device.c 32.5 KB
Newer Older
L
Linus Torvalds 已提交
1 2
/*
 * Copyright (c) 2004 Topspin Communications.  All rights reserved.
3
 * Copyright (c) 2005 Sun Microsystems, Inc. All rights reserved.
L
Linus Torvalds 已提交
4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36
 *
 * 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>
37
#include <linux/kernel.h>
L
Linus Torvalds 已提交
38 39
#include <linux/slab.h>
#include <linux/init.h>
40
#include <linux/mutex.h>
41
#include <linux/netdevice.h>
42 43
#include <linux/security.h>
#include <linux/notifier.h>
R
Roland Dreier 已提交
44
#include <rdma/rdma_netlink.h>
45 46
#include <rdma/ib_addr.h>
#include <rdma/ib_cache.h>
L
Linus Torvalds 已提交
47 48 49 50 51 52 53 54 55 56 57

#include "core_priv.h"

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

struct ib_client_data {
	struct list_head  list;
	struct ib_client *client;
	void *            data;
58 59 60
	/* The device or client is going down. Do not call client or device
	 * callbacks other than remove(). */
	bool		  going_down;
L
Linus Torvalds 已提交
61 62
};

63
struct workqueue_struct *ib_comp_wq;
T
Tejun Heo 已提交
64 65 66
struct workqueue_struct *ib_wq;
EXPORT_SYMBOL_GPL(ib_wq);

67 68 69
/* The device_list and client_list contain devices and clients after their
 * registration has completed, and the devices and clients are removed
 * during unregistration. */
L
Linus Torvalds 已提交
70 71 72 73
static LIST_HEAD(device_list);
static LIST_HEAD(client_list);

/*
74 75 76 77 78 79 80
 * device_mutex and lists_rwsem protect access to both device_list and
 * client_list.  device_mutex protects writer access by device and client
 * registration / de-registration.  lists_rwsem protects reader access to
 * these lists.  Iterators of these lists must lock it for read, while updates
 * to the lists must be done with a write lock. A special case is when the
 * device_mutex is locked. In this case locking the lists for read access is
 * not necessary as the device_mutex implies it.
81 82
 *
 * lists_rwsem also protects access to the client data list.
L
Linus Torvalds 已提交
83
 */
84
static DEFINE_MUTEX(device_mutex);
85 86
static DECLARE_RWSEM(lists_rwsem);

87 88 89 90 91 92 93 94
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,
};
L
Linus Torvalds 已提交
95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120

static int ib_device_check_mandatory(struct ib_device *device)
{
#define IB_MANDATORY_FUNC(x) { offsetof(struct ib_device, 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(query_gid),
		IB_MANDATORY_FUNC(alloc_pd),
		IB_MANDATORY_FUNC(dealloc_pd),
		IB_MANDATORY_FUNC(create_ah),
		IB_MANDATORY_FUNC(destroy_ah),
		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),
121 122
		IB_MANDATORY_FUNC(dereg_mr),
		IB_MANDATORY_FUNC(get_port_immutable)
L
Linus Torvalds 已提交
123 124 125
	};
	int i;

126
	for (i = 0; i < ARRAY_SIZE(mandatory_table); ++i) {
L
Linus Torvalds 已提交
127
		if (!*(void **) ((void *) device + mandatory_table[i].offset)) {
P
Parav Pandit 已提交
128 129
			pr_warn("Device %s is missing mandatory function %s\n",
				device->name, mandatory_table[i].name);
L
Linus Torvalds 已提交
130 131 132 133 134 135 136
			return -EINVAL;
		}
	}

	return 0;
}

137
static struct ib_device *__ib_device_get_by_index(u32 index)
138 139 140 141 142 143 144 145 146 147
{
	struct ib_device *device;

	list_for_each_entry(device, &device_list, core_list)
		if (device->index == index)
			return device;

	return NULL;
}

148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163
/*
 * Caller is responsible to return refrerence count by calling put_device()
 */
struct ib_device *ib_device_get_by_index(u32 index)
{
	struct ib_device *device;

	down_read(&lists_rwsem);
	device = __ib_device_get_by_index(index);
	if (device)
		get_device(&device->dev);

	up_read(&lists_rwsem);
	return device;
}

L
Linus Torvalds 已提交
164 165 166 167 168 169 170 171 172 173 174 175 176
static struct ib_device *__ib_device_get_by_name(const char *name)
{
	struct ib_device *device;

	list_for_each_entry(device, &device_list, core_list)
		if (!strncmp(name, device->name, IB_DEVICE_NAME_MAX))
			return device;

	return NULL;
}

static int alloc_name(char *name)
{
177
	unsigned long *inuse;
L
Linus Torvalds 已提交
178 179 180 181
	char buf[IB_DEVICE_NAME_MAX];
	struct ib_device *device;
	int i;

182
	inuse = (unsigned long *) get_zeroed_page(GFP_KERNEL);
L
Linus Torvalds 已提交
183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206
	if (!inuse)
		return -ENOMEM;

	list_for_each_entry(device, &device_list, core_list) {
		if (!sscanf(device->name, name, &i))
			continue;
		if (i < 0 || i >= PAGE_SIZE * 8)
			continue;
		snprintf(buf, sizeof buf, name, i);
		if (!strncmp(buf, device->name, IB_DEVICE_NAME_MAX))
			set_bit(i, inuse);
	}

	i = find_first_zero_bit(inuse, PAGE_SIZE * 8);
	free_page((unsigned long) inuse);
	snprintf(buf, sizeof buf, name, i);

	if (__ib_device_get_by_name(buf))
		return -ENFILE;

	strlcpy(name, buf, IB_DEVICE_NAME_MAX);
	return 0;
}

207 208 209 210
static void ib_device_release(struct device *device)
{
	struct ib_device *dev = container_of(device, struct ib_device, dev);

