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 | 10 | #include <linux/fs.h> |
b446b60e | 11 | #include <linux/kdev_t.h> |
1da177e4 LT |
12 | #include <linux/kernel.h> |
13 | #include <linux/blkdev.h> | |
66114cad | 14 | #include <linux/backing-dev.h> |
1da177e4 LT |
15 | #include <linux/init.h> |
16 | #include <linux/spinlock.h> | |
f500975a | 17 | #include <linux/proc_fs.h> |
1da177e4 LT |
18 | #include <linux/seq_file.h> |
19 | #include <linux/slab.h> | |
20 | #include <linux/kmod.h> | |
b81e0c23 | 21 | #include <linux/major.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> |
82d981d4 | 27 | #include <linux/part_stat.h> |
dd7de370 | 28 | #include <linux/blktrace_api.h> |
1da177e4 | 29 | |
dd7de370 | 30 | #include "blk-throttle.h" |
ff88972c | 31 | #include "blk.h" |
2aa7745b | 32 | #include "blk-mq-sched.h" |
8e141f9e | 33 | #include "blk-rq-qos.h" |
1059699f | 34 | #include "blk-cgroup.h" |
ff88972c | 35 | |
31eb6186 | 36 | static struct kobject *block_depr; |
1da177e4 | 37 | |
cf179948 MC |
38 | /* |
39 | * Unique, monotonically increasing sequential number associated with block | |
40 | * devices instances (i.e. incremented each time a device is attached). | |
41 | * Associating uevents with block devices in userspace is difficult and racy: | |
42 | * the uevent netlink socket is lossy, and on slow and overloaded systems has | |
43 | * a very high latency. | |
44 | * Block devices do not have exclusive owners in userspace, any process can set | |
45 | * one up (e.g. loop devices). Moreover, device names can be reused (e.g. loop0 | |
46 | * can be reused again and again). | |
47 | * A userspace process setting up a block device and watching for its events | |
48 | * cannot thus reliably tell whether an event relates to the device it just set | |
49 | * up or another earlier instance with the same name. | |
50 | * This sequential number allows userspace processes to solve this problem, and | |
51 | * uniquely associate an uevent to the lifetime to a device. | |
52 | */ | |
53 | static atomic64_t diskseq; | |
54 | ||
bcce3de1 | 55 | /* for extended dynamic devt allocation, currently only one major is used */ |
ce23bba8 | 56 | #define NR_EXT_DEVT (1 << MINORBITS) |
22ae8ce8 | 57 | static DEFINE_IDA(ext_devt_ida); |
bcce3de1 | 58 | |
a782483c CH |
59 | void set_capacity(struct gendisk *disk, sector_t sectors) |
60 | { | |
83794367 | 61 | bdev_set_nr_sectors(disk->part0, sectors); |
a782483c CH |
62 | } |
63 | EXPORT_SYMBOL(set_capacity); | |
64 | ||
e598a72f | 65 | /* |
449f4ec9 CH |
66 | * Set disk capacity and notify if the size is not currently zero and will not |
67 | * be set to zero. Returns true if a uevent was sent, otherwise false. | |
e598a72f | 68 | */ |
449f4ec9 | 69 | bool set_capacity_and_notify(struct gendisk *disk, sector_t size) |
e598a72f BS |
70 | { |
71 | sector_t capacity = get_capacity(disk); | |
a782483c | 72 | char *envp[] = { "RESIZE=1", NULL }; |
e598a72f BS |
73 | |
74 | set_capacity(disk, size); | |
e598a72f | 75 | |
a782483c CH |
76 | /* |
77 | * Only print a message and send a uevent if the gendisk is user visible | |
78 | * and alive. This avoids spamming the log and udev when setting the | |
79 | * initial capacity during probing. | |
80 | */ | |
81 | if (size == capacity || | |
50b4aecf CH |
82 | !disk_live(disk) || |
83 | (disk->flags & GENHD_FL_HIDDEN)) | |
a782483c | 84 | return false; |
e598a72f | 85 | |
a782483c | 86 | pr_info("%s: detected capacity change from %lld to %lld\n", |
452c0bf8 | 87 | disk->disk_name, capacity, size); |
7e890c37 | 88 | |
a782483c CH |
89 | /* |
90 | * Historically we did not send a uevent for changes to/from an empty | |
91 | * device. | |
92 | */ | |
93 | if (!capacity || !size) | |
94 | return false; | |
95 | kobject_uevent_env(&disk_to_dev(disk)->kobj, KOBJ_CHANGE, envp); | |
96 | return true; | |
e598a72f | 97 | } |
449f4ec9 | 98 | EXPORT_SYMBOL_GPL(set_capacity_and_notify); |
e598a72f | 99 | |
0d02129e CH |
100 | static void part_stat_read_all(struct block_device *part, |
101 | struct disk_stats *stat) | |
ea18e0f0 KK |
102 | { |
103 | int cpu; | |
104 | ||
105 | memset(stat, 0, sizeof(struct disk_stats)); | |
106 | for_each_possible_cpu(cpu) { | |
0d02129e | 107 | struct disk_stats *ptr = per_cpu_ptr(part->bd_stats, cpu); |
ea18e0f0 KK |
108 | int group; |
109 | ||
110 | for (group = 0; group < NR_STAT_GROUPS; group++) { | |
111 | stat->nsecs[group] += ptr->nsecs[group]; | |
112 | stat->sectors[group] += ptr->sectors[group]; | |
113 | stat->ios[group] += ptr->ios[group]; | |
114 | stat->merges[group] += ptr->merges[group]; | |
115 | } | |
116 | ||
117 | stat->io_ticks += ptr->io_ticks; | |
ea18e0f0 KK |
118 | } |
119 | } | |
ea18e0f0 | 120 | |
8446fe92 | 121 | static unsigned int part_in_flight(struct block_device *part) |
f299b7c7 | 122 | { |
b2f609e1 | 123 | unsigned int inflight = 0; |
1226b8dd | 124 | int cpu; |
f299b7c7 | 125 | |
1226b8dd | 126 | for_each_possible_cpu(cpu) { |
e016b782 MP |
127 | inflight += part_stat_local_read_cpu(part, in_flight[0], cpu) + |
128 | part_stat_local_read_cpu(part, in_flight[1], cpu); | |
1226b8dd | 129 | } |
e016b782 MP |
130 | if ((int)inflight < 0) |
131 | inflight = 0; | |
1226b8dd | 132 | |
e016b782 | 133 | return inflight; |
f299b7c7 JA |
134 | } |
135 | ||
8446fe92 CH |
136 | static void part_in_flight_rw(struct block_device *part, |
137 | unsigned int inflight[2]) | |
bf0ddaba | 138 | { |
1226b8dd MP |
139 | int cpu; |
140 | ||
1226b8dd MP |
141 | inflight[0] = 0; |
142 | inflight[1] = 0; | |
143 | for_each_possible_cpu(cpu) { | |
144 | inflight[0] += part_stat_local_read_cpu(part, in_flight[0], cpu); | |
145 | inflight[1] += part_stat_local_read_cpu(part, in_flight[1], cpu); | |
146 | } | |
147 | if ((int)inflight[0] < 0) | |
148 | inflight[0] = 0; | |
149 | if ((int)inflight[1] < 0) | |
150 | inflight[1] = 0; | |
bf0ddaba OS |
151 | } |
152 | ||
1da177e4 LT |
153 | /* |
154 | * Can be deleted altogether. Later. | |
155 | * | |
156 | */ | |
133d55cd | 157 | #define BLKDEV_MAJOR_HASH_SIZE 255 |
1da177e4 LT |
158 | static struct blk_major_name { |
159 | struct blk_major_name *next; | |
160 | int major; | |
161 | char name[16]; | |
fbdee71b | 162 | #ifdef CONFIG_BLOCK_LEGACY_AUTOLOAD |
a160c615 | 163 | void (*probe)(dev_t devt); |
fbdee71b | 164 | #endif |
68eef3b4 | 165 | } *major_names[BLKDEV_MAJOR_HASH_SIZE]; |
e49fbbbf | 166 | static DEFINE_MUTEX(major_names_lock); |
dfbb3409 | 167 | static DEFINE_SPINLOCK(major_names_spinlock); |
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 | |
dfbb3409 | 180 | spin_lock(&major_names_spinlock); |
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); |
dfbb3409 | 184 | spin_unlock(&major_names_spinlock); |
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 |
26e06f5b LC |
194 | * @probe: pre-devtmpfs / pre-udev callback used to create disks when their |
195 | * pre-created device node is accessed. When a probe call uses | |
196 | * add_disk() and it fails the driver must cleanup resources. This | |
197 | * interface may soon be removed. | |
9e8c0bcc MN |
198 | * |
