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 | |
cf179948 MC |
32 | /* |
33 | * Unique, monotonically increasing sequential number associated with block | |
34 | * devices instances (i.e. incremented each time a device is attached). | |
35 | * Associating uevents with block devices in userspace is difficult and racy: | |
36 | * the uevent netlink socket is lossy, and on slow and overloaded systems has | |
37 | * a very high latency. | |
38 | * Block devices do not have exclusive owners in userspace, any process can set | |
39 | * one up (e.g. loop devices). Moreover, device names can be reused (e.g. loop0 | |
40 | * can be reused again and again). | |
41 | * A userspace process setting up a block device and watching for its events | |
42 | * cannot thus reliably tell whether an event relates to the device it just set | |
43 | * up or another earlier instance with the same name. | |
44 | * This sequential number allows userspace processes to solve this problem, and | |
45 | * uniquely associate an uevent to the lifetime to a device. | |
46 | */ | |
47 | static atomic64_t diskseq; | |
48 | ||
bcce3de1 | 49 | /* for extended dynamic devt allocation, currently only one major is used */ |
ce23bba8 | 50 | #define NR_EXT_DEVT (1 << MINORBITS) |
22ae8ce8 | 51 | static DEFINE_IDA(ext_devt_ida); |
bcce3de1 | 52 | |
a782483c CH |
53 | void set_capacity(struct gendisk *disk, sector_t sectors) |
54 | { | |
cb8432d6 | 55 | struct block_device *bdev = disk->part0; |
a782483c | 56 | |
0f472277 | 57 | spin_lock(&bdev->bd_size_lock); |
a782483c | 58 | i_size_write(bdev->bd_inode, (loff_t)sectors << SECTOR_SHIFT); |
0f472277 | 59 | spin_unlock(&bdev->bd_size_lock); |
a782483c CH |
60 | } |
61 | EXPORT_SYMBOL(set_capacity); | |
62 | ||
e598a72f | 63 | /* |
449f4ec9 CH |
64 | * Set disk capacity and notify if the size is not currently zero and will not |
65 | * be set to zero. Returns true if a uevent was sent, otherwise false. | |
e598a72f | 66 | */ |
449f4ec9 | 67 | bool set_capacity_and_notify(struct gendisk *disk, sector_t size) |
e598a72f BS |
68 | { |
69 | sector_t capacity = get_capacity(disk); | |
a782483c | 70 | char *envp[] = { "RESIZE=1", NULL }; |
e598a72f BS |
71 | |
72 | set_capacity(disk, size); | |
e598a72f | 73 | |
a782483c CH |
74 | /* |
75 | * Only print a message and send a uevent if the gendisk is user visible | |
76 | * and alive. This avoids spamming the log and udev when setting the | |
77 | * initial capacity during probing. | |
78 | */ | |
79 | if (size == capacity || | |
50b4aecf CH |
80 | !disk_live(disk) || |
81 | (disk->flags & GENHD_FL_HIDDEN)) | |
a782483c | 82 | return false; |
e598a72f | 83 | |
a782483c | 84 | pr_info("%s: detected capacity change from %lld to %lld\n", |
452c0bf8 | 85 | disk->disk_name, capacity, size); |
7e890c37 | 86 | |
a782483c CH |
87 | /* |
88 | * Historically we did not send a uevent for changes to/from an empty | |
89 | * device. | |
90 | */ | |
91 | if (!capacity || !size) | |
92 | return false; | |
93 | kobject_uevent_env(&disk_to_dev(disk)->kobj, KOBJ_CHANGE, envp); | |
94 | return true; | |
e598a72f | 95 | } |
449f4ec9 | 96 | EXPORT_SYMBOL_GPL(set_capacity_and_notify); |
e598a72f | 97 | |
5cbd28e3 | 98 | /* |
abd2864a CH |
99 | * Format the device name of the indicated block device into the supplied buffer |
100 | * and return a pointer to that same buffer for convenience. | |
101 | * | |
102 | * Note: do not use this in new code, use the %pg specifier to sprintf and | |
103 | * printk insted. | |
5cbd28e3 | 104 | */ |
abd2864a | 105 | const char *bdevname(struct block_device *bdev, char *buf) |
5cbd28e3 | 106 | { |
abd2864a CH |
107 | struct gendisk *hd = bdev->bd_disk; |
108 | int partno = bdev->bd_partno; | |
109 | ||
5cbd28e3 CH |
110 | if (!partno) |
111 | snprintf(buf, BDEVNAME_SIZE, "%s", hd->disk_name); | |
112 | else if (isdigit(hd->disk_name[strlen(hd->disk_name)-1])) | |
113 | snprintf(buf, BDEVNAME_SIZE, "%sp%d", hd->disk_name, partno); | |
114 | else | |
115 | snprintf(buf, BDEVNAME_SIZE, "%s%d", hd->disk_name, partno); | |
116 | ||
117 | return buf; | |
118 | } | |
5cbd28e3 | 119 | EXPORT_SYMBOL(bdevname); |
e598a72f | 120 | |
0d02129e CH |
121 | static void part_stat_read_all(struct block_device *part, |
122 | struct disk_stats *stat) | |
ea18e0f0 KK |
123 | { |
124 | int cpu; | |
125 | ||
126 | memset(stat, 0, sizeof(struct disk_stats)); | |
127 | for_each_possible_cpu(cpu) { | |
0d02129e | 128 | struct disk_stats *ptr = per_cpu_ptr(part->bd_stats, cpu); |
ea18e0f0 KK |
129 | int group; |
130 | ||
131 | for (group = 0; group < NR_STAT_GROUPS; group++) { | |
132 | stat->nsecs[group] += ptr->nsecs[group]; | |
133 | stat->sectors[group] += ptr->sectors[group]; | |
134 | stat->ios[group] += ptr->ios[group]; | |
135 | stat->merges[group] += ptr->merges[group]; | |
136 | } | |
137 | ||
138 | stat->io_ticks += ptr->io_ticks; | |
ea18e0f0 KK |
139 | } |
140 | } | |
ea18e0f0 | 141 | |
8446fe92 | 142 | static unsigned int part_in_flight(struct block_device *part) |
f299b7c7 | 143 | { |
b2f609e1 | 144 | unsigned int inflight = 0; |
1226b8dd | 145 | int cpu; |
f299b7c7 | 146 | |
1226b8dd | 147 | for_each_possible_cpu(cpu) { |
e016b782 MP |
148 | inflight += part_stat_local_read_cpu(part, in_flight[0], cpu) + |
149 | part_stat_local_read_cpu(part, in_flight[1], cpu); | |
1226b8dd | 150 | } |
e016b782 MP |
151 | if ((int)inflight < 0) |
152 | inflight = 0; | |
1226b8dd | 153 | |
e016b782 | 154 | return inflight; |
f299b7c7 JA |
155 | } |
156 | ||
8446fe92 CH |
157 | static void part_in_flight_rw(struct block_device *part, |
158 | unsigned int inflight[2]) | |
bf0ddaba | 159 | { |
1226b8dd MP |
160 | int cpu; |
161 | ||
1226b8dd MP |
162 | inflight[0] = 0; |
163 | inflight[1] = 0; | |
164 | for_each_possible_cpu(cpu) { | |
165 | inflight[0] += part_stat_local_read_cpu(part, in_flight[0], cpu); | |
166 | inflight[1] += part_stat_local_read_cpu(part, in_flight[1], cpu); | |
167 | } | |
168 | if ((int)inflight[0] < 0) | |
169 | inflight[0] = 0; | |
170 | if ((int)inflight[1] < 0) | |
171 | inflight[1] = 0; | |
bf0ddaba OS |
172 | } |
173 | ||
1da177e4 LT |
174 | /* |
175 | * Can be deleted altogether. Later. | |
176 | * | |
177 | */ | |
133d55cd | 178 | #define BLKDEV_MAJOR_HASH_SIZE 255 |
1da177e4 LT |
179 | static struct blk_major_name { |
180 | struct blk_major_name *next; | |
181 | int major; | |
182 | char name[16]; | |
a160c615 | 183 | void (*probe)(dev_t devt); |
68eef3b4 | 184 | } *major_names[BLKDEV_MAJOR_HASH_SIZE]; |
e49fbbbf | 185 | static DEFINE_MUTEX(major_names_lock); |
