1 // SPDX-License-Identifier: GPL-2.0
2 #include <linux/capability.h>
3 #include <linux/blkdev.h>
4 #include <linux/export.h>
6 #include <linux/blkpg.h>
7 #include <linux/hdreg.h>
8 #include <linux/backing-dev.h>
10 #include <linux/blktrace_api.h>
12 #include <linux/uaccess.h>
14 static int blkpg_ioctl(struct block_device *bdev, struct blkpg_ioctl_arg __user *arg)
16 struct block_device *bdevp;
18 struct hd_struct *part, *lpart;
19 struct blkpg_ioctl_arg a;
20 struct blkpg_partition p;
21 struct disk_part_iter piter;
22 long long start, length;
25 if (!capable(CAP_SYS_ADMIN))
27 if (copy_from_user(&a, arg, sizeof(struct blkpg_ioctl_arg)))
29 if (copy_from_user(&p, a.data, sizeof(struct blkpg_partition)))
32 if (bdev != bdev->bd_contains)
38 case BLKPG_ADD_PARTITION:
40 length = p.length >> 9;
41 /* check for fit in a hd_struct */
42 if (sizeof(sector_t) == sizeof(long) &&
43 sizeof(long long) > sizeof(long)) {
44 long pstart = start, plength = length;
45 if (pstart != start || plength != length
46 || pstart < 0 || plength < 0 || partno > 65535)
49 /* check if partition is aligned to blocksize */
50 if (p.start & (bdev_logical_block_size(bdev) - 1))
53 mutex_lock(&bdev->bd_mutex);
56 disk_part_iter_init(&piter, disk,
57 DISK_PITER_INCL_EMPTY);
58 while ((part = disk_part_iter_next(&piter))) {
59 if (!(start + length <= part->start_sect ||
60 start >= part->start_sect + part->nr_sects)) {
61 disk_part_iter_exit(&piter);
62 mutex_unlock(&bdev->bd_mutex);
66 disk_part_iter_exit(&piter);
69 part = add_partition(disk, partno, start, length,
70 ADDPART_FLAG_NONE, NULL);
71 mutex_unlock(&bdev->bd_mutex);
72 return PTR_ERR_OR_ZERO(part);
73 case BLKPG_DEL_PARTITION:
74 part = disk_get_part(disk, partno);
78 bdevp = bdget(part_devt(part));
83 mutex_lock(&bdevp->bd_mutex);
84 if (bdevp->bd_openers) {
85 mutex_unlock(&bdevp->bd_mutex);
91 invalidate_bdev(bdevp);
93 mutex_lock_nested(&bdev->bd_mutex, 1);
94 delete_partition(disk, partno);
95 mutex_unlock(&bdev->bd_mutex);
96 mutex_unlock(&bdevp->bd_mutex);
100 case BLKPG_RESIZE_PARTITION:
101 start = p.start >> 9;
102 /* new length of partition in bytes */
103 length = p.length >> 9;
104 /* check for fit in a hd_struct */
105 if (sizeof(sector_t) == sizeof(long) &&
106 sizeof(long long) > sizeof(long)) {
107 long pstart = start, plength = length;
108 if (pstart != start || plength != length
109 || pstart < 0 || plength < 0)
112 part = disk_get_part(disk, partno);
115 bdevp = bdget(part_devt(part));
120 mutex_lock(&bdevp->bd_mutex);
121 mutex_lock_nested(&bdev->bd_mutex, 1);
122 if (start != part->start_sect) {
123 mutex_unlock(&bdevp->bd_mutex);
124 mutex_unlock(&bdev->bd_mutex);
130 disk_part_iter_init(&piter, disk,
131 DISK_PITER_INCL_EMPTY);
132 while ((lpart = disk_part_iter_next(&piter))) {
133 if (lpart->partno != partno &&
134 !(start + length <= lpart->start_sect ||
135 start >= lpart->start_sect + lpart->nr_sects)
137 disk_part_iter_exit(&piter);
138 mutex_unlock(&bdevp->bd_mutex);
139 mutex_unlock(&bdev->bd_mutex);
145 disk_part_iter_exit(&piter);
146 part_nr_sects_write(part, (sector_t)length);
147 i_size_write(bdevp->bd_inode, p.length);
148 mutex_unlock(&bdevp->bd_mutex);
149 mutex_unlock(&bdev->bd_mutex);
159 * This is an exported API for the block driver, and will not
160 * acquire bd_mutex. This API should be used in case that
161 * caller has held bd_mutex already.
