fs/ioctl.c

   1 /*
   2  *  linux/fs/ioctl.c
   3  *
   4  *  Copyright (C) 1991, 1992  Linus Torvalds
   5  */
   6
   7 #include <linux/syscalls.h>
   8 #include <linux/mm.h>
   9 #include <linux/capability.h>
  10 #include <linux/file.h>
  11 #include <linux/fs.h>
  12 #include <linux/security.h>
  13 #include <linux/module.h>
  14 #include <linux/uaccess.h>
  15 #include <linux/writeback.h>
  16 #include <linux/buffer_head.h>
  17 #include <linux/falloc.h>
  18
  19 #include <asm/ioctls.h>
  20
  21 /* So that the fiemap access checks can't overflow on 32 bit machines. */
  22 #define FIEMAP_MAX_EXTENTS      (UINT_MAX / sizeof(struct fiemap_extent))
  23
  24 /**
  25  * vfs_ioctl - call filesystem specific ioctl methods
  26  * @filp:       open file to invoke ioctl method on
  27  * @cmd:        ioctl command to execute
  28  * @arg:        command-specific argument for ioctl
  29  *
  30  * Invokes filesystem specific ->unlocked_ioctl, if one exists; otherwise
  31  * returns -ENOTTY.
  32  *
  33  * Returns 0 on success, -errno on error.
  34  */
  35 static long vfs_ioctl(struct file *filp, unsigned int cmd,
  36                       unsigned long arg)
  37 {
  38         int error = -ENOTTY;
  39
  40         if (!filp->f_op || !filp->f_op->unlocked_ioctl)
  41                 goto out;
  42
  43         error = filp->f_op->unlocked_ioctl(filp, cmd, arg);
  44         if (error == -ENOIOCTLCMD)
  45                 error = -EINVAL;
  46  out:
  47         return error;
  48 }
  49
  50 static int ioctl_fibmap(struct file *filp, int __user *p)
  51 {
  52         struct address_space *mapping = filp->f_mapping;
  53         int res, block;
  54
  55         /* do we support this mess? */
  56         if (!mapping->a_ops->bmap)
  57                 return -EINVAL;
  58         if (!capable(CAP_SYS_RAWIO))
  59                 return -EPERM;
  60         res = get_user(block, p);
  61         if (res)
  62                 return res;
  63         res = mapping->a_ops->bmap(mapping, block);
  64         return put_user(res, p);
  65 }
  66
  67 /**
  68  * fiemap_fill_next_extent - Fiemap helper function
  69  * @fieinfo:    Fiemap context passed into ->fiemap
  70  * @logical:    Extent logical start offset, in bytes
  71  * @phys:       Extent physical start offset, in bytes
  72  * @len:        Extent length, in bytes
  73  * @flags:      FIEMAP_EXTENT flags that describe this extent
  74  *
  75  * Called from file system ->fiemap callback. Will populate extent
  76  * info as passed in via arguments and copy to user memory. On
  77  * success, extent count on fieinfo is incremented.
  78  *
  79  * Returns 0 on success, -errno on error, 1 if this was the last
  80  * extent that will fit in user array.