199 | * The @name must be unique within the system. | |
200 | * | |
0e056eb5 | 201 | * The return value depends on the @major input parameter: |
202 | * | |
f33ff110 SB |
203 | * - if a major device number was requested in range [1..BLKDEV_MAJOR_MAX-1] |
204 | * then the function returns zero on success, or a negative error code | |
0e056eb5 | 205 | * - if any unused major number was requested with @major = 0 parameter |
9e8c0bcc | 206 | * then the return value is the allocated major number in range |
f33ff110 SB |
207 | * [1..BLKDEV_MAJOR_MAX-1] or a negative error code otherwise |
208 | * | |
209 | * See Documentation/admin-guide/devices.txt for the list of allocated | |
210 | * major numbers. | |
e2b6b301 CH |
211 | * |
212 | * Use register_blkdev instead for any new code. | |
9e8c0bcc | 213 | */ |
a160c615 CH |
214 | int __register_blkdev(unsigned int major, const char *name, |
215 | void (*probe)(dev_t devt)) | |
1da177e4 LT |
216 | { |
217 | struct blk_major_name **n, *p; | |
218 | int index, ret = 0; | |
219 | ||
e49fbbbf | 220 | mutex_lock(&major_names_lock); |
1da177e4 LT |
221 | |
222 | /* temporary */ | |
223 | if (major == 0) { | |
224 | for (index = ARRAY_SIZE(major_names)-1; index > 0; index--) { | |
225 | if (major_names[index] == NULL) | |
226 | break; | |
227 | } | |
228 | ||
229 | if (index == 0) { | |
dfc76d11 KP |
230 | printk("%s: failed to get major for %s\n", |
231 | __func__, name); | |
1da177e4 LT |
232 | ret = -EBUSY; |
233 | goto out; | |
234 | } | |
235 | major = index; | |
236 | ret = major; | |
237 | } | |
238 | ||
133d55cd | 239 | if (major >= BLKDEV_MAJOR_MAX) { |
dfc76d11 KP |
240 | pr_err("%s: major requested (%u) is greater than the maximum (%u) for %s\n", |
241 | __func__, major, BLKDEV_MAJOR_MAX-1, name); | |
133d55cd LG |
242 | |
243 | ret = -EINVAL; | |
244 | goto out; | |
245 | } | |
246 | ||
1da177e4 LT |
247 | p = kmalloc(sizeof(struct blk_major_name), GFP_KERNEL); |
248 | if (p == NULL) { | |
249 | ret = -ENOMEM; | |
250 | goto out; | |
251 | } | |
252 | ||
253 | p->major = major; | |
fbdee71b | 254 | #ifdef CONFIG_BLOCK_LEGACY_AUTOLOAD |
a160c615 | 255 | p->probe = probe; |
fbdee71b | 256 | #endif |
20d09975 | 257 | strscpy(p->name, name, sizeof(p->name)); |
1da177e4 LT |
258 | p->next = NULL; |
259 | index = major_to_index(major); | |
260 | ||
dfbb3409 | 261 | spin_lock(&major_names_spinlock); |
1da177e4 LT |
262 | for (n = &major_names[index]; *n; n = &(*n)->next) { |
263 | if ((*n)->major == major) | |
264 | break; | |
265 | } | |
266 | if (!*n) | |
267 | *n = p; | |
268 | else | |
269 | ret = -EBUSY; | |
dfbb3409 | 270 | spin_unlock(&major_names_spinlock); |
1da177e4 LT |
271 | |
272 | if (ret < 0) { | |
f33ff110 | 273 | printk("register_blkdev: cannot get major %u for %s\n", |
1da177e4 LT |
274 | major, name); |
275 | kfree(p); | |
276 | } | |
277 | out: | |
e49fbbbf | 278 | mutex_unlock(&major_names_lock); |
1da177e4 LT |
279 | return ret; |
280 | } | |
a160c615 | 281 | EXPORT_SYMBOL(__register_blkdev); |
1da177e4 | 282 | |
f4480240 | 283 | void unregister_blkdev(unsigned int major, const char *name) |
1da177e4 LT |
284 | { |
285 | struct blk_major_name **n; | |
286 | struct blk_major_name *p = NULL; | |
287 | int index = major_to_index(major); | |
1da177e4 | 288 | |
e49fbbbf | 289 | mutex_lock(&major_names_lock); |
dfbb3409 | 290 | spin_lock(&major_names_spinlock); |
1da177e4 LT |
291 | for (n = &major_names[index]; *n; n = &(*n)->next) |
292 | if ((*n)->major == major) | |
293 | break; | |
294462a5 AM |
294 | if (!*n || strcmp((*n)->name, name)) { |
295 | WARN_ON(1); | |
294462a5 | 296 | } else { |
1da177e4 LT |
297 | p = *n; |
298 | *n = p->next; | |
299 | } | |
dfbb3409 | 300 | spin_unlock(&major_names_spinlock); |
e49fbbbf | 301 | mutex_unlock(&major_names_lock); |
1da177e4 | 302 | kfree(p); |
1da177e4 LT |
303 | } |
304 | ||
305 | EXPORT_SYMBOL(unregister_blkdev); | |
306 | ||
7c3f828b | 307 | int blk_alloc_ext_minor(void) |
bcce3de1 | 308 | { |
bab998d6 | 309 | int idx; |
bcce3de1 | 310 | |
d1868328 | 311 | idx = ida_alloc_range(&ext_devt_ida, 0, NR_EXT_DEVT - 1, GFP_KERNEL); |
c4b2b7d1 CH |
312 | if (idx == -ENOSPC) |
313 | return -EBUSY; | |
314 | return idx; | |
bcce3de1 TH |
315 | } |
316 | ||
7c3f828b | 317 | void blk_free_ext_minor(unsigned int minor) |
bcce3de1 | 318 | { |
c4b2b7d1 | 319 | ida_free(&ext_devt_ida, minor); |
6fcc44d1 YY |
320 | } |
321 | ||
bc359d03 CH |
322 | void disk_uevent(struct gendisk *disk, enum kobject_action action) |
323 | { | |
bc359d03 | 324 | struct block_device *part; |
3212135a | 325 | unsigned long idx; |
bc359d03 | 326 | |
3212135a CH |
327 | rcu_read_lock(); |
328 | xa_for_each(&disk->part_tbl, idx, part) { | |
329 | if (bdev_is_partition(part) && !bdev_nr_sectors(part)) | |
330 | continue; | |
498dcc13 | 331 | if (!kobject_get_unless_zero(&part->bd_device.kobj)) |
3212135a CH |
332 | continue; |
333 | ||
334 | rcu_read_unlock(); | |
bc359d03 | 335 | kobject_uevent(bdev_kobj(part), action); |
498dcc13 | 336 | put_device(&part->bd_device); |
3212135a CH |
337 | rcu_read_lock(); |
338 | } | |
339 | rcu_read_unlock(); | |
bc359d03 CH |
340 | } |
341 | EXPORT_SYMBOL_GPL(disk_uevent); | |
342 | ||
05bdb996 | 343 | int disk_scan_partitions(struct gendisk *disk, blk_mode_t mode) |
9301fe73 | 344 | { |
190f676a | 345 | struct file *file; |
e5cfefa9 | 346 | int ret = 0; |
9301fe73 | 347 | |
9f18db57 | 348 | if (disk->flags & (GENHD_FL_NO_PART | GENHD_FL_HIDDEN)) |
e16e506c | 349 | return -EINVAL; |
b9684a71 CH |
350 | if (test_bit(GD_SUPPRESS_PART_SCAN, &disk->state)) |
351 | return -EINVAL; | |
e16e506c CH |
352 | if (disk->open_partitions) |
353 | return -EBUSY; | |
9301fe73 | 354 | |
e5cfefa9 YK |
355 | /* |
356 | * If the device is opened exclusively by current thread already, it's | |
357 | * safe to scan partitons, otherwise, use bd_prepare_to_claim() to | |
358 | * synchronize with other exclusive openers and other partition | |
359 | * scanners. | |
360 | */ | |
05bdb996 | 361 | if (!(mode & BLK_OPEN_EXCL)) { |
0718afd4 CH |
362 | ret = bd_prepare_to_claim(disk->part0, disk_scan_partitions, |
363 | NULL); | |
e5cfefa9 YK |
364 | if (ret) |
365 | return ret; | |
366 | } | |
367 | ||
3723091e | 368 | set_bit(GD_NEED_PART_SCAN, &disk->state); |
190f676a CB |
369 | file = bdev_file_open_by_dev(disk_devt(disk), mode & ~BLK_OPEN_EXCL, |
370 | NULL, NULL); | |
371 | if (IS_ERR(file)) | |
372 | ret = PTR_ERR(file); | |
e5cfefa9 | 373 | else |
190f676a | 374 | fput(file); |
e5cfefa9 | 375 | |
3723091e YK |
376 | /* |
377 | * If blkdev_get_by_dev() failed early, GD_NEED_PART_SCAN is still set, | |
378 | * and this will cause that re-assemble partitioned raid device will | |
379 | * creat partition for underlying disk. | |
380 | */ | |
381 | clear_bit(GD_NEED_PART_SCAN, &disk->state); | |
05bdb996 | 382 | if (!(mode & BLK_OPEN_EXCL)) |
e5cfefa9 YK |
383 | bd_abort_claiming(disk->part0, disk_scan_partitions); |
384 | return ret; | |
9301fe73 CH |
385 | } |
386 | ||
1da177e4 | 387 | /** |
d1254a87 | 388 | * device_add_disk - add disk information to kernel list |
