block/ioctl.c

   1 // SPDX-License-Identifier: GPL-2.0
   2 #include <linux/capability.h>
   3 #include <linux/blkdev.h>
   4 #include <linux/export.h>
   5 #include <linux/gfp.h>
   6 #include <linux/blkpg.h>
   7 #include <linux/hdreg.h>
   8 #include <linux/backing-dev.h>
   9 #include <linux/fs.h>
  10 #include <linux/blktrace_api.h>
  11 #include <linux/pr.h>
  12 #include <linux/uaccess.h>
  13
  14 static int blkpg_ioctl(struct block_device *bdev, struct blkpg_ioctl_arg __user *arg)
  15 {
  16         struct block_device *bdevp;
  17         struct gendisk *disk;
  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;
  23         int partno;
  24
  25         if (!capable(CAP_SYS_ADMIN))
  26                 return -EACCES;
  27         if (copy_from_user(&a, arg, sizeof(struct blkpg_ioctl_arg)))
  28                 return -EFAULT;
  29         if (copy_from_user(&p, a.data, sizeof(struct blkpg_partition)))
  30                 return -EFAULT;
  31         disk = bdev->bd_disk;
  32         if (bdev != bdev->bd_contains)
  33                 return -EINVAL;
  34         partno = p.pno;
  35         if (partno <= 0)
  36                 return -EINVAL;
  37         switch (a.op) {
  38                 case BLKPG_ADD_PARTITION:
  39                         start = p.start >> 9;
  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)
  47                                         return -EINVAL;
  48                         }
  49                         /* check if partition is aligned to blocksize */
  50                         if (p.start & (bdev_logical_block_size(bdev) - 1))
  51                                 return -EINVAL;
  52
  53                         mutex_lock(&bdev->bd_mutex);
  54
  55                         /* overlap? */
  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);
  63                                         return -EBUSY;
  64                                 }
  65                         }
  66                         disk_part_iter_exit(&piter);
  67
  68                         /* all seems OK */
  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);
  75                         if (!part)
  76                                 return -ENXIO;
  77
  78                         bdevp = bdget(part_devt(part));
  79                         disk_put_part(part);
  80                         if (!bdevp)
  81                                 return -ENOMEM;
  82
  83                         mutex_lock(&bdevp->bd_mutex);
  84                         if (bdevp->bd_openers) {
  85                                 mutex_unlock(&bdevp->bd_mutex);
  86                                 bdput(bdevp);
  87                                 return -EBUSY;
  88                         }
  89                         /* all seems OK */
  90                         fsync_bdev(bdevp);
  91                         invalidate_bdev(bdevp);
  92
  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);
  97                         bdput(bdevp);
  98
  99                         return 0;
 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)
 110                                         return -EINVAL;
 111                         }
 112                         part = disk_get_part(disk, partno);
 113                         if (!part)
 114                                 return -ENXIO;
 115                         bdevp = bdget(part_devt(part));
 116                         if (!bdevp) {
 117                                 disk_put_part(part);
 118                                 return -ENOMEM;
 119                         }
 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);
 125                                 bdput(bdevp);
 126                                 disk_put_part(part);
 127                                 return -EINVAL;
 128                         }
 129                         /* overlap? */
 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)
 136                                    ) {
 137                                         disk_part_iter_exit(&piter);
 138                                         mutex_unlock(&bdevp->bd_mutex);
 139                                         mutex_unlock(&bdev->bd_mutex);
 140                                         bdput(bdevp);
 141                                         disk_put_part(part);
 142                                         return -EBUSY;
 143                                 }
 144                         }
