]>
Commit | Line | Data |
---|---|---|
1da177e4 LT |
1 | /* |
2 | * gendisk handling | |
3 | */ | |
4 | ||
1da177e4 LT |
5 | #include <linux/module.h> |
6 | #include <linux/fs.h> | |
7 | #include <linux/genhd.h> | |
b446b60e | 8 | #include <linux/kdev_t.h> |
1da177e4 LT |
9 | #include <linux/kernel.h> |
10 | #include <linux/blkdev.h> | |
11 | #include <linux/init.h> | |
12 | #include <linux/spinlock.h> | |
f500975a | 13 | #include <linux/proc_fs.h> |
1da177e4 LT |
14 | #include <linux/seq_file.h> |
15 | #include <linux/slab.h> | |
16 | #include <linux/kmod.h> | |
17 | #include <linux/kobj_map.h> | |
2ef41634 | 18 | #include <linux/buffer_head.h> |
58383af6 | 19 | #include <linux/mutex.h> |
bcce3de1 | 20 | #include <linux/idr.h> |
1da177e4 | 21 | |
ff88972c AB |
22 | #include "blk.h" |
23 | ||
edfaa7c3 | 24 | static DEFINE_MUTEX(block_class_lock); |
edfaa7c3 | 25 | struct kobject *block_depr; |
1da177e4 | 26 | |
bcce3de1 TH |
27 | /* for extended dynamic devt allocation, currently only one major is used */ |
28 | #define MAX_EXT_DEVT (1 << MINORBITS) | |
29 | ||
30 | /* For extended devt allocation. ext_devt_mutex prevents look up | |
31 | * results from going away underneath its user. | |
32 | */ | |
33 | static DEFINE_MUTEX(ext_devt_mutex); | |
34 | static DEFINE_IDR(ext_devt_idr); | |
35 | ||
1826eadf AB |
36 | static struct device_type disk_type; |
37 | ||
e71bf0d0 TH |
38 | /** |
39 | * disk_get_part - get partition | |
40 | * @disk: disk to look partition from | |
41 | * @partno: partition number | |
42 | * | |
43 | * Look for partition @partno from @disk. If found, increment | |
44 | * reference count and return it. | |
45 | * | |
46 | * CONTEXT: | |
47 | * Don't care. | |
48 | * | |
49 | * RETURNS: | |
50 | * Pointer to the found partition on success, NULL if not found. | |
51 | */ | |
52 | struct hd_struct *disk_get_part(struct gendisk *disk, int partno) | |
53 | { | |
540eed56 TH |
54 | struct hd_struct *part = NULL; |
55 | struct disk_part_tbl *ptbl; | |
e71bf0d0 | 56 | |
540eed56 | 57 | if (unlikely(partno < 0)) |
e71bf0d0 | 58 | return NULL; |
540eed56 | 59 | |
e71bf0d0 | 60 | rcu_read_lock(); |
540eed56 TH |
61 | |
62 | ptbl = rcu_dereference(disk->part_tbl); | |
63 | if (likely(partno < ptbl->len)) { | |
64 | part = rcu_dereference(ptbl->part[partno]); | |
65 | if (part) | |
66 | get_device(part_to_dev(part)); | |
67 | } | |
68 | ||
e71bf0d0 TH |
69 | rcu_read_unlock(); |
70 | ||
71 | return part; | |
72 | } | |
73 | EXPORT_SYMBOL_GPL(disk_get_part); | |
74 | ||
75 | /** | |
76 | * disk_part_iter_init - initialize partition iterator | |
77 | * @piter: iterator to initialize | |
78 | * @disk: disk to iterate over | |
79 | * @flags: DISK_PITER_* flags | |
80 | * | |
81 | * Initialize @piter so that it iterates over partitions of @disk. | |
82 | * | |
83 | * CONTEXT: | |
84 | * Don't care. | |
85 | */ | |
86 | void disk_part_iter_init(struct disk_part_iter *piter, struct gendisk *disk, | |
87 | unsigned int flags) | |
88 | { | |
540eed56 TH |
89 | struct disk_part_tbl *ptbl; |
90 | ||
91 | rcu_read_lock(); | |
92 | ptbl = rcu_dereference(disk->part_tbl); | |
93 | ||
e71bf0d0 TH |
94 | piter->disk = disk; |
95 | piter->part = NULL; | |
96 | ||
97 | if (flags & DISK_PITER_REVERSE) | |
540eed56 | 98 | piter->idx = ptbl->len - 1; |
71982a40 | 99 | else if (flags & (DISK_PITER_INCL_PART0 | DISK_PITER_INCL_EMPTY_PART0)) |
e71bf0d0 | 100 | piter->idx = 0; |
b5d0b9df TH |
101 | else |
102 | piter->idx = 1; | |
e71bf0d0 TH |
103 | |
104 | piter->flags = flags; | |
540eed56 TH |
105 | |
106 | rcu_read_unlock(); | |
e71bf0d0 TH |
107 | } |
108 | EXPORT_SYMBOL_GPL(disk_part_iter_init); | |
109 | ||
110 | /** | |
111 | * disk_part_iter_next - proceed iterator to the next partition and return it | |
112 | * @piter: iterator of interest | |
113 | * | |
114 | * Proceed @piter to the next partition and return it. | |
115 | * | |
116 | * CONTEXT: | |
117 | * Don't care. | |
118 | */ | |
119 | struct hd_struct *disk_part_iter_next(struct disk_part_iter *piter) | |
120 | { | |
540eed56 | 121 | struct disk_part_tbl *ptbl; |
e71bf0d0 TH |
122 | int inc, end; |
123 | ||
124 | /* put the last partition */ | |
125 | disk_put_part(piter->part); | |
126 | piter->part = NULL; | |
127 | ||
540eed56 | 128 | /* get part_tbl */ |
e71bf0d0 | 129 | rcu_read_lock(); |
540eed56 | 130 | ptbl = rcu_dereference(piter->disk->part_tbl); |
e71bf0d0 TH |
131 | |
132 | /* determine iteration parameters */ | |
133 | if (piter->flags & DISK_PITER_REVERSE) { | |
134 | inc = -1; | |
71982a40 TH |
135 | if (piter->flags & (DISK_PITER_INCL_PART0 | |
136 | DISK_PITER_INCL_EMPTY_PART0)) | |
b5d0b9df TH |
137 | end = -1; |
138 | else | |
139 | end = 0; | |
e71bf0d0 TH |
140 | } else { |
141 | inc = 1; | |
540eed56 | 142 | end = ptbl->len; |
e71bf0d0 TH |
143 | } |
144 | ||
145 | /* iterate to the next partition */ | |
146 | for (; piter->idx != end; piter->idx += inc) { | |
147 | struct hd_struct *part; | |
148 | ||
540eed56 | 149 | part = rcu_dereference(ptbl->part[piter->idx]); |
e71bf0d0 TH |
150 | if (!part) |
151 | continue; | |
71982a40 TH |
152 | if (!part->nr_sects && |
153 | !(piter->flags & DISK_PITER_INCL_EMPTY) && | |
154 | !(piter->flags & DISK_PITER_INCL_EMPTY_PART0 && | |
155 | piter->idx == 0)) | |
e71bf0d0 TH |
156 | continue; |
157 | ||
ed9e1982 | 158 | get_device(part_to_dev(part)); |
e71bf0d0 TH |
159 | piter->part = part; |
160 | piter->idx += inc; | |
161 | break; | |
162 | } | |
163 | ||
164 | rcu_read_unlock(); | |
165 | ||
166 | return piter->part; | |
167 | } | |
168 | EXPORT_SYMBOL_GPL(disk_part_iter_next); | |
169 | ||
170 | /** | |
171 | * disk_part_iter_exit - finish up partition iteration | |
172 | * @piter: iter of interest | |
173 | * | |
174 | * Called when iteration is over. Cleans up @piter. | |
175 | * | |
176 | * CONTEXT: | |
177 | * Don't care. | |
178 | */ | |
179 | void disk_part_iter_exit(struct disk_part_iter *piter) | |
180 | { | |
181 | disk_put_part(piter->part); | |
182 | piter->part = NULL; | |
183 | } | |
184 | EXPORT_SYMBOL_GPL(disk_part_iter_exit); | |
185 | ||