dfbb3409 | 186 | static DEFINE_SPINLOCK(major_names_spinlock); |
1da177e4 LT |
187 | |
188 | /* index in the above - for now: assume no multimajor ranges */ | |
e61eb2e9 | 189 | static inline int major_to_index(unsigned major) |
1da177e4 | 190 | { |
68eef3b4 | 191 | return major % BLKDEV_MAJOR_HASH_SIZE; |
7170be5f NH |
192 | } |
193 | ||
68eef3b4 | 194 | #ifdef CONFIG_PROC_FS |
cf771cb5 | 195 | void blkdev_show(struct seq_file *seqf, off_t offset) |
7170be5f | 196 | { |
68eef3b4 | 197 | struct blk_major_name *dp; |
7170be5f | 198 | |
dfbb3409 | 199 | spin_lock(&major_names_spinlock); |
133d55cd LG |
200 | for (dp = major_names[major_to_index(offset)]; dp; dp = dp->next) |
201 | if (dp->major == offset) | |
cf771cb5 | 202 | seq_printf(seqf, "%3d %s\n", dp->major, dp->name); |
dfbb3409 | 203 | spin_unlock(&major_names_spinlock); |
1da177e4 | 204 | } |
68eef3b4 | 205 | #endif /* CONFIG_PROC_FS */ |
1da177e4 | 206 | |
9e8c0bcc | 207 | /** |
e2b6b301 | 208 | * __register_blkdev - register a new block device |
9e8c0bcc | 209 | * |
f33ff110 SB |
210 | * @major: the requested major device number [1..BLKDEV_MAJOR_MAX-1]. If |
211 | * @major = 0, try to allocate any unused major number. | |
9e8c0bcc | 212 | * @name: the name of the new block device as a zero terminated string |
e2b6b301 | 213 | * @probe: allback that is called on access to any minor number of @major |
9e8c0bcc MN |
214 | * |
215 | * The @name must be unique within the system. | |
216 | * | |
0e056eb5 | 217 | * The return value depends on the @major input parameter: |
218 | * | |
f33ff110 SB |
219 | * - if a major device number was requested in range [1..BLKDEV_MAJOR_MAX-1] |
220 | * then the function returns zero on success, or a negative error code | |
0e056eb5 | 221 | * - if any unused major number was requested with @major = 0 parameter |
9e8c0bcc | 222 | * then the return value is the allocated major number in range |
f33ff110 SB |
223 | * [1..BLKDEV_MAJOR_MAX-1] or a negative error code otherwise |
224 | * | |
225 | * See Documentation/admin-guide/devices.txt for the list of allocated | |
226 | * major numbers. | |
e2b6b301 CH |
227 | * |
228 | * Use register_blkdev instead for any new code. | |
9e8c0bcc | 229 | */ |
a160c615 CH |
230 | int __register_blkdev(unsigned int major, const char *name, |
231 | void (*probe)(dev_t devt)) | |
1da177e4 LT |
232 | { |
233 | struct blk_major_name **n, *p; | |
234 | int index, ret = 0; | |
235 | ||
e49fbbbf | 236 | mutex_lock(&major_names_lock); |
1da177e4 LT |
237 | |
238 | /* temporary */ | |
239 | if (major == 0) { | |
240 | for (index = ARRAY_SIZE(major_names)-1; index > 0; index--) { | |
241 | if (major_names[index] == NULL) | |
242 | break; | |
243 | } | |
244 | ||
245 | if (index == 0) { | |
dfc76d11 KP |
246 | printk("%s: failed to get major for %s\n", |
247 | __func__, name); | |
1da177e4 LT |
248 | ret = -EBUSY; |
249 | goto out; | |
250 | } | |
251 | major = index; | |
252 | ret = major; | |
253 | } | |
254 | ||
133d55cd | 255 | if (major >= BLKDEV_MAJOR_MAX) { |
dfc76d11 KP |
256 | pr_err("%s: major requested (%u) is greater than the maximum (%u) for %s\n", |
257 | __func__, major, BLKDEV_MAJOR_MAX-1, name); | |
133d55cd LG |
258 | |
259 | ret = -EINVAL; | |
260 | goto out; | |
261 | } | |
262 | ||
1da177e4 LT |
263 | p = kmalloc(sizeof(struct blk_major_name), GFP_KERNEL); |
264 | if (p == NULL) { | |
265 | ret = -ENOMEM; | |
266 | goto out; | |
267 | } | |
268 | ||
269 | p->major = major; | |
a160c615 | 270 | p->probe = probe; |
1da177e4 LT |
271 | strlcpy(p->name, name, sizeof(p->name)); |
272 | p->next = NULL; | |
273 | index = major_to_index(major); | |
274 | ||
dfbb3409 | 275 | spin_lock(&major_names_spinlock); |
1da177e4 LT |
276 | for (n = &major_names[index]; *n; n = &(*n)->next) { |
277 | if ((*n)->major == major) | |
278 | break; | |
279 | } | |
280 | if (!*n) | |
281 | *n = p; | |
282 | else | |
283 | ret = -EBUSY; | |
dfbb3409 | 284 | spin_unlock(&major_names_spinlock); |
1da177e4 LT |
285 | |
286 | if (ret < 0) { | |
f33ff110 | 287 | printk("register_blkdev: cannot get major %u for %s\n", |
1da177e4 LT |
288 | major, name); |
289 | kfree(p); | |
290 | } | |
291 | out: | |
e49fbbbf | 292 | mutex_unlock(&major_names_lock); |
1da177e4 LT |
293 | return ret; |
294 | } | |
a160c615 | 295 | EXPORT_SYMBOL(__register_blkdev); |
1da177e4 | 296 | |
f4480240 | 297 | void unregister_blkdev(unsigned int major, const char *name) |
1da177e4 LT |
298 | { |
299 | struct blk_major_name **n; | |
300 | struct blk_major_name *p = NULL; | |
301 | int index = major_to_index(major); | |
1da177e4 | 302 | |
e49fbbbf | 303 | mutex_lock(&major_names_lock); |
dfbb3409 | 304 | spin_lock(&major_names_spinlock); |
1da177e4 LT |
305 | for (n = &major_names[index]; *n; n = &(*n)->next) |
306 | if ((*n)->major == major) | |
307 | break; | |
294462a5 AM |
308 | if (!*n || strcmp((*n)->name, name)) { |
309 | WARN_ON(1); | |
294462a5 | 310 | } else { |
1da177e4 LT |
311 | p = *n; |
312 | *n = p->next; | |
313 | } | |
dfbb3409 | 314 | spin_unlock(&major_names_spinlock); |
e49fbbbf | 315 | mutex_unlock(&major_names_lock); |
1da177e4 | 316 | kfree(p); |
1da177e4 LT |
317 | } |
318 | ||
319 | EXPORT_SYMBOL(unregister_blkdev); | |
320 | ||
7c3f828b | 321 | int blk_alloc_ext_minor(void) |
bcce3de1 | 322 | { |
bab998d6 | 323 | int idx; |
bcce3de1 | 324 | |
22ae8ce8 | 325 | idx = ida_alloc_range(&ext_devt_ida, 0, NR_EXT_DEVT, GFP_KERNEL); |
c4b2b7d1 CH |
326 | if (idx == -ENOSPC) |
327 | return -EBUSY; | |
328 | return idx; | |
bcce3de1 TH |
329 | } |
330 | ||
7c3f828b | 331 | void blk_free_ext_minor(unsigned int minor) |
bcce3de1 | 332 | { |
c4b2b7d1 | 333 | ida_free(&ext_devt_ida, minor); |
6fcc44d1 YY |
334 | } |
335 | ||
1f014290 TH |
336 | static char *bdevt_str(dev_t devt, char *buf) |
337 | { | |
338 | if (MAJOR(devt) <= 0xff && MINOR(devt) <= 0xff) { | |
339 | char tbuf[BDEVT_SIZE]; | |
340 | snprintf(tbuf, BDEVT_SIZE, "%02x%02x", MAJOR(devt), MINOR(devt)); | |
341 | snprintf(buf, BDEVT_SIZE, "%-9s", tbuf); | |
342 | } else | |
343 | snprintf(buf, BDEVT_SIZE, "%03x:%05x", MAJOR(devt), MINOR(devt)); | |
344 | ||
345 | return buf; | |
346 | } | |
347 | ||
bc359d03 CH |
348 | void disk_uevent(struct gendisk *disk, enum kobject_action action) |
349 | { | |
bc359d03 | 350 | struct block_device *part; |
3212135a | 351 | unsigned long idx; |
bc359d03 | 352 | |
3212135a CH |
353 | rcu_read_lock(); |
354 | xa_for_each(&disk->part_tbl, idx, part) { | |
355 | if (bdev_is_partition(part) && !bdev_nr_sectors(part)) | |
356 | continue; | |
498dcc13 | 357 | if (!kobject_get_unless_zero(&part->bd_device.kobj)) |
