5 #include <linux/module.h>
7 #include <linux/genhd.h>
8 #include <linux/kdev_t.h>
9 #include <linux/kernel.h>
10 #include <linux/blkdev.h>
11 #include <linux/init.h>
12 #include <linux/spinlock.h>
13 #include <linux/seq_file.h>
14 #include <linux/slab.h>
15 #include <linux/kmod.h>
16 #include <linux/kobj_map.h>
17 #include <linux/buffer_head.h>
18 #include <linux/mutex.h>
20 struct subsystem block_subsys
;
21 static DEFINE_MUTEX(block_subsys_lock
);
24 * Can be deleted altogether. Later.
27 static struct blk_major_name
{
28 struct blk_major_name
*next
;
31 } *major_names
[BLKDEV_MAJOR_HASH_SIZE
];
33 /* index in the above - for now: assume no multimajor ranges */
34 static inline int major_to_index(int major
)
36 return major
% BLKDEV_MAJOR_HASH_SIZE
;
41 void blkdev_show(struct seq_file
*f
, off_t offset
)
43 struct blk_major_name
*dp
;
45 if (offset
< BLKDEV_MAJOR_HASH_SIZE
) {
46 mutex_lock(&block_subsys_lock
);
47 for (dp
= major_names
[offset
]; dp
; dp
= dp
->next
)
48 seq_printf(f
, "%3d %s\n", dp
->major
, dp
->name
);
49 mutex_unlock(&block_subsys_lock
);
53 #endif /* CONFIG_PROC_FS */
55 int register_blkdev(unsigned int major
, const char *name
)
57 struct blk_major_name
**n
, *p
;
60 mutex_lock(&block_subsys_lock
);
64 for (index
= ARRAY_SIZE(major_names
)-1; index
> 0; index
--) {
65 if (is_lanana_major(index
))
67 if (major_names
[index
] == NULL
)
72 printk("register_blkdev: failed to get major for %s\n",
81 p
= kmalloc(sizeof(struct blk_major_name
), GFP_KERNEL
);
88 strlcpy(p
->name
, name
, sizeof(p
->name
));
90 index
= major_to_index(major
);
92 for (n
= &major_names
[index
]; *n
; n
= &(*n
)->next
) {
93 if ((*n
)->major
== major
)
102 printk("register_blkdev: cannot get major %d for %s\n",
107 mutex_unlock(&block_subsys_lock
);
111 EXPORT_SYMBOL(register_blkdev
);
113 /* todo: make void - error printk here */
114 int unregister_blkdev(unsigned int major
, const char *name
)
116 struct blk_major_name
**n
;
117 struct blk_major_name
*p
= NULL
;
118 int index
= major_to_index(major
);
121 mutex_lock(&block_subsys_lock
);
122 for (n
= &major_names
[index
]; *n
; n
= &(*n
)->next
)
123 if ((*n
)->major
== major
)
125 if (!*n
|| strcmp((*n
)->name
, name
))
131 mutex_unlock(&block_subsys_lock
);
137 EXPORT_SYMBOL(unregister_blkdev
);
139 static struct kobj_map
*bdev_map
;
142 * Register device numbers dev..(dev+range-1)
143 * range must be nonzero
144 * The hash chain is sorted on range, so that subranges can override.
146 void blk_register_region(dev_t dev
, unsigned long range
, struct module
*module
,
147 struct kobject
*(*probe
)(dev_t
, int *, void *),
148 int (*lock
)(dev_t
, void *), void *data
)
150 kobj_map(bdev_map
, dev
, range
, module
, probe
, lock
, data
);
153 EXPORT_SYMBOL(blk_register_region
);
155 void blk_unregister_region(dev_t dev
, unsigned long range
)
157 kobj_unmap(bdev_map
, dev
, range
);
160 EXPORT_SYMBOL(blk_unregister_region
);
162 static struct kobject
*exact_match(dev_t dev
, int *part
, void *data
)
164 struct gendisk
*p
= data
;
168 static int exact_lock(dev_t dev
, void *data
)
170 struct gendisk
*p
= data
;
178 * add_disk - add partitioning information to kernel list
179 * @disk: per-device partitioning information
181 * This function registers the partitioning information in @disk
184 void add_disk(struct gendisk
*disk
)
186 disk
->flags
|= GENHD_FL_UP
;
187 blk_register_region(MKDEV(disk
->major
, disk
->first_minor
),
188 disk
->minors
, NULL
, exact_match
, exact_lock
, disk
);
190 blk_register_queue(disk
);
193 EXPORT_SYMBOL(add_disk
);
194 EXPORT_SYMBOL(del_gendisk
); /* in partitions/check.c */
196 void unlink_gendisk(struct gendisk
*disk
)
198 blk_unregister_queue(disk
);
199 blk_unregister_region(MKDEV(disk
->major
, disk
->first_minor
),
203 #define to_disk(obj) container_of(obj,struct gendisk,kobj)
206 * get_gendisk - get partitioning information for a given device
207 * @dev: device to get partitioning information for
209 * This function gets the structure containing partitioning
210 * information for the given device @dev.