a6f23657 JA |
186 | static inline int sector_in_part(struct hd_struct *part, sector_t sector) |
187 | { | |
188 | return part->start_sect <= sector && | |
189 | sector < part->start_sect + part->nr_sects; | |
190 | } | |
191 | ||
e71bf0d0 TH |
192 | /** |
193 | * disk_map_sector_rcu - map sector to partition | |
194 | * @disk: gendisk of interest | |
195 | * @sector: sector to map | |
196 | * | |
197 | * Find out which partition @sector maps to on @disk. This is | |
198 | * primarily used for stats accounting. | |
199 | * | |
200 | * CONTEXT: | |
201 | * RCU read locked. The returned partition pointer is valid only | |
202 | * while preemption is disabled. | |
203 | * | |
204 | * RETURNS: | |
074a7aca | 205 | * Found partition on success, part0 is returned if no partition matches |
e71bf0d0 TH |
206 | */ |
207 | struct hd_struct *disk_map_sector_rcu(struct gendisk *disk, sector_t sector) | |
208 | { | |
540eed56 | 209 | struct disk_part_tbl *ptbl; |
a6f23657 | 210 | struct hd_struct *part; |
e71bf0d0 TH |
211 | int i; |
212 | ||
540eed56 TH |
213 | ptbl = rcu_dereference(disk->part_tbl); |
214 | ||
a6f23657 JA |
215 | part = rcu_dereference(ptbl->last_lookup); |
216 | if (part && sector_in_part(part, sector)) | |
217 | return part; | |
218 | ||
540eed56 | 219 | for (i = 1; i < ptbl->len; i++) { |
a6f23657 | 220 | part = rcu_dereference(ptbl->part[i]); |
e71bf0d0 | 221 | |
a6f23657 JA |
222 | if (part && sector_in_part(part, sector)) { |
223 | rcu_assign_pointer(ptbl->last_lookup, part); | |
e71bf0d0 | 224 | return part; |
a6f23657 | 225 | } |
e71bf0d0 | 226 | } |
074a7aca | 227 | return &disk->part0; |
e71bf0d0 TH |
228 | } |
229 | EXPORT_SYMBOL_GPL(disk_map_sector_rcu); | |
230 | ||
1da177e4 LT |
231 | /* |
232 | * Can be deleted altogether. Later. | |
233 | * | |
234 | */ | |
235 | static struct blk_major_name { | |
236 | struct blk_major_name *next; | |
237 | int major; | |
238 | char name[16]; | |
68eef3b4 | 239 | } *major_names[BLKDEV_MAJOR_HASH_SIZE]; |
1da177e4 LT |
240 | |
241 | /* index in the above - for now: assume no multimajor ranges */ | |
242 | static inline int major_to_index(int major) | |
243 | { | |
68eef3b4 | 244 | return major % BLKDEV_MAJOR_HASH_SIZE; |
7170be5f NH |
245 | } |
246 | ||
68eef3b4 | 247 | #ifdef CONFIG_PROC_FS |
cf771cb5 | 248 | void blkdev_show(struct seq_file *seqf, off_t offset) |
7170be5f | 249 | { |
68eef3b4 | 250 | struct blk_major_name *dp; |
7170be5f | 251 | |
68eef3b4 | 252 | if (offset < BLKDEV_MAJOR_HASH_SIZE) { |
edfaa7c3 | 253 | mutex_lock(&block_class_lock); |
68eef3b4 | 254 | for (dp = major_names[offset]; dp; dp = dp->next) |
cf771cb5 | 255 | seq_printf(seqf, "%3d %s\n", dp->major, dp->name); |
edfaa7c3 | 256 | mutex_unlock(&block_class_lock); |
1da177e4 | 257 | } |
1da177e4 | 258 | } |
68eef3b4 | 259 | #endif /* CONFIG_PROC_FS */ |
1da177e4 | 260 | |
9e8c0bcc MN |
261 | /** |
262 | * register_blkdev - register a new block device | |
263 | * | |
264 | * @major: the requested major device number [1..255]. If @major=0, try to | |
265 | * allocate any unused major number. | |
266 | * @name: the name of the new block device as a zero terminated string | |
267 | * | |
268 | * The @name must be unique within the system. | |
269 | * | |
270 | * The return value depends on the @major input parameter. | |
271 | * - if a major device number was requested in range [1..255] then the | |
272 | * function returns zero on success, or a negative error code | |
273 | * - if any unused major number was requested with @major=0 parameter | |
274 | * then the return value is the allocated major number in range | |
275 | * [1..255] or a negative error code otherwise | |
276 | */ | |
1da177e4 LT |
277 | int register_blkdev(unsigned int major, const char *name) |
278 | { | |
279 | struct blk_major_name **n, *p; | |
280 | int index, ret = 0; | |
281 | ||
edfaa7c3 | 282 | mutex_lock(&block_class_lock); |
1da177e4 LT |
283 | |
284 | /* temporary */ | |
285 | if (major == 0) { | |
286 | for (index = ARRAY_SIZE(major_names)-1; index > 0; index--) { | |
287 | if (major_names[index] == NULL) | |
288 | break; | |
289 | } | |
290 | ||
291 | if (index == 0) { | |
292 | printk("register_blkdev: failed to get major for %s\n", | |
293 | name); | |
294 | ret = -EBUSY; | |
295 | goto out; | |
296 | } | |
297 | major = index; | |
298 | ret = major; | |
299 | } | |
300 | ||
301 | p = kmalloc(sizeof(struct blk_major_name), GFP_KERNEL); | |
302 | if (p == NULL) { | |
303 | ret = -ENOMEM; | |
304 | goto out; | |
305 | } | |
306 | ||
307 | p->major = major; | |
308 | strlcpy(p->name, name, sizeof(p->name)); | |
309 | p->next = NULL; | |
310 | index = major_to_index(major); | |
311 | ||
312 | for (n = &major_names[index]; *n; n = &(*n)->next) { | |
313 | if ((*n)->major == major) | |
314 | break; | |
315 | } | |
316 | if (!*n) | |
317 | *n = p; | |
318 | else | |
319 | ret = -EBUSY; | |
320 | ||
321 | if (ret < 0) { | |
322 | printk("register_blkdev: cannot get major %d for %s\n", | |
323 | major, name); | |
324 | kfree(p); | |
325 | } | |
326 | out: | |
edfaa7c3 | 327 | mutex_unlock(&block_class_lock); |
1da177e4 LT |
328 | return ret; |
329 | } | |
330 | ||
331 | EXPORT_SYMBOL(register_blkdev); | |
332 | ||
f4480240 | 333 | void unregister_blkdev(unsigned int major, const char *name) |
1da177e4 LT |
334 | { |
335 | struct blk_major_name **n; | |
336 | struct blk_major_name *p = NULL; | |
337 | int index = major_to_index(major); | |
1da177e4 | 338 | |
edfaa7c3 | 339 | mutex_lock(&block_class_lock); |
1da177e4 LT |
340 | for (n = &major_names[index]; *n; n = &(*n)->next) |
341 | if ((*n)->major == major) | |
342 | break; | |
294462a5 AM |
343 | if (!*n || strcmp((*n)->name, name)) { |
344 | WARN_ON(1); | |
294462a5 | 345 | } else { |
1da177e4 LT |
346 | p = *n; |
347 | *n = p->next; | |
348 | } | |
edfaa7c3 | 349 | mutex_unlock(&block_class_lock); |
1da177e4 | 350 | kfree(p); |
1da177e4 LT |
351 | } |
352 | ||
353 | EXPORT_SYMBOL(unregister_blkdev); | |
354 | ||
355 | static struct kobj_map *bdev_map; | |
356 | ||
870d6656 TH |
357 | /** |
