2 * fs/partitions/check.c
4 * Code extracted from drivers/block/genhd.c
5 * Copyright (C) 1991-1998 Linus Torvalds
6 * Re-organised Feb 1998 Russell King
8 * We now have independent partition support from the
9 * block drivers, which allows all the partition code to
10 * be grouped in one location, and it to be mostly self
13 * Added needed MAJORS for new pairs, {hdi,hdj}, {hdk,hdl}
16 #include <linux/init.h>
17 #include <linux/module.h>
19 #include <linux/slab.h>
20 #include <linux/kmod.h>
21 #include <linux/ctype.h>
22 #include <linux/genhd.h>
23 #include <linux/blktrace_api.h>
42 #ifdef CONFIG_BLK_DEV_MD
43 extern void md_autodetect_dev(dev_t dev);
46 int warn_no_part = 1; /*This is ugly: should make genhd removable media aware*/
48 static int (*check_part[])(struct parsed_partitions *) = {
50 * Probe partition formats with tables at disk address 0
51 * that also have an ADFS boot block at 0xdc0.
53 #ifdef CONFIG_ACORN_PARTITION_ICS
56 #ifdef CONFIG_ACORN_PARTITION_POWERTEC
57 adfspart_check_POWERTEC,
59 #ifdef CONFIG_ACORN_PARTITION_EESOX
64 * Now move on to formats that only have partition info at
65 * disk address 0xdc0. Since these may also have stale
66 * PC/BIOS partition tables, they need to come before
69 #ifdef CONFIG_ACORN_PARTITION_CUMANA
70 adfspart_check_CUMANA,
72 #ifdef CONFIG_ACORN_PARTITION_ADFS
76 #ifdef CONFIG_EFI_PARTITION
77 efi_partition, /* this must come before msdos */
79 #ifdef CONFIG_SGI_PARTITION
82 #ifdef CONFIG_LDM_PARTITION
83 ldm_partition, /* this must come before msdos */
85 #ifdef CONFIG_MSDOS_PARTITION
88 #ifdef CONFIG_OSF_PARTITION
91 #ifdef CONFIG_SUN_PARTITION
94 #ifdef CONFIG_AMIGA_PARTITION
97 #ifdef CONFIG_ATARI_PARTITION
100 #ifdef CONFIG_MAC_PARTITION
103 #ifdef CONFIG_ULTRIX_PARTITION
106 #ifdef CONFIG_IBM_PARTITION
109 #ifdef CONFIG_KARMA_PARTITION
112 #ifdef CONFIG_SYSV68_PARTITION
119 * disk_name() is used by partition check code and the genhd driver.
120 * It formats the devicename of the indicated disk into
121 * the supplied buffer (of size at least 32), and returns
122 * a pointer to that same buffer (for convenience).
125 char *disk_name(struct gendisk *hd, int partno, char *buf)
128 snprintf(buf, BDEVNAME_SIZE, "%s", hd->disk_name);
129 else if (isdigit(hd->disk_name[strlen(hd->disk_name)-1]))
130 snprintf(buf, BDEVNAME_SIZE, "%sp%d", hd->disk_name, partno);
132 snprintf(buf, BDEVNAME_SIZE, "%s%d", hd->disk_name, partno);
137 const char *bdevname(struct block_device *bdev, char *buf)
139 return disk_name(bdev->bd_disk, bdev->bd_part->partno, buf);
142 EXPORT_SYMBOL(bdevname);
145 * There's very little reason to use this, you should really
146 * have a struct block_device just about everywhere and use
147 * bdevname() instead.
149 const char *__bdevname(dev_t dev, char *buffer)
151 scnprintf(buffer, BDEVNAME_SIZE, "unknown-block(%u,%u)",
152 MAJOR(dev), MINOR(dev));
156 EXPORT_SYMBOL(__bdevname);
158 static struct parsed_partitions *
159 check_partition(struct gendisk *hd, struct block_device *bdev)
161 struct parsed_partitions *state;
164 state = kzalloc(sizeof(struct parsed_partitions), GFP_KERNEL);
167 state->pp_buf = (char *)__get_free_page(GFP_KERNEL);
168 if (!state->pp_buf) {
172 state->pp_buf[0] = '\0';
175 disk_name(hd, 0, state->name);
176 snprintf(state->pp_buf, PAGE_SIZE, " %s:", state->name);
177 if (isdigit(state->name[strlen(state->name)-1]))
178 sprintf(state->name, "p");
180 state->limit = disk_max_parts(hd);
182 while (!res && check_part[i]) {
183 memset(&state->parts, 0, sizeof(state->parts));
184 res = check_part[i++](state);
186 /* We have hit an I/O error which we don't report now.
