Merge branch 'k.o/for-4.12' into k.o/for-4.12-rdma-netdevice

drivers/infiniband/core/Makefile

@@ -29,4 +29,5 @@ ib_umad-y :=			user_mad.o
 
 ib_ucm-y :=			ucm.o
 
-ib_uverbs-y :=			uverbs_main.o uverbs_cmd.o uverbs_marshall.o
+ib_uverbs-y :=			uverbs_main.o uverbs_cmd.o uverbs_marshall.o \
+				rdma_core.o uverbs_std_types.o

drivers/infiniband/core/rdma_core.c

+/*
+ * Copyright (c) 2016, Mellanox Technologies inc.  All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include <linux/file.h>
+#include <linux/anon_inodes.h>
+#include <rdma/ib_verbs.h>
+#include <rdma/uverbs_types.h>
+#include <linux/rcupdate.h>
+#include "uverbs.h"
+#include "core_priv.h"
+#include "rdma_core.h"
+
+void uverbs_uobject_get(struct ib_uobject *uobject)
+{
+	kref_get(&uobject->ref);
+}
+
+static void uverbs_uobject_free(struct kref *ref)
+{
+	struct ib_uobject *uobj =
+		container_of(ref, struct ib_uobject, ref);
+
+	if (uobj->type->type_class->needs_kfree_rcu)
+		kfree_rcu(uobj, rcu);
+	else
+		kfree(uobj);
+}
+
+void uverbs_uobject_put(struct ib_uobject *uobject)
+{
+	kref_put(&uobject->ref, uverbs_uobject_free);
+}
+
+static int uverbs_try_lock_object(struct ib_uobject *uobj, bool exclusive)
+{
+	/*
+	 * When a shared access is required, we use a positive counter. Each
+	 * shared access request checks that the value != -1 and increment it.
+	 * Exclusive access is required for operations like write or destroy.
+	 * In exclusive access mode, we check that the counter is zero (nobody
+	 * claimed this object) and we set it to -1. Releasing a shared access
+	 * lock is done simply by decreasing the counter. As for exclusive
+	 * access locks, since only a single one of them is is allowed
+	 * concurrently, setting the counter to zero is enough for releasing
+	 * this lock.
+	 */
+	if (!exclusive)
+		return __atomic_add_unless(&uobj->usecnt, 1, -1) == -1 ?
+			-EBUSY : 0;
+
+	/* lock is either WRITE or DESTROY - should be exclusive */
+	return atomic_cmpxchg(&uobj->usecnt, 0, -1) == 0 ? 0 : -EBUSY;
+}
+
+static struct ib_uobject *alloc_uobj(struct ib_ucontext *context,
+				     const struct uverbs_obj_type *type)
+{
+	struct ib_uobject *uobj = kzalloc(type->obj_size, GFP_KERNEL);
+
+	if (!uobj)
+		return ERR_PTR(-ENOMEM);
+	/*
+	 * user_handle should be filled by the handler,
+	 * The object is added to the list in the commit stage.
+	 */
+	uobj->context = context;
+	uobj->type = type;
+	atomic_set(&uobj->usecnt, 0);
+	kref_init(&uobj->ref);
+
+	return uobj;
+}
+
+static int idr_add_uobj(struct ib_uobject *uobj)
+{
+	int ret;
+
+	idr_preload(GFP_KERNEL);
+	spin_lock(&uobj->context->ufile->idr_lock);
+
+	/*
+	 * We start with allocating an idr pointing to NULL. This represents an
+	 * object which isn't initialized yet. We'll replace it later on with
+	 * the real object once we commit.
+	 */
+	ret = idr_alloc(&uobj->context->ufile->idr, NULL, 0,
+			min_t(unsigned long, U32_MAX - 1, INT_MAX), GFP_NOWAIT);
+	if (ret >= 0)
+		uobj->id = ret;
+
+	spin_unlock(&uobj->context->ufile->idr_lock);
+	idr_preload_end();
+
+	return ret < 0 ? ret : 0;
+}
+
+/*
+ * It only removes it from the uobjects list, uverbs_uobject_put() is still
+ * required.
+ */
+static void uverbs_idr_remove_uobj(struct ib_uobject *uobj)
+{
+	spin_lock(&uobj->context->ufile->idr_lock);
+	idr_remove(&uobj->context->ufile->idr, uobj->id);
+	spin_unlock(&uobj->context->ufile->idr_lock);
+}
+
+/* Returns the ib_uobject or an error. The caller should check for IS_ERR. */
+static struct ib_uobject *lookup_get_idr_uobject(const struct uverbs_obj_type *type,
+						 struct ib_ucontext *ucontext,
+						 int id, bool exclusive)
+{
+	struct ib_uobject *uobj;
+
+	rcu_read_lock();
+	/* object won't be released as we're protected in rcu */
+	uobj = idr_find(&ucontext->ufile->idr, id);
+	if (!uobj) {
+		uobj = ERR_PTR(-ENOENT);
+		goto free;
+	}
+
+	uverbs_uobject_get(uobj);
+free:
+	rcu_read_unlock();
+	return uobj;
+}
+
+static struct ib_uobject *lookup_get_fd_uobject(const struct uverbs_obj_type *type,
+						struct ib_ucontext *ucontext,
+						int id, bool exclusive)
+{
+	struct file *f;
+	struct ib_uobject *uobject;
+	const struct uverbs_obj_fd_type *fd_type =
+		container_of(type, struct uverbs_obj_fd_type, type);
+
+	if (exclusive)
+		return ERR_PTR(-EOPNOTSUPP);
+
+	f = fget(id);
+	if (!f)
+		return ERR_PTR(-EBADF);
+
+	uobject = f->private_data;
+	/*
+	 * fget(id) ensures we are not currently running uverbs_close_fd,
+	 * and the caller is expected to ensure that uverbs_close_fd is never
+	 * done while a call top lookup is possible.
+	 */
+	if (f->f_op != fd_type->fops) {
+		fput(f);
+		return ERR_PTR(-EBADF);
+	}
+
+	uverbs_uobject_get(uobject);
+	return uobject;
+}
+
+struct ib_uobject *rdma_lookup_get_uobject(const struct uverbs_obj_type *type,
+					   struct ib_ucontext *ucontext,
+					   int id, bool exclusive)
+{
+	struct ib_uobject *uobj;
+	int ret;
+
+	uobj = type->type_class->lookup_get(type, ucontext, id, exclusive);
+	if (IS_ERR(uobj))
+		return uobj;
+
+	if (uobj->type != type) {
+		ret = -EINVAL;
+		goto free;
+	}
+
+	ret = uverbs_try_lock_object(uobj, exclusive);
+	if (ret) {
+		WARN(ucontext->cleanup_reason,
+		     "ib_uverbs: Trying to lookup_get while cleanup context\n");
+		goto free;
+	}
+
+	return uobj;
+free:
+	uobj->type->type_class->lookup_put(uobj, exclusive);
+	uverbs_uobject_put(uobj);
+	return ERR_PTR(ret);
+}
+
+static struct ib_uobject *alloc_begin_idr_uobject(const struct uverbs_obj_type *type,
+						  struct ib_ucontext *ucontext)
+{
+	int ret;
+	struct ib_uobject *uobj;
+
+	uobj = alloc_uobj(ucontext, type);
+	if (IS_ERR(uobj))
+		return uobj;
+
+	ret = idr_add_uobj(uobj);
+	if (ret)
+		goto uobj_put;
+
+	ret = ib_rdmacg_try_charge(&uobj->cg_obj, ucontext->device,
+				   RDMACG_RESOURCE_HCA_OBJECT);
+	if (ret)
+		goto idr_remove;
+
+	return uobj;
+
+idr_remove:
+	uverbs_idr_remove_uobj(uobj);
+uobj_put:
+	uverbs_uobject_put(uobj);
+	return ERR_PTR(ret);
+}
+
+static struct ib_uobject *alloc_begin_fd_uobject(const struct uverbs_obj_type *type,
+						 struct ib_ucontext *ucontext)
+{
+	const struct uverbs_obj_fd_type *fd_type =
+		container_of(type, struct uverbs_obj_fd_type, type);
+	int new_fd;
+	struct ib_uobject *uobj;
+	struct ib_uobject_file *uobj_file;
+	struct file *filp;
+
+	new_fd = get_unused_fd_flags(O_CLOEXEC);
+	if (new_fd < 0)
+		return ERR_PTR(new_fd);
+
+	uobj = alloc_uobj(ucontext, type);
+	if (IS_ERR(uobj)) {
+		put_unused_fd(new_fd);
+		return uobj;
+	}
+
+	uobj_file = container_of(uobj, struct ib_uobject_file, uobj);
+	filp = anon_inode_getfile(fd_type->name,
+				  fd_type->fops,
+				  uobj_file,
+				  fd_type->flags);
+	if (IS_ERR(filp)) {
+		put_unused_fd(new_fd);
+		uverbs_uobject_put(uobj);
+		return (void *)filp;
+	}
+
+	uobj_file->uobj.id = new_fd;
+	uobj_file->uobj.object = filp;
+	uobj_file->ufile = ucontext->ufile;
+	INIT_LIST_HEAD(&uobj->list);
+	kref_get(&uobj_file->ufile->ref);
+
+	return uobj;
+}
+
+struct ib_uobject *rdma_alloc_begin_uobject(const struct uverbs_obj_type *type,
+					    struct ib_ucontext *ucontext)
+{
+	return type->type_class->alloc_begin(type, ucontext);
+}
+
+static void uverbs_uobject_add(struct ib_uobject *uobject)
+{
+	mutex_lock(&uobject->context->uobjects_lock);
+	list_add(&uobject->list, &uobject->context->uobjects);
+	mutex_unlock(&uobject->context->uobjects_lock);
+}
+
+static int __must_check remove_commit_idr_uobject(struct ib_uobject *uobj,
+						  enum rdma_remove_reason why)
+{
+	const struct uverbs_obj_idr_type *idr_type =
+		container_of(uobj->type, struct uverbs_obj_idr_type,
+			     type);
+	int ret = idr_type->destroy_object(uobj, why);
+
+	/*
+	 * We can only fail gracefully if the user requested to destroy the
+	 * object. In the rest of the cases, just remove whatever you can.
+	 */
+	if (why == RDMA_REMOVE_DESTROY && ret)
+		return ret;
+
+	ib_rdmacg_uncharge(&uobj->cg_obj, uobj->context->device,
+			   RDMACG_RESOURCE_HCA_OBJECT);
+	uverbs_idr_remove_uobj(uobj);
+
+	return ret;
+}
+
+static void alloc_abort_fd_uobject(struct ib_uobject *uobj)
+{
+	struct ib_uobject_file *uobj_file =
+		container_of(uobj, struct ib_uobject_file, uobj);
+	struct file *filp = uobj->object;
+	int id = uobj_file->uobj.id;
+
+	/* Unsuccessful NEW */
+	fput(filp);
+	put_unused_fd(id);
+}
+
+static int __must_check remove_commit_fd_uobject(struct ib_uobject *uobj,
+						 enum rdma_remove_reason why)
+{
+	const struct uverbs_obj_fd_type *fd_type =
+		container_of(uobj->type, struct uverbs_obj_fd_type, type);
+	struct ib_uobject_file *uobj_file =
+		container_of(uobj, struct ib_uobject_file, uobj);
+	int ret = fd_type->context_closed(uobj_file, why);
+
+	if (why == RDMA_REMOVE_DESTROY && ret)
+		return ret;
+
+	if (why == RDMA_REMOVE_DURING_CLEANUP) {
+		alloc_abort_fd_uobject(uobj);
+		return ret;
+	}
+
+	uobj_file->uobj.context = NULL;
+	return ret;
+}
+
+static void lockdep_check(struct ib_uobject *uobj, bool exclusive)
+{
+#ifdef CONFIG_LOCKDEP
+	if (exclusive)
+		WARN_ON(atomic_read(&uobj->usecnt) > 0);
+	else
+		WARN_ON(atomic_read(&uobj->usecnt) == -1);
+#endif
+}
+
+static int __must_check _rdma_remove_commit_uobject(struct ib_uobject *uobj,
+						    enum rdma_remove_reason why)
+{
+	int ret;
+	struct ib_ucontext *ucontext = uobj->context;
+
+	ret = uobj->type->type_class->remove_commit(uobj, why);
+	if (ret && why == RDMA_REMOVE_DESTROY) {
+		/* We couldn't remove the object, so just unlock the uobject */
+		atomic_set(&uobj->usecnt, 0);
+		uobj->type->type_class->lookup_put(uobj, true);
+	} else {
+		mutex_lock(&ucontext->uobjects_lock);
+		list_del(&uobj->list);
+		mutex_unlock(&ucontext->uobjects_lock);
+		/* put the ref we took when we created the object */
+		uverbs_uobject_put(uobj);
+	}
+
+	return ret;
+}
+
+/* This is called only for user requested DESTROY reasons */
+int __must_check rdma_remove_commit_uobject(struct ib_uobject *uobj)
+{
+	int ret;
+	struct ib_ucontext *ucontext = uobj->context;
+
+	/* put the ref count we took at lookup_get */
+	uverbs_uobject_put(uobj);
+	/* Cleanup is running. Calling this should have been impossible */
+	if (!down_read_trylock(&ucontext->cleanup_rwsem)) {
+		WARN(true, "ib_uverbs: Cleanup is running while removing an uobject\n");
+		return 0;
+	}
+	lockdep_check(uobj, true);
+	ret = _rdma_remove_commit_uobject(uobj, RDMA_REMOVE_DESTROY);
+
+	up_read(&ucontext->cleanup_rwsem);
+	return ret;
+}
+
+static void alloc_commit_idr_uobject(struct ib_uobject *uobj)
+{
+	uverbs_uobject_add(uobj);
+	spin_lock(&uobj->context->ufile->idr_lock);
+	/*
+	 * We already allocated this IDR with a NULL object, so
+	 * this shouldn't fail.
+	 */
+	WARN_ON(idr_replace(&uobj->context->ufile->idr,
+			    uobj, uobj->id));
+	spin_unlock(&uobj->context->ufile->idr_lock);
+}
+
+static void alloc_commit_fd_uobject(struct ib_uobject *uobj)
+{
+	struct ib_uobject_file *uobj_file =
+		container_of(uobj, struct ib_uobject_file, uobj);
+
+	uverbs_uobject_add(&uobj_file->uobj);
+	fd_install(uobj_file->uobj.id, uobj->object);
+	/* This shouldn't be used anymore. Use the file object instead */
+	uobj_file->uobj.id = 0;
+	/* Get another reference as we export this to the fops */
+	uverbs_uobject_get(&uobj_file->uobj);
+}
+
+int rdma_alloc_commit_uobject(struct ib_uobject *uobj)
+{
+	/* Cleanup is running. Calling this should have been impossible */
+	if (!down_read_trylock(&uobj->context->cleanup_rwsem)) {
+		int ret;
+
+		WARN(true, "ib_uverbs: Cleanup is running while allocating an uobject\n");
+		ret = uobj->type->type_class->remove_commit(uobj,
+							    RDMA_REMOVE_DURING_CLEANUP);
+		if (ret)
+			pr_warn("ib_uverbs: cleanup of idr object %d failed\n",
+				uobj->id);
+		return ret;
+	}
+
+	uobj->type->type_class->alloc_commit(uobj);
+	up_read(&uobj->context->cleanup_rwsem);
+
+	return 0;
+}
+
+static void alloc_abort_idr_uobject(struct ib_uobject *uobj)
+{
+	uverbs_idr_remove_uobj(uobj);
+	ib_rdmacg_uncharge(&uobj->cg_obj, uobj->context->device,
+			   RDMACG_RESOURCE_HCA_OBJECT);
+	uverbs_uobject_put(uobj);
+}
+
+void rdma_alloc_abort_uobject(struct ib_uobject *uobj)
+{
+	uobj->type->type_class->alloc_abort(uobj);
+}
+
+static void lookup_put_idr_uobject(struct ib_uobject *uobj, bool exclusive)
+{
+}
+
+static void lookup_put_fd_uobject(struct ib_uobject *uobj, bool exclusive)
+{
+	struct file *filp = uobj->object;
+
+	WARN_ON(exclusive);
+	/* This indirectly calls uverbs_close_fd and free the object */
+	fput(filp);
+}
+
+void rdma_lookup_put_uobject(struct ib_uobject *uobj, bool exclusive)
+{
+	lockdep_check(uobj, exclusive);
+	uobj->type->type_class->lookup_put(uobj, exclusive);
+	/*
+	 * In order to unlock an object, either decrease its usecnt for
+	 * read access or zero it in case of exclusive access. See
+	 * uverbs_try_lock_object for locking schema information.
+	 */
+	if (!exclusive)
+		atomic_dec(&uobj->usecnt);
+	else
+		atomic_set(&uobj->usecnt, 0);
+
+	uverbs_uobject_put(uobj);
+}
+
+const struct uverbs_obj_type_class uverbs_idr_class = {
+	.alloc_begin = alloc_begin_idr_uobject,
+	.lookup_get = lookup_get_idr_uobject,
+	.alloc_commit = alloc_commit_idr_uobject,
+	.alloc_abort = alloc_abort_idr_uobject,
+	.lookup_put = lookup_put_idr_uobject,
+	.remove_commit = remove_commit_idr_uobject,
+	/*
+	 * When we destroy an object, we first just lock it for WRITE and
+	 * actually DESTROY it in the finalize stage. So, the problematic
+	 * scenario is when we just started the finalize stage of the
+	 * destruction (nothing was executed yet). Now, the other thread
+	 * fetched the object for READ access, but it didn't lock it yet.
+	 * The DESTROY thread continues and starts destroying the object.
+	 * When the other thread continue - without the RCU, it would
+	 * access freed memory. However, the rcu_read_lock delays the free
+	 * until the rcu_read_lock of the READ operation quits. Since the
+	 * exclusive lock of the object is still taken by the DESTROY flow, the
+	 * READ operation will get -EBUSY and it'll just bail out.
+	 */
+	.needs_kfree_rcu = true,
+};
+
+static void _uverbs_close_fd(struct ib_uobject_file *uobj_file)
+{
+	struct ib_ucontext *ucontext;
+	struct ib_uverbs_file *ufile = uobj_file->ufile;
+	int ret;
+
+	mutex_lock(&uobj_file->ufile->cleanup_mutex);
+
+	/* uobject was either already cleaned up or is cleaned up right now anyway */
+	if (!uobj_file->uobj.context ||
+	    !down_read_trylock(&uobj_file->uobj.context->cleanup_rwsem))
+		goto unlock;
+
+	ucontext = uobj_file->uobj.context;
+	ret = _rdma_remove_commit_uobject(&uobj_file->uobj, RDMA_REMOVE_CLOSE);
+	up_read(&ucontext->cleanup_rwsem);
+	if (ret)
+		pr_warn("uverbs: unable to clean up uobject file in uverbs_close_fd.\n");
+unlock:
+	mutex_unlock(&ufile->cleanup_mutex);
+}
+
+void uverbs_close_fd(struct file *f)
+{
+	struct ib_uobject_file *uobj_file = f->private_data;
+	struct kref *uverbs_file_ref = &uobj_file->ufile->ref;
+
+	_uverbs_close_fd(uobj_file);
+	uverbs_uobject_put(&uobj_file->uobj);
+	kref_put(uverbs_file_ref, ib_uverbs_release_file);
+}
+
+void uverbs_cleanup_ucontext(struct ib_ucontext *ucontext, bool device_removed)
+{
+	enum rdma_remove_reason reason = device_removed ?
+		RDMA_REMOVE_DRIVER_REMOVE : RDMA_REMOVE_CLOSE;
+	unsigned int cur_order = 0;
+
+	ucontext->cleanup_reason = reason;
+	/*
+	 * Waits for all remove_commit and alloc_commit to finish. Logically, We
+	 * want to hold this forever as the context is going to be destroyed,
+	 * but we'll release it since it causes a "held lock freed" BUG message.
+	 */
+	down_write(&ucontext->cleanup_rwsem);
+
+	while (!list_empty(&ucontext->uobjects)) {
+		struct ib_uobject *obj, *next_obj;
+		unsigned int next_order = UINT_MAX;
+
+		/*
+		 * This shouldn't run while executing other commands on this
+		 * context. Thus, the only thing we should take care of is
+		 * releasing a FD while traversing this list. The FD could be
+		 * closed and released from the _release fop of this FD.
+		 * In order to mitigate this, we add a lock.
+		 * We take and release the lock per order traversal in order
+		 * to let other threads (which might still use the FDs) chance
+		 * to run.
+		 */
+		mutex_lock(&ucontext->uobjects_lock);
+		list_for_each_entry_safe(obj, next_obj, &ucontext->uobjects,
+					 list) {
+			if (obj->type->destroy_order == cur_order) {
+				int ret;
+
+				/*
+				 * if we hit this WARN_ON, that means we are
+				 * racing with a lookup_get.
+				 */
+				WARN_ON(uverbs_try_lock_object(obj, true));
+				ret = obj->type->type_class->remove_commit(obj,
+									   reason);
+				list_del(&obj->list);
+				if (ret)
+					pr_warn("ib_uverbs: failed to remove uobject id %d order %u\n",
+						obj->id, cur_order);
+				/* put the ref we took when we created the object */
+				uverbs_uobject_put(obj);
+			} else {
+				next_order = min(next_order,
+						 obj->type->destroy_order);
+			}
+		}
+		mutex_unlock(&ucontext->uobjects_lock);
+		cur_order = next_order;
+	}
+	up_write(&ucontext->cleanup_rwsem);
+}
+
+void uverbs_initialize_ucontext(struct ib_ucontext *ucontext)
+{
+	ucontext->cleanup_reason = 0;
+	mutex_init(&ucontext->uobjects_lock);
+	INIT_LIST_HEAD(&ucontext->uobjects);
+	init_rwsem(&ucontext->cleanup_rwsem);
+}
+
+const struct uverbs_obj_type_class uverbs_fd_class = {
+	.alloc_begin = alloc_begin_fd_uobject,
+	.lookup_get = lookup_get_fd_uobject,
+	.alloc_commit = alloc_commit_fd_uobject,
+	.alloc_abort = alloc_abort_fd_uobject,
+	.lookup_put = lookup_put_fd_uobject,
+	.remove_commit = remove_commit_fd_uobject,
+	.needs_kfree_rcu = false,
+};
+

drivers/infiniband/core/rdma_core.h

+/*
+ * Copyright (c) 2005 Topspin Communications.  All rights reserved.
+ * Copyright (c) 2005, 2006 Cisco Systems.  All rights reserved.
+ * Copyright (c) 2005-2017 Mellanox Technologies. All rights reserved.
+ * Copyright (c) 2005 Voltaire, Inc. All rights reserved.
+ * Copyright (c) 2005 PathScale, Inc. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#ifndef RDMA_CORE_H
+#define RDMA_CORE_H
+
+#include <linux/idr.h>
+#include <rdma/uverbs_types.h>
+#include <rdma/ib_verbs.h>
+#include <linux/mutex.h>
+
+/*
+ * These functions initialize the context and cleanups its uobjects.
+ * The context has a list of objects which is protected by a mutex
+ * on the context. initialize_ucontext should be called when we create
+ * a context.
+ * cleanup_ucontext removes all uobjects from the context and puts them.
+ */
+void uverbs_cleanup_ucontext(struct ib_ucontext *ucontext, bool device_removed);
+void uverbs_initialize_ucontext(struct ib_ucontext *ucontext);
+
+/*
+ * 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.
+ */
+void uverbs_uobject_get(struct ib_uobject *uobject);
+
+/*
+ * In order to indicate we no longer needs this uobject, uverbs_uobject_put
+ * is called. When the reference count is decreased, the uobject is freed.
+ * For example, this is used when attaching a completion channel to a CQ.
+ */
+void uverbs_uobject_put(struct ib_uobject *uobject);
+
+/* Indicate this fd is no longer used by this consumer, but its memory isn't
+ * necessarily released yet. When the last reference is put, we release the
+ * memory. After this call is executed, calling uverbs_uobject_get isn't
+ * allowed.
+ * This must be called from the release file_operations of the file!
+ */
+void uverbs_close_fd(struct file *f);
+
+#endif /* RDMA_CORE_H */

drivers/infiniband/core/uverbs.h

@@ -76,12 +76,13 @@
  * an asynchronous event queue file is created and released when the
  * event file is closed.
  *
- * struct ib_uverbs_event_file: One reference is held by the VFS and
- * released when the file is closed.  For asynchronous event files,
- * another reference is held by the corresponding main context file
- * and released when that file is closed.  For completion event files,
- * a reference is taken when a CQ is created that uses the file, and
- * released when the CQ is destroyed.
+ * struct ib_uverbs_event_queue: Base structure for
+ * struct ib_uverbs_async_event_file and struct ib_uverbs_completion_event_file.
+ * One reference is held by the VFS and released when the file is closed.
+ * For asynchronous event files, another reference is held by the corresponding
+ * main context file and released when that file is closed.  For completion
+ * event files, a reference is taken when a CQ is created that uses the file,
+ * and released when the CQ is destroyed.
  */
 
 struct ib_uverbs_device {
@@ -101,18 +102,26 @@ struct ib_uverbs_device {
 	struct list_head			uverbs_events_file_list;
 };
 
-struct ib_uverbs_event_file {
-	struct kref				ref;
-	int					is_async;
-	struct ib_uverbs_file		       *uverbs_file;
+struct ib_uverbs_event_queue {
 	spinlock_t				lock;
 	int					is_closed;
 	wait_queue_head_t			poll_wait;
 	struct fasync_struct		       *async_queue;
 	struct list_head			event_list;
+};
+
+struct ib_uverbs_async_event_file {
+	struct ib_uverbs_event_queue		ev_queue;
+	struct ib_uverbs_file		       *uverbs_file;
+	struct kref				ref;
 	struct list_head			list;
 };
 
+struct ib_uverbs_completion_event_file {
+	struct ib_uobject_file			uobj_file;
+	struct ib_uverbs_event_queue		ev_queue;
+};
+
 struct ib_uverbs_file {
 	struct kref				ref;
 	struct mutex				mutex;
@@ -120,9 +129,13 @@ struct ib_uverbs_file {
 	struct ib_uverbs_device		       *device;
 	struct ib_ucontext		       *ucontext;
 	struct ib_event_handler			event_handler;
-	struct ib_uverbs_event_file	       *async_file;
+	struct ib_uverbs_async_event_file       *async_file;
 	struct list_head			list;
 	int					is_closed;
+
+	struct idr		idr;
+	/* spinlock protects write access to idr */
+	spinlock_t		idr_lock;
 };
 
 struct ib_uverbs_event {
@@ -159,6 +172,8 @@ struct ib_usrq_object {
 
 struct ib_uqp_object {
 	struct ib_uevent_object	uevent;
+	/* lock for mcast list */
+	struct mutex		mcast_lock;
 	struct list_head 	mcast_list;
 	struct ib_uxrcd_object *uxrcd;
 };
@@ -176,32 +191,18 @@ struct ib_ucq_object {
 	u32			async_events_reported;
 };
 
-extern spinlock_t ib_uverbs_idr_lock;
-extern struct idr ib_uverbs_pd_idr;
-extern struct idr ib_uverbs_mr_idr;
-extern struct idr ib_uverbs_mw_idr;
-extern struct idr ib_uverbs_ah_idr;
-extern struct idr ib_uverbs_cq_idr;
-extern struct idr ib_uverbs_qp_idr;
-extern struct idr ib_uverbs_srq_idr;
-extern struct idr ib_uverbs_xrcd_idr;
-extern struct idr ib_uverbs_rule_idr;
-extern struct idr ib_uverbs_wq_idr;
-extern struct idr ib_uverbs_rwq_ind_tbl_idr;
-
-void idr_remove_uobj(struct idr *idp, struct ib_uobject *uobj);
-
-struct file *ib_uverbs_alloc_event_file(struct ib_uverbs_file *uverbs_file,
-					struct ib_device *ib_dev,
-					int is_async);
+extern const struct file_operations uverbs_event_fops;
+void ib_uverbs_init_event_queue(struct ib_uverbs_event_queue *ev_queue);
+struct file *ib_uverbs_alloc_async_event_file(struct ib_uverbs_file *uverbs_file,
+					      struct ib_device *ib_dev);
 void ib_uverbs_free_async_event_file(struct ib_uverbs_file *uverbs_file);
-struct ib_uverbs_event_file *ib_uverbs_lookup_comp_file(int fd);
 
 void ib_uverbs_release_ucq(struct ib_uverbs_file *file,
-			   struct ib_uverbs_event_file *ev_file,
+			   struct ib_uverbs_completion_event_file *ev_file,
 			   struct ib_ucq_object *uobj);
 void ib_uverbs_release_uevent(struct ib_uverbs_file *file,
 			      struct ib_uevent_object *uobj);
+void ib_uverbs_release_file(struct kref *ref);
 
 void ib_uverbs_comp_handler(struct ib_cq *cq, void *cq_context);
 void ib_uverbs_cq_event_handler(struct ib_event *event, void *context_ptr);
@@ -210,9 +211,12 @@ void ib_uverbs_wq_event_handler(struct ib_event *event, void *context_ptr);
 void ib_uverbs_srq_event_handler(struct ib_event *event, void *context_ptr);
 void ib_uverbs_event_handler(struct ib_event_handler *handler,
 			     struct ib_event *event);
-void ib_uverbs_dealloc_xrcd(struct ib_uverbs_device *dev, struct ib_xrcd *xrcd);
+int ib_uverbs_dealloc_xrcd(struct ib_uverbs_device *dev, struct ib_xrcd *xrcd,
+			   enum rdma_remove_reason why);
 
 int uverbs_dealloc_mw(struct ib_mw *mw);
+void ib_uverbs_detach_umcast(struct ib_qp *qp,
+			     struct ib_uqp_object *uobj);
 
 struct ib_uverbs_flow_spec {
 	union {

drivers/infiniband/core/uverbs_cmd.c

@@ -40,270 +40,29 @@
 
 #include <linux/uaccess.h>
 
+#include <rdma/uverbs_types.h>
+#include <rdma/uverbs_std_types.h>
+#include "rdma_core.h"
+
 #include "uverbs.h"
 #include "core_priv.h"
 
-struct uverbs_lock_class {
-	struct lock_class_key	key;
-	char			name[16];
-};
-
-static struct uverbs_lock_class pd_lock_class	= { .name = "PD-uobj" };
-static struct uverbs_lock_class mr_lock_class	= { .name = "MR-uobj" };
-static struct uverbs_lock_class mw_lock_class	= { .name = "MW-uobj" };
-static struct uverbs_lock_class cq_lock_class	= { .name = "CQ-uobj" };
-static struct uverbs_lock_class qp_lock_class	= { .name = "QP-uobj" };
-static struct uverbs_lock_class ah_lock_class	= { .name = "AH-uobj" };
-static struct uverbs_lock_class srq_lock_class	= { .name = "SRQ-uobj" };
-static struct uverbs_lock_class xrcd_lock_class = { .name = "XRCD-uobj" };
-static struct uverbs_lock_class rule_lock_class = { .name = "RULE-uobj" };
-static struct uverbs_lock_class wq_lock_class = { .name = "WQ-uobj" };
-static struct uverbs_lock_class rwq_ind_table_lock_class = { .name = "IND_TBL-uobj" };
-
-/*
- * The ib_uobject locking scheme is as follows:
- *
- * - ib_uverbs_idr_lock protects the uverbs idrs themselves, so it
- *   needs to be held during all idr write operations.  When an object is
- *   looked up, a reference must be taken on the object's kref before
- *   dropping this lock.  For read operations, the rcu_read_lock()
- *   and rcu_write_lock() but similarly the kref reference is grabbed
- *   before the rcu_read_unlock().
- *
- * - Each object also has an rwsem.  This rwsem must be held for
- *   reading while an operation that uses the object is performed.
- *   For example, while registering an MR, the associated PD's
- *   uobject.mutex must be held for reading.  The rwsem must be held
- *   for writing while initializing or destroying an object.
- *
- * - In addition, each object has a "live" flag.  If this flag is not
- *   set, then lookups of the object will fail even if it is found in
- *   the idr.  This handles a reader that blocks and does not acquire
- *   the rwsem until after the object is destroyed.  The destroy
- *   operation will set the live flag to 0 and then drop the rwsem;
- *   this will allow the reader to acquire the rwsem, see that the
- *   live flag is 0, and then drop the rwsem and its reference to
- *   object.  The underlying storage will not be freed until the last
- *   reference to the object is dropped.
- */
-
-static void init_uobj(struct ib_uobject *uobj, u64 user_handle,
-		      struct ib_ucontext *context, struct uverbs_lock_class *c)
-{
-	uobj->user_handle = user_handle;
-	uobj->context     = context;
-	kref_init(&uobj->ref);
-	init_rwsem(&uobj->mutex);
-	lockdep_set_class_and_name(&uobj->mutex, &c->key, c->name);
-	uobj->live        = 0;
-}
-
-static void release_uobj(struct kref *kref)
-{
-	kfree_rcu(container_of(kref, struct ib_uobject, ref), rcu);
-}
-
-static void put_uobj(struct ib_uobject *uobj)
-{
-	kref_put(&uobj->ref, release_uobj);
-}
-
-static void put_uobj_read(struct ib_uobject *uobj)
-{
-	up_read(&uobj->mutex);
-	put_uobj(uobj);
-}
-
-static void put_uobj_write(struct ib_uobject *uobj)
-{
-	up_write(&uobj->mutex);
-	put_uobj(uobj);
-}
-
-static int idr_add_uobj(struct idr *idr, struct ib_uobject *uobj)
-{
-	int ret;
-
-	idr_preload(GFP_KERNEL);
-	spin_lock(&ib_uverbs_idr_lock);
-
-	ret = idr_alloc(idr, uobj, 0, 0, GFP_NOWAIT);
-	if (ret >= 0)
-		uobj->id = ret;
-
-	spin_unlock(&ib_uverbs_idr_lock);
-	idr_preload_end();
-
-	return ret < 0 ? ret : 0;
-}
-
-void idr_remove_uobj(struct idr *idr, struct ib_uobject *uobj)
-{
-	spin_lock(&ib_uverbs_idr_lock);
-	idr_remove(idr, uobj->id);
-	spin_unlock(&ib_uverbs_idr_lock);
-}
-
-static struct ib_uobject *__idr_get_uobj(struct idr *idr, int id,
-					 struct ib_ucontext *context)
-{
-	struct ib_uobject *uobj;
-
-	rcu_read_lock();
-	uobj = idr_find(idr, id);
-	if (uobj) {
-		if (uobj->context == context)
-			kref_get(&uobj->ref);
-		else
-			uobj = NULL;
-	}
-	rcu_read_unlock();
-
-	return uobj;
-}
-
-static struct ib_uobject *idr_read_uobj(struct idr *idr, int id,
-					struct ib_ucontext *context, int nested)
-{
-	struct ib_uobject *uobj;
-
-	uobj = __idr_get_uobj(idr, id, context);
-	if (!uobj)
-		return NULL;
-
-	if (nested)
-		down_read_nested(&uobj->mutex, SINGLE_DEPTH_NESTING);
-	else
-		down_read(&uobj->mutex);
-	if (!uobj->live) {
-		put_uobj_read(uobj);
-		return NULL;
-	}
-
-	return uobj;
-}
-
-static struct ib_uobject *idr_write_uobj(struct idr *idr, int id,
-					 struct ib_ucontext *context)
-{
-	struct ib_uobject *uobj;
-
-	uobj = __idr_get_uobj(idr, id, context);
-	if (!uobj)
-		return NULL;
-
-	down_write(&uobj->mutex);
-	if (!uobj->live) {
-		put_uobj_write(uobj);
-		return NULL;
-	}
-
-	return uobj;
-}
-
-static void *idr_read_obj(struct idr *idr, int id, struct ib_ucontext *context,
-			  int nested)
-{
-	struct ib_uobject *uobj;
-
-	uobj = idr_read_uobj(idr, id, context, nested);
-	return uobj ? uobj->object : NULL;
-}
-
-static struct ib_pd *idr_read_pd(int pd_handle, struct ib_ucontext *context)
-{
-	return idr_read_obj(&ib_uverbs_pd_idr, pd_handle, context, 0);
-}
-
-static void put_pd_read(struct ib_pd *pd)
-{
-	put_uobj_read(pd->uobject);
-}
-
-static struct ib_cq *idr_read_cq(int cq_handle, struct ib_ucontext *context, int nested)
-{
-	return idr_read_obj(&ib_uverbs_cq_idr, cq_handle, context, nested);
-}
-
-static void put_cq_read(struct ib_cq *cq)
-{
-	put_uobj_read(cq->uobject);
-}
-
-static struct ib_ah *idr_read_ah(int ah_handle, struct ib_ucontext *context)
-{
-	return idr_read_obj(&ib_uverbs_ah_idr, ah_handle, context, 0);
-}
-
-static void put_ah_read(struct ib_ah *ah)
-{
-	put_uobj_read(ah->uobject);
-}
-
-static struct ib_qp *idr_read_qp(int qp_handle, struct ib_ucontext *context)
-{
-	return idr_read_obj(&ib_uverbs_qp_idr, qp_handle, context, 0);
-}
-
-static struct ib_wq *idr_read_wq(int wq_handle, struct ib_ucontext *context)
+static struct ib_uverbs_completion_event_file *
+ib_uverbs_lookup_comp_file(int fd, struct ib_ucontext *context)
 {
-	return idr_read_obj(&ib_uverbs_wq_idr, wq_handle, context, 0);
-}
+	struct ib_uobject *uobj = uobj_get_read(uobj_get_type(comp_channel),
+						fd, context);
+	struct ib_uobject_file *uobj_file;
 