212 struct gendisk
*get_gendisk(dev_t dev
, int *part
)
214 struct kobject
*kobj
= kobj_lookup(bdev_map
, dev
, part
);
215 return kobj
? to_disk(kobj
) : NULL
;
218 #ifdef CONFIG_PROC_FS
220 static void *part_start(struct seq_file
*part
, loff_t
*pos
)
225 mutex_lock(&block_subsys_lock
);
226 list_for_each(p
, &block_subsys
.kset
.list
)
228 return list_entry(p
, struct gendisk
, kobj
.entry
);
232 static void *part_next(struct seq_file
*part
, void *v
, loff_t
*pos
)
234 struct list_head
*p
= ((struct gendisk
*)v
)->kobj
.entry
.next
;
236 return p
==&block_subsys
.kset
.list
? NULL
:
237 list_entry(p
, struct gendisk
, kobj
.entry
);
240 static void part_stop(struct seq_file
*part
, void *v
)
242 mutex_unlock(&block_subsys_lock
);
245 static int show_partition(struct seq_file
*part
, void *v
)
247 struct gendisk
*sgp
= v
;
249 char buf
[BDEVNAME_SIZE
];
251 if (&sgp
->kobj
.entry
== block_subsys
.kset
.list
.next
)
252 seq_puts(part
, "major minor #blocks name\n\n");
254 /* Don't show non-partitionable removeable devices or empty devices */
255 if (!get_capacity(sgp
) ||
256 (sgp
->minors
== 1 && (sgp
->flags
& GENHD_FL_REMOVABLE
)))
258 if (sgp
->flags
& GENHD_FL_SUPPRESS_PARTITION_INFO
)
261 /* show the full disk and all non-0 size partitions of it */
262 seq_printf(part
, "%4d %4d %10llu %s\n",
263 sgp
->major
, sgp
->first_minor
,
264 (unsigned long long)get_capacity(sgp
) >> 1,
265 disk_name(sgp
, 0, buf
));
266 for (n
= 0; n
< sgp
->minors
- 1; n
++) {
269 if (sgp
->part
[n
]->nr_sects
== 0)
271 seq_printf(part
, "%4d %4d %10llu %s\n",
272 sgp
->major
, n
+ 1 + sgp
->first_minor
,
273 (unsigned long long)sgp
->part
[n
]->nr_sects
>> 1 ,
274 disk_name(sgp
, n
+ 1, buf
));
280 struct seq_operations partitions_op
= {
284 .show
= show_partition
289 extern int blk_dev_init(void);
291 static struct kobject
*base_probe(dev_t dev
, int *part
, void *data
)
293 if (request_module("block-major-%d-%d", MAJOR(dev
), MINOR(dev
)) > 0)
294 /* Make old-style 2.4 aliases work */
295 request_module("block-major-%d", MAJOR(dev
));
299 static int __init
genhd_device_init(void)
303 bdev_map
= kobj_map_init(base_probe
, &block_subsys_lock
);
305 err
= subsystem_register(&block_subsys
);
307 printk(KERN_WARNING
"%s: subsystem_register error: %d\n",
312 subsys_initcall(genhd_device_init
);
317 * kobject & sysfs bindings for block devices
319 static ssize_t
disk_attr_show(struct kobject
*kobj
, struct attribute
*attr
,
322 struct gendisk
*disk
= to_disk(kobj
);
323 struct disk_attribute
*disk_attr
=
324 container_of(attr
,struct disk_attribute
,attr
);
328 ret
= disk_attr
->show(disk
,page
);
332 static ssize_t
disk_attr_store(struct kobject
* kobj
, struct attribute
* attr
,
333 const char *page
, size_t count
)
335 struct gendisk
*disk
= to_disk(kobj
);
336 struct disk_attribute
*disk_attr
=