358 | * blk_mangle_minor - scatter minor numbers apart | |
359 | * @minor: minor number to mangle | |
360 | * | |
361 | * Scatter consecutively allocated @minor number apart if MANGLE_DEVT | |
362 | * is enabled. Mangling twice gives the original value. | |
363 | * | |
364 | * RETURNS: | |
365 | * Mangled value. | |
366 | * | |
367 | * CONTEXT: | |
368 | * Don't care. | |
369 | */ | |
370 | static int blk_mangle_minor(int minor) | |
371 | { | |
372 | #ifdef CONFIG_DEBUG_BLOCK_EXT_DEVT | |
373 | int i; | |
374 | ||
375 | for (i = 0; i < MINORBITS / 2; i++) { | |
376 | int low = minor & (1 << i); | |
377 | int high = minor & (1 << (MINORBITS - 1 - i)); | |
378 | int distance = MINORBITS - 1 - 2 * i; | |
379 | ||
380 | minor ^= low | high; /* clear both bits */ | |
381 | low <<= distance; /* swap the positions */ | |
382 | high >>= distance; | |
383 | minor |= low | high; /* and set */ | |
384 | } | |
385 | #endif | |
386 | return minor; | |
387 | } | |
388 | ||
bcce3de1 TH |
389 | /** |
390 | * blk_alloc_devt - allocate a dev_t for a partition | |
391 | * @part: partition to allocate dev_t for | |
bcce3de1 TH |
392 | * @devt: out parameter for resulting dev_t |
393 | * | |
394 | * Allocate a dev_t for block device. | |
395 | * | |
396 | * RETURNS: | |
397 | * 0 on success, allocated dev_t is returned in *@devt. -errno on | |
398 | * failure. | |
399 | * | |
400 | * CONTEXT: | |
401 | * Might sleep. | |
402 | */ | |
403 | int blk_alloc_devt(struct hd_struct *part, dev_t *devt) | |
404 | { | |
405 | struct gendisk *disk = part_to_disk(part); | |
406 | int idx, rc; | |
407 | ||
408 | /* in consecutive minor range? */ | |
409 | if (part->partno < disk->minors) { | |
410 | *devt = MKDEV(disk->major, disk->first_minor + part->partno); | |
411 | return 0; | |
412 | } | |
413 | ||
414 | /* allocate ext devt */ | |
415 | do { | |
416 | if (!idr_pre_get(&ext_devt_idr, GFP_KERNEL)) | |
417 | return -ENOMEM; | |
418 | rc = idr_get_new(&ext_devt_idr, part, &idx); | |
419 | } while (rc == -EAGAIN); | |
420 | ||
421 | if (rc) | |
422 | return rc; | |
423 | ||
424 | if (idx > MAX_EXT_DEVT) { | |
425 | idr_remove(&ext_devt_idr, idx); | |
426 | return -EBUSY; | |
427 | } | |
428 | ||
870d6656 | 429 | *devt = MKDEV(BLOCK_EXT_MAJOR, blk_mangle_minor(idx)); |
bcce3de1 TH |
430 | return 0; |
431 | } | |
432 | ||
433 | /** | |
434 | * blk_free_devt - free a dev_t | |
435 | * @devt: dev_t to free | |
436 | * | |
437 | * Free @devt which was allocated using blk_alloc_devt(). | |
438 | * | |
439 | * CONTEXT: | |
440 | * Might sleep. | |
441 | */ | |
442 | void blk_free_devt(dev_t devt) | |
443 | { | |
444 | might_sleep(); | |
445 | ||
446 | if (devt == MKDEV(0, 0)) | |
447 | return; | |
448 | ||
449 | if (MAJOR(devt) == BLOCK_EXT_MAJOR) { | |
450 | mutex_lock(&ext_devt_mutex); | |
870d6656 | 451 | idr_remove(&ext_devt_idr, blk_mangle_minor(MINOR(devt))); |
bcce3de1 TH |
452 | mutex_unlock(&ext_devt_mutex); |
453 | } | |
454 | } | |
455 | ||
1f014290 TH |
456 | static char *bdevt_str(dev_t devt, char *buf) |
457 | { | |
458 | if (MAJOR(devt) <= 0xff && MINOR(devt) <= 0xff) { | |
459 | char tbuf[BDEVT_SIZE]; | |
460 | snprintf(tbuf, BDEVT_SIZE, "%02x%02x", MAJOR(devt), MINOR(devt)); | |
461 | snprintf(buf, BDEVT_SIZE, "%-9s", tbuf); | |
462 | } else | |
463 | snprintf(buf, BDEVT_SIZE, "%03x:%05x", MAJOR(devt), MINOR(devt)); | |
464 | ||
465 | return buf; | |
466 | } | |
467 | ||
1da177e4 LT |
468 | /* |
469 | * Register device numbers dev..(dev+range-1) | |
470 | * range must be nonzero | |
471 | * The hash chain is sorted on range, so that subranges can override. | |
472 | */ | |
edfaa7c3 | 473 | void blk_register_region(dev_t devt, unsigned long range, struct module *module, |
1da177e4 LT |
474 | struct kobject *(*probe)(dev_t, int *, void *), |
475 | int (*lock)(dev_t, void *), void *data) | |
476 | { | |
edfaa7c3 | 477 | kobj_map(bdev_map, devt, range, module, probe, lock, data); |
1da177e4 LT |
478 | } |
479 | ||
480 | EXPORT_SYMBOL(blk_register_region); | |
481 | ||
edfaa7c3 | 482 | void blk_unregister_region(dev_t devt, unsigned long range) |
1da177e4 | 483 | { |
edfaa7c3 | 484 | kobj_unmap(bdev_map, devt, range); |
1da177e4 LT |
485 | } |
486 | ||
487 | EXPORT_SYMBOL(blk_unregister_region); | |
488 | ||
cf771cb5 | 489 | static struct kobject *exact_match(dev_t devt, int *partno, void *data) |
1da177e4 LT |
490 | { |
491 | struct gendisk *p = data; | |
edfaa7c3 | 492 | |
ed9e1982 | 493 | return &disk_to_dev(p)->kobj; |
1da177e4 LT |
494 | } |
495 | ||
edfaa7c3 | 496 | static int exact_lock(dev_t devt, void *data) |
1da177e4 LT |
497 | { |
498 | struct gendisk *p = data; | |
499 | ||
500 | if (!get_disk(p)) | |
501 | return -1; | |
502 | return 0; | |
503 | } | |
504 | ||
d2bf1b67 TH |
505 | void register_disk(struct gendisk *disk) |
506 | { | |
507 | struct device *ddev = disk_to_dev(disk); | |
508 | struct block_device *bdev; | |
509 | struct disk_part_iter piter; | |
510 | struct hd_struct *part; | |
511 | int err; | |
512 | ||
513 | ddev->parent = disk->driverfs_dev; | |
514 | ||
515 | dev_set_name(ddev, disk->disk_name); | |
516 | ||
517 | /* delay uevents, until we scanned partition table */ | |
518 | dev_set_uevent_suppress(ddev, 1); | |
519 | ||
520 | if (device_add(ddev)) | |
521 | return; | |
522 | if (!sysfs_deprecated) { | |
523 | err = sysfs_create_link(block_depr, &ddev->kobj, | |
524 | kobject_name(&ddev->kobj)); | |
525 | if (err) { | |
526 | device_del(ddev); | |
527 | return; | |
528 | } | |
529 | } | |
530 | disk->part0.holder_dir = kobject_create_and_add("holders", &ddev->kobj); | |
531 | disk->slave_dir = kobject_create_and_add("slaves", &ddev->kobj); | |
532 | ||
533 | /* No minors to use for partitions */ | |
534 | if (!disk_partitionable(disk)) | |
535 | goto exit; | |
536 | ||
537 | /* No such device (e.g., media were just removed) */ | |
538 | if (!get_capacity(disk)) | |
539 | goto exit; | |
540 | ||
541 | bdev = bdget_disk(disk, 0); | |
542 | if (!bdev) | |
543 | goto exit; | |
544 | ||
545 | bdev->bd_invalidated = 1; | |
546 | err = blkdev_get(bdev, FMODE_READ, NULL); | |
547 | if (err < 0) | |