163 int __blkdev_reread_part(struct block_device *bdev)
165 struct gendisk *disk = bdev->bd_disk;
167 if (!disk_part_scan_enabled(disk) || bdev != bdev->bd_contains)
169 if (!capable(CAP_SYS_ADMIN))
172 lockdep_assert_held(&bdev->bd_mutex);
174 return rescan_partitions(disk, bdev);
176 EXPORT_SYMBOL(__blkdev_reread_part);
179 * This is an exported API for the block driver, and will
180 * try to acquire bd_mutex. If bd_mutex has been held already
181 * in current context, please call __blkdev_reread_part().
183 * Make sure the held locks in current context aren't required
184 * in open()/close() handler and I/O path for avoiding ABBA deadlock:
185 * - bd_mutex is held before calling block driver's open/close
187 * - reading partition table may submit I/O to the block device
189 int blkdev_reread_part(struct block_device *bdev)
193 mutex_lock(&bdev->bd_mutex);
194 res = __blkdev_reread_part(bdev);
195 mutex_unlock(&bdev->bd_mutex);
199 EXPORT_SYMBOL(blkdev_reread_part);
201 static int blk_ioctl_discard(struct block_device *bdev, fmode_t mode,
202 unsigned long arg, unsigned long flags)
206 struct request_queue *q = bdev_get_queue(bdev);
207 struct address_space *mapping = bdev->bd_inode->i_mapping;
210 if (!(mode & FMODE_WRITE))
213 if (!blk_queue_discard(q))
216 if (copy_from_user(range, (void __user *)arg, sizeof(range)))
227 if (start + len > i_size_read(bdev->bd_inode))
229 truncate_inode_pages_range(mapping, start, start + len - 1);
230 return blkdev_issue_discard(bdev, start >> 9, len >> 9,
234 static int blk_ioctl_zeroout(struct block_device *bdev, fmode_t mode,
238 struct address_space *mapping;
239 uint64_t start, end, len;
241 if (!(mode & FMODE_WRITE))
244 if (copy_from_user(range, (void __user *)arg, sizeof(range)))
249 end = start + len - 1;
255 if (end >= (uint64_t)i_size_read(bdev->bd_inode))
260 /* Invalidate the page cache, including dirty pages */
261 mapping = bdev->bd_inode->i_mapping;
262 truncate_inode_pages_range(mapping, start, end);
264 return blkdev_issue_zeroout(bdev, start >> 9, len >> 9, GFP_KERNEL,
265 BLKDEV_ZERO_NOUNMAP);
268 static int put_ushort(unsigned long arg, unsigned short val)
270 return put_user(val, (unsigned short __user *)arg);
273 static int put_int(unsigned long arg, int val)
275 return put_user(val, (int __user *)arg);
278 static int put_uint(unsigned long arg, unsigned int val)
280 return put_user(val, (unsigned int __user *)arg);
283 static int put_long(unsigned long arg, long val)
285 return put_user(val, (long __user *)arg);
288 static int put_ulong(unsigned long arg, unsigned long val)
290 return put_user(val, (unsigned long __user *)arg);
293 static int put_u64(unsigned long arg, u64 val)
295 return put_user(val, (u64 __user *)arg);
298 int __blkdev_driver_ioctl(struct block_device *bdev, fmode_t mode,
299 unsigned cmd, unsigned long arg)
301 struct gendisk *disk = bdev->bd_disk;
303 if (disk->fops->ioctl)
304 return disk->fops->ioctl(bdev, mode, cmd, arg);
309 * For the record: _GPL here is only because somebody decided to slap it
310 * on the previous export. Sheer idiocy, since it wasn't copyrightable
311 * at all and could be open-coded without any exports by anybody who cares.