  81  */
  82 #define SET_UNKNOWN_FLAGS       (FIEMAP_EXTENT_DELALLOC)
  83 #define SET_NO_UNMOUNTED_IO_FLAGS       (FIEMAP_EXTENT_DATA_ENCRYPTED)
  84 #define SET_NOT_ALIGNED_FLAGS   (FIEMAP_EXTENT_DATA_TAIL|FIEMAP_EXTENT_DATA_INLINE)
  85 int fiemap_fill_next_extent(struct fiemap_extent_info *fieinfo, u64 logical,
  86                             u64 phys, u64 len, u32 flags)
  87 {
  88         struct fiemap_extent extent;
  89         struct fiemap_extent __user *dest = fieinfo->fi_extents_start;
  90
  91         /* only count the extents */
  92         if (fieinfo->fi_extents_max == 0) {
  93                 fieinfo->fi_extents_mapped++;
  94                 return (flags & FIEMAP_EXTENT_LAST) ? 1 : 0;
  95         }
  96
  97         if (fieinfo->fi_extents_mapped >= fieinfo->fi_extents_max)
  98                 return 1;
  99
 100         if (flags & SET_UNKNOWN_FLAGS)
 101                 flags |= FIEMAP_EXTENT_UNKNOWN;
 102         if (flags & SET_NO_UNMOUNTED_IO_FLAGS)
 103                 flags |= FIEMAP_EXTENT_ENCODED;
 104         if (flags & SET_NOT_ALIGNED_FLAGS)
 105                 flags |= FIEMAP_EXTENT_NOT_ALIGNED;
 106
 107         memset(&extent, 0, sizeof(extent));
 108         extent.fe_logical = logical;
 109         extent.fe_physical = phys;
 110         extent.fe_length = len;
 111         extent.fe_flags = flags;
 112
 113         dest += fieinfo->fi_extents_mapped;
 114         if (copy_to_user(dest, &extent, sizeof(extent)))
 115                 return -EFAULT;
 116
 117         fieinfo->fi_extents_mapped++;
 118         if (fieinfo->fi_extents_mapped == fieinfo->fi_extents_max)
 119                 return 1;
 120         return (flags & FIEMAP_EXTENT_LAST) ? 1 : 0;
 121 }
 122 EXPORT_SYMBOL(fiemap_fill_next_extent);
 123
 124 /**
 125  * fiemap_check_flags - check validity of requested flags for fiemap
 126  * @fieinfo:    Fiemap context passed into ->fiemap
 127  * @fs_flags:   Set of fiemap flags that the file system understands
 128  *
 129  * Called from file system ->fiemap callback. This will compute the
 130  * intersection of valid fiemap flags and those that the fs supports. That
 131  * value is then compared against the user supplied flags. In case of bad user
 132  * flags, the invalid values will be written into the fieinfo structure, and
 133  * -EBADR is returned, which tells ioctl_fiemap() to return those values to
 134  * userspace. For this reason, a return code of -EBADR should be preserved.
 135  *
 136  * Returns 0 on success, -EBADR on bad flags.
 137  */
 138 int fiemap_check_flags(struct fiemap_extent_info *fieinfo, u32 fs_flags)
 139 {
 140         u32 incompat_flags;
 141
 142         incompat_flags = fieinfo->fi_flags & ~(FIEMAP_FLAGS_COMPAT & fs_flags);
 143         if (incompat_flags) {
 144                 fieinfo->fi_flags = incompat_flags;
 145                 return -EBADR;
 146         }
 147         return 0;
 148 }
 149 EXPORT_SYMBOL(fiemap_check_flags);
 150
 151 static int fiemap_check_ranges(struct super_block *sb,
 152                                u64 start, u64 len, u64 *new_len)
 153 {
 154         u64 maxbytes = (u64) sb->s_maxbytes;
 155
 156         *new_len = len;
 157
 158         if (len == 0)
 159                 return -EINVAL;
 160
 161         if (start > maxbytes)
 162                 return -EFBIG;
 163
 164         /*
 165          * Shrink request scope to what the fs can actually handle.
