dm-ioctl.c 41.6 KB
Newer Older
L
Linus Torvalds 已提交
1 2
/*
 * Copyright (C) 2001, 2002 Sistina Software (UK) Limited.
3
 * Copyright (C) 2004 - 2006 Red Hat, Inc. All rights reserved.
L
Linus Torvalds 已提交
4 5 6 7 8 9 10 11 12 13 14 15 16
 *
 * This file is released under the GPL.
 */

#include "dm.h"

#include <linux/module.h>
#include <linux/vmalloc.h>
#include <linux/miscdevice.h>
#include <linux/init.h>
#include <linux/wait.h>
#include <linux/slab.h>
#include <linux/dm-ioctl.h>
D
Darrick J. Wong 已提交
17
#include <linux/hdreg.h>
M
Milan Broz 已提交
18
#include <linux/compat.h>
L
Linus Torvalds 已提交
19 20 21

#include <asm/uaccess.h>

22
#define DM_MSG_PREFIX "ioctl"
L
Linus Torvalds 已提交
23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38
#define DM_DRIVER_EMAIL "dm-devel@redhat.com"

/*-----------------------------------------------------------------
 * The ioctl interface needs to be able to look up devices by
 * name or uuid.
 *---------------------------------------------------------------*/
struct hash_cell {
	struct list_head name_list;
	struct list_head uuid_list;

	char *name;
	char *uuid;
	struct mapped_device *md;
	struct dm_table *new_map;
};

M
Mikulas Patocka 已提交
39 40 41 42 43 44 45 46
/*
 * A dummy definition to make RCU happy.
 * struct dm_table should never be dereferenced in this file.
 */
struct dm_table {
	int undefined__;
};

L
Linus Torvalds 已提交
47 48 49 50 51 52 53 54 55 56 57 58 59
struct vers_iter {
    size_t param_size;
    struct dm_target_versions *vers, *old_vers;
    char *end;
    uint32_t flags;
};


#define NUM_BUCKETS 64
#define MASK_BUCKETS (NUM_BUCKETS - 1)
static struct list_head _name_buckets[NUM_BUCKETS];
static struct list_head _uuid_buckets[NUM_BUCKETS];

60
static void dm_hash_remove_all(int keep_open_devices);
L
Linus Torvalds 已提交
61 62 63 64 65 66

/*
 * Guards access to both hash tables.
 */
static DECLARE_RWSEM(_hash_lock);

M
Mikulas Patocka 已提交
67 68 69 70 71
/*
 * Protects use of mdptr to obtain hash cell name and uuid from mapped device.
 */
static DEFINE_MUTEX(dm_hash_cells_mutex);

L
Linus Torvalds 已提交
72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88
static void init_buckets(struct list_head *buckets)
{
	unsigned int i;

	for (i = 0; i < NUM_BUCKETS; i++)
		INIT_LIST_HEAD(buckets + i);
}

static int dm_hash_init(void)
{
	init_buckets(_name_buckets);
	init_buckets(_uuid_buckets);
	return 0;
}

static void dm_hash_exit(void)
{
89
	dm_hash_remove_all(0);
L
Linus Torvalds 已提交
90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116
}

/*-----------------------------------------------------------------
 * Hash function:
 * We're not really concerned with the str hash function being
 * fast since it's only used by the ioctl interface.
 *---------------------------------------------------------------*/
static unsigned int hash_str(const char *str)
{
	const unsigned int hash_mult = 2654435387U;
	unsigned int h = 0;

	while (*str)
		h = (h + (unsigned int) *str++) * hash_mult;

	return h & MASK_BUCKETS;
}

/*-----------------------------------------------------------------
 * Code for looking up a device by name
 *---------------------------------------------------------------*/
static struct hash_cell *__get_name_cell(const char *str)
{
	struct hash_cell *hc;
	unsigned int h = hash_str(str);

	list_for_each_entry (hc, _name_buckets + h, name_list)
117 118
		if (!strcmp(hc->name, str)) {
			dm_get(hc->md);
L
Linus Torvalds 已提交
119
			return hc;
120
		}
L
Linus Torvalds 已提交
121 122 123 124 125 126 127 128 129 130

	return NULL;
}

static struct hash_cell *__get_uuid_cell(const char *str)
{
	struct hash_cell *hc;
	unsigned int h = hash_str(str);

	list_for_each_entry (hc, _uuid_buckets + h, uuid_list)
131 132
		if (!strcmp(hc->uuid, str)) {
			dm_get(hc->md);
L
Linus Torvalds 已提交
133
			return hc;
134
		}
L
Linus Torvalds 已提交
135 136 137 138

	return NULL;
}

139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156
static struct hash_cell *__get_dev_cell(uint64_t dev)
{
	struct mapped_device *md;
	struct hash_cell *hc;

	md = dm_get_md(huge_decode_dev(dev));
	if (!md)
		return NULL;

	hc = dm_get_mdptr(md);
	if (!hc) {
		dm_put(md);
		return NULL;
	}

	return hc;
}

L
Linus Torvalds 已提交
157 158 159 160 161 162 163 164 165 166 167 168
/*-----------------------------------------------------------------
 * Inserting, removing and renaming a device.
 *---------------------------------------------------------------*/
static struct hash_cell *alloc_cell(const char *name, const char *uuid,
				    struct mapped_device *md)
{
	struct hash_cell *hc;

	hc = kmalloc(sizeof(*hc), GFP_KERNEL);
	if (!hc)
		return NULL;

169
	hc->name = kstrdup(name, GFP_KERNEL);
L
Linus Torvalds 已提交
170 171 172 173 174 175 176 177 178
	if (!hc->name) {
		kfree(hc);
		return NULL;
	}

	if (!uuid)
		hc->uuid = NULL;

	else {
179
		hc->uuid = kstrdup(uuid, GFP_KERNEL);
L
Linus Torvalds 已提交
180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208
		if (!hc->uuid) {
			kfree(hc->name);
			kfree(hc);
			return NULL;
		}
	}

	INIT_LIST_HEAD(&hc->name_list);
	INIT_LIST_HEAD(&hc->uuid_list);
	hc->md = md;
	hc->new_map = NULL;
	return hc;
}

static void free_cell(struct hash_cell *hc)
{
	if (hc) {
		kfree(hc->name);
		kfree(hc->uuid);
		kfree(hc);
	}
}

/*
 * The kdev_t and uuid of a device can never change once it is
 * initially inserted.
 */
static int dm_hash_insert(const char *name, const char *uuid, struct mapped_device *md)
{
209
	struct hash_cell *cell, *hc;
L
Linus Torvalds 已提交
210 211 212 213 214 215 216 217 218 219 220 221

	/*
	 * Allocate the new cells.
	 */
	cell = alloc_cell(name, uuid, md);
	if (!cell)
		return -ENOMEM;

	/*
	 * Insert the cell into both hash tables.
	 */
	down_write(&_hash_lock);
222 223 224
	hc = __get_name_cell(name);
	if (hc) {
		dm_put(hc->md);
L
Linus Torvalds 已提交
225
		goto bad;
226
	}
L
Linus Torvalds 已提交
227 228 229 230

	list_add(&cell->name_list, _name_buckets + hash_str(name));

	if (uuid) {
231 232
		hc = __get_uuid_cell(uuid);
		if (hc) {
L
Linus Torvalds 已提交
233
			list_del(&cell->name_list);
234
			dm_put(hc->md);
L
Linus Torvalds 已提交
235 236 237 238 239
			goto bad;
		}
		list_add(&cell->uuid_list, _uuid_buckets + hash_str(uuid));
	}
	dm_get(md);
M
Mikulas Patocka 已提交
240
	mutex_lock(&dm_hash_cells_mutex);
L
Linus Torvalds 已提交
241
	dm_set_mdptr(md, cell);
M
Mikulas Patocka 已提交
242
	mutex_unlock(&dm_hash_cells_mutex);
L
Linus Torvalds 已提交
243 244 245 246 247 248 249 250 251 252
	up_write(&_hash_lock);

	return 0;

 bad:
	up_write(&_hash_lock);
	free_cell(cell);
	return -EBUSY;
}

M
Mikulas Patocka 已提交
253
static struct dm_table *__hash_remove(struct hash_cell *hc)
L
Linus Torvalds 已提交
254
{
255
	struct dm_table *table;
M
Mikulas Patocka 已提交
256
	int srcu_idx;
257

L
Linus Torvalds 已提交
258 259 260
	/* remove from the dev hash */
	list_del(&hc->uuid_list);
	list_del(&hc->name_list);
M
Mikulas Patocka 已提交
261
	mutex_lock(&dm_hash_cells_mutex);
L
Linus Torvalds 已提交
262
	dm_set_mdptr(hc->md, NULL);
M
Mikulas Patocka 已提交
263
	mutex_unlock(&dm_hash_cells_mutex);
264

M
Mikulas Patocka 已提交
265 266
	table = dm_get_live_table(hc->md, &srcu_idx);
	if (table)
267
		dm_table_event(table);
M
Mikulas Patocka 已提交
268
	dm_put_live_table(hc->md, srcu_idx);
269