e63a46be | 389 | * @parent: parent device for the disk |
1da177e4 | 390 | * @disk: per-device partitioning information |
fef912bf | 391 | * @groups: Additional per-device sysfs groups |
1da177e4 LT |
392 | * |
393 | * This function registers the partitioning information in @disk | |
394 | * with the kernel. | |
395 | */ | |
278167fd LC |
396 | int __must_check device_add_disk(struct device *parent, struct gendisk *disk, |
397 | const struct attribute_group **groups) | |
d1254a87 | 398 | |
1da177e4 | 399 | { |
52b85909 | 400 | struct device *ddev = disk_to_dev(disk); |
7c3f828b | 401 | int ret; |
cf0ca9fe | 402 | |
69fe0f29 ML |
403 | /* Only makes sense for bio-based to set ->poll_bio */ |
404 | if (queue_is_mq(disk->queue) && disk->fops->poll_bio) | |
405 | return -EINVAL; | |
406 | ||
737eb78e DLM |
407 | /* |
408 | * The disk queue should now be all set with enough information about | |
409 | * the device for the elevator code to pick an adequate default | |
410 | * elevator if one is needed, that is, for devices requesting queue | |
411 | * registration. | |
412 | */ | |
d1254a87 | 413 | elevator_init_mq(disk->queue); |
737eb78e | 414 | |
9f4107b0 JA |
415 | /* Mark bdev as having a submit_bio, if needed */ |
416 | disk->part0->bd_has_submit_bio = disk->fops->submit_bio != NULL; | |
417 | ||
7c3f828b CH |
418 | /* |
419 | * If the driver provides an explicit major number it also must provide | |
420 | * the number of minors numbers supported, and those will be used to | |
421 | * setup the gendisk. | |
422 | * Otherwise just allocate the device numbers for both the whole device | |
423 | * and all partitions from the extended dev_t space. | |
3e1a7ff8 | 424 | */ |
02341a08 | 425 | ret = -EINVAL; |
7c3f828b | 426 | if (disk->major) { |
83cbce95 | 427 | if (WARN_ON(!disk->minors)) |
02341a08 | 428 | goto out_exit_elevator; |
2e3c73fa CH |
429 | |
430 | if (disk->minors > DISK_MAX_PARTS) { | |
431 | pr_err("block: can't allocate more than %d partitions\n", | |
432 | DISK_MAX_PARTS); | |
433 | disk->minors = DISK_MAX_PARTS; | |
434 | } | |
4c434392 LN |
435 | if (disk->first_minor > MINORMASK || |
436 | disk->minors > MINORMASK + 1 || | |
437 | disk->first_minor + disk->minors > MINORMASK + 1) | |
02341a08 | 438 | goto out_exit_elevator; |
7c3f828b | 439 | } else { |
83cbce95 | 440 | if (WARN_ON(disk->minors)) |
02341a08 | 441 | goto out_exit_elevator; |
3e1a7ff8 | 442 | |
7c3f828b | 443 | ret = blk_alloc_ext_minor(); |
83cbce95 | 444 | if (ret < 0) |
02341a08 | 445 | goto out_exit_elevator; |
7c3f828b | 446 | disk->major = BLOCK_EXT_MAJOR; |
539711d7 | 447 | disk->first_minor = ret; |
3e1a7ff8 | 448 | } |
7c3f828b | 449 | |
52b85909 CH |
450 | /* delay uevents, until we scanned partition table */ |
451 | dev_set_uevent_suppress(ddev, 1); | |
452 | ||
453 | ddev->parent = parent; | |
454 | ddev->groups = groups; | |
455 | dev_set_name(ddev, "%s", disk->disk_name); | |
8235b5c1 CH |
456 | if (!(disk->flags & GENHD_FL_HIDDEN)) |
457 | ddev->devt = MKDEV(disk->major, disk->first_minor); | |
83cbce95 LC |
458 | ret = device_add(ddev); |
459 | if (ret) | |
99d8690a CH |
460 | goto out_free_ext_minor; |
461 | ||
462 | ret = disk_alloc_events(disk); | |
463 | if (ret) | |
464 | goto out_device_del; | |
465 | ||
721da5ce GKH |
466 | ret = sysfs_create_link(block_depr, &ddev->kobj, |
467 | kobject_name(&ddev->kobj)); | |
468 | if (ret) | |
469 | goto out_device_del; | |
52b85909 CH |
470 | |
471 | /* | |
472 | * avoid probable deadlock caused by allocating memory with | |
473 | * GFP_KERNEL in runtime_resume callback of its all ancestor | |
474 | * devices | |
475 | */ | |
476 | pm_runtime_set_memalloc_noio(ddev, true); | |
477 | ||
478 | disk->part0->bd_holder_dir = | |
479 | kobject_create_and_add("holders", &ddev->kobj); | |
fe7d064f LC |
480 | if (!disk->part0->bd_holder_dir) { |
481 | ret = -ENOMEM; | |
ff53cd52 | 482 | goto out_del_block_link; |
fe7d064f | 483 | } |
52b85909 | 484 | disk->slave_dir = kobject_create_and_add("slaves", &ddev->kobj); |
fe7d064f LC |
485 | if (!disk->slave_dir) { |
486 | ret = -ENOMEM; | |
83cbce95 | 487 | goto out_put_holder_dir; |
fe7d064f | 488 | } |
52b85909 | 489 | |
83cbce95 LC |
490 | ret = blk_register_queue(disk); |
491 | if (ret) | |
492 | goto out_put_slave_dir; | |
75f4dca5 | 493 | |
9f18db57 | 494 | if (!(disk->flags & GENHD_FL_HIDDEN)) { |
8235b5c1 CH |
495 | ret = bdi_register(disk->bdi, "%u:%u", |
496 | disk->major, disk->first_minor); | |
83cbce95 LC |
497 | if (ret) |
498 | goto out_unregister_queue; | |
8235b5c1 | 499 | bdi_set_owner(disk->bdi, ddev); |
83cbce95 LC |
500 | ret = sysfs_create_link(&ddev->kobj, |
501 | &disk->bdi->dev->kobj, "bdi"); | |
502 | if (ret) | |
503 | goto out_unregister_bdi; | |
8235b5c1 | 504 | |
e5cfefa9 YK |
505 | /* Make sure the first partition scan will be proceed */ |
506 | if (get_capacity(disk) && !(disk->flags & GENHD_FL_NO_PART) && | |
507 | !test_bit(GD_SUPPRESS_PART_SCAN, &disk->state)) | |
508 | set_bit(GD_NEED_PART_SCAN, &disk->state); | |
509 | ||
9d5ee676 | 510 | bdev_add(disk->part0, ddev->devt); |
e16e506c | 511 | if (get_capacity(disk)) |
05bdb996 | 512 | disk_scan_partitions(disk, BLK_OPEN_READ); |
52b85909 CH |
513 | |
514 | /* | |
515 | * Announce the disk and partitions after all partitions are | |
8235b5c1 | 516 | * created. (for hidden disks uevents remain suppressed forever) |
52b85909 CH |
517 | */ |
518 | dev_set_uevent_suppress(ddev, 0); | |
519 | disk_uevent(disk, KOBJ_ADD); | |
a0a6314a CH |
520 | } else { |
521 | /* | |
522 | * Even if the block_device for a hidden gendisk is not | |
523 | * registered, it needs to have a valid bd_dev so that the | |
524 | * freeing of the dynamic major works. | |
525 | */ | |
526 | disk->part0->bd_dev = MKDEV(disk->major, disk->first_minor); | |
52b85909 CH |
527 | } |
528 | ||
75f4dca5 | 529 | disk_update_readahead(disk); |
77ea887e | 530 | disk_add_events(disk); |
76792055 | 531 | set_bit(GD_ADDED, &disk->state); |
83cbce95 LC |
532 | return 0; |
533 | ||
534 | out_unregister_bdi: | |
535 | if (!(disk->flags & GENHD_FL_HIDDEN)) | |
536 | bdi_unregister(disk->bdi); | |
537 | out_unregister_queue: | |
538 | blk_unregister_queue(disk); | |
fa81cbaf | 539 | rq_qos_exit(disk->queue); |
83cbce95 LC |
540 | out_put_slave_dir: |
541 | kobject_put(disk->slave_dir); | |
d90db3b1 | 542 | disk->slave_dir = NULL; |
83cbce95 LC |
543 | out_put_holder_dir: |
544 | kobject_put(disk->part0->bd_holder_dir); | |
83cbce95 | 545 | out_del_block_link: |
721da5ce | 546 | sysfs_remove_link(block_depr, dev_name(ddev)); |
5fa3d1a0 | 547 | pm_runtime_set_memalloc_noio(ddev, false); |
83cbce95 LC |
548 | out_device_del: |
549 | device_del(ddev); | |
83cbce95 LC |
550 | out_free_ext_minor: |
551 | if (disk->major == BLOCK_EXT_MAJOR) | |
552 | blk_free_ext_minor(disk->first_minor); | |
02341a08 YK |
553 | out_exit_elevator: |
554 | if (disk->queue->elevator) | |
555 | elevator_exit(disk->queue); | |
278167fd | 556 | return ret; |
1da177e4 | 557 | } |
e63a46be | 558 | EXPORT_SYMBOL(device_add_disk); |
1da177e4 | 559 | |
