Commit | Line | Data |
---|---|---|
3dcf60bc | 1 | // SPDX-License-Identifier: GPL-2.0 |
1da177e4 LT |
2 | /* |
3 | * gendisk handling | |
7b51e703 CH |
4 | * |
5 | * Portions Copyright (C) 2020 Christoph Hellwig | |
1da177e4 LT |
6 | */ |
7 | ||
1da177e4 | 8 | #include <linux/module.h> |
3ad5cee5 | 9 | #include <linux/ctype.h> |
1da177e4 LT |
10 | #include <linux/fs.h> |
11 | #include <linux/genhd.h> | |
b446b60e | 12 | #include <linux/kdev_t.h> |
1da177e4 LT |
13 | #include <linux/kernel.h> |
14 | #include <linux/blkdev.h> | |
66114cad | 15 | #include <linux/backing-dev.h> |
1da177e4 LT |
16 | #include <linux/init.h> |
17 | #include <linux/spinlock.h> | |
f500975a | 18 | #include <linux/proc_fs.h> |
1da177e4 LT |
19 | #include <linux/seq_file.h> |
20 | #include <linux/slab.h> | |
21 | #include <linux/kmod.h> | |
58383af6 | 22 | #include <linux/mutex.h> |
bcce3de1 | 23 | #include <linux/idr.h> |
77ea887e | 24 | #include <linux/log2.h> |
25e823c8 | 25 | #include <linux/pm_runtime.h> |
99e6608c | 26 | #include <linux/badblocks.h> |
1da177e4 | 27 | |
ff88972c AB |
28 | #include "blk.h" |
29 | ||
31eb6186 | 30 | static struct kobject *block_depr; |
1da177e4 | 31 | |
bcce3de1 | 32 | /* for extended dynamic devt allocation, currently only one major is used */ |
ce23bba8 | 33 | #define NR_EXT_DEVT (1 << MINORBITS) |
22ae8ce8 | 34 | static DEFINE_IDA(ext_devt_ida); |
bcce3de1 | 35 | |
a782483c CH |
36 | void set_capacity(struct gendisk *disk, sector_t sectors) |
37 | { | |
cb8432d6 | 38 | struct block_device *bdev = disk->part0; |
a782483c | 39 | |
0f472277 | 40 | spin_lock(&bdev->bd_size_lock); |
a782483c | 41 | i_size_write(bdev->bd_inode, (loff_t)sectors << SECTOR_SHIFT); |
0f472277 | 42 | spin_unlock(&bdev->bd_size_lock); |
a782483c CH |
43 | } |
44 | EXPORT_SYMBOL(set_capacity); | |
45 | ||
e598a72f | 46 | /* |
449f4ec9 CH |
47 | * Set disk capacity and notify if the size is not currently zero and will not |
48 | * be set to zero. Returns true if a uevent was sent, otherwise false. | |
e598a72f | 49 | */ |
449f4ec9 | 50 | bool set_capacity_and_notify(struct gendisk *disk, sector_t size) |
e598a72f BS |
51 | { |
52 | sector_t capacity = get_capacity(disk); | |
a782483c | 53 | char *envp[] = { "RESIZE=1", NULL }; |
e598a72f BS |
54 | |
55 | set_capacity(disk, size); | |
e598a72f | 56 | |
a782483c CH |
57 | /* |
58 | * Only print a message and send a uevent if the gendisk is user visible | |
59 | * and alive. This avoids spamming the log and udev when setting the | |
60 | * initial capacity during probing. | |
61 | */ | |
62 | if (size == capacity || | |
63 | (disk->flags & (GENHD_FL_UP | GENHD_FL_HIDDEN)) != GENHD_FL_UP) | |
64 | return false; | |
e598a72f | 65 | |
a782483c | 66 | pr_info("%s: detected capacity change from %lld to %lld\n", |
452c0bf8 | 67 | disk->disk_name, capacity, size); |
7e890c37 | 68 | |
a782483c CH |
69 | /* |
70 | * Historically we did not send a uevent for changes to/from an empty | |
71 | * device. | |
72 | */ | |
73 | if (!capacity || !size) | |
74 | return false; | |
75 | kobject_uevent_env(&disk_to_dev(disk)->kobj, KOBJ_CHANGE, envp); | |
76 | return true; | |
e598a72f | 77 | } |
449f4ec9 | 78 | EXPORT_SYMBOL_GPL(set_capacity_and_notify); |
e598a72f | 79 | |
5cbd28e3 | 80 | /* |
abd2864a CH |
81 | * Format the device name of the indicated block device into the supplied buffer |
82 | * and return a pointer to that same buffer for convenience. | |
83 | * | |
84 | * Note: do not use this in new code, use the %pg specifier to sprintf and | |
85 | * printk insted. | |
5cbd28e3 | 86 | */ |
abd2864a | 87 | const char *bdevname(struct block_device *bdev, char *buf) |
5cbd28e3 | 88 | { |
abd2864a CH |
89 | struct gendisk *hd = bdev->bd_disk; |
90 | int partno = bdev->bd_partno; | |
91 | ||
5cbd28e3 CH |
92 | if (!partno) |
93 | snprintf(buf, BDEVNAME_SIZE, "%s", hd->disk_name); | |
94 | else if (isdigit(hd->disk_name[strlen(hd->disk_name)-1])) | |
95 | snprintf(buf, BDEVNAME_SIZE, "%sp%d", hd->disk_name, partno); | |
96 | else | |
97 | snprintf(buf, BDEVNAME_SIZE, "%s%d", hd->disk_name, partno); | |
98 | ||
99 | return buf; | |
100 | } | |
5cbd28e3 | 101 | EXPORT_SYMBOL(bdevname); |
e598a72f | 102 | |
0d02129e CH |
103 | static void part_stat_read_all(struct block_device *part, |
104 | struct disk_stats *stat) | |
ea18e0f0 KK |
105 | { |
106 | int cpu; | |
107 | ||
108 | memset(stat, 0, sizeof(struct disk_stats)); | |
109 | for_each_possible_cpu(cpu) { | |
0d02129e | 110 | struct disk_stats *ptr = per_cpu_ptr(part->bd_stats, cpu); |
ea18e0f0 KK |
111 | int group; |
112 | ||
113 | for (group = 0; group < NR_STAT_GROUPS; group++) { | |
114 | stat->nsecs[group] += ptr->nsecs[group]; | |
115 | stat->sectors[group] += ptr->sectors[group]; | |
116 | stat->ios[group] += ptr->ios[group]; | |
117 | stat->merges[group] += ptr->merges[group]; | |
118 | } | |
119 | ||
120 | stat->io_ticks += ptr->io_ticks; | |
ea18e0f0 KK |
121 | } |
122 | } | |
ea18e0f0 | 123 | |
8446fe92 | 124 | static unsigned int part_in_flight(struct block_device *part) |
f299b7c7 | 125 | { |
b2f609e1 | 126 | unsigned int inflight = 0; |
1226b8dd | 127 | int cpu; |
f299b7c7 | 128 | |
1226b8dd | 129 | for_each_possible_cpu(cpu) { |
e016b782 MP |
130 | inflight += part_stat_local_read_cpu(part, in_flight[0], cpu) + |
131 | part_stat_local_read_cpu(part, in_flight[1], cpu); | |
1226b8dd | 132 | } |
e016b782 MP |
133 | if ((int)inflight < 0) |
134 | inflight = 0; | |
1226b8dd | 135 | |
e016b782 | 136 | return inflight; |
f299b7c7 JA |
137 | } |
138 | ||
8446fe92 CH |
139 | static void part_in_flight_rw(struct block_device *part, |
140 | unsigned int inflight[2]) | |
bf0ddaba | 141 | { |
1226b8dd MP |
142 | int cpu; |
143 | ||
1226b8dd MP |
144 | inflight[0] = 0; |
145 | inflight[1] = 0; | |
146 | for_each_possible_cpu(cpu) { | |
147 | inflight[0] += part_stat_local_read_cpu(part, in_flight[0], cpu); | |
148 | inflight[1] += part_stat_local_read_cpu(part, in_flight[1], cpu); | |
149 | } | |
150 | if ((int)inflight[0] < 0) | |
151 | inflight[0] = 0; | |
152 | if ((int)inflight[1] < 0) | |
153 | inflight[1] = 0; | |
bf0ddaba OS |
154 | } |
155 | ||
1da177e4 LT |
156 | /* |
157 | * Can be deleted altogether. Later. | |
158 | * | |
159 | */ | |
133d55cd | 160 | #define BLKDEV_MAJOR_HASH_SIZE 255 |
1da177e4 LT |
161 | static struct blk_major_name { |
162 | struct blk_major_name *next; | |
163 | int major; | |
164 | char name[16]; | |
a160c615 | 165 | void (*probe)(dev_t devt); |
68eef3b4 | 166 | } *major_names[BLKDEV_MAJOR_HASH_SIZE]; |