 166          */
 167         if (len > maxbytes || (maxbytes - len) < start)
 168                 *new_len = maxbytes - start;
 169
 170         return 0;
 171 }
 172
 173 static int ioctl_fiemap(struct file *filp, unsigned long arg)
 174 {
 175         struct fiemap fiemap;
 176         struct fiemap __user *ufiemap = (struct fiemap __user *) arg;
 177         struct fiemap_extent_info fieinfo = { 0, };
 178         struct inode *inode = filp->f_path.dentry->d_inode;
 179         struct super_block *sb = inode->i_sb;
 180         u64 len;
 181         int error;
 182
 183         if (!inode->i_op->fiemap)
 184                 return -EOPNOTSUPP;
 185
 186         if (copy_from_user(&fiemap, ufiemap, sizeof(fiemap)))
 187                 return -EFAULT;
 188
 189         if (fiemap.fm_extent_count > FIEMAP_MAX_EXTENTS)
 190                 return -EINVAL;
 191
 192         error = fiemap_check_ranges(sb, fiemap.fm_start, fiemap.fm_length,
 193                                     &len);
 194         if (error)
 195                 return error;
 196
 197         fieinfo.fi_flags = fiemap.fm_flags;
 198         fieinfo.fi_extents_max = fiemap.fm_extent_count;
 199         fieinfo.fi_extents_start = ufiemap->fm_extents;
 200
 201         if (fiemap.fm_extent_count != 0 &&
 202             !access_ok(VERIFY_WRITE, fieinfo.fi_extents_start,
 203                        fieinfo.fi_extents_max * sizeof(struct fiemap_extent)))
 204                 return -EFAULT;
 205
 206         if (fieinfo.fi_flags & FIEMAP_FLAG_SYNC)
 207                 filemap_write_and_wait(inode->i_mapping);
 208
 209         error = inode->i_op->fiemap(inode, &fieinfo, fiemap.fm_start, len);
 210         fiemap.fm_flags = fieinfo.fi_flags;
 211         fiemap.fm_mapped_extents = fieinfo.fi_extents_mapped;
 212         if (copy_to_user(ufiemap, &fiemap, sizeof(fiemap)))
 213                 error = -EFAULT;
 214
 215         return error;
 216 }
 217
 218 #ifdef CONFIG_BLOCK
 219
 220 static inline sector_t logical_to_blk(struct inode *inode, loff_t offset)
 221 {
 222         return (offset >> inode->i_blkbits);
 223 }
 224
 225 static inline loff_t blk_to_logical(struct inode *inode, sector_t blk)
 226 {
 227         return (blk << inode->i_blkbits);
 228 }
 229
 230 /**
 231  * __generic_block_fiemap - FIEMAP for block based inodes (no locking)
 232  * @inode: the inode to map
 233  * @fieinfo: the fiemap info struct that will be passed back to userspace
 234  * @start: where to start mapping in the inode
 235  * @len: how much space to map
 236  * @get_block: the fs's get_block function
 237  *
 238  * This does FIEMAP for block based inodes.  Basically it will just loop
 239  * through get_block until we hit the number of extents we want to map, or we
 240  * go past the end of the file and hit a hole.
 241  *
 242  * If it is possible to have data blocks beyond a hole past @inode->i_size, then
 243  * please do not use this function, it will stop at the first unmapped block
 244  * beyond i_size.
 245  *
 246  * If you use this function directly, you need to do your own locking. Use
 247  * generic_block_fiemap if you want the locking done for you.
 248  */
 249
 250 int __generic_block_fiemap(struct inode *inode,
 251                            struct fiemap_extent_info *fieinfo, loff_t start,
 252                            loff_t len, get_block_t *get_block)
 253 {
 254         struct buffer_head map_bh;
 255         sector_t start_blk, last_blk;
 256         loff_t isize = i_size_read(inode);
 257         u64 logical = 0, phys = 0, size = 0;
 258         u32 flags = FIEMAP_EXTENT_MERGED;
 259         bool past_eof = false, whole_file = false;
 260         int ret = 0;
 261
 262         ret = fiemap_check_flags(fieinfo, FIEMAP_FLAG_SYNC);
 263         if (ret)
 264                 return ret;
 265
 266         /*
 267          * Either the i_mutex or other appropriate locking needs to be held
 268          * since we expect isize to not change at all through the duration of
 269          * this call.
 270          */
 271         if (len >= isize) {
 272                 whole_file = true;
 273                 len = isize;
 274         }
 275
 276         /*
 277          * Some filesystems can't deal with being asked to map less than
 278          * blocksize, so make sure our len is at least block length.