211 212 213 214 215 216 217 218 219 220
	WARN_ON(dev->reg_state == IB_DEV_REGISTERED);
	if (dev->reg_state == IB_DEV_UNREGISTERED) {
		/*
		 * In IB_DEV_UNINITIALIZED state, cache or port table
		 * is not even created. Free cache and port table only when
		 * device reaches UNREGISTERED state.
		 */
		ib_cache_release_one(dev);
		kfree(dev->port_immutable);
	}
221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244
	kfree(dev);
}

static int ib_device_uevent(struct device *device,
			    struct kobj_uevent_env *env)
{
	struct ib_device *dev = container_of(device, struct ib_device, dev);

	if (add_uevent_var(env, "NAME=%s", dev->name))
		return -ENOMEM;

	/*
	 * It would be nice to pass the node GUID with the event...
	 */

	return 0;
}

static struct class ib_class = {
	.name    = "infiniband",
	.dev_release = ib_device_release,
	.dev_uevent = ib_device_uevent,
};

L
Linus Torvalds 已提交
245 246 247 248 249 250 251 252 253 254 255 256
/**
 * ib_alloc_device - allocate an IB device struct
 * @size:size of structure to allocate
 *
 * Low-level drivers should use ib_alloc_device() to allocate &struct
 * ib_device.  @size is the size of the structure to be allocated,
 * including any private data used by the low-level driver.
 * ib_dealloc_device() must be used to free structures allocated with
 * ib_alloc_device().
 */
struct ib_device *ib_alloc_device(size_t size)
{
257 258 259 260 261 262 263 264 265
	struct ib_device *device;

	if (WARN_ON(size < sizeof(struct ib_device)))
		return NULL;

	device = kzalloc(size, GFP_KERNEL);
	if (!device)
		return NULL;

266 267
	rdma_restrack_init(&device->res);

268 269 270 271 272 273 274 275 276 277
	device->dev.class = &ib_class;
	device_initialize(&device->dev);

	dev_set_drvdata(&device->dev, device);

	INIT_LIST_HEAD(&device->event_handler_list);
	spin_lock_init(&device->event_handler_lock);
	spin_lock_init(&device->client_data_lock);
	INIT_LIST_HEAD(&device->client_data_list);
	INIT_LIST_HEAD(&device->port_list);
L
Linus Torvalds 已提交
278

279
	return device;
L
Linus Torvalds 已提交
280 281 282 283 284 285 286 287 288 289 290
}
EXPORT_SYMBOL(ib_alloc_device);

/**
 * ib_dealloc_device - free an IB device struct
 * @device:structure to free
 *
 * Free a structure allocated with ib_alloc_device().
 */
void ib_dealloc_device(struct ib_device *device)
{
291 292
	WARN_ON(device->reg_state != IB_DEV_UNREGISTERED &&
		device->reg_state != IB_DEV_UNINITIALIZED);
293
	put_device(&device->dev);
L
Linus Torvalds 已提交
294 295 296 297 298 299 300 301 302
}
EXPORT_SYMBOL(ib_dealloc_device);

static int add_client_context(struct ib_device *device, struct ib_client *client)
{
	struct ib_client_data *context;
	unsigned long flags;

	context = kmalloc(sizeof *context, GFP_KERNEL);
303
	if (!context)
L
Linus Torvalds 已提交
304 305 306 307
		return -ENOMEM;

	context->client = client;
	context->data   = NULL;
308
	context->going_down = false;
L
Linus Torvalds 已提交
309

310
	down_write(&lists_rwsem);
L
Linus Torvalds 已提交
311 312 313
	spin_lock_irqsave(&device->client_data_lock, flags);
	list_add(&context->list, &device->client_data_list);
	spin_unlock_irqrestore(&device->client_data_lock, flags);
314
	up_write(&lists_rwsem);
L
Linus Torvalds 已提交
315 316 317 318

	return 0;
}

319 320 321 322 323 324
static int verify_immutable(const struct ib_device *dev, u8 port)
{
	return WARN_ON(!rdma_cap_ib_mad(dev, port) &&
			    rdma_max_mad_size(dev, port) != 0);
}

325
static int read_port_immutable(struct ib_device *device)
326
{
327
	int ret;
328 329 330 331 332 333 334 335 336 337 338 339 340 341 342
	u8 start_port = rdma_start_port(device);
	u8 end_port = rdma_end_port(device);
	u8 port;

	/**
	 * device->port_immutable is indexed directly by the port number to make
	 * access to this data as efficient as possible.
	 *
	 * Therefore port_immutable is declared as a 1 based array with
	 * potential empty slots at the beginning.
	 */
	device->port_immutable = kzalloc(sizeof(*device->port_immutable)
					 * (end_port + 1),
					 GFP_KERNEL);
	if (!device->port_immutable)
343
		return -ENOMEM;
344

345 346 347
	for (port = start_port; port <= end_port; ++port) {
		ret = device->get_port_immutable(device, port,
						 &device->port_immutable[port]);
348
		if (ret)
349
			return ret;
350