337 container_of(attr
,struct disk_attribute
,attr
);
340 if (disk_attr
->store
)
341 ret
= disk_attr
->store(disk
, page
, count
);
345 static struct sysfs_ops disk_sysfs_ops
= {
346 .show
= &disk_attr_show
,
347 .store
= &disk_attr_store
,
350 static ssize_t
disk_uevent_store(struct gendisk
* disk
,
351 const char *buf
, size_t count
)
353 kobject_uevent(&disk
->kobj
, KOBJ_ADD
);
356 static ssize_t
disk_dev_read(struct gendisk
* disk
, char *page
)
358 dev_t base
= MKDEV(disk
->major
, disk
->first_minor
);
359 return print_dev_t(page
, base
);
361 static ssize_t
disk_range_read(struct gendisk
* disk
, char *page
)
363 return sprintf(page
, "%d\n", disk
->minors
);
365 static ssize_t
disk_removable_read(struct gendisk
* disk
, char *page
)
367 return sprintf(page
, "%d\n",
368 (disk
->flags
& GENHD_FL_REMOVABLE
? 1 : 0));
371 static ssize_t
disk_size_read(struct gendisk
* disk
, char *page
)
373 return sprintf(page
, "%llu\n", (unsigned long long)get_capacity(disk
));
376 static ssize_t
disk_stats_read(struct gendisk
* disk
, char *page
)
379 disk_round_stats(disk
);
382 "%8lu %8lu %8llu %8u "
383 "%8lu %8lu %8llu %8u "
386 disk_stat_read(disk
, ios
[READ
]),
387 disk_stat_read(disk
, merges
[READ
]),
388 (unsigned long long)disk_stat_read(disk
, sectors
[READ
]),
389 jiffies_to_msecs(disk_stat_read(disk
, ticks
[READ
])),
390 disk_stat_read(disk
, ios
[WRITE
]),
391 disk_stat_read(disk
, merges
[WRITE
]),
392 (unsigned long long)disk_stat_read(disk
, sectors
[WRITE
]),
393 jiffies_to_msecs(disk_stat_read(disk
, ticks
[WRITE
])),
395 jiffies_to_msecs(disk_stat_read(disk
, io_ticks
)),
396 jiffies_to_msecs(disk_stat_read(disk
, time_in_queue
)));
398 static struct disk_attribute disk_attr_uevent
= {
399 .attr
= {.name
= "uevent", .mode
= S_IWUSR
},
400 .store
= disk_uevent_store
402 static struct disk_attribute disk_attr_dev
= {
403 .attr
= {.name
= "dev", .mode
= S_IRUGO
},
404 .show
= disk_dev_read
406 static struct disk_attribute disk_attr_range
= {
407 .attr
= {.name
= "range", .mode
= S_IRUGO
},
408 .show
= disk_range_read
410 static struct disk_attribute disk_attr_removable
= {
411 .attr
= {.name
= "removable", .mode
= S_IRUGO
},
412 .show
= disk_removable_read
414 static struct disk_attribute disk_attr_size
= {
415 .attr
= {.name
= "size", .mode
= S_IRUGO
},
416 .show
= disk_size_read
418 static struct disk_attribute disk_attr_stat
= {
419 .attr
= {.name
= "stat", .mode
= S_IRUGO
},
420 .show
= disk_stats_read
423 #ifdef CONFIG_FAIL_MAKE_REQUEST
425 static ssize_t
disk_fail_store(struct gendisk
* disk
,
426 const char *buf
, size_t count
)
430 if (count
> 0 && sscanf(buf
, "%d", &i
) > 0) {
432 disk
->flags
&= ~GENHD_FL_FAIL
;
434 disk
->flags
|= GENHD_FL_FAIL
;
439 static ssize_t
disk_fail_read(struct gendisk
* disk
, char *page
)
441 return sprintf(page
, "%d\n", disk
->flags
& GENHD_FL_FAIL