d8530de5 | 560 | static void blk_report_disk_dead(struct gendisk *disk, bool surprise) |
f55e017c CH |
561 | { |
562 | struct block_device *bdev; | |
563 | unsigned long idx; | |
564 | ||
f6103339 CB |
565 | /* |
566 | * On surprise disk removal, bdev_mark_dead() may call into file | |
567 | * systems below. Make it clear that we're expecting to not hold | |
568 | * disk->open_mutex. | |
569 | */ | |
570 | lockdep_assert_not_held(&disk->open_mutex); | |
571 | ||
f55e017c CH |
572 | rcu_read_lock(); |
573 | xa_for_each(&disk->part_tbl, idx, bdev) { | |
574 | if (!kobject_get_unless_zero(&bdev->bd_device.kobj)) | |
575 | continue; | |
576 | rcu_read_unlock(); | |
577 | ||
d8530de5 | 578 | bdev_mark_dead(bdev, surprise); |
f55e017c CH |
579 | |
580 | put_device(&bdev->bd_device); | |
581 | rcu_read_lock(); | |
582 | } | |
583 | rcu_read_unlock(); | |
584 | } | |
585 | ||
d8530de5 | 586 | static void __blk_mark_disk_dead(struct gendisk *disk) |
7a5428dc | 587 | { |
66fddc25 CH |
588 | /* |
589 | * Fail any new I/O. | |
590 | */ | |
a4f75764 CH |
591 | if (test_and_set_bit(GD_DEAD, &disk->state)) |
592 | return; | |
593 | ||
66fddc25 CH |
594 | if (test_bit(GD_OWNS_QUEUE, &disk->state)) |
595 | blk_queue_flag_set(QUEUE_FLAG_DYING, disk->queue); | |
71b26083 CH |
596 | |
597 | /* | |
598 | * Stop buffered writers from dirtying pages that can't be written out. | |
599 | */ | |
66fddc25 CH |
600 | set_capacity(disk, 0); |
601 | ||
602 | /* | |
603 | * Prevent new I/O from crossing bio_queue_enter(). | |
604 | */ | |
605 | blk_queue_start_drain(disk->queue); | |
d8530de5 | 606 | } |
f55e017c | 607 | |
d8530de5 CH |
608 | /** |
609 | * blk_mark_disk_dead - mark a disk as dead | |
610 | * @disk: disk to mark as dead | |
611 | * | |
612 | * Mark as disk as dead (e.g. surprise removed) and don't accept any new I/O | |
613 | * to this disk. | |
614 | */ | |
615 | void blk_mark_disk_dead(struct gendisk *disk) | |
616 | { | |
617 | __blk_mark_disk_dead(disk); | |
618 | blk_report_disk_dead(disk, true); | |
7a5428dc CH |
619 | } |
620 | EXPORT_SYMBOL_GPL(blk_mark_disk_dead); | |
621 | ||
b5bd357c LC |
622 | /** |
623 | * del_gendisk - remove the gendisk | |
624 | * @disk: the struct gendisk to remove | |
625 | * | |
626 | * Removes the gendisk and all its associated resources. This deletes the | |
627 | * partitions associated with the gendisk, and unregisters the associated | |
628 | * request_queue. | |
629 | * | |
630 | * This is the counter to the respective __device_add_disk() call. | |
631 | * | |
632 | * The final removal of the struct gendisk happens when its refcount reaches 0 | |
633 | * with put_disk(), which should be called after del_gendisk(), if | |
634 | * __device_add_disk() was used. | |
e8c7d14a LC |
635 | * |
636 | * Drivers exist which depend on the release of the gendisk to be synchronous, | |
637 | * it should not be deferred. | |
638 | * | |
639 | * Context: can sleep | |
b5bd357c | 640 | */ |
d2bf1b67 | 641 | void del_gendisk(struct gendisk *disk) |
1da177e4 | 642 | { |
8e141f9e | 643 | struct request_queue *q = disk->queue; |
eec1be4c CH |
644 | struct block_device *part; |
645 | unsigned long idx; | |
8e141f9e | 646 | |
e8c7d14a LC |
647 | might_sleep(); |
648 | ||
9f286992 | 649 | if (WARN_ON_ONCE(!disk_live(disk) && !(disk->flags & GENHD_FL_HIDDEN))) |
6b3ba976 CH |
650 | return; |
651 | ||
77ea887e TH |
652 | disk_del_events(disk); |
653 | ||
eec1be4c | 654 | /* |
d8530de5 | 655 | * Prevent new openers by unlinked the bdev inode. |
eec1be4c | 656 | */ |
a8698707 | 657 | mutex_lock(&disk->open_mutex); |
d8530de5 | 658 | xa_for_each(&disk->part_tbl, idx, part) |
eec1be4c | 659 | remove_inode_hash(part->bd_inode); |
a8698707 | 660 | mutex_unlock(&disk->open_mutex); |
c76f48eb | 661 | |
d8530de5 CH |
662 | /* |
663 | * Tell the file system to write back all dirty data and shut down if | |
664 | * it hasn't been notified earlier. | |
665 | */ | |
666 | if (!test_bit(GD_DEAD, &disk->state)) | |
667 | blk_report_disk_dead(disk, false); | |
668 | __blk_mark_disk_dead(disk); | |
8e141f9e | 669 | |
eec1be4c CH |
670 | /* |
671 | * Drop all partitions now that the disk is marked dead. | |
672 | */ | |
673 | mutex_lock(&disk->open_mutex); | |
674 | xa_for_each_start(&disk->part_tbl, idx, part, 1) | |
675 | drop_partition(part); | |
676 | mutex_unlock(&disk->open_mutex); | |
677 | ||
6b3ba976 | 678 | if (!(disk->flags & GENHD_FL_HIDDEN)) { |
8ddcd653 | 679 | sysfs_remove_link(&disk_to_dev(disk)->kobj, "bdi"); |
6b3ba976 | 680 | |
90f16fdd JK |
681 | /* |
682 | * Unregister bdi before releasing device numbers (as they can | |
683 | * get reused and we'd get clashes in sysfs). | |
684 | */ | |
edb0872f | 685 | bdi_unregister(disk->bdi); |
90f16fdd | 686 | } |
d2bf1b67 | 687 | |
6b3ba976 | 688 | blk_unregister_queue(disk); |
d2bf1b67 | 689 | |
cb8432d6 | 690 | kobject_put(disk->part0->bd_holder_dir); |
d2bf1b67 | 691 | kobject_put(disk->slave_dir); |
d90db3b1 | 692 | disk->slave_dir = NULL; |
d2bf1b67 | 693 | |
8446fe92 | 694 | part_stat_set_all(disk->part0, 0); |
cb8432d6 | 695 | disk->part0->bd_stamp = 0; |
721da5ce | 696 | sysfs_remove_link(block_depr, dev_name(disk_to_dev(disk))); |
25e823c8 | 697 | pm_runtime_set_memalloc_noio(disk_to_dev(disk), false); |
d2bf1b67 | 698 | device_del(disk_to_dev(disk)); |
d308ae0d | 699 | |
4c66a326 CH |
700 | blk_mq_freeze_queue_wait(q); |
701 | ||
cad9266a | 702 | blk_throtl_cancel_bios(disk); |
8f9e7b65 | 703 | |
d308ae0d ML |
704 | blk_sync_queue(q); |
705 | blk_flush_integrity(); | |
219cf43c JC |
706 | |
707 | if (queue_is_mq(q)) | |
708 | blk_mq_cancel_work_sync(q); | |
50e34d78 CH |
709 | |
710 | blk_mq_quiesce_queue(q); | |
711 | if (q->elevator) { | |
712 | mutex_lock(&q->sysfs_lock); | |
713 | elevator_exit(q); | |
714 | mutex_unlock(&q->sysfs_lock); | |
715 | } | |
716 | rq_qos_exit(q); | |
717 | blk_mq_unquiesce_queue(q); | |
718 | ||
d308ae0d | 719 | /* |
6f8191fd CH |
720 | * If the disk does not own the queue, allow using passthrough requests |
721 | * again. Else leave the queue frozen to fail all I/O. | |
d308ae0d | 722 | */ |
6f8191fd CH |
723 | if (!test_bit(GD_OWNS_QUEUE, &disk->state)) { |
724 | blk_queue_flag_clear(QUEUE_FLAG_INIT_DONE, q); | |
725 | __blk_mq_unfreeze_queue(q, true); | |
726 | } else { | |
727 | if (queue_is_mq(q)) | |
728 | blk_mq_exit_queue(q); | |
729 | } | |
1da177e4 | 730 | } |
d2bf1b67 | 731 | EXPORT_SYMBOL(del_gendisk); |
1da177e4 | 732 | |
f059a1d2 XY |
733 | /** |
734 | * invalidate_disk - invalidate the disk | |
735 | * @disk: the struct gendisk to invalidate | |
736 | * | |
737 | * A helper to invalidates the disk. It will clean the disk's associated | |
738 | * buffer/page caches and reset its internal states so that the disk | |
739 | * can be reused by the drivers. | |
740 | * | |
741 | * Context: can sleep | |
742 | */ | |
743 | void invalidate_disk(struct gendisk *disk) | |
744 | { | |
745 | struct block_device *bdev = disk->part0; | |
746 | ||
747 | invalidate_bdev(bdev); | |
748 | bdev->bd_inode->i_mapping->wb_err = 0; | |
749 | set_capacity(disk, 0); | |
750 | } | |
751 | EXPORT_SYMBOL(invalidate_disk); | |