 279          */
 280         if (logical_to_blk(inode, len) == 0)
 281                 len = blk_to_logical(inode, 1);
 282
 283         start_blk = logical_to_blk(inode, start);
 284         last_blk = logical_to_blk(inode, start + len - 1);
 285
 286         do {
 287                 /*
 288                  * we set b_size to the total size we want so it will map as
 289                  * many contiguous blocks as possible at once
 290                  */
 291                 memset(&map_bh, 0, sizeof(struct buffer_head));
 292                 map_bh.b_size = len;
 293
 294                 ret = get_block(inode, start_blk, &map_bh, 0);
 295                 if (ret)
 296                         break;
 297
 298                 /* HOLE */
 299                 if (!buffer_mapped(&map_bh)) {
 300                         start_blk++;
 301
 302                         /*
 303                          * We want to handle the case where there is an
 304                          * allocated block at the front of the file, and then
 305                          * nothing but holes up to the end of the file properly,
 306                          * to make sure that extent at the front gets properly
 307                          * marked with FIEMAP_EXTENT_LAST
 308                          */
 309                         if (!past_eof &&
 310                             blk_to_logical(inode, start_blk) >= isize)
 311                                 past_eof = 1;
 312
 313                         /*
 314                          * First hole after going past the EOF, this is our
 315                          * last extent
 316                          */
 317                         if (past_eof && size) {
 318                                 flags = FIEMAP_EXTENT_MERGED|FIEMAP_EXTENT_LAST;
 319                                 ret = fiemap_fill_next_extent(fieinfo, logical,
 320                                                               phys, size,
 321                                                               flags);
 322                         } else if (size) {
 323                                 ret = fiemap_fill_next_extent(fieinfo, logical,
 324                                                               phys, size, flags);
 325                                 size = 0;
 326                         }
 327
 328                         /* if we have holes up to/past EOF then we're done */
 329                         if (start_blk > last_blk || past_eof || ret)
 330                                 break;
 331                 } else {
 332                         /*
 333                          * We have gone over the length of what we wanted to
 334                          * map, and it wasn't the entire file, so add the extent
 335                          * we got last time and exit.
 336                          *
 337                          * This is for the case where say we want to map all the
 338                          * way up to the second to the last block in a file, but
 339                          * the last block is a hole, making the second to last
 340                          * block FIEMAP_EXTENT_LAST.  In this case we want to
 341                          * see if there is a hole after the second to last block
 342                          * so we can mark it properly.  If we found data after
 343                          * we exceeded the length we were requesting, then we
 344                          * are good to go, just add the extent to the fieinfo
 345                          * and break
 346                          */
 347                         if (start_blk > last_blk && !whole_file) {
 348                                 ret = fiemap_fill_next_extent(fieinfo, logical,
 349                                                               phys, size,
 350                                                               flags);
 351                                 break;
 352                         }
 353
 354                         /*
 355                          * if size != 0 then we know we already have an extent
 356                          * to add, so add it.
 357                          */
 358                         if (size) {
 359                                 ret = fiemap_fill_next_extent(fieinfo, logical,
 360                                                               phys, size,
 361                                                               flags);
 362                                 if (ret)
 363                                         break;
 364                         }
 365
 366                         logical = blk_to_logical(inode, start_blk);
 367                         phys = blk_to_logical(inode, map_bh.b_blocknr);
 368                         size = map_bh.b_size;
 369                         flags = FIEMAP_EXTENT_MERGED;
 370
 371                         start_blk += logical_to_blk(inode, size);
 372
 373                         /*
 374                          * If we are past the EOF, then we need to make sure as
 375                          * soon as we find a hole that the last extent we found
 376                          * is marked with FIEMAP_EXTENT_LAST
 377                          */
 378                         if (!past_eof && logical + size >= isize)
 379                                 past_eof = true;
 380                 }
 381                 cond_resched();
 382         } while (1);
 383
 384         /* If ret is 1 then we just hit the end of the extent array */
 385         if (ret == 1)
 386                 ret = 0;
 387
 388         return ret;
 389 }
 390 EXPORT_SYMBOL(__generic_block_fiemap);
 391
 392 /**
 393  * generic_block_fiemap - FIEMAP for block based inodes
 394  * @inode: The inode to map
 395  * @fieinfo: The mapping information
 396  * @start: The initial block to map
 397  * @len: The length of the extect to attempt to map
 398  * @get_block: The block mapping function for the fs
 399  *
 400  * Calls __generic_block_fiemap to map the inode, after taking
 401  * the inode's mutex lock.