351 352
		if (verify_immutable(device, port))
			return -EINVAL;
353
	}
354
	return 0;
355 356
}

357
void ib_get_device_fw_str(struct ib_device *dev, char *str)
358 359
{
	if (dev->get_dev_fw_str)
360
		dev->get_dev_fw_str(dev, str);
361 362 363 364 365
	else
		str[0] = '\0';
}
EXPORT_SYMBOL(ib_get_device_fw_str);

366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389
static int setup_port_pkey_list(struct ib_device *device)
{
	int i;

	/**
	 * device->port_pkey_list is indexed directly by the port number,
	 * Therefore it is declared as a 1 based array with potential empty
	 * slots at the beginning.
	 */
	device->port_pkey_list = kcalloc(rdma_end_port(device) + 1,
					 sizeof(*device->port_pkey_list),
					 GFP_KERNEL);

	if (!device->port_pkey_list)
		return -ENOMEM;

	for (i = 0; i < (rdma_end_port(device) + 1); i++) {
		spin_lock_init(&device->port_pkey_list[i].list_lock);
		INIT_LIST_HEAD(&device->port_pkey_list[i].pkey_list);
	}

	return 0;
}

390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406
static void ib_policy_change_task(struct work_struct *work)
{
	struct ib_device *dev;

	down_read(&lists_rwsem);
	list_for_each_entry(dev, &device_list, core_list) {
		int i;

		for (i = rdma_start_port(dev); i <= rdma_end_port(dev); i++) {
			u64 sp;
			int ret = ib_get_cached_subnet_prefix(dev,
							      i,
							      &sp);

			WARN_ONCE(ret,
				  "ib_get_cached_subnet_prefix err: %d, this should never happen here\n",
				  ret);
407 408
			if (!ret)
				ib_security_cache_change(dev, i, sp);
409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424
		}
	}
	up_read(&lists_rwsem);
}

static int ib_security_change(struct notifier_block *nb, unsigned long event,
			      void *lsm_data)
{
	if (event != LSM_POLICY_CHANGE)
		return NOTIFY_DONE;

	schedule_work(&ib_policy_change_work);

	return NOTIFY_OK;
}

425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448
/**
 *	__dev_new_index	-	allocate an device index
 *
 *	Returns a suitable unique value for a new device interface
 *	number.  It assumes that there are less than 2^32-1 ib devices
 *	will be present in the system.
 */
static u32 __dev_new_index(void)
{
	/*
	 * The device index to allow stable naming.
	 * Similar to struct net -> ifindex.
	 */
	static u32 index;

	for (;;) {
		if (!(++index))
			index = 1;

		if (!__ib_device_get_by_index(index))
			return index;
	}
}

L
Linus Torvalds 已提交
449 450 451 452 453 454 455 456 457
/**
 * ib_register_device - Register an IB device with IB core
 * @device:Device to register
 *
 * Low-level drivers use ib_register_device() to register their
 * devices with the IB core.  All registered clients will receive a
 * callback for each device that is added. @device must be allocated
 * with ib_alloc_device().
 */
458 459 460
int ib_register_device(struct ib_device *device,
		       int (*port_callback)(struct ib_device *,
					    u8, struct kobject *))
L
Linus Torvalds 已提交
461 462
{
	int ret;
463
	struct ib_client *client;
464
	struct ib_udata uhw = {.outlen = 0, .inlen = 0};
465 466
	struct device *parent = device->dev.parent;

467 468 469 470 471 472 473 474
	WARN_ON_ONCE(device->dma_device);
	if (device->dev.dma_ops) {
		/*
		 * The caller provided custom DMA operations. Copy the
		 * DMA-related fields that are used by e.g. dma_alloc_coherent()
		 * into device->dev.
		 */
		device->dma_device = &device->dev;
475 476 477 478 479 480 481 482 483 484 485 486 487
		if (!device->dev.dma_mask) {
			if (parent)
				device->dev.dma_mask = parent->dma_mask;
			else
				WARN_ON_ONCE(true);
		}
		if (!device->dev.coherent_dma_mask) {
			if (parent)
				device->dev.coherent_dma_mask =
					parent->coherent_dma_mask;
			else
				WARN_ON_ONCE(true);
		}
488 489 490 491 492
	} else {
		/*
		 * The caller did not provide custom DMA operations. Use the
		 * DMA mapping operations of the parent device.
		 */
493
		WARN_ON_ONCE(!parent);
494 495
		device->dma_device = parent;
	}
L
Linus Torvalds 已提交
496

497
	mutex_lock(&device_mutex);
L
Linus Torvalds 已提交
498 499 500 501 502 503 504 505 506 507 508 509

	if (strchr(device->name, '%')) {
		ret = alloc_name(device->name);
		if (ret)
			goto out;
	}

	if (ib_device_check_mandatory(device)) {
		ret = -EINVAL;
		goto out;
	}

510
	ret = read_port_immutable(device);
511
	if (ret) {
P
Parav Pandit 已提交
512 513
		pr_warn("Couldn't create per port immutable data %s\n",
			device->name);
514 515 516
		goto out;
	}

517 518 519 520 521 522
	ret = setup_port_pkey_list(device);
	if (ret) {
		pr_warn("Couldn't create per port_pkey_list\n");
		goto out;
	}