 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);
 150                         bdput(bdevp);
 151                         disk_put_part(part);
 152                         return 0;
 153                 default:
 154                         return -EINVAL;
 155         }
 156 }
 157
 158 static int blkdev_reread_part(struct block_device *bdev)
 159 {
 160         int ret;
 161
 162         if (!disk_part_scan_enabled(bdev->bd_disk) || bdev != bdev->bd_contains)
 163                 return -EINVAL;
 164         if (!capable(CAP_SYS_ADMIN))
 165                 return -EACCES;
 166
 167         mutex_lock(&bdev->bd_mutex);
 168         ret = bdev_disk_changed(bdev, false);
 169         mutex_unlock(&bdev->bd_mutex);
 170
 171         return ret;
 172 }
 173
 174 static int blk_ioctl_discard(struct block_device *bdev, fmode_t mode,
 175                 unsigned long arg, unsigned long flags)
 176 {
 177         uint64_t range[2];
 178         uint64_t start, len;
 179         struct request_queue *q = bdev_get_queue(bdev);
 180         struct address_space *mapping = bdev->bd_inode->i_mapping;
 181
 182
 183         if (!(mode & FMODE_WRITE))
 184                 return -EBADF;
 185
 186         if (!blk_queue_discard(q))
 187                 return -EOPNOTSUPP;
 188
 189         if (copy_from_user(range, (void __user *)arg, sizeof(range)))
 190                 return -EFAULT;
 191
 192         start = range[0];
 193         len = range[1];
 194
 195         if (start & 511)
 196                 return -EINVAL;
 197         if (len & 511)
 198                 return -EINVAL;
 199
 200         if (start + len > i_size_read(bdev->bd_inode))
 201                 return -EINVAL;
 202         truncate_inode_pages_range(mapping, start, start + len - 1);
 203         return blkdev_issue_discard(bdev, start >> 9, len >> 9,
 204                                     GFP_KERNEL, flags);
 205 }
 206
 207 static int blk_ioctl_zeroout(struct block_device *bdev, fmode_t mode,
 208                 unsigned long arg)
 209 {
 210         uint64_t range[2];
 211         struct address_space *mapping;
 212         uint64_t start, end, len;
 213
 214         if (!(mode & FMODE_WRITE))
 215                 return -EBADF;
 216
 217         if (copy_from_user(range, (void __user *)arg, sizeof(range)))
 218                 return -EFAULT;
 219
 220         start = range[0];
 221         len = range[1];
 222         end = start + len - 1;
 223
 224         if (start & 511)
 225                 return -EINVAL;
 226         if (len & 511)
 227                 return -EINVAL;
 228         if (end >= (uint64_t)i_size_read(bdev->bd_inode))
 229                 return -EINVAL;
 230         if (end < start)
 231                 return -EINVAL;
 232
 233         /* Invalidate the page cache, including dirty pages */
 234         mapping = bdev->bd_inode->i_mapping;
 235         truncate_inode_pages_range(mapping, start, end);
 236
 237         return blkdev_issue_zeroout(bdev, start >> 9, len >> 9, GFP_KERNEL,
 238                         BLKDEV_ZERO_NOUNMAP);
 239 }
 240
 241 static int put_ushort(unsigned long arg, unsigned short val)
 242 {
 243         return put_user(val, (unsigned short __user *)arg);
 244 }
 245
 246 static int put_int(unsigned long arg, int val)
 247 {
 248         return put_user(val, (int __user *)arg);
 249 }
 250
 251 static int put_uint(unsigned long arg, unsigned int val)
 252 {
 253         return put_user(val, (unsigned int __user *)arg);
 254 }
 255
 256 static int put_long(unsigned long arg, long val)
 257 {
 258         return put_user(val, (long __user *)arg);
 259 }
 260
 261 static int put_ulong(unsigned long arg, unsigned long val)
 262 {
 263         return put_user(val, (unsigned long __user *)arg);
 264 }
 265
 266 static int put_u64(unsigned long arg, u64 val)
 267 {
 268         return put_user(val, (u64 __user *)arg);
 269 }
 270
 271 int __blkdev_driver_ioctl(struct block_device *bdev, fmode_t mode,
 272                         unsigned cmd, unsigned long arg)
 273 {
 274         struct gendisk *disk = bdev->bd_disk;
 275
 276         if (disk->fops->ioctl)
 277                 return disk->fops->ioctl(bdev, mode, cmd, arg);
 278
 279         return -ENOTTY;
 280 }
 281 /*
 282  * For the record: _GPL here is only because somebody decided to slap it
 283  * on the previous export.  Sheer idiocy, since it wasn't copyrightable
 284  * at all and could be open-coded without any exports by anybody who cares.