 402  */
 403
 404 int generic_block_fiemap(struct inode *inode,
 405                          struct fiemap_extent_info *fieinfo, u64 start,
 406                          u64 len, get_block_t *get_block)
 407 {
 408         int ret;
 409         mutex_lock(&inode->i_mutex);
 410         ret = __generic_block_fiemap(inode, fieinfo, start, len, get_block);
 411         mutex_unlock(&inode->i_mutex);
 412         return ret;
 413 }
 414 EXPORT_SYMBOL(generic_block_fiemap);
 415
 416 #endif  /*  CONFIG_BLOCK  */
 417
 418 /*
 419  * This provides compatibility with legacy XFS pre-allocation ioctls
 420  * which predate the fallocate syscall.
 421  *
 422  * Only the l_start, l_len and l_whence fields of the 'struct space_resv'
 423  * are used here, rest are ignored.
 424  */
 425 int ioctl_preallocate(struct file *filp, void __user *argp)
 426 {
 427         struct inode *inode = filp->f_path.dentry->d_inode;
 428         struct space_resv sr;
 429
 430         if (copy_from_user(&sr, argp, sizeof(sr)))
 431                 return -EFAULT;
 432
 433         switch (sr.l_whence) {
 434         case SEEK_SET:
 435                 break;
 436         case SEEK_CUR:
 437                 sr.l_start += filp->f_pos;
 438                 break;
 439         case SEEK_END:
 440                 sr.l_start += i_size_read(inode);
 441                 break;
 442         default:
 443                 return -EINVAL;
 444         }
 445
 446         return do_fallocate(filp, FALLOC_FL_KEEP_SIZE, sr.l_start, sr.l_len);
 447 }
 448
 449 static int file_ioctl(struct file *filp, unsigned int cmd,
 450                 unsigned long arg)
 451 {
 452         struct inode *inode = filp->f_path.dentry->d_inode;
 453         int __user *p = (int __user *)arg;
 454
 455         switch (cmd) {
 456         case FIBMAP:
 457                 return ioctl_fibmap(filp, p);
 458         case FIONREAD:
 459                 return put_user(i_size_read(inode) - filp->f_pos, p);
 460         case FS_IOC_RESVSP:
 461         case FS_IOC_RESVSP64:
 462                 return ioctl_preallocate(filp, p);
 463         }
 464
 465         return vfs_ioctl(filp, cmd, arg);
 466 }
 467
 468 static int ioctl_fionbio(struct file *filp, int __user *argp)
 469 {
 470         unsigned int flag;
 471         int on, error;
 472
 473         error = get_user(on, argp);
 474         if (error)
 475                 return error;
 476         flag = O_NONBLOCK;
 477 #ifdef __sparc__
 478         /* SunOS compatibility item. */
 479         if (O_NONBLOCK != O_NDELAY)
 480                 flag |= O_NDELAY;
 481 #endif
 482         spin_lock(&filp->f_lock);
 483         if (on)
 484                 filp->f_flags |= flag;
 485         else
 486                 filp->f_flags &= ~flag;
 487         spin_unlock(&filp->f_lock);
 488         return error;
 489 }
 490
 491 static int ioctl_fioasync(unsigned int fd, struct file *filp,
 492                           int __user *argp)
 493 {
 494         unsigned int flag;
 495         int on, error;
 496
 497         error = get_user(on, argp);
 498         if (error)
 499                 return error;
 500         flag = on ? FASYNC : 0;
 501
 502         /* Did FASYNC state change ? */
 503         if ((flag ^ filp->f_flags) & FASYNC) {
 504                 if (filp->f_op && filp->f_op->fasync)
 505                         /* fasync() adjusts filp->f_flags */
 506                         error = filp->f_op->fasync(fd, filp, on);
 507                 else
 508                         error = -ENOTTY;
 509         }
 510         return error < 0 ? error : 0;
 511 }
 512
 513 static int ioctl_fsfreeze(struct file *filp)
 514 {
 515         struct super_block *sb = filp->f_path.dentry->d_inode->i_sb;
 516
 517         if (!capable(CAP_SYS_ADMIN))
 518                 return -EPERM;
 519
 520         /* If filesystem doesn't support freeze feature, return. */
 521         if (sb->s_op->freeze_fs == NULL)
 522                 return -EOPNOTSUPP;
 523
 524         /* Freeze */
 525         return freeze_super(sb);
 526 }
 527
 528 static int ioctl_fsthaw(struct file *filp)
 529 {
 530         struct super_block *sb = filp->f_path.dentry->d_inode->i_sb;
 531
 532         if (!capable(CAP_SYS_ADMIN))
 533                 return -EPERM;
 534
 535         /* Thaw */
 536         return thaw_super(sb);
 537 }
 538
 539 /*
 540  * When you add any new common ioctls to the switches above and below
 541  * please update compat_sys_ioctl() too.