3212135a CH |
358 | continue; |
359 | ||
360 | rcu_read_unlock(); | |
bc359d03 | 361 | kobject_uevent(bdev_kobj(part), action); |
498dcc13 | 362 | put_device(&part->bd_device); |
3212135a CH |
363 | rcu_read_lock(); |
364 | } | |
365 | rcu_read_unlock(); | |
bc359d03 CH |
366 | } |
367 | EXPORT_SYMBOL_GPL(disk_uevent); | |
368 | ||
9301fe73 CH |
369 | static void disk_scan_partitions(struct gendisk *disk) |
370 | { | |
371 | struct block_device *bdev; | |
372 | ||
373 | if (!get_capacity(disk) || !disk_part_scan_enabled(disk)) | |
374 | return; | |
375 | ||
376 | set_bit(GD_NEED_PART_SCAN, &disk->state); | |
377 | bdev = blkdev_get_by_dev(disk_devt(disk), FMODE_READ, NULL); | |
378 | if (!IS_ERR(bdev)) | |
379 | blkdev_put(bdev, FMODE_READ); | |
380 | } | |
381 | ||
1da177e4 | 382 | /** |
d1254a87 | 383 | * device_add_disk - add disk information to kernel list |
e63a46be | 384 | * @parent: parent device for the disk |
1da177e4 | 385 | * @disk: per-device partitioning information |
fef912bf | 386 | * @groups: Additional per-device sysfs groups |
1da177e4 LT |
387 | * |
388 | * This function registers the partitioning information in @disk | |
389 | * with the kernel. | |
390 | */ | |
83cbce95 | 391 | int device_add_disk(struct device *parent, struct gendisk *disk, |
d1254a87 CH |
392 | const struct attribute_group **groups) |
393 | ||
1da177e4 | 394 | { |
52b85909 | 395 | struct device *ddev = disk_to_dev(disk); |
7c3f828b | 396 | int ret; |
cf0ca9fe | 397 | |
737eb78e DLM |
398 | /* |
399 | * The disk queue should now be all set with enough information about | |
400 | * the device for the elevator code to pick an adequate default | |
401 | * elevator if one is needed, that is, for devices requesting queue | |
402 | * registration. | |
403 | */ | |
d1254a87 | 404 | elevator_init_mq(disk->queue); |
737eb78e | 405 | |
7c3f828b CH |
406 | /* |
407 | * If the driver provides an explicit major number it also must provide | |
408 | * the number of minors numbers supported, and those will be used to | |
409 | * setup the gendisk. | |
410 | * Otherwise just allocate the device numbers for both the whole device | |
411 | * and all partitions from the extended dev_t space. | |
3e1a7ff8 | 412 | */ |
7c3f828b | 413 | if (disk->major) { |
83cbce95 LC |
414 | if (WARN_ON(!disk->minors)) |
415 | return -EINVAL; | |
2e3c73fa CH |
416 | |
417 | if (disk->minors > DISK_MAX_PARTS) { | |
418 | pr_err("block: can't allocate more than %d partitions\n", | |
419 | DISK_MAX_PARTS); | |
420 | disk->minors = DISK_MAX_PARTS; | |
421 | } | |
7c3f828b | 422 | } else { |
83cbce95 LC |
423 | if (WARN_ON(disk->minors)) |
424 | return -EINVAL; | |
3e1a7ff8 | 425 | |
7c3f828b | 426 | ret = blk_alloc_ext_minor(); |
83cbce95 LC |
427 | if (ret < 0) |
428 | return ret; | |
7c3f828b | 429 | disk->major = BLOCK_EXT_MAJOR; |
539711d7 | 430 | disk->first_minor = ret; |
0d1feb72 | 431 | disk->flags |= GENHD_FL_EXT_DEVT; |
3e1a7ff8 | 432 | } |
7c3f828b | 433 | |
83cbce95 LC |
434 | ret = disk_alloc_events(disk); |
435 | if (ret) | |
436 | goto out_free_ext_minor; | |
9f53d2fe | 437 | |
52b85909 CH |
438 | /* delay uevents, until we scanned partition table */ |
439 | dev_set_uevent_suppress(ddev, 1); | |
440 | ||
441 | ddev->parent = parent; | |
442 | ddev->groups = groups; | |
443 | dev_set_name(ddev, "%s", disk->disk_name); | |
8235b5c1 CH |
444 | if (!(disk->flags & GENHD_FL_HIDDEN)) |
445 | ddev->devt = MKDEV(disk->major, disk->first_minor); | |
83cbce95 LC |
446 | ret = device_add(ddev); |
447 | if (ret) | |
448 | goto out_disk_release_events; | |
52b85909 CH |
449 | if (!sysfs_deprecated) { |
450 | ret = sysfs_create_link(block_depr, &ddev->kobj, | |
451 | kobject_name(&ddev->kobj)); | |
83cbce95 LC |
452 | if (ret) |
453 | goto out_device_del; | |
52b85909 CH |
454 | } |
455 | ||
456 | /* | |
457 | * avoid probable deadlock caused by allocating memory with | |
458 | * GFP_KERNEL in runtime_resume callback of its all ancestor | |
459 | * devices | |
460 | */ | |
461 | pm_runtime_set_memalloc_noio(ddev, true); | |
462 | ||
83cbce95 LC |
463 | ret = blk_integrity_add(disk); |
464 | if (ret) | |
465 | goto out_del_block_link; | |
bab53f6b | 466 | |
52b85909 CH |
467 | disk->part0->bd_holder_dir = |
468 | kobject_create_and_add("holders", &ddev->kobj); | |
83cbce95 LC |
469 | if (!disk->part0->bd_holder_dir) |
470 | goto out_del_integrity; | |
52b85909 | 471 | disk->slave_dir = kobject_create_and_add("slaves", &ddev->kobj); |
83cbce95 LC |
472 | if (!disk->slave_dir) |
473 | goto out_put_holder_dir; | |
52b85909 | 474 | |
83cbce95 LC |
475 | ret = bd_register_pending_holders(disk); |
476 | if (ret < 0) | |
477 | goto out_put_slave_dir; | |
52b85909 | 478 | |
83cbce95 LC |
479 | ret = blk_register_queue(disk); |
480 | if (ret) | |
481 | goto out_put_slave_dir; | |
75f4dca5 | 482 | |
8235b5c1 CH |
483 | if (disk->flags & GENHD_FL_HIDDEN) { |
484 | /* | |
485 | * Don't let hidden disks show up in /proc/partitions, | |
486 | * and don't bother scanning for partitions either. | |
487 | */ | |
488 | disk->flags |= GENHD_FL_SUPPRESS_PARTITION_INFO; | |
489 | disk->flags |= GENHD_FL_NO_PART_SCAN; | |
490 | } else { | |
491 | ret = bdi_register(disk->bdi, "%u:%u", | |
492 | disk->major, disk->first_minor); | |
83cbce95 LC |
493 | if (ret) |
494 | goto out_unregister_queue; | |
8235b5c1 | 495 | bdi_set_owner(disk->bdi, ddev); |
83cbce95 LC |
496 | ret = sysfs_create_link(&ddev->kobj, |
497 | &disk->bdi->dev->kobj, "bdi"); | |
498 | if (ret) | |
499 | goto out_unregister_bdi; | |
8235b5c1 | 500 | |
9d5ee676 | 501 | bdev_add(disk->part0, ddev->devt); |
52b85909 CH |
502 | disk_scan_partitions(disk); |
503 | ||
504 | /* | |
505 | * Announce the disk and partitions after all partitions are | |
8235b5c1 | 506 | * created. (for hidden disks uevents remain suppressed forever) |
52b85909 CH |
507 | */ |
508 | dev_set_uevent_suppress(ddev, 0); | |
509 | disk_uevent(disk, KOBJ_ADD); | |
52b85909 CH |
510 | } |
511 | ||
75f4dca5 | 512 | disk_update_readahead(disk); |
77ea887e | 513 | disk_add_events(disk); |
83cbce95 LC |
514 | return 0; |
515 | ||
516 | out_unregister_bdi: | |
517 | if (!(disk->flags & GENHD_FL_HIDDEN)) | |
518 | bdi_unregister(disk->bdi); | |
519 | out_unregister_queue: | |
520 | blk_unregister_queue(disk); | |
521 | out_put_slave_dir: | |
522 | kobject_put(disk->slave_dir); | |
523 | out_put_holder_dir: | |
524 | kobject_put(disk->part0->bd_holder_dir); | |
525 | out_del_integrity: | |
526 | blk_integrity_del(disk); | |
527 | out_del_block_link: | |
528 | if (!sysfs_deprecated) | |
