5 #include <linux/module.h>
7 #include <linux/genhd.h>
8 #include <linux/kernel.h>
9 #include <linux/blkdev.h>
10 #include <linux/init.h>
11 #include <linux/spinlock.h>
12 #include <linux/seq_file.h>
13 #include <linux/slab.h>
14 #include <linux/kmod.h>
15 #include <linux/kobj_map.h>
16 #include <linux/buffer_head.h>
17 #include <linux/mutex.h>
19 struct subsystem block_subsys
;
20 static DEFINE_MUTEX(block_subsys_lock
);
23 * Can be deleted altogether. Later.
26 static struct blk_major_name
{
27 struct blk_major_name
*next
;
30 } *major_names
[BLKDEV_MAJOR_HASH_SIZE
];
32 /* index in the above - for now: assume no multimajor ranges */
33 static inline int major_to_index(int major
)
35 return major
% BLKDEV_MAJOR_HASH_SIZE
;
40 void blkdev_show(struct seq_file
*f
, off_t offset
)
42 struct blk_major_name
*dp
;
44 if (offset
< BLKDEV_MAJOR_HASH_SIZE
) {
45 mutex_lock(&block_subsys_lock
);
46 for (dp
= major_names
[offset
]; dp
; dp
= dp
->next
)
47 seq_printf(f
, "%3d %s\n", dp
->major
, dp
->name
);
48 mutex_unlock(&block_subsys_lock
);
52 #endif /* CONFIG_PROC_FS */
54 int register_blkdev(unsigned int major
, const char *name
)
56 struct blk_major_name
**n
, *p
;
59 mutex_lock(&block_subsys_lock
);
63 for (index
= ARRAY_SIZE(major_names
)-1; index
> 0; index
--) {
64 if (major_names
[index
] == NULL
)
69 printk("register_blkdev: failed to get major for %s\n",
78 p
= kmalloc(sizeof(struct blk_major_name
), GFP_KERNEL
);
85 strlcpy(p
->name
, name
, sizeof(p
->name
));
87 index
= major_to_index(major
);
89 for (n
= &major_names
[index
]; *n
; n
= &(*n
)->next
) {
90 if ((*n
)->major
== major
)
99 printk("register_blkdev: cannot get major %d for %s\n",
104 mutex_unlock(&block_subsys_lock
);
108 EXPORT_SYMBOL(register_blkdev
);
110 /* todo: make void - error printk here */
111 int unregister_blkdev(unsigned int major
, const char *name
)
113 struct blk_major_name
**n
;
114 struct blk_major_name
*p
= NULL
;
115 int index
= major_to_index(major
);
118 mutex_lock(&block_subsys_lock
);
119 for (n
= &major_names
[index
]; *n
; n
= &(*n
)->next
)
120 if ((*n
)->major
== major
)
122 if (!*n
|| strcmp((*n
)->name
, name
))
128 mutex_unlock(&block_subsys_lock
);
134 EXPORT_SYMBOL(unregister_blkdev
);
136 static struct kobj_map
*bdev_map
;
139 * Register device numbers dev..(dev+range-1)
140 * range must be nonzero
141 * The hash chain is sorted on range, so that subranges can override.
143 void blk_register_region(dev_t dev
, unsigned long range
, struct module
*module
,
144 struct kobject
*(*probe
)(dev_t
, int *, void *),
145 int (*lock
)(dev_t
, void *), void *data
)
147 kobj_map(bdev_map
, dev
, range
, module
, probe
, lock
, data
);
150 EXPORT_SYMBOL(blk_register_region
);
152 void blk_unregister_region(dev_t dev
, unsigned long range
)
154 kobj_unmap(bdev_map
, dev
, range
);
157 EXPORT_SYMBOL(blk_unregister_region
);
159 static struct kobject
*exact_match(dev_t dev
, int *part
, void *data
)
161 struct gendisk
*p
= data
;
165 static int exact_lock(dev_t dev
, void *data
)
167 struct gendisk
*p
= data
;
175 * add_disk - add partitioning information to kernel list
176 * @disk: per-device partitioning information
178 * This function registers the partitioning information in @disk
181 void add_disk(struct gendisk
*disk
)
183 disk
->flags
|= GENHD_FL_UP
;
184 blk_register_region(MKDEV(disk
->major
, disk
->first_minor
),
185 disk
->minors
, NULL
, exact_match
, exact_lock
, disk
);
187 blk_register_queue(disk
);
190 EXPORT_SYMBOL(add_disk
);
191 EXPORT_SYMBOL(del_gendisk
); /* in partitions/check.c */
193 void unlink_gendisk(struct gendisk
*disk
)
195 blk_unregister_queue(disk
);
196 blk_unregister_region(MKDEV(disk
->major
, disk
->first_minor
),
200 #define to_disk(obj) container_of(obj,struct gendisk,kobj)
203 * get_gendisk - get partitioning information for a given device
204 * @dev: device to get partitioning information for
206 * This function gets the structure containing partitioning
207 * information for the given device @dev.