-static void put_wq_read(struct ib_wq *wq)
-{
-	put_uobj_read(wq->uobject);
-}
+	if (IS_ERR(uobj))
+		return (void *)uobj;
 
-static struct ib_rwq_ind_table *idr_read_rwq_indirection_table(int ind_table_handle,
-							       struct ib_ucontext *context)
-{
-	return idr_read_obj(&ib_uverbs_rwq_ind_tbl_idr, ind_table_handle, context, 0);
-}
+	uverbs_uobject_get(uobj);
+	uobj_put_read(uobj);
 
-static void put_rwq_indirection_table_read(struct ib_rwq_ind_table *ind_table)
-{
-	put_uobj_read(ind_table->uobject);
-}
-
-static struct ib_qp *idr_write_qp(int qp_handle, struct ib_ucontext *context)
-{
-	struct ib_uobject *uobj;
-
-	uobj = idr_write_uobj(&ib_uverbs_qp_idr, qp_handle, context);
-	return uobj ? uobj->object : NULL;
-}
-
-static void put_qp_read(struct ib_qp *qp)
-{
-	put_uobj_read(qp->uobject);
-}
-
-static void put_qp_write(struct ib_qp *qp)
-{
-	put_uobj_write(qp->uobject);
-}
-
-static struct ib_srq *idr_read_srq(int srq_handle, struct ib_ucontext *context)
-{
-	return idr_read_obj(&ib_uverbs_srq_idr, srq_handle, context, 0);
-}
-
-static void put_srq_read(struct ib_srq *srq)
-{
-	put_uobj_read(srq->uobject);
-}
-
-static struct ib_xrcd *idr_read_xrcd(int xrcd_handle, struct ib_ucontext *context,
-				     struct ib_uobject **uobj)
-{
-	*uobj = idr_read_uobj(&ib_uverbs_xrcd_idr, xrcd_handle, context, 0);
-	return *uobj ? (*uobj)->object : NULL;
-}
-
-static void put_xrcd_read(struct ib_uobject *uobj)
-{
-	put_uobj_read(uobj);
+	uobj_file = container_of(uobj, struct ib_uobject_file, uobj);
+	return container_of(uobj_file, struct ib_uverbs_completion_event_file,
+			    uobj_file);
 }
 
 ssize_t ib_uverbs_get_context(struct ib_uverbs_file *file,
@@ -348,17 +107,10 @@ ssize_t ib_uverbs_get_context(struct ib_uverbs_file *file,
 
 	ucontext->device = ib_dev;
 	ucontext->cg_obj = cg_obj;
-	INIT_LIST_HEAD(&ucontext->pd_list);
-	INIT_LIST_HEAD(&ucontext->mr_list);
-	INIT_LIST_HEAD(&ucontext->mw_list);
-	INIT_LIST_HEAD(&ucontext->cq_list);
-	INIT_LIST_HEAD(&ucontext->qp_list);
-	INIT_LIST_HEAD(&ucontext->srq_list);
-	INIT_LIST_HEAD(&ucontext->ah_list);
-	INIT_LIST_HEAD(&ucontext->wq_list);
-	INIT_LIST_HEAD(&ucontext->rwq_ind_tbl_list);
-	INIT_LIST_HEAD(&ucontext->xrcd_list);
-	INIT_LIST_HEAD(&ucontext->rule_list);
+	/* ufile is required when some objects are released */
+	ucontext->ufile = file;
+	uverbs_initialize_ucontext(ucontext);
+
 	rcu_read_lock();
 	ucontext->tgid = get_task_pid(current->group_leader, PIDTYPE_PID);
 	rcu_read_unlock();
@@ -382,7 +134,7 @@ ssize_t ib_uverbs_get_context(struct ib_uverbs_file *file,
 		goto err_free;
 	resp.async_fd = ret;
 
-	filp = ib_uverbs_alloc_event_file(file, ib_dev, 1);
+	filp = ib_uverbs_alloc_async_event_file(file, ib_dev);
 	if (IS_ERR(filp)) {
 		ret = PTR_ERR(filp);
 		goto err_fd;
@@ -565,19 +317,9 @@ ssize_t ib_uverbs_alloc_pd(struct ib_uverbs_file *file,
 		   (unsigned long) cmd.response + sizeof resp,
 		   in_len - sizeof cmd, out_len - sizeof resp);
 
-	uobj = kmalloc(sizeof *uobj, GFP_KERNEL);
-	if (!uobj)
-		return -ENOMEM;
-
-	init_uobj(uobj, 0, file->ucontext, &pd_lock_class);
-	ret = ib_rdmacg_try_charge(&uobj->cg_obj, ib_dev,
-				   RDMACG_RESOURCE_HCA_OBJECT);
-	if (ret) {
-		kfree(uobj);
-		return ret;
-	}
-
-	down_write(&uobj->mutex);
+	uobj  = uobj_alloc(uobj_get_type(pd), file->ucontext);
+	if (IS_ERR(uobj))
+		return PTR_ERR(uobj);
 
 	pd = ib_dev->alloc_pd(ib_dev, file->ucontext, &udata);
 	if (IS_ERR(pd)) {
@@ -591,10 +333,6 @@ ssize_t ib_uverbs_alloc_pd(struct ib_uverbs_file *file,
 	atomic_set(&pd->usecnt, 0);
 
 	uobj->object = pd;
-	ret = idr_add_uobj(&ib_uverbs_pd_idr, uobj);
-	if (ret)
-		goto err_idr;
-
 	memset(&resp, 0, sizeof resp);
 	resp.pd_handle = uobj->id;
 
@@ -604,25 +342,15 @@ ssize_t ib_uverbs_alloc_pd(struct ib_uverbs_file *file,
 		goto err_copy;
 	}
 
-	mutex_lock(&file->mutex);
-	list_add_tail(&uobj->list, &file->ucontext->pd_list);
-	mutex_unlock(&file->mutex);
-
-	uobj->live = 1;
-
-	up_write(&uobj->mutex);
+	uobj_alloc_commit(uobj);
 
 	return in_len;
 
 err_copy:
-	idr_remove_uobj(&ib_uverbs_pd_idr, uobj);
-
-err_idr:
 	ib_dealloc_pd(pd);
 
 err:
-	ib_rdmacg_uncharge(&uobj->cg_obj, ib_dev, RDMACG_RESOURCE_HCA_OBJECT);
-	put_uobj_write(uobj);
+	uobj_alloc_abort(uobj);
 	return ret;
 }
 
@@ -633,45 +361,19 @@ ssize_t ib_uverbs_dealloc_pd(struct ib_uverbs_file *file,
 {
 	struct ib_uverbs_dealloc_pd cmd;
 	struct ib_uobject          *uobj;
-	struct ib_pd		   *pd;
 	int                         ret;
 
 	if (copy_from_user(&cmd, buf, sizeof cmd))
 		return -EFAULT;
 
-	uobj = idr_write_uobj(&ib_uverbs_pd_idr, cmd.pd_handle, file->ucontext);
-	if (!uobj)
-		return -EINVAL;
-	pd = uobj->object;
-
-	if (atomic_read(&pd->usecnt)) {
-		ret = -EBUSY;
-		goto err_put;
-	}
-
-	ret = pd->device->dealloc_pd(uobj->object);
-	WARN_ONCE(ret, "Infiniband HW driver failed dealloc_pd");
-	if (ret)
-		goto err_put;
-
-	ib_rdmacg_uncharge(&uobj->cg_obj, ib_dev, RDMACG_RESOURCE_HCA_OBJECT);
-
-	uobj->live = 0;
-	put_uobj_write(uobj);
-
-	idr_remove_uobj(&ib_uverbs_pd_idr, uobj);
-
-	mutex_lock(&file->mutex);
-	list_del(&uobj->list);
-	mutex_unlock(&file->mutex);
+	uobj  = uobj_get_write(uobj_get_type(pd), cmd.pd_handle,
+			       file->ucontext);
+	if (IS_ERR(uobj))
+		return PTR_ERR(uobj);
 
-	put_uobj(uobj);
-
-	return in_len;
+	ret = uobj_remove_commit(uobj);
 
-err_put:
-	put_uobj_write(uobj);
-	return ret;
+	return ret ?: in_len;
 }
 
 struct xrcd_table_entry {
@@ -808,16 +510,13 @@ ssize_t ib_uverbs_open_xrcd(struct ib_uverbs_file *file,
 		}
 	}
 
-	obj = kmalloc(sizeof *obj, GFP_KERNEL);
-	if (!obj) {
-		ret = -ENOMEM;
+	obj  = (struct ib_uxrcd_object *)uobj_alloc(uobj_get_type(xrcd),
+						    file->ucontext);
+	if (IS_ERR(obj)) {
+		ret = PTR_ERR(obj);
 		goto err_tree_mutex_unlock;
 	}
 
-	init_uobj(&obj->uobject, 0, file->ucontext, &xrcd_lock_class);
-
-	down_write(&obj->uobject.mutex);
-
 	if (!xrcd) {
 		xrcd = ib_dev->alloc_xrcd(ib_dev, file->ucontext, &udata);
 		if (IS_ERR(xrcd)) {
@@ -835,10 +534,6 @@ ssize_t ib_uverbs_open_xrcd(struct ib_uverbs_file *file,
 
 	atomic_set(&obj->refcnt, 0);
 	obj->uobject.object = xrcd;
-	ret = idr_add_uobj(&ib_uverbs_xrcd_idr, &obj->uobject);
-	if (ret)
-		goto err_idr;
-
 	memset(&resp, 0, sizeof resp);
 	resp.xrcd_handle = obj->uobject.id;
 
@@ -847,7 +542,7 @@ ssize_t ib_uverbs_open_xrcd(struct ib_uverbs_file *file,
 			/* create new inode/xrcd table entry */
 			ret = xrcd_table_insert(file->device, inode, xrcd);
 			if (ret)
-				goto err_insert_xrcd;
+				goto err_dealloc_xrcd;
 		}
 		atomic_inc(&xrcd->usecnt);
 	}
@@ -861,12 +556,7 @@ ssize_t ib_uverbs_open_xrcd(struct ib_uverbs_file *file,
 	if (f.file)
 		fdput(f);
 
-	mutex_lock(&file->mutex);
-	list_add_tail(&obj->uobject.list, &file->ucontext->xrcd_list);
-	mutex_unlock(&file->mutex);
-
-	obj->uobject.live = 1;
-	up_write(&obj->uobject.mutex);
+	uobj_alloc_commit(&obj->uobject);
 
 	mutex_unlock(&file->device->xrcd_tree_mutex);
 	return in_len;
@@ -878,14 +568,11 @@ ssize_t ib_uverbs_open_xrcd(struct ib_uverbs_file *file,
 		atomic_dec(&xrcd->usecnt);
 	}
 
-err_insert_xrcd:
-	idr_remove_uobj(&ib_uverbs_xrcd_idr, &obj->uobject);
-
-err_idr:
+err_dealloc_xrcd:
 	ib_dealloc_xrcd(xrcd);
 
 err:
-	put_uobj_write(&obj->uobject);
+	uobj_alloc_abort(&obj->uobject);
 
 err_tree_mutex_unlock:
 	if (f.file)
@@ -903,75 +590,41 @@ ssize_t ib_uverbs_close_xrcd(struct ib_uverbs_file *file,
 {
 	struct ib_uverbs_close_xrcd cmd;
 	struct ib_uobject           *uobj;
-	struct ib_xrcd              *xrcd = NULL;
-	struct inode                *inode = NULL;
-	struct ib_uxrcd_object      *obj;
-	int                         live;
 	int                         ret = 0;
 
 	if (copy_from_user(&cmd, buf, sizeof cmd))
 		return -EFAULT;
 
-	mutex_lock(&file->device->xrcd_tree_mutex);
-	uobj = idr_write_uobj(&ib_uverbs_xrcd_idr, cmd.xrcd_handle, file->ucontext);
-	if (!uobj) {
-		ret = -EINVAL;
-		goto out;
-	}
-
-	xrcd  = uobj->object;
-	inode = xrcd->inode;
-	obj   = container_of(uobj, struct ib_uxrcd_object, uobject);
-	if (atomic_read(&obj->refcnt)) {
-		put_uobj_write(uobj);
-		ret = -EBUSY;
-		goto out;
-	}
-
-	if (!inode || atomic_dec_and_test(&xrcd->usecnt)) {
-		ret = ib_dealloc_xrcd(uobj->object);
-		if (!ret)
-			uobj->live = 0;
+	uobj  = uobj_get_write(uobj_get_type(xrcd), cmd.xrcd_handle,
+			       file->ucontext);
+	if (IS_ERR(uobj)) {
+		mutex_unlock(&file->device->xrcd_tree_mutex);
+		return PTR_ERR(uobj);
 	}
 
-	live = uobj->live;
-	if (inode && ret)
-		atomic_inc(&xrcd->usecnt);
-
-	put_uobj_write(uobj);
-
-	if (ret)
-		goto out;
-
-	if (inode && !live)
-		xrcd_table_delete(file->device, inode);
-
-	idr_remove_uobj(&ib_uverbs_xrcd_idr, uobj);
-	mutex_lock(&file->mutex);
-	list_del(&uobj->list);
-	mutex_unlock(&file->mutex);
-
-	put_uobj(uobj);
-	ret = in_len;
-
-out:
-	mutex_unlock(&file->device->xrcd_tree_mutex);
-	return ret;
+	ret = uobj_remove_commit(uobj);
+	return ret ?: in_len;
 }
 
-void ib_uverbs_dealloc_xrcd(struct ib_uverbs_device *dev,
-			    struct ib_xrcd *xrcd)
+int ib_uverbs_dealloc_xrcd(struct ib_uverbs_device *dev,
+			   struct ib_xrcd *xrcd,
+			   enum rdma_remove_reason why)
 {
 	struct inode *inode;
+	int ret;
 
 	inode = xrcd->inode;
 	if (inode && !atomic_dec_and_test(&xrcd->usecnt))
-		return;
+		return 0;
 
-	ib_dealloc_xrcd(xrcd);
+	ret = ib_dealloc_xrcd(xrcd);
 
-	if (inode)
+	if (why == RDMA_REMOVE_DESTROY && ret)
+		atomic_inc(&xrcd->usecnt);
+	else if (inode)
 		xrcd_table_delete(dev, inode);
+
+	return ret;
 }
 
 ssize_t ib_uverbs_reg_mr(struct ib_uverbs_file *file,
@@ -1004,14 +657,11 @@ ssize_t ib_uverbs_reg_mr(struct ib_uverbs_file *file,
 	if (ret)
 		return ret;
 
-	uobj = kmalloc(sizeof *uobj, GFP_KERNEL);
-	if (!uobj)
-		return -ENOMEM;
-
-	init_uobj(uobj, 0, file->ucontext, &mr_lock_class);
-	down_write(&uobj->mutex);
+	uobj  = uobj_alloc(uobj_get_type(mr), file->ucontext);
+	if (IS_ERR(uobj))
+		return PTR_ERR(uobj);
 
-	pd = idr_read_pd(cmd.pd_handle, file->ucontext);
+	pd = uobj_get_obj_read(pd, cmd.pd_handle, file->ucontext);
 	if (!pd) {
 		ret = -EINVAL;
 		goto err_free;
@@ -1025,10 +675,6 @@ ssize_t ib_uverbs_reg_mr(struct ib_uverbs_file *file,
 			goto err_put;
 		}
 	}
-	ret = ib_rdmacg_try_charge(&uobj->cg_obj, ib_dev,
-				   RDMACG_RESOURCE_HCA_OBJECT);
-	if (ret)
-		goto err_charge;
 
 	mr = pd->device->reg_user_mr(pd, cmd.start, cmd.length, cmd.hca_va,
 				     cmd.access_flags, &udata);
@@ -1043,9 +689,6 @@ ssize_t ib_uverbs_reg_mr(struct ib_uverbs_file *file,
 	atomic_inc(&pd->usecnt);
 
 	uobj->object = mr;
-	ret = idr_add_uobj(&ib_uverbs_mr_idr, uobj);
-	if (ret)
-		goto err_unreg;
 
 	memset(&resp, 0, sizeof resp);
 	resp.lkey      = mr->lkey;
@@ -1058,32 +701,20 @@ ssize_t ib_uverbs_reg_mr(struct ib_uverbs_file *file,
 		goto err_copy;
 	}
 
-	put_pd_read(pd);
-
-	mutex_lock(&file->mutex);
-	list_add_tail(&uobj->list, &file->ucontext->mr_list);
-	mutex_unlock(&file->mutex);
+	uobj_put_obj_read(pd);
 
-	uobj->live = 1;
-
-	up_write(&uobj->mutex);
+	uobj_alloc_commit(uobj);
 
 	return in_len;
 
 err_copy:
-	idr_remove_uobj(&ib_uverbs_mr_idr, uobj);
-
-err_unreg:
 	ib_dereg_mr(mr);
 
 err_put:
-	ib_rdmacg_uncharge(&uobj->cg_obj, ib_dev, RDMACG_RESOURCE_HCA_OBJECT);
-
-err_charge:
-	put_pd_read(pd);
+	uobj_put_obj_read(pd);
 
 err_free:
-	put_uobj_write(uobj);
+	uobj_alloc_abort(uobj);
 	return ret;
 }
 
@@ -1119,11 +750,10 @@ ssize_t ib_uverbs_rereg_mr(struct ib_uverbs_file *file,
 	     (cmd.start & ~PAGE_MASK) != (cmd.hca_va & ~PAGE_MASK)))
 			return -EINVAL;
 
-	uobj = idr_write_uobj(&ib_uverbs_mr_idr, cmd.mr_handle,
-			      file->ucontext);
-
-	if (!uobj)
-		return -EINVAL;
+	uobj  = uobj_get_write(uobj_get_type(mr), cmd.mr_handle,
+			       file->ucontext);
+	if (IS_ERR(uobj))
+		return PTR_ERR(uobj);
 
 	mr = uobj->object;
 
@@ -1134,7 +764,7 @@ ssize_t ib_uverbs_rereg_mr(struct ib_uverbs_file *file,
 	}
 
 	if (cmd.flags & IB_MR_REREG_PD) {
-		pd = idr_read_pd(cmd.pd_handle, file->ucontext);
+		pd = uobj_get_obj_read(pd, cmd.pd_handle, file->ucontext);
 		if (!pd) {
 			ret = -EINVAL;
 			goto put_uobjs;
@@ -1167,11 +797,10 @@ ssize_t ib_uverbs_rereg_mr(struct ib_uverbs_file *file,
 
 put_uobj_pd:
 	if (cmd.flags & IB_MR_REREG_PD)
-		put_pd_read(pd);
+		uobj_put_obj_read(pd);
 
 put_uobjs:
-
-	put_uobj_write(mr->uobject);
+	uobj_put_write(uobj);
 
 	return ret;
 }
@@ -1182,39 +811,20 @@ ssize_t ib_uverbs_dereg_mr(struct ib_uverbs_file *file,
 			   int out_len)
 {
 	struct ib_uverbs_dereg_mr cmd;
-	struct ib_mr             *mr;
 	struct ib_uobject	 *uobj;
 	int                       ret = -EINVAL;
 
 	if (copy_from_user(&cmd, buf, sizeof cmd))
 		return -EFAULT;
 
-	uobj = idr_write_uobj(&ib_uverbs_mr_idr, cmd.mr_handle, file->ucontext);
-	if (!uobj)
-		return -EINVAL;
-
-	mr = uobj->object;
-
-	ret = ib_dereg_mr(mr);
-	if (!ret)
-		uobj->live = 0;
-
-	put_uobj_write(uobj);
-
-	if (ret)
-		return ret;
-
-	ib_rdmacg_uncharge(&uobj->cg_obj, ib_dev, RDMACG_RESOURCE_HCA_OBJECT);
-
-	idr_remove_uobj(&ib_uverbs_mr_idr, uobj);
-
-	mutex_lock(&file->mutex);
-	list_del(&uobj->list);
-	mutex_unlock(&file->mutex);
+	uobj  = uobj_get_write(uobj_get_type(mr), cmd.mr_handle,
+			       file->ucontext);
+	if (IS_ERR(uobj))
+		return PTR_ERR(uobj);
 
-	put_uobj(uobj);
+	ret = uobj_remove_commit(uobj);
 
-	return in_len;
+	return ret ?: in_len;
 }
 
 ssize_t ib_uverbs_alloc_mw(struct ib_uverbs_file *file,
@@ -1236,14 +846,11 @@ ssize_t ib_uverbs_alloc_mw(struct ib_uverbs_file *file,
 	if (copy_from_user(&cmd, buf, sizeof(cmd)))
 		return -EFAULT;
 
-	uobj = kmalloc(sizeof(*uobj), GFP_KERNEL);
-	if (!uobj)
-		return -ENOMEM;
-
-	init_uobj(uobj, 0, file->ucontext, &mw_lock_class);
-	down_write(&uobj->mutex);
+	uobj  = uobj_alloc(uobj_get_type(mw), file->ucontext);
+	if (IS_ERR(uobj))
+		return PTR_ERR(uobj);
 
-	pd = idr_read_pd(cmd.pd_handle, file->ucontext);
+	pd = uobj_get_obj_read(pd, cmd.pd_handle, file->ucontext);
 	if (!pd) {
 		ret = -EINVAL;
 		goto err_free;
@@ -1254,11 +861,6 @@ ssize_t ib_uverbs_alloc_mw(struct ib_uverbs_file *file,
 		   in_len - sizeof(cmd) - sizeof(struct ib_uverbs_cmd_hdr),
 		   out_len - sizeof(resp));
 
-	ret = ib_rdmacg_try_charge(&uobj->cg_obj, ib_dev,
-				   RDMACG_RESOURCE_HCA_OBJECT);
-	if (ret)
-		goto err_charge;
-
 	mw = pd->device->alloc_mw(pd, cmd.mw_type, &udata);
 	if (IS_ERR(mw)) {
 		ret = PTR_ERR(mw);
@@ -1271,9 +873,6 @@ ssize_t ib_uverbs_alloc_mw(struct ib_uverbs_file *file,
 	atomic_inc(&pd->usecnt);
 
 	uobj->object = mw;
-	ret = idr_add_uobj(&ib_uverbs_mw_idr, uobj);
-	if (ret)
-		goto err_unalloc;
 
 	memset(&resp, 0, sizeof(resp));
 	resp.rkey      = mw->rkey;
@@ -1285,32 +884,17 @@ ssize_t ib_uverbs_alloc_mw(struct ib_uverbs_file *file,
 		goto err_copy;
 	}
 
-	put_pd_read(pd);
-
-	mutex_lock(&file->mutex);
-	list_add_tail(&uobj->list, &file->ucontext->mw_list);
-	mutex_unlock(&file->mutex);
-
-	uobj->live = 1;
-
-	up_write(&uobj->mutex);
+	uobj_put_obj_read(pd);
+	uobj_alloc_commit(uobj);
 
 	return in_len;
 
 err_copy:
-	idr_remove_uobj(&ib_uverbs_mw_idr, uobj);
-
-err_unalloc:
 	uverbs_dealloc_mw(mw);
-
 err_put:
-	ib_rdmacg_uncharge(&uobj->cg_obj, ib_dev, RDMACG_RESOURCE_HCA_OBJECT);
-
-err_charge:
-	put_pd_read(pd);
-
+	uobj_put_obj_read(pd);
 err_free:
-	put_uobj_write(uobj);
+	uobj_alloc_abort(uobj);
 	return ret;
 }
 
@@ -1320,39 +904,19 @@ ssize_t ib_uverbs_dealloc_mw(struct ib_uverbs_file *file,
 			     int out_len)
 {
 	struct ib_uverbs_dealloc_mw cmd;
-	struct ib_mw               *mw;
 	struct ib_uobject	   *uobj;
 	int                         ret = -EINVAL;
 
 	if (copy_from_user(&cmd, buf, sizeof(cmd)))
 		return -EFAULT;
 
-	uobj = idr_write_uobj(&ib_uverbs_mw_idr, cmd.mw_handle, file->ucontext);
-	if (!uobj)
-		return -EINVAL;
-
-	mw = uobj->object;
-
-	ret = uverbs_dealloc_mw(mw);
-	if (!ret)
-		uobj->live = 0;
-
-	put_uobj_write(uobj);
+	uobj  = uobj_get_write(uobj_get_type(mw), cmd.mw_handle,
+			       file->ucontext);
+	if (IS_ERR(uobj))
+		return PTR_ERR(uobj);
 
-	if (ret)
-		return ret;
-
-	ib_rdmacg_uncharge(&uobj->cg_obj, ib_dev, RDMACG_RESOURCE_HCA_OBJECT);
-
-	idr_remove_uobj(&ib_uverbs_mw_idr, uobj);
-
-	mutex_lock(&file->mutex);
-	list_del(&uobj->list);
-	mutex_unlock(&file->mutex);
-
-	put_uobj(uobj);
-
-	return in_len;
+	ret = uobj_remove_commit(uobj);
+	return ret ?: in_len;
 }
 
 ssize_t ib_uverbs_create_comp_channel(struct ib_uverbs_file *file,
@@ -1362,8 +926,8 @@ ssize_t ib_uverbs_create_comp_channel(struct ib_uverbs_file *file,
 {
 	struct ib_uverbs_create_comp_channel	   cmd;
 	struct ib_uverbs_create_comp_channel_resp  resp;
-	struct file				  *filp;
-	int ret;
+	struct ib_uobject			  *uobj;
+	struct ib_uverbs_completion_event_file	  *ev_file;
 
 	if (out_len < sizeof resp)
 		return -ENOSPC;
@@ -1371,25 +935,23 @@ ssize_t ib_uverbs_create_comp_channel(struct ib_uverbs_file *file,
 	if (copy_from_user(&cmd, buf, sizeof cmd))
 		return -EFAULT;
 
-	ret = get_unused_fd_flags(O_CLOEXEC);
-	if (ret < 0)
-		return ret;
-	resp.fd = ret;
+	uobj = uobj_alloc(uobj_get_type(comp_channel), file->ucontext);
+	if (IS_ERR(uobj))
+		return PTR_ERR(uobj);
 
-	filp = ib_uverbs_alloc_event_file(file, ib_dev, 0);
-	if (IS_ERR(filp)) {
-		put_unused_fd(resp.fd);
-		return PTR_ERR(filp);
-	}
+	resp.fd = uobj->id;
+
+	ev_file = container_of(uobj, struct ib_uverbs_completion_event_file,
+			       uobj_file.uobj);
+	ib_uverbs_init_event_queue(&ev_file->ev_queue);
 
 	if (copy_to_user((void __user *) (unsigned long) cmd.response,
 			 &resp, sizeof resp)) {
-		put_unused_fd(resp.fd);
-		fput(filp);
+		uobj_alloc_abort(uobj);
 		return -EFAULT;
 	}
 
-	fd_install(resp.fd, filp);
+	uobj_alloc_commit(uobj);
 	return in_len;
 }
 
@@ -1407,7 +969,7 @@ static struct ib_ucq_object *create_cq(struct ib_uverbs_file *file,
 				       void *context)
 {
 	struct ib_ucq_object           *obj;
-	struct ib_uverbs_event_file    *ev_file = NULL;
+	struct ib_uverbs_completion_event_file    *ev_file = NULL;
 	struct ib_cq                   *cq;
 	int                             ret;
 	struct ib_uverbs_ex_create_cq_resp resp;
@@ -1416,21 +978,21 @@ static struct ib_ucq_object *create_cq(struct ib_uverbs_file *file,
 	if (cmd->comp_vector >= file->device->num_comp_vectors)
 		return ERR_PTR(-EINVAL);
 
-	obj = kmalloc(sizeof *obj, GFP_KERNEL);
-	if (!obj)
-		return ERR_PTR(-ENOMEM);
-
-	init_uobj(&obj->uobject, cmd->user_handle, file->ucontext, &cq_lock_class);
-	down_write(&obj->uobject.mutex);
+	obj  = (struct ib_ucq_object *)uobj_alloc(uobj_get_type(cq),
+						  file->ucontext);
+	if (IS_ERR(obj))
+		return obj;
 
 	if (cmd->comp_channel >= 0) {
-		ev_file = ib_uverbs_lookup_comp_file(cmd->comp_channel);
-		if (!ev_file) {
-			ret = -EINVAL;
+		ev_file = ib_uverbs_lookup_comp_file(cmd->comp_channel,
+						     file->ucontext);
+		if (IS_ERR(ev_file)) {
+			ret = PTR_ERR(ev_file);
 			goto err;
 		}
 	}
 
+	obj->uobject.user_handle = cmd->user_handle;
 	obj->uverbs_file	   = file;
 	obj->comp_events_reported  = 0;
 	obj->async_events_reported = 0;
@@ -1443,13 +1005,7 @@ static struct ib_ucq_object *create_cq(struct ib_uverbs_file *file,
 	if (cmd_sz > offsetof(typeof(*cmd), flags) + sizeof(cmd->flags))
 		attr.flags = cmd->flags;
 
-	ret = ib_rdmacg_try_charge(&obj->uobject.cg_obj, ib_dev,
-				   RDMACG_RESOURCE_HCA_OBJECT);
-	if (ret)
-		goto err_charge;
-
-	cq = ib_dev->create_cq(ib_dev, &attr,
-					     file->ucontext, uhw);
+	cq = ib_dev->create_cq(ib_dev, &attr, file->ucontext, uhw);
 	if (IS_ERR(cq)) {
 		ret = PTR_ERR(cq);
 		goto err_file;
@@ -1459,14 +1015,10 @@ static struct ib_ucq_object *create_cq(struct ib_uverbs_file *file,
 	cq->uobject       = &obj->uobject;
 	cq->comp_handler  = ib_uverbs_comp_handler;
 	cq->event_handler = ib_uverbs_cq_event_handler;
-	cq->cq_context    = ev_file;
+	cq->cq_context    = &ev_file->ev_queue;
 	atomic_set(&cq->usecnt, 0);
 
 	obj->uobject.object = cq;
-	ret = idr_add_uobj(&ib_uverbs_cq_idr, &obj->uobject);
-	if (ret)
-		goto err_free;
-
 	memset(&resp, 0, sizeof resp);
 	resp.base.cq_handle = obj->uobject.id;
 	resp.base.cqe       = cq->cqe;
@@ -1478,32 +1030,19 @@ static struct ib_ucq_object *create_cq(struct ib_uverbs_file *file,
 	if (ret)
 		goto err_cb;
 
-	mutex_lock(&file->mutex);
-	list_add_tail(&obj->uobject.list, &file->ucontext->cq_list);
-	mutex_unlock(&file->mutex);
-
-	obj->uobject.live = 1;
-
-	up_write(&obj->uobject.mutex);
+	uobj_alloc_commit(&obj->uobject);
 
 	return obj;
 
 err_cb:
-	idr_remove_uobj(&ib_uverbs_cq_idr, &obj->uobject);
-
-err_free:
 	ib_destroy_cq(cq);
 
 err_file:
-	ib_rdmacg_uncharge(&obj->uobject.cg_obj, ib_dev,
-			   RDMACG_RESOURCE_HCA_OBJECT);
-
-err_charge:
 	if (ev_file)
 		ib_uverbs_release_ucq(file, ev_file, obj);
 
 err:
-	put_uobj_write(&obj->uobject);
+	uobj_alloc_abort(&obj->uobject);
 
 	return ERR_PTR(ret);
 }
@@ -1626,7 +1165,7 @@ ssize_t ib_uverbs_resize_cq(struct ib_uverbs_file *file,
 		   (unsigned long) cmd.response + sizeof resp,
 		   in_len - sizeof cmd, out_len - sizeof resp);
 
