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> |
8f9e7b65 | 28 | #include "blk-throttle.h" |
1da177e4 | 29 | |
ff88972c | 30 | #include "blk.h" |
2aa7745b | 31 | #include "blk-mq-sched.h" |
8e141f9e | 32 | #include "blk-rq-qos.h" |
1059699f | 33 | #include "blk-cgroup.h" |
ff88972c | 34 | |
31eb6186 | 35 | static struct kobject *block_depr; |
1da177e4 | 36 | |
cf179948 MC |
37 | /* |
38 | * Unique, monotonically increasing sequential number associated with block | |
39 | * devices instances (i.e. incremented each time a device is attached). | |
40 | * Associating uevents with block devices in userspace is difficult and racy: | |
41 | * the uevent netlink socket is lossy, and on slow and overloaded systems has | |
42 | * a very high latency. | |
43 | * Block devices do not have exclusive owners in userspace, any process can set | |
44 | * one up (e.g. loop devices). Moreover, device names can be reused (e.g. loop0 | |
45 | * can be reused again and again). | |
46 | * A userspace process setting up a block device and watching for its events | |
47 | * cannot thus reliably tell whether an event relates to the device it just set | |
48 | * up or another earlier instance with the same name. | |
49 | * This sequential number allows userspace processes to solve this problem, and | |
50 | * uniquely associate an uevent to the lifetime to a device. | |
51 | */ | |
52 | static atomic64_t diskseq; | |
53 | ||
bcce3de1 | 54 | /* for extended dynamic devt allocation, currently only one major is used */ |
ce23bba8 | 55 | #define NR_EXT_DEVT (1 << MINORBITS) |
22ae8ce8 | 56 | static DEFINE_IDA(ext_devt_ida); |
bcce3de1 | 57 | |
a782483c CH |
58 | void set_capacity(struct gendisk *disk, sector_t sectors) |
59 | { | |
83794367 | 60 | bdev_set_nr_sectors(disk->part0, sectors); |
a782483c CH |
61 | } |
62 | EXPORT_SYMBOL(set_capacity); | |
63 | ||
e598a72f | 64 | /* |
449f4ec9 CH |
65 | * Set disk capacity and notify if the size is not currently zero and will not |
66 | * be set to zero. Returns true if a uevent was sent, otherwise false. | |
e598a72f | 67 | */ |
449f4ec9 | 68 | bool set_capacity_and_notify(struct gendisk *disk, sector_t size) |
e598a72f BS |
69 | { |
70 | sector_t capacity = get_capacity(disk); | |
a782483c | 71 | char *envp[] = { "RESIZE=1", NULL }; |
e598a72f BS |
72 | |
73 | set_capacity(disk, size); | |
e598a72f | 74 | |
a782483c CH |
75 | /* |
76 | * Only print a message and send a uevent if the gendisk is user visible | |
77 | * and alive. This avoids spamming the log and udev when setting the | |
78 | * initial capacity during probing. | |
79 | */ | |
80 | if (size == capacity || | |
50b4aecf CH |
81 | !disk_live(disk) || |
82 | (disk->flags & GENHD_FL_HIDDEN)) | |
a782483c | 83 | return false; |
e598a72f | 84 | |
a782483c | 85 | pr_info("%s: detected capacity change from %lld to %lld\n", |
452c0bf8 | 86 | disk->disk_name, capacity, size); |
7e890c37 | 87 | |
a782483c CH |
88 | /* |
89 | * Historically we did not send a uevent for changes to/from an empty | |
90 | * device. | |
91 | */ | |
92 | if (!capacity || !size) | |
93 | return false; | |
94 | kobject_uevent_env(&disk_to_dev(disk)->kobj, KOBJ_CHANGE, envp); | |
95 | return true; | |
e598a72f | 96 | } |
449f4ec9 | 97 | EXPORT_SYMBOL_GPL(set_capacity_and_notify); |
e598a72f | 98 | |
0d02129e CH |
99 | static void part_stat_read_all(struct block_device *part, |
100 | struct disk_stats *stat) | |
ea18e0f0 KK |
101 | { |
102 | int cpu; | |
103 | ||
104 | memset(stat, 0, sizeof(struct disk_stats)); | |
105 | for_each_possible_cpu(cpu) { | |
0d02129e | 106 | struct disk_stats *ptr = per_cpu_ptr(part->bd_stats, cpu); |
ea18e0f0 KK |
107 | int group; |
108 | ||
109 | for (group = 0; group < NR_STAT_GROUPS; group++) { | |
110 | stat->nsecs[group] += ptr->nsecs[group]; | |
111 | stat->sectors[group] += ptr->sectors[group]; | |
112 | stat->ios[group] += ptr->ios[group]; | |
113 | stat->merges[group] += ptr->merges[group]; | |
114 | } | |
115 | ||
116 | stat->io_ticks += ptr->io_ticks; | |
ea18e0f0 KK |
117 | } |
118 | } | |
ea18e0f0 | 119 | |
8446fe92 | 120 | static unsigned int part_in_flight(struct block_device *part) |
f299b7c7 | 121 | { |
b2f609e1 | 122 | unsigned int inflight = 0; |
1226b8dd | 123 | int cpu; |
f299b7c7 | 124 | |
1226b8dd | 125 | for_each_possible_cpu(cpu) { |
e016b782 MP |
126 | inflight += part_stat_local_read_cpu(part, in_flight[0], cpu) + |
127 | part_stat_local_read_cpu(part, in_flight[1], cpu); | |
1226b8dd | 128 | } |
e016b782 MP |
129 | if ((int)inflight < 0) |
130 | inflight = 0; | |
1226b8dd | 131 | |
e016b782 | 132 | return inflight; |
f299b7c7 JA |
133 | } |
134 | ||
8446fe92 CH |
135 | static void part_in_flight_rw(struct block_device *part, |
136 | unsigned int inflight[2]) | |
bf0ddaba | 137 | { |
1226b8dd MP |
138 | int cpu; |
139 | ||
1226b8dd MP |
140 | inflight[0] = 0; |
141 | inflight[1] = 0; | |
142 | for_each_possible_cpu(cpu) { | |
143 | inflight[0] += part_stat_local_read_cpu(part, in_flight[0], cpu); | |
144 | inflight[1] += part_stat_local_read_cpu(part, in_flight[1], cpu); | |
145 | } | |
146 | if ((int)inflight[0] < 0) | |
147 | inflight[0] = 0; | |
148 | if ((int)inflight[1] < 0) | |
149 | inflight[1] = 0; | |
bf0ddaba OS |
150 | } |
151 | ||
1da177e4 LT |
152 | /* |
153 | * Can be deleted altogether. Later. | |
154 | * | |
155 | */ | |
133d55cd | 156 | #define BLKDEV_MAJOR_HASH_SIZE 255 |
1da177e4 LT |
157 | static struct blk_major_name { |
158 | struct blk_major_name *next; | |
159 | int major; | |
160 | char name[16]; | |
fbdee71b | 161 | #ifdef CONFIG_BLOCK_LEGACY_AUTOLOAD |
a160c615 | 162 | void (*probe)(dev_t devt); |
fbdee71b | 163 | #endif |
68eef3b4 | 164 | } *major_names[BLKDEV_MAJOR_HASH_SIZE]; |
e49fbbbf | 165 | static DEFINE_MUTEX(major_names_lock); |
dfbb3409 | 166 | static DEFINE_SPINLOCK(major_names_spinlock); |
1da177e4 LT |
167 | |
168 | /* index in the above - for now: assume no multimajor ranges */ | |
e61eb2e9 | 169 | static inline int major_to_index(unsigned major) |
1da177e4 | 170 | { |
68eef3b4 | 171 | return major % BLKDEV_MAJOR_HASH_SIZE; |
7170be5f NH |
172 | } |
173 | ||
68eef3b4 | 174 | #ifdef CONFIG_PROC_FS |
cf771cb5 | 175 | void blkdev_show(struct seq_file *seqf, off_t offset) |
7170be5f | 176 | { |
68eef3b4 | 177 | struct blk_major_name *dp; |
7170be5f | 178 | |
dfbb3409 | 179 | spin_lock(&major_names_spinlock); |
133d55cd LG |
180 | for (dp = major_names[major_to_index(offset)]; dp; dp = dp->next) |
181 | if (dp->major == offset) | |
cf771cb5 | 182 | seq_printf(seqf, "%3d %s\n", dp->major, dp->name); |
dfbb3409 | 183 | spin_unlock(&major_names_spinlock); |
1da177e4 | 184 | } |
68eef3b4 | 185 | #endif /* CONFIG_PROC_FS */ |
1da177e4 | 186 | |
9e8c0bcc | 187 | /** |
e2b6b301 | 188 | * __register_blkdev - register a new block device |
9e8c0bcc | 189 | * |
f33ff110 SB |
190 | * @major: the requested major device number [1..BLKDEV_MAJOR_MAX-1]. If |
191 | * @major = 0, try to allocate any unused major number. | |
9e8c0bcc | 192 | * @name: the name of the new block device as a zero terminated string |
26e06f5b LC |
193 | * @probe: pre-devtmpfs / pre-udev callback used to create disks when their |
194 | * pre-created device node is accessed. When a probe call uses | |
195 | * add_disk() and it fails the driver must cleanup resources. This | |