752 | ||
99e6608c VV |
753 | /* sysfs access to bad-blocks list. */ |
754 | static ssize_t disk_badblocks_show(struct device *dev, | |
755 | struct device_attribute *attr, | |
756 | char *page) | |
757 | { | |
758 | struct gendisk *disk = dev_to_disk(dev); | |
759 | ||
760 | if (!disk->bb) | |
761 | return sprintf(page, "\n"); | |
762 | ||
763 | return badblocks_show(disk->bb, page, 0); | |
764 | } | |
765 | ||
766 | static ssize_t disk_badblocks_store(struct device *dev, | |
767 | struct device_attribute *attr, | |
768 | const char *page, size_t len) | |
769 | { | |
770 | struct gendisk *disk = dev_to_disk(dev); | |
771 | ||
772 | if (!disk->bb) | |
773 | return -ENXIO; | |
774 | ||
775 | return badblocks_store(disk->bb, page, len, 0); | |
776 | } | |
777 | ||
fbdee71b | 778 | #ifdef CONFIG_BLOCK_LEGACY_AUTOLOAD |
22ae8ce8 | 779 | void blk_request_module(dev_t devt) |
bd8eff3b | 780 | { |
a160c615 CH |
781 | unsigned int major = MAJOR(devt); |
782 | struct blk_major_name **n; | |
783 | ||
784 | mutex_lock(&major_names_lock); | |
785 | for (n = &major_names[major_to_index(major)]; *n; n = &(*n)->next) { | |
786 | if ((*n)->major == major && (*n)->probe) { | |
787 | (*n)->probe(devt); | |
788 | mutex_unlock(&major_names_lock); | |
789 | return; | |
790 | } | |
791 | } | |
792 | mutex_unlock(&major_names_lock); | |
793 | ||
bd8eff3b CH |
794 | if (request_module("block-major-%d-%d", MAJOR(devt), MINOR(devt)) > 0) |
795 | /* Make old-style 2.4 aliases work */ | |
796 | request_module("block-major-%d", MAJOR(devt)); | |
797 | } | |
fbdee71b | 798 | #endif /* CONFIG_BLOCK_LEGACY_AUTOLOAD */ |
bd8eff3b | 799 | |
1da177e4 LT |
800 | #ifdef CONFIG_PROC_FS |
801 | /* iterator */ | |
def4e38d | 802 | static void *disk_seqf_start(struct seq_file *seqf, loff_t *pos) |
68c4d4a7 | 803 | { |
def4e38d TH |
804 | loff_t skip = *pos; |
805 | struct class_dev_iter *iter; | |
806 | struct device *dev; | |
68c4d4a7 | 807 | |
aeb3d3a8 | 808 | iter = kmalloc(sizeof(*iter), GFP_KERNEL); |
def4e38d TH |
809 | if (!iter) |
810 | return ERR_PTR(-ENOMEM); | |
811 | ||
812 | seqf->private = iter; | |
813 | class_dev_iter_init(iter, &block_class, NULL, &disk_type); | |
814 | do { | |
815 | dev = class_dev_iter_next(iter); | |
816 | if (!dev) | |
817 | return NULL; | |
818 | } while (skip--); | |
819 | ||
820 | return dev_to_disk(dev); | |
68c4d4a7 GKH |
821 | } |
822 | ||
def4e38d | 823 | static void *disk_seqf_next(struct seq_file *seqf, void *v, loff_t *pos) |
1da177e4 | 824 | { |
edfaa7c3 | 825 | struct device *dev; |
1da177e4 | 826 | |
def4e38d TH |
827 | (*pos)++; |
828 | dev = class_dev_iter_next(seqf->private); | |
2ac3cee5 | 829 | if (dev) |
68c4d4a7 | 830 | return dev_to_disk(dev); |
2ac3cee5 | 831 | |
1da177e4 LT |
832 | return NULL; |
833 | } | |
834 | ||
def4e38d | 835 | static void disk_seqf_stop(struct seq_file *seqf, void *v) |
27f30251 | 836 | { |
def4e38d | 837 | struct class_dev_iter *iter = seqf->private; |
27f30251 | 838 | |
def4e38d TH |
839 | /* stop is called even after start failed :-( */ |
840 | if (iter) { | |
841 | class_dev_iter_exit(iter); | |
842 | kfree(iter); | |
77da1605 | 843 | seqf->private = NULL; |
5c0ef6d0 | 844 | } |
1da177e4 LT |
845 | } |
846 | ||
def4e38d | 847 | static void *show_partition_start(struct seq_file *seqf, loff_t *pos) |
1da177e4 | 848 | { |
06768067 | 849 | void *p; |
def4e38d TH |
850 | |
851 | p = disk_seqf_start(seqf, pos); | |
b9f985b6 | 852 | if (!IS_ERR_OR_NULL(p) && !*pos) |
def4e38d TH |
853 | seq_puts(seqf, "major minor #blocks name\n\n"); |
854 | return p; | |
1da177e4 LT |
855 | } |
856 | ||
cf771cb5 | 857 | static int show_partition(struct seq_file *seqf, void *v) |
1da177e4 LT |
858 | { |
859 | struct gendisk *sgp = v; | |
ad1eaa53 | 860 | struct block_device *part; |
ecc75a98 | 861 | unsigned long idx; |
1da177e4 | 862 | |
3b5149ac | 863 | if (!get_capacity(sgp) || (sgp->flags & GENHD_FL_HIDDEN)) |
1da177e4 LT |
864 | return 0; |
865 | ||
ecc75a98 CH |
866 | rcu_read_lock(); |
867 | xa_for_each(&sgp->part_tbl, idx, part) { | |
868 | if (!bdev_nr_sectors(part)) | |
869 | continue; | |
a291bb43 | 870 | seq_printf(seqf, "%4d %7d %10llu %pg\n", |
ad1eaa53 | 871 | MAJOR(part->bd_dev), MINOR(part->bd_dev), |
a291bb43 | 872 | bdev_nr_sectors(part) >> 1, part); |
ecc75a98 CH |
873 | } |
874 | rcu_read_unlock(); | |
1da177e4 LT |
875 | return 0; |
876 | } | |
877 | ||
f500975a | 878 | static const struct seq_operations partitions_op = { |
def4e38d TH |
879 | .start = show_partition_start, |
880 | .next = disk_seqf_next, | |
881 | .stop = disk_seqf_stop, | |
edfaa7c3 | 882 | .show = show_partition |
1da177e4 LT |
883 | }; |
884 | #endif | |
885 | ||
1da177e4 LT |
886 | static int __init genhd_device_init(void) |
887 | { | |
e105b8bf DW |
888 | int error; |
889 | ||
e105b8bf | 890 | error = class_register(&block_class); |
ee27a558 RM |
891 | if (unlikely(error)) |
892 | return error; | |
1da177e4 | 893 | blk_dev_init(); |
edfaa7c3 | 894 | |
561ec68e ZY |
895 | register_blkdev(BLOCK_EXT_MAJOR, "blkext"); |
896 | ||
edfaa7c3 | 897 | /* create top-level block dir */ |
721da5ce | 898 | block_depr = kobject_create_and_add("block", NULL); |
830d3cfb | 899 | return 0; |
1da177e4 LT |
900 | } |
901 | ||
902 | subsys_initcall(genhd_device_init); | |
903 | ||
edfaa7c3 KS |
904 | static ssize_t disk_range_show(struct device *dev, |
905 | struct device_attribute *attr, char *buf) | |
1da177e4 | 906 | { |
edfaa7c3 | 907 | struct gendisk *disk = dev_to_disk(dev); |
1da177e4 | 908 | |
edfaa7c3 | 909 | return sprintf(buf, "%d\n", disk->minors); |
1da177e4 LT |
910 | } |
911 | ||
1f014290 TH |
912 | static ssize_t disk_ext_range_show(struct device *dev, |
913 | struct device_attribute *attr, char *buf) | |
914 | { | |
915 | struct gendisk *disk = dev_to_disk(dev); | |
916 | ||
1ebe2e5f CH |
917 | return sprintf(buf, "%d\n", |
918 | (disk->flags & GENHD_FL_NO_PART) ? 1 : DISK_MAX_PARTS); | |
1f014290 TH |
919 | } |
920 | ||
edfaa7c3 KS |
921 | static ssize_t disk_removable_show(struct device *dev, |
922 | struct device_attribute *attr, char *buf) | |
a7fd6706 | 923 | { |
edfaa7c3 | 924 | struct gendisk *disk = dev_to_disk(dev); |
a7fd6706 | 925 | |
edfaa7c3 KS |
926 | return sprintf(buf, "%d\n", |
927 | (disk->flags & GENHD_FL_REMOVABLE ? 1 : 0)); | |
a7fd6706 KS |
928 | } |
929 | ||
8ddcd653 CH |
930 | static ssize_t disk_hidden_show(struct device *dev, |
931 | struct device_attribute *attr, char *buf) | |
932 | { | |
933 | struct gendisk *disk = dev_to_disk(dev); | |
934 | ||
935 | return sprintf(buf, "%d\n", | |
936 | (disk->flags & GENHD_FL_HIDDEN ? 1 : 0)); | |
937 | } | |
938 | ||
1c9ce527 KS |
939 | static ssize_t disk_ro_show(struct device *dev, |
940 | struct device_attribute *attr, char *buf) | |
941 | { | |
942 | struct gendisk *disk = dev_to_disk(dev); | |
943 | ||
b7db9956 | 944 | return sprintf(buf, "%d\n", get_disk_ro(disk) ? 1 : 0); |
1c9ce527 KS |
945 | } |
946 | ||
3ad5cee5 CH |
947 | ssize_t part_size_show(struct device *dev, |
948 | struct device_attribute *attr, char *buf) | |
949 | { | |
