genhd.c 22.8 KB
Newer Older
L
Linus Torvalds 已提交
1 2 3 4 5 6 7
/*
 *  gendisk handling
 */

#include <linux/module.h>
#include <linux/fs.h>
#include <linux/genhd.h>
8
#include <linux/kdev_t.h>
L
Linus Torvalds 已提交
9 10 11 12 13 14 15 16
#include <linux/kernel.h>
#include <linux/blkdev.h>
#include <linux/init.h>
#include <linux/spinlock.h>
#include <linux/seq_file.h>
#include <linux/slab.h>
#include <linux/kmod.h>
#include <linux/kobj_map.h>
17
#include <linux/buffer_head.h>
18
#include <linux/mutex.h>
L
Linus Torvalds 已提交
19

20 21
#include "blk.h"

22 23 24 25
static DEFINE_MUTEX(block_class_lock);
#ifndef CONFIG_SYSFS_DEPRECATED
struct kobject *block_depr;
#endif
L
Linus Torvalds 已提交
26

A
Adrian Bunk 已提交
27 28
static struct device_type disk_type;

29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180
/**
 * disk_get_part - get partition
 * @disk: disk to look partition from
 * @partno: partition number
 *
 * Look for partition @partno from @disk.  If found, increment
 * reference count and return it.
 *
 * CONTEXT:
 * Don't care.
 *
 * RETURNS:
 * Pointer to the found partition on success, NULL if not found.
 */
struct hd_struct *disk_get_part(struct gendisk *disk, int partno)
{
	struct hd_struct *part;

	if (unlikely(partno < 1 || partno > disk_max_parts(disk)))
		return NULL;
	rcu_read_lock();
	part = rcu_dereference(disk->__part[partno - 1]);
	if (part)
		get_device(&part->dev);
	rcu_read_unlock();

	return part;
}
EXPORT_SYMBOL_GPL(disk_get_part);

/**
 * disk_part_iter_init - initialize partition iterator
 * @piter: iterator to initialize
 * @disk: disk to iterate over
 * @flags: DISK_PITER_* flags
 *
 * Initialize @piter so that it iterates over partitions of @disk.
 *
 * CONTEXT:
 * Don't care.
 */
void disk_part_iter_init(struct disk_part_iter *piter, struct gendisk *disk,
			  unsigned int flags)
{
	piter->disk = disk;
	piter->part = NULL;

	if (flags & DISK_PITER_REVERSE)
		piter->idx = disk_max_parts(piter->disk) - 1;
	else
		piter->idx = 0;

	piter->flags = flags;
}
EXPORT_SYMBOL_GPL(disk_part_iter_init);

/**
 * disk_part_iter_next - proceed iterator to the next partition and return it
 * @piter: iterator of interest
 *
 * Proceed @piter to the next partition and return it.
 *
 * CONTEXT:
 * Don't care.
 */
struct hd_struct *disk_part_iter_next(struct disk_part_iter *piter)
{
	int inc, end;

	/* put the last partition */
	disk_put_part(piter->part);
	piter->part = NULL;

	rcu_read_lock();

	/* determine iteration parameters */
	if (piter->flags & DISK_PITER_REVERSE) {
		inc = -1;
		end = -1;
	} else {
		inc = 1;
		end = disk_max_parts(piter->disk);
	}

	/* iterate to the next partition */
	for (; piter->idx != end; piter->idx += inc) {
		struct hd_struct *part;

		part = rcu_dereference(piter->disk->__part[piter->idx]);
		if (!part)
			continue;
		if (!(piter->flags & DISK_PITER_INCL_EMPTY) && !part->nr_sects)
			continue;

		get_device(&part->dev);
		piter->part = part;
		piter->idx += inc;
		break;
	}

	rcu_read_unlock();

	return piter->part;
}
EXPORT_SYMBOL_GPL(disk_part_iter_next);

/**
 * disk_part_iter_exit - finish up partition iteration
 * @piter: iter of interest
 *
 * Called when iteration is over.  Cleans up @piter.
 *
 * CONTEXT:
 * Don't care.
 */
void disk_part_iter_exit(struct disk_part_iter *piter)
{
	disk_put_part(piter->part);
	piter->part = NULL;
}
EXPORT_SYMBOL_GPL(disk_part_iter_exit);