548 | goto exit; | |
549 | blkdev_put(bdev, FMODE_READ); | |
550 | ||
551 | exit: | |
552 | /* announce disk after possible partitions are created */ | |
553 | dev_set_uevent_suppress(ddev, 0); | |
554 | kobject_uevent(&ddev->kobj, KOBJ_ADD); | |
555 | ||
556 | /* announce possible partitions */ | |
557 | disk_part_iter_init(&piter, disk, 0); | |
558 | while ((part = disk_part_iter_next(&piter))) | |
559 | kobject_uevent(&part_to_dev(part)->kobj, KOBJ_ADD); | |
560 | disk_part_iter_exit(&piter); | |
561 | } | |
562 | ||
1da177e4 LT |
563 | /** |
564 | * add_disk - add partitioning information to kernel list | |
565 | * @disk: per-device partitioning information | |
566 | * | |
567 | * This function registers the partitioning information in @disk | |
568 | * with the kernel. | |
3e1a7ff8 TH |
569 | * |
570 | * FIXME: error handling | |
1da177e4 LT |
571 | */ |
572 | void add_disk(struct gendisk *disk) | |
573 | { | |
cf0ca9fe | 574 | struct backing_dev_info *bdi; |
3e1a7ff8 | 575 | dev_t devt; |
6ffeea77 | 576 | int retval; |
cf0ca9fe | 577 | |
3e1a7ff8 TH |
578 | /* minors == 0 indicates to use ext devt from part0 and should |
579 | * be accompanied with EXT_DEVT flag. Make sure all | |
580 | * parameters make sense. | |
581 | */ | |
582 | WARN_ON(disk->minors && !(disk->major || disk->first_minor)); | |
583 | WARN_ON(!disk->minors && !(disk->flags & GENHD_FL_EXT_DEVT)); | |
584 | ||
1da177e4 | 585 | disk->flags |= GENHD_FL_UP; |
3e1a7ff8 TH |
586 | |
587 | retval = blk_alloc_devt(&disk->part0, &devt); | |
588 | if (retval) { | |
589 | WARN_ON(1); | |
590 | return; | |
591 | } | |
592 | disk_to_dev(disk)->devt = devt; | |
593 | ||
594 | /* ->major and ->first_minor aren't supposed to be | |
595 | * dereferenced from here on, but set them just in case. | |
596 | */ | |
597 | disk->major = MAJOR(devt); | |
598 | disk->first_minor = MINOR(devt); | |
599 | ||
01ea5063 SJK |
600 | /* Register BDI before referencing it from bdev */ |
601 | bdi = &disk->queue->backing_dev_info; | |
602 | bdi_register_dev(bdi, disk_devt(disk)); | |
603 | ||
f331c029 TH |
604 | blk_register_region(disk_devt(disk), disk->minors, NULL, |
605 | exact_match, exact_lock, disk); | |
1da177e4 LT |
606 | register_disk(disk); |
607 | blk_register_queue(disk); | |
cf0ca9fe | 608 | |
ed9e1982 TH |
609 | retval = sysfs_create_link(&disk_to_dev(disk)->kobj, &bdi->dev->kobj, |
610 | "bdi"); | |
6ffeea77 | 611 | WARN_ON(retval); |
1da177e4 | 612 | } |
1da177e4 | 613 | EXPORT_SYMBOL(add_disk); |
1da177e4 | 614 | |
d2bf1b67 | 615 | void del_gendisk(struct gendisk *disk) |
1da177e4 | 616 | { |
d2bf1b67 TH |
617 | struct disk_part_iter piter; |
618 | struct hd_struct *part; | |
619 | ||
620 | /* invalidate stuff */ | |
621 | disk_part_iter_init(&piter, disk, | |
622 | DISK_PITER_INCL_EMPTY | DISK_PITER_REVERSE); | |
623 | while ((part = disk_part_iter_next(&piter))) { | |
624 | invalidate_partition(disk, part->partno); | |
625 | delete_partition(disk, part->partno); | |
626 | } | |
627 | disk_part_iter_exit(&piter); | |
628 | ||
629 | invalidate_partition(disk, 0); | |
630 | blk_free_devt(disk_to_dev(disk)->devt); | |
631 | set_capacity(disk, 0); | |
632 | disk->flags &= ~GENHD_FL_UP; | |
633 | ||
ed9e1982 | 634 | sysfs_remove_link(&disk_to_dev(disk)->kobj, "bdi"); |
cf0ca9fe | 635 | bdi_unregister(&disk->queue->backing_dev_info); |
1da177e4 | 636 | blk_unregister_queue(disk); |
f331c029 | 637 | blk_unregister_region(disk_devt(disk), disk->minors); |
d2bf1b67 TH |
638 | |
639 | part_stat_set_all(&disk->part0, 0); | |
640 | disk->part0.stamp = 0; | |
641 | ||
642 | kobject_put(disk->part0.holder_dir); | |
643 | kobject_put(disk->slave_dir); | |
644 | disk->driverfs_dev = NULL; | |
645 | if (!sysfs_deprecated) | |
646 | sysfs_remove_link(block_depr, dev_name(disk_to_dev(disk))); | |
647 | device_del(disk_to_dev(disk)); | |
1da177e4 | 648 | } |
d2bf1b67 | 649 | EXPORT_SYMBOL(del_gendisk); |
1da177e4 | 650 | |
1da177e4 LT |
651 | /** |
652 | * get_gendisk - get partitioning information for a given device | |
710027a4 | 653 | * @devt: device to get partitioning information for |
496aa8a9 | 654 | * @partno: returned partition index |
1da177e4 LT |
655 | * |
656 | * This function gets the structure containing partitioning | |
710027a4 | 657 | * information for the given device @devt. |
1da177e4 | 658 | */ |
cf771cb5 | 659 | struct gendisk *get_gendisk(dev_t devt, int *partno) |
1da177e4 | 660 | { |
bcce3de1 TH |
661 | struct gendisk *disk = NULL; |
662 | ||
663 | if (MAJOR(devt) != BLOCK_EXT_MAJOR) { | |
664 | struct kobject *kobj; | |
665 | ||
666 | kobj = kobj_lookup(bdev_map, devt, partno); | |
667 | if (kobj) | |
668 | disk = dev_to_disk(kobj_to_dev(kobj)); | |
669 | } else { | |
670 | struct hd_struct *part; | |
671 | ||
672 | mutex_lock(&ext_devt_mutex); | |
870d6656 | 673 | part = idr_find(&ext_devt_idr, blk_mangle_minor(MINOR(devt))); |
bcce3de1 TH |
674 | if (part && get_disk(part_to_disk(part))) { |
675 | *partno = part->partno; | |
676 | disk = part_to_disk(part); | |
677 | } | |
678 | mutex_unlock(&ext_devt_mutex); | |
679 | } | |
edfaa7c3 | 680 | |
bcce3de1 | 681 | return disk; |
1da177e4 | 682 | } |
b6ac23af | 683 | EXPORT_SYMBOL(get_gendisk); |
1da177e4 | 684 | |
f331c029 TH |
685 | /** |
686 | * bdget_disk - do bdget() by gendisk and partition number | |
687 | * @disk: gendisk of interest | |
688 | * @partno: partition number | |
689 | * | |
690 | * Find partition @partno from @disk, do bdget() on it. | |
691 | * | |
692 | * CONTEXT: | |
693 | * Don't care. | |
694 | * | |
695 | * RETURNS: | |
696 | * Resulting block_device on success, NULL on failure. | |
697 | */ | |
aeb3d3a8 | 698 | struct block_device *bdget_disk(struct gendisk *disk, int partno) |
f331c029 | 699 | { |
548b10eb TH |
700 | struct hd_struct *part; |
701 | struct block_device *bdev = NULL; | |
f331c029 | 702 | |
548b10eb | 703 | part = disk_get_part(disk, partno); |
2bbedcb4 | 704 | if (part) |
548b10eb TH |
705 | bdev = bdget(part_devt(part)); |
706 | disk_put_part(part); | |
f331c029 | 707 | |
548b10eb | 708 | return bdev; |
f331c029 TH |