209 struct gendisk
*get_gendisk(dev_t dev
, int *part
)
211 struct kobject
*kobj
= kobj_lookup(bdev_map
, dev
, part
);
212 return kobj
? to_disk(kobj
) : NULL
;
215 #ifdef CONFIG_PROC_FS
217 static void *part_start(struct seq_file
*part
, loff_t
*pos
)
222 mutex_lock(&block_subsys_lock
);
223 list_for_each(p
, &block_subsys
.kset
.list
)
225 return list_entry(p
, struct gendisk
, kobj
.entry
);
229 static void *part_next(struct seq_file
*part
, void *v
, loff_t
*pos
)
231 struct list_head
*p
= ((struct gendisk
*)v
)->kobj
.entry
.next
;
233 return p
==&block_subsys
.kset
.list
? NULL
:
234 list_entry(p
, struct gendisk
, kobj
.entry
);
237 static void part_stop(struct seq_file
*part
, void *v
)
239 mutex_unlock(&block_subsys_lock
);
242 static int show_partition(struct seq_file
*part
, void *v
)
244 struct gendisk
*sgp
= v
;
246 char buf
[BDEVNAME_SIZE
];
248 if (&sgp
->kobj
.entry
== block_subsys
.kset
.list
.next
)
249 seq_puts(part
, "major minor #blocks name\n\n");
251 /* Don't show non-partitionable removeable devices or empty devices */
252 if (!get_capacity(sgp
) ||
253 (sgp
->minors
== 1 && (sgp
->flags
& GENHD_FL_REMOVABLE
)))
255 if (sgp
->flags
& GENHD_FL_SUPPRESS_PARTITION_INFO
)
258 /* show the full disk and all non-0 size partitions of it */
259 seq_printf(part
, "%4d %4d %10llu %s\n",
260 sgp
->major
, sgp
->first_minor
,
261 (unsigned long long)get_capacity(sgp
) >> 1,
262 disk_name(sgp
, 0, buf
));
263 for (n
= 0; n
< sgp
->minors
- 1; n
++) {
266 if (sgp
->part
[n
]->nr_sects
== 0)
268 seq_printf(part
, "%4d %4d %10llu %s\n",
269 sgp
->major
, n
+ 1 + sgp
->first_minor
,
270 (unsigned long long)sgp
->part
[n
]->nr_sects
>> 1 ,
271 disk_name(sgp
, n
+ 1, buf
));
277 struct seq_operations partitions_op
= {
281 .show
= show_partition
286 extern int blk_dev_init(void);
288 static struct kobject
*base_probe(dev_t dev
, int *part
, void *data
)
290 if (request_module("block-major-%d-%d", MAJOR(dev
), MINOR(dev
)) > 0)
291 /* Make old-style 2.4 aliases work */
292 request_module("block-major-%d", MAJOR(dev
));
296 static int __init
genhd_device_init(void)
300 bdev_map
= kobj_map_init(base_probe
, &block_subsys_lock
);
302 err
= subsystem_register(&block_subsys
);
304 printk(KERN_WARNING
"%s: subsystem_register error: %d\n",
309 subsys_initcall(genhd_device_init
);
314 * kobject & sysfs bindings for block devices
316 static ssize_t
disk_attr_show(struct kobject
*kobj
, struct attribute
*attr
,
319 struct gendisk
*disk
= to_disk(kobj
);
320 struct disk_attribute
*disk_attr
=
321 container_of(attr
,struct disk_attribute
,attr
);
325 ret
= disk_attr
->show(disk
,page
);
329 static ssize_t
disk_attr_store(struct kobject
* kobj
, struct attribute
* attr
,
330 const char *page
, size_t count
)
332 struct gendisk
*disk
= to_disk(kobj
);
333 struct disk_attribute
*disk_attr
=