/**
 * disk_map_sector_rcu - map sector to partition
 * @disk: gendisk of interest
 * @sector: sector to map
 *
 * Find out which partition @sector maps to on @disk.  This is
 * primarily used for stats accounting.
 *
 * CONTEXT:
 * RCU read locked.  The returned partition pointer is valid only
 * while preemption is disabled.
 *
 * RETURNS:
 * Found partition on success, NULL if there's no matching partition.
 */
struct hd_struct *disk_map_sector_rcu(struct gendisk *disk, sector_t sector)
{
	int i;

	for (i = 0; i < disk_max_parts(disk); i++) {
		struct hd_struct *part = rcu_dereference(disk->__part[i]);

		if (part && part->start_sect <= sector &&
		    sector < part->start_sect + part->nr_sects)
			return part;
	}
	return NULL;
}
EXPORT_SYMBOL_GPL(disk_map_sector_rcu);

L
Linus Torvalds 已提交
181 182 183 184 185 186 187 188
/*
 * Can be deleted altogether. Later.
 *
 */
static struct blk_major_name {
	struct blk_major_name *next;
	int major;
	char name[16];
189
} *major_names[BLKDEV_MAJOR_HASH_SIZE];
L
Linus Torvalds 已提交
190 191 192 193

/* index in the above - for now: assume no multimajor ranges */
static inline int major_to_index(int major)
{
194
	return major % BLKDEV_MAJOR_HASH_SIZE;
195 196
}

197
#ifdef CONFIG_PROC_FS
198
void blkdev_show(struct seq_file *seqf, off_t offset)
199
{
200
	struct blk_major_name *dp;
201

202
	if (offset < BLKDEV_MAJOR_HASH_SIZE) {
203
		mutex_lock(&block_class_lock);
204
		for (dp = major_names[offset]; dp; dp = dp->next)
205
			seq_printf(seqf, "%3d %s\n", dp->major, dp->name);
206
		mutex_unlock(&block_class_lock);
L
Linus Torvalds 已提交
207 208
	}
}
209
#endif /* CONFIG_PROC_FS */
L
Linus Torvalds 已提交
210 211 212 213 214 215

int register_blkdev(unsigned int major, const char *name)
{
	struct blk_major_name **n, *p;
	int index, ret = 0;

216
	mutex_lock(&block_class_lock);
L
Linus Torvalds 已提交
217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260

	/* temporary */
	if (major == 0) {
		for (index = ARRAY_SIZE(major_names)-1; index > 0; index--) {
			if (major_names[index] == NULL)
				break;
		}

		if (index == 0) {
			printk("register_blkdev: failed to get major for %s\n",
			       name);
			ret = -EBUSY;
			goto out;
		}
		major = index;
		ret = major;
	}

	p = kmalloc(sizeof(struct blk_major_name), GFP_KERNEL);
	if (p == NULL) {
		ret = -ENOMEM;
		goto out;
	}

	p->major = major;
	strlcpy(p->name, name, sizeof(p->name));
	p->next = NULL;
	index = major_to_index(major);

	for (n = &major_names[index]; *n; n = &(*n)->next) {
		if ((*n)->major == major)
			break;
	}
	if (!*n)
		*n = p;
	else
		ret = -EBUSY;

	if (ret < 0) {
		printk("register_blkdev: cannot get major %d for %s\n",
		       major, name);
		kfree(p);
	}
out:
261
	mutex_unlock(&block_class_lock);
L
Linus Torvalds 已提交
262 263 264 265 266
	return ret;
}

EXPORT_SYMBOL(register_blkdev);

A
Akinobu Mita 已提交
267
void unregister_blkdev(unsigned int major, const char *name)
L
Linus Torvalds 已提交
268 269 270 271 272
{
	struct blk_major_name **n;
	struct blk_major_name *p = NULL;
	int index = major_to_index(major);

273
	mutex_lock(&block_class_lock);
L
Linus Torvalds 已提交
274 275 276
	for (n = &major_names[index]; *n; n = &(*n)->next)
		if ((*n)->major == major)
			break;
277 278 279
	if (!*n || strcmp((*n)->name, name)) {
		WARN_ON(1);
	} else {
L
Linus Torvalds 已提交
280 281 282
		p = *n;
		*n = p->next;
	}
283
	mutex_unlock(&block_class_lock);
L
Linus Torvalds 已提交
284 285 286 287 288 289 290 291 292 293 294 295
	kfree(p);
}

EXPORT_SYMBOL(unregister_blkdev);

static struct kobj_map *bdev_map;

/*
 * Register device numbers dev..(dev+range-1)
 * range must be nonzero
 * The hash chain is sorted on range, so that subranges can override.
 */
296
void blk_register_region(dev_t devt, unsigned long range, struct module *module,
L
Linus Torvalds 已提交
297 298 299
			 struct kobject *(*probe)(dev_t, int *, void *),
			 int (*lock)(dev_t, void *), void *data)
{
300
	kobj_map(bdev_map, devt, range, module, probe, lock, data);
L
Linus Torvalds 已提交
301 302 303 304
}