e49fbbbf | 167 | static DEFINE_MUTEX(major_names_lock); |
1da177e4 LT |
168 | |
169 | /* index in the above - for now: assume no multimajor ranges */ | |
e61eb2e9 | 170 | static inline int major_to_index(unsigned major) |
1da177e4 | 171 | { |
68eef3b4 | 172 | return major % BLKDEV_MAJOR_HASH_SIZE; |
7170be5f NH |
173 | } |
174 | ||
68eef3b4 | 175 | #ifdef CONFIG_PROC_FS |
cf771cb5 | 176 | void blkdev_show(struct seq_file *seqf, off_t offset) |
7170be5f | 177 | { |
68eef3b4 | 178 | struct blk_major_name *dp; |
7170be5f | 179 | |
e49fbbbf | 180 | mutex_lock(&major_names_lock); |
133d55cd LG |
181 | for (dp = major_names[major_to_index(offset)]; dp; dp = dp->next) |
182 | if (dp->major == offset) | |
cf771cb5 | 183 | seq_printf(seqf, "%3d %s\n", dp->major, dp->name); |
e49fbbbf | 184 | mutex_unlock(&major_names_lock); |
1da177e4 | 185 | } |
68eef3b4 | 186 | #endif /* CONFIG_PROC_FS */ |
1da177e4 | 187 | |
9e8c0bcc | 188 | /** |
e2b6b301 | 189 | * __register_blkdev - register a new block device |
9e8c0bcc | 190 | * |
f33ff110 SB |
191 | * @major: the requested major device number [1..BLKDEV_MAJOR_MAX-1]. If |
192 | * @major = 0, try to allocate any unused major number. | |
9e8c0bcc | 193 | * @name: the name of the new block device as a zero terminated string |
e2b6b301 | 194 | * @probe: allback that is called on access to any minor number of @major |
9e8c0bcc MN |
195 | * |
196 | * The @name must be unique within the system. | |
197 | * | |
0e056eb5 | 198 | * The return value depends on the @major input parameter: |
199 | * | |
f33ff110 SB |
200 | * - if a major device number was requested in range [1..BLKDEV_MAJOR_MAX-1] |
201 | * then the function returns zero on success, or a negative error code | |
0e056eb5 | 202 | * - if any unused major number was requested with @major = 0 parameter |
9e8c0bcc | 203 | * then the return value is the allocated major number in range |
f33ff110 SB |
204 | * [1..BLKDEV_MAJOR_MAX-1] or a negative error code otherwise |
205 | * | |
206 | * See Documentation/admin-guide/devices.txt for the list of allocated | |
207 | * major numbers. | |
e2b6b301 CH |
208 | * |
209 | * Use register_blkdev instead for any new code. | |
9e8c0bcc | 210 | */ |
a160c615 CH |
211 | int __register_blkdev(unsigned int major, const char *name, |
212 | void (*probe)(dev_t devt)) | |
1da177e4 LT |
213 | { |
214 | struct blk_major_name **n, *p; | |
215 | int index, ret = 0; | |
216 | ||
e49fbbbf | 217 | mutex_lock(&major_names_lock); |
1da177e4 LT |
218 | |
219 | /* temporary */ | |
220 | if (major == 0) { | |
221 | for (index = ARRAY_SIZE(major_names)-1; index > 0; index--) { | |
222 | if (major_names[index] == NULL) | |
223 | break; | |
224 | } | |
225 | ||
226 | if (index == 0) { | |
dfc76d11 KP |
227 | printk("%s: failed to get major for %s\n", |
228 | __func__, name); | |
1da177e4 LT |
229 | ret = -EBUSY; |
230 | goto out; | |
231 | } | |
232 | major = index; | |
233 | ret = major; | |
234 | } | |
235 | ||
133d55cd | 236 | if (major >= BLKDEV_MAJOR_MAX) { |
dfc76d11 KP |
237 | pr_err("%s: major requested (%u) is greater than the maximum (%u) for %s\n", |
238 | __func__, major, BLKDEV_MAJOR_MAX-1, name); | |
133d55cd LG |
239 | |
240 | ret = -EINVAL; | |
241 | goto out; | |
242 | } | |
243 | ||
1da177e4 LT |
244 | p = kmalloc(sizeof(struct blk_major_name), GFP_KERNEL); |
245 | if (p == NULL) { | |
246 | ret = -ENOMEM; | |
247 | goto out; | |
248 | } | |
249 | ||
250 | p->major = major; | |
a160c615 | 251 | p->probe = probe; |
1da177e4 LT |
252 | strlcpy(p->name, name, sizeof(p->name)); |
253 | p->next = NULL; | |
254 | index = major_to_index(major); | |
255 | ||
256 | for (n = &major_names[index]; *n; n = &(*n)->next) { | |
257 | if ((*n)->major == major) | |
258 | break; | |
259 | } | |
260 | if (!*n) | |
261 | *n = p; | |
262 | else | |
263 | ret = -EBUSY; | |
264 | ||
265 | if (ret < 0) { | |
f33ff110 | 266 | printk("register_blkdev: cannot get major %u for %s\n", |
1da177e4 LT |
267 | major, name); |
268 | kfree(p); | |
269 | } | |
270 | out: | |
e49fbbbf | 271 | mutex_unlock(&major_names_lock); |
1da177e4 LT |
272 | return ret; |
273 | } | |
a160c615 | 274 | EXPORT_SYMBOL(__register_blkdev); |
1da177e4 | 275 | |
f4480240 | 276 | void unregister_blkdev(unsigned int major, const char *name) |
1da177e4 LT |
277 | { |
278 | struct blk_major_name **n; | |
279 | struct blk_major_name *p = NULL; | |
280 | int index = major_to_index(major); | |
1da177e4 | 281 | |
e49fbbbf | 282 | mutex_lock(&major_names_lock); |
1da177e4 LT |
283 | for (n = &major_names[index]; *n; n = &(*n)->next) |
284 | if ((*n)->major == major) | |
285 | break; | |
294462a5 AM |
286 | if (!*n || strcmp((*n)->name, name)) { |
287 | WARN_ON(1); | |
294462a5 | 288 | } else { |
1da177e4 LT |
289 | p = *n; |
290 | *n = p->next; | |
291 | } | |
e49fbbbf | 292 | mutex_unlock(&major_names_lock); |
1da177e4 | 293 | kfree(p); |
1da177e4 LT |
294 | } |
295 | ||
296 | EXPORT_SYMBOL(unregister_blkdev); | |
297 | ||
870d6656 TH |
298 | /** |
299 | * blk_mangle_minor - scatter minor numbers apart | |
300 | * @minor: minor number to mangle | |
301 | * | |
302 | * Scatter consecutively allocated @minor number apart if MANGLE_DEVT | |
303 | * is enabled. Mangling twice gives the original value. | |
304 | * | |
305 | * RETURNS: | |
306 | * Mangled value. | |
307 | * | |
308 | * CONTEXT: | |
309 | * Don't care. | |
310 | */ | |
311 | static int blk_mangle_minor(int minor) | |
312 | { | |
313 | #ifdef CONFIG_DEBUG_BLOCK_EXT_DEVT | |
314 | int i; | |
315 | ||
316 | for (i = 0; i < MINORBITS / 2; i++) { | |
317 | int low = minor & (1 << i); | |
318 | int high = minor & (1 << (MINORBITS - 1 - i)); | |
319 | int distance = MINORBITS - 1 - 2 * i; | |
320 | ||
321 | minor ^= low | high; /* clear both bits */ | |
322 | low <<= distance; /* swap the positions */ | |
323 | high >>= distance; | |
324 | minor |= low | high; /* and set */ | |
325 | } | |
326 | #endif | |
327 | return minor; | |
328 | } | |
329 | ||
7c3f828b | 330 | int blk_alloc_ext_minor(void) |
bcce3de1 | 331 | { |
bab998d6 | 332 | int idx; |
bcce3de1 | 333 | |
22ae8ce8 | 334 | idx = ida_alloc_range(&ext_devt_ida, 0, NR_EXT_DEVT, GFP_KERNEL); |
7c3f828b CH |
335 | if (idx < 0) { |
336 | if (idx == -ENOSPC) | |
337 | return -EBUSY; | |
338 | return idx; | |
339 | } | |
340 | return blk_mangle_minor(idx); | |
bcce3de1 TH |
341 | } |
342 | ||
7c3f828b | 343 | void blk_free_ext_minor(unsigned int minor) |
bcce3de1 | 344 | { |
7c3f828b | 345 | ida_free(&ext_devt_ida, blk_mangle_minor(minor)); |