M
Mikulas Patocka 已提交
270
	table = NULL;
L
Linus Torvalds 已提交
271
	if (hc->new_map)
M
Mikulas Patocka 已提交
272
		table = hc->new_map;
M
Mike Anderson 已提交
273
	dm_put(hc->md);
L
Linus Torvalds 已提交
274
	free_cell(hc);
M
Mikulas Patocka 已提交
275 276

	return table;
L
Linus Torvalds 已提交
277 278
}

279
static void dm_hash_remove_all(int keep_open_devices)
L
Linus Torvalds 已提交
280
{
281
	int i, dev_skipped;
L
Linus Torvalds 已提交
282
	struct hash_cell *hc;
283
	struct mapped_device *md;
M
Mikulas Patocka 已提交
284
	struct dm_table *t;
285 286 287

retry:
	dev_skipped = 0;
L
Linus Torvalds 已提交
288 289

	down_write(&_hash_lock);
290

L
Linus Torvalds 已提交
291
	for (i = 0; i < NUM_BUCKETS; i++) {
292 293 294
		list_for_each_entry(hc, _name_buckets + i, name_list) {
			md = hc->md;
			dm_get(md);
295

296 297
			if (keep_open_devices && dm_lock_for_deletion(md)) {
				dm_put(md);
298 299 300
				dev_skipped++;
				continue;
			}
301

M
Mikulas Patocka 已提交
302
			t = __hash_remove(hc);
303

304
			up_write(&_hash_lock);
305

M
Mikulas Patocka 已提交
306 307 308 309
			if (t) {
				dm_sync_table(md);
				dm_table_destroy(t);
			}
310
			dm_put(md);
311 312 313 314
			if (likely(keep_open_devices))
				dm_destroy(md);
			else
				dm_destroy_immediate(md);
315 316 317 318 319 320 321 322 323

			/*
			 * Some mapped devices may be using other mapped
			 * devices, so repeat until we make no further
			 * progress.  If a new mapped device is created
			 * here it will also get removed.
			 */
			goto retry;
		}
324 325
	}

L
Linus Torvalds 已提交
326
	up_write(&_hash_lock);
327 328 329

	if (dev_skipped)
		DMWARN("remove_all left %d open device(s)", dev_skipped);
L
Linus Torvalds 已提交
330 331
}

332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366
/*
 * Set the uuid of a hash_cell that isn't already set.
 */
static void __set_cell_uuid(struct hash_cell *hc, char *new_uuid)
{
	mutex_lock(&dm_hash_cells_mutex);
	hc->uuid = new_uuid;
	mutex_unlock(&dm_hash_cells_mutex);

	list_add(&hc->uuid_list, _uuid_buckets + hash_str(new_uuid));
}

/*
 * Changes the name of a hash_cell and returns the old name for
 * the caller to free.
 */
static char *__change_cell_name(struct hash_cell *hc, char *new_name)
{
	char *old_name;

	/*
	 * Rename and move the name cell.
	 */
	list_del(&hc->name_list);
	old_name = hc->name;

	mutex_lock(&dm_hash_cells_mutex);
	hc->name = new_name;
	mutex_unlock(&dm_hash_cells_mutex);

	list_add(&hc->name_list, _name_buckets + hash_str(new_name));

	return old_name;
}

367 368
static struct mapped_device *dm_hash_rename(struct dm_ioctl *param,
					    const char *new)
L
Linus Torvalds 已提交
369
{
370
	char *new_data, *old_name = NULL;
L
Linus Torvalds 已提交
371
	struct hash_cell *hc;
372
	struct dm_table *table;
373
	struct mapped_device *md;
374
	unsigned change_uuid = (param->flags & DM_UUID_FLAG) ? 1 : 0;
M
Mikulas Patocka 已提交
375
	int srcu_idx;
L
Linus Torvalds 已提交
376 377 378 379

	/*
	 * duplicate new.
	 */
380 381
	new_data = kstrdup(new, GFP_KERNEL);
	if (!new_data)
382
		return ERR_PTR(-ENOMEM);
L
Linus Torvalds 已提交
383 384 385 386 387 388

	down_write(&_hash_lock);

	/*
	 * Is new free ?
	 */
389 390 391 392 393
	if (change_uuid)
		hc = __get_uuid_cell(new);
	else
		hc = __get_name_cell(new);

L
Linus Torvalds 已提交
394
	if (hc) {
395 396 397
		DMWARN("Unable to change %s on mapped device %s to one that "
		       "already exists: %s",
		       change_uuid ? "uuid" : "name",
398
		       param->name, new);
399
		dm_put(hc->md);
L
Linus Torvalds 已提交
400
		up_write(&_hash_lock);
401
		kfree(new_data);
402
		return ERR_PTR(-EBUSY);
L
Linus Torvalds 已提交
403 404 405 406 407
	}

	/*
	 * Is there such a device as 'old' ?
	 */
408
	hc = __get_name_cell(param->name);
L
Linus Torvalds 已提交
409
	if (!hc) {
410 411
		DMWARN("Unable to rename non-existent device, %s to %s%s",
		       param->name, change_uuid ? "uuid " : "", new);
L
Linus Torvalds 已提交
412
		up_write(&_hash_lock);
413
		kfree(new_data);
414
		return ERR_PTR(-ENXIO);
L
Linus Torvalds 已提交
415 416 417
	}

	/*
418
	 * Does this device already have a uuid?
L
Linus Torvalds 已提交
419
	 */
420 421 422 423 424 425 426 427 428 429 430 431 432 433
	if (change_uuid && hc->uuid) {
		DMWARN("Unable to change uuid of mapped device %s to %s "
		       "because uuid is already set to %s",
		       param->name, new, hc->uuid);
		dm_put(hc->md);
		up_write(&_hash_lock);
		kfree(new_data);
		return ERR_PTR(-EINVAL);
	}

	if (change_uuid)
		__set_cell_uuid(hc, new_data);
	else
		old_name = __change_cell_name(hc, new_data);
L
Linus Torvalds 已提交
434

435 436 437
	/*
	 * Wake up any dm event waiters.
	 */
M
Mikulas Patocka 已提交
438 439
	table = dm_get_live_table(hc->md, &srcu_idx);
	if (table)
440
		dm_table_event(table);
M
Mikulas Patocka 已提交
441
	dm_put_live_table(hc->md, srcu_idx);
442

443 444
	if (!dm_kobject_uevent(hc->md, KOBJ_CHANGE, param->event_nr))
		param->flags |= DM_UEVENT_GENERATED_FLAG;
445

446
	md = hc->md;
L
Linus Torvalds 已提交
447 448
	up_write(&_hash_lock);
	kfree(old_name);
449 450

	return md;
L
Linus Torvalds 已提交
451 452 453 454 455 456 457 458 459 460 461 462 463
}

/*-----------------------------------------------------------------
 * Implementation of the ioctl commands
 *---------------------------------------------------------------*/
/*
 * All the ioctl commands get dispatched to functions with this
 * prototype.
 */
typedef int (*ioctl_fn)(struct dm_ioctl *param, size_t param_size);

static int remove_all(struct dm_ioctl *param, size_t param_size)
{
464
	dm_hash_remove_all(1);
L
Linus Torvalds 已提交
465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537
	param->data_size = 0;
	return 0;
}

/*
 * Round up the ptr to an 8-byte boundary.
 */
#define ALIGN_MASK 7
static inline void *align_ptr(void *ptr)
{
	return (void *) (((size_t) (ptr + ALIGN_MASK)) & ~ALIGN_MASK);
}

/*
 * Retrieves the data payload buffer from an already allocated
 * struct dm_ioctl.
 */
static void *get_result_buffer(struct dm_ioctl *param, size_t param_size,
			       size_t *len)
{
	param->data_start = align_ptr(param + 1) - (void *) param;

	if (param->data_start < param_size)
		*len = param_size - param->data_start;
	else
		*len = 0;

	return ((void *) param) + param->data_start;
}

static int list_devices(struct dm_ioctl *param, size_t param_size)
{
	unsigned int i;
	struct hash_cell *hc;
	size_t len, needed = 0;
	struct gendisk *disk;
	struct dm_name_list *nl, *old_nl = NULL;

	down_write(&_hash_lock);

	/*
	 * Loop through all the devices working out how much
	 * space we need.
	 */
	for (i = 0; i < NUM_BUCKETS; i++) {
		list_for_each_entry (hc, _name_buckets + i, name_list) {
			needed += sizeof(struct dm_name_list);
			needed += strlen(hc->name) + 1;
			needed += ALIGN_MASK;
		}
	}