-	cq = idr_read_cq(cmd.cq_handle, file->ucontext, 0);
+	cq = uobj_get_obj_read(cq, cmd.cq_handle, file->ucontext);
 	if (!cq)
 		return -EINVAL;
 
@@ -1641,7 +1180,7 @@ ssize_t ib_uverbs_resize_cq(struct ib_uverbs_file *file,
 		ret = -EFAULT;
 
 out:
-	put_cq_read(cq);
+	uobj_put_obj_read(cq);
 
 	return ret ? ret : in_len;
 }
@@ -1688,7 +1227,7 @@ ssize_t ib_uverbs_poll_cq(struct ib_uverbs_file *file,
 	if (copy_from_user(&cmd, buf, sizeof cmd))
 		return -EFAULT;
 
-	cq = idr_read_cq(cmd.cq_handle, file->ucontext, 0);
+	cq = uobj_get_obj_read(cq, cmd.cq_handle, file->ucontext);
 	if (!cq)
 		return -EINVAL;
 
@@ -1720,7 +1259,7 @@ ssize_t ib_uverbs_poll_cq(struct ib_uverbs_file *file,
 	ret = in_len;
 
 out_put:
-	put_cq_read(cq);
+	uobj_put_obj_read(cq);
 	return ret;
 }
 
@@ -1735,14 +1274,14 @@ ssize_t ib_uverbs_req_notify_cq(struct ib_uverbs_file *file,
 	if (copy_from_user(&cmd, buf, sizeof cmd))
 		return -EFAULT;
 
-	cq = idr_read_cq(cmd.cq_handle, file->ucontext, 0);
+	cq = uobj_get_obj_read(cq, cmd.cq_handle, file->ucontext);
 	if (!cq)
 		return -EINVAL;
 
 	ib_req_notify_cq(cq, cmd.solicited_only ?
 			 IB_CQ_SOLICITED : IB_CQ_NEXT_COMP);
 
-	put_cq_read(cq);
+	uobj_put_obj_read(cq);
 
 	return in_len;
 }
@@ -1757,44 +1296,38 @@ ssize_t ib_uverbs_destroy_cq(struct ib_uverbs_file *file,
 	struct ib_uobject		*uobj;
 	struct ib_cq               	*cq;
 	struct ib_ucq_object        	*obj;
-	struct ib_uverbs_event_file	*ev_file;
+	struct ib_uverbs_event_queue	*ev_queue;
 	int                        	 ret = -EINVAL;
 
 	if (copy_from_user(&cmd, buf, sizeof cmd))
 		return -EFAULT;
 
-	uobj = idr_write_uobj(&ib_uverbs_cq_idr, cmd.cq_handle, file->ucontext);
-	if (!uobj)
-		return -EINVAL;
+	uobj  = uobj_get_write(uobj_get_type(cq), cmd.cq_handle,
+			       file->ucontext);
+	if (IS_ERR(uobj))
+		return PTR_ERR(uobj);
+
+	/*
+	 * Make sure we don't free the memory in remove_commit as we still
+	 * needs the uobject memory to create the response.
+	 */
+	uverbs_uobject_get(uobj);
 	cq      = uobj->object;
-	ev_file = cq->cq_context;
+	ev_queue = cq->cq_context;
 	obj     = container_of(cq->uobject, struct ib_ucq_object, uobject);
 
-	ret = ib_destroy_cq(cq);
-	if (!ret)
-		uobj->live = 0;
-
-	put_uobj_write(uobj);
+	memset(&resp, 0, sizeof(resp));
 
-	if (ret)
+	ret = uobj_remove_commit(uobj);
+	if (ret) {
+		uverbs_uobject_put(uobj);
 		return ret;
+	}
 
-	ib_rdmacg_uncharge(&uobj->cg_obj, ib_dev, RDMACG_RESOURCE_HCA_OBJECT);
-
-	idr_remove_uobj(&ib_uverbs_cq_idr, uobj);
-
-	mutex_lock(&file->mutex);
-	list_del(&uobj->list);
-	mutex_unlock(&file->mutex);
-
-	ib_uverbs_release_ucq(file, ev_file, obj);
-
-	memset(&resp, 0, sizeof resp);
 	resp.comp_events_reported  = obj->comp_events_reported;
 	resp.async_events_reported = obj->async_events_reported;
 
-	put_uobj(uobj);
-
+	uverbs_uobject_put(uobj);
 	if (copy_to_user((void __user *) (unsigned long) cmd.response,
 			 &resp, sizeof resp))
 		return -EFAULT;
@@ -1816,7 +1349,7 @@ static int create_qp(struct ib_uverbs_file *file,
 	struct ib_device		*device;
 	struct ib_pd			*pd = NULL;
 	struct ib_xrcd			*xrcd = NULL;
-	struct ib_uobject		*uninitialized_var(xrcd_uobj);
+	struct ib_uobject		*xrcd_uobj = ERR_PTR(-ENOENT);
 	struct ib_cq			*scq = NULL, *rcq = NULL;
 	struct ib_srq			*srq = NULL;
 	struct ib_qp			*qp;
@@ -1830,18 +1363,20 @@ static int create_qp(struct ib_uverbs_file *file,
 	if (cmd->qp_type == IB_QPT_RAW_PACKET && !capable(CAP_NET_RAW))
 		return -EPERM;
 
-	obj = kzalloc(sizeof *obj, GFP_KERNEL);
-	if (!obj)
-		return -ENOMEM;
+	obj  = (struct ib_uqp_object *)uobj_alloc(uobj_get_type(qp),
+						  file->ucontext);
+	if (IS_ERR(obj))
+		return PTR_ERR(obj);
+	obj->uxrcd = NULL;
+	obj->uevent.uobject.user_handle = cmd->user_handle;
+	mutex_init(&obj->mcast_lock);
 
-	init_uobj(&obj->uevent.uobject, cmd->user_handle, file->ucontext,
-		  &qp_lock_class);
-	down_write(&obj->uevent.uobject.mutex);
 	if (cmd_sz >= offsetof(typeof(*cmd), rwq_ind_tbl_handle) +
 		      sizeof(cmd->rwq_ind_tbl_handle) &&
 		      (cmd->comp_mask & IB_UVERBS_CREATE_QP_MASK_IND_TABLE)) {
-		ind_tbl = idr_read_rwq_indirection_table(cmd->rwq_ind_tbl_handle,
-							 file->ucontext);
+		ind_tbl = uobj_get_obj_read(rwq_ind_table,
+					    cmd->rwq_ind_tbl_handle,
+					    file->ucontext);
 		if (!ind_tbl) {
 			ret = -EINVAL;
 			goto err_put;
@@ -1865,8 +1400,15 @@ static int create_qp(struct ib_uverbs_file *file,
 		has_sq = false;
 
 	if (cmd->qp_type == IB_QPT_XRC_TGT) {
-		xrcd = idr_read_xrcd(cmd->pd_handle, file->ucontext,
-				     &xrcd_uobj);
+		xrcd_uobj = uobj_get_read(uobj_get_type(xrcd), cmd->pd_handle,
+					  file->ucontext);
+
+		if (IS_ERR(xrcd_uobj)) {
+			ret = -EINVAL;
+			goto err_put;
+		}
+
+		xrcd = (struct ib_xrcd *)xrcd_uobj->object;
 		if (!xrcd) {
 			ret = -EINVAL;
 			goto err_put;
@@ -1878,8 +1420,8 @@ static int create_qp(struct ib_uverbs_file *file,
 			cmd->max_recv_sge = 0;
 		} else {
 			if (cmd->is_srq) {
-				srq = idr_read_srq(cmd->srq_handle,
-						   file->ucontext);
+				srq = uobj_get_obj_read(srq, cmd->srq_handle,
+							file->ucontext);
 				if (!srq || srq->srq_type != IB_SRQT_BASIC) {
 					ret = -EINVAL;
 					goto err_put;
@@ -1888,8 +1430,8 @@ static int create_qp(struct ib_uverbs_file *file,
 
 			if (!ind_tbl) {
 				if (cmd->recv_cq_handle != cmd->send_cq_handle) {
-					rcq = idr_read_cq(cmd->recv_cq_handle,
-							  file->ucontext, 0);
+					rcq = uobj_get_obj_read(cq, cmd->recv_cq_handle,
+								file->ucontext);
 					if (!rcq) {
 						ret = -EINVAL;
 						goto err_put;
@@ -1899,10 +1441,11 @@ static int create_qp(struct ib_uverbs_file *file,
 		}
 
 		if (has_sq)
-			scq = idr_read_cq(cmd->send_cq_handle, file->ucontext, !!rcq);
+			scq = uobj_get_obj_read(cq, cmd->send_cq_handle,
+						file->ucontext);
 		if (!ind_tbl)
 			rcq = rcq ?: scq;
-		pd  = idr_read_pd(cmd->pd_handle, file->ucontext);
+		pd  = uobj_get_obj_read(pd, cmd->pd_handle, file->ucontext);
 		if (!pd || (!scq && has_sq)) {
 			ret = -EINVAL;
 			goto err_put;
@@ -1954,11 +1497,6 @@ static int create_qp(struct ib_uverbs_file *file,
 			goto err_put;
 		}
 
-	ret = ib_rdmacg_try_charge(&obj->uevent.uobject.cg_obj, device,
-				   RDMACG_RESOURCE_HCA_OBJECT);
-	if (ret)
-		goto err_put;
-
 	if (cmd->qp_type == IB_QPT_XRC_TGT)
 		qp = ib_create_qp(pd, &attr);
 	else
@@ -1966,7 +1504,7 @@ static int create_qp(struct ib_uverbs_file *file,
 
 	if (IS_ERR(qp)) {
 		ret = PTR_ERR(qp);
-		goto err_create;
+		goto err_put;
 	}
 
 	if (cmd->qp_type != IB_QPT_XRC_TGT) {
@@ -1994,9 +1532,6 @@ static int create_qp(struct ib_uverbs_file *file,
 	qp->uobject = &obj->uevent.uobject;
 
 	obj->uevent.uobject.object = qp;
-	ret = idr_add_uobj(&ib_uverbs_qp_idr, &obj->uevent.uobject);
-	if (ret)
-		goto err_destroy;
 
 	memset(&resp, 0, sizeof resp);
 	resp.base.qpn             = qp->qp_num;
@@ -2018,54 +1553,41 @@ static int create_qp(struct ib_uverbs_file *file,
 		obj->uxrcd = container_of(xrcd_uobj, struct ib_uxrcd_object,
 					  uobject);
 		atomic_inc(&obj->uxrcd->refcnt);
-		put_xrcd_read(xrcd_uobj);
+		uobj_put_read(xrcd_uobj);
 	}
 
 	if (pd)
-		put_pd_read(pd);
+		uobj_put_obj_read(pd);
 	if (scq)
-		put_cq_read(scq);
+		uobj_put_obj_read(scq);
 	if (rcq && rcq != scq)
-		put_cq_read(rcq);
+		uobj_put_obj_read(rcq);
 	if (srq)
-		put_srq_read(srq);
+		uobj_put_obj_read(srq);
 	if (ind_tbl)
-		put_rwq_indirection_table_read(ind_tbl);
-
-	mutex_lock(&file->mutex);
-	list_add_tail(&obj->uevent.uobject.list, &file->ucontext->qp_list);
-	mutex_unlock(&file->mutex);
+		uobj_put_obj_read(ind_tbl);
 
-	obj->uevent.uobject.live = 1;
-
-	up_write(&obj->uevent.uobject.mutex);
+	uobj_alloc_commit(&obj->uevent.uobject);
 
 	return 0;
 err_cb:
-	idr_remove_uobj(&ib_uverbs_qp_idr, &obj->uevent.uobject);
-
-err_destroy:
 	ib_destroy_qp(qp);
 
-err_create:
-	ib_rdmacg_uncharge(&obj->uevent.uobject.cg_obj, device,
-			   RDMACG_RESOURCE_HCA_OBJECT);
-
 err_put:
-	if (xrcd)
-		put_xrcd_read(xrcd_uobj);
+	if (!IS_ERR(xrcd_uobj))
+		uobj_put_read(xrcd_uobj);
 	if (pd)
-		put_pd_read(pd);
+		uobj_put_obj_read(pd);
 	if (scq)
-		put_cq_read(scq);
+		uobj_put_obj_read(scq);
 	if (rcq && rcq != scq)
-		put_cq_read(rcq);
+		uobj_put_obj_read(rcq);
 	if (srq)
-		put_srq_read(srq);
+		uobj_put_obj_read(srq);
 	if (ind_tbl)
-		put_rwq_indirection_table_read(ind_tbl);
+		uobj_put_obj_read(ind_tbl);
 
-	put_uobj_write(&obj->uevent.uobject);
+	uobj_alloc_abort(&obj->uevent.uobject);
 	return ret;
 }
 
@@ -2201,17 +1723,22 @@ ssize_t ib_uverbs_open_qp(struct ib_uverbs_file *file,
 		   (unsigned long) cmd.response + sizeof resp,
 		   in_len - sizeof cmd, out_len - sizeof resp);
 
-	obj = kmalloc(sizeof *obj, GFP_KERNEL);
-	if (!obj)
-		return -ENOMEM;
+	obj  = (struct ib_uqp_object *)uobj_alloc(uobj_get_type(qp),
+						  file->ucontext);
+	if (IS_ERR(obj))
+		return PTR_ERR(obj);
 
-	init_uobj(&obj->uevent.uobject, cmd.user_handle, file->ucontext, &qp_lock_class);
-	down_write(&obj->uevent.uobject.mutex);
+	xrcd_uobj = uobj_get_read(uobj_get_type(xrcd), cmd.pd_handle,
+				  file->ucontext);
+	if (IS_ERR(xrcd_uobj)) {
+		ret = -EINVAL;
+		goto err_put;
+	}
 
-	xrcd = idr_read_xrcd(cmd.pd_handle, file->ucontext, &xrcd_uobj);
+	xrcd = (struct ib_xrcd *)xrcd_uobj->object;
 	if (!xrcd) {
 		ret = -EINVAL;
-		goto err_put;
+		goto err_xrcd;
 	}
 
 	attr.event_handler = ib_uverbs_qp_event_handler;
@@ -2226,15 +1753,11 @@ ssize_t ib_uverbs_open_qp(struct ib_uverbs_file *file,
 	qp = ib_open_qp(xrcd, &attr);
 	if (IS_ERR(qp)) {
 		ret = PTR_ERR(qp);
-		goto err_put;
+		goto err_xrcd;
 	}
 
-	qp->uobject = &obj->uevent.uobject;
-
 	obj->uevent.uobject.object = qp;
-	ret = idr_add_uobj(&ib_uverbs_qp_idr, &obj->uevent.uobject);
-	if (ret)
-		goto err_destroy;
+	obj->uevent.uobject.user_handle = cmd.user_handle;
 
 	memset(&resp, 0, sizeof resp);
 	resp.qpn       = qp->qp_num;
@@ -2243,32 +1766,25 @@ ssize_t ib_uverbs_open_qp(struct ib_uverbs_file *file,
 	if (copy_to_user((void __user *) (unsigned long) cmd.response,
 			 &resp, sizeof resp)) {
 		ret = -EFAULT;
-		goto err_remove;
+		goto err_destroy;
 	}
 
 	obj->uxrcd = container_of(xrcd_uobj, struct ib_uxrcd_object, uobject);
 	atomic_inc(&obj->uxrcd->refcnt);
-	put_xrcd_read(xrcd_uobj);
-
-	mutex_lock(&file->mutex);
-	list_add_tail(&obj->uevent.uobject.list, &file->ucontext->qp_list);
-	mutex_unlock(&file->mutex);
+	qp->uobject = &obj->uevent.uobject;
+	uobj_put_read(xrcd_uobj);
 
-	obj->uevent.uobject.live = 1;
 
-	up_write(&obj->uevent.uobject.mutex);
+	uobj_alloc_commit(&obj->uevent.uobject);
 
 	return in_len;
 
-err_remove:
-	idr_remove_uobj(&ib_uverbs_qp_idr, &obj->uevent.uobject);
-
 err_destroy:
 	ib_destroy_qp(qp);
-
+err_xrcd:
+	uobj_put_read(xrcd_uobj);
 err_put:
-	put_xrcd_read(xrcd_uobj);
-	put_uobj_write(&obj->uevent.uobject);
+	uobj_alloc_abort(&obj->uevent.uobject);
 	return ret;
 }
 
@@ -2294,7 +1810,7 @@ ssize_t ib_uverbs_query_qp(struct ib_uverbs_file *file,
 		goto out;
 	}
 
-	qp = idr_read_qp(cmd.qp_handle, file->ucontext);
+	qp = uobj_get_obj_read(qp, cmd.qp_handle, file->ucontext);
 	if (!qp) {
 		ret = -EINVAL;
 		goto out;
@@ -2302,7 +1818,7 @@ ssize_t ib_uverbs_query_qp(struct ib_uverbs_file *file,
 
 	ret = ib_query_qp(qp, attr, cmd.attr_mask, init_attr);
 
-	put_qp_read(qp);
+	uobj_put_obj_read(qp);
 
 	if (ret)
 		goto out;
@@ -2398,7 +1914,7 @@ static int modify_qp(struct ib_uverbs_file *file,
 	if (!attr)
 		return -ENOMEM;
 
-	qp = idr_read_qp(cmd->base.qp_handle, file->ucontext);
+	qp = uobj_get_obj_read(qp, cmd->base.qp_handle, file->ucontext);
 	if (!qp) {
 		ret = -EINVAL;
 		goto out;
@@ -2470,7 +1986,7 @@ static int modify_qp(struct ib_uverbs_file *file,
 	}
 
 release_qp:
-	put_qp_read(qp);
+	uobj_put_obj_read(qp);
 
 out:
 	kfree(attr);
@@ -2557,42 +2073,27 @@ ssize_t ib_uverbs_destroy_qp(struct ib_uverbs_file *file,
 
 	memset(&resp, 0, sizeof resp);
 
-	uobj = idr_write_uobj(&ib_uverbs_qp_idr, cmd.qp_handle, file->ucontext);
-	if (!uobj)
-		return -EINVAL;
+	uobj  = uobj_get_write(uobj_get_type(qp), cmd.qp_handle,
+			       file->ucontext);
+	if (IS_ERR(uobj))
+		return PTR_ERR(uobj);
+
 	qp  = uobj->object;
 	obj = container_of(uobj, struct ib_uqp_object, uevent.uobject);
+	/*
+	 * Make sure we don't free the memory in remove_commit as we still
+	 * needs the uobject memory to create the response.
+	 */
+	uverbs_uobject_get(uobj);
 
-	if (!list_empty(&obj->mcast_list)) {
-		put_uobj_write(uobj);
-		return -EBUSY;
-	}
-
-	ret = ib_destroy_qp(qp);
-	if (!ret)
-		uobj->live = 0;
-
-	put_uobj_write(uobj);
-
-	if (ret)
+	ret = uobj_remove_commit(uobj);
+	if (ret) {
+		uverbs_uobject_put(uobj);
 		return ret;
-
-	ib_rdmacg_uncharge(&uobj->cg_obj, ib_dev, RDMACG_RESOURCE_HCA_OBJECT);
-
-	if (obj->uxrcd)
-		atomic_dec(&obj->uxrcd->refcnt);
-
-	idr_remove_uobj(&ib_uverbs_qp_idr, uobj);
-
-	mutex_lock(&file->mutex);
-	list_del(&uobj->list);
-	mutex_unlock(&file->mutex);
-
-	ib_uverbs_release_uevent(file, &obj->uevent);
+	}
 
 	resp.events_reported = obj->uevent.events_reported;
-
-	put_uobj(uobj);
+	uverbs_uobject_put(uobj);
 
 	if (copy_to_user((void __user *) (unsigned long) cmd.response,
 			 &resp, sizeof resp))
@@ -2636,7 +2137,7 @@ ssize_t ib_uverbs_post_send(struct ib_uverbs_file *file,
 	if (!user_wr)
 		return -ENOMEM;
 
-	qp = idr_read_qp(cmd.qp_handle, file->ucontext);
+	qp = uobj_get_obj_read(qp, cmd.qp_handle, file->ucontext);
 	if (!qp)
 		goto out;
 
@@ -2672,7 +2173,8 @@ ssize_t ib_uverbs_post_send(struct ib_uverbs_file *file,
 				goto out_put;
 			}
 
-			ud->ah = idr_read_ah(user_wr->wr.ud.ah, file->ucontext);
+			ud->ah = uobj_get_obj_read(ah, user_wr->wr.ud.ah,
+						   file->ucontext);
 			if (!ud->ah) {
 				kfree(ud);
 				ret = -EINVAL;
@@ -2779,11 +2281,11 @@ ssize_t ib_uverbs_post_send(struct ib_uverbs_file *file,
 		ret = -EFAULT;
 
 out_put:
-	put_qp_read(qp);
+	uobj_put_obj_read(qp);
 
 	while (wr) {
 		if (is_ud && ud_wr(wr)->ah)
-			put_ah_read(ud_wr(wr)->ah);
+			uobj_put_obj_read(ud_wr(wr)->ah);
 		next = wr->next;
 		kfree(wr);
 		wr = next;
@@ -2900,21 +2402,21 @@ ssize_t ib_uverbs_post_recv(struct ib_uverbs_file *file,
 	if (IS_ERR(wr))
 		return PTR_ERR(wr);
 
-	qp = idr_read_qp(cmd.qp_handle, file->ucontext);
+	qp = uobj_get_obj_read(qp, cmd.qp_handle, file->ucontext);
 	if (!qp)
 		goto out;
 
 	resp.bad_wr = 0;
 	ret = qp->device->post_recv(qp->real_qp, wr, &bad_wr);
 
-	put_qp_read(qp);
-
-	if (ret)
+	uobj_put_obj_read(qp);
+	if (ret) {
 		for (next = wr; next; next = next->next) {
 			++resp.bad_wr;
 			if (next == bad_wr)
 				break;
 		}
+	}
 
 	if (copy_to_user((void __user *) (unsigned long) cmd.response,
 			 &resp, sizeof resp))
@@ -2950,14 +2452,14 @@ ssize_t ib_uverbs_post_srq_recv(struct ib_uverbs_file *file,
 	if (IS_ERR(wr))
 		return PTR_ERR(wr);
 
-	srq = idr_read_srq(cmd.srq_handle, file->ucontext);
+	srq = uobj_get_obj_read(srq, cmd.srq_handle, file->ucontext);
 	if (!srq)
 		goto out;
 
 	resp.bad_wr = 0;
 	ret = srq->device->post_srq_recv(srq, wr, &bad_wr);
 
-	put_srq_read(srq);
+	uobj_put_obj_read(srq);
 
 	if (ret)
 		for (next = wr; next; next = next->next) {
@@ -3004,14 +2506,11 @@ ssize_t ib_uverbs_create_ah(struct ib_uverbs_file *file,
 		   (unsigned long)cmd.response + sizeof(resp),
 		   in_len - sizeof(cmd), out_len - sizeof(resp));
 
-	uobj = kmalloc(sizeof *uobj, GFP_KERNEL);
-	if (!uobj)
-		return -ENOMEM;
-
-	init_uobj(uobj, cmd.user_handle, file->ucontext, &ah_lock_class);
-	down_write(&uobj->mutex);
+	uobj  = uobj_alloc(uobj_get_type(ah), file->ucontext);
+	if (IS_ERR(uobj))
+		return PTR_ERR(uobj);
 
-	pd = idr_read_pd(cmd.pd_handle, file->ucontext);
+	pd = uobj_get_obj_read(pd, cmd.pd_handle, file->ucontext);
 	if (!pd) {
 		ret = -EINVAL;
 		goto err;
@@ -3030,28 +2529,20 @@ ssize_t ib_uverbs_create_ah(struct ib_uverbs_file *file,
 	memset(&attr.dmac, 0, sizeof(attr.dmac));
 	memcpy(attr.grh.dgid.raw, cmd.attr.grh.dgid, 16);
 
-	ret = ib_rdmacg_try_charge(&uobj->cg_obj, ib_dev,
-				   RDMACG_RESOURCE_HCA_OBJECT);
-	if (ret)
-		goto err_charge;
-
 	ah = pd->device->create_ah(pd, &attr, &udata);
 
 	if (IS_ERR(ah)) {
 		ret = PTR_ERR(ah);
-		goto err_create;
+		goto err_put;
 	}
 
 	ah->device  = pd->device;
 	ah->pd      = pd;
 	atomic_inc(&pd->usecnt);
 	ah->uobject  = uobj;
+	uobj->user_handle = cmd.user_handle;
 	uobj->object = ah;
 
-	ret = idr_add_uobj(&ib_uverbs_ah_idr, uobj);
-	if (ret)
-		goto err_destroy;
-
 	resp.ah_handle = uobj->id;
 
 	if (copy_to_user((void __user *) (unsigned long) cmd.response,
@@ -3060,32 +2551,19 @@ ssize_t ib_uverbs_create_ah(struct ib_uverbs_file *file,
 		goto err_copy;
 	}
 
-	put_pd_read(pd);
-
-	mutex_lock(&file->mutex);
-	list_add_tail(&uobj->list, &file->ucontext->ah_list);
-	mutex_unlock(&file->mutex);
-
-	uobj->live = 1;
-
-	up_write(&uobj->mutex);
+	uobj_put_obj_read(pd);
+	uobj_alloc_commit(uobj);
 
 	return in_len;
 
 err_copy:
-	idr_remove_uobj(&ib_uverbs_ah_idr, uobj);
-
-err_destroy:
 	ib_destroy_ah(ah);
 
-err_create:
-	ib_rdmacg_uncharge(&uobj->cg_obj, ib_dev, RDMACG_RESOURCE_HCA_OBJECT);
-
-err_charge:
-	put_pd_read(pd);
+err_put:
+	uobj_put_obj_read(pd);
 
 err:
-	put_uobj_write(uobj);
+	uobj_alloc_abort(uobj);
 	return ret;
 }
 
@@ -3094,38 +2572,19 @@ ssize_t ib_uverbs_destroy_ah(struct ib_uverbs_file *file,
 			     const char __user *buf, int in_len, int out_len)
 {
 	struct ib_uverbs_destroy_ah cmd;
-	struct ib_ah		   *ah;
 	struct ib_uobject	   *uobj;
 	int			    ret;
 
 	if (copy_from_user(&cmd, buf, sizeof cmd))
 		return -EFAULT;
 
-	uobj = idr_write_uobj(&ib_uverbs_ah_idr, cmd.ah_handle, file->ucontext);
-	if (!uobj)
-		return -EINVAL;
-	ah = uobj->object;
-
-	ret = ib_destroy_ah(ah);
-	if (!ret)
-		uobj->live = 0;
-
-	put_uobj_write(uobj);
-
-	if (ret)
-		return ret;
-
-	ib_rdmacg_uncharge(&uobj->cg_obj, ib_dev, RDMACG_RESOURCE_HCA_OBJECT);
+	uobj  = uobj_get_write(uobj_get_type(ah), cmd.ah_handle,
+			       file->ucontext);
+	if (IS_ERR(uobj))
+		return PTR_ERR(uobj);
 
-	idr_remove_uobj(&ib_uverbs_ah_idr, uobj);
-
-	mutex_lock(&file->mutex);
-	list_del(&uobj->list);
-	mutex_unlock(&file->mutex);
-
-	put_uobj(uobj);
-
-	return in_len;
+	ret = uobj_remove_commit(uobj);
+	return ret ?: in_len;
 }
 
 ssize_t ib_uverbs_attach_mcast(struct ib_uverbs_file *file,
@@ -3142,12 +2601,13 @@ ssize_t ib_uverbs_attach_mcast(struct ib_uverbs_file *file,
 	if (copy_from_user(&cmd, buf, sizeof cmd))
 		return -EFAULT;
 
-	qp = idr_write_qp(cmd.qp_handle, file->ucontext);
+	qp = uobj_get_obj_read(qp, cmd.qp_handle, file->ucontext);
 	if (!qp)
 		return -EINVAL;
 
 	obj = container_of(qp->uobject, struct ib_uqp_object, uevent.uobject);
 
+	mutex_lock(&obj->mcast_lock);
 	list_for_each_entry(mcast, &obj->mcast_list, list)
 		if (cmd.mlid == mcast->lid &&
 		    !memcmp(cmd.gid, mcast->gid.raw, sizeof mcast->gid.raw)) {
@@ -3171,7 +2631,8 @@ ssize_t ib_uverbs_attach_mcast(struct ib_uverbs_file *file,
 		kfree(mcast);
 
 out_put:
-	put_qp_write(qp);
+	mutex_unlock(&obj->mcast_lock);
+	uobj_put_obj_read(qp);
 
 	return ret ? ret : in_len;
 }
@@ -3190,16 +2651,17 @@ ssize_t ib_uverbs_detach_mcast(struct ib_uverbs_file *file,
 	if (copy_from_user(&cmd, buf, sizeof cmd))
 		return -EFAULT;
 
-	qp = idr_write_qp(cmd.qp_handle, file->ucontext);
+	qp = uobj_get_obj_read(qp, cmd.qp_handle, file->ucontext);
 	if (!qp)
 		return -EINVAL;
 
+	obj = container_of(qp->uobject, struct ib_uqp_object, uevent.uobject);
+	mutex_lock(&obj->mcast_lock);
+
 	ret = ib_detach_mcast(qp, (union ib_gid *) cmd.gid, cmd.mlid);
 	if (ret)
 		goto out_put;
 
-	obj = container_of(qp->uobject, struct ib_uqp_object, uevent.uobject);
-
 	list_for_each_entry(mcast, &obj->mcast_list, list)
 		if (cmd.mlid == mcast->lid &&
 		    !memcmp(cmd.gid, mcast->gid.raw, sizeof mcast->gid.raw)) {
@@ -3209,8 +2671,8 @@ ssize_t ib_uverbs_detach_mcast(struct ib_uverbs_file *file,
 		}
 
 out_put:
-	put_qp_write(qp);
-
+	mutex_unlock(&obj->mcast_lock);
+	uobj_put_obj_read(qp);
 	return ret ? ret : in_len;
 }
 
@@ -3402,20 +2864,18 @@ int ib_uverbs_ex_create_wq(struct ib_uverbs_file *file,
 	if (cmd.comp_mask)
 		return -EOPNOTSUPP;
 
-	obj = kmalloc(sizeof(*obj), GFP_KERNEL);
-	if (!obj)
-		return -ENOMEM;
+	obj  = (struct ib_uwq_object *)uobj_alloc(uobj_get_type(wq),
+						  file->ucontext);
+	if (IS_ERR(obj))
+		return PTR_ERR(obj);
 
-	init_uobj(&obj->uevent.uobject, cmd.user_handle, file->ucontext,
-		  &wq_lock_class);
-	down_write(&obj->uevent.uobject.mutex);
-	pd  = idr_read_pd(cmd.pd_handle, file->ucontext);
+	pd  = uobj_get_obj_read(pd, cmd.pd_handle, file->ucontext);
 	if (!pd) {
 		err = -EINVAL;
 		goto err_uobj;
 	}
 
-	cq = idr_read_cq(cmd.cq_handle, file->ucontext, 0);
+	cq = uobj_get_obj_read(cq, cmd.cq_handle, file->ucontext);
 	if (!cq) {
 		err = -EINVAL;
 		goto err_put_pd;
@@ -3450,9 +2910,6 @@ int ib_uverbs_ex_create_wq(struct ib_uverbs_file *file,
 	atomic_inc(&cq->usecnt);
 	wq->uobject = &obj->uevent.uobject;
 	obj->uevent.uobject.object = wq;
-	err = idr_add_uobj(&ib_uverbs_wq_idr, &obj->uevent.uobject);
-	if (err)
-		goto destroy_wq;
 
 	memset(&resp, 0, sizeof(resp));
 	resp.wq_handle = obj->uevent.uobject.id;
@@ -3465,27 +2922,19 @@ int ib_uverbs_ex_create_wq(struct ib_uverbs_file *file,
 	if (err)
 		goto err_copy;
 
-	put_pd_read(pd);
-	put_cq_read(cq);
-
-	mutex_lock(&file->mutex);
-	list_add_tail(&obj->uevent.uobject.list, &file->ucontext->wq_list);
-	mutex_unlock(&file->mutex);
-
-	obj->uevent.uobject.live = 1;
-	up_write(&obj->uevent.uobject.mutex);
+	uobj_put_obj_read(pd);
+	uobj_put_obj_read(cq);
+	uobj_alloc_commit(&obj->uevent.uobject);
 	return 0;
 
 err_copy:
-	idr_remove_uobj(&ib_uverbs_wq_idr, &obj->uevent.uobject);
-destroy_wq:
 	ib_destroy_wq(wq);
 err_put_cq:
-	put_cq_read(cq);
+	uobj_put_obj_read(cq);
 err_put_pd:
-	put_pd_read(pd);
+	uobj_put_obj_read(pd);
 err_uobj:
-	put_uobj_write(&obj->uevent.uobject);
+	uobj_alloc_abort(&obj->uevent.uobject);
 
 	return err;
 }
@@ -3526,36 +2975,26 @@ int ib_uverbs_ex_destroy_wq(struct ib_uverbs_file *file,
 		return -EOPNOTSUPP;
 
 	resp.response_length = required_resp_len;
-	uobj = idr_write_uobj(&ib_uverbs_wq_idr, cmd.wq_handle,
-			      file->ucontext);
-	if (!uobj)
-		return -EINVAL;
+	uobj  = uobj_get_write(uobj_get_type(wq), cmd.wq_handle,
+			       file->ucontext);
+	if (IS_ERR(uobj))
+		return PTR_ERR(uobj);
 
 	wq = uobj->object;
 	obj = container_of(uobj, struct ib_uwq_object, uevent.uobject);
-	ret = ib_destroy_wq(wq);
-	if (!ret)
-		uobj->live = 0;
-
-	put_uobj_write(uobj);
-	if (ret)
-		return ret;
-
-	idr_remove_uobj(&ib_uverbs_wq_idr, uobj);
-
-	mutex_lock(&file->mutex);
-	list_del(&uobj->list);
-	mutex_unlock(&file->mutex);
+	/*
+	 * Make sure we don't free the memory in remove_commit as we still
+	 * needs the uobject memory to create the response.
+	 */
+	uverbs_uobject_get(uobj);
 