187 * But record it, and let the others do their job.
195 printk(KERN_INFO "%s", state->pp_buf);
197 free_page((unsigned long)state->pp_buf);
200 if (state->access_beyond_eod)
203 /* The partition is unrecognized. So report I/O errors if there were any */
206 strlcat(state->pp_buf, " unknown partition table\n", PAGE_SIZE);
207 else if (warn_no_part)
208 strlcat(state->pp_buf, " unable to read partition table\n", PAGE_SIZE);
210 printk(KERN_INFO "%s", state->pp_buf);
212 free_page((unsigned long)state->pp_buf);
217 static ssize_t part_partition_show(struct device *dev,
218 struct device_attribute *attr, char *buf)
220 struct hd_struct *p = dev_to_part(dev);
222 return sprintf(buf, "%d\n", p->partno);
225 static ssize_t part_start_show(struct device *dev,
226 struct device_attribute *attr, char *buf)
228 struct hd_struct *p = dev_to_part(dev);
230 return sprintf(buf, "%llu\n",(unsigned long long)p->start_sect);
233 ssize_t part_size_show(struct device *dev,
234 struct device_attribute *attr, char *buf)
236 struct hd_struct *p = dev_to_part(dev);
237 return sprintf(buf, "%llu\n",(unsigned long long)p->nr_sects);
240 ssize_t part_ro_show(struct device *dev,
241 struct device_attribute *attr, char *buf)
243 struct hd_struct *p = dev_to_part(dev);
244 return sprintf(buf, "%d\n", p->policy ? 1 : 0);
247 ssize_t part_alignment_offset_show(struct device *dev,
248 struct device_attribute *attr, char *buf)
250 struct hd_struct *p = dev_to_part(dev);
251 return sprintf(buf, "%llu\n", (unsigned long long)p->alignment_offset);
254 ssize_t part_discard_alignment_show(struct device *dev,
255 struct device_attribute *attr, char *buf)
257 struct hd_struct *p = dev_to_part(dev);
258 struct gendisk *disk = dev_to_disk(dev);
260 return sprintf(buf, "%u\n",
261 queue_limit_discard_alignment(&disk->queue->limits,
265 ssize_t part_stat_show(struct device *dev,
266 struct device_attribute *attr, char *buf)
268 struct hd_struct *p = dev_to_part(dev);
271 cpu = part_stat_lock();
272 part_round_stats(cpu, p);
275 "%8lu %8lu %8llu %8u "
276 "%8lu %8lu %8llu %8u "
279 part_stat_read(p, ios[READ]),
280 part_stat_read(p, merges[READ]),
281 (unsigned long long)part_stat_read(p, sectors[READ]),
282 jiffies_to_msecs(part_stat_read(p, ticks[READ])),
283 part_stat_read(p, ios[WRITE]),
284 part_stat_read(p, merges[WRITE]),
285 (unsigned long long)part_stat_read(p, sectors[WRITE]),
286 jiffies_to_msecs(part_stat_read(p, ticks[WRITE])),
288 jiffies_to_msecs(part_stat_read(p, io_ticks)),
289 jiffies_to_msecs(part_stat_read(p, time_in_queue)));
292 ssize_t part_inflight_show(struct device *dev,
293 struct device_attribute *attr, char *buf)
295 struct hd_struct *p = dev_to_part(dev);
297 return sprintf(buf, "%8u %8u\n", atomic_read(&p->in_flight[0]),
298 atomic_read(&p->in_flight[1]));
301 #ifdef CONFIG_FAIL_MAKE_REQUEST
302 ssize_t part_fail_show(struct device *dev,
303 struct device_attribute *attr, char *buf)
305 struct hd_struct *p = dev_to_part(dev);
307 return sprintf(buf, "%d\n", p->make_it_fail);
310 ssize_t part_fail_store(struct device *dev,
311 struct device_attribute *attr,
312 const char *buf, size_t count)
314 struct hd_struct *p = dev_to_part(dev);
317 if (count > 0 && sscanf(buf, "%d", &i) > 0)
318 p->make_it_fail = (i == 0) ? 0 : 1;
324 static DEVICE_ATTR(partition, S_IRUGO, part_partition_show, NULL);
325 static DEVICE_ATTR(start, S_IRUGO, part_start_show, NULL);
326 static DEVICE_ATTR(size, S_IRUGO, part_size_show, NULL);
327 static DEVICE_ATTR(ro, S_IRUGO, part_ro_show, NULL);
328 static DEVICE_ATTR(alignment_offset, S_IRUGO, part_alignment_offset_show, NULL);