	/*
	 * Grab our output buffer.
	 */
	nl = get_result_buffer(param, param_size, &len);
	if (len < needed) {
		param->flags |= DM_BUFFER_FULL_FLAG;
		goto out;
	}
	param->data_size = param->data_start + needed;

	nl->dev = 0;	/* Flags no data */

	/*
	 * Now loop through filling out the names.
	 */
	for (i = 0; i < NUM_BUCKETS; i++) {
		list_for_each_entry (hc, _name_buckets + i, name_list) {
			if (old_nl)
				old_nl->next = (uint32_t) ((void *) nl -
							   (void *) old_nl);
			disk = dm_disk(hc->md);
538
			nl->dev = huge_encode_dev(disk_devt(disk));
L
Linus Torvalds 已提交
539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555
			nl->next = 0;
			strcpy(nl->name, hc->name);

			old_nl = nl;
			nl = align_ptr(((void *) ++nl) + strlen(hc->name) + 1);
		}
	}

 out:
	up_write(&_hash_lock);
	return 0;
}

static void list_version_get_needed(struct target_type *tt, void *needed_param)
{
    size_t *needed = needed_param;

556
    *needed += sizeof(struct dm_target_versions);
L
Linus Torvalds 已提交
557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633
    *needed += strlen(tt->name);
    *needed += ALIGN_MASK;
}

static void list_version_get_info(struct target_type *tt, void *param)
{
    struct vers_iter *info = param;

    /* Check space - it might have changed since the first iteration */
    if ((char *)info->vers + sizeof(tt->version) + strlen(tt->name) + 1 >
	info->end) {

	info->flags = DM_BUFFER_FULL_FLAG;
	return;
    }

    if (info->old_vers)
	info->old_vers->next = (uint32_t) ((void *)info->vers -
					   (void *)info->old_vers);
    info->vers->version[0] = tt->version[0];
    info->vers->version[1] = tt->version[1];
    info->vers->version[2] = tt->version[2];
    info->vers->next = 0;
    strcpy(info->vers->name, tt->name);

    info->old_vers = info->vers;
    info->vers = align_ptr(((void *) ++info->vers) + strlen(tt->name) + 1);
}

static int list_versions(struct dm_ioctl *param, size_t param_size)
{
	size_t len, needed = 0;
	struct dm_target_versions *vers;
	struct vers_iter iter_info;

	/*
	 * Loop through all the devices working out how much
	 * space we need.
	 */
	dm_target_iterate(list_version_get_needed, &needed);

	/*
	 * Grab our output buffer.
	 */
	vers = get_result_buffer(param, param_size, &len);
	if (len < needed) {
		param->flags |= DM_BUFFER_FULL_FLAG;
		goto out;
	}
	param->data_size = param->data_start + needed;

	iter_info.param_size = param_size;
	iter_info.old_vers = NULL;
	iter_info.vers = vers;
	iter_info.flags = 0;
	iter_info.end = (char *)vers+len;

	/*
	 * Now loop through filling out the names & versions.
	 */
	dm_target_iterate(list_version_get_info, &iter_info);
	param->flags |= iter_info.flags;

 out:
	return 0;
}

static int check_name(const char *name)
{
	if (strchr(name, '/')) {
		DMWARN("invalid device name");
		return -EINVAL;
	}

	return 0;
}

634 635 636 637 638
/*
 * On successful return, the caller must not attempt to acquire
 * _hash_lock without first calling dm_table_put, because dm_table_destroy
 * waits for this dm_table_put and could be called under this lock.
 */
M
Mikulas Patocka 已提交
639
static struct dm_table *dm_get_inactive_table(struct mapped_device *md, int *srcu_idx)
640 641 642 643
{
	struct hash_cell *hc;
	struct dm_table *table = NULL;

M
Mikulas Patocka 已提交
644 645 646
	/* increment rcu count, we don't care about the table pointer */
	dm_get_live_table(md, srcu_idx);

647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662
	down_read(&_hash_lock);
	hc = dm_get_mdptr(md);
	if (!hc || hc->md != md) {
		DMWARN("device has been removed from the dev hash table.");
		goto out;
	}

	table = hc->new_map;

out:
	up_read(&_hash_lock);

	return table;
}

static struct dm_table *dm_get_live_or_inactive_table(struct mapped_device *md,
M
Mikulas Patocka 已提交
663 664
						      struct dm_ioctl *param,
						      int *srcu_idx)
665 666
{
	return (param->flags & DM_QUERY_INACTIVE_TABLE_FLAG) ?
M
Mikulas Patocka 已提交
667
		dm_get_inactive_table(md, srcu_idx) : dm_get_live_table(md, srcu_idx);
668 669
}

L
Linus Torvalds 已提交
670 671 672 673
/*
 * Fills in a dm_ioctl structure, ready for sending back to
 * userland.
 */
674
static void __dev_status(struct mapped_device *md, struct dm_ioctl *param)
L
Linus Torvalds 已提交
675 676 677
{
	struct gendisk *disk = dm_disk(md);
	struct dm_table *table;
M
Mikulas Patocka 已提交
678
	int srcu_idx;
L
Linus Torvalds 已提交
679 680 681 682

	param->flags &= ~(DM_SUSPEND_FLAG | DM_READONLY_FLAG |
			  DM_ACTIVE_PRESENT_FLAG);

683
	if (dm_suspended_md(md))
L
Linus Torvalds 已提交
684 685
		param->flags |= DM_SUSPEND_FLAG;

686
	param->dev = huge_encode_dev(disk_devt(disk));
L
Linus Torvalds 已提交
687

688 689 690 691 692 693
	/*
	 * Yes, this will be out of date by the time it gets back
	 * to userland, but it is still very useful for
	 * debugging.
	 */
	param->open_count = dm_open_count(md);
L
Linus Torvalds 已提交
694 695

	param->event_nr = dm_get_event_nr(md);
696
	param->target_count = 0;
L
Linus Torvalds 已提交
697

M
Mikulas Patocka 已提交
698
	table = dm_get_live_table(md, &srcu_idx);
L
Linus Torvalds 已提交
699
	if (table) {
700 701 702 703 704 705 706 707
		if (!(param->flags & DM_QUERY_INACTIVE_TABLE_FLAG)) {
			if (get_disk_ro(disk))
				param->flags |= DM_READONLY_FLAG;
			param->target_count = dm_table_get_num_targets(table);
		}

		param->flags |= DM_ACTIVE_PRESENT_FLAG;
	}
M
Mikulas Patocka 已提交
708
	dm_put_live_table(md, srcu_idx);
709 710

	if (param->flags & DM_QUERY_INACTIVE_TABLE_FLAG) {
M
Mikulas Patocka 已提交
711 712
		int srcu_idx;
		table = dm_get_inactive_table(md, &srcu_idx);
713 714 715 716 717
		if (table) {
			if (!(dm_table_get_mode(table) & FMODE_WRITE))
				param->flags |= DM_READONLY_FLAG;
			param->target_count = dm_table_get_num_targets(table);
		}
M
Mikulas Patocka 已提交
718
		dm_put_live_table(md, srcu_idx);
719
	}
L
Linus Torvalds 已提交
720 721 722 723
}

static int dev_create(struct dm_ioctl *param, size_t param_size)
{
724
	int r, m = DM_ANY_MINOR;
L
Linus Torvalds 已提交
725 726 727 728 729 730 731
	struct mapped_device *md;

	r = check_name(param->name);
	if (r)
		return r;

	if (param->flags & DM_PERSISTENT_DEV_FLAG)
732
		m = MINOR(huge_decode_dev(param->dev));
L
Linus Torvalds 已提交
733

734
	r = dm_create(m, &md);
L
Linus Torvalds 已提交
735 736 737 738
	if (r)
		return r;

	r = dm_hash_insert(param->name, *param->uuid ? param->uuid : NULL, md);
739 740 741 742 743
	if (r) {
		dm_put(md);
		dm_destroy(md);
		return r;
	}
L
Linus Torvalds 已提交
744 745 746

	param->flags &= ~DM_INACTIVE_PRESENT_FLAG;

747 748
	__dev_status(md, param);

L
Linus Torvalds 已提交
749 750
	dm_put(md);

751
	return 0;
L
Linus Torvalds 已提交
752 753 754 755 756
}

/*
 * Always use UUID for lookups if it's present, otherwise use name or dev.
 */
757
static struct hash_cell *__find_device_hash_cell(struct dm_ioctl *param)
L
Linus Torvalds 已提交
758
{
759
	struct hash_cell *hc = NULL;
A
Alasdair G Kergon 已提交
760

761
	if (*param->uuid) {
762 763 764
		if (*param->name || param->dev)
			return NULL;

765 766 767
		hc = __get_uuid_cell(param->uuid);
		if (!hc)
			return NULL;
768
	} else if (*param->name) {
769 770 771
		if (param->dev)
			return NULL;

772 773 774
		hc = __get_name_cell(param->name);
		if (!hc)
			return NULL;
775 776 777 778 779
	} else if (param->dev) {
		hc = __get_dev_cell(param->dev);
		if (!hc)
			return NULL;
	} else
780
		return NULL;
A
Alasdair G Kergon 已提交
781