-	ib_uverbs_release_uevent(file, &obj->uevent);
+	ret = uobj_remove_commit(uobj);
 	resp.events_reported = obj->uevent.events_reported;
-	put_uobj(uobj);
-
-	ret = ib_copy_to_udata(ucore, &resp, resp.response_length);
+	uverbs_uobject_put(uobj);
 	if (ret)
 		return ret;
 
-	return 0;
+	return ib_copy_to_udata(ucore, &resp, resp.response_length);
 }
 
 int ib_uverbs_ex_modify_wq(struct ib_uverbs_file *file,
@@ -3588,7 +3027,7 @@ int ib_uverbs_ex_modify_wq(struct ib_uverbs_file *file,
 	if (cmd.attr_mask > (IB_WQ_STATE | IB_WQ_CUR_STATE | IB_WQ_FLAGS))
 		return -EINVAL;
 
-	wq = idr_read_wq(cmd.wq_handle, file->ucontext);
+	wq = uobj_get_obj_read(wq, cmd.wq_handle, file->ucontext);
 	if (!wq)
 		return -EINVAL;
 
@@ -3599,7 +3038,7 @@ int ib_uverbs_ex_modify_wq(struct ib_uverbs_file *file,
 		wq_attr.flags_mask = cmd.flags_mask;
 	}
 	ret = wq->device->modify_wq(wq, &wq_attr, cmd.attr_mask, uhw);
-	put_wq_read(wq);
+	uobj_put_obj_read(wq);
 	return ret;
 }
 
@@ -3677,7 +3116,8 @@ int ib_uverbs_ex_create_rwq_ind_table(struct ib_uverbs_file *file,
 
 	for (num_read_wqs = 0; num_read_wqs < num_wq_handles;
 			num_read_wqs++) {
-		wq = idr_read_wq(wqs_handles[num_read_wqs], file->ucontext);
+		wq = uobj_get_obj_read(wq, wqs_handles[num_read_wqs],
+				       file->ucontext);
 		if (!wq) {
 			err = -EINVAL;
 			goto put_wqs;
@@ -3686,14 +3126,12 @@ int ib_uverbs_ex_create_rwq_ind_table(struct ib_uverbs_file *file,
 		wqs[num_read_wqs] = wq;
 	}
 
-	uobj = kmalloc(sizeof(*uobj), GFP_KERNEL);
-	if (!uobj) {
-		err = -ENOMEM;
+	uobj  = uobj_alloc(uobj_get_type(rwq_ind_table), file->ucontext);
+	if (IS_ERR(uobj)) {
+		err = PTR_ERR(uobj);
 		goto put_wqs;
 	}
 
-	init_uobj(uobj, 0, file->ucontext, &rwq_ind_table_lock_class);
-	down_write(&uobj->mutex);
 	init_attr.log_ind_tbl_size = cmd.log_ind_tbl_size;
 	init_attr.ind_tbl = wqs;
 	rwq_ind_tbl = ib_dev->create_rwq_ind_table(ib_dev, &init_attr, uhw);
@@ -3713,10 +3151,6 @@ int ib_uverbs_ex_create_rwq_ind_table(struct ib_uverbs_file *file,
 	for (i = 0; i < num_wq_handles; i++)
 		atomic_inc(&wqs[i]->usecnt);
 
-	err = idr_add_uobj(&ib_uverbs_rwq_ind_tbl_idr, uobj);
-	if (err)
-		goto destroy_ind_tbl;
-
 	resp.ind_tbl_handle = uobj->id;
 	resp.ind_tbl_num = rwq_ind_tbl->ind_tbl_num;
 	resp.response_length = required_resp_len;
@@ -3729,26 +3163,18 @@ int ib_uverbs_ex_create_rwq_ind_table(struct ib_uverbs_file *file,
 	kfree(wqs_handles);
 
 	for (j = 0; j < num_read_wqs; j++)
-		put_wq_read(wqs[j]);
-
-	mutex_lock(&file->mutex);
-	list_add_tail(&uobj->list, &file->ucontext->rwq_ind_tbl_list);
-	mutex_unlock(&file->mutex);
+		uobj_put_obj_read(wqs[j]);
 
-	uobj->live = 1;
-
-	up_write(&uobj->mutex);
+	uobj_alloc_commit(uobj);
 	return 0;
 
 err_copy:
-	idr_remove_uobj(&ib_uverbs_rwq_ind_tbl_idr, uobj);
-destroy_ind_tbl:
 	ib_destroy_rwq_ind_table(rwq_ind_tbl);
 err_uobj:
-	put_uobj_write(uobj);
+	uobj_alloc_abort(uobj);
 put_wqs:
 	for (j = 0; j < num_read_wqs; j++)
-		put_wq_read(wqs[j]);
+		uobj_put_obj_read(wqs[j]);
 err_free:
 	kfree(wqs_handles);
 	kfree(wqs);
@@ -3761,10 +3187,8 @@ int ib_uverbs_ex_destroy_rwq_ind_table(struct ib_uverbs_file *file,
 				       struct ib_udata *uhw)
 {
 	struct ib_uverbs_ex_destroy_rwq_ind_table	cmd = {};
-	struct ib_rwq_ind_table *rwq_ind_tbl;
 	struct ib_uobject		*uobj;
 	int			ret;
-	struct ib_wq	**ind_tbl;
 	size_t required_cmd_sz;
 
 	required_cmd_sz = offsetof(typeof(cmd), ind_tbl_handle) + sizeof(cmd.ind_tbl_handle);
@@ -3784,31 +3208,12 @@ int ib_uverbs_ex_destroy_rwq_ind_table(struct ib_uverbs_file *file,
 	if (cmd.comp_mask)
 		return -EOPNOTSUPP;
 
-	uobj = idr_write_uobj(&ib_uverbs_rwq_ind_tbl_idr, cmd.ind_tbl_handle,
-			      file->ucontext);
-	if (!uobj)
-		return -EINVAL;
-	rwq_ind_tbl = uobj->object;
-	ind_tbl = rwq_ind_tbl->ind_tbl;
-
-	ret = ib_destroy_rwq_ind_table(rwq_ind_tbl);
-	if (!ret)
-		uobj->live = 0;
-
-	put_uobj_write(uobj);
-
-	if (ret)
-		return ret;
-
-	idr_remove_uobj(&ib_uverbs_rwq_ind_tbl_idr, uobj);
+	uobj  = uobj_get_write(uobj_get_type(rwq_ind_table), cmd.ind_tbl_handle,
+			       file->ucontext);
+	if (IS_ERR(uobj))
+		return PTR_ERR(uobj);
 
-	mutex_lock(&file->mutex);
-	list_del(&uobj->list);
-	mutex_unlock(&file->mutex);
-
-	put_uobj(uobj);
-	kfree(ind_tbl);
-	return ret;
+	return uobj_remove_commit(uobj);
 }
 
 int ib_uverbs_ex_create_flow(struct ib_uverbs_file *file,
@@ -3882,15 +3287,13 @@ int ib_uverbs_ex_create_flow(struct ib_uverbs_file *file,
 		kern_flow_attr = &cmd.flow_attr;
 	}
 
-	uobj = kmalloc(sizeof(*uobj), GFP_KERNEL);
-	if (!uobj) {
-		err = -ENOMEM;
+	uobj  = uobj_alloc(uobj_get_type(flow), file->ucontext);
+	if (IS_ERR(uobj)) {
+		err = PTR_ERR(uobj);
 		goto err_free_attr;
 	}
-	init_uobj(uobj, 0, file->ucontext, &rule_lock_class);
-	down_write(&uobj->mutex);
 
-	qp = idr_read_qp(cmd.qp_handle, file->ucontext);
+	qp = uobj_get_obj_read(qp, cmd.qp_handle, file->ucontext);
 	if (!qp) {
 		err = -EINVAL;
 		goto err_uobj;
@@ -3931,24 +3334,14 @@ int ib_uverbs_ex_create_flow(struct ib_uverbs_file *file,
 		err = -EINVAL;
 		goto err_free;
 	}
-
-	err = ib_rdmacg_try_charge(&uobj->cg_obj, ib_dev,
-				   RDMACG_RESOURCE_HCA_OBJECT);
-	if (err)
-		goto err_free;
-
 	flow_id = ib_create_flow(qp, flow_attr, IB_FLOW_DOMAIN_USER);
 	if (IS_ERR(flow_id)) {
 		err = PTR_ERR(flow_id);
-		goto err_create;
+		goto err_free;
 	}
 	flow_id->uobject = uobj;
 	uobj->object = flow_id;
 
-	err = idr_add_uobj(&ib_uverbs_rule_idr, uobj);
-	if (err)
-		goto destroy_flow;
-
 	memset(&resp, 0, sizeof(resp));
 	resp.flow_handle = uobj->id;
 
@@ -3957,30 +3350,20 @@ int ib_uverbs_ex_create_flow(struct ib_uverbs_file *file,
 	if (err)
 		goto err_copy;
 
-	put_qp_read(qp);
-	mutex_lock(&file->mutex);
-	list_add_tail(&uobj->list, &file->ucontext->rule_list);
-	mutex_unlock(&file->mutex);
-
-	uobj->live = 1;
-
-	up_write(&uobj->mutex);
+	uobj_put_obj_read(qp);
+	uobj_alloc_commit(uobj);
 	kfree(flow_attr);
 	if (cmd.flow_attr.num_of_specs)
 		kfree(kern_flow_attr);
 	return 0;
 err_copy:
-	idr_remove_uobj(&ib_uverbs_rule_idr, uobj);
-destroy_flow:
 	ib_destroy_flow(flow_id);
-err_create:
-	ib_rdmacg_uncharge(&uobj->cg_obj, ib_dev, RDMACG_RESOURCE_HCA_OBJECT);
 err_free:
 	kfree(flow_attr);
 err_put:
-	put_qp_read(qp);
+	uobj_put_obj_read(qp);
 err_uobj:
-	put_uobj_write(uobj);
+	uobj_alloc_abort(uobj);
 err_free_attr:
 	if (cmd.flow_attr.num_of_specs)
 		kfree(kern_flow_attr);
@@ -3993,7 +3376,6 @@ int ib_uverbs_ex_destroy_flow(struct ib_uverbs_file *file,
 			      struct ib_udata *uhw)
 {
 	struct ib_uverbs_destroy_flow	cmd;
-	struct ib_flow			*flow_id;
 	struct ib_uobject		*uobj;
 	int				ret;
 
@@ -4007,29 +3389,12 @@ int ib_uverbs_ex_destroy_flow(struct ib_uverbs_file *file,
 	if (cmd.comp_mask)
 		return -EINVAL;
 
-	uobj = idr_write_uobj(&ib_uverbs_rule_idr, cmd.flow_handle,
-			      file->ucontext);
-	if (!uobj)
-		return -EINVAL;
-	flow_id = uobj->object;
-
-	ret = ib_destroy_flow(flow_id);
-	if (!ret) {
-		ib_rdmacg_uncharge(&uobj->cg_obj, ib_dev,
-				   RDMACG_RESOURCE_HCA_OBJECT);
-		uobj->live = 0;
-	}
-
-	put_uobj_write(uobj);
-
-	idr_remove_uobj(&ib_uverbs_rule_idr, uobj);
-
-	mutex_lock(&file->mutex);
-	list_del(&uobj->list);
-	mutex_unlock(&file->mutex);
-
-	put_uobj(uobj);
+	uobj  = uobj_get_write(uobj_get_type(flow), cmd.flow_handle,
+			       file->ucontext);
+	if (IS_ERR(uobj))
+		return PTR_ERR(uobj);
 
+	ret = uobj_remove_commit(uobj);
 	return ret;
 }
 
@@ -4046,31 +3411,37 @@ static int __uverbs_create_xsrq(struct ib_uverbs_file *file,
 	struct ib_srq_init_attr          attr;
 	int ret;
 
-	obj = kmalloc(sizeof *obj, GFP_KERNEL);
-	if (!obj)
-		return -ENOMEM;
-
-	init_uobj(&obj->uevent.uobject, cmd->user_handle, file->ucontext, &srq_lock_class);
-	down_write(&obj->uevent.uobject.mutex);
+	obj  = (struct ib_usrq_object *)uobj_alloc(uobj_get_type(srq),
+						   file->ucontext);
+	if (IS_ERR(obj))
+		return PTR_ERR(obj);
 
 	if (cmd->srq_type == IB_SRQT_XRC) {
-		attr.ext.xrc.xrcd  = idr_read_xrcd(cmd->xrcd_handle, file->ucontext, &xrcd_uobj);
-		if (!attr.ext.xrc.xrcd) {
+		xrcd_uobj = uobj_get_read(uobj_get_type(xrcd), cmd->xrcd_handle,
+					  file->ucontext);
+		if (IS_ERR(xrcd_uobj)) {
 			ret = -EINVAL;
 			goto err;
 		}
 
+		attr.ext.xrc.xrcd = (struct ib_xrcd *)xrcd_uobj->object;
+		if (!attr.ext.xrc.xrcd) {
+			ret = -EINVAL;
+			goto err_put_xrcd;
+		}
+
 		obj->uxrcd = container_of(xrcd_uobj, struct ib_uxrcd_object, uobject);
 		atomic_inc(&obj->uxrcd->refcnt);
 
-		attr.ext.xrc.cq  = idr_read_cq(cmd->cq_handle, file->ucontext, 0);
+		attr.ext.xrc.cq  = uobj_get_obj_read(cq, cmd->cq_handle,
+						     file->ucontext);
 		if (!attr.ext.xrc.cq) {
 			ret = -EINVAL;
 			goto err_put_xrcd;
 		}
 	}
 
-	pd  = idr_read_pd(cmd->pd_handle, file->ucontext);
+	pd  = uobj_get_obj_read(pd, cmd->pd_handle, file->ucontext);
 	if (!pd) {
 		ret = -EINVAL;
 		goto err_put_cq;
@@ -4086,11 +3457,6 @@ static int __uverbs_create_xsrq(struct ib_uverbs_file *file,
 	obj->uevent.events_reported = 0;
 	INIT_LIST_HEAD(&obj->uevent.event_list);
 
-	ret = ib_rdmacg_try_charge(&obj->uevent.uobject.cg_obj, ib_dev,
-				   RDMACG_RESOURCE_HCA_OBJECT);
-	if (ret)
-		goto err_put_cq;
-
 	srq = pd->device->create_srq(pd, &attr, udata);
 	if (IS_ERR(srq)) {
 		ret = PTR_ERR(srq);
@@ -4115,9 +3481,7 @@ static int __uverbs_create_xsrq(struct ib_uverbs_file *file,
 	atomic_set(&srq->usecnt, 0);
 
 	obj->uevent.uobject.object = srq;
-	ret = idr_add_uobj(&ib_uverbs_srq_idr, &obj->uevent.uobject);
-	if (ret)
-		goto err_destroy;
+	obj->uevent.uobject.user_handle = cmd->user_handle;
 
 	memset(&resp, 0, sizeof resp);
 	resp.srq_handle = obj->uevent.uobject.id;
@@ -4133,44 +3497,32 @@ static int __uverbs_create_xsrq(struct ib_uverbs_file *file,
 	}
 
 	if (cmd->srq_type == IB_SRQT_XRC) {
-		put_uobj_read(xrcd_uobj);
-		put_cq_read(attr.ext.xrc.cq);
+		uobj_put_read(xrcd_uobj);
+		uobj_put_obj_read(attr.ext.xrc.cq);
 	}
-	put_pd_read(pd);
-
-	mutex_lock(&file->mutex);
-	list_add_tail(&obj->uevent.uobject.list, &file->ucontext->srq_list);
-	mutex_unlock(&file->mutex);
-
-	obj->uevent.uobject.live = 1;
-
-	up_write(&obj->uevent.uobject.mutex);
+	uobj_put_obj_read(pd);
+	uobj_alloc_commit(&obj->uevent.uobject);
 
 	return 0;
 
 err_copy:
-	idr_remove_uobj(&ib_uverbs_srq_idr, &obj->uevent.uobject);
-
-err_destroy:
 	ib_destroy_srq(srq);
 
 err_put:
-	ib_rdmacg_uncharge(&obj->uevent.uobject.cg_obj, ib_dev,
-			   RDMACG_RESOURCE_HCA_OBJECT);
-	put_pd_read(pd);
+	uobj_put_obj_read(pd);
 
 err_put_cq:
 	if (cmd->srq_type == IB_SRQT_XRC)
-		put_cq_read(attr.ext.xrc.cq);
+		uobj_put_obj_read(attr.ext.xrc.cq);
 
 err_put_xrcd:
 	if (cmd->srq_type == IB_SRQT_XRC) {
 		atomic_dec(&obj->uxrcd->refcnt);
-		put_uobj_read(xrcd_uobj);
+		uobj_put_read(xrcd_uobj);
 	}
 
 err:
-	put_uobj_write(&obj->uevent.uobject);
+	uobj_alloc_abort(&obj->uevent.uobject);
 	return ret;
 }
 
@@ -4255,7 +3607,7 @@ ssize_t ib_uverbs_modify_srq(struct ib_uverbs_file *file,
 	INIT_UDATA(&udata, buf + sizeof cmd, NULL, in_len - sizeof cmd,
 		   out_len);
 
-	srq = idr_read_srq(cmd.srq_handle, file->ucontext);
+	srq = uobj_get_obj_read(srq, cmd.srq_handle, file->ucontext);
 	if (!srq)
 		return -EINVAL;
 
@@ -4264,7 +3616,7 @@ ssize_t ib_uverbs_modify_srq(struct ib_uverbs_file *file,
 
 	ret = srq->device->modify_srq(srq, &attr, cmd.attr_mask, &udata);
 
-	put_srq_read(srq);
+	uobj_put_obj_read(srq);
 
 	return ret ? ret : in_len;
 }
@@ -4286,13 +3638,13 @@ ssize_t ib_uverbs_query_srq(struct ib_uverbs_file *file,
 	if (copy_from_user(&cmd, buf, sizeof cmd))
 		return -EFAULT;
 
-	srq = idr_read_srq(cmd.srq_handle, file->ucontext);
+	srq = uobj_get_obj_read(srq, cmd.srq_handle, file->ucontext);
 	if (!srq)
 		return -EINVAL;
 
 	ret = ib_query_srq(srq, &attr);
 
-	put_srq_read(srq);
+	uobj_put_obj_read(srq);
 
 	if (ret)
 		return ret;
@@ -4321,53 +3673,39 @@ ssize_t ib_uverbs_destroy_srq(struct ib_uverbs_file *file,
 	struct ib_srq               	 *srq;
 	struct ib_uevent_object        	 *obj;
 	int                         	  ret = -EINVAL;
-	struct ib_usrq_object		 *us;
 	enum ib_srq_type		  srq_type;
 
 	if (copy_from_user(&cmd, buf, sizeof cmd))
 		return -EFAULT;
 
-	uobj = idr_write_uobj(&ib_uverbs_srq_idr, cmd.srq_handle, file->ucontext);
-	if (!uobj)
-		return -EINVAL;
+	uobj  = uobj_get_write(uobj_get_type(srq), cmd.srq_handle,
+			       file->ucontext);
+	if (IS_ERR(uobj))
+		return PTR_ERR(uobj);
+
 	srq = uobj->object;
 	obj = container_of(uobj, struct ib_uevent_object, uobject);
 	srq_type = srq->srq_type;
+	/*
+	 * Make sure we don't free the memory in remove_commit as we still
+	 * needs the uobject memory to create the response.
+	 */
+	uverbs_uobject_get(uobj);
 
-	ret = ib_destroy_srq(srq);
-	if (!ret)
-		uobj->live = 0;
-
-	put_uobj_write(uobj);
+	memset(&resp, 0, sizeof(resp));
 
-	if (ret)
+	ret = uobj_remove_commit(uobj);
+	if (ret) {
+		uverbs_uobject_put(uobj);
 		return ret;
-
-	ib_rdmacg_uncharge(&uobj->cg_obj, ib_dev, RDMACG_RESOURCE_HCA_OBJECT);
-
-	if (srq_type == IB_SRQT_XRC) {
-		us = container_of(obj, struct ib_usrq_object, uevent);
-		atomic_dec(&us->uxrcd->refcnt);
 	}
-
-	idr_remove_uobj(&ib_uverbs_srq_idr, uobj);
-
-	mutex_lock(&file->mutex);
-	list_del(&uobj->list);
-	mutex_unlock(&file->mutex);
-
-	ib_uverbs_release_uevent(file, obj);
-
-	memset(&resp, 0, sizeof resp);
 	resp.events_reported = obj->events_reported;
+	uverbs_uobject_put(uobj);
+	if (copy_to_user((void __user *)(unsigned long)cmd.response,
+			 &resp, sizeof(resp)))
+		return -EFAULT;
 
-	put_uobj(uobj);
-
-	if (copy_to_user((void __user *) (unsigned long) cmd.response,
-			 &resp, sizeof resp))
-		ret = -EFAULT;
-
-	return ret ? ret : in_len;
+	return in_len;
 }
 
 int ib_uverbs_ex_query_device(struct ib_uverbs_file *file,

drivers/infiniband/core/uverbs_main.c

@@ -52,6 +52,7 @@
 
 #include "uverbs.h"
 #include "core_priv.h"
+#include "rdma_core.h"
 
 MODULE_AUTHOR("Roland Dreier");
 MODULE_DESCRIPTION("InfiniBand userspace verbs access");
@@ -67,19 +68,6 @@ enum {
 
 static struct class *uverbs_class;
 
-DEFINE_SPINLOCK(ib_uverbs_idr_lock);
-DEFINE_IDR(ib_uverbs_pd_idr);
-DEFINE_IDR(ib_uverbs_mr_idr);
-DEFINE_IDR(ib_uverbs_mw_idr);
-DEFINE_IDR(ib_uverbs_ah_idr);
-DEFINE_IDR(ib_uverbs_cq_idr);
-DEFINE_IDR(ib_uverbs_qp_idr);
-DEFINE_IDR(ib_uverbs_srq_idr);
-DEFINE_IDR(ib_uverbs_xrcd_idr);
-DEFINE_IDR(ib_uverbs_rule_idr);
-DEFINE_IDR(ib_uverbs_wq_idr);
-DEFINE_IDR(ib_uverbs_rwq_ind_tbl_idr);
-
 static DEFINE_SPINLOCK(map_lock);
 static DECLARE_BITMAP(dev_map, IB_UVERBS_MAX_DEVICES);
 
@@ -168,37 +156,37 @@ static struct kobj_type ib_uverbs_dev_ktype = {
 	.release = ib_uverbs_release_dev,
 };
 
-static void ib_uverbs_release_event_file(struct kref *ref)
+static void ib_uverbs_release_async_event_file(struct kref *ref)
 {
-	struct ib_uverbs_event_file *file =
-		container_of(ref, struct ib_uverbs_event_file, ref);
+	struct ib_uverbs_async_event_file *file =
+		container_of(ref, struct ib_uverbs_async_event_file, ref);
 
 	kfree(file);
 }
 
 void ib_uverbs_release_ucq(struct ib_uverbs_file *file,
-			  struct ib_uverbs_event_file *ev_file,
+			  struct ib_uverbs_completion_event_file *ev_file,
 			  struct ib_ucq_object *uobj)
 {
 	struct ib_uverbs_event *evt, *tmp;
 
 	if (ev_file) {
-		spin_lock_irq(&ev_file->lock);
+		spin_lock_irq(&ev_file->ev_queue.lock);
 		list_for_each_entry_safe(evt, tmp, &uobj->comp_list, obj_list) {
 			list_del(&evt->list);
 			kfree(evt);
 		}
-		spin_unlock_irq(&ev_file->lock);
+		spin_unlock_irq(&ev_file->ev_queue.lock);
 
-		kref_put(&ev_file->ref, ib_uverbs_release_event_file);
+		uverbs_uobject_put(&ev_file->uobj_file.uobj);
 	}
 
-	spin_lock_irq(&file->async_file->lock);
+	spin_lock_irq(&file->async_file->ev_queue.lock);
 	list_for_each_entry_safe(evt, tmp, &uobj->async_list, obj_list) {
 		list_del(&evt->list);
 		kfree(evt);
 	}
-	spin_unlock_irq(&file->async_file->lock);
+	spin_unlock_irq(&file->async_file->ev_queue.lock);
 }
 
 void ib_uverbs_release_uevent(struct ib_uverbs_file *file,
@@ -206,16 +194,16 @@ void ib_uverbs_release_uevent(struct ib_uverbs_file *file,
 {
 	struct ib_uverbs_event *evt, *tmp;
 
-	spin_lock_irq(&file->async_file->lock);
+	spin_lock_irq(&file->async_file->ev_queue.lock);
 	list_for_each_entry_safe(evt, tmp, &uobj->event_list, obj_list) {
 		list_del(&evt->list);
 		kfree(evt);
 	}
-	spin_unlock_irq(&file->async_file->lock);
+	spin_unlock_irq(&file->async_file->ev_queue.lock);
 }
 
-static void ib_uverbs_detach_umcast(struct ib_qp *qp,
-				    struct ib_uqp_object *uobj)
+void ib_uverbs_detach_umcast(struct ib_qp *qp,
+			     struct ib_uqp_object *uobj)
 {
 	struct ib_uverbs_mcast_entry *mcast, *tmp;
 
@@ -227,138 +215,11 @@ static void ib_uverbs_detach_umcast(struct ib_qp *qp,
 }
 
 static int ib_uverbs_cleanup_ucontext(struct ib_uverbs_file *file,
-				      struct ib_ucontext *context)
+				      struct ib_ucontext *context,
+				      bool device_removed)
 {
-	struct ib_uobject *uobj, *tmp;
-
 	context->closing = 1;
-
-	list_for_each_entry_safe(uobj, tmp, &context->ah_list, list) {
-		struct ib_ah *ah = uobj->object;
-
-		idr_remove_uobj(&ib_uverbs_ah_idr, uobj);
-		ib_destroy_ah(ah);
-		ib_rdmacg_uncharge(&uobj->cg_obj, context->device,
-				   RDMACG_RESOURCE_HCA_OBJECT);
-		kfree(uobj);
-	}
-
-	/* Remove MWs before QPs, in order to support type 2A MWs. */
-	list_for_each_entry_safe(uobj, tmp, &context->mw_list, list) {
-		struct ib_mw *mw = uobj->object;
-
-		idr_remove_uobj(&ib_uverbs_mw_idr, uobj);
-		uverbs_dealloc_mw(mw);
-		ib_rdmacg_uncharge(&uobj->cg_obj, context->device,
-				   RDMACG_RESOURCE_HCA_OBJECT);
-		kfree(uobj);
-	}
-
-	list_for_each_entry_safe(uobj, tmp, &context->rule_list, list) {
-		struct ib_flow *flow_id = uobj->object;
-
-		idr_remove_uobj(&ib_uverbs_rule_idr, uobj);
-		ib_destroy_flow(flow_id);
-		ib_rdmacg_uncharge(&uobj->cg_obj, context->device,
-				   RDMACG_RESOURCE_HCA_OBJECT);
-		kfree(uobj);
-	}
-
-	list_for_each_entry_safe(uobj, tmp, &context->qp_list, list) {
-		struct ib_qp *qp = uobj->object;
-		struct ib_uqp_object *uqp =
-			container_of(uobj, struct ib_uqp_object, uevent.uobject);
-
-		idr_remove_uobj(&ib_uverbs_qp_idr, uobj);
-		if (qp == qp->real_qp)
-			ib_uverbs_detach_umcast(qp, uqp);
-		ib_destroy_qp(qp);
-		ib_rdmacg_uncharge(&uobj->cg_obj, context->device,
-				   RDMACG_RESOURCE_HCA_OBJECT);
-		ib_uverbs_release_uevent(file, &uqp->uevent);
-		kfree(uqp);
-	}
-
-	list_for_each_entry_safe(uobj, tmp, &context->rwq_ind_tbl_list, list) {
-		struct ib_rwq_ind_table *rwq_ind_tbl = uobj->object;
-		struct ib_wq **ind_tbl = rwq_ind_tbl->ind_tbl;
-
-		idr_remove_uobj(&ib_uverbs_rwq_ind_tbl_idr, uobj);
-		ib_destroy_rwq_ind_table(rwq_ind_tbl);
-		kfree(ind_tbl);
-		kfree(uobj);
-	}
-
-	list_for_each_entry_safe(uobj, tmp, &context->wq_list, list) {
-		struct ib_wq *wq = uobj->object;
-		struct ib_uwq_object *uwq =
-			container_of(uobj, struct ib_uwq_object, uevent.uobject);
-
-		idr_remove_uobj(&ib_uverbs_wq_idr, uobj);
-		ib_destroy_wq(wq);
-		ib_uverbs_release_uevent(file, &uwq->uevent);
-		kfree(uwq);
-	}
-
-	list_for_each_entry_safe(uobj, tmp, &context->srq_list, list) {
-		struct ib_srq *srq = uobj->object;
-		struct ib_uevent_object *uevent =
-			container_of(uobj, struct ib_uevent_object, uobject);
-
-		idr_remove_uobj(&ib_uverbs_srq_idr, uobj);
-		ib_destroy_srq(srq);
-		ib_rdmacg_uncharge(&uobj->cg_obj, context->device,
-				   RDMACG_RESOURCE_HCA_OBJECT);
-		ib_uverbs_release_uevent(file, uevent);
-		kfree(uevent);
-	}
-
-	list_for_each_entry_safe(uobj, tmp, &context->cq_list, list) {
-		struct ib_cq *cq = uobj->object;
-		struct ib_uverbs_event_file *ev_file = cq->cq_context;
-		struct ib_ucq_object *ucq =
-			container_of(uobj, struct ib_ucq_object, uobject);
-
-		idr_remove_uobj(&ib_uverbs_cq_idr, uobj);
-		ib_destroy_cq(cq);
-		ib_rdmacg_uncharge(&uobj->cg_obj, context->device,
-				   RDMACG_RESOURCE_HCA_OBJECT);
-		ib_uverbs_release_ucq(file, ev_file, ucq);
-		kfree(ucq);
-	}
-
-	list_for_each_entry_safe(uobj, tmp, &context->mr_list, list) {
-		struct ib_mr *mr = uobj->object;
-
-		idr_remove_uobj(&ib_uverbs_mr_idr, uobj);
-		ib_dereg_mr(mr);
-		ib_rdmacg_uncharge(&uobj->cg_obj, context->device,
-				   RDMACG_RESOURCE_HCA_OBJECT);
-		kfree(uobj);
-	}
-
-	mutex_lock(&file->device->xrcd_tree_mutex);
-	list_for_each_entry_safe(uobj, tmp, &context->xrcd_list, list) {
-		struct ib_xrcd *xrcd = uobj->object;
-		struct ib_uxrcd_object *uxrcd =
-			container_of(uobj, struct ib_uxrcd_object, uobject);
-
-		idr_remove_uobj(&ib_uverbs_xrcd_idr, uobj);
-		ib_uverbs_dealloc_xrcd(file->device, xrcd);
-		kfree(uxrcd);
-	}
-	mutex_unlock(&file->device->xrcd_tree_mutex);
-
-	list_for_each_entry_safe(uobj, tmp, &context->pd_list, list) {
-		struct ib_pd *pd = uobj->object;
-
-		idr_remove_uobj(&ib_uverbs_pd_idr, uobj);
-		ib_dealloc_pd(pd);
-		ib_rdmacg_uncharge(&uobj->cg_obj, context->device,
-				   RDMACG_RESOURCE_HCA_OBJECT);
-		kfree(uobj);
-	}
-
+	uverbs_cleanup_ucontext(context, device_removed);
 	put_pid(context->tgid);
 
 	ib_rdmacg_uncharge(&context->cg_obj, context->device,
@@ -372,7 +233,7 @@ static void ib_uverbs_comp_dev(struct ib_uverbs_device *dev)
 	complete(&dev->comp);
 }
 
-static void ib_uverbs_release_file(struct kref *ref)
+void ib_uverbs_release_file(struct kref *ref)
 {
 	struct ib_uverbs_file *file =
 		container_of(ref, struct ib_uverbs_file, ref);
@@ -392,58 +253,54 @@ static void ib_uverbs_release_file(struct kref *ref)
 	kfree(file);
 }
 
-static ssize_t ib_uverbs_event_read(struct file *filp, char __user *buf,
-				    size_t count, loff_t *pos)
+static ssize_t ib_uverbs_event_read(struct ib_uverbs_event_queue *ev_queue,
+				    struct ib_uverbs_file *uverbs_file,
+				    struct file *filp, char __user *buf,
+				    size_t count, loff_t *pos,
+				    size_t eventsz)
 {
-	struct ib_uverbs_event_file *file = filp->private_data;
 	struct ib_uverbs_event *event;
-	int eventsz;
 	int ret = 0;
 
-	spin_lock_irq(&file->lock);
+	spin_lock_irq(&ev_queue->lock);
 
-	while (list_empty(&file->event_list)) {
-		spin_unlock_irq(&file->lock);
+	while (list_empty(&ev_queue->event_list)) {
+		spin_unlock_irq(&ev_queue->lock);
 
 		if (filp->f_flags & O_NONBLOCK)
 			return -EAGAIN;
 
-		if (wait_event_interruptible(file->poll_wait,
-					     (!list_empty(&file->event_list) ||
+		if (wait_event_interruptible(ev_queue->poll_wait,
+					     (!list_empty(&ev_queue->event_list) ||
 			/* The barriers built into wait_event_interruptible()
 			 * and wake_up() guarentee this will see the null set
 			 * without using RCU
 			 */
-					     !file->uverbs_file->device->ib_dev)))
+					     !uverbs_file->device->ib_dev)))
 			return -ERESTARTSYS;
 
 		/* If device was disassociated and no event exists set an error */
-		if (list_empty(&file->event_list) &&
-		    !file->uverbs_file->device->ib_dev)
+		if (list_empty(&ev_queue->event_list) &&
+		    !uverbs_file->device->ib_dev)
 			return -EIO;
 
-		spin_lock_irq(&file->lock);
+		spin_lock_irq(&ev_queue->lock);
 	}
 
-	event = list_entry(file->event_list.next, struct ib_uverbs_event, list);
-
-	if (file->is_async)
-		eventsz = sizeof (struct ib_uverbs_async_event_desc);
-	else
-		eventsz = sizeof (struct ib_uverbs_comp_event_desc);
+	event = list_entry(ev_queue->event_list.next, struct ib_uverbs_event, list);
 
 	if (eventsz > count) {
 		ret   = -EINVAL;
 		event = NULL;
 	} else {
-		list_del(file->event_list.next);
+		list_del(ev_queue->event_list.next);
 		if (event->counter) {
 			++(*event->counter);
 			list_del(&event->obj_list);
 		}
 	}
 