196 | * interface may soon be removed. | |
9e8c0bcc MN |
197 | * |
198 | * The @name must be unique within the system. | |
199 | * | |
0e056eb5 | 200 | * The return value depends on the @major input parameter: |
201 | * | |
f33ff110 SB |
202 | * - if a major device number was requested in range [1..BLKDEV_MAJOR_MAX-1] |
203 | * then the function returns zero on success, or a negative error code | |
0e056eb5 | 204 | * - if any unused major number was requested with @major = 0 parameter |
9e8c0bcc | 205 | * then the return value is the allocated major number in range |
f33ff110 SB |
206 | * [1..BLKDEV_MAJOR_MAX-1] or a negative error code otherwise |
207 | * | |
208 | * See Documentation/admin-guide/devices.txt for the list of allocated | |
209 | * major numbers. | |
e2b6b301 CH |
210 | * |
211 | * Use register_blkdev instead for any new code. | |
9e8c0bcc | 212 | */ |
a160c615 CH |
213 | int __register_blkdev(unsigned int major, const char *name, |
214 | void (*probe)(dev_t devt)) | |
1da177e4 LT |
215 | { |
216 | struct blk_major_name **n, *p; | |
217 | int index, ret = 0; | |
218 | ||
e49fbbbf | 219 | mutex_lock(&major_names_lock); |
1da177e4 LT |
220 | |
221 | /* temporary */ | |
222 | if (major == 0) { | |
223 | for (index = ARRAY_SIZE(major_names)-1; index > 0; index--) { | |
224 | if (major_names[index] == NULL) | |
225 | break; | |
226 | } | |
227 | ||
228 | if (index == 0) { | |
dfc76d11 KP |
229 | printk("%s: failed to get major for %s\n", |
230 | __func__, name); | |
1da177e4 LT |
231 | ret = -EBUSY; |
232 | goto out; | |
233 | } | |
234 | major = index; | |
235 | ret = major; | |
236 | } | |
237 | ||
133d55cd | 238 | if (major >= BLKDEV_MAJOR_MAX) { |
dfc76d11 KP |
239 | pr_err("%s: major requested (%u) is greater than the maximum (%u) for %s\n", |
240 | __func__, major, BLKDEV_MAJOR_MAX-1, name); | |
133d55cd LG |
241 | |
242 | ret = -EINVAL; | |
243 | goto out; | |
244 | } | |
245 | ||
1da177e4 LT |
246 | p = kmalloc(sizeof(struct blk_major_name), GFP_KERNEL); |
247 | if (p == NULL) { | |
248 | ret = -ENOMEM; | |
249 | goto out; | |
250 | } | |
251 | ||
252 | p->major = major; | |
fbdee71b | 253 | #ifdef CONFIG_BLOCK_LEGACY_AUTOLOAD |
a160c615 | 254 | p->probe = probe; |
fbdee71b | 255 | #endif |
1da177e4 LT |
256 | strlcpy(p->name, name, sizeof(p->name)); |
257 | p->next = NULL; | |
258 | index = major_to_index(major); | |
259 | ||
dfbb3409 | 260 | spin_lock(&major_names_spinlock); |
1da177e4 LT |
261 | for (n = &major_names[index]; *n; n = &(*n)->next) { |
262 | if ((*n)->major == major) | |
263 | break; | |
264 | } | |
265 | if (!*n) | |
266 | *n = p; | |
267 | else | |
268 | ret = -EBUSY; | |
dfbb3409 | 269 | spin_unlock(&major_names_spinlock); |
1da177e4 LT |
270 | |
271 | if (ret < 0) { | |
f33ff110 | 272 | printk("register_blkdev: cannot get major %u for %s\n", |
1da177e4 LT |
273 | major, name); |
274 | kfree(p); | |
275 | } | |
276 | out: | |
e49fbbbf | 277 | mutex_unlock(&major_names_lock); |
1da177e4 LT |
278 | return ret; |
279 | } | |
a160c615 | 280 | EXPORT_SYMBOL(__register_blkdev); |
1da177e4 | 281 | |
f4480240 | 282 | void unregister_blkdev(unsigned int major, const char *name) |
1da177e4 LT |
283 | { |
284 | struct blk_major_name **n; | |
285 | struct blk_major_name *p = NULL; | |
286 | int index = major_to_index(major); | |
1da177e4 | 287 | |
e49fbbbf | 288 | mutex_lock(&major_names_lock); |
dfbb3409 | 289 | spin_lock(&major_names_spinlock); |
1da177e4 LT |
290 | for (n = &major_names[index]; *n; n = &(*n)->next) |
291 | if ((*n)->major == major) | |
292 | break; | |
294462a5 AM |
293 | if (!*n || strcmp((*n)->name, name)) { |
294 | WARN_ON(1); | |
294462a5 | 295 | } else { |
1da177e4 LT |
296 | p = *n; |
297 | *n = p->next; | |
298 | } | |
dfbb3409 | 299 | spin_unlock(&major_names_spinlock); |
e49fbbbf | 300 | mutex_unlock(&major_names_lock); |
1da177e4 | 301 | kfree(p); |
1da177e4 LT |
302 | } |
303 | ||
304 | EXPORT_SYMBOL(unregister_blkdev); | |
305 | ||
7c3f828b | 306 | int blk_alloc_ext_minor(void) |
bcce3de1 | 307 | { |
bab998d6 | 308 | int idx; |
bcce3de1 | 309 | |
d1868328 | 310 | idx = ida_alloc_range(&ext_devt_ida, 0, NR_EXT_DEVT - 1, GFP_KERNEL); |
c4b2b7d1 CH |
311 | if (idx == -ENOSPC) |
312 | return -EBUSY; | |
313 | return idx; | |
bcce3de1 TH |
314 | } |
315 | ||
7c3f828b | 316 | void blk_free_ext_minor(unsigned int minor) |
bcce3de1 | 317 | { |
c4b2b7d1 | 318 | ida_free(&ext_devt_ida, minor); |
6fcc44d1 YY |
319 | } |
320 | ||
1f014290 TH |
321 | static char *bdevt_str(dev_t devt, char *buf) |
322 | { | |
323 | if (MAJOR(devt) <= 0xff && MINOR(devt) <= 0xff) { | |
324 | char tbuf[BDEVT_SIZE]; | |
325 | snprintf(tbuf, BDEVT_SIZE, "%02x%02x", MAJOR(devt), MINOR(devt)); | |
326 | snprintf(buf, BDEVT_SIZE, "%-9s", tbuf); | |
327 | } else | |
328 | snprintf(buf, BDEVT_SIZE, "%03x:%05x", MAJOR(devt), MINOR(devt)); | |
329 | ||
330 | return buf; | |
331 | } | |
332 | ||
bc359d03 CH |
333 | void disk_uevent(struct gendisk *disk, enum kobject_action action) |
334 | { | |
bc359d03 | 335 | struct block_device *part; |
3212135a | 336 | unsigned long idx; |
bc359d03 | 337 | |
3212135a CH |
338 | rcu_read_lock(); |
339 | xa_for_each(&disk->part_tbl, idx, part) { | |
340 | if (bdev_is_partition(part) && !bdev_nr_sectors(part)) | |
341 | continue; | |
498dcc13 | 342 | if (!kobject_get_unless_zero(&part->bd_device.kobj)) |
3212135a CH |
343 | continue; |
344 | ||
345 | rcu_read_unlock(); | |
bc359d03 | 346 | kobject_uevent(bdev_kobj(part), action); |
498dcc13 | 347 | put_device(&part->bd_device); |
3212135a CH |
348 | rcu_read_lock(); |
349 | } | |
350 | rcu_read_unlock(); | |
bc359d03 CH |
351 | } |
352 | EXPORT_SYMBOL_GPL(disk_uevent); | |
353 | ||
0f77b29a | 354 | int disk_scan_partitions(struct gendisk *disk, fmode_t mode) |
9301fe73 CH |
355 | { |
356 | struct block_device *bdev; | |
e5cfefa9 | 357 | int ret = 0; |
9301fe73 | 358 | |
9f18db57 | 359 | if (disk->flags & (GENHD_FL_NO_PART | GENHD_FL_HIDDEN)) |
e16e506c | 360 | return -EINVAL; |
b9684a71 CH |
361 | if (test_bit(GD_SUPPRESS_PART_SCAN, &disk->state)) |
362 | return -EINVAL; | |
e16e506c CH |
363 | if (disk->open_partitions) |
364 | return -EBUSY; | |
9301fe73 | 365 | |
e5cfefa9 YK |
366 | /* |
367 | * If the device is opened exclusively by current thread already, it's | |
368 | * safe to scan partitons, otherwise, use bd_prepare_to_claim() to | |
369 | * synchronize with other exclusive openers and other partition | |
370 | * scanners. | |
371 | */ | |
372 | if (!(mode & FMODE_EXCL)) { | |
373 | ret = bd_prepare_to_claim(disk->part0, disk_scan_partitions); | |
374 | if (ret) | |
375 | return ret; | |
376 | } | |
377 | ||
3723091e | 378 | set_bit(GD_NEED_PART_SCAN, &disk->state); |
e5cfefa9 | 379 | bdev = blkdev_get_by_dev(disk_devt(disk), mode & ~FMODE_EXCL, NULL); |
e16e506c | 380 | if (IS_ERR(bdev)) |
e5cfefa9 YK |
381 | ret = PTR_ERR(bdev); |
382 | else | |
428913bc | 383 | blkdev_put(bdev, mode & ~FMODE_EXCL); |
e5cfefa9 | 384 | |
3723091e YK |
385 | /* |
386 | * If blkdev_get_by_dev() failed early, GD_NEED_PART_SCAN is still set, | |
387 | * and this will cause that re-assemble partitioned raid device will | |
388 | * creat partition for underlying disk. | |
389 | */ | |
390 | clear_bit(GD_NEED_PART_SCAN, &disk->state); | |
e5cfefa9 YK |