329 static DEVICE_ATTR(discard_alignment, S_IRUGO, part_discard_alignment_show,
331 static DEVICE_ATTR(stat, S_IRUGO, part_stat_show, NULL);
332 static DEVICE_ATTR(inflight, S_IRUGO, part_inflight_show, NULL);
333 #ifdef CONFIG_FAIL_MAKE_REQUEST
334 static struct device_attribute dev_attr_fail =
335 __ATTR(make-it-fail, S_IRUGO|S_IWUSR, part_fail_show, part_fail_store);
338 static struct attribute *part_attrs[] = {
339 &dev_attr_partition.attr,
340 &dev_attr_start.attr,
343 &dev_attr_alignment_offset.attr,
344 &dev_attr_discard_alignment.attr,
346 &dev_attr_inflight.attr,
347 #ifdef CONFIG_FAIL_MAKE_REQUEST
353 static struct attribute_group part_attr_group = {
357 static const struct attribute_group *part_attr_groups[] = {
359 #ifdef CONFIG_BLK_DEV_IO_TRACE
360 &blk_trace_attr_group,
365 static void part_release(struct device *dev)
367 struct hd_struct *p = dev_to_part(dev);
373 struct device_type part_type = {
375 .groups = part_attr_groups,
376 .release = part_release,
379 static void delete_partition_rcu_cb(struct rcu_head *head)
381 struct hd_struct *part = container_of(head, struct hd_struct, rcu_head);
383 part->start_sect = 0;
385 part_stat_set_all(part, 0);
386 put_device(part_to_dev(part));
389 void __delete_partition(struct hd_struct *part)
391 call_rcu(&part->rcu_head, delete_partition_rcu_cb);
394 void delete_partition(struct gendisk *disk, int partno)
396 struct disk_part_tbl *ptbl = disk->part_tbl;
397 struct hd_struct *part;
399 if (partno >= ptbl->len)
402 part = ptbl->part[partno];
406 blk_free_devt(part_devt(part));
407 rcu_assign_pointer(ptbl->part[partno], NULL);
408 rcu_assign_pointer(ptbl->last_lookup, NULL);
409 kobject_put(part->holder_dir);
410 device_del(part_to_dev(part));
415 static ssize_t whole_disk_show(struct device *dev,
416 struct device_attribute *attr, char *buf)
420 static DEVICE_ATTR(whole_disk, S_IRUSR | S_IRGRP | S_IROTH,
421 whole_disk_show, NULL);
423 struct hd_struct *add_partition(struct gendisk *disk, int partno,
424 sector_t start, sector_t len, int flags,
425 struct partition_meta_info *info)
428 dev_t devt = MKDEV(0, 0);
429 struct device *ddev = disk_to_dev(disk);
431 struct disk_part_tbl *ptbl;
435 err = disk_expand_part_tbl(disk, partno);
438 ptbl = disk->part_tbl;
440 if (ptbl->part[partno])
441 return ERR_PTR(-EBUSY);
443 p = kzalloc(sizeof(*p), GFP_KERNEL);
445 return ERR_PTR(-EBUSY);
447 if (!init_part_stats(p)) {
451 pdev = part_to_dev(p);
453 p->start_sect = start;
454 p->alignment_offset =
455 queue_limit_alignment_offset(&disk->queue->limits, start);
458 p->policy = get_disk_ro(disk);
461 struct partition_meta_info *pinfo = alloc_part_info(disk);
464 memcpy(pinfo, info, sizeof(*info));
468 dname = dev_name(ddev);
469 if (isdigit(dname[strlen(dname) - 1]))
470 dev_set_name(pdev, "%sp%d", dname, partno);
472 dev_set_name(pdev, "%s%d", dname, partno);
474 device_initialize(pdev);
475 pdev->class = &block_class;
476 pdev->type = &part_type;
479 err = blk_alloc_devt(p, &devt);
484 /* delay uevent until 'holders' subdir is created */
485 dev_set_uevent_suppress(pdev, 1);
486 err = device_add(pdev);
491 p->holder_dir = kobject_create_and_add("holders", &pdev->kobj);
495 dev_set_uevent_suppress(pdev, 0);
496 if (flags & ADDPART_FLAG_WHOLEDISK) {
497 err = device_create_file(pdev, &dev_attr_whole_disk);
502 /* everything is up and running, commence */
503 rcu_assign_pointer(ptbl->part[partno], p);
505 /* suppress uevent if the disk suppresses it */
506 if (!dev_get_uevent_suppress(ddev))
507 kobject_uevent(&pdev->kobj, KOBJ_ADD);
520 kobject_put(p->holder_dir);