782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797
	/*
	 * Sneakily write in both the name and the uuid
	 * while we have the cell.
	 */
	strlcpy(param->name, hc->name, sizeof(param->name));
	if (hc->uuid)
		strlcpy(param->uuid, hc->uuid, sizeof(param->uuid));
	else
		param->uuid[0] = '\0';

	if (hc->new_map)
		param->flags |= DM_INACTIVE_PRESENT_FLAG;
	else
		param->flags &= ~DM_INACTIVE_PRESENT_FLAG;

	return hc;
L
Linus Torvalds 已提交
798 799
}

800
static struct mapped_device *find_device(struct dm_ioctl *param)
L
Linus Torvalds 已提交
801 802 803 804 805 806
{
	struct hash_cell *hc;
	struct mapped_device *md = NULL;

	down_read(&_hash_lock);
	hc = __find_device_hash_cell(param);
807
	if (hc)
L
Linus Torvalds 已提交
808 809 810 811 812 813 814 815 816
		md = hc->md;
	up_read(&_hash_lock);

	return md;
}

static int dev_remove(struct dm_ioctl *param, size_t param_size)
{
	struct hash_cell *hc;
817
	struct mapped_device *md;
818
	int r;
M
Mikulas Patocka 已提交
819
	struct dm_table *t;
L
Linus Torvalds 已提交
820 821 822 823 824

	down_write(&_hash_lock);
	hc = __find_device_hash_cell(param);

	if (!hc) {
825
		DMDEBUG_LIMIT("device doesn't appear to be in the dev hash table.");
L
Linus Torvalds 已提交
826 827 828 829
		up_write(&_hash_lock);
		return -ENXIO;
	}

830 831
	md = hc->md;

832 833 834 835 836
	/*
	 * Ensure the device is not open and nothing further can open it.
	 */
	r = dm_lock_for_deletion(md);
	if (r) {
837
		DMDEBUG_LIMIT("unable to remove open device %s", hc->name);
838 839 840 841 842
		up_write(&_hash_lock);
		dm_put(md);
		return r;
	}

M
Mikulas Patocka 已提交
843
	t = __hash_remove(hc);
L
Linus Torvalds 已提交
844
	up_write(&_hash_lock);
M
Milan Broz 已提交
845

M
Mikulas Patocka 已提交
846 847 848 849 850
	if (t) {
		dm_sync_table(md);
		dm_table_destroy(t);
	}

851 852
	if (!dm_kobject_uevent(md, KOBJ_REMOVE, param->event_nr))
		param->flags |= DM_UEVENT_GENERATED_FLAG;
M
Milan Broz 已提交
853

854
	dm_put(md);
855
	dm_destroy(md);
L
Linus Torvalds 已提交
856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874
	return 0;
}

/*
 * Check a string doesn't overrun the chunk of
 * memory we copied from userland.
 */
static int invalid_str(char *str, void *end)
{
	while ((void *) str < end)
		if (!*str++)
			return 0;

	return -EINVAL;
}

static int dev_rename(struct dm_ioctl *param, size_t param_size)
{
	int r;
875
	char *new_data = (char *) param + param->data_start;
876
	struct mapped_device *md;
877
	unsigned change_uuid = (param->flags & DM_UUID_FLAG) ? 1 : 0;
L
Linus Torvalds 已提交
878

879
	if (new_data < param->data ||
880
	    invalid_str(new_data, (void *) param + param_size) || !*new_data ||
881 882
	    strlen(new_data) > (change_uuid ? DM_UUID_LEN - 1 : DM_NAME_LEN - 1)) {
		DMWARN("Invalid new mapped device name or uuid string supplied.");
L
Linus Torvalds 已提交
883 884 885
		return -EINVAL;
	}

886 887 888 889 890
	if (!change_uuid) {
		r = check_name(new_data);
		if (r)
			return r;
	}
L
Linus Torvalds 已提交
891

892
	md = dm_hash_rename(param, new_data);
893 894
	if (IS_ERR(md))
		return PTR_ERR(md);
895

896 897 898 899
	__dev_status(md, param);
	dm_put(md);

	return 0;
L
Linus Torvalds 已提交
900 901
}

D
Darrick J. Wong 已提交
902 903 904 905 906 907 908
static int dev_set_geometry(struct dm_ioctl *param, size_t param_size)
{
	int r = -EINVAL, x;
	struct mapped_device *md;
	struct hd_geometry geometry;
	unsigned long indata[4];
	char *geostr = (char *) param + param->data_start;
909
	char dummy;
D
Darrick J. Wong 已提交
910 911 912 913 914

	md = find_device(param);
	if (!md)
		return -ENXIO;

A
Alasdair G Kergon 已提交
915
	if (geostr < param->data ||
D
Darrick J. Wong 已提交
916 917 918 919 920
	    invalid_str(geostr, (void *) param + param_size)) {
		DMWARN("Invalid geometry supplied.");
		goto out;
	}

921 922
	x = sscanf(geostr, "%lu %lu %lu %lu%c", indata,
		   indata + 1, indata + 2, indata + 3, &dummy);
D
Darrick J. Wong 已提交
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

	if (x != 4) {
		DMWARN("Unable to interpret geometry settings.");
		goto out;
	}

	if (indata[0] > 65535 || indata[1] > 255 ||
	    indata[2] > 255 || indata[3] > ULONG_MAX) {
		DMWARN("Geometry exceeds range limits.");
		goto out;
	}

	geometry.cylinders = indata[0];
	geometry.heads = indata[1];
	geometry.sectors = indata[2];
	geometry.start = indata[3];

	r = dm_set_geometry(md, &geometry);

	param->data_size = 0;

out:
	dm_put(md);
	return r;
}

L
Linus Torvalds 已提交
949 950 951
static int do_suspend(struct dm_ioctl *param)
{
	int r = 0;
952
	unsigned suspend_flags = DM_SUSPEND_LOCKFS_FLAG;
L
Linus Torvalds 已提交
953 954 955 956 957 958
	struct mapped_device *md;

	md = find_device(param);
	if (!md)
		return -ENXIO;

959
	if (param->flags & DM_SKIP_LOCKFS_FLAG)
960
		suspend_flags &= ~DM_SUSPEND_LOCKFS_FLAG;
961 962
	if (param->flags & DM_NOFLUSH_FLAG)
		suspend_flags |= DM_SUSPEND_NOFLUSH_FLAG;
963

964
	if (!dm_suspended_md(md)) {
965
		r = dm_suspend(md, suspend_flags);
966 967 968
		if (r)
			goto out;
	}
L
Linus Torvalds 已提交
969

970
	__dev_status(md, param);
L
Linus Torvalds 已提交
971

972
out:
L
Linus Torvalds 已提交
973
	dm_put(md);
974

L
Linus Torvalds 已提交
975 976 977 978 979 980
	return r;
}

static int do_resume(struct dm_ioctl *param)
{
	int r = 0;
981
	unsigned suspend_flags = DM_SUSPEND_LOCKFS_FLAG;
L
Linus Torvalds 已提交
982 983
	struct hash_cell *hc;
	struct mapped_device *md;
984
	struct dm_table *new_map, *old_map = NULL;
L
Linus Torvalds 已提交
985 986 987 988 989

	down_write(&_hash_lock);

	hc = __find_device_hash_cell(param);
	if (!hc) {
990
		DMDEBUG_LIMIT("device doesn't appear to be in the dev hash table.");
L
Linus Torvalds 已提交
991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005
		up_write(&_hash_lock);
		return -ENXIO;
	}

	md = hc->md;

	new_map = hc->new_map;
	hc->new_map = NULL;
	param->flags &= ~DM_INACTIVE_PRESENT_FLAG;

	up_write(&_hash_lock);

	/* Do we need to load a new map ? */
	if (new_map) {
		/* Suspend if it isn't already suspended */
1006
		if (param->flags & DM_SKIP_LOCKFS_FLAG)
1007
			suspend_flags &= ~DM_SUSPEND_LOCKFS_FLAG;
1008 1009
		if (param->flags & DM_NOFLUSH_FLAG)
			suspend_flags |= DM_SUSPEND_NOFLUSH_FLAG;
1010
		if (!dm_suspended_md(md))
1011
			dm_suspend(md, suspend_flags);
L
Linus Torvalds 已提交
1012

1013 1014
		old_map = dm_swap_table(md, new_map);
		if (IS_ERR(old_map)) {
M
Mikulas Patocka 已提交
1015
			dm_sync_table(md);
1016
			dm_table_destroy(new_map);
L
Linus Torvalds 已提交
1017
			dm_put(md);
1018
			return PTR_ERR(old_map);
L
Linus Torvalds 已提交
1019 1020 1021 1022 1023 1024 1025 1026
		}

		if (dm_table_get_mode(new_map) & FMODE_WRITE)
			set_disk_ro(dm_disk(md), 0);
		else
			set_disk_ro(dm_disk(md), 1);
	}

1027
	if (dm_suspended_md(md)) {
L
Linus Torvalds 已提交
1028
		r = dm_resume(md);
1029 1030
		if (!r && !dm_kobject_uevent(md, KOBJ_CHANGE, param->event_nr))
			param->flags |= DM_UEVENT_GENERATED_FLAG;
1031
	}
L
Linus Torvalds 已提交
1032