391 | if (!(mode & FMODE_EXCL)) |
392 | bd_abort_claiming(disk->part0, disk_scan_partitions); | |
393 | return ret; | |
9301fe73 CH |
394 | } |
395 | ||
1da177e4 | 396 | /** |
d1254a87 | 397 | * device_add_disk - add disk information to kernel list |
e63a46be | 398 | * @parent: parent device for the disk |
1da177e4 | 399 | * @disk: per-device partitioning information |
fef912bf | 400 | * @groups: Additional per-device sysfs groups |
1da177e4 LT |
401 | * |
402 | * This function registers the partitioning information in @disk | |
403 | * with the kernel. | |
404 | */ | |
278167fd LC |
405 | int __must_check device_add_disk(struct device *parent, struct gendisk *disk, |
406 | const struct attribute_group **groups) | |
d1254a87 | 407 | |
1da177e4 | 408 | { |
52b85909 | 409 | struct device *ddev = disk_to_dev(disk); |
7c3f828b | 410 | int ret; |
cf0ca9fe | 411 | |
69fe0f29 ML |
412 | /* Only makes sense for bio-based to set ->poll_bio */ |
413 | if (queue_is_mq(disk->queue) && disk->fops->poll_bio) | |
414 | return -EINVAL; | |
415 | ||
737eb78e DLM |
416 | /* |
417 | * The disk queue should now be all set with enough information about | |
418 | * the device for the elevator code to pick an adequate default | |
419 | * elevator if one is needed, that is, for devices requesting queue | |
420 | * registration. | |
421 | */ | |
d1254a87 | 422 | elevator_init_mq(disk->queue); |
737eb78e | 423 | |
9f4107b0 JA |
424 | /* Mark bdev as having a submit_bio, if needed */ |
425 | disk->part0->bd_has_submit_bio = disk->fops->submit_bio != NULL; | |
426 | ||
7c3f828b CH |
427 | /* |
428 | * If the driver provides an explicit major number it also must provide | |
429 | * the number of minors numbers supported, and those will be used to | |
430 | * setup the gendisk. | |
431 | * Otherwise just allocate the device numbers for both the whole device | |
432 | * and all partitions from the extended dev_t space. | |
3e1a7ff8 | 433 | */ |
02341a08 | 434 | ret = -EINVAL; |
7c3f828b | 435 | if (disk->major) { |
83cbce95 | 436 | if (WARN_ON(!disk->minors)) |
02341a08 | 437 | goto out_exit_elevator; |
2e3c73fa CH |
438 | |
439 | if (disk->minors > DISK_MAX_PARTS) { | |
440 | pr_err("block: can't allocate more than %d partitions\n", | |
441 | DISK_MAX_PARTS); | |
442 | disk->minors = DISK_MAX_PARTS; | |
443 | } | |
e338924b | 444 | if (disk->first_minor + disk->minors > MINORMASK + 1) |
02341a08 | 445 | goto out_exit_elevator; |
7c3f828b | 446 | } else { |
83cbce95 | 447 | if (WARN_ON(disk->minors)) |
02341a08 | 448 | goto out_exit_elevator; |
3e1a7ff8 | 449 | |
7c3f828b | 450 | ret = blk_alloc_ext_minor(); |
83cbce95 | 451 | if (ret < 0) |
02341a08 | 452 | goto out_exit_elevator; |
7c3f828b | 453 | disk->major = BLOCK_EXT_MAJOR; |
539711d7 | 454 | disk->first_minor = ret; |
3e1a7ff8 | 455 | } |
7c3f828b | 456 | |
52b85909 CH |
457 | /* delay uevents, until we scanned partition table */ |
458 | dev_set_uevent_suppress(ddev, 1); | |
459 | ||
460 | ddev->parent = parent; | |
461 | ddev->groups = groups; | |
462 | dev_set_name(ddev, "%s", disk->disk_name); | |
8235b5c1 CH |
463 | if (!(disk->flags & GENHD_FL_HIDDEN)) |
464 | ddev->devt = MKDEV(disk->major, disk->first_minor); | |
83cbce95 LC |
465 | ret = device_add(ddev); |
466 | if (ret) | |
99d8690a CH |
467 | goto out_free_ext_minor; |
468 | ||
469 | ret = disk_alloc_events(disk); | |
470 | if (ret) | |
471 | goto out_device_del; | |
472 | ||
721da5ce GKH |
473 | ret = sysfs_create_link(block_depr, &ddev->kobj, |
474 | kobject_name(&ddev->kobj)); | |
475 | if (ret) | |
476 | goto out_device_del; | |
52b85909 CH |
477 | |
478 | /* | |
479 | * avoid probable deadlock caused by allocating memory with | |
480 | * GFP_KERNEL in runtime_resume callback of its all ancestor | |
481 | * devices | |
482 | */ | |
483 | pm_runtime_set_memalloc_noio(ddev, true); | |
484 | ||
485 | disk->part0->bd_holder_dir = | |
486 | kobject_create_and_add("holders", &ddev->kobj); | |
fe7d064f LC |
487 | if (!disk->part0->bd_holder_dir) { |
488 | ret = -ENOMEM; | |
ff53cd52 | 489 | goto out_del_block_link; |
fe7d064f | 490 | } |
52b85909 | 491 | disk->slave_dir = kobject_create_and_add("slaves", &ddev->kobj); |
fe7d064f LC |
492 | if (!disk->slave_dir) { |
493 | ret = -ENOMEM; | |
83cbce95 | 494 | goto out_put_holder_dir; |
fe7d064f | 495 | } |
52b85909 | 496 | |
83cbce95 LC |
497 | ret = blk_register_queue(disk); |
498 | if (ret) | |
499 | goto out_put_slave_dir; | |
75f4dca5 | 500 | |
9f18db57 | 501 | if (!(disk->flags & GENHD_FL_HIDDEN)) { |
8235b5c1 CH |
502 | ret = bdi_register(disk->bdi, "%u:%u", |
503 | disk->major, disk->first_minor); | |
83cbce95 LC |
504 | if (ret) |
505 | goto out_unregister_queue; | |
8235b5c1 | 506 | bdi_set_owner(disk->bdi, ddev); |
83cbce95 LC |
507 | ret = sysfs_create_link(&ddev->kobj, |
508 | &disk->bdi->dev->kobj, "bdi"); | |
509 | if (ret) | |
510 | goto out_unregister_bdi; | |
8235b5c1 | 511 | |
e5cfefa9 YK |
512 | /* Make sure the first partition scan will be proceed */ |
513 | if (get_capacity(disk) && !(disk->flags & GENHD_FL_NO_PART) && | |
514 | !test_bit(GD_SUPPRESS_PART_SCAN, &disk->state)) | |
515 | set_bit(GD_NEED_PART_SCAN, &disk->state); | |
516 | ||
9d5ee676 | 517 | bdev_add(disk->part0, ddev->devt); |
e16e506c | 518 | if (get_capacity(disk)) |
0f77b29a | 519 | disk_scan_partitions(disk, FMODE_READ); |
52b85909 CH |
520 | |
521 | /* | |
522 | * Announce the disk and partitions after all partitions are | |
8235b5c1 | 523 | * created. (for hidden disks uevents remain suppressed forever) |
52b85909 CH |
524 | */ |
525 | dev_set_uevent_suppress(ddev, 0); | |
526 | disk_uevent(disk, KOBJ_ADD); | |
a0a6314a CH |
527 | } else { |
528 | /* | |
529 | * Even if the block_device for a hidden gendisk is not | |
530 | * registered, it needs to have a valid bd_dev so that the | |
531 | * freeing of the dynamic major works. | |
532 | */ | |
533 | disk->part0->bd_dev = MKDEV(disk->major, disk->first_minor); | |
52b85909 CH |
534 | } |
535 | ||
75f4dca5 | 536 | disk_update_readahead(disk); |
77ea887e | 537 | disk_add_events(disk); |
76792055 | 538 | set_bit(GD_ADDED, &disk->state); |
83cbce95 LC |
539 | return 0; |
540 | ||
541 | out_unregister_bdi: | |
542 | if (!(disk->flags & GENHD_FL_HIDDEN)) | |
543 | bdi_unregister(disk->bdi); | |
544 | out_unregister_queue: | |
545 | blk_unregister_queue(disk); | |
fa81cbaf | 546 | rq_qos_exit(disk->queue); |
83cbce95 LC |
547 | out_put_slave_dir: |
548 | kobject_put(disk->slave_dir); | |
d90db3b1 | 549 | disk->slave_dir = NULL; |
83cbce95 LC |
550 | out_put_holder_dir: |
551 | kobject_put(disk->part0->bd_holder_dir); | |
83cbce95 | 552 | out_del_block_link: |
721da5ce | 553 | sysfs_remove_link(block_depr, dev_name(ddev)); |
83cbce95 LC |
554 | out_device_del: |
555 | device_del(ddev); | |
83cbce95 LC |
556 | out_free_ext_minor: |
557 | if (disk->major == BLOCK_EXT_MAJOR) | |
558 | blk_free_ext_minor(disk->first_minor); | |
02341a08 YK |
559 | out_exit_elevator: |
560 | if (disk->queue->elevator) | |
561 | elevator_exit(disk->queue); | |
278167fd | 562 | return ret; |
1da177e4 | 563 | } |
e63a46be | 564 | EXPORT_SYMBOL(device_add_disk); |
1da177e4 | 565 | |
7a5428dc CH |
566 | /** |
567 | * blk_mark_disk_dead - mark a disk as dead | |
568 | * @disk: disk to mark as dead | |
569 | * | |
