_mkey[0x1];
@@ -1211,7 +1212,8 @@ struct mlx5_ifc_cmd_hca_cap_bits {
 	u8         reserved_at_130[0xa];
 	u8         log_max_ra_res_dc[0x6];
 
-	u8         reserved_at_140[0xa];
+	u8         reserved_at_140[0x9];
+	u8         roce_accl[0x1];
 	u8         log_max_ra_req_qp[0x6];
 	u8         reserved_at_150[0xa];
 	u8         log_max_ra_res_qp[0x6];
@@ -3428,7 +3430,9 @@ struct mlx5_ifc_mkc_bits {
 
 	u8         translations_octword_size[0x20];
 
-	u8         reserved_at_1c0[0x1b];
+	u8         reserved_at_1c0[0x19];
+	u8         relaxed_ordering_read[0x1];
+	u8         reserved_at_1d9[0x1];
 	u8         log_page_size[0x5];
 
 	u8         reserved_at_1e0[0x20];
@@ -4889,7 +4893,19 @@ struct mlx5_ifc_query_q_counter_out_bits {
 
 	u8         req_cqe_flush_error[0x20];
 
-	u8         reserved_at_620[0x1e0];
+	u8         reserved_at_620[0x20];
+
+	u8         roce_adp_retrans[0x20];
+
+	u8         roce_adp_retrans_to[0x20];
+
+	u8         roce_slow_restart[0x20];
+
+	u8         roce_slow_restart_cnps[0x20];
+
+	u8         roce_slow_restart_trans[0x20];
+
+	u8         reserved_at_6e0[0x120];
 };
 
 struct mlx5_ifc_query_q_counter_in_bits {
diff --git a/include/rdma/ib_cm.h b/include/rdma/ib_cm.h
index b01a8a8d4de9..8ec482e391aa 100644
--- a/include/rdma/ib_cm.h
+++ b/include/rdma/ib_cm.h
@@ -500,21 +500,6 @@ int ib_send_cm_mra(struct ib_cm_id *cm_id,
 		   u8 private_data_len);
 
 /**
- * ib_send_cm_lap - Sends a load alternate path request.
- * @cm_id: Connection identifier associated with the load alternate path
- *   message.
- * @alternate_path: A path record that identifies the alternate path to
- *   load.
- * @private_data: Optional user-defined private data sent with the
- *   load alternate path message.
- * @private_data_len: Size of the private data buffer, in bytes.
- */
-int ib_send_cm_lap(struct ib_cm_id *cm_id,
-		   struct sa_path_rec *alternate_path,
-		   const void *private_data,
-		   u8 private_data_len);
-
-/**
  * ib_cm_init_qp_attr - Initializes the QP attributes for use in transitioning
  *   to a specified QP state.
  * @cm_id: Communication identifier associated with the QP attributes to
@@ -534,25 +519,6 @@ int ib_cm_init_qp_attr(struct ib_cm_id *cm_id,
 		       struct ib_qp_attr *qp_attr,
 		       int *qp_attr_mask);
 
-/**
- * ib_send_cm_apr - Sends an alternate path response message in response to
- *   a load alternate path request.
- * @cm_id: Connection identifier associated with the alternate path response.
- * @status: Reply status sent with the alternate path response.
- * @info: Optional additional information sent with the alternate path
- *   response.
- * @info_length: Size of the additional information, in bytes.
- * @private_data: Optional user-defined private data sent with the
- *   alternate path response message.
- * @private_data_len: Size of the private data buffer, in bytes.
- */
-int ib_send_cm_apr(struct ib_cm_id *cm_id,
-		   enum ib_cm_apr_status status,
-		   void *info,
-		   u8 info_length,
-		   const void *private_data,
-		   u8 private_data_len);
-
 struct ib_cm_sidr_req_param {
 	struct sa_path_rec	*path;
 	const struct ib_gid_attr *sgid_attr;
diff --git a/include/rdma/ib_verbs.h b/include/rdma/ib_verbs.h
index e2cc62217cc2..1f779fad3a1e 100644
--- a/include/rdma/ib_verbs.h
+++ b/include/rdma/ib_verbs.h
@@ -72,11 +72,16 @@
 #define IB_FW_VERSION_NAME_MAX	ETHTOOL_FWVERS_LEN
 
 struct ib_umem_odp;
+struct ib_uqp_object;
+struct ib_usrq_object;
+struct ib_uwq_object;
 
 extern struct workqueue_struct *ib_wq;
 extern struct workqueue_struct *ib_comp_wq;
 extern struct workqueue_struct *ib_comp_unbound_wq;
 
+struct ib_ucq_object;
+
 __printf(3, 4) __cold
 void ibdev_printk(const char *level, const struct ib_device *ibdev,
 		  const char *format, ...);
@@ -1413,8 +1418,11 @@ enum ib_access_flags {
 	IB_ZERO_BASED = IB_UVERBS_ACCESS_ZERO_BASED,
 	IB_ACCESS_ON_DEMAND = IB_UVERBS_ACCESS_ON_DEMAND,
 	IB_ACCESS_HUGETLB = IB_UVERBS_ACCESS_HUGETLB,
+	IB_ACCESS_RELAXED_ORDERING = IB_UVERBS_ACCESS_RELAXED_ORDERING,
 
-	IB_ACCESS_SUPPORTED = ((IB_ACCESS_HUGETLB << 1) - 1)
+	IB_ACCESS_OPTIONAL = IB_UVERBS_ACCESS_OPTIONAL_RANGE,
+	IB_ACCESS_SUPPORTED =
+		((IB_ACCESS_HUGETLB << 1) - 1) | IB_ACCESS_OPTIONAL,
 };
 
 /*
@@ -1544,7 +1552,7 @@ enum ib_poll_context {
 
 struct ib_cq {
 	struct ib_device       *device;
-	struct ib_uobject      *uobject;
+	struct ib_ucq_object   *uobject;
 	ib_comp_handler   	comp_handler;
 	void                  (*event_handler)(struct ib_event *, void *);
 	void                   *cq_context;
@@ -1558,6 +1566,11 @@ struct ib_cq {
 	};
 	struct workqueue_struct *comp_wq;
 	struct dim *dim;
+
+	/* updated only by trace points */
+	ktime_t timestamp;
+	bool interrupt;
+
 	/*
 	 * Implementation details of the RDMA core, don't use in drivers:
 	 */
@@ -1567,7 +1580,7 @@ struct ib_cq {
 struct ib_srq {
 	struct ib_device       *device;
 	struct ib_pd	       *pd;
-	struct ib_uobject      *uobject;
+	struct ib_usrq_object  *uobject;
 	void		      (*event_handler)(struct ib_event *, void *);
 	void		       *srq_context;
 	enum ib_srq_type	srq_type;
@@ -1612,7 +1625,7 @@ enum ib_wq_state {
 
 struct ib_wq {
 	struct ib_device       *device;
-	struct ib_uobject      *uobject;
+	struct ib_uwq_object   *uobject;
 	void		    *wq_context;
 	void		    (*event_handler)(struct ib_event *, void *);
 	struct ib_pd	       *pd;
@@ -1728,7 +1741,7 @@ struct ib_qp {
 	atomic_t		usecnt;
 	struct list_head	open_list;
 	struct ib_qp           *real_qp;
-	struct ib_uobject      *uobject;
+	struct ib_uqp_object   *uobject;
 	void                  (*event_handler)(struct ib_event *, void *);
 	void		       *qp_context;
 	/* sgid_attrs associated with the AV's */
@@ -2147,11 +2160,6 @@ struct ib_port_cache {
 	enum ib_port_state     port_state;
 };
 
-struct ib_cache {
-	rwlock_t                lock;
-	struct ib_event_handler event_handler;
-};
-
 struct ib_port_immutable {
 	int                           pkey_tbl_len;
 	int                           gid_tbl_len;
@@ -2627,13 +2635,18 @@ struct ib_device {
 	struct rcu_head rcu_head;
 
 	struct list_head              event_handler_list;
-	spinlock_t                    event_handler_lock;
+	/* Protects event_handler_list */
+	struct rw_semaphore event_handler_rwsem;
+
+	/* Protects QP's event_handler calls and open_qp list */
+	spinlock_t qp_open_list_lock;
 
 	struct rw_semaphore	      client_data_rwsem;
 	struct xarray                 client_data;
 	struct mutex                  unregistration_lock;
 