313 EXPORT_SYMBOL_GPL(__blkdev_driver_ioctl);
315 static int blkdev_pr_register(struct block_device *bdev,
316 struct pr_registration __user *arg)
318 const struct pr_ops *ops = bdev->bd_disk->fops->pr_ops;
319 struct pr_registration reg;
321 if (!capable(CAP_SYS_ADMIN))
323 if (!ops || !ops->pr_register)
325 if (copy_from_user(®, arg, sizeof(reg)))
328 if (reg.flags & ~PR_FL_IGNORE_KEY)
330 return ops->pr_register(bdev, reg.old_key, reg.new_key, reg.flags);
333 static int blkdev_pr_reserve(struct block_device *bdev,
334 struct pr_reservation __user *arg)
336 const struct pr_ops *ops = bdev->bd_disk->fops->pr_ops;
337 struct pr_reservation rsv;
339 if (!capable(CAP_SYS_ADMIN))
341 if (!ops || !ops->pr_reserve)
343 if (copy_from_user(&rsv, arg, sizeof(rsv)))
346 if (rsv.flags & ~PR_FL_IGNORE_KEY)
348 return ops->pr_reserve(bdev, rsv.key, rsv.type, rsv.flags);
351 static int blkdev_pr_release(struct block_device *bdev,
352 struct pr_reservation __user *arg)
354 const struct pr_ops *ops = bdev->bd_disk->fops->pr_ops;
355 struct pr_reservation rsv;
357 if (!capable(CAP_SYS_ADMIN))
359 if (!ops || !ops->pr_release)
361 if (copy_from_user(&rsv, arg, sizeof(rsv)))
366 return ops->pr_release(bdev, rsv.key, rsv.type);
369 static int blkdev_pr_preempt(struct block_device *bdev,
370 struct pr_preempt __user *arg, bool abort)
372 const struct pr_ops *ops = bdev->bd_disk->fops->pr_ops;
375 if (!capable(CAP_SYS_ADMIN))
377 if (!ops || !ops->pr_preempt)
379 if (copy_from_user(&p, arg, sizeof(p)))
384 return ops->pr_preempt(bdev, p.old_key, p.new_key, p.type, abort);
387 static int blkdev_pr_clear(struct block_device *bdev,
388 struct pr_clear __user *arg)
390 const struct pr_ops *ops = bdev->bd_disk->fops->pr_ops;
393 if (!capable(CAP_SYS_ADMIN))
395 if (!ops || !ops->pr_clear)
397 if (copy_from_user(&c, arg, sizeof(c)))
402 return ops->pr_clear(bdev, c.key);
406 * Is it an unrecognized ioctl? The correct returns are either
407 * ENOTTY (final) or ENOIOCTLCMD ("I don't know this one, try a
408 * fallback"). ENOIOCTLCMD gets turned into ENOTTY by the ioctl
409 * code before returning.
411 * Confused drivers sometimes return EINVAL, which is wrong. It
412 * means "I understood the ioctl command, but the parameters to
415 * We should aim to just fix the broken drivers, the EINVAL case
418 static inline int is_unrecognized_ioctl(int ret)
420 return ret == -EINVAL ||
425 static int blkdev_flushbuf(struct block_device *bdev, fmode_t mode,
426 unsigned cmd, unsigned long arg)
430 if (!capable(CAP_SYS_ADMIN))
433 ret = __blkdev_driver_ioctl(bdev, mode, cmd, arg);
434 if (!is_unrecognized_ioctl(ret))
438 invalidate_bdev(bdev);
442 static int blkdev_roset(struct block_device *bdev, fmode_t mode,
443 unsigned cmd, unsigned long arg)
447 if (!capable(CAP_SYS_ADMIN))
450 ret = __blkdev_driver_ioctl(bdev, mode, cmd, arg);
451 if (!is_unrecognized_ioctl(ret))
453 if (get_user(n, (int __user *)arg))
455 set_device_ro(bdev, n);
459 static int blkdev_getgeo(struct block_device *bdev,
460 struct hd_geometry __user *argp)
462 struct gendisk *disk = bdev->bd_disk;
463 struct hd_geometry geo;
468 if (!disk->fops->getgeo)
472 * We need to set the startsect first, the driver may
473 * want to override it.