570 | * Mark as disk as dead (e.g. surprise removed) and don't accept any new I/O | |
571 | * to this disk. | |
572 | */ | |
573 | void blk_mark_disk_dead(struct gendisk *disk) | |
574 | { | |
575 | set_bit(GD_DEAD, &disk->state); | |
576 | blk_queue_start_drain(disk->queue); | |
71b26083 CH |
577 | |
578 | /* | |
579 | * Stop buffered writers from dirtying pages that can't be written out. | |
580 | */ | |
581 | set_capacity_and_notify(disk, 0); | |
7a5428dc CH |
582 | } |
583 | EXPORT_SYMBOL_GPL(blk_mark_disk_dead); | |
584 | ||
b5bd357c LC |
585 | /** |
586 | * del_gendisk - remove the gendisk | |
587 | * @disk: the struct gendisk to remove | |
588 | * | |
589 | * Removes the gendisk and all its associated resources. This deletes the | |
590 | * partitions associated with the gendisk, and unregisters the associated | |
591 | * request_queue. | |
592 | * | |
593 | * This is the counter to the respective __device_add_disk() call. | |
594 | * | |
595 | * The final removal of the struct gendisk happens when its refcount reaches 0 | |
596 | * with put_disk(), which should be called after del_gendisk(), if | |
597 | * __device_add_disk() was used. | |
e8c7d14a LC |
598 | * |
599 | * Drivers exist which depend on the release of the gendisk to be synchronous, | |
600 | * it should not be deferred. | |
601 | * | |
602 | * Context: can sleep | |
b5bd357c | 603 | */ |
d2bf1b67 | 604 | void del_gendisk(struct gendisk *disk) |
1da177e4 | 605 | { |
8e141f9e CH |
606 | struct request_queue *q = disk->queue; |
607 | ||
e8c7d14a LC |
608 | might_sleep(); |
609 | ||
9f286992 | 610 | if (WARN_ON_ONCE(!disk_live(disk) && !(disk->flags & GENHD_FL_HIDDEN))) |
6b3ba976 CH |
611 | return; |
612 | ||
77ea887e TH |
613 | disk_del_events(disk); |
614 | ||
a8698707 | 615 | mutex_lock(&disk->open_mutex); |
d7a66574 | 616 | remove_inode_hash(disk->part0->bd_inode); |
d3c4a43d | 617 | blk_drop_partitions(disk); |
a8698707 | 618 | mutex_unlock(&disk->open_mutex); |
c76f48eb | 619 | |
45611837 CH |
620 | fsync_bdev(disk->part0); |
621 | __invalidate_device(disk->part0, true); | |
622 | ||
8e141f9e CH |
623 | /* |
624 | * Fail any new I/O. | |
625 | */ | |
626 | set_bit(GD_DEAD, &disk->state); | |
6f8191fd CH |
627 | if (test_bit(GD_OWNS_QUEUE, &disk->state)) |
628 | blk_queue_flag_set(QUEUE_FLAG_DYING, q); | |
d2bf1b67 | 629 | set_capacity(disk, 0); |
d2bf1b67 | 630 | |
8e141f9e CH |
631 | /* |
632 | * Prevent new I/O from crossing bio_queue_enter(). | |
633 | */ | |
634 | blk_queue_start_drain(q); | |
8e141f9e | 635 | |
6b3ba976 | 636 | if (!(disk->flags & GENHD_FL_HIDDEN)) { |
8ddcd653 | 637 | sysfs_remove_link(&disk_to_dev(disk)->kobj, "bdi"); |
6b3ba976 | 638 | |
90f16fdd JK |
639 | /* |
640 | * Unregister bdi before releasing device numbers (as they can | |
641 | * get reused and we'd get clashes in sysfs). | |
642 | */ | |
edb0872f | 643 | bdi_unregister(disk->bdi); |
90f16fdd | 644 | } |
d2bf1b67 | 645 | |
6b3ba976 | 646 | blk_unregister_queue(disk); |
d2bf1b67 | 647 | |
cb8432d6 | 648 | kobject_put(disk->part0->bd_holder_dir); |
d2bf1b67 | 649 | kobject_put(disk->slave_dir); |
d90db3b1 | 650 | disk->slave_dir = NULL; |
d2bf1b67 | 651 | |
8446fe92 | 652 | part_stat_set_all(disk->part0, 0); |
cb8432d6 | 653 | disk->part0->bd_stamp = 0; |
721da5ce | 654 | sysfs_remove_link(block_depr, dev_name(disk_to_dev(disk))); |
25e823c8 | 655 | pm_runtime_set_memalloc_noio(disk_to_dev(disk), false); |
d2bf1b67 | 656 | device_del(disk_to_dev(disk)); |
d308ae0d | 657 | |
4c66a326 CH |
658 | blk_mq_freeze_queue_wait(q); |
659 | ||
cad9266a | 660 | blk_throtl_cancel_bios(disk); |
8f9e7b65 | 661 | |
d308ae0d ML |
662 | blk_sync_queue(q); |
663 | blk_flush_integrity(); | |
219cf43c JC |
664 | |
665 | if (queue_is_mq(q)) | |
666 | blk_mq_cancel_work_sync(q); | |
50e34d78 CH |
667 | |
668 | blk_mq_quiesce_queue(q); | |
669 | if (q->elevator) { | |
670 | mutex_lock(&q->sysfs_lock); | |
671 | elevator_exit(q); | |
672 | mutex_unlock(&q->sysfs_lock); | |
673 | } | |
674 | rq_qos_exit(q); | |
675 | blk_mq_unquiesce_queue(q); | |
676 | ||
d308ae0d | 677 | /* |
6f8191fd CH |
678 | * If the disk does not own the queue, allow using passthrough requests |
679 | * again. Else leave the queue frozen to fail all I/O. | |
d308ae0d | 680 | */ |
6f8191fd CH |
681 | if (!test_bit(GD_OWNS_QUEUE, &disk->state)) { |
682 | blk_queue_flag_clear(QUEUE_FLAG_INIT_DONE, q); | |
683 | __blk_mq_unfreeze_queue(q, true); | |
684 | } else { | |
685 | if (queue_is_mq(q)) | |
686 | blk_mq_exit_queue(q); | |
687 | } | |
1da177e4 | 688 | } |
d2bf1b67 | 689 | EXPORT_SYMBOL(del_gendisk); |
1da177e4 | 690 | |
f059a1d2 XY |
691 | /** |
692 | * invalidate_disk - invalidate the disk | |
693 | * @disk: the struct gendisk to invalidate | |
694 | * | |
695 | * A helper to invalidates the disk. It will clean the disk's associated | |
696 | * buffer/page caches and reset its internal states so that the disk | |
697 | * can be reused by the drivers. | |
698 | * | |
699 | * Context: can sleep | |
700 | */ | |
701 | void invalidate_disk(struct gendisk *disk) | |
702 | { | |
703 | struct block_device *bdev = disk->part0; | |
704 | ||
705 | invalidate_bdev(bdev); | |
706 | bdev->bd_inode->i_mapping->wb_err = 0; | |
707 | set_capacity(disk, 0); | |
708 | } | |
709 | EXPORT_SYMBOL(invalidate_disk); | |
710 | ||
99e6608c VV |
711 | /* sysfs access to bad-blocks list. */ |
712 | static ssize_t disk_badblocks_show(struct device *dev, | |
713 | struct device_attribute *attr, | |
714 | char *page) | |
715 | { | |
716 | struct gendisk *disk = dev_to_disk(dev); | |
717 | ||
718 | if (!disk->bb) | |
719 | return sprintf(page, "\n"); | |
720 | ||
721 | return badblocks_show(disk->bb, page, 0); | |
722 | } | |
723 | ||
724 | static ssize_t disk_badblocks_store(struct device *dev, | |
725 | struct device_attribute *attr, | |
726 | const char *page, size_t len) | |
727 | { | |
728 | struct gendisk *disk = dev_to_disk(dev); | |
729 | ||
730 | if (!disk->bb) | |
731 | return -ENXIO; | |
732 | ||
733 | return badblocks_store(disk->bb, page, len, 0); | |
734 | } | |
735 | ||
fbdee71b | 736 | #ifdef CONFIG_BLOCK_LEGACY_AUTOLOAD |
22ae8ce8 | 737 | void blk_request_module(dev_t devt) |
bd8eff3b | 738 | { |
a160c615 CH |
739 | unsigned int major = MAJOR(devt); |
740 | struct blk_major_name **n; | |
741 | ||
742 | mutex_lock(&major_names_lock); | |
743 | for (n = &major_names[major_to_index(major)]; *n; n = &(*n)->next) { | |
744 | if ((*n)->major == major && (*n)->probe) { | |
745 | (*n)->probe(devt); | |
746 | mutex_unlock(&major_names_lock); | |
747 | return; | |
748 | } | |
749 | } | |
750 | mutex_unlock(&major_names_lock); | |
751 | ||
bd8eff3b CH |
752 | if (request_module("block-major-%d-%d", MAJOR(devt), MINOR(devt)) > 0) |
753 | /* Make old-style 2.4 aliases work */ | |
754 | request_module("block-major-%d", MAJOR(devt)); | |
755 | } | |
fbdee71b | 756 | #endif /* CONFIG_BLOCK_LEGACY_AUTOLOAD */ |
bd8eff3b | 757 | |
5c6f35c5 GKH |
758 | /* |
759 | * print a full list of all partitions - intended for places where the root | |
760 | * filesystem can't be mounted and thus to give the victim some idea of what | |
761 | * went wrong | |
762 | */ | |
763 | void __init printk_all_partitions(void) | |
764 | { | |
def4e38d TH |
765 | struct class_dev_iter iter; |