334 container_of(attr
,struct disk_attribute
,attr
);
337 if (disk_attr
->store
)
338 ret
= disk_attr
->store(disk
, page
, count
);
342 static struct sysfs_ops disk_sysfs_ops
= {
343 .show
= &disk_attr_show
,
344 .store
= &disk_attr_store
,
347 static ssize_t
disk_uevent_store(struct gendisk
* disk
,
348 const char *buf
, size_t count
)
350 kobject_uevent(&disk
->kobj
, KOBJ_ADD
);
353 static ssize_t
disk_dev_read(struct gendisk
* disk
, char *page
)
355 dev_t base
= MKDEV(disk
->major
, disk
->first_minor
);
356 return print_dev_t(page
, base
);
358 static ssize_t
disk_range_read(struct gendisk
* disk
, char *page
)
360 return sprintf(page
, "%d\n", disk
->minors
);
362 static ssize_t
disk_removable_read(struct gendisk
* disk
, char *page
)
364 return sprintf(page
, "%d\n",
365 (disk
->flags
& GENHD_FL_REMOVABLE
? 1 : 0));
368 static ssize_t
disk_size_read(struct gendisk
* disk
, char *page
)
370 return sprintf(page
, "%llu\n", (unsigned long long)get_capacity(disk
));
373 static ssize_t
disk_stats_read(struct gendisk
* disk
, char *page
)
376 disk_round_stats(disk
);
379 "%8lu %8lu %8llu %8u "
380 "%8lu %8lu %8llu %8u "
383 disk_stat_read(disk
, ios
[READ
]),
384 disk_stat_read(disk
, merges
[READ
]),
385 (unsigned long long)disk_stat_read(disk
, sectors
[READ
]),
386 jiffies_to_msecs(disk_stat_read(disk
, ticks
[READ
])),
387 disk_stat_read(disk
, ios
[WRITE
]),
388 disk_stat_read(disk
, merges
[WRITE
]),
389 (unsigned long long)disk_stat_read(disk
, sectors
[WRITE
]),
390 jiffies_to_msecs(disk_stat_read(disk
, ticks
[WRITE
])),
392 jiffies_to_msecs(disk_stat_read(disk
, io_ticks
)),
393 jiffies_to_msecs(disk_stat_read(disk
, time_in_queue
)));
395 static struct disk_attribute disk_attr_uevent
= {
396 .attr
= {.name
= "uevent", .mode
= S_IWUSR
},
397 .store
= disk_uevent_store
399 static struct disk_attribute disk_attr_dev
= {
400 .attr
= {.name
= "dev", .mode
= S_IRUGO
},
401 .show
= disk_dev_read
403 static struct disk_attribute disk_attr_range
= {
404 .attr
= {.name
= "range", .mode
= S_IRUGO
},
405 .show
= disk_range_read
407 static struct disk_attribute disk_attr_removable
= {
408 .attr
= {.name
= "removable", .mode
= S_IRUGO
},
409 .show
= disk_removable_read
411 static struct disk_attribute disk_attr_size
= {
412 .attr
= {.name
= "size", .mode
= S_IRUGO
},
413 .show
= disk_size_read
415 static struct disk_attribute disk_attr_stat
= {
416 .attr
= {.name
= "stat", .mode
= S_IRUGO
},
417 .show
= disk_stats_read
420 #ifdef CONFIG_FAIL_MAKE_REQUEST
422 static ssize_t
disk_fail_store(struct gendisk
* disk
,
423 const char *buf
, size_t count
)
427 if (count
> 0 && sscanf(buf
, "%d", &i
) > 0) {
429 disk
->flags
&= ~GENHD_FL_FAIL
;
431 disk
->flags
|= GENHD_FL_FAIL
;
436 static ssize_t
disk_fail_read(struct gendisk
* disk
, char *page
)
438 return sprintf(page
, "%d\n", disk
->flags
& GENHD_FL_FAIL