 542  *
 543  * do_vfs_ioctl() is not for drivers and not intended to be EXPORT_SYMBOL()'d.
 544  * It's just a simple helper for sys_ioctl and compat_sys_ioctl.
 545  */
 546 int do_vfs_ioctl(struct file *filp, unsigned int fd, unsigned int cmd,
 547              unsigned long arg)
 548 {
 549         int error = 0;
 550         int __user *argp = (int __user *)arg;
 551         struct inode *inode = filp->f_path.dentry->d_inode;
 552
 553         switch (cmd) {
 554         case FIOCLEX:
 555                 set_close_on_exec(fd, 1);
 556                 break;
 557
 558         case FIONCLEX:
 559                 set_close_on_exec(fd, 0);
 560                 break;
 561
 562         case FIONBIO:
 563                 error = ioctl_fionbio(filp, argp);
 564                 break;
 565
 566         case FIOASYNC:
 567                 error = ioctl_fioasync(fd, filp, argp);
 568                 break;
 569
 570         case FIOQSIZE:
 571                 if (S_ISDIR(inode->i_mode) || S_ISREG(inode->i_mode) ||
 572                     S_ISLNK(inode->i_mode)) {
 573                         loff_t res = inode_get_bytes(inode);
 574                         error = copy_to_user(argp, &res, sizeof(res)) ?
 575                                         -EFAULT : 0;
 576                 } else
 577                         error = -ENOTTY;
 578                 break;
 579
 580         case FIFREEZE:
 581                 error = ioctl_fsfreeze(filp);
 582                 break;
 583
 584         case FITHAW:
 585                 error = ioctl_fsthaw(filp);
 586                 break;
 587
 588         case FS_IOC_FIEMAP:
 589                 return ioctl_fiemap(filp, arg);
 590
 591         case FIGETBSZ:
 592                 return put_user(inode->i_sb->s_blocksize, argp);
 593
 594         default:
 595                 if (S_ISREG(inode->i_mode))
 596                         error = file_ioctl(filp, cmd, arg);
 597                 else
 598                         error = vfs_ioctl(filp, cmd, arg);
 599                 break;
 600         }
 601         return error;
 602 }
 603
 604 SYSCALL_DEFINE3(ioctl, unsigned int, fd, unsigned int, cmd, unsigned long, arg)
 605 {
 606         struct file *filp;
 607         int error = -EBADF;
 608         int fput_needed;
 609
 610         filp = fget_light(fd, &fput_needed);
 611         if (!filp)
 612                 goto out;
 613
 614         error = security_file_ioctl(filp, cmd, arg);
 615         if (error)
 616                 goto out_fput;
 617
 618         error = do_vfs_ioctl(filp, fd, cmd, arg);
 619  out_fput:
 620         fput_light(filp, fput_needed);
 621  out:
 622         return error;
 623 }