-	spin_unlock_irq(&file->lock);
+	spin_unlock_irq(&ev_queue->lock);
 
 	if (event) {
 		if (copy_to_user(buf, event, eventsz))
@@ -457,87 +314,158 @@ static ssize_t ib_uverbs_event_read(struct file *filp, char __user *buf,
 	return ret;
 }
 
-static unsigned int ib_uverbs_event_poll(struct file *filp,
+static ssize_t ib_uverbs_async_event_read(struct file *filp, char __user *buf,
+					  size_t count, loff_t *pos)
+{
+	struct ib_uverbs_async_event_file *file = filp->private_data;
+
+	return ib_uverbs_event_read(&file->ev_queue, file->uverbs_file, filp,
+				    buf, count, pos,
+				    sizeof(struct ib_uverbs_async_event_desc));
+}
+
+static ssize_t ib_uverbs_comp_event_read(struct file *filp, char __user *buf,
+					 size_t count, loff_t *pos)
+{
+	struct ib_uverbs_completion_event_file *comp_ev_file =
+		filp->private_data;
+
+	return ib_uverbs_event_read(&comp_ev_file->ev_queue,
+				    comp_ev_file->uobj_file.ufile, filp,
+				    buf, count, pos,
+				    sizeof(struct ib_uverbs_comp_event_desc));
+}
+
+static unsigned int ib_uverbs_event_poll(struct ib_uverbs_event_queue *ev_queue,
+					 struct file *filp,
 					 struct poll_table_struct *wait)
 {
 	unsigned int pollflags = 0;
-	struct ib_uverbs_event_file *file = filp->private_data;
 
-	poll_wait(filp, &file->poll_wait, wait);
+	poll_wait(filp, &ev_queue->poll_wait, wait);
 
-	spin_lock_irq(&file->lock);
-	if (!list_empty(&file->event_list))
+	spin_lock_irq(&ev_queue->lock);
+	if (!list_empty(&ev_queue->event_list))
 		pollflags = POLLIN | POLLRDNORM;
-	spin_unlock_irq(&file->lock);
+	spin_unlock_irq(&ev_queue->lock);
 
 	return pollflags;
 }
 
-static int ib_uverbs_event_fasync(int fd, struct file *filp, int on)
+static unsigned int ib_uverbs_async_event_poll(struct file *filp,
+					       struct poll_table_struct *wait)
+{
+	return ib_uverbs_event_poll(filp->private_data, filp, wait);
+}
+
+static unsigned int ib_uverbs_comp_event_poll(struct file *filp,
+					      struct poll_table_struct *wait)
 {
-	struct ib_uverbs_event_file *file = filp->private_data;
+	struct ib_uverbs_completion_event_file *comp_ev_file =
+		filp->private_data;
 
-	return fasync_helper(fd, filp, on, &file->async_queue);
+	return ib_uverbs_event_poll(&comp_ev_file->ev_queue, filp, wait);
 }
 
-static int ib_uverbs_event_close(struct inode *inode, struct file *filp)
+static int ib_uverbs_async_event_fasync(int fd, struct file *filp, int on)
 {
-	struct ib_uverbs_event_file *file = filp->private_data;
+	struct ib_uverbs_event_queue *ev_queue = filp->private_data;
+
+	return fasync_helper(fd, filp, on, &ev_queue->async_queue);
+}
+
+static int ib_uverbs_comp_event_fasync(int fd, struct file *filp, int on)
+{
+	struct ib_uverbs_completion_event_file *comp_ev_file =
+		filp->private_data;
+
+	return fasync_helper(fd, filp, on, &comp_ev_file->ev_queue.async_queue);
+}
+
+static int ib_uverbs_async_event_close(struct inode *inode, struct file *filp)
+{
+	struct ib_uverbs_async_event_file *file = filp->private_data;
+	struct ib_uverbs_file *uverbs_file = file->uverbs_file;
 	struct ib_uverbs_event *entry, *tmp;
 	int closed_already = 0;
 
-	mutex_lock(&file->uverbs_file->device->lists_mutex);
-	spin_lock_irq(&file->lock);
-	closed_already = file->is_closed;
-	file->is_closed = 1;
-	list_for_each_entry_safe(entry, tmp, &file->event_list, list) {
+	mutex_lock(&uverbs_file->device->lists_mutex);
+	spin_lock_irq(&file->ev_queue.lock);
+	closed_already = file->ev_queue.is_closed;
+	file->ev_queue.is_closed = 1;
+	list_for_each_entry_safe(entry, tmp, &file->ev_queue.event_list, list) {
 		if (entry->counter)
 			list_del(&entry->obj_list);
 		kfree(entry);
 	}
-	spin_unlock_irq(&file->lock);
+	spin_unlock_irq(&file->ev_queue.lock);
 	if (!closed_already) {
 		list_del(&file->list);
-		if (file->is_async)
-			ib_unregister_event_handler(&file->uverbs_file->
-				event_handler);
+		ib_unregister_event_handler(&uverbs_file->event_handler);
+	}
+	mutex_unlock(&uverbs_file->device->lists_mutex);
+
+	kref_put(&uverbs_file->ref, ib_uverbs_release_file);
+	kref_put(&file->ref, ib_uverbs_release_async_event_file);
+
+	return 0;
+}
+
+static int ib_uverbs_comp_event_close(struct inode *inode, struct file *filp)
+{
+	struct ib_uverbs_completion_event_file *file = filp->private_data;
+	struct ib_uverbs_event *entry, *tmp;
+
+	spin_lock_irq(&file->ev_queue.lock);
+	list_for_each_entry_safe(entry, tmp, &file->ev_queue.event_list, list) {
+		if (entry->counter)
+			list_del(&entry->obj_list);
+		kfree(entry);
 	}
-	mutex_unlock(&file->uverbs_file->device->lists_mutex);
+	spin_unlock_irq(&file->ev_queue.lock);
 
-	kref_put(&file->uverbs_file->ref, ib_uverbs_release_file);
-	kref_put(&file->ref, ib_uverbs_release_event_file);
+	uverbs_close_fd(filp);
 
 	return 0;
 }
 
-static const struct file_operations uverbs_event_fops = {
+const struct file_operations uverbs_event_fops = {
+	.owner	 = THIS_MODULE,
+	.read	 = ib_uverbs_comp_event_read,
+	.poll    = ib_uverbs_comp_event_poll,
+	.release = ib_uverbs_comp_event_close,
+	.fasync  = ib_uverbs_comp_event_fasync,
+	.llseek	 = no_llseek,
+};
+
+static const struct file_operations uverbs_async_event_fops = {
 	.owner	 = THIS_MODULE,
-	.read	 = ib_uverbs_event_read,
-	.poll    = ib_uverbs_event_poll,
-	.release = ib_uverbs_event_close,
-	.fasync  = ib_uverbs_event_fasync,
+	.read	 = ib_uverbs_async_event_read,
+	.poll    = ib_uverbs_async_event_poll,
+	.release = ib_uverbs_async_event_close,
+	.fasync  = ib_uverbs_async_event_fasync,
 	.llseek	 = no_llseek,
 };
 
 void ib_uverbs_comp_handler(struct ib_cq *cq, void *cq_context)
 {
-	struct ib_uverbs_event_file    *file = cq_context;
+	struct ib_uverbs_event_queue   *ev_queue = cq_context;
 	struct ib_ucq_object	       *uobj;
 	struct ib_uverbs_event	       *entry;
 	unsigned long			flags;
 
-	if (!file)
+	if (!ev_queue)
 		return;
 
-	spin_lock_irqsave(&file->lock, flags);
-	if (file->is_closed) {
-		spin_unlock_irqrestore(&file->lock, flags);
+	spin_lock_irqsave(&ev_queue->lock, flags);
+	if (ev_queue->is_closed) {
+		spin_unlock_irqrestore(&ev_queue->lock, flags);
 		return;
 	}
 
 	entry = kmalloc(sizeof *entry, GFP_ATOMIC);
 	if (!entry) {
-		spin_unlock_irqrestore(&file->lock, flags);
+		spin_unlock_irqrestore(&ev_queue->lock, flags);
 		return;
 	}
 
@@ -546,12 +474,12 @@ void ib_uverbs_comp_handler(struct ib_cq *cq, void *cq_context)
 	entry->desc.comp.cq_handle = cq->uobject->user_handle;
 	entry->counter		   = &uobj->comp_events_reported;
 
-	list_add_tail(&entry->list, &file->event_list);
+	list_add_tail(&entry->list, &ev_queue->event_list);
 	list_add_tail(&entry->obj_list, &uobj->comp_list);
-	spin_unlock_irqrestore(&file->lock, flags);
+	spin_unlock_irqrestore(&ev_queue->lock, flags);
 
-	wake_up_interruptible(&file->poll_wait);
-	kill_fasync(&file->async_queue, SIGIO, POLL_IN);
+	wake_up_interruptible(&ev_queue->poll_wait);
+	kill_fasync(&ev_queue->async_queue, SIGIO, POLL_IN);
 }
 
 static void ib_uverbs_async_handler(struct ib_uverbs_file *file,
@@ -562,15 +490,15 @@ static void ib_uverbs_async_handler(struct ib_uverbs_file *file,
 	struct ib_uverbs_event *entry;
 	unsigned long flags;
 
-	spin_lock_irqsave(&file->async_file->lock, flags);
-	if (file->async_file->is_closed) {
-		spin_unlock_irqrestore(&file->async_file->lock, flags);
+	spin_lock_irqsave(&file->async_file->ev_queue.lock, flags);
+	if (file->async_file->ev_queue.is_closed) {
+		spin_unlock_irqrestore(&file->async_file->ev_queue.lock, flags);
 		return;
 	}
 
 	entry = kmalloc(sizeof *entry, GFP_ATOMIC);
 	if (!entry) {
-		spin_unlock_irqrestore(&file->async_file->lock, flags);
+		spin_unlock_irqrestore(&file->async_file->ev_queue.lock, flags);
 		return;
 	}
 
@@ -579,13 +507,13 @@ static void ib_uverbs_async_handler(struct ib_uverbs_file *file,
 	entry->desc.async.reserved   = 0;
 	entry->counter               = counter;
 
-	list_add_tail(&entry->list, &file->async_file->event_list);
+	list_add_tail(&entry->list, &file->async_file->ev_queue.event_list);
 	if (obj_list)
 		list_add_tail(&entry->obj_list, obj_list);
-	spin_unlock_irqrestore(&file->async_file->lock, flags);
+	spin_unlock_irqrestore(&file->async_file->ev_queue.lock, flags);
 
-	wake_up_interruptible(&file->async_file->poll_wait);
-	kill_fasync(&file->async_file->async_queue, SIGIO, POLL_IN);
+	wake_up_interruptible(&file->async_file->ev_queue.poll_wait);
+	kill_fasync(&file->async_file->ev_queue.async_queue, SIGIO, POLL_IN);
 }
 
 void ib_uverbs_cq_event_handler(struct ib_event *event, void *context_ptr)
@@ -603,7 +531,7 @@ void ib_uverbs_qp_event_handler(struct ib_event *event, void *context_ptr)
 	struct ib_uevent_object *uobj;
 
 	/* for XRC target qp's, check that qp is live */
-	if (!event->element.qp->uobject || !event->element.qp->uobject->live)
+	if (!event->element.qp->uobject)
 		return;
 
 	uobj = container_of(event->element.qp->uobject,
@@ -648,15 +576,23 @@ void ib_uverbs_event_handler(struct ib_event_handler *handler,
 
 void ib_uverbs_free_async_event_file(struct ib_uverbs_file *file)
 {
-	kref_put(&file->async_file->ref, ib_uverbs_release_event_file);
+	kref_put(&file->async_file->ref, ib_uverbs_release_async_event_file);
 	file->async_file = NULL;
 }
 
-struct file *ib_uverbs_alloc_event_file(struct ib_uverbs_file *uverbs_file,
-					struct ib_device	*ib_dev,
-					int is_async)
+void ib_uverbs_init_event_queue(struct ib_uverbs_event_queue *ev_queue)
 {
-	struct ib_uverbs_event_file *ev_file;
+	spin_lock_init(&ev_queue->lock);
+	INIT_LIST_HEAD(&ev_queue->event_list);
+	init_waitqueue_head(&ev_queue->poll_wait);
+	ev_queue->is_closed   = 0;
+	ev_queue->async_queue = NULL;
+}
+
+struct file *ib_uverbs_alloc_async_event_file(struct ib_uverbs_file *uverbs_file,
+					      struct ib_device	*ib_dev)
+{
+	struct ib_uverbs_async_event_file *ev_file;
 	struct file *filp;
 	int ret;
 
@@ -664,16 +600,11 @@ struct file *ib_uverbs_alloc_event_file(struct ib_uverbs_file *uverbs_file,
 	if (!ev_file)
 		return ERR_PTR(-ENOMEM);
 
-	kref_init(&ev_file->ref);
-	spin_lock_init(&ev_file->lock);
-	INIT_LIST_HEAD(&ev_file->event_list);
-	init_waitqueue_head(&ev_file->poll_wait);
+	ib_uverbs_init_event_queue(&ev_file->ev_queue);
 	ev_file->uverbs_file = uverbs_file;
 	kref_get(&ev_file->uverbs_file->ref);
-	ev_file->async_queue = NULL;
-	ev_file->is_closed   = 0;
-
-	filp = anon_inode_getfile("[infinibandevent]", &uverbs_event_fops,
+	kref_init(&ev_file->ref);
+	filp = anon_inode_getfile("[infinibandevent]", &uverbs_async_event_fops,
 				  ev_file, O_RDONLY);
 	if (IS_ERR(filp))
 		goto err_put_refs;
@@ -683,64 +614,33 @@ struct file *ib_uverbs_alloc_event_file(struct ib_uverbs_file *uverbs_file,
 		      &uverbs_file->device->uverbs_events_file_list);
 	mutex_unlock(&uverbs_file->device->lists_mutex);
 
-	if (is_async) {
-		WARN_ON(uverbs_file->async_file);
-		uverbs_file->async_file = ev_file;
-		kref_get(&uverbs_file->async_file->ref);
-		INIT_IB_EVENT_HANDLER(&uverbs_file->event_handler,
-				      ib_dev,
-				      ib_uverbs_event_handler);
-		ret = ib_register_event_handler(&uverbs_file->event_handler);
-		if (ret)
-			goto err_put_file;
-
-		/* At that point async file stuff was fully set */
-		ev_file->is_async = 1;
-	}
+	WARN_ON(uverbs_file->async_file);
+	uverbs_file->async_file = ev_file;
+	kref_get(&uverbs_file->async_file->ref);
+	INIT_IB_EVENT_HANDLER(&uverbs_file->event_handler,
+			      ib_dev,
+			      ib_uverbs_event_handler);
+	ret = ib_register_event_handler(&uverbs_file->event_handler);
+	if (ret)
+		goto err_put_file;
+
+	/* At that point async file stuff was fully set */
 
 	return filp;
 
 err_put_file:
 	fput(filp);
-	kref_put(&uverbs_file->async_file->ref, ib_uverbs_release_event_file);
+	kref_put(&uverbs_file->async_file->ref,
+		 ib_uverbs_release_async_event_file);
 	uverbs_file->async_file = NULL;
 	return ERR_PTR(ret);
 
 err_put_refs:
 	kref_put(&ev_file->uverbs_file->ref, ib_uverbs_release_file);
-	kref_put(&ev_file->ref, ib_uverbs_release_event_file);
+	kref_put(&ev_file->ref, ib_uverbs_release_async_event_file);
 	return filp;
 }
 
-/*
- * Look up a completion event file by FD.  If lookup is successful,
- * takes a ref to the event file struct that it returns; if
- * unsuccessful, returns NULL.
- */
-struct ib_uverbs_event_file *ib_uverbs_lookup_comp_file(int fd)
-{
-	struct ib_uverbs_event_file *ev_file = NULL;
-	struct fd f = fdget(fd);
-
-	if (!f.file)
-		return NULL;
-
-	if (f.file->f_op != &uverbs_event_fops)
-		goto out;
-
-	ev_file = f.file->private_data;
-	if (ev_file->is_async) {
-		ev_file = NULL;
-		goto out;
-	}
-
-	kref_get(&ev_file->ref);
-
-out:
-	fdput(f);
-	return ev_file;
-}
-
 static int verify_command_mask(struct ib_device *ib_dev, __u32 command)
 {
 	u64 mask;
@@ -986,6 +886,8 @@ static int ib_uverbs_open(struct inode *inode, struct file *filp)
 	}
 
 	file->device	 = dev;
+	spin_lock_init(&file->idr_lock);
+	idr_init(&file->idr);
 	file->ucontext	 = NULL;
 	file->async_file = NULL;
 	kref_init(&file->ref);
@@ -1019,10 +921,11 @@ static int ib_uverbs_close(struct inode *inode, struct file *filp)
 
 	mutex_lock(&file->cleanup_mutex);
 	if (file->ucontext) {
-		ib_uverbs_cleanup_ucontext(file, file->ucontext);
+		ib_uverbs_cleanup_ucontext(file, file->ucontext, false);
 		file->ucontext = NULL;
 	}
 	mutex_unlock(&file->cleanup_mutex);
+	idr_destroy(&file->idr);
 
 	mutex_lock(&file->device->lists_mutex);
 	if (!file->is_closed) {
@@ -1032,7 +935,8 @@ static int ib_uverbs_close(struct inode *inode, struct file *filp)
 	mutex_unlock(&file->device->lists_mutex);
 
 	if (file->async_file)
-		kref_put(&file->async_file->ref, ib_uverbs_release_event_file);
+		kref_put(&file->async_file->ref,
+			 ib_uverbs_release_async_event_file);
 
 	kref_put(&file->ref, ib_uverbs_release_file);
 	kobject_put(&dev->kobj);
@@ -1231,7 +1135,7 @@ static void ib_uverbs_free_hw_resources(struct ib_uverbs_device *uverbs_dev,
 					struct ib_device *ib_dev)
 {
 	struct ib_uverbs_file *file;
-	struct ib_uverbs_event_file *event_file;
+	struct ib_uverbs_async_event_file *event_file;
 	struct ib_event event;
 
 	/* Pending running commands to terminate */
@@ -1268,7 +1172,9 @@ static void ib_uverbs_free_hw_resources(struct ib_uverbs_device *uverbs_dev,
 			 * (e.g mmput).
 			 */
 			ib_dev->disassociate_ucontext(ucontext);
-			ib_uverbs_cleanup_ucontext(file, ucontext);
+			mutex_lock(&file->cleanup_mutex);
+			ib_uverbs_cleanup_ucontext(file, ucontext, true);
+			mutex_unlock(&file->cleanup_mutex);
 		}
 
 		mutex_lock(&uverbs_dev->lists_mutex);
@@ -1278,21 +1184,20 @@ static void ib_uverbs_free_hw_resources(struct ib_uverbs_device *uverbs_dev,
 	while (!list_empty(&uverbs_dev->uverbs_events_file_list)) {
 		event_file = list_first_entry(&uverbs_dev->
 					      uverbs_events_file_list,
-					      struct ib_uverbs_event_file,
+					      struct ib_uverbs_async_event_file,
 					      list);
-		spin_lock_irq(&event_file->lock);
-		event_file->is_closed = 1;
-		spin_unlock_irq(&event_file->lock);
+		spin_lock_irq(&event_file->ev_queue.lock);
+		event_file->ev_queue.is_closed = 1;
+		spin_unlock_irq(&event_file->ev_queue.lock);
 
 		list_del(&event_file->list);
-		if (event_file->is_async) {
-			ib_unregister_event_handler(&event_file->uverbs_file->
-						    event_handler);
-			event_file->uverbs_file->event_handler.device = NULL;
-		}
+		ib_unregister_event_handler(
+			&event_file->uverbs_file->event_handler);
+		event_file->uverbs_file->event_handler.device =
+			NULL;
 
-		wake_up_interruptible(&event_file->poll_wait);
-		kill_fasync(&event_file->async_queue, SIGIO, POLL_IN);
+		wake_up_interruptible(&event_file->ev_queue.poll_wait);
+		kill_fasync(&event_file->ev_queue.async_queue, SIGIO, POLL_IN);
 	}
 	mutex_unlock(&uverbs_dev->lists_mutex);
 }
@@ -1396,13 +1301,6 @@ static void __exit ib_uverbs_cleanup(void)
 	unregister_chrdev_region(IB_UVERBS_BASE_DEV, IB_UVERBS_MAX_DEVICES);
 	if (overflow_maj)
 		unregister_chrdev_region(overflow_maj, IB_UVERBS_MAX_DEVICES);
-	idr_destroy(&ib_uverbs_pd_idr);
-	idr_destroy(&ib_uverbs_mr_idr);
-	idr_destroy(&ib_uverbs_mw_idr);
-	idr_destroy(&ib_uverbs_ah_idr);
-	idr_destroy(&ib_uverbs_cq_idr);
-	idr_destroy(&ib_uverbs_qp_idr);
-	idr_destroy(&ib_uverbs_srq_idr);
 }
 
 module_init(ib_uverbs_init);

drivers/infiniband/core/uverbs_std_types.c

+/*
+ * Copyright (c) 2017, Mellanox Technologies inc.  All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include <rdma/uverbs_std_types.h>
+#include <rdma/ib_user_verbs.h>
+#include <rdma/ib_verbs.h>
+#include <linux/bug.h>
+#include <linux/file.h>
+#include "rdma_core.h"
+#include "uverbs.h"
+
+int uverbs_free_ah(struct ib_uobject *uobject,
+		   enum rdma_remove_reason why)
+{
+	return ib_destroy_ah((struct ib_ah *)uobject->object);
+}
+
+int uverbs_free_flow(struct ib_uobject *uobject,
+		     enum rdma_remove_reason why)
+{
+	return ib_destroy_flow((struct ib_flow *)uobject->object);
+}
+
+int uverbs_free_mw(struct ib_uobject *uobject,
+		   enum rdma_remove_reason why)
+{
+	return uverbs_dealloc_mw((struct ib_mw *)uobject->object);
+}
+
+int uverbs_free_qp(struct ib_uobject *uobject,
+		   enum rdma_remove_reason why)
+{
+	struct ib_qp *qp = uobject->object;
+	struct ib_uqp_object *uqp =
+		container_of(uobject, struct ib_uqp_object, uevent.uobject);
+	int ret;
+
+	if (why == RDMA_REMOVE_DESTROY) {
+		if (!list_empty(&uqp->mcast_list))
+			return -EBUSY;
+	} else if (qp == qp->real_qp) {
+		ib_uverbs_detach_umcast(qp, uqp);
+	}
+
+	ret = ib_destroy_qp(qp);
+	if (ret && why == RDMA_REMOVE_DESTROY)
+		return ret;
+
+	if (uqp->uxrcd)
+		atomic_dec(&uqp->uxrcd->refcnt);
+
+	ib_uverbs_release_uevent(uobject->context->ufile, &uqp->uevent);
+	return ret;
+}
+
+int uverbs_free_rwq_ind_tbl(struct ib_uobject *uobject,
+			    enum rdma_remove_reason why)
+{
+	struct ib_rwq_ind_table *rwq_ind_tbl = uobject->object;
+	struct ib_wq **ind_tbl = rwq_ind_tbl->ind_tbl;
+	int ret;
+
+	ret = ib_destroy_rwq_ind_table(rwq_ind_tbl);
+	if (!ret || why != RDMA_REMOVE_DESTROY)
+		kfree(ind_tbl);
+	return ret;
+}
+
+int uverbs_free_wq(struct ib_uobject *uobject,
+		   enum rdma_remove_reason why)
+{
+	struct ib_wq *wq = uobject->object;
+	struct ib_uwq_object *uwq =
+		container_of(uobject, struct ib_uwq_object, uevent.uobject);
+	int ret;
+
+	ret = ib_destroy_wq(wq);
+	if (!ret || why != RDMA_REMOVE_DESTROY)
+		ib_uverbs_release_uevent(uobject->context->ufile, &uwq->uevent);
+	return ret;
+}
+
+int uverbs_free_srq(struct ib_uobject *uobject,
+		    enum rdma_remove_reason why)
+{
+	struct ib_srq *srq = uobject->object;
+	struct ib_uevent_object *uevent =
+		container_of(uobject, struct ib_uevent_object, uobject);
+	enum ib_srq_type  srq_type = srq->srq_type;
+	int ret;
+
+	ret = ib_destroy_srq(srq);
+
+	if (ret && why == RDMA_REMOVE_DESTROY)
+		return ret;
+
+	if (srq_type == IB_SRQT_XRC) {
+		struct ib_usrq_object *us =
+			container_of(uevent, struct ib_usrq_object, uevent);
+
+		atomic_dec(&us->uxrcd->refcnt);
+	}
+
+	ib_uverbs_release_uevent(uobject->context->ufile, uevent);
+	return ret;
+}
+
+int uverbs_free_cq(struct ib_uobject *uobject,
+		   enum rdma_remove_reason why)
+{
+	struct ib_cq *cq = uobject->object;
+	struct ib_uverbs_event_queue *ev_queue = cq->cq_context;
+	struct ib_ucq_object *ucq =
+		container_of(uobject, struct ib_ucq_object, uobject);
+	int ret;
+
+	ret = ib_destroy_cq(cq);
+	if (!ret || why != RDMA_REMOVE_DESTROY)
+		ib_uverbs_release_ucq(uobject->context->ufile, ev_queue ?
+				      container_of(ev_queue,
+						   struct ib_uverbs_completion_event_file,
+						   ev_queue) : NULL,
+				      ucq);
+	return ret;
+}
+
+int uverbs_free_mr(struct ib_uobject *uobject,
+		   enum rdma_remove_reason why)
+{
+	return ib_dereg_mr((struct ib_mr *)uobject->object);
+}
+
+int uverbs_free_xrcd(struct ib_uobject *uobject,
+		     enum rdma_remove_reason why)
+{
+	struct ib_xrcd *xrcd = uobject->object;
+	struct ib_uxrcd_object *uxrcd =
+		container_of(uobject, struct ib_uxrcd_object, uobject);
+	int ret;
+
+	mutex_lock(&uobject->context->ufile->device->xrcd_tree_mutex);
+	if (why == RDMA_REMOVE_DESTROY && atomic_read(&uxrcd->refcnt))
+		ret = -EBUSY;
+	else
+		ret = ib_uverbs_dealloc_xrcd(uobject->context->ufile->device,
+					     xrcd, why);
+	mutex_unlock(&uobject->context->ufile->device->xrcd_tree_mutex);
+
+	return ret;
+}
+
+int uverbs_free_pd(struct ib_uobject *uobject,
+		   enum rdma_remove_reason why)
+{
+	struct ib_pd *pd = uobject->object;
+
+	if (why == RDMA_REMOVE_DESTROY && atomic_read(&pd->usecnt))
+		return -EBUSY;
+
+	ib_dealloc_pd((struct ib_pd *)uobject->object);
+	return 0;
+}
+
+int uverbs_hot_unplug_completion_event_file(struct ib_uobject_file *uobj_file,
+					    enum rdma_remove_reason why)
+{
+	struct ib_uverbs_completion_event_file *comp_event_file =
+		container_of(uobj_file, struct ib_uverbs_completion_event_file,
+			     uobj_file);
+	struct ib_uverbs_event_queue *event_queue = &comp_event_file->ev_queue;
+
+	spin_lock_irq(&event_queue->lock);
+	event_queue->is_closed = 1;
+	spin_unlock_irq(&event_queue->lock);
+
+	if (why == RDMA_REMOVE_DRIVER_REMOVE) {
+		wake_up_interruptible(&event_queue->poll_wait);
+		kill_fasync(&event_queue->async_queue, SIGIO, POLL_IN);
+	}
+	return 0;
+};
+
+const struct uverbs_obj_fd_type uverbs_type_attrs_comp_channel = {
+	.type = UVERBS_TYPE_ALLOC_FD(sizeof(struct ib_uverbs_completion_event_file), 0),
+	.context_closed = uverbs_hot_unplug_completion_event_file,
+	.fops = &uverbs_event_fops,
+	.name = "[infinibandevent]",
+	.flags = O_RDONLY,
+};
+
+const struct uverbs_obj_idr_type uverbs_type_attrs_cq = {
+	.type = UVERBS_TYPE_ALLOC_IDR_SZ(sizeof(struct ib_ucq_object), 0),
+	.destroy_object = uverbs_free_cq,
+};
+
+const struct uverbs_obj_idr_type uverbs_type_attrs_qp = {
+	.type = UVERBS_TYPE_ALLOC_IDR_SZ(sizeof(struct ib_uqp_object), 0),
+	.destroy_object = uverbs_free_qp,
+};
+
+const struct uverbs_obj_idr_type uverbs_type_attrs_mw = {
+	.type = UVERBS_TYPE_ALLOC_IDR(0),
+	.destroy_object = uverbs_free_mw,
+};
+
+const struct uverbs_obj_idr_type uverbs_type_attrs_mr = {
+	/* 1 is used in order to free the MR after all the MWs */
+	.type = UVERBS_TYPE_ALLOC_IDR(1),
+	.destroy_object = uverbs_free_mr,
+};
+
+const struct uverbs_obj_idr_type uverbs_type_attrs_srq = {
+	.type = UVERBS_TYPE_ALLOC_IDR_SZ(sizeof(struct ib_usrq_object), 0),
+	.destroy_object = uverbs_free_srq,
+};
+
+const struct uverbs_obj_idr_type uverbs_type_attrs_ah = {
+	.type = UVERBS_TYPE_ALLOC_IDR(0),
+	.destroy_object = uverbs_free_ah,
+};
+
+const struct uverbs_obj_idr_type uverbs_type_attrs_flow = {
+	.type = UVERBS_TYPE_ALLOC_IDR(0),
+	.destroy_object = uverbs_free_flow,
+};
+
+const struct uverbs_obj_idr_type uverbs_type_attrs_wq = {
+	.type = UVERBS_TYPE_ALLOC_IDR_SZ(sizeof(struct ib_uwq_object), 0),
+	.destroy_object = uverbs_free_wq,
+};
+
+const struct uverbs_obj_idr_type uverbs_type_attrs_rwq_ind_table = {
+	.type = UVERBS_TYPE_ALLOC_IDR(0),
+	.destroy_object = uverbs_free_rwq_ind_tbl,
+};
+
+const struct uverbs_obj_idr_type uverbs_type_attrs_xrcd = {
+	.type = UVERBS_TYPE_ALLOC_IDR_SZ(sizeof(struct ib_uxrcd_object), 0),
+	.destroy_object = uverbs_free_xrcd,
+};
+
+const struct uverbs_obj_idr_type uverbs_type_attrs_pd = {
+	/* 2 is used in order to free the PD after MRs */
+	.type = UVERBS_TYPE_ALLOC_IDR(2),
+	.destroy_object = uverbs_free_pd,
+};

drivers/infiniband/hw/hfi1/chip.c

 /*
- * Copyright(c) 2015, 2016 Intel Corporation.
+ * Copyright(c) 2015 - 2017 Intel Corporation.
  *
  * This file is provided under a dual BSD/GPLv2 license.  When using or
  * redistributing this file, you may do so under either license.
@@ -64,6 +64,7 @@
 #include "platform.h"
 #include "aspm.h"
 #include "affinity.h"
+#include "debugfs.h"
 
 #define NUM_IB_PORTS 1
 
@@ -1045,6 +1046,7 @@ static void dc_start(struct hfi1_devdata *);
 static int qos_rmt_entries(struct hfi1_devdata *dd, unsigned int *mp,
 			   unsigned int *np);
 static void clear_full_mgmt_pkey(struct hfi1_pportdata *ppd);
+static int wait_link_transfer_active(struct hfi1_devdata *dd, int wait_ms);
 
 /*
  * Error interrupt table entry.  This is used as input to the interrupt
@@ -7165,7 +7167,7 @@ static void get_link_widths(struct hfi1_devdata *dd, u16 *tx_width,
 	 * set the max_rate field in handle_verify_cap until v0.19.
 	 */
 	if ((dd->icode == ICODE_RTL_SILICON) &&
-	    (dd->dc8051_ver < dc8051_ver(0, 19))) {
+	    (dd->dc8051_ver < dc8051_ver(0, 19, 0))) {
 		/* max_rate: 0 = 12.5G, 1 = 25G */
 		switch (max_rate) {
 		case 0:
@@ -7350,7 +7352,7 @@ void handle_verify_cap(struct work_struct *work)
 	}
 
 	ppd->link_speed_active = 0;	/* invalid value */
-	if (dd->dc8051_ver < dc8051_ver(0, 20)) {
+	if (dd->dc8051_ver < dc8051_ver(0, 20, 0)) {
 		/* remote_tx_rate: 0 = 12.5G, 1 = 25G */
 		switch (remote_tx_rate) {
 		case 0:
@@ -7897,6 +7899,9 @@ static void handle_dcc_err(struct hfi1_devdata *dd, u32 unused, u64 reg)
 		reg &= ~DCC_ERR_FLG_EN_CSR_ACCESS_BLOCKED_HOST_SMASK;
 	}
 
+	if (unlikely(hfi1_dbg_fault_suppress_err(&dd->verbs_dev)))
+		reg &= ~DCC_ERR_FLG_LATE_EBP_ERR_SMASK;
+
 	/* report any remaining errors */
 	if (reg)
 		dd_dev_info_ratelimited(dd, "DCC Error: %s\n",
@@ -8343,6 +8348,52 @@ static int read_lcb_via_8051(struct hfi1_devdata *dd, u32 addr, u64 *data)
 	return 0;
 }
 