709 | } |
710 | EXPORT_SYMBOL(bdget_disk); | |
711 | ||
5c6f35c5 GKH |
712 | /* |
713 | * print a full list of all partitions - intended for places where the root | |
714 | * filesystem can't be mounted and thus to give the victim some idea of what | |
715 | * went wrong | |
716 | */ | |
717 | void __init printk_all_partitions(void) | |
718 | { | |
def4e38d TH |
719 | struct class_dev_iter iter; |
720 | struct device *dev; | |
721 | ||
722 | class_dev_iter_init(&iter, &block_class, NULL, &disk_type); | |
723 | while ((dev = class_dev_iter_next(&iter))) { | |
724 | struct gendisk *disk = dev_to_disk(dev); | |
e71bf0d0 TH |
725 | struct disk_part_iter piter; |
726 | struct hd_struct *part; | |
1f014290 TH |
727 | char name_buf[BDEVNAME_SIZE]; |
728 | char devt_buf[BDEVT_SIZE]; | |
b5af921e | 729 | u8 uuid[PARTITION_META_INFO_UUIDLTH * 2 + 1]; |
def4e38d TH |
730 | |
731 | /* | |
732 | * Don't show empty devices or things that have been | |
733 | * surpressed | |
734 | */ | |
735 | if (get_capacity(disk) == 0 || | |
736 | (disk->flags & GENHD_FL_SUPPRESS_PARTITION_INFO)) | |
737 | continue; | |
738 | ||
739 | /* | |
740 | * Note, unlike /proc/partitions, I am showing the | |
741 | * numbers in hex - the same format as the root= | |
742 | * option takes. | |
743 | */ | |
074a7aca TH |
744 | disk_part_iter_init(&piter, disk, DISK_PITER_INCL_PART0); |
745 | while ((part = disk_part_iter_next(&piter))) { | |
746 | bool is_part0 = part == &disk->part0; | |
def4e38d | 747 | |
b5af921e WD |
748 | uuid[0] = 0; |
749 | if (part->info) | |
750 | part_unpack_uuid(part->info->uuid, uuid); | |
751 | ||
752 | printk("%s%s %10llu %s %s", is_part0 ? "" : " ", | |
1f014290 | 753 | bdevt_str(part_devt(part), devt_buf), |
f331c029 | 754 | (unsigned long long)part->nr_sects >> 1, |
b5af921e | 755 | disk_name(disk, part->partno, name_buf), uuid); |
074a7aca TH |
756 | if (is_part0) { |
757 | if (disk->driverfs_dev != NULL && | |
758 | disk->driverfs_dev->driver != NULL) | |
759 | printk(" driver: %s\n", | |
760 | disk->driverfs_dev->driver->name); | |
761 | else | |
762 | printk(" (driver?)\n"); | |
763 | } else | |
764 | printk("\n"); | |
765 | } | |
e71bf0d0 | 766 | disk_part_iter_exit(&piter); |
def4e38d TH |
767 | } |
768 | class_dev_iter_exit(&iter); | |
dd2a345f DG |
769 | } |
770 | ||
1da177e4 LT |
771 | #ifdef CONFIG_PROC_FS |
772 | /* iterator */ | |
def4e38d | 773 | static void *disk_seqf_start(struct seq_file *seqf, loff_t *pos) |
68c4d4a7 | 774 | { |
def4e38d TH |
775 | loff_t skip = *pos; |
776 | struct class_dev_iter *iter; | |
777 | struct device *dev; | |
68c4d4a7 | 778 | |
aeb3d3a8 | 779 | iter = kmalloc(sizeof(*iter), GFP_KERNEL); |
def4e38d TH |
780 | if (!iter) |
781 | return ERR_PTR(-ENOMEM); | |
782 | ||
783 | seqf->private = iter; | |
784 | class_dev_iter_init(iter, &block_class, NULL, &disk_type); | |
785 | do { | |
786 | dev = class_dev_iter_next(iter); | |
787 | if (!dev) | |
788 | return NULL; | |
789 | } while (skip--); | |
790 | ||
791 | return dev_to_disk(dev); | |
68c4d4a7 GKH |
792 | } |
793 | ||
def4e38d | 794 | static void *disk_seqf_next(struct seq_file *seqf, void *v, loff_t *pos) |
1da177e4 | 795 | { |
edfaa7c3 | 796 | struct device *dev; |
1da177e4 | 797 | |
def4e38d TH |
798 | (*pos)++; |
799 | dev = class_dev_iter_next(seqf->private); | |
2ac3cee5 | 800 | if (dev) |
68c4d4a7 | 801 | return dev_to_disk(dev); |
2ac3cee5 | 802 | |
1da177e4 LT |
803 | return NULL; |
804 | } | |
805 | ||
def4e38d | 806 | static void disk_seqf_stop(struct seq_file *seqf, void *v) |
27f30251 | 807 | { |
def4e38d | 808 | struct class_dev_iter *iter = seqf->private; |
27f30251 | 809 | |
def4e38d TH |
810 | /* stop is called even after start failed :-( */ |
811 | if (iter) { | |
812 | class_dev_iter_exit(iter); | |
813 | kfree(iter); | |
5c0ef6d0 | 814 | } |
1da177e4 LT |
815 | } |
816 | ||
def4e38d | 817 | static void *show_partition_start(struct seq_file *seqf, loff_t *pos) |
1da177e4 | 818 | { |
def4e38d TH |
819 | static void *p; |
820 | ||
821 | p = disk_seqf_start(seqf, pos); | |
243294da | 822 | if (!IS_ERR(p) && p && !*pos) |
def4e38d TH |
823 | seq_puts(seqf, "major minor #blocks name\n\n"); |
824 | return p; | |
1da177e4 LT |
825 | } |
826 | ||
cf771cb5 | 827 | static int show_partition(struct seq_file *seqf, void *v) |
1da177e4 LT |
828 | { |
829 | struct gendisk *sgp = v; | |
e71bf0d0 TH |
830 | struct disk_part_iter piter; |
831 | struct hd_struct *part; | |
1da177e4 LT |
832 | char buf[BDEVNAME_SIZE]; |
833 | ||
1da177e4 | 834 | /* Don't show non-partitionable removeable devices or empty devices */ |
b5d0b9df | 835 | if (!get_capacity(sgp) || (!disk_partitionable(sgp) && |
f331c029 | 836 | (sgp->flags & GENHD_FL_REMOVABLE))) |
1da177e4 LT |
837 | return 0; |
838 | if (sgp->flags & GENHD_FL_SUPPRESS_PARTITION_INFO) | |
839 | return 0; | |
840 | ||
841 | /* show the full disk and all non-0 size partitions of it */ | |
074a7aca | 842 | disk_part_iter_init(&piter, sgp, DISK_PITER_INCL_PART0); |
e71bf0d0 | 843 | while ((part = disk_part_iter_next(&piter))) |
1f014290 | 844 | seq_printf(seqf, "%4d %7d %10llu %s\n", |
f331c029 TH |
845 | MAJOR(part_devt(part)), MINOR(part_devt(part)), |
846 | (unsigned long long)part->nr_sects >> 1, | |
847 | disk_name(sgp, part->partno, buf)); | |
e71bf0d0 | 848 | disk_part_iter_exit(&piter); |
1da177e4 LT |
849 | |
850 | return 0; | |
851 | } | |
852 | ||
f500975a | 853 | static const struct seq_operations partitions_op = { |
def4e38d TH |
854 | .start = show_partition_start, |
855 | .next = disk_seqf_next, | |
856 | .stop = disk_seqf_stop, | |
edfaa7c3 | 857 | .show = show_partition |
1da177e4 | 858 | }; |
f500975a AD |
859 | |
860 | static int partitions_open(struct inode *inode, struct file *file) | |
861 | { | |
862 | return seq_open(file, &partitions_op); | |
863 | } | |
864 | ||
865 | static const struct file_operations proc_partitions_operations = { | |
866 | .open = partitions_open, | |
867 | .read = seq_read, | |
868 | .llseek = seq_lseek, | |
869 | .release = seq_release, | |
870 | }; | |
1da177e4 LT |
871 | #endif |
872 | ||