 285  */
 286 EXPORT_SYMBOL_GPL(__blkdev_driver_ioctl);
 287
 288 static int blkdev_pr_register(struct block_device *bdev,
 289                 struct pr_registration __user *arg)
 290 {
 291         const struct pr_ops *ops = bdev->bd_disk->fops->pr_ops;
 292         struct pr_registration reg;
 293
 294         if (!capable(CAP_SYS_ADMIN))
 295                 return -EPERM;
 296         if (!ops || !ops->pr_register)
 297                 return -EOPNOTSUPP;
 298         if (copy_from_user(&reg, arg, sizeof(reg)))
 299                 return -EFAULT;
 300
 301         if (reg.flags & ~PR_FL_IGNORE_KEY)
 302                 return -EOPNOTSUPP;
 303         return ops->pr_register(bdev, reg.old_key, reg.new_key, reg.flags);
 304 }
 305
 306 static int blkdev_pr_reserve(struct block_device *bdev,
 307                 struct pr_reservation __user *arg)
 308 {
 309         const struct pr_ops *ops = bdev->bd_disk->fops->pr_ops;
 310         struct pr_reservation rsv;
 311
 312         if (!capable(CAP_SYS_ADMIN))
 313                 return -EPERM;
 314         if (!ops || !ops->pr_reserve)
 315                 return -EOPNOTSUPP;
 316         if (copy_from_user(&rsv, arg, sizeof(rsv)))
 317                 return -EFAULT;
 318
 319         if (rsv.flags & ~PR_FL_IGNORE_KEY)
 320                 return -EOPNOTSUPP;
 321         return ops->pr_reserve(bdev, rsv.key, rsv.type, rsv.flags);
 322 }
 323
 324 static int blkdev_pr_release(struct block_device *bdev,
 325                 struct pr_reservation __user *arg)
 326 {
 327         const struct pr_ops *ops = bdev->bd_disk->fops->pr_ops;
 328         struct pr_reservation rsv;
 329
 330         if (!capable(CAP_SYS_ADMIN))
 331                 return -EPERM;
 332         if (!ops || !ops->pr_release)
 333                 return -EOPNOTSUPP;
 334         if (copy_from_user(&rsv, arg, sizeof(rsv)))
 335                 return -EFAULT;
 336
 337         if (rsv.flags)
 338                 return -EOPNOTSUPP;
 339         return ops->pr_release(bdev, rsv.key, rsv.type);
 340 }
 341
 342 static int blkdev_pr_preempt(struct block_device *bdev,
 343                 struct pr_preempt __user *arg, bool abort)
 344 {
 345         const struct pr_ops *ops = bdev->bd_disk->fops->pr_ops;
 346         struct pr_preempt p;
 347
 348         if (!capable(CAP_SYS_ADMIN))
 349                 return -EPERM;
 350         if (!ops || !ops->pr_preempt)
 351                 return -EOPNOTSUPP;
 352         if (copy_from_user(&p, arg, sizeof(p)))
 353                 return -EFAULT;
 354
 355         if (p.flags)
 356                 return -EOPNOTSUPP;
 357         return ops->pr_preempt(bdev, p.old_key, p.new_key, p.type, abort);
 358 }
 359
 360 static int blkdev_pr_clear(struct block_device *bdev,
 361                 struct pr_clear __user *arg)
 362 {
 363         const struct pr_ops *ops = bdev->bd_disk->fops->pr_ops;
 364         struct pr_clear c;
 365
 366         if (!capable(CAP_SYS_ADMIN))
 367                 return -EPERM;
 368         if (!ops || !ops->pr_clear)
 369                 return -EOPNOTSUPP;
 370         if (copy_from_user(&c, arg, sizeof(c)))
 371                 return -EFAULT;
 372
 373         if (c.flags)
 374                 return -EOPNOTSUPP;
 375         return ops->pr_clear(bdev, c.key);
 376 }
 377
 378 /*
 379  * Is it an unrecognized ioctl? The correct returns are either
 380  * ENOTTY (final) or ENOIOCTLCMD ("I don't know this one, try a
 381  * fallback"). ENOIOCTLCMD gets turned into ENOTTY by the ioctl
 382  * code before returning.
 383  *
 384  * Confused drivers sometimes return EINVAL, which is wrong. It
 385  * means "I understood the ioctl command, but the parameters to
 386  * it were wrong".
 387  *
 388  * We should aim to just fix the broken drivers, the EINVAL case
 389  * should go away.