EXPORT_SYMBOL(blk_register_region);

305
void blk_unregister_region(dev_t devt, unsigned long range)
L
Linus Torvalds 已提交
306
{
307
	kobj_unmap(bdev_map, devt, range);
L
Linus Torvalds 已提交
308 309 310 311
}

EXPORT_SYMBOL(blk_unregister_region);

312
static struct kobject *exact_match(dev_t devt, int *partno, void *data)
L
Linus Torvalds 已提交
313 314
{
	struct gendisk *p = data;
315 316

	return &p->dev.kobj;
L
Linus Torvalds 已提交
317 318
}

319
static int exact_lock(dev_t devt, void *data)
L
Linus Torvalds 已提交
320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336
{
	struct gendisk *p = data;

	if (!get_disk(p))
		return -1;
	return 0;
}

/**
 * add_disk - add partitioning information to kernel list
 * @disk: per-device partitioning information
 *
 * This function registers the partitioning information in @disk
 * with the kernel.
 */
void add_disk(struct gendisk *disk)
{
337
	struct backing_dev_info *bdi;
338
	int retval;
339

L
Linus Torvalds 已提交
340
	disk->flags |= GENHD_FL_UP;
341 342 343
	disk->dev.devt = MKDEV(disk->major, disk->first_minor);
	blk_register_region(disk_devt(disk), disk->minors, NULL,
			    exact_match, exact_lock, disk);
L
Linus Torvalds 已提交
344 345
	register_disk(disk);
	blk_register_queue(disk);
346 347

	bdi = &disk->queue->backing_dev_info;
348
	bdi_register_dev(bdi, disk_devt(disk));
349 350
	retval = sysfs_create_link(&disk->dev.kobj, &bdi->dev->kobj, "bdi");
	WARN_ON(retval);
L
Linus Torvalds 已提交
351 352 353 354 355 356 357
}

EXPORT_SYMBOL(add_disk);
EXPORT_SYMBOL(del_gendisk);	/* in partitions/check.c */

void unlink_gendisk(struct gendisk *disk)
{
358 359
	sysfs_remove_link(&disk->dev.kobj, "bdi");
	bdi_unregister(&disk->queue->backing_dev_info);
L
Linus Torvalds 已提交
360
	blk_unregister_queue(disk);
361
	blk_unregister_region(disk_devt(disk), disk->minors);
L
Linus Torvalds 已提交
362 363 364 365
}

/**
 * get_gendisk - get partitioning information for a given device
366 367
 * @devt: device to get partitioning information for
 * @part: returned partition index
L
Linus Torvalds 已提交
368 369
 *
 * This function gets the structure containing partitioning
370
 * information for the given device @devt.
L
Linus Torvalds 已提交
371
 */
372
struct gendisk *get_gendisk(dev_t devt, int *partno)
L
Linus Torvalds 已提交
373
{
374
	struct kobject *kobj = kobj_lookup(bdev_map, devt, partno);
375 376 377
	struct device *dev = kobj_to_dev(kobj);

	return  kobj ? dev_to_disk(dev) : NULL;
L
Linus Torvalds 已提交
378 379
}

380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399
/**
 * bdget_disk - do bdget() by gendisk and partition number
 * @disk: gendisk of interest
 * @partno: partition number
 *
 * Find partition @partno from @disk, do bdget() on it.
 *
 * CONTEXT:
 * Don't care.
 *
 * RETURNS:
 * Resulting block_device on success, NULL on failure.
 */
extern struct block_device *bdget_disk(struct gendisk *disk, int partno)
{
	dev_t devt = MKDEV(0, 0);

	if (partno == 0)
		devt = disk_devt(disk);
	else {
400
		struct hd_struct *part;
401

402
		part = disk_get_part(disk, partno);
403 404
		if (part && part->nr_sects)
			devt = part_devt(part);
405
		disk_put_part(part);
406 407 408 409 410 411 412 413
	}

	if (likely(devt != MKDEV(0, 0)))
		return bdget(devt);
	return NULL;
}
EXPORT_SYMBOL(bdget_disk);

414 415 416 417 418 419 420
/*
 * print a full list of all partitions - intended for places where the root
 * filesystem can't be mounted and thus to give the victim some idea of what
 * went wrong
 */
void __init printk_all_partitions(void)
{
421 422 423 424 425 426
	struct class_dev_iter iter;
	struct device *dev;

	class_dev_iter_init(&iter, &block_class, NULL, &disk_type);
	while ((dev = class_dev_iter_next(&iter))) {
		struct gendisk *disk = dev_to_disk(dev);
427 428
		struct disk_part_iter piter;
		struct hd_struct *part;
429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444
		char buf[BDEVNAME_SIZE];