-	struct ib_cache               cache;
+	/* Synchronize GID, Pkey cache entries, subnet prefix, LMC */
+	rwlock_t cache_lock;
 	/**
 	 * port_data is indexed by port number
 	 */
@@ -2942,7 +2955,7 @@ bool ib_modify_qp_is_ok(enum ib_qp_state cur_state, enum ib_qp_state next_state,
 
 void ib_register_event_handler(struct ib_event_handler *event_handler);
 void ib_unregister_event_handler(struct ib_event_handler *event_handler);
-void ib_dispatch_event(struct ib_event *event);
+void ib_dispatch_event(const struct ib_event *event);
 
 int ib_query_port(struct ib_device *device,
 		  u8 port_num, struct ib_port_attr *port_attr);
@@ -4309,6 +4322,9 @@ static inline int ib_check_mr_access(int flags)
 	    !(flags & IB_ACCESS_LOCAL_WRITE))
 		return -EINVAL;
 
+	if (flags & ~IB_ACCESS_SUPPORTED)
+		return -EINVAL;
+
 	return 0;
 }
 
diff --git a/include/rdma/iba.h b/include/rdma/iba.h
new file mode 100644
index 000000000000..6a1115b02a0d
--- /dev/null
+++ b/include/rdma/iba.h
@@ -0,0 +1,146 @@
+/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */
+/*
+ * Copyright (c) 2020, Mellanox Technologies inc.  All rights reserved.
+ */
+#ifndef _IBA_DEFS_H_
+#define _IBA_DEFS_H_
+
+#include <linux/kernel.h>
+#include <linux/bitfield.h>
+#include <asm/unaligned.h>
+
+static inline u32 _iba_get8(const u8 *ptr)
+{
+	return *ptr;
+}
+
+static inline void _iba_set8(u8 *ptr, u32 mask, u32 prep_value)
+{
+	*ptr = (*ptr & ~mask) | prep_value;
+}
+
+static inline u16 _iba_get16(const __be16 *ptr)
+{
+	return be16_to_cpu(*ptr);
+}
+
+static inline void _iba_set16(__be16 *ptr, u16 mask, u16 prep_value)
+{
+	*ptr = cpu_to_be16((be16_to_cpu(*ptr) & ~mask) | prep_value);
+}
+
+static inline u32 _iba_get32(const __be32 *ptr)
+{
+	return be32_to_cpu(*ptr);
+}
+
+static inline void _iba_set32(__be32 *ptr, u32 mask, u32 prep_value)
+{
+	*ptr = cpu_to_be32((be32_to_cpu(*ptr) & ~mask) | prep_value);
+}
+
+static inline u64 _iba_get64(const __be64 *ptr)
+{
+	/*
+	 * The mads are constructed so that 32 bit and smaller are naturally
+	 * aligned, everything larger has a max alignment of 4 bytes.
+	 */
+	return be64_to_cpu(get_unaligned(ptr));
+}
+
+static inline void _iba_set64(__be64 *ptr, u64 mask, u64 prep_value)
+{
+	put_unaligned(cpu_to_be64((_iba_get64(ptr) & ~mask) | prep_value), ptr);
+}
+
+#define _IBA_SET(field_struct, field_offset, field_mask, num_bits, ptr, value) \
+	({                                                                     \
+		field_struct *_ptr = ptr;                                      \
+		_iba_set##num_bits((void *)_ptr + (field_offset), field_mask,  \
+				   FIELD_PREP(field_mask, value));             \
+	})
+#define IBA_SET(field, ptr, value) _IBA_SET(field, ptr, value)
+
+#define _IBA_GET_MEM_PTR(field_struct, field_offset, type, num_bits, ptr)      \
+	({                                                                     \
+		field_struct *_ptr = ptr;                                      \
+		(type *)((void *)_ptr + (field_offset));                       \
+	})
+#define IBA_GET_MEM_PTR(field, ptr) _IBA_GET_MEM_PTR(field, ptr)
+
+/* FIXME: A set should always set the entire field, meaning we should zero the trailing bytes */
+#define _IBA_SET_MEM(field_struct, field_offset, type, num_bits, ptr, in,      \
+		     bytes)                                                    \
+	({                                                                     \
+		const type *_in_ptr = in;                                      \
+		WARN_ON(bytes * 8 > num_bits);                                 \
+		if (in && bytes)                                               \
+			memcpy(_IBA_GET_MEM_PTR(field_struct, field_offset,    \
+						type, num_bits, ptr),          \
+			       _in_ptr, bytes);                                \
+	})
+#define IBA_SET_MEM(field, ptr, in, bytes) _IBA_SET_MEM(field, ptr, in, bytes)
+
+#define _IBA_GET(field_struct, field_offset, field_mask, num_bits, ptr)        \
+	({                                                                     \
+		const field_struct *_ptr = ptr;                                \
+		(u##num_bits) FIELD_GET(                                       \
+			field_mask, _iba_get##num_bits((const void *)_ptr +    \
+						       (field_offset)));       \
+	})
+#define IBA_GET(field, ptr) _IBA_GET(field, ptr)
+
+#define _IBA_GET_MEM(field_struct, field_offset, type, num_bits, ptr, out,     \
+		     bytes)                                                    \
+	({                                                                     \
+		type *_out_ptr = out;                                          \
+		WARN_ON(bytes * 8 > num_bits);                                 \
+		if (out && bytes)                                              \
+			memcpy(_out_ptr,                                       \
+			       _IBA_GET_MEM_PTR(field_struct, field_offset,    \
+						type, num_bits, ptr),          \
+			       bytes);                                         \
+	})
+#define IBA_GET_MEM(field, ptr, out, bytes) _IBA_GET_MEM(field, ptr, out, bytes)
+
+/*
+ * The generated list becomes the parameters to the macros, the order is:
+ *  - struct this applies to
+ *  - starting offset of the max
+ *  - GENMASK or GENMASK_ULL in CPU order
+ *  - The width of data the mask operations should work on, in bits
+ */
+
+/*
+ * Extraction using a tabular description like table 106. bit_offset is from
+ * the Byte[Bit] notation.
+ */
+#define IBA_FIELD_BLOC(field_struct, byte_offset, bit_offset, num_bits)        \
+	field_struct, byte_offset,                                             \
+		GENMASK(7 - (bit_offset), 7 - (bit_offset) - (num_bits - 1)),  \
+		8
+#define IBA_FIELD8_LOC(field_struct, byte_offset, num_bits)                    \
+	IBA_FIELD_BLOC(field_struct, byte_offset, 0, num_bits)
+
+#define IBA_FIELD16_LOC(field_struct, byte_offset, num_bits)                   \
+	field_struct, (byte_offset)&0xFFFE,                                    \
+		GENMASK(15 - (((byte_offset) % 2) * 8),                        \
+			15 - (((byte_offset) % 2) * 8) - (num_bits - 1)),      \
+		16
+
+#define IBA_FIELD32_LOC(field_struct, byte_offset, num_bits)                   \
+	field_struct, (byte_offset)&0xFFFC,                                    \
+		GENMASK(31 - (((byte_offset) % 4) * 8),                        \
+			31 - (((byte_offset) % 4) * 8) - (num_bits - 1)),      \
+		32
+
+#define IBA_FIELD64_LOC(field_struct, byte_offset)                             \
+	field_struct, byte_offset, GENMASK_ULL(63, 0), 64
+/*
+ * In IBTA spec, everything that is more than 64bits is multiple
+ * of bytes without leftover bits.
+ */
+#define IBA_FIELD_MLOC(field_struct, byte_offset, num_bits, type)              \
+	field_struct, byte_offset, type, num_bits
+