? 1 : 0);
443 static struct disk_attribute disk_attr_fail
= {
444 .attr
= {.name
= "make-it-fail", .mode
= S_IRUGO
| S_IWUSR
},
445 .store
= disk_fail_store
,
446 .show
= disk_fail_read
451 static struct attribute
* default_attrs
[] = {
452 &disk_attr_uevent
.attr
,
454 &disk_attr_range
.attr
,
455 &disk_attr_removable
.attr
,
456 &disk_attr_size
.attr
,
457 &disk_attr_stat
.attr
,
458 #ifdef CONFIG_FAIL_MAKE_REQUEST
459 &disk_attr_fail
.attr
,
464 static void disk_release(struct kobject
* kobj
)
466 struct gendisk
*disk
= to_disk(kobj
);
469 free_disk_stats(disk
);
473 static struct kobj_type ktype_block
= {
474 .release
= disk_release
,
475 .sysfs_ops
= &disk_sysfs_ops
,
476 .default_attrs
= default_attrs
,
479 extern struct kobj_type ktype_part
;
481 static int block_uevent_filter(struct kset
*kset
, struct kobject
*kobj
)
483 struct kobj_type
*ktype
= get_ktype(kobj
);
485 return ((ktype
== &ktype_block
) || (ktype
== &ktype_part
));
488 static int block_uevent(struct kset
*kset
, struct kobject
*kobj
, char **envp
,
489 int num_envp
, char *buffer
, int buffer_size
)
491 struct kobj_type
*ktype
= get_ktype(kobj
);
492 struct device
*physdev
;
493 struct gendisk
*disk
;
494 struct hd_struct
*part
;
498 if (ktype
== &ktype_block
) {
499 disk
= container_of(kobj
, struct gendisk
, kobj
);
500 add_uevent_var(envp
, num_envp
, &i
, buffer
, buffer_size
,
501 &length
, "MINOR=%u", disk
->first_minor
);
502 } else if (ktype
== &ktype_part
) {
503 disk
= container_of(kobj
->parent
, struct gendisk
, kobj
);
504 part
= container_of(kobj
, struct hd_struct
, kobj
);
505 add_uevent_var(envp
, num_envp
, &i
, buffer
, buffer_size
,
507 disk
->first_minor
+ part
->partno
);
511 add_uevent_var(envp
, num_envp
, &i
, buffer
, buffer_size
, &length
,
512 "MAJOR=%u", disk
->major
);
514 /* add physical device, backing this device */
515 physdev
= disk
->driverfs_dev
;
517 char *path
= kobject_get_path(&physdev
->kobj
, GFP_KERNEL
);
519 add_uevent_var(envp
, num_envp
, &i
, buffer
, buffer_size
,
520 &length
, "PHYSDEVPATH=%s", path
);
524 add_uevent_var(envp
, num_envp
, &i
,
525 buffer
, buffer_size
, &length
,
530 add_uevent_var(envp
, num_envp
, &i
,
531 buffer
, buffer_size
, &length
,
533 physdev
->driver
->name
);
536 /* terminate, set to next free slot, shrink available space */
540 buffer
= &buffer
[length
];
541 buffer_size
-= length
;
546 static struct kset_uevent_ops block_uevent_ops
= {
547 .filter
= block_uevent_filter
,
548 .uevent
= block_uevent
,
551 decl_subsys(block
, &ktype_block
, &block_uevent_ops
);
554 * aggregate disk stat collector. Uses the same stats that the sysfs
555 * entries do, above, but makes them available through one seq_file.
556 * Watching a few disks may be efficient through sysfs, but watching
557 * all of them will be more efficient through this interface.