523 524
	ret = ib_cache_setup_one(device);
	if (ret) {
P
Parav Pandit 已提交
525
		pr_warn("Couldn't set up InfiniBand P_Key/GID cache\n");
526
		goto port_cleanup;
527 528
	}

529 530 531
	ret = ib_device_register_rdmacg(device);
	if (ret) {
		pr_warn("Couldn't register device with rdma cgroup\n");
532
		goto cache_cleanup;
533 534
	}

535 536 537
	memset(&device->attrs, 0, sizeof(device->attrs));
	ret = device->query_device(device, &device->attrs, &uhw);
	if (ret) {
P
Parav Pandit 已提交
538
		pr_warn("Couldn't query the device attributes\n");
539
		goto cg_cleanup;
540 541
	}

542
	ret = ib_device_register_sysfs(device, port_callback);
L
Linus Torvalds 已提交
543
	if (ret) {
P
Parav Pandit 已提交
544 545
		pr_warn("Couldn't register device %s with driver model\n",
			device->name);
546
		goto cg_cleanup;
L
Linus Torvalds 已提交
547 548 549 550
	}

	device->reg_state = IB_DEV_REGISTERED;

551
	list_for_each_entry(client, &client_list, list)
552
		if (!add_client_context(device, client) && client->add)
553
			client->add(device);
L
Linus Torvalds 已提交
554

555
	device->index = __dev_new_index();
556 557 558
	down_write(&lists_rwsem);
	list_add_tail(&device->core_list, &device_list);
	up_write(&lists_rwsem);
559 560 561
	mutex_unlock(&device_mutex);
	return 0;

562 563
cg_cleanup:
	ib_device_unregister_rdmacg(device);
564 565 566 567 568
cache_cleanup:
	ib_cache_cleanup_one(device);
	ib_cache_release_one(device);
port_cleanup:
	kfree(device->port_immutable);
569
out:
570
	mutex_unlock(&device_mutex);
L
Linus Torvalds 已提交
571 572 573 574 575 576 577 578 579 580 581 582 583 584 585
	return ret;
}
EXPORT_SYMBOL(ib_register_device);

/**
 * ib_unregister_device - Unregister an IB device
 * @device:Device to unregister
 *
 * Unregister an IB device.  All clients will receive a remove callback.
 */
void ib_unregister_device(struct ib_device *device)
{
	struct ib_client_data *context, *tmp;
	unsigned long flags;

586
	mutex_lock(&device_mutex);
L
Linus Torvalds 已提交
587

588 589
	down_write(&lists_rwsem);
	list_del(&device->core_list);
590 591 592 593 594
	spin_lock_irqsave(&device->client_data_lock, flags);
	list_for_each_entry_safe(context, tmp, &device->client_data_list, list)
		context->going_down = true;
	spin_unlock_irqrestore(&device->client_data_lock, flags);
	downgrade_write(&lists_rwsem);
595

596 597 598 599 600 601
	list_for_each_entry_safe(context, tmp, &device->client_data_list,
				 list) {
		if (context->client->remove)
			context->client->remove(device, context->data);
	}
	up_read(&lists_rwsem);
L
Linus Torvalds 已提交
602

603 604
	rdma_restrack_clean(&device->res);

605
	ib_device_unregister_rdmacg(device);
606
	ib_device_unregister_sysfs(device);
607 608 609

	mutex_unlock(&device_mutex);

610
	ib_cache_cleanup_one(device);
611

612 613 614
	ib_security_destroy_port_pkey_list(device);
	kfree(device->port_pkey_list);

615
	down_write(&lists_rwsem);
L
Linus Torvalds 已提交
616 617 618 619
	spin_lock_irqsave(&device->client_data_lock, flags);
	list_for_each_entry_safe(context, tmp, &device->client_data_list, list)
		kfree(context);
	spin_unlock_irqrestore(&device->client_data_lock, flags);
620
	up_write(&lists_rwsem);
L
Linus Torvalds 已提交
621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642

	device->reg_state = IB_DEV_UNREGISTERED;
}
EXPORT_SYMBOL(ib_unregister_device);

/**
 * ib_register_client - Register an IB client
 * @client:Client to register
 *
 * Upper level users of the IB drivers can use ib_register_client() to
 * register callbacks for IB device addition and removal.  When an IB
 * device is added, each registered client's add method will be called
 * (in the order the clients were registered), and when a device is
 * removed, each client's remove method will be called (in the reverse
 * order that clients were registered).  In addition, when
 * ib_register_client() is called, the client will receive an add
 * callback for all devices already registered.
 */
int ib_register_client(struct ib_client *client)
{
	struct ib_device *device;

643
	mutex_lock(&device_mutex);
L
Linus Torvalds 已提交
644 645

	list_for_each_entry(device, &device_list, core_list)
646
		if (!add_client_context(device, client) && client->add)
L
Linus Torvalds 已提交
647 648
			client->add(device);

649 650 651 652
	down_write(&lists_rwsem);
	list_add_tail(&client->list, &client_list);
	up_write(&lists_rwsem);

653
	mutex_unlock(&device_mutex);
L
Linus Torvalds 已提交
654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672

	return 0;
}
EXPORT_SYMBOL(ib_register_client);

/**
 * ib_unregister_client - Unregister an IB client
 * @client:Client to unregister
 *
 * Upper level users use ib_unregister_client() to remove their client
 * registration.  When ib_unregister_client() is called, the client
 * will receive a remove callback for each IB device still registered.
 */
void ib_unregister_client(struct ib_client *client)
{
	struct ib_client_data *context, *tmp;
	struct ib_device *device;
	unsigned long flags;