? 1 : 0);
440 static struct disk_attribute disk_attr_fail
= {
441 .attr
= {.name
= "make-it-fail", .mode
= S_IRUGO
| S_IWUSR
},
442 .store
= disk_fail_store
,
443 .show
= disk_fail_read
448 static struct attribute
* default_attrs
[] = {
449 &disk_attr_uevent
.attr
,
451 &disk_attr_range
.attr
,
452 &disk_attr_removable
.attr
,
453 &disk_attr_size
.attr
,
454 &disk_attr_stat
.attr
,
455 #ifdef CONFIG_FAIL_MAKE_REQUEST
456 &disk_attr_fail
.attr
,
461 static void disk_release(struct kobject
* kobj
)
463 struct gendisk
*disk
= to_disk(kobj
);
466 free_disk_stats(disk
);
470 static struct kobj_type ktype_block
= {
471 .release
= disk_release
,
472 .sysfs_ops
= &disk_sysfs_ops
,
473 .default_attrs
= default_attrs
,
476 extern struct kobj_type ktype_part
;
478 static int block_uevent_filter(struct kset
*kset
, struct kobject
*kobj
)
480 struct kobj_type
*ktype
= get_ktype(kobj
);
482 return ((ktype
== &ktype_block
) || (ktype
== &ktype_part
));
485 static int block_uevent(struct kset
*kset
, struct kobject
*kobj
, char **envp
,
486 int num_envp
, char *buffer
, int buffer_size
)
488 struct kobj_type
*ktype
= get_ktype(kobj
);
489 struct device
*physdev
;
490 struct gendisk
*disk
;
491 struct hd_struct
*part
;
495 if (ktype
== &ktype_block
) {
496 disk
= container_of(kobj
, struct gendisk
, kobj
);
497 add_uevent_var(envp
, num_envp
, &i
, buffer
, buffer_size
,
498 &length
, "MINOR=%u", disk
->first_minor
);
499 } else if (ktype
== &ktype_part
) {
500 disk
= container_of(kobj
->parent
, struct gendisk
, kobj
);
501 part
= container_of(kobj
, struct hd_struct
, kobj
);
502 add_uevent_var(envp
, num_envp
, &i
, buffer
, buffer_size
,
504 disk
->first_minor
+ part
->partno
);
508 add_uevent_var(envp
, num_envp
, &i
, buffer
, buffer_size
, &length
,
509 "MAJOR=%u", disk
->major
);
511 /* add physical device, backing this device */
512 physdev
= disk
->driverfs_dev
;
514 char *path
= kobject_get_path(&physdev
->kobj
, GFP_KERNEL
);
516 add_uevent_var(envp
, num_envp
, &i
, buffer
, buffer_size
,
517 &length
, "PHYSDEVPATH=%s", path
);
521 add_uevent_var(envp
, num_envp
, &i
,
522 buffer
, buffer_size
, &length
,
527 add_uevent_var(envp
, num_envp
, &i
,
528 buffer
, buffer_size
, &length
,
530 physdev
->driver
->name
);
533 /* terminate, set to next free slot, shrink available space */
537 buffer
= &buffer
[length
];
538 buffer_size
-= length
;
543 static struct kset_uevent_ops block_uevent_ops
= {
544 .filter
= block_uevent_filter
,
545 .uevent
= block_uevent
,
548 decl_subsys(block
, &ktype_block
, &block_uevent_ops
);
551 * aggregate disk stat collector. Uses the same stats that the sysfs
552 * entries do, above, but makes them available through one seq_file.
553 * Watching a few disks may be efficient through sysfs, but watching
554 * all of them will be more efficient through this interface.