873 | ||
cf771cb5 | 874 | static struct kobject *base_probe(dev_t devt, int *partno, void *data) |
1da177e4 | 875 | { |
edfaa7c3 | 876 | if (request_module("block-major-%d-%d", MAJOR(devt), MINOR(devt)) > 0) |
1da177e4 | 877 | /* Make old-style 2.4 aliases work */ |
edfaa7c3 | 878 | request_module("block-major-%d", MAJOR(devt)); |
1da177e4 LT |
879 | return NULL; |
880 | } | |
881 | ||
882 | static int __init genhd_device_init(void) | |
883 | { | |
e105b8bf DW |
884 | int error; |
885 | ||
886 | block_class.dev_kobj = sysfs_dev_block_kobj; | |
887 | error = class_register(&block_class); | |
ee27a558 RM |
888 | if (unlikely(error)) |
889 | return error; | |
edfaa7c3 | 890 | bdev_map = kobj_map_init(base_probe, &block_class_lock); |
1da177e4 | 891 | blk_dev_init(); |
edfaa7c3 | 892 | |
561ec68e ZY |
893 | register_blkdev(BLOCK_EXT_MAJOR, "blkext"); |
894 | ||
edfaa7c3 | 895 | /* create top-level block dir */ |
e52eec13 AK |
896 | if (!sysfs_deprecated) |
897 | block_depr = kobject_create_and_add("block", NULL); | |
830d3cfb | 898 | return 0; |
1da177e4 LT |
899 | } |
900 | ||
901 | subsys_initcall(genhd_device_init); | |
902 | ||
edfaa7c3 KS |
903 | static ssize_t disk_range_show(struct device *dev, |
904 | struct device_attribute *attr, char *buf) | |
1da177e4 | 905 | { |
edfaa7c3 | 906 | struct gendisk *disk = dev_to_disk(dev); |
1da177e4 | 907 | |
edfaa7c3 | 908 | return sprintf(buf, "%d\n", disk->minors); |
1da177e4 LT |
909 | } |
910 | ||
1f014290 TH |
911 | static ssize_t disk_ext_range_show(struct device *dev, |
912 | struct device_attribute *attr, char *buf) | |
913 | { | |
914 | struct gendisk *disk = dev_to_disk(dev); | |
915 | ||
b5d0b9df | 916 | return sprintf(buf, "%d\n", disk_max_parts(disk)); |
1f014290 TH |
917 | } |
918 | ||
edfaa7c3 KS |
919 | static ssize_t disk_removable_show(struct device *dev, |
920 | struct device_attribute *attr, char *buf) | |
a7fd6706 | 921 | { |
edfaa7c3 | 922 | struct gendisk *disk = dev_to_disk(dev); |
a7fd6706 | 923 | |
edfaa7c3 KS |
924 | return sprintf(buf, "%d\n", |
925 | (disk->flags & GENHD_FL_REMOVABLE ? 1 : 0)); | |
a7fd6706 KS |
926 | } |
927 | ||
1c9ce527 KS |
928 | static ssize_t disk_ro_show(struct device *dev, |
929 | struct device_attribute *attr, char *buf) | |
930 | { | |
931 | struct gendisk *disk = dev_to_disk(dev); | |
932 | ||
b7db9956 | 933 | return sprintf(buf, "%d\n", get_disk_ro(disk) ? 1 : 0); |
1c9ce527 KS |
934 | } |
935 | ||
edfaa7c3 KS |
936 | static ssize_t disk_capability_show(struct device *dev, |
937 | struct device_attribute *attr, char *buf) | |
86ce18d7 | 938 | { |
edfaa7c3 KS |
939 | struct gendisk *disk = dev_to_disk(dev); |
940 | ||
941 | return sprintf(buf, "%x\n", disk->flags); | |
86ce18d7 | 942 | } |
edfaa7c3 | 943 | |
c72758f3 MP |
944 | static ssize_t disk_alignment_offset_show(struct device *dev, |
945 | struct device_attribute *attr, | |
946 | char *buf) | |
947 | { | |
948 | struct gendisk *disk = dev_to_disk(dev); | |
949 | ||
950 | return sprintf(buf, "%d\n", queue_alignment_offset(disk->queue)); | |
951 | } | |
952 | ||
86b37281 MP |
953 | static ssize_t disk_discard_alignment_show(struct device *dev, |
954 | struct device_attribute *attr, | |
955 | char *buf) | |
956 | { | |
957 | struct gendisk *disk = dev_to_disk(dev); | |
958 | ||
dd3d145d | 959 | return sprintf(buf, "%d\n", queue_discard_alignment(disk->queue)); |
86b37281 MP |
960 | } |
961 | ||
edfaa7c3 | 962 | static DEVICE_ATTR(range, S_IRUGO, disk_range_show, NULL); |
1f014290 | 963 | static DEVICE_ATTR(ext_range, S_IRUGO, disk_ext_range_show, NULL); |
edfaa7c3 | 964 | static DEVICE_ATTR(removable, S_IRUGO, disk_removable_show, NULL); |
1c9ce527 | 965 | static DEVICE_ATTR(ro, S_IRUGO, disk_ro_show, NULL); |
e5610521 | 966 | static DEVICE_ATTR(size, S_IRUGO, part_size_show, NULL); |
c72758f3 | 967 | static DEVICE_ATTR(alignment_offset, S_IRUGO, disk_alignment_offset_show, NULL); |
86b37281 MP |
968 | static DEVICE_ATTR(discard_alignment, S_IRUGO, disk_discard_alignment_show, |
969 | NULL); | |
edfaa7c3 | 970 | static DEVICE_ATTR(capability, S_IRUGO, disk_capability_show, NULL); |
074a7aca | 971 | static DEVICE_ATTR(stat, S_IRUGO, part_stat_show, NULL); |
316d315b | 972 | static DEVICE_ATTR(inflight, S_IRUGO, part_inflight_show, NULL); |
c17bb495 | 973 | #ifdef CONFIG_FAIL_MAKE_REQUEST |
edfaa7c3 | 974 | static struct device_attribute dev_attr_fail = |
eddb2e26 | 975 | __ATTR(make-it-fail, S_IRUGO|S_IWUSR, part_fail_show, part_fail_store); |
c17bb495 | 976 | #endif |
581d4e28 JA |
977 | #ifdef CONFIG_FAIL_IO_TIMEOUT |
978 | static struct device_attribute dev_attr_fail_timeout = | |
979 | __ATTR(io-timeout-fail, S_IRUGO|S_IWUSR, part_timeout_show, | |
980 | part_timeout_store); | |
981 | #endif | |
edfaa7c3 KS |
982 | |
983 | static struct attribute *disk_attrs[] = { | |
984 | &dev_attr_range.attr, | |
1f014290 | 985 | &dev_attr_ext_range.attr, |
edfaa7c3 | 986 | &dev_attr_removable.attr, |
1c9ce527 | 987 | &dev_attr_ro.attr, |
edfaa7c3 | 988 | &dev_attr_size.attr, |
c72758f3 | 989 | &dev_attr_alignment_offset.attr, |
86b37281 | 990 | &dev_attr_discard_alignment.attr, |
edfaa7c3 KS |
991 | &dev_attr_capability.attr, |
992 | &dev_attr_stat.attr, | |
316d315b | 993 | &dev_attr_inflight.attr, |
edfaa7c3 KS |
994 | #ifdef CONFIG_FAIL_MAKE_REQUEST |
995 | &dev_attr_fail.attr, | |
581d4e28 JA |
996 | #endif |
997 | #ifdef CONFIG_FAIL_IO_TIMEOUT | |
998 | &dev_attr_fail_timeout.attr, | |
edfaa7c3 KS |
999 | #endif |
1000 | NULL | |
1001 | }; | |
1002 | ||
1003 | static struct attribute_group disk_attr_group = { | |
1004 | .attrs = disk_attrs, | |
1005 | }; | |
1006 | ||
a4dbd674 | 1007 | static const struct attribute_group *disk_attr_groups[] = { |
edfaa7c3 KS |
1008 | &disk_attr_group, |
1009 | NULL | |
1da177e4 LT |
1010 | }; |
1011 | ||
540eed56 TH |
1012 | static void disk_free_ptbl_rcu_cb(struct rcu_head *head) |
1013 | { | |
1014 | struct disk_part_tbl *ptbl = | |
1015 | container_of(head, struct disk_part_tbl, rcu_head); | |
1016 | ||
1017 | kfree(ptbl); | |
1018 | } | |
1019 | ||
1020 | /** | |