673
	mutex_lock(&device_mutex);
L
Linus Torvalds 已提交
674

675 676 677 678
	down_write(&lists_rwsem);
	list_del(&client->list);
	up_write(&lists_rwsem);

L
Linus Torvalds 已提交
679
	list_for_each_entry(device, &device_list, core_list) {
680
		struct ib_client_data *found_context = NULL;
L
Linus Torvalds 已提交
681

682
		down_write(&lists_rwsem);
L
Linus Torvalds 已提交
683 684 685
		spin_lock_irqsave(&device->client_data_lock, flags);
		list_for_each_entry_safe(context, tmp, &device->client_data_list, list)
			if (context->client == client) {
686 687 688
				context->going_down = true;
				found_context = context;
				break;
L
Linus Torvalds 已提交
689 690
			}
		spin_unlock_irqrestore(&device->client_data_lock, flags);
691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708
		up_write(&lists_rwsem);

		if (client->remove)
			client->remove(device, found_context ?
					       found_context->data : NULL);

		if (!found_context) {
			pr_warn("No client context found for %s/%s\n",
				device->name, client->name);
			continue;
		}

		down_write(&lists_rwsem);
		spin_lock_irqsave(&device->client_data_lock, flags);
		list_del(&found_context->list);
		kfree(found_context);
		spin_unlock_irqrestore(&device->client_data_lock, flags);
		up_write(&lists_rwsem);
L
Linus Torvalds 已提交
709 710
	}

711
	mutex_unlock(&device_mutex);
L
Linus Torvalds 已提交
712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741
}
EXPORT_SYMBOL(ib_unregister_client);

/**
 * ib_get_client_data - Get IB client context
 * @device:Device to get context for
 * @client:Client to get context for
 *
 * ib_get_client_data() returns client context set with
 * ib_set_client_data().
 */
void *ib_get_client_data(struct ib_device *device, struct ib_client *client)
{
	struct ib_client_data *context;
	void *ret = NULL;
	unsigned long flags;

	spin_lock_irqsave(&device->client_data_lock, flags);
	list_for_each_entry(context, &device->client_data_list, list)
		if (context->client == client) {
			ret = context->data;
			break;
		}
	spin_unlock_irqrestore(&device->client_data_lock, flags);

	return ret;
}
EXPORT_SYMBOL(ib_get_client_data);

/**
742
 * ib_set_client_data - Set IB client context
L
Linus Torvalds 已提交
743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762
 * @device:Device to set context for
 * @client:Client to set context for
 * @data:Context to set
 *
 * ib_set_client_data() sets client context that can be retrieved with
 * ib_get_client_data().
 */
void ib_set_client_data(struct ib_device *device, struct ib_client *client,
			void *data)
{
	struct ib_client_data *context;
	unsigned long flags;

	spin_lock_irqsave(&device->client_data_lock, flags);
	list_for_each_entry(context, &device->client_data_list, list)
		if (context->client == client) {
			context->data = data;
			goto out;
		}

P
Parav Pandit 已提交
763 764
	pr_warn("No client context found for %s/%s\n",
		device->name, client->name);
L
Linus Torvalds 已提交
765 766 767 768 769 770 771 772 773 774 775 776 777 778 779

out:
	spin_unlock_irqrestore(&device->client_data_lock, flags);
}
EXPORT_SYMBOL(ib_set_client_data);

/**
 * ib_register_event_handler - Register an IB event handler
 * @event_handler:Handler to register
 *
 * ib_register_event_handler() registers an event handler that will be
 * called back when asynchronous IB events occur (as defined in
 * chapter 11 of the InfiniBand Architecture Specification).  This
 * callback may occur in interrupt context.
 */
780
void ib_register_event_handler(struct ib_event_handler *event_handler)
L
Linus Torvalds 已提交
781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797
{
	unsigned long flags;

	spin_lock_irqsave(&event_handler->device->event_handler_lock, flags);
	list_add_tail(&event_handler->list,
		      &event_handler->device->event_handler_list);
	spin_unlock_irqrestore(&event_handler->device->event_handler_lock, flags);
}
EXPORT_SYMBOL(ib_register_event_handler);

/**
 * ib_unregister_event_handler - Unregister an event handler
 * @event_handler:Handler to unregister
 *
 * Unregister an event handler registered with
 * ib_register_event_handler().
 */
798
void ib_unregister_event_handler(struct ib_event_handler *event_handler)
L
Linus Torvalds 已提交
799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842
{
	unsigned long flags;

	spin_lock_irqsave(&event_handler->device->event_handler_lock, flags);
	list_del(&event_handler->list);
	spin_unlock_irqrestore(&event_handler->device->event_handler_lock, flags);
}
EXPORT_SYMBOL(ib_unregister_event_handler);

/**
 * ib_dispatch_event - Dispatch an asynchronous event
 * @event:Event to dispatch
 *
 * Low-level drivers must call ib_dispatch_event() to dispatch the
 * event to all registered event handlers when an asynchronous event
 * occurs.
 */
void ib_dispatch_event(struct ib_event *event)
{
	unsigned long flags;
	struct ib_event_handler *handler;

	spin_lock_irqsave(&event->device->event_handler_lock, flags);

	list_for_each_entry(handler, &event->device->event_handler_list, list)
		handler->handler(handler, event);

	spin_unlock_irqrestore(&event->device->event_handler_lock, flags);
}
EXPORT_SYMBOL(ib_dispatch_event);