556 * The output looks suspiciously like /proc/partitions with a bunch of
561 static void *diskstats_start(struct seq_file
*part
, loff_t
*pos
)
566 mutex_lock(&block_subsys_lock
);
567 list_for_each(p
, &block_subsys
.kset
.list
)
569 return list_entry(p
, struct gendisk
, kobj
.entry
);
573 static void *diskstats_next(struct seq_file
*part
, void *v
, loff_t
*pos
)
575 struct list_head
*p
= ((struct gendisk
*)v
)->kobj
.entry
.next
;
577 return p
==&block_subsys
.kset
.list
? NULL
:
578 list_entry(p
, struct gendisk
, kobj
.entry
);
581 static void diskstats_stop(struct seq_file
*part
, void *v
)
583 mutex_unlock(&block_subsys_lock
);
586 static int diskstats_show(struct seq_file
*s
, void *v
)
588 struct gendisk
*gp
= v
;
589 char buf
[BDEVNAME_SIZE
];
593 if (&sgp->kobj.entry == block_subsys.kset.list.next)
594 seq_puts(s, "major minor name"
595 " rio rmerge rsect ruse wio wmerge "
596 "wsect wuse running use aveq"
601 disk_round_stats(gp
);
603 seq_printf(s
, "%4d %4d %s %lu %lu %llu %u %lu %lu %llu %u %u %u %u\n",
604 gp
->major
, n
+ gp
->first_minor
, disk_name(gp
, n
, buf
),
605 disk_stat_read(gp
, ios
[0]), disk_stat_read(gp
, merges
[0]),
606 (unsigned long long)disk_stat_read(gp
, sectors
[0]),
607 jiffies_to_msecs(disk_stat_read(gp
, ticks
[0])),
608 disk_stat_read(gp
, ios
[1]), disk_stat_read(gp
, merges
[1]),
609 (unsigned long long)disk_stat_read(gp
, sectors
[1]),
610 jiffies_to_msecs(disk_stat_read(gp
, ticks
[1])),
612 jiffies_to_msecs(disk_stat_read(gp
, io_ticks
)),
613 jiffies_to_msecs(disk_stat_read(gp
, time_in_queue
)));
615 /* now show all non-0 size partitions of it */
616 for (n
= 0; n
< gp
->minors
- 1; n
++) {
617 struct hd_struct
*hd
= gp
->part
[n
];
619 if (hd
&& hd
->nr_sects
)
620 seq_printf(s
, "%4d %4d %s %u %u %u %u\n",
621 gp
->major
, n
+ gp
->first_minor
+ 1,
622 disk_name(gp
, n
+ 1, buf
),
623 hd
->ios
[0], hd
->sectors
[0],
624 hd
->ios
[1], hd
->sectors
[1]);
630 struct seq_operations diskstats_op
= {
631 .start
= diskstats_start
,
632 .next
= diskstats_next
,
633 .stop
= diskstats_stop
,
634 .show
= diskstats_show
637 struct gendisk
*alloc_disk(int minors
)
639 return alloc_disk_node(minors
, -1);
642 struct gendisk
*alloc_disk_node(int minors
, int node_id
)
644 struct gendisk
*disk
;
646 disk
= kmalloc_node(sizeof(struct gendisk
), GFP_KERNEL
, node_id
);
648 memset(disk
, 0, sizeof(struct gendisk
));
649 if (!init_disk_stats(disk
)) {
654 int size
= (minors
- 1) * sizeof(struct hd_struct
*);
655 disk
->part
= kmalloc_node(size
, GFP_KERNEL
, node_id
);
660 memset(disk
->part
, 0, size
);
662 disk
->minors
= minors
;
663 kobj_set_kset_s(disk
,block_subsys
);
664 kobject_init(&disk
->kobj
);
665 rand_initialize_disk(disk
);
670 EXPORT_SYMBOL(alloc_disk
);
671 EXPORT_SYMBOL(alloc_disk_node
);
673 struct kobject
*get_disk(struct gendisk
*disk
)
675 struct module
*owner
;
676 struct kobject
*kobj
;
680 owner
= disk
->fops
->owner
;
681 if (owner
&& !try_module_get(owner
))
683 kobj
= kobject_get(&disk
->kobj
);
692 EXPORT_SYMBOL(get_disk
);
694 void put_disk(struct gendisk
*disk
)
697 kobject_put(&disk
->kobj
);
700 EXPORT_SYMBOL(put_disk
);
702 void set_device_ro(struct block_device
*bdev
, int flag
)
704 if (bdev
->bd_contains
!= bdev
)
705 bdev
->bd_part
->policy
= flag
;
707 bdev
->bd_disk
->policy
= flag
;
710 EXPORT_SYMBOL(set_device_ro
);
712 void set_disk_ro(struct gendisk
*disk
, int flag
)
716 for (i
= 0; i
< disk
->minors
- 1; i
++)
717 if (disk
->part
[i
]) disk
->part
[i
]->policy
= flag
;
720 EXPORT_SYMBOL(set_disk_ro
);
722 int bdev_read_only(struct block_device
*bdev
)
726 else if (bdev
->bd_contains
!= bdev
)
727 return bdev
->bd_part
->policy
;
729 return bdev
->bd_disk
->policy
;
732 EXPORT_SYMBOL(bdev_read_only
);
734 int invalidate_partition(struct gendisk
*disk
, int index
)
737 struct block_device
*bdev
= bdget_disk(disk
, index
);
740 res
= __invalidate_device(bdev
);
746 EXPORT_SYMBOL(invalidate_partition
);