766 | struct device *dev; | |
767 | ||
768 | class_dev_iter_init(&iter, &block_class, NULL, &disk_type); | |
769 | while ((dev = class_dev_iter_next(&iter))) { | |
770 | struct gendisk *disk = dev_to_disk(dev); | |
ad1eaa53 | 771 | struct block_device *part; |
1f014290 | 772 | char devt_buf[BDEVT_SIZE]; |
e559f58d | 773 | unsigned long idx; |
def4e38d TH |
774 | |
775 | /* | |
776 | * Don't show empty devices or things that have been | |
25985edc | 777 | * suppressed |
def4e38d | 778 | */ |
3b5149ac | 779 | if (get_capacity(disk) == 0 || (disk->flags & GENHD_FL_HIDDEN)) |
def4e38d TH |
780 | continue; |
781 | ||
782 | /* | |
e559f58d CH |
783 | * Note, unlike /proc/partitions, I am showing the numbers in |
784 | * hex - the same format as the root= option takes. | |
def4e38d | 785 | */ |
e559f58d CH |
786 | rcu_read_lock(); |
787 | xa_for_each(&disk->part_tbl, idx, part) { | |
788 | if (!bdev_nr_sectors(part)) | |
789 | continue; | |
a9e7bc3d | 790 | printk("%s%s %10llu %pg %s", |
e559f58d | 791 | bdev_is_partition(part) ? " " : "", |
ad1eaa53 | 792 | bdevt_str(part->bd_dev, devt_buf), |
a9e7bc3d | 793 | bdev_nr_sectors(part) >> 1, part, |
ad1eaa53 CH |
794 | part->bd_meta_info ? |
795 | part->bd_meta_info->uuid : ""); | |
e559f58d | 796 | if (bdev_is_partition(part)) |
074a7aca | 797 | printk("\n"); |
e559f58d CH |
798 | else if (dev->parent && dev->parent->driver) |
799 | printk(" driver: %s\n", | |
800 | dev->parent->driver->name); | |
801 | else | |
802 | printk(" (driver?)\n"); | |
074a7aca | 803 | } |
e559f58d | 804 | rcu_read_unlock(); |
def4e38d TH |
805 | } |
806 | class_dev_iter_exit(&iter); | |
dd2a345f DG |
807 | } |
808 | ||
1da177e4 LT |
809 | #ifdef CONFIG_PROC_FS |
810 | /* iterator */ | |
def4e38d | 811 | static void *disk_seqf_start(struct seq_file *seqf, loff_t *pos) |
68c4d4a7 | 812 | { |
def4e38d TH |
813 | loff_t skip = *pos; |
814 | struct class_dev_iter *iter; | |
815 | struct device *dev; | |
68c4d4a7 | 816 | |
aeb3d3a8 | 817 | iter = kmalloc(sizeof(*iter), GFP_KERNEL); |
def4e38d TH |
818 | if (!iter) |
819 | return ERR_PTR(-ENOMEM); | |
820 | ||
821 | seqf->private = iter; | |
822 | class_dev_iter_init(iter, &block_class, NULL, &disk_type); | |
823 | do { | |
824 | dev = class_dev_iter_next(iter); | |
825 | if (!dev) | |
826 | return NULL; | |
827 | } while (skip--); | |
828 | ||
829 | return dev_to_disk(dev); | |
68c4d4a7 GKH |
830 | } |
831 | ||
def4e38d | 832 | static void *disk_seqf_next(struct seq_file *seqf, void *v, loff_t *pos) |
1da177e4 | 833 | { |
edfaa7c3 | 834 | struct device *dev; |
1da177e4 | 835 | |
def4e38d TH |
836 | (*pos)++; |
837 | dev = class_dev_iter_next(seqf->private); | |
2ac3cee5 | 838 | if (dev) |
68c4d4a7 | 839 | return dev_to_disk(dev); |
2ac3cee5 | 840 | |
1da177e4 LT |
841 | return NULL; |
842 | } | |
843 | ||
def4e38d | 844 | static void disk_seqf_stop(struct seq_file *seqf, void *v) |
27f30251 | 845 | { |
def4e38d | 846 | struct class_dev_iter *iter = seqf->private; |
27f30251 | 847 | |
def4e38d TH |
848 | /* stop is called even after start failed :-( */ |
849 | if (iter) { | |
850 | class_dev_iter_exit(iter); | |
851 | kfree(iter); | |
77da1605 | 852 | seqf->private = NULL; |
5c0ef6d0 | 853 | } |
1da177e4 LT |
854 | } |
855 | ||
def4e38d | 856 | static void *show_partition_start(struct seq_file *seqf, loff_t *pos) |
1da177e4 | 857 | { |
06768067 | 858 | void *p; |
def4e38d TH |
859 | |
860 | p = disk_seqf_start(seqf, pos); | |
b9f985b6 | 861 | if (!IS_ERR_OR_NULL(p) && !*pos) |
def4e38d TH |
862 | seq_puts(seqf, "major minor #blocks name\n\n"); |
863 | return p; | |
1da177e4 LT |
864 | } |
865 | ||
cf771cb5 | 866 | static int show_partition(struct seq_file *seqf, void *v) |
1da177e4 LT |
867 | { |
868 | struct gendisk *sgp = v; | |
ad1eaa53 | 869 | struct block_device *part; |
ecc75a98 | 870 | unsigned long idx; |
1da177e4 | 871 | |
3b5149ac | 872 | if (!get_capacity(sgp) || (sgp->flags & GENHD_FL_HIDDEN)) |
1da177e4 LT |
873 | return 0; |
874 | ||
ecc75a98 CH |
875 | rcu_read_lock(); |
876 | xa_for_each(&sgp->part_tbl, idx, part) { | |
877 | if (!bdev_nr_sectors(part)) | |
878 | continue; | |
a291bb43 | 879 | seq_printf(seqf, "%4d %7d %10llu %pg\n", |
ad1eaa53 | 880 | MAJOR(part->bd_dev), MINOR(part->bd_dev), |
a291bb43 | 881 | bdev_nr_sectors(part) >> 1, part); |
ecc75a98 CH |
882 | } |
883 | rcu_read_unlock(); | |
1da177e4 LT |
884 | return 0; |
885 | } | |
886 | ||
f500975a | 887 | static const struct seq_operations partitions_op = { |
def4e38d TH |
888 | .start = show_partition_start, |
889 | .next = disk_seqf_next, | |
890 | .stop = disk_seqf_stop, | |
edfaa7c3 | 891 | .show = show_partition |
1da177e4 LT |
892 | }; |
893 | #endif | |
894 | ||
1da177e4 LT |
895 | static int __init genhd_device_init(void) |
896 | { | |
e105b8bf DW |
897 | int error; |
898 | ||
e105b8bf | 899 | error = class_register(&block_class); |
ee27a558 RM |
900 | if (unlikely(error)) |
901 | return error; | |
1da177e4 | 902 | blk_dev_init(); |
edfaa7c3 | 903 | |
561ec68e ZY |
904 | register_blkdev(BLOCK_EXT_MAJOR, "blkext"); |
905 | ||
edfaa7c3 | 906 | /* create top-level block dir */ |
721da5ce | 907 | block_depr = kobject_create_and_add("block", NULL); |
830d3cfb | 908 | return 0; |
1da177e4 LT |
909 | } |
910 | ||
911 | subsys_initcall(genhd_device_init); | |
912 | ||
edfaa7c3 KS |
913 | static ssize_t disk_range_show(struct device *dev, |
914 | struct device_attribute *attr, char *buf) | |
1da177e4 | 915 | { |
edfaa7c3 | 916 | struct gendisk *disk = dev_to_disk(dev); |
1da177e4 | 917 | |
edfaa7c3 | 918 | return sprintf(buf, "%d\n", disk->minors); |
1da177e4 LT |
919 | } |
920 | ||
1f014290 TH |
921 | static ssize_t disk_ext_range_show(struct device *dev, |
922 | struct device_attribute *attr, char *buf) | |
923 | { | |
924 | struct gendisk *disk = dev_to_disk(dev); | |
925 | ||
1ebe2e5f CH |
926 | return sprintf(buf, "%d\n", |
927 | (disk->flags & GENHD_FL_NO_PART) ? 1 : DISK_MAX_PARTS); | |
1f014290 TH |
928 | } |
929 | ||
edfaa7c3 KS |
930 | static ssize_t disk_removable_show(struct device *dev, |
931 | struct device_attribute *attr, char *buf) | |
a7fd6706 | 932 | { |
edfaa7c3 | 933 | struct gendisk *disk = dev_to_disk(dev); |
a7fd6706 | 934 | |
edfaa7c3 KS |
935 | return sprintf(buf, "%d\n", |
936 | (disk->flags & GENHD_FL_REMOVABLE ? 1 : 0)); | |
a7fd6706 KS |
937 | } |
938 | ||
8ddcd653 CH |
939 | static ssize_t disk_hidden_show(struct device *dev, |
940 | struct device_attribute *attr, char *buf) | |
941 | { | |
942 | struct gendisk *disk = dev_to_disk(dev); | |
943 | ||
944 | return sprintf(buf, "%d\n", | |
945 | (disk->flags & GENHD_FL_HIDDEN ? 1 : 0)); | |
946 | } | |
947 | ||
1c9ce527 KS |
948 | static ssize_t disk_ro_show(struct device *dev, |
949 | struct device_attribute *attr, char *buf) | |
950 | { | |
951 | struct gendisk *disk = dev_to_disk(dev); | |
952 | ||
b7db9956 | 953 | return sprintf(buf, "%d\n", get_disk_ro(disk) ? 1 : 0); |
1c9ce527 KS |
954 | } |
955 | ||
3ad5cee5 CH |
956 | ssize_t part_size_show(struct device *dev, |
957 | struct device_attribute *attr, char *buf) | |
958 | { | |
0d02129e | 959 | return sprintf(buf, "%llu\n", bdev_nr_sectors(dev_to_bdev(dev))); |
3ad5cee5 CH |
960 | } |
961 | ||
962 | ssize_t part_stat_show(struct device *dev, | |
963 | struct device_attribute *attr, char *buf) | |