0d02129e | 950 | return sprintf(buf, "%llu\n", bdev_nr_sectors(dev_to_bdev(dev))); |
3ad5cee5 CH |
951 | } |
952 | ||
953 | ssize_t part_stat_show(struct device *dev, | |
954 | struct device_attribute *attr, char *buf) | |
955 | { | |
0d02129e | 956 | struct block_device *bdev = dev_to_bdev(dev); |
ed6cddef | 957 | struct request_queue *q = bdev_get_queue(bdev); |
ea18e0f0 | 958 | struct disk_stats stat; |
3ad5cee5 CH |
959 | unsigned int inflight; |
960 | ||
b2f609e1 | 961 | if (queue_is_mq(q)) |
0d02129e | 962 | inflight = blk_mq_in_flight(q, bdev); |
b2f609e1 | 963 | else |
0d02129e | 964 | inflight = part_in_flight(bdev); |
ea18e0f0 | 965 | |
86d73312 ZW |
966 | if (inflight) { |
967 | part_stat_lock(); | |
968 | update_io_ticks(bdev, jiffies, true); | |
969 | part_stat_unlock(); | |
970 | } | |
971 | part_stat_read_all(bdev, &stat); | |
3ad5cee5 CH |
972 | return sprintf(buf, |
973 | "%8lu %8lu %8llu %8u " | |
974 | "%8lu %8lu %8llu %8u " | |
975 | "%8u %8u %8u " | |
976 | "%8lu %8lu %8llu %8u " | |
977 | "%8lu %8u" | |
978 | "\n", | |
ea18e0f0 KK |
979 | stat.ios[STAT_READ], |
980 | stat.merges[STAT_READ], | |
981 | (unsigned long long)stat.sectors[STAT_READ], | |
982 | (unsigned int)div_u64(stat.nsecs[STAT_READ], NSEC_PER_MSEC), | |
983 | stat.ios[STAT_WRITE], | |
984 | stat.merges[STAT_WRITE], | |
985 | (unsigned long long)stat.sectors[STAT_WRITE], | |
986 | (unsigned int)div_u64(stat.nsecs[STAT_WRITE], NSEC_PER_MSEC), | |
3ad5cee5 | 987 | inflight, |
ea18e0f0 | 988 | jiffies_to_msecs(stat.io_ticks), |
8cd5b8fc KK |
989 | (unsigned int)div_u64(stat.nsecs[STAT_READ] + |
990 | stat.nsecs[STAT_WRITE] + | |
991 | stat.nsecs[STAT_DISCARD] + | |
992 | stat.nsecs[STAT_FLUSH], | |
993 | NSEC_PER_MSEC), | |
ea18e0f0 KK |
994 | stat.ios[STAT_DISCARD], |
995 | stat.merges[STAT_DISCARD], | |
996 | (unsigned long long)stat.sectors[STAT_DISCARD], | |
997 | (unsigned int)div_u64(stat.nsecs[STAT_DISCARD], NSEC_PER_MSEC), | |
998 | stat.ios[STAT_FLUSH], | |
999 | (unsigned int)div_u64(stat.nsecs[STAT_FLUSH], NSEC_PER_MSEC)); | |
3ad5cee5 CH |
1000 | } |
1001 | ||
1002 | ssize_t part_inflight_show(struct device *dev, struct device_attribute *attr, | |
1003 | char *buf) | |
1004 | { | |
0d02129e | 1005 | struct block_device *bdev = dev_to_bdev(dev); |
ed6cddef | 1006 | struct request_queue *q = bdev_get_queue(bdev); |
3ad5cee5 CH |
1007 | unsigned int inflight[2]; |
1008 | ||
b2f609e1 | 1009 | if (queue_is_mq(q)) |
0d02129e | 1010 | blk_mq_in_flight_rw(q, bdev, inflight); |
b2f609e1 | 1011 | else |
0d02129e | 1012 | part_in_flight_rw(bdev, inflight); |
b2f609e1 | 1013 | |
3ad5cee5 CH |
1014 | return sprintf(buf, "%8u %8u\n", inflight[0], inflight[1]); |
1015 | } | |
1016 | ||
edfaa7c3 KS |
1017 | static ssize_t disk_capability_show(struct device *dev, |
1018 | struct device_attribute *attr, char *buf) | |
86ce18d7 | 1019 | { |
e81cd5a9 CH |
1020 | dev_warn_once(dev, "the capability attribute has been deprecated.\n"); |
1021 | return sprintf(buf, "0\n"); | |
86ce18d7 | 1022 | } |
edfaa7c3 | 1023 | |
c72758f3 MP |
1024 | static ssize_t disk_alignment_offset_show(struct device *dev, |
1025 | struct device_attribute *attr, | |
1026 | char *buf) | |
1027 | { | |
1028 | struct gendisk *disk = dev_to_disk(dev); | |
1029 | ||
640f2a23 | 1030 | return sprintf(buf, "%d\n", bdev_alignment_offset(disk->part0)); |
c72758f3 MP |
1031 | } |
1032 | ||
86b37281 MP |
1033 | static ssize_t disk_discard_alignment_show(struct device *dev, |
1034 | struct device_attribute *attr, | |
1035 | char *buf) | |
1036 | { | |
1037 | struct gendisk *disk = dev_to_disk(dev); | |
1038 | ||
4e1462ff | 1039 | return sprintf(buf, "%d\n", bdev_alignment_offset(disk->part0)); |
86b37281 MP |
1040 | } |
1041 | ||
13927b31 MC |
1042 | static ssize_t diskseq_show(struct device *dev, |
1043 | struct device_attribute *attr, char *buf) | |
1044 | { | |
1045 | struct gendisk *disk = dev_to_disk(dev); | |
1046 | ||
1047 | return sprintf(buf, "%llu\n", disk->diskseq); | |
1048 | } | |
1049 | ||
5657a819 JP |
1050 | static DEVICE_ATTR(range, 0444, disk_range_show, NULL); |
1051 | static DEVICE_ATTR(ext_range, 0444, disk_ext_range_show, NULL); | |
1052 | static DEVICE_ATTR(removable, 0444, disk_removable_show, NULL); | |
1053 | static DEVICE_ATTR(hidden, 0444, disk_hidden_show, NULL); | |
1054 | static DEVICE_ATTR(ro, 0444, disk_ro_show, NULL); | |
1055 | static DEVICE_ATTR(size, 0444, part_size_show, NULL); | |
1056 | static DEVICE_ATTR(alignment_offset, 0444, disk_alignment_offset_show, NULL); | |
1057 | static DEVICE_ATTR(discard_alignment, 0444, disk_discard_alignment_show, NULL); | |
1058 | static DEVICE_ATTR(capability, 0444, disk_capability_show, NULL); | |
1059 | static DEVICE_ATTR(stat, 0444, part_stat_show, NULL); | |
1060 | static DEVICE_ATTR(inflight, 0444, part_inflight_show, NULL); | |
1061 | static DEVICE_ATTR(badblocks, 0644, disk_badblocks_show, disk_badblocks_store); | |
13927b31 | 1062 | static DEVICE_ATTR(diskseq, 0444, diskseq_show, NULL); |
3ad5cee5 | 1063 | |
c17bb495 | 1064 | #ifdef CONFIG_FAIL_MAKE_REQUEST |
3ad5cee5 CH |
1065 | ssize_t part_fail_show(struct device *dev, |
1066 | struct device_attribute *attr, char *buf) | |
1067 | { | |
0d02129e | 1068 | return sprintf(buf, "%d\n", dev_to_bdev(dev)->bd_make_it_fail); |
3ad5cee5 CH |
1069 | } |
1070 | ||
1071 | ssize_t part_fail_store(struct device *dev, | |
1072 | struct device_attribute *attr, | |
1073 | const char *buf, size_t count) | |
1074 | { | |
3ad5cee5 CH |
1075 | int i; |
1076 | ||
1077 | if (count > 0 && sscanf(buf, "%d", &i) > 0) | |
0d02129e | 1078 | dev_to_bdev(dev)->bd_make_it_fail = i; |
3ad5cee5 CH |
1079 | |
1080 | return count; | |
1081 | } | |
1082 | ||
edfaa7c3 | 1083 | static struct device_attribute dev_attr_fail = |
5657a819 | 1084 | __ATTR(make-it-fail, 0644, part_fail_show, part_fail_store); |
3ad5cee5 CH |
1085 | #endif /* CONFIG_FAIL_MAKE_REQUEST */ |
1086 | ||
581d4e28 JA |
1087 | #ifdef CONFIG_FAIL_IO_TIMEOUT |
1088 | static struct device_attribute dev_attr_fail_timeout = | |
5657a819 | 1089 | __ATTR(io-timeout-fail, 0644, part_timeout_show, part_timeout_store); |
581d4e28 | 1090 | #endif |
edfaa7c3 KS |
1091 | |
1092 | static struct attribute *disk_attrs[] = { | |
1093 | &dev_attr_range.attr, | |
1f014290 | 1094 | &dev_attr_ext_range.attr, |
edfaa7c3 | 1095 | &dev_attr_removable.attr, |
8ddcd653 | 1096 | &dev_attr_hidden.attr, |
1c9ce527 | 1097 | &dev_attr_ro.attr, |
edfaa7c3 | 1098 | &dev_attr_size.attr, |
c72758f3 | 1099 | &dev_attr_alignment_offset.attr, |
86b37281 | 1100 | &dev_attr_discard_alignment.attr, |
edfaa7c3 KS |
1101 | &dev_attr_capability.attr, |
1102 | &dev_attr_stat.attr, | |
316d315b | 1103 | &dev_attr_inflight.attr, |
99e6608c | 1104 | &dev_attr_badblocks.attr, |
2bc8cda5 CH |
1105 | &dev_attr_events.attr, |
1106 | &dev_attr_events_async.attr, | |
1107 | &dev_attr_events_poll_msecs.attr, | |
13927b31 | 1108 | &dev_attr_diskseq.attr, |
edfaa7c3 KS |
1109 | #ifdef CONFIG_FAIL_MAKE_REQUEST |
1110 | &dev_attr_fail.attr, | |
581d4e28 JA |
1111 | #endif |
1112 | #ifdef CONFIG_FAIL_IO_TIMEOUT | |