M
Mikulas Patocka 已提交
1033 1034 1035 1036
	/*
	 * Since dm_swap_table synchronizes RCU, nobody should be in
	 * read-side critical section already.
	 */
1037 1038
	if (old_map)
		dm_table_destroy(old_map);
M
Milan Broz 已提交
1039

1040
	if (!r)
1041
		__dev_status(md, param);
L
Linus Torvalds 已提交
1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070

	dm_put(md);
	return r;
}

/*
 * Set or unset the suspension state of a device.
 * If the device already is in the requested state we just return its status.
 */
static int dev_suspend(struct dm_ioctl *param, size_t param_size)
{
	if (param->flags & DM_SUSPEND_FLAG)
		return do_suspend(param);

	return do_resume(param);
}

/*
 * Copies device info back to user space, used by
 * the create and info ioctls.
 */
static int dev_status(struct dm_ioctl *param, size_t param_size)
{
	struct mapped_device *md;

	md = find_device(param);
	if (!md)
		return -ENXIO;

1071
	__dev_status(md, param);
L
Linus Torvalds 已提交
1072
	dm_put(md);
1073 1074

	return 0;
L
Linus Torvalds 已提交
1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087
}

/*
 * Build up the status struct for each target
 */
static void retrieve_status(struct dm_table *table,
			    struct dm_ioctl *param, size_t param_size)
{
	unsigned int i, num_targets;
	struct dm_target_spec *spec;
	char *outbuf, *outptr;
	status_type_t type;
	size_t remaining, len, used = 0;
1088
	unsigned status_flags = 0;
L
Linus Torvalds 已提交
1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100

	outptr = outbuf = get_result_buffer(param, param_size, &len);

	if (param->flags & DM_STATUS_TABLE_FLAG)
		type = STATUSTYPE_TABLE;
	else
		type = STATUSTYPE_INFO;

	/* Get all the target info */
	num_targets = dm_table_get_num_targets(table);
	for (i = 0; i < num_targets; i++) {
		struct dm_target *ti = dm_table_get_target(table, i);
1101
		size_t l;
L
Linus Torvalds 已提交
1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125

		remaining = len - (outptr - outbuf);
		if (remaining <= sizeof(struct dm_target_spec)) {
			param->flags |= DM_BUFFER_FULL_FLAG;
			break;
		}

		spec = (struct dm_target_spec *) outptr;

		spec->status = 0;
		spec->sector_start = ti->begin;
		spec->length = ti->len;
		strncpy(spec->target_type, ti->type->name,
			sizeof(spec->target_type));

		outptr += sizeof(struct dm_target_spec);
		remaining = len - (outptr - outbuf);
		if (remaining <= 0) {
			param->flags |= DM_BUFFER_FULL_FLAG;
			break;
		}

		/* Get the status/table string from the target driver */
		if (ti->type->status) {
1126 1127
			if (param->flags & DM_NOFLUSH_FLAG)
				status_flags |= DM_STATUS_NOFLUSH_FLAG;
1128
			ti->type->status(ti, type, status_flags, outptr, remaining);
L
Linus Torvalds 已提交
1129 1130 1131
		} else
			outptr[0] = '\0';

1132 1133 1134 1135 1136 1137 1138
		l = strlen(outptr) + 1;
		if (l == remaining) {
			param->flags |= DM_BUFFER_FULL_FLAG;
			break;
		}

		outptr += l;
L
Linus Torvalds 已提交
1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155
		used = param->data_start + (outptr - outbuf);

		outptr = align_ptr(outptr);
		spec->next = outptr - outbuf;
	}

	if (used)
		param->data_size = used;

	param->target_count = num_targets;
}

/*
 * Wait for a device to report an event
 */
static int dev_wait(struct dm_ioctl *param, size_t param_size)
{
1156
	int r = 0;
L
Linus Torvalds 已提交
1157 1158
	struct mapped_device *md;
	struct dm_table *table;
M
Mikulas Patocka 已提交
1159
	int srcu_idx;
L
Linus Torvalds 已提交
1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177

	md = find_device(param);
	if (!md)
		return -ENXIO;

	/*
	 * Wait for a notification event
	 */
	if (dm_wait_event(md, param->event_nr)) {
		r = -ERESTARTSYS;
		goto out;
	}

	/*
	 * The userland program is going to want to know what
	 * changed to trigger the event, so we may as well tell
	 * him and save an ioctl.
	 */
1178
	__dev_status(md, param);
L
Linus Torvalds 已提交
1179

M
Mikulas Patocka 已提交
1180 1181
	table = dm_get_live_or_inactive_table(md, param, &srcu_idx);
	if (table)
L
Linus Torvalds 已提交
1182
		retrieve_status(table, param, param_size);
M
Mikulas Patocka 已提交
1183
	dm_put_live_table(md, srcu_idx);
L
Linus Torvalds 已提交
1184

1185
out:
L
Linus Torvalds 已提交
1186
	dm_put(md);
1187

L
Linus Torvalds 已提交
1188 1189 1190
	return r;
}

1191
static inline fmode_t get_mode(struct dm_ioctl *param)
L
Linus Torvalds 已提交
1192
{
1193
	fmode_t mode = FMODE_READ | FMODE_WRITE;
L
Linus Torvalds 已提交
1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254

	if (param->flags & DM_READONLY_FLAG)
		mode = FMODE_READ;

	return mode;
}

static int next_target(struct dm_target_spec *last, uint32_t next, void *end,
		       struct dm_target_spec **spec, char **target_params)
{
	*spec = (struct dm_target_spec *) ((unsigned char *) last + next);
	*target_params = (char *) (*spec + 1);

	if (*spec < (last + 1))
		return -EINVAL;

	return invalid_str(*target_params, end);
}

static int populate_table(struct dm_table *table,
			  struct dm_ioctl *param, size_t param_size)
{
	int r;
	unsigned int i = 0;
	struct dm_target_spec *spec = (struct dm_target_spec *) param;
	uint32_t next = param->data_start;
	void *end = (void *) param + param_size;
	char *target_params;

	if (!param->target_count) {
		DMWARN("populate_table: no targets specified");
		return -EINVAL;
	}

	for (i = 0; i < param->target_count; i++) {

		r = next_target(spec, next, end, &spec, &target_params);
		if (r) {
			DMWARN("unable to find target");
			return r;
		}

		r = dm_table_add_target(table, spec->target_type,
					(sector_t) spec->sector_start,
					(sector_t) spec->length,
					target_params);
		if (r) {
			DMWARN("error adding target to table");
			return r;
		}

		next = spec->next;
	}

	return dm_table_complete(table);
}

static int table_load(struct dm_ioctl *param, size_t param_size)
{
	int r;
	struct hash_cell *hc;
M
Mikulas Patocka 已提交
1255
	struct dm_table *t, *old_map = NULL;
M
Mike Anderson 已提交
1256
	struct mapped_device *md;
1257
	struct target_type *immutable_target_type;
M
Mike Anderson 已提交
1258 1259 1260 1261

	md = find_device(param);
	if (!md)
		return -ENXIO;
L
Linus Torvalds 已提交
1262

M
Mike Anderson 已提交
1263
	r = dm_table_create(&t, get_mode(param), param->target_count, md);
L
Linus Torvalds 已提交
1264
	if (r)
1265
		goto err;
L
Linus Torvalds 已提交
1266

1267 1268
	/* Protect md->type and md->queue against concurrent table loads. */
	dm_lock_md_type(md);
L
Linus Torvalds 已提交
1269
	r = populate_table(t, param, param_size);
1270 1271
	if (r)
		goto err_unlock_md_type;
L
Linus Torvalds 已提交
1272

1273 1274 1275 1276 1277 1278
	immutable_target_type = dm_get_immutable_target_type(md);
	if (immutable_target_type &&
	    (immutable_target_type != dm_table_get_immutable_target_type(t))) {
		DMWARN("can't replace immutable target type %s",
		       immutable_target_type->name);
		r = -EINVAL;
1279
		goto err_unlock_md_type;
1280 1281
	}

1282 1283 1284 1285 1286 1287
	if (dm_get_md_type(md) == DM_TYPE_NONE)
		/* Initial table load: acquire type of table. */
		dm_set_md_type(md, dm_table_get_type(t));
	else if (dm_get_md_type(md) != dm_table_get_type(t)) {
		DMWARN("can't change device type after initial table load.");
		r = -EINVAL;
1288
		goto err_unlock_md_type;
1289
	}
1290 1291 1292 1293 1294

	/* setup md->queue to reflect md's type (may block) */
	r = dm_setup_md_queue(md);
	if (r) {
		DMWARN("unable to set up device queue for new table.");
1295
		goto err_unlock_md_type;
1296
	}
1297 1298 1299
	dm_unlock_md_type(md);

	/* stage inactive table */
L
Linus Torvalds 已提交
1300
	down_write(&_hash_lock);
M
Mike Anderson 已提交
1301 1302 1303 1304 1305
	hc = dm_get_mdptr(md);
	if (!hc || hc->md != md) {
		DMWARN("device has been removed from the dev hash table.");
		up_write(&_hash_lock);
		r = -ENXIO;
1306
		goto err_destroy_table;
L
Linus Torvalds 已提交
1307 1308 1309
	}

	if (hc->new_map)
M
Mikulas Patocka 已提交
1310
		old_map = hc->new_map;
L
Linus Torvalds 已提交
1311
	hc->new_map = t;
M
Mike Anderson 已提交
1312 1313
	up_write(&_hash_lock);