/**
 * ib_query_port - Query IB port attributes
 * @device:Device to query
 * @port_num:Port number to query
 * @port_attr:Port attributes
 *
 * ib_query_port() returns the attributes of a port through the
 * @port_attr pointer.
 */
int ib_query_port(struct ib_device *device,
		  u8 port_num,
		  struct ib_port_attr *port_attr)
{
843 844 845
	union ib_gid gid;
	int err;

846
	if (!rdma_is_port_valid(device, port_num))
847 848
		return -EINVAL;

849 850 851 852 853
	memset(port_attr, 0, sizeof(*port_attr));
	err = device->query_port(device, port_num, port_attr);
	if (err || port_attr->subnet_prefix)
		return err;

854 855 856
	if (rdma_port_get_link_layer(device, port_num) != IB_LINK_LAYER_INFINIBAND)
		return 0;

857 858 859 860 861 862
	err = ib_query_gid(device, port_num, 0, &gid, NULL);
	if (err)
		return err;

	port_attr->subnet_prefix = be64_to_cpu(gid.global.subnet_prefix);
	return 0;
L
Linus Torvalds 已提交
863 864 865 866 867 868 869 870 871
}
EXPORT_SYMBOL(ib_query_port);

/**
 * ib_query_gid - Get GID table entry
 * @device:Device to query
 * @port_num:Port number to query
 * @index:GID table index to query
 * @gid:Returned GID
872 873
 * @attr: Returned GID attributes related to this GID index (only in RoCE).
 *   NULL means ignore.
L
Linus Torvalds 已提交
874 875 876 877
 *
 * ib_query_gid() fetches the specified GID table entry.
 */
int ib_query_gid(struct ib_device *device,
878 879
		 u8 port_num, int index, union ib_gid *gid,
		 struct ib_gid_attr *attr)
L
Linus Torvalds 已提交
880
{
881
	if (rdma_cap_roce_gid_table(device, port_num))
882 883 884 885
		return ib_get_cached_gid(device, port_num, index, gid, attr);

	if (attr)
		return -EINVAL;
886

L
Linus Torvalds 已提交
887 888 889 890
	return device->query_gid(device, port_num, index, gid);
}
EXPORT_SYMBOL(ib_query_gid);

891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954
/**
 * ib_enum_roce_netdev - enumerate all RoCE ports
 * @ib_dev : IB device we want to query
 * @filter: Should we call the callback?
 * @filter_cookie: Cookie passed to filter
 * @cb: Callback to call for each found RoCE ports
 * @cookie: Cookie passed back to the callback
 *
 * Enumerates all of the physical RoCE ports of ib_dev
 * which are related to netdevice and calls callback() on each
 * device for which filter() function returns non zero.
 */
void ib_enum_roce_netdev(struct ib_device *ib_dev,
			 roce_netdev_filter filter,
			 void *filter_cookie,
			 roce_netdev_callback cb,
			 void *cookie)
{
	u8 port;

	for (port = rdma_start_port(ib_dev); port <= rdma_end_port(ib_dev);
	     port++)
		if (rdma_protocol_roce(ib_dev, port)) {
			struct net_device *idev = NULL;

			if (ib_dev->get_netdev)
				idev = ib_dev->get_netdev(ib_dev, port);

			if (idev &&
			    idev->reg_state >= NETREG_UNREGISTERED) {
				dev_put(idev);
				idev = NULL;
			}

			if (filter(ib_dev, port, idev, filter_cookie))
				cb(ib_dev, port, idev, cookie);

			if (idev)
				dev_put(idev);
		}
}

/**
 * ib_enum_all_roce_netdevs - enumerate all RoCE devices
 * @filter: Should we call the callback?
 * @filter_cookie: Cookie passed to filter
 * @cb: Callback to call for each found RoCE ports
 * @cookie: Cookie passed back to the callback
 *
 * Enumerates all RoCE devices' physical ports which are related
 * to netdevices and calls callback() on each device for which
 * filter() function returns non zero.
 */
void ib_enum_all_roce_netdevs(roce_netdev_filter filter,
			      void *filter_cookie,
			      roce_netdev_callback cb,
			      void *cookie)
{
	struct ib_device *dev;

	down_read(&lists_rwsem);
	list_for_each_entry(dev, &device_list, core_list)
		ib_enum_roce_netdev(dev, filter, filter_cookie, cb, cookie);
	up_read(&lists_rwsem);
955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979
}

/**
 * ib_enum_all_devs - enumerate all ib_devices
 * @cb: Callback to call for each found ib_device
 *
 * Enumerates all ib_devices and calls callback() on each device.
 */
int ib_enum_all_devs(nldev_callback nldev_cb, struct sk_buff *skb,
		     struct netlink_callback *cb)
{
	struct ib_device *dev;
	unsigned int idx = 0;
	int ret = 0;

	down_read(&lists_rwsem);
	list_for_each_entry(dev, &device_list, core_list) {
		ret = nldev_cb(dev, skb, cb, idx);
		if (ret)
			break;
		idx++;
	}

	up_read(&lists_rwsem);
	return ret;
980 981
}

L
Linus Torvalds 已提交
982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010
/**
 * ib_query_pkey - Get P_Key table entry
 * @device:Device to query
 * @port_num:Port number to query
 * @index:P_Key table index to query
 * @pkey:Returned P_Key
 *
 * ib_query_pkey() fetches the specified P_Key table entry.
 */
int ib_query_pkey(struct ib_device *device,
		  u8 port_num, u16 index, u16 *pkey)
{
	return device->query_pkey(device, port_num, index, pkey);
}
EXPORT_SYMBOL(ib_query_pkey);

