path: root/drivers/infiniband/core/device.c

/*
 * Copyright (c) 2004 Topspin Communications.  All rights reserved.
 * Copyright (c) 2005 Sun Microsystems, 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/module.h>
#include <linux/string.h>
#include <linux/errno.h>
#include <linux/kernel.h>
#include <linux/slab.h>
#include <linux/init.h>
#include <linux/netdevice.h>
#include <linux/security.h>
#include <linux/notifier.h>
#include <rdma/rdma_netlink.h>
#include <rdma/ib_addr.h>
#include <rdma/ib_cache.h>

#include "core_priv.h"
#include "restrack.h"

MODULE_AUTHOR("Roland Dreier");
MODULE_DESCRIPTION("core kernel InfiniBand API");
MODULE_LICENSE("Dual BSD/GPL");

struct workqueue_struct *ib_comp_wq;
struct workqueue_struct *ib_comp_unbound_wq;
struct workqueue_struct *ib_wq;
EXPORT_SYMBOL_GPL(ib_wq);

/*
 * Each of the three rwsem locks (devices, clients, client_data) protects the
 * xarray of the same name. Specifically it allows the caller to assert that
 * the MARK will/will not be changing under the lock, and for devices and
 * clients, that the value in the xarray is still a valid pointer. Change of
 * the MARK is linked to the object state, so holding the lock and testing the
 * MARK also asserts that the contained object is in a certain state.
 *
 * This is used to build a two stage register/unregister flow where objects
 * can continue to be in the xarray even though they are still in progress to
 * register/unregister.
 *
 * The xarray itself provides additional locking, and restartable iteration,
 * which is also relied on.
 *
 * Locks should not be nested, with the exception of client_data, which is
 * allowed to nest under the read side of the other two locks.
 *
 * The devices_rwsem also protects the device name list, any change or
 * assignment of device name must also hold the write side to guarantee unique
 * names.
 */

/*
 * devices contains devices that have had their names assigned. The
 * devices may not be registered. Users that care about the registration
 * status need to call ib_device_try_get() on the device to ensure it is
 * registered, and keep it registered, for the required duration.
 *
 */
static DEFINE_XARRAY_FLAGS(devices, XA_FLAGS_ALLOC);
static DECLARE_RWSEM(devices_rwsem);
#define DEVICE_REGISTERED XA_MARK_1

static LIST_HEAD(client_list);
#define CLIENT_REGISTERED XA_MARK_1
static DEFINE_XARRAY_FLAGS(clients, XA_FLAGS_ALLOC);
static DECLARE_RWSEM(clients_rwsem);

/*
 * If client_data is registered then the corresponding client must also still
 * be registered.
 */
#define CLIENT_DATA_REGISTERED XA_MARK_1
/*
 * xarray has this behavior where it won't iterate over NULL values stored in
 * allocated arrays.  So we need our own iterator to see all values stored in
 * the array. This does the same thing as xa_for_each except that it also
 * returns NULL valued entries if the array is allocating. Simplified to only
 * work on simple xarrays.
 */
static void *xan_find_marked(struct xarray *xa, unsigned long *indexp,
			     xa_mark_t filter)
{
	XA_STATE(xas, xa, *indexp);
	void *entry;

	rcu_read_lock();
	do {
		entry = xas_find_marked(&xas, ULONG_MAX, filter);
		if (xa_is_zero(entry))
			break;
	} while (xas_retry(&xas, entry));
	rcu_read_unlock();

	if (entry) {
		*indexp = xas.xa_index;
		if (xa_is_zero(entry))
			return NULL;
		return entry;
	}
	return XA_ERROR(-ENOENT);
}
#define xan_for_each_marked(xa, index, entry, filter)                          \
	for (index = 0, entry = xan_find_marked(xa, &(index), filter);         \
	     !xa_is_err(entry);                                                \
	     (index)++, entry = xan_find_marked(xa, &(index), filter))

static int ib_security_change(struct notifier_block *nb, unsigned long event,
			      void *lsm_data);
static void ib_policy_change_task(struct work_struct *work);
static DECLARE_WORK(ib_policy_change_work, ib_policy_change_task);

static struct notifier_block ibdev_lsm_nb = {
	.notifier_call = ib_security_change,
};

static int ib_device_check_mandatory(struct ib_device *device)
{
#define IB_MANDATORY_FUNC(x) { offsetof(struct ib_device_ops, x), #x }
	static const struct {
		size_t offset;
		char  *name;
	} mandatory_table[] = {
		IB_MANDATORY_FUNC(query_device),
		IB_MANDATORY_FUNC(query_port),
		IB_MANDATORY_FUNC(query_pkey),
		IB_MANDATORY_FUNC(alloc_pd),
		IB_MANDATORY_FUNC(dealloc_pd),
		IB_MANDATORY_FUNC(create_qp),
		IB_MANDATORY_FUNC(modify_qp),
		IB_MANDATORY_FUNC(destroy_qp),
		IB_MANDATORY_FUNC(post_send),
		IB_MANDATORY_FUNC(post_recv),
		IB_MANDATORY_FUNC(create_cq),
		IB_MANDATORY_FUNC(destroy_cq),
		IB_MANDATORY_FUNC(poll_cq),
		IB_MANDATORY_FUNC(req_notify_cq),
		IB_MANDATORY_FUNC(get_dma_mr),
		IB_MANDATORY_FUNC(dereg_mr),
		IB_MANDATORY_FUNC(get_port_immutable)
	};
	int i;

	device->kverbs_provider = true;
	for (i = 0; i < ARRAY_SIZE(mandatory_table); ++i) {
		if (!*(void **) ((void *) &device->ops +
				 mandatory_table[i].offset)) {
			device->kverbs_provider = false;
			break;
		}
	}

	return 0;
}

/*
 * Caller must perform ib_device_put() to return the device reference count
 * when ib_device_get_by_index() returns valid device pointer.
 */
struct ib_device *ib_device_get_by_index(u32 index)
{
	struct ib_device *device;

	down_read(&devices_rwsem);
	device = xa_load(&devices, index);
	if (device) {
		if (!ib_device_try_get(device))
			device = NULL;
	}
	up_read(&devices_rwsem);
	return device;
}

/**
 * ib_device_put - Release IB device reference
 * @device: device whose reference to be released
 *
 * ib_device_put() releases reference to the IB device to allow it to be
 * unregistered and eventually free.
 */
void ib_device_put(struct ib_device *device)
{
	if (refcount_dec_and_test(&device->refcount))
		complete(&device->unreg_completion);
}
EXPORT_SYMBOL(ib_device_put);

static struct ib_device *__ib_device_get_by_name(const char *name)
{
	struct ib_device *d