L
Linus Torvalds 已提交
1314
	param->flags |= DM_INACTIVE_PRESENT_FLAG;
1315
	__dev_status(md, param);
M
Mike Anderson 已提交
1316

M
Mikulas Patocka 已提交
1317 1318 1319 1320 1321
	if (old_map) {
		dm_sync_table(md);
		dm_table_destroy(old_map);
	}

M
Mike Anderson 已提交
1322
	dm_put(md);
L
Linus Torvalds 已提交
1323

1324 1325 1326 1327 1328 1329 1330 1331 1332
	return 0;

err_unlock_md_type:
	dm_unlock_md_type(md);
err_destroy_table:
	dm_table_destroy(t);
err:
	dm_put(md);

L
Linus Torvalds 已提交
1333 1334 1335 1336 1337 1338
	return r;
}

static int table_clear(struct dm_ioctl *param, size_t param_size)
{
	struct hash_cell *hc;
1339
	struct mapped_device *md;
M
Mikulas Patocka 已提交
1340
	struct dm_table *old_map = NULL;
L
Linus Torvalds 已提交
1341 1342 1343 1344 1345

	down_write(&_hash_lock);

	hc = __find_device_hash_cell(param);
	if (!hc) {
1346
		DMDEBUG_LIMIT("device doesn't appear to be in the dev hash table.");
L
Linus Torvalds 已提交
1347 1348 1349 1350 1351
		up_write(&_hash_lock);
		return -ENXIO;
	}

	if (hc->new_map) {
M
Mikulas Patocka 已提交
1352
		old_map = hc->new_map;
L
Linus Torvalds 已提交
1353 1354 1355 1356 1357
		hc->new_map = NULL;
	}

	param->flags &= ~DM_INACTIVE_PRESENT_FLAG;

1358
	__dev_status(hc->md, param);
1359
	md = hc->md;
L
Linus Torvalds 已提交
1360
	up_write(&_hash_lock);
M
Mikulas Patocka 已提交
1361 1362 1363 1364
	if (old_map) {
		dm_sync_table(md);
		dm_table_destroy(old_map);
	}
1365
	dm_put(md);
1366 1367

	return 0;
L
Linus Torvalds 已提交
1368 1369 1370 1371 1372 1373 1374 1375 1376 1377 1378
}

/*
 * Retrieves a list of devices used by a particular dm device.
 */
static void retrieve_deps(struct dm_table *table,
			  struct dm_ioctl *param, size_t param_size)
{
	unsigned int count = 0;
	struct list_head *tmp;
	size_t len, needed;
M
Mikulas Patocka 已提交
1379
	struct dm_dev_internal *dd;
L
Linus Torvalds 已提交
1380 1381 1382 1383 1384 1385 1386 1387 1388 1389 1390 1391 1392 1393 1394 1395 1396 1397 1398 1399 1400 1401 1402 1403 1404
	struct dm_target_deps *deps;

	deps = get_result_buffer(param, param_size, &len);

	/*
	 * Count the devices.
	 */
	list_for_each (tmp, dm_table_get_devices(table))
		count++;

	/*
	 * Check we have enough space.
	 */
	needed = sizeof(*deps) + (sizeof(*deps->dev) * count);
	if (len < needed) {
		param->flags |= DM_BUFFER_FULL_FLAG;
		return;
	}

	/*
	 * Fill in the devices.
	 */
	deps->count = count;
	count = 0;
	list_for_each_entry (dd, dm_table_get_devices(table), list)
M
Mikulas Patocka 已提交
1405
		deps->dev[count++] = huge_encode_dev(dd->dm_dev.bdev->bd_dev);
L
Linus Torvalds 已提交
1406 1407 1408 1409 1410 1411 1412 1413

	param->data_size = param->data_start + needed;
}

static int table_deps(struct dm_ioctl *param, size_t param_size)
{
	struct mapped_device *md;
	struct dm_table *table;
M
Mikulas Patocka 已提交
1414
	int srcu_idx;
L
Linus Torvalds 已提交
1415 1416 1417 1418 1419

	md = find_device(param);
	if (!md)
		return -ENXIO;

1420
	__dev_status(md, param);
L
Linus Torvalds 已提交
1421

M
Mikulas Patocka 已提交
1422 1423
	table = dm_get_live_or_inactive_table(md, param, &srcu_idx);
	if (table)
L
Linus Torvalds 已提交
1424
		retrieve_deps(table, param, param_size);
M
Mikulas Patocka 已提交
1425
	dm_put_live_table(md, srcu_idx);
L
Linus Torvalds 已提交
1426 1427

	dm_put(md);
1428 1429

	return 0;
L
Linus Torvalds 已提交
1430 1431 1432 1433 1434 1435 1436 1437 1438 1439
}

/*
 * Return the status of a device as a text string for each
 * target.
 */
static int table_status(struct dm_ioctl *param, size_t param_size)
{
	struct mapped_device *md;
	struct dm_table *table;
M
Mikulas Patocka 已提交
1440
	int srcu_idx;
L
Linus Torvalds 已提交
1441 1442 1443 1444 1445

	md = find_device(param);
	if (!md)
		return -ENXIO;

1446
	__dev_status(md, param);
L
Linus Torvalds 已提交
1447

M
Mikulas Patocka 已提交
1448 1449
	table = dm_get_live_or_inactive_table(md, param, &srcu_idx);
	if (table)
L
Linus Torvalds 已提交
1450
		retrieve_status(table, param, param_size);
M
Mikulas Patocka 已提交
1451
	dm_put_live_table(md, srcu_idx);
L
Linus Torvalds 已提交
1452 1453

	dm_put(md);
1454 1455

	return 0;
L
Linus Torvalds 已提交
1456 1457
}

1458
/*
M
Mikulas Patocka 已提交
1459 1460
 * Process device-mapper dependent messages.  Messages prefixed with '@'
 * are processed by the DM core.  All others are delivered to the target.
1461 1462 1463 1464 1465 1466
 * Returns a number <= 1 if message was processed by device mapper.
 * Returns 2 if message should be delivered to the target.
 */
static int message_for_md(struct mapped_device *md, unsigned argc, char **argv,
			  char *result, unsigned maxlen)
{
M
Mikulas Patocka 已提交
1467 1468 1469 1470 1471 1472 1473 1474 1475 1476 1477
	int r;

	if (**argv != '@')
		return 2; /* no '@' prefix, deliver to target */

	r = dm_stats_message(md, argc, argv, result, maxlen);
	if (r < 2)
		return r;

	DMERR("Unsupported message sent to DM core: %s", argv[0]);
	return -EINVAL;
1478 1479
}

L
Linus Torvalds 已提交
1480 1481 1482 1483 1484 1485 1486 1487 1488 1489 1490
/*
 * Pass a message to the target that's at the supplied device offset.
 */
static int target_message(struct dm_ioctl *param, size_t param_size)
{
	int r, argc;
	char **argv;
	struct mapped_device *md;
	struct dm_table *table;
	struct dm_target *ti;
	struct dm_target_msg *tmsg = (void *) param + param->data_start;
1491 1492
	size_t maxlen;
	char *result = get_result_buffer(param, param_size, &maxlen);
M
Mikulas Patocka 已提交
1493
	int srcu_idx;
L
Linus Torvalds 已提交
1494 1495 1496 1497 1498

	md = find_device(param);
	if (!md)
		return -ENXIO;

M
Milan Broz 已提交
1499
	if (tmsg < (struct dm_target_msg *) param->data ||
L
Linus Torvalds 已提交
1500 1501 1502 1503 1504 1505 1506 1507 1508 1509 1510 1511
	    invalid_str(tmsg->message, (void *) param + param_size)) {
		DMWARN("Invalid target message parameters.");
		r = -EINVAL;
		goto out;
	}

	r = dm_split_args(&argc, &argv, tmsg->message);
	if (r) {
		DMWARN("Failed to split target message parameters");
		goto out;
	}

1512 1513
	if (!argc) {
		DMWARN("Empty message received.");
1514
		goto out_argv;
1515 1516
	}

1517 1518 1519 1520
	r = message_for_md(md, argc, argv, result, maxlen);
	if (r <= 1)
		goto out_argv;

M
Mikulas Patocka 已提交
1521
	table = dm_get_live_table(md, &srcu_idx);
L
Linus Torvalds 已提交
1522
	if (!table)
M
Mikulas Patocka 已提交
1523
		goto out_table;
L
Linus Torvalds 已提交
1524

1525 1526 1527 1528 1529
	if (dm_deleting_md(md)) {
		r = -ENXIO;
		goto out_table;
	}