529 | sysfs_remove_link(block_depr, dev_name(ddev)); | |
530 | out_device_del: | |
531 | device_del(ddev); | |
532 | out_disk_release_events: | |
533 | disk_release_events(disk); | |
534 | out_free_ext_minor: | |
535 | if (disk->major == BLOCK_EXT_MAJOR) | |
536 | blk_free_ext_minor(disk->first_minor); | |
537 | return WARN_ON_ONCE(ret); /* keep until all callers handle errors */ | |
1da177e4 | 538 | } |
e63a46be | 539 | EXPORT_SYMBOL(device_add_disk); |
1da177e4 | 540 | |
b5bd357c LC |
541 | /** |
542 | * del_gendisk - remove the gendisk | |
543 | * @disk: the struct gendisk to remove | |
544 | * | |
545 | * Removes the gendisk and all its associated resources. This deletes the | |
546 | * partitions associated with the gendisk, and unregisters the associated | |
547 | * request_queue. | |
548 | * | |
549 | * This is the counter to the respective __device_add_disk() call. | |
550 | * | |
551 | * The final removal of the struct gendisk happens when its refcount reaches 0 | |
552 | * with put_disk(), which should be called after del_gendisk(), if | |
553 | * __device_add_disk() was used. | |
e8c7d14a LC |
554 | * |
555 | * Drivers exist which depend on the release of the gendisk to be synchronous, | |
556 | * it should not be deferred. | |
557 | * | |
558 | * Context: can sleep | |
b5bd357c | 559 | */ |
d2bf1b67 | 560 | void del_gendisk(struct gendisk *disk) |
1da177e4 | 561 | { |
e8c7d14a LC |
562 | might_sleep(); |
563 | ||
9f286992 | 564 | if (WARN_ON_ONCE(!disk_live(disk) && !(disk->flags & GENHD_FL_HIDDEN))) |
6b3ba976 CH |
565 | return; |
566 | ||
25520d55 | 567 | blk_integrity_del(disk); |
77ea887e TH |
568 | disk_del_events(disk); |
569 | ||
a8698707 | 570 | mutex_lock(&disk->open_mutex); |
d7a66574 | 571 | remove_inode_hash(disk->part0->bd_inode); |
d3c4a43d | 572 | blk_drop_partitions(disk); |
a8698707 | 573 | mutex_unlock(&disk->open_mutex); |
c76f48eb | 574 | |
45611837 CH |
575 | fsync_bdev(disk->part0); |
576 | __invalidate_device(disk->part0, true); | |
577 | ||
d2bf1b67 | 578 | set_capacity(disk, 0); |
d2bf1b67 | 579 | |
6b3ba976 | 580 | if (!(disk->flags & GENHD_FL_HIDDEN)) { |
8ddcd653 | 581 | sysfs_remove_link(&disk_to_dev(disk)->kobj, "bdi"); |
6b3ba976 | 582 | |
90f16fdd JK |
583 | /* |
584 | * Unregister bdi before releasing device numbers (as they can | |
585 | * get reused and we'd get clashes in sysfs). | |
586 | */ | |
edb0872f | 587 | bdi_unregister(disk->bdi); |
90f16fdd | 588 | } |
d2bf1b67 | 589 | |
6b3ba976 | 590 | blk_unregister_queue(disk); |
d2bf1b67 | 591 | |
cb8432d6 | 592 | kobject_put(disk->part0->bd_holder_dir); |
d2bf1b67 | 593 | kobject_put(disk->slave_dir); |
d2bf1b67 | 594 | |
8446fe92 | 595 | part_stat_set_all(disk->part0, 0); |
cb8432d6 | 596 | disk->part0->bd_stamp = 0; |
d2bf1b67 TH |
597 | if (!sysfs_deprecated) |
598 | sysfs_remove_link(block_depr, dev_name(disk_to_dev(disk))); | |
25e823c8 | 599 | pm_runtime_set_memalloc_noio(disk_to_dev(disk), false); |
d2bf1b67 | 600 | device_del(disk_to_dev(disk)); |
1da177e4 | 601 | } |
d2bf1b67 | 602 | EXPORT_SYMBOL(del_gendisk); |
1da177e4 | 603 | |
99e6608c VV |
604 | /* sysfs access to bad-blocks list. */ |
605 | static ssize_t disk_badblocks_show(struct device *dev, | |
606 | struct device_attribute *attr, | |
607 | char *page) | |
608 | { | |
609 | struct gendisk *disk = dev_to_disk(dev); | |
610 | ||
611 | if (!disk->bb) | |
612 | return sprintf(page, "\n"); | |
613 | ||
614 | return badblocks_show(disk->bb, page, 0); | |
615 | } | |
616 | ||
617 | static ssize_t disk_badblocks_store(struct device *dev, | |
618 | struct device_attribute *attr, | |
619 | const char *page, size_t len) | |
620 | { | |
621 | struct gendisk *disk = dev_to_disk(dev); | |
622 | ||
623 | if (!disk->bb) | |
624 | return -ENXIO; | |
625 | ||
626 | return badblocks_store(disk->bb, page, len, 0); | |
627 | } | |
628 | ||
22ae8ce8 | 629 | void blk_request_module(dev_t devt) |
bd8eff3b | 630 | { |
a160c615 CH |
631 | unsigned int major = MAJOR(devt); |
632 | struct blk_major_name **n; | |
633 | ||
634 | mutex_lock(&major_names_lock); | |
635 | for (n = &major_names[major_to_index(major)]; *n; n = &(*n)->next) { | |
636 | if ((*n)->major == major && (*n)->probe) { | |
637 | (*n)->probe(devt); | |
638 | mutex_unlock(&major_names_lock); | |
639 | return; | |
640 | } | |
641 | } | |
642 | mutex_unlock(&major_names_lock); | |
643 | ||
bd8eff3b CH |
644 | if (request_module("block-major-%d-%d", MAJOR(devt), MINOR(devt)) > 0) |
645 | /* Make old-style 2.4 aliases work */ | |
646 | request_module("block-major-%d", MAJOR(devt)); | |
647 | } | |
648 | ||
5c6f35c5 GKH |
649 | /* |
650 | * print a full list of all partitions - intended for places where the root | |
651 | * filesystem can't be mounted and thus to give the victim some idea of what | |
652 | * went wrong | |
653 | */ | |
654 | void __init printk_all_partitions(void) | |
655 | { | |
def4e38d TH |
656 | struct class_dev_iter iter; |
657 | struct device *dev; | |
658 | ||
659 | class_dev_iter_init(&iter, &block_class, NULL, &disk_type); | |
660 | while ((dev = class_dev_iter_next(&iter))) { | |
661 | struct gendisk *disk = dev_to_disk(dev); | |
ad1eaa53 | 662 | struct block_device *part; |
1f014290 | 663 | char devt_buf[BDEVT_SIZE]; |
e559f58d | 664 | unsigned long idx; |
def4e38d TH |
665 | |
666 | /* | |
667 | * Don't show empty devices or things that have been | |
25985edc | 668 | * suppressed |
def4e38d TH |
669 | */ |
670 | if (get_capacity(disk) == 0 || | |
671 | (disk->flags & GENHD_FL_SUPPRESS_PARTITION_INFO)) | |
672 | continue; | |
673 | ||
674 | /* | |
e559f58d CH |
675 | * Note, unlike /proc/partitions, I am showing the numbers in |
676 | * hex - the same format as the root= option takes. | |
def4e38d | 677 | */ |
e559f58d CH |
678 | rcu_read_lock(); |
679 | xa_for_each(&disk->part_tbl, idx, part) { | |
680 | if (!bdev_nr_sectors(part)) | |
681 | continue; | |
a9e7bc3d | 682 | printk("%s%s %10llu %pg %s", |
e559f58d | 683 | bdev_is_partition(part) ? " " : "", |
ad1eaa53 | 684 | bdevt_str(part->bd_dev, devt_buf), |
a9e7bc3d | 685 | bdev_nr_sectors(part) >> 1, part, |
ad1eaa53 CH |
686 | part->bd_meta_info ? |
687 | part->bd_meta_info->uuid : ""); | |
e559f58d | 688 | if (bdev_is_partition(part)) |
074a7aca | 689 | printk("\n"); |
e559f58d CH |
690 | else if (dev->parent && dev->parent->driver) |
691 | printk(" driver: %s\n", | |
692 | dev->parent->driver->name); | |
693 | else | |
694 | printk(" (driver?)\n"); | |
074a7aca | 695 | } |
e559f58d | 696 | rcu_read_unlock(); |
def4e38d TH |
697 | } |
698 | class_dev_iter_exit(&iter); | |
dd2a345f DG |