1021 | * disk_replace_part_tbl - replace disk->part_tbl in RCU-safe way | |
1022 | * @disk: disk to replace part_tbl for | |
1023 | * @new_ptbl: new part_tbl to install | |
1024 | * | |
1025 | * Replace disk->part_tbl with @new_ptbl in RCU-safe way. The | |
1026 | * original ptbl is freed using RCU callback. | |
1027 | * | |
1028 | * LOCKING: | |
1029 | * Matching bd_mutx locked. | |
1030 | */ | |
1031 | static void disk_replace_part_tbl(struct gendisk *disk, | |
1032 | struct disk_part_tbl *new_ptbl) | |
1033 | { | |
1034 | struct disk_part_tbl *old_ptbl = disk->part_tbl; | |
1035 | ||
1036 | rcu_assign_pointer(disk->part_tbl, new_ptbl); | |
a6f23657 JA |
1037 | |
1038 | if (old_ptbl) { | |
1039 | rcu_assign_pointer(old_ptbl->last_lookup, NULL); | |
540eed56 | 1040 | call_rcu(&old_ptbl->rcu_head, disk_free_ptbl_rcu_cb); |
a6f23657 | 1041 | } |
540eed56 TH |
1042 | } |
1043 | ||
1044 | /** | |
1045 | * disk_expand_part_tbl - expand disk->part_tbl | |
1046 | * @disk: disk to expand part_tbl for | |
1047 | * @partno: expand such that this partno can fit in | |
1048 | * | |
1049 | * Expand disk->part_tbl such that @partno can fit in. disk->part_tbl | |
1050 | * uses RCU to allow unlocked dereferencing for stats and other stuff. | |
1051 | * | |
1052 | * LOCKING: | |
1053 | * Matching bd_mutex locked, might sleep. | |
1054 | * | |
1055 | * RETURNS: | |
1056 | * 0 on success, -errno on failure. | |
1057 | */ | |
1058 | int disk_expand_part_tbl(struct gendisk *disk, int partno) | |
1059 | { | |
1060 | struct disk_part_tbl *old_ptbl = disk->part_tbl; | |
1061 | struct disk_part_tbl *new_ptbl; | |
1062 | int len = old_ptbl ? old_ptbl->len : 0; | |
1063 | int target = partno + 1; | |
1064 | size_t size; | |
1065 | int i; | |
1066 | ||
1067 | /* disk_max_parts() is zero during initialization, ignore if so */ | |
1068 | if (disk_max_parts(disk) && target > disk_max_parts(disk)) | |
1069 | return -EINVAL; | |
1070 | ||
1071 | if (target <= len) | |
1072 | return 0; | |
1073 | ||
1074 | size = sizeof(*new_ptbl) + target * sizeof(new_ptbl->part[0]); | |
1075 | new_ptbl = kzalloc_node(size, GFP_KERNEL, disk->node_id); | |
1076 | if (!new_ptbl) | |
1077 | return -ENOMEM; | |
1078 | ||
540eed56 TH |
1079 | new_ptbl->len = target; |
1080 | ||
1081 | for (i = 0; i < len; i++) | |
1082 | rcu_assign_pointer(new_ptbl->part[i], old_ptbl->part[i]); | |
1083 | ||
1084 | disk_replace_part_tbl(disk, new_ptbl); | |
1085 | return 0; | |
1086 | } | |
1087 | ||
edfaa7c3 | 1088 | static void disk_release(struct device *dev) |
1da177e4 | 1089 | { |
edfaa7c3 KS |
1090 | struct gendisk *disk = dev_to_disk(dev); |
1091 | ||
1da177e4 | 1092 | kfree(disk->random); |
540eed56 | 1093 | disk_replace_part_tbl(disk, NULL); |
074a7aca | 1094 | free_part_stats(&disk->part0); |
6d1d8050 | 1095 | free_part_info(&disk->part0); |
1da177e4 LT |
1096 | kfree(disk); |
1097 | } | |
edfaa7c3 KS |
1098 | struct class block_class = { |
1099 | .name = "block", | |
1da177e4 LT |
1100 | }; |
1101 | ||
e454cea2 | 1102 | static char *block_devnode(struct device *dev, mode_t *mode) |
b03f38b6 KS |
1103 | { |
1104 | struct gendisk *disk = dev_to_disk(dev); | |
1105 | ||
e454cea2 KS |
1106 | if (disk->devnode) |
1107 | return disk->devnode(disk, mode); | |
b03f38b6 KS |
1108 | return NULL; |
1109 | } | |
1110 | ||
1826eadf | 1111 | static struct device_type disk_type = { |
edfaa7c3 KS |
1112 | .name = "disk", |
1113 | .groups = disk_attr_groups, | |
1114 | .release = disk_release, | |
e454cea2 | 1115 | .devnode = block_devnode, |
1da177e4 LT |
1116 | }; |
1117 | ||
a6e2ba88 | 1118 | #ifdef CONFIG_PROC_FS |
cf771cb5 TH |
1119 | /* |
1120 | * aggregate disk stat collector. Uses the same stats that the sysfs | |
1121 | * entries do, above, but makes them available through one seq_file. | |
1122 | * | |
1123 | * The output looks suspiciously like /proc/partitions with a bunch of | |
1124 | * extra fields. | |
1125 | */ | |
1126 | static int diskstats_show(struct seq_file *seqf, void *v) | |
1da177e4 LT |
1127 | { |
1128 | struct gendisk *gp = v; | |
e71bf0d0 TH |
1129 | struct disk_part_iter piter; |
1130 | struct hd_struct *hd; | |
1da177e4 | 1131 | char buf[BDEVNAME_SIZE]; |
c9959059 | 1132 | int cpu; |
1da177e4 LT |
1133 | |
1134 | /* | |
ed9e1982 | 1135 | if (&disk_to_dev(gp)->kobj.entry == block_class.devices.next) |
cf771cb5 | 1136 | seq_puts(seqf, "major minor name" |
1da177e4 LT |
1137 | " rio rmerge rsect ruse wio wmerge " |
1138 | "wsect wuse running use aveq" | |
1139 | "\n\n"); | |
1140 | */ | |
1141 | ||
71982a40 | 1142 | disk_part_iter_init(&piter, gp, DISK_PITER_INCL_EMPTY_PART0); |
e71bf0d0 | 1143 | while ((hd = disk_part_iter_next(&piter))) { |
074a7aca | 1144 | cpu = part_stat_lock(); |
c9959059 | 1145 | part_round_stats(cpu, hd); |
074a7aca | 1146 | part_stat_unlock(); |
1f014290 | 1147 | seq_printf(seqf, "%4d %7d %s %lu %lu %llu " |
28f39d55 | 1148 | "%u %lu %lu %llu %u %u %u %u\n", |
f331c029 TH |
1149 | MAJOR(part_devt(hd)), MINOR(part_devt(hd)), |
1150 | disk_name(gp, hd->partno, buf), | |
28f39d55 JM |
1151 | part_stat_read(hd, ios[0]), |
1152 | part_stat_read(hd, merges[0]), | |
1153 | (unsigned long long)part_stat_read(hd, sectors[0]), | |
1154 | jiffies_to_msecs(part_stat_read(hd, ticks[0])), | |
1155 | part_stat_read(hd, ios[1]), | |
1156 | part_stat_read(hd, merges[1]), | |
1157 | (unsigned long long)part_stat_read(hd, sectors[1]), | |
1158 | jiffies_to_msecs(part_stat_read(hd, ticks[1])), | |
316d315b | 1159 | part_in_flight(hd), |
28f39d55 JM |
1160 | jiffies_to_msecs(part_stat_read(hd, io_ticks)), |
1161 | jiffies_to_msecs(part_stat_read(hd, time_in_queue)) | |
1162 | ); | |
1da177e4 | 1163 | } |
e71bf0d0 | 1164 | disk_part_iter_exit(&piter); |
1da177e4 LT |
1165 | |
1166 | return 0; | |
1167 | } | |
1168 | ||
31d85ab2 | 1169 | static const struct seq_operations diskstats_op = { |
def4e38d TH |
1170 | .start = disk_seqf_start, |
1171 | .next = disk_seqf_next, | |
1172 | .stop = disk_seqf_stop, | |
1da177e4 LT |
1173 | .show = diskstats_show |
1174 | }; | |
f500975a | 1175 | |
31d85ab2 AD |
1176 | static int diskstats_open(struct inode *inode, struct file *file) |