/**
 * ib_modify_device - Change IB device attributes
 * @device:Device to modify
 * @device_modify_mask:Mask of attributes to change
 * @device_modify:New attribute values
 *
 * ib_modify_device() changes a device's attributes as specified by
 * the @device_modify_mask and @device_modify structure.
 */
int ib_modify_device(struct ib_device *device,
		     int device_modify_mask,
		     struct ib_device_modify *device_modify)
{
1011 1012 1013
	if (!device->modify_device)
		return -ENOSYS;

L
Linus Torvalds 已提交
1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033
	return device->modify_device(device, device_modify_mask,
				     device_modify);
}
EXPORT_SYMBOL(ib_modify_device);

/**
 * ib_modify_port - Modifies the attributes for the specified port.
 * @device: The device to modify.
 * @port_num: The number of the port to modify.
 * @port_modify_mask: Mask used to specify which attributes of the port
 *   to change.
 * @port_modify: New attribute values for the port.
 *
 * ib_modify_port() changes a port's attributes as specified by the
 * @port_modify_mask and @port_modify structure.
 */
int ib_modify_port(struct ib_device *device,
		   u8 port_num, int port_modify_mask,
		   struct ib_port_modify *port_modify)
{
1034
	int rc;
1035

1036
	if (!rdma_is_port_valid(device, port_num))
1037 1038
		return -EINVAL;

1039 1040 1041 1042 1043 1044
	if (device->modify_port)
		rc = device->modify_port(device, port_num, port_modify_mask,
					   port_modify);
	else
		rc = rdma_protocol_roce(device, port_num) ? 0 : -ENOSYS;
	return rc;
L
Linus Torvalds 已提交
1045 1046 1047
}
EXPORT_SYMBOL(ib_modify_port);

1048 1049
/**
 * ib_find_gid - Returns the port number and GID table index where
1050
 *   a specified GID value occurs. Its searches only for IB link layer.
1051 1052
 * @device: The device to query.
 * @gid: The GID value to search for.
1053
 * @ndev: The ndev related to the GID to search for.
1054 1055 1056 1057 1058
 * @port_num: The port number of the device where the GID value was found.
 * @index: The index into the GID table where the GID was found.  This
 *   parameter may be NULL.
 */
int ib_find_gid(struct ib_device *device, union ib_gid *gid,
1059
		struct net_device *ndev, u8 *port_num, u16 *index)
1060 1061 1062 1063
{
	union ib_gid tmp_gid;
	int ret, port, i;

1064
	for (port = rdma_start_port(device); port <= rdma_end_port(device); ++port) {
1065
		if (rdma_cap_roce_gid_table(device, port))
1066 1067
			continue;

1068
		for (i = 0; i < device->port_immutable[port].gid_tbl_len; ++i) {
1069
			ret = ib_query_gid(device, port, i, &tmp_gid, NULL);
1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097
			if (ret)
				return ret;
			if (!memcmp(&tmp_gid, gid, sizeof *gid)) {
				*port_num = port;
				if (index)
					*index = i;
				return 0;
			}
		}
	}

	return -ENOENT;
}
EXPORT_SYMBOL(ib_find_gid);

/**
 * ib_find_pkey - Returns the PKey table index where a specified
 *   PKey value occurs.
 * @device: The device to query.
 * @port_num: The port number of the device to search for the PKey.
 * @pkey: The PKey value to search for.
 * @index: The index into the PKey table where the PKey was found.
 */
int ib_find_pkey(struct ib_device *device,
		 u8 port_num, u16 pkey, u16 *index)
{
	int ret, i;
	u16 tmp_pkey;
1098
	int partial_ix = -1;
1099

1100
	for (i = 0; i < device->port_immutable[port_num].pkey_tbl_len; ++i) {
1101 1102 1103
		ret = ib_query_pkey(device, port_num, i, &tmp_pkey);
		if (ret)
			return ret;
1104
		if ((pkey & 0x7fff) == (tmp_pkey & 0x7fff)) {
1105 1106 1107 1108 1109 1110 1111
			/* if there is full-member pkey take it.*/
			if (tmp_pkey & 0x8000) {
				*index = i;
				return 0;
			}
			if (partial_ix < 0)
				partial_ix = i;
1112 1113 1114
		}
	}

1115 1116 1117 1118 1119
	/*no full-member, if exists take the limited*/
	if (partial_ix >= 0) {
		*index = partial_ix;
		return 0;
	}
1120 1121 1122 1123
	return -ENOENT;
}
EXPORT_SYMBOL(ib_find_pkey);