699 | } |
700 | ||
1da177e4 LT |
701 | #ifdef CONFIG_PROC_FS |
702 | /* iterator */ | |
def4e38d | 703 | static void *disk_seqf_start(struct seq_file *seqf, loff_t *pos) |
68c4d4a7 | 704 | { |
def4e38d TH |
705 | loff_t skip = *pos; |
706 | struct class_dev_iter *iter; | |
707 | struct device *dev; | |
68c4d4a7 | 708 | |
aeb3d3a8 | 709 | iter = kmalloc(sizeof(*iter), GFP_KERNEL); |
def4e38d TH |
710 | if (!iter) |
711 | return ERR_PTR(-ENOMEM); | |
712 | ||
713 | seqf->private = iter; | |
714 | class_dev_iter_init(iter, &block_class, NULL, &disk_type); | |
715 | do { | |
716 | dev = class_dev_iter_next(iter); | |
717 | if (!dev) | |
718 | return NULL; | |
719 | } while (skip--); | |
720 | ||
721 | return dev_to_disk(dev); | |
68c4d4a7 GKH |
722 | } |
723 | ||
def4e38d | 724 | static void *disk_seqf_next(struct seq_file *seqf, void *v, loff_t *pos) |
1da177e4 | 725 | { |
edfaa7c3 | 726 | struct device *dev; |
1da177e4 | 727 | |
def4e38d TH |
728 | (*pos)++; |
729 | dev = class_dev_iter_next(seqf->private); | |
2ac3cee5 | 730 | if (dev) |
68c4d4a7 | 731 | return dev_to_disk(dev); |
2ac3cee5 | 732 | |
1da177e4 LT |
733 | return NULL; |
734 | } | |
735 | ||
def4e38d | 736 | static void disk_seqf_stop(struct seq_file *seqf, void *v) |
27f30251 | 737 | { |
def4e38d | 738 | struct class_dev_iter *iter = seqf->private; |
27f30251 | 739 | |
def4e38d TH |
740 | /* stop is called even after start failed :-( */ |
741 | if (iter) { | |
742 | class_dev_iter_exit(iter); | |
743 | kfree(iter); | |
77da1605 | 744 | seqf->private = NULL; |
5c0ef6d0 | 745 | } |
1da177e4 LT |
746 | } |
747 | ||
def4e38d | 748 | static void *show_partition_start(struct seq_file *seqf, loff_t *pos) |
1da177e4 | 749 | { |
06768067 | 750 | void *p; |
def4e38d TH |
751 | |
752 | p = disk_seqf_start(seqf, pos); | |
b9f985b6 | 753 | if (!IS_ERR_OR_NULL(p) && !*pos) |
def4e38d TH |
754 | seq_puts(seqf, "major minor #blocks name\n\n"); |
755 | return p; | |
1da177e4 LT |
756 | } |
757 | ||
cf771cb5 | 758 | static int show_partition(struct seq_file *seqf, void *v) |
1da177e4 LT |
759 | { |
760 | struct gendisk *sgp = v; | |
ad1eaa53 | 761 | struct block_device *part; |
ecc75a98 | 762 | unsigned long idx; |
1da177e4 | 763 | |
1da177e4 | 764 | /* Don't show non-partitionable removeable devices or empty devices */ |
d27769ec | 765 | if (!get_capacity(sgp) || (!disk_max_parts(sgp) && |
f331c029 | 766 | (sgp->flags & GENHD_FL_REMOVABLE))) |
1da177e4 LT |
767 | return 0; |
768 | if (sgp->flags & GENHD_FL_SUPPRESS_PARTITION_INFO) | |
769 | return 0; | |
770 | ||
ecc75a98 CH |
771 | rcu_read_lock(); |
772 | xa_for_each(&sgp->part_tbl, idx, part) { | |
773 | if (!bdev_nr_sectors(part)) | |
774 | continue; | |
a291bb43 | 775 | seq_printf(seqf, "%4d %7d %10llu %pg\n", |
ad1eaa53 | 776 | MAJOR(part->bd_dev), MINOR(part->bd_dev), |
a291bb43 | 777 | bdev_nr_sectors(part) >> 1, part); |
ecc75a98 CH |
778 | } |
779 | rcu_read_unlock(); | |
1da177e4 LT |
780 | return 0; |
781 | } | |
782 | ||
f500975a | 783 | static const struct seq_operations partitions_op = { |
def4e38d TH |
784 | .start = show_partition_start, |
785 | .next = disk_seqf_next, | |
786 | .stop = disk_seqf_stop, | |
edfaa7c3 | 787 | .show = show_partition |
1da177e4 LT |
788 | }; |
789 | #endif | |
790 | ||
1da177e4 LT |
791 | static int __init genhd_device_init(void) |
792 | { | |
e105b8bf DW |
793 | int error; |
794 | ||
795 | block_class.dev_kobj = sysfs_dev_block_kobj; | |
796 | error = class_register(&block_class); | |
ee27a558 RM |
797 | if (unlikely(error)) |
798 | return error; | |
1da177e4 | 799 | blk_dev_init(); |
edfaa7c3 | 800 | |
561ec68e ZY |
801 | register_blkdev(BLOCK_EXT_MAJOR, "blkext"); |
802 | ||
edfaa7c3 | 803 | /* create top-level block dir */ |
e52eec13 AK |
804 | if (!sysfs_deprecated) |
805 | block_depr = kobject_create_and_add("block", NULL); | |
830d3cfb | 806 | return 0; |
1da177e4 LT |
807 | } |
808 | ||
809 | subsys_initcall(genhd_device_init); | |
810 | ||
edfaa7c3 KS |
811 | static ssize_t disk_range_show(struct device *dev, |
812 | struct device_attribute *attr, char *buf) | |
1da177e4 | 813 | { |
edfaa7c3 | 814 | struct gendisk *disk = dev_to_disk(dev); |
1da177e4 | 815 | |
edfaa7c3 | 816 | return sprintf(buf, "%d\n", disk->minors); |
1da177e4 LT |
817 | } |
818 | ||
1f014290 TH |
819 | static ssize_t disk_ext_range_show(struct device *dev, |
820 | struct device_attribute *attr, char *buf) | |
821 | { | |
822 | struct gendisk *disk = dev_to_disk(dev); | |
823 | ||
b5d0b9df | 824 | return sprintf(buf, "%d\n", disk_max_parts(disk)); |
1f014290 TH |
825 | } |
826 | ||
edfaa7c3 KS |
827 | static ssize_t disk_removable_show(struct device *dev, |
828 | struct device_attribute *attr, char *buf) | |
a7fd6706 | 829 | { |
edfaa7c3 | 830 | struct gendisk *disk = dev_to_disk(dev); |
a7fd6706 | 831 | |
edfaa7c3 KS |
832 | return sprintf(buf, "%d\n", |
833 | (disk->flags & GENHD_FL_REMOVABLE ? 1 : 0)); | |
a7fd6706 KS |
834 | } |
835 | ||
8ddcd653 CH |
836 | static ssize_t disk_hidden_show(struct device *dev, |
837 | struct device_attribute *attr, char *buf) | |
838 | { | |
839 | struct gendisk *disk = dev_to_disk(dev); | |
840 | ||
841 | return sprintf(buf, "%d\n", | |
842 | (disk->flags & GENHD_FL_HIDDEN ? 1 : 0)); | |
843 | } | |
844 | ||
1c9ce527 KS |
845 | static ssize_t disk_ro_show(struct device *dev, |
846 | struct device_attribute *attr, char *buf) | |
847 | { | |
848 | struct gendisk *disk = dev_to_disk(dev); | |
849 | ||
b7db9956 | 850 | return sprintf(buf, "%d\n", get_disk_ro(disk) ? 1 : 0); |
1c9ce527 KS |
851 | } |
852 | ||
3ad5cee5 CH |
853 | ssize_t part_size_show(struct device *dev, |
854 | struct device_attribute *attr, char *buf) | |
855 | { | |
0d02129e | 856 | return sprintf(buf, "%llu\n", bdev_nr_sectors(dev_to_bdev(dev))); |
3ad5cee5 CH |
857 | } |
858 | ||
859 | ssize_t part_stat_show(struct device *dev, | |
860 | struct device_attribute *attr, char *buf) | |
861 | { | |
0d02129e CH |
862 | struct block_device *bdev = dev_to_bdev(dev); |
863 | struct request_queue *q = bdev->bd_disk->queue; | |
ea18e0f0 | 864 | struct disk_stats stat; |
3ad5cee5 CH |
865 | unsigned int inflight; |
866 | ||
0d02129e | 867 | part_stat_read_all(bdev, &stat); |
b2f609e1 | 868 | if (queue_is_mq(q)) |
0d02129e | 869 | inflight = blk_mq_in_flight(q, bdev); |
b2f609e1 | 870 | else |
0d02129e | 871 | inflight = part_in_flight(bdev); |
ea18e0f0 | 872 | |
3ad5cee5 CH |
873 | return sprintf(buf, |
874 | "%8lu %8lu %8llu %8u " | |
875 | "%8lu %8lu %8llu %8u " | |
876 | "%8u %8u %8u " | |
877 | "%8lu %8lu %8llu %8u " | |
878 | "%8lu %8u" | |