+/*
+ * Provide a cache for some of the LCB registers in case the LCB is
+ * unavailable.
+ * (The LCB is unavailable in certain link states, for example.)
+ */
+struct lcb_datum {
+	u32 off;
+	u64 val;
+};
+
+static struct lcb_datum lcb_cache[] = {
+	{ DC_LCB_ERR_INFO_RX_REPLAY_CNT, 0},
+	{ DC_LCB_ERR_INFO_SEQ_CRC_CNT, 0 },
+	{ DC_LCB_ERR_INFO_REINIT_FROM_PEER_CNT, 0 },
+};
+
+static void update_lcb_cache(struct hfi1_devdata *dd)
+{
+	int i;
+	int ret;
+	u64 val;
+
+	for (i = 0; i < ARRAY_SIZE(lcb_cache); i++) {
+		ret = read_lcb_csr(dd, lcb_cache[i].off, &val);
+
+		/* Update if we get good data */
+		if (likely(ret != -EBUSY))
+			lcb_cache[i].val = val;
+	}
+}
+
+static int read_lcb_cache(u32 off, u64 *val)
+{
+	int i;
+
+	for (i = 0; i < ARRAY_SIZE(lcb_cache); i++) {
+		if (lcb_cache[i].off == off) {
+			*val = lcb_cache[i].val;
+			return 0;
+		}
+	}
+
+	pr_warn("%s bad offset 0x%x\n", __func__, off);
+	return -1;
+}
+
 /*
  * Read an LCB CSR.  Access may not be in host control, so check.
  * Return 0 on success, -EBUSY on failure.
@@ -8354,9 +8405,13 @@ int read_lcb_csr(struct hfi1_devdata *dd, u32 addr, u64 *data)
 	/* if up, go through the 8051 for the value */
 	if (ppd->host_link_state & HLS_UP)
 		return read_lcb_via_8051(dd, addr, data);
-	/* if going up or down, no access */
-	if (ppd->host_link_state & (HLS_GOING_UP | HLS_GOING_OFFLINE))
-		return -EBUSY;
+	/* if going up or down, check the cache, otherwise, no access */
+	if (ppd->host_link_state & (HLS_GOING_UP | HLS_GOING_OFFLINE)) {
+		if (read_lcb_cache(addr, data))
+			return -EBUSY;
+		return 0;
+	}
+
 	/* otherwise, host has access */
 	*data = read_csr(dd, addr);
 	return 0;
@@ -8371,7 +8426,7 @@ static int write_lcb_via_8051(struct hfi1_devdata *dd, u32 addr, u64 data)
 	int ret;
 
 	if (dd->icode == ICODE_FUNCTIONAL_SIMULATOR ||
-	    (dd->dc8051_ver < dc8051_ver(0, 20))) {
+	    (dd->dc8051_ver < dc8051_ver(0, 20, 0))) {
 		if (acquire_lcb_access(dd, 0) == 0) {
 			write_csr(dd, addr, data);
 			release_lcb_access(dd, 0);
@@ -8677,13 +8732,20 @@ static void read_remote_device_id(struct hfi1_devdata *dd, u16 *device_id,
 			& REMOTE_DEVICE_REV_MASK;
 }
 
-void read_misc_status(struct hfi1_devdata *dd, u8 *ver_a, u8 *ver_b)
+void read_misc_status(struct hfi1_devdata *dd, u8 *ver_major, u8 *ver_minor,
+		      u8 *ver_patch)
 {
 	u32 frame;
 
 	read_8051_config(dd, MISC_STATUS, GENERAL_CONFIG, &frame);
-	*ver_a = (frame >> STS_FM_VERSION_A_SHIFT) & STS_FM_VERSION_A_MASK;
-	*ver_b = (frame >> STS_FM_VERSION_B_SHIFT) & STS_FM_VERSION_B_MASK;
+	*ver_major = (frame >> STS_FM_VERSION_MAJOR_SHIFT) &
+		STS_FM_VERSION_MAJOR_MASK;
+	*ver_minor = (frame >> STS_FM_VERSION_MINOR_SHIFT) &
+		STS_FM_VERSION_MINOR_MASK;
+
+	read_8051_config(dd, VERSION_PATCH, GENERAL_CONFIG, &frame);
+	*ver_patch = (frame >> STS_FM_VERSION_PATCH_SHIFT) &
+		STS_FM_VERSION_PATCH_MASK;
 }
 
 static void read_vc_remote_phy(struct hfi1_devdata *dd, u8 *power_management,
@@ -8891,8 +8953,6 @@ int send_idle_sma(struct hfi1_devdata *dd, u64 message)
  */
 static int do_quick_linkup(struct hfi1_devdata *dd)
 {
-	u64 reg;
-	unsigned long timeout;
 	int ret;
 
 	lcb_shutdown(dd, 0);
@@ -8915,19 +8975,9 @@ static int do_quick_linkup(struct hfi1_devdata *dd)
 		write_csr(dd, DC_LCB_CFG_RUN,
 			  1ull << DC_LCB_CFG_RUN_EN_SHIFT);
 
-		/* watch LCB_STS_LINK_TRANSFER_ACTIVE */
-		timeout = jiffies + msecs_to_jiffies(10);
-		while (1) {
-			reg = read_csr(dd, DC_LCB_STS_LINK_TRANSFER_ACTIVE);
-			if (reg)
-				break;
-			if (time_after(jiffies, timeout)) {
-				dd_dev_err(dd,
-					   "timeout waiting for LINK_TRANSFER_ACTIVE\n");
-				return -ETIMEDOUT;
-			}
-			udelay(2);
-		}
+		ret = wait_link_transfer_active(dd, 10);
+		if (ret)
+			return ret;
 
 		write_csr(dd, DC_LCB_CFG_ALLOW_LINK_UP,
 			  1ull << DC_LCB_CFG_ALLOW_LINK_UP_VAL_SHIFT);
@@ -9091,7 +9141,7 @@ static int set_local_link_attributes(struct hfi1_pportdata *ppd)
 	if (ret)
 		goto set_local_link_attributes_fail;
 
-	if (dd->dc8051_ver < dc8051_ver(0, 20)) {
+	if (dd->dc8051_ver < dc8051_ver(0, 20, 0)) {
 		/* set the tx rate to the fastest enabled */
 		if (ppd->link_speed_enabled & OPA_LINK_SPEED_25G)
 			ppd->local_tx_rate = 1;
@@ -9494,8 +9544,11 @@ static int test_qsfp_read(struct hfi1_pportdata *ppd)
 	int ret;
 	u8 status;
 
-	/* report success if not a QSFP */
-	if (ppd->port_type != PORT_TYPE_QSFP)
+	/*
+	 * Report success if not a QSFP or, if it is a QSFP, but the cable is
+	 * not present
+	 */
+	if (ppd->port_type != PORT_TYPE_QSFP || !qsfp_mod_present(ppd))
 		return 0;
 
 	/* read byte 2, the status byte */
@@ -10082,6 +10135,64 @@ static void check_lni_states(struct hfi1_pportdata *ppd)
 	decode_state_complete(ppd, last_remote_state, "received");
 }
 
+/* wait for wait_ms for LINK_TRANSFER_ACTIVE to go to 1 */
+static int wait_link_transfer_active(struct hfi1_devdata *dd, int wait_ms)
+{
+	u64 reg;
+	unsigned long timeout;
+
+	/* watch LCB_STS_LINK_TRANSFER_ACTIVE */
+	timeout = jiffies + msecs_to_jiffies(wait_ms);
+	while (1) {
+		reg = read_csr(dd, DC_LCB_STS_LINK_TRANSFER_ACTIVE);
+		if (reg)
+			break;
+		if (time_after(jiffies, timeout)) {
+			dd_dev_err(dd,
+				   "timeout waiting for LINK_TRANSFER_ACTIVE\n");
+			return -ETIMEDOUT;
+		}
+		udelay(2);
+	}
+	return 0;
+}
+
+/* called when the logical link state is not down as it should be */
+static void force_logical_link_state_down(struct hfi1_pportdata *ppd)
+{
+	struct hfi1_devdata *dd = ppd->dd;
+
+	/*
+	 * Bring link up in LCB loopback
+	 */
+	write_csr(dd, DC_LCB_CFG_TX_FIFOS_RESET, 1);
+	write_csr(dd, DC_LCB_CFG_IGNORE_LOST_RCLK,
+		  DC_LCB_CFG_IGNORE_LOST_RCLK_EN_SMASK);
+
+	write_csr(dd, DC_LCB_CFG_LANE_WIDTH, 0);
+	write_csr(dd, DC_LCB_CFG_REINIT_AS_SLAVE, 0);
+	write_csr(dd, DC_LCB_CFG_CNT_FOR_SKIP_STALL, 0x110);
+	write_csr(dd, DC_LCB_CFG_LOOPBACK, 0x2);
+
+	write_csr(dd, DC_LCB_CFG_TX_FIFOS_RESET, 0);
+	(void)read_csr(dd, DC_LCB_CFG_TX_FIFOS_RESET);
+	udelay(3);
+	write_csr(dd, DC_LCB_CFG_ALLOW_LINK_UP, 1);
+	write_csr(dd, DC_LCB_CFG_RUN, 1ull << DC_LCB_CFG_RUN_EN_SHIFT);
+
+	wait_link_transfer_active(dd, 100);
+
+	/*
+	 * Bring the link down again.
+	 */
+	write_csr(dd, DC_LCB_CFG_TX_FIFOS_RESET, 1);
+	write_csr(dd, DC_LCB_CFG_ALLOW_LINK_UP, 0);
+	write_csr(dd, DC_LCB_CFG_IGNORE_LOST_RCLK, 0);
+
+	/* call again to adjust ppd->statusp, if needed */
+	get_logical_state(ppd);
+}
+
 /*
  * Helper for set_link_state().  Do not call except from that routine.
  * Expects ppd->hls_mutex to be held.
@@ -10098,6 +10209,8 @@ static int goto_offline(struct hfi1_pportdata *ppd, u8 rem_reason)
 	int do_transition;
 	int do_wait;
 
+	update_lcb_cache(dd);
+
 	previous_state = ppd->host_link_state;
 	ppd->host_link_state = HLS_GOING_OFFLINE;
 	pstate = read_physical_state(dd);
@@ -10135,15 +10248,18 @@ static int goto_offline(struct hfi1_pportdata *ppd, u8 rem_reason)
 			return ret;
 	}
 
-	/* make sure the logical state is also down */
-	wait_logical_linkstate(ppd, IB_PORT_DOWN, 1000);
-
 	/*
 	 * Now in charge of LCB - must be after the physical state is
 	 * offline.quiet and before host_link_state is changed.
 	 */
 	set_host_lcb_access(dd);
 	write_csr(dd, DC_LCB_ERR_EN, ~0ull); /* watch LCB errors */
+
+	/* make sure the logical state is also down */
+	ret = wait_logical_linkstate(ppd, IB_PORT_DOWN, 1000);
+	if (ret)
+		force_logical_link_state_down(ppd);
+
 	ppd->host_link_state = HLS_LINK_COOLDOWN; /* LCB access allowed */
 
 	if (ppd->port_type == PORT_TYPE_QSFP &&

drivers/infiniband/hw/hfi1/chip.h

 #ifndef _CHIP_H
 #define _CHIP_H
 /*
- * Copyright(c) 2015, 2016 Intel Corporation.
+ * Copyright(c) 2015 - 2017 Intel Corporation.
  *
  * This file is provided under a dual BSD/GPLv2 license.  When using or
  * redistributing this file, you may do so under either license.
@@ -394,7 +394,8 @@
 #define LAST_REMOTE_STATE_COMPLETE   0x13
 #define LINK_QUALITY_INFO            0x14
 #define REMOTE_DEVICE_ID	     0x15
-#define LINK_DOWN_REASON	     0x16
+#define LINK_DOWN_REASON	     0x16 /* first byte of offset 0x16 */
+#define VERSION_PATCH		     0x16 /* last byte of offset 0x16 */
 
 /* 8051 lane specific register field IDs */
 #define TX_EQ_SETTINGS		0x00
@@ -524,10 +525,12 @@ enum {
 #define SUPPORTED_CRCS (CAP_CRC_14B | CAP_CRC_48B)
 
 /* misc status version fields */
-#define STS_FM_VERSION_A_SHIFT 16
-#define STS_FM_VERSION_A_MASK  0xff
-#define STS_FM_VERSION_B_SHIFT 24
-#define STS_FM_VERSION_B_MASK  0xff
+#define STS_FM_VERSION_MINOR_SHIFT 16
+#define STS_FM_VERSION_MINOR_MASK  0xff
+#define STS_FM_VERSION_MAJOR_SHIFT 24
+#define STS_FM_VERSION_MAJOR_MASK  0xff
+#define STS_FM_VERSION_PATCH_SHIFT 24
+#define STS_FM_VERSION_PATCH_MASK  0xff
 
 /* LCB_CFG_CRC_MODE TX_VAL and RX_VAL CRC mode values */
 #define LCB_CRC_16B			0x0	/* 16b CRC */
@@ -698,7 +701,8 @@ void fabric_serdes_reset(struct hfi1_devdata *dd);
 int read_8051_data(struct hfi1_devdata *dd, u32 addr, u32 len, u64 *result);
 
 /* chip.c */
-void read_misc_status(struct hfi1_devdata *dd, u8 *ver_a, u8 *ver_b);
+void read_misc_status(struct hfi1_devdata *dd, u8 *ver_major, u8 *ver_minor,
+		      u8 *ver_patch);
 void read_guid(struct hfi1_devdata *dd);
 int wait_fm_ready(struct hfi1_devdata *dd, u32 mstimeout);
 void set_link_down_reason(struct hfi1_pportdata *ppd, u8 lcl_reason,

drivers/infiniband/hw/hfi1/debugfs.c

@@ -51,8 +51,12 @@
 #include <linux/export.h>
 #include <linux/module.h>
 #include <linux/string.h>
+#include <linux/types.h>
+#include <linux/ratelimit.h>
+#include <linux/fault-inject.h>
 
 #include "hfi.h"
+#include "trace.h"
 #include "debugfs.h"
 #include "device.h"
 #include "qp.h"
@@ -1063,6 +1067,222 @@ DEBUGFS_SEQ_FILE_OPS(sdma_cpu_list);
 DEBUGFS_SEQ_FILE_OPEN(sdma_cpu_list)
 DEBUGFS_FILE_OPS(sdma_cpu_list);
 
+#ifdef CONFIG_FAULT_INJECTION
+static void *_fault_stats_seq_start(struct seq_file *s, loff_t *pos)
+{
+	struct hfi1_opcode_stats_perctx *opstats;
+
+	if (*pos >= ARRAY_SIZE(opstats->stats))
+		return NULL;
+	return pos;
+}
+
+static void *_fault_stats_seq_next(struct seq_file *s, void *v, loff_t *pos)
+{
+	struct hfi1_opcode_stats_perctx *opstats;
+
+	++*pos;
+	if (*pos >= ARRAY_SIZE(opstats->stats))
+		return NULL;
+	return pos;
+}
+
+static void _fault_stats_seq_stop(struct seq_file *s, void *v)
+{
+}
+
+static int _fault_stats_seq_show(struct seq_file *s, void *v)
+{
+	loff_t *spos = v;
+	loff_t i = *spos, j;
+	u64 n_packets = 0, n_bytes = 0;
+	struct hfi1_ibdev *ibd = (struct hfi1_ibdev *)s->private;
+	struct hfi1_devdata *dd = dd_from_dev(ibd);
+
+	for (j = 0; j < dd->first_user_ctxt; j++) {
+		if (!dd->rcd[j])
+			continue;
+		n_packets += dd->rcd[j]->opstats->stats[i].n_packets;
+		n_bytes += dd->rcd[j]->opstats->stats[i].n_bytes;
+	}
+	if (!n_packets && !n_bytes)
+		return SEQ_SKIP;
+	if (!ibd->fault_opcode->n_rxfaults[i] &&
+	    !ibd->fault_opcode->n_txfaults[i])
+		return SEQ_SKIP;
+	seq_printf(s, "%02llx %llu/%llu (faults rx:%llu faults: tx:%llu)\n", i,
+		   (unsigned long long)n_packets,
+		   (unsigned long long)n_bytes,
+		   (unsigned long long)ibd->fault_opcode->n_rxfaults[i],
+		   (unsigned long long)ibd->fault_opcode->n_txfaults[i]);
+	return 0;
+}
+
+DEBUGFS_SEQ_FILE_OPS(fault_stats);
+DEBUGFS_SEQ_FILE_OPEN(fault_stats);
+DEBUGFS_FILE_OPS(fault_stats);
+
+static void fault_exit_opcode_debugfs(struct hfi1_ibdev *ibd)
+{
+	debugfs_remove_recursive(ibd->fault_opcode->dir);
+	kfree(ibd->fault_opcode);
+	ibd->fault_opcode = NULL;
+}
+
+static int fault_init_opcode_debugfs(struct hfi1_ibdev *ibd)
+{
+	struct dentry *parent = ibd->hfi1_ibdev_dbg;
+
+	ibd->fault_opcode = kzalloc(sizeof(*ibd->fault_opcode), GFP_KERNEL);
+	if (!ibd->fault_opcode)
+		return -ENOMEM;
+
+	ibd->fault_opcode->attr.interval = 1;
+	ibd->fault_opcode->attr.require_end = ULONG_MAX;
+	ibd->fault_opcode->attr.stacktrace_depth = 32;
+	ibd->fault_opcode->attr.dname = NULL;
+	ibd->fault_opcode->attr.verbose = 0;
+	ibd->fault_opcode->fault_by_opcode = false;
+	ibd->fault_opcode->opcode = 0;
+	ibd->fault_opcode->mask = 0xff;
+
+	ibd->fault_opcode->dir =
+		fault_create_debugfs_attr("fault_opcode",
+					  parent,
+					  &ibd->fault_opcode->attr);
+	if (IS_ERR(ibd->fault_opcode->dir)) {
+		kfree(ibd->fault_opcode);
+		return -ENOENT;
+	}
+
+	DEBUGFS_SEQ_FILE_CREATE(fault_stats, ibd->fault_opcode->dir, ibd);
+	if (!debugfs_create_bool("fault_by_opcode", 0600,
+				 ibd->fault_opcode->dir,
+				 &ibd->fault_opcode->fault_by_opcode))
+		goto fail;
+	if (!debugfs_create_x8("opcode", 0600, ibd->fault_opcode->dir,
+			       &ibd->fault_opcode->opcode))
+		goto fail;
+	if (!debugfs_create_x8("mask", 0600, ibd->fault_opcode->dir,
+			       &ibd->fault_opcode->mask))
+		goto fail;
+
+	return 0;
+fail:
+	fault_exit_opcode_debugfs(ibd);
+	return -ENOMEM;
+}
+
+static void fault_exit_packet_debugfs(struct hfi1_ibdev *ibd)
+{
+	debugfs_remove_recursive(ibd->fault_packet->dir);
+	kfree(ibd->fault_packet);
+	ibd->fault_packet = NULL;
+}
+
+static int fault_init_packet_debugfs(struct hfi1_ibdev *ibd)
+{
+	struct dentry *parent = ibd->hfi1_ibdev_dbg;
+
+	ibd->fault_packet = kzalloc(sizeof(*ibd->fault_packet), GFP_KERNEL);
+	if (!ibd->fault_packet)
+		return -ENOMEM;
+
+	ibd->fault_packet->attr.interval = 1;
+	ibd->fault_packet->attr.require_end = ULONG_MAX;
+	ibd->fault_packet->attr.stacktrace_depth = 32;
+	ibd->fault_packet->attr.dname = NULL;
+	ibd->fault_packet->attr.verbose = 0;
+	ibd->fault_packet->fault_by_packet = false;
+
+	ibd->fault_packet->dir =
+		fault_create_debugfs_attr("fault_packet",
+					  parent,
+					  &ibd->fault_opcode->attr);
+	if (IS_ERR(ibd->fault_packet->dir)) {
+		kfree(ibd->fault_packet);
+		return -ENOENT;
+	}
+
+	if (!debugfs_create_bool("fault_by_packet", 0600,
+				 ibd->fault_packet->dir,
+				 &ibd->fault_packet->fault_by_packet))
+		goto fail;
+	if (!debugfs_create_u64("fault_stats", 0400,
+				ibd->fault_packet->dir,
+				&ibd->fault_packet->n_faults))
+		goto fail;
+
+	return 0;
+fail:
+	fault_exit_packet_debugfs(ibd);
+	return -ENOMEM;
+}
+
+static void fault_exit_debugfs(struct hfi1_ibdev *ibd)
+{
+	fault_exit_opcode_debugfs(ibd);
+	fault_exit_packet_debugfs(ibd);
+}
+
+static int fault_init_debugfs(struct hfi1_ibdev *ibd)
+{
+	int ret = 0;
+
+	ret = fault_init_opcode_debugfs(ibd);
+	if (ret)
+		return ret;
+
+	ret = fault_init_packet_debugfs(ibd);
+	if (ret)
+		fault_exit_opcode_debugfs(ibd);
+
+	return ret;
+}
+
+bool hfi1_dbg_fault_suppress_err(struct hfi1_ibdev *ibd)
+{
+	return ibd->fault_suppress_err;
+}
+
+bool hfi1_dbg_fault_opcode(struct rvt_qp *qp, u32 opcode, bool rx)
+{
+	bool ret = false;
+	struct hfi1_ibdev *ibd = to_idev(qp->ibqp.device);
+
+	if (!ibd->fault_opcode || !ibd->fault_opcode->fault_by_opcode)
+		return false;
+	if (ibd->fault_opcode->opcode != (opcode & ibd->fault_opcode->mask))
+		return false;
+	ret = should_fail(&ibd->fault_opcode->attr, 1);
+	if (ret) {
+		trace_hfi1_fault_opcode(qp, opcode);
+		if (rx)
+			ibd->fault_opcode->n_rxfaults[opcode]++;
+		else
+			ibd->fault_opcode->n_txfaults[opcode]++;
+	}
+	return ret;
+}
+
+bool hfi1_dbg_fault_packet(struct hfi1_packet *packet)
+{
+	struct rvt_dev_info *rdi = &packet->rcd->ppd->dd->verbs_dev.rdi;
+	struct hfi1_ibdev *ibd = dev_from_rdi(rdi);
+	bool ret = false;
+
+	if (!ibd->fault_packet || !ibd->fault_packet->fault_by_packet)
+		return false;
+
+	ret = should_fail(&ibd->fault_packet->attr, 1);
+	if (ret) {
+		++ibd->fault_packet->n_faults;
+		trace_hfi1_fault_packet(packet);
+	}
+	return ret;
+}
+#endif
+
 void hfi1_dbg_ibdev_init(struct hfi1_ibdev *ibd)
 {
 	char name[sizeof("port0counters") + 1];
@@ -1112,12 +1332,22 @@ void hfi1_dbg_ibdev_init(struct hfi1_ibdev *ibd)
 					    !port_cntr_ops[i].ops.write ?
 					    S_IRUGO : S_IRUGO | S_IWUSR);
 		}
+
+#ifdef CONFIG_FAULT_INJECTION
+	debugfs_create_bool("fault_suppress_err", 0600,
+			    ibd->hfi1_ibdev_dbg,
+			    &ibd->fault_suppress_err);
+	fault_init_debugfs(ibd);
+#endif
 }
 
 void hfi1_dbg_ibdev_exit(struct hfi1_ibdev *ibd)
 {
 	if (!hfi1_dbg_root)
 		goto out;
+#ifdef CONFIG_FAULT_INJECTION
+	fault_exit_debugfs(ibd);
+#endif
 	debugfs_remove(ibd->hfi1_ibdev_link);
 	debugfs_remove_recursive(ibd->hfi1_ibdev_dbg);
 out:

drivers/infiniband/hw/hfi1/debugfs.h

@@ -53,23 +53,79 @@ void hfi1_dbg_ibdev_init(struct hfi1_ibdev *ibd);
 void hfi1_dbg_ibdev_exit(struct hfi1_ibdev *ibd);
 void hfi1_dbg_init(void);
 void hfi1_dbg_exit(void);
+
+#ifdef CONFIG_FAULT_INJECTION
+#include <linux/fault-inject.h>
+struct fault_opcode {
+	struct fault_attr attr;
+	struct dentry *dir;
+	bool fault_by_opcode;
+	u64 n_rxfaults[256];
+	u64 n_txfaults[256];
+	u8 opcode;
+	u8 mask;
+};
+
+struct fault_packet {
+	struct fault_attr attr;
+	struct dentry *dir;
+	bool fault_by_packet;
+	u64 n_faults;
+};
+
+bool hfi1_dbg_fault_opcode(struct rvt_qp *qp, u32 opcode, bool rx);
+bool hfi1_dbg_fault_packet(struct hfi1_packet *packet);
+bool hfi1_dbg_fault_suppress_err(struct hfi1_ibdev *ibd);
+#else
+static inline bool hfi1_dbg_fault_packet(struct hfi1_packet *packet)
+{
+	return false;
+}
+
+static inline bool hfi1_dbg_fault_opcode(struct rvt_qp *qp,
+					 u32 opcode, bool rx)
+{
+	return false;
+}
+
+static inline bool hfi1_dbg_fault_suppress_err(struct hfi1_ibdev *ibd)
+{
+	return false;
+}
+#endif
+
 #else
 static inline void hfi1_dbg_ibdev_init(struct hfi1_ibdev *ibd)
 {
 }
 
-void hfi1_dbg_ibdev_exit(struct hfi1_ibdev *ibd)
+static inline void hfi1_dbg_ibdev_exit(struct hfi1_ibdev *ibd)
+{
+}
+
+static inline void hfi1_dbg_init(void)
 {
 }
 
-void hfi1_dbg_init(void)
+static inline void hfi1_dbg_exit(void)
 {
 }
 
-void hfi1_dbg_exit(void)
+static inline bool hfi1_dbg_fault_packet(struct hfi1_packet *packet)
 {
+	return false;
 }
 
+static inline bool hfi1_dbg_fault_opcode(struct rvt_qp *qp,
+					 u32 opcode, bool rx)
+{
+	return false;
+}
+
+static inline bool hfi1_dbg_fault_suppress_err(struct hfi1_ibdev *ibd)
+{
+	return false;
+}
 #endif
 
 #endif                          /* _HFI1_DEBUGFS_H */

drivers/infiniband/hw/hfi1/driver.c

@@ -59,6 +59,7 @@
 #include "trace.h"
 #include "qp.h"
 #include "sdma.h"
+#include "debugfs.h"
 
 #undef pr_fmt
 #define pr_fmt(fmt) DRIVER_NAME ": " fmt
@@ -1354,6 +1355,9 @@ void handle_eflags(struct hfi1_packet *packet)
  */
 int process_receive_ib(struct hfi1_packet *packet)
 {
+	if (unlikely(hfi1_dbg_fault_packet(packet)))
+		return RHF_RCV_CONTINUE;
+
 	trace_hfi1_rcvhdr(packet->rcd->ppd->dd,
 			  packet->rcd->ctxt,
 			  rhf_err_flags(packet->rhf),
@@ -1363,6 +1367,11 @@ int process_receive_ib(struct hfi1_packet *packet)
 			  packet->updegr,
 			  rhf_egr_index(packet->rhf));
 
+	if (unlikely(
+		 (hfi1_dbg_fault_suppress_err(&packet->rcd->dd->verbs_dev) &&
+		 (packet->rhf & RHF_DC_ERR))))
+		return RHF_RCV_CONTINUE;
+
 	if (unlikely(rhf_err_flags(packet->rhf))) {
 		handle_eflags(packet);
 		return RHF_RCV_CONTINUE;
@@ -1398,6 +1407,12 @@ int process_receive_bypass(struct hfi1_packet *packet)
 
 int process_receive_error(struct hfi1_packet *packet)
 {
+	/* KHdrHCRCErr -- KDETH packet with a bad HCRC */
+	if (unlikely(
+		 hfi1_dbg_fault_suppress_err(&packet->rcd->dd->verbs_dev) &&
+		 rhf_rcv_type_err(packet->rhf) == 3))
+		return RHF_RCV_CONTINUE;
+
 	handle_eflags(packet);
 
 	if (unlikely(rhf_err_flags(packet->rhf)))
@@ -1409,6 +1424,8 @@ int process_receive_error(struct hfi1_packet *packet)
 
 int kdeth_process_expected(struct hfi1_packet *packet)
 {
+	if (unlikely(hfi1_dbg_fault_packet(packet)))
+		return RHF_RCV_CONTINUE;
 	if (unlikely(rhf_err_flags(packet->rhf)))
 		handle_eflags(packet);
 
@@ -1421,6 +1438,8 @@ int kdeth_process_eager(struct hfi1_packet *packet)
 {
 	if (unlikely(rhf_err_flags(packet->rhf)))
 		handle_eflags(packet);
+	if (unlikely(hfi1_dbg_fault_packet(packet)))
+		return RHF_RCV_CONTINUE;
 
 	dd_dev_err(packet->rcd->dd,
 		   "Unhandled eager packet received. Dropping.\n");

drivers/infiniband/hw/hfi1/firmware.c

 /*
- * Copyright(c) 2015, 2016 Intel Corporation.
+ * Copyright(c) 2015 - 2017 Intel Corporation.
  *
  * This file is provided under a dual BSD/GPLv2 license.  When using or
  * redistributing this file, you may do so under either license.
@@ -1004,7 +1004,9 @@ static int load_8051_firmware(struct hfi1_devdata *dd,
 {
 	u64 reg;
 	int ret;
-	u8 ver_a, ver_b;
+	u8 ver_major;
+	u8 ver_minor;
+	u8 ver_patch;
 
 	/*
 	 * DC Reset sequence
@@ -1073,10 +1075,10 @@ static int load_8051_firmware(struct hfi1_devdata *dd,
 		return -ETIMEDOUT;
 	}
 
-	read_misc_status(dd, &ver_a, &ver_b);
-	dd_dev_info(dd, "8051 firmware version %d.%d\n",
-		    (int)ver_b, (int)ver_a);
-	dd->dc8051_ver = dc8051_ver(ver_b, ver_a);
+	read_misc_status(dd, &ver_major, &ver_minor, &ver_patch);
+	dd_dev_info(dd, "8051 firmware version %d.%d.%d\n",
+		    (int)ver_major, (int)ver_minor, (int)ver_patch);
+	dd->dc8051_ver = dc8051_ver(ver_major, ver_minor, ver_patch);
 
 	return 0;
 }

drivers/infiniband/hw/hfi1/hfi.h

@@ -1020,7 +1020,7 @@ struct hfi1_devdata {
 	u8 qos_shift;
 
 	u16 irev;	/* implementation revision */
-	u16 dc8051_ver; /* 8051 firmware version */
+	u32 dc8051_ver; /* 8051 firmware version */
 
 	spinlock_t hfi1_diag_trans_lock; /* protect diag observer ops */
 	struct platform_config platform_config;
@@ -1167,15 +1167,16 @@ struct hfi1_devdata {
 	bool eprom_available;	/* true if EPROM is available for this device */
 	bool aspm_supported;	/* Does HW support ASPM */
 	bool aspm_enabled;	/* ASPM state: enabled/disabled */
-	struct rhashtable sdma_rht;
+	struct rhashtable *sdma_rht;
 
 	struct kobject kobj;
 };
 
 /* 8051 firmware version helper */
-#define dc8051_ver(a, b) ((a) << 8 | (b))
-#define dc8051_ver_maj(a) ((a & 0xff00) >> 8)
-#define dc8051_ver_min(a)  (a & 0x00ff)
+#define dc8051_ver(a, b, c) ((a) << 16 | (b) << 8 | (c))
+#define dc8051_ver_maj(a) (((a) & 0xff0000) >> 16)
+#define dc8051_ver_min(a) (((a) & 0x00ff00) >> 8)
+#define dc8051_ver_patch(a) ((a) & 0x0000ff)
 
 /* f_put_tid types */
 #define PT_EXPECTED 0

drivers/infiniband/hw/hfi1/init.c

@@ -1425,6 +1425,16 @@ static int init_one(struct pci_dev *pdev, const struct pci_device_id *ent)
 	/* First, lock the non-writable module parameters */
 	HFI1_CAP_LOCK();
 
+	/* Validate dev ids */
+	if (!(ent->device == PCI_DEVICE_ID_INTEL0 ||
+	      ent->device == PCI_DEVICE_ID_INTEL1)) {
+		hfi1_early_err(&pdev->dev,
+			       "Failing on unknown Intel deviceid 0x%x\n",
+			       ent->device);
+		ret = -ENODEV;
+		goto bail;
+	}
+
 	/* Validate some global module parameters */
 	ret = init_validate_rcvhdrcnt(&pdev->dev, rcvhdrcnt);
 	if (ret)
@@ -1470,15 +1480,6 @@ static int init_one(struct pci_dev *pdev, const struct pci_device_id *ent)
 	if (ret)
 		goto bail;
 
-	if (!(ent->device == PCI_DEVICE_ID_INTEL0 ||
-	      ent->device == PCI_DEVICE_ID_INTEL1)) {
-		hfi1_early_err(&pdev->dev,
-			       "Failing on unknown Intel deviceid 0x%x\n",
-			       ent->device);
-		ret = -ENODEV;
-		goto clean_bail;
-	}
-
 	/*
 	 * Do device-specific initialization, function table setup, dd
 	 * allocation, etc.

drivers/infiniband/hw/hfi1/rc.c

@@ -1028,13 +1028,17 @@ void hfi1_rc_send_complete(struct rvt_qp *qp, struct ib_header *hdr)
 		    cmp_psn(qp->s_sending_psn, qp->s_sending_hpsn) <= 0)
 			break;
 		s_last = qp->s_last;
+		trace_hfi1_qp_send_completion(qp, wqe, s_last);
 		if (++s_last >= qp->s_size)
 			s_last = 0;
 		qp->s_last = s_last;
 		/* see post_send() */
 		barrier();
 		rvt_put_swqe(wqe);
-		rvt_qp_swqe_complete(qp, wqe, IB_WC_SUCCESS);
+		rvt_qp_swqe_complete(qp,
+				     wqe,
+				     ib_hfi1_wc_opcode[wqe->wr.opcode],
+				     IB_WC_SUCCESS);
 	}
 	/*
 	 * If we were waiting for sends to complete before re-sending,
@@ -1076,12 +1080,16 @@ static struct rvt_swqe *do_rc_completion(struct rvt_qp *qp,
 
 		rvt_put_swqe(wqe);
 		s_last = qp->s_last;
+		trace_hfi1_qp_send_completion(qp, wqe, s_last);
 		if (++s_last >= qp->s_size)
 			s_last = 0;
 		qp->s_last = s_last;
 		/* see post_send() */
 		barrier();
-		rvt_qp_swqe_complete(qp, wqe, IB_WC_SUCCESS);
+		rvt_qp_swqe_complete(qp,
+				     wqe,
+				     ib_hfi1_wc_opcode[wqe->wr.opcode],
+				     IB_WC_SUCCESS);
 	} else {
 		struct hfi1_pportdata *ppd = ppd_from_ibp(ibp);
 

drivers/infiniband/hw/hfi1/ruc.c

@@ -909,8 +909,10 @@ void hfi1_send_complete(struct rvt_qp *qp, struct rvt_swqe *wqe,
 
 	last = qp->s_last;
 	old_last = last;
+	trace_hfi1_qp_send_completion(qp, wqe, last);
 	if (++last >= qp->s_size)
 		last = 0;
+	trace_hfi1_qp_send_completion(qp, wqe, last);
 	qp->s_last = last;
 	/* See post_send() */
 	barrier();
@@ -920,7 +922,10 @@ void hfi1_send_complete(struct rvt_qp *qp, struct rvt_swqe *wqe,
 	    qp->ibqp.qp_type == IB_QPT_GSI)
 		atomic_dec(&ibah_to_rvtah(wqe->ud_wr.ah)->refcount);
 