6fcc44d1 YY |
346 | } |
347 | ||
1f014290 TH |
348 | static char *bdevt_str(dev_t devt, char *buf) |
349 | { | |
350 | if (MAJOR(devt) <= 0xff && MINOR(devt) <= 0xff) { | |
351 | char tbuf[BDEVT_SIZE]; | |
352 | snprintf(tbuf, BDEVT_SIZE, "%02x%02x", MAJOR(devt), MINOR(devt)); | |
353 | snprintf(buf, BDEVT_SIZE, "%-9s", tbuf); | |
354 | } else | |
355 | snprintf(buf, BDEVT_SIZE, "%03x:%05x", MAJOR(devt), MINOR(devt)); | |
356 | ||
357 | return buf; | |
358 | } | |
359 | ||
bc359d03 CH |
360 | void disk_uevent(struct gendisk *disk, enum kobject_action action) |
361 | { | |
bc359d03 | 362 | struct block_device *part; |
3212135a | 363 | unsigned long idx; |
bc359d03 | 364 | |
3212135a CH |
365 | rcu_read_lock(); |
366 | xa_for_each(&disk->part_tbl, idx, part) { | |
367 | if (bdev_is_partition(part) && !bdev_nr_sectors(part)) | |
368 | continue; | |
498dcc13 | 369 | if (!kobject_get_unless_zero(&part->bd_device.kobj)) |
3212135a CH |
370 | continue; |
371 | ||
372 | rcu_read_unlock(); | |
bc359d03 | 373 | kobject_uevent(bdev_kobj(part), action); |
498dcc13 | 374 | put_device(&part->bd_device); |
3212135a CH |
375 | rcu_read_lock(); |
376 | } | |
377 | rcu_read_unlock(); | |
bc359d03 CH |
378 | } |
379 | EXPORT_SYMBOL_GPL(disk_uevent); | |
380 | ||
9301fe73 CH |
381 | static void disk_scan_partitions(struct gendisk *disk) |
382 | { | |
383 | struct block_device *bdev; | |
384 | ||
385 | if (!get_capacity(disk) || !disk_part_scan_enabled(disk)) | |
386 | return; | |
387 | ||
388 | set_bit(GD_NEED_PART_SCAN, &disk->state); | |
389 | bdev = blkdev_get_by_dev(disk_devt(disk), FMODE_READ, NULL); | |
390 | if (!IS_ERR(bdev)) | |
391 | blkdev_put(bdev, FMODE_READ); | |
392 | } | |
393 | ||
fef912bf HR |
394 | static void register_disk(struct device *parent, struct gendisk *disk, |
395 | const struct attribute_group **groups) | |
d2bf1b67 TH |
396 | { |
397 | struct device *ddev = disk_to_dev(disk); | |
d2bf1b67 TH |
398 | int err; |
399 | ||
e63a46be | 400 | ddev->parent = parent; |
d2bf1b67 | 401 | |
ffc8b308 | 402 | dev_set_name(ddev, "%s", disk->disk_name); |
d2bf1b67 TH |
403 | |
404 | /* delay uevents, until we scanned partition table */ | |
405 | dev_set_uevent_suppress(ddev, 1); | |
406 | ||
fef912bf HR |
407 | if (groups) { |
408 | WARN_ON(ddev->groups); | |
409 | ddev->groups = groups; | |
410 | } | |
d2bf1b67 TH |
411 | if (device_add(ddev)) |
412 | return; | |
413 | if (!sysfs_deprecated) { | |
414 | err = sysfs_create_link(block_depr, &ddev->kobj, | |
415 | kobject_name(&ddev->kobj)); | |
416 | if (err) { | |
417 | device_del(ddev); | |
418 | return; | |
419 | } | |
420 | } | |
25e823c8 ML |
421 | |
422 | /* | |
423 | * avoid probable deadlock caused by allocating memory with | |
424 | * GFP_KERNEL in runtime_resume callback of its all ancestor | |
425 | * devices | |
426 | */ | |
427 | pm_runtime_set_memalloc_noio(ddev, true); | |
428 | ||
cb8432d6 CH |
429 | disk->part0->bd_holder_dir = |
430 | kobject_create_and_add("holders", &ddev->kobj); | |
d2bf1b67 TH |
431 | disk->slave_dir = kobject_create_and_add("slaves", &ddev->kobj); |
432 | ||
9ec49144 | 433 | if (disk->flags & GENHD_FL_HIDDEN) |
8ddcd653 | 434 | return; |
8ddcd653 | 435 | |
9301fe73 | 436 | disk_scan_partitions(disk); |
d2bf1b67 | 437 | |
bc359d03 | 438 | /* announce the disk and partitions after all partitions are created */ |
d2bf1b67 | 439 | dev_set_uevent_suppress(ddev, 0); |
bc359d03 | 440 | disk_uevent(disk, KOBJ_ADD); |
8ddcd653 | 441 | |
4d7c1d3f | 442 | if (disk->queue->backing_dev_info->dev) { |
443 | err = sysfs_create_link(&ddev->kobj, | |
444 | &disk->queue->backing_dev_info->dev->kobj, | |
445 | "bdi"); | |
446 | WARN_ON(err); | |
447 | } | |
d2bf1b67 TH |
448 | } |
449 | ||
1da177e4 | 450 | /** |
fa70d2e2 | 451 | * __device_add_disk - add disk information to kernel list |
e63a46be | 452 | * @parent: parent device for the disk |
1da177e4 | 453 | * @disk: per-device partitioning information |
fef912bf | 454 | * @groups: Additional per-device sysfs groups |
fa70d2e2 | 455 | * @register_queue: register the queue if set to true |
1da177e4 LT |
456 | * |
457 | * This function registers the partitioning information in @disk | |
458 | * with the kernel. | |
3e1a7ff8 TH |
459 | * |
460 | * FIXME: error handling | |
1da177e4 | 461 | */ |
fa70d2e2 | 462 | static void __device_add_disk(struct device *parent, struct gendisk *disk, |
fef912bf | 463 | const struct attribute_group **groups, |
fa70d2e2 | 464 | bool register_queue) |
1da177e4 | 465 | { |
7c3f828b | 466 | int ret; |
cf0ca9fe | 467 | |
737eb78e DLM |
468 | /* |
469 | * The disk queue should now be all set with enough information about | |
470 | * the device for the elevator code to pick an adequate default | |
471 | * elevator if one is needed, that is, for devices requesting queue | |
472 | * registration. | |
473 | */ | |
474 | if (register_queue) | |
475 | elevator_init_mq(disk->queue); | |
476 | ||
7c3f828b CH |
477 | /* |
478 | * If the driver provides an explicit major number it also must provide | |
479 | * the number of minors numbers supported, and those will be used to | |
480 | * setup the gendisk. | |
481 | * Otherwise just allocate the device numbers for both the whole device | |
482 | * and all partitions from the extended dev_t space. | |
3e1a7ff8 | 483 | */ |
7c3f828b CH |
484 | if (disk->major) { |
485 | WARN_ON(!disk->minors); | |
2e3c73fa CH |
486 | |
487 | if (disk->minors > DISK_MAX_PARTS) { | |
488 | pr_err("block: can't allocate more than %d partitions\n", | |
489 | DISK_MAX_PARTS); | |
490 | disk->minors = DISK_MAX_PARTS; | |
491 | } | |
7c3f828b CH |
492 | } else { |
493 | WARN_ON(disk->minors); | |
3e1a7ff8 | 494 | |
7c3f828b CH |
495 | ret = blk_alloc_ext_minor(); |
496 | if (ret < 0) { | |
497 | WARN_ON(1); | |
498 | return; | |
499 | } | |
500 | disk->major = BLOCK_EXT_MAJOR; | |
501 | disk->first_minor = MINOR(ret); | |
0d1feb72 | 502 | disk->flags |= GENHD_FL_EXT_DEVT; |
3e1a7ff8 | 503 | } |
7c3f828b CH |
504 | |
505 | disk->flags |= GENHD_FL_UP; | |
3e1a7ff8 | 506 | |
9f53d2fe SG |
507 | disk_alloc_events(disk); |
508 | ||
8ddcd653 CH |
509 | if (disk->flags & GENHD_FL_HIDDEN) { |
510 | /* | |
511 | * Don't let hidden disks show up in /proc/partitions, | |
512 | * and don't bother scanning for partitions either. | |
513 | */ | |
514 | disk->flags |= GENHD_FL_SUPPRESS_PARTITION_INFO; | |
515 | disk->flags |= GENHD_FL_NO_PART_SCAN; | |
516 | } else { | |
3c5d202b CH |
517 | struct backing_dev_info *bdi = disk->queue->backing_dev_info; |