1113 | &dev_attr_fail_timeout.attr, | |
edfaa7c3 KS |
1114 | #endif |
1115 | NULL | |
1116 | }; | |
1117 | ||
9438b3e0 DW |
1118 | static umode_t disk_visible(struct kobject *kobj, struct attribute *a, int n) |
1119 | { | |
1120 | struct device *dev = container_of(kobj, typeof(*dev), kobj); | |
1121 | struct gendisk *disk = dev_to_disk(dev); | |
1122 | ||
1123 | if (a == &dev_attr_badblocks.attr && !disk->bb) | |
1124 | return 0; | |
1125 | return a->mode; | |
1126 | } | |
1127 | ||
edfaa7c3 KS |
1128 | static struct attribute_group disk_attr_group = { |
1129 | .attrs = disk_attrs, | |
9438b3e0 | 1130 | .is_visible = disk_visible, |
edfaa7c3 KS |
1131 | }; |
1132 | ||
a4dbd674 | 1133 | static const struct attribute_group *disk_attr_groups[] = { |
edfaa7c3 | 1134 | &disk_attr_group, |
cc5c516d CH |
1135 | #ifdef CONFIG_BLK_DEV_IO_TRACE |
1136 | &blk_trace_attr_group, | |
ff53cd52 TW |
1137 | #endif |
1138 | #ifdef CONFIG_BLK_DEV_INTEGRITY | |
1139 | &blk_integrity_attr_group, | |
cc5c516d | 1140 | #endif |
edfaa7c3 | 1141 | NULL |
1da177e4 LT |
1142 | }; |
1143 | ||
b5bd357c LC |
1144 | /** |
1145 | * disk_release - releases all allocated resources of the gendisk | |
1146 | * @dev: the device representing this disk | |
1147 | * | |
1148 | * This function releases all allocated resources of the gendisk. | |
1149 | * | |
b5bd357c LC |
1150 | * Drivers which used __device_add_disk() have a gendisk with a request_queue |
1151 | * assigned. Since the request_queue sits on top of the gendisk for these | |
1152 | * drivers we also call blk_put_queue() for them, and we expect the | |
1153 | * request_queue refcount to reach 0 at this point, and so the request_queue | |
1154 | * will also be freed prior to the disk. | |
e8c7d14a LC |
1155 | * |
1156 | * Context: can sleep | |
b5bd357c | 1157 | */ |
edfaa7c3 | 1158 | static void disk_release(struct device *dev) |
1da177e4 | 1159 | { |
edfaa7c3 KS |
1160 | struct gendisk *disk = dev_to_disk(dev); |
1161 | ||
e8c7d14a | 1162 | might_sleep(); |
a2041761 | 1163 | WARN_ON_ONCE(disk_live(disk)); |
e8c7d14a | 1164 | |
dd7de370 YK |
1165 | blk_trace_remove(disk->queue); |
1166 | ||
c5db2cfc CH |
1167 | /* |
1168 | * To undo the all initialization from blk_mq_init_allocated_queue in | |
1169 | * case of a probe failure where add_disk is never called we have to | |
1170 | * call blk_mq_exit_queue here. We can't do this for the more common | |
1171 | * teardown case (yet) as the tagset can be gone by the time the disk | |
1172 | * is released once it was added. | |
1173 | */ | |
1174 | if (queue_is_mq(disk->queue) && | |
1175 | test_bit(GD_OWNS_QUEUE, &disk->state) && | |
1176 | !test_bit(GD_ADDED, &disk->state)) | |
1177 | blk_mq_exit_queue(disk->queue); | |
1178 | ||
b6553bef CH |
1179 | blkcg_exit_disk(disk); |
1180 | ||
46754bd0 | 1181 | bioset_exit(&disk->bio_split); |
2a19b28f | 1182 | |
77ea887e | 1183 | disk_release_events(disk); |
1da177e4 | 1184 | kfree(disk->random); |
5d400665 | 1185 | disk_free_zone_bitmaps(disk); |
a33df75c | 1186 | xa_destroy(&disk->part_tbl); |
1059699f | 1187 | |
d152c682 | 1188 | disk->queue->disk = NULL; |
61a35cfc | 1189 | blk_put_queue(disk->queue); |
76792055 CH |
1190 | |
1191 | if (test_bit(GD_ADDED, &disk->state) && disk->fops->free_disk) | |
1192 | disk->fops->free_disk(disk); | |
1193 | ||
2f4731dc | 1194 | iput(disk->part0->bd_inode); /* frees the disk */ |
1da177e4 | 1195 | } |
87eb7107 | 1196 | |
23680f0b | 1197 | static int block_uevent(const struct device *dev, struct kobj_uevent_env *env) |
87eb7107 | 1198 | { |
23680f0b | 1199 | const struct gendisk *disk = dev_to_disk(dev); |
87eb7107 MC |
1200 | |
1201 | return add_uevent_var(env, "DISKSEQ=%llu", disk->diskseq); | |
1202 | } | |
1203 | ||
f8c7511d | 1204 | const struct class block_class = { |
edfaa7c3 | 1205 | .name = "block", |
87eb7107 | 1206 | .dev_uevent = block_uevent, |
1da177e4 LT |
1207 | }; |
1208 | ||
a9b12f8b | 1209 | static char *block_devnode(const struct device *dev, umode_t *mode, |
050a4f34 JA |
1210 | kuid_t *uid, kgid_t *gid) |
1211 | { | |
1212 | struct gendisk *disk = dev_to_disk(dev); | |
1213 | ||
1214 | if (disk->fops->devnode) | |
1215 | return disk->fops->devnode(disk, mode); | |
1216 | return NULL; | |
1217 | } | |
1218 | ||
ef45fe47 | 1219 | const struct device_type disk_type = { |
edfaa7c3 KS |
1220 | .name = "disk", |
1221 | .groups = disk_attr_groups, | |
1222 | .release = disk_release, | |
050a4f34 | 1223 | .devnode = block_devnode, |
1da177e4 LT |
1224 | }; |
1225 | ||
a6e2ba88 | 1226 | #ifdef CONFIG_PROC_FS |
cf771cb5 TH |
1227 | /* |
1228 | * aggregate disk stat collector. Uses the same stats that the sysfs | |
1229 | * entries do, above, but makes them available through one seq_file. | |
1230 | * | |
1231 | * The output looks suspiciously like /proc/partitions with a bunch of | |
1232 | * extra fields. | |
1233 | */ | |
1234 | static int diskstats_show(struct seq_file *seqf, void *v) | |
1da177e4 LT |
1235 | { |
1236 | struct gendisk *gp = v; | |
ad1eaa53 | 1237 | struct block_device *hd; |
e016b782 | 1238 | unsigned int inflight; |
ea18e0f0 | 1239 | struct disk_stats stat; |
7fae67cc | 1240 | unsigned long idx; |
1da177e4 LT |
1241 | |
1242 | /* | |
ed9e1982 | 1243 | if (&disk_to_dev(gp)->kobj.entry == block_class.devices.next) |
cf771cb5 | 1244 | seq_puts(seqf, "major minor name" |
1da177e4 LT |
1245 | " rio rmerge rsect ruse wio wmerge " |
1246 | "wsect wuse running use aveq" | |
1247 | "\n\n"); | |
1248 | */ | |
9f5e4865 | 1249 | |
7fae67cc CH |
1250 | rcu_read_lock(); |
1251 | xa_for_each(&gp->part_tbl, idx, hd) { | |
1252 | if (bdev_is_partition(hd) && !bdev_nr_sectors(hd)) | |
1253 | continue; | |
b2f609e1 | 1254 | if (queue_is_mq(gp->queue)) |
ad1eaa53 | 1255 | inflight = blk_mq_in_flight(gp->queue, hd); |
b2f609e1 | 1256 | else |
ad1eaa53 | 1257 | inflight = part_in_flight(hd); |
ea18e0f0 | 1258 | |
86d73312 ZW |
1259 | if (inflight) { |
1260 | part_stat_lock(); | |
1261 | update_io_ticks(hd, jiffies, true); | |
1262 | part_stat_unlock(); | |
1263 | } | |
1264 | part_stat_read_all(hd, &stat); | |
26e2d7a3 | 1265 | seq_printf(seqf, "%4d %7d %pg " |
bdca3c87 MC |
1266 | "%lu %lu %lu %u " |
1267 | "%lu %lu %lu %u " | |
1268 | "%u %u %u " | |
b6866318 KK |
1269 | "%lu %lu %lu %u " |
1270 | "%lu %u" | |
1271 | "\n", | |
26e2d7a3 | 1272 | MAJOR(hd->bd_dev), MINOR(hd->bd_dev), hd, |
ea18e0f0 KK |
1273 | stat.ios[STAT_READ], |
1274 | stat.merges[STAT_READ], | |
1275 | stat.sectors[STAT_READ], | |
1276 | (unsigned int)div_u64(stat.nsecs[STAT_READ], | |
1277 | NSEC_PER_MSEC), | |
1278 | stat.ios[STAT_WRITE], | |
1279 | stat.merges[STAT_WRITE], | |
1280 | stat.sectors[STAT_WRITE], | |
1281 | (unsigned int)div_u64(stat.nsecs[STAT_WRITE], | |
1282 | NSEC_PER_MSEC), | |
e016b782 | 1283 | inflight, |
ea18e0f0 | 1284 | jiffies_to_msecs(stat.io_ticks), |
8cd5b8fc KK |
1285 | (unsigned int)div_u64(stat.nsecs[STAT_READ] + |
1286 | stat.nsecs[STAT_WRITE] + | |
1287 | stat.nsecs[STAT_DISCARD] + | |
1288 | stat.nsecs[STAT_FLUSH], | |
1289 | NSEC_PER_MSEC), | |
ea18e0f0 KK |
1290 | stat.ios[STAT_DISCARD], |