964 | { | |
0d02129e | 965 | struct block_device *bdev = dev_to_bdev(dev); |
ed6cddef | 966 | struct request_queue *q = bdev_get_queue(bdev); |
ea18e0f0 | 967 | struct disk_stats stat; |
3ad5cee5 CH |
968 | unsigned int inflight; |
969 | ||
b2f609e1 | 970 | if (queue_is_mq(q)) |
0d02129e | 971 | inflight = blk_mq_in_flight(q, bdev); |
b2f609e1 | 972 | else |
0d02129e | 973 | inflight = part_in_flight(bdev); |
ea18e0f0 | 974 | |
86d73312 ZW |
975 | if (inflight) { |
976 | part_stat_lock(); | |
977 | update_io_ticks(bdev, jiffies, true); | |
978 | part_stat_unlock(); | |
979 | } | |
980 | part_stat_read_all(bdev, &stat); | |
3ad5cee5 CH |
981 | return sprintf(buf, |
982 | "%8lu %8lu %8llu %8u " | |
983 | "%8lu %8lu %8llu %8u " | |
984 | "%8u %8u %8u " | |
985 | "%8lu %8lu %8llu %8u " | |
986 | "%8lu %8u" | |
987 | "\n", | |
ea18e0f0 KK |
988 | stat.ios[STAT_READ], |
989 | stat.merges[STAT_READ], | |
990 | (unsigned long long)stat.sectors[STAT_READ], | |
991 | (unsigned int)div_u64(stat.nsecs[STAT_READ], NSEC_PER_MSEC), | |
992 | stat.ios[STAT_WRITE], | |
993 | stat.merges[STAT_WRITE], | |
994 | (unsigned long long)stat.sectors[STAT_WRITE], | |
995 | (unsigned int)div_u64(stat.nsecs[STAT_WRITE], NSEC_PER_MSEC), | |
3ad5cee5 | 996 | inflight, |
ea18e0f0 | 997 | jiffies_to_msecs(stat.io_ticks), |
8cd5b8fc KK |
998 | (unsigned int)div_u64(stat.nsecs[STAT_READ] + |
999 | stat.nsecs[STAT_WRITE] + | |
1000 | stat.nsecs[STAT_DISCARD] + | |
1001 | stat.nsecs[STAT_FLUSH], | |
1002 | NSEC_PER_MSEC), | |
ea18e0f0 KK |
1003 | stat.ios[STAT_DISCARD], |
1004 | stat.merges[STAT_DISCARD], | |
1005 | (unsigned long long)stat.sectors[STAT_DISCARD], | |
1006 | (unsigned int)div_u64(stat.nsecs[STAT_DISCARD], NSEC_PER_MSEC), | |
1007 | stat.ios[STAT_FLUSH], | |
1008 | (unsigned int)div_u64(stat.nsecs[STAT_FLUSH], NSEC_PER_MSEC)); | |
3ad5cee5 CH |
1009 | } |
1010 | ||
1011 | ssize_t part_inflight_show(struct device *dev, struct device_attribute *attr, | |
1012 | char *buf) | |
1013 | { | |
0d02129e | 1014 | struct block_device *bdev = dev_to_bdev(dev); |
ed6cddef | 1015 | struct request_queue *q = bdev_get_queue(bdev); |
3ad5cee5 CH |
1016 | unsigned int inflight[2]; |
1017 | ||
b2f609e1 | 1018 | if (queue_is_mq(q)) |
0d02129e | 1019 | blk_mq_in_flight_rw(q, bdev, inflight); |
b2f609e1 | 1020 | else |
0d02129e | 1021 | part_in_flight_rw(bdev, inflight); |
b2f609e1 | 1022 | |
3ad5cee5 CH |
1023 | return sprintf(buf, "%8u %8u\n", inflight[0], inflight[1]); |
1024 | } | |
1025 | ||
edfaa7c3 KS |
1026 | static ssize_t disk_capability_show(struct device *dev, |
1027 | struct device_attribute *attr, char *buf) | |
86ce18d7 | 1028 | { |
e81cd5a9 CH |
1029 | dev_warn_once(dev, "the capability attribute has been deprecated.\n"); |
1030 | return sprintf(buf, "0\n"); | |
86ce18d7 | 1031 | } |
edfaa7c3 | 1032 | |
c72758f3 MP |
1033 | static ssize_t disk_alignment_offset_show(struct device *dev, |
1034 | struct device_attribute *attr, | |
1035 | char *buf) | |
1036 | { | |
1037 | struct gendisk *disk = dev_to_disk(dev); | |
1038 | ||
640f2a23 | 1039 | return sprintf(buf, "%d\n", bdev_alignment_offset(disk->part0)); |
c72758f3 MP |
1040 | } |
1041 | ||
86b37281 MP |
1042 | static ssize_t disk_discard_alignment_show(struct device *dev, |
1043 | struct device_attribute *attr, | |
1044 | char *buf) | |
1045 | { | |
1046 | struct gendisk *disk = dev_to_disk(dev); | |
1047 | ||
4e1462ff | 1048 | return sprintf(buf, "%d\n", bdev_alignment_offset(disk->part0)); |
86b37281 MP |
1049 | } |
1050 | ||
13927b31 MC |
1051 | static ssize_t diskseq_show(struct device *dev, |
1052 | struct device_attribute *attr, char *buf) | |
1053 | { | |
1054 | struct gendisk *disk = dev_to_disk(dev); | |
1055 | ||
1056 | return sprintf(buf, "%llu\n", disk->diskseq); | |
1057 | } | |
1058 | ||
5657a819 JP |
1059 | static DEVICE_ATTR(range, 0444, disk_range_show, NULL); |
1060 | static DEVICE_ATTR(ext_range, 0444, disk_ext_range_show, NULL); | |
1061 | static DEVICE_ATTR(removable, 0444, disk_removable_show, NULL); | |
1062 | static DEVICE_ATTR(hidden, 0444, disk_hidden_show, NULL); | |
1063 | static DEVICE_ATTR(ro, 0444, disk_ro_show, NULL); | |
1064 | static DEVICE_ATTR(size, 0444, part_size_show, NULL); | |
1065 | static DEVICE_ATTR(alignment_offset, 0444, disk_alignment_offset_show, NULL); | |
1066 | static DEVICE_ATTR(discard_alignment, 0444, disk_discard_alignment_show, NULL); | |
1067 | static DEVICE_ATTR(capability, 0444, disk_capability_show, NULL); | |
1068 | static DEVICE_ATTR(stat, 0444, part_stat_show, NULL); | |
1069 | static DEVICE_ATTR(inflight, 0444, part_inflight_show, NULL); | |
1070 | static DEVICE_ATTR(badblocks, 0644, disk_badblocks_show, disk_badblocks_store); | |
13927b31 | 1071 | static DEVICE_ATTR(diskseq, 0444, diskseq_show, NULL); |
3ad5cee5 | 1072 | |
c17bb495 | 1073 | #ifdef CONFIG_FAIL_MAKE_REQUEST |
3ad5cee5 CH |
1074 | ssize_t part_fail_show(struct device *dev, |
1075 | struct device_attribute *attr, char *buf) | |
1076 | { | |
0d02129e | 1077 | return sprintf(buf, "%d\n", dev_to_bdev(dev)->bd_make_it_fail); |
3ad5cee5 CH |
1078 | } |
1079 | ||
1080 | ssize_t part_fail_store(struct device *dev, | |
1081 | struct device_attribute *attr, | |
1082 | const char *buf, size_t count) | |
1083 | { | |
3ad5cee5 CH |
1084 | int i; |
1085 | ||
1086 | if (count > 0 && sscanf(buf, "%d", &i) > 0) | |
0d02129e | 1087 | dev_to_bdev(dev)->bd_make_it_fail = i; |
3ad5cee5 CH |
1088 | |
1089 | return count; | |
1090 | } | |
1091 | ||
edfaa7c3 | 1092 | static struct device_attribute dev_attr_fail = |
5657a819 | 1093 | __ATTR(make-it-fail, 0644, part_fail_show, part_fail_store); |
3ad5cee5 CH |
1094 | #endif /* CONFIG_FAIL_MAKE_REQUEST */ |
1095 | ||
581d4e28 JA |
1096 | #ifdef CONFIG_FAIL_IO_TIMEOUT |
1097 | static struct device_attribute dev_attr_fail_timeout = | |
5657a819 | 1098 | __ATTR(io-timeout-fail, 0644, part_timeout_show, part_timeout_store); |
581d4e28 | 1099 | #endif |
edfaa7c3 KS |
1100 | |
1101 | static struct attribute *disk_attrs[] = { | |
1102 | &dev_attr_range.attr, | |
1f014290 | 1103 | &dev_attr_ext_range.attr, |
edfaa7c3 | 1104 | &dev_attr_removable.attr, |
8ddcd653 | 1105 | &dev_attr_hidden.attr, |
1c9ce527 | 1106 | &dev_attr_ro.attr, |
edfaa7c3 | 1107 | &dev_attr_size.attr, |
c72758f3 | 1108 | &dev_attr_alignment_offset.attr, |
86b37281 | 1109 | &dev_attr_discard_alignment.attr, |
edfaa7c3 KS |
1110 | &dev_attr_capability.attr, |
1111 | &dev_attr_stat.attr, | |
316d315b | 1112 | &dev_attr_inflight.attr, |
99e6608c | 1113 | &dev_attr_badblocks.attr, |
2bc8cda5 CH |
1114 | &dev_attr_events.attr, |
1115 | &dev_attr_events_async.attr, | |
1116 | &dev_attr_events_poll_msecs.attr, | |
13927b31 | 1117 | &dev_attr_diskseq.attr, |
edfaa7c3 KS |
1118 | #ifdef CONFIG_FAIL_MAKE_REQUEST |
1119 | &dev_attr_fail.attr, | |
581d4e28 JA |
1120 | #endif |
1121 | #ifdef CONFIG_FAIL_IO_TIMEOUT | |
1122 | &dev_attr_fail_timeout.attr, | |
edfaa7c3 KS |
1123 | #endif |
1124 | NULL | |
1125 | }; | |
1126 | ||
9438b3e0 DW |
1127 | static umode_t disk_visible(struct kobject *kobj, struct attribute *a, int n) |
1128 | { | |
1129 | struct device *dev = container_of(kobj, typeof(*dev), kobj); | |
1130 | struct gendisk *disk = dev_to_disk(dev); | |
1131 | ||
1132 | if (a == &dev_attr_badblocks.attr && !disk->bb) | |