518 | struct device *dev = disk_to_dev(disk); | |
3a92168b | 519 | |
8ddcd653 | 520 | /* Register BDI before referencing it from bdev */ |
7c3f828b CH |
521 | dev->devt = MKDEV(disk->major, disk->first_minor); |
522 | ret = bdi_register(bdi, "%u:%u", | |
523 | disk->major, disk->first_minor); | |
3a92168b | 524 | WARN_ON(ret); |
3c5d202b | 525 | bdi_set_owner(bdi, dev); |
7c3f828b | 526 | bdev_add(disk->part0, dev->devt); |
8ddcd653 | 527 | } |
fef912bf | 528 | register_disk(parent, disk, groups); |
fa70d2e2 MS |
529 | if (register_queue) |
530 | blk_register_queue(disk); | |
cf0ca9fe | 531 | |
523e1d39 TH |
532 | /* |
533 | * Take an extra ref on queue which will be put on disk_release() | |
534 | * so that it sticks around as long as @disk is there. | |
535 | */ | |
958229a7 CH |
536 | if (blk_get_queue(disk->queue)) |
537 | set_bit(GD_QUEUE_REF, &disk->state); | |
538 | else | |
539 | WARN_ON_ONCE(1); | |
523e1d39 | 540 | |
77ea887e | 541 | disk_add_events(disk); |
25520d55 | 542 | blk_integrity_add(disk); |
1da177e4 | 543 | } |
fa70d2e2 | 544 | |
fef912bf HR |
545 | void device_add_disk(struct device *parent, struct gendisk *disk, |
546 | const struct attribute_group **groups) | |
547 | ||
fa70d2e2 | 548 | { |
fef912bf | 549 | __device_add_disk(parent, disk, groups, true); |
fa70d2e2 | 550 | } |
e63a46be | 551 | EXPORT_SYMBOL(device_add_disk); |
1da177e4 | 552 | |
fa70d2e2 MS |
553 | void device_add_disk_no_queue_reg(struct device *parent, struct gendisk *disk) |
554 | { | |
fef912bf | 555 | __device_add_disk(parent, disk, NULL, false); |
fa70d2e2 MS |
556 | } |
557 | EXPORT_SYMBOL(device_add_disk_no_queue_reg); | |
558 | ||
b5bd357c LC |
559 | /** |
560 | * del_gendisk - remove the gendisk | |
561 | * @disk: the struct gendisk to remove | |
562 | * | |
563 | * Removes the gendisk and all its associated resources. This deletes the | |
564 | * partitions associated with the gendisk, and unregisters the associated | |
565 | * request_queue. | |
566 | * | |
567 | * This is the counter to the respective __device_add_disk() call. | |
568 | * | |
569 | * The final removal of the struct gendisk happens when its refcount reaches 0 | |
570 | * with put_disk(), which should be called after del_gendisk(), if | |
571 | * __device_add_disk() was used. | |
e8c7d14a LC |
572 | * |
573 | * Drivers exist which depend on the release of the gendisk to be synchronous, | |
574 | * it should not be deferred. | |
575 | * | |
576 | * Context: can sleep | |
b5bd357c | 577 | */ |
d2bf1b67 | 578 | void del_gendisk(struct gendisk *disk) |
1da177e4 | 579 | { |
e8c7d14a LC |
580 | might_sleep(); |
581 | ||
6b3ba976 CH |
582 | if (WARN_ON_ONCE(!disk->queue)) |
583 | return; | |
584 | ||
25520d55 | 585 | blk_integrity_del(disk); |
77ea887e TH |
586 | disk_del_events(disk); |
587 | ||
a8698707 | 588 | mutex_lock(&disk->open_mutex); |
d7a66574 | 589 | remove_inode_hash(disk->part0->bd_inode); |
6c60ff04 | 590 | disk->flags &= ~GENHD_FL_UP; |
d3c4a43d | 591 | blk_drop_partitions(disk); |
a8698707 | 592 | mutex_unlock(&disk->open_mutex); |
c76f48eb | 593 | |
45611837 CH |
594 | fsync_bdev(disk->part0); |
595 | __invalidate_device(disk->part0, true); | |
596 | ||
d2bf1b67 | 597 | set_capacity(disk, 0); |
d2bf1b67 | 598 | |
6b3ba976 | 599 | if (!(disk->flags & GENHD_FL_HIDDEN)) { |
8ddcd653 | 600 | sysfs_remove_link(&disk_to_dev(disk)->kobj, "bdi"); |
6b3ba976 | 601 | |
90f16fdd JK |
602 | /* |
603 | * Unregister bdi before releasing device numbers (as they can | |
604 | * get reused and we'd get clashes in sysfs). | |
605 | */ | |
6b3ba976 | 606 | bdi_unregister(disk->queue->backing_dev_info); |
90f16fdd | 607 | } |
d2bf1b67 | 608 | |
6b3ba976 | 609 | blk_unregister_queue(disk); |
d2bf1b67 | 610 | |
cb8432d6 | 611 | kobject_put(disk->part0->bd_holder_dir); |
d2bf1b67 | 612 | kobject_put(disk->slave_dir); |
d2bf1b67 | 613 | |
8446fe92 | 614 | part_stat_set_all(disk->part0, 0); |
cb8432d6 | 615 | disk->part0->bd_stamp = 0; |
d2bf1b67 TH |
616 | if (!sysfs_deprecated) |
617 | sysfs_remove_link(block_depr, dev_name(disk_to_dev(disk))); | |
25e823c8 | 618 | pm_runtime_set_memalloc_noio(disk_to_dev(disk), false); |
d2bf1b67 | 619 | device_del(disk_to_dev(disk)); |
1da177e4 | 620 | } |
d2bf1b67 | 621 | EXPORT_SYMBOL(del_gendisk); |
1da177e4 | 622 | |
99e6608c VV |
623 | /* sysfs access to bad-blocks list. */ |
624 | static ssize_t disk_badblocks_show(struct device *dev, | |
625 | struct device_attribute *attr, | |
626 | char *page) | |
627 | { | |
628 | struct gendisk *disk = dev_to_disk(dev); | |
629 | ||
630 | if (!disk->bb) | |
631 | return sprintf(page, "\n"); | |
632 | ||
633 | return badblocks_show(disk->bb, page, 0); | |
634 | } | |
635 | ||
636 | static ssize_t disk_badblocks_store(struct device *dev, | |
637 | struct device_attribute *attr, | |
638 | const char *page, size_t len) | |
639 | { | |
640 | struct gendisk *disk = dev_to_disk(dev); | |
641 | ||
642 | if (!disk->bb) | |
643 | return -ENXIO; | |
644 | ||
645 | return badblocks_store(disk->bb, page, len, 0); | |
646 | } | |
647 | ||
22ae8ce8 | 648 | void blk_request_module(dev_t devt) |
bd8eff3b | 649 | { |
a160c615 CH |
650 | unsigned int major = MAJOR(devt); |
651 | struct blk_major_name **n; | |
652 | ||
653 | mutex_lock(&major_names_lock); | |
654 | for (n = &major_names[major_to_index(major)]; *n; n = &(*n)->next) { | |
655 | if ((*n)->major == major && (*n)->probe) { | |
656 | (*n)->probe(devt); | |
657 | mutex_unlock(&major_names_lock); | |
658 | return; | |
659 | } | |
660 | } | |
661 | mutex_unlock(&major_names_lock); | |
662 | ||
bd8eff3b CH |
663 | if (request_module("block-major-%d-%d", MAJOR(devt), MINOR(devt)) > 0) |
664 | /* Make old-style 2.4 aliases work */ | |
665 | request_module("block-major-%d", MAJOR(devt)); | |
666 | } | |
667 | ||
5c6f35c5 GKH |
668 | /* |
669 | * print a full list of all partitions - intended for places where the root | |
670 | * filesystem can't be mounted and thus to give the victim some idea of what | |
671 | * went wrong | |
672 | */ | |
673 | void __init printk_all_partitions(void) | |
674 | { | |
def4e38d TH |
675 | struct class_dev_iter iter; |
676 | struct device *dev; | |
677 | ||
678 | class_dev_iter_init(&iter, &block_class, NULL, &disk_type); | |
679 | while ((dev = class_dev_iter_next(&iter))) { | |
680 | struct gendisk *disk = dev_to_disk(dev); | |
ad1eaa53 | 681 | struct block_device *part; |
1f014290 | 682 | char devt_buf[BDEVT_SIZE]; |
e559f58d | 683 | unsigned long idx; |
def4e38d TH |
684 | |
685 | /* | |