879 | "\n", | |
ea18e0f0 KK |
880 | stat.ios[STAT_READ], |
881 | stat.merges[STAT_READ], | |
882 | (unsigned long long)stat.sectors[STAT_READ], | |
883 | (unsigned int)div_u64(stat.nsecs[STAT_READ], NSEC_PER_MSEC), | |
884 | stat.ios[STAT_WRITE], | |
885 | stat.merges[STAT_WRITE], | |
886 | (unsigned long long)stat.sectors[STAT_WRITE], | |
887 | (unsigned int)div_u64(stat.nsecs[STAT_WRITE], NSEC_PER_MSEC), | |
3ad5cee5 | 888 | inflight, |
ea18e0f0 | 889 | jiffies_to_msecs(stat.io_ticks), |
8cd5b8fc KK |
890 | (unsigned int)div_u64(stat.nsecs[STAT_READ] + |
891 | stat.nsecs[STAT_WRITE] + | |
892 | stat.nsecs[STAT_DISCARD] + | |
893 | stat.nsecs[STAT_FLUSH], | |
894 | NSEC_PER_MSEC), | |
ea18e0f0 KK |
895 | stat.ios[STAT_DISCARD], |
896 | stat.merges[STAT_DISCARD], | |
897 | (unsigned long long)stat.sectors[STAT_DISCARD], | |
898 | (unsigned int)div_u64(stat.nsecs[STAT_DISCARD], NSEC_PER_MSEC), | |
899 | stat.ios[STAT_FLUSH], | |
900 | (unsigned int)div_u64(stat.nsecs[STAT_FLUSH], NSEC_PER_MSEC)); | |
3ad5cee5 CH |
901 | } |
902 | ||
903 | ssize_t part_inflight_show(struct device *dev, struct device_attribute *attr, | |
904 | char *buf) | |
905 | { | |
0d02129e CH |
906 | struct block_device *bdev = dev_to_bdev(dev); |
907 | struct request_queue *q = bdev->bd_disk->queue; | |
3ad5cee5 CH |
908 | unsigned int inflight[2]; |
909 | ||
b2f609e1 | 910 | if (queue_is_mq(q)) |
0d02129e | 911 | blk_mq_in_flight_rw(q, bdev, inflight); |
b2f609e1 | 912 | else |
0d02129e | 913 | part_in_flight_rw(bdev, inflight); |
b2f609e1 | 914 | |
3ad5cee5 CH |
915 | return sprintf(buf, "%8u %8u\n", inflight[0], inflight[1]); |
916 | } | |
917 | ||
edfaa7c3 KS |
918 | static ssize_t disk_capability_show(struct device *dev, |
919 | struct device_attribute *attr, char *buf) | |
86ce18d7 | 920 | { |
edfaa7c3 KS |
921 | struct gendisk *disk = dev_to_disk(dev); |
922 | ||
923 | return sprintf(buf, "%x\n", disk->flags); | |
86ce18d7 | 924 | } |
edfaa7c3 | 925 | |
c72758f3 MP |
926 | static ssize_t disk_alignment_offset_show(struct device *dev, |
927 | struct device_attribute *attr, | |
928 | char *buf) | |
929 | { | |
930 | struct gendisk *disk = dev_to_disk(dev); | |
931 | ||
932 | return sprintf(buf, "%d\n", queue_alignment_offset(disk->queue)); | |
933 | } | |
934 | ||
86b37281 MP |
935 | static ssize_t disk_discard_alignment_show(struct device *dev, |
936 | struct device_attribute *attr, | |
937 | char *buf) | |
938 | { | |
939 | struct gendisk *disk = dev_to_disk(dev); | |
940 | ||
dd3d145d | 941 | return sprintf(buf, "%d\n", queue_discard_alignment(disk->queue)); |
86b37281 MP |
942 | } |
943 | ||
13927b31 MC |
944 | static ssize_t diskseq_show(struct device *dev, |
945 | struct device_attribute *attr, char *buf) | |
946 | { | |
947 | struct gendisk *disk = dev_to_disk(dev); | |
948 | ||
949 | return sprintf(buf, "%llu\n", disk->diskseq); | |
950 | } | |
951 | ||
5657a819 JP |
952 | static DEVICE_ATTR(range, 0444, disk_range_show, NULL); |
953 | static DEVICE_ATTR(ext_range, 0444, disk_ext_range_show, NULL); | |
954 | static DEVICE_ATTR(removable, 0444, disk_removable_show, NULL); | |
955 | static DEVICE_ATTR(hidden, 0444, disk_hidden_show, NULL); | |
956 | static DEVICE_ATTR(ro, 0444, disk_ro_show, NULL); | |
957 | static DEVICE_ATTR(size, 0444, part_size_show, NULL); | |
958 | static DEVICE_ATTR(alignment_offset, 0444, disk_alignment_offset_show, NULL); | |
959 | static DEVICE_ATTR(discard_alignment, 0444, disk_discard_alignment_show, NULL); | |
960 | static DEVICE_ATTR(capability, 0444, disk_capability_show, NULL); | |
961 | static DEVICE_ATTR(stat, 0444, part_stat_show, NULL); | |
962 | static DEVICE_ATTR(inflight, 0444, part_inflight_show, NULL); | |
963 | static DEVICE_ATTR(badblocks, 0644, disk_badblocks_show, disk_badblocks_store); | |
13927b31 | 964 | static DEVICE_ATTR(diskseq, 0444, diskseq_show, NULL); |
3ad5cee5 | 965 | |
c17bb495 | 966 | #ifdef CONFIG_FAIL_MAKE_REQUEST |
3ad5cee5 CH |
967 | ssize_t part_fail_show(struct device *dev, |
968 | struct device_attribute *attr, char *buf) | |
969 | { | |
0d02129e | 970 | return sprintf(buf, "%d\n", dev_to_bdev(dev)->bd_make_it_fail); |
3ad5cee5 CH |
971 | } |
972 | ||
973 | ssize_t part_fail_store(struct device *dev, | |
974 | struct device_attribute *attr, | |
975 | const char *buf, size_t count) | |
976 | { | |
3ad5cee5 CH |
977 | int i; |
978 | ||
979 | if (count > 0 && sscanf(buf, "%d", &i) > 0) | |
0d02129e | 980 | dev_to_bdev(dev)->bd_make_it_fail = i; |
3ad5cee5 CH |
981 | |
982 | return count; | |
983 | } | |
984 | ||
edfaa7c3 | 985 | static struct device_attribute dev_attr_fail = |
5657a819 | 986 | __ATTR(make-it-fail, 0644, part_fail_show, part_fail_store); |
3ad5cee5 CH |
987 | #endif /* CONFIG_FAIL_MAKE_REQUEST */ |
988 | ||
581d4e28 JA |
989 | #ifdef CONFIG_FAIL_IO_TIMEOUT |
990 | static struct device_attribute dev_attr_fail_timeout = | |
5657a819 | 991 | __ATTR(io-timeout-fail, 0644, part_timeout_show, part_timeout_store); |
581d4e28 | 992 | #endif |
edfaa7c3 KS |
993 | |
994 | static struct attribute *disk_attrs[] = { | |
995 | &dev_attr_range.attr, | |
1f014290 | 996 | &dev_attr_ext_range.attr, |
edfaa7c3 | 997 | &dev_attr_removable.attr, |
8ddcd653 | 998 | &dev_attr_hidden.attr, |
1c9ce527 | 999 | &dev_attr_ro.attr, |
edfaa7c3 | 1000 | &dev_attr_size.attr, |
c72758f3 | 1001 | &dev_attr_alignment_offset.attr, |
86b37281 | 1002 | &dev_attr_discard_alignment.attr, |
edfaa7c3 KS |
1003 | &dev_attr_capability.attr, |
1004 | &dev_attr_stat.attr, | |
316d315b | 1005 | &dev_attr_inflight.attr, |
99e6608c | 1006 | &dev_attr_badblocks.attr, |
2bc8cda5 CH |
1007 | &dev_attr_events.attr, |
1008 | &dev_attr_events_async.attr, | |
1009 | &dev_attr_events_poll_msecs.attr, | |
13927b31 | 1010 | &dev_attr_diskseq.attr, |
edfaa7c3 KS |
1011 | #ifdef CONFIG_FAIL_MAKE_REQUEST |
1012 | &dev_attr_fail.attr, | |
581d4e28 JA |
1013 | #endif |
1014 | #ifdef CONFIG_FAIL_IO_TIMEOUT | |
1015 | &dev_attr_fail_timeout.attr, | |
edfaa7c3 KS |
1016 | #endif |
1017 | NULL | |
1018 | }; | |
1019 | ||
9438b3e0 DW |
1020 | static umode_t disk_visible(struct kobject *kobj, struct attribute *a, int n) |
1021 | { | |
1022 | struct device *dev = container_of(kobj, typeof(*dev), kobj); | |
1023 | struct gendisk *disk = dev_to_disk(dev); | |
1024 | ||
1025 | if (a == &dev_attr_badblocks.attr && !disk->bb) | |
1026 | return 0; | |
1027 | return a->mode; | |
1028 | } | |
1029 | ||
edfaa7c3 KS |
1030 | static struct attribute_group disk_attr_group = { |
1031 | .attrs = disk_attrs, | |
9438b3e0 | 1032 | .is_visible = disk_visible, |
edfaa7c3 KS |