+#endif /* _IBA_DEFS_H_ */
diff --git a/include/rdma/ibta_vol1_c12.h b/include/rdma/ibta_vol1_c12.h
new file mode 100644
index 000000000000..269904425d3f
--- /dev/null
+++ b/include/rdma/ibta_vol1_c12.h
@@ -0,0 +1,213 @@
+/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */
+/*
+ * Copyright (c) 2020, Mellanox Technologies inc. All rights reserved.
+ *
+ * This file is IBTA volume 1, chapter 12 declarations:
+ *  CHAPTER 12: COMMUNICATION MANAGEMENT
+ */
+#ifndef _IBTA_VOL1_C12_H_
+#define _IBTA_VOL1_C12_H_
+
+#include <rdma/iba.h>
+
+#define CM_FIELD_BLOC(field_struct, byte_offset, bits_offset, width)           \
+	IBA_FIELD_BLOC(field_struct,                                           \
+		       (byte_offset + sizeof(struct ib_mad_hdr)), bits_offset, \
+		       width)
+#define CM_FIELD8_LOC(field_struct, byte_offset, width)                        \
+	IBA_FIELD8_LOC(field_struct,                                           \
+		       (byte_offset + sizeof(struct ib_mad_hdr)), width)
+#define CM_FIELD16_LOC(field_struct, byte_offset, width)                       \
+	IBA_FIELD16_LOC(field_struct,                                          \
+			(byte_offset + sizeof(struct ib_mad_hdr)), width)
+#define CM_FIELD32_LOC(field_struct, byte_offset, width)                       \
+	IBA_FIELD32_LOC(field_struct,                                          \
+			(byte_offset + sizeof(struct ib_mad_hdr)), width)
+#define CM_FIELD64_LOC(field_struct, byte_offset)                              \
+	IBA_FIELD64_LOC(field_struct, (byte_offset + sizeof(struct ib_mad_hdr)))
+#define CM_FIELD_MLOC(field_struct, byte_offset, width, type)                  \
+	IBA_FIELD_MLOC(field_struct,                                           \
+		       (byte_offset + sizeof(struct ib_mad_hdr)), width, type)
+#define CM_STRUCT(field_struct, total_len)                                     \
+	field_struct                                                           \
+	{                                                                      \
+		struct ib_mad_hdr hdr;                                         \
+		u32 _data[(total_len) / 32 +                                   \
+			  BUILD_BUG_ON_ZERO((total_len) % 32 != 0)];           \
+	}
+
+/* Table 106 REQ Message Contents */
+#define CM_REQ_LOCAL_COMM_ID CM_FIELD32_LOC(struct cm_req_msg, 0, 32)
+#define CM_REQ_SERVICE_ID CM_FIELD64_LOC(struct cm_req_msg, 8)
+#define CM_REQ_LOCAL_CA_GUID CM_FIELD64_LOC(struct cm_req_msg, 16)
+#define CM_REQ_LOCAL_Q_KEY CM_FIELD32_LOC(struct cm_req_msg, 28, 32)
+#define CM_REQ_LOCAL_QPN CM_FIELD32_LOC(struct cm_req_msg, 32, 24)
+#define CM_REQ_RESPONDER_RESOURCES CM_FIELD8_LOC(struct cm_req_msg, 35, 8)
+#define CM_REQ_LOCAL_EECN CM_FIELD32_LOC(struct cm_req_msg, 36, 24)
+#define CM_REQ_INITIATOR_DEPTH CM_FIELD8_LOC(struct cm_req_msg, 39, 8)
+#define CM_REQ_REMOTE_EECN CM_FIELD32_LOC(struct cm_req_msg, 40, 24)
+#define CM_REQ_REMOTE_CM_RESPONSE_TIMEOUT                                      \
+	CM_FIELD8_LOC(struct cm_req_msg, 43, 5)
+#define CM_REQ_TRANSPORT_SERVICE_TYPE CM_FIELD_BLOC(struct cm_req_msg, 43, 5, 2)
+#define CM_REQ_END_TO_END_FLOW_CONTROL                                         \
+	CM_FIELD_BLOC(struct cm_req_msg, 43, 7, 1)
+#define CM_REQ_STARTING_PSN CM_FIELD32_LOC(struct cm_req_msg, 44, 24)
+#define CM_REQ_LOCAL_CM_RESPONSE_TIMEOUT CM_FIELD8_LOC(struct cm_req_msg, 47, 5)
+#define CM_REQ_RETRY_COUNT CM_FIELD_BLOC(struct cm_req_msg, 47, 5, 3)
+#define CM_REQ_PARTITION_KEY CM_FIELD16_LOC(struct cm_req_msg, 48, 16)
+#define CM_REQ_PATH_PACKET_PAYLOAD_MTU CM_FIELD8_LOC(struct cm_req_msg, 50, 4)
+#define CM_REQ_RDC_EXISTS CM_FIELD_BLOC(struct cm_req_msg, 50, 4, 1)
+#define CM_REQ_RNR_RETRY_COUNT CM_FIELD_BLOC(struct cm_req_msg, 50, 5, 3)
+#define CM_REQ_MAX_CM_RETRIES CM_FIELD8_LOC(struct cm_req_msg, 51, 4)
+#define CM_REQ_SRQ CM_FIELD_BLOC(struct cm_req_msg, 51, 4, 1)
+#define CM_REQ_EXTENDED_TRANSPORT_TYPE                                         \
+	CM_FIELD_BLOC(struct cm_req_msg, 51, 5, 3)
+#define CM_REQ_PRIMARY_LOCAL_PORT_LID CM_FIELD16_LOC(struct cm_req_msg, 52, 16)
+#define CM_REQ_PRIMARY_REMOTE_PORT_LID CM_FIELD16_LOC(struct cm_req_msg, 54, 16)
+#define CM_REQ_PRIMARY_LOCAL_PORT_GID                                          \
+	CM_FIELD_MLOC(struct cm_req_msg, 56, 128, union ib_gid)
+#define CM_REQ_PRIMARY_REMOTE_PORT_GID                                         \
+	CM_FIELD_MLOC(struct cm_req_msg, 72, 128, union ib_gid)
+#define CM_REQ_PRIMARY_FLOW_LABEL CM_FIELD32_LOC(struct cm_req_msg, 88, 20)
+#define CM_REQ_PRIMARY_PACKET_RATE CM_FIELD_BLOC(struct cm_req_msg, 91, 2, 6)
+#define CM_REQ_PRIMARY_TRAFFIC_CLASS CM_FIELD8_LOC(struct cm_req_msg, 92, 8)
+#define CM_REQ_PRIMARY_HOP_LIMIT CM_FIELD8_LOC(struct cm_req_msg, 93, 8)
+#define CM_REQ_PRIMARY_SL CM_FIELD8_LOC(struct cm_req_msg, 94, 4)
+#define CM_REQ_PRIMARY_SUBNET_LOCAL CM_FIELD_BLOC(struct cm_req_msg, 94, 4, 1)
+#define CM_REQ_PRIMARY_LOCAL_ACK_TIMEOUT CM_FIELD8_LOC(struct cm_req_msg, 95, 5)
+#define CM_REQ_ALTERNATE_LOCAL_PORT_LID                                        \
+	CM_FIELD16_LOC(struct cm_req_msg, 96, 16)
+#define CM_REQ_ALTERNATE_REMOTE_PORT_LID                                       \
+	CM_FIELD16_LOC(struct cm_req_msg, 98, 16)
+#define CM_REQ_ALTERNATE_LOCAL_PORT_GID                                        \
+	CM_FIELD_MLOC(struct cm_req_msg, 100, 128, union ib_gid)
+#define CM_REQ_ALTERNATE_REMOTE_PORT_GID                                       \
+	CM_FIELD_MLOC(struct cm_req_msg, 116, 128, union ib_gid)
+#define CM_REQ_ALTERNATE_FLOW_LABEL CM_FIELD32_LOC(struct cm_req_msg, 132, 20)
+#define CM_REQ_ALTERNATE_PACKET_RATE CM_FIELD_BLOC(struct cm_req_msg, 135, 2, 6)
+#define CM_REQ_ALTERNATE_TRAFFIC_CLASS CM_FIELD8_LOC(struct cm_req_msg, 136, 8)
+#define CM_REQ_ALTERNATE_HOP_LIMIT CM_FIELD8_LOC(struct cm_req_msg, 137, 8)
+#define CM_REQ_ALTERNATE_SL CM_FIELD8_LOC(struct cm_req_msg, 138, 4)
+#define CM_REQ_ALTERNATE_SUBNET_LOCAL                                          \
+	CM_FIELD_BLOC(struct cm_req_msg, 138, 4, 1)
+#define CM_REQ_ALTERNATE_LOCAL_ACK_TIMEOUT                                     \
+	CM_FIELD8_LOC(struct cm_req_msg, 139, 5)
+#define CM_REQ_SAP_SUPPORTED CM_FIELD_BLOC(struct cm_req_msg, 139, 5, 1)
+#define CM_REQ_PRIVATE_DATA CM_FIELD_MLOC(struct cm_req_msg, 140, 736, void)
+CM_STRUCT(struct cm_req_msg, 140 * 8 + 736);
+
+/* Table 107 MRA Message Contents */
+#define CM_MRA_LOCAL_COMM_ID CM_FIELD32_LOC(struct cm_mra_msg, 0, 32)
+#define CM_MRA_REMOTE_COMM_ID CM_FIELD32_LOC(struct cm_mra_msg, 4, 32)
+#define CM_MRA_MESSAGE_MRAED CM_FIELD8_LOC(struct cm_mra_msg, 8, 2)
+#define CM_MRA_SERVICE_TIMEOUT CM_FIELD8_LOC(struct cm_mra_msg, 9, 5)
+#define CM_MRA_PRIVATE_DATA CM_FIELD_MLOC(struct cm_mra_msg, 10, 1776, void)