		/*
		 * Don't show empty devices or things that have been
		 * surpressed
		 */
		if (get_capacity(disk) == 0 ||
		    (disk->flags & GENHD_FL_SUPPRESS_PARTITION_INFO))
			continue;

		/*
		 * Note, unlike /proc/partitions, I am showing the
		 * numbers in hex - the same format as the root=
		 * option takes.
		 */
		printk("%02x%02x %10llu %s",
445
		       MAJOR(disk_devt(disk)), MINOR(disk_devt(disk)),
446 447 448 449 450 451 452 453 454 455
		       (unsigned long long)get_capacity(disk) >> 1,
		       disk_name(disk, 0, buf));
		if (disk->driverfs_dev != NULL &&
		    disk->driverfs_dev->driver != NULL)
			printk(" driver: %s\n",
			       disk->driverfs_dev->driver->name);
		else
			printk(" (driver?)\n");

		/* now show the partitions */
456 457
		disk_part_iter_init(&piter, disk, 0);
		while ((part = disk_part_iter_next(&piter)))
458
			printk("  %02x%02x %10llu %s\n",
459 460 461
			       MAJOR(part_devt(part)), MINOR(part_devt(part)),
			       (unsigned long long)part->nr_sects >> 1,
			       disk_name(disk, part->partno, buf));
462
		disk_part_iter_exit(&piter);
463 464
	}
	class_dev_iter_exit(&iter);
465 466
}

L
Linus Torvalds 已提交
467 468
#ifdef CONFIG_PROC_FS
/* iterator */
469
static void *disk_seqf_start(struct seq_file *seqf, loff_t *pos)
470
{
471 472 473
	loff_t skip = *pos;
	struct class_dev_iter *iter;
	struct device *dev;
474

475 476 477 478 479 480 481 482 483 484 485 486 487
	iter = kmalloc(GFP_KERNEL, sizeof(*iter));
	if (!iter)
		return ERR_PTR(-ENOMEM);

	seqf->private = iter;
	class_dev_iter_init(iter, &block_class, NULL, &disk_type);
	do {
		dev = class_dev_iter_next(iter);
		if (!dev)
			return NULL;
	} while (skip--);

	return dev_to_disk(dev);
488 489
}

490
static void *disk_seqf_next(struct seq_file *seqf, void *v, loff_t *pos)
L
Linus Torvalds 已提交
491
{
492
	struct device *dev;
L
Linus Torvalds 已提交
493

494 495
	(*pos)++;
	dev = class_dev_iter_next(seqf->private);
496
	if (dev)
497
		return dev_to_disk(dev);
498

L
Linus Torvalds 已提交
499 500 501
	return NULL;
}

502
static void disk_seqf_stop(struct seq_file *seqf, void *v)
503
{
504
	struct class_dev_iter *iter = seqf->private;
505

506 507 508 509
	/* stop is called even after start failed :-( */
	if (iter) {
		class_dev_iter_exit(iter);
		kfree(iter);
510
	}
L
Linus Torvalds 已提交
511 512
}

513
static void *show_partition_start(struct seq_file *seqf, loff_t *pos)
L
Linus Torvalds 已提交
514
{
515 516 517 518 519 520
	static void *p;

	p = disk_seqf_start(seqf, pos);
	if (!IS_ERR(p) && p)
		seq_puts(seqf, "major minor  #blocks  name\n\n");
	return p;
L
Linus Torvalds 已提交
521 522
}

523
static int show_partition(struct seq_file *seqf, void *v)
L
Linus Torvalds 已提交
524 525
{
	struct gendisk *sgp = v;
526 527
	struct disk_part_iter piter;
	struct hd_struct *part;
L
Linus Torvalds 已提交
528 529 530
	char buf[BDEVNAME_SIZE];

	/* Don't show non-partitionable removeable devices or empty devices */
531 532
	if (!get_capacity(sgp) || (!disk_max_parts(sgp) &&
				   (sgp->flags & GENHD_FL_REMOVABLE)))
L
Linus Torvalds 已提交
533 534 535 536 537
		return 0;
	if (sgp->flags & GENHD_FL_SUPPRESS_PARTITION_INFO)
		return 0;