686 | * Don't show empty devices or things that have been | |
25985edc | 687 | * suppressed |
def4e38d TH |
688 | */ |
689 | if (get_capacity(disk) == 0 || | |
690 | (disk->flags & GENHD_FL_SUPPRESS_PARTITION_INFO)) | |
691 | continue; | |
692 | ||
693 | /* | |
e559f58d CH |
694 | * Note, unlike /proc/partitions, I am showing the numbers in |
695 | * hex - the same format as the root= option takes. | |
def4e38d | 696 | */ |
e559f58d CH |
697 | rcu_read_lock(); |
698 | xa_for_each(&disk->part_tbl, idx, part) { | |
699 | if (!bdev_nr_sectors(part)) | |
700 | continue; | |
a9e7bc3d | 701 | printk("%s%s %10llu %pg %s", |
e559f58d | 702 | bdev_is_partition(part) ? " " : "", |
ad1eaa53 | 703 | bdevt_str(part->bd_dev, devt_buf), |
a9e7bc3d | 704 | bdev_nr_sectors(part) >> 1, part, |
ad1eaa53 CH |
705 | part->bd_meta_info ? |
706 | part->bd_meta_info->uuid : ""); | |
e559f58d | 707 | if (bdev_is_partition(part)) |
074a7aca | 708 | printk("\n"); |
e559f58d CH |
709 | else if (dev->parent && dev->parent->driver) |
710 | printk(" driver: %s\n", | |
711 | dev->parent->driver->name); | |
712 | else | |
713 | printk(" (driver?)\n"); | |
074a7aca | 714 | } |
e559f58d | 715 | rcu_read_unlock(); |
def4e38d TH |
716 | } |
717 | class_dev_iter_exit(&iter); | |
dd2a345f DG |
718 | } |
719 | ||
1da177e4 LT |
720 | #ifdef CONFIG_PROC_FS |
721 | /* iterator */ | |
def4e38d | 722 | static void *disk_seqf_start(struct seq_file *seqf, loff_t *pos) |
68c4d4a7 | 723 | { |
def4e38d TH |
724 | loff_t skip = *pos; |
725 | struct class_dev_iter *iter; | |
726 | struct device *dev; | |
68c4d4a7 | 727 | |
aeb3d3a8 | 728 | iter = kmalloc(sizeof(*iter), GFP_KERNEL); |
def4e38d TH |
729 | if (!iter) |
730 | return ERR_PTR(-ENOMEM); | |
731 | ||
732 | seqf->private = iter; | |
733 | class_dev_iter_init(iter, &block_class, NULL, &disk_type); | |
734 | do { | |
735 | dev = class_dev_iter_next(iter); | |
736 | if (!dev) | |
737 | return NULL; | |
738 | } while (skip--); | |
739 | ||
740 | return dev_to_disk(dev); | |
68c4d4a7 GKH |
741 | } |
742 | ||
def4e38d | 743 | static void *disk_seqf_next(struct seq_file *seqf, void *v, loff_t *pos) |
1da177e4 | 744 | { |
edfaa7c3 | 745 | struct device *dev; |
1da177e4 | 746 | |
def4e38d TH |
747 | (*pos)++; |
748 | dev = class_dev_iter_next(seqf->private); | |
2ac3cee5 | 749 | if (dev) |
68c4d4a7 | 750 | return dev_to_disk(dev); |
2ac3cee5 | 751 | |
1da177e4 LT |
752 | return NULL; |
753 | } | |
754 | ||
def4e38d | 755 | static void disk_seqf_stop(struct seq_file *seqf, void *v) |
27f30251 | 756 | { |
def4e38d | 757 | struct class_dev_iter *iter = seqf->private; |
27f30251 | 758 | |
def4e38d TH |
759 | /* stop is called even after start failed :-( */ |
760 | if (iter) { | |
761 | class_dev_iter_exit(iter); | |
762 | kfree(iter); | |
77da1605 | 763 | seqf->private = NULL; |
5c0ef6d0 | 764 | } |
1da177e4 LT |
765 | } |
766 | ||
def4e38d | 767 | static void *show_partition_start(struct seq_file *seqf, loff_t *pos) |
1da177e4 | 768 | { |
06768067 | 769 | void *p; |
def4e38d TH |
770 | |
771 | p = disk_seqf_start(seqf, pos); | |
b9f985b6 | 772 | if (!IS_ERR_OR_NULL(p) && !*pos) |
def4e38d TH |
773 | seq_puts(seqf, "major minor #blocks name\n\n"); |
774 | return p; | |
1da177e4 LT |
775 | } |
776 | ||
cf771cb5 | 777 | static int show_partition(struct seq_file *seqf, void *v) |
1da177e4 LT |
778 | { |
779 | struct gendisk *sgp = v; | |
ad1eaa53 | 780 | struct block_device *part; |
ecc75a98 | 781 | unsigned long idx; |
1da177e4 | 782 | |
1da177e4 | 783 | /* Don't show non-partitionable removeable devices or empty devices */ |
d27769ec | 784 | if (!get_capacity(sgp) || (!disk_max_parts(sgp) && |
f331c029 | 785 | (sgp->flags & GENHD_FL_REMOVABLE))) |
1da177e4 LT |
786 | return 0; |
787 | if (sgp->flags & GENHD_FL_SUPPRESS_PARTITION_INFO) | |
788 | return 0; | |
789 | ||
ecc75a98 CH |
790 | rcu_read_lock(); |
791 | xa_for_each(&sgp->part_tbl, idx, part) { | |
792 | if (!bdev_nr_sectors(part)) | |
793 | continue; | |
a291bb43 | 794 | seq_printf(seqf, "%4d %7d %10llu %pg\n", |
ad1eaa53 | 795 | MAJOR(part->bd_dev), MINOR(part->bd_dev), |
a291bb43 | 796 | bdev_nr_sectors(part) >> 1, part); |
ecc75a98 CH |
797 | } |
798 | rcu_read_unlock(); | |
1da177e4 LT |
799 | return 0; |
800 | } | |
801 | ||
f500975a | 802 | static const struct seq_operations partitions_op = { |
def4e38d TH |
803 | .start = show_partition_start, |
804 | .next = disk_seqf_next, | |
805 | .stop = disk_seqf_stop, | |
edfaa7c3 | 806 | .show = show_partition |
1da177e4 LT |
807 | }; |
808 | #endif | |
809 | ||
1da177e4 LT |
810 | static int __init genhd_device_init(void) |
811 | { | |
e105b8bf DW |
812 | int error; |
813 | ||
814 | block_class.dev_kobj = sysfs_dev_block_kobj; | |
815 | error = class_register(&block_class); | |
ee27a558 RM |
816 | if (unlikely(error)) |
817 | return error; | |
1da177e4 | 818 | blk_dev_init(); |
edfaa7c3 | 819 | |
561ec68e ZY |
820 | register_blkdev(BLOCK_EXT_MAJOR, "blkext"); |
821 | ||
edfaa7c3 | 822 | /* create top-level block dir */ |
e52eec13 AK |
823 | if (!sysfs_deprecated) |
824 | block_depr = kobject_create_and_add("block", NULL); | |
830d3cfb | 825 | return 0; |
1da177e4 LT |
826 | } |
827 | ||
828 | subsys_initcall(genhd_device_init); | |
829 | ||
edfaa7c3 KS |
830 | static ssize_t disk_range_show(struct device *dev, |
831 | struct device_attribute *attr, char *buf) | |
1da177e4 | 832 | { |
edfaa7c3 | 833 | struct gendisk *disk = dev_to_disk(dev); |
1da177e4 | 834 | |
edfaa7c3 | 835 | return sprintf(buf, "%d\n", disk->minors); |
1da177e4 LT |
836 | } |
837 | ||
1f014290 TH |
838 | static ssize_t disk_ext_range_show(struct device *dev, |
839 | struct device_attribute *attr, char *buf) | |
840 | { | |
841 | struct gendisk *disk = dev_to_disk(dev); | |
842 | ||
b5d0b9df | 843 | return sprintf(buf, "%d\n", disk_max_parts(disk)); |
1f014290 TH |
844 | } |
845 | ||
edfaa7c3 KS |
846 | static ssize_t disk_removable_show(struct device *dev, |
847 | struct device_attribute *attr, char *buf) | |
a7fd6706 | 848 | { |
edfaa7c3 | 849 | struct gendisk *disk = dev_to_disk(dev); |
a7fd6706 | 850 | |
edfaa7c3 KS |
851 | return sprintf(buf, "%d\n", |
852 | (disk->flags & GENHD_FL_REMOVABLE ? 1 : 0)); | |
a7fd6706 KS |
853 | } |
854 | ||
8ddcd653 CH |
855 | static ssize_t disk_hidden_show(struct device *dev, |
856 | struct device_attribute *attr, char *buf) | |
857 | { | |
858 | struct gendisk *disk = dev_to_disk(dev); | |
859 | ||
860 | return sprintf(buf, "%d\n", | |
861 | (disk->flags & GENHD_FL_HIDDEN ? 1 : 0)); | |
862 | } | |