1033 | }; |
1034 | ||
a4dbd674 | 1035 | static const struct attribute_group *disk_attr_groups[] = { |
edfaa7c3 KS |
1036 | &disk_attr_group, |
1037 | NULL | |
1da177e4 LT |
1038 | }; |
1039 | ||
b5bd357c LC |
1040 | /** |
1041 | * disk_release - releases all allocated resources of the gendisk | |
1042 | * @dev: the device representing this disk | |
1043 | * | |
1044 | * This function releases all allocated resources of the gendisk. | |
1045 | * | |
b5bd357c LC |
1046 | * Drivers which used __device_add_disk() have a gendisk with a request_queue |
1047 | * assigned. Since the request_queue sits on top of the gendisk for these | |
1048 | * drivers we also call blk_put_queue() for them, and we expect the | |
1049 | * request_queue refcount to reach 0 at this point, and so the request_queue | |
1050 | * will also be freed prior to the disk. | |
e8c7d14a LC |
1051 | * |
1052 | * Context: can sleep | |
b5bd357c | 1053 | */ |
edfaa7c3 | 1054 | static void disk_release(struct device *dev) |
1da177e4 | 1055 | { |
edfaa7c3 KS |
1056 | struct gendisk *disk = dev_to_disk(dev); |
1057 | ||
e8c7d14a LC |
1058 | might_sleep(); |
1059 | ||
77ea887e | 1060 | disk_release_events(disk); |
1da177e4 | 1061 | kfree(disk->random); |
a33df75c | 1062 | xa_destroy(&disk->part_tbl); |
d152c682 | 1063 | disk->queue->disk = NULL; |
61a35cfc | 1064 | blk_put_queue(disk->queue); |
2f4731dc | 1065 | iput(disk->part0->bd_inode); /* frees the disk */ |
1da177e4 | 1066 | } |
87eb7107 MC |
1067 | |
1068 | static int block_uevent(struct device *dev, struct kobj_uevent_env *env) | |
1069 | { | |
1070 | struct gendisk *disk = dev_to_disk(dev); | |
1071 | ||
1072 | return add_uevent_var(env, "DISKSEQ=%llu", disk->diskseq); | |
1073 | } | |
1074 | ||
edfaa7c3 KS |
1075 | struct class block_class = { |
1076 | .name = "block", | |
87eb7107 | 1077 | .dev_uevent = block_uevent, |
1da177e4 LT |
1078 | }; |
1079 | ||
3c2670e6 | 1080 | static char *block_devnode(struct device *dev, umode_t *mode, |
4e4098a3 | 1081 | kuid_t *uid, kgid_t *gid) |
b03f38b6 KS |
1082 | { |
1083 | struct gendisk *disk = dev_to_disk(dev); | |
1084 | ||
348e114b CH |
1085 | if (disk->fops->devnode) |
1086 | return disk->fops->devnode(disk, mode); | |
b03f38b6 KS |
1087 | return NULL; |
1088 | } | |
1089 | ||
ef45fe47 | 1090 | const struct device_type disk_type = { |
edfaa7c3 KS |
1091 | .name = "disk", |
1092 | .groups = disk_attr_groups, | |
1093 | .release = disk_release, | |
e454cea2 | 1094 | .devnode = block_devnode, |
1da177e4 LT |
1095 | }; |
1096 | ||
a6e2ba88 | 1097 | #ifdef CONFIG_PROC_FS |
cf771cb5 TH |
1098 | /* |
1099 | * aggregate disk stat collector. Uses the same stats that the sysfs | |
1100 | * entries do, above, but makes them available through one seq_file. | |
1101 | * | |
1102 | * The output looks suspiciously like /proc/partitions with a bunch of | |
1103 | * extra fields. | |
1104 | */ | |
1105 | static int diskstats_show(struct seq_file *seqf, void *v) | |
1da177e4 LT |
1106 | { |
1107 | struct gendisk *gp = v; | |
ad1eaa53 | 1108 | struct block_device *hd; |
e016b782 | 1109 | unsigned int inflight; |
ea18e0f0 | 1110 | struct disk_stats stat; |
7fae67cc | 1111 | unsigned long idx; |
1da177e4 LT |
1112 | |
1113 | /* | |
ed9e1982 | 1114 | if (&disk_to_dev(gp)->kobj.entry == block_class.devices.next) |
cf771cb5 | 1115 | seq_puts(seqf, "major minor name" |
1da177e4 LT |
1116 | " rio rmerge rsect ruse wio wmerge " |
1117 | "wsect wuse running use aveq" | |
1118 | "\n\n"); | |
1119 | */ | |
9f5e4865 | 1120 | |
7fae67cc CH |
1121 | rcu_read_lock(); |
1122 | xa_for_each(&gp->part_tbl, idx, hd) { | |
1123 | if (bdev_is_partition(hd) && !bdev_nr_sectors(hd)) | |
1124 | continue; | |
0d02129e | 1125 | part_stat_read_all(hd, &stat); |
b2f609e1 | 1126 | if (queue_is_mq(gp->queue)) |
ad1eaa53 | 1127 | inflight = blk_mq_in_flight(gp->queue, hd); |
b2f609e1 | 1128 | else |
ad1eaa53 | 1129 | inflight = part_in_flight(hd); |
ea18e0f0 | 1130 | |
26e2d7a3 | 1131 | seq_printf(seqf, "%4d %7d %pg " |
bdca3c87 MC |
1132 | "%lu %lu %lu %u " |
1133 | "%lu %lu %lu %u " | |
1134 | "%u %u %u " | |
b6866318 KK |
1135 | "%lu %lu %lu %u " |
1136 | "%lu %u" | |
1137 | "\n", | |
26e2d7a3 | 1138 | MAJOR(hd->bd_dev), MINOR(hd->bd_dev), hd, |
ea18e0f0 KK |
1139 | stat.ios[STAT_READ], |
1140 | stat.merges[STAT_READ], | |
1141 | stat.sectors[STAT_READ], | |
1142 | (unsigned int)div_u64(stat.nsecs[STAT_READ], | |
1143 | NSEC_PER_MSEC), | |
1144 | stat.ios[STAT_WRITE], | |
1145 | stat.merges[STAT_WRITE], | |
1146 | stat.sectors[STAT_WRITE], | |
1147 | (unsigned int)div_u64(stat.nsecs[STAT_WRITE], | |
1148 | NSEC_PER_MSEC), | |
e016b782 | 1149 | inflight, |
ea18e0f0 | 1150 | jiffies_to_msecs(stat.io_ticks), |
8cd5b8fc KK |
1151 | (unsigned int)div_u64(stat.nsecs[STAT_READ] + |
1152 | stat.nsecs[STAT_WRITE] + | |
1153 | stat.nsecs[STAT_DISCARD] + | |
1154 | stat.nsecs[STAT_FLUSH], | |
1155 | NSEC_PER_MSEC), | |
ea18e0f0 KK |
1156 | stat.ios[STAT_DISCARD], |
1157 | stat.merges[STAT_DISCARD], | |
1158 | stat.sectors[STAT_DISCARD], | |
1159 | (unsigned int)div_u64(stat.nsecs[STAT_DISCARD], | |
1160 | NSEC_PER_MSEC), | |
1161 | stat.ios[STAT_FLUSH], | |
1162 | (unsigned int)div_u64(stat.nsecs[STAT_FLUSH], | |
1163 | NSEC_PER_MSEC) | |
28f39d55 | 1164 | ); |
1da177e4 | 1165 | } |
7fae67cc | 1166 | rcu_read_unlock(); |
9f5e4865 | 1167 | |
1da177e4 LT |
1168 | return 0; |
1169 | } | |
1170 | ||
31d85ab2 | 1171 | static const struct seq_operations diskstats_op = { |
def4e38d TH |
1172 | .start = disk_seqf_start, |
1173 | .next = disk_seqf_next, | |
1174 | .stop = disk_seqf_stop, | |
1da177e4 LT |
1175 | .show = diskstats_show |
1176 | }; | |
f500975a AD |
1177 | |
1178 | static int __init proc_genhd_init(void) | |
1179 | { | |
fddda2b7 CH |
1180 | proc_create_seq("diskstats", 0, NULL, &diskstats_op); |
1181 | proc_create_seq("partitions", 0, NULL, &partitions_op); | |
f500975a AD |
1182 | return 0; |
1183 | } | |
1184 | module_init(proc_genhd_init); | |
a6e2ba88 | 1185 | #endif /* CONFIG_PROC_FS */ |
1da177e4 | 1186 | |
c97d93c3 CH |
1187 | dev_t part_devt(struct gendisk *disk, u8 partno) |
1188 | { | |
0e0ccdec | 1189 | struct block_device *part; |
c97d93c3 CH |
1190 | dev_t devt = 0; |
1191 | ||
0e0ccdec CH |
1192 | rcu_read_lock(); |
1193 | part = xa_load(&disk->part_tbl, partno); | |
1194 | if (part) | |
c97d93c3 | 1195 | devt = part->bd_dev; |
0e0ccdec | 1196 | rcu_read_unlock(); |
c97d93c3 CH |
1197 | |
1198 | return devt; | |
1199 | } | |
1200 | ||
cf771cb5 | 1201 | dev_t blk_lookup_devt(const char *name, int partno) |
a142be85 | 1202 | { |