1133 | return 0; | |
1134 | return a->mode; | |
1135 | } | |
1136 | ||
edfaa7c3 KS |
1137 | static struct attribute_group disk_attr_group = { |
1138 | .attrs = disk_attrs, | |
9438b3e0 | 1139 | .is_visible = disk_visible, |
edfaa7c3 KS |
1140 | }; |
1141 | ||
a4dbd674 | 1142 | static const struct attribute_group *disk_attr_groups[] = { |
edfaa7c3 | 1143 | &disk_attr_group, |
cc5c516d CH |
1144 | #ifdef CONFIG_BLK_DEV_IO_TRACE |
1145 | &blk_trace_attr_group, | |
ff53cd52 TW |
1146 | #endif |
1147 | #ifdef CONFIG_BLK_DEV_INTEGRITY | |
1148 | &blk_integrity_attr_group, | |
cc5c516d | 1149 | #endif |
edfaa7c3 | 1150 | NULL |
1da177e4 LT |
1151 | }; |
1152 | ||
b5bd357c LC |
1153 | /** |
1154 | * disk_release - releases all allocated resources of the gendisk | |
1155 | * @dev: the device representing this disk | |
1156 | * | |
1157 | * This function releases all allocated resources of the gendisk. | |
1158 | * | |
b5bd357c LC |
1159 | * Drivers which used __device_add_disk() have a gendisk with a request_queue |
1160 | * assigned. Since the request_queue sits on top of the gendisk for these | |
1161 | * drivers we also call blk_put_queue() for them, and we expect the | |
1162 | * request_queue refcount to reach 0 at this point, and so the request_queue | |
1163 | * will also be freed prior to the disk. | |
e8c7d14a LC |
1164 | * |
1165 | * Context: can sleep | |
b5bd357c | 1166 | */ |
edfaa7c3 | 1167 | static void disk_release(struct device *dev) |
1da177e4 | 1168 | { |
edfaa7c3 KS |
1169 | struct gendisk *disk = dev_to_disk(dev); |
1170 | ||
e8c7d14a | 1171 | might_sleep(); |
a2041761 | 1172 | WARN_ON_ONCE(disk_live(disk)); |
e8c7d14a | 1173 | |
c5db2cfc CH |
1174 | /* |
1175 | * To undo the all initialization from blk_mq_init_allocated_queue in | |
1176 | * case of a probe failure where add_disk is never called we have to | |
1177 | * call blk_mq_exit_queue here. We can't do this for the more common | |
1178 | * teardown case (yet) as the tagset can be gone by the time the disk | |
1179 | * is released once it was added. | |
1180 | */ | |
1181 | if (queue_is_mq(disk->queue) && | |
1182 | test_bit(GD_OWNS_QUEUE, &disk->state) && | |
1183 | !test_bit(GD_ADDED, &disk->state)) | |
1184 | blk_mq_exit_queue(disk->queue); | |
1185 | ||
b6553bef CH |
1186 | blkcg_exit_disk(disk); |
1187 | ||
46754bd0 | 1188 | bioset_exit(&disk->bio_split); |
2a19b28f | 1189 | |
77ea887e | 1190 | disk_release_events(disk); |
1da177e4 | 1191 | kfree(disk->random); |
5d400665 | 1192 | disk_free_zone_bitmaps(disk); |
a33df75c | 1193 | xa_destroy(&disk->part_tbl); |
1059699f | 1194 | |
d152c682 | 1195 | disk->queue->disk = NULL; |
61a35cfc | 1196 | blk_put_queue(disk->queue); |
76792055 CH |
1197 | |
1198 | if (test_bit(GD_ADDED, &disk->state) && disk->fops->free_disk) | |
1199 | disk->fops->free_disk(disk); | |
1200 | ||
2f4731dc | 1201 | iput(disk->part0->bd_inode); /* frees the disk */ |
1da177e4 | 1202 | } |
87eb7107 | 1203 | |
23680f0b | 1204 | static int block_uevent(const struct device *dev, struct kobj_uevent_env *env) |
87eb7107 | 1205 | { |
23680f0b | 1206 | const struct gendisk *disk = dev_to_disk(dev); |
87eb7107 MC |
1207 | |
1208 | return add_uevent_var(env, "DISKSEQ=%llu", disk->diskseq); | |
1209 | } | |
1210 | ||
edfaa7c3 KS |
1211 | struct class block_class = { |
1212 | .name = "block", | |
87eb7107 | 1213 | .dev_uevent = block_uevent, |
1da177e4 LT |
1214 | }; |
1215 | ||
a9b12f8b | 1216 | static char *block_devnode(const struct device *dev, umode_t *mode, |
050a4f34 JA |
1217 | kuid_t *uid, kgid_t *gid) |
1218 | { | |
1219 | struct gendisk *disk = dev_to_disk(dev); | |
1220 | ||
1221 | if (disk->fops->devnode) | |
1222 | return disk->fops->devnode(disk, mode); | |
1223 | return NULL; | |
1224 | } | |
1225 | ||
ef45fe47 | 1226 | const struct device_type disk_type = { |
edfaa7c3 KS |
1227 | .name = "disk", |
1228 | .groups = disk_attr_groups, | |
1229 | .release = disk_release, | |
050a4f34 | 1230 | .devnode = block_devnode, |
1da177e4 LT |
1231 | }; |
1232 | ||
a6e2ba88 | 1233 | #ifdef CONFIG_PROC_FS |
cf771cb5 TH |
1234 | /* |
1235 | * aggregate disk stat collector. Uses the same stats that the sysfs | |
1236 | * entries do, above, but makes them available through one seq_file. | |
1237 | * | |
1238 | * The output looks suspiciously like /proc/partitions with a bunch of | |
1239 | * extra fields. | |
1240 | */ | |
1241 | static int diskstats_show(struct seq_file *seqf, void *v) | |
1da177e4 LT |
1242 | { |
1243 | struct gendisk *gp = v; | |
ad1eaa53 | 1244 | struct block_device *hd; |
e016b782 | 1245 | unsigned int inflight; |
ea18e0f0 | 1246 | struct disk_stats stat; |
7fae67cc | 1247 | unsigned long idx; |
1da177e4 LT |
1248 | |
1249 | /* | |
ed9e1982 | 1250 | if (&disk_to_dev(gp)->kobj.entry == block_class.devices.next) |
cf771cb5 | 1251 | seq_puts(seqf, "major minor name" |
1da177e4 LT |
1252 | " rio rmerge rsect ruse wio wmerge " |
1253 | "wsect wuse running use aveq" | |
1254 | "\n\n"); | |
1255 | */ | |
9f5e4865 | 1256 | |
7fae67cc CH |
1257 | rcu_read_lock(); |
1258 | xa_for_each(&gp->part_tbl, idx, hd) { | |
1259 | if (bdev_is_partition(hd) && !bdev_nr_sectors(hd)) | |
1260 | continue; | |
b2f609e1 | 1261 | if (queue_is_mq(gp->queue)) |
ad1eaa53 | 1262 | inflight = blk_mq_in_flight(gp->queue, hd); |
b2f609e1 | 1263 | else |
ad1eaa53 | 1264 | inflight = part_in_flight(hd); |
ea18e0f0 | 1265 | |
86d73312 ZW |
1266 | if (inflight) { |
1267 | part_stat_lock(); | |
1268 | update_io_ticks(hd, jiffies, true); | |
1269 | part_stat_unlock(); | |
1270 | } | |
1271 | part_stat_read_all(hd, &stat); | |
26e2d7a3 | 1272 | seq_printf(seqf, "%4d %7d %pg " |
bdca3c87 MC |
1273 | "%lu %lu %lu %u " |
1274 | "%lu %lu %lu %u " | |
1275 | "%u %u %u " | |
b6866318 KK |
1276 | "%lu %lu %lu %u " |
1277 | "%lu %u" | |
1278 | "\n", | |
26e2d7a3 | 1279 | MAJOR(hd->bd_dev), MINOR(hd->bd_dev), hd, |
ea18e0f0 KK |
1280 | stat.ios[STAT_READ], |
1281 | stat.merges[STAT_READ], | |
1282 | stat.sectors[STAT_READ], | |
1283 | (unsigned int)div_u64(stat.nsecs[STAT_READ], | |
1284 | NSEC_PER_MSEC), | |
1285 | stat.ios[STAT_WRITE], | |
1286 | stat.merges[STAT_WRITE], | |
1287 | stat.sectors[STAT_WRITE], | |
1288 | (unsigned int)div_u64(stat.nsecs[STAT_WRITE], | |
1289 | NSEC_PER_MSEC), | |
e016b782 | 1290 | inflight, |
ea18e0f0 | 1291 | jiffies_to_msecs(stat.io_ticks), |
8cd5b8fc KK |
1292 | (unsigned int)div_u64(stat.nsecs[STAT_READ] + |
1293 | stat.nsecs[STAT_WRITE] + | |
1294 | stat.nsecs[STAT_DISCARD] + | |
1295 | stat.nsecs[STAT_FLUSH], | |
1296 | NSEC_PER_MSEC), | |
ea18e0f0 KK |
1297 | stat.ios[STAT_DISCARD], |
1298 | stat.merges[STAT_DISCARD], | |
1299 | stat.sectors[STAT_DISCARD], | |
1300 | (unsigned int)div_u64(stat.nsecs[STAT_DISCARD], | |
1301 | NSEC_PER_MSEC), | |
1302 | stat.ios[STAT_FLUSH], | |
1303 | (unsigned int)div_u64(stat.nsecs[STAT_FLUSH], | |
1304 | NSEC_PER_MSEC) | |
28f39d55 | 1305 | ); |
1da177e4 | 1306 | } |
7fae67cc | 1307 | rcu_read_unlock(); |
9f5e4865 | 1308 | |
1da177e4 LT |
1309 | return 0; |
1310 | } | |
1311 | ||
31d85ab2 | 1312 | static const struct seq_operations diskstats_op = { |
def4e38d TH |
1313 | .start = disk_seqf_start, |
1314 | .next = disk_seqf_next, | |
1315 | .stop = disk_seqf_stop, | |
1da177e4 LT |
1316 | .show = diskstats_show |
1317 | }; | |