528 static bool disk_unlock_native_capacity(struct gendisk *disk)
530 const struct block_device_operations *bdops = disk->fops;
532 if (bdops->unlock_native_capacity &&
533 !(disk->flags & GENHD_FL_NATIVE_CAPACITY)) {
534 printk(KERN_CONT "enabling native capacity\n");
535 bdops->unlock_native_capacity(disk);
536 disk->flags |= GENHD_FL_NATIVE_CAPACITY;
539 printk(KERN_CONT "truncated\n");
544 int rescan_partitions(struct gendisk *disk, struct block_device *bdev)
546 struct parsed_partitions *state = NULL;
547 struct disk_part_iter piter;
548 struct hd_struct *part;
551 if (state && !IS_ERR(state)) {
556 if (bdev->bd_part_count)
558 res = invalidate_partition(disk, 0);
562 disk_part_iter_init(&piter, disk, DISK_PITER_INCL_EMPTY);
563 while ((part = disk_part_iter_next(&piter)))
564 delete_partition(disk, part->partno);
565 disk_part_iter_exit(&piter);
567 if (disk->fops->revalidate_disk)
568 disk->fops->revalidate_disk(disk);
569 check_disk_size_change(disk, bdev);
570 bdev->bd_invalidated = 0;
571 if (!get_capacity(disk) || !(state = check_partition(disk, bdev)))
575 * I/O error reading the partition table. If any
576 * partition code tried to read beyond EOD, retry
577 * after unlocking native capacity.
579 if (PTR_ERR(state) == -ENOSPC) {
580 printk(KERN_WARNING "%s: partition table beyond EOD, ",
582 if (disk_unlock_native_capacity(disk))
588 * If any partition code tried to read beyond EOD, try
589 * unlocking native capacity even if partition table is
590 * successfully read as we could be missing some partitions.
592 if (state->access_beyond_eod) {
594 "%s: partition table partially beyond EOD, ",
596 if (disk_unlock_native_capacity(disk))
600 /* tell userspace that the media / partition table may have changed */
601 kobject_uevent(&disk_to_dev(disk)->kobj, KOBJ_CHANGE);
603 /* Detect the highest partition number and preallocate
604 * disk->part_tbl. This is an optimization and not strictly
607 for (p = 1, highest = 0; p < state->limit; p++)
608 if (state->parts[p].size)
611 disk_expand_part_tbl(disk, highest);
614 for (p = 1; p < state->limit; p++) {
616 struct partition_meta_info *info = NULL;
618 size = state->parts[p].size;
622 from = state->parts[p].from;
623 if (from >= get_capacity(disk)) {
625 "%s: p%d start %llu is beyond EOD, ",
626 disk->disk_name, p, (unsigned long long) from);
627 if (disk_unlock_native_capacity(disk))
632 if (from + size > get_capacity(disk)) {
634 "%s: p%d size %llu extends beyond EOD, ",
635 disk->disk_name, p, (unsigned long long) size);
637 if (disk_unlock_native_capacity(disk)) {
638 /* free state and restart */
642 * we can not ignore partitions of broken tables
643 * created by for example camera firmware, but
644 * we limit them to the end of the disk to avoid
645 * creating invalid block devices
647 size = get_capacity(disk) - from;
651 if (state->parts[p].has_info)
652 info = &state->parts[p].info;
653 part = add_partition(disk, p, from, size,
654 state->parts[p].flags,
655 &state->parts[p].info);
657 printk(KERN_ERR " %s: p%d could not be added: %ld\n",
658 disk->disk_name, p, -PTR_ERR(part));
661 #ifdef CONFIG_BLK_DEV_MD
662 if (state->parts[p].flags & ADDPART_FLAG_RAID)
663 md_autodetect_dev(part_to_dev(part)->devt);
670 unsigned char *read_dev_sector(struct block_device *bdev, sector_t n, Sector *p)
672 struct address_space *mapping = bdev->bd_inode->i_mapping;
675 page = read_mapping_page(mapping, (pgoff_t)(n >> (PAGE_CACHE_SHIFT-9)),
681 return (unsigned char *)page_address(page) + ((n & ((1 << (PAGE_CACHE_SHIFT - 9)) - 1)) << 9);
683 page_cache_release(page);
689 EXPORT_SYMBOL(read_dev_sector);