+CM_STRUCT(struct cm_mra_msg, 10 * 8 + 1776);
+
+/* Table 108 REJ Message Contents */
+#define CM_REJ_LOCAL_COMM_ID CM_FIELD32_LOC(struct cm_rej_msg, 0, 32)
+#define CM_REJ_REMOTE_COMM_ID CM_FIELD32_LOC(struct cm_rej_msg, 4, 32)
+#define CM_REJ_MESSAGE_REJECTED CM_FIELD8_LOC(struct cm_rej_msg, 8, 2)
+#define CM_REJ_REJECTED_INFO_LENGTH CM_FIELD8_LOC(struct cm_rej_msg, 9, 7)
+#define CM_REJ_REASON CM_FIELD16_LOC(struct cm_rej_msg, 10, 16)
+#define CM_REJ_ARI CM_FIELD_MLOC(struct cm_rej_msg, 12, 576, void)
+#define CM_REJ_PRIVATE_DATA CM_FIELD_MLOC(struct cm_rej_msg, 84, 1184, void)
+CM_STRUCT(struct cm_rej_msg, 84 * 8 + 1184);
+
+/* Table 110 REP Message Contents */
+#define CM_REP_LOCAL_COMM_ID CM_FIELD32_LOC(struct cm_rep_msg, 0, 32)
+#define CM_REP_REMOTE_COMM_ID CM_FIELD32_LOC(struct cm_rep_msg, 4, 32)
+#define CM_REP_LOCAL_Q_KEY CM_FIELD32_LOC(struct cm_rep_msg, 8, 32)
+#define CM_REP_LOCAL_QPN CM_FIELD32_LOC(struct cm_rep_msg, 12, 24)
+#define CM_REP_LOCAL_EE_CONTEXT_NUMBER CM_FIELD32_LOC(struct cm_rep_msg, 16, 24)
+#define CM_REP_STARTING_PSN CM_FIELD32_LOC(struct cm_rep_msg, 20, 24)
+#define CM_REP_RESPONDER_RESOURCES CM_FIELD8_LOC(struct cm_rep_msg, 24, 8)
+#define CM_REP_INITIATOR_DEPTH CM_FIELD8_LOC(struct cm_rep_msg, 25, 8)
+#define CM_REP_TARGET_ACK_DELAY CM_FIELD8_LOC(struct cm_rep_msg, 26, 5)
+#define CM_REP_FAILOVER_ACCEPTED CM_FIELD_BLOC(struct cm_rep_msg, 26, 5, 2)
+#define CM_REP_END_TO_END_FLOW_CONTROL                                         \
+	CM_FIELD_BLOC(struct cm_rep_msg, 26, 7, 1)
+#define CM_REP_RNR_RETRY_COUNT CM_FIELD8_LOC(struct cm_rep_msg, 27, 3)
+#define CM_REP_SRQ CM_FIELD_BLOC(struct cm_rep_msg, 27, 3, 1)
+#define CM_REP_LOCAL_CA_GUID CM_FIELD64_LOC(struct cm_rep_msg, 28)
+#define CM_REP_PRIVATE_DATA CM_FIELD_MLOC(struct cm_rep_msg, 36, 1568, void)
+CM_STRUCT(struct cm_rep_msg, 36 * 8 + 1568);
+
+/* Table 111 RTU Message Contents */
+#define CM_RTU_LOCAL_COMM_ID CM_FIELD32_LOC(struct cm_rtu_msg, 0, 32)
+#define CM_RTU_REMOTE_COMM_ID CM_FIELD32_LOC(struct cm_rtu_msg, 4, 32)
+#define CM_RTU_PRIVATE_DATA CM_FIELD_MLOC(struct cm_rtu_msg, 8, 1792, void)
+CM_STRUCT(struct cm_rtu_msg, 8 * 8 + 1792);
+
+/* Table 112 DREQ Message Contents */
+#define CM_DREQ_LOCAL_COMM_ID CM_FIELD32_LOC(struct cm_dreq_msg, 0, 32)
+#define CM_DREQ_REMOTE_COMM_ID CM_FIELD32_LOC(struct cm_dreq_msg, 4, 32)
+#define CM_DREQ_REMOTE_QPN_EECN CM_FIELD32_LOC(struct cm_dreq_msg, 8, 24)
+#define CM_DREQ_PRIVATE_DATA CM_FIELD_MLOC(struct cm_dreq_msg, 12, 1760, void)
+CM_STRUCT(struct cm_dreq_msg, 12 * 8 + 1760);
+
+/* Table 113 DREP Message Contents */
+#define CM_DREP_LOCAL_COMM_ID CM_FIELD32_LOC(struct cm_drep_msg, 0, 32)
+#define CM_DREP_REMOTE_COMM_ID CM_FIELD32_LOC(struct cm_drep_msg, 4, 32)
+#define CM_DREP_PRIVATE_DATA CM_FIELD_MLOC(struct cm_drep_msg, 8, 1792, void)
+CM_STRUCT(struct cm_drep_msg, 8 * 8 + 1792);
+
+/* Table 115 LAP Message Contents */
+#define CM_LAP_LOCAL_COMM_ID CM_FIELD32_LOC(struct cm_lap_msg, 0, 32)
+#define CM_LAP_REMOTE_COMM_ID CM_FIELD32_LOC(struct cm_lap_msg, 4, 32)
+#define CM_LAP_REMOTE_QPN_EECN CM_FIELD32_LOC(struct cm_lap_msg, 12, 24)
+#define CM_LAP_REMOTE_CM_RESPONSE_TIMEOUT                                      \
+	CM_FIELD8_LOC(struct cm_lap_msg, 15, 5)
+#define CM_LAP_ALTERNATE_LOCAL_PORT_LID                                        \
+	CM_FIELD16_LOC(struct cm_lap_msg, 20, 16)
+#define CM_LAP_ALTERNATE_REMOTE_PORT_LID                                       \
+	CM_FIELD16_LOC(struct cm_lap_msg, 22, 16)
+#define CM_LAP_ALTERNATE_LOCAL_PORT_GID                                        \
+	CM_FIELD_MLOC(struct cm_lap_msg, 24, 128, union ib_gid)
+#define CM_LAP_ALTERNATE_REMOTE_PORT_GID                                       \
+	CM_FIELD_MLOC(struct cm_lap_msg, 40, 128, union ib_gid)
+#define CM_LAP_ALTERNATE_FLOW_LABEL CM_FIELD32_LOC(struct cm_lap_msg, 56, 20)
+#define CM_LAP_ALTERNATE_TRAFFIC_CLASS CM_FIELD8_LOC(struct cm_lap_msg, 59, 8)
+#define CM_LAP_ALTERNATE_HOP_LIMIT CM_FIELD8_LOC(struct cm_lap_msg, 60, 8)
+#define CM_LAP_ALTERNATE_PACKET_RATE CM_FIELD_BLOC(struct cm_lap_msg, 61, 2, 6)
+#define CM_LAP_ALTERNATE_SL CM_FIELD8_LOC(struct cm_lap_msg, 62, 4)
+#define CM_LAP_ALTERNATE_SUBNET_LOCAL CM_FIELD_BLOC(struct cm_lap_msg, 62, 4, 1)
+#define CM_LAP_ALTERNATE_LOCAL_ACK_TIMEOUT                                     \
+	CM_FIELD8_LOC(struct cm_lap_msg, 63, 5)
+#define CM_LAP_PRIVATE_DATA CM_FIELD_MLOC(struct cm_lap_msg, 64, 1344, void)
+CM_STRUCT(struct cm_lap_msg, 64 * 8 + 1344);
+
+/* Table 116 APR Message Contents */
+#define CM_APR_LOCAL_COMM_ID CM_FIELD32_LOC(struct cm_apr_msg, 0, 32)
+#define CM_APR_REMOTE_COMM_ID CM_FIELD32_LOC(struct cm_apr_msg, 4, 32)
+#define CM_APR_ADDITIONAL_INFORMATION_LENGTH                                   \
+	CM_FIELD8_LOC(struct cm_apr_msg, 8, 8)
+#define CM_APR_AR_STATUS CM_FIELD8_LOC(struct cm_apr_msg, 9, 8)
+#define CM_APR_ADDITIONAL_INFORMATION                                          \
+	CM_FIELD_MLOC(struct cm_apr_msg, 12, 576, void)
+#define CM_APR_PRIVATE_DATA CM_FIELD_MLOC(struct cm_apr_msg, 84, 1184, void)
+CM_STRUCT(struct cm_apr_msg, 84 * 8 + 1184);
+
+/* Table 119 SIDR_REQ Message Contents */
+#define CM_SIDR_REQ_REQUESTID CM_FIELD32_LOC(struct cm_sidr_req_msg, 0, 32)
+#define CM_SIDR_REQ_PARTITION_KEY CM_FIELD16_LOC(struct cm_sidr_req_msg, 4, 16)
+#define CM_SIDR_REQ_SERVICEID CM_FIELD64_LOC(struct cm_sidr_req_msg, 8)
+#define CM_SIDR_REQ_PRIVATE_DATA                                               \
+	CM_FIELD_MLOC(struct cm_sidr_req_msg, 16, 1728, void)
+CM_STRUCT(struct cm_sidr_req_msg, 16 * 8 + 1728);
+
+/* Table 120 SIDR_REP Message Contents */
+#define CM_SIDR_REP_REQUESTID CM_FIELD32_LOC(struct cm_sidr_rep_msg, 0, 32)
+#define CM_SIDR_REP_STATUS CM_FIELD8_LOC(struct cm_sidr_rep_msg, 4, 8)
+#define CM_SIDR_REP_ADDITIONAL_INFORMATION_LENGTH                              \
+	CM_FIELD8_LOC(struct cm_sidr_rep_msg, 5, 8)
+#define CM_SIDR_REP_QPN CM_FIELD32_LOC(struct cm_sidr_rep_msg, 8, 24)
+#define CM_SIDR_REP_SERVICEID CM_FIELD64_LOC(struct cm_sidr_rep_msg, 12)
+#define CM_SIDR_REP_Q_KEY CM_FIELD32_LOC(struct cm_sidr_rep_msg, 20, 32)
+#define CM_SIDR_REP_ADDITIONAL_INFORMATION                                     \
+	CM_FIELD_MLOC(struct cm_sidr_rep_msg, 24, 576, void)
+#define CM_SIDR_REP_PRIVATE_DATA                                               \
+	CM_FIELD_MLOC(struct cm_sidr_rep_msg, 96, 1088, void)
+CM_STRUCT(struct cm_sidr_rep_msg, 96 * 8 + 1088);
+
+#endif /* _IBTA_VOL1_C12_H_ */
diff --git a/include/rdma/rdmavt_qp.h b/include/rdma/rdmavt_qp.h
index b550ae89bf85..0d5c70e2d8ab 100644
--- a/include/rdma/rdmavt_qp.h
+++ b/include/rdma/rdmavt_qp.h
@@ -640,34 +640,14 @@ static inline int rvt_cmp_msn(u32 a, u32 b)
 	return (((int)a) - ((int)b)) << 8;
 }
 