	/* show the full disk and all non-0 size partitions of it */
538
	seq_printf(seqf, "%4d  %4d %10llu %s\n",
539
		MAJOR(disk_devt(sgp)), MINOR(disk_devt(sgp)),
L
Linus Torvalds 已提交
540 541
		(unsigned long long)get_capacity(sgp) >> 1,
		disk_name(sgp, 0, buf));
542 543 544

	disk_part_iter_init(&piter, sgp, 0);
	while ((part = disk_part_iter_next(&piter)))
545
		seq_printf(seqf, "%4d  %4d %10llu %s\n",
546 547 548
			   MAJOR(part_devt(part)), MINOR(part_devt(part)),
			   (unsigned long long)part->nr_sects >> 1,
			   disk_name(sgp, part->partno, buf));
549
	disk_part_iter_exit(&piter);
L
Linus Torvalds 已提交
550 551 552 553

	return 0;
}

554
const struct seq_operations partitions_op = {
555 556 557
	.start	= show_partition_start,
	.next	= disk_seqf_next,
	.stop	= disk_seqf_stop,
558
	.show	= show_partition
L
Linus Torvalds 已提交
559 560 561 562
};
#endif


563
static struct kobject *base_probe(dev_t devt, int *partno, void *data)
L
Linus Torvalds 已提交
564
{
565
	if (request_module("block-major-%d-%d", MAJOR(devt), MINOR(devt)) > 0)
L
Linus Torvalds 已提交
566
		/* Make old-style 2.4 aliases work */
567
		request_module("block-major-%d", MAJOR(devt));
L
Linus Torvalds 已提交
568 569 570 571 572
	return NULL;
}

static int __init genhd_device_init(void)
{
573 574 575 576
	int error;

	block_class.dev_kobj = sysfs_dev_block_kobj;
	error = class_register(&block_class);
R
Roland McGrath 已提交
577 578
	if (unlikely(error))
		return error;
579
	bdev_map = kobj_map_init(base_probe, &block_class_lock);
L
Linus Torvalds 已提交
580
	blk_dev_init();
581 582 583 584 585

#ifndef CONFIG_SYSFS_DEPRECATED
	/* create top-level block dir */
	block_depr = kobject_create_and_add("block", NULL);
#endif
586
	return 0;
L
Linus Torvalds 已提交
587 588 589 590
}

subsys_initcall(genhd_device_init);

591 592
static ssize_t disk_range_show(struct device *dev,
			       struct device_attribute *attr, char *buf)
L
Linus Torvalds 已提交
593
{
594
	struct gendisk *disk = dev_to_disk(dev);
L
Linus Torvalds 已提交
595

596
	return sprintf(buf, "%d\n", disk->minors);
L
Linus Torvalds 已提交
597 598
}

599 600
static ssize_t disk_removable_show(struct device *dev,
				   struct device_attribute *attr, char *buf)
601
{
602
	struct gendisk *disk = dev_to_disk(dev);
603

604 605
	return sprintf(buf, "%d\n",
		       (disk->flags & GENHD_FL_REMOVABLE ? 1 : 0));
606 607
}

K
Kay Sievers 已提交
608 609 610 611 612 613 614 615
static ssize_t disk_ro_show(struct device *dev,
				   struct device_attribute *attr, char *buf)
{
	struct gendisk *disk = dev_to_disk(dev);

	return sprintf(buf, "%d\n", disk->policy ? 1 : 0);
}

616 617
static ssize_t disk_size_show(struct device *dev,
			      struct device_attribute *attr, char *buf)
618
{
619
	struct gendisk *disk = dev_to_disk(dev);
L
Linus Torvalds 已提交
620

621
	return sprintf(buf, "%llu\n", (unsigned long long)get_capacity(disk));
L
Linus Torvalds 已提交
622
}
623 624 625

static ssize_t disk_capability_show(struct device *dev,
				    struct device_attribute *attr, char *buf)
626
{
627 628 629
	struct gendisk *disk = dev_to_disk(dev);

	return sprintf(buf, "%x\n", disk->flags);
630
}
631 632 633

static ssize_t disk_stat_show(struct device *dev,
			      struct device_attribute *attr, char *buf)
L
Linus Torvalds 已提交
634
{
635 636
	struct gendisk *disk = dev_to_disk(dev);