475 memset(&geo, 0, sizeof(geo));
476 geo.start = get_start_sect(bdev);
477 ret = disk->fops->getgeo(bdev, &geo);
480 if (copy_to_user(argp, &geo, sizeof(geo)))
485 /* set the logical block size */
486 static int blkdev_bszset(struct block_device *bdev, fmode_t mode,
491 if (!capable(CAP_SYS_ADMIN))
495 if (get_user(n, argp))
498 if (!(mode & FMODE_EXCL)) {
500 if (blkdev_get(bdev, mode | FMODE_EXCL, &bdev) < 0)
504 ret = set_blocksize(bdev, n);
505 if (!(mode & FMODE_EXCL))
506 blkdev_put(bdev, mode | FMODE_EXCL);
511 * always keep this in sync with compat_blkdev_ioctl()
513 int blkdev_ioctl(struct block_device *bdev, fmode_t mode, unsigned cmd,
516 void __user *argp = (void __user *)arg;
518 unsigned int max_sectors;
522 return blkdev_flushbuf(bdev, mode, cmd, arg);
524 return blkdev_roset(bdev, mode, cmd, arg);
526 return blk_ioctl_discard(bdev, mode, arg, 0);
528 return blk_ioctl_discard(bdev, mode, arg,
529 BLKDEV_DISCARD_SECURE);
531 return blk_ioctl_zeroout(bdev, mode, arg);
533 return blkdev_report_zones_ioctl(bdev, mode, cmd, arg);
535 return blkdev_reset_zones_ioctl(bdev, mode, cmd, arg);
537 return put_uint(arg, bdev_zone_sectors(bdev));
539 return put_uint(arg, blkdev_nr_zones(bdev));
541 return blkdev_getgeo(bdev, argp);
546 return put_long(arg, (bdev->bd_bdi->ra_pages*PAGE_SIZE) / 512);
548 return put_int(arg, bdev_read_only(bdev) != 0);
549 case BLKBSZGET: /* get block device soft block size (cf. BLKSSZGET) */
550 return put_int(arg, block_size(bdev));
551 case BLKSSZGET: /* get block device logical block size */
552 return put_int(arg, bdev_logical_block_size(bdev));
553 case BLKPBSZGET: /* get block device physical block size */
554 return put_uint(arg, bdev_physical_block_size(bdev));
556 return put_uint(arg, bdev_io_min(bdev));
558 return put_uint(arg, bdev_io_opt(bdev));
560 return put_int(arg, bdev_alignment_offset(bdev));
561 case BLKDISCARDZEROES:
562 return put_uint(arg, 0);
564 max_sectors = min_t(unsigned int, USHRT_MAX,
565 queue_max_sectors(bdev_get_queue(bdev)));
566 return put_ushort(arg, max_sectors);
568 return put_ushort(arg, !blk_queue_nonrot(bdev_get_queue(bdev)));
571 if(!capable(CAP_SYS_ADMIN))
573 bdev->bd_bdi->ra_pages = (arg * 512) / PAGE_SIZE;
576 return blkdev_bszset(bdev, mode, argp);
578 return blkpg_ioctl(bdev, argp);
580 return blkdev_reread_part(bdev);
582 size = i_size_read(bdev->bd_inode);
583 if ((size >> 9) > ~0UL)
585 return put_ulong(arg, size >> 9);
587 return put_u64(arg, i_size_read(bdev->bd_inode));
591 case BLKTRACETEARDOWN:
592 return blk_trace_ioctl(bdev, cmd, argp);
593 case IOC_PR_REGISTER:
594 return blkdev_pr_register(bdev, argp);
596 return blkdev_pr_reserve(bdev, argp);
598 return blkdev_pr_release(bdev, argp);
600 return blkdev_pr_preempt(bdev, argp, false);
601 case IOC_PR_PREEMPT_ABORT:
602 return blkdev_pr_preempt(bdev, argp, true);
604 return blkdev_pr_clear(bdev, argp);
606 return __blkdev_driver_ioctl(bdev, mode, cmd, arg);
609 EXPORT_SYMBOL_GPL(blkdev_ioctl);