-	rvt_qp_swqe_complete(qp, wqe, status);
+	rvt_qp_swqe_complete(qp,
+			     wqe,
+			     ib_hfi1_wc_opcode[wqe->wr.opcode],
+			     status);
 
 	if (qp->s_acked == old_last)
 		qp->s_acked = last;

drivers/infiniband/hw/hfi1/sdma.c

@@ -868,7 +868,7 @@ struct sdma_engine *sdma_select_user_engine(struct hfi1_devdata *dd,
 
 	cpu_id = smp_processor_id();
 	rcu_read_lock();
-	rht_node = rhashtable_lookup_fast(&dd->sdma_rht, &cpu_id,
+	rht_node = rhashtable_lookup_fast(dd->sdma_rht, &cpu_id,
 					  sdma_rht_params);
 
 	if (rht_node && rht_node->map[vl]) {
@@ -962,7 +962,12 @@ ssize_t sdma_set_cpu_to_sde_map(struct sdma_engine *sde, const char *buf,
 			continue;
 		}
 
-		rht_node = rhashtable_lookup_fast(&dd->sdma_rht, &cpu,
+		if (vl >= ARRAY_SIZE(rht_node->map)) {
+			ret = -EINVAL;
+			goto out;
+		}
+
+		rht_node = rhashtable_lookup_fast(dd->sdma_rht, &cpu,
 						  sdma_rht_params);
 		if (!rht_node) {
 			rht_node = kzalloc(sizeof(*rht_node), GFP_KERNEL);
@@ -982,7 +987,7 @@ ssize_t sdma_set_cpu_to_sde_map(struct sdma_engine *sde, const char *buf,
 			rht_node->map[vl]->ctr = 1;
 			rht_node->map[vl]->sde[0] = sde;
 
-			ret = rhashtable_insert_fast(&dd->sdma_rht,
+			ret = rhashtable_insert_fast(dd->sdma_rht,
 						     &rht_node->node,
 						     sdma_rht_params);
 			if (ret) {
@@ -1025,7 +1030,7 @@ ssize_t sdma_set_cpu_to_sde_map(struct sdma_engine *sde, const char *buf,
 		if (cpumask_test_cpu(cpu, mask))
 			continue;
 
-		rht_node = rhashtable_lookup_fast(&dd->sdma_rht, &cpu,
+		rht_node = rhashtable_lookup_fast(dd->sdma_rht, &cpu,
 						  sdma_rht_params);
 		if (rht_node) {
 			bool empty = true;
@@ -1049,7 +1054,7 @@ ssize_t sdma_set_cpu_to_sde_map(struct sdma_engine *sde, const char *buf,
 			}
 
 			if (empty) {
-				ret = rhashtable_remove_fast(&dd->sdma_rht,
+				ret = rhashtable_remove_fast(dd->sdma_rht,
 							     &rht_node->node,
 							     sdma_rht_params);
 				WARN_ON(ret);
@@ -1108,7 +1113,7 @@ void sdma_seqfile_dump_cpu_list(struct seq_file *s,
 	struct sdma_rht_node *rht_node;
 	int i, j;
 
-	rht_node = rhashtable_lookup_fast(&dd->sdma_rht, &cpuid,
+	rht_node = rhashtable_lookup_fast(dd->sdma_rht, &cpuid,
 					  sdma_rht_params);
 	if (!rht_node)
 		return;
@@ -1322,6 +1327,12 @@ static void sdma_clean(struct hfi1_devdata *dd, size_t num_engines)
 	synchronize_rcu();
 	kfree(dd->per_sdma);
 	dd->per_sdma = NULL;
+
+	if (dd->sdma_rht) {
+		rhashtable_free_and_destroy(dd->sdma_rht, sdma_rht_free, NULL);
+		kfree(dd->sdma_rht);
+		dd->sdma_rht = NULL;
+	}
 }
 
 /**
@@ -1341,12 +1352,14 @@ int sdma_init(struct hfi1_devdata *dd, u8 port)
 {
 	unsigned this_idx;
 	struct sdma_engine *sde;
+	struct rhashtable *tmp_sdma_rht;
 	u16 descq_cnt;
 	void *curr_head;
 	struct hfi1_pportdata *ppd = dd->pport + port;
 	u32 per_sdma_credits;
 	uint idle_cnt = sdma_idle_cnt;
 	size_t num_engines = dd->chip_sdma_engines;
+	int ret = -ENOMEM;
 
 	if (!HFI1_CAP_IS_KSET(SDMA)) {
 		HFI1_CAP_CLEAR(SDMA_AHG);
@@ -1378,7 +1391,7 @@ int sdma_init(struct hfi1_devdata *dd, u8 port)
 	/* alloc memory for array of send engines */
 	dd->per_sdma = kcalloc(num_engines, sizeof(*dd->per_sdma), GFP_KERNEL);
 	if (!dd->per_sdma)
-		return -ENOMEM;
+		return ret;
 
 	idle_cnt = ns_to_cclock(dd, idle_cnt);
 	if (!sdma_desct_intr)
@@ -1507,18 +1520,27 @@ int sdma_init(struct hfi1_devdata *dd, u8 port)
 	dd->flags |= HFI1_HAS_SEND_DMA;
 	dd->flags |= idle_cnt ? HFI1_HAS_SDMA_TIMEOUT : 0;
 	dd->num_sdma = num_engines;
-	if (sdma_map_init(dd, port, ppd->vls_operational, NULL))
+	ret = sdma_map_init(dd, port, ppd->vls_operational, NULL);
+	if (ret < 0)
 		goto bail;
 
-	if (rhashtable_init(&dd->sdma_rht, &sdma_rht_params))
+	tmp_sdma_rht = kzalloc(sizeof(*tmp_sdma_rht), GFP_KERNEL);
+	if (!tmp_sdma_rht) {
+		ret = -ENOMEM;
 		goto bail;
+	}
+
+	ret = rhashtable_init(tmp_sdma_rht, &sdma_rht_params);
+	if (ret < 0)
+		goto bail;
+	dd->sdma_rht = tmp_sdma_rht;
 
 	dd_dev_info(dd, "SDMA num_sdma: %u\n", dd->num_sdma);
 	return 0;
 
 bail:
 	sdma_clean(dd, num_engines);
-	return -ENOMEM;
+	return ret;
 }
 
 /**
@@ -1604,7 +1626,6 @@ void sdma_exit(struct hfi1_devdata *dd)
 		sdma_finalput(&sde->state);
 	}
 	sdma_clean(dd, dd->num_sdma);
-	rhashtable_free_and_destroy(&dd->sdma_rht, sdma_rht_free, NULL);
 }
 
 /*

drivers/infiniband/hw/hfi1/trace_misc.h

@@ -72,6 +72,54 @@ TRACE_EVENT(hfi1_interrupt,
 		      __entry->src)
 );
 
+#ifdef CONFIG_FAULT_INJECTION
+TRACE_EVENT(hfi1_fault_opcode,
+	    TP_PROTO(struct rvt_qp *qp, u8 opcode),
+	    TP_ARGS(qp, opcode),
+	    TP_STRUCT__entry(DD_DEV_ENTRY(dd_from_ibdev(qp->ibqp.device))
+			     __field(u32, qpn)
+			     __field(u8, opcode)
+			     ),
+	    TP_fast_assign(DD_DEV_ASSIGN(dd_from_ibdev(qp->ibqp.device))
+			   __entry->qpn = qp->ibqp.qp_num;
+			   __entry->opcode = opcode;
+			   ),
+	    TP_printk("[%s] qpn 0x%x opcode 0x%x",
+		      __get_str(dev), __entry->qpn, __entry->opcode)
+);
+
+TRACE_EVENT(hfi1_fault_packet,
+	    TP_PROTO(struct hfi1_packet *packet),
+	    TP_ARGS(packet),
+	    TP_STRUCT__entry(DD_DEV_ENTRY(packet->rcd->ppd->dd)
+			     __field(u64, eflags)
+			     __field(u32, ctxt)
+			     __field(u32, hlen)
+			     __field(u32, tlen)
+			     __field(u32, updegr)
+			     __field(u32, etail)
+			     ),
+	     TP_fast_assign(DD_DEV_ASSIGN(packet->rcd->ppd->dd);
+			    __entry->eflags = rhf_err_flags(packet->rhf);
+			    __entry->ctxt = packet->rcd->ctxt;
+			    __entry->hlen = packet->hlen;
+			    __entry->tlen = packet->tlen;
+			    __entry->updegr = packet->updegr;
+			    __entry->etail = rhf_egr_index(packet->rhf);
+			    ),
+	     TP_printk(
+		"[%s] ctxt %d eflags 0x%llx hlen %d tlen %d updegr %d etail %d",
+		__get_str(dev),
+		__entry->ctxt,
+		__entry->eflags,
+		__entry->hlen,
+		__entry->tlen,
+		__entry->updegr,
+		__entry->etail
+		)
+);
+#endif
+
 #endif /* __HFI1_TRACE_MISC_H */
 
 #undef TRACE_INCLUDE_PATH

drivers/infiniband/hw/hfi1/trace_rc.h

 /*
-* Copyright(c) 2015, 2016 Intel Corporation.
+* Copyright(c) 2015, 2016, 2017 Intel Corporation.
 *
 * This file is provided under a dual BSD/GPLv2 license.  When using or
 * redistributing this file, you may do so under either license.
@@ -104,11 +104,6 @@ DEFINE_EVENT(hfi1_rc_template, hfi1_ack,
 	     TP_ARGS(qp, psn)
 );
 
-DEFINE_EVENT(hfi1_rc_template, hfi1_timeout,
-	     TP_PROTO(struct rvt_qp *qp, u32 psn),
-	     TP_ARGS(qp, psn)
-);
-
 DEFINE_EVENT(hfi1_rc_template, hfi1_rcv_error,
 	     TP_PROTO(struct rvt_qp *qp, u32 psn),
 	     TP_ARGS(qp, psn)

drivers/infiniband/hw/hfi1/trace_tx.h

@@ -633,6 +633,49 @@ DEFINE_EVENT(hfi1_bct_template, bct_get,
 	     TP_PROTO(struct hfi1_devdata *dd, struct buffer_control *bc),
 	     TP_ARGS(dd, bc));
 
+TRACE_EVENT(
+	hfi1_qp_send_completion,
+	TP_PROTO(struct rvt_qp *qp, struct rvt_swqe *wqe, u32 idx),
+	TP_ARGS(qp, wqe, idx),
+	TP_STRUCT__entry(
+		DD_DEV_ENTRY(dd_from_ibdev(qp->ibqp.device))
+		__field(struct rvt_swqe *, wqe)
+		__field(u64, wr_id)
+		__field(u32, qpn)
+		__field(u32, qpt)
+		__field(u32, length)
+		__field(u32, idx)
+		__field(u32, ssn)
+		__field(enum ib_wr_opcode, opcode)
+		__field(int, send_flags)
+	),
+	TP_fast_assign(
+		DD_DEV_ASSIGN(dd_from_ibdev(qp->ibqp.device))
+		__entry->wqe = wqe;
+		__entry->wr_id = wqe->wr.wr_id;
+		__entry->qpn = qp->ibqp.qp_num;
+		__entry->qpt = qp->ibqp.qp_type;
+		__entry->length = wqe->length;
+		__entry->idx = idx;
+		__entry->ssn = wqe->ssn;
+		__entry->opcode = wqe->wr.opcode;
+		__entry->send_flags = wqe->wr.send_flags;
+	),
+	TP_printk(
+		"[%s] qpn 0x%x qpt %u wqe %p idx %u wr_id %llx length %u ssn %u opcode %x send_flags %x",
+		__get_str(dev),
+		__entry->qpn,
+		__entry->qpt,
+		__entry->wqe,
+		__entry->idx,
+		__entry->wr_id,
+		__entry->length,
+		__entry->ssn,
+		__entry->opcode,
+		__entry->send_flags
+	)
+);
+
 #endif /* __HFI1_TRACE_TX_H */
 
 #undef TRACE_INCLUDE_PATH

drivers/infiniband/hw/hfi1/user_sdma.c

@@ -1615,9 +1615,10 @@ static inline void set_comp_state(struct hfi1_user_sdma_pkt_q *pq,
 {
 	hfi1_cdbg(SDMA, "[%u:%u:%u:%u] Setting completion status %u %d",
 		  pq->dd->unit, pq->ctxt, pq->subctxt, idx, state, ret);
-	cq->comps[idx].status = state;
 	if (state == ERROR)
 		cq->comps[idx].errcode = -ret;
+	smp_wmb(); /* make sure errcode is visible first */
+	cq->comps[idx].status = state;
 	trace_hfi1_sdma_user_completion(pq->dd, pq->ctxt, pq->subctxt,
 					idx, state, ret);
 }

drivers/infiniband/hw/hfi1/verbs.c

@@ -60,6 +60,7 @@
 #include "trace.h"
 #include "qp.h"
 #include "verbs_txreq.h"
+#include "debugfs.h"
 
 static unsigned int hfi1_lkey_table_size = 16;
 module_param_named(lkey_table_size, hfi1_lkey_table_size, uint,
@@ -296,6 +297,22 @@ static inline bool wss_exceeds_threshold(void)
 	return atomic_read(&wss.total_count) >= wss.threshold;
 }
 
+/*
+ * Translate ib_wr_opcode into ib_wc_opcode.
+ */
+const enum ib_wc_opcode ib_hfi1_wc_opcode[] = {
+	[IB_WR_RDMA_WRITE] = IB_WC_RDMA_WRITE,
+	[IB_WR_RDMA_WRITE_WITH_IMM] = IB_WC_RDMA_WRITE,
+	[IB_WR_SEND] = IB_WC_SEND,
+	[IB_WR_SEND_WITH_IMM] = IB_WC_SEND,
+	[IB_WR_RDMA_READ] = IB_WC_RDMA_READ,
+	[IB_WR_ATOMIC_CMP_AND_SWP] = IB_WC_COMP_SWAP,
+	[IB_WR_ATOMIC_FETCH_AND_ADD] = IB_WC_FETCH_ADD,
+	[IB_WR_SEND_WITH_INV] = IB_WC_SEND,
+	[IB_WR_LOCAL_INV] = IB_WC_LOCAL_INV,
+	[IB_WR_REG_MR] = IB_WC_REG_MR
+};
+
 /*
  * Length of header by opcode, 0 --> not supported
  */
@@ -501,6 +518,35 @@ static inline opcode_handler qp_ok(int opcode, struct hfi1_packet *packet)
 	return NULL;
 }
 
+static u64 hfi1_fault_tx(struct rvt_qp *qp, u8 opcode, u64 pbc)
+{
+#ifdef CONFIG_FAULT_INJECTION
+	if ((opcode & IB_OPCODE_MSP) == IB_OPCODE_MSP)
+		/*
+		 * In order to drop non-IB traffic we
+		 * set PbcInsertHrc to NONE (0x2).
+		 * The packet will still be delivered
+		 * to the receiving node but a
+		 * KHdrHCRCErr (KDETH packet with a bad
+		 * HCRC) will be triggered and the
+		 * packet will not be delivered to the
+		 * correct context.
+		 */
+		pbc |= (u64)PBC_IHCRC_NONE << PBC_INSERT_HCRC_SHIFT;
+	else
+		/*
+		 * In order to drop regular verbs
+		 * traffic we set the PbcTestEbp
+		 * flag. The packet will still be
+		 * delivered to the receiving node but
+		 * a 'late ebp error' will be
+		 * triggered and will be dropped.
+		 */
+		pbc |= PBC_TEST_EBP;
+#endif
+	return pbc;
+}
+
 /**
  * hfi1_ib_rcv - process an incoming packet
  * @packet: data packet information
@@ -583,6 +629,11 @@ void hfi1_ib_rcv(struct hfi1_packet *packet)
 			rcu_read_unlock();
 			goto drop;
 		}
+		if (unlikely(hfi1_dbg_fault_opcode(packet->qp, opcode,
+						   true))) {
+			rcu_read_unlock();
+			goto drop;
+		}
 		spin_lock_irqsave(&packet->qp->r_lock, flags);
 		packet_handler = qp_ok(opcode, packet);
 		if (likely(packet_handler))
@@ -781,7 +832,6 @@ static int build_verbs_tx_desc(
 		if (ret)
 			goto bail_txadd;
 	}
-
 	/* add the ulp payload - if any. tx->ss can be NULL for acks */
 	if (tx->ss)
 		ret = build_verbs_ulp_payload(sde, length, tx);
@@ -800,7 +850,6 @@ int hfi1_verbs_send_dma(struct rvt_qp *qp, struct hfi1_pkt_state *ps,
 	struct hfi1_ibdev *dev = ps->dev;
 	struct hfi1_pportdata *ppd = ps->ppd;
 	struct verbs_txreq *tx;
-	u64 pbc_flags = 0;
 	u8 sc5 = priv->s_sc;
 
 	int ret;
@@ -809,12 +858,16 @@ int hfi1_verbs_send_dma(struct rvt_qp *qp, struct hfi1_pkt_state *ps,
 	if (!sdma_txreq_built(&tx->txreq)) {
 		if (likely(pbc == 0)) {
 			u32 vl = sc_to_vlt(dd_from_ibdev(qp->ibqp.device), sc5);
+			u8 opcode = get_opcode(&tx->phdr.hdr);
+
 			/* No vl15 here */
 			/* set PBC_DC_INFO bit (aka SC[4]) in pbc_flags */
-			pbc_flags |= (!!(sc5 & 0x10)) << PBC_DC_INFO_SHIFT;
+			pbc |= (!!(sc5 & 0x10)) << PBC_DC_INFO_SHIFT;
 
+			if (unlikely(hfi1_dbg_fault_opcode(qp, opcode, false)))
+				pbc = hfi1_fault_tx(qp, opcode, pbc);
 			pbc = create_pbc(ppd,
-					 pbc_flags,
+					 pbc,
 					 qp->srate_mbps,
 					 vl,
 					 plen);
@@ -917,7 +970,6 @@ int hfi1_verbs_send_pio(struct rvt_qp *qp, struct hfi1_pkt_state *ps,
 	u32 plen = hdrwords + dwords + 2; /* includes pbc */
 	struct hfi1_pportdata *ppd = ps->ppd;
 	u32 *hdr = (u32 *)&ps->s_txreq->phdr.hdr;
-	u64 pbc_flags = 0;
 	u8 sc5;
 	unsigned long flags = 0;
 	struct send_context *sc;
@@ -942,9 +994,14 @@ int hfi1_verbs_send_pio(struct rvt_qp *qp, struct hfi1_pkt_state *ps,
 
 	if (likely(pbc == 0)) {
 		u8 vl = sc_to_vlt(dd_from_ibdev(qp->ibqp.device), sc5);
+		struct verbs_txreq *tx = ps->s_txreq;
+		u8 opcode = get_opcode(&tx->phdr.hdr);
+
 		/* set PBC_DC_INFO bit (aka SC[4]) in pbc_flags */
-		pbc_flags |= (!!(sc5 & 0x10)) << PBC_DC_INFO_SHIFT;
-		pbc = create_pbc(ppd, pbc_flags, qp->srate_mbps, vl, plen);
+		pbc |= (!!(sc5 & 0x10)) << PBC_DC_INFO_SHIFT;
+		if (unlikely(hfi1_dbg_fault_opcode(qp, opcode, false)))
+			pbc = hfi1_fault_tx(qp, opcode, pbc);
+		pbc = create_pbc(ppd, pbc, qp->srate_mbps, vl, plen);
 	}
 	if (cb)
 		iowait_pio_inc(&priv->s_iowait);
@@ -1220,12 +1277,14 @@ int hfi1_verbs_send(struct rvt_qp *qp, struct hfi1_pkt_state *ps)
 static void hfi1_fill_device_attr(struct hfi1_devdata *dd)
 {
 	struct rvt_dev_info *rdi = &dd->verbs_dev.rdi;
-	u16 ver = dd->dc8051_ver;
+	u32 ver = dd->dc8051_ver;
 
 	memset(&rdi->dparms.props, 0, sizeof(rdi->dparms.props));
 
-	rdi->dparms.props.fw_ver = ((u64)(dc8051_ver_maj(ver)) << 16) |
-				    (u64)dc8051_ver_min(ver);
+	rdi->dparms.props.fw_ver = ((u64)(dc8051_ver_maj(ver)) << 32) |
+		((u64)(dc8051_ver_min(ver)) << 16) |
+		(u64)dc8051_ver_patch(ver);
+
 	rdi->dparms.props.device_cap_flags = IB_DEVICE_BAD_PKEY_CNTR |
 			IB_DEVICE_BAD_QKEY_CNTR | IB_DEVICE_SHUTDOWN_PORT |
 			IB_DEVICE_SYS_IMAGE_GUID | IB_DEVICE_RC_RNR_NAK_GEN |
@@ -1504,10 +1563,10 @@ static void hfi1_get_dev_fw_str(struct ib_device *ibdev, char *str,
 {
 	struct rvt_dev_info *rdi = ib_to_rvt(ibdev);
 	struct hfi1_ibdev *dev = dev_from_rdi(rdi);
-	u16 ver = dd_from_dev(dev)->dc8051_ver;
+	u32 ver = dd_from_dev(dev)->dc8051_ver;
 
-	snprintf(str, str_len, "%u.%u", dc8051_ver_maj(ver),
-		 dc8051_ver_min(ver));
+	snprintf(str, str_len, "%u.%u.%u", dc8051_ver_maj(ver),
+		 dc8051_ver_min(ver), dc8051_ver_patch(ver));
 }
 
 static const char * const driver_cntr_names[] = {
@@ -1524,6 +1583,7 @@ static const char * const driver_cntr_names[] = {
 	"DRIVER_EgrHdrFull"
 };
 
+static DEFINE_MUTEX(cntr_names_lock); /* protects the *_cntr_names bufers */
 static const char **dev_cntr_names;
 static const char **port_cntr_names;
 static int num_driver_cntrs = ARRAY_SIZE(driver_cntr_names);
@@ -1578,6 +1638,7 @@ static struct rdma_hw_stats *alloc_hw_stats(struct ib_device *ibdev,
 {
 	int i, err;
 
+	mutex_lock(&cntr_names_lock);
 	if (!cntr_names_initialized) {
 		struct hfi1_devdata *dd = dd_from_ibdev(ibdev);
 
@@ -1586,8 +1647,10 @@ static struct rdma_hw_stats *alloc_hw_stats(struct ib_device *ibdev,
 				      num_driver_cntrs,
 				      &num_dev_cntrs,
 				      &dev_cntr_names);
-		if (err)
+		if (err) {
+			mutex_unlock(&cntr_names_lock);
 			return NULL;
+		}
 
 		for (i = 0; i < num_driver_cntrs; i++)
 			dev_cntr_names[num_dev_cntrs + i] =
@@ -1601,10 +1664,12 @@ static struct rdma_hw_stats *alloc_hw_stats(struct ib_device *ibdev,
 		if (err) {
 			kfree(dev_cntr_names);
 			dev_cntr_names = NULL;
+			mutex_unlock(&cntr_names_lock);
 			return NULL;
 		}
 		cntr_names_initialized = 1;
 	}
+	mutex_unlock(&cntr_names_lock);
 
 	if (!port_num)
 		return rdma_alloc_hw_stats_struct(
@@ -1823,9 +1888,13 @@ void hfi1_unregister_ib_device(struct hfi1_devdata *dd)
 	del_timer_sync(&dev->mem_timer);
 	verbs_txreq_exit(dev);
 
+	mutex_lock(&cntr_names_lock);
 	kfree(dev_cntr_names);
 	kfree(port_cntr_names);
+	dev_cntr_names = NULL;
+	port_cntr_names = NULL;
 	cntr_names_initialized = 0;
+	mutex_unlock(&cntr_names_lock);
 }
 
 void hfi1_cnp_rcv(struct hfi1_packet *packet)

drivers/infiniband/hw/hfi1/verbs.h

@@ -195,6 +195,11 @@ struct hfi1_ibdev {
 	struct dentry *hfi1_ibdev_dbg;
 	/* per HFI symlinks to above */
 	struct dentry *hfi1_ibdev_link;
+#ifdef CONFIG_FAULT_INJECTION
+	struct fault_opcode *fault_opcode;
+	struct fault_packet *fault_packet;
+	bool fault_suppress_err;
+#endif
 #endif
 };
 

drivers/infiniband/hw/hns/hns_roce_hw_v1.c

@@ -33,6 +33,7 @@
 #include <linux/platform_device.h>
 #include <linux/acpi.h>
 #include <linux/etherdevice.h>
+#include <linux/of.h>
 #include <rdma/ib_umem.h>
 #include "hns_roce_common.h"
 #include "hns_roce_device.h"

drivers/infiniband/hw/qib/qib_rc.c

@@ -938,7 +938,10 @@ void qib_rc_send_complete(struct rvt_qp *qp, struct ib_header *hdr)
 		/* see post_send() */
 		barrier();
 		rvt_put_swqe(wqe);
-		rvt_qp_swqe_complete(qp, wqe, IB_WC_SUCCESS);
+		rvt_qp_swqe_complete(qp,
+				     wqe,
+				     ib_qib_wc_opcode[wqe->wr.opcode],
+				     IB_WC_SUCCESS);
 	}
 	/*
 	 * If we were waiting for sends to complete before resending,
@@ -983,7 +986,10 @@ static struct rvt_swqe *do_rc_completion(struct rvt_qp *qp,
 		qp->s_last = s_last;
 		/* see post_send() */
 		barrier();
-		rvt_qp_swqe_complete(qp, wqe, IB_WC_SUCCESS);
+		rvt_qp_swqe_complete(qp,
+				     wqe,
+				     ib_qib_wc_opcode[wqe->wr.opcode],
+				     IB_WC_SUCCESS);
 	} else
 		this_cpu_inc(*ibp->rvp.rc_delayed_comp);
 

drivers/infiniband/hw/qib/qib_ruc.c

@@ -769,7 +769,10 @@ void qib_send_complete(struct rvt_qp *qp, struct rvt_swqe *wqe,
 	    qp->ibqp.qp_type == IB_QPT_GSI)
 		atomic_dec(&ibah_to_rvtah(wqe->ud_wr.ah)->refcount);
 
-	rvt_qp_swqe_complete(qp, wqe, status);
+	rvt_qp_swqe_complete(qp,
+			     wqe,
+			     ib_qib_wc_opcode[wqe->wr.opcode],
+			     status);
 
 	if (qp->s_acked == old_last)
 		qp->s_acked = last;

drivers/infiniband/hw/qib/qib_verbs.c

@@ -113,6 +113,19 @@ static unsigned int ib_qib_disable_sma;
 module_param_named(disable_sma, ib_qib_disable_sma, uint, S_IWUSR | S_IRUGO);
 MODULE_PARM_DESC(disable_sma, "Disable the SMA");
 
+/*
+ * Translate ib_wr_opcode into ib_wc_opcode.
+ */
+const enum ib_wc_opcode ib_qib_wc_opcode[] = {
+	[IB_WR_RDMA_WRITE] = IB_WC_RDMA_WRITE,
+	[IB_WR_RDMA_WRITE_WITH_IMM] = IB_WC_RDMA_WRITE,
+	[IB_WR_SEND] = IB_WC_SEND,
+	[IB_WR_SEND_WITH_IMM] = IB_WC_SEND,
+	[IB_WR_RDMA_READ] = IB_WC_RDMA_READ,
+	[IB_WR_ATOMIC_CMP_AND_SWP] = IB_WC_COMP_SWAP,
+	[IB_WR_ATOMIC_FETCH_AND_ADD] = IB_WC_FETCH_ADD
+};
+
 /*
  * System image GUID.
  */

drivers/infiniband/sw/rdmavt/cq.c

@@ -50,6 +50,7 @@
 #include <linux/kthread.h>
 #include "cq.h"
 #include "vt.h"
+#include "trace.h"
 
 /**
  * rvt_cq_enter - add a new entry to the completion queue
@@ -93,6 +94,7 @@ void rvt_cq_enter(struct rvt_cq *cq, struct ib_wc *entry, bool solicited)
 		}
 		return;
 	}
+	trace_rvt_cq_enter(cq, entry, head);
 	if (cq->ip) {
 		wc->uqueue[head].wr_id = entry->wr_id;
 		wc->uqueue[head].status = entry->status;
@@ -482,6 +484,7 @@ int rvt_poll_cq(struct ib_cq *ibcq, int num_entries, struct ib_wc *entry)
 		if (tail == wc->head)
 			break;
 		/* The kernel doesn't need a RMB since it has the lock. */
+		trace_rvt_cq_poll(cq, &wc->kqueue[tail], npolled);
 		*entry = wc->kqueue[tail];
 		if (tail >= cq->ibcq.cqe)
 			tail = 0;

drivers/infiniband/sw/rdmavt/mr.c

@@ -191,8 +191,9 @@ static int rvt_alloc_lkey(struct rvt_mregion *mr, int dma_region)
 
 		tmr = rcu_access_pointer(dev->dma_mr);
 		if (!tmr) {
-			rcu_assign_pointer(dev->dma_mr, mr);
 			mr->lkey_published = 1;
+			/* Insure published written first */
+			rcu_assign_pointer(dev->dma_mr, mr);
 			rvt_get_mr(mr);
 		}
 		goto success;
@@ -224,8 +225,9 @@ static int rvt_alloc_lkey(struct rvt_mregion *mr, int dma_region)
 		mr->lkey |= 1 << 8;
 		rkt->gen++;
 	}
-	rcu_assign_pointer(rkt->table[r], mr);
 	mr->lkey_published = 1;
+	/* Insure published written first */
+	rcu_assign_pointer(rkt->table[r], mr);
 success:
 	spin_unlock_irqrestore(&rkt->lock, flags);
 out:
@@ -253,23 +255,24 @@ static void rvt_free_lkey(struct rvt_mregion *mr)
 	spin_lock_irqsave(&rkt->lock, flags);
 	if (!lkey) {
 		if (mr->lkey_published) {
-			RCU_INIT_POINTER(dev->dma_mr, NULL);
+			mr->lkey_published = 0;
+			/* insure published is written before pointer */
+			rcu_assign_pointer(dev->dma_mr, NULL);
 			rvt_put_mr(mr);
 		}
 	} else {
 		if (!mr->lkey_published)
 			goto out;
 		r = lkey >> (32 - dev->dparms.lkey_table_size);
-		RCU_INIT_POINTER(rkt->table[r], NULL);
+		mr->lkey_published = 0;
+		/* insure published is written before pointer */
+		rcu_assign_pointer(rkt->table[r], NULL);
 	}
-	mr->lkey_published = 0;
 	freed++;
 out:
 	spin_unlock_irqrestore(&rkt->lock, flags);
-	if (freed) {
-		synchronize_rcu();
+	if (freed)
 		percpu_ref_kill(&mr->refcount);
-	}
 }
 
 static struct rvt_mr *__rvt_alloc_mr(int count, struct ib_pd *pd)
@@ -822,16 +825,21 @@ int rvt_lkey_ok(struct rvt_lkey_table *rkt, struct rvt_pd *pd,
 		goto ok;
 	}
 	mr = rcu_dereference(rkt->table[sge->lkey >> rkt->shift]);
-	if (unlikely(!mr || atomic_read(&mr->lkey_invalid) ||
-		     mr->lkey != sge->lkey || mr->pd != &pd->ibpd))
+	if (!mr)
 		goto bail;
+	rvt_get_mr(mr);
+	if (!READ_ONCE(mr->lkey_published))
+		goto bail_unref;
+
+	if (unlikely(atomic_read(&mr->lkey_invalid) ||
+		     mr->lkey != sge->lkey || mr->pd != &pd->ibpd))
+		goto bail_unref;
 
 	off = sge->addr - mr->user_base;
 	if (unlikely(sge->addr < mr->user_base ||
 		     off + sge->length > mr->length ||
 		     (mr->access_flags & acc) != acc))
-		goto bail;
-	rvt_get_mr(mr);
+		goto bail_unref;
 	rcu_read_unlock();
 
 	off += mr->offset;
@@ -867,6 +875,8 @@ int rvt_lkey_ok(struct rvt_lkey_table *rkt, struct rvt_pd *pd,
 	isge->n = n;
 ok:
 	return 1;
+bail_unref:
+	rvt_put_mr(mr);
 bail:
 	rcu_read_unlock();
 	return 0;
@@ -922,15 +932,20 @@ int rvt_rkey_ok(struct rvt_qp *qp, struct rvt_sge *sge,
 	}
 
 	mr = rcu_dereference(rkt->table[rkey >> rkt->shift]);
-	if (unlikely(!mr || atomic_read(&mr->lkey_invalid) ||
-		     mr->lkey != rkey || qp->ibqp.pd != mr->pd))
+	if (!mr)
 		goto bail;
+	rvt_get_mr(mr);
+	/* insure mr read is before test */
+	if (!READ_ONCE(mr->lkey_published))
+		goto bail_unref;
+	if (unlikely(atomic_read(&mr->lkey_invalid) ||
+		     mr->lkey != rkey || qp->ibqp.pd != mr->pd))
+		goto bail_unref;
 
 	off = vaddr - mr->iova;
 	if (unlikely(vaddr < mr->iova || off + len > mr->length ||
 		     (mr->access_flags & acc) == 0))
-		goto bail;
-	rvt_get_mr(mr);
+		goto bail_unref;
 	rcu_read_unlock();
 
 	off += mr->offset;
@@ -966,6 +981,8 @@ int rvt_rkey_ok(struct rvt_qp *qp, struct rvt_sge *sge,
 	sge->n = n;
 ok:
 	return 1;
+bail_unref:
+	rvt_put_mr(mr);
 bail:
 	rcu_read_unlock();
 	return 0;

drivers/infiniband/sw/rdmavt/qp.c

 /*
- * Copyright(c) 2016 Intel Corporation.
+ * Copyright(c) 2016, 2017 Intel Corporation.
  *
  * This file is provided under a dual BSD/GPLv2 license.  When using or
  * redistributing this file, you may do so under either license.
@@ -117,23 +117,6 @@ const int ib_rvt_state_ops[IB_QPS_ERR + 1] = {
 };
 EXPORT_SYMBOL(ib_rvt_state_ops);
 