L
Linus Torvalds 已提交
637 638 639
	preempt_disable();
	disk_round_stats(disk);
	preempt_enable();
640
	return sprintf(buf,
641 642
		"%8lu %8lu %8llu %8u "
		"%8lu %8lu %8llu %8u "
L
Linus Torvalds 已提交
643 644
		"%8u %8u %8u"
		"\n",
645 646 647 648 649 650 651 652
		disk_stat_read(disk, ios[READ]),
		disk_stat_read(disk, merges[READ]),
		(unsigned long long)disk_stat_read(disk, sectors[READ]),
		jiffies_to_msecs(disk_stat_read(disk, ticks[READ])),
		disk_stat_read(disk, ios[WRITE]),
		disk_stat_read(disk, merges[WRITE]),
		(unsigned long long)disk_stat_read(disk, sectors[WRITE]),
		jiffies_to_msecs(disk_stat_read(disk, ticks[WRITE])),
L
Linus Torvalds 已提交
653 654 655 656 657
		disk->in_flight,
		jiffies_to_msecs(disk_stat_read(disk, io_ticks)),
		jiffies_to_msecs(disk_stat_read(disk, time_in_queue)));
}

658
#ifdef CONFIG_FAIL_MAKE_REQUEST
659 660 661 662 663 664 665
static ssize_t disk_fail_show(struct device *dev,
			      struct device_attribute *attr, char *buf)
{
	struct gendisk *disk = dev_to_disk(dev);

	return sprintf(buf, "%d\n", disk->flags & GENHD_FL_FAIL ? 1 : 0);
}
666

667 668
static ssize_t disk_fail_store(struct device *dev,
			       struct device_attribute *attr,
669 670
			       const char *buf, size_t count)
{
671
	struct gendisk *disk = dev_to_disk(dev);
672 673 674 675 676 677 678 679 680 681 682 683 684 685
	int i;

	if (count > 0 && sscanf(buf, "%d", &i) > 0) {
		if (i == 0)
			disk->flags &= ~GENHD_FL_FAIL;
		else
			disk->flags |= GENHD_FL_FAIL;
	}

	return count;
}

#endif

686 687
static DEVICE_ATTR(range, S_IRUGO, disk_range_show, NULL);
static DEVICE_ATTR(removable, S_IRUGO, disk_removable_show, NULL);
K
Kay Sievers 已提交
688
static DEVICE_ATTR(ro, S_IRUGO, disk_ro_show, NULL);
689 690 691
static DEVICE_ATTR(size, S_IRUGO, disk_size_show, NULL);
static DEVICE_ATTR(capability, S_IRUGO, disk_capability_show, NULL);
static DEVICE_ATTR(stat, S_IRUGO, disk_stat_show, NULL);
692
#ifdef CONFIG_FAIL_MAKE_REQUEST
693 694
static struct device_attribute dev_attr_fail =
	__ATTR(make-it-fail, S_IRUGO|S_IWUSR, disk_fail_show, disk_fail_store);
695
#endif
696 697 698 699

static struct attribute *disk_attrs[] = {
	&dev_attr_range.attr,
	&dev_attr_removable.attr,
K
Kay Sievers 已提交
700
	&dev_attr_ro.attr,
701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716
	&dev_attr_size.attr,
	&dev_attr_capability.attr,
	&dev_attr_stat.attr,
#ifdef CONFIG_FAIL_MAKE_REQUEST
	&dev_attr_fail.attr,
#endif
	NULL
};

static struct attribute_group disk_attr_group = {
	.attrs = disk_attrs,
};

static struct attribute_group *disk_attr_groups[] = {
	&disk_attr_group,
	NULL
L
Linus Torvalds 已提交
717 718
};

719
static void disk_release(struct device *dev)
L
Linus Torvalds 已提交
720
{
721 722
	struct gendisk *disk = dev_to_disk(dev);

L
Linus Torvalds 已提交
723
	kfree(disk->random);
724
	kfree(disk->__part);
L
Linus Torvalds 已提交
725 726 727
	free_disk_stats(disk);
	kfree(disk);
}
728 729
struct class block_class = {
	.name		= "block",
L
Linus Torvalds 已提交
730 731
};

A
Adrian Bunk 已提交
732
static struct device_type disk_type = {
733 734 735
	.name		= "disk",
	.groups		= disk_attr_groups,
	.release	= disk_release,
L
Linus Torvalds 已提交
736 737
};

738
#ifdef CONFIG_PROC_FS
739 740 741 742 743 744 745 746
/*
 * aggregate disk stat collector.  Uses the same stats that the sysfs
 * entries do, above, but makes them available through one seq_file.
 *
 * The output looks suspiciously like /proc/partitions with a bunch of
 * extra fields.
 */
static int diskstats_show(struct seq_file *seqf, void *v)
L
Linus Torvalds 已提交
747 748
{
	struct gendisk *gp = v;
749 750
	struct disk_part_iter piter;
	struct hd_struct *hd;
L
Linus Torvalds 已提交
751 752 753
	char buf[BDEVNAME_SIZE];