1530 1531
	ti = dm_table_find_target(table, tmsg->sector);
	if (!dm_target_is_valid(ti)) {
L
Linus Torvalds 已提交
1532 1533
		DMWARN("Target message sector outside device.");
		r = -EINVAL;
1534
	} else if (ti->type->message)
L
Linus Torvalds 已提交
1535 1536 1537 1538 1539 1540
		r = ti->type->message(ti, argc, argv);
	else {
		DMWARN("Target type does not support messages");
		r = -EINVAL;
	}

1541
 out_table:
M
Mikulas Patocka 已提交
1542
	dm_put_live_table(md, srcu_idx);
L
Linus Torvalds 已提交
1543 1544 1545
 out_argv:
	kfree(argv);
 out:
1546 1547 1548 1549 1550
	if (r >= 0)
		__dev_status(md, param);

	if (r == 1) {
		param->flags |= DM_DATA_OUT_FLAG;
M
Mikulas Patocka 已提交
1551
		if (dm_message_test_buffer_overflow(result, maxlen))
1552 1553 1554 1555 1556 1557
			param->flags |= DM_BUFFER_FULL_FLAG;
		else
			param->data_size = param->data_start + strlen(result) + 1;
		r = 0;
	}

L
Linus Torvalds 已提交
1558 1559 1560 1561
	dm_put(md);
	return r;
}

1562 1563 1564 1565 1566 1567 1568 1569
/*
 * The ioctl parameter block consists of two parts, a dm_ioctl struct
 * followed by a data buffer.  This flag is set if the second part,
 * which has a variable size, is not used by the function processing
 * the ioctl.
 */
#define IOCTL_FLAGS_NO_PARAMS	1

L
Linus Torvalds 已提交
1570 1571 1572 1573
/*-----------------------------------------------------------------
 * Implementation of open/close/ioctl on the special char
 * device.
 *---------------------------------------------------------------*/
1574
static ioctl_fn lookup_ioctl(unsigned int cmd, int *ioctl_flags)
L
Linus Torvalds 已提交
1575 1576 1577
{
	static struct {
		int cmd;
1578
		int flags;
L
Linus Torvalds 已提交
1579 1580
		ioctl_fn fn;
	} _ioctls[] = {
1581 1582 1583 1584 1585 1586 1587 1588 1589 1590 1591 1592 1593 1594 1595 1596 1597 1598 1599 1600
		{DM_VERSION_CMD, 0, NULL}, /* version is dealt with elsewhere */
		{DM_REMOVE_ALL_CMD, IOCTL_FLAGS_NO_PARAMS, remove_all},
		{DM_LIST_DEVICES_CMD, 0, list_devices},

		{DM_DEV_CREATE_CMD, IOCTL_FLAGS_NO_PARAMS, dev_create},
		{DM_DEV_REMOVE_CMD, IOCTL_FLAGS_NO_PARAMS, dev_remove},
		{DM_DEV_RENAME_CMD, 0, dev_rename},
		{DM_DEV_SUSPEND_CMD, IOCTL_FLAGS_NO_PARAMS, dev_suspend},
		{DM_DEV_STATUS_CMD, IOCTL_FLAGS_NO_PARAMS, dev_status},
		{DM_DEV_WAIT_CMD, 0, dev_wait},

		{DM_TABLE_LOAD_CMD, 0, table_load},
		{DM_TABLE_CLEAR_CMD, IOCTL_FLAGS_NO_PARAMS, table_clear},
		{DM_TABLE_DEPS_CMD, 0, table_deps},
		{DM_TABLE_STATUS_CMD, 0, table_status},

		{DM_LIST_VERSIONS_CMD, 0, list_versions},

		{DM_TARGET_MSG_CMD, 0, target_message},
		{DM_DEV_SET_GEOMETRY_CMD, 0, dev_set_geometry}
L
Linus Torvalds 已提交
1601 1602
	};

1603 1604 1605 1606 1607
	if (unlikely(cmd >= ARRAY_SIZE(_ioctls)))
		return NULL;

	*ioctl_flags = _ioctls[cmd].flags;
	return _ioctls[cmd].fn;
L
Linus Torvalds 已提交
1608 1609 1610 1611 1612 1613 1614 1615 1616 1617 1618 1619 1620 1621 1622 1623 1624 1625 1626 1627 1628 1629 1630 1631 1632 1633 1634 1635 1636 1637 1638 1639 1640 1641 1642 1643
}

/*
 * As well as checking the version compatibility this always
 * copies the kernel interface version out.
 */
static int check_version(unsigned int cmd, struct dm_ioctl __user *user)
{
	uint32_t version[3];
	int r = 0;

	if (copy_from_user(version, user->version, sizeof(version)))
		return -EFAULT;

	if ((DM_VERSION_MAJOR != version[0]) ||
	    (DM_VERSION_MINOR < version[1])) {
		DMWARN("ioctl interface mismatch: "
		       "kernel(%u.%u.%u), user(%u.%u.%u), cmd(%d)",
		       DM_VERSION_MAJOR, DM_VERSION_MINOR,
		       DM_VERSION_PATCHLEVEL,
		       version[0], version[1], version[2], cmd);
		r = -EINVAL;
	}

	/*
	 * Fill in the kernel version.
	 */
	version[0] = DM_VERSION_MAJOR;
	version[1] = DM_VERSION_MINOR;
	version[2] = DM_VERSION_PATCHLEVEL;
	if (copy_to_user(user->version, version, sizeof(version)))
		return -EFAULT;

	return r;
}

1644 1645
#define DM_PARAMS_KMALLOC	0x0001	/* Params alloced with kmalloc */
#define DM_PARAMS_VMALLOC	0x0002	/* Params alloced with vmalloc */
1646 1647 1648 1649 1650 1651 1652
#define DM_WIPE_BUFFER		0x0010	/* Wipe input buffer before returning from ioctl */

static void free_params(struct dm_ioctl *param, size_t param_size, int param_flags)
{
	if (param_flags & DM_WIPE_BUFFER)
		memset(param, 0, param_size);

1653 1654
	if (param_flags & DM_PARAMS_KMALLOC)
		kfree(param);
1655 1656 1657 1658
	if (param_flags & DM_PARAMS_VMALLOC)
		vfree(param);
}

1659 1660 1661
static int copy_params(struct dm_ioctl __user *user, struct dm_ioctl *param_kernel,
		       int ioctl_flags,
		       struct dm_ioctl **param, int *param_flags)
L
Linus Torvalds 已提交
1662
{
1663
	struct dm_ioctl *dmi;
1664
	int secure_data;
1665
	const size_t minimum_data_size = sizeof(*param_kernel) - sizeof(param_kernel->data);
L
Linus Torvalds 已提交
1666

1667
	if (copy_from_user(param_kernel, user, minimum_data_size))
L
Linus Torvalds 已提交
1668 1669
		return -EFAULT;

1670
	if (param_kernel->data_size < minimum_data_size)
L
Linus Torvalds 已提交
1671 1672
		return -EINVAL;

1673
	secure_data = param_kernel->flags & DM_SECURE_DATA_FLAG;
1674

1675 1676
	*param_flags = secure_data ? DM_WIPE_BUFFER : 0;

1677 1678 1679 1680 1681 1682
	if (ioctl_flags & IOCTL_FLAGS_NO_PARAMS) {
		dmi = param_kernel;
		dmi->data_size = minimum_data_size;
		goto data_copied;
	}

M
Mikulas Patocka 已提交
1683 1684
	/*
	 * Try to avoid low memory issues when a device is suspended.
1685
	 * Use kmalloc() rather than vmalloc() when we can.
M
Mikulas Patocka 已提交
1686
	 */
1687
	dmi = NULL;
1688 1689 1690 1691 1692
	if (param_kernel->data_size <= KMALLOC_MAX_SIZE) {
		dmi = kmalloc(param_kernel->data_size, GFP_NOIO | __GFP_NORETRY | __GFP_NOMEMALLOC | __GFP_NOWARN);
		if (dmi)
			*param_flags |= DM_PARAMS_KMALLOC;
	}
1693 1694

	if (!dmi) {
1695 1696
		unsigned noio_flag;
		noio_flag = memalloc_noio_save();
1697
		dmi = __vmalloc(param_kernel->data_size, GFP_NOIO | __GFP_REPEAT | __GFP_HIGH | __GFP_HIGHMEM, PAGE_KERNEL);
1698
		memalloc_noio_restore(noio_flag);
1699 1700
		if (dmi)
			*param_flags |= DM_PARAMS_VMALLOC;
1701 1702
	}

1703
	if (!dmi) {
1704
		if (secure_data && clear_user(user, param_kernel->data_size))
1705
			return -EFAULT;
L
Linus Torvalds 已提交
1706
		return -ENOMEM;
1707
	}
L
Linus Torvalds 已提交
1708

1709
	if (copy_from_user(dmi, user, param_kernel->data_size))
1710
		goto bad;
L
Linus Torvalds 已提交
1711

1712
data_copied:
1713 1714 1715
	/*
	 * Abort if something changed the ioctl data while it was being copied.
	 */
1716
	if (dmi->data_size != param_kernel->data_size) {
1717 1718 1719 1720
		DMERR("rejecting ioctl: data size modified while processing parameters");
		goto bad;
	}