863 | ||
1c9ce527 KS |
864 | static ssize_t disk_ro_show(struct device *dev, |
865 | struct device_attribute *attr, char *buf) | |
866 | { | |
867 | struct gendisk *disk = dev_to_disk(dev); | |
868 | ||
b7db9956 | 869 | return sprintf(buf, "%d\n", get_disk_ro(disk) ? 1 : 0); |
1c9ce527 KS |
870 | } |
871 | ||
3ad5cee5 CH |
872 | ssize_t part_size_show(struct device *dev, |
873 | struct device_attribute *attr, char *buf) | |
874 | { | |
0d02129e | 875 | return sprintf(buf, "%llu\n", bdev_nr_sectors(dev_to_bdev(dev))); |
3ad5cee5 CH |
876 | } |
877 | ||
878 | ssize_t part_stat_show(struct device *dev, | |
879 | struct device_attribute *attr, char *buf) | |
880 | { | |
0d02129e CH |
881 | struct block_device *bdev = dev_to_bdev(dev); |
882 | struct request_queue *q = bdev->bd_disk->queue; | |
ea18e0f0 | 883 | struct disk_stats stat; |
3ad5cee5 CH |
884 | unsigned int inflight; |
885 | ||
0d02129e | 886 | part_stat_read_all(bdev, &stat); |
b2f609e1 | 887 | if (queue_is_mq(q)) |
0d02129e | 888 | inflight = blk_mq_in_flight(q, bdev); |
b2f609e1 | 889 | else |
0d02129e | 890 | inflight = part_in_flight(bdev); |
ea18e0f0 | 891 | |
3ad5cee5 CH |
892 | return sprintf(buf, |
893 | "%8lu %8lu %8llu %8u " | |
894 | "%8lu %8lu %8llu %8u " | |
895 | "%8u %8u %8u " | |
896 | "%8lu %8lu %8llu %8u " | |
897 | "%8lu %8u" | |
898 | "\n", | |
ea18e0f0 KK |
899 | stat.ios[STAT_READ], |
900 | stat.merges[STAT_READ], | |
901 | (unsigned long long)stat.sectors[STAT_READ], | |
902 | (unsigned int)div_u64(stat.nsecs[STAT_READ], NSEC_PER_MSEC), | |
903 | stat.ios[STAT_WRITE], | |
904 | stat.merges[STAT_WRITE], | |
905 | (unsigned long long)stat.sectors[STAT_WRITE], | |
906 | (unsigned int)div_u64(stat.nsecs[STAT_WRITE], NSEC_PER_MSEC), | |
3ad5cee5 | 907 | inflight, |
ea18e0f0 | 908 | jiffies_to_msecs(stat.io_ticks), |
8cd5b8fc KK |
909 | (unsigned int)div_u64(stat.nsecs[STAT_READ] + |
910 | stat.nsecs[STAT_WRITE] + | |
911 | stat.nsecs[STAT_DISCARD] + | |
912 | stat.nsecs[STAT_FLUSH], | |
913 | NSEC_PER_MSEC), | |
ea18e0f0 KK |
914 | stat.ios[STAT_DISCARD], |
915 | stat.merges[STAT_DISCARD], | |
916 | (unsigned long long)stat.sectors[STAT_DISCARD], | |
917 | (unsigned int)div_u64(stat.nsecs[STAT_DISCARD], NSEC_PER_MSEC), | |
918 | stat.ios[STAT_FLUSH], | |
919 | (unsigned int)div_u64(stat.nsecs[STAT_FLUSH], NSEC_PER_MSEC)); | |
3ad5cee5 CH |
920 | } |
921 | ||
922 | ssize_t part_inflight_show(struct device *dev, struct device_attribute *attr, | |
923 | char *buf) | |
924 | { | |
0d02129e CH |
925 | struct block_device *bdev = dev_to_bdev(dev); |
926 | struct request_queue *q = bdev->bd_disk->queue; | |
3ad5cee5 CH |
927 | unsigned int inflight[2]; |
928 | ||
b2f609e1 | 929 | if (queue_is_mq(q)) |
0d02129e | 930 | blk_mq_in_flight_rw(q, bdev, inflight); |
b2f609e1 | 931 | else |
0d02129e | 932 | part_in_flight_rw(bdev, inflight); |
b2f609e1 | 933 | |
3ad5cee5 CH |
934 | return sprintf(buf, "%8u %8u\n", inflight[0], inflight[1]); |
935 | } | |
936 | ||
edfaa7c3 KS |
937 | static ssize_t disk_capability_show(struct device *dev, |
938 | struct device_attribute *attr, char *buf) | |
86ce18d7 | 939 | { |
edfaa7c3 KS |
940 | struct gendisk *disk = dev_to_disk(dev); |
941 | ||
942 | return sprintf(buf, "%x\n", disk->flags); | |
86ce18d7 | 943 | } |
edfaa7c3 | 944 | |
c72758f3 MP |
945 | static ssize_t disk_alignment_offset_show(struct device *dev, |
946 | struct device_attribute *attr, | |
947 | char *buf) | |
948 | { | |
949 | struct gendisk *disk = dev_to_disk(dev); | |
950 | ||
951 | return sprintf(buf, "%d\n", queue_alignment_offset(disk->queue)); | |
952 | } | |
953 | ||
86b37281 MP |
954 | static ssize_t disk_discard_alignment_show(struct device *dev, |
955 | struct device_attribute *attr, | |
956 | char *buf) | |
957 | { | |
958 | struct gendisk *disk = dev_to_disk(dev); | |
959 | ||
dd3d145d | 960 | return sprintf(buf, "%d\n", queue_discard_alignment(disk->queue)); |
86b37281 MP |
961 | } |
962 | ||
5657a819 JP |
963 | static DEVICE_ATTR(range, 0444, disk_range_show, NULL); |
964 | static DEVICE_ATTR(ext_range, 0444, disk_ext_range_show, NULL); | |
965 | static DEVICE_ATTR(removable, 0444, disk_removable_show, NULL); | |
966 | static DEVICE_ATTR(hidden, 0444, disk_hidden_show, NULL); | |
967 | static DEVICE_ATTR(ro, 0444, disk_ro_show, NULL); | |
968 | static DEVICE_ATTR(size, 0444, part_size_show, NULL); | |
969 | static DEVICE_ATTR(alignment_offset, 0444, disk_alignment_offset_show, NULL); | |
970 | static DEVICE_ATTR(discard_alignment, 0444, disk_discard_alignment_show, NULL); | |
971 | static DEVICE_ATTR(capability, 0444, disk_capability_show, NULL); | |
972 | static DEVICE_ATTR(stat, 0444, part_stat_show, NULL); | |
973 | static DEVICE_ATTR(inflight, 0444, part_inflight_show, NULL); | |
974 | static DEVICE_ATTR(badblocks, 0644, disk_badblocks_show, disk_badblocks_store); | |
3ad5cee5 | 975 | |
c17bb495 | 976 | #ifdef CONFIG_FAIL_MAKE_REQUEST |
3ad5cee5 CH |
977 | ssize_t part_fail_show(struct device *dev, |
978 | struct device_attribute *attr, char *buf) | |
979 | { | |
0d02129e | 980 | return sprintf(buf, "%d\n", dev_to_bdev(dev)->bd_make_it_fail); |
3ad5cee5 CH |
981 | } |
982 | ||
983 | ssize_t part_fail_store(struct device *dev, | |
984 | struct device_attribute *attr, | |
985 | const char *buf, size_t count) | |
986 | { | |
3ad5cee5 CH |
987 | int i; |
988 | ||
989 | if (count > 0 && sscanf(buf, "%d", &i) > 0) | |
0d02129e | 990 | dev_to_bdev(dev)->bd_make_it_fail = i; |
3ad5cee5 CH |
991 | |
992 | return count; | |
993 | } | |
994 | ||
edfaa7c3 | 995 | static struct device_attribute dev_attr_fail = |
5657a819 | 996 | __ATTR(make-it-fail, 0644, part_fail_show, part_fail_store); |
3ad5cee5 CH |
997 | #endif /* CONFIG_FAIL_MAKE_REQUEST */ |
998 | ||
581d4e28 JA |
999 | #ifdef CONFIG_FAIL_IO_TIMEOUT |
1000 | static struct device_attribute dev_attr_fail_timeout = | |
5657a819 | 1001 | __ATTR(io-timeout-fail, 0644, part_timeout_show, part_timeout_store); |
581d4e28 | 1002 | #endif |
edfaa7c3 KS |
1003 | |
1004 | static struct attribute *disk_attrs[] = { | |
1005 | &dev_attr_range.attr, | |
1f014290 | 1006 | &dev_attr_ext_range.attr, |
edfaa7c3 | 1007 | &dev_attr_removable.attr, |
8ddcd653 | 1008 | &dev_attr_hidden.attr, |
1c9ce527 | 1009 | &dev_attr_ro.attr, |
edfaa7c3 | 1010 | &dev_attr_size.attr, |
c72758f3 | 1011 | &dev_attr_alignment_offset.attr, |
86b37281 | 1012 | &dev_attr_discard_alignment.attr, |
edfaa7c3 KS |
1013 | &dev_attr_capability.attr, |
1014 | &dev_attr_stat.attr, | |
316d315b | 1015 | &dev_attr_inflight.attr, |