-/**
- * rvt_compute_aeth - compute the AETH (syndrome + MSN)
- * @qp: the queue pair to compute the AETH for
- *
- * Returns the AETH.
- */
 __be32 rvt_compute_aeth(struct rvt_qp *qp);
 
-/**
- * rvt_get_credit - flush the send work queue of a QP
- * @qp: the qp who's send work queue to flush
- * @aeth: the Acknowledge Extended Transport Header
- *
- * The QP s_lock should be held.
- */
 void rvt_get_credit(struct rvt_qp *qp, u32 aeth);
 
-/**
- * rvt_restart_sge - rewind the sge state for a wqe
- * @ss: the sge state pointer
- * @wqe: the wqe to rewind
- * @len: the data length from the start of the wqe in bytes
- *
- * Returns the remaining data length.
- */
 u32 rvt_restart_sge(struct rvt_sge_state *ss, struct rvt_swqe *wqe, u32 len);
 
 /**
+ * rvt_div_round_up_mtu - round up divide
  * @qp - the qp pair
  * @len - the length
  *
diff --git a/include/rdma/uverbs_named_ioctl.h b/include/rdma/uverbs_named_ioctl.h
index 3447bfe356d6..6ae6cf8e4c2e 100644
--- a/include/rdma/uverbs_named_ioctl.h
+++ b/include/rdma/uverbs_named_ioctl.h
@@ -76,7 +76,7 @@
 #define DECLARE_UVERBS_NAMED_OBJECT(_object_id, _type_attrs, ...)              \
 	static const struct uverbs_method_def *const UVERBS_OBJECT_METHODS(    \
 		_object_id)[] = { __VA_ARGS__ };                               \
-	const struct uverbs_object_def UVERBS_OBJECT(_object_id) = {           \
+	static const struct uverbs_object_def UVERBS_OBJECT(_object_id) = {    \
 		.id = _object_id,                                              \
 		.type_attrs = &_type_attrs,                                    \
 		.num_methods = ARRAY_SIZE(UVERBS_OBJECT_METHODS(_object_id)),  \
@@ -88,10 +88,10 @@
  * identify all uapi methods with a (object,method) tuple. However, they have
  * no type pointer.
  */
-#define DECLARE_UVERBS_GLOBAL_METHODS(_object_id, ...)	\
+#define DECLARE_UVERBS_GLOBAL_METHODS(_object_id, ...)                         \
 	static const struct uverbs_method_def *const UVERBS_OBJECT_METHODS(    \
 		_object_id)[] = { __VA_ARGS__ };                               \
-	const struct uverbs_object_def UVERBS_OBJECT(_object_id) = {           \
+	static const struct uverbs_object_def UVERBS_OBJECT(_object_id) = {    \
 		.id = _object_id,                                              \
 		.num_methods = ARRAY_SIZE(UVERBS_OBJECT_METHODS(_object_id)),  \
 		.methods = &UVERBS_OBJECT_METHODS(_object_id)                  \
diff --git a/include/rdma/uverbs_std_types.h b/include/rdma/uverbs_std_types.h
index 05eabfd5d0d3..1b28ce1aba07 100644
--- a/include/rdma/uverbs_std_types.h
+++ b/include/rdma/uverbs_std_types.h
@@ -104,16 +104,6 @@ static inline void uobj_put_write(struct ib_uobject *uobj)
 	rdma_lookup_put_uobject(uobj, UVERBS_LOOKUP_WRITE);
 }
 
-static inline int __must_check
-uobj_alloc_commit(struct ib_uobject *uobj, struct uverbs_attr_bundle *attrs)
-{
-	int ret = rdma_alloc_commit_uobject(uobj, attrs);
-
-	if (ret)
-		return ret;
-	return 0;
-}
-
 static inline void uobj_alloc_abort(struct ib_uobject *uobj,
 				    struct uverbs_attr_bundle *attrs)
 {
@@ -124,8 +114,7 @@ static inline struct ib_uobject *
 __uobj_alloc(const struct uverbs_api_object *obj,
 	     struct uverbs_attr_bundle *attrs, struct ib_device **ib_dev)
 {
-	struct ib_uobject *uobj =
-		rdma_alloc_begin_uobject(obj, attrs->ufile, attrs);
+	struct ib_uobject *uobj = rdma_alloc_begin_uobject(obj, attrs);
 
 	if (!IS_ERR(uobj))
 		*ib_dev = attrs->context->device;
diff --git a/include/rdma/uverbs_types.h b/include/rdma/uverbs_types.h
index d57a5ba00c74..f1cbdae67250 100644
--- a/include/rdma/uverbs_types.h
+++ b/include/rdma/uverbs_types.h
@@ -83,9 +83,9 @@ enum rdma_lookup_mode {
  */
 struct uverbs_obj_type_class {
 	struct ib_uobject *(*alloc_begin)(const struct uverbs_api_object *obj,
-					  struct ib_uverbs_file *ufile);
+					  struct uverbs_attr_bundle *attrs);
 	/* This consumes the kref on uobj */
-	int (*alloc_commit)(struct ib_uobject *uobj);
+	void (*alloc_commit)(struct ib_uobject *uobj);
 	/* This does not consume the kref on uobj */
 	void (*alloc_abort)(struct ib_uobject *uobj);
 