	/*
754
	if (&gp->dev.kobj.entry == block_class.devices.next)
755
		seq_puts(seqf,	"major minor name"
L
Linus Torvalds 已提交
756 757 758 759 760 761 762 763
				"     rio rmerge rsect ruse wio wmerge "
				"wsect wuse running use aveq"
				"\n\n");
	*/
 
	preempt_disable();
	disk_round_stats(gp);
	preempt_enable();
764
	seq_printf(seqf, "%4d %4d %s %lu %lu %llu %u %lu %lu %llu %u %u %u %u\n",
765 766
		MAJOR(disk_devt(gp)), MINOR(disk_devt(gp)),
		disk_name(gp, 0, buf),
767 768 769 770 771 772
		disk_stat_read(gp, ios[0]), disk_stat_read(gp, merges[0]),
		(unsigned long long)disk_stat_read(gp, sectors[0]),
		jiffies_to_msecs(disk_stat_read(gp, ticks[0])),
		disk_stat_read(gp, ios[1]), disk_stat_read(gp, merges[1]),
		(unsigned long long)disk_stat_read(gp, sectors[1]),
		jiffies_to_msecs(disk_stat_read(gp, ticks[1])),
L
Linus Torvalds 已提交
773 774 775 776 777
		gp->in_flight,
		jiffies_to_msecs(disk_stat_read(gp, io_ticks)),
		jiffies_to_msecs(disk_stat_read(gp, time_in_queue)));

	/* now show all non-0 size partitions of it */
778 779
	disk_part_iter_init(&piter, gp, 0);
	while ((hd = disk_part_iter_next(&piter))) {
780 781 782
		preempt_disable();
		part_round_stats(hd);
		preempt_enable();
783
		seq_printf(seqf, "%4d %4d %s %lu %lu %llu "
784
			   "%u %lu %lu %llu %u %u %u %u\n",
785 786
			   MAJOR(part_devt(hd)), MINOR(part_devt(hd)),
			   disk_name(gp, hd->partno, buf),
787 788 789 790 791 792 793 794 795 796 797 798
			   part_stat_read(hd, ios[0]),
			   part_stat_read(hd, merges[0]),
			   (unsigned long long)part_stat_read(hd, sectors[0]),
			   jiffies_to_msecs(part_stat_read(hd, ticks[0])),
			   part_stat_read(hd, ios[1]),
			   part_stat_read(hd, merges[1]),
			   (unsigned long long)part_stat_read(hd, sectors[1]),
			   jiffies_to_msecs(part_stat_read(hd, ticks[1])),
			   hd->in_flight,
			   jiffies_to_msecs(part_stat_read(hd, io_ticks)),
			   jiffies_to_msecs(part_stat_read(hd, time_in_queue))
			);
L
Linus Torvalds 已提交
799
	}
800
	disk_part_iter_exit(&piter);
L
Linus Torvalds 已提交
801 802 803 804
 
	return 0;
}

805
const struct seq_operations diskstats_op = {
806 807 808
	.start	= disk_seqf_start,
	.next	= disk_seqf_next,
	.stop	= disk_seqf_stop,
L
Linus Torvalds 已提交
809 810
	.show	= diskstats_show
};
811
#endif /* CONFIG_PROC_FS */
L
Linus Torvalds 已提交
812

813 814 815 816 817 818 819 820 821 822
static void media_change_notify_thread(struct work_struct *work)
{
	struct gendisk *gd = container_of(work, struct gendisk, async_notify);
	char event[] = "MEDIA_CHANGE=1";
	char *envp[] = { event, NULL };

	/*
	 * set enviroment vars to indicate which event this is for
	 * so that user space will know to go check the media status.
	 */
823
	kobject_uevent_env(&gd->dev.kobj, KOBJ_CHANGE, envp);
824 825 826
	put_device(gd->driverfs_dev);
}

A
Adrian Bunk 已提交
827
#if 0
828 829 830 831 832 833
void genhd_media_change_notify(struct gendisk *disk)
{
	get_device(disk->driverfs_dev);
	schedule_work(&disk->async_notify);
}
EXPORT_SYMBOL_GPL(genhd_media_change_notify);
A
Adrian Bunk 已提交
834
#endif  /*  0  */
835

836
dev_t blk_lookup_devt(const char *name, int partno)
837
{
838 839 840
	dev_t devt = MKDEV(0, 0);
	struct class_dev_iter iter;
	struct device *dev;
841

842 843
	class_dev_iter_init(&iter, &block_class, NULL, &disk_type);
	while ((dev = class_dev_iter_next(&iter))) {
844 845
		struct gendisk *disk = dev_to_disk(dev);