1177 | { | |
1178 | return seq_open(file, &diskstats_op); | |
1179 | } | |
1180 | ||
1181 | static const struct file_operations proc_diskstats_operations = { | |
1182 | .open = diskstats_open, | |
1183 | .read = seq_read, | |
1184 | .llseek = seq_lseek, | |
1185 | .release = seq_release, | |
1186 | }; | |
1187 | ||
f500975a AD |
1188 | static int __init proc_genhd_init(void) |
1189 | { | |
31d85ab2 | 1190 | proc_create("diskstats", 0, NULL, &proc_diskstats_operations); |
f500975a AD |
1191 | proc_create("partitions", 0, NULL, &proc_partitions_operations); |
1192 | return 0; | |
1193 | } | |
1194 | module_init(proc_genhd_init); | |
a6e2ba88 | 1195 | #endif /* CONFIG_PROC_FS */ |
1da177e4 | 1196 | |
cf771cb5 | 1197 | dev_t blk_lookup_devt(const char *name, int partno) |
a142be85 | 1198 | { |
def4e38d TH |
1199 | dev_t devt = MKDEV(0, 0); |
1200 | struct class_dev_iter iter; | |
1201 | struct device *dev; | |
a142be85 | 1202 | |
def4e38d TH |
1203 | class_dev_iter_init(&iter, &block_class, NULL, &disk_type); |
1204 | while ((dev = class_dev_iter_next(&iter))) { | |
a142be85 | 1205 | struct gendisk *disk = dev_to_disk(dev); |
548b10eb | 1206 | struct hd_struct *part; |
a142be85 | 1207 | |
3ada8b7e | 1208 | if (strcmp(dev_name(dev), name)) |
f331c029 | 1209 | continue; |
f331c029 | 1210 | |
41b8c853 NB |
1211 | if (partno < disk->minors) { |
1212 | /* We need to return the right devno, even | |
1213 | * if the partition doesn't exist yet. | |
1214 | */ | |
1215 | devt = MKDEV(MAJOR(dev->devt), | |
1216 | MINOR(dev->devt) + partno); | |
1217 | break; | |
1218 | } | |
548b10eb | 1219 | part = disk_get_part(disk, partno); |
2bbedcb4 | 1220 | if (part) { |
f331c029 | 1221 | devt = part_devt(part); |
e71bf0d0 | 1222 | disk_put_part(part); |
548b10eb | 1223 | break; |
def4e38d | 1224 | } |
548b10eb | 1225 | disk_put_part(part); |
5c0ef6d0 | 1226 | } |
def4e38d | 1227 | class_dev_iter_exit(&iter); |
edfaa7c3 KS |
1228 | return devt; |
1229 | } | |
edfaa7c3 KS |
1230 | EXPORT_SYMBOL(blk_lookup_devt); |
1231 | ||
1da177e4 LT |
1232 | struct gendisk *alloc_disk(int minors) |
1233 | { | |
1946089a CL |
1234 | return alloc_disk_node(minors, -1); |
1235 | } | |
689d6fac | 1236 | EXPORT_SYMBOL(alloc_disk); |
1946089a CL |
1237 | |
1238 | struct gendisk *alloc_disk_node(int minors, int node_id) | |
1239 | { | |
1240 | struct gendisk *disk; | |
1241 | ||
94f6030c CL |
1242 | disk = kmalloc_node(sizeof(struct gendisk), |
1243 | GFP_KERNEL | __GFP_ZERO, node_id); | |
1da177e4 | 1244 | if (disk) { |
074a7aca | 1245 | if (!init_part_stats(&disk->part0)) { |
1da177e4 LT |
1246 | kfree(disk); |
1247 | return NULL; | |
1248 | } | |
bf91db18 | 1249 | disk->node_id = node_id; |
540eed56 TH |
1250 | if (disk_expand_part_tbl(disk, 0)) { |
1251 | free_part_stats(&disk->part0); | |
b5d0b9df TH |
1252 | kfree(disk); |
1253 | return NULL; | |
1da177e4 | 1254 | } |
540eed56 | 1255 | disk->part_tbl->part[0] = &disk->part0; |
b5d0b9df | 1256 | |
1da177e4 | 1257 | disk->minors = minors; |
1da177e4 | 1258 | rand_initialize_disk(disk); |
ed9e1982 TH |
1259 | disk_to_dev(disk)->class = &block_class; |
1260 | disk_to_dev(disk)->type = &disk_type; | |
1261 | device_initialize(disk_to_dev(disk)); | |
1da177e4 LT |
1262 | } |
1263 | return disk; | |
1264 | } | |
1946089a | 1265 | EXPORT_SYMBOL(alloc_disk_node); |
1da177e4 LT |
1266 | |
1267 | struct kobject *get_disk(struct gendisk *disk) | |
1268 | { | |
1269 | struct module *owner; | |
1270 | struct kobject *kobj; | |
1271 | ||
1272 | if (!disk->fops) | |
1273 | return NULL; | |
1274 | owner = disk->fops->owner; | |
1275 | if (owner && !try_module_get(owner)) | |
1276 | return NULL; | |
ed9e1982 | 1277 | kobj = kobject_get(&disk_to_dev(disk)->kobj); |
1da177e4 LT |
1278 | if (kobj == NULL) { |
1279 | module_put(owner); | |
1280 | return NULL; | |
1281 | } | |
1282 | return kobj; | |
1283 | ||
1284 | } | |
1285 | ||
1286 | EXPORT_SYMBOL(get_disk); | |
1287 | ||
1288 | void put_disk(struct gendisk *disk) | |
1289 | { | |
1290 | if (disk) | |
ed9e1982 | 1291 | kobject_put(&disk_to_dev(disk)->kobj); |
1da177e4 LT |
1292 | } |
1293 | ||
1294 | EXPORT_SYMBOL(put_disk); | |
1295 | ||
e3264a4d HR |
1296 | static void set_disk_ro_uevent(struct gendisk *gd, int ro) |
1297 | { | |
1298 | char event[] = "DISK_RO=1"; | |
1299 | char *envp[] = { event, NULL }; | |
1300 | ||
1301 | if (!ro) | |
1302 | event[8] = '0'; | |
1303 | kobject_uevent_env(&disk_to_dev(gd)->kobj, KOBJ_CHANGE, envp); | |
1304 | } | |
1305 | ||
1da177e4 LT |
1306 | void set_device_ro(struct block_device *bdev, int flag) |
1307 | { | |
b7db9956 | 1308 | bdev->bd_part->policy = flag; |
1da177e4 LT |
1309 | } |
1310 | ||
1311 | EXPORT_SYMBOL(set_device_ro); | |
1312 | ||
1313 | void set_disk_ro(struct gendisk *disk, int flag) | |
1314 | { | |
e71bf0d0 TH |
1315 | struct disk_part_iter piter; |
1316 | struct hd_struct *part; | |
1317 | ||
e3264a4d HR |
1318 | if (disk->part0.policy != flag) { |
1319 | set_disk_ro_uevent(disk, flag); | |
1320 | disk->part0.policy = flag; | |
1321 | } | |
1322 | ||
1323 | disk_part_iter_init(&piter, disk, DISK_PITER_INCL_EMPTY); | |
e71bf0d0 TH |
1324 | while ((part = disk_part_iter_next(&piter))) |
1325 | part->policy = flag; | |
1326 | disk_part_iter_exit(&piter); | |
1da177e4 LT |
1327 | } |
1328 | ||
1329 | EXPORT_SYMBOL(set_disk_ro); | |
1330 | ||
1331 | int bdev_read_only(struct block_device *bdev) | |
1332 | { | |
1333 | if (!bdev) | |
1334 | return 0; | |
b7db9956 | 1335 | return bdev->bd_part->policy; |
1da177e4 LT |
1336 | } |
1337 | ||
1338 | EXPORT_SYMBOL(bdev_read_only); | |
1339 | ||
cf771cb5 | 1340 | int invalidate_partition(struct gendisk *disk, int partno) |
1da177e4 LT |
1341 | { |
1342 | int res = 0; | |
cf771cb5 | 1343 | struct block_device *bdev = bdget_disk(disk, partno); |
1da177e4 | 1344 | if (bdev) { |
2ef41634 CH |
1345 | fsync_bdev(bdev); |
1346 | res = __invalidate_device(bdev); | |
1da177e4 LT |
1347 | bdput(bdev); |
1348 | } | |
1349 | return res; | |
1350 | } | |
1351 | ||
1352 | EXPORT_SYMBOL(invalidate_partition); |