@@ -98,7 +98,6 @@ struct uverbs_obj_type_class {
 				       enum rdma_remove_reason why,
 				       struct uverbs_attr_bundle *attrs);
 	void (*remove_handle)(struct ib_uobject *uobj);
-	u8    needs_kfree_rcu;
 };
 
 struct uverbs_obj_type {
@@ -138,23 +137,34 @@ struct ib_uobject *rdma_lookup_get_uobject(const struct uverbs_api_object *obj,
 void rdma_lookup_put_uobject(struct ib_uobject *uobj,
 			     enum rdma_lookup_mode mode);
 struct ib_uobject *rdma_alloc_begin_uobject(const struct uverbs_api_object *obj,
-					    struct ib_uverbs_file *ufile,
 					    struct uverbs_attr_bundle *attrs);
 void rdma_alloc_abort_uobject(struct ib_uobject *uobj,
 			      struct uverbs_attr_bundle *attrs);
-int __must_check rdma_alloc_commit_uobject(struct ib_uobject *uobj,
-					   struct uverbs_attr_bundle *attrs);
+void rdma_alloc_commit_uobject(struct ib_uobject *uobj,
+			       struct uverbs_attr_bundle *attrs);
+
+/*
+ * uverbs_uobject_get is called in order to increase the reference count on
+ * an uobject. This is useful when a handler wants to keep the uobject's memory
+ * alive, regardless if this uobject is still alive in the context's objects
+ * repository. Objects are put via uverbs_uobject_put.
+ */
+static inline void uverbs_uobject_get(struct ib_uobject *uobject)
+{
+	kref_get(&uobject->ref);
+}
+void uverbs_uobject_put(struct ib_uobject *uobject);
 
 struct uverbs_obj_fd_type {
 	/*
 	 * In fd based objects, uverbs_obj_type_ops points to generic
 	 * fd operations. In order to specialize the underlying types (e.g.
 	 * completion_channel), we use fops, name and flags for fd creation.
-	 * context_closed is called when the context is closed either when
-	 * the driver is removed or the process terminated.
+	 * destroy_object is called when the uobject is to be destroyed,
+	 * because the driver is removed or the FD is closed.
 	 */
 	struct uverbs_obj_type  type;
-	int (*context_closed)(struct ib_uobject *uobj,
+	int (*destroy_object)(struct ib_uobject *uobj,
 			      enum rdma_remove_reason why);
 	const struct file_operations	*fops;
 	const char			*name;
@@ -163,11 +173,11 @@ struct uverbs_obj_fd_type {
 
 extern const struct uverbs_obj_type_class uverbs_idr_class;
 extern const struct uverbs_obj_type_class uverbs_fd_class;
-void uverbs_close_fd(struct file *f);
+int uverbs_uobject_fd_release(struct inode *inode, struct file *filp);
 