def4e38d TH |
1203 | dev_t devt = MKDEV(0, 0); |
1204 | struct class_dev_iter iter; | |
1205 | struct device *dev; | |
a142be85 | 1206 | |
def4e38d TH |
1207 | class_dev_iter_init(&iter, &block_class, NULL, &disk_type); |
1208 | while ((dev = class_dev_iter_next(&iter))) { | |
a142be85 | 1209 | struct gendisk *disk = dev_to_disk(dev); |
a142be85 | 1210 | |
3ada8b7e | 1211 | if (strcmp(dev_name(dev), name)) |
f331c029 | 1212 | continue; |
f331c029 | 1213 | |
41b8c853 NB |
1214 | if (partno < disk->minors) { |
1215 | /* We need to return the right devno, even | |
1216 | * if the partition doesn't exist yet. | |
1217 | */ | |
1218 | devt = MKDEV(MAJOR(dev->devt), | |
1219 | MINOR(dev->devt) + partno); | |
c97d93c3 CH |
1220 | } else { |
1221 | devt = part_devt(disk, partno); | |
1222 | if (devt) | |
1223 | break; | |
def4e38d | 1224 | } |
5c0ef6d0 | 1225 | } |
def4e38d | 1226 | class_dev_iter_exit(&iter); |
edfaa7c3 KS |
1227 | return devt; |
1228 | } | |
edfaa7c3 | 1229 | |
4a1fa41d CH |
1230 | struct gendisk *__alloc_disk_node(struct request_queue *q, int node_id, |
1231 | struct lock_class_key *lkclass) | |
1946089a CL |
1232 | { |
1233 | struct gendisk *disk; | |
1234 | ||
61a35cfc CH |
1235 | if (!blk_get_queue(q)) |
1236 | return NULL; | |
1237 | ||
c1b511eb | 1238 | disk = kzalloc_node(sizeof(struct gendisk), GFP_KERNEL, node_id); |
f93af2a4 | 1239 | if (!disk) |
61a35cfc | 1240 | goto out_put_queue; |
6c23a968 | 1241 | |
edb0872f CH |
1242 | disk->bdi = bdi_alloc(node_id); |
1243 | if (!disk->bdi) | |
1244 | goto out_free_disk; | |
1245 | ||
cb8432d6 CH |
1246 | disk->part0 = bdev_alloc(disk, 0); |
1247 | if (!disk->part0) | |
edb0872f | 1248 | goto out_free_bdi; |
22ae8ce8 | 1249 | |
f93af2a4 | 1250 | disk->node_id = node_id; |
a8698707 | 1251 | mutex_init(&disk->open_mutex); |
a33df75c CH |
1252 | xa_init(&disk->part_tbl); |
1253 | if (xa_insert(&disk->part_tbl, 0, disk->part0, GFP_KERNEL)) | |
1254 | goto out_destroy_part_tbl; | |
f93af2a4 | 1255 | |
f93af2a4 CH |
1256 | rand_initialize_disk(disk); |
1257 | disk_to_dev(disk)->class = &block_class; | |
1258 | disk_to_dev(disk)->type = &disk_type; | |
1259 | device_initialize(disk_to_dev(disk)); | |
cf179948 | 1260 | inc_diskseq(disk); |
4a1fa41d | 1261 | disk->queue = q; |
d152c682 | 1262 | q->disk = disk; |
4dcc4874 | 1263 | lockdep_init_map(&disk->lockdep_map, "(bio completion)", lkclass, 0); |
0dbcfe24 CH |
1264 | #ifdef CONFIG_BLOCK_HOLDER_DEPRECATED |
1265 | INIT_LIST_HEAD(&disk->slave_bdevs); | |
1266 | #endif | |
1da177e4 | 1267 | return disk; |
f93af2a4 | 1268 | |
a33df75c CH |
1269 | out_destroy_part_tbl: |
1270 | xa_destroy(&disk->part_tbl); | |
2f4731dc | 1271 | iput(disk->part0->bd_inode); |
edb0872f CH |
1272 | out_free_bdi: |
1273 | bdi_put(disk->bdi); | |
f93af2a4 CH |
1274 | out_free_disk: |
1275 | kfree(disk); | |
61a35cfc CH |
1276 | out_put_queue: |
1277 | blk_put_queue(q); | |
f93af2a4 | 1278 | return NULL; |
1da177e4 | 1279 | } |
e319e1fb | 1280 | EXPORT_SYMBOL(__alloc_disk_node); |
1da177e4 | 1281 | |
4dcc4874 | 1282 | struct gendisk *__blk_alloc_disk(int node, struct lock_class_key *lkclass) |
f525464a CH |
1283 | { |
1284 | struct request_queue *q; | |
1285 | struct gendisk *disk; | |
1286 | ||
1287 | q = blk_alloc_queue(node); | |
1288 | if (!q) | |
1289 | return NULL; | |
1290 | ||
4a1fa41d | 1291 | disk = __alloc_disk_node(q, node, lkclass); |
f525464a CH |
1292 | if (!disk) { |
1293 | blk_cleanup_queue(q); | |
1294 | return NULL; | |
1295 | } | |
f525464a CH |
1296 | return disk; |
1297 | } | |
1298 | EXPORT_SYMBOL(__blk_alloc_disk); | |
1299 | ||
b5bd357c LC |
1300 | /** |
1301 | * put_disk - decrements the gendisk refcount | |
0d20dcc2 | 1302 | * @disk: the struct gendisk to decrement the refcount for |
b5bd357c LC |
1303 | * |
1304 | * This decrements the refcount for the struct gendisk. When this reaches 0 | |
1305 | * we'll have disk_release() called. | |
e8c7d14a LC |
1306 | * |
1307 | * Context: Any context, but the last reference must not be dropped from | |
1308 | * atomic context. | |
b5bd357c | 1309 | */ |
1da177e4 LT |
1310 | void put_disk(struct gendisk *disk) |
1311 | { | |
1312 | if (disk) | |
efdc41c8 | 1313 | put_device(disk_to_dev(disk)); |
1da177e4 | 1314 | } |
1da177e4 LT |
1315 | EXPORT_SYMBOL(put_disk); |
1316 | ||
f525464a CH |
1317 | /** |
1318 | * blk_cleanup_disk - shutdown a gendisk allocated by blk_alloc_disk | |
1319 | * @disk: gendisk to shutdown | |
1320 | * | |
1321 | * Mark the queue hanging off @disk DYING, drain all pending requests, then mark | |
1322 | * the queue DEAD, destroy and put it and the gendisk structure. | |
1323 | * | |
1324 | * Context: can sleep | |
1325 | */ | |
1326 | void blk_cleanup_disk(struct gendisk *disk) | |
1327 | { | |
1328 | blk_cleanup_queue(disk->queue); | |
1329 | put_disk(disk); | |
1330 | } | |
1331 | EXPORT_SYMBOL(blk_cleanup_disk); | |
1332 | ||
e3264a4d HR |
1333 | static void set_disk_ro_uevent(struct gendisk *gd, int ro) |
1334 | { | |
1335 | char event[] = "DISK_RO=1"; | |
1336 | char *envp[] = { event, NULL }; | |
1337 | ||
1338 | if (!ro) | |
1339 | event[8] = '0'; | |
1340 | kobject_uevent_env(&disk_to_dev(gd)->kobj, KOBJ_CHANGE, envp); | |
1341 | } | |
1342 | ||
52f019d4 CH |
1343 | /** |
1344 | * set_disk_ro - set a gendisk read-only | |
1345 | * @disk: gendisk to operate on | |
7f31bee3 | 1346 | * @read_only: %true to set the disk read-only, %false set the disk read/write |
52f019d4 CH |
1347 | * |
1348 | * This function is used to indicate whether a given disk device should have its | |
1349 | * read-only flag set. set_disk_ro() is typically used by device drivers to | |
1350 | * indicate whether the underlying physical device is write-protected. | |
1351 | */ | |
1352 | void set_disk_ro(struct gendisk *disk, bool read_only) | |
1da177e4 | 1353 | { |
52f019d4 CH |
1354 | if (read_only) { |
1355 | if (test_and_set_bit(GD_READ_ONLY, &disk->state)) | |
1356 | return; | |
1357 | } else { | |
1358 | if (!test_and_clear_bit(GD_READ_ONLY, &disk->state)) | |
1359 | return; | |
e3264a4d | 1360 | } |
52f019d4 | 1361 | set_disk_ro_uevent(disk, read_only); |
1da177e4 | 1362 | } |
1da177e4 LT |
1363 | EXPORT_SYMBOL(set_disk_ro); |
1364 | ||
1365 | int bdev_read_only(struct block_device *bdev) | |
1366 | { | |
947139bf | 1367 | return bdev->bd_read_only || get_disk_ro(bdev->bd_disk); |
1da177e4 | 1368 | } |
1da177e4 | 1369 | EXPORT_SYMBOL(bdev_read_only); |
cf179948 MC |
1370 | |
1371 | void inc_diskseq(struct gendisk *disk) | |
1372 | { | |
1373 | disk->diskseq = atomic64_inc_return(&diskseq); | |
1374 | } |