 390  */
 391 static inline int is_unrecognized_ioctl(int ret)
 392 {
 393         return  ret == -EINVAL ||
 394                 ret == -ENOTTY ||
 395                 ret == -ENOIOCTLCMD;
 396 }
 397
 398 static int blkdev_flushbuf(struct block_device *bdev, fmode_t mode,
 399                 unsigned cmd, unsigned long arg)
 400 {
 401         int ret;
 402
 403         if (!capable(CAP_SYS_ADMIN))
 404                 return -EACCES;
 405
 406         ret = __blkdev_driver_ioctl(bdev, mode, cmd, arg);
 407         if (!is_unrecognized_ioctl(ret))
 408                 return ret;
 409
 410         fsync_bdev(bdev);
 411         invalidate_bdev(bdev);
 412         return 0;
 413 }
 414
 415 static int blkdev_roset(struct block_device *bdev, fmode_t mode,
 416                 unsigned cmd, unsigned long arg)
 417 {
 418         int ret, n;
 419
 420         if (!capable(CAP_SYS_ADMIN))
 421                 return -EACCES;
 422
 423         ret = __blkdev_driver_ioctl(bdev, mode, cmd, arg);
 424         if (!is_unrecognized_ioctl(ret))
 425                 return ret;
 426         if (get_user(n, (int __user *)arg))
 427                 return -EFAULT;
 428         set_device_ro(bdev, n);
 429         return 0;
 430 }
 431
 432 static int blkdev_getgeo(struct block_device *bdev,
 433                 struct hd_geometry __user *argp)
 434 {
 435         struct gendisk *disk = bdev->bd_disk;
 436         struct hd_geometry geo;
 437         int ret;
 438
 439         if (!argp)
 440                 return -EINVAL;
 441         if (!disk->fops->getgeo)
 442                 return -ENOTTY;
 443
 444         /*
 445          * We need to set the startsect first, the driver may
 446          * want to override it.
 447          */
 448         memset(&geo, 0, sizeof(geo));
 449         geo.start = get_start_sect(bdev);
 450         ret = disk->fops->getgeo(bdev, &geo);
 451         if (ret)
 452                 return ret;
 453         if (copy_to_user(argp, &geo, sizeof(geo)))
 454                 return -EFAULT;
 455         return 0;
 456 }
 457
 458 /* set the logical block size */
 459 static int blkdev_bszset(struct block_device *bdev, fmode_t mode,
 460                 int __user *argp)
 461 {
 462         int ret, n;
 463
 464         if (!capable(CAP_SYS_ADMIN))
 465                 return -EACCES;
 466         if (!argp)
 467                 return -EINVAL;
 468         if (get_user(n, argp))
 469                 return -EFAULT;
 470
 471         if (!(mode & FMODE_EXCL)) {
 472                 bdgrab(bdev);
 473                 if (blkdev_get(bdev, mode | FMODE_EXCL, &bdev) < 0)
 474                         return -EBUSY;
 475         }
 476
 477         ret = set_blocksize(bdev, n);
 478         if (!(mode & FMODE_EXCL))
 479                 blkdev_put(bdev, mode | FMODE_EXCL);
 480         return ret;
 481 }
 482
 483 /*
 484  * always keep this in sync with compat_blkdev_ioctl()
 485  */
 486 int blkdev_ioctl(struct block_device *bdev, fmode_t mode, unsigned cmd,
 487                         unsigned long arg)
 488 {
 489         void __user *argp = (void __user *)arg;
 490         loff_t size;
 491         unsigned int max_sectors;
 492
 493         switch (cmd) {
 494         case BLKFLSBUF:
 495                 return blkdev_flushbuf(bdev, mode, cmd, arg);
 496         case BLKROSET:
 497                 return blkdev_roset(bdev, mode, cmd, arg);
 498         case BLKDISCARD:
 499                 return blk_ioctl_discard(bdev, mode, arg, 0);
 500         case BLKSECDISCARD:
 501                 return blk_ioctl_discard(bdev, mode, arg,
 502                                 BLKDEV_DISCARD_SECURE);
 503         case BLKZEROOUT:
 504                 return blk_ioctl_zeroout(bdev, mode, arg);
 505         case BLKREPORTZONE:
 506                 return blkdev_report_zones_ioctl(bdev, mode, cmd, arg);
 507         case BLKRESETZONE:
 508         case BLKOPENZONE:
 509         case BLKCLOSEZONE:
 510         case BLKFINISHZONE:
 511                 return blkdev_zone_mgmt_ioctl(bdev, mode, cmd, arg);
 512         case BLKGETZONESZ:
 513                 return put_uint(arg, bdev_zone_sectors(bdev));
 514         case BLKGETNRZONES:
 515                 return put_uint(arg, blkdev_nr_zones(bdev->bd_disk));
 516         case HDIO_GETGEO:
 517                 return blkdev_getgeo(bdev, argp);
 518         case BLKRAGET:
 519         case BLKFRAGET:
 520                 if (!arg)
 521                         return -EINVAL;
 522                 return put_long(arg, (bdev->bd_bdi->ra_pages*PAGE_SIZE) / 512);
 523         case BLKROGET:
 524                 return put_int(arg, bdev_read_only(bdev) != 0);
 525         case BLKBSZGET: /* get block device soft block size (cf. BLKSSZGET) */
 526                 return put_int(arg, block_size(bdev));
 527         case BLKSSZGET: /* get block device logical block size */
 528                 return put_int(arg, bdev_logical_block_size(bdev));
 529         case BLKPBSZGET: /* get block device physical block size */
 530                 return put_uint(arg, bdev_physical_block_size(bdev));
 531         case BLKIOMIN:
 532                 return put_uint(arg, bdev_io_min(bdev));
 533         case BLKIOOPT:
 534                 return put_uint(arg, bdev_io_opt(bdev));
 535         case BLKALIGNOFF:
 536                 return put_int(arg, bdev_alignment_offset(bdev));
 537         case BLKDISCARDZEROES:
 538                 return put_uint(arg, 0);
 539         case BLKSECTGET:
 540                 max_sectors = min_t(unsigned int, USHRT_MAX,
 541                                     queue_max_sectors(bdev_get_queue(bdev)));
 542                 return put_ushort(arg, max_sectors);
 543         case BLKROTATIONAL:
 544                 return put_ushort(arg, !blk_queue_nonrot(bdev_get_queue(bdev)));
 545         case BLKRASET:
 546         case BLKFRASET:
 547                 if(!capable(CAP_SYS_ADMIN))
 548                         return -EACCES;
 549                 bdev->bd_bdi->ra_pages = (arg * 512) / PAGE_SIZE;
 550                 return 0;
 551         case BLKBSZSET:
 552                 return blkdev_bszset(bdev, mode, argp);
 553         case BLKPG:
 554                 return blkpg_ioctl(bdev, argp);
 555         case BLKRRPART:
 556                 return blkdev_reread_part(bdev);
 557         case BLKGETSIZE:
 558                 size = i_size_read(bdev->bd_inode);
 559                 if ((size >> 9) > ~0UL)
 560                         return -EFBIG;
 561                 return put_ulong(arg, size >> 9);
 562         case BLKGETSIZE64:
 563                 return put_u64(arg, i_size_read(bdev->bd_inode));
 564         case BLKTRACESTART:
 565         case BLKTRACESTOP:
 566         case BLKTRACESETUP:
 567         case BLKTRACETEARDOWN:
 568                 return blk_trace_ioctl(bdev, cmd, argp);
 569         case IOC_PR_REGISTER:
 570                 return blkdev_pr_register(bdev, argp);
 571         case IOC_PR_RESERVE:
 572                 return blkdev_pr_reserve(bdev, argp);
 573         case IOC_PR_RELEASE:
 574                 return blkdev_pr_release(bdev, argp);
 575         case IOC_PR_PREEMPT:
 576                 return blkdev_pr_preempt(bdev, argp, false);
 577         case IOC_PR_PREEMPT_ABORT:
 578                 return blkdev_pr_preempt(bdev, argp, true);
 579         case IOC_PR_CLEAR:
 580                 return blkdev_pr_clear(bdev, argp);
 581         default:
 582                 return __blkdev_driver_ioctl(bdev, mode, cmd, arg);
 583         }
 584 }
 585 EXPORT_SYMBOL_GPL(blkdev_ioctl);