559 * The output looks suspiciously like /proc/partitions with a bunch of
564 static void *diskstats_start(struct seq_file
*part
, loff_t
*pos
)
569 mutex_lock(&block_subsys_lock
);
570 list_for_each(p
, &block_subsys
.kset
.list
)
572 return list_entry(p
, struct gendisk
, kobj
.entry
);
576 static void *diskstats_next(struct seq_file
*part
, void *v
, loff_t
*pos
)
578 struct list_head
*p
= ((struct gendisk
*)v
)->kobj
.entry
.next
;
580 return p
==&block_subsys
.kset
.list
? NULL
:
581 list_entry(p
, struct gendisk
, kobj
.entry
);
584 static void diskstats_stop(struct seq_file
*part
, void *v
)
586 mutex_unlock(&block_subsys_lock
);
589 static int diskstats_show(struct seq_file
*s
, void *v
)
591 struct gendisk
*gp
= v
;
592 char buf
[BDEVNAME_SIZE
];
596 if (&sgp->kobj.entry == block_subsys.kset.list.next)
597 seq_puts(s, "major minor name"
598 " rio rmerge rsect ruse wio wmerge "
599 "wsect wuse running use aveq"
604 disk_round_stats(gp
);
606 seq_printf(s
, "%4d %4d %s %lu %lu %llu %u %lu %lu %llu %u %u %u %u\n",
607 gp
->major
, n
+ gp
->first_minor
, disk_name(gp
, n
, buf
),
608 disk_stat_read(gp
, ios
[0]), disk_stat_read(gp
, merges
[0]),
609 (unsigned long long)disk_stat_read(gp
, sectors
[0]),
610 jiffies_to_msecs(disk_stat_read(gp
, ticks
[0])),
611 disk_stat_read(gp
, ios
[1]), disk_stat_read(gp
, merges
[1]),
612 (unsigned long long)disk_stat_read(gp
, sectors
[1]),
613 jiffies_to_msecs(disk_stat_read(gp
, ticks
[1])),
615 jiffies_to_msecs(disk_stat_read(gp
, io_ticks
)),
616 jiffies_to_msecs(disk_stat_read(gp
, time_in_queue
)));
618 /* now show all non-0 size partitions of it */
619 for (n
= 0; n
< gp
->minors
- 1; n
++) {
620 struct hd_struct
*hd
= gp
->part
[n
];
622 if (hd
&& hd
->nr_sects
)
623 seq_printf(s
, "%4d %4d %s %u %u %u %u\n",
624 gp
->major
, n
+ gp
->first_minor
+ 1,
625 disk_name(gp
, n
+ 1, buf
),
626 hd
->ios
[0], hd
->sectors
[0],
627 hd
->ios
[1], hd
->sectors
[1]);
633 struct seq_operations diskstats_op
= {
634 .start
= diskstats_start
,
635 .next
= diskstats_next
,
636 .stop
= diskstats_stop
,
637 .show
= diskstats_show
640 struct gendisk
*alloc_disk(int minors
)
642 return alloc_disk_node(minors
, -1);
645 struct gendisk
*alloc_disk_node(int minors
, int node_id
)
647 struct gendisk
*disk
;
649 disk
= kmalloc_node(sizeof(struct gendisk
), GFP_KERNEL
, node_id
);
651 memset(disk
, 0, sizeof(struct gendisk
));
652 if (!init_disk_stats(disk
)) {
657 int size
= (minors
- 1) * sizeof(struct hd_struct
*);
658 disk
->part
= kmalloc_node(size
, GFP_KERNEL
, node_id
);
663 memset(disk
->part
, 0, size
);
665 disk
->minors
= minors
;
666 kobj_set_kset_s(disk
,block_subsys
);
667 kobject_init(&disk
->kobj
);
668 rand_initialize_disk(disk
);
673 EXPORT_SYMBOL(alloc_disk
);
674 EXPORT_SYMBOL(alloc_disk_node
);
676 struct kobject
*get_disk(struct gendisk
*disk
)
678 struct module
*owner
;
679 struct kobject
*kobj
;
683 owner
= disk
->fops
->owner
;
684 if (owner
&& !try_module_get(owner
))
686 kobj
= kobject_get(&disk
->kobj
);
695 EXPORT_SYMBOL(get_disk
);
697 void put_disk(struct gendisk
*disk
)
700 kobject_put(&disk
->kobj
);
703 EXPORT_SYMBOL(put_disk
);
705 void set_device_ro(struct block_device
*bdev
, int flag
)
707 if (bdev
->bd_contains
!= bdev
)
708 bdev
->bd_part
->policy
= flag
;
710 bdev
->bd_disk
->policy
= flag
;
713 EXPORT_SYMBOL(set_device_ro
);
715 void set_disk_ro(struct gendisk
*disk
, int flag
)
719 for (i
= 0; i
< disk
->minors
- 1; i
++)
720 if (disk
->part
[i
]) disk
->part
[i
]->policy
= flag
;
723 EXPORT_SYMBOL(set_disk_ro
);
725 int bdev_read_only(struct block_device
*bdev
)
729 else if (bdev
->bd_contains
!= bdev
)
730 return bdev
->bd_part
->policy
;
732 return bdev
->bd_disk
->policy
;
735 EXPORT_SYMBOL(bdev_read_only
);
737 int invalidate_partition(struct gendisk
*disk
, int index
)
740 struct block_device
*bdev
= bdget_disk(disk
, index
);
743 res
= __invalidate_device(bdev
);
749 EXPORT_SYMBOL(invalidate_partition
);