1721
	/* Wipe the user buffer so we do not return it to userspace */
1722
	if (secure_data && clear_user(user, param_kernel->data_size))
1723 1724
		goto bad;

L
Linus Torvalds 已提交
1725 1726
	*param = dmi;
	return 0;
1727 1728

bad:
1729
	free_params(dmi, param_kernel->data_size, *param_flags);
1730

1731
	return -EFAULT;
L
Linus Torvalds 已提交
1732 1733 1734 1735 1736 1737
}

static int validate_params(uint cmd, struct dm_ioctl *param)
{
	/* Always clear this flag */
	param->flags &= ~DM_BUFFER_FULL_FLAG;
1738
	param->flags &= ~DM_UEVENT_GENERATED_FLAG;
1739
	param->flags &= ~DM_SECURE_DATA_FLAG;
1740
	param->flags &= ~DM_DATA_OUT_FLAG;
L
Linus Torvalds 已提交
1741 1742 1743 1744 1745 1746 1747 1748 1749 1750 1751 1752 1753 1754 1755 1756 1757 1758 1759 1760 1761 1762 1763 1764

	/* Ignores parameters */
	if (cmd == DM_REMOVE_ALL_CMD ||
	    cmd == DM_LIST_DEVICES_CMD ||
	    cmd == DM_LIST_VERSIONS_CMD)
		return 0;

	if ((cmd == DM_DEV_CREATE_CMD)) {
		if (!*param->name) {
			DMWARN("name not supplied when creating device");
			return -EINVAL;
		}
	} else if ((*param->uuid && *param->name)) {
		DMWARN("only supply one of name or uuid, cmd(%u)", cmd);
		return -EINVAL;
	}

	/* Ensure strings are terminated */
	param->name[DM_NAME_LEN - 1] = '\0';
	param->uuid[DM_UUID_LEN - 1] = '\0';

	return 0;
}

A
Alasdair G Kergon 已提交
1765
static int ctl_ioctl(uint command, struct dm_ioctl __user *user)
L
Linus Torvalds 已提交
1766 1767
{
	int r = 0;
1768
	int ioctl_flags;
1769
	int param_flags;
L
Linus Torvalds 已提交
1770
	unsigned int cmd;
A
Andrew Morton 已提交
1771
	struct dm_ioctl *uninitialized_var(param);
L
Linus Torvalds 已提交
1772
	ioctl_fn fn = NULL;
1773
	size_t input_param_size;
1774
	struct dm_ioctl param_kernel;
L
Linus Torvalds 已提交
1775 1776 1777 1778 1779 1780 1781 1782 1783 1784 1785 1786 1787 1788 1789 1790 1791 1792 1793 1794 1795 1796 1797 1798

	/* only root can play with this */
	if (!capable(CAP_SYS_ADMIN))
		return -EACCES;

	if (_IOC_TYPE(command) != DM_IOCTL)
		return -ENOTTY;

	cmd = _IOC_NR(command);

	/*
	 * Check the interface version passed in.  This also
	 * writes out the kernel's interface version.
	 */
	r = check_version(cmd, user);
	if (r)
		return r;

	/*
	 * Nothing more to do for the version command.
	 */
	if (cmd == DM_VERSION_CMD)
		return 0;

1799
	fn = lookup_ioctl(cmd, &ioctl_flags);
L
Linus Torvalds 已提交
1800 1801 1802 1803 1804 1805 1806 1807
	if (!fn) {
		DMWARN("dm_ctl_ioctl: unknown command 0x%x", command);
		return -ENOTTY;
	}

	/*
	 * Copy the parameters into kernel space.
	 */
1808
	r = copy_params(user, &param_kernel, ioctl_flags, &param, &param_flags);
L
Linus Torvalds 已提交
1809

1810 1811
	if (r)
		return r;
L
Linus Torvalds 已提交
1812

1813
	input_param_size = param->data_size;
L
Linus Torvalds 已提交
1814 1815 1816 1817 1818
	r = validate_params(cmd, param);
	if (r)
		goto out;

	param->data_size = sizeof(*param);
1819
	r = fn(param, input_param_size);
L
Linus Torvalds 已提交
1820

1821 1822 1823 1824
	if (unlikely(param->flags & DM_BUFFER_FULL_FLAG) &&
	    unlikely(ioctl_flags & IOCTL_FLAGS_NO_PARAMS))
		DMERR("ioctl %d tried to output some data but has IOCTL_FLAGS_NO_PARAMS set", cmd);

L
Linus Torvalds 已提交
1825 1826 1827 1828 1829 1830
	/*
	 * Copy the results back to userland.
	 */
	if (!r && copy_to_user(user, param, param->data_size))
		r = -EFAULT;

1831
out:
1832
	free_params(param, input_param_size, param_flags);
L
Linus Torvalds 已提交
1833 1834 1835
	return r;
}

A
Alasdair G Kergon 已提交
1836 1837 1838 1839 1840
static long dm_ctl_ioctl(struct file *file, uint command, ulong u)
{
	return (long)ctl_ioctl(command, (struct dm_ioctl __user *)u);
}

M
Milan Broz 已提交
1841 1842 1843 1844 1845 1846 1847 1848 1849
#ifdef CONFIG_COMPAT
static long dm_compat_ctl_ioctl(struct file *file, uint command, ulong u)
{
	return (long)dm_ctl_ioctl(file, command, (ulong) compat_ptr(u));
}
#else
#define dm_compat_ctl_ioctl NULL
#endif

1850
static const struct file_operations _ctl_fops = {
A
Arnd Bergmann 已提交
1851
	.open = nonseekable_open,
A
Alasdair G Kergon 已提交
1852
	.unlocked_ioctl	 = dm_ctl_ioctl,
M
Milan Broz 已提交
1853
	.compat_ioctl = dm_compat_ctl_ioctl,
L
Linus Torvalds 已提交
1854
	.owner	 = THIS_MODULE,
1855
	.llseek  = noop_llseek,
L
Linus Torvalds 已提交
1856 1857 1858
};

static struct miscdevice _dm_misc = {
P
Peter Rajnoha 已提交
1859
	.minor		= MAPPER_CTRL_MINOR,
L
Linus Torvalds 已提交
1860
	.name  		= DM_NAME,
P
Peter Rajnoha 已提交
1861
	.nodename	= DM_DIR "/" DM_CONTROL_NODE,
L
Linus Torvalds 已提交
1862 1863 1864
	.fops  		= &_ctl_fops
};

P
Peter Rajnoha 已提交
1865 1866 1867
MODULE_ALIAS_MISCDEV(MAPPER_CTRL_MINOR);
MODULE_ALIAS("devname:" DM_DIR "/" DM_CONTROL_NODE);

L
Linus Torvalds 已提交
1868 1869 1870 1871 1872 1873 1874 1875 1876 1877 1878 1879 1880 1881 1882 1883 1884 1885 1886 1887 1888 1889 1890 1891 1892 1893 1894 1895 1896 1897 1898
/*
 * Create misc character device and link to DM_DIR/control.
 */
int __init dm_interface_init(void)
{
	int r;

	r = dm_hash_init();
	if (r)
		return r;

	r = misc_register(&_dm_misc);
	if (r) {
		DMERR("misc_register failed for control device");
		dm_hash_exit();
		return r;
	}

	DMINFO("%d.%d.%d%s initialised: %s", DM_VERSION_MAJOR,
	       DM_VERSION_MINOR, DM_VERSION_PATCHLEVEL, DM_VERSION_EXTRA,
	       DM_DRIVER_EMAIL);
	return 0;
}

void dm_interface_exit(void)
{
	if (misc_deregister(&_dm_misc) < 0)
		DMERR("misc_deregister failed for control device");

	dm_hash_exit();
}
M
Mike Anderson 已提交
1899 1900 1901 1902 1903 1904 1905 1906 1907 1908 1909 1910 1911 1912 1913

/**
 * dm_copy_name_and_uuid - Copy mapped device name & uuid into supplied buffers
 * @md: Pointer to mapped_device
 * @name: Buffer (size DM_NAME_LEN) for name
 * @uuid: Buffer (size DM_UUID_LEN) for uuid or empty string if uuid not defined
 */
int dm_copy_name_and_uuid(struct mapped_device *md, char *name, char *uuid)
{
	int r = 0;
	struct hash_cell *hc;

	if (!md)
		return -ENXIO;

M
Mikulas Patocka 已提交
1914
	mutex_lock(&dm_hash_cells_mutex);
M
Mike Anderson 已提交
1915 1916 1917 1918 1919 1920
	hc = dm_get_mdptr(md);
	if (!hc || hc->md != md) {
		r = -ENXIO;
		goto out;
	}

1921 1922 1923 1924
	if (name)
		strcpy(name, hc->name);
	if (uuid)
		strcpy(uuid, hc->uuid ? : "");
M
Mike Anderson 已提交
1925 1926

out:
M
Mikulas Patocka 已提交
1927
	mutex_unlock(&dm_hash_cells_mutex);
M
Mike Anderson 已提交
1928 1929 1930

	return r;
}