1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168
/**
 * ib_get_net_dev_by_params() - Return the appropriate net_dev
 * for a received CM request
 * @dev:	An RDMA device on which the request has been received.
 * @port:	Port number on the RDMA device.
 * @pkey:	The Pkey the request came on.
 * @gid:	A GID that the net_dev uses to communicate.
 * @addr:	Contains the IP address that the request specified as its
 *		destination.
 */
struct net_device *ib_get_net_dev_by_params(struct ib_device *dev,
					    u8 port,
					    u16 pkey,
					    const union ib_gid *gid,
					    const struct sockaddr *addr)
{
	struct net_device *net_dev = NULL;
	struct ib_client_data *context;

	if (!rdma_protocol_ib(dev, port))
		return NULL;

	down_read(&lists_rwsem);

	list_for_each_entry(context, &dev->client_data_list, list) {
		struct ib_client *client = context->client;

		if (context->going_down)
			continue;

		if (client->get_net_dev_by_params) {
			net_dev = client->get_net_dev_by_params(dev, port, pkey,
								gid, addr,
								context->data);
			if (net_dev)
				break;
		}
	}

	up_read(&lists_rwsem);

	return net_dev;
}
EXPORT_SYMBOL(ib_get_net_dev_by_params);

1169
static const struct rdma_nl_cbs ibnl_ls_cb_table[RDMA_NL_LS_NUM_OPS] = {
1170
	[RDMA_NL_LS_OP_RESOLVE] = {
1171
		.doit = ib_nl_handle_resolve_resp,
1172 1173
		.flags = RDMA_NL_ADMIN_PERM,
	},
1174
	[RDMA_NL_LS_OP_SET_TIMEOUT] = {
1175
		.doit = ib_nl_handle_set_timeout,
1176 1177
		.flags = RDMA_NL_ADMIN_PERM,
	},
1178
	[RDMA_NL_LS_OP_IP_RESOLVE] = {
1179
		.doit = ib_nl_handle_ip_res_resp,
1180 1181
		.flags = RDMA_NL_ADMIN_PERM,
	},
1182 1183
};

L
Linus Torvalds 已提交
1184 1185 1186 1187
static int __init ib_core_init(void)
{
	int ret;

T
Tejun Heo 已提交
1188 1189 1190 1191
	ib_wq = alloc_workqueue("infiniband", 0, 0);
	if (!ib_wq)
		return -ENOMEM;

1192
	ib_comp_wq = alloc_workqueue("ib-comp-wq",
1193
			WQ_HIGHPRI | WQ_MEM_RECLAIM | WQ_SYSFS, 0);
1194 1195 1196 1197 1198
	if (!ib_comp_wq) {
		ret = -ENOMEM;
		goto err;
	}

1199
	ret = class_register(&ib_class);
1200
	if (ret) {
P
Parav Pandit 已提交
1201
		pr_warn("Couldn't create InfiniBand device class\n");
1202
		goto err_comp;
1203
	}
L
Linus Torvalds 已提交
1204

1205
	ret = rdma_nl_init();
R
Roland Dreier 已提交
1206
	if (ret) {
1207
		pr_warn("Couldn't init IB netlink interface: err %d\n", ret);
R
Roland Dreier 已提交
1208 1209 1210
		goto err_sysfs;
	}

1211 1212 1213 1214 1215 1216
	ret = addr_init();
	if (ret) {
		pr_warn("Could't init IB address resolution\n");
		goto err_ibnl;
	}

1217 1218 1219 1220 1221 1222
	ret = ib_mad_init();
	if (ret) {
		pr_warn("Couldn't init IB MAD\n");
		goto err_addr;
	}

1223 1224 1225 1226 1227 1228
	ret = ib_sa_init();
	if (ret) {
		pr_warn("Couldn't init SA\n");
		goto err_mad;
	}

1229 1230 1231
	ret = register_lsm_notifier(&ibdev_lsm_nb);
	if (ret) {
		pr_warn("Couldn't register LSM notifier. ret %d\n", ret);
1232
		goto err_sa;
1233 1234
	}

1235
	nldev_init();
1236
	rdma_nl_register(RDMA_NL_LS, ibnl_ls_cb_table);
1237
	ib_cache_setup();
L
Linus Torvalds 已提交
1238

1239 1240
	return 0;

1241 1242
err_sa:
	ib_sa_cleanup();
1243 1244
err_mad:
	ib_mad_cleanup();
1245 1246
err_addr:
	addr_cleanup();
1247
err_ibnl:
1248
	rdma_nl_exit();
1249
err_sysfs:
1250
	class_unregister(&ib_class);
1251 1252
err_comp:
	destroy_workqueue(ib_comp_wq);
1253 1254
err:
	destroy_workqueue(ib_wq);
L
Linus Torvalds 已提交
1255 1256 1257 1258 1259 1260
	return ret;
}

static void __exit ib_core_cleanup(void)
{
	ib_cache_cleanup();
1261
	nldev_exit();
1262 1263
	rdma_nl_unregister(RDMA_NL_LS);
	unregister_lsm_notifier(&ibdev_lsm_nb);
1264
	ib_sa_cleanup();
1265
	ib_mad_cleanup();
1266
	addr_cleanup();
1267
	rdma_nl_exit();
1268
	class_unregister(&ib_class);
1269
	destroy_workqueue(ib_comp_wq);
1270
	/* Make sure that any pending umem accounting work is done. */
T
Tejun Heo 已提交
1271
	destroy_workqueue(ib_wq);
L
Linus Torvalds 已提交
1272 1273
}

1274 1275
MODULE_ALIAS_RDMA_NETLINK(RDMA_NL_LS, 4);

1276
subsys_initcall(ib_core_init);
L
Linus Torvalds 已提交
1277
module_exit(ib_core_cleanup);