99e6608c | 1016 | &dev_attr_badblocks.attr, |
2bc8cda5 CH |
1017 | &dev_attr_events.attr, |
1018 | &dev_attr_events_async.attr, | |
1019 | &dev_attr_events_poll_msecs.attr, | |
edfaa7c3 KS |
1020 | #ifdef CONFIG_FAIL_MAKE_REQUEST |
1021 | &dev_attr_fail.attr, | |
581d4e28 JA |
1022 | #endif |
1023 | #ifdef CONFIG_FAIL_IO_TIMEOUT | |
1024 | &dev_attr_fail_timeout.attr, | |
edfaa7c3 KS |
1025 | #endif |
1026 | NULL | |
1027 | }; | |
1028 | ||
9438b3e0 DW |
1029 | static umode_t disk_visible(struct kobject *kobj, struct attribute *a, int n) |
1030 | { | |
1031 | struct device *dev = container_of(kobj, typeof(*dev), kobj); | |
1032 | struct gendisk *disk = dev_to_disk(dev); | |
1033 | ||
1034 | if (a == &dev_attr_badblocks.attr && !disk->bb) | |
1035 | return 0; | |
1036 | return a->mode; | |
1037 | } | |
1038 | ||
edfaa7c3 KS |
1039 | static struct attribute_group disk_attr_group = { |
1040 | .attrs = disk_attrs, | |
9438b3e0 | 1041 | .is_visible = disk_visible, |
edfaa7c3 KS |
1042 | }; |
1043 | ||
a4dbd674 | 1044 | static const struct attribute_group *disk_attr_groups[] = { |
edfaa7c3 KS |
1045 | &disk_attr_group, |
1046 | NULL | |
1da177e4 LT |
1047 | }; |
1048 | ||
b5bd357c LC |
1049 | /** |
1050 | * disk_release - releases all allocated resources of the gendisk | |
1051 | * @dev: the device representing this disk | |
1052 | * | |
1053 | * This function releases all allocated resources of the gendisk. | |
1054 | * | |
b5bd357c LC |
1055 | * Drivers which used __device_add_disk() have a gendisk with a request_queue |
1056 | * assigned. Since the request_queue sits on top of the gendisk for these | |
1057 | * drivers we also call blk_put_queue() for them, and we expect the | |
1058 | * request_queue refcount to reach 0 at this point, and so the request_queue | |
1059 | * will also be freed prior to the disk. | |
e8c7d14a LC |
1060 | * |
1061 | * Context: can sleep | |
b5bd357c | 1062 | */ |
edfaa7c3 | 1063 | static void disk_release(struct device *dev) |
1da177e4 | 1064 | { |
edfaa7c3 KS |
1065 | struct gendisk *disk = dev_to_disk(dev); |
1066 | ||
e8c7d14a LC |
1067 | might_sleep(); |
1068 | ||
7c3f828b CH |
1069 | if (MAJOR(dev->devt) == BLOCK_EXT_MAJOR) |
1070 | blk_free_ext_minor(MINOR(dev->devt)); | |
77ea887e | 1071 | disk_release_events(disk); |
1da177e4 | 1072 | kfree(disk->random); |
a33df75c | 1073 | xa_destroy(&disk->part_tbl); |
958229a7 | 1074 | if (test_bit(GD_QUEUE_REF, &disk->state) && disk->queue) |
523e1d39 | 1075 | blk_put_queue(disk->queue); |
2f4731dc | 1076 | iput(disk->part0->bd_inode); /* frees the disk */ |
1da177e4 | 1077 | } |
edfaa7c3 KS |
1078 | struct class block_class = { |
1079 | .name = "block", | |
1da177e4 LT |
1080 | }; |
1081 | ||
3c2670e6 | 1082 | static char *block_devnode(struct device *dev, umode_t *mode, |
4e4098a3 | 1083 | kuid_t *uid, kgid_t *gid) |
b03f38b6 KS |
1084 | { |
1085 | struct gendisk *disk = dev_to_disk(dev); | |
1086 | ||
348e114b CH |
1087 | if (disk->fops->devnode) |
1088 | return disk->fops->devnode(disk, mode); | |
b03f38b6 KS |
1089 | return NULL; |
1090 | } | |
1091 | ||
ef45fe47 | 1092 | const struct device_type disk_type = { |
edfaa7c3 KS |
1093 | .name = "disk", |
1094 | .groups = disk_attr_groups, | |
1095 | .release = disk_release, | |
e454cea2 | 1096 | .devnode = block_devnode, |
1da177e4 LT |
1097 | }; |
1098 | ||
a6e2ba88 | 1099 | #ifdef CONFIG_PROC_FS |
cf771cb5 TH |
1100 | /* |
1101 | * aggregate disk stat collector. Uses the same stats that the sysfs | |
1102 | * entries do, above, but makes them available through one seq_file. | |
1103 | * | |
1104 | * The output looks suspiciously like /proc/partitions with a bunch of | |
1105 | * extra fields. | |
1106 | */ | |
1107 | static int diskstats_show(struct seq_file *seqf, void *v) | |
1da177e4 LT |
1108 | { |
1109 | struct gendisk *gp = v; | |
ad1eaa53 | 1110 | struct block_device *hd; |
e016b782 | 1111 | unsigned int inflight; |
ea18e0f0 | 1112 | struct disk_stats stat; |
7fae67cc | 1113 | unsigned long idx; |
1da177e4 LT |
1114 | |
1115 | /* | |
ed9e1982 | 1116 | if (&disk_to_dev(gp)->kobj.entry == block_class.devices.next) |
cf771cb5 | 1117 | seq_puts(seqf, "major minor name" |
1da177e4 LT |
1118 | " rio rmerge rsect ruse wio wmerge " |
1119 | "wsect wuse running use aveq" | |
1120 | "\n\n"); | |
1121 | */ | |
9f5e4865 | 1122 | |
7fae67cc CH |
1123 | rcu_read_lock(); |
1124 | xa_for_each(&gp->part_tbl, idx, hd) { | |
1125 | if (bdev_is_partition(hd) && !bdev_nr_sectors(hd)) | |
1126 | continue; | |
0d02129e | 1127 | part_stat_read_all(hd, &stat); |
b2f609e1 | 1128 | if (queue_is_mq(gp->queue)) |
ad1eaa53 | 1129 | inflight = blk_mq_in_flight(gp->queue, hd); |
b2f609e1 | 1130 | else |
ad1eaa53 | 1131 | inflight = part_in_flight(hd); |
ea18e0f0 | 1132 | |
26e2d7a3 | 1133 | seq_printf(seqf, "%4d %7d %pg " |
bdca3c87 MC |
1134 | "%lu %lu %lu %u " |
1135 | "%lu %lu %lu %u " | |
1136 | "%u %u %u " | |
b6866318 KK |
1137 | "%lu %lu %lu %u " |
1138 | "%lu %u" | |
1139 | "\n", | |
26e2d7a3 | 1140 | MAJOR(hd->bd_dev), MINOR(hd->bd_dev), hd, |
ea18e0f0 KK |
1141 | stat.ios[STAT_READ], |
1142 | stat.merges[STAT_READ], | |
1143 | stat.sectors[STAT_READ], | |
1144 | (unsigned int)div_u64(stat.nsecs[STAT_READ], | |
1145 | NSEC_PER_MSEC), | |
1146 | stat.ios[STAT_WRITE], | |
1147 | stat.merges[STAT_WRITE], | |
1148 | stat.sectors[STAT_WRITE], | |
1149 | (unsigned int)div_u64(stat.nsecs[STAT_WRITE], | |
1150 | NSEC_PER_MSEC), | |
e016b782 | 1151 | inflight, |
ea18e0f0 | 1152 | jiffies_to_msecs(stat.io_ticks), |
8cd5b8fc KK |
1153 | (unsigned int)div_u64(stat.nsecs[STAT_READ] + |
1154 | stat.nsecs[STAT_WRITE] + | |
1155 | stat.nsecs[STAT_DISCARD] + | |
1156 | stat.nsecs[STAT_FLUSH], | |
1157 | NSEC_PER_MSEC), | |
ea18e0f0 KK |
1158 | stat.ios[STAT_DISCARD], |
1159 | stat.merges[STAT_DISCARD], | |
1160 | stat.sectors[STAT_DISCARD], | |
1161 | (unsigned int)div_u64(stat.nsecs[STAT_DISCARD], | |
1162 | NSEC_PER_MSEC), | |
1163 | stat.ios[STAT_FLUSH], | |
1164 | (unsigned int)div_u64(stat.nsecs[STAT_FLUSH], | |
1165 | NSEC_PER_MSEC) | |
28f39d55 | 1166 | ); |
1da177e4 | 1167 | } |
7fae67cc | 1168 | rcu_read_unlock(); |
9f5e4865 | 1169 | |
1da177e4 LT |
1170 | return 0; |
1171 | } | |
1172 | ||
31d85ab2 | 1173 | static const struct seq_operations diskstats_op = { |
def4e38d TH |
1174 | .start = disk_seqf_start, |
1175 | .next = disk_seqf_next, | |
1176 | .stop = disk_seqf_stop, | |
1da177e4 LT |
1177 | .show = diskstats_show |
1178 | }; | |
f500975a AD |
1179 | |