-/*
- * Translate ib_wr_opcode into ib_wc_opcode.
- */
-const enum ib_wc_opcode ib_rvt_wc_opcode[] = {
-	[IB_WR_RDMA_WRITE] = IB_WC_RDMA_WRITE,
-	[IB_WR_RDMA_WRITE_WITH_IMM] = IB_WC_RDMA_WRITE,
-	[IB_WR_SEND] = IB_WC_SEND,
-	[IB_WR_SEND_WITH_IMM] = IB_WC_SEND,
-	[IB_WR_RDMA_READ] = IB_WC_RDMA_READ,
-	[IB_WR_ATOMIC_CMP_AND_SWP] = IB_WC_COMP_SWAP,
-	[IB_WR_ATOMIC_FETCH_AND_ADD] = IB_WC_FETCH_ADD,
-	[IB_WR_SEND_WITH_INV] = IB_WC_SEND,
-	[IB_WR_LOCAL_INV] = IB_WC_LOCAL_INV,
-	[IB_WR_REG_MR] = IB_WC_REG_MR
-};
-EXPORT_SYMBOL(ib_rvt_wc_opcode);
-
 static void get_map_page(struct rvt_qpn_table *qpt,
 			 struct rvt_qpn_map *map,
 			 gfp_t gfp)
@@ -1789,11 +1772,14 @@ static int rvt_post_one_wr(struct rvt_qp *qp,
 					0);
 		qp->s_next_psn = wqe->lpsn + 1;
 	}
-	trace_rvt_post_one_wr(qp, wqe);
-	if (unlikely(reserved_op))
+	if (unlikely(reserved_op)) {
+		wqe->wr.send_flags |= RVT_SEND_RESERVE_USED;
 		rvt_qp_wqe_reserve(qp, wqe);
-	else
+	} else {
+		wqe->wr.send_flags &= ~RVT_SEND_RESERVE_USED;
 		qp->s_avail--;
+	}
+	trace_rvt_post_one_wr(qp, wqe);
 	smp_wmb(); /* see request builders */
 	qp->s_head = next;
 
@@ -2069,8 +2055,12 @@ static void rvt_rc_timeout(unsigned long arg)
 	spin_lock_irqsave(&qp->r_lock, flags);
 	spin_lock(&qp->s_lock);
 	if (qp->s_flags & RVT_S_TIMER) {
+		struct rvt_ibport *rvp = rdi->ports[qp->port_num - 1];
+
 		qp->s_flags &= ~RVT_S_TIMER;
+		rvp->n_rc_timeouts++;
 		del_timer(&qp->s_timer);
+		trace_rvt_rc_timeout(qp, qp->s_last_psn + 1);
 		if (rdi->driver_f.notify_restart_rc)
 			rdi->driver_f.notify_restart_rc(qp,
 							qp->s_last_psn + 1,

drivers/infiniband/sw/rdmavt/trace.h

 /*
- * Copyright(c) 2016 Intel Corporation.
+ * Copyright(c) 2016, 2017 Intel Corporation.
  *
  * This file is provided under a dual BSD/GPLv2 license.  When using or
  * redistributing this file, you may do so under either license.
@@ -52,3 +52,5 @@
 #include "trace_qp.h"
 #include "trace_tx.h"
 #include "trace_mr.h"
+#include "trace_cq.h"
+#include "trace_rc.h"

drivers/infiniband/sw/rdmavt/trace_cq.h

+/*
+ * Copyright(c) 2016 Intel Corporation.
+ *
+ * This file is provided under a dual BSD/GPLv2 license.  When using or
+ * redistributing this file, you may do so under either license.
+ *
+ * GPL LICENSE SUMMARY
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of version 2 of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful, but
+ * WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+ * General Public License for more details.
+ *
+ * BSD LICENSE
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions
+ * are met:
+ *
+ *  - Redistributions of source code must retain the above copyright
+ *    notice, this list of conditions and the following disclaimer.
+ *  - Redistributions in binary form must reproduce the above copyright
+ *    notice, this list of conditions and the following disclaimer in
+ *    the documentation and/or other materials provided with the
+ *    distribution.
+ *  - Neither the name of Intel Corporation nor the names of its
+ *    contributors may be used to endorse or promote products derived
+ *    from this software without specific prior written permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+ * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+ * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+ * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+ * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ *
+ */
+#if !defined(__RVT_TRACE_CQ_H) || defined(TRACE_HEADER_MULTI_READ)
+#define __RVT_TRACE_CQ_H
+
+#include <linux/tracepoint.h>
+#include <linux/trace_seq.h>
+
+#include <rdma/ib_verbs.h>
+#include <rdma/rdmavt_cq.h>
+
+#undef TRACE_SYSTEM
+#define TRACE_SYSTEM rvt_cq
+
+#define wc_opcode_name(opcode) { IB_WC_##opcode, #opcode  }
+#define show_wc_opcode(opcode)                                \
+__print_symbolic(opcode,                                      \
+	wc_opcode_name(SEND),                                 \
+	wc_opcode_name(RDMA_WRITE),                           \
+	wc_opcode_name(RDMA_READ),                            \
+	wc_opcode_name(COMP_SWAP),                            \
+	wc_opcode_name(FETCH_ADD),                            \
+	wc_opcode_name(LSO),                                  \
+	wc_opcode_name(LOCAL_INV),                            \
+	wc_opcode_name(REG_MR),                               \
+	wc_opcode_name(MASKED_COMP_SWAP),                     \
+	wc_opcode_name(RECV),                                 \
+	wc_opcode_name(RECV_RDMA_WITH_IMM))
+
+#define CQ_PRN \
+"[%s] idx %u wr_id %llx status %u opcode %u,%s length %u qpn %x"
+
+DECLARE_EVENT_CLASS(
+	rvt_cq_entry_template,
+	TP_PROTO(struct rvt_cq *cq, struct ib_wc *wc, u32 idx),
+	TP_ARGS(cq, wc, idx),
+	TP_STRUCT__entry(
+		RDI_DEV_ENTRY(cq->rdi)
+		__field(u64, wr_id)
+		__field(u32, status)
+		__field(u32, opcode)
+		__field(u32, qpn)
+		__field(u32, length)
+		__field(u32, idx)
+	),
+	TP_fast_assign(
+		RDI_DEV_ASSIGN(cq->rdi)
+		__entry->wr_id = wc->wr_id;
+		__entry->status = wc->status;
+		__entry->opcode = wc->opcode;
+		__entry->length = wc->byte_len;
+		__entry->qpn = wc->qp->qp_num;
+		__entry->idx = idx;
+	),
+	TP_printk(
+		CQ_PRN,
+		__get_str(dev),
+		__entry->idx,
+		__entry->wr_id,
+		__entry->status,
+		__entry->opcode, show_wc_opcode(__entry->opcode),
+		__entry->length,
+		__entry->qpn
+	)
+);
+
+DEFINE_EVENT(
+	rvt_cq_entry_template, rvt_cq_enter,
+	TP_PROTO(struct rvt_cq *cq, struct ib_wc *wc, u32 idx),
+	TP_ARGS(cq, wc, idx));
+
+DEFINE_EVENT(
+	rvt_cq_entry_template, rvt_cq_poll,
+	TP_PROTO(struct rvt_cq *cq, struct ib_wc *wc, u32 idx),
+	TP_ARGS(cq, wc, idx));
+
+#endif /* __RVT_TRACE_CQ_H */
+
+#undef TRACE_INCLUDE_PATH
+#undef TRACE_INCLUDE_FILE
+#define TRACE_INCLUDE_PATH .
+#define TRACE_INCLUDE_FILE trace_cq
+#include <trace/define_trace.h>

drivers/infiniband/sw/rdmavt/trace_rc.h

+/*
+ * Copyright(c) 2017 Intel Corporation.
+ *
+ * This file is provided under a dual BSD/GPLv2 license.  When using or
+ * redistributing this file, you may do so under either license.
+ *
+ * GPL LICENSE SUMMARY
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of version 2 of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful, but
+ * WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+ * General Public License for more details.
+ *
+ * BSD LICENSE
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions
+ * are met:
+ *
+ *  - Redistributions of source code must retain the above copyright
+ *    notice, this list of conditions and the following disclaimer.
+ *  - Redistributions in binary form must reproduce the above copyright
+ *    notice, this list of conditions and the following disclaimer in
+ *    the documentation and/or other materials provided with the
+ *    distribution.
+ *  - Neither the name of Intel Corporation nor the names of its
+ *    contributors may be used to endorse or promote products derived
+ *    from this software without specific prior written permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+ * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+ * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+ * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+ * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ *
+ */
+#if !defined(__RVT_TRACE_RC_H) || defined(TRACE_HEADER_MULTI_READ)
+#define __RVT_TRACE_RC_H
+
+#include <linux/tracepoint.h>
+#include <linux/trace_seq.h>
+
+#include <rdma/ib_verbs.h>
+#include <rdma/rdma_vt.h>
+
+#undef TRACE_SYSTEM
+#define TRACE_SYSTEM rvt_rc
+
+DECLARE_EVENT_CLASS(rvt_rc_template,
+		    TP_PROTO(struct rvt_qp *qp, u32 psn),
+		    TP_ARGS(qp, psn),
+		    TP_STRUCT__entry(
+			RDI_DEV_ENTRY(ib_to_rvt(qp->ibqp.device))
+			__field(u32, qpn)
+			__field(u32, s_flags)
+			__field(u32, psn)
+			__field(u32, s_psn)
+			__field(u32, s_next_psn)
+			__field(u32, s_sending_psn)
+			__field(u32, s_sending_hpsn)
+			__field(u32, r_psn)
+			),
+		    TP_fast_assign(
+			RDI_DEV_ASSIGN(ib_to_rvt(qp->ibqp.device))
+			__entry->qpn = qp->ibqp.qp_num;
+			__entry->s_flags = qp->s_flags;
+			__entry->psn = psn;
+			__entry->s_psn = qp->s_psn;
+			__entry->s_next_psn = qp->s_next_psn;
+			__entry->s_sending_psn = qp->s_sending_psn;
+			__entry->s_sending_hpsn = qp->s_sending_hpsn;
+			__entry->r_psn = qp->r_psn;
+			),
+		    TP_printk(
+			"[%s] qpn 0x%x s_flags 0x%x psn 0x%x s_psn 0x%x s_next_psn 0x%x s_sending_psn 0x%x sending_hpsn 0x%x r_psn 0x%x",
+			__get_str(dev),
+			__entry->qpn,
+			__entry->s_flags,
+			__entry->psn,
+			__entry->s_psn,
+			__entry->s_next_psn,
+			__entry->s_sending_psn,
+			__entry->s_sending_hpsn,
+			__entry->r_psn
+			)
+);
+
+DEFINE_EVENT(rvt_rc_template, rvt_rc_timeout,
+	     TP_PROTO(struct rvt_qp *qp, u32 psn),
+	     TP_ARGS(qp, psn)
+);
+
+#endif /* __RVT_TRACE_RC_H */
+
+#undef TRACE_INCLUDE_PATH
+#undef TRACE_INCLUDE_FILE
+#define TRACE_INCLUDE_PATH .
+#define TRACE_INCLUDE_FILE trace_rc
+#include <trace/define_trace.h>

drivers/infiniband/sw/rdmavt/trace_tx.h

@@ -71,10 +71,20 @@ __print_symbolic(opcode,                                   \
 	wr_opcode_name(RDMA_READ_WITH_INV),                \
 	wr_opcode_name(LOCAL_INV),                         \
 	wr_opcode_name(MASKED_ATOMIC_CMP_AND_SWP),         \
-	wr_opcode_name(MASKED_ATOMIC_FETCH_AND_ADD))
+	wr_opcode_name(MASKED_ATOMIC_FETCH_AND_ADD),       \
+	wr_opcode_name(RESERVED1),                         \
+	wr_opcode_name(RESERVED2),                         \
+	wr_opcode_name(RESERVED3),                         \
+	wr_opcode_name(RESERVED4),                         \
+	wr_opcode_name(RESERVED5),                         \
+	wr_opcode_name(RESERVED6),                         \
+	wr_opcode_name(RESERVED7),                         \
+	wr_opcode_name(RESERVED8),                         \
+	wr_opcode_name(RESERVED9),                         \
+	wr_opcode_name(RESERVED10))
 
 #define POS_PRN \
-"[%s] wr_id %llx qpn %x psn 0x%x lpsn 0x%x length %u opcode 0x%.2x,%s size %u avail %u head %u last %u"
+"[%s] wqe %p wr_id %llx send_flags %x qpn %x qpt %u psn %x lpsn %x ssn %x length %u opcode 0x%.2x,%s size %u avail %u head %u last %u pid %u num_sge %u"
 
 TRACE_EVENT(
 	rvt_post_one_wr,
@@ -83,7 +93,9 @@ TRACE_EVENT(
 	TP_STRUCT__entry(
 		RDI_DEV_ENTRY(ib_to_rvt(qp->ibqp.device))
 		__field(u64, wr_id)
+		__field(struct rvt_swqe *, wqe)
 		__field(u32, qpn)
+		__field(u32, qpt)
 		__field(u32, psn)
 		__field(u32, lpsn)
 		__field(u32, length)
@@ -92,11 +104,17 @@ TRACE_EVENT(
 		__field(u32, avail)
 		__field(u32, head)
 		__field(u32, last)
+		__field(u32, ssn)
+		__field(int, send_flags)
+		__field(pid_t, pid)
+		__field(int, num_sge)
 	),
 	TP_fast_assign(
 		RDI_DEV_ASSIGN(ib_to_rvt(qp->ibqp.device))
+		__entry->wqe = wqe;
 		__entry->wr_id = wqe->wr.wr_id;
 		__entry->qpn = qp->ibqp.qp_num;
+		__entry->qpt = qp->ibqp.qp_type;
 		__entry->psn = wqe->psn;
 		__entry->lpsn = wqe->lpsn;
 		__entry->length = wqe->length;
@@ -105,20 +123,30 @@ TRACE_EVENT(
 		__entry->avail = qp->s_avail;
 		__entry->head = qp->s_head;
 		__entry->last = qp->s_last;
+		__entry->pid = qp->pid;
+		__entry->ssn = wqe->ssn;
+		__entry->send_flags = wqe->wr.send_flags;
+		__entry->num_sge = wqe->wr.num_sge;
 	),
 	TP_printk(
 		POS_PRN,
 		__get_str(dev),
+		__entry->wqe,
 		__entry->wr_id,
+		__entry->send_flags,
 		__entry->qpn,
+		__entry->qpt,
 		__entry->psn,
 		__entry->lpsn,
+		__entry->ssn,
 		__entry->length,
 		__entry->opcode, show_wr_opcode(__entry->opcode),
 		__entry->size,
 		__entry->avail,
 		__entry->head,
-		__entry->last
+		__entry->last,
+		__entry->pid,
+		__entry->num_sge
 	)
 );
 

drivers/infiniband/ulp/ipoib/ipoib_fs.c

@@ -281,8 +281,11 @@ void ipoib_delete_debug_files(struct net_device *dev)
 {
 	struct ipoib_dev_priv *priv = netdev_priv(dev);
 
+	WARN_ONCE(!priv->mcg_dentry, "null mcg debug file\n");
+	WARN_ONCE(!priv->path_dentry, "null path debug file\n");
 	debugfs_remove(priv->mcg_dentry);
 	debugfs_remove(priv->path_dentry);
+	priv->mcg_dentry = priv->path_dentry = NULL;
 }
 
 int ipoib_register_debugfs(void)

drivers/infiniband/ulp/ipoib/ipoib_main.c

@@ -108,6 +108,33 @@ static struct ib_client ipoib_client = {
 	.get_net_dev_by_params = ipoib_get_net_dev_by_params,
 };
 
+#ifdef CONFIG_INFINIBAND_IPOIB_DEBUG
+static int ipoib_netdev_event(struct notifier_block *this,
+			      unsigned long event, void *ptr)
+{
+	struct netdev_notifier_info *ni = ptr;
+	struct net_device *dev = ni->dev;
+
+	if (dev->netdev_ops->ndo_open != ipoib_open)
+		return NOTIFY_DONE;
+
+	switch (event) {
+	case NETDEV_REGISTER:
+		ipoib_create_debug_files(dev);
+		break;
+	case NETDEV_CHANGENAME:
+		ipoib_delete_debug_files(dev);
+		ipoib_create_debug_files(dev);
+		break;
+	case NETDEV_UNREGISTER:
+		ipoib_delete_debug_files(dev);
+		break;
+	}
+
+	return NOTIFY_DONE;
+}
+#endif
+
 int ipoib_open(struct net_device *dev)
 {
 	struct ipoib_dev_priv *priv = netdev_priv(dev);
@@ -1674,8 +1701,6 @@ void ipoib_dev_cleanup(struct net_device *dev)
 
 	ASSERT_RTNL();
 
-	ipoib_delete_debug_files(dev);
-
 	/* Delete any child interfaces first */
 	list_for_each_entry_safe(cpriv, tcpriv, &priv->child_intfs, list) {
 		/* Stop GC on child */
@@ -2090,8 +2115,6 @@ static struct net_device *ipoib_add_port(const char *format,
 		goto register_failed;
 	}
 
-	ipoib_create_debug_files(priv->dev);
-
 	if (ipoib_cm_add_mode_attr(priv->dev))
 		goto sysfs_failed;
 	if (ipoib_add_pkey_attr(priv->dev))
@@ -2106,7 +2129,6 @@ static struct net_device *ipoib_add_port(const char *format,
 	return priv->dev;
 
 sysfs_failed:
-	ipoib_delete_debug_files(priv->dev);
 	unregister_netdev(priv->dev);
 
 register_failed:
@@ -2191,6 +2213,12 @@ static void ipoib_remove_one(struct ib_device *device, void *client_data)
 	kfree(dev_list);
 }
 
+#ifdef CONFIG_INFINIBAND_IPOIB_DEBUG
+static struct notifier_block ipoib_netdev_notifier = {
+	.notifier_call = ipoib_netdev_event,
+};
+#endif
+
 static int __init ipoib_init_module(void)
 {
 	int ret;
@@ -2243,6 +2271,9 @@ static int __init ipoib_init_module(void)
 	if (ret)
 		goto err_client;
 
+#ifdef CONFIG_INFINIBAND_IPOIB_DEBUG
+	register_netdevice_notifier(&ipoib_netdev_notifier);
+#endif
 	return 0;
 
 err_client:
@@ -2260,6 +2291,9 @@ static int __init ipoib_init_module(void)
 
 static void __exit ipoib_cleanup_module(void)
 {
+#ifdef CONFIG_INFINIBAND_IPOIB_DEBUG
+	unregister_netdevice_notifier(&ipoib_netdev_notifier);
+#endif
 	ipoib_netlink_fini();
 	ib_unregister_client(&ipoib_client);
 	ib_sa_unregister_client(&ipoib_sa_client);

drivers/infiniband/ulp/ipoib/ipoib_vlan.c

@@ -86,8 +86,6 @@ int __ipoib_vlan_add(struct ipoib_dev_priv *ppriv, struct ipoib_dev_priv *priv,
 		goto register_failed;
 	}
 
-	ipoib_create_debug_files(priv->dev);
-
 	/* RTNL childs don't need proprietary sysfs entries */
 	if (type == IPOIB_LEGACY_CHILD) {
 		if (ipoib_cm_add_mode_attr(priv->dev))
@@ -108,7 +106,6 @@ int __ipoib_vlan_add(struct ipoib_dev_priv *ppriv, struct ipoib_dev_priv *priv,
 
 sysfs_failed:
 	result = -ENOMEM;
-	ipoib_delete_debug_files(priv->dev);
 	unregister_netdevice(priv->dev);
 
 register_failed:

include/rdma/ib_pack.h

@@ -80,6 +80,8 @@ enum {
 	IB_OPCODE_UD                                = 0x60,
 	/* per IBTA 1.3 vol 1 Table 38, A10.3.2 */
 	IB_OPCODE_CNP                               = 0x80,
+	/* Manufacturer specific */
+	IB_OPCODE_MSP                               = 0xe0,
 
 	/* operations -- just used to define real constants */
 	IB_OPCODE_SEND_FIRST                        = 0x00,

include/rdma/ib_verbs.h

@@ -1357,6 +1357,17 @@ struct ib_fmr_attr {
 
 struct ib_umem;
 
+enum rdma_remove_reason {
+	/* Userspace requested uobject deletion. Call could fail */
+	RDMA_REMOVE_DESTROY,
+	/* Context deletion. This call should delete the actual object itself */
+	RDMA_REMOVE_CLOSE,
+	/* Driver is being hot-unplugged. This call should delete the actual object itself */
+	RDMA_REMOVE_DRIVER_REMOVE,
+	/* Context is being cleaned-up, but commit was just completed */
+	RDMA_REMOVE_DURING_CLEANUP,
+};
+
 struct ib_rdmacg_object {
 #ifdef CONFIG_CGROUP_RDMA
 	struct rdma_cgroup	*cg;		/* owner rdma cgroup */
@@ -1365,19 +1376,16 @@ struct ib_rdmacg_object {
 
 struct ib_ucontext {
 	struct ib_device       *device;
-	struct list_head	pd_list;
-	struct list_head	mr_list;
-	struct list_head	mw_list;
-	struct list_head	cq_list;
-	struct list_head	qp_list;
-	struct list_head	srq_list;
-	struct list_head	ah_list;
-	struct list_head	xrcd_list;
-	struct list_head	rule_list;
-	struct list_head	wq_list;
-	struct list_head	rwq_ind_tbl_list;
+	struct ib_uverbs_file  *ufile;
 	int			closing;
 
+	/* locking the uobjects_list */
+	struct mutex		uobjects_lock;
+	struct list_head	uobjects;
+	/* protects cleanup process from other actions */
+	struct rw_semaphore	cleanup_rwsem;
+	enum rdma_remove_reason cleanup_reason;
+
 	struct pid             *tgid;
 #ifdef CONFIG_INFINIBAND_ON_DEMAND_PAGING
 	struct rb_root      umem_tree;
@@ -1407,9 +1415,16 @@ struct ib_uobject {
 	struct ib_rdmacg_object	cg_obj;		/* rdmacg object */
 	int			id;		/* index into kernel idr */
 	struct kref		ref;
-	struct rw_semaphore	mutex;		/* protects .live */
+	atomic_t		usecnt;		/* protects exclusive access */
 	struct rcu_head		rcu;		/* kfree_rcu() overhead */
-	int			live;
+
+	const struct uverbs_obj_type *type;
+};
+
+struct ib_uobject_file {
+	struct ib_uobject	uobj;
+	/* ufile contains the lock between context release and file close */
+	struct ib_uverbs_file	*ufile;
 };
 
 struct ib_udata {

include/rdma/rdmavt_qp.h

@@ -2,7 +2,7 @@
 #define DEF_RDMAVT_INCQP_H
 
 /*
- * Copyright(c) 2016 Intel Corporation.
+ * Copyright(c) 2016, 2017 Intel Corporation.
  *
  * This file is provided under a dual BSD/GPLv2 license.  When using or
  * redistributing this file, you may do so under either license.
@@ -526,7 +526,6 @@ static inline void rvt_qp_wqe_reserve(
 	struct rvt_qp *qp,
 	struct rvt_swqe *wqe)
 {
-	wqe->wr.send_flags |= RVT_SEND_RESERVE_USED;
 	atomic_inc(&qp->s_reserved_used);
 }
 
@@ -550,7 +549,6 @@ static inline void rvt_qp_wqe_unreserve(
 	struct rvt_swqe *wqe)
 {
 	if (unlikely(wqe->wr.send_flags & RVT_SEND_RESERVE_USED)) {
-		wqe->wr.send_flags &= ~RVT_SEND_RESERVE_USED;
 		atomic_dec(&qp->s_reserved_used);
 		/* insure no compiler re-order up to s_last change */
 		smp_mb__after_atomic();
@@ -574,6 +572,7 @@ extern const enum ib_wc_opcode ib_rvt_wc_opcode[];
 static inline void rvt_qp_swqe_complete(
 	struct rvt_qp *qp,
 	struct rvt_swqe *wqe,
+	enum ib_wc_opcode opcode,
 	enum ib_wc_status status)
 {
 	if (unlikely(wqe->wr.send_flags & RVT_SEND_RESERVE_USED))
@@ -586,7 +585,7 @@ static inline void rvt_qp_swqe_complete(
 		memset(&wc, 0, sizeof(wc));
 		wc.wr_id = wqe->wr.wr_id;
 		wc.status = status;
-		wc.opcode = ib_rvt_wc_opcode[wqe->wr.opcode];
+		wc.opcode = opcode;
 		wc.qp = &qp->ibqp;
 		wc.byte_len = wqe->length;
 		rvt_cq_enter(ibcq_to_rvtcq(qp->ibqp.send_cq), &wc,

include/rdma/uverbs_std_types.h

+/*
+ * Copyright (c) 2017, Mellanox Technologies inc.  All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#ifndef _UVERBS_STD_TYPES__
+#define _UVERBS_STD_TYPES__
+
+#include <rdma/uverbs_types.h>
+
+extern const struct uverbs_obj_fd_type uverbs_type_attrs_comp_channel;
+extern const struct uverbs_obj_idr_type uverbs_type_attrs_cq;
+extern const struct uverbs_obj_idr_type uverbs_type_attrs_qp;
+extern const struct uverbs_obj_idr_type uverbs_type_attrs_rwq_ind_table;
+extern const struct uverbs_obj_idr_type uverbs_type_attrs_wq;
+extern const struct uverbs_obj_idr_type uverbs_type_attrs_srq;
+extern const struct uverbs_obj_idr_type uverbs_type_attrs_ah;
+extern const struct uverbs_obj_idr_type uverbs_type_attrs_flow;
+extern const struct uverbs_obj_idr_type uverbs_type_attrs_mr;
+extern const struct uverbs_obj_idr_type uverbs_type_attrs_mw;
+extern const struct uverbs_obj_idr_type uverbs_type_attrs_pd;
+extern const struct uverbs_obj_idr_type uverbs_type_attrs_xrcd;
+
+static inline struct ib_uobject *__uobj_get(const struct uverbs_obj_type *type,
+					    bool write,
+					    struct ib_ucontext *ucontext,
+					    int id)
+{
+	return rdma_lookup_get_uobject(type, ucontext, id, write);
+}
+
+#define uobj_get_type(_type) uverbs_type_attrs_##_type.type
+
+#define uobj_get_read(_type, _id, _ucontext)				\
+	 __uobj_get(&(_type), false, _ucontext, _id)
+
+#define uobj_get_obj_read(_type, _id, _ucontext)			\
+({									\
+	struct ib_uobject *uobj =					\
+		__uobj_get(&uobj_get_type(_type),			\
+			   false, _ucontext, _id);			\
+									\
+	(struct ib_##_type *)(IS_ERR(uobj) ? NULL : uobj->object);	\
+})
+
+#define uobj_get_write(_type, _id, _ucontext)				\
+	 __uobj_get(&(_type), true, _ucontext, _id)
+
+static inline void uobj_put_read(struct ib_uobject *uobj)
+{
+	rdma_lookup_put_uobject(uobj, false);
+}
+
+#define uobj_put_obj_read(_obj)					\
+	uobj_put_read((_obj)->uobject)
+
+static inline void uobj_put_write(struct ib_uobject *uobj)
+{
+	rdma_lookup_put_uobject(uobj, true);
+}
+
+static inline int __must_check uobj_remove_commit(struct ib_uobject *uobj)
+{
+	return rdma_remove_commit_uobject(uobj);
+}
+
+static inline void uobj_alloc_commit(struct ib_uobject *uobj)
+{
+	rdma_alloc_commit_uobject(uobj);
+}
+
+static inline void uobj_alloc_abort(struct ib_uobject *uobj)
+{
+	rdma_alloc_abort_uobject(uobj);
+}
+
+static inline struct ib_uobject *__uobj_alloc(const struct uverbs_obj_type *type,
+					      struct ib_ucontext *ucontext)
+{
+	return rdma_alloc_begin_uobject(type, ucontext);
+}
+
+#define uobj_alloc(_type, ucontext)	\
+	__uobj_alloc(&(_type), ucontext)
+
+#endif
+

include/rdma/uverbs_types.h

+/*
+ * Copyright (c) 2017, Mellanox Technologies inc.  All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#ifndef _UVERBS_TYPES_
+#define _UVERBS_TYPES_
+
+#include <linux/kernel.h>
+#include <rdma/ib_verbs.h>
+
+struct uverbs_obj_type;
+
+struct uverbs_obj_type_class {
+	/*
+	 * Get an ib_uobject that corresponds to the given id from ucontext,
+	 * These functions could create or destroy objects if required.
+	 * The action will be finalized only when commit, abort or put fops are
+	 * called.
+	 * The flow of the different actions is:
+	 * [alloc]:	 Starts with alloc_begin. The handlers logic is than
+	 *		 executed. If the handler is successful, alloc_commit
+	 *		 is called and the object is inserted to the repository.
+	 *		 Once alloc_commit completes the object is visible to
+	 *		 other threads and userspace.
+	 e		 Otherwise, alloc_abort is called and the object is
+	 *		 destroyed.
+	 * [lookup]:	 Starts with lookup_get which fetches and locks the
+	 *		 object. After the handler finished using the object, it
+	 *		 needs to call lookup_put to unlock it. The exclusive
+	 *		 flag indicates if the object is locked for exclusive
+	 *		 access.
+	 * [remove]:	 Starts with lookup_get with exclusive flag set. This
+	 *		 locks the object for exclusive access. If the handler
+	 *		 code completed successfully, remove_commit is called
+	 *		 and the ib_uobject is removed from the context's
+	 *		 uobjects repository and put. The object itself is
+	 *		 destroyed as well. Once remove succeeds new krefs to
+	 *		 the object cannot be acquired by other threads or
+	 *		 userspace and the hardware driver is removed from the
+	 *		 object. Other krefs on the object may still exist.
+	 *		 If the handler code failed, lookup_put should be
+	 *		 called. This callback is used when the context
+	 *		 is destroyed as well (process termination,
+	 *		 reset flow).
+	 */
+	struct ib_uobject *(*alloc_begin)(const struct uverbs_obj_type *type,
+					  struct ib_ucontext *ucontext);
+	void (*alloc_commit)(struct ib_uobject *uobj);
+	void (*alloc_abort)(struct ib_uobject *uobj);
+
+	struct ib_uobject *(*lookup_get)(const struct uverbs_obj_type *type,
+					 struct ib_ucontext *ucontext, int id,
+					 bool exclusive);
+	void (*lookup_put)(struct ib_uobject *uobj, bool exclusive);
+	/*
+	 * Must be called with the exclusive lock held. If successful uobj is
+	 * invalid on return. On failure uobject is left completely
+	 * unchanged
+	 */
+	int __must_check (*remove_commit)(struct ib_uobject *uobj,
+					  enum rdma_remove_reason why);
+	u8    needs_kfree_rcu;
+};
+
+struct uverbs_obj_type {
+	const struct uverbs_obj_type_class * const type_class;
+	size_t	     obj_size;
+	unsigned int destroy_order;
+};
+
+/*
+ * Objects type classes which support a detach state (object is still alive but
+ * it's not attached to any context need to make sure:
+ * (a) no call through to a driver after a detach is called
+ * (b) detach isn't called concurrently with context_cleanup
+ */
+
+struct uverbs_obj_idr_type {
+	/*
+	 * In idr based objects, uverbs_obj_type_class points to a generic
+	 * idr operations. In order to specialize the underlying types (e.g. CQ,
+	 * QPs, etc.), we add destroy_object specific callbacks.
+	 */
+	struct uverbs_obj_type  type;
+
+	/* Free driver resources from the uobject, make the driver uncallable,
+	 * and move the uobject to the detached state. If the object was
+	 * destroyed by the user's request, a failure should leave the uobject
+	 * completely unchanged.
+	 */
+	int __must_check (*destroy_object)(struct ib_uobject *uobj,
+					   enum rdma_remove_reason why);
+};
+
+struct ib_uobject *rdma_lookup_get_uobject(const struct uverbs_obj_type *type,
+					   struct ib_ucontext *ucontext,
+					   int id, bool exclusive);
+void rdma_lookup_put_uobject(struct ib_uobject *uobj, bool exclusive);
+struct ib_uobject *rdma_alloc_begin_uobject(const struct uverbs_obj_type *type,
+					    struct ib_ucontext *ucontext);
+void rdma_alloc_abort_uobject(struct ib_uobject *uobj);
+int __must_check rdma_remove_commit_uobject(struct ib_uobject *uobj);
+int rdma_alloc_commit_uobject(struct ib_uobject *uobj);
+
+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.
+	 */
+	struct uverbs_obj_type  type;
+	int (*context_closed)(struct ib_uobject_file *uobj_file,
+			      enum rdma_remove_reason why);
+	const struct file_operations	*fops;
+	const char			*name;
+	int				flags;
+};
+
+extern const struct uverbs_obj_type_class uverbs_idr_class;
+extern const struct uverbs_obj_type_class uverbs_fd_class;
+
+#define UVERBS_BUILD_BUG_ON(cond) (sizeof(char[1 - 2 * !!(cond)]) -	\
+				   sizeof(char))
+#define UVERBS_TYPE_ALLOC_FD(_size, _order)				 \
+	{								 \
+		.destroy_order = _order,				 \
+		.type_class = &uverbs_fd_class,				 \
+		.obj_size = (_size) +					 \
+			  UVERBS_BUILD_BUG_ON((_size) <			 \
+					      sizeof(struct ib_uobject_file)),\
+	}
+#define UVERBS_TYPE_ALLOC_IDR_SZ(_size, _order)				\
+	{								\
+		.destroy_order = _order,				\
+		.type_class = &uverbs_idr_class,			\
+		.obj_size = (_size) +					\
+			  UVERBS_BUILD_BUG_ON((_size) <			\
+					      sizeof(struct ib_uobject)), \
+	}
+#define UVERBS_TYPE_ALLOC_IDR(_order)					\
+	 UVERBS_TYPE_ALLOC_IDR_SZ(sizeof(struct ib_uobject), _order)
+#endif