 #define UVERBS_BUILD_BUG_ON(cond) (sizeof(char[1 - 2 * !!(cond)]) -	\
 				   sizeof(char))
-#define UVERBS_TYPE_ALLOC_FD(_obj_size, _context_closed, _fops, _name, _flags)\
+#define UVERBS_TYPE_ALLOC_FD(_obj_size, _destroy_object, _fops, _name, _flags) \
 	((&((const struct uverbs_obj_fd_type)				\
 	 {.type = {							\
 		.type_class = &uverbs_fd_class,				\
@@ -175,7 +185,7 @@ void uverbs_close_fd(struct file *f);
 			UVERBS_BUILD_BUG_ON((_obj_size) <               \
 					    sizeof(struct ib_uobject)), \
 	 },								\
-	 .context_closed = _context_closed,				\
+	 .destroy_object = _destroy_object,				\
 	 .fops = _fops,							\
 	 .name = _name,							\
 	 .flags = _flags}))->type)
diff --git a/include/trace/events/rdma_core.h b/include/trace/events/rdma_core.h
new file mode 100644
index 000000000000..17642aa54437
--- /dev/null
+++ b/include/trace/events/rdma_core.h
@@ -0,0 +1,394 @@
+/* SPDX-License-Identifier: GPL-2.0-only */
+/*
+ * Trace point definitions for core RDMA functions.
+ *
+ * Author: Chuck Lever <chuck.lever@oracle.com>
+ *
+ * Copyright (c) 2019, Oracle and/or its affiliates. All rights reserved.
+ */
+
+#undef TRACE_SYSTEM
+#define TRACE_SYSTEM rdma_core
+
+#if !defined(_TRACE_RDMA_CORE_H) || defined(TRACE_HEADER_MULTI_READ)
+#define _TRACE_RDMA_CORE_H
+
+#include <linux/tracepoint.h>
+#include <rdma/ib_verbs.h>
+
+/*
+ * enum ib_poll_context, from include/rdma/ib_verbs.h
+ */
+#define IB_POLL_CTX_LIST			\
+	ib_poll_ctx(DIRECT)			\
+	ib_poll_ctx(SOFTIRQ)			\
+	ib_poll_ctx(WORKQUEUE)			\
+	ib_poll_ctx_end(UNBOUND_WORKQUEUE)
+
+#undef ib_poll_ctx
+#undef ib_poll_ctx_end
+
+#define ib_poll_ctx(x)		TRACE_DEFINE_ENUM(IB_POLL_##x);
+#define ib_poll_ctx_end(x)	TRACE_DEFINE_ENUM(IB_POLL_##x);
+
+IB_POLL_CTX_LIST
+
+#undef ib_poll_ctx
+#undef ib_poll_ctx_end
+
+#define ib_poll_ctx(x)		{ IB_POLL_##x, #x },
+#define ib_poll_ctx_end(x)	{ IB_POLL_##x, #x }
+
+#define rdma_show_ib_poll_ctx(x) \
+		__print_symbolic(x, IB_POLL_CTX_LIST)
+
+/**
+ ** Completion Queue events
+ **/
+
+TRACE_EVENT(cq_schedule,
+	TP_PROTO(
+		struct ib_cq *cq
+	),
+
+	TP_ARGS(cq),
+
+	TP_STRUCT__entry(
+		__field(u32, cq_id)
+	),
+
+	TP_fast_assign(
+		cq->timestamp = ktime_get();
+		cq->interrupt = true;
+
+		__entry->cq_id = cq->res.id;
+	),
+
+	TP_printk("cq.id=%u", __entry->cq_id)
+);
+
+TRACE_EVENT(cq_reschedule,
+	TP_PROTO(
+		struct ib_cq *cq
+	),
+
+	TP_ARGS(cq),
+
+	TP_STRUCT__entry(
+		__field(u32, cq_id)
+	),
+
+	TP_fast_assign(
+		cq->timestamp = ktime_get();
+		cq->interrupt = false;
+
+		__entry->cq_id = cq->res.id;
+	),
+
+	TP_printk("cq.id=%u", __entry->cq_id)
+);
+
+TRACE_EVENT(cq_process,
+	TP_PROTO(
+		const struct ib_cq *cq
+	),
+
+	TP_ARGS(cq),
+
+	TP_STRUCT__entry(
+		__field(u32, cq_id)
+		__field(bool, interrupt)
+		__field(s64, latency)
+	),
+
+	TP_fast_assign(
+		ktime_t latency = ktime_sub(ktime_get(), cq->timestamp);
+
+		__entry->cq_id = cq->res.id;
+		__entry->latency = ktime_to_us(latency);
+		__entry->interrupt = cq->interrupt;
+	),
+
+	TP_printk("cq.id=%u wake-up took %lld [us] from %s",
+		__entry->cq_id, __entry->latency,
+		__entry->interrupt ? "interrupt" : "reschedule"
+	)
+);
+
+TRACE_EVENT(cq_poll,
+	TP_PROTO(
+		const struct ib_cq *cq,
+		int requested,
+		int rc
+	),
+
+	TP_ARGS(cq, requested, rc),
+
+	TP_STRUCT__entry(
+		__field(u32, cq_id)
+		__field(int, requested)
+		__field(int, rc)
+	),
+
+	TP_fast_assign(
+		__entry->cq_id = cq->res.id;
+		__entry->requested = requested;
+		__entry->rc = rc;
+	),
+
+	TP_printk("cq.id=%u requested %d, returned %d",
+		__entry->cq_id, __entry->requested, __entry->rc
+	)
+);
+
+TRACE_EVENT(cq_drain_complete,
+	TP_PROTO(
+		const struct ib_cq *cq
+	),
+
+	TP_ARGS(cq),
+
+	TP_STRUCT__entry(
+		__field(u32, cq_id)
+	),
+
+	TP_fast_assign(
+		__entry->cq_id = cq->res.id;
+	),
+
+	TP_printk("cq.id=%u",
+		__entry->cq_id
+	)
+);
+
+
+TRACE_EVENT(cq_modify,
+	TP_PROTO(
+		const struct ib_cq *cq,
+		u16 comps,
+		u16 usec
+	),
+
+	TP_ARGS(cq, comps, usec),
+
+	TP_STRUCT__entry(
+		__field(u32, cq_id)
+		__field(unsigned int, comps)
+		__field(unsigned int, usec)
+	),
+
+	TP_fast_assign(
+		__entry->cq_id = cq->res.id;
+		__entry->comps = comps;
+		__entry->usec = usec;
+	),
+
+	TP_printk("cq.id=%u comps=%u usec=%u",
+		__entry->cq_id, __entry->comps, __entry->usec
+	)
+);
+
+TRACE_EVENT(cq_alloc,
+	TP_PROTO(
+		const struct ib_cq *cq,
+		int nr_cqe,
+		int comp_vector,
+		enum ib_poll_context poll_ctx
+	),
+
+	TP_ARGS(cq, nr_cqe, comp_vector, poll_ctx),
+
+	TP_STRUCT__entry(
+		__field(u32, cq_id)
+		__field(int, nr_cqe)
+		__field(int, comp_vector)
+		__field(unsigned long, poll_ctx)
+	),
+
+	TP_fast_assign(
+		__entry->cq_id = cq->res.id;
+		__entry->nr_cqe = nr_cqe;
+		__entry->comp_vector = comp_vector;
+		__entry->poll_ctx = poll_ctx;
+	),
+
+	TP_printk("cq.id=%u nr_cqe=%d comp_vector=%d poll_ctx=%s",
+		__entry->cq_id, __entry->nr_cqe, __entry->comp_vector,
+		rdma_show_ib_poll_ctx(__entry->poll_ctx)
+	)
+);
+
+TRACE_EVENT(cq_alloc_error,
+	TP_PROTO(
+		int nr_cqe,
+		int comp_vector,
+		enum ib_poll_context poll_ctx,
+		int rc
+	),
+
+	TP_ARGS(nr_cqe, comp_vector, poll_ctx, rc),
+
+	TP_STRUCT__entry(
+		__field(int, rc)
+		__field(int, nr_cqe)
+		__field(int, comp_vector)
+		__field(unsigned long, poll_ctx)
+	),
+
+	TP_fast_assign(
+		__entry->rc = rc;
+		__entry->nr_cqe = nr_cqe;
+		__entry->comp_vector = comp_vector;
+		__entry->poll_ctx = poll_ctx;
+	),
+
+	TP_printk("nr_cqe=%d comp_vector=%d poll_ctx=%s rc=%d",
+		__entry->nr_cqe, __entry->comp_vector,
+		rdma_show_ib_poll_ctx(__entry->poll_ctx), __entry->rc
+	)
+);
+
+TRACE_EVENT(cq_free,
+	TP_PROTO(
+		const struct ib_cq *cq
+	),
+
+	TP_ARGS(cq),
+
+	TP_STRUCT__entry(
+		__field(u32, cq_id)
+	),
+
+	TP_fast_assign(
+		__entry->cq_id = cq->res.id;
+	),
+
+	TP_printk("cq.id=%u", __entry->cq_id)
+);
+
+/**
+ ** Memory Region events
+ **/
+
+/*
+ * enum ib_mr_type, from include/rdma/ib_verbs.h
+ */
+#define IB_MR_TYPE_LIST				\
+	ib_mr_type_item(MEM_REG)		\
+	ib_mr_type_item(SG_GAPS)		\
+	ib_mr_type_item(DM)			\
+	ib_mr_type_item(USER)			\
+	ib_mr_type_item(DMA)			\
+	ib_mr_type_end(INTEGRITY)
+
+#undef ib_mr_type_item
+#undef ib_mr_type_end
+
+#define ib_mr_type_item(x)	TRACE_DEFINE_ENUM(IB_MR_TYPE_##x);
+#define ib_mr_type_end(x)	TRACE_DEFINE_ENUM(IB_MR_TYPE_##x);
+
+IB_MR_TYPE_LIST
+
+#undef ib_mr_type_item
+#undef ib_mr_type_end
+
+#define ib_mr_type_item(x)	{ IB_MR_TYPE_##x, #x },
+#define ib_mr_type_end(x)	{ IB_MR_TYPE_##x, #x }
+
+#define rdma_show_ib_mr_type(x) \
+		__print_symbolic(x, IB_MR_TYPE_LIST)
+
+TRACE_EVENT(mr_alloc,
+	TP_PROTO(
+		const struct ib_pd *pd,
+		enum ib_mr_type mr_type,
+		u32 max_num_sg,
+		const struct ib_mr *mr
+	),
+
+	TP_ARGS(pd, mr_type, max_num_sg, mr),
+
+	TP_STRUCT__entry(
+		__field(u32, pd_id)
+		__field(u32, mr_id)
+		__field(u32, max_num_sg)
+		__field(int, rc)
+		__field(unsigned long, mr_type)
+	),
+
+	TP_fast_assign(
+		__entry->pd_id = pd->res.id;
+		if (IS_ERR(mr)) {
+			__entry->mr_id = 0;
+			__entry->rc = PTR_ERR(mr);
+		} else {
+			__entry->mr_id = mr->res.id;
+			__entry->rc = 0;
+		}
+		__entry->max_num_sg = max_num_sg;
+		__entry->mr_type = mr_type;
+	),
+
+	TP_printk("pd.id=%u mr.id=%u type=%s max_num_sg=%u rc=%d",
+		__entry->pd_id, __entry->mr_id,
+		rdma_show_ib_mr_type(__entry->mr_type),
+		__entry->max_num_sg, __entry->rc)
+);
+
+TRACE_EVENT(mr_integ_alloc,
+	TP_PROTO(
+		const struct ib_pd *pd,
+		u32 max_num_data_sg,
+		u32 max_num_meta_sg,
+		const struct ib_mr *mr
+	),
+
+	TP_ARGS(pd, max_num_data_sg, max_num_meta_sg, mr),
+
+	TP_STRUCT__entry(
+		__field(u32, pd_id)
+		__field(u32, mr_id)
+		__field(u32, max_num_data_sg)
+		__field(u32, max_num_meta_sg)
+		__field(int, rc)
+	),
+
+	TP_fast_assign(
+		__entry->pd_id = pd->res.id;
+		if (IS_ERR(mr)) {
+			__entry->mr_id = 0;
+			__entry->rc = PTR_ERR(mr);
+		} else {
+			__entry->mr_id = mr->res.id;
+			__entry->rc = 0;
+		}
+		__entry->max_num_data_sg = max_num_data_sg;
+		__entry->max_num_meta_sg = max_num_meta_sg;
+	),
+
+	TP_printk("pd.id=%u mr.id=%u max_num_data_sg=%u max_num_meta_sg=%u rc=%d",
+		__entry->pd_id, __entry->mr_id, __entry->max_num_data_sg,
+		__entry->max_num_meta_sg, __entry->rc)
+);
+
+TRACE_EVENT(mr_dereg,
+	TP_PROTO(
+		const struct ib_mr *mr
+	),
+
+	TP_ARGS(mr),
+
+	TP_STRUCT__entry(
+		__field(u32, id)
+	),
+
+	TP_fast_assign(
+		__entry->id = mr->res.id;
+	),
+
+	TP_printk("mr.id=%u", __entry->id)
+);
+
+#endif /* _TRACE_RDMA_CORE_H */
+
+#include <trace/define_trace.h>
diff --git a/include/uapi/rdma/ib_user_ioctl_cmds.h b/include/uapi/rdma/ib_user_ioctl_cmds.h
index 64f0e3aacd3f..d4ddbe4e696c 100644
--- a/include/uapi/rdma/ib_user_ioctl_cmds.h
+++ b/include/uapi/rdma/ib_user_ioctl_cmds.h
@@ -56,6 +56,7 @@ enum uverbs_default_objects {
 	UVERBS_OBJECT_FLOW_ACTION,
 	UVERBS_OBJECT_DM,
 	UVERBS_OBJECT_COUNTERS,
+	UVERBS_OBJECT_ASYNC_EVENT,
 };
 