1291 | stat.merges[STAT_DISCARD], | |
1292 | stat.sectors[STAT_DISCARD], | |
1293 | (unsigned int)div_u64(stat.nsecs[STAT_DISCARD], | |
1294 | NSEC_PER_MSEC), | |
1295 | stat.ios[STAT_FLUSH], | |
1296 | (unsigned int)div_u64(stat.nsecs[STAT_FLUSH], | |
1297 | NSEC_PER_MSEC) | |
28f39d55 | 1298 | ); |
1da177e4 | 1299 | } |
7fae67cc | 1300 | rcu_read_unlock(); |
9f5e4865 | 1301 | |
1da177e4 LT |
1302 | return 0; |
1303 | } | |
1304 | ||
31d85ab2 | 1305 | static const struct seq_operations diskstats_op = { |
def4e38d TH |
1306 | .start = disk_seqf_start, |
1307 | .next = disk_seqf_next, | |
1308 | .stop = disk_seqf_stop, | |
1da177e4 LT |
1309 | .show = diskstats_show |
1310 | }; | |
f500975a AD |
1311 | |
1312 | static int __init proc_genhd_init(void) | |
1313 | { | |
fddda2b7 CH |
1314 | proc_create_seq("diskstats", 0, NULL, &diskstats_op); |
1315 | proc_create_seq("partitions", 0, NULL, &partitions_op); | |
f500975a AD |
1316 | return 0; |
1317 | } | |
1318 | module_init(proc_genhd_init); | |
a6e2ba88 | 1319 | #endif /* CONFIG_PROC_FS */ |
1da177e4 | 1320 | |
c97d93c3 CH |
1321 | dev_t part_devt(struct gendisk *disk, u8 partno) |
1322 | { | |
0e0ccdec | 1323 | struct block_device *part; |
c97d93c3 CH |
1324 | dev_t devt = 0; |
1325 | ||
0e0ccdec CH |
1326 | rcu_read_lock(); |
1327 | part = xa_load(&disk->part_tbl, partno); | |
1328 | if (part) | |
c97d93c3 | 1329 | devt = part->bd_dev; |
0e0ccdec | 1330 | rcu_read_unlock(); |
c97d93c3 CH |
1331 | |
1332 | return devt; | |
1333 | } | |
1334 | ||
4a1fa41d CH |
1335 | struct gendisk *__alloc_disk_node(struct request_queue *q, int node_id, |
1336 | struct lock_class_key *lkclass) | |
1946089a CL |
1337 | { |
1338 | struct gendisk *disk; | |
1339 | ||
c1b511eb | 1340 | disk = kzalloc_node(sizeof(struct gendisk), GFP_KERNEL, node_id); |
f93af2a4 | 1341 | if (!disk) |
aa0c680c | 1342 | return NULL; |
6c23a968 | 1343 | |
46754bd0 CH |
1344 | if (bioset_init(&disk->bio_split, BIO_POOL_SIZE, 0, 0)) |
1345 | goto out_free_disk; | |
1346 | ||
edb0872f CH |
1347 | disk->bdi = bdi_alloc(node_id); |
1348 | if (!disk->bdi) | |
46754bd0 | 1349 | goto out_free_bioset; |
edb0872f | 1350 | |
17220ca5 PB |
1351 | /* bdev_alloc() might need the queue, set before the first call */ |
1352 | disk->queue = q; | |
1353 | ||
cb8432d6 CH |
1354 | disk->part0 = bdev_alloc(disk, 0); |
1355 | if (!disk->part0) | |
edb0872f | 1356 | goto out_free_bdi; |
22ae8ce8 | 1357 | |
f93af2a4 | 1358 | disk->node_id = node_id; |
a8698707 | 1359 | mutex_init(&disk->open_mutex); |
a33df75c CH |
1360 | xa_init(&disk->part_tbl); |
1361 | if (xa_insert(&disk->part_tbl, 0, disk->part0, GFP_KERNEL)) | |
1362 | goto out_destroy_part_tbl; | |
f93af2a4 | 1363 | |
b6553bef CH |
1364 | if (blkcg_init_disk(disk)) |
1365 | goto out_erase_part0; | |
1366 | ||
f93af2a4 CH |
1367 | rand_initialize_disk(disk); |
1368 | disk_to_dev(disk)->class = &block_class; | |
1369 | disk_to_dev(disk)->type = &disk_type; | |
1370 | device_initialize(disk_to_dev(disk)); | |
cf179948 | 1371 | inc_diskseq(disk); |
d152c682 | 1372 | q->disk = disk; |
4dcc4874 | 1373 | lockdep_init_map(&disk->lockdep_map, "(bio completion)", lkclass, 0); |
0dbcfe24 CH |
1374 | #ifdef CONFIG_BLOCK_HOLDER_DEPRECATED |
1375 | INIT_LIST_HEAD(&disk->slave_bdevs); | |
1376 | #endif | |
1da177e4 | 1377 | return disk; |
f93af2a4 | 1378 | |
b6553bef CH |
1379 | out_erase_part0: |
1380 | xa_erase(&disk->part_tbl, 0); | |
a33df75c CH |
1381 | out_destroy_part_tbl: |
1382 | xa_destroy(&disk->part_tbl); | |
06cc978d | 1383 | disk->part0->bd_disk = NULL; |
2f4731dc | 1384 | iput(disk->part0->bd_inode); |
edb0872f CH |
1385 | out_free_bdi: |
1386 | bdi_put(disk->bdi); | |
46754bd0 CH |
1387 | out_free_bioset: |
1388 | bioset_exit(&disk->bio_split); | |
f93af2a4 CH |
1389 | out_free_disk: |
1390 | kfree(disk); | |
1391 | return NULL; | |
1da177e4 | 1392 | } |
1da177e4 | 1393 | |
74fa8f9c CH |
1394 | struct gendisk *__blk_alloc_disk(struct queue_limits *lim, int node, |
1395 | struct lock_class_key *lkclass) | |
f525464a | 1396 | { |
74fa8f9c | 1397 | struct queue_limits default_lim = { }; |
f525464a CH |
1398 | struct request_queue *q; |
1399 | struct gendisk *disk; | |
1400 | ||
74fa8f9c | 1401 | q = blk_alloc_queue(lim ? lim : &default_lim, node); |
ad751ba1 | 1402 | if (IS_ERR(q)) |
74fa8f9c | 1403 | return ERR_CAST(q); |
f525464a | 1404 | |
4a1fa41d | 1405 | disk = __alloc_disk_node(q, node, lkclass); |
f525464a | 1406 | if (!disk) { |
6f8191fd | 1407 | blk_put_queue(q); |
74fa8f9c | 1408 | return ERR_PTR(-ENOMEM); |
f525464a | 1409 | } |
6f8191fd | 1410 | set_bit(GD_OWNS_QUEUE, &disk->state); |
f525464a CH |
1411 | return disk; |
1412 | } | |
1413 | EXPORT_SYMBOL(__blk_alloc_disk); | |
1414 | ||
b5bd357c LC |
1415 | /** |
1416 | * put_disk - decrements the gendisk refcount | |
0d20dcc2 | 1417 | * @disk: the struct gendisk to decrement the refcount for |
b5bd357c LC |
1418 | * |
1419 | * This decrements the refcount for the struct gendisk. When this reaches 0 | |
1420 | * we'll have disk_release() called. | |
e8c7d14a | 1421 | * |
c5db2cfc CH |
1422 | * Note: for blk-mq disk put_disk must be called before freeing the tag_set |
1423 | * when handling probe errors (that is before add_disk() is called). | |
1424 | * | |
e8c7d14a LC |
1425 | * Context: Any context, but the last reference must not be dropped from |
1426 | * atomic context. | |
b5bd357c | 1427 | */ |
1da177e4 LT |
1428 | void put_disk(struct gendisk *disk) |
1429 | { | |
1430 | if (disk) | |
efdc41c8 | 1431 | put_device(disk_to_dev(disk)); |
1da177e4 | 1432 | } |
1da177e4 LT |
1433 | EXPORT_SYMBOL(put_disk); |
1434 | ||
e3264a4d HR |
1435 | static void set_disk_ro_uevent(struct gendisk *gd, int ro) |
1436 | { | |
1437 | char event[] = "DISK_RO=1"; | |
1438 | char *envp[] = { event, NULL }; | |
1439 | ||
1440 | if (!ro) | |
1441 | event[8] = '0'; | |
1442 | kobject_uevent_env(&disk_to_dev(gd)->kobj, KOBJ_CHANGE, envp); | |
1443 | } | |
1444 | ||
52f019d4 CH |
1445 | /** |
1446 | * set_disk_ro - set a gendisk read-only | |
1447 | * @disk: gendisk to operate on | |
7f31bee3 | 1448 | * @read_only: %true to set the disk read-only, %false set the disk read/write |
52f019d4 CH |
1449 | * |
1450 | * This function is used to indicate whether a given disk device should have its | |
1451 | * read-only flag set. set_disk_ro() is typically used by device drivers to | |
1452 | * indicate whether the underlying physical device is write-protected. | |
1453 | */ | |
1454 | void set_disk_ro(struct gendisk *disk, bool read_only) | |
1da177e4 | 1455 | { |
52f019d4 CH |
1456 | if (read_only) { |
1457 | if (test_and_set_bit(GD_READ_ONLY, &disk->state)) | |
1458 | return; | |
1459 | } else { | |
1460 | if (!test_and_clear_bit(GD_READ_ONLY, &disk->state)) | |
1461 | return; | |
e3264a4d | 1462 | } |
52f019d4 | 1463 | set_disk_ro_uevent(disk, read_only); |
1da177e4 | 1464 | } |
1da177e4 LT |
1465 | EXPORT_SYMBOL(set_disk_ro); |
1466 | ||
cf179948 MC |
1467 | void inc_diskseq(struct gendisk *disk) |
1468 | { | |
1469 | disk->diskseq = atomic64_inc_return(&diskseq); | |
1470 | } |