f500975a AD |
1318 | |
1319 | static int __init proc_genhd_init(void) | |
1320 | { | |
fddda2b7 CH |
1321 | proc_create_seq("diskstats", 0, NULL, &diskstats_op); |
1322 | proc_create_seq("partitions", 0, NULL, &partitions_op); | |
f500975a AD |
1323 | return 0; |
1324 | } | |
1325 | module_init(proc_genhd_init); | |
a6e2ba88 | 1326 | #endif /* CONFIG_PROC_FS */ |
1da177e4 | 1327 | |
c97d93c3 CH |
1328 | dev_t part_devt(struct gendisk *disk, u8 partno) |
1329 | { | |
0e0ccdec | 1330 | struct block_device *part; |
c97d93c3 CH |
1331 | dev_t devt = 0; |
1332 | ||
0e0ccdec CH |
1333 | rcu_read_lock(); |
1334 | part = xa_load(&disk->part_tbl, partno); | |
1335 | if (part) | |
c97d93c3 | 1336 | devt = part->bd_dev; |
0e0ccdec | 1337 | rcu_read_unlock(); |
c97d93c3 CH |
1338 | |
1339 | return devt; | |
1340 | } | |
1341 | ||
cf771cb5 | 1342 | dev_t blk_lookup_devt(const char *name, int partno) |
a142be85 | 1343 | { |
def4e38d TH |
1344 | dev_t devt = MKDEV(0, 0); |
1345 | struct class_dev_iter iter; | |
1346 | struct device *dev; | |
a142be85 | 1347 | |
def4e38d TH |
1348 | class_dev_iter_init(&iter, &block_class, NULL, &disk_type); |
1349 | while ((dev = class_dev_iter_next(&iter))) { | |
a142be85 | 1350 | struct gendisk *disk = dev_to_disk(dev); |
a142be85 | 1351 | |
3ada8b7e | 1352 | if (strcmp(dev_name(dev), name)) |
f331c029 | 1353 | continue; |
f331c029 | 1354 | |
41b8c853 NB |
1355 | if (partno < disk->minors) { |
1356 | /* We need to return the right devno, even | |
1357 | * if the partition doesn't exist yet. | |
1358 | */ | |
1359 | devt = MKDEV(MAJOR(dev->devt), | |
1360 | MINOR(dev->devt) + partno); | |
c97d93c3 CH |
1361 | } else { |
1362 | devt = part_devt(disk, partno); | |
1363 | if (devt) | |
1364 | break; | |
def4e38d | 1365 | } |
5c0ef6d0 | 1366 | } |
def4e38d | 1367 | class_dev_iter_exit(&iter); |
edfaa7c3 KS |
1368 | return devt; |
1369 | } | |
edfaa7c3 | 1370 | |
4a1fa41d CH |
1371 | struct gendisk *__alloc_disk_node(struct request_queue *q, int node_id, |
1372 | struct lock_class_key *lkclass) | |
1946089a CL |
1373 | { |
1374 | struct gendisk *disk; | |
1375 | ||
c1b511eb | 1376 | disk = kzalloc_node(sizeof(struct gendisk), GFP_KERNEL, node_id); |
f93af2a4 | 1377 | if (!disk) |
aa0c680c | 1378 | return NULL; |
6c23a968 | 1379 | |
46754bd0 CH |
1380 | if (bioset_init(&disk->bio_split, BIO_POOL_SIZE, 0, 0)) |
1381 | goto out_free_disk; | |
1382 | ||
edb0872f CH |
1383 | disk->bdi = bdi_alloc(node_id); |
1384 | if (!disk->bdi) | |
46754bd0 | 1385 | goto out_free_bioset; |
edb0872f | 1386 | |
17220ca5 PB |
1387 | /* bdev_alloc() might need the queue, set before the first call */ |
1388 | disk->queue = q; | |
1389 | ||
cb8432d6 CH |
1390 | disk->part0 = bdev_alloc(disk, 0); |
1391 | if (!disk->part0) | |
edb0872f | 1392 | goto out_free_bdi; |
22ae8ce8 | 1393 | |
f93af2a4 | 1394 | disk->node_id = node_id; |
a8698707 | 1395 | mutex_init(&disk->open_mutex); |
a33df75c CH |
1396 | xa_init(&disk->part_tbl); |
1397 | if (xa_insert(&disk->part_tbl, 0, disk->part0, GFP_KERNEL)) | |
1398 | goto out_destroy_part_tbl; | |
f93af2a4 | 1399 | |
b6553bef CH |
1400 | if (blkcg_init_disk(disk)) |
1401 | goto out_erase_part0; | |
1402 | ||
f93af2a4 CH |
1403 | rand_initialize_disk(disk); |
1404 | disk_to_dev(disk)->class = &block_class; | |
1405 | disk_to_dev(disk)->type = &disk_type; | |
1406 | device_initialize(disk_to_dev(disk)); | |
cf179948 | 1407 | inc_diskseq(disk); |
d152c682 | 1408 | q->disk = disk; |
4dcc4874 | 1409 | lockdep_init_map(&disk->lockdep_map, "(bio completion)", lkclass, 0); |
0dbcfe24 CH |
1410 | #ifdef CONFIG_BLOCK_HOLDER_DEPRECATED |
1411 | INIT_LIST_HEAD(&disk->slave_bdevs); | |
1412 | #endif | |
1da177e4 | 1413 | return disk; |
f93af2a4 | 1414 | |
b6553bef CH |
1415 | out_erase_part0: |
1416 | xa_erase(&disk->part_tbl, 0); | |
a33df75c CH |
1417 | out_destroy_part_tbl: |
1418 | xa_destroy(&disk->part_tbl); | |
06cc978d | 1419 | disk->part0->bd_disk = NULL; |
2f4731dc | 1420 | iput(disk->part0->bd_inode); |
edb0872f CH |
1421 | out_free_bdi: |
1422 | bdi_put(disk->bdi); | |
46754bd0 CH |
1423 | out_free_bioset: |
1424 | bioset_exit(&disk->bio_split); | |
f93af2a4 CH |
1425 | out_free_disk: |
1426 | kfree(disk); | |
1427 | return NULL; | |
1da177e4 | 1428 | } |
1da177e4 | 1429 | |
4dcc4874 | 1430 | struct gendisk *__blk_alloc_disk(int node, struct lock_class_key *lkclass) |
f525464a CH |
1431 | { |
1432 | struct request_queue *q; | |
1433 | struct gendisk *disk; | |
1434 | ||
80bd4a7a | 1435 | q = blk_alloc_queue(node); |
f525464a CH |
1436 | if (!q) |
1437 | return NULL; | |
1438 | ||
4a1fa41d | 1439 | disk = __alloc_disk_node(q, node, lkclass); |
f525464a | 1440 | if (!disk) { |
6f8191fd | 1441 | blk_put_queue(q); |
f525464a CH |
1442 | return NULL; |
1443 | } | |
6f8191fd | 1444 | set_bit(GD_OWNS_QUEUE, &disk->state); |
f525464a CH |
1445 | return disk; |
1446 | } | |
1447 | EXPORT_SYMBOL(__blk_alloc_disk); | |
1448 | ||
b5bd357c LC |
1449 | /** |
1450 | * put_disk - decrements the gendisk refcount | |
0d20dcc2 | 1451 | * @disk: the struct gendisk to decrement the refcount for |
b5bd357c LC |
1452 | * |
1453 | * This decrements the refcount for the struct gendisk. When this reaches 0 | |
1454 | * we'll have disk_release() called. | |
e8c7d14a | 1455 | * |
c5db2cfc CH |
1456 | * Note: for blk-mq disk put_disk must be called before freeing the tag_set |
1457 | * when handling probe errors (that is before add_disk() is called). | |
1458 | * | |
e8c7d14a LC |
1459 | * Context: Any context, but the last reference must not be dropped from |
1460 | * atomic context. | |
b5bd357c | 1461 | */ |
1da177e4 LT |
1462 | void put_disk(struct gendisk *disk) |
1463 | { | |
1464 | if (disk) | |
efdc41c8 | 1465 | put_device(disk_to_dev(disk)); |
1da177e4 | 1466 | } |
1da177e4 LT |
1467 | EXPORT_SYMBOL(put_disk); |
1468 | ||
e3264a4d HR |
1469 | static void set_disk_ro_uevent(struct gendisk *gd, int ro) |
1470 | { | |
1471 | char event[] = "DISK_RO=1"; | |
1472 | char *envp[] = { event, NULL }; | |
1473 | ||
1474 | if (!ro) | |
1475 | event[8] = '0'; | |
1476 | kobject_uevent_env(&disk_to_dev(gd)->kobj, KOBJ_CHANGE, envp); | |
1477 | } | |
1478 | ||
52f019d4 CH |
1479 | /** |
1480 | * set_disk_ro - set a gendisk read-only | |
1481 | * @disk: gendisk to operate on | |
7f31bee3 | 1482 | * @read_only: %true to set the disk read-only, %false set the disk read/write |
52f019d4 CH |
1483 | * |
1484 | * This function is used to indicate whether a given disk device should have its | |
1485 | * read-only flag set. set_disk_ro() is typically used by device drivers to | |
1486 | * indicate whether the underlying physical device is write-protected. | |
1487 | */ | |
1488 | void set_disk_ro(struct gendisk *disk, bool read_only) | |
1da177e4 | 1489 | { |
52f019d4 CH |
1490 | if (read_only) { |
1491 | if (test_and_set_bit(GD_READ_ONLY, &disk->state)) | |
1492 | return; | |
1493 | } else { | |
1494 | if (!test_and_clear_bit(GD_READ_ONLY, &disk->state)) | |
1495 | return; | |
e3264a4d | 1496 | } |
52f019d4 | 1497 | set_disk_ro_uevent(disk, read_only); |
1da177e4 | 1498 | } |
1da177e4 LT |
1499 | EXPORT_SYMBOL(set_disk_ro); |
1500 | ||
cf179948 MC |
1501 | void inc_diskseq(struct gendisk *disk) |
1502 | { | |
1503 | disk->diskseq = atomic64_inc_return(&diskseq); | |
1504 | } |