1180 | static int __init proc_genhd_init(void) | |
1181 | { | |
fddda2b7 CH |
1182 | proc_create_seq("diskstats", 0, NULL, &diskstats_op); |
1183 | proc_create_seq("partitions", 0, NULL, &partitions_op); | |
f500975a AD |
1184 | return 0; |
1185 | } | |
1186 | module_init(proc_genhd_init); | |
a6e2ba88 | 1187 | #endif /* CONFIG_PROC_FS */ |
1da177e4 | 1188 | |
c97d93c3 CH |
1189 | dev_t part_devt(struct gendisk *disk, u8 partno) |
1190 | { | |
0e0ccdec | 1191 | struct block_device *part; |
c97d93c3 CH |
1192 | dev_t devt = 0; |
1193 | ||
0e0ccdec CH |
1194 | rcu_read_lock(); |
1195 | part = xa_load(&disk->part_tbl, partno); | |
1196 | if (part) | |
c97d93c3 | 1197 | devt = part->bd_dev; |
0e0ccdec | 1198 | rcu_read_unlock(); |
c97d93c3 CH |
1199 | |
1200 | return devt; | |
1201 | } | |
1202 | ||
cf771cb5 | 1203 | dev_t blk_lookup_devt(const char *name, int partno) |
a142be85 | 1204 | { |
def4e38d TH |
1205 | dev_t devt = MKDEV(0, 0); |
1206 | struct class_dev_iter iter; | |
1207 | struct device *dev; | |
a142be85 | 1208 | |
def4e38d TH |
1209 | class_dev_iter_init(&iter, &block_class, NULL, &disk_type); |
1210 | while ((dev = class_dev_iter_next(&iter))) { | |
a142be85 | 1211 | struct gendisk *disk = dev_to_disk(dev); |
a142be85 | 1212 | |
3ada8b7e | 1213 | if (strcmp(dev_name(dev), name)) |
f331c029 | 1214 | continue; |
f331c029 | 1215 | |
41b8c853 NB |
1216 | if (partno < disk->minors) { |
1217 | /* We need to return the right devno, even | |
1218 | * if the partition doesn't exist yet. | |
1219 | */ | |
1220 | devt = MKDEV(MAJOR(dev->devt), | |
1221 | MINOR(dev->devt) + partno); | |
c97d93c3 CH |
1222 | } else { |
1223 | devt = part_devt(disk, partno); | |
1224 | if (devt) | |
1225 | break; | |
def4e38d | 1226 | } |
5c0ef6d0 | 1227 | } |
def4e38d | 1228 | class_dev_iter_exit(&iter); |
edfaa7c3 KS |
1229 | return devt; |
1230 | } | |
edfaa7c3 | 1231 | |
e319e1fb | 1232 | struct gendisk *__alloc_disk_node(int minors, int node_id) |
1946089a CL |
1233 | { |
1234 | struct gendisk *disk; | |
1235 | ||
c1b511eb | 1236 | disk = kzalloc_node(sizeof(struct gendisk), GFP_KERNEL, node_id); |
f93af2a4 CH |
1237 | if (!disk) |
1238 | return NULL; | |
6c23a968 | 1239 | |
cb8432d6 CH |
1240 | disk->part0 = bdev_alloc(disk, 0); |
1241 | if (!disk->part0) | |
22ae8ce8 CH |
1242 | goto out_free_disk; |
1243 | ||
f93af2a4 | 1244 | disk->node_id = node_id; |
a8698707 | 1245 | mutex_init(&disk->open_mutex); |
a33df75c CH |
1246 | xa_init(&disk->part_tbl); |
1247 | if (xa_insert(&disk->part_tbl, 0, disk->part0, GFP_KERNEL)) | |
1248 | goto out_destroy_part_tbl; | |
f93af2a4 CH |
1249 | |
1250 | disk->minors = minors; | |
1251 | rand_initialize_disk(disk); | |
1252 | disk_to_dev(disk)->class = &block_class; | |
1253 | disk_to_dev(disk)->type = &disk_type; | |
1254 | device_initialize(disk_to_dev(disk)); | |
1da177e4 | 1255 | return disk; |
f93af2a4 | 1256 | |
a33df75c CH |
1257 | out_destroy_part_tbl: |
1258 | xa_destroy(&disk->part_tbl); | |
2f4731dc | 1259 | iput(disk->part0->bd_inode); |
f93af2a4 CH |
1260 | out_free_disk: |
1261 | kfree(disk); | |
1262 | return NULL; | |
1da177e4 | 1263 | } |
e319e1fb | 1264 | EXPORT_SYMBOL(__alloc_disk_node); |
1da177e4 | 1265 | |
f525464a CH |
1266 | struct gendisk *__blk_alloc_disk(int node) |
1267 | { | |
1268 | struct request_queue *q; | |
1269 | struct gendisk *disk; | |
1270 | ||
1271 | q = blk_alloc_queue(node); | |
1272 | if (!q) | |
1273 | return NULL; | |
1274 | ||
1275 | disk = __alloc_disk_node(0, node); | |
1276 | if (!disk) { | |
1277 | blk_cleanup_queue(q); | |
1278 | return NULL; | |
1279 | } | |
1280 | disk->queue = q; | |
1281 | return disk; | |
1282 | } | |
1283 | EXPORT_SYMBOL(__blk_alloc_disk); | |
1284 | ||
b5bd357c LC |
1285 | /** |
1286 | * put_disk - decrements the gendisk refcount | |
0d20dcc2 | 1287 | * @disk: the struct gendisk to decrement the refcount for |
b5bd357c LC |
1288 | * |
1289 | * This decrements the refcount for the struct gendisk. When this reaches 0 | |
1290 | * we'll have disk_release() called. | |
e8c7d14a LC |
1291 | * |
1292 | * Context: Any context, but the last reference must not be dropped from | |
1293 | * atomic context. | |
b5bd357c | 1294 | */ |
1da177e4 LT |
1295 | void put_disk(struct gendisk *disk) |
1296 | { | |
1297 | if (disk) | |
efdc41c8 | 1298 | put_device(disk_to_dev(disk)); |
1da177e4 | 1299 | } |
1da177e4 LT |
1300 | EXPORT_SYMBOL(put_disk); |
1301 | ||
f525464a CH |
1302 | /** |
1303 | * blk_cleanup_disk - shutdown a gendisk allocated by blk_alloc_disk | |
1304 | * @disk: gendisk to shutdown | |
1305 | * | |
1306 | * Mark the queue hanging off @disk DYING, drain all pending requests, then mark | |
1307 | * the queue DEAD, destroy and put it and the gendisk structure. | |
1308 | * | |
1309 | * Context: can sleep | |
1310 | */ | |
1311 | void blk_cleanup_disk(struct gendisk *disk) | |
1312 | { | |
1313 | blk_cleanup_queue(disk->queue); | |
1314 | put_disk(disk); | |
1315 | } | |
1316 | EXPORT_SYMBOL(blk_cleanup_disk); | |
1317 | ||
e3264a4d HR |
1318 | static void set_disk_ro_uevent(struct gendisk *gd, int ro) |
1319 | { | |
1320 | char event[] = "DISK_RO=1"; | |
1321 | char *envp[] = { event, NULL }; | |
1322 | ||
1323 | if (!ro) | |
1324 | event[8] = '0'; | |
1325 | kobject_uevent_env(&disk_to_dev(gd)->kobj, KOBJ_CHANGE, envp); | |
1326 | } | |
1327 | ||
52f019d4 CH |
1328 | /** |
1329 | * set_disk_ro - set a gendisk read-only | |
1330 | * @disk: gendisk to operate on | |
7f31bee3 | 1331 | * @read_only: %true to set the disk read-only, %false set the disk read/write |
52f019d4 CH |
1332 | * |
1333 | * This function is used to indicate whether a given disk device should have its | |
1334 | * read-only flag set. set_disk_ro() is typically used by device drivers to | |
1335 | * indicate whether the underlying physical device is write-protected. | |
1336 | */ | |
1337 | void set_disk_ro(struct gendisk *disk, bool read_only) | |
1da177e4 | 1338 | { |
52f019d4 CH |
1339 | if (read_only) { |
1340 | if (test_and_set_bit(GD_READ_ONLY, &disk->state)) | |
1341 | return; | |
1342 | } else { | |
1343 | if (!test_and_clear_bit(GD_READ_ONLY, &disk->state)) | |
1344 | return; | |
e3264a4d | 1345 | } |
52f019d4 | 1346 | set_disk_ro_uevent(disk, read_only); |
1da177e4 | 1347 | } |
1da177e4 LT |
1348 | EXPORT_SYMBOL(set_disk_ro); |
1349 | ||
1350 | int bdev_read_only(struct block_device *bdev) | |
1351 | { | |
947139bf | 1352 | return bdev->bd_read_only || get_disk_ro(bdev->bd_disk); |
1da177e4 | 1353 | } |
1da177e4 | 1354 | EXPORT_SYMBOL(bdev_read_only); |