 enum {
@@ -67,6 +68,7 @@ enum uverbs_methods_device {
 	UVERBS_METHOD_INVOKE_WRITE,
 	UVERBS_METHOD_INFO_HANDLES,
 	UVERBS_METHOD_QUERY_PORT,
+	UVERBS_METHOD_GET_CONTEXT,
 };
 
 enum uverbs_attrs_invoke_write_cmd_attr_ids {
@@ -80,6 +82,11 @@ enum uverbs_attrs_query_port_cmd_attr_ids {
 	UVERBS_ATTR_QUERY_PORT_RESP,
 };
 
+enum uverbs_attrs_get_context_attr_ids {
+	UVERBS_ATTR_GET_CONTEXT_NUM_COMP_VECTORS,
+	UVERBS_ATTR_GET_CONTEXT_CORE_SUPPORT,
+};
+
 enum uverbs_attrs_create_cq_cmd_attr_ids {
 	UVERBS_ATTR_CREATE_CQ_HANDLE,
 	UVERBS_ATTR_CREATE_CQ_CQE,
@@ -241,4 +248,12 @@ enum uverbs_attrs_flow_destroy_ids {
 	UVERBS_ATTR_DESTROY_FLOW_HANDLE,
 };
 
+enum uverbs_method_async_event {
+	UVERBS_METHOD_ASYNC_EVENT_ALLOC,
+};
+
+enum uverbs_attrs_async_event_create {
+	UVERBS_ATTR_ASYNC_EVENT_ALLOC_FD_HANDLE,
+};
+
 #endif
diff --git a/include/uapi/rdma/ib_user_ioctl_verbs.h b/include/uapi/rdma/ib_user_ioctl_verbs.h
index 9019b2d906ea..a640bb814be0 100644
--- a/include/uapi/rdma/ib_user_ioctl_verbs.h
+++ b/include/uapi/rdma/ib_user_ioctl_verbs.h
@@ -41,6 +41,13 @@
 #define RDMA_UAPI_PTR(_type, _name)	__aligned_u64 _name
 #endif
 
+#define IB_UVERBS_ACCESS_OPTIONAL_FIRST (1 << 20)
+#define IB_UVERBS_ACCESS_OPTIONAL_LAST (1 << 29)
+
+enum ib_uverbs_core_support {
+	IB_UVERBS_CORE_SUPPORT_OPTIONAL_MR_ACCESS = 1 << 0,
+};
+
 enum ib_uverbs_access_flags {
 	IB_UVERBS_ACCESS_LOCAL_WRITE = 1 << 0,
 	IB_UVERBS_ACCESS_REMOTE_WRITE = 1 << 1,
@@ -50,6 +57,11 @@ enum ib_uverbs_access_flags {
 	IB_UVERBS_ACCESS_ZERO_BASED = 1 << 5,
 	IB_UVERBS_ACCESS_ON_DEMAND = 1 << 6,
 	IB_UVERBS_ACCESS_HUGETLB = 1 << 7,
+
+	IB_UVERBS_ACCESS_RELAXED_ORDERING = IB_UVERBS_ACCESS_OPTIONAL_FIRST,
+	IB_UVERBS_ACCESS_OPTIONAL_RANGE =
+		((IB_UVERBS_ACCESS_OPTIONAL_LAST << 1) - 1) &
+		~(IB_UVERBS_ACCESS_OPTIONAL_FIRST - 1)
 };
 
 enum ib_uverbs_query_port_cap_flags {
diff --git a/include/uapi/rdma/mlx5_user_ioctl_cmds.h b/include/uapi/rdma/mlx5_user_ioctl_cmds.h
index 20d88307f75f..afe7da6f2b8e 100644
--- a/include/uapi/rdma/mlx5_user_ioctl_cmds.h
+++ b/include/uapi/rdma/mlx5_user_ioctl_cmds.h
@@ -115,6 +115,22 @@ enum mlx5_ib_devx_obj_methods {
 	MLX5_IB_METHOD_DEVX_OBJ_ASYNC_QUERY,
 };
 
+enum mlx5_ib_var_alloc_attrs {
+	MLX5_IB_ATTR_VAR_OBJ_ALLOC_HANDLE = (1U << UVERBS_ID_NS_SHIFT),
+	MLX5_IB_ATTR_VAR_OBJ_ALLOC_MMAP_OFFSET,
+	MLX5_IB_ATTR_VAR_OBJ_ALLOC_MMAP_LENGTH,
+	MLX5_IB_ATTR_VAR_OBJ_ALLOC_PAGE_ID,
+};
+
+enum mlx5_ib_var_obj_destroy_attrs {
+	MLX5_IB_ATTR_VAR_OBJ_DESTROY_HANDLE = (1U << UVERBS_ID_NS_SHIFT),
+};
+
+enum mlx5_ib_var_obj_methods {
+	MLX5_IB_METHOD_VAR_OBJ_ALLOC = (1U << UVERBS_ID_NS_SHIFT),
+	MLX5_IB_METHOD_VAR_OBJ_DESTROY,
+};
+
 enum mlx5_ib_devx_umem_reg_attrs {
 	MLX5_IB_ATTR_DEVX_UMEM_REG_HANDLE = (1U << UVERBS_ID_NS_SHIFT),
 	MLX5_IB_ATTR_DEVX_UMEM_REG_ADDR,
@@ -156,6 +172,7 @@ enum mlx5_ib_objects {
 	MLX5_IB_OBJECT_FLOW_MATCHER,
 	MLX5_IB_OBJECT_DEVX_ASYNC_CMD_FD,
 	MLX5_IB_OBJECT_DEVX_ASYNC_EVENT_FD,
+	MLX5_IB_OBJECT_VAR,
 };
 
 enum mlx5_ib_flow_matcher_create_attrs {
diff --git a/include/uapi/rdma/qedr-abi.h b/include/uapi/rdma/qedr-abi.h
index c022ee26089b..a0b83c9d4498 100644
--- a/include/uapi/rdma/qedr-abi.h
+++ b/include/uapi/rdma/qedr-abi.h
@@ -48,6 +48,18 @@ struct qedr_alloc_ucontext_req {
 	__u32 reserved;
 };
 
+#define QEDR_LDPM_MAX_SIZE	(8192)
+#define QEDR_EDPM_TRANS_SIZE	(64)
+
+enum qedr_rdma_dpm_type {
+	QEDR_DPM_TYPE_NONE		= 0,
+	QEDR_DPM_TYPE_ROCE_ENHANCED	= 1 << 0,
+	QEDR_DPM_TYPE_ROCE_LEGACY	= 1 << 1,
+	QEDR_DPM_TYPE_IWARP_LEGACY	= 1 << 2,
+	QEDR_DPM_TYPE_RESERVED		= 1 << 3,
+	QEDR_DPM_SIZES_SET		= 1 << 4,
+};
+
 struct qedr_alloc_ucontext_resp {
 	__aligned_u64 db_pa;
 	__u32 db_size;
@@ -59,10 +71,12 @@ struct qedr_alloc_ucontext_resp {
 	__u32 sges_per_recv_wr;
 	__u32 sges_per_srq_wr;
 	__u32 max_cqes;
-	__u8 dpm_enabled;
+	__u8 dpm_flags;
 	__u8 wids_enabled;
 	__u16 wid_count;
-	__u32 reserved;
+	__u16 ldpm_limit_size;
+	__u8 edpm_trans_size;
+	__u8 reserved;
 };
 
 struct qedr_alloc_pd_ureq {