846 847 848 849 850 851 852 853
		if (strcmp(dev->bus_id, name))
			continue;
		if (partno < 0 || partno > disk_max_parts(disk))
			continue;

		if (partno == 0)
			devt = disk_devt(disk);
		else {
854
			struct hd_struct *part;
855

856 857 858
			part = disk_get_part(disk, partno);
			if (!part || !part->nr_sects) {
				disk_put_part(part);
859
				continue;
860
			}
861 862

			devt = part_devt(part);
863
			disk_put_part(part);
864
		}
865
		break;
866
	}
867
	class_dev_iter_exit(&iter);
868 869 870 871
	return devt;
}
EXPORT_SYMBOL(blk_lookup_devt);

L
Linus Torvalds 已提交
872 873
struct gendisk *alloc_disk(int minors)
{
874 875 876 877 878 879 880
	return alloc_disk_node(minors, -1);
}

struct gendisk *alloc_disk_node(int minors, int node_id)
{
	struct gendisk *disk;

881 882
	disk = kmalloc_node(sizeof(struct gendisk),
				GFP_KERNEL | __GFP_ZERO, node_id);
L
Linus Torvalds 已提交
883 884 885 886 887 888 889
	if (disk) {
		if (!init_disk_stats(disk)) {
			kfree(disk);
			return NULL;
		}
		if (minors > 1) {
			int size = (minors - 1) * sizeof(struct hd_struct *);
890
			disk->__part = kmalloc_node(size,
891
				GFP_KERNEL | __GFP_ZERO, node_id);
892
			if (!disk->__part) {
893
				free_disk_stats(disk);
L
Linus Torvalds 已提交
894 895 896 897 898 899
				kfree(disk);
				return NULL;
			}
		}
		disk->minors = minors;
		rand_initialize_disk(disk);
900 901 902
		disk->dev.class = &block_class;
		disk->dev.type = &disk_type;
		device_initialize(&disk->dev);
903 904
		INIT_WORK(&disk->async_notify,
			media_change_notify_thread);
L
Linus Torvalds 已提交
905 906 907 908 909
	}
	return disk;
}

EXPORT_SYMBOL(alloc_disk);
910
EXPORT_SYMBOL(alloc_disk_node);
L
Linus Torvalds 已提交
911 912 913 914 915 916 917 918 919 920 921

struct kobject *get_disk(struct gendisk *disk)
{
	struct module *owner;
	struct kobject *kobj;

	if (!disk->fops)
		return NULL;
	owner = disk->fops->owner;
	if (owner && !try_module_get(owner))
		return NULL;
922
	kobj = kobject_get(&disk->dev.kobj);
L
Linus Torvalds 已提交
923 924 925 926 927 928 929 930 931 932 933 934 935
	if (kobj == NULL) {
		module_put(owner);
		return NULL;
	}
	return kobj;

}

EXPORT_SYMBOL(get_disk);

void put_disk(struct gendisk *disk)
{
	if (disk)
936
		kobject_put(&disk->dev.kobj);
L
Linus Torvalds 已提交
937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952
}

EXPORT_SYMBOL(put_disk);

void set_device_ro(struct block_device *bdev, int flag)
{
	if (bdev->bd_contains != bdev)
		bdev->bd_part->policy = flag;
	else
		bdev->bd_disk->policy = flag;
}

EXPORT_SYMBOL(set_device_ro);

void set_disk_ro(struct gendisk *disk, int flag)
{
953 954 955
	struct disk_part_iter piter;
	struct hd_struct *part;

L
Linus Torvalds 已提交
956
	disk->policy = flag;
957 958 959 960
	disk_part_iter_init(&piter, disk, DISK_PITER_INCL_EMPTY);
	while ((part = disk_part_iter_next(&piter)))
		part->policy = flag;
	disk_part_iter_exit(&piter);
L
Linus Torvalds 已提交
961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976
}

EXPORT_SYMBOL(set_disk_ro);

int bdev_read_only(struct block_device *bdev)
{
	if (!bdev)
		return 0;
	else if (bdev->bd_contains != bdev)
		return bdev->bd_part->policy;
	else
		return bdev->bd_disk->policy;
}

EXPORT_SYMBOL(bdev_read_only);

977
int invalidate_partition(struct gendisk *disk, int partno)
L
Linus Torvalds 已提交
978 979
{
	int res = 0;
980
	struct block_device *bdev = bdget_disk(disk, partno);
L
Linus Torvalds 已提交
981
	if (bdev) {
982 983
		fsync_bdev(bdev);
		res = __invalidate_device(bdev);
L
Linus Torvalds 已提交
984 985 986 987 988 989
		bdput(bdev);
	}
	return res;
}

EXPORT_SYMBOL(invalidate_partition);